1. Atomic structure at the nanoscale: a 21st century materials challenge
Pb (s.g. Fm-3m)
3.479
4.920
6.026
6.958
Emil S. Bozin
Applied Physics and Applied Mathematics Department, Columbia University
Condensed Matter Physics and Material Science Department, BNL
Workshop on Characterization of Advanced Materials under Extreme Environments for Next Generation Energy Systems
September 25-26, 2009
25 September 2009

2. Complex nanostructured materials
INTRO
Nanostructured bulk crystals
Nanoporous (mesoporous) materials
Nanoparticles
Images: Igor Levin/Tom Pinnavaia/Sandra Rosenthal
Physical properties often critically depend on the nano-scale structure, rather than the long-range structure!
- Complex bulk systems with interesting physical properties are often inhomogeneous on a nanometer lengthscale
high-temperature superconductors, colossal magnetoresistive materials,
high performance thermoelectric materials …
- Nano-particles, nano-tubes, nano-wires etc. important for applications optoelectronics, nanosensors, programmed release drug delivery systems…

3. Crystallography gives average structure
ROUTINE
Rietveld method
in powders
Figures: J.S.O. Evans et al.
and M. Tucker et al.
Bragg peak info ONLY

4. Crystallography challenged: materials w/ disorder
Rietveld approach assumption: crystals are perfectly periodic…
…but this is not always the case!
Example: Ho2(Ti2‑xHox)O7-x/2 ”stuffed spin ice”
300K neutron diffraction patterns (GPPD, IPNS, Argonne)
x=0.67
x=0.00
x=0.30
x=0.50
pyrochlore
fluorite
x=0.3

5. Crystallography challenged: nano-crystals
Rietveld method
in powders
Figures: J.S.O. Evans et al.
and M. Tucker et al.
Bragg peak info ONLY

6. Global Approach Add complementary information Extra experimental data
FUTURE
Add complementary information
Extra experimental data
Theoretical constraints
Science, 316, 561 (2007).

7. What is PDF?
APPROACH
5.11Å
4.92Å
4.26Å
Pair distribution function (PDF) gives the probability of finding an atom at a distance “r” from a given atom.
3.76Å
2.84Å
2.46Å
1.42Å

8. Pair Distribution Function Method: Applicability
APPROACH
Lanthanum Aluminate glass
Th. Proffen et al, Los Alamos Science (2006)
Gold: bulk vs nanoparticle
Originally: short range order in liquids and glasses
Since late 1980’s: disorder in crystalline materials
Recently: nanocrystalline materials
J. Du and L.R. Corales, J. Non-Cryst. Solids (2007)

9. Pair Distribution Function from total scattering experiments
APPROACH
Structure function
Raw data
PDF

10. Pair Distribution Function from total scattering experiments
APPROACH
How can we get short range structural information?
Bragg intensity → long range order
Diffuse intensity →short range order

11. Pair Distribution Function: info on different lengthscales
APPROACH
Strength of the Pair Distribution Function
short range information
intermediate range information

12. Pair Distribution Function: info on different lengthscales
APPROACH
PDF primer: C60

13. Pair Distribution Function: Q-space resolution effect
APPROACH

14. Pair Distribution Function: rapid data collection mode
APPROACH
Rapid Acquisition PDF (RAPDF) at a synchrotoron:
Measuring PDFs in a few seconds
P. Chupas et al, J. Appl. Cryst. 36, (2003).

15. Nickel data, 0.03s collection time, Qmax 28 Å-1
Pair Distribution Function: rapid data collection mode
APPROACH
Application of a large-area, high-sensitivity, fast readout,
flat-panel GE detector based on an amorphous silicon
Time-resolved PDF measurements:
Reduction of PtIV Pt0
Nickel data, 0.03s collection time, Qmax 28 Å-1
P. Chupas et al, J. Appl. Cryst. 40, (2007).

16. Pair Distribution Function: modeling
APPROACH Ni
300K Ni data from GPPD at IPNS, ANL
data
model
difference
PDF Programs
PDFgetX1
PDFgetN1
PDFgui2 (PDFfit) 1 2

17. Nanometer scale structure of CuIr2S4
HIGHLIGHT
Local structural aspects of the metal-insulator transition
in CuIr2S4
- X-ray total scattering study -
Collaboration with: J.F. Mitchell, MSD, Argonne Nat. Lab.
Credits: Y.S. Hor, A.S. Masadeh, H.J. Kim, P. Juhas, S.J.L. Billinge

18. The role of lattice geometry – frustration
HIGHLIGHT
MgAl2O4
X sites
B octahedral sites
A tetrahedral sites
Spinel structure: AB2X Crystallizes in the cubic (isometric) crystal system B-sublattice – pyrochlore – corner-shared tetrahedra This sublattice promotes frustration, interesting physics arises in spin (AF interactions) and charge (half-integer valence) sectors!

