Cu(In,Ga)Se2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$_2$$\end{document}-based solar cells for space applications: proton irradiation and annealing recovery

In this work, we present an experimental study of a Cu(In,Ga)Se2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$_2$$\end{document} (CIGS)-based solar cell (SC), irradiated with protons of energy 80 and 180 keV and with fluences of 1012\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{12}$$\end{document}, 1013\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{13}$$\end{document}, and 1014\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10^{14}$$\end{document} cm-2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^{-2}$$\end{document}, as well as a strategy to recover the induced damage. The possible modifications of the structural, electrical, and optical properties, induced by the proton irradiation, were investigated. Although the irradiation did not promote any major modification in the crystalline structure, it did induce the creation of defects responsible for changes in the electronic structure which caused a partial PL quenching and significant changes in the PL spectral shape, as well as a reduction of the power conversion efficiency and open-circuit voltage of up to 30% as revealed by J–V measurements. The photoluminescence results showed a broadening, redshift and decrease in the signal-to-noise ratio. The recovery of damage induced by irradiation in several SCs was tested through annealing steps performed at different temperatures and time intervals. It was found that the best recovery strategy for the investigated irradiation parameters was carrying out several isothermal annealing at 200 °C for 30 min. This strategy is compatible with the intermittent variation of the temperature in space and allowed to recover a power conversion efficiency comparable to that of the as grown cell. In particular, it must be highlighted that keeping the SC at room temperature in ambient atmosphere and in the dark, did not promote significant recovery in contradiction with some previous reports. This recovery methodology was applied in parallel for non-irradiated SCs and no increase in power conversion efficiency was found, but rather a slight decrease. The dominant radiative recombination channel was, apparently, unchanged with the irradiation and the subsequent recovery process. Nonetheless, changes in the concentration of defects of different types cannot be excluded, which is in line with a significant influence of fluctuating potentials in both as grown and after recovery stages of the solar cell. This work constitutes a first systematic study that simultaneously encompasses the influence of proton irradiation on the optical and electrical properties of CIGS SCs and a damage recovery methodology with a high potential to be explored in space applications. Additionally, it contributes to reinforcing the high potential of CIGS technology in the context of creating constellations of small satellites that are being developed by different entities, particularly private ones.

different SCs technologies [3,4,6,13].For CIGS SCs, some tests have been carried out in satellites, focusing on the analysis of their electrical performance [12,14].Overall, it has been assumed that the CIGS technology does not suffer from degradation when irradiated with high-energy electrons.However, in the case of low and high-energy protons, the degradation is measurable, although being clearly weaker than in Si, GaAs, and multijunction-based SCs [2,[15][16][17][18].Despite the electrical performance of CIGS SCs being reasonably well understood in terrestrial applications, much is unknown concerning the origin of their high radiation hardness, the nature and influence of radiation-induced defects, and the radiation tolerance of the individual SC layers [15,16,19,20].Some studies have shown that, in the case of implantation with lowenergy protons, there is significant degradation of the layers interfaces and when the degradation reaches the depletion region, the cells' performance is significantly affected [14-16, 19, 21].In this context, photoluminescence is a particularly suitable technique for studying the role of defects in the optical properties of cells and, consequently, in the dynamics of charge carriers with the consequent impact on the performance of the devices [15,19,[22][23][24][25].

