Match! Tips & Tricks
The information given below will provide you with a lot of background knowledge that will help you to apply the Match! software more efficiently:
The most important issue you should keep in mind when using Match! is that the
quality of your results strongly depends on the quality of the peak data that are derived
from your initial raw data during the so-called "raw data processing" step. For example,
if your peak data contain 2theta-errors (e.g. caused by sample displacement or 2theta-shift
of the detector), or if there are far too many peaks detected because your raw data contain
a lot of noise, you most certainly will get poor results (if any).
The background is that during the first step of the phase identification process (search),
only the peak data are compared to the reference database. This is reasonable since the
reference databases themselves (like the PDF or the COD) also contain peak data only.
Hence, it should be clear that the quality of your peak data is most important to get good results.
What does this mean in practice? Although the determination of the peak data from the raw
data can be run automatically using Match!, the computer is still not able to equal the "eye"
of an experienced powder diffraction specialist, who is normally much better in distinguishing
between noise and peaks. Hence, it is definitely worth while to invest some time in a critical evaluation of the peak data with regard to the raw data. Match! offers several functions which make this procedure as comfortable as possible.
Here are some suggestions for a reasonable workflow:
- Once you have imported the raw diffraction data, you should subtract the alpha2-contribution
if necessary (menu "Pattern/Subtract alpha2"). If the pattern looks very rugged, you might also
want to run one or two steps of "raw data smoothing (menu "Pattern/Raw data smoothing"), but
keep in mind that small peaks might be extinguished if you overdue it, resulting in the potential
loss of minority phases.
- If there is a strong background radiation, you should check and probably correct the background
curve that was initially determined automatically. In order to do so, first make sure that the
background curve is currently displayed in the diffraction pattern, e.g. by selecting the
corresponding option from the pattern graphic's context menu (which can be opened by pressing the
right mouse button within the pattern graphics). Afterwards, you can shift the background control
points (small squares on the background curve) using the mouse, and/or add more control points.
Please have a look at the chapter "Table of Mouse Button and Key Combinations" in the online help (or
on p. 64 of the manual's appendix) in order to learn how to modify the background control points
using the mouse.
- When the pattern looks all right, run a peak search (menu "Pattern/Peak searching") and/or a
profile fitting calculation (menu "Pattern/Profile fitting), in order to get an initial set of
- Now it is time to closely inspect your data: Zoom e.g. into the left-hand side (low 2theta
region) of the pattern, and inspect peaks and their positions while slowly moving to the right
(high 2theta region). Match! offers a variety of keyboard shortcuts and mouse operation options
for zooming and tracking; please have a look at the corresponding chapter "Table of Mouse
Button and Key Combinations" in the online help (or on p. 64 in the manual's appendix).
Add or delete peaks while scrolling through the enlarged pattern, and maybe fit individual peaks
to the profile data in order to obtain accurate intensity values.
- Take care that all maxima in the raw data that you estimate as being "real" peaks are "covered"
by peaks, while at the same time using as few peaks as possible. Keep in mind that the "empty"
spaces between the peaks are as important to exclude non-matching phases as the peaks are important
to find the matching ones! Hence, by only using the peaks that you are really certain about, you
can increase the separative power of the search-match calculation.
- Once you have investigated the whole pattern and have settled on a set of peaks, check and
(if necessary) correct the specimen displacement error as well as the zero point, using the
corresponding commands in the "Pattern" menu. Use the histograms to decide on a certain correction
value, and do not hesitate to come back and try a different correction if the search-match calculation
did not give a reasonable result.
- In order to correct potential 2theta errors, you can also use an internal standard if you know
that a certain phase is present in the sample. Simply press <Ctrl+F> (<Cmd+F> on the Mac) (or click
into the corresponding field "Find phases/entries" to the right of the toolbar at the top), then
type the name, the formula sum or the entry number of the known compounds, and finally press <Return>.
Match! will make sure that the corresponding entries are displayed in the candidate list at the
bottom of the screen, and then automatically highlight the first entry matching your criterion
(e.g. name or formula sum). As a result, the corresponding diffraction pattern will be displayed
in the pattern graphics above. You can now adjust the 2theta axis according to the highlighted
internal standard, simply by pressing <Ctrl+T> (<Cmd+T> on the Mac).
- At this point, you are now ready to run the search-match calculation, investigate the resulting
entries, and select the matching phases.
Using the entry searching and restrainting facilities of Match! is rather straight forward in most cases,
however, there may be circumstances in which an in depth knowledge of the logical concepts behind the
"Restraints", "Additional entries" and "Find phases/entries" functionality may be helpful:
First of all, the candidate list in the bottom left corner of the Match! program window displays a list of
all entries in the current answer set, the entry number of which is always displayed in the status bar at
the bottom to the left of the reference database name (in the corner on the lower right-hand side).
