Physicochemical characterization of point defects in fluorine doped tin oxide films
Transcript of Physicochemical characterization of point defects in fluorine doped tin oxide films
Physicochemical characterization of point defects in fluorine doped tin oxide filmsFikry El Akkad and Sudeep Joseph Citation: Journal of Applied Physics 112, 023501 (2012); doi: 10.1063/1.4736798 View online: http://dx.doi.org/10.1063/1.4736798 View Table of Contents: http://scitation.aip.org/content/aip/journal/jap/112/2?ver=pdfcov Published by the AIP Publishing
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
93.180.53.211 On: Tue, 18 Feb 2014 13:57:34
Physicochemical characterization of point defects in fluorinedoped tin oxide films
Fikry El Akkada) and Sudeep JosephPhysics Department, Kuwait University, P.O. Box 5969, Safat 13060, Kuwait
(Received 12 February 2012; accepted 9 June 2012; published online 16 July 2012)
The physical and chemical properties of spray deposited FTO films are studied using FESEM,
x-ray diffraction (XRD), x-ray photoelectron spectroscopy (XPS), electrical and optical
measurements. The results of XRD measurements showed that the films are polycrystalline (grain
size 20–50 nm) with Rutile structure and mixed preferred orientation along the (200) and (110)
planes. An angular shift of the XRD peaks after F-doping is observed and interpreted as being due
to the formation of substitutional fluorine defects (FO) in presence of high concentration of oxygen
vacancies (VO) that are electrically neutral. The electrical neutrality of oxygen vacancies is
supported by the observation that the electron concentration n is two orders of magnitude lower
than the VO concentration calculated from chemical analyses using XPS measurements. It is shown
that an agreement between XPS, XRD, and Hall effect results is possible provided that the degree
of deviation from stoichiometry is calculated with the assumption that the major part of the bulk
carbon content is involved in O-C bonds. High temperature thermal annealing is found to cause an
increase in the FO concentration and a decrease in both n and VO concentrations with the increase
of the annealing temperature. These results could be interpreted in terms of a high temperature
chemical exchange reaction between the SnO2 matrix and a precipitated fluoride phase. In this
reaction, fluorine is released to the matrix and Sn is trapped by the fluoride phase, thus creating
substitutional fluorine FO and tin vacancy VSn defects. The enthalpy of this reaction is determined
to be approximately 2.4 eV while the energy of formation of a VSn through the migration of SnSn
host atom to the fluoride phase is approximately 0.45 eV. VC 2012 American Institute of Physics.
[http://dx.doi.org/10.1063/1.4736798]
I. INTRODUCTION
The transparent conducting oxides (TCO) are wide band
gap semiconductors that have the remarkable property of
possessing both high metallic-type electrical conductivity
and high insulator-type optical transmittance in the visible
part of the spectrum. This group of oxides has wide applica-
tions in the field of optoelectronic devices such as solar cells,
flat panel displays, gas sensors, and heat reflecting
mirrors.1–4 SnO2 (TO) is one of the prominent members of
this group that has been extensively studied particularly in
relation with doping effects in view of improving its per-
formance as TCO. Fluorine doping is known to have a favor-
able impact on the performance because it enhances the
electrical conductivity without significantly reducing the
transmittance.5–8 Therefore, a large amount of work has
been devoted to the study of the physical properties of FTO
thin films prepared using various techniques including mag-
netron sputtering,9 chemical vapor deposition,10 sol-gel,11
electron beam evaporation,12 and spray pyrolysis (SP).13–16
Despite the work already conducted, there are few litera-
ture investigations on the relationship between the physical
and the chemical properties of FTO films.17,18 Consequently,
the role of point defects on the optoelectronic properties of
FTO has not been adequately studied. The assignment of
electrical conduction to the substitutional fluorine donor is
well documented,19,20 but the role of intrinsic defects did not
have an equivalent attention.21,22 One of these defects is the
oxygen vacancy VO, which has long been invoked in order to
interpret the electrical properties of undoped as well as
doped tin oxide.18,20–25 However, Singh et al.22 showed that
VO is a deep donor in TO with energy level depth of �1.8 eV
below the conduction band edge, so that it does not contrib-
ute to the electron conduction of the material. Additionally,
evidence was obtained for the creation of compensating
acceptors in FTO with concentrations depending on the fluo-
rine content and on the processing parameters.26,27 As to the
nature of these acceptor defects, some authors proposed in-
terstitial fluorine, but the possibility of an intrinsic defect
such as interstitial oxygen Oi or tin vacancy VSn cannot be
excluded.21,22,26–28 This calls for a revision of our view on
the electrical properties of FTO and on the variation of these
properties with the preparation parameters and post-
preparation heat treatments.
X-ray photoelectron spectroscopy (XPS) is an important
and widely used technique for chemical analysis of materi-
als, but the reported analyses of fluorine in FTO using XPS
are few, essentially because the studied samples did not have
high enough concentration.6,23,24 Until the work of Martinez
et al.18 in 2006, there had been no XPS signal detected for
fluorine in FTO films prepared by SP. Later, more or less
clearer XPS peaks were reported by some authors.19,29,30
Besides the main chemical state identified as being due to
a)Author to whom correspondence should be addressed. Electronic mail:
0021-8979/2012/112(2)/023501/10/$30.00 VC 2012 American Institute of Physics112, 023501-1
JOURNAL OF APPLIED PHYSICS 112, 023501 (2012)
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
93.180.53.211 On: Tue, 18 Feb 2014 13:57:34
the Sn-F bond, the presence of other fluoride states was
suggested.30
In the present work, we report on an investigation of the
morphological, structural, electrical and optical properties of
spray-deposited FTO films with known intrinsic and extrin-
sic dominant point defects content. The defects concentra-
tions are determined via XPS chemical analyses and are
varied using high temperature post-deposition annealing.
