Electron microscopy, spectroscopy, and first-principles calculations of Cs2O

Article abstract of DOI:10.1016/j.jssc.2004.09.022

Oxides of cesium play a key role in ameliorating the photoelectron emission of various opto-electronic devices. However, due to their extreme reactivity, their electronic and optical properties have hardly been touched upon. With the objective of better understanding the electronic and optical properties of Cs₂O in relationship to its structure, an experimental and theoretical study of this compound was undertaken. First-principles density functional theory calculations were performed. The preferred structural motif for this compound was found to be anti-CdCl₂. Here three Cs-O-Cs molecular layers are stacked together through relatively weak van-der-Waals forces. The energy bands were also calculated. The lowest transition at 1.45 eV, was found to be between the K point in the valence band to the Γ point in the conduction band. A direct transition at 2 eV was found in the center (Γ) of the Brillouin zone. X-ray powder diffraction, transmission electron microscopy and selected area electron diffraction were used to analyze the synthesized material. These measurements showed good agreement with the calculated structure of this compound. Absorption measurements at 4.2 K indicated two optical transitions with somewhat higher energy (indirect one at 1.65 and a direct transition at 2.2 eV, respectively). Photoluminescence measurements also showed similar transitions, suggesting that the lower indirect transition is enhanced by three nearby minima at 1.5 eV in the Brillouin zone.

Full text of DOI:10.1016/j.jssc.2004.09.022

ARTICLE IN PRESS
Journal of Solid State Chemistry 178 (2005) 1190–1196
www.elsevier.com/locate/jssc
Electron microscopy, spectroscopy, and first-principles
calculations of Cs2O
Ã
S. Gemminga, G. Seiferta, C. Muhleb, M. Jansenb, A. Albu-Yaronc, T. Aradc, R. Tennec,
a
+
Institut fur Physikalische Chemie, Technische Universita¨t, Dresden, 01062 Dresden, Germany
bMax-Planck-Institut fu¨r Festko¨rperforschung, Heisenbergstrasse 1, 70569 Stuttgart, Germany
cDepartment of Materials and Interfaces, The Weizmann Institute of Science, Faculty of Chemistry, PO Box 26, 76100 Rehovot, Israel
Received 27 June 2004; received in revised form 5 September 2004; accepted 10 September 2004
Abstract
Oxides of cesium play a key role in ameliorating the photoelectron emission of various opto-electronic devices. However, due to
their extreme reactivity, their electronic and optical properties have hardly been touched upon. With the objective of better
understanding the electronic and optical properties of Cs2O in relationshipto its structure, an experimental and theoretical study of
this compound was undertaken. First-principles density functional theory calculations were performed. The preferred structural
motif for this compound was found to be anti-CdCl2. Here three Cs–O–Cs molecular layers are stacked together through relatively
weak van-der-Waals forces. The energy bands were also calculated. The lowest transition at 1.45 eV, was found to be between the K
point in the valence band to the G point in the conduction band. A direct transition at 2 eV was found in the center (G) of the
Brillouin zone. X-ray powder diffraction, transmission electron microscopy and selected area electron diffraction were used to
analyze the synthesized material. These measurements showed good agreement with the calculated structure of this compound.
Absorption measurements at 4.2 K indicated two optical transitions with somewhat higher energy (indirect one at 1.65 and a direct
transition at 2.2 eV, respectively). Photoluminescence measurements also showed similar transitions, suggesting that the lower
indirect transition is enhanced by three nearby minima at 1.5 eV in the Brillouin zone.
r 2004 Elsevier Inc. All rights reserved.
Keywords: Cesium oxide
1. Introduction
study of these materials very difficult. In particular, the
study of cesium oxides using transmission electron
microscopy (TEM) and its auxiliary techniques, like
electron-diffraction (ED) and energy dispersive X-ray
spectroscopy (EDS), requires a special means to transfer
the sample safely, which was not possible thus far.
