1.Introduction.

Semiconducting silicides are extensively studied in the recent years because of their promising properties as Si-based electronic devices [1]. Among them, beta-iron disilicide, β-FeSi2, formed on a Si substrate, is known to exhibit photoluminescence (PL) peak at around 0.8 eV. However, substrate Si also has a luminescence peak at 0.80 eV, which nearly coincides with that of β-FeSi2, so that the origin of PL peak from β-FeSi2 / Si is always point of issue. On the other hand, the ion beam sputter deposition (IBSD) method was successfully applied to fabrication of β-FeSi2 film on Si substrate [2]. It is interest to see how the ion-induced processes in IBSD affect the PL properties. In this study, the fabrication process and PL properties of IBSD-grown β-FeSi2 were investigated with reference to ion-induced effects.

2. Experimental.

The β-FeSi2 films were grown by the IBSD method. In Fig.1 the schematic diagram of IBSD device is shown. The typical deposition procedures are as follows: (1) ultrasonic cleaning of the Si samples in acetone, ethylene and water, before installing in a ultra-high vacuum (10-7 Pa) vessel, (2) sputter etching (SE) of the sample by Ne+ ion beam (3 keV, 5 μA, 20 min), annealing the sample for 1 hour at 973 K, and (3) sputter deposition of Fe target by Ar+ ion beam (35 keV, 180 μA, 60 min). In photoluminescence (PL) measurement, the samples were exited by 532 nm solid-state laser with an output power of 50-100 mW, and PL spectra were obtained in the temperature range of 10-300 K. Luminescence was analyzed in the wavelength region of 1100-1700 nm (0.73-1.12 eV). The scheme of β-FeSi2 film growth process is shown in Fig. 2.

3. Photoluminescence (PL) characterization of β-FeSi2 prepared by ion beam sputter deposition (IBSD) method.

The various stages of IBSD method to prepare β-FeSi2 film on Si substrate were characterized by PL measurement. It has been found that sputter etching of the substrate surface is essential to obtain highly oriented β-FeSi2 film [3]. Fig. 3 shows the PL spectra of Si substrate, SE-treated Si and β-FeSi2 film prepared on SE-treated Si substrate for CZ samples. All the samples shown in the figure were annealed in air at 1153 K for 20 hours and the spectra were measured at 34 K.

It is clearly shown that the annealing drastically increased the PL intensity at around 0.81 eV. It should be noted that it was the samples that underwent Ne+ sputter etching that showed most drastic increase of PL peak upon thermal annealing in air. For CZ sample, the largest intensity of 0.81 eV peak was obtained for SE-treated Si, followed by β-FeSi2 formed on Si. Comparison between Si substrate (a) and SE-treated Si (b) indicates that annealing alone did not lead to such PL enhancement. It is thus clear that SE and thermal annealing cooperatively enhanced the observed PL emission at 0.81 eV.

A previous study indicated that after high temperature annealing (above 1073 K) the morphology of β-FeSi2 film changed to island structure [4], which fact indicates that some portion of the PL signal originated from the bulk of Si. Therefore, the PL signal from the annealed β-FeSi2 formed on Si substrate may include contributions from both silicide and Si. From comparison of spectra b and c in Fig. 3, it is considered that contribution of Si substrate to the observed PL enchantment cannot be ignored.

Fig. 4 shows temperature dependence of the PL peak energy of the CZ samples used in this study, together with the data reported by Maeda et al. [5] and Binetti et al. [6].

For Si substrate the temperature dependence of PL peak at around 0.8 eV showed a complicated dependence on temperature. But in the cases where SE was applied to substrate, the PL characteristics changed drastically and the position of the peak shifted to lower energy. Interestingly, the temperature dependence for β-FeSi2 / Si, SE-treated Si and data for bands of D1 line reported by Binetti et al. [6] appeared to be similar, and the peak energy was located at higher energy than those reported in ref. [5], which is considered as an intrinsic peak of β-FeSi2. On the other hand, Fig. 5 compares the temperature dependence of the PL intensity of the samples shown in Fig. 4. The temperature dependence for β-FeSi2 / Si was similar to that reported in the literature [5], while it appeared to quench at slightly lower temperature than SE-treated Si.

The quenching processes shown in Fig. 5 were analyzed, which takes into account two non-radiative processes as shown by the following equation:

I (T) = I0 / [1 + C1 exp ~(E1 / kT) + C2 exp ~(E2 / kT)],

where C1,2 are the constants for first and second non-radiative processes, E1,2 the activation energies for these processes, I0 the intensity at T = 0 K and k the Boltzmann constant. Using this equation the activation energies; E1 = 4 meV and E2 = 70 meV, and the dimensionless constants; C1 = 2.4 and C2 = 9.8×103, which correspond to the number density of non-radiative centers, were obtained for the β-FeSi2 / Si sample after 20 hours of annealing (curve c), whereas E1 = 9 meV, C1 = 2 and E2 = 116 meV, C2 = 4×104 were obtained for the annealed Si (SE-treated) (curve b). However, the fact the observed peak energy in this study was higher than the literature data indicates that it is unlikely to originate from the intrinsic band of β-FeSi2.

