Role of hydrogen during rapid vapor-phase doping analyzed by x-ray photoelectron spectroscopy and Fourier-transform infrared-attenuated total reection

Yukihiro Kiyotaa) and Fumiko Yano

Central Research Laboratory, Hitachi Ltd., 1-280 Higashi-Koigakubo, Kokubunji, Tokyo, Japan

Seiji Suzuki and Taroh Inada

College of Engineering, Hosei University, Kajino-cho, Koganei, Japan

Received 12 March 1997; accepted 3 October 1997 The surface of boron-doped layers formed by rapid vapor-phase doping was analyzed by x-ray photoelectron spectroscopy XPS and Fourier-transform infrared-attenuated total reection FTIR-ATR, to determine the role of the hydrogen carrier gas. Boron doping was carried out with a B2H6 source gas and a hydrogen carrier gas at 800 and 900 C. A nitrogen carrier gas was also used for comparison. Using hydrogen carrier gas, no evidence of boron segregation was observed in the XPS spectra. FTIR-ATR analysis conrmed that the hydrogen termination of the surface was maintained during doping. Using nitrogen carrier gas, layers that included segregated boron and silicon nitride were produced on the surface, which led to poor controllability of the boron concentration. When a hydrogen carrier gas is used, the hydrogen termination should promote the surface migration of adsorbed species. The hydrogen carrier gas plays an important role in terminating the silicon dangling bonds, thus preventing excessive chemisorption of boron. 1998 American Vacuum Society. S0734-21019803401-0

I. INTRODUCTION As the performance of high-speed silicon devices continues to advance, interest in developing processes for ultrashallow pn junctions with a depth of less than 50 nm has been increasing. For example, source and drain extensions of 0.1 m metaloxidesemiconductor transistors should have a diffusion depth of 20 nm and a surface carrier concentration of 1020 cm3. It is especially difcult to achieve such shallow and high-concentration doping through ion implantation of boron because of the channeling effect of the boron ions. Hence, a large amount of research has focused on nding a process to form shallow p -type layers.13 Among them, ultrahigh vacuum chemical vapor deposition CVD has an excellent potential for shallow boron doping with precise control of the concentration.4 To meet the requirements of shallow junctions for future large-scale integrated circuits, we have developed rapid vapor-phase doping RVD and have applied it to device fabrication.5,6 RVD is a kind of gas-source diffusion in which a hydrogen carrier gas is used. Most other methods that use B2H6 as a diffusion source for boron doping have some drawbacks such as boron segregation, SiB6 formation, and poor controllability of the concentration.7,8 In RVD, however, the boron concentration can be controlled over a wide range, and by choosing suitable doping conditions, the segregation of boron can be prevented.9,10 The difference between RVD and other methods is the use of a hydrogen carrier gas. A hydrogen carrier gas is usually used in silicon epitaxial growth because it cleans the silicon surface, that is, it removes native oxides. Aoyama et al. have reported the
a

results of a surface analysis after hydrogen annealing and claried that the reconstructed surface was covered with adsorbed hydrogen.11 In this article, to clarify the role of hydrogen during RVD, we characterized the shallow boron-doped layers using secondary ion mass spectroscopy SIMS, x-ray photoelectron spectroscopy XPS, and Fourier-transform infrared spectroscopy-attenuated total reection FTIR-ATR. II. EXPERIMENT The doping experiments were carried out using equipment that included a lamp annealing system5 as shown in Fig. 1. CZ100, n -type 10 cm wafers were used in our experiments. The off angle from the 100 plane was 4. Before doping, each wafer was treated by a conventional method that involved chemical cleaning, HF dipping, and rinsing with deionized water. The doping sequence was reported previously.6 The most notable feature of the RVD process is that B2H6 gas is introduced into the reaction chamber with a hydrogen carrier gas at room temperature, then, a wafer is rapidly heated for the doping. The doping conditions are summarized in Table I. After the B2H6 injection is stopped, 1 min postannealing was done to reduce the boron adsorption during cooling. To clarify the role of hydrogen in this experiment, a nitrogen carrier gas was also used. The oxygen concentration in the chamber was reduced to less than 0.1%, which was the detection limit of the oxygen detector attached to the reaction chamber, both for hydrogen- and nitrogencarried gas to prevent the oxidation of the sample during doping. Boron-doped layers were analyzed by SIMS Atomika 3000, XPS Shimadzu Kratos XSAM 800, and FTIR-ATR Nicolet model 730. The primary ion for SIMS analysis was 12 keV O2. The x-ray source for XPS analysis
1998 American Vacuum Society 1

