Pushing data quality for laboratory pair

distribution function experiments

Cite as: Rev. Sci. Instrum. 90, 043905 (2019); https://doi.org/10.1063/1.5093714
Submitted: 24 February 2019 . Accepted: 26 March 2019 . Published Online: 17 April 2019

Sabrina L. J. Thomae, Nils Prinz, Thomas Hartmann, Michael Teck, Sascha Correll, and Mirijam Zobel

ARTICLES YOU MAY BE INTERESTED IN

High-performance small- and wide-angle X-ray scattering (SAXS/WAXS) experiments on a

multi-functional laboratory goniometer platform with easily exchangeable X-ray modules
Review of Scientific Instruments 89, 085115 (2018); https://doi.org/10.1063/1.5041949

High-throughput powder diffraction measurement system consisting of multiple MYTHEN

detectors at beamline BL02B2 of SPring-8
Review of Scientific Instruments 88, 085111 (2017); https://doi.org/10.1063/1.4999454

Benchmark oxygen-oxygen pair-distribution function of ambient water from x-ray

diffraction measurements with a wide Q-range
The Journal of Chemical Physics 138, 074506 (2013); https://doi.org/10.1063/1.4790861

Rev. Sci. Instrum. 90, 043905 (2019); https://doi.org/10.1063/1.5093714 90, 043905

© 2019 Author(s).
 Review of ARTICLE scitation.org/journal/rsi
Scientific Instruments

Pushing data quality for laboratory pair

distribution function experiments
Cite as: Rev. Sci. Instrum. 90, 043905 (2019); doi: 10.1063/1.5093714
Submitted: 24 February 2019 • Accepted: 26 March 2019 •
Published Online: 17 April 2019

Sabrina L. J. Thomae,1 Nils Prinz,1 Thomas Hartmann,2 Michael Teck,2 Sascha Correll,2 and Mirijam Zobel1,a)

AFFILIATIONS
1
Chemistry Department, University Bayreuth, Universitaetsstr. 30, Bayreuth 95447, Germany
2
STOE & Cie GmbH, Hilpertstr. 10, Darmstadt 64295, Germany

a)
Author to whom correspondence should be addressed: mirijam.zobel@uni-bayreuth.de

ABSTRACT
Over the last decade, some studies with laboratory pair distribution function (PDF) data emerged. Yet, limited Qmax or instrumental res-
olution impeded in-depth structural refinements. With more advanced detector technologies, the question arose how to design novel PDF
equipment for laboratories that will allow decent PDF refinements over r = 1–70 Å. It is crucial to reflect the essential requirements, namely,
monochromatic X-rays, suppression of air scattering, instrumental resolution, and overall measurement times. The result is a novel PDF
setup based on a STOE STADI P powder diffractometer in transmission-/Debye-Scherrer geometry with monochromatic Ag Kα1 radiation,
featuring a MYTHEN2 4K detector covering a Q range of 0.3–20.5 Å−1 . PDF data are collected in a moving PDF mode within 6 h. Structural
signatures of liquids can be satisfactorily resolved in the PDF as shown for the ionic liquid hmimPF6 . The high instrumental resolution is
mirrored in low qdamp values determined from LaB6 measurements. PDF data from a powder sample of ca. 7 nm TiO2 nanoparticles were
successfully refined over up to 70 Å with goodness-of-fit values Rw < 0.22 (respectively Rw = 0.18 over 30 Å), thanks to the low background
and high instrumental resolution, hereby enlarging the accessible r range by several tens of Angstroms compared to previous laboratory PDF
studies.
Published under license by AIP Publishing. https://doi.org/10.1063/1.5093714

