- Email: [email protected]
Hydrogen production via hydrolysis reaction from ball-milled Mg-based materials
International Journal of Hydrogen Energy 31 (2006) 109 – 119 www.elsevier.com/locate/ijhydene
Hydrogen production via hydrolysis reaction from ball-milled Mg-based materials M.-H. Grosjeana , M. Zidounea , L. Rouéa,∗ , J.-Y. Huotb,1 a INRS-Énergies, Matériaux et Télécommunications, 1650 boulevard Lionel Boulet, Varennes (QC), Canada J3X 1S2 b HERA Hydrogen Storage Systems Inc, 577 rue Le Breton, Longueuil (QC), Canada J4G 1R9
Available online 3 March 2005
Abstract Hydrogen generation by hydrolysis of Mg and MgH2 has been investigated in pure water and 1 M KCl. It has been found that hydrolysis reaction of Mg and Mg–Ni composite, both obtained by high-energy ball milling, is faster and extensive when they are immersed in 1 M KCl. In contrast, milled Mg and Mg–Ni composite in pure water, MgH2 and MgH2 –Ni composites in pure water and in 1 M KCl show low yield and reactivity. Hydrolysis kinetics and yield are maximum with Mg–10 at% Ni composite milled for 30 min, so reaction is fully completed within an hour in the presence of chloride ions. It is related to the creation of micro-galvanic cells between Mg and dispersed Ni elements, accentuating greatly Mg corrosion in highly conductive aqueous media. A significant increase of the H2 production is also observed with 30 min milled Mg sample, likely because of the accentuation in the pitting corrosion resulting from the creation of numerous defects and fresh surfaces through the milling process. On the other hand, intensive ball milling of pure magnesium has no effect on the Mg reactivity in pure water. Ball milling effect is likely masked by the significant Mg passivation in pure water. A correlation is established between the conversion yield of ball-milled MgH2 powder in pure water and its effective surface area, which is increased by the milling process. Ni addition has no effect on the hydrolysis reaction in nonconductive media (i.e. pure water) and with nonconductive material (i.e. MgH2 ). 䉷 2005 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights reserved. Keywords: Hydrogen storage materials; Magnesium; High-energy ball milling; Hydrolysis reaction; Corrosion
1. Introduction Small PEM fuel cells are being considered as battery replacement for various mobile applications, such as cellular phones and laptop computers [1]. Their development requires the conception of a compact, safe and inexpensive hydrogen source and storage systems. Different sources of hydrogen can be considered, including compressed H2 , carbon-based H2 storage, metal hydrides, chemical hydrides ∗ Corresponding author. Tel.: +1 450 929 8185; fax: +1 450 929 8102. E-mail address: [email protected] (L. Roué). 1 Present address: Timcal Ltd., 6743 Bodio, Switzerland.
and methanol. Among them, chemical hydrides (e.g. NaBH4 , LiAlH4 , MgH2 ) [2–8] appear very promising due to their high theoretical hydrogen yield by weight. For example, pure hydrogen can be obtained by reacting MgH2 with water through the reaction: MgH2 + 2H2 O → Mg(OH)2 + 2H2 Hr = −277 kJ mol−1
(1)
The hydrogen yield of this hydrolysis reaction is 6.4 wt% when stoichiometric water is included in the calculation. The hydrogen yield increases to 15.2 wt% if produced water from the fuel cell is recovered for the above hydrolysis reaction and is not included in the calculation. Moreover, this reaction has the advantage to produce residual Mg(OH)2 ,
0360-3199/$30.00 䉷 2005 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.ijhydene.2005.01.001
110
M.-H. Grosjean et al. / International Journal of Hydrogen Energy 31 (2006) 109 – 119
which is environmental friendly. It is also possible to produce hydrogen by using very cheap Mg material by the following reaction: Mg + 2H2 O → Mg(OH)2 + H2 Hr = −354 kJ mol−1
(2)
It this case, the hydrogen yield is 3.3 wt% (including water mass) and 8.2 wt% (no water included in the calculation). However, the hydrolysis reaction using conventional Mg or MgH2 is rapidly interrupted because of the formation of a passive magnesium hydroxide layer onto the reactive material. Acid can be added to form soluble Mg2+ species but this might be detrimental to the equipment and it provides a potential hazard for users. Recent works have shown that the hydrolysis of MgH2 can be improved in terms of yield and kinetics by using nanocomposites MgH2 –X (X = Ca, Li, LiAlH4 , CaH2 ) materials prepared by ball milling [7,8]. The best performance was obtained with MgH2 + 20 mol% CaH2 powder mixture milled during 10 h, leading to a reaction yield close to 80% after 30 min of hydrolysis with excess water. It was stated that CaH2 hydrolysis is beneficial to break down the passive layer on particles, favouring the extensive reaction between freshly exposed MgH2 surface and water. The nanostructure formed upon extensive milling is essential to obtain high yield and fast kinetics of the hydrolysis reaction. The material appears very promising but it involves highly reactive CaH2 compound and extended milling time that is unfavourable to produce economically viable and safe hydrogen source materials. It should be noticed that the price of the chemical hydride and its recycling or recovery cost must be very low since the hydrolysis reaction is irreversible. In this work, we present an investigation of the hydrogen production using Mg- and MgH2 -based materials elaborated by high-energy ball milling. The aim is to conceive highly corroding and reactive materials to ensure a full completion of the hydrolysis reaction in neutral and saline aqueous solutions. In particular, our approach is to create micro-galvanic cells between magnesium and nickel additive. To our knowledge, no scientific work has been published on this matter but a few patents propose similar strategy to produce H2 [9–12] without emphasizing the mechanism and clarifying the influence of the microstructure of ball-milled Mg-based materials on the hydrolysis reactivity. This is the major aim of the present study. 2. Experimental 2.1. Sample preparation MgH2 (Th. Goldschmidt, 95 wt% MgH2 –5 wt% Mg, 20 m), Mg (Alpha Aesar, 99.8 wt%, −325 mesh) and Ni (Alpha Aesar, 99.8 wt%, −325 mesh) were used as starting materials. Powders (3 g) were introduced in a stainless-steel vial sealed under argon atmosphere and containing two
11.1 mm and one 14.3 mm diameter steel balls, corresponding to a ball-to-powder mass ratio of 8:1. MgH2 , MgH2 + 10 at% Ni, Mg and Mg + 10 at% Ni samples were milled for 30 min, 3 and 10 h using a SPEX 8000 ball mixer. 2.2. Sample characterization The samples were characterized by X-ray diffraction (XRD) using on a Bruker AXSD8 Siemens diffractometer with Cu K radiation. Scanning electron microscopy (SEM) observations were made using a Jeol JSM-6300F microscope. The distribution of Ni particles in composite powders was checked by energy dispersive X-ray (EDX) mapping. The specific surface area of the powders was measured by N2 adsorption (multipoint BET) using a Quantachrome Autosorb Automated Gas Sorption system. The oxygen content in the powders was measured by inert gas fusion technique with a TC-600 detector from LECO (detection limit of 10 ppm, measurement accuracy of 5%). The iron content was determined by neutron activation analysis (detection limit of 50 ppm, measurement accuracy of 5%). The hydrolysis reactions were carried out at room temperature and atmospheric pressure in a flask reactor of 1000 ml with two openings, one for water addition using a funnel pressure equalizing and the other one for hydrogen exhaust. The produced gas was flown through a condenser and drierite to remove all water vapour before to pass through a flowmeter (ADM 3000, Agilent Technologies). The flowmeter was connected to a computer recording gas flow and volume as a function of time. An aqueous solution of 100 ml was added to react with 250 mg of powder. A magnetic stirrer agitated the solution continuously during the test. The background flow (0.15 ml/min) was subtracted from the data. Each test was repeated at least two times and the precision of the measurement was estimated at 5%. Electrochemical measurements were performed at room temperature using a potentiostat/galvanostat/FRA (VoltaLab 40 from Radiometer Analytical) using a three electrodes cell. The counter electrode was a Pt wire placed in a separated compartment. The reference electrode was a saturated calomel electrode (SCE). The working electrode was a pellet made by cold pressing of 500 mg of milled powder in a 16 mm diameter stainless-steel cylindrical die. A load of 10,000 kg cm−2 was applied during 10 min leading to a nearly fully dense pellet. A Luggin capillary was used to reduce the ohmic drop. Experiments were performed in borate buffer solution (0.3 M Na2 B4 O7 /H3 BO3 , pH = 8.4) with 1 M KCl. Oxygen was removed from the solution before the experiment by a nitrogen flux. After 15 min in open circuit conditions, the electrode was polarized from −0.3 V vs. open circuit potential to 1.5 V vs. SCE at a scan rate of 0.5 mV s−1 . Polarization curves were corrected for the uncompensated resistance determined by AC impedance spectroscopy.
M.-H. Grosjean et al. / International Journal of Hydrogen Energy 31 (2006) 109 – 119
111
14
Mg MgH2
2
specific surface area (m /g)
12 10 8 6 4 2 0 0
2
4
6
8
10
milling time (h)
Fig. 1. Specific surface area of Mg and MgH2 powders as a function of the milling time.
