A multifactor study of catalyzed hydrolysis of solid NaBH4 on cobalt nanoparticles: Thermodynamics and kinetics

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 3 4 ( 2 0 0 9 ) 9 3 8 – 9 5 1

Avai lab le at www.sc iencedi rect .com

journa l homepage : www.e lsev ie r . com/ loca te /he

A multifactor study of catalyzed hydrolysis of solid NaBH4

on cobalt nanoparticles: Thermodynamics and kinetics

Jerome Andrieuxa,*, Dariusz Swierczynskib, Laetitia Laversennea,Anthony Garronb, Simona Bennicib, Christelle Goutaudiera, Philippe Mielea,Aline Aurouxb, Bernard Bonnetota,1

aUniversite de Lyon, Laboratoire des Multimateriaux et Interfaces, CNRS UMR 5615, F-69622, Villeurbanne, FrancebUniversite de Lyon, Institut de Recherches sur la Catalyse et l’Environnement de Lyon, CNRS UMR 5256, F-69622, Villeurbanne, France

a r t i c l e i n f o

Article history:

Received 28 July 2008

Received in revised form

16 September 2008

Accepted 17 September 2008

Available online 30 November 2008

Keywords:

Hydrogen storage

Solid sodium borohydride

Hydrolysis reaction

Catalysis

Cobalt

Kinetics

Thermodynamics

* Corresponding author. Tel.: þ33 472431234;E-mail address: jerome.andrieux@univ-ly

1 Deceased December 2006.0360-3199/$ – see front matter ª 2008 Interndoi:10.1016/j.ijhydene.2008.09.102

a b s t r a c t

In the present work, hydrogen generation through hydrolysis of a NaBH4(s)/catalyst(s) solid

mixture was realized for the first time as a solid/liquid compact hydrogen storage system

using Co nanoparticles as a model catalyst. The performance of the system was analysed

from both the thermodynamic and kinetic points of view and compared with the classical

catalyzed hydrolysis of a NaBH4 solution. The kinetic analysis of the NaBH4(s)/catalyst(s)/

H2O(l) system shows that the reaction is first order with respect to the catalyst concen-

tration, and the activation energy equal to 35 kJ molNaBH4�1 . Additionally, calorimetric

measurements of the heat evolved during the hydrolysis of NaBH4 solutions evidence the

global process energy (�217 kJ molNaBH4�1 ). Characterization of the cobalt nanoparticles

before and after the hydrolysis associated with the calorimetric measurements suggests

the ‘‘in situ’’ formation of a catalytically active CoxB phase through ‘‘reduction’’ of an outer

protective oxide layer that is regenerated at the end of reaction.

ª 2008 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights

reserved.

1. Introduction the catalyzed hydrolysis of sodium borohydride solutions

Now that technologies to use hydrogen as a clean fuel are

readily available, like the Proton Exchange Membrane Fuel

Cell (PEMFC), and can be developed at an industrial scale,

research mainly focuses on the barrier of development which

is hydrogen storage for delayed use.

The potential applications of sodium borohydride have

been widely studied and have been recently reviewed by

Cakanyıldırım and Guru [1]. Most of the research efforts and

industrial devices developed up to now have been aimed at

fax: þ33 472440618.on1.fr (J. Andrieux).

ational Association for H

(stabilised with NaOH), according to the reaction (1)

BH�4ðaqÞ þ 4H2OðlÞ �!catBðOHÞ�4ðaqÞþ4H2ðgÞ (1)

Thehydrogenstoragecapacityofsodiumborohydridesolutions

depends on the quantity of water involved in the whole storage

system. Assuming a quantitative reaction, a standard commercial

solution containing 20 wt.% of NaBH4 allows a storage capacity of

only 4.2 wt.% H2. Moreover, in order to prevent self-hydrolysis of

the fuel, NaOH is usually added in a concentration of 1–5 wt.% that

further lowers the reactivity and the storage capacity.

ydrogen Energy. Published by Elsevier Ltd. All rights reserved.

mailto:[email protected]

http://www.elsevier.com/locate/he

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 3 4 ( 2 0 0 9 ) 9 3 8 – 9 5 1 939

Higher hydrogen storage capacities can be achieved by the

reaction of stoichiometric quantity of water with solid sodium

borohydride according to the Eq. (2) (NaBH4(s)/catalyst(s)/H2O(l)

system), which leads to the formation of solid hydrated

metaborates and the release of four moles of hydrogen [2,3].

NaBH4ðsÞ þ ð2þ xÞH2OðlÞ �!catNaBO2$xH2OðsÞ þ 4H2ðgÞ (2)

where x¼ 0, 2 and 4 [3].

The first advantage of using the solid NaBH4–liquid water

system is that both reagents are stored separately, thus

avoiding the problem of instability encountered for NaBH4

solutions. In this case, the storage capacity depends on the

borate hydration degree [3]. x¼ 0 leads to a theoretical amount

of generated H2 equivalent to 10.8 wt.% (117 kg H2 m�3)

considering the fuel (NaBH4þH2O).

If the water content in the NaBH4(s)/catalyst(s)/H2O(l) system

is high enough to dissolve the metaborate product at room

temperature (25 �C), the advantages of this configuration are

the possibility to store the reagents separately and to remove

the products of the reaction from the system in liquid form.

Schlesinger et al. were among the first to study NaBH4

hydrolysis [4] from stabilised NaBH4 solution. They have

shown that non-catalyzed sodium borohydride hydrolysis has

a very slow kinetics, making it inappropriate for a solution to

produce hydrogen on demand. A linear evolution of hydrogen

as a function of time is obtained at the beginning of the

hydrolysis. For the first step of the reaction, the activation

energy is found to be approximately 134 kJ molNaBH4�1 [5]. A

second step is then observed where the rate of hydrogen

release falls down rapidly to reach a plateau corresponding to

the end of the reaction. The shape of the experimental

hydrogen yield as a function of time has been clearly explained.

This behaviour could be explain by an increase of the pH of the

solution during the course of the reaction, due to the formation

of the strongly basic metaborate ion B(OH)4� [4,6,7].

The potential outlets of this technology have motivated

many studies on different catalysts with the aim to improve

the kinetics and the yield of NaBH4 hydrolysis.

A literature review concerning kinetic studies, focused on

transition metal catalysts, is summarized in Table 1 and Fig. 1.

It shows first of all the diversity of configurations tested by the

authors of these studies. The kinetic studies have been carried

out on aqueous solutions with NaBH4 concentrations varying

from 0.05 to 25 wt.%, stabilized by NaOH (0.1–10 wt.%) or non-

stabilized, using batch reactors. The activation energies and

the hydrogen generation rates reported by the authors are

presented in Table 1. Hydrolysis experiments have been

carried out in a temperature range from 10 to 50 �C on average,

pointing out a common temperature domain easy to realize for

industrial applications. Hydrogen generation rates have been

recalculated from the literature data in L min�1 gcatalyst�1 and

L min�1 gmetal�1 for comparison, and show the influence of the

support on the catalyst efficiency.

Noble metal-based catalysts under the form of slurry

dispersed in the solution have been the most widely studied.

Platinum supported on LiCoO2 presents the highest hydrogen

generation rate at room temperature [8]. Other noble metal-

based catalysts, stabilised ruthenium nanoclusters and

palladium supported on activated carbon, show interesting

properties: activation energies of 28.5 and 28 kJ molNaBH4�1 ,

respectively, and hydrogen generation rates at room temper-

ature of 3.65 and 0.5 L min�1 gmetal�1 , respectively [9,10].

As indicated in Table 1 the quantity of catalyst used in most

studies is quite high with respect to the quantity of NaBH4.

