Low-Temperature Fabrication of Polycrystalline Si Thin Film Using Al

0 downloads 0 Views 318KB Size Report
Using Al-Induced Crystallization without Native Al Oxide at Amorphous Si/Al ... The a-Si/Al/SiO2/Si structure was then heated at a low temperature of 400 C to ...
Japanese Journal of Applied Physics Vol. 44, No. 7A, 2005, pp. 4770–4775 #2005 The Japan Society of Applied Physics

Low-Temperature Fabrication of Polycrystalline Si Thin Film Using Al-Induced Crystallization without Native Al Oxide at Amorphous Si/Al Interface Youhei SUGIMOTO, Naoki TAKATA, Takeshi H IROTA, Ken-ichi I KEDA, Fuyuki Y OSHIDA, Hideharu N AKASHIMA and Hiroshi NAKASHIMA1  Interdisciplinary Graduate School of Engineering Sciences, Kyushu University, 6-1 Kasuga-Koen, Kasuga, Fukuoka 816-8580, Japan 1 Art, Science and Technology Center for Cooperative Research, Kyushu University, 6-1 Kasuga-Koen, Kasuga, Fukuoka 816-8580, Japan (Received October 6, 2004; revised February 10, 2005; accepted March 6, 2005; published July 8, 2005)

Low-temperature fabrication of polycrystalline silicon (poly-Si) thin film has been performed by Al-induced crystallization (AIC), and the structural properties have been investigated. In our experiments, to prevent native oxidation of Al film, an amorphous silicon (a-Si)/Al bilayer was formed on the SiO2 /Si substrate by electron beam evaporation without breaking the vacuum. The a-Si/Al/SiO2 /Si structure was then heated at a low temperature of 400 C to induce AIC. It was confirmed that layer exchange of the a-Si/Al bilayer is induced even though there is no native oxidation of Al film, which was demonstrated by scanning transmission electron microscopy and energy dispersive X-ray analysis. The mechanism for layer exchange of the a-Si/Al bilayer has been discussed. Furthermore, it was verified by scanning electron microscopy and spectroscopic ellipsometry that the a-Si/Al thickness ratio of roughly 1 : 1 is suitable to achieve a flat surface morphology of poly-Si. In addition, it was found, by X-ray diffraction and orientation imaging microscopy, that the Si(111)-oriented grain becomes dominant with decreasing thickness of the a-Si/Al bilayer. [DOI: 10.1143/JJAP.44.4770] KEYWORDS: poly-Si, low-temperature process, aluminum-induced crystallization, solid-phase crystallization, structural properties

1.

Introduction

Si thin films on glass substrate are widely applied to a variety of electronic devices. Recently, low-temperature polycrystalline silicon (poly-Si) films on inexpensive substrates (non-alkali glass) have been expected to be applied in practice owing to the good device performance and low fabrication cost. The materials are generally fabricated at low temperatures below 600 C. Among the several approaches to poly-Si film fabrication, solid-phase crystallization (SPC) is conventionally adopted to crystallize amorphous silicon (a-Si) films. However, this technique has a trade-off relationship between crystallization temperature and time at temperatures below 600 C.1) Therefore, as alternative approaches to low-temperature crystallization of a-Si films, laser annealing crystallization (LAC) and metalinduced crystallization (MIC) have been intensively studied as promising candidates to replace the standard SPC.2,3) MIC using a metal/a-Si bilayer structure is effective for decreasing the activation energy of crystallization, and can easily induce transformation from a-Si to poly-Si below 600 C. Among the MIC techniques, Al-induced crystallization (AIC), in which Al thin film is used as a catalyst, is known to cause the transformation within 1.5 h even at a low temperature of 500 C.4–6) The basic process is a-Si/Al/ substrate annealing below the eutectic temperature (577 C) on the Al–Si binary system. Performing this technique, Nast et al.4) observed a new phenomenon of layer exchange of the a-Si/Al bilayer induced on the glass substrate, resulting in the direct fabrication of poly-Si film on the substrate. The developed AIC technique is called Al-induced layer exchange (ALILE).5) Nast and Hartmann noted that native Al oxide inserted into the a-Si/Al interface plays an essential role in the final crystal quality of poly-Si film.7) The AIC technique with a native Al oxide layer, i.e., the deposition of an a-Si/Al bilayer with breaking the vacuum, has recently been focused on and studied under various conditions in 

