Introduction

A goal of catalytic chemistry is to understand key issues to control dynamic catalysis of nanoparticles and surfacesby in situ characterization techniques, because new catalysts in industry have been developed in try-and-errortype investigation, and rational design of nanoparticle catalysts and surfaces is still hard to achieve in a molecularlevel. Among many characterization techniques, x-ray absorption fine structure (XAFS) is one of the most powerfultechniques for characterization of structures and electronic states of supported metal catalysts which have been employedin a variety of industrial chemical processes as well as in sustainable green processes, environmentally benignprocesses, automobile exhaust cleaning processes, fuel cells. Recently, time-resolved Dispersive-XAFS(DXAFS) hasreceived much attention as an inevitable technique to discover the origin and mechanism of the genesis of efficientsolid catalysis.

In this study, I have explored dynamic changes in the structure and electronic state of supported metal catalystsin redox processes by in situ time-resolved DXAFS, choosing two types of supported metal catalysts, a Re10 clustercatalyst supported on HZSM-5 zeolite(Re/HZSM-5) for catalytic phenol synthesis from benzene and O2 and twokinds of PtSn bimetallic nanoparticle catalysts supported on !-Al2O3 for reforming process and on active carbon forfuel cell.

Experimental

Sample preparation ; A 0.6 wt% Re/HZSM-5 catalyst was prepared by Chemical Vapor Deposition of CH3ReO3. A !-Al2O3-supported Pt-Sn catalyst was prepared by a conventional impregnation method using 0.4 M HCl aq.A carbon (KETJEN BLACK ECP600JD)-supported Pt-Sn catalyst was also prepared by a conventional adsorptionmethod as follows; typically 0.14 g H2PtCl・6H2O was dissolved into 0.1 M HCl aq. or 1 M HCl aq., and 0.061 gSnCl2・2H2O was added to each solvent under continuously stirring. The solution was stirred for 5 h at R.T. and thenfiltered. The remaining powder was kept under vacuum over night. The samples prepared with 1 M HCl aq. and 0.1M HCl aq. are denoted as 1 M PtSn/C and 0.1 M PtSn/C, respectively.

In situ XAFS measurement ; All in situ XAFS measurements were conducted at the Photon Factory in KEK. Insitu DXAFS experiments at Re LIII-edge were performed by a Bragg-type curved Si(111) crystal polychromator(R = 2.5 m) and a photo-diode array equipped with a phosphorescent device. In situ step-scan and quick XAFSmeasurements at Pt LIII edge were mainly carried out at the beamline of 9A, 9C, and 12C with Si(111) double crystalmonochromators (DCM), and some XAFS measurements at Pt-LIII edge and all in situ XAFS measurements at SnK-edge were conducted at the beamline of NW10A with Si(311) DCM. In situ DXAFS measurements at Pt LIII-edgeand Sn K-edge were measured at the beamline of NW2A. A Bragg-type Si(311) curved crystal polychromator(R =1.5 m) was used for Pt LIII-edge measurements and a Laue-type Si(511) curved crystal polychromator(R = 0.9 m) wasused for Sn K-edge measurements.

Results and Discussions

The oxidation of active Re10 clusters supported on HZSM-5

Phenol is one of the most important industrial chemicals, but the its chemical process, called as cumene process,has some drawbacks such as low-one path efficiency or treatments of lots of by-products. The direct phenolsynthesis by the the selective oxidation of benzene with molecular oxygen has been the most desired processand explored for decades in the light of economy and environment.The Re10 cluster supported on HZSM-5 is the first promising catalyst which transforms benzene to phenol with 93.9% selectivity at 9.9% conversion.

Fig. 1(a) shows Re LIII-edge time-resolved XAFSspectra for the Re10 cluster/HZSM-5 catalyst under O2atmosphere at 553 K, where the active Re10 clustertransforms to the inert ReO4 monomer. In the timeresolvedXAFS spectra, there were observed threeisosbestic points. It indicates that the XAFS spectracan be described as μt(E, t) = c(t)μt(E, 0) +(1 - c(t))μt(E,"); μt(E, 0) is the initial spectrum andμt(E,") is the final spectrum. The coefficient c(t)was described as c(t) = Ae-kt from the fitting results,where k# is the apparent rate constant. This impliesthat the oxidation of the Re10 cluster on HZSM-5 isof the first order to [Re10]. Thus, it is suggested that a Re10 cluster is swiftly oxidized to ReO4 monomers withoutformation of any undesirable intermediates and this should be a key of the high selectivity of the Re/HZSM-5 catalyst.In situ DXAFS measurements were conducted under O2 and Bz + O2 atmosphere at several temperature. The similaranalysis of these results gives Fig. 1(b) shows the Arrhenius plots of the oxidation of Re10 species under O2 and Bz+ O2. Their activation energies (Ea) were calculated to be 49 kJ mol-1 and 74 kJ mol-1, respectively. Although thereaction steps under the both atmospheres are the same, the activation energy was different from each other. Benzeneshows a positive effect on stabilizing the active cluster structure under the reaction atmosphere and this is a key of itshigh catalytic performance.

