Introduction

Perovskite oxides (with the crystal structure ABO3, as shown in Fig. 1a) are characterized by a widediversity of physical properties originating from the electronic d-orbits, ranging from strong electroniccorrelation effects to catalytic behavior. Crucially this crystal structure, supported by an oxygen framework, isrobust against ionic substitution, enabling these diverse physical properties to be incorporated in epitaxialheterostructures. Perovskite heterointerfaces and surfaces have shown exotic electronic phases and devicefunctionalities, including two-dimensional electronic states, magnetic Schottky junctions, and artificialmultiferroics. In this context, it is important to note that the electrostatic description of a heterointerface has onlytwo boundary parameters: an interface charge and an interface dipole. While the former, causing interfacialelectronic depletion or accumulation, is widely used to create novel interface physics, the latter, modifying theinterfacial band offset by an electrostatic dipole on the atomic scale (Fig. 1b and c), has not been fullyinvestigated in perovskite heterostructures. This is due to a combination of the relatively complex concept, aswell as the difficulties of experimentally evaluating the accurate band diagram. This thesis presents a systematicstudy of engineering interface dipoles at perovskite oxide heterointerfaces, dealing with both fundamentalaspects, and device applications. Although such dipole engineering has been a longstanding problem inconventional semiconductors, here a sizable but well-controlled interface dipole is demonstrated using an ioniccharge layer, which is readily incorporated in perovskite heterostructures. Because the design of interfacialelectronic phases and devices are based on the interfacial band alignment, the ability to arbitrarily control bandoffsets using interface dipoles can be extremely powerful for both pure and applied oxide research.

Theoretical background

The conceptual complexity of the surface or interface dipole comes from the fact that its absolute magnitude cannot be simply extracted from a measurement of the work function of the bulk materials. In a givensystem, by systematically changing the interface structure, only relative changes in the interface dipole can bemeasured. This restriction stems from the need to consider the electrostatic boundary conditions on an atomicscale. While first principle calculations have succeeded in extracting the absolute magnitude of the interfacedipole1, reproducing the interfacial discontinuity of core level energies in photoemission spectra, a moregeneralized theoretical framework is needed to address the universality of this issue. Here we discuss the mostgeneralized form for the absolute magnitude of the interface dipole, providing a conceptual foothold for thediscussion of interface dipole engineering in perovskite heterostructures

The interface dipole is defined as the difference of the electrostatic potential standard in each materialconstituting the interface. Although this potential standard is usually defined for a given material with respect tothe vacuum at infinity, the existence of a surface dipole modifies the electrostatic potential in the materials,making this definition defunct. Here, starting from the Green's function of the Poisson's equation, q r 0 40γ( q : charge, 0 vacuum permittivity, r : distance from the charge), the potential standard at vacuum infinityis recalculated as the average electrostatic potential in each local unit cell (uc). This defines the electrostaticpotential standard in a material, reflecting the interface dipole, but not the periodic potential variation on theatomic scale. The interface dipole (.) is calculated as:

.〓e is the elementary charge, and Azz Q' ( Az P ) and Bzz Q'( Bz P ) are the zz (z) component of the quadrapole(dipole) moment density in the unit cell of each material.z, describes the overall interface charge, where the first term, z, is the microscopic charge density inthe interface region arbitrarily defined between z = z1 and z = z2, and the second term is the contribution from thepolarization in each material. The first and second term in is the interface dipole from the quadrapolemoment in each material and that fromz, respectively (Fig. 2). , thus calculated, is invariant toarbitrary definitions of the interface region (z1 and z2) or the unit cell in each material, but at the same time,clearly shows the physical origin of the interface dipole. In the following experiments of the interface dipoleengineering, we designed the term of zat perovskite oxide heterointerfaces, as is also the case for theinterface dipole engineering in conventional semiconductors.

