Limitations of ultra-thin transparent conducting oxides for integration into plasmonic-enhanced thin-film solar photovoltaic devices
This study investigates ultra-thin transparent conducting oxides (TCO) of indium tin oxide (ITO), aluminum-doped zinc oxide (AZO) and zinc oxide (ZnO) to determine their viability as candidate materials for use in plasmonic-enhanced thin-film amorphous silicon solar photovoltaic (PV) devices. First a sensitivity analysis of the optical absorption for the intrinsic layer of a nano-disk patterned thin-film amorphous silicon-based solar cell as a function of TCO thickness (10–50 nm) was performed by simulation. These simulation results were then used to guide the design of the experimental work which investigated both optical and electrical properties of ultra-thin (10 nm on average) films simultaneously deposited on both glass and silicon substrates using conventional rf sputtering. The effects of deposition and post-processing parameters on material properties of ITO, AZO and ZnO ultra-thin TCOs were probed and the suitability of TCOs for integration into plasmonic-enhanced thin-film solar PV devices was assessed. The results show that ultra-thin TCOs present a number of challenges for use as thin top contacts on plasmonic-enhanced PV devices: (1) optical and electrical parameters differ greatly from those of thicker (bulk) films deposited under the same conditions, (2) the films are delicate due to their thickness, requiring very long annealing times to prevent cracking, and (3) reactive gases require careful monitoring to maintain stoichiometry. The results presented here found a trade-off between conductivity and transparency of the deposited films. Although the sub 50 nm TCO films investigated exhibited desirable optical properties (transmittance greater than 80 %), their resistivity was too high to be considered as materials for the top contact of conventional PV devices. Future work is necessary to improve thin TCO properties, or alternative materials, and geometries are needed in plasmonic-based amorphous silicon solar cells. The stability of ultra-thin TCO films also needs to be experimentally investigated under normal device operating conditions.
KeywordsTransparent conducting oxide (TCOs) Plasmonics Solar photovoltaics Indium tin oxide Zinc oxide Aluminum-doped zinc oxide
Despite the material, sustainability, economic and technical benefits of thin-film solar photovoltaic (PV) devices [1, 2, 3], conventional crystalline silicon (c-Si) modules dominate the market . The cost of c-Si PV has fallen to the point that the balance of systems (BOS) and thus the efficiency of the modules plays a major role in the levelized cost of electricity for solar . There is thus a clear need to improve the efficiency of thin-film devices further . Recent developments in plasmonics theory promise new methods with great potential to enhance light trapping in thin-film PV devices [7, 8, 9, 10, 11, 12, 13, 14]. To fully exploit these potential benefits offered by plasmonic-based devices, TCOs with high transmittance (low loss) and low enough resistivity are to be used as device top contacts. However, for current transparent conducting oxides (TCOs) to be successfully integrated into the novel proposed plasmonic-enhanced PV devices, ultra-thin TCOs films are required . For example, simulations by Vora et al. showed a 19.65 % increase in short circuit current (JSC) for nano-cylinder patterned solar cell (NCPSC) in which the ITO layer thickness was kept at 10 nm to minimize the parasitic Ohmic losses and simultaneously act as a buffer layer while helping to tune the resonance for maximum absorption . TCOs such as the most established indium tin oxide (ITO), aluminum-doped zinc oxide (AZO) and zinc oxide (ZnO) are standard integral materials in current thin-film solar PV devices [15, 16, 17, 18]. Bulk material properties for common TCOs including ITO have been well researched and documented for different processing conditions and substrates [15, 16, 19, 20, 21, 22, 23]; however, this is not the case for ultra-thin TCOs. The few exceptions include Sychkova et al. , who reported both optical and electrical properties of 9–80 nm ITO films deposited by pulsed DC sputtering varied with thickness and showed a general increase in resistivity with decrease in film thickness . Other notable studies on ultra-thin ITO films using various deposition techniques include the following: Chen et al. who used filtered cathodic vacuum arc (FCVA) to deposit 30–50 nm on heated quartz and Si substrates ; Tseng and Lo, who used DC magnetron sputter for 34.71–71.64 nm ITO film on PET (polyethylene terephthalate) ; Kim et al. who used RF magnetron sputter for films between 40 and 280 nm deposited on PMMA substrate heated at 70 °C ; Alam and Cameron, who used sol–gel process for 50–250 nm film deposited on titanium dioxide film ; and Betz et al. who used planar DC magnetron sputtering for 50, 100 and 300 nm films on glass substrates . The results from these few thin TCO studies reveal a pattern in which resistivity increases rapidly as film thickness decreases from 50 to 10 nm.
