Stability of structural defects of polycrystalline silicon grown by rapid thermal annealing of amorphous silicon films

Stability of structural defects of polycrystalline silicon grown by rapid thermal annealing of amorphous silicon films

ELSEVIER Thin Solid Films 268 ( 1995) 14 Stability of structural defects of polycrystalline silicon grown by rapid thermal annealing of amorphous si...

474KB Sizes 0 Downloads 46 Views

ELSEVIER

Thin Solid Films 268 ( 1995) 14

Stability of structural defects of polycrystalline silicon grown by rapid thermal annealing of amorphous silicon films D. Girginoudi a7*,S. Girginoudi a, A. Thanailakis a, N. Georgoulas a, J. Stoemenos b, J. Antonopoulos b aDepartment ofElectrical and Computer Engineering, Democritus University of Thrace, 67100 Xanthi, Greece b Department of Physics. Aristotle University of Thessaloniki, 54006 Thessaloniki, Greece

Received 18 January 1995; accepted 3 May 1995

Abstract The crystallization of a-Si films, grown by low-pressure chemical vapor deposition and annealed by rapid thermal annealing (RTA), at 850 “C in conjunction with conventional heating at 600 “C for 6 h, has been studied using transmission electron microscopy. The results of RTA at 850 “C showed that the improvement of the poly-Si structure (large crystallites with low density of microtwins) was maximized for an annealing time of 150 s. The same results were also obtained by RTA at 850 “C with successive steps of 30 s duration each (5 X 30 s). A multiple-step annealing is sufficient to activate the movement of twin boundaries within the grains resulting in their annihilation. This process is compatible with the technology of growing good-quality poly-Si on low-cost glass substrates. Keywords: Structural properties; Silicon; Annealing; Crystallization

1. Introduction The growth of poly-Si films on glass substrates is a process of great technological interest for the fabrication of low-cost solar cells [ 11, flat-panel displays for television and videographic applications [ 2,3], and large-scale artificial neural networks [4]. If low-cost glass substrates are to be used in all these applications, the growth temperature of poly-Si films as well as the processing temperature during the fabrication of thin film transistors should be kept below the softening point of glass, which is around 650 “C. The various methods used by research workers to grow poly-Si films are: the crystallization of a-Si, grown at 530 “C by low-pressure chemical vapor deposition (LPCVD) and subsequently annealed at 600 “C for 6 h [ 51; the crystallization by excimer laser of aSi [ 21; and the crystallization by rapid thermal annealing of a-Si [ 6,7]. In the case of annealing at 6OO”C, large crystallites are formed with a large number of in-grain defects, mainly twins, which are unstable and can be eliminated by annealing at 850 “C [ 81. However, long annealing at this temperature is forbidden for glass substrates. This problem can be overcome by using the rapid thermal annealing (RTA) technique. It has recently been shown that crystallization of a-$ grown * Corresponding author. 0040-6090/95/$09.50 0 1995 Elsevier Science S.A. All rights reserved SSD10040-6090(95)06677-2

on glass substrates, by annealing at 600 “C for 6 h and by subsequent rapid thermal annealing at 850 “C for 45 s, significantly improves the electrical properties of poly-Si films [ 91. For annealing times longer than 45 s, the glass substrate temperature exceeds its softening point, i.e. 650 “C. Despite the significant improvement in the quality of poly-Si, the RTA time of 45 s is not sufficient to remove all microtwins from the interior of crystallites [ 8,9]. In this work the crystallization studies of a-Si films deposited on SiO, by LPCVD, at 600 “C for 6 h, and subsequently subjected to a multi-step RTA at 850 “C are presented. The effect of RTA duration on the elimination of microtwins in poly-Si films is studied by transmission electron microscopy (TEM) . It is shown that the improvement of the poly-Si was maximized for an annealing time of 150 s. The same results were also obtained using RTA at 850 “C with multiple successive equal time steps of 30 s duration each (5 X 30 s) . The second process is compatible with the technology of growing good-quality poly-Si on glass substrates.

