The role of sulfur during Mo etching using SF6 and Cl2 gas chemistries

Abstract

As the design rule of integrated circuits (IC) has decreased in size, the requirements for making interconnections of these circuits have become more stringent. When fabricating ®ne pattern interconnections, the resistance and contact reliability of interconnection materials have to improve in quality. In the past, ®ne pattern interconnection has been conducted with the use of polysilicon; however, resistance and contact reliability have now called for the use of refractory metal silicide [1, 2]. The use of refractory metal has brought some valuable characteristics into question. Among the variety of refractory metals, Mo has an extremely high melting point (2620 8C), good thermal stability and a low resistivity. Also, the thermal expansion coef®cient is similar to that of silicon (Mo ˆ 5:0 3 10y6 8C, Si ˆ 3:0 3 10y6 8C), so when it is heated, Mo experiences little stress. Due to these favorable qualities, Mo has been used in many other applications such as hard coatings, superconductors, Al diffusion barriers, gate materials and high dielectric electrode materials [3, 4]. Thus, we will continue to see Mo compounds used in many more applications. When Mo compounds are to be used in semiconductor processing, the dry etching of Mo is an essential process. Meanwhile, the etching of Mo compounds has been generally carried out with SF6 or SF6 mixed gas plasma. This is due to the high vapor pressure of MoF6 compounds. At the same time, the role of S in the etching of Si or W with SF6 plasma has recently been reported. Ninomiya et al. [5] concluded that S promotes the etching of Si. Moreover, Turban et al. [6] explained that in the case of an SiO2 cathode, S reacted with oxygen generated from cathode materials to form SO2, and SO2 was removed on the W surface. During the etching process of Mo compounds, however, the role of S has not been investigated thoroughly. To improve the etching characteristics, the role of plasma species such as F and S in the etch reaction has to be examined closely. Therefore, the role of S on the surface reaction during Mo etching with SF6 gas plasma was investigated in this work. In this study, the surface reaction of Mo etching was investigated with SF6=Cl2 gas plasma by using X-ray photoelectron spectroscopy (XPS). To examine the role of S during the etching of Mo with SF6 gas plasma, pure Mo was used in this process. The role of S and the relationship between the etch rate and gas-mixing ratio will be further discussed in this letter. The 5-inch Si substrates used for this study were doped with B (0.85±1.15 U-cm), oriented (1 0 0) and chemically etched for 60 s using 1% HF : H2O prior to chemical vapor deposition (CVD) growth. The substrates were coated with a 600-nm-thick layer of SiO2 grown by low-pressure-CVD (SiH4 ‡ O2, 420 8C, 240 mTorr). Deposition of the Mo ®lms was performed using a Varian 3180 d.c. sputtering system equipped with a 7-inch conical magnetron sputtering source. The Mo sputtering target was speci®ed at 99.999% purity. Sputtering was performed in research-grade Ar at a pressure of 8 mTorr, and the distance from source to substrate was ,3.3 inch. Typical sputtering power was 1 kW. During deposition, the substrate was grounded, and the substrate temperature was held at 100 8C using gas conduction heating. The ®nal thickness of the sputtered Mo ®lm was ,300 nm and was ,95% uniform across the surface of the 5-inch wafers. Plasma etching of the Mo ®lms was performed using an Applied Materials Model P-5000 etching system. Wafers were placed on a bottom electrode. Substrate holder temperature during the Mo etch was held at ,45 8C using the circulation of cooling water. A turbo pump reduced residual gas pressure to below 5 3 10y5 Torr before the processing step. The Cl2=SF6 gas mixture was used for Mo etch, and the total gas ow was 24 sccm. The Cl2=(SF6 ‡ Cl2) mixing ratio was changed from 0 to 1. RF power, chamber pressure and magnetic ®eld were constant at 250 watts, 100 mTorr, and 30 Gauss, respectively. The etch rate was measured using a four-point probe. After removing the plasma-etched samples from the etching system, the samples were exposed to the atmospheric environment for approximately 10 min prior to XPS analysis. Compositional analysis of the Mo surface was performed using a VG Scienti®c