Structural Properties of (Ti, Zr)(Mn, Cr)₂M_0.1 (M = None, Fe, Co, Ni, and Cu) Hydrogen Storage Alloys: Composition Distribution and Occupied Site of Doped Element

Abstract

CO₂ tolerance of hydrogen storage alloys of AB₂-type (C14 Laves phase) Ti_0.515Zr_0.485Mn_1.2Cr_0.8M_0.1 (AB₂-M, M = none, Fe, Co, Ni, and Cu) depends on dopant M. Since our goal is to clarify this mechanism, we determined the elemental analysis using X-ray absorption spectroscopy (XAS), scanning electron microscope (SEM) coupled with energy dispersive X-ray spectroscopy (EDX), powder X-ray diffraction (XRD), and neutron powder diffraction (NPD) with Rietveld refinement in this study. As a result of XAS analysis, a strong evidence of all doped elements occupying B site in AB₂ was obtained. SEM-EDX showed inhomogeneous composition with vacancy in B site and linear correlation of Ti/Zr and Mn/Cr ratio. The peak width in XRD patterns of AB₂-M depends on the magnitude of homogeneity, therefore the Rietveld analysis using NPD patterns could not be well refined. Thus, homogeneity is not important but element of B site would be important for CO₂ tolerance as well as AB₅ type alloys.

Additive metal M occupies B site in AB₂-type hydrogen storage alloys. Therefore, CO₂ tolerance depends on B element.

1. Introduction

Hydrogen produced from hydrocarbon using steam reforming contains CO, H₂O, and CO₂.¹⁾ A CO Adsorbent and Metal hydride Intermediate-Buffer (COA-MIB) purifies hydrogen by removing CO and H₂O at the COA unit and then CO₂ using hydrogen storage alloys (MH).^2–4) MH enables flexible hydrogen supply into polymer electrolyte fuel cells (PEFC) in order to follow PEFC demand. First generation of MH for COA-MIB system has used expensive AB₅-type alloys which contains rare earth metal. To minimize the system cost, we employed rare earth free AB₂-type alloys due to generally cheaper and higher capacity than AB₅-type alloys.⁵⁾ Most of hydrogen storage alloys except some AB₂-type alloys have a drawback of easy deactivation by CO₂ (poisoning), which means slow kinetics and small H₂ capacity.⁶⁾ We have recently clarified that small amount of doped element M into AB₂-type alloys (AB₂-M: Ti_0.515Zr_0.485Mn_1.2Cr_0.8-M_0.1, M = Fe, Co, Ni, Cu) effects CO₂ tolerance.⁷⁾ While Fe as M of AB₂-M (AB₂-Fe) improved CO₂ tolerance, AB₂-Ni showed significant poisoning. The tendency of CO₂ tolerance of AB₂-M is AB₂-Ni < AB₂ < AB₂-Co < AB₂-Fe. We also found that B element in AB₅-type alloys is important role of CO₂ tolerance.⁸⁾ Considering this fact and above doped elements are generally easy to occupy B site,⁹⁾ AB₂-M can be applied this analogy. However, there is no evidence of which M occupies B site in AB₂-M and some elements can occupy both A and B sites.^10,11) The purpose in this study is to determine the occupied site of doped metal in the crystal of AB₂-M in order to discuss the mechanism of additive effect on CO₂ tolerance. Although XAS is powerful tool to determine the occupation site of specific element, it is difficult to analyze if the samples are composed in multiple phases. Thus, the distribution of composition of AB₂ and AB₂-M have been investigated and the occupation site of additive element also have been determined in this study.

