Self-textured transparent conductive oxide film improves efficiency of solar cells

Aluminum-doped zinc oxide provides a promising alternative to both fluorine-doped tin oxide and indium tin oxide for microcrystalline and amorphous thin-film photovoltaics.
08 February 2010
Doyoung Kim and Hyungjun Kim

Satisfying the universal goal of improving solar-energy conversion efficiency will require modulating the path taken by light on the surface of solar cells. Greater surface roughness is one way to make light scattering more efficient.1 A rough surface both increases the length of a photon's light path and also enhances absorption of light by reducing reflection. Typical surface-texturing methods include wet etching the front electrode using a chemical solution.2 However, repeated wet etching increases processing costs and thickens the transparent conductive oxide film. Chemical-vapor deposition represents an alternative method of creating a rough surface. This so-called self-texturing technique also does away with the need for additional chemical-texturing steps. Figure 1 shows schematically the difference in Light scatting between nontextured and textured surfaces.


Figure 1. Schematic diagram showing light absorption as a function of surface morphology of (a) nontextured and (b) textured surfaces.

According to the commonly known chemical-vapor-deposition growth model,3 formation of grains (similarly oriented crystalline clusters) depends strongly on process conditions and nucleation densities. For example, if nucleation density decreases, grain size tends to increase and vice versa. Nucleation density is determined by surface mobility and the surface reaction rate of chemical precursors such as diethylzinc. These characteristics, in turn, are controlled by chemical-vapor-deposition process parameters such as substrate temperature, pressure, and impurities. Enhancing surface roughness requires formation of large grains, which is achieved by optimizing process conditions. The resistivity of the film is particularly important in application of transparent conductive oxide in photovoltaics. Low resistivity promotes energy-conversion efficiency owing to improved photocurrent and can be achieved by injecting impurities such as aluminum (Al), gallium, or boron, which increases the concentration of electrons.

An early example of self-textured transparent conductive oxide for solar-cell fabrication was reported by the group of Fay using boron as a doping source.4,5 High surface roughness and low resistivity were achieved. However, this self-texturing approach has rarely been studied in conjunction with other dopant materials. Because Al is more cost-effective than boron, in our work we investigated self-texturing of Al-doped zinc oxide (ZnO) using low-pressure chemical-vapor deposition. Figure 2 shows the surface morphology of Al-doped ZnO prepared in this manner on a glass substrate. The surface morphologies were measured by atomic-force microscopy and grown as a function of Al content. All samples showed rough surface morphology that resembles random trigonal prisms. These shapes play an important role in improving light scattering in photovoltaic applications. We then performed a statistical evaluation by calculating the rms average roughness (Rrms) of a surface from an integral of the height profile. The Rrms value of undoped ZnO films—see Figure 2(a)—was 48.1nm. For Al content of 5 and 7% by weight (wt%), the value increased abruptly to 71.0 and 82.0nm, respectively. However, at 8wt% the value of Rrms was reduced because Al was incorporated over the solubility limit.6 Thus, we conclude that the regularity and sensitivity of the surface morphology of ZnO films prepared by chemical-vapor deposition depend on the amount of Al-doping content.


Figure 2. Surface morphology of aluminum (Al)-doped zinc oxide (ZnO)-prepared using chemical-vapor deposition was evaluated by atomic-force microscopy as a function of Al content. (a) 0% by weight (wt%), (b) 5wt%, (c) 7wt%, and (d) 8wt%. R: Surface roughness.

Figure 3 shows the total and diffused transmittance (TT and DT) of undoped ZnO and Al-doped ZnO films as a function of Al content. For undoped ZnO, the TT is above 83% for the wide range between 600 and 1100nm. As Al content increases, however, the TT value reduces to 73% at 600nm. TT values in the wavelength range from 600 to 1100nm are similar, indicating that Al-doped ZnO transmits light uniformly in the visible and IR range for solar-cell applications. In particular, the introduction of trimethylaluminum doping is very effective in raising the DT value to 29% at 600nm (see Figure 3). The haze factor, defined as the DT/TT ratio, quantifies the light-scattering capability in air of a ZnO thin film.5 The calculated maximum haze factor was 43% at a wavelength of 600nm for an Al content of 43mTorr. This result is 2.5 times higher than the haze factor of undoped ZnO and 2.8 times higher than for mass-produced fluorine-doped tin oxide.7


Figure 3. Total- and diffused-transmittance spectroscopy of Al-doped ZnO films as a function of Al content. (a) 0wt%, (b) 5wt%, (c) 7wt%, and (d) 8wt%.

In summary, we have achieved a self-textured Al-doped ZnO surface without chemical etching. The surface roughness of the resulting films proved to be a function of Al content. We obtained a maximum roughness of 82.1nm at an Al content of 7wt%. Based on the surface optical characteristics, we calculated a haze factor of Al-doped ZnO as large as 43% at a wavelength of 600nm. These findings indicate that Al-doped ZnO has a higher light-scattering capacity than fluorine-doped tin oxide, which makes it a plausible alternative in the mass production of transparent conductive oxide for use in solar cells. In the future we will focus on preparing Al-ZnO films with low resistivity and testing their practical application in solar cells.

This work is supported by the Yonsei University Institute of TMS (Telecommunications, Multimedia, State-of-charge) Information Technology, a Brain Korea 21 program.


Doyoung Kim, Hyungjun Kim
Yonsei University
Seoul, South Korea

References:
PREMIUM CONTENT
Sign in to read the full article
Create a free SPIE account to get access to
premium articles and original research