i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 8 7 e9 3

Available online at www.sciencedirect.com

ScienceDirect

journal homepage: www.elsevier.com/locate/he

Short Communication

Modification of the shrinking core model for

hydrogen generation by reaction of aluminum
particles with water

S.S. Razavi-Tousi*, J.A. Szpunar

Department of Mechanical Engineering, University of Saskatchewan, S7N 5A9, Saskatoon, Saskatchewan, Canada

article info abstract

Article history: The assumptions of the traditional shrinking core model cannot be adopted for the core
Received 2 July 2015 eshell structure formed during the reaction of aluminum particles and water. The model
Received in revised form assumes that diffusion coefficient is constant and the particle size is unchanged during the
8 September 2015 reaction. This study modifies the traditional shrinking core model by including the effect of
Accepted 18 November 2015 the growing size of the particles and the changes in porosity of the hydroxide layer on
Available online 21 December 2015 hydrogen generation. The modified model is fitted to the experimental data and the results
are compared to that of the traditional model. The flux of water molecules in the hydroxide
Keywords: layer is calculated using the parameters obtained by the hydrogen generation data.
Hydrogen Copyright © 2015, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights
Controlling step reserved.
Diminishing core
Diffusion
Pores

temperature of water. The hydroxide species formed as

Introduction
Aluminumewater reaction
Researchers have extensively investigated the reaction of
aluminum particles with water for the purpose of hydrogen
generation [1e9]. Because aluminum particles are usually
covered by a thin native oxide film, the reaction does not start
immediately after immersion of particles in water. The thin
oxide film should be hydrated/dissolved in water and then
reduction of water molecules by water initiates. The rate and
byproducts of the reaction strongly depends on pH and
byproduct is soluble in solutions with a pH higher than 11 or
lower than 2 [10,11]. However, within a pH range of 2e11, a
hydroxide layer precipitates on the aluminum surface, which
makes the diffusion of ion species/water molecules a critical
step in determining the kinetics of hydrogen generation.
At the presence of a hydroxide layer on the surface of
aluminum particles, which results in a coreeshell structure
[12], three parameters can control the rate of the reaction: 1)
diffusion of ion species/water molecules in the solution to the
particles surface, 2) diffusion of ion species/water molecules
from the particle surface to the substrate aluminum and vice
versa, 3) the reaction between aluminum and water. If one

* Corresponding author. Tel.: þ1 (306) 290 9084.

