Effects of material properties of band-gap-graded Cu(In,Ga)Se 2 thin films on the onset of the quantum efficiency spectra of corresponding solar cells

Polycrystalline Cu(In,Ga)Se 2 (CIGSe) thin-film solar cells exhibit gradual onset in their external quantum efficiency ( EQE ) spectra whose shape can be affected by various CIGSe material properties. Apart from influences on the charge-carrier collection, a broadening of the EQE onset leads to enhanced radiative losses in open-circuit voltage ( V oc ). In the present work, Gaussian broadening of parameters describing the EQE onset of thin-film solar cells, represented by the standard deviation, σ total , was evaluated to study the impacts of the effective band-gap energy, the electron diffusion length, and the Ga/In gradient in the CIGSe absorber. It is shown that σ total can be disentangled into contributions of these material properties, in addition to a residual component σ residual . Effectively, σ total depends only on a contribution related to the Ga/In gradient as well as on σ residual . The present work highlights the connection of this compositional gradient, the microstructure in the polycrystalline CIGSe absorber, and the luminescence emission with the residual component σ residual . It is demonstrated that a flat band-gap with no compositional gradient in the bulk of the CIGSe absorber is essential to obtain the lowest σ total values and thus result in lower recombination losses and gains in V oc .


| INTRODUCTION
Cu(In,Ga)Se 2 (CIGSe) thin-film solar cells (front contact/CIGSe/Mo/ glass stacks) with record conversion efficiencies of more than 23% 1 exhibit Ga/In gradients with higher [Ga] towards the back and higher [In] towards the front contact.Since the band-gap energy (E g ) of CuInSe 2 (1.04 eV) 2 is smaller than that of CuGaSe 2 (1.68 eV), the varying [Ga]/([Ga] + [In]) (GGI) ratio perpendicular to the substrate leads to a variation of the local E g in the solid solution CIGSe. 2 The GGI gradients in CIGSe layers for high-efficiency devices (>20%) exhibit V-shapes with a local minimum of [Ga] and thus of E g at a position between front and back contact.This position (on an axis perpendicular to the substrate) is what determines the onset of the light absorption as a function of the photon energy, and the minimum or effective E g of a graded CIGSe thin film can be obtained from the onsets in absorption or in external quantum efficiency (EQE) spectra.
While the evaluation of absorption spectra offers the advantage of the absence of any transport-related effects (which need to be taken into account in EQE spectra), such measurements cannot be performed easily on completed solar cells owing to the opaque Mo back contact.Since CIGSe layers on transparent substrates need not to exhibit the identical materials properties as those in completed solar cells with Mo/glass substrates, it is not possible to correlate directly the absorption behavior of a CIGSe layer on a transparent substrate with the device performance of the front contact/CIGSe/ Mo/glass solar cell.Therefore, the absorption behaviors of CIGSe layers were investigated by the evaluation of EQE spectra of the corresponding solar cells with Mo back contacts in the present work.
Onsets of the EQE spectra are not only affected by charge-carrier transport. 3Other contributing factors are the CIGSe film thickness, as well as E g fluctuations in the CIGSe alloy. 4The shape of the EQE onset has an impact on the V oc via radiative losses and on the collection. 5wever, it is unclear whether E g fluctuations can be determined easily by evaluation of the onset in EQE spectra for solar cells with various CIGSe layer thicknesses (in case transport effects can be neglected).We note that in the following, the CIGSe compound with x = GGI exhibits a band-gap energy E g (x) = (1-x) E g (CuInSe 2 ) + x E g (CuGaSe 2 ) -bx (1-x), where b = 0.2 is the bowing parameter. 6 the present work, the onset of the EQE spectra of CIGSe solar cells with varying effective E g and Ga gradients was evaluated by fitting Gaussians to the first derivatives of the EQE spectra, which provides the effective E g values from the mean and σ total from the standard deviations of the Gaussians. 7,8We will show that the influence of the CIGSe layer thickness, low diffusion length and Ga gradient on the shape of the EQE onset is substantial.In order to facilitate optimal carrier collection and reduced radiative V oc losses, compositional gradients in CIGSe layers should be restricted to the regions close to the interfaces to the front and back contacts.

