Assessment of transparent conductive oxides as back contacts for inline-fabricated Cu(In,Ga)Se2 solar cells

J. Phys. Energy 7(2025) 045018 https://doi.org/10.1088/2515-7655/adfd86 Journal of Physics: Energy OPEN ACCESS RECEIVED 31 March 2025 REVISED 11 June 2025 ACCEPTED FOR PUBLICATION 20 August 2025 PUBLISHED 16 September 2025 Original content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. PAPER Assessment of transparent conductive oxides as back contacts for inline-fabricated Cu(In,Ga)Se2solar cells Rico Gutzler1,∗, Chang-Yun Song2, Dimitrios Hariskos1, Heiko Kempa2, Roland Scheer2, Roland Wuerz1, Erik Ahlswede1, Stefan Paetel1and Wolfram Witte1 1Zentrum für Sonnenenergieund Wasserstoff-Forschung Baden-Württemberg (ZSW), Meitnerstr. 1, Stuttgart 70563, Germany 2Martin-Luther-Universität Halle-Wittenberg, Institut für Physik, von-Danckelmann-Platz 3, Halle (Saale) 06120, Germany ∗Author to whom any correspondence should be addressed. E-mail: [email protected] Keywords: Cu(In, Ga)Se2, thin-film photovoltaics, transparent back contact, bifacial solar cell, inline fabrication Abstract The integration of transparent back contacts (TBCs) into Cu(In,Ga)Se2(CIGS) and (Ag,Cu)(In,Ga)Se2thin-film solar cells presents significant opportunities for enhancing device performance and enabling applications such as bifacial and tandem solar cells. However, the implementation of these TBCs in industry-relevant inline deposition processes poses several technological challenges, in particular regarding their thermal and chemical stability during the CIGS process at elevated temperatures while at the same time maintaining high electrical conductivity and high optical transparency. We investigate various transparent conductive oxides as TBC candidates and the use of a thin Mo protection layer to address these challenges, focusing on optimizing CIGS process parameters to ensure seamless integration within existing industrial frameworks. Our results demonstrate that under appropriate CIGS growth conditions and TBC selection, TBCs with high transparency and low free carrier absorption in the near-infrared wavelength region can be successfully incorporated into an inline CIGS deposition process. We establish a laser-scribing process for cell definition that replaces mechanical scribing, which often showed poor cell definition, and photolithography, which is a slow multi-step process. Indium tin oxide requires a thin Mo protection layer to remain functional after CIGS deposition. Zirconium-doped indium oxide (IZrO), however, does not require a protection/sacrificial layer and remains optically transparent and conductive, withstanding our CIGS deposition conditions under which other TBCs already degrade. For CIGS cells with a band gap of 1.1 eV, 14.2% efficiency and high back contact transparency can be demonstrated with IZrO. Our analysis paves the way for the industrial implementation of advanced photovoltaic CIGS devices with improved efficiency and new functionalities. 1. Introduction A requirement for both photovoltaic (PV) top cells in a tandem device and bifacial solar cells is electrically conductive contacts that are optically transparent. Whereas the integration of transparent back contacts (TBCs) instead of routinely employing opaque molybdenum (Mo) into Cu(In,Ga)Se2(CIGS) and (Ag,Cu)(In,Ga)Se2(ACIGS) thin-film solar cells has been shown to work well on the laboratory scale [1–4], the implementation of TBCs in scalable inline deposition processes has received no attention so far. Numerous materials have been explored in lab-scale systems. Nakada and coworkers investigated indium tin oxide (ITO) and SnO2:F (FTO) as TBCs for CIGS solar cells already in 2004, achieving 13.7% efficiency on FTO and 13.0% on ITO, both nearly identical to reference cells with Mo back contact [5]. At the same time, Young and coworkers presented a four-terminal CuGaSe2/Cu(In,Ga)Se2tandem device using FTO as transparent back electrodes [6], and shortly after ITO was reported in a mechanically stacked tandem [7]. Current record cells use ITO as back contact, achieving 19.77% efficiency with