Electronic structure of Cr1-δS (δ=0, 0.17) with NiAs-type crystal structure

Article abstract of DOI:10.1016/S0038-1098(02)00806-2

Valence-band and conduction-band the electronic structure of the CrS (δ=0) and Cr₅S₆ (δ=0.17) has been investigated by means of photoemission and inverse-photoemission spectroscopies. The bandwidth of the valence bands of Cr₅S₆ (8.5 eV) is wider than that of CrS (8.1 eV), though the Cr 3d partial density of states evaluated from the Cr 3p-3d resonant photoemission spectroscopy is almost unchanged between the two compounds concerning shapes as well as binding energies. The Cr 3d (t_2g) exchange splitting energies of CrS and Cr₅S₆ are determined to be 3.9 and 3.3 eV, respectively.

Full text of DOI:10.1016/S0038-1098(02)00806-2

Solid State Communications 125 (2003) 243–246
www.elsevier.com/locate/ssc
Electronic structure of Cr12dS (d ¼ 0, 0.17)
with NiAs-type crystal structure
a
M. Koyama , H. Sato , Y. Ueda , C. Hirai , M. Taniguchi
b,
a
b
b,c
*
aKure National College of Technology, Agaminami 2-2-11, Kure 737-8506, Japan
bGraduate School of Science, Hiroshima University, Kagamiyama 1-3-1, Higashi-Hiroshima 739-8526, Japan
cHiroshima Synchrotron Radiation Center, Hiroshima University, Kagamiyama 2-313, Higashi-Hiroshima 739-8526, Japan
Received 11 November 2002; accepted 22 November 2002 by H. Akai
Abstract
Valence-band and conduction-band the electronic structure of the CrS (d ¼ 0) and Cr5S6 (d ¼ 0.17) has been investigated by
means of photoemission and inverse-photoemission spectroscopies. The bandwidth of the valence bands of Cr5S6 (8.5 eV) is
wider than that of CrS (8.1 eV), though the Cr 3d partial density of states evaluated from the Cr 3p–3d resonant photoemission
spectroscopy is almost unchanged between the two compounds concerning shapes as well as binding energies. The Cr 3d (t2g)
exchange splitting energies of CrS and Cr5S6 are determined to be 3.9 and 3.3 eV, respectively.
q 2003 Elsevier Science Ltd. All rights reserved.
PACS: 71.20.Be; 79.60. 2 i
Keywords: A. Transition metal compounds; D. Electronic states; E. Photoemission spectroscopy
It is well known that chromium chalcogenides based on the
NiAs-type structure exhibit various physical properties
depending on the chalcogen and ordered Cr vacancy
[1–6]. Cr12dTe are ferromagnetic metals with the d-
dependent Curie temperatures, while Cr12dSe and Cr12dS
exhibit predominantly antiferromagnetic behavior [7].
Among the chromium chalcogenides, CrS is an antiferro-
hybridization between the Cr 3d and S 3p states. Some band-
structure calculations have been carried out on CrS so far
[8–10]. However, there is no experimental information on
electronic structure of Cr12dS.
In this study, we report the valance-band and conduc-
tion-band densities of states (DOSs) of Cr12dS (d ¼ 0; 0.17)
obtained by means of the ultraviolet photoemission and
inverse-photoemission spectroscopies (UPS and IPES) as
well as the Cr 3d partial DOSs of Cr12dS deduced from the
resonant photoemission spectroscopy (RPES) in the Cr 3p–
3d excitation region. These DOSs are compared with the
result of the band-structure calculation and discussed in the
light of the effect of the vacancies at the Cr site on electronic
structure.
´
magnet with the Neel temperature of 450 K, while Cr5S6 a
ferrimagnet with the Curie temperature of 300 K and shows
the antiferromagnetic behavior below 150 K [2]. The
metal–semiconductor transition is observed for Cr12dS
with d # 0:1 at ,620 K, while Cr12dS with d . 0:1 are
metals. The crystal structures of CrS and Cr5S6 are shown in
Fig. 1, where the Cr vacancies of Cr5S6 order regularly and
the unit cell becomes twice of that of CrS along the c-axis.
A wide variety of physical properties of Cr12dS, which
change also with the applied pressure, is derived from the
electronic structure with respect to the Cr 3d states and
Samples used for the present experiments were poly-
crystals of Cr12dS (d ¼ 0, 0.17) prepared by the following
procedure. An appropriate amount of Cr (99.999%) and S
(99.999%) elements sealed in an evacuated quartz ampoule
was heated at 1173 K for a week. The reacted products were
crushed and mixed well in Ar atmosphere. The products
sealed in the evacuated ampoule were again heated at
1773 K for several hours, cooled down to 973 K in 24 h and
*
Corresponding author. Tel.: þ81-824-247471; fax: þ81-824-
240719.
E-mail address: jinjin@hiroshima-u.ac.jp (H. Sato).
0038-1098/03/$ - see front matter q 2003 Elsevier Science Ltd. All rights reserved.
PII: S0038-1098(02)00806-2

