Unveiling a Near‐Total Rear Passivation for High Performance of Ultra‐Thin ACIGS Xavier L Pinheiro, António J N Oliveira, André F Violas, Kevin Oliveira, Paulo A Fernandes, Marika Edoff, Pedro M P Salomé, Jennifer P Teixeira Abstract Thin‐film Cu(In,Ga)Se2 solar cells are a key photovoltaic technology for addressing global energy and environmental challenges. Reducing absorber thickness lowers manufacturing cost and critical raw material consumption but introduces optical and electrical losses due to incomplete light absorption and increased rear interface recombination. Rear recombination can be mitigated through Ga grading and/or rear contact passivation, for example using tunnel oxide as in c‐Si TOPCon technology. In this work, a full coverage plasma enhanced chemically vapor deposited 8 nm porous SiOx passivation layer was implemented in ultra‐thin (Ag,Cu)(In,Ga)Se2 (ACIGS) solar cells and compared with patterned passivation layers (∼70% coverage) and an unpassivated reference. Standard current‐voltage and external quantum efficiency characterization methods show that the complete passivation yielded the highest performance, achieving 16.4% efficiency, a 2% absolute improvement over the reference. Patterned samples showed enhanced performance to a lower extent. Elemental and structural analysis showed that the SiOx layers remained intact after 550°C deposition of the 700 nm ACIGS absorber. Notably, a MoSe2 interfacial layer formed between the full SiOx layer and Mo contact, likely due to Na and/or Se diffusion through SiOx pores. Combining near‐complete passivation and the ohmic MoSe2 contact may explain the improved performance. Ultra‐thin solar cells of (Ag,Cu)(In,Ga)Se2 (ACIGS) suffer from rear interface recombination, limiting their performance. To tackle it, a full coverage plasma enhanced chemically vapor‐deposited 8 nm porous SiOx passivation layer was implemented and evaluated. It is shown that a combination of the near‐complete passivation from silica and the formation of MoSe2 below may explain the higher performance. Introduction The relentless pace of technological advancement, the increasing global population, climate change, and the world's geopolitical context have fuelled a surge in energy demand in the past few years. Global energy demand increased by 2.3% in 2024, while electricity demand grew even more rapidly, by 4.3%, approximately twice the rate of total energy demand []. This impressive and challenging requirement to fulfil can be partially answered with one of the primary sources of clean energy—photovoltaics (PV), allowing it to also meet sustainable energy targets defined by the European Union []. Specifically, Building Integrated PV (BIPV) brings energy production directly to the consumer and having the advantage of decreased land usage []. In late 2024, the BIPV market accounted for 2 GW of installed power globally, while standard PV was above 1.8 TW [, , ]. In order to increase the BIPV footprint, PV technologies must meet the necessary criteria of efficiency, stability, weight, flexibility, and aesthetics []. Despite wafer‐based crystalline silicon (Si) dominance, specifically monocrystalline, over the PV market – 97% at the end of 2025 [], thin‐film technologies can be key players in BIPV applications. Specifically, Cu(In,Ga)Se2 (CIGS) solar cells provide several advantages, such as great thermodynamic stability [], high absorption coefficient [], and mature commercialization in flexible substrates []. Moreover, it holds a world record efficiency of 23.6% [, ] and has a low energy payback time []. One extra step can be taken if, instead of a thin‐film (∼2000 nm thick absorber), the ultra‐thin‐film range in CIGS is considered (≤ 750 nm). Such alteration allows to decrease the production costs and time, increase throughput, and reduce the usage of critical raw materials, meeting sustainable goals [, ]. The ultra‐thin regime further enables the production of lightweight and flexible modules, easing their application in BIPV, which will reduce the building ́s CO2 emissions, leading to a net‐zero scenario. However, such drastic reduction in the absorber thickness has led to a performance loss from 23.6% to 16.2% or 18.2% for 750 nm, considering the full or active area, respectively, and to 15.2% for 490 nm [, ]. The main causes of such loss are incomplete light absorption, with a severe decrease for thickness values below 500 nm, and rear interface recombination, which becomes more impactful at this level [, , ]. Two main strategies have been employed to mitigate recombination at the rear interface: (i) Ga grading with a higher content at the rear, which is a demanding task in an ultra‐thin absorber due to the challenging elemental diffusion control and reproducibility [, ], and (ii) the addition of a nanostructured passivation layer []. The