3. Results and discussion 3.1. Effects of ball milling on Mg and MgH2 properties Fig. 1 shows the evolution of the specific surface area of Mg and MgH2 powders as a function of the milling time. The specific surface area of Mg powder decreases from ∼0.7 to ∼0.1 m2 g−1 after 10 h of milling. In contrast, the specific surface area of MgH2 powder increases significantly from 1.1 to 12.2 m2 g−1 after 30 min of milling then decreases to 7.8 m2 g−1 after 10 h of milling. This reflects the major difference in the mechanical properties between MgH2 and Mg. The brittle characteristic of MgH2 powder favours its fracturing during the milling process whereas the cold welding phenomenon predominates with ductile Mg material. This is confirmed by SEM observations (Fig. 2) showing the formation, after 30 min of milling, of large platelets with a mean size larger than 100 m for Mg powder in comparison to small granular particles with a mean diameter lower than 10 m for MgH2 . Fig. 3 presents XRD patterns of Mg and MgH2 powders as a function of the milling time. In the case of Mg powder (Fig. 3a), no major structural modification appears with milling, except the accentuation of the peak broadening reflecting the decrease of the Mg crystallite size as the milling time increases. Peaks attributed to MgO phase are also discernable and increases slightly with prolonged milling. For MgH2 powder (Fig. 3b), the XRD profiles are more complex with the appearance of several phases. Initially, two series of peaks related to Mg and -MgH2 phases are discernable. After 30 min of milling, additional peaks related to MgO and -MgH2 phases appear. Their intensity increases after 3 h of milling and does not change significantly with further milling. An accentuation of the peak broadening with
Fig. 2. SEM micrographs of (a) Mg and (b) MgH2 obtained after 30 min of milling.
milling is obvious and more important than observed with Mg powder. XRD patterns were fitted using Topaz software to estimate the phase abundance and to determine the crystallite size and internal strain from the Williamson–Hall plot. The results are summarized in Table 1. The Mg crystallite size is reduced to 54 nm after 10 h of milling compared to 14 nm for MgH2 , confirming the better efficiency of the fracturing process with brittle MgH2 powder. These values are in accordance with literature data on milled Mg [13] and milled MgH2 [14,15]. The strain into Mg after 10 h of milling is low (=0.21%) that indicates a high recovery rate during milling. Close value was obtained for MgH2 ( = 0.25% after 10 h milling) but it must be noted that the strain reaches a maximum value of 0.66% in the first stage of milling (30 min) and then decreases for further milling. This behaviour has been observed with various milled materials such as Ru and RuAl [16], Fe and W [17] and it can be explained by a change in the deformation mechanism from plastic deformation via generation and motion of dislocations to a gliding along the grain boundaries. The phase abundance presented in Table 1 confirms that the ball milling process accentuates Mg and MgH2 oxidation. Indeed, the proportion of MgO phase increases from 1 wt% for unmilled Mg to 13 wt% after 10 h of milling. MgO content increase with milling is
112
M.-H. Grosjean et al. / International Journal of Hydrogen Energy 31 (2006) 109 – 119
Intensity (a. u.)
Mg MgO
Table 1 Phase abundance, crystallite size and strain of Mg and MgH2 powders for different milling times Sample
Phase
Abundance (wt%)
Mg (0 h)
Mg Mgo Mg MgO Mg MgO Mg MgO
-MgH2
0h
0.5h
Mg (0.5 h) 3h
Mg (3 h) Mg (10 h)
10h
20
30
(a)
40
50
60
MgH2 (0 h)
2 Theta (˚) β-MgH2 X γ -MgH2
Mgh2 (0.5 h)
Intensity (a. u.)
Mg MgO
MgH2 (3 h)
0h
MgH2 (10 h) 0.5h
X
X
3h
Mg -MgH2 -MgH2 Mg MgO -MgH2 -Mgh2 Mg MgO -MgH2 -MgH2 Mg Mgo
Crystallite size (nm)
Strain (%)
99 1 99 1 90 10 88 12
> 100 — ∼100 — 82 — 54 —
0.02 — 0.10 — 0.17 — 0.21 —
95 5 83 4 12 1 66 15 14 5 60 18 16 6
> 100 — 31 — — — 21 — — — 14 — — —
0.05 — 0.66 — — — 0.25 — — — 0.24 — ——
10h
20
(b)
30
40
50
60
2 Theta (˚)
Fig. 3. X-ray diffraction patterns of (a) Mg and (b) MgH2 powders as a function of the milling time.
also observed with MgH2 powder but it appears less important (6 wt% of MgO after 10 h of milling). Previous works indicated from XPS and Auger analyses that MgO is dispersed into the bulk of the milled Mg powder rather than to be segregated on its surface [18]. This demonstrates that MgO formation is not related to the surface oxidation of the powder with ambient atmosphere once the sample has been taken out of the vial but it is rather due to its bulk oxidation during the milling process. In other words, the particle fracturing by ball-to-powder impacts creates clean new surfaces; such reactive surfaces readily oxidize with residual/leaking air present in the container or with residual oxides present at the surface of the milling tools and then, MgO is incorporated into the bulk of the material by subsequent interparticle welding events. In addition, MgH2 milling leads to the formation of -MgH2 phase. Its proportion reaches 15 wt% after 3 h of milling and does not increases drastically with further milling. Such a phase is usually formed under highpressure and high-temperature conditions [19]. Its appear-
ance during ball milling, i.e. at room temperature and atmospheric pressure, illustrates the capacity of the milling process to overcome the activation energy barrier by creating deformations and structural defects into the material [14]. The partial decomposition of MgH2 into Mg is also discernable with the increase of the Mg phase abundance from 5 wt% before milling to 16 wt% after 10 h of milling. Hydrogen desorption might be induced by the local increase of the temperature in the microscopic impact zones. The increase of the oxygen content in Mg and MgH2 with increasing milling time is confirmed by the chemical analysis of the powders as seen in Fig. 4. An oxygen concentration of 2.6 wt% is measured for Mg powder milled 10 h compared to 2.1 wt% for 10 h milled MgH2 . Assuming that all oxygen is in MgO form, it corresponds to a MgO content of 6.5 and 5.3 wt%. Thus, on the basis of the O contamination measurements, no major difference appears in the oxidation sensitivity of Mg and MgH2 through the milling process. This difference is even inexistent in the first stages of milling (30 min). In addition, as shown in Fig. 4, the iron contamination of Mg and MgH2 powders is very low (< 0.1wt% Fe in Mg and MgH2 after 10 h of milling). This reflects the low strength/hardness of the Mg and MgH2 powders compared to the steel milling tools, which limits the erosion of the vial and the balls during milling.
M.-H. Grosjean et al. / International Journal of Hydrogen Energy 31 (2006) 109 – 119
O in Mg O in MgH2 Fe in Mg Fe in MgH2
contaminant concentration (wt. %)
2.5
2.0
1.5
1.0
0.5
0.0 0
2
4
6
8
10
milling time (h)
Fig. 4. Variation of the concentration of contaminants (O, Fe) into Mg and MgH2 powders as a function of the milling time.
3.2. Hydrolysis of Mg and MgH2 Hydrogen production is expressed as conversion yield (%) defined as the volume of produced hydrogen over the theoretical volume of hydrogen that should be released assuming that all material is hydrolyzed. The conversion yields and hydrogen generation rates (ml H2 min−1 g−1 ) at different stages of the hydrolysis reaction for the different powders and media tested in the present work are presented in Table 2. Fig. 5 shows the hydrogen production profiles for reaction with pure water of Mg and MgH2 powders milled for different times. As seen in Fig. 5a, the ball milling treatment has no effect on the Mg reactivity in pure water. All curves show a very rapid H2 release in the first 30–60 s of hydrolysis followed by a drastic interruption of the reaction, leading to a very low conversion yield (< 15%). The abrupt stop of the hydrolysis reaction is due to the wellknown formation of passive Mg(OH)2 layer onto the Mg powder surface, which prevent further contact between water and un-reacted Mg material. The formation of Mg(OH)2 occurs although the bulk pH of the solution is much lower than the pH 10.5 of Mg(OH)2 formation [20] because local pH at the powder/solution interface can be greater than 10. This local alkalization is attributed to the cathodic component of the hydrolysis reaction leading to the formation of hydroxide ions through the hydrogen evolution reaction: 2H2 O + 2e = 2OH− + H2 . In contrast to that observed with Mg, the ball milling affects the MgH2 hydrolysis reactivity in pure water (Fig. 5b). The highest reactivity is obtained with 30 min milled MgH2 , which displays a conversion yield of 26% compared to 9% for unmilled MgH2 powder. The decrease of the MgH2 hydrolysis reactivity with further milling time (conversion yield of 21% and 16% for 3 h and 10 milled
113
