Abstract

Laves compounds have been considered as one of the prominent candidates of hydrogen storage materials, since they have relatively good hydrogen capacity compared to AB5 type compounds such as LaNi5 and also can charge/discharge hydrogen under ambient conditions. Laves AB2 structures are effectively viewed as tetrahedral close-packing of constituent A and B atoms, where their atomic size ratio, i.e., RA/RB (R denotes the relevant atomic radius) plays a critical role for a stability of the compounds. Empirically, hydrogen storage AB2 compounds are designed toward less-dense packing conditions in order to incorporate hydrogen atoms into interstitial spaces within the structure. However, there is a definite limit for this strategy; it is a well-known phenomenon that the Laves structures are transformed into amorphous during hydrogenation when the RA/RB exceeds critical value of approximately 1.37. This is referred to as hydrogen-induced amorphization (HIA), which was indeed confirmed for a large number of AB2 compounds with C15-type cubic structure. Therefore, understanding microscopic origin of the HIA phenomenon is an important key to improve furthers the hydrogen-storage ability of the Laves compounds. So far, HIA have been studied mostly by X-ray diffraction. In the present study, we investigate the details of microstructural changes of Laves RENi2 (RE = Pr, Gd, La) compounds using mainly conventional transmission electron microscopy observation (CTEM) and scanning transmission electron microscopy (STEM) equipped with electron energy loss spectrum (EELS) detector.

In chapter 1, general background of the present study is introduced. Metal hydride the most prominent type of hydrogen storage is explained briefly. Characteristics of Laves and pseudo-Laves compounds are elucidated including HIA.

In chapter 2, experimental procedures are described. For sample synthesis, arc-melting was used for RENi2 (RE = Pr, Gd, La) and induction melting for pseudo-Laves (Pr(1-x)Mgx)Ni2 compounds. Sieverts apparatus was used to measure hydrogen capacities of each compound. To investigate the structural changes of Laves and pseudo-Laves compounds, (S)TEM observation was performed mainly.

In chapter 3-5, the main results of the present study are described. Chapter 3 contains investigation of microstructure changes of Laves RENi2 (RE = Pr, Gd, La) compounds. It is turned out that hydrogenation-processed microstructures of RENi2 compounds, supposed to be amorphized according to their RA/RB, are not pure amorphous structure but composed of Ni nanocrystals embedded in an amorphous matrix of PrH2. On this basis, we propose hydrogenation-induced micro-phase separation (HIMPS), which provides a more precise/correct view of the phenomenon that has so far been referred to as HIA.

For hydride properties, the amounts of hydrogen absorption of Laves RENi2 compounds were proportional to the atomic size ratio (RA/RB), since the interstitial space increases with RA/RB. But most of absorbed hydrogen cannot be desorbed due to HIA (i.e, HIMPS in the present study). It seems that the entire amorphization occurred based on X-ray diffraction patterns of Laves RENi2 compounds. However, the electron diffraction patterns reveal significant features, which are hardly detected by X-ray diffraction. For the hydrogenated Laves RENi2 compounds, the patterns are not single halo-ring typical of amorphous structure but definitely composed of multiple rings even after the disappearance of distinct Bragg reflection in the X-ray patterns. This immediately suggests that the structure is not pure amorphous but perhaps microcrystalline states responsible for generating the powder diffraction pattern (i.e., Debye-Scherrer rings). In fact, the multi-rings can be successfully indexed by the crystalline Ni with face-centered-cubic (fcc) structure and the amorphous hydride REH2. Based on the results, HIMPS can be described as follow;

Dark field images obtained from hydrogenated RENi2 compounds directly shows HIMPS phenomenon in Laves phase. By imaging with the several parts of the innermost strong-ring indexed as (111)(REH2) in the dark field image obtained from RENi2 compound, the contrast always appear to be homogeneous all over the matrix. On the contrary, when the DF images are formed with the second strong-ring indexed as (111)(fcc-Ni) in the same manner, there appear to be significant variations with strong diffraction contrast. On the basis of these DF-TEM observations, it can be concluded that the microstructure is not pure amorphous but composed of nano-crystals embedded in an amorphous matrix, and they are reasonably identified as fcc-Ni crystals and REH2 amorphous by indexing the relevant peaks in the electron diffraction pattern. Atomic-resolution TEM/STEM images also confirmed the presence of Ni nanocrystals embedded amorphous structure of hydrogenated Laves RENi2 compounds. Additionally, chemical analysis using EELS was performed to support HIMPS phenomenon. The nanocrystals are identified to be almost pure Ni whose plasmon peak appears at ~25.0 eV. The amorphous matrix reveals two peaks in EEL spectra, which are basically interpreted as due to plasmon excitations (~ 15 eV for Pr as an example) and excitations of the 5p electrons (~ 30 eV for Pr) of rare-earth metals. It is empirically known that, after hydrogenation (i.e., forming hydrides), the plasmon peak of rare-earth metals generally reveal upward shift about 2~3 eV, a tendency of which is qualitatively understood as a mean increase of valence electrons provided by hydrogen atoms based on nearly free electron approximations. For the present amorphous matrix, the relevant peak shift is in fact observed when compared with the peaks from the original RENi2 compound, supporting that the amorphous matrix is composed of PrH2.