front illumination and © 2025 The Author(s). Published by IOP Publishing Ltd J. Phys. Energy 7(2025) 045018 R Gutzler et al 10.89% with rear illumination [1]. Hydrogen-doped In2O3was investigated by Keller et al for both CIGS [8] and ACIGS [4] solar cells, as well as wide-gap (A)CIGS [2] absorbers, which offers the advantage of low near-infrared absorption compared to ITO or FTO. The same group used tungsten-doped In2O3to fabricate wide-gap ACIGS cells with 13.6% efficiency, likewise with low near-infrared absorption [9]. Wide-gap CIGS cells were also fabricated on FTO with the help of a thin Mo layer in between TBC and absorber [10,11]. ITO was also shown to yield good backside efficiency with wide-gap Cu(In,Ga)S2[12]. Further oxides explored as TBC include Al-doped ZnO, with reports that a thin MoSe2layer is required to form an Ohmic contact [13, 14], whereas for other groups an Ohmic contact is achievable without interfacial Mo layer [15]. ITO, hydrogen-doped In2O3, and zirconium-doped In2O3were recently investigated regarding their suitability for bifacial applications, with the latter two showing better performance under backside illumination [16]. The observation of GaOxformation at the CIGS/TBC interface is often reported and its negative effect on the cell’s fill factor is discussed in view of its n-type conductivity and the resulting counter diode at the back contact [17,18]. The presence of Na during absorber growth is reported to enhance GaOxformation and can be used to control its thickness [4,8]. GaOxformation can be limited by inserting a thin Mo layer between TBC and CIGS prior to absorber growth [13,19], which is oxidized to MoSe2[13,17] or MoSe2and MoOx [20] during absorber growth at elevated temperatures under selenium atmosphere. The presence of an Mo or MoSe2interlayer, however, negatively impacts transparency of the device [21] and its thickness needs to be carefully controlled. Next to Mo, other materials were also investigated for interface modification on TCOs, such as a ∼10 nm thick MoO3layer on ITO, which improves both open-circuit voltage (VOC) and fill factor (FF) [22], in spite of the formation of a thin GaOxlayer. A 6 nm thin WOxlayer was used as hole extraction layer on ITO, resulting in improved short-circuit current density (JSC) and FF [23]. In order to thermodynamically limit the formation of GaOx, sulfur was deposited on ITO in superstrate [24] and substrate applications [25], in both cases leading to a gain in VOC. A common observation is that none of the intermediate layers Mo, MoO3, WOx, or S completely inhibit the formation of GaOxdue to its comparatively low standard molar enthalpy of formation of −1089 kJ mol−1[26]. Depending on the choice of TBC, process temperatures below 550 ◦C are used to preserve the TBC’s properties and to limit the formation of GaOx, with a few exceptions reporting higher CIGS growth temperatures. Fonoll-Rubio and coworkers assume that the selenization of an ITO back contact forms an amorphous In-Se phase, which degrades the optical transparency of the TBC and the device’s optoelectronic properties, in particular at high process temperatures >540 ◦C [21]. Industrial pilot lines fabricating CIGS modules can operate at higher temperatures up to 650 ◦C [27] and also the inline coevaporation process used in the present study routinely requires substrate temperatures above 550 ◦C [28]. The aim of this work is to bridge the gap between the fabrication of CIGS devices on TBCs with small static laboratory coaters at comparatively low temperatures and the high-temperature inline processes often used in industrial settings. Several technological challenges, such as a high thermal budget that interrelates with the TBC’s thermal and chemical stability during the CIGS coevaporation process at elevated temperatures under a selenium atmosphere, need to be addressed to preserve high electrical conductivity and optical transparency under these harsh conditions. Our assessment identifies transparent conductive oxides that fulfill these requirements. We also discuss the adjustment of the ACIGS process parameters, such as lowering the overall thermal budget, while ensuring seamless integration within existing industrial frameworks. Finally, the performance of semitransparent cells is discussed in view of different absorber thicknesses. 