[CCP14 Home: (Frames | No Frames)]
CCP14 Mirrors: [UK] | [CA] | [US] | [AU]

(This Webpage Page in No Frames Mode)

CCP14

Methods, Problems and Solutions

Mounting Samples in Powder X-ray Diffraction

The CCP14 Homepage is at http://www.ccp14.ac.uk

[Back to Problems and Solutions]

Links


Via sci.techniques.xtallography newsgroup

From: david.mccready@pnl.gov (Dave McCready)
Newsgroups: sci.techniques.xtallography
Subject: Specimen Mounts for XRPD
Date: 9 Sep 1999 21:41:19 GMT
Organization: Battelle PNNL


As evidenced by recent publications, there appears to be considerable
interest in techniques for the preparation of specimens for X-ray powder
diffraction analysis.  As such, I have decided to throw my 2 cents in here.

First, I will mention my preference in specimen holders.  I almost exclusively
use the off-axis quartz plates produced by the Gem Dugout in State College, PA
(Phone/FAX 814.238.4069).  These are high-quality, durable specimen holders,
which yield low background.  Rather expensive perhaps, but well worth the cost.
For most applications, the standard, front-loading shallow cavity versions of
these plates are ideal.

Proper X-ray powder diffraction specimens are, of course, flat, densely packed,
very fine-grained powders.  This is a fairly short order, but I have noticed
many people have difficulty accomplishing the task.  In particular, achieving 
high packing density can be a challenge.

The usual approach is to press powder into a cavity using an ordinary microscope
slide.  However, many materials will agglomerate and pile up upon compression.
This tendency interferes with one's ability to also make the specimen mount 
flat.

My solution has been to use _frosted_ microscope slides.

By frosted, I mean slides that are sandblasted to provide a surface for marking 
the slide with a pen (not slides that are _painted_ for the same purpose).

I favor slides that are frosted on one end and only on one side.
Full-frosted cytology slides might also work, but I have not yet tried them.

Anyway, these partially frosted slides work (for me, anyway) as follows:

------
STEP 1
------
A small amount of powder approximating (or, ideally, slightly exceeding) the 
volume of the cavity is placed in the holder.

The _smooth_ portion of the microscope slide is first used to lightly compress
the powder into the cavity.

Excess sample material is then scraped away from just the edges of the cavity
using a razor blade.

This leaves a slightly elevated specimen surface.

------
STEP 2
------
Now, the frosted end of the slide is brought into use.

Placing the frosted end of the slide atop the elevated specimen surface,
one can drive the remaining excess sample material into the cavity using
moderate-to-considerable downward force combined with twisting motion.

Whatever additional excess sample material might then be present appears
as a kind of "halo," which will be visible through the slide around the
specimen cavity.

If no halo appears, the cavity is not yet "full" (densely packed).

------
STEP 3
------
A small amount of sample material can be picked up on the corner of a razor
blade and deposited on or near the center of the cavity.

Then, the frosted portion of the slide is again used to work this additional
sample material into the cavity.

If no halo appears around the cavity, this process is repeated.

Once the halo appears, the cavity is full.

Any remaining excess sample material on the surface of the holder around the
cavity can then be (carefully) scraped away with the razor.

Now one has a densely packed, flat specimen mount.

However, its surface is very rough from contact with the frosted microscope 
slide.

------
STEP 4
------
As a final step, the surface of the specimen mount can be smoothed using the 
plain, smooth surface of the microscope slide.

Inspection of the specimen should then reveal a very smooth study surface.

Pits in the specimen surface indicate that the cavity can accept additional
sample material.  (In that case, return to Step 3.)

--------------
Additional Tip
--------------
Certain microscope slides will have frosted surfaces that are a bit too rough
for this application.  Small particles of glass will be peeled off the slide
and mixed into the specimen if very much force is applied in Step 3.

The frosting on such slides can be toned down a bit (smoothed) using a
Scotchbrite pad and a little water.

This problem will also occur if the sample material is both coarse 
(large-grained) and harder than glass.  In that case, grind the sample 
to a finer consistency.

