Characterization of oxides of cesium

Article abstract of DOI:10.1021/jp036432o

Cesium oxides are materials of great interest to the photodetection industry because of their relatively low work function (a??1 eV). Used mainly as coating films for photoemissive devices, they provide high wavelength thresholds and high photocurrents. However, they are unstable, air-sensitive, and hygroscopic, rendering them short-lived and limiting their applications. Although the technology of these devices is highly developed, their characterization on the micro- and nanoscale suffers from their poor chemical stability and poor crystallinity. In the present study, cesium oxides were synthesized from the elements and were characterized using a combination of chemical and structural analysis techniques. Because the reaction products were extremely sensitive to humidity, sample analysis without atmospheric exposure was essential, and techniques were developed for the transfer of the samples to the measurements systems. Extensive data obtained from X-ray energy-dispersive spectroscopy (EDS), X-ray photoelectron spectroscopy (XPS), transmission electron microscopy (TEM), selected-area electron diffraction (SAED), X-ray diffraction (XRD), and Raman microscopy were obtained. Raman spectra with bands at 103, 742, and 1134 cm^-1 strongly confirmed the presence of the oxide, peroxide, and Superoxide ions, respectively, as well as the absence of carbonate as an impurity. The A_1g mode of Cs₂O was detected as an anti-Stokes band at 103 cm^-1. This study provides further insight into the reactivity of the various cesium oxides.

Full text of DOI:10.1021/jp036432o

12360
J. Phys. Chem. B 2004, 108, 12360-12367
Characterization of Oxides of Cesium
A. Band,† A. Albu-Yaron,† T. Livneh,†,‡ H. Cohen,§ Y. Feldman,§ L. Shimon,§
R. Popovitz-Biro,† V. Lyahovitskaya,† and R. Tenne*,†
Department of Materials and Interfaces and Chemical SerVices Unit, Weizmann Institute of Science,
RehoVot 76100, Israel
ReceiVed: August 14, 2003; In Final Form: March 10, 2004
Cesium oxides are materials of great interest to the photodetection industry because of their relatively low
work function (∼1 eV). Used mainly as coating films for photoemissive devices, they provide high wavelength
thresholds and high photocurrents. However, they are unstable, air-sensitive, and hygroscopic, rendering them
short-lived and limiting their applications. Although the technology of these devices is highly developed,
their characterization on the micro- and nanoscale suffers from their poor chemical stability and poor
crystallinity. In the present study, cesium oxides were synthesized from the elements and were characterized
using a combination of chemical and structural analysis techniques. Because the reaction products were
extremely sensitive to humidity, sample analysis without atmospheric exposure was essential, and techniques
were developed for the transfer of the samples to the measurements systems. Extensive data obtained from
X-ray energy-dispersive spectroscopy (EDS), X-ray photoelectron spectroscopy (XPS), transmission electron
microscopy (TEM), selected-area electron diffraction (SAED), X-ray diffraction (XRD), and Raman microscopy
were obtained. Raman spectra with bands at 103, 742, and 1134 cm-1 strongly confirmed the presence of the
oxide, peroxide, and superoxide ions, respectively, as well as the absence of carbonate as an impurity. The
A1g mode of Cs2O was detected as an anti-Stokes band at 103 cm-1. This study provides further insight into
the reactivity of the various cesium oxides.
Introduction
in air. In particular, the reaction of cesium and its oxides with
residual CO2 gas and humidity is highly exothermic. Therefore,
The oxidation of alkali metals is a field of great technological
interest as alkali metal-oxygen compounds play a crucial role
in the engineering of photocathodes, in the promotion of
catalytic reactions, and in the oxidation enhancement of
semiconductor surfaces. In particular, the oxides of cesium play
an important role in many practical photoemissive devices. The
oxides of cesium are low-work-function semiconductors. The
low work function enables the electrons to escape from the solid
upon low-energy excitations and create photoemissive surfaces
suitable for light detection with high quantum efficiency and
high threshold wavelengths.
The complex cesium-oxygen phase diagram consists of an
unusually large number of oxides: the suboxides (e.g., Cs7O,
Cs4O, Cs3O, and Cs7O2), the normal oxide (Cs2O), the sesqui-
oxide (Cs2O3), and the per- and superoxides (Cs2O2, CsO3, and
Cs2O4 or CsO2). Furthermore, no clear relationship has been
found between the Cs/O ratios and the measured work functions.
Various studies during the early 1900s described procedures
for the synthesis of pure cesium oxides.1-4 Later studies revealed
the crystal structures of these oxides.5-9 More recent investiga-
tions of these materials applying ultraviolet and X-ray photo-
electron spectroscopies (UPS and XPS, respectively) and nuclear
magnetic resonance (NMR) techniques suggest that the earlier
characterizations of the various cesium oxide compounds were,
in fact, oversimplified.10-14 Cesium oxides are highly unstable
these materials must be studied under a strictly inert or vacuum
environment. This limitation and the complexity of the phase
diagram makes the comprehensive characterization of these
materials rather important but a very difficult task.
