A low resistance salicide technology which includes a formation of Selective W on TiSi, and an mnealing after W deposition (SWAN) has been proposed. A remarkable reduction in sheet resistance (down to 1/10) and thermal stability (up to 850°C) have been achieved for quarter-micron CMOS.
1.INTRODUCTIONAs the device size shrinks, TiSi, thickness should also be reduced to realize shallower source/drain junctions. In quarter micron devices, the TiSi, can never have a low resistivity. This is because the reduced line width and film thickness demand a higher phase transition temperature[ 11, while agglomeration temperature of TiSbdecreases in thinner TiSiJ21.In this paper, we present a new salicide structure which consists of a tungsten layer selectively deposited on TiSi,. The tungsten layer gives a sufficiently low sheet resistance, making the phase transition from C49 to C54 TiSi, unnecessary. The major role of TiSi, is as a barrier layer (against the diffusion of W and Si) rather than a low resistance layer. This new salicide structm has been accomplished by developing W selective chemical vapor deposition technology on metastable C49 TiSi,.
--
2.TECHNOLOGYIn the fabrication of W-covered TiSi,, there are two major difficulties. First, fluoride compound, which inhibits good contact, is formed on TiSi, films exposed to WF,[3]. Second, the nucleation of W is very poor on TiSi,. Other studies showed that W could be deposited on the surface-nitrided TiSi, without forming fluoride compounds[4], while the poor nucleation problem still remains [5].To solve these problems, the novel technology, SWAN includes two key points$) the formation of W on C49 TiSi, to enhance nucleation of W, (ii) the rapid lhermal -annealing(RTA) after W deposition to remove fluoride from the nucleation layer in the Tis&. An outline of the SWAN process is shown in Fig.1. Fig.2 shows a TEM cross-sectional view of a 0 . 2 5~ &OS transistor using SWAN technology.
3.RESULTS AND DISCUSSION
Nucleation of W on TiSi,Fig.:3(a) shows a SEM micrograph after W deposition on C49TiSi,. Gr~od selectivity and morphology of W have been achieved on C49 TiSi, However, on C54 TiSi, the morphology was very poor (Fig:.3(b)). The possible explanation of the improved nucleation in Fig.3(a) is the favorable reaction of WF, with C49 Tis4 compared with that of C54 TiSG. The estimated W growth rate on C49 TiSi, was 4nm/sec., and that of C54 TiSi, was under OSnm/sec. Fig.4 shows cross-sectional TEM micrographs after selective W deposition on C49 TiSi, and on C54 TiSi,. A nucleation layer of approximately lOnm thickness exists on C49 Tis&, while on C54 TiSi, there are few observable nucleation sites. AES measurement detected Ti, Si, and F in the nucleation layer.
Removal offluorine at the nucleation layerFig.5 shows the AES depth profiles of W/riSi, layer. The asdeposited profile (Fig.S(a)) indicates that the fluoride layer exists between W layer and TiSi, layer. However, the 2nd RTA at 800°C could successfully remove fluorine (FigS(b)), thus greatly reducing resistivity of the ...