The synthesis of pure cesium oxides was studied in the
early days of the 20th century [1–3]. Later studies
revealed the crystal structures of these oxides [4,5]. The
complex cesium–oxygen phase diagram consists of an
unusually large number of oxides among them Cs7O,
Cs11O3 [6], the stoichiometric Cs2O, and also Cs2O2,
Cs2O3 [5], CsO2 [7–9], and CsO3 [7] where the oxygen
can be present in the form of molecular ions such as
superoxide OÀ2 or peroxide O22À. Cesium oxide (Cs2O)
crystallizes in the 3R anti-CdCl2 structure. The unit cell
Oxides of alkali metals and particularly those of
cesium play an important role in reducing the work-
function and thus enhancing the electrical current of
photo-cathodes, and in promoting catalytic reactions
and oxidation enhancement of various semiconductor
surfaces. These technological aspects make the study of
cesium oxides of special importance. However, the
oxides of cesium are extremely sensitive to the ambient
atmosphere, converting into white cesium carbonate
upon exposure. This obstacle makes a comprehensive
Ã
Corresponding author. Fax: +972 8 934 4137.
E-mail addresses: gotthard.seifert@chemie.tu-dresden.de
(S. Gemming), reshef.tenne@weizmann.ac.il (R. Tenne).
0022-4596/$ - see front matter r 2004 Elsevier Inc. All rights reserved.
doi:10.1016/j.jssc.2004.09.022

ARTICLE IN PRESS
S. Gemming et al. / Journal of Solid State Chemistry 178 (2005) 1190–1196
1191
consists of three Cs2O molecular layers with rhombohe-
¯
dral arrangement (R3M) and unit cell parameters:
(1 Ha=24.2 eV) and only minor changes occur beyond
EPcuWt =100 Ha.
a=b=4.256 A; c=18.990 A. In each molecular sheet
the layer of oxygen atoms is sandwiched between two
layers of cesium atoms in octahedral arrangement [4a,7].
The present work deals with the structure of Cs2O,
which was investigated through both theoretical calcu-
lations and experimental means.
For an investigation of the structural properties and
stacking fault energies the Hellman-Feynman forces on
the atoms and the stress tensor components must also
converge with respect to EPcuWt . While the forces are
already converged for EPcuWt =64 Ha, the stress tensor
components vanish only for EPcuWt =121 Ha or better.
Thus the electronic properties are analysed with
EPcuWt =64 Ha, whereas EcPuWt =121 Ha is employed for
the structural and elastic properties. The k-point
sampling was 10 Â 10 Â 3 for the 3R anti-CdCl2
(layered) structure with three formula units in the
periodically repeated unit cell and 10 Â 10 Â 10 for the
simple 1H stacking sequence with one formula unit per
unit cell.
The energy gaps (Eg) of the various cesium oxides and
their electron affinities (EA) are not accurately known.
Reasonable estimates from light absorption and photo-
emission measurements for Cs2O are: Eg=2 eV and
EA=0.8 eV [10]. The measured work function for
mixtures of cesium oxides was found to be as low as
1 eV (ꢀ1200 nm) [9]. Many studies were performed in
order to improve the properties of cesium oxide films as
coatings for photocathode front surfaces. Aging of the
photocathodes is often related to degradation of the
cesium oxide coating. It was found that the highest
photo-emissivity and greatest stability of photocathodes
is obtained with a cesium to oxygen ratio very close to
2:1 [10–12]. However, this ratio does not necessarily
indicate that merely Cs2O is present in the film.
Therefore, the study of the electronic and optical
properties of Cs2O in relationshipwith its structure is
highly warranted.
In the present study, a combination of optical
measurements and first-principles density functional
theory (DFT) calculations was used to study the
electronic structure of Cs2O in relationshipto its
structure. High-resolution transmission electron micro-
scopy (HRTEM) imaging was attempted to reveal the
previously hidden aspect of the microstructure on a
range of scales extending down to atomic dimensions.
Absorption and photoluminescence measurements at
room and cryogenic temperatures were carried out.
Satisfactory agreement was obtained between the
calculated and experimental observations.
2.2. Synthesis and measurements
Cesium oxide was prepared from the elements.
Cesium metal was produced from CsCl (Fluka, p.a.