19. Thiospinel CuIr2S4
HIGHLIGHT
At ~226 K it exhibits electronic, magnetic & structural transitions

20. Thiospinel CuIr2S4 properties: structural transition
HIGHLIGHT
- Electron diffraction
- TS226K
- in-situ TEM micrographs
Cubic
Fd-3m
Tetragonal
I41/amd
W. Sun et al., J. Phys.l Soc. Jpn., 70, 2817 (2001).

21. Thiospinel CuIr2S4 properties
HIGHLIGHT
- Metallic at high T
Insulating at low T
TMI226K
PES, NMR, LDA:
Cu Ir3.5+ (half-integer!)
PES of the insulating phase:
~0.1 eV gap near EF
Ir 5d DOS strongly distorted
S. Nagata et al., Physica B. 194, 1077 (1994).
J. Matsuno et al., Phys. Rev. B. 55, R15979 (1997).
T. Furubayashi et al., Solid State Comm. 126, 617 (2003).

22. Thiospinel CuIr2S4 properties
HIGHLIGHT
- High T: Pauli paramagnetic
Low T: nonmagnetic
(NMR, Mössbauer, magnetic susceptibility)
T. Furuyabashi et al., J. Phys. Soc. Japan. 63, 3333 (1994).
Ir8S24 octamer

23. CuIr2S4: background information
HIGHLIGHT
D.I. Khomskii and T. Mizokawa,
PRL 94, (2005).
P.G. Radaelli et al.,
Nature 416, 155 (2002).
Low-T structure
Red octahedra Ir3+
Blue octahedra Ir4+
- Metal at high temperature (Ir is nominally 3.5+)
- Insulator at low temperature (Ir’s are 3+ AND 4+)
- T-induced Metal-Insulator transition at ~226K
- Charge ordering and spin dimerization at low-T
- Long range charge ordered patterns of isomorphic octamers
- Associated structural change (Peierls distortion)
- Ir4+: short Ir-Ir distances appear (~3.0Å) - DIMER
- Ir3+: no short Ir-Ir distances (~3.5Å)

24. Reduced dimensionality due to specifics of xy-orbitals
HIGHLIGHT
CuIr2S4
D.I. Khomskii, Physica Scripta 72, CC8-14 (2005)

25. CuIr2S4: effect of x-ray irradiation
HIGHLIGHT
Long range
ordered dimers IC C
Triclinic
P-1
Cubic
Fd-3m
Insulator
Metal
V. Kiryukhin et al.,
PRL 97, (2006).
Tetragonal
I41/amd
Tetragonal
I41/amd
- Incommensurate (IC) short range
state below ~40K
- Commensurate (C) short range
state 40K<T<~100K
Melting of LRO dimers at low T!

26. CuIr2S4: effect of x-ray irradiation
HIGHLIGHT
When x-ray irradiated
no long range ordered
dimers
Long range
ordered dimers IC C
Triclinic
P-1
Cubic
Fd-3m
Insulator
Metal
V. Kiryukhin et al.,
PRL 97, (2006).
Tetragonal
I41/amd
- Incommensurate (IC) short range
state below ~40K
- Commensurate (C) short range
state 40K<T<~100K
T. Furubayashi et al., Solid State Comm. 126, 617 (2003).

27. CuIr2S4: local structure view of Metal-Insulator transition
HIGHLIGHT
Issues:
Do the dimers survive locally when the long range order is removed by:
temperature
Cr-doping
x-ray irradiation
Total Scattering Approach:
Crystallography – sensitive
to long range ordered dimers
Atomic PDF – sensitive to presence of dimers
no long range ordered
dimers
Long range ordered
dimers IC C
Triclinic
P-1
Cubic
Fd-3m
Insulator
Metal
Tetragonal
I41/amd