| EVALUATION OF THE ONSET IN EQE SPECTRA
The onset of an EQE spectrum can be evaluated in various ways.One possible and established approach is to extract the Urbach energy (E u ) from the subgap region of the internal QE (IQE) spectrum, which needs the input of the EQE and reflectance R spectra, since IQE(E) = EQE(E) / (1-R(E)) (E is the photon energy). 9The E u is then given as the inverse of the slope of the subgap region in the IQE spectra (i.e., ln (IQE(E)) = c + E/E u , where c is a constant). 9t another way is to evaluate not EQE(E) but its first derivative dEQE/dE to obtain a E g distribution as proposed by Rau et al. 7 followed by a Gaussian fitting of the E g distribution. 8,10A Gaussian (Figure 1, green curve) is fitted to the function dEQE/dE (Figure 1, red curve) around its local maximum.According to Rau et al., the mean of the Gaussian gives the effective band-gap energy E g of the solar absorber, and the standard deviation the amplitude of band-gap fluctuations (σ g ), when electrostatic potential fluctuations can be neglected. 4,8,11It is visible in Figure 1 that the Gaussian fit deviates substantially from the experimental dEQE/dE below the full width at half maximum (FWHM), that is, for dEQE/dE <0.3 (normalized).Nevertheless, the fit agrees well with the experimental data for dEQE/dE >0.3 (normalized), at the center of the peak and FWHM which is where the effective (E g ) and the standard deviation (σ total ) are extracted.
][14][15] It will be an essential part of the present work to disentangle the standard deviation σ total (determined by the Gaussian fit) to obtain its contributions by the diffusion length (σ diff ), the (effective) absorber thickness (σ thick ), the E g gradient (σ grad ), and a residual component (σ residual ).
A straightforward approach would be to consider these individual quantities as independent contributions to σ total : However, in a real CIGSe solar cell, these contributions are not independent from one another.Indeed, the Ga/In gradient affects considerably the effectively lightabsorbing thickness of the CIGSe layer.Also, the Ga/In gradient may influence the diffusion length of the charge carriers, for example, in case it exhibits a decreasing GGI from the Mo/CIGSe interface towards the CIGSe/buffer interface (since then, electrons are repelled from the back contact).At any rate, when regarding CIGSe solar cells with absorbers free of any band-gap gradient, it becomes clear that the impact of σ diff and σ thick on σ total can be neglected (see supporting information S1 and S2).Since the present work deals only with graded CIGSe absorbers, σ grad contains contributions also of σ diff and σ thick , and σ total can be expressed via: Experimental external quantum efficiency (EQE) spectrum (black) for one of the solar cells studied in the present work.Gaussian fitting (green) of the first derivative of EQE (red) (normalized to 1) gives an effective band-gap energy E g = 1.19 eV as mean and σ total of 49 meV as standard deviation It is important to note that an experimental EQE spectra exhibit influences of the subgap absorption, the diffusion length, and the effective thickness of the absorber, leading to a Gaussian-type dEQE/ dE curve with standard deviation σ total .In contrast, when simulating the effects of only the E g gradient on the EQE onset, no Gaussian-type distribution dEQE/dE is obtained.A corresponding Gaussian fit for the determination of the standard deviation seems to be inappropriate at first glance.However, in the present work, we will simply use the Gaussian fit as a tool to quantify the broadening of the dEQE/dE curve and to be able to determine eventually the residual broadening σ residual in Equation ( 1).
This residual broadening will be related (mainly) to local strain and/or compositional variations in the polycrystalline CIGSe layer.
These variations are present at various length scales (0.1 nm to 1 μm 4 ) and can, unfortunately, not always be assessed easily via microscopic analyses; however, they were detected and quantified at grain boundaries and dislocations in polycrystalline CIGSe. 5,16What we assume is that the local strain and/or compositional variations induce corresponding changes in the electronic structure, that is, in the conduction-band and valence-band edges, and thus lead to local variations of E g , that is, to E g fluctuations. 17r all the solar cells studied in the present work, the EQE spectra feature subgap absorption visible as band tails (at E < E g ) at the EQE onset.The Urbach energies (E u ) were determined from all experimental EQE spectra in addition to the broadening σ total to establish a quantitative representation of this subgap absorption (Table 1).In contrast to E g fluctuations, Urbach tails may originate from phonon interactions and optical transitions between defect and tail states. 18 visible in Figure S3, the inclusion of Urbach tails leads to a flattening of the EQE onset and a corresponding broadening of the dEQE/dE distribution.In contrast to the experimental EQE spectrum, fitting a Gaussian to a simulated dEQE/dE distribution has the only purpose to determine the standard deviation (and not the effective E g ).As visible from Figure S3, the Gaussian fit to dEQE/dE is not good; nevertheless, it allows for extracting a broadening via the standard deviation.If this broadening were always included in the simulated EQE spectra, the determination of the individual impact of a CIGSe material property (E g gradient, film thickness, and diffusion length) would be very complicated, and it would also cause conflict with the concept of disentanglement (Equation ( 1)).Therefore, we assume that the contribution of Urbach energy to the broadening is present in the component σ residual and do not include any Urbach tails in simulated EQE spectra.
Apart from affecting the integral absorption of the incident sun light and thus the carrier collection of the solar cell, σ total is detrimental to the device performance by leading to a decreased open-circuit voltage (V oc ). 4,5 oc ¼ σ Here, k B is Boltzmann constant, e is the elemental charge, and T is the absolute temperature of the solar cell.Note that Equation (2) provides the V oc deficit due to radiative losses only.The gradient and the possible origins of σ residual also lead to nonradiative recombination via defect states resulting in additional V oc losses.

