Two-dimensional electron gas (2DEG) provides a fascinating playground for low-dimensional physics. Representatives of 2DEG are fabricated at heterostructures of semiconductors, the metal-oxide semiconductors, and thin metal films. Another candidate for producing 2DEG was surfaces of metals or semiconductors. The surface state is created at a few layers of the top of materials, i.e. the interface between the material and vacuum, so that it is intrinsically two-dimensional. Moreover, the surface state is so sensitive with extrinsic effects, for example, localization or scattering centers given by adsorption of foreign atoms. Therefore, the surface 2DEG has attracted wide attention of researchers of condensed matter physics and been intensively studied by various techniques, such as scanning tunneling spectroscopy and photoemission spectroscopy.

However, study of the surface state by transport measurements has been still developing because of technical difficulty such as leakage of current through the bulk state or fabricating good contacts to the state in ultra high vacuum (UHV). However, these problems have been addressed by applying a micro-four-point probe method with a narrow probe spacing of ~ 20μm on a sample surface, fabricated on a substrate of semiconductor such as silicon. This method enables to electrically decouple the surface state with the bulk state.

By using the micro-four-point probe method, transport properties of various superstructures have been investigated. Among them, Si(111) √7×√3-In-rec surface, which was formed by depositing In atoms of 1.2 monatomic layer (ML) thick on a Si(111) substrate, was found to show a metallic character in transport from room temperature to ~ 10 K without any phase transitions. In addition, this surface was reported to have a metallic and parabolic dispersing band by photoemission spectroscopy. Therefore, this surface provides a good system to study the interaction of adatoms with the surface state.

In the present thesis, I focused on studying the interaction between magnetic adatoms and the surface two-dimensional electron state provided by Si(111)√7×√3-In-rec surface as a prototype of surface dilute magnetic systems. Since the screening in two-dimensional electronic systems is generally weaker than that in three dimensional systems, such interactions may be more significant in the surface state. Concretely, by using the micro-four-point probe method, I conducted surface-sensitive and temperature-dependent resistivity measurements for the Si(111)√7×√3-In-rec surface with various concentrations of Co adatoms in situ in UHV. From applying the appropriate theories for the observed temperature dependences of resistivity, magnetic effects in the presence of disorder were found to play a crucial role. Moreover, the observed behavior has a possibility of indicating a change of the magnetic state of the sample surfaces. In the following, I describe the details.

In the dilute concentration regime of the Co coverage from ~ 0.00025 ML to ~ 0.00125 ML, the resistivity deviation from the pristine Si(111)√7×√3-In-rec surface was first observed at temperatures lower than 100 K for all the sample surfaces, while the pristine Si(111) √7×√3-In-rec surface remained metallic from room temperature to ~ 10 K with high residual resistivity ratio, indicating a high defect density. At the lowest concentration of Co adatoms of ~ 0.00025 ML, the deviation in resistivity monotonically increased down to the lowest temperature, as shown in Fig. 1(a). At the slightly higher Co coverage, the deviation in resistivity occurred and increased along with the decrease of temperature, exhibiting a resistivity maximum at a certain temperature T(ρmax), followed by the decrease in resistivity again by further cooling. In addition, T(ρmax) was linearly proportional to the Co coverage shown in Fig. 1 (b), which indicates that the observed temperature dependence of resistivity was associated with spins induced by Co adatoms.

To clarify the origin of the observed phenomena, I focused on the interaction of conduction electrons and magnetic impurities in the presence of disorder, such as weak localization, electron-electron interaction, and s-d interaction (Kondo effect, RKKY interaction, and competition between them). The calculated resistivity by quantum corrections of weak localization and electron-electron interaction in the presence of magnetic impurities were too small to explain the magnitude of the observed resistivity deviation. In addition, the calculation did not reproduce the observed character of Co coverage dependence of T(ρmax). However, a competition between Kondo effect and RKKY interaction can explain the observed phenomena, indicating that it was a dominant mechanism. From fitting for the respective samples, I deduced Kondo temperature TK of the Co/Si(111) √7×√3-In-rec surface as ~ 4 K which was intrinsically independent on Co coverages, and typical RKKY temperature Tsg (corresponding typical RKKY interaction energy) which depended on the Co coverage. The values indicated that the ground state of the sample with Co coverage of 0.00025 ML preferred Kondo state, in which spins are isolated with negligible mutual interaction, while at higher coverages, RKKY interaction between the spins could not be ignored. Moreover, by considering the enhancement factor of coefficient of Kondo logarithmic resistivity by disorder, I found the quantitative agreement between the calculation and the observed deviation of resistivity for the sample with the lowest Co coverage, by using a parameter of the deduced TK only. With increase of Co coverage, however, the calculations showed larger deviation from the observations in the magnitude. It means that it was necessary for the correction of enhancement factor, including the effect of RKKY interaction in the presence of disorder.

