Yuan and Chan (Psychometrika, 76, 670–690, 2011) recently showed how to compute the covariance matrix of standardized regression coefficients from covariances. In this paper, we describe a method for computing this covariance matrix from correlations. Next, we describe an asymptotic distribution-free (ADF; Browne in British Journal of Mathematical and Statistical Psychology, 37, 62–83, 1984) method for computing the covariance matrix of standardized regression coefficients. We show that the ADF method works well with nonnormal data in moderate-to-large samples using both simulated and real-data examples. R code (R Development Core Team, 2012) is available from the authors or through the Psychometrika online repository for supplementary materials.

One of the difficulties in analyzing atomic resolution electron microscope images is that the sample thickness is usually unknown or has to be fitted from parameters that are not precisely known. An accurate measure of thickness, ideally on a column-by-column basis, parameter free, and with single atom accuracy, would be of great value for many applications, such as matching to simulations. Here we propose such a quantification method for annular dark field scanning transmission electron microscopy by using the single electron intensity level of the detector. This method has the advantage that we can routinely quantify annular dark field images operating at both low and high beam currents, and under high dynamic range conditions, which is useful for the quantification of ultra-thin or light-element materials. To facilitate atom counting at the atomic scale we use the mean intensity in an annular dark field image averaged over a primitive cell, with no free parameters to be fitted. To illustrate the potential of our method, we demonstrate counting the number of Al (or N) atoms in a wurtzite-type aluminum nitride single crystal at each primitive cell over the range of 3–99 atoms.

Multislice simulations in the transmission electron microscope (TEM) were used to examine changes in annular-dark-field scanning TEM (ADF-STEM) images, conventional bright-field TEM (BF-CTEM) images, and selected-area electron diffraction (SAED) patterns as atomically thin hexagonal boron nitride (h-BN) samples are tilted up to 500 mrad off of the [0001] zone axis. For monolayer h-BN the contrast of ADF-STEM images and SAED patterns does not change with tilt in this range, while the contrast of BF-CTEM images does change; h-BN multilayer contrast varies strongly with tilt for ADF-STEM imaging, BF-CTEM imaging, and SAED. These results indicate that tilt series analysis in ADF-STEM image mode or SAED mode should permit identification of h-BN monolayers from raw TEM data as well as from quantitative post-processing.

The relation between image resolution and information transfer is explored. It is shown that the existence of higher frequency transfer in the image is just a necessary but not sufficient condition for the achievement of higher resolution. Adopting a two-point resolution criterion, we suggest that a 10% contrast level between two features in an image should be used as a practical definition of resolution. In the context of scanning transmission electron microscopy, it is shown that the channeling effect does not have a direct connection with image resolution because sharp channeling peaks do not move with the scanning probe. Through a quantitative comparison between experimental image and simulation, a Fourier-space approach is proposed to estimate defocus and sample thickness. The effective atom size in Z-contrast imaging depends on the annular detector's inner angle. Therefore, an optimum angle exists for the highest resolution as a trade-off between reduced atom size and reduced signal with limited information transfer due to noise.