Академический Документы
Профессиональный Документы
Культура Документы
Technical Scale of
Electrochemistry
..
Klans J uttner
Karl-Winnacker Institute, Dechema e.v., Frankfurtan Main, Germany
1.1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.2
1.3
1.4
1.4.1
1.4.2
Fundamentals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Current Efciency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Space-time Yield . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6
7
8
1.5
1.5.1
1.5.2
Electrochemical Thermodynamics . . . . . . . . . . . . . . . . . . . . . . .
Energy Yield . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Specic Electrical Energy Consumption . . . . . . . . . . . . . . . . . . . .
8
11
12
1.6
12
19
1.1
Introduction
galvanic industry, or electrophoretic coating in the automotive industry. The environmental aspect has become a major
issue and a crucial factor in the development of industrial processes to meet
the requirements of sustainable development. In this eld, electrochemistry
offers promising approaches due to its
environmental compatibility and use of
the electron as a clean reagent [2028].
There is common agreement among scientists that electrochemically based processes will be of increasing importance
in the future to meet the economic and
social challenges resulting from urgent
demands of low-grade raw materials utilization, energy savings, and protection of
the environment.
1.2
Figure 1 shows the cell room of the electrolyzer at Bayer Materials Science AG in
Brunsbuttel, Germany.
The main advantages of electrochemical
processes are as follows:
Versatility: Direct or indirect oxidation
and reduction, phase separation, concentration or dilution, biocide functionality, applicability to a variety of media
and pollutants in gases, liquids, and
solids, and treatment of small to large
volumes from microliters up to millions
of liters.
Energy efciency: Lower temperature requirements than their nonelectrochemical counterparts, for example, anodic
destruction of organic pollutants instead
of thermal incineration; power losses
caused by inhomogeneous current distribution, voltage drop, and side reactions being minimized by optimization
of electrode structure and cell design.
Amenability to automation: The system
inherent variables of electrochemical
processes, for example, electrode potential and cell current, are particularly
suitable for facilitating process automation and control.
Cost effectiveness: Cell constructions and
peripheral equipment are generally simple and, if properly designed, also
inexpensive.
Since electrochemical reactions take
place at the interface of an electronic conductor the electrode and an ion conductor the electrolyte electrochemical processes are of heterogeneous nature. This
implies that, despite the advantages mentioned earlier, the performance of electrochemical processes suffers from masstransport limitations and the size of the
specic electrode area. A crucial point is
also the stability of cell components in
..
Cell room of the hydrochloric acid electrolyzer with GDE in Brunsb uttel [28]
(photo: Bayer Material Science AG).
Fig. 1
contact with aggressive media, in particular, the durability and long-term stability
of the electrode material and the catalysts.
1.3
Macro
kinetics
Micro
kinetics
Charge
transfer
Reaction
mechanism
Mass
transport
Heat
transfer
Cell design
1.4
Fundamentals
r=
i, j. (2)
Ai Ai + e e =
Bj Bj ,
(3)
1.4 Fundamentals
Tab. 1 Electrode reactions and standard
potentials Eo [18]
Electrode reaction
Standard
potential
Eo /V vs
SHE
Li+ + e Li
Na+ + e Na
Al3+ + 3e Al
2H2 O + 2e H2 + 2OH
Zn2+ + 2e Zn
Fe2+ + 2e Fe
Ni2+ + 2e Ni
Pb2+ + 2e Pb
2H3 O+ + 2e H2 + 2H2 O
Cu2+ + 2e Cu
[Fe(CN)6 ]3 + e [Fe(CN)6 ]4
O2 + 2H3 O+ + 2e
H2 O2 + 2H2 O
Fe3+ + e Fe2+
Ag+ + e Ag
O2 + 4H3 O+ + 4e 6H2 O
Cl2 + 2e 2Cl
Ce4+ + e Ce3+
+
MnO
4 + 8H3 O + 5e
2+
Mn + 12H2 O
O3 + 2H3 O+ + 2e
O2 + 3H2 O
F2 + 2e 2F
3.045
2.711
1.706
0.8277
0.763
0.409
0.230
0.126
0
0.340
0.356
0.682
0.771
0.799
1.229
1.37
1.443
1.491
1 dne
Ae e dt
2.87
(4)
Qt
F
(5)
(7)
P MP
Qt
e F
reactant R
mR =
R MR
Qt
e F
(8)
Current Efciency
1 dne
1 dnP
=
P dt
e dt
2.07
i
,
e F
(6)
electrochemical reaction.
eP =
mP e F
nP /P
=
P MP Qt
ne /e
(9)
(10)
Space-time Yield
1 dmP
Vr dt
P MP e
ae i(t)
e F
id,lim =
e
F km cR
R
(13)
P
MP e ae km cR
R
(14)
1.5
Electrochemical Thermodynamics
(11)
from charge transfer to a mass-transportcontrolled regime, where the diffusionlimited current density, as the maximum
of the reaction rate, can be reached
(12)
RT
i ln ai
e F
(15)
= nF
Uco
(16)
i Gof,i
o
o
Go298 = H298
T S298
(18)
where T S o is the reversible heat exchanged with the environment. The entropy change S o can be either positive
or negative and is directly related to the
temperature coefcient of the cell
(17)
Uco
T
=
p
1
1 Go298
o
=
S298
nF T
nF
(19)
Reactant
H2
Cl2
H3 O+
Cl
O2
H2 O
Zn0
Zn++
HCl
C
CO
State of aggregation
g
g
aq
aq
g
l
s
aq
g
s
g
Ho298 [kJmol1 ]
0
0
0
167.54
0
285.25
0
152.51
92.53
0
110.5
o [JK 1 mol1 ]
S298
130.74
223.09
0
55.13
205.25
70.12
41.65
106.54
186.79
5.69
198.0
10
o
i Hf,i
T
o
i Sf,i
Oxygen electrode:
1
O2 + 2H+ + 2e = H2 O with
2
o
= +1.229 V versus SHE
EO
2 /H2 O
(21)
(20)
If Go298 is negative, the reaction is exergonic and should occur spontaneously,
otherwise it is endergonic and does not
occur or occurs in the reverse direction
only. The negative sign in Eq. (16) implies
that the cell voltage or standard potential
of an exergonic reaction must be positive.
This calls for a denition of the cell voltage
measurement. According to IUPAC [51],
the cell voltage is the potential difference
between the right half cell I and the left
half cell II, as shown in Fig. 3. The direction of the electronic current in the outer
electronic circuit is dened from left to
right, which means that cathodic reduction
must take place in half cell I and anodic
oxidation in half cell II.
This is demonstrated by combining an
oxygen electrode and a hydrogen electrode
as the two half cells of an electrochemical
cell. The half cell reactions are commonly
formulated as reductions:
Hydrogen electrode:
1
o
H2 with EH
+ /H
2
2
= 0.0 V versus SHE
(22)
H+ + e =
1
O 2 + H2 = H2 O
2
Half cell II
Half cell I
Fig. 3
Oxidation
Reduction
(25)
(24)
Cell voltage Uc
[V]
(23)
The cell voltage under standard conditions is the potential difference between the right and left half cell: Uco =
o
o
o
EH
EO
+ /H = 1.229 V. Since Uc
2 /H2 O
2
is positive, the Gibbs free energy of
the reaction is negative according to
Eq. (16)
Go298
= 2.96485
Uth =
(27)
= 237.160 kJ mol1 .
The cell reaction (25) occurs spontaneously and the cell delivers electrical work
nF Uco , which is utilized in H2 /O2 fuel cells
(see also Chapter 8). Calculation of Go298
from thermodynamic data in Table 2 yields
1
o
o
HH
HOo 2,g )
Go298 = (HH
2 O
2,g
2
1 o
o
o
SH
SO
)
T (SH
2 O
2,g
2 2,g
(26)
Go298 = (285.25 0 0) 298
1
(70.12 130.74 205.25)
2
= 236.603 kJ mol1
The
reversible
heat
T S o =
1
48.647 kJ mol is negative. It is released
by the system and cannot be exploited as
electrical work.
If the two half cells are arranged opposite
with the hydrogen electrode on the right
and the oxygen electrode on the left, the
corresponding reactions are reversed and
the overall process refers to the electrolysis
of water with hydrogen and oxygen as the
products. In this case, the cell voltage
Uco is negative and Go298 is positive.
The reversible heat T S o is positive and
is taken up from the environment; the
cell cools down in the course of the
Energy Yield
Ge
Uc,o e
=
,
nF Uc
Uc
(28)
H e
Uth e
=
.
nF Uc
Uc
(29)
11
12
1.5.2
Uc e F
(J kg1 )
P MP e
(30)
1.6
Fig. 4
+
Flow
+
Multicathode cell
Waste
water
(b)
Anolyte
Multimesh
cathode
Mesh
electrodes
Ion-exchange membrane
Pump cell
Chemelec cell
Inlets
Distributor
Anolyte
Metal
powder
Electrolyte
outlet
Rod guidance
Rod cathodes
Separator
Swiss-role cell
Electrode
Beat-rod cell
Rotation
axes
Electrolyte
inlet
Catholyte Concentric
anodes
Fluidized bed
level
Electrolyte
level
ECO cell
+ + + +
Baffle
plates
Catholyte
Stator
Rotor
DSA mesh
anode
(a)
Flow
+
Current
connections
Flow
Anolyte
Fig. 4
(Continued).
(c)
Inlet
Packed-bed
cathode
Packed-bed
enViro cell
Overflow channel
Catholyte
outlet
Flow distributor
Large
particle outlet
Outlet
Small
particle inlet
+
Anode
Anolyte
Fluidized-bed cell
Catholyte
Cathode
feeder electrode
Ion-exchange membrane
Gas outlet
Rotation axes
Membrane
Anolyte
outlet
+
Rolling particles
(graphite)
Tankhouse
14
(31)
15
conversion per pass can be achieved. Continuous operation with recirculation of the
solution and cascade arrangement of cells
has therefore been used in practical applications.
Moving and circulating bed electrodes, also
denoted as a spouted bed, are described
in the literature [87]. As an example, the
rolling tube cell is shown in Fig. 4(c) [77, 78].
In principle, this cell resembles the wellknown plating process for piece goods,
using slowly rotating barrels. The rotating
drum is only partially lled with particles to achieve thorough agitation. This
cell has proved to be especially useful for
silver and gold recovery. Further developments of moving bed electrodes and their
application can be found elsewhere [88].
In Fig. 5, the specic electrode area
ae (cm1 ) of various electrochemical cells
is shown as a function of a characteristic
length (interelectrode distance or particle diameter). It can be seen that the
electrochemical cells differ signicantly
in their specic area and characteristic
length by orders of magnitude. At a current
I
= j .
Vr
(32)
100
100
Fluidizedbed cell
10
Current concentration j *
[A cm3]
Plate-andframe cell
Specific electrode area ae
[cm1]
16
Packed-bed cell
Swiss-roll cell
1.0
Plate-andframe cell
0.10
Swiss-roll cell
Packed-bed cell
0.01
Chloralkali cell
(mercury process)
0.1
00.1
0.1
Chloralkali cell
(mercury process)
Fluidizedbed cell
1
Characteristic length
[cm]
10
0.001
00.1
0.1
Characteristic length
[cm]
Fig. 5
10
Fig. 7
Cell room with capillary gap electrolyzers [28] (photo: BASF AG).
enters via a central channel and is radially distributed between the circular disk
electrodes. In the bipolar conguration,
the disks are isolated from each other and
connected at the end plates to the voltage supply. To avoid reconversion of the
product at the counterelectrodes in the
undivided cells, the disk geometry and
ow velocity must be carefully tuned to
ensure that the diffusion layer thickness,
17
18
Electrodes
Schematic
distribution
of current
lines
Gas bubbles
Diaphragm
Fig. 9
Cathode with
catalyst layer
Metal mesh
Bipolar
plate
Nafion 120
Anodic current
collector
Cathodic current
collector
9. F. Hine, Electrode Processes and Electrochemical Engineering, Plenum Press, New York,
1985.
10. T. Z. Fahidy, Principles of Electrochemical
Reactor Analysis, Elsevier, Amsterdam, 1985.
11. I. Rouzar, K. Micka, A. Kimla, Electrochemical Engineering, Elsevier, Amsterdam, 1986,
Vol. 1 and 2.
12. E. Heitz, G. Kreysa, Principles of Electrochemical Engineering, VCH, Weinheim, 1986.
13. F. Beck, H. Goldacker, G. Kreysa et al., Electrochemistry in Ullmanns Encyclopedia of
Industrial Chemistry, VCH, Weinheim, 1987,
pp. 183254, Vol. A9.
14. K. Scott, Electrochemical Reaction Engineering,
Academic Press, London, 1991.
15. D. Pletcher, F. C. Walsh, Industrial Electrochemistry, Chapman and Hall, London, 1993.
16. F. Goodridge, K. Scott, Electrochemical Process Engineering, Plenum Press, New York,
1995.
17. H. Wendt, G. Kreysa, Electrochemical Engineering, Springer, Berlin, 1999.
18. V. M. Schmidt, Elektrochemische Verfahrenstechnik, Wiley-VCH, Weinheim, 2003.
19. J. O. M. Bockris, Electrochemistry of Cleaner
Environments, Plenum, New York, 1972.
20. G. Kreysa, Reactor design for electrochemical water treatment in Process Technologies
for Water Treatment (Ed.: S. Stucki), Plenum
Publishing Corporation, New York, 1988.
21. D. Pletcher, N. L. Weinberg, The Green
Potential of Electrochemistry, Chemical
19
20
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42. J. O. M. Bockris, A. K. N. Reddy, M. GamboaAldeco, Modern Electrochemistry 2A Fundamentals of Electronics, Kluwer, New York,
1998.
43. M. Paunovic, Fundamentals in Electrochemical Deposition, Wiley, New York, 1998.
44. W. J. Lorenz, W. Plieth, Electrochemical Nanotechnology, Wiley-VCH, Weinheim, 1998.
45. J. Grimshaw, Electrochemical Reactions and
Mechanisms in Organic Chemistry, Elsevier,
Amsterdam, 2000.
46. R. Memming, Semiconductor Electrochemistry, Wiley-VCH, Weinheim, 2000.
47. J. Bard, L. R. Faulkner, Electrochemical Methods, 2nd ed., Wiley, New York, 2001.
48. J. W. Schultze, G. Staikov, Scaling Down in
Electrochemistry: Electrochemical Micro- and
Nanosystem Technology, Elsevier, Amsterdam, 2001.
49. H. Kaesche, Corrosion of Metals, Physicochemical Principles and Current Problems, Springer,
Berlin, 2003.
50. H. H. Girault, Analytical and Physical Electrochemistry, EPFI Press, Marcel Dekker, 2004.
51. International Union of Pure and Applied
Chemistry: Groen, Einheiten und Symbole
in der Physikalischen Chemie (Ed.: K.-H.
Homann), VCH, Weinheim, 1996.
52. G. Gritzner, G. Kreysa, Nomenclature, symbols and denitions in electrochemical engineering, J. Electroanal. Chem. 1993, 360,
351.
53. Handbook of Chemistry and Physics, Robert
and Weast, 1977.
54. I. Barin, O. Knacke, O. Kubaschewski, Thermodynamical Properties of Inorganic Substances, Springer, Berlin, 1973.
55. R. E. W. Janson, R. J. Marshall, Chem. Eng.
1976, 315, 769.
56. R. E. W. Janson, N. R. Tomov, Chem. Eng.
1977, 327, 867.
57. C. L. Lopez-Cacicedo, Inst. Chem. Eng. Symp.
Ser. 1975, 42, 29.
58. D. R. Gabe, J. Appl. Electrochem. 1974, 4, 91.
59. F. S. Holland, Chem. Ind. (London), 1978, 13,
453.
60. L. J. Ricci, Chem. Eng. 1975, 82(27), 29.
61. R. Kammel, H.-W. Lieber, Galvanotechnik
1977, 68, 710.
62. W. Gotzelmann, Galvanotechnik 1977, 68,
789.
63. D. Bruhn, W. Dietz, K.-J. Muller, C. Reynvaan, EPA 86109265.8, 1986.
21
23
Electrocatalysis
A. Katsaounis, S. Brosda, and C.G. Vayenas
University of Patras, Patras, Greece
2.1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
25
2.2
Adsorption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
28
2.3
2.3.1
2.3.2
2.3.3
2.3.4
2.3.5
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
34
34
36
38
43
47
2.4
2.4.1
2.4.2
2.4.3
2.4.4
2.4.5
Theory of Electrocatalysis . . . . . . . . . . . . . . .
Overpotentials and the ButlerVolmer Equation
Volcano Plots . . . . . . . . . . . . . . . . . . . . . . .
Transition State Approaches . . . . . . . . . . . . .
Quantum Mechanicsbased Models . . . . . . . .
Promotional Effects . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
48
48
51
53
54
57
2.5
58
2.6
60
2.7
63
2.8
66
2.9
69
2.10
76
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
24
2 Electrocatalysis
2.11
Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
80
80
25
2.1
Introduction
The term electrocatalysis was rst introduced by Grubb in 1963 [1] in connection
with the anodic and cathodic chargetransfer reactions in fuel cells. However,
the rst systematic experimental investigations of various electrocatalysts had
been carried out in the 1920s by Bowden and Rideal [2] and the concept and
rst interpretation of electrocatalysis had
been introduced in the 1930s by Horiuti
and Polanyi [3]. Their theory and its extensions and improvements have been
analyzed lucidly by Bockris [46]. Recent
reviews of progress in electrocatalysis can
be found in more general [68] or specialized books [911].
An electrocatalyst is an electrode that can
accelerate the approach to equilibrium of
an electrochemical process, which involves
a net charge transfer, without itself being
consumed in the process.
This denition is in analogy to that of a
catalyst which is a substance (quite often a
solid) which can accelerate, without being
consumed, the approach to equilibrium of
a chemical process in a gaseous or liquid
mixture.
In heterogeneous catalysis, the specic activity of a catalyst is usually expressed as turnover frequency (TOF) (s1 )
2 Electrocatalysis
26
1.4
1.2
II
2.1 Introduction
The effect of the
variation of the H on the
potential energydistance
diagram. (Reprinted with
permission from Ref. 4,
Copyright 1994 by Elsevier).
Fig. 2
PE
A
B
HAox
H
HBox
H ox = bHAds
XAB
Effect of catalyst
potential on the oxygen
desorption activation energy,
Ed , calculated from the Redhead
analysis of O2 TPD spectra for
Pt, Ag, and Au electrodes
deposited on YSZ. (Reprinted
with permission from Ref. 24,
Copyright 1997 by Springer).
Fig. 3
[eV]
3
Ed
[eV]
2.5
Au
2
Pt
1.5
Ag
0
0.2
0.4
0.6 0.8
eU*WR
[eV]
3.5
80
70
60
2.5
50
2
40
Na
K STD
K
Cs STD
1.5
[eV]
30
0
1.2
Ed
[kcd mol1]
Ed
[eV]
27
28
2 Electrocatalysis
2.2
Adsorption
(1)
(2)
2H
2M + H2
+ 2e + 2M
2M
H
(3)
(5)
2.2 Adsorption
Electrochemical
Langmuir isotherm. Plots of
fractional coverage versus
(a) concentration and
(b) potential or log c, for three
values of the equilibrium
constant K = 1, 5, and 25.
(Reprinted with permission
from Ref. 29, Copyright 1993 by
Wiley-VCH).
Fig. 5
1
K
25
Fractional coverge
[]
0.8
5
0.6
0.4
0.2
0.4
0.8
(a)
0
1
0.8
Fractional coverge
[]
1.2
1.6
C
[mM]
U vs arbitary reference
[mV]
60
120
180
240
25
300
0.6
0.4
0.2
0
3
(b)
log C
[mM]
29
2 Electrocatalysis
G = Go + r
(6)
(7)
1
8
Fractional coverage
[]
30
16
r
f =_
RT
0.1
0.01
0.001
200
100
100
f
[V]
200
300
2.2 Adsorption
a signicant role in the operation of COpoisoned low-temperature polymer electrolyte membrane (PEM) fuel cells [32].
In aqueous electrochemistry, two more
factors are important to consider when
studying adsorption or using adsorption
isotherms. The rst is the presence of
H2 O.
The state-of-the-art knowledge of water
adsorption in ultrahigh vacuum (UHV),
up to the late 1980s has been reviewed in
detail by Thiel and Madey [33] and in a
very lucid and concise form by Leiva [34].
Water bonding occurs with charge transfer
to the surface, lowering the work function
of the metal signicantly, while the internal molecular structure and vibrational
properties of water are only slightly perturbed. The adsorption energy of water
varies between 0.4 and 0.7 eV. Thus, water is a weakly chemisorbed species when
compared to other adsorbates such as CO.
The low adsorption energy implies that
adsorbed water molecules tend to make
hydrogen bonds with each other, even at
low coverages. Thus, it is experimentally
very difcult to study isolated adsorbed
molecules. Water clustering is practically
unavoidable. The water structure of the
rst two monolayers, established from
diffraction experiments over hexagonal
surfaces, is shown in Fig. 7. This structure
consists of an icelike hydrogen-bonded
network in which the rst layer is bound to
First
layer
Second
layer
31
2 Electrocatalysis
Fig. 8 Oxygen distribution functions in
the direction parallel to a Ag(111)
electrode as obtained from X-ray
diffraction experiments [35]. (a,b)
Potentials of 0.23 and +0.52 V with
respect to the potential of zero charge
respectively. Z = 0 corresponds to the
center of the rst Ag plane. Density is
referred to the bulk density of water.
(Reprinted with permission from
Ref. 35, Copyright 1995 by Elsevier).
10
12
10
12
0
(b)
Z
[]
(a)
32
6
Z
[]
2.2 Adsorption
1
0.8
Relative intensity
[a.u.]
1648 cm1
Strongly
H - bonded
0.6
Weakly H - bonded
1612 cm1
Icelike
1642 cm1
0.4
0.2
0
pzc
0.2
0.4
0.6
0.8
1.2
1.4
U vs RHE
[V]
Potential dependence
of the water bending infrared
band center frequency as
function of applied potential for
water adsorbed at a Pt(111)
electrode. (Reprinted with
permission from Ref. 36,
Copyright 1997 by the American
Chemical Society).
Fig. 10
1640
1630
1620
1610
1600
dHOH
1590
kb T
i
(8)
he has obtained Fig. 11(a), which shows
the very signicant deviation of the ratio
c/cbulk from unity when signicant potential differences exist at the metalsolution
interface. This implies that the surface
concentration values, to be used in the
electrochemical Langmuir or Frumkin
2 (r) =
0.2
0.4
0.6 0.8
Potential
[V]
1.2
33
2 Electrocatalysis
(a) Effect of distance from
electrode on ion concentration. Excess
concentration of attracted and repelled
ions with respect to bulk concentration
in the vicinity of an electrode surface
calculated from Gouy-Chapman
model [34]. Curves correspond to 0.025,
0.050, and 0.100 V potential differences.
(b) Outer Helmholz plane (OHP) and
inner Helmholz plane (IHP). Anions
adsorbed on the surface have lost part
of their solvation layer and are contact
adsorbed (IHP); cations remain fully
solvated and are water-separated
adsorbed (OHP). (Reprinted with
permission from Ref. 34, Copyright 2003
by John Wiley & Sons Ltd).
Fig. 11
14
12
10
c/cbulk
34
Attracted
ions
8
6
4
Repelled
ions
(a)
Distance
OHP
IHP
(b)
2.3
Iac
12
Il
10
Current density
[mA cm2]
a
8
Infinite
Activation
resistance control
b
Mixed
control
Masstransport
control
4
2
0
Potential
[V]
35
2 Electrocatalysis
V RHE
1.6
2OH
Working potential of
oxygen electrode
Pt 1
_
+ H2O + 2e
2 O2
Total overpotential at
the oxygen electrode
1.23
Electrode of potential
36
Activation
overpotential of
oxygen
(electrode)
Cell voltage UC
Total overpotential at the
hydrogen electrode
0
Working potential
of the H2 electrode
Pt 1
_ H2 + OH)
2(H2O+e
2
Working current density
Current density
Fig. 13
2.3.2
dened from
WC = UWC UoWC
(9)
(10)
l=0
l=1
U WC
hohmic, wc = IRwc
hWC
hW + hC
U WC
(a)
l=0
l=0
l=1
UWR
hohmic, WR
hW
U WR
(b)
vanishes within less than 1 s and the remaining part of the overpotential which
vanishes much slower is W + C .
It is worth emphasizing that although
overpotentials are usually associated with
electrodeelectrolyte interfaces, in reality
they refer to, and are measured as,
deviations of the potential of the electrodes
37
38
2 Electrocatalysis
(11)
(12)
Cyclic Voltammetry
Cyclic voltammetry is still the most commonly used technique for the investigation of electrocatalysis. It involves the
repeated imposition of a triangular potential function, against time, usually created
by a function generator and applied via
a potentiostat between the working and
reference electrodes, while recording the
39
R1
100
C
5
42
C
8
39
C
9
36
10
I
[A]
eg
ion
0.1
0.01
R2
p C2H4 = 0.40 102 bar
Ta
fe
lr
0
C
3
296 32 60
C
C 4
C
200
400
(a)
600 800
UWR
[mV]
1000 1200
500
300
200
100
I
[A]
50
30
20
10
T = 436 C
T = 398 C
T = 368 C
5
3
2
1
(b)
100
200
300
400
UWR
[mV]
39
2 Electrocatalysis
Hydrogen
Double - layer
Oxygen region
0.4
O2
PtH - oxidation
Current density j
[mA cm2]
0.2
mA cm2
0.5
1.5 V RHE
1.0
Reversible
0.2
O2 - potential
PtH formation
Pt O - reduction
0.4
H2
0.6
(a)
Potential U
0.8
0.6
0.4
a
Current density j
[mA cm2]
40
0.2
0.5
1.0
1.5
V RHE
0
0.2
0.4
0.6
0.8
1.0
Potential U
(b)
(13)
followed by
Pt OH + OH Pt O + H2 O + e
(14)
above 0.8 V. Oxygen evolution, which is
thermodynamically possible above 1.23 V,
starts usually above 1.6 V, that is, there is
a signicant activation overpotential for
O2 evolution, and thus also for oxide
reduction, as can be seen in Fig. 16, even
when using Pt electrodes [8].
In the cathodic region (I < 0), after the
oxide reduction peak
Pt O + H2 O + 2e Pt + 2OH
(15)
one obtains H chemisorption via
Pt + H2 O + e Pt H + OH (16)
close to the thermodynamic potential
(RHE = 0) of the H2 /H2 O couple, followed by H2 evolution. At the beginning of
the anodic scan (I > 0), one observes three
PtH oxidation peaks, assigned to the reverse of reaction (16) and corresponding
to three different H adsorption sites.
These hydrogen reduction/oxidation
peaks near the low-potential end of the
CV are used routinely for computing the
41
2 Electrocatalysis
true surface area of polycrystalline electrodes by comparing their area with that
obtained with Pt single-crystal electrodes.
By comparing Figure 16(a) and (b), one
observes that the electrolyte has a rather
minor effect on the CV of Pt, provided it
does not introduce a redox couple. On the
other hand, the CV of Au (Figure 16(b)) is
substantially different: The H adsorption
peaks are very severely suppressed, the
double-layer region is very signicantly
1 M H2SO4
1.5
current density j
[mA cm2]
42
1 M KOH
3
CH3OH
1.0
0.5
30
CH2O
60
20
40
10
20
CH3OH
CH2O
0.2
4
HCOOH
0.1
HCOO
0
0.1
0.6
1.2
0.6
1.2
Fig. 17
Fig. 18
Current
0.5 mA
+
3 mA
+
600 C
750 C
30 mA
+
900 C
0.4
(a)
100 mA
0
0.2
Potential
[V]
0.2
40
30
20
I
[mA]
450 C
10
0
0s
10
20
30
1
(b)
20 s
400 s
600 s
60 s
40 s
160 s
100 s
260 s
0.5
0.8
UWR
[V]
AC-impedance Spectroscopy
43
2 Electrocatalysis
ZRe
R0
R1
Zlm
C1
R0 + R1
R0 = 4 k, R1= 2 k,
R2 = 1 k,
R0
C1 = 104 F
C1
C2
R1 R2
C2 = 107 F
R0
0
R1C1 = 1/2pfm
C2 = 105 F
C2 = 106 F
R0 + R1
R0
(b)
Zlm
[K]
log f
(a)
Zlm
44
ZRe
0
(c)
ZRe
[K]
Bode (a) and corresponding Nyquist plot (b) of the circuit, shown in inset, which is
frequently used to model a metal/solid electrolyte interface. Effect (c) of capacitance C2 on the
Nyquist plot at xed R0 , R1 , and R2 . (Reprinted with permission from Ref. 23, Copyright 2001
by Kluwer/Plenum Publisher).
Fig. 19
O2
O(a)
+ 2e
(17)
Cd,1
Cd,2
(18)
45
2 Electrocatalysis
1V
1.4V
2V
2
zim
[k]
2V
zim
[k]
C1
0.6 V
C2
ZRe
o.c = 0.13 V
C1
0.6 V
2 V
[k]
2
C1
(a)
C1
0
0
C2
0.1 Hz
1 Hz
0.1 Hz
10
12
14
ZRe
[k]
103
400
Pt/YSZ
Cd,2
102
300
Cd
Cd,i
[mF cm2]
10
10
[mF cm ]
Cd,1
200
100
102
2
UWR
[V]
(b)
0
0.6
0.4
0.2
0.2
0.4
0.6
h
[V]
(c)
300
Ni/YSZ
Au/YSZ
Cd
[mF cm2]
200
Cd
[mF cm2]
46
100
100
(d)
0.6
0.4
0.2
h
[V]
10
0.6
0.2
(e)
0.4
0.2
0.2
h
[V]
(a) Complex impedance spectra (Nyquist plots) of the CH4 , O2 , and Pd|YSZ system
at different Pd catalyst potentials. Open-circuit potential UoWR = 0.13 V. (b) Dependence on
catalyst potential of Cd,1 and Cd,2 . (Reprinted with permission from Ref. 23, Copyright 2003 by
Kluwer/Plenum Publisher) (c) Double-layer capacitance as a function of overpotential of the
systems Pt/YSZ, (d) Ni/YSZ, and (e) Au/YSZ, the latter before () and after () prolonged
anodic overpotential application [42, 43]. (Reprinted with permission from Ref. 23, Copyright
2003 by Kluwer/Plenum Publisher).
Fig. 20
2.3.5
60
0.5
( j jdl)dU
0.05
30
j
[mA cm2]
jdl
0 0
0.2
j/ V
30
60
(a)
(100)
j
[mA cm2]
(100)
0
0.25
(110)
0
(b)
0.5
U vs
[V]
47
2 Electrocatalysis
Cyclic voltammogram for CO
stripping at sputter-cleaned electrodes
in 0.5 M H2 SO4 , dashed lines after CO
removal. (a) Pt and (b) Ru. Immersion
and adsorption at 250 mV, sweep rate:
20 mV s1 . (Reprinted with permission
from Refs 45, 46, Copyright 1994 by the
American Chemical Society).
Fig. 22
j
[mA cm2]
Pt/CO
(a)
50
Ru/CO
j
[mA cm2]
48
0
(b)
0.5
U
[V]
2.4
Theory of Electrocatalysis
2.4.1
(19)
(21)
(22)
(23)
A + ne B
(24)
a = (n )/ ;
c = / + ; a + c = n/ (25)
In extension of the denition of reaction
order in heterogeneous catalysis, the order
of an electrocatalytic reaction, ai , with
respect to a reactant, j , is dened as
log i0
log
i
=
ai =
log cj
log cj j =k
j =k,U
(26)
which implies that the comparison is made
at constant potential U , or overpotential
, so that ai expresses the variation
of the exchange current with varying
concentration cj .
Therefore, from Eq. (26) it follows that
the exchange current i0 , which is the
measure of the activity of an electrocatalyst,
is not a simple temperature-dependent
constant, but also depends on reactant
concentration.
Thus, in solid-state electrochemistry,
the exchange current I0 is known to be
strongly dependent both on temperature
and on gaseous composition. It increases
with temperature with an activation energy
which is typically 3545 kcal mol1 for
Pt and 2025 kcal mol1 for Ag lms
deposited on stabilized zirconia [23].
The I0 dependence on gaseous composition is usually complex. It has therefore
49
50
2 Electrocatalysis
1/2
1/2
KO pO2 /(1 + KO pO2 )
(27)
(28)
or, equivalently,
1/4
1/2
(29)
(30)
(31)
where r0 is the regular, that is, opencircuit, catalytic rate. When using Eq. (31)
to estimate the order of magnitude of
the enhancement factor expected for
a given catalytic reaction, one must use
the I0 value measured in the presence of
the reacting gas mixture. The fact that I0
increases exponentially with temperature
in conjunction with the fact that is
inversely proportional to I0 explains why
EPOC is limited usually to temperatures
below 600 C [23].
As already noted, the performance of a
galvanic cell (for electricity production) or
of an electrolytic cell (for the production of
chemicals) is dictated by the magnitude of
the overpotentials both at the anode and at
the cathode, a and c , by the ohmic drop
between the anode and the cathode and
by the diffusional overpotentials, d,a and
d,c , at the anode and cathode respectively,
resulting from slow mass transfer at the
anode and cathode. The latter are described
by
RT
ln(1 i/i,a );
F
RT
ln(1 i/i,c )
=
F
d,a =
d,c
(32)
Volcano Plots
(I0 /F )2 = kf kr B (1 B )
A+ + e B(a)
where the adsorbed intermediate
is in equilibrium with gaseous B,
expressing the rate of the forward
reverse charge-transfer reactions, rF
(33)
B(a)
and
and
and
Fig. 23
therefore
I0 = (kf kr )1/2 F [B (1 B )]1/2
(37)
One thus observes that the exchange current I0 which expresses the electrocatalytic
activity is maximized when B 0.5. If
one uses a Langmuir isotherm to relate B
with pB [56], then Eq. (37) gives
1/2 1/2
I0 = (kf kr )1/2 F
KB pB
(1 + KB pB )
(38)
Electrocatalytic activity, In j
(36)
qB 1
qB 0
(2)
(1)
Weakly adsorbed
Strongly adsorbed
0
Adsorption energy, G B
51
2 Electrocatalysis
Pt
Rh
4
log jo
[A cm2]
Fig. 24
1
Re
lr
Au
Cu Ni
Co
p-block and
Noble
d-electron
Fe
UPD H W
rich
Sn Bi
Mo
Ti
Ag
Pb Zn
Fe- group
Ta
Nb
Ga and noble
Cd
Form
d-electron
In
TI
hydride
rich
A
B
C phase D
5
6
7
8
9
10
11
125.4
209
292.6
376.2
GB = RT ln KB
(41)
MH bond strength
[kJ mol1]
0.1
Fig. 25
RuO2
0.2
Overpotential
[V]
52
0.3
PtO2 IrO2
Co3o4
MnO2
0.4
NiOx
0.5
Fe3O4
0.6
PbO2
0.7
100
200
300
400
Enthalpy change
lower-to-higher oxide transition
[KJ mol1]
GB
(42)
2RT
2O2 2O(a) + 2e ;
2O(a) O2 (g)
(39)
I0 = (kf kr )1/2 F
KO pO2
1/2
(1 + KO pO2 )
(40)
1/2
GB
(43)
2RT
2.4.3
(44)
can be expressed as
rred = cA+ ce kred exp
=o
Gred (M )
RT
(45)
where cA+ is the concentration of H+ at the
OHP, ce is the concentration of electrons in
the metal electrode, kred is a constant with
=o
dimensions of frequency, and Gred is the
standard free energy of the transition state
which is a function of the Galvani potential
inside the metal, M . This leads [5, 6, 66] to
Scheme of diabatic
(nonadiabatic) potential energy surfaces
U(R) as functions of coordinate R for
the process Ox + e Red at different
overpotentials . The intersection of the
curves corresponding to the oxidized
Uox and reduced Ured states are marked
with dots. (Reprinted with permission
from Ref. 34, Copyright 2003 by John
Wiley & Sons Ltd).
Fig. 26
h >0
h =0
U [R]
h <0
Uox
Ured
R
53
54
2 Electrocatalysis
= H
t
(47)
e0v2
OHP
jc = ()f ()
( + o )2
exp
d
4kb T
ja = ()(1 f ())
( + o )2
d
exp
4kb T
Fermi level
B0v1
Barrier
Energy, E f
Metal
electrode
Electron
finds less
acceptor
states
Fig. 27
Exponentially
distributed
electron level
+
of H3O ions
corresponding to
vib.-rot.states
+
of H H2O bond
(49)
E0 of H3O+ions
e0v2
(48)
Electronic
energy
Electronic
energy
Wox
Wox
Fermi - level
eoh
l Fermi- level
Wred
Wred
Fig. 28
55
2 Electrocatalysis
Energy
More negative
electrode potential
2p*
ef
d band
5s
sp band
Density of states
Fig. 29
0
DFT-GGA chemisorption energy
[eV]
56
Au(111)
Metals
Ag(111)
Overlayers
0.5
Cu(111)
Pd/Ru(0001) Cu/Pt(111)
Pd(111)
Ni(111)
1.5
Ni/Ru(0001)
Ru(0001)
4.5
3.5
3
ed
[eV]
2.5
Fig. 30
Pt(111)
1.5
2.4.5
Promotional Effects
Fig. 31
1.2
CO tolerant
0.8
4f72
4d52
0.4
Pt95Fe05
Pt40Fe60
4f52
4d32
Pt56Fe44
Pt85Fe15
0
1.2
0.8
0.4
Pt76Ni24
Pt82Co18
Pt57Ru43
Pt45Mo55
0
Not CO tolerant
0
0.4
BE shifts from pure Pt
[eV]
Pt55Nb45 Pt65Pd35
Pt76Ti24
Pt70Ge30
0.8
1.2
0
0.4
0.8
1.2
Pt51Au49
Pt66ln34
Pt76sb24
Pt65Bi35
57
2 Electrocatalysis
Effect of chemical shift of Pt
4d5/2 binding energy in Pt skin layer,
prepared on a bare Pt50 Fe50 alloy
surface by sputtering deposition, onto
(top) the kinetically controlled H2
oxidation current (Jk ) and (bottom) the
steady-state CO coverage (CO ) at room
temperature in 0.1 M HClO4 solution
saturated with 100 ppm CO/H2
balance [96, 98]. (Reprinted with
permission from Ref. 96, Copyright 2003
by John Wiley & Sons Ltd).
Fig. 32
qCO
jk
[mA]
3
2
1
0
0.8
0.6
0.4
0.2
0
0.1
0.2
0.3
0.4
Chemical shift
[eV]
S
0.02
i
[mA real cm2]
58
Se
2.5
Fig. 33
0.01
Hg
Ag Pt Bi
0
1.5
Te
2
Electronegativity
2.5
Pt + H2 O + e
Pt
Hupd + OH (pH > 7)
U vs RHE
[V]
0
0.16
0.7
0.31
0.6
0.4
0.3
0.001
0.98
0.01
0.1
I
[mA real cm2]
H2 + Pt
2Pt
H
(53)
(54)
H2 + Pt H + Pt H + e
Cyclic voltammogram of
platinum single-crystal surfaces in the
hydrogen adsorption region in 0.5 M
H2 SO4 under nitrogen purging;
|| = 50 mV s1 [117, 122]. (Reprinted
with permission from Ref. 122,
Copyright 1991 by EDP Sciences).
(55)
(52)
0.8
Pt(110)
Pt(100)
Current density j
PtH
0.9
Pt(111)
0.3
V RHE
Potential U
59
2 Electrocatalysis
Arrhenius plots of the exchange
current densities (i0 ) for the HER/HOR
on Pt(hkl) in acid solution. (Reprinted
with permission from Ref. 117,
Copyright 2003 by John Wiley & Sons
Ltd).
Fig. 36
2.8
3.2
3.4
Pt(110)
Pt(100)
Pt(111)
3.6
3.8
3.2
3.4
3.6
3.8
T 1
[103 K1]
Hupd
3
2.6
This electrocatalytic system is of great importance [123136] both for efcient fuel
cell operation [123, 125] and for efcient
electrolysis [123]. The progress in understanding oxygen electrocatalysis has been
reviewed by Yeager [129], Appleby [130],
Hopd
(a)
(b)
3
Hupd Hopd
Pt(110)
jo
[mA cm2]
0
1
2
0.1
1
Hupd
Pt(111)
Hopd
0.01
3 3.2 3.4 3.6
0.2
10
U vs RHE
[V]
j
[mA cm2]
log(jo)
[A cm2]
j
[mA cm2]
60
0.2
U vs RHE
[V]
Damjanovic [131], Kinoshita [132], Bockris [92], Adjic [134], MacDougall [123], and
Ross [124] among others for aqueous
systems and by McEvoy [28] and IversTiffee [125] for solid-state electrochemical
systems.
The oxygen reduction reaction (ORR)
at the cathode of an electrochemical cell
involves the transfer of four electrons from
the cathode to an O2 molecule followed by
removal of the reduced product from the
interface. Reduction of an O2 molecule
to water or to hydroxyl anions at an
electrode/aqueous electrolyte interface is
to be distinguished from reduction of O2
to 2O2 at the surface of an oxide or at
an electrode/solid electrolyte interface in
solid-state electrochemistry. In the latter
case, the ORR is generally different at the
surface of a noble metal like Pt from that
at the surface of a metallic oxide.
The ORR requires a weakening of the
oxygenoxygen bond of the O2 molecule
and therefore a strong interaction with
the surface of the catalytic electrode. A
noble metal like Pt has too large a work
function to be able to donate two electrons
to an adsorbed oxygen atom to create O2
ions at the surface; but it can transfer one
electron per oxygen atom, and this electron
is charge compensated by a proton from
the solution. If both oxygen atoms of the
O2 molecule are adsorbed on the surface,
a bidentate adsorption occurs [137]. Thus,
in acid solution
O2 + 2e + 2H+ 2OHads
(56)
(57)
(58)
(60)
(61)
(62)
(63)
where
H2 O2 + 2H+ + 2e 2H2 O,
U o = 1.77 V
(64)
(65)
61
62
2 Electrocatalysis
(66)
M(m1)+ M(m1)+ + O2
Mm+ (O2 )2 Mm+ (72)
(67)
O2 + H2 O + 2e
HO
2
+ OH ,
U = 0.076 V
(68)
with
HO
2
+ H2 O + 2e 3OH ,
U o = 0.88 V
(69)
(73)
followed by
2Mm+ + 2e 2M(m1)+
(74)
2.7
15
i 104
[Acm2]
2
1
0
1
2
UH
[V]
20 years in connection with the development of direct methanol fuel cells (DMFCs)
Table 1 shows the equilibrium oxidation potentials for H2 , CH3 OH, H2 CO,
HCOOH, and CO. One observes that the
open-circuit potentials of the corresponding fuel cells utilizing these molecules
as the fuel at the anode are close to
each other and thus show similar promise
from a thermodynamic point of view [142].
However, as shown in Fig. 22, the oxidation of CO exhibits an overpotential of
0.7 V on Pt and 0.6 V on Ru anodes in
acidic environments. The overpotential for
CO oxidation is signicantly smaller in
5
Regions almost A1 A2 A3
6
thermodynamically 5 7 8 9
4
reversible 3
Pt
Pt-8 Ru
Pd Pt-8 Ru
Pt-40 Ru
6
7
Pt-60
lr,Rh
Ru
Os
Pt/Ru alloys are
expressed in atom %
9
10
Au
11 Ag
1
d-band vacancies
63
2 Electrocatalysis
Atomic-resolution images of
oxygen islands on Pt(111). These
images illustrate the enhanced
precursor dissociation probability at
adsorbed oxygen atoms and the
resulting quasi-one-dimensional growth
for (a) T = 140 K (3 L dose, image
100 60 A 2 ); (b) T = 105 K with
chainlike islands corresponding to
strings of oxygen pairs (1 L dose, image
70 40 A 2 ); (c) model illustrating the
directional growth of Oad (black
spheres) chains for Pt(111); (i) an
oxygen precursor molecule (shaded)
approaches an atom island; (ii) on its
dissociation a chain segment evolves.
(Reprinted with permission from
Ref. 140, Copyright 1997 by Nature
Publishing Group).
Fig. 40
T = 140 K
(a)
T = 105 K
(b)
(c)
0.8
0.7
0.6
Current density
[mA cm2]
64
0.5
Pt
Pt4.5Ru4Ir0.5
Pt4Ru4Ir1
Pt/Ru
Pt/Ir
0.4
0.3
alkaline environments [117] but such environments are of no interest for polymer
electrolyte membrane fuel cell (PEMFC),
applications. The same trend is observed
when examining the oxidation of small
organic molecules and comparing the performance of Pt electrocatalysts in acidic
and alkaline environments, as shown earlier in Fig. 17. The same gure showed the
multipeak CVs obtained primarily in acidic
environments resulting from the oxidation
of passivating oxide and subsequent easier oxidation on the cleaned surface in the
reverse scan.
Since such high anodic overpotentials
are prohibitive for practical applications,
a very intense research effort has been
launched during the last 20 years for nding more efcient anodic electrocatalysts.
The source of the problem is that the
direct anodic six-electron oxidation
0.2
0.1
0
Fig. 41
400 600 800 1000 1200 1400 1600 1800
Potential vs RHE
[mV]
Tab. 1
Ho [kcal mol1 ] So [cal mol1 ] G [kcal mol1 ] n Uo [V]
H2 + O2 H2 O(l)
C + 1/2O2 CO
CO + 1/2O2 CO2
CH3 OH + 3/2O2 CO2 + 2H2 O
CH2 O + O2 CO2 + H2 O
HCOOH + 1/2O2 CO2 + H2 O
68.14
26.4
67.62
173.8
134.28
64.66
(76)
39.0
+21.4
20.7
23.5
32.1
+11.8
2
2
2
6
4
2
1.23
0.71
1.33
1.21
1.35
1.48
56.69
32.81
61.45
166.8
124.7
68.2
Current
2 mA
(m s1 = 60)
2 1013 A
Fig. 42
4 1011 A
(m s1 = 40)
0.5
1.0
Potential vs RHE
[V]
1.5
65
2 Electrocatalysis
0.1
0.06
10
0.04
2.8
i
[mA cm2]
0.08
i
[mA cm2]
66
Methanol electrooxidation
current densities as a function of
ruthenium coverage at 0.56 V () and at
0.38 V () on the Pt(111)/Ru electrode.
(Reprinted with permission from
Ref. 147, Copyright 1998 by the
American Chemical Society).
Fig. 43
0.02
5
0
0.1
0.15
0.2
Ru
[mL]
0.25
0.3
50
j
[mA cm2]
40
30
20
10
0
0
0.2
(a)
0.4 0.6
1 qH
0.8
30
j
[mA cm2]
25
20
15
10
5
0
0
0.2
0.4 0.6
1 qH
0.8
0.2
0.4 0.6
1 qH
0.8
(b)
j
[mA cm2]
1.5
0.5
0
(c)
(78)
67
2 Electrocatalysis
Polarization curves for the
oxidation of CO (A), HCOOH (B),
HCHO (C), and CH3 OH (D) at 40 C on
Pt with deposited Sn atoms. (Reprinted
with permission from Ref. 150,
Copyright 1976 by Elsevier Science).
Fig. 45
100
i
[mA real cm2]
68
10
B
D
0.5
Pt
qSn
(79)
MO M + 1/2O2
(80)
(81)
M( OH) M + 1/2O2 + H+ + e
(82)
The distinction between active and
nonactive behavior and the underlying
mechanistic explanation are supported by
several experimental observations, including measurement of the concentration
of reactive intermediates in the oxygen
evolution reaction, such as hydroxyl radicals produced by discharge of water. In
fact, hydroxyl radicals produced at nonactive electrodes can be intercepted by
using p-nitrosodimethylaniline as a selective scavenger [170].
In practice, however, most anodes will
exhibit a mixed behavior since both parallel reaction paths will participate in
organics oxidation and oxygen evolution
reactions. As a general rule, the closer
the reversible potential of the surface redox couple to the potential of oxygen
evolution, the higher the active character
of the anode. In Ref. 170, dimensionally stabletype anodes (DSA) based on
IrO2 (Ti/IrO2 ) and synthetic boron-doped
diamond (p-Si/BDD) electrodes are compared in the oxidation of organics in
acid media (1 M H2 SO4 or 1 M HC1O4 ).
Ti/IrO2 may serve as a typical example of
active anode (the redox potential of the
IrO3 /IrO2 couple, 1.35 V/RHE, is close
to the standard potential of O2 evolution,
1.23 V/RHE) and is thus suitable for electrosynthesis, while p-Si/BDD, a typical
nonactive BDD, is known to be inert
and to have weak adsorption properties.
Fig. 46
2.9
69
2 Electrocatalysis
Electrochemical
(electrocatalytic)
oxidation
rate re = l/nF
Catalytic oxidation
rate r
current-induced rate change
r > >l/nF
C2H4 + 6O2
C2H4 + 3O2
2CO2 + 2H2O
e
O2
O2
O2
Catalyst electrode
Solid electrolyte
Counterelectrode
O2 (g)
(a)
50
100
I=0
I = + 1 mA
I=0
800
600
40
20
0
r
[108 mol Os1]
60
30
400
20
200
ro + (I/2F) 104
10
0
200
0
0
(b)
40
80
TOF
[s1]
70
2FNG/I 1200
Time
[seconds]
Electrochemical reaction
2H+ (Nafion) + 2e+1-butene
Catalytic reaction
1-butene cis-2-butene
trans-2-butene
butene
Pt
e
H+
H+
H+
H+
UWR
e
(a)
AU
AU
1.4
25
30 Wt% Pd/C
rcis
rtrans
rbutane
i (mA cm2)
r
[106 mol s1]
20
15
0.8
0.6
i
[mA cm2]
1.2
10
0.4
5
0.2
0
(b)
0.1
0.2
0.3
0.4
Cell voltage
[V]
71
2 Electrocatalysis
100
80
cis-butene
trans-butene
butene
50
60
30
40
20
20
10
0.05
0.1
0.15
0.2
0.25
0
0.3
Cell voltage
[V]
(c)
Fig. 48
Faradaic efficiency
[]
40
% Selectivity
72
Continued.
(83)
where rcatalytic is the current- or potentialinduced change in catalytic rate, I is the applied current, and F is Faradays constant.
Second, the rate enhancement, , dened from
(84)
= r/ro
where r is the electropromoted catalytic
rate and ro is the unpromoted (open
circuit) catalytic rate.
A reaction exhibits electrochemical promotion when || > 1, whereas electrocatalysis is limited when || 1. A reaction is termed electrophobic when > 1
and electrophilic when < 1. In the former case, the rate increases with catalyst
potential, U , whereas in the latter case the
rate decreases with catalyst potential. values up to 3 105 [23, 210] and values up
to 150 [23] have been found for several systems. More recently, values between 300
and 1200 [211, 212] have been measured
for C2 H4 oxidation on Pt.
(86)
73
74
2 Electrocatalysis
O
Gas reactants
(e.g., O2, CO)
O
O
O +
++
O2
Gas reactants
(e.g., O2, CO)
O
+
O
+
Metal
++
++
O2 O2
O
+
Double layer
O
+
++
O2
Na
+ O
Na+
Na+
Na+
Metal
Na+
Na+
Double layer
Na+
I=0
40
r
[108 mol O/s]
80
TOF
[s1]
30
60
20
40
20
10
800
Strongly bonded
backspillover state gets
populated at a rate I/2F,
Catalytic rate is 99%due
to weakly bonded state
0
200
2FNG/I 1200
(a)
200
Catalyst potential U WR
[mV]
I = +1mA
I=0
100
2400
6600 78 000
Time [Seconds]
T
[K]
600
20
700
800
900
3900 s
Weakly bonded
highly reactive
O state
(dN/dT )
[101 mol/s1]
16
Strongly bonded O
backspillover state
2030 s
12
700 s
2FNG/I = 2500 s
340 s
8
125 s
4
(b)
0s
300
400
500
600
T
[C]
75
2 Electrocatalysis
40
2FNG/I = 1200 s
30
20
I
[mA]
10
0
0s
10
160 s
20
Strongly bonded O
backspillover state
30
0.7
0.4
20 s
60 s 40 s
100 s
Weakly bonded
Higly reactive
O state
0.2
0.5
400 s 260 s
600 s
0.1
0.8
UWR
[V]
(c)
Fig. 50
Continued.
425
250
2.10
102
10
Electrochemical
promotion
Electrocatalysis
0.1
0.8
1.2
Fig. 51
T
[C]
76
1.4 1.6
(1000/T )
[K1]
1.8
In recent years, and starting from electrochemical promotion studies [23], certain
rules have been extracted for the selection
of promoters for catalytic and electrocatalytic systems.
From Eq. (86), one can study the dependence of catalytic reactions on catalyst
work function. Four types of rate-work
function dependence have been identied experimentally, that is, electrophobic
(r/ > 0), electrophilic (r/ < 0),
volcano (where the rate exhibits a maximum with varying ) and inverted volcano
type, where the rate exhibits a minimum
with varying (Fig. 52).
80
0.4
UWR
[V]
0
1.6 1.2
0.4
T = 700 C
0.8 0.4
10
r/r0
r/r0
0.4
T = 450 C
pNO = 1.3 kPa
pC2H4 = 3.7 kPa
5
10
10
0.6
(a)
(b)
25 20 15 10 5
UWR
[V]
8
0.1
0.1
UWR
[V]
0.2
0.8 0.4
0.3
0.8
0.4
10
r/r0
r/r0
T = 560 C
po2 = 0.2 kPa
pCO = 11 kPa
1
2
(c)
T = 425C
po2 = 19 kPa
pCH3OH = 0.9 kPa
1
0
15 10 5
(d)
10
15
77
2 Electrocatalysis
UWR,
[V] [eV]
UWR,
[V] [eV]
0.5
0.5
1.0
1.0
1.0
0.5
lD = 0.15
10
r/r0
1
0.4
0.1
A
0.2
0.01
10
10
0
20
20
(a)
10
1.0
0.5
0.5
1.0
1.0
1.0
0.8
D
A
0.5
0.5
1.0
kD = 0.01, kA = 0.01
pD = 1, pA = 1
lD = 0.15, lA = 0.15
0.1
JD JA
r/r0
0.6
0.4
0.2
20
10
0.8
kD = 100,
pD = 1
lD = 0.15
10
UWR,
[V] [eV]
0.6
0.01
(b)
UWR,
[V] [eV]
1.0
0.1
0.2
D
0
20
10
lA = 0.15
0.6
0.4
100
JD JA
JD JA
0.6
1.0
kA = 0.01, pA = 1
0.8
kD = 0.01, pD = 1
kA = 100, p A = 1
lD = 0.15
lA = 0.15
0.5
kD = 100, pD = 1
100
0.8
r /r 0
1.0
r/r0
1.0
JD JA
78
0.4
kA = 100
pA = 1
lA = 0.15
0.2
1
0
20
(c)
10
10
20
0.01
Fig. 53
0
20
(d)
10
10
20
(88)
expression is:
r=
kR kA kD pA pD exp[(D + A )]
(89)
[1 + kD pD exp(D )
+ kA pA exp(A )]2
79
80
2 Electrocatalysis
2.11
Concluding Remarks
The urgent need to develop more efcient fuel cell anodes and cathodes has
brought the electrochemical, catalytic, and
surface science communities closer than
ever before and has made electrocatalysis a
rapidly growing eld both in experimental
new ndings and in theoretical understanding. It is very likely that the rapid
advances in catalyst and electrocatalyst
nanoparticle preparation and characterization [10], together with the utilization
of new powerful in situ techniques, such
as electrochemical nuclear magnetic resonance (NMR) [216] in conjunction with
the rapidly developing theoretical and computational methods [89] will soon become
valuable tools not only for a better fundamental understanding of electrocatalysis
but also for the design of more efcient
practical electrocatalysts.
References
1. W. T. Grubb, 17th Annual Power Sources
Conference, Atlantic City, 1963.
2. F. B. Bowden, E. Rideal, Proc. R. Soc. 1928,
A120, 5979.
3. (a) I. Horiuti, M. Polanyi, Acta Physicohimica U.R.S.S. 1935, 2, 505532; (b) Trans.
Faraday Soc. 1934, 30, 11641172.
4. J. O. M. Bockris, Z. S. Minevski, Electrochim. Acta 1994, 39, 14711479.
5. J. O. M. Bockris, A. K. N. Reddy, M. Gamboa-Aldeco, Modern Electrochemistry, 2nd
ed., Kluwer Academic/Plenum Publishers,
New York, 2000
6. J. O. M. Bockris, S. U. M. Khan, Surface
Electrochemistry, A Molecular Level Approach, Plenum Press, New York, 1993,
pp. 319349, 394405.
7. A. J. Bard, L. R. Faulkner, Electrochemical
Methods. Fundamentals and Applications,
John Wiley and Sons, New York, 2001.
8. C. H. Hamann, A. Hamnett, W. Vielstich,
Electrochemistry, 2nd ed., Wiley-VCH, Weinheim, 1998.
9. W. Vielstich, A. Lamm, H. A. Gasteiger,
(Eds.), Handbook of Fuel Cells: Fundamentals
Technology and Applications, Electrocatalysis,
John Wiley and Sons, Chichester, 2003, Vol.
2.
10. A. Wieckowski, E. Savinova, C. G. Vayenas,
(Eds.), Catalysis and Electrocatalysis at
Nanoparticles, Marcel Dekker, New York,
2003.
11. P. N. Ross, in Electrocatalysis (Eds.:
J. Lipkowski, P. N. Ross), John Wiley and
Sons-VCH, New York, 1998, pp. 4374.
12. R. Parsons, J. O. M. Bockris, Trans. Faraday
Soc. 1951, 47, 914.
13. B. E. Conway, J. O. M. Bockris, Can. J.
Chem. 1957, 35, 1124.
14. P. George, J. Grifth, in Enzymes (Eds.:
P. D. Boyer, H. Lardy, K. Myrback), Academic Press, New York, 1959, p. 347, Vol. 1.
81
82
2 Electrocatalysis
54.
55.
56.
57.
58.
59.
60.
61.
62.
63.
64.
65.
66.
67.
68.
69.
70.
71.
83
84
2 Electrocatalysis
130.
131.
132.
133.
134.
135.
136.
137.
138.
139.
140.
141.
142.
143.
144.
145.
146.
147.
148.
149.
150.
151.
152.
153.
154.
155.
156.
157.
158.
159.
160.
161.
162.
163.
164.
165.
166.
85
86
2 Electrocatalysis
205. B. Grzybowska-Swierkosz, J. Haber, Annual Reports on the Progress of Chemistry,
The Royal Society of Chemistry, Cambridge,
1994.
206. D. Tsiplakides, S. Balomenou, A. Katsaounis et al., Catal. Today 2005, 100,
133144.
207. C. G. Vayenas, S. Brosda, C. Pliangos, 40th
Anniversary issue, J. Catal. 2003, 216,
487504.
208. C. Sanchez, E. Leiva, in Handbook of Fuel
CellsFundamentals, Technology and Applications (Eds.: W. Vielstich, A. Lamm, H. A.
Gasteiger), John Wiley and Sons, New York,
2003, pp. 6568, Vol. 2.
209. E. Leiva, C. Sanchez, in Handbook of Fuel
CellsFundamentals, Technology and Applications (Eds.: W. Vielstich, A. Lamm, H. A.
Gasteiger), John Wiley and Sons, New York,
2003, pp. 145149, Vol. 2.
87
Electrochemical Deposition
Milan Paunovic
IBM Research Division Yorktown Heights, New York
3.1
3.1.1
3.1.2
3.1.3
3.1.3.1
3.1.3.1.1
3.1.3.1.2
3.1.3.2
3.1.3.3
3.1.3.4
3.1.4
3.1.4.1
3.1.4.2
3.1.5
3.1.5.1
3.1.5.2
Electrodeposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Metal/Metal-ion (M/Mz+ ) Interphase at Equilibrium . . . . . .
Electrode Potential Departs from Equilibrium: Overpotential
High-overpotential Approximation . . . . . . . . . . . . . . . . . .
Large Cathodic Current . . . . . . . . . . . . . . . . . . . . . . . . . .
Large Anodic Current . . . . . . . . . . . . . . . . . . . . . . . . . . .
Low-overpotential Approximation . . . . . . . . . . . . . . . . . . .
Inuence of Mass Transport on Electrode Kinetics . . . . . . .
Underpotential Deposition . . . . . . . . . . . . . . . . . . . . . . .
Atomistic Aspects of Electrodeposition . . . . . . . . . . . . . . .
Step-edge Ion-transfer Mechanism . . . . . . . . . . . . . . . . . .
Terrace Ion-transfer Mechanism . . . . . . . . . . . . . . . . . . .
Techniques for the Study on Electrode Processes . . . . . . . . .
Galvanostatic Transient Technique . . . . . . . . . . . . . . . . . .
Potentiostatic Transient Technique . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
91
91
92
93
95
95
95
97
97
98
99
99
99
101
101
102
3.2
3.2.1
3.2.2
3.2.2.1
3.2.2.2
3.2.3
3.2.3.1
3.2.3.2
3.2.3.3
3.2.3.4
3.2.4
3.2.4.1
3.2.4.2
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
103
103
103
104
105
106
106
106
107
107
108
108
108
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
88
3 Electrochemical Deposition
3.2.5
3.2.5.1
3.2.5.2
3.2.5.3
3.2.5.4
3.2.5.5
3.2.5.6
3.2.5.7
3.2.5.8
Effect of Additives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Chemisorption and Physisorption . . . . . . . . . . . . . . . . . . . . . . .
Adsorption Equilibrium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Effect of Additives on Kinetics and Mechanism of Electrodeposition .
Effect of Additives on Nucleation and Growth . . . . . . . . . . . . . . . .
Dependence of Types of Deposit on the Surface Coverage by Additive
Leveling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Brightening . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Consumption of Additives . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
109
109
110
110
110
111
111
111
111
3.3
3.3.1
3.3.2
3.3.2.1
3.3.2.2
3.3.2.3
3.3.3
3.3.3.1
3.3.3.2
3.3.3.3
3.3.3.4
3.3.3.5
3.3.4
3.3.4.1
3.3.4.2
3.3.4.3
3.3.4.4
Electroless Deposition . . . . . . . . . . . . . . . . . . . . . . .
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Electrochemical Model: the Mixed-potential Theory . . . .
WagnerTraud Diagram . . . . . . . . . . . . . . . . . . . . . .
Evans Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Mixed Potential, Emp . . . . . . . . . . . . . . . . . . . . . . . .
Kinetics and Mechanism . . . . . . . . . . . . . . . . . . . . . .
The Cathodic Partial Reaction. Kinetic Scheme . . . . . . .
Kinetics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
The Anodic Partial Reaction. Mechanism . . . . . . . . . . .
Parallel Reactions . . . . . . . . . . . . . . . . . . . . . . . . . .
Kinetics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Activation of Noncatalytic Surfaces . . . . . . . . . . . . . . .
Electrochemical Activation . . . . . . . . . . . . . . . . . . . .
Photochemical Activation . . . . . . . . . . . . . . . . . . . . .
Activation by Displacement Deposition . . . . . . . . . . . .
Activation by Thermal Decomposition of Metallic Oxides
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
112
112
113
114
115
115
117
117
117
117
118
118
119
119
119
119
119
3.4
3.4.1
3.4.2
3.4.2.1
120
120
121
3.4.3
Displacement Deposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Electrochemical Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Predictions of the Feasibility of the Reaction. Complexed Metal Ions
in Displacement Deposition . . . . . . . . . . . . . . . . . . . . . . . . . . .
Kinetics and Mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.5
3.5.1
3.5.2
3.5.2.1
3.5.2.2
3.5.2.3
3.5.2.4
3.5.3
123
123
124
124
124
125
126
128
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
122
123
3 Electrochemical Deposition
3.6
3.6.1
3.6.2
3.6.3
3.6.4
3.6.4.1
3.6.4.2
3.6.4.3
3.6.5
3.6.6
3.6.7
3.6.8
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
129
129
129
129
131
131
131
132
132
132
133
134
3.7
3.7.1
3.7.2
3.7.3
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
134
134
136
138
3.8
3.8.1
3.8.2
3.8.3
3.8.4
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
142
142
143
145
149
152
156
A1
156
A2
157
A3
Magnetic Anisotropy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
157
A4
Magnetostriction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
158
89
91
3.1
Electrodeposition
3.1.1
Introduction
(1)
(2)
92
3 Electrochemical Deposition
Metal/Metal-ion
Equilibrium
(M/Mz+ )
Interphase at
(5)
(6)
Metal/metal-ion
interphase at equilibrium: an
equal number of metal ions
Mz+ crossing in both directions
across the metal/solution
interphase.
Solution Fig. 1
Mz +
n = n
Mz +
Metal
(3)
3.1 Electrodeposition
RT
ln[Mz+ ]
zF
(7)
(8)
(9)
0.0592
log[Mz+ ]
z
(10)
(11a)
(11b)
(12)
93
94
3 Electrochemical Deposition
kB T
exp(G= /RT )
h
(13)
Gec = f (E)
(14)
=
=
=
(15)
where Gpin and Gpd are the potentialindependent and the potential-dependent
parts of the electrochemical activation energy, respectively. The potential-dependent
=
part of the activation energy Gpd can be
estimated (approximated) by the introduction of the transfer coefcient , which
was introduced in 1930 by Erdey-Gruz and
Volmer [5]. With this approximation, the
potential-dependent contribution for the
reduction reaction is zF where is
i = io exp(zF /RT )
(16)
(17)
(18)
i= i i
Figure 2 illustrates the metal/metal-ion
electrode in nonequilibrium state. From
Eqs (1618), we get the ButlerVolmer
3.1 Electrodeposition
Fig. 2 Metal/metal-ion electrode
deviates from the equilibrium potential:
two currents owing in opposite
directions.
Mz +
Mz +
Mz +
equation
h>0
i>0
Net anodic reaction
Difference of two
Solution
Metal
High-overpotential Approximation
3.1.3.1.1
Mz +
Mz +
h<0
i<0
Net cathodic reaction
Mz +
= a b log |i|
(20)
95
3 Electrochemical Deposition
i /i
i
1.0
0.8
0.6
ic
400
300
200
0.4
0.2
100
Total current
100
200
300
400
[mV]
0.2
0.4
io
ia
i
0.6
0.8
1.0
Fig. 3
slope b as
240
180
Cathodic polarization
II
[mV]
96
120
60
Anodic polarization
0
0.4
0.8
1.6
Log i
[i in mA cm 2]
2.4
3.2
Fig. 4
3.1 Electrodeposition
Low-overpotential Approximation
Under these conditions, the exponentials
in the ButlerVolmer equation (19) can
be approximated using a power series,
and taking only the rst-order terms in
, we get
3.1.3.2
i = io (zF /RT )
(23)
i = io f
(24)
or
where f = F /RT .
Thus, for small values of , less than
about 0.01 V, when the electrode potential
is near the reversible potential, the current
varies linearly with the overpotential.
3.1.3.3 Inuence of Mass Transport on
Electrode Kinetics
The currentpotential relationships dened by Eqs (16, 17, and 19) are valid
Density
[i ]
ButlerVolmer
exponential
relationship
(the purely activation
controlled current)
iL
iL
Current
3
Mixed control
(activation and
mass transport)
Linear
region
2
1
Exponential region
Potential
97
98
3 Electrochemical Deposition
Concentration
cox(x )
Nearest diffusion
layer model (linear profile)
c bOx
Actual concentration
(activity profile)
x=0
Diffusion
layer
Underpotential Deposition
We have seen earlier that metal M will be
deposited on the cathode from the solution
3.1.3.4
(26)
(27)
3.1 Electrodeposition
Initial state
Mz +
H2O
Solution
Kink site
final state
99
100
3 Electrochemical Deposition
Fig. 8 Step-edge ion-transfer
mechanism.
Mz+
Solution
H2O
Kink site
Step edge
Solution
Mz +
deposit are the atomic structure of the surface, the population density of the surface
defects (dislocations, growth edges, kinks,
and vacancies) and adsorbed species, and
the electronic structure of surfaces (Ref. 1;
Chapter 4 in Volume 2).
The initial theoretical treatment of these
mechanisms of deposition was given by
Lorenz [6, 810]. The initial experimental
studies on surface diffusion were published by Mehl and Bockris [7, 11]. Conway
and Bockris [12, 13] calculated activation
energies for the ion-transfer process of
various surface sites. The simulation of
3.1 Electrodeposition
The study of the charge-transfer processes, free from the effects of mass
transport, is possible by the use of transient techniques. In the simplest transient
techniques, the interface at equilibrium
is abruptly changed from an equilibrium state to a steady state characterized
by a new potential difference . This
abrupt change is imposed by applying either (1) the constant current, galvanostatic
transient technique or (2) the constant potential, potentiostatic transient technique.
