- Email: [email protected]
Ultrahigh vacuum system for atomic-scale planarization of 6 inch Si(001) substrate
__ __ l!iE
J-s3
applied
surface science
ELSEVIER
Applied Surface Science lOO/lOl
(1996) 31 l-315
Ultrahigh vacuum system for atomic-scale planarization of 6 inch Sit00 1j substrate Ken Idota a** $’, Masaaki Niwa ‘, Isao Surnita a’2 ’ Matsushita h Semiconductor
Rrsetrrch
Research Insrime Center. Matsushita
Tokyo Inc.. 3-10-I Electric
Higashirnita,
Ind. Co.. Ltd.. 3-l-l
Received 22 August 1995; accepted
Tama-Xu. Kawasaki
Yaglrltlo-nakaa~achi.
13 November
214, Japan
Moriguchi.
Osaka 570. Japan
1995
Abstract Ultrahigh vacuum (UHV) system equipped with aluminum radiation shields has been developed in order to obtain atomically flat Si(OOl) surfaces in a large area. Aluminum radiation shields were employed to suppress the temperature rise of chamber walls. The UHV system has enabled a 6 inch substrate to be heated up to 700°C under a pressure of less than IO-’ Pa range. The surface structure of the substrate was confirmed by the reflection high energy electron diffraction (RHEED). when the 6 inch substrate was heated at 600°C in UHV after a modified RCA cleaning, the RHEED pattern indicated oxide free surface consisted of 2 X I structure. The roughness was evaluated from numerous observations by cross-sectional transmission electron microscopy (XTEM). Root-mean-square roughness of the Si surface was less than 0. I nm for almost all over the 6 inch substrate. This implies that the heating in this system gives rise to the atomic-scale planarization of the 6 inch substrate.
1. Introduction Atomic-scale planarization of Si surfaces has become a very important technique for the ultralargescale integrated circuit (ULSI) processing, since the mobility of electrons within the inversion layer and the dielectric breakdown characteristics strongly depend on the surface microroughness [ 1,2]. Recently, planarization of Si surfaces has been extensively studied by a wet treatment [3-71. On the other, an atomic-scale planarization of the small pieces of Si(OO1) substrate by heating in ultra-
_ Corresponding author. Tel.: + 81-6-9064863: 906345 1. ’ Fax: +8l-449111491: Tel.: +81-44.9116351. ‘Fax: +81-44-9111491:Tel.: +81-44-9116351. 0169.4332/96/$15.00 Copyright P/I SO169-4331(96)00233-4
fax:
+ 81-6.
high vacuum (UHV) has been investigated [8,9]. In view of the practical application for ULSI technology, it is required to develop a system which has the ability to heat a Si substrate by a wafer size in UHV with lower temperature. This paper describes a newly developed UHV system for atomic-scale planarization of a 6 inch Si substrate by modified wet treatment prior to heating in UHV.
2. Aluminum
radiation
shield
In order to obtain an atomic-scale planarized Si surface, high temperature heating of Si substrate in UHV is necessary. However, when the substrate is heated at high temperature in vacuum chamber, a
0 1996 Elsevier Science B.V. All rights reserved.
large amount of radiant heat evolves from the 6 inch Si substrate and the substrate heater. This will lead vacuum degradation inside the chamber. In order to prevent the radiant heat from raising the temperature of chamber wall, we tried to use an aluminum radiation shield. The effect of aluminum radiation shield was confirmed experimentally using the chamber shown in Fig. l(a). A conventional liquid nitrogen shroud made of stainless steel was also prepared for comparison, as shown in Fig. l(b). The aluminum radiation shield made of pure aluminum is cylindrical pane1 finished to mirror surface. Pure aluminum is preferable for the radiation shield because of its high reflectance against infrared rays. In Fig. 1, each chamber is evacuated by ion pump with a pumping speed of 500 1 s- ‘. A base pressure in each chamber was 1 X lop8 Pa. Fig. 2 shows temperature dependence on pressure for the cases when aluminum radiation shield and the liquid nitrogen shroud are used. The temperature was measured with the thermocouple near the substrate heater made from tantalum. The result for the shroud without the liquid nitrogen is also shown in this figure. When the shroud with the liquid nitrogen was used, the pressure revealed almost constant up to 5OO”C, since the shroud acts as a cryopump. On the other hand, the pressure increases at temperatures above 500°C with increasing the temperature due to desorption of molecule absorbed on the shroud. It is necessary to supply additional liquid nitrogen to
lo-5
, , , (. ! -DA1 radiation shield .-3SUS shroud wlthout Liq.N, USUS shroud with Liq.N,
,o-‘0
200
400
600
600
I
1000
-
1 DO
Temperature (“C) Fig. 2. Temperature
dependence on pressure in a vacuum system with respect to the introducing of the aluminum radiation shield. the shroud with liquid nitrogen. and the shroud without liquid nitrogen.
