Bull. Mater. Sci., Vol. 24, No. 3, June 2001, pp. 297–300. © Indian Academy of Sciences.
Grain size dependent optical band gap of CdI2 films
PANKAJ TYAGI and A G VEDESHWAR*
Department of Physics and Astrophysics, University of Delhi, Delhi 110 007, India
MS received 26 September 2000
Abstract. The thermally evaporated stoichiometric CdI2 films show good c-axis alignment normal to substrate
plane for film thickness up to 200 nm. The optical absorption data indicate an allowed direct interband transi-
tion across a gap of 3⋅⋅6 eV in confirmation with earlier band structure calculations. However, part of the
absorption data near band edge can be fitted to an indirect band gap of 3 eV. The dependence of band gap on
film thickness (> 200 nm) can be explained qualitatively in terms of decreasing grain boundary barrier height
with grain size.
Keywords. Optical properties; thin films; halides.
1. Introduction isostructural compound to many halides and chalco-
genides.
The cadmium iodide is a layer structured compound with
a hexagonal unit cell, in which each hexagonal sheet of
2. Experimental
Cd atoms sandwiched between two similar sheets of I
atoms, the Cd atoms being octahedrally coordinated (Hul-
The CdI2 thin films were grown on glass substrates at
liger et al 1976). The stacking of these three sandwiched
room temperature by thermal evaporation using a moly-
sheets form the three-dimensional compound. CdI2 is a
bdenum boat at a pressure better than 10–6 Torr. The start-
well known polytype material having as high as 200
ing material was analar grade powder, which was
polytypes out of which only very few are commonly occur-
pelletized for evaporation. The quartz crystal thickness
ring (Trigunayat 1991). The possibility of different
monitor was used for monitoring the film thickness during
stacking sequences of three layer sandwiches along the
evaporation. Uniformity and thickness of the film was
third direction due to the weak bonding between layers
subsequently checked and confirmed by DEKTAK IIA
leads to polytype structure. The reported studies on CdI2
surface profiler. The deposition rate was slow (2–4 nm/s)
concerning structure and optical properties are scattered.
as higher deposition rates led to nonuniform film growth
The optical absorption measurements carried out on
and low sticking. Films up to 100 nm were completely
single crystal samples were fitted to an indirect gap of
transparent and become translucent for higher thicknesses.
3⋅2 eV (Greenaway and Nitsche 1965) and many sub-
The films of thickness below 50 nm were nonuniform,
sequent reports indicate an indirect band gap of similar
discontinuous and above 600 nm peel off the substrate.
values (Lee et al 1969; Takemura et al 1975). However,
Small pieces of 0⋅5 × 0⋅5 cm were cut for various mea-
the band structure calculations show the presence of both
surements so as to carry out various analyses on the same
direct and indirect band gaps, the difference between the
film. The structural studies of the films were carried out
two is only about 0⋅3–0⋅6 eV (McCanny et al 1977;
by X-ray diffraction analysis (PHILIPS X-Pert model
Bordas et al 1978; Robertson 1979; Coehoorn et al 1985;
PW-1830 generator diffractometer). The film composition
Pollini et al 1986). The electronic structure determined by
was found to be stoichiometric as determined by EDAX
optical reflectivity on CdI2 crystals shows allowed direct
(JEOL–840). The optical absorption measurements were
transition of 4⋅3 eV at 30 K and forbidden direct transi-
carried out in UV/VIS region using Shimadzu UV-260
tion of 3⋅8 eV at 300 K (Pollini et al 1986). However,
spectrophotometer.
there are very little studies on CdI2 films regarding their
optical properties (Brahms 1965; Nakagawa et al 1987;
Kondo et al 1994, 1998a). Therefore, we thought it worth-
while to carry out the optical studies on CdI2 films in 3. Results and discussion
order to understand the material better as it is a basic
All CdI2 thin films grown at room temperature were poly-
crystalline and stoichiometric (as per X-ray diffraction
*Author for correspondence and EDAX analysis) irrespective of the film thickness.
297
298 Pankaj Tyagi and A G Vedeshwar
However, the earlier report (Yu 1967) on the structure of where A is the absorbance and t the film thickness,
CdI2 films shows that thinner films (< 100 nm) are amor- neglecting the reflection coefficient, which is negligible
phous and thicker films tend to crystallize, while as per and insignificant near the absorption edge. The depend-
Kondo et al (1998b), the CdI2 films grown at substrate ence of α on hν near the band edge is shown in figure 3
temperature < 300 K are amorphous and those either for 250 nm thick film. In a crystalline or polycrystalline
grown or annealed above 300 K are polycrystalline. Our material both direct or indirect optical transitions are
X-ray diffraction data is in agreement with ASTM No. possible depending on the band structure of the material.
