Reduction of tin oxide by hydrogen radicals

Article abstract of DOI:10.1021/jp981447l

The effect of a reducing hydrogen ambient on textured tin oxide thin films on glass substrates has been investigated. Hydrogen treatments were done at 230 and 430 °C by hot wire (HW) and rf plasma-decomposed hydrogen with pure H₂ as source gas. By these treatments the possible reduction of the substrate during the deposition of a-Si:H for solar cells is simulated. Ion beam techniques revealed that the exposure to HW-decomposed H-radicals leads to the formation of a tin-rich surface layer of 40 nm in 1 min at both 230 and at 430 °C. The loss of oxygen is higher for the high-temperature treatment. The optical transmission at a wavelength of 800 nm is reduced from 80% to less than 20%, while the sheet resistance increases from 6 to 8 Ω/□. At both temperatures the reduction of fluorine-doped tin oxide (FTO) by a HW-treatment occurs faster than by rf plasma-decomposed H. The H radical concentration, which is higher for the HW-decomposed hydrogen as compared to rf plasma-decomposed hydrogen, is the most important factor in determining the rate of the reduction process. For short exposures to H radicals, the transparency and conductivity of the tin oxide may be completely restored by means of reoxidation in air at 400 °C. In contrast, prolonged exposure to H-radicals induces irreversible loss of transparency and conductivity, concomitant with formation of granule-like particles of metallic tin on the surface. A thin plasma-deposited a-Si:H-layer was found to effectively protect the FTO-layer against reduction due to HW-generated H-radicals.

Full text of DOI:10.1021/jp981447l

J. Phys. Chem. B 1998, 102, 6219-6224
6219
Reduction of Tin Oxide by Hydrogen Radicals
J. Wallinga,* W. M. Arnoldbik, A. M. Vredenberg, R. E. I. Schropp, and W. F. van der Weg
Interface Physics, Debye Institute, P.O. Box 80000, NL-3508 TA Utrecht, The Netherlands
ReceiVed: March 11, 1998; In Final Form: May 18, 1998
The effect of a reducing hydrogen ambient on textured tin oxide thin films on glass substrates has been
investigated. Hydrogen treatments were done at 230 and 430 °C by hot wire (HW) and rf plasma-decomposed
hydrogen with pure H2 as source gas. By these treatments the possible reduction of the substrate during the
deposition of a-Si:H for solar cells is simulated. Ion beam techniques revealed that the exposure to HW-
decomposed H-radicals leads to the formation of a tin-rich surface layer of 40 nm in 1 min at both 230 and
at 430 °C. The loss of oxygen is higher for the high-temperature treatment. The optical transmission at a
wavelength of 800 nm is reduced from 80% to less than 20%, while the sheet resistance increases from 6 to
8 Ω/0. At both temperatures the reduction of fluorine-doped tin oxide (FTO) by a HW-treatment occurs
faster than by rf plasma-decomposed H. The H radical concentration, which is higher for the HW-decomposed
hydrogen as compared to rf plasma-decomposed hydrogen, is the most important factor in determining the
rate of the reduction process. For short exposures to H radicals, the transparency and conductivity of the tin
oxide may be completely restored by means of reoxidation in air at 400 °C. In contrast, prolonged exposure
to H-radicals induces irreversible loss of transparency and conductivity, concomitant with formation of granule-
like particles of metallic tin on the surface. A thin plasma-deposited a-Si:H-layer was found to effectively
protect the FTO-layer against reduction due to HW-generated H-radicals.
1. Introduction
is well above the melt temperature of tin, i.e., 232 °C. Above
the melt temperature, clustering of tin after the reduction of the
tin oxide may occur, which causes irreversible changes to the
tin oxide matrix.
Wanka et al.6 studied the interface between FTO and PECVD
a-Si:H by XPS, using sputtering to obtain depth profiles. It
was found that an SiOx barrier layer with a deposition-rate-
dependent thickness had formed at the interface during deposi-
tion. For a high deposition rate, the resulting SiOx layer was
thinner, indicating that shielding of the FTO-layer by the a-Si:H
reduces the amount of reduction taking place.
