TSF-28672; No of Pages 3

Thin Solid Films xxx (2011) xxxxxx

Contents lists available at ScienceDirect

Thin Solid Films

Advantages of transparent conducting oxide thin lms with controlled permittivity for thin lm photovoltaic solar cells
T.A. Gessert , J. Burst, X. Li, M. Scott, T.J. Coutts
National Renewable Energy Laboratory (NREL), Golden, CO 80401, USA

a r t i c l e
Available online xxxx

i n f o

a b s t r a c t
Our recent investigations have identied a pathway to produce transparent conducting oxide (TCO) lms that demonstrate higher infrared transparency. The technique involves controlling the dielectric permittivity of the TCO lm such that the electrical properties are maintained, but the plasma frequency (p) is shifted to longer wavelength. This has the effect of reducing free-carrier absorption in the visible and near-infrared spectral region, thus producing a TCO lm with higher optical transmission. The technique has been demonstrated for sputtered lms of indium tin oxide by adding small amounts of ZrO2 to a ceramic sputtering target, and for SnO2:F lms deposited by chemical vapor deposition using a metalorganic Zr source. 2011 Published by Elsevier B.V.

Keywords: Transparent conducting oxides CdTe thin lm photovoltaic solar cells

1. Introduction Recent reports have described that transparent conducting oxide (TCO) lms optimized for photovoltaic (PV) applications should be designed to achieve low resistivity by maximizing mobility rather than electron concentration [1,2]. Specically, for two TCO lms with similar resistivity, the lm with higher mobility requires a lower freeelectron concentration to achieve the same resistivity. The lm with lower free-electron concentration presents fewer electrons to interact with the incident electromagnetic eld (i.e., light). This produces less optical reection and absorption. The process is described by the Drude free-electron theory and is shown schematically in Fig. 1. In Fig. 1, the interaction of essentially free electrons in a TCO with an electromagnetic eld is physically similar to a driven, damped harmonic oscillator; this can be envisioned as a light mass (electron) connected to a much heavier mass (ion core) by a simple spring that is operating in a dampening medium. In this case, the driving frequency is the frequency of electromagnetic radiation (), the light mass is the effective mass of the electrons (m*), and the damping constant of the medium is related to the mobility () and the real part of the dielectric permittivity of the particular TCO material (1). Although most Drude-theory modeling accounts for most of these parameters, the dielectric permittivity is typically assumed to be similar for all TCOs (~4) and is not regarded as a variable parameter. In this study, we demonstrate that dielectric permittivity is likely an important parameter in understanding differences in TCO materials. Further, we show that control of dielectric permittivity in TCO is possible and may be an underappreciated design avenue in the development of new TCO materials.

The Drude theory formalisms that have been found to describe TCO optical and electrical properties can be shown to yield the following relationship for the frequency () dependence for the real part (1) of the complex permittivity [2]: 1 = ne2 o m 2

where is the high-frequency permittivity for a particular material, o is the permittivity of free space, m* is the effective mass of the carriers, n is the carrier density, and e is the electronic charge; 1() describes the ability of the constituents of the material (i.e., ion cores and electrons) to become polarized in the presence of electromagnetic oscillations. The plasma frequency (p) is obtained at the frequency where 1() in the above equation equals zero, s 2 ne p = : o m The plasma wavelength (p) is p = s 2c m = 2c o 2 ; p ne

Corresponding author. E-mail address: [email protected] (T.A. Gessert). 0040-6090/$ see front matter 2011 Published by Elsevier B.V. doi:10.1016/j.tsf.2011.01.143

where c is the speed of light. Numerical modeling of these predictions indicates that variation of could have a signicant effect on near-infrared (NIR) absorption, transmission, and reection [3]. Increasing of any material implies compositional and/or structural modications that increase the material's polarizibility within the frequency range of interest. Although polarizibility is determined by the combined dipole response from both ion cores

Please cite this article as: T.A. Gessert, et al., Thin Solid Films (2011), doi:10.1016/j.tsf.2011.01.143

T.A. Gessert et al. / Thin Solid Films xxx (2011) xxxxxx

Fig. 1. Simplied schematic of free-carrier absorption in a TCO.

