Synthesis and characterization of K8-x(H2) ySi46

Article abstract of DOI:10.1021/ic9004592

A hydrogen-containing inorganic clathrate with the nominal composition, K₇(H₂)₃Si₄₆, has been prepared in 98% yield by the reaction of K₄Si₄ with NH₄Br. Rietveld refinement of the powder X-ray diffraction data is consistent with the clathrate type I structure. Elemental analysis and ¹H MAS NMR confirmed the presence of hydrogen in this material. Type I clathrate structure is built up from a Si framework with two types of cages where the guest species, in this case K and H₂, can reside: a large cage composed of 24 Si, in which the guest resides in the-6d position, and a smaller one composed of 20 Si, in which the guest occupies the 2a position (cubic space group Pm3 n). Potassium occupancy was examined using spherical aberration (Cs) corrected scanning transmission electron microscopy (STEM). The highangle annular dark-field (HAADF) STEM experimental and simulated images indicated that the K is deficient in both the 2a and the 6dsites. ¹H and ²⁹Si MAS NMR are consistent with the presence of H₂ in a restricted environment and the clathrate I structure, respectively. FTIR and ²⁹Si{¹H} CP MAS NMR results show no evidence for a Si-H bond, suggesting that hydrogen is present as H₂ in interstitial sites. Thermal gravimetry (TG) mass spectrometry (MS) provide additional confirmation of H₂ with hydrogen loss at ～400 °C.

Full text of DOI:10.1021/ic9004592

Inorg. Chem. 2010, 49, 815–822 815
DOI: 10.1021/ic9004592
Synthesis and Characterization of K8-x(H ) Si
2 y 46
†
‡
§
†
Doinita Neiner, Norihiko L. Okamoto, Ping Yu, Sharon Leonard, Cathie L. Condron, Michael F. Toney,
^
‡,^,#
,†
Quentin M. Ramasse, Nigel D. Browning,
and Susan M. Kauzlarich*
†
‡
§
Department of Chemistry, Department of Chemical Engineering and Materials Science, and NMR Facility,
University of California, One Shields Avenue, Davis, California 95616, Stanford Synchrotron Radiation
^
Lightsource, SLAC National Accelerator Laboratory, Menlo Park, California 94025, National Center of
Electron Microscopy, Lawrence Berkeley Laboratory, 1 Cyclotron Road, Berkeley, California 94270, and
#
Physical and Life Sciences Directorate, Lawrence Livermore National Laboratory, Livermore,
California 94550
Received March 9, 2009
A hydrogen-containing inorganic clathrate with the nominal composition, K (H ) Si , has been prepared in 98% yield
2 3 46
7
by the reaction of K Si with NH Br. Rietveld refinement of the powder X-ray diffraction data is consistent with the
4
4
4
1
clathrate type I structure. Elemental analysis and H MAS NMR confirmed the presence of hydrogen in this material.
Type I clathrate structure is built up from a Si framework with two types of cages where the guest species, in this case K
and H , can reside: a large cage composed of 24 Si, in which the guest resides in the 6d position, and a smaller one
2
composed of 20 Si, in which the guest occupies the 2a position (cubic space group Pm 3h n). Potassium occupancy was
examined using spherical aberration (Cs) corrected scanning transmission electron microscopy (STEM). The high-
angle annular dark-field (HAADF) STEM experimental and simulated images indicated that the K is deficient in both
1
29
the 2a and the 6d sites. H and Si MAS NMR are consistent with the presence of H in a restricted environment and
2
2
9
1
the clathrate I structure, respectively. FTIR and Si{ H} CP MAS NMR results show no evidence for a Si-H bond,
suggesting that hydrogen is present as H in interstitial sites. Thermal gravimetry (TG) mass spectrometry (MS)
provide additional confirmation of H with hydrogen loss at ∼ 400 ꢀC.
2
2
Introduction
have been very extensively investigated. Most of type I
clathrates, M T , crystallize in Pm3n, cubic cell consisting
8
46
Clathrates constitute a unique class of inclusion com-
pounds with a rich crystal chemistry that allows for a variety
of physical properties such as thermoelectricity, supercon-
of two dodecahedra (T cage) and six tetrakaidecahedra (T
24
20
cage) encapsulating guests (M) at 2a and 6d sites. Type II
clathrates, M T136, have a cubic cell, Fd3m, made of penta-
1-3
24
ductivity, and hydrogen storage.
The crystal structure of
gonal dodecahedra (T ) and hexakaidecahedra (T ) while
20
28
clathrates consists of three-dimensional host frameworks
6,7
4,5
the 24 guest atoms M, occupy 8b and 16c positions. An
illustration of the different clathrate cages in Type I is
presented in Figure 1. The 6d site is shaded.
that encapsulate different guest molecules. The framework
is formed of polyhedra with pentagonal and/or hexagonal
faces. The variation in these structures arises from the type of
polyhedra that host atoms form and the polyhedra con-
Inorganic clathrates (type I and II) in which group 14
elements make up the framework structure and alkali metals
reside in the guest sites have been studied for their possible
4-6
nectivity.
Among the numerous structure types of the
inorganic clathrate compounds, type I and type II clathrates
2,3,5,8
thermoelectric applications.
The type II clathrate
hydrates in which H O molecules comprise the framework
2
*
To whom correspondence should be addressed. E-mail: smkauzlarich@
structure with H2 molecules as the guest species have
been of interest for hydrogen storage applications. We
have reported the synthesis of a new type-I Si clathrate,
1,9
ucdavis.edu.
