Interpreting nanovoids in atom probe tomography data for accurate local compositional measurements

Xing Wang, Constantinos Hatzoglou, Brian Sneed, Zhe Fan, Wei Guo, Ke Jin, Di Chen, Hongbin Bei, Yongqiang Wang, William J. Weber, Yanwen Zhang, Baptiste Gault, Karren L. More, Francois Vurpillot & Jonathan D. Poplawsky

Abstract
Quantifying chemical compositions around nanovoids is a fundamental task for research and development of various materials. Atom probe tomography (APT) and scanning transmission electron microscopy (STEM) are currently the most suitable tools because of their ability to probe materials at the nanoscale. Both techniques have limitations, particularly APT, because of insufficient understanding of void imaging. Here, we employ a correlative APT and STEM approach to investigate the APT imaging process and reveal that voids can lead to either an increase or a decrease in local atomic densities in the APT reconstruction. Simulated APT experiments demonstrate the local density variations near voids are controlled by the unique ring structures as voids open and the different evaporation fields of the surrounding atoms. We provide a general approach for quantifying chemical segregations near voids within an APT dataset, in which the composition can be directly determined with a higher accuracy than STEM-based techniques.

Introduction
Nanosized voids, including pores, cavities, and bubbles, are common defects found in materials1. Voids can be generated under a variety of service conditions and are most often detrimental to the material’s structural properties. For example, a high strain rate leads to the nucleation and growth of nanovoids, which can coalesce and cause the fracture of ductile metals2,3. Helium (He) bubbles are formed in materials used in nuclear reactors owing to the synergistic effects of neutron bombardment and the nuclear (n,α) transmutation reactions, resulting in swelling of the material4,5. Besides such undesirable voids, nanosized voids can be intentionally introduced to optimize a material’s properties6. For example, nanostructured porous silicon has found extensive application in optoelectronics and biomedicine owing to its high-efficiency photoluminescence and variable surface chemistries7. The inherent pore structures in zeolites are exploited for catalysis owing to their high surface-to-volume ratio8.

Voids are essentially empty volumes and their interactions with the surrounding matrix are controlled by the surrounding atoms. An accurate characterization of the material chemistry around voids is important for both mitigating the formation of unintended voids and taking advantage of the presence of intended voids. Scanning transmission electron microscopy (STEM)-based spectroscopy methods, e.g., electron-energy loss spectroscopy (EELS) and energy dispersive X-ray spectroscopy (EDS), are powerful tools for performing compositional analysis at the Å-level9,10; however, when the size of a target object is smaller than the specimen thickness, the obtained composition of the target is an average value that includes all of the material within the full sample thickness, not just the target because the electron beam is transmitted through the entire sample thickness, leading to errors in the compositional measurement of the targeted feature. Atom probe tomography (APT) provides sub-nm-scale three-dimensional (3D) chemical information with a much higher compositional sensitivity than that measured by analytical STEM techniques11. APT analyses are enabled by applying a large electrical field to a needle-shaped specimen tip that ionizes and pulls the atoms off the tip’s surface one-by-one, i.e., field evaporation11. The type and position of the ionized atoms are determined using time-of-flight mass spectrometry, a position-sensitive detector, and a reconstruction model12. This makes APT a promising approach for determining local compositions without the detrimental influence of sample thickness. APT has been applied to characterize the material chemistry near nanoscale defect features such as grain boundaries and precipitates13,14,15,16. Further details regarding the APT experiment can be found in the Methods section.

During an APT experiment, the specimen must have a perfectly hemispherical cap shape for the magnification to be consistent on the detector. Any deviation from a hemispherical surface causes the projected surface image to be distorted (aberrated). Aberrations are commonly encountered in APT reconstructions near structural or chemical heterogeneities that may change the normal field evaporative behaviors of atoms. These aberrations are likely exacerbated for voids because the cavity structure can abruptly deform the APT tip morphology. Overall, APT data analysis is challenging for materials that cause the APT tip surface to deviate from a hemisphere during the experiment and is particularly challenging for void-containing materials.

A few studies have employed APT to analyze voids in materials; unfortunately, the results suggest that producing accurate reconstructions near these cavity structures and interpreting the 3D data are challenging17,18,19,20,21. Larson et al.17 indicated that the local magnification effect near a void can introduce large aberrations in the atom positions that broaden the void using field-evaporation simulations. Edmondson et al.19 studied nanosized bubbles in ferritic alloys using APT and found that the bubbles could be visualized using iso-density surfaces at approximately half the atomic density of the alloy, indicating bubbles were shown as low-density regions. Porous zeolite catalysts were recently investigated using APT by two independent research teams22,23,24,25. Although pores have zero atomic density inside, i.e., empty space, no such regions were found in the APT reconstructions23. Rather, spatial correlations of regions with high densities and high concentrations of segregated elements in the proximity of the pores suggest that pores may present as high-density regions in the 3D reconstruction. Thus, it remains unclear whether voids will appear as low-density or high-density regions in APT reconstructions. To obtain accurate compositional information around nanovoids using APT, it is necessary to more thoroughly understand the field-evaporation process near voids and its effects on the atom coordinates during APT data reconstruction.

To this end, we conducted correlative APT-STEM experiments that were combined with simulated APT experiments for a series of void-containing alloys. Because visualizing voids in an APT reconstruction is a concern, a needle-shaped APT tip was imaged by STEM prior to conducting the APT experiment. The size and location of each void within the needle-shaped specimen were accurately determined from STEM images acquired at different rotation angles. The same tip was then field-evaporated and analyzed using APT such that the local atomic density variations shown in the APT reconstruction could be directly correlated with the voids observed in the STEM images. A simulated APT data set reproduced and explained the observed atomic density variations by using experimentally measured void sizes and compositional segregation as input parameters. Based on the insights obtained from the experiments and simulations, we provide approaches for interpreting compositional and dimensional information of nanovoids using APT, in which the compositional measurements can achieve a much higher accuracy than conventional STEM-based EDS and EELS measurements.

For any more information, please log on https://www.nature.com/articles/s41467-020-14832-w