The deposition of a-Si:H by hot wire chemical vapor
deposition4 (HWCVD) usually takes place at substrate temper-
atures above 400 °C.7 This deposition method enables the
preparation of device-quality films with a low hydrogen content,8
and the layers are reported to be less sensitive to the creation
of metastable defects7 than those deposited by PECVD.
Together with the high deposition rates of typically 1 nm/s,8
this makes the material very promising for application in a-Si:H
solar cells. During the deposition of a-Si:H by HWCVD, the
FTO-substrate is exposed to a mixture of silicon and hydrogen
radicals, while no ions are involved.7 So far, it is not known
whether the rate of the reduction of the FTO-layer by this
method is comparable to what is observed with PECVD.
For tin-doped indium oxide layers (ITO) without surface
texture, the effects of HW-decomposed and rf plasma-
decomposed hydrogen have been compared by Lan and Kan-
icki.9 They showed that the effects of reduction of the indium
oxide, i.e., a decrease in optical transmission, and an increase
in sheet resistance are qualitatively the same for HW-
decomposed and rf plasma-decomposed hydrogen treatments
at 300 °C. However, because a textured layer is essential in a
p+-i-n+ a-Si:H solar cell structure in order to increase the
light absorption in a thin-film solar cell due to light scattering,10
An amorphous silicon p+-i-n+ solar cell is usually deposited
on a rough fluorine-doped tin oxide (FTO) front contact by
plasma-enhanced chemical vapor deposition (PECVD).1 During
the initial stages of deposition of the hydrogenated amorphous
silicon (a-Si:H), the FTO-layer is exposed to a silane plasma,
which contains hydrogen radicals and ions, next to Si-containing
species. From experiments where only H2 was used as a process
gas, and thus no layer is deposited, it is known that the hydrogen
radicals and ions in the plasma interact with the FTO-layer2
and cause a reduction of the tin oxide. This leads to a decrease
in the transmission and an increase in the sheet resistance of
the layers.
Schade et al.3 found that the effect of a hydrogen plasma
exposure is influenced by an external dc-bias voltage. The dc-
bias was interpreted to lead to a change of the ion flux. The
results indicated that an additional bombardment with hydrogen
ions during an H-radical treatment leads to enhanced reduction.
In our opinion, however, applying an external bias, which leads
to an increase of the deposition rate, will increase the radical
flux.4 This implies that we consider it doubtful whether ions
are the cause of the observed enhancement of the reduction of
tin oxide.
In the range of substrate temperatures used for PECVD, i.e.,
from 100 to 250 °C, the loss of transmission and increase of
sheet resistance become more prominent as temperature in-
creases.3,5 Reoxidizing the samples is possible by a thermal
annealing treatment in air at 400 °C for a few hours.2 This
treatment results in an almost complete recovery of the
optoelectronic properties to the nontreated state. This restoration
is certainly not obvious, since the temperature of the treatment
* Corresponding author. E-mail: J.Wallinga@phys.uu.nl. FAX: +31 30
254 3165.
S1089-5647(98)01447-3 CCC: $15.00 © 1998 American Chemical Society
Published on Web 07/17/1998

6220 J. Phys. Chem. B, Vol. 102, No. 32, 1998
Wallinga et al.
TABLE 1: Conditions Used for Tin Oxide Treatmenta
The chemical binding state of tin and oxygen in the first ∼10
nm at the surface of an FTO-layer before treatment and of
sample HTHW1 were determined by X-ray photoelectron
spectroscopy (XPS) with the KR X-rays obtained from electron
bombardment of an Al-target. The XPS data were taken ex
situ.
The surface morphology of FTO-layers before treatment, the
H-treated samples, and the reoxidized samples was examined
by scanning electron microscopy (SEM), using a Philips XL30-
FEG electron microscope. Element analysis was performed on
these samples, by energy dispersive X-ray (EDX) analyses of
the Sn L-lines and the O K-lines, using electron energies of 7
keV.