This physicochemical study of point defects is aimed at
improving our knowledge of the role in FTO films, which
may provide some guidelines for the control the physical
properties of this type of films for device applications. We
show that the results can be interpreted on the basis of a
point defect model, which assumes the release of fluorine
from a fluoride phase and the creation of tin vacancies during
high temperature annealing. The enthalpies of these reactions
are determined.
II. EXPERIMENTAL
SnO2 thin films were prepared from spray solutions con-
sisting of mixtures of SnCl4: propanol:H2O with molar ratio
1:9:2. Doping with F was accomplished by adding the
desired amount of F in the form of NH4F to the spray solu-
tion. The films were deposited on borosilicate glass sub-
strates heated at a temperature of 450 6 3 �C using IR heater
and temperature controller. Spraying was performed using
N2 as carrier gas (pressure � 2.5 psi) in pulses of duration
�1 s and time interval �40 s in order to allow the substrate
to be reheated to the preparation temperature. With 100
pulses, films of thickness of 230 6 13 nm were obtained
(measured using auto-nulling spectroscopic imaging ellip-
someter type Nanofilm—EP3SE). A glass sprayer type
Thomas Scientific model 1702F61 was used. The distance
between the spray nozzle and the substrate was approxi-
mately 25 cm.
High temperature annealing was performed by placing
the sample into an open-ended Pyrex tube heated inside a
horizontal furnace. A flow of air or pure argon gas was main-
tained during annealing while monitoring the temperature
using a chromel–alumel thermocouple. After an annealing
time of 20 min, the Pyrex tube was pulled out of the furnace
and the sample was left to cool down to room temperatures
while being kept in the annealing atmosphere.
The sample morphology was studied using a field emis-
sion scanning electron microscope (FESEM) type JEOL
JSM—7001F. X-ray diffraction (XRD) measurements were
used to study the structural properties of the films. The spec-
tra were obtained using x-ray diffractometer type Siemens
D5000, which uses the Cu Ka (k¼ 1.5406 A) with Bragg-
Brentano measurements geometry.
X-ray photoelectron spectroscopy (XPS) measurements
were carried out on model Thermo ESCALAB 250Xi spec-
trometer using monochromator with Al Ka radiation
(1486.6 eV) with x-ray spot size 380 lm. The spectral acqui-
sition and processing were carried out by means of A van-
tage V 4.74 data system. The sample was carefully
introduced into the preparation chamber with the sample
holder. It is then degassed until good vacuum was achieved,
then it was transferred into the analysis chamber. The analy-
ses were carried out with the parameters: Analysis chamber
pressure 10�9 Torr, step size 0.1 eV, dwell time 100 ms, and
pass energy of 20 eV. All binding energy (BE) values were
determined with respect to C1s line (284.6 eV) originating
from adventitious carbon. Etching was performed using an
argon ion gun with voltage of 2 kV, current of 2 lA, and ras-
ter size of 2 mm2.
Electrical measurements (resistivity, Hall effect) were
carried out at room temperature using the Van Der Pau
method in MMR technologies type system. For this, four Al
contacts each of area 2 mm2 and thickness 50 nm were de-
posited on the sample surface by thermal evaporation. Leads
to the external circuit were made by soldering Au wires to
the Al contacts using indium. Currents in the range 0.5–5.0
mA and a magnetic field of 0.3 Tesla were used.
Transmission (T) measurements in the wavelength range
200� k� 3000 nm were carried out using a double beam
spectrophotometer type Cary SE (a borosilicate glass sub-
strate was used as reference).
III. RESULTS AND COMMENTS
The fluorine content of the spray solution was raised up
to a ratio F/Sn¼ 80 wt. % whereby an intense XPS signal of
F was observed from the films. This concentration of fluorine
in the solution was used for the preparation of the films stud-
ied here. The concentration in the films, determined from the
ratio of the principal F1s peak intensity to that of the Sn3d5/2
peak, was found to be in the range 2.0–7.8 at. % depending
on the annealing temperature. The samples were divided into
two groups for annealing in air and in argon atmospheres at
temperatures TH¼ 500 �C, 550 �C, and 600 �C for 20 min.
They were then characterized using FESEM, XRD, XPS,
electrical and optical measurements.
A. Morphology
All the films were uniform and free of pinholes. The
FESEM images (Fig. 1) show that the films surfaces consist
of well defined grains of shape and size depending on the
annealing temperature. The average grain size decreases
from �50 nm at 450 �C (as deposited) to �20 nm at 550 �C,
then it increases slightly at TH¼ 600 �C. The packing density
of the films also varies with annealing temperature, showing
a decrease up to TH¼ 550 �C then a slight increase at
TH¼ 600 �C. While the surface of the as-deposited sample
consists of grains of nearly circular shape, the surface of the
annealed samples shows the appearance of an additional type
of grains of elongated shape. The presence of grains of dif-
ferent shapes in FTO films has been observed previously and
correlated with the crystallite orientation as determined from
XRD spectra.5 The appearance of these elongated grains
may have a contribution, among other factors (Sec. III B), to
the observed slight change in the relative intensity
Z¼ I(200)/I(110) of the XRD peaks (Fig. 2) between as-
deposited (Z¼ 0.89) and annealed (Z¼ 0.77) samples
(Fig. 2(d)). The size and shape of grains in FTO films is
known to depend strongly on the technique of preparation
and the processing parameters.5,14,24,26 The observed
023501-2 F. E. Akkad and S. Joseph J. Appl. Phys. 112, 023501 (2012)
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
93.180.53.211 On: Tue, 18 Feb 2014 13:57:34
variations in crystallites size and shape in our films (Fig. 1)
may be related to a variation in the concentration of point
defects and other chemical species created during annealing
(Secs. III C and III D). Similar FESEM images were obtained
for Ar-annealed samples.
