Proton irradiation of CdTe thin film photovoltaics deposited on cerium‐doped space glass

Space photovoltaics is dominated by multi‐junction (III‐V) technology. However, emerging applications will require solar arrays with high specific power (kW/kg), flexibility in stowage and deployment, and a significantly lower cost than the current III‐V technology offers. This research demonstrates direct deposition of thin film CdTe onto the radiation‐hard cover glass that is normally laminated to any solar cell deployed in space. Four CdTe samples, with 9 defined contact device areas of 0.25 cm2, were irradiated with protons of 0.5‐MeV energy and varying fluences. At the lowest fluence, 1 × 1012 cm−2, the relative efficiency of the solar cells was 95%. Increasing the proton fluence to 1 × 1013 cm−2 and then 1 × 1014 cm−2 decreased the solar cell efficiency to 82% and 4%, respectively. At the fluence of 1 × 1013 cm−2, carrier concentration was reduced by an order of magnitude. Solar Cell Capacitance Simulator (SCAPS) modelling obtained a good fit from a reduction in shallow acceptor concentration with no change in the deep trap defect concentration. The more highly irradiated devices resulted in a buried junction characteristic of the external quantum efficiency, indicating further deterioration of the acceptor doping. This is explained by compensation from interstitial H+ formed by the proton absorption. An anneal of the 1 × 1014 cm−2 fluence devices gave an efficiency increase from 4% to 73% of the pre‐irradiated levels, indicating that the compensation was reversible. CdTe with its rapid recovery through annealing demonstrates a radiation hardness to protons that is far superior to conventional multi‐junction III‐V solar cells.


