The characterization of photoresist shrinkage difference in X-Y directions with CDSEM metrology

As IC industry moves forward along with critical dimension (CD) continuously scaling down, the inline CD error budgets become tighter to meet device demands, and thus its performance requirements for CD measurement also become more stringent. And Scanning electron microscope for CD metrology (CD-SEM) has been a mainstream metrology tool for CD measurement during inline production and offline CD data collections for the model setup of optical proximity correction (OPC). As reported in many papers and well known to the industry, photoresist (PR) shrinkage has been a big issue during CD measurement after development inspections (ADI) since it greatly impacts the accuracy and precision of CD metrology. In this paper, we focus on one interesting phenomenon about PR X-Y shrinkage at different rates under the exposure of e-beam irradiations when measuring two-dimensional (2D) features with CDSEM. In current applications of advanced devices, there are more and more complicated 2D structures, which requires efficient and accuracy CD metrology during productions or OPC model creations. During the measurement of square 2D features in ArF or ArF immersion processes, we have found there are about 2∼8nm X-Y CD difference in one ADI image, even if CDSEM had been well calibrated for X-Y magnification and astigmatism. We confirmed above results in neighboring exposure shots of same wafers by rotating 90 degree, and same CD difference found in X-Y directions. Plus, we also verified inline wafers after etching inspections (AEI), but no obvious X-Y CD difference was found. These facts told us one story that above X-Y CD difference should be from measurement itself and dedicated to ADI process. After a series of elaborate DOE experiments based on different substrates with ADI and AEI splits, we have clarified the underlying mechanism. The results of our study will be presented and discussed.