dielectrics used as passivation layers, alumina (AlOx), silica (SiOx), titania (TiOx), and hafnia (HfOx) [, , ] allow to increase the performance of the overall solar cell by reducing the impact of rear interface defects, but these have to be patterned, e.g., with lines or holes, to allow a direct contact between the absorber and the metallic back contact [, , , , , ] since they commonly are insulators. Nevertheless, it may be possible to implement a full‐layer strategy by replicating what has been done in Si technology [, , ]. The implementation of carrier‐selective contacts at critical interfaces provides passivation, conductivity, and selectivity, reducing recombination losses and enhancing the charge carrier collection at the corresponding interfaces [, ]. It is no easy feat to provide all these requirements with one single layer, so architectures combining passivation (dielectrics), selectivity (e.g., transition metal oxides), and conductivity (e.g., transparent conductive oxides) layers have been tested in Si solar cells [, ]. Regarding CIGS solar cells, some progress has been made by applying selective layers at the front and rear interfaces with proven performance increase [, , , , ]. While the development and implementation of selective contacts may require an individual thin layer, or even the combination of several distinct thin layers, the integration of one single ultra‐thin layer may be capable of providing similar benefits to performance. Tunnel oxide passivated contact (TOPCon) is one of the most promising crystalline Si solar cell architectures; in this structure, an ultra‐thin (1–2 nm) silicon oxide layer is added to the rear contact, enabling interface passivation and charge transport through tunnelling [, , , , ]. TOPCon technology is at a critical point with a proven record efficiency of 27.8% [], comparable with 27.8% for heterojunction interdigitated back contact (HIBC) solar cells [] and 24.5% for Passivated Emitter Rear Contact (PERC) []. However, creating a passivation layer with such reduced thickness requires highly controllable and conformal deposition techniques, such as Atomic Layer Deposition (ALD) [, ]. Furthermore, in CIGS based solar cells an ultra‐thin layer may not be able to provide the necessary passivation effect, depending on the material: for Al2O3, full passivation layers with 2 and 5 nm still led to a blocking behavior [, , ]; considering HfO2, thickness values below 2.5 nm do not show a blocking behavior but the solar cell performance is not improved []. Jian et al. stated that in ultra‐thin passivation layers two transport mechanisms may occur: (i) pure tunnelling and (ii) type ohmic through the pinholes or pores in the oxide layer []. Therefore, a porous passivation layer can be implemented in a full‐area architecture, possibly delivering passivation effects while maintaining charge conduction. The present study tests the possibility of reaching a significant passivation effect in an ultra‐thin (Ag,Cu)(In,Ga)Se2 (ACIGS) solar cell with a full coverage thin layer of a porous dielectric. Additionally, it is compared with a conventional device with no passivation scheme, as a reference, and two line‐patterned passivated devices around 70% coverage and different thickness values. Silica (SiOx), deposited by plasma‐enhanced chemical vapor deposition (PECVD), was chosen as a passivating material due to its effective rear interface passivation, and the thickness was around 8 nm to maximize the passivation effect, as previously reported [, ]. Current density‐voltage (JV) and external quantum efficiency (EQE) measurements were performed under standard conditions for solar cell characterization. The results show a clear light‐to‐power conversion efficiency improvement through the application of any of the studied schemes, with the highest efficiency for the porous full‐layer. Through electron microscopy, a uniform distribution of MoSe2 was identified at the SiOx/Mo interface of the 8 nm patterned and porous full layers, which was assumed to be formed during the diffusion of Se and Na through the pores of the dielectric layer. The higher performance is attributed to the combination of near 100% passivation area with the ohmic‐type contact of the Mo/MoSe2/(SiOx)/ACIGS enabled by the pores. Experimental Section The solar cells fabrication started with 1.1 mm thick Soda‐Lime Glass (SLG) substrates with a sputtered 350 nm Mo (molybdenum) layer on top. Four devices were fabricated, with the substrates depicted in Figure : a conventional reference (REF) with no passivation layer; and three passivated samples in which two have a porous 8 nm SiOx layer, one full coverage (F8) and the other line patterned (P8), and the latter is a 25 nm line patterned SiOx layer (P25). The SiOx layers were deposited by PECVD using an SPTS MPX CVD system, with a plasma