| INTRODUCTION
The Centre for Solar Energy Research (CSER), Swansea University in collaboration with the University of Surrey has developed a thin film cadmium telluride (CdTe) solar cell technology for use in Space. [1][2][3] Working with industrial partners Qioptiq Space Technology (QST) and Surrey Satellite Technology Ltd, this solar cell technology is designed to meet the emerging demands of new space applications.
There are currently over 2000 operational satellites in Earth 0 s orbit, with power requirements from a few Watts to 10 0 s of kW in the most part satisfied by solar photovoltaics (PV). The power-to-weight ratio, the operational lifetime, and the cost/Watt of PV are critical parameters for the success of extra-terrestrial missions. Initially, for space application, silicon PV was employed; starting with the Vanguard 1 mission in 1958, but since the late 1990s, space missions have favoured multi-junction (III-V) PV with its high-power density (kW/m 2 ) and beginning-of-life (BOL) efficiency of around 30%.
However, there are emerging applications which will require solar arrays with high specific power (kW/kg), flexibility in stowage and deployment, and a significantly lower cost than is currently available. 4 New space PV technologies need to be developed to meet the needs of future advances in space exploration and energy harvesting. Some of these predicted advances include large constellations in space or fixed lunar/Martian bases, solar electric propulsion (SEP), 5 and Spacebased Solar Power (SBSP). 6 The latter is a method of collecting solar power in space for use on Earth exploiting the exposure to higher (than terrestrial) intensity sunlight, with near 24 hour-a-day operation and no climatic interference. Solar cells deployed in space are subject to high intensity radiation which is conventionally mitigated by covering with a cerium doped cover glass. The cover glass provides a pathway for the absorption of high intensity radiation that would darken any conventional glass. 7 The thickness of this glass determines the intensity of proton and electrons that are absorbed and is chosen depending upon the mission orbit and hence radiation environment that the solar cells are to be deployed in.
The innovative step of this research is to directly deposit thin film CdTe onto the cover glass thus saving on the weight and additional cost of having to use a substrate material: • The cover glass is flexible allowing it to be "rolled up" before and after the solar cell is applied to it.
• This flexibility enables a cost-reducing roll-to-roll manufacturing process.
• A flexible solar cell technology for space will enable reduction of stowage volume and new pathways for subsequent deployment.
A 100 micron, chemically toughened and cerium-doped cover glass has been supplied by Qioptiq Space Technology for this research. 8 However, this and previous studies indicate that the polycrystalline CdTe material could be more radiation hard than other materials. Bätzner et al stated that, for CdTe, "onset of cell degradation typically occurs at particle fluences, which are 2 orders of magnitude higher than that conventionally experienced by monocrystalline space solar cells of Si or III-V compounds". 9 This high level of radiation hardness, for CdTe, will potentially allow for a far thinner and therefore lighter cover glass than is used for conventional III-V devices.
Previous research into the proton degradation of CdTe solar cells has been influenced by generation of colour centres within the glass superstrates typically used. 9-11 G. Yang et al deposited CdTe onto Corning™ ultra-thin 100-micron glass and subjected samples to 15-MeV energy of protons through the glass side. 11 Using fluences of between 1 × 10 12 cm −2 and 1 × 10 15 cm −2 , their results were affected by a reduction in short-circuit current through a darkening of the non-radiation hard glass superstrate. This effective darkening of the glass reduces transmission of photons through to the PV material. It is a fast process and rapidly reduces the photo-current of the PV devices. This paper is the first to report the proton radiation hardness of CdTe deposited onto cerium-doped cover glass. Unlike any previous studies, the proton irradiation will not appreciably darken the glass superstrate, and hence the superstrate will not contribute to any additional loss of short circuit current within the device.
For these experiments, a proton energy of 0.5 MeV and irradiation directly applied to the CdTe face has been chosen to ensure penetration through the active layers into the glass substrate, and yet to maximise any potential proton damage within the polycrystalline semiconductor layer. Four fluences have been used to simulate different orbital environments and durations of missions.
The solar cell materials were deposited using metal organic chemical vapour deposition (MOCVD). The cells follow a superstrate configuration, see Figure 1, where 800 nm of Al-doped ZnO (AZO) and 100 nm of undoped ZnO buffer layer were deposited directly onto the chemically toughened 100-μm cerium-doped cover glass using MOCVD. The AZO/ZnO layers are followed by a 25-nm CdS seed layer and 125 nm CdZnS window layer before deposition of 3.25 μm of As-doped CdTe absorber layer. The As-doping of the CdTe layer is graded to produce an As concentration of 3 × 10 18   For the work presented in this paper, 9 × 0.25 cm 2 cells were prepared on each of the 5 deposited samples. The 9 cells share a common front contact (the AZO/ZnO) which is revealed by removing the CdTe/ CdZnS on 2 opposite sides of the sample followed by gold evaporation onto the exposed AZO/ZnO. Electrical probing is then made possible by contact to both the revealed AZO/ZnO strips with copper tape clamped in place and a gold probe contact to each of the 0.25-cm 2 gold square back contacts (see Figure 2).
In this configuration, it was possible to measure the electrical continuity across the TCO providing a useful metric to be collected at different stages of the irradiation. This bus-bar-to-bus-bar (B2B) value was found to be between 4 and 9 Ω for the 5 samples and did not vary significantly before or after the proton irradiation.

| Proton irradiation
Proton irradiation of the cells was performed at normal incidence to the CdTe/gold back contact face. The proton irradiation was carried out at the Surrey Ion Beam Centre using the 2 MV van de Graaff ion implanter, which is an Engineering and Physical Sciences Research Council (EPSRC) Central Facility. The implanter is capable of implanting ions at energies between 2 keV and 4 MeV, into sample sizes ranging from~mm 2 to 40 cm 2 . The samples may be held at constant temperatures ranging from~1270 to~77 K. Beam currents up to 10 mA were available, so that, even at the high particle fluences, the irradiations could be completed within a day.

| Solar cell characterisation
The AM1.5G solar cell performance was measured using an ABET Sun

| Inert atmosphere anneal of irradiated solar cells
A post-proton irradiation anneal was carried out on Sample 3, which experienced a fluence of 1 × 10 14 cm −2 and Sample 5, the control.
Using a Carbolite™ tube furnace, under a nitrogen atmosphere and at atmospheric pressure, the samples were held at 100°C for 168 hours.

