Preferred Orientation Analysis using the

Rietveld method: a simple MAUD Tutorial

Last updated: 20-Mar-08.

This tutorial tries to explain in a simple way to obtain Pole Figures from XRPD data. I’ll assume
you have already done your experiment: you have collected your sample data and you have a
good standard material data collection too.

A (very) few words about MAUD:

MAUD is a software written by Luca Lutterotti (http://www.ing.unitn.it/~maud/index.html) and is
freely available for download from his website. It is a very useful tool for Rietveld Analysis and
implements powerful algorithms for preferred orientation analysis as well.
In this tutorial the version 2.058 of the software is used. That is the very last version at the time
I’m writing this tutorial (note: during the last revision the latest version is 2.062). I’m using it on a
MacOSX system, but MAUD works on windows and linux as well.

Input files preparation

The standard format used by MAUD to input XRPD profiles is the .esg. There are two convenient
ways to prepare these kinds of files from your image data (I’ll assume you have MAR data or
different Image Plate data):
1) You can use Fit2D (http://www.esrf.eu/computing/scientific/FIT2D/) coupled with a useful
Java script from Sebastien Merkel: fit2d2maud
(http://merkel.zoneo.net/RDX/index.php?n=Software.Fit2d2maud)
For the preparation of the .esg files using this method I invite you to go to Merkel’s
website and follow carefully his tutorial: it is very clear and it will follow you step-by-step
in data integration and files conversion.
2) You can input the images file directly in MAUD and obtain directly the .esg files you need
using a ImageJ script (an image processing software embedded into MAUD). To access
this method you simply edit (eye icon) your dataset, select “datafiles” from the tab above,
click the “image manager” button on the right (this will open ImageJ), open your image
file (from the menu), select “Plugins -> Multispectra from Transmission Image” and then
you can setup your data from the panel and then integrate.

Which one is the best? I’m used to obtain my .esg files with Merkel’s method for different
reasons: I started using his tutorial, I already used Fit2D, the second method didn’t work on my
laptop , I can make some corrections directly in Fit2D to feed MAUD with better data. The
second method btw is much faster and the corrections you make with Fit2D like detector tilt,
beam center, sample-detector distance… can be done during the refinement. You can refine
these parameters with .esg obtained with Fit2D too: you’ll have a better initial guess…
To end this paragraph I invite you to go to Merkel’s website and check his nice tutorials. A good
idea would be to check the tutorials that are present in MAUD webpages too! The present tutorial
will follow +/- Merkel’s style, but it will focus on Rietveld refinement and LPO studies step-by-step
and will avoid the data input preparation since it would be difficult to be clearer than him.
So, I’ll assume you have your .esg data (standard and sample) ready to be imported into MAUD
and you have all the instrumental parameters and sample data you need…

This tutorial will follow step-by-step a LPO (Lattice Preferred Orientation) Rietveld analysis on a
clay mixture. Starting from the calibration with a standard.
If you find errors or steps that are no precise/clear enough, please tell me at
[email protected] so I can improve the tutorial.

1
STEP 1: Loading standard data
In this experiment we have collected CeO2 as the standard. A good data collection on this
material has been used in Fit2D to refine beam center, sample detector distance and detector
non-orthogonality. You should also have the .esg file concerning this data collection.
To do this analysis you need a folder that contains: The datafile you obtained from fit2d2maud or
from MAUD image manager (I have CeO2.esg, obtained from the first one), and a phase file that
describes you phase information (symmetry, atoms positions…) that is a .cif file (I have ceria.cif).
These .cif files can be obtained from commercial databases (es. ICSD), from the COD database
(http://sdpd.univ-lemans.fr/cod/), or from the little database that comes with the MAUD software
(structures.mdb). If you don’t have a .cif file you can input your structure directly into MAUD by
hand… This case is not so frequent, luckily.
OK, now that you have everything you need you can run MAUD (for installation requirements
have a look at MAUD website). I’m doing my analysis while writing, so for you it is like following
my job while reading: I think it is a good approach. And it’s much faster…
Once MAUD is started you need to load your ceria (or whatever standard you used) .esg file:
select “Datasets” (1) on the tab and select “DatafileSet_x” (2) in the panel, then click the “eye”
button (3) on the top. From now I’ll refer at these actions as “Edit the DatafileSet_x”, since the
“eye” (“edit”) button will be used quite often.

