The autoindexing problem requires the most precise possible peak positions, which are obtained by peak fitting. The process for fitting has been presented in Chapter 17. Peak fitting is performed from the “Peak List” data tree subentry of a histogram and results from the fitting are found as well in the “Peak List” data tree entry. When performing peak fitting, it helps to be on the lookout for peaks with shoulders from a nearly overlapped reflection. Inclusion of the second peak, even when the peaks have minimal separation will provide peak positions that are closer to the reflection positions. When fit as a single peak, the reported position will be something of a weighted average of the overlapped reflections.
The peaks that will be used for autoindexing are found in a separate data tree subentry, “Index Peak List.” The reason that this second list is present is that this allows for selection of peaks and possible editing as well as tracking the assignment of hkl indices. Use the Operations/“Load/Reload” command to copy the peak positions from the “Peak List” to the “Index Peak List”. Note that each peak in this list has a “Use” flag that can be turned off to omit the peak position from the indexing process.
You want to use some care with the choice and number of peak positions. The lowest Q peaks are typically key for indexing as they help define the length of the axes. Autoindexing typically requires a minimum of 20 peak positions but inclusion of more than \(\sim 30\) peaks usually makes indexing more difficult. It is usually better to leave out “suspect” peaks than include them. What should make you suspicious about a peak? Certainly if you see a peak that is much narrower than the others, this could be arising from a single crystal that is not representative of the sample. Significantly broader peaks could be from a more poorly crystallized impurity or could be from more than one partly overlapped reflection. In this latter case, since the position matches neither reflection, including that will likely degrade the indexing process.
Once you have populated the “Index Peak List” and have selected which peaks to use, you can select the “Unit Cells List” data tree tab. The upper part of the resulting window has options for autoindexing and the process used is documented in the “Fit Peaks/Autoindexing in GSAS-II” tutorial (https://advancedphotonsource.github.io/GSAS-II-tutorials/FitPeaks/Fit∖%20Peaks.htm). While it is possible to attempt indexing in all the Bravais lattice options at one time, I recommend using a small number at a time.
Trial unit cells located in the autoindexing process are placed in the lower section of the “Index Peak List” window. You will then want to review the located cells visually, to see how well each cell indexes the pattern. When you select a unit cell from this lower list, it is loaded into the middle portion of the window where the reflections generated by that cell can be visualized superimposed on the observed powder diffraction pattern. Note when the “Unit Cells List” data tree entry is selected, the peak positions from the “Index Peak List” window are shown as blue lines, when the “Use” flag is selected and are not shown otherwise. The lines generated from the selected unit cell are shown as a orange dashed line. Note that the upper data limit is also shown as a red dashed line, but the dashes have double the spacing. Thus, when looking at the plot, lines that have red dashes superimposed on blue lines indicate locations where the observed peaks have been indexed. A blue line that does not have red dashes is an unindexed line – you will want to make note of these and this can indicate the location of an impurity or that the unit cell is a subcell of one that will index all the lines, but could also be just a cell that does not match the pattern. Locations that have only the orange dashed lines indicate generated reflection positions that do not have associated peaks. Sometimes you will notice small peaks in those locations once you know where to look. An indexed location without intensity is not in itself a serious problem. It could be an extinct reflection or may just be allowed but very weak, but too many unobserved reflections is probably an indication that the cell being evaluated is a supercell of the correct unit cell.
Note the keep column in the unit cells list. If you find a cell that shows promise, check this. Then if any subsequent searches are made, those entries will be retained while all others are deleted to make room for the new search results.