244
M. Koyama et al. / Solid State Communications 125 (2003) 243–246
Fig. 1. Crystal structures of (a) CrS and (b) Cr5S6.
then quenched into iced water. The crystal structure and
composition were checked by the X-ray powder diffraction
and electron-probe micro-analysis (EPMA), respectively.1
Lattice parameters of CrS and Cr5S6 with single phase of the
Fig. 2. (a) UPS spectra measured at hn ¼ 21:2 eV and IPES spectra
of Cr12dS. (b) The total DOS of CrS derived by the band-structure
calculation [8].
˚
NiAs-type structure at room temperature are a0 ¼ 3:469 A
˚
˚
˚
and c0 ¼ 5:786 A, and a0 ¼ 3:448 A and c0 ¼ 5:751 A,
respectively, where a0 and c0 are defined in Fig. 1. The
smaller lattice parameters of Cr5S6 are noticed due to the Cr
vacancies. Clean surfaces for the UPS, IPES and RPES
measurements were in situ obtained by scraping with a
diamond file.
CrS exhibits a weak shoulder at 20.7 eV, two peaks at
21.5 and 26.7 eV and a broad structure around 24.7 eV
between the two peaks. The IPES spectrum, on the other
hand, shows a structure at 2.4 eV and a broad structure
between 6 and 11 eV. The whole features of the UPS and
IPES spectra of Cr5S6 are similar to those of CrS. We find a
weak shoulder at 20.7 eV, peak structures at 21.5, 24.7,
26.4, and 1.8 eV and a broad structure between 6 and 11 eV
in the spectra of Cr5S6.
Apparatus for UPS and IPES is composed of three
chambers for sample preparation, UPS experiments and
IPES experiments with base pressures of 2 £ 10210
,
2 £ 10210 and 5 £ 10210 Torr, respectively. The UPS
spectrometer is made up of an He discharge lump ðhn ¼
21:2 eVÞ and a double-cylindrical mirror-analyzer (DCMA).
Pass energy of photoelectrons through DCMA was set to
16.0 eV with the corresponding energy resolution of
0.20 eV. The IPES spectrometer consists of an Erdmann–
Zipf low-energy electron gun and a bandpass photon
detector centered at hn ¼ 9:43 eV [11,12]. Total energy
resolution is 0.56 eV. The energy of the UPS and IPES
spectra was defined with respect to the Fermi level (EF) and
both spectra were connected at EF.
Fig. 2(b) shows theoretical total DOS for the antiferro-
magnetic phase of CrS by means of the band-structure
calculation using the augmented spherical wave method [8].
The peak structures appear at 27.1, 24.7, 20.7, and
1.9 eV. The hybridization between the Cr 3d and S 3p states
exists over the calculated energy region. In the hybridization
bands, the Cr 3d " and Cr 3d # states with t2g symmetry
mainly contribute to the peaks at 20.7 and 1.9 eV,
respectively, providing the Cr 3d (t2g) exchange splitting
energy of 2.6 eV. The S 3p states hybridized with the Cr 3d
states with eg symmetry, on the other hand, contribute to the
peak at 24.7 eV. The peak at 27.1 eV is predominantly
derived from the S 3p states.
The Cr 3p–3d RPES experiments were carried out on the
beamline BL7 at Hiroshima Synchrotron Radiation Center
(HSRC). Light from the storage ring (HiSOR) is mono-
chromatized by the Dragon-type monochromator [13]. We
used a photoemission spectrometer with a hemispherical
photoelectron analyzer (Gammadata Scienta SES100)
attached to the end station of BL7. The total energy
resolution was set below 0.1 eV around hn ¼ 50 eV. All
experiments were performed at room temperature.
For the valence bands of Cr12dS mainly composed of the
Cr 3d and S 3p states, the photoionization cross-section of
the Cr 3d state is only twice larger than that of the S 3p state
at hn ¼ 21:2 eV [14]. The UPS spectra of Cr12dS measured
at hn ¼ 21:2 eV, thus, provide information on the Cr 3d
states as well as the S 3p states. The experimental results
suggest that the Cr 3d–S 3p hybridization bands spread over
the top ,8 eV region in the valence bands. The whole
features of the UPS and IPES spectra are in good agreement
with the theoretical DOS. On the basis of the band-structure
calculation, we assign the two peaks near EF in the UPS and
IPES spectra of Cr12dS to the occupied Cr 3d (t2g) " and
Fig. 2(a) shows the valence-band UPS spectra measured
at hn ¼ 21:2 eV and the conduction-band IPES spectra of
CrS ðd ¼ 0Þ and Cr5S6 ðd ¼ 0:17Þ: The UPS spectrum of
1
Although the mixed NiAs-type and monoclinic phases for CrS
has been reported [6], the reflections from the monoclinic phase
were not observed in the present X-ray diffraction patterns.
unoccupied Cr 3d (t2g)
#
states with nearly localized