2. Experimental Procedure

AB₂ (target: Ti_0.515Zr_0.485Mn_1.2Cr_0.8) and AB₂-M (target: Ti_0.515Zr_0.485Mn_1.2Cr_0.8M_0.1; M = Fe, Co, Ni, and Cu), were synthesized by arc melting (NEV-AD 03, Nisshin Giken Co.) under an argon atmosphere. Starting materials of Zr (99.7%), Mn (99.99%), Cr (99.99%), Fe (99.9%), Co (99.9%), Ni (99.9%), and Cu (99.9%) were purchased from Furuuchi Chemical Co. and Ti (99.5%) was purchased from The Nilaco Corporation. After the arc melting, alloys were annealed at 1100 °C for 24 h in the argon atmosphere for homogenization. Composition distribution of alloys were determined by SEM-EDX (SEM: TM3030, Hitachi High-Technologies; EDX: Quantax 70, Bruker). More than 35 spots (∼20 µm) were randomly quantified in each alloy. Powder X-ray diffraction (XRD) patterns of AB₂-M at RT were acquired by Rint-ultima (Rigaku, Cu-Kα). Neutron powder diffraction (NPD) has been performed using NOVA in BL21, the Japan Proton Accelerator Research Complex (J-PARC). Each 2∼3 g of sample was filled in a cylindrical null-scattering V₉₆Ni₄ container (6.00 mm of diameter and 0.10 mm of thickness). The data was collected with 500 kW beam for 6 h at RT up to Q_max = 68.48 Å⁻¹. Then, we tried to determine the detail of occupation site of elements using Rietveld refinement (Z-Rietveld).^12,13) The compositions for Rietveld refinement were fixed to the SEM-EDX results. X-ray absorption spectroscopy (XAS) has been performed at BL14B2 in SPring-8 in order to determine the distance from target element to the nearest atom. XAS spectra of Ti, Zr, Cr, Mn, Fe, Co, Ni, and Cu K-edge in each AB₂-M were acquired using transmission method.

3. Results and Discussions

3.1 XAS analysis

X-ray Absorption Near Edge Structure (XANES) spectra of AB₂ and AB₂-M for K-edge of each element are shown in Fig. 1.¹⁴⁾ Absorption starting energies of all elements (even doped metals) for all samples are quite close to those for the elemental metal foils. All profiles have pre-edge as well as metal foils. These facts indicate that all samples are metallic phase and no or quite small oxidation proceeds. Since Mn is easy to be oxidized, absorption edge in Mn foil slightly shifts higher energy than AB₂-Mn probably due to oxidation of Mn foil. These detailed profiles including doped metals are different from elemental metal foil, indicating that the major phases of all samples are alloy without unreacted starting material.

Fig. 1

XANES profiles of (a) Ti-K (b) Zr-K, (c) Cr-K, and (d) Mn-K for AB₂ and AB₂-M and (e) Fe-K, (f) Co-K, (g) Ni-K, and (h) Cu-K for AB₂-M.¹⁴⁾

Structural information can be obtained from EXAFS spectrum. Radial distribution function (RDF) obtained by EXAFS profile is generally used for determination of the distance between target and the nearest atom. The nearest atom of the C14 Laves phase is the own element (ex. the nearest atom of A site is element of A site) and the distance between sphere radii of A site is larger than that of B site.¹⁵⁾ While Ti- and Zr-K profiles in all samples have a peak at ∼2.3 Å, Mn and Cr have a peak at ∼2 Å (Fig. 2). The distances between Ti and Zr from nearest atom are larger than Mn and Cr atoms, which means that Ti and Zr occupy A site and Mn and Cr atoms occupy B site. Considering all samples have almost identical profiles for each Ti, Zr, Mn, and Cr element, the structure of AB₂ and AB₂-M could be identical. For the A elements, Zr profiles for all samples are completely identical. Peak positions of Ti for all samples in Fig. 2 are the same (2.33 Å), which are the close to Zr (2.36 Å). On the other hand, peak positions of Mn profiles are slightly different (AB₂, AB₂-Fe, and AB₂-Cu: 1.99 Å, AB₂-Co: 1.96 Å, AB₂-Ni: 2.02 Å). The peak positions of Cr in AB₂-M are not same (AB₂ and AB₂-Co: 2.06 Å, AB₂-Fe, AB₂-Ni, and AB₂-Cu: 2.09 Å), and slightly larger than Mn position. This small difference of peak positions of Mn and Cr profiles could be effect of doped metal because doped metals are the different atomic radius from Mn and Cr. The distances from the nearest atom for all doped M elements were approximately 2 Å, which are the quite similar distance to Mn and Cr cases. This fact would be a strong evidence of that all doped metals occupy B site. M in the AB₂-Fe, AB₂-Co, and AB₂-Cu have close peak position to Mn, while peak position of M in the AB₂-Ni is close to (or slightly shorter than) Cr. We are not sure the reason of such difference (or Ni may migrate into A site) because tendency of the distance does not fit with order of atomic radius of metals. Since EXAFS profiles of A elements in AB₂-M have no peak (or quite small shoulder) at ∼2 Å in Fig. 2, migration of A elements into B site does not occur (or slightly occur). B elements also seem not to migrate into A site (or Ni may partially migrate).

