Changes in output parameters of 1 MeV electron irradiated upright metamorphic GaInP/GaInAs/Ge triple junction solar cell

The changes in output parameters of 1 MeV electron irradiated MOCVD grown upright metamorphic (UMM) GaInP/GaInAs/Ge triple junction solar cells have been studied. Non-ionizing energy loss (NIEL) approach and MULASSIS simulation were applied for analyzing the effects of irradiation induced displacement damage on cell performance. The influence of base thickness on radiation resistance has been studied by changing the base thickness of top GaInP and middle GaInAs subcell, respectively. The experimental results show that the electrical parameters, Voc, Isc, and Pmax of UMM cell degrade with the increase of electron fluence. The change of spectra response indicates middle GaInAs subcell degrades more severe than top GaInP subcells, and the base thickness of two subcells has different effects on spectra response of UMM cell.


I. INTRODUCTION
The fundamental objectives of developing space solar cells are to increase the conversion efficiency, improve the radiation tolerance, reduce the mass and cost of solar cells and solar arrays. Space solar cells have undergone significant development from Si to III-V based materials, 1 from single junction to multijunction configurations, 2 from lattice matched to lattice-mismatched structures for higher conversion efficiencies. 3 III-V multijunction solar cells deliver the highest performance among the current available solar cell technologies, and lattice-matched (LM) GaInP/Ga(In)As/Ge triple junction solar cells are the most matured technology for space photovoltaic applications. However, the conversion efficiency of LM cell is limited by the non-optimal bandgap combinations to the given AM0 solar spectrum. Bandgap matching within the subcells can be obtained by using different materials, and several approaches such as wafer bonding, 4 mechanical stacking, 5 inverted metamorphic (IMM) 6,7 and upright metamorphic (UMM) 8,9 have been proposed to solve the lattice mismatch problems between the subcells. These methods, except the UMM, on the other hand, require additional fabrication processes and increase the cost. 40.7 % conversion efficiency of a metamorphic three-junction GaInP/GaInAs/Ge cell under high concentration terrestrial application has been demonstrated. 10 Furthermore, latest GaInP/Ga(In)As/Ge UMM solar cell product of AZUR has been reported with efficiency of 31.0 % under AM0 spectrum, 11 while the highest efficiency of traditional LM cell is about 30 %. Therefore, UMM solar cell is a promising alternative for space application.
However, the harsh space environment, such as vacuum, strong ultraviolet, electron and proton radiation, temperature cycling, brings additional challenges for space solar cell designing. Electron and proton irradiation is the main reason of solar cell degradation in space. The degradation mechanisms of conventional triple junction LM cells upon high energy particles have been studied extensively in past decades. 12,13,14 Although there are some reports of radiation effects of IMM solar cells, 15,16 unfortunately, reports of space radiation testing of UMM solar cells are scarce, and a number of design challenges still remain to be addressed.
In this paper, overall performance of 1 MeV electron irradiated metal-organic chemical vapor deposition (MOCVD) grown GaInP/GaInAs/Ge UMM triple junction solar cells have been investigated. The degradation of both electrical and spectral output parameters of UMM cell have been discussed. Fig. 1 shows the major structure of UMM GaInP/GaInAs/Ge triple junction solar cells used in this work. All samples were grown by an AIXTRON MOCVD reactor on p-type Ge substrate, and solar cells were fabricated with standard lithography processes. The details of epitaxial growth and device fabrication has been reported in our previous work. 17 Cell structure contains a 3-µm-thick distributed Bragg reflector (DBR) between middle and bottom cell, which is consists of 15 pairs AlGaInAs/GaInAs layers. A compositionally step-graded monolithic GaInAs, with a strain gradient of 0.5 %/µm and total thickness of 2 µm, was used as a buffer layer between DBR and Ge bottom cell. All the devices studied in this work are 4 cm × 8 cm in size and with initial conversion efficiency of 30.5 %, under standard AM0 solar spectrum.

II. EXPERIMENTAL
The electron irradiation was conducted by an ELV-8 vertical electron accelerator at room temperature. Samples were placed at a uniform electron flux 5 × 10 10 e/cm 2 ·s area during the irradiation without bias. Electrical and optical parameters of solar cells were measured prior to irradiation and after irradiated by 1 MeV electron with total fluence of 5 × 10 14 , 1 × 10 15 and 1.5 × 10 15 e/cm 2 , respectively. The applied electron fluence has been selected based on the space solar cell evaluation irradiation test standard. 18 Electrical output parameters of solar cell, short circuit current (I sc ), open circuit voltage (V oc ), and maximum output power (P max ) were measured under the standard test conditions  (AM0, 136.7 mW/cm 2 at 25 • C) by Spectrosun X25A solar simulator. The external quantum efficiency (EQE) was measured by a home-build setup. Besides, the structural properties of UMM solar cell have been studied by high-resolution X-ray diffraction (XRD), and, threading dislocation density (TDD) was confirmed from cathodoluminescence (CL) measurement. Result of these measurement, which indicates high quality of as-grown solar cell epilayers, have been reported in our previous work. 17