The depth profile of tin upon reduction due to a H-treatment
was studied by means of Rutherford backscattering spectrometry
(RBS). Elastic recoil detection (ERD) was used to determine
the oxygen concentration profiles. The analysis of the ion
spectroscopy measurements is complicated by the native surface
texture of the FTO-layers and the change of morphology upon
treatment. Because of the spot sizes of at least 1 × 1 mm2, the
obtained signals are an average over the various facet angle
orientations at the sample surface.
For the RBS-analysis a 1 MeV He+ ion beam produced by a
3 MeV single-ended Van de Graaff accelerator was used. The
angle between the beam and the sample surface amounted to
90°. Scattered particles were detected under an angle θ of 170°,
relative to the beam, as indicated in the inset of Figure 4a. The
RBS-spectra were analyzed using the computer code RUMP.13
Six curves have been calculated for different facet angles of
plus and minus 23, 33, and 43° for all samples, thus mimicking
the surface roughness as known from XTEM. The presented
simulations are the average of these six curves and thus account
for the surface roughness of the FTO-layers. From the height
of the tin profiles of the treated samples compared to that of
the nontreated ones, we extracted the relative variation in the
tin content.
The change in oxygen concentration of the top layer is
determined by elastic recoil detection (ERD).14 A 45 MeV
63Cu-beam produced by a 6.5 MeV tandem Van de Graaff
accelerator was used. The angle R of the incoming beam with
the surface was varied between 17.5 and 24.5°, and the recoil
angle φ was 30-37°. A 12 µm Mylar foil in front of the stop-
detector was used to obtain background-free ERD signals.
Because of the small angles involved in these measurements,
the influence of surface roughness was expected to be large.15
However, in comparing the ERD-signal of a rough and a
polished FTO-layer, we found relatively small changes. More-
over, because the results for the different measurement con-
figurations are consistent, qualitative comparison of the results
is possible. We present here data for R ) 17.5° and φ ) 30°.
The measurement configuration is schematically depicted in the
inset of Figure 6.
treatment
HTPE10,
deposition
HTPE4,
samples
LTHW1 HTHW1 LTPE10 HTPE1 HWSi PESib
time (min)
1
1
10
10, 4, 1
430
PE
0.17
430
HW
1
230
PE
Ttreat (°C)
230
HW
20
430
HW
20
230
PE
50
HW or PE
H2 (sccm)
SiH4 (sccm)
p (µbar)
50
90
20
40
700
20
20
700
700
a LT and HT denote low and high temperature. PE and HW denote
plasma-decomposed and hot wire decomposed hydrogen treatment,
respectively. b Sample PESi was treated with HW-hydrogen radicals
at 430 °C for 1 min after deposition.
we study FTO-layers here. They have inherent surface texture,
arising from the deposition of the layers on glass substrates by
atmospheric pressure CVD.
The purpose of this study is to compare the degradation of
APCVD tin oxide due to a treatment with hydrogen radicals
from a hot wire and with a hydrogen plasma. Both the changes
in chemical state and in surface morphology of textured SnO2:F
substrates were examined. Using ion beam techniques, depth
profiles of the compositions have been obtained. Annealing
treatments in air were performed to study the reoxidation
behavior of the samples after the H treatments.
2. Experimental Section
For all experiments U-type SnO2:F transparent conducting
oxide layers on glass substrates from Asahi Glass Co. were used.
The thickness of the APCVD-deposited polycrystalline FTO-
layers is ∼1 µm, and the height variations are around 50 nm
over one crystal with a diameter of roughly 150 nm. The angle
between the surface plane and the facets of the FTO-crystals,
as determined by cross-sectional transmission electron micros-
copy (XTEM), is 33° on average but is distributed between 12
and 54°.
Both the plasma- and the HW-decomposed H-treatments of
the FTO-substrates were performed in the PASTA11 deposition
system. The used hot-wire assembly consists of two tungsten
wires with an effective length of 12 cm, situated 4 cm apart.12
A wire temperature of 1850 °C was used for the HW
decomposition of H2. This wire temperature is well above the
minimum temperature at which H2 is effectively decomposed.
All plasma-decomposed hydrogen treatments were performed
at 13.56 MHz at a power of 3 W. The preparation conditions
and sample denotations are summarized in Table 1.
