A program (SEM-PSF) has been written to estimate the Point Spread Function (PSF) from a suitable SEM image. The PSF is modelled using a Gaussian and/or Lorentzian function obtained by fitting a line profile model (offset plus Erf and/or ArcTan function) to the average of the best of many edge profiles measured at a range of angles. PSF measurements comparing SEM performance under a range of conditions will be discussed.
SEM resolution is a critical performance specification but there is no widely agreed definition or standard measurement method.
The two most common approaches for measuring SEM resolution are based on Fourier transform methods [1] and line profiles measured across edges [2]. Resolution is typically given as a single number assuming Gaussian probe shape. This “resolution” is essentially the width of the Gaussian but unfortunately, there is no agreement on where this width is measured. The measured “resolution” could vary by a factor of three or more depending on the choice of width. In addition, the probe shape may not be Gaussian or symmetrical and so cannot be described by a single “resolution” parameter.
The shape of the electron beam focussed at the sample is a better measure of resolution. The SEM image is a convolution of the ideal object image with the probe shape. In principle the probe shape can be found from the image of a point source, the PSF. In practice this is very difficult. Program SEM-PSF allows the PSF to be measured using edge profiles from a suitable sample.
References
1. SMART – a program to measure SEM resolution and imaging performance, D.C. Joy. Journal of Microscopy, Vol. 208, Pt 1 October 2002, pp 24–34
2. Modelling PSF of scanning electron microscopes for image restoration, J. Swindells; M. Razaz; K. Tovey. 9th European Signal Processing Conference (EUSIPCO 1998)