Observation of Rapid Change of Crystalline Structure during the Phase Transition of the Palladium-Hydrogen System
Akio Kawasaki, Satoshi Itoh, Kunihiro Shima, Kenichi Kato, Haruhiko Ohashi, Tetsuya Ishikawa, Toshimitsu Yamazaki
aa r X i v : . [ c ond - m a t . m t r l - s c i ] M a y Observation of Rapid Change of Crystalline Structure during the Phase Transition ofthe Palladium-Hydrogen System
Akio Kawasaki, ∗ Satoshi Itoh,
2, 3
Kunihiro Shima, Kenichi Kato, Haruhiko Ohashi, Tetsuya Ishikawa, and Toshimitsu Yamazaki
2, 3 Department of Physics, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA Department of Physics, University of Tokyo, Bunkyo, Tokyo 113-0033, Japan RIKEN, Nishina Center, Wako, Saitama-ken 351-0198, Japan Tanaka Kikinzoku Kogyo K.K., Tomioka, Gunma-ken 370-2452, Japan RIKEN SPring-8 Center, Sayo, Hyogo-ken, 679-5148, Japan (Dated: August 30, 2018)We performed an X-ray diffraction experiment while palladium bulk absorbed and desorbed hy-drogen to investigate the behavior of the crystalline lattice during the phase transition between the α phase and β phase. Fast growth of β phase was observed around x = 0 . x = 0 .
45 of PdH x . Inaddition, slight compression of the lattice at high hydrogen concentration and increase in the latticeconstant and the line width of the α phase after a cycle of absorption and desorption of hydrogenwas observed. These behavior correlated with the change in the sample length, which may inferthat the change in shape was related to the phase transition. PACS numbers: 61.50.Ks,64.70.K-,64.70.kd, 81.05.Bx
I. INTRODUCTION
Palladium is known as a metal that absorbs largeamount of hydrogen. This property opened its appli-cations to storage and filter of hydrogen, and intense re-search on this property has been performed [1, 2]. Thelarge amount of hydrogen absorption is related to twophases of the palladium-hydrogen (Pd-H) system, onecalled α phase and the other called β phase. The α phase has smaller hydrogen fraction x = H / Pd, and itincludes pure palladium. The β phase contains morehydrogen atoms. When palladium metal is exposed tohydrogen gas, typically at high temperature such as 100 ◦ C or more, palladium absorbs hydrogen. The absorp-tion induces phase transition from α phase to β phasethrough α + β phase, where the α phase and the β phasecoexist, as the phase diagram [3–5] shows.This phase transition from the α phase to the β phaseis believed to be the cause of deformation of a palla-dium bulk reported in Refs. [5–13]. These changes inthe shape significantly larger than the ordinary plasticdeformation are explained as a phenomenon connectedto a phase transition, sometimes with a relation to super-plasticity. We previously found a change in the shape ofpalladium metal in the direction of minimizing its surfacearea [13], and subsequently we observed large bending ofa horizontal palladium plate with only a small externalforce, and warping back and forth of vertical palladiumplate [5]. However, it is not clear why the change in theshape of the bulk happened so as to minimize its surfacearea and why the palladium plate warped back and forth.In this paper, we report an X-ray diffraction experi- ∗ [email protected] Side ViewTop View Imaging PlateChamber wall(Aluminum)Palladium SampleViewport Laser displacement sensorInjected X ray X ray End capScattered X ray 10 mm1 mm 1 mm70 mm ΔLJig
FIG. 1. (Color online) Experiment Setup ment at a synchrotron radiation facility SPring-8 to ob-tain microscopic information of the Pd-H system duringthe phase transition. We took X-ray diffraction spectrawhile a palladium bulk absorbed and desorbed hydrogengas with the information on the length of the sample.The analysis was performed to get microscopic informa-tion, such as lattice constant, crystal grain size, and theintensity of the diffraction from the two phases, againstthe time and the hydrogen fraction x , in order for us toobtain microscopic understanding of the change in theshape of the palladium bulk during the treatment withhydrogen.Microscopic study of the Pd-H system has long beenperformed [2, 14–16]. Initially, X-ray diffraction studywas performed in the static states. The lattice con-stants of the α phase and the β phase are measured as a α = 3 .
