Multi Factorial Study of Megasonic Enhanced Photoresist Strip with DiO3

Abstract

Keywords: Multi factorial screening, heated DiO3, dissolved ozone, megasonic, photoresist stripping, cavitation, OH° radicalsIntroduction Optimization of photoresist stripping recipes involves a number of parameters. But which factor or combination of factors has the most impact on removal rate of different photo resists? Cost of ownership considerations is also very important. In this study, a design of experiments (DoE) supported by software involving a number of controlled factors was performed. The different photoresists were treated with ozonized UPW (DiO3). In order to increase the strip rate parameters like dissolved ozone concentration, megasonic power density, temperature, spin speed and flow rate were varied. In order to obtain a better understanding a two level fractional factorial DoE approach was chosen. This multi factorial study provided a design space (allowed process window) showing optimization potential for different types of resist. Direct correlations between different parameters were found. Experimental Details APPARATUS— Silicon wafers were processed on a POLOS™ Spin200 advanced tabletop single substrate spin processor. A megasonic transducer and DiO3 source were both integrated with the Spin200 for these experiments. A radially uniform, large area sapphire megasonic transducer, the MegPie® was used. The DiO3 was supplied by an MKS LIQUOZON® Single Ozonated Water Delivery System. The liquid was heated with a Heateflex®. Details are described in [1] and Figure 1. MATERIALS — For the tests 8” Si wafers were used, coated with different types of Photoresist both positive and negative (see Table 1). The thickness was always 10 µm to ensure residual layer for strip rate measurement. Priming using a spin-on primer and coating of the PR was done on EVG®101 200mm Advanced Spin- and Spray Coating System, the process was optimized in order to achieve best coating uniformity in order to guarantee a minimum influence of the TTV (total thickness variation) on the strip rate evaluation. MEASUREMENT— PR film thickness was measured before and after the DiO3 stripping process using a FRT® Micro Glider Fully Automated Optical Surface Metrology, 44 point pattern along the x-axis through the center of the wafer. The average was used for before/after comparison. The difference divided by process time expressed as strip rate in nanometers per minute (nm/min). A baseline screening with AZ 9260 using MKS MODDE® design of experiments software was done with a total of 19 runs: 16 runs with changing factors from minimum to maximum level. Three runs were repeated at mean settings in order to determine the reproducibility of the tests (Table II). After validation of the program, more challenging photoresists (AZ nL of 2070, SU 8 UV + PEB, BCB 4024-40) were tested. Target was a removal rate of 150 to 300 nm/min. A very wide scope of process variables was tested for the impact on removal rate with several types of PR e.g. DiO3 concentration 40 – 120 mg/l, temperature 20 – 60°C. Results The screening provided both the design space visualized in a MODDE 4D contour plot and the ranking of most influencing factors (see Figure 2). All values within 4 sigma standard deviation, no outliers found, regression R² = 0.8718 (see Figure 3). Good fit of the software model was obtained which allowed interpolation of the design space.Targeted removal rate of 150-300 nm could be achieved both with baseline and verification test and even surpassed applying extended factors with up to 565 nm/min (AZ nLOF). Even for a challenging resist like SU 8 UV-PEB, a significant removal rate of 146 nm/min could be noted.The concentration of DiO3, temperature, flow rate, speed, megasonic power density and time were the ranking of factors by effect on the strip rate.In general the strip rates were enhanced by increasing the dissolved ozone concentration, DiO3 temperature as well as spin speed. Up to 70 % higher strip rate were observed by increasing the DiO3 concentration from 80 to 110 mg/l and doubling the spin speed. On the other side spin speed showed an allowed maximum in case of adding megasonic power (see Figure 4+5). The paper will present the interaction of the analyzed parameters and potential optimization for tool set-up.