Preliminary Testing Results For A New X-Ray Stepper

Abstract

ABSTRACT Preliminary testing results for a new, commercially available soft x-ray proximity stepper are presented. Total system performance (i.e., overlay preci­ sion, alignment precision, and critical dimension control) is under investiga­ tion using electrical probe structures and scanning electron microscopy (SEM). Preliminary results indicate that the total system performance is consistent with that required for 0.5 /zm device design rules. Process latitude data will be presented for several conventional novolak-based resists as well as for an e-beam novolak-based resist. Sensitivities have been obtained over the range of 5 to 400 mJ/cm^. The absence of measurable reflectivity coefficients for nearly normal incident soft x-rays allows one to perform stepper testing with nonstan- dard substrates. The stable source, large process latitude for conventional resists, and substrate insensitivity allow stepper use without the need to run send ahead wafers. The high MTF and lack of imaging aberrations allow the use of unbiased reticle patterns with excellent pattern replication on a variety of substrates. Preliminary performance of the alignment system, stage, and gap control imply that overlay consistent with 0.50 /zm design rules is achievable.1. INTRODUCTIONThe present industry standard for lithography at -1.0 /zm design rules con­ sists of ultraviolet (UV) exposure of cresol novolak resin that contains a UV sensitive dissolution inhibitor. In general, it may be easier to replace the lithographic technology, if the new tool is capable of being used with the pre­ sent resist, than it is to replace the photoresist due to the dramatic impact the resist has on subsequent wafer processes. As a consequence, strict quality control is placed on incoming resist to insure compatibility with painstakingly developed wafer processes (i.e., substrate etch and ion implant). Substantial yield losses occur if the resist is not compatible with subsequent processes.Optical technologies have limitations that are difficult to circumvent for patterns with features of 1.0 /xm or less (i.e., pitch < 2.0 /zm). Nonstandard resist processes (i.e., multi-level resist, contrast enhancement, and new resist materials) have been investigated to accommodate optical lithographic shortcom­ ings for high resolution applications; however, they are not generally being used due to increased capital cost and inherently higher defect densities. Developments to improve optical systems, primarily through the use of shorter wavelengths (i.e., i-line and deep UV), have required a reintroduction of most of the above nonstandard resist processes. Conventional resists have too high an absorption to be utilized effectively at shorter optical wavelengths.Other lithographic technologies have been investigated, but they are gener­ ally not capable of using existing, production-proven, novolak-based resists due to insufficient wafer throughputs (i.e., unacceptable sensitivities). In addi­ tion, most non-optical exposure modalities typically produce low average irra-