Generation, transport and collection of carriers in polycrystalline (PX) solar cells and their constituent materials are poorly understood, and significantly different than in their single-crystal counterparts. Recent theoretical and experimental results have put forth the expectation that grain boundaries in PX-solar cell materials such as CdTe and CuInGaSe2, either as-grown or after appropriate post-growth treatment, may have electronic properties which are advantageous to charge separation and solar cell operation[1-3]. However, a microscopic picture of the spatial variations in the optoelectronic properties of these materials is, for the most part, still lacking. The goal of the work reported here is to explore the optoelectronic and spectroscopic properties of grain-boundaries in these materials at the nanometer length-scale, via novel, high-resolution optical techniques. Towards this end, a significant enhancement in photo-response near grain boundaries in CdTe solar cells, consistent with models put forth in reference 2, was observed via near-field Optical Beam Induced Current (n-OBIC) [4]. A systematic μ-PL study of the effect of CdC12-treatment on recombination in CdTe/CdS solar cell structures of varying thickness directly examined the variation in optoelectronic properties at grain-boundaries in this material, revealing the grain-boundary and surface passivation effects of this important post-growth processing step. For comparison, we also studied the effects of SiNx post-growth treatment and annealing on the photo-response of PX-silicon solar cells using n-OBIC. These results and our most-recent n-OBIC measurements in CdTe and CuInGaSe2 solar cells are discussed.

Solar cells based on CuInS2 still suffer from Voc limitation compared to the value expected from the bandgap. The open circuit voltage can be enhanced via controlled doping of small amounts of Zinc or Magnesium (< 1 at. %), however the underlying physical reason for this improvement is not understood so far. We have systematically studied the effect of diffusion of Zn and Mg into the absorber layers with photoluminescence and quantum efficiency measurements. ZnO (ZnMgO, MgF2) has been deposited onto the absorber layers and photoluminescence (PL) spectra at room temperature were recorded as a function of the annealing temperature. The modified emission spectra reveal a defect related transition with a strong luminescence signal. At the same time the open-circuit voltage of the Zn (Mg) doped CuInS2 cells are around 100 mV higher than the values found for the undoped cells from the same device process. We interpret these results as a reduced defect state density at the surface and an increased bulk defect concentration induced by the incorporation of Zn or Mg. Comparison of external quantum efficiency (EQE) and I/U measurements with the PL results support observed changes of the interface/bulk properties.

Using the combination of high-resolution transmission electron microscopy, first-principles density-functional total-energy calculations, and image simulations, we studied the atomic structure and passivation effects of double-positioning (DP) twin boundaries in CdTe. The DP twin boundaries are found to contain more Te dangling bonds than Cd dangling bonds, resulting in energy states in the bandgap that are detrimental to the electronic properties of CdTe. We found that I, Br, Cl, S, and O atoms present passivation effects on the DP twin boundaries to differing degrees, whereas H does not passivate the boundaries. Of all these impurities, I and Cl atoms present the best passivation effects on the DP twin boundaries. The superior passivation effects are realized by either terminating the Cd atoms with dangling bonds, or substituting the Te atoms with dangling bonds in the DP twin boundaries in CdTe by Cl and I atoms.

To characterize the influence of sodium diffusion on the Cu(In,Ga)Se2 (CIGS) layer, samples with and without an A12O3 diffusion barrier were investigated by transmission electron microscopy. The soda lime glass acts as a sodium source for samples without diffusion barrier while a post deposition treatment with NaF adds sodium to the CIGS layer for samples with diffusion barrier. For low deposition temperatures in the three-stage growth process for CIGS slightly columnar CIGS grains near the Mo back contact are found and larger grains near the top surface. Samples subject to high-temperature post deposition treatments show slightly larger grain sizes in the CIGS layer than samples prepared at the same temperature but with sodium from the soda lime glass. The surface smoothness of the CIGS layer and therefore the continuity of the following layers is improved for samples with higher deposition temperature. Occasional defects are found on the surface of the ZnO layer which can locally alter reflectivity and absorption properties of these solar cells. Despite the high CIGS surface roughness, the CdS layer with an average thickness of only 50 nm appears continuous and even fills trenches between CIGS grains near the interface.