suppress the desorption. For the aluminum radiation shield, the pressure also increases with increasing the temperature. However, the increasing rate is smaller than that for the shroud with the liquid nitrogen in the range from 500°C to 900°C. This indicates that the aluminum radiation shield is promising method to suppress the increase of the outgassing at high temperature. For the shroud without the liquid nitrogen, the outgassing is greater than that of the other two cases under the whole temperature region. This is because stainless steel is easily be heated up compared with aluminum.
3. UHV system
Heater
Heater
%
1 Radiation Shield (Pure Aluminum) Liquid Nitrogen Shroud (SUS) (a)
(W
Fig. 1. Schematic diagram of the experimental chambers for equipped aluminum radiation shield (a) and equipped liquid nitrogen shroud (b). Each chamber is evacuated by ion pump with a pumping speed of 500 1 s- ’
Fig. 3 shows the UHV system newly developed for the planarization of 6 inch Si substrate. This system consists of three chambers, i.e., a cleaning, a preheating, and a load-lock. Aluminum radiation shields made from mechanochemical polished pure aluminum plates are placed both in the cleaning and the preheating chambers, and pipes for water cooling are mounted on the aluminum radiation shields. This system is evacuated by turbomolecular pumps, ion pump, and titanium sublimation pumps. An extractor gauge and a B-A gauge were installed on the cleaning and on the preheating chambers, respectively. A base pressure of 3 X lo-” Pa was achieved in the cleaning chamber. Tantalum resistive heaters were introduced for the 6 inch substrate heating from the
K. Idota et al./Applied
Surfucr Science lOO/ 101 (14%~ 311-315
4. Planarization
B
313
process for 6 inch substrate
2 P)
T 5 I=
I
Preheating ch. .#;OOX600
‘\
IP I (10001/s)
EHEED-Gufl
\
L-l
Aluminum Radiation Shield
Fig. 3. UHV system for planarization
of the 6 inch Si substrate.
back side of the substrate. Temperature on Si substrate surfaces was measured by thermocouples and pyrometer. A reflection high energy electron diffraction (RHEED) was equipped in the cleaning chamber. Fig. 4 illustrates the temperature dependence on pressure when the 6 inch Si substrate is heated by tantalum resistive heater in the cleaning chamber. The pressure increased abruptly at above 700°C with increasing the temperature. However, the substrate was heated up to 700°C under a pressure of less than 5 X lop8 Pa.
10-B 0
’
= 200
400
600
Temperature Fig. 4. Temperature dependence substrate is heated by tantalum chamber.
800
P-type, 6 inch Si(OO1) substrates (p = 5-10 0 cm) with a misorientation angle of less than $0.5” were used in this experiment. At first, the substrates were cleaned by a ‘modified RCA cleaning (m-RCA cleaning)‘. The m-RCA cleaning consists of three steps similar to the standard RCA cleaning [lo]. NH,OH/H,O* ratio is decreased to less than 0.2 in the first step. As the second step, the chemical oxide layer on the surface was removed in a diluted HF solution. In order to form a very thin oxide film on the substrate, the final step treatment was performed in a HCl:H,O,:H,O (1: 1:4) solution for 2 min. The m-RCA cleaning was carried out at room temperature. After this m-RCA cleaning, the substrates were introduced into the system and heated in the cleaning chamber. Si surface structure was observed by the RHEED with an acceleration voltage of 29 kV. Incident direction of primary electron beam was (1 IO).
1000
(‘C)
on pressure when the 6 inch Si resistive heater in the cleaning
Fig. 5. RHEED patterns from the 6 inch Si(OO1) substrate after the modified RCA cleaning. (a) Before heating, (b) heating at 600°C for 90 h in UHV.