Electronic mail: kiyota@crl.hitachi.co.jp J. Vac. Sci. Technol. A 161, Jan/Feb 1998

0734-2101/98/161/1/5/$10.00

Kiyota et al.: Role of hydrogen during rapid vaporphase doping

FIG. 1. Experimental setup.

was 12 keV Mg K, and the detection angle for the photoelectrons was 30. In FTIR-ATR analysis, a germanium prism attached to the samples was used as a multiple internal reections element and the spectra were obtained with p -polarized incident light in a N2 ambient. The FTIR analysis was done twice, just after doping and after HF treatment of the doped samples. III. EXPERIMENTAL RESULTS
A. SIMS analysis
FIG. 2. SIMS proles of boron; a hydrogen carrier case; b nitrogen carrier case.

The boron proles from the SIMS analysis are shown in Fig. 2. These samples were doped under the same conditions except for the carrier gas, so the differences in the spectra show the different effects of the carrier gases. The oxygen level in the doped layers was analyzed using cesium primary ion to conrm the level was identical to the bulk silicon and we could ignore the effect of oxidation-enhanced diffusion. When the hydrogen carrier gas was used, the diffusion depths were less than 50 nm and the surface boron concentrations were between 1019 and 1021 cm3. The surface boron concentrations increased at the lower doping temperature. When the nitrogen carrier gas was used, however, the surface boron concentrations exceeded 1022 cm3 regardless of the doping temperature. These concentrations exceeded the solid solubility and were close to the concentration of silicon itself. This suggests that the surface of the silicon was covered with distinct layer which includes boron wtih high concentration. From the SIMS proles, the sheet boron concentrations in the doped layers were calculated and shown in Fig. 3. When the hydrogen carrier gas was used, the sheet concentration decreased as the temperature increased. This indicates as inTABLE I. Doping conditions. H2 or N2 B2H6 /H2 0.1% Doping temperature Doping pressure Doping time Postannealing 25 l /min 100 ml /min 800 C, 900 C Atmospheric 120 s 60 s

creased probability of re-evaporation and outdiffusion for the adsorbed species, that is, the activation energies for these reactions are high. In this case, boron hydrides generated by the thermal dissociation of the B2H6 gas physically adsorbed on the surface. On the other hand, when the nitrogen carrier was used, the sheet concentrations exceeded the hydrogen carrier case, and did not change with temperature. This result indicates a low probability of re-evaporation for the adsorbed species. In this case, the species chemically adsorbed on the surface.
B. XPS analysis

The XPS spectra for Si (2 p ), B (1 s ), and N (1 s ) are shown in Fig. 4. Only the results for the samples doped at

FIG. 3. Sheet boron concentration calculated from SIMS data. J. Vac. Sci. Technol. A, Vol. 16, No. 1, Jan/Feb 1998

Kiyota et al.: Role of hydrogen during rapid vaporphase doping

FIG. 5. FTIR-ATR spectra measured just after doping.

eV,13 which is close to the peak in our data. When the nitrogen was used, the peak intensity for the adsorbed nitrogen at 401 eV is higher than that of the hydrogen carrier case. This is due to the absence of the terminated hydrogen as described later. From these results, we conclude that the surface of the samples doped in nitrogen is covered with silicon nitride that includes boron. Boron on the surface may be bonding with silicon or existing alone, but the details on the bonding conguration are not yet clear.
C. FTIR-ATR analysis
FIG. 4. XPS spectra for Si (2 p ), B (1 s ), and N (1 s ) peaks.

800 C are shown. For the samples doped at 900 C, almost identical spectra were obtained. Before the XPS analysis, the samples were dipped in HF and rinsed in deionized water to reduce the effect of adsorbed species from the air. When a hydrogen carrier gas was used, the B (1 s ) peak at 188 eV was very small, showing that the boron concentration at the surface was below the detection limit of XPS which is about 3% of silicon. The Si (2 p ) peak at 99 eV, which is almost identical to that of the Si substrate can be seen. The peak around 401 eV is due to the nitrogen that adsorbed on the surface from the air. The detection limit of nitrogen is 0.5%. From these spectra, we can see the surface conditions for the samples doped in hydrogen were almost identical to the usual boron-doped silicon. On the other hand, in the case of nitrogen, a B (1 s ) peak at 188 eV appears. Since the B (1 s ) peaks from boron nitride or boro-silicated glass appears at 191 and 194 eV, respectively,12 the peak at 188 eV must be due to either boron that bonded with silicon or metallic boron that segregated on the surface. At the higher energy shoulder of the Si (2 p ) peak, we can see another chemically shifted peak at 102 eV. Also in the N (1 s ) spectra, we can see another peak at 398 eV as well as the peak at 401 eV which is due to the adsorbed nitrogen. The N (1 s ) peak from silicon nitride has been reported to be between 397 and 398
JVST A - Vacuum, Surfaces, and Films