Introduction
The current trend in the space industry toward the miniaturization of satellites to reduce mission costs implies a reduction in the size and weight of the instruments installed aboard, including the solar cells (SCs).The use of Si and III-V (like GaAs) materials or multijunction SCs in small satellites is disadvantageous due to their bulkiness, resulting in low specific power (W/kg) and low stowability, which are crucial for the figures of merit in any space mission [1][2][3].Though, on Earth, Si SCs show a clear dominance of the photovoltaic market over the thin film-based SCs, they have a low tolerance to radiation in space environment.Indeed, already with the launch into space of the first satellite equipped with a SC in 1958, the detrimental effects of cosmic radiation were identified as a major issue [4].This fact motivated the search for combining high power conversion efficiency ( ) with high radiation tolerance.Nowadays, Cu(In,Ga)Se 2 (CIGS) SCs are one of the best technological options for space applications since they combine high efficiency (23.35%), comparable to that of polycrystalline Si (24.4%) and close to that of monocrystalline Si, with a radiation tolerance higher than of Si, GaAs or multijunction-based SCs [2,[5][6][7][8][9][10].
Besides high-energy photons, the orbital radiation environment includes fast electrons, protons, neutrons, and heavy ions, e.g., ejected by solar events and later trapped by the Earth's magnetic field or transiting through the domains of artificial satellites [3,11,12].In the case of SCs installed externally on the satellites, their location makes them particularly exposed to radiation damage.In recent decades, various studies focused on evaluating the radiation hardness of recovery in contradiction with some previous reports.This recovery methodology was applied in parallel for non-irradiated SCs and no increase in power conversion efficiency was found, but rather a slight decrease.The dominant radiative recombination channel was, apparently, unchanged with the irradiation and the subsequent recovery process.Nonetheless, changes in the concentration of defects of different types cannot be excluded, which is in line with a significant influence of fluctuating potentials in both as grown and after recovery stages of the solar cell.This work constitutes a first systematic study that simultaneously encompasses the influence of proton irradiation on the optical and electrical properties of CIGS SCs and a damage recovery methodology with a high potential to be explored in space applications.Additionally, it contributes to reinforcing the high potential of CIGS technology in the context of creating constellations of small satellites that are being developed by different entities, particularly private ones. in particular Cu vacancies [27].It must be noted that the CIGS used for photovoltaic applications is Cu-poor with a [Cu]/([Ga]+[In]) ratio in the range 0.88-0.92[28,29].Also, the [Ga]/([Ga]+[In]) ratio changes in depth [30].This significant deviation from stoichiometry and intentional change of composition during the growth, results in the formation of many types of acceptor and donor defects, being assumed in the literature that the Cu vacancy ( V Cu ) is the one responsible for the p-type behavior [31,32].Such a high density of defects promote a sufficiently high interaction between them and the semiconductor is heavily doped and strongly compensated.The expected doping level should be, at least, in the order of 10 18 cm −3 [22,25].Given the many types of defects in the lattice and the high values of densities present, the discussion about the identification of the types of defects in the lattice is very difficult to carry out.
Recent studies revealed that the degradation caused by irradiation in CIGS SCs can be partially reversed by keeping the cell at room temperature or by annealing steps and they recover from proton irradiation damage more quickly than other types of SCs [14,17,26,[33][34][35].Khatri et al. [14] suggested that shallow-type defects at the surface created by small proton fluxes can be compensated for by annealing under illumination, although this is not the case of deep defects generated by a high proton flux.Despite those results suggesting the SCs' recovery being dependent on cell illumination and annealing temperature, the few studies of cells irradiated with low-energy protons focus on the analysis of electrical parameters measured at room temperature and considering limited time intervals of annealing.In particular, no systematic studies addressing the optical properties of SCs that explore the defect repair of CIGS [18,36], have been conducted.
In the present work, a set of seven CIGS-based SCs, with standard architecture, were irradiated with protons with energies of 80 and 180 keV and with fluences of 10 12 , 10 13 and 10 14 cm −2 .The induced proton damage was evaluated through structural, electrical, and optical measurements involving X-ray diffraction (XRD), Raman spectroscopy, current density-voltage (J-V), and photoluminescence (PL).The recovery from damage was investigated through annealing steps at room temperature and in the temperature range 100-200 °C.A strategy involving the annealing of the SCs in ambient atmosphere and in the dark, at moderately high temperatures, showed a notable recovery of the power conversion efficiency and optical properties after annealing for 30 min at 200 °C.

SCs preparation
A set of eight CIGS-based SCs were grown at the Ångström Laboratory, with a linear Ga profile and without any antireflective coating or post-deposition alkali treatment, according to the cell architecture: soda lime glass/Mo/CIGS/CdS/i-ZnO/ZnO:Al/Ni-Al-Ni grid.More details can be found elsewhere [37].

Physical characterization
The X-ray diffractograms were measured under the Bragg-Brentano geometry with a PANalytical Philips X'Pert MPD diffractometer using the Cu-K line ( = 1.54060Å).Indexation of the diffractograms was performed using the International Centre for Diffraction Data (ICDD) database.
Raman spectroscopy measurements were performed on a Jobin-Yvon LabRam HR 800 UV spectrometer in the backscattering configuration.The spectrometer was equipped with an Olympus BX-41 microscope and a Peltier-cooled CCD detector (Andor), and the measurements were performed with an objective of 100× magnification, with numerical aperture of 0.9, and a 2400 lines mm −1 grating.A solid state laser (Material Laser Quantum) with a wavelength of 532 nm was used.All spectra were corrected considering the optical phonon of bulk Si at 521 cm −1 .
The SCs performance was characterized by J-V measurements under standard AM 1.5 illumination, using a Sciencetech class 3A solar simulator.
PL measurements were carried out using a Bruker Vertex 80v Fourier Transform Infrared (FTIR) spectrometer equipped with InGaAs and InSb detectors.The SCs were inserted in a nitrogen gas flow cryostat (OptistatCF, Oxford Instruments) which allowed to change the temperature (T) in the range 70-300 K.The excitation source was the 532 nm laser line (MGL-F-532, CNI), and the laser power (P) was measured at the front of the cryostat window by means of an optical power meter (842-PE, Newport), equipped with a Si detector (818-SL, Newport).The diameter of the laser spot was approximately 1 mm.

Proton irradiation
The proton irradiation was carried out in a Danfysik 1090 implanter at the Laboratory of Accelerators and Radiation Technologies (LATR) of Instituto Superior Técnico [38].Two energies for the protons were used, 80 and 180 keV, in order to locate the implantation profile of protons and, consequently, to maximize the damage, near the pn junction or inside the CIGS layer, respectively.The expected positions of the implanted protons were simulated with the SRIM (Stopping and Range of Ions in Matter) software [39] (see Fig. 1 in Supporting Information for additional details).For both energies, three fluences ( 10 12 , 10 13 and 10 14 cm −2 ) were investigated.With the exception of one SC (#12), all the other seven SCs were subjected to irradiation with protons (see Supporting Information for the irradiation parameters of each SC).