To determine the influence of the substrate temperature on
the reduction reaction, a low-temperature (LTHW1) treatment
at 230 °C and a high-temperature (HTHW1) treatment at 430
°C were performed by hot wire decomposed hydrogen for 1
min. For comparison a plasma-decomposed hydrogen treatment
was performed for 10 min at 230 °C (LTPE10) and 430 °C
(HTPE10). Shorter plasma-decomposed H-treatments of 1 and
4 min were performed at 430 °C (HTPE1 and HTPE4).
Reoxidation of these samples was performed in air at 400 °C
for 120 min. In addition, two samples were made to study the
influence of the deposition of ∼15 nm a-Si:H on a tin oxide
substrate. One layer was deposited by HWCVD at 430 °C
(HWSi), and the other one was made by PECVD at 230 °C
(PESi). The deposition times were 10 s and 1 min, respectively.
After deposition the low-temperature PECVD-sample was
heated to 430 °C and a 1 min HW-decomposed hydrogen
treatment was performed.
The influence of the tin oxide reduction on the transmission
of the FTO-layers was measured using a Perkin-Elmer Lambda
2S spectrophotometer. We quote here the transmission at a
wavelength of 800 nm. The sheet resistance of the FTO-layers
is determined using a four-point measurement.
3. Results
3.1. HW-Decomposed Hydrogen Radicals. Figure 1 shows
the measured XPS-data of the SnO2:F sample before any
treatment and of the HTHW1-sample. Before treatment no
metallic tin is detected; tin is only present in the SnO2-
configuration.16 However, the HW hydrogen radical treated film

Reduction of Tin Oxide by Hydrogen Radicals
J. Phys. Chem. B, Vol. 102, No. 32, 1998 6221
Figure 2. SEM-pictures of (a) tin oxide before treatment, (b) HTHW1,
(c) LTHW1. Magnification for all pictures as indicated in (a).
temperature, there was no sphere formation, indicating that the
reduction at lower temperatures is less severe.
In Figure 3a the optical transmission at a wavelength of 800
nm is plotted, both after the different H-treatments and after
the subsequent reoxidation in air. For the 1 min treatments with
HW-decomposed hydrogen, both at low and high temperature,
the decrease in transmission is fully reversible.
Figure 1. XPS spectra of (a) tin and (b) oxygen. The peaks at a binding
energy of 486.6 and 495.1 eV for tin 3d5/2 and 3d3/2 relate to SnO2.
The metallic tin peaks are 3d5/2 at 484.9 eV and 3d3/2 at 493.4 eV.16
The 1s oxygen binding energy of SnO2 is found at 530.6 eV, while the
binding energy of 532.9 eV relates to the 1s oxygen binding energy of
SiO2.
Upon hydrogen radical treatment the sheet resistance of the
samples increases, as is shown in Figure 3b together with the
effect of subsequent heat treatments at 400 °C in air. The
starting value was 6 Ω/0. After the low and high temperature
HW-decomposed hydrogen treatment, the resistance increased
slightly to 8 Ω/0. No significant change was observed upon
reoxidation.
shows tin peaks at the binding energy corresponding to metallic
tin. The corresponding peak heights are comparable in size to
the peaks due to binding of Sn4+ in SnO2, even though some
surface oxidation could have taken place during transportation
of the samples in air. The lower intensity of both peaks with
respect to the sample before treatment is due to the presence of
silicon at the surface of the sample. This is caused by a small
release of silicon from the walls of the reactor while the sample
is heated and treated. The only oxygen-related peak in the
spectrum of the sample before treatment is due to oxygen in
SnO2. In the oxygen signal of sample HTHW1, no signal due
to SnO2 is found. It shows one distinct peak in agreement with
SiO2. It is thus clear that HW-decomposed hydrogen radicals
cause a reduction reaction at the surface of a tin oxide layer.
