Optical analysis of polycrystalline semiconductors and transparent conducting oxides (TCOs) used in thin film photovoltaics is a considerable challenge due to surface roughness and nonuniformities that may exist over many orders of magnitude in the in-plane scale L, from tenths of nanometers to hundreds of microns. Here we describe the new optical technique of multichannel Mueller matrix ellipsometry, that enables one to extract the dielectric function (ε1, ε2) of such a film, as well as the time evolution of surface roughness layer thicknesses on (i) microscopic (L <<λ, whereλ is the probe wavelength), (ii) macroscopic (L λ), and (iii) “geometric optical” (L <<λ) scales, in an analysis of data collected during deposition or processing. We present as an example, analysis results from spectra collected during chemical etching of the TCO zinc oxide that leads to detectable surface roughness on all three scales. The optical results are in good agreement with direct measurements by atomic force microscopy (AFM) and profilometry.

The microstructure of high-efficiency CdTe solar cells on commercial SnO2/soda-lime glass was investigated by scanning transmission electron microscopy. The CdTe solar cells have a structure of soda-lime glass/SnO2/ZTO/CdS:O/CdTe. We found no interdiffusion between the SnO2 layer and ZTO layer. Weak diffusion of Zn from the ZTO layer into the CdS:O layer was observed; however, the diffusion was not uniform. Interdiffusion also occurred at the CdTe/CdS:O interface. In the back-side of the CdTe, a thin layer of Te was found, which formed during the nitric-phosphoric etching. In addition, a very thin layer of CdHgTe was observed at the CdTe/Te interface.

Optical losses in thin film solar cells arise due to reflections at the top interfaces where dielectric discontinuities may be significant, e.g., between the glass and transparent conducting oxide (TCO) contact and between the TCO and semiconductor structure. Advanced optical engineering approaches are needed to minimize such losses. One approach is to incorporate multilayered or graded-index TCO films designed to act as broad-band anti-reflectors. Thus, it is important to be able to modulate the near-infrared/visible index of refraction of the TCO over a relatively wide range (e.g., 1.3 < n < 2.0) without increasing its extinction coefficient k or significantly degrading its electrical conductance. Here we report an investigation of SnO2 and ZnO sculptured thin films (STFs) under development for this purpose Sculptured thin films are deposited under low surface mobility conditions using stepwise or continuous variations in the polar and/or azimuthal angles of the deposition flux impinging on the surface. Deposition at a glancing polar angle leads to columnar growth, optical anisotropy, and low ordinary indices of refraction, whereas normal incidence deposition under the same conditions leads to dense isotropic films and high indices. In this study, we explore the dependence of the optical properties of SnO2 and ZnO, including index and birefringence spectra, on the polar deposition angle. Optical modeling reported here assesses the ability of the STF concept to provide tailored TCOs for advanced optical engineering of CdTe solar cell structures.

A type-I (“spike”) conduction-band offset (CBO) greater than a few tenths of an eV at the n/p interface of a solar cell can lead to significant distortion of the current-voltage (J-V) curve. Such distortion has been observed in CdS/CIS cells, low-gallium CdS/CIGS cells, and CIGS cells with alternative windows that increase the CBO. The basic feature is reduced current collection in forward bias. The distortion is mitigated by photoconductivity in the CdS or other window layer, and it is therefore more severe if the illumination contains no photons with energies greater than the band gap of the window layer. The device-physics analysis of such distortion is numerical simulation incorporating a three-layer [TCO/CdS/CI(G)S] approximation for the solar cell. The parameters that influence the barrier height, and hence the distortion, are the magnitude of the CBO, the doping of the p- and n- layers, the defect density of the CdS, and the thicknesses of the CdS and TCO layers. The key value, however, is the energy difference between the quasi-Fermi level for electrons and the conduction band at the CdS/CIS interface. Thermionic emission across the interface will limit the current collection, if the difference exceeds approximately 0.48 eV at 300 K and one-sun illumination. This constraint is consistent with experiment, and strategies to satisfy the 0.48-eV rule when designing solar cells are enumerated.

We study the electric current through metal-semiconductor junctions of a type used in thin-film PV for back contacts. To concentrate on one type of junction we used the symmetric structures of rf-sputtered CdTe layer sandwiched between two Cr contacts. Along with the conventional measurements, the current-sensing contact mode AFM was employed to measure the current-voltage characteristics and current variations with time under fixed voltage. We found that (i) the electric current flow is laterally strongly nonuniform; (ii) it chaotically varies over time; (iii) this behavior did not correlate with surface topography. We interpret our observations in terms of defect assisted tunneling through time-dependent defect pathways.