The cesium oxides’ band-gap energies (Eg) and electron
affinities (EA) are not accurately known. From light absorption
and photoemission measurements, Borzyak et al. made reason-
able estimates for Cs2O: Eg ) 2 eV and EA ) 0.8 eV.15
Sommer measured work functions as low as 1 eV (∼1200 nm)
for mixtures of cesium oxides.10 Many studies have been
performed to improve the properties of cesium oxide films as
coatings for photocathode front surfaces. Aging of the photo-
cathodes is often related to degradation of the cesium oxide
coating. It was found that the highest photoemissivity and
greatest stability of photocathodes is obtained with a cesium-
to-oxygen ratio very close to 2:1.16 However, this ratio does
not necessarily indicate the exclusive presence of Cs2O. The
engineering of devices such as photocathodes is focused mostly
on achieving the lowest work function and the highest quantum
efficiency and not necessarily on optimizing the structure and
chemical composition of the deposited films. Furthermore,
cesium oxide films are prepared under moderate conditions, and
therefore, their crystallinity is quite poor, and their phase not
necessarily well-defined, which makes these films yet more
sensitive to the atmosphere or to heating.
In the present study, an extensive effort at the characterization
of cesium oxides, which was not available before, is presented.
The compounds were synthesized directly from the elements.5
Techniques for manipulating and transferring samples from the
reaction vessel to the analysis apparatus under strictly inert
* To whom correspondence should be addressed. E-mail: reshef.tenne@
weizmann.ac.il.
† Department of Materials and Interfaces.
‡ Permanent address: Nuclear Research Center-Negev, Beer Sheva, P.O.
Box 9001, Beer Sheva, Israel 84190.
§ Chemical Services Unit.
10.1021/jp036432o CCC: $27.50 © 2004 American Chemical Society
Published on Web 07/27/2004

Characterization of Oxides of Cesium
J. Phys. Chem. B, Vol. 108, No. 33, 2004 12361
conditions of an anhydrous/anaerobic atmosphere were devel-
oped, enabling their characterization by transmission electron
microscopy, which had not been reported previously for these
materials. Chemical analyses of the products were carried out
by X-ray energy-dispersive spectroscopy (EDS) and X-ray
photoelectron spectroscopy (XPS) measurements. Structural
analysis was achieved by X-ray diffraction (XRD) and electron
diffraction (ED), as well as by transmission electron microscopy
(TEM). Raman microscopy and absorption and luminescence
measurements were carried out to gain additional information
on the products. This unique combination of divergent analytical
techniques and the extensive data obtained allowed for a firm
identification of the major reaction products and sources of
contamination, such as carbonates, that occurred during the
synthesis and handling of the samples.
with a Phillips CM120 TEM (tungsten filament) instrument
operated at 120 kV, fitted with an EDAX system (model
Phoenix) and application programs. The samples were trans-
ferred from the reaction ampule to a desiccator in the glovebox.
The desiccator was subsequently loaded into an inflatable
glovebag filled with argon for mounting of the specimens onto
copper grids coated with either carbon or lacey carbon films.
A sleeve connected the glovebag to the goniometer of the
microscope, permitting transfer of the grids to the TEM
instrument without exposing them to the room atmosphere.
X-ray Photoelectron Spectroscopy (XPS). Measurements were
performed in an AXIS-HS Kratos Analytical instrument, using
a monochromatized Al (KR) source, hν ) 1486.6 eV. The line
shapes were analyzed by curve fitting of the experimental data
using Gaussian-Lorentzian functions and Shirley background
subtraction. A special sample holder was manufactured to
transfer the air-sensitive samples from the argon glovebox into
the XPS chamber under medium (10-2 Torr) vacuum conditions.
A rough estimate of the transfer quality could be obtained
visually from the color of the powder within the analysis
chamber. Of the four samples analyzed by XPS, two samples
underwent a successful transfer. The analysis of the best sample,
number 10, is presented below. Light Ar sputtering was still
needed (4 keV, rate of ∼50 Å/min) to remove adsorbed carbon
and oxygen from the surface, enabling a distinction between
surface and bulk compositions.
Notwithstanding the lengthy synthetic process, the chemical
composition of the samples was not uniform and was found to
vary from one location to another within the ampule. The wide
array of analytical methods engaged in this work, however, was
undertaken to achieve the characterization of these inhomoge-
neous and reactive oxide phases. Each analytical technique
probed a different typical sample area (volume): EDS on TEM
and micro-Raman and XPS measurements probe areas of
approximately 100 nm2, 10 µm2, and 1 mm2, respectively. XRD
measurements were performed on areas of a few square
millimeters. Therefore, the EDS and to some extent also the
micro-Raman techniques were used to characterize single
phases. The differences in depth sensitivity of the above
techniques were used as a complementary source of information
regarding the depth distribution of the various products. For
example, EDS was not sensitive to the superficial cesium
carbonate film that was found by the XPS and TEM/ED
analyses. This film could have been obtained during the transfer
of the sample to the XPS, or it might have always been present
on the sample surface.
Experimental Section
Materials. Cesium metal (>98%) was purchased from Metals
Basis, Alfa Aesar, and oxygen gas (>98%) from Hamercaz
Lehamzan Ltd., Herzliya, Israel.