E-mail address: [email protected]

order to form poly-Si having a large grain size at a low temperature.4–10) If ALILE can be achieved for the a-Si/Al bilayer without native Al oxide, ALILE would have the advantage of crystallization at a lower temperature and/or for a shorter time, because the activation energy of crystallization becomes close to that (0.8 eV) for the diffusion of Si into Al,11,12) which is much lower than that (1.04–1.8 eV) for a bilayer with native Al oxide.5,10) Such a low-temperature poly-Si fabrication is promising, because the formed poly-Si could be used in thin film transistors or solar cells on lowcost plastic substrates. Recently, Kim et al. indirectly demonstrated layer exchange for AIC without breaking the vacuum by image contrast using a focused ion beam spectroscope,6) but studies on layer exchange using directly structural analyses such as scanning transmission electron microscopy (STEM) have not yet been carried out. Thus, it is important to assess the AIC without native Al oxide and compare it with that with native Al oxide. It is our purpose to show exactly, by STEM and energy dispersive X-ray analysis, that an a-Si/Al bilayer, formed on SiO2 by electron beam evaporation without breaking the vacuum, changes to an Al/poly-Si bilayer even at a low temperature of 400 C. Through the investigation of the influence of the a-Si/Al thickness ratio on poly-Si formation, it is demonstrated that the a-Si/Al thickness ratio of roughly 1 : 1 is suitable to obtain a flat surface morphology. Also, it is shown that the crystal direction is strongly influenced by the thicknesses of the a-Si/Al bilayer and that Si(111)oriented grains become dominant with decreasing thicknesses of the a-Si/Al bilayer. 2.

Experimental

SiO2 was grown onto single-crystal silicon (c-Si) substrates by dry oxidation at 1000 C to form the SiO2 /Si substrates. The SiO2 thickness was approximately 130 nm. Both Al and a-Si films were deposited on the SiO2 /Si substrate by electron beam evaporation without breaking the

4770

Jpn. J. Appl. Phys., Vol. 44, No. 7A (2005)

Y. SUGIMOTO et al.

3.

Results and Discussion

3.1

Layer exchange of a-Si/Al bi-layer without native Al oxide at interface Prior to Al etching, bright-field TEM observation was carried out to investigate the cross-sectional structure of the a-Si (100 nm)/Al (100 nm) bilayer after annealing. The TEM image is shown in Fig. 1, where the sample was annealed at 400 C for 10 h. The TEM image indicates that layer 1/ layer 2 structures exist under the glue. It was verified by electron diffraction that both layers are crystalline. In order to further examine this structure in detail, the depth distribution analysis of each element was also performed using the STEM-equipped EDX system. Figure 2(a) shows a high-angle annular dark-field (HAADF) image. The HAADF image has a characteristic feature, i.e., the contrast is strongly affected by the average atomic number.13) Thus, it

Glue layer1 layer2

100nm Fig. 1. TEM image after annealing of a-Si (100 nm)/Al (100 nm) sample. Annealing was performed at 400 C for 10 h. Al etching was not carried out.

count

400

(b)

(a) Al

300

Distance (nm)

vacuum, where Al thickness (dAl ) and a-Si thickness (da-Si ) were varied within 40–100 and 40–280 nm, respectively, and the deposition rates of Al and a-Si were controlled within 16–18 nm/min; the ultimate base pressures were 2:0  106 and 7:0  107 Torr for the depositions of Al and a-Si, respectively. Heat treatment was performed in an electric furnace tube in dry N2 ambient for a given period, with the temperature kept at 400 C. The annealed sample was taken out from the furnace tube and then dipped in standard Al etching solution (16 parts phosphoric acid, 1 parts nitric acid, 1 parts acetic acid and 2 parts deionized water) at 53 C for 3 min. We observed the film structures by field emission scanning electron microscopy (FE-SEM) and transmission electron microscopy (TEM). Moreover, to analyze the depth distribution of elements in the annealed sample, a scanning transmission electron microscopy (STEM)-equipped energy dispersive X-ray (EDX) system was employed. The optical properties were obtained from the extinction coefficient (k) spectrum measured by spectroscopic ellipsometry (SE). Xray diffraction (XRD) and orientation imaging microscopy (OIM) were also used to obtain crystallographic information.