The study of transient processes utilizes
the three-electrode cell design (Fig. 10).
Galvanostatic Transient Technique
In the galvanostatic technique, the current
between the test electrode and the auxiliary
3.1.5.1
(29)
Solution
level
RE
CE
LC
GF
TE
Fig. 10
101
3 Electrochemical Deposition
Variation of the potential of the
test electrode, E, with time during
galvanostatic electrolysis (millisecond
range); Eo , equilibrium potential; Ei ,
potential of the test electrode at the
beginning of electrolysis at constant
current density i.
Fig. 11
Input
i
t=0
Time
[ms]
the potential Ei when the electrode reaction begins at a measurable rate. The time
necessary to charge capacitor Cdl is in the
range of microseconds. Thus, in the galvanostatic transient technique the duration
of the input current density pulse is of the
order of milliseconds (ms). From a series
of measurements of Ei for a set of i values,
one can construct the currentpotential relationship for an electrochemical process.
Figure 4 was constructed in this way.
E
Ei
Response
Overpotential h
Eo
t=0
t1
Time
iDL
t=0
Time
Input
E
h
Eo
t=0
i
i = f (t )
Response
102
Variation of current
with time during potentiostatic
electrolysis.
Fig. 12
t=0
Time
Introduction
(30)
(31)
103
104
3 Electrochemical Deposition
Free energy of
formation of a cluster as a
function of size N (a cluster of
N atoms); Nc , the size of the
critical cluster (nucleus).
Fig. 13
Gc
Nc
(32)
from solution onto the STM tip and subsequently transferring Cu from the tip to the
substrate by an appropriate tip approach
(Fig. 14). This technique can be fully automated and used for the fabrication of a variety of products in nano-electrochemical
technology (e.g. conductors in microelectronics).
Nucleation of Surface Nuclei
The nucleation law for a uniform probability with time t of conversion of a site on
the metal electrode into nuclei is given by
3.2.2.1
N = No [1 exp(At)]
(33)
(34)
Cu2+
Cu cluster
Substrate surface
Schematic diagram for the transfer of electrodeposited Cu from a Cu-covered STM tip to
the Au substrate during a tip approach (from Ref. 18 with permission from Elsevier Science Inc.).
Fig. 14
(35)
indicating that all electrode sites are converted to nuclei instantaneously. Hence,
this is referred to as instantaneous nucleation. Second, for small A and small t,
Eq. (33) reduces to
N ANo t
(36)
Overlap of diffusion
zones of cylindrical nuclei
growing on a surface. Shaded
regions, two zones overlapping;
black region, three zones
overlapping.
Fig. 15
105
106
3 Electrochemical Deposition
Microsteps
Propagation of copper microsteps on real
surfaces of Pt was observed by Rynders
and Alkire [27] using in situ AFM. They
used the Pt single crystal as the substrate
that was cut not exactly parallel to the
(100) plane but with a misorientation of
about 2 . This misorientation in cutting
resulted in microsteps of 25 to 50 A in
height and terraces approximately 1 in
length. Copper was electrodeposited from
the CuSO4 /H2 SO4 electrolyte. A sequence
of potential (or current) pulses (10 ms to
10 s) was used to obtain a sequence of
data that represents the initial growth
processes. AFM images show that the
3.2.3.2
Substrate
(a)
(b)
(c)
Fig. 16
t1
t2
Fig. 17
Bunching of steps.
107
108
3 Electrochemical Deposition
Nucleation-coalescence Growth
Mechanism
M, TDC
Madi overlayer
Substrate
Fig. 18
Stranski-Krastanov mechanism.
(37)
aS aM
aM
(38)
Thus, if
(Madi S) (Madi M)
(39)
Effect of Additives
Most solutions used in the electrodeposition of metals and alloys contain one or
more inorganic or organic additives that
have a specic function in the deposition
process. These additives (and impurities)
affect the deposition and crystal-building
processes as adsorbates (substances that
are adsorbed) at the surface of the cathode.
Adsorption and the factors that determine the adsorbatesurface interaction
are discussed in detail in Volume 1 and
Volume 2.
Chemisorption and Physisorption
One classication of adsorption phenomena is based on the adsorption energy, that
3.2.5.1
109
110
3 Electrochemical Deposition
is, the energy of the adsorbatesurface interaction. In this classication there are
two basic types of adsorption: chemisorption (an abbreviation of chemical adsorption) and physisorption (an abbreviation
of physical adsorption). In chemisorption,
the chemical attractive forces of adsorption
act between the surface and the adsorbate (usually covalent bonds). Thus, there
is a chemical combination between the
substrate and the adsorbate where electrons are shared and/or transferred. New
electronic congurations are formed by
this sharing of electrons. In physisorption,
the physical forces of adsorption, van der
Waals or electrostatic forces, operate between the surface and the adsorbate; there
is no electron transfer and no electron
sharing.
Adsorption energy for chemisorbed
species is larger than that for physisorbed
species. Typical values for chemisorption
are in the range of 20100 kcal mol1
and for physisorption in the range of
5 kcal mol1 .
Adsorption Equilibrium
Since in many cases, electrodeposition
in the presence of an additive does not
use up the additive (no incorporation
of the additive in the deposit), one can
conclude that the adsorption equilibrium
is a dynamic one. In a dynamic adsorption
equilibrium state, the adsorbed molecules
continually desorb at a rate equal to the
rate at which dissolved molecules from
the solution become adsorbed. If the
rates of the adsorption and desorption
processes are high and of the same order
of magnitude as that of the cathodic
deposition process, then no incorporation,
entrapment, of additives in the deposit will
occur. However, if they are much smaller,
additive molecules will be entrapped
in the deposit via propagating steps
3.2.5.2
Leveling
Leveling was initially dened as the progressive reduction of the surface roughness during deposition. Surface roughness
may be the result of a coarse mechanical polishing. In this case, scratches on
the cathode represent the initial roughness, and the result of cathodic leveling
is a smooth (at) deposit or a deposit of
reduced roughness. During this type of
leveling, more metal is deposited in recesses than on peaks. This leveling is of
great value in the metal-nishing industry.
It is also of great importance in the electronics industry for the deposition through
polymeric masks and in the electroforming of micromechanical devices (sensors,
actuators, micromotors). In these cases,
the leveling process means the uniform
deposition in microproles (lling up recesses), which can be dened with masks
3.2.5.6
Consumption of Additives
Additives can be consumed at the cathode
by incorporation into the deposit and/or
by the electrochemical reaction at the cathode or anode. Consumption of coumarin
3.2.5.8
111
3 Electrochemical Deposition
Rate of coumarin consumption
[108 mol cm2 s1]
112
2.0
1.5
t
1125 mV
1.0
960 mV
0.5
0
1.0
2.0
3.0
4.0
Fig. 19
in the deposition of nickel from a Wattstype solution was studied extensively; for
example, Roger and Taylor found that the
coumarin concentration decreases linearly
with time and that the rate of coumarin
consumption is a function of coumarin
bulk concentration [42]. Figure 19 shows
that the rate of consumption increases
with an increase in the bulk concentration
of coumarin. Thus, in the case of the electrodeposition of nickel from Watts-type
solution the total current density is the
sum of the current densities for nickel deposition iNi , hydrogen evolution iH , and
additive reduction iR .
In general, control of an electrodeposition solution includes monitoring the
concentration of additives and by-products
of the reaction of additives at the electrode.
from an aqueous solution are power supply, two metal electrodes (M1 and M2 ), water containing the dissolved ions, and two
metalsolution interfaces; M1 solution
and M2 solution. An electrolytic cell for
the electroless deposition is shown in
Fig. 20. It can be seen that in electroless deposition there is no power supply, and the
system has only one electrode. However,
the solution is more complex. It contains
water, a metal salt MA (Mz+ ; Az ), and a
reducing agent Red as basic components.
The overall reactions of electrodeposition and electroless deposition may be
used to compare these two processes. The
process of electrodeposition of metal M is
represented by
3.3
Electroless Deposition
3.3.1
Introduction
Electrode
Mz+
solution + ze Mlattice
(40)
Catalytic surface
Mz+
solution + Redsolution
Mlattice + Oxsolution
(41)
Fig. 20
(42)
Catalytic
substrate
M+
e Red
Solution:
H2O, salt MA
Red, reducing
agent
Mz+
Mlattice
solution + ze
(43)
and one oxidation reaction, the anodic
partial reaction
Catalytic surface
Redsolution
Oxsolution + me
(44)
113
3 Electrochemical Deposition
3.3.2.1
1. itotal = f (E) curve intersects the potential axis. At this intersection, the current
Oxidation
of Red
ia
Current density
(I )
114
ia
0
E Red
Electroless solution
Potential
Emp
[V]
ic
E 0Mz +
Reduction of Mz +, ic
Fig. 21
is zero and
ic = ia for itotal = 0
(45)
(46)
(+)
i M = f ()
Eeq,M
Emp
iM = f ()
= f (
i Red
Eeq,Red
()
iRe
Red
= f(
)
Log idep
Log i
Fig. 22
115
3 Electrochemical Deposition
(47)
(48)
0.3
0.4
Reduction of Cu2+
0.5
eq
Em
E versus. SCE
[V]
116
iCu2+ = f ()
0.6
0.7
Emp
0.8
eq
0.9
ERed
iHCHO = f ()
1.0
1.1
5
idep
Log i
[I in A cm2]
Fig. 23
Mz+ + ze M
(51)
3.3.3
(50)
RDS
Mz+ + e M(z1)+
(52)
117
118
3 Electrochemical Deposition
(53)
Redinterm R
(54)
(56)
(57)
(58)
Kinetics
The major factors determining the rate
of the anodic partial reaction are pH and
additives. Since OH ions are reactants
in the charge-transfer step, for example,
Eq. (59), the effect of pH is direct and
signicant (e.g. see Ref. 55). Additives may
have an inhibiting or an accelerating effect.
3.3.3.5
3.3.4
Mz+ + Red M + Ox
(60)
(61)
(62)
(63)
119
120
3 Electrochemical Deposition
of Al2 O3 results in the generation of aluminum particles, which are catalytic for
the electroless deposition of Cu (the rst
reaction probably being displacement deposition).
3.4
Displacement Deposition
3.4.1
Introduction
Zn
Cu
Zn2+
e
Cu2+
CuSo4 solution,
e.g. a (Cu2+) = 1
A displacement
deposition of Cu on Zn.
Fig. 24
(64)
Cu2+ + 2e Cu
(65)
and
(66)
Fig. 25
Increasing affinity of
ions for electrons
e
Flow of electrons
Electrochemical Model
The overall displacement deposition reaction in general Ox/Red (Mz+ /M) terms is
given by
z+
Ml + Mz+
r Mr + Ml
(67)
(68)
121
122
3 Electrochemical Deposition
Higher
electrode
potential
+
Er
Mr2+/Mr
Displacement
deposition
Mr2+ + ze
Mr
e
Flow of
electrons
Ei
Mi2+/Mi
Substrate metal
Mi2+ + ze
oxidation: Mi
Lower
electrode
potential
Fig. 26
The overall displacement deposition reaction according to Eq. (66) can be considered as the reaction of the electrochemical
cell
Zn/ZnSO4 (aq)//CuSO4 (aq)/Cu
(71)
Sn2+ + Cu Sn + Cu2+
(72)
E (Cu
2+
/Cu) = +0.34 V
(73)
(74)
(75)
(76)
where 0.55 is the standard electrode potential for Cu+ /Cu. Thus, in this case Cu+ /Cu
is more negative than Sn2+ /Sn and reaction (72) will go spontaneously from left
to right, resulting in the displacement
deposition of Sn on Cu. In this approximate calculation, we considered that only
[Cu(CN)3 ]2 complexed ions are present
in the solution and that the CN ions
do not affect the electrode potential of
the Sn2+ /Sn electrode. However, in reality, there is a mixture of different CN
complexes of copper in the solution.
Reaction (72) is a very important displacement deposition reaction in the
printed circuit industry. It is used to help
the soldering capability of copper.
3.4.3
3.5
Introduction
123
3 Electrochemical Deposition
Electrodeposition
Feasibility
The feasibility of the simultaneous deposition of two or more metals at the cathode
is solved on the basis of currentpotential
curves for individual, single (pure) metals.
Figure 27 shows currentpotential curves
for a general case of deposition of single
metals M1 and M2 . It can be seen that if
electrodes are immersed in the solution of
both M1 and M2 ions, and if the cathode
i1
M1
=
M2
i2
3.5.2.1
M1
i1
The simultaneous
deposition of metals M1 and M2
in which the rest (equilibrium)
potential ER,1 is more positive
than ER,2 and a1 = a2 .
Fig. 27
i2
E1 E(i )' E2
E (i )
Potential E
negative
(77)
M2
Current density
[i ]
124
This difference in potentials can be reduced by complexing Cu2+ and Zn2+ ions
in the solution. The preferred complexing agent is the CN ion. In a solution
containing CN ions (NaCN, KaCN), the
concentration of Cu2+ is reduced much
more than that of Zn2+ because of the
difference in the dissociation constant of
the cyanide complex. The dissociation constant of the [Cu(CN)3 ]2 complex is
(Cu+ )(CN )3
= 5.6 1028
([CuCN)3 ]2 )
(79)
(80)
Current density
[i ]
M2
M1
i2
i1
E1
E2
Negative
125
3 Electrochemical Deposition
Current density
[i ]
Cu
Zn
0.2
0.0
0.2
0.4
0.6
0.8
1.0
E 0
E 0(Cu2+/Cu)
0.34 V
E 0 = 1.10 V
E 0(Zn2+/Zn)
0.76 V
(a)
Cu
Zn
Current density
[i ]
126
0.2
0.0
0.2
0.6
1.0
E
E = 0.42 V
(b)
E(Cu+/Cu)
0.43 V
E(Zn2+/Zn)
0.85 V
Fig. 29
[FeOH]+ + e FeOH
(84)
FeOH + H+ + e Fe + H2 O
(85)
where RDS stands for rate-determining
step. The overall electrode reaction is
Fe2+ + 2e Fe
(86)
127
3 Electrochemical Deposition
16
12
Calculated partial
currents I of nickel and iron,
single versus alloy at 1600 rpm
(from Ref. 70 with permission
from the Electrochemical
Society).
Fig. 30
20
i
[mA cm2]
128
1. iFe, alloy
2. iFe, single
3. iNi, alloy
4. iNi, single
8
1,2
0
0.52 0.56 0.60 0.64 0.68 0.72 0.76 0.80
VVNHE
[V]
shows calculated partial current densities of Ni and Fe, single versus alloy
deposition. Thus, the model of Hessami
and Tobias explains the basic features
of anomalous deposition: (1) the ratio of
the less positive (noble) metal, Fe, is
much higher in the alloy than in the
electrolyte, and (2) the presence of the
less positive (noble) metal inhibits the
discharge of the more positive (noble)
metal, Ni.
The model based on metal-hydroxide
ions ([MOH]+ ) was further developed by
Grande and Talbot [71]. Sasaki and Talbot [72] demonstrated the extendibility of
this model to the electrodeposition of
CoFe and NiCo alloys. They found
that there is a slight inhibition of the
more positive metal deposition and a promotion (acceleration) of the less positive
metal deposition for all binary iron-group
alloys.
Golodnitsky et al. found that inhibition
of the more positive metal deposition
by the less noble one does not depend
on the anion composition of the electrolyte [73].
3.5.3
Electroless Deposition
3.6
Introduction
Columnar Microstructure
129
130
3 Electrochemical Deposition
Polycrystalline
+ lamellae
[mA cm2]
Lamellae
Cubic layers
25 20 17 15 12 10
Pyramids
3
50
100
150
[mv]
Schematic cross
section (perpendicular to the
substrate) of the columnar
deposit.
Fig. 32
Substrate
Development of Texture
3.6.4.1
(a)
(b)
Fig. 33
131
132
3 Electrochemical Deposition
crystallographic axes of grains constituting a polycrystal are xed (have the same
orientation) with respect to the axes of
the reference system, the polycrystalline
material exhibits preferred orientation or
texture (Fig. 33b). The development of texture can occur during the deposition or the
postdeposition processing.
The competitive growth model of the development of texture during deposition is
based on the idea that different crystal
faces have different rates of growth. Thus,
there is a growth rate competition between crystallites of various orientations.
Crystallites of various orientations could
be generated either during the preferential
nucleation process or during the competitive growth mechanism subsequent to
the stage of coalescence. The type of
texture depends on the composition of
electrolyte, substrate, the overpotential,
and so on.
Amorphous Substrate
An amorphous substrate (e.g. vitreous
carbon) is without any epitaxial inuence;
it is inert with respect to the growth process
of the deposit. Froment and coworkers [84]
showed that in the initial stages of the Ni
growth from Watts solution on vitreous
carbon substrate, orientation of individual
3D nuclei is random. They also showed
that a newly coalesced compact deposit
has perfectly random orientation. From
these observations, they concluded that
the texture of thicker Ni deposits in
this case is the result of a competitive
growth mechanism occurring in a stage
of growth subsequent to the coalescence
stage.
In Sect. 3.7.1, we present one example
which shows how texture may determine
the physical properties and reliability of
thin lms.
3.6.4.3
3.6.5
Dendrites
Dendritic deposits grow under mass transportcontrolled electrodeposition conditions. These conditions involve low
concentration of electrolyte and high current density. A dendrite is a skeleton
of a monocrystal consisting of stem and
branches. The shapes of the dendrites are
mainly determined by the directions of preferred growth in the lattice. The simplest
dendrites consist of the stem and primary
branches. The primary branches may develop secondary and tertiary branches. The
angles between the stem and the branches,
or between different branches, assume certain denite values in accordance with the
space lattice. Thus, dendrites can be two dimensional (2D) or three dimensional (3D).
Figure 34 shows the fractal structure of
a at 2D electrodeposited copper leaves
at increasing magnication: 64, 256, 523,
and 1024. Fractal objects are the same on
different observation scales. The thickness
of the deposit in Fig. 34 is of the order
of 200 nm. Copper is in its normal fcc
structure, but has many microcrystals
whose size is not larger than 20 nm.
The dendrite growth process may be described on the basis of cluster growth model
of diffusion-limited aggregation (DLA) and
fractal concepts in surface growth [83, 85].
3.6.6
Whiskers
Sequence of pictures showing different magnications of the same leaf (from Ref. 85 with
permission from Materials Research Society).
Fig. 34
formation [87]. The two important characteristics of whiskers are (1) large ratio
between the longitudinal and the lateral
dimensions and (2) preservation of the lateral dimensions during the growth [30].
3.6.7
(87)
Electrical Resistance
According to the modern quantum electronic theory [1, 88, 89], electrical resistivity
of a metal results from the scattering of
electrons by the lattice. In a perfect lattice, electrons experience no scattering,
and they can carry current without any attenuation. A real metal lattice departs from
perfect long-range order, and electrons are
(88)
133
134
3 Electrochemical Deposition
Mechanical Properties
3.7
Introduction
The most active areas in the modern applications of electrochemical deposition are
semiconductor technology and magnetic
recording.
One major recent advance in siliconbased semiconductor industry is the development of copper interconnects on chips.
This new technology replaces the traditional aluminum or aluminum alloy (e.g.
AlCu) conductors produced by physical
vapor deposition (PVD) with copper conductors manufactured by electrodeposition. Copper has been replacing aluminum
since 1999 owing to its low bulk electrical
(89)
where Ea is the activation energy for grainboundary diffusion, n is the current density
exponent, J is the current density, T is
the temperature, k is the Boltzmanns
constant, and A is a constant [92, 95].
The number of transistors per chip is
increasing continuously (for a microprocessor, the total number of transistors per
chip was 11, 21, and 40 M (million) for
years 1997, 1999, and 2001, respectively)
and the physical feature size of transistors
is decreasing; consequently the dimensions of interconnections (interconnects)
on the chip are scaling down; for example, linewidths were 0.25, 0.18, 0.15, and
0.10 m in the years 1997, 1999, 2001,
and 2006, respectively. This scaling down
of interconnects on chips requires interconnect metal of high quality and better
understanding of physical properties of
thin lms [96].
The required degree of understanding of
physical properties of thin metal lms used
for interconnects on chips is illustrated in
(90)
135
136
3 Electrochemical Deposition
Aluminum
(a)
W
Stud
W
Stud
Al
Dielectric
Al
W
Stud
W
Stud
(b)
Dielectric
W
Stud
W
Stud
Wire
Wire
W
Stud
W
Stud
(c)
(d)
Fig. 35
(a)
Barrier (Ta)
Cu seed layer
Dielectric
(e)
(b)
Resist
Electrodeposited
Cu
Cu
(c)
(f)
(d)
(g)
Cu
via
Cu line
Cu
via
(h)
Fig. 36
137
138
3 Electrochemical Deposition
Dielectric
Dielectric
(a)
Substrate
(b)
Cu seed layer
Barrier (Ta)
(c)
Cu line
Cu
via
(d)
Substrate
(b)
Cu seed layer
Resist
(c)
Cu
(d)
Cu
(e)
Dielectric
Cu
(f)
Fig. 38
shorts between interconnects. Thus, diffusion barrier layers are an integral part
of the fabrication of copper interconnects
(Figs 36, 37). Barrier lms isolate (encapsulate) Cu interconnects from adjacent
dielectric material. The most studied diffusion barriers are Ta [110, 111], Ti, and
TiN [112, 113].
Since barrier metals have relatively high
electrical resistivity (Ta 12.4; Ti 80 cm)
139
140
3 Electrochemical Deposition
Resist
Dielectric
(a)
Substrate
(b)
Cu seed layer
(c)
(d)
(e)
Fig. 39
Cu seed layer
Dielectric
(a)
Substrate
Dielectric
Substrate
(b)
Cu seed layer
Fig. 40
Dielectric
Cu
Barrier (Ta)
(a)
Substrate
Top dielectric barrier
Dielectric
Cu
(b)
Barrier (Ta)
Substrate
Fig. 41
141
142
3 Electrochemical Deposition
al
Second
interleve
l dielectr
ctric
le
evel die
Via
terl
Third in
Second le
vel metal
Via
ic
First lev
el meta
l
First in
terlev
el die
lectric
Cu
Conta
ct
Cu
Transis
to
r regio
Silico
n sub
Fig. 42
strate
Introduction
Read current
Coil
Read head
read element
MR or GMR
sensor
Magnetic core
(pole, yoke)
Shield
Track width
S S
N N
Magnetization
SS
N N
S S
NN
Gap
Inductive write
element write
head
Recording
medium
Fig. 43
Write Heads
143
144
3 Electrochemical Deposition
(91)
(92)
(93)
Read Heads
145
146
3 Electrochemical Deposition
(94)
HS
0.7
0.6
HS
(Fe 30 /Cr 9 )40
0.5
40
30
20
10
H<0
F
HS
10
20
0
Magnetic field H
[kG]
H=0
AF
30
40
H>0
F
Fig. 44
147
3 Electrochemical Deposition
50
295 K
40
Under/over
H parallel I
1. 50 Fe/50 Fe
2
R/R
(%)
Overlayer
30
20
2. 50 Fe/50 Cu
3. 50 Cu/50 Fe
4. 50 Cu/50 Cu
4
10
0
4
0
H
[kOe]
Fig. 45
148
0
10
20
30
40
50
60
10
20
30
40
Cu thickness
[]
Theoretical curve
Experimental points
50
Fig. 46
There are two types of magnetic recording media: particulate media and thin-lm
media. Particulate media are composed
of magnetic particles embedded in an
organic polymer matrix. The most commonly used particles are -Fe2 O3 (gamma
ferric oxide), cobalt-modied gamma ferric oxide, -Fe2 O3 + Co, CrO2 (chromium
dioxide), metal particles (FeNi, FeCo),
and barium ferrite (BaFe12 2x Cox Tix O19 ).
The metal particles and barium ferrite are
preferred materials for high-density applications. Particulate magnetic media are
prepared by chemical precipitation techniques. The coercivity of particulate media
is in the range 250 to 2300 Oe [144].
In the early 1960s, particulate media
dominated the rigid disk as well as the
(95)
149
3 Electrochemical Deposition
14
12
10
Hc
[102 Oe]
150
8
6
4
2
0
1000
2000
3000
4000
5000
Deposit thickness
[]
of 450680 A.
Electrodeposited CoNi(P) alloys
show coercivity of 1020 Oe at the thickness
of 3200 A [150].
Thickness
Process
Lubricant
10
Dipping
150
Sputtering
~500
Electroless
deposition
Underlayer (NiP)
~10 m
Electroless
deposition
Substrate
(AlMg alloy)
~0.6 mm
Schematic cross
section of a general electroless
deposited thin-lm disk
magnetic recording medium.
Fig. 48
Magnetic
layer
structure
Thickness
Process
Lubricant
10
Dipping
150
Sputtering
300
Sputtering
Underlayer (Cr)
500
Sputtering
Sublayer (NiP)
~10 m
Electroless
deposition
Substrate
(AlMg or glass)
~0.6 mm
Thickness
(approximate)
Process
20
Dipping
75
100
Spacer layer
520
100
Intermediate layer
50
Underlayer
150
Seed layer
50
Underlayer
structure
Sputtering
Schematic cross
section of a dominant thin-lm
magnetic disk structure.
Fig. 49
Substrate
Fig. 50
151
152
3 Electrochemical Deposition
153
154
3 Electrochemical Deposition
72. K. Y. Sasaki, J. B. Talbot, J. Electrochem. Soc.
1995, 142, 775782; 2000, 146, 189197.
73. D. Golodnitsky, N. V. Gudin, G. A.
Volyanuk, J. Electrochem. Soc. 2000, 147,
41564163.
74. I. Ohno, O. Wakabayashi, S. Haruyama, J.
Electrochem. Soc. 1985, 132, 23232330.
75. G. E. Georgiou, F. A. Baiocchi, H. S. Luftman et al., J. Electrochem. Soc. 1991, 138,
20612069.
76. Y. Shacham-Diamand, S. Lopatin, Microelectron. Eng. 1997, 37/38, 7778.
77. H. Sawai, T. Kanamori, I. Koiwa et al., J.
Electrochem. Soc. 1990, 137, 36533660.
78. T. Osaka, T. Homma, T. Yokoshima, in
Modern Electroplating, 4th ed. (Eds.:
M. Schlesinger, M. Paunovic), Wiley, New
York, 2000, Chapter 19.
79. I. Ohno, in Modern Electroplating, 4th ed.
(Eds.: M. Schlesinger, M. Paunovic), Wiley,
New York, 2000, Chapter 22.
80. H. Seiter, H. Fischer, L. Albert, Electrochim.
Acta 1960, 2, 97120.
81. S. C. Barnes, G. G. Storey, H. J. Pick,
Electrochim. Acta 1960, 2, 195206.
82. D. J. Srolovitz, A. Mazor, G. G. Bukiet, J.
Vac. Sci. Technol. 1988, A6, 23712380.
83. G. L. M. K. S. Kahandra, X. Zou, R. Farrell
et al., Phys. Rev. Lett. 1992, 56, 3741.
84. M. Amblart, G. Froment, N. Maurin et al.,
Electrochim. Acta 1983, 28, 909915.
85 A.-L. Barabasi, H. E. Stanly, Fractal Concepts
in Surface Growth, Cambridge University
Press, Cambridge, 1995.
86. P. B. Price, D. A. Vermilyea, M. B. Webb,
Acta Metall. Mater. 1958, 6, 524531.
87. Y. Zhang, J. A. Abys, in Modern Electroplating, 4th ed. (Eds.: M. Schlesinger,
M. Paunovic), Wiley, New York, 2000.
88. Z. Fisk, G. W. Webb, in Treatise on Material
Science and Technology (Ed.: F. W. Fradin),
Academic Press, New York, 1981, p. 297,
Vol. 21.
89. P. L. Rossiter, The Electrical Resistivity of
Metals and Alloys, Cambridge University
Press, Cambridge, 1987.
90. M. Paunovic, L. A. Clevenger, J. Gupta
et al., J. Electrochem. Soc. 1993, 140,
26902694.
91. J. M. E. Harper, E. G. Colgan, C.-K. Hu
et al., MRS Bull. 1994, 19(8), 2329.
92. D. S. Gardner, J. Omuki, K. Kudoo et al.,
Thin Solid Films 1995, 262, 104119.
136.
137.
138.
139.
140.
141.
142.
143.
144.
145.
146.
147.
148.
149.
150.
151.
152.
153.
154.
155.
156.
157.
155
156
3 Electrochemical Deposition
158. T. Osaka, T. Homma, T. Yokoshima, in
Modern Electroplating (Eds.: M. Schlesinger,
M. Paunovic), Wiley, New York, 2000,
Chapter 19.
159. L. Callegaro, E. Puppin, P. L. Cavallotti
et al., J. Magn. Magn. Mater. 1996, 155,
190198.
160. V. Dubin, Y. Shacham-Diamand, B. Zhao
et al., Mater. Res. Soc. Symp. Proc. 1996,
427, 179182.
Appendix
A1
Fig. 51
A3 Magnetic Anisotropy
Hysteresis loop of hard
magnetic material.
Fig. 52
Ms
Mr
b
a
Hc
+Hc
A2
Mr
Ms
Fig. 53
M
Ms
b
a
Mr
Hc
H
+Hc
H
b
Mr
A3
Magnetic Anisotropy
Ms
M
157
3 Electrochemical Deposition
1400
[0001]
Easy
1200
[0001]
1000
M
[emu cm3]
158
Hard
Easy
0]
1
[10
800
600
400
[1010]
Hard
200
0
5
6
H
[103 Oe]
10
Fig. 54 Magnetization curve for easy [0001 c axis] and hard [1010] axis for
cobalt showing large magnetocrystalline anisotropy (from Ref. 122).
Magnetostriction
l
l
(96)
161
4.1
Electrowinning, electrorening, and electroplating involve the exchange of electrons between a solid electrode and ions
or molecules dissolved in solution. The
underlying physical and chemical laws are
the same for most metals and applications.
The rate of the reaction involved depends
on the electrode potential, the electrode
area, and the rate of mass transport of
the electroactive species to the electrode
surface. The main differences are in the
construction of the cells, their geometry, construction materials, and operating
practice. The process examples presented
have been selected on the basis of their
historical or technological importance.
The background of electrolysis processes
is the Faradays law that connects the
amount of electricity and mass of reacted
material. The quantity of electricity in
coulombs equals the number of electrons
involved in the reaction multiplied by
the Faradays constant (F ). The value of
is mass (g)
is current (A)
is time (s)
is atomic weight (g mol1 )
is valency of the metal (number
of electrons)
F is Faradays constant
(96 485 C mol1 ).
I t
zF
(2)
162
4 Electrowinning of Metals
where
is current efciency (%).
A value of current efciency below
100% usually indicates that by-products
are formed. Another possibility is that
reverse reaction happens, for example,
dissolution of deposited metal. In metal
recovery from acidic solutions, the hydrogen evolution reaction is a common side
reaction that lowers the current efciency.
The current efciency can also vary with
the current density. Current efciency is a
widely used measure of tankhouse prociency in producing metal, but it does not
provide a direct measure of metal quality.
Using Faradays law it is possible
to calculate a so-called electrochemical
equivalent for a metal as in Eq. (3)
ekv(g A1 S1 ) =
M
zF
(3)
Metal
Ag
Au
Au
Cd
Co
Cu
Cu
Ni
Pb
Pb
Pt
Sn
Sn
Zn
Valence
Electrochemical
equivalent
[mg A1 s1 ]
Electrochemical
equivalent
[g A1 h1 ]
107.9
197.0
197.0
112.4
58.9
63.5
63.5
58.7
207.2
207.2
195.1
118.7
118.7
65.4
1
1
3
2
2
1
2
2
2
4
2
2
4
2
1.12
2.04
0.68
0.58
0.31
0.66
0.33
0.30
1.07
0.54
1.01
0.62
0.31
0.34
4.03
7.34
2.45
2.09
1.12
2.38
1.19
1.08
3.85
1.94
3.64
2.23
1.12
1.22
Electrolyte
Fig. 1
Metal deposition
Gas evolution
Permanent anode
Purified cathode
Metal deposition
Metal dissolution
Impure anode
Deposited cathode
Electrorefining
Electrolyte
2Cl Cl2 + 2e
(4)
(5)
z+
+ ze Me
(6)
(7)
Table 2 lists standard electrode potentials for some metals and some other
reactions common in electrorening and
electrowinning. The metals with high equilibrium potential are noble metals. They
are often difcult to dissolve but deposit
easily. The metals with low equilibrium
potentials are active metals that dissolve
easily but are more difcult to reduce.
163
164
4 Electrowinning of Metals
Standard electrode potentials of
reactions of interest in metal electrolysis [1]
Reaction
Component
Tab. 2
Au+ + e Au
Au+ + 3e Au
Cl2 + 2e 2Cl
O2 + 4H+ + 4e 2H2 O
Pt2+ + 2e Pt
Pd2+ + 2e Pd
Hg2 2+ + 2e 2Hg
Ag+ + e Ag
Cu+ + e Cu
SbO2 + 4H+ + 3e Sb + 2H2 O
AsO2 + 4H+ + 3e As + 2H2 O
Cu2+ + 2e Cu
Bi3+ + 3e Bi
2H+ + 2e H2
Fe3+ + 3e Fe
Pb2+ + 2e Pb
Sn2+ + 2e Sn
Ni2+ + 2e Ni
Co2+ + 2e Co
Cd2+ + 2e Cd
Fe2+ + 2e Fe
Zn2+ + 2e Zn
Cr2+ + 2e Cr
Mn2+ + 2e Mn
Ti2+ + 2e Ti
Al3+ + 3e Al
Mg2+ + 2e Mg
Standard
electrode
potential,
Eo [V]
1.692
1.498
1.395
1.228
1.188
0.987
0.788
0.799
0.520
0.446
0.429
0.337
0.215
0.000
0.037
0.126
0.136
0.250
0.277
0.403
0.440
0.763
0.913
1.179
1.630
1.663
2.363
An electrowinning or electrorening
electrolyte is not a simple aqueous solution
of a metal salt. The electrolyte contains
many compounds that all have specic
functions. Some compounds have a positive effect, whereas others are detrimental.
The electrolyte component types and their
roles are listed in Table 3.
Basic Electrochemistry
The probability of an electrochemical reaction happening is evaluated as electrode
potential. The reaction rate is measured as
4.1.1.1
Metal ion
Anion
Auxiliary salt
Impurities
Inorganic
additives
Organic
additives
Water
Suspended
solids
Role
Source of metal
Maximize solubility
Provide anode reaction
Conductivity
pH control
Ionic strength and
conductivity
Buffering and pH control
By-product for recovery
Detrimental to efciency or
purity
Assist anode life
Impurity removal
Cathode leveling
Impurity control and
improved efciency
Grain rener
Mist suppressor
Solvent
Incorporation into cathode
or remove as cell sludge
G) than in the cathodic direction (positive G). If the equilibrium potential is
high, the reaction will more likely proceed
in the cathodic direction.
The electrochemical series tabulates
standard electrode potentials. Some
sources call the electrochemical series oxidation/reduction potentials, electromotive
series, and so on. The reference state of
electrochemical series is the hydrogen evolution reaction, or H+ /H2 reaction. Its
standard electrode potential has been universally assigned as 0 V. This electrode
is the standard hydrogen electrode (SHE)
against which all others are compared.
For example, the standard electrode potential of the Fe/Fe2+ reaction is 0.440 V
and that of Cu/Cu2+ reaction is +0.337 V.
The standard electrode potentials are calculated from Gibbs free energy values by
Eq. (8) that is applicable only in the abovementioned standard state.
Eo =
G
zF
(8)
(9)
where
R is the molar gas constant =
8.3143 J mol1 K1
T is absolute temperature (K)
cR and cO refer to the concentrations of
reduced and oxidized species,
for example, solid metal and
metal ion.
165
166
4 Electrowinning of Metals
actual electrode potential and its equilibrium potential is overpotential. The signs
of the overpotential and current are positive for anodic reactions and negative for
cathodic reactions.
Increasing polarization increases the
current and vice versa. If the polarization
is low, that is, the potential shift per unit
increase in current density is small, then
no factors will retard the electrochemical
reaction rate. If the polarization is high,
the electrode potential must change considerably to obtain higher currents.
Polarization has various meanings and
interpretations depending on the system
under study. For an electrochemical reaction, this is the difference between actual
electrode potential and reaction equilibrium potential. Anodic polarization is the
shift of anode potential to the positive direction, and cathodic polarization is the
shift of cathode potential to the negative direction. In an electrochemical production
system driven with an external current
source, polarization is a harmful phenomenon. It will increase the cell voltage
and therefore production costs. A system
that polarizes easily will not pass high
currents even at high overpotentials. The
reaction rates are therefore small.
Electrode polarization is not a simple
phenomenon. Electrochemical reactions
are heterogeneous reactions. They occur
in several phases in subsequent steps. The
reaction path of metal deposition usually
includes the following steps:
Transfer of reactive species from electrolyte to electrode surface
Adsorption
Charge-transfer step at the surface
Surface diffusion
Growth of the metal atom into the metal
crystal structure.
i = i0 e RT e RT
(10)
where i is the current density and i0 is the
exchange current density corresponding
to the reaction rate in the dynamic
equilibrium. In Eq. (10), the factor is
the symmetry factor with values between
0 and 1. If is higher than 0.5, the rate of
the anodic partial reaction increases faster
with increasing overpotential than the rate
of the cathodic reaction. At values 0 < <
0.5, the rate of the cathodic partial reaction
increases faster. The high overpotential
(and high current density) approximation
of the ButlerVolmer equation is known
as Tafel equation (11)
= a + b log(i)
(11)
167
168
4 Electrowinning of Metals
(13)
In Eq. (13) km is the local mass transport coefcient D/, that is, diffusion
coefcient divided by the thickness of diffusion layer. High mass transport rates
can be achieved by electrode movement,
Anode
Cathode
Anode
Cathode
34 cm
CA
CC
CMez+
(bulk)
The concentration gradients in electrorening cell with vertical plate electrodes and
the resulting natural convective ow pattern.
Fig. 2
(14)
10
Current density
[mA cm2]
Current density
[mA cm2]
0.1
0
(a)
50
100
150
Overpotential
[mV]
200
0
(b)
50
100
150
Overpotential
[mV]
200
Polarization curves for different overpotentials. In (a) the current density is on linear scale
and in (b) on logarithmic scale.
Fig. 3
169
4 Electrowinning of Metals
+
+
+ +
+
+
+
+ +
+
+ +
+
+
+
+
+
+
+
+ +
Hydrated
cation
Dehydration of
metal cation
Charge
transfer reaction
+
+
+ +
+
+
+ +
+
+
+
+
+
+
Growth site
+
+
+ +
+
+
+ +
+
+
+
+
+
+
+
Adsorbed atom
+
+
+
Metal electrode
Electrochemical
double layer
Fig. 4
+
+
+ +
Diffusion layer
Hydrodynamic layer
+
+
+ +
+
+
+
170
/I
L
(15)
171
4 Electrowinning of Metals
< a /2
> a /2
172
a
Primary current
distribution
Fig. 5
Secondary current
distribution
Tertiary current
distribution
The effect of primary, secondary, and tertiary current distribution on deposit thickness.
coefcient, that is, on the diffusion coefcient and diffusion layer thickness. The
effect of tertiary current distribution depends on the characteristic length with
respect to diffusion layer thickness. If the
characteristic length (distance between two
peaks) is larger than the diffusion layer
thickness, the tertiary current distribution
follows the surface prole. If the characteristic length is smaller than the diffusion
layer thickness, the tertiary current distribution will decrease the smoothing effect
of secondary current distribution. The diffusion length from the bulk electrolyte to
the peaks is smaller, and, thus, more metal
can deposit on the peaks, giving a more
rough deposit [3, p. 126].
Figure 5 shows the effect of different current distributions on the potential distribution on the electrode surface. In primary
current distribution, the current concentrates on protruding areas, because they
are closer to the opposite electrode. Protruding areas also have lateral accessibility
to current, and primary current distribution would give innite current density on
ideally sharp peaks. In secondary current
distribution, the protruding areas polarize
more owing to the higher local current
density. This greater overpotential tends to
oppose current ow and divert current to
the recessed areas. The ohmic resistance
to be overcome decreases with decreasing
prole depth, and the equalizing effect is
stronger on smaller prole depths. The
current efciency is often dependent on
current density, and if the current efciency decreases with increasing current
density, it has a smoothing effect. If the
current density is close to the limiting
current density, the tertiary current distribution prevails. The higher the ratio of
current to limiting current, the stronger
will be the preferential deposition on microprole peaks [4].
173
174
4 Electrowinning of Metals
z F Ecell
M
(17)
where Ecell is the cell voltage (V). The specic energy consumption for a particular
metal and process is calculated using the
electrochemical equivalent of the metal,
the voltage needed to drive the process,
and the current efciency. The common
way to express energy consumption is to
use kWh kg1 by dividing the right-hand
side of Eq. (16) by 3600. To lower the specic energy consumption, it is necessary
to lower the cell voltage and/or increase
the current efciency.
Energy costs are an important component of any electrolysis process. The
energy consumption depends on the operating current density and cell voltage.
The operating current densities are well
below the limiting current density set by
the mass-transfer rate. It has been found
that operation at some fraction of the limiting current density is necessary to avoid
excessive cell voltage and operating problems such as uneven deposit growth or
low current efciency. The optimum current density is also a compromise between
reduced capital costs and higher operating
costs as current density is increased.
Examples of the current densitycell
voltage dependence are shown in Figs 6
and 7. In Fig. 6, the system is the electrorening of copper. The thermodynamical
cell voltage is zero. When the cell is operated, polarization effects determine the
cell voltage. Increasing the current density
increases polarization. Upon increasing
cathode polarization, the deposit morphology changes to more rough and nally
to powder when limiting current density is
reached. The anode polarizes continuously
unless anode impurities cause anode passivation, after which the anode potential
increases sharply. In Fig. 7, the system is
the electrowinning of copper. The thermodynamical cell voltage is 0.89 V. The total
cell voltage depends again on the electrode
polarization, which depends on the operating current density. The effect of cathode
polarization is the same as in electrorening. The anode polarizes continuously as
the anode reaction is the decomposition of
water, Eq. (4).
Table 4 shows examples of the energy
consumption of some electrorening and
electrowinning processes [5, p. 163].
Factors Affecting Deposit Quality
A metal deposit consists of crystals, which
originate as nucleus. Formation of a new
crystal begins with the adsorption of a
single ion from the electrolyte reducing to
an adatom. This adatom is loosely bound
to the surface and at any given moment, a
certain number of adatoms are present on
the surface. Adatoms can dissolve again,
form a new nucleus, or grow into an
existing crystal. The adatoms move around
4.1.1.4
Anodic
limiting
current
Ox
evo ygen
luti
on
Anodic current
Anode passivation
IK = IA
Electrode
potential
Cathodic
limiting
current
Hy
evo drog
luti en
on
Smooth
cathode
Rough
cathode
Powder
Cathodic current
Oxy
evo gen
lutio
n
Fig. 6
IK = IA
Electrode
potential
Cathodic
limiting
current
Hy
evo drog
luti en
on
Smooth
cathode
Rough
cathode
Powder
Cathodic current
Fig. 7 Current densitycell voltage dependence for the electrowinning
of copper.
175
176
4 Electrowinning of Metals
Tab. 4
Metal
Comparison of energy use in some electrorening and electrowinning operations [5, p. 163]
Electrolyte
Electrorening
Ag
Nitrate
Cu
Sulfate
Ni
Sulfate/chloride
Pb
Fluosilicic
Electrowinning
Cu
Sulfate
Co
Sulfate
Zn
Sulfate
Temperature
[ C]
Thermodynamical
voltage [V]
Cell
voltage
[V]
Current
density
[A m2 ]
Current
efciency
[%]
Energy
consumption
[kWh kg1 ]
25
60
60
40
0
0
0
0
2.02.8
0.20.3
1.9
0.45
300400
150350
200
200
9398
9298
98
9092
0.7
0.20.3
2.0
0.2
40
60
35
0.90
1.52
2.0
1.92.0
3.54.0
3.6
170
160
4001000
8590
8090
9094
1.82.0
3.7
3.2
Fig. 8
FI
BR
FT
UD
177
4 Electrowinning of Metals
J/C M e z +
Very
low
Low
Medium
No deposit
or FI or screw
dislocation
No nucleation
increasing
BR
BR
increasing Nc,2
Very
low
Low
Inhibition intensity
BR
High
FI dendrites
Nc,2
Nc,2
increasing
BR
Very
high
FI powder
increasing Nc,3
BR
Z or FT
FI
or
FI powder
or
UD if bad
crystallization
increasing Nc,3
FT
UD
UD
UD
Medium
FT
FT
Hydrogen
evolution
or
Discharge
of
another
ion
FT
Very
High
UD
UD
increasing Nc,3
High
increasing Nc,3
178
in powder
Hydrogen evolution
or
discharge of another ion
in powder
Fig. 9
Additives that increase the deposition overpotential at a given current density, for instance, by altering the Tafel constants, can
be considered deposit-leveling additives.
Since additives are typically present in very
small concentrations, their transport to the
electrode is nearly always under diffusion
control and sensitive to ow variations.
There are several factors affecting cathode purity that are related to the additives.
During the electrolysis process, incorrect
additive levels can cause excessive edge
growth, nodular growth, or dendrites.
Electrolysis Technology
The design and construction of an electrolysis cell depend on the cell reactions, metal
concentration, and production rate. Most
of the cell designs operate continuously
and at steady state. In electrorening operations, the electrolyte composition does
not change much. In electrowinning operations, the metal concentration decreases
and acidity often increases. Most of the
cell design parameters deal with mass
transport of the reacting species and maintaining the purity and composition of the
electrolyte.
Cell Construction
The traditional electrolytic cells are simple, rectangular, open tanks with parallelplate electrodes. Cells are generally made
from concrete and lined with lead or
polyvinylchloride (PVC). The traditional
4.1.2.1
conguration is lead-lined concrete. Concrete cells with plastic liners were popular
in the 1970s and 1980s. The corrosionresistant lining is necessary to protect the
tank from the acid corrosive electrolyte.
Polymer-modied concrete is nowadays
used to avoid the lining of cells. The polymer concrete cells became popular in the
late 1980s and 1990s in copper production with conversions to permanent cathode technology and in new tankhouses.
Polymer concrete is a corrosion-resistant
construction material that withstands the
mechanical wear in the electrode handling.
In the traditional cells, the cathodes
and anodes are at plates that are placed
parallel to each other. Anodes and cathodes
are suspended alternately in the cells
with precise spacing. In some cases,
it is necessary to separate anode and
cathode electrolytes. This is done by
using separators, with diaphragms that
allow solution ow or membranes that
allow only anion or cation transfer. The
deposition of metal onto the cathodes
is a batch process. The cathodes are
removed from the cell when the deposit
has grown thick enough. The ow rate of
the electrolyte is kept low so that possible
solids fall to the bottom of the cell and do
not contaminate the cathodes. Figure 10
shows an example of the traditional
parallel-plate cell.
Diaphragm or membrane cells are used
in special cases where impurities are easily
codeposited on the cathode, anode slimes
are a problem, or the anolyte and catholyte
have a different composition. A diaphragm
cell can be made by enclosing either the
anodes or the cathodes in a rigid bag.
Enclosing anodes in a bag is an efcient
method to collect anode slimes, but the
anode compartments must be cleaned
quite often. Enclosing cathodes in a bag
requires purication of the anolyte in a
179
180
4 Electrowinning of Metals
Electrolyte in
Electrolyte out
Fig. 10
181
4 Electrowinning of Metals
Packed bed
Fluidized bed
Spouted bed
Perforated plates
182
Anode
Diaphragm
Cathode bed
Fig. 11
(a) +
Fig. 12
E = Ecell
(b) +
E = 5 Ecell
183
184
4 Electrowinning of Metals
Insulator
Busbar
Insulator
Anode
Fig. 13
Cathode
Cell wall
i+ t+
t+ + t0
(18)
ieff =
i+ t+ i t
t+ + t
(19)
185
186
4 Electrowinning of Metals
The controlled grain structure allows forming an adherent protective oxide coating
within days when placed in an electrowinning cell without the necessity of being
preconditioned [13]. In sulfuric acid electrolytes, the introduction of cobalt ions into
the solution reduces the anode overvoltage
and the corrosion rate of lead.
The electrocatalytically active layer of
lead dioxide has a high overpotential
toward oxygen evolution, and this causes
a signicant increase in energy use.
The formation of lead dioxide layer
happens at potentials of 1.6 V. This means
an additional overvoltage of 0.4 V as
compared to the equilibrium potential of
oxygen evolution. The formation of lead
dioxide requires corrosion of lead, which
in sulfate solutions results rst in the
formation of lead sulfate. Crystallization of
lead sulfate increases the anode potential
to such high values that formation of
lead dioxide becomes possible. During
operation, it is important to keep a PbO2
electrode well above the PbO2 PbSO4
potential; otherwise, the corrosion of the
anode will occur. Despite the formation
of a passive lm of lead dioxide, there
is generally a slow continued corrosion,
which leads to the thickening of the
PbO2 lm.
The DSA-type anodes are inert, coated
anodes made of a valve metal (titanium,
niobium, or tantalum) base coated with
an electrochemically active coating. The
active coating is made either of noble
metals or of mixed metal oxides. Noble
metals in active coatings are usually
platinum or platinum alloys. Mixed metaloxide coatings contain active oxides and
inert oxides; the active components are
usually ruthenium dioxide (RuO2 ) and
iridium dioxide (IrO2 ) and the inert
components are mostly titanium dioxide
(TiO2 ) and other oxides such as tantalum
Copper Production
187
188
4 Electrowinning of Metals
Pyrometallurgical
Hydrometallurgical
Sulfide ores
Comminution
Leaching
Flotation
Pregnant leachate
Precipitation
Concentrate
Solvent
extraction
Cement copper
Copper-rich electrolyte
Smelting
Electrowinning
Converting
Blister copper
Copper cathodes
Anode
refining and
casting
Copper anodes
Electrorefining
Copper cathodes
Fig. 14
Cu2+
Cu2+
Ni2+
Ni2+
Fe3+
Zn2+
Bi3+
Pb2+
NiO Au
Ag Se
Cathode
Anode
Electrolyte
Principle of copper
electrorening.
Fig. 15
Fe3+
Zn2+
Bi3+
SO42
PbSO4
Anode slime
slow
189
190
4 Electrowinning of Metals
(21)
(22)
The reactions (20) to (22) form the copper equilibrium on the electrode surfaces.
Concentration of Cu(I) on the cathode
surface affects the deposition rate. The
maximum net rate of Cu+ production is
at about 50 mV versus Cu/CuSO4 and at
higher overpotentials it decreases. Disturbing the Cu(II)Cu(I)Cu equilibrium
can cause the formation of copper powder,
but this is more a problem on the anode. For the current densities commonly
used in electrorening, the cathode overpotential is between 50 and 100 mV. The
system is mainly charge transfer controlled
and the effect of mass-transfer polarization is small. If Cu(I) concentration on the
cathode surface decreases, mass-transfer
polarization will increase, causing more
uneven deposit.
In order to produce a cast anode that
is dimensionally consistent and smooth
surfaced, some oxygen must be present
in the molten copper. During cooling,
this oxygen forms cuprous oxide (Cu2 O).
During the electrorening process, Cu2 O
dissolves (23)
Cu2 O + H2 SO4 = CuSO4 + H2 O + Cu
(23)
Because of this chemical dissolution,
copper builds up in the electrolyte. In
order to maintain a constant copper concentration in the electrolyte, continuous
to the electrodes. The Outokumpu system uses polymer concrete cells with 80
permanent cathodes at 100-mm spacing
[22]. Jumbo cells are very wide cells (up
to 28 m) with several electrode racks, each
rack having 40 electrode pairs. The racks
are connected in the usual Walker current feed design. The electrolyte ow is
from one side of the cell to the other
[23].
The anodes remain in the electrolytic
cells until they are almost completely
dissolved, usually for two batches of
cathodes. They are then replaced with
new ones. Typically, 1318% of the
Anode furnace
Anode
scrap
Anode
scrap
Anodes
Starting sheet
cells
Starting
sheets
Electrorefining
cells
Anode
slime
Anode
slime
Electrolyte
bleed
Liberator
cells
Precious metal
recovery
Decopperized
electrolyte
Precious metals
As, Sb, Bi
removal cells
Electrolyte
Evaporation
Nickel sulfate
Fig. 16
Cathodes
191
192
4 Electrowinning of Metals
Tab. 5
Element
Anode concentration
variation
[g g1 ]
Attainable
concentrations
[g g1 ]
ASTM
B115-91,
grade 1, max.
[g g1 ]
EN1978:1998,
Cu-CATH-1,
max. [g g1 ]
Cu
Se
Te
Bi
Total Se+Te+Bi
Cr
Mn
Sb
Cd
As
P
Total Cr+Mn+Sb+Cd+As+P
Pb
S
Sn
Ni
Fe
Si
Zn
Co
Total Sn+Ni+Fe+Si+Zn+Co
Ag
O
98.599.6%
103000
Balance
<0.2
<0.3
0.030.3
Balance
2
2
4
1001500
1004000
0.51.0
0.20.7
1004000
104000
1005000
0.10.3
3.06.0
<0.2
<0.30.5
0.51.0
5
15
5
10
10
1003000
10004000
4.015
25
Balance
2
2
2
3
15
5
15
10
20
25
of the controlling impurity and the impurity concentration that is tolerable in the
electrolyte. The purication of the bleed
stream is done in three sequential steps.
The excess copper is rst removed and
recovered, usually by electrowinning in
liberator cells. Then As, Sb, and Bi are
removed via electrowinning to an impure
Cu deposit. As the copper concentration
decreases, the deposition potential gradually increases until rst bismuth and
then antimony are reduced to the elemental form at the cathode. Because arsenic
deposits at or near the hydrogen potential, the electrical energy requirements are
high, and the formation of highly toxic hydrogen arsenide (arsine, AsH3 ) may occur.
After the removal of arsenic, the nickel sulfatesulfuric acid solution passes to nickel
recovery and acid treatment. Nickel is conventionally recovered by the evaporation of
the electrolyte to concentrate and crystallize the nickel sulfate [25].
The insoluble anode impurities are
collected as anode slimes from the bottom
of the cells and from washing the scrap
anodes. The slimes are treated in a
precious metals plant for recovery of gold,
silver, and other valuable by-products. In
conventional slime treatment, the rst
step is usually the decopperizing in
an autoclave at high temperatures and
pressures in sulfuric acid media. The
ltrate contains tellurium and chlorides.
Tellurium is removed by copper, after
which the ltrate returns to electrolysis.
The slimes are smelted to Dore metal and
cast to anodes. The anodes are processed
by silver electrolysis. The slime of silver
electrolysis contains gold and PGMs.
The slime is collected and leached with
hydrochloric acid and a strong oxidant. The
dissolved gold is precipitated with SO2 gas,
and the PGMs can be cemented.
193
194
4 Electrowinning of Metals
Typical voltage components in a
copper electrorening cell [5, p. 603]
Tab. 6
[5, p. 601].
= 1.1 [1.0 0.007 (T 55)
Potential source
Electrolyte resistance
Cathode overpotential
Anode overpotential
Contact resistance
Total
Overpotential
range [V]
0.110.13
0.040.08
0.030.06
0.010.02
0.190.29
(24)
4.1.3.1.3
(a)
(b)
(c)
(d)
195
196
4 Electrowinning of Metals
Fig. 18
Leaching
Pregnant
leachate
Raffinate
Solvent
extraction
Loaded
organic
Stripped
organic
Stripping
Copper-rich
electrolyte
Spent
electrolyte
Electrowinning
Copper cathodes
acid. The electrowinning is done in covered air-sparged cells using lead anodes
and titanium cathodes. Current density in
electrowinning is high, 430650 A m2 .
The cell voltage is 2.5 V and current efciency over 90% [32].
4.1.3.3 Halide-based Copper
Electrowinning
Chloride leaching of copper suldes has
received a great deal of attention in
past decades. The chloride processes have
distinct advantages over sulfate-based processes. The chloride systems have faster
kinetics, making atmospheric leaching
197
198
4 Electrowinning of Metals
Metclor cells with DSA-type anodes, hollow titanium cathodes, and cation-selective
membranes. Copper is deposited on the
cathode and chlorine is evolved on the
anode. Sodium ions transport through
the membrane to the catholyte. The operating principle of the Metclor cell was
quite close to that of the chlor-alkali diaphragm cell. Copper was deposited at
the current density of 1500 A m2 . The
spent catholyte contained cupric, cuprous,
and sodium ions. Chlorine generated at
the anode was recovered and used to
reoxidize the cuprous chloride generated in the catholyte during electrowinning. The recovered anolyte was depleted
from copper and used again as anolyte
[37].
The Intec Copper Process is a chloridebased method of converting copper suldes to high-purity copper and elemental
sulfur. The main components of the leaching solution are sodium chloride and
bromide. The lixiviant, BrCl2 , also known
as Halex , is produced anodically in the
electrowinning cell. The leaching is carried
out at atmospheric pressure using air as
the oxidant. The process utilizes a threestage countercurrent leaching at 8085 C.
Copper is leached in stages 1 and 2 and
gold is leached in stage 3. The pregnant
solution is puried in several stages. First,
all the copper is converted to cuprous by
contact with copper powder. Second, silver is removed by an amalgam process
using mercury and aluminum additions.
The nal step is neutralization with limestone to pH 4.04.5 to precipitate the
residual impurities. The puried cuprous
chloridesodium halide solution contains
80 g L1 Cu+ , 250 g L1 NaCl, 30 g L1
CaCl, and 30 g L1 NaBr at 75 C. The solution is fed to a diaphragm cell where
Cu dendrites are deposited on corrugated
titanium cathode. The depleted catholyte
Nickel Production
199
200
4 Electrowinning of Metals
+
9
10
6
8
10
3
7
2
1
5
10
10
10
201
202
4 Electrowinning of Metals
203
204
4 Electrowinning of Metals
Anolyte
overflow
Anode
Cathode
Catholyte distribution
Diaphragm frame
Fig. 20
also occurs to yield small quantities of hydrogen at the cathode and oxygen at the
anode.
Falconbridge Nikkelverk A/S in Kristiansand, Norway, uses the chlorine leach
and electrowinning process. The matte is
rst ground and then chlorine leached to
dissolve the contained nickel. The chlorine
leach solution is oxidized with chlorine
gas and neutralized with nickel carbonate to precipitate iron hydroxide. The
ltrate is pumped to solvent extraction
of cobalt. The cobalt solution is sent to
cobalt electrowinning. The rafnate contains 220 g L1 Ni and it is diluted to
85 g L1 Ni for Pb and Mn removal, and
after that, it is diluted to 60 g L1 Ni for
electrowinning. Typical operating parameters for electrowinning are current density
of 220 A m2 and current efciency of
9899%. The catholyte contains 60 g L1
Ni and the anolyte contains 54 g L1 Ni
and the electrolyte temperature is 60 C.
The renery produces both regular nickel
cathodes and Falconbridge Crowns . The
regular cathodes are produced by plating
onto nickel starting sheets. Crowns weighing 3060 g are round pieces produced by
plating nickel onto nonconductive cathode
blanks with evenly spaced conductive spots
[52, 53].
Nickel electrowinning is done in diaphragm cells using DSA anodes and
nickel starting sheets. Each cell contains
46 anodes and 45 cathodes with an anode
spacing of 145 mm. In the Falconbridge
system, the anodes are enclosed in a
polyester diaphragm bag to collect the chlorine gas generated in the anode reaction.
Figure 21 shows the Falconbridge nickel
electrowinning cell design. The chlorine
gas is channeled to a vacuum hood tted
to the top of the anode. Ducts, connecting the top of the hood with a manifold
that runs alongside the tank, allow the
205
206
4 Electrowinning of Metals
Fig. 21
207
208
4 Electrowinning of Metals
4.1.5
Zinc Production
Electrodeposition of Zinc
In principle, zinc is more difcult to electrowin than other base metals because the
equilibrium potential of zinc is considerably lower than that of hydrogen. The
evolution of hydrogen from an acidic zinc
sulfate solution is thermodynamically favored over zinc deposition. Zinc has high
hydrogen evolution overpotential, and this
makes the electrolytic production of zinc
possible. The initial deposition of zinc on
the fresh cathode surface is inuenced by
simultaneous zinc deposition and hydrogen evolution. The overpotential must be
high enough for zinc deposits to nucleate.
Impurities, sometimes even at concentrations as low as the parts per billion level,
could decrease hydrogen evolution overpotential and thus partly or entirely stop the
electrodeposition of zinc. The electrolyte
impurities can be grouped into four groups
on the basis of their effect [58]:
4.1.5.1
1. Ag, Cu, and Cd are capable of codepositing with zinc at their limiting current
density. They will induce zinc redissolution by forming a cathodic area.
2. Ni, Co, and Fe codeposit with Zn. They
will lower the hydrogen overvoltage
and initiate cyclic zinc deposition and
redissolution.
3. As, Sb, and Ge act as catalyst for
hydrogen evolution. They will increase
the critical current density at which
cathode formation begins, but once zinc
has deposited, they will not cause zinc
redissolution.
4. Al and Pb will not cause zinc
redissolution.
The hydrogen overvoltage is also inuenced by the condition of the surface of
the electrodes. The starter sheets must be
smooth as smooth surfaces lead to higher
hydrogen evolution overvoltages. For this
209
210
4 Electrowinning of Metals
7090 to the cathode surface [61]. Stripping of cathode zinc is more difcult with
basal or vertical orientations.
Energy Consumption
In electrowinning processes using sulfuric
acid solutions, the main anodic reaction is
the decomposition of water with evolution
of oxygen and formation of hydrogen
protons. The largest component of the
cell potential is the decomposition voltage
of the electrolyte. The polarization of the
anode strongly affects the cell potential,
because the evolution of oxygen always
needs certain overpotential. Table 7 lists
factors affecting the cell voltage which is
a sum of thermodynamical equilibrium
potentials of decomposition of water and
zinc deposition, electrode overpotentials,
and ohmic potential drops.
The current density is between 400
and 800 A m2 giving the cell voltage
3.33.5 V. The energy consumption is
about 3.13.3 kWh kg1 . The cathodic
current efciency is 88 to 93%, the
difference being due to H2 evolution, short
circuits, and current lost to the ground.
The anodic current efciency for oxygen
evolution is 98%, and the difference
is due to MnO2 and PbO2 formation
and Cl2 evolution. When operated at
a current density of 400 to 500 A m2
4.1.5.2
Tab. 7
Potential source
Thermodynamical voltage
Electrolyte resistance
Cathode overpotential
Anode overpotential
Hardware resistance
Total
Overpotential
[V]
1.99
>0.6
0.1
>0.6
0.2
3.5
zinc dust. Zinc will displace more noble elements by cementation reaction. The
puried ZnSO4 solution feed to the cellhouse contains 150175 g L1 Zn. It is
diluted by circulating the cell solution.
The electrowinning cell concentrations are
then 5060 g L1 Zn and 170210 g L1
H2 SO4 . The cells are undivided. The anodes are usually lead alloyed with silver and
the cathodes are aluminum plates. A typical tankhouse may contain 20 00035 000
cathodes and anodes with an additional anode in each cell. The area of the electrodes
varies from 1.1 to 1.5 m2 , but the trend is
to use larger electrodes of area 3.03.5 m2 .
The electrode spacing is 80100 mm. The
cathodes are grown for 2448 hours only
to prevent dendrite formation. After stripping, the zinc cathodes are melted and cast
into ingots.
The quality of zinc deposit depends on
the purity of the electrolyte. With a pure
electrolyte, it is possible to use higher temperatures, and thereby lower electrolyte
resistance and decrease electrode overvoltages. With an impure electrolyte, the
temperature must be lowered to 3035 C
to hinder hydrogen evolution caused by the
impurities. Temperatures below 30 C can
cause formation of calcium sulfate; temperatures above 40 C can increase lead
corrosion, and above 45 C can increase
sticking of the deposit.
Demands on the anode material are low
corrosion rates for a long lifetime, low oxygen overvoltage to minimize the specic
energy consumption, and good mechanical stability. The lead anodes are a good
compromise for anodes in zinc electrowinning. Lead is a low-cost material with good
corrosion resistance in strong corrosive
environments. The anodes used in the
zinc electrowinning process are usually
conditioned before use. The conditioning
process can be passivation in a KFH2 SO4
Noble Metals
211
212
4 Electrowinning of Metals
(32)
(34)
213
214
4 Electrowinning of Metals
Stainless steel
cathode
+ Silver anodes
Anode basket
Current feeder
Anode slime
Silver crystals
Silver crystals
Fig. 22
..
M obius and Balbach-Thum cells for silver rening.
215
216
4 Electrowinning of Metals
4.1.7
Other Metals
Lead
World production of rened lead was
6.39 million tons in 2002 [73]. Primary lead
production has usually two distinct operations: smelting lead concentrate to bullion
and rening the bullion. Depending on
the nature of the concentrate, the bullion
contains impurities, such as iron, copper,
nickel, cobalt, zinc, arsenic, antimony, tin,
bismuth, selenium, tellurium, silver, and
gold, as well as sulfur and oxygen. In conventional lead rening, the bullion from
the smelting is drossed by cooling the bullion to within 50 C of the solidication
point of lead. Iron, copper, and many impurities from the bullion are rejected by
crystallization partly or fully. For the most
part, conventional lead pyrometallurgical
rening is conducted in gas-red steel kettles in batch operations.
The electrolytic rening of lead bullion
anodes has been practiced since the beginning of the twentieth century. In practice,
the useful solutions have been restricted
to the soluble lead salts of uosilicic
acid (H2 SiF6 ), uoboric acid (HBF4 ), and
sulfamic acid (NH3 SO3 ). The most important of these is the uosilicic acid
electrolyte. All these electrolytes require
the use of additives for deposit control
and the relatively expensive electrolytes
cannot be regenerated. The electrolytes
and operating parameters are quite similar. The electrolytes contain 80100 g L1
Pb and 5070 g L1 free acid at temperature of 3040 C. The current densities
are between 160 and 200 A m2 and cell
voltage 0.30.4 V. Current efciencies are
high, 9698%. The energy consumption
is 0.110.13 kWh kg1 [41].
The size of the lead anodes is limited
owing to the metals low strength. The
4.1.7.1
217
218
4 Electrowinning of Metals
Pb(BF4 )2 + 2e = Pb + 2BF4
(37)
nickel and cobalt, is the preferred process, replacing the reduction roasting and
electrorening process route. Today, most
of the worlds cobalt is produced by electrowinning [46, 77]. In 2003, the world
production was 40 300 tons of cobalt metal,
metal powder, and compounds [78].
Many processes utilize electrowinning
for cobalt recovery. Cobalt can be deposited
as high-purity metal from sulfate or
chloride solutions. The electrowinning
solution must be puried from metallic
impurities. The following methods have
been used [5, p. 820]:
Removal of copper by cementation with
cobalt, deposition as sulde using H2 S,
solvent extraction, and electrowinning.
Removal of iron by oxidation with
air to Fe3+ followed by hydroxide
precipitation.
Removal of arsenic by crystallization
as As2 O3 or deposition with iron as
arsenate.
Removal of nickel by deposition as
amine-sulfate double salt or by solvent
extraction.
After solution purication, cobalt can
be deposited as Co(II) or Co(III) hydroxide. The cobalt solutions are often too
low in cobalt for direct electrowinning.
In this case, cobalt hydroxide is precipitated, thickened, and redissolved to cobalt
tankhouse electrolyte and then electrowon
[77]. The cobalt hydroxide has been reduced in electric furnaces to an alloy with
80% Co, 18% Fe, and 0.8% Ni, cast to
anodes, and electrorened. The process
was done in Hybinette cells, and the solution contained 60120 g L1 Co, pH was
3.348, and the temperature was 60 C.
The current density was 160240 A m2 ,
cell voltage was 1.21.8 V, and the current efciency was 95%. This process is
not used anymore because of the lack of
use anodes made from cobalt alloy containing 82% Co, 14% Si, and 4% Mn [46, 77].
Large amounts of cobalt are produced by chloride hydrometallurgy. The
large nickel-producing companies, Falconbridge in Norway and Sumitomo in Japan,
have similar processes for cobalt production [77].