Annealing treatments
The annealing steps were carried out by placing the SC in an oven previously heated to the desired temperature, under the ambient atmosphere and in the dark.After removing the SC from the oven, the temperature measured on the front surface (optical window) of the SC, showed that for all annealing steps, the time required to reach room temperature was ∼ 30 s.

Results
The results presented in this work, will be mainly focused in SC#13, which was irradiated with protons of 180 keV with a fluence of 10 13 cm −2 and, occasionally, results obtained for the remaining irradiated SCs will be discussed.Additionally, a nonirradiated SC (#12) was used as a reference SC for the discussion of the recovery methodology.

Irradiated SCs
The electrical properties of the as grown SC#13 were investigated through J-V measurements (see Fig. 1a and Table 1).After proton irradiation, the electrical evaluation was repeated.The estimated parameters show a reduction of the short-circuit current density ( J SC ), the open circuit voltage ( V OC ), power conversion efficiency ( ) and ideality factor ( n id ) by ∼16%, ∼27%, ∼30% and 39%, respectively, whereas the fill factor (FF) increased by ∼14%.Clearly, V OC , and n id are the electric parameters most affected by the irradiation.
The potential structural changes in the SC, induced by the proton irradiation, were evaluated by XRD and Raman spectroscopy measurements.Figure 2a shows the XRD curves before and after the irradiation with    101) planes at the ZnO layer.Additionally, there is a diffraction peak at 2 ∼ 33.0 • , denoted by ⋆ in Fig. 2a, which is likely due to the forbidden Si (200) reflection [41] from the Si sample holder used in the diffractometer.The comparison of the diffractograms measured before and after irradiation does not evidence significant changes.The only exception is the peak ascribed to the forbidden reflection from the Si sample holder, due to possible changes in cell assembly compared to previous measurement.Figure 2b shows the results obtained by Raman spectroscopy before and after irradiation.Both spectra show peaks at ∼ 176 cm −1 , ∼ 218 cm −1 and ∼ 301 cm −1 .As reported in the literature [42,43], these peaks are associated with the A 1 (CIGS), B 2 ∕E (CIGS), and LO (CdS) vibration modes, respectively.No significant changes induced by irradiation are observed from the inspection of the vibrational modes.Therefore, the global structural analysis based on the XRD and Raman spectroscopy measurements supports the absence of any sizeable modification in the crystalline structure of the inspected individual layers, induced by the proton irradiation.
Regarding the possible modifications in the electronic structure of the CIGS layer, PL measurements were performed.Figure 1b shows the superposition of two normalized PL spectra of the SC#13 measured at 70 K, before and after irradiation, under the same excitation power ( ∼7 mW).A significant shift to lower energies of the luminescence, a strong increase of the full-width-at-half-maximum (FWHM), and a notable decrease in the signal-tonoise ratio (SNR), are observed after irradiation.A similar behavior was observed for SCs irradiated with both energies in all fluences investigated, i.e., SCs#1, #2, #4, #5, #7, #8 (see Figs. 2 and 3 in Supporting Information).Therefore, the degradation in the luminescence reported in Fig. 1b is reproducible for the whole range of irradiation parameters studied in this work.It is important to emphasize that the value of the SC#13 also diminished from 10.2 to 7.2% after irradiation (see Table 1).Therefore, these facts suggest that the poorer electrical performance of the SC after irradiation is related to the creation of defects in the CIGS layer, some of them deep in the bandgap, which are involved in both radiative and nonradiative recombination channels that contribute to a lower collection of charge carriers.Those radiative channels are responsible for luminescence at lower photon energies in comparison with the ones measured in the as grown stage.These results are not in contradiction with the absence of changes in the crystalline structure since optical and electrical properties are much more dependent on the density of defects than the structural ones.Also, they are in line with a proton implantation profile located in the CIGS layer of the SC as shown by the SRIM simulations (see Figure 1 in Supporting Information).Following the above results for SC#13, irradiation caused a decrease in the performance of the SC cell, which was accompanied by a remarkable change in luminescence originating from the CIGS layer.It is therefore important to identify strategies that may result in the recovery of the SCs' performance.In this sense, a study was developed in two distinct phases.Firstly, we focused on identifying the annealing parameters that maximize the recovery of the luminescence properties after annealing in comparison with the measured luminescence in the as grown stage.The modifications on the electronic structure induced by annealing steps at different temperatures (including room temperature), for different time intervals, were investigated through PL measurements (see Supporting Information).This phase was carried out on six irradiated SCs (SC#1, #2, #4, #5, #7, #8)).Secondly, we applied the experimental conditions identified in the first phase to tentatively recover the SC#13 performance, evaluating the electrical and optical properties after each annealing step.
As shown in Figs. 4 and 6 in Supporting Information for SC#5 and SC#2, a few annealing steps at 100 • C and also at room temperature, in ambient atmosphere and in the dark, for different time intervals, resulted in marginal changes of the shape of the luminescence and an increase of the SNR.Nevertheless, the luminescence after those steps still resembles the luminescence measured after irradiation which undoubtedly shows that the electronic structure of the CIGS layer continues to be strongly influenced by energy levels related to defects created by irradiation.Increasing the annealing temperature to 150 • C (Fig. 6 in Supporting Information for SC#2) showed a significant change in the luminescence shape, evidenced by a blueshift and a reduction of the FWHM.The shape of the luminescence obtained is similar to that measured in the as grown stage.However, the peak energy of the luminescence remained below the value observed in the as grown state and the SNR is low.Therefore, despite a decrease in the influence of defects in the electronic structure of the CIGS layer, these results suggests that a significant fraction of the defects created by irradiation are still present in the lattice and are responsible for keeping nonradiactive recombination channels that compete for the photoexcited charge carriers.A further increase of the annealing temperature to 180 • C did not improve significantly the optical properties of the SC.On the other hand, the annealing at 200 • C of the SC#5 and SC#2 (Figs. 5 and 6, respectively, in Supporting Information), showed that the damage recovery is clearly greater since a small FWHM and a high SNR were obtained as well as the shape and peak energy of the luminescence become closer to the observed in the as grown state.Finally, the annealing at 240 • C of the SC#5 (Fig. 5 in Supporting Information) proved to be harmful since it led to the appearance of luminescence for energies lower than the energy range observed in the as grown state, which suggests the creation of sufficiently deep recombination channels in the bandgap.
