Thin film cadmium telluride solar cells on ultra‐thin glass in low earth orbit—3 years of performance data on the AlSat‐1N CubeSat mission

This paper details 3 years of cadmium telluride (CdTe) photovoltaic performance onboard the AlSat‐1N CubeSat in low earth orbit. These are the first CdTe solar cells to yield I–V measurements from space and help to strengthen the argument for further development of this technology for space application. The data have been collected over some 17 000 orbits by the CubeSat with the cells showing no signs of delamination, no deterioration in short circuit current or series resistance. The latter indicating that the aluminium‐doped zinc oxide transparent front electrode performance remained stable over the duration. Effects of temperature on open circuit voltage (Voc) were observed with a calculated temperature coefficient for Voc of −0.19%/°C. Light soaking effects were also observed to increase the Voc. The fill factor decreased over the duration of the mission with a major contribution being a decrease in shunt resistance of all four of the cells. The decrease in shunt resistance is speculated to result from gold diffusion from the rear contacts into the absorber and through to the front interface. This has likely resulted in the formation of a deep trap state within the CdTe and micro shunts formed between the rear and front contact. Further development of this technology should therefore utilise more stable back contacting methodologies more commonly employed for terrestrial CdTe modules.


| INTRODUCTION
Thin film cadmium telluride (CdTe) photovoltaics (PVs) are a welldeveloped technology for terrestrial applications but have previously been untested in space. This paper reports on 3 years in a low earth orbit (LEO) of the first operational CdTe solar cell to be deployed in space. The novel approach is to directly deposit the CdTe PV material onto ultra-thin (100-micron) radiation hard cover glass, yielding a lightweight and flexible solar cell. 1 The in-orbit flight test has produced 3-year data, allowing evaluation of the durability of this type of technology towards space power applications.
Currently, multijunction gallium arsenide-based PVs dominate the space power sector with its very high conversion efficiencies providing the high-power density that satellites require. However, emerging space-based applications are driving demand for higher specific power (W/kg) and high peak output (>50-kW peak), greater stability to radiation and finally, an order of magnitude lower cost. The direct application of CdTe PV to space grade ultra-thin cover glass has the potential to meet all these requirements and to be a game changer technology.
The cover glass is a cerium-doped aluminosilicate glass, provided by Qioptiq Space Technology, and is laminated on the front surface of most PV solar cells used in space. It remains fully transparent whilst protecting the underlying cells from the high energy protons and electrons of the space environment. Previous reports of laboratory measurements have demonstrated that the CdTe on cover glass has achieved an air mass zero (AM0) conversion efficiency of 12.4%, yielding a cell level specific power of 600 W/kg. 1 This CdTe solar cell structure was shown by the authors to potentially be two orders of magnitude more radiation stable to protons than the multijunction gallium arsenide-based cell. 2 The cells were intentionally irradiated with 0.5-MeV protons incident on the CdTe side of the structure and not through the cover glass to deliberately maximise the damage at the PV junction. The cells were found to show little degradation up to a fluence of 1 Â 10 13 protons cm À2 . The stability of the CdTe structure under proton irradiation tests correlates with earlier work by Bätzner et al. who showed that radiation degradation, in thin film CdTe, occurred at two orders of magnitude higher particle fluence than for silicon or triple junction III-V space solar cells. 3 The opportunity to flight test the thin film CdTe on cover glass arose through a competitive bid process to build a payload for a joint mission between the United Kingdom and Algerian Space Agencies.
The AlSat-1N mission is a 3-unit CubeSat, built by the Surrey Space Centre, which flew three U.K.-based experimental payloads. 4 The CdTe on cover glass experiment will be referred to herein as the Thin Film Solar Cell Payload (TFSC) Payload. The TFSC Payload was developed by the authors over a 9-month period and needed not only to be able to perform and communicate current voltage (I-V) measurements but to be robust enough to withstand the rigours of launch, deployment and exposure to the space environment. The CubeSat was launched in September 2016 into a 661 Â 700 km, 98.20 Sun synchronous orbit. Initial data, in the first few weeks of orbit, showed the TFSC Payload to be fully operational, and the first survey under strong illumination was conducted on the 5 January 2017. 5 The authors now present 3 years of data from the TFSC Payload.
At the point of the last data entry in this paper, September 2019, the AlSat-1N had completed over 17 000 Earth orbits with the TFSC Payload experiencing temperature cycles of between À4 C and 51 C throughout. Over the 3 years, 135 surveys of the four CdTe solar cells comprising the payload were completed yielding a rare insight into the relatively long-term performance in space. Indeed, before this mission, there is no reported solar cell performance for CdTe. Historically, the only attempt to test a CdTe solar cell in space was a part of a suite of solar cells technologies to be tested on STRV-1B, launched in 1994. However, the 47 test cells (including one CdTe cell) did not communicate their I-V data back to the ground station. 6 Delamination, of thin film PV for space applications, has been considered to be of particular concern to the research community since a research programme, conducted by Dutch Space B.V., investigating thin film copper indium gallium diselenide (CIGS), on titanium foil, suffered delamination from the substrate at the interconnects when subject to the thermal cycling that might be encountered in space. 7 However, after these results, new interconnects were developed and a flight experiment conducted on the Delfi-C 3 , 3-unit CubeSat in a LEO. 8 The CIGS cells, on titanium foil, showed no signs of degradation over the initial 3-month monitoring period during which time they were subject to temperature changes of 70 C twice a minute due to the quick rotation of the CubeSat and the cells' low heat capacity. This demonstrated that the revised interconnect method had resolved the delamination issue. 8

