Transient enhanced diffusion of implanted boron in 4H-silicon carbide

Transient enhanced diffusion of implanted boron in 4H-silicon carbide M. S. Janson, M. K. Linnarsson, A. Hallén, B. G. Svensson, N. Nordell et al. Citation: Appl. Phys. Lett. 76, 1434 (2000); doi: 10.1063/1.126055 View online: http://dx.doi.org/10.1063/1.126055 View Table of Contents: http://apl.aip.org/resource/1/APPLAB/v76/i11 Published by the AIP Publishing LLC.

Additional information on Appl. Phys. Lett. Journal Homepage: http://apl.aip.org/ Journal Information: http://apl.aip.org/about/about_the_journal Top downloads: http://apl.aip.org/features/most_downloaded Information for Authors: http://apl.aip.org/authors

Downloaded 02 Oct 2013 to 202.116.1.149. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://apl.aip.org/about/rights_and_permissions

APPLIED PHYSICS LETTERS

VOLUME 76, NUMBER 11

13 MARCH 2000

Transient enhanced diffusion of implanted boron in 4H-silicon carbide M. S. Janson,a) M. K. Linnarsson, A. Halle´n, and B. G. Svensson Royal Institute of Technology, Department of Electronics, Electrum 229, S-164 40 Kista-Stockholm, Sweden

N. Nordellb) IMC, Electrum 233, S-164 40 Kista-Stockholm, Sweden

H. Bleichner ABB Corporate Research, Electrum 215, S-164 40 Kista-Stockholm, Sweden

共Received 5 October 1999; accepted for publication 20 January 2000兲 Experimental evidence is given for transient enhanced diffusion of boron 共B兲 in ion-implanted silicon carbide 共SiC兲. The implanted B is diffusing several ␮m into the samples when annealed at 1600 and 1700 °C for 10 min, but the in-diffused tails remain unaffected when the annealing times are increased to 30 min at the same temperatures. A lower limit of the effective B diffusivity at 1600 °C is determined to 7⫻10⫺12 cm2/s, which is 160 times larger than the equilibrium B diffusivity given in the literature. © 2000 American Institute of Physics. 关S0003-6951共00兲03111-9兴

project. They consisted of n-type 4H–SiC epitaxial layers, doped with nitrogen to an effective carrier concentration of 5⫻1015 cm⫺3, which were implanted with boron ( 11B⫹) and aluminum ( 27Al⫹). A multiple implantation scheme was used to obtain a B profile with an average concentration of 2⫻1018 cm⫺3 and a depth of 0.8 ␮m, while the implanted Al extended approximately 0.4 ␮m into the epi-layer. The implantations were preformed at an elevated temperature of 500 and 800 °C for B and Al, respectively. In addition to the implanted samples, an epitaxial structure was grown with a highly doped boron layer surrounded by undoped material. The B concentration in this layer was 8⫻1019 cm⫺3 and the width of the layer was 0.4 ␮m. The reactor used for epitaxy is described in detail elsewhere.16 The B and Al implanted samples were then heat treated at 1600 and 1700 °C for 10 and 30 min, and at 2050 °C for 60 min. The sample with the buried B layer was annealed at 1700 °C for 60 min. The 1600 and 1700 °C treatments were performed in a vapor phase epitaxy 共VPE兲 reactor with a controlled SiH4 flow while the 2050 °C anneal was done in a sublimation reactor. Acceptor atom concentrations as a function of sample depth were determined by secondary ion mass spectrometry 共SIMS兲 measurements utilizing a Cameca IMS 4f microanalyzer. A primary sputtering beam of 8 keV ( 16O) ⫹ 2 ions was rastered over an area of 200⫻200 ␮m2 and secondary ions of 11 ⫹ B and 27Al⫹ were collected from the central region of the sputtered crater. Figure 1 shows the SIMS profiles of the implanted samples before and after annealing. In the implanted region, close to the surface, the B concentration is strongly reduced in the annealed samples indicating an out-diffusion of B through the sample surface. This out-diffusion increases both as a function of the duration and, more prominently, of the temperature used for the anneal. Only 3% of the asimplanted B dose remains in the sample annealed at the highest temperature. However, in this study, we focus on the B diffusion into the undamaged part of the samples, beyond the as-implanted distribution. Starting at a concentration of 7⫻1016 cm⫺3, a distinct kink in the 1600 °C profiles indi-

