Abstract:

Si-based spintronics will lead to a new progress beyond the conventional Si technology, because spin degrees of freedom provide new and unimaginable functionalities in well-established Si devices and systems. Here, injection and detection of spin-polarized current in actively-controlled channels are indispensable fundamental operations to obtain new functionalities in Si-based spintronic devices such as spin MOSFETs. Group-IV ferromagnetic semiconductors have many advantages to achieve these operations in comparison with commonly-used ferromagnetic metals, because they can be grown on Si platform with atomically flat and abrupt interfaces, they are compatible with Si device processes, and the conductivity mismatch problem does not have to be considered. Thus, group-IV ferromagnetic semiconductors are promising for emerging Si-technology-based spintronic devices.

Ferromagnetic semiconductors have been given simultaneously dual advantages of the semiconductor properties and the magnetic properties. Fundamental properties of "intrinsic" ferromagnetic semiconductors have been investigated in III-V-based ferromagnetic semiconductors Ga1-xMnxAs and In1-xMnxAs. The magnetic ions were substituted for the lattice sites of host semiconductors keeping the single crystal structure of the type of hosts, and as a result, the band structure of ferromagnetic semiconductors reflects that of hosts. Ferromagnetic ordering is induced by the s,p-d exchange interactions, not by the intermetallic precipitates. The s,p-d exchange interactions provide unique properties such as large magneto-optical effects induced by spin-splitting of the band-edge, and controllability of their magnetism by varying the hole density. When a new candidate material is fabricated, investigation of these fundamental characters is indispensable to test whether it is an intrinsic ferromagnetic semiconductor or not.

The first successful growth of group-IV ferromagnetic semiconductor was reported by Park et al. They demonstrated the control of ferromagnetic ordering in gated Ge1-xMnx films by applying an electric field. Although this result implied that Ge1-xMnx had the behavior of intrinsic ferromagnetic semiconductors, the investigation of fundamental properties was not enough. Then controversy arose over the origin of the ferromagnetism in epitaxial Ge1-xMnx films, since some reports suggested that the Ge2Mn nanocolumn structures, the intermetallic Mn5Ge3, and amorphous Ge1-yMny clusters were easily formed in epitaxial Ge1-xMnx films.

In this thesis, Fe was chosen as magnetic dopants in place of commonly used Mn atoms. Fe-doped Ge (Ge1-xFex) films have been grown on Ge(001) and Si(001) substrates by low-temperature molecular beam epitaxy (LT-MBE) and thoroughly investigated their crystal structure and magnetic properties. Furthermore, by studying group-IV magnetic heterostructures, possibility of using Ge1-xFex films in actual spintronic devices is explored.

At first, Ge1-xFex films were epitaxially grown on Ge(001) substrates by low-temperature molecular beam epitaxy (LT-MBE). During growth, the surface morphology of the samples was monitored by in-situ reflection high energy electron diffraction (RHEED) observations. RHEED patterns of the samples showed the diffraction pattern of the diamond-type lattice structure without extra spots caused by the formation of Fe-Ge precipitates. Detailed crystallographic analyses were carried out by transmission electron microscopy (TEM), transmission electron diffraction (TED), energy dispersive X-ray spectroscopy (EDX), and X-ray diffraction spectroscopy (XRD) observations. Epitaxially-grown Ge1-xFex films on Ge(001) substrates maintained the diamond-type lattice structure including the fluctuation of Fe distribution and tiny stacking fault defects without any other ferromagnetic Fe-Ge precipitates, when they were grown at the substrate temperature (TS) of less than 200 °C. The lattice constant (along the growth direction) evaluated from XRD spectra of the samples linearly decreased with increasing the Fe content (x) up to 13.0% and was saturated at x more than 13.0%. This behavior indicates that Fe atoms were substituted for the lattice sites of host Ge up to at least x = 13.0%. Magneto-optical measurements were carried out by magnetic circular dichroism (MCD) technique. The MCD spectra of Ge1-xFex films reflected the band structure of host Ge and the MCD peak at the critical point (especially E1 transition energy) was largely enhanced by Fe doping, although broad offset-like MCD signals were observed. This indicates that the band-edge spin-splitting was induced by the s,p-d exchange interactions as described above. Magnetic field dependence of MCD intensity at any photon energies including E1 and other points exhibited clear ferromagnetic hysteresis loops, and these shapes were identical with one another, indicating that the MCD spectral features and ferromagnetic ordering of the samples came from a magnetically-homogeneous single ferromagnetic phase, without any other ferromagnetic Fe-Ge precipitates. The Curie temperature (TC) of the Ge1-xFex films on Ge(001) evaluated from the temperature dependence of MCD hysteresis loops linearly increased with increasing x up to 13.0%, and was saturated at x more than 13.0%. The behavior of TC had a good correlation with the behavior of lattice constant as a function of x. This correlation indicates that TC increases in proportion to the number of substitutional Fe atoms, and that the crystal structure and the magnetic properties of Ge1-xFex films on Ge(001) have linear relation with each other, which is one of the essentials for intrinsic ferromagnetic semiconductors. All the results presented here confirm that epitaxial Ge1-xFex films on Ge(001) are "intrinsic" ferromagnetic semiconductor.