E-mail addresses: ser105@mail.usask.ca, s.razavitousi@gmail.com (S.S. Razavi-Tousi).
http://dx.doi.org/10.1016/j.ijhydene.2015.11.080
0360-3199/Copyright © 2015, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.
88 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 8 7 e9 3
assumes that an abundant amount of ion species/water
molecules is available in the solution, the diffusion in the
solution is not the controlling step and thus either the
chemical reaction or diffusion in hydroxide layer controls the
kinetics. In a case where a chemical reaction is the controlling
step, the reaction rate is explained by the “diminishing
sphere” model. However the diffusion of ion species is the
controlling step when a hydroxide layer is present and the
reaction rate is explained by the “shrinking core” model.
Few studies attempted to fit either the diminishing sphere
or shrinking core model to the measured hydrogen evolution
data from the reaction of aluminum and water [13e16]. How-
ever, the results of fittings cannot be considered satisfactory
because of the considerable deviation of the theoretical values
from experimental data. A. I. Ratko et al. [17] developed a
diffusion based model for the reaction of aluminum and water,
which later was used by H. Nie et al. [18]. Although this model
was developed considering the mechanism of the reaction, H.
Nie et al. concluded that the diffusion coefficient changes with
the time of the reaction, which indicated “inadequacy of the
simplified model” [18]. Another approach to model the kinetics
of reaction of solid particles with a liquid is to use empirical
formula [19e21]. Although the numerical models result in a
good fitting to the experimental data, they do not provide any
insight to the mechanism of the reaction.
New assumptions for the shrinking core model
In this study, considering the growth of the hydroxide film on
aluminum particles during the reaction, we attempt to modify
the shrinking core model. There are two main differences
between the assumptions of the shrinking-core model and the
proposed model in explaining the growth of hydroxide layer.
The first difference is that the diffusion coefficient of the shell
(hydroxide layer) formed on the core (aluminum) in our model
changes with time. Researchers found that the grown hy-
droxide layer is not uniform across its thickness [22e25]. This
was explained by density of the formed hydroxide that
changes during the reaction. R.S. Alwitt claimed that when the
reaction starts, the formed hydroxide phase is permeable to
water [26]. As reaction continues, the hydroxide densifies and
pores become filled, therefore, water penetration rate through
the shell reduces. The observations of pores in the hydroxide
layer, the hydroxide densification during the reaction and the
resulting change in the reaction rate are discussed in our
previous work [27]. The change in the porosity of the shell and
its influence on the reaction rate have to be incorporated into
formulating the kinetics of the reaction.
The second difference between the assumptions of our
model and the shrinking-core model concerns the thickness
of the formed shell around the particle. The shrinking-core
model was developed for the leaching process [28], in which
the thickness of the shell is the same as that of the consumed
core. However for the reaction of aluminum particles with
water, considering the difference between the density of the
produced hydroxide layer and the consumed aluminum core,
the thickness of the hydroxide layer is bigger than that of the
consumed core.
The main body of the model that we developed in this
study is similar to the traditional shrinking core model. The
above mentioned assumptions will be applied in this model
and the results will be compared with the experimental data.
A comparison will be made between the modified model and
traditional shrinking core model, and the rate of mass transfer
of species/water molecules will be calculated based on the
modified model.
Experimental procedure
Aluminum powder of 99.8% purity (MFCD00134029) provided
by Alfa Aeasar Company was mixed with 3 wt% of Stearic acid
(97% purity, Fisher Scientific-AC17449-0010) as the process
control agent (PCA), and the mixture was activated by ball
milling in a planetary ball mill (Torrey HillseND2L) for 19 h in
an argon atmosphere.
The procedure for measurement of hydrogen production
rate is similar to our previous work [29]. We added 2 g
aluminum powder into an Erlenmeyer flask contained 200 ml
distilled water at temperature of 80 ± 3  C with a constant
stirring rate of 120 rpm. The produced hydrogen gas passed
through a desiccant (CoCl2) to absorb moisture, and then an
ADM2000 flowmeter measured the flow with an accuracy of
0.1 ml/min. The flowmeter was connected to a computer and
the ADM Trend software acquired the data. We measured the
flow for the first hour of the reaction for each sample and then
normalized the results for 1 g aluminum powder. The
hydrogen yield curves vs. time were obtained by integrating
the flow data over time. We did not continue measuring
hydrogen flow after 1 h of the reaction because the collected
data was sufficient for comparison with the models.
Development of the model
The schematic diagram of the model is presented in Fig. 1. The
particle on the right represents the situation before the reac-
tion, and the particle on the left shows how the reaction forms
the coreeshell structure. One can see that in the traditional
shrinking core model the overall size of the particle is un-
changed during the reaction, however, in our model the size of
the particle increases and the radius of the particle changes
from Ro to R. The assumption of a coreeshell structure was
adopted based on our SEM observations of the cross section of
reacted aluminum particles [12].
Water reaches to the core of the particle and reacts with
aluminum. The mass transfer equation at a given diameter
within the hydroxide layer can be expressed as:
fH2 Otowards the core g  fH2 Oconsumed by reaction g þ fH2 Ogenerated by reaction g
¼ fH2 Oaccumulated g
(1)
H2 Ogenerated by reaction ¼ H2 Oaccumulated ¼ 0 (2)
Therefore;
H2 Otowards the core ¼ H2 Oconsumed by reaction (3)
Eq. (3) means that the amount of water that diffuses to-
ward the core should be the same as the amount of water
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 8 7 e9 3 89

consumed by the reaction to generate hydrogen. This equa- dNA dCA

tion relates the rate of consumption of aluminum core to the  ¼ 4pr2 DA (8)
dt dr
diffusion equation of water in the shell.
Integrating Eq. (8) over the thickness of the shell and con-
The first steps of the new model are the same as those
centration of A atoms from the surface of the particle to the
already presented in the literature [28]. We write the balance
core gives:
of reaction as Eq. (4):