I. INTRODUCTION scattering cross-sections of all contributing atoms. Thus, intrinsi-

Pair distribution function (PDF) analysis is employed to struc-
turally characterize crystallographically challenging materials, such
as glasses, liquids, or bulk materials with nanoscale disorder.1 Thus,
the PDF enjoys increasing popularity in the field of nanotechnol-
ogy as it is capable to access the structure of molecular clusters and
nanocrystalline materials.2 X-ray PDF data are obtained by Fourier
transformation of high-energy X-ray scattering data over a reason-
ably large Q range, at synchrotron radiation facilities, typically ca.
0.8–30 Å−1 ; see the following equation:
2 Qmax
G(r) = ∫ Q[S(Q) − 1] sin(Qr)dQ, (1)
π Qmin
where Q = 4π(sin θ/λ) is the wave vector transfer with wave-
length λ and scattering angle 2θ and S(Q) is the (total) scattering
structure function, which is obtained from the measured intensity
I(Q) by correcting for the background and experimental aspects
such as detector sensitivity and applying a normalization for the
cally, the PDF technique uses both, Bragg and diffuse scattering,
to retrieve structural information from powder diffraction data. In
order to obtain a most accurate S(Q), each measurement requires
a background dataset measured under identical experimental con-
ditions in order to properly account for the various background
contributions such as the sample container, Compton scattering, or
absorption.
Although the X-ray PDF technique had mostly been restricted
to synchrotron data, a couple of studies employing laboratory PDF
data emerged over the last decade, primarily, comprising feasibil-
ity studies.3–6 Some studies used laboratory PDF data for compar-
ison with neutron data7,8 or studied solutions7 and glasses with
low Qmax .9 Most frequently, strong instrumental damping and lim-
ited Qmax ≤ 17 Å−1 prevail,9,10 and the laboratory PDF data are
used for qualitative conclusions rather than pushing the goodness
of fit, Rw , to minimal values for detailed structural insights.5,11 Rw
is defined by the following equation, where Gobs (r) and Gcalc (r)
are the observed and modeled PDFs, respectively. w(r) is the