99%) and Ca (Sigma-Aldrich, 99%) by a reaction under
vacuum at T=800 1C and was distilled twice [13,14].
Then, liquid cesium was reacted with a substoichio-
metric amount of dry oxygen [15]. The raw product was
heated at T=180 1C for 3 days under argon (p=1 atm)
and was subsequently ground in a stream of dry argon.
This cycle was repeated three times. Finally, the excess
cesium was removed by sublimation in vacuum at
T=200 1C. The as-obtained air-sensitive product has a
clear orange color and consists of a single phase.
Optical-absorption and luminescence measurements
were carried out with sealed ampoules having flat front
surfaces. A double-grating monochromator (HRD,
Jobin-Yvon) with a standard tungsten lampwas used
along with the photomultiplier (RG942, Hamamatsu).
Both absorption and photoluminescence-PL (excitation
by Ar ion laser at 514.5 nm) were measured in the range
of 500–900 nm at both cryogenic and room tempera-
tures. A special holder was constructed to allow
insertion of the sealed quartz ampoule containing the
material into the cryostat. The X-ray powder diffraction
data of Cs2O were collected with a linear position-
sensitive detector on a Stoe Stadi-P diffractometer (Stoe
& Cie GmbH, Germany, Ge-monochromated MoKa1
radiation (wave length=0.7093 A), 2y range 4–401, step
0.011, 0.2 mm capillary diameter). The data has been
calibrated with respect to an external Si-standard. Le-
Bail fit has been done using JANA2000 program (V.
Petricek, M. Dusek, Jana2000, with the crystallographic
computing system: Institute of Physics, Prague, Czech
Republic, 2003) with 17 variables refined (background
has been modelled by a Legendre 10-terms polynom).
Imaging, electron diffraction (ED) and chemical analy-
sis by X-ray energy dispersive spectroscopy (EDS) were
performed with a Philips CM120 (120 kV) TEM,
2. Experimental
2.1. Computational details
Self-consistent density-functional band structure cal-
culations were carried out in the local density approx-
imation with the program package ABINIT. A plane-
wave basis set and norm-conserving pseudopotentials
are employed to represent the valence electron density
and the core-valence interaction. The semicore shell of
Cs was also treated explicitly, because the Cs 5s and 5p
bands are quite close to the Fermi level, thus an
interaction with the O levels has to be accounted for.
With these pseudopotentials converged densities of
PW
states are obtained for a kinetic energy cutoff E
cut
of the plane waves which is better than 64 Ha

ARTICLE IN PRESS
S. Gemming et al. / Journal of Solid State Chemistry 178 (2005) 1190–1196
1192
equipped with an EDS system (EDAX model Phoenix
microanalyzer). Fast Fourier transform (FFT) of the
high-resolution images was obtained by Digital Micro-
graph software (Gatan). Sample analysis without atmo-
spheric exposure was essential and techniques for
manipulation, transferring and mounting (onto gold
grids) the sample from the ampoule to the TEM under
inert conditions of anhydrous/anaerobic atmosphere
were developed.
ison with the interaction between adjacent triple layers.
In order to quantify this result, the simpler AbC AbC
stacked (1H) structure was also studied and was found
to be less stable than the 3R structure by 0.176 eV per
formula unit. This value corresponds to a very low
stacking fault energy between the 3R and the 1H
structures of only 0.16 J/m2. Therefore the interaction
between adjacent Cs–O–Cs triple layers can be char-
acterized as rather weak. In contrast, the intra-layer
stacking fault, which leads to a trigonally prismatic
arrangement of Cs ions around O has a considerably
higher energy of 2 J/m2. These stacking fault energies are
actually comparable with the values obtained earlier for
MoS2 [16].