Synthesis. All syntheses in this work were carried out in
quartz or Pyrex ampules. Cesium was transferred to the ampule,
which was terminated with a high-pressure Teflon valve
(Rotaflo, Stoke-on-Trent, Staffordshire, U.K.), under argon
atmosphere in a dry glovebox. The ampules were first evacuated
to 10-5 Torr, and a measured (corresponding to Cs2O stoichi-
ometry) amount of oxygen gas was added. After the addition
of oxygen, the ampules were sealed, and their contents were
heated to 250 °C overnight. Cooling to room temperature was
done at a controlled rate of 0.1 °C/min. In the attempt to reach
mainly a cesium/oxygen ratio of 2:1 of the final synthesized
product, modifications of the above basic procedure were tried
by varying the oxygen content, the reaction temperature (100-
600 °C), and/or the final cooling rate or in a few cases by
quenching in liquid nitrogen. Over 70 synthesis experiments
were carried out under these various experimental conditions.
An initial visual inspection overview of the resulting solid
powders of all of the performed syntheses indicates inhomo-
geneities in their respective colors (between and within the
ampules). The chemical and structural analysis reported here,
however, refers to the materials resulting from the same
synthesis procedure, described above, but each performed on
different batches (ampules). The direct cesium oxidation
products issued from the synthesis procedure used in this work
exhibited mainly dark purple and to a lesser extent orange colors.
Analysis. Optical Absorption and Luminescence. Measure-
ments were carried out on sealed ampules with flat surfaces. A
double-grating monochromator (HRD, Jobin-Yvon) with a
standard tungsten lamp was used along with a photomultiplier
(RG942, Hamamatsu). Both absorption and luminescence
(excitation with an Ar ion laser at 514.5 nm) were measured in
the range of 500-900 nm at room temperature.
Results
Over 70 synthesis experiments were carried out under various
conditions. To rationalize the experimental findings, the results
presented in this study are restricted only to the synthesis
performed at a Cs/O ratio of 2:1, at a heating temperature of
250 °C, and at a cooling rate of 0.1 °C/min. The inhomogeneity
of the resulting solid powders was manifested by their different
colors. For example, the products of samples 1, 2, 5-8, 12, 13,
and 60 were dark purple, whereas those of samples 3, 4, 9-11,
14, 15, 22, and 39 were partially orange, with most of the
material nevertheless being purple. Furthermore, the colors of
samples that were exposed to ambient atmosphere changed to
white instantaneously, and their composition matched that of
cesium carbonate. Comprehensive examination of the products
by chemical and structural analysis techniques leading to the
identification of a mixture of several oxides is hereby presented.
Note that, because of the very high reactivity of the material
Raman Microscopy. Measurements were carried out directly
on the sealed reaction ampules. Raman measurements were
performed with a Raman imaging microscope (Renishaw 1000)
equipped with a 25-mW He-Ne 632.8-nm laser.
X-ray diffraction was performed on samples transferred to
thin-wall quartz capillaries. An Enraf Nonius FR590 generator
with sealed-tube molybdenum radiation (λ ) 0.71 Å) was used
for powder X-ray diffraction. The generator was set to a voltage
of 50 kV and a current of 30 mA.
Transmission Electron Microscopy (TEM). Imaging, electron
diffraction (ED), and chemical analysis by X-ray energy-
dispersive spectroscopy (EDS) of the products were performed

12362 J. Phys. Chem. B, Vol. 108, No. 33, 2004
Band et al.
TABLE 1: Cs/O Ratios in Different Synthesized Materials
Calculated from EDS Spectra
is ascribed to the presence of adsorbed oxygen on the TEM
grid itself and/or on the sample surface (possibly hydrates).
sample
cesium (%) oxygen (%) Cs/O oxide formulaa
Chemical analysis by XPS provided further evidence of the
inhomogeneity of the samples. The analyzed powder (sample
10) was mostly orange-tinted, but the ampule contained
substantial amounts of the deep purple phase, which could have
been mixed with the orange powder during processing of the
ampule contents and transfer to the XPS instrument. The
presence of Cs2O, in addition to Cs2O2 and CsO2, was confirmed
by these measurements. No significant material degradation was
detected during the XPS measurements. Although the samples
were measured under ultrahigh vacuum (10-8-10-9 Torr), the
outer layer of the material reacted to give cesium carbonate and
cesium carbide, which were removed by a very short low dose
Ar ion sputtering.
1
2
3
39.2
42.0
36.8
46.5
23.2
22.8
24.0
32.3
32.9
62.9
60.8
58.0
63.2
53.5
76.8
77.2
76.0
67.7
67.1
37.1
1.0:1.5
1.0:1.4
1.0:1.7
1.0:1.1
1.0:3.3
1.0:3.4
1.0:3.2
1.0:2.1
1.0:2.0
1.7:1.0
Cs2O3
4 (position a)
4 (position b)
5
6
7
8
9
Cs2O2
CsO3
CsO2
Cs2O
a Suggested oxide according to the O/Cs stoichiometric ratio.