Real time spectroscopic ellipsometry (RTSE) based on rotating-compensator modulation and multichannel detection has been implemented to characterize polycrystalline thin film CdTe deposition for photovoltaic applications. RTSE is capable of providing routine deposition information on substrate temperature T and deposition rate. It is also capable of providing detailed information on the thickness evolution of microstructure and optical properties. In this study, we highlight the differences in nucleation that occur under different CdTe deposition conditions on smooth crystalline Si wafer substrates. Differing behavior in the initial stages of deposition has been observed, ranging from layer-by-layer growth to nucleation and coalescence of 45 Å thick clusters. We also consider the thickness and substrate dependence of the microstructure, comparing depositions on smooth Si wafer and rough thin film Mo substrates.

In situ sensors are an important tool for process control, optimization, and documentation, both in the laboratory and industrial environments. Their further application to deposition of CuInxGa1-xSe2 (CIGS) for photovoltaics is particularly important, as record device efficiencies produced in the laboratory have yet to be replicated in manufacturing. This paper provides an overview of the current state of the art of in situ diagnostics for devices based on coevaporated CIGS.

Glancing incidence x-ray diffraction (GIXRD) is a powerful diagnostic tool for nondestructive analysis of thin film materials and structures. GIXRD fundamentals and experimental techniques and results are presented with respect to polycrystalline thin-films. The refraction index and critical incident beam angle are calculated for selected elemental and inorganic compounds. Quantitative GIXRD analysis with Cu Kα radiation is applied to several problems: recrystallization of ultra-thin chemical surface deposited CdS films; compositional grading in CuIn1-xGaxSe2 films; formation of native oxides on CdTe films; interdiffusion between CdTe and CdS; Te film thickness determination on etched CdTe; and correlation of inplane residual stress with deposition conditions for sputtered Mo films.

In thin film solar cells interfaces between lattice mismatched or dissimilar materials are used for the front and the back contact. A p-i-n device structure should be possible as most simple but ideally suited thin film solar cell. In contrast the interfaces in CdTe solar cells are found to be much more complex containing interdiffused phase boundaries at the front as well as at the back contact. By comparison to non-interdiffused interfaces using contact phases of adapted work functions it can be shown that the contact potentials of the front contact but also of the back contact are dominated by Fermi level pinning. The pinning states are evidently due to dislocation defects at the boundary of CdTe to the contact phases. Based on these results it is concluded that interdiffused phase boundaries or appropriate passivation layers are a precondition for efficient solar cells whenever strongly lattice mismatched or dissimilar materials are combined.

Polycrystalline thin-film CdTe is one of the leading materials used in photovoltaic solar cells. One way to improve device performance and stability is through understanding how various process steps alter defect states in the CdTe layer. Low-temperature photoluminescence (PL) studies show a 1.456-eV PL peak in single-crystal CdTe that is likely due to a Cui-OTe defect complex. A similar peak, observed in as-deposited glass/SnO2:F/CdS/CdTe thin-film structures, strongly suggests a common origin. The 1.456-eV peak is also seen in a thin-film sample after performing the CdCl2 treatment needed for high efficiencies.

Variable angle spectroscopic ellipsometry has been used to characterize Cu(InGa)Se2 thin films as a function of relative Ga content and to study the effects of Cu off-stoichiometry. Uniform Cu(InGa)Se2 films were deposited on Mo-coated soda lime glass substrates by elemental evaporation with a wide range of relative Cu and Ga concentrations. Optical constants of Cu(InGa)Se2 were determined over the energy range of 0.75–C4.6 eV for films with 0 ≤ Ga/(In+Ga) ≤ 1 and used to determine electronic transition energies. Further, the changes in the optical constants and electronic transitions as a function of Cu off-stoichiometry were determined in Cu-In-Ga-Se films with Cu atomic concentration varying from 10 to 25 % and Ga/(In+Ga) = 0.3. Films with Cu in the range 16–24 % are expected to contain 2 phases so an effective medium approximation is used to model the data. This enables the relative volume fractions of the two phases, and hence composition, to be determined. Two distinctive features are observed in the optical spectra as the Cu concentration decreases. First, the fundamental bandgaps are shifted to higher energies. Second, the critical point features at higher energies become broader suggesting degradation of the crystalline quality of the material.