3. Results and discussion
3.1. Theory and calculations
3.1.1. Structure
3.1.2. Electronic structure
The most stable optimized structure is depicted in
Fig. 1. It has the 3R anti-CdCl2 crystal structure, which
is characterized by an AbC BcA CaB stacking sequence
of Cs and O layers as indicated in Fig. 1. Thus, in
accordance with the experimental crystal structure data,
the O atoms are surrounded by six Cs neighbors in
octahedral arrangement. The lattice parameters of the
hexagonal unit cell were optimized and amount to
a=b=4.28 A and c=15.49 A. The calculated value of c
is lower than the value deduced from the TEM
experiment (see Section 3.2), which indicates that the
non-bonding interaction across the van-der-Waals gap
of the structures is overestimated as a consequence of
the local-density approximation. The spacing along the c
axis is dc(Cs–O)=2.48 A within a Cs–O–Cs triple layer
and dc(Cs–Cs)=2.69 A between adjacent Cs–O–Cs
triple layers. This finding is indicative of a stronger
binding interaction within the triple layers in compar-
The density of electronic states (DOS) is given in
Fig. 2 for the 3R structure (top) and for the 1H structure
(bottom). Four peaks in the occupied region of 3R
correspond to semicore and valence states, namely to the
Cs 5s states at À22.6 eV, O 2s at À16.7 eV, Cs 5p at
À9.3 eV, and the combined DOS of O 2p and Cs 6s at
À2.5 eV. The latter value compares well with UV
photoemission data of Cs2O, where a very narrow peak
appears at 2.6 eV below the Fermi level during the
oxidation of a Cs film [17]. For the 1H structure also
four peaks are obtained: Cs 5s at À22.6 eV, O 2s at
À15.9 eV, Cs 5p at À9.4 eV, and the combined DOS of
O 2p and Cs 6s at À1.7 eV. The subvalence 5s and 5p
levels of Cs are insensitive towards the particular inter-
layer arrangement, because these states are rather
localized at the Cs centers and do not participate in
the inter-layer interaction. O 2s, O 2p, and Cs 6s
experience shifts to lower binding energies when going
from the 3R structure to the less stable 1H stacking.
Fig. 2. Calculated density of states of Cs2O. 3R—top; 1H—bottom.
The DOS curves are broadened by Gaussian functions of 0.5 eV full
width at half maximum.
Fig. 1. Schematic illustration of the anti-CdCl2 (3R) structure of Cs2O.
Oxygen is in black and cesium in gray.

ARTICLE IN PRESS
S. Gemming et al. / Journal of Solid State Chemistry 178 (2005) 1190–1196
1193
This result indicates that the weak inter-layer interaction
is conveyed mainly by the outermost valence electrons.
For both structures an energy gapseparates occupied
and unoccupied states; in order to analyse the states
spanning this gap in more detail, the band structure of
the valence and conduction states were calculated for
both the 3R (Fig. 3) and the 1H (Fig. 4) structure. The
band structure for the 3R structure is depicted in Fig. 3
within an energy range from À12 to 6 eV. The k points
are labelled according to the nomenclature for hexago-
nal systems. In the 3R structure the Cs 5p levels exhibit
an overall splitting of 2.7 eV, whereas the valence states
are only split by at most 1 eV at the K and H points. This
low splitting reflects the weak ligand field of the rather
soft Cs cations coordinated to the O anion. In
accordance with the symmetry of the 3R stacking
sequence the valence levels exhibit no dispersion along
the G–A and M–L lines. The unoccupied states
approach the Fermi level at the G point with a gap of
2 eV to the uppermost occupied state at G. According to
a projection of the corresponding wave function into
atomic spheres this lowest unoccupied state is mainly
derived from the Cs 6s valence levels with only a low
admixture of O states. As the valence band maximum
occurs at the K points a smaller, indirect energy gap of
1.45 eV results. Slightly larger values of 1.5 eV are
obtained for indirect transitions from M, H, and L to
G. Hence, according to the present DFT calculations
Cs2O can be classified as a semiconductor with an
indirect gap.