Table 3 presents the results of XPS measurements carried
out on one of the synthesized materials (sample 10). The C 1s
and the O 1s spectra of carbon and oxygen consist of broad
lines that could be decomposed into several components. The
carbon peak at ∼282 eV is assigned to cesium carbide,17
whereas the carbon adsorbed on the surface of the material
appears at a binding energy of 284.8 eV and was used to
calibrate the energy scale. This signal is dramatically decreased
after sputtering. The carbon peak at approximately 288 eV is
attributed to CO32- groups.18,19 The oxygen peaks are assigned
to the negative oxygen ions according to Jupille et al.,13 who
studied thick oxidized Cs layers by XPS. The present results
are in very good agreement with previously published data. The
presence of these ions clearly supports the claim that micro-
crystalline particles of cesium oxide, cesium peroxide, and
cesium superoxide are constituents of the reaction products.
From the assignment of the different peaks, it is possible to
conclude that, in the area of the measurements, CsO2 is the most
abundant product as the superoxide ion is detected with the
highest percentage. Yet, Ar sputtering shows that the sample
surface is typically more oxidized than the bulk material. The
outermost surface consists of CsO2, whereas the depth profile
tends to lead to 1:1 stoichiometry of Cs/O. This ratio is
sometimes spoiled by reoxidation of the sputtered surface by
residual oxygen in the vacuum chamber. Cs2O and Cs2O2 are
identified by the O2- and O2 binding energies (525.3 and
2-
530.5 eV), respectively. The distribution between the oxides
can be estimated from the ratio between the oxygen ions
peaks: Cs2O/Cs2O2/CsO2 ) 1.1:1.0:5.0. Because of the large
analyzed area (1 mm2), microscopic crystals could not be
examined separately. As expected, less successful transfers of
the sample exhibited enhanced surface oxidation, i.e., enrichment
with respect to CsO2, and the presence of carbonates.
Figure 1. XRD pattern of cesium oxide powder from sample 11. (a)
Raw data, (b) intensity profile, (c) peak positions according to PDF
#09-0104.
An interesting in situ result emerged from XPS measurements
obtained under illumination. The light-induced XPS (LIXPS)
analysis20 provides information on photoconductivity and hence
on the optical properties of the sample. Using a HeNe laser
source (1.96 eV), the studied powder exhibited significant light-
induced line shifts, which indicate efficient optical absorption
of red light. These complementary in situ data confirm the
relevance of the XPS analysis to the optical study of the sealed
material (see Figure 6 below).
and the location (sometime 500 m apart) of the different
instruments used, it was not possible to perform more than one
analysis on a single opened ampule (sample).
Chemical Characterization. Results obtained by EDS
analyses of various structural features observed in some of the
synthesized materials are summarized in Table 1: apparent
relative concentrations are listed as Cs/O ratios and represent
the average of three to five measurements. EDS analysis on
typical areas of ∼100 nm2 revealed that the examined samples
had more than one type of oxide present, as reflected also in
the XRD and ED analyses (Figure 1 and Table 2). The variations
in the Cs/O ratio at different locations of a sample and from
sample to sample reflect the inhomogeneity of the synthesized
materials. Also, in most cases, the oxygen content was higher
than expected according to the assigned oxides. This discrepancy
Structural Characterization. XRD and TEM. The X-ray
diffraction pattern of an orange sample (11) held in a quartz
capillary, obtained with an Enraf Nonius diffractometer, is
presented in Figure 1a. The peaks in the pattern (Figure 1b)
can be indexed according to (hkl) lines of Cs2O (PDF #09-0104)
(Figure 1c). The widths of the peaks reflect the small size of
the Cs2O nanoparticles. The XRD patterns of the two other

Characterization of Oxides of Cesium
J. Phys. Chem. B, Vol. 108, No. 33, 2004 12363
TABLE 2: Experimental d Spacings (Å) Calculated from Electron Diffraction Data of the Two Sets of Ring Patterns Compared
with d Spacings of the Corresponding PDFs of the Cesium Oxide Phases
calculated d spacings
calculated d spacings
set 1
d spacings from PDFs
set 2
sample
cesium oxide phase
sample
12
13
CsO3
CsO2
Cs2O3
Cs2O2
Cs2O
Cs2CO3
5.14
4.23
14
15
5.16
4.29
5.12
4.23
3.77
3.19
3.81a
3.16a
(-b) 3.57a
3.49a
3.12a
3.43a
3.16
3.45
3.15
3.46
3.13
3.50
3.14
(-b) 3.10a
3.05a
3.05
2.89
(-b) 2.99a
2.91a
2.89
2.78
2.79a
2.70
2.27
2.63a
2.47
2.58a
2.43a
2.61a
2.45
2.57
2.46
2.58
2.47
2.39
(-b) 2.42a
2.34
1.99
2.34
2.38a
2.23
2.10
2.15
2.12
2.00
2.17
1.95
1.65
2.17
1.98
2.16
1.92
1.97
1.93
1.93a
1.92
1.78
1.62a
1.48
1.76
1.58
1.45
1.73
1.61
1.46
1.66
1.42
1.35
1.19
1.06
0.97
1.63
1.38
1.32
1.16
1.03
0.95
1.62a
1 1.41
1.33
1.60
1.45
1.41
1.32
1.20a
0.94
a Asterisks (*) denote peaks of highest intensity in the corresponding PDFs. b Dashes (-) denote high-intensity peaks not observed in the electron
diffraction patterns.