Rev. Sci. Instrum. 90, 043905 (2019); doi: 10.1063/1.5093714 90, 043905-1
Published under license by AIP Publishing
 Review of
Scientific Instruments
weighting factor and N is the scale factor of the model to the
data:12
2 1/2
∑i w(ri )[Gobs (ri ) − NGcalc (ri )]
Rw = { 2
} .
∑i w(ri )[Gobs (ri )]
Previous comparison of synchrotron with laboratory PDF data was,
in part, even based on the simulation of the laboratory data.3 Fur-
thermore, fit ranges often did not exceed 20 Å.6 With previous
improvements in monochromatization and detector technology,
here we report laboratory PDF data which provide very high good-
ness of fit (low Rw ) for fit ranges even up to 70 Å for nanoparticle
powders.
II. GENERAL INSTRUMENTAL CONSIDERATIONS
With decreasing Qmax value used for the Fourier transforma-
tion [see Eq. (1)], the PDF peaks may shift, their peak heights
change, and their full width half maxima (FWHM) increase. This
becomes important, in particular, when Qmax gets below 17 Å−1 , see
the behavior of the PDFs of TiO2 nanoparticles for varying Qmax
in Fig. S1 in the supplementary material. With molybdenum (Mo)
X-ray sources on laboratory diffractometers, Qmax = 17 Å−1 can be
reached at best, but the experimentally useful Qmax on average is ca.
1–2 Å−1 lower. Therefore, a silver (Ag) tube has to be used reach-
ing ideally Qmax = 21 Å−1 in order to build an instrument capable of
reaching Qmax values for which the PDF peak broadening due to low
Qmax cut-off values is negligible.
The quantum efficiency of detector materials is decreasing with
higher X-ray energies. High Z materials such as CdTe are outstand-
ing candidates for implementation at synchrotrons, where energies
of up to 80 keV are detected with 77% efficiency with 1000 µm
thick CdTe chips.17 Naturally, this also holds for laboratory setups.
However, as Ag X-rays of 22.16 keV are only one third in energy
compared to standard synchrotron setups of 60–90 keV, the gain
in quantum efficiency with CdTe is not as large. On the contrary,
CdTe is a rather temperature-sensitive material and requires cool-
ing to prevent charge accumulation at grain boundaries, which is
FIG. 1. Picture of STADI P in vertical geometry with MYTHEN2 4K module (a) and schemes for stationary mode (b) and moving mode (c). In both schemes, optical elements
such as monochromators and slits are omitted for clarity, as well as the beamstop on the secondary side. The sample is represented by the gray circle in the middle and the
X-ray path is represented by the blue arrows. The four MYTHEN modules are displayed in red together with their trapezoidal fans for reduced background in front of them in
gray. In the stationary mode (b), the 3rd and 4th modules detect the data in the negative 2θ range, which is folded into the positive 2θ range afterward. In the moving mode
(c), a continuous 2θ range is covered by the movement of the detector arm, see the additional light red module positions.
Rev. Sci. Instrum. 90, 043905 (2019); doi: 10.1063/1.5093714
Published under license by AIP Publishing
ARTICLE scitation.org/journal/rsi
likely to cause additional background in measurements. For labo-
ratory setups with Ag X-rays, 1 mm Si chips feature 50% quan-
tum efficiency (for Mo X-rays 75%)13 and are provided in several
detectors, e.g., Bruker LynxEYE XE-T or the Hybrid Photon Count-
ing (HPC) technology in the MYTHEN series. Additionally, those
(2)
Si detectors offer negligible background due to the availability of
energy thresholds in the X-ray detection, producing completely fea-
tureless background baselines over the entire 2θ range. Despite the
lower flux (ca. half)14 and lower detection efficiency of Ag X-rays
compared to Mo X-rays from X-ray tubes, Ag X-ray tubes are
beneficial for the overall data quality because of the achiev-
able Q-range >17 Å−1 which contains additional structural
information.