Fig. 2

Radial distance from target atoms to the nearest atom in AB₂ and AB₂-M.¹⁴⁾

3.2 Elemental analysis

Next, we investigated the detail composition of alloys using SEM-EDX in order to determine the degree of uniformity of alloys and to find the effect of doped metal in AB₂-M. In the previous study,⁷⁾ the average compositions of AB₂ and AB₂-M were close to the target composition (Ti_0.515Zr_0.485Mn_1.2Cr_0.8M_0.1) but all samples were composed of multiple phases, which are Ti-rich and Zr-rich phases. Samples synthesized in this study also seemed to be multiple phases because the composition determined by EDX depends on the spot in all samples. This result stands for that we cannot determine the proper composition of main phase if we use the average composition of all area. Therefore, we tried to determine the composition of each phase. First, ratio of A and B elements in AB₂ alloys was investigated, and the results are shown in Fig. 3. M elements in the alloys were counted as B elements as XAS results concluded. For all samples, concentrations of A elements have negative linear correlation to concentrations of B elements because the total mole of elements were normalized by A + B = 3. The order of the width of distribution of A to B ratio, is AB₂ < AB₂-Cu < AB₂-Ni < AB₂-Co < AB₂-Fe. Almost all phases in AB₂ and AB₂-M have less than 2 of B concentration. Since migration could not occur except Ni from XAFS results, vacancy could be doped in the B site when M elements were doped. Concentration of B elements in some spots for AB₂-Ni partially exceed to 2, also indicating partially Ni migrates into A site.

Fig. 3

Relationship of total amount of A and B elements in AB₂ and AB₂-M (normalized by A + B = 3).

Figure 4 shows the distribution of concentration for each element in AB₂-M. None of them looks single phase because their distributions are wide. Moreover, Ti/Zr in AB₂-Fe has wide distribution with no trends (or two peaks), while most of Ti/Zr, Mn/Cr, and M distributions of AB₂ and AB₂-M are Gaussian-like shape. The number of Ti/Zr > 1 spots are more than that of Ti/Zr < 1 spots, which stands for that Ti-rich phase is main phase (with minor Zr-rich phase) in all AB₂-M. Their Ti/Zr value in average and main phases (= Max. Frequency in the Fig. 4) are slightly higher than the target value. For the distribution of Mn/Cr, all samples have Gaussian shape but both of their main and average compositions are not identical and were less than the target. The distributions of M in all AB₂-M are also slightly similar to Gaussian shape, but AB₂-Cu has quite wide-range distribution. Both main and average compositions in all samples are different. Moreover, some of them have 2 peaks such as Ti/Zr in AB₂-Fe and AB₂-Ni, or main composition locates out of peak of the Gaussian.

Fig. 4

Distribution of composition ratio of A and B elements in AB₂ and AB₂-M.

In order to determine the main phase, relationship between A and B elements has been investigated. The Ti/Zr for all samples have an excellent correlation with Mn/Cr as shown in Fig. 5: the higher Ti concentration, the higher Mn concentration. AB₂ distributes in wide range (Ti/Zr: 0.71–1.82, Mn/Cr: 1.24–1.58) and no dense area, indicating AB₂ has wide composition. On the other hand, AB₂-M samples distribute narrower than AB₂ except for AB₂-Fe, indicating M has a role of forming homogeneous phase except for Fe. AB₂-Ni and AB₂-Cu have dense area although they are still composed of multiple phases as discussed above. AB₂-Cu could be composed of mainly 2 phases because there are 2 dense areas, which are surrounded in dashed squares in Fig. 5. The main phase of AB₂-Cu could be Ti-rich phase. Thus, the order of uniformity, which is width of Ti/Zr-Mn/Cr, is AB₂ (1.11–0.36) = AB₂-Fe (1.65–0.31) < AB₂-Co (0.86–0.19) < AB₂-Ni (0.37–0.19) ≤ AB₂-Cu (0.17–0.14, 0.26–0.17). This trend is similar to the trend of atomic radius of B element (AB₂ includes Cr and Mn).

