Thin Solid Films 567 (2014) 20–31

Contents lists available at ScienceDirect

Thin Solid Films

journal homepage: www.elsevier.com/locate/tsf

Study of wet etching thin films of indium tin oxide

in oxalic acid by monitoring the resistance
Suelene S. Mammana a,⁎, Alessandra Greatti a, Francis H. Luiz a, Francisca I. da Costa a, Alaide P. Mammana a,
Guilherme A. Calligaris b, Lisandro P. Cardoso b, Carlos I.Z. Mammana a, Daniel den Engelsen a
a
Brazilian Association for Informatics - ABINFO, Rua Deusdete Martins Gomes 163, CEP 13084-723, Campinas, SP, Brazil
b
Institute of Physics Gleb Wataghin, State University of Campinas-UNICAMP, CEP 13083-859, Campinas, SP, Brazil

a r t i c l e i n f o a b s t r a c t

Article history: We describe a study on wet etching of thin films of indium tin oxide (ITO) using a simple method by monitoring
Received 9 September 2013 the resistance of the thin film in aqueous solutions of oxalic acid and hydrochloric acid. Generally three different
Received in revised form 15 July 2014 regimes can be distinguished during etching ITO in acids: (1) initial etching, which is slow, (2) a fast etching
Accepted 17 July 2014
phase and (3) slow etching stage at the end. These regimes are explained in terms of a porosity–roughness
Available online 26 July 2014
model. This porosity model has been confirmed largely by X-ray reflection measurements at grazing incidence,
Keywords:
roughness measurements and scanning electron microscopy (SEM).
Etch rate A reliable method for monitoring the resistance during etching has been developed. This method is based
Activation energy on a 2-strips measuring jig with a very low series contact resistance.
X-ray reflectivity The activation energy of the etch rate of ITO films was found to be 80 ± 5 kJ/mol for oxalic acid and 56 ± 5 kJ/mol
Morphology for HCl. SEM analyses in the final stage of the etching process indicate an enrichment of Sn in the residual film
Selective etching material. These observations are explained in terms of preferential etching of In2O3. X-ray analyses showed
Electrical resistance that the density of the ITO film decreased by etching. By adding ferric chloride to the oxalic acid solution we
could accelerate the etch rate substantially.
© 2014 Elsevier B.V. All rights reserved.

1. Introduction investment in equipment than others, for instance plasma etching or

Transparent conductors continue to be of extreme importance for
several applications such as displays, touch screens, photovoltaic devices
and flexible electronics. Many of these devices employ one or more layers
of a transparent conductive material. The high electrical conductivity and
optical transparency of thin films of indium tin oxide (ITO) make them
the most widely used transparent conductors for these devices.
The impressive growth of the area of the substrates coated with ITO
and the increasingly smaller dimensions of the devices built on them
are requiring higher control of the processes for both depositing and pat-
terning the ITO films. Furthermore, besides the high cost of ITO, its depo-
sition and patterning processes represent a reasonable percentage of the
final cost of the devices, reason why less costly processes continue to be
searched as alternatives. At the same time, environmental responsibility
is driving the search for processes that are less harmful to the environ-
ment as well as less aggressive to people enrolled in the processing.
Patterning of thin films of ITO is usually done by lithography, which
includes an etching step that is mostly wet etching since it requires less
⁎ Corresponding author at: Arquiteto José Augusto Silva 761, ap: 63 TO, Jardim
Santa Cândida, Campinas, CEP: 13087-570, São Paulo, Brazil. Tel.: + 55 19 33059283.
E-mail addresses: ssmammana@abinfo.com.br, ssmammana@gmail.com
(S.S. Mammana).
laser ablation.
Hydrochloric acid (HCl) and HCl with additional nitric acid (HNO3) or
ferric chloride (FeCl3) are often used as wet etchants at slightly elevated
temperature for ITO films [1–4]. Etch rates of poly crystalline ITO films in
these etchants vary between 5 and 50 nm/min, depending on tempera-
ture and concentration. It should be stressed that amorphous ITO films
show about a factor of 100 larger etch rates [5–7], indicating that the
structure of the ITO film is a paramount parameter for the etch rate.
Device makers for displays etc. often prefer to apply amorphous ITO
films, because of the high etch rate.
HCl-based etchants have some drawbacks such as: uncontrolled
under or lateral etching, attack of the metal layers beneath the ITO-top
film, corrosion of equipment and fume exhaust facilities. Organic acids
such as oxalic acid, acetic acid, formic acid, citric acid and tartaric acid
are less corrosive and generally do not attack the metallization of a
backplane with thin film transistors. For that reason, these acids are
now being investigated for wet etching of ITO and other transparent
conductive oxides such as indium zinc oxide and indium gallium zinc
oxide [6,8–11].
ITO typically consists of 90% In2O3 and 10% SnO2; its crystal structure
is bixbyite (cubic structure), which is identical to the crystal structure of
In2O3 [12]. In ITO, Sn4+ ions sit on In3 + lattice sites: therefore ITO is
sometimes indicated as Sn-doped In2O3. Since In2O3 is well dissolving