(
1) Lee, H.; Lee, J. W.; Kim, D. Y.; Park, J.; Seo, Y. T.; Zeng, H.;
Moudrakovski, I. L.; Ratcliffe, C. I.; Ripmeester, J. A. Nature 2005, 434,
7
43–746.
2) Iversen, B. B.; Palmqvist, A. E. C.; Cox, D. E.; Nolas, G. S.; Stucky,
G. D.; Blake, N. P.; Metiu, H. J. Solid State Chem. 2000, 149, 455–458.
(
(
7) Ramachandran, G. K.; Dong, J. J.; Diefenbacher, J.; Gryko, J.;
(
3) Bobev, S.; Sevov, S. C. J. Solid State Chem. 2000, 153, 92–105.
4) Kasper, J. S.; Hagenmuller, P.; Pouchard, M.; Cros, C. Science 1965,
Marzke, R. F.; Sankey, O. F.; McMillan, P. F. J. Solid State Chem. 1999,
145, 716–730.
(
1
50, 1713–1714.
(8) Qiu, L. Y.; White, M. A.; Li, Z. Q.; Tse, J. S.; Ratcliffe, C. I.; Tulk,
C. A.; Dong, J. J.; Sankey, O. F. Phys. Rev. B 2001, 6402.
(9) Hester, K. C.; Strobel, T. A.; Sloan, E. D.; Koh, C. A.; Huq, A.;
Schultz, A. J. J. Phys. Chem. B 2006, 110, 14024–14027.
(
(
5) Kovnir, K. A.; Shevelkov, A. V. Uspekhi Khimii 2004, 73, 999–1015.
6) Pouchard, M.; Cros, C.; Hagenmuller, P.; Reny, E.; Ammar, A.;
Menetrier, M.; Bassat, J. M. Solid State Sci. 2002, 4, 723–729.
r 2009 American Chemical Society
Published on Web 12/29/2009
pubs.acs.org/IC

8
16 Inorganic Chemistry, Vol. 49, No. 3, 2010
Neiner et al.
and dwelled there for 3-4 days. The purity of the obtained K Si
4
4
material was confirmed by powder X-ray diffraction. NH Br
4
(
99.99% Aldrich) was dried under dynamic vacuum at 100 ꢀC
for 14 h prior to usage. The synthesis of the potassium silicon
clathrate was achieved by the solid state reaction between the
Zintl salt, K
of 0.25:0.8 molar ratio, 0.755 g of K Si and 0.8825 g of NH Br,
4 4 4
Si , and NH Br. The optimized reaction consisted
4
4
4
ground in an agate mortar in a nitrogen filled glovebox and the
resulting powder pressed into pellets. The pellets were loaded in
a 20 mm long alumina boat and placed in a closed end glass tube
with adaptor. The glass tube dimensions must be carefully
considered since both ammonia and hydrogen gases are gene-
rated in this reaction. The tube was 14.6 long with an internal
diameter of 3/4 mm, and the adapter provided an additional
1
.5 in length. The tube was connected to a Schlenk line and left
under vacuum for 30 min. The tube was then inserted in a
furnace preheated at 300 ꢀC with the sample positioned in the
center of the furnace and the adapter outside of the furnace. The
reaction vessel dwelled at 300 ꢀC for 13 h and then was removed
from the furnace and air-cooled to room temperature (RT).
In all reactions, the solid product was determined to be KBr,
mainly by X-ray powder diffraction. The salt byproduct is
removed by washing first with ethanol and then with warm
HPLC grade water. The resulting dark gray powder was dried
under flowing argon at 80-100 ꢀC overnight.
Figure 1. View of the clathrate I structure type down the c axis. The
framework atoms are silicon, and the shaded polyhedra are centered by K
atoms ofthe samecolor inthe 6d site. The K atominthe 2a siteisindicated
by red, and the red lines provide the outline of the corresponding
polyhedra (the K atoms at the corners of the unit cell have been omitted
for clarity).
Caution! Traces of K Si spontaneously ignite in air; the
4
4
products present in the tube have to cool down at RT under vacuum
or argon. After the tube is cold it should be disconnected from the
Schlenk and allowed to oxidize in air. Only after that the alumina
boat should be removed from the glass tube, the products ground
into a powder and ethanol addition done slowly. If the product
ignites, then the heat from this reaction is sufficient to convert
some of the potassium clathrate into silicon.
Na5.5(H ) Si , that encapsulates hydrogen at ambient
2 2.15
46
temperature and pressure from the reaction of Na Si with
4
4
10
NH Br. A clear advantage of inorganic clathrate com-
4
pounds such as Na5.5(H ) Si46 over their hydrate counter-
2 2.15
2. Characterization. Elemental Analysis. Elemental ana-
lysis was performed on three different reaction batches by
parts is their stability at ambient atmosphere.
Semiconducting clathrate materials are typically
synthesized by thermal decomposition of their parent
17
Desert Analytics Laboratories. HF dissolution was requested
for the third batch. Elemental analysis results for three batches
were as follows.
3
,5,7,11,12
Zintl salts
or from elements with the use of high
1
3
pressure techniques. Recently, clathrate formation has
been reported by reacting Na Si with HCl. NH X salts
1
4
batch
H
N
Br
K
Si
4
4
4
provide a convenient acid/oxidation reagent for metal
1
2
3
0.4%
0.21%
0.12%
1.9%
1.7%
0.1%
10%
10%
14.2%
62.6%
73%
71%
1
0,15,16
silicides.
To further explore the ammonium halide
as a reagent in the low temperature synthesis of clathrate
phases, we have prepared the K-containing Si clathrate.