Tin
Tin production in 2003 was 279 000 tons,
of which 11 900 tons was from secondary
sources [80]. Today, most tin is used as
a protective coating or as an alloy with
other metals. Tin is used as a coating for
steel cans, in solders for joining pipes or
electrical conductors, and in bearing alloys.
The common method for tin production is
reduction of cassiterite (SnO2 ) with carbon
or carbon dioxide. The reduction smelting
is followed by re rening of the molten
tin to remove iron, arsenic, antimony, and
copper. Even though the pyrometallurgical
treatment is fairly complicated, it is used
more often, as electrorening is a slow
process. Very impure tin concentrates
are rened by electrolysis after smelting
because re rening is not efcient for
producing high-purity metal from lowgrade raw material.
The high electrochemical equivalent of
Sn2+ favors the use of an electrometallurgical rening process. The main impurities Au, Ag, Cu, Bi, As, and Sb do not
dissolve under electrorening conditions,
but will remain in the anode slime. The
other impurities Ni, Fe, Zn, and Al can
be removed by a pyrometallurgical rening operation. Only lead lies close to tin
in the electrochemical series. There are
practical problems as simple and inexpensive sulfate or chloride electrolytes lead
to spongy or dendritic deposits, and the
deposit morphology can be changed only
by large additions of additives. Therefore,
4.1.7.3
219
220
4 Electrowinning of Metals
obtained without additives. However, current efciencies are very low, and the process often must be carried out at 90 C [81].
A special application for alkaline electrolysis is detinning of tinplate. Tinplate, a
coating of tin over a steel or iron sheet, is a
widely used packaging material. The plating can be done either by an electrolytic or
hot dipping process. The thickness of the
steel substrate is usually in the range of
0.150.55 mm. Because of improved coating technology, the tin content has gradually decreased and the average tin layer has
a thickness of 0.20.3 m. Extraction of tin
from tinned plates serves the double purpose of preparing steel scrap for remelting
and producing secondary tin. In sodium
hydroxide solutions, tin dissolves selectively from the steel substrate, mainly as
stannate, and it is further oxidized to stannite. Compact tin deposits can be produced
from alkaline electrolytes if the Sn2+ concentration is kept below 0.50.75 g L1 ,
but tin sponge is quite suitable for tin
extraction. The current density must be
carefully controlled to ensure primary
Sn2+ formation at the anode, to obtain satisfactory deposits, and to achieve 7585%
current efciency at the cathode [83].
Cadmium
Cadmium production is often connected
with zinc production. The impure cadmium metal produced in the purication
of zinc electrowinning electrolyte is dissolved in 150 g L1 sulfuric acid solution
and electrowon in a process similar to
that for zinc. The impure metals that
would disturb electrowinning, for example, Cu, Ni, As, are removed by cementation with impure cadmium. The solution
is neutralized with lime. The electrolyte
contains 140180 g L1 Cd, 3040 g L1
Zn, some chlorides, and animal glue as
additives. The electrolysis is done using
4.1.7.4
12. R. R. Moskalyk, A. Alfantazi, A. S. Tombalakian et al., Miner. Eng. 1999, 12(1), 6573.
13. A. Siegmund, D. Prengaman, Zinc electrowinning using novel rolled Pb-Ag-Ca
anodes, Hydrometallurgy 2003: 5th International Symposium, Honoring Professor Ian
M. Ritchie, Minerals, Metals and Materials Society (TMS), Vancouver, Canada, 2003,
pp. 12791288.
14. H. B. Beer, Improvements in or relating to
electrodes for electrolysis. Br. Pat. 1,147,442.
1965, p. 10.
15. H. B. Beer, Electrode having platinum metal
oxide coating thereon, and method of use
thereof. U.S. Pat. 3,711,385. 1970, p. 16.
16. M. Moats, K. Hardee, C. Brown, JOM 2003,
55(7), 4648.
17. D. L. Edelstein, Copper. U.S. Geological Survey, Minerals Yearbook 2002, 2004. On-line,
http://minerals.usgs.gov/minerals/pubs/
commodity/copper/.
18. P. J. Mackey, A. E. Wraith, Miner. Process.
Extractive Metall. 2004, 113, C25C37.
19. P. J. Bartos, Resour. Policy 2002, 28, 8594.
20. T. Robinson, J. Quinn, W. Davenport, Electrolytic copper rening 2003 world tankhouse operating data in Copper 2003 Cobre
2003, (Eds.: J. E. Dutrizac, C. G. Clement),
Canadian Institute of Mining, Metallurgy
and Petroleum, Santiago, Chile, 2003,
pp. 366.
21. M. Stelter, H. Bombach, Erzmetall 2001,
54(9), 432438.
22. T. Marttila, L. Palmu, H. Virtanen, Advanced
tankhouse systems by Outokumpu in Copper
2003, (Eds.: J. E. Dutrizac, C. G. Clement),
Canadian Institute of Mining, Metallurgy
and Petroleum, Santiago, Chile, 2003,
pp. 121135.
23. M. P. Amsden, R. M. Sweetin, D. G. Treilhard, J. Met. 1978, 30(7), 1626.
24. K., Rinne, Moderne Entwicklungen auf
dem Gebiet der Kupferrafnationselektrolyse, Elektrolyse der Nichteisenmetalle, Ed. Verlag Chemie, Weinheim, 1982, pp. 6792.
25. J. E. Hoffmann, JOM 2004, 56(7), 3033.
26. P. Stantke, JOM 2002, 54(4), 1922.
27. O. Forsen, The role of organic additives
on the quality of copper in electrorening
process, Copper 90. Rening, Fabrication,
221
222
4 Electrowinning of Metals
29. M. Stelter, H. Bombach, Investigations on
copper electrorening at high current densities, Copper 2003 Cobre 2003, Canadian
Institute of Mining, Metallurgy and Petroleum, Santiago, Chile, 2003, pp. 555567.
30. T. Robinson, J. Jenkins, S. Rasmussen et al.,
Copper electrowinning 2003 world tankhouse operating data, Copper 2003 Cobre
2003, Canadian Institute of Mining, Metallurgy and Petroleum, Santiago, Chile, 2003,
pp. 421472.
31. J. Peacey, X. J. Guo, E. Robles, Copper hydrometallurgy. Current status, preliminary
economics, future direction and positioning versus smelting, Copper 2003 Cobre
2003, Canadian Institute of Mining, Metallurgy and Petroleum, Santiago, Chile, 2003,
pp. 205222.
32. D. E. G. Maschmeyer, P. Kawulka, E. F. G.
Milner et al., J. Met. 1978, 30(7), 2731.
33. D. S. Flett, CIM Bull. 2002, 95(1065), 95103.
34. H. K. Lin, X. J. Wu, P. D. Rao, JOM 1991,
43(8), 6065.
35. J. Hoffmann, JOM 1991, 43(8), 4849.
36. G. E. Atwood, R. W. Livingston, Erzmetall
1980, 33, 251255.
37. R. F. Dalton, G. Diaz, R. Price et al., JOM
1991, 43(8), 5156.
38. J. Moyes, J. Lam, V. Santos, The intec copper
process: development of a new electrowinning cell in Copper 2003 Cobre 2003, (Eds.:
J. E. Dutrizac, C. G. Clement), Canadian Institute of Mining, Metallurgy and Petroleum,
Santiago, Chile, 2003, pp. 487508.
39. P. H. Kuck, Nickel. U.S. Geological Survey, Minerals Yearbook 2002, 2004. on-line,
http://minerals.usgs.gov/minerals/pubs/
commodity/nickel/.
40. D. J. MacKinnon, CIM Bull. 1989, 82(927),
6582.
41. R. Kammel, Nichteisen-Schwermetalle in
Chemische Technologie, Band 6, Metallurgie
(Ed.: K. Winnacker, L. Kuchler), Carl Hansen
Verlag, Munchen, 1973, pp. 215367.
42. A. I. Levin, V. M. Gabov, Sov. J. Non-ferr. Met.
1968, 9(1), 4044.
43. J. R. Boldt Jr., The Winning of Nickel,
Methuen & Company, London, 1967, p. 487.
44. D. K. E. Kerfoot, Nickel, Ullmanns Encyclopedia of Industrial Chemistry, Ed. John Wiley
& Sons, Weinheim, 2000.
45. L. S. Renzoni, R. C. McQuire, W. V. Barker,
J. Met. 1958, 10(6), 414418.
73.
74.
75.
76.
77.
78.
79.
80.
81.
82.
83.
84.
223
224
4 Electrowinning of Metals
4.2
6000
5000
4000
3000
2000
1000
0
1970
1975
1980
1985
1990
Year
1995
2000
2005
2010
Fig. 1
Tab. 1
Brazil
Venezuela
United States
Canada
France
Germany
Norway
Spain
Oceania
Cents
[kWha ]
MWh/longb
[ton]
1.89
1.15
2.10
1.05
1.96
3.03
1.70
3.40
1.29
16.1
15.7
15.8
15.6
14.7
16.0
16.6
16.0
15.7
225
4 Electrowinning of Metals
Bauxite
(impure aluminum oxide)
Calcination
(thermal
decomposition)
Crushing
Carbon
anodes
AI2O3
Electrolysis
in Hall Hroult
cells
Grinding
Electric
power
Leaching in
autoclaves
Solid liquid
separation
Recycled NaOH solution
226
Precipitation
of hydrated
alumina
Solidliquid
separation
Alloying
Alloying
agents
"Red mud"
waste
Casting
Seed
crystals
Rolling/
extrusion, etc.
Hydrated
alumina
crystals
User
Schematic diagram of the steps entailed in producing aluminum from bauxite. From
Evans and DeJonghe [12].
Fig. 2
Studs
Baked carbon
Electrolyte
Shell
Carbon
Anode
Metal
Cathode
Studs
Baked carbon
Electrolyte
Shell
Carbon
Cathode
Metal
Anode
Point
breaker
feeder
Anode
Electrolyte
Metal
Cathode
Shell
227
228
4 Electrowinning of Metals
F + 2AlF6 3 + Al2 O3
= 3AlOF3 2 + AlF4
and
3F + 4AlF6 3 + Al2 O3
= 3AlOF5 4 + 3AlF4
Furthermore the electrolyte usually contains additives (AlF3 can be so regarded)
that are intended to reduce the vapor
pressure of the electrolyte, and therefore
vapor losses/emissions of uorides, or to
improve such properties as electrical conductivity, liquidus temperature, or density.
A representative additive is calcium uoride. For example cryolite plus 5 wt%
229
4 Electrowinning of Metals
1080
1060
1040
t
[C]
230
1020
1000
980
965.9
960
10
15
[Al2O3]
[wt%]
Fig. 4
(1)
0 now includes the enthalpy
where Htot
required to take alumina and carbon at
2.194ROS3 + 0.8686ROS4
(2)
where
ROS = Mass% Al2 O3 /Mass% Al2 O3
(3)
at saturation
1.0
a(Al2O3)
0.8
0.6
0.4
ROS1.5
ROS 2.77
0.2
Fourth-degree
polynomial
0.0
0.0
0.2
0.4
0.6
0.8
1.0
ROS
Fig. 5 Correlations (see text for equations) given by Haupin and
Kvande for the activity of alumina in HallHeroult electrolytes [17].
231
4 Electrowinning of Metals
Fig. 6 Overvoltages in a
HallHeroult cell versus
electrolyte alumina content.
From the work of Haupin [16].
1.4
1.2
Equilibrium potential
1.0
Volts
0.8
Anode surface overvoltage
0.6
0.4
Bubble overvoltage
0.2
0.0
1
4
5
%Al2O3
232
%Al2O3
5
15
0.15 V
0.35 V
0.25 V
1.334 V
F_Ho = 2.037 V
1.756 V
Bemf
Fig. 8
10
0.032 V
Cathode Concentration : OV
0.036 V
Anode Concentration : OV
0.466 V
C + O2 = CO2
1.026 V
1.222 V
Anode Surface : OV
2.248 V
Reversible potential
Volts
kWh/kg Al
Equivalent to 2.013 V
0.35 V
Heat losses, 6.250 kWh/kg Al
0.466 kWh/kg Al
233
4 Electrowinning of Metals
5
15
0.35 V
0.35 V ?
0.25 V ?
0.471 kWh/kg Al
1.126 V
0.15 V
Heat losses
3.533 kWh/kg Al
10
kWh/kg Al
EHo = 2.924 V
2.380 V
Bemf
Cathode overvoltage
Reversible potential
2.248 V
Anode overvoltage
0.10 V 0.032 V
234
Cell voltages and energy consumption for a cell equipped with an inert
(oxygen evolving) anode from the work of Haupin and Kvande [17].
Fig. 9
(4)
where a and b are constants and i is the anodic current density. However, as pointed
out recently by Yang et al. [18], there is
scatter in the published values, which
was attributed to experimental difculties.
Within the range i = 0.051 A cm2 , a
ranges from 0.2 to 0.75 while b ranges
from 0.09 to 0.6.
Modern electroanalytical techniques
have been used to investigate the anodic
reaction and Zhu and Sadoway [19] have
provided a list.
Sorensen and Kjelstrup [20] have suggested that the anodic reaction should be
considered as two reactions: a two-electron
process producing CO and a four-electron
process producing CO2 . The former was
thought to dominate at low overpotentials while the latter would be dominant
at the high current densities and overpotentials representative of the HallHeroult
cell. Furthermore, these authors suggested
that analysis of cell data should be made
using the full ButlerVolmer (BV) equation rather than just the one (anodic)
term represented by the Tafel equation.
Figure 10 shows the t of their two reaction BV equations to the experimental
data of Solli [21].
A long established method for measuring the CE is the PearsonWaddington
equation [22]:
CE = 100% 0.5(%CO)
(5)
1.9
1.8
Ecorr
[V]
900 C
1.6
1.4
1045 C
970 C
1.2
1.2
3
3.0
1
log (i )
[A cm2]
0
0.0
Fig. 10
235
4 Electrowinning of Metals
C + CO2 = 2CO
j0
[A cm2]
236
6
4
2
0
0
Al2 O3
[Wt%]
Fig. 11
237
1.36
0.89
1996
1995
1994
1993
1992
1991
1.60
1.40
1.20
1.00
0.80
0.60
0.40
0.20
0.00
Frequency of anode
effects (anode effects per cell
per day) at Alcoa versus year.
From the paper of Marks [28].
Fig. 12
1.53 1.60
1990
4 Electrowinning of Metals
1.0
2.0
3.0
4.0
I
[mA cm2]
6.0
Forward scan
From top
0.60 wt%
0.40 wt%
0.25 wt%
0.15 wt%
0.01 wt%
Reverse scan
From top
0.60 wt%
0.40 wt%
0.25 wt%
0.15 wt%
0.01 wt%
2000
1500
1000
500
(a)
0
2500
2000
I
[mA cm2]
238
1500
1000
500
0
0.0
(b)
1.0
2.0
3.0
4.0
E vs Al / AlF3
[V]
5.0
6.0
Fig. 13
239
4 Electrowinning of Metals
0.0
250
5.0
10.0
15.0
I
[mA cm2]
200
150
100
50
0
(a)
50 000
CO
CO2
C
[ppm]
40 000
30 000
20 000
10 000
0
(b)
1000
800
600
C2F6
80
60
400
40
200
20
0
0.0
(c)
100
CF4
5.0
10.0
E /Al/AlF3
C (C2F6)
[ppm]
C (CF4)
[ppm]
240
0
15.0
Fig. 14
Cell voltage
4.9
4.8
4.7
4.6
4.5
3
10
30
100 300 1000 3000
Time (seconds before anode effect)
Fig. 15
1.0E-06
Anode effect
Partial pressure
[ Torr]
1.0E-07
28 (CO)
1.0E-08
69 (CF4 /C2F6)
44 (CO2)
1.0E-09
50 (CF4 /C2F6)
60 (COS)
1.0E-10
25 (CF4)
1.0E-11
102
104
106
108
110 112
Time
[min]
114
116
118
120
Fig. 16
241
242
4 Electrowinning of Metals
as those discussed earlier. One notable result of this investigation was the detection
of COF2 starting a few minutes before the
anode effect and continuing throughout
the effect.
Utne and coworkers [36] measured the
emissions of carbonyl sulde from industrial cells during normal operation.
High levels of COS were found in the
undiluted/unoxidized anode gas; amounts
corresponded to 3.310.2 kg tonne1 of
aluminum produced. Most of the sulfur
emitted from the cells appears to originate
from sulfur in the anodes. However, the
COS is oxidized as the anode gasses leave
the cell. After the gasses pass through dry
and then wet scrubbing, the COS content
corresponds to 0.210.26 kg tonne1 Al.
Hyland et al. [37] have recently reviewed sulfur and uoride emissions
(including particulate emissions) from
HallHeroult cells and concluded that operational changes over the past few years,
such as a tendency toward lower ratio
(more volatile) electrolyte may have made
emission control more difcult.
Gas Bubbles in HallHeroult Cells
Many effects of gas bubbles released at
electrodes (on electrolyte ow, mass and
heat transport, conduction, etc.) have been
well studied in the past. A text with an
extensive treatment of this topic is that of
Hine [38]. However, in HallHeroult cells
these effects are worthy of special mention
because the relatively high current density,
of the order of 1 A cm2 , and temperature
make the volumetric gas evolution rate
from the anode large. Furthermore, difculties of measurement on actual cells
mean less knowledge of these effects than
in many other electrochemical cells. Finally, one effect of the bubble is to make
the task difcult in reducing the enormous
4.2.4.4
Fig. 17
Cell resistance
[]
0.5
Bath Only
3 LS
0.4
6 LS
0.3 6 LS + 44 SS
0.2
0.1
0
Interelectrode gap
[cm]
that the spheres introduce a xed additional resistance into the interelectrode gap
independently of the size of that gap. These
investigators found that this additional resistance was proportional to the volume of
the bubbles and nearly independent of
their size and shape.
From the description of bubble growth
and release in the penultimate paragraph
one would expect individual bubbles to
be detectable, for example, in the cell
voltage trace for small anodes, such as
those used in laboratory cells. Aaberg and
colleagues [41] used a cell with a 10-cm diameter graphite anode and measured both
cell voltage and electrolyte level as a function of time. Both these variables showed
periodic uctuations and the latter uctuation enabled the bubble volume at release
to be measured. Bubble release frequency
was found to increase with increasing current density, ranging from 0.13 to 3.2 Hz.
Bubble volume at release also increased
with current density but only slightly.
Simultaneous measurement of the cell resistance by current interruption yielded
resistances for the anodic bubble layer of
0.51.5 cm2 .
Wang and Tabereaux [42] too used cell
voltage uctuations (for a 15-cm diameter
anode) to study bubble release. Figure 18
shows the voltage trace for an anode after
243
4 Electrowinning of Metals
Voltage traces for a
new anode (a) and an older
(15-cm diameter) anode (b) as
reported by Wang and
Tabereaux. The release of
individual bubbles can be seen
in the former trace [42].
Fig. 18
Cell voltage
[V]
4.4
4.3
4.2
4.1
4.0
(a)
12
Time
[seconds]
16
20
Cell voltage
[V]
3.90
3.85
3.80
12
15
Time
[seconds]
(b)
1600
Interpolar resistance
[]
244
Close-packed bars
Flat plates
1200
R=
800
ro ACD
400
ACD
[cm]
245
246
4 Electrowinning of Metals
The arrangement of
busbars and anode rods for six
end riser cells mathematically
modeled by Lympany and
Evans. The cells are in two
potlines and each cell has 18
anodes [50].
Fig. 20
Fig. 21
247
248
4 Electrowinning of Metals
0.3
0.25
Sediment velocity
[cm s1]
Fig. 22 Results of an
investigation, by water
modeling, of Kalgraf and
Trklep, into transport of
alumina particles (sediment)
along the bottom of a cell. The
parameters on the curves are
the diameters (millimeters) of
nylon particles (simulating
alumina particles). Solid lines
are calculated curves [64].
0.2
0.85
0.33
0.15
0.1
0.05
0.85
0.33
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Water velocity
[cm s1]
249
250
4 Electrowinning of Metals
various unstable modes are linked to distinct Fourier components of the vertical
magnetic eld distribution, allowing more
extensive, yet still practically simplied,
application of this approach.
In more computationally oriented work,
Segatz and Droste [76], have examined
the dependence of instabilities on design
and operating parameters. Figure 23 is
representative of their results; the growth
rate of various modes of oscillation is
plotted against cell current. Below a
certain current (approximately 57 kA in
the present case) interface waves have
essentially zero growth rate but beyond this
current, instability occurs. Interestingly,
the form of the unstable oscillation
changes as the current is increased. In
a similar way, but simplied by the use
of the shallow water model (described
in the previous section), Droste and
colleagues [77]) calculated results not very
different from those from the more
complicated 3D model.
0
50
150
200
0.06
0.06
Growth rate
[l s1]
A
B
0.04
0.04
C
D
0.02
0.02
0.00
0
50
100
Current
[kA]
150
0.00
200
Fig. 23
251
4 Electrowinning of Metals
0
10
20
30
40
50
60
70
80
90
100
by
[G]
bx
[G]
Voltage
[mv]
bz
[G]
252
200
200
400
400
10
10
10
10
10
10
10
10
15
15
15
15
0
10
20
30
40
50
60
Time
[seconds]
70
80
90
100
Fig. 24
253
4 Electrowinning of Metals
1000
Temperature
[C]
254
990
980
Measured
Calculated
970
0
12
Time
[hours]
16
20
Comparison of the results of the mathematical model of Tabsh and coworkers for the
thermal behavior of a HallHeroult cell with measurements [90].
Fig. 25
24
Cell voltage
[V]
6.0
5.5
AIF3 added
5.0
Simulated
Measured
4.5
Bath temperature
[C]
4.0
980
Measured
940
920
Superheat
30
Bath ratio
Simulated
960
12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44
Simulated
20
Measured
10
0
1.20
1.15
1.10
1.05
1.00
0.95
Simulated
Measured
12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44
Al2O3 concentration
[%]
Bath ledge
[cm]
20
Simulated
15
10
5
6
5
Simulated
Measured
3
2
12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44
Time
[hours]
Fig. 26
255
4 Electrowinning of Metals
10 cm gap
measured shape
initial shape
25.00
0.19
20.00
0.19
0.20
0.23
0.28
15.00
Y
[cm]
256
0.39
0.54
0.60
0.71 0.76 0.75 0.75
10.00
5.00
0.00
5.00
10.00
15.00
20.00
25.00
30.00
X
[cm]
Fig. 27
6.25 cm s1
Fig. 28
257
258
4 Electrowinning of Metals
Nearhorizontal anode (5)
ACD = 2 cm ; CD = 0.9 A cm2 ; submergence = 2.5 cm; 0.5 vol% butanol
6.25 cm s1
Results similar to those of Fig. 28 but with the bottom of the anode inclined at 5 to the
horizontal [44, 45].
Fig. 29
Bus bars
(copper)
To dc
power
Steel
shell
Stainless
steel
Collector
bars
Anode
Wood's metal
Hot water bath
35 cm
35 cm
Fig. 30
Stainless steel
The enormous electrical energy consumption of HallHeroult cells and their emissions of greenhouse gasses might be
moderated by the use of very different
anodes and cathodes. This has led to many
investigations of inert anodes and wettable cathodes. The former electrodes
are ones where the anodic reaction is
the generation of oxygen and therefore
the anode ought not to be consumed.
This anodic reaction has equilibrium potential (versus an aluminum cathode) of
2.22.3 V compared to 1.71.8 V for a carbon anode evolving CO2 . Consequently,
a switch to inert anodes would entail
an increase in electrical energy consumption unless other changes are made. One
such additional change would be the reduction in the ACD thereby reducing the
wasteful IR loss in conducting electricity
through this distance containing relatively
poorly conducting electrolyte intermixed
with bubbles. This reduction in ACD is
more likely to be possible with inert anodes because the location of the bottom of
the anode is known more precisely than in
the case of the consumable carbon anode.
Consequently, the anode can be brought
closer to the cathode without danger of
short circuits. This reduction of the ACD
can be further enhanced if materials wetted by aluminum are used in the bottom of
the cell. The normal lining of the bottom
of the cells is carbon, which is not wetted
259
4 Electrowinning of Metals
0.24
0.2
Corrosion rate
[g cm2h1]
260
0.16
0.12
0.08
0.04
6
8
Al2O3
[wt%]
10
12
Fig. 31
14
Concluding Remarks
261
262
4 Electrowinning of Metals
12. J. W. Evans, L. C. De Jonghe, The Production
and Processing of Inorganic Materials, TMS,
Warrendale, PA, USA, 2002.
13. F. McGravie, W. W. Robertson, S. T. Solinski, in 25th Conference on Metallurgists
(Eds.: E. Ozberk, D. W. Macmillan, R. I. L.
Guthrie), The Metallurgical Society of the
Canadian Institute of Mining and Metallurgy, Series 25-7, No. 2, 1986, pp. 319.
14. C. Vanvoren, P. Homsi, J. L. Basquin, in
Light Metals 2001 (Ed.: J. L. Angier), The
Minerals, Metals and Materials Society,
Warrendale, PA, 2001, pp. 221226.
15. E. Skybakmoen, A. Solheim, A. Sterten,
Metall. Mater. Trans. B 1997, 28, 8186.
16. W. Haupin, in Light Metals 1998 (Ed.:
B. Welch), The Minerals, Metals and Materials Society, Warrendale, PA, 1998,
pp. 531537.
17. W. Haupin, H. Kvande, in Light Metals 2000
(Ed.: R. D. Peterson), The Minerals, Metals
and Materials Society, Warrendale, PA,
2000, pp. 379384.
18. J.-H. Yang, Q.-X. Zhang, J. Thonstad, in
Light Metals 1999 (Ed.: C. E. Eckert), The
Minerals, Metals and Materials Society,
Warrendale, PA, 1999, pp. 453460.
19. H. Zhu, D. R. Sadoway, in Light Metals 1999
(Ed.: C. E. Eckert), The Minerals, Metals and
Materials Society, Warrendale, PA, 1999,
pp. 241246.
20. T. S. Srenson, S. Kjelstrup Sadoway, in
Light Metals 1999 (Ed.: C. E. Eckert, The
Minerals, Metals and Materials Society,
Warrendale, PA, 1999, pp. 415422.
21. L. N. Solli, Personal communication, May
26, 1998.
22. T. G. Pearson, J. Waddington, Discuss. Faraday Soc. 1947, 1, 307320.
23. M. M. R. Dorreen, M. M. Hyland, B. J.
Welch, in Light Metals 1998 (Ed.: B. Welch),
The Minerals, Metals and Materials Society,
Warrendale, PA, 1998, pp. 483489.
24. A. Kisza, J. Kazmierczak, J. Thonstad, in
Light Metals 1999 (Ed.: C. E. Eckert), The
Minerals, Metals and Materials Society,
Warrendale, PA, 1999, pp. 423429.
25. J. Marks, R. Roberts, V. Bakshi, in Light
Metals 2000 (Ed.: R. D. Peterson), The
Minerals, Metals and Materials Society,
Warrendale, PA, 2000, pp. 365371.
26. A. T. Tabereaux, J. Metals 1994, 11, 3234.
27. J. Thonstad, T. A. Utigard, H. Vogt Dolin,
in Light Metals 2000 (Ed.: R. D. Peterson),
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
49.
50.
51.
52.
53.
54.
55.
56.
57.
263
264
4 Electrowinning of Metals
76. M. Segatz, C. Droste, in Light Metals 1994
(Ed.: U. Mannweiler), The Minerals, Metals
and Materials Society, Warrendale, PA,
1994, pp. 313322.
77. Ch. Droste, M. Segatz, D. Vogelsang, in
Light Metals 1998 (Ed.: B. Welch, The
Minerals, Metals and Materials Society,
Warrendale, PA, 1998, pp. 419428.
78. J. Descloux, M. Flueck, M. V. Romerio, in
Light Metals 1994 (Ed.: U. Mannweiler),
The Minerals, Metals and Materials Society,
Warrendale, PA, 1994, pp. 275281.
79. J. P. Antille, M. Flueck, M. V. Romerio, in
Light Metals 1994 (Ed.: U. Mannweiler), The
Minerals, Metals and Materials Society,
Warrendale, PA, 1994, pp. 305312.
80. J. P. Antille, J. Descloux, M. Flueck, in
Light Metals 1999 (Ed.: C. E. Eckert), The
Minerals, Metals and Materials Society,
Warrendale, PA, 1999, pp. 333338.
81. P. A. Davidson, Eur. J. Mech., B/Fluids 1994,
13, 1532.
82. P. A. Davidson, R. I. Lindsay, in Light Metals
1997 (Ed.: R. Huglen), The Minerals, Metals
and Materials Society, Warrendale, PA,
1997, pp. 437442.
83. P. A. Davidson, R. I. Lindsay, J. Fluid Mech.
1998, 362, 273.
84. P. A. Davidson, R. I. Lindsay, Mater. Sci.
Technol. 2000, 16, 475.
85. W. E. Wahnsiedler, in Light Metals 1987
(Ed.: R. D. Zabreznik), The Minerals,
Metals and Materials Society, Warrendale,
PA, 1987, pp. 269287.
86. V. Potocnik, in Light Metals 1989 (Ed.:
P. G. Campbell), The Minerals, Metals and
Materials Society, Warrendale, PA, 1989,
pp. 227235.
87. M. Segatz, D. Vogelsang, C. Droste, in Light
Metals 1993 (Ed.: S. K. Das), The Minerals,
Metals and Materials Society, Warrendale,
PA, 1993, pp. 361.
88. M. Segatz, Ch. Droste, D. Vogelsang, in
Light Metals 1997 (Ed.: R. Huglen), The
Minerals, Metals and Materials Society,
Warrendale, PA, 1997, pp. 429435.
89. O. Zikanov, A. Thess, P. A. Davidson,
et al.,Metall. Mater. Trans. B 2000, 31, 1541.
90. Tabsh, M. Dupuis, A. Gomes, Light Metals
1996 (Ed.: W. Hale), The Minerals, Metals
and Materials Society, Warrendale, PA,
1996, pp. 451457.
91. H. Q. Tang, N. Urata, C. M. Read, in Light
Metals 1998 (Ed.: B. Welch), The Minerals,
92.
93.
94.
95.
96.
97.
98.
99.
100.
101.
102.
103.
104.
105.
265
267
Electrowinning of Other
Inorganic Compounds
Carl Heinz Hamann
..
Institut f ur Reine und Angewandte Chemie, Oldenburg, Germany
..
Thorsten R opke
Hydrogen Institute of Applied Technologies, Schwerin, Germany
Peter Schmittinger
Formerly Degussa AG, Germany
5.1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.2
5.2.1
5.2.2
5.2.3
5.2.3.1
5.2.3.2
5.2.3.3
5.2.3.4
5.2.3.4.1
5.2.3.4.2
5.2.3.4.3
5.2.3.5
5.2.3.6
5.2.3.6.1
5.2.3.6.2
5.2.3.6.3
5.2.3.7
5.2.3.7.1
5.2.3.7.2
5.2.3.7.3
5.2.4
269
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
269
269
270
272
272
273
276
278
278
280
280
281
283
284
286
287
287
287
288
.
.
290
291
268
5.2.5
5.2.5.1
5.2.5.2
5.2.5.3
5.2.5.4
5.2.5.5
5.2.6
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
292
293
293
294
295
296
296
5.3
5.3.1
5.3.2
5.3.3
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
297
297
297
297
5.4
5.4.1
5.4.2
5.4.3
5.4.3.1
5.4.3.2
5.4.3.3
5.4.3.4
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
299
299
300
301
301
302
304
305
5.5
5.5.1
5.5.2
5.5.3
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
305
305
305
306
5.6
5.6.1
5.6.2
5.6.3
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
307
307
307
307
5.7
5.7.1
5.7.2
5.7.3
5.7.4
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
308
309
309
310
312
312
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
269
5.1
Introduction
5.2
270
34%
20%
6%
6%
4%
30%
18%
18%
15%
12%
8%
5%
24%
chlorate ClO3 , perchlorate ClO4 , and chlorine dioxide ClO2 . With the exception of
chlorite, all chlorineoxygen compounds
mentioned can be obtained as a direct
product of electrolysis, but electrolytic production is not dominating in all cases.
There are chemical pathways, too, using
molecular chlorine and caustic alkali as
basic educts which have been produced
electrolytically before.
Chlorineoxygen
compounds
are
used besides
chlorine
itself predominantly for bleaching and disinfection
purposes. With water they form solutions of hypochloric acid, these solutions
showing oxidizing, chlorinating, or hydrochlorinating properties, depending on
the pH-value established [10].
Since it became clear that the direct
use of chlorine in the bleaching of pulp
and paper leads to highly poisonous
chlorinated organic species in the waste
water, chlorine dioxide is used instead
in the so-called ECF (Elemental Chlorine
Free) process. The competing process is
the TCF (Total Chlorine Free) process,
which uses peroxides [11].
Chlorine dioxide is produced from
sodium chlorate in situ, the latter becoming the most important technical chlorineoxygen compound in this way (production in the year 2002: 2.8 106 tons).
Ninety-ve percent of the chlorate is used
in the described way; another usage is, for
instance, the production of herbicides.
Perchlorate as a strong oxidant is widely
used. The usage in pyrotechnics is only
one example.
5.2.2
(1)
(2)
Using iron as cathode material, hydrogen is split off from the water molecule
2H2 O + 2e H2 + 2OH
(3)
Considering reactions (1) and (2), chlorine and sodium amalgam are formed in
the electrolysis cell. The amalgam can be
extracted and undergoes a decomposition
reaction in a separate reactor (decomposer,
denuder). The electrocatalytic process is
2NaHgx + 2H2 O
2Na+ + 2OH + 2xHg + H2 (4)
So, the addition of Eqs. (1), (2), and (4)
gives the overall reaction of the so-called
amalgam process for the production of
chlorine and alkali from sodium chloride
2NaCl + 2H2 O
2NaOH + H2 + Cl2
(5)
(6)
3ClO + H2 O ClO
3 + 2Cl
2
3
+ 3H+ + O2 + 3e
(7)
4
From chlorate, at the end, perchlorate
can be produced anodically
ClO
3 + H2 O ClO4 + 2H + 2e
(8)
(For reason of simplicity, we wrote H+
instead of H3 O+ .)
So, the sodium hydroxide generated will
be contaminated with chlorineoxygen
compounds, under rapid reduction of the
current efciency. The elemental chlorine
bubbling from the anode will contain
amounts of oxygen. Thus it is necessary to
prevent OH -migration from the cathode
to the anode in order to receive an
economic technical process. This can be
done by using divided cells, leading to
the so-called diaphragm and membrane
processes.
On the other hand, if reactions (1)
and (3) are performed in undivided cells,
chlorineoxygen compounds are accessible (see Sect. 5.2.5).
271
272
5.2.3
Pure brine
Chlorine gas
c
C
i
d
Hydrogen gas
Demineralized
Hg decomposer water
CW
Depleted
brine
CW
h
NaHgx
NaOH
NaHgx
g
Hg
f
2NaHgx + 2H2O 2Na+ + 2xHg + H2 + 2OH
Fig. 1
NaOH
Demineralized
decomposer
water
Brine
CI2 CI2 H2O
l
e
H 2O
c
b
h
a
H2O
H2O
H2
f
g
Hg
H2O
i NaOH
Anolyte
Uhde mercury cell. (a) Cell base; (b) anode; (c) cover seal; (d) cell cover;
(e) group adjusting gear; (f ) intercell busbar; (g) short-circuit switch; (h) hydrogen
cooler; (i) vertical decomposer; (j) mercury pump; (k) anode adjusting gear; (l) inlet
box; (m) end box, compare Fig. 1. Please note: adjusting gears (e) and (k) are
needed to maintain a distance of ca. 1.52.5 mm between anodes and the cathode.
Greater distances lead to an increase of ohmic losses. In case of distances too
small, the anodecathode system will be short circuited and the cell will be
damaged. In this case, the cell must be repaired and has to be short circuited in total
by a suitable switch (g). In case of the anodes (b), the structure described in the text
can be seen left, in the middle the substructure is displayed [3, p. 38, 13].
Fig. 2
273
274
h
a
Fig. 3
i
k
h
k
g
j
c
a
e
g
Mercury cell room (cross section, schematic). (a) Basement oor; (b) oor
drains; (c) cell supports with insulators; (d) supply pipes; (e) cells; (f ) decomposers;
(g) service walkways; (h) crane; (i) ridge ventilator; (j) ventilation air supply;
(k) windows/lighting [3, p. 43].
Fig. 4
cathode compartment that serves to impede OH -transport (see Sect. 5.2.2). This
is reenforced by a continuous ow of the
NaCl electrolyte into this compartment,
thus equalizing the effects of diffusion
and migration of OH -ions into the anode
compartment. Figure 5 shows a cut-away
view of this arrangement.
The cathodes themselves are made, for
instance, of steel mesh or woven steel
and the anodes are DSA in form of
expanded titanium meshes. The brine
(310320 g NaCl/L) enters the anode
compartment and completely covers the
anodes and the cathodes (called ngers
or tubes). The chlorine leaves the cell
through an outlet in the cell head. The
NaCl-brine
Chlorine
l
Cl2
Cl2
a
H2
H2
Na+
Cl
H2O
b
c
d
NaCl
NaCl
e
Cutaway view of a diaphragm cell. (a) Anode; (b) base plate protector;
(c) base plate; (d) current busbars; (e) cathode; (f ) deposited diaphragm; (g) anolyte
compartment; (h) catholyte compartment; (i) catholyte level; (j) hydrogen gas-room;
(k) anolyte level; (l) cell head. Not included: outlet for cell liquor and hydrogen;
compare text [3, p. 52].
Fig. 5
275
276
Hydrogen
Chlorine
c
d
f
g
h
Caustic cell
liquor
ELTECH Systems MDC cells. (a) Brine feed rotometer; (b) head sight glass; (c) cell
head; (d) cathode frame assembly; (e) tube sheet; (f ) base plate; (g) cathode tube; (h) base
plate protector; (i) DSA [3, p. 65].
Fig. 6
Cl2
NaOH + H2
Membrane
2Na+(H2O)x
Cl2
2NaCl
+
+2Na +
(1)
2e
Cl
NaCl
-brine
2H2O + 2e
H2 + 2OH
(3)
OH
H2O
277
278
Membrane
Electrode
Contact surface
Corrugated
strip
Electrode
NaClfeed
H2Ofeed
Fig. 8
Brine
Starting point of the
chlor-alkali process is usually solid NaCl
in form of rock salt, solar salt or prepared from solution-mined brine. These
materials of course contain impurities.
5.2.3.4.1
a
b
c
g
d
f
e
Fig. 9
279
280
Chlorine
Chlorine gas produced in the various electrochemical processes is saturated with water vapor at
high temperatures and may also contain
brine mist and traces of chlorinated hydrocarbons (originating from piping and
rubber-lined vessels), and is normally at
atmospheric pressure. Before the chlorine
can be used it must (in this order) be
cooled, puried, dried, compressed, and if
necessary liqueed.
The cooling can be performed indirectly
in heat exchangers (made of titanium), or
directly in packed towers, where the cooling medium water is sprayed into the top
and ows counter-currently to the chlorine. The cooling of course must not go
below ca. 10 C, otherwise solid chlorine
hydrate will be formed, as water is still
present. After cooling, water droplets and
brine mist are removed by ltering elements using glass wool or porous quartz
granules, alternatively electrostatic purication can take place. Organic impurities
can be adsorbed by carbon lters.
The drying of the chlorine is carried
out almost exclusively with sulfuric acid
of 9698 wt% in a multistage process (for
instance packed towers, counter-current
ow). After the drying, another demisting
step, now from sulfuric acid, is necessary.
Compression is carried out with the
usual mechanical means for this purpose.
Care must be exercised to prevent the
heat of compression from increasing the
temperature enough to ignite materials in
contact with the chlorine.
The quality of the compressed and
dried chlorine gas from the electrolysis
plant is sufcient for most applications,
for example, the production of EDC for
PVC. Liquefaction is necessary, if gaseous
components like oxygen, carbon dioxide,
nitrogen, or hydrogen have to be removed,
or if chlorine is stored as a liquid. In the
5.2.3.4.2
market (41 106 t/a) depended on the diaphragm process, 22% on mercury cells
and 21% on membrane cells, 5% others [1,
p. 248]. In 2001, these data had changed
to: chlorine capacity 53.5 106 t/a, diaphragm process 42%, amalgam process
19%, membrane process 36%, 3% others.
Considering the year 2005, the share of
the membrane process is estimated to be
45% [19].
Newly built membrane cell plants will
cost ca. 300 000 $ per metric ton per
day chlorine capacity (green-eld basis).
A comparison with the investment costs
for the other two varieties does not make
sense at the moment, because since
1991 only a few diaphragm cell plants
and no mercury cell plants were built
[3, p. 123].
Table 3 summarizes the advantages and
disadvantages of the three chlor-alkali
processes.
The last point in this comparison will
be the remarks use of mercury, costly
Mercury
process
Diaphragm
process
Membrane
process
3.11
2470
2.15
1705
2.15
1705
4
10
31003400
3.5
2.5
23002900
33.2
46
22002600b
8001000
200400
31003400
31003900
24003000
100
106
84
281
282
Product qualities: typical compositions of chlorine, caustic, and hydrogen [3, p. 118]
>99.9
Membrane
Puried
96.598
0.52.0
0.10.3
0.10.5
1.03.0
50.0
1.0
0.1
0.01
0.1
0.02
0.001
0.0015
0.0005
0.0007
0.0002
Nonea
Process
Mercury
9899
0.10.3
0.20.5
0.10.5
0.20.5
50.0
0.025
0.1
0.01
0.001
0.02
0.001
0.0015
0.0005
0.0007
0.0002
Nonea
0.001
9799.5
0.52.0
0.030.3
50.0
0.005
0.05
0.0005
0.0005
<0.001
0.001
0.0002
0.0005
0.0005
0.00001
0.00001
50.0
0.005
0.04
0.0001
0.001
0.002
0.0001
0.0001
0.0001
0.0004
0.0001
Nonea
>99.9b
>99.9
a <106 %.
Hydrogen gas from the mercury process contains mercury: 110 mg m3 , depending on the
purication process. The hydrogen gas from the other two processes is free of mercury.
Process
Advantages
Disadvantages
Diaphragm
process
Mercury
process
Membrane
process
283
14
Wastewater
Products
Mercury cell process
g Hg / t Cl2
284
12
0
1978
1980
1982
1984
1986
1988
1990
1992
1994
1996
Year
Fig. 10
Rev. H2
potential
2.5
ca.3.65 V
Current density
Amalgam
process
3.11 V
ca.2.4 V
Membrane
process
ODC, Ca-1.6V
Rev. O2
potential
Diaphragma
process
2.15 V
Electrode potential
+0.82
ca.2.6 V
+1.33
10 kA m2
0.64
0.8
f
e
d
1.78
1.94
c 2NaCl + 2x Hg + 2e 2NaHgx + 2Cl (2)
Fig. 11
285
0.8
V NHE
Electrode potential
Fig. 12
(3)
a.u.
Current density
286
H2O + e
OH + 1/2H2
Solution
Na
Graphite
(catalyst)
Na+ + e
NaHgx
Amalgam
Fig. 13
Closing Remarks
287
288
+ H2 O + 2e 2OH
(10)
NaOH
Membrane
2Na+(H2O)x
Cl2
2NaCl
+2Na+
2e
Cl
NaCl
brine
1/2O2 + H2O
+ 2e
2OH
OH
H2O
O2
Fig. 14
Fig. 15
ODC
Atmospherical
value
Catholyte
Hydrostatic
pressure
O2-gasroom
Gas pressure
289
290
O2
Catholyte
Gas
pocket
2CH3 OH + 2e H2 + 2CH3 O
(12)
(compare Fig. 12). In combination with
Eq. (9) then we have instead of Eq. (4)
2NaHgx + 2CH3 OH 2CH3 ONa
+ H2 + 2xHg
Hydrostatic
pressure
O2
Gas pressure
(13)
5.2.4
2Cl Cl2 + 2e
(14)
Cathode : 2H + 2e H2
(15)
Cathode :
+
1
2 O2 + 2H + 2e H2 O (18)
2HCl Cl2 + H2
(16)
Reaction (18) represents the oxygen reduction process in acidic solution, the
reversible potential then being +1.23 V.
This leads to a reversible cell voltage of
0.2 V. The theoretical energy usage per
tonne of chlorine in this way is unbeatable
160 kWh.
The ODC in this case may form together with the membrane an unit, that
is, the porous electrode is (including
the current distributor) directly attached
to the membrane surface. The electrocatalyst needed for oxygen reduction
in this case is located at the membrane/electrode boundary. Concerning the
anode, the nite gap principle may be
used. Figure 17 gives a scheme of the
principle.
The setup becomes possible because
the cathodic reaction product (water) can
easily be drained through the porous ODC
without affecting the water content of
the membrane. And there is no pressure
compensation needed due to the fact that
the ODC is not affected directly by the
hydrostatic pressure.
Producing chlorine from hydrochloric acid using ODCs in the described
291
292
Fig. 17
H2O
2H+
2HCI
1/2O2
ODC
Catalyst layer
Membrane
5.2.5
2H2 O + 2e H2 + 2OH
(1)
(3)
(6)
3ClO + H2 O ClO
3 + 2Cl
2
3
+ 3H+ + O2 + 3e
(7)
4
ClO
3 + H2 O ClO4 + 2H + 2e
(8)
the spacious variations of using the chlorine oxygen compounds for bleaching, disinfection or water treatment, information
on these items is only given in those few
cases, where it is needed for better understanding.
In case of electrolytic production of chlorine oxygen compounds, questions like
purication of brine and of products arise,
too. These questions are solved in analogy
to the means described in Sect. 5.2.3.4 and
shall not be repeated here.
Hypochlorite
According to Eqs. (1), (3), and (6),
hypochlorite will be formed in undivided
electrolysis cells fed with sodium chloride.
Combining the equations mentioned, we
nd as overall reaction
5.2.5.1
ClO2 + 2NaOH + H2 O2
2NaClO2 + O2 + H2 O
(22)
293
294
Cl2 + H2 O HOCl + H+ + Cl
(23)
The acid formed undergoes dissociation
HOCl ClO + H+
(24)
and ClO
3 is formed according to
+ 2H+
(25)
ClO
3 + 3H2 O + 6e Cl + 6OH
(27)
Both reduction processes are inhibited by addition of chromate (36 g L1
sodium chromate) to the cell liquor.
Rupture
disc
Sight
glass
Sight
Rupture glass
disc
Hydrogen to
hydrogen cooler
Reactor overflow
Degasifier
Riser
Electrolyte
cooler
Chlorate
reactor
Strong chlorate
feed tank
Electrolyzer
Strong chlorate to
strong chlorate cooler
Cell feed
Fig. 18
Example for a technical chlorate cell (Kvaerner Chemetics) [4, p. 532; 36].
ClO
3 + H2 O ClO4 + H2
(28)
295
296
(29)
But ClO2 can be produced by electrolysis, too. In this case, sodium chlorite
solution is introduced into the anode compartment of a divided cell and oxidized
anodically [4, p. 535].
5.2.6
Additional Aspects
(30)
Sodium ions are transferred into the cathode compartment via the membrane and
form the desired sodium hydroxide as before. The anolyte leaves the cell in the form
of a mixture of Na2 SO4 and sulfuric acid.
But the conventional membrane setup
described is only one possibility of a
sodium sulfate electrolysis. Other proposals include the use of an ODC or even an
HDA (Hydrogen Depolarized Anode) for
cell voltage reduction or the use of anion
exchange membranes instead of the cation
variety [7]. Last but not least electrodialysis
may be used: in a three compartment unit
the salt is converted into free acid and the
free base [40].
In the case of producing alcoholates or
other chemicals from alkali amalgam (see
5.3 Fluorine
Introduction
(31)
5.3.2
5.3.3
Electrochemical Reactions
Technical Process
Figure 19 shows an example for a classical uorine cell developed for the French
297
298
HF
e
d
f
c
b
g
Fig. 19
5.4
Introduction
Production
Consumption
Steam reforming of
natural gas or naphta
109m3
N
190
Partial oxidation
of heavy oil
120
Benzine
reforming
90
Coal
gasification
50
Synthesis of ethen
33
Chlor-alkali
industry
10
Other
Water electrolysis
1.3
Fig. 20
H
Y
D
R
O
G
E
N
Synthesis of
ammonia
109m3
N
200
Process heat
generation
150
Processing of
mineral oils
100
Synthesis of
methanol
25
FischerTropsch
synthesis
15
299
300
Electrochemical Reactions
System
Alkaline
Acidic
Ion-conducting
Cathode
2OH 1/2O2 + H2 O + 2e
3H2 O 1/2O2 + 2H3 O+ + 2e
O2 1/2O2 + 2e
Anode
(10) 2H2 O + 2e H2 + 2OH
(18) 2H3 O+ + 2e H2 + 2H2 O
(32) H2 O + 2e H2 + O2
(3)
(15)
(33)
(34)
1
O3 + H2
3
(35)
Technical Processes
Classical Water Electrolysis
The major part of water electrolyzers
operated today use alkaline electrolytes
and follow the bipolar design. Figure 21
represents the principle setup.
5.4.3.1
H2
H2
O2
Filter
Inlet pipe
Principal setup of a
classical water electrolyzer.
Fig. 21
301
H2
Conventional
V
3
Voltage
302
5 mm Asbestos separator
O2
2.23
2
H2
Advanced
1.67
1.23
1
0.5 mm
NiO-separator
0
0
4
6
8
Current density
10 kA m2
Fig. 22
In addition to a compared with a diaphragm better conductivity of the membrane, advanced electrocatalysis is applied
for cell voltage reduction: electrocatalyst
particles (for instance platinum or NiZn
alloy) are deposited in the boundary layer
adjacent to a porous electrode material [55].
Figure 23 shows the principle.
100 kW electrolyzers have been built
commercially, showing a cell voltage of
only 1.55 V at a load of 5 kA m2 . An additional advantage is a reduced volume of
the unit compared to the classical variety.
But membrane electrolysis suffers a severe drawback: the high costs for the solid
polymer electrolyte. If membranes being
SPE-membrane
O2
Contact
surface
H2
H+
Water
inlet
2H+ +
2e
H2
Catalyst
layer
Fig. 23
303
H2
Cathode
ZrO2-solid-electrolyte (YSZ)
Anode
O2
Electronic conductive
interconnection
material
10 mm
304
H2O(g)
Fig. 24
5.5 Peroxodisulfates
5.5
5.5.1
Introduction
Electrochemical Reactions
2SO2
4 S2 O8 + 2e
or
2HSO
4
S2 O2
8
(36)
+
+ 2H + 2e
(37)
305
306
Technical Processes
e
c
g
d
f
g
Fig. 25
b
e
Introduction
Electrochemical Reactions
13 00017 000 kWh per ton of H2 O2 (calculated as 100 wt%), as well as the necessity to use noble metals as anode material
and a yield of only 70%. So, due to
high costs this production line has become insignicant compared with catalytic
processes developed later, such as the anthraquinone process.
But there is another possibility to
generate peroxide by an electrochemical
process, namely, the indirect reduction
of oxygen (see this encyclopaedia Vol. 7,
or [24, p. 281]). In both acidic and
alkaline solutions, peroxide is formed as an
intermediate. In todays practice, alkaline
solution (NaOH) is used.
O2 + H2 O + 2e HO
2
+ OH
(E0 = 0.076 V)
(39)
HO
2 + H2 O + 2e 3OH
(E0 = 0.88 V)
(40)
Technical Process
307
308
O2
Diaphragm
O2
2OH
1/2O2
+ H2O + 2e
Trickle bed,
O2 + H2O + 2e
HO2 + OH
NaOH/H2O2
Anolyte
Fig. 26
Cr2 O2
4 + 2H
7 + H2 O2CrO
(41)
(42)
The standard redox potential of this process is 1.36 V. Equation (42) includes the
possibility to regenerate Cr(III) solutions
generated in using Cr(VI) as an oxidant.
The cathodic reaction is
6H+ + 6e 3H2
(43)
Manganese Dioxide
Manganese is contained in many minerals; they generally contain more than 35%
Mn, and consist of hydrated or anhydrous
oxides. Only 5% of the world production of manganese ore is consumed in
nonmetallurgy applications. From these
5% amounting to 1.35 106 tons per
year (1976) 0.5 106 tons per year were
associated with the manufacturing of dry
cell batteries and a similar amount was
used for the production of manganese
chemicals. These data are probably still
applicable, because worldwide ore consumption did not change remarkably.
Natural manganese dioxide used in
dry cells must be of battery grade.
This implies a content of 7085% MnO2
(4455% Mn) in the ore, going along with
a content of less than 0.05% concerning
metals more electronegative than zinc,
such as Cu, Ni, Co, or As (which
309
310
2+
+ 2H2 O MnO2 + 4H + 2e
(44)
The cathode reaction is
Mn
2H+ + 2e H2
(15)
Permanganate
In the preceding section, the production of manganese chemicals was mentioned. These chemicals refer to the oxidation states up to VI. The preparation
of potassium permanganate (oxidation
state VII) is performed using electrolysis.
The procedure starts with MnO2 ore
nely ground, which is subjected to
alkaline oxidative fusion, giving K2 MnO4 .
From this species, the permanganate
may be produced chemically by reaction
with elemental chlorine or with carbon
dioxide.
2K2 MnO4 + Cl2 2KMnO4
+ 2KCl
(45)
2MnO2
4 2MnO4 + 2e
(47)
Cathode reaction is
2H2 O + 2e H2 + 2OH
(3)
Spent electrolyte
Electrolyte
a
g
c
b
d
e
Fig. 27
311
312
Copper Oxide
(3)
(50)
und
Tendenzen, Chemie Ingenieur Technik, 2005,
pp. 13631376 Vol. 77.
8. (a) Chemical Week, February 9, 2005, p. 28.
(b) Chemical Week, February 19, 2003, p. 57.
9. H. Schultz, G. Bauer, E. Schachl et al., Potassium compounds in Ullmanns encyclopedia
of industrial chemistry, 6th ed., Electronic
Release, (Eds.: B. Elvers, S. Hawkins), WileyVCH, Weinheim, 1999, pp. 281328, Vol. 8.
10. H. Vogt, J. Balej, J. E. Bennett et al., Chlorine
oxides and chlorine oxygen acids in Ullmanns encyclopedia of industrial chemistry,
6th ed., Electronic Release, (Eds.: B. Elvers,
S. Hawkins), Wiley-VCH, Weinheim, 1999,
pp. 281328, Vol. 8.
11. info@cellchem.ekachemicals.com, 2003.
12. S. Trasatti, G. Lodi, in Elektrodes of Conductive Metal Oxides, part A, (Ed.: S. Trasatti),
Elsevier Scientic Publishing Comp., Amsterdam - Oxford - New York, 1980, p. 301 ff.
13. UhdeNora, Alkalichloridelektrolyse nach
dem Quecksilberverfahren, Brochure.
14. C. H. Hamann, W. Vielstich, Elektrochemie,
Wiley-VCH, Weinheim, 4. Auage 2005.
15. T. A. Davis, J. D. Genders, D. Pletcher, Ion
permeable membranes, Alresford, 1997.
16. UhdeNora, Company Information, Mai
2003; Company Pamphlet, 2002; a) AsahiKasei, Company Information, July 2003;
Company Pamphlet, 2003; b) Chlorine
Engineers Corp., Company Information,
June 2003; Company Pamphlet, 2003; c)
www.eltechsystems.com/mcre.html, 2003;
www.eltechsystems.com/3-1m2mcre. html,
2003; d) Electrochemical Technology
Business, www.etbusiness.com/html/chlortech/fm 1500.htm; www.etbusiness.com/
html/chlor-tech/bichlor.htm.
17. C. Kotzo et al., Sulfate Removal from Brine,
presented at the Chlorine Institute Annual
Meeting Houston, Texas, 2000.
18. G. A. Carlson, E. E. Estep, Porous cathode
cell for metals removal from aqueous
solutions, Paper presented at the ECS
meeting, May 1972.
19. UhdeNora, Chlorine Symposium, 2004
see 4, p. 446.
20. EuroChlor Publication Environmental Protection, 11, Mercury Housekeeping, 4th ed.,
1998.
313
314
57.
58.
59.
60.
61.
62.
63.
64.
65.
66.
67.
68.
69.
70.
71.
72.
315
317
Electrochemistry Applied to
Organic Synthesis: Principles
and Main Achievements
Jacques Simonet
University of Rennes, Rennes, France
6.1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
321
6.2
6.2.1
6.2.2
6.2.3
6.2.3.1
6.2.3.2
6.2.4
.
.
.
.
.
.
322
322
323
323
323
324
....
324
6.3
6.3.1
6.3.1.1
.......
.......
.......
.......
.......
.......
A Survey
.......
...
...
...
...
...
...
of
...
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
6.3.2.1
6.3.2.2
6.3.2.3
6.4
6.3.1.2
6.3.1.3
6.3.1.4
6.3.1.5
6.3.1.6
6.3.1.7
6.3.2
325
325
325
327
327
330
332
333
334
335
335
337
339
339
318
6.5
6.5.1
6.5.1.1
6.5.1.2
6.5.1.3
6.5.1.4
6.6
341
341
341
342
344
345
6.6.3.3
6.7
354
6.8
6.8.1
6.8.2
356
356
358
6.9
6.9.1
6.9.2
358
358
359
6.10
6.10.4
361
362
6.11
Oxidation of Halides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
363
6.12
Anodic Fluorination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
364
6.13
Zero-current Electrolyses . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
365
6.6.1
6.6.2
6.6.3
6.6.3.1
6.6.3.2
6.10.1
6.10.2
6.10.3
346
346
348
349
349
351
353
360
360
361
6.14
365
6.15
6.15.1
6.15.2
6.16
369
371
319
321
6.1
Introduction
322
separator;
6.2
6.2.1
...
... ... ...
...
.. .. ..
...
... ... ... ... ... ... ...
Anolyte
.
.
.
.
Catholyte
Separator
(membrane)
The Nernst equation takes into account the thermodynamics of the two
redox reactions (fast in the sense of
the kinetics of the heterogeneous electron transfer, but here in the absence
of associated chemical reactions). The
thermodynamic properties of a redox
reaction are described by the standard
potentials, E , and the concentrations of
the redox species at the electrode surface. At the anode, we have an electron
extraction from the electroactive species,
and at the cathode the electron is donated to an electroactive species to effect
a reduction. These processes may be
represented in the general form as:
Oxidation
A
ne
+ne
E = EAB
+
RT
nF
ln
B s
A s
Reduction
C
+n e
n e
E = ECD
+
RT ln
n'F
C
D
s
s
Choice of a Cell
The design of the cell determines the success of the electrochemical process. Very
generally, most of cells possess a separator (porous glass or organic membrane)
between the anodic and cathodic compartments. For achieving experiments on the
laboratory scale (0.1 to 10 g), even rudimentary geometries can be quite efcient.
It must also be emphasized that a cell giving the best results for one reaction might
not be the one well adapted for another.
For laboratory use, most electrolyses can
be achieved with a U-shaped cell equipped
with a glass frit separator. In electrochemistry, the notion of the working electrode
dipping in a well-dened surrounding is
essential. The electrochemist denes the
reaction (oxidation at the anode in the
anodic compartment or reduction at the
cathode in the cathodic compartment) to
be achieved, and what happens at the other
electrode (called counter electrode) does not
really matter (although one has to avoid efciently any diffusion of efuents formed
in the other compartment). Some examples of electrolysis cells are shown and
commented on at the end of this chapter
(Sect. 6.15).
On the laboratory scale, it may be helpful to use three-electrode cells, where the
electrochemical potential of the working
electrode is set against that of a reference
electrode, itself connected to a potentiostat.
The purpose of a potentiostat is to x precisely the reduction (or oxidation) potential
necessary for the complete conversion of
an organic substrate. At the end of the
reaction, the electrochemical current has
totally vanished and the concomitant use
323
324
Reduction
H+
E+ Anion radicals
Oxidation
+e
Cleavage
Free radicals
+e
+e
Organic
substrates
Organic ions
Cations radicals
H+
e
Cations Anions
Nu Substitution
reactions
Free radicals
e
Dimer
Dimer
Or addition to
cathode surface
Or addition to
anode surface
Organic cations
Organic anions
Scheme 1
6.2.4
Aromatic
substitution
ne
Metallic cations me
M n+
RX
M(n + m)+
(Mn3+, Ce4+, Ag2+, Tl3+)
Indirect oxidation
(in-cell or ex-cell
processes)
Specific reduction
processes
Free radicals
Scheme 2
but only in the case where the electron transfer occurs to generate free
radicals (induction of radical chains,
e.g. using NiII or CoII complex catalysts) or charged radicals (reactions of
cycloadditions or cyclodimerization under
the conditions chosen, so as to keep these
reactions catalytic in electron). Reactions
involving zero-valent metals can be useful for processes leading to the formation
of carboncarbon bonds from organic
halides.
The ease of formation of free radicals
brings chemists to consider electrochemistry as a convenient method to afford
dimers. Obviously, reactions of this type
are totally stoichiometric and demand
1 mol of electron per mole of a substrate.
In some cases, especially in oxidation,
the use of mediators by the electrogeneration of high-valency metallic cations can
be helpful. On the other hand, if similar,
indirect reactions are possible in the reduction range, organic mediators have to
be used. Such redox catalysis reactions are
perturbed by the fact that couplings may
occur between -acceptor anion radicals
used as electrogenerated reducing species
and a free radical issuing from the cleavage
reactions. Schemes 1 and 2 summarizes
the principal reactions achievable by electrochemistry. Obviously, the key words
are radicals, charged radicals, electrogenerated acids (EGA) and bases (EGB), nucleophiles, and electrophiles, catalysis by
zero-valency metals or by valency change
of metallic cores of metallo complexes.
The use of specic, electrochemically
charged interfaces also appears to be of
utmost importance, since structure and
properties of these conducting materials
can be changed with an applied potential.
In summary, the electrochemical technique is in fact quite simple but is also rich
in possibilities. Recent exhaustive reviews
RCO2
ads
CO2
R
ads
1/2 R R
2e
CO2
R+
Nu
Solvent
RNu
325
326
RCOO
However, many examples in the literature concern mixed electrolyses of carboxylates that can afford dissymmetric dimers
(in mixtures with symmetric ones)
e
R COO + RCOO R R
Reactions of this type are interesting
when any cheap and trivial carboxylates are
Y
R
Y
R
Dimerization
R
Nu
Nu
CF3
Radical addition
N
+ further oxidation
+
Y
+
CO2
HN
O
CH3
CF3
HN
O
Radical addition
+ further oxidation
N
CF3COO
CF3COO
R/Coupling
e
CO2
N
CH3
Fast
O
CH3CO2
R()
Nu
Coupling
1/2 RR
RNu
ER+ /R )). Many examples of dimerization [1] can be drawn from the recent
literature:
dimerization of stabilized anions from
arylacetylenes
2PhC C
Li
2e
THF
PhC
C C C
Ph
2e
RR
R = Ph, yield = 55%
R = C18H37, yield = 54%
ArH
dimerization of malonates
2EtCH(CO2Et)2
ArH
EtO2C CEt
EtO2C CEt
CO2Et
RSO2CHSO2R
RSO2CHSO2R
MeOH/Na
LiClO4
RSO2CSO2R
RSO2CSO2R
H
(if R = Et, yield: 78% at room temperature)
CO2Et
2e
EtONa
in CH3CN
+ +
HArArH
2H+
ArAr
ArH
Addition
+
2H+
HArArH
e
327
328
2CH3
CH3
Oxidation
2e / 2H+ Anode
H 3C
Me
CH3
CH3
Me
Me
CH3
H3C
Me
CH3
71%
CH3
H3 C
CH3
Me
64%
Anthracene
+ anisole
Oxidation
OMe
2e / 2H+ Anode
H3C
H3C
H3C
CH2
CH3
CH3
CH3
70%
OMe
85%
itself
in a
since
may
CF3COOH/CH2Cl2
nBu4NBF4
MeO
OMe
MeO
OMe
OMe
Anodic
oxidation
OMe
OMe
MeO
CH2Cl2 or CH3CN
nBu4NBF4
MeO
OMe
OMe
329
330
OR
n
O
O
O
OR2
R2O
R 1O
OR1
OR2
n2
O
O
O
O
O
O
O
O
O
O
O
n1
Lastly, such a trimerization reaction until now specic for orthodialkoxybenzenes could be applied [11] to the
polymerization (as polytriphenylenes) of
dibenzocrown ethers leading to resins
supported by conducting materials. The
ionophoric properties of such products
have been used for the extraction of
heavy metal cations from water when
the concentrations of these cations are
extremely small.
A
e, Ni or Pd as catalyst
DMF
Y
Y
5098%
MeOHEtOH, NaBr
e, Iron anode
20 C
NiBr2bipy as catalyst
PdIIY2L2
2e (cathode)
ArNu
ArX
Pd0L2
(4)
(1)
(cis) ArPdIIL
ArPdIIX (trans)
Nu
(3)
Nu
(2)
L
ArPdIINu
L
(trans)
Other examples can be given concerning the coupling of arylbromides [19] with
the use of nickel metal catalysts. Reactions of this type may be achieved in
protic media (20:80 Dimethylformamide
(DMF)/ethanol or 50:50 ethanol/methanol
mixture) with yields comparable with
those usually found in aprotic media. (In
Formation of arylaryl links was demonstrated to occur between aromatic bromides and 2-chloro or 2-bromopyridines.
These reactions are efcient, since they
allow in one step substitution reactions
that are obviously much more difcult to achieve by the usual chemical
methods.
331
332
Br
+
N
Br
e, Al anode
NiBr2 bipy 10%
DMF 20 C
Y
DMF, 4070 C e, Aluminium anode
ClCH(R)CO2R NiBr2, bipy 510%
CH(R)CO2R
Y
5080%
5090%
Br
Same conditions
Br
+ ClCH(R)COR
Y
DMF , 4070 C
e , Al anode
NiBr2, bipy 515%
S
80%
CH(R)COR
ArX
e,
Y
5080%
ArR
Ni catalyst
Sacrificial
anode
RCH=CHA
(b)
ArCH2CH2A
e, CoII or NiII
Br
+ CH2=CHA
RCH(R)CH2A
FG
DMFpyridine (9:1)
1) e, Fe anode
60 C
NiBr2 10%
2) H3O+
CH2CH2A
It is important to stress that the electrochemical reaction is here fully stoichiometric. The term catalytic used for Ni or Co
intermediates (constantly regenerated in
the course of the electrochemical reaction)
has therefore to be carefully differentiated
from reactions using NiII or PdII as listed
above.
Aromatic Carboxylations
The electroreduction of aromatic halides
in the presence of catalytic amounts of
[NiPPh3 ] [27] or [Nidppn] [28] mainly
yields arylcarboxylates. Such a reduction
process may also be catalyzed by palladium
complexes.
6.3.1.6
Br
MeO
RCH=CHCH2CH2COCH3
CO2
FG
e, PdCl2(PPh3)2
CO2H
FG
FG = Cl, CF3, CO2Et, F, CH3
(FG) functional group
333
334
Isopropanol
Et4NOTs
2e, 2H+
Sn cathode
CH3
H
OH
cis/trans: 100 : 1
OH
O
(CH2)n
CO2Me
H
trans/cis: 2.5/1
95% yield
for dimethylcyclobutylmalonate;
Br(CH2)CH=C(CO2CH3)2
starting molecule):
RH
RH
H+
R
e
Hg pool,
(dry DMF/nBu4NBr) 1.85 V vs SCE
CH(CO2CH3)2
R+
Nu
RNu
Substitution
+
H
R(H)
Unsaturation
65%
CH2OTs
6.3.2
In a general way, although many exceptions can exist that depend on intrinsic
structures of organic substrates, the anodic oxidation of functional groups with
mobile hydrogen is depicted as follows
(the process implies a radical cation
as primary intermediate, whose acidity
is strongly enhanced compared to the
6.3.2.1 Functionalizations of
Nitrogen-containing Compounds
Generally speaking, the electrochemical
oxidation of amines yields iminium salts,
which are easily hydrolized. However, efcient nucleophiles can be added to the
electrolysis mixtures in order to react
on the iminium before its hydrolysis by
residual water. The nucleophiles can be
cyanide [36] or methanolate [37]:
335
336
Pt anode
methanol/water
CH3
N CH3
NaCN/2e/H+
62%
Pt anode
acetonitrile
CH3
N
C2H5
Et4NCN/2e /H
CH2CN
38%
CH3
CH 2 CN
+
N
CH
C2 H 5
CN
64%
CH3
34%
CH3
N
MeOHLiClO4
2e
CH2OMe
N
(CH3)2N
N
N
N(CH2)2
CH3
N
N(CH3)2
N
(CH3)2N
MeOHNaOAc
N
(CH3)2N
4e
CH2OMe
N
N
N(CH3)2
CH3
N CH2OMe
CH3
N
CH2OMe
MeOHNaOAc
CH3N
6e
MeOCH2
N
N
N
CH2OMe
CH3
In the example given above, experimental conditions and the control of the
electrical charge may orientate the reaction selectively to the mono-, di-, and tri
substitution. The action of Lewis acid on
substituted amine functions allows one to
get back the iminium intermediates [38,
39]. Thus, the nucleophilic action in the
course of the electrolysis could be seen
as an iminium protection, and renders
possible easy synthesis of the heterocycles
as outlined below:
Ph
+
N
CH2
CH3
CH3
Y
Y
N
CH3
CHOCH3
R3
Oxidation of CH Functions
The anodic conversion of tertiary CH
functions (without any specic activation)
is possible, yet it needs rather high potentials. One typical example is certainly that
of adamantane [42], the oxidation of which
can be achieved in acetonitrile. Under
these conditions, the corresponding acetamide (via Ritter reaction) was obtained.
