We have theoretically evaluated the phase stability and electronic structure of Cu2ZnSnSe4 (CZTSe) and Cu2ZnSnS4 (CZTS). The enthalpies of formation for kesterite, stannite and wurtz-stannite phases of CZTSe and CZTS were calculated using a plane-wave pseudopotential method within the density functional formalism. For CZTSe, the calculated formation enthalpy (ΔH) of the kesterite phase (−312.7 kJ/mol) is a little smaller than that of the stannite phase (−311.3 kJ/mol) and much smaller than that of the wurtz-stannite phase (−305.7 kJ/mol). For CZTS, the ΔH of the kesterite phase (−361.9 kJ/mol) is smaller than that of the stannite phase (−359.9 kJ/mol) and much smaller than that of the wurtz-stannite phase (−354.6 kJ/mol). The difference of ΔH between the kesterite and stannite phases for CZTS is greater than that for CZTSe. This indicates the kesterite phase is more stable than the stannite phase in CZTS compared with CZTSe. The valence band maximums (VBMs) of both the kesterite- and stannite-type CZTSe(CZTS) are antibonding orbitals of Cu 3d and Se 4p (S 3p). The conduction band minimums (CBMs) are antibonding orbitals of Sn 5s and Se 4p (S 3p). The Zn atom does not affect the VBM or the CBM in either CZTSe(CZTS). The theoretical band gap of the kesterite phase calculated with sX-LDA in both CZTSe and CZTS is a little wider than that of the wurtz-stannite phase and much wider than that of the stannite phase.

We report the first studies of electroabsorption in Cu(InGa)Se2 (CIGS) solar cell devices. We utilized a bifacial CIGS device with a Ga/(In+Ga) ratio of 0.8 (bandgap of 1.5 eV) deposited onto semi-transparent (40 nm thick) Mo coated glass as the back contact. By modulating the electric field using a small sinusoidal potential of amplitude δV across the CIGS layer, we were able to detect the modulation ΔT of the transmitted light. This was examined as a function of photon energy, DC bias, temperature, and modulation frequency (100 Hz to 10 kHz) and had a maximum amplitude of ΔT/T ≈ 10−5 for δV = 0.3 V. Very different characteristics were obtained for near bandgap light (1.3 eV) compared to photon energies considerable smaller (<0.95 eV). While the latter exhibited a strong temperature and frequency dependence, indicating an important role for deep defects in the effect, the former exhibited very little change with temperature or frequency, indicating the predominance of transitions involving bandtail states. Different metastable states of the CIGS layer produced by prolonged light soaking above the bandgap energy were also examined.

Chemical bath deposition (CBD) is a commonly used method of depositing cadmium sulfide (CdS) films for photovoltaic application. The method is based on decomposition of a sulfur source in an alkaline solution of a cadmium source on the surface of the Cu(In,Ga)Se2 (CIGS) substrate. On the lab scale the CdS film is deposited by submerging a 1” square CIGS substrate in a heated beaker containing the chemical bath. This batch processing method is the one used for record-performing devices. There is an ongoing effort at the National Renewable Energy Laboratory to scale-up the CBD process to deposit CdS films on 6” square substrate. Efforts are focused at designing both batch and flow reactors for depositing uniform, device quality CdS films on larger substrates. Batch reactor designs involve reproducing the deposition process in the beaker on a bigger scale with minimal chemical waste, while flow reactors are designed for continuous processing, such as encountered in roll-to-roll manufacturing lines.