B. X-ray diffraction (XRD)
Fig. 2 shows typical XRD spectra for SnO2 films with
and without F-doping. It also shows a spectrum obtained af-
ter annealing at TH¼ 600 �C in air. The spectra show that the
films are polycrystalline with a single phase tetragonal Rutile
structure and a mixed preferred orientation along the (110)
and (200) planes. These results are independent of annealing
temperature/atmosphere. Based on structural factor calcula-
tions, Zaouk et al.31 showed that the intensity ratio
Z¼ I(200)/I(110) in FTO films depends on the site occupied by
fluorine in the SnO2 lattice. For an occupation probability in
the range 0%–100%, Z2 is almost constant with value �0.50
for substitutional site while varying in the range 0.50–1.19
and 0.50–1.34 for interstitial and mixed substitutional-
interstitial sites, respectively. Our results show that
Z2� 0.6–0.8, which suggests that the site occupied by F in
our samples is not uniquely substitutional, in agreement with
the work of Canestraro et al.26 Similar values of Z were
reported by Miao et al.32 on FTO films grown at tempera-
tures between 400 �C and 550 �C using SP method.
The XRD peaks were found to shift to lower values of
Bragg angle h after F-doping. Fig. 2(d) shows this behavior
for the (200) peak. The shift is towards increasing lattice
constant indicating an expansion of the lattice. This lattice
expansion cannot be explained by the replacement of the
host oxygen ion by fluorine in the lattice, since the ionic ra-
dius of F (1.33 A) is smaller than that of O (1.4 A). However,
if our undoped samples contain large concentration of oxy-
gen vacancies, a lattice expansion should occur by the incor-
poration of F into the oxygen vacancies with the creation of
substitutional fluorine FO defects. A similar peak shift
induced by an increase in the growth temperature has been
observed on spray deposited FTO films13 as well as on other
oxides.33,34 The initial vacancies concentration is expected
to be strongly dependent on the technique and parameters of
preparation.
The observed peak shift may also be attributed to the
change in stress in the films. The in-plane stress is propor-
tional to the strain given by (co-c)/co; where co and c are the
lattice constants obtained from the (200) reflection for
undoped and F-doped samples respectively. In our case,
co¼ 2.312 A and c¼ 2.321 A, which yields a strain
of�3.9� 10�3. The negative value of the strain indicates
that the stress is compressive while the magnitude falls in the
range reported for SnO2 and other oxides.13,34,35 A compres-
sive stress implies an expansion of the lattice, which is con-
sistent with the expected creation of FO defects. The
presence of stress has been correlated with the creation of
point defects in SnO2 (Ref. 36) and ZnO (Ref. 35) films.
The grain size, with and without F-doping, calculated
from the FWHM of the (200) peak using the Scherrer equa-
tion37 is found to be in the range 30–45 nm. This range of
values corresponds to the grain dimension in a direction nor-
mal to the substrate. It is of the same order of magnitude as
the planar dimension obtained using FESEM. On the other
hand, the effect of annealing on the broadening of the peaks
was found to be insignificant for temperatures up to 600 �C(Fig. 2(d)). The slight shift of the peaks toward higher Bragg
angle in the annealed samples (Fig. 2(d)) indicates a
FIG. 1. FESEM micrographs of FTO
films; (a) As-deposited, TH¼ 450 �C;
and air-annealed at (b) TH¼ 500 �C, (c)
TH¼ 550 �C, and (d) TH¼ 600 �C.
023501-3 F. E. Akkad and S. Joseph J. Appl. Phys. 112, 023501 (2012)
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
93.180.53.211 On: Tue, 18 Feb 2014 13:57:34
tendency toward lattice contraction following high tempera-
ture annealing. It will be shown later (Secs. III A and IV B)
that this lattice contraction is likely to be due to the creation
of tin vacancies during annealing rather than due to an
increase in the concentration of oxygen vacancies.
It should be noted that the interpretation of the lattice
expansion as being due to the incorporation of fluorine into
an oxygen vacancy implies that the lattice was initially con-
tracted by the absence of the host oxygen atom. According
to Singh et al.,22 this is expected to be the case when oxygen
vacancies are in the neutral state (V0O). In this case, the three
nearest-neighbor Sn atoms relax inward by 2.5%, while for
VþO and V2þO , the relaxations are outward by 5.6% and 10% of
the equilibrium Sn-O bond length, respectively. The compar-
ison between the (200) lattice spacing in our undoped sam-
ples and in the Cassiterite crystal (2.312 A and 2.350 A,
respectively) revealed that the lattice is relaxed inward by
1.6% in our undoped samples indicating that the oxygen
vacancies are neutral.
C. XPS
Figure 3 shows a typical XPS survey spectrum for FTO
films. The main detected elements are O, Sn, F, and C. The
presence of C in all samples is due to contamination from
the environment. Fig. 4 shows a typical depth profile for O,
Sn, F, and C, obtained using argon ion etching of an air
annealed sample. The elements O, Sn, and F show an initial
increase of concentration with depth until an etching time of
about 60 s then a change of less than 8% beyond that time.
The carbon concentration, on the other hand, shows an
FIG. 2. XRD spectra for (a) as-deposited TO (b) as-deposited FTO and (c) air annealed FTO (TH¼ 600 �C); (d) the (200) peak in TO, FTO, and annealed FTO
samples.
FIG. 3. Typical XPS survey spectrum for FTO films with the main peaks
indicated.