http://dx.doi.org/10.1016/j.tsf.2014.07.027
0040-6090/© 2014 Elsevier B.V. All rights reserved.
 S.S. Mammana et al. / Thin Solid Films 567 (2014) 20–31
in acids whereas SnO2 is not, it is a priori not clear how the etching of a
thin ITO film in an acidic solution will proceed. So, the following ques-
tion is relevant: is In2O3 dissolving preferentially from the thin film of
ITO? To answer this question one needs to study the kinetics of the etch-
ing process of ITO in detail.
The purpose of our work is twofold:
(1) Presenting a simple and cost effective method to study the kinetics
of etching thin films of ITO in acidic solutions;
(2) Illustrating the usefulness of our method for studying the etching
process with etchants consisting of oxalic acid and mixtures of
oxalic acid and oxidizing or reducing agents.
Some preliminary results of our study have been presented in the
Proceedings of LatinDisplay 2012/IDRC 2012 [13,14].
In studies on wet etching of ITO, the procedure to evaluate the etch
rate is very often not explicitly mentioned [1,3,4,6–10]. Since the focus
in those studies is on overall etch rates, it is likely that the etch rate is
evaluated by dividing the film thickness by the total etch time. Howev-
er, it is not mentioned how the total etch time is determined. The under-
lying assumption for the evaluation of the etch rate in those studies is
that it is constant during the etching process. Kinetic studies of the
etching process of thin films of ITO and other transparent conductive
materials such as SnO2 and ZnO require measuring the thickness of
the film or the mass removed during the process.
Optical methods for monitoring the thickness were used, such as
ellipsometry [15], and diffraction on gratings engraved in the film. In
the latter method, a lamellar grating built in the film is monitoring the
dry etching of very thin layers through the measurement of the depth
of the grating grooves. This technique can also be used to precisely
detect the end point of the etching process, in situ and in real time,
which makes it a powerful tool to control the etching [16–18]. Never-
theless, this method has the disadvantage of requiring the patterning
of a diffraction grating on the film.
Jacobs et al. [19] and van den Meerakker et al. [2] employed
profilometry to measure the decrease of the thickness of an ITO-film
covered with a photoresist pattern with steps. Since this method also
requires the patterning of the photoresist, the measurement can be
cumbersome for very thin ITO-films because of surface roughness and
formation of isolated ITO-residuals on the substrate surface.
For monitoring the etch rate, we propose to measure the electrical
resistance of the film as it is a function of the thickness of the film and
taking advantage that it does not require photolithographic processes
to prepare the samples. Moreover, the electrical conductivity is one of
the most important properties of transparent oxide films such as ITO.
So, resistance is a convenient parameter to be measured and it is directly
related to the electrical properties of the devices obtained by an etching
process. Our method is not limited to ITO, but may be applied to study
the kinetics of etching of various types of conducting films or to monitor
industrial etch processes.
Monitoring the resistance during etching generates information not
only on the etch rate but also on the resistivity of the ITO-film as a func-
tion of z, being the axis in the direction normal to the substrate. If the
resistivity of the film does not change with z, a linear increase of the
resistance with time during etching indicates that the etch rate
is constant during the process. However, if the resistance does not
increase linearly with time, then either the etch rate may vary during
the process or the resistivity may change with z.
The difference between surface and bulk properties of ITO films
has been described extensively in the literature on thin films of ITO
[20–24]. The objective of that work is to increase the work function of
the ITO-surface, which facilitates the hole-injection from an ITO-anode
into the hole-injection layer of an organic light emitting diode (OLED).
King et al. [25] describe that electrons accumulate in the surface area of
crystalline pure indium oxide, whereas in the surface zone of ITO they
find a depletion of electrons, leading to a higher surface resistance. For
ITO this depletion zone is very thin (0.2 nm). The work of King et al. refers
21
to a macroscopic single crystal and it is unknown whether the boundary
conditions of their treatment may be applied to the nanocrystals of ITO in
thin films during etching.
At the end of the etching process in an industrial ambiance, it is
important to know what residues could adhere to the substrate, because
these could impair the isolation between tracks or parts of devices
produced with etching processes. For this purpose we characterized
the ITO-films with respect to topography, crystalline structure, micro-
chemical composition before and after different etch times, by using
techniques as X-ray reflectivity (XRR), grazing incidence X-ray diffrac-
tion (GIXRD), atomic force microscopy (AFM), scanning electron
microscopy (SEM), energy dispersive X-ray spectroscopy (EDX) and
profilometry.
We applied our resistance monitoring method to study the etching
process of ITO thin films, employing oxalic acid that is less harmful to
people and to the ambient than inorganic acids normally used in labora-
tories and in the industry. Oxalic acid is a weaker acid than HCl or HNO3;
so, it might be expected that etching of ITO films in oxalic acid will be
slower.
The effect of adding oxidizing agents to HCl-containing etchants is
well documented in the literature [2,19]. Especially FeCl3 enhances the
etch rate of ITO in concentrated HCl. The addition of oxidants to oxalic
acid etchants has not been studied and we considered it interesting to
study whether FeCl3 has an enhancing effect for this etchant as well.
Lee [26] found that the addition of a mild reducing compound such as
ascorbic acid to an etching bath containing oxalic acid also accelerates
the etch rate of ITO films. This observation is not understood, because
oxalic acid is also a reducing agent. Nevertheless, we have included
the effect of adding another reducing agent, i.e. sodium hypophosphite,
to the oxalic acid bath in our research.
2. Experiments and characterization
Our samples were cut from soda lime glass plates (1 mm thickness)
industrially deposited with thin films of ITO of 25 nm and 175 nm thick-
nesses and nominal sheet resistances of ~100 and ~10 Ω/square, which
correspond to resistivities of ~2.5 × 10−4 Ω.cm and ~1.8 × 10−4 Ω.cm
respectively. All samples were highly transparent. The 25 nm ITO film
was deposited directly on the soda lime glass, while the substrate
glass of the 175 nm ITO film had a thin layer of silicon dioxide (SiO2)
as a barrier to prevent diffusion of sodium (Na) ions into the ITO film.
The samples were thoroughly cleaned before etching for 15 min in a
solution of 5% Extran® MA02 in an ultrasonic bath, then rinsed with
distilled water and, in sequence, with absolute ethyl alcohol, dried
with hot air and stored in a dry box. Some samples were also cleaned
with a common detergent with anionic surfactants. We did not use
aggressive reagents (notably acids) to clean the samples in order to
preserve the surface structure of the ITO films.
To study the etch rate as a function of the concentration and of the
temperature, the ITO films were etched in solutions of oxalic acid in
water. We chose to take concentrations in the range of 0.02 to 1.35 m,
in which m stands for molality, while the temperature of the solutions
was adjusted to 40, 50 and 60 °C during the etching process. The solubil-
ity of oxalic acid in water at room temperature is about 1.3 m, but at the
temperatures during etching it is much higher.
Etching was also studied in solutions of oxalic acid containing ascor-
bic acid and sodium hypophosphite in concentrations of 0.01 to 0.15 m,
at temperatures of 40, 50 and 60 °C. Furthermore, etching experiments
were done with oxalic acid and iron(III) nitrate in concentrations
between 0.08 and 0.5 m at 60 °C as well as with iron(III) chloride in con-
centrations of 0.3 to 1.0 m, at 60 °C. This study was completed by etch-
ing ITO in oxalic acid with potassium chloride. For comparison with
oxalic acid, the etching was also done in hydrochloric acid at a concen-
tration of 3.2 M (where M stands for molarity) at 40, 50, 60 and 70 °C.
The etching solutions were prepared with distilled water and high
purity (p.a.) reagents. Each etching experiment was done with a freshly
22 S.S. Mammana et al. / Thin Solid Films 567 (2014) 20–31
prepared solution to ensure high reproducibility and the solution was
agitated by a flow of air to guarantee uniformity of the etching. Evapora-
tion losses of water during the etching were in most cases counteracted
by adding balanced quantities of distilled water to the etching bath.
The etching was monitored after each minute by measuring the
resistance of the ITO film to determine the film thickness in function
of time. This measurement was done after removing the sample from
the solution, cleaning it for 1 min in an ultrasonic bath of distilled
water at 50 °C, rinsing it with ethyl alcohol and drying it in a jet of hot
air. The time needed to remove the sample from the solution and to
wash it in distilled water to stop the reaction was routinely 7 s, therefore
not contributing significantly to the error in the etch time of 1 min. We
chose to measure the resistance with the sample out of the etching so-
lution, avoiding the parasitic currents in the conductive etching solu-
tion. No alkali was used to stop the etching reactions, avoiding that
additional reagents could be adsorbed on the ITO-surface and affect
the measurements.
For measuring the resistance of the samples we developed a jig with
two parallel contact strips of 20 mm length at a distance of 10 mm,
yielding 2 squares in parallel. The jig had a cantilever pulled by a string
of adjustable length to apply a force of up to 45 N to a tool that presses
the sample against two conducting strips (ropes) made of tinned
copper-woven strands. These strips are parallel within 100 μm (deviation
from parallelism b 1%) determined by a polyvinyl chloride separator. The
effective contact area for two strips was 0.2 cm2 and the load was 2.3
× 106 Pa. We refer to this resistance measurement as the 2-strips method.
The resistance was measured with a conventional ohm meter. The jig
permitted fast and accurate measurements of the resistance during
etching.
The effect and the value of the contact resistance in this measuring
method were determined in the following way. Three measuring jigs
were fabricated with different distances (D) of 5.25, 8.3 and 11.5 mm
respectively between the strips. Measurements of the resistance of
ITO-films were made before and after cleaning, and after etching for 2,
7 and 17 min respectively in 1.35 M oxalic acid at 60 °C. The result of
measuring 10 samples is depicted in Fig. 1a.
The resistance values of the samples before etching virtually coincid-
ed with those measured after etching for two minutes. From the calcu-
lated linear fittings the series resistance, being the sum of the contact
resistance at each strip, at D = 0 mm is obtained. We attribute the
low contact resistance of only 0.3 Ω to the application of woven wires,
which provide a great number of contact points with the film along its
length (20 mm) distributing the current flow uniformly over the macro-
scopic contact area. Fig. 1b shows that the contact resistance is not a
function of cleaning or etching. From this result we conclude that the
two-strips measuring method is a reliable way to monitor the sheet
resistance of ITO films during etching.
Repeatability of the 2-strips measurements was tested for a series of
20 measurements at the same sample, applying and releasing the
pressure on the sample to simulate the conditions of the experiment.
The mean deviation of the resistance measurements was better than
0.5%. For monitoring the resistance of the film during etching, it is not
necessary to measure the sheet resistance as long as the geometries of
sample and 2-strips measuring jig (distance between and length of
the contact strips) do not change.
For comparison reasons, resistance measurements of the ITO
films with the 2-strips and 4-point probes were done before and
after etching. Two different 4-point probes were used for this com-
parison, viz. the HM21 (Jandel Engineering Ltd.) and the FPP-2000
(Thin Film Devices Inc.). We found an excellent agreement between
the 2-strips probe and the 4-point probes: the deviations between the
sheet resistance measured with the 2-strips probe and 4-point probes
before and after etching are less than 6% for the same sample being the
standard deviation of 4% for the measurements with the 4-point probe.
The 4-point probes measurements were done at three different positions
on the ITO-film. The etching curves R0/R, to be introduced in Eq. (2) in
Section 3, determined with the 2-strips probe and the 4-point FPP-
2000 probe, coincide. This latter observation indicates that the etch
rate is uniform over the film surface.
The thickness, topography, crystalline structure and chemical compo-
sition of the films were investigated during the etching process.
Topographic characterization was done by atomic force microscopy
(AFM) with a Digital Instruments Nanoscope III AFM, and by scanning
electron microscopy (SEM) with an FEI Quanta 650 microscope,
equipped with an energy dispersive X-Ray (EDX) spectrometer (Oxford
Instruments) for making semi-quantitative surface analyses. The thick-
ness and roughness of the films were measured by profilometry with a
Veeco Dektak 150 Surface profilometer. The poly-crystallinity, rough-
ness, composition and density of the ITO films were investigated by X
ray diffraction techniques, specifically with X-ray reflectivity (XRR) and
grazing incidence X-ray diffraction (GIXRD), using a PANalytical X'Pert
MRD diffractometer with CuKα (0.15418 nm) radiation, a 0.27° parallel
plate collimator and a graphite flat crystal as a beam monochromator
[27]. The XRR-measurements of the ITO films were done after various
etching times.
3. Porosity model of etching ITO
We shall first consider the etch rate of an ITO film that is uniform,
horizontally and in the z-direction, in a solution that is well stirred
and has a large excess (~ 104) of etchant molecules as compared to
ITO-molecules in the film. If the area to be etched is constant, the etched
volume or mass is proportional to the decrease of thickness, indicated
by e. If there is no autocatalytic effect of reactants, then the etch rate is
constant: in other words, e decreases linearly with time, according to:
eðt Þ ¼ e0 −kt; ð1Þ
where e0 is the initial thickness of the film and k is the etch rate, indicat-
ed here in nm/min. If the initial resistance at t = 0 is denoted by R0, then
the normalized conductance, R0/R(t), shortly represented by R0/R, can
be written as:
R0 =R ¼ eðt Þ=e0 : ð2Þ
The etching process can be monitored by plotting R0/R as a function
of time. It should be mentioned that R0 is generally not equal to the
sheet resistance of the film: it is the initial resistance of the film before
etching whose value depends on the area covered by the two measuring
strips. For a uniform film the R0/R curve will be a straight line going
down from 1 at t = 0 to 0 at the end of the etch process. Deviations
from linearity in the R0/R versus time curves may indicate changes in
the properties of the ITO-film, such as structure, resistance, composition
and density as a function of z. Deviations from linearity in the R0/R
curves were normally observed in our experiments. Before describing
the porosity–roughness model of ITO films in detail, we show some typ-
ical R0/R curves of etching ITO in oxalic acid.
Fig. 2a shows the normalized conductance R0/R of ITO films with ini-
tial sheet resistance of 100 Ω/square as a function of time, obtained from
the measurement of the resistance of the ITO films during etching in
oxalic acid with concentrations of 0.02, 0.04, 0.08, 0.32 and 1.35 m at
50 °C, while Fig. 2b is depicting the normalized conductance versus
time of etching an ITO film in 0.08 m oxalic acid at 40, 50 and 60 °C.
In these figures the slope of the R0/R-curves is almost zero at the be-
ginning of the process; then, the curves show an almost linear region in
the middle, while at the end of the etching the slope is again decreasing
substantially. Both the beginning and the end of the R0/R curves indicate
that the ITO films are non-uniform, either in resistivity or in etch rate.
We shall now introduce a model that allows a simple description of
the phenomena observed in Fig. 2a and b. We refer to this model as “po-
rosity model”, because the most important feature of this model is ero-
sion of the ITO bulk layer creating a rough and porous layer. The
 S.S. Mammana et al. / Thin Solid Films 567 (2014) 20–31 23

