Journal of Alloys and Compounds 400 (2005) 188–193
Hydriding characteristics of FeTi/Pd films A.M. Vredenberg a,∗ , E.M.B. Heller a,1 , D.O. Boerma b a
Debye Institute, Surfaces, Interfaces and Devices, Universiteit Utrecht, P.O. Box 80.000, 3508 TA Utrecht, The Netherlands Centro de Micro-An´alisis, Universidad Aut´onoma de Madrid, Cantoblanco, 28049 Madrid, Spain Received 15 March 2005; accepted 31 March 2005 Available online 25 May 2005
Abstract The hydrogen storage properties of thin films of FeTi evaporated on Si substrates and covered with 20 nm Pd were studied. The films serve as a model system for powdered FeTi, with grains that are (partly) covered with Pd. This material could serve as a practical hydrogen storage material. The 20 nm Pd layer prevents the oxidation of the FeTi layer during air exposure up to temperatures of 200 ◦ C and during H charging and discharging in impure hydrogen. The FeTi is a mixture of amorphous and nano-crystalline material. Two FeTi compositions (43 at.% Fe, i.e. Ti-rich, and 56 at.% Fe, i.e. Fe-rich) were studied. The H charging and discharging characteristics as a function of temperature and pressure are determined from a differential pressure measurement for Fe and Ti-rich material before and after annealing. After discharging in vacuum at a temperature of 150 ◦ C a H residue of H/M ∼ 0.12 is observed. The recoverable charging capacity of FeTi (also after many cycles) is 0.9 H/M (H atoms per metal atom) for RT charging at 2700 mbar and vacuum discharging at 150 ◦ C. © 2005 Elsevier B.V. All rights reserved. Keywords: Hydrogen storage materials; Thin films; Intermetallics; Disordered systems; Gas–solid reactions
1. Introduction To exploit the hydrogen storage properties of FeTi, the problems of corrosion and embrittlement must be resolved. The corrosion resistance of the material can be improved by coating the surface with a H2 permeable, non-corrosive material [1,2]. Without such protection, air-exposed FeTi is covered with a surface oxide that blocks the uptake of hydrogen. A high temperature activation treatment can dissolve this oxide, but the material remains very reactive to impurities in the H2 gas (H2 O and O2 ), or exposure to air . The problem of embrittlement can be overcome by using FeTi in the form of a powder with grain sizes of a few micrometer diameter. As a storage system, one could think of micron sized FeTi particles partly covered with a protective coating . As was shown by Sanders and Tartarchuk, a Pd coating can prevent ∗
Corresponding author. Tel.: +31 30 2534249; fax: +31 30 2543165. E-mail address: [email protected]
(A.M. Vredenberg). 1 Present address: Ecofys bv, Postbus 8408, 3503 RK Utrecht, The Netherlands. 0925-8388/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.jallcom.2005.03.081
the oxidation of FeTi by impurities in H2 gas . In previous research, we showed that up to 150 ◦ C 4 nm of Pd is sufficient to act as a protective layer against oxidation of the FeTi in air . It is known that the structure of the material plays an important role in both the reversible storage capacity and the need for an activation treatment. In amorphous materials, mechanical integrity upon hydriding is better. However, H absorption is not fully reversible at RT and an activation treatment is still necessary [1,6,7]. Nano-crystalline FeTi exhibits easier activation and a lower plateau pressure in comparison to microcrystalline FeTi . In this article, the hydriding characteristics of FeTi/Pd bilayers of two FeTi compositions are investigated. Such a layered system may serve as a model system to investigate the effect of structure (composition, crystallinity, capping layers, impurities, etc.) on the hydriding characteristics in a controlled way. The use of thin films allows the application of a combination of diagnostic tools to monitor the evolution of structure and hydrogen absorption. In this paper, we will focus on the pressure–composition (p–c) diagram
A.M. Vredenberg et al. / Journal of Alloys and Compounds 400 (2005) 188–193
in relation to the alloy composition and the disordered structure of the film. The kinetics of the H charging and discharging process is the subject of a forthcoming paper.