powders, respectively) may be related to the formation of MgO and Mg as shown previously (see Table 1). However, the major reason to explain the reactivity decay with prolonged milling seems to be the decrease of the powder surface area. Indeed, the conversion yield of MgH2 powders appears to depend upon their specific surface area (Fig. 6). In other words, higher the reactive surface area of MgH2 powder in water, larger the fraction of powder will be reacted prior to its passivation. From extrapolation of the fitting curve in Fig. 6, a conversion yield of 100% might be reached for MgH2 powder having an effective surface area of 23 m2 /g. For that purpose, works are in progress to increase the effective surface area of the MgH2 powder through different approaches, such as the addition of a process-control agent (e.g. graphite) in order to prevent the cold welding of the particles during the milling process. 3.3. Influence of chloride ions on Mg and MgH2 hydrolysis When hydrolysis is performed in 1 M KCl aqueous solution rather than pure water, the results are drastically different for Mg powders. Indeed, as seen in Table 2 and Fig. 7a, the ball milling has a marked influence on the Mg powder reactivity in contrast to that observed previously in pure water. Mg powder milled 30 min displays the best efficiency with a conversion yield reaching 89% after 1 h of hydrolysis. Additional experiments performed with Mg powders milled for 15 and 45 min confirms that 30 min is the optimal milling duration to obtain highly reactive Mg material as clearly shown in Fig. 8 where the conversion yield after 1 h of hydrolysis in 1 M KCl is plotted versus the Mg milling duration. The hydrolysis curves for milled Mg in KCl solution presents a distinctive shape with the appearance of an induction period. That is, after the first 30 s of rapid hydrolysis (region I in Fig. 7a), the hydrolysis rate decreases for few minutes (region II) and reaccelerates again (region III) before stopping progressively (region IV). The duration of this induction period (region II) differs with milling time (i.e. ∼1 min for 0.5 h Mg powder compared to ∼3 min for 3 h Mg powder and ∼7 min for 10 h Mg powder). This period may correspond to the incubation time for pit initiation. The increase of the Mg conversion yield with KCl addition is associated with the destabilization of the Mg(OH)2 passive layer by chloride ions as demonstrated through corrosion studies [21,22]. The Cl− ions substitute OH− ions to form MgCl2 . This salt is more soluble than Mg(OH)2 and localized breakdown of the passive layer occurs leading to a pitting corrosion process. Song, et al. [23,24] suggested that the presence of Cl− might also accelerate the electrochemical reaction rate from magnesium to magnesium univalent ions. The increase of the H2 production with milled Mg powders (in particular with 0.5 h milled Mg sample) may reflect their higher susceptibility to pitting corrosion. In order to confirm this issue, corrosion experiments in presence of Cl− were performed. These experiments were done in borate buffer solution (pH = 8.5) in order to maintain the
114
M.-H. Grosjean et al. / International Journal of Hydrogen Energy 31 (2006) 109 – 119
Table 2 Conversion yields (%), total H2 volume (ml H2 g−1 ) after 1 h of hydrolysis and hydrogen generation rates (ml H2 min−1 g−1 ) at different stages of the hydrolysis reaction per gram of the different powders tested in the present work H2 O Mg
rate (ml/min.g)
0h
0.5 h
0–1 min 1–15 min 15–60 min
113 0.98 0.70 158 16
115 0.57 0.61 152 16
72 0.14 0.22 88 9
116 0.39 0.48 148 15
0–1 min 1–15 min 15–60 min
114 1.04 0.46 151 9
184 12.5 2.15 460 26
182 9.55 1.13 368 21
135 6.27 1.44 290 16
0–1 min 1–15 min 15–60 min
126 1.09 0.48 164 22
115 1.08 0.71 164 22
90 0.40 0.39 116 15
122 0.16 0.30 140 18
0–1 min 1–15 min 15–60 min
120 1.33 0.88 184 13
165 9.93 1.70 384 27
132 5.33 1.23 296 21
136 3.62 0.85 228 16
0h
0.5 h
3h
10 h
0–1 min 1–15 min 15–60 min
118 0.63 0.76 164 17
113 24.5 8.82 856 89
86 3.87 8.53 524 54
88 4.68 1.82 240 25
0–1 min 1–15 min 15–60 min
107 0.71 1.49 188 11
205 21.6 3.25 656 37
128 14.9 2.28 444 25
128 17.8 2.64 500 28
0-1 min 1–15 min 15–60 min
115 2.15 0.83 184 24
452 18.2 1.22 763 100
171 27.2 1.50 624 82
158 4.48 1.14 276 36
0–1 min 1–15 min 15–60 min
107 0.93 2.08 216 15
178 17.7 3.12 564 39
132 13.2 1.93 408 28
116 10.7 3.07 408 29
H2 produced (ml/g) Conversion yield (%) MgH2
rate (ml/min.g)
H2 produced (ml/g) Conversion yield (%) Mg+10%Ni
rate (ml/min.g)
H2 produced (ml/g) Conversion yield (%) MgH2 +10%Ni
rate (ml/min.g)
H2 produced (ml/g) Conversion yield (%) KCl 1 M Mg
rate (ml/min.g)
H2 produced (ml/g) Conversion yield (%) MgH2
rate (ml/min.g)
H2 produced (ml/g) Conversion yield (%) Mg+10%Ni
rate (ml/min.g)
H2 produced (ml/g) Conversion yield (%) MgH2 +10%Ni
rate (ml/min.g)
H2 produced (ml/g) Conversion yield (%)
3h
10 h
M.-H. Grosjean et al. / International Journal of Hydrogen Energy 31 (2006) 109 – 119
115
30
20
0h 0.5h 10h 10
3h
conversion yield (%) after 1 h of hydrolysis in water
conversion yield (%)
28
0.5 h
24
3h 20
10 h
16
12
0h 8
0
4
0
10
20
30
40
50
60
Time (minute)
(a)
0
2
4
6
8
10
12
14
specific surface area (m2/g)
Fig. 6. Conversion yield of MgH2 powders (0, 0.5, 3 and 10 h milled) as a function of their effective surface area.
30
conversion yield (%)
0.5 h
3h
20
10 h
10
0h
0 0
(b)
10
20
30
40
50
60
Time (minute)
Fig. 5. Hydrogen production profiles for reaction with pure water of (a) Mg and (b) MgH2 powders milled for different times.
pH at a constant value during the experiment and, therefore, enabling an exact corrosion-resistance measurement. Moreover, the reactivity of milled Mg electrodes was so high in un-buffered KCl solution that corrosion experiments were not able to be completed before the collapse of the electrode. Fig. 9 shows polarization curves of unmilled (curve a) and 0.5 h milled (curve b) magnesium electrodes in borate buffer solution with 1 M KCl. Both curves display a similar shape but several major differences are discernable. Indeed, with 0.5 h milled Mg electrode: (i) the current developed in the cathodic branch is much higher; (ii) the anodic peak current related to the active–passive transition is more intense; (iii) the mean value of the anodic current in the pseudo-passive region is about five times higher; (iv) the current oscillations reflecting a metastable pitting are much more numerous and intense; (v) the breakdown potential appears at about 1 V more negative potential; (vi) the current corrosion estimated by extrapolation of the Tafel slopes (first Stern method) is
2770 A cm −2 for 0.5 h milled Mg electrode compared to 490 A cm−2 for unmilled Mg electrode. All these features confirm the much higher sensitivity of 0.5 h milled Mg to passivity breakdown. It can be ascribed to the formation of defects (grains boundaries, dislocations, vacancies, etc.) through the milling process, which may favour localized chloride enrichment inducing an enhancement of the pitting corrosion. The breaking of the native surface oxide layer on Mg powder during milling may also create fresh active surfaces. On the other hand, the fact that Mg powders milled 3 and 10 h are less reactive than 0.5 h milled Mg powder can be ascribed to the accentuation of the powder oxidation during prolonged milling (i.e. > 30 min) as seen previously. Prolonged milling may also induce a more uniform distribution of the defects compared to 0.5 h Mg sample, resulting in a better distribution of Cl− on the powder surface and thus, localized chloride enrichment favourable to pitting corrosion might be less marked. In contrast, KCl addition induces no major improvement of the hydrolysis activity of MgH2 powders as shown in Fig. 7b and Table 2. For example, the conversion yield of 0.5 h milled MgH2 powder is 37% in 1 M KCl compared to 26% in pure water. It can be related to the insulator behaviour of MgH2 material with a band gap of 5 eV [25]. This is confirmed by the absence of current response through its electrochemical characterization, while it is a serious limitation for the galvanic and pitting corrosion processes. This is also consistent with the significant improvement of the corrosion resistance of Mg and Mg alloys in NaCl solution after cathodic treatment to form magnesium hydride layer [26]. 3.4. Influence of Ni addition on Mg and MgH2 hydrolysis It is well known that magnesium is highly susceptible to impurities that promote internal galvanic attacks [22]. Nickel displays low hydrogen overpotential and constitutes
116
M.-H. Grosjean et al. / International Journal of Hydrogen Energy 31 (2006) 109 – 119 0.25
100
0.5h
3h
III
40
20
b
0.15
60
0
0.20
IV
II
Current (A)
conversion yield (%)
80
0.05
10h
0.00
0h
-0.05
I
a
0.10
-0.10
0
10
20
(a)
30
40
50
-2.0
60
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
Potential (V. vs. SCE)
Time (minute)
Fig. 9. Polarization curves of (a) unmilled Mg and (b) 0.5 h milled Mg electrodes in 1 M KCl solution. Scan rate: 0.5 mV/s.
100
60
* 40
Mg MgO Ni
0.5h 10h 3h
20
0h 0 0
10
20
(b)
30
40
50
Intensity (a. u.)
conversion yield (%)
80
*
0h
*
0.5h
60
3h
Time (minute)
10h
Fig. 7. Hydrogen production profiles for reaction with 1 M KCl of (a) Mg and (b) MgH2 powders milled for different times.
20
30
40
50
60
2 Theta (°)
Fig. 10. X-ray diffraction patterns of Mg–10 at% Ni composite powder as a function of the milling time.
100 90
conversion yield (%) after 1 h of hydrolysis
80 70 60 50 40 30 20 10 0
1
2
3
4
5
6
7
8
9
10
milling time (h)
Fig. 8. Conversion yield of Mg powders after 1 h of hydrolysis in 1 M KCl vs. the milling time.
an appropriate cathode material to induce severe galvanic corrosion of Mg. Thus, Mg–Ni composite materials were elaborated to increase the yield and the kinetic of the hydrolysis reaction. Fig. 10 shows the XRD patterns of Mg–10 at% Ni powder as a function of the milling time. Only peaks associated with Mg and Ni phases are observed. Their constant position with milling indicates that alloy (e.g. Mg2 Ni) or solid solution such as Mg(Ni) are not formed. In addition, the Mg peak broadening is more important than observed previously with milled Mg alone (see Fig. 3a), reflecting an accentuation of the fracturing process in presence of Ni powder due to its higher hardness compared to Mg. The Mg crystallite size after 10 h of milling with 10 at% Ni is 17 nm compared to
M.-H. Grosjean et al. / International Journal of Hydrogen Energy 31 (2006) 109 – 119
117
hydrolysis. In short, 30 min is the optimum milling duration because:
Fig. 11. EDX mapping of Ni in Mg–10 at% Ni composite materials as a function of the milling time: (a) unmilled, (b) 0.5 h milled, (c) 3 h milled, (d) 10 h milled.
