Chemical Engineering Science 60 (2005) 5830 5835

www.elsevier.com/locate/ces

CO2 absorption in carbonate/bicarbonate solutions: The

Danckwerts-criterion revisited
A.H.G. Cents1 , D.W.F. Brilman1 , G.F. Versteeg
University of Twente, PO Box 217, 7500 AE Enschede, The Netherlands

Received 7 July 2004; received in revised form 27 April 2005; accepted 5 May 2005
Available online 1 July 2005

Abstract
In industrial applications CO2 is frequently removed from gas streams at elevated pressures by absorption and subsequent chemical
reaction in carbonate/bicarbonate solutions (e.g. Beneld process). The criterion that determines whether or not this reaction can be
regarded as pseudo-rst order is given by Danckwerts and Sharma (1966), but its derivation has never been published. In the present
study, a consistent set of equations and physico-chemical parameters is presented to describe the CO2 absorption/reaction process. It was
found that the criterion was justied, but should be somewhat conned to reduce possible errors. Furthermore, the model was validated
by experiments in a stirred cell reactor up to CO2 partial pressures of 16 bar.
2005 Elsevier Ltd. All rights reserved.

Keywords: Danckwerts-plot; CO2 absorption; Mass transfer; Stirred cell; Elevated pressure

1. Introduction intercept matches the square of the volumetric mass transfer

coefcient (kL a)2 . A commonly used reaction system for
The Danckwerts-plot technique (Danckwerts et al., 1963) this method is a carbonate/bicarbonate buffer solution, in
is used in chemical engineering to simultaneously obtain which the reaction between CO2 and water can be catalyzed
the mass transfer parameters, kL and a, from mass transfer by a number of agents, e.g. hypochlorite and arsenite. The
experiments. From the measurements of the gas absorption criterion that determines whether or not the reaction of CO2
rate, RA (mole s1 ), at different apparent rst order reaction in a carbonate/bicarbonate buffer solution can be regarded
rate constants the values of kL and a can be determined as pseudo-rst order is given by Danckwerts and Sharma
simultaneously using the Danckwerts surface renewal model (1966):
(Danckwerts, 1950) with a (pseudo) rst order reaction:  
1 2

2 mCO 2 cCO 2 (G) +
RA cCO 2 cH CO
= (kL a)2 + k1,app DA a 2 . (1)

3
m A c A VL  
3

 DCO 2 k1,app
When the left-hand side of Eq. (1) is plotted versus the ap- 1 + 1  1. (2)
kL2
parent rst order rate constant times the diffusion coef-
cient of the gas into the liquid (k1,app DA ), the slope equals
The reaction can be regarded as pseudo-rst order when the
the squared specic gas liquid interfacial area (a 2 ) and the
concentrations of all the ions are uniform throughout the
mass transfer zone. Although this Danckwerts-criterion is
Corresponding author. Tel.: +31 53 489 3327; fax: +31 53 489 4774.
often referred to in literature, its derivation has never been
E-mail address: g.f.versteeg@utwente.nl (G.F. Versteeg).
1 Present address: SASOL Technology Netherlands B.V., c/o Faculty published. Moreover, it was found that although the left-hand
of Chemical Technology, University of Twente, PO Box 217, 7500 AE side of Eq. (2) is much smaller than 1, still large concentra-
Enschede, The Netherlands. tion gradients could exist, and thus considerable deviations

0009-2509/$ - see front matter 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ces.2005.05.020
 A.H.G. Cents et al. / Chemical Engineering Science 60 (2005) 5830 5835 5831

from the pseudo-rst order region may occur. The criterion rH CO = k11 cCO 2 (L) cOH k12 cH CO + k22 cCO 2
3 3 3
is based on the fact that the transport of carbonate from k21 cH CO cOH + k41 cCO 2 (L)
3
the bulk can become the limiting step in the mass transfer k42 cH CO cH + , (12)
process. However, due to the inuence of both the carbonate 3

and bicarbonate ions on the hydroxyl concentration in the rCO 2 = k21 cH CO cOH k22 cCO 2 , (13)
mass transfer zone, a possible reduction in ux is difcult to 3 3 3

predict on beforehand. Information regarding this criterion rH + = k32 k31 cOH cH + + k41 cCO 2 (L)
is therefore very important, both for research activities (the
k42 cH CO cH + . (14)
above-mentioned Danckwerts-plot technique), as well as for 3