Herein we report the synthesis and characterization of H2
encapsulated silicon type I clathrate: K8 (H ) Si .
0.27%
The variability in composition from sample to sample sug-
gests that the amount of hydrogen may be difficult to control.
The total composition is less than 100%, and this is attributed to
surface oxidation of the sample, as observed in the TEM, Si
NMR (described below), and FTIR (Supporting Information).
If the missing mass is attributed to oxygen and the correspond-
-x
2 y 46
2
9
Experimental Section
. Synthesis. All manipulations for the synthesis of the
potassium silicon clathrate were performed in a N filled glove-
1
2
ing Si for SiO is removed, along with assuming that Br is present
2
box or vacuum/inert atmosphere Schlenk line. Si (99.999%
Aldrich) and K (lump 99% Aldrich) were used as received.
K Si was prepared from K and Si. A 1:1 mixture of the two was
welded in a niobium tube, and the niobium tube was sealed in a
quartz jacket under a partial argon pressure. The quartz tube
was then placed in a furnace and heated with 1 ꢀC/min to 650 ꢀC
because of KBr that was not completely removed and the
corresponding K was removed, then we arrive at the follo-
wing K/Si/H ratios: 7.38:46:12.73, 5.07:46:4.5, and 8.0:46:2.63.
Averaging these leads to the formula K6.82(H ) Si . It is
possible that the surface oxidation prevents complete dissolu-
tion of the sample. It is highly likely that small changes
in pressure, temperature, and time provide slightly different
average compositions. Therefore, this phase will be referred to
4
4
2 3.3
46
(
10) Neiner, D.; Okamoto, N. L.; Condron, C. L.; Ramasse, Q. M.; Yu,
P.; Browning, N. D.; Kauzlarich, S. M. J. Am. Chem. Soc. 2007, 129, 13857–
3862.
11) McMillan, P. F.; Gryko, J.; Bull, C.; Arledge, R.; Kenyon, A. J.;
2 y
as having the general formula K8-x(H ) Si46 throughout the
1
manuscript. The microscopy was performed on samples 1 and 2.
The solid state NMR, FTIR, and X-ray powder diffraction
results were reproduced on all samples. The synchrotron and the
TG/DSC/MS data were from sample 3.
(
Cressey, B. A. J. Solid State Chem. 2005, 178, 937–949.
(
(
12) Sangster, J. J. Phase Equilib. Diffus. 2006, 27, 190–191.
13) Tanigaki, K.; Shimizu, T.; Itoh, K. M.; Teraoka, J.; Moritomo, Y.;
Yamanaka, S. Nat. Mater. 2003, 2, 653–655.
14) Bohme, B.; Guloy, A.; Tang, Z. J.; Schnelle, W.; Burkhardt, U.;
Baitinger, M.; Grin, Y. J. Am. Chem. Soc. 2007, 129, 5348.
EDX/STEM HAADF. Approximately 0.1 g of the synthe-
sized powder has been sonicated for 30 min in n-butanol and
spread over a gold-mesh grid covered by a thin perforated
(
(15) Meyer, G. Z. Anorg. Allg. Chem. 2008, 634, 201–222.
(16) Neiner, D.; Chiu, H. W.; Kauzlarich, S. M. J. Am. Chem. Soc. 2006,
1
28, 11016–11017.
(17) http://www.desertanalytics.com/index.htm.

Article
Inorganic Chemistry, Vol. 49, No. 3, 2010 817
carbon film. The grid was heated under dynamic vacuum at
2
energy dispersive X-ray spectroscopy (EDX) were performed
under a JEOL JEM-2500SE electron microscope operated at
of 0.002 deg. The peak profile was chosen to be pseudo-Voigt
(Howard asymmetry), and the background was fit to a fifth-
order polynomial. Atomic site occupancy and atomic displace-
ment parameters were modeled either by allowing parameters to
refine free of manipulation or by setting the model according to
the results provided by STEM. The refinement results are
provided as Supporting Information.
00 ꢀC overnight. Selected area electron diffraction (SAED) and
2
00 kV. The sample was also examined using a spherical aberra-
tion (Cs) corrected VG HB501 scanning transmission electron
microscope (STEM) operated at 100 kV. High-angle annular
dark-field (HAADF) images were taken with the convergence
semiangle of 20 mrad and collection semiangle of 70-210 mrad,
Differential Scanning Calorimetry (DSC)/Thermogravimetric
Analysis (TGA)/Mass Spectrometry (TG/MS). TGA/DSC were
performed on a TGA/DSC STA 409 Netzsch instrument by
heating the sample under argon. The 409 Netzsch TGA/DSC
instrument was coupled with an Aelos QMS 300 MS to identify
the evolved gases. The MS utilizes a standard electron impact
ionization detector. The transfer line between the TGA/DSC
and MS was maintained at 120 ꢀC for the duration of the
experiment. The sample was brought under inert atmosphere
and loaded in the TGA/DSC/MS instrument. The data were
obtained by heating the samples under Ar gas (10 mL/min) from
RT to 650 ꢀC with 20 K/min. These particular experimental
conditions are necessary to observe low concentrated elements
in a sample, for example, low inert gas flow rate and high heating
ramp rate. The MS channels observed were mass 2 (hydrogen),
17 (ammonia), 18 (water), 39 (potassium), 40 (potassium
hydride), 81 (hydrobromic acid). Only mass 2 and mass
81 channels provided an observable signal over the temperature
range measured. The signal for mass 81 occurred at about 50 ꢀC
higher than that for mass 2, centered at ∼450 ꢀC, and was
2 orders of magnitude weaker than the mass 2 signal under the
same MS experimental conditions.