valve Kx is opened, the uptake starts. The H uptake was examined in 0.1–5 bar of H2 at RT. From the measured pressure difference, the dimensions of the sample and the volumes Vl , Vx and Vr , the amount of H taken up by the sample can be calculated using the ideal gas law. By combining this with RBS data on the thickness of the layer, we find the hydrogen to metal ratio (H/M). The accuracy in the calculation of H/M is around 4%. After charging, we assume a constant H/Pd = 0.65 to calculate H/M (with M = Fe + Ti). This is based on published p–c diagrams of thin Pd films at RT and the observation that the kinetics of Pd is faster than for FeTi/Pd layers [10,11]. After discharging in vacuum we assume that practically all H has desorbed from the Pd layer [11,12]. X-ray photoelectron spectroscopy (XPS) was used to analyze the chemical state of the elements present in the first few nm of the surface. XPS was done using a CLAM-2 hemispherical sector analyzer and a Vacuum Generators XR2E2 Twin Anode X-ray Source employing an Al K␣ source operating at 120 W. X-ray diffraction (XRD) was used to investigate the crystal structure of the layers. XRD has been performed with a Nonius PDS 120 powder diffractometer equipped with a position sensitive detector of 120◦ 2θ, using Co K␣ radiation.
Thin films of 100 nm FeTi coated with 20 nm Pd were deposited on Si wafers by e-beam evaporation (at p ∼ 5–10−7 mbar). Two different FeTi compositions were deposited, Fe43 Ti57 and Fe56 Ti44 (as determined with Rutherford backscattering spectrometry (RBS)). Depth profiles of C, N and O were measured with elastic recoil detection (ERD) using a 66 or 72 MeV Ag beam and a E–E telescope . Contamination levels of C, N and O in the as-deposited FeTi/Pd layer were below 0.5 at.%. The H uptake of the layers was measured in two ways. With ERD (using a 2.8 MeV He beam and a 13.8 m Mylar foil in front of the detector) the concentration profile was measured in situ after H uptake in 1 bar H2 at room temperature (RT). Transport from the gas-loading chamber to the ERD chamber was done within minutes. The goniometer of the ERD measurement chamber is cooled with l-N2 to prevent loss of hydrogen during the measurement. The hydrogen uptake was also measured by using a volumetric device especially designed for gas uptake in thin films. In our high pressure hydrogen reactor (HPHR), the change in pressure difference between a reference volume (Vr ) and a small volume containing the sample (Vl ) is monitored using a differential pressure gauge (see Fig. 1). The HPHR is connected to a gas manifold via the valve Kout . Attached to the manifold is a cylinder with 99.9999% pure H2 . To measure the H uptake of a sample, H2 is introduced in the volume Vx . A comparable pressure of H2 is introduced in the volumes Vr 1 + Vr 2. Valve Kout is closed and when the
XRD spectra of an as-deposited FeTi layer containing 43 at.% Fe and an as-deposited FeTi/Pd layer containing 56 at.% Fe are shown in Fig. 2. Pd peaks are visible, but no clear peaks of FeTi, Fe2 Ti, Fe or Ti are observed. A broad peak is seen around 50–53◦ 2θ, where peaks due to FeTi and Fe2 Ti are located . The broadness of this peak indicates
Fig. 1. The volumes of the HPHR. The sensor (in Vs ) measures the pressure difference that arises when a sample (in Vl ) takes up hydrogen, while the volumes Vx + Vl and the reference volumes Vr 1 + Vr 2 are closed off by the differential valve (Kdiff ). In that situation, the Kx valve is open and the outlet valve Kout is closed.
Fig. 2. XRD spectra of a 110 nm FeTi layer containing 43 at.% Fe (bottom spectrum) and a 90 nm FeTi layer containing 56 at.% Fe coated with 20 nm Pd (top spectrum). The angular position of all the peaks of Pd, FeTi, Fe2 Ti and the high intensity peaks of Fe and Ti are indicated in the bar diagram (top). The intensity scale is logarithmic.
