Oxygen Precipitation and Thermal Donors in Silicon
- PDF / 131,878 Bytes
- 2 Pages / 420.48 x 639 pts Page_size
- 80 Downloads / 254 Views
OXYGEN PRECIPITATION AND THERMAL DONORS IN SILICON RONALD C. NEWMAN J. J. Thomson Physical Whiteknights, Reading,
Laboratory University of Berks, RG6 2AF, U.K.
Reading,
P. O.
Box 220.
ABSTRACT with second order kinetics at the Oxygen in silicon precipitates 0 normal diffusion rate at 450 C and there Is evidence for self-Interstitial generation. These findings are discussed In relation to thermal donors. A REVIEW OF RECENT MEASUREMENTS AND THEIR ANALYSIS 0 Heating CZ silicon in the range 650-1050 C leads to the precipitation of S10 2 particles wtih a number density n = 0. 15 exp 12. 65eV/kT] cm- 3 at To [1]. oxygen O diffusion by normal interstitial a rate limited accommodate the volume increase around precipitates, self-interstitials Si. 0 are generated. If we extrapolate to 450 C, the value of n (perhaps naively) However the indicates that each particle contains only two oxygen atoms. traditional view is that up to say ten or more 0, agglomerate sequentially and the defects correspond to the observed sequence of thermal donors (T. D. 's). Enhanced diffusion of 0i has to be invoked, while the role of Sti is unclear. The general validity of such an extrapolation is therefore in doubt. Nevertheless we should seek experimental information to answer the questions [2,3]: (a) how large are 0i agglomerates? (b) is 0. diffusion (d) are Si. involved in T. D. 's? enhanced? (c) are Si, generated? Heating samples at 4500 C for 500h leads to a loss of half the [Oi( 3 originally present (2], but no new vibrational absorption 1018 atom cmMaybe the characteristic of SI-0 bonds of S102 has been reported. complexes formed have a small dipole moment as for a free di-oxygen molecule. Evidence for precipitates in these samples has been sought using neutron scattering and TEM. but without success [4]. indicating that the particles are Indeed very small. The first defects to form must be oxygen pairs. During this period the loss of 0,. measured by IR, should be described by second order kinetics. A plot of tc(t)]-' - [c(o)]-' versus time t should be linear with a gradient of (8TwDr B ). where r. (taken as 10A) is the capture radius for pair formation. The linear dependence amazingly extends to very long times (200h) [3]. Values of the diffusion coefficient D = 6.7 x 10-20 and 3.2 x 10-19 cm 2 s- 1 have been determined 0 from 420 and 450 C data respectively and agree with the normal values of 2 1 (1, 5]. The conclusions are that 5. 9 x 10-2o and 3. 4 x 10-19 cm senhanced diffusion of 0, does not occur, and the formation of 02 defects is the rate limiting process for 0, precipitation. The overall process must be more complicated because T.D. production must be explained. Resistivity measurements made on our sample heated at 450 0 C showed that one T. D. was formed for every 16ten 301 atoms lost for all times up to 150h, when [T. D.] reached 2 x 10 cm- . We next describe evidence for the formation of Si,. Heating CZ SI doped with carbon ([C.] = 7 x 10 17 cm- 3 ) at 4500 C does not modify the rate of is also lost at about half this rate
Data Loading...