In the 1H structure the semicore 5p levels of Cs
exhibit a maximum dispersion of about 2.7 eV at G, H,
and L, and a minimum of 0.5 eV at A (Fig. 4). The set of
mainly O 2p valence bands starts at À2.2 eV (G, A). As
for the 3R structure the energy gapat the G point
amounts to about 2 eV. For the 1H structure metal-
derived bands contribute to unoccupied states close to
the Fermi level especially at the G, K, and M points. On
the average the conduction band edge is located above
the valence band by 2 eV for kc=0.0 1/c and by 4 eV for
kc=0.5 1/c. Thus an indirect gapof 1.5 eV is obtained
for the transition from K (and M) to G, which is similar
to the electronic properties of the ground state 3R
phase. These values should be regarded as a lower
bound for the experimentally observed band gaps,
because DFT calculations intrinsically tend to under-
estimate the size of the band gap.
Fig. 5 gives the electron density distribution in a cut
plane which is spanned by the [11–20] and [0001]
directions. In this plane all three basis atoms are
comprised as indicated in the figure. The AbC AbC
AbC stacking sequence of the 1H structure (left panel)
and the CaB AbC BcA stacking of the 3R structure
(right panel) can be clearly distinguished. The valence
electron density is strongly delocalized with a peak
located at the O sites, and some smaller maxima within
the octahedral coordination sphere around the O site
and also in the interlayer region. As already concluded
on the basis of the DOS plots, the subvalence density
remains at the Cs sites, has only minor overlapwith the
valence density maximum at the O site, and does not
participate in a directed interaction across the interlayer
region.
Fig. 3. Band structure for the 3R structure of Cs2O.
Fig. 5. Electron density distribution in Cs2O with two different
structures. Left—1H; right—3R.
Fig. 4. Band structure of 1H structure of Cs2O.

ARTICLE IN PRESS
S. Gemming et al. / Journal of Solid State Chemistry 178 (2005) 1190–1196
1194
3.2. XRD analysis
Fig. 6 shows the experimentally determined X-ray
powder diffraction of the powder and the fitted pattern
according to 3R-Cs2O (PDF #09-0104) file. This
analysis clearly shows that the probed material is the
layered Cs2O with 3R-anti CdCl2 structure. The
calculated cell parameters are: a=4.2514(1) A,
c=18.984(1), which agree very well with the published
data: a=4.256(4) A, c=18.990(2) A [4a].
3.3. TEM analysis
TEM analysis of the Cs2O specimens was particularly
difficult, because of the bending and/or of the random
movement of the crystals (attributed to electron beam
induced charging), during examination. Thus, in prac-
tice, it was quite difficult to tilt a crystal to the exact
zone axis, because of its drift and continuous change of
orientation relative to the electron beam. Fig. 7 shows
the (HRTEM) image of a Cs2O crystallite taken from
the Cs2O powder sample (a), the Fourier transform
(FFT) (inset b), and the intensity profile from the black
framed region in a (inset c). The HRTEM image clearly
reveals the intrinsic lamellar structure of the material.
The distance between the layers (6.32 A) matches the
Fourier Transform (FFT) results and the line profile
results (6.38 A) for the (0003) points, a typical value for
c/3 in Cs2O. Fig. 8 is a typical selected-area electron
diffraction (SAED) pattern of the Cs2O sample. The
SAED pattern exhibits rings made of the superposition
of spotted rings on fairly sharp and continuous rings,
indicative of both small particles as well as larger ones.
The diffraction pattern can be readily indexed as of the
hexagonal Cs2O structure. These results are in good
Fig. 7. (a) TEM image showing the typical crystalline microstructure
of the Cs2O sample; note that at locations when orientation is at
optimum defocus conditions in the electron beam of the microscope,
its intrinsic layered structure is clearly disclosed; (b) FFT and (c)
intensity line profile from the black framed marked region in (a).
Fig. 8. Selected Area Electron diffraction pattern of the Cs2O sample.
agreement with those of the X-ray diffraction measure-
ments. Elemental analysis by EDS further confirmed the
presence of both Cs and O in the ratio Cs:Oꢀ2:1,
suggesting the Cs2O phase.
Fig. 6. XRD of the product and fitting with peak indexing according
to the file: 3R-Cs2O. D represents the deviation of the fitted pattern
from the experimentally determined one.