Opposed to the common assumption that high-energy X-rays
do not scatter significantly in air, proper air shielding is in fact the
key to low-background laboratory PDF data. This involves (i) a long
collimator reaching very close to the sample and (ii) a beamstop with
optimized positioning close to the sample to prevent air scattering.
III. LAYOUT AND OPERATION OF THE STADI P PDF
MYTHEN2 4K
The novel PDF setup is based on a vertically mounted standard
STOE STADI P diffractometer in transmission-/Debye-Scherrer
geometry with a curved Ge(111) monchromator yielding pure Kα1
radiation from the line focus window of a standard sealed tube
(Malvern Panalytical C-Tech, Ag, 2200 W). The beam is focused
on the detector (Dectris MYTHEN2 4K) in 260 mm distance with
the sample position halfway. The MYTHEN2 4K consists of four
Dectris MYTHEN2 R 1K modules, which are Si-strip detectors with
1280 strips/module, 50 µm strip width, and 1000 µm sensor thick-
ness connected to a one detector control system (Dectris DCS4). The
four modules are fixed on a curved arm positioning them exactly
on the focusing circle of the STADI P. In order to minimize air
scattering, the use of a collimator ending with a slit close to the cap-
illary is as mandatory as a fan shaped air scattering protection in
front of each detector module. Each module offers an angular aper-
ture close to 18.5○ 2θ, and they are positioned with a gap of 18○ in
between them, see Fig. 1. Both values are rounded as they depend
90, 043905-2
 Review of
Scientific Instruments
on the individual calibration curves of each module and have to be
an integer multiple of the internal step width. The MYTHEN2 4K
can be used in two different modes, a stationary one offering a 2θ
range from 0.76○ to 73○ (0.3–13 Å−1 ) and a “moving” mode up to
141○ 2θ (Qmax = 20.52 Å−1 ).
In the stationary setup, the detector arm stays at a fixed posi-
tion with the modules at −54.5○ to −36.0○ , −18.5○ to 0.0○ , 18.0○
to 36.5○ , and 54.0○ to 72.5○ 2θ. After the exposure, the pattern is
assembled by the software (STOE WinXPOW 3.12), which folds the
data of the modules from the negative 2θ range [module number
3 and 4 in Fig. 1(b)] into the corresponding gaps in the positive 2θ
range.
In the “moving” mode, the detector arm operates only in the
positive 2θ range, see Fig. 1(c). The detector arm shifts by 18○ 2θ
after half of the measurement time and creates two datasets at dif-
ferent positions, which are combined to one full pattern of 141○ 2θ
by the software afterward. To achieve better background statis-
tics, however, it is better not to shift the detector arm once by
18○ , but to shift only in increments of 2.65○ 2θ (module width
divided by seven). Hence, eight individual datasets with offsets of
2.65○ 2θ are measured. As only the overlapping 2θ areas of all
8 ranges can be finally added to one pattern (for reasons of iden-
tical statistics in each angular range), the resulting 2θ range yields
132○ (Q = 20.5 Å−1 ).
The sample can be fixed between two acetate foils or in a cap-
illary, while permanent spinning improves the statistics. To avoid
possible shadowing of the scattered signal by the transmission sam-
ple holder operated in 2θ/ω mode at high angles, capillaries are
favorable for the PDF measurements.
To estimate measuring times for a data file of sufficient qual-
ity, datasets of 6 and 20 h overall measurement time were compared.
The corresponding I(Q) and derived PDFs for 6 and 20 h are almost
identical with the 6 h raw data featuring more noise, in particu-
lar, for high Q, see Fig. S2. The resulting higher termination ripples
in the 6 h PDF data are most pronounced with increasing inter-
atomic distances, but do not affect structural refinements. Conse-
quently, a measurement time of 6 h was used for the data herein, see
Secs. V A and V B.
IV. EXPERIMENTAL DETAILS
A. Chemicals