023501-4 F. E. Akkad and S. Joseph J. Appl. Phys. 112, 023501 (2012)
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
93.180.53.211 On: Tue, 18 Feb 2014 13:57:34
abrupt decrease with depth until an etching time of about
60 s, then a little decrease up to 360 s where it reaches about
4% of the initial value. The large change in concentration
upon approaching the surface was not observed on as-
deposited films. For example, the atomic percentage of O,
Sn, F, and C in an as-deposited sample show, respectively,
the following values for etching time t¼ 0 and t¼ 60 s:
41.8%, 42.2%; 16.7%,18.1%; 0.70%, 0.69%; and 40.9%,
39.1%. This indicates that the observed change in elemental
concentrations with depth takes place during annealing. The
reduction of the Sn and O concentration in the surface region
may be due to the congruent evaporation of SnO2 during
annealing, while the reduction of fluorine concentration may
be due to the formation of volatile fluoride species at the
surface.
For annealing temperature TH� 500 �C, the F1s BE
peak at etching time t¼ 60 s could be deconvoluted into two
Gaussian peaks as shown for the air annealed samples in Fig.
5. The principal peak (labeled FA) is located at
684.90 6 0.18 eV and the auxiliary peak (FB) is located at
684.13 eV, 684.42 eV, and 687.45 eV for TH¼ 500 �C,
550 �C, and 600 �C, respectively (Table I). Martinez et al.,18
Wu et al.,19 and Zhi et al.29 reported weaker FA signals in
SP-deposited FTO films and assigned it to Sn-F bond. The
maximum F/Sn ratio in those films was 2 at. %., which
explains the weaker signals as compared to the present work
for which F/Sn is in the range 2.0–7.8 at. %. Wu et al.19
reported that the intensity of this peak is proportional to the
amount of NH4F in the spray solution due to the substitution
of oxygen by fluorine, i.e., creation of FO defects. The chem-
ical state giving rise to the FB peak whose position varies
with TH is unknown. Nevertheless, it must be associated
with another fluoride phase. Table I summarizes the XPS
results for fluorine in our FTO samples.
Fig. 6 shows that the C1s peak for TH¼ 600 �C is decon-
voluted into three Gaussian peaks indicating the presence of
three chemical states for carbon. One of these states gives
rise to the peak CA (at 284.6 eV) used to calibrate the bind-
ing energy scale of the spectrum and usually attributed to
hydrocarbons or hydrocarbon groups (C-C, C-H bonds).19,30
A second peak CB at 285.8 6 0.5 eV can possibly be
assigned to other hydrocarbon complexes. The third chemi-
cal state produces the peak (CC) at 287.75 eV, which was
previously observed by a number of authors and assigned to
the C-O bond.19,30 Fig. 6 shows also the binding energy cor-
responding to the spin-orbit Sn3d3/2 and Sn3d5/2 peaks situ-
ated at 494.72 eV and 486.33 eV, respectively. The position
of the Sn3d5/2 peak corresponds to the Sn-O bond in which
Sn is in the tetravalent oxidation state as in SnO2.38 The
peak is perfectly Gaussian, which indicates that the presence
of the Sn-F bonds has negligible effect on its shape. This is
because of the fact that the peak due to the Sn-F bond coin-
cides with that of the Sn-O bond39 and due to the much lower
intensity for the Sn-F peak. The O1s peak (Fig. 6), on the
other hand, consists of two Gaussian peaks at 530.26 eV
(OA) and 530.92 eV (OB). The binding energies of both the
OA and the Sn3d5/2 peaks agree with the literature values for
the Sn-O bond with divalent and tetravalent oxidation states
for O and Sn, respectively.19,38 This confirms the successful
formation of SnO2 phase in our samples. The OB peak can
be attributed to the O-C bond.40
The atomic percentage of C, Sn, and O for some sam-
ples annealed at different temperatures/atmospheres is given
in Table II. The ratio O/Sn is larger than 2 (with one excep-
tion), which may lead to the illusive conclusion that our sam-
ples contain excess oxygen contrary to the XRD results.
FIG. 4. Depth profile of the main elements in FTO film annealed in air at
500 �C (etch rate � 2 A/s).
FIG. 5. F1s spectra after deconvolution
showing the FA (dotted line) and FB (solid
line) peaks for as deposited (TH¼ 450 �C)
and air annealed samples at TH¼ 500 �C,
550 �C, and 600 �C. The etching time is
60 s.
TABLE I. Summary of XPS results for fluorine in FTO films. The BE of the
FA and FB peaks and the atomic ratio FA/Sn for samples annealed in air at
different temperatures/atmospheres are listed. The etching time is 60 s.
BE (eV)
TH ( �C) FA FB FA/Sn (at. %)
450 684.92 — 2.0, 3.8
500 685.16 684.13 1.8, 3.1
550 685.07 684.42 3.4
600 684.8 687.45 6.1, 7.1
023501-5 F. E. Akkad and S. Joseph J. Appl. Phys. 112, 023501 (2012)
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
93.180.53.211 On: Tue, 18 Feb 2014 13:57:34
However, it will be shown later (Sec. IV) that it is very likely
that a non-negligible part of the oxygen population is chemi-
cally involved in bonds with other elements particularly car-
bon. An O/Sn ratio larger than 2 were reported on FTO films
by a number of investigators without indicating the carbon
content.23,41
D. Electrical properties
The results of room temperature electrical measure-
ments are summarized in Table III. It is noticed that, regard-
less of the annealing atmosphere, the electron concentration
n decreases with the increase of the annealing temperature
TH. A decrease of n with the increase of growth temperature
above 450 �C was reported by Martinez and Acosta42 on
FTO films prepared using the SP method. It was attributed to
a decrease in the VO concentration due to approach to the
stoichiometric composition at higher temperatures. Although
our results of XPS measurements indicate that the VO con-
centration decreases with the increase of the annealing tem-
perature (Table II and Sec. IV), this behavior is not expected
to induce a corresponding decrease in n, since the VO defects
are electrically neutral according to the more recent theoreti-
cal work of Singh et al.,22 which is supported by our XRD
results. Therefore, this reduction of n is most likely caused
by the creation of compensating acceptors at high annealing
temperatures. More discussion of this behavior of n will be
presented later (Sec. IV).