˚
for an estimated probe size of 1.0 A, sufficient to resolve the 6d
1
and 2a sites of the clathrate in projection. The STEM image
8
1
9
simulations were performed using Kirkland’s code.
Solid State NMR. All Solid-state NMR experiments were
performed on a Bruker Avance 500 spectrometer equipped with
an 11.75 T magnet. The samples were loaded in Bruker zirconia
rotors with kel-F caps in a glovebox filled with dry flowing
nitrogen, and TMS was used as the external chemical shift
1
reference. For H MAS NMR, a Bruker 4 mm CP-MAS probe
was used with a MAS rate of 15 kHz. A Hahn-echo pulse
sequence (90-t-180-acquisition) was used to suppress the signal
from the probe background; the 90 degree pulse width was
2
.5 μs; the interpulse delay time (t) was synchronized with rotor
time; the spectrum width was 500 kHz; the relaxation delay was
s; a total of 2000 transients was averaged; a block size of 8192
was used and zero-filled to 16384 with 1 Hz line broadening.
5
29
For Si MAS NMR the sample was loaded in a 4 mm
zirconia rotor and a 4 mm Bruker MAS probe was used with a
spinning rate of 10 kHz. A Hahn-echo pulse sequence was
utilized; the 90 degree pulse length was 2.7 μs, the interpulse
delay was synchronized with the rotor time; the spectrum
width was 500 kHz; the recycle delay was 300 s; a total of
Results and Discussion
2
2
006 transients were averaged; the data were processed with
000 Hz line broadening. For Si{ H} CP MAS NMR experi-
2
9
1
A hydrogen-encapsulated potassium silicon clathrate has
been synthesized from the solid state reaction of the Zintl salt,
K Si , and NH Br. This reaction is heterogeneous and hence
4 4 4
ment, the contact time was 1 ms, the recycle delay time was 5 s,
the spectrum width was 250 kHz, the total number of scan
was 62206.
complex as it involves both solid and gaseous phases. As with
many reactions in the solid state, this reaction is an example
of an entropy driven process pushed forward by the evolution
of hydrogen, ammonia, and possibly other volatiles. More-
over, the imposed dynamic vacuum helps remove the gaseous
phases and hence greatly enhances the rate of the forward
reaction. The reaction with molar ratio K Si /NH Br 0.25:1
FTIR. FTIR data were obtained using a Shimadzu IR
Prestige 21 equipped with a diffusive reflectance accessory.
The powders were mixed with the KBr using a 1:100 molar
ratio dilution of the sample in the KBr matrix. The samples were
mixed with KBr in a nitrogen-filled drybox. The FTIR measure-
ments were performed in air.
4
4
4
Powder X-ray Diffraction Data. The powder X-ray diffrac-
tion patterns were obtained on a PAD V Scintag instrument (Cu
at 300 ꢀC produces a mixture of potassium silicon clathrate,
crystalline silicon, and potassium bromide. However, when
the molar ratio K Si /NH Br is changed to 0.25:0.8, the
˚
KR radiation λ = 1.5418 A) equipped with a graphite mono-
chromator. Data were collected in a step scan mode between 20ꢀ
and 100ꢀ 2θ, with a step size of 0.02ꢀ. Structural refinement was
done by the Rietveld method with the GSAS package of
4
4
4
diamond-structure silicon byproduct is avoided, allowing
the formation of pure phase type I potassium silicon clathrate
and potassium bromide. It is possible that in the case of the
20
programs. The refined parameters include background, peak
shape, cell, atom positions, scale factor, atomic displacement
slight deficiency of NH Br that some of the potassium leaves
4
p
parameters and occupancies. The R-factor (R ), the weighted
as K metal, as the reaction is under dynamic vacuum. A
general reaction scheme is presented below:
2
R-factor (wR ), and the goodness of fit, χ , are defined as
p
P
P
P
P
2
follows: R
p
=
[yio - yic]/
y
io, wR
p
= [
w
i
(yio - yic) /
P
2
1/2
2
2
w
wiyio
i
(yio) ] , and χ = [wR
p
/Rexp] , where Rexp = [(N - P)/
and y and yic are the observed and the calculated
1
1:5K Si þ ð38 þ xÞNH Br f K ðH Þ Si þ
4
4
4
8 -x
2
y
46
2
1/2
]
intensities, w is the weighting factor, N is the total number of
io
ð38 þ xÞKBr þ ð38 þ xÞNH þ ð19 -y þ x=2ÞH
ð1Þ
i
3
2
observed intensities when the background is refined, and P is the
number of adjusted parameters.
High-resolution powder X-ray diffraction experiments were
performed at Stanford Synchrotron Radiation Lightsource
Elemental analysis supports the reaction (1) pathway
indicating that hydrogen is present in all samples, and the
general formula K8 (H ) Si will be used throughout. In
-x
2 y 46
(
SSRL) on Beamline 2-1. Data were collected in reflection
the elemental analysis data sets a small amount of Br was
geometry using X-ray energy of 13.5 keV. A Ta-doped Si(111)
crystal analyzer was used to define the diffraction beam.
Data were collected between 10 and 42 deg 2θ with a step size
detected. Silicon clathrates encapsulating Cl, Br, and I
5,21,22
have been prepared;
however, the small amount of Br
(
18) Krivanek, O. L.; Dellby, N.; Lupini, A. R. Ultramicroscopy 1999,
8, 1.
19) Advanced Computing in Electron Microscopy; Kirkland, E. J., Ed.;
Plenum Press, NY, 1998.