2. Experimental

procedure

Amorphous Si films of 50 nm thickness were deposited at 530 “C, by the LPCVD technique, on thermally oxidized 4

D. Girginoudi et al. /Thin Solid Films 268 (1995) 1-f

2

3. Results and discussion

0

40

200

240

Fig. 1. Schematic illustration of RTA at 850 “C with multiple successive steps, each of 30 s duration (5 X 30 s).

inch Si wafers. The thickness of the thermal oxide was 100 nm. The reaction gas was a mixture of SiI-L,and H2 with flow rates of 20 and 100 seem, respectively, and the total pressure was 266.6 Pa. The deposition was carried out in a cold stainless-steel wall, single-wafer RTLPCVD JIPELEC system. The substrates were heated, through double-quartz windows, by 12 W halogen lamps and the temperature was measured and controlled by an optical pyrometer focused at the centre of the sample back side. The pyrometer was calibrated using a K-type thermocouple. The heating up rate was 250 ‘C s-’ and the cooling down rate was around 80 “C s- ‘, the total time taken to reach room temperature being less than 1 min. The a-Si films were subsequently crystallized by means of one of the following processes. 1. Annealing at 600 “C/6 h. 2. Annealing at 600 “C/6 h and subsequent RTA at 850 “C for 30 s, or 150 s. 3. Annealing at 600 “C/6 h and subsequent RTA at 850 “C for multiple successive steps of 30 s each (5x30 s) , as is shown in Fig. 1. 4. RTA at 850 “C for 30 s, or 150 s. 5. RTA at 850 “C for multiple successive steps of 30 s each (5x30 s) (Fig. 1). The RTA processing was carried out in a pure argon flow using the RTLPCVD system described above. The structural characterization of these poly-Si films was carried out by combining cross-sectional and plan-view TEM observations, using a JEM12OCX and a JEM2OOOFXmicroscope.

The TEM morphological and structural characteristics of poly-Si films, formed by the different annealing processes as described in Section 2, are presented in Table 1. Fig. 2 shows the structural features of the five most representative poly-Si samples. Three photographs of the same sample are included in each column, presenting the plan-view micrograph (mainly dark-field images), its corresponding diffraction pattern and the cross-sectional micrograph, respectively. In the case of sample D,, which was subjected to rapid thermal annealing at 850 “C for successive steps of 30 s each, the mean size of crystallites is relatively small, as it is shown in Fig. 2(a), but the crystallites are free from internal defects, as it is clearly evident from the particularly sharp diffraction rings shown in Fig. 2(f) . This morphology is also supported by the corresponding cross-sectional micrograph shown in Fig. 2(k) . Despite the small thickness of the films, the crystallites do not reach the surface. Samples Do and D2 have characteristics similar to those of sample Di and, thus, they are not included in Fig. 2. Sample Dgr which was only annealed at 600 “C for 6 h, is presented in Fig. 2(b), 2(g) and 2( 1). In this case, very large crystallites with a mean size of 210 nm were grown. However, within these grains a high density of defects, mainly twins with a mean size of about 30 nm is evident. These multiple twins cause a broadening of the diffraction rings, which is clearly evident on the first ring d, 1, = 0.38 nm of Fig. 2 (g) . In the same figure, forbidden diffraction spots can be observed and they are marked with an arrow. These spots are attributed to double electron diffraction, owing to the presence of microtwins. Moreover, there appear diffraction lines, denoted by the letter R in Fig. 2(g) . The existence of these lines is attributed to the presence of parallel twins. This becomes evident in the high-magnification micrograph (Fig. 3) taken from a large crystallite. Inside the grain a large number of small twins is observed, resulting in a tweed-like structure. Thin { 111 } multiple twins result in the formation of streak intensities in the reciprocal lattice, shown by the diffraction pattern in the inset of Fig. 3. The streak intensity along the ( 111) direction is attributed to the size effect, owing to the very high density of the thin lamellae twins [ lo]. In sample Ds, which was annealed at 600 “C for 6 h and then subjected to rapid thermal annealing at 850 “C for 30 s,

