1.Introduction

In this thesis, two experimental setups are presented which I developed. One is a time-of-flight measurement system following laser induced desorption of noble gas atoms from metal surfaces. The other is an ultra-high vacuum chamber installed in a synchrotron radiation facility (SPring-8 BL09XU, Japan) for measuring nuclear resonant scattering of synchrotron radiation by Kr multi- and monolayer physisorbed on a solid surface. I used the former instrument to investigate the photon stimulated desorption of Xe atoms from a Au(001) surface. The latter is used to investigate the local electronic structure of Kr at a TiO2(110) surface using nuclear energy levels of 83Kr as a probe.

2.Photon stimulated desorption of Xe

The laser induced system consists of pulse laser sources, sample specimen and quadrupole mass spectrometer as a detector, the latter two of which are enclosed in a ultra-high vacuum chamber. The sample specimen is a clean Au(001) surface cooled at 20 K, ontop of which physisorbed Xe layers are prepared by backfilling the chamber. The quadrupole mass spectrometer is tuned to acquire a high sensitivity to Xe atoms and is synchronized with the laser pulse source. The desorption signals of Xe following pulse laser irradiations are recorded with a digital oscilloscope.

2.1. Thermal and non-thermal desorption of Xe

I experimentally obtained a series of time-of-flight spectra of Xe atoms following the pulse laser irradiations at 2.3 and 6.4 eV and at various laser fluence. The time-of-flight spectra are well analyzed with a sum of two Maxwell-Boltzmann distributions. The minor component is much smaller than the other and its origin is sill unclear. Thus, we discuss only the major component. With this analysis, I obtained the desorption yield and translational temperature of desorbing Xe as a function of a wide range of the laser fluence.

The representative results are that at a large photon fluence the desorption of Xe is induced both at 2.3 and 6.4 eV photons indicating a desorption process independent of photon energy, whereas at a small photon fluence region the desorption of Xe is detected only at 6.4 eV indicating a desorption process dependent of photon energy. To further clarify the desorption process, I carried out numerical calculations of the time evolution of the surface temperature rise and desorption flux following the pulse laser irradiations. The calculated desorption yield and translational temperature both well reproduced the experimentally observed values at large laser fluence, indicating that laser induced thermal desorption of Xe is a dominant desorption process at this laser fluence. At the same time, the analysis highlights that the laser induced thermal desorption is inconsistent with the desorption process observed at a small laser fluence where translational temperature of Xe is much higher than the surface temperature and that the desorption yield linearly depend on the photon fluence. The desorption process at a small laser fluence region is, thus, concluded to be a non-thermal process.

I speculate that the non-thermal desorption of Xe observed only at 6.4 eV involves a transient Xe- formation on Au. The non-thermal process should involve an electronic excitation process which is reachable with 6.4 eV and not with 2.3 eV photons. The possibilities of direct excitations, ionization of Xe and other elemental excitations are safely excluded, for they do not explain the observed photon energy dependence. The desorption dynamics following electronic excitation is assumed to be the Antoniewicz model. In order to clarify the validity of the model, I carried out numerical calculations to obtain the time-of-flight spectra and desorption probability based on a classical dynamics of Xe atoms on an adiabatic potential. The calculated results of the time-of-flight and desorption cross sections well reproduces when the lifetime of Xe- state on Au is assumed to be about ~15 fs.

2.2. Post-desorption collision effect

Another experiment using the laser desorption system concerns the post-desorption collision effects on the time-of-flight spectra. In the experiment, I fixed the laser fluence at the large laser fluence region in order to intentionally induce the thermal desorption of Xe. On the other hand, the initially prepared Xe coverage is varied from 0.3 monolayer to 10 monolayers, which is effectively a variation of desorption flux.

I obtained a series of time-of-flight spectra as a function of the initial Xe coverage. The spectrum close to 0 is well analyzed with a Maxwell-Boltzmann distribution with a translational temperature close to the surface temperature. On the other hand, with increasing the initial Xe coverage, the peak position of time-of-flight spectra shifts to the smaller time-of-flight regions. Furthermore, the time-of-flight spectrum become unable to be analyzed with a Maxwell-Boltzmann velocity distributions with increasing initial Xe coverage, whereas they become able to be analyzed with a shifted Maxwell-Boltzmann velocity distributions which is characterized with a stream velocity.

Analyzing the obtained time-of-flight spectra with the shifted Maxwell-Boltzmann velocity distribution, we deduced the translational temperatures and the stream velocities as a function of the initial Xe coverage. The translational temperature is at 300 K at the initial Xe coverage close to 0. It decreased with increasing initial Xe coverage and become constant at about 190 K with the initial Xe coverage above 3 ML. The stream velocity is close to 0 at the initial Xe coverage close to 0. It increases from 0 to 140 m/s with increasing initial Xe coverage and become constant about 140 m/s.