The above study, shows that the recovery of the optical properties is greater when the annealing of the SC is carried out in ambient atmosphere and in the dark at 200 • C for approximately 30 mins.Hence, the recovery of SC#13 was tentatively promoted through four annealing steps carried out at 200 • C for 30 mins each one.The performance of the SC and the luminescence properties were evaluated after each step.Figure 3a shows a superposition of the J-V curves measured after each annealing step.The corresponding curves for the as grown and after the irradiation stages are also included for comparison.The electrical parameters extracted from the measured J-V curves are presented in Table 1.While the annealing of the SC resulted in almost no change of the J SC , the V OC evidenced a significant increase ( ∼110 meV) with the first annealing and a lower increase after the second and third annealings.After this stage, the V OC did not improved further.As for the values, a continuous increase was obtained reaching, after the fourth annealing, a value comparable with the one for the as grown stage.With annealings, FF presents an initial decrease followed by a continuous increase, reaching a higher value than the initial one.Regarding the ideality factor, an initial increase followed by a continuous decrease, is registered.It must be mentioned that none of the SC performance parameters recovered to the initial values except for .
As discussed above, the degradation of the electrical performance of the SC induced by the irradiation was accompanied by a significant change of the luminescence's shape and SNR (see Fig. 1b).The normalized PL spectra measured after each annealing are presented in Fig. 3b.It is observed that a single annealing at 200 • C for 30 min is enough to promote significant changes in the luminescence properties, notably a blueshift, an increase of SNR, and a reduction of the FWHM.Actually, the shape of the emission is close to the one observed in the as grown stage.After the second annealing, the PL spectra emission blueshifts even further.After the third and fourth annealings, the peak energy ( E p ) of the luminescence is very close to the value obtained after the first annealing suggesting the stabilization of the E p value.However, despite the shift to higher energies with annealing, the value of E p is still lower than the value observed in the as grown stage.A similar behavior was described for the V OC parameter which suggests a correlation between V OC and E p .
In order to better understand the changes in the recombination channels accomplished by the irradiation and subsequent annealing steps, a more indepth PL study was carried out for SC#13 at the as grown and after the fourth annealing stages.Before proceeding, it is important to note that the deconvolution of radiative transitions after the irradiation stage in SC#13, was not possible due to two factors: (1) the luminescence spectrum spans a broad range of energies that extends to energies beyond the detection range of the InGaAs detector, as shown in Fig. 2b in Supporting Information through the measurement of the luminescence with the InSb detector; we must note that the sensitivity of the InSb detector is much lower than the one of the InGaAs detector which prevents its use in dependencies with the excitation power and temperature for the SCs studied in this work; (2) the shape and width of the luminescence does not allow a correct identification of the radiative transitions present in the luminescence; Fig. 7c in Supporting Information shows a tentative fitting model involving seven Gaussian functions; nevertheless, other models with different number of components, describe reasonably well the measured luminescence, thus preventing a correct deconvolution of the radiative transitions.Therefore, there is no physical support to fix a fitting model and proceed with the analysis of the luminescence after irradiation.
The deconvolution of the luminescence in the as grown and after the fourth annealing stages with Gaussian functions, was performed according to the procedure described in Supporting Information (Fig. 7a, b).Two and three radiative transitions were identified in the as grown and after the fourth annealing stages, respectively.The relative intensity of the non dominant radiative transitions is low which is reflected in a higher uncertainty on the peak energy and intensity as compared to the dominant transition.As a consequence, the following analysis is focused on the dominant transition in each of the two stages of the SC.The resulting peak energy and PL integrated intensity of this transition are plotted as a function of the normalized excitation power in Fig. 4. With the increase of excitation power, a blueshift of the luminescence and an increase of its intensity, are observed.The peak energy as a function of excitation power can be fitted by the equation [44][45][46][47][48]: where is a coefficient that parameterizes the energy shift and P 0 is a fitting parameter.As shown in Table 2, the estimated values of were 9.5 ± 0.1 and 9.0 ± 0.03 meV for the as grown and after the fourth annealing stages, respectively, which are close to each other.Such high values are strong hints to radiative recombinations in a highly doped and strongly compensated semiconductor [37,44,45,[49][50][51].Thus, the electronic structure is influenced by fluctuating potentials and the density of states involved in the radiative recombination channels is low.
Regarding the dependence of the PL integrated intensity, the following power law is commonly used [37,45,47,48,52]: where m is an adjustable parameter.The estimated m values, 1.40 ± 0.02 and 1.15 ± 0.01 for the as grown and after the fourth annealing stages, respectively, suggest a higher localization of the charge carriers in the lattice after the damage recovery.
The temperature dependence of the luminescence was also investigated in the range 70-300 K. Figure 5a, b shows the dependencies of the peak energy and PL integrated intensity, respectively, of the dominant radiative transition.As the temperature is increased,