| EXPERIMENTAL
A unique approach for space solar cells was taken, where the CdTe device structure was deposited directly onto the 100-micron cover glass using the superstrate configuration. Metalorganic chemical vapour deposition (MOCVD) was used to deposit all semiconductor layers except the evaporated gold back contacts (see Figure 1A).
Aluminium-doped zinc oxide was the first layer deposited and formed an extremely well-adhered transparent conducting electrode (TCE). This is followed by a thin, high resistance, layer of zinc oxide to reduce micro shunts. Both these initial layers were deposited at atmospheric pressure, using a nitrogen carrier gas. The glass/TCE structure was then moved to another atmospheric pressure MOCVD reactor, but using hydrogen carrier gas, for deposition of CdZnS n-type and arsenic-doped CdTe p-type layers and finally a chlorine heat treatment to passivate grain boundaries. 10 The surface was then rinsed with deionized water and the structure annealed in air for 90 min at 170 C before revealing the TCE bus bars by mechanical scraping.
Gold contacts were evaporated onto both the bus bars and the CdTe back surface defining four separate cells, as shown in Figure 1B.
The 60 Â 60-mm cover glass sample therefore contained 4 Â 1 cm 2 cells, defined by the back contacts, with a common front TCE. Using 4 Â 1 cm 2 defined cells increased the complexity of the experiment but generated more data and built in a certain degree of redundancy if one of the cells were to fail during the mission. External electrical contacting was then made possible through use of a second piece of cover glass with evaporated gold tracks that were offset by 5 mm and secured using a space qualified double side adhesive polyimide provided by GTS Flexible Materials Ltd. The polyimide had holes over each of the contacts that were filled with a conductive silver epoxy to enable rugged external connections to the printed circuit board (PCB). The silver epoxy introduced an increase to the series resistance; however, in the short time frame given to prepare the payload, it was decided that this approach offered the most durable electrical connection between the cells and the external circuit. Further details can be found in Underwood et al. 5 The external PCB, Figure 2A, also had an LM35 temperature sensor configured to measure in the range of À56 C to +148 C mounted centrally in intimate contact with the back of the cover glass. This sensor then allowed the TFSC Payload temperature to be measured each time the four cells were surveyed.
The internal PCB, Figure 2B, housed the electronic circuitry required to receive commands from the CubeSats onboard computer (OBC), survey each of the four cells I-V curves and the LM35 and store the data ready for upload to the CubeSats OBC. Table 1 shows the I-V measurements for the final flight model selected to fly on the CubeSat. The measurements were carried out at the Surrey Space Centre to ensure basic functionality of the assembled TFSC Payload with the onboard I-V measurement circuit. The laboratory was not equipped with a solar simulator; however, an uncalibrated 25-W halogen light source was used to verify the functionality of the control PCB ( Figure 2B). The low short circuit current (I sc ) was expected due F I G U R E 1 (A) The CdTe device architecture on 100-micron cover glass. (B) A CdTe device deposited on to 60 Â 60-mm 100-micron-thick cover glass; 4 Â 1 cm 2 cells defined by gold contacts to the CdTe with two common contacts to the TCE F I G U R E 2 Figure 2A shows the external PCB of the encapsulated TFSC Payload. Gold contacts for the four cells and two common bus bars can be seen on the top and bottom of the glass. These were connected to the PCB electronic circuit via indium/tin solder and gold wires. Figure 2B shows the internal PCB that, when commanded automatically, measures the four cells and the LM35 temperature sensor. Figure    The Earth's albedo is a function of cloud cover and season. 11 Average values for solar cell power from the albedo effect can be more than 10% of that derived from direct insolation alone.
Data surveys were carefully filtered from the analysis if they met the following criteria: • Surveys with the cells displaying close to zero I sc and low cell temperature were inferred to have been completed in a dark phase of the CubeSats orbit. • Surveys with low I sc but a higher cell temperature were considered to be lit by the Earth's albedo rather than direct solar radiation.
• Cells with low I sc values and were shown by the Sun-sensor readings to have a very shallow angle relative to the Sun.
• Telemetry files that were found to have corrupted data over part of the I-V curve.
Taking these criteria into account resulted in 66% of the surveys being removed from analysis. It should be pointed out, however, that this was not unexpected, as there was no active control of the pointing of the test cells towards the Sun, and so for each survey, it was a matter of chance whether there was favourable illumination of the cells. We did consider putting an optical trigger into the circuit so the cells were only measured when the Sun was at normal incidence-however, we abandoned this, as we could not be sure that the spacecraft would ever be in such a configuration. We instead decided to obtain what cell measurement data we could and derive the solar incidence angle retrospectively from the spacecraft's attitude determination systemprincipally the analogue Sun-sensor mounted on the same facet as the test cells.   Figure 5A). Figure 5B is a plot of the TFSC Payload cell temperature for each of the valid surveys between January 2017 and September 2019. The cell temperatures, as measured by the LM35 temperature sensor, ranged between À1.8 C and +36.6 C with the AlSat-1N core temperature ranging between À2.5 C and +31.6 C for the equivalent survey points. There was a linear relationship between the two temperature readings with the TFSC Payload being a few degrees Celsius higher than the AlSat-1N core. It should be noted that the surveys receiving solar flux above 115 mW/cm 2 and thus subject to more detailed analysis had cell temperatures across the whole range. CdTe solar cell V oc to be À0.24%/ C. 15 V oc , in CdTe, has also been shown to improve by $4% following light soaking due to filling of trap states in the absorber. 16 Figure 7 is a plot of V oc (mV) versus cell  The steady decrease, in fill factor, is caused by a decrease in shunt resistance over time seen in Figure 10. These two factors will act to decrease shunt resistance and as previously stated could explain the decrease in V oc seen in Figure 8B.