Clean and ordered chalcopyrite CuInSe2 surfaces are a precondition for the study of the electronic structure by angle-resolved photoelectron spectroscopy. The preparation of welldefined CuInSe2(001) surfaces by the combination of molecular beam epitaxy and a selenium capping and decapping process is described. The surface structure of CuInSe2 epilayers with different bulk composition is compared and analysed by low-energy electron diffraction.

Employing near-stoichiometric surfaces, the valence electronic structure of CuInSe2 was investigated by angle-resolved photoelectron spectroscopy at the synchrotron source BESSY 2. This is the first study of the valence band structure of a copper chalcopyrite semiconductor material by photoelectron spectroscopy. The valence band dispersion along τT, i.e. the [001] direction, was investigated by a variation of the excitation energy from 10 to 35 eV under normal emission, and the band dispersion along τT, i.e. the [110] direction, was analysed by angular scans with hv = 13 eV.

The valence bands derived from antibonding and bonding Se4p-Cu3d hybrid orbitals, nonbonding Cu3d states and In-Se hybrid states are clearly indentified. The strongest dispersion is found for the topmost valence band with a bandwidth of ∼0.7 eV from τ to T. From τ to N, the observed dispersion was 0.5 eV. The experimental valence bands are discussed in relation to calculated band structures in the literature.

We discuss physical mechanisms underlying the performance and stability of CdTe based thin-film PV. The processes in (i) photovoltaic junction, (ii) back contact, (iii) nonuniformities, (iv) grain boundaries, and (v) light-induced degradation are addressed including their interactions. The physics of thin film PV turns out to be quite different from that of crystalline PV. High surface-volume ratio and lack of crystallinity result in strong interfacial effects, lateral nonuniformities, and shunting-like and adhesion instabilities in thin film structures. This paper is aimed at presenting a ‘big picture'; also, it suggests practical ways of improving thin-film PV.

We have investigated metastable changes to Cu(InGa)Se2 solar cells in response to infrared optical exposure and forward injection current. Using the drive level capacitance profiling and admittance spectroscopy techniques, we are able to distinguish changes in the free hole carrier concentration from those of a deeper bulk trap. For both types of treatment, changes in these concentrations clearly follow a 1:1 ratio, in agreement with recent predictions of a microscopic model proposed by Lany and Zunger. The creation kinetics follow a strongly sub-linear power law at 250K, and we demonstrate that we can account for this using rate equations in which the metastable effect is initiated by electron capture into an empty defect state. The dependence on time and light intensity is (time) 0.22± 0.03 and (intensity) 0.53± 0.06, except at high light intensities where saturation is observed.

Open-Circuit Voltage Decay (OCVD) is a common technique used to characterize numerous semiconductor devices. However, to the knowledge of the authors, the technique has not previously been applied to CdTe-based solar cells. We use a simple setup consisting of a function generator, rectifying diode, and digital oscilloscope to measure the dark open circuit voltage decay as a function of time across a CdTe solar cell. We find the decay to be described by the equation v(t) = v0 + A1exp (–t/τ1) + A2exp (–t/τ2) where v is the voltage, t is time, τ1 and τ2 are characteristic decay times, and A1, A2 and v0 are constants. The two characteristic decay times are on the order of 10 μs and 500 μs. The relative contribution of the two decay times depends on the magnitude of the initial applied voltage pulse. We will describe preliminary results on the correlation between the OCVD and solar cell performance, including the effects of light-soaking on OCVD behavior.

Using EBIC and EDX measurements, CIGS solar cells prepared under several different conditions were observed and characterized. The results of EBIC and EDX measurements suggest that Cd plays an important role in the forming of a buried pn-junction in the CIGS layer via diffusion, and de-emphasize the possibility of the formation of the hetero pn-junction at the CdS/CIGS heterointerface. The correlation of the extent of the space charge region and the observed shift in the pn-junction location with the diffusion of the constituent elements in CIGS was investigated.

The absorption, photoconductivity and photoluminescence spectra of Cd0.5Mn0.5Te thin films, deposited by “flash” evaporation and vapor phase deposition have been analyzed. The energy gap and the energetic position of the recombination levels in CdMnTe thin films for different substrate temperatures have been determined. The investigation results have been compared with the same parameters for CdMnTe single crystals used as evaporation material.