The reaction mechanism of phase separation of the Pt-Sn nanoparticles

Supported Pt-Sn bimetal catalysts are used as a multifunctional catalyst for reforming, alkane dehydrogenation,fuel cell electrode. Pt-Sn bimetal catalysts are also tolerant to CO poisoning, which is especially beneficial forcommercial fuel cell electrodes.Although the Pt-Sn bimetal catalysts have advantages in catalytic properties andeconomical efficiency, the bimetal phase can be dephased with dissociation of nano-alloy phase by oxidation, andthis is supposed to induce leaching of the Pt and Sn ions on the electrode and decrease the catalytic activity. Theinformation on the dynamic formation and dissociation of nano-alloy particles is crucial to understand and improvethe leaching processes.

The kinetic analysis of the oxidation of PtSn/!-Al2O3

From XRD pattern and in situ EXAFS experiments, the Pt-Sn alloy was formed n !-Al2O3 after reduction andit was mainly the Pt3Sn1 structure, which has the fcc structure and longer lattice constants than Pt bulk. Pt and Sn havebeen oxidized in 30 min under O2 atmosphere at 673 K. Fig. 2(a) shows the results of in situ DXAFS measurementsat Pt LIII-edge for PtSn/!-Al2O3.

The μt(E1, t) of Pt LIII-edge was fitted well with the double exponential function as μt(E1, t) = c1e-k'1 t + c2e'-k'2t.This indicates that the oxidation of Pt in PtSn/!-Al2O3 proceeds through at least two successive reactions. Fig. 2(c)shows the change of the white line intensity of Pt LIII-edge at different PO2.The kinetic analyses were performed inthe following manner; the white line intensity, μt(E1, t), was plotted against the reaction time(t) in Fig. 2 (b) and thefunction μt(E1, t) was fitted by μt(E1, t) = #cie-k'i t+y0 with the minimum i. Here k'i is an apparent rate constant, whichdepends on the pressure of gaseous oxygen.Supposing O2 adsorbs molecularly, the number of O2(ad) is described asNKPO2(1 + KPO2 )-1, thus k#i is described as k'i = ki(T)NKPO2(1 + KPO2 )-1. N is the number of adsorption point onthe surface of nanoparticle and K is adsorption equilibrium constant for O2 and ki is a true rate constant. The k'i plotagainst PO2 was fitted with the function to extract ki(T).

Each line has a turnoff point(pointed in Fig. 2(c)). It takes longer time for μt(E1, t) to reach the point as PO2becomes lower. Fig. 2(d) shows the dependence of k' for the oxidation of Pt on PO2 . As the k'1 is proportional to PO2 ,KPO2 must be much smaller than 1 and the number of O2 adsorbed on the surface of the nanoparticles is much lowerthan that at saturation. Thus k' can be described as k' = kNKPO2 . On the other hand, k'2 is be independent of PO2 .This means the equilibrium constant K is much lager than that of the first oxidation process.

Fig. 3(a) shows the change of the white line intensity,μt(E1, t), at Sn K-edge for PtSn/!-Al2O3. Theintensity of the white line of Sn K-edge is be describedby a single step scheme c・ e-k't in contrast to the oxidationof Pt. And the apparent rate constant of theoxidation of PO2(Fig. 3 (b)). These facts indicate thatPt and Sn can be oxidized simultaneously at the beginningof the oxidation process. Although Sn is oxidizedthrough single process, Pt is not completely oxidizedin the first step. Sn is comparable to that of the firstoxidation step of Pt and the dependence of k' is alsoin proportion to PO2 .

It is supposed that Pt nanoparticles are constructed after the first oxidation step and it is gradually oxidized. HighresolutionTEM images for Pt/SnO2 showed formation of Pt nanoparticles surrounded by amorphous SnOx on SnO2under the similar oxidation condition to the present XAFS experiment. The whole oxidation process of PtSn/!-Al2O3may be described as follows; Pt and Sn in the PtSn/!-Al2O3 are oxidized simultaneously at the beginning, which isthe phase separation process, and Pt becomes nanoparticles surrounded by amorphous SnOx.