314
K. Idoturtal./AppliedSu~acrScirncr
lOO/lOl
(19961311-315
Fig. 6. XTEM micrographs of the 6 inch substrates after the modified RCA cleaning (a) and after the UHV-heating The hatched area (left side) shows the planarized area with rms roughness of less than 0. I nm.
Fig. 5(a) shows RHEED pattern from the Si surface prior to heating after the m-RCA cleaning. On the other, Fig. 5(b) shows the case when heated at 600°C under a pressure of less than 5 X 10m8 Pa after the m-RCA cleaning. In Fig. 5(a), RHEED pattern indicates 1 X 1 structure. The surface structure began to convert from 1 X 1 to 2 X 1 at 585°C in UHV, although the 2 X 1 pattern was indistinct. Then, the 2 X 1 RHEED pattern was becoming clear during heating until 600°C. Finally, the RHEED pattern revealed clear 2 X 1 reconstructed structure as shown in Fig. 5(b). The 2 X 1 pattern with sharp streaks indicates the oxide free reconstructed clean surface. Surface contaminated structures like SIC were not observed in Fig. 5(b). Si surface roughness was observed by cross-sectional transmission electron microscopy (XTEM), and root-mean-square (rrns) roughness was evaluated by digitizing the lattice site on the Si(OO1) surface. Fig. 6(a) and (b) show the XTEM micrographs of the 6 inch substrate after the m-RCA cleaning and after the UHV-heating at 600°C for 90 h, respectively. Typical micrographs selected from numerous XTEM observations are shown in this figure. As for
at 600°C for 90 h (b).
surface roughness, apparent improvement was found for the substrate heated in UHV in comparison with the m-RCA cleaned substrate. The rms roughness in Fig. 6(b) was less than 0.1 nm. These results indicate that atomic-scale planarization is caused by the UHV-heating at 600°C for 90 h. UHV-heating at 600°C might cause the extremely slow migration. The planarized area (hatched) where the rms roughness was at most 0.1 nm are shown in the left side of this figure. In addition, no dislocations were generated on the surface at all. From these considerations, it was clarified that the atomic-scale planarization was attained within a whole practical area generally used for device fabrication.
5. Conclusion A newly developed UHV system equipped with aluminum radiation shields was demonstrated. By means of this system, a 6 inch substrate was heated at above 700°C under the pressure of the order of lo-’ Pa.
K. Idota et al./Applied
Sur$ace Science lOO/ 101 (1996) 311-315
Combined with the modified RCA cleaning, heating in this system provides an atomic planarization of the large diameter Si substrates for practical use. This atomically flat surface (rms < 0.1 nm) was realized at relatively low temperature (600°C) under a pressure of less than 5 X lo-’ Pa.
Acknowledgements This work was performed under the management of FED as a part of the MIT1 R&D program (Quantum Functional Devices project) supported by NEDO. The authors greatly appreciate ElKO Engineering Co., Ltd. throughout the generous support of the fabrication of the UHV system.
315
References [l] N.D. Arora and G.S. Gildenblad, IEEE Trans. Electron Devices ED-34 (19871 89. [2] S. Takagi, M. Iwase and A. Toriumi, in: Ext. Abstr. 22nd Int. Conf. on Solid State Devices and Materials, Sendai, 1990, p. 275. [3] T. Yasaka. K. Kanda. K. Sawara, S. Miyazaki and M. Hirose, Jpn. J. Appl. Phys. 30 (1991) 3567. [4] T. Ohmi, M. Miyashita, M. Itano, T. Imaoka and I. Kawanabe. IEEE Trans. Electron Devices ED-39 (1992) 537. [5] G.S. Higashi, R.S. Becher, Y.J. Chabal and A.J. Becher, Appl. Phys. Lett. 58 (1991) 1656. [6] S. Watanabe, N. Nakayama and T. Ito, Appl. Phys. Lett. 59 (1991) 1458. [7] P. Jakob. P. Dumas and Y.J. Chabal, Appl. Phys. Lett. 59 (1991) 2968. [8] M. Niwa. M. Udagawa, K. Okada, T. Kouzaki and R. Sinclair. Appl. Phys. Lett. 63 (1993) 675. [9] M. Niwa, T. Kouzaki, K. Okada, M. Udagawa and R. Sinclair, Jpn. J. Appl. Phys. 33 (1994) 388. [lo] W. Kern and D.A. Puotinen. RCA Rev. 31 (1970) 187.