Silver antimony selenide (AgSbSe2) thin films were prepared by heating sequentially deposited antimony sulphide (Sb2S3), silver selenide (Ag2Se) and Ag thin films in close contact with a selenium thin film. Sb2S3 thin film was prepared from chemical bath containing SbCl3 and Na2S2O3, Ag2Se from the bath containing AgNO3 and Na2SeSO3 and Se thin films from an acidified solution of Na2SeSO3, at room temperature on cleaned glass substrates. Ag thin film was deposited by vacuum thermal evaporation. The annealing temperature was varied from 300-390°C in vacuum (∼10−3 Torr) for 1 h. X-ray diffraction analysis showed the films formed at 350 °C was polycrystalline AgSb(S,Se)2 or AgSbSe2 depending on selenium thin film thickness. Morphology of these films was analyzed using Atomic Force Microscopy and Scanning Electron Microscopy. The elemental analysis was done using Energy Dispersive X-ray technique. Optical characterization of the thin films was done by optical transmittance spectra. The electrical characterizations were done using Hall effect and photocurrent measurements. A photovoltaic structure: Glass/ITO/CdS/AgSbSe2/Ag was formed, in which CdS was deposited by chemical bath deposition. J-V characteristics of this PV structure showed Voc=370 mV and Jsc=0.5 mA/cm2 under illumination using a tungsten halogen lamp.

Several wet-processing steps are used in fabricating high-efficiency CdTe/CdS solar cells. These steps can hinder in-line processing; thus, developing an all-dry processing option is attractive for a manufacturing-friendly process. In this study, we systematically modified the baseline process used in our laboratory to replace CdS deposited by chemical-bath deposition (CBD) with sputter-deposited CdS and Cu-doped graphite paste back-contact with Cu-doped ZnTe deposited by radio-frequency sputtering. In addition to CdTe deposited by close-spaced sublimation, we also used conventionally evaporated CdTe. The results show that replacing only CBD CdS with oxygenated CdS deposited by sputtering produces devices with performance comparable to baseline devices if the front bilayer SnO2 is replaced by a Cd2SnO4/ZnSnO alloy. Replacing the graphite paste back-contact with sputter-deposited Cu-doped ZnTe resulted in device performance comparable to baseline devices. Incorporating both dry processing steps gave performance comparable to the devices with sputtered CdS with a SnO2 front contact. We used capacitance-voltage and minority-carrier lifetime measurements to analyze the factors affecting device performance and we present the results here.

A variety of junction capacitance-based characterization methods were used to investigate alloys of Ag into Cu(In1-xGax)Se2 photovoltaic solar cells over a broad range of compositions. Alloys show encouraging trends of increasing VOC with increasing Ag content, opening the possibility of wide-gap cells for use in tandem device applications. Drive level capacitance profiling (DLCP) has shown very low free carrier concentrations for all Ag-alloyed devices, in some cases less than 1014 cm−3, which is roughly an order of magnitude lower than that of CIGS devices. Transient photocapacitance spectroscopy has revealed very steep Urbach edges, with energies between 10 meV and 20 meV, in the Ag-alloyed samples. This is in general lower than the Urbach edges measured for standard CIGS samples and suggests a significantly lower degree of structural disorder.

Technology variations involving Cu and Cl impurities are among the major performance influencing factors for CdS/CdTe thin film solar cells. CuCl and CdCl2 influence on the energetic diagram of impurity levels with respect to variation of deposition parameters has been investigated. A comparative analysis has been carried out by using low temperature photoluminescence (PL) studies (17-98K) of CdTe thin films in the device configuration (from CdS/CdTe inteface and CdTe sides). To study the effect of CuCl influence, as-deposited, annealed heterojunctions, with CuCl treatment of CdS have been investigated.

