Statistical parameter evaluation for swing curves for the 1.2 μm and 1.8 μm resist thickness in CMOS photolithography process technology

Abstract

The swing curve is depending on the rotation speed of the spinner and the thickness of the photoresist. The thickness should be thick enough to prevent the unexposed area from being etched. It is also need to be uniform so that the image of the pattern to be developed later is in focus at all points on the resist surface. The variation of the photoresist thickness will impact the overall yield of the devices. In this respect, the photoresist thickness needs to be characterized carefully. Therefore, an experiment was set up to determine the right thickness and exposure time to control the critical dimension of the pattern. Photoresist coater and stepper are the main equipment used in this experiment. Positive photoresist is utilized as a pattern transfer medium and elipsometer is employed for film thickness measurement. 24 wafers are used in this experiment. The wafers are separated into 2 sets, each set which consists of 12 wafers have all the thickness variation. This to test the effect of standing resist before exposure and development to the dose-to-clear. The first set is coated, exposed and development. However, the second set is exposed and developed two days after being coated. After the wafer went through the standard cleaning procedure, the wafers were then coated using standard recipes which the spin speed ranging from 6500 to 7600 rpm in 100 rpm incremental with the designated rpm fro 1.2 µm photoresist thicknesses is in the middle. Subsequently, the photoresist thickness of each wafer is measured using elipsometer. Then, the exposure process is done by creating job that will expose the wafers in 20 by 10 arrays of small squares of open field with size 2x2 mm. The exposure is done in such a way where the exposure time will step up to the right and step down to the left from the center after each column. The center of the exposure energy time/step is 240/5. Then the graph is plotted to see the different. From the results we obtained that the minima for the dose-to-clear is at 7200 rpm where the thickness is 1.22 µm. whereby the first set and the second set of wafers produce a slight variation in clearing point.