Professional Documents
Culture Documents
Transmission Photoacoustic Spectroscopy of Flash-Evaporated
Transmission Photoacoustic Spectroscopy of Flash-Evaporated
www.elsevier.com/locate/tsf
Abstract
Photoacoustic spectroscopy (PAS) has proved to be an effective technique for the evaluation of inherent defect population in a wide range of
materials for various applications. This paper demonstrates the use of this technique in transmission mode and hence evaluates the optical
properties of flash evaporated CuIn0.75Ga0.25Se2 (CIGS) thin films. Both the photoacoustic and transmission spectra were recorded at room
temperature using high resolution near-infrared of the gas-microphone type PAS which revealed a very broad transmission region (about 300
meV) near the fundamental band edge in the as-grown CIGS thin films due to the presence of several shallow defect levels. The post-deposition
heat treatment of the samples under selenium ambient followed by annealing under inert and forming gas ambient showed significant changes in
the behavior of the PAS spectra particularly near the fundamental band edge. The absorption coefficient has been derived from these spectra to
determine the band gap energy values and the activation energies for several defect related energy levels. Using the PAS, the energy band gap
values were in the range of 1.197 to 1.202 eV. The optical transmission spectra were also recorded from the spectrophotometer. The transmission
data was used to determine the energy band gap values which were calculated to be in the range of 1.159 to 1.194 eV. These values were found to
be in good agreement to each other as well as to those reported in the literature.
D 2005 Elsevier B.V. All rights reserved.
contributes to the measured spectra [4]. This technique relies cell geometry and microphone response by normalizing the
on the generation of an acoustic signal when a sample, placed output signal of the specimen to that of fine carbon black
inside a closed cell, is illuminated with chopped monochro- powder. The optical transmission spectra were also obtained
matic light. The measured acoustic signal, detected by a from a Hitachi 2000 double beam spectrophotometer in the
sensitive microphone, is a function of the optical absorption wavelength range from 800 to 1800 nm.
properties of the sample. PAS has proved to be an effective Various post-deposition annealing regimes were adopted:
technique for evaluating not only the standard optical annealing in vacuum, annealing in an inert gas, annealing in an
properties but also inherent shallow and deep level defects in ambient of 9 : 1 mixture of N2 : H2 and annealing in a selenium
a wide range of materials [5]. For single crystals or thick films, atmosphere. The influence of various temperatures in the range
where the thermal diffusion length is less than the thickness of 200 to 500 -C was investigated for 30 min. The two stage
the sample, the PAS signal is proportional to the absorption annealing process was then employed to optimize the general
coefficient. However for thin films, where the thermal diffusion properties in particular the optical properties of the CIGS thin
length is greater than the film thickness, the effects of the films.
substrate and the multiple reflections within the film have to be
taken into account. 3. Results and discussions
In this paper we have demonstrated that the same PAS
system can be used both in the photoacoustic and transmission Fig. 1 show the photoacoustic signal plotted against the
modes to gain detailed information including deep and shallow wavelength for standard carbon black powder from which all of
defect level states and the band gap energies of semiconductor the other spectra of single crystal and thin film samples were
materials and in particular CIGS compound in thin film form. normalized. The average particle size of the standard carbon
The samples were deposited using a modified flash evaporation black powder was in the range of 1 –10 Am and the spectra was
technique onto glass substrates. The effect of deposition measured at room temperature with a modulating frequency of
temperature and the post-deposition annealing under various 112 Hz in the wavelength range of 800 to 1800 nm.
ambient at different temperature and times on the optical Fig. 2 illustrates the normalized transmission photoacoustic
properties of CIGS thin film samples have been examined. We spectra of as-grown CIGS thin films (approximately 1 Am
have shown that by using a PAS system in transmission mode thick) as a function of photon energy hr. It can be seen that all
we can produce identical results to those obtained using the spectra are dominated by defect levels both near the
standard PAS. fundamental band edge as well as in the low photon energy
range. Due to the presence of one or more shallow defect levels
2. Experimental close to the band edge, the nature of the fundamental band-to-
a flat response at all wavelengths [4]. The sample has the effect
of filtering out certain wavelengths so that when compared 4
with the PA spectra obtained with that of the standard carbon
black spectra the missing values of wavelengths can be
ascertained and related to absorption levels in the sample.