4. In search of intrinsic photoluminescence from β-FeSi2.

4.1 PL from β-FeSi2 formed on SOI substrates.

A substrate with SOI structure is widely applied in semiconductor industry because of its benefit for low power consumption. As a substrate, so-called SIMOX (Silicon IMplanted by OXigen) was employed, where a thin (approximately 100nm) overlayer is situated on insulating BOX (Buried OXide) layer of 100-150 nm in thickness. These samples were thermally annealed in a low vacuum (10(-4) Pa) furnace at 1123 K for 24 h, and then PL measurements are carried out. The results are illustrated in Fig. 6.

The effect of the thermal annealing on the sample without the template (fabricated by deposition of Fe with the approximate thickness of 50 nm) is to quench the 0.82 eV peak. For this sample, the XPS (X-ray photoelectron spectroscopy) analysis and the TEM (transmission electron microscopy) observation showed that the β-FeSi2 film was considerably aggregated into smaller particles with the size of 20-30 nm. In addition, the interface of the β-FeSi2 and Si is disappeared where the particles were directly placed on the buried oxide (BOX) layer. On the other hand, it is seen that the film fabricated with the template shows the enhancement of the PL peak intensity at 0.82 eV after annealing. Whereas for the sample without the template, aggregation due to annealing produces the small particles directly placed on the BOX layer. From the temperature dependence of the photoluminescence peak energy and intensity it can be concluded that the PL origin can be same as for the SE Si case from the previous section.

It is deduced that employing template method in fabrication of β-FeSi2 films on a SIMOX substrate shows the favorable advantage to enhance the PL intensity.

4.2 PL from β-FeSi2 formed on bulk crystal.

It is still unclear from the previous section whether the observed PL signal originated from the β-FeSi2 or from some other sources. In this section, the PL signal from β-FeSi2 film, which was grown by molecular beam epitaxy (MBE) on β-FeSi2 substrate. The β-FeSi2 substrate was an ingot grown from Ga solvent. The surface of substrates was polished then wet-etched to make it smooth. The PL signal from these samples was measured using the same equipment described before. The PL intensity was found to be very weak, below the detection limit of the PL measurements. One of possibilities of such weak signal can be attributed to some impurities in this substrate and/or film during preparation process.

5. The dependence of PL characteristics on SE energy to Si substrate.

In this section, how the irradiation effect on Si substrate affects the PL properties of β-FeSi2 film on it. In Fig. 7 the PL spectra of annealed β-FeSi2 samples fabricated with different incident SE energies are compared.

The PL intensity of the annealed β-FeSi2 on Si substrate increased as the SE energy applied to the substrate increased, thereby indicating the contribution of irradiation defects to the observed PL spectra. Evidence of "critical" sputter etching energy located near of 3 keV sputter etching energy value was found. If the energy of SE ion beam was higher than this "critical" energy the impact of Si related PL in resulting signal is highest. Therefore, Fig. 7 may indicate that SE can be employed to control the PL properties of β-FeSi2 film on Si substrate.

Conclusions.

1.A strong photoluminescence peak at around 0.8 eV was observed for β-FeSi2 films deposited on Si substrates that were sputter etched by Ne+, and then thermally annealed in air at elevated temperature. It was revealed that the PL peak at 0.8 eV observed in this study was mainly from D1 emission bands in Si substrate.

2.The β-FeSi2 films were successfully grown on a thin (100 nm) Si over-layer of SOI substrate by IBSD method, and the β-FeSi2 was grown on β-FeSi2 substrate, in order to exclude the contribution of Si substrate to the PL characteristics. It was found that the PL intensity was at least an order of magnitude small than the case when Si substrate was employed.

3.The PL intensity of the annealed β-FeSi2 on Si substrate increased as the SE energy applied to the substrate increased, thereby indicating the contribution of irradiation defects to the observed PL spectra. Evidence of "critical" sputter etching energy located near of 3 keV sputter etching energy value was found. If the energy of SE ion beam was higher than this "critical" energy the impact of Si related PL in resulting signal is highest.

Fig. 1: Schematic diagram of IBSD device.

Fig. 2: Typical scheme of β-FeSi2 film fabrication and characterization processes.

Fig. 3: PL spectra of (a) Si substrate, (b) SE-treated Si and (c) β-FeSi2 film prepared on SE-treated Si substrate for CZ samples

Fig. 4: Temperature dependence of the PL peak energy of the samples using CZ Si as a substrate; (a) CZ Si, (b) SE-treated CZ Si and (c) β-FeSi2 film prepared on SE-treated CZ Si, together with literature data: (x) Maeda et al. [5] and (y1, y2, y3) Binetti et al. [6]. The samples (a), (b) and (c) were annealed in air for 20 hours.

