Effects of Growth Kinetics and Surface Emissivity on Chemical Vapor Deposition of Silicon in a Lamp-Heated Single-Wafer Reactor

Abstract

Chemical vapor deposition of Si in a lamp-heated single-wafer reactor is studied by monitoring the growth rate of films deposited on bare Si, silicon-on-insulator ͑SOI͒, and oxidized Si wafers. The growth rate is consistently higher for deposition of polycrystalline Si than for epitaxy of Si, due to different kinetics dictating the two depositions. Epitaxy of Si on SOI shows a higher overall growth rate than on bare Si. Likewise, deposition of polycrystalline Si on oxidized Si wafer with a 4000 Å thick oxide has a higher growth rate than on that with 15 Å oxide. These are attributed to surface emissivity variation during Si deposition. The difference in kinetics plays a more dominant role than the surface emissivity variation in affecting the growth rate for the depositions studied.Chemical vapor deposition ͑CVD͒ of epitaxial Si ͑epi-Si͒ and that of polycrystalline Si ͑poly-Si͒ are characterized with different apparent activation energies in the surface reaction limited regime. 1-5 In our recent study of CVD Si at 600-750°C and 20-80 Torr in a lamp-heated single-wafer reactor, 5 the apparent activation energy for epi-Si is 2.1 eV and that for poly-Si 1.6 eV, in agreement with earlier published results. 1,3,6 Under these deposition conditions, first-order reaction kinetics was found for the ͑average͒ growth rate of epi-Si as well as for poly-Si with respect to the partial pressures of SiH 4 . However, the ͑average͒ growth rate depended more strongly on the partial pressure of H 2 for epi-Si compared to poly-Si. These differences in growth kinetics were attributed to a higher H surface coverage for epi-Si than for poly-Si. Hence the growth rate of epi-Si is limited by H 2 desorption whereas the growth rate of poly-Si is limited by the balance between the SiH 4 adsorption and the H 2 desorption. 5 As a result, an appreciable difference in thickness of the two types of Si layers is observed under identical deposition conditions. 5 Since the susceptor for bedding the Si wafer is heated by lamp radiation, variation of wafer surface emissivity during Si deposition leading to wafer temperature alternation could also result in different growth rates for epi-Si on bare Si substrates and poly-Si on oxidized Si wafers. [7][8][9] The purpose of the present study is to identify the dominant effect on growth rate, i.e., deposition kinetics vs. wafer surface emissivity.The single-wafer reactor used in this study was a commercial ASM Epsilon-2000 system whose cross section is shown schematically in Fig. 1. The whole graphite susceptor of 240 ϫ 280 mm size is coated with SiC to prevent contamination. It is heated by two tungsten-halogen lamp-banks above and below together with four lamps ͑not shown͒ located under the susceptor and centered around the rotating axis. A part of the susceptor is set to rotation to improve thickness uniformity over a wafer. The susceptor temperature is measured with three thermocouples placed in the surrounding fixed part ͑front, rear, and side͒ and one encapsulated in the middle of the rotating part. ...

Growth kinetics of SiGe/Si superlattices on bulk and silicon-on-insulator substrates for multi-channel devices

“…6. As expected, the Si and SiGe growth rates are thermally activated [27], with an increase for Si (for SiGe) from 3 (4) up to 22 (18) nm min À1 as growth temperature increases from 615 1C up to 675 1C. The associated activation energies, E a (Si)=54.8 kcal mol À1 and E a (SiGe)=42.3 kcal mol À1 , are close to the Si-H bond strength (47 kcal mol À1 [28]) and in-between the Si-H bond strength and the Ge-H bond strength (37 kcal mol À1 [29,30]), respectively.…”

“…We have thus investigated the specifics of the reduced pressure-chemical vapor deposition (RP-CVD) of Si and SiGe on both types of substrates. Some limited literature data have indeed shown that the Si growth rate can either increase [18] or decrease [19,20] when switching from bulk to SOI substrates. Both the buried oxide thickness and the overlayer thickness seem to have an impact on it [19,20], hinting at changes in surface temperature depending on the stack.…”

Selective epitaxial growth of boron- and phosphorus-doped Si and SiGe for raised sources and drains

“…A significant Si growth rate reduction was also noted when switching from 1000 to 1500 ( A and finally 4000 ( A-thick BOX layers [16], in agreement with point (ii). As far as trend (iii) is concerned, it fits rather well with the data reported by Pejnefors et al [29] for large deposited Si thickness. They have indeed obtained Si growth rates that were slightly lower than on bulk for the first 150 nm deposited on top of the 50 nm thick Si over-layer of their SOI substrate, and then slightly higher.…”

“…2 is much lower on SOI substrates with 4000 ( Athick BOX layers than for other thickness. This is most probably due to changes in absorptivity and emissivity when the BOX thickness changes, which induce some surface temperature (and thus growth rates) variations for the same nominal growth temperature [29]; (iii) For longer deposition times, the Si growth rate converges towards the one on bulk, fullsheet wafers (7.6 nm min À1 ). The Si growth rate extracted from the next 30 min of growth (i.e.…”

Growth of SiGe/Si superlattices on silicon-on-insulator substrates for multi-bridge channel field effect transistors

“…Some limited literature data have indeed shown that the Si growth rate can be vastly modified (usually reduced) when switching from bulk to SOI substrates [10][11][12]. Both the buried oxide thickness and the over-layer thickness seem to have an impact on it [11,12], hinting at changes in the surface temperature depending on the stack.…”