- Email: [email protected]
The catalytic effect of Fe and Cr on hydrogen and deuterium absorption in Mg thin films
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 7 ( 2 0 1 2 ) 3 5 4 0 e3 5 4 7
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/he
The catalytic effect of Fe and Cr on hydrogen and deuterium absorption in Mg thin films H. Fritzsche a,*, W.P. Kalisvaart b, B. Zahiri b, R. Flacau a, D. Mitlin b a
National Research Council Canada, SIMS, Canadian Neutron Beam Centre, Chalk River Laboratories, building 459, Chalk River, Ontario K0J 1J0, Canada b Chemical and Materials Engineering, University of Alberta, T6G 2V4, and National Research Council Canada, National Institute for Nanotechnology, T6G 2M9, Edmonton, Alberta, Canada
article info
abstract
Article history:
We examined the deuterium absorption and desorption of 55 nm thick Mg films alloyed
Received 27 April 2011
with Fe and Cr using in-situ neutron reflectometry. The Mg alloy films were covered with
Received in revised form
bimetallic catalyst layers and could be fully absorbed at room temperature at a pressure of
31 May 2011
8 mbar. The NR experiments revealed a deuterium gradient within the Mg alloy layers
Accepted 2 June 2011
during absorption and a large deuterium uptake up to a D/M ratio of about 0.45 before the
Available online 1 September 2011
layer started to expand and form magnesium deuteride (MgD2). Our NR data suggest that the catalytic effect of the FeeCr alloy is based on the avoidance of the formation of
Keywords:
a blocking MgD2 layer in the early stages of the absorption process leading to a fast
Low temperature hydrogen
hydrogen absorption kinetics.
absorption
Crown Copyright ª 2011, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All
Magnesium
rights reserved.
Thin film Neutron reflectometry Bilayer catalysts
1.
Introduction
MgH2 is a very attractive hydrogen storage material because it possesses a high gravimetric storage density of 7.6 wt.% and is relatively inexpensive. However, the temperatures needed to overcome the slow kinetics are too high for most practical applications. Therefore, a lot of experimental work has been done to improve the sorption kinetics and lower the temperatures needed for absorption and desorption. Most approaches consisted of using high energy ball milling to reduce the grain sizes and adding metals or metal oxides as catalysts [1e9]. Especially combinations of 3d metals like FeeTi [10,11], FeeV [12], and CreTi [13] showed very fast sorption kinetics when alloyed with Mg.
In the last years research on thin films has attracted a lot of attention. Using thin films as model systems makes it possible to discriminate between the catalytic surface effect of a cap layer on top of a Mg film and the catalytic bulk effect of alloying the Mg film with other elements. These thin film systems have the big advantage that they can be fabricated in a controlled and reproducible way so that e.g. a Mg film can be completely covered with a surface catalyst layer. A 5 nm thick Pd cap layer has turned out to be sufficient to avoid oxidation of the hydrogen storage film underneath and to dissociate the hydrogen molecules [14e16]. As a consequence, thin films do not need to be activated at high temperature and they always show much better sorption behaviour than the corresponding bulk material. Thin films
* Corresponding author. E-mail address: [email protected] (H. Fritzsche). 0360-3199/$ e see front matter Crown Copyright ª 2011, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.ijhydene.2011.06.014
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 7 ( 2 0 1 2 ) 3 5 4 0 e3 5 4 7
absorb hydrogen fast even at room temperature and hydrogen pressures below 1 bar [15,17,18]. Using the thin film approach it was possible to even improve the performance of a single Pd layer by adding another layer between the Pd cap layer and the Mg film. Bilayer catalysts like Nb/Pd [19e21], Ta/Pd [17,22], V/ Pd [20,23], and Ti/Pd [18,20,21,23,24] improved the absorption and desorption kinetics substantially. The second catalyst layer acts as a diffusion barrier and avoids the formation of a MgePd alloy as was observed in a previous study [25]. Compared to the brute force intermixing of hydrogen storage materials with catalysts caused by ball milling, coevaporation during thin film sputtering enables one to fabricate well-defined alloy compositions and modify only the bulk storage material leaving the surface catalyst layer unchanged. Using this approach, several classes of readily cyclable ternary alloys with rapid low temperature kinetics were discovered: MgeFeeTi [26], MgeAleTi [26], MgeFeeV [12], MgeCreTi [13], and MgeCreV [27]. Here we present a new alloy, MgeFeeCr, which shows rapid absorption and desorption kinetics. We applied in-situ Neutron Reflectometry (NR) to determine the deuterium concentration in the film structure along with the change of the film structure during the absorption and desorption. With its high sensitivity to deuterium and its nanometer resolution along the film normal, NR is the technique of choice to determine the deuterium concentration profile in thin films. Whereas PcT curves and sorption cycling experiments only measure the total hydrogen content, NR gives deeper insight into the absorption and desorption process by revealing hydrogen or deuterium gradients in thin films with subnanometer precision.
2.
Experimental
The thin films were prepared by co-sputtering onto a Si(100) wafer in a confocal sputtering chamber (Orion 5 instrument from AJA International) operating at an Ar (purity 99.999%) pressure of 5 103 mbar which had been previously evacuated to a pressure less than 3 108 mbar. For the NR experiments, first a 10 nm Ta buffer layer was deposited onto the wafer and, without interruption, a 50 nm Mg0.8Fe0.1Cr0.1 alloy layer was co-sputtered followed by either a (5 nm Ta/5 nm Pd) or a (5 nm FeCr/5 nm Pd) bilayer. For the volumetric measurements, which were always performed at 200 C, a 1.5 mm thick Mg0.8Fe0.1Cr0.1 film was used with a (5 nm Ta/5 nm Pd) bilayer both, at the top and bottom of the Mg alloy layer. Hydrogen was absorbed at a starting pressure of 3 bar and desorbed at a starting pressure of 0.005 bar. The system (HyEnergy PCTPro 2000) automatically switched from absorption to desorption, and vice versa, once the sorption rate fell below 0.004 wt%/min. Transmission electron microscopy was performed using the JEOL JEM-2100 microscope, operating at 200 kV acceleration voltage. The sample was cycled for 140 times before taking it out in the desorbed state. The neutron reflectometry experiments were performed on the D3 reflectometer at the neutron research reactor NRU in Chalk River. We used a focusing pyrolytic graphite (PG) monochromator at a neutron wavelength of l ¼ 0.237 nm
3541
along with a PG filter [28] to reduce the higher order contamination (l/2 and l/3). In an NR experiment the neutron beam hits the surface of a sample at the scattering angle q and is specularly reflected off the sample at the same angle, which is typically in the range between 0 and 2 . The interfaces of the samples are arranged perpendicular to the scattering vector. ! ! 4p j! q j ¼ kr ki ¼ sinq (1) l ! ! with kr and ki being the reflected and incoming neutron wave vector. At grazing incidence the interaction of the neutron with the sample can be described with a neutron index of refraction [29] analogous to optical reflectivity. The neutron index of refraction depends on the strength of the interaction of neutrons with a specific isotope in the film and can be represented by n2 ¼ 1
l2 Nj bj ; p
(2)
where Nj is the number density and bj the nuclear scattering length [30,31] of the elements/isotopes in layer j. The product Njbj is called scattering length density (SLD). We used the software PARRATT32, which is based on the Parratt recursion algorithm [32], to fit our NR data by varying the SLD, layer thickness, and interface roughness of each individual layer j. In our NR experiments we used deuterium because of its large coherent scattering length, which leads to a dramatic increase in the SLD of the absorbing layer, thus leading to a change in the index of refraction and finally in the measured reflectivity curve. Total reflection of neutrons occurs up to a critical scattering vector qc: pffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffi qc ¼ 4 pNb ¼ 4 pS:
(3)
The SLD S of our Mg-based alloy films is larger than the SLD of the Si substrate. Therefore, the increase of qc is a good measure of the average deuterium content absorbed by the film. The deuterium concentration cD in D/metal can be calculated as follows [33]: cD ¼
SMþD tMþD bM , 1 : SM tM bD
(4)
SM and SMþD are the SLD of the as-prepared and deuterium loaded film, respectively, with the corresponding film thicknesses tM and tMþD, the average scattering length of the film bM, and the scattering length of deuterium [30,31] bD ¼ 6.67 fm.