894 ˚A and a β = 4 .
040 ˚A [2] by the standard X-ray diffraction technique after putting the sample backto the room temperature. The figures in Ref. [15] wasobtained by placing X-ray diffraction pictures of differentsamples with different hydrogen fraction together. Thedynamic observation of the phase transition has also beenperformed [17–20]. Most of them use the information ofa single peak to estimate the amount of one phase inthe bulk. Lattice constant information on these reportsis simply plotted against temperature or pressure of hy-drogen, and time dependent analysis was not performed.In addition, most of the reports are for nano particle ofpalladium [18–20], and the time dependent microscopicstudy of the palladium bulk is lacking. Our report fillsthis blank and connects the relation between the macro-scopic behavior and the microscopic parameter.
II. EXPERIMENTAL METHODA. Experiment setup
The X-ray diffraction experiment was performed atRIKEN Materials Science beamline BL44B2 of SPring-8.The X-ray energy was 20 keV with the energy resolutionof 10 − . The photon flux was 10 s − and the beamsize was 0.5 mm (vertical) × L was measuredwith a laser displacement sensor through a viewport. Thechamber was made of stainless steel, except for the pathfor the X-ray made of 1 mm thick aluminum. The cham-ber was connected to a pumping system consisting of aturbo molecular pump and a rotary pump. The wholechamber was wrapped with heating tapes to heat thesample up to 120 ◦ C.Prior to the start of the experiment, the chamber wasevacuated out to below 10 − Pa. Hydrogen gas was in-troduced through a flow meter, the start of which is de-fined as t = 0. At first, the introduction was at a max-imum rate of 40 ml/min, and once the pressure reached0.2 MPa, the flow rate was reduced in order to keep thepressure around 0.2 MPa, as shown in Fig. 5 (e). Afterthe saturation of the pressure and ∆ L , by which we re-garded the phase transition as finished, we pumped the hydrogen out through the pumping system. We finishedthe evacuation when the decrease in pressure accelerated,which is the sign of the complete outgassing of hydrogen.We introduced hydrogen again to see the behavior at thebeginning of the second cycle of absorption/desorption ofhydrogen. Throughout the experiment, we recorded thepressure and the temperature of the chamber and ∆ L .The scattered X-ray was recorded with an imagingplate covering the diffraction angle 2 θ of 0 ≤ θ ≤ ◦ .One imaging plate recorded data of 18 different expo-sures and therefore one set of measurements consisted of18 data. Each exposure was for 30 seconds. We startedthe first exposure of the first set right after we started tointroduce hydrogen. B. Characteristics of the setup
Although the result and its interpretation are de-scribed assuming the sample was in the ideal conditionfor the powder X-ray diffraction, we have to consider thedifference between the ideal condition and our system.First, palladium sample was large and certain amountof the X-ray was absorbed by the sample. Second, thenumber of crystal grains might be small and their ori-entation might not be random. Third, the sample mightnot have been uniform, as it takes certain amount of timefor hydrogen to diffuse into the depth of the palladium.The Attenuation coefficient of the palladium is 207.9cm − for 20 keV X-ray. This means that the X-ray pass-ing through 0.1 mm bulk is attenuated to 13% of theincoming flux. Thus, we basically looked at the phasetransition of the thin surface layer whose thickness is atmost 0.1 mm. As the X-ray hit the top point of the trian-gular cross section of the sample, we can assume that thereasonable amount of the diffracted light is transmittedfor all the diffraction angles.The volume we observed is roughly 0.5 mm × × µ m in Fig. 7. It is therefore possible that the numberof grains is too small for the diffraction pattern to havecircular symmetry. This can cause the suppression of thesignal from certain crystal planes.Given that we looked at the thin surface layer, thediffusion would not take that long time. Fig. 2 showsthe numerical calculation of the absorption of hydrogenby 1 mm thick palladium plate when the density of thehydrogen gas linearly increased, with the diffusion coef-ficient of 3 . × − for hydrogen in palladium at 120 ◦ C. Although it took certain amount of time for hydro-gen to reach the center of the sample, the 0.1 mm thicksurface layer got the amount of hydrogen comparable tothe very surface in short time; in 1 minutes, the densitybecomes 40% of the outermost area and after 20 minutes,the density is 90% of the surface one. Thus, except forthe first few minutes, hydrogen density can be assumedto be roughly uniform.