power of 30 W at 13.56 MHz, a substrate temperature of 300°C, using SiH4 and N2O precursor gases, and 0.9 Torr working pressure. Majee et al. reported silica films with high porosity levels (25%) with similar PECVD deposition parameters [], so the deposition conditions were the same for all passivation layers, and the thickness was the critical parameter. After this step, the samples were prepared separately, according to the required patterning, while the full‐layer sample was ready for the absorber deposition. The line pattern samples were coated with 600 nm of photoresist (AZ1505) to be exposed using direct write lithography (DWL), a photolithography laser system from Heidelberg – DWL 2000. After exposure, the photoresist was developed in AZ 400K 1:4, revealing the open lines with approximately 900 nm of width and 2800 nm pitch. The exposed lines were then etched by reactive ion etching using an SPTS ICP system with a CF4 chemistry. The remaining photoresist was removed from the samples by an acetone bath for 30 min, followed by an isopropanol and deionized water cleaning in an ultrasound bath. Prior to the absorber deposition, a 15 nm layer of sodium fluoride (NaF) was evaporated on top of the substrates []. Subsequently, the ACIGS absorber layers were deposited on all four devices in a single co‐evaporation run at 550°C, resulting in a nominal 700 nm thick layer, compatible with an ultra‐thin solar cell configuration. The composition values of this absorber layer, determined by x‐ray fluorescence (XRF) performed in a Panalytical Epsilon 5, are ([Ag]+[Cu])/([Ga]+[In]) = 0.79, [Ag]/([Ag]+[Cu]) = 0.11, and [Ga]/([Ga]+[In]) = 0.40. An intentional Ga profile was implemented with a high‐Ga and low‐In content near the rear interface. Afterward, the devices were completed with a 50 nm cadmium sulphide (CdS) layer deposited by chemical bath deposition and an optical window of 50 and 200 nm sputtered bi‐layer of zinc oxide (ZnO) i‐ZnO/ZnO:Al, respectively []. The individual cells, which have an area of 0.1 cm2, were separated via a lithography step and chemical etching of the CdS/i‐ZnO/ZnO:Al stack []. No post‐deposition treatments or anti‐reflection coating were applied in the studied devices. After the passivated substrate preparation, these were characterized by scanning electron microscopy (SEM) through a FEI Nova Nano SEM 650 system with an acceleration voltage of 5 kV. X‐Ray Reflectometry (XRR) analysis was performed on a PANalytical X'Pert PRO Materials Research Diffractometer equipment (Cu Kα x‐ray tube λ = 0.154 nm, 45 kV, and 40 mA) to assess the film density and porosity level. With a complete device, the solar cell elemental variation and Ga grading were measured with a depth profile by Glow‐discharge optical emission spectroscopy (GDOES). Then, the devices were characterized under standard conditions by JV measurements, and EQE was taken at 0 V. Capacitance‐voltage (CV) measurements were also performed with an AC signal with an amplitude of 25 mV and a frequency of 10 kHz. Optical simulations allow us to study the optical impact of the passivated layers on the solar cell performance and, in this way, decouple the possible electrical gain. Hence, the commercial software Ansys|Lumerical was used to run the 3D finite‐difference time‐domain (FDTD) modelling of the solar cell devices in this work [, ]. The simulations were designed with the smallest possible mesh size, considering time and memory constraints. Besides, for critical interfaces, a mesh size of 1 nm was considered. The simulation included both refractive index (n) and extinction coefficient (k) values (n(λ) = n+ik), taken from the literature [, , , ]. Cross‐sectional samples from the produced devices were prepared using a FEI Helios NanoLab 450S focused ion beam (FIB) to have electron‐transparent lamellae (thickness < 100 nm) for subsequent scanning transmission electron microscopy (STEM) coupled with energy‐dispersive x‐ray spectroscopy (EDS) analysis. To ensure defect‐free lamella fabrication preserving its condition, a protective layer of platinum was deposited before any etching. Since a gallium (Ga) ion source was used to etch, this element was excluded from the EDS study. High‐angle annular dark‐field (HAADF) STEM images were acquired using a double‐corrected FEI Titan G3 Cubed Themis; images of 2048 × 2048 pixels were recorded using a convergence angle of 21 mrad with pixel dwell time set to 8 μs. EDS‐mapping was performed with the same instrument using a Super‐X EDS detector. Iterative maps of 2048 × 2048 pixels were recorded with a dwell time per pixel of 5 μs at 200 keV, under a current between 150 and 250 pA and collection times of 15 min. The Velox software from Thermofisher Scientific was used for acquiring and processing the STEM‐EDS data. For the acquired elemental composition line scans, the