The kinetic analysis of the oxidation of 1 M PtSn/C(Pt3Sn1/C) and 0.1 M PtSn/C(Pt1Sn1/C)

The PtSn structure in the reduced 1M PtSn/C and the reduced 0.1MPtSn/C were Pt3Sn1 and Pt1Sn1 determinedby XRD and EXAFS, respectively. PtSn alloy has several kinds of structure according to the atomic ratio of Sn andPt. Especially the Pt3Sn1 structure is considered as the active structure of the electrode of fuel cell. Here the structureof PtSn alloy is formed selectively with changing the concentration of HCl. Fig. 4(a) shows the results of in situDXAFS experiments at Pt-LIII edge for 1M PtSn/C, whose main structure is Pt3Sn1. The change of the intensity ofwhite line is analyzed under the assumption of the single step scheme. Fig. 4 (b) shows the plot of k# against PO2 ,where k# is described as kNKPO2(1 + KPO2 )-1 and k = 1.3 s-1 and K = 0.044. In situ DXAFS measurements at SnK-edge for Pt3Sn1/C were also conducted. The oxidation process of Sn was also fitted by a single exponential curve,resulted in k = 1.3 s-1.

By examining the change of white line intensity over longer time, it was obtained that Sn was completelyoxidized in 10 s and that Pt was gradually oxidized over 600 s like the oxidation of PtSn/!-Al2O3. This indicates thatPt is not oxidized in the single process. This is because SnOx covers Pt after the first oxidation step like the case ofPtSn/!-Al2O3.

In situ Pt LIII-edge DXAFS spectra of 0.1 M PtSn/C, whose main structure is Pt1Sn1, are shown in Fig. 5(a).The change of white line intensity is described as the single step scheme, but Pt of Pt1Sn1/C is less oxidized than thatof Pt3Sn1/C. The pressure dependence of k# is indicated in Fig. 5(b). k was calculated as 0.75 s-1 from the plot of k#against PO2 . The rate constant of the Pt oxidation in Pt1Sn1/C is smaller than that of Pt3Sn1/C. But, the rate constantof the Sn oxidation was comparable to that for Pt3Sn1/C. Sn of the Pt1Sn1/C was completely oxidized. Consideringthe results of high-resolution TEM of Pt/SnO2, SnOx is produced in the oxidation of Pt1Sn1/C and the Pt nanoparticlesin Pt1Sn1/C. Thus the Pt nanoparticle is kept away from the gas phase O2 due to the SnOx coverage. The oxidationof Pt nanoparticle in PtSn/C is strongly affected by the content of Sn in the initial PtSn bimetal nanoparticles. The insitu DXAFS revealed the kinetic parameters and dynamic oxidation behaviors for each of Pt and Sn sites in the Pt-Snbimetal nanoparticles supported on active carbon.

Fig. 1 (a) Time-resolved XAFS spectra of Re LIII-edge underO2 atmosphere, where the spectrum was measured in 0.1 s every1 s. (b) Arrhenius plots of the oxidation of the Re10 clustersunder O2 and a mixture of benzene and O2.

Fig. 2 (a) The time-resolved XAFS spectra of PtSn/!-Al2O3 Pt LIII edge in the oxidation process at PO2 = 21.0 kPa and T = 673K. (b) the change in the intensity of the white line of (a). (c) the variation of the intensity of white line under several PO2 . (d) therate constants for the oxidation of PtSn/!-Al2O3.

Fig. 3(a) The change of the white line intensity at Sn K-edgeat PO2 = 23.6 kPa and T = 673 K. (b) the rate constants for theoxidaion of Sn in PtSn/!-Al2O3.

Fig. 4 (a) The Pt LIII-edge time-resolved XAFS spectra ofPt3Sn/C at pO2 = 20 kPa and T = 573 K. (b) the rate constantsfor the oxidation of Pt3Sn/C.

Fig. 5 (a) The Pt LIII-edge time-resolved XAFS spectra ofPtSn/C at PO2 = 20 kPa and T = 573 K. (b) the rate constantsof oxidation of PtSn/C.