2. Methods 2.1. CIGS deposition and cell fabrication The CIGS solar cells were produced through a standardized workflow. The CIGS absorbers were co-evaporated in an industry-oriented inline coater designed for substrates up to 30 ×30 cm2[28]. As back contacts, different TCOs were used as detailed below. In all deposition campaigns, reference samples with a ∼550 nm thick opaque Mo back contact were fabricated. The substrate heater temperature was systematically varied in the third stage of the inline growth process. The absorber’s composition was controlled by adjusting evaporator temperatures while keeping the growth time constant. An RbF post-deposition treatment was applied to all absorbers. For CIGS cells, the [Ga]/([Ga]+[In]) ratio (GGI) was 0.32. The CIGS absorber had thicknesses of 2.2 µm (referred to as 'standard'in the following), and a 530 nm thin absorber was fabricated with four times faster inline speed without changing the rates of the evaporation sources. A chemical bath deposition (CBD) method was used to grow ∼50 nm of CdS. Subsequently, a Zn0.85Mg0.15O (ZMO) layer was RF-sputtered on top of the CdS layer followed by DC-sputtered Al-doped ZnO (AZO) as the front contact. All processes, except for the CBD buffer, were 2 J. Phys. Energy 7(2025) 045018 R Gutzler et al inline. A Ni/Al/Ni grid was deposited by electron-beam evaporation. Cells on TBCs were laser-scribed with a total cell area ∼0.48 cm2. No antireflective coating was used. 2.2. Transparent back contacts ITO was purchased from Solems S.A. with 370 nm thickness and 8–12 Ω/□sheet resistance (Rsq) (ITO Sol 12, 10 wt% SnO2in In2O3, corresponding to 17 mol% SnO2in In2O3, molecular formula In2(1−x)SnxO3−x with x=0.17). A selection of ITO samples was sputter-coated with a ∼10 nm thick Mo sacrificial layer before absorber growth. FTO with a thickness of 1 µm and ∼13 Ω/□sheet resistance was sourced from Calyxo GmbH. IZO and IZrO were sputter-deposited in-house on 3 mm soda-lime glass. About 220 nm thick IZrO layers were RF sputtered at room temperature from an In2O3target with 2 wt% ZrO2(corresponding to 4.4 mol% ZrO2in In2O3, molecular formula In2(1−x)ZrxO3−xwith x=0.044), yielding Rsq ⪆20 Ω/□(an Ar/O2/H2-gas mixture was used during sputtering). IZO was RF sputtered from an In2O3with 7 wt% ZnO (corresponding to 20.4 mol% ZnO in In2O3, In2(1−x)ZnxO3−2xwith x=0.204) target at room temperature for layers with ∼320 nm thickness and Rsq 10 Ω/□. 2.3. Laser-scribing For cell area definition, the AZO front contact on the CIGS layer was laser-scribed using a microFLEX system by 3D-Micromac with a femtosecond (fs) laser with a wavelength of 515 nm and a scan velocity of 1500 mm s−1at a frequency of 100 kHz. The pulse-to-pulse distance was 15 µm. A comparison of mechanically scribed cells to laser-scribed cells on one sample with opaque Mo back contact revealed that the laser scribed cells reached 95% of the cell efficiency of the mechanically-scribed cells. The main loss was observed in the FF. 2.4. Current–density voltage (JV) measurement JV was measured under one sun AM1.5G at 25 ◦C using a WACOM Super Solar Simulator WXS-90 S-L2 (grade AAA). A Si-reference cell was used for calibration. The bifacial JV measurements were performed in a setup of two Wavelabs Sinus 70 LED solar simulators, which were mounted opposite to each other with the sample on a glass plate between them, so that illumination from the front and from the rear side can be adjusted independently. Significant deviations of the sample temperature were avoided by limiting the illumination to the duration of the measurement. Cell JSC and efficiency is reported on total cell area. 2.5. External quantum efficiency (EQE) measurement EQE measurements were conducted using a Bentham PVE300 system, with a Xenon short-arc lamp and/or a Quartz Halogen lamp as light sources. A TMC300-FU monochromator was used to select the wavelength, and measurements were performed over the 250–1250 nm range with 5 nm or 10 nm step size. Data acquisition and analysis were carried out using BenWin+software. 