dNA 1 1
AðfluidÞ þ bBðsolidÞ ¼ CðsolidÞ þ dDðfluidÞ (4)   ¼ 4pDA CAO (9)
dt Rc R
The rate at which A atoms (water atoms) react is deter-
where CAO is the concentration of A at the surface of the
mined by the flux of A atoms directed to the core. Flux by
particle.
definition is the number of moles (or mass) that passes an area
In this step, the first modification to the model will be
(S) over time (t). At any given radius r (see Fig. 1) during the
applied. We need to calculate how the size of a particle
steady state, the flux of water directed towards the core can be
changes when the reaction is taking place. The overall volume
written as:
of the particle (shell þ core) can be written as:
dN
Q¼ (5) VR ¼ VRc þ Vx (10)
S  dt
where N, Q, and S are the number of moles, flux and surface where VR, VRC and Vx are respectively the overall volume of
area, respectively. the particle, volume of the remained core and volume of the
Considering the spherical shape of the particle, S can be produced shell on the core.
replaced by 4pr2 . Moreover, based on Eq. (3), the flux of water If “n” is the ratio of volume of the produced shell to volume
toward core is equal to the consumption (consumption has of consumed core, one can write:
the negative sign) of water by the reaction, therefore: Vx ¼ nðVRo  VRc Þ (11)
dNA where VRO is the volume the original particle.
 ¼ 4pr2 QA ¼ 4pR2 QAs ¼ 4pR2c QAc ¼ constan t (6)
dt Therefore, considering Eqs. (10) and (11) one can write:
where QAS and QAC are the flux at the surface of the particle
VR ¼ VRc ð1  nÞ þ nVR0 (12)
and the boundary of the core, respectively. That means that
the amount of water that enters the shell is consumed at the Eq. (12) can be then written as:
coreeshell boundary by the reaction with the aluminum core.
4 3 4 4
The flux of fluid can be written by Fick's law: pR ¼ pR3c ð1  nÞ þ n  pR3o (13)
3 3 3
dCA and therefore:
QA ¼ DA (7)
dr
where CA is the concentration of A atoms. Combining Eq. (7)  1=3
R ¼ ð1  nÞR3c þ nR3o (14)
and Eq. (6) yields:

Fig. 1 e The schematic of the modified model; the unreacted particle is shown on right and the particle during the reaction is
shown on left.
90 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 8 7 e9 3

Eq. (14) predicts the overall radius of a particle with an

n þ ð1  nÞð1  XB Þ2=3  ½n þ ð1  nÞð1  XB Þ2=3
initial radius of RO, which after the reaction the radius of the
2bDo CAo ð1  nÞ
core reduced to RC. ¼ 2
lnðkt þ 1Þ (23)
krRo
Now that the equation for the changes in overall radius of
the particle is established, Eq. (9) can be modified by replacing Eq. (23) predicts the degree of conversion for a coreeshell
the R value from Eq. (14): structured particle with a changing size and diffusion coeffi-
! cient during the reaction period.
dNA 1 1
  ¼ 4pDA CAO (15)
dt Rc ð1  nÞR3 þ nR3 1=3
c o

Number of B moles in the core material (NB) is related to the Results and discussion
density of B atoms (rB: number of B moles in the unit volume
of the core material) through Eq. (16): Fitting the model

NB ¼ rB Vc (16) Eq. (23) was fitted to the experimental data using the Origin
software. r is the number of aluminum moles in unit volume
Considering Eq. (3), the amount of water that diffuses to-
of a particle, which equals to 0.1 mol/cm3. CAO is the con-
ward the aluminum core should be the same as the amount
centration of water molecules (the fluid A in Eq. (4)) at the
of water that is consumed by the reaction. The amount of
surface (distance ¼ 0) of the particle. 1 cm3 of water includes 1/
water consumed by the reaction is related to the amount of
18 mol of water, thus the concentration of water molecules
consumed aluminum by Eq. (4). Therefore, the consumption
equals 1/18 ¼ 0.056 mol/cm3. b is the ratio of aluminum to
rate of B atoms (atoms from aluminum core) can be written
water for the reaction (b ¼ 0.5) obtained from the condition of
as:
stoichiometry of the reaction of aluminum and water at 80  C:


4pR3c
dNB ¼ bdNA ¼ rB dVc ¼ rB d ¼ 4rB pR2c dRc (17)
3
Al þ 2H2O ¼ AlOOH þ 3/2H2 (24)
Therefore:
k and Do are obtained from the fitting to the experimental
4rB pR2c dRc
dNA ¼ (18) results. Based on Eq. (20), Do is the diffusion coefficient of
b
water (fluid) in the hydroxide layer (shell) at the time zero,
Replacing Eq. (18) in Eq. (15) gives: while the magnitude of k determines how fast diffusion co-
! efficient of the hydroxide layer decreases because of densifi-
4r pR2 dRc 1 1
 B c  ¼ 4pDA CAO dt (19) cation, and n is the ratio of volume of the produced shell to
b Rc ð1  nÞR3 þ nR3 1=3
c o volume of consumed core. Based on Eq. (24) and 1 mole of
In this step the next modification in the model is applied. aluminum hydroxide is produced by consumption of 1 mol of
Our SEM and electrochemical observations on reacted aluminum. Aluminum has a density of 2.7, and the calculated
aluminum particles revealed that not only the hydroxide layer density of boehmite is 3.03. Considering molar weights of 27 g
thickens by time, but also it densifies as reaction continues and 60 g for aluminum and boehmite respectively, the volume
[27]. Since the porosity of the hydroxide layer decreases over of produced hydroxide is two times larger than that of
time, the diffusion coefficient of water in the shell should consumed aluminum, thus n ¼ 2. Nevertheless, since the
change with time rather than being a constant value. We produced boehmite is porous, the volume of the produced
consider that DA changes by a hyperbolic function of time (Eq. hydroxide depends on the amount of porosity in the shell
(20)). In this equation, Do is the diffusion coefficient at the layer and that is bigger than in a pore-free boehmite. There-
time ¼ 0, k is a constant that determines the effect of time on fore, one can assume that n  2. Accordingly, instead of
the diffusion coefficient and þ1 is required to avoid an infinite considering n ¼ 2, we left the n value to be determined by
DA at the beginning of the reaction: fitting the model to the experimental data.
The result of the fitting is presented in Fig. 2. Moreover, the
Do fitting of the traditional shrinking core model (SCM) to the
DA ¼ (20)
kt þ 1 experimental data is also provided. Table 1 provides the pa-
And Eq. (19) turns to: rameters obtained from fitting the modified model to the
experimental data. One can see that the traditional model
"  2=3 #Rc
R2c ð1  nÞR3c þ nR3o bDo CAO results in a poor fit to the experimental data, while the
 ¼ lnðt þ 1Þ (21)
2 2ð1  nÞ krB modified model fits with an R-square of 0.994. The value of
Ro
n ¼ 2.8 was obtained from the fitting, which suggests that
Defining XB as the fractional conversion: because of presence of pores, the hydroxide layer formed on
the surface of the particle is thicker than it could be (n ¼ 2) in a

3 pore-free state.
Rc
1 ¼ XB (22) The reaction of aluminum and water is described in three
Ro
stages [27]; a) induction period, b) fast reaction and c) slow
Eq. (21) can be re-written as: reaction. The induction period is the time that the oxide layer
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 8 7 e9 3 91

reaction, the volume of the produced shell can be larger (n > 1)

or smaller (n < 1) than the volume of the consumed core. For a
given diffusion coefficient, the value of n determines the rate
of the reaction. If diffusion coefficient is fixed over time, Eq.
(19) is integrated with the value of DA that is constant. Thus,
the conversion equation will be:

2 "
3 #2=3
rc rc 2bDA CAO tð1  nÞ
n þ ð1  nÞ  n þ ð1  nÞ ¼
Ro Ro rB R2o
(25)

When conversion is completed, the core diameter is zero,

thus the complete conversion time can be calculated as;
 
n  n2=3 rB R2o
rc ¼ 0 ¼ > t ¼  (26)
2bDA CAO ð1  nÞ
Fig. 2 e The results of fitting of the traditional shrinking
and Eq. (25) turns to:
core model (SCM) and the modified model (Eq. (23)) on the
data obtained from reaction of 19 h milled aluminum  t
n þ ð1  nÞð1  XB Þ2=3  ½n þ ð1  nÞð1  XB Þ2=3 ¼ n  n2=3
powder and hot water. t
(27)