-------
Summary
-------
Specimens prepared for X-ray powder diffraction as outlined above generally
yield very good, reproducible patterns.  With practice, such specimens can be
prepared in a very short time.  Only common and inexpensive tools are required.

The limitations of front-loading specimen mounts are part of the package,
but the supposed negative effects of such specimen mounts are, in my opinion,
overrated.  Specimens prepared by this method will certainly suffice for the
vast majority of X-ray powder diffraction analyses.

I hope this information is received in the spirit it is provided, which is
that it be of some general interest and utility.

Regards,
David E. McCready
Battelle EMSL
Richland, WA
http://www.emsl.pnl.gov/

Via sci.techniques.xtallography newsgroup

From: Armel Le Bail [lebail@aviion.univ-lemans.fr]
Newsgroups: sci.techniques.xtallography
Subject: Re: Specimen Mounts for XRPD
Date: Wed, 15 Sep 1999 09:42:42 +0200
Organization: Universite du Maine - France
Xref: daresbury sci.techniques.xtallography:5655


Dave McCready wrote:

> As evidenced by recent publications, there appears to be considerable
> interest in techniques for the preparation of specimens for X-ray powder
> diffraction analysis.  As such, I have decided to throw my 2 cents in here.

Well, and what about preferred orientation, the main problem
in XRPD for those refining structures by the Rietveld method ?
Your process will certainly produce a lot of preferred orientation,
if the sample is prone to it.

--
Armel Le Bail - Universite du Maine, Laboratoire des Fluorures,
CNRS ESA 6010, Av. O. Messiaen, 72085 Le Mans Cedex 9, France
http://www.cristal.org/

Via sci.techniques.xtallography newsgroup

From: Jason G Wolf [jwolf+@andrew.cmu.edu]
Newsgroups: sci.techniques.xtallography
Subject: Re: Specimen Mounts for XRPD
Date: Wed, 15 Sep 1999 14:35:22 -0400
Organization: Materials Science and Engineering, Carnegie Mellon, Pittsburgh, PA

Excerpts from netnews.sci.techniques.xtallography: 9-Sep-99 Specimen
Mounts for XRPD 

> ...one can drive the remaining excess sample material into the cavity using
> moderate-to-considerable downward force combined with twisting motion.

This is exactly what you should NOT do!

I use a similar technique when top-loading, but I press lightly to avoid
packing.  Any sort of top-loading is a short cut. It should only be used
when you already know (or have no concern over) the sample composition. 
Top-loading has no effect on peak position which is often the only
matter of interest.

Back-loading is a must for quantitative composition analysis.  I was
recently trying out the I/Ic technique for an Al2O3-NaCl mixture and
top-loading was a disaster.  For some projects, even back-loading is not
good enough.

Le Bail already mentioned the problem top-loading can create for
Rietveld analysis.  The method will also hinder phase identification by
screwing up relative peak heights.

That being said, I am interested to see if the frosted glass will reduce
the static electricity I occasionally encounter while mounting very
light specimens.

Via sci.techniques.xtallography newsgroup

From: Mike Meier [mlmeier@ucdavis.edu]
Newsgroups: sci.techniques.xtallography
Subject: Re: Specimen Mounts for XRPD
Date: Wed, 15 Sep 1999 18:04:16 -0700
Organization: U.C. Davis


Dave McCready wrote:

> >Well, and what about preferred orientation, the main problem
> >in XRPD for those refining structures by the Rietveld method ?
> >Your process will certainly produce a lot of preferred orientation,
> >if the sample is prone to it.
>
> Well, besides the fact that I anticipated exactly this kind of
> retort to my post, I will say that I strongly feel the preferred
> orientation problem is _highly_ overrated.

I may try your method, but preferred orientation may not be so overrated.
Take for example two recent experiences in our lab.  In one, a student
tested three methods of sample preration for his material (pure white,
collected from the shore of a lake, don't know what it was).  Each method
gave different relative intensities.  One method produced only one
diffraction peak.  Clearly a very oriented material.  In the other case a
student sprinkled a littel boron nitride on a quartz plate and obtained only
one diffraction peak.  Turns out that his material consited of
single-crystal particles having only {111} facets.