You could change the DataFileSet_x name by double-clicking it.

On the new window select “Datafiles” from the tab and click the “browse” button. Then locate and
select your standard .esg file (CeO2.esg).
Here you can set the tilt angle information: in our sample you should set “phi” to 0 (default). More
information about sample position will be given later.
In this panel you can also see the “Image Manager” button you can use to create .esg files.

2
Once you have loaded the datafiles you have to setup you instrumental parameters: click on
“General” on the tab at the top and click the “Edit” button in the “Instrument” panel.

In the “Diffraction Instrument” window you have to set some parameters:

3
My samples have been measured in a synchrotron beamline using a MAR detector. I’m giving an
instrument name (“Clays2”: because we are studying a clay sample).
- For Intensity calibration select “none cal”.
- For Angular calibration select “Flat Image Transmission” and then click the “options” button: set
the refined sample-detector distance you got from Fit2D (in mm) or the sample detector distance
estimated at the beamline, if you used MAUD image manager (in cm, in this case!).
- For Geometry I use the Lorentz corrected transmission image.
- Measurement is 2Theta.
- Source: select “synchrotron” and set the proper wavelength clicking the “option” button.
- Detector: you can leave Scintillation.
- Instrument broadening: leave the Caglioti PseudoVoigt but click the “options” button and set the
Asimmetry coefficients to 0 (Asimmetry tab), otherwise you’ll get weird peak shapes, in this case.

Now it is would be wise to save your analysis: MAUD uses .par extensions for its analysis files. I’ll
call my analysis “CeO2_standard.par”.

At this time I’m also used to calculate and store an averaged profile: it is useful for preliminary
analyses, phases identification (if you have any doubts: you should already know the mineral
phases in your sample)… Refining a single profile is much faster than refining many at one time!
And data obtained from a single profile can be used as a good starting guess when you analyze
the whole dataset.
To do that you have to select all the profiles entries in the “Spectra list” you have when you “edit”
your dataset and press the “Sum spectra” button. Save your file (mine is “CeO2_profile.cif”),
remove all spectra from the panel and browse for your new datafile to load it. Now you can use it
to check what you need to quickly.
For this standard we don’t need to check anything so we can continue with the multiple spectra
dataset.
Looking at the main window you can see the averaged profile: it would be a good idea to set a
proper 2-theta range for the analysis. You can do that editing the dataset and setting proper
values in the “Computation range” panel. I’ll use 1.6 and 4.7°.

Now we need to input phase data: we are working on a standard so e need to load only ceria
data. Click the “Load an object from a (CIF) database or file” (the “cylinder” icon on the left),

4
browse to select the “ceria.cif” file, select the proper phase in the new window and click the
“choose” button. Now your phase is loaded (you can see it in the panel on the top-left.

Now the preliminary steps are complete. It is time to save, and to check your input data.
Especially useful would be to look at the “Plot 2D” in the main window to look for anything
anomalous. If one, or more, of your spectra show anomalous zones, peaks, whatever problem…
it’d be a good idea to disable it. You can do that editing your dataset and unchecking the
checkbox “enabled” relative to the spectrum (“Selected spectrum options” panel). From this
window you can have also a look at one o more files simply selecting it (or them) and clicking on
the “View” button. This is probably the best method to look for anomalous features in your
spectra.
Note: MAUD comes with a small database of crystal structures. It is called “structures.mdb” and is
located in the MAUD installation directory. You can browse, open it and select the phase you
need, if present.

5
Now we can set the sample position, this depends on your data collection geometry: in our case
we need to choose the “sample” tab, edit it, go to the “sample position” tab, and add proper
values (for me 90 for omega and chi). With correct values you’ll see the maxima in pole figures
with the same orientation as your physical sample. In case you need to apply any symmetry to
your ODF you should position your sample accordingly (e.g. for axial symmetry, with es. [001]
axis, you need the (001) maximum at the center of your 001 pole figure)

STEP 2: Starting the standard sample refinement

Now you are in MAUD main window. The first thing to do is to calculate a spectrum: to do so you
simply press the “calculator” button on the top. Probably your results will be very bad: don’t worry,
it’s normal. The first thing to check is the intensity. In my sample the calculated spectrum is by far
too intense. You can adjust it manually by going in the lower panel, open the DatafileSet_x menu,
open Clay2 menu (the name I gave to the instrument), select the Intensity (scale factor) line and
play with the arrows to change intensity values.