20) http://www.ccp14.ac.uk/solution/gsas/
(21) Kovnir, K. A.; Abramchuk, N. S.; Zaikina, J. V.; Baitinger, M.;
Burkhardt, U.; Schnelle, W.; Olenev, A. V.; Lebedev, O. I.; Van Tendeloo,
G.; Dikarev, E. V.; Shevelkov, A. V. Z. Kristallogr. 2006, 221, 527–532.
(22) Yakimchuk, A. V.; Zaikina, J. V.; Reshetova, L. N.; Ryabova, L. I.;
Khokhlov, D. R.; Shevelkov, A. V. Low Temp. Phys. 2007, 33, 276–279.
7
(
(

8
18 Inorganic Chemistry, Vol. 49, No. 3, 2010
Neiner et al.
Figure 2. (a) Secondary electron image of a K8-x(H
particle. (b) SAED pattern taken along the [100] direction. (c) EDX
spectrum of a K8-x(H Si46 clathrate particle.
2 y
) Si46 clathrate
2 y
)
observed in the elemental analysis suggests that the Br is
probably a result of KBr that was not completely removed by
washing, and the quantity is below the detection limit of
X-ray powder diffraction. The total weights were consistently
lower than 100%, with the highest totals for the elemental
analysis observed when HF was employed to dissolve the
solid. This suggests that there is an oxide coating on
the surface of the powder. These inferences concerning the
identity of the Br and the presence of a thin oxide surface are
consistent with the microscopy, NMR, and FTIR results, as
described below.
Figures 2(a) and 2(b) show a secondary electron image and
a SAED pattern of a typical clathrate particle obtained from
reaction 1. The SAED patterns observed for the majority of
particles examined corresponded to the type I clathrate
structure, although an electron diffraction pattern consistent
with an amorphous structure was observed for a small
number of particles. Figure 2(c) shows an EDX spectrum
obtained from the clathrate particle shown in Figure 2(a). No
difference was observed between the spectra obtained for
shorter and longer acquisition times, indicating that the
electron beam did not damage the sample during data
acquisition. Only peaks corresponding to Si, K, O, and Au
are observed. The Au signature peaks are attributed to the
gold-mesh grid supporting the particles. The O peak can be
attributed to a thin amorphous layer covering the clathrate
particles because a higher O concentration was obtained at a
very thin area of the particle. The K content obtained from
EDX quantification for eight different particles is 5.5 ( 0.3
per formula unit, when the Si sites are assumed to be fully
occupied (K/Si=5.5:46). Further characterization was car-
ried out for this material by HAADF-STEM to determine
the K occupancies in the 2a and 6d sites.
Figure 3. (a) Raw and (b) low-pass filtered HAADF-STEM images
of the K8 (H ) Si clathrate compound taken along the [100] direction.
-x
2 y 46
Figure 4(a) shows an experimental image averaged by
using 2dx software and an intensity profile along the yellow
23
line in the image. The image size is one and a half unit cells
to allow both the 2a and 6d sites to be visible in a single image.
An intensity peak is observed at the 6d site, whereas a dip is
seen at the 2a site. Figures 4(b)-(e) show calculated images
and intensity profiles with different K occupancies in the
2a and 6d sites. The intensities at the 2a and 6d sites decrease
with the decreasing K occupancies: the intensity of HAADF-
th
n
STEM images is indeed proportional to the n power Z (n=
.5-2) of the average atomic number Z of the material being
observed, and they are therefore often referred to as
Figure 3 shows raw and low-pass filtered HAADF-STEM
images of the potassium silicon clathrate taken along the
1
[
100] direction. Since the guest atoms are considered to be
24,25
Z-contrast images.
The best way to estimate occupancies
5
weakly bonded to the cage framework, we anticipated the K
guest atoms would be easily damaged by the intense electron
probe in the aberration corrected STEM. However, contrary
to our expectations, the compound was quite stable, and no
beam damage was observed. The ring pattern and the center
of the ring correspond to the Si24 cage and the 6d sites,
respectively (the Si24 cage with a guest atom is overlaid in
Figure 3(b)).
from STEM-HAADF is to compare the integrated intensities
at the sites. The peak width at the 6d site in Figure 4(a)
[experiment] is broader than that in Figure 4(d) [6d=0.8].
(23) Renault, L.; Chou, H. T.; Chiu, P. L.; Hill, R. M.; Zeng, X. Y.;
Gipson, B.; Zhang, Z. Y.; Cheng, A. C.; Unger, V.; Stahlberg, H. J. Comp.-
Aided Mol. Des. 2006, 20, 519–527.
(
24) Jesson, D.; Pennycook, S. J. Proc. R. Soc. London 1995, A 449, 273.
(25) Hatrel, P.; Rose, H.; Dinges, C. Ultramicroscopy 1996, 63, 93.

Article
Inorganic Chemistry, Vol. 49, No. 3, 2010 819
Figure 4. (a) Experimental HAADF-STEM image averaged by using the 2dx software and intensity profile along the yellow line indicated in the image.
(b)-(e) Simulated images with different potassium occupancies at the 2a and 6d guest sites and their intensity profiles.
1
Figure 5. Magic angle spinning H solid state NMR spectrum for
2
9
K8-x(H
2
)
y
Si46
.
Figure 6. Magic angle spinning Si solid state NMR spectrum for
Si46
K
2
8-x(H )
y
.
