Laboratory instrumentation the last 10 years
Energy, Mining and Environment
Pamela Whitfield
Accuracy in Powder Diffraction IV, NIST, April 2013
A reality check?
Overview
•Where we’ve come from
•The goniometer – heart of the system
•Kinematic-type mounts
•PSDs
•Multi-layer optics
•Selected example application
Diffraction in the lab
• Lab diffractometers often in multi-user environments
• No single setup has been able to satisfy everyone • However CuKα Bragg-Brentano with graphite monochromator
came closest and was the de-facto standard for years…
• Simple
• Built like tanks
• Minimal electronics • The Scintag XDS2000 has 1 safety
circuit that can be defeated with a magnet!
The march of analogue to digital….
• Systems have become much more complex • Systems designed to be flexible and modular
– The electronics themselves sometimes less so…..
• More sophisticated microprocessor electronics and safety systems • Architectures familiar to anyone who`s looked inside a PC
Old New
The Goniometer heart of the diffractometer
• Basic concept unchanged • Axes attached to very large bearings… usually ball races
• Drive gears do eventually wear
• Goniometers still largely mechanical but… • Optical encoders now widespread to improve positioning accuracy • Goniometer arms → `optical benches` • Realignments reduced or even eliminated
Heidenhain RON905 optical encoder accuracy ±0.4”
The Goniometer • When pushing the limits of accuracy every little matters…
• Flexing of goniometer arm limits accuracy in vertical setup • → detector/source weight & counterbalancing become issues
• θ−2θ often preferable in terms of pure positional accuracy
Jim Cline being fussy! Stress analysis on Jim’s custom D5000 detector arm
System alignment • Manual system alignment can be…
• boring • frustrating • source of much rude language! • impediment to switching configurations
– increases my ‘activation energy’
• Don’t have someone with the ‘Midas touch’?
• 2 main alternative approaches • Fully motorized and automated alignment • Kinematic-type mounts for reproducibility
Automated alignment
• Alignment automated at the touch of a button • Everything motorized and software controlled • Easy for the user
• However, system never exactly the same twice after realigning • No different from manual realignment in this regard
• System reproducibility shouldn’t change over time • Mechanical wear excepted
Kinematic mounts • Kinetic mounts used routinely for mounting optical mirrors
• 3-leg stool (tripod) concept
• Dirty environments? • Forces should force dirt out from between surfaces – no effect • However, turbidity from environment/surfaces can prevent
equilibrium being reached – Chocking effect (www.precisionball.com/kinematic_encyclopedia.php)
3-point kinematic mirror mount
• Precision engineering – expensive!
• Hardened surfaces desirable (e.g. WC) • Reduce contact distortion + low friction • High shear strength to resist lateral forces (chocking)
• Mounts need to be robust • Delicate stuff tends to not fare well in a university environment!
• Factory-aligned • Portability between multiple systems?
From the old to the new….. • Bragg-Brentano + graphite mono/scintillation detector
• gives excellent data but low counts
• Fear not! • Many new toys to spend your money on!
• Seems every component has a possible upgrade • X-ray source
– Microfocus, rotating anode, ‘liquid metal jet’
• Optics – Mirrors, monocaps, polycaps, monochromators (primary and secondary)
• Detectors – PSDs, 2D detectors, energy-dispersive
X-ray sources
• Standard X-ray tube still the most common source • Ceramic insulation used increasingly • Ceramic is a better heat-conductor • Better tube lifetime claimed
• Rotating anode • Still quite a niche product for powder diffraction • Large increase in tube power (↑ 18kW) • Generally very heavy and bulky
– Not normally used for θ−θ systems due to issues mentioned previously but….
Rigaku TTRAX
GE X-ray tubes
Spot-focus X-ray sources • Microfocus
• Widely available • Standard anode materials • High brilliance
– Comparable to older synchrotron beamlines
• Liquid-metal jet • Don`t have to worry about melting the
anode as it`s already liquid • Wavelength GaKα (other metals as
dopants in liquid) – GaKα similar to CuKα but users often
uncomfortable changing wavelengths..
