LOW-ENERGY IMAGING OF NONCONDUCTIVE SURFACES IN SEM

If any not absolutely smooth and homogeneous nonconductive surface is observed in SEM, a locally variable charging appears which produces lo cal electrical fields that destroy the image. The surface charge depen ds on the total electron yield sigma, i.e. the number of outgoing elec trons divided by the number of the incoming ones, on the primary curre nt, resistance between the field of view and the earth, surrounding ga s pressure and time. A charge balance is established between primary a nd emitted electron currents, the specimen current taking charges off the surface, and the charged particle bombardment from the atmosphere, which can be neglected in high vacuum. sigma changes with the landing energy of primary electrons: near to zero energy, sigma < 1, and the surfaces charge negatively. For nearly all materials, between two char acteristic energy values E(I) and E(II), a exceeds unity and the charg e is positive, while above E(II), sigma < 1 again and a negative charg e appears. E(I) can be found usually between 80 and 300 eV and E(II) a mounts to one or a few keV and depends on the electron impact angle. A novel method is proposed consisting in approaching one of the critica l landing energies E(I) or E(II) where sigma --> 1 and the surface cha rge formation is suppressed to a level corresponding to the intensity of spontaneous discharging. The retarding field principle of low energ y SEM realization, which minimizes the point resolution changes with t he landing energy, makes both E(I) and E(II) equally well accessible. In order to find the critical energy without getting the specimen inte nsively charged, we recommend to measure, quickly enough from the poin t of view of the charging process duration, the time dependence of the signal in a single pixel by stepwise approaching the landing energy o f the slowest change, i.e. the acceptably approximated critical energy . A model of the image signal vs. time development is presented, toget her with a description of algorithms and software for the critical ene rgy measurement. The preliminary results presented confirm our basic i deas regarding the method.