The RBS-spectra of the HW-treated samples are depicted in
Figure 4a,b together with the simulations. From the tin signal
in this measurement, we find that a surface layer of 40 nm has
changed in composition from SnO2 to SnO for the LTHW1
sample and to SnO0.6 for the HTHW1 sample. Below this
surface layer another layer of 55 nm is present, which contains
slightly more tin than the bulk SnO2. Thus RBS-measurements
show relatively more tin for the sample treated at high
temperature than for that treated at low temperature. For both
samples the reoxidation is complete after the heat treatment in
air at 400 °C, as is also shown in Figure 4a,b.
The SEM-image for sample HTHW1 shows an onset of
sphere formation, as can be seen from the appearance of a few
‘white’ particles in Figure 2b. The surface of the nontreated
sample is shown in Figure 2a. In addition to the formation of
particles at the surface, also the morphology has changed from
the polycrystalline structure to a more rounded, and larger,
structure upon the treatment with HW-decomposed hydrogen
radicals. It is confirmed that the tin particles that formed on
sample HTHW1 do not disappear upon annealing. The surface
morphology for sample LTHW1 is depicted in Figure 2c and
is comparable to that for sample HTHW1. At the lower
The results of the ERD-measurements are shown in Figure
5. The surface morphology inhibits a quantitative interpretation
of the data. Owing to the changes in surface morphology upon
the hydrogen radical treatments, care should also be taken in
the qualitative comparison of the data. Nevertheless, for the
LTHW1 and HTHW1 samples we observe that the oxygen
content at the surface has dropped, over a depth range in
agreement with the RBS-data. The largest change is observed
for the high temperature sample. From Figure 5 it is seen that
the oxygen concentration in the reoxidized samples is completely
restored.

6222 J. Phys. Chem. B, Vol. 102, No. 32, 1998
Wallinga et al.
Figure 4. RBS spectra for samples (a) LTHW1, (b) HTHW1 after
H-treatment and after reoxidation, (c) PESi and HWSi. In the plots the
surface energy of Sn is indicated. In the inset of (a) the measurement
configuration is depicted.
Figure 3. (a) Transmission and (b) sheet resistance after H-treatment
and after subsequent heat treatment in air at 400 °C for 120 min.
No changes in conductivity and transmission were observed
for a sample that was only heated to 430 °C in a vacuum.
3.2. Plasma-Decomposed Hydrogen Radicals and Ions.
The effect of plasma-decomposed H-radicals and ions on the
surface morphology of textured tin oxide after a 1 min treatment
at 430 °C, sample HTPE1, is shown in Figure 7a. The rounded
surface structure of sample HTPE1 is changed into a polycrys-
talline morphology after reoxidation in air at 400 °C for 120
min as is shown in Figure 7b.
From the SEM-image in Figure 7c, it is clear that the exposure
to plasma-decomposed hydrogen radicals and ions at 430 °C
leads to the formation of spheres on the sample surface after 4
min (HTPE4). This effect increases, and after 10 min tile-like
structures have grown to a size of 2 µm (HTPE10), as is shown
in Figure 7d. This formation of surface spheres is similar to
what is reported for ITO.9,17 Using EDX we found that the
spheres consist of tin; the oxygen concentration in the large
spheres on the sample surface was below the detection limit of
1%.
The transmission of sample HTPE1 is fully restored upon
the heat treatment, as is shown in Figure 3a. However, the
transmission of the sample HTPE10, which is comparable to
HTPE1 after the hydrogen treatment, is still below 40% after
the anneal in air. The onset of an irreversible change in
transmission is seen in the HTPE4 sample, where the transmis-
sion is slightly lower after the heat treatment.
Figure 5. Oxygen depth profiles as measured by ERD before treatment,
H-treated, and annealed samples. The line at 66.7 at. % O denotes
stoichiometric SnO2.
HTPE10 samples the resistance increased to 10-12 Ω/0. After
the heat treatments in air the resistance of these samples rose
to values of 16 and 60 Ω/0, respectively.
The LTPE10 sample did not show any changes in transmis-
sion, sheet resistance, or tin-profile after the plasma-decomposed
hydrogen treatment.