In chapter 4, we suggest new sight to interpret HIMPS (i.e., HIA in the previous study). So far, HIA phenomenon of Laves compounds has been well described by topological way (i.e., critical point of RA/RB, ~1.37). We confirmed that HIMPS occurs instead of HIA in Laves compounds in a previous chapter. Accordingly, the question arises how the micro-phase separation proceeds in Laves compounds. In the present study, we assume that a driving force for the phase separation may be related to enthalpy changes during hydride formation. Miedema's semi-empirical model was used to estimate the enthalpy changes of Laves compounds caused by HIMPS in Laves compounds. The relation of Miedema's model is as follow;

where "1-F" term is a correction term supplementing the original relation of Miedema's model. The detail explanation of the correction term is described in chapter 4. Since this relation seems to describe phase separation between A and B elements very well, we used this model to estimate the enthalpy changes of Laves compounds induced by HIMPS. As a consequence, we confirmed that the enthalpy change of HIMPS on Laves compounds correlates to the atomic size ratio. It is shown that out that the HIMPS phenomenon can be explained well by thermodynamic criteria as well as topological way. Additionally, we find a particular group of the hydrogenated Laves compounds, which can be explained by thermodynamic criteria but not by topological criteria. After the studies by Aoki et. al, which it was immediately pointed out that the RA/RB criterion cannot be applied for some of the AB2 compounds. For some compounds of which atomic size ratio exceeds the critical point of RA/RB, HIMPS does not occur at room temperature, but occurs only with heat-assisted condition. We find that three groups of Laves compounds with occurrence of HIMPS are divided by different range of the enthalpy changes estimated by Miedema's model. Accordingly, we assume that HIMPS may occur when the phase separation is rather stable than hydride phase maintain C15 Laves structure. To ensure our assumption, estimation of possible enthalpy changes of the hydride phase with Laves structure is necessary. We use first principles calculation to estimate the enthalpy changes of Laves phase hydride and compare to those of HIMPS. There are three tetrahedral hydrogen sites with different chemical environment (B4, AB3, A2B2). Generally hydrogen atoms cannot be positioned in the B4 site since "B" element is non-hydriding element. Thus, we considered two hydrogen sites (AB3, A2B2) for the calculation. The results were obtained with self-consistent density functional theory (DFT) calculation using the Vienna Ab initio Simulation package (VASP). As a result, we found that until the atomic size ratio up to ~ 1.37, the state of hydride phase maintaining Laves structure is more stable than the state of HIMPS in Laves compounds. In contrast, as the atomic size ratio exceeds the critical point, it reveals that the state of HIMPS is more stable than that of hydride Laves phase.

In chapter 5, we try to find the Laves compounds available for hydrogen charge-discharge. By choosing the element A and B, atomic-size ratio can be controlled. However, a precise control is difficult due to the selecting limitation of elements forming Laves phase. In pseudo-Laves compounds, a precise control of atomic-size ratio is possible by the composition change. Pseudo-Laves (Pr(1-x)Mgx)Ni2 (x = 0.3, 0.5. 0.7) compounds were prepared. We checked the occurrence of HIMPS as increasing amounts of Mg in pseudo-Laves (Pr(1-x)Mgx)Ni2 compound. When x reaches 0.5, the pseudo-Laves compounds firstly can desorb hydrogen. The atomic size ratio of pseudo-Laves (Pr(0.5)Mg(0.5))Ni2 is 1.37, the exact critical point in a topological criteria. In addition, we presume that at the critical point of HIMPS some microstructural changes may occur. It is found that the precursor structural changes already occur in the hydrogenation-processed (Pr0.5Mg0.5)Ni2 compound. From a combination of STEM and EELS analysis, Ni-rich regions and Pr-rich regions were observed in the (Pr(0.5)Mg(0.5))Ni2 compound. It seems that fcc Ni clusters were formed in the matrix of the (Pr0.5Mg0.5)Ni2. It is assumed that hydrogen atoms may be captured at Pr-rich regions. Accordingly, we investigated (Pr(0.5)Mg(0.5))Ni2 compound as increasing a cycle of hydrogen charge-discharge. We confirmed that amount of captured hydrogen gradually increases with hydrogen charge-discharge cycles from the upward shift of plasmon peaks of Pr.

Finally, in chapter 6, we summarize the present thesis: Hydrogen-induced microstructure changes of Laves compounds, i.e., the previous HIA should be replaced with HIMPS. The HIMPS phenomenon can be successfully described by enthalpy changes during the hydrogenation. Energetically, the HIMPS state is more stable than of hydride state Laves phase for Laves compounds, whose RA/RB exceeds the critical point.