(a) (b)
the luminescence measured in the as grown stage blueshifts by 29 meV, whereas after the fourth annealing stage evidences an apparent redshift followed by a blueshift of 23 meV (see Table 2).Such large shifts are common in highly doped and strongly compensated semiconductors [24,37].
Concerning the dependence involving the PL integrated intensity (see Fig. 5b), as the temperature is increased, in both stages of the SC, a progressive extinction of the luminescence is observed.The experimental behavior was fitted with the equation [53]: where k is the Boltzmann constant, and I 0 is the PL integrated intensity at 0 K.The second term describes a type of de-excitation channel involving discrete high energy levels.The energy difference between an hypothetical high energy (nonradiative) level and the radiative level is E i and the parameter c i corresponds to the degeneracy ratio between the high energy state and the radiative state.The third term describes another type of nonradiative de-excitation channel, one involving an energy band.The activation energy of this channel is E bx and the parameter c bx T 3 2 accounts for the temperature dependence of the effective state density of the band involved, being c bx a fitting parameter.The best fitting model for the as grown stage consists in a single de-excitation channel involving a band, with an estimated activation energy of 72.3 ± 0.2 meV.No evidence for a de-excitation channel involving any discrete energy level was found.A similar model was identified for ( 3) , the thermal quenching of the luminescence after the fourth annealing, and the estimated activation energy was 60 ± 2 meV.These two activation energy values are associated with a type of de-excitation channel that involves the release of a charge carrier from the radiative state into an energy band.The proximity between the two activation energy values must be noted.

Non-irradiated SC (#12)
In order to deepen the discussion of damage recovery in SC#13, a non-irradiated SC (#12) was annealed, first to 150 • C, and then to 200 • C, in ambient atmosphere and in the dark, for 30 mins.SC#12, as all other SCs used in this study, belongs to the same growth set.The chosen temperatures correspond to those at which a clear recovery of luminescence properties was observed in the various irradiated SCs studied in this work (see discussion above and Supporting Information), therefore determinant for the changes in the electronic structure of the CIGS layer.Performance and photoluminescence were evaluated in each of the SC stages: as grown, after annealing at 150 • C; after a second annealing at 200 • C.
From the J-V measurements (see Fig. 6a and Table 3), both annealing steps revealed detectable variations on the electrical parameters.After the annealing at 150 • C, it was observed the maintenance of the value of J SC and , a decrease of ∼ 4% of V OC and of 15% of n id , and an increase in FF of 2%.After annealing at 200 • C, there were small decreases of J SC , V OC , and of 2%, 1%, and 3%, respectively, whereas FF remains practically unchanged and n id increased 6%.After the two annealing steps, the results showed a continuous

(a) (b)
decrease of both (0.6%) and V OC (30 mV).In the case of J SC , n id and FF, no particular trends were identified.These results are compatible with a small increase of the total density of defects on the CIGS layer with the two annealing steps.Figure 6(b) shows a superposition of normalized PL spectra measured in the three stages of the SC, at 70 K, under the same excitation power ( ∼7.4 mW).The experimental behavior reveals a shift in luminescence toward lower energies, more significant for the first annealing at 150 • C. With regard to the shape of the luminescence, after the first annealing, no change is perceptible, while after the second annealing, an increase in the asymmetry of the luminescence is observed for lower energies.Both facts, redshift and asymmetry, are strong hints that both annealing steps promoted a change of the total density of defects in the CIGS layer, compatible with an increase of this density [22,25,37,47,54].This interpretation is in line with the results of the electrical characterization of the SC, since a tendency of reduction of the values of and V OC was observed with the two annealing steps.
In each of the SC stages, the dependence on the excitation power of the luminescence was studied, which allows us to evaluate in greater depth the changes in the total density of defects.Following a similar approach discussed for SC#13, the deconvolution of the luminescence with Gaussian functions was performed in order to identify the possible radiative transitions (see Fig. 8 in Supporting Information).Two radiative transitions were identified in the three stages of the SC, which is comparable to the deconvolution in the as grown stage of SC#13.The luminescence is dominated by one transition, in which the following analysis will be focused.Figure 7a, b show the corresponding dependencies of the peak energy and PL integrated intensity, respectively.With increasing excitation power, a strong blueshift is observed in the three stages of SC.Fitting Eq. ( 1) to the experimental data gave values of 9.5 ± 0.1 , 8.5 ± 0.1 , and 10.8 ± 0.2 for the as grown, after annealing at 150 • C, and after annealing at 200 • C stages, respectively (see Table 4).The value for the as grown stage is the same as the estimated one for the equivalent stage of SC#13, which is expected since both SCs were manufactured from the same growth batch.On the other hand, a small decrease in the value of was obtained after the first annealing and a significant increase after the second annealing.This increase is compatible with the increase in asymmetry observed at low temperature, discussed earlier.
With regard to the behavior of the PL integrated intensity (see Fig. 7b), Eq. ( 2) was fitted to the experimental data in each stage of the SC.The estimated m values were 1.33 ± 0.01 , 1.18 ± 0.01 , and 1.20 ± 0.01 for the as grown, after annealing at 150 • C, and after annealing at 200 • C stages, respectively.Although the changes in m are not significant, apparently, there is a tendency for its value to decrease after annealing, which may indicate a greater localization of charge carriers after the annealing steps.It should be noted that the values of 1.33 ± 0.01 and 1.20 ± 0.01 are close to the ones estimated for the luminescence of SC#13 at the as grown and after the fourth annealing stages, respectively.
Overall, the above results for the non-irradiated SC, confirmed the potential of annealing at temperatures of 150 and 200 • C to promote changes of the density of defects in the CIGS layer, as the electrical performance evaluation in this SC, suggested.

