Systematic Errors and Sample

Preparation for X-Ray Powder

Diffraction

Jim Connolly
EPS400-001, Spring 2010
 Introduction
 Most systematic errors in diffraction experiments
are related to the characteristics, preparation
and placement of the specimen. Today we will:
– Differentiate and define those errors
– Present preparation techniques to minimize those
errors
– Discuss different types of sample mounting strategies
and trade-offs of various methods

 Always remember the distinction between

sample and specimen
 Footnote question: What is the difference between a
random and systematic error?
 Goals of Specimen Preparation
 Overall Rule: The time and effort put into
specimen preparation should not be more
than is required by the experiment objective
 Basic information in the diffraction pattern:
– The position of the diffraction peaks
– The peak intensities, and shape of peaks
– The intensity distribution as a function of
diffraction angle
 The utility of this information depends on both
the experiment parameters and the sample
preparation
 Communicate with your “client” about the
objectives of your experiment
 Design your experiment to achieve those
objectives
 Specimens and Experimental Errors
 Axial Divergence: The X-ray beam diverges out of the plane of the
focusing circle
 Flat Specimen Error: The specimen is flat, and does not follow the
curvature of the focusing circle.
 Compositional Variations between Sample and Specimen
 Specimen Displacement: Position of the sample mount causes
deviation of the focusing circle
 Specimen Transparency: Beam penetration into a “thick” specimen
changes diffraction geometry
 Specimen Thickness: Trade-offs between accuracy of peak positions
and intensities
 Particle Inhomogeneity: Can significantly alter diffraction intensities
 Preferred Orientation: Can produce large variations in intensity and
limit the peaks seen.
 Beam Path from Source to Detector
Path from X-
ray source to
detector is
shown at right
Beam path:
 From horizontal source F to
 vertical soller slits SS1 to
 Divergence slit D5
 Specimen S (A=center of diffractometer circle) to
 Receiving Scatter slit RS to
 Receiving soller slits SS2 to
 scatter slit SS to
 Monochromator and Detector (not in picture)
 Axial Divergence
 Detector sees the arc of the
Debye ring not just the
diffractions along the 2D
diffractometer circle
 Leads to a notable peak
asymmetry, particularly
pronounced at low 2θ
 Axial Divergence error for
Silver Behenate is shown at
right
 Can be minimized by
closely spaced soller slits
(at the cost of reduced
intensity)
 Flat Specimen Error
 The extreme edges of the specimen lie on
another focusing circle (rf’) which results in
the overall diffracted intensity being skewed
to a lower value of 2.
 This is related to the divergence of the
incident beam by the equation below where
 is the angular aperture of divergence slit in
degrees

 2 cot 
 2   Div Slit 2min MoK CuK CrK
343.8 0.25 4.6 8.86 19.24 28.58
0.50 9.2 4.45 5.61 14.35
The table at right shows
specimen irradiation 1.00 18.4 2.22 4.83 7.18
lengths (in mm) for a
diffractometer of a 2.00 37.2 1.11 2.42 3.59
particular radius (not ours) 4.00 78.0 0.56 1.22 1.81
Differences between Sample and Specimen
 Grinding Effects
– Problem: Excessive percussive grinding produces
extremely small particle size  peak broadening
– Remedy: Be careful (or use non-percussive
grinding techniques)
 Irradiation Effects
– Interaction with beam changes specimen
– Rare in inorganics; significant issue in organics
and phases with poorly-bound H2O
 Environmental Effects
– Strain effects in materials at elevated temperature
– Chemical reactivity of specimen
 Sensitivity to water, air or other solvents
 Usually reversible, sometimes not
 Systematically used as a tool in clay analysis
 Specimen Displacement
 Cause: Specimen is higher or
lower than it should be (i.e., not
at center of diffractometer circle
or tangent to focusing circle)
 Effect: 2θ error defined by the
following equation:
114.59 s cos 
 2  
R
(R is the radius of the diffractometer
circle; s is the deviation from the correct
position on the focusing circle measured
as the difference between r and r’)
 Can be a significant cause of errors in 2θ
 More pronounced at low θ values (cosine function)
 Can produce asymmetric peak broadening at low angles (resembling axial
divergence)
 2θ can be as much as 0.01º for each 15m of displacement at low angles
 Can be caused by poor diffractometer alignment
 Specimen Transparency
 Caused by diffraction occurring at
depth within a thick specimen
 Results in 2θ related to effective
penetration depth of the specimen
 The error is defined:

sin 2
 2 
2R   is dependent on mass and the x-ray
where  is the linear attenuation wavelength
(a.k.a. linear absorption) coefficient  For SiO2 and CuK,  = 97.6/cm, or approx.
for the x-ray wavelength, R is the 100/cm. Thus t0.5 thus is about 0.01 cm or
radius of the diffractometer circle 100 m.
and 2 is in radians  For high-density, high / materials (metals,
alloys), t0.5 will be on the order of 10 m
1  For low-density organics, t0.5 will be on the
t 0 .5  Defines the working order of 1,000 m, and a thick sample will
 depth induce very significant displacement errors
 Loose packing of powders can add reduce
density and thus increase t0.5
 Specimen Thickness
Bottom line for specimens is:
 Thin specimens
– Yield the best angular measurements (i.e. most
accurate peak positions)
– Do not yield accurate intensity measurements
(because of bad particle statistics)
– Tend to be more susceptible to preferred
orientation effects

 Thick specimens
– Can yield good intensity measurements (better
particle statistics, less susceptible to preferred
orientation)
– Susceptible to angular measurement errors
 Sample Inhomogeneity

 Multi-phase samples
may be inhomogeneous
 In example at right
chalcopyrite CuFeS2 (A)
partially oxidized to
cuprospinel CuFe2O4 (C)
 A and C have different
mass attenuation
coefficients (143.2
and116.1, respectively)

 The result will be diffraction intensities from the two phases that are
not directly proportional to the amounts of the phases present
 This effect is called the absorption effect or the particle
inhomogeneity effect.
 Preferred Orientation

 Preferred Orientation (lack of random

orientation in the powder) is usually the
dominant cause of intensity variations in
a diffraction pattern.

 Effect is most pronounced for crystals

with anisotropic shapes (or habits)

 This significantly affects the diffracted

intensities from the specimen.
 Preferred Orientation
 Occurs as a consequence
of the “spotty” nature of
diffraction in a non-random
specimen
 The effect is that the
Debye ring is uneven in
intensity
 Preferred Orientation
 How the preferred orientation is manifest in the diffraction
pattern varies with the material
– Clay minerals have a platy habit and will orient perpendicular to
(00l).
– Equant cubes (NaCl) orient parallel to their cubic crystal faces
– Bladed (most pyroxenes and amphiboles) or fibrous (most asbestos
minerals and some zeolites) materials orient parallel to their
elongation direction
– Some engineered polymers use preferred orientation for specialized
qualities
 Severe preferred orientation in a specimen will result in
“invisible” diffraction peaks
 In most specimens, all of the diffraction peaks will be seen
but their relative intensities will differ from the “ideal”
pattern.
 Careful specimen preparation can minimize the effect
 Whole-pattern refinements can use preferred orientation as
another parameter to be fit to the data
 Particle Statistics
 Quantitative (and semi-quantitative) X-ray powder diffraction
is based on the principle that quantities are proportional to
intensity.
 Accurate intensities require:
– Random orientation of crystallites in the specimen
– Sufficient number of particles for good crystallite statistics
Note: Particle size is frequently (erroneously) equated with crystallite size

 At right is a schematic pole

plot of diffractions from two
powder specimens on a
sphere
 A random pattern indicates a
random orientation
 A particle statistics exercise
 It is instructive to understand how crystallite statistics will quantitatively
affect intensity, i.e., what size particles are required to achieve a
repeatability in intensity measurements?
 Assume a powdered quartz (SiO2) specimen:
– Volume = (area of beam) x (2x half-depth of penetration)
– Assume area = 1cm x 1cm = 100mm 2
– t½ = 1/, where  = linear absorption coefficient
– SiO2 = 97.6 /cm or ~100 /cm = 10 /mm
– V = (100) (2) / 10 mm3
– Thus Volume = 20 mm3
 Estimate number of particles at different particle sizes:

Particle Diameter  40 m 10 m 1 m

V/particle 3.35 x 10-5 mm3 5.24 x 10-7 5.24 x 10-10

Particles/mm3 2.98 x 104 1.91 x 106 1.91 x 109

Particles in 20 mm3 5.97 x 105 3.82 x 107 3.82 x 1010

 Particle Statistics Exercise (cont.)