6
Now you see that there’s no background, the peak profiles are wrong and the peak positions are
not very well estimated. We’ll correct this quickly…
Let’s start with the peak positions. I have previously refined beam center, sample-detector
distance and detector tilt with Fit2D. Results are pretty close to the correct ones but we need to
refine them again in MAUD.
To set the parameters to refine there are some different techniques: I’m used to open a window
(Apple+L in OSX, Ctrl+L in MS Windows and linux) and set all the values that need to be refined
here. You can do that in the lower panel in the main window or by right-clicking values in the
windows. I find easier to use the “tree table window”. You can access this also from the main
window menu, Analysis -> Parameters list.
In simple cases like this one you could also use one of the “wizards” available with success.
But let’s refine the Flat Image Transmission options (access local menu by clicking “fixed” on the
right, under the “status” column).

7
Close and go to the main window; now we can refine for a few (in this case 7 is a good value)
cycles: simply click the “hammer” icon-button and set a value around 7 moving the cursor on the
left.
Now our peaks are perfectly centered. But we need a background function, better intensities and
definitely more accurate peak shape parameters.

Let’s adjust background and intensities first. Here the background is quite simple, but I’d add
some more parameters to the polynomial that will describe the background. Edit your dataset,
choose the “Background function” tab and add some more parameters. I’ll add three more.

8
Now you can open the Parameters list window, the one with the tree view, and click on the “Fix all
parameters” button. Then click on “Free backgrounds” and “Free scale pars” buttons. Close and
run a few refinement cycles (hammer button).
The background now is reasonable, as well as intensities, but the peak shape is completely
wrong.

The standard collection is very important to find good values for the instrumental component of
the pseudovoigt profile function. Since it is an instrumental contribution it should be kept fixed for
all the data collected in that particular data collection.
Now we refine these values. We need to open the parameters list window, expand all (button),
and refine the Caglioti (and gaussianity) parameters. Then run the refinment.

9
You can see quite an improvement, but the profile function is still incomplete: we need now to
refine the sample contribution to the peak shape. In MAUD you can do so by refining the
“crystallite size” and “microstrain” parameters. There are different crystallite size and microstrain
parameters: to have a look you can edit your ceria in the “phases” tab, then choose
“microstructure” in the tab in the new window. We need to set a line broadening model and a
size-strain model. For our ceria the default (Delf and Isotropic) choices are good. I have a
personal preference about Popa model, so I’ll use it anyway. You can use the default choice: in
this sample, results should be almost the same.
Since you are measuring instrumental parameters with a standard that should have ideally no
defects you could also set (fixed) the microstrain to 0. This is an assumption…

Once you have set your favorite model, open the parameters list window and set crystallite size
as refined (you can find them in the phase-specific parameters). Run the refinement.

10
The profile fit now improved quite a lot, but it is still not perfect. Now it would be a good idea to
look at the Plot2D in the main window.

The calculated and measured datasets are pretty close, but now we’ll improve the backgrounds a
little. Since each spectrum has a slightly different background we can take it into account adding
some additional background parameters to each spectrum. To do that you have to edit your
dataset, select the datafiles tab, select all your spectra in the “Spectra list” panel, click on “Add
bkg param” button a few times. Each time you press the button a new parameter is set. I’ll click
two times.
The go to the parameter list window and refine scale factor and backgrounds (“Fix all
parameters”, “Free scale pars”, “Free backgrounds” buttons, in sequence). Run the refinement.
You can see an improvement:

11
Now it is time to improve it a little more: you can refine again the Flat Image Transmission
parameters (if you don’t trust too much Fit2D calibration), to run another refinement with peak
profile parameters (instrumental or sample-depending), scale factors, backgrounds. You fit will
improve again, but not too much: this is fine-tuning. In this case it is fundamental NOT to refine
cell parameters: you are working on a standard! Otherwise it would be useless… 
Usually when you work on the single profile instead of a series you get an almost perfect fit. But
I’d like to keep the same conditions I use with my samples i.e. multiple profiles.
This fit is from the spectra series:

Now it is a good idea to store an instrumental parameters file to store the instrumental
parameters you have refined here. I usually create a blank .cif file in the analysis folder. Then I
edit the dataset, click on the “store” button in the “Instrument” panel, and then I select my blank

12
file (I called it clays_inst.cif). It is very useful to do that so you don’t have to re-enter by hand all
the instrumental parameters when you’d need to do that.