Discussion
In this work, a set of seven SCs was irradiated with protons of energies 80 and 180 keV with fluences of 10 12 , 10 13 and 10 14 cm −2 .For all energies and fluences, Table 3 Short-circuit current density ( J SC ), open circuit voltage ( V OC ), fill factor (FF), efficiency ( ) and ideality factor ( n id ) values for the non-irradiated SC#12 in the as grown, after the first annealing ( 150 • C), and after the second annealing ( 200 • C) stages irradiation was found to cause a significant deterioration of the optical properties.The recovery of these properties was tested through annealing steps in ambient atmosphere and in the dark at temperatures in the range 100-240 • C. It is important to note that keeping an irradiated SC for 134 days at room temperature, in ambient atmosphere and in the dark, was not reflected in the recovery of the cell's optical properties.A similar behavior was observed after keeping a SC, previously irradiated and annealed at 100 • C, at room temperature for 19 days.The temperature of 200 • C was identified as the one that maximizes the recovery of the luminescence properties.These results seems not be in line with Boden et al. [34] that showed an increase of ∼ 10 % in the efficiency for a CIGS SC irradiated with protons of 100 keV in a fluence of 10 13 cm −2 , and also of Kawakita et al. [26] that reported a recovery of the short-circuit current of CIGS SC irradiated with protons of 3 MeV with a fluence rate of protons of 3 × 10 11 cm −2 s −1 , under light illumination.A deep investigation of both electrical and optical properties was performed for a SC irradiated with low energy protons (180 keV) to maximize the damage in the CIGS layer, using a fluence of 10 13 cm −2 , which is much higher than the annual exposure in space.As a comparison, we can refer that a typical equipment in a low earth orbit (LEO) mission is exposed to an integrated fluence of protons with energy below 1 MeV of ∼ 10 9 cm −2 per day [11], meaning that an irradiation with 10 13 cm −2 simulates about 30 years in space.The main goal with the energy and fluence used, is to accelerate in the laboratory, the degradation process of a CIGS-based SC, concentrating the damage mainly on the CIGS layer [34].The structural study by XRD and Raman measurements gave no evidence concerning a measurable structural damage induced by the proton irradiation.Conversely, the electronic structure was clearly affected by the proton irradiation, as shown by the degradation of the electrical performance of the SC as well as by the strong modification of the luminescence with origin in the CIGS layer.This point will be further discussed below.
With the proton irradiation (180 keV, 10 13 cm −2 ), the damages in the SC were reflected in a decrease in J SC and, mainly, in V OC in line with the literature [34].Particularly, in the case of V OC , this is an expected result since this parameter critically depends on the defect density in the CIGS layer [25].Thus, the changes in V OC and J SC confirm a significant creation of defects with irradiation.It is interesting to note the increase in FF despite the decrease in efficiency.Clearly, the observed behavior of J SC and V OC , had a dominant and harmful effect on cell performance.Nevertheless, the SC was still working which suggests that the irradiation with protons for the above parameters, does not critically disturb the charge collection mechanisms in the SC to the point of preventing the cell from functioning.Also, taking into account the implantation profile for the energy of 180 keV (see Supporting Information), we speculate that the irradiation may have created damage at the interfaces of the top layers of the SC and a concentration of defects in the depletion region, regardless of a maximized damage inside the CIGS layer [14-16, 19, 55, 56].
A close relationship between the previous reported changes in the electrical performance of the SC, in particular of the V OC , and the radiative channels involved in the luminescence processes, was found.With the irradiation, a reduction of up to about 30% was obtained in and V OC , in accordance with the literature [2,14,17,57], which was accompanied by a broadening, redshift of the PL and a reduction of the SNR.Clearly, irradiation created defects in the CIGS lattice involved in new recombination channels.A significant fraction of the new channels are radiative and involve deep energy levels in the bandgap as shown by the significant increase in FWHM, which resulted in an extension of the luminescence to lower energies.Such high number of new radiative channels hindered a physically meaningful analysis of the luminescence in the stage after irradiation of the cell.Nevertheless, the decrease in SNR shows a higher competition for the photogenerated charge carriers, thus allowing us to assume that most of the induced recombination channels are nonradiative.All these channels will negatively influence the charge collection, but not hinder the partial operation of the SC as shown by the J-V measurements.
Despite the changes in cell performance, it was found that the damage caused by irradiation is not irreversible and that it is possible to promote its recovery through annealing steps.Indeed, a single annealing at 200 • C for 30 mins led to an increase in both V OC and and a drastic change in luminescence, compatible with the partial recovery of the CIGS lattice from radiation damage.Furthermore, it was observed that additional annealing steps with the same parameters allow to further converge to the value prior to irradiation, which shows that cell operation can be recovered to a performance comparable to the initial value.
The annealing of the non-irradiated SC at 150 and 200 • C showed a measurable impact on its electrical and optical properties, namely on the V OC , efficiency, and peak energy of the luminescence, which evidence that this type of treatment has the potential to change the total density of defects in an as grown SC.The observed changes point to an increase of the total density of defects.Interestingly, comparable annealing steps performed on irradiated SCs, showed an opposite effect, i.e., a reduction of the total density of defects, based on two experimental results: (1) increase of V OC and efficiency; (2) quite strong reduction of the FWHM of the luminescence, its blueshift and the increase of the SNR.Therefore, the induced effects promoted by the annealing steps depends critically on the starting point in terms of the density of defects and dominant type of defects present on the CIGS layer.Indeed, this work shows that although annealing at 150 and 200 • C can be harmful for SCs, they are clearly useful for recovering from radiation damage.Additionally, the difference in behavior observed for irradiated and non-irradiated cells with annealing, shows that irradiation has a decisive effect on the SC, in accordance with the electrical and optical measurements.
We speculate that the cause of such behaviors is related with possible metastable defects in each case.After irradiation, a significant fraction of metastable defects can be created in the CIGS and the annealing steps at a proper temperature are able to promote a global relaxation of the layer.This is compatible with the results of Khatri et al. [14] regarding the possible compensation of shallow-type defects at the surface.In the case of a non-irradiated CIGS, the annealing steps can promote the creation of other type of defects, some of which can also be metastable.Nevertheless, this possibility requires additional studies that are out of the scope of the present work.
The analysis of the evolution of the J SC , V OC and n id values, as well as of the E p of the luminescence, with irradiation and annealing steps, shows that the values of these parameters are not the same as before irradiation.Thus, these results suggest that the types of defects and the overall defects density are close but not the same as before irradiation, which certainly reflects the high density of defects characteristic of the Cu-poor CIGS films used in SCs, in accordance with the high estimated values.As a consequence of this property of CIGS, on the one hand, the creation of defects in the lattice does not prevent the partial operation of the SC and, on the other hand, it allows the material to adapt to a new equilibrium when damage recovery is promoted.The various annealing steps showed that there is a significant increase in FF, which is accompanied by a reduction in the ideality factor ( n id ), as shown in Table 1.This is compatible with restoring the dominant charge transport mechanisms in the cell.With annealing, the drastic change of the shape of the luminescence is a consequence of a diminishing of density of defects that should increase the shunt resistance and reduce the series resistance.Another aspect that may contribute to the variation observed in the FF, is the location of the defects.After irradiation, the defects may be deeper in the CIGS, as suggested by the simulation of the implantation profile of protons with energy of 180 keV (see Supporting Information) and by the reduction in n id .The first annealing can promote diffusion of defects toward the depletion zone, justifying the decrease in the FF.However, the modification of the distribution of defects in the lattice ends up favoring the increase of V OC , being this contribution determinant for the small increase in cell performance.Also, the increase of the SNR of the luminescence is compatible with a partial suppression of nonradiative defects quite efficient in capturing photogenerated charge carriers.The continuous increase of the FF with successive annealing steps, supports this interpretation.
Although, electrical measurements suggested that the SC state after recovery is different from the state before irradiation, the PL results showed that the heavily doped character of CIGS is not affected after the recovery process from radiation damage as observed by the estimated values.Indeed, the excitation power dependence of the luminescence showed a low density of states involved in the recombination channels in both the as grown and after the fourth annealing stages of the SC.Also, the thermal activation energy of the dominant nonradiative de-excitation channel for the as grown cell is similar to the one obtained after the annealing of the irradiated cell, which suggests the involvement of the same dominant radiative state in both stages of the SC.
It was found that the study of recovery through the evaluation of PL gives a strong indication regarding the performance of the solar cell.To the best of our knowledge, the type of approach presented here, which shows and explores the proximity between electrical performance and luminescence in studying the radiation hardness of CIGS SCs, is absent from the literature and is in line with a recent study of the electronic structure of this type of SCs [25].
The results obtained in this work show that the CIGS cell recovers from damage caused by irradiation with protons of 180 keV, when subjected to thermal cycles at 200 • C. Taking into account that the studied proton fluences are clearly superior to those that satellites are exposed, and that in orbit satellites are subject to repetitive cycling of temperatures whose range is greater than 150 • C, this work contributes to the establishment of strategies in space for recovering from radiation damage, thus allowing the extension of the operability of satellites.