54 nm for Mg milled without Ni. The positive effect of the added Ni on the mechanical grinding of the Mg powder was also confirmed by the formation of smaller particles with a narrower distribution than observed with pure milled Mg. EDX mapping of Ni in Mg–10 at% Ni composite materials as a function of the milling time is shown in Fig. 11. After 30 min of milling, Ni inclusions with a size of about 5 m are clearly discernable and their distribution is rather heterogeneous. For longer milling time, the size of the Ni inclusions decreases and their distribution becomes more homogeneous and so, after 10 h of milling, the size of the Ni inclusions is lower that 1 m and they are very well distributed in the Mg matrix. Table 2 and Fig. 12a present the hydrolysis activities in 1 M KCl of Mg–10 at% Ni for different milling times. The conversion yield evolution with milling time is similar to the one observed previously with pure Mg. That is the hydrolysis efficiency is maximal for material milled 30 min and decreases for longer milling durations. The conversion yields as well as the hydrolysis kinetic are significantly improved with Ni addition. Indeed, with 0.5 h milled Mg–10 at% Ni, a conversion yield of 100% is obtained after 1 h of hydrolysis with a hydrogen production rate of 452 ml min−1 g−1 of powder in the first minute of reaction compared to a conversion yield of 89% with a hydrogen production rate of 113 ml min−1 g−1 of powder in the first minute of reaction for pure Mg material. Moreover, no induction period is observed during hydrolysis. In addition, the hydrolysis reactivity with 0.5 h milled Mg mixed manually in a mortar with 10 at% Ni (curve e in Fig. 12a) is high but remains lower than the one of Mg–10 at% Ni mixture milled 30 min, confirming the advantage of the intensive milling process to produce an highly reactive Mg–Ni composite powder for
• it corresponds to the milling time inducing the larger increase of the pitting corrosion with chloride as demonstrated previously with pure Mg • it is the optimal milling duration to form a composite Mg–Ni material with an appropriate distribution of Ni and Mg elements, i.e. with a close proximity between anode (Mg) and cathode (Ni) components and with a large cathode surface area in contact with the electrolyte.Such a microstructure favours the creation of numerous microgalvanic cells with minimized cell resistance resulting in accentuation of the corrosion (hydrolysis) reaction. For longer milling time (i.e. > 30 min), Ni material present initially in large proportion at the surface of the Mg powder is introduced into the bulk of the Mg matrix.Thus, the amount of Ni exposed to the KCl solution becomes lower, making the galvanic corrosion reaction less favourable. Moreover, when materials of disparate starting harnesses (such as Ni and Mg) are submitted to prolonged ball milling, the encapsulation of the hard phase (Ni) by the soft one (Mg) often occurs [27], accentuating the decrease of the fraction of Ni material in contact with the electrolyte. However, despite its lower conversion yield, the 0.5 h milled Mg powder without Ni produces more hydrogen per gram of material than the Mg–Ni composite, i.e. 856 ml H2 compared to 763 ml H2 (see Table 2), corresponding to a hydrogen yield of 7.6 and 6.8 wt%, respectively (the KCl solution weight is not included in the calculation). Lastly, the hydrolysis experiments performed with Mg–10 at% Ni in pure water and the ones done with MgH2 –10 at% Ni in KCl solution or pure water (see Fig. 12b–d and Table 2) do not show any significant reactivity improvement compared to the ones obtained without added Ni. This confirms the requirement to have an electronic conductive material and an ionic conductive media to induce the galvanic corrosion process. 4. Conclusion We have elaborated very effective and cheap H2 source materials, which consisted of Mg powder milled for 30 min or a mixture of Mg + 10 at%. Ni milled for 30 min. The hydrolysis reaction requires the presence of chloride ions to be effective. The hydrogen production is much higher than observed with MgH2 -based materials. The outcome of the present study can be summarized as follows: • high-energy ball milling of pure magnesium has no effect on the Mg reactivity for hydrolysis in pure water. Ball milling effect is assumed to be masked by the significant Mg passivation in pure water. • a correlation is established between the conversion yield of milled MgH2 powder in pure water and its effective
118
M.-H. Grosjean et al. / International Journal of Hydrogen Energy 31 (2006) 109 – 119 100
0.5 h
100
e 3h
60
40 10 h
80
conversion yield (%)
conversion yield (%)
80
60
40 0.5h 0h 10h 3h
0h
20
20
0
0 0
10
20
(a)
30
40
50
60
0
20
80
80
40
0.5h 3h 10h
conversion yield (%)
100
60
30
40
50
60
40 0.5h 3h 10h 0h
20
20 0h
0
0 0
(c)
60
Time (minute)
100
conversion yield (%)
10
(b)
Time (minute)
10
20
30
40
50
0
60
Time (minute)
10
20
(d)
30
40
50
60
Time (minute)
Fig. 12. Hydrogen production curves of (a) Mg–10 at% Ni in 1 M KCl, (b) Mg–10 at% Ni in pure water, (c) MgH2 –10 at% Ni in 1 M KCl, (d) MgH2 –10 at% Ni in pure water for different milling times.
surface area, which is increased by the milling process. • when hydrolysis is performed in the presence of chloride ions, a significant increase of the H2 production is observed with 30 min milled Mg sample. It is explained by the accentuation of the pitting corrosion due to the creation of numerous defects and fresh surfaces through the 30 min milling process. Further milling induces a decrease of the Mg reactivity for hydrolysis due to the accentuation of the powder oxidation and the homogeneity of the defects distribution into the material after prolonged milling. • a full completion of the hydrolysis reaction in the presence of chloride ions is observed with Mg–10 at% Ni composite material milled for 30 min. It is related to the creation of micro-galvanic cells between well-distributed Mg and Ni elements, which accentuates greatly the Mg corrosion in conductive aqueous media. • Ni addition has no effect on the hydrolysis reaction in nonconductive media (i.e. pure water) and with nonconductive material (i.e. MgH2 ).
Acknowledgements This work has been financially supported by HERA Hydrogen Storage Systems Inc. and the National Sciences and Engineering Research Council (NSERC) of Canada. We thank D. Cossement from the Hydrogen Research Institute for the BET analyses.
References [1] Linden D. In: Reddy TB, editor. Handbook of batteries. 3rd ed., New York: McGraw-Hill Inc.; 2001. p. 42.3–43.20. [2] Aiello R, Matthews MA, Reger DL, Collins JE. Int J Hydrogen Energy 1998;23:1103. [3] Kong VCY, Foulkes FR, Kirk DW, Hinatsu JT. Int J Hydrogen Energy 1999;24:665. [4] Kong VCY, Foulkes FR, Kirk DW, Hinatsu JT. Int J Hydrogen Energy 2003;28:205. [5] Kojima Y, Suzuki KI, Kawai Y. J Mater Science 2004;39:2227.
M.-H. Grosjean et al. / International Journal of Hydrogen Energy 31 (2006) 109 – 119 [6] Kojima Y, Suzuki KI, Fukimoto K, Kawai Y, Kimbara M, Nakanishi H, Matsumoto S. J Power Sources 2004;125:22. [7] Huot J, Schulz R, Liang G. J Alloys Compounds 2003;353:L12. [8] Tessier JP, Palau P, Huot J, Schulz R, Guay D. J Alloys Compounds 2004;376:180. [9] Black SA, et al. US patent 3,942,511, 1976. [10] Sergev S, et al. US Patent 4,223,661, 1980. [11] Sergev S, et al. US Patent 4,264, 362, 1981. [12] Kanoya I, et al. US patent 2003/0173229 A1, 2003. [13] Hwang S, Nishimura C, McCormick PG. Mater Sci Eng A 2001;318:22. [14] Huot J, Liang G, Boily S, Van Neste A, Schulz R. J Alloys Compounds 1999;293–295:495. [15] Hanada N, Ichikawa T, Orimo SI, Fujii H. J Alloys Compounds 2004;366:269. [16] Hellstren E, Fecht HJ, Fu Z, Johnson WL. J Appl Phys 1989;65:305.
119
[17] Oleszk D, Shingu PH. J Appl Phys 1996;79:2975. [18] Grosjean MH, Zidoune M, Roue L, Huot J, Schulz R. Electrochim Acta 2004;49:2461. [19] Bortz M, Berthevilee B, Böttger G, Yvon K. J Alloys Compounds 1999;287:L4. [20] Pourbaix M. In: Atlas of the electrochemical Equilibria in aqueous solutions. 2nd ed., Houston, Texas: National Association of Corrosion Engineers; 1974. p. 143. [21] Makar GL, Kruger J. J Electrochem Soc 1990;137(2):414. [22] Song GL, Atrens A. Adv Eng Mater 1999;1:11. [23] Song G, Atrens A, ST John D, Wu X, Nairn J. Corr Sci 1997;39:1981. [24] Song G, Atrens A, ST John D, Nairn J, Li Y. Corr Sci 1997;39:855. [25] Yu R, Lam PK. Phys Rev B 1988;37:8730. [26] Nakatsugawa I. US patent 6,117,298, 2000. [27] Maurice D, Courtney TH. Metal Mater Trans A 1996;27A:1981.