M. Koyama et al. / Solid State Communications 125 (2003) 243–246
245
character, respectively. The structures around 24.7 eV in
the UPS spectra are attributed to the S 3p states hybridized
with the Cr 3d (eg) " states, while the most deeper peaks at
26.7 eV for CrS and at 26.4 eV for Cr5S6 are almost
derived from the S 3p states.
Photoemission intensities have been normalized to the
monochromator output. Vertical bars indicate structures due
to emissions of the Cr MVV Auger electrons. One can
recognize that the peak at 21.5 eV exhibits the prominent
resonance. With increasing hn from 38.3 eV, the intensity of
the main peak at 21.5 eV decreases to the minimum at
hn ¼ 40:8 eV, reaches sharply to the maximum at
hn ¼ 47:5 eV, and then decreases gradually with hn. The
resonance enhancement takes place as a result of an
interference between the direct excitation process of the
Cr 3d electrons ð3p63d4 þ hn ! 3p63d3 þ 1fÞ and the
discrete Cr 3p–3d core excitation process followed by a
super Coster–Kronig decay ð3p63d3 þ hn ! 3p53d5 !
3p63d3 þ 1fÞ; where 1f represents emitted photoelectrons.
Since only the Cr 3d states are resonantly enhanced for hn
near the Cr 3p–3d excitation region, we can estimate a
measure of the Cr 3d contribution to the valence bands (Cr
3d partial DOS) by subtracting the spectrum at antireso-
nance ðhn ¼ 40:8 eVÞ from that on resonance ðhn ¼ 47:5
eVÞ:
Although the UPS and IPES spectra of Cr12dS are
similar, we find several differences with respect to the
following points: (i) The bandwidth of the valence bands of
Cr5S6 (8.5 eV) are wider than that of CrS (8.1 eV). (ii) The
energy position of the S 3p-derived peak of CrS shifts
toward the EF side by 0.3 eV for Cr5S6. (iii) The Cr 3d (t2g)
exchange splitting energy of Cr5S6 (3.9 eV) is larger than
that of CrS (3.3 eV), which is larger than the theoretical
value of 2.6 eV. It is noted that the energy position of the Cr
3d (t2g) " peak is at 21.5 eV for both compounds in spite of
the energy shift of the S 3p-derived peak.
In order to deduce the Cr 3d contributions to the valence
bands of Cr12dS, we have carried out the RPES experiments
in the Cr 3p–3d excitation region. Fig. 3(a) and (b) shows a
series of the RPES spectra of CrS for hn between 38.3 and
60.0 eV and the Cr 3p–3d core absorption spectrum
measured by means of total electron yield, respectively.
The derived Cr 3d partial DOSs of Cr12dS are shown in
Fig. 4, where the background contributions due to the
secondary electrons have been removed. The Cr 3d DOS is
composed of a weak shoulder near EF, a main peak at
21.5 eV and a broad structure between 22.5 and 26 eV.
The features of the Cr 3d DOSs of Cr12dS are the same with
respect to the peak-energy position and relative intensity.
The RPES results again support the assignment of the peak
at 21.5 eV in the UPS spectra to the Cr 3d (t2g) " states. It
should be noticed that almost no contribution of the Cr 3d
states to the valence-band region deeper than 26 eV. The
Cr 3d–S 3p hybridization bands are in the top 6 eV, and the
peaks at 26.7 eV (CrS) and 26.4 eV (Cr5S6) are derived
from the S 3p states.
Here, we discuss the Cr vacancy-dependence of the
electronic structure of Cr12dS. The Cr ion has nominally 3d4
electron configuration in the CrS (Cr2þS22) compound.
With an increase in d, the number of electrons occupying the
Cr 3d orbitals decreases and becomes 3d3 configuration in
the Cr2S3 compounds under the assumption that the valence
of the S ions remains 2 2 . Hole doping into the Cr 3d bands
due to the Cr vacancies would lead to the energy shift of the
Fig. 3. (a) A series of photoemission spectra of CrS in the Cr 3p–3d
excitation region. The photoemission intensities have been normal-
ized to monochromator output. (b) The Cr 3p–3d core absorption
spectrum measured by means of total electron yield.
Fig. 4. Cr 3d partial DOSs of CrS and Cr5S6. Background
contribution due to the secondary electrons has been removed.