4771

Si 200

100

O 100nm 1

0

50

100

0

Fig. 2. (a) HAADF-STEM image after annealing of a-Si (100 nm)/Al (100 nm) sample. (b) Depth distribution of each element determined by EDX along line 1 in (a).

can be clearly seen that there are several contrasts on the image, as shown in Fig. 2(a). These contrasts also correlate with the results of EDX investigation, as shown in Fig. 2(b), which was scanned along line 1 in Fig. 2(a). The results also show that the depth distributions of Al and Si are reversed, and furthermore, the counts of Si inside the Al layer are much higher than that of Al inside the Si layer. These results indicate that the exchange of the a-Si/Al bilayer was induced while maintaining similar tendencies as those of the Al–Si binary system below 577 C. Based on the above results, it is concluded that layer 1/ layer 2 structures in Fig. 1 are Al/poly-Si layers, respectively. The thickness of the poly-Si layer is roughly 100 nm, which is equal to the thickness of the evaporated Al film. In addition, since the bright-field TEM image is influenced by crystal qualities such as orientation,14) the grain boundary can be clearly observed, as shown by the arrow in Fig. 1. The height is comparable to the thickness of the poly-Si layer, which means that the vertical grain size is roughly 100 nm. Thus, it is concluded that ALILE takes place without native Al oxide. The fact that the thickness of the poly-Si film after ALILE is equal to that of the evaporated Al film is similar to the experimental results of using native Al oxide obtained by Nast.15) However, our results of layer exchange are contrary to Nast’s experimental evidence that AIC is possible for the case without native Al oxide although ALILE is not. The experiment without native Al oxide was performed on an aSi (250 nm)/Al (200 nm)/glass structure where a-Si was deposited by magnetron sputtering and Al by thermal evaporation. Annealing was carried out at 450 C for 2 h, resulting in the fabrication of mixed films of Al and crystallized poly-Si on the glass substrate.15) Therefore, Nast concluded that an Al oxide layer introduced at the a-Si/Al interface is essential for layer exchange of the a-Si/Al bilayer. The difference between our results and Nast’s results is most likely attributable to the initial states of the a-Si/Al interface during the formation of the a-Si/Al bilayer. According to the literature,16–18) a model for AIC has been

4772

Jpn. J. Appl. Phys., Vol. 44, No. 7A (2005)

Y. S UGIMOTO et al.

Al-Si mixed phase (interdiffusion of Al and Si ) c-Si nucleation (inside a-Si film)

Mixed film of Al and poly-Si (no layer exchange)

a-Si film Al

(a)

Al film

poly-Si Al

Al

c-Si nucleation (inside Al film) no native Al oxide (dissociative diffusion of Si) a-Si film

(b)

Al /poly-Si layer (layer exchange)

Al diffuse out

Al film

a-Si film Al

Al film Poly-Si film (small grain)

c-Si nucleation of high density (inside Al film) Lateral growth of c-Si

Fig. 3. (a) Schematic AIC models having Al–Si mixed phase and (b) no native Al oxide (or very thin Al oxide film) at a-Si/Al interface.

