Direct Detection of Electron Backscatter Diffraction Patterns

We report the first use of direct detection for recording electron backscatter diffraction patterns. We demonstrate the following advantages of direct detection: the resolution in the patterns is such that higher order features are visible; patterns can be recorded at beam energies below those at which conventional detectors usefully operate; high precision in cross-correlation based pattern shift measurements needed for high resolution electron backscatter diffraction strain mapping can be obtained. We also show that the physics underlying direct detection is sufficiently well understood at low primary electron energies such that simulated patterns can be generated to verify our experimental data.

Figure 1

EBSD patterns recorded from Si (001) at 20 keV, $\sim 10\mathrm{nA}$ probe current and 20 s exposure. The bottom right of each pattern (inset) shows the power spectrum obtained from a Fourier transform of a $512\times 512$ pixel region at the center of each image. (a) Direct electron detection and (b) conventional indirect detection. (c) Calculated dynamical diffraction simulation. (d),(e),(f) Enlargement of features near the 114 zone axis. In (a),(d) the vertical 220 band at the center of the pattern exhibits both 440 and 660 Kikuchi lines, the latter often not being visible in (b),(e). Intensities in the extracted $512\times 512$ images were rescaled to zero mean and unit variance before transforming in order to normalize the power spectra.

Figure 2

EBSD patterns from Si (100) recorded by direct electron detection (a),(d) before and (b),(e) after plasma etching, compared to (c),(f) conventional indirect detection. (a),(b),(c),(d),(e) 20 s exposure. (f) 180 s exposure. (a),(b),(c) 10 keV primary energy. (d),(e),(f) 5 keV primary energy. The bottom right of each pattern (inset) shows the power spectrum obtained from a Fourier transform of a $512\times 512$ pixel region at the center of each image.

Figure 3

Precision of Si (100) EBSD pattern shifts measured between adjacent points on a line scan along the sample tilt axis, as a function of the number of frames averaged. Each frame used a 200 ms exposure and patterns were recorded at 20 keV primary energy with a 9 nA probe current.