Demonstration

A Schottky junction, formed between a metal and a semiconductor, is defined by the Schottky barrierheight, SBH W-x+Δ(W : metal work function, iconductor electron affinity). Since SBH canbe measured in a variety of ways, this junction provides a device platform for studying the control of.Historically, in Schottky junctions formed with covalent semiconductors, SBH is almost independent of Wdue to an caused by unintended interface charges. Ionic semiconductors, such as the oxides, tend to mitigatethis problem2, facilitating as a controllable degree of freedom at heterointerfaces. In the followingexperiments, was tuned in Schottky junctions using perovskite oxide by inserting an ionic charge sheet at aheterointerface. This charge sheet induces a counter screening charge in the metal, creating the required toshift the SBH (Fig. 3a).

The Schottky junctions were fabricated by pulsed laser deposition (PLD), growing SrRuO3, a readilyavailable oxide with good metallicity, on {100} Nb-doped SrTiO3, a widely used N-type oxide semiconductor. Ahighly concentrated laser beam ablates the target material and deposits it onto a heated substrate to grow a singlecrystal thin film. Monitoring the diffraction intensity of an electron beam grazing the surface enables us tocontrol the film thickness at an atomic level during growth. 0-2 uc of LaTiO3 or SrAlOx were grown prior toSrRuO3, introducing (LaO)+ or (AlO2)- ionic charges at the SrRuO3/Nb:SrTiO3 heterointerface. This insertionwas confirmed by scanning transmission electron microscope (STEM) images as shown in Fig. 3b. Themagnitude of the shift of the SBH due to D was valuated through current-voltage (IV), capacitance-voltage(CV), internal photoemission (IPE), and X-ray photoemission (PES) measurements.

Fig. 4a shows PES spectra of Ti 2p3/2 core levels in 1 uc (LaO)+-inserted, non-inserted, and 2 uc (AlO2)--inserted Schottky junctions, plotting the photoemission intensity as a function of the binding energy withrespect to the Fermi level. Because the shift of the Ti 2p3/2 core level peak reflects changes in D, a shift tohigher binding energy by (LaO)+ insertion means a smaller SBH, and vice versa for (AlO2)-. The SBHs from allthe four measurements are summarized in Fig. 4b, showing an evolution from 0 eV to 1.7 eV by (LaO)+ or(AlO2)- insertion with respect to the original SBH of 1.2 eV for no insertion. This magnitude of D isremarkable with respect to conventional semiconductors, highlighting the advantage of ovskiteheterostructures. D was also controlled in a range of other heterointerfaces: SrRuO3/LaAlO3/Nb:SrTiO3,La0.5Sr0.5TiO3/SrAlOx/Nb:SrTiO3, and a0.7Sr0.3MnO3/SrMnO3/Nb:SrTiO3, where fundamental issues such as theeffects of in-plane inhomogeneity, surface charges, interfacial charged defects, and the interface terminationwere addressed.

Device application

A fundamental application of artificial interface dipole engineering is the controlled creation of Ohmic orrectifying interfaces. In particular, having a highly rectifying (insulating) interface is a key issue in transistors, inwhich the current channel should be electrically isolated from the control electrodes. A hot electron transistor(HET)3 has a tri-layer structure of semiconductor(emitter)/metal(base)/semiconductor(collector) as shown in Fig.5a. Hot electrons are ballistically transferred from the emitter to the collector, and can be controlled by the basecurrent. By incorporating ferromagnetic or ferroelectric oxides, multifunctional transistors can be expected, andmoreover, the hot electron mean free path (MFP), extracted from this device, provides the direct measure ofstrong electronic correlation transition metal oxides.

HETs with a heterostructure of {100} SrTiO3/La0.7Sr0.3MnO3/Nb:SrTiO3 were fabricated, where aferromagnetic metallic La0.7Sr0.3MnO3 was chosen for possible magnetic functionalities. Because the transistorcharacteristics are defined by the emitter current, we must suppress the collector/base leakage current, namely the reverse bias current in the collector/base Schottky junction. In the as-grown device, however, this leakagecurrent (black arrow in Fig. 5b) was as large as the emitter current (grey arrow), prohibiting the transistoroperation. An interface dipole was induced by inserting 1 uc of SrMnO3, to increase the collector/base SBH,successfully suppressing the leakage current by five orders of magnitude (black arrow in Fig. 5c). Fig. 6 showsthe common emitter output characteristics at room temperature, which shows the clear modulation of thecollector current by the base current.