proposed, as shown in Fig. 3(a), in which an Al–Si mixed layer having the Alx Si phase is assumed to exist at the a-Si/ Al interface during annealing of the a-Si/Al bilayer. The Al– Si mixed layer accelerates the interdiffusion of Al and Si, leading to the crystallization of a-Si on both sides of the Al– Si mixed layer. Thus, the results obtained by Nast would be interpreted as indicating that an Al–Si mixed layer was formed at the a-Si/Al interface during the deposition of the a-Si/Al bilayer because of the supply of rather high-energy particles to Al films in sputter deposition. Consequently, the overall layer exchange may be interrupted as being due to the presence of the Al–Si mixed phase. In order to clarify the role of the Al–Si mixed layer, however, further studies are needed. By contrast, the present results clearly showed layer exchange, and the grain size of the poly-Si formed was typically 0.2–3 mm, as demonstrated by OIM evaluation described in §3.3. These results are consistent with the results obtained by Kim et al.6) Here, it should be noted that an electron beam having low deposition energy was used in bilayer formation without the breaking the vacuum in both studies. In the following, we discuss a model for ALILE without native Al oxide. According to the report by Sieber et al.,8) an oxide layer introduced at the a-Si/Al interface is very important from the viewpoint of dissociative diffusion of Si into Al film and the local nucleation of Si in the Al film, which lead to ALILE of the a-Si/Al bilayer and poly-Si formation of large grain size. We believe that a similar behavior occurs in the present ALILE without native Al oxide. The model is shown in Fig. 3(b). There is no Al–Si mixed layer and no native Al oxide (or very thin Al oxide). The Si atoms dissociatively diffuse with high flux into the Al film and nucleation of c-Si occurs with high density in the Al layer. The nucleated Si grows in the vertical direction and arrives at the SiO2 substrate within a short annealing time because of the high diffusivity of Si in Al. With annealing time, c-Si grows to 0.2–3 mm in the lateral direction, maintaining the initial

thickness of the Al film, and the lateral growth stops when the growth is interrupted by that of neighboring c-Si. Thus, layer exchange is performed effectively and the poly-Si formed by the present ALILE has high density and small grain size. 3.2

Influence of a-Si/Al thickness ratio on poly-Si formation Figures 4(a), 4(b), and 4(c) show FE-SEM surface images for da-Si =dAl ¼ 40=100 nm, da-Si =dAl ¼ 100=100 nm, and

(a)

(d) poly-Si island n (dendritic growth) small grain

SiO2 substrate

Poly-Si island 100nm 2.5 µ m

SiO2

(b)

(e) poly-Si film Si protrusion (some pores exist)

pore

Poly-Si film Si protrusion

100nm 2.5 µ m

(c) porous film (crystalline Si)

SiO2

Si protrusion

porous film (crystalline Si)) poly-Si film (some pores exist)

2.5 µ m

(f)

100nm

SiO2

Fig. 4. FE-SEM image for (a) da-Si =dAl ¼ 40=100 nm, (b) 100/100 nm and (c) 280/100 nm. (d), (e) and (f) are the schematic structural representations corresponding to SEM images (a), (b) and (c), respectively. Annealing was performed at 400 C for 10 h.

Jpn. J. Appl. Phys., Vol. 44, No. 7A (2005)

Y. SUGIMOTO et al.

da-Si =dAl ¼ 280=100 nm samples, respectively. Annealing was performed at 400 C for 10 h and then Al was etched. The image of sample (a) shows that Si islands having dendritic structure can be locally observed on the SiO2 substrate, where the height of islands is equal to the thickness of Al films. Furthermore, it is found that the island of dendritic structure consists of grains smaller than 3 mm. Thus, the dendritic island is composed of poly-Si. The schematic representation of such structural formation is given in Fig. 4(d). The model in Fig. 3(b) well explains that the vertical growth of c-Si proceeds rapidly to the same thickness (100 nm) as the Al thickness and that the lateral growth stops because the quantity of a-Si is much smaller than that of Al, leading to poly-Si islands having dendritic structure. For da-Si =dAl ¼ 100=100 nm, the poly-Si film with continuous surface morpholpgy seems to be formed, as shown in Fig. 4(b), although the film has some protrusions and pores. The height of protrusions is 85–100 nm and the depth of pores is roughly 60–100 nm at each position. In the case of da-Si =dAl ¼ 280=100 nm, it can be clearly seen that porous films or protrusions are observed on continuous poly-Si films, as shown in Fig. 4(c). The schematic representations of the results in Figs. 4(b) and 4(c) are given in Figs. 4(e) and 4(f), respectively. The model in Fig. 3(b) again well explains that the continuous poly-Si film can be formed on the SiO2 substrate because the quantity of a-Si is equal to or much larger than that of Al. Therefore it is concluded that the a-Si/Al thickness ratio of roughly 1 : 1 is suitable to obtain a flat surface morphology, since the appearance of porous film or protrusions on continuous poly-Si films can be minimized. The same conclusion is given for ALILE using native Al oxide.9) It is well known that the SE measurement is sensitive to surface morphology and crystal quality. Knowledge of the correlation between the FE-SEM observation and SE results is very convenient for characterizing the poly-Si films formed by ALILE. Figure 5 shows the k spectra obtained by

