16 132
C. H. HO, Y. S. HUANG, K. K. TIONG, AND P. C. LIAO
PRB 58
values of E and E could be obtained by fitting a different
g
p
energy range, i.e., by rejecting some points at lower or
higher photon energies. From this selective omission of data,
an error of the order Ϯ0.02 eV can be deduced for the esti-
mation of E . Fitting data on different samples give similar
g
parameters, even though there are differences in the absorp-
tion spectra due to differences in sample thicknesses.
From our experiments, the indirect gaps at room tempera-
ture, denoted as Eg
ʈ
(EgЌ), are, respectively, determined to
be 1.35Ϯ0.02 (1.38Ϯ0.02) eV for ReS and 1.17Ϯ0.02
2
(
1.20Ϯ0.02) eV for ReSe . Here, Eg
ʈ
and EgЌ refer, respec-
b and EЌb polarizations.
It is noticed that the values for EgЌ are larger than those of
Eg , and are similar to those of our previous report of Eg ,
which determined from the absorption data of the unpolar-
2
tively, to the indirect gap of the E
ʈ
ʈ
13
ized incident light. Our values differ slightly from previous
published works.2
,14–16
The first available data, 1.33 eV for
ReS and 1.15 eV for ReSe , were obtained by Wildervanck
2
2
2
and Jellinek from transmittance measurements. Koffyberg,
Ouight, and Wold14 analyzed the photoelectrochemical
spectral-response data resulting in a lowest-energy indirect-
optical band gap of 1.4 eV for ReS . From optical-absorption
2
measurements, Marzik et al.15 determined the indirect-
optical band gaps of ReS and ReSe to be 1.32͑5͒ eV for
2
2
n-type ReS and 1.17͑5͒ eV for n-type ReSe . Recent at-
2
2
tempts to determine the energy gaps of p-type ReS and
2
p-type ReSe using the photoresponse spectra obtained at
2
semiconductor-electrolyte interfaces were carried out by
Wheeler, Leland, and Bard. The values obtained were
FIG. 3. Experimental points of (␣ប)1/2 vs ប that are deduced
from polarization-dependent absorption measurements for ͑a͒ ReS2
16
1.31Ϯ0.01 and 1.22Ϯ0.01 eV for ReS and ReSe , respec-
2
2
and ͑b͒ ReSe at several representative temperatures between 25
2
tively.
ReX ͑XϭS, Se͒ can be thought of as distorted 1T-MX
and 500 K. The hollow circles ͑solid squares͒ are data points from
2
2
the E
ʈ
b (EЌb) polarization measurements and the solid lines are
17
dichalcogenides. The 1T-MX2 phases consist of edge-
shared MX octahedra. In each MX layer, the metal-atom
the least-squares fits to Eq. ͑2͒.
6
2
sheet is sandwiched by chalcogen-atom sheets, and the metal
A and B are constants. The first term on the right-hand side
of Eq. ͑2͒ corresponds to an absorption of a photon and a
phonon, whereas the second term corresponds to an absorp-
tion of a photon and emission of a phonon and contributes
only when បуE ϩE . There is a large residual absorp-
atoms of an undistorted MX layer form a hexagonal lattice.
2
3
ReS and ReSe have a d electron count, and their metal-
2
2
atom sheets exhibit a clustering pattern of ‘‘diamond
chains.’’ The atoms comprising the Re diamonds are copla-
4
nar, but each diamond unit is canted at a small angle from
the basal plane. This canting results in a small variation of
the Re atom heights perpendicular to the basal plane. The
distortion of the Re atom sheet from perfect hexagonal sym-
metry further creates a distortion of the chalcogen sheets,
both perpendicular and parallel to the basal plane. One con-
sequence of this distortion is the opening of an energy gap in
the band structure due to the mutual repulsion of the orbitals
g
p
tion at photon energies below the absorption edge. The large
values of the absorption coefficient ␣ below absorption edge
of ReS and ReSe most probably indicate the existence of
2
2
impurities or defects in the materials. At this point, we have
not considered in detail the effect of these impurity or defect
states. For simplicity, in our present study, the residual ab-
sorption is assumed to be a constant and subtracted out for
the evaluation of the band gap E and phonon energy E .
18
around the Fermi energy level. Recently, Kelty et al., and
Kertesz and Hoffmann19 employed simple tight-binding elec-
tronic band-structure calculations of the extended Huckel-
type on ReS and ReSe , respectively. The results show that
g
p
The data of ReS and ReSe at different temperatures were
2
2
then to be fitted to Eq. ͑2͒. Representative results are shown
in Fig. 3, where the hollow circles and solid squares are
representative experimental points deduced from different
2
2
the band levels around the band gap are dominated by the Re
polarization ͑E
ʈ
b and EЌb polarizations, respectively͒ trans-
metal d orbitals. The contribution of chalcogen p orbitals
z
mittance spectra and the solid lines are fitted to Eq. ͑2͒. The
are much stronger at the top portion of the valence band than
at the bottom portion of the conduction band. For the chal-
cogen p and p orbitals, the top portion of the valence band
results strongly indicate that ReS and ReSe are indirect
2
2
band-gap semiconductors, in which E
ʈ
b polarization exhibits
x
y
a smaller band gap and a single phonon makes important
contributions in assisting the indirect transitions. The non-
uniform thicknesses and unsmooth sample surface will tend
to deviate the incident angles from the normal direction, re-
sulting in some variations in the absorption spectra. Differing
has nearly equal contributions from all chalcogen atoms. The
values of the theoretical indirect gap for a single ReS2
layer,18 the three-dimensional ReS ͑Ref. 18͒ and ReSe2
2
͑Ref. 19͒ lattices are 1.27, 0.81, and 0.87 eV, respectively.
The value of 1.27 eV for a single layer of ReS ͑Ref. 18͒ is
2