Solid State Communications, Vol. 34, PP. 891—894.

Pergamon Press Ltd. 1980. Printed in Great Britain.

ON THE PHYSICAL BASIS FOR URBACH’S RULE

J.D. Wiley
Dept. of Electrical and Computer Engr., University of Wisconsin, Madison 53706, U.S.A.
and
E. Schonherr and A. Breitschwerdt
Max-Planck-Institut für Festkorperforschung, Stuttgart, Federal Republic of Germany
(Received 11 January 1980 by M. Cardona)

The low-energy tail of the E U a exciton in GeS obeys Urbach’s rule, with
Urbach parameters a0 = 1.45 ±0.05, and hw~= 8.7 ±0.6 meV. The
energy ~ corresponds to a previously measured rigid-layer vibrational
mode which has no associated electric field. This finding is inconsistent
with Dow and Redfield’s unified theory of Urbach and exponential
absorption edges. Our results are consistent with Sumi and Toyazawa’s
theory of Urbach edges, and with Fivaz and Mooser’s model for electron—
phonon interactions in layered compounds.

IN 1953, URBACH [1] reported that the absorption with

edge of AgBr exhibits a low-energy tail which is accu-
rately exponential over several orders of maguitude of
the absorption coefficient. Subsequent investigation
[2] has revealed similar behavior in a great number and
variety of insulators and semiconductors, including (to
name only a few) the silver halide, alkali halide, and
oxide insulators; and (at least some members of) the
group VI, III—V, Il—VI, IV—VI, II—v2, V2V13, ~
IV—V2, and organic semiconductors. It is therefore
clear that the Urbach edge is a macroscopic manifesta-
thin of some fundamental aspect of the optical proper-
ties of solids. This observation has led to a considerable
amount of theoretical work, and numerous models
which predict exponential absorption edges [2, 3].
Unfortunately, although these theories often invoke
radically different microscopic mechanisms, their
macroscopic predictions are sufficiently similar that it
is difficult to distinguish among them experimentally,
This is particularly true of the two most prominent
theories: those of Sumi and Toyazawa [4], and Dow
and Redfield [51 Thus, the microscopic mechanism
. the Urbach edge can be observed even at 0 K.
underlying Urbach’s rule has remained quite contro-
versial. In this paper we present the first clear example
of an Urbach edge which is consistent with Sumi and
Toyazawa’s theory, but inconsistent with that of Dow
and Redfield.
In its most general form, Urbach’s rule is given by
[3]
2kT\ (hwD~
U = (—-j
0~ tanh \2kT/ (2)
where a is the linear absorption coefficient; E is the
photon energy; and a0, E0, a0, and w~,are parameters
which depend on the material.
Equations (1) and (2) have been derived by Sumi
and Toyazawa on the basis of a model in which exciton—
phonon interactions are assumed to be responsible for
the exponential spectral dependence [4]. In Sumi and
Toyazawa’s theory, the exciton is treated as being
partially mobile and partially localized, due to momen-
tary trapping by lattice deformations. The Urbach edge
is then a consequence of the interplay between the free
and localized natures of the exciton. The parameter
hw~is the energy of the dominant phonon involved in
the exciton—phonon interaction, and a0 is inversely
proportional to the strength of that interaction. In
materials with a sufficiently strong exciton—phonon
interaction, the excitons can become self-trapped, and
Dow and Redfield [5] have noted that exponential
absorption edges [not necessarily obeying the full
Urbach’s rule of equations (1) and (2)] are often ob-
served in heavily-doped, damaged, or disordered
materials. Using this as a clue, and postulating the
existence of a common underlying mechanism for all
exponential and Urbach edges, they propose a unified

~/ a(Eo E)\
— theory
is inby
caused which the exponential
electric-field spectral
ionization dependence
of excitons. For
a(E) = a0 exp kT )‘
__________
(1) pure (undoped) materials, the fields are caused by LO

