Comparison of Close-Spaced Sublimated and Chemical Bath Deposited CdS Films: Effects on CdTe Solar Cells

Close-spaced-sublimated (CSS) CdS films exhibit strong fundamental edge luminescence, high optical absorption, and a bandgap of ~2.41 eV. Structurally, these films show good crystallinity with thickness-dependent grain sizes that vary between 100-400 nm. In contrast, chemical-bath-deposited (CBD) CdS exhibits subband luminescence, lower absorption, and a thicknessdependent bandgap. These films have CdS grains typically less than 50 nm in size and poorer crystallinity. However, CdTe devices fabricated with these lower “quality” CBD CdS films yield higher Voc’s and fill factors. Carrier lifetimes in finished CSS CdS devices measured between 100 and 200 ps while lifetimes in CBD CdS devices were much higher (>500 ps). Compositional differences in the Cd/(S+Te) ratio at the interface suggest the possibility of lower CdS doping and higher CdTe compensation as one reason for lower Voc’s in CSS CdS devices. INTRODUCTION CdTe superstrate solar cells require the use of a thin n-type CdS window layer between the n-type transparent conductor oxide (TCO) and the p-type CdTe absorber. The current process for fabricating the highestefficiency CdTe solar cells uses close-spaced-sublimation and chemical-bath-deposition for growing the CdTe and CdS layers respectively [1]. The latter CBD process is questionable as a manufacturing process which has promoted recent interest in both CSS and chemical-vapordeposited (CVD) CdS [2,3]. Both processes require the use of higher substrate temperatures than what is encountered in films grown by solution chemistry (typ.~100°C). Until recently, CdTe devices made from CSS-grown CdS yielded lower Voc’s and consequently lower efficiencies than devices made with CBD-grown CdS films [4]. The latter improvements were based upon incorporating oxygen during the CdS growth leading to significant CdO formation. In this paper, we first characterize differences in the morphology and optical properties of as-grown CBD and CSS CdS films grown entirely in He ambients. Next, we then fabricate CdS/CdTe devices using these films and identify differences that exist at CSS CdS/CdTe and CBD CdS/CdTe interfaces. By this approach, the groundwork for improving the performance of CSS CdS/CdTe devices is established. EXPERIMENTAL APPROACH Corning 7059 glass was used as the substrate in all film depositions. TCO’s consisted of SnO2 grown by either tetramethyl-tin or stannous chloride CVD. CBD CdS films were grown by a “standard” process consisting of a solution containing 550 ml H2O, 8 ml of 0.033 M Cd-acetate solution, 4.7 ml of 1.0 M ammonium acetate solution, and 15 ml of 14.76 M NH4OH solution titrated with 8 ml of 0.067 M thiourea solution. CSS CdS films were grown by sublimating pulverized 99.999% CdS powder in 10 torr of helium onto SnO2/glass substrates heated to ~475 °C. In both cases, CdS film thickness was controlled by varying deposition time. CBD film depositions typically required 3540 minutes deposition time while CSS film depositions required 4-15 minutes. With exception of the type of CdS used, film thickness was the primary CdS variable used in this study. To provide a true comparison of the effects of CSS and CBD CdS films, the CdTe film deposition and post deposition device processing were kept fairly constant and have been described in detail previously [5]. CdTe depositions typically were performed by pre-annealing the CdS/SnO2/Glass substrates in H2 at 300°C followed by the CdTe deposition in an ambient of 1 torr O2 and 15 torr helium. CdTe thickness for devices was approximately 10 μm. A large number of CdTe depositions were also performed in which the CdTe film thickness was controlled to ~1000-3000 Å, with substrate temperatures of 525 and 600°C, and oxygen partial pressures of 0, 1, and 2 torr. These “thin-film couples” were used as samples for optical and grazing-incidence X-ray diffraction (GIXRD) characterization of the CdS/CdTe interface. A “lift-off” procedure was also developed for separating CdS/thick CdTe interfaces near the metallurgical junction. Subsequent analysis could then be performed on the exposed alloy surface directly. These latter thick-film “liftoff samples” consisted of 10-μm-thick CdTe films and consequently, should be more representative of actual device interfaces. Optical measurements consisted of both timeresolved photoluminescence (TRPL) to measure carrier lifetimes in CdTe both before and after wet-CdCl2 treatment of the CdS/CdTe layers, and specular (non-integrating) R,T measurements of thin-film couples. I-V device measurements were performed using a xenon-arc lamp simulator calibrated with NREL-confirmed devices. I-V curves were also used to extract the series-resistancecorrected, reverse saturation dark current, Jo. External quantum efficiency (QE) measurements were performed using zero-light bias with total integrated current set to the measured Jsc of each device. Finally, morphology and compositional data were obtained with scanning electron microscopy (SEM), atomicforce microscopy (AFM), and Auger electron spectroscopy (AES). RESULTS AND DISCUSSION The higher substrate temperatures available to CSS-grown CdS films relative to CBD CdS films resulted in marked differences in grain size, crystallinity, and optical properties. As shown in Fig. 1, CdS grown by CSS is much larger-grained than CdS grown by CBD . Although substrate temperature cannot significantly change CBD CdS grain size [6], CSS grain size can be varied by controlling both substrate temperature and film thickness. In this study, grain size was observed to vary from ~100 nm for very thin (575 Å) CSS CdS films up to nearly 300-400 nm for thicker (3060 Å) films. CBD CdS grain sizes were limited to around 30-50 nm for films thinner than 1500 Å. Thinner CBD CdS films result in a slight decrease in grain size. CSS-grown CdS films were shown to have better crystallinity (as shown by higher XRD intensities) than CBD-grown CdS. CSS films consisted primarily of hexagonal phase CdS with CBD CdS films containing primarily the cubic form.