In the early stages of SiC device processing, p-n junctions were formed by thermal diffusion of acceptor atoms, boron or aluminum, from the surface into n-type bulk material. This procedure is, however, impractical because of the very high temperatures 共⬎1700 °C兲 needed to diffuse sufficient amounts of the slowly migrating dopants into the samples.1,2 For instance, the high temperatures made standard SiO2 masking impossible. A far more attractive way to introduce doping atoms is the use of ion implantation, which also adds the advantage of well-controlled doping profiles. Unfortunately, the lattice damage caused by the implanted ions has proven difficult to anneal in SiC. The crystalline order as studied by Rutherford backscattering spectrometry is usually restored after a thermal anneal at 1500 °C,3,4 but defect agglomerates are still found by transmission electron microscopy5–7 and positron annihilation spectroscopy8 even after anneals at 1700 and 1500 °C, respectively. High temperature post-implantation anneals are also needed to position the implanted ions at substitutional, electrically active, lattice sites. In a study by Kimoto et al.,9 it was found that aluminum implanted to a dose of 1⫻1014 cm⫺2 required a 1600 °C furnace anneal to reach full 共⬎90%兲 activation while the corresponding temperature for boron was 1700 °C. As stated above, boron diffusion is not negligible at these high temperatures and several authors have recently reported both a strong out-diffusion of implanted boron atoms, as well as diffusion into the undamaged bulk region of the sample.4,6,10–14 The in-diffusion of B is expected to be of great importance for device performance since diffused B may compensate n-type doping and form almost intrinsic zones.15 In this letter, we show that the migration of implanted boron into the sample is both a transient and an enhanced diffusion process 共TED兲 at temperatures around 1700 °C. Diode test pieces were obtained from the ABB, SiC a兲

Electronic mail: [email protected] Present address: Royal Institute of Technology, Semiconductor Laboratory, Electrum 229, S-164 40 Kista, Sweden.

b兲

0003-6951/2000/76(11)/1434/3/$17.00

1434

© 2000 American Institute of Physics

Downloaded 02 Oct 2013 to 202.116.1.149. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://apl.aip.org/about/rights_and_permissions

Appl. Phys. Lett., Vol. 76, No. 11, 13 March 2000

Janson et al.

1435

FIG. 2. SIMS measurements of 11B concentration vs depth in a 4H–SiC epitaxial layer with a buried B layer 共solid line兲. The epitaxial structure was annealed at 1700 °C for 60 min 共open circles兲 and a Gaussian function was fitted to the lower concentration part of the diffusion tail 共broken line兲.

FIG. 1. SIMS measurements of 11B concentration vs depth in a 4H–SiC epitaxial layer, implanted with 11B⫹ and 27Al⫹ ions 共solid line兲. The samples were post-implant annealed at 共a兲 1600 and 共b兲 1700 °C for 10 and 30 min 共open and closed circles, respetively兲. An additional sample was annealed at 2050 °C for 60 min 共c, open circles兲. Gaussian functions has been fitted to the lower concentration part of the diffusion tails 共broken lines兲.

cates the beginning of an almost exponential tail extending at least 3 ␮m into sample. When comparing the profiles in the 10 and 30 min annealed samples, one finds that the two diffusion tails are close to identical. This means that the diffusion process responsible for the exponential tails has ceased within the first 10 min of the anneal and is not active at longer times. In other words, the diffusion into the sample at 1600 °C is governed by a transient process. The diffusion tails into the samples annealed at 1700 °C 关Fig. 1共b兲兴 exhibit the same transient behavior observed at 1600 °C, i.e., within the experimental accuracy, there is no change in the profile when going from 10 to 30 min. Furthermore, the shape of the curves at 1600 and 1700 °C are almost identical, but the absolute concentration values at 1700 °C are twice that at 1600 °C. In contrast, after the diffusion at 2050 °C, the distribution of boron is radically different compared to that at 1600 and 1700 °C 关Fig. 1共c兲兴. The diffused B extends considerably deeper into the epitaxial layer and the profile exhibits a concave shape, typical for a ‘‘classical’’ diffusion process as discussed below. Figure 2 shows the SIMS profiles of as-grown and annealed samples having a buried boron layer. The as-grown profile has the shape of a box with a maximum concentration