equal contact resistances of Rs/2 between each contact strip and the
ITO film. The total contact resistance Rs is in series with the film resis-
tance R0.
Fig. 3b shows R0/R curves calculated according to the equivalent
circuit depicted in Fig. 3a with a parameter selection that adequately
fits the resistance measurements and shape of the R0/R curves. Etching
of the bulk ITO layer starts after the original top layer has been removed
completely. In the simulations we did not allow fp N kb, because of
conservation of mass. After removing of the bulk layer, quite some addi-
tional time is necessary to consume the porous layer completely. We
can only indicate the end of the etch process as the moment that the
resistance becomes very large (N107 Ω), or R0/R → 0. Complete removal
of all porous material cannot be detected by resistance measurements
beyond the percolation limit, i.e. the point that isolated islands of ITO
material appear on the substrate.
The first and second breakpoints in the R0/R curves in Fig. 3b indicate
the moments that the top and bulk layers respectively have been etched
away: the time after passing the second breakpoint is necessary to etch
the remaining rough and porous layer. The parameters that determine
the time necessary for the initial etching, in which the resistance chang-
es hardly, are et and kt and the time necessary for removing the porous
layer after the second breakpoint is determined by ep, fp and kp. We have
insufficient experimental data to determine these parameters indepen-
dently. The values of R0/R at which the breakpoints occur are largely
determined by the resistivities of the various layers. In the simulations
presented in Fig. 3b we did not vary ρb and kb, because the first quantity
largely determines the value of R0, while the latter is dominant for the
etch time between the breakpoints. However, from the simulations it
can definitely be concluded that the original top layer must have a larger
resistivity than that of bulk ITO: a small value of ρt would show up in the
R0/R-t diagram as a negative slope starting at t = 0, in other words, no
plateau would show up.

Fig. 1. Resistance measurements for non-etched ITO-films with 2-strips method. (A) The
horizontal axis represents the distance between the strips and the resistance measured
before etching (at t = 0) coincides with that measured after 2 min and (B) contact resistance
as a function of etching time.

permeation of etchant through this rough and porous layer is assumed

to be large to enable a fast etching of the underlying ITO-bulk layer. So,
what is actually happening is that the ITO bulk layer is etched, while a
porous layer is being formed as a consequence of the etching process.
This porous and rough layer becomes the new surface layer of the film.
In the model presented here we assume this erosion of the bulk ITO to
be starting after the original top layer has been removed. In reality
roughening will start right from the beginning, probably at a low rate
and accelerating after the initial top layer has been removed. However,
this consideration does not affect the basic conclusions that can be
derived from the model.
When the bulk ITO has been consumed by the etchant molecules,
the remaining rough and porous layer is slowly removed by the etchant.
This model is explained in Fig. 3a–c.
These figures illustrate our model consisting of 3 layers: a thin top
layer, indicated by the subscript t, the porous layer indicated by the
subscript p and the virgin ITO in the bulk, indicated by the subscript b.
The ITO-film is characterized by the thicknesses et, ep and eb, resistivities
ρt, ρp and ρb and etch rates kt, kp and kb. Since the porous layer is formed
while the bulk material is being consumed, we need also to introduce
Fig. 2. Normalized conductance R0 /R versus time of etching 25 nm ITO films (A) at
the formation rate of the porous film, being fp. The three layers t, p 50 °C in oxalic acid at various concentrations (indicated in molality) and (B) at various
and b are supposed to be parallel. To complete the model, we assume temperatures in 0.08 m oxalic acid solution.
24 S.S. Mammana et al. / Thin Solid Films 567 (2014) 20–31

Fig. 3. (A) Porosity model for etching ITO films where the thickness is indicated by e, resistivity by ρ and etch rate by k (the subscripts t, p and b refer to top, porous and bulk layers
respectively), (B) simulation of R0/R curves for porosity model with a parameter selection that represents the experimental curves and (C) simulation of R0/R curves for porosity model
of the ITO film with a series resistance of 0, 0.3xR0 and R0 respectively.