Studies have thus also focused on less expensive catalysts

with the aim to get the same catalytic activity as noble metal

catalysts. Non-noble metal catalysts have been intensively

studied in the form of metal salts [2,4,11], Raney metals [11,12]

or metal borides [11–16]. The first studies have shown the poor

efficiency of iron or copper as catalysts in the hydrolysis of

NaBH4, and reveal the better activity of nickel and cobalt.

Kaufman and co-workers found an activation energy of

71 kJ molNaBH4�1 for bulk nickel catalyst, and 63 kJ molNaBH4

�1 for

Raney nickel [12]. Nickel based catalysts are more efficient

under micrometric or nanometric sizes, which demonstrates

the key influence of the catalyst surface area/volume ratio. In

order to maximize this parameter, Metin and co-workers have

synthesized water-dispersible nickel (0) nanoclusters [17]

which were found to be highly active catalysts even at room

temperature, with an activation energy of 54 kJ molNaBH4�1 .

Cobalt as a catalyst for the hydrolysis of sodium borohydride

was first studied by Kaufman and co-workers [12]. They

found for this catalyst a quite high activation energy of

75 kJ molNaBH4�1 , but provided no details on the catalyst form or

morphology. Liu and co-workers have studied different cobalt-

based catalysts [11]. A commercially available fine powder of

cobalt showed a good catalytic activity with an activation

energy of 62.7 kJ molNaBH4�1 . Ye and co-workers have compared

different supported cobalt catalysts [18]. They found that a g-

Al2O3 supported cobalt catalyst was very efficient, with an

activation energy of 32.63 kJ molNaBH4�1 and a rate of hydrogen

generation of 1.15 L min�1 gmetal�1 . As shown in Fig. 1, this result

is close to the best results obtained for noble metal catalysts

[9,10]. Recent studies have demonstrated also the efficiency of

CoxB phase in the catalysis of NaBH4 hydrolysis. A rate of

2.32 L min�1 gmetal�1 was measured with CoxB supported on

activated carbon [19] while a rate of 11 L min�1 gmetal�1 was

obtained for CoxB/Ni foam [20]. Finally, recent promising

works have shown the efficiency of CoxB doped with other

metals. Pd–CoxB catalyst has been reported by Liang et al.

with the hydrogen generation rate of 2.87 L min�1 gmetal�1 [21].

Dai et al. have reported excellent performances of

W–CoxB catalysts [22] with a hydrogen generation rate

of 15 L min�1 gmetal�1 at room temperature and an activation

energy of 29 kJ molNaBH4�1 , equal to the best result obtained with

noble metal catalysts.

Finally, the CoxB phase has been claimed to be the active

phase in the hydrolysis [11,12,18–20,23,25]. However, in spite

of the abundance of the literature studies on Co-based cata-

lysts, few details are given neither on the possible catalytic

mechanism that occurs during the hydrolysis nor on the

identification of the active phase.

For this purpose, we have started a detailed study on

a nano-dispersed cobalt-based catalyst.

2. Experimental

Experiments were carried out with commercial sodium

borohydride (Acros organics, 98% purity, powder, average

Table 1 – Comparison of chosen kinetic studies of catalyzed NaBH4 hydrolysis from the literature and this work

Nature ofthe catalyst

Form Hydrolysis configurationa Quantity ofcatalyst used

(wt.% of NaBH4)

Activationenergy

(kJ molNaBH4�1 )

Temperaturerange (�C)

rcatb

(L min�1 gcatalyst�1 )

(20 �C< T< 25 �C)

rmetalc

(L.min�1.gmetal�1 )

(20 �C < T< 25 �C)

Reference

NaBH4 (wt%) NaOH(wt.%)

N.CPt/LiCoO2

(1.5 wt.% Pt)

LiCoO2

supported metal20 10

4.3 (Pt/LiCoO2)

0.06 (Pt)N.C N.C 3.9 260.4 [8]

Ru Nanoclusters 0.57 0 0.7 28.5 30–45 3.65 3.65 [9]

Ru/IRA400

(5 wt.% Ru)

IRA 400 resin


4.3 (Ru/IRA400)

0.2 (Ru)47 25–55 0.199 3.98 [24]

Ru/IRA400

(5 wt.% Ru)

IRA 400 resin

supported metal7.5 1

11 (Ru/IRA400)

0.5 (Ru)56 0–40 0.378 7.56 [7]

Pd/C

(10 wt.% Pd)

Activated Carbon

supported metal0.05 0.1

307 (Pd/C)

30.7 (Pd)28 N.C. 0.05 0.5 [10]

Co Powder 0.4 8 13 75 0–35 0.373 0.373 [12]

Co Powder 1 10 250 41.9 10–50 0.126 0.126 [11]

Co Raney form 1 10 50 53.7 10–30 0.267 0.267 [11]

Co/Al2O3

(9 w.t%)

g-Al2O3

supported metal5 5

73 (Co/Al2O3)

6.6 (Co)32.63 30–50 0.103 1.15 [18]

Co/C (9 wt.%)Active carbon

supported metal5 5

73 (Co/C)

6.6 (Co)45.64 30–50 0.018 0.2 [18]

Co–B Powder 20 5 N.C. (0.05 g) 64.87 10–30 0.875 0.875 [23]

Co–B/C

(30 wt.%)

Active carbon

supported metal0.76 8

26.7 (Co–B/C)

8 (Co–B)57.8 25–40 0.53 2.32 [19]

(continued on next page)

in

te

rn

at

io

na

ljo

ur

na

lo

fh

yd

ro

ge

ne

ne

rg

y3

4(2

00

9)

93

8–

95

19

40

Table 1 (continued )

Nature ofthe catalyst

Form Hydrolysis configurationa Quantity ofcatalyst used

(wt.% of NaBH4)

Activationenergy

(kJ molNaBH4�1 )

Temperaturerange (�C)

rcatb

(L min�1 gcatalyst�1 )

(20 �C< T< 25 �C)

rmetalc

(L.min�1.gmetal�1 )

(20 �C < T< 25 �C)

Reference

NaBH4 (wt%) NaOH(wt.%)

Co–B/Ni Foam

(50 wt.%)

Ni foam

supported metal25 3

60 (Co–B/NiFoam)

0.3 (Co–B)45 20–40 0.55 1.11 [25]

Co–B/Ni

Foam (N.C.)

Ni foam


N.C. (Co–B/NiFoam)

2.56 (Co-B)33 25–45 N.C 11 (30 �C) [20]

Co–W–B/Ni

Foam (N.C.)

Ni foam

supported metal20 5

N.C. (Co–W–B/NiFoam)

2.25 (Co–W–B)29 25–45 N.C 15 (30 �C) [22]

N.C. Co–B/Pd–NiFS

(N.C.)

Pd modified

Ni foam supported

metal

20 4 N.C. (32 mg Co–B) N.C N.C N.C 2.87 (30 �C) [21]

Ni Powder 0.4 8 15.8 71 0–35 0.114 0.114 [12]

Ni Powder 1 10 250 62.7 10–50 0.02 0.02 [11]

Ni Nanoclusters 0.57 0 1.45 54 25–45 5 5 [17]

Ni Raney form 1 10 50 50.7 10–30 0.23 0.23 [11]

Ni Raney form 0.4 8 7.9 63 0–35 0.156 0.156 [12]

Ni–B Powder 1,5 10 33.3 38 20–60 0.155 0.155 [14]

Co Powder w10 nm 19 0 10 35 40–80 N.C N.C This work

N.C.¼non-communicated.

a The wt.% refers to the component content in aqueous solution.

b rcat¼Hydrogen generation specific rate based on catalyst (metalþ support) mass (L min�1 gcatalyst�1 ).

c rmetal¼Hydrogen generation specific rate based on metal mass (L min�1 gmetal�1 ).

in

te

rn

at

io

na

ljo

ur

na

lo

fh

yd

ro

ge

ne

ne

rg

y3

4(2

00

9)

93

8–

95

19

41

Fig. 1 – Comparison of activation energies obtained for

different catalysts in previous studies (4 activation energy

from the present work).


particle size: 200 mm). Prior to the experiments the powder

was pre-treated at 180 �C for 2 h under primary vacuum

(w10�2 mbar) in order to avoid the presence of hydrated

species (NaBH4$2H2O). Commercial sodium hydroxide (Sigma–

Aldrich, 98%) was used to stabilize aqueous solutions of

sodium borohydride.