3.
Neutron reflectometry data
The NR curve of an as-prepared 55 nm thick Mg0.8Fe0.1Cr0.1 film capped with a (5 nm FeCr/5 nm Pd) bilayer catalyst is shown in Fig. 1. The open circles are the measured data points, the solid line is the fit based on the SLD profile depicted in the inset. The reflectivity curve contains a series of maxima and minima, the so-called Kiessig fringes [34], which correspond to characteristic thicknesses in the specimen. The oscillations in Fig. 1 are due to our sample’s total film thickness of 76 nm. The SLD profile provides a real space representation of the film and is
3542
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 7 ( 2 0 1 2 ) 3 5 4 0 e3 5 4 7
Fig. 1 e Measured neutron reflectivity curve (open circles) as a function of the scattering vector q for a 55 nm Mg0.8Fe0.1Cr0.1 film capped with a FeCr/Pd bilayer. The solid line represents a fit corresponding to the SLD profile shown in the inset.
plotted as a function of z, the film normal, with z ¼ 0 at the film surface. The SLD profile shows the film structure with ˚ 2) followed by a 5.3 nm Pd surface layer (SLD ¼ 4.1 106 A 6 ˚ 2 A ), a 54.7 nm a 5.4 nm FeCr layer (SLD ¼ 6.0 10 ˚ 2), a 10.7 nm Ta layer Mg0.8Fe0.1Cr0.1 layer (SLD ¼ 2.7 106 A ˚ 2), and a bulk Si crystal (SLD ¼ 2.1 106 (SLD ¼ 3.6 106 A 2 ˚ A ). The SLD values determined from the fit are within 5% of the calculated values using tabulated data [30,31]. The kinetics of the deuterium absorption can be easily and quickly monitored by measuring the NR curve in a narrow q range around the critical scattering vector qc. This is displayed in Fig. 2, where NR curves of the sample with the FeCr/Pd catalyst bilayer are plotted at different times after a deuterium pressure of 8 mbar was introduced into the sample cell at t ¼ 0 at 300 K. The critical scattering vector qc increases continuously with time until reaching saturation after about 4 h. This shift of qc can be taken as a direct measure of deuterium absorption because according to Eq. (3) the deuterium
Fig. 3 e Neutron reflectivity curves close to the critical scattering vector qc for a 52 nm thick Mg0.8Fe0.1Cr0.1 film capped with a Ta/Pd bilayer, measured at room temperature in a deuterium pressure of 10 mbar for different exposure times.
a
b
c
Fig. 2 e Neutron reflectivity curves close to the critical scattering vector qc for a 55 nm thick Mg0.8Fe0.1Cr0.1 film capped with a FeCr/Pd bilayer, measured at room temperature in a deuterium pressure of 8 mbar for different exposure times.
Fig. 4 e Measured neutron reflectivity curves (open circles) for a 55 nm thick Mg0.8Fe0.1Cr0.1 film capped with a FeCr/Pd bilayer in a deuterium pressure of 8 mbar after exposure of a) 20 min, b) 57 min, and c) 84 min. The solid lines represent fits corresponding to the SLD profiles shown in the inset.
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 7 ( 2 0 1 2 ) 3 5 4 0 e3 5 4 7
absorption leads to an increase of the SLD in the Mg alloy layer due to the large scattering length of deuterium. The reflectivity curve measured after 222 min exposure time corresponds to a D/M ratio of c ¼ 1.5, very close to the maximum expected saturation value of 1.6, calculated under the assumption that all Mg atoms react to form MgD2. We also performed the same experiments for the samples with a Ta/Pd bilayer, however, in a slightly higher deuterium pressure of 10 mbar. The data are plotted in Fig. 3. Both samples show similar behaviour, the slightly faster kinetics of the Ta/Pd catalyzed sample is most likely due to the slightly higher pressure of 10 mbar for the Ta/Pd sample. The full range of the measured NR curves for the films with the FeCr/Pd bilayer catalyst is shown in Fig. 4 and Fig. 5 at various exposure times as indicated in the graph. The open circles are the experimental data, the solid lines are fits as described above, and the corresponding SLD profiles are shown as insets. The SLD of the Mg alloy layer increases ˚ 2 for the ascontinuously from the value of 2.7 106 A prepared layer with a noticeable deuterium gradient during the early stages of the deuterium absorption process (Fig. 4b,c and Fig. 5a).
a
b
3543
Fig. 6 e Measured neutron reflectivity curve (open circles) for a 55 nm thick Mg0.8Fe0.1Cr0.1 film capped with a FeCr/Pd bilayer in a deuterium pressure of 8 mbar after exposure of 84 min (same exp. data as in Fig. 4c). The solid line represents a fit based on a homogeneous SLD in the Mg alloy layer as shown in the inset.
The introduction of these deuterium gradients are necessary to fit the data. It is impossible to obtain a good fit for these NR curves assuming a Mg alloy layer with a constant SLD. This is shown in Fig. 6, where a fit (solid line) to the same experimental data as in Fig. 4c is shown assuming a homogeneous SLD in the MgeFeeCr layer. As can be seen, the fit with the homogeneous SLD cannot fit the data close to qc at all and has deficiencies at the second and fourth fringe. ˚ 2, The saturation value for the SLD amounts to 6.2 106 A calculated from an NR curve measured in 1 bar deuterium pressure at 300 K (see Fig. 8a). This corresponds to a D/M ratio of 1.57, which is (within the errors) equal to the maximum expected value of 1.6. Both, the SLD of the Pd and Ta layer ˚ 2 to increases during the absorption process, from 4.1 106 A 6 ˚ 2 6 ˚ 2 6 ˚ 2 5.2 10 A and from 3.6 10 A to 4.9 10 A ,
c
Fig. 5 e Measured neutron reflectivity curves (open circles) for a 55 nm thick Mg0.8Fe0.1Cr0.1 film capped with a FeCr/Pd bilayer in a deuterium pressure of 8 mbar after exposure of a) 110 min, b) 170 min, and c) 222 min. The solid lines represent fits corresponding to the SLD profiles shown in the inset.
Fig. 7 e Deuterium concentration in a 55 nm thick Mg0.8Fe0.1Cr0.1 film capped with a FeCr/Pd bilayer, displayed as a function of exposure time in 8 mbar D2. The deuterium concentration was determined from the fits to the NR curves shown in Figs. 4 and 5. The dashed line at c [ 1.6 corresponds to the maximum achievable D/M ratio.