3
Using this set up, measurements were registered for the
flash-evaporated CIGS thin films. Thin film samples were
deposited by the modified flash evaporation technique onto
2
glass substrates; the temperature of selected samples was varied
between room temperature and 200 -C [6]. These were
compared with spectra recorded from clean virgin glass
substrates and with spectra recorded using the standard PAS 1
technique. All the measurements were recorded at room
temperature using a chopping frequency of 112 Hz and a
photon energy range of 0.7 to 1.4 eV. The final output was read 0
800 1000 1200 1400 1600 1800 2000
by a computer which was also configured to control the
Wavelength (nm)
monochromator. The photoacoustic spectra were corrected to
account for the spectral distribution of the optical system, the Fig. 1. Photoacoustic spectrum of the standard carbon black powder.
E. Ahmed et al. / Thin Solid Films 515 (2006) 239 – 244 241
0.60
0.40
Uncoated glass slide
CIGS-1
CIGS-2
0.20 CIGS-3
0.00
0.60 0.80 1.00 1.20 1.40
Photon Energy hv (eV)
Fig. 2. Normalized transmission photoacoustic spectra of virgin glass slide and CIGS thin films.
band transition is difficult to establish as this region is extended sition throughout the depth of the film [6]. In contrast the
across a 300 meV photon energy range. This was not surprising normalized transmission photoacoustic spectra from the CIS
as the photoacoustic signal depends strongly on the structure and CIGS single crystal samples always showed the region
and uniformity of the sample. In the case of thin films grown near the fundamental band edge free from defect states and
by flash-evaporation in this study the existence of voids, grain therefore displayed a steep transition edge [7].
boundaries and deviation from the stoichiometry [6] gave rise In order to confirm that these transition peaks in the
to these defects. The structural properties of the as-grown normalized transmission photoacoustic spectra are characteris-
samples showed a preferred orientation in the form of a less tic of CIGS thin films a cleaned virgin glass slide of 800 Am
intense and relatively broad single <112> diffraction line and thickness was also analyzed using the same setup and for
the grain size was found to be less than 0.2 Am. The respective comparison the normalized transmission photoacoustic ampli-
random RBS spectra of these samples further confirmed the tude signal recorded in the same photon energy range is plotted
observed PAS results as they revealed a non-uniform compo- in Fig. 2 along with the spectra of various thin films. It can be
0.80
0.60
0.40
CIGS-4 (annealed)
0.20 CIGS-5 (annealed)
0.00
0.60 0.80 1.00 1.20 1.40
Photon Energy hv (eV)
Fig. 3. Normalized transmission photoacoustic spectra of two stage annealed CIGS thin films.
242 E. Ahmed et al. / Thin Solid Films 515 (2006) 239 – 244
15
of indium tin oxide (ITO) prepared by electron beam
evaporation technique [8]. These observations support the idea
CuIn0.75Ga0.25Se2 Thin Films that the transitions in the low photon energy range of the
normalized transmission photoacoustic amplitude signal of
12 CIGS thin film are characteristic of the sample since they do
not appear in the photoacoustic amplitude spectrum of the
virgin substrates.
The as-grown CIGS thin films when annealed under
maximum selenium vapour pressure at 300 -C for 120 min
(αhv)2 (10+4 cm-1 eV)2
1.00
Transmission PAS
0.80
Normalized Signal
E1 E2 E3 E4 E5 E6 E7
0.60
0.40
PAS
0.20
0.70 0.80 0.90 1.00 1.10
Photon Energy hv (eV)
Fig. 5. Comparison of PAS and TPAS spectra of deep level transitions in CIGS thin films.
E. Ahmed et al. / Thin Solid Films 515 (2006) 239 – 244 243
70
CuIn0.75Ga0.25Se2 Thin Films
60 200°C
100°C
50
Transmission (%)
40
25°C
30
20
10
0
800 1000 1200 1400 1600 1800
Wavelength (nm)
Fig. 6. Transmission spectra of CIGS thin films deposited at different substrate temperature.