28. Neutron PDF experiment: Ir resonance issue
HIGHLIGHT

29. CuIr2S4: T-driven Metal-Insulator transition
HIGHLIGHT
Dramatic changes observed in the local structure, consistent with crystallography (T=10K)
R. Endoh et al.,
PRB 68, (2003).
I changed this one. I didn’t think the added much, in fact took away from the spectacular result of the 10K change, which is enhanced with respect to the 10 change 10k higher.
SJB: This is really very good now. Make sure you emphasize that both plots show a 10K temperature difference (and only 10K apart), so the signal of the phase transition is very clear in the local structure. Why is this a surprise? It won’t be to most people who haven’t done PDF so it may be better to give a counter example…the T-dependent phase transition in undoped manganite where the local JT distortion doesn’t disappear? On second thoughts, this may confuse people in the middle of a 10 minute talk, but having that slide as a backup if there is a question would be good.
30s
Compare the case where the MI
transition is not crossed! (T=10K)

30. Structure of LaMnO3 across the JT-transition at 720 K
LaMnO3: utilizing intuitiveness of PDF - simplicity
HIGHLIGHT
Structure of LaMnO3 across the JT-transition at 720 K
700 K data (blue) vs 750 K data (red)
This is a great plot. I think it would be stronger if you limited the range to say 15 so the Ir-Ir peak disappearance is amplified. It is in-your-face evidence…you don’t need to say any more. Forget the modeling, everyone will believe it if you just show the data! Well, don’t forget the modeling, put it in, but you get my meaning…really play this aspect up as it is great something like this can really be seen in the “raw” data.
30s
Distortions persist locally!

31. CuIr2S4: T-driven Metal-Insulator transition
HIGHLIGHT
Dramatic changes observed in the local structure, consistent with crystallography (T=10K)
R. Endoh et al.,
PRB 68, (2003).
I changed this one. I didn’t think the added much, in fact took away from the spectacular result of the 10K change, which is enhanced with respect to the 10 change 10k higher.
SJB: This is really very good now. Make sure you emphasize that both plots show a 10K temperature difference (and only 10K apart), so the signal of the phase transition is very clear in the local structure. Why is this a surprise? It won’t be to most people who haven’t done PDF so it may be better to give a counter example…the T-dependent phase transition in undoped manganite where the local JT distortion doesn’t disappear? On second thoughts, this may confuse people in the middle of a 10 minute talk, but having that slide as a backup if there is a question would be good.
30s
Compare the case where the MI
transition is not crossed! (T=10K)

32. CuIr2S4: Hysteretic structural behavior observed
HIGHLIGHT
R. Endoh et al.,
PRB 68, (2003).
Phase 1: Cubic K
Fraction of dimerized sample
from 2-phase fit
Emil, you can get the dimerized fraction directly from the short Ir-Ir peak if you assume it is fully there at below the transition and fully gone above. Does this lie on top of the result from the modeling? It should, and this is good indication that the modeling is doing an ok job. It is hard to see from these plots because they are on a different scale
Phase 2: Triclinic K

33. CuIr2S4: Cr-doping driven Metal-Insulator transition (5% Cr-doping)
HIGHLIGHT
R. Endoh et al.,
PRB 68, (2003).
This is a great plot. I think it would be stronger if you limited the range to say 15 so the Ir-Ir peak disappearance is amplified. It is in-your-face evidence…you don’t need to say any more. Forget the modeling, everyone will believe it if you just show the data! Well, don’t forget the modeling, put it in, but you get my meaning…really play this aspect up as it is great something like this can really be seen in the “raw” data.
30s
Dimers disappear locally when Insulator-Metal transition is invoked by Cr-doping at 200K

34. CuIr2S4: Melting the long range ordered dimerization pattern by x-rays
HIGHLIGHT
0.5s 1s
MAR345 exposures: 0.5s, 1s, 30s,
~2 minute break in exposing, 0.5s
Yes, this is good.
0.5s
30s

35. CuIr2S4: Melting the long range ordered dimerization pattern by x-rays
HIGHLIGHT
0.5s 1s
MAR345 exposures: 0.5s, 1s, 30s,
~2 minute break in exposing, 0.5s
Yes, this is good.
0.5s
30s

36. CuIr2S4: Melting the long range ordered dimerization pattern by x-rays
HIGHLIGHT
0.5s 1s
MAR345 exposures: 0.5s, 1s, 30s,
~2 minute break in exposing, 0.5s
Yes, this is good.
0.5s
30s

