• PDF / 461,504 Bytes
• 6 Pages / 420.48 x 639 pts Page_size
• 43 Downloads / 265 Views

DOWNLOAD

REPORT

---

Si. Although the gain associated with the absence of channeling is real, the situation regarding the presence or absence of TED is more controversial and appears to strongly depend on the exact experimental details. The main drawback of the pre-amorphisation process is the formation of a band of extended defects near the original amorphous-crystalline (a/c) interface [1] which survive the SPE regrowth and are known as "End Of Range" (EOR) or "type II" defects [2]. Among other detrimental effects these EOR defects have been shown to strongly affect boron redistribution [3,4], and thus their presence impedes an accurate prediction of the dopant profiles. In this paper, after a brief review of the existing data concerning TED in pre-amorphised Si, we report on our recent work on the behavior of the EOR defects. We are then able to suggest a picture explaining some apparently contradictory results. MAIN EXPERIMENTAL RESULTS ON TED IN PRE-AMORPHISED SI A detailed recent review of the literature published during the last 15-20 years on this subject can be found in [5], and thus we give here only a brief overview. For a long time it has been thought that the existing data, exhibiting either TED or normal redistribution, could be rationalized in the following way:

55 Mat. Res. Soc. Symp. Proc. Vol. 532 ©1998 Materials Research Society

10 21

--.I

~~ ~ ,/~I. ~..

o-.* I

SI

.... .

nhred diffusion coonl.1 so coft.

-,,Brlff

101o0 0o . U10

r 10......

,

.

---

]i-]

'i-iil '.

1tprofile

10 0

0.04

0.08

0.12

depth (-ini)

Figure 1: Comparison between experimental and simulated results

i) If the initial profile extends beyond the initial a/c interface, the boron redistribution

exhibits

a

strong

TED.

The

diffusion

enhancement was attributed to the release of self-interstitials in the crystalline side of the wafer, which was damaged by the amorphisation stepii)performed by heavy ionwhen implantation. On the other hand, the initial B is totally inside the pre-amorphised layer, the diffusion is normal. This was attributed to the perfect crystallinity restored by the SPE regrowth at the very beginning of the

anneal, and it was suggested that the band of EOR defects act as a perfect barrier for the transmission of the self-interstitials from the

bulk side towards the regrown layer [3, 6]. More recently it has been shown that significant TED can occurs even if the as-implanted B profile is inside the amorphised layer, its magnitude depending on the boron dose [7]. This is illustrated on Fig. 1 (from [8]), showing a comparison between experimental and simulated profiles of B implanted (3 keV, 5x1014 cmn2 ) in a 175 nm thick pre-amorphised Si layer, annealed at 1000°C, 10 sec. It is clear from that figure, that the equilibrium simulation strongly underestimates the actual boron diffusion, and that an enhancement by a factor of 22 is necessary to obtain a satisfactory fit. On the other hand recent results by Chao et al. seems to contradict these observations [9]. These authors compare the redistribution of two bor