Cobalt nanoparticles from Strem chemicals (ref. 27-0020)

were used as catalyst in this study. These nanoparticles have

a very homogeneous particle size distribution with an average

diameter of 10–12 nm, and are covered with a protective oxide

layer [26]. The catalyst was mixed mechanically with the pre-

treated NaBH4 powder in a mortar under Ar atmosphere prior

to the experiments of catalyzed hydrolysis of solid NaBH4.

Hydrolyses have been carried out with 1, 5, and 10 wt.% of

cobalt with respect to sodium borohydride. The fresh Co

nanoparticles will be referred to as ‘‘nCoF’’. In order to under-

stand the evolution of the catalyst during hydrolysis of sodium

borohydride, the catalyst has been characterized after a stan-

dard test with 10 wt.% of catalyst at 30 �C. The mass of pre-

catalyzed mixture was 3 g and the hydrolysis was carried out in

a 250 mL flask connected to a reflux condenser to avoid water

loss during the hydrolysis. After test, the nanoparticles were

washed five times with distilled/deoxygenated water, and then

dried under vacuum. These nanoparticles after hydrolysis test

will be named ‘‘nCoT’’. Finally, as received Co nanoparticles

(nCoF) have been oxidized in air in ambient conditions for 50

days in order to study the influence of the oxide layer. These

nanoparticles after oxidation will be referred to as ‘‘nCoFox’’.

2.1. Calorimetric measurements for catalyzed hydrolysisof NaBH4 solution

A Differential Reaction Calorimeter (DRC, SETARAM) has been

used to quantify the heat effects and to follow the hydrolysis

reaction by measurement of the total heat evolved at 30 �C. A

detailed description of the experimental set-up was previ-

ously described [27]. The volume of released gas has been

measured using a gas-meter coupled to the DRC. Details

concerning the use of DRC system are reported by Nogent

et al. [28]. The common experimental protocol was the

following. Firstly 40 mg of catalyst was suspended in 20 mL of

1 M NaOH solution under nitrogen flow directly in the DRC

system. The catalyst was then activated by addition of 10 mL

of stabilised NaBH4 solution containing 212 mg (2 wt.%,

5.36 mmol) of sodium borohydride and 4 wt.% of NaOH. The

catalytic–calorimetric tests were performed after generation

of the active phase by four successive additions of 10 mL of the

stabilized solution of NaBH4.

There are very scarce literature data concerning experi-

mental calorimetric study of the reaction of borohydride

hydrolysis. Davis et al. [29] measured the heat of reaction for

sodium borohydride hydrolysis with hydrochloric acid giving

the value of 267 kJ molNaBH4�1 . More recently Zhang et al. [30]

measured the enthalpy of reaction of �212.1 kJ molNaBH4�1 cor-

responding to the global reaction (3).

The standard-state enthalpy (4) change for sodium boro-

hydride hydrolysis in aqueous solution (4) can be calculated

from standard formation enthalpies at 25 �C: 48.2 kJ mol�1

(BH4�), �285.8 kJ mol�1 (H2O), �1345.5 kJ mol�1 (B(OH)4

�) [31].

BH�4ðaqÞ þ 4H2OðlÞ /DrH

�

BðOHÞ�4ðaqÞþ4H2ðgÞ (3)

DrH� ¼ �250:5 kJ mol�1

NaBH4(4)

2.2. Catalyzed hydrolysis of solid NaBH4/catalystmixture

Hydrolysis experiments were performed in a 20 mL test-tube

closed by a silicon stopper and placed in a thermostatic bath.

No stirring is used in this configuration. For the determination

of kinetic parameters the overall system can be considered as

a non-steady state slurry batch reactor. The H2O/NaBH4 molar

ratio was equal to 9 (19 wt.% of sodium borohydride) for all the

tests. Prior to the experiment the mixture of catalyst and

NaBH4 was charged into the reactor in a glove box under argon

atmosphere. Distilled water was purged with argon prior to

use in order to remove oxygen. In a typical experiment, 500 ml

of water was injected into the reactor with a needle placed

directly inside the bed of the pre-catalyzed mixture containing

110 mg of NaBH4. Automated burette was used to control the

quantity of water. A second needle evacuated generated gases

to the outlet tube connected to an inverted, water filled,

graduated cylinder, situated in a water filled tank. Hydrogen

generated volumes were measured as a function of time, by

video monitoring water displaced from the graduated cylinder

as the reaction proceeded. Thermal probe placed inside the

NaBH4 bed measured the temperature during the reaction. As

a consequence reactor’s temperature evolution can be recor-

ded in relation with the hydrogen generated, as a function of

time.

The hydrogen generation yield is defined as the ratio of the

experimental hydrogen generated volume to the theoretical

one (the latter being a function of the initial number of

reagent’s moles). Kinetic parameters were based on the

measurement of the rate of hydrogen generated, which is

defined as the slope in the catalytic step of the hydrolysis. For


an easier comparison with the literature data, the reaction

rate will be expressed as hydrogen generation rate ‘‘r’’ (mL s�1)

while, taking into account the amount of catalyst (cobalt)

involved, the reaction rate will be defined as hydrogen genera-

tion specific rate based on metal mass ‘‘rCo’’ (L min�1 gCo�1).

The kinetic characterization of the catalyzed NaBH4

hydrolysis provides the reaction orders with respect to the

catalyst and the reagents in the reaction, and to the activation

energy Ea. Following the reaction (5), the rate law can be

expressed as follows

�4$d½NaBH4�

dt¼ d½H2�

dt¼ k$½Co�a$½NaBH4�b$½H2O�d (5)

where a, b, d are reaction orders and k the rate constant.

Previous studies have shown that the NaBH4 hydrolysis is

a zero order reaction with respect to both the reagents NaBH4

and H2O, for a catalyzed or a non-catalyzed reaction [12,14]. If

the ratio NaBH4/H2O is kept constant, Eq. (5) can be simplified

and leads to the following equation which allows the deter-

mination of the reaction order a with respect to the quantity of

catalyst used.

�4$d½NaBH4�

dt¼ d½H2�

dt¼ k$½Co�a (6)

Moreover, assuming that hydrolysis is a thermally activated

phenomenon, the values of the rate constant k obtained for

different temperatures can be fitted by the Arrhenius law,

where Ea is the activation energy (kJ molNaBH4�1 ), A the pre-

exponential factor, and R the ideal gas constant (Eq. (7)):

k ¼ A$exp

��Ea

RT

�(7)

2.3. Catalyst characterization

2.3.1. Chemical analysisChemical composition was determined by inductively

coupled plasma atomic emission spectroscopy (ICP-AES) with

a Flamme Perkin–Elmer M 1100 after solubilisation of the

samples in H2SO4:HNO3:HCl solutions.

2.3.2. X-ray photoelectron spectroscopyThe XPS experiments were carried out with a KRATOS AXIS

Ultra DLD spectrometer using a hemispherical analyzer and

working under vacuum with a pressure lower than 10�9 mbar.