6.3.2.2
Lewis acid
H
R2
R1CON
+
CHR3
Nu
R2
R1CON
CHR3
Nu
NHCOCH3
2e
CH2
H+
CH3
N C
CH3
+
N C
NuCH2
2e
CO2CH3
N C
CH3
H+
H
Nu
(90%)
C C
+
N+
CO2CH3
e /Nu H+
Nu
Nu
CO2CH3
C C C
Nu
337
338
carbon of the double bond affected by oxidation becomes the sole reaction to be considered.
Thus, alternatively two chemical routes can be
envisaged: (1) disubstitution of the (one) double bond and (2) dimerization of the radical
cations.
C
C
e
Nu
C Nu
+
C
Nu, dim
OMe
Nu, e, Nu
NuCCCCNu
2e
Et4NOTs/NaOMe
CH3OH
Diene
Solvent
electrolyte
Secondary
product
Main product
OCH3
CH3OH
Et4NOTs
OCH3
OCH3
OCH3
OCH3
OCH3
CH3OH
Et4NOTs
OCH3
OCH3
OCOCH3
OCOCH3
CH3COOH
Et4NOTs
OCOCH3
OCOCH3
H3CO
CH3OH
Et4NOTs
H3CO
OCH3
2e/2H+
CH3OH
R1 +
S
R2
R2
Dim
R1
+
S
+
S
R1
R2
R2
OCH3
6.4
S H2O/2H+ 2e
R2
S
+
2ClO4
R1
S
R2
S
+
H2O/2H
2e
R1
e(LiClO4)
S
+
Dimerization
S
+
S
S
R2
Let us note that until now electrochemistry has not been able to yield chiral sulfoxides from the starting prochiral thioethers
even though the use of chiral (chemically
modied) electrodes was considered.
1/2
S+
S
2ClO4
339
340
SR
+
SR
e, H+
[LiClO4]
PhSR
R +
S
Ph
SR
Lutidine
e,
Pt anode
24%
MeCN
LiClO4
S
K2CO3
ClO4
S
96%
2MeO
SMe
Bu4NBF4/acetonitrile
2e
platinum anode
MeO
S
BF4
MeS
Me
OMe
Lastly, the oxidation of dibenzothiophenes with electron donating substituents, such as alkoxy or amino groups,
unexpectedly affords extremely stable radical cations [48] that can be used in organic
chemistry as convenient oxidizing species
with the oxidative abilities depending on
their standard potential (E ). The salt given
below is electrically conducting and for
this reason is readily obtained in very large
amounts (yield: 73%) as a thick deposit
on graphite anodes (electrolyte: acetonitrile +0.1 M LiClO4 ). Many other radicals
have been obtained, and their standard
potentials have been published [49].
6.5 Electrochemical Conversion of Main Organic Functions in Aqueous and Protic Organic Media
O
O
+
S
O
ClO4
6.5
R2
R1
6.5.1
Nitro Group
The cathodic reduction of nitroalkanes
(primary and secondary) affords [50] (fourelectron reduction in acidic or neutral
media) the corresponding hydroxylamines
that are often quite difcult to be converted
into amines.
CHNO
R2
1
Hydrolysis R
C=NOH
R2
R1
2e, 2H+
H2O
CHNO2
C=O
R2
oximes and therefore the regular fourelectron reduction allows a selective generation of hydroxylamines. Conjugated
nitroalkenes are converted (four-electron
reduction) into oximes [51] through the
unsaturated hydroxylamine.
R1CH=C
R2
NO2
4e, 4H+
H2O
R2
R1CH=C
NHOH
6.5.1.1
R1
CHNO2
R1CH2C
R2
NOH
R2
Ar
H2O 4e , 4H+
CH CH2 CR
R1
CNO2
R2
R1
H2O 4e, 4H+
CHNHOH
R2
Ar
Ar
R1
R2
CH CH2 CR
C NHOH O
R1
H2O
R2
N
O
341
342
4e, 4H+
H2O
ArNO2
ArNHOH Anode
Cathode
ArNO
Solution
ArNHOH
Spacer
ArNH2
ArN(OH)2
Aromatic
hydroxylamines
bearing
electron-donor substituents (OR, NR2 , SR)
exhibit a limited stability. Thus, reduction
products are generally the corresponding
anilines. The global mechanism is
illustrated [56] below for the case of
nitrophenol.
H2O
NHOH
NO2
ArNO
2e,
H2O
2e, 2H+
H2O
2H+
OH
ArNHOH
OH
2e,
NH
2H+
O
Redox
process
NO
NH2
OH
OH
6.5 Electrochemical Conversion of Main Organic Functions in Aqueous and Protic Organic Media
R
C NOH
4e, 4H+
H2O
CH
NH2
N
H
N NO
CH3
CH3
N
H
CH3
2e, 2H+
NH2
NH +
N
H
1
N O + 2OH
2 2
CH3
H3C
CH3
CH3
NH
H3C
NH2
(Basic) 2e + 1.5H2O
NH3, NH4+
H2O
NH
R
CH3
H3C
O
H3C
H2O
+
(Acidic) 4e , 4H
ArN NAr
CH3
4e, 4H+
H2O
ArNHNHAr
O
Oxidation
ArN NAr
343
344
2e
Basic
ArH + CN
2R'CH
2e
2H+
CHCOR
Ar
Ar
HC
CH
OH OH
2e
2H+
Ar
C O
R
2e, 2H+
Ar
OH
C
ArCH2NH2
ArCN
2ArCOR
meso + DL
RCHCH2 COR
2
6.5 Electrochemical Conversion of Main Organic Functions in Aqueous and Protic Organic Media
Electrochemical
Saturation
H
H
H
OR
2e, 2H+
ArCH
OH
ROH
Ar
ArCOH
C O
2e, 2H+
ArCH2OH
R
Chirality agent
(homogeneous or heterogeneous)
+2e,
2H+
Ar
*
CH OH
R
In most of the cases [6368], whatever the chosen mode of the chirality
induction and the nature of the considered medium, experimental enantiomeric
excesses generally are low (ee < 20%)
and this technique remains to be optimized.
ArCONHR
H2 O
ArCONHR
ArCH2 OH + NH2 R (Less acidic)
345
346
6.6
ArX + e
ArX
Ar + X
Ar
ArH
or Ar
H-atom transfer
ArH
RX
R + X
RX
e, H+
RH
>
CX2
C X
>
ArSR
ArSO2R
(But ArSOR
t-Bu
Me
ArSR
Cl
ArCH2OR
2e/DMF
2H+(proton donor)
mercury cathode
ArCH3 + HOR
while
OR
+ R
e
H+
RH
e
H+
RH
ArSR + H2O)
+
MeSMe
ter-Butylation reactions
e,
Below are exhibited some useful reactions based on the cathodic cleavage of
bonds, which can be seen often as a source
of free radicals (homogeneous or heterogeneous cleavage reactions):
Lastly, intramolecular cyclizations (presumably provoked by the transient radical)
Me
R4N+, X
4/H2NMe
ArSO2 +
t-Bu
RO
2e/LiClO
ArS
ArSO2R
2e, 2H+
[R4N+, M]
(Amalgam
if M = Hg)
R3N
347
348
Ph
C(CH2)4I
PhC
Ph
C(CH2)3CH2
Ph
SH
S
Ar'
Ar'
C
Ar
OH OH Ar
2B
2BH
Ar'
Ar
Ar'
Ar'
Ar
Ar
C
O
Disproportionation
and protonation by BH
Ar'
C
Ar
6.6.2
OH
Ar'
C
Ar
As to the oxidation, an opposite situation may exist [74a] since only strongly
electron-enriched bonds are able to undergo scission.
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
2e
2MeOH
(2H+)
OMe
range.
Ph
Ph
N
N
Ac
N Ac
N Ac
Ac
4e
4Ac2O
Ph
Ph
Oxidation
4e, Ac+
6.6.3
In many cases, the electrochemical technique recently developed has been shown
to be a fruitful alternative to chemical deprotection. Thus, organic chemistry
functions that are mainly involved in protection/deprotection processes are mostly
aldehydes and ketones, amines, alcohols,
and thiols; carboxylic acids can also be
quoted, although their esterication as well
as their ester hydrolyses are not straightforward.
The cleavage of a protecting group
by electrochemical means, when such
a method is applicable, can be considered as advantageous for the following
reasons:
The
equipment
is
simple
(a
two-compartment cell with adequate electrodes and a source of
current potentiostat or galvanostat).
The reaction needs, in principle, only
electron transfer. There are therefore
no predictable side reactions, but the
choice of the solvent (nucleophilic
for anodic deprotection) can be crucial.
The potential or current can be adjusted
(and controlled carefully during the
course of the reaction) in order to
render the deprotection regioselective
(control of the potential or the amount
of charge).
Sometimes the use of redox mediators
(with electron transfer now taking
place in the homogeneous phase) may
contribute to better selectivity. Thus,
deprotection can be carried out under
very mild conditions.
349
350
S
C
R2
H2O/H+
R2
R1
R2
OH
S
S
OH
C
R2
R1
H2O
1e
R1
OH
H2O
R1
C
R2
quantitative yields.
S
Ph3P
R2
H R1
ClO4
Platinum or glassy
carbon anode
O
+
ClO4
Ph3P
R2
R1
CHCOX + R2 CHCOP2
NHP1
NH2
XH
R1
NHP1
O
C2H5
CHCONHCHR2
COP2
351
352
manner.
Ts
Ts
SO2
CH3
COOC(CH3)3
COOCH2C6H5
: Paratoluene sulfonyl
: Boc (alkyl or
aryloxycarbonyl)
R1
N
R2
SO2Ar
R1
N
R2
R1
SO2Ar
+ ArSO2
N
R2
Ts
(CH2)n N
Ts
4Ts, 4H+ 8e
(CH2)n
H
N
(CH2)n
HN
NH
(CH2)n N
H
n = 1 96%
n = 2 95%
e, H+
Ms
N
R1
NH
Ts
R2
Ts
N
Ms
2Ts, 2H+ 4e
Ms
N
NH
HN
N
Ms
R1
H
CHNTs
R2
R1
2e
(Aprotic
solvent)
CHNH
R2
R1
R1
Starting
Sulfonamide
R2
R1
CHNH2
H
CNTs
Ts
R2
H2O
NH3
CHNTs
R2
R1
C NH
R2
R1
C O
R2
NO2
1 Reduction ROCH2
O
O
O
N COtBu
R
+
2H
2e
(Solvent)
O
O
NH + CO2 + tBu4
R
NO2
NO2
ROCH2
NH2
ROCH
NH
Deprotection
phase
2 Oxidation ROCH2
NH2
ROH
353
354
OBn
O
O
6.7
OCH3
2) +1.55 V
1) 2e / CH3CN/H2O,
(vs Ag/AgCl/Cl )
Bu4NPF6
BnO
BnO
BnO
OBn
O
OH
C6H11SMBZl
CH3CN/lutidine
(4BrC6H4)3N
e
+
C6H11SMBZL
Since the 1960s, the tremendous activity devoted to solid phase chemistry has
found a renewal of interest with the emergence of a new concept, the combinatorial
synthesis [96], which now nds various applications in both academic research and
in industry. On the basis of the difference in solubility between a polymeric
material and a simple reactant or impurity, the following cleavage from the
support and the recovery of the desired
product have led to the development of
Fast
C6H11SMBZL
(4BrC6H4)3N
Anode
C6H11S +
Coupling
Ritter
reaction
C6H11S
2
89%
4CH3 OC6H4CH2
H2O
Acetamide
Fast
e
Ar SO2 XRR'
+ RR'X
Ar SO2
RR'X
X : CH, N
NH2
RHN
NHSO2
R: hexCO,PhCO, BzCO
RHN
N
H
SO2NEt2
H2N
PT
Ar
SO2Cl
NHR
R: PhCO
CH3
355
356
partly because of the presence of a sulfonamide moiety [99]. This new interfacial
process, starting from the electrogenerated
sulfonylchloride PT, has been validated in
the synthesis of various polyamine precursors, which were obtained with yields
ranging from 40 to 60%.
Thus, the deprotected secondary amine
shown in the scheme below has been
obtained in 60% yield according to the
following sequence:
N3
O
SO2Cl
SO2N
N3
(1)
S
(2)
NH2
O
SO2N
NH2
(3)
NHCOPh
O
SO2N
NHCOPh
(4)
NHCOPh
HN
NHCOPh
(1) [N3(CH2)3]2NH, pyridine; (2) SnCl2, PhSH, Et3N; (3) PhCOCl, pyridine; (4) 2e, proton source
6.8
Electroreduction
6.8 What is the State of the Art for Organic Silicon Derivatives?
R = Alk, Ar
Similarly, reducing mono- and dichlorosilanes and disilanes taken in
different combinations (e.g. R2 SiCl2 +
R3 SiCl; R3 SiSiR2 Cl + R3 SiCl; 2R3 SiSiR2
Cl; 2R3 SiSiR2 Cl + R2 SiCl2 , etc.) yield tri-,
tetra-, and pentasilanes [102, 104, 105].
Silyl anions, electrogenerated by the
cathodic reduction of polyfunctional
chlorosilanes, react with remaining SiCl
bonds of the starting molecules or of
intermediately formed oligomers to give
linear polysilanes. The molecular mass of
the resulting polysilanes depends on the
electrolysis conditions (solvent, electrode
material, supporting salt, etc.) [102,
105108].
4e to 8e
R1R2SiCl2
2Cl
R1
Si
R2
R1, R2 = Ar, Me
357
358
+2e
+X
R +
SiCl
R
R3SiR
Cl
SiR
Electrooxidation
Si
2e
Et3N 3HF
2e
Et4NBF4
R3SiF
Si
RCHSiMe3 RCH2 Nu
R = Ph, CH2 CH;
Nu = ROH, RCOOH, H2 O.
A similar process with -silyl amines,
carbamates [129], ethers [130], suldes
[130], or selenides [131] also results in the
displacement of the Me3 Si group with an
appropriate nucleophile.
Indirect anodic substitution was used
for a functionalization of heteroatom
(O, S, N) -substituted organylstannanes
and organylsilanes, and of triphenylphosphine [127].
6.9
Electroreduction
Likewise, in organic chemistry, the elimination of selenoxides is very important when electrochemically oxidizing
organylselenides containing a (Se)Csp3 H
fragment. This reaction, combined with
the above selenylation, is a key step in
a whole series of selenium-catalyzed processes of the functionalization of organic
compounds [138a, 139].
2e/Nu
R
Nu
SePh
2e
Nu
Ph2Se2
+2e
2Se R2 Se2
2RX, 2X
6.9.2
Electrooxidation
RSe
2e, Nu
Nu
Electrooxidation of diphenylselenide
[140] and of a number of cyclic aryl
selenides (having no acidic protons at
the -carbon prone to syn elimination),
potentially interesting from the biological
point of view, gives selenoxides in different
yields [141].
High acidity of a proton in the position to selenium, increasing even more
when the molecule obtains the positive
charge upon oxidation, can be used for
its oxidative replacement. Thus, the reactions of -methoxylation, acetoxylation,
and uorination (Nu = MeO, AcO, F) of
alkylarylselenides [142] have been realized.
2e
PhSeCH2 R PhSeCH(Nu)R
Nu, H+
RSe
Nu
359
360
6.10
Potassium superoxide KO2 is commercially available. But since dioxygen is reversibly reduced in aprotic organic solvents
at many types of electrodes (e.g. platinum,
glassy carbon), it can be easily formed in
situ [144]. The superoxide can be readily
protonated: therefore aprotic solvents possessing an efcient proton donor and aqueous solvents permit a two-electron step to
occur. This step affords the formation of
hydrogen peroxide. Its further reduction
O2
+
O2
H+
O2H +
H+
O2H
aprotic
medium
O2
O2H
Fast
O2H
O2H2
The use of molecular oxygen in electroorganic synthesis must take into account the fact that superoxide can act both
as a reducing reagent and nucleophile. Additionally, in the presence of fairly acidic
substrates (but not electroactive at the
standard potential of O2 (in acetonitrile
EO2 = 1.06 V versus SCE)) acidobasic
reactions may take place in the bulk of
the catholyte (owing to the basicity of O2
and O2 H ). Below are described some
standard reactions employing dioxygen (or
its reduced forms) generated in DMF,
principally in the presence of tetraalkylammonium salts.
acidic substrates (carbon substitution)
[145 146]
R
Ph
C
CO2H
Ph H
2
H
O
+
O2 , NR4
e, coupling
[146] O2
OH
Ph
OH
C CO2H
Ph OH
OH
OH
OH [146]
O2
O2 + H2O2
OH + OH + O2
OH +
Phenoxy derivatives
R2
+
RNH C O
RNH
Anode
O O
2
1 e
1 + + O2
1 O2
1+
1 O2
6.10.2
6.10.3
361
362
R2
R2
R1
+2.0 V vs SCE
S8 8S2+ + 16e
Tetraethylammonium
tetrauoroborate/CH2 Cl2
Pt anode
R2
R1
R1 = C6H5
R2 = CHO or CO2Me
Yield: 8285%
2XArH + S2+
XArSArX + 2H+
OH
2
HO
S2+
OH
95%
OH
1 M Cl3CCOOH
OH
37%
With other nucleophiles, such as thiols [156], SSS linkages are generated
in good yield. However, the side formation
of SS bonds could be provoked by the
oxidative properties of S2+ .
2RSH
RSSSR + 2H+
S2+
6.10.4
CH2 S
S2+
CH2 S
S
48%
SCH 2 CH 2 S
SCH 2 CH 2 S
20%
t-BuSSBu-t
2e
2e
2tBu+ + S2
2RS+
R'CN
H2O/H+
OH(R)
SMe
CH3S+
H+ [157]
+ RS+
Acetonitrile RS
N
[158]
CCH3
+
H2O/H+
RS
tBuNCOR'
6.11
Oxidation of Halides
NCOCH3
OH
H+ CH2Cl2
[159]
O
SAr
[157159]
However, the above schemes do not work
with a tertiary R. The oxidation of the corresponding disuldes yields essentially tertiary
carbocation. Consequently, oxidations performed in nitriles [160] can be considered
as an elegant route (via the Ritter reaction)
OCH3
2e
+ ArS+
NH4Br/CH3OH
CH3O
H
C6H13CHCH3
OH
2e
KI/H2O
C6H13CCH3
O
363
364
6.12
R1
NH3
Anodic Fluorination
NH
R2
R2
I/I+ 2e
R1
N
R1
R2
N
H
R2
CO2H
HO
OH
2e, H2O
OH
Br
HO
H
OH
OH
HO
H
CO2H
CHO
R
C
C
R
2e
Nu
MeO
Nu
CH2
EWG
CH
EWG
F 2e
R
R
MeO
ArS CH2
EWG
Acetonitrile Et N3HF
3
Pt anode
ArSCHEWG
F
SCH2R
EWG
Acetonitrile Et N3HF
3
Pt anode
SCHR
EWG
Zero-current Electrolyses
Cathode
DMF
PhSO2
Electrogenerated O2 As EGB
DMF
Cathode
PhO2S
SO2Ph
(Induction
at the cathode)
ArNu
ArX
X
ArX
ArNu
SO2Ph
6.14
ArX
e
SO2Ph
Ar
Nu
(Propagation)
in solution
365
366
H H
H
Anodic
aoxydation
R
O
Hydrolysis
O
OMe
OMe H
H H
O
S
OMe
Br
Br
Br
6.15
The present chapter gives a long, but farfrom-exhaustive, list of reactions achievable by means of electrochemistry at
cathodes or anodes under specic experimental conditions (choice of the solvent,
the electrolyte, the material of the working electrode, etc.). However, the geometry
of the cell, strictly depending on the distance and the volume between the working
electrode and the counterelectrode, and
the nature of the separator, if any, can
play a fundamental role in the success of
the electrochemical method for achieving
a synthesis.
Below are presented a few cells, some
made of glass, of small volume, which
could be convenient for organic chemists
to test on small amounts of substrates
(e.g. from 0.1 to 10 g), their possible interest, and the feasibility of electrochemical
conversion. Some other cells usable for
Laboratory Cells
367
368
Working solid
electrode
Counter electrode
(carbon or Pt grid)
Inert gas Additional
inlet
entry
Reference
electrode
Outlet
Porous glass
separator
Contact for
Hg pool
cm
Inert gas
outlet
Location for
reference
electrode
Inert gas
inlet
Porous glass
separator
Pt contact
for Hg pool
with the starting material and/or the reaction product formed at the working
electrode. Most of the time, the cathodic
reaction leads to hydrogen evolution.
Lastly, cells equipped of porous electrodes are useful in organic electrochemistry. Redox ow cells were efciently
developed by Moinet [55], and this concept
can be considered as a great advantage
toward conventional reduction and oxidation methods in organic chemistry. It is
essentially because they can be particularly
useful to obtain electroactive intermediates
in reduction and oxidation multistep
processes. A striking example had been
mentioned in Sect. 6.5.1.1 in the synthesis
of aryl nitroso compounds. However, the
ow of the solution through the two porous
electrodes is difcult to regulate.
6.15.2
Industrial Cells
Within the last three decades, a large number of versatile electrochemical cells have
been developed [185]. Beside their suitable
geometry and principle of operation of the
cell, the availability of more stable electrodes and separators (mostly membranes
of acceptable lifetime) has considerably
improved.
1 : Solid anode
2 : Anolyte frame with inlet/outlet for anolyte
3 : Separator
4 : Catholyte frame with catholyte inlet/outlet
5 : Solid cathode
6.16
Spent
oxidizing
agent
Reactor
Scrubber
Electrochemical
cell
Separator
Oxidizing
agent
storage
Product
Fig. 5
369
370
Product (cathodic
(C) or anodic (A))
C Adiponitrile
C 2,2-Dichloro-1-(4hydroxyphenyl)
ethene
C 3-Hydroxybenzyl alcohol
C Tetrauoro-p-xylene
A
A
A
A
A
A
A
C
C
A
A
Dimethyl sebacate
Anisic aldehyde
Tolualdehyde
Anthraquinone
2,5-Dimethoxy-2,5
dihydrofuran
Calcium gluconate
Fluorocarbons
4-Aminophenol
Azoxybenzene
Mucic acid
Acetoin
Precursor
Company/Country
Scale (metric
tons per year)
Acrylonitrile
3,3,3-Trichloro-1hydroxy-1-(4hydroxyphenyl)
ethene
3-Hydroxy benzoic acid
Bis-quaternary
ammonium salt
Monomethyl adipate
4-Methoxytoluene
p-Xylene
Anthracene
Furan
Mitsui Taatsu
Europe
100
200
Asahi, other
BASF
BASF
L.B. Holliday
BASF
?
3500
1000
100
100
Glucose
Carboxylic acids
Nitrophenol
Nitrobenzene
Galacturonic acid
Butanone
India, Sandoz
3 M, Bayer, Phillips
India
India
EDF
BASF
>1000
>10
?
3
?
?
OMe
OH
2e,
2H+
2MeOH/I
Graphite anode
371
372
373
374
138.
139.
140.
141.
142.
143.
144.
145.
146.
147.
148.
149.
150.
151.
152.
153.
154.
155.
156.
157.
158.
159.
160.
161.
162.
163.
164.
165.
166.
167.
168.
169.
170.
171.
172.
173.
174.
175.
176.
5767. (b) A. Bewick, D. E. Coe, J. M. Mellord et al., J. Chem. Soc., Chem. Commun.
1980, 5152. (c) A. Bewick, D. E. Coe,
M. Libert et al., J. Electroanal. Chem. 1983,
144, 235250. (d) A. Bewick, D. E. Coe,
J. M. Mellord et al., J. Chem. Soc., Perkin
Trans. 1 1985, 10331038.
S. Toteberg-Kaulen, E. Steckhan, Tetrahedron 1988, 44, 43894397.
D. Elothmani, Q. T. Do, J. Simonet et al., J.
Chem. Soc., Chem. Commun. 1993, 715717.
N. Clauson-Kass, F. Limborg, K. Glens, Acta
Chem. Scand. 1952, 6, 531534.
T. Shono, Y. Matsumura, J. Hayashi et al.,
Tetrahedron Lett. 1979, 165168.
T. Chiba, H. Sakagami, M. Murata et al., J.
Org. Chem. 1995, 60, 67646770.
(a) S. Torii, K. Uneyama, M. Ono et al.,
Tetrahedron Lett. 1979, 20, 46614662; (b)
J. Y. Becker, D. Zemach, J. Chem. Soc.,
Perkin 2 1981, 336; (c) K. Rossen, R. P.
Volante, R. J. Reider, Tetrahedron Lett. 1997,
38, 777778.
J. F. Fauvarque, G. Guerin, S. Petit et al.,
Process and electrolytic cell for preparing
galactaric acid, Fr. Demande, Fr 2699937,
1994.
N. L. Weinberg, K. Hoffman, A. Kentaro,
Can. J. Chem. 1971, 49, 740745.
J. Gourcy, J. Simonet, M. Jaccaud, Electrochim. Acta 1979, 24, 10391046.
L. L. Miller, B. F. Watkins, J. Am. Chem. Soc.
1976, 98, 15151519.
T. Fuchigami, in Organic Electrochemistry,
4th ed. (Eds.: H. Lund, O. Hammerich),
Marcel Dekker, NY, Basel, 2000,
pp. 10351050.
J. M. Saveant, Acc. Chem. Res. 1980, 13,
323329.
J. Pinson, J. M. Saveant, in Electroorganic
Synthesis. Festschrift for Manuel M. Baizer
(Eds.: R. D. Little, N. L. Weinberg), Marcel
Dekker, New York, 1991, pp. 2944.
J. Delaunay, A. Orliac, J. Simonet, Tetrahedron Lett. 1986, 27, 62056206; 1995, 36,
20832084.
J. Delaunay, J. Simonet, C. R. Acad. Sci.
Paris 1997, 324II, 269273.
M. F. Nielsen, B. Batanero, T. Lohl et al.,
Chem. Eur. 1997, 3, 2011.
D. Caras-Quintero, P. Baeuerle, Chem.
Commun. 2004, 8, 926927.
A. M. Marthe, G. Mousset, P. Pouillen
et al., Electrochim. Acta 1991, 36, 1911.
375
376
188. D. Pletcher, F. Walsh, Industrial Electrochemistry, 2nd ed., Chapman & Hall, London, 1990, pp. 313323.
189. M. M. Baizer, J. Electrochem. Soc. 1964, 111,
215.
190. M. M. Baizer, D. E. Danly, CHEMTECH
1980, 10, 302.
191. D. Degner, H. Hannebaum, H. Nohe et al.,
AlChE Symp Ser. 1979, 75, 1418.
192. M. Seko, A. Yomiyama, T. Isoya, Hydrocarbon Process. 1979, 117.
193. H. Putter, J. Botzem, H. Hannebaum et al.,
Elektrochemische Methoxylierung von
Ketronen, Annual Conference of the GDCh.
Fachgruppe Angewandte Elektrochemie,
Frankfurt, September 1998, GDCh
Monograph, pp. 458465.
194. BASF information 15.10.98, Company
brochure, BASF, Ludwigshafen.
195. P. Vaudano, E. Plattner, C. Comninellis,
Chimia 1995, 49, 1216.
196. (a) S. Dapperheld, R. Rossmeissel, Process
for partial electrolyte dehalogenation of
dichloro- and trichloroacetic acids and
electrolysis solutions, Geo. Offen. DE
4029068, 1992.
197. J. Chaussard, M. Troupel, Y. Robin et al.,
J. Appl. Electrochem. 1984, 19, 345348.
377
7.1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
379
7.2
Basic Principles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
379
7.3
History of Batteries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
384
7.4
388
7.5
7.5.1
7.5.2
391
391
396
7.6
397
7.7
Characteristics of Batteries . . . . . . . . . . . . . . . . . . . . . . . . . . . .
398
7.8
7.8.1
7.8.2
400
400
402
7.9
7.9.1
7.9.1.1
7.9.1.2
7.9.1.3
7.9.1.4
7.9.1.5
7.9.1.6
7.9.1.6.1
7.9.1.6.2
Battery Technologies . . . . . . . . . . . .
Primary Batteries . . . . . . . . . . . . . . .
Leclanches Cells . . . . . . . . . . . . . . .
Magnesium Cells . . . . . . . . . . . . . . .
Alkaline Manganese Dioxide Batteries .
Silver Oxide Cells . . . . . . . . . . . . . . .
Zinc/Air Cells . . . . . . . . . . . . . . . . .
Lithium Batteries . . . . . . . . . . . . . . .
Li/SO2 Cells . . . . . . . . . . . . . . . . . .
Li/MnO2 Cells . . . . . . . . . . . . . . . .
402
402
402
404
405
406
406
407
407
408
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
378
7.9.2
7.9.2.1
7.9.2.2
7.9.2.3
7.9.2.4
7.9.2.5
Secondary Batteries . . . . . . . . . . . . . . .
Lead-acid Batteries . . . . . . . . . . . . . . . .
NickelCadmium Batteries . . . . . . . . . .
NickelHydrogen Batteries . . . . . . . . . .
NickelMetal Hydride (Ni/MH) Batteries
Lithium-ion Batteries (Li-ion) . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
409
409
411
412
413
414
7.10
7.10.1
7.10.1.1
7.10.1.2
7.10.2
7.10.3
7.10.4
7.10.5
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
415
416
416
417
417
418
418
418
7.11
418
7.12
List of Symbols
Greek . . . . . .
Subscripts . . .
Superscripts . .
References . . .
420
420
421
421
421
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
379
7.1
Introduction
7.2
Basic Principles
This section presents the denition of specic terms used in battery technology and
the basic principles for the regular operation of batteries. Lithium-ion batteries will
be used as an example to illustrate these
concepts because they possess attributes
that exhibit the most recent technologies
available in batteries.
A battery is a complex device that
delivers electrical energy by transforming
chemical energy. The electrical energy
is provided by electrochemical reactions
(oxidationreduction reactions) that take
place at the anode and the cathode of
the battery. While the term battery is
often used, the basic electrochemical unit
being referred to is the cell [1]. A battery
is composed of several cell units that
are connected in series or in parallel
Negative
electrode
(Lix C6)
Separator
Li+
Positive
electrode
(Li1x AB)
(+) Tab
DV
() Current collector
() Tab
380
Electrolyte, binder,
and filler
Active material
Lix C6
xLi + C6 + xe
(1)
Charge
xLi+ + xe + Li1x AB
LiAB
Charge
(2)
where AB represents a composite material. The reactions given in Eqs (1)
and (2) are reversible, that is the reason
why lithium-ion batteries are rechargeable. Theoretically, the deintercalation and
intercalation fractions of lithium (x) in
both the electrodes should be the same (as
shown in Eqs 1 and 2). However, because
of side reactions or decomposition reactions that take place during the charge and
discharge processes lithium may react with
other species present in the battery affecting the lithium balance shown in Eqs (1)
and (2) (e.g. the formation of the solid electrolyte interface, SEI). These details have
been suppressed to avoid complications
that are not necessary to illustrate the objectives of this section. More details about
side reactions in lithium-ion batteries can
be found in the literature [3].
There are two types of electrodes,
metal electrodes and porous electrodes.
For instance, lithium-ion batteries possess
porous electrodes (as shown in Fig. 1),
while lead-acid batteries have metal electrodes. The selection between metal and
porous electrodes depends on the chemistry and on the safety of the battery.
Porous electrodes in a battery are usually
very thin to reduce the internal resistance
of the battery. They are typically made
of small particles to reduce internal diffusion limitations and to provide a large
381
382
Lid
Gasket
Tabs
Cathode lead
Burst disk
PTC
Separator
Insulator
Can with
plastic
insulator
Separator
Anode
lead
Can
Insulator
Jelly roll
Cathode
(a)
Anode
(b)
Schematic diagram of a complete lithium-ion battery. (a) Jelly roll, (b) other components and
safety devices (from Ref. 5).
Fig. 2
383
384
7.3
History of Batteries
Fig. 3
Gaston Plant
developed the first
rechargeable battery
(lead acid battery).
1859
1973
1947
Neumann successfully
sealed the Ni/Cd
battery. His design is
currently used in
electronic applications
1940
Rechargeable alkaline
manganese (RAM)
cells were developed.
1975
1991
Sony commercialized
the first lithium ion
battery.
1959
1970
1990
Sanyo and Matsushita
commercialized Ni/MH
batteries
Eveready commercialized
the first alkaline battery
developed by Lew Urry.
1878
Alessandro Volta
described the first
electrochemical
battery (voltaic pile).
1800
1866
George Lionel
Leclanch developed
the first prototype of
the modern dry cell
(Leclanchs cell).
1791
Galvani observed that
electricity is conducted
through animal tissue
(frog leg experiment).
Edison (USA)
invented the Ni/Fe
battery (1901)
1899-1901
Waldmar Jungner
(Sweden) invented the
Ni/Cd battery (1899).
1836
John Fredric Daniell
developed the first
battery able to produce
a constant source of
electricity (Daniells
cell).
386
quantities of zinc oxide in aqueous solution for the electrolyte [7]. These batteries
are now available in two different sizes: relatively large cylindrical cells and miniature
button cells.
Long cycle batteries were needed for
satellite applications. In 1970, the rst
nickelhydrogen battery (Ni/H2 ) was developed for this purpose. These batteries
are able to store hydrogen at very high
pressures but they are very expensive. The
long cycle capability of these batteries catalyzed the research and development of
a similar type of battery for consumer
applications. Different alloys were developed to store hydrogen; however, they were
very unstable. It was not until the 1980s,
when enough stable materials were developed for use in commercial cells [6], that
the nickelmetal hydride cells (Ni/MH)
were designed. In 1990, a Ni/MH battery
was commercialized in Japan by Sanyo
and Matsushita [11]. These two companies
started the large-scale commercialization
of small sealed Ni/MH batteries. The
companies associated themselves with Duracell, Toshiba, and Varta in a consortium
known as the 3C alliance (camcorders,
cellular phones, and computers) [7]. The
idea of the 3C alliance was to use the
Ni/MH batteries for applications such
as communications, cordless consumer
products, and other high rate long cycle
life applications. This cell system is currently under research and development
for electric vehicles applications.
A revolutionary event in the primary
battery market was the development of
primary lithium batteries in 1973. Lithium
is a very attractive metal for battery applications due to its low atomic mass (6.94), its
high specic capacity (3.86 Ah g1 ), and its
high electrochemical reduction potential
(3.035 V) [10]. Research in lithium batteries started in the late 1950s. It was then
387
388
of the material. The lithium metal electrode formed a rough surface susceptible
to violent reactions. In 1991, Sony commercialized the rst lithium-ion battery,
which uses a carbon intercalation material for the negative electrode instead of
lithium metal. This design tremendously
improved the safety of these batteries by
reducing the lithium metal deposition. Today, the lithium-ion batteries have invaded
the electronic applications market and they
are under research and development for
large-scale applications such as electric vehicles and satellites.
7.4
Battery
Applications
Primary batteries
Secondary batteries
the world, information about their market is used as reference in this analysis.
Figure 4 shows the sales of small-size batteries in Japan from 1992 to 1997 [12].
The sales followed a parabolic growth tendency. In 1997, the total worldwide sales
of batteries were about $14 billion [13],
while the total battery sales in Japan were
$6.74 billion, which represent 48% of the
global battery sales in the world. The percentage left (52%) was distributed among
US, China, Indonesia, Singapore, UK, and
other countries in Europe.
In 1997, 25% of the battery sales in
Japan corresponded to primary batteries ($1.69 billion) and the other 75%
corresponded to secondary batteries
($5.05 billion) [12]. It is expected that in
the coming years the sales in secondary
batteries will increase even more and the
sales in primary batteries will decrease, as
Sales, billion $
10
8
6
4
2
0
1992
1993
1994 1995
Year
1996
1997
389
Others
2%
Manganese dry
batteries
25%
Lead storage
batteries
30%
Lithium-ion
batteries
34%
Nickel-cadmium
batteries
18%
Nickel hydride
batteries
17%
Other alkaline
storage batteries
1%
Fig. 5
1000
Production, million cells
390
Ni Cd
Ni MH
Li ion
800
600
400
200
0
1991
1992
1993
1994
Year
1995
1996
1997
7.5
(3)
e
V
K(s) K+ + e
(+)
()
Hg
HgO
K
OH
K+
KOH/water
Fig. 7
K+
OH
OH
K+
K+
OH
391
392
given by
K(s) KOH/H O
2
HgO Hg(l) Pt
Pt
(5)
(6)
e = e
(7)
e
(8)
e =
K+ K(s) = e
(11)
= 2e
(12)
We have specied all the equilibrium relations. The last step in the derivation is
to connect the chemical equilibrium with
the electrochemical potential; this can be
done using the denition of electrochemical potential. The electrochemical potential
of an ion was dened for rst time by
Guggenheim [16] as follows: the difference
in the electrochemical potential of an ion
between two phases is dened as the work
of transferring reversibly, at constant temperature and constant volume, 1 mol from
one phase to the other. Hence [15]
w = i i = zi F ( )
(13)
F U = e e
(15)
(17)
393
394
(18)
1 0
+ RT
2 Hg
1
0HgO
2
ln(KOH ) 0K
1 0
2 H2 O
(20)
(21)
Pt/H2 (g) H+ /H2 O K(s) Pt (22)
and for the mercuric oxide electrode the
half-cell schematic is given by
Pt/H2 (g) H+ /H2 O HgO(s)
(23)
Hg(l) Pt
Following the procedure described previously (using ideal gas as secondary
reference state for the SHE) it is easy
to show that for the anode half-cell reaction (Eq. 3), the standard cell potential
with respect to SHE is given by
K+
1
0
F U = K + H2 + RT ln
2
H+
(24)
where the superscript * indicates ideal gas
secondary reference state. The tabulated
value for this standard cell potential is
U = 2.925 V [14]. For the cathode halfcell reaction (Eq. 4) the standard cell
potential with respect to SHE is given by
FU =
1
( + 0HgO(s) + 0H2 O
2 H2
0Hg(l) ) RT ln(H+ OH )
(25)
(26)
395
396
Electrode Kinetics
K(s)
K+
OH
K+
OH
OH
Water electrolyte
interface
(27)
si
i
nm
Double layer
F s = e
H
H
H
K+
Fig. 8 Schematic
representation of the double
layer for the electrode reaction
given in Eq. (3). The thickness
of the double layer is of the
order of nanometers.
RT
e
I = io e RT
(28)
where io is the exchange current density
(amperes per square meter) which depends on the composition, temperature,
and nature of electrode surface; a and c
are the anodic and cathodic transfer coefcients, respectively; and I is the current
density (amperes per square meter). The
transfer coefcients are the measure of
the symmetry of the energy barrier of the
transition state [14].
If the surface overpotential is very small
(s 0), Eq. (28) can be approximated by
I = io
(a + c )
F s
RT
The double layer can affect the distribution of the current density. When the
electrode potential changes the current
that ows will be distributed, part of it will
be used to charge the double-layer capacity
and another part will go to the faradic reaction (see Eq. 9). For a given current, the
double layer will decrease the potential of
the cell acting as an internal resistance (a
fraction of the current will go to the double
layer).
The electrode kinetics is measured experimentally. Some common
measurement techniques are impedance
spectroscopy, potential sweep rates, and
intermittent methods. Details about these
techniques are given elsewhere [14].
7.6
(29)
s
RT
I = i e
(31)
o
The transport laws used in electrolytic solutions can be applied to describe the transport mechanisms in a battery. A detailed
description of transport mechanisms in
electrochemical systems is available elsewhere [14, 15]. Some general comments
will be given in this section.
A material balance in a unit cell is given
by
Ci
(32)
= Ni + i
t
where Ci is the concentration of species
i (moles per cubic meter), t is the time
(seconds), Ni is the molar ux of species i
(moles per square meter second), and i
is the homogenous reaction rate of species
i (moles per cubic meter second).
The molar ux of species i can be calculated using Nernst-Plank equation: [15]
Ni = zi ui F Ci Di Ci + Ci
(33)
397
398
(34)
(37)
7.7
Characteristics of Batteries
399
400
7.8
Theoretical Capacity
A-h 1
eq - g Me
(38)
(39)
1
1/Ce,negative + 1/Ce,positive
(42)
401
402
7.8.2
Theoretical Voltage
7.9
Battery Technologies
There are many different battery technologies available but only some of the most
widespread ones will be discussed in this
section. The literature about this topic is
very extensive; only some of the most important aspects have been summarized in
this section. The interested reader is referred to Refs 1, 7, 10 for more details.
7.9.1
Primary Batteries
Cylindrical shape
Button shape
b At
a Data
1.92.0
1.6
1.2
3
Aqueous
3
Aqueous
1
Magnesium
65
1.51.75
1.5
0.9
10
Leclanche
Characteristics
Tab. 2
Aqueous
2
125
38
1.51.6
1.5
0.9
4
Alkaline
manganese
dioxide
Aqueous
5
120
1.6
1.5
1.0
6
Silver oxide
340
1.45
1.5
0.9
3c
Zn/air
Aqueous
3
Primary batteries
230
200
3.5
3.0
2.0
12
Not aqueouse
4
Not aqueousd
5
Li/MnO2
260
3.1
3.0
2.0
2
Li/SO2
404
(43)
(44)
(45)
Leclanches cells are produced in cylindrical and at (prismatic) designs in different sizes. Their energy density at 20 C
is very low (65 Wh kg1 for the cylindrical
size, see Table 2) compared to the other primary battery technologies. However, they
are the cheapest batteries available in the
market (see Table 2). Their nominal voltage is 1.5 V, which is comparable with
other battery technologies (see Table 2).
The applications of Leclanches cells include gas lighter, toys, watches, alarms,
emergency lighting, lanterns, Geiger counters, and mass spectrometers. One of the
main advantages of Leclanches cells is
that they have a long tradition of reliability; other advantages include low cell cost,
large selection of shapes, and diverse sizes,
voltages, and capacities. Disadvantages of
these cells are low energy density, low efciency, and poor storage life. It has also
been reported that Leclanches cells do
not operate well under low-temperature
conditions, they leak easily under abuse
conditions, and their voltage falls steadily
with discharge.
Magnesium Cells
The chemistry of Leclanches cells was
modied by changing the zinc electrode
for a magnesium electrode. This change
brought a new type of cells known as
magnesium cells into the battery market.
Magnesium cells use a manganese dioxide cathode (like the Leclanches cells), a
magnesium anode, and an aqueous electrolyte. Typical electrolyte mixtures include
solutions of magnesium perchlorate with
lithium chromate or magnesium bromide
dissolved in water. Kraft paper is used as
separator. The cell reactions are
2MnO2 + H2 O + 2e
More information about the cell reactions and their mechanisms can be found
elsewhere [1, 7]. For cells with ammonium
chloride as electrolyte, the overall reaction
for light discharge rates can be summarized by
Zn + 2MnO2 + 2NH4 Cl
2MnOOH + Zn(NH3 )2 Cl2 (46)
2MnOOH + 2Zn(OH)Cl
(47)
7.9.1.2
Mn2 O3 + 2OH
(48)
Mg + 2OH
Mg(OH)2 + 2e
(49)
(50)
(51)
Zn + 2OH
Zn(OH)2 + 2e
(52)
(53)
405
406
output, good shelf life, good leakage resistance, good low-temperature performance,
and good shock resistance. Their disadvantages include high initial production
costs and a slightly higher operating cost
than carbon-zinc cells (cylindrical type),
see Table 2.
7.9.1.5
7.9.1.4
Zn + 2OH Zn(OH)2 + 2e
(55)
(56)
Zinc/Air Cells
Zinc/air cells were commercialized in
1932. They use oxygen from the environment to produce electrical energy. Oxygen
diffuses into the cell and is used as cathode,
powdered zinc is used as anode, and an
aqueous solution of potassium hydroxide
is used as electrolyte. Excess of water vapor
can degrade the performance of the cell;
to minimize this situation a hydrophobic
membrane is used as separator. The cell
reactions are
1
O2 + H2 O + 2e 2OH (57)
2
Zn Zn+2 + 2e
+2
Zn
(58)
Zn(OH)2 ZnO + H2 O
(60)
(61)
Their primary market is for applications such as: man-pack radios, radar
systems, electronic watches, and hearingaid devices. Their advantages include high
energy density, at discharge voltage, long
shelf life, low cost, capacity independent
of temperature, and load within the operating range. Another advantage is that the
cathode material does not run out from
the system because it is obtained from the
environment. The capacity of the battery
is limited by the amount of zinc available
in the cell. The cells are environmental
friendly. Their main disadvantage is the
limited power output. In addition, the cells
have a short life once activated if the humidity of the air used is higher than the
equilibrium water vapor pressure of the
electrolyte mixture.
Lithium Batteries
Primary lithium batteries were introduced
into the market in 1973, although research
in this area started 20 years earlier. The
major concern with the use of lithium
in batteries was the stabilization and
passivation of the material. Several properties make lithium a very attractive metal
for battery applications, such as its light
weight, high electrochemical equivalence,
high voltage, and good conductivity. However, lithium is a very reactive material,
especially in the presence of water. Nonaqueous electrolytes are used in lithium
batteries to avoid violent reactions. Even
though the electrolytes are mostly dry,
lithium still reacts with the electrolyte
forming a passivation layer called solid electrolyte interface (SEI). The formation of the
SEI is convenient from a safety standpoint;
it makes the material stable during the operation of the battery and provides a longer
shelf life to the battery. On the other hand,
the SEI causes a decrease in the capacity
of the cell and it signicantly modies the
7.9.1.6
electrochemical behavior of lithium. Kinetics and details about the formation of the
SEI are still unknown.
Lithium primary batteries can be classied into three different categories: soluble
cathode cells, solid cathode cells, and solid
electrolyte cells. Their classication depends on the type of cathode material and
the electrolytes used. Soluble cathode cells
use liquid or gaseous cathode materials
that are dissolved in the electrolyte. Because of the physical state of the cathode,
they have a low resistance, which gives
them a high rate capability. Nevertheless,
the pressure in these cells can go really
high due to the presence of gases. Solid
cathode cells use solid materials as cathodes. They have the advantage of not being
highly pressurized; however, they do not
have the high rate capability that soluble
cathode cells possess. Solid cathode cells
are usually designed to handle medium
discharge rates. Solid electrolyte cells use
solid electrolytes, which give them an extremely long storage life, and practically
eliminate pressure problems in lithium
batteries. However, the internal resistance
of these cells is the highest and therefore
they are capable of handling only low discharge rates. The most common primary
lithium batteries available in the market
belong to the soluble cathode cells and
solid cathode cells categories. A brief description of the most popular designs in
these two categories is given next.
Li/SO2 Cells
Lithium/sulfur
dioxide cells (Li/SO2 ) are perhaps one
of the most advanced lithium battery systems. They belong to the soluble cathode
cells category. Liquid SO2 is used as cathode; a lithium foil is used as anode, and
lithium bromide dissolved in acetonitrile is
used as electrolyte. The active cathode material is held on an aluminum mesh with
7.9.1.6.1
407
408
2Li 2Li + 2e
(62)
(63)
(64)
Li Li + e
(66)
disadvantage is the moderate and low capability rate due to the internal resistance
of the cells. Because of the low discharge
rate limitation, these cells are not available
in large sizes.
Pb + PbO2 + 2H2 SO4
Charge
7.9.2
2PbSO4 + 2H2 O
Secondary Batteries
Discharge
PbO2 + H2 SO4 + 2H+ + 2e
Charge
PbSO4 + 2H2 O
(68)
Discharge
Pb + H2 SO4
PbSO4
Charge
+ 2H + 2e
(69)
(70)
409
35
2.0 1.8
40 to 55
23c
200700
Aqueous
Good
1
3035
1.251.0
40 to 50
1520
300700
Aqueous
Good
2
NickelCadmium
(sealed design)
55
1.31.15
0 to 50
60
15006000
Aqueous
Good
4
NickelHydrogen
Secondary batteries
b At
a Data
Characteristics
Tab. 3
50
1.251.10
20 to 50
20
300600
Aqueous
Good
3
NickelMetal
hydride
90
4.02.5
20 to 55
510
5001000
Not aqueousd
Moderate
3
Lithium-ion
(C/LiCoO2
system)
410
(71)
Pb + H+ + HSO
4 PbSO4 + H2
(72)
The self-discharge reactions cause hydrogen and oxygen evolution, which may
end up causing thermal runaway in the
battery. The rates of the self-discharge
processes (Eqs 71 and 72) depend on temperature, electrolyte volume and concentration, and impurity content (specially
antimony content). More details about the
self-discharge process in these batteries is
given in the literature [7].
NickelCadmium Batteries
Nickelcadmium (NiCd) batteries were
invented in 1899 by Jungner. They use
nickel hydroxide as cathode active material,
cadmium and iron as anode, and an aqueous solution of potassium hydroxide as
electrolyte. Materials such as polystyrene,
nonwoven nylon, and cellulose are used as
7.9.2.2
2NiO(OH) + 2H2 O + 2e
Charge
2Ni(OH)2 + 2OH
(73)
Discharge
Cd + 2OH
Cd(OH)2 + 2e
Charge
(74)
Equations (73) and (74) take place at
the cathode and anode, respectively. In
practice, these reactions are more complex
with the formation of different NiO(OH)
modications and a series of higher nickel
oxides with different degrees of hydration.
The overall reaction is summarized by
Discharge
Cd + 2NiO(OH) + 2H2 O
Charge
Cd(OH)2 + 2Ni(OH)2
(75)
411
412
batteries (vented or sealed) have the following common advantages: long life cycles, reliability, low maintenance, strong
and sturdy, great long-term storage, good
charge retention, and they can be operated at a wide range of temperatures. The
vented sintered plate has 50% higher energy density than the pocket plate. The
sealed type has the advantage of being
closed. Disadvantages of NiCd batteries
include: they are more expensive than leadacid batteries, they require a controlled
charging system to prevent thermal runaway, and they go through the memory
effect.
The term memory effect is used to
describe a reversible process that may
result in a gradual reduction of both
power and capacity of the NiCd batteries with cycling (it may also take
place in Nickelmetal hydride batteries).
The effect takes place following shallow
chargedischarge cycles, that is, the effect is usually noticed when the batteries
are partially discharged and recharged
repetitively without the benet of a full discharge. For instance, suppose that a NiCd
battery is usually discharged in three hours
for a particular application when submitted to complete charge/discharge cycles;
if the battery is not allowed to discharge
completely before charging it, the user will
notice that the next time the battery is
used it will discharge in a time shorter
than 3 hours. The phenomenon is called
memory effect because the battery seems
to remember the lower capacity. The total
capacity of the cell is not affected during
this process, because the process is reversible and can be xed by submitting the
battery to a regular full charge/discharge
cycle. What appears to be happening is
that some morphological change occurs
in the undischarged active material, which
causes the reduction of the working cell
NiO(OH) + H2 O + e
Charge
Ni(OH)2 + OH
(76)
Discharge
1
H2 + OH
H2 O + e
2
Charge
(77)
Equations (76) and (77) take place at
the cathode and anode, respectively. The
overall cell reaction is given by
Discharge
1
H2 + NiO(OH)
Ni(OH)2
2
Charge
(78)
Ni/H2 batteries are built in a particular
shape, especially designed for applications such as geosynchronous earth orbit
(GEO) commercial communications satellites and low earth orbit (LEO) satellites.
Ni/H2 batteries have a high energy density (55 Wh kg1 ) as shown in Table 3,
and they have the best cycle life among
secondary batteries. Another advantage of
these cells is that they can tolerate overcharge. Additionally, the state of charge of
the battery is correlated with the pressure
of H2 , which makes it easy to have an indication of the state of charge of the battery.
The specic technology used in these batteries, their design, and range of operation
make these batteries the most expensive to
build among secondary batteries. Another
disadvantage is that their self-discharge is
proportional to hydrogen pressure. Some
scientists [22] have concluded that the selfdischarge in Ni/H2 batteries is due to a
direct reaction of hydrogen with nickel oxide, and that the reaction rate depends
linearly on hydrogen pressure indicating
that the reaction is rst order with respect
NiO(OH) + H2 O + e
Discharge
Ni(OH)2 + OH
(79)
413
414
MH + OH
M + H2 O + e
Discharge
(80)
Equations (79) and (80) take place at
the cathode and anode, respectively. The
overall reaction is summarized by
Charge
MH + NiO(OH)
M + Ni(OH)2
Discharge
(81)
The overall reaction does not involve
the electrolyte; therefore, the electrolyte
volume does not change in the battery
during discharge and charge cycles, as it
does in Ni/Cd batteries (see Eq. 75).
Nickelmetal hydride batteries are produced in cylindrical, button, and at
designs in different sizes. Some characteristics of these batteries are summarized in Table 3. Ni/MH batteries have
a much higher energy density than Ni/Cd
batteries. Their energy density is about
50 Wh kg1 at 20 C.
Applications of Ni/MH batteries include
computers, camcorders, cellular phones,
communication equipment, variety of
cordless consumer products, high rate
long cycle life applications, electric vehicles
(under development), and so on. Ni/MH
batteries are more environmental friendly
than Ni/Cd batteries, and they are easy to
dispose. Disadvantages of Ni/MH batteries
include lower rate capability, poorer charge
retention, and less tolerance for overcharge
than NiCd batteries. Like Ni/Cd batteries,
Ni/MH batteries are also subject to the
memory effect; a description of this
phenomenon can be found in Sect. 7.9.2.2.
Ni/MH batteries are subject to selfdischarge and capacity fade due to the
reaction of the residual hydrogen in
the cell with the positive electrode, as
well as the slow decomposition of both
Discharge
xLi + xe + Li1x AB
LiAB
Charge
(82)
Discharge
Lix C6
xLi + C6 + xe
(83)
Charge
Lix C6 + Li1x AB
C6 + LiAB
Charge
(84)
Lithium-ion batteries are built in cylindrical and at designs in different sizes.
Table 3 summarizes some of the characteristics of Li-ion batteries. Li-ion batteries possess the highest energy density
(90 Wh kg1 ) and the highest voltage
range (4.02.5 V) among secondary batteries. Other advantages include lightweight,
and no memory effect. Applications of
lithium-ion batteries include cell phones,
laptops, electric vehicles (under development), and satellites (under development).
Disadvantages of these batteries are no
high rate capability (nonaqueous electrolyte increases the internal resistance of
the battery) and capacity fade. One of the
mechanisms associated with capacity fade
is the formation of the SEI layer; more
details about this are given in Sect. 7.9.1.6.
Other mechanisms associated with capacity fade are electrolyte decomposition,
active material dissolution, lithium deposition, electrolyte oxidation, phase change,
and so on. A summary of the different
mechanisms is given by Arora et al. [3].
7.10
415
416
R int
V
+
Analog circuit
representation of a battery.
Fig. 9
circuit representation of a
battery representation has
used in battery modeling.
the battery performance is
the following equation:
battery. This
been widely
In this case,
described by
V = U + Iapp Rint
7.10.2
(85)
417
418
419
420
7.12
List of Symbols
AB
C
Ce
C0
Ci
CT
D
Di
F
io
I
Iapp
Me
Mi
mi
Ni
n
Pi
R
i
Rint
composite material
electrolyte concentration, moles
per cubic meter
theoretical capacity of the
electrode, amperes hour per gram
solvent concentration, moles per
cubic meter
concentration of species i, moles
per cubic meter
total capacity of the electrode,
ampere hour
diffusion coefcient of the electrolyte, square meter per second
diffusion coefcient of species i,
square meter per second
Faraday constant,
96487 C equiv1 .
exchange current density,
amperes per square meter
current density, amperes per
square meter
total applied current, amperes
equivalent weight, grams per
equivalent gram
compound formula or ion
participating in electrode reaction
molality of species i, moles
per kilogram
molar ux of species i, moles per
square meter second
number of electrodes transferred
during electrode reaction
partial pressure of species i,
atmosphere
universal gas constant,
8.314 J mol1 K1
homogenous reaction rate of
species i, moles per cubic meter
second
internal resistance of the battery,
ohms
si
T
t
0
t+
U
U
ui
V
v
v+
We
w
x
zi
stoichiometric coefcient of
species i in the electrode
reaction, dimensionless
temperature, kelvin
time, seconds
transfer number of cations in
the electrolyte relative to the
solvent velocity, dimensionless
open-circuit potential of the
cell, voltage
standard open-circuit potential
of the cell, voltage
mobility of species i, square
meter mole per joule second
potential of the battery, voltage
velocity of the bulk, meters per
second
number of cations in which the
species dissociates
weight of the active material in the
electrode, grams
reversible work transferred in an
electrochemical cell, joules per
mole
lithium intercalation fraction,
dimensionless
charge number of species i,
dimensionless
Greek
electric potential, voltage
a
anodic transfer coefcients for
ButlerVolmer equation,
dimensionless
c
cathodic transfer coefcients for
ButlerVolmer equation,
dimensionless
i
fugacity coefcient, dimensionless
i
the molal activity coefcient of
species i, dimensionless
s
surface overpotential, voltage
i
proportionality constant
expression at secondary
reference state, kilograms per mol
i,g
Subscripts
i
species in equilibrium
+ cation
anion
Superscripts
0 pure phases at the same temperature
and pressure
* ideal gas secondary reference state
References
1. D. Linden, Handbook of Batteries, McGrawHill, New York, 1995.
2. G. G. Botte, Mathematical Modeling and
Thermal Stability of Lithium Ion Batteries,
University of South Carolina, Columbia,
2000.
3. P. Arora, R. E. White, J. Electrochem. Soc.
1998, 145, 36473668.
4. D. Pletcher, F. C. Walsh, Industrial Electrochemistry, Chapman & Hall, New York, 1990.
5. B. A. Johnson, R. E. White, J. Power Sources
1998, 70, 4854.
6. Buchmann I. Batteries in a portable world,
2001, http://www.buchmann.ca/.
7. C. A. Vincent, B. Scrosati, Modern Batteries,
John Wiley and Sons, New York, 1997.
8. Corrosion Doctors, 1999, http://www.
corrosion-doctors.org/Batteries/History.
htm.
9. Encyclopaedia Britannica Incorporation. The
New Encyclopaedia Britannica, Encyclopaedia
Britannica, Inc, Chicago, 1987.
10. R. M. Dell, Solid State Ionics 2000, 134,
139158.
11. T. Sakai, I. Uehara, H. Ishikawa, J. Alloys
Compd. 1999, 293, 762769.
12. Japanese Market News, 2003, http://www.
wtcosaka.com/market/item/batteries.html.
13. Battery Technologies Incorporation, 2006,
http://www.batterytech.com.
14. A. J. Bard, L. R. Faulkner, Electrochemical
Methods Fundamentals and Applications, John
Wiley and Sons, New York, 2001.
421
422
423
429
8.1
8.1.1
Introduction
Energy conversion systems are the principal source of pollution on our planet. In the
industrialized countries, the electric power
plants which convert chemical energy into
electrical energy are the main producers
of CO2 and SO2 emissions; private cars,
on the other hand, are the principal producers of CO and NOx emissions. An
8 Fuel Cells
Load
+
2e
H2
2e
2H+
H2
O2
O2
H2O
H2
O2
Anode
Anode:
Electrolyte
Cathode
2H+ + 2e
H2
H2O
Overall
reaction: H2 + 1/2O2
H2O
as oxidant. Hydrogen and oxygen are supplied to porous electrodes, the gases reach
the electrodeelectrolyte interface and react such that water and an electrical current
are produced.
Because of the direct energy conversion,
fuel cells work at a higher efciency
than energy conversion processes, which
1.0
0.8
Theoretical efficiency
of a H2 /O2-fuel cell
0.6
Efficiency
430
0.4
0.2
0
200
Carnot cycle
lower heat reservoir at 300 K
Carnot cycle
Fuel cell, theoretical
600
1000
Temperature
[K]
70
form protons which move through the gasimpermeable electrolyte. At the cathode,
oxygen is reduced and water is produced,
while electrons ow in an external circuit.
Water is the only by-product of the energy
conversion in a hydrogenoxygen fuel cell.
This water is produced at the cathode in
a fuel cell with proton-conducting electrolytes such as a proton exchange membrane or phosphoric acid, and at the anode
in fuel cells with an alkaline electrolyte or
an oxygen-ion-conducting electrolyte such
as yttrium stabilized zirconia (YSZ).
Figure 1 shows only the electrochemical
core of a fuel cell. However, a single electrodeelectrolyte assembly is not sufcient
for electricity generation, since the typical
voltage of a single cell is below 1 V. In
order to obtain suitable voltages, a number of cells are serially connected. Thus,
in addition to the electrochemical components, a fuel cell needs interconnector
plates, which connect two electrically adjacent electrochemical units and usually also
SOFC
60
PAFC
50
Combined cycle
power plant
Efficiency
[%]
40
Steam power
plant
30
20
Diesel engine
Industrial
gas turbine
Gas engine
10
0
0.1
0.5
10
50
100
500 1000
Efciency of energy conversion as a function of the size of the power plant; conventional
systems are compared with projected fuel cell systems such as PAFC and SOFC.
Fig. 3
431
8 Fuel Cells
120
T = 850 C/O2
T = 950 C/O2
T = 850 C/Air
T = 950 C/Air
100
80
Stack voltage
[V]
432
60
40
20
0
200
400
Current density
[mA cm2]
600
Examples of current
densityvoltage curves for a solid oxide
(see Sect. 8.1.3.2.2) fuel cell stack as a
function of temperatures. These curves
are measured directly. Parameter list:
interconnector CrFe5 Y; anode Ni cermet
(Ni + ZrO2 ); cathode lanthan strontium
manganite (LSM); electrolyte YSZ
800 (ZrO2 Y2 O3 ); fuel H2 (1 bar); oxidant
O2 /air (1 bar); stacks of 80 cells in serial
connection.
Fig. 4
el = id
UF
(1)
el
stack
= el
(2)
8.1.3
433
434
8 Fuel Cells
Tab. 1
Fuel cell
Electrolyte
Electrodes
(anode/cathode)
Interconnector
Concentrate KOH
in porous
matrix
Proton exchange
membrane
Ni/Ag (Pt/Pt) on
porous Ni
Nickel
Pt on graphite
Graphite sheet
Steel
metal (stainless
steel)
Graphite
Inconel
Proton exchange
membrane fuel
cell (PEMFC)
Phosphoric acid
fuel cell (PAFC)
Concentrate
Pt/Pt
phosphoric acid
in porous
matrix
Molten carbonate Molten Li and K
Ni/NiO
fuel cell (MCFC)
carbonate in
porous matrix
Solid oxide fuel
Yttrium stabilized Ni cermet,
SrLaMnO3 ,
cell (SOFC)
zirconia
other
ZrO2 (Y2 O3 )
perovskites
Construction
materials
Graphite, Inconel
Inconel
Inconel, Ni,
ceramic
Chromium based
alloys or
chromium
based ceramics
Ceramic, Ni and
Fe alloys
(3)
Characteristic AFC
PEMFC
PAFC
MCFC
SOFC
160220 C
Pure
hydrogen,
reformate
620660 C
Natural gas,
reformed or
directly fed,
biogas, coal
gas
Air
8001000 C
Natural gas,
reformed or
directly fed,
biogas, coal
gas
Air
Temperature
Fuel
6090 C
Pure
hydrogen
5090 C
Pure
hydrogen,
reformatea
Oxidant
Pure oxygen
Application
Space and
military
Pure oxygen
Air
or air
Space,
Cogeneration Cogeneration or
military, aupower plant
combined
tomotive,
depending on
and
size
stationary
50250 kW
11 MW
2 MW
demonstration
plant
System power
at present
Electrical efciency
Stack (%) 6970
System
62
(%)
a Produced
5068
4358
55
40
Cycle power
plants
100 kW
demonstration
plant
65
6065
54 (cogeneration) >50
(cogeneration)
6065 (combined 6570 (combined
cycle)
cycle)
1
O2 + 2e + H2 O 2OH
2
(reduction of the oxidant at the cathode)
(4)
The mobile ions are the OH ions of
the alkaline solution, which move from
the cathode to the anode. The water is
produced at the anode; it can be removed
together with unreacted hydrogen or with
the circulating electrolyte, depending on
stack design.
Reagents: The AFCs work only with
pure reagents. The presence of impurities
and nitrogen reduces the electrochemical efciency of the cell. The presence
of CO2 , which reacts easily with the alkaline electrolyte to form carbonates, may
be detrimental for the cell, resulting in
435
8 Fuel Cells
1.4
CO + 0.502
CO2
1.2
1.0
CH4+ 2O2
CO2 + 2H2O
H2 + 0.502
H2O
1.0
0.8
CH4+ 2O2
CO2 + 2H2O
0.9
0.6
H2 + 0.502
H2O
0.4
0.8
0.2
CO + 0.502
200
400
0.7
CO2
600
DG/DH
Cell voltage
[V]
436
800
1000
0.6
Temperature
[C]
Fig. 5
known as polymer electrolyte fuel cells (PEFCs), ion exchange membrane fuel cells
(IEMFCs), and solid polymer electrolyte
fuel cells (SPEFCs). The rst prototype of
PEMFC was produced by General Electric
(GE) in 1963 [7]. A 1 kW system was realized for the Gemini mission of NASA
with a sulfonated polystyrene membrane
as electrolyte, and the cell was operated
with pure gases. In 1969, Naon (Du Pont)
was used as electrolyte in a GE fuel cell for
the rst time.
The operating principles of a PEMFC
comprise the following reactions:
H2 2H+ + 2e (oxidation of the
fuel at the anode)
(5)
1
O2 + 2e + 2H+ H2 O
2
(reduction of the oxidant at the cathode)
(6)
Porous conductor
(coal)
+
H2O
O2 (air)
H2
Anode
Proton conductive
membrane (e.g., Nafion)
Cathode
Catalyst
(platinum)
Seal
Fig. 6
437
8 Fuel Cells
Cell voltage
[mV]
1000
800
600
400
200
0
Current densityvoltage
curves of a PEMFC with Pt as
electrocatalyst as a function of
CO content of the fuel; the
poisoning of the electrocatalyst
is evident in the reduction of the
voltage of the cell at constant
current density. Anode:
1 mg cm2 Pt; cathode:
1 mg cm2 Pt; electrolyte:
200 400 600 800 1000 1200 1400 Naon 117; p(H2 ): 2.2 bar
Current density
(abs.); p(O2 : 2.4 bar (abs).
Fig. 7
H2
H2 + 25 L/L CO
H2 + 50 L/L CO
H2 + 100 L/L CO
H2 + 250 L/L CO
1200
[mA cm2]
Power density
[W cm2]
0.8
0.6
0.4
0.2
0
0.12
Conductivity
[S cm1]
438
400
800 1200
Current density
[mA cm2]
1600
Performance plots of
PEM single cells with Pt,
Pt0.7 Ru0.3 , and Pt0.5 Ru0.5
anodes. The performance of
different electrocatalysts in
presence of CO in the fuel are
shown as current
densitypower curves. Anode:
1 mg cm2 Pt or PtRu; Cathode:
1 mg cm2 Pt; Electrolyte:
Naon 117; Fuel gas:
2000 H + 100 ppmv CO;
2
Temperature: 80 C; (pH2 ): 2.2
bar (abs.); (pO2 ): 2.4 bar (abs).
Fig. 8
Nafion 117
0.08
0.4
0
10
20
N(H2O)/N(SO3H)
30
Electricity Heat
CH3OH or
other fuels
(natural gas)
POR
Partial
oxidation
reformer
H2
HSR
PEMFC
Heated
steam
reformer
H2O
Air
DMFC
Fuel processing
Electricity Heat
Fig. 10
Two different routes for using methanol as fuel for a fuel cell.
