Yoshiyuki Seike

Silicon Wafer Cleaning Using New Liquid Aerosol with Controlled Droplet Velocity and Size by Rotary Atomizer Method

A liquid aerosol, which sprays cleaning liquid with a carrier gas, is widely used for cleaning semiconductor devices. The liquid aerosol using a conventional two-fluid nozzle may cause pattern damage on the wafer. To resolve this problem, we have made a prototype new rotary atomizing two-fluid cleaning nozzle (RAC nozzle), which can control the velocity distribution and size distribution of flying liquid droplets separately. It was confirmed by measuring flying liquid droplets using a shadow Doppler particle analyzer system that the mean volumetric diameter of the droplets could be atomized to 20 µm or less at a rotational speed of the air turbine of 50,000 min-1 and that the mean velocity of the flying liquid droplets could be controlled in the range under 65 m/s independently. It was confirmed in a cleaning experiment using polystyrene latex (PSL) particles on a wafer that particle removal efficiency increased when shaping air pressure increased. Also, the particle removal efficiency was improved with the finer atomization promoted by a higher rotational speed of the air turbine.

Development and Analysis of a High-Pressure Micro Jet Pad Conditioning System for Interlayer Dielectric Chemical Mechanical Planarization

Conventional diamond disc pad conditioning methods employed in chemical mechanical planarization (CMP) have presented several problems for integrated circuit (IC) manufacturers. These include diamond wear, which reduces pad life, and diamond fracture, which causes the semiconductor devices to be scratched by loose diamond fragments. In order to attempt to overcome these problems, a high-pressure micro jet (HPMJ) conditioning system, in which pressurized ultra pure water (UPW) ranging from 3–30 MPa is sprayed on the pad surface, is proposed and developed. This study first analyzes the extent of the kinetic energy of water droplets ejecting from the HPMJ system and its utility in conditioning the pad surface. Subsequently, CMP is used to polish interlayer dielectric (ILD) films using both conventional diamond discs as well as HPMJ conditioning methods. Results, reported in the form of coefficient of friction (COF), removal rate, pad surface roughness and pad surface quality, highlight both the advantages as well as disadvantages of the HPMJ method compared to conventional conditioning schemes.

Development of a PAD conditioning process for interlayer dielectric CMP using high-pressure micro jet technology

Diamond conditioning was compared to an alternative method, namely high-pressure micro jet (HPMJ) conditioning, through a series of interlayer dielectric chemical mechanical planarization (ILD CMP) marathon tests. The two systems were compared individually and in combination on the basis of ILD removal rate (RR), coefficient of friction (COF), and the physical appearance of the pad surface (both on the top areas as well as inside the grooves). Results indicated that diamond conditioning alone was effective in causing RR and COF stability during extended runs, but it could not clean the slurry residues and other by-products from the surface of the pad (especially inside the grooves). Results also showed that HPMJ conditioning was able to effectively clean the pad surface, despite not providing enough energy to abrade the surface of the pad and maintain constant RR and COF during extended polishing. Based on these findings, a new pad conditioning method based on a combination of diamond and HPMJ conditioning was proposed. Results showed that this new method allowed for stable polish results in terms of RR and COF during extended marathon runs, and also yielded substantially residue-free surfaces, which could extend pad life and reduce wafer-level defect.

Thermal, Tribological, and Removal Rate Characteristics of Pad Conditioning in Copper CMP

High Pressure Micro Jet (HPMJ) pad conditioning system was investigated as an alternative to diamond disc conditioning in copper CMP. A series of comparative 50-wafer marathon runs were conducted at constant wafer pressure and sliding velocity using Rohm & Haas IC1000 and Asahi-Kasei EMD Corporation (UNIPAD) concentrically grooved pads under ex-situ diamond conditioning or HPMJ conditioning. SEM images indicated that fibrous surface was restored using UNIPAD pads under both diamond and HPMJ conditioning. With IC1000 pads, asperities on the surface were significantly collapsed. This was believed to be due to differences in pad wear rates for the two conditioning methods. COF and removal rate were stable from wafer to wafer using both diamond and HPMJ conditioning when UNIPAD pads were used. Also, HPMJ conditioning showed higher COF and removal rate when compared to diamond conditioning for UNIPAD. On the other hand, COF and removal rates for IC1000 pads decreased significantly under HPMJ conditioning. Regardless of pad conditioning method adopted and the type of pad used, linear correlation was observed between temperature and COF, and removal rate and COF.

Characterization of Slurry Residues in Pad Grooves for Diamond Disc and High Pressure Micro Jet Pad Conditioning Processes

Slurry residues inside pad grooves are quantified by UV-enhanced fluorescence (UVEF) technique for conventional diamond disc and high pressure micro jet (HPMJ) pad conditioning methods. Results show that with HPMJ pad conditioning, slurry residues inside pad grooves decrease with time at twice the rate compared to diamond disc conditioning. Therefore, this suggests that HPMJ pad conditioning is a viable method to extend pad life and possibly improve wafer-level defectivity.

Conformal photo-resist coating technique in the through-silicon via of the semiconductor devices with the rotary atomising aerosol spray

Semiconductor devices are increasingly sophisticated in their application of three-dimensional laminated chips inside through-silicon via (TSV). TSV is a technology that connects the stacked chips using through electrodes instead of higher integrated circuits densities. In this report, we invented a new photo-resist coating method inside the TSV using the rotary atomising aerosol spray. We measured the characteristics of the flying droplets from the atomising aerosol spray nozzle by a shadow Doppler particle analyser (SDPA) to indicate that the new method is capable of coating the photo-resist inside the TSV. Furthermore, we tried to make the prototype coating system with the rotary atomising aerosol spray nozzle and run the experiments in to coat the photo-resist inside the TSV test element group (TEG) wafer. Results indicate that the photo-resist is coated on the TSV TEG wafer having opening diameter of 50 µm to 200 µm, depth of 50 µm and square pyramid via holes with tapered angle of 54.7 degrees by using the prototype equipment. It is confirmed that coating along the inner walls of via holes is feasible.

Development of a Pad Conditioning Method for ILD CMP using a High Pressure Micro Jet System

The goal of this study is to determine if High Pressure Micro Jet (HPMJ) conditioning can be used as a substitute for, or in conjunction with, conventional diamond pad conditioning. Five conditioning methods were studied during which 50 ILD wafers were polished successively in a 100-mm scaled polisher and removal rate (RR), coefficient of friction (COF), pad flattening ratio (PFR) and scanning electron microscopy (SEM) measurements were obtained. Results indicated that PFR increased rapidly, and COF and removal rate decreased significantly, when conditioning was not employed. With diamond conditioning, both removal rate and COF were stable from wafer to wafer, and low PFR values were observed. SEM images indicated that clean grooves could be achieved by HPMJ pad conditioning, suggesting that HPMJ may have the potential to reduce micro scratches and defects caused by slurry abrasive particle residues inside grooves. Regardless of different pad conditioning methods, a linear correlation was observed between temperature, COF and removal rate, while an inverse relationship was seen between COF and PFR.