3. Results and discussion Several TCOs were tested as back contacts during standard growth conditions for a 2.2 µm thick CIGS layer with an integral GGI of 0.32 and deposited at a reduced substrate heater temperature of 580 ◦C maximum, roughly 100 ◦C below the standard temperature for opaque Mo back contacts. The tested TBCs include bare ITO, ITO with a ∼10 nm thick Mo sacrificial layer, FTO, IZO, and IZrO. For this deposition at reduced substrate heater temperature, reference cells with an opaque Mo back contact achieve an efficiency of 18.0%. Transmittance measurements of the TBCs before absorber deposition are shown in figure 1(a) and after absorber deposition in figure 1(b), with the absorber being mechanically removed after absorber deposition. Even at the reduced substrate temperature, the transmittance of ITO (orange), IZO (light blue), and FTO (red) is significantly reduced in the low-wavelength part. The strongest degradation over the whole wavelength range is observed for FTO. A thin Mo layer deposited on ITO, which before CIGS deposition significantly suppressed transmittance (gray), is converted to MoSexand MoOxduring the growth process [20] and becomes more transparent due to the larger band gaps of these materials. IZrO (dark blue) shows a slight gain in transmittance around 350 nm but reduced transmittance for other wavelengths. A possible degradation mechanism may be due to the formation of an amorphous In–Se phase, as reported by Fonoll-Rubio et al [21]. IZO and IZrO are high-mobility TCOs with low carrier density. As a result, they exhibit low free carrier absorption and high transmittance for wavelengths >1000 nm. After the absorber growth, both materials become even more transparent for wavelengths >1200 nm, which might indicate a further reduction in free carrier density. The IZrO back contact (blue line) shows no indications of free carrier absorption up to 3 J. Phys. Energy 7(2025) 045018 R Gutzler et al Figure 1. (a) Transmittance of IZrO, FTO, IZO, and ITO with and without a ∼10 nm Mo layer before CIGS growth at 580 ◦C (2.2 µm thickness). (b) Transmittance after CIGS growth, the CIGS layer was mechanically removed before the measurement. Figure 2. (a) Reflectance of ITO and IZrO back contact after absorber growth with absorber (dashed lines) and with mechanical removed absorber (solid lines). Reflectance is measured through the glass (back side), i.e. light passes through glass/TBC /absorber. (b) T/(1 −R) (dashed line) and absorption A =1−R−T (solid line) as a function of wavelength for ITO and IZrO after absorber removal. 1600 nm. In contrast, the large free carrier absorption in ITO compared to IZrO can be observed in figure 2(a) as an increasing reflectance starting at 900 nm (orange line). The reflectance of the IZrO and ITO samples after CIGS growth is shown in figure 2(a). It was measured through the glass side of the sample, i.e. light travels through glass/TBC/absorber and is reflected in this order from the interfaces, just as in a bifacial application. Reflectance over the whole spectrum is below 20% for both TBCs and changes only slightly when the CIGS absorber is removed. The reflectance corrected transmittance T/(1 −R) and the absorption A=1−R−Tare plotted in figure 2(b). Again, due to free carrier absorption, ITO shows inferior performance with large absorption and decreasing T/(1 −R) with increasing wavelength above 600 nm. IZrO has A<0.2 and T/(1 −R)>0.8 for the spectrum above 600 nm. Hence, IZrO is the more suitable choice for applications where light with large wavelengths needs to be efficiently transmitted, such as TCOs in tandem top cells. For bifacial applications, the early onset of free carrier absorption at 900 nm in case of ITO will impact transmittance, especially in CIGS solar cells with a band gap Eg⩽1.4 eV. We note in addition that the sheet resistance of FTO and IZO increases significantly during absorber growth, whereas the other