3.3. Deposition of Thin a-Si:H-Layers. As is shown in
Figure 4c, the RBS-spectra from sample HWSi and sample PESi
are comparable, apart from a slightly thicker a-Si:H-layer in
case of the PECVD-sample. The thicknesses of the a-Si:H-
layer derived from these measurements are 17 and 12 nm for
The HTPE1 sample showed no increase in resistance upon
treatment. However, the value increased after the heat treatment
in air, which restored the transmission. For the HTPE4 and

Reduction of Tin Oxide by Hydrogen Radicals
J. Phys. Chem. B, Vol. 102, No. 32, 1998 6223
surface is higher in case of HW as compared to a plasma. It is
therefore expected that radical-related effects on the tin oxide
layer can only be meaningfully compared for plasma-treated
samples with a treatment time that is longer than for HW-
decomposed hydrogen treated samples.
During an actual HWCVD deposition, when pure SiH4 is fed
into the process chamber, the hydrogen-radical concentration
typically is a factor of 10 lower than in the case of pure
hydrogen.19 The effect of hydrogen radicals due to decomposi-
tion of SiH4 will thus be much less than that due to decomposi-
tion of H2 for the same duration. Although a lower radical
concentration with the use of SiH4 instead of H2 is also expected
in the case of PECVD, we have not found evidence for this.
Apart from this, upon deposition the substrate is only
gradually shielded from the influence of hydrogen radicals and
ions. Recent measurements in our group20 showed that HW-
decomposed hydrogen radical treatments affect the hydrogen
concentration in a-Si:H-layers up to a depth of 200 nm.
4.2. Reduction Behavior. Besides the measurements pre-
sented in the previous sections, there is also a clearly visible
difference between the samples. The samples that show
darkening after the hydrogen treatment become transparent upon
a heat treatment in air. The ones that show whitening after
H-treatment retain a low transmission after the heat treatment.
The whitening results from strong light scattering from the
surface, which is in agreement with the morphology observed
by SEM.
Figure 6. Oxygen depth profiles as measured by ERD without
deposition (nontreated) and after deposition of a thin PECVD-layer
with subsequent HW-decomposed H-treatment (HWSi). The inset shows
the measurement configuration.
At a temperature of 230 °C, the difference between the
plasma- and HW-decomposed H-treatments appears to be large.
Sample LTPE10, exposed to 10 min plasma-decomposed H,
does not show any reduction. In contrast to this, the sample
that was treated for 1 min by HW-decomposed hydrogen at
approximately the same temperature shows strong reduction of
the surface tin oxide layer. This is thought to be mainly caused
by small variations in temperature between the PECVD- and
HWCVD-treatments, as a result of differences in substrate
heating. The absence of reduction for sample LTPE10 at 230
°C is different from what is observed by Schade et al. at
temperatures below 250 °C.3 This might indicate the superior
quality of the FTO used in our experiments. At a higher
temperatures, i.e., above the melt temperature of tin, the stability
in plasma-decomposed hydrogen is lost and we observe a strong
temperature dependence of the reduction, similar to Schade et
al.3
Figure 7. SEM-pictures of (a) HTPE1, (b) HTPE1 after reoxidation,
We have shown that at 430 °C, up to a threshold that is
between 1 and 4 min for plasma-decomposed H and is slightly
less than 1 min for HW-decomposed hydrogen, the damage done
to the FTO-substrates by these treatments is almost fully
reversible. Even at temperatures well above the melt temper-
ature of tin, the decrease in optical transmission due to a
hydrogen treatment can be recovered by oxidation in air at 400
°C for 2 h. It is clear from Figure 7b that the polycrystalline
nature of the tin oxide has to a large extent been recovered owing
to the annealing. The time for an HW H-treatment to form Sn
surface spheres is one to four times shorter than for a hydrogen
plasma. This is in agreement with the ratio of the hydrogen
radical fluxes of the two deposition methods.
(c) HTPE4, (d) HTPE10. Magnification for all pictures as indicated in
(a).
the PECVD and HWCVD a-Si:H-layers, respectively. The
amount of tin in the Si-layer is below the detection limit of 5
at. %.