sition towards that of the starting material. This indicates that 0.895, 0.96 and 1.035 eV, respectively. In the case of
the annealing process has resulted in the redistribution of the transmission photoacoustic spectra these peaks are better
atoms and many of the defect states are annealed out. resolved indicating sharp transitions at their respective photon
The absorption coefficient calculated from the photoacous- energy values. However, in photoacoustic spectra the transi-
tic spectra of annealed CIGS thin films was used to determine tions are less pronounced and some of the peaks (e.g. E2 and
the band gap energies from the room temperature plots of E3) appear to be a combination of two small peaks whose
(ahr)2 against the photon energy hr and the representative
curves are shown in Fig. 4. By the extrapolation of curves 10
down to zero level the measured energy band gap value CuIn0.75Ga0.25Se2 Thin Films
(å 1.202 eV) of sample annealed in forming gas was found to
25°C
be slightly higher than that observed (å 1.197 eV) from argon 100°C
200°C
annealed thin film. The observed higher value of the band gap 8
energy can be due to several reasons including the measure-
ment error (å 0.5%), compositional variations through the
depth of the sample or due to the appearance of some defect
(αhv)2 (10+4 cm-1 eV)2
photon energy values are very close to each other. A similar 4. Conclusions
type of spectrum was observed from CIGS single crystals [10]
however the intensity of the peaks was less in single crystal We have demonstrated that a photoacoustic spectrometer in
compared to those observed in thin films. Since the peak the transmission mode can produce identical results to those
positions are same in single crystal and thin films their origin is obtained using the standard photoacoustic system or spectro-
also likely to be the same and can be attributed due to photometer, with the analysis of the data from thin film
transitions from one defect state to another or transitions samples is greatly simplified. An added advantage of this new
between defect states and the conduction or valence bands [11]. configuration is its ease of operation; in the standard set up
Fig. 6 shows the transmission spectra of CIGS thin films each new sample has to be resealed before spectra can be
deposited at various substrate temperatures. The spectrum of obtained. In transmission mode, the cell is initially filled with
the film deposited at room temperature is smeared with a carbon black and then sealed for life. Each new sample,
maximum transmission level of approximately 40%. However regardless of its size, is simply placed on top of the
the spectra representing films deposited at higher substrate photoacoustic cell quartz window and the spectra obtained.
temperature were sharp with a higher maximum level of The calculated energy band gap values of flash-evaporated
transmission approaching to 60%. These changes were found CIGS thin films were found to be in good agreement to each
to be in good agreement with enhancements in the physical other as well as with those reported in the literature.
properties of the samples deposited at higher substrate
temperatures or annealed after deposition [9,12]. This strongly References
indicates that at higher temperatures processes involving
redistribution of atoms and the reduction of defects are [1] K. Ramanathan, M.A. Contreras, C.L. Perkins, S. Asher, F.S. Hasoon, J.
occurring more rapidly. Keane, D. Young, M. Romero, W. Metzger, R. Noufi, J. Ward, A. Duda,
Prog. Photovolt. 11 (2003) 225.
In determining the respective energy band gap values of the [2] J. Manifacier, J. Gasiot, J. Fillard, J. Phys. E: Sci. Instrum. 9 (1976) 1002.
flash-evaporated CIGS thin film samples, (ahr)2 was plotted [3] R. Swanepoel, J. Phys. E: Sci. Instrum. 17 (1984) 896.
against the photon energy and Fig. 7 shows the representative [4] A. Rosencwaig (Ed.), Photoacoustics and Photoacoustic Spectroscopy,
curves obtained. Extrapolation of the curves down to a zero John Wiley & Sons, New York, 1980.
(ahr)2 level gave energy value of 1.159 eV for the film [5] E. Ahmed, A. Zegadi, A.E. Hill, R.D. Pilkington, R.D. Tomlinson, Thin
Solid Films 268 (1995) 144.
deposited at room temperature. This band gap value increased [6] E. Ahmed, A.E. Hill, R.D. Pilkington, R.D. Tomlinson, J. Phys. D: Appl.
as the substrate temperature was raised. The resultant shift was Phys. 26 (1993) 1787.
1.175 eV for 100 -C and 1.194 eV for 200 -C. The band gap [7] E. Ahmed: PhD Thesis, University of Salford, UK, (1995).
value of 1.194 eV for films deposited at 200 -C substrate [8] I.E. Elfallal: PhD Thesis, University of Salford, UK, (1993).
[9] E. Ahmed, A. Zegadi, A.E. Hill, R.D. Pilkington, R.D. Tomlinson,
temperature is in good agreement with the reported value
J. Mater. Sci., Mater. Electron. 7 (1996) 213.
[5,11]. The low band gap value for the film deposited at room [10] E. Ahmed, A. Zegadi, A.E. Hill, R.D. Pilkington, R.D. Tomlinson, W.
temperature appears to be linked to structural imperfections. Ahmed, J. Mater. Process. Technol. 77 (1996) 260.
The films deposited by such methods contain additional states [11] A. Zegadi, M.A. Slifkin, M. Djamin, R.D. Tomlinson, H. Neumann, Solid
in the vicinity of the band edge due to structural defects within State Commun. 83 (1992) 587.
the grains and grain boundaries causing a smaller effective [12] K.L. Chopra (Ed.), Thin Film Phenomina, Krieger, New York, 1979.
band edge. This behavior is consistent with results obtained
from a variety of analytical techniques including scanning
electron microscopy, XRD and Raman spectra [9].