There is no definite trend of variation in the Hall mobil-
ity l with TH (Table III). However, for both air-and argon-
annealed samples, there is a tendency towards decreasing labove 550 �C. The dominant carrier scattering mechanism at
room temperature for n� 2� 1020 cm�3 was reported by
Martinez et al.18 to be that of charged impurities similar to
the case of ZnO.43 The literature values of l for FTO films
containing carrier concentration of �2� 1020 cm�3 vary
from 5 to 84 cm2/V s depending on the fluorine concentra-
tion and the processing parameters.18,44,45 This wide disper-
sion in l—values for samples with similar net impurity
concentration is an indication of non-negligible compensa-
tion in FTO films. This may explain the behavior of l above
550 �C in our samples. The compensating acceptors may be
extrinsic (e.g., interstitial fluorine Fi) or intrinsic (e.g., inter-
stitial oxygen Oi or vacancy of tin VSn).
E. Optical properties
The transmission spectra were measured in the wave-
length range 200� k� 3000 nm. The results for samples
annealed in air at different TH values are shown in Fig. 7.
The transmission T shows a plateau at about 88% in the
wavelength range 700� k� 1200 nm with little variation in
value with annealing temperature/atmosphere. Above
k� 1500 nm, T decreases steadily and a tail extends over the
near IR region up to k¼ 3000 nm.
The gradual decrease of the transmission with k in the
near IR region is a characteristic phenomenon of free carrier
absorption. The absorption coefficient af for this mechanism
is described by the classical formula46
af ¼nq2k2
sm8p2nrc3; (1)
where s is the relaxation time, m* is the effective mass of
carriers, c is the speed of light, and nr is the refractive index.
FIG. 6. C1s, Sn3d and O1s binding energy
peaks after deconvolution into Gaussian peaks
for FTO film annealed in air at TH¼ 500 �C.
The etching time is 60 s.
TABLE II. Atomic percentage of C, Sn, O and calculated stoichiometry pa-
rameters for FTO films for an etching time of 60 s. d is the amount of oxygen
deficiency and [VO] is the oxygen vacancy concentration.
Sample TH ( �C) Atmosphere C Sn O O/Sn d[VO]
(1022 cm�3)
A1 450 as-deposited 39.00 18.10 42.20 2.33 0.88 2.44
A2a 500 Air 26.70 23.90 48.70 2.04 0.69 1.91
A2b 500 Air 26.77 15.94 38.13 2.39 0.70 1.94
A3 550 Air 27.46 15.94 35.67 2.24 0.80 2.21
A4a 600 Air 24.00 23.10 51.30 2.22 0.54 1.50
A4b 600 Air 8.60 33.00 55.50 1.68 0.51 1.41
R2 500 Argon 6.09 23.02 36.30 1.58 0.61 1.68
R3 550 Argon 17.38 20.59 43.94 2.13 0.50 1.38
TABLE III. Summary of electrical and optical results in FTO films: electron
concentration n, electron mobility l, optical absorption coefficient at
2990 nm at, and energy band gap Eg.
Sample TH ( �C) Atmosphere
n
(1020 cm�3)
l(cm2/Vs)
at
(104 cm�1)
Eg
(eV)
A1 450 as-deposited 5.62 14.1 8.36 4.422
A2 500 Air 3.45 22.2 6.51 4.335
A3 550 Air 2.06 21.8 6.50 4.314
A4 600 Air 2.16 18.3 5.98 4.361
R1 450 as-deposited 8.58 8.9 8.32 4.422
R2 500 Argon 6.78 16.9 6.39 4.511
R3 550 Argon 2.62 19.8 5.67 4.531
R4 600 Argon 3.29 9.3 5.58 4.512
023501-6 F. E. Akkad and S. Joseph J. Appl. Phys. 112, 023501 (2012)
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
93.180.53.211 On: Tue, 18 Feb 2014 13:57:34
This equation predicts that the absorption coefficient at a
given wavelength is proportional to n.
The absorption coefficient was calculated using the
method of Swanepoel.47 This method is valid for an absorb-
ing film on a transparent substrate in the region of strong
absorption. Therefore, it is applicable in the band edge
region as well as in the region of the transmission tail. Fig. 8
shows the absorption coefficient at at k¼ 2990 nm as a func-
tion of the electron concentration n in the sample. The rela-
tion can be represented by a straight line as predicted by Eq.
(1). We have also verified that for a given sample a power
law relationship at ! kb with b close to 2 prevails, thus sup-
porting the dominant role of free carrier absorption. This
agreement between optical and electrical properties (Fig. 8)
can be taken to confirm that the observed electrical quantities
(q and n) are bulk properties of the films.
In the region of the fundamental absorption, the absorp-
tion coefficient a is given by the formula46
ðah�Þ1 ¼ Aðh� � EgÞ; (2)
where A is a constant, Eg is the energy gap and l¼ 2 for
direct allowed transitions. Fig. 9 shows (ah�)2 vs. h� plots in
the region of the fundamental absorption for air annealed
samples. A linear relationship is obtained at high photon
energy indicating a direct energy band gap. The values of Eg
determined from the intersection of the straight line with the
h� axis for both air- and Ar-annealed samples are listed in
Table III. They fall in the range 4.312 eV–4.531 eV. The lit-
erature value of Eg shows wide dispersion between different
investigators. Values in the range 3.60 eV–4.76 eV have
been reported.48–51 This dispersion can be attributed to two
opposing factors; on one hand, the Moss-Burstein shift (band
filling) causes a UV shift of the optical band gap and, on the
other hand, the many body interaction leads to shrinkage of
gap due to the appearance of band tails (Urbach tail).48
IV. DISCUSSION
A. Point defects concentrations
The question of whether the high temperature defects
(Vo, FO) are quenched down to room temperature can be dis-
cussed on the basis of the kinetics of the film-atmosphere
equilibrium during cooling the sample. Using the profile of
Sn and O near the surface (Fig. 4), which is likely to be
diffusion-limited, we determine a diffusion length for VO in
the range 20–26 A in the temperature range 500–600 �C(using an approximate value of 2 A/s for the rate of etching).