J-s3
applied
surface science
ELSEVIER
Applied Surface Science lOO/lOl
(1996) 31 l-315
Ultrahigh vacuum system for atomic-scale planarization of 6 inch Sit00 1j substrate Ken Idota a** $’, Masaaki Niwa ‘, Isao Surnita a’2 ’ Matsushita h Semiconductor
Rrsetrrch
Research Insrime Center. Matsushita
Tokyo Inc.. 3-10-I Electric
Higashirnita,
Ind. Co.. Ltd.. 3-l-l
Received 22 August 1995; accepted
Tama-Xu. Kawasaki
Yaglrltlo-nakaa~achi.
13 November
214, Japan
Moriguchi.
Osaka 570. Japan
1995
Abstract Ultrahigh vacuum (UHV) system equipped with aluminum radiation shields has been developed in order to obtain atomically flat Si(OOl) surfaces in a large area. Aluminum radiation shields were employed to suppress the temperature rise of chamber walls. The UHV system has enabled a 6 inch substrate to be heated up to 700°C under a pressure of less than IO-’ Pa range. The surface structure of the substrate was confirmed by the reflection high energy electron diffraction (RHEED). when the 6 inch substrate was heated at 600°C in UHV after a modified RCA cleaning, the RHEED pattern indicated oxide free surface consisted of 2 X I structure. The roughness was evaluated from numerous observations by cross-sectional transmission electron microscopy (XTEM). Root-mean-square roughness of the Si surface was less than 0. I nm for almost all over the 6 inch substrate. This implies that the heating in this system gives rise to the atomic-scale planarization of the 6 inch substrate.
1. Introduction Atomic-scale planarization of Si surfaces has become a very important technique for the ultralargescale integrated circuit (ULSI) processing, since the mobility of electrons within the inversion layer and the dielectric breakdown characteristics strongly depend on the surface microroughness [ 1,2]. Recently, planarization of Si surfaces has been extensively studied by a wet treatment [3-71. On the other, an atomic-scale planarization of the small pieces of Si(OO1) substrate by heating in ultra-
_ Corresponding author. Tel.: + 81-6-9064863: 906345 1. ’ Fax: +8l-449111491: Tel.: +81-44.9116351. ‘Fax: +81-44-9111491:Tel.: +81-44-9116351. 0169.4332/96/$15.00 Copyright P/I SO169-4331(96)00233-4
fax:
+ 81-6.
high vacuum (UHV) has been investigated [8,9]. In view of the practical application for ULSI technology, it is required to develop a system which has the ability to heat a Si substrate by a wafer size in UHV with lower temperature. This paper describes a newly developed UHV system for atomic-scale planarization of a 6 inch Si substrate by modified wet treatment prior to heating in UHV.
2. Aluminum
radiation
shield
In order to obtain an atomic-scale planarized Si surface, high temperature heating of Si substrate in UHV is necessary. However, when the substrate is heated at high temperature in vacuum chamber, a
0 1996 Elsevier Science B.V. All rights reserved.
large amount of radiant heat evolves from the 6 inch Si substrate and the substrate heater. This will lead vacuum degradation inside the chamber. In order to prevent the radiant heat from raising the temperature of chamber wall, we tried to use an aluminum radiation shield. The effect of aluminum radiation shield was confirmed experimentally using the chamber shown in Fig. l(a). A conventional liquid nitrogen shroud made of stainless steel was also prepared for comparison, as shown in Fig. l(b). The aluminum radiation shield made of pure aluminum is cylindrical pane1 finished to mirror surface. Pure aluminum is preferable for the radiation shield because of its high reflectance against infrared rays. In Fig. 1, each chamber is evacuated by ion pump with a pumping speed of 500 1 s- ‘. A base pressure in each chamber was 1 X lop8 Pa. Fig. 2 shows temperature dependence on pressure for the cases when aluminum radiation shield and the liquid nitrogen shroud are used. The temperature was measured with the thermocouple near the substrate heater made from tantalum. The result for the shroud without the liquid nitrogen is also shown in this figure. When the shroud with the liquid nitrogen was used, the pressure revealed almost constant up to 5OO”C, since the shroud acts as a cryopump. On the other hand, the pressure increases at temperatures above 500°C with increasing the temperature due to desorption of molecule absorbed on the shroud. It is necessary to supply additional liquid nitrogen to
lo-5
, , , (. ! -DA1 radiation shield .-3SUS shroud wlthout Liq.N, USUS shroud with Liq.N,
,o-‘0
200
400
600
600
I
1000
-
1 DO
Temperature (“C) Fig. 2. Temperature
dependence on pressure in a vacuum system with respect to the introducing of the aluminum radiation shield. the shroud with liquid nitrogen. and the shroud without liquid nitrogen.
