This thesis presents a theoretical study about electron transport and dynamics of molecules on metal surfaces based on first-principles calculations. Molecular/metal interface is one of the most important subjects in surface and nano science because it is the spot where chemical reactions happen and is the place which determines functions of organic devices. Controlling catalytic reactions and functions of devices is a goal for many scientists although this is hindered by unpredictable behaviors of molecules on metal surfaces. This unpredictability comes from the fact that a molecular orbital (MO), of which energy is well defined in the gas phase, is shifted and broadened due to the interactions with metal surfaces. The degree of the modification of the MO varies a lot with metal species. Although the MO on metal surfaces can be described using a conventional first-principles calculation such as density functional theory (DFT), it is difficult to connect the information of the MO directly to macroscopic properties of devices such as current (I)-voltage (V) curves or reaction rates. One reason for the difficulty is that usually molecular layers are required to measure I-V curves. The orientations of the surrounding molecules, which are also affected by metal surfaces, significantly modify the energy level alignment of the MO. The observation of electronic structures at "buried" interfaces is experimentally difficult. Another reason is the presence of an external driving force to transport carriers or induce the atomic dynamics. No electric devices start to work without an applied bias. Such an external driving force also modifies the MO and then the consequent electric current forces the system to leave from the ground state.

However, once we can understand observable properties from microscopic information by overcoming above difficulties, it will be possible to design and control a molecular device and chemical reaction by adjusting the position and width of MOs from tremendous combination of molecules and metals. This is the chemical engineering in the literal meaning. To achieve this goal, constructing a correct model without ad hoc parameters is mostly required. The first-principles calculation is a useful method to describe the electronic state of a molecule near the metal surface without empirical parameters. If the first-principles calculation is able to identify MOs contributing mainly to transport, find a vibrational motion promoting the reaction and determine those properties in nonequilibrium conditions in addition to the ground state properties, more experimental results can be explained from a microscopic point of view. We list three scientific objects to be studied in this thesis: the surface Peierls transition, the energy level alignment at metal/organic interfaces and the molecular switch. Through these works, quantum schemes to shed light on the origins of those phenomena are developed by extending the conventional DFT to calculate physical quantities which are not obtained before.

In this work, properties of atoms and molecules on metal surfaces in contexts of above phenomena are investigated using DFT. The electron transport properties under nonequilibrium conditions are studied mainly by the nonequilibrium Green's function (NEGF) formalism implemented into the HiRUNE module. Within the scheme of the NEGF, the inelastic electron tunneling spectroscopy (IETS) signal is also calculated as well as the electron-phonon coupling matrix. The nudged elastic band method is generalized to be able to calculate the reaction path under current and is implemented into the SMEAGOL package. The resonance model to treat the vibrational heating is extended for systems having a high barrier and wide working bias.

This thesis consists of 8 chapters and chapter 1 gives the general introduction. Chapter 2 presents the basic concepts of DFT and its practical implementation in the SIEATA program, which is used throughout in this work. We show that the electronic states can be solved in the form of the eigenvalue problem using Kohn-Sham orbitals, under the scheme of DFT. In SIESTA the Kohn-Sham orbitals are expanded by localized basis functions, which are suitable to calculate transport properties because interactions between molecules and electrodes can be straightforwardly written in the matrix form. Therefore it is important to understand how to describe electronic states in the periodic cell using localized basis sets. We start from basic concepts such as the pseudopotential and construction of basis functions, and then describe how to practically calculate the total energy and atomic forces.

In chapter 3 we investigate the charge density wave (CDW) on Cu(001) covered by In, Pb and Bi using DFT. Although this study is not related to the transport properties, it contains some important concepts such as surface states, surface phonons and electron-phonon coupling between them. We explain the origins of surface states of In, Pb and Bi/Cu(001)-0.5 ML having c(2×2) structure that cause a phase transition via the generation of a CDW. Band-structure analysis suggests a common mechanism of CDW transition for In and Pb/Cu(001)-0.5 ML. The absorption of In and Pb on Cu(001) leads to a reduction in the energy of the band composing the edge of bulk band gap. As a result, the band enters the bulk band gap and has the character of the surface state. Since these surface states compose a well-nested Fermi surface, CDW transition would be induced. In addition, we study the relationship between the lattice distortions and the band dispersions of the surface states of In/Cu(001)-0.5 ML and W(001) in terms of the Jahn-Teller effect (JTE). While we find promoting modes that resolve the degeneracy of surface states of W(001), any surface-localized mode of In/Cu(001) does not split the band of surface state. Therefore, the static electron-phonon interaction of In/Cu(001) is weak. We propose that the dynamic JTE contributes to the CDW transition of In/Cu(001) and the dynamic nature is related to the spatial coherence length of the CDW. Our strategy of analysis based on the real-space movement of atoms and MOs will be useful in terms of chemical engineering of the Peierls transition.

Chapter 4 describes NEGF method, which is useful to study transport properties of molecules between electrodes. The HiRUNE module is designed to bring out the advantage of MO theory. The difference between HiRUNE and existing programs for transport calculations is explained. Not only the transport without scattering by nuclear vibrations (ballistic transport), but also corrections caused by the electron-phonon interaction are introduced.

In chapter 5 we study charge transport of PTCDA molecular layer on Ag and Al electrodes using NEGF-DFT method. To analyze roles of organic/metal interfacial states for transport, we examine two kinds of electrodes: Ag(111) and Al(111). By quantitative evaluation of the coupling strength between PTCDA molecular orbitals and electrodes, we find the creation of the Shockley-type state at the interface of PTCDA and Ag(111). In contrast, the Al(111) surface forms a strong chemical bond with PTCDA. A clear Shockley-type state is not created, and an ohmic I-V is found for contacts consisting of thin PTCDA layers and Al(111) electrodes. We also predict that further stacking of PTCDA layers will make I-V characteristics more Schottky-like for both Ag and Al electrodes, regardless the different microscopic mechanism.