(12–574) and with the earlier report (Bringans and Liang Assuming parabolic bands, the relation between α and Eg
1981). The XRD analysis shows a high degree of orienta- for a direct transition is given by (Mott and Davis 1979)
tion with the basal plane parallel to substrate and c-axis
normal to substrate plane as indicated by the total absence αhν = constant (hν – Eg)n, (2)
of (001), (101) or (111) reflections. Earlier studies (Yu
1967; Bringans and Liang 1981) have also observed and for indirect transition by (Mott and Davis 1979)
this kind of c-axis alignment in CdI2 films. However, a
small amount of (111) reflection and other non zero l- A(hν − Eg + Ep ) n B(hν − Eg − Ep ) n
reflections were observed for higher film thicknesses than αhν = + , (3)
exp(θ D / T ) − 1 1 − exp(−θ D / T )
160 nm. This indicates a slight misalignment among the
grains. We have determined the average grain size from
where Ep is the phonon energy assisting the transition, θD
the full width at half maximum (FWHM) of the most
the Debye temperature and A, B are constants. For a direct
intense peak using the Scherrer formula (Cullity 1978) as
transition n = 1/2 or 3/2 depending on whether the transi-
a function of film thickness as shown in figure 1. The
tion is allowed or forbidden in quantum mechanical sense.
inset of figure 1 shows for comparison the grain size
Similarly, n = 2 or 3 for indirect allowed and forbidden
dependence on thickness as determined by SEM.
transition respectively. The usual method of determining
The optical absorption measurements were carried out
band gap is to plot a graph between (αhν)1/n and hν and
on a large number of samples of various thicknesses.
look for that value of n which gives best linear graph in
Figure 2 shows some representative optical absorbance
the band edge region. Obviously, there will be a single
curves as a function of wavelength for different thick-
linear region in direct transition and two linear portions in
nesses. The wavy nature of the absorbance away from the
indirect transition as can be seen from (2) and (3). We
fundamental absorption edge is due to the interference
plotted (αhν)1/n vs hν for CdI2 films for various thick-
fringes arising from the substrate–film and film–air inter-
nesses and the best fit was obtained for n = 1/2 indicating
faces. The fringes smoothen out as film thickness increases.
a direct allowed transition as shown for 250 nm thick film
The steep rise in the absorbance near the absorption edge
in the inset A of figure 3.
hints at a direct type transition. The absorption coefficient
In earlier reports (Greenaway and Nitsche 1965; Lee
α was calculated by
et al 1969), the absorption data on single crystals were
fitted to n = 2 to obtain an indirect gap of 3⋅2 eV by the plot
α = (2⋅303A)/t, (1)
of (α)1/2 vs hν. Using their data of α vs hν, we have found
that the best fit is obtained for n = 1/2, i.e. a direct transi-
tion. A peak corresponding to a direct band to band tran-
Figure 1. Grain size variation with thickness of CdI2 films
determined from the full width at half maximum of most intense
X-ray diffraction peak using Scherrer formula. Inset shows
similar variation as determined from scanning electron micro- Figure 2. The optical absorption spectra for different thick-
graph. nesses of CdI2 films.
Optical band gap of CdI2 films 299
sition (at room temperature) has been identified in the size as shown in figure 5 is exactly similar to its thick-
reflectivity spectra at 3⋅8 eV (Greenaway and Nitsche ness dependence and indicates the decreasing barrier
1965) and at 4⋅3 eV (at 30 K) (Pollini et al 1986). The height with increasing grain size. The variation of grain
energy band structure calculations (McCanny et al 1977; boundary barrier height with grain size is given by (Slater
Bordas et al 1978; Robertson 1979; Bringans and Liang 1956)
1981; Coehoorn et al 1985; Pollini et al 1986) clearly
show the existence of both direct and indirect band gaps Eb = Ebo + C(X – fd)2, (4)
of similar magnitudes, the difference between the two is
about 0⋅3–0⋅6 eV. However, the band structure (Coehoorn where Ebo is the original barrier height, C is a constant
et al 1985) and band gaps (Slater 1956) of CdI2 are shown depending on the density of charge carriers and dielectric
to be insensitive to its polytypes. Therefore, we can constant of the material, X the barrier width (20–30 nm),
realize that both indirect and direct band gaps exist in d the grain size and f is a fraction of the order of 1/15 to
CdI2 and they are separated by just 0⋅3–0⋅6 eV. Since 1/50 depending on the charge accumulation and carrier
indirect gap is just less than direct gap, it lies near the concentration. We can estimate the change in barrier
onset of direct gap and can hardly be noticed in (αhν)2 height as a function of grain size in CdI2. We take
vs hν plot of much dominated direct transition. Therefore, X = 20 nm as the average barrier width, f = 1/15 consider-
the part of the optical absorption data near the knee ing the low carrier concentration of CdI2 and the grain
or tail of the direct absorption edge have to be size as a function of thickness from figure 1. Thus, the
replotted as (αhν)1/2 vs hν to determine indirect gap as calculated variation factor (X – fd)2 as a function of grain
shown in the inset B of figure 3. However, we still believe size is compared qualitatively with the experimentally
that the optical absorption data reveals clearly a direct observed band gap variation with grain size in figure 5.
band gap showing the best fit to n = 1/2. Thus determined The striking agreement indicates the dominance of barrier
value of Eg (indirect) agrees well with earlier experimen- height contribution to the observed band gap variation
tal results as well as band structure calculations. Our with thickness. The band gap variation with either thick-
value of Eg (direct) of 3⋅6 eV determined for thicknesses ness or grain size is nominal for film thickness < 250 nm.