Fig. 3 shows the comparison of the normalized result of the

on the surface of aluminum is dissolved/hydrated by water
[11,26,30e32]. There is no reaction between aluminum and
water during the induction period and no hydrogen is gener-
ated, therefore this stage should be excluded from the fitting.
The second stage (fast reaction) occurs when a porous hy-
droxide layer forms on the surface of aluminum. Following
the fast reaction, the hydroxide layer is densified and becomes
less porous, which causes the decrease in the diffusion coef-
ficient and reducing the hydrogen generation rate. Contrary to
the traditional shrinking model where the diffusion coeffi-
cient is constant and different stages of the reaction cannot be
adopted, the current model allows formulating a transition
from the fast to slow reaction by the change in the diffusion
coefficient.
Comparison of the models
modified shrinking core model at constant DA (Eq. (27)),
traditional shrinking core model (Eq. (28)) and the reaction
control model (Eq. (29)) [28].
t
¼ 1  ð1  XB Þ1=3 (28) Está
t
trocado
t
¼ 1  3ð1  XB Þ2=3 þ 2ð1  XB Þ (29)
t
The X axis in Fig. 3 is t/t, which t is time and t is complete
conversion time. The Y axis is the fraction of conversion. For
the sake of comparison, different “n” values between 0.25 and
4 were applied. One can see that at the beginning of the re-
action, the rate of the reaction control model is slower
compared to the traditional and modified shrinking core
model. However, as the reaction continues, the reaction rate
of traditional and modified shrinking core model slows down
faster compared to the reaction control model. This change in

Formation of a shell around the reacting particle, which con-

trols the reaction rate, transforms the reaction control model
to the shrinking core model [28]. For a given constant diffusion
coefficient of the fluid in the shell, the rate of the reaction
depends on thickness of the shell. For leaching processes, the
overall diameter of the particles does not change during the
conversion, and the conversion follows the assumptions of
the traditional shrinking core model. For such a reaction in
that the overall diameter of the particles changes during the