Now our refinement using the standard is complete: we have seen that all our parameters were
good and we’ve got refined instrumental parameters. It’s time now to get our hands on a “real”
multiphase sample and to study also the preferred orientation. Now things unfortunately will not
be so easy and straightforward…
Sometimes you’ll find useful to use the wizards MAUD includes. To access them, simply go to the
menu Analysis -> Wizard. They usually work well, but you have to be careful. The main problem I
see when using the wizards is that the B factor is considered a “basic phase parameter”. I usually
don’t refine this parameter with synchrotron data, especially when dealing with complex phases
and/or mixtures. In these cases you can add very large errors in your analysis. My advice is to
avoid the wizards here…

Step 3: loading your sample data and basic refinement

Now it’d be a good idea to create a new folder and put in the sample .esg file(s), the standard .par
file (our CeO2_standard.par e.g.), the .cif files of your phases (optional: if you have your phases
e.g. in a database located in a different directory you can use them. I only prefer to have
everything I use in the same folder).
Summarizing: now in my folder I have: 7 .esg files (I have 7 images at 7 different tilt angles),
several .cif files (I’d like to compare them and choose the better ones: I have several kaolinites
here, e.g.), the renamed standard file (I renamed it “clay1.par”), the instrumental parameter file
(not strictly necessary now, but it could be useful if you need to reset/check some things).
This is one of the images I’ve treated with Fit2D and fit2d2maud (look for Merkel’s tutorial!) to
obtain the .esg files. The texture is pretty evident.

13
Now let’s open the clay1.par file (the CeO2 renamed…). If you use OSX you can simply drag ‘n
drop on MAUD icon. Otherwise you can start MAUD and open your file from the menu.
We need now to load datafiles, so we “edit” the DataFileSet_x to load all 7 the .esg files.
Ok, now we need a little setup. First we need to choose a proper 2-theta interval: the one you
used with the standard probably it’s not the best for the new sample (e.g. in our case we need all
the low 2-theta angle part).
Now I’ll explain a little “trick”. To do the first analyses and speed up the procedure it is useless to
use all the spectra simultaneously: it would take a lot of time. To have a look at our data and
choose the proper mineral phases it is convenient to work on a single profile, averaged. So:
select ALL your spectra in the “Spectra List” box and press the “Sum Spectra” button. Give a
proper name (clay1_profile works well for us) and save. You now have a .cif file that is your
averaged profile. Now remove (button…) all your datafiles and load this new clay1_profile.cif
datafile.
Save your data now (clay1_profile.par , e.g.).

14
So, we have now loaded a single profile. It’s time to check phases. In our example I know that
I’ve a mixture of illite and kaolinite. Please before refining be sure to know exactly the mineral
species you sample contains (unlike you want to look for residual peaks in the difference plot to
look for some unknown phase/s, or so on… I’ll assume you know what you’ve in your sample). In
this case we have only one illite structure available, so I guess I’ll choose that one. For kaolinite I
will check a few structures instead. I’ve downloaded from the COD website (look at the beginning
of this document) all the .cif files, I want to test and I’ve put them in my experiment folder. Let’s
load kaolinites and illite (remember to get rid, using the red ‘X’ button, of the cerium oxide
structure, if you have it still in your phases box). Once loaded all your phases calculate the
spectrum. Done? Ok, it’s not a good fit indeed, but we don’t care. Now. I’ll only refine the
background and scale factor a little to have a better look…

15
After some very basic refinement:

Now we have to choose the best kaolinite among the three available here. They are very similar
but I think I’ll choose the first one. Ok, let’s eliminate the other two and run a basic refinement to
have reasonable values for background and scale factor (forget to obtain immediately a good fit
with such a strong texture!). Now with the profile you can try to look for any suspicious feature
that may have happened during the data collection, e.g. is the sample-detector distance still
correct? Or is there any unknown phase I’ve not considered before? Here we don’t see any
evident problem so we can delete our profile form the Spectra list and load again all the .esg files.