3. Results and discussion 3.1. Structural changes in FeTi/Pd upon hydriding
A.M. Vredenberg et al. / Journal of Alloys and Compounds 400 (2005) 188–193
that the material does not have long-range order and must be amorphous or nano-crystalline in nature. The diffractograms do not allow distinguishing these two, but the hydrogenation p–c diagrams presented below suggest that the films cannot be fully amorphous. Indeed, additional diffraction observations in an electron microscope revealed a heavily disordered structure, but with a ring pattern that was too sharp for fully amorphous material. No change in the spectrum was observed after hydrogen charging. 3.2. H cycling at RT Upon the first H2 exposure the FeTi/Pd layers are charged to H/M values of 0.58 and 0.75 in 56 and 43 at.% Fe (in the HPHR in 800 mbar H2 at RT), respectively. Typical uptake curves are shown in the left panel of Fig. 3. Uptake curves measured on other samples show the same characteristics, but with a small spread in H/M from sample to sample (∼0.03). The time necessary for charging is ∼2 min for 56 at.% Fe and ∼10 min for 43 at.% Fe. In the right panel of Fig. 3, two later cycles of 56 and 43 at.% Fe are shown. The final H uptake is 0.37 H/M for 56 at.% Fe and 0.35 H/M for 43 at.% Fe. In comparison with a first charge it is clear that the uptake rate is faster, but the total uptake is less. As we will show below, this is partly caused by a residue (H/M ∼ 0.12) that remains after discharging at 150 ◦ C. The HPHR only measures the relative uptake (H/M), and thus second and later charges show an uptake that is lowered by at least the residue from previous chargings. In order to investigate the effect of the anneal separately, we annealed under vacuum a fresh 43 at.% Fe sample at 150 ◦ C before charging. As can be seen in the left panel of Fig. 3, annealing before the first H charging does not lead to a lower uptake, only to a higher uptake rate . In Fig. 4, the H concentration profile of a charged and a discharged FeTi/Pd layer is shown. The charged sample shows two levels, one at H/M = 0.15 in the region of the Pd layer and one at 0.25 in the FeTi layer. The plateau level in
Fig. 3. HPHR data of H charging at RT. Left: first charge for 56 and 43 at.% Fe, the latter also after an anneal at 150 ◦ C (in 800 mbar H2 ). Right: later charge for 43 at.% Fe (in 850 mbar H2 ) and 56 at.% Fe (in 900 mbar H2 ).
Fig. 4. ERD H depth profiles in FeTi/Pd with 43 at.% Fe, as deposited and after H charging and after 150 ◦ C discharging. The boundaries between the layers are indicated by vertical lines on the x-axis.