My experience is not as extensive as yours, but the experiences I have had
have taught me to watch out for this problem.

Do you anticipate problems with uniform mixing?  I once had a very
interesting time trying to prepare a 50/50 copper/corundum sample for a
quantitative analysis.

Mike Meier

Via sci.techniques.xtallography newsgroup

Newsgroups: sci.techniques.xtallography
Date: Wed, 15 Sep 1999 11:10:41 -0700
Subject: Re: Specimen Mounts for XRPD
From: david.mccready@pnl.gov (Dave McCready)
Organization: Battelle PNNL


>Well, and what about preferred orientation, the main problem
>in XRPD for those refining structures by the Rietveld method ?
>Your process will certainly produce a lot of preferred orientation,
>if the sample is prone to it.

Well, besides the fact that I anticipated exactly this kind of
retort to my post, I will say that I strongly feel the preferred
orientation problem is _highly_ overrated.

If we believe what we read, one would probably be led to think
that preferred orientation infected every single XRPD analysis
ever attempted.  Having done in excess of 10,000 XRPD analyses
myself, this has not been my experience.  Relatively few materials
are "prone" to preferred orientation so extreme that it really
matters.

As for the Rietveld method, it too is highly overrated.  One does
not get angle information from XRPD data (i.e., one does not measure
it).  No amount of corrections, modeling, etc. can make up for data
that simply isn't there in the first place.

Standardless Rietveld refinement for structure analysis (which appears
to be the "standard" approach) is even more laughable.  If one cannot
accurately measure the lattice parameters (which one _cannot_ without
internal standardization), what exactly is the point of refining atom
positions?

In addition, if the XRPD pattern of a specimen clearly indicates
the material is an isomorph of some other compound (a prerequisite
for doing a Rietveld refinement), who really cares what the atom
positions of the study material are?  If you can do the refinement
at all, the positions are, by definition, very close to those of
the isomorph.  But then, I suppose these "refinements" _do_ make
for nice graphs, tables, etc. for "publication."  (Wow!)

Now, if you want to spend _hours_ preparing a specimen for the purpose
of doing a Rietveld refinement, that's just fine.  On the other hand,
if you have some real work to do and limited time to do it, I think
you will find my specimen preparation procedure will "suffice for the
vast majority of X-ray powder diffraction analyses."  At the very least,
why don't you try it before you knock it?

Regards,
Dave McCready
Battelle EMSL
Richland, WA
http://www.emsl.pnl.gov/

Via sci.techniques.xtallography newsgroup

From: Armel Le Bail [lebail@aviion.univ-lemans.fr]
Newsgroups: sci.techniques.xtallography
Subject: Re: Specimen Mounts for XRPD
Date: Tue, 21 Sep 1999 15:38:49 +0200
Organization: Universite du Maine - France
Xref: daresbury sci.techniques.xtallography:5680


Frank May wrote:

> I wholeheartedly agree with David regarding specimen preps, and I've been doing
> it pretty much full time since 1972 - and if one calculates running 500-1,0000
> patterns per year, that's a bunch and gives a data base on which to make some
> conclusion.

OK, I made only 30-100 long-time exposure, high resolution
patterns per year since 1974.

> I find "preferred orientations effects" are the exception rather than the rule.

When the sample is packed, I find more than 75% affected ; 25% quite
strongly  affected. In all cases for which I perform Rietveld refinements,
I record at least 2 patterns because it is essential to be sure that
no preferred orientation occurred. Here are my 2 cents for performing
a pattern almost free of preferred orientation : deposit the sample
on the holder through a 63 microns sieve (with grain size <10 microns).
Improve the surface by gently applying a sheet of paper on it, only once
(better ways are known, but more expensive). Never never never pack.


> The absolute test for preferred orientation effects in an xrd pattern is to run
> it multiple times - dumping out the sample between runs and repacking.  If it
> reproduces, then there is no problem with p.o.