following characteristic x‐ray emission lines were used: Kα for Si and Se, and Lα for Mo. Regarding the x‐ray photoelectron spectroscopy (XPS) analysis, the system ESCALAB 250 Xi Thermo Scientific equipped with a monochromatic Al Kα source (1 487 eV, 500 μm radius spot) was used. The acquired spectra include a low‐resolution Survey to assess all present elements and high‐resolution for quantification of C 1s, Si 2p, O 1s, Ag 3d, Cu 2p, In 3d, Ga 2p, Se 3d, Mo 3p, and Na 1s elemental orbitals. Results and Discussion Figure shows a representative SEM top‐view image of the developed line contact passivation scheme, consisting of a direct contact region with the Mo back contact (in black) and dielectric lines for passivation (in light gray). The developed scheme features a passivation line width of 1945 ± 22 nm with a pitch of 2800 nm for both P8 and P25 samples. The contact openings have a width of 855 ± 33 nm, accounting for a passivation area of around 69.5%. Concerning the full porous layer, XRR was used to determine its density and thus porosity. For a similar 20 nm SiOx thin‐film on a Si substrate (chosen to have sufficient interference in the XRR pattern), a density (ρ) of 1.3 g∙cm−3 was determined using the critical angle θc (θc≈2δ∝ρ) []. The typical density of SiO2 deposited through PECVD revolves around 2.3 g/cm3 [], indicating the deposited thin‐film has a porosity level above 40%. The remaining solar cell layers (ACIGS/CdS/ZnO bi‐layer) were deposited on the prepared substrates. The first approach to evaluate the incorporation of the passivation layers, full or patterned, into the ACIGS solar cell architecture was based on an in‐depth elemental analysis. The GDOES elemental plots are shown in Figure for the REF, F8, P8, and P25 solar cells. The topmost layers (CdS/i‐ZnO/ZnO:Al) were chemically etched so the elemental profiling starts at the ACIGS absorber. Regarding the Si signal, associated with the passivation layers, it is possible to observe a Si peak at the interface where the layer is located for all the passivated samples. As expected, the higher the SiOx thickness, the more intense the Si signal. Moreover, despite the same passivation layer thickness value, F8 presents a higher Si signal than P8, which might be related to the higher passivation area attained for the F8 device. Altogether, these results suggest that the passivation layer was confined to the rear interface following the ACIGS deposition, although this analysis does not ensure that the developed schemes remained intact. Another tendency to highlight is related with sodium (Na), where the presence of a Na peak at the rear interface can be attributed to accumulation at the rear interface of the ACIGS for the passivated samples. This result has already been previously reported for passivated ultra‐thin cells with a NaF pre‐deposition layer, since the SiOx layers can partially block its diffusion; however, it is difficult to pin‐point the origin of Na accumulation, either from the NaF or diffusion from the SLG substrate, which is not a crucial point []. In all samples, throughout the ACIGS layer thickness, Ag, Cu, and Se are mostly kept constant; for Ga and In, there is a slight increase and decrease, respectively, toward the Mo rear contact. Such linear grading is compatible with the commonly implemented Ga‐profile that increases both open circuit voltage (VOC) and fill factor (FF) by repelling photogenerated electrons from the rear interface, further reducing rear recombination [, ]. This strategy can be considered one of the first steps to develop a carrier selective contact for ACIGS solar cells since there is an electronic engineering that changes the conductivity of one of the charge carriers and hence the selectivity at the interface. We highlight that this Ga profile is the same in all samples. Additionally, the final devices were characterized via SEM top‐view and cross‐section to access the surface and layer's morphology. In Figure to C the top‐view images of the F8, P8, and P25 samples surface are presented, in which a uniform surface morphology is found in Figure arising from the full coverage passivation layer, while for the patterned samples in Figure the line pattern can be noted. For sample P25, it is more noticeable since the thickness is higher, resulting in a transference of the pattern to the top layer, but for sample P8, since only 8 nm were used, only a small contrast can be noted. Despite the pattern transfer, in the cross‐sectional characterization of Figure , a clean coverage of the buffer and window layers on top of the absorber was attained; therefore, with no significant impact on the optoelectronic properties of the devices. In all samples, the columnar Mo rear contact can be seen, after which the ACIGS absorber is with similar size large grains. The solar cells fabricated