No abstract

No abstract

No abstract

No abstract

An aid to focusing an electron microscope is described, employing the signal obtained by scanning the image. It is shown that use can be made of certain frequency components from the resulting signal, in order to provide an indication of the focus condition. A voltage is generated which is proportional to the mean power of these frequency components, and which, for most objects, has an extremum value at focus. The greatest sensitivity is obtained when phase-contrast effects are present, when the above voltage has a minimum value at focus. An experimental system has been built using a selenium electron-bombardment-induced-conductivity (e.b.i.c.) target scanned by a conventional raster, and, although targets of less than the highest quality were used, it has been shown that typical biological specimens, exhibiting phase contrast, can be focused with a reproducibility greater than is possible by eye. Simple theory predicts that, for such objects, all frequency components are equally useful as a measure of focus, but practical results show that only the upper half of the spectrum should be used, the lower frequencies tending to give rise to error in the direction of over-focus. The arrangement has also been used, in conjunction with a peak-holding controller, in an automatic focusing system.List of principal symbols V = output of mean squarer V o = value of V at minimum contrast / = focal length of microscope / 0 = focal length at minimum contrast N = mean-square noise added to V s(t) = signal resulting from scanning k = contrast /" = intensity in element of image I m = intensity in adjacent element I au = average, or background, intensity = phase angle between background and scattered waves 0 = angle between ray and optic axis at image M = magnification A = wavelength of electrons G ss (cu) = ideal power spectrum of s(t) 2tj = time taken to scan one resolution element t s = time interval between successive scans N o = electrons collected per second per unit area G nn (to) -power spectrum of image noise n(t) = noise added after filtering <«(0 2 ) = mean-square value of n(t)