891
892 ON THE PHYSICAL BASIS FOR URBACH’S RULE
I surfaces of all samples were mirror-like, and free of
- ~___—~
LU 0~ -
E II b
0
2 absorption edge at 4.2 K for the E la and E lb polar-
0 -
z izations. The extraordinary in-plane dichroism is
0
H caused by a selection-rule in which E H a transitions are
0.. Ge S
4.21<
0 -
I I
.7 .8 1.9 2.0 2.1
ENERGY, hw 1eV)
Fig. 1. The absorption edge of GeS in the a and b
polarizations, where a and b lie parallel to the principal
(001) cleavage planes. The E a absorption is direct-
allowed, and rises steeply to a sharp free-exciton peak
at 1.7365 eV. The E lb absorption is direct-forbidden,
causing a dramatic dichroism near the absorption thres-
hold.
phonons or (in the case of piezoelectric materials) by
acoustic phonons. If the phonon which is responsible
for a given Urbach edge has an associated electric field,
it will clearly be difficult to determine whether the
edge was caused by electric-field ionization, or by some
other exciton—phonon interaction. If, on the other
hand, a case could be found in which an Urbach edge is
caused by a phonon which has no associated electric
field, this would provide a clear refutation of the
claimed universality of the electric-field ionization
model. GeS provides such an example.
GeS is a member of the group IV—VI compound
semiconductors [6] crystallizing in an orthorhombic
, edge (see Fig. 3).
(D1~) structure with lattice constants [7] a = 4.30 A,
b = 3.65 A, and c = 10.44 A. This is a layered structure
in which Ge and S atoms are arranged in double-layer
planes perpendicularto
coordinated, the c-axis. Each
with 3 strongly-bonded andatom is 6-fold
2 weakly-
bonded neighbors in its own double-layer, and a single
weakly-bonded neighbor in an adjacent double-layer.
The interlayer bonding is sufficiently weak to allow
micaceous cleavage, but not nearly as weak as the inter-
layer bonding in (say) MoS
2 or graphite.
Optical transmission measurements were performed
on samples cleaved (IC) from single-crystal ingots of
undoped GeS grown by vacuum sublimation [10].
Sample thicknesses ranged from 4.2 jim to 8.8 mm. The
Vol. 34, No. 11
cleavage-steps over areas of at least 1 cm2. An infrared
viewer was used to verify that the samples contained no
internal cracks, voids, inclusions, or other visible defects.
After X-ray orientation, to locate the a and b axes, the
samples were carefully mounted in a variable-
temperature cryostat, and transmission measurements
- were performed at T 4.2, 60, 120, 180, and 240 K,
using a Cary model 17 spectrophotometer. The data
were analyzed as described in [9]. Figure 1 shows the
direct-allowed and E II b transitions are direct-forbidden
[9, 11]. The detailed shape of the E lb edge appears to
be determined by energy-dependent matrix elements or
by band non-parabolicity. It is not well described by
any simple theoretical expression for absorption edge-
2,2 shapes and, in particular, does not obey Urbach’s rule.
The E II a edge, on the other hand, obeys Urbach’s rule
quite accurately over at least three orders of magnitude
in a for 4.2 ( T ~ 240 K. This is demonstrated in Figs.
2 and 3. The heavy lines in Fig. 2 are measured absorp-
tion coefficients, and the light lines are linear, corre-
sponding to equation (1), with a focal point at a
1 and E 0 =
1.1 x io~ cm 0 = 1.753 x 0.002 eV. The steep-
ness parameter, a, as deduced from Fig. 2, is plotted as
a function of temperature in Fig. 3. It is noteworthy
that the E Ia Urbach edge appears on the low-energy
tail of the exciton absorption line, while the E lb edge
shows no exciton absorption, and no Urbach behavior.
Similar results (an Urbach edge only in that polarization
for which an exciton is observed) have been reported
for trigonal Se [13] and GeSe [14]. This is strong
evidence for the importance of excitons in the forma-
tion of Urbach edges. A more detailed discussion of the
GeS absorption edge will be presented elsewhere [12].
Here, we shall concentrate on the significance of the
8.7 meV phonon which is responsible for the Urbach
Essentially all of the optical modes of GeS have
been measured by infrared reflectivity and Raman
scattering [8]. One of these, the B2g rigid-layer mode,
1, in excellent
has a measured frequency of 76 ± 5 cm
agreement with the Urbach phonon, hw~= 8.7 ±0.6
meV = 70 ±5 cm1 The nearest (zone-center) phonons
.
above and below the 76 cm1 phonon are at 96 cm1
and 55 cm1 respectively [81. The 76 cm’ B2g mode
,
corresponds to rigid, compressive vibrations of the
layers along the c-axis. The space-group of the GeS
lattice (D~)includes a center of inversion symmetry
midway between the double-layer planes. Thus, rigid-
layer vibrations normal to the planes cannot produce
any electric field. This is consistent with the fact that
 Vol. 34, No. 11 ON THE PHYSICAL BASIS FOR URBACH’S RULE 893

6 P - 1.5 -

1~i05 ~I1I) - 11.0 -

E - /1/ 1.45
// /!~ b 8.7x10
3eV

~io~ - -

/ 05-

IL
LU
0
0
iO~-
O 0 I I I I I

I— - 0 60 120 180 240 300

T(°K)—’
~ 102 - Fig. 3. A plot of the Urbach parameter, a vs temperature.
The a values were obtained from the exponential fits
- shown in Fig. 2. The solid line was calculated using
120K equation (2) with a
0 = 1.45 and hc~= 8.7 meV. This
figure together with Fig. 2, confirms Urbach behavior
10 240K 80K 60K - for the E 11 a absorption edge in GeS. Estimated fitting
4.2K errors are ±0.05 for a0 and ±0.6 meV for hwy.