with the aforementioned substrates were characterized with JV and EQE, with the representative JV illuminated curves being shown in Figure , and the averaged values from the figures of merit, along with the best cell parameters, presented in Table . All devices exhibit diode‐like JV characteristics, with no evidence of significant shunt paths or excessive series resistance. By strict best cell comparison, the highest efficiency value of 16.4% is obtained for the F8 sample, followed by the line pattern samples, P8 and P25, with the REF sample presenting the lowest efficiency value of 14.6%. It is important to note that the value of 16.4% efficiency is in the same range as the current record efficiency for sub‐750 nm cells, which is an indication of an effective coupling of an optimized ACIGS absorber with an efficient rear interface passivation strategy [, ]. The same trend is obtained for the average efficiency values. The integration of a rear interface passivation scheme, independently of the architecture, enables for improved VOC, FF, and short circuit current density (JSC) values, in comparison to the REF device. The highest VOC increase, which corresponded to a value of around 60 mV, was obtained for the F8 sample, which might have arisen from a larger passivation area with respect to the patterned substrates. In order to differentiate the passivation VOC gains, CV characteristics were acquired for each sample, allowing the determination of the net free carrier concentration (Ncv) at the short circuit condition [, ]: 5.1 × 1015 cm−3 for REF, 6.1 × 1015 cm−3 for F8, 5.0 × 1015 cm−3 for P8, and 2.8 × 1015 cm−3 for P25. With these values, the expected VOC variation from the NCV differences (ΔVOC−NCV) can be calculated for each of the passivated samples [], and then the effective passivation gain (ΔVOC−Pass.=ΔVOC−JV−ΔVOC−NCV) may be estimated. For the patterned passivation samples, P8 and P25, a ΔV OC − Pass. gain of around 45 mV was attained, while for F8 a gain of 55 mV was obtained. A similar effective passivation gain is expected for the patterned samples, as both have an equal nominal passivation area coverage. The higher effective gain in sample F8, in the same order as the JV gain, further showcasing the potential of this near‐total rear passivation scheme. Moreover, the VOC enhancement is in the upper range with respect to the ones reported in the literature, indicating an effective passivation of recombination mechanisms, especially considering that all devices have approximately the same bandgap energy value around 1.32 eV, obtained through the first derivative of the EQE curves, with the values shown in Table [, ]. To gain deeper insight into the recombination mechanisms and diode behavior of the fabricated devices, a one‐diode model fit was applied to the light JV curves, following Hegedus et al. [], enabling extraction of advanced electrical parameters including the ideality factor (A), saturation current density (J0), shunt resistance (Rshunt), and series resistance (RS) that allow a more comprehensive assessment of junction quality and recombination pathways across the different passivation strategies. The obtained A value is similar across all devices, with a value around 1.5 – 1.7, showing the consistency of the developed ACIGS absorbers. The obtained values suggest that dominant recombination mechanisms can be described by the Shockley–Read–Hall (SRH) model []. With respect to J0, the higher value is obtained for the REF sample, suggesting recombination is in fact reduced with the implementation of any of the studied passivation strategies. Regarding the Rshunt parameter, an increase is obtained for all passivated samples, regardless of the architecture, with respect to the REF one. Notably, the higher shunt resistance (Rshunt) average value is attained for the F8 sample, in which the larger passivated area was designed. A reduction of shunting paths, seen through a larger Rshunt, has already been previously reported for ultra‐thin devices emplyoing a dielectric passivation strategy, and this has been attributed to a reduction of the rear interface conductive area [, , , , ]. Such an increase in Rshunt is important as with the thinning down of the absorber, the probability of increased shunts increases, and therefore the dielectric layer brings additional benefits other than just of passivation. Finally, for RS, this parameter may increase in the passivated samples since we are adding a “non‐conducting layer” to the rear interface; however, a nontrivial trend is observed. Although no significant variations are obtained in the Rs values between all samples, comparing the REF and the patterned passivated samples, P8 and P25, the RS does increase but only slightly; for the sample F8, the RS decreases, being the lowest. Such a decrease in the RS suggests that no