The broadening comes from thermal vibration of atoms and
sample vibration during the observation. When integrated
intensities were compared, the occupancy is larger: the 6d site
is 0.9 and the 2a site is 0.6. Thus, the overall K content
obtained with the HAADF-STEM image is approximately
Na5.5(H ) Si . The broad resonance observed in Figure 5
2 2.15 46
at 5.26 ppm is consistent with H that has restricted motion
2
and may be interacting with both cages. The assignment of
these peaks to H molecules rather than H covalently bound
2
to Si is consistent with the absence of any Si-H stretches in
6
.6 per unit formula, which is slightly larger than the EDX
29
1
the FTIR and the lack of a signal in Si{ H} CP MAS NMR.
result (5.5). The K content from HAADF-STEM image
provides the analysis for one particle, whereas the EDX is
for a larger sample size; however, the errors should be less in
the HAADF-STEM image.
It is possible that the resonance at 5.26 ppm observed in the
1
H NMR is a result of H in both cages in varying amounts,
2
and hence moved to lower fields compared with that ob-
1
1
served for the Na5.5(H ) Si46 H spectrum. This hypothesis
H MAS NMR spectrum is presented in Figure 5. There
2 2.15
would also be consistent with the HAADF-STEM analysis
presented herein indicating that both 2a and 6d sites are
are five distinct resonances in the proton NMR spectrum at
5
.26 ppm, 1.8 ppm, 1.4 ppm, 0.94 ppm, and 0.05 ppm. The
partially occupied in the H encapsulated potassium clathrate
data show that hydrogen is present in this material; however,
the structure of the spectrum differs from that reported for
2
while only the 6d sites are partially occupied in the sodium
10
1
0
compound.
Na5.5(H ) Si . In the case of Na5.5(H ) Si461clathrate,
2 2.15
46
2 2.15
29
Si MAS NMR is presented in Figure 6 and shows four
there were also five distinct resonances in the H NMR
spectrum at 4.2 ppm, 3.8 ppm 1.3 ppm, 0.79 ppm and 0.08
ppm, with the largest resonance at 4.2 ppm. The resonance
resonances at 1274, 886, 509, and -74 ppm; three are
attributed to the clathrate framework and one at -74 ppm
is attributed to oxide. The chemical shifts for the three
resonances are similar to those published for K7.66Si46 pre-
paredby vacuumdecomposition of thebinary silicide, KSi, at
10
at 4.2 ppm was attributed to H encapsulation in the large
2
Si24 cage, based on evaluation of similar assignments for a
2
6
type II THF H O clathrate. There is a downfield shift of
3
2
27
elevated temperatures (1384, 940, and 556 ppm), but shifted
the hydrogen resonance in K (H ) Si as compared to
7
2 3 46
(
26) Florusse, L. J.; Peters, C. J.; Schoonman, J.; Hester, K. C.; Koh,
(27) Ferguson, M. J.; Moudrakovski, I. L.; Ratcliffe, C. I.; Tse, J. S. Mat.
Res. Soc. Symp. 2002, 691, G14.4.1–G 14.4.6.
C. A.; Dec, S. F.; Marsh, K. N.; Sloan, E. D. Science 2004, 306, 46C9–471.

8
20 Inorganic Chemistry, Vol. 49, No. 3, 2010
Neiner et al.
to higher shielding field and broader. In reference 27, it was
noted that the resonance at 940 ppm showed a distinct
anisotropy and therefore was best assigned to the Si in the
6c site. The higher shielding field and broader resonances for
K8 (H ) Si may be attributed to relative K depletion and
-x
2 y 46
local disordering in our sample compared with the stoichio-
metric samples prepared by vacuum decomposition.
Similar to previously published assignments for K7.66Si48,
the broad feature at 509 ppm has been assigned to the 16i Si
site, and the signals at about 1274 and 886 ppm are for Si in
2
7
2
4k and 6c sites. The broad feature of the signal for Si 24k
compared with Si 16i indicates that K 6d sites are more
disordered than K 2a. Therefore the composition of the guest
species inside this cage is more variable. This result is
consistent with the relatively larger displacement parameter
in XRD (discussed below) and HAADF-STEM data. Our
˚
2
9
Figure 7. Rietveld profile fit for K8-x(H
2
)
y
Si46 (λ = 1.5418 A). Calcu-
Si MAS NMR spectrum shows different intensities for
lated (solid green line), experimental (red crosses) and the difference
lower magenta line). The data were refined for the space group Pm3n.
these resonances compared with reference 27. While there is
no detailed information about the previously published
NMR experiments, we speculate that the difference could
be due to the following reasons: first, the recycle delay might
not be long enough; with 300 s recycle delay, it might saturate
(
analysis on the solid sample also indicated that there is no
Si-H stretch present, consistent with the proposed encapsu-
lation of H rather than a hydride. The FTIR data are
2
provided in the Supporting Information.
some signals with long T relaxation times. Second, the
1
Laboratory and synchrotron X-ray data were collected on
K8 (H ) Si obtained in this study. The synchrotron data
confirms the clathrate structure and absence of significant
amounts of amorphous SiO (Supporting Information). The
refined lattice parameter obtained was a =10.266(2) A. The
X-ray powder diffraction data along with the Rietveld refine-
ment are presented in Figure 7. The diffraction peaks were
indexed to a cubic cell, space group Pm3n, a=10.2932(3) A,
slightly larger than that obtained from the synchrotron data
˚
refinement and reported for K Si (a = 10.27518(5) A),
8 46
prepared via high temperature decomposition of K Si .
Atomic site occupancies were modeled according to
the occupancy obtained by HAADF-STEM. In this case
the refinement did not improve beyond a χ ∼ 5, and the
Hahn-echo pulse sequence used in our experiment might
cause the loss of some broad signals. Although the relative
intensities of our NMR data are different than those in the
literature, the peak positions are consistent with the pub-
lished assignments and provide strong evidence about the
-x
2 y 46
2
˚
27
29
presence of these cages in clathrate structure. The Si
NMR spectrum is quite different for the potassium clathrate
as compared with the sodium clathrate, suggesting the con-
duction electron density is higher in this Si framework as
˚
10,27
compared to the Na analogue.
