Metrology Package for E-Beam Exposure Tool Evaluation.

Throughout the stages of e-beam exposure tool development and application, the machine performance needs to be accurately characterized. A series of tests have been developed and implemented to diagnose and optimize electron-beam exposure systems (EBES) for both mask-making and direct wafer writing. The techniques employed for these tests will be described, the evaluation results and how they lend themselves to machine diagnostics will be illustrated for the AT&T in-house EBES4 system. The tests consist of a complement of optical inspection, SEM imaging, MARKET(R) analysis, and electrical probe metrology. Optical and SEM techniques served to examine the machine writing characteristics and exposure pattern quality. Partially developed images in positive resists accentuated errors due to improper exposure. Electrical probe metrology was used to quantify and optimize machine accuracy. Compared to conventional optical and SEM measurement techniques, electrical metrology offers advantages of higher throughput, superior resolution and sensitivity, much better accuracy and precision, and abundant statistics. The standard MEBES MARKET(R) program was utilized to measure pattern misplacement due to intrafield and interfield distortions. The magnitude of stripe shear and butting errors was quantified by electrical linewidth measurement. The 256microns stripe butting error on EBES4 has a mean value of 0.011microns and 3 sigma of 0.066microns. Employing simple electrical testers such as box-in-box and bridge resistors, EBES4 demonstrated pattern placement precision of 0.069microns and 0.066microns, ``|mean| + 3 sigma'', in the x and y directions respectively. Aligning globally to zero-level fiducial marks, machine level- to-level registration capability for direct writing was determined to be 0.093microns and 0.110microns, ``|mean| + 3 sigma'', in the x- and y-directions respectively. Again, with the aid of electrical testers, linewidth control across a 90 mm square writing area at 20 kV on EBES4 was measured to be 0.027microns (3 sigma) for 0.375microns isolated features. Such machine evaluation data can be used to prove-in system specifications, and can be fed back to both the machine designers and the users for system optimization and IC design rule development respectively.