Table 1 The. mean grain size and the density of microtwins in poly-Si films of 50 nm thickness for different annealing conditions Sample identification DO Dl D2 D3 D4 D5 D6

Conventional heating at 600°C for 6 h

Yes Yes Yes Yes

RTA at 850 “C

Mean grain size (nm)

Density of microtwins

30 set 5X3Osec 150 see 150 set 5X30sec 30 set -

45 50 50 235 230 210 210

Small Small Small Small Small Medium High

D. Girginoudi et al. /Thin Solid Films 268 (1995) 14

the mean size of crystallites is about the same as in sample Dg. However, the density of twins has been reduced significantly, as shown in Fig. 2(c). The absence of diffraction lines from the diffraction pattern, shown in Fig. 2(h) , verifies this fact. However, the application of RTA for 30 s is not sufficient to eliminate the microtwins, as it is evident from the broadening of the diffraction rings. In the case of sample D4, which was crystallized by annealing at 600 “C for 6 h and, then, was further annealed by multiple successive RTA steps of 30 s each (5 X 30 s), the microtwins have been eliminated, as it shown in Fig. 2(d). This is also verified by the diffraction pattern shown in Fig. 2(i), where sharp diffraction rings are evident, without forbidden spots owing to the double diffraction observed in sample D,. The most important difference is that the crystallites in sample D4 are about five times larger than those of sample D,, as may be seen from the cross-sectional micrographs shown in Fig. 2(k) and 2(n). Sample D3 was annealed in a conventional furnace at 600 “C for 6 h and, then was further subjected to RTA for 150 s. As it can be seen from Fig. 2(e), the crystallites are similar in size to those of sample D4. The same result is derived from the diffraction patterns shown in Fig. 2(j) and 2(i). It is concluded that the RTA process of multiple successive steps, each of 30 s duration, is equivalent to the continuous RTA process of the same total duration. Furthermore, it is very important to stress that the process of multiple successive RTA steps results in improvement of the quality of poly-Si without any danger of glass-substrate deformation, which would have occurred in the case of long-term exposure to high temperatures. The samples Do, D, and Dz, which were subjected only to the RTA process, contain crystallites of approximately the

3

Fig. 3. High magnification micrograph taken from a crystallite with the electron beam parallel to the [ 1lo] crystallographic direction, as is shown in the diffraction pattern in the inset of the micrograph. Long twins, at the centre of the gram, are evident. Microtwins. also running along the same direction at the side bands of the long central twins, result in a tweed-like stmcture. Streaks in the reciprocal space, along the [ I 111 direction, are due to the size effect produced by the high density of twins. Satellite spots of the l/3( 111) and 2/3( 111) type are also evident, owing to double diffraction of the electron beam by the superimposed microtwins.

same size, despite the great differences in annealing times. The size of these crystallites is much smaller compared with that of sample Dg, which was annealed at a lower temperature (Table 1) . This type of behaviour suggests that in the cases of samples Do, D, and Dz nucleation is homogeneous, i.e. a large number of nuclei is activated owing to the high RTA temperature. Since crystallization follows an Arrhenius law, the crystallization process at 850 “C is completed in times shorter than 30 s, but the size of the crystallites is small.

Fig. 2. Plan-view micrographs, their corresponding diffraction patterns and the cross-sectional micrographs of poly-Si films of 50 nm thickness, grown by conventional heating at 600 “C for 6 h and/or by RTA at 850 “C for different times, according to Table 1. The sample identification symbol D, is shown at the upper right-hand comer of the plan-view micrographs.

D. Girginoudi et al. /Thin Solid Films 268 (1995) l-4

tiple steps, in order to avoid the transfer of excessive heat to the glass substrates.

4. Conclusions Transmission electron microscopy studies of poly-Si films grown by conventional heating of a-Si films at 600 “C for 6 h, and subsequently by further rapid thermal annealing at 850 “C for different time intervals, have shown that a RTA time of 150 s results in crystallites of large size with low density of microtwins. The same results were also obtained by RTA at 850 “C with successive steps of 30 s duration each (5 X 30 s) . This process is compatible with the technology of growing good-quality poly-Si films on low-cost glass substrates. Fig. 4. Illustration of the TPRE mechanism, which permits a fast incorporation of atomsat the comer of the twin plane. At the left twin boundary two atoms are sufficient to complete a ring, instead of three for a defect-free ( 111) surface Thus, the formation of multiple twins accelerates the growth of crystallites.