of 8⫻1019 cm⫺3 and a width of 0.4 ␮m. A broadening of the as-grown profile starts at 3⫻1016 cm⫺3 and extends 0.4 ␮m on each side of the box where it reaches the B background level in the epi-layer at 2⫻1015 cm⫺3. This broadening is most likely due to diffusion during the growth of the epitaxial layer (T growth⫽1620 °C) and the slightly larger broadening towards the surface can be explained by the memory effect from the highly doped layer.17 In the annealed sample, two diffusion tails develop symmetrically on each side of the B layer. The diffusion profiles have a similar, almost exponential, shape as observed in the implanted samples, but the penetration length of the diffused B atoms is now much smaller. The almost exponential diffusion tails of Figs. 1 and 2 clearly does not follow the distribution nor the temporal development of a standard, concentration independent Fickian diffusion process, i.e., the Gaussian or erfc solutions,18 but the experimental data presented here are too limited to make more elaborated modeling meaningful. However, in order to obtain a quantitative measure of the B diffusion observed in this experiment, a more phenomenological approach is applied. As a first approximation, the diffusion profiles of Figs. 1共a兲, 1共b兲, and Fig. 2 can be reasonably well described by a sum of two Gaussian functions which would then represent a low concentration, fast diffusing, and a high concentration, slow diffusing part of the diffusion profile, respectively. A physical interpretation of this division may be that the high concentration part of the diffusion profiles consists of B-related complexes that are stable at temperatures up to 1700 °C but dissociate at 2050 °C, where the high concentration part of the profile is absent 关Fig. 1共c兲兴. Considering only the faster diffusing, low concentration part, Gaussian functions are fitted to this section of the diffusion profiles in Figs. 1 and 2. Assuming a limited diffusion source in a 共semi-兲 infinite medium, the extracted values of the broadening, ␴, can now be used to associate effective diffusivities, B , to the low concentration part of the profiles. The relaD eff B , the diffusion time t and ␴ then writes: tion between D eff B 18 2 D eff⫽ ␴ /2t. The fitted Gaussian functions are plotted as broken lines in Figs. 1 and 2, and the corresponding values

Downloaded 02 Oct 2013 to 202.116.1.149. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://apl.aip.org/about/rights_and_permissions

1436

Janson et al.

Appl. Phys. Lett., Vol. 76, No. 11, 13 March 2000

B TABLE I. Comparison between effective diffusivities D eff obtained in this B B study and corresponding literature values D lit. D eff are calculated from ␴, extracted from the SIMS profiles in Figs. 1 and 2.

Sample

␴ 共␮m兲

B 共cm2/s) D eff

B (cm2/s) D lit

B B D eff /D lit

Implanted 1600 °C/10 min 1700 °C/10 min 2050 °C/60 min

0.90 1.3 4.3

⭓6.7⫻10⫺12 ⭓1.3⫻10⫺11 2.5⫻10⫺11

4.3⫻10⫺14 2.5⫻10⫺13 3.5⫻10⫺11

⭓160 ⭓50 0.7

Buried layer 1700 °C/60 min

0.38

2.0⫻10⫺13

2.5⫻10⫺13

0.8

B of ␴ and D eff are listed in Table I. It is clearly seen that the fitted functions poorly represent the diffusion profiles except for the implanted sample annealed at 2050 °C where the fit is good for the entire diffusion tail. It should also be noted that B values of the 1600 and 1700 °C for 10 min annealed the D eff samples are lower limit values since the corresponding 30 min annealed samples show that the diffusion process has stopped at some time before 10 min have elapsed. Atomic state diffusion in SiC is generally not a very well documented subject in the literature. However, some work regarding boron diffusion has been performed in the seventies and early eighties by Mokhov and Vodakov et al.19–21 A representative expression of the B diffusivity in their work using low-doped or n-type SiC is given in Ref. 20: D Blit⫽50 ⫻exp(⫺5.6 eV/k B T) cm2/s, which is valid in the temperature range of 1500–2550 °C. This equation is used here to B . compare with D eff B /D Blit ratios are listed in Table I and D Blit as well as the D eff B and D Blit for the imthe agreement is good between D eff planted 2050 °C annealed sample and the sample with the buried B epi-layer. This agreement is noteworthy considering the different experimental procedures and the crude asB . On the other hand, the sumptions made when extracting D eff extracted effective diffusivities for the implanted samples annealed at 1600 and 1700 °C are enhanced by at least a factor of 160 and 50, respectively, when compared to the literature data. The good agreement with D Blit for the buried layer sample means that the enhanced diffusivities observed in the implanted samples, annealed at the same or lower temperature, are related to the implantation process. In fact, a similar transient enhanced diffusion is well established in B-implanted crystalline silicon and attributed to an interaction of the B atoms with implantation-induced excess silicon self-interstitials through the interstitialcy or kick-out mechanism.22 The diffusion mechanism for B in SiC is not yet established and arguments have been given both for a vacancy20 and an interstitial type of mechanism.23 It is still an open question whether the reported transient enhanced B diffusion in SiC is due to an implantation induced excess of vacancies or self-interstitials.