Fig. 5

Distribution of composition for Mn/Cr vs. Ti/Zr in AB₂ and AB₂-M.

Using average and max composition in Fig. 4, we estimated the molar ratio of B to A elements (B/A) and average/main composition which are denoted in the Table 1. We employed denser phase as the main phase if two peaks showed in the distribution. The order of B/A in main phase is interestingly the same order as CO₂ poisoning tolerance of AB and AB₂-M (AB₂-Ni < AB₂ < AB₂-Co < AB₂-Fe).⁷⁾ The difference between composition of average and main phase are less than 0.2. Average compositions in all samples were closer to the target composition than those of main phases. However, composition of both average and main phases were Ti-rich (different from previous study) and small Mn composition (similar to previous study). Reason of small Mn composition is probably because of evaporation during the arc melting.

Table 1 Composition distribution of AB₂-M by spot analysis.

From these results, AB₂ and AB₂-M synthesized in this study were composed by multiple phases with vacancy although main and average compositions were close to the target.

3.3 XRD and NPD

Figure 6 shows the XRD profiles of AB₂ and AB₂-M. Since all samples are corresponding to C14 Laves structure,¹⁶⁾ they have quite similar XRD (The full-scale of XRD profiles are shown in Fig. A1 in appendix) and NPD patterns (Fig. A2). However, their peak sharpness of XRD profiles are different. The full width at half maximum (FWHM) of 103 peak at ∼39° ¹⁷⁾ are AB₂ (0.486) > AB₂-Fe (0.438) > AB₂-Co (0.403) > AB₂-Ni (0.395) > AB₂-Cu (0.234). The expanse of FWHM is affected by smaller grain size but also inhomogeneous composition. Since there is no significant difference in grain size among AB₂ and AB₂-M, such wide FWHM of peak can be explained by inhomogeneous composition, which is also obtained from SEM-EDX results. As AB₂ and AB₂-M are composed of multiple phases, it is difficult to determine the crystal structure by Rietveld refinement. We tried to refine their structures from NPD profiles but none of them reached less than 5.9% of R_wp due to inhomogeneous composition (Fig. A2). We have obtained the tendency that the R_wp of M = B case are always less than that of M = A case. This is also one of the evidences that M locates at B site.

Fig. 6

XRD profiles of AB₂-M (33–44°).

3.4 Discussion

The main issue for AB₂-M is why dopant M drastically affects CO₂ tolerance. The composition distribution seems not to affect the CO₂ tolerance from SEM-EDX results. The only significant difference between AB₂-M is dopant, which locates on B site. We have already obtained the results that electron state of B element is important for CO₂ tolerance in the case of AB₅.⁸⁾ These facts lead to a conclusion that element occupying B site in AB₂-type alloys also could strongly affect CO₂ tolerance even if doping small amount despite composition is not uniform. Therefore, CO₂ tolerance could increase if concentration of Fe in AB₂-M increases, which can be future work.

4. Conclusion

The approach in this study is to determine the detail of composition of AB₂-type hydrogen storage alloys including small amount of doped metal (M) in order to understand the mechanism of CO₂ tolerance. XANES profiles of AB₂ and AB₂-M were confirmed as metallic alloy phase. EXAFS revealed that all doped M locates in B site, where the distance between target element and the nearest atom was the same as Mn and Cr (∼2 Å). SEM-EDX results showed that AB₂-M synthesized in this study had close average/main composition to the target composition, which is Ti_0.515Zr_0.485Mn_1.2Cr_0.8M_0.1 (M = Fe, Co, Ni, and Cu). However, all of them have inhomogeneous phase. The result of elemental analysis using SEM-EDX also revealed that Ti/Zr ratio in AB₂-M increases when Mn/Cr ratio increases. The homogeneity of composition affects the FWHM of 103 peak in their XRD patterns, which is consistent of SEM-EDX. Rietveld refinement using NPD patterns of AB₂-M could not obtain good R_wp value probably due to inhomogeneous composition. From these results, CO₂ tolerance of AB₂-type has almost no relationship with composition but it could depend on the doped M, which is the same as AB₅-type alloys.