Further annealing at 850 “C does not have any significant effect on the size of the grains. This is expected, because secondary crystallization in silicon, which implies movement of the grain boundaries, occurs at temperatures above 1150 “C. Low-temperature annealing (around 600 “C) results in heterogeneous nucleation, and consequently the number of activated nuclei is small. Crystallization proceeds through the formation of microtwins and, consequently, crystallites of large size are formed, which contain a large number of defects, as is the case with specimen D6. The formation of twins accelerates crystallization. Solid-state crystallization of Si by the twin-plane re-entrant edge (TPRE) mechanism has been proposed by other workers [ 11,121. The TPRE mechanism facilitates the growth, permitting a fast incorporation of atoms at low-energy sites at the corners of the twin planes [ 131, as shown diagrammatically in Fig. 4. The twin planes which are created are faults with low formation energy of the order of 30 erg cm-‘, while the formation energy for a high angle boundary is about 1000 erg cm - * [ 81. Consequently, the crystallization of a-Si through microtwins results in a rapid reduction of the free energy of the system, by creating large but strongly twinned grains. However, twin boundaries are very mobile above 750 “C and can be eliminated by a short-time annealing above this temperature. Thus, the low temperature annealing at 600 “C/6 h followed by RTA at 850 “C results in the formation of larger grains free of defects. We have shown that this RTA process can proceed by mul-

Acknowledgements The financial support from the Greek Ministry of Industry, Energy and Technology, General Secretariat of Research and Technology (research program “STRIDE HELLAS 8”) is gratefully acknowledged.

References [ 11 Z. Shi, T.L. Young, G.F. Zheng and M.A. Green, Sol. Energy Mater. Solar Cells. 31( 1) (1993) 51. [2] S.D. Brotherton, J.R. Ayres and N.D. Young, Solid-State Electron., 34 (1991) 671. [3] I. Wei Wu, Solid State Phenom., 37-38 (1994) 553. [4] P. Migliorato, in H.E. Maes, R.P. Mertens. R.J. Van Overstraeten (eds.), Eur. Solid Stare Device Research ConjY, Leuven 14-17 Sept. 1992, Elsevier, Amsterdam, 1992, p. 89. [S] T. Noma, T. Yonehara and H. Kumomi, Appl. Phys. Lett., 59 (1991)

653. [6] M. Bonnel, N. Duhamel, M. Guendouz, L. Haji, B. Loisel and P. Ruoult, Jpn. J. Appl. Phys., 30 (1991) L1924. 171 H. Pattyn, J. Poortmans, P. Debenest, M. Caymax, P. Vetter, M. Elliq, E. Fogarassy, Z. Nenyei, J. Nijs and R. Mertens, in H.E. Maes, R.P. Mertens and R.J. Van Overstaeten (eds.), Eur. Solid State Device Research Cant, Leuven IC17Sept 1992, Elsevier, Amsterdam, 1992, p. 97. [8] L. Haji, P. Joubert, J. Stoemenos and N.A. Economou, J. Appl. Phys., 75 ( 1994) 3944. 191 J. Stoemenos, N.A. Economou, L. Haji, M. Bonnel, N. Duhamel and B. Loisel, Solid State Phenom., 37-38 (1994) 287. [lo] P.B. Hirsch, A. Howie. R.B. Nicholson, D.W. Pashley, Electron Microscopy ofThin Crysrals, Chapter 6. Butterworths, London, 1965. [l II C.H. Chu, Y.M. Lu and M.H. Hon. J. Mater. Sci., 27 (1992) 3883. 1121D.J. Cheng, W.J. Styg, D.H. Kuo and M.H. Hon, J. Elecrrochem. Sot.. 134 (1987) 3145. t 131 R. Drosd and J. Washburn, J. Appl. Phys., 53 (1982) 397.