16
Once we have loaded them we have to set the tilt angles correctly for each .esg dataset (the
position, omega and chi values, in our experimental setup, is the same we used for the standard).
To do that, we select all the profiles with the same tilt angle, we click on the “modify angles”
button and we input the correct values in the boxes. In our experiment we tilted the phi axis so we
need to input that values, for each dataset we got from the single .esg. You can check if you have
input the values correctly by looking at the box below once you have selected a spectrum in the
box above.

Done? OK, save (save often: MAUD can be merciless), I’ll give the “clay1_complete.par” name,
and calculate your plot. Yes, it’s the same, but now it’s better to switch to the “Plot 2D” window.
During the refinement it is good to look at both the profile and the Plot 2D window: the profile
window is the best to look for peak shape/position errors, the 2D plot is the best to look for
texture/intensity-related errors.
Now in the 2D plot we also need to look for bad profiles: once we locate a bad profile we can
disable it from the spectra list (unchecking the “enabled” checkbox, the “View” button can be very
helpful to be sure to disable the correct spectrum…).
OK, we still have a lot of work to do…

17
First I’d start with “many” cycles of the background and scale parameters, have a look at the box
on your left: you can see the errors for each cycle, At this point you can stop before convergence
to spare some machine-time.
Now we can refine some sample-dependent features to improve the peak shape a little. MAUD
offers different models for crystallite size and microstrain. If you work with isotropic or simple, well
crystalline materials the isotropic mode will work well and quickly, for more complex crystals or
particular analyses the Popa model will provide more data. If you are in trouble you can try an
anisotropic model without rules as well. I personally like the Popa model: you can obtain
direction-dependent crystallite size and microstrain parameters. If you have a good sample
results are usually very good. We have here clays i.e. troublesome mineral species, but the
crystallinity seems acceptable and only two phases are present. So the Popa model should work
quite well.
Set these parameters and launch the refinement (Parameter List window…); time for a cup of
coffee.
Not a great improvement… but let’s be patient.

18
Now it’s time to fix the different intensities for each dataset: when you collect with different tilt
angles a flat sample you’ll have different intensities for each dataset (at high tilt angles the XR
beam passes through a larger quantity of diffracting material). Or you may have problems due to
the storage ring refill, and so on.
So, let’s open the parameter list window and press the “free all spectra monitors” button and then
the “bound datafiles by angle” and select “phi” (in our measurement this is the correct angle, the
tilt angle). Refine again. Time for a cup of tea (too much coffee is bad).
Now we can add multiple backgrounds: they will improve the background at this time. So if we go
to the “spectra list” (edit the DataFileSet_x), we select all the spectra and we press the “Add bkg
param” three times we’ll add three single profile-dependent background parameters. If you have
good backgrounds you can avoid this or add only one parameter. Refine. In the meanwhile you
can read some literature it will take time (no more drinking…).

(this is before adding the individual background params)

It’s getting better and better, slowly. To boost the improvement now we introduce the texture. We
need to edit each phase, select the “Advanced Models” tab and select a proper texture model.
For texture analysis the EWIMV model is suggested. You can set the options by clicking the
“Options” button (intuitive enough, isn’t it?). I’d set the ODF resolution equal to the angular
resolution you used for slicing the images. From this window you can also export your ODFs in
beartex format for further analysis on the Beartex software. But before we need to obtain good
ODFs for our components…
In the options panel you can also impose a symmetry: if you are sure your sample texture has
really a symmetry you can set it to spare some time and some eventual headaches… We are not
so lucky, so we work free from any symmetry and look what happens.
Ok, set the texture parameters for each mineral and run a first refinement (texture + backgrounds
+ scale factors) and monitor the analysis from the 2D plot. Now it’s time for a short walk, some
work in the lab, watching a movie, … or a faster computer!

19
After the texture refinement we get this: much better but we need to improve it again.

You can see the texture from the Plot 2D:

20
Now we have to improve again the peak shape: free microstructure. After a few cycles (and a
couple hours at least) you get a nicer result.

Now it’s time to improve the structure: touching crystal structural parameters can provide good
results or some headaches as well. To start we free the crystal cell parameters. In this
measurement it seems that crystal structures are quite good, but it seems also that illite will have
some benefits with a little fine-tuning. Be aware that MAUD considers the B-factors as basic
phase parameters. IMHO this is not a very good choice: if you free those parameters too early
you’ll probably get bad numbers (but maybe a good fit…) especially working with X-ray data. My
technique (I don’t have to do fine structure refinement/solution) is to fix the B-factors to a
reasonable number and then avoid touching them. It’s usually a very good approximation and can
save you from unwanted surprises.