Cu-doped ZnTe films grown by galvanic deposition were developed to provide transparent ohmic contacts to CdTe solar cells. Control of the Cu doping with triethanolamine was critical to limit the free Cu in order to achieve high transparency (>60% in the visible) and to minimize shunting. Devices with ZnTe:Cu contacts had comparable performance to devices with Cu/Au or Cu/graphite contacts, achieving Voc of 0.79 V and fill factor (FF) of 68% for standard front illumination. Bifacial spectral response (SR) measurements, through front and back contacts, were analyzed and yielded a diffusion length (L) of 0.8 μm and depletion width (W) of 2.5 μm. Backwall SR measurements made through the transparent ZnTe contact are much more sensitive to L and W than are measurements for standard front illumination. ZnTe:Cu is a promising material for bifacial characterization as well as tandem cell interconnects and more stable Cu-doped contacts.

Zn1-yMgyO bandgap controllable transparent conducting films were used for the wide-gap Cu(In1-xGax)Se2 thin film solar cells. Undoped Zn1-yMgyO and Al doped Zn1-yMgyO films were deposited by co-sputtering using a carousel type sputtering apparatus. Zn1-yMgyO films with Mg content y of up to 0.10 were examined. For Cu(In1-xGax)Se2 with band gap energy ˜1.38 eV, the cell performance was slightly improved by using Zn1-yMgyO and Al doped Zn1-yMgyO instead of ZnO and Al doped ZnO. An unexpected improvement of short circuit current density was observed.

The electronic properties of sulfur containing CIS chalcopyrite alloys have been characterized using junction capacitance methods. Two devices were examined; one containing CuIn(S,Se)2 alloy with a 1:2 S:Se ratio and a bandgap near 1.15eV, and the other an endpoint CuInS2 alloy with a bandgap slightly above 1.5eV. Drive-level capacitance profiling measurements indicated hole carrier densities of less than 1 x 1015 cm-3 and 1.5 x 1016 cm-3, respectively. Transient photocapacitance (TPC) sub-bandgap spectroscopic measurements revealed sharp bandtails plus a broad defect band within the bandgap of each alloy. The TPC spectra for the CuInS2 sample revealed a couple of unusual features, including a bandtail signal that reversed sign below 250K. This indicated poorer hole collection than electron collection in the low temperature regime. Comparing these results to TPC spectra obtained previously for Cu(InGa)Se2 alloys indicate some similarities but also some striking differences.

Direct characterization of band alignment at chemical bath deposition (CBD)-CdS/Cu0.93 (In1-xGax)Se2 has been carried out by photoemission spectroscopy (PES) and inverse photoemission spectroscopy (IPES). Ar ion beam etching at the condition of the low ion kinetic energy of 350 eV yields a removal of surface contamination as well as successful measurement of the intrinsic properties of each layer and the interfaces. Especially interior regions of the wide gap CIGS layers with a band gap of 1.4 ∼ 1.6 eV were successfully exposed. IPES spectra revealed that the conduction band offset (CBO) at the interface region of the wide gap CIGS with x = 0.60 and 0.75 was negative, where the conduction band minimum of CdS was lower than that of CIGS. It was also observed that the energy spacing between conduction band minimum (CBM) of CdS layer and valence band maximum (VBM) of Cu0.93(In0.25Ga0.75)Se2 layer at interface region was no wider than that of the interface over the Cu0.93(In0.60Ga0.40)Se2 layer.

Sn-doped indium oxide (ITO) films deposited at temperatures <250°C by r.f. magnetron sputtering are investigated for use as transparent back-contact electrodes to CdS/CdTe top cells in two-junction, mechanically stacked tandem-cell designs. Results show that carrier mobility increases with substrate temperature while beneficial changes in carrier activation occurs at a substrate temperature ~230°C. Small additions of oxygen to the sputtering environment are effective at reducing near infrared (NIR) absorption. Control of these deposition parameters has produced ITO films with NIR absorption (860–1300 nm) < 10%, and electrical resistivity of <1x10-3 ohm-cm.

The complementary techniques of electron paramagnetic resonance (EPR) and photoelectron spectroscopy (PES) have been used to study the native oxidation of CuGaSe2 crystals and polycrystalline thin films. After storage of specimen under ambient conditions for few months, an EPR signal occurred which is assigned to Cu2+ and is found independently from the sample morphology. This signal is due to the formation of a copper hydroxide surface phase, Cu(OH)2, observed only after long term oxidation. Chemical etching in KCN removes and thermal reduction by annealing in vacuum at 200°C reduces this Cu(OH)2 surface phase as proved by EPR and PES. Implications for solar cell device performance will be discussed.