The electrical properties of ITO thin films depend on the preparation method, the deposition parameters used for a given deposition technique and the subsequent heat treatments. Key factors for the low resistivity have not been clearly documented because of the complex structure of the unit cell of crystalline In2O3 formed by 80 atoms and the complex nature of the conducting mechanisms in polycrystalline films . The issue is further complicated by the large number of processing parameters, even for a single technique.
The optical effects of TCO thickness
Sensitivity analysis for the proposed silver nano-disk patterned solar cell (NDPSC) was performed in the 300–750 nm spectral range to determine the optimum ITO layer thickness which would promote maximum enhancement and minimize Ohmic losses. Having a TCO spacer layer with as low as possible Ohmic losses is desirable for efficient coupling of light from the silver nano-discs into the active layers of the device. The results are shown in Fig. 1b and theoretically show 10 nm films offer the best absorption and hence the greatest potential to improve efficiency in plasmonic-based PV devices. From these results, AZO and ITO offer the best potential due to lower Ohmic losses and ZnO, despite having the greatest Ohmic losses among the three TCOs, is still promising particularly for the sub 20 nm films since its absorption (>250 W/m2) is still higher than that expected of a standard PV device.
The focus of the study was to investigate ways of improving material properties of ultra-thin TCOs for integration into plasmonic-enhanced thin-film solar PV devices by studying the effects of different process parameters on both optical and electrical properties of sub 50 nm films. A comparative study of the three most commonly used TCOs in thin-film commercial solar cells is undertaken, and a more in-depth study of ITO is performed.
Sample preparation and fabrication
Samples of ITO with thickness ranging from 10 to 50 nm were deposited on both glass and n-doped silicon (with a 32-nm thermally grown oxide layer) substrates using rf sputter deposition techniques previously described in refs [30, 31, 32]. A 99.99 % 4-inch pressed ITO (Sn2O:In2O3 10:90 % wt) target was used, and an average base pressure of 7 × 10−8 torr was achieved before deposition. Both the glass and silicon substrates were ultrasonically cleaned in isopropanol for 5 min. All other process parameters such as target bias [900 V (ITO and ZnO) and −500 V (AZO)] and substrate distance (75 mm) were kept constant through the experiment. Substrates and target were sputter pre-cleaned in an argon environment for 5 min before each run. The protocol for pre-cleaning is described in Ref. . To investigate substrate dependency, ITO was deposited on a pair of substrates for 1 min with 0 % oxygen ratio and 100 W rf power. ZnO samples were processed at rf power of 100 W on glass and silicon substrates in an argon environment and 0 % oxygen in the same system as ITO using a stoichiometric 99.99 % 4-inch pressed ZnO target. The process pressure was maintained at 7.1 × 10−3 torr and the deposition rate was calculated to be 8 nm/min.
Summary of process parameters for the TCOs
RF power (W)
Target bias (V)
Process gases flow rates (sccm)
Film thickness (nm)
To investigate the effects of post-processing treatment on both optical and electrical effects, additional samples of ITO films on sodalime glass (SLG) substrates were processed using a different instrument  to obtain a pair of film samples with varying thicknesses from 10 to 50 nm in steps of 10 nm. The system is a four-gun sputtering system with a target to substrate spacing of approximately 4″. An ITO (90 % In2O3/10 % SnO2 from Lesker) target was used. The material was sputtered using 100 W rf under 4 mTorr of Ar. Deposition time was varied for film thickness with 36 s resulting in 10 nm (~3 A/sec). This deposition rate was determined by depositing for a set amount of time and measuring the resulting film thickness using stylus profilometry (Veeco Dektak 150).
One sample for each as-deposited pair was divided into three samples using a diamond scriber. The three pieces were then annealed separately at 400 °C for 10, 20 and 30 min, respectively, using UHP forming gas (FG) (95 % N2/5 % H2 from Air Gas) in a sealed (by vacuum coupling components) quartz tube inside a tube furnace. The furnace was equilibrated at the heating temperature prior to sample introduction. The samples were placed in the quartz tube; then the tube was purged with FG at 5 scfm for 5 min—this was approximately four exchanges of tube volume. After purging, the samples were introduced into the hot zone with a vacuum-sealed push rod, and the flow rate was reduced to approximately 150 sccm for the duration of heating. After heating, the samples were removed from the hot zone and cooled by increasing the gas flow. After characterization, the sample previously annealed at 400 °C for 30 min was further annealed at 500 °C for 10 min.