The Na atoms in the cages
7
4
4
of clathrate are considered nearly neutral, with a small
amount of electron transfer to the framework, and the 3s
electrons cause the strong Knight shift observed for the Si in
2
28
the framework structure. K has more electrons and there-
fore is more metallic than sodium; the depleted K atoms in
the 6d and 2a cages would cause a stronger Knight shift than
the Na analogue. This is consistent with what is observed
for the rubidium clathrate where the positions and line widths
difference map indicated discrepancies between the calcu-
lated and the observed powder patterns. Three approaches
were taken to refine the model. In all these approaches the Si
and K 2a sites were fully refined, for example, atomic
positions, site occupancies, and displacements. In one ap-
proach the atomic displacement of K in the 6d site was fixed
to be similar to all the other atoms and the occupancy of the
site was refined. This produced an improved fit of the
experimental with the observed powder patterns; the occu-
pancy of the 6d site decreases, but the refinement was not
stable and did not converge. Moreover, opening the displace-
ment flag for the K 6d site caused the displacement parameter
to increase to 3.0, double those for the other atoms. The
second approach was to consider an anisotropic displace-
ment on the K 6d site. It is known that anisotropic displace-
ments are not adequate with X-ray powder diffraction data,
however; because this has been employed with K Si , a
similar approach was explored here. It was found that the
displacement for this site did not improve beyond U =2.2
and U = U = 3.4, while U = U = U = 0. A third
approach was to move K from the center of the Si24 cage, the
d site, to an off-center position in the 24k site (0.24, 0.51, 0).
29
of the Si NMR signals are even more different than the
27
sodium clathrate. Introduction of Ba in sodium clathrate
29
also causes broadening, shifts, and intensity changes in Si
29
NMR, providing further support for our spectral assign-
ments.
The resonance at -74 ppm is very close to -100 ppm
which is the normal Si-O resonance, and it may be assigned
28-30
to oxidation of the surface.
This observation is further
supported by our EDX data, in which oxygen was detected;
however, amorphous silica has not been observed by X-ray
diffraction and thus it is most likely present as a thin oxide
coating rather than amorphous SiO2 (see Figure 7 and
synchrotron data in Supporting Information).
31
8
46
2
9
1
Si{ H} CP MAS NMR spectrum showed no resonance
29
1
11
consistent with the absence of a Si-H bond. The Si{ H}
spectrum is presented in the Supporting Information. FTIR
22
33
12
13
23
6
(
28) Gryko, J.; McMillan, P. F.; Sankey, O. Phys. Rev. B 1996, 54(5),
037–3039.
29) Shimizu, F.; Maniwe, Y.; Kume, K.; Kawaji, H.; Yamanaka, S.;
Ishikawa, M. Synth. Met. 1997, 86, 2141–2142.
30) Onischuk, A. A.; Strunin, V. P.; Samoilova, R. I.; Nosov, A. V.;
Ushakova, M. A.; Panfilov, V. N. J. Aerosol Sci. 1997, 28, 1425–1441.
This has been proposed before by Tse to model the changes in
3
(
(
(31) McCusker, L. B.; Von Dreele, R. B.; Cox, D. E.; Louer, D.; Scardi, P.
J. Appl. Crystallogr. 1999, 32, 36–50.

Article
Inorganic Chemistry, Vol. 49, No. 3, 2010 821
Table 1. Crystallographic Refinement Data for the K8-x(H
Phase, Space Group Pm3n
2
)
y
Si46 Clathrate
a
1
00*Uiso
˚
(A)
atom Wyckoff
x
y
z
occupancy
K
K
Si
Si
Si
2a
6d
0
0
0.5
0
0
0.834(6)
0.965(4)
1.6(0)
3.0(6)
1.5(3)
1.4(1)
1.3(2)
0.25
0
24k
16i
6c
0.3067(8) 0.1183(8)
1
1
1
0.1842(6) 0.1842(6) 0.1842(6)
0.25 0.5
0
a
˚
Lab X-ray source, 8.75 keV. a = 10.2932(3) A, R
p
= 3.34%, Rwp =
2
4
.49%, χ = 2.933.
3
2
the K Si crystal structure at high pressures. In doing so,
8
46
the isotropic displacement improved, 2.0, and the occupancy
was 0.25 for the K in a 24k site, which gave the overall
32
2 y
Figure 8. TGA/DSC/MS data for K8-x(H ) Si46. Traces are as indi-
cated onthe figure. TGcurve is black, DSCcurve isred, andhydrogenMS
signal is shown in blue.
composition K7.6Si . The refinement presented in Figure 7
46
and the cell parameter and atomic positions presented in
Table 1 are provided for the model where all sites refine
freely, and the occupancies for the K were allowed to vary
while those of Si were fixed. This corresponds to the stoichio-
results indicate that the sample is not homogeneous, that is, K
content not only varies from one unit cell to another, but it is
possible that some parts of the sample are K-rich while others
are K-deficient. As STEM is not dependent upon the average,
this provides a local composition and structure information.
This variability is probably induced through the processing, as
metry, K7.5Si , consistent with the refinement of the struc-
46
ture of potassium silicon clathrate prepared by other
7
methods. It is known that the occupancy, displacement,
and the scale factor are highly correlated, and can be very
31
difficult to refine with X-ray powder diffraction data.