Titanium dioxide (TiO2 ) particles of 5 nm diameter were pur-
chased from NanoAmor in 2014, and particle size was checked
with Transmission electron microscopy (TEM) (no surface stabiliza-
tion by organic ligands, anatase structure; Nanostructured & Amor-
phous Materials, Inc., Houston, TX 77084, USA). The ionic liq-
uid hmimPF6 (1-hexyl-3-methylimidazolium hexafluorophosphate,
C10 H19 F6 N2 P) has been provided by Eric Meshot, Lawrence Liver-
more National Laboratory.
B. Transmission electron microscopy
TEM images of the TiO2 nanoparticles for the evaluation of the
particle size were acquired with a Zeiss Leo EM922 Omega TEM
with an accelerating voltage of 200 kV. 4 µl of very diluted NP dis-
persion were drop-casted on a carbon-coated meshed copper grid
and dried in air.
Rev. Sci. Instrum. 90, 043905 (2019); doi: 10.1063/1.5093714
Published under license by AIP Publishing
ARTICLE scitation.org/journal/rsi
C. Laboratory PDF measurements
All laboratory PDF data were collected in the moving mode on
the STOE STADI P setup as described in Sec. II with a total mea-
surement time of 6 h for the LaB6 and TiO2 , and 40 h for the ionic
liquid.
D. Synchrotron PDF experiments
TiO2 data were collected in a Kapton capillary (diameter
®
1 mm) at beamline 11-ID-B at the Advanced Photon Source (APS,
Chicago) at 58.6 keV on a Perkin Elmer detector at the end of 2015.
The data were collected in 30 s. The hmimPF6 synchrotron data
were collected for 2 min at beamline ID15-ERH at the European
Synchrotron Radiation Facility (ESRF, Grenoble) at 68 keV on a
Pilatus 2M CdTe detector. Distance calibration was done with NIST
CeO2 at the APS and Si standards at the ESRF. All empty capil-
laries were taken under the same experimental conditions as the
samples.
E. Data analysis
Both synchrotron and laboratory PDFs were calculated with
PDFgetX3,4 using Qmax values of ca. 19 Å−1 for means of compar-
ison. The Qmax values were chosen for the best possible PDF data
for the individual datasets. Qmax of the ceria data from APS was
19.9 Å−1 , and that of the TiO2 one it was 19.3 Å−1 . For the cor-
responding laboratory data, the values were 18.5 Å−1 for LaB6 and
18.6 Å−1 for TiO2. For the hmimPF6 data, Qmax for laboratory data
was 17.7 Å−1 and for the ESRF data it was 18.9 Å−1 , as higher Qmax
values in both cases resulted only in more high frequency ripples
in the PDF data without smoothing of the raw data. Fits were done
with PDFgui.15
V. RESULTS
A. Instrumental resolution with LaB6
Refinements of the laboratory LaB6 data from STOE have
been carried out over varying r ranges starting at rmin = 1 Å
and ending at rmax = 30, 40, 50, 60, 70, and 80 Å to derive the
largest meaningful range of the PDF data. Figure S3 and Table S1
show the results and details for two different refinement strate-
gies (keeping qdamp and qbroad constant as determined over 80 Å
or letting them refine for each rmax ). As a result, a fit range of
80 Å can be used by providing parameters for the instrumental
resolution of qdamp = 0.011 and qbroad = 0.010. Figure 2 shows
this refinement and clearly illustrates that the data are only little
damped for increasing distances. This allows to derive nanoparticle
sizes well within the general limitations of size determination from
PDF data.
Previous literature on PDF from laboratory sources merely
contains refinements over data ranges of 20 Å.4,6,16 To our knowl-
edge, being able to fit standards with an Rw = 0.14 over 80 Å is unique
at the present point for laboratory PDF diffractometers.
B. Nanoparticle powders—Showcase TiO2
The diffraction data of a powder of TiO2 anatase nanoparti-
cles with diameters of ca. 7 nm are displayed in Fig. S5, together
with the empty glass capillary, showing the very low scattering back-
ground. Despite some noise at large Q (>50 Å), the instrument is
90, 043905-3
 Review of ARTICLE scitation.org/journal/rsi
Scientific Instruments