Data were acquired using a monochromatic source of Al Ka X-

rays (1486.6 eV, 150 W), a pass energy of 40 eV, and a hybrid

lens mode. The analysed area is 700 mm� 300 mm. Charge

neutralization was required for all samples. The peaks were

referenced to the C–(C,H) components of the C1s band at

284.6 eV.

2.3.3. X-ray diffraction (XRD)The crystalline phases were determined by X-ray diffraction

using a Bruker D5005 powder diffractometer where the

sample is fixed while the X-ray tube (Cu Ka1þa2;

l¼ 0.154184 nm) and the detector rotate. X-ray diffraction

patterns were recorded between 5� and 105� (2q) with a step

size of 0.02� and an acquisition time of 8 s/step. In order to

increase the signal/background ratio, a zero background

holder was used (made from a commercial semiconductor

grade of silicon wafer grown and cut along the [100]-axis, i.e.,

Si(100)).

2.3.4. Magnetic measurementsMagnetic measurements were performed on the Co nano-

particles by the Weiss extraction method in an electromagnet

providing fields up to 21 kOe (2.1 T) at 25 �C. The amount of

metallic cobalt present in the sample was determined by

comparing the saturation magnetization thus obtained with

the specific saturation magnetization of bulk cobalt [32].

2.3.5. High-resolution transmission electronmicroscopy (HRTEM)High-resolution transmission electron microscopy (HRTEM)

was performed with a 200 kV TOPCON 002B microscope, with

a point resolution of 0.195 nm. All samples were ultrasonically

dispersed in hexane at room temperature. A drop of this

suspension was then deposited on the grid and allowed to dry

in air. The particle size was determined from the TEM images.

2.3.6. Fourier Transformed InfraRed (FT-IR)Fourier Transformed InfraRed (FT-IR) spectra were recorded

on a Nicolet 560 spectrometer equipped with DTGS/CsI

detector by transmission through a KBr pellet containing 1

wt.% of Co nanoparticles. KBr powder, pre-heated at 140 �C

was used to collect background. Prior to pellet preparation, the

KBr powder and the nanoparticles were mixed together by

mechanical grinding in a glove box (1 wt.% of cobalt nano-

particles). Thirty-two scans were acquired at a 4 cm�1 reso-

lution for each sample.

3. Results

3.1. Calorimetric measurements for catalyzed hydrolysisof NaBH4 solution

Catalyzed hydrolysis of NaBH4 solution has been performed

using the cobalt nanoparticles as catalyst in order to compare

with the literature data and to be used as a reference for the

catalyzed hydrolysis of solid NaBH4/catalyst mixture. The DRC

coupled with a gas-meter has been used for simultaneous

measurement of the reaction heat and the evolved hydrogen.

Catalytic tests have been performed by subsequent additions

of NaBH4 solution (2 wt.%) in order to evaluate the catalyst

stability [23]. In order to avoid the precipitation of NaBO2$4H2O,

low concentration NaBH4 solutions were used. As the catalytic

activity is not related to the NaBH4 concentration, the results

can be compared with other studies carried out at higher

concentrations.

The measurement of the heat evolved during a catalytic

reaction is important from both the practical and funda-

mental point of view, giving the thermal risk assessment and

the thermodynamic data for the determination of reaction

mechanism.

The heat measured with the DRC set-up results from the

contribution of all heat effects in the reacting system. Calo-

rimetric tests performed using low concentration NaBH4

solutions permit to verify the presence of other heat effects

Table 2 – Hydrogen generation rates and experimentalreaction enthalpies determined by DRC measurementsfor the commercial Co nanoparticles as received (nCoF),oxidized in air at ambient conditions during 50 days(nCoFox) and reduced by NaBH4 and then filtered (nCoT)

Number ofinjectiona

nCoF nCoFox nCoT

rCob DHexp

c rCob DHexp

c rCob DHexp

c

1 0.78 �246 0.43 �320 0.47 �246

2 1.15 �209 0.74 �221 0.68 �246

3 1.07 �215 0.79 �211 0.75 �219

4 1.11 �210 0.79 �214 0.85 �221

5 0.90 �223 0.80 �215 0.86 �219

Average (2–5) 1.06 �214 0.78 �215 0.79 �219

a ((2 wt.% NaBH4 4 wt.% (1 M) NaOH)) ((L min�1 gCo�1)).

b Hydrogen generation specific rate based on Co mass, ‘‘rCo’’

(L min�1 gCo�1).

c Experimental reaction enthalpy, ‘‘DHexp’’ ((kJ molNaBH4�1 )).


that can be related with the catalyst evolution, especially with

the reduction/oxidation phenomena [27].

In order to determine the activity and stability of the

catalyst, cobalt nanoparticles have been studied in three

different initial states. That is, (1) as received commercial Co

nanoparticles (nCoF), (2) Co nanoparticles oxidized in air at

ambient conditions during 50 days (nCoFox) and, (3) reduced

by NaBH4 and then filtered (nCoT ). Fig. 2 shows the typical

evolution of H2 volume as a function of time for five subse-

quent injections of NaBH4 solution into the calorimetric vessel

containing the nCoF catalyst. A summary of the hydrogen

generation rates and the measured reaction enthalpies for all

the tests are presented in Table 2.

We observe that for each type of sample the first injection

of stabilized sodium borohydride solutions results in lower

hydrogen generation rates and higher exothermicity of the

reaction compared to the four subsequent additions. We

assume that the first addition could be attributed to the

reduction of an outer oxide layer present on the Co nano-

particles and secondly to the formation of an active phase.

Both the hydrogen generation rate and the evolved energy

tend to have similar values after successive injections. In the

case of the nCoF catalyst we have observed an increase of the

hydrogen generation rate up to 1.06 L min�1 gCo�1. This activity

is comparable with that generally observed for Co-based

catalyst [23].

The evolved energies for the last injections are similar in

the three cases with a mean value between 214 and

219 kJ molNaBH4�1 . These results differ from the theoretical value

of 250 kJ molNaBH4�1 . It is interesting to observe that the first

injection leads to higher energies compared to the following

ones, showing that additional reactions impact the enthalpy

measurement. These reactions might correspond to the

formation of a cobalt boride phase as suggested by Bonne-

mann [33]. In the case of the oxidized catalyst, the high value

of the evolved energy (�320 kJ molNaBH4�1 ) observed for the first

injection can be explained by the reduction of the outer layer.

From the analysis of catalytic performance of the Co

nanoparticles with the NaBH4 stabilized solution using the

Fig. 2 – Evolution of generated H2 volume vs. time for

catalyzed hydrolysis of NaBH4 solution using the nCoF

catalyst at 30 8C.

‘’multi-addition’’ calorimetric method we can conclude the

following:

(1) The active phase is formed in situ, most probably by

formation of cobalt boride species [4].

(2) The activity of the catalyst after ‘‘activation’’ in terms of

hydrogengenerationrate isclose to1 L min�1 gCo�1 atT¼ 30 �C

in the specific experimental conditions of this work.

(3) The average measured energy, evolved during the reac-

tion, is �217 kJ molNaBH4�1 .

Results of catalyst performances in the stabilized solution

of NaBH4 will stand as a reference during the analysis of the

more complex NaBH4(s)/catalyst(s)/H2O(l) system.

3.2. Kinetics of catalyzed hydrolysis of solid NaBH4/catalyst mixture (NaBH4(s)/catalyst(s)/H2O(l) system)

In order to establish the rate law for catalyzed hydrolysis of

NaBH4 using cobalt nanoparticles (nCoF), two sets of experi-

ments were performed at a molar ratio of H2O/NaBH4¼ 9.