SE measurement, where the samples were the same as those in Fig. 4. The k spectrum for poly-Si under the condition of da-Si =dAl ¼ 100=100 nm indicates a similar variation to that of the c-Si spectrum obtained from a Si(100) single crystal, showing a peak at the photon energy of 4.35 eV. Here, k is directly proportional to the absorption coefficient of light, and the spectrum tends to exhibit two peaks near 3.4 and 4.3 eV depending on the crystalline quality of Si films.19) The k spectrum indicates that poly-Si film formed by ALILE has a high quality. The SE spectrum for da-Si =dAl ¼ 280=100 nm exhibits similar variation to that for da-Si =dAl ¼ 100=100 nm, which suggests that Al which appeared on the top layer after ALILE acts as a catalyst for the crystallization of excess a-Si. However, the k-values are very low because most of the top surface layers consist of porous films or protrusions. By contrast, the SE spectrum for da-Si =dAl ¼ 40=100 nm is quite different from that for da-Si =dAl ¼ 100=100 nm, because most of the surface layer consists of SiO2 substrates. Thus, it can be concluded that the shape and relative values of the k-spectrum are in good correlation with the surface morphology observed by SEM. 3.3

Influence of a-Si/Al thickness on crystal orientation of poly-Si Figure 6 shows the XRD results for the samples with different film thicknesses (dAl : da-Si ¼ 1 : 1), where the samples were annealed at 400 C for 12 h. We used the SiO2 / Si(511) substrate to avoid detecting the Si(400) peak from the Si(100) substrate in the measured range. In the case of dAl ¼ da-Si ¼ 100 nm, the XRD profile shows several peaks that indicate the crystallization of a-Si film: Si(111), (220), (311) and (400). These diffraction intensities are measured as relative intensities in the XRD system. The integral intensities of (220), (311) and (400) to (111) peaks are generally known to be 55.0, 30.0 and 6.0% for Si powder with randomly oriented grains. In addition, it was found that the Si(111) diffraction intensity increases for dAl ¼ da-Si ¼ 40 nm, compared with that for dAl ¼ da-Si ¼ 60 and 100 nm.

c-Si(100) da-Si dAl

Si (111)

4 3 2

a-Si/Al= 100 /100 nm

1 280 /100 nm

0

a-Si Al SiO2

Si (311)

4

Si (220)

Intensity (arb. units)

Extinction Coefficient k

5

4773

Si (400) dAl=da-Si=100nm

Si (400) dAl=da-Si=60nm

40 /100 nm

Photon Energy (eV)

dAl=da-Si=40nm

5

Fig. 5. Extinction coefficient spectra for c-Si(100) and poly-Si films obtained by ALILE. The poly-Si samples were the same as those in Fig. 4.

20

30

40

50

60

70

80

90

Diffraction angle 2θ (deg) Fig. 6. XRD profiles with different film thicknesses for da-Si ¼ dAl . Annealing was performed at 400 C for 12 h. Measurement was performed with 2= scan.

4774

Jpn. J. Appl. Phys., Vol. 44, No. 7A (2005)

(a)

2.5 µ m

Y. S UGIMOTO et al.

(b)

2.5 µ m

Fig. 7. OIM images for (a) da-Si ¼ dAl ¼ 100 nm and (b) da-Si ¼ dAl ¼ 40 nm samples. Red and blue colors represent areas of Si(100)- and (111)oriented grains having a misorientation less than 15 , respectively. White and black colors represent grains other than those of two dominant orientations and areas whose Kikuchi lines can not be analyzed, respectively. The samples were the same as those in Fig. 6.