| MATERIALS AND METHODS
Solar cells fabricated at both ZSW, Stuttgart, Germany, and PVcomB, Berlin, Germany, were investigated.All photovoltaic parameters of the nine investigated cells are summarized in Table S4.At PVcomB, CIGSe layers were fabricated on Mo-coated soda lime glass (SLG) substrates following a three-stage-type thermal co-evaporation process as described in Heinemann et al. 19 The range of substrate temperatures (T s ) was 500-650 C. The device had no antireflection coating.
Other PVcomB absorbers were prepared by a sequential process using sputtered metal precursor layers, which were annealed in a nitrogen atmospheric pressure conveyor oven manufactured by Smit Thermal Solutions first, in Se vapor, and then in H 2 S at 580-600 C. 20 At ZSW, CIGSe absorbers were deposited using an in-line multistage co-evaporation process on Mo-coated SLG substrates with an Note: The GGI and CGI ratios for all the cells were calculated from the elemental concentrations measured by XRF.Solar cells having the same effective band-gap energy E g in the CIGSe absorbers exhibited different standard deviations of dEQE/dE, σ total .The V oc loss was calculated using Equation 2, and E u as described in Section 2.
industrially relevant 30 x 30 cm 2 coater. 21No additional alkali PDT process was used, and Na (partially K) was only provided from SLG during CIGS deposition at elevated temperatures.Integral GGI ratios of 0. were analyzed by means of glow-discharge optical emission spectroscopy (GDOES) 22 and energy dispersive X-ray spectroscopy (EDX).
EDX mapping was performed under an acceleration voltage between 7 and 10 kV, and the data were evaluated using Oxford Instruments Aztec software packages.(We do not show the EDX data in the present work since they provide the same elemental distributions as the GDOES results.)The average grain size distribution was determined by electron backscatter diffraction (EBSD).Also, the lateral emission fluctuation on micrometer scale was investigated via cathodoluminescence (CL) spectroscopy. 23For EQE simulations (λ = 300-1400 nm) via SCAPS 24 (see Table S5 for simulation details), the absorption coefficient (α) in the absorption model was calculated considering the case of a direct gap semiconductor where Here, A is a material-dependent parameter calculated for every cell using their corresponding, integral [Cu]/([Ga] + [In]) (CGI) and GGI values. 25For the device simulations via SCAPS, we varied the electron diffusion length (via the electron mobilities, the effective defect density, and the electron lifetimes in the CIGSe layer) or the absorber layer thickness or the GGI gradient perpendicular to the substrate (via the E g gradient and that of the electron affinity, calculated from the gradient in GGI) as input parameters.