Position [mm]0 0.2 0.4 H ga s / n H bu l k R e l a t i v e D en s i t y n FIG. 2. (Color online) Density of hydrogen in 1mm thick pal-ladium bulk relative to the gas phase 1, 5, 10 and 20 minutesafter the start of the introduction of hydrogen: initial condi-tion is all point was zero at t = 0 and the number at the twoends increased at a constant rate to simulate the increasinghydrogen pressure. III. RESULTS
Figure 3 shows diffraction spectra at five representa-tive moments during the experiment. Some conspicu-ous diffraction lines before the absorption of hydrogengot weaker during the absorption, and disappeared af-ter the absorption. When palladium desorbed hydrogen,these peaks reappeared roughly at the same 2 θ . Duringthe absorption, new diffraction lines appeared at slightlysmaller 2 θ , which corresponds to the lattice constant a larger than the vanishing diffraction lines. These newlines disappeared during the desorption. Based on theobservation, we identified the phase and the crystal planefor the diffraction lines. The intense diffraction lines atthe beginning were identified as α phase lines, as thesample was annealed before the experiment to outgas hy-drogen. The new lines appearing during the absorptionwere identified as β phase lines because these had slightlylarger lattice constant a β than a α , which matched withthe previously reported ratio of a α and a β in Ref. [2](subscription α and β show parameters for the α andthe β phase, respectively). In addition, the differencebetween a α and a β was significantly larger than the fluc-tuation of a α and a β during the absorption and the des-orption of hydrogen. Diffraction lines without any signif-icant change over the experiment were regarded to be thebackground originated from other materials on the pathof X-ray, such as the aluminum chamber wall and resis-tive material in the heating tape. These lines were usedto remove systematic errors of the diffraction angles.Figure 4 is magnified plots of the time dependent be-havior of the diffraction lines from the (111) and (200)planes. The top half shows the first 60 minutes of the ab-sorption stage. The initially intense α phase diffractionlines gradually disappeared over an hour and β phasediffraction lines grew up rapidly at significantly smaller2 θ after several minutes. It is notable that most of the growth of the β phase diffraction lines finished in severalminutes. The bottom half of Fig. 4 shows the behav-ior during the desorption stage. The β phase diffrac-tion lines gradually disappeared and the α phase diffrac-tion lines grew up. The diffraction line angle changedas time evolved. Changes were slower during the des-orption stage than that during the absorption stage, butthey had a common behavior that the growing peaks grewup smoothly, whereas the disappearing peaks had fluctu-ation in their position and the intensity during its fastvanish.For a close look at the time dependent behavior of thediffraction line parameters, we fitted each diffraction linewith a Gaussian plus a linear background function, ob-taining the center of the diffraction line 2 θ α and 2 θ β ,the diffraction line width Γ α and Γ β as the full width athalf maximum (FWHM) and the diffraction line inten-sity I α and I β as the area of the Gaussian. If two ormore diffraction lines were close to each other, the