TBCs exhibit only minor changes (see table 1). JV curves of solar cells with the different back contacts are shown in figure 3(a). Most cells are characterized by significant blocking behavior apparent in the JV curves. The solar cell with an FTO back contact (red) shows strong blocking, while ITO (orange) leads to full blocking of the current. The blocking in the case of ITO can be reduced by adding a thin Mo sacrificial layer of 10 nm thickness (gray). The IZO-based cell (turquoise) also shows reduced blocking compared to FTO and bare ITO, while the least 4 J. Phys. Energy 7(2025) 045018 R Gutzler et al Table 1. JV parameters of CIGS solar cells (total area), Rsq of TBC before and after CIGS growth (standard thickness with Eg=1.1 eV), maximum transmittance at wavelengths <1200 nm (TRNmax) of all tested TBCs. PCE (%) FF (%) VOC (mV) JSC (mA cm−2) Rsq (Ω/□) (pristine TBC) Rsq (Ω/□) (after CIGS growth) TRNmax (%) (<1200 nm, pristine TBC) TRNmax (%) (<1200 nm, after CIGS growth) IZrO 14.0 (14.2∗) 63.8 659 33.3 (33.8∗) 20 19 89 77 FTO 2.1 26.5 538 14.6 13 813 85 49 IZO 5.3 27.9 628 30.4 10 134 83 71 ITO 0.0 23.5 115 0.4 8 11 79 75 ITO/Mo 7.4 36.7 622 32.3 7 7 22 59 ∗PCE corrected for JSC derived from EQE measurement. 5 J. Phys. Energy 7(2025) 045018 R Gutzler et al Figure 3. (a) Light JV curves of CIGS solar cells (standard thickness) fabricated on IZrO (blue), FTO (red), IZO (turquoise), ITO (orange), and ITO/Mo (10 nm) (gray). (b) Relative instability (difference in enthalpy of formation) of the used TBC materials and MoO3with respect to Ga2O3. blocking appears for the IZrO back contact (blue). In view of the often observed formation of GaOxat the TBC/absorber interface, relative thermodynamic stability of a TCO material with respect to Ga2O3has been discussed [4]. Figure 3(b) shows the difference in enthalpy of formation for the TBC materials used here with respect to the more stable Ga2O3. Due to a lack of literature values, the enthalpy of formation of the oxide mixtures ITO, IZO, and IZrO was interpolated from the values of the pure compounds taken from the literature [26], weighted by their respective molar fraction. Thus, alloying ZnO into In2O3(IZO) destabilizes the indium oxide due to its less negative enthalpy of formation. ZrO2is more stable than In2O3and hence makes the mixture more stable. For FTO, the value of pure SnO2was taken as an approximation due to the lack of other data. MoO3is also shown in view of the Mo sacrificial layer used. All TBCs are less stable than pure Ga2O3, i.e. at elevated temperatures with Ga present, all materials will form GaOx. This relative instability correlates with the extent of the blocking behavior in figure 3(a): the least stable material FTO significantly blocks current, likely due to an extensive GaOxlayer at the interface. The most stable material IZrO shows the least blocking, and materials with intermediate stability block to some extent. MoOx, while thermodynamically less stable than ITO, improves device performance. Previous studies have shown that a Mo sacrificial layer reacts with Se and O to form MoSexand MoOx[13,17], protecting the underlying ITO from reacting with Ga from the absorber to form GaOx[20]. The role of GaOxat the back contact is currently not fully understood since it appears to be often detrimental but can also be benign. If the GaOxis of n-type, it is reasonable to assume that a counter diode forms at the back contact if it is thick enough. A very thin layer might not form a pn-junction observable in JV measurements. However, doping Ga2O3with Cu induces p-type conductivity and shifts the valence band towards the Fermi level [29,30]. It is possible that some of the Cu provided during CIGS growth is doping the GaOxlayer and alters its electronic properties sufficiently to change it from detrimental to benign. The observed blocking then comes about naturally by taking together thermodynamic stability of the different TBCs leading to extensive GaOx formation during absorber growth, and the resulting formation of a counter diode with n-type GaOxat the back contact. We note, however, that the