A Comparison of Transient Boron Diffusion in Silicon, Silicon Carbide and Diamond

Share Embed


Descripción

Materials Science Forum Vols. 600-603 (2009) pp 453-456 online at http://www.scientific.net © (2009) Trans Tech Publications, Switzerland

A Comparison of Transient Boron Diffusion in Silicon, Silicon Carbide and Diamond M.K. Linnarsson1a, J. Isberg2b, A. Schöner3c, and A. Hallén1d 1

Royal Institute of Technology, Lab of Materials and Semiconductor Physics, P.O. Box Electrum 229, SE-164 40 Kista, Sweden

2

Uppsala University, Department of Engineering Science, Box 534, SE-751 21, Uppsala, Sweden 3

Acreo AB, P.O. Box Electrum 236, SE-164 40 Kista, Sweden

a

[email protected], [email protected], [email protected], [email protected]

Keywords: Boron, SIMS, Diffusion, 4H-SiC, Diamond

Abstract. The boron diffusion in three kinds of group IV semiconductors: silicon, silicon carbide and synthetic diamond has been studied by secondary ion mass spectrometry. Ion implantation of 300 keV, 11B-ions to a dose of 2×1014 cm-2 has been performed. The samples are subsequently annealed at temperatures ranging from 800 to 1650 °C for 5 minutes up to 8 hours. In silicon and silicon carbide, the boron diffusion is attributed to a transient process and the level of out-diffusion is correlated to intrinsic carrier concentration. No transient, out-diffused, boron tail is revealed in diamond at these temperatures. Introduction An increasing interest in dopant diffusion in semiconductors has emerged driven by smaller sizes and more demanding device applications. For example, it is difficult to form shallow boron junctions in silicon by ion implantation due to an enhanced out-diffusion ("out-diffusion" refers to out diffusion of boron from the implanted region, deeper into the sample). The level of out diffused boron is close to the intrinsic carrier concentration at temperatures ≤ 800 °C and approaches the solubility limit at higher diffusion temperatures, i.e. ≥850 °C [1]. In 1990, a model for the enhanced boron diffusion in implanted silicon was proposed by Fair [1]. In this model, the enhanced diffusivity is correlated to the point defects generated from the annealing of implanted damage. According to this model the diffusion follows a kick-out mechanism, predominately assisted by neutral self-interstitials below 800 °C, while donor self-interstitials becomes dominant above 800 °C. A transient, boron out diffusion is also revealed in “intrinsic” silicon carbide [2, 3]. However, the diffusion mechanism for B in SiC is not yet established. According to electrical measurements, the main part of the boron in the out-diffused tail is not on substitutional Si sites [4]. Hence, the ideas from transient boron diffusion in silicon can not be transferred to silicon carbide without modifications. For single crystalline diamond, the other constituent of silicon carbide, experimental diffusion data is limited [5]. In this investigation we have compared out-diffusion of implanted boron in silicon, silicon carbide and diamond. Secondary ion mass spectrometry (SIMS) was used to determine the depth distribution of boron for a range of diffusion times and temperatures. Experimental Three kinds of single crystalline semiconductors have been used, silicon, 4H-silicon carbide and CVD (chemical vapor deposition) diamond. All samples are n-type with doping levels below 1×1015 cm-3. A boron diffusion source has been introduced into the samples by ion implantation with 300 keV 11B ions to a dose of 2×1014 cm-2. Subsequent heat treatments were then carried out. The samples were heat treated between 800 °C and 1650 °C for 5 minutes up to 8 hours. The annealing All rights reserved. No part of contents of this paper may be reproduced or transmitted in any form or by any means without the written permission of the publisher: Trans Tech Publications Ltd, Switzerland, www.ttp.net. (ID: 130.238.198.20-26/08/08,09:00:37)