Acknowledgments

XAS measurement in this study was performed in SPring-8 (Proposal No. 2015B1776) with funding support from the Foundation for high energy accelerator science, Japan. This work was partially supported by the Neutron Scattering Program Advisory Committee of IMSS, KEK (Proposal No. 2014S06). The authors thank Itsuki Miyazato from the University of the Ryukyus (master student at this time) for XAS measurement. Figure 1 and 2 are from his master thesis.

REFERENCES

• 1)   S.  Ahmed and  M.  Krumpelt: Int. J. Hydrogen Energ. 26 (2001) 291–301.
• 2)   S.  Miura,  A.  Fujisawa and  M.  Ishida: Int. J. Hydrogen Energ. 37 (2012) 2794–2799.
• 3)   S.  Miura,  A.  Fujisawa,  S.  Tomekawa,  Y.  Taniguchi,  N.  Hanada and  M.  Ishida: J. Alloy. Compd. 580 (2013) S414–S417.
• 4)   A.  Fujisawa,  S.  Miura,  Y.  Mitsutake and  M.  Monde: J. Alloy. Compd. 580 (2013) S423–S426.
• 5)   H.  Homma,  H.  Saitoh,  A.  Kamegawa and  M.  Okada: Mater. Trans. 43 (2002) 2741–2747.
• 6)   G.D.  Sandrock and  P.D.  Goodell: J. Less-Common Met. 104 (1984) 159–173.
• 7)   N.  Hanada,  H.  Asada,  T.  Nakagawa,  H.  Higa,  M.  Ishida,  D.  Heshiki,  T.  Toki,  I.  Saita,  K.  Asano,  Y.  Nakamura,  A.  Fujisawa and  S.  Miura: J. Alloy. Compd. 705 (2017) 507–516.
• 8)   N.  Hanada,  T.  Nakagawa,  H.  Asada,  M.  Ishida,  K.  Takahashi,  S.  Isobe,  I.  Saita,  K.  Asano,  Y.  Nakamura,  A.  Fujisawa and  S.  Miura: J. Alloy. Compd. 647 (2015) 198–203.
• 9)   H.  Nakano and  S.  Wakao: J. Alloy. Compd. 231 (1995) 587–593.
• 10)   R.M.  Waterstrat,  B.N.  Das and  P.A.  Beck: Trans. Metall. AIME 224 (1962) 512–518.
• 11)   R.  Hempelmann,  E.  Wicke,  G.  Hilscher and  G.  Wiesinger: Ber. Bunsenges. Phys. Chem. 87 (1983) 48–55.
• 12)   R.  Oishi,  M.  Yonemura,  Y.  Nishimaki,  S.  Torii,  A.  Hoshikawa,  T.  Ishigaki,  T.  Morishima,  K.  Mori and  T.  Kamiyama: Nucl. Instrum. Methods Phys. Res. A 600 (2009) 94–96.
• 13)   R.  Oishi-Tomiyasu,  M.  Yonemura,  T.  Morishima,  A.  Hoshikawa,  S.  Torii,  T.  Ishigaki and  T.  Kamiyama: J. Appl. Cryst. 45 (2012) 299–308.
• 14)  I. Miyazato: Master thesis, University of the Ryukyus, (2016).
• 15)   S.  Ishida,  S.  Asano and  J.  Ishida: J. Phys. Soc. Jpn. 54 (1985) 3925–3933.
• 16)   J.M.  Park and  J.Y.  Lee: J. Less-Common Met. 167 (1991) 245–253.
• 17)   J.L.  Bobet,  B.  Chevalier and  B.  Darriet: Intermetallics 8 (2000) 359–363.

Appendix

Fig. A1

Full-scale of XRD profiles of AB₂-M.

Fig. A2

The results of Rietveld refinement for neutron diffraction profiles of AB₂ and AB₂-M. The compositions are fixed to (a) average and (b) main phase of SEM-EDX results corresponding to Table 1. NPD profiles, which are normalized to the profiles of the vanadium standard sample and the background. The Bragg reflection positions are shown for the hexagonal Laves phase.