Shock wave assisted removal of micron size dust. particles from silicon wafer surfaces.

Transcription

1 Shock wave assisted removal of micron size dust particles from silicon wafer surfaces G. Jagadeesh 1, M. Mizunaga 2, K. Shibasaki 2, S. Shibasaki 2, T. Saito 3 and K. Takayama 4 1 Dept. of Aerospace Engineering,Indian Institute of Science, Bangalore,India 2 Maruwa Electronic Company Inc., Chiba, Japan. 3 Muroran Institute of Technology, Hokkoido, Japan 4 Interdisciplinary Shock Wave Research Center, Tohoku University, Sendai, Japan. Abstract. A new high speed rotor device has been designed, fabricated and tested for cleaning (removal of sub-micron size dust particles) silicon wafer surfaces. The 140 mm diameter aluminium rotor with grooves spins at a maximum speed of 50,000 rpm over silicon wafer surfaces placed at a distance of 1 mm from the rotor surface. A sonic argon jet coming out of the rotor surface is also used to enhance the wafer cleaning process. The tungsten particles (0.7 µ) are removed successfully from both plain and patterned silicon wafer surfaces. However it was not possible to remove 0.1 µ aluminium dioxide particles from the wafer surfaces. Details of the study are presented in this paper. 1 Introduction Micron and sub-micron particulate contamination is believed to be responsible for more than 80 percent yield loss in semi-conductor industries. With the ever shrinking device sizes in the micro-electronic industry there is an urgent need to come up with more effective dry cleaning techniques. Shibasaki et al (2001) had proposed the possibility of using a high speed (supersonic edge speed) rotor device for removing micron size dust particles from silicon wafer surfaces. This paper describes the development of a novel shock wave assisted technique (Takayama et al 2003) for removing sub-micron dust particles from silicon wafer surfaces. Silicon wafers are mounted in the vicinity (1-2 mm stand-off distance) of a high speed (60,000 r.p.m.) corrugated cylinder(140 mm diameter; 45 mm length). The high strength aluminium alloy rotor is powered by a bi-directional 3 phase A.C. induction water cooled motor that operates at 240 V and maximum power of 5 kw. The rotor itself is supported on both the sides with ceramic bearings with the maximum permissible axis vibration of about 30 microns. The focus of this study is to carry out sub-micron size dust removal experiments by using the corrugated cylinder spinning at supersonic edge speeds (380 m/s) in the vicinity of plane and patterned silicon wafer surfaces. Further Infra-red thermography, speckle interferometry and pressure measurements on a plate at close stand-off distances from the spinning rotor using PVDF pressure sensors have also been carried out to characterize the flow field around the rotor. The details of the experimental study along with few important results are presented in the subsequent sections. 2 Experiments, Results and Discussions Figure 1 shows the photograph of the high speed rotor with grooves along with the pneumatic fixtures for mounting the silicon wafers for cleaning experiments. The depth