Furthermore, it should be noted that the magnetic states of Co/Si(111) √7×√3-In-rec surface in the dilute range of Co adatoms. It was reported that at TK > Tsg the ground state was Kondo state, while at TK < Tsg a spin-glass state possibly overcame the Kondo state due to the enhanced RKKY interaction. According to this assumption, the magnetic state experienced a transition from the Kondo state to a spin-glass state by increasing the Co coverage.

In the dense regime of Co coverage from ~0.0125 ML to ~ 0.58 ML, the transport results were different from those in the dilute regime. For the sample of Si(111)√7×√3-In-rec surface with Co adatoms of 0.58 ML, the temperature dependence of resistivity showed a semiconducting character, which was often observed for granular metal. However, for the samples with Co coverage of ~ 0.0125 ML to ~ 0.2 ML, the resistivity first showed a decrease by cooling, indicating a metallic character. Then, it passed through the resistivity minimum, and then turned to increase by further cooling. All of them shows a resistivity higher than that of the pristine Si(111)√7×√3-In-rec or the dilute Co/Si(111) √7×√3-In-rec surface at room temperature, indicating an enhanced disorder. In addition, the temperature dependence of resistivity showed a change in the slope, leading to the change of effective mass or Fermi wave vector of the samples. From extrapolation of resistivity at higher temperature, I estimated the resistivity raised by spins of Co adatoms for the respective samples. All of them were larger than the resistivity values expected by weak localization and electron-electron interaction, and fitted well with a logarithmic function of temperature. A two-level model of tunneling electrons, which was a significant mechanism for strongly disordered systems, was found to be inappropriate to explain the phenomena. In addition, the Si(111) √7×√3-In-rec surface with adsorption of Ag of ~ 0.05 ML, which showed almost the same resistivity with Si(111) √7×√3-In-rec surface with Co of ~ 0.03 ML at room temperature, did not exhibit the logarithmic rise in resistivity at lower temperature. This supports that the observed deviated resistivity has a magnetic origin.

The conduction mechanism of the present samples in the dense Co regime was considered to be the same as that of Kondo-like resistivity, which was observed in strongly disordered bulk systems with denser magnetic concentrations than in ordinal dilute magnetic alloys. The normalized resistivity by the resistivity minimum and the temperature at the resistivity minimum showed a characteristic dependence of Co coverages. This possibly indicates that the magnetic state showed a transition from spin-glass to inhomogeneous ferromagnetism.

As described as far, in the present thesis, I have carried out temperature dependence of Co/Si(111) √7×√3-In-rec with various Co coverages. From analysis by using available theories, I found that s-d interaction in the presence of disorder was essential to explain the observation in the regime of dilute concentration of Co adatoms. In the dense regime, moreover, I found that magnetic interaction in the presence of strong disorder was also important to expalin the observation. Finally, I suggest that I have detected three phase of Co adatoms on a metallic surface 2DEG of Si(111) √7×√3-In-rec by transport measurements with changing the adatom concentration; a dilute phase in which Kondo effect is dominated, a medium-concentration phase in which RKKY interaction occurs and a spin-glass state is possibly formed, and a dense phase where an inhomogeneous ferromagnetism begins to appear with stronger spin-spin interaction. These findings provide an important example of fundamental physics of dilute magnetism on surfaces and also can have possible applications to spintronics.