In conclusion, we have shown that the bulk diffusion of implanted boron in 4H–SiC is both a transient and highly enhanced process for annealing temperatures of 1600 and 1700 °C. Comparison with diffusion at 1700 °C in a sample having a buried epitaxial B layer, where no enhanced migration was found, gives strong evidence for that the transient enhanced boron diffusion is related to the implantation process. This work was funded by the Swedish Foundation for Strategic Research and ABB Corporate Research. Ulf Linde˚ strand are acknowledged for their support of felt and Bo¨rje A this project. 1

H. C. Chang, C. Z. L. May, and L. F. Wallace, Silicon Carbide 1595 共Permagon, Oxford, 1960兲, pp. 496–507. 2 L. G. Griffiths, J. Appl. Phys. 36, 571 共1965兲. 3 T. Kimoto, A. Itoh, H. Matsunami, T. Nakata, and M. Watanabe, J. Electron. Mater. 25, 879 共1996兲. 4 M. V. Rao, J. A. Gardner, P. H. Chi, O. W. Holland, G. Kelner, J. Kretchmer, and M. Ghezzo, J. Appl. Phys. 81, 6635 共1997兲. 5 E. Glaser, A. Heft, J. Heindl, U. Kaiser, T. Baumann, W. Wesch, H. P. Strunk, and W. Wendler, Inst. Phys. Conf. Ser. 142, 557 共1996兲. 6 G. Pensl, V. V. Afanas’ev, M. Bassler, M. Schadt, T. Troffer, J. Heindl, H. P. Strunk, M. Maier, and W. J. Choyke, Inst. Phys. Conf. Ser. 142, 275 共1996兲. 7 ˚ . Persson, Q. Wahab, L. Hultman, N. Nordell, A. Scho¨ner, K. P. O. A Rottner, E. Olsson, and M. K. Linnarsson, Mater. Sci. Forum 264-268, 413 共1998兲. 8 G. Brauer, W. Anwand, P. G. Coleman, J. Stormer, F. Plazaola, J. M. Campillo, Y. Pacaud, and W. Skorupa, J. Phys.: Condens. Matter 10, 1147 共1998兲. 9 T. Kimoto, O. Takemura, H. Matsunami, T. Nakata, and M. Inoue, J. Electron. Mater. 27, 358 共1998兲. 10 T. Troffer, M. Schadt, T. Frank, H. Itoh, G. Pensl, J. Heindl, H. P. Strunk, and M. Maier, Phys. Status Solidi A 162, 277 共1997兲. 11 A. Scho¨ner, S. Karlsson, T. Schmitt, A. Halle´n, M. Frischholz, and K. Rottner, Mater. Res. Soc. Symp. Proc. 512, 469 共1998兲. 12 M. Gong, C. V. Reddy, C. D. Beling, S. Fung, G. Brauer, H. Wirth, and W. Skorupa, Appl. Phys. Lett. 72, 2739 共1998兲. 13 S. Seshadri, G. W. Eldridge, and A. K. Agarwal, Appl. Phys. Lett. 72, 2026 共1998兲. 14 M. Laube, G. Pensl, and A. Itoh, Appl. Phys. Lett. 74, 2292 共1999兲. 15 S. Ortolland, C. Raynaud, J. P. Cahnte, M. L. Locatelli, A. A. Lebedev, A. N. Andreev, N. S. Savkina, V. E. Chelnokov, M. G. Rastagaeva, and A. L. Syrkin, J. Appl. Phys. 80, 5464 共1996兲. 16 N. Nordell, A. Scho¨ner, and S. G. Andersson, J. Electrochem. Soc. 143, 2910 共1996兲. 17 N. Nordell, A. Scho¨ner, and M. K. Linnarsson, J. Electron. Mater. 26, 187 共1997兲. 18 J. Philibert, Atom Movements Diffusion and Mass Transport in Solids 共les e´ditions de physique, Les Ulis, France, 1991兲. 19 E. N. Mokhov, Y. A. Vodakov, G. A. Lomakina, V. G. Oding, G. F. Kholuyanov, and V. V. Semenov, Sov. Phys. Semicond. 6, 414 共1972兲. 20 E. N. Mokhov, E. E. Goncharov, and G. G. Ryabova, Sov. Phys. Semicond. 18, 27 共1984兲. 21 Y. A. Vodakov, N. Zhumaev, B. P. Zverev, G. A. Lomakina, E. N. Mokhov, V. G. Oding, V. V. Sermenov, and Y. F. Simakhin, Sov. Phys. Semicond. 11, 214 共1977兲. 22 P. A. Stolk, H.-J. Gossman, D. J. Eaglesham, D. C. Jacobson, C. S. Rafferty, G. H. Gilmer, M. Jaraiz, J. M. Poate, H. S. Luftman, and T. E. Haynes, J. Appl. Phys. 81, 6031 共1997兲. 23 A. O. Konstantinov, Sov. Phys. Semicond. 26, 151 共1992兲.

Downloaded 02 Oct 2013 to 202.116.1.149. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://apl.aip.org/about/rights_and_permissions