454

Silicon Carbide and Related Materials 2007

temperatures were chosen to guaranty intrinsic carrier conditions under the treatment (i.e. outside the implanted area the intrinsic carrier concentration, at the used annealing temperatures, is well above the dopant concentration as well as the boron detection limit in SIMS). Except for diamond, where the crystal stability limits the heat treatment to 1650°C. An RF-furnace was used for heat treatments above 1200 °C and a vacuum furnace was employed below 1200 °C. Our Cameca 4f instrument has been utilized for secondary ion mass spectrometry (SIMS) to obtain the depth distribution of 11B. Results and discussion Fig. 1a-c displays the boron out-diffusion from an ion implanted distribution made in three types of single crystalline semiconductors; silicon, 4H-SiC and diamond. The silicon sample (Fig. 1a) has been annealed at 800 °C for 30 minutes, 2 h and 8 h. As the time increases the diffusivity decreases and only a minor change in the boron distribution is revealed between the 2 h and the 8 h anneals. This means that the diffusion process at 800 ºC has ceased within the first 8 h and the migration can be attributed to a transient process. The solubility limit of boron in silicon at 800 °C (~ 2×1019 cm3 , from ref [6]) is not exceeded in any part of the implanted boron profile. During heat treatment at 800 °C, enhanced diffusion occurs at a doping concentration below the intrinsic carrier concentration, 2×1018 cm-3. Our results are in agreement with previous published data supporting the kick-out mechanism for enhanced boron diffusion in silicon [see for example ref. 1, 7 and references therein]. In Fig. 1b, an implanted 4H-SiC sample is shown before and after heat treatment at 1300 °C for 5 minutes. No pronounced change is seen in the out-diffusion tail, if the annealing time is increased (not shown in the figure). The migration has a transient characteristic where the diffusion process ceased within a few minutes. Hence, the diffusion process can not be described by a “classic” diffusion process with Gaussian (i.e. thin film solution), or erfc (i.e. infinite source) solutions. The shape of the diffusion curve at 1300 °C is almost exponential and the out-diffusion occurs at a doping concentration below the intrinsic carrier concentration. Apart from the boron level of out diffusion (close to the intrinsic carrier concentration), the shape of diffusion curves differ between silicon and silicon carbide. At 1300 °C, the solubility limit of for boron in 4H-SiC is 3×1018 cm-3 (extracted from ref. 8). Hence, the solubility is exceeded in the boron implantation peak but the level of out-diffusion is 100 times smaller when the boron solubility. The same type of out-diffusion tail is seen if the dose is decreased not to exceed the solubility limit (not shown in the figure). In analogy with silicon, the level of out-diffusion is below the intrinsic carrier concentration. However, the picture is more complex in silicon carbide compared to silicon due to the presence of more kinds of point defects. According to theoretical analysis, the vacancy related diffusion in SiC is hindered by high activation energies and interstitial-mediated boron diffusion has been suggested [9]. If the diffusion in silicon carbide follows the same route as in silicon, the boron in the tail region should occupy the silicon sublattice and be electrical active. However, electrical measurements by Negoro et al. [4] shows a depression in acceptor profile compared to the total boron concentration in the first 0.2 µm of the tail region. Hence, electrical measurement indicates that the model from silicon has to be modified and that the carbon self-interstitials and/or vacancies can not be ignored. It is also interesting to note that the presence of ~1019 cm-3 carbon in silicon suppress transient enhanced boron diffusion by eliminating the source [7]. Fig. 1c shows a boron implanted, single crystalline, diamond sample, which has been heat treated at 1650 °C for 1 h. No diffusion of boron is revealed, except from a small pile up of boron at the surface. According to the carbon phase diagram, graphite is the stable phase at any temperature at atmospheric pressure. However, the threshold for transformation is very high and the diamond

Materials Science Forum Vols. 600-603

a)

Fig.1. SIMS measurement of 11 B versus depth for asimplanted and annealed samples. The samples are implanted with 300 keV 11B+ ions, dose 2×1014 cm-2, in ntype (1×1015 cm-3) silicon (a), 4H-SiC (b), and diamond (c).