Table 1 e Parameters obtained from fitting the modified

model (Eq. (23)) to the experimental data from hydrogen
production of aluminum powder samples reacting with
water.
R-Square Do k n
(m2/sec) (sec1)
19 h milled- as produced 0.994 3.39E-11 0.00336 2.83 Fig. 3 e Degree of conversion vs. normalized time for the
19 h milled-1 min annealed 0.987 3.55E-11 0.00368 4.51 traditional shrinking core model (SCM), reaction control
19 h milled- 3 h annealed 0.995 2.97E-11 0.00266 5.45
and the modified model (Eq. (27)).
92 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 8 7 e9 3
the reaction rate is because at the end of the reaction,
shrinking core model considers that the core particle is
covered by a thick shell that slows down the reaction rate.
Comparison of traditional and modified shrinking core model
reveals that for a given DA, the rate of the reaction for a par-
ticle with a shell depends on the ratio of volume of the pro-
duced shell compared to the consumed core. Depending on
“n”, the modified model predicts a faster or slower reaction
rate compared to the traditional shrinking core model. At
n ¼ 1, both the traditional and modified model predicts the
same reaction rate.
Rate of mass transfer
Based on the first Fick's law, flux of ion species (or water
molecule for this case) can be calculated if the diffusion co-
efficient, concentration of the ions and the thickness for
diffusion are known:
vC CAO  Cc
J ¼ D ¼ DA
vx xt  0
where D (diffusion coefficient) is replaced by DA (time
dependant coefficient, Eq. (20)), CAO is the concentration of
water molecules at the surface of the particle
(CAO ¼ 0.056 mol/cm3), Cc is the concentration of water mol-
ecules at the coreeshell interface (Cc ¼ 0), and xt is the
thickness of the hydroxide layer at the time ¼ t:
xt ¼ R  rc
The relation between the core radius and fractional con-
version can be written as:
rc ¼ Ro ð1  XB Þ1=3
And R is obtained from Eq. (14), therefore:
 1=3
xt ¼ nR3 þ R3o ð1  XB Þð1  nÞ  Ro ð1  XB Þ1=3
The relation between XB and time is obtained from Eq. (23).
As diffusion parameters (k and Do) can be obtained from
fitting the experimental results, the flux of water can be
plotted versus time of the reaction.
In order to study the dependence of flux on the reaction
time and microstructure of the sample, the modified model is
Fig. 4 e (a) The hydrogen generation vs. time for the as produced 19 h milled aluminum powder and the milled powders
annealed for 1 min and 3 h. (b) The corresponding calculated diffusion flux of water in the hydroxide layer.
fitted with the data from hydrogen production the 19 h milled
sample annealed for 1 min and 3 h at 550  C at argon atmo-
sphere from Ref. [29]. The modified model (Eq. (23)) fitted well
to the experimental data and the fitting parameters are re-
ported in Table 1. Using the fitting parameters, and incorpo-
rating Eq. (33) and Eq. (23) into Fick's law (Eq. (30)), effect of the
reaction time on the flux of diffusing water in different sam-
ples was obtained (Fig. 4(a) and (b)).
One can see that at the beginning of the reaction, there is a
significant difference between the flux of water molecules in
the hydroxide layer for different samples. However as the
reaction continues, the difference becomes negligible. This
implies that at the beginning of the reaction where the hy-
droxide layer is thin and porous, the affinity of the micro-
structure of aluminum to react with water determines flux of
water molecules. As the reaction continues, the effect of the
grown hydroxide layer prevails over the effect of microstruc-
ture and all samples have similar values of the flux. This result
obtained from the modified model is consistent with the
electrochemical findings explaining the effect of grain size of
(30)
aluminum on open circuit potential (OCP) of the reaction;
samples of different grain size show markedly different OCPs
at the beginning of the reaction while the potential becomes
similar for the different samples after 1 h of the reaction [27].
Similarly, the hydrogen production results obtained from the
as produced and aged aluminum powders confirm that the
microstructure of the aluminum has a considerable effect on
the hydrogen production rate at the beginning of the reaction,
(31) while this difference becomes negligible as the reaction con-
tinues and the hydroxide layer thickens on the aluminum
surface [29].
(32)
Conclusion
(33)
The shrinking core model was modified based on the as-
sumptions that are applicable to the reaction of aluminum
particles with hot water. The first modification was that it was
assumed that the size of the reacting particle changes during
the reaction because of the difference between the volume of
the consumed core and the produced shell. Accordingly,
depending on the volume of the produced shell, the diffusion
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 8 7 e9 3
distance for the ion species (water molecules for alumi-
numewater reaction) could be larger or smaller compared to
the traditional shrinking core model. The second modification
was introduced because densification of the boehmite during
the reaction decreases the volume of the pores in the hy-
droxide layer (shell), and as a result the overall diffusion co-
efficient is not constant over time. The initial diffusion
coefficient and the magnitude of the change in its value during