The FTIR-ATR spectra that we measured just after the doping are shown in Fig. 5. Due to the absorption by free carriers, broad peak at around 900 cm1 can be seen in all spectra. The peaks at between 2900 and 3000 cm1 are due to carbon contamination from the air. In the spectra of the sample doped in hydrogen, there is a sharp peak due to the SiHx bond at 2099 cm1. This peak is due to either the SiH bond on the Si (2 1) dimer11 or the SiH2 bond.14 The existence of the peak related to the SiHx bond indicates that the hydrogen termination of the dangling bonds of silicon was maintained during doping, as it is in epitaxial growth,11 even though the processing temperature was much higher than that at which terminated hydrogen desorbs. Aoyama et al. have concluded that the hydrogen termination that occurred after annealing in hydrogen was due to the adsorption of the species generated by the dissociation of H2 in the atmosphere.11 On the other hand, the spectra of the sample doped in nitrogen show no peak due to the SiHx bond, which indicates that the terminated hydrogen had desorbed. Instead, a BO peak was observed. The BO peak is due to the oxidation of segregated boron at the surface which occurred while the samples were being carried to the FTIR spectrometer. As shown in the XPS spectra, there should be a silicon nitride layer on the surface of the samples doped in nitrogen. However in the FTIR spectra, SiN peaks between 800 and 1000 cm1 cannot be seen because they are overlapped by the absorption by free carriers. A small BH peak was seen at 2540 cm1 regardless of the carrier gas when the

Kiyota et al.: Role of hydrogen during rapid vaporphase doping

FIG. 7. Doping reaction model for the hydrogen carrier case.

2 Reection SiHBH3SiHBH3, 3 Migration SiHBH3SiHSiHSiHBH3,

FIG. 6. FTIR-ATR spectra measured after HF dip.

4 Recombination SiHBH3SiHBH32Si-HB2H6, 5 Chemisorption SiBH3SiBHH2

doping temperature was 800 C. This suggests that B2H6 gas does not completely dissociate at this temperature and the precursors adsorb on the silicon surface as a form of boron hydrides. The doped samples were dipped in 0.5% HF and rinsed with deionized water, then they were analyzed by FTIR-ATR again. At that time, the samples doped in hydrogen repelled the water, while those doped in nitrogen did not. The changes in the FTIR spectra caused by the 0.5% HF treatment of the samples doped in hydrogen are shown in Fig. 6. After HF treatment, a SiH2 peak at 2110 cm1 and a SiH3 peak at 2140 cm1 appeared instead of the sharp peak at 2099 cm1. The wavenumbers of these peaks are identical to those for the sample before doping. This indicates the existence of Si atoms at the surface without boron segregation when the hydrogen carrier gas was used. When the nitrogen carrier gas was used, there were no SiHx peaks even after HF treatment, although the BO peak decreased. This also suggests that boron segregation or silicon nitride formation occurred when the nitrogen carrier gas was used. IV. DISCUSSION As mentioned above, when a hydrogen carrier gas is used, hydrogen termination of Si dangling bonds is maintained during doping and boron segregation is suppressed. Considering these results, the main surface reactions that occur during doping in hydrogen can be described as follows. This model is based on the surface reactions during CVD of a -Si proposed by Matsuda.15 Here, SiH stands for a silicon atom at the surface whose dangling bonds are terminated by hydrogen, and Si stands for a silicon atom with dangling bonds. BH3 molecules are assumed to be the precursors generated by thermal dissociation of B2H6 gas.16 Reactions between the adsorbed species and hydrogen in the atmosphere are ignored. 1 Physisorption SiHBH3SiHBH3,
J. Vac. Sci. Technol. A, Vol. 16, No. 1, Jan/Feb 1998