the FeTi part implies that the equilibrium distribution in the FeTi is homogeneous in depth. Comparison with the H depth profile of the as-deposited layer and the indicated depth of the layers suggests that the level in the Pd region is most likely due to H present in the surface region and is not caused by a plateau in the Pd layer. This is in agreement with thermal desorption measurements, which indicate that H is not stable in Pd in UHV . After in situ discharging in vacuum at 150 ◦ C, ERD measurements reveal a residue of H/M = 0.12, as shown in Fig. 4, indicating the presence of deep trapping sites in the material. Repeated in situ charge/discharge cycles consistently show a residue of 0.12 ± 0.01 after each cycle. The same residue was found for the other composition (56 at.% Fe), for samples that were prepared in the HPHR and measured with ERD immediately after discharging. During long air exposures after discharging the residue decreases for Fe-rich material (H/M = 0.04 was measured after two weeks), while that of the Ti-rich material remains the same in the same period. The presence of deep trapping sites, responsible for the residue, will also become clear below, where we investigate the RT p–c isotherms for both compositions. These sites, which are most probably located in amorphous zones in the material, are filled at low pressures, and are not completely emptied at 150 ◦ C. To get a picture of the hydrogen cycle capacity of the FeTi/Pd layers, we must look at both charging and discharging. The complete hydrogen charging and discharging cycles measured in the HPHR are shown in Fig. 5 for the first two cycles. Upon the first exposure, the layers are charged to a total H/M of ∼0.61 (56 at.% Fe) and 0.78 (43 at.% Fe). During subsequent pumping (for 2 min) an amount H/M of ∼0.21 (56 at.% Fe) and 0.03 (43 at.% Fe) is discharged. The amount of discharging cannot be measured directly (since the system is pumped during this time), thus the decrease in H/M was determined in a separate experiment by recharging the layer after various pumping times and measuring the amount of uptake. Annealing at 150 ◦ C leads
A.M. Vredenberg et al. / Journal of Alloys and Compounds 400 (2005) 188–193
Fig. 5. First two cycles of hydrogen charging in the HPHR in 800–900 mbar H2 at RT. From top to bottom: total H/M vs. time for 56 at.% Fe and for 43 at.% Fe; schematic representation of the pressure (mbar) and temperature excursions (◦ C). For both cycles, the H residue (dotted line), as measured with ERD, is added to the value measured with the HPHR. The vertical lines indicate the amount of H desorbed during pumping at RT.
to an extra discharging H/M of ∼0.26 (56 at.% Fe) and 0.58 (43 at.% Fe). Again, after this charging and discharging a residue of H/M of 0.12 is left in the layer as observed with ERD. In the second exposure to H2 an amount H/M of ∼0.37 (56 at.% Fe) and 0.35 (43 at.% Fe) is taken up (see also Fig. 3). In the following discharging, the layer is discharged to the level of the residue. After the first charging and discharging, both the H uptake and the residue of H were found to be stable in subsequent cycles. Thus, the hydrogen cycle capacity (at RT charging in 800 mbar H2 and discharging at 150 ◦ C) is H/M ∼ 0.36 for both FeTi compositions. The cycle capacity can be increased either by using a pressure just above the plateau pressure pp , or by choosing a higher temperature for discharging. Indeed, at 2700 mbar we found a cycle capacity of H/M ∼ 0.9, assuming the same residue at 150 ◦ C. In addition, we found that at 200 ◦ C an extra amount of H/M = 0.06 could be discharged (after charging at 800 mbar). We conclude that during the first charge more H is taken up than during later charges. After exposure to air for a longer time (days), again the same amount of H is taken up as in the first charge. This indicates that the process causing the reduction in uptake is reversible. Apparently, the reduction in H uptake after the first cycle is due to structural changes that occur during the first cycle. These changes cannot be induced by a 150 ◦ C anneal only, as is demonstrated in Fig. 3. This seems to suggest that the combination of annealing and the presence of H in the sample must be the cause of the lower H uptake in the second charge. During the long airexposure after cycling, the structure relaxes from the quasiequilibrium state induced during discharging, resulting in a higher uptake. Hydrogen-assisted relaxation mechanisms in elemental and alloyed metals have been observed before. Many of
Fig. 6. p–c diagrams for different FeTi compositions measured with the HPHR at RT. The solid lines are guides to the eye.
