Methods for Tomographic Segmentation in Pseudo-Cylindrical Coordinates for Bobbin-Type Batteries

Abstract

High-resolution X-ray computed tomography (CT) has become an invaluable tool in battery research for its ability to probe phase distributions in sealed samples. The Cartesian coordinates used in describing the CT image stack are not appropriate for understanding radial dependencies, like that seen in bobbin-type batteries. The most prominent of these bobbin-type batteries is alkaline Zn-MnO₂, which dominates the primary battery market. To understand material radial dependencies within these batteries, a method is presented to approximate the Cartesian coordinates of CT data into pseudo-cylindrical coordinates. This is important because radial volume fractions are the output of computational battery models, and this will allow the correlation of a battery model to CT data. A selection of 10 anodes inside Zn-MnO₂ AA batteries are used to demonstrate the method. For these, the pseudo-radius is defined as the relative distance in the anode between the central current collecting pin and the separator. Using these anodes, we validate that this method results in averaged one-dimensional material profiles that, when compared to other methods, show a better quantitative match to individual local slices of the anodes in the polar θ-direction. The other methods tested are methods that average to an absolute center point based on either the pin or the separator. The pseudo-cylindrical method also corrects for slight asymmetries observed in bobbin-type batteries because the pin is often slightly off-center and the separator often has a noncircular shape.

Figure 1

(a) Geometry of a bobbin-type cylindrical Zn–MnO₂ battery. (b) Schematic representation of the microstructure and discharge reactions of the Zn anode. (c) Schematic representation of the microstructure and discharge reactions of the MnO₂ cathode.

Figure 2

Diagram of the setup for CT experiments at beamline 6-BM-A at Argonne National Laboratory. The axes shown are the APS coordinate system, which was different than the coordinate system used in subsequent battery analysis.

Figure 3

Segmented Zn–MnO₂ X-ray CT results. (a) Segmented 3D volume of an AA anode pulse discharged at 750 mA to 1420 mA h (cell #10). The data has been cut away to show the interior, and a black line has been added to visually define the edge. (b) Cross section of a pristine AA anode (cell #1). (c) Cross section of the same AA anode shown in panel (a) (cell #10).

Figure 4

Terminology used in the analysis. (a) A single anode cross section, which is a 2.93 μm high slice along the battery z-direction or z-slice. To analyze homogeneity in the polar θ-direction, cross sections were divided into θ-slices. (b) A data set for one anode contained 400 cross sections. These stacked along the battery z-direction to form a 3D volume. The total height probed was thus 1.172 mm.

Figure 5

Methods for determining the radial position of a voxel at (x,y) marked with an orange circle. The centers of the pin (x_0,pin,y_0,pin) and separator (x_0,sep,y_0,sep) are marked with blue and red lines. The size of the orange circle is exaggerated for viewing. (a) Measuring the absolute distance from either the current collecting pin or center of the separator. (b) Measuring the relative position between the current collecting pin and the edge of the anode (marked by the tripled white line).

Figure 6

Cross sections of segmented CT data with false-colored phases. (a) Anode continuously discharged at 250 mA to 1420 mA h (cell #6), with a current collecting pin that was 105 μm off-center. (b) Anode pulse discharged at 750 mA to 1420 mA h (cell #10), with a current collecting pin that was 183 μm off-center.

Figure 7

Radial volume fraction profiles for Zn and ZnO within partially discharged anodes, using the three methods described in Figure 5. (a) The anode in Figure 6a (cell #6), with a current collecting pin slightly off-center. (b) The anode in Figure 6b (cell #10), with a current collecting pin significantly off-center.

Figure 8

Segmented θ-slices of a partially discharged anode. (a) False-colored segmented CT data for a battery pulsed discharged at 750 mA to 1420 mA h (cell #10), annotated to show eight θ-slices. (b) Smoothed 1D radial volume fraction profile of ZnO for each of the eight θ-slices and their average.

Figure 9

Mean Euclidean distances for the three methods of determining the radial phase volume profiles. The symbols used match those reported in Table 1.

Figure 10

False-colored segmented cross section of the first cell in Figure 9 continuously discharged to 1420 mA h (cell #3). The anode was divided into eight θ-slices along θ_sep, with a significant void defect in slices 7 and 8.

Figure 11

Smoothed radial phase volume fractions of the cell featured in Figure 10 (cell #3), with θ-slices 7 and 8 showing significant deviation in pore and ZnO content from the average θ-slice due to a manufacturing defect.