TABLE 3: XPS Data for Sample 10
atomic concentration (%)
fresh 30-s 1-min
sample sputtered sputtered compound
typical of a mixture of oxide phases and of carbonate. Whereas
Cs2O and Cs2CO3 (PDF #35-0962) can be assigned with a great
degree of confidence, the presence of Cs2O2 (PDF #10-0262)
and Cs2O3 (PDF #10-248) is less firmly established. As a rule,
the characteristic 6.33 Å between the (003) planes of Cs2O was
observed only infrequently in the electron diffraction patterns,
likely because of the high background intensity close to the
central spot. The intensity of the (003) line in the PDF of Cs2O
is rather weak (see Figure 1), which can also explain the
difficulty in observing this peak in the ED pattern. It can
nevertheless be concluded that the present synthesis yields a
mixture of oxide phases.
The source of the carbonate in samples 14 and 15 is not
completely clear. However, these samples most likely were
exposed to air at some point during their transfer from the quartz
ampule to the TEM instrument. However, as far as we are aware,
there was no apparent difference in the preparation or transfer
of these samples compared to those of set 1. The SAED results
are in reasonable agreement with the XRD and XPS measure-
ments, even though they were carried out on different samples
CsO2 and Cs2O2 have tetragonal and orthorhombic unit cells,
respectively.7,8 In both cases, the crystallographic c direction is
parallel to the O2 ion internuclear axis.21 The cesium superoxide
crystallizes in the calcium carbide structure with c ) 7.20 Å
and a ) 4.44 Å, and the cesium peroxide geometry is a distorted
fluorite (CaF2) structure (O2 replacing Ca) with a ) 4.322 Å,
b ) 7.517 Å, and c ) 6.430 Å. The Cs2O has the layered “anti-
CdCl2” structure with a rhombohedral unit cell of dimensions
4.256 × 18.99 Å (18.99 is 3 times the Cs-Cs distance between
adjacent layers, i.e., 3R symmetry). The oxygen atoms are
sandwiched between two cesium layers, and Cs-Cs van der
Waals interactions link adjacent “sandwiches”. CsO3 has two
known phases: a monoclinic form stable at low temperatures
element
BE (eV)
Cs 3d
O 1s
723.5 ( 0.2
525.3 ( 0.1
529.5 ( 0.2
530.5 ( 0.1
282.7 ( 0.2
284.8 ( 0.1
288.1 ( 0.1
26.8
1.9
25.6
0
10.2
26.8
8.7
43.2
4.7
25.9
2.5
8.4
9.1
46.0
5.4
24.1
4.8
9.0
5.8
O2-
-
O2
O2
2-
C 1s
CsC
C
Cs2CO3
6.2
4.9
samples studied by this method coincided clearly with CsO2 as
the major phase, along with Cs2O2.
Typical selected-area electron diffraction (SAED) patterns
obtained by TEM with samples 1-9 and 12-15 are shown in
Figure 2. A background of fairly sharp and continuous rings
can be discerned in all of the diffraction patterns examined,
indicative of the polycrystalline nature of the product. It is
noticed that these diffraction rings belong to two sets of
diffraction patterns (see Figure 2a,b). Table 2 shows the
calculated d spacings of the two sets of patterns, together with
the corresponding values taken from the powder diffraction files
(PDFs) of the various cesium oxide and cesium carbonate phases
in samples 12-15. Note that peaks of high intensity in the PDF
files are denoted by asterisks (*); high-intensity peaks that do
not appear in the ED pattern are denoted by dashes (-). It is
clearly observed that there is relatively little overlap between
the d lines of the two series and that there exist in fact two
separate sets of diffraction patterns. “Set 1” (Figure 2a and Table
2) appears to correspond to the characteristic CsO2 diffraction
lines (PDF #26-0395). The presence of CsO3 (PDF #19-0314)
is established with a lesser degree of certainty in this set. The
patterns of “set 2” (Figure 2b) contain the diffraction rings

12364 J. Phys. Chem. B, Vol. 108, No. 33, 2004
Band et al.
Figure 3. TEM micrographs of some typical particles in sample 15.
In a, the interplanar spacing is 2.99 ( 0.04 Å; in b, the interplanar
spacing is 3.55 ( 0.05 Å.
dioxide entities in the lattice (O2-, O22-).22-25 In some cases,
the various cesium compounds were distributed in well-separated
zones, and in other cases, mixtures of the compounds were
detected on the scale of a few micrometers. The ability to
distinguish between the two can be used to nondestructively
monitor the quality of the cesium oxidation product.
Figure 2. Selected-area electron diffraction (SAED) patterns showing
the two typical sets of diffraction rings: (a) set 1, (b) set 2 (see text).
with cell dimensions of a ) 6.751 Å, b ) 6.267 Å, and c )
9.015 Å and a tetragonal form at room temperature with unit
cell dimensions of a ) 9.73 Å and c ) 8.76 Å. The presence
of the monoclinic CsO3 in the reaction mixture has not been
confirmed.