In Section 4 we describe XRR-results that substantiate the presence of
a top layer with a slightly lower density. The thickness of this layer is
found to be 7.42 nm. A simulation with the parameter combination
et(0) = 7.42 and kt = 1.5 nm/min (all other parameters identical) gener-
ates a curve that coincides with curve 2 in Fig. 3b. Curve 2 in Fig. 3b can
also be represented when the erosion rate fp is much smaller: for exam-
ple when fp = 0.5 nm/min, then ρp = 3.5 × 10–4 and kp = 0.5 nm/min,
while the other parameters don't change.
What we can see in Fig. 3b is that the slope of the R0/R curve between
the breakpoints changes when the parameters of the porous layer are
 S.S. Mammana et al. / Thin Solid Films 567 (2014) 20–31
varied: in other words, it is impossible to determine the etch rate of the
bulk ITO-layer, kb, from this slope. From simulations as presented in
Fig. 3b we estimate that kb obtained from the slope of the linear part
of the R0/R-t diagram can be 25% smaller than kb calculated from curve
1 in Fig. 3b, in which the effect of a porous layer has been nullified.
Nevertheless, we have chosen the slope of the R0/R curve between the
breakpoints to determine the effective etch rate, because it includes
the slowing-down effects imposed by the rough, porous layer. The
results and analyses presented in the next sections are based on this
effective etch rate.
In Fig. 3c we show the effect of the series contact resistance on R0/R.
When the contact resistance Rs N 0.2R0, the effect becomes noticeable. In
our case, in which Rs is only 6% of R0, the effect of Rs may be neglected.
The experimental R0/R curves shown in Fig. 2a and b have smooth
transitions between the three etch phases, whereas the simulations
depicted in Fig. 3b and c have abrupt transitions. This behavior can
also be simulated by introducing more layers with gradual changes of
the relevant parameters.
The initial etching behavior of ITO films is strongly affected by the
cleaning process before etching. Fig. 4 shows that rigorous degreasing
of the ITO thin film affects the initial etching behavior. For that reason
we decided to degrease our samples in 5% Extran® MA02 using an ultra-
sonic bath before etching. This process reduced the initial etch time in
most cases and therefore we assume that a slow initial etch rate is
also related to surface cleaning (grease, dust, etc.) and does not provide
information on the bulk etch rate.
4. Morphology and structure analyses
We analyzed the ITO-samples with an X-ray diffractometer, measur-
ing reflectivity and X-ray diffraction pattern at grazing incidence, an
atomic force microscope, a scanning electron microscope and a surface
profilometer. The soda lime glass with 25 nm ITO films without a SiO2
diffusion barrier between glass and ITO had a roughness of the glass sub-
strate of about 0.48 nm, measured after etching off the ITO film. The
175 nm ITO films, having a thin SiO2 barrier layer between ITO and
glass, were very smooth. The surface analyses were made before and
after etching during various times in 0.3 m oxalic acid at 50 °C for the
25 nm samples and 1.2 m oxalic acid at 60 °C for the 175 nm samples.
For the GIXRD measurement the angle with respect to the sample
surface was adjusted to 1° in order to identify the ITO film on the
substrate [27]. The results showed the predominant occurrence of the
In2O3 (cubic crystal system), as expected for poly crystalline ITO films
and shown in Fig. 5.
Fig. 4. R0/R versus time of etching a 25 nm ITO film in 0.32 m oxalic acid at 60 °C. Curve 1:
with Extran degreasing. Curve 2: without degreasing in Extran.
25
Fig. 5. GIXRD diagram of 175 nm ITO film before etching.
The XRR measurements provide information on the structure of the
surface layer, since the specular X-ray reflection (XRR) is related to the
scattering effect that depends on the change of the average electron
density across an interface and not upon the periodic nature of electron
density associated with the crystal structure. It is possible to determine
thickness using the period of the interference fringes, roughness using
the amplitude and shape of the interference fringes and density from
the critical angle. Experimental and fitted XRR-curves are represented
in Fig. 6a for the 175 nm ITO-film before etching; Fig. 6b depicts the
XRR curves after 15 min etching and Fig. 6c shows these curves after
25 min etching. The relevant parameters for the films of 25 and
175 nm calculated from the XRR spectra are summarized in Table 1.
Fitting of the XRR curves was performed using the X'Pert Reflectivity
program provided by the PANalytical software suite in order to evaluate
density and layer thickness of the non-etched and etched films. By
representing the ITO film by more than one layer, allowance is made
for a grading density. So, the parameters to be fitted in this procedure
are density, number and thickness of the layers that represent the ITO
film.
The fitting process has been carried out the ITO film before etching
(Fig. 6a) with a major restriction on the density. It was constrained be-
tween 6.95 and 7.18 g/cm3, these are the values for the density of
In2O3-SnO2 (40:60) and pure In2O3 respectively [28,29]. After estimation
of the thickness value through the oscillation periods, the X'pert program
changed this value, also taking into account the interface/surface rough-
ness, providing the best fit through a least squares algorithm.
Fig. 6a shows two simulations: (1) an ITO film represented by a
single layer and (2) and ITO film represented by two layers, a top
layer of 7.4 nm with a slightly lower density than that of the bulk. This
latter fit shows that the two-layer representation is better than the
one-layer representation of the ITO film before etching. This top layer
with a slightly lower density could be ascribed to ITO enriched with
SnO2. SnO2 enrichment of the surface layer of an ITO film was reported
by Wu et al. [22].
For simulating the XRR-curves for ITO films after 15 and 25 min etch-
ing we used a three-layer model for the ITO film to account for the den-
sity gradient having 3 different layer thicknesses and interface
roughness. The only constraint on the density was that it was at maxi-
mum 7.18 g/cm3. Nevertheless, we kept our model as simple as possible,
i.e. added the minimum number of parameters. Fig. 6b and c show that
the best fits are obtained for the three-layer model, i.e. a grading of the
density.
The top layer has the lowest density, indicating a high porosity,
whereas the subsequent layers show higher density, indicating a less
open structure. The various ITO layers are indicated by 1, 2 and 3 in
Table 1. Besides the grading of the density, we can see a reduction of
26 S.S. Mammana et al. / Thin Solid Films 567 (2014) 20–31
a) before etching
b) 15min etching
c) 25min etching
Fig. 6. X-ray Reflectivity (XRR) profiles of 175 nm ITO film, (a) before etching, the inset
shows the experimental and calculated XRR profiles for one-layer and two-layers,
(b) 15 min etching, the inset shows the experimental and calculated XRR profiles for one-,
two- and three-layers and (c) 25 min etching, the inset shows the experimental and
calculated XRR profiles for one-, two- and three-layers.
the total thickness of the ITO film as the etching time increases. It should
be mentioned that the analysis of the 25 nm ITO film shows a reduction
in density and thickness as well, but the roughness remains almost the
same after 12 min etching in 0.3 m oxalic acid at 50 °C.
The 175 nm ITO sample before etching was investigated with an
atomic force microscope in order to measure the magnitude of the ITO
grain size. The results showed that the size of the grains was not
uniform, presenting a dispersion. We measured the order of magnitude
of the ITO crystallites using roughness and cross section analysis and we
found that the height of the grains (Rmax) is in the range of 3.1 nm to
16 nm and the lateral dimension of the grains is between 18.5 nm and
35 nm. This latter measurement enabled a comparison with the SEM
analyses.
Various 175 nm ITO samples were investigated with a scanning
electron microscope (SEM) before and after the etching process. The
SEM photographs are presented in Fig. 7. The initial surface of the ITO
film before etching presents clusters of crystallites, which are non-
uniformly distributed on the surface in different planes with mean
boundaries dimensions of approximately 500 nm, as we can see in
Fig. 7a. After 25 min of etching in 1.2 m oxalic acid at 60 °C the surface
roughness increases, as shown in Fig. 7b. Fig. 7c shows the residue of
an ITO film after 35 min of etching in 1.2 m oxalic acid at 60 °C: we
see unconnected clusters of crystallites non-uniformly distributed on
the surface. This explains the “infinite” resistance value measured for
this sample, as the film is no longer continuous and uniform. After 1 h
etching the ITO film is completely removed from the surface of the
glass substrate (Fig. 7d). This 1 h etching sample has been coated with
17 nm Au-film to prevent charging during the SEM analysis. We can
see in this micrograph the smooth surface of the SiO2 layer on top of
the soda lime glass substrate.
Fig. 8 shows a graph of the surface roughness (Ra) of the same
samples presented in Fig. 7 measured by profilometry. The surface
roughness was measured in three different points of the sample and
the average value increased with the etching time as can be observed
in the graph. This result agrees qualitatively well with the roughness
determined from the XRR-analyses presented in Table 1.
Fig. 9 shows a graph of the ratio of chemical elements using EDX
technique for the samples shown in Fig. 7. The EDX signals indicated
the dominant effect of the substrate: the Si-signal showed only a varia-
tion of approximately 10% as a function of etching time. For this reason,
the Si signal was taken as reference to represent the quantities of In and
Sn in the films in order to obtain a relative analysis. We maximized the
x-ray counts gathered in the EDX- spectra, moreover we counted for a
long time to get high accuracy. We took the standard deviation as root
square of the number of counts in the peaks. The error bars in Fig. 9
were determined using the relative deviation and the propagation in
the ratios of the signals [30].
We can see in Fig. 9 that the In/Si ratio decreases more strongly than
the Sn/Si ratio. The Sn/In ratio remains approximately constant for low
etching time but increases substantially at longer etching time indicat-
ing a preferential etching of In2O3 over SnO2 in ITO. From these results
it can be concluded that the uniformity assumption underlying the
linear part of the R0/R-curve breaks down at the end of the etching
process because of the preferential etching of In2O3.
5. ITO etching in oxalic acid
Fig. 10a shows the etch rates of 25 nm ITO-film in oxalic acid at 40,
50 and 60 °C as a function of the concentration of oxalic acid and
Fig. 10b shows the same etch rates as a function of the H+ concentration
of the oxalic acid solution. The proton concentration of the oxalic acid
solution depicted in Fig. 10b is calculated from the concentration of
oxalic acid (as weighted) with Ka1 = 0.05, being the first ionization
constant of oxalic acid in water:
þ ‐
H2 C2 O4 →H þ HC2 O4 : ð3Þ
The behavior shown in Fig. 10a and b is different from the results
described by Tsai and Wu [9]: they found that the etch rate becomes
constant at oxalic acid concentrations N 0.3 m for all temperatures.
 S.S. Mammana et al. / Thin Solid Films 567 (2014) 20–31 27