measured using ellipsometry (Wollam Auto SE). Optical measurements were performed using a spectrophotometer (Varian Cary 5G) in a wavelength range of 3002500 nm. The lm resistivity, mobility, and carrier concentration were obtained at room temperature from Hall analysis using the van der Pauw method (BioRad HL5500). Spectroscopic ellipsometry (SE) measurements were performed to analyze dielectric constants using a J.A. Woollam M-2000NIR rotating compensator ellipsometer at three angles of incidence (55, 60, and 65) in a spectral range of 8001700 nm. The TCO lms were represented by a Drude oscillator to model free-carrier absorption at long wavelengths. Additional SE parameters are provided elsewhere [8]. For the ZrO2 lm (not a TCO), a Lorentz oscillator was used, which gave a straight-line t in the NIR region. 3. Results In these initial studies, Zr was added to both sputtered ITO lms and FTO lms produced by LPCVD. However, we expect that other traditional TCO materials (e.g., ZnO:Al, In2O3:ZnO, Cd2SnO4, etc.) will produce similar results if an -enhancing addition can be appropriately incorporated into the TCO host. Both sets of lms (ITO/ITZO and FTO/FTZO) were produced during the same time period so that lm variation could be minimized, allowing effects due to the high- component (i.e., Zr) to be isolated. It should also be noted that both the ITO and FTO lms produced for this comparison are of high optical and electrical quality, and so any observed material improvement means improvement over already high-quality parameters. Figs. 2 and 3 compare two ITO and ITZO lms with different ranges of carrier concentration. Note that each gure compares lms that demonstrate similar thicknesses and electrical properties (carrier concentrations and mobility, measured using room-temperature Hall measurements). This degree of lm similarity will minimize effects caused by the free-carrier absorption discussed earlier. The gures show that the ITZO lms demonstrate signicantly higher NIR transmission compared to the ITO lms. Similarly, Figs. 4 and 5 compare two sets of FTO and FTZO lms (each set has similar carrier concentrations). In both gures, we see that the lms that contain Zr demonstrate superior optical quality [911]. Closer inspection of the FTZO lms also reveals that the lms containing Zr have a slightly wider optical bandgap than the FTO lms. Because the carrier concentrations are similar for the lms being compared, the wider optical bandgap probably results from the FTZO alloy having a higher bandgap than FTO due to alloying, rather than being caused by a BursteinMoss shift. Fig. 6 shows SE analyses for the sputtered ITO-based lms, whereas Fig. 7 shows similar analysis for the LPCVD FTO-based lms. The gures show that as Zr is added to the ITO or FTO host materials, 1 increases with increasing Zr content within the wavelength range probed. Recalling that

and electrons, in the frequency range of interest for TCO applications (NIR and optical, 10131015 s1), the dipole response of the electrons is expected be more important [4]. Nevertheless, changes in TCO permittivity are not well understood at this time, and so contributions from ion cores may also be signicant. Insight into modifying the dielectric constant of TCOs can be gained through studies of SiO2-based lms used in metal-oxide-semiconductor (MOS) devices. In this case, new process designs to reduce MOS feature size would benet signicantly from increased permittivity of the oxide layer at an operating frequency of 1081010 s1. These studies have demonstrated that relatively small additions of high-permittivity oxides, such as ZrO2 and HfO2, can yield permittivity values signicantly higher than expected from a linear extrapolation between SiO2 and ZrO2 [5]. Our efforts to increase TCO have pursued a similar approach involving fabrication of thin lms using traditional TCO materials (e.g., ITO and FTO) and incorporating a small amount of a high- constituent (e.g., Zr). The high- addition was also chosen to maintain either electrical quality (i.e., Zr has been shown to demonstrate similar doping in In2O3 as does Sn [6,7]) or produce an alloy, as in the case of FTZO. In both cases, the goal of Zr addition is to gain an advantage in the optical properties from the high- material without sacricing the electrical quality of the TCO lm. 2. Experimental details Indium tin oxide (ITO) and ITO:Zr (ITZO) lms onto glass were prepared using radio-frequency (r.f.) magnetron sputtering at 13.56 MHz in a sputter-down conguration from 2-in. planar targets of 99.99% purity. The sourcesubstrate separation was ~ 1.27 cm. Doped targets were fabricated by hot pressing followed by ring in oxygen at 760 C. The target compositions compared in this study are shown in Table 1. Other experimental details of the sputtering process are provided elsewhere [8]. SnO2:F (FTO) and SnO2:F:Zr (FTZO) lms were deposited onto glass by low-pressure chemical vapor deposition (LPVCD) through reaction of tetrametyl tin (TMT), zirconium tertbutoxide (ZTB), and bromotriuormethane (uorine source) in oxygen at a temperature range of 500550 C. The thickness of the ~2001200-nm lms was measured using stylus prolometry (Veeco Dektak 3), and the 100200-nm lms were
Table 1 Nominal composition for ITO and ITZO sputtering targets used in this study. Dopant concentration was calculated assuming specied target compositions. Target Specied target composition (Weight %) In2O3 ITO ITZO 91 90 SnO2 9 9 ZrO2 0 1 Dopant concentration (Atomic %) Sn 3.29 3.27 Zr 0 0.49

80

T, R, A (%)

T
60 40

A
20 0 500 1000 1500

2000

2500

Wavelength (nm)
Fig. 2. Comparison of ITO and ITZO lms with similar thickness, mobility, and n = ~7 1020 cm3. Solid line: ITZO-005 (0.448 m, 7.92 1020 cm3, 35.0 cm2V1s1). Dashed Line ITO-004 (0.451 m, 7.29 1020 cm3, 36.9 cm2 V1s1).