At the beginning of the 1980s, the interest in DMFC arose and methanol was used
as fuel for high-temperature fuel cells [12].
Nowadays, PEM fuel cells with Naon as
electrolyte appear to be well suited for the
direct oxidation of methanol. There are,
however, some major problems in adapting a PEMFC to operate with methanol.
The catalytic material of the anode has to be
improved in order to avoid the loss of activity because of the formation of by-products.
As in the case of a PEMFC using hydrogen from a reformer as fuel, a method
to improve the anodic material is the
use of PtRu or PtRuSn mixtures [8].
A particular problem of the DMFC is
cross-diffusion of methanol through the
electrolyte.
A possibility of avoiding the poisoning
of the catalysts is to operate at a higher
temperature as long as the electrolyte is
stable and the cross over of methanol is
tolerable. The interaction between Naon
and methanol is complex; methanol causes
an increase of the ohmic drop of the
439
8 Fuel Cells
1000
H2 at 80 C
CH3OH (liquid,
CH3OH (vapor,
CH3OH (vapor,
CH3OH (vapor,
Current
densityvoltage curves for
PEMFC operating with
hydrogen and methanol. The
efciency of a PEMFC operating
as direct methanol fuel cell is
about half of that of a PEMFC
operating with pure hydrogen.
Fig. 11
80 C)
120 C)
120 C)
130 C)
800
Cell voltage
[mV]
440
600
400
200
0
10
100
Current density
[mA cm2]
1000
H2 2H+ + 2e
(oxidation of the fuel at the anode) (9)
1
O2 + 2e + 2H+ H2 O
2
(reduction of the oxidant at the cathode,
air is the most common oxidant)
(10)
The mobile ions are the H+ which move
from the anode to the cathode through
100% phosphoric acid immobilized in
a SiCpoly(tetrauoroethylene) (PTFE)
H2 + CO2
3 H2 O + CO2 + 2e
(11)
1
O2 + CO2 + 2e CO2
3
2
(reduction of the oxidant at the cathode;
air is the most common oxidant)
(12)
H2O
H2
H2O
CO
Ni anode
H+
H+
O2
Cathode
Electrolyte
Fig. 12
441
442
8 Fuel Cells
CH4 + H2 O 3H2 + CO
(13)
3H2 + 2CO2
3 3H2 O + 3CO2 + 6e
(14)
CO + CO2
3 2CO2 + 2e
(15)
H2O, CO2
Fuel
electrode
Electrolyte
(ZrO2/Y2 O3)
O2
O2
O2
O2
Air electrode
Air
N2
Fig. 13
Fig. 14
H2 + Ox0 H2 O + Voo
0 + 2e
H2O
YSZ
Oox
oo
VO
443
444
8 Fuel Cells
Nickel felt
Fuel
Fuel
Oxidant
Nickel
plated on
interconnector
Oxidant
Interconnector
Nickel
felt
Anode
Electrolyte
Oxidant
Oxidant
Fuel
Cathode
Porous
support tube
Nickel felt
Fig. 15
0.05 mm
0.01 mm
0.05 mm
Electrolyte
0.10 mm
0.05 mm
Anode
50 50 mm2
1.50 mm
..
Conventional at-plate (Siemens, left) and planar (Research Center J ulich, Germany, right)
SOFC stack concepts.
Fig. 16
Electrochemical
reactions
H2
CO
Steam reforming
Shift reaction
CO H2O
CH4 CO CO2 H2
H2O
Zone for
reforming
reaction
Anode
CO
CO2 H2 H2O
Electrochemically
active area
Electrolyte
Fig. 17
445
446
8 Fuel Cells
References
1. M. Cappadonia, U. Stimming, K. V. Kordesch et al., Ullmanns Encyclopedia of Industrial
Chemistry, Wiley-VCH, Weinheim, 2003,
Vol. 15, pp. 95117.
2. K. Kordesch, G. Simader, Fuel Cells, VCH,
Verlagsgesellschaft, Weinheim, 1996.
3. K. Ledjeff (Ed.), Brennstoffzellen, C.F. Muller,
Heidelberg, 1995.
4. R. B. Ferguson, Pratt & Whitney Aircraft,
Design and Development of H2 /O2 -Fuel cell
power plants for Apollo Command Module, NASA/Houston Final report 6/16/1962,
Apollo Fuel Cell System, Proceedings of 23rd
Annual Power Sources Conference, 1969,
pp. 1113.
5. M. Schautz, G. Dudley, Testing of Space fuel
cells at ESTEC reprinted article from ESA
Bulletin no. 60, 1992.
6. O. Fuhrer et al., Int. J. Hydrogen Energy 1994,
19(3), 343348.
7. R. Cohen, Gemini fuel cell system, Proceedings 20th Power Sources Conference, 24-26
May 1996, pp. 2124.
8. A. Aramata, M. Masuda, J. Electrochem. Soc.
1991, 138 1949.
9. Brochure DuPont Naon peruorinated
membranes, DuPont, Fayetteville, 1994.
10. T. E. Springer, T. A. Zawodzinski, S. Gottesfeld, J. Electrochem. Soc. 1991, 138 2334.
11. T. E. Springer, M. S. Wilson, S. Gottesfeld, J.
Electrochem. Soc. 1993, 140 3513.
12. D. J. Wheeler et al., Proceedings Workshop
on Direct Methanol Air Fuel Cells, The
Electrochemical Society, Pennington, 1992,
p. 193, Vol. 92-14.
13. M. Waidhas et al., Research and development of low temperature fuel cells at
Siemens, 1994-Fuel Cell Seminar, San
Diego, 1994, p. 477.
14. L. G. J. Haart, J. Divisek, E. Riensche,
U. Stimming, EP-A 1 0 829 103, 1998.
8.2
8.2.1
Introduction
447
448
8 Fuel Cells
Anode
Cathode
H2, fuel
(a)
O2, air
FF
GDL CL
PEM
Fundamental understanding:
structure formation and
transport in PEM and CL
Structure vs function
CL GDL FF
Materials research:
New membranes (high-T)
and catalysts
Advanced design?
Theory
Novel tools, criteria
Diagnostics:
Performance evaluation,
in situ vs. ex situ
Global optimum?
Engineering:
assembling, testing,
operation conditions
(b)
(a) Principal layout of a PEM fuel cell with the main functional
components, viz. proton-conducting polymer-electrolyte membrane
(PEM), catalyst layers on anode (ACL) and cathode sides (CCL),
gas-diffusion layers (GDL) and ow elds (FF). (b) Disciplines in fuel
cell research and how they are connected by the theory.
Fig. 1
449
450
8 Fuel Cells
8.2.2
The Membrane
8.2.2.1 Structure and Function of the
Membrane
Currently, the best proton-conducting materials contain water as the vital component, utilizing the advantages of the
anomalous proton mobility in hydrogenbonded water networks [2022]. PEMs are
such materials, used as the key components in PEFC and DMFC [23, 24]. Their
function is to provide a stable separation
of gaseous reactants on the anode and
cathode sides and good proton conductivity. State-of-the-art PEMs fulll these
demands due to a bifunctional composition: a solid matrix of polymer backbones
is responsible for stability; a network of
water-containing pores, emerging in this
matrix, provides pathways for protons.
There are practically no, or very few,
ultrasmall pores available to gas permeation, since pores open up mainly upon
the membrane water uptake and swelling
that are driven by hydration of hydrophilic
residues. The aqueous pathways for protons lead to the unfavorable side effect of
molecular diffusion of water or any other
small molecules dissolvable in water, such
as, for example, methanol.
The effect of polymer morphology on
membrane structure and conductance has
been shown recently. In Ref. 25 hydrogenbased graft-copolymer membranes were
compared in terms of morphology and
performance to random copolymer membranes with the same ion content. For the
hydrated grafted membranes TEM micrographs revealed a picture of a continuous
phase-separated network of water-lled
channels with diameters of 5 nm. In
contrast to that, the random copolymer
membranes exhibit a reduced tendency
toward microphase separation; water is
451
452
8 Fuel Cells
5 nm
Network of pores
PTFE (hydrophobic)
1 nm
SO3
SO3
H+
H+
SO3
SO3
Single pore
H+
H+
Acidwater cluster
There exists a trade-off between the expense at which protons are available and
the ease with which they move in different pore regions. Proton concentration
decreases from the surface to the center,
but their mobility is highest in the pore
center and low in the proximity of anionic
groups. The distributions of proton density
and mobility within the pore determine the
dependence of proton conductivity on pore
radius. This dependence was studied in
detail in Refs. 40, 43. Increase of the pore
radius shifts the balance from surface-type
to bulk-type conductivity.
On the largest length scale, top picture
of Fig. 2, the distribution of water in the
membrane is depicted as a porous network. The latter is characterized by a pore
size distribution (psd) and a tortuousity
factor , which accounts for multiple interconnectivity and bending of pathways in
the network. The distribution of pore radii
translates into a distribution of pore conductivities. Via this correspondence, the
distribution of water in the membrane nally determines its transport properties,
namely, proton conductivity and water diffusivity.
Since PEM, like living organisms, need
water to function and the amount and
state of water are critical for an efcient
operation, secondary requirements on this
type of fuel cell membranes emerge. These
include the necessity of sufcient humidication and the ability to retain water
under operation conditions. Associated
problems under fuel cell operation include the electroosmotic transport of water
to the cathode side accompanied by dehydration at the anode side [45]. In the
cathode the accumulation of water at high
current densities, typically greater than
1 A cm2 , causes performances losses due
to blocking of catalytically active sites and
restriction of oxygen transport.
Membrane Conductance
Several models of the random pore space
and its evolution upon membrane swelling
are suggested in Refs. 39, 40 to bridge the
gap between the conductivity of a single
pore and the conductance of a membrane.
Combined with experimental data on psds
they reveal the features that cause good or
bad membrane performance.
It was possible to rationalize the family of
Arrhenius plots measured for Naon 117
at different water contents [46]. Under an
assumption that the surface conductivity
has higher activation energy, supported
by microscopic considerations in Refs. 40,
43, the Arrhenius slope should become
steeper with the decreasing amount of
water in the membrane [39], that is,
the smaller the amount of the bulk
water that we have in pores. Activation
energies obtained from these plots are
0.1 eV for the largest possible water
contents (Activation energies of proton
transfer in water, estimated from nuclear
magnetic resonance relaxation times, are
0.1 eV [47].) and 0.30.4 eV at small
water contents. How to rationalize this
variation?
8.2.2.2
453
454
8 Fuel Cells
equation [48],
Ga =
(Er
+ G)2
4Er
(1)
455
456
8 Fuel Cells
457
458
8 Fuel Cells
Fig. 3
459
460
8 Fuel Cells
461
462
8 Fuel Cells
Load
Anode
H2
H2O
H+
jel-osm
jp
Cathode
O2
Electroosmotic drag
jwa
jwc
jw= jel-osm + jhydr
Pag
Pcg
jhydr
Hydraulic flux
used to simulate the membrane performance inside the cell (cf. Fig. 4).
For this purpose the electrochemically
active electrode layers at x = 0 (anode side)
and x = L (cathode side) are considered as
innitely thin. Gas transport in the backing
layers is assumed to be facile (high permeability). Inhomogeneous distributions of
feed gases along the electrode planes may
thereby be ignored. All local properties,
thus, depend only on the coordinate x
perpendicular to these boundary planes,
and the problem is reduced to a onedimensional one. Moreover, heat transfer
phenomena are ignored, assuming that
thermostatic conditions are externally provided.
The heterogeneous model of PEMs implies the existence of a water-lled porous
network, which reorganizes upon water
uptake. This reorganization has two major
impacts on transport properties: the increase of single pore cross-sectional areas,
available for proton and water transport,
463
464
8 Fuel Cells
and the evolution of orientation and topology of pores in the network. Therefore,
the phenomenological description of proton current and water management relies
on the following experimental information: (1) water content w as a function of
liquid pressure P l in the pores (or more
generally as a function of the free energy of
water binding with the membrane material), (2) pore-radius distribution w(r) [60,
81], (3) electrolyte conductivity as a function of water content, el (w) [39, 46, 85],
(4) electroosmotic drag coefcient, n, that
is, the number of water molecules transported through the membrane together
with each proton, and its dependence
on w [86, 87]. The dependencies (1) and
(2) determine mutual relations between
water content, membrane morphology,
and water transport.
Concept of Capillary Pressure
Pore-radius distributions and ab-/
desorption isotherms are important structural characteristics of generic porous
media [80, 88]. The absorption isotherm
provides a relation for the liquid uptake
of a porous medium under controlled external conditions, viz., the pressure of an
external uid. Within a bounded system,
such as a cylindrical tube, a discontinuity
of the pressure eld across the interface
between two uid phases exists. The corresponding pressure difference is called
capillary pressure, P c . In the case of contact
between gas phase, P g , and liquid water
phase, P l , the capillary pressure is given by
8.2.2.7.1
Pc = Pg Pl
(2)
AVm cos
r
(3)
(4)
P g = Pc Pa
(5)
and, therefore,
x
g
wx = f Pxl + Pc P g 1
L
(6)
The external gas pressure gradient will,
thus, be superimposed as a driving force
of water ux on the internal pressure gradients that develop due to electroosmotic
effect and capillary forces. Note that gas
pressures in anode and cathode chambers
can be controllably varied.
465
466
8 Fuel Cells
Basic Equations
In the following sections, internal mechanisms of
water ux in the membrane due to capillary forces will be explored, considering
P g = 0. The effect of nonzero P g is
presented separately in Sect. 8.2.2.8.2 Two
equations determine the distribution of
water in the membrane and its effect on
currentvoltage performance under operation conditions.
The balance equation of water ow is
determined by
8.2.2.7.2
K(wx ) dPxl
dx
(7)
(10)
(11)
del
dx
(12)
Substituting dP l /dx from Eq. (7) and using (del /dx) = (del /dPxl )(dPxl /dx) we
obtain
jp
del
Fcw K(wx )
=
el (wx ) n(wx )jp jw
dPxl
(13)
Note that convection of protons with
the ow of water has been neglected in
Eq. (12). (The effect of a convection term
in Eq. (12) was considered in Ref. 16.
The net effect due to proton convection
is a slight worsening of the effective
membrane conductivity.)
The following percolation dependence
of proton conductivity will be used for
deniteness,
with
g(jp , rx ) =
b
[w(r) wc ](r rc )(r)
1
dr
,
r1 ws rx
[n(w(r)) jw /jp ]r 2
Analytical Solution
Only the
essential parts of the solution will be presented in this section. A more detailed
consideration including the discussion
of analytical expressions for the asymptotic behavior can be found in Refs.
16, 83.
The general solution of Eqs. (7) and (13)
is given by
8.2.2.7.3
L
(15)
g(jp , rx )
Lx
L
U = Jm G(jp , a), U = el (L) el (0)
0
j p = Jm
(16)
G(jp , a) =
b
[w(r) wc ](r rc )(r)
0
dr
,
r 1 ws a
el (w(r))[n(w(r)) jw /jp ]r 2
(17)
valid for n(w(r))jp > jw (locally).
rx is the radius up to which pores are lled
in the x-cross section. In pores with these
radii the capillary equilibrium prevails. On
the cathode side pores are lled up to the
radius rL = b, corresponding to the saturation water content. At the anode side we
have r0 a. This solution is valid for r >
rc in all REVs, where rc is the pore radius at
the percolation threshold wc . r1 is the rst
moment of the pore-radius distribution,
1
dw(r)
drr
(18)
r1 =
ws 0
dr
An important membrane parameter Jm
(in units A cm2 ) is dened by
Jm =
F c w ws r 1
4
L
(19)
(20)
467
8 Fuel Cells
Fig. 5 Pore size distribution function of
microporous membranes, parameterization with logarithmic normal
distribution, for the parameter sets
specied in Table 4. (a) Differential psd,
(b) cumulative psd. Experimental data in
(b) have been measured by the method
of standard porosimetry [44] for Naon
117 at 30 C. Curves correspond to the
set of parameters in Table 1.
1.0
1.0
0.8
0.8
0.6
0.4
0.6
0.2
0.0
0.4
0
log(r )
[nm]
1
(1)
(2)
(3)
0.2
0.0
0
10
15
20
25
r
[nm]
(a)
1.0
0.8
w/ws
468
0.6
0.4
0.2
0.0
0.5
0.0
0.5
1.0
1.5
2.0
2.5
log(r )
[nm]
(b)
log(r/rm ) 2
ws
dw(r)
exp
=
dr
log s
r
log(rmax /rm ) 2
exp
rmax
log s
(21)
Parameters of the psd. Values of the critical current density jpc are calculated on the basis of
the membrane parameters specied in Table 2
Tab. 1
Set
rm
[nm]
rmax
[nm]
r1
[nm]
ks
[nm2 ]
rc
[nm]
jpc /Jm
jpc
[A cm2 ]
(1)
(2)
(3)
(4)
(5)
1.0
1.0
1.0
1.0
1.0
0.15
0.30
0.60
0.15
0.15
100
50
14.9
3.0
1.2
10.8
8.2
1.2 103
1.6 10
1.5
2.7 102
1.2 102
1.08
0.69
0.72
1.05
1.00
0.67
0.65
0.66
0.45
0.40
5.0
1.0
0.4
2.5
1.7
h(rx )
1
=
r 1 ws
H (a)
=
0
r 1 ws
dr
dr
b
rx
b
a
(22)
and the parameter
jpc =
1
{jw + Jm h(rc )}
n
(23)
jp
L
Jm
H (a)
0
njp jw
(25)
469
8 Fuel Cells
njpc jw jp
Uc
njp jw jpc
L njp
(jp jpc )
r njp jw
(27)
0.00
0.05
U/Jm
[VA1cm2]
470
0.10
0.15
Ohm's law
(1)
(2)
(3)
0.20
0.25
0.0
0.2
0.4
Jp /Jm
0.6
Fig. 6
Tab. 2
Thickness
Conductivity parameter (Eq. 12)
Residual conductivity (Eq. 12)
Saturation water content
Percolation water content
Surface tension
Viscosity
Inverse tortuousity
Concentration of water
Electroosmotic coefcient
L = 0.02 cm
0 = 0.07 Scm1
r = 0.0007 Scm1
ws = 0.4(vol. fraction)
wc = 0.1ws
= 0.05 Jm2
103 kgs1 m1
= 1/3
cw = 55 mol l1
n = 1.2
(28)
471
8 Fuel Cells
1.0
0.8
0.6
w/ws
472
0.4
jp /Jm = 0.109
jp /Jm = 0.288
jp /Jm = 0.390
jp /Jm = 0.441
jpc /Jm = 0.454
0.2
0.0
0.000
Anode
0.005
0.010
x
[cm]
0.015
0.020( = L)
Cathode
Fig. 7
Moreover, a pronounced effect of the maximum pore size is evident from this table.
The existence of large pores facilitates water management and thereby gives larger
values of jpc .
This conclusion is straightforward as far
as the electroosmotic drag coefcient is
a weak function of r1 . If it decreases
dramatically with the decrease of r1 , the
conclusion will depend on the competition
between the drag and permeation coefcients. If, however, the whole variation of
the drag coefcient is between 1 and 1.4,
the conclusions made will stay valid.
8.2.2.8.2 Modes of Water Management in
The model gives hints
the Membrane
for water management. Supplying an
appropriate ux of water from the anode
side to compensate the electroosmotic ux,
that is, positive jw (Eq. 23), will increase
the critical current density. Alternatively,
one may apply a gradient in gas pressure,
P g /L, as explored in Refs. 16, 83. The
effect of this external pressure gradient
is superimposed linearly on the internal
driving force provided by the capillary
pressure gradient.
Indeed, the optimum membrane performance is obtained under saturation
conditions, where
d(Pxc )
= 0,
dx
P g = P l ,
wx = ws
(31)
and
U =
=
L
jp for jp < jps
s
jw
Fcw Ks P g
+
n
n
L
(32)
L
jp
Fcw K(ws )
(33)
(35)
473
474
8 Fuel Cells
jw = n(wx )jp
L
dwx
(36)
J D(wx )
ws m
dx
del (x)
dx
el (wx ) n(wx )jp jw del
=
LJm
D(wx )
dwx
jp = el (wx )
(37)
Structural characteristics of the membrane are now incorporated into the transport coefcients D(wx ), n(wx ), el (wx ).
Formally, the generalized model looks like
jp =
dwD(w) ,
j w + Jm
n
1 (x/L) wx
ws
jp
D(w)
dw
U = LJm
(38)
njp jw wa
el (w)
(for jp > jpc it is the same as in Eq. (27))
The critical current density, dened via
ws
1
dwD(w)
(39)
jpc =
j w + Jm
n
wc
is proportional to the area under the
D(w) curve within the limits w = wc and
w = ws . The task of physical models and of
ex-situ diagnostics is to determine detailed
D(w) dependences.
8.2.2.9.2 Hydraulic Permeation versus Diffusion and Comparison with Experiments
In earlier approaches to the problem of
membrane water management, an activitydependent diffusion coefcient was considered which was transformed into a
w-dependent diffusion coefcient, D (in
cm2 s1 ) [7]. Within the diffusion approach
the nondimensional effective coefcient
D is related to the dimensional diffusion
constant D (in cm2 s1 ) via the identity
D = Fc w D/L.
Jm
w s
Parameterizations of effective
dimensionless diffusion constants D(w)
in the model of hydraulic permeation
(cf. Eq. 2.40), using parameterization
(4) in Table 1) and in the diffusion
model. In the latter, D is obtained from
the dimensional diffusion constant D
(in cm2 s1 ) via the identity
Jm D = Fcw ws D/L. Absolute values have
been adjusted in such a way, that both
parameterizations will give the same
value of jpc /Jm .
Hydr. perm.
Diffusion
Fig. 8
0
0.0
0.2
0.4
0.6
w /ws
0.8
1.0
475
8 Fuel Cells
0.0
0.2
0.4
U/Jm
[VA1cm2]
Hydr. perm.
Diffusion
0.6
0.8
1.0
0.0
0.1
0.2
0.3
0.4
jp /Jm
(a)
1.0
0.8
0.6
w/ws
476
0.4
jp /Jm = 0.109
jp /Jm = 0.288
jp /Jm = 0.390
jp /Jm = 0.441
jpc /Jm = 0.454
0.2
0.0
0.000
(b)
0.005
0.010
x
[cm]
0.015
0.020
Fig. 9
=
dx
0
1
el (wx )
(41)
1.3
Exp.
Hydr. perm
Diff.
/s
1.2
2
1
0.0
1.1
0.5
1.0
1.0
0.0
0.2
0.4
0.6
0.8
1.0
jp
[A cm2]
Fig. 10
477
478
8 Fuel Cells
polymer host. This will fortify the diffusional component of water backow.
8.2.2.10 Membrane Performance
The key concept used in
Guidelines
this chapter on membrane performance
in PEFC is the perception of presentday PEMs as phase-separated systems. A
hydrophobic phase of polymer backbones
that provides mechanical stability and
a hydrophilic phase of water-containing
pathways for proton and water mobility
are distinguished.
This notion is supported by a large
number of independent experimental data,
related to structure and mobility in these
membranes. It implies furthermore a distinction of proton mobility in various water
environments, strongly bound surface water and liquidlike bulk water, and the
existence of water-lled pores as network
forming elements. Appropriate theoretical treatment of such systems involves
random network models of proton conductivity and concepts from percolation
theory, and includes hydraulic permeation
as a prevailing mechanism of water transport under operation conditions. On the
basis of these concepts a consistent approach to membrane performance can be
presented.
The main conclusions for membrane
water management are:
479
480
8 Fuel Cells
Carbon grains
Pt nanoparticles
Ionomer
PEM
GDL
1020 m
Micropores
(510 nm)
Fig. 11
Mesopores
(10100 nm)
(b) Agglomerate
50 nm
Ionomer
- proton conductor
- binder
Reaction spots
- nonideal
- optimal
the creation of a large electroactive contact area between carbon/Pt particles and
electrolyte particles.
Because of the intricate composition,
various transport modes contribute to the
supply of O2 to the reaction sites, involving
viscous ow, Knudsen diffusion and ordinary molecular diffusion [103]. The relative
importance of the distinct mechanisms depends decisively on the psd. Dominating
transport though micropores favors Knudsen ow, whereas molecular diffusion with
considerably larger diffusion constants will
I=
4F PO2 D
,
RT l
(42)
Here, the oxygen partial pressure p is normalized to the absolute O2 -partial pressure
PO2 at the interface between catalyst layer
and GDL (at = 1), p = PO2 /PO2 . D is
an effective oxygen diffusion constant (in
cm2 s1 ). jp () is the local proton current
density (in A cm2 ) and j0 = jp ( = 0)
is its value at the interface with the membrane, where j0 is equal to the total current
density through the cell.
The parameter I (in A cm2 ) is a characteristic current density of oxygen diffusion.
The factor 4 in the denition of I accounts for the number of protons that
are consumed by one oxygen molecule
in the cathode reaction. Equation (42) is
based on Knudsen diffusion as the prevailing mechanism of oxygen transport in a
predominantly microporous catalyst layer,
that is, oxygen molecules are assumed to
collide more frequently with pore walls
than with other molecules [104]. A variant which, instead, considered molecular
diffusion as the prevailing mode of oxygen transport was studied in Ref. 17. This
would be more realistic in a layer with a
considerable porosity due to macropores.
Both variants give similar results. Analytical expressions are, however, simpler in
the variant presented here.
Proton transport in the layer is described
by the migration equation
d()
1
= jp (),
d
el
l
(43)
481
482
8 Fuel Cells
= i p() exp
d
RT
(1 c )nt F ()
(44)
exp
RT
Here, nt is the number of electrons that
are transferred in the rate-determining
reaction step. At a Pt| PFSI interface nt = 1
has been identied [99, 110]. The standard
exchange current density (in A cm2 )
i = 2F S0 lk (cH+ ) cO2
is an effective reaction rate parameter,
determined by the externally provided
reactant concentrations. The factor 2 accounts for the number of electrons that
are transferred in one elementary reaction.
Moreover, i is proportional to the total effective surface S0 of the electroactive Pt|
PFSI contact area per unit volume (unit
cm1 ) and to the reaction rate constant
k . The dimensionless property = S0 l
corresponds, thus, to the real-to-apparent
surface area ratio.
Due to the assumption of electroneutrality in any REV of the layer, the proton
(45)
whereas at 2b it is conventionally
approximated by the Tafel expression,
djp ()
()
= i p() exp
,
d
b
b=
RT
c nt F
(46)
djp ()
() d()
1
,
= i
= jp (),
d
d
b
RT
b =
(48)
nt F
(47)
l
(1 )
lp
j0 ,
jp () =
l
sinh
lp
l
cosh
(1 )
lp
1
j0 (49)
() =
l
p
sinh
lp
sinh
483
484
8 Fuel Cells
i 1/2
b
b
i
(52)
1/2
(53)
8.2.3.4
dp
= (1 i)
d
(54)
d
= gi
d
(55)
di
= p
d
(56)
j0
i
b ln
I
I
(59)
( = 0) = 1
(60)
gives j0 and,
j0
0 = =
I
(62)
(63)
(64)
(65)
2 bi exp
2b
c l
1
(1 ) exp
tan
(66)
2 lp
2b
jp () =
8.2.3.4.3
2
d2 ()
()
l
=
b
exp
dx 2
lp
b
For xed
p( = 0) = 1, i( = 0) = 0,
0 = b ln 0 + 2b ln
where the working point is xed by specifying a value 1 , the electrode potential
at = 0. The potential losses in the catalyst layer are obtained by putting = 0 in
these equations and calculating the corresponding 0 = (0) and j0 = j
p (0) at the
given 1 . Note that the factor 2 bi is
independent of l, since l 1 and i l.
An alternative way toward this solution
was pursued in Ref. 108. Differentiating
Eq. (46) with respect to and substituting
d/d from Eq. (43) an expression leads to
d
d
djp
1
+
jp 2
d
2 b
=0
(67)
485
486
8 Fuel Cells
and 0 [108],
j0
2 b
j 2
c l 2
0
0
=
exp
2 lp
b
2 b
2
j 2
l
c
0
0
tan
exp
2 lp
b
2 b
(69)
It is not possible to resolve this transcendental equation for the explicit 0 (j0 )
dependence. However, two limiting cases
can be conveniently studied: in the limit
j0 2 b the simple Tafel kinetics is
recovered,
0 = b ln
j0
i
,
j0 2 b
(70)
j0 2 b
(71)
Note that according to Eq. (65) the latter
limit is attained when the local electrode
10
j0
2sb
lp
1,max = 2b ln
2c l
j0
0 1 = 2b ln
,
b
i
d2 p()
0
=
exp
p()
d 2
I
b
0.1
0.01
0.01
j0
h
i*
=
exp ( 0 )
2sb 2sb
b
0.1
1
10
i * exp h0
(b)
2sb
100
j0 2 b
(73)
Figure 12 shows the currentvoltage
plot of the catalyst layer in loglog
coordinates. The transition from j0 =
i exp(0 /b) at small current densities to
j0 = 2 bi exp (0 /2b) at large current
densities is clearly seen in this plot. Near
j0 = 2 b the transition region described
by Eq. (69) couples the two limiting
straight lines.
h
j0
i * exp 0
=
( )
2sb
2sb
b
(72)
(74)
(75)
and
j0 = I 0 tanh(0 )
(76)
with
j0
p() = exp (1 ) ,
I
j0
jp () = j0 1 exp (1 )
I
(78)
487
8 Fuel Cells
0.4
Oxygen depletion
[V]
Intermediate
4
Kinetic
3
0.3
0.2
h0
2b ln (j0 /i )
ln(G0)
0.1
0.0
1
0
(a)
j0 /i
1.0
0.8
j /j0
0.6
1 (0.001)
0.4
0.2
0.0
0.0
2
3
4
5
6
0.2
(0.5)
(1.0)
(1.6)
(2.5)
(4.0)
0.4
(b)
Dimensionless plot of
polarization characteristics of the
catalyst layer and proles of current
density and pressure along the thickness
coordinate . Here, these properties are
depicted in the mixed regime, when
proton and oxygen transport limitations
are of equal importance, I = b (g = 1).
(a) The potential drop o and particular
contributions according to Eq. (81)
(2b ln(j0 /I), b ln c ) are shown, helping
to distinguish the three indicated
regimes of performance. (b) The proles
of jp /j0 and (c) p are depicted at the
points 16 marked in (a). Local
distributions in electrode potential,
which are also calculated in the model,
are not shown here. The general trend
observed from them is an increase
towards the membrane interface. At
point 6 the total increase is 2b.
Fig. 13
0.6
0.8
1.0
0.6
0.8
1.0
x/l
1.0
0.8
0.6
488
0.4
0.2
0.0
0.0
(c)
0.2
0.4
x /l
(80)
These voltage losses are most pronounced
for low proton conductivity, that is,
small . The ohmic contribution ( 1 )
results from a simple fact: if there is a layer
close to the membrane with no oxygen,
the protons must pass this layer (without any reaction) to reach another layer,
close to the backing layer, where there is
oxygen to react with. Actually, at current
densities j0 I the catalyst layer performs
as if it had one exible boundary at =
1 eff / l. While the boundary on one side,
= 1, remains xed, the boundary at =
1 eff / l moves away from the membrane
interface with increasing current density.
The electrode potential for eff l is
() =1eff / l +
0 b ln c + 2b ln
j0
j0
i
+
b ln
I
I
(81)
489
490
8 Fuel Cells
j0
0 (j0 ) = 2b ln
2b ln
I i
j0
= 2b ln
I i
j0
= 0 (j0 ).
= 2b ln
I i
(82)
(83)
(84)
Fig. 14
10
8
Oxygen depletion regime
6
4
Intermediate regime
Range of optimum
performance (k = 1)
2
0
Kinetic regime
0.0
0.5
1.0
1.5
2.0
j0 /I
Composition Dependence
In Sect. 8.2.3.5 we have discussed how
for the given parameters that characterize
ionic transport, gas transport, and reactivity of the catalyst layer, and a specied
target current density, one can choose the
layer thickness, which provides minimal
voltage losses. The parameters, however,
depend on composition, which can be
varied to optimize the performance. By
composition one may imply the chemical composition of the components. Such
variation is a subject of material chemistry.
Here we will discuss only the variation of
the relative amounts of the distinct components (carbon, Pt, PFSI, PTFE) which is
a subject of physics of composites.
In spite of the widely recognized importance of an advanced catalyst layer
design, detailed structural data for catalyst layers are still scarce in the open
literature on fuel cells [116, 117]. In one
of the rare experimental studies, Uchida
et al. showed the effect of the variation
of the PFSI (and PTFE) content on catalyst layer performance [101]. An attempt
to rationalize the experimentally observed
composition dependence theoretically was
rst undertaken in Ref. 17. The prerequisites for an adequate theoretical study
8.2.3.6
491
492
8 Fuel Cells
Xc
(85)
poly
r 4
1
r 2 2
(86)
(Xel Xc )
(Xel Xc )
(1 Xc )
(87)
1 Xc
Is is the diffusion parameter in the absence
of any network constraints, say the one
at very high porosity. The exponent
here can not be exactly the same as in
Eq. (78), but we will take the same value for
X
[1 + exp(a(X Xc ))]b
(90)
493
8 Fuel Cells
Dependence of basic catalyst
layer parameters on composition,
according to Eqs. (8789). The
percolation-type dependencies on
electrolyte content Xel are depicted for
the normalized parameters of (a) proton
conductivity, , and oxygen diffusivity, I,
as well as (b) for the exchange current
density at the indicated values of ec
(residual reactivity), the percolation
threshold Xc = 0.1, and the residual
diffusivity parameter d = 0.01.
Fig. 15
0.4
I/Is
s/ss
0.3
0.2
0.1
0.0
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Xel
(a)
c=0
c = 0.01
0.12
c = 0.1
c = 0.2
i */is*
494
0.08
0.04
0.00
0.0
(b)
0.2
0.4
0.6
0.8
1.0
Xel
1.0
h0
[V]
0.8
0.6
0.105
0.125
0.300
0.500
0.600
0.4
0.2
0.0
0.2
0.4
0.6
0.8
j0
[A cm2]
Fig. 16
495
8 Fuel Cells
0.1 mg cm2
0.6 mg cm2
1.0 mg cm2
0.8
0.6
Xel = 0.5
0.4
Xel = 0.125
Xel = 0.3
0.2
Xel = 0.105
0.0
0.0
0.2
0.4
0.6
0.8
1.0
j0
[A cm2]
Fig. 17
cm2
j0 (91)
S
1.0
Ufc
[V]
496
497
498
8 Fuel Cells
jp (, t)
(, t)
= J (, t) Cdl ()
t
(92)
Here J (, t) is the rate of current generation which is either given by the RHS
of Eq. (45) or (46), and Cdl is the doublelayer capacitance per unit volume of the
distributed electrode. In other words, it is
the capacity per unit surface area of the
contact between the metal and electrolyte
Sdl l
,
dl
(93)
1
Z() =
p
i
dl
bC
(95)
(96)
499
500
8 Fuel Cells
1
d()
1
p = c p exp
(104)
= (jp )
(98)
2b
d
d(jp )
i
()
Due to the exponential acceleration of
=
exp
+ ICdl ()
the reaction rate with electrode potential,
d
b
b
(99) the penetration depth decreases, whereas
the characteristic frequency as well as the
The substitution of () from Eq. (65) effective electrode conductivity increase exgives
ponentially.
The spatial variation of the stationary
2
2
l
d ()
=
electrode
potential, (), is small, as long
d 2
Lp 2 (1 )
as Lp > 2l. Therefore, the validity criterion
of Eq. (103) is
l
2
(1 )
cos
2Lp (1 )
lp
c
2b < 1 < 2b ln
(105)
l
+I
()
p (1 )
If 1 exceeds this interval, the spatial
(100) dependence of () cannot be neglected.
In this case the solution of Eq. (100) leads
The denitions used here are
to a hypergeometric equation, which, due
lp
to its complexity, is not shown here.
1
(101)
Lp (1 ) = exp
At = 0 the real differential resistance
c
2b
of the layer is recovered (cf. Eq. 51),
1
p (1 ) = c p exp
(102)
b
1
l
(106)
coth
diff =
A simple analytical solution can be
p
Lp
obtained, when the renormalized reactionSeveral approximations for the calculapenetration depth Lp is large compared
tion
of the effect of O2 -transport limitato l. Indeed, for this case the variation
of () with can be neglected, that tions on the complex impedance response
is, the electrolyte phase in the electrode of the electrode have been studied [144].
is equipotential. Therefore, the dynamic In one of these approximations, only oxyresponse in the Tafel regime is given by a gen diffusion limitations were considered,
similar expression as in the linear regime, whereas proton transport limitations were
neglected (equipotential electrolyte phase),
1
1/2
cf. Sect. 8.2.3.4.4. The stationary solution
Z() =
1+I
p
p
for this case provides the denition of a
gas-diffusion-limited reaction-penetration
1/2
l
(103) depth,
coth
1+I
Lp
p
I
eff
0
exp
=
(107)
however, with renormalized penetration
l
i
2b
depth Lp , (cf. Eq.(101)) characteristic frequency p , (cf. Eq.(102)) and conductance as discussed in Sect. 8.2.3.4.4 (cf. Eq. 79).
2b
Lp
bi
(108)
and for I b
b
l
0
exp
diff
coth
2b
eff
I i
(109)
Note that the replacement b I
transforms Eq. (101) into Eq. (107) and
Eq. (108) into Eq. (109), emphasizing similar impacts of proton and oxygen transport
limitations.
The most accurate solution, which does
not use any of the aforementioned approximations, can be obtained only numerically;
it utilizes the rst-order expressions for the
spatial variations of the stationary parts of
local electrode potential, (), as well as
local oxygen partial pressure, p(). These
solutions will not be reproduced here. Instead this section will be concluded with a
discussion of limiting frequency and thickness regimes, exemplifying the possible
use of impedance spectroscopy for catalyst
layer diagnosis.
All the approximate analytical expressions that were considered in Ref. 144 give
the best results when transport limitations can be considered as insignicant
for the stationary electrode performance,
corresponding to large renormalized characteristic lengths as compared to the
electrode thickness, eff , Lp 2l. These
conditions are most likely to be realized
at small current densities.
Limiting
Regimes
Highfrequency limit, c . In this regime, the
complex impedance has a linear branch
at an angle of 45 in the ColeCole
representation, which is prescribed by
1
2
Z() =
1/2
(1 I). (110)
Cdl
2
8.2.3.7.2
Here, the impedance response is independent of the working point, and the frequency dependence is determined solely
by the material parameters of the composite. For l Lp , the linear branch appears
only at frequencies > (/Cdl ). Doublelayer charging and proton transport dominate the overall electrode response in this
regime, whereas Faradaic processes are
insignicant due to the high frequencies.
An equivalent representation of this system is an RC-transmission line [130]. Since
no fractality or branching of the network
is assumed, the response resembles that
of a Warburg impedance with a characteristic proportionality Z 1/2 , where
the proportionality constant in Eq. (110) is
a characteristic function of the composition. This proportionality constant could
be determined in a plot of |Z| versus
1/2 . Therefore, if were known from
independent ex situ conductivity measurements or from theoretical considerations
based on percolation theory, Cdl could
be inferred from this kind of plot (or
vice versa). Important conclusions may
be drawn about the distribution of electrolyte and its effect on performance.
A frequency-dependent penetration depth
of the sinusoidal signal can be dened
by
l
(111)
() =
Cdl
Poor proton conductivity will give a small
value of this penetration depth. This will
501
502
8 Fuel Cells
Z() = diff 1 A
Lp
p
l
I
B
(112)
p
Lp
with
3/2 + coth( )
3
+ 2
,
4
sinh(2 )
2
1
1+
, = l/Lp .
B( ) =
2
sinh(2 )
A( ) =
1
2
(113)
This asymptotic expansion forms a part
of a semicircle in the complex plane of
a ColeCole plot. This kind of response
indicates charge-transfer limitations, bypassed through the double-layer charging.
The center of the semicircle is located on
the real axis at
B2
M = diff 1
(114)
2A
and its radius is given by
R = diff
B2
2A
(115)
For Lp / l 1 the limitations due to proton transport are practically absent and the
impedance response forms a perfect semicircle in the ColeCole representation with
M = R = diff /2, due to the parallel processes of charge-transfer and double-layer
charging, which are distributed homogeneously within the layer. The frequency
in the turning point of the semicircle is
p , the approximation being the better
the larger the ratio Lp / l. For Lp / l
= 1 the
error of this estimate is about 10%.
The linear response to a small perturbation in the Tafel regime is characterized by
the renormalized values Lp , p , and p .
Close to the lower boundary of the potential
range, specied in Eq. (105), the penetration depth Lp will be large compared to
the electrode thickness. With the increase
of the electrode potential, however, Lp decreases exponentially, while p increases
exponentially. Once Lp approaches l, the
differential electrode resistance takes the
form
1
1
diff
exp
(116)
c p
2b
that is, the Tafel slope doubles if proton
transport limitations become as important
as kinetic losses.
Thus, the graphical representation of
impedance spectra provides denite tools
for the determination of catalyst layer parameters via the identication of the linear
branch in the high-frequency limit and the
branches of semicircular character.
In
Fig. 18
composition-dependent
impedance spectra are depicted. Catalyst
layer properties have been parameterized
as functions of the composition as
described in Sect. 8.2.3.6.1. For the doublelayer capacitance a parameterization in the
form.
s
P (Xel )P (Xm )
Cdl = Cdl
(117)
Im(Z )
[ cm2]
Xel = 0.15
Xel = 0.30
Xel = 0.45
Xel = 0.60
Re(Z )
[ cm2]
Fig. 18
s is the double-layer
is used. Here, Cdl
capacitance of a system that is completely
lled with electrolyte. The accessibility to
the innite macroporous network is not
a prerequisite for a piece of the contact
area in order to contribute to the doublelayer charging. The impedance spectra
were calculated for j0 = 0.01 A cm2 using the approximations that only include
proton transport limitations, based on
Eq. (100). Since oxygen transport limitations were supposed to be negligible
(I j0 ) in this gure, it was consistent to
put d = 1 in Eq. (89). The reference parameters used in this plot are b = 30 mV,
i / l = 2 104 A cm3 , = 20 S cm2 ,
and l = 10 m.
The high-frequency domain of the
impedance spectra clearly exhibits the linear branch of the response, prescribed by
Eq. (110). It is the more pronounced, the
more severe proton transport limitations
are, that is, the smaller Xel is. Calculated
parameters corresponding to different
compositions are listed in Table 3. At small
Xel the renormalized penetration depth
Lp is smaller than the electrode thick-
Xel
0.15
0.30
0.45
0.60
l
[V]
0
[V]
lp /l
Lp /l
c s1
0.225
0.211
0.198
0.189
0.248
0.214
0.199
0.190
18.3
42.0
58.1
72.0
0.60
1.77
3.04
4.33
902
568
365
277
503
504
8 Fuel Cells
505
8 Fuel Cells
Quasi-three-dimensional Simulation of a
Cell
Goals and Methods
The cross section of a fuel cell is shown in
Fig. 19. The MEA, which includes backing
layers, catalyst layers, and the membrane,
is clamped between two metallic or
graphite plates with machined channels
for feed gases supply. The gases then
8.2.4.1
Catalyst layer
Membrane
Catalyst layer
Backing layer
x
Fig. 19
3D element
Backing layer
2D element
506
507
508
8 Fuel Cells
Element
8.2.4.2.1 Basic Assumptions In order to
demonstrate the capabilities of the Q3D
model we consider in this chapter a basic
Ni = cDi i + ci v D Arcy
where
v D Arcy =
kp
p
(118)
(119)
509
510
8 Fuel Cells
=
B
Dij
Di
j
(121)
1
Di = DiB + (DiK DiB ) 1 tanh
2
x x
(122)
Here x is the location of the backing/catalyst layer interface, is the characteristic width of the interface smearing,
which must be much smaller than the catalyst layer thickness for this interpolation to
(123)
where Q is the rate of electrochemical reaction, n is the number of electrons participating in the reaction and Si
is stoichiometry coefcient. Substituting
Eqs. (118) and (119) into Eq. (123) we obtain an equation for i , which contains the
pressure gradient.
The equation for pressure p follows from
the following arguments. Neglecting the
mass of protons we can consider that there
are no volume sources and sinks of mass
on either side of the cell. The continuity
equation for the total mass ux thus reads
Si
Q
nF
(v D Arcy ) = 0
(124)
law p =
RT we obtain
M
M
pp = 0
(125)
RT
Note that the gas composition varies
over space; M, therefore, depends on the
coordinates.
Since the membrane is assumed to be
impermeable to gases, Eqs. (118), (123),
and (125) can be written for the anode and
the cathode compartments separately with
the boundary condition of zero ux at the
catalyst layer/membrane interfaces.
(el el ) = Q
(126)
(a a ) = Qa
(127)
(c c ) = Qc
(128)
511
512
8 Fuel Cells
composition of the catalyst). The parameters, which are used in the simulations,
are indicated in the corresponding tables.
All that the models of the class described above need is the conversion
function that denes the relationship
between the local values of the reacting gas concentration, overpotential, and
the produced current. The subject of
the microscopic theory is to justify this
relationship.
Sufciently far from equilibrium, which
is the only interesting regime for current
generation, the Tafel law (129), (130) with
effective parameters is used in the simulations. Measurements performed on model
electrocatalysts suggest different values
for these parameters depending on the
particular catalysts used or just their different structure. The mesoscopic effects
in electrocatalysis are the hot topic nowadays [127, 187189] and they are probable
candidates for one of the reasons of this
variance. The main cause of this variance
may be associated with the complicated
interplay between adsorption and reaction
stages, and a modeler interested in fuel
Tafel slopes of the methanol oxidation on technical Pt/Ru anodes. The anodes are prepared
by wet spraying. Methanol concentration: 1 M, potential range used for the calculation of the b
factors: 0.250.4 V (RHE) (A. Havranek and K. Wippermann, unpublished results)
Tab. 4
Catalyst
Fractions of
catalyst
and Naon
[wt%]
Pt/Ruloading
[mga cm2 ]
Press forcea
[kNa cm2 ]
Operating
temperature
[ C]
Tafel
slope
[mV]
70/30
1.3
0.5
50
60
70/30
1.0
0.5
50
88
93/7
0.5
0.5
60
56
force applied to the samples during the hot-pressing process, hot-pressing temperature:
130 C.
Fig. 20
Finite-difference Approximations
Equations (125128) are particular cases
of the prototype equation
u
(u)
+
(u)
=q
x
x
y
y
(131)
where (u) stands for the corresponding
kinetic coefcient. In order to cast this
equation into the nite-difference form,
we introduce an orthogonal grid in the
domain shown in Fig. 20 and approximate
Eq. (131) by
m+1
ui+1,j um+1
i,j
m
0.5(m
i+1,j + i,j )
xi
m+1
ui,j um+1
i1,j
m
m
0.5(i,j + i1,j )
xi1
8.2.4.2.6
+
=
0.5(xi1 + xi )
m+1
ui,j +1 um+1
i,j
m
m
0.5(i,j +1 + i,j )
yj
m+1
ui,j um+1
i,j 1
m
m
0.5(i,j + i,j 1 )
yj 1
0.5(yj 1 + yj )
qijm
(132)
513
514
8 Fuel Cells
(133)
WEm
exp(Em ) 1
m m+1
(um+1
i+1,j exp(E )ui,j ),
Em =
hi WEm
E
(134)
NNm+1 NSm+1
m
= qi,j
, (135)
0.5(yj 1 + yj )
i,j + 1
NN
NW
i 1,j
i,j i,j
i,j
NS
i,j 1
NE
c vc
z
z
hc
4F
(137)
where pc (z) is the pressure.
Detailed analysis shows [192] that in the
practically interesting range of inlet velocities, being much smaller than the speed
of sound, the ow in the channel is incompressible, that is, c = const. Solving
Eq. (136) for velocity we immediately obtain
1
2Mw MO2
vc (z) = vc0 +
c hc
4F
z
j (z) dz
(138)
j (z) dz
(139)
0
where cO2 is the oxygen molar concentration. Under given vc (z) and j (z) integration of Equation (140) is trivial. Analogous
algebra determines the methanol or hydrogen concentration in the anode channel.
For the case, for example, of methanol we
have
j (z) 1
(cM va )
=
z
ha 6F
The velocity in the cathode channel, therefore, increases. Physically, oxygen molecules are replaced with water
molecules in the ow; the total mass of
(140)
(141)
Iteration Procedure
Having described the approaches to the
internal and channel problems we now
can describe the iteration procedure:
8.2.4.4
515
516
8 Fuel Cells
Step 0:
In all elements put cO2 (x, y, z) =
0 , c (x, y, z) = c0 , (x, y) = 0,
cO2
M
c
M
a (x, y) = a0 , el (x, y) = ka0 (where
the constant k < 1 is determined from
lc Qc = la Qa ; Qc and Qa are given by
Eqs. 129 and 130).
Solve Eqs. (123), (125), and (126) to
Eq. (128) for each element (Fig. 19)
Each element sends boundary values
(along vertical dashed lines in Fig. 19)
to the neighbors and receives their
boundary values. The values along the
left boundary of the leftmost element
and along the right boundary of the
rightmost element are determined by
physical boundary conditions (zero ux
along y axis).
Calculate the local current density in
every element, using the relation
j=
1
l
dx
o
dy Qc
(142)
1500
2500
3500
50
150
250
350
Qa, Qc, jp
Current collector
Methanol
ja
Qa
jp
Qc
jc
Oxygen
Current collector
Fig. 22
517
8 Fuel Cells
ja, jc
500
1500
2500
3500
50
150
250
350
Qa,Qc, jp
Collector
Collector
Methanol
ja
Qa
jp
Collector
Qc
Collector
518
jc
Oxygen
Fig. 23
Tab. 5
Anode
side
Cathode
side
Cell temperature T [ C]
110
110
Inlet gas pressure [atm]
1.5
2
100
100
Inlet gas velocity [cm s1 ]
Oxygen molar fraction at
0.2
the inlet
Water vapor molar fraction
0.79
0.01
at the inlet
Nitrogen molar fraction at
0.79
the inlet
Methanol molar fraction at
0.2
the inlet
CO2 molar fraction at the
0.01
inlet
Exchange current density i 1 102 1 105
[A cm3 ]
Transfer coefcient
0.5
2.0
Reaction order
1.0
0.5
0.0315
0.0315
Membrane phase
conductivity [S cm1 ]
Carbon phase conductivity
40
40
[S cm1 ]
Catalyst layer thickness
10
10
[m]
Backing layer thickness
100
100
[m]
Membrane thickness [m]
20
20
Feed channel length [cm]
80
80
Feed channel height [cm]
0.1
0.1
vc0
cO2
S j (z)
=
z
nF hc
(144)
519
8 Fuel Cells
jel
285
305
325
345
Qa,Qc
150
300
450
600
750
Inlet
10
12
CH3OH
O2
Outlet
0.000
0.002
0.004
CH3OH
0.006
0.008
0.010
Catalyst
layer
Qa
Membrane
jel
Catalyst
layer
0.014
Qc
0.016
O2
0.018
0.020
0.022
0.024
Inlet
Outlet
Fig. 24
12
11
100
10
9
200
8
7
6
Oxygen concentration
along the channel
(106 mol cm3 ) for indicated
values of the mean current
density in the cell (mA cm2 )
Fig. 25
13
O2 concentration
[106 mole cm3]
520
300
0
20
40
60
Distance
[cm]
80
100
jlim =
0
nF hc vc0 cO2
S
L
(148)
Cell voltage
[V]
0.8
Voltagecurrent curves
of the cell with a conventional
meander channel for the three
indicated values of the channel
length (cm).
0.7
0.6
400
Fig. 26
0.5
100
200
200
100
300
400
Current density
[mA cm2]
500
521
522
8 Fuel Cells
L
ln 1 1
(149)
side [198]
j
j
= ln
ln k ln 1
b
j
jD
(152)
where j = 2el b/ l is characteristic current density, k = (li /j )(cO2 /cO2ref ) and
jD =
j (z)
1
1 z/L
= ln 1
1
j
0
cO2
(151)
Recently, the proles of j (z) have been
directly measured [197]. The results [197]
obtained in conditions of a well humidied
membrane are in excellent agreement with
this simple theory [199].
(153)
is the limiting current density due to oxygen transport through the backing layer.
Here Db is oxygen diffusion coefcient in
the backing layer of a thickness lb . Relation (152) describes the limiting behavior
of cell polarization curve: as j approaches
jD , tends to innity.
Equation (152) contains oxygen concentration in the channel cO2 = cO2 (z), which
obeys Eq. (144). Solution to Eqs. (152) and
(144) at the assumption z independent is
given by [199]
a
z
j
=
a
exp
(154)
0
jD
where
z = z/L (along-the-channel coordinate given in the units of the portion of the
total length of the channel) a is a function
of polarization voltage:
a=
nF Db cO2
Slb
8.2.4.5.4
1
1+
1 /b
Ke
(155)
and
0
=
jD
0
nF Db cO2
,
Slb
lb hvc0
Db L
K=
Slb li
,
nF Db cO2ref
(156)
are constants.
Equation (154) gives the local polarization curve of individual segment at a
distance z from the channel inlet. The
set of curves for different
z is shown in
Fig. 27.
Fig. 27
Polarization voltage
0.9
10
0.7 0.5
0
0.1
0.3
0.2
0.1
0.5
0.6
z
= ln K
1
b
(157)
523
524
8 Fuel Cells
Conditions and parameters for DMFC
with differential pressure ow eld
Tab. 6
Anode side
Cathode side
1.5
2 (1.95 red)
0.034
0.034
19.35 (red)
19.35 (red)
8.2.4.5.6
Inlet
Outlet
Fig. 28
10
Qa, Qc
6
2
4
6
Inlet Inlet
10
8
12
10
12
200
400
600
800 1000
P
1.95 1.96 1.97 1.98 1.99 2.00
Anode
Outlet Outlet Outlet
H2 O
O2
Qa Mem Qc CH3OH
CH3OH
0
O2
4
H2O
0
Inlet
A
Inlet
Inlet
Inlet
C
Cathode
D
Outlet Outlet Outlet
Fig. 29
525
8 Fuel Cells
Conditions and parameters for
hydrogenoxygen fuel cell
Tab. 7
Anode
side
Cell temperature T [ C]
85
Inlet gas pressure [atm]
1.5
Inlet gas velocity [cm s1 ]
50
Oxygen molar fraction at
the inlet
Water vapor molar
0.40
fraction at the inlet
Nitrogen molar fraction
at the inlet
Hydrogen molar fraction
0.45
at the inlet
CO2 molar fraction at the
0.15
inlet
Exchange current density 1.4 105
i [A cm3 ]
Transfer coefcient
0.5
Reaction order
1.0
Red channel length [cm]
25.8
Cathode
side
85
2
50
0.2
0.01
0.79
1 105
2.0
1.0
25.8
,
(158)
la =
2ia
cH2
which is similar to the well-known
reaction-penetration depth in the context
of GEDs [11, 106]. Figure 34(a) shows the
exponential t of the simulation data. The
slope of the tting curves is in excellent
agreement with that given by Eq. (158).
Thus the concept of reaction-penetration
depth is valid for PEFC anodes with long
feed channels at practically interesting current densities.
8.2.5
120
D
C
B
A
100
Water flux
[mA cm2]
526
80
60
20
40
100
200
300
Current density
[mA cm2]
400
Outlet
Inlet
Cell element
Cell element
Inlet
Outlet
Sketch of the simulated fuel cell. On both sides of the cell the ow eld is
formed by three parallel meander channels (shown on top). The cell cross section is
shown below. The computational domain consists of 15 identical elements (one
element is depicted by dashed lines).
Fig. 31
527
8 Fuel Cells
Inlet
Outlet
ha
Qa
H2
19
1800
1.2
15
1200
0.8
11
600
0.4
H2
Qa
X
[cm]
0.010
ha
0.011
Membrane
0.013
X
[cm]
528
hc
12
250
0.014
567
10
200
Qc
8
563
150
O2
559
100
O2
Inlet
Qc
hc
Outlet
Maps of hydrogen concentration (106 mol cm3 ), reaction rates (A cm3 ), electrode
potentials (mV), and oxygen concentration (106 mol cm3 ) in the anode and cathode catalyst
layers. Shown are the maps for the elements at the inlet of the White channel and at the outlet
of the Grey channel (Fig. 31). Mean current density in the cell is 200 mA cm2 .
Fig. 32
Fig. 33
103
Reaction rate
[A cm3]
200
102
101
1
0.01
200
Membrane
100
100
0.011
0.012
0.013
0.014
Distance
[cm]
529
8 Fuel Cells
Despite all the complications, catalyst layer theory provides a few clear
instructions for improvements. Composition and thickness can be readily
adjusted, following recent optimization
studies. The theory was used, for instance, to study upgraded design with
nonuniform compositions [123, 124]. Performance enhancements measured for
correspondingly fabricated functionally
graded CCLs are in line with predictions
of the theory.
The theory-based estimate, that Pt utilization in conventional three-phase composite CCLs reaches only 1020% of
its full potential, gives rise to optimistic
speculations on how much better catalyst layers could be made. Exploring
these reserves requires innovative fabrication procedures. Such efforts could
be facilitated decisively, however, by upgrades in the theoretical understanding
of catalyst layer functioning. Aspects
of partial liquid water saturation, water
formation/transformation and two-phase
transport in porous structures should be
included in the macro-homogeneous electrode theory. At microscopic or atomistic
resolution, on the other hand, improved
theoretical understanding of relevant reaction mechanisms in nanoparticle systems
could reveal routes toward nanostructured
electrocatalysts with highest activities.
(a)
200
101
Membrane
Anode potential
[mV]
100
102
103
0.0100
0.0105
0.0110
101
Outlet
Inlet
102
103
(b)
Membrane
Fig. 34
Anode potential
[mV]
530
200
0.0100
0.0105
Distance
[cm]
0.0110
cw
Concentration
of water, 55 mol l1
D(wx )
jw , jwa , jwc
jp
jpc
jm
K(wx )
L
n(wx )
P g, P l, P c
r
r1
U
w
wc
ws
x
Water diffusion
coefcient (dimensionless)
Water ow (membrane,
anode, cathode), A cm2
Proton
current density, A cm2
Critical proton
current density, A cm2
Characteristic parameter,
jm jpc , A cm2
Hydraulic
permeability, nm2
Membrane thickness, m
Electroosmotic coefcient
Gas, liquid,
capillary pressure, atm
Pore radius, nm
First moment of the pore
radius distribution, nm
Membrane
resistance, cm
Membrane
potential drop, V
Water content (volume
fraction) in the membrane
Percolation water content
Saturation water content
Coordinate perpendicular
to membrane surface, m
Greek symbols
Viscosity, mJ s cm2
Tortuousity factor
Membrane
el (wz )
conductivity, S cm1
el
Local
membrane potential, V
psd
Pore size distribution
REV
Representative
elementary volume
531
532
8 Fuel Cells
Xel , Xm , Xc
b
b
cO2
Cdl
g
I
i
jp
j0
l
lp
Lp
PO2
p
c
diff
S0
Apparent Tafel
slope, b = RT /c nt F , V
Tafel parameter in linear
regime, b = RT /nt F , V
Oxygen
concentration at diffusion
layer interface, mol l1
Double-layer
capacitance, Fcm2
Ratio of
O2 diffusion versus proton
conductivity, g = I / b
Characteristic parameter
of O2 diffusion, A cm2
Exchange current
density parameter, A cm2
Local proton
current density, A cm2
Proton
current density at interface
to membrane, A cm2
Catalyst
layer thickness, m
Reaction penetration depth
(proton transport), m
Renormalized reaction
penetration depth, m
Oxygen
partial pressure, atm
Nondimensional
pressures,p = PO2 /PO2
Linear
electrode polarization
resistance, cm2
Differential
resistivity, cm2
Effective
catalyst|electrolyte
contact area, cm1
Coordinate
perpendicular to
catalyst layer surface, m
Z()
Nondimensional
coordinate, = x/ l
Volume portions
(electrolyte, metal,
percolation threshold)
Complex
impedance, cm2
Greek symbols
Dimensionless parameter,
= (i I /j02 ) exp(/b)
eff
Thickness of active part
of the catalyst layer, m
Proton
conductance per surface
area, = el / l, S cm2
p
Mixed
conductance parameter,
S cm2 (def. in Eq. (52))
p
Renormalized
conductance, S cm2
el
Local
electrolyte potential, V
Potential
c
of carbon/catalyst phase, V
Parameter of residual
ec (d )
reactivity (diffusivity)
Frequency
of sinusoidal signal, s1
p
Characteristic frequency
(renormalized: p ), s1
Symbols Q3D Model, Section 4
c
h
i
Molar
concentration, mol cm3
Channel height, cm
Exchange (volumetric)
current density, A cm3
j
j
jlim
kp
K1
l
la
L
L1
M
M
n
Ni
p
Q
S
v
x, y, z
Local
current density, A cm2
Mean current
density in a cell, A cm2
Limiting
current density due to feed
gas exhaustion, A cm2
Hydraulic
permeability, cm2
Dimensionless
constant in (147)
Catalyst
layer thickness, cm
Characteristic
thickness (158) (anode
catalyst layer of PEFC), cm
Channel length, cm
Characteristic length of
feed gas consumption, cm
Molecular weight, g mol1
Mass of the molecule, g
Number
of electrons, participating
in the reaction
Molar ux of ith
component, mol cm2 s1
Pressure, g cm1 s2
Rate of charged
particles production
in electrochemical
reaction, cm3 s1
Stoichiometric coefcient
Celocity of the ow
in the channel, cm s1
Coordinates, cm
Greek symbols
ButlerVolmer
transfer coefcient
Correction
factor for porosity (binary
diffusion coefcient).
Reaction order
Kinetic coefcient
Conductivity, S cm1
Correction factor
for tortuosity (Knudsen
diffusion coefcient).
533
534
8 Fuel Cells
Membrane Models
Our section on membrane modeling focuses primarily on functional requirements and performance modeling of polymer electrolyte membranes (PEM) for fuel
cells. There have not been decisive new
insights for this type of device modeling since the writing of this chapter.
The very basics of the pertinent understanding of the structure of Naon and
its water sorption properties are revisited
in Ref. 211. The relevant phenomenology
of various transport phenomena and reactions in proton conductors for fuel cell
applications has been reviewed recently in
Ref. 212. An encyclopedic survey of existing models of transport phenomena in
membranes can be found in Ref. 213.
An avalanche-like rush of new research
activities has evolved around new routes
in membrane synthesis and strategies in
theoretical and computational modeling
that could facilitate a deliberate design of
highly functionalized fuel cell membranes.
Comprehensive reviews on membrane
synthesis highlight principles and fabrication of new membranes particularly
those applicable for DMFC [214], those
that are feasible for operation at elevated
temperatures (up to 200 C under ambient
pressures) [215], and those that are based
on various modications of Naon-type
membranes [216].
This chapter focuses mainly on macrohomogeneous approaches that could rationalize the effects of catalyst layer thickness
and composition on catalyst utilization and
performance. It is relevant for the operation of random three-phase composite
layers, which are still the state of the art
in fuel cell technology. It is widely recognized that activity and utilization of the
catalyst will have to be understood at ner
structural resolution in order to be able to
reduce catalyst loadings to the levels that
are acceptable for industrial applications.
Determining the true extent to which catalyst is utilized in current layers remains
a challenge for experimental diagnostics.
Theoretical estimates suggest that the effectiveness of catalyst utilization could be
improved more than 10-fold.
A number of fundamental studies explore catalyst activity at an atomistic scale.
DFT calculations can reveal how the rates
of surface processes depend on the local electronic structure of surface atoms
[233238]. Monte Carlo simulations and
mean-eld approaches can incorporate
this information in order to rationalize the
effects of nanoparticle sizes and surface
structure on the overall rates of current
conversion [233, 239]. Thereby the nontrivial dependence of reactivity on particle
size could be explained.
535
536
8 Fuel Cells
Refs. 242, 243). Another issue is monitoring the structure and its changes with
degradation. We are not aware of any substantial results on this front. The relevant
structures and their changes during operation are not only poorly predictable
but they are also not sufciently well
characterized.
At high anodic overpotentials, methanol
oxidation reaction exhibits strongly nonTafel behavior owing to nite and
potential-independent rate of methanol
adsorption on catalyst surface [244]. The
equations of Section 8.2.3 can be modied to take into account the non-Tafel
kinetics of methanol oxidation. The results
reveal an interesting regime of the anode
catalyst layer operation featuring a variable thickness of the current-generating
domain [245]. The experimental verication of this effect, however, has not yet
been performed.
The Q2D model of a DMFC [248] revealed a new effect: localization of current
close to the inlet of the oxygen channel
at small airow rate. The effect is due
to methanol crossover through the membrane: a system with crossover current is
generated only close to the inlet of the
oxygen channel, where oxygen is in excess (galvanic domain). In the rest of the
cell, all available oxygen is consumed in a
direct burning with permeated methanol.
Further theoretical [255, 256] and experimental [257260] studies of this regime
have shown that the oxygen-depleted part
of the cell works as an electrolytic cell:
it consumes current produced in the galvanic domain to split methanol on the
cathode side and to produce hydrogen on
the anode side.
The Q2D approach was utilized to construct the models of PEFC [261263] and
DMFC [264, 265] stacks. Assembling the
cells into a stack leads to another overhead:
one needs to transport the current through
the bipolar plates (BPs) separating individual cells. The respective voltage loss can be
calculated on the basis of the model [262].
Generally, in-plane current in BPs can ow
in any direction; the plates are thus true
2D objects [265]. A promising approach to
stack modeling would be a combination of
Q2D models for the description of individual cells with fully 2D models of current
and heat transport in the BPs.
Progress in computer technology facilitates 3D modeling of cells [266, 267]. The
3D results help understand the role of
3D effects in cell function and to explore
and optimize complex geometries of ow
elds. Reduced models [268, 269], which
utilize the idea of splitting the fully 3D or
2D problem into a number of problems
of lower dimensionality seem to be very
promising (this idea is similar to that described in Section 8.2.4). These models are
much faster, than their fully 3D counterparts, which is vitally important for cell
optimization and massive comparison of
model predictions with experiment.
Final Remarks
21. T. Erdey-Gruz,
Transport Phenomena in
Aqueous Solutions, Adam Hilger, London,
1974.
22. N. Agmon, Chem. Phys. Lett. 1995, 244, 456.
23. O. Savadogo, J. New Mater. Electrochem.
Syst. 1998, 1, 47.
24. M. Doyle, S. K. Choi, G. Proulx, J. Electrochem. Soc. 2000, 147, 34.
537
538
8 Fuel Cells
25. J. Ding, C. Chuy, S. Holdcroft, Chem.
Mater. 2001, 13, 2231.
26. W. Grot, Chem.-Ing.-Tech. 1972, 44, 167.
27. G. A. Eisman, in Diaphragms, Separators,
and Ion-Exchange Membranes (Eds.: J. W.
Van Zee, R. E. White, K. Kinoshita et al.),
Proc. Vol. 86-13, The Electrochemical
Society, Pennington, 1986, p. 52.
28. A. Eisenberg, Macromolecules 1990, 23,
4098.
29. S. J. Paddison, J. New Mater. Electrochem.
Syst. 2001, 4, 197.
30. M. Eikerling, S. J. Paddison, T. A. Zawodzinski Jr., J. New Mater. Electrochem.
Syst. 2002, 5, 1.
31. J. J. Sumner, S. E. Creager, J. J. Ma et al., J.
Electrochem. Soc. 1998, 145, 107.
32. H. K. Christensen, J. Fang, J. N. Israelachvili, Phys. Rev. B 1989, 39, 11750.
33. K.-D. Kreuer, S. J. Paddison, E. Spohr et al.,
Chem. Rev. 2004, 104, 4637.
34. S. Walbran, A. A. Kornyshev, J. Chem. Phys.
2001, 114, 10039.
35. E. Spohr, Mol. Sim. 2004, 30, 107.
36. P. Commer, C. Hartnig, D. Seeliger et al.,
Mol. Sim. 2004, 30, 755.
37. E. Spohr, P. Commer, A. A. Kornyshev, J.
Phys. Chem. B 2002, 106, 10560.
38. P. Commer, A. G. Cherstvy, E. Spohr et al.,
Fuel Cells 2002, 2, 127.
39. M. Eikerling, A. A. Kornyshev, U. Stimming, J. Phys. Chem. B 1997, 101, 10807.
40. M. Eikerling, A. A. Kornyshev, A. M.
Kuznetsov et al., J. Phys. Chem. B 2001,
105, 3646.