industrial purposes, in which CO2 is frequently removed The bulk concentrations of all species are calculated using
from gas streams at high pressure using chemical absorption the equilibrium constraints, the overall carbon balance and
in a liquid (e.g. the Beneld process, Benson et al., 1954). the electroneutrality constraint:
At these elevated CO2 pressures the Danckwerts-criterion
is violated more easily and accurate knowledge regarding K1 /KW cCO 2 (L),bulk cOH ,bulk = cH CO ,bulk , (15)
3
the solution of the complete reaction system is therefore
required. K2 cOH ,bulk cH CO ,bulk = cCO 2 ,bulk , (16)
3 3
To study the validity of the above criterion, the ux of
CO2 from the gas to the liquid phase was determined nu- cH CO ,ini + cCO 2 ,ini = cCO 2 (L),bulk + cH CO ,bulk
3 3 3
merically by solving the mass balances for all components.
+ cCO 2 ,bulk , (17)
Furthermore, the developed model was compared to mass 3

transfer experiments up to a CO2 partial pressure of 16 bar.

KW cH + ,bulk cOH ,bulk = 0, (18)

cH + ,bulk + cK + ,bulk = cOH ,bulk + cH CO ,bulk

3
2. Reaction system
+ 2cCO 2 ,bulk . (19)
3
The following reactions occur in the carbonate/bicarbonate
The concentration proles of all species in the mass trans-
solution:
fer zone can be calculated as a function of time using the
k11 following equation:
CO 2 + OH
H CO
3, (3)
k12
jcA (x, t) j2 cA (x, t) F j(
(x, t)cA (x, t))
= DA z A DA
k21 jt jx 2 RT jx
H CO
3 + OH
CO 3 + H2 O,
2
(4) + rA (x, t). (20)
k22

k31 The electrostatic potential gradient (
) can be calculated by

OH + H +
H2 O, (5) the use of the NernstEinstein equation (Newman, 1973)
k32 assuming dynamic electroneutrality:
k41 NC jcq (x,t)
CO 2 + H2 O
H CO +
3 +H . (6) RT q=1 zq Dq jx
k42
(x, t) = NC 2 . (21)
F q=1 zq Dq cq (x, t)
The corresponding equilibrium constants are dened as:
cH + cH CO The following boundary conditions are used:
K1 = 3
, (7)
cCO 2 (L) t = 0, x > 0 : cA (x, t) = cA,bulk ,
cCO 2
K2 = 3
, (8) t > 0, x = 0 : cCO 2 (x, t) = mCO 2 cCO 2 (G) ,
cH CO cOH
3
t > 0, x = 0:
KW = cH + cOH . (9) jcion (x, t) F
Dion zion Dion
(x, t)cion (x, t) = 0,
The reaction rate per component is then given by: jx RT

rCO 2 = k12 cH CO k11 cCO 2 (L) cOH t > 0, x = : cA (x, t) = cA,bulk . (22)
3
+ k42 cH CO cH + k41 cCO 2 (L) , (10) The ux of CO2 through the gasliquid interface is calcu-
3
lated using the Higbie penetration model (Higbie, 1935). The
rOH = k12 cH CO k11 cCO 2 (L) cOH + k22 cCO 2
3 3 complete system of equations was solved using the method
k21 cH CO cOH + k32 k31 cOH cH + , (11) of lines.
3
5832 A.H.G. Cents et al. / Chemical Engineering Science 60 (2005) 5830 5835

3. Physical parameters Table 1

Used physical and chemical parameters at 297 K for the absorption of
CO2 in a 0.5 M KHCO3 /0.5 M K2 CO3 solution
The solubility and diffusivity of CO2 in the 0.5 M/0.5 M
potassium carbonate/potassium bicarbonate solution are Parameter Value Parameter Value
given elsewhere (Cents et al., 2001). All physical parameters
k11 , m3 mol1 s1 1.41 101 DCO 2 , m2 s1 1.59 109
at 24 C are given in Table 1. k12 , s1 3.11 104 DOH , m2 s1 5.17 109
The ionic diffusion coefcients have been estimated using k21 , m3 mol1 s1 6.00 106 DH CO , m2 s1 1.18 109
3
the Nernst equation (Horvath, 1985): k22 , s1 1.22 106 D , m2 s1 9.20 1010
CO 2
3
RT k31 , m3 mol1 s1 1.40 108 DH + , m2 s1 9.21 109
DA = 
A . (23) k32 , mol m3 s1 1.29 100 DK + , m2 s1 1.95 109
zA F 2
k41 , s1 0.24 101 mCO 2 , - 5.24 101
The ionic conductivities at innite dilution, 
A , have been k42 , m3 mol1 s1 5.71 101 W , kg m3 9.97 102
tted to the experimental values of Horvath (1985).