The image shown in Figure 4a (sample 22) is constructed
from two different zones that are clearly distinguished by their
colors. Mapping the image with 3 µm spatial resolution indicated
the presence of Cs2O2 in the “orange” zone exclusively, although
the presence of other oxides at greater (>1 µm) depths cannot
be definitively excluded. A typical spectrum of the orange zone
is shown in Figure 4a. The attribution of the band at 742 cm-1
Figure 3 shows transmission electron microscopy (TEM)
micrographs taken from specimens 12-15. As mentioned above,
TEM observations of all of the specimens were particularly
difficult to obtain because of the random movement of the
crystals, as well as the bending of the crystals during observa-
tion. The random movements of the crystallites could be
attributed to electron beam induced charging or dehydration
(outgassing) of the crystallites. Thus, in practice, it was almost
impossible to use the goniometer to tilt a crystal to the exact
zone axis because of its drift and continuous change of
orientation relative to the electron beam. In addition, prolonged
electron irradiation led to changes of the morphology of the
material: a lump with an irregular shape was formed instead
of the original agglomerate consisting of individual particles
(Figure 3a). The interplanar distances in Figure 3a are about
2.90-3.00 Å, and the interplanar distances in Figure 3b are
about 3.45-3.55 Å. The fringes in Figure 3a correspond to
interplanar distances of the (104) plane of Cs2O (2.91 Å) or of
the (130) plane of Cs2CO3 (2.99 Å). The fringes in Figure 3b
correspond to the (012) plane of Cs2O (3.43 Å) or the (121)
plane of Cs2CO3 (3.45 Å).
2-
to the vibration of O2 and the analysis of the peroxide ion
spectrum in terms of its higher overtones and anharmonicity
constant were discussed in detail elsewhere.25 All bands other
than the 742 cm-1 band are attributed to the Pyrex ampule. No
Raman signal was obtained from the “blue” part of the sample.
Figure 4b (sample 60) presents another spectrum from a
different sample. Unlike the former case, here, in addition to
the peroxide ion signal at 742 cm-1, a band that is attributed to
O2- is observed at 1134 cm-1. This assignment is based on the
vibrational analysis of the O2- ion in CsO2 that was detected at
1132 cm-1 23
.
Cs2O crystallizes in the 3R anti-CdCl2 layered structure.5 This
structure has two interlayer Raman active modes of A1g and Eg
symmetries. In these optical modes, the outer sublayers of Cs+
cations vibrate out of phase with each other while the middle
sublayer of O2- remains at rest. The displacement vectors are
perpendicular and parallel to the basal plane in the A1g and Eg
modes, respectively, with the A1g mode vibrating at a substan-
tially higher frequency than the Eg mode.26
Raman Microanalysis. Identification of the Various Phases.
CsO2, Cs2O2, and Cs2O were the three main compounds
identified in the synthesized mixtures. The Raman spectra reflect
the crystal lattice vibrations, which are based on crystal
symmetry assignments,5,7,8 and the local vibrations of the various
In Figure 4c, the Raman spectrum from one of the ampules
(sample 39) is shown, revealing an anti-Stokes band at 103
cm-1, attributed to the A1g mode of Cs2O.26 Because of
spectrometer characteristics, this band could not be detected in

Characterization of Oxides of Cesium
J. Phys. Chem. B, Vol. 108, No. 33, 2004 12365
Figure 5. Raman spectrum from sample 39 showing bands that are
attributed to cesium hydrate impurities.
developed during this study provided sufficient protection so
that transformation of the cesium oxides to cesium carbonate
was only superficial and the oxides could be discerned and
analyzed.
Another possible process that might contaminate the present
samples is the hydration of the oxide by residual humidity.
Figure 5 presents a Raman spectrum taken from a sealed ampule
(sample 39). Here, a strong band is detected at 848 cm-1 with
weaker bands at 770 and 887 cm-1, as well as some lower-
frequency modes at 283, 395, 453, and 594 cm-1. In accordance
with Evans,22 the band at 848 cm-1 is attributed to a high hydrate
(possibly octahydrate) and the other bands to lower hydrates.
The residual water content in the quartz ampules was shown to
have a great influence on various chemical and physical
processes.28
Optical Absorption and Luminescence. The focused excitation
light sources for both absorption and luminescence measure-
ments permitted analysis of zones with a homogeneous orange
appearance (sample 9). The absorption and luminescence spectra
of the orange-tinted product are presented in Figure 6a. The
luminescence of the product suggests the dominance of a
material with a direct band gap. The first luminescence peak,
at approximately 660 nm, is due to the band-gap emission
(∼1.88 eV). The second emission peak, at 760 nm (1.63 eV),
could be associated with a donor-acceptor relaxation of a
relatively deep state. The approximate band gap of the products
can be calculated from the absorption data by using the
following equation29
Figure 4. (a) Raman spectra from two distinct zones in sample 22
observed in the displayed image and pointed out in different colors:
purple, dark appearance (dotted line); orange, bright appearance (solid
line). Mapping of the image with 3-µm spatial resolution indicated the
existence of Cs2O2 in the orange zone exclusively. All bands other than
the 742 cm-1 band are denoted with asterisks and are attributed to the
Pyrex ampule. (b) Raman spectrum (background subtracted) from
sample 60 showing a mixture of CsO2 (1134 cm-1) and Cs2O2 (742
cm-1). The origin of the band at 755 cm-1 is not well understood. (c)
Anti-Stokes Raman spectrum from sample 39, showing a band at 103
cm-1 that is attributed to Cs2O. No traces of CsO2 and Cs2O2 are
observed.
the Stokes side of the spectrum. The A1g vibrational frequency
of Cs2O nicely correlates with a rough estimate that is acquired
by comparing it to the A1 vibration at 111 cm-1 of CdI2 that
crystallizes in a 4H-CdCl2 structure,26 after considering a simple
linear chain dynamical model.27 The Eg mode, which appears
below 100 cm-1, cannot be detected with the present instru-
mentation.