Hydrogen production via hydrolysis reaction from ball-milled Mg-based materials M.-H. Grosjeana , M. Zidounea , L. Rouéa,∗ , J.-Y. Huotb,1 a INRS-Énergies, Matériaux et Télécommunications, 1650 boulevard Lionel Boulet, Varennes (QC), Canada J3X 1S2 b HERA Hydrogen Storage Systems Inc, 577 rue Le Breton, Longueuil (QC), Canada J4G 1R9
Available online 3 March 2005
Abstract Hydrogen generation by hydrolysis of Mg and MgH2 has been investigated in pure water and 1 M KCl. It has been found that hydrolysis reaction of Mg and Mg–Ni composite, both obtained by high-energy ball milling, is faster and extensive when they are immersed in 1 M KCl. In contrast, milled Mg and Mg–Ni composite in pure water, MgH2 and MgH2 –Ni composites in pure water and in 1 M KCl show low yield and reactivity. Hydrolysis kinetics and yield are maximum with Mg–10 at% Ni composite milled for 30 min, so reaction is fully completed within an hour in the presence of chloride ions. It is related to the creation of micro-galvanic cells between Mg and dispersed Ni elements, accentuating greatly Mg corrosion in highly conductive aqueous media. A significant increase of the H2 production is also observed with 30 min milled Mg sample, likely because of the accentuation in the pitting corrosion resulting from the creation of numerous defects and fresh surfaces through the milling process. On the other hand, intensive ball milling of pure magnesium has no effect on the Mg reactivity in pure water. Ball milling effect is likely masked by the significant Mg passivation in pure water. A correlation is established between the conversion yield of ball-milled MgH2 powder in pure water and its effective surface area, which is increased by the milling process. Ni addition has no effect on the hydrolysis reaction in nonconductive media (i.e. pure water) and with nonconductive material (i.e. MgH2 ). 䉷 2005 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights reserved. Keywords: Hydrogen storage materials; Magnesium; High-energy ball milling; Hydrolysis reaction; Corrosion
1. Introduction Small PEM fuel cells are being considered as battery replacement for various mobile applications, such as cellular phones and laptop computers [1]. Their development requires the conception of a compact, safe and inexpensive hydrogen source and storage systems. Different sources of hydrogen can be considered, including compressed H2 , carbon-based H2 storage, metal hydrides, chemical hydrides ∗ Corresponding author. Tel.: +1 450 929 8185; fax: +1 450 929 8102. E-mail address: [email protected] (L. Roué). 1 Present address: Timcal Ltd., 6743 Bodio, Switzerland.
and methanol. Among them, chemical hydrides (e.g. NaBH4 , LiAlH4 , MgH2 ) [2–8] appear very promising due to their high theoretical hydrogen yield by weight. For example, pure hydrogen can be obtained by reacting MgH2 with water through the reaction: MgH2 + 2H2 O → Mg(OH)2 + 2H2 Hr = −277 kJ mol−1
(1)
The hydrogen yield of this hydrolysis reaction is 6.4 wt% when stoichiometric water is included in the calculation. The hydrogen yield increases to 15.2 wt% if produced water from the fuel cell is recovered for the above hydrolysis reaction and is not included in the calculation. Moreover, this reaction has the advantage to produce residual Mg(OH)2 ,
0360-3199/$30.00 䉷 2005 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.ijhydene.2005.01.001
110
M.-H. Grosjean et al. / International Journal of Hydrogen Energy 31 (2006) 109 – 119
which is environmental friendly. It is also possible to produce hydrogen by using very cheap Mg material by the following reaction: Mg + 2H2 O → Mg(OH)2 + H2 Hr = −354 kJ mol−1
(2)
It this case, the hydrogen yield is 3.3 wt% (including water mass) and 8.2 wt% (no water included in the calculation). However, the hydrolysis reaction using conventional Mg or MgH2 is rapidly interrupted because of the formation of a passive magnesium hydroxide layer onto the reactive material. Acid can be added to form soluble Mg2+ species but this might be detrimental to the equipment and it provides a potential hazard for users. Recent works have shown that the hydrolysis of MgH2 can be improved in terms of yield and kinetics by using nanocomposites MgH2 –X (X = Ca, Li, LiAlH4 , CaH2 ) materials prepared by ball milling [7,8]. The best performance was obtained with MgH2 + 20 mol% CaH2 powder mixture milled during 10 h, leading to a reaction yield close to 80% after 30 min of hydrolysis with excess water. It was stated that CaH2 hydrolysis is beneficial to break down the passive layer on particles, favouring the extensive reaction between freshly exposed MgH2 surface and water. The nanostructure formed upon extensive milling is essential to obtain high yield and fast kinetics of the hydrolysis reaction. The material appears very promising but it involves highly reactive CaH2 compound and extended milling time that is unfavourable to produce economically viable and safe hydrogen source materials. It should be noticed that the price of the chemical hydride and its recycling or recovery cost must be very low since the hydrolysis reaction is irreversible. In this work, we present an investigation of the hydrogen production using Mg- and MgH2 -based materials elaborated by high-energy ball milling. The aim is to conceive highly corroding and reactive materials to ensure a full completion of the hydrolysis reaction in neutral and saline aqueous solutions. In particular, our approach is to create micro-galvanic cells between magnesium and nickel additive. To our knowledge, no scientific work has been published on this matter but a few patents propose similar strategy to produce H2 [9–12] without emphasizing the mechanism and clarifying the influence of the microstructure of ball-milled Mg-based materials on the hydrolysis reactivity. This is the major aim of the present study. 2. Experimental 2.1. Sample preparation MgH2 (Th. Goldschmidt, 95 wt% MgH2 –5 wt% Mg, 20 m), Mg (Alpha Aesar, 99.8 wt%, −325 mesh) and Ni (Alpha Aesar, 99.8 wt%, −325 mesh) were used as starting materials. Powders (3 g) were introduced in a stainless-steel vial sealed under argon atmosphere and containing two
11.1 mm and one 14.3 mm diameter steel balls, corresponding to a ball-to-powder mass ratio of 8:1. MgH2 , MgH2 + 10 at% Ni, Mg and Mg + 10 at% Ni samples were milled for 30 min, 3 and 10 h using a SPEX 8000 ball mixer. 2.2. Sample characterization The samples were characterized by X-ray diffraction (XRD) using on a Bruker AXSD8 Siemens diffractometer with Cu K radiation. Scanning electron microscopy (SEM) observations were made using a Jeol JSM-6300F microscope. The distribution of Ni particles in composite powders was checked by energy dispersive X-ray (EDX) mapping. The specific surface area of the powders was measured by N2 adsorption (multipoint BET) using a Quantachrome Autosorb Automated Gas Sorption system. The oxygen content in the powders was measured by inert gas fusion technique with a TC-600 detector from LECO (detection limit of 10 ppm, measurement accuracy of 5%). The iron content was determined by neutron activation analysis (detection limit of 50 ppm, measurement accuracy of 5%). The hydrolysis reactions were carried out at room temperature and atmospheric pressure in a flask reactor of 1000 ml with two openings, one for water addition using a funnel pressure equalizing and the other one for hydrogen exhaust. The produced gas was flown through a condenser and drierite to remove all water vapour before to pass through a flowmeter (ADM 3000, Agilent Technologies). The flowmeter was connected to a computer recording gas flow and volume as a function of time. An aqueous solution of 100 ml was added to react with 250 mg of powder. A magnetic stirrer agitated the solution continuously during the test. The background flow (0.15 ml/min) was subtracted from the data. Each test was repeated at least two times and the precision of the measurement was estimated at 5%. Electrochemical measurements were performed at room temperature using a potentiostat/galvanostat/FRA (VoltaLab 40 from Radiometer Analytical) using a three electrodes cell. The counter electrode was a Pt wire placed in a separated compartment. The reference electrode was a saturated calomel electrode (SCE). The working electrode was a pellet made by cold pressing of 500 mg of milled powder in a 16 mm diameter stainless-steel cylindrical die. A load of 10,000 kg cm−2 was applied during 10 min leading to a nearly fully dense pellet. A Luggin capillary was used to reduce the ohmic drop. Experiments were performed in borate buffer solution (0.3 M Na2 B4 O7 /H3 BO3 , pH = 8.4) with 1 M KCl. Oxygen was removed from the solution before the experiment by a nitrogen flux. After 15 min in open circuit conditions, the electrode was polarized from −0.3 V vs. open circuit potential to 1.5 V vs. SCE at a scan rate of 0.5 mV s−1 . Polarization curves were corrected for the uncompensated resistance determined by AC impedance spectroscopy.
M.-H. Grosjean et al. / International Journal of Hydrogen Energy 31 (2006) 109 – 119
111
14
Mg MgH2
2
specific surface area (m /g)
12 10 8 6 4 2 0 0
2
4
6
8
10
milling time (h)
Fig. 1. Specific surface area of Mg and MgH2 powders as a function of the milling time.