Conclusions
This work represents a systematic and innovative study of the influence of proton irradiation on the optical and electrical properties of CIGS SCs, as well as the development of a SC recovery strategy based on annealing.The damage due to proton irradiation to which a CIGS SC is subjected to in orbit was simulated with irradiation with protons of low energy, 80 and 180 keV, and with fluences of 10 12 , 10 13 and 10 14 cm −2 .A standard architecture of the CIGS SC was investigated.The irradiation did not promote any measurable modification of the crystalline structure.The protons created defects that modified significantly the electronic structure of the SC that led to a reduction of the power conversion efficiency and V OC as well as to a severe modification of the radiative recombination channels.Keeping the irradiated SCs at room temperature, in ambient atmosphere and in the dark, for tens of days or annealing the SC at temperatures other than 200 • C, did not prove to be decisive in the recovery of the cells' optical properties in contradiction with previous works.In fact, it was found that the recovery of the radiation damage can be done through isothermal annealing steps of the SC at 200 • C for 30 mins, thus accelerating the defect repair mechanisms in CIGS.This beneficial effect of annealing steps, noticed for irradiated cells, is contrary to the harmful outcome observed in the case of non-irradiated cells.It is worth noting that objects in space, such as satellites, are under large thermal amplitude and that 200 • C is close to the maximum temperature reached in orbit.Thus, the intermittent temperature variation experienced by CIGS cells in space contributes significantly to its recovery from the damage caused by proton irradiation.
Along the irradiation and recovery of the cell, a correlated behavior between V OC and the luminescence properties was observed.The results showed that the electronic structure of CIGS, after irradiation and damage recovery, is that of a strongly doped and compensated material and also that the dominant radiative recombination channel apparently does not undergo changes with the several treatments of the SC.Nevertheless, evidence was found that the concentration of the different types of defects in the as grown and after recovery stages are distinct, reflecting both the highly doped and compensated nature and the self-healing characteristic of the CIGS.
A comprehensive understanding of the response of CIGS SCs to proton irradiation and their recovery requires further studies to investigate the effect of higher proton energies, covering the whole range of energies found in space.It is also important to develop recovery strategies compatible with proton energies besides the ones addressed in this work.Our results show the potential of using CIGS SCs in the space environment in which the damage caused by irradiation with protons, and its recovery, are decisive in the performance of the devices.

