Electrical characterization of low temperature pulsed-electron-deposited CU(IN,GA)SE2 solar cells

Abstract

Thin Film Solar Cells (TFSC’s) with Cu(In1-x,Gax)Se2¬ (CIGS)-absorber deposited by means of the novel Low Temperature Pulsed Electron Deposition (LTPED) technique developed at IMEM-CNR [1] were investigated with Admittance Spectroscopy (AS), Capacitance-Voltage profiling (C-V), temperature-dependent current voltage measurements (IV-T) and temperature dependent open-circuit voltage measurements (VOC-T) to obtain information about doping densities, defect characteristics and recombination mechanisms. The novelty of the LTPED technique is the achievement of a single stage deposition at low substrate temperature (250°C as opposed to the >350°C used in other processes) by using a stoichiometric target in the Pulsed Electron Deposition (PED) technique. The characterizations performed during this thesis allowed a more detailed comparison of the behaviour of CIGS-based devices created using LTPED (LTPED devices) and CIGS-based devices produced using more conventional deposition techniques. A methodology was developed to minimize metastable effects often reported for CIGS-based devices [2] [3] [4] [5] from influencing the measurement results, and for each technique the most relevant frequency-, temperature- and bias- were determined. The best performing LTPED devices were found to have doping densities in the range of 2-3∙1015 cm-3, corresponding to depletion layer widths of w~0.5-0.75 μm [6]. Very often lower efficiencies could be attributed to insufficient doping. Although the AS-spectra varied significantly between investigated cells, in nearly each case a high defect density was observed and two separate steps in the capacitance similar to the ‘N1’ and ‘N2’ steps often reported in literature [7] [6] could be distinguished. For the step identified as the N1 step, the activation energy E_d of the trap-level was found to lie in the range of 40 meV ≤E_d≤ 250 meV. From VOC-T and IV-T measurements the activation energy E_a of the dominant recombination mechanism in the temperature range 200 K ≤T≤ 350 K was found to lie close to the absorber bandgap E_G (1000-1200 meV) for nearly all cells, suggesting Shockly-Read-Hall (SRH) recombination in the bulk of the absorber layer. For T< 200 K a lower activation energy was found, most likely indicating increased interface recombination or tunnelling enhanced interface recombination. For a small number of low efficiency cells (η< 10%) values for E_a<E_G were found for T> 200 K, indicating alternative recombination paths such as interface recombination limited the efficiency also at normal operating temperatures. To test the feasibility of creating a homojunction device rather than the presumed P-N heterojunction between CdS and CIGS in standard devices, attempts were performed to deposit a layer of Ordered Defect Compound (ODC) phase of CIGS directly onto the surface of the P-type CIGS absorber using LTPED. Comparison of samples with expected ODC-layer thicknesses of d= 0, 10, 16, 40 and 800 nm only showed a significant loss in performance in the sample with the 800 nm ODC-layer, and an increase in the R_S and R_Sh values in the dark with increasing ODC-layer thickness. The increased R_S and R_Sh values in the dark were found to reduce to similar values as those observed on devices without ODC-layer upon illumination. These effects were attributed to a high photoconductivity in the ODC-layer due to a high concentration of defects, as also proposed in [8]. In this picture, the variations in the R_S values might be explained as a barrier for forward current being formed by the resistive ODC-layer in the dark, which reduces upon illumination as the conductivity of the ODC-layer increases. The variations in the R_Sh values would be explained by the additional ODC-layer covering shunting paths in the CIGS absorber, leading to reduced shunting in the dark. Under illumination, the increased conductivity of the ODC-layer would then explain the large reduction in R_Sh. Accelerated lifetime tests (ALT’s) in damp heat (DH, 85% relative humidity and 85°C) were performed on two LTPED devices with Ga/(Ga+In) ratio (GGI) of 0.375 and 0.3 for 60 and 80 hours respectively. As often reported in literature [9] [10] [11] [10] [12] the clearest effect of DH treatment was the degradation of the Mo and ZnO layers causing an increase in R_S. Corrosion and oxidation of the Mo might also explain the observed losses in EQE at longer wavelengths (λ> 750 nm) as discoloration of the Mo surface might cause a reduction in the amount of light reflected back into the CIGS absorber after reaching the back-contact. The generally observed decrease in R_Sh was attributed to degradation and gradual disappearance of the absorber layer resulting in short-circuits between TCO and Mo back-contact. The observed disappearance of CIGS and formation of ‘craters’ in the absorber layer is expected to be a problem more relevant in LTPED devices, since in depositions by LTPED larger particulates/debris are known to be incorporated into the bulk of the CIGS film, which can eventually cause partial detachment [1]. In separate experiments the relevance of Na doping for formation of an ohmic contact in LTPED devices and the metastability and photoconductivity in CdS layers were investigated. Comparison of Mo/CIGS/Au and Mo/NaF/CIGS/Au samples strongly suggested that NaF can indeed allow a reduction of a back-contact barrier at the CIGS/Mo contact at temperatures for which no MoSe2 is expected to form [13]. The underlying mechanism is thought to be formation of a thin, highly doped P+-layer at the CIGS/Mo interface allowing tunnelling of charge carriers. However, significant non-uniformity of was observed in this effect in the investigated samples. Measurements performed on a ~1 μm layer of CdS showed occurrence of photoconductivity with slow transients. This suggested a significant concentration of deep defects in the CdS buffer layers used in LTPED devices, which could have an important role in the metastability observed in LTPED devices