fittingfunction contained multiple Gaussian functions. Amongthese data, the (111) diffraction line is plotted in Fig. 5as the representative. Fig. 5 (a) shows a α and a β calcu-lated from 2 θ α and 2 θ β . A systematic error of 2 θ α and2 θ β presumably due to the slight fluctuation of the po-sition of the whole chamber against the beam and theimaging plate was removed by adding a correction so asto keep the diffraction angle of BG1 line in Fig. 3 con-stant at the average value over all the data points. Fig.5 (b) shows Γ α and Γ β in the unit of angle. Fig. 5 (c)displays I α and I β . A systematic fluctuation of I α and I β was removed by a compensation factor that made anaverage of I β over the five last sets of absorption stageconstant.∆ L is plotted in Fig. 5 (d). The temperature of thesample, pressure inside the chamber and the hydrogenflow rate are plotted in Fig. 5 (d) and (e).Figure 6 shows I α and I β of (111) line against the hy-drogen fraction x , together with ∆ L . The x was calcu-lated from the chamber volume, the hydrogen flow rateand the pressure. IV. DISCUSSIONA. Lattice constant a α and a β a α and a β based on the spectrum before and after theabsorption in Fig. 3 are summarized in Table I. Ourresult gave larger number than Ref. [2]. Part of thereason of the difference is the thermal expansion of thelattice. The thermal expansion coefficient for pure palla-dium, 1 . × − K − at 293 K, and 95 K temperaturedifference (Ref. [2]: 25 ◦ C, us: 120 ◦ C) gives 0.004 ˚A ex-pansion of the lattice. However, this still leaves discrep-ancy significantly larger than the statistical fluctuation,particularly for a α . The remaining difference should beexplained by the different condition to prepare the sam-ple, as the difference is in the order of magnitude same [deg] θ
215 20 25 30 I n t en s i t y [ a . u .]
111 200 B G B G Before Absorption During Absorption After Absorption During Desorption After Desorption
FIG. 3. (Color online) X-ray diffraction spectra over experiment run: the middle of the absorption and the desorption spectrumwas 12 minutes and 90 minutes after the start of the absorption and the desorption, respectively. Three-digit numbers on thetop represent the crystal plane identification for the α phase (black) and the β phase (red). BG1 and BG2 are backgroundlines. Blue horizontal axes show the lattice constant for different crystal plane. The gray lines superposed onto the four spectraare the spectrum before absorption. [deg] θ
215 15.5 16 I n t en s i t y [ a . u .] T i m e [ m i n ] (111) ]Åa [ [deg] θ I n t en s i t y [ a . u .] T i m e [ m i n ] (200) ]Åa [ [deg] θ
215 15.5 I n t en s i t y [ a . u .] T i m e [ m i n ] (111) [deg] θ I n t en s i t y [ a . u .] T i m e [ m i n ] (200) FIG. 4. (Color online) Magnified X-ray diffraction spectrumof (111) line (top left: absorption stage, bottom left: des-orption stage) and (200) line (top right: absorption stage,bottom right: desorption stage). The blue axes shows thelattice constant. as the fluctuation of a α and a β shown in Fig. 5.The time dependent behavior of a α and a β of the (111)diffraction line is shown in Fig. 5 (a). Notable features TABLE I. Lattice constant [˚A]: TE column shows the amountof the thermal expansion of the lattice.Phase This result Ref. [2] Difference TE α ± .
005 3.894 0.034 0.004 β ± .