them can be explained by a hydrogen-enhanced diffusion effect, first described by Berry and Pritchet in a NiZr metallic glass . The enhanced diffusion and relaxation is a result of a dramatic increase in the equilibrium vacancy concentration at high hydrogen concentration due to a strong hydrogen–interstitial/vacancy interaction . The enhancement is reversible, but with slow kinetics at room temperature. Many relaxation phenomena in metallic systems can be traced back to this mechanism , and although we do not provide a mechanism by which the hydrogen uptake is reduced, it seems likely that reversible relaxation involving a hydrogen/vacancy interaction is operative in our FeTi alloys as well. 3.3. p–c dependence of H in FeTi/Pd The pressure–composition RT isotherm of the two different Fe/Ti compositions was measured in the HPHR for samples that had been charged several times. Each point is measured by charging the sample at a given pressure and temperature. After discharging at 150 ◦ C and cooling down again, the H uptake at a different pressure is measured. After completion of the experiments in the HPHR, the amount of residual H was measured with ERD (H/M ∼ 0.12, as above) and added to the H/M as measured with the HPHR. The p–c diagrams shown in Fig. 6 show several general features. First, at low H2 pressures (<800 mbar) already a considerable amount of H is taken up in the material. For both compositions this amounts to H/M of 0.3–0.4. Secondly, above H/M ∼ 0.3–0.4 a sloping plateau is observed. The pressure pp in the middle of the plateau lies at 1700 mbar for samples with 56 at.% Fe, and at 1300 mbar for samples with 43 at.% Fe. The compositional range of this plateau is H/M ∼ 0.4–0.8 for 56 at.% Fe and ∼0.4–1.0 in the case of 43 at.% Fe. Characteristically, amorphous materials do not show a plateau in the p–c isotherm. This is due to a wide distribution of available sites for hydrogen [6,7], in contrast to crystalline
A.M. Vredenberg et al. / Journal of Alloys and Compounds 400 (2005) 188–193
material, where the filling of a particular site gives rise to a pressure plateau. Therefore, the p–c diagrams in Fig. 6 indicate that our deposited FeTi films are not fully amorphous, but are presumably a mixture of nano-crystalline and amorphous phases, in accordance with XRD and TEM observations. Both pressure plateaus in Fig. 6 are inclined and lie below that of coarse-grained, poly-crystalline material (>10 bar). A variety of factors modify the characteristics of the plateau (i.e. location and inclination), and we will discuss a number of them below. The chemical composition is one of the most important factors affecting the metal–hydrogen interaction. In fact, it may seem surprising that the two isotherms in Fig. 6 are so similar, whereas the alloy compositions are different. Yet, both compositions are away from the stoichiometric compound, and away from ordered crystallinity. It has been observed  and calculated  that excess Ti gives rise to a lowering of the plateau pressure. Two effects may be operative in the plateau reduction: first, the excess Ti results in lattice expansion, and second, excess Ti modifies the electronic structure of the alloy. Both will affect the energy of the Ti4 Fe2 octahedral binding site for hydrogen. The Ti4 Fe2 octahedron is believed to be the binding site in crystalline FeTi, with an associated plateau that lies above 10 bar. It is imaginable that excess Fe has similar effects, i.e. lattice expansion, and a lowering of the binding energy of the Ti4 Fe2 site. The presence of amorphous material, reflected by the high absorption at low pressure, will also affect the pressure plateau. This has been observed before in partially amorphous FeTi, obtained by ball-milling [8,17,19]. Here, it was argued that the filling of the amorphous regions exerts a pressure on the nano-grains, thereby changing the chemical potential. Further, it could be shown that this leads to lowering and inclination of the plateau. A similar mechanism may be operative in deposited films. Chemical disorder is expected to modify the H binding characteristics as well. Disorder results in the appearance of Ti5 Fe1 and Ti6 Fe0 octahedrons with supposedly increased binding energies. Such deep trapping sites will be filled at low pressures and expand the crystal, which will tend to shift the plateau to lower pressures .