Chapter 6 outlines two important concepts related to current-induced dynamics, current induced forces and vibrational heating. The high bias applied to the molecule between electrodes induces changes in structures and excites molecular vibrations. These effects can be observed as a breaking of metal wires or switching of conformations probed by changes in the electronic conductance. We present how we can interpret such a phenomenon from a microscopic view. Finally we describe a practical way to address these phenomena using first-principles calculations.

In chapter 7 we study dynamics of STM-induced switch of melamine/Cu(001) using all of methods described in previous chapters. We evaluate the conformation dependent transport properties, the reaction path and the bias dependent energy barrier from first principles by using the non-equilibrium Green's function formalism combined with density functional theory. Furthermore we calculate the IETS signal to identify the modes promoting the conformational changes of the molecule. The IETS signal also reveals why the high-energy molecular conformations do not relax back to the lowest energy state even for large applied bias voltage. We then formulate a vibrational heating model to describe the STM-induced switching in this system, where the switching rates and their dependence on the voltage and current are obtained by using parameters extracted from our first principles results. We find that the different switching behavior for positive and negative bias originates from the different electronic properties of the empty and filled states of melamine on Cu, combined with changes of barrier height with the applied bias.

Chapter 8 gives the general conclusion.

本論文は『Theoretical Study of Electron Transport and Dynamics of Molecules on Metal Surfaces Based on First-Principles Calculations (第一原理計算に基づく金属上の分子の電気伝導とダイナミクスの理論的研究)』と題し、全8章からなる。STMからの電流は、金属表面上の単分子の構造変化を引き起こすことがある。その反応確率を分子軌道の観点から理解することは、表面反応を制御していく上で非常に重要であるが、印加する電圧などの影響のため、その定量的な予測は困難である。申請者は、密度汎関数法(DFT)と非平衡グリーン関数法(NEGF)を組み合わせ、さらに電流存在下での反応活性障壁を求める非平衡Nudged Elastic Band(NeqNEB)法を開発した。これによって、伝導電子が分子振動を励起することで引き起こされる化学反応について、すべてのパラメータを第一原理計算から得てシミュレーションすることが可能となった。

第1章は序論であり、金属表面上に吸着した分子の分子軌道について、実験測定上での困難さと第一原理計算による予測の可能性について概説している。この分子の振動と分子軌道との相関によって引き起こされる現象について述べ、これらが有機半導体デバイスや表面化学反応の制御にとって重要であることを説明している。

第2章では、本論文で用いている第一原理的手法に共通するDFT法と、局在基底関数を用いたSIESTAパッケージへの実装について背景を記述している。

第3章では、第2章で説明したDFTを用いて、Cu(001)表面上に形成された金属単原子膜の電荷密度波転移の機構について論じている。Cu(001)表面のIn単原子膜は、温度によって可逆な電荷密度波転移によって対称性の低い低温相と対称性の高い高温相を行き来する。特定の振動モード方向に沿って原子構造を変化させ、バンド図をプロットすることにより、実空間の視点から電子・格子相互作用を解析している。電荷密度波相関長が短いとされるW(001)では、特定のモードに沿ってバンドの縮退が解け、パイエルス転移が誘起されると示唆している。一方電荷密度波相関長が長いとされるIn/Cu(001)では、静的な構造変化によっては縮退が解けなかったことから、In/Cu(001)の電荷密度波転移は動的ヤーン・テラー効果によって誘起されると結論している。

第4章では、金属電極に挟まれた分子あるいは分子膜の伝導度を計算するためのNEGFの基礎理論を記述している。DFTから得られた電子状態を元に、電流―電圧曲線、伝導度、さらに非弾性トンネル電子分光(IETS)の強度までを計算することが可能であると結論している。

第5章では、NEGF-DFTを用いて、さらに射影分子軌道を援用し、Ag(111)あるいはAl(111)電極に挟まれたPTCDA分子膜の伝導特性について論じている。電極がAgの場合は、Ag(111)の表面状態と分子軌道との混成による界面準位が形成されるため、理想的なSchottky型の界面準位接続となる。一方電極がAlの場合、強い軌道の混成によりAlと接するPTCDA分子の軌道が界面に閉じ込められ、膜厚を増やすとSchottky型に漸近することを明らかにしている。以上の結果から、分子軌道法による伝導機構の理解が重要であると結論している。

第6章では、伝導電子が引き起こす原子核のマイグレーションと分子振動励起についての基礎理論を記述している。NEGF-DFTによって電流が原子核に及ぼす力を計算することが可能であり、また振動励起についても分子軌道や分子振動の情報を元にモデルを構築できると結論している。

第7章では、NEGF-DFTを用いてCu(001)上のメラミン分子のスイッチ機構について論じている。電流存在下での反応活性障壁を求めるNeqNEB法を導入、実装し、NeqNEB法とIETS計算から、反応座標と励起される振動モードを確定している。一つの伝導電子が多数の分子振動を励起する過程を取り込んだ共鳴モデルを構築することにより、第一原理計算から得たパラメータによってスイッチ確率の再現に成功している。ミクロな情報をもとにした共鳴モデルによってスイッチ確率を再現できると結論している。

第8章は総括であり、本論文の結果をまとめている。

以上のように本論文は、従来のNEGF-DFT法を大きく発展拡張し、予測可能な理論計算の現実系への応用を可能にした。本論文で得られた理論的知見は、有機デバイスや触媒反応の理論化学、化学システム工学に大きく貢献する。

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。