suppress the desorption. For the aluminum radiation shield, the pressure also increases with increasing the temperature. However, the increasing rate is smaller than that for the shroud with the liquid nitrogen in the range from 500°C to 900°C. This indicates that the aluminum radiation shield is promising method to suppress the increase of the outgassing at high temperature. For the shroud without the liquid nitrogen, the outgassing is greater than that of the other two cases under the whole temperature region. This is because stainless steel is easily be heated up compared with aluminum.
3. UHV system
Heater
Heater
%
1 Radiation Shield (Pure Aluminum) Liquid Nitrogen Shroud (SUS) (a)
(W
Fig. 1. Schematic diagram of the experimental chambers for equipped aluminum radiation shield (a) and equipped liquid nitrogen shroud (b). Each chamber is evacuated by ion pump with a pumping speed of 500 1 s- ’
Fig. 3 shows the UHV system newly developed for the planarization of 6 inch Si substrate. This system consists of three chambers, i.e., a cleaning, a preheating, and a load-lock. Aluminum radiation shields made from mechanochemical polished pure aluminum plates are placed both in the cleaning and the preheating chambers, and pipes for water cooling are mounted on the aluminum radiation shields. This system is evacuated by turbomolecular pumps, ion pump, and titanium sublimation pumps. An extractor gauge and a B-A gauge were installed on the cleaning and on the preheating chambers, respectively. A base pressure of 3 X lo-” Pa was achieved in the cleaning chamber. Tantalum resistive heaters were introduced for the 6 inch substrate heating from the
K. Idota et al./Applied
Surfucr Science lOO/ 101 (14%~ 311-315
4. Planarization
B
313
process for 6 inch substrate
2 P)
T 5 I=
I
Preheating ch. .#;OOX600
‘\
IP I (10001/s)
EHEED-Gufl
\
L-l
Aluminum Radiation Shield
Fig. 3. UHV system for planarization
of the 6 inch Si substrate.
back side of the substrate. Temperature on Si substrate surfaces was measured by thermocouples and pyrometer. A reflection high energy electron diffraction (RHEED) was equipped in the cleaning chamber. Fig. 4 illustrates the temperature dependence on pressure when the 6 inch Si substrate is heated by tantalum resistive heater in the cleaning chamber. The pressure increased abruptly at above 700°C with increasing the temperature. However, the substrate was heated up to 700°C under a pressure of less than 5 X lop8 Pa.
10-B 0
’
= 200
400
600
Temperature Fig. 4. Temperature dependence substrate is heated by tantalum chamber.
800
P-type, 6 inch Si(OO1) substrates (p = 5-10 0 cm) with a misorientation angle of less than $0.5” were used in this experiment. At first, the substrates were cleaned by a ‘modified RCA cleaning (m-RCA cleaning)‘. The m-RCA cleaning consists of three steps similar to the standard RCA cleaning [lo]. NH,OH/H,O* ratio is decreased to less than 0.2 in the first step. As the second step, the chemical oxide layer on the surface was removed in a diluted HF solution. In order to form a very thin oxide film on the substrate, the final step treatment was performed in a HCl:H,O,:H,O (1: 1:4) solution for 2 min. The m-RCA cleaning was carried out at room temperature. After this m-RCA cleaning, the substrates were introduced into the system and heated in the cleaning chamber. Si surface structure was observed by the RHEED with an acceleration voltage of 29 kV. Incident direction of primary electron beam was (1 IO).
1000
(‘C)
on pressure when the 6 inch Si resistive heater in the cleaning
Fig. 5. RHEED patterns from the 6 inch Si(OO1) substrate after the modified RCA cleaning. (a) Before heating, (b) heating at 600°C for 90 h in UHV.