It was previously ensured that the thickness of the poly-Si film formed is comparable to that of evaporated Al film under the above condition (dAl : da-Si ¼ 1 : 1), as shown in Fig. 1, which implies that the crystallized volume should be decreased with decreasing thickness of Al film. Nevertheless, the intensity of the Si(111) peak for dAl ¼ da-Si ¼ 40 nm becomes strong. Therefore, the XRD results indicate that the fraction of grains having the Si(111) orientation is increased for dAl ¼ da-Si ¼ 40 nm. When ALILE was performed at 350 C, the intensity of the Si(111) peak became strong even for dAl ¼ da-Si ¼ 60 nm. Regarding these samples of dAl ¼ da-Si ¼ 40, 60 and 100 nm, we also performed the evaluation of local crystal orientation distribution using OIM equipment. The mapping was accomplished to analyze the Kikuchi lines,19) and the boundaries whose rotation angle was more than 15 were defined as high-angle boundaries. Figures 7(a) and 7(b) show two OIM images obtained from dAl ¼ da-Si ¼ 100 and 40 nm samples, respectively. The area of Si(111)-oriented grains is colored blue, and that of Si(100)-oriented grains is colored red. In the present study, (111) and (100)-oriented grains were defined as the grains whose misorientation from ideal (111) and (100) plane is less than 15 . Those OIM images show that the blue and red parts occupied 25.7 and 34.0% of the total area of the sample of dAl ¼ da-Si ¼ 100 nm [Fig. 7(a)]. However, for dAl ¼ da-Si ¼ 40 nm, the blue and red parts, corresponding to the (111) and (100)

101

(a) 100 nm

4.

Conclusion

We fabricated of poly-Si thin film at 400 C by AIC without native Al oxide and demonstrated the structural

111

111

001

grains, reversely occupied 61.3 and 4.6%, respectively, [Fig. 7(b)], which is consistent with the XRD results. The sample of dAl ¼ da-Si ¼ 60 nm had partial fractions of 31.4 and 27.9% for (111) and (100) grains, respectively, which were halfway between those of dAl ¼ da-Si ¼ 40 and 100 nm samples. It was confirmed from the results in Figs. 7(a) and 7(b) that the grain sizes for dAl ¼ da-Si ¼ 100 and 40 nm samples are within 0.23–3.2 and 0.11–3.7 mm, respectively, which are smaller than that in the case of ALILE using native Al oxide. Figures 8(a), 8(b) and 8(c) show inverse pole figures for 100-, 60- and 40-nm-thick samples, respectively, where the samples were the same as those in Fig. 6. The inverse pole figures indicate that the distribution of grain orientation around the Si(111) direction becomes marked with decreasing film thickness. In particular, the inverse pole figure of the 40-nm-thick sample indicates a major distribution around the Si(111) direction corresponding to the exact surface direction. These results suggest that the crystal direction is strongly influenced by the thickness of the a-Si/Al bilayer in addition to annealing temperature. Regarding the crystal direction of poly-Si formed by SPC, Haji et al.20) pointed out that the crystal direction is influenced by the ratio between a-Si film thickness (d1 ) and the mean distance of nuclei (d2 ) at the film/substrate interface. When the distance d2 is much larger than the film thickness d1 , which is the case of a very low nucleation density, the crystallites reach the surface within a short period and growth proceeds two-dimensionally. In this case, the grain shows preferred orientation with Si(111), i.e., very thin poly-Si films after crystallization exhibit (111)-preferred orientation due to the strongly anisotropic rate of grain growth.20) The growth model reported by Haji et al. is similar to the present model shown in Fig. 3(b). We showed that the height of poly-Si grains is decided by the thickness of the Al film in the range of 40–100 nm, and the two-dimensional size of grains is in the range of 100–4000 nm. This situation is satisfied under the condition of d2  d1 , which is enhanced by thinning dAl and da-Si . Therefore, we can explain why Si(111)-oriented grains are dominant after ALILE of a very thin a-Si/Al bilayer.

001

101

(b) 60 nm

111

001

101

(c) 40 nm

Fig. 8. Inverse pole figures indicating the distribution of crystal orientation relative to surface direction: (a) da-Si ¼ dAl ¼ 100 nm, (b) da-Si ¼ dAl ¼ 60 nm and (c) da-Si ¼ dAl ¼ 40 nm samples. The samples were the same as those in Fig. 6.