However, the device properties fluctuated due to the stochastic existence of pinholes through the baselayer. These pinholes can be mitigated by using a scanning probe tip as a nanoscale emitter as shown in Fig. 7a.This approach, known as ballistic electron emission microscopy (BEEM), enhanced the reproducibility of thedevice properties, making it possible to achieve hot electron spectroscopy. Fig. 7b shows the collector current(transmission current) as a function of the emitter probe bias, which was increased by decreasing temperature,possibly reflecting the spin scattering in La0.7Sr0.3MnO3.

Conclusion

A systematic study of interface dipole engineering at perovskite oxide heterointerfaces was performed,addressing both fundamental issues and the device applications. Firstly, the conceptual groundwork for theinterface dipole was presented, calculating its absolute magnitude in a general form, invariant to the arbitrarydefinition of the interface region. Secondly, artificial interface dipoles were demonstrated in {100}SrRuO3/Nb:SrTiO3Schottky junctions with (LaO)+ or (AlO2)- ionic charges inserted, as well as in several otherheterointerfaces. The interface dipole was evaluated via four different measurements: IV, CV, IPE, and PES,showing the arbitrary band offset control over a range of up to 1.7 eV. Fundamental issues including the effectsof in-plane inhomogeneity, surface charges, interfacial charged defects, and the interface termination wereexperimentally investigated, elucidating a methodology for the continuous control and the maximization of theinterface dipole. Finally, this technique was applied to the fabrication of HETs, establishing a device platformfor hot electron spectroscopy in strongly correlated electronic systems, highlighting the power of interfacialdipole engineering in oxide electronics, which cannot be fully exploited in other materials systems.

1. J. Junquera, M. H. Cohen, and K. M. Rabe, J. Phys.: Condens. Matter 19, 213203 (2007).

2. S. Kurtin, T. C. McGill, and C. A. Mead, Phys. Lev. Lett. 22, 1433 (1969).

3. S. M. Sze, C. R. Crowell, G. P. Carey, and E. E. LaBate, J. Appl. Phys. 37, 2690 (1966).

Fig. 1 Schematic illustrations of (a) a perovskite structure, ABO3, and (b, c) electrostatic interface dipoles,decreasing and increasing respectively, the electrostatic potential from Material A to B.

Fig. 2 A schematic illustration of an interfacedipole. V denotes the electrostatic potential.

Fig. 3 (a) A schematic illustration of a Schottkyjunction with interface dipoles between the insertedionic charge and the counter screening charge. (b)STEM images of 1 uc (LaO)+-inserted, non-inserted,and 2 uc (AlO2)--inserted SrRuO3/Nb:SrTiO3.

Fig. 4 (a) PES spectra of Ti 2p3/2 core levels in 1 uc(LaO)+-inserted, non-inserted, and 2 uc (AlO2)--inserted SrRuO3/Nb:SrTiO3 Schottky junctions. (b) ASBH plot as a function of inserted layers by fourmeasurements: IV, CV, IPE, and PES.

Fig. 5 (a) A schematic band diagram of aSrTiO3/La0.7Sr0.3MnO3/Nb:SrTiO3 HET. IVcharacteristics of the base/collector (BC) andthe base/emitter (BE) Schottky junctions (b)without and (c) with BC interface dipole.

Fig. 6 Room-temperaturecommon emitter outputcharacteristics of the HET,with the fixed base currentvaried from 0 to 1 mA.

Fig. 7 (a) A schematicillustration of BEEM. (b)Transmission current versusbias from BEEM fordifferent temperatures.