CdTe, with a direct band gap of 1.45 eV is well suited to the terrestrial AM1.5 solar irradiance and currently makes up half of the thin film (TF) photovoltaic (PV) market. There are 4 main factors that determine final cost of PV modules: the conversion efficiency, materials amount per unit area of module, production yield, and the economy of scale. It is therefore valuable to investigate alternative and/or innovative deposition techniques and processes which have the potential to impact on these factors. Metal organic chemical vapour deposition (MOCVD) is a powerful technique offering increased process repeatability, achieving a high level of control of materials characteristics. Recent improvements in CdTe devices using atmospheric–pressure (AP) MOCVD have led to efficiencies of 13.3 % using 2 μm absorber layers and 11 % with a 1 μm absorber layer [11, 12]. These results were achieved: by extending the optical band gap of the window layer, using a ternary alloy (Cd0.9Zn0.1S), with intentional p-type doping of the CdTe layers with As, and the use of an in situ (dry) deposited CdCl2 layer and anneal. All layers, except the front and back contact, are grown by a sequential MOCVD process. Furthermore this is a dry process without the need for any etch treatment. The inherent design of the horizontal MOCVD laboratory chambers do not lend themselves well to large scale production. However, the CSER group has designed and built an experimental inline reaction chamber to evaluate AP-MOCVD as an inline production process. Discussion is made based on kinetically limited growth and molar supply models to assess the suitability of the MOCVD process to deposit fast enough for an inline process. The AP-MOCVD inline reactor uses a showerhead to deliver the precursors onto a moving 5 × 7.5 cm2 substrate and preliminary results for deposited layers are given. From these preliminary results it has been extrapolated that a 1 μm thick CdTe layer can be deposited on substrates moving at 60 cm/min.

We studied CuGaSe2 (CGS) thin films with different Cu contents by means of X-ray diffraction (XRD) and micro-Raman spectroscopy. The CGS absorbers were deposited by co-evaporation on Mo/glass substrates. We found a clear shift of the CGS Raman mode frequencies to lower values with increasing Cu/Ga ratio. This is in direct correlation with the increasing lattice constants a and c extracted from XRD patterns. Influence of stress on the obtained results can be neglected, because very small stress values below 50 MPa were determined with the sin2Ψ method.

We have investigated different methods for preparing CdTe/CdS cross sections for electrical measurements, including the following: cleaving; using GaAs substrates; and sandwiching the structure between the substrate and a glass slide, and polishing with diamond discs and alumina suspension. The latter method proved to be the most reliable, with a success rate of over 90%.

We investigated cross sections of CdTe/CdS samples with scanning Kelvin probe microscopy (SKPM) using two different methods: applying the alternate bias with a frequency equal to 18.5 kHz, or equal to the frequency of the second cantilever resonance peak. The results showed that using the second resonance frequency produced a smoother signal, allowing the calculation of the electric field inside the device using just the raw SKPM data.

We were able to measure the distribution of the electrical potential inside working devices. Then, by taking the first derivative of the potential, we calculated the electric field and determined the location of the p-n junction.

The early stage formation mechanisms operating during the sublimation growth of CdTe films on CdS has been evaluated using a growth interrupt methodology for deposition under 100 Torr of N2. Key stages of the growth were identified and are discussed in terms of the processes of island nucleation, island growth/coalescence, channel formation and secondary nucleation that have been reported for other materials systems. It was demonstrated that the grain size could be manipulated by means of controlling the gas pressure in the range 2 – 200 Torr, with the grain diameter increasing with pressure linearly as D (μm) = 0.027(± 0.011) × P (Torr) + 0.90(± 0.31). For a series of solar cells made using such material, the performance parameters were seen to increase with grain size up to a plateau corresponding to grains of ∼4 μm in this case. Equivalent circuit parameters for resistive components arising from grain boundaries, and the contact to the CdTe, were measured. It is considered that grain boundary barriers in CdTe are harmful to PV performance, and that the plateau in performance occurs when the grain size is increased to the level where the contact resistance is greater than that due to grain boundaries.