This means that the vacancies diffuse over few lattice con-
stants during cooling the sample at the experimental average
rate of about 30 �C/min. In addition, the diffusion barrier
deduced from these results amounts to approximately 1.4 eV,
which is close to the value 1.7–2.2 eV computed theoreti-
cally using density functional theory.52 With such high diffu-
sion barrier, one can neglect the departure of vacancies to
the atmosphere during cooling. Also the change in the FO
concentration can be neglected, since this defect is likely to
diffuse via vacancy mechanism, which keeps the total con-
centration unchanged. Therefore, it seems plausible that the
creation and annihilation of defects during annealing takes
place via chemical reactions inside the sample rather than
through atomic exchange with the environment. This
explains the observation that the physical properties of the
samples are almost independent of the annealing atmosphere
(air or argon).
1. Oxygen vacancies
As noted above (Sec. III C), the atomic ratio O/Sn points
toward the presence of excess oxygen in our samples.
FIG. 7. The transmission spectra for FTO films; as-deposited (450 �C) and
annealed in air atmosphere at 500 �C, 550 �C, and 600 �C. The films thick-
ness is in the range 217–243 nm.
FIG. 8. Dependence of the absorption coefficient at 2990 nm (at) on the
electron concentration n in FTO films.
FIG. 9. Dependence of (a h�)2 on h� for as-deposited and air annealed FTO
samples.
023501-7 F. E. Akkad and S. Joseph J. Appl. Phys. 112, 023501 (2012)
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
93.180.53.211 On: Tue, 18 Feb 2014 13:57:34
However, the calculation of a reliable value of the stoichiom-
etry in the “bulk” of the film cannot be done before evaluat-
ing the effect of possible creation of chemical bonds
between the two extremely active elements; oxygen and car-
bon. The presence of O-C bonds is suggested from both the
OB and CC peaks and was reported by previous investiga-
tors.40 Since each C ion consumes half the number of O ions
that a Sn ion consumes, then we assume that If [OSn]¼ x[Sn] then [OC]¼ 1=2 x {e[C]} where [OSn] and [OC] are the
concentrations of oxygen involved in chemical bonds with
tin and carbon, respectively, [Sn] and [C] are the concentra-
tions of tin and carbon, respectively, and e is the fraction of
C involved in the O-C bonds (0� e� 1). The total oxygen
concentration is [O]¼ [OSn]þ [OC], which leads to
x ¼ 2½O2½Sn þ e½C : (3)
We define d as the amount of oxygen deficiency (d¼ 2�x),
so that d is proportional to the vacancy concentration in
SnOx films. Fig. 10 shows the profile of d with different val-
ues of e for the same sample whose elemental profiles are
shown in Fig. 4. It is seen that the profile converges towards
an asymptotic value at large depths. At etching time t¼ 60 s
and e¼ 1, the calculated value of d is indicative of the “bulk”
value to within 64%. The relative change of d from one
sample to another is expected to have accuracy better than
64%. The obtained values of d are shown in Table II. It is
noticed that, contrary to what the ratio O/Sn indicates, the
films are in fact oxygen-deficient with values of d in the
range 0� d� 1 as expected for SnO2�d. This result is con-
sistent with the XRD results, which indicate that our films
contain large concentrations of oxygen vacancies. Therefore,
it may be concluded that the ratio O/Sn that has often been
used to determine the stoichiometry in TO is not a good in-
dicative of this property. This is confirmed by noticing that
the value of d calculated on different points of the same sam-
ple (Table II) is almost the same while the O/Sn ratio shows
significant dispersion (samples FTOA2a & b and FTOA4a &
b). The vacancy concentration [VO] was calculated using
[VO]¼ Pd where P¼ 2.77� 1022 cm�3 mol�1 for tin oxide.
The obtained values of [VO] (Table II) are about two orders
of magnitudes higher than the electron concentration n (Ta-
ble III), which confirms that oxygen vacancies are electri-
cally neutral in agreement with our XRD results and the
work of Singh et al.22
2. Substitutional fluorine
It was pointed out in Sec. III C that the fluorine peak FA
is due to Sn-F bonds and that previous reports showed that
its intensity is proportional to the amount of NH4F in the
spray solution suggesting that it is due to the substitutional
fluorine defect FO.18,19,29 The concentration of substitutional
fluorine was calculated from the relative intensity of the FA
peak to the Sn3d5/2 peak (FA/Sn) using [FO]¼P (FA/Sn). It
was found to be in the range 5–20� 1020 cm�3 which is
close to the range of values for the electron concentration
(2–9� 1020 cm�3) as shown is Table III. This is in line with
the expectation that substitutional fluorine is the source of
electrons in FTO films. However, contrary to expectations,
the increase in [FO] with TH is associated with a decrease in
n (Tables I and III, Fig. 11).