3. Results and discussion 3.1. Effects of ball milling on Mg and MgH2 properties Fig. 1 shows the evolution of the specific surface area of Mg and MgH2 powders as a function of the milling time. The specific surface area of Mg powder decreases from ∼0.7 to ∼0.1 m2 g−1 after 10 h of milling. In contrast, the specific surface area of MgH2 powder increases significantly from 1.1 to 12.2 m2 g−1 after 30 min of milling then decreases to 7.8 m2 g−1 after 10 h of milling. This reflects the major difference in the mechanical properties between MgH2 and Mg. The brittle characteristic of MgH2 powder favours its fracturing during the milling process whereas the cold welding phenomenon predominates with ductile Mg material. This is confirmed by SEM observations (Fig. 2) showing the formation, after 30 min of milling, of large platelets with a mean size larger than 100 m for Mg powder in comparison to small granular particles with a mean diameter lower than 10 m for MgH2 . Fig. 3 presents XRD patterns of Mg and MgH2 powders as a function of the milling time. In the case of Mg powder (Fig. 3a), no major structural modification appears with milling, except the accentuation of the peak broadening reflecting the decrease of the Mg crystallite size as the milling time increases. Peaks attributed to MgO phase are also discernable and increases slightly with prolonged milling. For MgH2 powder (Fig. 3b), the XRD profiles are more complex with the appearance of several phases. Initially, two series of peaks related to Mg and -MgH2 phases are discernable. After 30 min of milling, additional peaks related to MgO and -MgH2 phases appear. Their intensity increases after 3 h of milling and does not change significantly with further milling. An accentuation of the peak broadening with
Fig. 2. SEM micrographs of (a) Mg and (b) MgH2 obtained after 30 min of milling.
milling is obvious and more important than observed with Mg powder. XRD patterns were fitted using Topaz software to estimate the phase abundance and to determine the crystallite size and internal strain from the Williamson–Hall plot. The results are summarized in Table 1. The Mg crystallite size is reduced to 54 nm after 10 h of milling compared to 14 nm for MgH2 , confirming the better efficiency of the fracturing process with brittle MgH2 powder. These values are in accordance with literature data on milled Mg [13] and milled MgH2 [14,15]. The strain into Mg after 10 h of milling is low (=0.21%) that indicates a high recovery rate during milling. Close value was obtained for MgH2 ( = 0.25% after 10 h milling) but it must be noted that the strain reaches a maximum value of 0.66% in the first stage of milling (30 min) and then decreases for further milling. This behaviour has been observed with various milled materials such as Ru and RuAl [16], Fe and W [17] and it can be explained by a change in the deformation mechanism from plastic deformation via generation and motion of dislocations to a gliding along the grain boundaries. The phase abundance presented in Table 1 confirms that the ball milling process accentuates Mg and MgH2 oxidation. Indeed, the proportion of MgO phase increases from 1 wt% for unmilled Mg to 13 wt% after 10 h of milling. MgO content increase with milling is
112
M.-H. Grosjean et al. / International Journal of Hydrogen Energy 31 (2006) 109 – 119
Intensity (a. u.)
Mg MgO
Table 1 Phase abundance, crystallite size and strain of Mg and MgH2 powders for different milling times Sample
Phase
Abundance (wt%)
Mg (0 h)
Mg Mgo Mg MgO Mg MgO Mg MgO
-MgH2
0h
0.5h
Mg (0.5 h) 3h
Mg (3 h) Mg (10 h)
10h
20
30
(a)
40
50
60
MgH2 (0 h)
2 Theta (˚) β-MgH2 X γ -MgH2
Mgh2 (0.5 h)
Intensity (a. u.)
Mg MgO
MgH2 (3 h)
0h
MgH2 (10 h) 0.5h
X
X
3h
Mg -MgH2 -MgH2 Mg MgO -MgH2 -Mgh2 Mg MgO -MgH2 -MgH2 Mg Mgo
Crystallite size (nm)
Strain (%)
99 1 99 1 90 10 88 12
> 100 — ∼100 — 82 — 54 —
0.02 — 0.10 — 0.17 — 0.21 —
95 5 83 4 12 1 66 15 14 5 60 18 16 6
> 100 — 31 — — — 21 — — — 14 — — —
0.05 — 0.66 — — — 0.25 — — — 0.24 — ——
10h
20
(b)
30
40
50
60
2 Theta (˚)
Fig. 3. X-ray diffraction patterns of (a) Mg and (b) MgH2 powders as a function of the milling time.
also observed with MgH2 powder but it appears less important (6 wt% of MgO after 10 h of milling). Previous works indicated from XPS and Auger analyses that MgO is dispersed into the bulk of the milled Mg powder rather than to be segregated on its surface [18]. This demonstrates that MgO formation is not related to the surface oxidation of the powder with ambient atmosphere once the sample has been taken out of the vial but it is rather due to its bulk oxidation during the milling process. In other words, the particle fracturing by ball-to-powder impacts creates clean new surfaces; such reactive surfaces readily oxidize with residual/leaking air present in the container or with residual oxides present at the surface of the milling tools and then, MgO is incorporated into the bulk of the material by subsequent interparticle welding events. In addition, MgH2 milling leads to the formation of -MgH2 phase. Its proportion reaches 15 wt% after 3 h of milling and does not increases drastically with further milling. Such a phase is usually formed under highpressure and high-temperature conditions [19]. Its appear-
ance during ball milling, i.e. at room temperature and atmospheric pressure, illustrates the capacity of the milling process to overcome the activation energy barrier by creating deformations and structural defects into the material [14]. The partial decomposition of MgH2 into Mg is also discernable with the increase of the Mg phase abundance from 5 wt% before milling to 16 wt% after 10 h of milling. Hydrogen desorption might be induced by the local increase of the temperature in the microscopic impact zones. The increase of the oxygen content in Mg and MgH2 with increasing milling time is confirmed by the chemical analysis of the powders as seen in Fig. 4. An oxygen concentration of 2.6 wt% is measured for Mg powder milled 10 h compared to 2.1 wt% for 10 h milled MgH2 . Assuming that all oxygen is in MgO form, it corresponds to a MgO content of 6.5 and 5.3 wt%. Thus, on the basis of the O contamination measurements, no major difference appears in the oxidation sensitivity of Mg and MgH2 through the milling process. This difference is even inexistent in the first stages of milling (30 min). In addition, as shown in Fig. 4, the iron contamination of Mg and MgH2 powders is very low (< 0.1wt% Fe in Mg and MgH2 after 10 h of milling). This reflects the low strength/hardness of the Mg and MgH2 powders compared to the steel milling tools, which limits the erosion of the vial and the balls during milling.
M.-H. Grosjean et al. / International Journal of Hydrogen Energy 31 (2006) 109 – 119
O in Mg O in MgH2 Fe in Mg Fe in MgH2
contaminant concentration (wt. %)
2.5
2.0
1.5
1.0
0.5
0.0 0
2
4
6
8
10
milling time (h)
Fig. 4. Variation of the concentration of contaminants (O, Fe) into Mg and MgH2 powders as a function of the milling time.
3.2. Hydrolysis of Mg and MgH2 Hydrogen production is expressed as conversion yield (%) defined as the volume of produced hydrogen over the theoretical volume of hydrogen that should be released assuming that all material is hydrolyzed. The conversion yields and hydrogen generation rates (ml H2 min−1 g−1 ) at different stages of the hydrolysis reaction for the different powders and media tested in the present work are presented in Table 2. Fig. 5 shows the hydrogen production profiles for reaction with pure water of Mg and MgH2 powders milled for different times. As seen in Fig. 5a, the ball milling treatment has no effect on the Mg reactivity in pure water. All curves show a very rapid H2 release in the first 30–60 s of hydrolysis followed by a drastic interruption of the reaction, leading to a very low conversion yield (< 15%). The abrupt stop of the hydrolysis reaction is due to the wellknown formation of passive Mg(OH)2 layer onto the Mg powder surface, which prevent further contact between water and un-reacted Mg material. The formation of Mg(OH)2 occurs although the bulk pH of the solution is much lower than the pH 10.5 of Mg(OH)2 formation [20] because local pH at the powder/solution interface can be greater than 10. This local alkalization is attributed to the cathodic component of the hydrolysis reaction leading to the formation of hydroxide ions through the hydrogen evolution reaction: 2H2 O + 2e = 2OH− + H2 . In contrast to that observed with Mg, the ball milling affects the MgH2 hydrolysis reactivity in pure water (Fig. 5b). The highest reactivity is obtained with 30 min milled MgH2 , which displays a conversion yield of 26% compared to 9% for unmilled MgH2 powder. The decrease of the MgH2 hydrolysis reactivity with further milling time (conversion yield of 21% and 16% for 3 h and 10 milled
113