41. T. A. Zawodzinski Jr., C. Derouin, S.
Radzinski et al., J. Electrochem.Soc. 1993,
140, 1041.
42. K. D. Kreuer, T. Dippel, W. Meyer, J. Meier,
Mater. Res. Soc. Symp. Proc. 1993, 293, 273.
43. M. Eikerling, A. A. Kornyshev, J. Electroanal. Chem. 2001, 502, 1.
44. M. Eikerling, A. A. Kornyshev, J. Electroanal. Chem. 2002, 528, 196.
45. H. H. Voss, D. P. Wilkinson, P. G. Pickup
et al., Electrochim. Acta 1995, 40, 321.
46. M. Cappadonia, J. W. Erning, S. M. Saberi
Niaki et al., Solid State Ionics 1995, 77, 65.
47. Z. Luz, S. Meiboom, J. Am. Chem. Soc. 1964,
86, 4768.
48. L. I. Krishtalik, Charge Transfer Reactions
in Electrochemical and Chemical Processes,
Plenum, New York, 1986.
539
540
8 Fuel Cells
117. J. Xie, K. L. Moore, T. A. Zawodzinski et al.,
J. Electrochem. Soc. 2004, 151, A1841.
118. M. B. Isichenko, Rev. Mod. Phys. 1992, 64,
961.
119. A. S. Ioselevich, A. A. Kornyshev, Phys. Rev.
E 2002, 65, 2131.
120. A. Ioselevich, A. A. Kornyshev, W. Lehnert,
Solid State Ionics 1999, 124, 221.
121. R. P. Simpraga, B. E. Conway, Electrochim.
Acta 1998, 43, 3045.
122. Q. Wang, M. Eikerling, D. Song et al., J.
Electrochem. Soc. 2004, 151, A950.
123. Z. Xie, T. Navessin, Z. Shi et al., J.
Electrochem. Soc. 2005, 152, A1171.
124. D. Song, Q. Wang, S. Liu et al., J. Power
Sources 2004, 126, 104.
125. T. Navessin, S. Holdcroft, Q. Wang et al., J.
Electroanal. Chem. 2004, 567, 111.
126. T. Navessin, M. Eikerling, Q. Wang et al., J.
Electrochem. Soc. 2005, 152, A796.
127. F. Maillard, M. Eikerling, O. V. Cherstiouk
et al., Faraday Discuss 2004, 125, 357.
128. A. S. Ioselevich, A. A. Kornyshev, Fuel Cells
2001, 1, 40.
129. Q. Wang, M. Eikerling, D. Song et al., J.
Electroanal. Chem. 2004, 573, 61.
130. J. J. Macdonald (Ed.), Impedance Spectroscopy, Wiley, New York, 1987.
131. Z. B. Stoynov, B. M. Grafov, B. S. SavovaStoynov et al., Electrochemical Impedance,
Nauka, Moscow, 1991.
132. R. De Levie, in Advances in Electrochemistry and Electrochemical Engineering (Ed.:
P. Delahay), Wiley, New York, 1967, p. 329,
Vol. 6.
133. F. Gutmann, J. Appl. Electrochem. 1965, 112,
94.
134. H. Keiser, K. Beccu, M. A. Gutjahr, Electrochim. Acta 1976, 21, 539.
135. M. Keddam, C. Rakotomavo, H. Takenouti,
Electrochim. Acta 1984, 14, 437.
136. N. A. Hampson, S. A. G. R. Karunathilaka,
R. Leek, J. Appl. Electrochem. 1980, 10, 3.
137. J. I. Zhang, F. Wenger, M. Jerome et al.,
C.R. Acad. Sci. C 1985, 995.
138. P. Lay, P. F. Lawrence, N. J. M. Wilkins
et al., J. Appl. Electrochem. 1985, 15, 755.
139. Proceedings of 4th International Symposium
on Electrochemical Impedance Spectroscopy,
Rio de Janeiro, Brazil, 1998.
140. Proceedings of.5th International Symposium
on Electrochemical Impedance Spectroscopy,
Marileva, Italy, 2001.
541
542
8 Fuel Cells
205. T. V. Nguyen, J. Electrochem. Soc. 1996, 143,
L103.
206. D. L. Wood, J. S. Yi, T. V. Nguyen III,
Electrochim. Acta 1998, 43, 3795.
207. A. Vishnyakov, A. V. Neimark, J. Phys.
Chem. B 2000, 104, 4471.
208. A. Vishnyakov, A. V. Neimark, J. Phys.
Chem. B 2001, 105, 7830.
209. J. Ennari, M. Elomaa, F. Sundholm, Polymer 1999, 40, 5035.
210. K. A. Friedrich, K. P. Geyzers, A. J. Dickinson et al., J. Electroanal. Chem. 2002,
524, 261.
211. K. A. Mauritz and R. B. Moore, Chem Rev.
2004, 104, 4535.
212. K. D. Kreuer, S. J. Paddison, E. Spohr et al.,
Chem Rev. yr2004, 104, 4637.
213. A. Z. Weber and J. Newman, Chem Rev.
2004, 104, 4679.
214. N. W. Deluca and Y. A. Elabd, J. Polym. Sci.
B, Polym. Phys. 2006, 44, 2201.
215. Q. F. Li, R. H. He, J. O. Jensen et al., Chem.
Mater. 2003, 15, 4896.
216. B. Smitha, S. Sridhar, and A. A. Khan, J.
Membrane Sci. 2005, 259, 10.
217. A. Vishnyakov and A. V. Neimark, J. Phys.
Chem. 2001, 105, 9586.
218. S. S. Jang, V. Molinero, T. Ca
gin et al., J.
Phys. Chem. B 2004, 108, 3149.
219. D. Seeliger, C. Hartnig, and E. Spohr, Electrochim. Acta 2005 50, 4234.
220. J. L. Matthews, E. K. Lada, L. M. Weiland
et al. Smart Mat. Struct. 2006, 15, 187.
221. (a) P. G. Khalatur, S. K. Talitskikh, and
A. R.Khokhlov, Macromol. Theory Sim.
2002, 11, 566586. (b) D. Y. Galperin and
A. R. Khokhlov, Macromol. Theory Simul.
2006, 15, 137.
222. J. T. Wescott, Y. Qi, L. Subramanian et al.,
J. Chem. Phys. 2006, 124, 134702.
223. (a) S. J. Paddison and J. A. Elliott, J. Phys.
Chem. A 2005, 109, 7583; S. J. Paddison,
Ann. Rev. Mater. Res. 2003, 33, 289.
224. M. K. Petersen, F. Wang, N. P. Blake et al.,
J. Phys. Chem. B 2005, 109, 37273730.
225. A. Roudgar, S. P. Narasimachary, and
M. Eikerling, J. Phys. Chem. B in press.
226. M. Schuster, T. Rager, A. Noda et al., J. Fuel
Cells 2005, 5, 355.
227. J. Teissie, M. Prats, P. Soucaille et al., Proc.
Natl. Acad. Sci. 1985, 82, 3217.
228. V. B. P. Leite, A. Cavalli, and O. N. Oliveira,
Jr., Phys. Rev. E 1998, 57, 6835.
543
544
8 Fuel Cells
8.3
Introduction
545
546
8 Fuel Cells
547
548
8 Fuel Cells
Anode
vent
Cathode
vent
2H++ 2e
Anode : H2
Cathode: O2 + 4e +
4H+
2H2O
H+
0.7 V
Anode
feed, H2
Cathode
feed, O2
Gas diffusion
backing
Fig. 1
Fig. 2
12
11
12
11
12
549
550
8 Fuel Cells
type of cell: PEM fuel cell = polymer electrolyte membrane fuel cell,
SPE fuel cell = solid polymer electrolyte
fuel cell, or PEFC = polymer electrolyte
fuel cell. In this chapter, we have preferred the latter acronym. The membrane
material commonly employed in most recent PEFC technology developments is
poly(PFSA). Naon made by DuPont
is the best-known material of this type.
Similar materials are produced by Asahi
Chemical and Asahi Glass (Japan), as
well as by 3M (United States). W.L.
Gore developed and provided a composite membrane for PEFCs, with expanded
PTFE as backbone and poly(PFSA) as
the active material. These structures are
aimed at providing improved lateral dimensional stability, particularly to avoid
excessive lateral dimensional changes during membrane drying/rehydration cycles.
The combined chemical and physical
properties of poly(PFSA) membranes give
them signicant advantages as membrane
electrolyte for PEFCs. These membranes
exhibit superior long-term chemical stability under both oxidative and reductive
environments owing to their Teon-like
molecular backbone. The specic protonic
conductivity achieved in well-humidied
membranes of this type is of the order of 0.1 S cm1 at typical cell operation
temperatures, which translates, for example, to an areal resistance of 0.05 cm2
for a membrane of 50-m thickness.
These thin membranes are, at the same
time, effective gas separators: the permeability of both oxygen and hydrogen
through the membrane is of the order of
1011 1010 mol cm1 s1 atm1 , which
translates to a gas crossover equivalent current density of 110 mA cm2 through a
membrane in an operating fuel cell. This
leakage current is at or below 1% of
551
552
8 Fuel Cells
(2)
O2 + 4e + 4H+ = 2H2 O
(3)
and
(5)
with ow elds on both sides. Metal bipolar plates have some important advantages
over carbon-based materials: better mechanical properties and higher electronic
conductivity. Consequently, stacks based
on metal hardware can achieve tighter
packaging by using thinner bipolar plates
and, therefore, achieve higher stack power
densities. However, corrosion susceptibility of metal hardware in PEFCs and
DMFCs, remains a challenge. Conicting
reports have described, on the one hand,
successful implementation of metal hardware for PEFC and DMFC stacks and,
on the other hand, testing showing insufcient immunity to corrosion. It seems
that the choice of a specic metal alloy or
surface-coated metal for PEFC hardware
may be critical in achieving long-term,
trouble-free performance. Reference 14
summarizes information on performance
and durability of graphite and carbon composite bipolar plate materials, in addition
to metal hardware including primarily
stainless steel and surface-coated titanium
and aluminum.
An MEA/bipolar plate pair is the building block repeat unit of the fuel-cell stack.
The length of the stack, that is, the number of MEA/bipolar plate pairs stacked in
series is determined by the source voltage required by the relevant load, with this
number reaching, for example, 300 cells in
transportation applications which require
a 200-V power source. The surface area of
each cell in the stack is then determined by
the maximum power demand, for example, in such a 300 cell, 200-V stack, a cell
active area of 250 cm2 (1/4 ft2 ) generates
50 kW of total power at current density
of 1 A cm2 , achievable at a PEFC single
cell voltage of 0.66 V. The PEFC has been
developed to date for a wide variety of applications, ranging from 50 to 200 kW stacks
553
554
8 Fuel Cells
Examining a large-scale, framed
MEA for a 75-kW PEFC stack (courtesy
of UTC Fuel Cells).
Fig. 3
for primary power sources in transportation all the way down to 1-W power sources
for portable power in consumer electronics
applications. These different applications
obviously result in a wide variety of stack
sizes and architectures.
An idea about stack sizes associated
with potential transportation applications
Fig. 4
Fig. 6
555
556
8 Fuel Cells
(6)
(7)
0 is calculated as
where Vcell
0
= Go /nF
Vcell
(8)
(9)
where Vcath (J ) and Van (J ) describe electrode potentials at current density J of the
cathode and the anode respectively versus a reversible hydrogen electrode (RHE),
cath (J ) and an (J ) describe the overall voltage loss at current density Jcell
at the cathode and anode respectively
(calculated versus the thermodynamically
expected electrode potential versus RHE at
Jcell = 0), and Rcell is the high-frequency
component of the resistance of the cell, determined by the sum of (1) the membrane
resistance (which, under some circumstances, exhibits dependence on cell current J ), (2) the cell plate/current collector
bulk resistance, and (3) the various contact resistances originating from stacking
and pressing together all cell components.
Jcell
aan J0,an
Jcell
acath J0,cath
J Rohm
bcath log10
Jcell
1
Jlim,cath
(10)
To highlight the major features of the
PEFC polarization characteristics and their
electrochemical origins, Eq. (10) is written
assuming simple behavior of the fuel-cell
anode and cathode behaving as ultrathin
electrodes of total specic catalyst surface area (square centimeter of Pt per
square centimeter of cross-sectional area)
set by mass loading and dispersion of
the catalyst. Additional losses in a PEFC
electrode due to transport through the
thickness dimension of the catalyst layer
(see Sect. 8.3.7.2C) are not considered in
Eq. (10) for the sake of simplicity. The
equation can be seen to assume a constant Tafel slope, bcath , for the interfacial
oxygen electrode process through the current range covered and a voltage loss at
the hydrogen anode calculated from the
relevant linear relationship between an
and Jcell , which applies because the cell
current densities (per square centimeter
of Pt) are of the same order or smaller
than the exchange current density of the
anode process. The PEFC-limiting current is set by the air cathode limiting
current, Jlim,cath which is generally determined, in turn, by the rate of (gas phase)
diffusion of oxygen from air through a
porous cathode-backing layer of about
300-m thickness, with the pores lled
RT
= V0
F
557
8 Fuel Cells
Hydrogen-air cell
1.2
1
0.8
0.6
0.4
0.2
0
0
0.25
0.5
(a)
0.75
1.25
1.5
1.75
12.5
25
37.5
50
62.5
75
87.5
100
1000
900
800
Areal power density
[mW cm2]
558
700
600
500
400
300
200
100
0
0
(b)
0.15
0.3
0.45
0.6
0.75
0.9
1.05
1.2
Cell voltage
[V]
Fig. 7
559
8 Fuel Cells
Polymer electrolyte fuel cells
1.0
Nafion 117
- HF Res
Memb C
- HF Res
Nafion 105
- HF Res
Dev. Dow
- HF Res
HF resistance [ cm2]
or cell voltage [V]
0.8
0.6
H2/O2 = 3/5 atm
Tcell = 80 C
0.12 0.16 mg Pt /cm2
0.4
0.2
0.5
1.0
2.0
1.5
2.5
3.0
Current density
[A cm2]
(a)
1.0
Nafion 117
- HF Res
Memb C
- HF Res
Nafion 105
- HF Res
Dev. Dow
- HF Res
0.8
HF resistance [ cm2]
or cell voltage [V]
560
0.6
H2/air = 3/5 atm
Tcell = 80 C
0.12 0.16 mg Pt /cm2
0.4
0.2
0
(b)
0.5
1.0
1.5
2.0
2.5
Current density
[A cm2]
Fig. 8
is given, in
enthalpy exchangers. Pstack
general, as
(11)
(12)
is, at the most, 10 mV at a cell current density of 1 A cm2 . With Vcell near 0.60 V at
cell current density of 1 A cm2 a typical
PEFC peak-power point at Tcell = 80 Cand with Rcell of typically 0.1 cm2 , Vcath
would be given to good approximation
as Vcath = 0.6 V + J Rcell = 0.70 V versus
RHE. At Vcath = 0.70 V, cath is over 0.5 V
and the ORR current density per square
centimeter Pt is expected to be 45 orders
of magnitude higher than the exchange
current density (see Eq. 10), that is, Jcell =
10 mA cm2 Pt. Consequently, a cell current of 1 A cm2 of MEA cross-sectional
area will be obtained at Vcath = 0.70 V
with 100 cm2 of catalyst area per square
centimeter of MEA cross-sectional area.
This degree of surface area amplication is achievable at 0.10.2 mg Pt/cm2
of MEA, assuming a Pt particle size of
2 nm and a high catalyst surface utilization. In summary, a low ohmic loss and
good technology of catalyst dispersion and
utilization enables to reach in a PEFC
an areal power density of 0.6 W cm2
together with a conversion efciency of
4050%, in spite of the intrinsically low
activity of the air electrode (J0,cath =
107 106 A cm2 Pt). The cathode voltage loss, cath , of over 0.5 V, required to
reach the maximum power point in the
PEFC, corresponds, however, to a loss of
over 40 percentage points (0.55 /1.23) in
voltage efciency, that is, in conversion
efciency. Another smaller loss of eight
percentage points (0.10/1.23) is incurred
by the PEFCs ohmic resistance, associated with the membrane electrolyte itself
and with inner cell contact resistances.
In the case of the DMFC, a typical design
point of Vcell = 0.45 V, in a state-of-theart cell, would allow a current density
of 0.2 A cm2 at cell temperature of
80 C. Under these conditions, the anode
potential would typically be 0.4 V versus
561
562
8 Fuel Cells
Naon
(H2CCH2)n
PSSA
CF3
SO3H
CF
CF2
(CF2CF2)n(CF2CF)m
CF2CF2SO3H
CF
CF2
(CF2CF2)n(CF2CF)m
SO3H
Chemical structure
Chemical structures of Ionomeric Membranes Employed in (or considered for) PEFCs and DMFCs
Membrane
Tab. 1
N
H
CF3
C
CF3
R2
H
N
CF2
R1
(CH2CH2)p(CF2CF2)n(CF2CF)m
SO2
SO3H
CF2
1x
co
CF3
C
CF3
HO3S
SO2
x
SO3H
564
8 Fuel Cells
565
566
8 Fuel Cells
continued work on lower cost alternatives. It is unclear how much the cost
of poly(PFSA) membranes could drop if
the demand for them increases, but it has
been argued that a 2 order-of-magnitude
increase in the market for such membranes their market is now limited to
chlor-alkali industry application and to
development, demonstration, and testing
of PEFC power system prototypes will
bring about an order-of-magnitude decrease in their price. For transportation,
considered the largest potential market for
PEFCs in the longer run, the target power
system cost is under 50 $/kW and that
would require the lowering of membrane
cost to well under 50 $/m2 . A drawback of
a different nature is the limited temperature range over which these membranes
can be effectively used. The upper temperature limit is usually considered around
100 C because of the low membrane water content above that temperature under
low internal over-pressure. Above 100 C
the water content in a poly[PFSA] membrane required to maintain high proton
conductivity, can be achieved only under
signicant water vapor pressure elevated
total pressure requiring, in turn, maintained in the fuel-cell stack. The low limit
of the useful temperature window is set
by the strong fall of the protonic conductivity in a hydrated PFSA membrane
under the freezing point of water. The inner PEFC temperature window, enabling
high performance, is consequently considered 0100 C (preferably 1080 C) and,
in the design of a PEFC stack subsystem, provision has to be made to raise the
cell temperature above the freezing point
of water when the ambient temperature
is lower than that, and proper shedding
of heat generated by the cell process is
required to limit the rise of the cell temperature to a maximum near 80 C.
For optimal use in fuel cells, in addition to the primary requirement of high
protonic conductivity, the membranes employed in PEFCs must be chemically and
mechanically stable in the fuel-cell environment. The chemical stability requirement has been an important issue since
the initial development of membranes for
PEFCs. Degradation of PSSA membranes
used initially for PEFCs in the early 1960s
is severe even under conditions signicantly milder than those encountered in
an operating fuel cell [20, 21]. Consequently, the introduction of the poly(PFSA)
membranes, where CH bonds are fully
replaced by CF bonds in the polymer
structure, enabled an important step-up in
membrane stability in the operating PEFC.
Subsequent attempts to develop alternative
membranes, not based on peruorinated
polymers, have included, for example,
ionomers with poly(aromatic) backbones
and uorinated, sulfonic acidterminated
sidechains. The latter membranes, as a
rule, have lower protonic conductivity at
a given degree of sulfonation, or given
equivalent weight (EW), because of smaller
tendency for the nanoscale, aqueous ionic
phase separation, which occurs most readily in ionomers with highly hydrophobic, peruorocarbon backbone. Otherwise,
such hydrocarbon membranes show,
as a rule, lower chemical stability versus
that exhibited by peruorinated membranes.
In spite of the documented, relatively
high chemical stability of poly(PFSA)
membranes in the fuel-cell environment,
recent extensive work looking into the origins of performance loss observed in PEFCs has revealed important mechanisms
of degradation that apply to peruorinated
membranes (while being further amplied in nonperuorinated membranes).
An important mechanism of membrane
to the PA-doped PBI facilitate dissociation, but the water content in question
is low and does not require signicant
steam pressure, thereby enabling PEFC
fuel cell operation at such higher temperatures at practically ambient pressure.
Blending PA with PBI is assisted by
acidbase interactions between the acid
and the base functions of the PBI building
block (see Table 1). Nevertheless, the composite, PA-doped PBI and the poly(PFSA)
membranes belong to different categories
of polymer electrolytes. The latter uses only
distilled water, whereas the former relies
on some liquid electrolyte blended/bonded
with a polymeric matrix. If the composite
membrane blend is stable under some relevant operation conditions, the dopant
electrolyte can provide some important advantages, rendered by a proton-conducting
fuel cell operating at over 120 C under
low pressure. These include much better
impurity tolerance than in operation at
or under 80 C and more effective heat
rejection. Such electrolytes would therefore be considered when signicant system
simplication is to be achieved in operation on some processed carbonaceous fuel.
In such cases the much better impurity tolerance of the fuel cell allows to eliminate
fuel processing stages required to strongly
lower the carbon monoxide level in the
anode feed. This advantage at the system
level could be sufcient to compensate
for some drawbacks of the PA-doped PBI
electrolyte, including the well-documented
lowering of the electrocatalytic activity of
a Pt catalyst when in contact with PA
electrolyte. This lower catalytic activity results in hydrogen/air cells with PA-PBI
membrabe electrolyte exhibiting, at some
given current density, cell voltages that
are typically 200300 mV lower versus hydrogen/air PEFCs employing poly(PFSA)
membranes.
567
568
8 Fuel Cells
Tab. 2
Rehydration temperature
27 C<T<94 C
27 C
Rehydration temperature
65 C
80 C
21 (34%)
21 (42%)
25 (56%)
12 (20%)
11 (22%)
16 (36%)
14 (23%)
15 (30%)
23 (52%)
16 (26%)
15 (30%)
25 (56%)
a Membrane
569
8 Fuel Cells
35
Membrane C
Dow
Nafion 117
30
25
20
15
10
5
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
Water activity
(a)
16
Uptake (# waters/sulfonate)
570
14
Dow
Membrane C
Nafion 117
12
10
8
6
4
0
(b)
T = 30 C
2
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
Water activity
Fig. 9
membrane in contact with liquid water, has been at the center of studies
of the poly(PFSA)/water system. Another
clear thermodynamic differentiation is that
membrane water at the lowest hydration
levels is associated with enthalpy of water uptake larger than the heat of water
condensation, whereas for water added
into a membrane of high water content,
the enthalpy of vapor uptake is signicantly lower than the enthalpy of water
condensation. This claries further that,
at low water levels in the membrane, water molecules are strongly bonded to ions,
whereas with further increase in water
level, the water becomes more and more
bulklike and the process of further hydration is driven by entropy rise rather than
by enthalpy drop. The consequence for
PEFC technology is that a strong difference in water (and hence proton) mobility
is to be expected with membrane water content, with a demarcation line
in water properties between the rst ve
to six waters per sulfonate group and
higher water content. A recent molecular dynamics work by Paddison et al. [26]
shows that another factor strongly affecting protonic mobility, namely the tendency
of the sulfonic acid groups to dissociate
according to SO3 H + H2 O SO3
(H3 O+ ), also dramatically increases as
water levels reach 6H2 O/SO3 H. Consequently, the general picture of the aqueous
subphase in the ionomeric membrane,
the higher hydration levels targeted for
high-performance PEFCs, is that of concentrated, well-dissociated sulfonic acid
(SO3 (H3 O+ )) in aqueous medium of
properties close to that of bulk water.
The water/membrane system behavior
at temperatures below the freezing point
of water is, for obvious reasons, very
signicant technologically. Some recent
reports on self-start of PEFC stacks from
571
572
8 Fuel Cells
(13)
0.14
sulfonic acid group to have a clearly dissociated H3 O+ ion around the sulfonic acid
head group. At lower water contents than
that, the proton is still strongly associated
with the counter charge on the sulfonic
acid group, as manifested by the very
short calculated protonsulfonate group
distances. According to these calculations,
protonic charge shielding is ineffective at
less than ve water molecules per sulfonic acid group, whereas around ve to
six waters per sulfonic acid group, water starts to provide sufcient shielding as
manifested by the appearance of the wellresolved H3 O+ group signicantly further
removed from the sulfonate group. This is
shown in Fig. 11. One cannot fail to notice that the number of water molecules
per sulfonate group required, according to
these rst principles calculations, for effective acid dissociation (hence signicant
protonic conductivity), corresponds to the
transition from membrane water associated with uptake enthalpies larger than
that of water condensation to membrane
water of uptake energy resembling that of
free water condensation [24].
Dow
Membrane C
Nafion 117
0.12
Conductivity
[S cm1]
0.10
0.08
0.06
0.04
0.02
0
10
15
20
25
Water content in # waters per sulfonate
30
573
8 Fuel Cells
Fig. 11 First principle calculations of
lowest energy structures for the triic
acid/water system. Left:
CF3 SO3 + 3H2 O; Right:
CF3 SO3 H + 6H2 O. For the case of
= 3 (left), dissociation just starts to
occur, however the H3 O+ moiety
remains very close to the SO
3 group:
r(O-H) = 1.56 Angstrom. In contrast, in
the case of = 6, separation of the
hydrated proton from the conjugated
anion occurs: r(O-H) jumps up from
1.74 Angstrom to 3.68 Angstrom with
the sixth water molecule added [29].
1e-4
T = 300 K
DH2O
D [cm2 sec1]
574
1e-5
Nafion117
(phase separated)
1e-6
Polyaromatic membrane
(homogeneously sulfonated)
D
DH2O
1e-7
2
10
12
14
16
[H2O]/[SO3H]
Fig. 12 Intradiffusion coefcients of water and of protons in concentrated
aqueous acid solution, homogeneously sulfonated polyaromatic ionomer, and
phase-separated poly(PFSA) ionomer [27].
10
20
T = 300 K
Hydrochloric acid
A = D /DH2O
[H2O] / [H3O+]
Nafion117
(phase separated)
Polyaromatic membrane
(homogeneously sulfonated)
12
16
20
Molarity
Fig. 13
Nw,drag = I ()/F
(14)
575
8 Fuel Cells
H2O production
H+(H2 O)n
Water diffusion
H+ transport
Anode
2H+ + 2e
H2(g)
Nafion
membrane
O2 (humidified)
Electroosmotic drag
H2 (humidified)
576
Cathode
2H2O
O2 + 4H+ + 4e
(15)
(16)
577
8 Fuel Cells
30
Diffusion coefficient ( 106)
[cm2 sec1]
578
DOW
Nafion 105
Nafion 117
25
20
15
10
Diffusion coefficients at 80 C
for three different membranes
5
0
25
10
15
20
Water content, N (H2O)/N (SO3H)
30
35
1H
Fig. 15
cathode side, with liquid water. The signicant difference in water content and
in water transport properties between a
membrane in contact with water vapor and
a membrane in contact with liquid water
is discussed in Sect. 8.3.4.2.1.
Reference 32 reports an activation energy for water transport in Naon on
the order of 0.17 eV (3.9 kcal mol1 )
for water contents ranging from =
8 14. At lower water content examined,
= 3.5, an activation energy of 0.22 eV
(5.1 kcal mol1 ) is reported over the temperature range 2090 C. The increase in
activation energy for water motion with
decreasing water content [32] accords well
with the notion that water mobility is increasingly hindered at low water contents
by shrunken pores and stronger electrostatic interactions. On the basis of these
numbers for the activation energy, an increase by about factor 2 between 20 and
80 C is expected in the water diffusion
coefcient in poly(PFSA) membranes of
higher water contents, and by about factor
4 at the lowest water content.
5.5
5
4.5
4
3.5
3
2.5
2
1.5
20
40
60
80
Temperature
[C]
100
120
140
579
580
8 Fuel Cells
(20)
(21)
581
8 Fuel Cells
16
StoichA = 2 @ 1 A cm2
14
12
10
8
6
Current density
[A cm2]
0.1
0.2
0.5
0.8
2
0
0.2
H+
0.6
0.4
Thickness fraction
0.8
Cathode
1.0
Anode
Fig. 17
1.0
tmem = 175 m
tmem = 140 m
tmem = 100 m
tmem = 50 m
0.8
0.6
[ cm2]
H2O flux /H+ flux
582
0.4
0.2
0
0.3
0.2
0.1
0
1
Current density A cm2
Fig. 18 Computed Rcell (upper panel) and the ratio water ux/proton ux
through the cell (lower panel) under the conditions described for
Fig. 17 [36].
583
584
8 Fuel Cells
conductivity and of uniform catalyst distribution, and have such layer very well
bonded to the ionomer membrane surface. Three types of tools are available, in
principle, to answer this overall challenge:
optimized metal alloy catalyst surface properties, maximized catalyst dispersion and
effective catalyst layer composition and
structure that enables high catalyst surface
utilization.
Optimization of catalyst composition
and preparation/fabrication for the PEFC
or DMFC cathode has been pursued during the last decade with one signicant
success being the replacement of Pt by
some Pt alloy catalysts, including PtNi and
PtCr [41]. By the mid-2000s, PtCo alloys
eventually emerged as the cathode catalysts
of choice for PEFCs. The latter binary cathode catalysts have been reported to provide
intrinsic catalytic activity in oxygen reduction in a PEFC cathode that is higher by a
factor of 35 versus that of a non-alloyed
Pt catalyst of similar Pt mass loading. This
is an advantage of high value, particularly
in PEFC applications of highest sensitivity
to the price of platinum. Catalyst composition optimization at the anode side of
the PEFC and DMFC has been primarily centered on the optimization of the
composition and the rened preparation
of PtRu alloy catalysts. The latter exhibit
signicant tolerance to 10100 ppm CO
in hydrogen-fueled PEFCs and high activity in electrooxidation of methanol in the
DMFC anode. These electrocatalysis R&D
issues are discussed in further detail in
Sects 8.3.7.1 and 8.3.8.1, respectively.
Perhaps the most important advancements in electrocatalysis that contributed
mostly to the development of lowtemperature fuel-cell technology belong to
the domain of practical electrocatalysis
and include maximized catalyst dispersion
and effective packaging of the required
585
586
8 Fuel Cells
surface area into thin catalyst layers of optimized composition structure, and of good
bonding to the substrate surface. Highly
dispersed Pt or Pt alloy catalysts can be
prepared using a variety of techniques. The
most common is the formation the Pt colloidal particles from solution onto partly
graphitized carbon powder (e.g., Vulcan
XC-72), with reduction by various possible
reducing agents taking place during and
following the formation of such colloidal
particles. The nal product is typically
metal particles of the order of 2 nm in
diameter, supported onto carbon power
of particle size around 10 nm. For catalysts consisting of spherical particles of
diameter 2r, application of catalyst mass
m per square centimeter of membrane
area results in overall catalyst surface area,
acat , per square centimeter of membrane
area of
(22)
acat = 3m/r
where is the mass density of the catalyst.
For example, at m = 0.4 mg Pt/cm2 of
membrane area, r = 1 nm, and (for Pt)
of 20 g cm3 , acat is calculated as 600 cm2
Pt/cm2 of membrane area. Such highly
dispersed catalysts thus allow, in principle,
to reach a factor 600 amplication of
the geometric surface area of the electrode
with mass loading of only 0.4 mg Pt/cm2 of
electrode area. Such a high level of catalyst
dispersion has been one of the most
important keys to the high power output
and conversion efciencies achieved in
PEFCs.
Equation (22) shows that the total surface area of the catalyst at given level
of dispersion (given r) increases linearly
with overall catalyst mass loading, m. The
same is not true, however, for the effective
or utilizable catalyst surface area. It has
been the experience of developers of PEFC
and DMFC MEAs that, catalyst utilization,
0.09
022
Water
content
(H2O/SO3)
6.5
0
Polymer electrolyte membrane (340 m)
Fig. 19 MRI of water distribution in a PEFC at zero current and under two current
densities. The cathode is on the right side of the MEA imaged [37].
shows that, as result of limited rate of reactant diffusion into the catalyst layer in the
case of the 10% hydrogen feed, the concentration of the reactant and, consequently,
the current production, are both calculated
to be very low at fractional widths of the
catalyst layer larger than 0.5, away from the
catalyst layer/gas interface. On the other
hand, if the effective protonic conductivity
of the catalyst layer is an order of magnitude lower than that of the neat ionomer, a
signicant drop in overpotential across the
catalyst layer could make the half of the
catalyst layer away from the membrane
signicantly less effective in current production. This is seen to be the case for the
combination of low protonic conductivity
and high reactant concentration. It can be
readily understood from Fig. 20 why, at
given transport characteristics typical for
composite, Pt/C//ionomer catalyst layers,
utilization of the catalyst could strongly
drop when the thickness of the layer exceeds 510 m, particularly in the case
of low reactant concentration. In contrast,
thicker layers could still enhance electrode
587
8 Fuel Cells
0.12
0.1
xh = 0.1
Va(y)
[V]
0.08
0.06
s = 0.01 S cm1
0.04
s = 0.1 S cm1
xh = 1
0.02
0
H2 concentration ratio
0.02
xh = 1
1
0.8
0.6
1
s = 0.1 S cm1 s = 0.01 S cm
0.4
xh = 0.1
0.2
0
xh = 0.1
3.5
3
588
s = 0.1 S cm1
2.5
2
s = 0.01 S cm1
1.5
xh = 1
1
0.5
0
0.2
0.6
0.4
Fractional length y
0.8
Fig. 21
589
590
8 Fuel Cells
Tab. 3
the catalyst layer subsequently being deposited onto one surface of the pretreated
carbon cloth/paper. Compared to catalysts
based on Pt black, an important advantage of the carbon-supported Pt catalyst
was the higher degree of Pt dispersion.
Pt particle sizes of 2 nm in diameter are
easily obtained in the carbon-supported
form by various protocols described in
the patent literature. This is compared
with a typical particle size of about 10 nm
obtainable in Pt blacks. TEM images of
carbon (Vulcan X-72)-supported Pt catalyst of 20% Pt by weight are given in
Fig. 21. Trying to take advantage of the
higher dispersion of the Pt/C catalyst,
Raistrick started his investigation by hotpressing gas-diffusion electrodes of this
type directly to an ionomeric membrane
and realized that the poor cell performance obtained consistently was caused
by lack of protonic access to the majority
of catalyst sites not in intimate contact
with the membrane. Effective protonic
access could not be met within the 50-mthick Pt/C//PTFE catalyst layer because Pt
sites distributed in a catalyst layer bonded
by PTFE are protonically disconnected
from the ionomeric membrane. To overcome this critical limitation, Raistrick
impregnated the Pt/C//PTFE catalyst layer
591
592
8 Fuel Cells
the ionomer in the catalyst layer. A thermoplastic form of the ionomer was proposed
as an effective solution for the lm robustness problem. Naon (and other peruorinated ionomers) can be converted to a
thermoplastic form by the ion-exchange inclusion of large, hydrophobic counterions such as tetrabutylammonium (TBA+ )
[11]. The material is, consequently, moderately melt-processable. In an improved
version of the original CCM fabrication
process [11], thin-lm catalyst layers are
cast from an ink that consists of supported platinum catalyst and solubilized
ionomer in the TBA+ form. A general
scheme describing the preparation of a
catalyzed membrane employing thin-lm
catalyst layers is given in Fig. 22.
As seen in Fig. 22, MEAs based on such
thin-lm catalyst layers can be constructed
using a decal process, in which the ink
is cast onto Teon blanks for transfer to
the membrane by hot-pressing. A second
approach is to cast the same type of ink
(TBA+ form of the ionomer) directly onto
the membrane [12]. The latter process has
an advantage over the decal process in
(c) Dry
Cell voltage
[V]
0.8
0.6
0.4
0.2
10
15
20
25
Fig. 23
593
594
8 Fuel Cells
set at the time without full consideration of the overall system optimization
for maximum efciency and minimum
size and weight. As of today (a decade
later), a similar performance at similar,
or even lower, catalyst loadings has been
reported with both gaseous reactants at
ambient pressure, as discussed in the following. Comparison of catalyst utilization
obtained with the different PEFC catalyzation techniques is given in Fig. 23, with
all numbers applying to pressurized operation of the cells. The advantage in catalyst
utilization with the thin-layer approach
can be clearly seen, the utilization increasing with this approach particularly at the
higher cell currents (lower cell voltage) owing to the minimized transport limitations
in the thin catalyst layer.
Fig. 24
10 000
(a)
150 000
(b)
595
8 Fuel Cells
Pt-coated nanowhiskers embedded into
the ionomeric membrane
Membrane
GDL
~0.5m
1.0
NSTF loading: A/C = 0.060 mg Pt/cm2 each
0.8
Cell voltage
[V]
596
0.6
0.4
0.2
0.0
0.0
0.4
0.6
0.8
J [A cm2]
1.0
1.2
sufcient to prevent damage, for example, pinholes, during prolonged operation (with such damage possibly leading
to hazardous, direct fuel/oxidant intermixing)?
3. Can the high protonic conductivity of
the membrane be maintained for thousands of hours in spite of any possible
source of trace ionic contaminants that
could replace protons in the ionomer?
Reasonably satisfactory, albeit by no
means complete, answers have been provided for the above questions. Life testing
of pressurized PEFC stacks developed for
space application at GE (later Hamilton
Standard) rst demonstrated thousands of
hours of operation without failure [54]. A
more recent, 11 000 hours test of a PEFC
eight-cell stack with neat oxygen and neat
hydrogen feeds showed a low decay rate
of 1.31.4 V hour1 under constant load
[55]. In the last two cases, the catalyst loadings employed were high: at least 4 mg
of Pt black per square centimeter on each
side of the cell. The decay rate with order
of magnitude lower catalyst loadings, as
required for cost competitiveness in most
PEFC applications, is still to be fully quantied for specic stack materials and builds
and under various operation conditions.
Oxidation of the carbon support material
used in the Pt/C catalysts, typically employed in PEFCs of lower loadings, is one
cause of faster deterioration in cells employing such lower loadings [56]. Highest
rates of cathode carbon oxidation are measured at or near the open-circuit potential
and are strongly enhanced at an elevated
cell temperature. Under open-circuit conditions, the cathode is not protected by the
lower electrode potential established under
current (providing cathodic protection)
and the Pt-catalyzed, carbon-support oxidation rate then becomes highest, with
597
598
8 Fuel Cells
possible results ranging from loss of cathode catalyst layer transport characteristics
to overall catalyst layer disintegration. Possible countermeasures are the use of
more graphitized carbon supports that are
less vulnerable to oxidation and/or optimization of the operation conditions, for
example, leaving the cell under load until
it cools down signicantly to provide maximum cathodic protection to the cathode
as long as the cell temperature is elevated.
The Pt catalyst itself is also found to
be vulnerable to some loss by corrosion,
involving oxidation of Pt metal atoms from
the catalyst particle to generate Pt ions,
that is, Pt particle dissolution. Recent
work [57, 58] has revealed a behavior
of Pt in this regard that is not unlike
that of more reactive metals: Pt can be,
within the potential range relevant for
PEFC or DMFC electrodes, in either a
passive state or in a more active state
with regard to metal dissolution rates.
A passive state of the Pt catalyst in a
PEFC, cathode, is established on steady
operation at potentials above 0.8 V versus
the hydrogen electrode, at which potential
a protective supercial surface oxide layer
forms on the Pt catalyst surface, reducing
the rate of metal dissolution to very low
levels. The active state of Pt occurs, on
the other hand, as result of the reduction
in this surface oxide layer, typically by
operation for some period of time at
a lower electrode potential, for example,
0.5 V versus hydrogen. Once a Pt particle
surface is activated in this way, fast
subsequent perturbation of the cathode
potential back to values higher than 0.8 V
results in unusually high instantaneous
rate of Pt dissolution, lasting until the
passive state is reestablished by surface
oxide formation.
Pt catalyst ripening in PEFCs electrodes,
resulting in loss of catalyst active area, was
0.4
Mass fraction (DIM)
(a)
0.3
0.2
(b)
(c)
0.1
8
12
Particle diameter
[nm]
16
20
599
600
8 Fuel Cells
during PEFC life tests according to measurements of cell impedance at 5 kHz [11].
However, the deionized water employed
in the humidication scheme in such tests
[11] had very low levels of metal ions
(e.g., Fe+2/+3 , Ca+2 , or Mg+2 ). Such multivalent ions exchange irreversibly with
protons in the poly(PFSA) membrane,
causing a drop in membrane conductivity. Deionizing the water used for PEFC
humidication is, therefore, required, and
appropriate plumbing materials should be
used to avoid generation of ionic contaminants by corrosion processes.
A critical issue of chemical instability of the ionomeric membranes most
commonly used in PEFC technology, repeatedly brought up recently, is the attack
of poly(PFSA) membranes by OH radicals, followed by further attack by hydrogen
peroxide [60]. There is good evidence for
the formation of OH radicals in the PEFC
cathode by recombination of hydrogen
crossing over the membrane with oxygen
at the cathode or vice versa with formation
of the OH radical at the anode. Although
the crossover rate of hydrogen and oxygen
through the membrane is equivalent of
only several milliamperes per square centimeter, it is apparently sufcient to cause
enough initial damage to the structure of
poly(PFSA) materials currently manufactured, and this initial attack by the OH.
radical apparently triggers further attack
of the membrane by hydrogen peroxide.
The latter is a not uncommon side product of the cathode process, particularly
when the cathode catalyst surface is partly
deactivated. Remedial efforts pursued at
this time, inclue minimizing the probability of OH radical formation in the
cell by maintenance of high membrane
hydration through the thickness dimension (hence use of thinner membranes) as
601
602
8 Fuel Cells
Combustor
BP valve
LT radiator
Recirculation
pump
Fc stack
Condenser
Reservoir
Humidifier
Humidifier
Hydrogen
Air
radiator
(a)
BP valve
Combustor
Recirculation
pump
Fc stack
Hydrogen
Air
radiator
(b)
the cell into two humidiers for the incoming air and the incoming hydrogen
streams, respectively, thereby ensuring
effective humidication across the active
area of each cell. The same function of
Desulfurizer
Shift
reactor
Selective
oxidator
Control
unit
Internal
load
or
Tail gas
burner
External
load
Power
conditioning
Reformer
or
Heat exchanger/
humidifier
Thermal
management
system
Fuel in
Fuel cell
Expander
Exhaust
Condenser
Air
compressor
Air in
Water
pump
Coolant
pump
Water tank
Heat
exchanger
Fig. 30 Scheme of a power system based on a PEFC stack operating on reformed carbonaceous fuel
(eg., natural gas) and air [63].
603
604
8 Fuel Cells
20
z
[ cm2]
15
10
10
15
20
25
30
35
z
[ cm2]
Fig. 31 Impedance spectrum measured for a bulk, smooth
Pt/Naon membrane interface in a hydrogen atmosphere at the
H2 /H+ equilibrium potential [65].
40
605
606
8 Fuel Cells
(23)
1000
800
600
400
200
0
0
200
400
600
800
1000
Fig. 32
CO + M
(M CO)
(24)
H2 + 2M
2(M H)
(25)
(M H) H+ + e + M
(26)
H2 O + (M CO)
kec
M + CO2 + 2H+ + 2e
(27)
The rst two equations represent the
competing processes of CO adsorption
and dissociative chemisorption of H2 . As
seen, it was assumed [66] that desorption rate for CO and the adsorption rate
for hydrogen could both be functions
607
608
8 Fuel Cells
CO = kfc xCO PA (1 CO h )
bfc kfc CO kec CO e
M + H2 O
M OH + H+ + e
(28)
M (CO) + M OH
CO2 + H+ + e + M + M
(29)
On a Pt catalyst particle, M and M are
both (obviously) Pt surface sites, although
CO adsorption and water activation may
take place on distinguishable types of sites
for example, CO adsorption on crystal
surface terraces and OH formation on
steps, or edges, with the process then being
completed by migration of the mobile,
adsorbed CO molecules along the terrace
to the step to recombine with the surface
OH groups. In the case of PtRu alloys,
known to enable onset of process (28) at
signicantly lower anode potentials versus
unalloyed Pt, M in Eq. (28) is likely a Ru
site, whereas M is likely an adjacent Pt
surface site.
On the basis of the scheme described
by Eq (24), Eqs. (30) and (31) represent the
steady-state catalyst surface balance of adsorption, desorption, and electrooxidation
uxes of carbon monoxide and of hydrogen. If the intermediate hydrogen step
is second order in catalyst sites, as was
assumed on the basis of the nature of
process (25) [66], then, in Eq. (31), n = 2.
Equations (30) and (31) determine CO , the
fraction of catalyst sites with adsorbed CO,
Va +Veq
bc
=0
(30)
Va +Veq
bc
(32)
Current density expressions for hydrogen and CO oxidation, jh and jCO , are
given in Eq. (32) as a function of anode
potential Va referenced to a SHE corrected
for local Nernstian potential shift associated with local values of pH2 different
than 1. Note that the same overpotential
is assumed at a given anodic potential for
both Hads and COads electrooxidation. This
means that both kfh and kfc describe the
rate of the anodic process at 0 V RHE (and
full surface coverage by the corresponding
surface species).
If all the rate constants are independent
of CO , that is, kfh and bfc are constant,
then an analytic solution for h and CO
is given by Eq. (33) with intermediate
variables dened in (34) [66]. From the
expressions for jh and jCO in Eq. (32), the
kinetic current densities of hydrogen and
of CO electrooxidation can be evaluated.
a12 s1 a2 kfh PA xH
kfh bfh a12 a2 PA xH
kfc PA xCO a1 bfh kfh + a1 s1
2 P x
a2 bfh kfh
A H
Va
(34)
(GCO )
RT
(36)
609
8 Fuel Cells
1
0.8
0 ppm CO
100
0.6
50
25
Vcell
[V]
610
0.4
250
0.2
0.2
0.4
0.6
0.8
seen in Figs 32 and 33, claries the signicant challenge of maintaining high
performance in PEFCs operating with anode feed streams derived from processed
carbonaceous fuels. When considering
losses in the fuel-processing steps upstream the PEFC and, in addition, the
signicant cathode losses in the PEFC
itself, it is readily understood that, to
maintain an attractive energy-conversion
efciency, the PEFC cannot be allowed
to operate at anode overpotentials signicantly higher than 50 mV. Therefore,
maximizing the critical current (Figs 32
and 33) at some given level of CO in the anode feed stream is essential. One possible
approach demonstrated to the mitigation
of CO poisoning at the PEFC anode catalyst
is to bleed low levels of oxygen or air into
the CO-contaminated anode feed stream
[68, 69]. It was rst demonstrated for
(39)
(40)
611
612
8 Fuel Cells
hydrogen feed stream take place in a separate catalytic lter upstream the anode
catalyst. With proper design of this active
prelter, very little oxygen would reach the
anode catalyst.
8.3.7.1.4 CO-tolerant PEFC Anode Catalysts
Although demonstrated successfully in PEFC stacks with COcontaminated feed streams, air bleed into
the anode is not considered a problemfree approach. Mixtures of hydrogen and
oxygen are of some concern even in
well-controlled systems. Consequently, the
search for a PEFC anode catalyst of better
CO tolerance has been pursued. At such
an anode catalyst, the hope is that a sufcient rate of CO electrooxidation could
be obtained closer to 0 V versus RHE,
such that a sufcient number of free
Pt sites could be maintained at steady
state at low anode overpotential. This requirement corresponds to values of kec in
Eq. (27) exceeding that for Pt catalyst by at
least 2 orders of magnitude. Alternatively,
or in addition, lower afnity of CO to an
optimized anode catalyst surface, corresponding to higher bfc in Eq. (24), would
also bring about improved tolerance. PtRu
catalysts have been most extensively researched, developed, and demonstrated in
recent years as CO-tolerant anode catalysts in PEFCs [7274]. Other alternative to
PtRu anode catalyst exhibiting signicant
CO tolerance, for example, PtMo, have also
been described [75]; however, such alternatives have not been fully demonstrated
to have superior catalytic properties and
long-term stability.
An approach described most recently for
maximizing CO tolerance at relatively low
anode catalyst loadings involves a combination of a CO-tolerant electrocatalyst,
for example, a PtRu alloy and a separate
chemical oxidation catalyst layer of zero
ionomer content, embedded within the anode structure itself in front of the (ionomer
rich) anode catalyst layer. The latter serves
the same function as that of the catalytic
prelter, originally described in Ref. 71.
Such properly positioned, bifunctional
anode catalyst structures allow to combine
oxidative lowering of Pco in the anode by
air bleed, with maximized tolerance of the
electrocatalyst to any remaining low level
of CO in the gas mixture reaching the electrocatalyst. PEFC performance level very
close to that achieved on neat hydrogen was
demonstrated with such a composite anode in operation with an anode feed stream
of 40% hydrogen containing 100 ppm CO
and is presented in Fig. 34 [42]. The gure
also highlights the fact that CO2 , a major constituent in the anode feed stream
in reformate/air PEFCs, is not playing a
role other than that of fuel diluent. This
is nontrivial because CO2 could also react
chemically with hydrogen (reverse water
shift) at the anode catalyst to form CO in
situ. Such reactivity of CO2 in the PEFC
anode was recorded with nonoptimized
anode catalysts/structures.
The PEFC Anode: Summary
The rate of hydrogen oxidation at an
impurity-free, well-humidied Pt/ionomer interface with catalyst surface area
of the order of 100 cm2 Pt/cm2 of membrane area is sufciently high to limit
anode voltage loss in PEFCs to several
millivolts at 1 A cm2 , as indeed observed
with hydrogen feed streams of high purity. However, anode catalyst poisoning by
CO is a major challenge that needs to
be addressed in each PEFC-based power
system according to the expected composition/residual contamination of the fuel
feed stream. Levels of CO up to 100 ppm
in the anode feed can be taken care of by
in situ bleeding of air, using a properly
8.3.7.1.5
Cell voltage
[V]
0.8
0.6
0.4
0.2
0.2
0.4
0.6
Current density
[A cm2]
0.8
1.0
613
614
8 Fuel Cells
in the PEFC. In a test of other possible contaminants that could result from methanol
reforming, for example, [76] strong and
irreversible effects of formic acid at a
PEFC were reported for a Pt-catalyzed
(high loading) anode, whereas methanol,
formaldehyde, and methyl formate were
found to have much smaller and reversible
effects. This is an example of a case where
the use of a more active anode electrocatalyst, such as PtRu, could signicantly
facilitate oxidation of intermediate several
signicant impurities resulting from the
coupling of natural gas (or gasoline) reformers of various types with PEFC stacks.
The use of the latter fuels requires to probe
and address, by either removal upstream
or with modied catalysts, possible detrimental effects of low levels of sulfur, H2 S,
COS, and NH3 . The effects of ppm levels of NH3 in the fuel feed stream on
PEFC performance have been recently reported in [77]. At the level of a few parts
per million, the effects of ammonia on
PEFC anode performance are major. They
can be reversed by switching to perfectly
ammonia-free hydrogen, provided the exposure to the ammonia is not longer than a
few hours. After longer exposure, the deleterious effect becomes irreversible. Unlike
CO, ammonia apparently does not deactivate the anode catalyst by adsorption at
the ppm levels concerned and the deleterious effects seem to result from exchanging
ionomer protons rst in the anode catalyst
layer(s) and subsequently also in the membrane by ammonium ions [77]. This lowers
strongly the interfacial activity in the anode rst and, with further exposure, the
membrane proton conductivity. The clear
conclusion from the information provided
in [77] is that the level of ammonia has
to be brought down to zero in the PEFC
anode feed stream and that the target can
(41)
Pt + O2 + H+ + e = Pt(O2 H)ads
(42)
615
616
8 Fuel Cells
Tab. 4
V versus
RHE
Bare RDE in
0.5 M H2 SO4
[mA cm2 ,
25 C]
Filmed RDE in
0.5 M H2 SO4
[mA cm2 ,
25 C]
0.056
0.27
0.74
1.8
0.11
0.34
0.71
1.1
0.95
0.90
0.85
0.80
Tab. 5
1.1
18
Bulk
Naon
3.1
916
4.0
0.10.4
617
8 Fuel Cells
it still does not mimic precisely the interfacial composition in PEFC cathode
structures based on Pt/C//recast ionomer
composites (see Table 5). Nevertheless,
this arrangement enabled preparation of
a Pt/ionomer interface and completion
of the cell by other electrodes in direct contact with the membrane without
use of any liquid electrolyte, employing
a well-dened geometry: microelectrode
of known diameter in direct contact with
an ionomeric membrane. This geometry
also enabled studies at elevated pressures and temperatures. Mass-transportcorrected results, derived from slow scan
(2 mV sec1 ) measurements of ORR currents in such cell, are shown in Fig. 35.
They are tted with two linear Tafel domains, as has been traditionally done in
the description of such ORR data. From
such plots, the authors concluded by extrapolation an exchange current density,
J0 , of J0 = 7.8 107 A cm2 associated
with the domain of high Tafel slope (high
polarization). The authors noted that the
latter is the highest exchange current
density reported for the ORR at room temperature at any Pt/electrolyte interface.
While the high apparent exchange current
density may look interesting, it should be
realized that direct comparison of measured
ORR rates at the Pt/ionomer interface and
at other Pt/dilute aqueous acid solutions
0.9
63 mV/decade
0.8
E vs DHE
618
118 mV/decade
0.7
0.6
0.5
4.5
3.5
ii i
log
(i i i )
2.5
(43)
(44)
Fig. 36
Oxygen
4
5
log Jo
[A cm2]
b = 120 mV/decade
b = 60 mV/decade
6
7
8
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
log Po2
[atm]
Air
5
6
log Jo
[A cm2]
b = 120 mV/decade
b = 60 mV/decade
8
9
0.4
0.2
0
log Po2
[atm]
0.2
0.4
619
620
8 Fuel Cells
Gases
Electrodes
A W
RC
Shrinkable
Teflon tubing
Nafion
tubing
Bottle
Pt flags
H2SO4
Nafon film
Liquid water
or humid air,
O2 or Ar
(a)
5mm
Counter electrode
(b)
Fig. 37 Cell employed for measurements of ORR kinetics at the
Pt microelectrode/recast ionomer interface (a) and a view of the
microelectrode/counterelectrode geometry (b) [59].
W designates the cathode (working) electrode, A, the anode,
and R and C designate the dynamic hydrogen and the
counterelectrode, respectively, for the DHE (dynamic hydrogen
electrode) reference.
20
10
0
10
Current
[nA]
(a)
20
10
0
10
20
30
100
400
7 C
50 C
600
0
200
200
20 C
300
0
Current
[nA]
100
400
60 C
600
0
200
100
200
30 C
300
0
400
70 C
600
0
200
100
200
300
0.2
200
300
0
(b)
Potential (V vs DHE)
200
Current
[nA]
20
40 C
0.4
0.6
0.8
1.0
Potential (V vs DHE)
1.2
1.4
400
600
80 C
0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
Potential (V vs DHE)
Fig. 39 Currentpotential characteristics obtained with the cell in Fig. 37 at several temperatures in
a water vaporsaturated oxygen atmosphere [59].
621
8 Fuel Cells
Kinetic current
[mA cm2]
(a)
Cathodic scan
250
200
150
100
50
0
50
Cathodic scan
(a)
101
10
101
(b)
250
200
150
100
50
0
50
Anodic scan
Anodic scan
101
100
101
(b)
0.30
0.25
0.25
Tafel SLP.
[V/dec.]
0.30
0.20
0.15
0.10
0.05
0.4
(c)
data in Fig. 36 are presented as a function of potential in Fig. 40, together with
the corresponding value of dVcath /d log J .
The latter slope was computer-derived
directly from the experimental plot of
log JORR versus Vcath [87]. Figure 37 shows
a continuously varying Tafel slope through
the cathode potential range of signicance (0.90.5 V), with a current density
of 0.1 mA cm2 at 0.9 V recorded in the
anodic scan (i.e., as the Pt electrode is
not covered by excess chemisorbed oxygen). It was consequently suggested [87]
that there is no good basis to force two
discrete Tafel slopes on this data and that
extrapolation of log JORR versusVcath data
to the thermodynamic potential for the
O2 /H2 O couple using some constant Tafel
slope can yield exchange current densities for the ORR of questionable meaning.
Kinetic current
[mA cm2]
with saturated water vapor clearly suggest that, to ensure the highest level of
cathode performance, the PEFC cathode
catalyst should be kept in contact with
some water in liquid state. This high
local water activity requirement at the
cathode catalyst/ionomer interface, was
further elucidated in a subsequent PEFC
impedance spectrum study [90] and ascribed to effective proton solvation being
prerequisite for high rates of interfacial proton transfer (process 42). This specic need
of good hydration of the cathode/ionomer
interface, comes in addition to the requirement of high water activity in the bulk of
the ionomer electrolyte to maintain high
bulk membrane proton conductivity.
Mass-transport-corrected currents at T =
30 C, that is, the kinetic current densities
evaluated from the measured T = 30 C
Tafel SLP.
[V/dec.]
622
0.6
0.8
1.0
Potential (V vs DHE)
0.20
0.15
0.10
0.05
0.4
1.2
(c)
0.6
0.8
1.0
Potential (V vs DHE)
1.2
Fig. 40 (a) ORR current, (b) kinetic (mass-transport-corrected) ORR current and (c) the
apparent slope dV/d log J as functions of cathode potential, obtained using the cell in Fig. 37
with water vaporsaturated oxygen atmosphere at 30 C [59].
623
8 Fuel Cells
Current density
624
Experimentally determined
surface oxygen coverage, ox , as a
function of Pt electrode potential during
a cathodic scan from 1.0 V and the
associated log JORR versus Vcath
dependence calculated (for ambient
conditions) from the combined effects
of enhancement of the rate of the ORR
at Pt metal surface sites with increase in
cathode overpotential and cathodic
stripping of a blocking surface oxygen
species between 1.0 and 0.6 V [88]. The
log JORR versus Vcath plot shown, as
calculated from Eq. 36, exhibits a
gradual change of dV/d log J (Tafel
slope) from around 40 mV/decade at
1.0 V to 120 mV/decade at potentials
lower than 0.65 V [88].
1.0 Fig. 41
102
101
100
101
102
103
104
105
106
107
0.8
0.6
0.4
0.2
0.0
0.4
0.5
0.6
0.7
0.8
0.9
1.0
Some ORR kinetic measurements reported most recently (2005) were performed directly on the dispersed catalyst of
an operating PEFC cathode at an elevated
temperature of 80 C and a somewhat elevated air pressure of 1.5 atm [81]. The
surface area of Pt available in such cathodes was evaluated by voltammetry, as
shown in Fig. 42. The latter report adheres to ORR rate evaluation based on
currents measured at 0.9 V versus RHE,
and these results are given in Table 7.
The iR-corrected, Vcell versus log Jcell
plot in Fig. 42 shows a uniform slope
of dVcell /d(log Jcell ), however not that the
plot is terminated near Vcath = 0.8 V (at
1 A cm2 ) and, consequently, the whole
Tab. 6 Measured ORR rates and apparent Tafel slopes at various Pt/acid electrolyte
interfaces at 25 C with gaseous atmosphere of O2 /saturated water vapor at P = 1 atm [87]
Pt/recast Naon
Pt/bulk ionomer
Pt/0.5 M H2 SO4
Pt/9.5 M TFMSA
Pt/6 M TFMSA
mA cm2 Pt at
0.90 V
mA cm2 Pt
0.85 V
b at
0.85 V mV/decade
0.1
0.2
0.4
0.4
0.2
0.2
0.4
80
63
0.6
118
61
EiR-free
[V]
0.85
70 mV/decade
20
im
[mA mg1 pt]
0.90
20
0.80
0.75
0.70
0.01
60 m2 pt/gpt
40
im = 0.11 A mg1 pt
is = 180 A cm2 pt
0.1
ieff
[Acm2] (b)
(a)
Fig. 42
Tab. 7
Catalyst
20% Pt/Ca
47% Pt/Cb
46% Pt/Cb
0.15/0.17
71
0.17
0.11
0.4/0.4
75
0.21
0.16
0.5/0.5
86
0.21
0.11
a Pt/C
b Pt/C
625
8 Fuel Cells
catalyst performance, that is, before longterm testing effects have been documented
for such ultralow Pt loadings.
The possible complete replacement of
Pt or Pt alloy catalysts employed in PEFC
cathodes by alternatives, which do not require any precious metal, is an appropriate
nal topic for this section. Some nonprecious metal ORR electrocatalysts, for
example, carbon-supported macrocyclics
of the type FeTMPP or CoTMPP [92], or
even carbon-supported iron complexes derived from iron acetate and ammonia [93],
have been examined as alternative cathode catalysts for PEFCs. However, their
specic ORR activity in the best cases is signicantly lower than that of Pt catalysts in
the acidic PFSA medium [93]. Their longterm stability also seems to be signicantly
inferior to that of Pt electrocatalysts in the
PFSA electrolyte environment [92]. As explained in Sect. 8.3.5.1, the key barrier to
compensation of low specic catalytic activity of inexpensive catalysts by a much
higher catalyst loading, is the limited mass
and/or charge transport rate through composite catalyst layers thicker than 10 m.
8.3.7.2.3 The PEFC Air Cathode: Comprehensive Analysis of Voltage Loss
The
overall voltage loss at the PEFC air cathode
is determined by combination of interfacial and transport rate limitations, where
0.90
Ei R-free
[V]
626
0.85
47% Pt/C (TKK)
Leached 45% Pt-Co/C (TKK)
0.80
Gas
diffusion
backing
Cathode
catalyst
layer
Membrane
O2
P *O 0
y
B
H+
PO(1) 1
0
A
n
o
d
e
Lonomer
Catalyst
Carbon
Fig. 44 Schematic of regions considered in comprehensive modeling
of the PEFC air electrode. From left to right: gas ow channel, GDL,
cathode catalyst layer. Oxygen in transported through the porous GDL
by gas-phase diffusion through an inert mixture of nitrogen and water
vapor. The catalyst layer is described in terms of effective transport
characteristics of gas, protons, and electrons [13].
627
628
8 Fuel Cells
follows[13]:
as follows:
I cl
Interfacial ORR losses:
=
= I Rcl
(47)
y
I Il
C/C
ar , the Pt surface area per unit geometric
(49)
=
y
p
ox ID
surface area (dimensionless); and
exp((0.9 V)/b), or exp(F (V )), the ORR Equations 4749 describe variations of
rate dependence on potential relative parameters along the y coordinate of
to 0.9 V described by either a con- the catalyst layer (y = z/ lcl ), where z
stant Tafel slope of 2.3 b (millivolts is the catalyst layer thickness coordiper decade of current density) or, al- nate, y = 0 species the catalyst layer/gas
ternatively, by an experimentally de- interface, and y = 1 species the catatermined b(V) measured at a model, lyst layer/ionomeric membrane interface
(see Fig. 44), in which Rcl (= lcl / ) is
smooth Pt/ionomer system [88].
the protonic resistance through a unit
Catalyst layer transport losses:
cross-sectional area of the catalyst layer
lcl , the thickness of the catalyst layer (cm); and I (= nF DC / l ) is a characterisD
cl
DC , the product of the effective diffusion tic diffusion-controlled current density in
coefcient of oxygen (cm2 s1 ), and the catalyst layer. The thickness of the
the equilibrium oxygen concentration catalyst layer disappears from the equa(moles cm3 atm1 ) in the catalyst tions by introducing Rcl , ar , and ID . The
layer at the interface with the backing experimental variables considered include
the overpotential , the current density I ,
(GDL),
and the oxygen concentration C , when
pox = 1 atm;
, the effective protonic conductivity pox = 1 atm at the catalyst layer/gas interface. The O2 partial pressure, pox , at
within the catalyst layer (S cm1 ).
the
catalyst layer/backing layer interface is
Backing layer transport losses:
determined,
in turn, by the cathode inlet
lb , the thickness of the backing layer
gas stream composition and stoichiomet(cm), and,
ric ow rate and by the backing layer (GDL)
(J), the effective porosity of the backing
transport characteristics.
(dimensionless).
Equations (4749) describe the nature
The rest of the parameters that dene of the losses incurred by transport within
the behavior of the cell for each set the catalyst layer. The catalyst surface area
of operating conditions include the (catalyst sites) active in the ORR process of
cathode inlet gas stream composition, conversion of oxygen ux to protonic curow rate and total pressure, the degree rent ux (Eq. 49) is distributed linearly
of humidication of the inlet gas stream, along the thickness of the catalyst layer.
Consequently, gas supply to a catalyst site
and the temperature of the cell.
removed from the backing layer/catalyst
The key equations describing the limita- layer interface is limited by the effective
tions in the catalyst layer in this model are permeability of oxygen in the catalyst layer,
whereas proton reach to a catalyst site removed from the ionomeric membrane is
limited by the effective protonic conductivity of the catalyst layer. As result of
the former, the catalyst sites experience
a distribution of oxygen concentrations
(Eq. 48, 49), whereas, as result of the
latter, different catalyst sites experience
different effective overpotentials (Eq. 47).
Calculated distributions of reactant and of
electrode potential within the catalyst layer
are described in Fig. 20 for the case of diluted hydrogen feed streams and similar
types of distributions for the case of an air
cathode were reported in Ref. 13.
(a)
(b)
1.5 A cm2 (b) [94]. The calculation was done using numerical
reconstruction by a statistical method of the layers microstructure.
The drop in oxygen concentration across the thickness dimension of
the catalyst layer is 20% in the top case (Jcell = 0.5 A cm2 ) and 80%
in the bottom case (Jcell = 1.5 A cm2 ). The concentration of oxygen at
the backing/catalyst layer interface (designated by red) corresponds in
the top and bottom case respectively to 70% and 20% of the
concentration of oxygen in air; this drop is caused by the oxygen
concentration gradient across the cathode GDL at some Jcell .
629
630
8 Fuel Cells
(50)
feed streams, demonstrate that the PEFClimiting current depends on oxygen mole
fraction, rather than on oxygen partial pressure. This is immediate indication that
transport in the air cathode is limited
at high current densities in the gas-phase
component of the porous GDL. On the
other hand, comparison of curves (a) and
(b) demonstrates the low enhancement
in current observed at low overpotentials
when changing cathode feed from 5-atm
air to 5-atm O2 . Instead of near fourfold
increase in current expected under control by interfacial kinetics (1st order in
PO2 ) the observed increase at low cathode
overpotentials (high cell voltage) is only
about twofold. This ratio could be rationalized only by the effect of limited protonic
conductivity within the catalyst layer, the
relative importance of each effects varies
1.0
Model fit
(a)
0.8
5 atm O2
VCath
(b)
(c)
0.6
(d)
5 atm air
0.4
O2 + N 2
5 atm 5.2% O2
O2 + N2
2 atm 13.5% O2
0.2
0
0.5
1.0
1.5
2.0
Current density
[A cm2]
Fig. 46 Simultaneous t to four polarization curves for a PEFC with
300-mm-thick cathode GDL and 7.5-m-thick cathode catalyst layer
applied to the membrane surface [12]. Different cathode feed stream
compositions were used and a simultaneous t demanded for all
characteristics using the same physical and transport parameters for the
GDL and the catalyst layer [13]. Fitted cathode parameters were:
icath = 90 mA/cm2 , b = 85 mV/decade, G = 1.3 103 S/cm,
DC = 1.9 108 mol/cm atm, and GDL porosity , varying between
0.25 at zero current and 0.19 at 15 A/cm2 .
631
632
8 Fuel Cells
Fig. 47 Microscopic imaging of liquid droplet emergence and growth within a hydrophobic
cathode-backing layer, followed by its removal into a hydrophilic air channel [38].
387.75 m
258.5 m
129.25 m
0.0 m
Fig. 48
633
8 Fuel Cells
V cath
0.81
0.77
0.7
0.64
0.56
0.48
0.3
0.2
Measured
10 Hz
0.1
100 Hz
1.6 Hz
1 kHz
0.1
0.2
0.3
(a)
0.4
0.4
0.5
z
[ cm2]
V cath
0.81
0.77
0.71
0.64
0.56
0.48
0.3
634
0.2
0.1
0.6
0.7
Simultaneous fit
1.6 Hz
100 Hz
10 Hz
1 kHz
(b)
0.1
0.2
0.8
0.3
0.4
0.5
z
[ cm2]
0.6
0.7
635
636
8 Fuel Cells
The direct methanol fuel cell
CH3OH + air
C
D
o
i
n
r
v
e
e
c
r
t
s
i
o
n
Electric power + benign products
CH3OH + 1.5O2 = CO2 + 2H2O +
+ h 4.8 Wh cm3 of liquid
methanol consumed
DMFC schematic
CO2
Anode
Catalyst
3H2O
Cathode
CH3OH + H2O
1.5O2
Membrane
L and GDL
GDL
(52)
+
4(PtHads ) = 4Pt + 4H + 4e
(53)
Ru + H2 O = RuOHads + H+ + e (54)
RuOHads + PtCOads
= CO2 + H+ + e + Pt + Ru
(55)
637
638
8 Fuel Cells
639
640
8 Fuel Cells
= 1/Jmem
+ 1/Jan,back.