≤ 250 nm agrees well with the predicted value of 3⋅8 eV This could be due to the better c-axis alignment of the
from band structure calculations. grains as observed in XRD for these thicknesses. As film
However, both types of band gaps showed thickness thickness becomes larger, the misalignment among the
dependence as shown in figure 4. In general, thickness grains starts leading to grain boundary structure. There-
dependence of band gap can arise due to one or combined fore, the data for higher thickness agree better with the
effect of the following causes: (i) a large density of dis- behaviour of grain boundary barrier height with grain
locations, (ii) quantum size effect and (iii) the change in sizes. It can be noted that the only earlier study focussed
barrier height due to change in grain size in polycrystal- on the structure of films was carried out in the range 5–
line films. However, first one looks reasonable cause in 100 nm and showed good c-axis alignment with which our
the present case with small contributions from dislocation results agree very well. Therefore, we can clearly realize
density as well. As the thickness of the films in the pre- that the thickness dependence of optical band gap is
sent study is quite large, the quantum size effect can com- dominated by the grain size dependence of grain boundary
pletely be ruled out. The decreasing band gap with grain barrier height for film thicknesses > 200 nm.
Figure 3. The absorption coefficient α as a function of hν.
The inset A shows (αhν)2 vs hν plot for the determination of Eg
(direct) while inset B shows (αhν)1/2 vs hν plot for the Figure 4. Thickness dependence of Eg (direct). Inset shows
determination of Eg (indirect). similar dependence of Eg (indirect).
300 Pankaj Tyagi and A G Vedeshwar
like to thank Mr P C Padmakshan, Department of Geo-
logy, University of Delhi, Delhi, for carrying out X-ray
diffraction measurements.
References
Bordas J, Robertson J and Jakobsson A 1978 J. Phys. C: Solid
State Phys. 11 2607
Brahms S 1965 Phys. Lett. 19 272
Bringans R D and Liang W Y 1981 J. Phys. C: Solid State
Phys. 14 1065
Coehoorn R, Sawatzky G A, Haas C and deGroot R A 1985
Phys. Rev. B31 6739
Cullity B D 1978 Elements of X-ray diffraction (California,
Figure 5. The grain size dependence of band gap along with USA: Addison-Wesley) 2nd Ed. p. 102
the estimated grain boundary barrier height variation factor (X – Greenaway D L and Nitsche R 1965 J. Phys. Chem. Solids 26
fd)2 with grain size. 1445
Hulliger F, Levy F and Reidel D 1976 in Structural chemistry
of layer-type phases (Dordrecht, Holland: Ed. publishing
4. Conclusions
company) pp 31–34 and 275
Kondo S, Matsuoka S and Saito T 1994 Phys. Status Solidi (b)
The cadmium iodide films show good c-axis alignment 186 K77
normal to substrate plane for film thicknesses up to about Kondo S, Matsuoka S and Saito T 1998a Phys. Status Solidi (a)
200 nm and a slight misalignment sets in for higher thick- 165 271
nesses as observed by XRD. The optical absorption data Kondo S, Suzuki T and Saito T 1998b J. Phys. D: Appl. Phys.
fit best to direct band to band type transition indicating a 31 2733
direct band gap in conformity with band structure calcula- Lee P A, Said G, Davis R and Lim T H 1969 J. Phys. Chem.
tions. However, a smaller indirect band gap can also be Solids 30 2719
determined from part of absorption data near the band McCanny J V, Williams R H, Murray R B and Kemeny P C
edge for the purpose of comparison with earlier analyses 1977 J. Phys. C: Solid State Phys. 10 4255
Mott N F and Davis E A 1979 Electronic processes in non-
of absorption data as well as the band structure calcula-
crystalline materials (Oxford: Clarendon Press) pp. 273–
tions. The decreasing band gap with film thickness can be
274
attributed to the grain size dependent grain boundary Nakagawa H, Yamada T, Matsumoto H and Hayashi T 1987 J.
barrier height, at least for film thickness > 200 nm, Phys. Soc. Jap. 56 1185
consistent with XRD analysis. Pollini I, Thomas J, Coehoorn R and Hass C 1986 Phys. Rev.
B33 5747
Robertson J 1979 J. Phys. C: Solid State Phys. 12 4753
Slater J C 1956 Phys. Rev. 103 1631
Acknowledgements
Takemura Y, Komatsu T and Kaifu Y 1975 Phys. Status Solidi (b)
72 K87
We would like to thank Dr P Arun, Vinod Kumar Paliwal Trigunayat G C 1991 Solid State Ionics 48 3
and Hina for their helpful discussions. Also, we would Yu R M 1967 Philos. Mag. 16 1167