 Equal distribution on a unit sphere (area = 4 steradians) yields

a radiating sheaf of pole vectors

 Calculating angular distribution on the sphere:

Particle Diameter  40 m 10 m 1 m

Area/pole, AP = 4/ #
2.11 x 10-5 3.27 x 10-7 6.58 x 10-10
particles
Angle between poles, 
0.297 0.037 0.005
=
 Particle Statistics Exercise (cont.)
 Geometry of diffraction of a single particle:

 R is the diffractometer radius (a range is shown), F the focal length of

the anode (a characteristic of the x-ray tube), and  the angular
divergence as shown. In the above example, L (= 0.5 mm) is the length
of source visible to the target.
 NP (number of diffracting particles) = (area on unit sphere corresponding
to divergence) / (area on unit sphere per particle) = AD/AP
 Particle Statistics Exercise (cont.)
 To determine AD requires relating effective source area, FxL, to
area on a unit sphere:

= 2.5 x 10-4

 Calculating AD/AP yields the number of particles diffracting in any

given unit area for our three particle sizes:

Particle Diameter  40 m 10 m 1 m

NP 12 760 38,000

Conclusion: The standard uncertainty in Poisson statistics is proportional

to n½, where n is the number of particles. To achieve a relative error of <
1%, we need 2.3  = 2.3 n½ / n < 1%. This requires n > 52, 900 particles!
Thus not even 1 m particles will succeed in achieving 1% accuracy in
intensity.
The Bottom Line: Easily achievable particle sizes will not routinely
yield high-precision, repeatable intensity measurements.
 Enough of this Particle Statistics Stuff
 Other Factors can degrade or improve intensity accuracy:
– Concentration: mixed phase specimens reduce particles of a given
phase in a unit area, increasing error
– Reflection multiplicity: Multiplicity in higher symmetry crystal
structures give more diffraction per unit cell, improving statistics
– Specimen thickness: may improve diffraction volume, limited by
maximum penetration depth
– Peak width (crystallite size): polycrystalline particles with random
orientation can greatly improve statistics, but extremely small size will
result in peak broadening.
– Specimen rotation/rocking: helps to get more particles in the beam.
Rocking combined with rotation is best.
Ultimately, quantitative analysis based on peak intensities
cannot reliably achieve 1% accuracy even under the most
favorable specimen conditions of randomly oriented 1m
particles
 From Sample to Specimen

From rock to powder

– Bico Jaw Crusher * (Loc. 1)
– Plattner Steel Mortar & Pestle (Loc. 2)
– Spex Shatterbox * (Loc. 2)
– Mortar & Pestle (Loc. 2 & 3)
– Retch-Brinkman Grinder * (Loc. 3)
– Sieves for sizing (Loc. 2)

* Manual for use available on class web page

Equipment Locations:
1 – Northrop Hall Rm 110
2 – Geochem Lab, Northrop Hall Rm 213 (see Dr. Mehdi-Ali)
3 – XRD Lab, Northrop Hall Rm B-25
Bico “Chipmunk”
Jaw Crusher
“Marvelous”
Mehdi
Plattner Mortar and Pestle
Spex Shatterbox
 Various Mortars and Pestles

“Diamonite” synthetic
alumina

Natural Agate
 Retch
Brinkman
Grinder Pestle
Motor
Pestle Up-
Down
Pestle

Mortar

Speed Adjustments
Retch Brinkman Grinder
 To sieve or not to sieve . . .
“Mesh” size of sieve Maximum diameter of
screen particle passed
200 74 m
325 45 m
400 38 m
600 25 m
1000 10 m

 Sieves can be metal, teflon or other synthetic

 Theoretically 10 m particles may be passed
(practically, it doesn’t work)
 Because of static forces, 325 mesh is smallest for
routine use though 400 and 600 may be used with a lot
of effort
 Specimen Mounts
 How your specimen is mounted should be determined by the
requirements of your experiment – i.e., don’t do more or less than is
necessary
 Know the characteristics of your holder – always run your mount
without any specimen to know your baseline “background”
conditions
Mount Types:
 Thin Mounts (best for accurate angular measurements)
– Slurry mounts (on any flat substrate)
– Double-stick tape
– Petroleum jelly “emulsion”
 Volume or “Bulk” Mounts (best for accurate intensity measurements)
– Side-drift mounts
– Top-load mounts
– Thin-film mounts
– Back-pack mounts
 Zero-background (off-axis quartz plate) mounts
 Special techniques to reduce
preferred orientation

 Aerosol Spray Drying using Clear Acrylic

Lacquer
(procedure outlined in class notes)

 Aqueous Spray Drying in a Heated Chamber

(details at http://www.macaulay.ac.uk/spraydrykit/index.html)
 Coming Attractions:
 After Spring Break on Mar 24: In-Class Exam:
– Open-book
– Start Promptly at 3:00 PM; do not be late
– Must complete exam in 1 hour
– Written short-answer format
– All reference materials okay
– Texts will be available for reference
– Includes everything through last week (Weeks 1 thru 6)
– Emphasis on demonstrating your understanding of material

 Followed by: Tour of the XRD Lab

– Layout, location of all equipment and computers
– Introduction to MDI DataScan program and scheduling of
hands-on lab training
– Radiation safety exam should be completed before class