Monitoring of Impurities. It is important to note that no
evidence for cesium carbonate was found from the Raman and
XRD measurements in the samples sealed in capillaries within
the glovebox. A sample of commercial Cs2CO3 sealed in a
quartz capillary exhibited a strong Raman-active vibration at
1045 cm-1. No such peak was detected in the current cesium
oxide samples. The XRD patterns did not show d spacings
corresponding to the carbonate ion either. However, TEM and
XPS measurements showed traces of cesium carbonate in several
samples in addition to the cesium oxides. The TEM samples
were handled during transfer under argon in inflatable glovebags.
It is believed that residual oxygen, water and CO2 from the
atmosphere could not be fully avoided in these glovebags.
Moreover, the vapor pressure of the pumping system, which
mainly consists of oil molecules, could itself be a source for
carbonate formation. The XPS samples were opened in the
airlock at a pressure of 0.75 bar. Here again, the residual air
concentration and the oil-based pumping system were sufficient
for the formation of some cesium carbonate from the extremely
reactive cesium oxides. Nevertheless, the handling methods
(Rhν)2 ) hν - Eg
where R is the absorption coefficient, which is proportional to
the measured optical density; hν is the photon energy (in eV);
and Eg is the calculated band gap. In Figure 6b, hν is plotted as
a function of (Rhν)2. The extrapolated band gap is Eg ) 1.9
eV, which agrees with the luminescence threshold and is
consistent with previous data for Cs2O.15 The good fit of the
experimental data suggests that the orange product is a
semiconductor with a direct band gap, most probably Cs2O or
Cs2O2 or mixtures thereof.
Discussion
The characterization of materials that are sensitive to the
ambient atmosphere is difficult. In fact, the use of multiple
techniques was found to be essential for unraveling the
complexity of the cesium oxide phases. Each of the above

12366 J. Phys. Chem. B, Vol. 108, No. 33, 2004
Band et al.
the various oxides are not as negative as that of the reaction of
the respective elements.
Conclusions
The present study shows a comprehensive analysis of the air-
sensitive products of the reaction between cesium and oxygen.
Both direct lattice imaging and electron diffraction modes of
TEM indicate that a mixture of oxides was obtained in this
reaction. The diffraction lines of the XRD patterns can be
associated with more than one oxide, again demonstrating the
heterogeneous character of the present product. Chemical
analysis by EDS in TEM showed the variation of the Cs/O ratios
from different locations on the grids. These ratios are consistent
with the oxide, peroxide, and superoxide of cesium. The
chemical shifts of the oxygen XPS lines clearly attest to the
presence of the oxide, peroxide, and superoxide ions, as well.
Surface oxidation of the samples indicates that an improved
handling of the samples would be needed for future studies.
The chemical analysis also establishes the absence of impurities
in most of the ampules, although the presence of residual water
in the ampule wall is problematic. Optical measurements
(absorption and luminescence) show that, in accordance with
previous works, direct band-gap semiconductors with Eg ≈ 1.9
eV were obtained for the orange-tinted phase, which could be
ascribed to Cs2O or Cs2O2. Raman spectroscopy provided
evidence for the unique spectral vibrations of the peroxide and
superoxide oxygen ions, giving further support to the previous
findings. Furthermore, these measurements established that the
products of the synthesis were cesium oxides with no traces of
cesium carbonate. Cesium carbonate was obtained only during
improper handling of the samples during transfer prior to
analysis. Also, the Raman bands at 103, 742, and 1134 cm-1
were assigned to the oxide, peroxide, and superoxide ions of
cesium, respectively.
Figure 6. (a) Absorption and luminescence spectra of the cesium
oxides mixture in sample 9. (b) Extrapolation of Eg from the absorption
spectrum.
techniques played an important role in the analysis. The ability
to make inferences from the results of one technique about the
results of the other analyses was also crucial in this context.
However, by all means, a most relevant tool for analysis of
inhomogeneous and water-sensitive samples, such as those of
the present work, was the TEM/EDS/ED. This kind of analysis
had not been used before for the analysis of the oxides of
cesium. Raman microscopy was also essential in establishing
the contamination of the samples by carbonate during manipula-
tion or handling. It was also useful for the analysis of hydrates,
which were formed by reaction of the cesium oxides with
residues of water adsorbed to the ampule walls. The XPS results
clearly showed the progressive surface oxidation.
A methodology of this kind could be very useful in the study
of cesium oxide coatings in photoemissive devices as a means
for their characterization, particularly with respect to their aging
mechanisms.
Acknowledgment. This work was supported by a grant in
aid from the Minerva Foundation (Munich) and the G.M.J.