21
But let’s launch this refinement. It will take quite long. If you set a large number of cycles you can
go home and see what happened the day after… Maybe it’s interesting to point out that the
texture recognition in MAUD is very fast and usually works very well, so for the texture itself you
don’t need many cycles. For other parameters it’d be better to refine as many cycles as you can,
but you can spare time stopping before the reaching the complete convergence. Or you can put
in less strict convergence criteria.
Now the fit is better again (not bad at all for a clay…):

Let’s try to improve it again. Now we can try to refine the structure, if possible. This is a delicate
process, in the sense that you can get errors not so easy to identify. But with the actual fit we can
provide the software a very good starting guess, so unwanted surprises should be kept to a
minimum (not so true with the sample presented here!). Let’s set a high number of cycles and see
what happens. For a complete structure refinement you’d need, in addition to the use the best
resolution data you can get, to set restrictions, if possible, and/or try to refine B factors. It could be
hard work…
In this sample I won’t refine anything here except for a couple of heavy atoms positions. Save
frequently: with non-ideal data it is very likely you’ll have problems (atoms bouncing all around the
cell and so on) and you want to be able to go back. Ideally the better the data, the higher the
symmetry, the lower the phase number the easier is to refine cell parameters. Set everything you
feel to refine in the refinement parameters window (ctrl+L). But be careful and check the results
for consistency.

At the end we can do a little fine-tuning to further improve (just a little: don’t expect any big
improvement here) the refinement. We need not to refine too many parameters together and pay
attention not to refine parameters that are directly in competition (e.g. two different 2theta-
dependent shift parameters). Unfortunately there is not a straightforward method that works well
for each sample. In MAUD some pre-selected refinement paths are present as wizards (menu:
Analysis -> Wizard). They can work pretty well. But sometimes they fail. I personally prefer not to
use them, but if you have a simple structure/sample you can use them without worrying too much.
With a click you can get your analysis completed, just be sure to carefully check the results.

22
OK, this is the final fit. Quite good for a clay sample. I hope you have to deal with less
troublesome mineral species . Now it’s time to have a look at the results and check if some
more or less evident errors are present.

Let’s have a look at the 2D plot (in log scale you can see better the smallest details):

Very nice indeed… luckily kaolinite here was very well ordered/crystallized.

23
But we can check the single spectra series by clicking the “Plot selected dataset” button on the
top-right of the main window. A new window will appear with all your fitted profiles: you have fitted
all of them simultaneously! You can zoom and check them in small groups: look for intensity
problems and other general bad fitting problems.
Some of my fitted profiles:

But we have refined X-ray data for one purpose: pole figures! So we need to plot them.
Open the texture plot window by clicking ctrl+T (or “apple”+T in OSX) or using the MAUD main
menu Graphics -> Texture plot.

24
As first we can check pole figure coverage (enable proper radio-buttons and checkboxes to
display what you want to).

Now you can select your phase, enable desired indices on the left panel, and plot the pole figures
(you can accept or enter custom max and min P.F. values).
Here it is basal and (200) for illite:

25
The same for kaolinite:

Once you have satisfactory pole figures, if you need to, you can export ODFs to work on them
using the Beartex software (I cannot avoid to cite my boss’s software ). In Beartex you have a
number of options: smoothing, plotting, rotating, calculating waves velocities… You can find some
info here: http://eps.berkeley.edu/~wenk/TexturePage/beartex.htm
To export ODF in Beartex format you need to edit a phase, click the “Advanced Models” tab, Click
the “Options” button, and then click the “Export ODF” button and save your .maa file.

If you need a number to measure the goodness of your fit (but for that the profiles and difference
plots give much more infos…) at the end of your refinement MAUD provides some numerical
data. I personally look at the weighted R-value, Rw(%). When a fit is good? When this value is
low. How much is low? That’s the point: it depends on your data… The best fit you can obtain is
the closest one to the LeBail fit.
To look at all your refinement parameters you can open a window (MAUD main menu: Analysis -
> Results) and you will get all your parameters values listed, with e.s.d.’s (if refined).