barrier blocking the charge carriers exists and that the porosity may have enabled pathways for efficient charge extraction. Therefore, this in‐depth one‐diode JV analysis supports that the observed electrical improvements stem from the passivation effects at the rear interface. However, in the case of the full SiOx layer, which achieved the best performance, the origin of the lower RS is still unclear. Regarding FF, since no significant differences are attained in the RS value, it follows the Rshunt one, with the highest FF average value of 76.2% being obtained for the F8 sample, and the lowest of 74% for the REF one. Looking into Jsc values, a maximum increase of around 0.6 mA·cm−2 is obtained for the P25 sample with respect to the REF one. The representative EQE data in Figure and the ACIGS absorptance spectra presented in Figure , obtained from FDTD 3D optical simulations, were used to decouple the passivation electronic effect from the optical improvement. In both measured and simulated EQE, a slightly higher EQE in the passivated samples from the REF one arises in the spectral region from 600 nm to the Near‐Infrared. Such an increase occurs due to an improved rear reflection from the difference between CIGS‐based (3.01 at 600 nm) and SiOx (1.47 at 600 nm) n values [, , ]. Overall, the simulated optical gain is small, only visible on the two peaks in the 750 – 900 nm region, and the introduction of any full or pattern SiOx layer should give a slight maximum gain in Jsc of 0.2 mA·cm−2. Hence, the overall JSC enhancement obtained in the JV measurements is attributed to improved rear interface defect passivation. Finally, a statistical representation of the figures of merit obtained from the JV characteristic curves is presented in Figure , highlighting the enhancements obtained with the passivation strategies, more predominant with the full‐layer coverage (F8). Moreover, the figures of merit of each sample have a small spread around the average value, as also seen by the standard deviation values in Table . Despite the measured Si signal in the GDOES measurements, there is still no definitive proof that the SiOx passivation schemes are fully intact following the ACIGS deposition. Moreover, a more in‐depth analysis of the contact nature is also key to better understand the solar cells' performance. Thus, to assess the SiOx layer integrity and the contact characteristics, a lamella of each of the passivated devices was prepared through FIB, and a STEM analysis was performed. While for the STEM only one lamella was studied, the complete near 5 μm side length was scrutinized to ensure the observations were uniform. In the STEM images HAADF of Figure is possible to see that for the line pattern passivated samples, P25 and P8, the SiOx lines are intact, with a thicker layer on the P25 sample. Moreover, the CdS buffer layer and both ZnOx‐based layers of the optical window are also visible, with a uniform coverage over the ACIGS absorber. In a standard Mo/CIGS interface, the selenization of the Mo surface that occurs during the first stages of the ACIGS deposition leads to the MoSe2 layer formation []. In Figure , two line scans in the contact line interface region were performed: line scan A in P8 and P25, being the EDS line profiles for Se, Mo, and Si elements shown in Figure , respectively. There is a clear overlap of the Mo and Se signals, suggesting the formation of the MoSe2 layer within the line contacts of the passivated samples. It is noteworthy that it has been demonstrated that this layer facilitates the formation of a quasi‐ohmic type contact between the CIGS and Mo [, ]. Furthermore, it is expected to find this layer only at the contact openings, and not where the SiOx was deposited. The EDS line scans performed in the passivation region, shown in Figure , line scan B in P25 in Figure and in P8 in Figure , present a peak associated with the passivation line and no clear Se and Mo signal overlapping, so no evidence of MoSe2. A Se elemental distribution EDS mapping was also conducted in both the P8 and P25 samples. In Figure , the Se mapping of sample P25 corroborates the line scan B profile of sample P25. However, in sample P8, Figure , several spots emerge with Se below the SiOx, indicating Se penetration through the porous passivation layer. Concerning the sample with a full 8 nm passivation layer, F8, the lower magnification STEM HAADF image in Figure shows the existence of a uniform SiOx layer across the sample. Although by close inspection at higher magnification Figure , small, blurred spots appear within the SiOx layer, which may be associated with a higher‐porosity SiOx region where there is a higher MoSe2 formation probability. Similar to the P8 sample (Figure ), in these blurred spots, the Se elemental distribution bypasses the SiOx layer and reaches the Mo, where