The highest performance CdS/CdTe thin film solar cells are generally completed with a Cucontaining back contact. The copper appears to be critical for achieving heavy p-type doping of the CdTe at the contact to permit the formation of a low resistance contact. In previous extended x-ray absorption fine structure (EXAFS) work we have inferred that most of the Cu in CdTe films resides as Cu2O at the boundaries of CdTe grains in films that have received a chloride treatment in the presence of oxygen, a critical step needed to improve the performance of all CdTe thin-film cells. This has suggested a mechanism for grain boundary passivation in thinfilm CdTe solar cells. We believe most of the diffused Cu decorates grain boundaries as oxides, consistent with the low doping densities typically observed in CdTe solar cells. The significance for grain boundary passivation will be discussed. We also find evidence that the grain-boundary Cu2O in CdCl2 treated CdTe films is unstable and tends to transform to CuO under some stress conditions.

CuGaSe2 absorber layers were prepared by evaporating elemental Cu, Ga and Se in three stage on Molybdenum coated soda lime glass. The composition of the resultant film was studied by monitoring the substrate temperature, which decreased when a Cu-Se secondary phase was formed. As the Ga supplement increased during the third stage the void that formed in the beginning of the third stage was removed, while a small grain Ga-rich layer was formed on the surface, resulting in a Cu deficient surface. Therefore, the Voc was improved because of the enhanced surface morphology and the Jsc was reduced, due to the Ga rich layer on of the surface. Under optimal conditions, we achieved a cell performance of Voc = 780 mV, Jsc = 12.9 mA/cm2, ff = 62.5 and ν = 7.3 %.

We have fabricated CIGS:Fe polycrystalline thin films using a standard three-stage method, and investigated the effects of Fe doping on cell performances. The Ga / (In+Ga) ratio was varied between 0.3 ˜ 1.0 (= CGS), and the Fe concentration was varied between 0.0 ˜ 1.2 mol%. The films were characterized by various means, including the cell performance. Increment of the grain size with higher Fe content was observed. Redshift with higher Fe content was observed in the absorbance spectra. The spectral response of the fabricated solar cells deteriorated with higher Fe content, from the long wavelength side.

Electrodeposition (ED) of CuInSe2-based thin films from a buffered single-bath on dcsputtered Mo layers has been investigated. In order to understand the film growth, cyclic voltammetry (CV) was used to identify mechanisms leading to the formation of CuInSe2 and Cu(In,Ga)Se2. Similar CV data were observed for deposition from Cu-Se, Cu-In-Se, Cu-Ga-Se, and Cu-In-Ga-Se baths. A preliminary mechanism for CuInSe2 and Cu(In,Ga)Se2 film growth has been proposed from CV, composition and glancing incidence x-ray diffraction (GIXRD) data. Incorporation of In into the growing films occurs via reaction with H2Se, formed by reduction of the initially deposited Cu3Se2, to form In2Se3, which is in turn rapidly assimilated into the film by reaction with copper selenides to form CuInSe2. The incorporation of Ga may occur via a similar mechanism, however, the precipitation of Ga(OH)3 can not be ruled out as a possible route for Ga uptake. All as-deposited ED CuInSe2–based films have poor crystallinity and require annealing in H2Se prior to device processing. Preliminary device results are presented, reporting a conversion efficiency of 6.5% and 6.2% for electrodeposited CuInSe2 and Cu(In,Ga)Se2, respectively.

To fabricate a high-efficiency polycrystalline thin-film tandem cell, the most critical work is to make a high-efficiency top cell (>15%) with high bandgap (Eg=1.5-1.8 eV) and high transmission (T>70%) in the near-infrared (NIR) wavelength region. The CdTe cell is one of the candidates for the top cell, because CdTe state-of-the-art single-junction devices with efficiencies of more than 16% are available, although its bandgap (1.48 eV) is slightly lower for a top cell in a dual-junction device. In this paper, we focus on the development of an ultra-thin, low-bandgap CuxTe transparent back-contact to produce high-efficiency CdTe cells with high NIR transmission. We have achieved an NREL-confirmed 13.9%-efficient CdTe transparent solar cell with an infrared transmission of ~50% and a CdTe/CIS polycrystalline mechanically stacked thin-film tandem cell with an NREL-confirmed efficiency of 15.3%.