B. Defects model
1. Proposed reaction
The unexpected simultaneous increase of [FO] and
decrease of n with annealing temperature indicates that the
same high temperature reaction that leads to the creation of
FO defects produces compensating acceptor defects with
larger charge concentration. The supply of FO defects during
annealing may take place from a fluoride second phase such
as for example SnF4. In this case, the change of the tin oxida-
tion state from the Sn4þ to the Sn2þ state may lead to the
release of fluorine and the creation of VSn acceptor according
to the reaction
2SnF4 þ 2V0O þ SnSn þ 2e! 3SnF2 þ 2FþO þ V4�
Sn : (i)
This reaction leads to a decrease in both VO and n concentra-
tions and to an increase in FO concentration as observed
experimentally. Applying the law of mass action to the
above reactions, one obtains
½VSn½FO2
½VO2n2¼ K (4)
with
K ¼ Koexpð�DH=kTHÞ; (5)
where [VSn], [FO], and [VO] are the concentrations of tin
vacancies, substitutional fluorine, and oxygen vacancies,
respectively. DH is the enthalpy of reaction (i) and Ko is a
constant.
2. Calculation of an approximate value for theenthalpy
By creating Arrhenius plots using the equation for a
thermally activated concentration,
Ci¼Co exp(�Ei/kTH), where Ei (i¼FA/Sn, d, n) is the
activation energy and Co is a constant, one can use Ei toFIG. 10. Depth profile of d in SnO2�d for different values of e (Eq. (4)).
023501-8 F. E. Akkad and S. Joseph J. Appl. Phys. 112, 023501 (2012)
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
93.180.53.211 On: Tue, 18 Feb 2014 13:57:34
determine the enthalpy of the above reaction. Figure 11
shows these Arrhenius plots, leading to EFO¼EFA/Sn
¼ 0.36 eV, EVO¼Ed¼�0.18 eV and En¼�0.42 eV, where
EFO and EVO are the formation energies of substitutional flu-
orine and oxygen vacancies, respectively.
The neutrality condition at room temperature is given
by,
FþO ¼ nþ 4V4�Sn ;
so that the formation energy of the tin vacancies (EVT) can
be determined by creating an Arrhenius plot for the quantity
([FO]� n). Fig. 11(d) shows this plot which yields
EVT¼ 0.45 eV. From these values of Ei and using Eqs. (4)
and (5), an enthalpy of 2.37 eV for the reaction (i) is
determined.
In the above calculation, we neglect the variation of nbetween the annealing temperature and the room tempera-
ture, since the substitutional fluorine defects are known to
act as shallow impurities that are completely ionizes above
room temperature.
V. CONCLUSION
The morphological, structural, electrical, optical, and
chemical properties of FTO films are studied. The role of ex-
trinsic and intrinsic point defects on these properties is
emphasizes. It is shown that the films contain large concen-
tration (�1022 cm�3) of oxygen vacancies that are electri-
cally neutral. The source of electrons is the partially
compensated FO donors. The results suggest that the ratio O/
Sn is not a reliable quantity for determining the stoichiome-
try of the films because a significant part of the oxygen popu-
lation is involved in chemical bonds with carbon. The high
temperature heat treatment causes an increase in FO and a
decrease in both n and VO. These results can be interpreted
in terms of a high temperature chemical exchange reaction
between the SnO2 matrix and a precipitated fluoride phase.
In this reaction, the release of fluorine from the fluoride
phase and subsequent formation of substitutional fluorine
defects is associated with the creation of tin vacancies. The
enthalpy of the reaction is determined to be approximately
2.4 eV of which 0.45 eV is consumed in the migration of Sn
from a lattice site to the fluoride phase with the creation of a
tin vacancy.
ACKNOWLEDGMENTS
The authors would like to thank the research administra-
tion of Kuwait University for funding the present work (Pro-
ject No. SP01/11). Their thanks are also due to the general
facility of the Faculty of Science (Projects GS 02/08 and
GS03/01) and of the Faculty of Engineering (Projects GE01/
07, GE01/08, and GE02/08) for their valuable technical
support.
1P. D. Paulson, B. E. McCandless, and R. W. Birkmire, J. Appl. Phys. 95,
3010 (2004).2K. S. Ramaiah, V. S. Raja, A. K. Bhatnaar, F. S. Juang, S. J. Chang, and
Y. K. Su, Mater. Lett. 45, 251 (2000).3C. H. Yang, S. C. Lee, S. C. Chen, and T. C. Lin, Mater. Sci. Eng. B 129,
154 (2006).4J. H. Spanggaard and F. C. Krebs, Sol. Energy Mater. Sol. Cells 83, 125
(2004).5E. Elangovan and K. Ramamurthi, J. Optoelectron. Adv. Mater. 5, 45
(2003).6R. Riveros, E. Romero, and G. Gordillo, Braz. J. Phys. 36, 1042 (2006).7N. Memarian, S. M. Rozati, E. Elamurugu, and E. Fortunato, Phys. Status
Solidi C 7, 2277 (2010).8B. Thangaraju, Thin Solid Films 402, 71 (2002).9B. H. Liao, C. C. Kuo, P. J. Chen, and C. C. Lee, Appl. Opt. 50, C106
(2011).10Z. Remes, M. Vanecek, H. M. Yates, P. Evans, and D. W. Sheel, Thin
Solid Films 1, 6287 (2009).11A. P. Rizzato, C. V. Santilli, S. H. Pulcinelli, P. Hammer, and V. Briois,
J. Eur. Ceram. Soc. 25, 2045 (2005).12D. C. Paine, T. Whitson, D. Janiac, R. Beresford, C. Ow Yang, and B.
Lewis, J. Appl. Phys. 85, 8445 (1999).13P. S. Shewale, S. I. Patil, and M. D. Uplane, Semicond. Sci. Technol. 25,
115008 (2010).14S. Vijayalakshmi, S. Venkataraj, M. Subramanian, and R. Jayavel, J Phys.
D: Appl. Phys. 41, 035505 (2008).15V. Bilgin, I. Akyuz, E. Ketenci, S. Kose, and F. Atay, Appl. Surf. Sci. 256,
6586 (2010).
FIG. 11. Arrhenius plots for (a) the electron concentration n, (b) the atomic ratio FA/Sn, (c) the amount of oxygen deficiency d, and (d) the quantity [FO]-n.