performance of a TBC might also depend on the TCO deposition process and other layer parameters; for example, the ITO Sol30 by the same commercial supplier as the ITO Sol12 shown here led to less blocking (data not shown), but still more than IZO or IZrO. The thermodynamic stability of a TCO gives a reasonable first hint toward its suitability as a back contact. However, other chemical and physical processes are likely also at play. Dopant segregation in the form of reduced F content was observed for FTO at CIGS deposition temperatures as low as 550 ◦C [17,19], and is a probable explanation for the here observed increased sheet resistance and low cell performance. Furthermore, selenization of the upper TBC layers [21], or changes in oxygen vacancies will also take place during high-temperature processes and will depend on the relative abundance of other elements like Se and Ga. Understanding the thermochemistry at the interface is essential for optimizing TBC stability and interface properties. This is exemplified by the solar cell based on an IZrO back contact (see table 1). The absorber’s band gap, calculated from the onset of the absorption edge of an EQE measurement (linear extrapolation of E×EQE2), is 1.1 eV and the cell efficiency calculated using the current density from EQE is 14.2% The observation that the thermodynamic stability of the TBC is a limiting parameter is confirmed by experiments in which the transport speed of the samples in the inline coater is increased by a factor of four. Without altering the metal source temperatures, this results in a roughly four times thinner absorber layer, 6 J. Phys. Energy 7(2025) 045018 R Gutzler et al Figure 4. Light JV curves of CIGS solar cells fabricated on IZrO (blue), ITO (orange), and FTO (red) with a thin CIGS absorber (530 nm thickness). Table 2. JV parameters of CIGS solar cells, Rsq of TBC after thin absorber growth (530 nm thick CIGS). PCE (%) FF (%) VOC (mV) JSC (mA cm−2) Rsq (Ω/□) (after absorber growth) IZrO 7.7 55.2 563 24.8 26 ITO 6.8 49.5 575 23.8 9 FTO 4.8 44.0 455 23.9 19 but also the TBC being exposed to elevated temperatures for a much shorter time. In this case, the TBCs that showed complete (ITO) or substantial (FTO) blocking under standard deposition times now yield functional devices with higher FF. Figure 4shows the corresponding JV curves of FTO (red), ITO (orange), and IZrO (blue) of solar cells with a 530 nm thick CIGS absorber. The blocking is substantially reduced, presumably from both thinner GaOxformation and fewer chemical changes in the TBC itself (e.g. loss of F-doping or formation of Se–In compounds). Table 2gives the corresponding JV parameters. The sheet resistance of FTO after absorber growth remains much lower compared to the slower deposition and remains similar for IZrO and ITO. The TBC’s sheet resistance has a direct impact on the cell’s series resistance, thus lowering the FF. However, the TBC with the largest sheet resistance (IZrO) has also the highest FF, highlighting that sheet resistance is only part of the picture and that further mechanisms are at play. Differences in VOC can be explained by different Na-doping from the soda-lime glass, as all three TBCs likely lead to different Na diffusion kinetics due to grain structure and thickness of the film. The combined observations that a reduced maximum process temperature is required to achieve functional PV devices with transparent back contact, that current blocking correlates with thermodynamic stability of the TBC, and that a faster processing speed is required to get otherwise problematic materials to work as back electrode strongly suggest that the overall thermal budget during the absorber growth process is the critical parameter which needs to be controlled. The kinetics of GaOxgrowth and other detrimental changes to the TBC can be slowed down by reducing the overall thermal load by either reducing maximum temperature and/or limiting the exposure time to high temperatures. Bifaciality is one of the goals that motivates the use of transparent back contacts. The bifaciality of the solar cells with a 530 