| Intersurvey variation
The TFSC Payload has been subject to recorded temperatures of up

| CONCLUSIONS
The TFSC Payload remains the only reported I-V data for CdTe solar cells from space. The TFSC Payload, originally designed for a 1-year in-orbit data acquisition, has provided 3 years of I-V measurements.
After the rigours of launch and deployment, 17 000 orbits and cell temperature variations between at least À3.8 C and +51 C, there is no evidence of delamination observed from the I-V data. The increase in V oc in space is explained by the lower cell temperatures and extended light soaking. The V oc temperature coefficient, for the relatively small range of temperature experienced by the TFSC Payload, is calculated to be À0.19%/ C. The aluminium-doped zinc oxide TCE performance remains unaffected after 3 years in space. The relative change in the I-V parameters over time is then attributed to ageing effects with the gold back contact, but maintenance of the high I sc value indicates good robustness of the CdTe device. It appears that in-diffusion of the gold back contact is producing microshorts leading to a decrease in the shunt resistance and thus the fill factor. If gold diffusion is the cause of the reduced shunt resistance, then it is likely that the creation of deep level traps within the CdTe that are acting to also reduce the V oc over time. Clearly, an alternative back contact architecture needs to be developed before the next in-orbit trials of this technology. With the current world record conversion efficiency for thin film CdTe being 22.1% AM1.5, the ultimate ambition of a space ready CdTe solar cell is a radiation and thermally stable device structure with 20% AM0 efficiency, a specific power of >1.5 kW/kg and a significantly lower production cost to current multijunction technology.
F I G U R E 9 Mean (circles) and standard deviation (lines) series resistance (Ω cm 2 ) over the duration of the mission for surveys receiving solar flux above 115 mW/cm 2 . Cell temperature ( C) (triangles) for each survey is shown on the secondary y-axis F I G U R E 1 0 Mean (circles) and standard deviation (lines) shunt resistance (Ω cm 2 ) over the duration of the mission for surveys receiving solar flux above 115 mW/cm 2 . Cell temperature ( C) (triangles) for each survey is shown on the secondary y-axis