水素エネルギーの有効利用促進にあたり,水素の安全かつ効率的な貯蔵を実現する水素吸蔵物質の開発は急務の課題である。水素吸蔵合金は,水素ガスとの反応により水素原子を直接結晶格子内へと取り込むため,単位体積辺りの吸蔵量に優れると共に,不純物を殆ど含まない高純度での水素貯蔵を可能とする。従来の研究により,その化学量論組成がAB5,またはAB2で表される金属間化合物群が水素吸蔵合金として有望であることが示されている。後者のAB2型は特にラーベス化合物と呼ばれ,非常に多くの金属元素の組み合わせで形成されるため,合金設計による特性改善が期待されてきた。ラーベス化合物の水素吸蔵量を増加させるための設計指針としては,化合物を構成するA,B原子のサイズ比をできるだけ大きくし,格子体積を大きくする方法がある。しかしながら,構成原子サイズ比が1.37を越えると,ラーベス化合物が水素吸蔵に伴ってアモルファス化してしまう(Hydrogenation-Induced Amorphization: HIA)という現象が見いだされ,AB2型化合物の水素吸蔵性能が頭打ちとなってしまった。また,HIA現象がなぜ起こるのかというメカニズムに関しても殆ど理解が進まなかったため,HIA克服の糸口もつかめない状況であった。

上記の問題を踏まえて,本研究ではHIA現象のメカニズム解明,およびその克服を目的とし,いくつかの典型的なAB2型ラーベス化合物の水素化過程に伴う微細構造変化を詳細に調べた。本論文は6章からなる。

第1章は序論であり,種々の水素吸蔵合金との比較に基づいてAB2型ラーベス化合物の水素吸蔵特性を述べると共に,HIAがどのような経緯で見いだされたのかを整理し,本研究の目的と位置づけを明確化している。

第2章では,本研究を進めるにあたっての実験方法が述べられている。従来,水素化に伴って生成したとされるアモルファス構造相が,粉末X線回折法による検証のみであったことから,本研究では電子顕微鏡(TEM/STEM)による微細構造直接観察を行ったことが特筆すべき点である。ラーベス化合物試料としては,原子サイズ比の影響を系統的に調べる観点から,希土類(RE)-Niの2元系合金,およびPr-Mg-Ni3元系合金を作成した。

第3章では,RENi2(RE = Pr, Gd, La)ラーベス化合物に関する観察結果が述べられている。これらの化合物はいずれも原子半径比が1.37を越えており,水素吸蔵に伴い粉末X線回折パターンにおいてブラッグピークが消失することを確認した。従来は,この観察事実に基づきアモルファス化との結論がなされていた。本研究での電子回折,暗視野TEM観察およびSTEM分光法により,水素化後の試料が粒径1nm程度のNi微結晶とREH2のアモルファス相からなることが判明した。すなわち,水素化に伴って微視的スケールでの相分離を生じていることが,現象の本質であることを看破したのである。この新たな知見に基づき,従来のHIAに替わってHydrogenation-Induced Micro-Phase Separation (HIMPS)モデルを提案している。

第4章では,本研究で見いだされたHIMPS現象について,そのメカニズムを議論している。以前のHIAに関する研究においても,結晶幾何学的パラメータである構成原子サイズ比がどのような物性値と相関を持つのか検討されていたが,明瞭な解が見いだせていなかった。本研究では,水素化に伴うラーベス構造不安定性が相分離に起因することから,熱力学的な相安定性との関連性に着目し,種々のラーベス化合物の水素化に伴うエンタルピー変化をMiedema経験則に基づいて算出した。その値を構成原子サイズ比に対してプロットしたところ,非常によい相関を見いだすことに成功した。さらに,種々のラーベス化合物の水素化挙動を第一原理計算に基づいて検討し,構成原子サイズ比が1.37近傍の化合物において,ラーベス水素化物が相分離に対して相対的に不安定となる熱力学的臨界点に相当すると結論づけた。

第5章では,格子定数を組成によって系統的に変化させた擬2元系ラーベス化合物(Pr(1-x)Mgx)Ni2 (x = 0.3, 0.5. 0.7)の水素化に伴う微細構造変化を調べている。この合金系では,擬構成原子サイズ比がおよそ1.37 となる(Pr(0.5)Mg(0.5))Ni2においても水素吸蔵-放出が可能となることが報告されていた。本研究により,その水素化初期段階において,微視的にはNiの微結晶生成,すなわち局所的な相分離が進行していることが判明した。1~100回の水素吸蔵-放出サイクルに伴う微細構造を詳細に調べた結果,局所相分離が数nmオーダーのPrH2ドメイン生成により徐々に進行することが見いだされた。なぜこのような形態を取るのかは明らかでは無いが,擬2元化の合金設計により相分離速度が抑制され,水素吸蔵特性の改善が実現されると推論された。

第6章は以上の総括である。

以上を要するに,本研究は電子顕微鏡による微視的構造の直接観察を通して,代表的な水素吸蔵合金であるAB2型ラーベス化合物の水素化時に生じうる格子不安定性・アモルファス化現象が,実際には微視的なスケールでの相分離であることをこと看破するとともに,それが熱力学的な因子に基づいて合理的に説明できることを示した。これは本研究によって初めて解明された知見であり,水素吸蔵物質研究のみならず,広く材料科学の観点からも極めて意義深いと言える。

よって本論文は博士(工学)の学位請求論文として合格と認められる。