3544
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 7 ( 2 0 1 2 ) 3 5 4 0 e3 5 4 7
respectively. This corresponds to a deuterium concentration of c ¼ 0.5 and c ¼ 0.4, respectively. The SLD of the FeCr layer ˚ 2 revealing that no deuterium is stays constant at 6 106 A stored in the FeCr layer. Interestingly, the film does not expand up to a concentration of c ¼ 0.45 (see Fig. 5a). The Mg alloy film expands by 25% in the fully saturated state. The deuterium concentration in the MgeFeeCr layer as a function of exposure time in 8 mbar D2 is displayed in Fig. 7. A perfectly linear increase of the deuterium concentration is observed up to D/M ¼ 0.45, i.e. in the range where we observe no expansion of the MgeFeeCr layer. Interestingly, once the layer starts to expand, i.e. MgD2 starts to form, the deuterium absorption rate increases substantially. The dashed line at c ¼ 1.6 marks the maximum achievable D/M ratio for a Mg0.8Fe0.1Cr0.1 layer as Fe and Cr do not form a deuteride. To desorb the film we first evacuated the sample cell at room temperature. The corresponding NR curve shown in Fig. 8b was measured 6 h after the evacuation. The SLD of the Mg0.8Fe0.1Cr0.1 layer stayed constant, whereas the SLD of the Pd ˚ 2 to 3.9 106 A ˚ 2, within layer decreased from 5.2 106 A the error the number for the as-prepared film. That means that the deuterium stays absorbed in the Mg alloy layer, whereas the Pd layer is being desorbed at room temperature.
Then we increased the temperature to 370 K. As qc was not changing, we increased the temperature further to 400 K. The NR curve measured after 7 h at 400 K is displayed in Fig. 8c. By comparing the NR curve to the fully absorbed sample in Fig. 8a, there are two obvious changes: (i) qc decreased substantially and (ii) the oscillation period increased. That means (i) that a considerable amount of deuterium was desorbed and (ii) that the film contracted back. From the fit we can conclude a remaining average deuterium concentration of c ¼ 0.06 in the Mg alloy film and a residual expansion of 1.5% compared to the as-prepared film. As the Pd and FeCr-layer thickness is the same (within errors) as in the as-prepared film, the remaining higher SLD seen in Fig. 8c cannot be attributed to the diffusion of atoms from the catalyst layer into the Mg alloy. The absorption and desorption behaviour of the Mg0.8Fe0.1Cr0.1 films capped with a Ta/Pd bilayer is very similar. From the NR curves shown in Fig. 9 we determined a 27% expansion in the fully absorbed state at 1 bar (Fig. 9b) with D/ M ¼ 1.5. The film could be completely desorbed at 400 K as can be seen from Fig. 9c. The film contracted back to its asprepared thickness - the same behaviour as we observed for the FeCr/Pd capped films.
a
a
b
b
c
c
Fig. 8 e Measured neutron reflectivity curves (open circles) for a 55 nm thick Mg0.8Fe0.1Cr0.1 film capped with a FeCr/Pd bilayer a) in a deuterium pressure of 1 bar, b) in vacuum at 300 K, and c) after 8 h at 400 K in a He pressure of 1 bar. The solid lines represent fits corresponding to the SLD profiles shown in the inset.
Fig. 9 e Measured neutron reflectivity curves (open circles) for a 52 nm thick Mg0.8Fe0.1Cr0.1 film capped with a Ta/Pd bilayer a) as-prepared, b) in deuterium pressure of 1 bar, measured at 300 K, and c) after 8 h at 400 K in a He pressure of 1 bar. The solid lines represent fits corresponding to the SLD profiles shown in the inset.
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 7 ( 2 0 1 2 ) 3 5 4 0 e3 5 4 7
4.
Absorption and desorption kinetics
Fig. 10 shows the absorption behaviour of a 1.5 mm thick Mg0.8Fe0.1Cr0.1 film capped with a (5 nm Ta/5 nm Pd) catalyst bilayer, measured at 200 C and a hydrogen pressure of 3 bar. For the first absorption it takes about 10 min to reach the fully absorbed state. However, an evident enhancement in absorption kinetics during the course of cycling was observed up to about 100 cycles. Thereafter, the absorption kinetics shows some deterioration. The desorption of the same film, also measured at 200 C, is displayed in Fig. 11. The first few desorption curves are relatively fast, taking almost 10 min to reach a fully desorbed state. However, the material shows a significant degradation in terms of kinetics for increasing cycle numbers up to 10 cycles. The desorption kinetics is fairly stable after this activation period, taking almost 20 min to achieve complete desorption.
5.
3545
Discussion
The continuous shift of qc as a function of deuterium exposure time nicely proves that the film continuously absorbs deuterium. However, the deuterium concentration is not homogeneous throughout the MgeFeeCr layer. The SLD profiles reveal that there is a deuterium concentration gradient in the MgeFeeCr film with a higher deuterium concentration towards the catalyst layer. The deuterium diffuses through the whole layer as evidenced by the continuous increase of the deuterium concentration at the MgeFeeCr/Ta interface and the fact that deuterium is being absorbed into the bottom Ta layer. Unlike observed in an earlier study on pure Mg films [27], a blocking MgD2 layer is not formed. So, similar to our previous study on MgeCreV alloys [27], we can conclude that the catalytic effect of the alloying is the prevention of the formation of a hydride diffusion barrier by the segregation of the FeCr. The TEM dark-field image at the bottom of Fig. 12 clearly shows the uniform dispersion of the FeCr precipitates, the TEM bright-field image of the same particle is shown
Fig. 10 e Absorption kinetics of a 1.5 mm thick Mg0.8Fe0.1Cr0.1 film capped with a Ta/Pd bilayer, measured at 200 C and a hydrogen pressure of 3 bar.
Fig. 11 e Desorption kinetics of a 1.5 mm thick Mg0.8Fe0.1Cr0.1 film capped with a Ta/Pd bilayer, measured at 200 C.
for comparison at the top of Fig. 12. These FeCr precipitates provide diffusion pathways for deuterium and nucleation sites for the formation of MgD2 similar to the MgeFeeTi system [11]. A slightly different mechanism, namely the
Fig. 12 e TEM pictures of a MgeFeeCr particle in the desorbed state after 140 cycles; top: bright-field image showing the particle, bottom: dark-field image showing the FeeCr precipitates as white spots.
3546
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 7 ( 2 0 1 2 ) 3 5 4 0 e3 5 4 7
formation of AleTi diffusion channels, was proposed for the fast kinetics of MgeAleTi [26,35]. We observed that the MgeFeeCr alloy film can absorb deuterium up to D/M ¼ 0.45 without expansion. This has also been observed in MgeCreV alloy films [27]. XRD measurements show that in the MgeCreV system the volume is contracted by only 4%, whereas Vegard’s law would result in a 10% volume contraction. So, it was argued by Kalisvaart et al. [27] that the constant film thickness up to D/M ¼ 0.5 might be the result of a balance between an expansion caused by deuterium uptake and a contraction caused by the segregation of V and Cr. This segregation had been concluded from XRD data taken in the fully absorbed state. From XRD data on MgeFeeCr films (not shown here) we can calculate a contraction of the molar volume of 12%, even more than the predicted 10% by Vegard’s law. So, a segregation of FeeCr at the early stages of the deuteration could not account for a volume contraction and compensate a potential film expansion. Our NR and XRD data on the MgeFeeCr system suggest that deuterium is stored in the alloy system as a solid solution up to D/M ¼ 0.45, and MgD2 is only formed at higher concentrations. The deuterium concentration as a function of time, displayed in Fig. 7, supports this interpretation because a linear increase in deuterium concentration is observed up to t ¼ 120 min and is followed by a much faster increase for t > 120 min, indicating the formation of MgD2.
6.
Conclusion
The results on Mg0:8 Fe0:1 Cr0:1 films show that the Mg0:8 Fe0:1 Cr0:1 alloy system represents a promising hydrogen storage system with (i) fast absorption kinetics at 200 C (<2 min after 140 cycles), ii) rapid desorption kinetics at 200 C (20 min after 140 cycles), iii) 5 wt% hydrogen storage capacity, and iv) a small degradation of the gravimetric storage capacity from 5 wt.% to 4.8 wt.% during cycling. Our NR data suggest that the catalytic effect of the Fe and Cr alloying is the prevention of the formation of a MgD2 layer. The FeeCr alloy allows the deuterium to diffuse through the whole Mg alloy film up to D/ M ¼ 0.45 before the formation of MgD2 sets in.