k22 = k21 /K2 , is given by Hikita et al. (1976):

4. Chemical parameters
1568.9
log(K2 ) = 2.5866 6.737 103 T. (28)
The value of k11 in the presence of co-electrolytes is given T
by Pohorecki and Moniuk (1988): The values of the backward reactions, k32 and k42 can be
k11 calculated from the equilibrium constants and are k31 /KW
log = 0.11 103 cK + + 0.17 103 cCO 2 . (24) and k41 /K1 , respectively.
k11 3

The inuence of the bicarbonate ion was not given by these

authors and was therefore neglected. The reaction rate con-
, is given by
stant at innite dilution, k11
2383
log k11 = 8.916 . (25)
T hanced due to the reaction, but this enhancement is less
Reaction 4 is fast as it involves only a proton transfer.
The reaction rate constant, k21 , was determined to be 6
106 m3 /(mol s) (Eigen, 1963). The neutralization rate con-
stant, k31 , was also determined by this author and was 1.4
108 m3 /(mol s). The reaction rate between CO2 and wa-
ter, k41 , is very slow (0.024 s1 , Danckwerts and Sharma,
1966). This reaction can, however, be catalyzed by a num-
ber of agents. Cents et al. (2001) used sodium hypochlo-
rite for which a reaction rate constant was determined to be
1.8 m3 mol1 s1 at 294 K.
The backward reaction rate of reaction 3, k12 , is dened
by the value of the equilibrium constant for this reaction
(k12 = k11 KW /K1 ). K1 is given as a function of temperature
by Edwards et al. (1978):


12092.1
K1 = exp 36.786 ln(T ) + 235.482 W ,
T assumption of a pseudo-rst order reaction is made.
(26)
where W is the density of water (kg/m3 ).
The value of the solubility product KW is taken from
Tsonopoulos (1976):
 
KW 5839.5
log = 22.4773 log(T ) + 61.2062.
W
2 T
(27)
The equilibrium constant (K2 ) at innite dilution that de-
termines the value of the backward reaction of reaction 4,
5. Results
In Fig. 1 typical concentration proles of the different
components are presented. The absorption of CO2 is en-
compared to the case when a pseudo-rst order reaction is
assumed, due to the depletion of hydroxyl ions near the in-
terface. The average ux of CO2 from the gas phase to the
liquid phase is 23% lower, when the concentration proles
of all components are taken into account, compared to case
with the assumption of a pseudo-rst order reaction.
The signicance of the Danckwerts criterion is shown in
Fig. 2. The ratio of the actual ux and the ux when a
pseudo-rst order reaction is assumed is plotted versus the
left-hand side of Eq. (2) for two different cases: by variation
in the mass transfer coefcient kL and by variation of the
CO2 partial pressure. In the variation of the kL , a constant
pressure of 1.0 bar was applied and in the variation of the
pressure a constant kL of 5105 m/s was taken. The results
show that the left-hand side of the equation should be less
than 0.1 in order to sustain an error less than 3% when the
To test the modelling results, CO2 absorption experiments
were carried out at 24 C in a 0.5 M KHCO3 /0.5 M K2 CO3
buffer solution in a 640 ml stirred autoclave, which contained
300 ml of liquid. A stirring speed of 60 rpm was used, which
ensured a at gasliquid interfacial area, as shown from ex-
periments with different stirring speeds. In the experiments
the CO2 partial pressure was varied from 0.5 to 16 bar to
change the value of the left-hand side in the Danckwerts
criterion (Eq. (2)).
The mass transfer coefcient, kL , of CO2 in water was de-
termined from the reactor pressure vs. time curve in which
 A.H.G. Cents et al. / Chemical Engineering Science 60 (2005) 5830 5835 5833

2 x 10-4
1
water
buffer solution
normalized concentration (-)

kL * EA (m/s) model
1.5 0.8
-
HCO3

0.6
1
2-
CO3
OH- 0.4

0.5
0.2
CO2

0
0 1 2 3 4 0
0 5 10 15 20
penetration depth (m) x 10-4 pressure (bar)

Fig. 1. Concentration proles in the 0.5 M KHCO3 /0.5 M K2 CO3 buffer Fig. 3. kL EA versus the pressure of CO2 in a stirred cell reactor. Model
solution at the end of the contact time. pCO 2 = 4 bar, kL = 2 105 m/s. line: 297 K.