4. Conclusion It is shown that a thin film FeTi/Pd structure is a proper model system to study the hydriding characteristics of storage systems with a large surface to volume ratio. By using beam techniques and a sensitive, but simple, gas cycling device we were able to measure hydrogen concentration measurements as a function of pressure and temperature, as well as characterize the layered structure itself. It is shown that thin films of FeTi can be charged with H without activation by applying a Pd coating. At the first hydrogen exposure, these layers exhibit a high H uptake of H/M = 0.75 for layers containing
43 at.% Fe and 0.58 for 56 at.% Fe, respectively (in 800 mbar H2 at RT). Later charges show a lower H uptake. It was shown that this decrease of H uptake in the second charge is only partly due to a residue of H remaining in the layer after 150 ◦ C annealing. Even when taking the residue into account, the uptake of the second charge H/M is ∼0.2 lower than that of the first charge. We think the reason for this is structural relaxation due to annealing in the presence of H in the material, probably mediated by hydrogen-enhanced host diffusion. After the first charging both the H uptake and the residue of H after 150 ◦ C annealing were found to be stable. The Pd remains intact even after repeated charging and discharging. For both compositions, the H cycle capacity is H/M ∼ 0.36 for RT charging in 800 mbar H2 and discharging at 150 ◦ C. By annealing at 200 ◦ C, the capacity can be enlarged to H/M ∼ 0.43 for the same charging conditions. We also conclude that the cycle capacity just above the plateau pressure is ∼0.9. H charging proved to be possible at higher temperatures and lower pressures than with micro-crystalline FeTi, comparable to what is observed by others in nano-crystalline FeTi. We think the enhanced absorption at low hydrogen pressure is due to highly disordered or amorphous regions. The lower plateau pressure can be ascribed to the stress induced by the hydrogen-filled amorphous regions, enhanced by additional strain due to compositional mismatch.
Acknowledgement This research was performed with financial support from the Stichting voor Fundamenteel Onderzoek der Materie (FOM).
References  K. Nakamura, Ber. Bunsenges. Phys. Chem. 89 (1985) 191.  J.H. Sanders, B. Tatarchuk, J. Phys. Met. Phys. 18 (1988) L67.  L. Schlapbach, T. Riesterer, Appl. Phys. A 32 (1983) 169.  A. Zaluska, L. Zaluski, J.O. Str¨om-Olsen, Appl. Phys. A 72 (2001) 157.  E.M.B. Heller, J.F. Suyver, A.M. Vredenberg, D.O. Boerma, Appl. Surf. Sci. 150 (1999) 227.  R.C.J. Bowman, Mater. Sci. Forum 31 (1988) 197.  E. Batalla, J.O. Str¨om-Olsen, Z. Altounian, D. Boothroyd, R. Harris, J. Mater. Res. 1 (1986) 765.  L. Zaluski, A. Zaluska, J.O. Str¨om-Olsen, J. Alloy. Compd. 253–254 (1997) 70.  W.M. Arnold Bik, F.H.P.M. Habraken, Rep. Prog. Phys. 56 (1993) 859.  M.-W. Lee, R. Glosser, Hydrogen in disordered and amorphous solids, in: G. Bambakidis, R.C.J. Bowman (Eds.), NATO, vol. 136, 1986, pp. 351–358.  E.M.B. Heller, A.M. Vredenberg, D.O. Boerma, in press.  F. Cuevas, M. Hirscher, Kronmiller, Presented at JA98 in Paris, 1998.
A.M. Vredenberg et al. / Journal of Alloys and Compounds 400 (2005) 188–193  R. Fischer, W. Haelg, L. Schlapbach, F. Stucki, A. Andresen, Mater. Res. Bull. 13 (1978) 931.  B.S. Berry, W.C. Pritchet, Mater. Sci. Eng. 97 (1988) 419.  E. Hayashi, Y. Kurokawa, Y. Fukai, Phys. Rev. Lett. 80 (1998) 5588.  H.-R. Sinning, G. Steckler, R. Scarfone, Def. Diff. For. 167 (1999) 1.
 P. Tessier, L. Zaluski, A. Zaluska, J.O. Str¨om-Olsen, R. Schulz, Mater. Sci. For. 225–227 (1996) 869.  H. Yukawa, Y. Takahasi, M. Morinaga, Comput. Mater. Sci. 14 (1999) 291.  L. Zaluski, A. Zaluska, P. Tessier, J.O. Str¨om-Olson, R. Schulz, J. Alloy Compd. 227 (1995) 53.