ARTICLE IN PRESS
S. Gemming et al. / Journal of Solid State Chemistry 178 (2005) 1190–1196
1195
3.4. Spectroscopy
measurements and the DFT calculations, two distinct
transition bands are observed. Whereas at room
temperature, a high energy transition is obtained at
app. 1.95 eV, a low energy transition is observed at
1.65 eV. Fig. 11 shows the influence of the light
excitation intensity on the PL spectrum at 4.2 K. Two
wide peaks at 1.65 and 1.95 eV are clearly revealed.
As the excitation intensity increases, the low energy
transition becomes more pronounced, which is quite
unusual for optical transitions in semiconductors.
The first-principles calculations indicated that the low
energy transition (1.5 eV) is indirect. One would not
expect strong PL from this kind of transition. None-
theless, the K–G lowest energy transition can be
enhanced by contributions from the H, M and L
points in the valence band to the G point in the
conduction band.
An alternative explanation to the appearance of the
two peaks would be some spatial non-uniformity, or an
impurity level. Since low temperature PL is very
sensitive to small (o0.1%) deviations from stoichiome-
try, like vacancies and interstitials, it is not impossible
that the low edge transition at 1.65 eV is associated with
deepimperfections of this kind. Local symmetry break-
ing by vacancies and impurities is known to be a major
factor in enhancing the PL of indirect bandgap
materials, such as GaP [20], or Cl-doped 2H-MoS2
[21]. The low energy transition is observed clearly also
by the absorption measurements (Fig. 8). These
measurements are known to be much less sensitive to
minute concentrations of impurities or lattice disorder
than the PL measurements. This discussion leads to the
conclusion that, in accordance with the theoretical
results, the low energy transition observed in this work
around 1.6 eV emanates from the genuine electronic
structure of Cs2O. It is well known that DF theory
underestimates the electronic transitions in solids, which
may explain the small discrepancy between theory and
experimental results. Further explanation for this
discrepancy lies in the uncertainty of the optical data
of 70.05 eV.
Fig. 9 shows a typical optical absorption spectrum of
Cs2O at room temperature. Two absorption gaps are
clearly observed a weak absorption edge around 1.65 eV
and a stronger absorption edge at 2.1 eV. This spectrum
coincides quite favorably with the results of the DFT
calculations, which suggest an indirect band gapat
1.45 eV and a direct gapat 2.0 eV. Indeed, a plot of the
(absorption)2 Â hn vs. hn yields a straight line [18].
Extrapolation of this line to zero absorption yields a
band gapof 2.27 eV. Similar measurements with a few
other samples yielded values for the band gap in the
range of 2.1–2.25 eV. The low absorption edge was
fitted using the formula (absorption)1/2 Â hn vs. hn,
yielding an indirect gapof 1.67 eV. The XRD peaks’
width (Fig. 6) clearly shows that the powder consists of
particles with micron-size large particles. Therefore,
scattering of the light, which is important for the
analysis of the optical properties of nanoparticles, is
not expected to influence the presently measured
absorption spectrum [19].
The PL spectra of Cs2O at various temperatures are
shown in Fig. 10. In accordance with the absorption
Fig. 9. Absorption spectrum of Cs2O. The indirect (1.67 eV) and the
direct gaps (2.27 eV) are clearly observed.
Fig. 11. Effect of the excitation light intensity on the photolumines-
cence spectra at 4.2 K.
Fig. 10. Photoluminescence spectra of Cs2O at various temperatures.

ARTICLE IN PRESS
S. Gemming et al. / Journal of Solid State Chemistry 178 (2005) 1190–1196
1196
4. Conclusions
optical measurements and to H. Waks of the Electron
Microscopy Unit for his valuable technical assistance
with the TEM. This work was supported by the Minerva
Foundation (Munich) and the G.M.J. Schmidt Minerva
Center for Supramolecular Architectures; The German-
Israeli Research Foundation (GIF) and the Israel
Science Foundation.