本論文は7章で構成されている。第1章は、イントロダクションであり、固体触媒の重要性と研究課題を述べてから、種々の分光学手法を用いた固体触媒の構造評価の研究をレビューしている。そこでは、X線吸収微細構造(XAFS)による測定手段の特徴と有用性を強調している。

第2章は、XAFSの原理と時間分解DXAFS法の詳細を述べている。時間分解DXAFS法で、固体触媒の反応条件下での構造変化を追跡するには、気体の拡散が律速となり実効的な時間分解能を制限していた。本研究では、新たに設計・製作した気体導入系からなるサンプルセルを用いることで、実効的な時間分解能を今までの1/50程度までに向上させ、秒オーダーで変化する構造の追跡を可能にしたことは特筆に値する。

第3章では、直接フェノール合成反応に活性なRe10クラスター/HZSM-5触媒が酸素雰囲気及びベンゼンと酸素の混合ガス雰囲気下で、Re10クラスターからReO4モノマー至る秒オーダーの時間変化を、Re LIII端の時間分解DXAFSスペクトルで測定した実験結果を記述している。実験結果の解析から、Re10クラスターの構造変化では、Re2やRe4などのフラグメント中間体が観測されないことを確認し、構造変化の反応機構を決定した。そして、Re10クラスター触媒の高選択性は、フラグメント中間体が存在しないことによるものであることを明らかにした。さらには、各々の反応雰囲気の下でのRe10クラスターの構造変化の反応速度定数と活性化エネルギーを導出した。そのうえで、構造変化の反応機構と活性化エネルギーの違いから、ベンゼンの存在がRe10クラスター触媒の活性の保持に寄与していることを解明した。本章の研究成果は、触媒能のメカニズムを解明するための新たなアプローチに先鞭をつけたものであり評価に値する。

第4章では、アルミナ上に担持した白金-スズ合金ナノ粒子の形成過程と合金ナノ粒子の酸化過程を、Pt LIII吸収端及びSn K吸収端の時間分解DXAFSスペクトルで測定した実験結果を記述している。スペクトルのwhite lineの時間変化から、Pt及びSnのそれぞれについて、合金形成過程と合金酸化過程の反応速度定数と活性化エネルギーを導出した。それらの反応速度定数と活性化エネルギーの違いから、合金形成過程と酸化過程における白金とスズそれぞれの反応性を化学反応速度論の観点から議論した。どちらの過程でも白金とスズが相互に影響し合い、その反応性を変えていることを解明した。これまでの研究では、元素毎に反応性を議論することは不可能であったが、本研究で用いた時間分解DXAFS法がそれを可能にしており、高い評価を与えることができる。

第5章は、活性炭上に担持した2種類の白金-スズ合金ナノ粒子(Pt3Sn, PtSn)の酸化過程を、Pt LIII吸収端及びSn K吸収端の時間分解DXAFSスペクトルで測定した実験結果を記述している。スペクトルのwhite lineの時間変化から、Pt3Sn相とPtSn相に含まれる白金及びスズのそれぞれについて、反応速度定数と活性化エネルギーを導出した。それらの反応速度定数と活性化エネルギーの違いに基づき、Pt3Sn相及びPtSn相に含まれる白金とスズ各々の反応性の違いを化学反応速度論の観点から議論し、合金中に含まれるスズの量が増加すると白金の酸化の進行が遅くなることを解明した。本研究は、実触媒材料の評価をするためには、時間分解DXAFSスペクトルに基づく化学反応速度論の議論が必要不可欠であることを明らかにしたものであり特筆に値する。

第6章では、酸化亜鉛上に担持したパラジウムの、PdZn合金の形成過程とPdZn合金の酸化過程をPd K吸収端のXAFSスペクトル及び時間分解DXAFSスペクトルを測定した実験結果を記述している。スペクトルのwhite lineの時間変化から、それぞれの過程の反応速度定数を導出した。また、XAFSスペクトルの構造解析から、反応時間毎のPdの構造を決定した。反応速度定数と構造パラメータから、合金の形成過程及び酸化過程では、パラジウムのナノ粒子が中間体として生成していることを明らかにした。合金形成及び酸化過程の構造変化を明瞭に示した前例はなく、本研究は、貴金属と遷移金属からなる合金ナノ粒子の形成過程について、新たな知見を与えるものである。

まとめの第7章では、元素毎の時間変化の情報が得られる時間分解DXAFS法は、反応条件下での触媒自身の構造及び反応性を解き明かすためには、なくてはならない手法であることが強調されている。また、本実験で得られた研究成果をベースとして、触媒自身の変化と触媒能の相関について議論している。

なお、第3章は、唯美津木 准教授、他6名、第4章は、稲田康宏 教授、他7名、第5章は、稲田康宏 教授、他7名、第6章は稲田康宏 教授、他5名との共同研究であるが、論文提出者が主体となって実験及び分析・検証を行ったもので、論文提出者の寄与が十分であると判断する。

したがって、博士(理学)の学位を授与できると認める。