Optical and electrical characterization process
The film thickness measurements and optical characterization were carried out using spectroscopic ellipsometry (J.A Woollam Co UV–VIS V-VASE with control module VB-400). In each case, a standard scan was performed ranging from 300 to 1000 nm in increments of 10 nm for the 65°, 70° and 75° incident angles. Random detailed scans were performed for the quality check purposes although they are normally not necessary for isotropic samples. Ellipsometry analysis was performed following the process by Synchkova . Intensity measurements were carried out using the VASE for normal transmission incidence (0° reflection angle) for the three TCOs on glass substrates for the same wavelength range as above. A baseline scan was obtained for the clean SLG substrate first followed by the main data scans using baseline data. Both the baseline and the data scans were acquired in close successions to minimize errors due to light source intensity fluctuations.
Electrical characterization was performed using a four-point probe system consisting of ITO optimized tips consisting of 500 micron tip radii set to 60 g pressure and an RM3000 test unit from Jandel Engineering Limited, UK. The sheet resistance of the 10 and 20 nm TCOs on glass and on silicon substrates with a spacer oxide layer was determined by direct measurement for both forward and reverse currents. For each TCO on glass sample, a mean sheet resistance value from three random points was used in the final results whilst a mean of only two points was used for the TCO on Si samples since they were smaller.
All samples were imaged for film quality and a compositional analysis was done using a Hitachi S4700 field emission scanning electron microscope (FE-SEM). Atomic force microscopy (AFM) was performed using a Veeco Dimension 3000 equipment with cantilever tips (Tap300Al-G) on a 1:1 acquisition aspect ratio. The field of view was 2 μm at 512 pixel width and scans performed at a speed of 0.5 Hz. Three randomly selected fields of view were acquired per sample and the analyzed areas were limited near to the center of the sample. Roughness analysis was then performed on a defect-free region.
Theory and calculations
The theoretical derivations of both the resistivity and attenuation coefficient of the ITO films are highlighted in Sects. 3.1 and 3.2 below to explain the underlying processes contributing to the results reported in this paper.
Results and discussion
The transmittance and resistivity measurement results for the TCOs are discussed below.
Transmittance results support the sensitivity analysis results. For the 20-nm films, AZO has greater than 90 % transmittance for the 300–1000 wavelength range, whilst ITO and ZnO show an average transmittance greater than 80 and 70 %, respectively, in the same spectral range.
Resistivity of 20 nm as-deposited ITO, AZO and ZnO films on SLG substrates
Sheet resistance, Rs (Ω/□) × 103
Resisitivity, ρ (Ω cm)
1.3 × 10−3
1.7 × 10−3
7.8 × 10−4
Sheet resistance of various as-deposited TCO samples
Sheet resistance, Rs (Ω/□) × 103
Resisitivity, ρ (Ω cm)
8.3 × 10−4
1.3 × 10−3
4.2 × 10−4
1.7 × 10−4
Transmittance measurements for ITO
As-deposited thinner ITO samples (10 and 20 nm) have the transmittance greater than 95 %. It is interesting to note that the 40-nm film sample does not seem to follow this general trend, particularly the sample annealed for 20 min. This sample film shows the greatest increase in mean roughness (vida infra) when all other films’ roughness is decreasing and it also has the best transmission for all the 40-nm film samples. The general trend is that the overall transmittance curve for the as-deposited ITO shifts down with increasing film thickness (i.e., the as-deposited film becomes less transparent with increasing thickness as expected). Around visible spectrum and at higher wavelengths, the transmittance for the as-deposited ITO approaches that of the annealed ITO (i.e., annealing is not much effective here in improving transparency). However, around UV wavelengths the transmittance for the as-deposited ITO shifts down below that of the annealed samples (i.e., at small wavelengths annealing is more effective as the annealed samples are more transparent). This is a well-known phenomenon (Burstein–Mess shift) which is a result of ITO optical band gap shifting towards higher energies when annealed in FG or H2 gas. This is attributed to increase in carrier concentration and is well documented . In addition, it appears that among the annealed samples, 20 min gives the optimum transmittance for thicknesses below 50 nm, especially at large wavelengths. There observed trend means that the use of thinner (10 nm), more transmitting and low loss (Ohmic losses) films will result in more light being coupled into the underlying i-a-Si:H layer rather than being absorbed in the TCO layer as is the case with thicker film (>20 nm).
Film morphology and roughness
Effect of substrate on ultra-thin ITO films
Summary of AFM results for as-deposited ITO films on glass and Si substrates
RMS roughness (Rq)
Mean roughness (Ra)
ITO on Si wafer, 10 nm
Uniform, small features
ITO on Si wafer, 20 nm
Uniform features larger than 10 nm
ITO on glass, 10 nm
Very fine features with some larger defects
ITO on glass, 20 nm
Fine features with many large defects
Effect of annealing time on ultra-thin ITO films
When ultra-thin ITO films were subjected to post-processing treatment at 400 °C in a FG environment, different treatment times produced different effects.