nm thin absorber was probed with both JV and EQE measurements several months after cell fabrication and initial JV measurements (unencapsulated cells were stored under ambient conditions). Figure 5(a) shows the JV curves of a cell with ITO back contact under front illumination (orange) and rear illumination (dark orange), and figure 5(b) of an IZrO cell under front illumination (blue) and rear illumination (purple). The barrier in the ITO cell also impacts the JV curve under rear illumination, although it appears less pronounced. The IZrO cell does not exhibit such a barrier. The series resistance is larger under rear illumination. The corresponding EQE measurements shown in figure 5(c) for the ITO cell and figure 5(d) for the IZrO cell show a reduced but non-zero EQE across the absorbed spectrum. The EQE 7 J. Phys. Energy 7(2025) 045018 R Gutzler et al Figure 5. Dark and light JV curves of a 530 nm thick CIGS solar cells with (a) front illumination (orange) and back illumination (dark orange) with ITO as TBC and (b) front illumination (blue) and back illumination (purple) with IZrO as TBC. EQE measured on the same CIGS cells with front and backside illumination (c) on ITO and (d) on IZrO (same color code). of the IZrO cell is higher both under front and rear illumination compared to the ITO cell and also the relative front/rear EQE is consistently larger for the IZrO cell across the entire spectrum, not only in the region above 900 nm. The bifaciality in JSC, defined as Jback illumination SC,EQE Jfront illumination SC,EQE , is 60% for IZrO and 46% for ITO cells. The better EQE for the IZrO cell is a direct consequence of its higher transmittance of the TBC in the visible and to a lesser extent in the red part of the spectrum due to lower free carrier absorption. Under rear illumination, both TBCs show an increased EQE around 350 nm compared to front illumination, which is due to the absence of the absorption of UV light in the CdS buffer layer. This effect is slightly larger in ITO compared to IZrO, possibly because of its larger optical gap as a consequence of the larger Burstein-Moss shift due to ITO’s higher doping. The lower EQE under back illumination for short wavelengths can be attributed to the short diffusion length of photo-generated carriers [1,8,31] and high recombination velocity at the back contact [14]. Carriers formed close to the back contact by high-energy photons and far away from the pn junction are more prone to recombination before contributing to charge-collection efficiency. For larger wavelengths, incomplete absorption, amplified here by the thin absorber, lowers not only carrier collection under back illumination, but is also observable in the sloping off of the EQE towards larger wavelengths under front illumination. The better optical properties of IZrO compared to ITO (higher transmittance especially for longer wavelengths and lower absorption, see figure 2) lead to improved charge collection for all wavelengths. Overall, the general shape of the EQEs under front and back illumination and the ratio between the current densities is comparable to reported absorbers with similar thickness [8,19,31,32]. A complete list of JV parameters under front and rear illumination is provided in table 3. The cell with ITO back contact shows stronger signs of degradation over time (several months between initial and bifacial measurements) in both VOC and FF compared to the cell with IZrO contact, but both cells remained fully functional. Whereas the FF remains roughly constant between front and rear illumination, VOC decreases by a few tens of mV and JSC decreases by 30%–40%. 8 J. Phys. Energy 7(2025) 045018 R Gutzler et al Table 3. JV parameters and JSC extracted from EQE of 530 nm thick CIGS solar cells under front and rear illumination with ITO and IZrO back contact. Front PCE (%) Front FF (%) Front VOC (mV) Front JSC (mA cm−2) Front JSC (mA cm−2) from EQE Rear PCE (%) Rear FF (%) Rear VOC mV) Rear JSC (mA cm−2) Rear JSC (mA cm−2) from EQE IZrO 8.4 56.9 549 26.9 27.1 4.9 57.1 517 16.7 16.3 ITO 5.1 40.3 512 24.6 23.5 3.3 38.6 478 17.7 10.8 9