in place of the monochromator, a rather broad wavelength band

FIG. 2. PDF refinements of LaB6 .
able to separate close peaks [inset (c)] and rise peaks above the
background even for Q > 13 Å−1 [inset (d)]. This improvement in
data quality is made possible by the Ge(111) monochromator which
provides pure Ag Kα1 radiation. If only ellipsoidal mirrors were used
contains Kα1,2 mixtures. Such a mixture not only smears out features
in the XRD but also smears out peaks in the PDF with increasing
distances, becoming pronounced above 20 Å.
Based on the high instrumental resolution and the low back-
ground of the STOE STADI P, nanoparticle powders can be refined
and their sizes determined over large fit ranges out to 70 Å, see
Fig. 3(a) [see Fig. S6 for corresponding S(Q) and F(Q)]. The refine-
ment of the laboratory PDF data over r = 70 Å resulted in Rw = 0.22.
The difference curve contains no more structural features. For com-
parison, data on the same TiO2 nanoparticle powder were acquired
at beamline 11-ID-B at the APS (see Fig. S7 for the refinement with
Rw = 0.16), and the particle size was furthermore confirmed by TEM
(Fig. S8).
In order to better compare with previous laboratory PDF stud-
ies, the TiO2 data were additionally refined over various r-ranges,
i.e., with rmax = 20, 30, 50, and 70 Å. For the refinements, we used
the instrumental resolution parameters, qdamp and qbroad , from the
LaB6 refinement, and the particle size (spdiameter) was held con-
stant at the value refined over 70 Å. The results of these refinements
are compared in Table I and the fit over 30 Å with Rw = 0.18 is shown
in Fig. 3(b). It has to be pointed out that the PDF peaks above 20 Å,
which are resolved, thanks to the Kα1 monochromatization, are
well described in the refinement. When the r range is reduced to

FIG. 3. PDF refinement of data from the STOE STADI P of TiO2 nanoparticles showing the experimental PDF (blue), the fit (red), and the difference (gray, in offset) for the
data range of 70 Å in (a), while the range 0–15 Å is enlarged. Refinement of the same data set over 30 Å (b). The fit range started each at r = 1 Å and fit curves were
extrapolated to r = 0 Å.