In a first set of experiments, the reaction temperature was

varied between 40 and 80 �C with a 10 �C step with a fixed

catalyst content (10 wt.%) (Fig. 3a). Reactor temperature,

measured by a thermal probe was quasi-constant in the range

from 30% to 80% of the hydrogen generated (Fig. 3b). The

measured temperatures are taken into account in the kinetic

calculations (Treactor) and have been used to determine the

activation energy.

In the second set of experiments (Fig. 5a), the temperature

was kept constant at 80 �C and the quantity of cobalt nano-

particles (nCoF) with respect to NaBH4 was varied from 1 to 10

wt.% (0.84–8.4 wt.% of Co according to the chemical analysis of

the nCoF; see Section 3.3). Those results allow to obtain the

reaction order with respect to the catalyst.

The hydrogen release rates were determined by linear

regression of the experimental points corresponding to the

range from 30% to 80% of the hydrogen generation yield,

corresponding to the catalytic step. Linear regression coeffi-

cients are upper than 0.95.

Fig. 3 – (a) Hydrogen generated as a function of time for several temperatures. Quantity of catalyst (nCoF) was held constant

at 10 wt.% and molar ratio H2O/NaBH4 [ 9. (b) Reactor temperature as a function of time. (c) Zoom of the initial reaction

period of (a). (d) Hydrogen generation rates, r and rCo as a function of the reactor temperature ( : hydrogen generation rate,

‘‘r’’ in mL sL1; : hydrogen generation specific rate based on metal mass, ‘‘rCo’’ in L minL1 gCoL1).


3.2.1. Activation energyInfluence of the temperature on the kinetics of catalyzed

NaBH4 hydrolysis is shown in Fig. 3a. The evolution of the

hydrogen generated versus time can be decomposed in

several steps. Step I corresponds to an induction time where

no hydrogen generation is detected, i.e.: the hydrogen

generation rate is very small comparable to the hydrogen

generation rate of the non-catalyzed hydrolysis. Fig 3c pres-

ents a zoom of the initial reaction period from Fig. 3a. Taking

into account that the hydrogen measurement is performed

every 5 s, this step is only observable for the ‘‘low’’ tempera-

tures (first 30 s at 40 and 50 �C in Fig. 3c). Our results show that

this induction time is temperature dependent. Step II corre-

sponds to the initial hydrogen release up to 30–40% of the

total hydrogen generated. During this step the hydrogen

generation rate is variable and characterized by occurrence of

periods where r diminishes or tends to 0, observed in Fig. 3a, c

as plateau at lower temperatures (40–60 �C). While the

temperature increases, the duration of this plateau decreases

with an increasing slope of the curve (see Fig. 3a). Step III is

defined as the catalytic step. We can observe that the

hydrogen generation rate during step III increases with the

temperature. It also increases linearly with time except for

40 �C, temperature at which a change of the slope occurred;

finally, step IV is defined by a decrease of the hydrogen

generation rate at the end of the reaction. Fig. 3b plotted in

parallel to Fig. 3a presents the evolution of the reactor

temperature as a function of time during the hydrolysis. An

increase of the reactor temperature during the steps I and II

can be observed. Step III is characterized by a quasi-constant

temperature of the reactor and the step IV by a decrease to

reach the temperature of thermostated bath. From these

results, an average reactor temperature is calculated for each

experiment, taken as an average of various temperatures of

the plateau, during the catalytic step (step III). These average

temperatures are needed to determine the activation energy

of the system. From step III of Fig. 3a, the hydrogen generation

rate was determined by linear regression of the curve. Rates (r

and rCo) as a function of the reactor temperature are plotted in

Fig. 3c. Both these rates linearly increase with the reactor

temperature.

The plot of the hydrogen generation rate (r) as a function of

the reciprocal absolute reactor temperature (1/Treactor) leads to

a straight line with a slope of �4286.3 K (Fig. 4).

Fig. 5 – (a) Hydrogen generated as a function of time for

non-catalyzed hydrolysis, and hydrolyses carried out with

1 wt.%, 5 wt.% and 10 wt.% of catalyst (nCoF) at 80 8C and

molar ratio and H2O/NaBH4 [ 9. (b) Hydrogen generation

rates, r and rCo as a function of the quantity of cobalt ( :

hydrogen generation rate, ‘‘r’’ in mL sL1; : hydrogen

generation specific rate based on metal mass, ‘‘rCo’’ in

L minL1 gCoL1).

Fig. 4 – The Arrhenius plot for NaBH4(s)/catalyst(s)/H2O(l)

system with 10 wt.% of catalyst in the NaBH4(s)/catalyst(s)

mixture and a molar ratio H2O/NaBH4 [ 9.

Fig. 6 – The graph of ln (r) vs. ln (molCo) at 80 8C and H2O/

NaBH4 [ 9.


This value is closely related to the activation energyby Eq. (8):

�4286:3 ¼ �Ea

R(8)

As a consequence, the apparent activation energy, for the

cobalt nanoparticles used (nCoF ), is about 35.6 kJ molNaBH4�1 .

3.2.2. Reaction order with respect to the catalystFig. 5a shows the percentage of hydrogen generated as

a function of time for different quantities of cobalt nano-

particles. Compared to the non-catalyzed hydrolysis, the

cobalt nanoparticles, nCoF, present a catalytic activity in the

hydrolysis of NaBH4.

The time dependence of hydrogen generated is similar for

different quantity of catalysts. Compared to the lower

temperature study (Fig. 3a), at 80 �C, only two steps are clearly

defined in Fig. 5a. This behaviour is also observed with NaBH4

solution hydrolysis (liquid system). At 80 �C the dissolution

and diffusion time of NaBH4 can be considered as negligible.

As a consequence, steps I and II described previously cannot

be detected. Step III proceeds as the hydrolysis is catalyzed by

cobalt nanoparticles nCoF. Step IV is reached close to the end

of the reaction with a decrease of the hydrogen generation

rate until a plateau indicating the end of the reaction. From

step III of Fig. 5a, we determine the hydrogen generation rate

(r) by linear regression of the curve. Fig. 5b shows the

hydrogen generation rate (r) versus the quantity of catalyst

used. This rate increases with the Co quantity. Hydrogen

generation specific rate based on metal mass (rCo) is also

reported in Fig. 5b. It presents a high value for the lowest

quantity of catalyst used (1 wt.%), about 35 L min�1 gCo�1, but

decreases and remains constant for higher quantity of cata-

lyst (5 and 10 wt.%), equals to 25 L min�1 gCo�1.

As presented in Fig. 6, the plot of the hydrogen generation

rate (r) as a function of cobalt concentration, both in loga-

rithmic scale, leads to a straight line by fitting the points. The

slope of this line is found to be 0.83, which indicates a pseudo

reaction order with respect to the concentration of cobalt.

Fig. 7 – X-ray diffractograms for the Co nanoparticles (a) as

received (nCoF) and (b) reduced by NaBH4 and then filtered

(nCoT).


3.3. Characterization of the catalyst

According to Bonnemann et al. [26] the Co nanoparticles used

are composed of a nanocrystallized metallic cobalt bulk,

coated by a protective oxide layer. In order to verify the nature

of this outer shell and its evolution during hydrolysis reaction,

the nanoparticles have been characterized in their initial state

(nCoF) and after hydrolysis test (nCoT). The bulk properties

were analysed by chemical analysis, XRD, TEM and magnetic

measurements. The surface layer was analysed using XPS, FT-

IR spectroscopy and TEM. Results are summarized in Table 3.