powders, respectively) may be related to the formation of MgO and Mg as shown previously (see Table 1). However, the major reason to explain the reactivity decay with prolonged milling seems to be the decrease of the powder surface area. Indeed, the conversion yield of MgH2 powders appears to depend upon their specific surface area (Fig. 6). In other words, higher the reactive surface area of MgH2 powder in water, larger the fraction of powder will be reacted prior to its passivation. From extrapolation of the fitting curve in Fig. 6, a conversion yield of 100% might be reached for MgH2 powder having an effective surface area of 23 m2 /g. For that purpose, works are in progress to increase the effective surface area of the MgH2 powder through different approaches, such as the addition of a process-control agent (e.g. graphite) in order to prevent the cold welding of the particles during the milling process. 3.3. Influence of chloride ions on Mg and MgH2 hydrolysis When hydrolysis is performed in 1 M KCl aqueous solution rather than pure water, the results are drastically different for Mg powders. Indeed, as seen in Table 2 and Fig. 7a, the ball milling has a marked influence on the Mg powder reactivity in contrast to that observed previously in pure water. Mg powder milled 30 min displays the best efficiency with a conversion yield reaching 89% after 1 h of hydrolysis. Additional experiments performed with Mg powders milled for 15 and 45 min confirms that 30 min is the optimal milling duration to obtain highly reactive Mg material as clearly shown in Fig. 8 where the conversion yield after 1 h of hydrolysis in 1 M KCl is plotted versus the Mg milling duration. The hydrolysis curves for milled Mg in KCl solution presents a distinctive shape with the appearance of an induction period. That is, after the first 30 s of rapid hydrolysis (region I in Fig. 7a), the hydrolysis rate decreases for few minutes (region II) and reaccelerates again (region III) before stopping progressively (region IV). The duration of this induction period (region II) differs with milling time (i.e. ∼1 min for 0.5 h Mg powder compared to ∼3 min for 3 h Mg powder and ∼7 min for 10 h Mg powder). This period may correspond to the incubation time for pit initiation. The increase of the Mg conversion yield with KCl addition is associated with the destabilization of the Mg(OH)2 passive layer by chloride ions as demonstrated through corrosion studies [21,22]. The Cl− ions substitute OH− ions to form MgCl2 . This salt is more soluble than Mg(OH)2 and localized breakdown of the passive layer occurs leading to a pitting corrosion process. Song, et al. [23,24] suggested that the presence of Cl− might also accelerate the electrochemical reaction rate from magnesium to magnesium univalent ions. The increase of the H2 production with milled Mg powders (in particular with 0.5 h milled Mg sample) may reflect their higher susceptibility to pitting corrosion. In order to confirm this issue, corrosion experiments in presence of Cl− were performed. These experiments were done in borate buffer solution (pH = 8.5) in order to maintain the
114
M.-H. Grosjean et al. / International Journal of Hydrogen Energy 31 (2006) 109 – 119
Table 2 Conversion yields (%), total H2 volume (ml H2 g−1 ) after 1 h of hydrolysis and hydrogen generation rates (ml H2 min−1 g−1 ) at different stages of the hydrolysis reaction per gram of the different powders tested in the present work H2 O Mg
rate (ml/min.g)
0h
0.5 h
0–1 min 1–15 min 15–60 min
113 0.98 0.70 158 16
115 0.57 0.61 152 16
72 0.14 0.22 88 9
116 0.39 0.48 148 15
0–1 min 1–15 min 15–60 min
114 1.04 0.46 151 9
184 12.5 2.15 460 26
182 9.55 1.13 368 21
135 6.27 1.44 290 16
0–1 min 1–15 min 15–60 min
126 1.09 0.48 164 22
115 1.08 0.71 164 22
90 0.40 0.39 116 15
122 0.16 0.30 140 18
0–1 min 1–15 min 15–60 min
120 1.33 0.88 184 13
165 9.93 1.70 384 27
132 5.33 1.23 296 21
136 3.62 0.85 228 16
0h
0.5 h
3h
10 h
0–1 min 1–15 min 15–60 min
118 0.63 0.76 164 17
113 24.5 8.82 856 89
86 3.87 8.53 524 54
88 4.68 1.82 240 25
0–1 min 1–15 min 15–60 min
107 0.71 1.49 188 11
205 21.6 3.25 656 37
128 14.9 2.28 444 25
128 17.8 2.64 500 28
0-1 min 1–15 min 15–60 min
115 2.15 0.83 184 24
452 18.2 1.22 763 100
171 27.2 1.50 624 82
158 4.48 1.14 276 36
0–1 min 1–15 min 15–60 min
107 0.93 2.08 216 15
178 17.7 3.12 564 39
132 13.2 1.93 408 28
116 10.7 3.07 408 29
H2 produced (ml/g) Conversion yield (%) MgH2
rate (ml/min.g)
H2 produced (ml/g) Conversion yield (%) Mg+10%Ni
rate (ml/min.g)
H2 produced (ml/g) Conversion yield (%) MgH2 +10%Ni
rate (ml/min.g)
H2 produced (ml/g) Conversion yield (%) KCl 1 M Mg
rate (ml/min.g)
H2 produced (ml/g) Conversion yield (%) MgH2
rate (ml/min.g)
H2 produced (ml/g) Conversion yield (%) Mg+10%Ni
rate (ml/min.g)
H2 produced (ml/g) Conversion yield (%) MgH2 +10%Ni
rate (ml/min.g)
H2 produced (ml/g) Conversion yield (%)
3h
10 h
M.-H. Grosjean et al. / International Journal of Hydrogen Energy 31 (2006) 109 – 119
115
30
20
0h 0.5h 10h 10
3h
conversion yield (%) after 1 h of hydrolysis in water
conversion yield (%)
28
0.5 h
24
3h 20
10 h
16
12
0h 8
0
4
0
10
20
30
40
50
60
Time (minute)
(a)
0
2
4
6
8
10
12
14
specific surface area (m2/g)
Fig. 6. Conversion yield of MgH2 powders (0, 0.5, 3 and 10 h milled) as a function of their effective surface area.
30
conversion yield (%)
0.5 h
3h
20
10 h
10
0h
0 0
(b)
10
20
30
40
50
60
Time (minute)
Fig. 5. Hydrogen production profiles for reaction with pure water of (a) Mg and (b) MgH2 powders milled for different times.
pH at a constant value during the experiment and, therefore, enabling an exact corrosion-resistance measurement. Moreover, the reactivity of milled Mg electrodes was so high in un-buffered KCl solution that corrosion experiments were not able to be completed before the collapse of the electrode. Fig. 9 shows polarization curves of unmilled (curve a) and 0.5 h milled (curve b) magnesium electrodes in borate buffer solution with 1 M KCl. Both curves display a similar shape but several major differences are discernable. Indeed, with 0.5 h milled Mg electrode: (i) the current developed in the cathodic branch is much higher; (ii) the anodic peak current related to the active–passive transition is more intense; (iii) the mean value of the anodic current in the pseudo-passive region is about five times higher; (iv) the current oscillations reflecting a metastable pitting are much more numerous and intense; (v) the breakdown potential appears at about 1 V more negative potential; (vi) the current corrosion estimated by extrapolation of the Tafel slopes (first Stern method) is
2770 A cm −2 for 0.5 h milled Mg electrode compared to 490 A cm−2 for unmilled Mg electrode. All these features confirm the much higher sensitivity of 0.5 h milled Mg to passivity breakdown. It can be ascribed to the formation of defects (grains boundaries, dislocations, vacancies, etc.) through the milling process, which may favour localized chloride enrichment inducing an enhancement of the pitting corrosion. The breaking of the native surface oxide layer on Mg powder during milling may also create fresh active surfaces. On the other hand, the fact that Mg powders milled 3 and 10 h are less reactive than 0.5 h milled Mg powder can be ascribed to the accentuation of the powder oxidation during prolonged milling (i.e. > 30 min) as seen previously. Prolonged milling may also induce a more uniform distribution of the defects compared to 0.5 h Mg sample, resulting in a better distribution of Cl− on the powder surface and thus, localized chloride enrichment favourable to pitting corrosion might be less marked. In contrast, KCl addition induces no major improvement of the hydrolysis activity of MgH2 powders as shown in Fig. 7b and Table 2. For example, the conversion yield of 0.5 h milled MgH2 powder is 37% in 1 M KCl compared to 26% in pure water. It can be related to the insulator behaviour of MgH2 material with a band gap of 5 eV [25]. This is confirmed by the absence of current response through its electrochemical characterization, while it is a serious limitation for the galvanic and pitting corrosion processes. This is also consistent with the significant improvement of the corrosion resistance of Mg and Mg alloys in NaCl solution after cathodic treatment to form magnesium hydride layer [26]. 3.4. Influence of Ni addition on Mg and MgH2 hydrolysis It is well known that magnesium is highly susceptible to impurities that promote internal galvanic attacks [22]. Nickel displays low hydrogen overpotential and constitutes
116
M.-H. Grosjean et al. / International Journal of Hydrogen Energy 31 (2006) 109 – 119 0.25
100
0.5h
3h
III
40
20
b
0.15
60
0
0.20
IV
II
Current (A)
conversion yield (%)
80
0.05
10h
0.00
0h
-0.05
I
a
0.10
-0.10
0
10
20
(a)
30
40
50
-2.0
60
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
Potential (V. vs. SCE)
Time (minute)
Fig. 9. Polarization curves of (a) unmilled Mg and (b) 0.5 h milled Mg electrodes in 1 M KCl solution. Scan rate: 0.5 mV/s.
100
60
* 40
Mg MgO Ni
0.5h 10h 3h
20
0h 0 0
10
20
(b)
30
40
50
Intensity (a. u.)
conversion yield (%)
80
*
0h
*
0.5h
60
3h
Time (minute)
10h
Fig. 7. Hydrogen production profiles for reaction with 1 M KCl of (a) Mg and (b) MgH2 powders milled for different times.
20
30
40
50
60
2 Theta (°)
Fig. 10. X-ray diffraction patterns of Mg–10 at% Ni composite powder as a function of the milling time.
100 90
conversion yield (%) after 1 h of hydrolysis
80 70 60 50 40 30 20 10 0
1
2
3
4
5
6
7
8
9
10
milling time (h)
Fig. 8. Conversion yield of Mg powders after 1 h of hydrolysis in 1 M KCl vs. the milling time.
an appropriate cathode material to induce severe galvanic corrosion of Mg. Thus, Mg–Ni composite materials were elaborated to increase the yield and the kinetic of the hydrolysis reaction. Fig. 10 shows the XRD patterns of Mg–10 at% Ni powder as a function of the milling time. Only peaks associated with Mg and Ni phases are observed. Their constant position with milling indicates that alloy (e.g. Mg2 Ni) or solid solution such as Mg(Ni) are not formed. In addition, the Mg peak broadening is more important than observed previously with milled Mg alone (see Fig. 3a), reflecting an accentuation of the fracturing process in presence of Ni powder due to its higher hardness compared to Mg. The Mg crystallite size after 10 h of milling with 10 at% Ni is 17 nm compared to
M.-H. Grosjean et al. / International Journal of Hydrogen Energy 31 (2006) 109 – 119
117
hydrolysis. In short, 30 min is the optimum milling duration because:
Fig. 11. EDX mapping of Ni in Mg–10 at% Ni composite materials as a function of the milling time: (a) unmilled, (b) 0.5 h milled, (c) 3 h milled, (d) 10 h milled.
