Ultrasonic Cure Monitoring of Photoresist During Pre-Exposure Bake Process

We have developed a sensor to measure the glass transition temperature (T,) of photoresist during its pre-exposure bake. During prebake, the resist must reach this temperature in order for significant solvent evaporation to occur. To determine T, we measured the change in phase of a high frequency pulse as it was reflected from the interface between an 8” silicon wafer and a photoresist film. This measurement was performed before and after a prebaked Shipley 1813 resist film was removed from the wafer as well as during the prebake of a subsequent film. The phase change measured when a film of resist was removed from the wafer agreed well with theoretical calculations. The phase change was also monitored during photoresist cure and was found to decrease to a repeatable minimum at about 48 degrees Celsius. We believe that the characteristic temperature that resulted in the phase minimum was the glass transition temperature for the resinlsolvent mixture, and that the nonlinear increase i n phase that followed this minimum was due to solvent evaporation. INTRODUCTION The purpose of the photoresist prebake process is to remove any excess solvent that is initially added to the resist as well as to anneal the polymer film, allowing the resin molecules to relax into their lowest energy state. This thin film of resist is typically on the order of 1-2pm thick, with the change in film thickness during prebake of about 0.5pm. time is too long, then the feature size may not be as small If the solvent isn’t fully evaporated or if the prebake as expected. It is important when heating the resist for prebake that the temperature of the film reach and exceed the glass transition temperature (T,) in order to facilitate the diffusion of solvent and polymer chains that will anneal the film for prebake [ l ] . none of them have been applied in situ for endpoint Several techniques are available to measure Tg but detection of photoresist softbake [2,3,4,5]. There has been some research i n endpoint detection of the prehake process. Metz, et al. (6) performed real-time measurement of resist film thickness on silicon wafers using multi-wavelength reflection interferometry. They determined resist thickness versus spin and bake time. This method was used to monitor non-uniformities for statistical process control. effect of thickness change, hut also the effect of the The technique we describe here monitors not only the change in the elastic properties of the resist as i t bakes. 0-7803-4153-8/97/$10.00