The ERD oxygen profile of the 17 nm PECVD-sample in
Figure 6 does not show any oxygen in the top ∼13 nm of the
a-Si:H-layer. From these measurements we cannot exclude the
formation of a thin silicon oxide layer at the interface between
the a-Si:H and the FTO.
4. Discussion
4.1. Radical Fluxes. The deposition rate of PECVD
a-Si:H-layers is mainly related to the flux of Si-containing
radicals to the surface.18 The same holds for deposition of
a-Si:H by HWCVD. Because the deposition rate for HWCVD
is typically five times higher than for PECVD, we infer that
also the radical flux is larger. When only H2 is used as source
gas, we assume that also the flux of hydrogen radicals to the
After prolonged treatment with plasma-decomposed hydrogen
radicals and ions, particles started to form at the tin oxide
surface. It is confirmed that these are tin droplets. This
nucleation into droplets of elemental tin at the surface of the
reduced tin oxide leads to the continued renewed exposure of
not-fully reduced tin oxide to hydrogen radicals. Consequently
a large loss of oxygen from the sample is possible. Once tin

6224 J. Phys. Chem. B, Vol. 102, No. 32, 1998
Wallinga et al.
particles have formed, the reduction process is not reversible
by reoxidation at 400 °C in air anymore.
Acknowledgment. This work was financially supported by
the Netherlands Agency for Energy and the Environment
(NOVEM). We gratefully acknowledge C. H. M. van der Werf
for the preparation of the samples, E. M. B. Heller for the XPS-
measurements, and E. A. G. Hamers for the extensive discus-
sions. We thank Asahi Glass Co. for the supply of tin oxide
substrates.
XTEM on an FTO-layer with an HWCVD a-Si:H-layer
deposited on top at a substrate temperature of 440 °C previously
showed that an interface layer with a thickness of 100 nm4 has
formed between the tin oxide and the a-Si:H, which indicates
a reduction of the top of the tin oxide.
In contrast to this, after the deposition of a 10 nm PECVD
a-Si:H-layer at 230 °C and subsequent HW-decomposed H-
treatment, no reduction of the tin oxide and no oxygen in the
upper region of the a-Si:H-layer are detected. We attribute this
stability against reduction to the formation of a thin silicon oxide
layer at the a-Si:H-FTO-interface. Because it is known that
10 nm a-Si:H does not form a diffusion barrier for H-radicals,
it is expected that during H-radical treatment at first a thin silicon
oxide layer is formed, which subsequently acts as a barrier layer.
It even is possible that such a layer will form at the FTO-
a-Si:H interface upon deposition of the a-Si:H, similar to what
was found by Wanka et al.6 According to a study by Lan and
Kanicki,9 a SiO2 barrier layer, which prevents a metal oxide
surface from reduction, can also be deposited by PECVD. Solar
cells with a 5 nm SiO barrier on top of the TCO have been
reported by de Nijs et al.21 We infer this could also be a suitable
method for depositing HWCVD a-Si:H on FTO.
References and Notes
(1) Crandall, R.; Luft, W. Prog. PhotoVoltaics 1995, 3, 315.
(2) Major, S.; Kumar, S.; Bhatnagar, M.; Chopra, K. L. Appl. Phys.
Lett. 1986, 49, 394.
(3) Schade, H.; Smith, Z. E.; Thomas, J. H., III; Catalano, A. Thin
Solid Films 1984, 117, 149.
(4) Wallinga, J.; Knoesen, D.; Hamers, E. A. G.; van Sark, W. G. J.
H. M.; van der Weg, W. F.; Schropp, R. E. I. Plasma induced changes to
TCO/a-Si:H interfaces. In Amorphous Silicon Technologys1996; Hack, M.,
Schiff, E. A., Wagner, S., Schropp, R., Matsuda, A., Eds.; Materials
Research Society: Pittsburgh, PA, 1996.
(5) Shiratsuchi, R.; Hirata, M.; Misonou, M.; Kawahara, H., Perfor-
mance of a-Si:H solar cells with H2 plasma pretreated SnO2:F films. In
Technical Digest of the International PVSEC-3; Publication Office
PVSEC: Tokyo, 1987; pp 619-622.
