Dose-Fractionated EELS Through Multipass In-Situ Spectrum Imaging

Traditionally, electron energy-loss spectrum images were acquired using a single pass with dwell times on the order of 10’s to 100’s of milliseconds. Drift, sample damage, and contamination were not apparent until several minutes into a scan or until the acquisition was finished. Multipass spectrum imaging was proposed, where rapidly acquired single passes were summed together to obtain the final dwell time [1]. This allows for rapid feedback of sample integrity and the ability to correct sample drift between frames, eliminating mid-frame shifting of the features in the spectrum image. Combining multipass spectrum imaging with direct detection eliminates the accumulation of read noise by adding counted spectra. This allows users to add multiple spectra together without compromising the signal-to-noise ratio (SNR). However, dose-sensitive specimens such as biological cells, polymers, and ceramics can damage rapidly under the electron beam [2], and during multipass spectrum imaging, a compromised set of spectra could be added to spectra acquired from the sample in a pristine state. Gatan recently introduced in-situ spectrum image acquisition modes, where each spectrum image of a multipass series is stored as a function of time and can be linked to varying levels of applied stimulus (e.g., temperature, bias, mechanical, or electron radiation). Since each spectrum image is saved individually, one can combine or remove data before, during, or after changes are observed in the material under the applied stress giving the ability to monitor chemical phase and composition changes both spatially and as a function of time/stimulus [3]. In addition, combining or removing passes gives the ability to enhance the SNR or segment off the passes that may be compromised by sample damage or contamination. Here we combine multipass in-situ spectrum imaging with our K3 direct electron detector to monitor and characterize the effect of total dose in beam sensitive samples. Calcium carbonate (CaCO3) is used as a model system as the total dose thresholds for reduction and mass loss are well characterized [4]. In Fig. 1 we observe spectrum images from a cluster of calcium carbonate particles, separated by increasing accumulated dose. Mass loss is accelerated at the surfaces and boundaries between particles as evident by the waning intensities of the C K, Ca L2,3, and O K elemental maps. Previous studies stated that mass loss occurred overall through the kinetic pathway of CO2 formation [4]. However, through spectrum imaging we can observe the reduction of the Ca L2,3 signal at the edge and boundary regions of the particles. It is not yet clear if Ca diffuses away from the region due to local heating and defect formation, or if knock-on damage causes sputtering of Ca at the boundaries. In Fig. 2, the ELNES of the C K, Ca L2,3, and the O K edge were extracted from the red and blue areas highlighted in the ADF images on Fig. 1. Clear differences in the C and O K-edge ELNES are observed between the two areas, suggesting that the original chemical configuration is different at the boundary compared to the middle of the particle. The differences in the Ca L2,3 edge extracted from each area are initially more subtle. However, changes with total accumulated dose are observed to occur faster in the boundary region (blue) compared to the middle of the particle (red). Specifically, the crystal field splitting of the L2,­3 edges fade in the spectra extracted from the boundary, suggesting that the octahedral coordination deteriorates faster near the boundary compared with the middle of the particle [5]. In-situ spectrum imaging combined with counting detectors offers a method to acquire spectral data from dose sensitive samples without compromising data quality. Our next goal will be to apply spectrum imaging to more dose-sensitive materials systems such as polymers and biological systems. Simultaneous ADF (top) and corresponding elemental maps acquired as a function of accumulated dose. C K (a), Ca L2,3 (b), and O K ionization edges extracted from the middle (red box) from the top CaCO3 particle in Fig. 1. C K (d), Ca L2,3 (e), and O K (f) ionization edges extracted from the boundary (blue box) between two CaCO3 particles in Fig. 1.