Right after you have started Match! or run the "File/New" command, the
"current answer set" contains all entries from the current reference database. Typically, these are 40.000 to
400.000 entries, so it would take a lot of time and memory (and wouldn't even make much sense) to load and display
the entry data of all reference database entries into the candidate list. Hence, Match! displays a message instead,
asking you to do something in order to delimit this large amount of entries in order to find out which entries
you are actually interested in, e.g. to import diffraction data or enter selection criteria (restraints).
For example, if you import diffraction data and run a search-match calculation, a figure-of-merit value (FoM)
is calculated for each entry, so the criterion "Minimum FoM" in the lower left-hand corner of the "Search-Match"
tab of the "Options" dialog can be applied to restrict the answer set to only those entries whose figure-of-merit
is above this user defined value, in order to select the entries that best match your diffraction pattern.
Another possibility to delimit the full amount of reference database entries to the ones you are interested in
is to enter selection criteria (e.g. name) on the "Restraints" tab on the right-hand side. This tab offers a large
amount of selection criteria like elements that must or must not be present.
Some of these selection criteria are text fields, like e.g. "Name". If you enter some text here, only reference
database entries that exactly contain this text in the corresponding database field are accepted to be included
in the answer set. In order to help you here, Match! can display a list of all corresponding values in the reference
database so that you can directly select from this list. This "List selection box" is displayed by pressing the
corresponding button to the right of a text field input line. You can also use the wild card character '*' to
tell Match! to disregard text before and/or after the fragment you have entered (e.g. "*corundum*"), however,
Match! still distinguishes between capital and small characters within the fragments.
Once you have delimited the full amount of reference database entries in one or the other way, you have obtained
an answer set (a set of selected entries below the full reference database) the entries of which can now be displayed
in the candidate list. However, there is another limitation for this, i.e. the "Max. number of entries" setting at
the top of the "Candidate list" tab sheet of the "Options" dialog. Normally, the candidate list will only be filled
with entries if the current answer set does not contain more than this number of entries. If the number of answer
set entries is larger than this value, a corresponding message "Too many matching entries" is displayed in the candidate
list instead of the entries, instructing you to further delimit the current answer set e.g. by entering additional entries.
So far for the candidate list, the current answer set and the "Restraints" tab.
The "Add. entries" (additional entries) tab does something different: While normally the current answer set (->
candidate list) only contains entries that match all "Restraints" criteria as well as the "Minimum FoM" criterion (if
applicable), the "Add entries" tab allows you to add specific entries regardless of these criteria. This may be useful
for example if you know that a certain phase/entry is present in your sample but has not been assigned a figure-of-merit
value larger than "Minimum FoM" in the last search-match calculation. Hence, you most probably would like to add this
entry to the answer set/candidate list, in order to check what caused the low FoM value, like missing/undetected peaks,
a strong 2theta shift, or a very low intensity scaling factor.
In order to do so, you can enter selection criteria (e.g. name) on the "Add. entries" tab, just like on the "Restraints"
tab. The number of entries that match these selection criteria is displayed in brackets behind "Add. entries" on the tab;
these entries are then added to the ones selected by "Restraints" and/or the "Minimum FoM" criterion. Typically, you
will find these additional entries at the bottom of the candidate list because their figure-of-merit values are low (quite
often close to 0.0).
As you can imagine from these explanations, the usage of the "Add. entries" tab only makes sense if the current answer
set does not already contain the full reference database (in which case there would be no more entries that could be added).
Hence, the "Add entries" tab is inactive in this case in some earlier versions of Match!.
The "Find phase/entry(s)" functionality in the top right corner of the Match! window is pretty much some kind of
mixture between these two facilities ("Restraints" and "Add. entries"): Once you have entered some criterion (e.g. name)
in the corresponding line at the top and pressed <Enter>, Match! searches for matching entries in the reference database.
It then checks if these entries are present in the current answer set. If this is not the case, it adds these matching
entries, using the "Add. entries" mechanism, by automatically entering the corresponding entry numbers on the "References"
sub-tab of the "Add. entries" tab.
Afterwards, Match! locates the first of these matching entries (typically the one with the highest FoM-value) in the
candidate list (answer set), and makes sure that it is marked and hence displayed in the powder diffraction pattern.
Hence, this functionality is some kind of abbreviation to the much more detailed "Add. entries" mechanism and at the
same time quickly locates the entry(s) of interest in the candidate list. In order to account for this "quick searching"
character of the functionality, it is not required that the name entered in the "Find phases/entries" field is matched
exactly (like it is the case on the "Add. entries" and "Restraints" tab). Instead, it is sufficient that an entry to be
regarded as "matching" contains the entered name as a fragment, also neglecting small and capital letters (in contrast
to the criterion "Name" on the "Restraints" and "Add. entries" tabs).