interval. It was veried that the composition of the buffer

JCO2 did not change signicantly during this time.
1.1 pfo The results are presented in Fig. 3 and more details are
JCO2
given in Table 2. A reasonable agreement between the model
1 and the experiments is obtained at low pressures (the model
predicts a ux that is approximately 4% higher compared
variation of pCO2
0.9 to the experiments). The effect of ionic strength on K2 ,
which was taken into account by Roberts and Danckw-
0.8 erts (1962) and Hikita et al. (1976) (for sodium ions) was
found to be too strong and caused an underestimation of
0.7 variation of kL the ux (via an underestimation of the OH concentration)
by almost a factor of 2 (kL EA = 3.83 105 m/s) when
0.6 making use of a correction for the presence of potassium
ions:
0.5

0 0.2 0.4 0.6 0.8 1 1.2 1.4 K2 1.01 cK + /1000
log = + 6.1 105 cK + . (30)
left hand side of equation 2 K2 1 + 1.49 cK + /1000
Fig. 2. Ratio of the ux calculated with the full model and the ux Not taking this ionic effect on K2 into account, was sup-
calculated with the assumption of a pseudo-rst order (pfo) reaction.
ported by two other measurements:

Measurement of the CO2 bulk concentration was per-

the dissolved CO2 concentration was taken into account. The formed by leading a small nitrogen stream through an
value of the mass transfer coefcient in the buffer solution intensely stirred buffer solution. It was veried that the
was estimated from experiments in pure water using a cor- outlet gas stream was saturated with CO2 by variation of
rection for the diffusion coefcient (the power of 0.6 was the ow rate. At 21 C the nitrogen stream was saturated
estimated from the work of Versteeg et al. (1987)): with 0.21% CO2 , which could be calculated back to a liq-

 uid concentration of 0.053 mol/m3 . With: K2 = K1 /KW
DCO 2 ,buffer 0.6 2
kL,buffer = kL,water . (29) [CO 2 ][CO 2
3 ]/[H CO 3 ] this leads to K2 = 5.8 (K2 in-
DCO 2 ,water nite dilution = 5.8, K2 ionic effect = 19.6).
The value of kL in pure water was not dependent on the The hydroxyl concentration was measured using a PH-
pressure and was found to be 1.93 105 m/s. Application meter. At 21.2 C a pH of 10.38 was measured. This can
of Eq. (29) lead to an estimated kL value of 1.75105 m/s be reworked to a hydroxyl concentration of 0.18 mol/m3 .

in the buffer solution. The ux of CO2 in the buffer solution Using K2 =[CO 2 3 ]/([H CO 3 ][OH ]) this leads to K2 =
was determined from the pressure drop in a certain time 5.6. (K2 innite dilution = 5.8, K2 ionic effect = 19.5).
5834 A.H.G. Cents et al. / Chemical Engineering Science 60 (2005) 5830 5835

Table 2
Detailed experimental and modelling results

Experimental Model

Pressure Temperature RA JA a kL E A kL E A Relative diff. OH

int /OHbulk
(bar) ( C) (106 mol/s) 103 mol/(m2 s) 105 m/s 105 m/s % b (-)

0.52 24.0 2.26 0.69 6.19 6.38 3.07 0.79

1.34 24.5 5.34 1.62 5.74 6.04 5.29 0.58
2.02 24.1 7.60 2.31 5.38 5.55 3.12 0.48
3.60 24.0 12.07 3.67 4.80 5.00 4.14 0.34
5.95 24.2 18.04 5.49 4.35 4.50 3.49 0.22
7.54 24.2 22.11 6.73 4.21 4.23 0.55 0.17
10.59 24.2 27.34 8.32 3.70 3.85 3.85 0.12
15.17 23.9 36.07 10.98 3.41 3.43 0.77 0.07
aA
cell = 32.9 cm .
2
bAt the end of a penetration time.