Table 1
XRR data of ITO films.

ITO films (nominal thickness) Etching time (min) Layers Density (g/cm3) Thickness (nm) Roughness (nm)

25 nm 0 ITO 7.13 22.1 0.8

25 nm 12 ITO 4.15 14.87 0.9
175 nm 0 Rich SnO2 6.95 7.42 3.40
ITO 7.17 175.99 1.65
SiO2 2.64 15.90 0.94
175 nm 15 ITO (1) 3.25 6.69 4.95
ITO (2) 6.24 5.13 5.91
ITO (3) 6.99 73.73 3.68
SiO2 2.64 15.90 0.94
175 nm 25 ITO (1) 0.85 3.54 3.36
ITO (2) 6.34 9.89 7.75
ITO (3) 6.99 23.68 6.57
SiO2 2.64 15.90 0.94
175 nm 35 ITO (1) 1.82 13.85 5.59
ITO (2) 3.24 2.21 0.50
SiO2 2.64 15.90 0.94

Fig. 10b indicates that between at 0.015 b [H+] b 0.23 m, which is equiv-
alent to 0.02 b [H2C2O4] b 1.35 m, the etch rate is almost linear with
[H+]; however, at low acidity this linearity breaks down. So, it cannot
be concluded that the etch rate of ITO in oxalic acid is first order with
[H+].
Another interesting difference between the etch rates of Tsai and
Wu and our results is the value of the etch rate. At 50 °C and 0.32 m
they found an etch rate of 180 nm/min, whereas we determined
2 nm/min, being a factor of 90 lower. It should be stressed that it is
precarious to compare the etch rates of ITO published by other authors,
since the etch rate can differ by more than a factor of 100, depending on
the ratio between amorphous and crystalline material in the ITO film
[5–7]. In other words, comparing etch rates of ITO films without consid-
ering the crystallinity does not make sense.

Fig. 7. SEM pictures of 175 nm ITO film after different etch times in 1.2 m oxalic acid at 60 °C (electric potential: 20.0 kV, spot size: 3.0, work distance: 8.0 mm) (a) without etching,
(b) 25 min etching, (c) 35 min etching and (d) 1 h etching (25.0 kV).
28 S.S. Mammana et al. / Thin Solid Films 567 (2014) 20–31

Fig. 8. Surface roughness (Ra) in different stages of etching 175 nm ITO film in 1.2 m oxalic
acid at 60 °C.

Fig. 11 shows the effect of adding ferric chloride (FeCl3) or ferric

nitrate (Fe(NO3)3) to a bath of oxalic acid with a concentration of 1.35
m. Ferric chloride is an oxidizing agent and frequently used in etching
ITO in combination with HCl. We found that the addition of these
two ferric salts to oxalic acid has a very different effect: FeCl3 is en-
hancing the etch rate substantially, whereas Fe(NO3)3 has no effect.
Fe(NO3)3 is a rather strong oxidizing agent, not only because of the
Fe3 + ion, but rather because of the nitrate ion according to the half
reaction
Fig. 10. (a) Etch rate of ITO in oxalic acid at 40, 50 and 60 °C as a function of oxalic acid
‐ þ concentration (in molality) and (b) same data plotted as a function of H+-concentration
NO3 þ 4H þ 3e→NO þ 2H2 O ð4aÞ (in molality).