The role of intrinsic point defects on radiative recombination in Cu(In,Ga)Se2 thin films was investigated by photoluminescence (PL) and photoluminescence excitation (PLE) spectroscopies. Experiments were performed on device-grade polycrystalline layers and single crystal thin films. PL transitions identified by others as indicating a shallow state with an ionization energy of ∼16 meV is proposed to be a transition into band tail states rather than a distinct shallow defect. The presence of deep levels contributing to radiative recombination does not necessarily preclude the material from producing a high efficiency device and may suggest the absence of dominant non-radiative recombination pathways. The band edge width as measured by PLE and the separation of this edge from defect states are suggested to be potentially effective indicators of the quality of a material. Luminescence that appears to be connected with the absence of Na in the growth process persists in high Ga alloy, Na containing materials, suggesting that Na may become ineffective in passivating or eliminating certain defects in high Ga material.

In this study, the pn hetero-interface between Zn(O,S,OH)x buffer and Cu(InGa)(SSe)2 (CIGSS) surface layers is discussed in order to achieve the fill factor (FF) over 0.73 and the circuit efficiency of 16 % on aperture area of over 800 cm2. Two resistances, i.e. shunt resistance (Rsh) and series resistance (Rs), in the circuits are employed as a yardstick to evaluate the interface quality. Since there are no realistic yardsticks on the Rs, the difference between Voc and optimum-power voltage (Vop) (i.e. Voc-Vop [V/cell]) is applied as a simple tool to evaluate the Rs. It is emphasized that it is important to reduce the Rs mainly correlated to the buffer deposition process and, as a result, the interface quality. We consider the Rs is dependent on the remaining Zn(OH)2 concentration in the Zn(O,S,OH)x buffer deposited by a chemical-bath deposition (CBD) technique. As an approach to make the Rs minimize and the Rsh maximize simultaneously, adjusting the thickness of a CBD-Zn(O,S,OH)x buffer layer and a non-doped ZnO layer deposited by a metal-organic chemical vapor deposition (MOCVD) technique has been effective to reduce the remaining Zn(OH)2 concentration. Determining the optimized deposition procedure to achieve the FF over 0.700 consistently, the circuit efficiency of 15.3 % with aperture area of 856 cm2 and the FF of 0.717 has been achieved.

The CdTe thin film module technology can be considered mature for large scale application.

First Solar is now capable of producing more than 400MW/year of CdTe modules and it is still increasing its production. First Solar also announced that they can get a production cost close to 1$/W. A further cost reduction will render this kind of energy production competitive with the energy obtained from fossil fuels.

In our laboratory we developed a novel dry process that could further reduce the production cost.

This process is now being used by a new company (Arendi Spa) in a line that will produce ∼15-18MW/year. Half of the line has been already built and the complete line should be ready in July 2009. The module production will start at the end of 2009.

Two types of commercial electrically conductive adhesive (CA) tapes, one isotropic and the other anisotropic, were investigated for their potential application as interleafing connects for flexible thin-film photovoltaic modules. The performance stability of their vacuum-laminated sample constructs between two 50-μm stainless steel (SS) foils and between the SS foil and bilayer ZnO/Mo-coated SS foil was evaluated upon damp-heat and dry-heat exposure. Preliminary results indicated that the isotropic CA tape was more stable than the anisotropic tape.

Alternative energy sources such as thin film photovoltaics can be accelerated by improving the rapid and successful transition from laboratory research innovation to commercial production. Most laboratory research and development is on a small scale and its production is in small volumes. It focuses on exploration, discovery, and understanding. When the successful innovation is commercialized, both the scale and the volume increase dramatically and the focus shifts to performance, reliability, yield and cost. This transformation can be accelerated by closely managing risk and by integrating the equipment design and the process development. Also, the cadmium telluride photovoltaic technology has properties that make it more amenable to rapid scale up to low cost and high volume manufacturing.