Position Sensitive Detectors rule? • Historically proportional detectors using a counting gas
e.g. mBraun, Inel
• Calibration required to relate ‘channel number’ to angle
• Defocussing in linear PSDs
• Inel known for their curved PSDs • No defocussing but calibration functions more complex
mBraun PSD
Inel 120° PSD
PSDs rule! • PSDs largely moved away from gas-based technologies
• PSDs now a common sight in XRD labs • What effect has this had on aspects of accuracy?
•GOOD • Count rates… • Excellent signal-to-noise
•BAD • Count rates! • Artefacts in data now visible
Newer PSD technologies
2 (degrees CoK)5 10 15 20 25 30 35
Inte
nsity
(cou
nts)
0
1000
2000
3000
4000
35kV, 40mA30kV, 45mA
Quartz stoneCo tube, Vantec, Fe filter
Xe escape peaks
K
K
• More advanced gas-based counting (e.g. Vantec-1) • 33keV Xe escape peaks can be problematic
– not all detectors the same…. one of ours less sensitive then the other • Works up to MoKα (Vantec AgKα )
Effect of generator voltage on appearance of 33keV Xe escape peaks without primary optic
Solid state strip detectors • Current mainstream PSD technology
• Cost effective
• PSD energy discrimination improving
• Higher X-ray energies require thicker Si films to maintain efficiency
• Image-plate based curved PSD - Stoe
2 (degrees - CuK)20 30 40 50 60 70 80 90
Arbi
trary
inte
nsity
20 30 40 50
Sqrt(
inte
nsity
)
* * * *
PSDs rule! – not so good? • More stuff you`d rather ignore
• Absorption-edges of Kβ-filters – Background fitting errors → inaccurate peak intensities
• Tungsten Lα lines & residual Kβ intensity becomes very obvious
LaB6 with CuKα and a nickel Kβ filter
Lin
(Cps
)
0
10
20
30
40
50
2-Theta - Scale20 30 40 50 6
Which dataset do you prefer?
PSDs • Normally poor energy discrimination
• In absence of monochromator fluorescence a major problem • Poor peak-to-background impacts accuracy (CPD Rietveld RR) • Tweaking detector electronic discriminators can help…
LiMn1.5Ni0.5O4 CoKα with varying detector discriminator settings
Discriminators…
Lin
(Cps
)
0
1
2
3
4
5
6
7
8
9
10
2-Theta - Scale21.2 22 23 24
PHA0.1 0.2 0.3 0.4 0.5 0.6
LL = 0.36WW = 0.06 P/B = 13.4
PHA0.1 0.2 0.3 0.4 0.5 0.6
LL = 0.28WW = 0.34
P/B = 4.5
PHA0.1 0.2 0.3 0.4 0.5 0.6
LL = 0.1WW = 0.5
P/B = 4.2
Rescaled to normalize background
Sacrifice intensity to improve P/B ratio
P/B still way off 50
• Recent developments in electronics have greatly improved possible energy discrimination of PSDs
Alternative low tech approach… • High-purity Al foil filters can reduce fluorescent background
• costs intensity • but once again less can be more!
Mn3O4 CuKα primary mirror mBraun PSD Al foil mounted on Ni filter
2 (degrees CuK)20 30 40 50 60
Inte
nsity
(cou
nts)
0
1000
2000
3000
4000
5000no Al5 Al layers/Ni10 Al layers/Ni
2 (degrees CuK)30 32 34 36 38 40
Inte
nsity
(cou
nts)
0
50
100
150
200
250no Al 15 Al layers 10 Al layers
Normalized to background
P/B = 20
2 (degrees CoK)0 20 40 60 80 100 120 140
Sqrt(
Inte
nsity
)
0
20
40
60
80
100
120
140
160 Rwp = 10.2%
15 20 25 30 35 40
0
10
20
30
40
50
60
Residual K
PSDs and monochromators… Data from Ian Madsen
Y2O3 data from an APD system Bragg-Brentano proportional 0D detector graphite monochromator
2 (degrees CoK)0 20 40 60 80 100 120 140
Sqrt(
Inte
nsity
)
0
50
100
150
200
250Rwp = 9.2%
15 20 25 30 35 40
0
10
20
30
40
50
60
Residual K
• Out-of-plane monochromators fitted to some PSDs
• Gets rid of fluorescence but can cause other issues
Y2O3 data from an MPD system Bragg-Brentano PSD strip detector with monochromator
2 (degrees AgK)10 15 20 25
Inte
nsity
(cou
nts)
-2000
0
2000
4000
6000
8000
10000
AgKα Pd filter
PSDs – flexibility? • Most lab-based PSDs optimized for CuKα
• MoKα – reduced efficiency even if it works…. • AgKα – good luck!