Fig. 5: Temperature dependence of the PL intensity of the samples shown in Fig. 4

Fig. 6: The PL spectra of annealed β-FeSi2 sample fabricated without template, with template + 25 nm FeSi2 deposition and with template + 50 nm FeSi2 deposition.

Fig. 7: The PL spectra of annealed samples fabricated with different spatter etching beam energy.

鉄シリサイド半導体(β-FeSi2)は、資源量が比較的豊富で人体への毒性が小さい鉄(Fe)とシリコン(Si)を構成元素とし、また、Si基板への薄膜成長が容易であるため、環境に優しい次世代の半導体材料として期待されている。鉄シリサイドに関しては、近赤外領域に発光ピークを有するものの発光強度が小さいこと、また、ほぼ同じ波長域にSi中の欠陥に起因する発光があるため、Si基板上に作製された鉄シリサイドの場合Si由来の発光との識別が困難であること、などが報告されている。本研究では、イオンビームを用いた手法により、Siをはじめとする種々の基板に対して鉄シリサイド薄膜を作製し、その発光特性を調べることにより、発光の起源の解明や発光強度の改善に資する知見を得ることを目的とした。

本論文は6章で構成されている。

第1章は研究の背景と目的が書かれている。

第2章は実験方法について説明している。本研究では、イオンビームスパッタ蒸着法により固体ターゲットからスパッタされた原子を加熱した基板と反応させ、厚さ 100 nm 程度の鉄シリサイド薄膜の作製を行った。成膜直後の発光強度は微弱であるため、さらに高温でのアニールを行った。実験では、Siのほか、SOI (Si層中に絶縁層を挿入した基板) や、鉄シリサイドのバルク単結晶を研磨して基板状にしたものも基板として用いた。

第3章はSi基板に作製した鉄シリサイド試料からの発光に対するSi基板の影響について記述している。成膜工程の随所で発光測定を行った結果、発光強度は成膜後に行うアニールにより飛躍的に増大することが明らかにされた。Si基板に対しては、蒸着前に表面に存在する自然酸化物層を除去するために数keVのイオンによるスパッタ洗浄を行っているが、この処理を行った試料からの発光強度が特に高いことが明らかにされた。しかし、全く同じ事象が成膜を行わないSi基の場合にも観測され、さらに、発光ピークの位置やピーク強度の温度依存性を、鉄シリサイド固有の発光とされる文献データと比較したところ、本研究における鉄シリサイド試料からの発光は、Si基板に由来する可能性が高いことが示された。

第4章は鉄シリサイドに固有の発光を追究する試みについて記述している。そのために、SOI基板において絶縁層により分離されたSi層の薄い方 (約 100 nm) に成膜し、Siの影響を極力抑えるとともに、鉄シリサイドのバルク結晶から作製した試料を研磨して基板にすることによりSiと共存しない鉄シリサイド試料を得た。しかし、前者の場合、発光強度はSi基板に成膜した場合に比べて約1桁低く、後者に至っては、現状では十分な大きさのバルク結晶が得られないこともあり、有意な発光ピークを観測するには至っていない。なお、SOI基板への成膜の場合、試料の元素組成分析から鉄シリサイドは過剰なSiと共存していると推定され、発光ピークの温度依存性の結果とも併せて、この場合もSiからの発光の寄与が無視できないことが示されている。

第5章は蒸着前の基板前処理条件が発光特性に与える影響について記述している。スパッタ洗浄されたSi基板に鉄シリサイド薄膜を作製する点では第3章での実験と同様であるが、本章では基板処理条件を変えて成膜を行っている。具体的には、スパッタ処理を行う入射イオンのエネルギーを変え、そのことが鉄シリサイド試料および基板単独からの発光特性にどのような影響を与えるかを調べている。実験の結果、鉄シリサイド試料では入射エネルギーに関係なくほぼ同程度のピーク強度を呈する一方で、基板からの発光に大きなエネルギー依存性があるという興味深い結果が示されている。すなわち、あるしきいエネルギー以下では基板からの発光がほとんどなく、この条件下では鉄シリサイド試料からの発光はほぼそれ自身によるものと結論できる。さらに、この条件で得られた発光ピークの挙動を調べたところ、これを鉄シリサイドに固有のピークと結論するまでには至っていないとはいえ、今後の研究の進むべき方向を提示した点では有意義な成果といえる。

第6章は前出の各章を総括して結論を述べている。

以上、本研究は、イオンビームスパッタ蒸着法で作製された鉄シリサイド試料の発光特性に関する唯一の研究であるだけでなく、イオンビームによる照射がその発光特性にどのような影響を及ぼすかを調べた上で、発光特性の制御を試みその改善に向けて有用な知見も提示しており、システム量子工学、特に量子ビームを利用した新材料創製分野への貢献は小さくない。

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。