Repacking, I must disagree ! The NIST SRM 1970 is Al2O3 highly packed.
Intensities are well reproducible, sure. But the preferred orientation
is enormous, and also well reproducible. You should run the pattern
at least two times, that is OK : but one pressed, and the other not.

> 2) is the pattern reproducible?  A reproducible pattern is one where the peaks
> absolutely overlay on an on-screen display of the patterns.  The degree of peak
> intensity differences is an indication of counting statistics and "texture
> effects" (which includes preferred orientation effects).

Degree of peak intensity difference as a function of packing, yes.
Having the peak intensities for packing = 0 is essential.

Check all your 500-1,0000 patterns by the Rietveld method,
and you will have an idea of the degree of preferred orientation
that you introduced by packing. By the way, are your samples
always different or almost always the same ?

About the Rietveld method : this is science. And science
evolves when the demonstration is made that something
was wrong. Introduction of size and microstrain or faulting
effects in the Rietveld method is also science, which
can be improved. It gives estimations of what happens
in your sample. Estimations can be more or less wrong,
depending of the state of the science development
toward the truth. Current approachs are phenomenological
ones, at least allowing to improve fits and probably
structure accuracy. Fortunately, some guys try
to improve the current state of knowledge.

Anyway, if you do not want to determine, or more simply
to refine, the structure from powder data, then pack your sample,
that is your problem. You will probably obtain a correct
identification from the JCPDS ICDD PDF-2 database (many
patterns in this database are highly affected by preferred
orientation, so your method is just fine for that ;-).

--
Armel Le Bail - Universite du Maine, Laboratoire des Fluorures,
CNRS ESA 6010, Av. O. Messiaen, 72085 Le Mans Cedex 9, France
http://www.cristal.org/

Via sci.techniques.xtallography newsgroup

From: Henry Barwood [hbarwood@indiana.edu]
Newsgroups: sci.techniques.xtallography
Subject: Re: Specimen Mounts for XRPD
Date: Tue, 21 Sep 1999 09:38:20 -0500
Organization: Indiana Geological Survey
Xref: daresbury sci.techniques.xtallography:5681


Armel Le Bail wrote:

> When the sample is packed, I find more than 75% affected ; 25% quite
> strongly  affected. In all cases for which I perform Rietveld refinements,
> I record at least 2 patterns because it is essential to be sure that
> no preferred orientation occurred. Here are my 2 cents for performing
> a pattern almost free of preferred orientation : deposit the sample
> on the holder through a 63 microns sieve (with grain size <10 microns).
> Improve the surface by gently applying a sheet of paper on it, only once
> (better ways are known, but more expensive). Never never never pack.

I would agree with Armel on the packing and add a technique I've found to be
surprisingly effective.

First I begin with a "zero background" plate (The quartz plates have been best, in my
experience). I coat this with several layers of a very thin solution of gelatin (2%)
and let it almost dry.

Second, when the gelatin is firm, but tacky, I sieve a very finely ground <10 micron
sample onto the plate. Let the sample sit for long enough for the gelatin to dry.

Third, invert the plate and gently tap it to dislodge the excess sample. If done
properly, the remaining sample adhering to the gelatin will be about 90-95% random
orientation, adequate for most diffraction work.

I originally developed this technique to mount <1 milligram samples for powder
diffraction analysis, but have found that it gives excellent results with the various
programs (Xfit, Crys, EXPO, etc.). It is one of those techniques that is also an art!
You have to judge the gelatin surface properly. Too wet and the sample orients, too
dry and not enough sample adheres.

Backgrounds are below 50 cps on my diffractometer, over the range 4-70+ degrees two
theta, for the combination of quartz plate and gelatin and I have managed to identify
unknown samples of around 10 micrograms with a diffractometer.

Henry Barwood

Via sci.techniques.xtallography newsgroup

From: Frank May [Frank.L.May@umsl.edu]
Newsgroups: sci.techniques.xtallography
Subject: Re: Specimen Mounts for XRPD
Date: 21 Sep 1999 11:06:02 -0700
Organization: University of MO-St. Louis
Xref: daresbury sci.techniques.xtallography:5683


        In response to Armel's response to my response to the subject.....