Schmidt Minerva Center for Supramolecular Architecture.
Careful homogenization of the reactants, a lengthy thermal
annealing process, and the gradual heating/cooling of the ampule
were used to attain chemical equilibrium. Furthermore, numer-
ous experiments (not reported in this work) were carried out
during which a number of experimental parameters were varied
in a systematic way. These experiments did not provide a clear
relationship between the experimental conditions and the
reaction products. The calculated free energies, at room tem-
perature, of the formation of both the peroxide (Cs2O2) and the
superoxide (CsO2) are very high (-327 and -387 kJ/mol,
respectively), whereas that of Cs2O is somewhat lower (-274
kJ/mol).30 Therefore, Cs2O can disproportionate into a mixture
of peroxides and elemental cesium, which could further react
with oxygen to give the superoxides. The relatively modest
difference between the free energies of formation of the cesium
peroxide and superoxide might explain the difficulty in obtaining
a pure reaction product. However, the predominance of CsO2
in the analyzed samples was expected because of its greater
thermodynamic stability. Also, the very high free energy of
formation of Cs2CO3 (-1019 kJ/mol) might explain the high
chemical reactivity of the cesium oxides with respect to ambient
CO2, although the free energies of the reactions of CO2 with
References and Notes
(1) Howe, J. L. J. Am. Chem. Soc. 1907, 29, 384.
(2) Marquis, R. Bull. Soc. Chem. 1907, 1, 169.
(3) Rengade, E. Ann. Chim. Phys. VIII 1907, 11, 379.
(4) Rengade, E. Ann. Chim. Phys. VIII 1907, 11, 381.
(5) Tsai, K. R. J. Phys. Chem. 1956, 60, 338.
(6) Tsai, K. R. J. Phys. Chem. 1956, 60, 347.
(7) Foppl, H. Z. Anorg. Allg. Chem. 1957, 291, 12.
(8) Dudarev, V. Ya.; Tsentsiper, A. B.; Dobrolyubova, M. S. SoV. Phys.
Crystallogr. 1974, 18, 477.
(9) Helms, A.; Klemm, W. Z. Anorg. Allg. Chem. 1939, 242, 201.
(10) Sommer, A. H. J. Appl. Phys. 1980, 51, 1254.
(11) Su, C. Y.; Lindau, I.; Spicer, W. E. Chem. Phys. Lett. 1982, 87,
523.
(12) Woratschek, B.; Sesselmann, W.; Kuppers, J.; Ertl, G.; Haberland,
H. J. Chem. Phys. 1987, 86, 4.
(13) Jupille, J.; Dolle, P.; Besancon, M. Surf. Sci. 1992, 260, 271.
(14) Krawietz, T. R.; Murray, D. K.; Haw, J. F. J. Phys. Chem. A 1998,
102, 8779.
(15) Borzyak, P. G.; Bibik, V. F.; Kramerenko, G. S. Bull. Acad. Sci.
USSR, Phys. Ser. 1956, 20, 939.
(16) Phillips, C. C.; Hughes, A. E.; Sibbett, W. J. Phys. D: Appl. Phys.
1984, 17, 611.
(17) Ramqvist, L.; Hamrin, K.; Johansson, G.; Fahlman, A.; Nordling,
C. J. Phys. Chem. Solids 1969, 30, 1835.

Characterization of Oxides of Cesium
J. Phys. Chem. B, Vol. 108, No. 33, 2004 12367
(18) Hammond, J. S.; Gaarenstroom, S. W.; Winograd, N. Anal. Chem.
1975, 47, 2194.
(25) Livneh, T.; Band, A.; Tenne, R. J. Raman Spectrosc. 2002, 33,
675.
(19) Gammond, J. S.; Holubka, J. W.; Devries, J. E.; Duckie, R. A.
Corros. Sci. 1981, 21, 239.
(20) Buller, R.; Cohen, H.; Popovitz-Biro, R.; Minkin, E.; Lifshitz, E.;
Lahav, M. AdV. Funct. Mater. 2002, 12, 713.
(21) Pedio, M.; Benfatto, M.; Aminpirooz, S.; Hasse, J. Phys. ReV. B
1994, 50, 6596.
(22) Evans, J. C. Chem. Commun. 1969, 682.
(23) Hunter-Saphir, S. A.; Creighton, J. A. J. Raman Spectrosc. 1998,
29, 417.
(26) Adams, D. M.; Tan, T. J. Phys. Chem. Solids 1981, 42, 559.
(27) Anderson, A.; Lo, Y. W.; Todoeschuck, J. P. Spectrosc. Lett. 1981,
14, 301.
(28) Scha¨fer, H.; Grofe, T.; Trenkel, M. J. Solid State Chem. 1973, 8,
14.
(29) Pankove, J. I. Optical Processes in Semiconductors; Dover Publica-
tions Inc.: New York, 1971; p 36.
(30) Bard, A. J.; Parsons, R.; Jordan, J. Standard Potentials in Aqueous
Solution; Marcel Dekker Inc.: New York, 1985; pp 755-757.
(24) Eysel, H. H.; Thym, S. Z. Anorg. Allg. Chem. 1975, 411, 97.