TABLE I. Fit results of the TiO2 refinement of laboratory STOE STADI P and synchrotron data. For both fits, the particle size
(spdiameter) was refined from the 1–70 Å range and then held constant for smaller fit ranges. STOE data are refined with
qdamp = 0.011, qbroad = 0.010, and spdiameter = 74 Å and APS data with qdamp = 0.036, qbroad = 0.0052, and spdiameter
= 68 Å.

Lab 1–70 Lab 1–50 Lab 1–30 Lab 1–20 APS 1–70 APS 1–30

Rw 0.22 0.20 0.18 0.15 0.16 0.15

a (Å) 3.7948 3.7949 3.7944 3.7944 3.7900 3.7897
c (Å) 9.5316 9.5315 9.5312 9.5293 9.5176 9.5159
δ1 (Å2 ) 1.2 1.2 1.2 1.2 1.3 1.3
uiso,Ti (Å2 ) 0.00837 0.00838 0.00807 0.00764 0.00703 0.00692
uiso,O (Å2 ) 0.0198 0.0198 0.0201 0.0205 0.0176 0.0175

Rev. Sci. Instrum. 90, 043905 (2019); doi: 10.1063/1.5093714 90, 043905-4
Published under license by AIP Publishing
 Review of
Scientific Instruments
FIG. 4. PDFs of hmimPF6 from STOE STADI P (Qmax = 17.7 Å−1 ) and beamline
ID15-ERH, ESRF (Qmax = 18.9 Å−1 ).
rmax = 20 Å as used in previous laboratory PDF reports, our
goodness-of-fit beats by far previous fit qualities of laboratory PDF
data.4,6,16
C. Liquids—Showcase ionic liquid
Liquids, glasses, and amorphous solids elude traditional crys-
tallographic analysis; however, the PDF allows to directly asses
intramolecular distances and intermolecular short-range order. One
key prerequisite for measuring PDF data is a sufficiently low Qmin ,
as the main features of the liquid structure are contained in the first
sharp diffraction peak (FSDP) at low Q. Figure 4 shows the PDF
of the ionic liquid hmimPF6 collected on the STOE STADI P with
Qmin = 0.3 Å−1 in comparison to the PDF data collected at ID15-
ERH and ESRF. There are more termination ripples in the labora-
tory PDF data because of the deficient signal-to-noise ratio in the
lab. Yet, the essential features of the liquid are well contained in
the laboratory PDF data [for corresponding S(Q) and F(Q) data,
see Fig. S9].
VI. CONCLUSIONS
In this work, a novel design of a laboratory PDF diffractome-
ter is laid out based on the STOE STADI P with an Ag anode
and a MYTHEN2 4K detector system, consisting of four Dectris
MYTHEN2 R 1K modules. Based on data collection times of 6 h
over an I(Q) range of 0.3–20.5 Å−1 , it is shown that besides high
instrumental resolution and low background, unprecedented PDF
data quality is achieved. All relevant features of the ionic liquid
hmimPF6 as a test case for liquid samples are satisfactorily resolved
in the PDF. PDF data from a powder sample of ca. 7 nm TiO2
nanoparticles (anatase structure) were successfully refined result-
ing in Rw values of as low as 0.18 over a fit range of 1–30 Å. It is
even possible to carry out the same laboratory PDF refinement over
Rev. Sci. Instrum. 90, 043905 (2019); doi: 10.1063/1.5093714
Published under license by AIP Publishing
ARTICLE scitation.org/journal/rsi
r = 1–70 Å with only slightly poorer fit qualities (Rw = 0.22), brought
about by the monochromatization of the Kα1 Ag X-rays. This is—
to our knowledge—the very first example of high-quality laboratory
PDF data to be fit beyond r = 30 Å and thus provide the basis for
in-depth structural refinements based on laboratory PDF data.
SUPPLEMENTARY MATERIAL
See supplementary material for additional figures and tables.
ACKNOWLEDGMENTS
We acknowledge the Deutsche Forschungsgemeinschaft (SFB
840 - C7 and SPP 2080, Grant No. ZO 369/2-1) and the Bavar-
ian Polymer Institute (BPI) for funding. M.Z. acknowledges M. W.
Terban for general discussions, D. S. Jung for discussions on detec-
tor technology, and E. Meshot for hmimPF6 samples. We acknowl-
edge the granted beamtime and support at 11-ID-B, Advanced Pho-
ton Source (K. Chapman, P. Chupas and K. Beyer), as all as at
ID15-ERH, European Synchrotron Radiation Facility (G. Vaughan).
REFERENCES
1
S. J. Billinge and M. G. Kanatzidis, Chem. Commun. 749–760 (2004).
2
S. J. Billinge and I. Levin, Science 316(5824), 561–565 (2007).
3
T. Dykhne, R. Taylor, A. Florence, and S. J. Billinge, Pharm. Res. 28(5),
1041–1048 (2011).
4
P. Juhás, T. Davis, C. L. Farrow, and S. J. L. Billinge, J. Appl. Crystallogr. 46(2),
560–566 (2013).
5
J. te Nijenhuis, M. Gateshki, and M. J. Fransen, Z. Kristallogr. Suppl. 2009(30),
163–169.
6
G. Confalonieri, M. Dapiaggi, M. Sommariva, M. Gateshki, A. N. Fitch, and
A. Bernasconi, Powder Diffr. 30(S1), S65–S69 (2015).
7
F. Meersman, D. Bowron, A. K. Soper, and M. H. Koch, Phys. Chem. Chem.
Phys. 13(30), 13765–13771 (2011).
8
T. D. Bennett, A. L. Goodwin, M. T. Dove, D. A. Keen, M. G. Tucker, E. R.
Barney, A. K. Soper, E. G. Bithell, J. C. Tan, and A. K. Cheetham, Phys. Rev. Lett.
104(11), 115503 (2010).
9
L. Cormier and G. J. Cuello, Geochim. Cosmochim. Acta 122, 498–510
(2013).
10
B. Bueken, N. Van Velthoven, T. Willhammar, T. Stassin, I. Stassen, D. A. Keen,
G. V. Baron, J. F. M. Denayer, R. Ameloot, S. Bals, D. De Vos, and T. D. Bennett,
Chem. Sci. 8(5), 3939–3948 (2017).
11
M. Sommariva, M. Gateshki, J.-A. Gertenbach, J. Bolze, U. König, B. Ş. Vasile,
and V.-A. Surdu, Powder Diffr. 29(S1), S47–S53 (2014).
12
P. F. Peterson, E. S. Bozin, T. Proffen, and S. J. L. Billinge, J. Appl. Crystallogr.
36, 53–64 (2003).
13
T. Donath, S. Brandstetter, L. Cibik, S. Commichau, P. Hofer,
M. Krumrey, B. Lüthi, S. Marggraf, P. Müller, M. Schneebeli, C. Schulze-Briese,
and J. Wernecke, J. Phys.: Conf. Ser. 425(6), 062001 (2013).
14
V. Honkimäki, J. Sleight, and P. Suortti, J. Appl. Crystallogr. 23(5), 412–417
(1990).
15
C. L. Farrow, P. Juhas, J. W. Liu, D. Bryndin, E. S. Bozin, J. Bloch, T. Proffen,
and S. J. Billinge, J. Phys.: Condens. Matter 19(33), 335219 (2007).
16
K. Galliez, P. Deniard, P.-E. Petit, D. Lambertin, F. Bart, and S. Jobic, J. Appl.
Crystallogr. 47(2), 552–560 (2014).
17
D. Šišak Jung, T. Donath, O. Magdysyuk, and J. Bednarcik, Powder Diffr. 32,
S22–S27 (2017).
90, 043905-5