• Detector electronics may need better shielding
• Absorption edge from β filter difficult to miss with MoKα & AgKα!
2 (degrees AgK)10 15 20 25 30 35 40 45
Inte
nsity
(cou
nts)
0
5000
10000
15000
20000
SRM1976 500µm Lynxeye, Pd β filter
2 (degrees AgK)10 15 20 25 30 35 40 45
Inte
nsity
(cou
nts)
0
5000
10000
15000
20000Fit when K-edge effects modelled
However - all is not lost….. • E.g. PSD not
optimised for MoKα …
• Calcium silicate hydrate (CSH) carbonation
• Fixed snaphots - 8°
• Damp sample vs dry
• Damp sample 3s time resolution
6.3 MPa (915psi) CO2 dry sample 24 hours 30s snapshots
2θ (MoKα) 10 12 14 16 18
Flowing 1atm CO2 damp sample 20 mins 3s snapshots
10 12 14 16 18
2D detectors • Common at synchrotrons
• Increasingly in lab as well…
• Positional accuracy more problematic as more sources of error in calibration • Complex aberration functions • With lab wavelengths integrated data from multiple frames may
have to be merged…
• We never used ours much – too many Mn-rich samples • Fluorescence a real problem with either CuKα or CoKα • Al foil filter also works with 2D detectors but not ideal… • Currently sat in a box!
Multi-layer Mirrors…. • Technologies for multilayers developed over last 10 years
• The layer materials in use have evolved • Mirror corrosion issues solved
• Originally only CuKα widely available
• Now available in all wavelengths in divergent/parallel/ convergent
• The right mirror for the right application can be great
Fraunhofer 40mm CuKα W-based mirror (circa 2002)
Incoatec sealed 60mm CuKα Ni-based mirror (circa 2006)
Multi-layer mirror optics mirrors versus monochromators – pros & cons • Pros
• Choose from convergent, parallel or divergent • Very high intensities possible • Excellent Kβ suppression • Good resolution can be achieved in focussing geometry
• Cons • Wavelength specific • Kα1/Kα2 doublet • Past history of surface corrosion • Temperature sensitive
• Despite cons I have an unsealed CuKα focussing mirror
Not for those of a weak disposition…! Horror pic from 2005
• Eventually explained by ozone attack
• Sealed units or inert flowing gas solved problem… • or reduce Bremmstralung energy by reducing voltage
A nice new ‘mirror-like’ mirror After intensities started to drop…
2 (degrees CuK)10 20 30 40 50 60
Inte
nsity
(cou
nts)
0
10000
20000
30000
CuKα focussing mirror – Pro (intensity) • Fluoroethylene carbonate (focussing mirror + 10° PSD)
• Structure indexed using a 4 minute dataset • Solved & refined with 3.5hr pseudo-VCT dataset → 140° 2θ
CuKα focussing mirror – Con (odd profile) • Can be tricky to get Kα2 at 50% during manual alignment
• The change easily modelled but not ideal
• Alignment for max intensity often doesn’t give best result….
2Th Degrees9796959493929190898887
Cou
nts
4,500
4,000
3,500
3,000
2,500
2,000
1,500
1,000
500
0
-500
Si 0.00 %
SRM640c Kα2 = 63% (simple 2 line emission profile)
Multi-layer mirrors • Make sure you have the right one for the job…
• Parallel-beam mirror should have been fine for capillary transmission work
Two theta (degrees)26 28 30 32 34
Arbi
trary
inte
nsity
focussingpowderpeak positions
• Divergent ‘powder mirrors’ can be problematic • strange peak shapes,
poor resolution
Comparison of capillary data from jadarite with a ‘powder mirror’ and a focussing mirror
Focussing for transmission
Twin mirrors • Second mirror focuses onto scintillation detector
• decent peak resolution • excellent backgrounds • the data is still student-proof!