This latter feature is something that might have caused some misunderstanding in some cases: For example, the current
"COD-Inorg" reference database contains 26 entries with name "Corundum" and 1 entry with name "corundum", i.e. with a
small letter at first position. As has already been mentioned, this difference does not matter when using the
"Find phases/entries" line, so 27 entries are added using the "Add. entries" mechanism.
However, this mechanism does not delimit the current answer set, like it would be the case if you used "Restraints",
it just makes sure the corresponding entries are present (maybe among others) in the current answer set, and
marks/displays the first one of these entries. If the current number of entries in the answer set is larger than the
corresponding parameter (so that normally nothing would be displayed), Match! uses an internal "work-around" mechanism
to nevertheless display a subset of entries (that also contain the ones matching the "Find phases/entries" criterion)
in the candidate list.
If you use the wild card expression "*corundum*" on the "Restraints" or "Add. entries" tab, Match! nevertheless
still distinguishes between capital and small characters. As has already been mentioned, there is only 1 entry with
a small 'c' and 26 with a capital 'C' at the beginning ("corundum" vs. "Corundum"), so that's why you find a single
entry when you enter "*corundum*", 26 entries when you enter "*Corundum", and 27 entries when you give "*orundum*" instead.
In certain circumstances it may be difficult to identify minor phases if one or more of their three
strongest peaks are not observed (or not found by the peak searching function) within the experimental
range. The background is that normally a reference pattern is only accepted as "candidate" if at least
the three strongest peaks are observed (i.e. can be correlated to the experimental diffraction pattern).
If you are in doubt that you might be missing one or more minor phases, you should deactivate the option
"Check 3 strongest peaks before running full match" on the "Search-Match" tab of the "Options" dialog.
Afterwards, please run a new search-match calculation, e.g. by selecting the "Search-Match" command from
the "Search" menu.
The drawback of deactivating this option is that the search-match calculation may take significantly longer.
Another option you should keep in mind when trying to identify minor phases is to check if there are peaks
in the raw diffraction data that have not been detected by Match!. You might want to try to increase the
peak searching sensitivity either by pressing the "F2" button on your keyboard or by selecting the corresponding
command from the "Pattern/Peak searching" submenu.
If you have to do so for many of your samples, you should consider to increase the default peak searching
sensitivity on the "Raw data" page of the "Options" dialog (which can be opened from the "Tools" menu).
If you are using a reference database that does not contain atomic coordinates for all entries, you might
face the problem how to create a cif-file providing the full crystal structure data for Rietveld refinement.
If you cannot export a reasonable cif-file e.g. from Pearson's Crystal Data, you might want to use a
crystal structure description from a corresponding scientific paper. However, in many cases there is no cif-file
available for this paper, so you have to create one on your own by manually entering the
data from the paper. The problem is that the cif-syntax is far from being trivial.
Fortunately, you can use the Match! User Database Manager to create the cif-file. In order to do
so, please open it e.g. by selecting the corresponding command from the "Tools" menu.
Once the dialog is displayed, please press the "Add" button in the toolbar at
the top of the window. The tabs at the bottom of the dialog will be activated,
with the "General" tab in front. Please click on the "Crystal structure" tab
to make it visible, then start to enter your crystallographic data by first
marking the appropriate crystal system. Afterwards, select the space group
from the corresponding combobox below, and fill in the unit cell parameters.
Finally, press the "Add" button below the table of atoms to the right, and
enter the element and coordinates of the first atom. Finish the input of the
first atom's parameters by pressing <Return>, then repeat this step
for the remaining atoms. Once you have finished entering the atomic parameters,
press the "Save changes" button at the bottom of the dialog.
Make sure that the new entry is marked in the table at the top, then press
the "Export" button in the toolbar at the top of the dialog. Select "CIF file (*.cif)"
as "Files of type:", select an appropriate directory and file name, and press "Save".
That's it! You can now leave the User Database Manager by pressing the "Quit" button
(you can save the new user database at this step if you like), and import
the cif-file as a response to the corresponding question next time you try to
run a Rietveld refinement.
If you are frequently using the same set of restraints (e.g. elements, density, inorganics only
etc.), you might get tired of
selecting these restraints in the corresponding dialog over and over again. Instead, it would
be much easier if you could somehow store the entries that correspond to these restraints,
in order to use them as a subset based on which you can
run search-match operations later on.
You can easily save the current selection criteria, and recall them later on with just
two mouse clicks, using the so-called "selection criteria presets".
You will find the corresponding control elements at the bottom of the "Restraints" (and also
"Additional entries") tab to the right of "Preset". Of course, it is possible to save several sets of selection criteria.
In order to save the current selection criteria, simply press the "Save" button.
Afterwards, you have to enter a name using which you can easily identify this set of
selection criteria later on ("Silicon compounds" in our example). Your selection criteria
will then be stored in your personal settings (e.g. Windows registry).