Jpn. J. Appl. Phys., Vol. 44, No. 7A (2005)

properties of the obtained poly-Si films. Even though native Al oxide was not inserted, layer exchange from the a-Si/Al to Al/poly-Si bilayer was successfully induced. However, the grain size was one order of magnitude smaller than that of poly-Si fabricated by ALILE using native Al oxide. EDX analysis revealed that the depth distributions of Al and Si inside the Al/poly-Si bilayer were similar to the relation of solid solubility on the Al–Si binary system below 577 C. The influence of the a-Si/Al thickness ratio on poly-Si formation was also investigated for ALILE without native Al oxide. It was verified by FE-SEM and SE that the a-Si/Al thickness ratio of roughly 1 : 1 is suitable for obtaining a flat surface morphology of poly-Si. Furthermore, the influence of a-Si/Al thickness on the crystal direction of poly-Si was investigated for ALILE without native Al oxide. The crystal direction was strongly influenced by the thicknesses of Al and Si and showed preferred orientation with Si(111) with decreasing thicknesses. The phenomenon was successfully explained with the aid of Haji et al.’s model. Acknowledgment This work was performed using the clean room facilities of the Art, Science and Technology Center for Cooperative Research in Kyushu University. This study was partially supported by a Grant-in-Aid for Science Research B (16360156) from the Ministry of Education, Culture, Sports, Science and Technology.

Y. SUGIMOTO et al.

4775

1) A. T. Voutsas: Appl. Surf. Sci. 208–209 (2003) 250. 2) A. Hara, K. Yoshino, F. Takeuchi and N. Sasaki: Jpn. J. Appl. Phys. 42 (2003) 23. 3) Z. Jin, G. A. Bhat, M. Yeung, H. S. Kwok and M. Wong: J. Appl. Phys. 84 (1998) 194. 4) O. Nast, T. Puzzer, L. M. Koschier, A. B. Sproul and S. R. Wenham: Appl. Phys. Lett. 73 (1998) 3214. 5) S. Gall, M. Muske, I. Sieber, O. Nast and W. Fuhs: J. Non-Cryst. Solids 299–302 (2002) 741. 6) H. Kim, D. Kim, G. Lee, D. Kim and S. H. Lee: Sol. Energy Mater. Sol. Cells 74 (2002) 323. 7) O. Nast and A. J. Hartmann: J. Appl. Phys. 88 (2000) 716. 8) I. Sieber, R. Schneider, I. Doerfel, P. Schubert-Bischoff, S. Gall and W. Fuhs: Thin Solid Films 427 (2003) 298. 9) O. Nast and S. R. Wenham: J. Appl. Phys. 88 (2000) 124. 10) P. I. Widenborg and A. G. Aberle: J. Cryst. Growth 242 (2002) 270. 11) J. O. McCaldin and H. Sankur: Appl. Phys. Lett. 19 (1971) 524. 12) H. Qingheng, E. S. Yang and H. Izmirliyan: Solid-State Electron. 25 (1982) 1187. 13) E. Carlino, S. Modesti, D. Furlanetto, M. Piccin, S. Rubini and A. Franciosi: Appl. Phys. Lett. 83 (2003) 662. 14) Y. Aya, A. Ando, S. Yamasaki and K. Wakisaka: Jpn. J. Appl. Phys. 42 (2003) L1302. 15) O. Nast: Dr. Thesis, Marburg, Germany, 2000 p. 51. 16) M. S. Haque, H. A. Nassem and W. D. Brown: J. Appl. Phys. 75 (1994) 3928. 17) M. S. Ashitikar and G. L. Sharma: J. Appl. Phys. 78 (1995) 913. 18) J. H. Kim and J. Y. Lee: Jpn. J. Appl. Phys. 35 (1996) 2052. 19) J. Gao, H. Nakashima, N. Sakai, D. Gao, J. Wang, K. Furukawa and K. Muraoka: Jpn. J. Appl. Phys. 38 (1999) L220. 20) L. Haji, P. Joubert, J. Stoemenos and N. A. Economou: J. Appl. Phys. 75 (1994) 3944.