or SiBH3SiB3H, 6 Diffusion SiBHSiHBdiffused or SiBSiBdiffused. This reaction model is shown schematically in Fig. 7. Hydrogen which terminates the dangling bonds of silicon reduces the reactivity of the surface. Thus, BH3 precursors can physisorb and migrate on the surface. When the BH3 precursors meet the small number of silicon dangling bonds, they chemisorb and make SiB bonds. Yu et al. have reported using XPS and time desorption spectroscopy that thermal desorption of hydrogen caused no loss of boron on the surface, and B2H6 adsorption was promoted by the desorption of hydrogen.17 In our experiments, the same reactions appear to have occurred between BH3 and the hydrogen-terminated surface of silicon. A hydrogen carrier gas probably plays an important role in terminating the dangling bonds, thus promoting the surface migration of adsorbed species and preventing excessive chemisorption of boron. We have reported the controllability of the surface boron concentration between 1018 and 1021 cm3 in this process using hydrogen carrier gas.8,9 The above-mentioned effects of hydrogen make it possible to control the boron concentration of the doped layers. On the other hand, in the nitrogen atmosphere, the adsorbed BH3 precursors can easily nd the silicon dangling bonds because there are no SiH bonds as shown in Fig. 5. Thus the precursors chemically adsorb which makes the probability for re-evaporation low. Since the boron chemisorption on silicon is not self-limiting,17 a surplus of dangling bonds causes continuous chemisorption and thus boron segregation. V. CONCLUSIONS To determine the role of the hydrogen carrier gas, the surface of boron-doped layers formed by rapid vapor-phase

Kiyota et al.: Role of hydrogen during rapid vaporphase doping

2

doping were analyzed by XPS and FTIR-ATR. Using hydrogen carrier gas, boron segregation was not observed in the XPS spectra. The FTIR-ATR analysis showed that the hydrogen termination of the surface was maintained during doping. Using nitrogen carrier gas, however, layers that included segregated boron and silicon nitride were produced on the surface which led to the poor controllability of the boron concentration. We proposed a doping model with a hydrogen carrier gas, in which hydrogen termination promotes the surface migration of adsorbed species. The hydrogen carrier gas plays an important role in terminating the dangling bonds, thus preventing excessive chemisorption of boron. These effects make it possible to control the boron concentration of the doped layers. ACKNOWLEDGMENT We express great thanks to Toshihiko Itoga and Toshiyuki Mine for their valuable discussion on the results of our FTIR-ATR analysis. The authors also thank Masaru Matsushima and Yutaka Kaneko who contributed to the design of the doping chamber.
1

M. Saito, T. Yoshitomi, H. Hara, M. Ono, Y. Akasaka, H. Nii, S. Matsuda, H. S. Momose, Y. Katsumata, Y. Ushiku, and H. Iwai, IEEE Trans. Electron Devices ED-40, 2264 1993.

J. Nishizawa, K. Aoki, and T. Akamine, Appl. Phys. Lett. 56, 1334 1990. 3 H. Kujirai, E. Murakami, and S. Kimura, Jpn. J. Appl. Phys., Part 1 34, 782 1995. 4 B. S. Meyerson, F. K. LeGoues, T. N. Nguyen, and D. L. Harame, Appl. Phys. Lett. 50, 113 1987. 5 Y. Kiyota, M. Matsushima, Y. Kaneko, M. Kanetomo, and Y. Tamaki, Appl. Phys. Lett. 64, 910 1994. 6 Y. Kiyota, T. Nakamura, S. Suzuki, and T. Inada, IEICE Trans. Electron. E79-C, 554 1996. 7 Y. W. Hsieh, J. Electrochem. Soc. 117, 807 1970. 8 N. Saito, T. Akamine, K. Aoki, and Y. Kojima, Jpn. J. Appl. Phys., Part 1 32, 4404 1993. 9 Y. Kiyota, T. Onai, T. Nakamura, T. Inada, A. Kuranouchi, and Y. Hirano, IEEE Trans. Electron Devices ED-39, 2077 1992. 10 Y. Kiyota, T. Nakamura, T. Inada, A. Kuranouchi, and Y. Hirano, J. Electrochem. Soc. 140, 1117 1993. 11 T. Aoyama, K. Goto, T. Yamazaki, and T. Ito, J. Vac. Sci. Technol. A 14, 2909 1996. 12 T. Kazahaya, and M. Hirose, Jpn. J. Appl. Phys., Part 2 25, L75 1986. 13 Handbook of X-ray Photoelectron Spectroscopy Perkin-Elmer, Eden Prairie, MN, 1992. 14 M. Hirose, M. Hiroshima, T. Yasaka, and S. Miyazaki, J. Vac. Sci. Technol. A 12, 1864 1994. 15 A. Matsuda, K. Nomoto, Y. Takeuchi, A. Suzuki, A. Yuuki, and J. Perrin, Surf. Sci. 227, 50 1990. 16 S. Komatsu, M. Kasamatsu, K. Yamada, and Y. Moriyoshi, J. Appl. Phys. 71, 5654 1992. 17 M. L. Yu, D. J. Vitkavage, and B. S. Meyerson, J. Appl. Phys. 59, 4032 1992.

JVST A - Vacuum, Surfaces, and Films