The layered compound Cs2O was synthesized and
investigated both theoretically by density-functional
band-structure calculations and experimentally by op-
tical absorption, luminescence, and transmission elec-
tron microscopy measurements. Cs2O crystallizes
preferably in the anti-CdCl2 structure (3R). The gross
structural parameters were obtained in good agreement
both by TEM experiments and density-functional
calculations. The van-der-Waals gapbetween adjacent
Cs–O–Cs triple layers is underestimated in the calcula-
tion as a consequence of the local density approxima-
tion. Nevertheless, the inter-layer interaction across the
van-der-Waals gapis weak as indicated by the low
stacking fault energy calculated for the 1H structure
(0.12 J/m2).
References
[1] J.L. Howe, J. Am. Chem. Soc. 29 (1907) 384.
[2] R. Marquis, Bull. Soc. Chem. 1 (1907) 169.
[3] (a) E. Rengade, Ann. Chim. Phys. [VIII] 11 (1907) 379;
(b) E. Rengade, Ann. Chim. Phys. [VIII] 11 (1907) 381.
[4] (a) K.R. Tsai, P.M. Harris, E.N. Lassettre, J. Phys. Chem. 60
(1956) 338;
(b) K.R. Tsai, P.M. Harris, E.N. Lassettre, J. Phys. Chem. 60
(1956) 347.
The analysis of the electronic and optical properties
shows that Cs2O is a large-gapsemiconductor, which
exhibits two intrinsic energy gaps at 1.5–1.7 eV and
1.9–2.2 eV with different temperature dependence in the
photoluminescence measurements. Two distinct transi-
tions are also obtained from the density-functional
calculations, an indirect K (H, M, L)—G one at 1.45 eV
and a direct one of roughly 2 eV at G. It is thus tempting
to assign the two signals in the absorption and
luminescence measurements to the indirect and direct
electronic transitions found in the calculation of the
pristine, defect-free 3R structure. Judging from the
temperature and excitation energy dependence of the
photoluminescence signals it can, however, not be
excluded, that the peak at lower transition energy is
enhanced by the presence of an impurity level.
[5] A. Helms, W.Z. Klemm, Z. Anorg. Allg. Chem. 242 (1939) 201.
[6] A. Simon, Struct. Bond. 36 (1979) 82.
[7] W. Hesse, W. Schnick, M. Jansen, Prog. Solid State Chem. 19
(1989) 47.
[8] R. Schwartz, G. Brauer, Z. Anorg. Chem. 255 (1947) 102.
[9] N.N. Greenwood, A. Earnshaw, Chemistry of the Elements,
Pergamon Press, Oxford, 1984, pp. 97–100.
[10] P.G. Borzyak, V.F. Bibik, G.S. Kramerenko, Bull. Acad. Sci.
USSR Phys. Ser. 20 (1956) 939.
[11] C.C. Phillips, A.E. Hughes, W. Sibbett, J. Phys. D: Appl. Phys. 17
(1984) 611.
[12] A.H. Sommer, J. Appl. Phys. 51 (1980) 1254.
[13] L. Hackspill, Helv. Chim. Acta 11 (1928) 1008.
[14] G. Brauer, Handbuch der Prap. Anorg. Chem. Enke-Verlag
Stuttgart 2 (1978) 938.
[15] G. Brauer, Handbuch der Prap. Anorg. Chem. Enke-Verlag
Stuttgart 2 (1978) 953.
[16] Y. Rosenfeld Hacohen, R. Popovitz-Biro, Y. Prior, S. Gemming,
G. Seifert, R. Tenne, Phys. Chem. Chem. Phys. 5 (2003) 1644.
[17] P.E. Gregory, P. Chye, H. Sunami, W.E. Spicer, J. Appl. Phys. 46
(1976) 3525–3529.
[18] J.I. Pankove, Optical Processes in Semiconductors, Dover
Publications, Inc., New York, 1971.
Acknowledgment
[19] A. Usami, Solar Energy Mater. Solar Cells 62 (2000) 239.
[20] L. Merz, R.A. Faulkner, P.J. Dean, Phys. Rev. 188 (1969) 1228.
[21] L. Kulyuk, L. Charron, E. Fortin, Phys. Rev. B 68 (2003) 075314.
We are grateful to Dr. R. Popovitz-Biro for useful
discussions, to F. Kupnov for the assistance with the