本論文は、多彩な電子物性を示すペロブスカイト酸化物の界面において、界面ダイポールによるバンドオフセット制御技術を確立し、さらにそれを利用して新規デバイス開発を行ったことを述べたものである。本文は英文で記され、全8 章からなる。

第 1 章は、ペロブスカイト酸化物、とくにその界面研究の歴史的経緯を概観したのち、それを踏まえて本研究の目的を述べている。バンドオフセットは界面を利用した電子デバイスの特性をしばしば支配するが、本研究は、一対の界面電荷が作る界面ダイポールによってバンドオフセットを制御する技術について系統的な研究を行うことを目的とした、ペロブスカイト酸化物では初めの試みである。

第 2 章は、界面ダイポールの物理的意味を理論的考察により明らかにしており、さらに他の物質と比較してペロブスカイト酸化物が界面ダイポール制御の容易な物質であることを指摘している。とくに前者に関しては、界面ダイポールをバンドオフセットの変化量という相対的な量として定義することは容易だが、静電学的な絶対量として定義し異なる物質界面間で比較することは自明でなかった。本研究は単位格子ごとに真空準位を定義することで界面が有するダイポールモーメントを絶対量として定式化し、界面ダイポールを考察する上での基本的枠組みを提供している

第 3 章は、本研究で用いた実験技術を薄膜作製、構造解析、電気測定の3 項に分け、一般的原理と本研究における実験において特徴的な点とを説明している。

第 4 章では、ペロブスカイト酸化物からなる金属半導体界面に原子スケールのイオン電荷層を挿入し、金属のスクリーニング電荷との間に界面ダイポールを作製したことを述べている。理想的なショットキー接合の範疇では1.7eV の範囲で、多少理想的な場合から外れた特性も許せば約3eV の範囲で、バンドオフセットを制御した。これらの値は他の物質界面で可能な制御範囲より大きく、ペロブスカイト酸化物界面の特長として構造設計に広く応用できると考えられる。

第 5 章では、前章で述べた方法ではイオン電荷と電子電荷の両方を用いたのに対し、イオン電荷同士の間に界面ダイポールを作製したことを述べている。この場合でもバンドオフセットを有効に変調することに成功し、電荷の種類に関わらず界面ダイポールを形成できることを実証した。

第 6 章では、界面ダイポールを用いた界面制御に関してより詳細な議論を行っている。界面ダイポールの大きさは基本的に挿入電荷量と電荷間距離によって決定されるが、面内不均質性や界面欠陥がある場合にはそれらの構造パラメータにも強く影響される。本章では、界面に面内不均質性を敢えて持たせることで電流値を抑制することなくバンドオフセットを変化させる、あるいは界面欠損量を熱力学的に制御することで界面ダイポールの大きさを最大化するなど、界面構造の精密な制御による高度なバンドオフセット制御を実現したことを述べている。

第 7 章では、トランジスタにおける制御電極への漏れ電流を界面ダイポールによるバンドオフセット制御によって抑制し、ペロブスカイト酸化物を用いたバイポーラ型トランジスタ動作に成功したことを述べている。ペロブスカイト酸化物の3 層構造では初めて、ホットエレクトロンのバリスティック伝導による電流伝達を実現している。このデバイス構造は、任意のペロブスカイト酸化物金属におけるホットエレクトロンの平均自由行程を直接検出できることから、強相関物理の解明に役立つと考えられる。

第 8 章は全体を総括した後に、本研究の成果を踏まえた研究の将来展望を述べている。本研究は、ペロブスカイト酸化物界面の界面ダイポールについて、構造設計の基礎研究、理論的考察、デバイス応用までを多角的かつ総合的に行ったものであり、酸化物エレクトロニクスの一要素技術を確立するものである。このバンドオフセット制御技術は、ペロブスカイト酸化物界面を利用したマルチフェロイックデバイスや低次元電子系の設計に大きな自由度を与えるだけでなく、触媒を初めとする表面デバイスへも応用可能なものと考えられる。

なお、本論文の第4 章はファンハロルド、疋田育之、ベルクリストファ、蓑原誠人、西川満、組頭広志、尾嶋正治、デイビッドミュラー、フィッティングコーコティスレナ、マンディジュリア、第5 章はファンハロルド、疋田育之、蓑原誠人、組頭広志、尾嶋正治、第6 章はファンハロルド、疋田育之、ベルクリストファ、秋山英文、吉田正裕、第7 章はファンハロルド、疋田育之、ベルクリストファ、バナジタマリカ、ラナゴーラブとの共同研究であるが、論文提出者が主体となって実験および解析を行ったもので、本人の寄与が十分であると判断される。