023501-9 F. E. Akkad and S. Joseph J. Appl. Phys. 112, 023501 (2012)
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
93.180.53.211 On: Tue, 18 Feb 2014 13:57:34
16R. Tala-Ighil, M. Boumaour, M. S. Belkaid, A. Maallemi, K. Melhani, and
A. Iratni, Sol. Energy Mater. Sol. Cells 90, 1797 (2006).17E. Shanthi, A. Banerjee, and K. L. Chopra, Thin Solid Films 88, 93
(1982).18A. I. Martinez, L. Huerta, J. M. O-Rueda de Leon, D. Acosta, and O.
Malik, J. Phys. D: Appl. Phys. 39, 5091 (2006).19S. Wu, S. Yuan, L. Shi, Y. Zhao, and J. Fang, J. Colloid Interface Sci. 346,
12 (2010).20H. Kim, R. C. Y. Auyeung, and A. Pique, Thin Solid Films 516, 5052
(2008).21C
¸. Kilic and A. Zunger, Phys. Rev. Lett. 88, 095501 (2002).
22A. K. Singh, A. Janotti, M. Scheffler, and C. G. Van de Walle, Phys. Rev.
Lett. 101, 055502 (2008).23Z. B. Zhou, R. Q. Cui, G. M. Hadi, W. Y. Li, and Z. M. Ding, J. Mater.
Sci.: Mater. Electron. 12, 417 (2001).24R. Tala-Ighil, M. Boumaour, M. S. Belkaid, A. Maallemi, K. Melhani, and
A. Iratni, Sol. Energy Mater. Sol. Cells 90, 1797 (2006).25E. C. P. E. Rodrigues and P. Olivi, J. Phys. Chem. Solids 64, 1105 (2003).26C. D. Canestraro, M. M. Oliveria, R. Valazki, M. V. S. da Silva, D. G. F.
David, I. Pepe, A. F. da Silva, L. S. Roman, and C. Persson, Appl. Surf.
Sci. 255, 1874 (2008).27M. Fantini and I. Torriani, Thin Solid Films 138, 255 (1986).28C. M. Ghimbeu, R. C. Van Landschoot, J. Schoonman, and M. Lumbreras,
J. Eur. Ceram. Soc. 27, 207 (2007).29X. Zhi, G. Zhao, T. Zhu, and Y. Li, Surf. Interface Anal. 40, 67 (2008).30J. H. Park, D. J. Byun, and J. K. Lee, J. Electroceram. 23, 506 (2009).31D. Zaouk, R. al Asmar, J. Podlecki, Y. Zaatar, A. Khoury, and A. Fou-
caran, Microelectron. J. 38, 884 (2007).32D. Miao, Q. Zhao, S. Wu, Z. Wang, X. Zang, and X. Zhao, J. Non-Cryst.
Solids 356, 2557 (2010).33J. H. Lee, Y. Y. Kim, H. K. Cho, and J. Y. Lee, J. Cryst. Growth 311,
4641 (2009).
34H. Izumi, F. O. Adurodija, T. Kaneyoshi, T. Ishihara, and H. Yoshioka,
J. Appl. Phys. 91, 1213 (2002).35Y.-J. Lin, M.-S. Wang, C.-J. Liu, and H.-J. Haung, Appl. Surf. Sci. 256,
7623 (2010).36S. Rani, N. K. Puri, S. C. Roy, M. C. Bhatnagar, and D. Kanjilal, Nucl.
Instrum. Methods Phys. Res. Sec. B 266, 1987 (2008).37A. L. Patterson, Phys. Rev. 56, 978 (1939).38H. J. Ahn, H. C. Choi, H. W. Park, S. B. Kim, and Y. E. Sung, J. Phys.
Chem. B 108, 9815 (2004).39See http://www.Nist.gov for NIST x-ray photoelectron spectroscopy
database.40F. Beguin, I. Rashkov, N. Manolova, R. Benoit, R. Erre, and S. Delpeux,
Eur. Polym. J. 34, 905 (1998).41E. Cetinorgu, S. Goldsmith, Y. Rosenberg, and R. L. Boxman, J. Non-
Cryst. Solids 353, 2595 (2007).42A. I. Martinez and D. R. Acosta, Thin Solid Films 483, 107 (2005).43T. Minami, MRS Bull. 25, 38 (2000).44M. Di Giulio, D. Manno, G. Micocci, R. Rella, P. Siciliano, and A. Tepore,
Sol. Energy Mater. Sol. Cells 31, 235 (1993).45J. Yang, W. Liu, L. Dong, Y. Li, C. Li, and H. Zhao, Appl. Surf. Sci. 257,
10499 (2011).46J. I. Pankove, Optical Processes in Semiconductors (Dover, 1975).47R. Swanepoel, J. Phys. E. Sci. Instrum. 16, 1214 (1983).48G. Sanon, R. Rup, and A. Mansigh, Phys. Rev. B 44, 5672 (1991).49L. V. A. Scalvi, F. R. Messias, A. E. Souza, M. S. Li, C. V. Santilli, and S.
H. Pulcinelli, J. Sol-Gel Sci. Technol. 13, 793 (1998).50R. Summitt, J. A. Marley, and N. F. Borrelli, J. Phys. Chem. Solids 25,
1465 (1964).51A. K. Abass, H. Bakr, S. A. Jassim, and T. A. Fahad, Sol. Energy Mater.
17, 425 (1988).52N. Lopez, J. Daniel Prades, F. Hernandez-Rami rez, J. R. Morante, J.
Pand, and S. Mathurd, Phys. Chem. Chem. Phys. 12, 2401 (2010).
023501-10 F. E. Akkad and S. Joseph J. Appl. Phys. 112, 023501 (2012)
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
93.180.53.211 On: Tue, 18 Feb 2014 13:57:34