Daily alignment of the microscope is a prerequisite to reaching optimal illumination lens and imaging lens conditions for high resolution imaging in cryo-EM. In contrast to the dramatic progress in automated image acquisition and post-image processing techniques, less attention has been paid to the improvement of microscope alignment before data collection. In this study, we have employed our recently published tool, s2stigmator, to study how image astigmatism varies with the imaging conditions (e.g. defocus, magnification). We have found that the large change of defocus/magnification between visual correction of astigmatism and subsequent data collection tasks, or even during data collection will inevitably result in undesirable residual astigmatism in the final images. Furthermore, the dependence of astigmatism on the imaging conditions varies significantly from time to time, so that it cannot be reliably compensated by pre-calibration of the microscope. These findings have essentially invalidated a basic assumption of current cryo-EM imaging strategies that assumes invariant astigmatism for different defocuses/magnifications used in the microscope alignment stage and the final data acquisition stage. Based on these findings, we recommend the same magnification and the median defocus of the intended defocus range for final data collection are used in the objective lens astigmatism correction task during microscope alignment and in the focus mode of the iterative low-dose imaging. It is also desirable to develop a fast, accurate method that can perform dynamic correction of the astigmatism for different intended defocuses during automated imaging.

Defocus and magnification dependent variation of TEM image astigmatism