H may be encapsulated as a gas and K is removed from the
2
Theoretical calculations for an empty Si46 framework
indicate that the minimum energy configuration is obtained
silicide as the salt, KBr, with concomitant clathrate formation.
X-ray diffraction not only analyzes a larger sample size but also
provides the average K and Si occupancies. Chemical analysis
for the three samples measured also suggests variability in
stoichiometry, and further efforts at controlling pressure may
result in a more well-defined sample composition.
33
˚
for a = 10.355 A, and K7.6Si46 prepared by vacuum de-
composition of K Si has a smaller lattice parameter of
4
4
12
˚
0.2751 A. The samples presented herein show lattice
1
parameters slightly larger to slightly smaller than that re-
ported for K7.6Si . This suggests that encapsulation with K,
46
The TGA/DSC/MS data are presented in Figure 8. The
TGA data show a weight loss of ∼1.1% that occurs
above 100 ꢀC; the DSC data show a very broad exotherm
at ∼350 ꢀC, and a broad and weak MS signal, hydrogen
channel, at ∼400 ꢀC. K or KH are not observed in the MS, in
agreement with the reported high thermal stability of potas-
sium type I silicon clathrate. The sample did not decompose
during heating under these conditions. Potassium clathrate,
K Si , is reported to be stable at higher temperatures than its
or any other guest provides some interaction with the frame-
work. There seems to be variation in exact stoichiometry in
the sample. It is possible that while all the sample is type I
clathrate, there is inhomogeneity that leads to different lattice
constants. The K atom, refined in the 6d site, had a large
displacement parameter, almost double the other atoms in the
structure. This large displacement parameter could indicate
significant variability in composition for this site. If H shares
2
8
46
this site, this may also contribute to the large displacement
parameter. These factors make it difficult, without very high
resolution data, to obtain accurate occupancy for the K on the
32
sodium analogue. It has been experimentally seen by us and
others that sodium type I clathrate decomposes to diamond-
structured silicon upon heating at 20 K/min to 650 ꢀC,
potassium type I clathrate maintains its cage-like structure
under the same experimental conditions. An X-ray diffraction
pattern of samples after the above-mentioned heat treatment
shows crystalline Si in the case of sodium clathrate and leaves
the potassium silicon clathrate almost unchanged.
6d site by refinement of X-ray powder diffraction data.
While elemental analysis, EDX, and HAADF-STEM
suggest a lower K content than the X-ray refinement where
the average K occupancy results in a stoichiometry closer to
7
.5, although the larger lattice parameter and displacement
parameter for K in the 6d site may indicate that the powder
X-ray data is not of high enough quality to provide a
definitive model of the guest site and an accurate stoichio-
metry. One cause of the discrepancy observed between the
elemental analysis, EDX, and HAADF-STEM on one hand
and the X-ray data on the other is the respective sample sizes
used by these techniques. While the elemental analysis and
microscopy used 10-20 mg, our X-ray experiment used
Summary
The presence of hydrogen is provided by elemental ana-
1
29
1
lysis, H MAS NMR, and TG/MS. FTIR and Si{ H} CP
MAS NMR show no evidence for Si-H bonds, suggesting
that the hydrogen is molecular in nature, and not bonded to
the silicon framework. These results are consistent with a
model of H entrapped in the guest site. HAADF-STEM
2
0
.5-1 g of the same sample and therefore should provide a
provides detailed structural information as to the framework
structure and guest site occupancy suggesting that the solid is
inhomogeneous and that at least some of the clathrate is
deficient in K. This model is further supported by Si MAS
NMR. Rietveld refinement of the X-ray diffraction data
more accurate average for the sample. Moreover, the STEM
(
32) Tse, J. S.; Desgreniers, S.; Li, Z. Q.; Ferguson, M. R.; Kawazoe, Y.
Phys. Rev. Lett. 2002, 89(19), 195507-1–195507-4.
33) Adams, G. B.; Okeeffe, M.; Demkov, A. A.; Sankey, O. F.; Huang,
Y. M. Phys. Rev. B 1994, 49, 8048–8053.
29
(
provides an averaged structure where K in the 6d site has a

8
22 Inorganic Chemistry, Vol. 49, No. 3, 2010
Neiner et al.
doubled displacement parameter with respect to the other
atoms in the refinement, suggesting variability in compo-
tract No. DE-FG02-03ER46057, and by the Japan
Society for the Promotion of Science for Young Scienti-
sts. Part of this work was carried out at the National
Center for Electron Microscopy, under the auspices of the
U.S. Department of Energy under contract number DE-
AC02-05CH11231. Portions of this research were carried
out at the Stanford Synchrotron Radiation Lightsource,
a national user facility operated by Stanford University
on behalf of the U.S. Department of Energy, Office of
Basic Energy Sciences.
sition for the hydrogen encapsulation clathrate, K8
-
-x
(
H ) Si . Further investigation of the synthesis with ammo-
2 y 46
nium salt along with in situ measurements would provide
insight into the formation of this hydrogen encapsulated
inorganic clathrate.
Acknowledgment. The authors gratefully acknowledge
Dr. Alexandra Navrotsky for use of the powder diffra-
ctometer. We wish to thank David G. Morgan, who
carried out the image processing using 2dx software. This
work was supported by the Center of Excellence for
Chemical Hydrides under Contract No. DE-FC36-
29
1
Supporting Information Available: Spectra of the Si{ H} CP
MAS NMR and FTIR, Rietveld refinement of the synchrotron
data. This material is available free of charge via the Internet at
http://pubs.acs.org.
0
5GO15055, by the Department of Energy under Con-