1.  The degree of "preferred orientation" one tolerates depends on the nature of
the work.  I do mostly qualitative analysis, and I depend on the presence and
location of peaks more than I do on their intensities.  Besides, when I compare
my patterns with ICDD PDF files, I know the "preferred orientation" in the
reference patterns is probably worse than mine (okay, there are exceptions).  If
data in the reference pattern is older than 10 years, I **expect** "intensity
anomalies".
2.  Never say "never never never...pack."  When determining unit cell
dimensions, preferred orientation may be desireable to improve the intensities
of critical reflections.  Subsequently, a "correct" pattern may be obtained
using the lattice paramenters previously obtained.  (I got this one at a DXC
workshop.)
3.  An "artistic" approach for specimen preparation is nice, but that means you
will obtain a pattern which is different from mine and from someone else's
pattern.  (Armel's is not the only posting to this effect.)  There has to be a
"correct" way to prepare specimens which not dependent on "art".
4.  SRM 1970 is a pressed specimen, but would you really consider it a powder? 
Do you use it as an intensity reference or as a peak position reference?  If
used as a peak position reference, then "preferred orientation" is not critical
(see #2 above).
5.  Rietveld analysis was originally developed for inherently low resolution
neutron diffraction data.  I posted previously my concern for how **relative**
peak breadths are treated for modern, high resolution x-ray data.  It is my
contention that the breadth of every peak in an XRD pattern is related to the
breadth of every peak by reason of the SHAPE OF THE CRYSTAL.  It matters not
that good instrument functions, profile functions, etc. are well defined if the
data are treated by a rule which is not physically meaningful.  Specimens which
exhibit peak broadening due to their crystallite size give data which cannot be
treated properly with the Cagliotti function.  What is really required is a
"crystal shape function".  Until such a function is employed, it is immaterial
if one has "preferred orientation" effects in the data.  By the way, using high
resolution data, one should easily discern that peaks are not "isometric" (or
whatever one chooses to call the effect).

REFERENCE:  Klug & Alexander 2nd Edition, Table 9-2, page 659.  Look it up!

        Keep the discussion going.
-----------------------------

Frank May
Department of Chemistry
University of Missouri-St. Louis
St. Louis, MO  63121
frank.l.may@umsl.edu

Via sci.techniques.xtallography newsgroup

From: Cristal 
Newsgroups: sci.techniques.xtallography
Subject: Re: Specimen Mounts for XRPD
Date: Wed, 22 Sep 1999 01:10:01 +0200
Xref: daresbury sci.techniques.xtallography:5685


Frank May wrote :

>   REFERENCE:  Klug & Alexander 2nd Edition, Table 9-2, page 659.  Look it up!

A not obsolete Bible, but there are New Wills, Science has evolved since those
heroic times without computers.

I have many more recent references in which the size/microstrain/
faulting/and-so-on are approached in the Rietveld method by a hkl dependency
(not the Caglioti law, of course). The story started in 1985 by the crude use
of ellipsoids hkl dependency. The current state of the art makes use of spherical
harmonics. So that the improvement that you expect is already there - see
GSAS, FULLPROF, WinMprof, etc.

But remember, if you intend to use the Rietveld method, never, never, never
pack the  sample (this is the only case where the rule "never say never" does not
apply ;-). I even know cases of preferred orientation occurring with
neutron data in spite of the cylinder shape of the sample container, due to
packing.

Good news, if you prefer to use the Pawley or Le Bail methods,
you can pack the sample if only the profile shapes and widths interest you.
Without the structure constraint, the fit will surely be even better.

--
Armel Le Bail - Université du Maine, Laboratoire des Fluorures,
CNRS ESA 6010, Av. O. Messiaen, 72085 Le Mans Cedex 9, France
http://www.cristal.org/


[Back to Problems and Solutions]

[CCP14 Home: (Frames | No Frames)]
CCP14 Mirrors: [UK] | [CA] | [US] | [AU]

(This Webpage Page in No Frames Mode)

If you have any queries or comments, please feel free to contact the CCP14