A 2-D numerical circuit model is used to analyze the impact of shunts on basic performance parameters of a CdTe thin-film module. A numerical estimate of module-efficiency loss in the worst-case scenario due to shunts of different severity and fractional module area is presented. It is shown that absolute module-efficiency loss Δη (%) varies in systematic fashion with these shunt parameters. Estimates of Δη based on simple area-weighted efficiency are typically low by 3-4 times. Furthermore, the distribution pattern of shunts over the module plays a significant role in the module loss. A reliable parameter P to characterize the distribution of shunts is introduced, and its effect on module-efficiency loss, as well as individual-parameter (FF, VOC and JSC) losses, is shown. Furthermore, higher transparent-conductive-oxide (TCO) sheet resistance is shown to increase shunt isolation and consequently mitigate the efficiency decrease.

CIS based chalcopyrite absorber materials are usually substituted in the cation and anion lattice to yield mixed pentanary crystals with the general composition Cu(In,Ga)(Se,S)2 to achieve an optimised adaptation of the semiconductor bandgap to the terrestrial solar spectrum. Real-time investigations during the annealing of stacked elemental layers (SEL) of sputtered metals Cu and In and evaporated chalcogens S and Se with varying ratios were performed by angle-dispersive time-resolved XRD (X-ray diffraction) measurements. After qualitative phase analysis the measured powder diagrams were quantitatively analysed by the Rietveld method, the phases formed determined and their reaction kinetics obtained. Ternary indium and copper sulfoselenides form by the sulfoselenisation of the intermetallic alloy yielding different educts for the chalcopyrite formation with varying sulfur content. For S/(S+Se) ≥ 0.5 the formation of the chalcopyrite CuIn(S,Se)2 is similar to the crystallisation path of CuInS2. With increasing amount of selenium (S/(S+Se) = 0.25) different ternary sulfoselenides contribute to the semiconductor formation. For small amounts of sulfur, i.e. S/(S+Se) ≤ 0.1, the chalcopyrite crystallisation proceeds comparable to the one observed for sulfur-free Cu-In-Se precursors. The formation of CuIn(S,Se)2 is accelerated and proceeds mainly after the peritectic decomposition of Cu(S,Se) to Cu2(S,Se). The sulfur content determines the crystallisation temperature of the semiconductor because Cu(S,Se) decomposes at higher temperatures with increasing sulfur. Upon heating S ↔ Se exchange reactions take place in the Cu-S-Se and Cu-In-S-Se system.

Indium (III) sulfide has recently attracted much attention due to its potential in optical sensors as a photoconducting material and in photovoltaic applications as a wide direct bandgap material. On the other hand, optical absorption properties are key parameters in developing highly photosensitive photodetectors and high efficiency solar cells. We show that indium sulfide nanorod arrays produced by glancing angle deposition techniques have superior absorption and low reflectance properties compared to conventional flat thin film counterparts. We observed an optical absorption value of approximately 96% for nanorods, in contrast to 80% for conventional amorphous-to-polycrystalline thin films of indium sulfide. A photoconductivity response was also observed in the nanorod samples, whereas no measurable photoresponse was detected in conventional thin films. We give a preliminary description of the enhanced light absorption properties of the nanorods by using Shirley-George Model that predicts enhanced diffuse scattering and reduced reflection of light due the rough morphology.

Transient photocurrent (TPI) and photocapacitance (TPC) spectroscopy have been applied to a set of compositional graded CuIn1-xGaxSe2 (CIGS) solar cell devices deposited by the vacuum co-evaporation method at the National Renewable Energy Laboratory. These measurements provide a spectral map of the optically induced release of carriers for photon energies from below 1 eV to 2 eV. By comparing the two types of spectra one can distinguish majority from minority carrier processes and they clearly reveal a higher degree of minority carrier collection for devices in which the Ga fraction increased monotonically with distance from the junction. This agrees with notions of how compositional grading improves overall cell performance. Minority carrier collection was even more strongly enhanced in sample devices incorporating v-shaped Ga-grading. Spatial profiles of the free hole carrier densities and deep acceptor concentrations were examined using drive-level capacitance profiling (DLCP). In the compositionally graded sample devices we found that the free carrier density decreased and that defect density increased with increasing Ga fraction toward back contact.