The nCoF sample contains 84.0 wt.% of Co. An important

quantity of carbon (6.0 wt.%) and about 0.9 wt.% of Al indicate

the presence of other phases than metallic Co. In the nCoT

sample the quantity of Co decreases to 76.5 wt.%. A decrease

of carbon content to 2.0 wt.% and Al to 0.2 wt.% can be also

observed. The sample after hydrolysis contains 0.8 wt.% of

incorporated boron. In order to obtain information about the

bulk crystalline phases and the particle diameter, X-ray

diffraction experiments have been performed. The XRD

patterns for the nCoF and nCoT samples are shown in Fig. 7. For

both samples the main diffraction line at 44.341� 2q corre-

sponds to the cubic (FCC) phase of Co. From the full-width at

half-maximum of this peak, the Sherrer equation allows to

determine a mean crystallite size of 2–3 nm. For the nCoT

sample we notice additionally the presence of large diffraction

lines of an amorphous or poorly crystallized phase around 37�

and 59� 2q which can be attributed to a cubic CoO phase (40

wt.%). Based on XRD experiments shown in Fig. 7, Riedvelt

refinement has been carried out and the result are reported in

Table 3. The morphology of Co nanoparticles has been

Table 3 – Comparison of the properties of the Conanoparticles as received (nCoF) and reduced by NaBH4

and then filtered (nCoT)

nCoF nCoT

Bulk properties

Co (wt.%) 84.0 76.5

Al (wt.%) 0.9 0.2

C (wt.%) 6.0 2.0

B (wt.%) 0.0 0.8

Quantity of metallic

Co (wt.%)

(magnetism)

65 45

Particle size

TEM d¼ 10–15 nm Aggregates

100–200 nm

Magnetism <12 nm <12 nm

XRD Cubic Co 64%, 3 nm 60% metallic Co

Hexagonal

Co 36%, 2 nm

10% hexagonal

Co of 3 nm

40% oxide CoO

of 2 nm.

Surface properties

XPS Co(0)/Co(II)¼ 50%/50%

surface

oxide layer with a

thickness< 2 nm

45% of C

Co(0)/Co(II)¼ 30%/70%

surface

oxide layer with

a thickness w3 nm

60% of C

FT-IR Carbonate No carbonate

examined by TEM. The micrograph of the nCoF sample (Fig. 8a)

shows a very homogeneous particle size distribution with

a mean size of 10–12 nm. After exposition to air during 50 days

(nCoFox), high-resolution images permit to evidence the

presence of a slightly thicker surface oxide layer (Fig. 8b).

The TEM micrographs for sample nCoT after hydrolysis are

given in Fig. 8c. We can observe agglomerates of particles

coated by a badly crystallized phase that could be attributed to

CoO as detected by XRD. In order to quantify the metallic Co

phase the specific magnetization of the particles was inves-

tigated by Weiss extraction method. The results are shown in

Fig. 9. The specific saturation magnetization (Ms) of the initial

nanoparticles (nCoF) was 104 uemcgs/g representing about

65% of the Co bulk value (Ms for bulk Co is 160 uemcgs/g). For

the sample after hydrolysis (nCoT) the specific saturation

magnetization diminished to 70 uemcgs/g corresponding to

about 45% of metallic Co. The comparison with Co content as

determined by AES–ICP, equals to 85 wt.%, permits to estimate

that 20 and 40 wt.% of Co is present within the oxide shell in

the initial and after hydrolysis samples, respectively. The

surface composition and the oxidation state of the elements

have been determined by XPS analysis. The Co 2p3/2 spectra

for the nCoF and nCoT cobalt nanoparticles are presented in

Fig. 10. The binding energy values are 777.9 eV for the Co(0)

and 781.0 eV for Co(II). For the sample before hydrolysis (nCoF),

the Co 2p3/2 peak presents two oxidation states, 0 and II. The

corresponding intensities are 50%/50%, which indicate that

the cobalt particles are covered with a surface oxide layer with

a thickness lower than 2 nm (considering the hypothesis of

a homogeneous surface layer). After the hydrolysis reaction

(nCoT), the proportions Co(0)/Co(II) are 30%/70%, which

correspond to a thicker surface layer (about 3 nm). Both

samples contain also a quite important proportion of carbon,

about 45% for the nCoF and 60% for the nCoT. The sample after

hydrolysis contained traces of sodium (<1%). However, no

boron was detected contrary to the results of chemical anal-

ysis, indicating that B should be probably present under the

surface layer in the nCoT sample. In order to study the nature

Fig. 8 – TEM of Co nanoparticles (a) as received (nCoF), (b)

oxidized in air at ambient conditions during 50 days

(nCoFox), (c) reduced by NaBH4 and then filtered (nCoT).

Fig. 9 – Specific magnetization curves of the Co

nanoparticles before hydrolysis (nCoF) and after hydrolysis

(nCoT) (T [ 25 8C).

Fig. 10 – XPS Co 2p3/2 spectra of Co nanoparticles (a) as

received (nCoF) and (b) reduced by NaBH4 and then filtered

(nCoT).


of the chemical bonds at the surface of the nanoparticles, both

samples were investigated by Fourier Transformed Infrared

spectroscopy. Infrared spectra for the pure KBr (background),

the nanoparticles as received (nCoF) and reduced by NaBH4

(nCoT) are shown in Fig. 11. Absorption peaks around

3400 cm�1 (hydroxyl group contribution), and 2400 cm�1

(carbon dioxide contribution) are attributed to impurities

contained in the KBr powder. Peaks around 2900, 1400, 800 and

600 cm�1 are found in the nCoF spectrum. A peak around

2900 cm�1 reveals carbon hydrogen stretching, according to

organometallic or surfactant compounds used during the

synthesis of the cobalt nanoparticles [26]. The other peaks can

be attributed to a carbonate phase [34]. The strong, charac-

teristic absorption bands are found around 1420 cm�1,

850 cm�1 and 700 cm�1. The two lower frequency bands are

very sharp. All of the band positions are somewhat dependent

on the hydration state of the sample which can explain the

difference observed. No characteristic peak is revealed in the

nCoT spectrum. These experiments point out the reduction of

cobalt carbonate during the hydrolysis. Table 3 presents

Fig. 11 – FT-IR spectra of KBr powder, nCoF, and nCoT.


a comparison of the properties of Co nanoparticles before and

after hydrolysis as characterized by the different physico-

chemical techniques.

4. Discussion

4.1. Temperature evolution during the reaction

The measured apparent activation energy should be consid-

ered with care. In fact, the exothermicity of the reaction

inducing local hot points, according to the high specific heat

capacity of cobalt, the temperature on the active site may be

much higher than the external temperature of the thermo-

stated bath. Fig. 3c points out this phenomenon. Reactor

temperature is plotted as a function of time, in parallel to

hydrogen released (Fig. 3b). An increase of the reactor

temperature after the beginning of hydrogen generation

(corresponding to the water injection) can be noticed for each

experiment. DT is found to increase with the experiment

temperature. Maximum DT is obtained for experiment carried

out at 80 �C with an increase of the reactor temperature of

4 �C. This phenomenon is due to the exothermic reaction of H2

release [4]. More recently, Zhao and co-workers [19] claimed

that the lower the ratio H2O/NaBH4 is, the higher the increase

of the reactor temperature will be. We conclude on an average

increase of temperature of 2–3 �C for our hydrolysis conditions

(H2O/NaBH4¼ 9; 19 wt.% NaBH4). Moreover, during catalytic

step (range from 30% to 80% of the hydrogen generated),

isothermal conditions can be assumed as the reactor

temperature is quasi-constant. We deduced from this step the

reactor temperature used for activation energy calculation.

4.2. Discussion about kinetics

Activation energies have been obtained in previous studies for

different catalysts (Fig. 1 and Table 1).