| E g and σ total from experimental EQE for graded CIGSe absorbers
As a first step, the approach described in Section 2 (Figure 1) was applied to the EQE spectra acquired from 50 different CIGSe solar cells.We note that all the studied CIGSe absorber layers in these cells exhibited Ga/In gradients perpendicular to the substrate.In Figure 2, the extracted σ total was plotted versus effective E g (here, E g is obtained from the mean of the Gaussian used to fit the E g distribution and this corresponds with the local minimum of the E g gradient).
In order to study systematically the influence of various CIGSe materials properties on σ total , a series of nine CIGSe solar cells was analyzed in more detail.The effective E g , the integral GGI and CGI ratios of the CIGSe absorber layer, as well as the extracted σ total (Section 2) and the corresponding V oc losses (Equation ( 2)) are given in Table 1.We note that the two cells with the largest σ total , cell nos. 5 and 8, exhibit the lowest CGI ratios (0.85 and 0.84); as we will see further below (Table 2), these are also the two cells with the largest σ residual values.Earlier work by Stephan et al. 26,27 suggests that the density of point defects is higher the smaller the CGI ratio.
Nevertheless, a clear dependency of σ total as a function of E g cannot be detected; as already outlined in Section 2, various materials properties contribute to σ total , of which E g may be one.Notably, one solar cell with the smallest σ total of 17 meV and E g = 1.03 eV particularly stood out; this cell was further investigated in detail in the present work to reveal what the material properties of the CIGSe absorber layer are for such a low σ total .

| Effect of the grading in CIGSe thin films
To investigate the effect of the E g gradient (or Ga/In gradient) in CIGSe on the EQE onset (and thus, on the broadening of E g distribution perpendicular to the substrate), again, SCAPS simulations were performed.We incorporated the experimental E g gradients that are given in Figures 3 and S6.Every solar cell had a unique absorber thickness (Table 2), which was used in the SCAPS simulations.
To obtain a parameter representing the pattern of grading at the location of minimum E g corresponding to the onset in the EQE spectrum, the curvature at the notch point was determined for all E g F I G U R E 2 σ total vs E g from the approach depicted in Figure 1, showing no specific dependency of σ total on E g .One solar cell (cell no. 2 in Table 1) exhibits the smallest σ total of 17 meV with effective E g of 1.03 eV.The experimental errors in the extracted σ total and E g values are about 2 and 10 meV (about the size of the symbols) gradients.For a curve f(x) = E g (x) with x a position on an axis perpendicular to the substrate, the curvature at the local minimum of the E g gradient is quantified by the second derivative d 2 f(x)/dx 2 . 28e gradient as such has a considerable effect on the shape of the EQE and thus contributes correspondingly to σ total .Cell nos.8 and 9 have identical E g ; however, the shapes of their E g gradients F I G U R E 3 (A, D, G) E g gradients calculated from glow-discharge optical emission spectroscopy (GDOES) depth profiles for solar cell nos.9, 8, and 2. Also given are the corresponding values for E g and the curvature.(B, E, H) Simulated external quantum efficiency (EQE) and (C, F, I) Gaussian fitted dEQE/dE for the cell nos.9, 8, and 2 by using each of their original GGI gradient to model the CIGSe absorber.There is no broadening of the E g distribution for homogeneous CIGSe layers; however, when a gradient is incorporated, there is a broadening of the dEQE/dE for all the cells except for cell no.2 (with a rather flat gradient) are different.As shown in Figure 3D, cell no.8 exhibits a stronger gradient and thus a larger broadening (Figure 3F) (σ grad = 21 meV) when compared with cell no. 9 featuring a gradient with relatively smaller curvature (Figure 3A).For all the nine solar cells, σ grad values were determined (see Table 2) by fitting Gaussians to the dEQE/dE dependencies (see Figure S6 for the remaining six cells).Cell no. 2 featuring a flat gradient with the lowest curvature has the smallest σ grad of the order of 0.001 meV (σ grad $ 0 meV) (Figure 3I).
The effect of the gradient on the EQE onset was further confirmed by modeling CIGSe layers of the same thickness and E g but having gradients of various curvatures ranging from 0 to 4 eV/μm 2 (see supporting information S7).
Cell no. 2 with the smallest σ total of 17 meV exhibits a flat E g gradient with curvature 0.3 eV/μm 2 , that is, the minimum E g was almost constant throughout a major part of the CIGSe absorber (Figure 3G).