Acknowledgment We acknowledge the financial support of Natural Resources Canada within the framework of the Clean Energy Funds. The research presented herein is made possible by a reflectometer jointly funded by Canada Foundation for Innovation (CFI), Ontario Innovation Trust (OIT), Ontario Research Fund (ORF), and the National Research Council Canada (NRC).
references
[1] Dornheim M, Eigen N, Barkhordarian G, Klassen T, Bormann R. Tayloring hydrogen storage materials towards application. Adv Eng Mater 2006;8:377.
[2] Sakintuna B, Lamari-Darkrim F, Hirscher M. Metal hydride materials for solid hydrogen storage: a review. Int J Hydrogen Energy 2007;32:1121. [3] Huot J, Liang G, Boily S, Neste AV, Schulz R. Structural study and hydrogen sorption kinetics of ball-milled magnesium hydride. J Alloys Compd 1999;293-295:495. [4] Zaluska A, Zaluski L, Stro¨m-Olsen JO. Synergy of hydrogen sorption in ball-milled hydrides of Mg and Mg2Ni. J Alloys Compd 1999;289:197. [5] Orimo S, Ikeda K, Fujii H, Fujikawa Y, Kitano Y, Yamamoto K. Structural and hydriding properties of the MgeNieH system with nano- and/or amorphous structures. Acta Mater 1997; 45:2271. [6] Liang G, Huot J, Boily S, Schulz R. Hydrogen desorption kinetics of a mechanically milled MgH2 þ 5 at.% V nanocomposite. J Alloys Compd 2008;460:559. [7] Yonkeu AL, Swainson IP, Dufour J, Huot J. Kinetic investigation of the catalytic effect of a body centered cubicalloy TiV1.1MN0.9 (BCC) on hydriding/dehydriding properties of magnesium. J Alloys Compd 2000;305:239. [8] Oelerich W, Klassen T, Bormann R. Metal oxides as catalysts for improved hydrogen sorption in nanocrystalline Mg-based materials. J Alloys Compd 2001;315:237. [9] Shang CX, Bououdina M, Song Y, Guo ZX. Mechanical alloying and electronic simulations of (MgH2 þ M) systems (M ¼ Al, Ti, Fe, Ni, Cu and Nb) for hydrogen storage. Int J Hydrogen Energy 2004;29:73. [10] Zahiri B, Harrower CT, Amirkhiz BS, Mitlin D. Rapid and reversible hydrogen sorption in Mg-Fe-Ti thin films. Appl Phys Lett 2009;95:103114. [11] Amirkhiz BS, Zahiri B, Kalisvaart P, Mitlin D. Synergy of elemental Fe and Ti promoting low temperature hydrogen sorption cycling of magnesium. Int J Hydrogen Energy 2011; 36:6711e22. [12] Zahiri B, Amirkhiz BS, Danaie M, Mitlin D. Bimetallic Fe-V catalyzed magnesium films exhibiting rapid and cycleable hydrogenation at 200 C. Appl Phys Lett 2010a; 96:013108. [13] Zahiri B, Amirkhiz BS, Mitlin D. Hydrogen storage cycling of MgH2 thin film nanocomposites catalyzed by bimetallic Cr Ti. Appl Phys Lett 2010b;97:083106. [14] Pick MA, Davenport JW, Strongin M, Dienes GJ. Enhancement of hydrogen uptake rates for Nb and Ta by thin surface overlayers. Phys Rev Lett 1979;43:286. [15] Borgschulte A, Westerwaal RJ, Rector JH, Dam B, Griessen R, Schoenes J. Effect of the strong metal-support interaction on hydrogen sorption kinetics of Pd-capped switchable mirrors. Phys Rev B 2004;70:155414. [16] Fritzsche H, Saudi M, Haagsma J, Ophus C, Luber E, Harrower CT, et al. Neutron reflectometry study of hydrogen desorption in destabilized MgAl alloy thin films. Appl Phys Lett 2008;92:121917. [17] Harrower C, Poirier E, Fritzsche H, Kalisvaart P, Satija S, Akgun B, et al. Early deuteration steps of Pd- and Ta/Pdcatalyzed Mg70Al30 thin films observed at room temperature. Int J Hydrogen Energy 2010;35:10343e8. [18] Poirier E, Harrower CT, Kalisvaart P, Bird A, Teichert A, Wallacher D, et al. Deuterium absorption in Mg70Al30 thin films with bilayer catalysts: a comparative neutron reflectometry study. J Alloys Compd 2011;509:5468. [19] Remhof A, Song G, Theis-Bro¨hl K, Zabel H. Reversible loading of epitaxial Y(00.1) films with hydrogen. Phys Rev B 1997;56: R2897. [20] Baldi A, Gonzalez-Silveira M, Palmisano V, Dam B, Griessen R. Destabilization of the MgeH system through elastic constraints. Phys Rev Lett 2009;102:226102. [21] Tan XH, Harrower CT, Amirkhiz BS, Mitlin D. Nano-scale bilayer Pd/Ta, Pd/Nb, Pd/Ti and Pd/Fe catalysts for hydrogen
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 7 ( 2 0 1 2 ) 3 5 4 0 e3 5 4 7
[22]
[23]
[24]
[25]
[26]
sorption in magnesium thin films. Int J Hydrogen Energy 2009;34:7741. Fritzsche H, Ophus C, Harrower CT, Luber E, Mitlin D. Low temperature hydrogen desorption in MgAl thin films achieved by using a nanoscale Ta/Pd bilayer catalyst. Appl Phys Lett 2009a;94:241901. Pasturel M, Wijngaarden RJ, Lohstroh W, Schreuders H, Slaman M, Dam B, et al. Influence of the chemical potential on the hydrogen sorption kinetics of MgNi/TM/Pd (TM ¼ transition metal) trilayers. Chem Mater 2007;19:624. Gremaud R, Borgschulte A, Lohstroh W, Schreuders H, Zu¨ttel A, Dam B, et al. Ti-catalyzed Mg(AlH4)2: a reversible hydrogen storage material. J Alloys Compd 2005;404e406: 775. Fritzsche H, Saudi M, Haagsma J, Ophus C, Harrower CT, Mitlin D. Structural changes of thin MgAl films during hydrogen desorption. Nucl Instr Meth A 2009b;600:301. Kalisvaart WP, Harrower CT, Haagsma J, Zahiri B, Luber EJ, Ophus C, et al. Hydrogen storage in binary and ternary Mgbased alloys: a comprehensive experimental study. Int J Hydrogen Energy 2010;35:2091.
3547
[27] Kalisvaart P, Luber E, Fritzsche H, Mitlin D. Effect of alloying magnesium with chromium and vanadium on hydrogenation kinetics studied with neutron reflectometry. Chem Commun 2011;47:4294. [28] Bergsma J, van Dijk C. Pyrolytic graphite as a second order neutron filter. Nucl Instr Meth 1967;51:121. [29] Zhou XL, Chen SH. Theoretical foundation of x-ray and neutron reflectometry. Phys Rep 1995;257:223. [30] Sears VF. Neutron scattering lengths and cross sections. Neutron News 1992;3:26. [31] www.nist.gov, 2003. [32] Parratt LG. Surface studies of solids by total reflection of x-rays. Phys Rev 1954;95:359. [33] Rehm C, Fritzsche H, Maletta H, Klose F. Hydrogen concentration and its relation to interplanar spacing and ˚ Nb(110) films during in situ layer thickness of 1000 A hydrogen charging experiments. Phys Rev B 1999;59:3142. [34] Kiessig H. Interferenz von Ro¨ntgenstrahlen an du¨nnen Schichten. Ann Physik 1931;402:769. [35] Hao S. The TiAl mechanism for enhanced (de)hydrogenation kinetics in Mg-based films. Appl Phys Lett 2010;97:111905.