The difficulty of the comparison is due to the large variety

of configurations chosen by the experimenters. Comparison

can be done between the activation energies keeping in mind

the quantity of catalyst used in the experiments. This

parameter found for the cobalt nanoparticles nCoF

(35.6 kJ molNaBH4�1 ) is one of the lowest values obtained. This

result strives towards best results obtained for noble metal

catalysts, where the lowest activation energy is found to be

28.5 kJ mol�1 [9]. Two explanations can be proposed to

understand this behaviour. Firstly, catalyst nCoF is used under

nanometric form. Secondly, the active catalytic phase is

formed in situ during the hydrolysis of sodium borohydride.

Finally, the interaction nanoparticles/substrate is one of the

fundamental parameters we must consider to obtain transi-

tion metal catalyst as efficient as noble metal catalysts

[18,20,22].

The variation in the reaction order, with respect to the

catalyst quantity, of 0.83 compared to theoretical 1 might be

due to the modification of the surface catalyst by decompo-

sition of carbonate and formation of oxide. The influence of

the ratio of catalyst with respect to NaBH4 is described in

Fig. 5a. In those conditions, the available quantity of NaBH4

used to reduce the ‘‘carbonate-oxide’’ layer differs when the

quantity of catalyst is 1 wt.% or 10 wt.%. As a consequence, the

quantity of active phase formed during step II with respect to

the total Co content (i.e. the ratio CoxB/Co) is more important

for 1 wt.% of catalyst. This phenomenon can be illustrated in

Fig. 5b: hydrogen generation rate, rCo, is equal to 35 L min�1 gCo�1

(high ‘‘active phase/Co ratio’’) whereas it decreases to

25 L min�1 gCo�1 for higher catalyst quantities (lower CoxB/Co

ratio). The reaction order thus determined can be considered

as a pseudo reaction order, including a partial reaction order

of 1 (according to kinetics behaviour obtained for other

heterogeneous catalysts [7,9,17]) and a partial reaction order

lower than 0.83 linked to secondary reactions described

previously.

4.3. Catalyst evolution

Characterizations performed on the nanoparticles before and

after hydrolysis allow to propose a possible catalyst evolution

during the hydrolysis reaction. Initially metallic Co particles

are coated with a protective layer made of cobalt carbonate

with a w2 nm thickness. This layer is removed by contact with

sodium borohydride solution to form the catalytically active

phase. The ‘‘reduction’’ of the outer shell results probably in

the formation of CoxB active phase on the particles surface,

which is stable under reducing conditions in the presence of

BH4� ions [4]. The active phase formation step results in the

lower initial hydrogen generation rate observed for the first

injection of NaBH4 solution in the ‘‘liquid’’ hydrolysis system

and at low temperatures in the ‘‘solid’’ hydrolysis system (step

II). The presence of Boron in the nCoT sample indicates also

a possible formation of CoxB active phase. At the end of the

hydrolysis reaction, cobalt-based active phase is not stable in

the basic media leading to reoxidation of CoxB and formation

of a cobalt oxide (CoO) layer observed by XPS, XRD and TEM.

The fact that no B and only Co0 and Co2þwere detected by XPS

may be explained by partial oxidation of the CoxB phase to

form Co0 and B2O3. The Co0 can be subsequently oxidized to

CoO and B2O3 removed from the surface by dissolution. The

Co2B phase is covered by the oxidized Co layer at the end of

the reaction and thus is not detectable by XPS surface

analysis.

Fig. 12 – Proposition of the reaction mechanism for the NaBH4(s)/catalyst(s)/H2O(l) system.


4.4. Proposition of the reaction mechanism for theNaBH4(s)/catalyst(s)/H2O(l) system

General mechanism ofcatalysis was proposed by Kojima and co-

workers [8].Thishydrolysisoccurs inthreedifferentsteps.ForPt/

LiCoO2, these authors propose an adsorption of borohydride on

the noble metal (which has a high electronic density favouring

the bondingformationwith the vacancies of boron) parallel to an

adsorption of water on the oxide support. Reaction of adsorbed

species occurs during the second step. Then products are des-

orbed in the last step. This detailed mechanism is now well

accepted for the catalyzed hydrolysis of sodium borohydride.

In the present work, the catalyst was mixed mechanically

with the NaBH4 powder prior to the addition of water. In this

configuration the reaction phases can be considered as pre-

sented in Fig. 12. We suppose that the reaction rate is suffi-

ciently high and does not depend on the NaBH4 concentration

during phases 1–6: zero order with respect to concentration of

NaBH4 [7]. At the end of the reaction, the NaBH4 concentration

is very low, leading to a limit of diffusion. This phenomenon

explains, with an increase of the pH solution, the shape of the

hydrogen generation curve (step IV of Figs. 3a and 5a). The

‘‘induction time’’ (step I) observed in Fig. 3a corresponds to an

overlapping of phases 1 and 2, and decreases with an increase

of reaction temperature as the dissolution of NaBH4 is

temperature dependent. It should also be noted that, in spite of

the thermostatic conditions, the real temperature on the

catalyst surface is much higher than the bulk temperature due

to the high exothermicity of the hydrolysis reaction. The

temperature dependent plateau pointed out in Fig. 3a (step II) is

clearly linked to phase 3. The oxide/carbonate layer coating

nCoF is reduced by BH4� ions adsorbed at the surface during step

II. Step III is associated to hydrolysis reaction under the influ-

ence of the catalytic active phase and is composed by a sum up

of phases 4–6. Finally, reoxidation of the catalyst at the same

time as the pH of the solution increases explains the decreasing

of the reaction rate at high concentration of reaction product

leading to a plateau at the end of the reaction (phases 7–9).

5. Conclusion

We have shown the feasibility of an efficient hydrogen

generation in a system where a powder mixture of NaBH4(s)/

catalyst(s) is hydrolysed by direct water addition. This system,

in comparison with the catalyzed hydrolysis of aqueous NaBH4

solution, presents the advantage of a long-term stability due to

the separate storage of the two reactants. The use of cobalt

nanoparticles as a model catalyst made it possible to identify

the main steps and parameters of the reaction mechanism. A

kinetic study with cobalt nanoparticles evidenced an apparent

activation energy of 35 kJ molNaBH4�1 and an order of reaction of

one with respect to the catalyst. The physico-chemical char-

acterizations of the catalyst before and after hydrolysis led to

the understanding of the catalytic phenomena that occur

during the reaction. Firstly, as evidenced by liquid calorimetry,

an initial ‘‘in situ’’ activation of the catalyst is needed. As

a matter of fact, reduction of the carbonate (oxide) layer was

the initial step preceding the catalyzed hydrolysis reaction.

Intermediate cobalt boride species may constitute the catalytic

active phase during the hydrolysis. Those results point out the

critical importance of the initial surface composition in the

case of cobalt-based catalysts.

A detailed elucidation of the catalytic step mechanism

requires the identification of the active phase, the formation

of which is one of the limiting steps of the hydrolysis system.

Acknowledgements

We dedicate this paper to the memory of Dr. Bernard Bonnetot

who initiated this research work. We thank A. Brioude for

performing the TEM experiments, C. Guimon for performing

and interpreting XPS experiments, G. Bergeret for carrying out

XRD and Rietveltd refinement for the obtained diffractograms

and J.-A. Dalmon and E. Landrivon for their kind help with the

magnetic measurements.Theauthors acknowledge theFrench

National Research Agency ANR for the financial support.

r e f e r e n c e s

[1] Cakanyıldırım C, Guru M. Hydrogen cycle with sodiumborohydride. Int J Hydrogen Energy 2008;. doi:10.1016/j.ijhydene.2008.05.084C.