Next, Ge1-xFex films were epitaxially grown at TS = 200 °C on Si(001) substrates by LT-MBE. To establish the epitaxial growth of Ge1-xFex films on Si is the next important step for realizing group-IV-based spintronic devices integrated into Si CMOS platform. In-situ RHEED during growth showed only the diffraction pattern of the diamond-type lattice structure. Crystallographic analyses revealed that the crystal quality of the Ge1-xFex layer was deteriorated including not only the fluctuation of Fe distribution and tiny stacking fault defects as well as the films on Ge(001) but also threading dislocations induced by large lattice mismatch between Ge1-xFex and Si, although there were no formation of intermetallic Fe-Ge precipitates within Ge1-xFex layer. The lattice constant (along the growth direction) evaluated from XRD spectra linearly decreased with increasing x up to 13.0% and was saturated at x more than 13.0%. MCD measurements indicate a largely enhancement of MCD peak intensity at E1, a magnetically-homogeneous single origin of their ferromagnetism, linear increase of TC depending on x, and good correlation between TC and the lattice constant. All the results of Ge1-xFex films on Si(001) were consistent with the case of Ge1-xFex films on Ge(001), confirming that epitaxial Ge1-xFex films on Si(001) are also intrinsic ferromagnetic semiconductor. Furthermore, magneto-transport properties were examined in the Ge1-xFex films on Si(001). Anomalous Hall effects (AHE) with ferromagnetic hysteresis loops was clearly observed at the temperature lower than TC evaluated from MCD, and the shapes of AHE was consistent with those of MCD. This means that anomalous Hall effect and MCD measurements detected the same ferromagnetic phase, that is, the intrinsic ferromagnetic semiconductor phase.

Finally, the observation of tunneling magneto-resistance (TMR) effect was attempted in tri-layer magnetic tunnel junction (MTJ) structures with ferromagnetic Ge1-xFex films as ferromagnetic electrode. TMR-like behaviors were observed at low temperature (~3K) in the tri-layer structure of top Fe / Si0.2Ge0.8 barrier / bottom Ge1-xFex (x = 13.0%), although TMR changed negatively and these TMR ratio were very low (~ 0.3%). This behavior indicated that Ge1-xFex films can actually function as a spin injector and detector in Si-technology-based spintronic devices.

本論文は、「A New Group-IV Ferromagnetic Semiconductor Ge1-xFex: Epitaxial Growth, Crystal Structure, Magnetic Properties, and Heterostructures(新しいIV族強磁性半導体Ge1-xFex薄膜の研究:エピタキシャル成長、結晶構造、磁性、およびヘテロ構造)」と題し、英文で書かれている。本論文は、著者の研究によって創成された新しいIV族強磁性半導体Ge1-xFex薄膜のエピタキシャル成長、結晶構造および結晶性、磁性、およびヘテロ構造の研究成果を記述しており、全5章から成る。

第1章は「Introduction」であり、スピントロニクスと強磁性半導体に関する研究の背景と状況を述べ、本論文の構成と目的を示している。

第2章は「Epitaxial growth and characterizations of Fe doped Ge (Ge1-xFex) thin films on Ge(001) substrates」であり、Ge(001)基板上へのFeドープGe (Ge1-xFex)薄膜の分子線エピタキシー(MBE)法によるエピタキシャル成長、X線回折および高分解能透過型電子顕微鏡(TEM)のよる微視的な構造評価、磁気円二色性(MCD)によるバンド構造とスピン分裂の評価、および磁化特性評価を行い、構造と物性を実験的に明らかにした結果を記している。これらの評価の結果、Ge1-xFex薄膜中のFe濃度分布に不均一性はあるものの、結晶全体がダイヤモンド構造を保っており、バンド構造もダイヤモンド型半導体の特徴を有していること、明瞭な強磁性を示し、強磁性転移温度(TC)はFe組成x=0.13まではxに比例して増大し最高で170 K程度であること、強磁性の起源は単一であり、強磁性金属の析出物など第2相によるものではないこと、などを明らかにした。以上により、Ge1-xFexは真性の強磁性半導体であることを示した。

第3章は「Epitaxial growth and characterization of Fe doped Ge (Ge1-xFex) thin films on Si(001) substrates」であり、MBEを用いてSi(001)基板上にGe1-xFex薄膜を成長し、詳細な構造評価と磁気光学効果測定を行い、さらに磁気輸送特性を調べた結果を記している。Si(001)基板上に成長したGe1-xFex薄膜は基板との格子不整による転位等の欠陥は存在するものの、結晶構造、バンド構造ともにダイヤモンド型半導体のそれであり、明瞭な強磁性を示し、強磁性転移温度(TC)はFe組成x=0.13まではxに比例して増大し最高で130 K程度であること、MCDと異常ホール効果の磁場依存性は完全に一致し強磁性の起源は単一であること、などを明らかにした。以上より、Si基板上に成長したGe1-xFexも真性の強磁性半導体であることを示した。

第4章は「Device Applications」であり、スピントロニクスの基本素子である強磁性トンネル接合をGe基板上に作製し、そのスピン依存トンネル伝導について調べた結果を記している。Fe/SiGe/Ge1-xFex から成る強磁性ヘテロ構造によってトンネル接合を形成し、トンネル磁気抵抗効果と思われる磁気抵抗変化を観測した。これにより、Ge1-xFex薄膜がスピン注入源およびスピン検出器として働く可能性を示唆した。

第5章は「Concluding remarks」であり、本論文で得られた結果のまとめと今後の展望を述べている。

以上これを要するに、本論文は、新物質であるGe1-xFex薄膜をエピタキシャル成長によってGe基板上およびSi基板上に形成し、Feの平均組成が10%程度以下ではその結晶構造とバンド構造がダイヤモンド型半導体の特徴を保つこと、強磁性を示すとともにその起源が単一の磁性相からなること、すなわちGe1-xFex薄膜は真性の強磁性半導体であることを示したものであり、電子工学上、寄与するところが少なくない。

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。