37. CuIr2S4: Melting the long range ordered dimerization pattern by x-rays
HIGHLIGHT
0.5s 1s
MAR345 exposures: 0.5s, 1s, 30s,
~2 minute break in exposing, 0.5s
Yes, this is good.
0.5s
30s

38. CuIr2S4: Melting the long range ordered dimerization pattern by x-rays
HIGHLIGHT
Melting observed in present study is approximately order of magnitude “faster” than that in the earlier reports
SJB: OK, this is good. If you look at the Ishibashi paper, the diffraction pattern is plotted on a log scale, which enhances small things happening near the baseline. It could work for you if you (1) subtract a background so the baseline is close to zero then (2) plot on a log scale. It might be worth a try.
Do you have any plots of YOUR data to compare to the top left panel? This would really strengthen this slide (and you can get rid of some of the text to make space…you already gave the conditions earlier. Here you are trying to establish that the pattern changed in the diffraction as later you will show that it doesn’t change in the PDF. One way to do it would be to plot blue and red diffraction data with a green difference curve below…this will emphasize the change in the diffraction. On the next slide no change in the PDF. Pick a range of the diffraction patter if you can that shows it well. Maybe this is not possible, but it would make it much stronger.
0.5s
30s
H. Ishibashi et al., PRB 66, (2002).

39. CuIr2S4: Melting the long range ordered dimerization pattern by x-rays – local aspects
HIGHLIGHT
15s
Cumulative
Data
Collection
collection
exposure
Something that is not coming through strongly to me is whether the changes in the diffraction pattern are comparable in the two cases…vs T and vs fluence. The story would be strengthened if you could do the following:
Show the low-Q diffraction plots and difference curves for the 3 cases side-by-side. They all look somewhat similar with the LRO peak going away
below that (or to the side with the other plots above one another), show the PDFs. The bottom line would be that the changes in the LRO look similar (if they do) but the changes in local structure are quite different. Conclusion? You should also check local structure as well as average structure.
0.5s
30s
Continuous
Exposure

40. the underlying local structures are the same!
CuIr2S4: Melting the long range ordered dimerization pattern by x-rays – local aspects
HIGHLIGHT
15s
Snapshot
Data
Collection
collection
23 sec
Collection: 250 msec snapshot
Exposure: continuous 23 sec
I added “local” to the text…check it is correct
0.5s
exposure
30s
Differences due to statistics only,
the underlying local structures are the same!

41. CuIr2S4: Melting the long range ordered dimerization pattern – overview
HIGHLIGHT
Diffraction
patterns
(long range order)
Temperature
Doping
Irradiation
PDF
profiles
(short range order)

42. CuIr2S4 thiospinel: Summary
HIGHLIGHT
Do the dimers survive locally when the long range order is removed by:
1. Temperature: NO: the local dimers are destroyed
Fraction of the sample that is dimerized has been mapped out through the hysteretic phase transition
2. Cr-doping: NO: the local dimers are destroyed
3. x-ray irradiation: YES: the local dimers survive
98keV x-rays also melt the long range order but the melting is faster than previous reports
Long range order recovers quickly at 100K as previously observed
Total Scattering Approach:
Crystallography – sensitive to long range order dimers
Atomic PDF – sensitive to presence of dimers
Dimer peaks clearly resolved and visible
Here is my suggestion for a summary

43. Acknowledgements People: Facilities: Funding: Simon Billinge
(Columbia/BNL)
Adam DeConinck
(MSU)
Pavol Juhas
(Columbia)
Xiangyun Qiu
(MSU, now at NIH)
Marek Schmidt
(MSU/ISIS, now at Polish Academy of Sciences)
Hyunjeong Kim
(LANSCE)
Ahmad Masadeh
(MSU, now at University of Jordan)
Gianluca Paglia
(MSU, now in industry, Australia)
Thomas Proffen
(LANSCE, Los Alamos)
John Mitchell
(MSD, Argonne)
Tapan Chatterji
(ILL, Grenoble, FR)
Paolo Radaelli
(ISIS, UK)
Peter Chupas, Douglas Robinson
(APS, Argonne)
Facilities:
Advanced Photon Source, Argonne
Intense Pulsed Neutron Source, Argonne
Los Alamos Neutron Scattering Center
ISIS, Rutherford Appleton Laboratory
Funding:
NSF DMR ,
DOE DE-AC02-06CH11375,
DOE DE-AC52-06NA25396,
DOE DE-AC02-98CH10886.

44. Thank you!