it may form MoSe2. The line scan in Figure , performed at the rear interface of sample F8 in the region of Figure , reveals the Mo, Se, and Si distribution in a spot where the Se distribution does not cross the SiOx layer to the Mo; hence, supposedly a lower porosity region, and so no clear Se and Mo overlapping is observed. In the high‐magnification STEM image of the interface below the SiOx layer, where pores are probably present – Figure , the characteristic layered structure of MoSe2 with a c‐axis orientation perpendicular to the Mo interface was detected [, , ]. Despite being recognizable at close, the crystal structure has low contrast. This can have happened due to surface amorphization during the FIB lamella fabrication, but also because these are local size formations that, due to the small thickness of the passivation layer and the roughness of the Mo rear contact, do not have a lateral and depth wise uniform distribution. It is suggested that the porosity of the full passivation layer is the key characteristic that enables the highest‐performance device, by decreasing recombination at the rear interface and still allowing charge transport to occur in a type‐ohmic contact formed with MoSe2. Alkali doping, in this case by diffusion of Na from the SLG and the pre‐deposition treatment (DT) of NaF, is a key element for high‐performance CIGS‐based solar cells [, ], and it has been demonstrated that Na is a driving force in the formation of MoSe2; when no Na source is present, the MoSe2 formation is seriously hindered [, ]. Ledinek et al. showed the importance of Na when having a full passivation layer of AlOx: when no source of Na exists, the solar cells exhibit a blocking behavior; also, if only a post‐DT is applied (after de CIGS growth), the performance is still hindered if a thick passivation layer is used. However, when Na is supplied and free to diffuse, the solar cell can benefit from enhanced performance via rear passivation []. These have been tested in Al2O3 and HfO2, and in both, a critical possibility of Na damaging the passivation layer arises; although no clear holes in the layer are observed [, ]. Nevertheless, Na supply is necessary to enable conductivity through the full passivation layers, although being a compromise between the benefits of Na and the rear passivation area coverage []. In previous works, when the alkali metals are supplied by a spin‐coated alkali fluoride on top of the passivation layer, the strategy developed is to intentionally perforate the passivation layers with alkali diffusion; the openings are randomly distributed and allow direct contact between the CIGS and the Mo while maintaining a passivation area above 95% [, , ]. Lastly, it can be stated that Na doping is essential for the high performance of the solar cell with the full passivation layer, since it may facilitate the formation of MoSe2 through the porous passivation layer. STEM‐EDS shows that all rear passivation strategies are locally intact, and the 8 nm SiOx layer in samples P8 and F8 allows the diffusion of Se and so the formation of MoSe2 below the passivation layer. To further investigate if this result extends across all area of sample F8, a peel‐off procedure was done to perform a top‐view analysis of a larger area of the rear interface. Such a procedure was performed as reported elsewhere [], with an extra cooling step with liquid nitrogen between the epoxy curing and the cleavage step. After such a procedure, two distinct sides of the sample were inspected: (i) the metallic sheet with the ZnO:Al/i‐ZnO/CdS/ACIGS structure (from bottom to top) and (ii) the original SLG with the Mo rear contact, the MoSe2, and the passivation layer. In Figure , a top‐view SEM image of the rear surface and corresponding Si, Se, and Mo elemental distribution EDS maps are presented for side (ii). The SiOx layer is intact, even after a peel‐off procedure. Across the surface, there are several debris that is believed to be remnants of the peel‐off procedure, including the darker spots. Nevertheless, the SiOx‐coated Mo rear contact has an as‐deposited appearance, showing that it was not damaged in the harsh ACIGS deposition. The Si elemental distribution across the rear interface surface (Figure ) displays a uniformly distributed SiOx layer with no interruption; also, the Se distribution is fairly uniform besides often accumulating in some areas. The Se distribution and its accumulation cannot be correlated with any change in the Si signal or any significant morphological modification of the rear interface. This SEM mapping analysis clearly shows the uniform, non‐localized distribution of MoSe2 across the rear interface of sample F8. Nevertheless, to further confirm its existence and to investigate what is the chemical state of these elements, XPS surface analysis was performed. The