(59)
and Jan,back
are respectively
where Jmem
limiting current densities for methanol
transport through the membrane and
through the anode backing under a concentration gradient equal to the methanol
concentration at the anode backing/ow
channel interface (Eq. 59 assumes a small
concentration drop across the catalyst layer
itself). When the cell is under current
Jcell , Jcrossover is given by the difference
between the current equivalent of the ux
of methanol across the anode backing layer
and the cell (= anode) current, that is
(60)
From the last two equations, the relationship between the rate of crossover from the
641
642
8 Fuel Cells
1 (Jcell /Jan,back
)
(61)
The last equation reveals that, to the degree that a DMFC can be operated close to
the anode-limiting current (corresponding
to the concentration of the feed stream),
the crossover rate in the cell under current
can be very low, even in cases of backing
and membrane combinations that allow
high rates of crossover at open circuit. With
fuel utilization in the DMFC given, in general, by fuel = Jcell /(Jcell + Jcrossover ), the
relationship Jcrossover (Jcell ) derived above
(Eq. (61) allows to calculate fuel utilization
at some Jcell in a DMFC of the given membrane and anode backing permeability of
methanol:
fuel = Jcell /{Jcell + Jcrossover @ Jcell =0
(1 Jcell /Jan,back
)}
(62)
This expression shows that fuel utilization levels as high as 90% can be
achieved even when the membrane is
quite leaky to fuel, that is, when
Jcrossover @ Jcell =0 is quite large. For example, with Jcrossover @ Jcell =0 as high as
50% of the cell current density targeted,
operation at Jcell /Jlim,an = 0.85 enables,
according to Eq. (62), fuel as high as 93%
with the leaky membrane.
A special case of Eq. (62) arises when the
DMFC anode is (methanol) dead-ended,
that is, all methanol fed into the anode at a
current equivalent rate of Jfeed either reacts
at the anode or recombines with oxygen at
the cathode. These operation conditions
apply in 13 W, passive DMFC platforms
developed for consumer electronics applications described later in this section. Under these conditions, fuel = Jcell /Jfuel feed
J
aan J0,an
J
acath J0,cath
bcath log10
J Rohm
J
1
Jlim,cath
Vcell
Jcrit
1
Jlim,an
J
= V0 ban log10
aan J0,an
J
1
Jlim,an
J
acath J0,cath
bcath log10
J Rohm
J
1
Jlim,cath
(64)
643
644
8 Fuel Cells
Before the next breakthrough in electrocatalysis of the methanol anode and/or the
air cathode, the only viable way to increase
the exchange current density by an order
of magnitude (without further increase in
the relatively high catalyst loading required
in the DMFC) is to increase the cell temperature. In fact, the lower and mid curves
in Fig. 52(a) describe well a DMFC operating close to ambient conditions and
near 80 C, respectively. Figure 53 describes a measured DMFC polarization
curve, recorded at 80 C with liquid aqueous 0.5 M methanol as anode feed and
air supplied near ambient pressure, showing a power output near 100 mW cm2
at 0.5 V close to four times higher than
generated at the same cell voltage at a cell
temperature near ambient.
Figure 52(b) describes the variations
of cell power output and cell energyconversion efciency with cell current in a DMFC with fuel utilization is
calculated as Jcell /Jlim,an (Eq. 63). This expression corresponds to the specic case
of a (methanol) dead-ended anode. Unlike the case of the hydrogen/air cell,
where fuel efciency is high at any cell
current, the power output and conversion
efciency of the DMFC increase hand in
hand with rise in cell current. Both power
output and conversion efciency reach
their peak slightly short of the cell (anode) limiting current and both drop with
further increase in cell current. In fact,
since cell = fuel v , fuel given in the
case of a dead-ended methanol anode as
0 ,
Jcell /Jlim,an and v is given by Vcell /Vcell
both cell and Pcell and increase with the
product Jcell Vcell and, consequently, peak
power and peak efciency are reached at
an identical point on the polarization curve
(Fig. 52b). A very similar pattern applies
also for the case of a DMFC with a recirculated anode feed stream of controlled
Anode
backing
Membrane
Cathode
backing
Fig. 51
of methanol and water from a DMFC anode fed by liquid aqueous methanol. This
liquid ux across the membrane is driven
primarily by a concentration gradient in
the case of methanol and by electroosmotic drag, that is, cell current in the case
of water. Possible impacts of methanol
penetration into the cathode start from the
simplest case [110], where the processes
of oxygen reduction and methanol oxidation each occur without interference by the
other and the component cathode currents
of opposite signs add algebraically. A more
severe situation at the DMFC cathode corresponds to cathode ooding, which can
occur as a result of loss of hydrophobic
nature of the cathode backing following
wetting by aqueous methanol.
Noninteracting processes of oxygen reduction and methanol oxidation at the
DMFC cathode result in a mixed potential corresponding to increase of cathode
overpotential at some given cell current.
/(Jcell Jcrossover )]
cath = bcath log[Jcell
/Jcell )]
= bcath log[Jcell
(65)
645
8 Fuel Cells
1.2
1.1
1
J0 = 1e-5 A cm2
0.9
Cell voltage
[V ]
0.8
J0 = 1e-6 A cm2
0.7
0.6
0.5
Base case
0.4
0.3
0.2
0.1
0
0
0.025
0.05
0.075
0.1
0.125
0.15
J0 = 1e-5 A cm
0.25
0.72
Note:
0.6
J0 = 1e-6 A cm2
0.2
0.48
0.36
0.15
Base case
101 mW cm2
0.1
0.24
74 mW cm2
46 mW cm2
0.05
0.025
0.05
0.075
0.1
0.125
(a)
646
0.12
0
0.15
(b)
Fig. 52 (a) DMFC polarization curves calculated according to Eq. (64) which assumes ultrathin
electrodes, negligible effect of methanol crossover, and zero-order anode process in methanol up to
Jcrit = Jlim,an 0.01 A cm2 . The three calculated polarization curves, corresponding to 3 orders of
magnitude of the exchange current density of the cathode and anode (assumed identical), reect
state-of-the-art polymer electrolyte DMFC performance near ambient conditions (bottom curve) and
near 80 C (mid curve), as well as the further improvement in cell performance expected at 80 C
if/when an increase by another order of magnitude of the catalytic activities of DMFC cathode and
anode is achieved; (b) calculated power output and energy-conversion efciency as function of cell
polarization, corresponding to the three cases in (a).
The specic parameters used for calculating the curves in Fig. 52 are as follows:
Rohm = 0.375 cm ,
Cell voltage
[V]
0.8
0.6
0.4
0.2
0.0
50
100
150
200
250
300
Current density
[mA cm2]
Fig. 53
because of the high rates of liquid penetration from a liquid-fed DMFC anode across
the membrane into the cathode. The water
ux is driven by an electroosmotic drag
of 2.53.5 water molecules per proton,
that is, nearly 20 molecules of water per
molecule of methanol reacting at the anode
travel across the membrane in a DMFC
with an aqueous methanol anode feed
[33]. Methanol carried along into the cathode further exacerbates cathode ooding
by negatively affecting the cathode wetproong characteristics. Water buildup in
the DMFC cathode can, in fact, be reduced by minimizing the rate of methanol
crossover through the range of DMFC operation conditions, that is, adjusting the
fuel feed rate to the cell current as the latter varies (see Eq. 61). Additionally, proper
design of the cathode ow eld, to achieve
effective liquid removal by airow across
the cathode surface can assist in sweeping excess cathode liquid with the air
exhaust. Cathode features determining the
647
8 Fuel Cells
efciency of water removal include a cathode backing with a proper degree of openness and strong hydrophobic surface characteristics and, at the same time, ow channels with hydrophilic surfaces, facilitating
transfer of water droplets from the cathode
GDL to the airow channel (see Fig. 47).
Figure 54 shows that, with effective design
of the DMFC air cathode, cell performance
is maintained high at very low stoichiometric ow of air. In fact, it is within 90% of
the performance measured with very high
airow rates, shown in Fig. 53. The ability
to operate at such low airow rates has
clear advantages in both minimizing parasitic power losses and in avoiding excessive
water losses from the cell.
8.3.9
Signicant interest has been shown recently by consumer electronics manufacturers and by the military sector in
low-temperature portable fuel cells, particularly portable DMFC. Both civilian and
military sectors are interested in a nextgeneration, small portable power source
of energy density that signicantly exceeds
that of leading-edge battery technologies.
This means a power system energy density
well beyond 400 Wh L1 (0.4 Wh cm3 )
and/or a specic energy well beyond
200 Wh kg1 (0.2 Wh g1 ) to be achieved
while providing the power for the full
dynamic range of the load in a given device. Advanced power sources able to meet
and beat these specications will allow
greatly extended run times per recharge
(refuel) for mobile phones, laptop computers, and digital cameras and also for
radios, night-vision goggles, global positioning systems, and laser-targeting units
in the military arena. Perhaps most even
more signicantly, higher-energy-density
power sources are likely to be essential
building blocks in converged handheld
communications devices that combine
0.9
0.8
0.7
Cell voltage
[V]
648
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0.00
0.05
0.10
0.15
0.20
0.25
Current density
[A cm2]
Fig. 54
color-display technology with voice, broadband data, and video functionality all in
a single power-hungry platform.
The strong potential advantage of fuelcell power sources over batteries originates from the high energy density
of the fuel. For low-temperature fuel
cells (Tcell < 100 C) applicable for manportable devices, the theoretical specic
energies of the two most suitable fuels are 33 Wh g1 for hydrogen and
6.1 Wh g1 for methanol. In comparison, advanced rechargeable lithium-ion
batteries have a specic energy of around
0.2 Wh g1 less than 1% of the energy
density of neat hydrogen and 3% that
of neat methanol. Such a comparison
is obviously skewed, because the battery
comprises a complete power source in one
package, whereas the fuel for a fuel cell
still has to be packaged with the chemicalto-electrical energy-conversion device (the
fuel cell itself) plus any further auxiliaries,
including power conditioning and power
system controls. Figure 55 is a building
block scheme of a portable power source
based on an air-breathing fuel cell fueled
by liquid methanol. The power system includes a fuel-containment chamber and
the volume fraction taken by the fuel is
shown to be relatively large, as this is
prerequisite for overall system energy density advantage over the incumbent battery
technology. The total volume of portable
power sources of this type, designed for
laptops, for example, will have to be limited
to about 200 cm3 , whereas for handsets, it
will have to ultimately drop under 30 cm3 .
These very small volumes will have to
accommodate enough methanol fuel to
provide signicantly more energy to the
load than an incumbent battery of similar volume. The methanol fuel will have
to share, however, the small total volume
available, with the fuel cell itself and all
auxiliaries required to generate power outputs in the range of 1030 W for a laptop
computer and 13 W for a handset.
To summarize the effects of these key
wt of the total
parameters, if a fraction ffuel
weight of a power system is the fuel itself
and the systems overall energy-conversion
efciency (chemical-to-electrical energy to
the load) is system , then the actual
wt
specic energy of the power pack, Esystem
1
(Wh g ), will be given by Eq. (1).
wt
wt
wt
= ffuel
Efuel
system
Esystem
(66)
Fuel
Battery
Power
management
and control
Water collection
and pumping
Fuel feed
Fuel cell array
649
650
8 Fuel Cells
(67)
(i) Fuel-cell stack technology of minimum weight (or volume) requirement to generate the average power
demand of a portable device under the benign operating conditions
dictated by handheld applications.
651
8 Fuel Cells
system volume versus that of a lithiumion battery of energy density 0.2 Wh cm3 .
The volumes are compared for a range of
overall system energy content demands
or overall use time (the number of Whs
equals the number of hours of use at the
1-W power output considered). At fuelcell stack power density of 100 mW cm3
(100 W L1 ), the energy density of a
DMFC system is projected to exceed
that of the lithium-ion battery for use
times exceeding 46 hours assuming
total auxiliaries volume of 10 cm3 , equal
to that of the fuel cell stack itself and
assuming system conversion efciency
near 30%. The additional parallel lines in
the gure show that, even if the volume
160
Prismatic Li-ion battery
@ 0.2 Wh cm3
140
120
100
volume
(cm3)
652
80
60
40
DMFC @
1.5 Wh per cm3 fuel
20
0
10
12
14
16
18
20
22
24
26
28
30
Energy
[Wh]
Fig. 56 System volume for a 1-W DMFC power source versus number of watt-hours
stored (= hours of use time at 1-W to the load per single fuel cartridge) compared to
the volume of a prismatic Li-ion battery of the same number of watt-hours. The three
cases shown are for three different volumes of the nonfuel-containing components of
the DMFC system, that is, stack + auxiliaries. Assumptions for the DMFC system are:
stack power density of 50100 W L1 (0.050.1 W cm3 ), auxiliaries volume = stack
volume, and system conversion efciency = 30%.
653
654
8 Fuel Cells
Control logic
Neat
methanol
Fluidics
and electronics
Fuel cell
Anode in
DC/DC
converter
Cathode
ambient
air
Pump driver
MeOH feed pump
Water recovery pump
Recirculation pump
Water
Anode out
(a)
Control logic
Neat
methanol
Fluidics
and electronics
Fuel cell
DC-DC
converter/
power conditioning
Cathode
ambient
air
MeOH feed
H2 O
Water
(b)
Fig. 57
655
656
8 Fuel Cells
MTI Micros MobionTM DMFC and methanol cartridge
integrated into intermec Inc.s RFID reader
DMFC system
characteristics:
Cartridge:
100% MeOH
Power out:
1W
Energy to load:
in continuous
discharge:
35 Wh*
(*Present battery is
8 Wh)
Fig. 58 Air-breathing DMFC cell array and methanol cartridge integrated
into the handle of an RFID tag reader made by Intermec, Inc.
Methanol
cartridge
Microfuel cell
FOMA handset
attached to
microfuel cell
(a)
Fuel
cell
Fuel
cartridge
Power cord
to laptop
Auxiliary
battery
(b)
657
658
8 Fuel Cells
References
659
660
8 Fuel Cells
31. M. Eikerling, A. A. Kornyshev, Y. Kharkats
et al., J. Electrochem. Soc. 1998, 145, 2684.
32. K.-D. Kreuer, T. Dippel, W. Meyer et al.,
Mat. Res. Soc. Symp. Proc. 1993, 293, 273.
33. X. Ren, W. Henderson, S. Gottesfeld, J.
Electrochem. Soc. 1997, 144.
34. T. Fuller, J. Newman, J. Electrochem. Soc.
1992, 139, 1332.
35. T. A. Zawodzinski, J. Davey, J. Valerio et al.,
Electrochim. Acta 1995, 40, 297.
36. T. E. Springer, T. A. Zawodzinski, S. Gottesfeld, J. Electrochem. Soc. 1991, 138, 2334.
37. Cover page of the Oct/Nov 2005 issue of Fuel
Cell Review, Institute of Physics Publishing.
38. C.-Y. Wang, Chem. Rev. 2004, 104, 4727.
39. D. Bernardi, M. Verbrugge, J. Electrochem.
Soc. 1992, 139, 2477.
40. G. J. M. Janssen, J. Electrochem. Soc. 2001,
148, A1313.
41. (a) S. Mukerjee, S. Srinivasan, J. Electroanal. Chem. 1993, 357, 201; (b) S. Mukerjee,
S. Srinivasan, M. P. Soriaga et al., J. Phys.
Chem. 1995, 99, 4577.
42. T. E. Springer, T. Rockward, T. A. Zawodzinski et al., J. Electrochem. Soc. 2001,
148(1), A11.
43. US Patent No. 3,297,484 (Niedrach, 1967).
44. US Patent No. 4,272,353, June 1981 (R.J.
Lawrence).
45. H. Takenaka, E. Torikai, Kokai Tokyo Koho
(Japan Patent) 1980, 55, 38934.
46. P. S. Fedkiw, W.-H. Her, J. Electrochem. Soc.
l989, 136, 899.
47. P. Aldbert, F. Novel-Cattin, M. Pineri et al.,
Solid State Ionics 1989, 35, 3.
48. S. Srinivasan, E. A. Ticianelli, C. R. Derouin
et al., J. Power Sources 1988, 22, 359.
49. E. A. Ticianelli, C. R. Derouin, S. Srinivasan, J. Electroanal. Chem. 1988, 251, 275.
50. S. Mukerjee, S. Srinivasan, A. J. Appleby,
Electrochim. Acta 1993, 38, 1661.
51. M. S. Wilson, F. Garzon, K. E. Sickafus et al., J. Electrochem. Soc. 1993, 140,
2872.
52. S. C. Thomas, X. Ren, S. Gottesfeld, J.
Electrochem. Soc. 1999, 146, 4354.
53. (a) M. K. Debe, in Ref. 5 Vol. 3, Chapter 45;
(b) Proceedings of the Fuel Cell Workshop
in Kofu, Japan, 2005.
54. J. F. McElroy, Abstracts of the Fuel Cell
Seminar, Courtesy Associates, Phoenix,
1990, p. 282.
55. D. P. Wilkinson, J. St-Piere, in Ref. 5, Vol. 3,
p. 618.
93.
94.
95.
96.
97.
98.
99.
100.
101.
102.
103.
104.
105.
106.
107.
108.
109.
110.
111.
112.
113.
661
663
9.1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
665
9.2
9.2.1
9.2.2
9.2.3
9.2.4
Water Radiolysis . . . . . . . . . . . .
Radiolytic Yield . . . . . . . . . . . . .
Chemical Reactions . . . . . . . . . .
Convection . . . . . . . . . . . . . . . .
General Solution of the Equations
.
.
.
.
.
668
669
671
672
672
9.3
673
9.4
678
9.5
691
9.6
711
9.7
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
719
720
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
665
9.1
Introduction
666
Shippingport
Dresden, Fermi I
Magnox
1950
Gen I
1960
Generation III
Commercial power
reactors
Advanced
LWRs
LWR-PWR, BWR
CANDU
VVER/RBMK
1970
Gen II
1980
1990
ABWR
System 80+
AP600
EPR
Gen III
2000
2010
Near-Term
Deployment
Generation IV
Generation III+
evolutionary
designs offering
improved
economics
Gen III+
2020
Highly
economical
Enhanced
safety
Minimal
waste
Proliferation
resistant
Gen IV
2030
Fig. 1
9.1 Introduction
Fig. 2
Containment/drywell
Reactor
vessel
Steam
line
Turbine building
Throttle
valve
Steam dryer
and
moisture
separater
Electrical
generator
Turbine
Reactor core
Condenser
Jet pump
To/from
river
Recirculation
pump
Pump
Containment suppression chamber
Fig. 3
667
668
9.2
Water Radiolysis
Radiolytic Yield
669
670
Number
Gamma4
285 C
Neutron4
285 C
Alpha5
270300 C
1
2
3
4
5
6
7
8
9
10
11
12
13
14
1.15
1.08
3.97
1.25
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.62
0.00
4.15
0.93
0.50
1.09
0.99
0.04
0.00
0.00
0.00
0.00
0.00
0.00
0.88
0.00
0.93
0.13
0.12
0.45
1.55
0.00
0.00
0.00
0.00
0.00
0.00
0.00
1.65
0.00
0.13
Species
e
H
OH
H2 O 2
HO2
HO2
O2
O2
O2
O
O
H2
OH
H+
Chemical Reactions
N
N
s=1 m=1
ksm Cs Cm Ci
N
ksi Cs
s=1
(2)
where ksm is the rate constant for the
reaction between species s and m, ksi is
the rate constant for the reaction between
species s and i, and Ci , Cm , and Cs
are the concentrations of species i, m,
and s, respectively. N is the number
of reactions in the model (i.e., N = 48).
Explicit expressions for the gain and the
loss of each species are summarized
elsewhere [6].
The rate constant, kj is a function of
coolant temperature. Since the temperature throughout the heat transport circuit
is not constant, the actual rate constant for
each chemical reaction must be calculated
for each specic position using Arrhenius
law:
1
Ea 1
k = ko exp
(3)
R To
T
where ko is the rate constant at temperature To , Ea is the activation energy [6], R
671
672
(4)
9.2.3
Convection
9.2.4
s=1
(7)
The equations must be solved iteratively
until satisfactory convergence is obtained.
Note that, because the mass ow rate
(dm/dt) in a single (unbranched) channel
9.3
After the concentration of each radiolysis species is calculated, the corrosion potential of the component can
be estimated using a mixed potential
model (MPM) [35]. The MPM is based on
the physico-electrochemical condition that
charge conservation must be fullled at
the corroding interface. Because electrochemical reactions transfer charge across
a metal/solution interface at rates measured by the partial currents, the following
673
674
(8)
j =1
(9)
where
X=
e(EEo )/bf
2.61 103 e4416/T +0.523(EEo )
(10)
0.5
and
Eo = 0.122 1.5286 103 T
(11)
3
k0
1
(13)
+
ln 30
3 K
k7
The parameters in these equations are
as follows: , , , and n are the
oxidation state of the cation in the barrier
layer; the polarizability of the lm/solution
interface (i.e., the dependence of the
potential drop across the lm/solution
interface on the applied voltage); the
dependence of the potential drop across
the lm/solution interface on the pH; and
the kinetic order of the lm dissolution
reaction with respect to hydrogen ion
concentration, respectively. Note that, in
deriving Eqs (12) and (13), the oxidation
state of the cation in the barrier layer
() is set equal to the oxidation state
of the same cation in the solution/outer
layer. The standard rate constants, ki0 ,
and i correspond to the reaction shown
in Fig. 4, is the electric eld strength,
= F /RT , and K = . The three terms
on the right side of Eq. (12) arise from
the transmission of cation interstitials,
the transmission of cation vacancies, and
the transmission of oxygen vacancies
(or dissolution of the lm), respectively.
Values for these parameters are readily
obtained by optimizing the PDM on
Barrier layer
(1) m + Vm
k1
MM + m + ce
(2) m
k2
Mi + + m + ce
(3) m
k3
Mm +
k4
(4) MM
(5) Mi+
c
V + ce
2 O
k6
(6) VO + H2O
k1
MG+ (aq) +
k6
OO + 2H+
c
H2O + (G c)e
2
Fig. 4 Summary of the defect generation and annihilation reactions envisioned at the
interfaces of the barrier oxide layer on a metal, according to the PDM [37]. VM cation vacancy
+
in the lm; vm vacancy in the metal substrate; Mi cation interstitial; V
O oxygen (anion)
vacancy; M+ (aq) cation in solution; MM cation in cation site on the metal sublattice; OO
oxide ion in anion site on the oxygen sublattice; and MO/2 stoichiometric barrier layer
oxide. Note that Reactions 1, 2, 4, 5, and 6 are lattice conservative processes (they do not result
in the movement of the interface), whereas Reaction 3 and 7 are nonconservative.
or
H2 O + e 1/2H2 + OH
+
O2 + 4H + 4e 2H2 O
(14)
or
H2 O2 + 2e 2HO
(16)
(17)
e
(EER/O
)/ba
(EE e
)/bc
R/O
e
e
1 (EER/O
1
)/ba
+
e
i0,R/O
il,f
e
1 (EER/O
)/bc
e
il,r
(18)
675
676
Col 8 vs Col 9
Col 8 vs Col 12
0.1
ECP
[VSHE]
0.0
0.1
0.2
0.3
0.4
0.5
0.0
0.5
1.0
1.5
2.0
2.5
Feedwater [H2]
[mg kg1]
Fig. 5
0.0
ECP
[VSHE]
0.2
O2
H2O2
0.4
0.6
0.8
1.0
1e-6
1e-5
1e-4
1e-3
1e-2
1e-1
1e+0
1e+1
[O2] or [H2O2]
[mg kg1]
677
678
9.4
679
Fluid flow
Oxygen transport
Positive current
Positive current
O2 + 4H+ + 4 e 2H2O
Electron current
Ls
680
O2 + 4H+ + 4 e 2H2O
Crack
advance
Electron current
and
i =
(21)
F
zi Di Ci Di Ci
RT
(22)
681
682
1
Q 1
(288 C) exp
R T
561.15
(23)
where (T ) and (288 C, 561.15K) are the
crack-tip strain rate at temperature T and
288 C, respectively, and Q is the thermal
activation energy for the crack-tip strain
rate.
In this numerical version of the CEFM,
a potential Emouth is assumed at the
crack mouth, the boundary between the
crack-internal and the crack-external environments, and this potential is then
iterated until the external current (Iext ,
obtained via solving Laplaces equation)
and the internal current (Iint , obtained
via solving the NernstPlanck equations)
match for the prevailing Ecorr (Fig. 8), that
is, until Iint = Iext . This is the practical
implementation of the CEFM statement
CGR
Emouth
Internal
current
External
current
No
Iint = Iect?
Yes
Electrochemical
crack advance
End
(26)
Internal current
Etip
Concentration profile
in crack
No
Yes
End
683
684
1.00E-07
1.00E-08
0
50
100
150 200
Temperature
[C]
250
300
350
Fig. 10
685
6.4
ECp
Emouth
Etip
PH
External polarization
6.3
6.2
6.1
0.1
0.2
Internal polarization
5.9
0.3
5.8
0.4
5.7
0.5
5.6
0.6
[-]Hd
Electrochemical potential
[V]
686
5.5
50
100
150
200
Temperature
[C]
250
300
350
Side grooves
Fatigue crack
Intergranular crack
Mechanical fracture
687
1e-6
1e-7
1e-8
1e-9
1e-10
1e-11
0.8
0.6
0.4
0.2
0.0
0.2
L = 0.1 cm
L = 0.3 cm
L = 0.5 cm
L = 0.7 cm
L = 1.0 cm
L = 1.5 cm
L = 2.0 cm
L = 2.5 cm
L = 3.0 cm
L = 4.0 cm
L = 5.0 cm
L = 7.0 cm
L = 10 cm
L = 15 cm
L = 20 cm
L = 30 cm
L = 50 cm
0.4
Corrosion potential
[VSHE]
Fig. 13
1e-4
1e-5
1e-6
Coupling current
[A]
688
1e-7
1e-8
1e-9
1e-10
1e-11
1e-12
1e-13
1e-10
L = 0.1cm
L = 0.3cm
L = 0.5cm
L = 0.7cm
L = 1.0cm
L = 1.5cm
L = 2.0cm
L = 2.5cm
L = 3.0cm
L = 4.0cm
L = 5.0cm
L = 7.0cm
L = 10 cm
L = 15 cm
L = 20 cm
L = 30 cm
L = 50 cm
1e-8
1e-7
1e-6
1e-5
Plots of coupling current versus crack growth rate for ECLs ranging from
0.1 to 50 cm for constant stress intensity and for the other, constant conditions
stated in the gure [54].
Fig. 14
Coupling current
[A]
1e-6
1e-7
L = 0.1 cm
L = 0.3 cm
L = 0.5 cm
L = 0.7 cm
L = 1.0 cm
L = 1.5 cm
L = 2.0 cm
L = 2.5 cm
L = 3.0 cm
L = 4.0 cm
L = 5.0 cm
L = 7.0 cm
L = 10 cm
L = 15 cm
L = 20 cm
L = 30 cm
L = 50 cm
10
cm s1.
1e-8
1e-9
1e-10
1e-11
1e-12
1e-13
0.8
0.6
0.4
0.2
0.0
0.2
0.4
Corrosion potential
[VSHE]
Fig. 15
689
Experimental data6
0.1 ms cm1- Ford's correlation6
0.5 ms cm1- Ford's correlation6
0.1 ms cm1- Congleton's correlation14
0.5 ms cm1- Congleton's correlation14
106
107
k = 0.3
k = 0.2
k = 0.1
690
288 C water
Oonstant load
108
Assumed condition
25 ksi-in (27.5 MPa,m)
Observed data
0.10.3 mS cm1
109
15 C cm2
Calibration datum
1010
0.8
0.6
0.4
0.2
0
Corrosionpotential
[VSHE]
0.2
0.4
0.6
IGSCC crack growth rate versus corrosion potential (ECP) for sensitized
Type 304 SS in high-temperature water as a function of solution conductivity (25 )
for given values for the degree of sensitization (15 C cm2 ) and stress intensity
(27.5 MPa.m1/2 ). The experimental data were measured for ambient temperature
conductivities, 25 , ranging from 0.1 to 0.3 S cm1 . The citations given in the
gure are those presented in Ref. 43.
Fig. 16
9.5
691
692
Steam dryer
assembly
Steam outlet
Steam separator
assembly
Feedwater sparger
Low-pressure coolant
injection inlet
Jet pumps
(beams)
Top guide
Full assemblies
Recirculation
inlet
(piping)
Core shroud
Control blade
(absorber tubes)
Recirculation
(outlet)
Manifold
(piping)
Recirculation
(pump motor)
Shutoff valve
Bypass line
Shutoff valve
Recirculation pump
Key components of the BWR. the parts names in red indicates areas of IGSCC.
Fig. 17
Main steam L.
Steam separator
Feedwater
4
1
6
Fig. 18
10
693
694
Feedwater
f2
f2
steam
Component 3
Component 4
1f 2
Liquid
Upper plenum
Mixing plenum
Component 5
Upper downcomer
Component 1
Component 2
Core channel
Core bypass
1f 3
Component 6
Upper downcomer
f3
Component 7
f1
1f 1
Recirculation
system
Component 10
Top lower plenum
Component 8
Component 9
Jet pump
Fig. 19
695
696
1000
100
Measured
Calculated
10
1
0.0
0.5
1.0
1.5
2.0
Measured
Calculated
100
10
1
0.0
0.5
1.0
1.5
2.0
Fig. 20
697
UP MP UD LD RS
CC
0.30
0.20
JP
BLP TLP
CB
0.10
ECP
[V]
0.00
CC : Core channel
CB : Core bypass
UP : Upper plenum
MP : Mixing plenum
UD : Upper downcomer
LD : Lower downcomer
RS : Recirculation system
JP : Jet pump
BLP : Bottom lower plenum
TLP : Top lower plenum
0.10
0.20
0.30
0.40
0.50
0.60
0.70
1000
4000
5000
2000
3000
Flow path distance from core inlet
[cm]
6000
7000
0.30
UP
MP
UD
0.20
LD
RS
0.10
CB
0.00
ECP
[V]
698
0.10
CC : Core channel
CB : Core bypass
JP BLP
UP : Upper plenum
MP : Mixing plenum
TLP
UD : Upper downcomer
LD : Lower downcomer
RS : Recirculation system
JP : Jet pump
0.20
0.30
0.40
0.50
0.60
0.70
1000
2000
3000
4000
5000
Flow path distance from core inlet
[cm]
6000
7000
Fig. 21
106
UP
107 CC
LD
MP UD
RS
JP BLP TLP
CB
108
CC : Core channel
CB : Core bypass
UP : Upper plenum
MP : Mixing plenum
UD : Upper downcomer
LD : Lower downcomer
RS : Recirculation system
109
1010
1011
1000
JP : Jet pump
BLP : Bottom lower plenum
TLP : Top lower plenum
2000
3000
4000
5000
Flow path distance from core inlet
[cm]
6000
7000
107 CC
108
MP
LD
UD
RS
CB
109
CC : Core channel
CB : Core bypass
UP : Upper plenum
JP BLP
MP : Mixing plenum
UD : Upper downcomer
LD : Lower downcomer
RS : Recirculation system
TLP
JP : Jet pump
BLP : Bottom lower plenum
TLP : Top lower plenum
1010
1011
1000
2000
3000
4000
5000
Flow path distance from core inlet
[cm]
6000
7000
Predicted crack growth rate versus ow path distance from the bottom of
the core for 0 and 1.2 ppm of hydrogen added to the feedwater of the Leibstadt
BWR [19].
Fig. 22
699
Fig. 23
3.0
Linear
2.5
Crack depth
(cm)
700
Slope =
L
t
t =0
2.0
1.5
ALERT
1.0
Outages
0.5
0
12 24 36 48 60 72 84 96 108 120
Time
[month]
operating history), because the CGR decreases as the crack length increases [54],
as shown in Fig. 23. This is due to an
increase in the potential drop down the
crack, even though the mechanical driving
force (the stress intensity) is maintained
constant. The calculations shown in this
gure were carried out on a crack on the
inside of the core shroud adjacent to the
H-3 weld near the top of the shroud over a
120-month operating history of a reactor in
service. If the dependence of CGR on crack
length is not recognized, the integrated
damage function (IDF) is overpredicted by
several hundred percent, thereby leading
to a much more pessimistic evolution of
damage (Fig. 23).
Application of the models at each state
point yields a CGR, but the CGR depends
upon the value of parameters along
the CEP and upon the stress intensity
(although only weakly) and the crack
and
KI,i = (t)Y Li
(29)
120%
Power
100%
80%
60%
40%
20%
0%
20
40
60
80
100
120
140
100
120
140
Time
[months]
Feedwater hydrogen injection
1.2
[H2]
[ppb]
1.0
0.8
0.6
0.4
0.2
0.0
20
40
60
80
Time
[months]
Corrosion evolutionary path (CEP) dened in terms of the prime variables of reactor power
and hydrogen added to the reactor feedwater.
Fig. 24
701
702
O2 concentration
250
200
150
100
50
0
20
40
60
80
100
120
140
80
100
120
140
80
100
120
140
80
100
120
140
Time
[months]
H2
30
H2 concentration
[ppb]
25
20
15
10
5
0
20
40
60
H2O2 concentration
[ppb]
Time
[months]
500
450
400
350
300
250
200
150
100
50
0
H2O2
20
40
60
ECP
[mVSHE]
Time
[months]
450
400
350
300
250
200
150
100
50
0
ECP
20
40
60
Time
[months]
Plots of the concentrations of oxygen (ppb), hydrogen (ppb), and hydrogen peroxide
(ppb), and the calculated ECP (mVSHE ) at the top inner surface of core shroud as a function of
time along the CEP. Note that the concentration of hydrogen is considerably lower than the
feedwater hydrogen, because of the mixing of ows from the lower plenum and because
hydrogen is stripped from the coolant by boiling in the core.
Fig. 25
703
CGR
[pm s1]
1400
1200
1000
800
600
400
200
0
20
40
60
80
100
120
140
100
120
140
100
120
140
Time
[months]
KI
[Mpa.m]
20
40
60
80
Time
[months]
Crack length
3.0
2.5
Crack depth
704
2.0
1.5
1.0
0.5
0.0
20
40
60
80
Time
[months]
Plots of the concentrations of crack growth rate (pm s1 ), stress intensity (MPa.m1/2 ), and
the crack length (mm) at the top inner surface of core shroud as a function of time along the CEP.
Fig. 26
3.0
Crack depth
[cm]
2.5
2.0
1.5
1.0
0.5
0.0
0
20
40
60
Time
[months]
80
100
120
Fig. 27
705
0.6
UP MP
UD LD
RS
0.4
JP
LP
ECP
[VSHE]
0.2
0.0
0.2
0.4
0.6
0.8
600
1200 1800 2400 3000
Flow path distance from core inlet
[cm]
3600
Fig. 28
106
MP
CC UP
CB
UD LD
RS
LP
CC : Core channel
CB : Core bypass
UP : Upper plenum
MP : Mixing plenum
UD : Upper dowmcomer
LD : Lower dowmcomer
RS : Recirculation system
JP : Jet pump
LP : Lower plenum
108
109
1010
1011
0
JP
107
Crack growth rate
[cm s1]
706
1200
1800
2400
3000
3600
CCUP MP
UD LD
CB
0.2
RS
LP
CC : Core channel
CB : Core bypass
UP : Upper plenum
MP : Mixing plenum
UD : Upper dowmcomer
LD : Lower dowmcomer
RS : Recirculation system
JP : Jet pump
LP : Lower plenum
0.0
0.2
ECP
[VSHE]
JP
0.4
0.6
[H2]FW = 0.0 ppm
0.8
600
1200
1800
2400
Flow path distance from core inlet
[cm]
3000
3600
Fig. 30
108
CC UP MP
UD LD
CB
109
[H2]FW = 0.0 ppm
RS
JP
LP
CC : Core channel
CB : Core bypass
UP : Upper plenum
MP : Mixing plenum
UD : Upper dowmcomer
LD : Lower dowmcomer
RS : Recirculation system
JP : Jet pump
LP : Lower plenum
1010
[H2]FW = 1.0 ppm
1011
0
600
1200
1800
2400
Flow path distance from core inlet
[cm]
3000
Fig. 31
3600
707
inhibition, because they are the only models known to the author that explicitly
consider the electrochemical kinetics of
the redox reactions that occur on the steel
surface. Again, we emphasize that the
accumulation of damage due to SCC is primarily an electrochemical phenomenon,
and any quantitative, deterministic theory
must address the kinetics of the chargetransfer processes in the system.
Reference was made above to ULCO,
since it is instructive to explore the veracity
of this remedial measure. By way of background, IGSCC damage/conductivity correlations have been noted between damage
accumulated in good plants that have operated with low coolant conductivity, and
bad plants, in which the coolant conductivity was excessively high. The reader
will recall from Sect. 9.4 that conductivity
affects the CGR by allowing the coupling
current exiting the crack to ow to a greater
area on the external surface, where it is
consumed by oxygen reduction. Thus, the
external surface can support a larger coupling current, which translates into an
enhanced CGR. The impact of coolant
conductivity on accumulated damage for a
crack adjacent to the H-3 weld on the inside
surface of the core shroud of a GE Model 6
BWR under NWC over a 20-month period
of constant, full power operation is shown
4.5
Uncoated
4.0
Crack length
[cm]
708
3.5
3.0
Coated
Inhibition of IGSCC in
sensitized Type 304 SS by a dielectric
ZrO2 coating. The experiments were
2.5
0 50 100 150 200 250 300 350 400 450 carried out with two C(T) specimens
Time
daisy-chained together in the same
[hours]
solution (0.005 M Na2 SO4 ) at 288 C.
Fig. 32
in Fig. 33. The conductivity values correspond to the ambient temperature (25 C)
values; of course, the values at the elevated temperature as calculated from the
concentrations of ionic species and their
limiting conductance values are employed
in the calculation. The reader will note
from Fig. 33 that coolant conductivity has
a signicant impact on the accumulated
damage and that the crack length might be
reduced from 1.9 cm to a little over 1.4 cm,
from an initial value of 0.5 cm, by reducing the ambient temperature conductivity
from 0.3 to 0.1 S cm1 . While signicant,
the gain is modest when compared against
HWC, particularly with NMCA, and a detailed cost/benet analysis would have to
be carried out to determine whether the
addition of the necessary facilities (additional ion exchange beds and larger ow
fraction through the reactor water cleanup
unit) are justied.
2.0
1.8
1.6
Crack depth
[cm]
1.4
1.2
1.0
0.10 ms cm1
0.8
0.15 ms cm
0.20 ms cm
0.6
0.30 ms cm1
0.4
0.2
0.0
0
10
Time
[months]
15
20
Crack depth versus time curves for a crack at the H-3 weld in the shroud as a
function of the coolant conductivity.
Fig. 33
709
Crack depth
[cm]
710
1.0
0.8
15.0 MPa.m
20.0 MPa.m
0.6
27.5 MPa.m
35.0 MPa.m
45.0 MPa.m
0.4
0.2
0.0
10
Time
[months]
15
20
Crack depth versus time curves for a crack at the H-3 weld in the shroud as a
function of the stress intensity factor.
Fig. 34
Crack depth
[cm]
2.0
1.5
ISI
ALERT
1.0
0.5
0.0
0
10
12
14
16
18
20
Depth of a crack at the H-3 weld in the shroud as a function of time (months)
after Outage 11. values calculated by ALERT. reported by Tang et al. [63].
Fig. 35
9.6
Compared with the work that has been reported on modeling BWR primary coolant
circuits, relatively little work has been
carried out to assess electrochemical effects in PWR primary circuits. This state
of affairs reects the fact that cracking
has not been as great a problem in
PWR primary coolant circuits as it has
been in BWR primary coolant circuits.
Secondly, because of the high hydrogen concentration (25 cc (STP)/kg (H2 O)
50 cc (STP)/kg (H2 O) corresponding
to 1.12 103 2.24 103 m) employed
in a PWR primary circuit to suppress
radiolysis, and in view of the lack of sustained boiling, it was generally believed
that the ECP was dominated by the hydrogen equilibrium potential and hence
that the coolant circuit acts as a giant
711
712
Pressurizer
spray
control and
needle
valves
Safety valves
(3)
Reactor drain
tank
Charging
line
Pressurizer
SG2
SG1
Loop 2
Reactor
Loop 1
Shutdown
cooling
Shutdown
cooling
Safety injection
Fig. 36
Letdown
Typical conditions that exist in the main loop of the primary coolant circuit of a PWR
Property
Value
Comment
Temperature
Pressure
Coolant composition
Typical
Typical
LiB trajectory over a typical
fuel cycle is shown in Fig. 38
Hydrogen concentration
295330 C
150 bar (2250 psi)
40000 ppm B as boric acid,
41 ppm Li as lithium
hydroxide, depending upon
the burn-up of the fuel and
the vendor.
2555 cc (STP)/kg (H2 O)
3 105 rad s1
6 105 rad s1
3 105 rad s1
18 000 kg s1
Typical
713
714
Holddown
ring
Cea shroud
assembly
Upper
guide
structure
Inlet
nozzle
Alignment
key
Core
support
barrel
Outlet
nozzle
Core
shroud
150"
active
core length
Instrumentation
assembly
Fuel
assembly
Lower
support
structure
Flow
baffle
Instrumentation
nozzle
Fig. 37
4.0
3.5
pH
7.0
pH
7.1
pH
7.2
pH
7.4
pH
6.9
3.0
2.5
2.0
1.5
1.0
0.5
0
2000
Principal 1
Do not operate
below pH = 6.9
Principal 2
Materials and fuel review
prior to operation
in this area
Principal 3
Maintain 2.2 0.15 ppm
Principal 4
li until selected pH
Maintain selected pH
is reached
at 0.15 ppm li
1600
1200
800
400
Fig. 38
operation. The radiolysis of water is predicted using the full 48-model radiolysis
model [32] with the radiolytic yields given
in Table 1. The calculations were carried
out at 10-cm increments, with the ECP being calculated at each increment using the
MPM (Sect. 9.3). Dose rates for photons,
neutrons, and particles, coolant path
geometry, and detailed thermal hydraulic
data for each location were provided by the
vendor for whom the modeling work was
performed.
The primary coolant circuit loops
were divided into 28 sections (cf, 10 or
11 sections in the BWR case) as shown
in Fig. 39. Each section is characterized
by an inlet and outlet temperature; , n,
and dose rates; ow fraction; linear ow
velocity; hydrodynamic diameter; section
length; material of construction (SS
715
716
Fig. 39
S10
SG
S16
S11
S9
S8
S17
S4 S5 S6 S7
S15
Press.
RHRS
S2 S3
CORE
S18
S14
S1
S19
S13
S21S27
S28
RWCU
S20
S12
stainless steel); and exchange current density multipliers, as in the case of the BWR
models discussed earlier. If the inlet and
outlet temperatures are different, it is assumed that the temperature varies linearly
with distance between the two values. The
code was used to model the ECP around
the primary coolant system for normal operation with deoxygenated (5 ppb O2 ) or
oxygenated makeup water (5.0 ppm O2 )
and for cold shutdown conditions. Facilities are included in the PWR ECP code
for modeling the periodic injection of oxygen and hydrogen peroxide into the RWCU
and oxygen into the RHRS. The concentration of hydrogen was set to 55 cc (STP)/kg
(H2 O) and the coolant mass ow rate was
17 880 kg s1 .
Figure 40 displays the predicted ECP
versus distance from the bottom of the core
for full power, normal operation, which
includes the injection of 5 mg kg1 O2 into
the feedwater. All of the sections, except
Section 7, display ECP values that lie
ECP
[V]
0.2
S2: Section 2
S5: Section 5
S3: Section 3
S6: Section 6
S4: Section 4
S7: Section 7
0.0
0.2
0.4
0.6
0.8
1.0
100
200
300
400
500
Fig. 40
0.8
0.6
0.4
ECP
[V]
0.2
S12
0.0
0.2
S8 S9
S10
S1
S11
S6
0.4
B
A
0.6
SG
0.8
1.0
Core
1000
2000
3000
4000
5000
6000
Flow path distance from core inlet
[cm]
Hot leg
7000
8000
Cold leg
ECP versus distance for the main primary coolant loop in a PWR under full
Normal power operation.
Fig. 41
717
0.6
0.4
ECP
[V]
0.2
0.0
S15 S16
0.2
S14
0.4
S13
0.6
0.8
1.0
1000
2000
3000
4000
5000
6000
Flow path from pressurizer inlet
[cm]
7000
8000
ECP versus distance for the pressurizer loop in a PWR under full power
operation.
Fig. 42
S25 S26
0.6
S27
S28
S22
0.4
S24
S23
0.2
ECP
[V]
718
0.0 S21
0.2
0.4
0.6
S20
0.8
1.0
ECP versus distance for the reactor water cleanup unit (RWCU) in a PWR
under full power operation.
Fig. 43
9.7 Summary
719
720
9.7 Summary
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
721
722
43.
44.
45.
46.
47.
48.
49.
50.
51.
52.
53.
54.
55.
56.
57.
58.
59.
60.
61.
62.
63.
64.
65.
66.
723
10
10.1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
725
10.2
10.2.1
10.2.2
10.2.3
10.2.4
.
.
.
.
.
726
730
732
735
736
10.3
High-temperature Potentiometry . . . . . . . . . . . . . . . . . . . . . . . .
736
10.4
High-temperature pH Measurements . . . . . . . . . . . . . . . . . . . . .
739
10.5
741
10.6
741
10.7
744
10.8
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
745
746
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
725
10.1
Introduction
726
10.2
(1)
(2)
O + ne R
kb
(3)
(4)
(7)
(8)
(13)
GO =
GoO
oxid nF E
727
728
as
red nF (E eq E )
o
nF k aO exp
RT
oxid nF (E eq E )
= nF k o aR exp
RT
= jo
(14)
eq
E =E +
(15)
ln
nF
aR
for the electrochemical reaction presented
by Eq. (3). This equation forms the basis
for potentiometric measurements so that
electrode systems are often referred to as
exhibiting Nernstian behavior or response
during their calibration when the measured potential, E eq , is a linear function
of ln(aO /aR ) with a theoretical slope of
RT /nF as dictated by Eq. (15).
Taking thermodynamics into account,
the electric potential of an individual
eq
electrode, Ei , can be related to the
Gibbs energy, r Gi , of the corresponding
electrochemical half-reaction as follows:
eq
r Gi = zi F Ei
(16)
(18)
Because E eq is an experimentally
measurable value and E 0 can be obtained using an appropriate extrapolation,
r G0 can be, in principle, experimentally
derived using a high-temperature potentiometric system, a system where the electrochemical reactions are reversible and
the electrode potential difference should
be measured using a high-impedance
electrometer. Moreover, if r G0 can be
calculated using available thermodynamic
data, the activities (and/or fugacities) of
the chemical species can be experimentally
derived. Therefore, the high-temperature
potentiometric measurements can be a
powerful tool in studying a number
of thermodynamic phenomena if (1) the
electrode electrochemical reactions are
reversible and (2) the high-temperature
potential measurements are reliable and
reproducible within a few millivolts or
less. It should be mentioned that, if the
high-temperature thermodynamic properties (the standard Gibbs energies and the
activity/fugacity coefcients) are available
for all species of an electrochemical reaction, both E 0 and E eq can be theoretically
calculated and compared with observed
potentiometric data. In this way, the reliability of an experimental potentiometric
system can be conrmed.
For reliable high-temperature potentiometric studies, the measured open-circuit
potential should be stable and reproducible
within a few millivolts or less. Similar
to the low-temperature electrochemical
measurements, both the indicator and
reference electrodes should be used in
devising a high-temperature potentiometric cell. Both the electrochemical probes
should be resistant to chemical degradation and pressure (mechanical) stress in
high-temperature aqueous environment.
The development of suitable indicator
and reference electrodes that can reliably
729
730
r G0Ag/AgCl
=
F
f G0Ag + f G0Cl f G0AgCl
F
(20)
731
732
Fig. 1
Ag|AgCl
PTFE tube
Zirconia tube
Metal tube
junction remains constant and any uncertainty in the thermal diffusion potential
can be eliminated for measurements at a
given temperature and pressure. If desirable, the thermal liquid junction potential
can be either calculated or experimentally
estimated and, in this way, the electrode
potential value can be evaluated with
respect to the SHE scale. In the studies carried out so far, the ow-through reference
electrode was used in a thermocell, which
has either platinum or yttria-stabilized
zirconia (YSZ) indicator electrodes as
pH sensors for potentiometric measurements up to 400 C [13]. In these tests,
0.1 mol kg1 NaCl aqueous solution was
used as a nonisothermal electrolyte bridge
(reference electrode solution) to electrically connect the Ag/AgCl electrochemical
The indicator electrode must have a stable and reproducible potential for a course
of measurement and should be able to
respond in a Nernstian manner to varying conditions in the high-temperature
aqueous environment. In other words,
the activity of the dissolved species, ai ,
and the standard open-circuit potential,
E 0 , should be, in principle, denable
by measuring the open-circuit potential between the indicator and reference electrodes and applying the Nernst
equation (17).
While a number of the indicator electrodes have been tested for operation over
a wide range of temperatures, only the
0
EYSZ(Hg/HgO) = EYSZ(Hg/HgO)
2
aH+
RT
+
ln
2F
aH2 O
0
should be mentioned that EYSZ(Hg/HgO)
does not depend on the YSZ membrane
properties and the only requirement is
a sufcient O2 conductivity at the temperature of interest. It was experimentally
found that when using a modern highimpedance (>1014 ) electrometer the lowest temperature for using an YSZ membrane can be around 100 C. Therefore, the
YSZ(Hg/HgO) electrode cannot be practically used at temperatures below 100 C.
The Pt(H2 ) indicator electrode has been
widely used in measuring pH in the
concentration cells at temperatures below
300 C [2, 14] and in the ow-through electrochemical systems at temperatures up
to 400 C [13]. The schematic of the owthrough Pt(H2 ) electrode, which has been
used together with the ow-through reference electrode described above, is given in
Fig. 2 [13]. The YSZ(Hg/HgO) indicator
(22)
Solution
0
EYSZ(Hg/HgO)
where
is the standard
YSZ(Hg/HgO) electrode potential, which
does not depend on the solution composition but depends on the standard
Gibbs energies of formation, f G0i , of
the species involved in the electrochemical reaction [H+ (aq), HgO(s), Hg(l), and
H2 O(l)]:
0
EYSZ(Hg/HgO)
r G0YSZ(Hg/HgO)
+f G0Hg
2O
0
f GHgO 2f G0 +
H
f G0H
ZrO2 tube
Ti tube
2F
2F
(23)
ZrO2 capillary
Pt coil
733
734
Solution
PEEK tube
Teflon
sealant
Stainless steel
fittings
Chemgrip epoxy
adhesive glue
Shrinkable teflon
tube
Hastelloy B tube
Copper
cooling coil
ZrO2 sand
HgO|Hg paste
Fig. 3
Pt coil
with the above-described ow-through reference electrode, is given in Fig. 3 [16, 18].
The usefulness of a number of
metal/metal-oxide (e.g. Ir/IrO2 , Zr/ZrO2 ,
W/WO2 , etc.) electrodes and the glass
electrode has been tested over a wide range
of temperatures. However, the existence
of the Nernstian behavior has not been
well demonstrated yet. The glass electrode
can probably be employed at temperatures
up to about 200 C but was found to
be impractical owing to an inconvenient
design for high-temperature subcritical
and supercritical aqueous solutions.
10.2.3
Undesirable Potentials
735
736
such that the concentration of NaCl is signicantly greater than the concentration
of another solute (such as HCl or NaOH).
The approach was theoretically and experimentally conrmed at temperatures up to
400 C and densities down to 0.17 g cm3
and does not require any knowledge of the
association constants [20].
The buffer solutions play a signicant
role in low-temperature potentiometric
and electrochemical kinetics measurements. It is highly desirable to estimate
a set of the buffer solutions that can be
used at temperatures above 100 C. Thus
far, little has been done for developing
a necessary set of high-temperature buffer
systems. However, aqueous 0.05 mol kg1
potassium hydrogen phthalate solution
has been shown as an appropriate buffer
system to be used at temperatures up to
about 225 C [21].
10.3
High-temperature Potentiometry
During the last 30 years, Oak Ridge National Laboratory (ORNL) and others have
used the hydrogen-electrode concentration cells (HECC) [14] for potentiometric
measurements of homogeneous [22] and
heterogeneous [23] protolytic aqueous systems to 300 C over a wide range of
ionic strength. Many acidbase, metal
ion hydrolysis, metal complexation, metaloxide solubility, and surface adsorption
reactions have been studied using this
technique [22, 23]. The most recent design of the stirred HECC is described in
Ref. 14.
While the electrochemical experimental techniques used at temperatures below
300 C are out of the scope of this chapter, a brief description of the HECC will
be given below, owing to the signicance
(25)
737
738
Test solution
inlet (PEEK tube)
Reference solution
inlet (PEEK tube)
A
C
High temperature
zone
Thermocouple
Outlet
(PEEK tube)
Fig. 4
(27)
10.4
High-temperature pH Measurements
(29)
(30)
739
740
E(X) E(S)
(31)
(RT/F) ln 10
(32)
E(X) E(S)
2.303RT/F
Ed (X) Ed (S)
1
aH2 O (S)
+
lg
2
aH2 O (X)
2.303RT/F
(33)
where S denotes the standard solution,
E(X) and E(S) are, respectively, the opencircuit potential for solutions X and S,
aH2 O (X) and aH2 O (S) are, respectively,
activities of water of solutions X and S,
and Ed (X) and Ed (S) are, respectively, the
diffusion potentials of solutions X and
S with respect to the reference electrode
potential. The calibration coefcient, ,
should be estimated in a range of pH of
interest using reference systems or buffer
solutions as described in Sect. 10.2.4. The
approach has been tested so far for a
temperature range up to about 300 C [27]
and additional efforts are needed to nd
10.5
10.6
741
742
PEEK tube
CONAX
glands
Stainless steel
fittings
Personal
computer
I/O card
ZRA
Cooling coil
ZrO2 cement
5 mm
EN probes
Platinum
Type 304 SS
1.5 mm
Platinum
Fig. 5
Type 304 SS
743
744
10.7
10.8 Conclusions
10.8
Conclusions
745
746
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
References
1. D. A. Palmer, R. Fernandez-Prini, A. H. Harvey, (Eds.), Aqueous Systems at Elevated Temperatures and Pressures: Physical Chemistry in
21.
22.
10.8 Conclusions
23.
24.
25.
26.
27.
28.
29.
747
749
11
Electrochemistry of Silicon
Etching
X. Gregory Zhang
Teck Cominco Metals Ltd., Mississauga, Canada
11.1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
751
11.2
11.2.1
11.2.2
Dissolution of Silicon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Fluoride Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
In Alkaline Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
753
753
755
11.3
Hydrogen Evolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
756
11.4
758
11.5
11.5.1
11.5.1.1
11.5.1.2
Reaction Mechanisms
A General Model . . . .
Elementary Steps . . . .
Reaction Paths . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
760
762
763
764
11.6
11.6.1
11.6.2
11.6.2.1
11.6.2.2
11.6.2.3
11.6.2.4
11.6.3
11.6.3.1
11.6.3.2
11.6.3.3
11.6.3.4
11.6.4
11.6.4.1
11.6.4.2
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
767
767
772
772
773
774
776
776
777
778
779
780
780
780
781
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
750
11.6.4.3
11.6.4.4
Ethanolamine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Tetramethyl Ammonium Hydroxide (TMAH) . . . . . . . . . . . . . . . .
782
782
11.7
11.7.1
783
784
11.8
11.8.1
11.8.2
11.8.2.1
11.8.3
Anisotropic Etching . . . . . . . . . . . . . . . . . . . . .
Rate-Limiting Process . . . . . . . . . . . . . . . . . . . .
Mechanism of Anisotropic Etching . . . . . . . . . . .
A General Model for both HF and KOH Solutions .
Basic Features of Anisotropically Etched Surfaces .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
785
787
788
789
792
11.9
11.9.1
11.9.2
11.9.3
11.9.4
Surface Roughness . . . . . . . . . . . . . . . . . . . . . . . .
Microroughness . . . . . . . . . . . . . . . . . . . . . . . . . .
Macroroughness . . . . . . . . . . . . . . . . . . . . . . . . . .
Crystallographic Characters and Formation of Hillocks
Origins of Roughness . . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
794
795
795
796
798
11.10
11.10.1
11.10.1.1
11.10.2
11.10.3
11.10.3.1
11.10.3.2
Applications of Etching . . .
Cleaning . . . . . . . . . . . .
RCA Cleaning . . . . . . . . .
Defect Etching . . . . . . . .
Material Removal . . . . . . .
Uniform Material Removal
Selective Material Removal
References . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
799
799
800
800
801
801
803
804
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
751
11.1
Introduction
752
Dissolution rate
Dislocation
Isotropic
etching
(a)
Fig. 1
(b)
Anisotropic
etching
(c)
Defect
etching
Solution
i
(a)
Crystal
(b)
e e
(c)
11.2
Dissolution of Silicon
The dissolution of silicon can be best characterized by i V curves. Neglecting the details associated with silicon substrate such
as doping, the currentpotential relationship of silicon in aqueous solutions can be
considered to be principally determined
Fluoride Solutions
Vp
OCP
753
Current density
[mA cm2 ]
2.5% HF
150
Fig. 4 Currentpotential
curves for dark and illuminated
p- and n-Si electrodes in a
2.5 wt% HF electrolyte (after
Lehmann and Foll [7]).
J4
100
n -Si, illuminated
50
J1
J2
0
2
vb
n -Si, dark
4
6
Potential
[VPt]
10
12
Current density
[mA cm2]
754
10
PS
8 formation
Electropolishing
J1
J2
4
2
0
200
200
400
Potential
[mVsec]
600
In Alkaline Solutions
755
n -Si at low I0
H2
VJ1
[V]
[V]
0.2
n -Si
p -Si
vp
0.1
0.1
Current density
[mA cm2]
Current density
[mA cm2]
756
ip
Dark
0.1
Illuminated
0.2
(a)
2 4 6
Potential
[V]
0
Dark
Illuminated
0.1
2
10
(b)
4
Potential
[V]
iV curves in 2 M KOH in the dark and illumination, (a) p-Si, (b) n-Si (after Chen
et al. [16]).
Fig. 7
In alkaline solutions, the effective dissolution valence at OCP is zero and changes
only very slightly with anodic polarization
before the passivation peak; it is less than
0.4 at Vp [18]. The dissolution reaction below the passivation potential is thus almost
completely chemical in nature [1820].
11.3
Hydrogen Evolution
Dark
20
Bare
p -Si
40
Platinized
p -Si
60
Illuminated
80
1.5
1.0
0.5
Potential
[Vsce]
30
Ni
Ba
Pb
20
Co
Au
Cd
iPhoto
[mA cm2]
10
0.0
Pt
re
Fig. 8 Steady-state
photocurrentvoltage curves for naked
p-type Si and for platinized p-Si.
Aqueous solution buffered to pH 6.6;
illumination is at 632.8 nm,
2.5 mW cm2 . The platinized p-type Si
was prepared by photoelectrochemical
redution from 1 103 M KPtCI in
0.1 M NaCIO/H2 O at 0.3 Vsce until
1.1 102 C cm2 had passed (after
Dominey et al. [21]).
Photocurrent
[A]
40
0.8
0.4
Potential
[VNHE]
0.4
757
758
11.4
% in total H2
Electrochemical
Chemical
100 %
OC
Si
r5
r1
r6
r4
r
r11
r
r9
r7
r2
r3
Space charge
layer
r10
r8
Oxide
Helmholtz layer
r1 and r2 majority and minority carrier transport, r3 transport of holes to the surface
r4 charge transfer across the Helmhotz layer, r5 electron injection
r6 chemical dissolution, r7 oxide formation, r8 ionic transport in oxide,
r9 injection of oxidants, r10 dissolution of oxide,
r11 mass transport in electrolyte
Fig. 11 Schematic illustration of the processes involving the transport of charge and
species in the different phases in the Si/electrolyte interface region (after Zhang [3]).
759
760
KOH
ra
p+ , n+
ra
p+ , n+
Dark
OCP
OCP < V < Vp
V Vp
0
Large
Large
r3
r4
r10 , r11
r3
r2 or r5b
r10 , r11
r3
r4
r10 , r11
Large
Large
Small
r6
r6
r10
r6
r6
r10
r6
r6
r10
Illuminated
OCP
OCP < V < Vp
V Vp
Large
Large
Large
r4
r4
r10 , r11
r3
r3 or r4 or r5c
r10 , r11
r4
r4
r10 , r11
Large
Large
Small
r6
r6
r10
r6
r6
r10
r6
r6
r10
a Oxidation
Reaction Mechanisms
Whether current doubling and quadrupling occur is determined by the competing processes depending on the availability
of holes. The observation of hydrogen termination of silicon surface in HF solutions
in the 1980s led to the conclusion that the
dissolution of silicon starts by adsorption
of hydrogen and then the adsorbed hydrogen atoms are replaced by uoride atoms.
The uoride terminated silicon atoms further react with HF molecules to break
the silicon atoms from the lattice and dissolve into the solution [32, 4042]. The
reaction of silicon atoms with hydrogen
results in hydrogen evolution, which is responsible for the chemical component of
the dissolution reaction and for an effective dissolution valence less than 4. The
relative contribution of chemical reactions
and electrochemical reactions depends on
pH of the uoride solution [41].
The individual steps in the dissolution
process in HF solutions were further considered in a number of studies in the
late-1980s and early-1990s to involve a hole
capturing in the rst step followed by an
electron injection as the second step [30,
42, 43]. The oxidized silicon intermediate,
having a valence of two, is further oxidized
by the injection of two electrons or reacting
with the hydrogen ions depending on light
intensity. More recently, an intermediate
Si(I), which is an electron decient surface
silicon atom and is mobile on the surface,
was proposed to act as a radical to catalyze
the hydrogen reduction reaction and to account for the transition of the dissolution
reaction from valence 4 to valence 2 with
increasing light intensity [44].
In alkaline solutions, an important issue involving the reaction mechanisms has
been whether the etching process at OCP
is of chemical (Fig. 2c) or electrochemical
nature (Fig. 2b), that is, whether chargetransfer processes involve carriers in the
761
762
A General Model
Dissociation of solutes
Concentration of OH
Activity of H2 O
Solubility of SiO2
Rate of SiO2 dissolution
Hydrogen termination at OCP
Dissolution rate of Si at OCP
OCP V < Vp
at V > Vp , rp
Effective dissolution valence
at OCP
OCP < V Vp
low light intensity, n-Si
high light intensity, n-Si
at V > Vp
Chemical reaction
at OCP
OCP < V Vp
at V > Vp
Photocurrent quantum efciency, n-Si
OCP V Vp
low light intensity
high light intensity
V > Vp
Difference in n-Si and p-Si near OCP
Band bending at OCP
Dependence on orientation
at V < Vp
at V > Vp
Dissolution uniformity
at V < Vp
at V > Vp
HF
KOH
K = 713 103
Extremely low
High
High
High
Yes
None
Increase with V, lowest
at OCP
High, rp rocp
Complete
High
Low at high [KOH]
High
Low
Yes
High
Decrease with V,
highest near OCP
Low, rp rocp
No dissolution
24
4
2
3.6
H2 evolution
Low
High, <50%
10%
0
<1
<1
<1
4
H2 evolution
High, 100%
High, >50%
0
4
2
1
Yes
p-Si downward, n-Si
upward
0
0
1
No
p-Si and n-Si downward
Yes
No
Yes
No
Porous silicon
formation
Uniform
Anisotropic
Uniform
Elementary Steps
The dissolution of a surface silicon atom
involves the replacement of the surface
hydrogen atom by an OH in KOH solutions and by F in HF solutions. The
difference between the hydrogen replacement by OH and that by F is whether
holes are involved. As shown in Fig. 12, in
KOH, upon the hydrogen replacement by
OH , the silicon atom becomes a radical,
which in the following reaction steps is
neutralized by the reduction of hydrogen
ions. In HF, the replacement of hydrogen
by F requires a hole, which results in
a neutralized SiF bonding. The valence
state of the adsorbed hydrogen before and
after the replacement, being zero, is not
changed in both solutions. The important
difference is that after being terminated by
an OH ion, the silicon atom is a radical
with two extra electrons, while in SiF
termination the bonds associated with the
silicon atom are neutralized.
The hydrogen adsorption onto a silicon
atom is a reduction process since the
valence of the hydrogen atom is changed
from +1 to 0. It occurs when the backbond
of SiSiF or SiSiOH is broken by
11.5.1.1
H2
OH
Si
Si
OH
Si
Si
Si
OH
+
(I)
Si
Si
Si
OH
Si
H2
H
H
Si
Si
Fig. 12
F
+F
Si
Si
Si
Si
F
Si
Si
Si
(II)
763
764
OH OH
Si
Si
F
HF
OH
Si
Si
Si
F
Si
Si
(III)
H
Si
F
H
OH
Si
F
Si
OH OH
F
Si
Si
e
Si
Si
Si
(IV)
H
Si
Si
Si
F
Si
H
H H
F
+2F + h e
Si
Si
Si
Si
(II)
Si
F H
(1)
+ 2HF
Si
Si
Si
Si
H
Si
H H
(IV)
Si
Si
Si
Si
+ 2H2O
FH
4H
Si
O
Si
H
Si
(2)
O
Si
Fig. 14
OH
OH
+2H+
OH
Si
OH
H2
H
HH
Si
Si
H
Si
Si
+ 2OH
Si
Si
(I)
OH OH H
Si
Si
Si
OH
H
H H
Si
Si
Si
OH
+2H2O
Si
(III)
+2H+
OH
H2
H
Si
Si
Si
H2
+ 2H2O, 2h
OH OH H
Si
OH
OH
(3)
OH
Si
Si
Si
OH OH H H
+2H2O 4h
Si
Si
Si
Si
Si
Si
H
OH
OH
HH
Si
Fig. 15
H
4H+
Si
Si
Si
(4)
(5)
O O
Si
765
766
Path
Conditions
(1)
(1) and (2)
(2)
2
2<n<4
4
HF
OCP < V Vp
OCP V < Vp
V Vp
(3)
(3), (4), (5)
(5)
0
<1
4
KOH
OCP
OCP < V < Vp
V > Vp
11.6
General
767
768
Sia
Solutionc
(100)
(111)
(110)
References
n, 2
p, 1
n, 3
HF solutions
9.6% HF, 15 hours
48% HF, 5 days
1 M NH4 F, pH = 4.2
1 M NH4 F, pH 8.25
rt
rt
rt
rt
4 104
8 103
3 104
0.03
103
51
52
53
54
25
rt
78
rt
rt
rt
rt
0.83
115
0.05
280 000
14 000
70
12 700
55
56
57
58
59
60
61
rt
rt
rt
2.2
4.2
7.2
62
63
64
65
66
830
830
67
52
68
200
69
9.7
70
Poly Si
Poly Si, 1010
p
p
n, 3
n, 3
n, 0.002, 3
923
923
p, n
n, 10100
Poly Si
p, 3
p, 3
n, >8 1018 cm3
n, <1018 cm3
p, >7 1018 cm3
p, <1018 cm3
p, n
0.615
13
13
n, p 0.116
HF + oxidants
0.2% HF + 50% HNO3
2% HF + 50% HNO3
1.3% NH4 F + 5% H2 O2
0.1% HF + 10% H2 O2
33.5% HF + 23.1% HNO3
10% HF + 56% HNO3
1% HF + 3.8
104 M KMnO4
0.01M CrO3 + 2M HF
2M CrO3 + 1M HF
10 M HF + 0.5M HBr
+ 10 mM Br2
3 M NH4 F + 1M Cu(NO3 )2
0.25% HF + 35% HNO3
+ 15% H2 O2
8% HF + 75% HNO3
+ 17% CH3 COOH
1.3% HF + 49% HNO3
+ 49.7% CH3 COOH
5% HF + 24% HNO3
+ 50% CH3 COOH
3% HF + 15% HNO3
+ 70% CH3 COOH
0.42% HF + 17.5% HNO3
+ 67% CH3 COOH
0.42% HF + 17.5% HNO3
+ 67% CH3 COOH
0.42% HF + 17.5% HNO3
+ 67% CH3 COOH
0.42% HF + 17.5% HNO3
+ 67% CH3 COOH
Secco etch
Schimmel etch
Sirtl etch
Wright etch
Yang etch
MEMC etch
Sopori etch
830
70 65
rt
13 300
22 31
35 0.3
rt
0.3
20
rt
18
rt
480b
71
rt
27b
71
rt
330b
71
rt
33b
71
rt
rt
rt
rt
rt
rt
rt
250
292
385
100
250
833
3500
72
73
74
74
75
76
77
(continued)
Sia
n, 1
Poly Si
n, p
n, p
n, p
n
p
n, p
p, 030
n, 13
p
p, 5
n, 13
p, 10
p, 10
n, 0.05
Deposited Si
Deposited Si
p, n, 0.220
0.001100
p, 1075
Solutionc
Inorganic alkaline solutions
20% KOH (4.2 M)
20% KOH
34% KOH
23.4% KOH + 13.3% IPA
23.4% KOH + 13.3% IPA
0.5 M KOH
0.5 M KOH, stirred
0.1 M KOH, stirred
2 M KOH
2 M KOH
2 M KOH
35% KOH, stirred
44% KOH
2 M LiOH
2 M NaOH
25% RbOH
1.9 M NH3
10% NH4 OH
3.7 wt% NH4 OH, stirred
3 103 M H2 O2
+ 1.9 M NH3
0.5 M NH4 OH
+ 0.09 M H2 O2
5 M NH4 OH + 5 M H2 O2
2.65 M NH3 + 0.01M H2 O2
NH4 OH + H2 O (1:5),
pH = 12.4
SC1
1:1 water:hydrazine
48% hydrazine + 20% IPA
+ 32% H2 O
Organic alkaline solutions
EDP, 8 ml H2 O
+ 17 ml NH2 (CH2 )2 NH2
+ 3 g C6 H4 (OH)2
B etch
F etch
S etch
Etch 1
25% TMAH,
(100)
(111)
(110)
References
30
100
70
80
80
23
23
23
rt
rt
70
80
80
25
70
50
70
70
75
70
6
560
100
130
38
0.5
6
3.5
3.5
3.5
50
390
1750
2
21
22
10.7
67
73
0.06
19
1.5
8
0.6
2
5.8
2.8
9
930
216
55
740
3500
7.5
14
33
22
45
45
78
79
79
49
49
49
80
80
49
81
82
49
49
83
84
85
86
84
80
0.28
87
80
80
70
0.12
333
0.23b
87
88
89
70
110
Boiling
0.14b
360
50
57
216
89
90
91
110
139
8.3
83
92
115
115
115
117
70
69
225
170
100
45.3
5.2
4.2
53
1.5
190
111
89.9
93
93
45
94
54
(continued overleaf )
769
770
(continued)
Sia
Solutionc
(100)
(111)
(110)
n, 40
n, p, 15 cm
5% TMAH
29% TMAH + 8.5% IPA
90
80
225
61
5.4
3.9
308
References
95
96
a cm3
or cm .
not specied.
c HF solutions are generally in volume concentration while OH solutions are in weight concentration
as the solvent water is not specied for its concentration.