Fig. 1 (a) Temperature dependence of the measured sheet resistivity for Si(111) √7×√3-In-rec surface with various Co coverages. The line on the sample is the guide to the eye. (b) The temperature at the resistivity maximum versus Co coverages.

本論文は、磁性原子を不純物として含む金属超薄膜の表面電気伝導度をマイクロ4端子プローブ法によって測定し、その温度依存性、濃度依存性を解析して2次元電子系の電気伝導の特徴を明らかにした研究で,全5章からなる。第1章は序論で、本研究の背景として,磁性不純物を含む金属の電気伝導の特徴とその起源、および Si(111)表面上に作成した金属超薄膜の電気伝導の特徴とこれまでの研究について述べている。また、本研究で金属的伝導を示す2次元電子系としてSi(111)√7x√3-In rec表面を選択した理由を述べ、そこに磁性原子(Co)を吸着させて電気伝導の機構を明らかにするという、研究目的を述べている。第2章では、この研究でおこなった実験の詳細を記述している。低速電子線回折(LEED)、反射高速電子線回折(RHEED)、走査トンネル顕微鏡(STM)による表面原子構造解析法について述べるとともに,マイクロ4端子プローブ法による表面電気伝導測定法とそのデータ解析によって得られる知見についてまとめられている。

第3章は、磁性原子が希薄な場合の表面電気伝導度測定とその解析結果の詳しい記述である。実験では,まずSi(111)√7x√3-In recを作成してその表面原子構造が√7x√3構造を示すことをRHEEDおよびSTMで確認した後,マイクロ4端子プローブ法によって電気伝導度の温度依存性を測定してSi(111)√7x√3-In recの電気抵抗が温度の低下とともに減少する金属的な振る舞いを示すことを明らかにした。また、バルク金属表面に較べて原子構造が不規則であるため残留抵抗が大きくなることも解った。次に,Co原子濃度の異なるいくつかの表面について電気伝導度の温度依存性を測定した結果,温度の低下とともに電気抵抗が50 Kー100 Kで極小を示すことを観測した。また、さらに低温にすると最もCo濃度(0,00025ML)の低いもの以外は、電気抵抗が 20 K-50 Kで極大値(ρM)を示して減少し、極大値を示す温度がCo濃度とともに増加する結果を得た。

磁性不純物を含む金属では、伝導電子と磁性不純物の局在スピンとの相互作用のために電気抵抗が低温で極小値を示すこと(近藤効果)、磁性不純物濃度が増加すると伝導電子を介して局在スピン間に働く相互作用(RKKY相互作用)のためにさらに低温で電気抵抗が極大値を示すことなどが知られている。本研究で用いたCoを含むSi(111)√7x√3-In rec表面の不規則性を考慮し,電気抵抗の温度依存性を、近藤効果とRKKY相互作用とが競合していることによると仮定して解析したところ,定性的に実験結果を説明できることがわかった。

第4章では、磁性不純物であるCo濃度が大きい(0.00125-0.58ML)領域の表面電気伝導度の実験とその解析結果について詳しく記述されている。この領域では,全ての試料で室温での電気抵抗が Coを含まないSi(111)√7x√3-In recに較べて大きい電気抵抗を示した。また、電気抵抗は温度の低下とともに低下して極小を示し,さらに低温に向かって温度の対数に比例して増加する温度依存性を示した。以上の実験結果は,磁性不純物濃度の大きい場合、バルクの近藤効果と同じ機構によるとして理解することができた。

第5章は、研究結果のまとめについて記述している。

以上のように、本論文はSi(111)√7x√3-In rec表面に磁性不純物としてCoを含む2次元金属超薄膜の作成し,電気伝導の特徴を表面敏感な様々な実験手法を用いて研究し、その温度依存性、濃度依存性を解析することによって2次元電子系の電気伝導における局在スピンの役割について新しい知見を与えたものである。

なお、本論文の第3および4章の一部は長谷川修司らとの共同研究であるが、論文提出者が主体となって実験し、結果の解析、検証を行ったもので、本論文が示す研究成果に関して論文提出者の寄与が十分であると判断する。

従って、博士(理学)の学位を授与できると認める。