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/he
The catalytic effect of Fe and Cr on hydrogen and deuterium absorption in Mg thin films H. Fritzsche a,*, W.P. Kalisvaart b, B. Zahiri b, R. Flacau a, D. Mitlin b a
National Research Council Canada, SIMS, Canadian Neutron Beam Centre, Chalk River Laboratories, building 459, Chalk River, Ontario K0J 1J0, Canada b Chemical and Materials Engineering, University of Alberta, T6G 2V4, and National Research Council Canada, National Institute for Nanotechnology, T6G 2M9, Edmonton, Alberta, Canada
article info
abstract
Article history:
We examined the deuterium absorption and desorption of 55 nm thick Mg films alloyed
Received 27 April 2011
with Fe and Cr using in-situ neutron reflectometry. The Mg alloy films were covered with
Received in revised form
bimetallic catalyst layers and could be fully absorbed at room temperature at a pressure of
31 May 2011
8 mbar. The NR experiments revealed a deuterium gradient within the Mg alloy layers
Accepted 2 June 2011
during absorption and a large deuterium uptake up to a D/M ratio of about 0.45 before the
Available online 1 September 2011
layer started to expand and form magnesium deuteride (MgD2). Our NR data suggest that the catalytic effect of the FeeCr alloy is based on the avoidance of the formation of
Keywords:
a blocking MgD2 layer in the early stages of the absorption process leading to a fast
Low temperature hydrogen
hydrogen absorption kinetics.
absorption
Crown Copyright ª 2011, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All
Magnesium
rights reserved.
Thin film Neutron reflectometry Bilayer catalysts
1.
Introduction
MgH2 is a very attractive hydrogen storage material because it possesses a high gravimetric storage density of 7.6 wt.% and is relatively inexpensive. However, the temperatures needed to overcome the slow kinetics are too high for most practical applications. Therefore, a lot of experimental work has been done to improve the sorption kinetics and lower the temperatures needed for absorption and desorption. Most approaches consisted of using high energy ball milling to reduce the grain sizes and adding metals or metal oxides as catalysts [1e9]. Especially combinations of 3d metals like FeeTi [10,11], FeeV [12], and CreTi [13] showed very fast sorption kinetics when alloyed with Mg.
In the last years research on thin films has attracted a lot of attention. Using thin films as model systems makes it possible to discriminate between the catalytic surface effect of a cap layer on top of a Mg film and the catalytic bulk effect of alloying the Mg film with other elements. These thin film systems have the big advantage that they can be fabricated in a controlled and reproducible way so that e.g. a Mg film can be completely covered with a surface catalyst layer. A 5 nm thick Pd cap layer has turned out to be sufficient to avoid oxidation of the hydrogen storage film underneath and to dissociate the hydrogen molecules [14e16]. As a consequence, thin films do not need to be activated at high temperature and they always show much better sorption behaviour than the corresponding bulk material. Thin films
* Corresponding author. E-mail address: [email protected] (H. Fritzsche). 0360-3199/$ e see front matter Crown Copyright ª 2011, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.ijhydene.2011.06.014
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 7 ( 2 0 1 2 ) 3 5 4 0 e3 5 4 7
absorb hydrogen fast even at room temperature and hydrogen pressures below 1 bar [15,17,18]. Using the thin film approach it was possible to even improve the performance of a single Pd layer by adding another layer between the Pd cap layer and the Mg film. Bilayer catalysts like Nb/Pd [19e21], Ta/Pd [17,22], V/ Pd [20,23], and Ti/Pd [18,20,21,23,24] improved the absorption and desorption kinetics substantially. The second catalyst layer acts as a diffusion barrier and avoids the formation of a MgePd alloy as was observed in a previous study [25]. Compared to the brute force intermixing of hydrogen storage materials with catalysts caused by ball milling, coevaporation during thin film sputtering enables one to fabricate well-defined alloy compositions and modify only the bulk storage material leaving the surface catalyst layer unchanged. Using this approach, several classes of readily cyclable ternary alloys with rapid low temperature kinetics were discovered: MgeFeeTi [26], MgeAleTi [26], MgeFeeV [12], MgeCreTi [13], and MgeCreV [27]. Here we present a new alloy, MgeFeeCr, which shows rapid absorption and desorption kinetics. We applied in-situ Neutron Reflectometry (NR) to determine the deuterium concentration in the film structure along with the change of the film structure during the absorption and desorption. With its high sensitivity to deuterium and its nanometer resolution along the film normal, NR is the technique of choice to determine the deuterium concentration profile in thin films. Whereas PcT curves and sorption cycling experiments only measure the total hydrogen content, NR gives deeper insight into the absorption and desorption process by revealing hydrogen or deuterium gradients in thin films with subnanometer precision.
2.
Experimental
The thin films were prepared by co-sputtering onto a Si(100) wafer in a confocal sputtering chamber (Orion 5 instrument from AJA International) operating at an Ar (purity 99.999%) pressure of 5 103 mbar which had been previously evacuated to a pressure less than 3 108 mbar. For the NR experiments, first a 10 nm Ta buffer layer was deposited onto the wafer and, without interruption, a 50 nm Mg0.8Fe0.1Cr0.1 alloy layer was co-sputtered followed by either a (5 nm Ta/5 nm Pd) or a (5 nm FeCr/5 nm Pd) bilayer. For the volumetric measurements, which were always performed at 200 C, a 1.5 mm thick Mg0.8Fe0.1Cr0.1 film was used with a (5 nm Ta/5 nm Pd) bilayer both, at the top and bottom of the Mg alloy layer. Hydrogen was absorbed at a starting pressure of 3 bar and desorbed at a starting pressure of 0.005 bar. The system (HyEnergy PCTPro 2000) automatically switched from absorption to desorption, and vice versa, once the sorption rate fell below 0.004 wt%/min. Transmission electron microscopy was performed using the JEOL JEM-2100 microscope, operating at 200 kV acceleration voltage. The sample was cycled for 140 times before taking it out in the desorbed state. The neutron reflectometry experiments were performed on the D3 reflectometer at the neutron research reactor NRU in Chalk River. We used a focusing pyrolytic graphite (PG) monochromator at a neutron wavelength of l ¼ 0.237 nm
3541
along with a PG filter [28] to reduce the higher order contamination (l/2 and l/3). In an NR experiment the neutron beam hits the surface of a sample at the scattering angle q and is specularly reflected off the sample at the same angle, which is typically in the range between 0 and 2 . The interfaces of the samples are arranged perpendicular to the scattering vector. ! ! 4p j! q j ¼ kr ki ¼ sinq (1) l ! ! with kr and ki being the reflected and incoming neutron wave vector. At grazing incidence the interaction of the neutron with the sample can be described with a neutron index of refraction [29] analogous to optical reflectivity. The neutron index of refraction depends on the strength of the interaction of neutrons with a specific isotope in the film and can be represented by n2 ¼ 1
l2 Nj bj ; p
(2)
where Nj is the number density and bj the nuclear scattering length [30,31] of the elements/isotopes in layer j. The product Njbj is called scattering length density (SLD). We used the software PARRATT32, which is based on the Parratt recursion algorithm [32], to fit our NR data by varying the SLD, layer thickness, and interface roughness of each individual layer j. In our NR experiments we used deuterium because of its large coherent scattering length, which leads to a dramatic increase in the SLD of the absorbing layer, thus leading to a change in the index of refraction and finally in the measured reflectivity curve. Total reflection of neutrons occurs up to a critical scattering vector qc: pffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffi qc ¼ 4 pNb ¼ 4 pS:
(3)
The SLD S of our Mg-based alloy films is larger than the SLD of the Si substrate. Therefore, the increase of qc is a good measure of the average deuterium content absorbed by the film. The deuterium concentration cD in D/metal can be calculated as follows [33]: cD ¼
SMþD tMþD bM , 1 : SM tM bD
(4)
SM and SMþD are the SLD of the as-prepared and deuterium loaded film, respectively, with the corresponding film thicknesses tM and tMþD, the average scattering length of the film bM, and the scattering length of deuterium [30,31] bD ¼ 6.67 fm.