2 2 Jagadeesh et al. of the angular groove is 5 mm and three such grooves are made on the rotor at 120 deg. apart. Although not discussed here various types of surface corrugations have been tried to arrive at the best possible geometric configuration for dust removal experiments. Provision is provided on the outer surface of the rotor for generating a sonic jet of air/argon during high speed spinning of the rotor over the silicon wafer surface. Two sets of three jet orifices 180 deg. apart are located at a distance of 30 mm from the inclined groove on the surface. Provision is also provided to generate a sonic jet coming from the axial direction in between the stationary wafer surface and the spinning rotor. Before carrying out the wafer cleaning experiments the surface static pressure is measured (Kistler 603B) at 5 different locations on a plate placed exactly in the same place where the silicon wafers would be housed for dust removal experiments. The typical pressure time history recorded from the these experiments at a stand-off distance (distance between the plate and the rotor surface) of 5 mm is shown in Fig.2. The instantaneous periodic pressure spikes seen essentially confirms the presence of shock trains continuously sweeping the surface when the rotor spins at 50,000 rpm. It appears that the pulsating flow field introduced by the spinning rotor in the gap between rotor and plate is the main reason for removing the sub micron dust particles. Although not discussed in this paper the presence of shock trains has also been confirmed from the electronic speckle interferometry and the temperature field measurements using Infra-red thermography. Further the dust removal experiments are carried out using 80 mm diameter circular silicon wafers. Both plain as well as patterned (with circuit etching) wafers are used and known size sub micron particles (tungsten and aluminium dioxide) are used to contaminate the silicon wafer surface. The particles are deposited on the wafer by preparing a solution of known quality of particles in anhydrous alcohol. The alcohol evaporates after some time leaving behind a stain of the particle clusters on the wafer surface. Microscopic images of the surface of the wafer is taken before and after the cleaning process. Figures 3 and 4 show the results from the cleaning experiments carried out using 0.7 µ tungsten particles. The photographs clearly show that we have been successful in removing the particles by the present technique in both plain and patterned silicon wafers. The current micro-electronic industry standard permits about 15 particles of 0.15 µ size to be present in 80 mm diameter wafer after the cleaning process. In practice it is the dust on the wafer surface originates mostly from the deposition of silicon dioxide particles during the circuit etching process carried out in clean rooms. Hence we also wanted to try removing 0.10 µ aluminium dioxide particles deposited on the wafer surface. But we have not been successful in these experiments as seen from Fig.5. In any case these are still early days before we integrate the device to the assembly lines in the micro-electronic industry. Issues such as the role played by the jet in removing the dust from the surface, the macro and micro characteristics of the complex fluid structure interaction in the narrow gap between the rotor and the wafer surface and the identification of critical control parameters influencing the cleaning process have to be resolved in future studies. 3 Conclusion A novel shock wave assisted sub micron size dust removal technique for dry cleaning of silicon wafer surfaces has been developed. The device comprises of high speed rotor with surface corrugations spinning over silicon wafer surface at supersonic edge speeds. The tungsten particles (0.7 µ) are removed successfully from both plain and patterned silicon wafer surfaces by the present method. However it was not possible to remove 0.1 µ

3 Dust removal from silicon wafer surfaces 3 aluminium dioxide particles from the wafer surfaces. Reducing the noise from the device as well as the temperature of the rotor surface are some of the important issues to be resolved before using the device in clean room environment for dust removal from silicon wafer surfaces. References 1. K. Shibasaki, S. Shibasaki, G. Jagadeesh, M. Sun and K. Takayama: High speed rotor device for removal of particles from solid surfaces using shock waves. Japanese Patent No , (2003) 2. K. Takayama, G. Jagadeesh and S. Shibasaki: Development of high speed cylindrical rotor device for industrial applications of shock waves. In: 23rd ISSW, Texas, USA, 2001

4 4 Jagadeesh et al. Fig. 1. A photograph of the high speed rotor used for dust removal from silicon wafer surfaces Fig. 2. The measured values of static pressure on the silicon wafer surface subjected to shock wave loading;50,000 rpm; stand-off distance=5 mm

5 Dust removal from silicon wafer surfaces 5 Fig. 3. Microscopic image of silicon wafer surfaces before and after shock wave loading;high speed rotor speed was maintained at 45,000 rpm for 15 s in the presence of a air sonic jet; 0.7 µ tungsten particles;stand-off distance = 1 mm Fig. 4. Microscopic image of patterned silicon wafer surfaces before and after shock wave loading;high speed rotor speed was maintained at 40,000 rpm for 30 s followed by 43,000 rpm for 15 s and finally at 43,000 rpm for 15 s in the presence of a air sonic jet for 15 s; 0.7 µ tungsten particles;stand-off distance = 1 mm

6 6 Jagadeesh et al. Fig. 5. Microscopic image of patterned silicon wafer surfaces before and after shock wave loading;high speed rotor speed was maintained at 40,000 rpm for 30 s followed by 43,000 rpm for 15 s and finally at 43,000 rpm for 15 s in the presence of a air sonic jet for 15 s; 0.1 µ aluminium dioxide particles;stand-off distance = 1 mm