The information upon which operators, technicians and engineers make decisions consists of measured data. Bad data usually lead to bad decisions. For this reason quality control professionals have always focused on metrology. The problem is particularly significant for lithographers, where the extremely small features of state-of-the art processes have pushed measurement capabilities to their limits, and, some might say, beyond. Some of the metrology issues and challenges faced by lithographers will be discussed in this chapter.
Defect detection will be used as an example to illustrate the importance of understanding the measurement process in order to properly use measurement tools. Consider a laser light scattering system used to detect defects (Fig. 6.2). A particle is detected because it scatters light. However, small particles scatter light primarily in the forward direction (Fig. 8.1). When defects are sitting on, or embedded in, thin films, optical interference will affect the strength of the signal produced by the scattered light, because the forward scattered light is detected only after it has reflected from the substrate (Fig. 8.2). Since particle size is determined by the strength of the scattered light signal in these systems, the measured particle size will depend upon the films on the substrate, which can modulate the reflectance (Fig. 8.3). A particle will scatter a certain amount of light, but the films on the substrate will affect the amount of light that actually reaches the detector. Consider, for example, the measured particle size in a controlled experiment. Polystyrene latex spheres were used to calibrate a system which used a HeNe laser (wavelength=632.8nm ) on bare silicon. The resulting measured particle sizes for spheres on bare silicon and substrates with different thickness of silicon nitride are shown in Table 8.1. Because of the thin film interference effect, the apparent particle size is generally smaller for the nitride films, since a reduced signal can result from either small particles or a low-reflectance substrate. In some instances, the signal was too small to provide a measurement. The opposite effect occurs on an aluminum substrate, where the highly reflective surface appears to enhance the particle size (Table 8.2). Interpretation of defects that consist of embedded particles or voids must be made with caution, as well.
It has just been shown that considerable misinterpretation of the data would occur if one did not take thin film optical interference effects into account. This is of particular importance to lithographers, who are usually concerned about defects in photoresist films or anti-reflection coatings. Correct decisions require an understanding of the measurement process.
Online access to SPIE eBooks is limited to subscribing institutions.