o

Solubility limit at 800 C

19

10 -3

B Concentration (cm )

Intrinsic carrier concentration 18

10

As imp. o

800 C/30m o

800 C/2h

17

10

o

800 C/8h

16

11

10

15

10

Silicon

14

10

0

0.5

1 1.5 Depth (µm)

2

b)

2.5

10

o

1300 C/5m o

Solubility limit at 1300 C -3

(a) The silicon sample has been annealed at 800 °C for 30 minutes, 2 and 4 hours. The solubility limit and the intrinsic carrier concentration in Si at 800 °C are indicated in the figure. (b) For 4H-SiC a heat treatment is performed at 1300 °C during 5 minutes. The solubility limit and the intrinsic carrier concentration in 4H-SiC at 1300 °C have been included.

As imp

19

B Concentration (cm )

455

18

10

17

10

Intrinsic carrier concentration 16

11

10

15

10

4H-SiC

14

10

0

1

2 3 Depth (µm)

10

10

18

-3

B Concentration (cm )

c) 19

4

5

(c) The diamond sample has been annealed at 1650 °C for 1 hour. The intrinsic carrier concentration in diamond at 1650 °C is indicated in the graph.

Diamond

As imp. o

1650 C/1h 10

17

10

16

11

Intrinsic carrier concentration

10

15

10

14

0

0.5

1 1.5 Depth (µm)

2

2.5

456

Silicon Carbide and Related Materials 2007

structure remains after the heat treatment at 1650 °C for 1 h. At 1650 °C, the intrinsic carrier concentration is ~8×1014 cm-3, a value in the same range as the n-doping. Hence, a higher anneal temperature has been preferable in this study. Unfortunately, a further increase of the annealing temperature will destroy the diamond structure. Under the used experimental conditions, the excess of point defects introduced into diamond during ion implantation do not support enhanced diffusion, at least not below 1650 °C. According to previously reported data, pronounced boron diffusion has been observed in diamond after heat treatment at 1200 °C for 20 hours [5]. The divergence from our results may, however, be explained by different experimental conditions, or the crystal quality. Summary Ion implanted boron exhibits transient enhanced diffusion in silicon and silicon carbide samples at 800 °C and 1300 °C, respectively. At theses temperature the intrinsic concentration exceeds the dopant concentration in the out-diffused tail. No transient, out-diffused, boron tail is revealed in the diamond sample after heat treatment at 1650 °C. However, the intrinsic carrier concentration in diamond at this temperature is in the same range as the n-type doping level. Acknowledgment One of the authors wishes to acknowledge “Ångpanneföreningens forskningsstiftelse” for financial support. References [1] [2] [3] [4] [5] [6] [7] [8] [9]

R.B. Fair, J. Electrochem. Soc. 137 (1990) p.67. M.S. Janson, M.K. Linnarsson, A. Hallén et al., Appl. Phys. Lett. 76 (2000) p.1434. H. Bracht, N.A. Stolwijk, M. Laube and G. Pensl, Appl. Phys. Lett. 77 (2000) p.3188. Y. Negoro, T. Kimoto and H. Matsunami, J. Appl. Phys. 98 (2005) p.43709. T. Sung, G. Popovici, M.A. Prelas and R.G. Wilson, Mat.Res.Soc.Symp.Proc. 416 (1996) p.467. D.Nobili in Properties of silicon, edited by C.Hilsum, INSPEC, London (1988). P.A. Stolk, H.-J. Gossmann, D.J. Eaglesham et al., J. Appl. Phys. 81 (1997) p.6031. M.K. Linnarsson, M.S. Janson, N. Nordell et al., Appl. Surf. Sci. 252 (2006) p.5316. M. Bockstedte, A. Mattausch and O. Pankratov, Mat, Sci. Forum 483-485 (2005) p. 527.

Lihat lebih banyak...

Comentarios

Copyright © 2017 DATOSPDF Inc.