|
Infrared spectroscopy is a powerful tool for studying chemical reactions which take place in photoresist films. It provides information about functional groups present in the material, is nondestructive to the sample, can be used to monitor reactions in real-time, and can be used as a monitor for real-time process control. Transmission spectroscopy has been used for determination of rate constants for the post exposure bake (PEB) step in both positive and negative chemically amplified x-ray resists, but in transmission the IR beam must pass through the wafer in order to reach the detector. Reflection spectroscopy is more suitable for on-line monitoring because all the optics are located on the same side of the wafer; however, traditional reflection spectra of thin films contain interference fringes which make quantitative analysis difficult. We have developed a method, called mirror-backed infrared reflection- absorption spectroscopy (MBIRRAS), that eliminates the problem of interference fringes in reflection spectra. The method is capable of providing quantitative information about the concentration of functional groups in resist films and produced a linear calibration curve for a commercial crosslinker (hexamethoxymethylmelamine) in a Novolac matrix over a 20% by weight range of concentrations. MBIRRAS was also used to monitor the height of an IR peak corresponding to formation of crosslinks in Shipley negative-tone resist SAL 605 at various processing conditions. The results of this experiment show that PEB time and temperature, prebake time and temperature, and delay between exposure and PEB, all have statistically significant effects on the extent of crosslinking during PEB. In order to explain the observed effect of prebake conditions on the amount of crosslinking during the PEB step and to define the acceptable range of prebaking conditions for resist/wafer adhesion studies, another experiment was performed to demonstrate that thermally generated crosslinking occurred during an additional prebake step without any exposure to x rays. In fact, an additional prebake for 60 seconds at 110 degree(s)C can generate the same number of crosslinks as an x- ray dose of 4 mJ/cm2, while baking for 60 minutes at 110 degree(s)C is equivalent to an x-ray dose that is greater than the resist sensitivity (42 mJ/cm2). Noticeable changes in crosslink concentration were detected at 30 seconds. This suggests that crosslinking occurs during the normal prebake step and explains the dependence of linewidth on prebake time. Finally, a correlation was demonstrated between printed linewidth and MBIRRAS peak height corresponding to crosslink formation in SAL 605. This correlation, along with the demonstrated precision of the MBIRRAS method (3 sigma <EQ 3%), suggests that MBIRRAS can be used as an on-line monitor for adaptive control of the processing of SAL 605, thus reducing critical dimension (CD) variation from wafer to wafer for 0.25 micrometers lithography.2438
|