As a conclusion let me emphasize the importance of data collection: from bad data it is
impossible to get good results without “cheating” (that is bad). Standard collection is important.
Sample preparation is important too. Also: don’t expect miracles. If you refine a material with 20
different triclinic, disordered, phases using one CCD image and you have a lot of troubles… it’s
just normal.
After you have done your first refinement with MAUD (see part 2 of this document), play around
with menus and options: you’ll find some useful options you need.

26
Some basic Q and A…

Q) In my spectrum all peaks are shifted and the shift seems to be 2-theta dependent.
A) Typically (but not always) is a sample position problem. In the figure above the distance
sample-detector is incorrect. (Single peak profiles show a “-/+”, as here, or “+/-“ errors in the
difference plot).

Q) Peak intensities are wrong…

A) The problem could be due to preferential orientation. In the figure above a wrong texture is
imposed in a non-textured sample and peaks intensities are completely wrong (check reflection
indices too). If you have no texture probably your structure (atoms species/coordinates) is wrong.

27
Q) Peaks shapes are wrong…
A) …so the peak profile function has some incorrect coefficients. If the instrumental peak profile
parameters are properly set, the problem is in the microstructure parameters. In the figure above
the crystallite size is too small so peaks become broader (this is a typical “+/-/+” error: look at the
difference plot).

Q) Only some peaks are shifted…

A) Probably you have a wrong cell parameter (check indices). In the figure above the quartz c
axis is wrong and (hk0) peaks are still good.
Warning: in all the examples I show here I’m assuming you have no macro-stress!

28
Q) Intensities are too high/low depending on 2-theta.
A) Check your instrumental parameters setting. How’s Lorentz correction? But before have a look
at the structure thermal parameters… ;-)

For much more complete infos about Rietveld refinement you can get Young’s (editor) book “The
Rietveld Method” and/or have a look at this paper: McCusker et al. “Rietveld refinement
guidelines”, J. Appl. Cryst. (1999). 32, 36-50.

In the future I’m planning to add some other +/- simple problems and solutions… Contributes and
advice are very welcome.

nd
Now start your MAUD: in the 2 part we are working together on some data!

29
Part 2

Hands on data: let’s refine a Mg-alloy sample

Preparation
In this second part of the tutorial YOU will do the hard work. You should have read what written
above and now you can try to fit actual data to get a texture analysis. In this analysis you have
only one phase, it is very simple and the texture is very simple as well. Moreover you’ll be able to
complete this analysis very quickly. I’ll provide some files for you (you should find them in the .zip
file where you found this tutorial):
- “instrument.ins”: this file contains all the information obtained from the analysis of the standard
plus instrumental setup parameters. It will save you the time and the trouble to analyze the
standard. How to obtain this file is clearly explained in the part 1 of this document, where a CeO2
standard is refined.
- “Magnesium_0.esg”: it is the file containing the profiles of our Mg sample. It has been obtained
by integrating the MAR image in 10° sectors. In this analysis texture is very simple, quite sharp
and the structure is very simple: we can use one image, it will speed up things without
compromising the quality of the analysis. This file was obtained using Fit2D and fit2d2maud.
- “Magnesium.cif”: the file with data about Mg structure.

Step 1: Input the data

Start MAUD; then Press (menu): File -> New -> General Analysis.
Change the name of “DataFileSet_x” by double clicking it and type “Mg_0_tilt”. Now edit it (“eye”
icon). Go in the “Instrument” box and click the “Import” button, then select your “instrument.ins”
file.
Click on the “Datafiles” tab in the same window. Click the “Browse” button in the “Spectra list” box
and open your “Magnesium_0.esg” file.
Click on the “Background” tab in the same window. Add a few more parameters by clicking the
“Add parameter” button (three more is OK).
Click on the “General” tab in the same window. In the “Computation range” box and use 2.0 as
2Theta min and 6.2 as 2Theta max.
Click the “OK” button.
Now you should see your window like this:

30
If it doesn’t, restart and try again. 
Now we can save (File -> Save analysis as) the analysis and call it “magnesium.par”.

31
Have a look at the Plot 2D (tab above the profile image): there’s texture for sure!

Now click on the “Phases” tab, then menu “Edit -> Load object from CIF”. Open your
“Magnesium.cif” file. Double click on “9008505” to call this phase “Mg alloy”.