observation of Urbach behavior at 4.2 K is taken as

I I
1.5 1.55 .6 .65 1.7
hweV —‘
Fig. 2. Absorption coefficient vs photon energy for GeS
in the Ella polarization at various temperatures. The
heavy lines are experimental. The light lines are linear,
coinciding with the measured data over approximately
3 decades of and extrapolating to a focal-point at
a0 = 1.1 x 10 cm’ ,E0 = 1.751 eV.
the B2g rigid-layer mode is Raman active, but not infra-
red active [8]. The (anisotropic) exciton radius in GeS
is estimated to be 50 A [12] Since exciton—phonon
. the GeS Urbach edge has been identified as one which
interactions typically involve phonons with wavevectors
q ~ 1 /a~,where a~is the exciton radius, the GeS
exciton
1. Thisinteracts primarily
is a small fractionwith phonons
(<0.07) of distance
of the q ~ 0.02to
fiC nearest zone boundary. We therefore conclude that
the
the Urbach edge in GeS is caused by zone-center
phonons having no associated electric fields. This finding
is inconsistent with the claim [5] that electric field
.75 1.8 evidence that self-trapping has occurred [4]. Further
evidence for self-trapping of excitons in GeS is provided
by the very nature of the 8.7 meV phonon. Fivaz and
Mooser [15] have proposed a model for electron—
phonon interactions in layer-type semiconductors, in
which fluctuations in the interlayer spacings give rise to
a strong deformation—potential interaction. They show,
further, that this interaction can easily lead to self-
trapping of electrons, and may explain the dominant
p-type conductivity of many layer-type semiconductors
(including Ge S). The 8.7 meV phonon which produces
causes precisely the rigid inter-layer vibrations en-
visioned by Fivaz and Mooser. We therefore conclude
that self-trapping
electrons) probablyof occurs
excitons
in (via
GeS,self-trapping
and that, at of
least for
layer-type compounds, Sumi and Toyazawa’s criterion
for self-trapping is slightly conservative.
Acknowledgements This work was supported, in part,
—

ionization of excitons is the universal mechanism by NATO (Research Grant No. 1125). The Research
responsible for Urbach edges. Corporation (Research Grant No. 8098), and the Univer-
sity of Wisconsin Graduate School.
Our results are in good agreement with the theory
of Sumi and Toyazawa, with the following intriguing
quantitative discrepancy. According to Sumi and REFERENCES
Toyazawa, a value of a
0 = 1.45 is too large to allow 1. F. Urbach,Phys. Rev. 92, 1325 (1953).
self-trapping of excitons. On the other hand, the 2. M.V. Kurik, Phys. Status Solidi (a) 8, 9 (1971).
894 ON THE PHYSICAL BASIS FOR URBACH’S RULE Vol. 34, No. 11
3. D.L. Dexter & R.S. Knox, Excitons. Wiley, New 9. J.D. Wiley, A. Breitschwerdt & E. Schönherr,
York (1965). Solid State Commun. 17, 355 (1975).
4. H. Sumi & Y. Toyazawa, J. Phys. Soc. Japan 31, 10. E. Schönherr & W. Stetter, J. Cryst. Growth 30,
342 (1971). 96(1975).
5. J.D. Dow & D. Redfield, Phys. Rev. B5, 594 11. R. Eymard & A. Otto, Phys. Rev. Bl6, 1616
(1972). (1977).
6. N.Kh. Abrikosov, V.F. Bankina, L.V. Poretskaya, 12. J.D. Wiley, D. Thomas, E. Schönherr & A.
L.V. Shelimova & E.V. Skudnova, Semiconducting Breitschwerdt (to appear in J. Phys. Chem. Solids).
II--- VI, IV-- VI, and V— VI Compounds, Cli. II 13. G.G. Roberts, S. Tutthasi & R.C. Keezer, Phys.
Plenum Press, New York (1969). Rev. 166, 637 (1968).
7. The notation used here for the crystallographic 14. M.P. Lisitsa, A.P. Zakharchuck, S.F. Terekhova,
axes follows that of [6] Other authors have
. G.G. Tsebulya, L.K. Mladov & S.M. Todorov,
followed different conventions, and care should Phys. Status Solidi(b) 75, K51 (1976).
be used in comparing results. 15. R. Fivaz & E. Mooser,Phys. Rev. 136, A833
8. J.D. Wiley, W.J. Buckel & R.L. Schmidt, Phys. (1964).
Rev. B13, 2489 (1976).