3.
Neutron reflectometry data
The NR curve of an as-prepared 55 nm thick Mg0.8Fe0.1Cr0.1 film capped with a (5 nm FeCr/5 nm Pd) bilayer catalyst is shown in Fig. 1. The open circles are the measured data points, the solid line is the fit based on the SLD profile depicted in the inset. The reflectivity curve contains a series of maxima and minima, the so-called Kiessig fringes [34], which correspond to characteristic thicknesses in the specimen. The oscillations in Fig. 1 are due to our sample’s total film thickness of 76 nm. The SLD profile provides a real space representation of the film and is
3542
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 7 ( 2 0 1 2 ) 3 5 4 0 e3 5 4 7
Fig. 1 e Measured neutron reflectivity curve (open circles) as a function of the scattering vector q for a 55 nm Mg0.8Fe0.1Cr0.1 film capped with a FeCr/Pd bilayer. The solid line represents a fit corresponding to the SLD profile shown in the inset.
plotted as a function of z, the film normal, with z ¼ 0 at the film surface. The SLD profile shows the film structure with ˚ 2) followed by a 5.3 nm Pd surface layer (SLD ¼ 4.1 106 A 6 ˚ 2 A ), a 54.7 nm a 5.4 nm FeCr layer (SLD ¼ 6.0 10 ˚ 2), a 10.7 nm Ta layer Mg0.8Fe0.1Cr0.1 layer (SLD ¼ 2.7 106 A ˚ 2), and a bulk Si crystal (SLD ¼ 2.1 106 (SLD ¼ 3.6 106 A 2 ˚ A ). The SLD values determined from the fit are within 5% of the calculated values using tabulated data [30,31]. The kinetics of the deuterium absorption can be easily and quickly monitored by measuring the NR curve in a narrow q range around the critical scattering vector qc. This is displayed in Fig. 2, where NR curves of the sample with the FeCr/Pd catalyst bilayer are plotted at different times after a deuterium pressure of 8 mbar was introduced into the sample cell at t ¼ 0 at 300 K. The critical scattering vector qc increases continuously with time until reaching saturation after about 4 h. This shift of qc can be taken as a direct measure of deuterium absorption because according to Eq. (3) the deuterium
Fig. 3 e Neutron reflectivity curves close to the critical scattering vector qc for a 52 nm thick Mg0.8Fe0.1Cr0.1 film capped with a Ta/Pd bilayer, measured at room temperature in a deuterium pressure of 10 mbar for different exposure times.
a
b
c
Fig. 2 e Neutron reflectivity curves close to the critical scattering vector qc for a 55 nm thick Mg0.8Fe0.1Cr0.1 film capped with a FeCr/Pd bilayer, measured at room temperature in a deuterium pressure of 8 mbar for different exposure times.
Fig. 4 e Measured neutron reflectivity curves (open circles) for a 55 nm thick Mg0.8Fe0.1Cr0.1 film capped with a FeCr/Pd bilayer in a deuterium pressure of 8 mbar after exposure of a) 20 min, b) 57 min, and c) 84 min. The solid lines represent fits corresponding to the SLD profiles shown in the inset.
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 7 ( 2 0 1 2 ) 3 5 4 0 e3 5 4 7
absorption leads to an increase of the SLD in the Mg alloy layer due to the large scattering length of deuterium. The reflectivity curve measured after 222 min exposure time corresponds to a D/M ratio of c ¼ 1.5, very close to the maximum expected saturation value of 1.6, calculated under the assumption that all Mg atoms react to form MgD2. We also performed the same experiments for the samples with a Ta/Pd bilayer, however, in a slightly higher deuterium pressure of 10 mbar. The data are plotted in Fig. 3. Both samples show similar behaviour, the slightly faster kinetics of the Ta/Pd catalyzed sample is most likely due to the slightly higher pressure of 10 mbar for the Ta/Pd sample. The full range of the measured NR curves for the films with the FeCr/Pd bilayer catalyst is shown in Fig. 4 and Fig. 5 at various exposure times as indicated in the graph. The open circles are the experimental data, the solid lines are fits as described above, and the corresponding SLD profiles are shown as insets. The SLD of the Mg alloy layer increases ˚ 2 for the ascontinuously from the value of 2.7 106 A prepared layer with a noticeable deuterium gradient during the early stages of the deuterium absorption process (Fig. 4b,c and Fig. 5a).
a
b
3543
Fig. 6 e Measured neutron reflectivity curve (open circles) for a 55 nm thick Mg0.8Fe0.1Cr0.1 film capped with a FeCr/Pd bilayer in a deuterium pressure of 8 mbar after exposure of 84 min (same exp. data as in Fig. 4c). The solid line represents a fit based on a homogeneous SLD in the Mg alloy layer as shown in the inset.
The introduction of these deuterium gradients are necessary to fit the data. It is impossible to obtain a good fit for these NR curves assuming a Mg alloy layer with a constant SLD. This is shown in Fig. 6, where a fit (solid line) to the same experimental data as in Fig. 4c is shown assuming a homogeneous SLD in the MgeFeeCr layer. As can be seen, the fit with the homogeneous SLD cannot fit the data close to qc at all and has deficiencies at the second and fourth fringe. ˚ 2, The saturation value for the SLD amounts to 6.2 106 A calculated from an NR curve measured in 1 bar deuterium pressure at 300 K (see Fig. 8a). This corresponds to a D/M ratio of 1.57, which is (within the errors) equal to the maximum expected value of 1.6. Both, the SLD of the Pd and Ta layer ˚ 2 to increases during the absorption process, from 4.1 106 A 6 ˚ 2 6 ˚ 2 6 ˚ 2 5.2 10 A and from 3.6 10 A to 4.9 10 A ,
c
Fig. 5 e Measured neutron reflectivity curves (open circles) for a 55 nm thick Mg0.8Fe0.1Cr0.1 film capped with a FeCr/Pd bilayer in a deuterium pressure of 8 mbar after exposure of a) 110 min, b) 170 min, and c) 222 min. The solid lines represent fits corresponding to the SLD profiles shown in the inset.
Fig. 7 e Deuterium concentration in a 55 nm thick Mg0.8Fe0.1Cr0.1 film capped with a FeCr/Pd bilayer, displayed as a function of exposure time in 8 mbar D2. The deuterium concentration was determined from the fits to the NR curves shown in Figs. 4 and 5. The dashed line at c [ 1.6 corresponds to the maximum achievable D/M ratio.
3544
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 7 ( 2 0 1 2 ) 3 5 4 0 e3 5 4 7
respectively. This corresponds to a deuterium concentration of c ¼ 0.5 and c ¼ 0.4, respectively. The SLD of the FeCr layer ˚ 2 revealing that no deuterium is stays constant at 6 106 A stored in the FeCr layer. Interestingly, the film does not expand up to a concentration of c ¼ 0.45 (see Fig. 5a). The Mg alloy film expands by 25% in the fully saturated state. The deuterium concentration in the MgeFeeCr layer as a function of exposure time in 8 mbar D2 is displayed in Fig. 7. A perfectly linear increase of the deuterium concentration is observed up to D/M ¼ 0.45, i.e. in the range where we observe no expansion of the MgeFeeCr layer. Interestingly, once the layer starts to expand, i.e. MgD2 starts to form, the deuterium absorption rate increases substantially. The dashed line at c ¼ 1.6 marks the maximum achievable D/M ratio for a Mg0.8Fe0.1Cr0.1 layer as Fe and Cr do not form a deuteride. To desorb the film we first evacuated the sample cell at room temperature. The corresponding NR curve shown in Fig. 8b was measured 6 h after the evacuation. The SLD of the Mg0.8Fe0.1Cr0.1 layer stayed constant, whereas the SLD of the Pd ˚ 2 to 3.9 106 A ˚ 2, within layer decreased from 5.2 106 A the error the number for the as-prepared film. That means that the deuterium stays absorbed in the Mg alloy layer, whereas the Pd layer is being desorbed at room temperature.