We report on the electronic properties of the state-of the art surface-modified CuGaSe2 solar cells. We compare between the 10.2%-efficient surface-modified and the 9.53%-efficient unmodified CuGaSe2 solar cells. We examined our cells using deep level transient spectroscopy (DLTS) and drive level capacitance profiling (DLCP). The DLTS data for the modified CGS exhibits minority traps with activation energies ranging from Ec — 0.3 to Ec — 0.6 eV, whereas that for the pure CGS exhibits minority traps with activation energies ranging from Ec — 0.06 to Ec — 0.1 eV (where Ec is the energy of electrons at the conduction band minimum). While varying the filling pulse duration, we observed the gradual increase in the amplitude of the DLTS signal for these states until it apparently saturates at a pulse duration of ˜1s for the 10.2% cell, and 0.05 s for the 9.53% cell. Increasing the duration of the filling pulse also results in broadening of the DLTS signals and shifting of the maximum of these signals towards lower temperature, whereas the high-temperature sides coincide. Using a model that allows us to distinguish between bandlike states and localized ones based on the dependence of the shape of their DLTS-signal on the filling-pulse duration, we relate the electron trap to bandlike states. The DLCP data shows that the 10.2% cell has a lower carrier concentration, a more uniform defect density profile, a larger depletion width, and a higher drift collection length for photo-generated carriers as compared to our 9.53% CuGaSe2 cell. We also recorded the transient capacitance versus time and found that the 10.2% has responded differently compared with the 9.53% one. The transient capacitance decay curves for these two cells are different.

A reactive sputtering process was developed for the production of Cu(In,Ga)Se2 films on a moving stainless steel substrate, in simulation of the operation of a roll coater. Cu, In and Ga fluxes were provided through magnetron sputtering and were reacted in a flux of Se on the heated substrate. CdS films were deposited either by chemical bath deposition (CBD) or by sputtering. Devices of the type steel/Cr/Mo/CIGS/CdS/ZnO/Ag were completed by sputtering ZnO layers and by screen printing grid lines. We made devices with efficiency values above 9%. A uniformity study was performed on a CIGS film and on small area devices made from it. The target length was 12”. Targets of this size are expected to produce a uniformly thick deposit over a range of 6-8”. The film thickness was 2.54 μm over a range of 6” with a standard deviation ó of 0.04 μm. The film composition was uniform over a range of 16”. The values of Cu/III and Ga/III were 0.84 and 0.31, with ó values of 0.02 and 0.01, respectively. The efficiency of allsputtered devices was uniform over a range of 12”, well beyond the 6” wide range of constant CIGS film thickness. Their efficiency was 6.6% on average with ó=0.6%.

CuInS2 thin films have been prepared by reactive magnetron sputtering from metallic targets in an Ar/H2S atmosphere. The Cu/In ratio and substrate temperature have been varied and absorber layers were processed to solar cell devices using the HMI baseline process. Device efficiencies up to 8.8 % have been achieved. The electronic and optical properties of the solar cell devices were analyzed by Scanning Electron Microscopy, Raman, photoluminescence, current-voltage and quantum efficiency measurements. Raman measurements indicate that Cu-Au defect ordering is present in some of the films, which can be eliminated by appropriate choice of the growth parameters. Photoluminescence measurements show spectra very similar to the spectra found for cells from the two-step sequential process indicating a similar defect structure in these films. No principle obstacles to establishing a one-step reactive magnetron sputtering process for CuInS2 solar cell absorbers have been identified.

The extent to which grain boundaries (GBs) in polycrystalline materials may be detrimental, benign, or even beneficial is explored with numerical simulations in two dimensions. We focus on the effects of GB recombination in Cu(In,Ga)Se2 (CIGS) solar cells and its effects on solar-cell performance. The simulations predict that (1) for device effciency exceeding 17%, the effective GB recombination velocity must be less than 104 cm/s; (2) grain boundaries within the space-charge region (SCR) lower the open-circuit voltage, whereas the short-circuit current is reduced by grain boundaries in the bulk material; and (3) horizontal GBs are relatively benign unless they are located in the SCR. Modifications to the electronic structure near grain boundaries show that charge-induced band-bending at grain boundaries will most likely have a negative effect on device performance, whereas a down-shift in the valence-band energy at the grain surface can effectively passivate the GBs and reduce the effective recombination velocity. For the models considered, GBs generally have a deleterious effect on effciency, and GBs alone can not explain the apparent superiority of polycrystalline over single-crystalline CIGS materials.