246
M. Koyama et al. / Solid State Communications 125 (2003) 243–246
Cr 3d-derived peak in the photoemission spectra toward the
EF side with d, in contradiction to the experimental results in
Figs. 2(a) and 4, where the energy position of the Cr 3d (t2g)-
derived peak is unchanged at 21.5 eV between the two
compounds. The energy shift by ,0.3 eV towards the EF
side is observed for the S 3p-derived peaks from CrS (d ¼ 0;
26.7 eV) to Cr5S6 (d ¼ 0:17; 26.4 eV), rather than for the
Cr 3d peaks. This suggests that the holes are mainly
introduced to the S 3p bands in the Cr12dS compounds.
Recently, we have investigated magnetic circular
dichroism (MCD) in the Cr 2p–3d absorption spectra of
Cr12dTe [15]. Theoretical analyses for the MCD spectra
using a configuration interaction theory with a CrTe6 cluster
also suggest that holes are mainly doped into the Te 5p states
rather than the Cr 3d states.
Acknowledgements
The authors are grateful to K. Takada and H. Okuda for
their experimental supports.
References
[1] F. Jellinek, Acta Crystallogr. 10 (1957) 620.
[2] T. Kamigaichi, J. Sci. Hiroshima Univ. A 24 (1960) 371.
[3] K. Masumoto, T. Kamigaichi, T. Hihara, J. Phys. Soc. Jpn 15
(1960) 1355.
[4] T.J.A. Popma, C.F. Van Bruggen, J. Inorg. Nucl. Chem. 31
(1969) 73.
The wider bandwidth of the Cr 3d–S 3p hybridization
bands and the smaller Cr 3d (t2g) exchange splitting energy
of Cr5S6 than those of CrS are qualitatively explained by
taking into account the smaller lattice constants of Cr5S6
caused by existence of the Cr vacancies. The difference of
the bandwidth is understood as originating from the higher
degree of the Cr 3d–S 3p hybridization in Cr5S6 in
comparison with that in CrS. As concerns the Cr 3d (t2g)
exchange splitting energy, the Cr 3d states directly interact
with the nearest Cr 3d states, since the Cr ions order in the c
plane in the NiAs-type structure (see Fig. 1), and the Cr 3d–
Cr 3d hybridization plays an important role in determining
the Cr 3d (t2g) exchange splitting energy [16]. The smaller
Cr 3d (t2g) exchange splitting energy for Cr5S6 than that for
CrS is assumed to be due to the higher degree of the Cr 3d–
Cr 3d hybridization in Cr5S6 in comparison with that in CrS.
In summary, the UPS and IPES spectra of Cr12dS
suggest that the holes due to the ordered Cr vacancies are
mainly doped into the S 3p bands rather than the Cr 3d
states. The wider bandwidth of the valence bands and the
smaller Cr 3d (t2g) exchange splitting energy of Cr5S6 are
qualitatively explained on the basis of the smaller lattice
constants of Cr5S6.
[5] H. Konno, M. Yuzuri, J. Phys. Soc. Jpn 57 (1988) 621.
[6] V.V. Sokolovich, Sov. Phys. Solid State 34 (1992) 371.
[7] J.B. Goodenough, Magnetism and the Chemical Bond,
Interscience, New York, 1963.
[8] J. Dijkstra, C.F. van Bruggen, C. Haas, R.A. de Groot, J. Phys.:
Condens. Matter 1 (1989) 9163.
[9] T. Kawakami, N. Tanaka, K. Motizuki, J. Magn. Magn. Mater.
196–197 (1999) 629.
[10] D. Hobbs, J. Hafner, J. Phys.: Condens. Matter 11 (1999)
8197.
[11] K. Yokoyama, K. Nishihara, K. Mimura, Y. Hari, M.
Taniguchi, Y. Ueda, M. Fujisawa, Rev. Sci. Instrum. 64
(1993) 87.
[12] Y. Ueda, K. Nishihara, K. Mimura, Y. Hari, M. Taniguchi, M.
Fujisawa, Nucl. Instrum. Methods A 330 (1993) 140.
[13] M. Taniguchi, J. Ghijsen, J. Syn. Rad. 5 (1998) 1176.
[14] J.J. Yeh, I. Lindau, At. Data Nucl. Data Tables 32 (1985) 1.
[15] K. Yaji, A. Kimura, C. Hirai, H. Sato, M. Taniguchi, M.
Koyama, K. Shimada, A. Tanaka, T. Muro, S. Imada, S. Suga,
unpublished.
[16] H. Sato, M. Taniguchi, K. Mimura, S. Senba, H. Namatame,
Y. Ueda, Phys. Rev. B 61 (2000) 10622.