006 4.040 0.014 0.004 for a α are (i) its initial increase, (ii) smaller a α than theinitial value when the α phase appeared again at thedesorption stage, and (iii) the increase of a α at the endof the desorption. The first feature should be simplydue to the hydrogen’s occupying interstitial space amongpalladium atoms to expand the palladium lattice. a α was 0.6% smaller than the initial valuewhen it ap-peared again during the desorption. This is significantlylarger than the uncertaintiy of the center position, and ispossibly due to the compression by the β phase. Becausethe sample length L increased by only 0.6% even whenthe sample completely turned into β phase that has 3%larger lattice constant than α phase, the crystal grainexperienced the huge internal stress[21], and it is possi-ble that the pressure compressed α phase lattice. Thefact that the rapid growth of the I β started at the sametime as the start of the decrease in a α during absorption-started also supports the idea of the compression by the β phase. a α increased by 0.3% after a cycle of the absorptionand the desorption of the hydrogen, which is signifi-cantly larger than the uncertainty of the center position.This means that there is an irreversible effect of hydro- La tt i c e C o Initial latticeconstant ] Å n s t an t [ Initial lattice [ deg ] Γ L i ne W i d t h L i ne I n t en s i t y I[ a . u .] C ] ° T e m pe r a t u r e [ Absorption P r e ss u r e [ k P a ] Introduction H 5 [ml/min] × Rate
350 400 450 500 550 600 650 α
350 400 450 500 550 600 650 constant β (a)
350 400 450 500 550 600 650 ] Å G r a i n S i z e [ (b) αβ
350 400 450 500 550 600 650 L [ mm ] ∆ C hange (c) αβ
350 400 450 500 550 600 650 .0.5.0 S a m p l e Leng t h -1.5-1.0-0.50.00.5 Desorption (d)
Time [min]
350 400 450 500 550 600 650050100150200 P r e ss u r e [ P a ] -3 -2 -1 (e) FIG. 5. (Color online) Time dependence of (a) the lattice constant a α and a β , (b) line width Γ α and Γ β and correspondinggrain size (right axis), and (c) line intensity I α and I β of the diffraction from (111) plane. Black points are for the α phase andred ones are for the β phase. The sample length change ∆ L is shown in (d) (brown line, positive ∆ L corresponds to largersample), and experiments condition are shown in (d) and (e). The H introduction rate is magnified by 5 times, and should beread with the left axis. Blue and green dotted lines show the start of absorption and desorption stage of the hydrogen. Datapoints are not shown if the diffraction line was too weak to perform the fit. Only initial 120 minutes is shown for the firstabsorption stage, as the last 200 minutes did not have much change. gen treatment on palladium bulk. The behavior of a α in the second cycle quite similar to the first absorptionstage suggests that the value at the end of the desorp-tion stage should be regarded as the value for the purepalladium, not the value with slight amount of hydrogenremaining in the bulk. Irreversible changes were also re-ported previously, such as the change in the shape andthe degradation of metallic luster [13]. The observationin this experiment revealed that such a change also hap-pened in microscopic scale. Note that the behavior isquite similar for a β in the desorption stage, but a β wentback to its initial value.The change of 2 θ β is large enough to be observed in Fig. 4, and detailed time dependent behavior of a β is shownin Fig. 5 (a). It had a local minimum during the phasetransition, when t = 60 min, which corresponds to thehydrogen fraction x ≃ .
45 in Fig. 6. These were 0.2%jump, and possible explanation is the growth of β phasefirst compressed its lattice and then the lattice graduallyexpanded. The expansion was not all the way back tothe initial number. This observation is quite differentfrom the report in Ref. [15], where a β increased as thehydrogen fraction x increased. The difference should bedue to the experimental condition, particularly if the X-ray diffraction is done during the phase transition or not. Hydrogen Fraction x=H/Pd0 0.2 0.4 0.6 D i ff r a c t i on L i ne I n t en s i t y I [ a . u .] L [ mm ] ∆ S a m p l e Leng t h C hange αβ FIG. 6. (Color online) Intensity of the (111) diffraction lineagainst the hydrogen fraction x .TABLE II. Line width and crystal size: the α phase has thevalue before and after a cycle of absorption and desorption ofhydrogen. The β phase value is for after the absorption.Phase Line width [deg] Crystal Size [˚A] α , before 0.062 ± ± β ± ± α , after 0.217 ± ± B. Line Width Γ α and Γ β Γ α , Γ β and the crystal grain size derived from Scher-rer formula are summarized in Table II. The trend wasthat the grain size got smaller as the absorption and thedesorption proceeded. The actual grain size in Fig. 7 is10-100 µ m, which is larger than the coherence length ofthe X-ray, 1000˚A. It is likely that the defect or the dis-order in the single crystal grain decreased the effectivegrain size.The change of Γ α and Γ β over time shown in Fig. 5 (b)tells how the reduction of the grain size happened. TheΓ α increased by 27% from the beginning after a cycle ofabsorption and desorption, which was the same trend asthe number in Table II. The increase started before the α phase line disappeared. The Γ β was approximatelythe same as Γ α both when the α phase disappeared, andwhen the α phase reappeared. Once α phase reappeared,Γ α was more or less constant. This means that mostincrease in Γ occurred during the absorption. Γ β changedboth during the absorption and the desorption, with alocal maximum at t = 60 min for absorption stage and t = 450 min for desorption stage, though they are notvery significant.This increase in Γ α is consistent with the observationof crystal orientation with the scanning electron micro-scope electron backscatter patterns (SEM-EBSP) in Fig.7. The palladium metal that went through a cycle of ab-sorption and desorption of hydrogen gas had much moresmall angle tilt grain boundary, between 1 . ◦ and 3 ◦ ,compared to the annealed palladium sample, whereas the FIG. 7. (Color online) SEM-EBSP image of palladium samplebefore (left) and after (right) hydrogen absorption: top twosection shows the crystal orientation, and each color showsdifferent orientation. Bottom two shows the angle differenceof crystal orientation at grain boundaries. size of the grain was roughly the same if the grain bound-ary is defined with tilt angle larger than 6 ◦ . The effectivedecrease of the grain size observed in the X-ray diffrac-tion was due to the formation of small angle tilt grainboundary. This tilt is likely to be formed when the α phase turned into the β phase during the absorption. C. Diffraction Line Intensity I α and I β Time dependent behavior of I α and I β of the (111)diffraction line is shown in Fig. 5 (c). The β phasediffraction line appeared at 8 minutes after the start ofthe absorption, and two thirds of the change was com-pleted in the first 18 minutes. During this time range,hydrogen fraction x changed from 0.03 to 0.15 as shownin Fig. 6. This x was close to that of when the suddenchange in the palladium shape occurred in Ref. [5]. It isinferred that the sudden change in the shape was inducedby the appearance of β phase. There was another quickincrease of I β around t = 60 min. This corresponds tothe hydrogen fraction between x = 0 .
40 and x = 0 . I β ≃ . I β ≃ . α phase line slowed down astime went by, but the last moment of the disappearancearound t = 60 min was faster than the earlier part. Thisjump to I α = 0 happened at the same time as the secondjump in the I β . Also, the fast change in I α and I β around x = 0 .
45 coincided with the local minimum for a β onboth the absorption and the desorption stage and thelocal maximum of Γ β . This whole behavior correspondsto the point of phase transition from the α + β phase to β phase. Combined with the appearance of β phase line, x for α + β phase should be 0 . ≤ x ≤ . I α + I β , which should hold if the amount of the materialthat scatters the X-ray is constant. In our observation, I α + I β had ±
20% fluctuation. It is possibly due to theimperfect randomness of the crystal orientation that re-sulted in the different amount X-ray scattered by the α phase and the β phase.Most β phase diffraction lines totally disappeared whenthe desorption was completed. This ensures that all thesample went back to α phase and the pumping was longenough to evacuate all hydrogen. D. Sample length change ∆ L ∆ L is plotted in Fig. 5 (d) and Fig. 6. The sam-ple expanded by 0.6 % when it absorbed hydrogen up toPdH . and then shrank more than its original length by-2.5% when hydrogen was released completely. This wasroughly consistent with Ref. [13]. The behavior of ∆ L was quite similar to that of I β , especially in the desorp-tion stage. Also, the slight change a α and a β coincidedwith the behavior of ∆ L , particularly in the desorptionstage. These implies the sample length change was re-lated to the phase transition. V. CONCLUSION AND OUTLOOK