[2] Brown HC, Brown CA. New, highly active metal catalysts forthe hydrolysis of borohydride. J Am Chem Soc 1962;84:1493–4.


[3] Marrero-Alfonso EY, Gray JR, Davis TA, Matthews MA.Minimizing water utilization in hydrolysis of sodiumborohydride: the role of sodium metaborate hydrates. Int JHydrogen Energy 2007;32:4723–30.

[4] Schlesinger HI, Brown HC, Finholt AE, Gilbreath JR,Hockstra HR, Hyde EK. Sodium borohydride, its hydrolysisand its use as a reducing agent and in the generation ofhydrogen. J Am Chem Soc 1953;75:215–9.

[5] Gardiner JA, Collat JW. Kinetics of the stepwise hydrolysis oftetrahydroborate ion. J Am Chem Soc 1965;87:1692–700.

[6] Kilpatrick M, McKinney CD. Kinetics of the reaction oflithium borohydride in aqueous acid solution. J Am ChemSoc 1950;72:5474–6.

[7] Amendola SC, Sharp-Goldman SL, Janjua MS, Spencer NC,Kelly MT, Petillo PJ, et al. A safe, portable, hydrogen gasgenerator using aqueous borohydride solution and Rucatalyst. Int J Hydrogen Energy 2000;25:969–75.

[8] Kojima Y, Suzuki K, Fukumoto K, Sakaki M, Yamamoto T,Kawai Y, et al. Hydrogen generation using sodiumborohydride solution and metal catalyst coated on metaloxide. Int J Hydrogen Energy 2002;27:1029–34.

[9] Ozkar S, Zahmakiran M. Hydrogen generation fromhydrolysis of sodium borohydride using Ru(0) nanoclustersas catalyst. J Alloys Compd 2005;404–406:728–31.

[10] Guella G, Zanchetta C, Patton B, Miotello A. New insights onthe mechanism of Palladium-catalyzed hydrolysis of sodiumborohydride from 11B NMR measurements. J Phys Chem B2006;110:17024–33.

[11] Liu BH, Li ZP, Suda S. Nickel- and cobalt-based catalysts forhydrogen generation by hydrolysis of borohydride. J AlloysCompd 2006;415:288–93.

[12] Kaufman CM, Sen B. Hydrogen generation by hydrolysis ofsodium tetrahydroborate : effects of acids and transitionmetals and their salts. J Chem Soc Dalton Trans 1985;2:307–13.

[13] Paul R, Buisson P, Joseph N. Catalytic activity of nickelborides. Ind Eng Chem 1952;44:1006.

[14] Hua D, Hanxi Y, Xingping A, Chuansin C. Hydrogenproduction from catalytic hydrolysis of sodium borohydridesolution using nickel boride catalyst. Int J Hydrogen Energy2003;28:1095–100.

[15] Wu C, Wu F, Bai Y, Yi BL, Zhang HM. Cobalt boride catalystsfor hydrogen generation from alkaline NaBH4 solution.Mater Lett. 2005;59:1748–51.

[16] Patel N, Guella G, Kale A, Miotello A, Patton B, Zanchetta C,et al. Thin films of Co–B prepared by pulsed laser depositionas efficient catalysts in hydrogen producing reactions. ApplCatal A: Gen 2007;323:18–24.

[17] Metin O, Ozkar S. Hydrogen generation from the hydrolysisof sodium borohydride by usingwater dispersible,hydrogenphosphate-stabilized nickel(0) nanoclusters ascatalyst. Int J Hydrogen Energy 2007;32:1707–15.

[18] Ye W, Zhang H, Xu D, Ma L, Yi B. Hydrogen generationutilizing alkaline sodium borohydride solution andsupported cobalt catalyst. J Power Sources 2007;164:544–8.

[19] Zhao J, Ma H, Chen J. Improved hydrogen generation fromalkaline NaBH4 solution using carbon-supported Co-B ascatalysts. Int J Hydrogen Energy 2007;32:4711–6.

[20] Dai HB, Liang Y, Wang P, Cheng HM. Amorphous cobalt–boron/nickel foam as an effective catalyst for hydrogengeneration from alkaline sodium borohydride solution.J Power Sources 2008;177:17–23.

[21] Liang J, Li Y, Huang Y, Yang J, Tang H, Wei Z, et al. Sodiumborohydride hydrolysis on highly efficient Co–B/Pd catalysts.Int J Hydrogen Energy 2008;. doi:10.1016/j.ijhydene.2008.05.082.

[22] Dai HB, Liang Y, Wang P, Yao XD, Rufford T, Lu M, et al. High-performance cobalt–tungsten–boron catalyst supported onNi foam for hydrogen generation from alkaline sodiumborohydride solution. Int J Hydrogen Energy 2008;. doi:10.1016/j.ijhydene.2008.05.080.

[23] Jeong SU, Kim RK, Cho EA, Kim HJ, Nam SW, Oh IH, et al. Astudy on hydrogen generation from NaBH4 solution using thehigh-performance Co-B catalyst. J Power Sources 2005;144:129–34.

[24] Amendola SC, Sharp-Goldman SL, Janjua MS, Kelly MT,Petillo PJ, Binder M. An ultrasafe hydrogen generator:aqueous, alkaline borohydride solutions and Ru catalyst.J Power Sources 2000;85:186–9.

[25] Lee J, Kong KY, Jung CR, Cho E, Yoon SP, Han J, et al. Astructured Co–B catalyst for hydrogen extraction fromNaBH4 solution. Catal Today 2007;120:305–10.

[26] Bonnemann H, Brijoux W, Brinkmann R, Matoussevitch N,Waldofner N, Palina N. A size-selective synthesis of air stablecolloidal magnetic cobalt nanoparticles. Inorg Chim Acta2003;350:617–24.

[27] Garron A, Swierczynski D, Bennici S, Auroux A. Newinsights into the mechanism of H2 generation throughNaBH4 hydrolysis on Co-based nanocatalysts studied bydifferential reaction calorimetry. Int J Hydrogen Energy 2008(accepted).

[28] Nogent H, Le Tacon X, Vincent L, Sbirrazzuoli N. DRC signaltreatment for heat flow and reagents accumulationdetermination. J Loss Prevention Process Ind 2005;18:43–4.

[29] Davis WD, Mason LS, Stegeman G. The heats of formation ofsodium borohydride, lithium borohydride and lithiumaluminum hydride. J Am Chem Soc 1949;71:2775–81.

[30] Zhang J, Fisher TS, Gore JP, Hazra D, Ramachandran PV. Heatof reaction measurements of sodium borohydridealcoholysis and hydrolysis. Int J Hydrogen Energy 2006;31:2292–8.

[31] Lide DR. In: CRC handbook of chemistry and physics. CRC;2000.

[32] Barbier A, Hanif A, Dalmon J-A, Martin GA. Preparation andcharacterization of well-dispersed and stable Co/SiO2catalysts using the ammonia method. Appl Catal A: Gen1998;168:333.

[33] Bonnemann H, Richard RM. Nanoscopic metal particles –synthetic methods and potential applications. Eur J InorgChem 2001:2045.

[34] Coblentz Society, Inc (Coblentz No. 508). Evaluated infraredreference spectra. In: Linstrom PJ, Mallard WG, editors.WebBook of chemistry NIST, vol. 69. Gaithersburg MD, 20899:National Institute of Standards and Technology, http://webbook.nist.gov; June 2005.

http://webbook.nist.gov

http://webbook.nist.gov

A multifactor study of catalyzed hydrolysis of solid NaBH4 on cobalt nanoparticles: Thermodynamics and kinetics

Documents