The time-of-flight spectra at the initial Xe coverage close to 0 is well explained by a normal laser induced thermal desorption, whereas those at the large initial Xe coverage is inconsistent with the thermal desorption. The observed modification of the time-of-flight spectra increases with increasing initial Xe coverage which is assumed to be proportional to the desorption flux. Hence, I assumed that the modifications of the time-of-flight spectra are due to the post-desorption effects, which is in good agreement with the previous reports.

The observed phenomena that the translational temperature and the stream velocity become constant at the large initial Xe coverage are not explained by the post-desorption collision effects. I speculated that at a large desorption flux the desorption flow become well described by a rarefied gas dynamics in which an appreciable atomic collisions occur leading to the thermal equilibration of inside the desorption gas flow. Knudsen layer formation theory is such a theory which describes the dynamics of the desorption flow from the surface in the strong evaporations. The theory predicts the if the Knudsen layer is formed the temperature and the velocity become constant at a characteristics values.

I compared the experimental values of translational temperature and the stream velocity with the prediction based the Knudsen layer formation theory. The comparison shows a very good agreement of experimental values with the theoretical values. Hence, it is conclude that the observed behavior of the translational temperature and the stream velocity at a large desorption flux is understood as a manifestation of the formation of the Knudsen layer above the surface.

3.Nuclear resonant scattering by (83)Kr at TiO2(110)

In the Mossbauer spectroscopy, resonant absorption of the nuclear energy levels are measured and one obtain the hyperfine splitting of its levels due to the local magnetic field or electric field gradient. The magnetic field or the electric field gradient is caused by the local magnetic state or electronic configurations of the atom. Thus, the method is of some significance when one is interested in local magnetic or electronic structures.

The local electronic structure of noble gas atoms at the surfaces is of some interest in view of the mechanism of the work function reduction after their adsorption. There were, however, several drawbacks for its application to studies on the physisorption systems. The method is not sensitive to the surface. The convenient radio-isotope source is not available for the noble gas atoms.

Recently, the use of the synchrotron radiation as a photon source for the Mossbauer spectroscopy has been developed which is called the nuclear resonant scattering of synchrotron radiation. This method has several advantages for the surface science. The wavelength is selective. Thus, the noble gas atoms like Kr are excitable. The synchrotron light is highly directive which is favorable as to the surface sensitivity, for it can be used with the glancing angle regime.

All the experiments were carried out in the SPring-8 BL09XU, where I installed an ultra-high vacuum chamber. The chamber is precisely positioned for the synchrotron light on a mobile table. The sample was a clean TiO2(110) cooled at 20 K suspended on a manipulator which is precisely rotated at the angle with the synchrotron light. The ground state of 83Kr possess I = 9/2 with parity +. The first excited state is located at 9.4 keV above the ground state which possess I = 7/2 with parity unchanged. It is the M1 transition.

The time spectra of the nuclear resonant scattering of synchrotron radiations by 83Kr mono- and multi-layers are obtained. The time spectrum at 5 monolayers shows a single exponential decay indicating the degenerate single peak in energy range. The time spectrum at monolayer shows, on the other hand, that the spectrum is well reproduced with a single exponential decay and a periodic function. The monolayer Kr is assumed to have a strong interaction with the substrate, which leads to the electronic modification of Kr.

The oscillation of the time spectrum observed at monolayer regime is analyzed assuming that it is a quantum beat originated from the quadrupole splitting of the nuclear energy levels due to the electric field gradient on the surface. By using the quadrupole moment in the previous reports and the quadruple splitting obtained in the present experiment, I deduced the electric field gradient applied to the Kr nuclei to be in the order of 1021 V/m2.

本論文は「Photon stimulated desorption of and nuclear resonant scattering by noble gas atoms at solid surfaces (表面吸着した希ガス原子からの光刺激脱離と核共鳴散乱)」と題し,論文提出者が行った研究の成果をまとめたものである.

論文は5章から成っている.

第1章は序論である.固体表面における希ガス分子の物理吸着と吸着系の電子状態に関する研究背景を概観した後,第一の課題である吸着分子の光刺激脱離に関する先行研究についてまとめている.先行研究では,エキシトン励起に起因した脱離機構が論じられてきたが,近年の分光研究の結果に言及し,電子親和準位を介した光刺激脱離の可能性を指摘している.続いて,第二の課題であるシンクロトロン放射光を用いた核共鳴散乱に関する研究背景をまとめ,従来のメスバウアー分光と比較し,その特徴と表面系への応用可能性を述べている.これらの研究背景を踏まえて,本研究の具体的な課題設定を行っている.