This half reaction has a standard electrode potential of 0.96 V [28]

and it could lead to the following reaction:

‐ þ ions are N 109 at room temperature [32]: this means that ferric oxalate
2NO3 þ 2H þ 3H2 C2 O4 →6CO2 þ 2NO þ 4H2 O; ð4bÞ is likely to be formed according to:

which indicates that Fe(NO3)3 could consume oxalic acid and thus 2‐ 3þ þ
C2 O4 þ Fe →FeC2 O4 ð5Þ
might reduce the oxalic acid concentration in the etch vessel, explaining
the behavior observed in Fig. 11.
If reaction (5) takes place, the equilibrium of reaction (3) is shifted to
We made some tests whether reaction (4b) occurred in our etching
the right side because the HC2O− 2−
4 ions produce H+ and C2O4 . In other
experiments with oxalic acid and Fe(NO3)3 by determining the oxalic +
words, the solution gets more acid and eventually the H concentration
acid content in the reaction vessels after etching by titration with
could be as high as two times the molarity of the weighted oxalic acid. If
KMnO4 [31]. We found that after etching of ITO-films in etch baths con-
this happens, the etch rate will increase strongly according to Fig. 10b.
taining H2C2O4 and Fe(NO3)3 no change in the concentration of H2C2O4
Although reaction (5) cannot be excluded for FeCl3, for Fe(NO3)3 reac-
could be detected: in other words, reaction (4b) does not take place.
tion (5) must be excluded, since we did not find an enhancement of
Oxalic acid is forming complexes with many metal ions, also with
the etch rate with this salt. If we accept this conclusion, then by similar-
ferric ions. The complex forming constants between ferric and oxalate
ity we also exclude reaction (5) for FeCl3, because there is not a good
reason why there should be a difference between these two salts.

Fig. 9. Ratio of EDX signals in different stages of etching 175 nm ITO film in 1.2 m oxalic Fig. 11. Etch rate of 175 nm ITO film as a function of the concentration of additional FeCl3
acid at 60 °C. In/Si ratio, (×10) Sn/Si ratio and (×10) Sn/In ratio. and Fe(NO3)3 in 1.35 m oxalic acid at 60 °C.
 S.S. Mammana et al. / Thin Solid Films 567 (2014) 20–31 29

We also measured the etch rate of ITO in oxalic acid with KCl and did
not find an enhancement. This result indicates that for enhancing the
etching in oxalic acid one needs both ferric and chloride ions. Adding
a non-chloride ferric salt or a non-ferric chloride to the etch bath does
not enhance the etch rate.
Fig. 11 indicates that the etch rate increases monotonously by
adding FeCl3 to the oxalic acid solution. This is quite different from the
behavior of adding FeCl3 to concentrated solutions of HCl, as described
by van den Meerakker et al. [2]. In the case of HCl, they observed a
large increase of the etch rate by adding only a small quantity of FeCl3.
In the discussion section we shall comment on the large difference
between FeCl3 and Fe(NO3)3 and the differences with the observations
of van den Meerakker et al.
Although most R0/R-curves obtained by us showed the behavior as
shown in Fig. 2, etching of ITO films in oxalic acid with FeCl3 yielded Fig. 13. Arrhenius plots of the etch rate of 25 nm ITO film in 0.02, 0.08 and 1.35 m oxalic
R0/R-curves that did not indicate a slowing down of the etch rate at acid.
the end. This might be explained by a reaction between oxalic acid
and ferric chloride; however, we did not investigate this deviation any
further. We did also some etching experiments of ITO films in 3.2 M HCl at
Fig. 12 shows that by adding only a small quantity of sodium 40, 50 and 60 °C. The etch rate at 50 °C in HCl is 3 times larger than in
hypophosphite (0.01 m) to the oxalic acid bath (0.32 m), we already ob- 1.35 m oxalic acid. At 40 °C and 60 °C it is 3.5 and 2.5 times larger re-
serve a drastic decrease of the etch rate of 30% and at 0.04 m NaH2PO2 spectively. We found that the activation energy for the etch rate of ITO
this reduction is more than 55%. Because of the large excess of oxalic in 3.2 M HCl is 56 ± 5 kJ/mol. This is identical with the result of van
acid at these conditions, a reaction between oxalic acid and sodium den Meerakker et al. [2].
hypophosphite can be excluded. Instead of that we assume that the
hypophosphite ion is inhibiting the reaction between H+ and In2O3 by 6. Discussion
forming an insoluble In-POx compound at the surface of the ITO-film.
In the case of adding ascorbic acid to an etch bath of 1.2 m oxalic acid The inverse S-shape of the R0/R-curves in Figs. 2 and 3 is explained in
of 60 °C, we did not find a change in the etch rate for concentrations
the porosity model by the slow etching of a thin top layer and the for-
of added ascorbic acid in the range between 0 and 0.6 m. This is different mation of a rough, porous layer that has a larger resistivity than the
from the results described by Lee [26], who found an acceleration of the
ITO bulk layer. The porosity model presented here is thus inherently
ITO etch rate by adding ascorbic acid to an oxalic acid etchant. Since connected to the etching process and is not related to the morphology
ascorbic acid is a weak reducing agent, it cannot enforce any reducing
or composition of the film before etching. Nevertheless, a static three-
activity of oxalic acid. This reasoning may explain our results. layer model of the ITO-film before etching may also represent the in-
Fig. 13 shows Arrhenius plots of the etch rates of ITO films in oxalic
verse S-shape of R0/R upon etching ITO films. Van den Meerakker et al.
acid at various concentrations. From the slopes of the Arrhenius plots [5] and Hoheisel et al. [7] found that at certain deposition conditions
the activation energy Ea of the etch rate was calculated, using the
of ITO a bottom layer with a more amorphous structure could be depos-
relation: ited. This would lead to an acceleration of the etch rate, in contrast with
our results. Based on the results presented in Section 4 we think that the
r t ¼ r 0 expð−Ea = RT Þ; ð6Þ dynamic porosity model of ITO-etching is a better representation of the
process.
Where rt is the etch rate at temperature t, r0 is the etch rate at a In Section 3 we have shown that the etch rate derived from the lin-
reference temperature, R is the gas constant and T is the temperature ear part of the R0/R curve cannot directly be related to the bulk etch rate
indicated in Kelvin (K). Fig. 14 shows the activation energy of the etch of ITO, but depends also on the properties of a rough and porous layer
rate of ITO-films determined in various concentrations of oxalic acid. It formed during etching. The effective etch rate derived from these curves
can be seen that the activation energy of ITO in a bath of oxalic acid is is considered to be a realistic representation of the etch rate: it ignores
80 ± 5 kJ/mol and does virtually not depend on the concentration. the slow start of the etching, but it rather includes the roughness effects
This value is larger than the activation energy, 66.8–70 kJ/mol, as deter- created during etching. The error in the effective etch rate by not taking
mined by Tsai and Wu [9]. into account the slow start of the etching is considered to be rather

