Some sequential fits will be for a series of datasets where the same refinement settings will be optimal for all of the histograms. The tutorial is an example of that. However, this is not always the case. As one example, if during a set of diffraction measurements, a new phase appears, you will need to change the refinement settings to include that phase at the point in the sequence where that phase is first seen. When the phase first appears, there may be too few peaks to refine more than the phase fraction. Once more peaks are present, you will likely wish to refine the lattice parameters. At a later stage, you may want to refine structural and/or sample broadening terms.
Unless your datasets are all very similar, you will likely want to add parameters to the model as the refinement progresses. Note that after you set the refinement flags in the data window for any histogram or HAP data tree entry, there is a menu command for “Copy flags.” This will allow you to set the same refinement flags for other histograms without having to visit the data tree entry for each histogram. The hard part is remembering to use the “Copy flags” menu command, but if you fail to do this, you can simply copy them in the next refinement. Note that in most of the menus, there is also a “Copy” and a “Copy Selected” menu command. The “Copy” or “Copy parameters” will copy all parameters and their refinement flags (with the exception that for Sample Parameters, only the refined values and a few other items, such as instrument name are copied; values that are expected to change such as temperature, setting angles, etc. are not copied.) You will be asked to which histograms to copy the values. The “Copy Selected” menu command offers a choice of which parameters to copy as well as the histograms to where they will be copied. For both cases the values that will be copied are coming from the currently-selected histogram.
There are two schools of thought about how to perform sequential fits. One is to review the data and look for where there are significant changes and to use that knowledge to group the histograms into sections that are likely to work well with the same refinement settings. The other is to find reasonable settings for the fit of the first histogram and then let the sequential fit go and see where it starts to fail as an indication of where the refinement settings need to change. Both approaches work. Personally, I prefer to know more about what is going on in my data.