The ionic effect on K2 was therefore not considered in the DA diffusion coefcient of component A, m2 s1
model. At higher pressures hydroxyl and carbonate ions are EA enhancement factor due to the chemical reac-
severely depleted in the mass transfer zone, which is shown tion, -
in the last column of Table 2 by the ratio of the concentration F Faraday constant, 96485 C mol1
of hydroxyl ions at the interface and in the bulk liquid (at the k1,app apparent (pseudo) rst order reaction rate con-
end of a penetration time). A maximum relative deviation stant, s1
of 5% between the model and the experiments was found, kL liquid phase mass transfer coefcient, m s1
which means that the depletions of ions near the interface 
i ionic conductivity at innite dilution,
are described well. m2
1 mol1
m ratio of solubility in the liquid phase and in the
gas phase, mol m3 /(mol m3 )
6. Conclusions rA reaction rate of component A, mol m3 s1
R gas constant, 8.314 J mol1 K 1
The results of this study show that the presented equa- RA absorption rate of component A, mol s1
tions and physico-chemical parameters form a consistent t time, s
set to describe CO2 absorption in (0.5 M/0.5 M) carbon- T temperature, K
ate/bicarbonate solutions. The Danckwerts criterion as pre- x spatial coordinate, m
sented in Eq. (2) is justied and could be conned to: z ionic charge, -
 
1 2
mCO 2 cCO 2 (G) +
cCO 2 cH CO Greek letters

3

3

 D k
1 +
CO 2 1,app
1 < 0.1 W density of water, kg m3
kL2
electrostatic potential gradient, J C1 m1
All other equilibrium and kinetic rate constants are dened
to reduce the possible error in the ux calculation to less in the text.
than 3% when a pseudo-rst order reaction is assumed.
Furthermore, experimental validation of the used model Subscript and superscript
was performed in a stirred cell reactor up to CO2 partial
pressures of 16 bar. The criterion becomes more important bulk in the bulk liquid
at these high CO2 partial pressures, which emphasizes its int at the interface
signicance for industrially applied CO2 removal units op-
erating at elevated pressure.
References

Notation Benson, H.E., Field, J.H., Jimeson, R.M., 1954. CO2 absorption employing
hot carbonate solutions. Chemical Engineering Progress 50, 356363.
Cents, A.H.G., Brilman, D.W.F., Versteeg, G.F., 2001. Gas absorption in
Acell geometric surface area of the stirred cell, m2 an agitated gasliquidliquid system. Chemical Engineering Science
cA concentration of component A, mol m3 56, 10751083.
 A.H.G. Cents et al. / Chemical Engineering Science 60 (2005) 5830 5835
Danckwerts, P.V., 1950. Absorption by simultaneous diffusion and
chemical reaction. Transactions of the Faraday Society 46, 300304.
Danckwerts, P.V., Sharma, M.M., 1966. Absorption of carbon dioxide into
solutions of alkalis and amines. The Chemical Engineer CE, 244280.
Danckwerts, P.V., Kennedy, A.M., Roberts, D., 1963. Kinetics of CO2
absorption in alkaline solutions- II absorption in a packed column and
tests of surface renewal models. Chemical Engineering Science 18, 63
72.
Edwards, T.J., Maurer, G., Newman, J., Prausnitz, J.M., 1978. Vapor-
liquid equilibria in multicomponent aqueous solution of volatile weak
electrolytes. A.I.Ch.E. Journal 24, 966976.
Eigen, M., 1963. Protonenbertagung, sure-base-katalyse und
enzymatische hydrolyse. teil I: Elementarvorgnge. Angewandte
Chemie 75, 489508.
Higbie, R., 1935. The rate of absorption of a pure gas into a still liquid
during short periods of exposure. Transactions of the A.I.Ch.E. 31, 365.
Hikita, H., Asai, S., Takatsuka, T., 1976. Absorption of carbon dioxide
into aqueous sodium hydroxide and sodium bicarbonate solutions.
Chemical Engineering Journal 11, 131141.
5835
Horvath, A.L., 1985. Handbook of aqueous electrolyte solutions: physical
properties, estimation, and correlation methods. Chichester, Ellis
Horwood.
Newman, J.S., 1973. Electrochemical systems. Prentice-Hall, Englewood
Cliffs, NJ.
Pohorecki, R., Moniuk, W., 1988. Kinetics of reaction between carbon
dioxide and hydroxyl ions in aqueous electrolyte solutions. Chemical
Engineering Science 43, 16771684.
Roberts, D., Danckwerts, P.V., 1962. Kinetics of CO2 absorption in
alkaline solutions. I. Transient absorption rates and catalysis by arsenite.
Chemical Engineering Science 17, 961969.
Tsonopoulos, C., 1976. Ionization constants of water pollutants. Journal
of Chemical and Engineering Data 21, 190193.
Versteeg, G.F., Blauwhoff, P.M.M., Swaaij, W.P.M., 1987. The effect of
diffusivity on gasliquid mass transfer in stirred, vessels. Experiments
at atmospheric and elevated pressures. Chemical Engineering Science
42, 11031119.