Analysis of spatial point patterns in electron-counting images. (9th February 2022)

• Record Type: Journal Article
• Authors: Kodama, Tetsuji; Nakashima, Yusuke; Akashi, Tetsuya; Takahashi, Yoshio; Mori, Shigeo; Harada, Ken
• Is Part Of: Microscopy. Volume 71:Number 3(2022)
• Page Start: 142
• Page End: 151
• Publication Date: 2022-02-09
• Subjects: electron counting -- direct detection sensor -- spatial point pattern -- spatial randomness; Microscopy -- Periodicals; 502.825
• DOI: 10.1093/jmicro/dfac006
• Languages: English
• ISSNs: 2050-5698

Abstract

Abstract: In this study, the spatial counting statistics of free electron beams, which were released via field emission from cold metal and propagated through a vacuum region, were investigated to examine the normal functioning of the counting equipment for electron correlation spectroscopy. The beam electrons were recorded separately according to the locations of individual events as they reached the direct detection transmission Complementary Metal Oxide Semiconductor (CMOS) sensor. We examined the spatial point patterns arising from the locations of the individual events of each primary electron being detected in the case of electrons in a state in which the wave function is constant on the sensor. The quadrat method, which compares the observed frequencies of the number of electron counts in the subsets of the study region with the predicted frequencies from a Poisson distribution, indicates a clustering-type departure from complete spatial randomness. To explore some of the basic principles governing the location of coherent electrons being counted, Ripley's K-function and the corresponding L-function of a stationary spatial point process were used to test the complete spatial randomness from the data. The maximum peak in the average of the L-functions was sensitive only to the mean counts per frame. Thus, clustering of spatial point patterns may result from abnormalities in the direct detection camera. When the interaction of the beam electrons with the sensor is included in the simulation, there is a reasonable match between the average of the L-functions and the experimental curves with the theoretically simulated curves.