Etch 1:100 g gallic acid + 317 ml ethanolamine + 144 ml water + 0.3 ml of 10% FC129
(3 M) surfactant.
SC1: NH4 OH:H2 O:H2 O2 mixtures.
IPA: iso-2,2-propyl alcohol.
EDP or EPW: NH2 (CH2 )2 NH2 + C6 H4 (OH)2 + H2 O.
B etch: 7.5 ml NH2 (CH2 )2 NH2 + 1.2 g C6 H4 (OH)2 + 2.4 ml H2 O [93].
F etch: 7.5 ml NH2 (CH2 )2 NH2 + 2.4 g C6 H4 (OH)2 + 2.4 ml H2 O [93].
S etch: 7.5 ml NH2 (CH2 )2 NH2 + 1.2 g C6 H4 (OH)2 + 1 ml H2 O + 0.045 g pyrazine [93].
Sirtl etch: 1 48% HF: 1 5 M CrO3 [73].
Secco etch: 2(49% HF):1(0.15 M K2 Crs O7 ) at room temperature with ultrasonic agitation [72].
Sopori etch: 36:20:2 [HF (49%):CH3 COOH (glacial):HNO3 (70%)].
Wright etch: 60 ml 49% HF, 30 ml HNO3 (69%), 30 ml 5 M CrO3 , 2 g Cu(NO3 )2 .3H2 O, 60 ml acetic
acid (glacial), and 60 ml H2 O [74].
Yang etch: 0.75 M CrO3 + 25% HF.
Schimmel etch: 0.33 M Cr6+ + 33% HF without ultrasonic agitation.
TMAH: tetramethyl ammonium hydroxide.
MEMC etch: 35 : 25 : 18 21(49% HF:70% HNO3 :HAc:H2 O) + 1g(NO3 )2 .3H2 O per 100 ml of mixed
solution [76].
b Orientation
Si
p(111), 3
n(111), 0.002
n(111), 5
p, n
n(100)
n(111), 30
p(111), 10
n(100), 550
(100)
(110)
(211)
n(100), 1
n(110), 1
n(111), 1
(100)
(100)
(110)
(111)
(100)
(110)
(111)
n(111), 1
(100)
(100)
(110)
(111)
(100), 1015 cm3
(100), 2 1019 cm3
(110)
(100)
(111)
a See
Solutionc
Acidic solutions
5% HF + 24% HNO3
+ 50% CH3 COOH
0.42% HF + 17.5% HNO3
+ 67% CH3 COOH
0.42% HF + 17.5% HNO3
+ 67% CH3 COOH
5% HF + 63% HNO3
7 M NH4 F + 1 M Cu(NO3 )2
69.5 ml
HNO3 + 8 ml HF + 22 ml H2 O
+ 80 ml glacial acetic acid
[CrO3 ]/[HF] of 0.030.72 with 50 g
CrO3 in 100 ml H2 O and 48% HF
91.5% H3 PO4
Inorganic alkaline solutions
2545% KOH
42% KOH
42% KOH
6 M KOH, 5090 C
23.4% KOH + 13.3% IPA
23.4% KOH + 13.3% IPA
23.4% KOH + 13.3% IPA
25% RbOH, 5080 C
3.7 wt% NH4 OH
3.7 wt% NH4 OH
3.7 wt% NH4 OH
1:1 water:hydrazine
1:1 water:hydrazine
1:1 water:hydrazine
Organic alkaline solutions
EDP, 8 ml H2 O +
17 ml NH2 (CH2 )2 NH2
+ 3 g C6 H4 (OH)2
B etch
EDP
EDP
EDP
25% TMAH
25% TMAH
22% TMAH, 6090
22% TMAH, 6090
22% TMAH, 6090
Ea [eV]
References
0.42
69
0.23
71
0.54
71
0.17
0.52
0.21
97
65
98
0.340
99
1.15
100
0.56
0.59
0.60
0.641
0.77
0.76
0.59
0.47
0.22
0.19
0.19
0.28
0.41
0.25
101
45
45
102
79
79
79
83
86
86
86
90
90
90
0.48
92
0.47
0.40
0.33
0.52
0.52
0.64
0.39
0.61
0.40
92
45
45
45
103
103
95
95
95
771
Etch rate
[ s1]
772
In Fluoride-based Solutions
Absence of Oxidants
The etch rate of silicon in HF solutions at
room temperature at OCP is very low, in
the order of 103 A s1 in concentrated solutions (>25% HF) [52]. Figure 17 shows
that dilution of concentrated HF results in
an increase of etch rate [51, 110]. It also
shows that the etch rate increases sharply
from pH about 4, peaks at 6.2, and then
decreases drastically to near zero at about
8.5 [53]. The etch peak has been attributed
to the involvement of HF2 as the etching
species, which shows a concentration peak
at the pH about 6 [110]. In HFNH4 F
solutions, the etch rate decreases with
increasing HF at a constant NH4 F concentration and it increases with increasing
NH4 F at constant HF concentration [111].
Etch rate depends on the time of immersion. For example, the etch rate of p-Si
in concentrated HF decreases with immersion time from about 0.5 A s1 at 1 minute
to about 2 104 A s1 at 10 days [52].
The relatively higher rate at the beginning
is attributed to the etching of native oxide
11.6.2.1
thermal
anodic
quartz
CVD
desified CVD
Si
10
1
0.1
0.01
0.1
1
HF concentration
[wt%]
10
Etch rate
[ s1]
Diluted HF
1 M NH4F
0.06
0.04
0.02
0.00
0
pH
773
10 M HF
100
Etch rate
[ s1]
774
2 M CrO3
0.05 M
CrO3
10
10 M HF
2 M HF
2 M HF
1
0.1
10
100
1000
[HF]/[CrO3]
(1)
(2)
90
HC2H3O2 diluent
H2O diluent
20
We
80
10
igh
t%
30
70
%)
F(
,H
t%
50
.51
(330)
60
40
We
igh
(69
50
(940)
NO
49
.2%
,H
40
60
(152)
(112)
(87)
30
20
70
(76.3)
80
(51)
90
(33)
(11.5)
10
[m min1]
90
80
70
60
50
40
Weight %,diluent
30
20
10
775
776
11.6.3
Alkaline Solutions
The most commonly used alkaline solutions have been KOH and EDP (or EPW)
Etch rate
[ s1]
8
6
4
2
23 C
Si (100)
Unstirred
Fig. 20
5
10
KOH concentration
[M]
15
777
778
600
p (100)
n (100)
OCP(n)
OCP(p)
pp(n)
pp(p)
400
200
(a) 0
100
n-Accumulation
p-Depletion
p (111)
n (111)
n-Depletion
80
60
OCP(n,p)
40
pp(n)
pp(p)
20
(b)
2.0
1.5
1.0
0.5
(110) wafers [130, 131]. At low KOH concentrations, below 30%, there is a tendency
to form pyramidal islands on (100) surface.
The density of hillocks varies with temperature, potential, and the presence of
chemical agents in the solution [34, 47].
Reaction Kinetics
The etching of silicon in KOH is a process
of dissolution and hydrogen evolution.
Two hydrogen molecules are evolved for
each silicon atom dissolved during the
etching [47]. The main etching product
is determined to be SiO2 (OH)2 2 which
can further be polymerized [48, 132]. OH
is an active etching species since OH
concentration decreases during etching.
Water is also an active species because
silicon does not etch in water-free alcohol
that contains KOH [47]. K+ as a cation
present in the solution is not directly
involved in the etching reaction because
the etching behavior of silicon is essentially
similar in different alkali solutions [33, 49].
In IPA buffered solutions, the IPA does
not alter the composition of the solutions
or participate in the etching reactions [48,
132]. The activation energy of the etching
process is 0.5662 eV for KOH of different
concentrations indicating that the etching
is controlled by a surface process [45].
The dependence of etch rate on KOH
concentration in the concentration range
of 10 to 60% has been found to best t with
11.6.3.2
an equation [45]
1
(4)
(5)
1.9 M NH3
70 C
Etch rate
[ s1]
80
60
40
20
0
0
1
2
3
Concentration of H2O2
[M]
779
780
Organic Solutions
EDP Solutions
EDP or EPW is a mixture of ethylenediamine (ED or E), pyrocatechol (P), and
water (W). This solution usually operates
at 110120 C (about boiling point) [93].
The typical compositions and etch rates
can be seen in Table 5. Figure 23 shows
the etch rate as a function of the water
content [92]. No etching occurs in solutions without the presence of water and
the maximum etch rate is observed with a
water molar fraction of about 0.6. The etch
rate increases with increasing pyrocatechol
content to about 5 mol% above which etch
rate becomes constant. There is a denite
etch rate without addition of pyrocatechol,
11.6.4.1
60
Etch rate
[ s1]
40
20
0.03 M pyrocatechol
100 C
0
0.0
suggesting that it is not an essential component of the etching system but rather
a catalyst. Addition of other agents such
as pyrazine and p-benzoquinone has a
similar effect [93].
EDP etch solutions can be affected by
many operation factors such as bath design, contamination level, silicon content
in the solution, and oxygen pressure [93].
Diffusion plays an important role in
the etching of silicon in EDP [45]. Trace
amount of chemical agents such as diazine, pyrazine, and quinone changes the
etch rate and the difference in the etch
rates among the crystal planes [46]. The
etch rate changes signicantly when exposed to air due to the effect of dissolved
oxygen. Also, insoluble residues and rough
raindrop-like patterns tend to form at
low temperatures and with aged solutions [93, 112]. Residue formation occurs
0.2
0.4
0.6
0.8
Mole fraction of water
1.0
500
Etch rate
[ s-1]
400
300
200
OCP
100
0
p -Si
n -Si
2.5
2.0
1.5
1.0
Potential
[Vsce]
0.5
0.0
781
782
(8)
Ethanolamine
Ethanolamine (NH2 CH2 CH2 OH) based
anisotropic etchants are developed aiming to replace the toxic EDP etchant [124,
140]. Linde and Austin [124] reported
a functional formula which consists of
100 g gallic acid + 305 ml ethanolamine,
NH2 CH2 CH2 OH, +140 water + 1.3 g
pyrazine + 0.26 ml FC129 (3 M) surfactant. The typical etch rate of (100) silicon in
this etchant at 118 C is about 300 A s1 .
It has high boron selectivity and good uniformity. Also, the etch rate ratios of silicon
to other materials such as carbide, nitride,
oxides, and most metals in ethanolamine
etchants are very large [108].
Gallic acid functions as an effective
silicon complexing agent in aqueous
ethanolamine solutions. In the absence of
it the etch rate is zero. Etch rate increases
with water content. The additions of
pyrazines, pyridazines, and triazoles show
various catalytic effects on the etching process [140]. The catalysts that lead to faster
oxidation result in faster etch rates, and the
difference among the catalysts is attributed
to a steric effect. Oxidative catalysts tend
to inuence the etching selectivity of the
major crystal orientations [94].
Tetramethyl Ammonium
Hydroxide (TMAH)
TMAH, (CH3 )4 NOH, etchant has been
developed more recently for anisotropic
etching of silicon [59, 109, 141]. It is relatively safe to use and presents no special
(6)
Si + 2OH + 4H2 O
Si(OH)6 = + 2H2 Oxidation
(7)
Si(OH)6 = + 3C6 H4 (OH)2
[Si(C6 H4 )3 ]=
+ 6H2 O Chelation
11.6.4.3
11.6.4.4
11.7
Fig. 25
0.1
0.01
B, EDP, 110 C
P, KOH, 21 C
Ge, EDP, 79 C
C, KOH
0.001
1017
1018
1019
1020
1021
Dopant concentration
[cm3]
1022
783
784
Mechanism
decrease of etch rate with boron concentration exhibits an inverse fourth power dependence on the boron concentration [33,
46, 147]. The etch rate reduction at high
boron doping levels is attributed by Raley
et al. [46] to the reduced electron concentration required for the reduction of
hydrogen in the etching process. The rate
of reduction reaction is a fourth order
reaction with respect to electron concentration n
(9)
R n4
Seidel et al. [148] proposed a similar
model, ascribing the reduced etch rate
of highly doped material to the reduced
electron concentration caused by the degeneracy induced by the heavy doping [33].
According to this model, the etch rate decreases at a concentration corresponding
to a doping level of 2.2 1019 at which the
material is degenerated and the Fermi level
moves into the valence band, making silicon behave like a metal. The fourth power
decrease of etch rate is explained by the
fact that for the dissolution of one silicon
atom, four charges are required to transfer
across the solid/electrolyte interface.
The shortened electron life model could
not explain the etch stop on n-Si, in which
there are abundant electrons in the conduction band. Also, this model assumes
electrochemical reactions responsible for
the etching process in alkaline solutions,
in which they are in fact largely in chemical
nature.
On the other hand, the passivation
model, proposed by Palik et al. [80, 149],
attributed the etch rate reduction to the
easier formation of an oxide lm on highly
doped silicon. It is supported by a number of experimental observations. First, the
difference between passivation potential
and OCP decreases with increasing doping
concentration, implying easier passivation
Fig. 26
11.8
Anisotropic Etching
Anisotropic etching, that is, different dissolution rates on different crystal planes,
is a characteristic feature of silicon etching in alkaline solutions. Strictly speaking,
the etch rate of silicon always depends,
to a various extent, on crystal orientation
in all etching solutions, acidic or alkaline.
However, the etch rate difference among
different planes is small in acidic HF
solutions compared to those in alkaline
solutions. Figure 26 shows the etch rate
ratios of (100)/(111) and (110)/(111) planes
in various solutions.
In HF-based solutions the ratios are
close to 1, varying only slightly with different solution compositions. An exception
is in NH4 F solution with the addition of
copper ions in which the etch rate ratio of (100)/(111) is found to be as high
as 15 [65]. Addition of silver or gold ions
does not have such an effect. The special effect of copper ions on anisotropicity
may indicate that the copper deposition
100
(100)/(111)
(110)/(111)
10
1
A B C D E F G H I J K M
Etchant
785
786
Tab. 6 Etch rates of different crystallographic planes in 25% TMAH and 34% KOH at 70 C, in
A s 1 [54, 78]
Orientation
(100)
(110)
(210)
(211)
(221)
(310)
(311)
(320)
(111)
TMAH
KOH
Etch rate
Ratio of X/(111)
45.3
88.7
99.5
87.8
79.5
98.2
96.0
95.7
1.5
30.2
59.1
66.3
58.5
53
65.5
64
63.8
1.0
Etch rate
100
216
212
151
94
181
179
220
1.5
Ratio of X/(111)
67
144
141
101
63
121
119
147
1.0
11.8.1
Rate-Limiting Process
12
10
EDP
69 C
8
6
20% TMAH
86 C
4
2
0
3
32% KOH
44 C
2
1
0
1
2
Angle relative to (111) plane
787
788
etching, when both are controlled by surface processes, is that the former is not
sensitive to variations in lattice structure
but the latter is.
11.8.2
the formation of oxide is an electrochemical one [18, 19, 158]. In addition, the etch
rate of (111) surface in KOH, although
much smaller than those of (100) and (110)
planes, have denite values, in the range
of 210 A s1 , which is much larger than
the dissolution rate of passive oxide lm
in KOH in the order of 0.01 A s1 [126].
The basic consideration in reaction kinetics models is that the reaction rate is
determined by the lattice structure on the
surface. The difference in the lattice structures of various crystal planes gives rise
to differences in surface bond density,
electron density, surface free energy, and
so on, which then determine the dissolution rate of the surface silicon atoms. The
rate of this reaction is determined by the
bonding condition of the surface silicon
atoms. The difference in etch rate between
(111) and (100) surfaces was considered
to relate to the bond densities on the two
surfaces in the early models [18]. It was
soon recognized that this effect alone will
not cause an etch rate difference of more
than a factor of 2. Some models attributed
the anisotropic etching to the different energy levels of surface states on (111) and
(100) surfaces arising from binding surface atoms to OH [45]. The difference
in the activation energy of various crystal
planes was related to the difference in the
energy states of the backbonds of differently oriented surfaces. Such a difference
in energy levels has also been attributed
to the difference in work function between
(111) and (100) planes [139].
A number of studies attributed lattice
defects to be the cause of anisotropic
etching. On the basis of the nding by
STM imaging, Allongue et al. [19, 157]
proposed that the anisotropic etching
occurs preferentially at the kink-site atoms
at the edges of terraces. Jakob and
Chabal [158] reasoned that the silicon
kinetic processes that actualize such physical cause, and (3) the surface condition
that determines the removal rate of the
surface atoms. Also, the model has to be
consistent with the data from the alkaline
etching and HF etching.
The physical cause in the rst aspect can
be considered to be, most convincingly,
associated with the backbond strength
theory [18, 157, 158]. Because of the adsorption of OH , which has a large
electronegativity, the backbond is polarized and weakened. The {111} plane has
only one SiOH surface bond, whereas
{100} surface has two bonds as characteristically represented by the scheme shown
in Fig. 28. The atoms on the (100) surface
thus have weaker backbonds compared to
those on the (111) surface due to the larger
polarization. This is considered the fundamental cause for the higher dissolution
rate of (100) surface than that of (111)
surface.
Regarding the second aspect, generally,
as shown in Fig. 29, three species may be
involved in the etching reaction: charge
carriers, that is, electrons and holes, at the
semiconductor surface; chemical species
such as OH , NO3 H2 O, and so on
near the surface in the solution; and active
surface silicon atoms which are favorable
sites for reaction and removal. Unlike the
other two species, charge carriers may or
may not be involved depending on whether
the reaction is of an electrochemical
nature. The concentrations of each of
these species are determined by different
processes such as diffusion, migration,
H
O
H
O
Si
Schematic illustration
of the bonding condition of
(100) and (111) surfaces.
H
O
H
O
Si
H
O
Si
H
O
H
O
H
O
H
O
Si
Si
Si
Fig. 28
(100)
(111)
789
790
Silicon
Charge carriers
h
h
A
A
A
A
small and etching on (111) surface proceeds only at lattice inhomogeneities such
as steps, kinks, and vacancies. Thus, a perfect (111) surface will have an extremely
low etch rate and the etch rate of real (111)
surfaces is determined by the etch rate at
the steps and the density of the steps.
The key points of this third aspect can be
summarized as (1) the etch rate of a perfect
(111) surface is negligibly small, (2) the
etching of real (111) surfaces is due to
the etching along the edges of (111) steps,
(3) the amount of the (111) steps depends
on the degree of misorientation from the
perfect (111) plane, (4) all crystallographic
planes can be viewed as composed of the
(111) terraces and steps at a microscopic
level. The etch rate of a surface with an
angle from (111) surface can then be
correlated with that at (111) steps as shown
in Fig. 30. The vertical etch rate, Vv , which
is the planar etch rate normally measured
in experiments, is
Vv = Vs a/b
(10)
(11)
(111) terrace
Terrace edge
Vv
Vs
L
Fig. 30 Relationship between vertical etch rate, Vv , step etch rate, Vs , and the angle
from the (111) plane, .
Rv = Rs an/ l
(12)
(13)
(14)
791
792
(a)
(b)
(c)
Mask
<110>
t Si
(1)
(a)
(2)
(3)
<110>
(b)
R(111)
WSi
Wox
R(100)
q
W0
q = 54.74
U = R (111) t/sin q
W0 = Wsi 2 tSi
793
Corner undercut
[m]
794
(a)
40
35
30
25
20
15
10
5
0
(111) step
Corner undercut
(111) planes
Apparent
etching plane
KOH, propanol,
H2O, at 80 C
0
10 20 30 40 50 60
Depth
[ m]
(b)
11.9
Surface Roughness
Microroughness
Macroroughness
795
796
(100), Ra = 0.011 m
(111), Ra = 0.052 m
1m
(110), Ra = 0.351 m
0
200
400
600
Lateral dimension
[m]
797
798
Origins of Roughness
(a)
Metal
deposit
(b)
Lattice
defect
(d)
Fig. 35
(c)
Terraces
and steps
(e)
Pore
(f)
11.10
Applications of Etching
Cleaning
A clean surface is essential for device reliability and performance [183]. It becomes
critical as the dimensions of devices become smaller and smaller as a result
of ever-increasing integration and complexity. It has been estimated that over
50% of yield losses in integrated circuit
(IC) fabrication are due to microcontamination [184]. Today, a typical process
ow for advanced ICs consists of 300
to 500 steps, 30% of which are wafer
cleaning steps [185]. The need for wafer
cleaning can be separated into three areas (1) preparation of the wafer surfaces
for oxidation, diffusion, deposition, and
metallization, (2) preparation for the application of photoresist, and (3) removal
of photoresist after the etching process [186, 187].
A cleaned silicon wafer is (1) particle
free, (2) without organic contaminants,
(3) with minimal metallic contamination,
(4) free of native oxide, and (5) with minimal surface roughness [24, 58]. A clean
surface has a surface concentration of
contaminants such as Fe, Cu, and Ni of
less than 1010 cm2 and microroughness
of about 0.2 nm [136]. As-received wafer
typically has several thousand particles, a
roughness of about 0.2 nm and metal concentrations of about 5 1010 1012 cm2
for Fe, Cu, and Zn [135, 188, 189]. The degree of cleanness is constantly improved as
cleaning technology continues to progress.
Contaminants on silicon surfaces can
be classied as molecular, ionic, and
atomic or as hydrocarbons, metals, and
particles [57, 190]. Typical molecular contaminants are waxes, resins, oils, and
organic compounds that are commonly
generated in the processes after sawing, or
from human skin and plastic containers.
They are usually attached to the surface
by weak electrostatic forces. Ionic contaminants such as Na+ , Cl , F , and I are
present after etching in HF or alkaline
etchants. They may be precipitated on the
surface by physical adsorption or chemical adsorption. The atomic contaminants
of concern are metals such as gold, iron,
copper, and nickel originating from acid
etchants. The metallic impurities can only
be effectively removed by wet cleaning processes [58, 136]. The ceramic particles such
as SiC, Si3 N4 , Al2 O3 , SiO2 , and C, originate to a large extent from wet processes
and cleaning baths. These particles are attracted by the electrical eld generated by
the wafers [190]. The contaminants may
physically locate in different phases at the
799
800
metallic contaminants that are not entirely removed in SC1 solution, such as
gold [57]. A preliminary cleaning step with
a hot H2 SO4 H2 O2 mixture (2:1) can be
used for grossly contaminated wafers having visible residues. The SC1 solutions are
continually modied for better results, dilution [55], ultrapure water [190], ozonized
water [58] and other additions [44, 136,
184]. More recently, this cleaning solution is often used with a concentration
only one-tenth of the original solution to
avoid surface roughness and to reduce
production cost and its effect on the environment [191].
Cleaning of metallic impurities on a silicon surface requires the oxidation and
dissolution of the metal atoms. Most metals have a standard potential that is much
higher than that of silicon. To oxidize the
metal atoms, the electrode potential, at
least in the local area of the deposits, has
to be raised to the level of the reversible
potential of the metal. In the SC1 cleaning
solutions the potential of silicon is 0.4 Vsce
at which the silicon surface is passivated
(It is 1.10 Vsce without H2 O2 ) [84]. This
means in the SC1 solution the species
with reversible potential less than 0.4 Vsce
will be oxidized. Passivation, which is responsible for the slow etch rate in the
SC1 solution [55], serves three purposes:
(1) it allows the surface to be anodically
polarized, which is required for oxidizing
of metal impurities, (2) it prevents much
etching of the substrate during the cleaning process, and (3) it helps to physically
dislodge the metal atoms from the surface
by forming a oxide layer between the metal
atoms and the silicon surface.
11.10.2
Defect Etching
Material Removal
801
802
(100)
Metal
(a)
(b)
Light
Current
V pd
Etch rate
p
V pi
V ni
p, n
p+
pe
(c)
p+
p, n
V nd
n
(d)
B doping
(e)
Current
Laser
Current
p,
Illuminated
n
Potential
n+
Metal V
Vapp
Unilluminated
(f)
p,n +
Potential
(g)
Current
Laser
Current
Unilluminated
Illuminated
(h)
Vn
Potential
Vp
Metal
Potential
(i)
Fig. 36 Schematic illustration of the different etching techniques for material removal :
(a) planar etching at OCP, (b) anisotropic etching of masked surface, (c) planar etching
under an anodic bias, (d) lateral selective etching of the p region of an illuminated pn
junction, (e) vertical selective etching using heavily doped material for an etch stop,
(f) preferential etching of illuminated area on an n-type material, (g) vertical selective
etching of p or n+ material under an anodic potential in the dark using the low etch rate of
n-Si as an etch stop, (h) preferential etching of the nonilluminated area on an illuminated
p-type material, and (i) vertical selective etching of p-Si of anodically biased pn junction
using the passivity of n-Si as an etch stop.
applying an external current, anodic potential can also be applied through galvanic
action by depositing a metal on a part of
the silicon surface [213].
Etching of the silicon wafer in mixtures
of HF and HNO3 is routinely used to
remove surface work damage and stress
from the slicing operation as well as to
provide a relatively nonporous and clean
803
804
805
806
807
808
809
12
Electrochemical Machining
Alexei D. Davydov
Frumkin Institute of Electrochemistry, Russian Academy of Sciences,
Moscow, Russia
Vladimir M. Volgin
Tula State University, Tula, Russia
12.1
12.1.1
12.1.2
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Denition and General Information . . . . . . . . . . . . . . . . . . . . . .
Advantages and Disadvantages . . . . . . . . . . . . . . . . . . . . . . . . . .
811
811
812
12.2
12.2.1
12.2.2
Electrode Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Anodic Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Cathodic Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
813
813
819
12.3
820
12.4
12.4.1
821
12.4.2
Accuracy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Methods of Lowering the Effects of Void Fraction and Heating on the
Distribution of Current Density Over the WP Surface . . . . . . . . . .
Methods Enhancing the Localization of Metal Dissolution . . . . . . .
821
822
12.5
12.5.1
12.5.1.1
12.5.1.2
12.5.2
12.5.2.1
12.5.2.1.1
12.5.2.1.2
12.5.2.2
12.5.2.3
12.5.3
823
826
826
827
828
829
829
831
831
832
833
810
12 Electrochemical Machining
12.5.3.1
12.5.3.2
12.5.3.3
12.5.4
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
835
837
837
838
12.6
12.6.1
12.6.2
12.6.3
12.6.4
12.6.5
12.6.6
12.6.7
12.6.8
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
839
839
841
845
845
846
847
847
850
12.7
850
12.8
851
12.9
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
852
852
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
811
12.1
Introduction
12.1.1
812
12 Electrochemical Machining
Fig. 1 Schematic of ECM
installation.
Feed
unit
TE
WP
Electrolyte
supply
unit
Control
unit
Power
supply
12.1.2
The absence of any physical contact between the TE and the WP is the reason
for both the advantages and disadvantages
of the method. The ECM productivity,
as well as the productivity of any other
electrochemical process, is governed by
Faradays and Maxwells laws and is independent of the mechanical properties of
the WP material, in particular, hardness.
Therefore, ECM is feasible for the machining of difcult-to-cut materials (hardened
steels, superalloys, magnetic alloys, Ti, W,
WC, and so forth). The absence of cutting forces virtually eliminates the tool
wear and allows one to machine fragile materials and thin-walled parts. The
TE may be of various designs that enable machining in the difcult-to-reach
areas, inside a complex-shaped part, for
example. After ECM, the product is free
from burrs. Three-dimensional products
can be machined in a single step. With
the application of ECM, it is possible
to generate more freedom of design for
the product. In contrast to other methods of machining, in ECM, as a rule,
decreasing roughness of the surface is
(a)
3
5
5
(b)
12.2
Electrode Processes
(c)
12.2.1
Anodic Processes
813
12 Electrochemical Machining
EocE #
Anodic potential Ea
(a)
Fig. 3
i#
Eoc
Eox E #
Current density i
Current density i
e
Current density i
814
Anodic potential Ea
(b)
E#
Anodic potential Ea
(c)
rate in the gap and the solution concentration at the inlet to the gap; therefore, an
increase of the ow rate and the optimization of concentration are the principal ways
of raising the limiting current. A review,
Ref. 22, is devoted to an analysis of the
problem of the limiting currents of metal
anodic dissolution. The potential range of
the limiting current cd (Fig. 3a) depends
on the properties of the anodic lm and
the parameters of the process (the ow
rate, the concentration, and the electrolyte
temperature). When a certain value of the
anodic potential is exceeded, the current
may increase again: the overlimiting current range de is reached.
The case in which the electrolyte anions
have a relatively weak activating action,
for example NO3 on nickel, is important
for ECM. A special feature of this case
is that E # is signicantly higher than
in NaCl and, most importantly, higher
than the potential at the onset of anodic
oxygen evolution Eox (Fig. 3b). The use
of weakly activating electrolytes enables
one to enhance the localization of metal
dissolution on the areas with the smallest
IEG and the highest current density i
(i > i # , Fig. 3b) and to reduce the metal
dissolution on the WP areas adjacent to
the area to be machined (overcutting).
On the adjacent areas, the current ows,
but it is lower than i # (the gap is
larger) and, hence, is consumed by the
815
816
12 Electrochemical Machining
Fig. 4 Schemes of ECM explaining
different degrees of anodic dissolution
localization: (a) the ECM scheme,
(b) the distributions of the current
density, i, and the dissolution rate, Vd ,
Machining area
over the machining area and the
Stray current area adjacent area (stray-current area) under
various conditions of ECM, (c,d) shapes
x
of cavities produced by the ECM with
the same tube TE under various
conditions of ECM. (1) The distributions
of current density i during the ECM in
NaCl and NaNO3 solutions and the
dissolution rate Vd in the NaCl solution.
1
(2) The distributions of the dissolution
2
rate, Vd , in ECM with a constant voltage
in the NaNO3 solution. (3) The
x
distribution of dissolution rate during
pulse ECM in NaNO3 solution.
Tool electrode
Workpiece
(a)
i
i2#
Vd
#
i1
3
(b)
(c)
(d)
(e)
Fig. 5
Ea
0.5
1.0
Cr(VI)
Ea
[V]
1.5
0
10
(a)
20
30
nCr, mass %
h, %; PCr(VI),%
40
50 100
h
Cr(VI)
100
1.5
Ea
Ea
[V]
50
2.0
0
(b)
The regularities of high-rate anodic dissolution of many materials were considered in Ref. 9.
As an example, the experimental results for the anodic dissolution of alloys
of the nickelchromium system are presented in Figs 5 and 6. This composition
is the basis of several superalloys, which
are machined using the ECM. Alloy components chromium and nickel exhibit
different tendencies to passivate. Nickel is
weakly passivated in the NaCl solution; its
tendency to passivate is much stronger
10
20
30
40
50 100
nCr, mass %
817
818
12 Electrochemical Machining
mF ni zi
Q
Ai
(1)
P, [g A1 min1 cm2]
0.4
0.3
2
0.2
0.1
20
40
60
nCr, mass %
80
100
12.2.2
Cathodic Processes
(3)
An important consequence of this cathodic reaction is the fact that the IEG becomes lled with a two-phase, gasliquid
mixture. This, along with the electrolyte
heating with the owing current, leads to
the variation of the electrolyte conductivity
in the gap and, correspondingly, of the current density at the anode. The latter makes
the production of a part with prescribed
sizes by the ECM method difcult. In the
case of too large a void fraction (it depends
on the current density, the electrolyte ow
rate, the pressure, and so forth), local electrical breakdown within the gap (sparking)
may occur, which can degrade the surface
nish and damage the TE surface [28].
In the nitrate electrolyte, the reduction of
nitrate ions takes place on the cathode. The
complete reduction of nitrate in the neutral
819
820
12 Electrochemical Machining
(4)
12.3
(6)
(U Ea + Ec )
S
(7)
12.4 Accuracy
the machining of large, sculptured surface areas, the problem of the proper
arrangement of inlet holes (normally in
the TE) to inject the electrolyte into the interelectrode space is of prime importance
[3, 10]. The problems of optimization of
voltage, electrolyte ow rate, and the TE
feed rate are also of importance. Theoretical and experimental studies, and ECM
practice, show that the process should be
conducted at the minimum required U
and with the highest possible feed rate.
This allows one to maintain a very small
IEG, to obtain high productivity, machining accuracy, and good surface nish.
However, there are some impediments to
the enhancement of feed rate, primarily
electrolyte overheating, leading to boiling.
Very intense heating and hydrogen generation in the gap can cause the onset of
unwanted electrical discharges, leading to
poor surface nish on the WP and the
electroerosive wear of the TE. Calculation
of the temperature is complicated by the
fact that the temperature varies along the
ow path as well as across the gap. At
present, a 2D model of the temperature
distribution has been developed, on the
basis of which computer simulations have
been generated in order to determine the
critical conditions leading to discharges
[29]. Using this model, the relationship
has been determined between the critical
local temperature in the gap and the average temperature at the outlet, which is
easily measured and used as a signal for a
control system for ECM.
12.4
Accuracy
Among these methods, pulse and pulsecyclic methods of ECM are of great
importance. The voltage pulse duration
is chosen such that the conductivity of
medium in the gap has no time to
change substantially, and the pulse-off
time assures the complete renewal of the
electrolyte in the gap. Pulse ECM enables
one to enhance the uniformity of current
density distribution, but reduces the ECM
productivity, because a considerable time
is taken by the pauses. A decline in
the productivity may become intolerably
great when a large surface area is being
machined: the larger the distance passed
by the electrolyte along the gap, the shorter
should be the pulse-on time and the longer
should be the pulse-off time. To reduce
this drawback, the pulse-cyclic conditions
of ECM have been elaborated. In this case,
the voltage pulse is applied when the gap is
small and, in the pause, the TE is removed
821
822
12 Electrochemical Machining
12.5
i on At
Va =
i=F
Na
Zk Jk (X ) on At
k=1
X
= div( JX ) + qX in t
t
(8)
I
C
(a)
(b)
Scheme of ECM:
is the initial WP surface; At is the WP surface
during the course of machining; C is the TE surface; 0 and t are the
interelectrode space bounded by the surfaces of the electrodes (C and A)
and insulators (I) before the machining (a) and during the course of
machining (b).
Fig. 7
A0
823
824
12 Electrochemical Machining
the temperature, the hydrodynamic velocity, and so forth, which are denoted by the
vector of generalized spatial elds X, over
the IEG. These elds vary in the region t
with moving boundary (Fig. 7) according
to the third equation of system (8), which
represents the conservation laws for the
components of vector X, that is, the conservation laws of mass, momentum, energy,
and so forth. Depending on the method of
describing the WP surface, the evolution
of the WP surface is determined by the
equations presented in Table 1.
The mathematical description of ECM
can be used for solving the direct and
inverse electrochemical shaping problems
[4, 9]. In the rst case, there is a need to
determine the shape and dimensions of
the machined WP surface. In this case,
the geometry of the TE working surface
is prescribed. In the second case, there
is a need to determine the shape and
dimensions of the TE working surface,
in order to produce the WP surface with a
required geometry.
In the general case, in order to solve
the direct and inverse ECM problems, it
is necessary to obtain the simultaneous
solution of the basic equations describing the WP surface evolution [9] and a
set of equations for the electrode reactions kinetics and transfer processes in
the gap, which determines the distribution
of current density over the WP surface.
Tab. 1
Method of description
Implicit coordinate
WP surface
(x, y, z, t) = 0
/t + Va |grad
| = 0
T(x, y, z) t = 0
Explicit coordinate
z = z(x, y, t)
Va |grad T| = 1
z/t + Va 1 + |grad z|2 = 0
Parametrical
/t = na Va
825
826
12 Electrochemical Machining
div(grad ) = 0 in t
grad grad
= 0 on At
t
(11)
see Table 1.
For simple-shaped electrodes (plane,
cylinder, and sphere), the problem can
be solved analytically [19]. For more
complex-shaped electrodes, numerical
methods are used. For the numerical
solution of the set of equations (11), both
iteration methods (Ref. 36, for example)
and the method of variation of inequalities
are used. The latter allows one to reduce the
initial problem to an equivalent problem in
the region with xed boundaries [37, 38].
In the steady-state case, the mathematical
description of the ideal ECM process
is reduced to the solution of Laplaces
equation for the electric potential in the
region with an unknown free boundary:
div(grad ) = 0 in t
grad grad
=
on At
Vc
= 0 on At
g ,
n
(12)
(U Ea + Ec )
Vc
(13)
43 2 1
y direction
[mm]
C
2
0
4
The variation in S0 leads to the variation in the shape of the machined surface
(Fig. 8). Consequently, control of machining parameters enables one to change the
shape and dimensions of the machined
surface. It is seen from (13) and Fig. 8 that
decreasing the voltage and raising the TE
feed rate allows one to decrease S0 and
improve the reproduction of TE shapes on
the WP.
2
0
x direction
[mm]
827
12 Electrochemical Machining
Fig. 9 The results of solving
the inverse ECM problem: A is
the desired WP surface given by
the equation y = x 2 ; 14 are
various TE surfaces that are
used to produce the desired WP
surface at the corresponding
equilibrium interelectrode gaps
S0 (mm): (1) 0.1, (2) 0.2,
(3) 0.3, and (4) 0.4.
3
4
A
y direction
[mm]
828
0.5
1
x direction
[mm]
1.5
Quasi-steady-state Approximation
The method of quasi-steady-state approximation (the step method) implies the use
of the explicit difference scheme for solving the problem with moving boundaries.
In doing so, the ECM rate eld at the instant time t is calculated from the transfer
equations (8) in the region t (Fig. 7) with
t+t =
t + tVat |grad
|
(14)
div( grad ) = 0
i = grad
(15)
g ,
=0
(16)
n
Various numerical methods are used
to solve Laplaces equation for ECM
including the method of nite differences,
the nite element method, the boundary
element method [9, 43, 44], and so
forth.
Since the regularities of gas transfer
from the electrode surface to the bulk
electrolyte are not clearly understood, it
is commonly assumed that = const
829
830
12 Electrochemical Machining
[4549]. In this case, the boundary element method is most suitable. Laplaces
equation is solved in the following way
(Fig. 10).
1. Three-dimensional, solid-state models
of the TE and WP are developed
(Fig. 10a).
(a)
5
3
5
5
(b)
Fig. 10
(c)
Ck
= div( jk + Ck (V + grad
))
t
k = 1, . . . , N
+ Rk
div(grad
) =
N
div(M k jk )
k=1
N
M k Rk
k=1
Q
+ div(Q[V + grad
])
t
= i grad + div( grad T )
N
zk Ck = 0
(17)
k=1
831
832
12 Electrochemical Machining
1
Va
(18)
(19)
(20)
K
ak fk (x, y)
(21)
k=1
(a)
(b)
(c)
(d)
Variation in the WP shape during ECM. The time after the machining
was started (min): (a) 1, (b) 2, (c) 3, (d) 4.
Fig. 11
=0
ak
ak
(22)
Here ak is a small perturbation
imposed on coefcient ak .
The derivatives in Eq. (22) are calculated
numerically; this requires K direct model
evaluations. To solve the set of nonlinear equations (22), it is advantageous to
use the NewtonRaphson method possessing quadratic rates of convergence and
requiring no more than four to seven iterations [45].
12.5.3
Within the local, one-dimensional approximation, the solution of the direct and
inverse ECM problems is considerably
simplied.
First, instead of determining the shape
of one electrode, it is sufcient to nd the
distribution of the IEG over the machining zone. This appears to be much simpler
than determining, immediately, the electrodes shape, because the IEG depends
only on three parameters (u, v, t), whereas
the electrode surface is determined by four
parameters (x, y, z, t).
Second, using the local, one-dimensional
approximation for the distribution of the
electrical potential, one can obtain the
following approximate equation for the
gradient of electrical potential:
grad
U Ea + Ec
=
(23)
833
12 Electrochemical Machining
60
i
H [A cm2]
834
40
20
10
20
30
x
[mm]
40
200
0.4
150
0.3
100
0.2
2
1
0.1
50
4
0
Q
[C cm2]
i
[A cm2]
3
1
t1
3 t2
t
[ms]
Fig. 13
r c = (x c , y c , zc )
(24)
(25)
835
836
12 Electrochemical Machining
(a)
(b)
2
(c)
(d)
Fig. 14
S
= Va 1+1 S|a Vc nc +div(SVc )|a
t
(27)
where Vc is the vector of the TE feed rate
and 1 S|a is the Beltrami rst differential
parameter of the IEG on the WP surface.
There are three terms in the right-hand
side of Eq. (27): the rst term allows for the
variation of the IEG due to the WP anodic
dissolution; the second term allows for the
variation of the IEG that is attributed due
to the normal to the TE surface component
of the feed rate; and the third term allows
for the variation of the IEG due to the
WP surface motion along the TE surface.
Using Eq. (27), the variation in the IEG
magnitude can be determined for any
scheme of quasi-equidistant ECM.
The numerical solution of Eq. (27) is
performed by the nite volume method
[63]. A mesh of triangular elements is
constructed on the working surface of
the TE. It is assumed that the curvature
of the TE surface is constant within
each element. The principal curvatures
and principal directions are determined
in the center of the element, when the
mesh is constructed. The calculations
are performed in the local system of
coordinates with the origin in the center
of the triangular element; the u and
v axes coincide with the directions of
principal curvatures. The set of ordinary
12.5.3.3
837
838
12 Electrochemical Machining
the design of the TE and the determination of the machining parameters. The
accuracy and productivity of nishing operations depend to a large extent on the
proper solution of these problems.
The difculty in the design of ECM operations is that the machining parameters
and the shape of the TE working surface
cannot be determined independently of
one another. When the machining parameters are changed, the IEG distribution
also changes; consequently, to produce a
WP of the desired shape and size, the
working surface of the TE should be
corrected.
Figure 15 gives the general scheme of
an automated design of ECM operations.
Using an integrated CAD/CAM/CAE system, the three-dimensional, solid-state or
surface model of the WP is created. To
elaborate an ECM operation, a geometrical
model of the WP and also other required
information (the WP material, the accuracy of dimensions and shape, and the
surface nish) are delivered from the universal CAD system to the subsystem of
ECM operation design.
The subsystem for the automatic design
of ECM operations in the dialog mode
enables the following problems to be
solved: (1) Analysis (the direct problem):
the shape and size of the WP surface are
determined by the given TE surface; and
(2) synthesis (the inverse problem): the
working surface of the TE is determined
in order to produce the desired WP
surface.
The use of the dialog mode in the design
of ECM operations is caused by the fact
that the mathematical models of ECM are
rather complicated, and the solution of
the problem depends on a large number of
parameters. Realization of the dialog mode
of design actually means the replacement
of the machining of test WPs (the trial
Solid-state
or
surface model
of the part,
material,
required
tolerance, and
surface finish
Fig. 15
12.6
Electrochemical Sinking
839
840
12 Electrochemical Machining
(a)
(b)
(a) Die produced by ECM sinking and (b) TE. Electrolyte feed holes and slots are
seen on the TE surface.
Fig. 16
Fig. 17
shaping.
Electrochemical Drilling
841
842
12 Electrochemical Machining
Electrolyte
TE rotation
(a)
Curved hole
(b)
Fig. 19
3
4
5
(a)
(b)
(c)
(d)
Fig. 20
843
844
12 Electrochemical Machining
Wire tension
Lead wire
Flexible tube
Curved hole
WP
Electrolyte
TE tip
Insulating cover
(a)
(b)
Fig. 21
12.6.3
Electrochemical Deburring
4
3
Before deburring
After deburring
1
Fig. 22
845
846
12 Electrochemical Machining
Fig. 23
applied for electrochemical nish machining. The power source (as a rule, pulse
one) and the electrolyte (usually NaNO3 )
differ from those for EDM.
The production of the razor shaving head
at one of the divisions (DAP) of Royal
Philips Electronics, where a production
line that is completely equipped with
ECM technology is employed [84], is a
good example of electrochemical nish
machining of shaped parts in mass
production.
12.6.5
Application of ECM in the eld of biomedical engineering is well known [64, 8587].
Stainless steel plates for use in bone
surgery and articial joints (e.g. hip
Subsea Application
In recent years, ECMM has developed intensively in accordance with the general
tendency toward the development of various methods of micromachining for manufacturing small-size parts or structural
elements with a characteristic dimension
of about 1 mm.
In the rst approach to the ECMM,
the required localization is achieved by
the use of inert photoresist masks on
the WP areas, which should not be
dissolved. This method has some specic
features (for instance, the problem of mask
undercutting and island formation) and
was considered, for example, in Ref. 91.
In the second approach, the required localization of metal dissolution is achieved
by other means. Some of them were considered in the previous sections. Here, TEs
of several types are used [9297]: (1) An
electrode with the shape and dimensions
corresponding to the required cavity in the
WP; (2) a TE in the form of a wire or needle, which penetrates into the WP at the
expense of its local anodic dissolution and
moves according to a certain program, in
order to obtain a given contour; and (3) a
high-speed jet beam of electrolyte from a
micronozzle.
847
848
12 Electrochemical Machining
Fig. 24
849
850
12 Electrochemical Machining
Hybrid Processes
ECM is performed without physical contact between the tool and the WP in
contrast to mechanical cutting and without
strong heating in the machining zone as
in EDM. Therefore, no surface metal layer
with mechanical and thermal distortion,
compressive stresses, and cracks form in
ECM. However, sometimes intergranular
attack (superalloys) or hydrogen embrittlement (titanium alloys) of the surface
layer occurs in ECM. This may degrade
the performance of the machined parts.
In order to eliminate the intergranular
attack and hydrogen embrittlement, the
electrolyte composition, temperature, and
the electrical conditions must be optimized
851
852
12 Electrochemical Machining
12.9
Conclusions
12.9 Conclusions
30. A. Zaytsev, I. Agafonov, N. Gimaev et al., J.
Mater. Process. Technol. 2004, 149, 419425.
31. A. V. Rybalko, S. I. Galanin, A. G. Atanasyants, Elektron. Obrab. Mater. 1993, (3), 36.
32. A. Davydov, 1st International Conference on
Machining and Measurements of Sculptured
Surfaces, Krakow, 1997, pp. 317330.
33. V. S. Krylov, A. D. Davydov, J. Kozak, Sov.
Electrochem. 1975, 11, 10751194.
34. A. L. Krylov, Dokl Akad Nauk SSSR 1968,
178, 321323.
35. J. M. Fitz-Gerald, J. A. McGeough, J. Inst.
Math. Appl. 1969, 5, 387408.
36. R. Hunt, Int. J. Number. Methods Eng. 1990,
29(6), 11771192.
37. C. M. Elliott, Free boundary problems: theory
and applications, Boston, 1983, pp. 505511.
38. C. M. Elliott, J. Inst. Math. Appl. 1980, 25,
121131.
39. A. A. Lacey, IMA J. Appl. Math. 1985, 34,
259267.
40. H. Tipton, V International Conference on
Machine Tool and Research, Birmingham,
1964, pp. 509522.
41. J. Kozak, Arch. Bud. Masz. 1967, 15,
240256.
42. A. Kh. Karimov, V. V. Klokov, E. I. Filatov,
Metody Rascheta Elektrokhimicheskogo Formoobrazovaniya, Kazanskii Universitet,
Kazan, 1990.
43. J. Deconinck, Current Distributions and
Electrode Shape Changes in Electrochemical
Systems, Lecture Notes in Engineering, Vol.
75, Springer-Verlag, Berlin, 1992, pp. 281.
44. G. A. Prentice, C. W. Tobias, J. Electrochem.
Soc. 1982, 129, 7278.
45. Y. Zhou, J. J. Derby, Chem. Eng. Sci. 1995,
50/17, 26792689.
46. H. Shirvani, A. Shirvani, I. Mirzaee, II International Conference on Advances in Production Engineering. Part II, Warsaw, 2001,
pp. 343352.
47. M. J. Noot, A. C. Telea, J. K. M. Jansen et al.,
Comput. Vis. Sci. 1998, 1, 105111.
48. M. Purcar, J. Deconinck, L. Bortels, J.
Mater. Process. Technol. 2004, 149, 472478.
49. L. Bortels, J. Deconinck, P. Van Ransbeeck,
J. Mater. Process. Technol. 2004, 149,
486492.
50. C. S. Chang, L. W. Hourng, C. T. Chung, J.
Appl. Electrochem. 1999, 29, 321330.
51. C. S. Chang, L. W. Hourng, J. Appl. Electrochem. 2001, 31, 145154.
853
854
12 Electrochemical Machining
75. E. J. Weller, (Ed.), Nontraditional Machining
Processes, 2nd ed., Society of Manufacturing
Engineers, Dearborn, 1984.
76. L. K. Gillespie, Deburring Technology for Improved Manufacturing, Society of Manufacturing Engineers, Dearborn, 1981.
77. H. Schein, SME Technical Paper MR87154, Society of Manufacturing Engineering,
Dearborn, 1987.
78. T. Wang, Int. J. Electr. Mach. 1996, (1), 912.
79. D. Ball, in Cost Guide for Automatic Finishing
Processes (Ed.: L. J. Rhoades), Society of
Manufacturing Engineers, Dearborn, 1981.
80. S. Sakai, T. Masuzawa, S. Ito, Proceedings of International Symposium ElectroMachining (ISEM-9), The Japan Society
of Electro-Machining Engineers, Nagoya,
1989, pp. 155158.
81. T. Masuzawa, S. Sakai, Ann. CIRP 1987, 36,
123126.
82. F. Klocke, A. Karden, XII International
Symposium for Electromachining (ISEM
XII), Aachen, 1998, VDI BERICHTE,
pp. 375385.
83. B. Lilly, J. Brevick, C.-S. Chen, Trans.
NAMRI/SME 1997, 25, 153158.
84. H. S. J. Altena, J. Mater. Process. Technol.
2004, 149, 1821.
85. J. Bannard, J. Appl. Electrochem. 1975, 5,
8990.
86. C. Madore, D. Landolt, J. Micromech. Microeng. 1997, 7, 270275.
87. P.-F. Chauvy, P. Hoffmann, D. Landolt,
Appl. Surf. Sci. 2003, 208209, 165170.
88. D. Clifton, M. B. Barker, R. W. Gusthart
et al., Underwater Technol. 1991, 17(4), 19.
89. J. A. McGeough, Ocean Ind. 1983, 18, 282.
90. T. J. S. Green, Underwater Syst. Des. 1986,
8(5), 2830.
91. M. Datta, D. Landolt, Electrochim. Acta
2000, 45, 25352558.
92. M. Suda, N. Watanabe, K. Furuta et al.,
2nd International Workshop on Microfactories (IWMF 2000), Fribourg, 2000,
pp. 163166.
93. B. Bhattacharyya, S. Mitra, A. K. Boro, Rob.
Comput. Integr. Manuf. 2002, 18, 283.
94. T. Masuzawa, T. Takawashi, XII International Symposium for Electromachining (ISEM XII), Aachen, 1998, VDI
BERICHTE, pp. 115.
95. R. Schuster, V. Kirchner, P. Allongue et al.,
Science 2000, 289, 98.
855
13
Environmental Electrochemistry
Barry MacDougall, C. Bock and M. Gattrell
National Research Council of Canada, Ottawa, Ontario, Canada
13.1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
857
13.2
858
13.3
860
13.4
866
13.5
869
13.6
871
13.7
872
13.8
874
13.9
876
13.10
877
13.11
Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
878
879
857
13.1
Introduction
858
13 Environmental Electrochemistry
13.2
Anode
Cathode
H
CI
CO2 + H2O
CI
CI
CI
CI
CI
Direct
(At electrode)
Aqueous
H2SO4; HCIO4
Organic
C6H6; C6Hx CI6x
PCB; PCP
H
Indirect
Ag+
Ag2+
Superoxide O2
O3
Nonaqueous
DMF; DMSO;
CH3CN
Inorganic
Pb2+; Zn2+; Cu2+;
Hg2+; Cd2+; Sn2+;
859
860
13 Environmental Electrochemistry
Cathodic dehalogenation
CI
OH
CI + 5H+ + 10e
CI
H + 5CI
H
H
CI
CI
OH
Pentachlorophenol
Phenol
Anodic oxidation
H
OH
H + 11H2O
H
Phenol
Fig. 2
13.3
destruction/modication of problem
toxins, both organic or inorganic, in
solution since this is a topic of
great academic interest with considerable
industrial potential at the present time.
It also illustrates very well the important
elements common to all electrochemical
reactor systems, that is, nature of
solution; geometry, activity, and lifetime
of anode catalyst; optimum current
and/or potential for system operation;
design of electrochemical cell and overall
treatment process (e.g. batch versus
continuous, divided versus undivided cell);
need for auxiliary equipment such as
pumps and valves; and the need for
nonelectrochemical pre- or posttreatment
such as ltration and biotreatment, if
a hybrid system is required [13, 14].
The overall cost of the system, broken
down into capital and operating cost, will
determine how competitive the process is,
and whether the electrochemical treatment
route has a chance of being chosen by
industry to do the particular environmental
S[OH] + R S[ ] + RO + H + e
(2)
S[OH] + H2 O S[ ] + O2
+ 3H+ + 3e (3)
(1) being discharge of water onto the
oxide surface S, (2) being oxygen transfer
to a reactant R in solution, and (3) being
parallel oxygen evolution. The process is
referred to as electrochemical incineration and ideally results in either complete
mineralization of the organic, R, to CO2 or
its conversion to a less-toxic form. Shortterm experiments were conducted on the
oxidation of organics such as phenol,
benzene, toluene, and p-benzoquinone,
as well as inorganics such as cyanide,
Cr(3+), and dimethylsulfoxide (DMSO)
(as a model for the oxygen-transfer process). The inuence of the presence of
various dopants in the -PbO2 , as well
as the doping level, was investigated. The
general observation was that the presence
861
862
13 Environmental Electrochemistry
H+ + OH* ads + e
Organic mineralization
OH
+ 34OH* ads
7CO2 + 21H2O
CH3
p-Cresol
Oxygen evolution
OH*ads
1/2O2 + H+ + e
863
864
13 Environmental Electrochemistry
2000
TDC (ppm)
IrO2
1500
Platinum
1000
500
PbO2
0
10
15
20
25
30
Time (h)
Fig. 4
865
866
13 Environmental Electrochemistry
13.4
represented as
RX + H+ + 2e RH + X
(4)
2.0
1.0
Cyclic voltammogram of
hexachlorobenzene at a glassy carbon
electrode in acetonitrile containing
0.1 M tetraethylammonium
tetrauoroborate (TEABF4). Sweep rate
is 50 mV s1 . The cathodic peaks
represent the stepwise dechlorination
process.
Fig. 5
4
5
6
Reduction current
867
868
13 Environmental Electrochemistry
(5)
The iron is in the form of a permeable reactive wall that is installed in the
ground and through which the contaminated groundwater must ow. Considerable research has been done in this eld
by Gillham and others [7780] and its
applicability in the eld has been demonstrated [81, 82]. The technique can be
used to treat a wide range of aliphatic
organohalides (e.g. CCl4 , C2 Cl6 , CH2 Cl4 ,
vinylchloride, etc.), the Fe serving as a
true catalyst in lowering the overvoltage
for dehalogenation. Pollutant levels have
been decreased from 100s of ppm to
1 ppm level. The approach does not,
however, appear to work for halogenated
aromatics. The technique has been proven
on both pilot demonstration and fullscale installations in the United States
and Europe. Investigations are now being made on its effectiveness at depths
>40 feet, the limit of standard excavation methods, using special engineering
injection methods. The attraction of the
zero-valent iron dechlorination method
is that it is extremely simple, requiring no outside electrochemical control
of the reactions, and it would appear to
have great future potential. For further
information on the practical applications
of the technology, the reader is referred
to the following representative website:
www.beak.com/eti.html.
13.5
869
870
13 Environmental Electrochemistry
(8)
13.6
871
872
13 Environmental Electrochemistry
13.7
873
874
13 Environmental Electrochemistry
are examples of success with the aqueousphase TiO2 approach for both organics
and inorganics such as H2 S and CN
[120124], the technology is not as yet mature (in comparison with homogeneous
photolysis), and more work is needed at
the engineering scale-up phase in terms of
both economics and performance. Other
AOTs for pollution treatment such as
electrohydraulic cavitation/sonolysis and
supercritical water oxidation are outside
the scope of this chapter, and further reading on this topic is recommended [125].
As with all other treatment technologies, considerable attention has been paid
to the pollutant degradation reactions and
the extent to which it proceeds to complete
mineralization. This is extremely important, especially when one considers the
possibility of generating intermediates that
may be more toxic than the starting material. Consequently, care has been taken
to identify intermediates and do proper
mass-balance calculations for the reaction.
Hybrid approaches should also be considered, for example, combining the AOTs
discussed earlier with more standard treatments such as GAC adsorption in a preor poststage. In the end, economic considerations will determine the approach that
is used. The reader is referred to Chapter 8 of Ref. 112 for a detailed evaluation
of the economics of the various treatments
discussed earlier.
13.8
Ion permeable (or ion exchange) membranes play a major role in electrochemical
synthesis and processing and have revolutionized electrochemical engineering
during the past 30 or so years. They are often composed of an organic polymer onto
875
876
13 Environmental Electrochemistry
Dilute
MX
Concentrated
HX
Anionpermeable
membrane
Concentrated
MOH
Cationpermeable
membrane
Cathode
Anode
X
e
H2O
H+
H+
Dilute
HX
OH
H+
H2O
OH
Concentrated
MX
Dilute
MOH
Fig. 6
Electrochemical Remediation
of Contaminated Soil
877
878
13 Environmental Electrochemistry
pollutants as well as monitor their concentrations in both the gaseous and liquid
phases. In the scientic laboratory, this
is accomplished by using analytical instruments such as HPLC and GCMS, in
conjunction with TOC and COD measurements, to follow the nature and level of
the pollutants as a function of treatment
conditions. The crucial importance of such
analysis hinges on the fact that, while complete mineralization of pollutants may not
always be necessary, one certainly has to
generate a product that is substantially less
toxic than the starting material. Alternatively, there may be considerable benet in
electrochemically forming a product that
can be easily handled by another treatment route, for example, the conversion
of a nonbiodegradable pollutant into one
that is readily biotreated [152]. It may, indeed, be economically advantageous to
only do a partial electrochemical treatment in light of the number of electrons
required for some reactions. This would
then be followed by a subsequent alternative treatment (chemical, biological, etc.).
To make these decisions with any degree
of intelligence, it is obviously necessary to
have information about the mechanism of
the treatment reaction, that is, to perform
product analysis and proper mass balances
during the course of the reaction.
In eld applications, the monitoring of
the type and level of pollutants is also
of great importance, and information is
often required in real time rather than the
several days of delay needed for sending
samples to the laboratory. In response
to this need, major advances have been
made in the development and application
of small-scale, portable sensors for in
situ monitoring of pollutants [153158].
Sensors can be of the solid-state, optical,
bio-, and electrochemical types, examples
of the latter being pH- and ion-selective
13.11
Concluding Remarks
be achieved by either using electrochemistry as a cleaner synthesis and manufacturing route (if this is indeed feasible) or
employing electrochemistry as part of an
overall, hybrid production system in order
to achieve cleaner manufacturing. In this
chapter, most of the discussion has concerned the more standard electrochemical
approach to detecting and treating pollution; nevertheless, electrochemistry can
have a real role in preventing pollution in
the rst place.
As with all other environmental treatment and prevention technologies, the
key to the utilization of electrochemistry
is cost. It matters little how scientically marvelous the technology is; if the
cost is prohibitive, it will not be used.
Factors inuencing the cost of electrochemical technology include reaction CE,
the electrocatalyst required and its lifetime,
design, and construction of electrochemical reactor (plus auxiliary equipment such
as pumps and valves), safety aspects of
the process, necessity for pre- or posttreatment, and cost of electricity. In some
instances, the latter may be the least concern, especially where electricity costs are
in the range of 3.5 c/kwh. The electrode
material and reactor system costs are critical since they represent a large fraction of
the overall system cost. If a useful product can be obtained from the treatment
method (e.g. metals from cathodic reduction), then this economic benet can help
to offset the treatment cost. The environmental benet, while being most difcult
to quantify than the economic return, is
probably the most important factor for society as a whole.
In this chapter, a number of electrochemical treatment technologies have
been discussed to address various environmental problems. Each problem will have
its own specic circumstances that will
879
880
13 Environmental Electrochemistry
11. S. I. Bailey, I. M. Ritchie, F. R. Hewgill,
J. Chem. Soc., Perkin Trans. 2 1983, 5,
645652.
12. S. D. Roso, M. Finkelstein, E. J. Rudd,
Anodic Oxidation, Academic Press, New
York, 1975, p. 273.
13. F. Goodridge, K. Scott, Electrochemical Process Engineering, Plenum, New York, 1995,
Chapter 6.
14. D. Danly, D. Genders, D. Pletcher, Principles of electrochemical engineering and
scale-up, Short Course at 9th International
Forum on Electrosynthesis in the Chemical
Industry, Clearwater Beach, 1995
15. I. H. Yeo, D. C. Johnson, J. Electrochem. Soc.
1987, 1341, 19731977.
16. Y. L. Hsiao, D. C. Johnson, J. Electrochem.
Soc. 1989, 136, 37043710.
17. K. T. Kawagoe, D. C. Johnson, J. Electrochem. Soc. 1994, 131, 34053409.
18. Y. L. Hsiao, J. E. Vitt, D. C. Johnson,
J. Electrochem. Soc. 1992, 139, 377380.
19. D. C. Johnson, N. Popovic, J. Feng et al.,
The Electrochem. Soc. Proc. Vol. 1995, 9526,
176188.
20. D. C. Johnson, L. L. Houk, J. Feng et al.,
Environ. Sci. Technol. 1999, 33, 26382643.
21. D. C. Johnson, H. Chang, J. Feng et al. in
Electrochemistry for a Cleaner Environment
(Eds.: T. L. Rose, E. Rudd, O. Murphy
et al.), The Electrochemical Society, 1994,
pp. 7586, PV9419.
22. N. B. Tahar, A. Savall, J. Appl. Electrochem.
1999, 29, 277283.
23. N. D. Popovic, J. A. Cox, D. C. Johnson,
J. Electroanal. Chem. 1998, 456, 202209.
24. C. Comninellis, C. Pulgarin, J. Appl. Electrochem. 1993, 28, 108112; 1991, 21,
703708.
25. C. Pulgarin, N. Alder, P. Peringer
et al.,
Water Res. 1994, 28, 887894.
26. C. Comninellis, Electrochimi. Acta 1994, 39,
18571862.
27. C. Comninellis in Water Purication
by Photocatalytic, Photoelectrochemical and
Electrochemical Processes (Eds.: T. L.
Rose, E. Rudd, O. Murphy et al.), The
Electrochemical Society, 1994, pp. 7586,
PV9419.
28. S. Stucki, R. Kotz, B. Carcer et al., J. Appl.
Electrochem. 1991, 21, 99104.
29. R. Kotz, S. Stucki, B. Carcer, J. Appl. Electrochem. 1991, 21, 1420.
30. A. Savall, Chimia 1995, 49, 2326.
54.
55.
56.
57.
58.
59.
60.
61.
62.
63.
64.
65.
66.
67.
881
882
13 Environmental Electrochemistry
90. J. J. Byerley, K. Enns, CIM Bull. 1984, 77,
8793, No. 861.
91. K. E. Haque, D. J. MacKinnon, CIM Bull.
1996, 89, 104106, No. 1001.
92. C. S. Fugivara, P. T. A. Sumodjo, A. A.
Castro et al., The Analyst 1996, 121,
541546.
93. M. L. Lin, Y. Y. Wang, C. C. Wan, J. Appl.
Electrochem. 1992, 22, 11971200.
94. H. Tamura, T. Arikado, H. Yoneyama et al.,
Electrochim. Acta 1974, 19, 273277.
95. T. Arikado, I. Iwakura, H. Yoneyama et al.,
Electrochim. Acta 1976, 21, 10211027.
96. P. Tissot, M. Fragniere, J. Appl. Electrochem.
1994, 24, 509512.
97. D. Pletcher, N. L. Weinberg, Chem. Eng.
1992, 132141.
98. D. Pletcher, F. C. Walsh in Electrochemistry
for a Cleaner Environment (Eds.: D. Genders,
N. Weinberg), The Electrosynthesis Company, New York, 1992, pp. 51100.
99. D. Pletcher, F. C. Walsh, Industrial Electrochemistry, Chapman & Hall, London and
New York, 1990, pp. 336346.
100. D. Simonsson, Chem. Soc. Rev. 1997, 26,
181189.
101. D. Pletcher, F. C. Walsh, Industrial Electrochemistry, Chapman & Hall, London and
New York, 1990, pp. 346350.
102. F. C. Walsh, D. R. Gabe, Trans. Inst. Chem.
Eng. 1990, 68B, 107117.
103. F. C. Walsh, as in ref. 98, pp. 101159.
104. D. R. Gabe, G. D. Wilson, J. GonzalezGarcia et al., J. Appl. Electrochem. 1998, 28,
759780.
105. J. M. Marciel, S. M. L. Agostinho, J. Appl.
Electrochem. 1999, 29, 745750.
106. B. Fleet, Collect. Czech. Chem. Commun.
1988, 53, 11071133.
107. C. J. Brown, as in ref. 98, pp. 183205.
108. D. Pletcher, as in ref. 98, pp. 1149.
109. T. A. Davies, as in ref. 98, pp. 173181.
110. S. Itoi, I. Nakamura, T. Kaahara, Desalination 1990, 32, 383389.
111. C. Von Sonntag, G. Mark, R. Martens et al.,
J. Water SRT-Aqua 1993, 42, 201210.
112. K. Rajeshwar, J. Ibanez, Environmental Electrochemistry, Academic Press, New York,
1997, Chapter 6.
113. See Chapter 8 in ref. [112].
114. M. I. Litter, Appl. Catal., B: Environ. 1999,
23, 89114.
883