| Assessing the residual component σ residual
Using the information of Section 4.2, we can now use Equation ( 1) and determine σ residual .Table 2 lists the residual components σ residual for these cells, together with the values for σ total , the curvature, and σ grad .The σ grad values remain between 0 (cell no.It is important to note that a E g gradient affects the broadening of the EQE onset in various ways.First, it contributes to σ grad .In addition, a E g gradient is always linked to microstructural and compositional variations perpendicular to the substrate that have a share in In Table 2, there are solar cells that exhibit identical minimum  This particular solar cell exhibits the smallest σ residual of 17 meV and a flat E g gradient (apart from the one at the back) in the CIGSe absorber (curvature 0.3 eV/μm 2 ; see Figure 3G).As visible in Figure 4B, the CIGSe absorber exhibits large grains in the top region of the absorber, where the E g (i.e., Ga/In) gradient remains flat and small grains only close to the Mo back contact, where the GGI ratio becomes much larger (the average grain size, d grain , is about 0.5 μm including all grains in the absorber).We note that the dependency of the average grain size on the GGI ratio in the CIGSe layer was investigated in an earlier work. 29The amplitude of the lateral fluctuations in the emission spectra of this cell was investigated via hyperspectral CL imaging with a pixel size of 50 nm.(Figure 4C).51 pixels (see also Table S10).From the CL peak energy distribution map (Figure 4C), it is apparent that the peak in the CL emission varies with the depth (owing to the Ga/In gradients) and also laterally.The CL emission occurs at higher energies towards the back contact because of the higher GGI ratio.
The lateral and perpendicular fluctuations of the CL emission exhibited very similar amplitudes σ CL,lateral and σ CL,perp of about 10 and 9 meV (see Table 3).This means that with hardly any compositional gradient present in the absorber, fluctuations with small amplitudes remain which seem to be isotropic in the CIGSe thin film and lead to only small, radiative V oc losses.E g .However, the σ residual value of cell no.8 (44 meV) is higher by 14 meV than that of cell no. 9 (31 meV).This difference in σ residual cannot be explained well via the slight differences in the lateral fluctuation of CL emission (σ CL is 15 meV vs. 12 meV; see Figure 5C,F).In contrast, the average grain size of the CIGSe layer in cell no.8 is 0.2 μm, while in cell no.9, it is 0.4 μm.This difference is probably also linked to the difference in curvature values (10 and 3 eV/μm 2 ) and elemental composition.We can assume considerable microstrain induced by the Ga/In gradients being present perpendicular to the substrate, as well as slight compositional variations and additional microstrain in the lateral directions, which is different for the CIGSe layers in cell nos.8 and 9.

| DISCUSSIONS
The evaluation of the EQE spectrum of a solar cell provides valuable insight into the limitations of the device performance.Apart from the direct link between the EQE spectrum and the short-circuit current density j sc , the broadening of the EQE onset can be related to a radiative V oc loss assessed by quantifying the standard deviation σ total of the dEQE/dE dependency.As apparent from the results further above, influences of the absorber thickness and the electron diffusion length on σ total can be neglected.In contrast, the E g gradient (perpendicular to the substrate) in the CIGSe layer as well as a residual component, attributed (at least in part) to E g fluctuations, exhibit the two quantities contributing to σ total .
It is convenient to divide the total broadening σ total into its components σ grad and σ residual (Equation 1).The contribution of σ thick and σ diff can be neglected for nongraded absorbers with thicknesses above 1 μm and diffusion lengths larger than 1.5 μm.We found that σ grad is substantial in case of large curvature values (>1 eV/μm 2 ), but that σ residual is always larger than σ grad .
T A B L E 3 Summary correlating the grain size, the lateral fluctuations in CL emission, and the curvature with the σ residual and radiative V oc loss of selected cells  As for the CL emission, the amplitudes of fluctuations perpendicular to the substrate (σ CL,perp ) are always larger than those in lateral directions (σ CL,lateral ); see Tables 3 and S10.We can attribute this fact at least in part to the much larger compositional changes perpendicular to the substrate mainly due to the Ga/In gradient.
It is interesting to compare the curvature, the average grain size, and the amplitudes of fluctuations in CL emission of a CIGSe layer with the σ residual value of the corresponding solar cell (Table 3).As visible from this comparison, σ residual is increased with increasing curvature and with decreasing average grain size.Since the average grain size/the microstructure in general depends substantially on the Ga/In gradient and thus on the curvature, we can attribute a major contribution to σ residual to local variations in composition and to (micro)strain perpendicular to the substrate, hence, by features which can be regarded as origins of E g fluctuations.
In contrast, the σ CL,lateral values are very similar to each other; it is unclear whether the given differences (10-15 meV) are really significant.Within the scope of the present work, it was not possible to detect any significant variations in chemical composition in the lateral directions via EDX mapping; we were not able to distinguish unambiguously between possible compositional variations and statistical fluctuations in the elemental distribution maps.
A E g -graded CIGSe absorber exhibits a gradual EQE onset resulting in a radiative V oc loss, determined by the total broadening σ total (containing contributions from σ grad and σ residual ).In addition, the device performance can also be affected by the V oc losses due to nonradiative recombination influenced strongly by the E g gradients in the CIGSe layer perpendicular to the substrate, since they affect the microstructure and hence the density of grain boundaries in the polycrystalline absorber, which act as centers of enhanced (nonradiative) recombination. 30Overall, the V oc deficit due to radiative recombination is quantified by σ total (containing σ grad and σ residual ) and not only by σ residual .
At this point, we would like to note that Ga/In gradients in the CIGSe layer in the regions close to the front and back contacts are known to be beneficial with respect to reducing interface recombination at these interfaces. 31However, we emphasize that a flat E g gradient (with small curvature) spanning across a major depth of the CIGSe layer 32 leads not only to enhanced absorption and collection but also to reduced broadening σ total , thus smaller radiative V oc losses.