To conclude, we performed the in-situ
X-ray diffractionexperiment during the phase transition in Pd-H system.The fast growth of I β and gradual decrease in I α wereobserved during the absorption of hydrogen. a α and a β got smaller as hydrogen concentration got higher with an initial increase in a α before β phase appeared. Thissuggests the compression of the lattice by the β phase.The small angle tilt inside the cyristal grain grew aftera cycle of hydrogen absorption and desorption, resultingin the increase of Γ α . The coincidence of the behavior inthe angle a α , a β , I α , I β and ∆ L implied that the changein the shape was due to the phase transition.The reproducibility of the behavior when we have mul-tiple cycles of the absorption and desorption of hydrogenwas not clear. The one hour of second cycle suggeststhat the behavior of parameters were basically the sameas the first cycle. However, it is still necessary to directlyshow what happens when we have more than one cycleof absorption and desorption of hydrogen.The behavior of the center of the sample also needsto be investigated. Since X-ray diffraction in principlecannot give the information of the deep inside the bulk,neutron scattering or some other method is expected toreveal how the center of the bulk changes.The discussion is only for our observation in Pd-H sys-tem. Lots of other systems show superplasticity or simi-lar kind of deformation over a phase transition, same asPd-H system, but this fact is not enough to show that themicroscopic behavior is universal in the phase transitionof polycrystalline materials. In order to show the univer-sality, one has to do the same experiment for differentmaterials. ACKNOWLEDGMENTS
This research is supported by the Strategic Programfor R&D of RIKEN. The synchrotron radiation experi-ment were performed at BL44B2 in SPring-8 with theapproval of RIKEN (Proposal No. 20100100), and theuse of RIKEN beamline at SPring-8 was supported byRIKEN SPring-8 Center. We would like to thank Dr.M. Sato for writing data acquisition. We are gratefulto Dr. Nishimura with insightful discussion and Prof.M. Iwasaki and Prof. R. S. Hayano for the stimulatingsupport. [1] F. A. Lewis,
The Palladium Hydrogen System (AcademicPress, New York, 1967).[2] T. B. Flanagan and W. A. Oates,Annu. Rev.Mater. Sci. , 269 (1991).[3] A. G. Knapton, Platin. Metals Rev. , 44 (1977).[4] E. Wicke and H. Brodowsky, in Hydrogen in Metals II,ed. G. Alefeld and J. Volkl (Springer, Berlin, 1978) p. 73.[5] A. Kawasaki, S. Itoh, K. Shima, and T. Yamazaki,Mater. Sci. Eng. A , 231 (2012).[6] W. Krause and L. Kahlenberg, Trans. Electrochem. Soc. , 449 (1935).[7] J. B. Hunter, Platinum Met. Rev. , 130 (1960).[8] A. Kufudakis and J. Cermak, Surf. Technol. , 57 (1982).[9] R. V. Kotelva and J. L. Glukhova, Int. J. Hydrogen En-ergy , 175 (1997).[10] Z. R. Xu and R. B. Mclellan, Acta Mater. , 4543(1998).[11] H. S. Y. Jung and Y. Sakai, J. Soc. Mater. Sci. Japan , 1242 (2000), in Japanese.[12] Y. Jung and Y. Sakai, J. Soc. Mater. Sci. Japan , 999(2001), in Japanese.[13] T. Yamazaki, M. Sato, and S. Itoh, Proc.Jpn Acad. Ser.B , 183 (2009).[14] E. A. Owen and J. I. Jones,Proc. Phys. Soc. , 587 (1937). [15] P. C. Aben and W. G. Burgers,Trans. Faraday Soc. , 1989 (1962).[16] Y. Sakamoto, K. Yuwasa, and K. Hirayama,J. of the Less Common Met. , 115 (1982).[17] G. Li, H. Kobayashi, J. M. Taylor, R. Ikeda, Y. Kub-ota, K. Kato, M. Takata, T. Yamamoto, S. Toh, S. Mat-sumura, et al. , Nature Mater. , 802 (2014).[18] R. Nandi, R. Pitchai, S. Wong, J. Cohen, R. Burwell Jr., and J. Butt, J. of Catalysis , 298 (1981).[19] W. Vogel, W. He, Q.-H. Huang,Z. Zou, X.-G. Zhang, and H. Yang,Int. J. of Hydrogen Energy , 8609 (2010).[20] P. Canton, C. Meneghini, P. Riello, A. Balerna, andA. Benedetti, J. Phys. Chem. B , 8088 (2001).[21] R. Delmelle and J. Proost,Phys. Chem. Chem. Phys.13