314
K. Idoturtal./AppliedSu~acrScirncr
lOO/lOl
(19961311-315
Fig. 6. XTEM micrographs of the 6 inch substrates after the modified RCA cleaning (a) and after the UHV-heating The hatched area (left side) shows the planarized area with rms roughness of less than 0. I nm.
Fig. 5(a) shows RHEED pattern from the Si surface prior to heating after the m-RCA cleaning. On the other, Fig. 5(b) shows the case when heated at 600°C under a pressure of less than 5 X 10m8 Pa after the m-RCA cleaning. In Fig. 5(a), RHEED pattern indicates 1 X 1 structure. The surface structure began to convert from 1 X 1 to 2 X 1 at 585°C in UHV, although the 2 X 1 pattern was indistinct. Then, the 2 X 1 RHEED pattern was becoming clear during heating until 600°C. Finally, the RHEED pattern revealed clear 2 X 1 reconstructed structure as shown in Fig. 5(b). The 2 X 1 pattern with sharp streaks indicates the oxide free reconstructed clean surface. Surface contaminated structures like SIC were not observed in Fig. 5(b). Si surface roughness was observed by cross-sectional transmission electron microscopy (XTEM), and root-mean-square (rrns) roughness was evaluated by digitizing the lattice site on the Si(OO1) surface. Fig. 6(a) and (b) show the XTEM micrographs of the 6 inch substrate after the m-RCA cleaning and after the UHV-heating at 600°C for 90 h, respectively. Typical micrographs selected from numerous XTEM observations are shown in this figure. As for
at 600°C for 90 h (b).
surface roughness, apparent improvement was found for the substrate heated in UHV in comparison with the m-RCA cleaned substrate. The rms roughness in Fig. 6(b) was less than 0.1 nm. These results indicate that atomic-scale planarization is caused by the UHV-heating at 600°C for 90 h. UHV-heating at 600°C might cause the extremely slow migration. The planarized area (hatched) where the rms roughness was at most 0.1 nm are shown in the left side of this figure. In addition, no dislocations were generated on the surface at all. From these considerations, it was clarified that the atomic-scale planarization was attained within a whole practical area generally used for device fabrication.
5. Conclusion A newly developed UHV system equipped with aluminum radiation shields was demonstrated. By means of this system, a 6 inch substrate was heated at above 700°C under the pressure of the order of lo-’ Pa.
K. Idota et al./Applied
Sur$ace Science lOO/ 101 (1996) 311-315
Combined with the modified RCA cleaning, heating in this system provides an atomic planarization of the large diameter Si substrates for practical use. This atomically flat surface (rms < 0.1 nm) was realized at relatively low temperature (600°C) under a pressure of less than 5 X lo-’ Pa.
Acknowledgements This work was performed under the management of FED as a part of the MIT1 R&D program (Quantum Functional Devices project) supported by NEDO. The authors greatly appreciate ElKO Engineering Co., Ltd. throughout the generous support of the fabrication of the UHV system.
315
References [l] N.D. Arora and G.S. Gildenblad, IEEE Trans. Electron Devices ED-34 (19871 89. [2] S. Takagi, M. Iwase and A. Toriumi, in: Ext. Abstr. 22nd Int. Conf. on Solid State Devices and Materials, Sendai, 1990, p. 275. [3] T. Yasaka. K. Kanda. K. Sawara, S. Miyazaki and M. Hirose, Jpn. J. Appl. Phys. 30 (1991) 3567. [4] T. Ohmi, M. Miyashita, M. Itano, T. Imaoka and I. Kawanabe. IEEE Trans. Electron Devices ED-39 (1992) 537. [5] G.S. Higashi, R.S. Becher, Y.J. Chabal and A.J. Becher, Appl. Phys. Lett. 58 (1991) 1656. [6] S. Watanabe, N. Nakayama and T. Ito, Appl. Phys. Lett. 59 (1991) 1458. [7] P. Jakob. P. Dumas and Y.J. Chabal, Appl. Phys. Lett. 59 (1991) 2968. [8] M. Niwa. M. Udagawa, K. Okada, T. Kouzaki and R. Sinclair. Appl. Phys. Lett. 63 (1993) 675. [9] M. Niwa, T. Kouzaki, K. Okada, M. Udagawa and R. Sinclair, Jpn. J. Appl. Phys. 33 (1994) 388. [lo] W. Kern and D.A. Puotinen. RCA Rev. 31 (1970) 187.