• Quite rare….and not cheap….
• Particle statistics aren’t great
2 Theta (degrees)34.0 34.5 35.0 35.5 36.0
Inte
nsity
(cou
nts)
0
200
400
600
800
1000
1200
1400 -0.6 mm -0.5 mm -0.4 mm -0.3 mm -0.2 mm-0.1 mm0.0 mm0.1 mm 0.2 mm0.3 mm0.4 mm0.5 mm0.6 mm
Al2O3 104 reflection with displacement for contoured sample
Oddity? 60mm Ni-based sealed primary with 40mm W-based unsealed secondary….
2 (degrees CuK)80 85 90 95 100 105 110
Inte
nsity
(cou
nts)
-20
0
20
40
60
80
100
120
140 5º grazing incidence (0.2+1+0.2mm slits)symmetrical
micronized yttria
Grazing incidence with twin mirrors… • Secondary mirror → no loss of peak resolution
• Complex intensity/angle relationship but can be modelled • Dependence gets quite ‘interesting’ with sample displacement…
2 (degrees CuK)0 20 40 60 80 100
I asym
m/I sy
mm
0.0
0.5
1.0
1.5
2.0
Toraya equation 1 for 5º incident angleToraya equation 3 for 0.2mm beam height and 1mm detector slit
Calculated intensity/angle curve for the 5º grazing setup
2 (degrees CuK)20 40 60 80 100
Inte
nsity
(cou
nts)
0
500
1000
1500
2000
2500
3000
Toraya corrections (no displacement)
Incident angle refined to 5.5ºMirror recieving slit refined to 1.3mm
Rwp = 19.8%, GOF = 1.1
Increasingly Popular Application Areas • Combinatorial screening
• 2D detectors with well plates
• Pharmaceuticals • Polymorphs and structure solution
• Tomography
• Residual stress • 2D detectors
• Pair Distribution Function (PDF) • High energy
Pair Distribution Function
• Synchrotron beam-time a finite resource • Lab studies viable for some systems where resolution not as vital
• Was always possible but painfully slow with lab instrumentation
• Need very low noise to high scattering angles • PSDs desirable but efficiency ↓ as energy ↑
Apologies for the pun!
PDF – instrument considerations • Want highest-energy to maximize Q-range, but…..
• MoKα – Qmax ~18 Å-1, 3kW LFF tubes • AgKα – Qmax ~22 Å-1, 1.5kW LFF tubes (50kV, 28mA)
• Capillary geometry desirable • Monochromator versus mirror?
• Trade-off between • Higher Q Ag Mo • Tube power Ag Mo • PSD efficiency Ag Mo
2 (degrees - MoK)20 40 60 80 100 120 140
Inte
nsity
(cou
nts)
0
10000
20000
30000
40000
50000
60000
2 (degrees - MoK)
100 110 120 130 140 150
Inte
nsity
(cou
nts)
0
200
400
600
800
1000
Example - LiMn1/3Co1/3Ni1/3O2 • MoKα with Vantec PSD (10° window)
• Efficiency ~50% in use versus CuKα (N.B. higher tube power)
3 hour scan time (reflection) Radial Soller slits to reduce background 40kV 55mA to minimize Xe escape peaks without primary optic (the ‘good’ Vantec)
LiMn1/3Co1/3Ni1/3O2
• PDFFIT just as happy with lab data
• Despite low resolution distortion away from average structure (red line) visible
Mn4+-O (1.91Å)
Ni2+-O (2.06Å)
Fit using PDFFIT of the average R-3m structure versus G(r )
Blowup shows the misfit in the TM-O environment
Conclusion • Faster is not always better…..
• Counts aren’t everything • less really can be more!
• Use best setup for the experiment, not the most expensive!
• In multi-user environments the single perfect instrument still doesn’t exist
• Competing demands of speed, flexibility and accuracy lead to trade-offs
Gazing into the future? • Better energy discrimination?
• Hopefully…
• Better angular resolution? • Probably not….
• Is the future in the lab 2D? • For some but not all
• High energy • As above….
• Do the instrument vendors have more rabbits to pull from the hat?
• Most certainly…
Acknowledgements…
• Ian Madsen
• Jim Cline
Questions…..?
Top Related