(6) Wanka, H. N.; Schubert, M. B.; Lotter, E., Investigation of TCO/
p-interface and surface morphology of a-Si:H solar cells by in-situ
ellipsometry. In First WCPEC, Hawaii; IEEE: Piscataway, NJ, 1994; pp
516-519.
5. Conclusion
(7) Papadopulos, P.; Scholz, A.; Bauer, S.; Schro¨der, B.; Oechsner,
H. J. Non-Cryst. Solids 1993, 164-166, 87.
At 430 °C HW-created hydrogen radicals lead to the reduction
of textured tin oxide and the formation of tin particles at the
surface within 1 min. This is one to four times faster than with
plasma-decomposed hydrogen at the same temperature. At
lower temperature, 230 °C, the stability of the FTO under
plasma-decomposed hydrogen is lost for HW-decomposed
hydrogen. Nevertheless at the lower temperature less reduction
is found, indicating a strong temperature dependence. The
chemical reduction reaction leads to a less transparent film with
a higher sheet resistance. The comparison between the reduction
by HW- and plasma-decomposed hydrogen shows that the
combined exposure to ions and radicals in the hydrogen plasma
does not induce different processes at the tin oxide surface than
the exposure to HW-decomposed hydrogen radicals alone.
When elemental tin aggregates into tin particles after prolonged
reduction, oxidation in air no longer leads to recovery of the
original optoelectronic properties. Before the onset of particle
formation these properties can be recovered completely, includ-
ing the polycrystalline nature of the film. A barrier layer of
silicon oxide prevents the reduction and was formed by
depositing a thin low-temperature plasma a-Si:H-layer, followed
by a HW-decomposed hydrogen treatment at high temperature.
The deposition of a PECVD SiO2-layer on the FTO substrate
before deposition of HW a-Si:H will be investigated in the
future.
(8) Mahan, A. H.; Carapella, J.; Nelson, B. P.; Crandall, R. S.; Balberg,
I. J. Appl. Phys. 1991, 69, 6728.
(9) Lan, J.-H.; Kanicki, J. Thin Solid Films 1997, 304, 123.
(10) Yablonovitch, E.; Cody, G. D. IEEE Trans. Electron DeVices 1982,
29, 300.
(11) Madan, A.; Rava, P.; Schropp, R. E. I.; von Roedern, B. Appl.
Surf. Sci. 1993, 70-71, 716.
(12) Schropp, R. E. I.; Feenstra, K. F.; Molenbroek, E. C.; Meiling, H.;
Rath, J. K. Philos. Mag. B. 1997, 76, 309.
(13) Doolittle, L. R. Nucl. Instrum. Methods Phys. Res. B 1986, 15,
227.
(14) Arnoldbik, W. M.; Habraken, F. H. P. M. Rep. Prog. Phys. 1993,
56, 859.
(15) Itoh, Y.; Maeda, T.; Nakajima, T.; Kitamura, A.; Ogiwara, N.;
Saidoh, M. Nucl. Instrum. Methods Phys. Res. B 1996, 117, 161.
(16) Chastrain, J., Ed. Handbook of X-ray photoelectron spectroscopy;
Perkin-Elmer: Eden Prairie, MN, 1992.
(17) Kuboi, O. Jpn. J. Appl. Phys. 1981, 20, L783.
(18) Hamers, E. A. G.; van Sark, W. G. J. H. M.; Bezemer, J.; van der
Weg, W. F.; Goedheer, W. J. Ion energy distributions in silane-hydrogen
plasmas. In Amorphous Silicon Technology-1996; Hack, M., Schiff, E.
A., Wagner, S., Schropp, R., Matsuda, A., Eds.; Materials Research
Society: Pittsburgh, PA, 1996.
(19) Tsuji, N.; Akiyama, T.; Komiyama, H. J. Non-Cryst. Solids 1996,
198-200, 1034.
(20) Feenstra, K. F. Private communication.
(21) de Nijs, M. M.; Carvalho, C.; Santos, M.; Martins, R. Appl. Surf.
Sci. 1991, 52, 339.