In this case we know that our sample displays axial symmetry and we need to orient it
accordingly to impose this symmetry in MAUD during the analysis i.e. we need to have the Mg
(002) pole figure maximum in the center of the pole figure. For the setup of this experiment we
need go to the “Sample” tab, edit “Sample_x” , go to the “Sample position” tab and set Chi to “90”
deg.

32
Let’s go back to the “Plot” window and calculate a spectrum (“calculator” icon):

It doesn’t look too bad. Now let’s do the analysis step by step…

33
Step 2: analysis
Let’s check calculated and measured peaks. There are no peaks missing in the calculated pattern
so our phase is OK and it’s alone. Good.
We now improve intensities and background a little. Menu: “Analysis -> Parameters list”. Click on
the “Free backgrounds” and “Free scale pars” buttons.

Then “Close” and let’s start the refinement by clicking the “hammer” button in the main window.
Let’s use 21 cycles.
At this point we see that peak positions are slightly wrong (it’s quite obvious also in the image
above). It’s OK: this is not pure Mg, it’s an alloy with some foreign atoms inside that change the
lattice parameters of the unit cell: so we’ll refine the unit cell. In this case we have a single phase,
with one atom in the asymmetric unit, we can proceed this way, even if it’s not formally correct. In
more complicated structures you’d need more info about your sample composition.
I’m assuming that we completely trust that sample-detector distance refined with CeO2 is the
same here. I didn’t make this measurement nor know the sample well, so I take no
responsibility… ;-P Let’s take the approximation here for good.

34
OK let’s edit the “Mg alloy”: “Phases” tab, select “Mg alloy” and click the “eye” button (this is the
last time I write the explicit procedure…). Right-click on both the Cell parameters boxes (a and c)
and set them as “refined”. “OK”.

Refine (button with “hammer” icon in the main window).

Now let’s start with the texture (microstructure is already acceptable, we’ll improve it later): Edit
the “Mg alloy”, go to the “Advanced models” tab and select EWIMV for the texture. Then click the
“Options button” and in the new window select “cylindrical” in the “Generate symmetry” pull-down
menu. In this window you can set the ODF resolution or export your ODF for beartex, if needed.
“Close” and “OK”.

35
Refine and then look at the 2D plot, it’s a good improvement:

36
Also the profile is good, but we can improve the profile by refining the crystallite size and
microstrain. We can do that by pressing the “Free microstructure” button in the Refinement
Parameter List window you get pressing Ctrl+L. Do you remember to save Frequently? Now the
profile is better:

(The small systematic error of the intensity with the two-theta is due to the foreign atoms in the
alloy, the correction of this error is beyond the purpose of this tutorial, I have to keep it simple)

37
Now the analysis is finished: we can have a look at our pole figures by Menu: “Graphics” ->
“Texture plot”. Let’s activate (100) and (002) pole figures by checking the proper checkboxes and
click on the “Plot” button”.

In the Texture plotting window you have three useful options (I highlighted them with the green
arrows) for plotting the pole figure coverage, the reconstructed intensities pole figures (image
above) or the inverse pole figures (image below).

38
I hope you didn’t believe me when I said the analysis was finished… 
Actually it is very difficult to say when a Rietveld analysis is finished. I think you can stop when
you think you cannot reasonably improve your fit anymore. When you get to this point you can
still improve things, but procedures that work for one dataset may be very dangerous applied to
different others. There’s no standard recipe. In this specific case we can easily improve a little our
fit by setting individual background parameters for the profiles. Then refining again for some
cycles. Then we can refine again microstructure, and so on. A good final result should be similar
to this one:

39
If you see the difference is not so evident and the pole figures are quite similar to the ones
obtained above too so I’m not showing them.

You can add more tilt images by adding a new object in the “Datafiles” window and proceed by
inputting the instrumental parameters file, 2Theta ranges, and so on (read the procedure above).
Only be sure to add the proper tilt angle by selecting all the profiles and clicking on the “Modify
angle” button. Then set the proper value for the tilt. Then you can continue with the refinement.
With this tutorial I provide only one tilt dataset to speed up things and because for this sample is
really enough. I tried with 10 different tilt datasets and the improvement is almost negligible. And
analysis is much slower… Usually you need more than one tilt, btw.

I hope you have enjoyed your Mg-alloy texture analysis. That was easy: now you can start to
have some headaches with some real samples… 

Marco

EOF

40