第2章では,光刺激脱離と核共鳴散乱の実験について,その原理と実験方法,さらに自身で開発した実験装置について述べている.光刺激脱離については,四重極質量分析器の高感度化と飛行時間スペクトルの測定・解析について詳述している.核共鳴散乱については,実験に用いた83Kr原子核の基本的性質について述べた後,入射X線の全反射角近傍で測定することで表面吸着系での測定が可能となることを述べている.83Krの基底状態(核スピン9/2)と励起状態(核スピン7/2)は電場勾配下でそれぞれ5重と4重に四極子分裂する.このためスペクトルが複雑になることが予想されるが,表面電場勾配の主軸を放射光の偏光と一致させることで,核共鳴の選択則から共鳴スペクトルの解析が容易になることを指摘している.

第3章は,実験結果である.

はじめに,第一の課題である光刺激脱離について,Au(100)表面でのXeに関する実験結果を述べている.まず,励起光強度が異なるときに測定した飛行時間スペクトルを示し,励起強度が強い場合には励起光の波長に依らず脱離が観測されるのに対して,励起強度が弱い場合には6.4eVでのみ脱離が生じることを示している.得られた飛行時間スペクトルから,脱離分子の並進温度と脱離強度を解析し,励起光強度依存性としてまとめている.1次元の熱拡散に基づく表面温度のシミュレーションから脱離分子の並進温度を計算し,実験結果と比較することで,励起光強度が強い領域では光誘起の熱脱離が生じていると結論している.また計算結果は,脱離分子の並進温度が励起光強度の増大に対して飽和傾向を示すこともよく再現している.続いてXeの被覆率を変化させたときの実験結果を示している.被覆率が増加するにつれて,飛行時間スペクトルのピーク位置がシフトし,スペクトルが修正マックスウェルボルツマン分布でよく表されることを示している.速度分布を特徴づける量として,分布の流速と並進温度を解析し,いずれも被覆率が5原子層で飽和傾向を示すことを明らかにしている.

次に第二の課題である核共鳴散乱について,83Kr/TiO2に対して行った実験結果を述べている.核共鳴散乱は,散乱の時間遅れ成分として観測される.入射光の波長を±50meVの範囲で掃引し,共鳴プロファイルを得ることに成功している.さらに共鳴成分強度をKr曝露量に対して測定することで,Krの表面への吸着確率を求めている.これらの結果をふまえ,膜厚が5原子層と1原子層の試料に対して,核共鳴時間スペクトルの測定を行っている.スペクトルから見かけの寿命を求め,いずれも自然寿命より短くなるダイナミカル効果が現れることを明らかにしている.

第4章では,実験結果に基づき考察を行っている.光刺激脱離について,励起光波長依存性の実験結果とXeの電子状態に基づき,初期電子励起状態として,基板金属からXeの電子親和準位への励起が生じていると議論している.続いて,吸着分子がイオン化したときの脱離機構として提唱されているAntoniewiczモデルに基づき,励起状態のポテンシャル上での原子ダイナミクスをシミュレーションし,実験結果を矛盾なく再現する励起状態の寿命として,15fsを導いている.孤立Xe原子の電子親和力が負であるにもかかわらず,吸着Xe原子の電子親和準位が安定化し長寿命となった要因として,鏡像力の影響と励起状態間の軌道混成効果を議論している.また,飛行時間スペクトルが被覆率依存性を示した要因として,脱離後の分子間衝突の効果を議論している.実験結果で得られた流速とマッハ数が,文献で報告されている理論計算とよい一致を示すことから,並進温度が飽和した5原子層で表面にはKnudsen層と呼ばれる熱平衡化層が形成されていると結論している.

続いて,核共鳴散乱について,実験で得られた時間スペクトルの解析を行っている.放射光の偏光と表面電場の主軸の関係を考察し,表面電場勾配強度を変化させて,共鳴スペクトルのシミュレーションを行っている.その結果に基づき実験の時間スペクトルと比較することで,有意な量子ビート構造は観測されていないと結論し,Krの吸着構造と表面電場勾配に関する議論を行っている.

第5章は本研究の結論であり,結果の要約とそこから得られた知見,今後の展望が述べられている.

以上を要約すると,本研究は,希ガス物理吸着層を対象とした光刺激脱離と核共鳴散乱の研究を,実験手法の開発をふまえて行ったもので,これらの成果は物理工学として顕著な寄与があったと評価できる.よって,本論文は博士(工学)の学位申請論文として合格と認められる.