Then we increased the temperature to 370 K. As qc was not changing, we increased the temperature further to 400 K. The NR curve measured after 7 h at 400 K is displayed in Fig. 8c. By comparing the NR curve to the fully absorbed sample in Fig. 8a, there are two obvious changes: (i) qc decreased substantially and (ii) the oscillation period increased. That means (i) that a considerable amount of deuterium was desorbed and (ii) that the film contracted back. From the fit we can conclude a remaining average deuterium concentration of c ¼ 0.06 in the Mg alloy film and a residual expansion of 1.5% compared to the as-prepared film. As the Pd and FeCr-layer thickness is the same (within errors) as in the as-prepared film, the remaining higher SLD seen in Fig. 8c cannot be attributed to the diffusion of atoms from the catalyst layer into the Mg alloy. The absorption and desorption behaviour of the Mg0.8Fe0.1Cr0.1 films capped with a Ta/Pd bilayer is very similar. From the NR curves shown in Fig. 9 we determined a 27% expansion in the fully absorbed state at 1 bar (Fig. 9b) with D/ M ¼ 1.5. The film could be completely desorbed at 400 K as can be seen from Fig. 9c. The film contracted back to its asprepared thickness - the same behaviour as we observed for the FeCr/Pd capped films.
a
a
b
b
c
c
Fig. 8 e Measured neutron reflectivity curves (open circles) for a 55 nm thick Mg0.8Fe0.1Cr0.1 film capped with a FeCr/Pd bilayer a) in a deuterium pressure of 1 bar, b) in vacuum at 300 K, and c) after 8 h at 400 K in a He pressure of 1 bar. The solid lines represent fits corresponding to the SLD profiles shown in the inset.
Fig. 9 e Measured neutron reflectivity curves (open circles) for a 52 nm thick Mg0.8Fe0.1Cr0.1 film capped with a Ta/Pd bilayer a) as-prepared, b) in deuterium pressure of 1 bar, measured at 300 K, and c) after 8 h at 400 K in a He pressure of 1 bar. The solid lines represent fits corresponding to the SLD profiles shown in the inset.
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 7 ( 2 0 1 2 ) 3 5 4 0 e3 5 4 7
4.
Absorption and desorption kinetics
Fig. 10 shows the absorption behaviour of a 1.5 mm thick Mg0.8Fe0.1Cr0.1 film capped with a (5 nm Ta/5 nm Pd) catalyst bilayer, measured at 200 C and a hydrogen pressure of 3 bar. For the first absorption it takes about 10 min to reach the fully absorbed state. However, an evident enhancement in absorption kinetics during the course of cycling was observed up to about 100 cycles. Thereafter, the absorption kinetics shows some deterioration. The desorption of the same film, also measured at 200 C, is displayed in Fig. 11. The first few desorption curves are relatively fast, taking almost 10 min to reach a fully desorbed state. However, the material shows a significant degradation in terms of kinetics for increasing cycle numbers up to 10 cycles. The desorption kinetics is fairly stable after this activation period, taking almost 20 min to achieve complete desorption.
5.
3545
Discussion
The continuous shift of qc as a function of deuterium exposure time nicely proves that the film continuously absorbs deuterium. However, the deuterium concentration is not homogeneous throughout the MgeFeeCr layer. The SLD profiles reveal that there is a deuterium concentration gradient in the MgeFeeCr film with a higher deuterium concentration towards the catalyst layer. The deuterium diffuses through the whole layer as evidenced by the continuous increase of the deuterium concentration at the MgeFeeCr/Ta interface and the fact that deuterium is being absorbed into the bottom Ta layer. Unlike observed in an earlier study on pure Mg films [27], a blocking MgD2 layer is not formed. So, similar to our previous study on MgeCreV alloys [27], we can conclude that the catalytic effect of the alloying is the prevention of the formation of a hydride diffusion barrier by the segregation of the FeCr. The TEM dark-field image at the bottom of Fig. 12 clearly shows the uniform dispersion of the FeCr precipitates, the TEM bright-field image of the same particle is shown
Fig. 10 e Absorption kinetics of a 1.5 mm thick Mg0.8Fe0.1Cr0.1 film capped with a Ta/Pd bilayer, measured at 200 C and a hydrogen pressure of 3 bar.
Fig. 11 e Desorption kinetics of a 1.5 mm thick Mg0.8Fe0.1Cr0.1 film capped with a Ta/Pd bilayer, measured at 200 C.
for comparison at the top of Fig. 12. These FeCr precipitates provide diffusion pathways for deuterium and nucleation sites for the formation of MgD2 similar to the MgeFeeTi system [11]. A slightly different mechanism, namely the
Fig. 12 e TEM pictures of a MgeFeeCr particle in the desorbed state after 140 cycles; top: bright-field image showing the particle, bottom: dark-field image showing the FeeCr precipitates as white spots.
3546
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 7 ( 2 0 1 2 ) 3 5 4 0 e3 5 4 7
formation of AleTi diffusion channels, was proposed for the fast kinetics of MgeAleTi [26,35]. We observed that the MgeFeeCr alloy film can absorb deuterium up to D/M ¼ 0.45 without expansion. This has also been observed in MgeCreV alloy films [27]. XRD measurements show that in the MgeCreV system the volume is contracted by only 4%, whereas Vegard’s law would result in a 10% volume contraction. So, it was argued by Kalisvaart et al. [27] that the constant film thickness up to D/M ¼ 0.5 might be the result of a balance between an expansion caused by deuterium uptake and a contraction caused by the segregation of V and Cr. This segregation had been concluded from XRD data taken in the fully absorbed state. From XRD data on MgeFeeCr films (not shown here) we can calculate a contraction of the molar volume of 12%, even more than the predicted 10% by Vegard’s law. So, a segregation of FeeCr at the early stages of the deuteration could not account for a volume contraction and compensate a potential film expansion. Our NR and XRD data on the MgeFeeCr system suggest that deuterium is stored in the alloy system as a solid solution up to D/M ¼ 0.45, and MgD2 is only formed at higher concentrations. The deuterium concentration as a function of time, displayed in Fig. 7, supports this interpretation because a linear increase in deuterium concentration is observed up to t ¼ 120 min and is followed by a much faster increase for t > 120 min, indicating the formation of MgD2.
6.
Conclusion
The results on Mg0:8 Fe0:1 Cr0:1 films show that the Mg0:8 Fe0:1 Cr0:1 alloy system represents a promising hydrogen storage system with (i) fast absorption kinetics at 200 C (<2 min after 140 cycles), ii) rapid desorption kinetics at 200 C (20 min after 140 cycles), iii) 5 wt% hydrogen storage capacity, and iv) a small degradation of the gravimetric storage capacity from 5 wt.% to 4.8 wt.% during cycling. Our NR data suggest that the catalytic effect of the Fe and Cr alloying is the prevention of the formation of a MgD2 layer. The FeeCr alloy allows the deuterium to diffuse through the whole Mg alloy film up to D/ M ¼ 0.45 before the formation of MgD2 sets in.