Please cite this article as: T.A. Gessert, et al., Thin Solid Films (2011), doi:10.1016/j.tsf.2011.01.143

T.A. Gessert et al. / Thin Solid Films xxx (2011) xxxxxx

100 80

Real Part Dielectric Constant (1)

4 3 2 1 0 -1 -2 800 1000 1200 1400 1600

T, R, A (%)

T
60 40

A
20 0 500 1000 1500

IZrO 9 wt.% Zr ITZO 1 wt.% Zr ITO 0 wt% Zr

2000

2500

Wavelength (nm)
Fig. 3. Comparison of ITO and ITZO lms with similar thickness, mobility, and n=~1.21021 cm3. Solid line ITZO-003 (0.437 m, 1.391020 cm3, 24.4 cm2V1s1). Dashed line ITO-005 (0.449 m, 1.231020 cm3, 27.3 cm2V1s1).

Wavelength (nm)
Fig. 6. Real part of permittivity for ITO, ITZO, and sputtered In2O3 lms containing 9 wt.% ZrO2 (but no Sn).

80

Transmission (%)

60 40 20 0 400 600 800 1000 1200 1400

SnO2:F (513 nm, 1.75x10 cm , 31 cm /V-s) SnO2:F:Zr (494 nm, 1.86x10 cm , 26 cm /V-s)
20 -3 2

20

-3

Wavelength (nm)
Fig. 4. Comparison of FTO and FTZO (~ 1% Zr) lms with similar carrier concentration and mobility. Solid line SnO2:F:Zr, dashed line SnO2:F. Fig. 7. Real part of permittivity for ~ 1%10% Zr in FTO. Undoped FTO (bottom) and ZrO2 (top) are included for comparison.

p is achieved when 1 equals zero, Fig. 6 illustrates that a ~1 wt.% addition of ZrO2 increases p of ITO from ~1300 to ~1550 nm. It is this shift in p that causes the increase in transmission shown in Figs. 2 and 3. Although Fig. 7 shows a similar trend of increasing 1 with increasing Zr concentration, identifying a change in p is not as apparent as in Fig. 6, because the value of 1 increases rapidly with increasing Zr. Reasons for this complication may be due to subtleties in the SE modeling, and are presently being investigated. 4. Conclusions We have shown that the dielectric permittivity of a TCO material can be controlled to yield benets for PV solar cells. The increase in

as Zr is added to sputtered ITO or LPCVD FTO can shift p to a longer wavelength and make it possible to improve NIR transmission without reducing the electrical quality of the lm. These results suggest that TCO lms with permittivity control may have advantages that could be exploited at manufacturing volumes consistent with commercial thin-lm PV production. Although this insight can assist in designing future TCO materials, we believe it may hold even greater signicance for present TCO materials that may have industrial advantages but in which the NIR optical transmission remains limited by low electron mobility. Acknowledgement This research was supported under DOE Contract No. DE-AC36-08GO28308 to NREL.

80

Transmission (%)

60

References
[1] T. Coutts, D.L. Young, X. Li, MRS Bull. 25 (8) (2000) 58. [2] T.J. Coutts, D.L. Young, T.A. Gessert, in: D.S. Ginley, H. Hosono, D.C. Paine (Eds.), Handbook of Transparent Conductors, Springer, 2010. [3] T.A. Gessert, Y. Yoshida, C.C. Fesenmaier, T.J. Coutts, Materials Research Society Symposium Proceedings, Vol. 1165, MRS, Warrendale, PA, 2009, p. 247. [4] G.D. Wilk, R.M. Wallace, J.M. Anthony, J. Appl. Phys. 89 (10) (2001) 5243. [5] G. Lucovsky, G.B. Rayner, J. Appl. Phys. Lett. 77 (18) (2000) 2912. [6] R. Groth, Phys. Status Solidi 14 (1966) 69. [7] T. Koida, M. Mondo, Appl. Phys. Lett. 89 (2006) 082104.1. [8] T.A. Gessert, Y. Yoshida, C.C. Fesenmaier, T.J. Coutts, J. Appl. Phys. 105 (2009) 083547.1. [9] T.A. Gessert, Y. Yoshida, and T.J. Coutts, U.S. Patent Application No. 11/718, 628, May 4, 2007. [10] T.A. Gessert, J. Duenow, T. Barnes, T.J. Coutts, U.S. Patent Application PCT 0742, September 2007. [11] X. Li, J. Burst, T.A. Gessert, U.S. Patent Application No. 61/314, 802, May 3, 2010.

40

SnO2:F (510 nm, 4.3x10 , 28 cm /V-s) SnO2:F:Zr (496 nm, 4.5x10 , 28 cm /V-s)
20 2

20

20

0 400 600 800 1000 1200 1400

Wavelength (nm)
Fig. 5. Comparison of FTO and FTZO (~2% Zr) lms with similar carrier concentration and mobility. Solid line SnO2:F:Zr, dashed line SnO2:F.

Please cite this article as: T.A. Gessert, et al., Thin Solid Films (2011), doi:10.1016/j.tsf.2011.01.143