the reaction can be determined by parameters obtained from
fitting the proposed model to the experimental data. Finally,
flux of water molecules was calculated using fitting parame-
ters, which revealed that the significance of effect of micro-
structure on the reaction rate decreases as the hydroxide layer
on the aluminum particles thickens.
Acknowledgment
We acknowledge the NSERC discovery grant and support of
Canada Research Chairs program.
references
[1] Soler L, Macanas J, Munoz M, Casado J. Aluminum and
aluminum alloys as sources of hydrogen for fuel cell
applications. J Power Sources 2007;169:144e9.
[2] Elitzur S, Rosenband V, Gany A. Study of hydrogen
production and storage based on aluminumewater reaction.
Int J Hydrogen Energy 2014;39:6328e34.
[3] Chen X, Zhao Z, Liu X, Hao M, Chen A, Tang Z. Hydrogen
generation by the hydrolysis reaction of ball-milled
aluminiumelithium alloys. J Power Sources 2014;254:345e52.
[4] Jia Y, Shen J, Meng H, Dong Y, Chai Y, Wang N. Hydrogen
generation using a ball-milled Al/Ni/NaCl mixture. J Alloy
Compd 2014;588:259e64.
[5] Xu F, Sun L, Lan X, Chu H, Sun Y, Zhou H, et al. Mechanism of
fast hydrogen generation from pure water using AleSnCl2
and bi-doped AleSnCl2 composites. Int J Hydrogen Energy
2014;39:5514e21.
[6] Yavor Y, Goroshin S, Bergthorson JM, Frost DL. Comparative
reactivity of industrial metal powders with water for
hydrogen production. Int J Hydrogen Energy
2015;40:1026e36.
[7] Razavi-Tousi SS, Nematollahi GA, Ebadzadeh T, Szpunar JA.
Modifying aluminumewater reaction to generate nano-sized
aluminum hydroxide particles beside hydrogen. Powder
Technol 2013;241:166e73.
[8] Gai W-Z, Shi Y, Deng Z-Y, Zhou J-G. Clarification of activation
mechanism in oxide-modified aluminum. Int J Hydrogen
Energy 2015;40:12057e62.
[9] Yavor Y, Goroshin S, Bergthorson JM, Frost DL, Stowe R,
Ringuette S. Enhanced hydrogen generation from
aluminumewater reactions. Int J Hydrogen Energy
2013;38:14992e5002.
[10] Vedder W, Vermilye DA. Aluminumþwater reaction. T
Faraday Soc 1969;65:561.
[11] Vermilyea DA, Vedder W. Inhibition of the aluminumþwater
reaction. T Faraday Soc 1970;66:2644e54.
[12] Razavi-Tousi SS, Szpunar JA. Effect of structural evolution of
aluminum powder during ball milling on hydrogen
93
generation in aluminumewater reaction. Int J Hydrogen
Energy 2013;38:795e806.
[13] Tikhov S, Sadykov V, Ratko A, Kouznetsova T,
Romanenkov V, Eremenko S. Kinetics of aluminum powder
oxidation by water at 100  C. React Kinet Catal L
2007;92:83e8.
[14] Soler L, Candela AM, Macanas J, Munoz M, Casado J.
Hydrogen generation from water and aluminum promoted
by sodium stannate. Int J Hydrogen Energy 2010;35:1038e48.
[15] Soler L, Candela AM, Macanas J, Munoz M, Casado J. In situ
generation of hydrogen from water by aluminum corrosion
in solutions of sodium aluminate. J Power Sources
2009;192:21e6.
[16] Streletskii AN, Kolbanev IV, Borunova AB, Butyagin PY.
Mechanochemical activation of aluminum: 3. Kinetics of
interaction between aluminum and water. Colloid J
2005;67:631e7.
[17] Rat'ko AI, Romanenkov VE, Bolotnikova EV, Krupen'kina ZV.
Hydrothermal synthesis of porous Al2O3/Al metal ceramics:
II. Mechanism of formation of a porous Al(OH)3/Al
composite. Kinet Catal 2004;45:149e55.
[18] Nie H, Schoenitz M, Dreizin EL. Calorimetric investigation of
the aluminumewater reaction. Int J Hydrogen Energy
2012;37:11035e45.
[19] Wang W, Chen DM, Yang K. Investigation on microstructure
and hydrogen generation performance of Al-rich alloys. Int J
Hydrogen Energy 2010;35:12011e9.
[20] Chambon M, Abanades S, Flamant G. Kinetic investigation of
hydrogen generation from hydrolysis of SnO and Zn solar
nanopowders. Int J Hydrogen Energy 2009;34:5326e36.
[21] Funke HH, Diaz H, Liang X, Carney CS, Weimer AW, Li P.
Hydrogen generation by hydrolysis of zinc powder aerosol.
Int J Hydrogen Energy 2008;33:1127e34.
[22] Mori S, Draley JE. Oxide dissolution and its effect on the
corrosion of 1100 aluminum in water at 70  C. J Electrochem
Soc 1967;114:352e3.
[23] Draley JE, Mori S, Loess RE. The corrosion of 1100 aluminum
in oxygen-saturated water at 70  C. J Electrochem Soc
1963;110:622e7.
[24] Rider AN. The influence of porosity and morphology of
hydrated oxide films on epoxy-aluminium bond durability. J
Adhes Sci Technol 2001;15:395e422.
[25] Baker BR, Balser JD. Formation of films on aluminum by
reaction with water. Aluminum 1976;52:197e200.
[26] Alwitt RS. The growth of hydrous oxide films on aluminum. J
Electrochem Soc 1974;121:1322e8.
[27] Razavi-Tousi SS, Szpunar JA. Mechanism of corrosion of
activated aluminum particles by hot water. Electrochim Acta
2014;127:95e105.
[28] Levenspiel O. Chemical reaction engineering. New York John
Wiley & Sons; 1999.
[29] Razavi-Tousi SS, Szpunar JA. Role of ball milling of
aluminum powders in promotion of aluminum-water
reaction to generate hydrogen. Metall Mater Trans E
2014;1:247e56.
[30] Bernard WJ, Randall JJ. An investigation of the reaction
between aluminum and water. J Electrochem Soc
1960;107:483e7.
[31] Bunker BC, Nelson GC, Zavadil KR, Barbour JC, Wall FD,
Sullivan JP, et al. Hydration of passive oxide films on
aluminum. J Phys Chem B 2002;106:4705e13.
[32] Stralin A, Hjertberg T. Influence of surface-composition on
initial hydration of aluminum in boiling water. Appl Surf Sci
1994;74:263e75.