The as-deposited films mean roughness for this second batch of ITO samples were observed to vary between 0.67 and 0.85 nm. The 10-nm film had the smallest mean roughness value whilst the 20-nm film had the largest value. This may be due to the presence of surface defect features which seem to be more pronounced on the 20-nm film compared to all the other samples. Generally, all samples show a varying degree of dust particles’ presence and potential artifacts. Sections of the film samples which exhibited heavy dust particles’ (and any other contaminants) presence, striations and potential artifacts that were not consistent with other areas on the sample were excluded from the analysis.
The images show a sharp increase in the mean roughness for generally all films after 10 min of heat treatment. Whilst the film roughness is small for both the 10- and 20-nm films, it is observed to increase by a factor of two for the 30- to 50-nm film thickness samples. There is a trend for all films showing a decrease in mean roughness after 20 min of post-processing treatment with the 30-nm film showing the greatest decrease from approximately 1.9 to 1 nm. Evidently, annealing for 30 min results in a slight improvement in film roughness for the 30- to 50-nm range of film. However, the thinner films (10 and 20 nm) show great deterioration in film mean roughness when annealed for longer periods of time (30 min or greater). This can be explained by the onset of islands on both of these films. Island formation is more pronounced on the quasi 2D 10-nm film resulting in the mean roughness increasing from the initial value of approximately 0.7–1.9 nm.
The detailed study on ITO showed some dependency of electrical properties and surface roughness with substrate type which is consistent with results from previous studies on slightly thicker films. Also ITO films on glass show a high degree of surface defects and finer amorphous-like features which may explain the high and oscillating values of sheet resistance on these films. Films grown on Si substrate have uniform, but large features. However, the same films have higher resistivity values. All samples, as-deposited and annealed, have a transmittance value greater than 80 % with the as-deposited films being superior except for the 40-nm films for which the annealing for 20 min gives the best transmittance. Further analysis of samples shows films annealed for 20 min generally have the lowest resistivity and lower roughness values.
Future work is needed to improve other TCOs such as AZO and ZnO and to engineer new high-conductivity low-loss materials for integration into plasmonic devices. AZO exhibited a transmittance superior to that of ITO while ZnO had the best sheet resistance among the three TCOs being compared. Further investigative work is needed to find the balance between films with useful resistivity and acceptable Ohmic losses in plasmonic-based PV devices. Future work should focus on different processing techniques such as DC sputtering as well as exploring other post-processing environments.
Ultra-thin TCOs and in particular ITO present a number of challenges for use as thin top contacts on plasmonic-enhanced PV devices. First, both ultra-thin TCO optical and electrical parameters differ greatly from those of thicker (bulk) films deposited under the same conditions. Second, they are delicate due to their thickness, requiring very long annealing times to prevent film cracking. The reactive gases (usually oxygen or hydrogen) require careful monitoring to avoid over-oxidizing or over-reducing the film as it impacts their stoichiometry. There is a trade-off between conductivity and transparency of the deposited films. The sub 50 nm thick TCO films investigated exhibited desirable optical properties (transmittance greater than 80 %), which makes them viable for plasmonic PV devices applications. However, all films evaluated here had resistivity values too high to be considered as materials for the top contact of conventional PV devices.
Authors would like to acknowledge the support from faculty start-up funds from the University of Toledo, Fulbright (Science and Technology) funds, and the National Science Foundation (CBET-1235750). Furthermore, authors would like to acknowledge the helpful discussion with Dr. J. Mayandi.
- 7.Gwamuri, J., Güney, D.Ö., Pearce, J.M.: Advances in plasmonic light trapping in thin-film solar photovoltaic devices. In: Tiwari, A., Boukherroub, R., Sharon, M. (eds.) Solar cell nanotechnology, pp. 243–270. Wiley, Hoboken (2013)Google Scholar
- 10.Green, M.A.: Third generation photovoltaics: advanced solar energy conversion (Springer Series in Photonics). Springer (2005)Google Scholar
- 12.Maier, S.A.: Plasmonics: fundamentals and applications. Springer, New York (2007)Google Scholar
- 32.Perkin-Elmer, R.F.: Sputtering System-6 Inch protocol: MOST. http://www.appropedia.org/Perkin-Elmer_RF_Sputtering_System-6_Inch_protocol:_MOST. Accessed 10 Jul 2015
- 33.Knickerbocker, S.A.: Ph.D. Dissertation, Michigan Technological University (1995)Google Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.