Figure 1 a
Figure 1 a J-V curves and b normalized PL spectra of the SC#13 measured before (orange) and after irradiation (blue) with protons of 180 keV with a fluence of 10 13 cm −2 .The spectra were measured at 70 K with an excitation wavelength of 532 nm and an excitation power of ∼ 7 mW.

Figure 2 a
Figure 2 a XRD diffractograms and b Raman spectra of the SC#13 before (orange) and after (blue) irradiation with protons with energy of 180 keV and with a fluence of 10 13 protons cm −2 .The peak marked in (a) with ⋆ is ascribed to the forbidden Si(200) reflection from the sample holder used in the experimental setup.

Figure 3 a
Figure 3 a J-V curves and b normalized PL spectra of the SC#13 measured before and after irradition, as well as after the annealing steps.The PL spectra were measured at 70 K with an excitation wavelength of 532 nm and an excitation power of ∼ 7 mW.

Figure 4
Figure 4 Excitation power dependence of the a peak energy and b PL integrated intensity of the dominant radiative transition observed in the as grown and fourth annealing stages of the SC#13.The excitation power is normalized to the maximum

Figure 5
Figure 5 Temperature dependence of a peak energy and b PL integrated intensity for the dominant radiative transition observed in the as grown and annealed stages of the SC#13.The solid lines in (b) illustrate the fits of Eq. 3 to the experimental data.

Figure 6 a
Figure 6 a J-V curves and b normalized PL spectra of the SC#12, non-irradiated, measured in three different stages: as grown (orange), after annealing at 150 • C (citrus green), and after annealing at 200 • C (violet).The spectra were measured at 70 K with an excitation wavelength of 532 nm and an excitation power of ∼ 7.4 mW.

Figure 7
Figure 7 Excitation power dependence of the a peak energy and b PL integrated intensity of the dominant radiative transition observed in three stages of the SC#12: as grown, after annealing at 150 • C, and after annealing at 200 • C. The excitation power

Table 1
• are due to reflections from the (110) and (211) planes in the Mo layer and the peaks at 2 ∼ 34.4 • and 36.4 • result from reflections in the (002) and (

Table 2
Values of , m, energy shift ( ΔE ) of the luminescence with the increase of temperature, and activation energy E bx estimated for the irradiated SC#13 in the as grown and after the fourth annealing stages

Table 4
and m values estimated for the non-irradiated SC#12 in the as grown, after the first annealing ( 150 • C), and after the second annealing ( 200 • C) stages