| RESULTS
Five CdTe on ultra-thin cover glass samples were prepared following the methodology described in Section 2.1. The samples were clamped into Perspex holders which serve 2 purposes: to facilitate electrical contact to the exposed TCO front contacts and to provide mechanical stability during characterisation and proton irradiation. The 5 samples were measured in the CSER laboratory for AM1.5G performance before being transported to the Surrey Ion Beam Centre for proton irradiation. Four of the samples were loaded onto a platen and subjected to the 4 different fluences of 0.5-MeV energy protons.
The fifth sample was not irradiated and served as a control. The 0.5-MeV protons were shown by simulation to penetrate the PV structure such that they cause peak ionisation in the AZO layer just before the glass substrate interface. Table 1 provides a summary of the J-V performance for each of the samples, pre, and post proton fluence irradiation. Table 1 shows that the lateral resistance of the TCO, the B2B resistance, remains unchanged, within experimental error, after proton irradiation. Small sample to sample difference arises due to variations in the underlying TCO conductivity, thickness of the high resistivity ZnO layer, and resistance between the TCO and copper tape contacts.
The stability of the B2B to such high levels of proton fluence confirms that any change in device performance is not attributable to TCO performance. Table 1 also shows the before and after average cell performance of the PV cells. Figure 4A shows the mean relative efficiency (Eff) of the 4 samples versus their respective proton fluence. For the lowest fluence of 1 × 10 12 cm −2 , the mean solar cell efficiency decreased by 5%. Figure 4A shows that, for the CSER deposited CdTe and moving to a fluence of 1 × 10 13 cm −2 , the mean efficiency drops by 18%, which is again above the 80% of B.O.L. performance that is required for a solar TABLE 1 Mean J/V parameters for 9 × 0.25 cm cells on each of the 5 samples before and after proton irradiation. Sample 4, subject to the highest proton fluence, did not produce a J-V curve after irradiation. The standard deviation (SD) for each average parameter did not show any significant change before and after irradiation. All SDs were in the range of Eff 0.3-1.4%, J sc 0.5-1.5 mA/cm 2 , V oc 7-17 mV, and FF 1.1-4.5%. The B2B value is the bus bar-to-bus bar resistance (the lateral resistance of the TCO)  where the devices reach their threshold of proton radiation tolerance.
The relative mean efficiency drops by 96% when subject to a fluence of 1 × 10 14 cm −2 and exhibits no photoresponse at fluence of 1 × 10 15 cm −2 . The large drop in efficiency at 1 × 10 14 cm −2 can be attributed to a large decrease in J sc , shown in Figure 4B. The FF, shown in Figure 4C, has not shown a large decrease for this proton dose and a smaller decrease for V oc than J sc as can be seen in Figure 4D. The large decrease in J sc shown in Figure 4B was further investigated with EQE measurements of the cells. The EQE spectra taken from the cells following the different proton doses are shown in Figure 5. No significant change was observed in the EQE in Figure 5A, consistent with the measured J sc for this cell. Some apparent increase is observed in Figure 5B following the 1 × 10 13 cm −2 dose; most of this improvement is seen at longer wavelength, and the overall improvement is consistent with the apparent increase in J sc . Figure 5C shows the before and after  Figure 6 confirms this carrier concentration in the control, Sample 5. When Sample 2 was subject to 1 × 10 13 cm −2 proton fluence, the CdTe device carrier concentration was reduced by an order of magnitude from 1 × 10 16 cm −3 to 1 × 10 15 cm −3 but has retained most of the initial device efficiency, as can be seen in Table 1 and Figure 5B.
The evidence of the reduced carrier concentration for Sample 2 and the buried junction for the higher proton fluence, Sample 3, points to one of the following:   The degradation mechanism was investigated further using an annealing treatment. Both Sample 3, after a proton fluence of 1 × 10 14 cm −2 , and Sample 5, the control, were subject to an inert atmosphere anneal at 100°C for 168 hours as described in the experimental section. Figure 7 shows how this simple and a relatively low temperature anneal had a dramatic effect on recovering the EQE of the heavily irradiated Sample 3. No significant change was observed in the control PV cell. The Sample 3 efficiency was increased from 4% to 73% of its original performance. The relative mean J sc of Sample 3, which had taken the most significant deterioration from the proton irradiation, was increased to above the initial mean J sc , from 22.0 to 23.0 mA/cm 2 . Figure 7 shows that this is due to an increase in the long wavelength EQE which is similar to the post irradiated EQE for the 1 × 10 13 cm −2 proton dose of Sample 2.
The annealing mechanism of proton radiation damage can be expected to occur in space where solar arrays typically experience these annealing conditions (inert atmosphere and exposure to temperatures ≥100°C). The doses of protons that the samples have been subject to in this study would take many years to accumulate in the space environment; hence, this annealing and recovery mechanism could be expected to offset a significant amount of the damage observed in the solar cell performance at these proton fluences.