Acknowledgment We acknowledge the financial support of Natural Resources Canada within the framework of the Clean Energy Funds. The research presented herein is made possible by a reflectometer jointly funded by Canada Foundation for Innovation (CFI), Ontario Innovation Trust (OIT), Ontario Research Fund (ORF), and the National Research Council Canada (NRC).
references
[1] Dornheim M, Eigen N, Barkhordarian G, Klassen T, Bormann R. Tayloring hydrogen storage materials towards application. Adv Eng Mater 2006;8:377.
[2] Sakintuna B, Lamari-Darkrim F, Hirscher M. Metal hydride materials for solid hydrogen storage: a review. Int J Hydrogen Energy 2007;32:1121. [3] Huot J, Liang G, Boily S, Neste AV, Schulz R. Structural study and hydrogen sorption kinetics of ball-milled magnesium hydride. J Alloys Compd 1999;293-295:495. [4] Zaluska A, Zaluski L, Stro¨m-Olsen JO. Synergy of hydrogen sorption in ball-milled hydrides of Mg and Mg2Ni. J Alloys Compd 1999;289:197. [5] Orimo S, Ikeda K, Fujii H, Fujikawa Y, Kitano Y, Yamamoto K. Structural and hydriding properties of the MgeNieH system with nano- and/or amorphous structures. Acta Mater 1997; 45:2271. [6] Liang G, Huot J, Boily S, Schulz R. Hydrogen desorption kinetics of a mechanically milled MgH2 þ 5 at.% V nanocomposite. J Alloys Compd 2008;460:559. [7] Yonkeu AL, Swainson IP, Dufour J, Huot J. Kinetic investigation of the catalytic effect of a body centered cubicalloy TiV1.1MN0.9 (BCC) on hydriding/dehydriding properties of magnesium. J Alloys Compd 2000;305:239. [8] Oelerich W, Klassen T, Bormann R. Metal oxides as catalysts for improved hydrogen sorption in nanocrystalline Mg-based materials. J Alloys Compd 2001;315:237. [9] Shang CX, Bououdina M, Song Y, Guo ZX. Mechanical alloying and electronic simulations of (MgH2 þ M) systems (M ¼ Al, Ti, Fe, Ni, Cu and Nb) for hydrogen storage. Int J Hydrogen Energy 2004;29:73. [10] Zahiri B, Harrower CT, Amirkhiz BS, Mitlin D. Rapid and reversible hydrogen sorption in Mg-Fe-Ti thin films. Appl Phys Lett 2009;95:103114. [11] Amirkhiz BS, Zahiri B, Kalisvaart P, Mitlin D. Synergy of elemental Fe and Ti promoting low temperature hydrogen sorption cycling of magnesium. Int J Hydrogen Energy 2011; 36:6711e22. [12] Zahiri B, Amirkhiz BS, Danaie M, Mitlin D. Bimetallic Fe-V catalyzed magnesium films exhibiting rapid and cycleable hydrogenation at 200 C. Appl Phys Lett 2010a; 96:013108. [13] Zahiri B, Amirkhiz BS, Mitlin D. Hydrogen storage cycling of MgH2 thin film nanocomposites catalyzed by bimetallic Cr Ti. Appl Phys Lett 2010b;97:083106. [14] Pick MA, Davenport JW, Strongin M, Dienes GJ. Enhancement of hydrogen uptake rates for Nb and Ta by thin surface overlayers. Phys Rev Lett 1979;43:286. [15] Borgschulte A, Westerwaal RJ, Rector JH, Dam B, Griessen R, Schoenes J. Effect of the strong metal-support interaction on hydrogen sorption kinetics of Pd-capped switchable mirrors. Phys Rev B 2004;70:155414. [16] Fritzsche H, Saudi M, Haagsma J, Ophus C, Luber E, Harrower CT, et al. Neutron reflectometry study of hydrogen desorption in destabilized MgAl alloy thin films. Appl Phys Lett 2008;92:121917. [17] Harrower C, Poirier E, Fritzsche H, Kalisvaart P, Satija S, Akgun B, et al. Early deuteration steps of Pd- and Ta/Pdcatalyzed Mg70Al30 thin films observed at room temperature. Int J Hydrogen Energy 2010;35:10343e8. [18] Poirier E, Harrower CT, Kalisvaart P, Bird A, Teichert A, Wallacher D, et al. Deuterium absorption in Mg70Al30 thin films with bilayer catalysts: a comparative neutron reflectometry study. J Alloys Compd 2011;509:5468. [19] Remhof A, Song G, Theis-Bro¨hl K, Zabel H. Reversible loading of epitaxial Y(00.1) films with hydrogen. Phys Rev B 1997;56: R2897. [20] Baldi A, Gonzalez-Silveira M, Palmisano V, Dam B, Griessen R. Destabilization of the MgeH system through elastic constraints. Phys Rev Lett 2009;102:226102. [21] Tan XH, Harrower CT, Amirkhiz BS, Mitlin D. Nano-scale bilayer Pd/Ta, Pd/Nb, Pd/Ti and Pd/Fe catalysts for hydrogen
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 7 ( 2 0 1 2 ) 3 5 4 0 e3 5 4 7
[22]
[23]
[24]
[25]
[26]
sorption in magnesium thin films. Int J Hydrogen Energy 2009;34:7741. Fritzsche H, Ophus C, Harrower CT, Luber E, Mitlin D. Low temperature hydrogen desorption in MgAl thin films achieved by using a nanoscale Ta/Pd bilayer catalyst. Appl Phys Lett 2009a;94:241901. Pasturel M, Wijngaarden RJ, Lohstroh W, Schreuders H, Slaman M, Dam B, et al. Influence of the chemical potential on the hydrogen sorption kinetics of MgNi/TM/Pd (TM ¼ transition metal) trilayers. Chem Mater 2007;19:624. Gremaud R, Borgschulte A, Lohstroh W, Schreuders H, Zu¨ttel A, Dam B, et al. Ti-catalyzed Mg(AlH4)2: a reversible hydrogen storage material. J Alloys Compd 2005;404e406: 775. Fritzsche H, Saudi M, Haagsma J, Ophus C, Harrower CT, Mitlin D. Structural changes of thin MgAl films during hydrogen desorption. Nucl Instr Meth A 2009b;600:301. Kalisvaart WP, Harrower CT, Haagsma J, Zahiri B, Luber EJ, Ophus C, et al. Hydrogen storage in binary and ternary Mgbased alloys: a comprehensive experimental study. Int J Hydrogen Energy 2010;35:2091.
3547
[27] Kalisvaart P, Luber E, Fritzsche H, Mitlin D. Effect of alloying magnesium with chromium and vanadium on hydrogenation kinetics studied with neutron reflectometry. Chem Commun 2011;47:4294. [28] Bergsma J, van Dijk C. Pyrolytic graphite as a second order neutron filter. Nucl Instr Meth 1967;51:121. [29] Zhou XL, Chen SH. Theoretical foundation of x-ray and neutron reflectometry. Phys Rep 1995;257:223. [30] Sears VF. Neutron scattering lengths and cross sections. Neutron News 1992;3:26. [31] www.nist.gov, 2003. [32] Parratt LG. Surface studies of solids by total reflection of x-rays. Phys Rev 1954;95:359. [33] Rehm C, Fritzsche H, Maletta H, Klose F. Hydrogen concentration and its relation to interplanar spacing and ˚ Nb(110) films during in situ layer thickness of 1000 A hydrogen charging experiments. Phys Rev B 1999;59:3142. [34] Kiessig H. Interferenz von Ro¨ntgenstrahlen an du¨nnen Schichten. Ann Physik 1931;402:769. [35] Hao S. The TiAl mechanism for enhanced (de)hydrogenation kinetics in Mg-based films. Appl Phys Lett 2010;97:111905.