atomic percentage values for each of the acquired elements are listed in Table , and the elemental orbital spectra of Mo 3d, Se 3d, Si 2p, and O 1s are presented in Figure . Note that contamination due to handling, such as carbon, was not considered in this calculation. The quantification reveals that 18.4% originates from ACIGS remnants due to peel‐off at the rear interface, with the majority of this contribution attributed to Ga, likely a result of the higher Ga profile implemented at the rear interface, and the Na signal is attributed to unreacted NaF from the pre‐deposition treatment. Considering first the Si 2p and O 1s spectra associated with the SiOx layer, the Si 2p spectrum exhibits two distinguishable peaks: one attributed to SiO2 at 103 eV, and the other to an aluminium loss feature, likely an artifact from the x‐ray source gun. The O 1s spectrum, in turn, shows three peaks, with the most intense attributed to SiO2 at 532.3 eV, while the remaining two are assigned to single and double carbon bonds (O─C and O═C, respectively). Considering the atomic percentages of the two peaks associated with SiO2 (Si 2p and O 1s), a stoichiometric ratio of 2 was calculated, consistent with and confirming the presence of the passivation layer. Second, the Mo 3d spectrum reveals two doublets: one associated with Mo in its metallic state (227.6 eV) corresponding to the rear contact, and the other associated with a Mo–Se bond (228.4 eV). In the Se 3d spectrum, only one doublet is observed, also associated with a Mo–Se bond (54.1 eV). Considering the atomic percentages of the peaks associated with the Mo–Se bond, a stoichiometric ratio of 2.1 was calculated, confirming the presence of the MoSe2 layer. This latter value should be regarded as an estimation, since the average XPS probing depth is approximately 10 nm, while the passivation layer itself is 8 nm thick—meaning only the outermost 2 nm below the SiOx layer were effectively probed. Overall, the XPS analysis suggests an intact SiO2 passivation layer and a rather uniform MoSe2 layer below it. Thus, the high‐performance enhancement can be related to the combination of near‐full passivation, translated into the highest Rshunt and a near 60 mV VOC increase, and the formation of MoSe2 through the pores, leading to an ohmic contact, resulting in the lowest Rs, leading to an average increase of almost 2% absolute efficiency. Conclusions Interface passivation has been a key driving force for performance improvements in solar cells, both for Si and CIGS‐based technologies, with patterned dielectric layers and tunnel oxide layers as key strategies. In the present study, the incorporation of a porous rear passivation layer with full coverage in ultra‐thin ACIGS solar cells is tested. Three passivated samples were developed: two line‐patterned, with around 70% of the passivated area but different SiOx thicknesses, 8 and 25 nm (P8 and P25), and one porous full, i.e., 100% passivation area, layer of 8 nm (F8). The samples were fully characterized by GDOES depth profiling, JV, EQE, and electron microscopy (SEM and STEM). First, GDOES and STEM analysis proved that all passivation architectures survived the harsh ACIGS deposition conditions. Then, performance‐wise, the full‐layer passivation sample (F8) has the highest efficiency, with a 2% absolute value increase and a maximum value of 16.4%. Also, both the VOC and FF were enhanced, which has been associated with a passivation effect since the optical gain is negligible, as seen through EQE and supported by optical simulations. The remaining passivated samples also had performance enhancements, but to a smaller extent. After One‐diode fitting, sample F8 was found to have the highest shunt resistance and the smallest series resistance. In the STEM analysis, both patterned samples have the characteristic MoSe2 of the Mo/CIGS interface in the contact openings. For the samples with an 8 nm passivation layer, patterned (P8) and full (F8), the MoSe2 was found below the passivation layer. A peel‐off procedure of sample F8 allowed a deeper assessment of the rear interface and showed that the full SiOx layer was found uniform, intact, and with no morphological deformation. Moreover, XPS analysis of this rear contact confirmed the existence of SiO2 and MoSe2, both with a stoichiometric ratio of 2. Summarizing, the high performance of the full porous passivated sample was attributed to the near‐complete rear interface passivation, since no defects were found in this layer, combined with the type‐ohmic of Mo/MoSe2(SiOx)/ACIGS across the rear interface. The diffusion of Na and Se through pores of the passivation layer is associated as the most probable cause of the MoSe2 formation. Contributor Information Pedro M. P. Salomé, Email: pedro.salome@inl.int. Jennifer P. Teixeira, Email: jennifer.teixeira@inl.int.