If another person is using the same computer, he can store his own selection criteria
independent from yours (of course assuming that he is using his own account on this machine...).
If you would like to recall a certain set of selection criteria, simply press the "arrow down" button
of the combo box to the left of the "Save" button, and select the name of the desired selection criteria
from the list that opens. That's it! The selection criteria that have been loaded will instantly be applied to the table of
candidate entries (candidate list), and of course also to the entries that are investigated during
the search-match calculation.
On the "Restraints" and "Additional entries" tabsheets, a so-called "list selection box" can be
opened in each case where a manual/keyboard
input of a numerical or text value is required, i.e. in each case where an edit/input line
is present. In these cases, you can open a list of all available values for the
corresponding field (the list selection box) by clicking on the
corresponding "table" button to the right of the input line.
You can easily select the value(s) you are interested in by clicking on the corresponding
lines. Besides this, you can shorten the search for a particular value/line by rapidly
typing the first few letters. For instance,
if you are looking for an author "Stolpovskaya V.", you can simply type stolp
without pausing between the letters, and the corresponding line will instantly be
If you made a mistake by typing a wrong letter, simply wait for two seconds (which
will automatically clear the input buffer) and try again.
If your search-match calculation gives a large number of candidate
entries, there may be two reasons for this:
- Your diffraction pattern contains a large number of
peaks. As a result, the probability
that there is an acceptable agreement between reference peak positions and the experimental
peaks is rather high, so a large number of entries may have an acceptable figure-of-merit (FoM)
Please inspect the peaks carefully one by one, and delete all those peaks that you are not
really sure about. Afterwards, run the search-match again.
- The "peak correlation 2theta window" is rather large. Normally, this window is adjusted
by Match! automatically before the search-match calculation, based on the average FWHM
(full width at half maximum) of the experimental peaks. As a consequence, broad peaks (large
average FWHM) result in large peak correlation windows, so a large number of peaks (entries)
is considered as "matching".
You can disable the automatic determination of the 2theta peak correlation window size on the
lower right-hand side ("Delta 2theta") of the "Search-Match" page of the "Tools/Options"
dialog, by deselecting the "Auto" check box below. Try to use/adjust a smaller value for
"Delta 2theta", press "Ok", and run the search-match again.
In general you should avoid to apply smoothing to your raw data. Normally it is not necessary and can cause
severe issues like peak extinction or convergence problems in Rietveld refinement (at least if you exaggerate).
If there really is a lot of noise present in your raw data (so that raw data smoothing can hardly
be avoided if you would like to keep the number of detected peaks at an acceptable rate),
please apply the raw data smoothing gently and carefully!
If far too many diffraction peaks are frequently determined during peak searching,
we recommend to enable the raw data smoothing for the
automatic raw data processing in the corresponding Options dialog page.
Start with a single "Number of repetitions" and maybe increase this number if the result is not sufficient.
Please keep in mind that if too many peaks are determined, this may cause difficulties when determining the zero-point error and
lead to more "false positive" phases in the candidate list after the search-match.
If too few peaks are detected, this may lead to less entries in the candidate list after the search-match,
due to a better separation between phases that may be present and those that are unlikely. However,
it is more likely that especially minor phases are not determined in this case.
- If some shoulder peaks are not detected, this may be caused by the fact that when detecting
a new peak (based on a second-derivative minimum), Match! searches for a maximum in the raw
data within two times the FWHM. This can cause shoulder peaks to be neglected if the default FWHM value is relatively large.
You should adjust the default FWHM value on the
"Raw data" tab of the "Options" dialog (maybe you have to deactivate the corresponding "Automatic" option in order to be able to modify the default FWHM value).
- If peak positions are not determined exactly, this may be caused by a relatively large step width in your raw/profile data. In order to fix this problem, you could try to increase the resolution
(i.e. decrease the step width) by interpolation, using the menu command "Pattern/Increase resolution...". You should start by trying a small factor (e.g. 2 or 3) first, because otherwise the noise may be "fixed" (i.e. fluctuations are described by too many data points, so they are no longer detected as "random noise" but as a "signal"). Repeat the peak detection after increasing the resolution.
- The best way to cope with missing or surplus peaks is still to visually inspect the diffraction pattern on your own and to decide about the presence or absence of a peak guided by both scientific experience and intuition.
This can be done in a very comfortable way using Match!'s zoom or zoom/track and peak editing funcionalities, as it
is demonstrated in our tutorial video. A careful check and probably correction of the peaks will safe a lot of time that can be wasted in performing a search-match on bad peak data!
Sometimes the pattern graphics in the report is split to two pages; unfortunately, this cannot
be prohibited from within Match!. As a work-around, please save the report as an HTML-file to
some temporary directory, then open or print this file using your web browser.