54 nm for Mg milled without Ni. The positive effect of the added Ni on the mechanical grinding of the Mg powder was also confirmed by the formation of smaller particles with a narrower distribution than observed with pure milled Mg. EDX mapping of Ni in Mg–10 at% Ni composite materials as a function of the milling time is shown in Fig. 11. After 30 min of milling, Ni inclusions with a size of about 5 m are clearly discernable and their distribution is rather heterogeneous. For longer milling time, the size of the Ni inclusions decreases and their distribution becomes more homogeneous and so, after 10 h of milling, the size of the Ni inclusions is lower that 1 m and they are very well distributed in the Mg matrix. Table 2 and Fig. 12a present the hydrolysis activities in 1 M KCl of Mg–10 at% Ni for different milling times. The conversion yield evolution with milling time is similar to the one observed previously with pure Mg. That is the hydrolysis efficiency is maximal for material milled 30 min and decreases for longer milling durations. The conversion yields as well as the hydrolysis kinetic are significantly improved with Ni addition. Indeed, with 0.5 h milled Mg–10 at% Ni, a conversion yield of 100% is obtained after 1 h of hydrolysis with a hydrogen production rate of 452 ml min−1 g−1 of powder in the first minute of reaction compared to a conversion yield of 89% with a hydrogen production rate of 113 ml min−1 g−1 of powder in the first minute of reaction for pure Mg material. Moreover, no induction period is observed during hydrolysis. In addition, the hydrolysis reactivity with 0.5 h milled Mg mixed manually in a mortar with 10 at% Ni (curve e in Fig. 12a) is high but remains lower than the one of Mg–10 at% Ni mixture milled 30 min, confirming the advantage of the intensive milling process to produce an highly reactive Mg–Ni composite powder for
• it corresponds to the milling time inducing the larger increase of the pitting corrosion with chloride as demonstrated previously with pure Mg • it is the optimal milling duration to form a composite Mg–Ni material with an appropriate distribution of Ni and Mg elements, i.e. with a close proximity between anode (Mg) and cathode (Ni) components and with a large cathode surface area in contact with the electrolyte.Such a microstructure favours the creation of numerous microgalvanic cells with minimized cell resistance resulting in accentuation of the corrosion (hydrolysis) reaction. For longer milling time (i.e. > 30 min), Ni material present initially in large proportion at the surface of the Mg powder is introduced into the bulk of the Mg matrix.Thus, the amount of Ni exposed to the KCl solution becomes lower, making the galvanic corrosion reaction less favourable. Moreover, when materials of disparate starting harnesses (such as Ni and Mg) are submitted to prolonged ball milling, the encapsulation of the hard phase (Ni) by the soft one (Mg) often occurs [27], accentuating the decrease of the fraction of Ni material in contact with the electrolyte. However, despite its lower conversion yield, the 0.5 h milled Mg powder without Ni produces more hydrogen per gram of material than the Mg–Ni composite, i.e. 856 ml H2 compared to 763 ml H2 (see Table 2), corresponding to a hydrogen yield of 7.6 and 6.8 wt%, respectively (the KCl solution weight is not included in the calculation). Lastly, the hydrolysis experiments performed with Mg–10 at% Ni in pure water and the ones done with MgH2 –10 at% Ni in KCl solution or pure water (see Fig. 12b–d and Table 2) do not show any significant reactivity improvement compared to the ones obtained without added Ni. This confirms the requirement to have an electronic conductive material and an ionic conductive media to induce the galvanic corrosion process. 4. Conclusion We have elaborated very effective and cheap H2 source materials, which consisted of Mg powder milled for 30 min or a mixture of Mg + 10 at%. Ni milled for 30 min. The hydrolysis reaction requires the presence of chloride ions to be effective. The hydrogen production is much higher than observed with MgH2 -based materials. The outcome of the present study can be summarized as follows: • high-energy ball milling of pure magnesium has no effect on the Mg reactivity for hydrolysis in pure water. Ball milling effect is assumed to be masked by the significant Mg passivation in pure water. • a correlation is established between the conversion yield of milled MgH2 powder in pure water and its effective
118
M.-H. Grosjean et al. / International Journal of Hydrogen Energy 31 (2006) 109 – 119 100
0.5 h
100
e 3h
60
40 10 h
80
conversion yield (%)
conversion yield (%)
80
60
40 0.5h 0h 10h 3h
0h
20
20
0
0 0
10
20
(a)
30
40
50
60
0
20
80
80
40
0.5h 3h 10h
conversion yield (%)
100
60
30
40
50
60
40 0.5h 3h 10h 0h
20
20 0h
0
0 0
(c)
60
Time (minute)
100
conversion yield (%)
10
(b)
Time (minute)
10
20
30
40
50
0
60
Time (minute)
10
20
(d)
30
40
50
60
Time (minute)
Fig. 12. Hydrogen production curves of (a) Mg–10 at% Ni in 1 M KCl, (b) Mg–10 at% Ni in pure water, (c) MgH2 –10 at% Ni in 1 M KCl, (d) MgH2 –10 at% Ni in pure water for different milling times.
surface area, which is increased by the milling process. • when hydrolysis is performed in the presence of chloride ions, a significant increase of the H2 production is observed with 30 min milled Mg sample. It is explained by the accentuation of the pitting corrosion due to the creation of numerous defects and fresh surfaces through the 30 min milling process. Further milling induces a decrease of the Mg reactivity for hydrolysis due to the accentuation of the powder oxidation and the homogeneity of the defects distribution into the material after prolonged milling. • a full completion of the hydrolysis reaction in the presence of chloride ions is observed with Mg–10 at% Ni composite material milled for 30 min. It is related to the creation of micro-galvanic cells between well-distributed Mg and Ni elements, which accentuates greatly the Mg corrosion in conductive aqueous media. • Ni addition has no effect on the hydrolysis reaction in nonconductive media (i.e. pure water) and with nonconductive material (i.e. MgH2 ).
Acknowledgements This work has been financially supported by HERA Hydrogen Storage Systems Inc. and the National Sciences and Engineering Research Council (NSERC) of Canada. We thank D. Cossement from the Hydrogen Research Institute for the BET analyses.
References [1] Linden D. In: Reddy TB, editor. Handbook of batteries. 3rd ed., New York: McGraw-Hill Inc.; 2001. p. 42.3–43.20. [2] Aiello R, Matthews MA, Reger DL, Collins JE. Int J Hydrogen Energy 1998;23:1103. [3] Kong VCY, Foulkes FR, Kirk DW, Hinatsu JT. Int J Hydrogen Energy 1999;24:665. [4] Kong VCY, Foulkes FR, Kirk DW, Hinatsu JT. Int J Hydrogen Energy 2003;28:205. [5] Kojima Y, Suzuki KI, Kawai Y. J Mater Science 2004;39:2227.
M.-H. Grosjean et al. / International Journal of Hydrogen Energy 31 (2006) 109 – 119 [6] Kojima Y, Suzuki KI, Fukimoto K, Kawai Y, Kimbara M, Nakanishi H, Matsumoto S. J Power Sources 2004;125:22. [7] Huot J, Schulz R, Liang G. J Alloys Compounds 2003;353:L12. [8] Tessier JP, Palau P, Huot J, Schulz R, Guay D. J Alloys Compounds 2004;376:180. [9] Black SA, et al. US patent 3,942,511, 1976. [10] Sergev S, et al. US Patent 4,223,661, 1980. [11] Sergev S, et al. US Patent 4,264, 362, 1981. [12] Kanoya I, et al. US patent 2003/0173229 A1, 2003. [13] Hwang S, Nishimura C, McCormick PG. Mater Sci Eng A 2001;318:22. [14] Huot J, Liang G, Boily S, Van Neste A, Schulz R. J Alloys Compounds 1999;293–295:495. [15] Hanada N, Ichikawa T, Orimo SI, Fujii H. J Alloys Compounds 2004;366:269. [16] Hellstren E, Fecht HJ, Fu Z, Johnson WL. J Appl Phys 1989;65:305.
119
[17] Oleszk D, Shingu PH. J Appl Phys 1996;79:2975. [18] Grosjean MH, Zidoune M, Roue L, Huot J, Schulz R. Electrochim Acta 2004;49:2461. [19] Bortz M, Berthevilee B, Böttger G, Yvon K. J Alloys Compounds 1999;287:L4. [20] Pourbaix M. In: Atlas of the electrochemical Equilibria in aqueous solutions. 2nd ed., Houston, Texas: National Association of Corrosion Engineers; 1974. p. 143. [21] Makar GL, Kruger J. J Electrochem Soc 1990;137(2):414. [22] Song GL, Atrens A. Adv Eng Mater 1999;1:11. [23] Song G, Atrens A, ST John D, Wu X, Nairn J. Corr Sci 1997;39:1981. [24] Song G, Atrens A, ST John D, Nairn J, Li Y. Corr Sci 1997;39:855. [25] Yu R, Lam PK. Phys Rev B 1988;37:8730. [26] Nakatsugawa I. US patent 6,117,298, 2000. [27] Maurice D, Courtney TH. Metal Mater Trans A 1996;27A:1981.