Fig. 12. Etch rate of 25 nm ITO film at 40, 50 and 60 °C as a function of the concentration of Fig. 14. Activation energy of etch rate of 25 nm ITO film in oxalic acid. The line is inserted to
sodium hypophosphite in 0.32 m oxalic acid. guide the eye.
30 S.S. Mammana et al. / Thin Solid Films 567 (2014) 20–31
small as follows from the curves presented in Figs. 2a, b and 4. Fig. 6a
and Table 1 show that the fit between the theoretical and experimental
XRR-curves of an ITO film before etching improves considerably by
taking into account a thin top layer with a slightly lower density. This ma-
terial with a lower density may tentatively be ascribed to ITO enriched
with SnO2. If this is the case, it would also explain why the surface layer
has a resistivity larger than that of bulk ITO (1.8 × 10−4 Ω.cm) and
why the etch rate kt is lower than the effective kb. The SEM investigations
described in Section 4 show that etching of ITO is a selective process: dur-
ing etching In2O3 is being dissolved faster than SnO2; so, the porous layer
is continuously enriched with SnO2. At the end of the etching process the
total SnO2 content of the film is larger than in the beginning. The residual
SnO2 is still conducting, but the removal rate from the substrate is largely
reduced because pure SnO2 does not dissolve in oxalic acid or HCl.
So, the most obvious explanation of the almost horizontal slopes of
the R0/R curve at the beginning and the end is Sn-enrichment of ITO.
Sn-enrichment in the surface layer of ITO films has been found by Wu
et al. [22]. Preferential etching or leaching of In from ITO-powders in
acidic solvents such as HCl or H2SO4 is well-known and was described
by various authors [33,34]. Preferential etching of In from poly crystal-
line ITO films has not been described in the literature. Therefore, let us
elaborate in more detail on this interesting observation.
The porous layer that remains on the substrate after the bulk ITO has
been consumed by the etchant is not pure SnO2 but still contains In2O3:
so, dissolving In3+ ions may drag Sn4+ ions from the lattice and thus the
residual SnO2 may be removed as well, albeit at a low rate. In Table 2 we
have listed some reactions that may take place in etching ITO. The
corresponding standard Gibbs energies for the reactions, calculated
from the data in Ref. [28] are listed as well. Both amorphous and crystal-
line forms of In2O3 dissolve in acids because of the negative Gibbs energy.
The Gibbs energy is a thermodynamic quantity, which does not provide
information on the kinetics, i.e. the etch rate. The second reaction of H+
with SnO2 shows a positive value for ΔG0 indicating that SnO2 does not
dissolve in acids. To dissolve SnO2 in acids, a reducing agent is necessary
that converts SnO2 into SnO, as shown by the third reaction. The differ-
ence in Gibbs energies between the first and second reactions explains
why it is to be expected that Sn-rich ITO will show a low etch rate in
acid etchants, whereas Sn-poor samples will etch readily. In other
words, the thermodynamic data in Table 2 and published data on
leaching rates of In and Sn from ITO [33,34] in acidic etchants agree
with the Sn-enrichment in the remainder of the ITO-film as shown in
Fig. 9.
Both In3 + and Sn2 + ions can form complex compounds with the
oxalate ion; however, the thermodynamic data are incomplete to make
firm conclusions. If the essential reaction of etching ITO in acids is
represented by the first reaction, then it is obvious that In2O3 is preferen-
tially etched and that residues on the glass are enriched with insoluble
SnO2. The quasi-linear dependence of the etch rate of ITO versus [H+]
in oxalic acid as shown in Fig. 10b supports the assumption that the
first reaction in Table 2 is the most important.
Finally, let us consider the possibility that the non-linearity of the R0/R
curves is related to humidity. Gas sensing devices based on ITO and SnO2
thin films were described in the literature [35,36]. The effect of oxygen-
containing vapors such as water, methanol and ethanol on ITO and SnO2
is decreasing the resistance of ITO and SnO2. An ITO film that has been
stored for a long time in humid air could get a surface layer with a
lower resistance, whereas a film stored in very dry air may get a high
Table 2
Reactions and Gibbs free energy (kJ/mol) [28].
Reaction ΔG0
+ 3+
In2O3 + 6H → 3H2O + 2In −88.15
SnO2 + 4H+ → 2H2O + Sn4+ +38.37
SnO + 2H+ → H2O + Sn2+ −6.12
resistance. Our samples were stored at ambient conditions of about
50% relative humidity; so, we can rule out possible effects of humidity
at the beginning of the etching. In the end phase the effect of humidity
cannot be completely ruled out, because if the Sn/In atomic ratio in the
remainder of the film is changing during etching, then a change in resis-
tance cannot be assigned conclusively to either composition change or
moisture effects.
The etch rates of ITO-films in oxalic acid, represented in Fig. 10a, are
much lower than those determined by Tsai and Wu [9]. Since we also
determined the etch rate of ITO in 3.2 M HCL, we can compare our
results with those of van den Meerakker et al. [2]. They found an etch
rate of 2 nm/min in 4 M HCL at 30 °C and an activation energy of the
etch rate of 55 kJ/mol in 6 M HCl. If we assume that the activation ener-
gy for etching ITO in HCL does not depend on the concentration, then
their etch rate of ITO in 4 M HCL at 40 °C would be 4 nm/min, which
is very similar to our result being 3.8 nm/min in 3.2 M HCl at 40 °C.
This indicates that the ITO-films used by van den Meerakker et al. and
us are similar regarding crystallinity.
Van den Meerakker et al. proposed a model [2], which explains the
enhancement of the etch rate by adding FeCl3 to concentrated HCl. The
basic idea is that Fe3+ ions generate positively charged indium sites at
the surface of the ITO film by forming OH radicals and Fe2+ ions. The
charged indium sites react with Cl− ions, while the OH radicals and
Fe2+ are immediately transformed to OH− and Fe3+. This is a catalysis
model that can also explain the enhancement of etching ITO in oxalic
acid, since the dependence of the etch rate with FeCl3 concentration,
shown in Fig. 11, also fits in the catalysis model of van den Meerakker
et al. Since our experiments with Fe(NO3)3 did not show any increase of
the ITO etch rate in oxalic acid, we tentatively conclude that the large
nitrate ion cannot come close enough to the charged In-site to get bond-
ed. In other words, accelerating the etching of ITO in oxalic acid also
requires (a) electron capture by the Fe3+ ion and (b) neutralization of
the formed In+-site with a Cl− ion.
Finally we like to make some comments on using our resistance mon-
itoring method in an industrial environment. The R/R0-curves that are
obtained using this method and the identification of the three regimes
can be used for: (a) comparing ITO film preparation processes: best pro-
cesses don't generate a top layer with a low etch rate; (b) comparing
etchants and etching processes: best processes have large kb and large
kp; (c) optimizing cleaning processes: best cleaning creates either small
et or large kt. The monitoring method does not answer the question of
the complete removal of residues beyond the percolation limit. For
quality assurance of the etching process, additional analyses such as
SEM have to be made.
7. Conclusions
We present a study on the kinetics of wet etching of conductive thin
films by monitoring the resistance. The method has been tested with
thin crystalline films of ITO of 25 and 175 nm in aqueous solutions of
oxalic acid: it yields insight in the kinetics of the etching process. The
most interesting result is that this simple monitoring method indicates
that etching ITO films in acidic solutions does not proceed linearly with
time, but is slow in the beginning, then accelerates and slows down at
the end. This behavior can be well represented by a porosity model of
ITO: after removal of a top layer from the ITO-film, a rough and porous
layer is created on top of the bulk ITO, which remains at the surface after