| DISCUSSION AND SOLAR CELL CAPACITANCE SIMULATOR (SCAPS) MODELLING
The degradation mechanism for the CdTe solar cells under intense pro-  The approach taken with the irradiated solar cells was to keep these baseline parameters the same, as far as possible, and observe if the changes in EQE and J-V parameters could be reproduced through changing the absorber layer N a and N t . It was found, using the Sample 2, 1 × 10 13 cm −2 data that a good fit could be obtained by reducing N a to 1.5 × 10 15 cm −3 (using the independently measured N a from the C-V profile in Figure 6) and keeping N t the same (for both CdTe:S and CdTe absorber layers), contrary to the expectation that proton damage would lead to an increase in N t . The J-V and EQE curve fits are shown in Figure 9A,B and the J-V fit parameters in Table 3.
It can be seen from Figure 9A and Table 3 Table 3 and did not require an increase in trap density as discussed as a possibility in Section 3. As the only change in the model parameters was the acceptor concentration in the CdTe:S and CdTe absorber layers, the observed reduction in efficiency of this proton irradiated cell can be clearly attributed to the reduction in N a that was confirmed by C-V measurement.
For the recovery of the EQE and J-V parameters, following the low temperature anneal, the same model parameters were used as for the un-irradiated and 1 × 10 13 cm −2 irradiated cells. The EQE fit, shown in Figure 10 and J-V parameters, shown in Table 4     making the bulk of the absorber layer n-type. The partial recovery after a 7-day low temperature anneal indicated that the compensation was reversible. These results can be explained by the proton irradiation creating interstitial hydrogen forming a shallow donor. The proton irradiation dose is sufficiently high to create up to 5 × 10 17 cm −3 donors if all the protons were absorbed in the CdTe layer, more than enough to compensate the active As acceptors. This is also consistent with the recovery following a low temperature anneal as hydrogen is a fast diffuser in CdTe.

| CONCLUSIONS
This study is the first to measure the effects of proton irradiation of CdTe solar cells deposited directly onto radiation hard cover glass.
Using the cover glass as the superstrate has removed the darkening effects observed for non-cerium doped glass superstrates. Different Once the proton dose was increased to 1 × 10 13 cm −2 and then 1 × 10 14 cm −2 , the solar cell relative efficiency decreased to 82% and 4%, respectively. This response to proton radiation was better than previous studies using CdTe and can be attributed to using cerium doped cover glass. Importantly, the efficiency following these doses is also more than 2 orders of magnitude better than conventional multi-junction devices. Device characterization by EQE of the irradiated cells showed that at the high dose of 1 × 10 14 cm −2 , a buried junction was forming with a spike in the EQE near the CdTe band edge (> 775 nm).
The effect of the high intensity proton doses on degradation of the CdTe solar cells can be explained using SCAPS modelling. An excellent fit was obtained for the 1 × 10 13 cm −2 dose sample with a reduction in N a and no change in trap density, which was surprising but supported with C-V measurement.
A low energy thermal anneal (100°C), in a nitrogen atmosphere, was carried out on the sample that received the 1 × 10 14 cm −2 dose.
The anneal had the effect of restoring the solar cell efficiency to 73% of its pre-irradiated value. The SCAPS modelling provided a good fit to the recovered EQE and J-V parameters with removal of the buried junction and N a restored to 8 × 10 14 cm −3 . This indicated that the compensation was readily reversible and is consistent with the proton irradiation creating interstitial hydrogen, resulting in a shallow donor level.
This study of 0.5-MeV proton irradiation for CdTe solar cells on a cerium doped cover glass over a wide range of fluences demonstrates a radiation hardness (to protons) that is far superior to conventional multi-junction III-V solar cells used for space. Further investigations will look at the effects of electron irradiation on the degradation mechanism of CdTe solar cells on the cover glass. This will be an opportunity to isolate radiation damage from the proton implantation mechanism.