| CONCLUSION
In the present work, we investigated the effect of various material 30   and 0.66 as confirmed by X-ray fluorescence were realized by adapting the Ga and In evaporation rates on different substrate carriers in the same deposition run.Subsequently, the CdS buffer layer was grown on the CIGS absorbers by chemical bath deposition with a thiourea-based process.Radio-frequency-sputtered i-ZnO was deposited as highly resistive (HR) layer, and direct-current-sputtered ZnO:Al was used as front contact.Solar cells with a total area of 0.5 cm 2 were completed with Ni/Al/Ni grid fingers on top without antireflective coating.EQE measurements were performed under light bias at zero and À1 V bias voltage over a wavelength range of 300 to 1400 nm, with silicon and indium gallium arsenide solar cells as references.The elemental distributions perpendicular to the substrate in the CIGSe layers

2 (
2) and 23 meV.The residual component σ residual exhibits much larger values (25-50 meV) for all cells with the exception of cell no.17 meV).
E g and different σ residual values and again other cells with the same σ residual but different E g .In order to shed more light on how the CIGSe materials properties may affect σ residual , we present case studies in the following Section 4.4, investigating selected solar cells (cell nos.2, 8, and 9) more in detail by means of EBSD and CL (note that corresponding results from cell nos.4 and 6 are provided in Figures S8 and S9).

4. 4 |
Case studies correlating the microstructure, the curvature, and the lateral fluctuations in luminescence emissions to the σ residual of selected cells 4.4.1 | Cell no.2: σ residual = 17 meV, E g = 1.03 eV The fluctuations of the CL emission were analyzed parallel to the substrate at the positions of the notch in the E g gradients by obtaining a standard deviation of the emission energy across 192 pixels.The minimum amplitude of fluctuations in the CL emission perpendicular to the substrate (σ CL,perp ) was also extracted by taking a standard deviation of the emission energy across F I G U R E 4 (A) Cross-section SEM image of the solar cell stack of cell no. 2. (B) Electron backscatter diffraction (EBSD) pattern-quality map from the same position.Larger grains are found in the top part of the CIGSe layer because of the flat Ga/in gradient, smaller grains towards the Mo back contact due to higher GGI.(C) Cathodoluminescence (CL) peak energy distribution map of the same position as in (A) and (B)

F
I G U R E 5 (A, D) SEM images of the investigated cross sections of cell nos.8 (A) and 9 (D).(B, E) Electron backscatter diffraction (EBSD) pattern-quality maps from the same areas as in (A) and (D).(C, F) Cathodoluminescence (CL) peak energy distribution maps from the same areas as in (A) and (D).Pixel size = 50 nm properties of CIGSe absorbers on the onsets of EQE spectra of corresponding solar cells.Compositional gradients perpendicular to the substrate in the CIGSe thin films were identified as substantial contributors to the broadening of these EQE onsets, leading to radiative V oc losses.Flat E g gradients across a major depth of the CIGSe layer yield steep EQE onsets and thus smaller current and voltage losses in the corresponding solar cells.The present work gives hints of how to change the materials design of CIGSe absorbers for power conversion efficiencies of 25% and beyond.
List of all solar cells investigated in more detail Calculated curvatures, σ grad and residual component σ residual for all investigated solar cells with different E g and total CIGSe thickness T A B L E 2 4.4.2| CIGSe solar cells with similar effective E g but different σ residual (cell no.8 with σ residual = 44 meV, E g = 1.19 eV and cell no.9 with σ residual = 31 meV, E g = 1.20 eV) Cell nos.8 and 9 contain CIGSe layers with nearly the same (effective)