the bulk layer has been consumed by the etchant. Complete removal of
this porous layer from the substrate requires substantial additional
time, which cannot be derived from the resistance monitoring method
itself. By combining the porosity model with the XRR-results we tenta-
tively conclude that the top layers of our ITO films were enriched with
SnO2, explaining the lower density and the slow etch rate.
The slow etch rate at the end of the etching process can certainly be
ascribed to enrichment of SnO2 in the remainder of the ITO film. This
conclusion is based on analyses of the SEM/EDX results.
 S.S. Mammana et al. / Thin Solid Films 567 (2014) 20–31
We found that oxalic acid is a suitable etchant for ITO, yielding 2–3
times lower etch rates than concentrated HCl at the same temperature.
The etch rate in oxalic acid can be accelerated substantially by adding
FeCl3 to oxalic acid, whereas adding Fe(NO3)3 to oxalic acid has no
effect. The acceleration of the etch rate by FeCl3 is explained in terms
of a catalysis model. The substantial slowing down of the etch rate
with a small amount of sodium hypophosphite is explained by assum-
ing the formation of an insoluble In-POx compound at the surface of
the ITO-film.
Because of the low cost, high efficiency and accuracy, we suggest
that our resistance monitoring process based on a sturdy two-strips-
measuring jig with a very low contact resistance can easily be applied
in an industrial environment.
Acknowledgements
We are grateful to the staff of the Electron Microscopy Laboratory
(LME) and Scanning Probe Microscopy Laboratory (SPM) of the Nation-
al Nanotechnology Laboratory (LNNano) of the National Materials and
Energy Research Center (CNPEM) in Campinas (Brazil) for technical
support during electron microscopy work and atomic force microscopy
work and to the staff of the Photovoltaic Research Laboratory of the Ap-
plied Physics Department of the State University of Campinas
(UNICAMP) for the roughness measurements with the profilometer.
References
[1] S. Morozumi, Materials and process of LCDs, in: B. Bahadur (Ed.), Liquid crystals —
applications and uses, vol. 1, World Scientific Publishing Co., Singapore, 1990, p. 183.
[2] J.E.A.M. van den Meerakker, P.C. Baarslag, M. Scholten, On the mechanism of ITO etching
in halogen acids: the influence of oxidizing agents, J. Electrochem. Soc. 142 (1995)
2321.
[3] M. Venkatesan, S. McGee, U. Mitra, Indium tin oxide thin films for metallization in
microelectronic devices, Thin Solid Films 170 (1989) 151.
[4] C.J. Huang, Y.K. Su, S.L. Wu, The effect of solvent on the etching of ITO electrode,
Mater. Chem. Phys. 84 (2004) 146.
[5] J.E.A.M. van den Meerakker, P.C. Baarslag, W. Walrave, T.J. Vink, J.L.C. Daams, On the
homogeneity of sputter-deposited ITO films. Part II. Etching behaviour, Thin Solid
Films 266 (1995) 152.
[6] E. Nishimura, H. Ohkawa, P.K. Song, Y. Shigesato, Microstructures of ITO films
deposited by d.c. magnetron sputtering with H2O introduction, Thin Solid Films
445 (2003) 235.
[7] M. Hoheisel, A. Mitwalsky, C. Mrotzek, Microstructure and etching properties of
sputtered indium-tin oxide (ITO), Phys. Status Solidi (a) 123 (1991) 461.
[8] T.-H. Tsai, Y.-F. Wu, Wet etching mechanisms of ITO films in oxalic acid, Microelectron.
Eng. 83 (2006) 536.
[9] T.-H. Tsai, Y.-F. Wu, Organic acid mixing to improve ITO film etching in flat panel
display manufacturing, J. Electrochem. Soc. 153 (2006) C86.
[10] P. Patnaik, Handbook of inorganic chemicals, McGraw Hill, New York, 2002. 394.
[11] C.-Y. Lee, C. Chang, W.P. Shih, C.L. Dai, Wet etching rates of InGaZnO for the fabrication
of transparent thin-film transistors on plastic substrates, Thin Solid Films 518 (2010)
3992.
[12] G.B. González, T.O. Mason, J.P. Quintana, O. Warschkow, D.E. Ellis, J.-H. Hwang, J.P.
Hodges, J.D. Jorgensen, Defect structure studies of bulk and nano-indium-tin
oxide, J. Appl. Phys. 96 (2004) 3912.
31
[13] S.S. Mammana, A. Greatti, F.H. Luiz, F.I. da Costa, A.P. Mammana, D. den Engelsen, C.I.
Z. Mammana, A new method for studying the kinetics of wet etching thin films of
ITO, Proc. LatinDisplay 2012/IDRC, 2012.
[14] S.S. Mammana, A. Greatti, F.H. Luiz, F.I. da Costa, A.P. Mammana, D. den Engelsen, G.
A. Galligaris, L.P. Cardoso, C.I.Z. Mammana, Etching of thin films of ITO in oxalic acid,
Proc. LatinDisplay 2012/IDRC, 2012.
[15] I. Zudans, C.J. Seliskar, W.R. Heineman, In situ measurements of sensor film dynamics
by spectroscopic ellipsometry. Demonstration of back-side measurements and the
etching of indium tin oxide, Thin Solid Films 426 (2003) 238.
[16] G.F. Mendes, L. Cescato, J. Frejlich, E.S. Braga, A.P. Mammana, Continuous optical
measurement of the dry etching of silicon using the diffraction of a lamellar grating,
J. Electrochem. Soc. 189 (1985) 190.
[17] E.S. Braga, G.F. Mendes, J. Frejlich, A.P. Mammana, Optical monitoring of the end
point in thin film plasma etching, Thin Solid Films 109 (1983) 363.
[18] G.F. Mendes, L. Cescato, E.S. Braga, J. Frejlich, A.P. Mammana, Plasma etching rate
measurement using the diffraction of a lamellar grating, Thin Solid Films 117
(1984) 107.
[19] J.W.M. Jacobs, H.F. van Rooijen, J.E.A.M. van den Meerakker, T.J. Vink, Microstructural
and electrochemical aspects of wet etching ITO films, in: Y. Kuo (Ed.), Thin film
transistor technologies III, Electrochem. Soc. Proc, 158, 1997.
[20] Y. Gassenbauer, A. Klein, Electronic and chemical properties of tin-doped indium
oxide (ITO) surfaces and ITO/ZnPc interfaces studied in-situ by photoelectron
spectroscopy, J. Phys. Chem. B110 (2006) 4793.
[21] S.P. Harvey, T.O. Mason, Y. Gassenbauer, R. Schafranek, A. Klein, Surface versus bulk
electronic/defect structures of transparent conducting oxides: I. Indium oxide and
ITO, J. Phys. D. Appl. Phys. 39 (2006) 3959.
[22] B. Low, F. Zhu, K. Zhang, S. Chua, Improvement of hole injection in phenyl-substituted
electroluminescent devices by reduction of oxygen deficiency near the indium tin
oxide surface, Appl. Phys. Lett. 80 (2002) 4659.
[23] M.G. Mason, L.S. Hung, C.W. Tang, S.T. Lee, K.W. Wong, M. Wang, Characterization of
treated indium-tin-oxide surfaces used in electroluminescent devices, J. Appl. Phys.
86 (1999) 1688.
[24] H.Y. Yu, X.D. Feng, D. Grozea, Z.H. Lu, R.N.S. Sodhi, A.-M. Hor, H. Aziz, Surface
electronic structure of plasma-treated indium tin oxides, Appl. Phys. Lett. 78
(2001) 2595.
[25] P.D.C. King, T.D. Veal, D.J. Payne, A. Bourlange, R.G. Egdell, C.F. McConville, Surface
electron accumulation and the charge neutrality level in In2O3, Phys. Rev. Lett. 101
(2008) 116808.
[26] K.W. Lee, ITO etching composition, patent nr. WO 00/11107 (2000).
[27] U. Pietsch, V. Holý, T. Baumbach, High-resolution X-ray scattering from thin films to
lateral nanostructures, Spinger-Verlag, New York, 2004.
[28] Handbook of chemistry and physics, 84th ed. CRC Press, Boca Raton, 2004.
[29] http://en.wikipedia.org/wiki/Indium(III)_oxide .
[30] D.B. Williams, C.B. Carter, Transmission electron microscopy — a text book for material
science, second ed. Springer Science Media and Business Media Press, New York, 2009.
[31] I.M. Issa, A.A. El-Aasser, M.M. El-Merzibani, Low temperature oxidation of oxalic acid
with potassium permanganate, Fresenius' Z. Anal. Chem. 178 (1960) 12.
[32] J.G. Speight, Lange's handbook of chemistry, 16th ed. McGraw-Hill, New York, 2005.
1373.
[33] Y. Li, Z. Liu, Q. Li, Z. Liu, L. Zeng, Recovery of indium from used indium-tin oxide
(ITO) targets, Hydrometallurgy 105 (2011) 207.
[34] S. Virolainen, D. Ibana, E. Paatero, Recovery of indium from indium tin oxide by
solvent extraction, Hydrometallurgy 107 (2011) 56.
[35] G.Y. Cha, W.W. Baek, S.T. Lee, J.O. Lim, J.S. Huh, Effects of crystal structure and particle
size on ethanol gas sensing characteristics of nanocrystalline ITO thick film, J. Ceram.
Soc. Jpn. 112 (2004) 252.
[36] M. Parthibavarmana, V. Hariharan, C. Sekar, High-sensitivity humidity sensor based
on SnO2 nanoarticles synthesized by microwave irradiation method, Mater. Sci. Eng.
C 31 (2011) 840.