A. Electron irradiation simulation
When charged particles go through a solar cell structure, part of its energy is transferred to crystal lattice and result in ionization or atomic displacement of the solar cell. The density of such defects increases rapidly with the increase of the irradiation fluence. These displacement damages introduce point defects in solar cell crystal structure, such as vacancies, interstitials, etc., which is the increase the decreasing the minority carrier diffusion length by increase recombination centres, and, consequently, cause the degradation of solar cell output parameters. 19 The degradation performance of solar cells can be predicted by displacement damage dose (D d ) approach. 20 The numerical value of D d in a given material can be evaluated by combining irradiation fluence and nonionizing energy loss (NIEL) of target material. Calculated NIEL values for GaInP2, GaAs and Ge materials upon electron and proton irradiation with different energy have been reported. 21 Since the NIEL values of these three materials are very close to each other for 1 MeV electron irradiation, we selected the NIEL value of GaAs and corresponding D d of our samples by 1 MeV electron irradiation is calculated by following equation, 22 where E e is electron energy, Φ e is the electron fluence. The applied electron fluences for irradiation in this study and its corresponding D d values are listed in Table I Table I.
During the irradiation, the energy transferred from the colliding electron to the target atoms and create defects via nuclear elastic interaction. If the initial energy of electron is much higher than the displacement energy threshold of target atom, then more energy will be transferred to the recoil atom, which will collide with other target atoms and create additional recoil atoms, and, consequently, generate more defects. 23 Therefore, the irradiation induced displacement damage by high energy particles increases with the distance away from surface. similar for all three parameters. When the electron fluence reached 1.5 × 10 15 e/cm 2 , the value of P max decreased 23.6%, and degradation of V oc and I sc are 11.8% and 9.5%, respectively. The degradation of V oc is bigger than I sc , and, P max degraded the most. The irradiation induced displacement damages in solar cell lattice act as recombination center, generation center, compensation center, and temporary trapping center, etc. 24 by creating additional energy level in energy bandgap of semiconductor materials. When the fluence of irradiation particles increased, the density of these defects increases rapidly and degrade the output parameters of solar cell severely. The degradation of I sc is mainly caused by the displacement damages in the active layers, base and emitter region, of solar cell. Defects in this region act as non-radiative recombination centers and separate the photo-generated electron-hole pairs when they arrive near to the p-n junction, and, consequently, reduce the minority carrier lifetime and diffusion length which result in the degradation of the I sc . 25 Fig . 4 show the degradation of EQE value of each subcell in UMM triple junction solar cell after 1 MeV electron irradiation. It can be seen from Fig. 4 that the EQE values of top and middle subcells decreased with the increase of electron fluence. However, in bottom Ge cells, the EQE value fluctuated with the increase of electron fluence in irregular order. This abnormal spectral response of Ge bottom cell was also observed by some other groups as artifact EQE response, and the main reasons for this phenomenon are the combined effects of shunt and luminescence coupling. 26,27 Typically, Ge bottom subcell in GaInP/GaInAs/Ge triple junction solar cell produces bigger I sc compared to top and middle subcells before and after irradiation, but the I sc of multijunction solar cell is determined by the smallest I sc among all of subcells. Therefore, the degradation of Ge bottom cell by irradiation will not affect the overall performance and can be ignored in cell performance evaluation. It also can be observed from Fig. 4   top and middle subcells happened in long-wavelength regions, and, the degradation of GaInAs middle subcell is larger compared to that of GaInP top subcell. This result can be seen more clearly in Fig. 5. The result of EQE spectral response degradation in each subcell is consistent with irradiation induced displacement damage distribution analysis result shown in Fig. 2(a). The value of D d in each subcell is bigger in the base layer compared to emitter layer. Therefore, the probability of collection of photo-generated carriers near to the bottom of base layer is low due to the reduction of carrier diffusion length. On contrary, the carriers generated near the depletion region, which is corresponding to short-wavelength region, can be collected more effectively. Top Ga 0.43 In 0.57 P subcell has relatively smaller degradation compared to middle Ga 0.92 In 0.08 As subcell due to the comparatively bigger In-P fraction. The comparison results of InGaP, InGaAsP, and InGaAs solar cells showed that the radiation resistance of these materials can be significantly improved by increasing In and P fractions. 28 The superior radiation resistance of InP based solar cell is originated from the room-temperature annealing and minority carrier injection-enhanced annealing phenomena of major radiation-induced defects in InP. 29

C. Effects of top and middle subcell active layer thickness
In order to investigate the effects of subcell active layer thickness on the initial conversion efficiency and radiation resistance of UMM cell, we prepared two groups of samples which are completely identical to previously studied samples except the thickness of the active layer of top GaInP subcell and GaInAs middle cell. Our reference sample, sample ID F00, has subcell active  layer thickness of 500 nm (top)/1.5 µm (middle) and current density of 19.13 mA/cm 2 (top)/ 18.65 mA/cm 2 (middle), respectively. In one group, the thickness of GaInP subcell active layer is decreased from 500 nm to 450 and 400 nm respectively, and in another group, the thickness of GaInAs subcell active layer is increased from 1.5 µm to 1.7 and 1.8 µm, respectively, as shown in Table II. Since the diffusion length of photo-generated carriers are larger than any of these chosen film thickness, the overall extraction of photo carriers will not be affected but only the degradation properties of the cell. All samples went through the same irradiation described as above with electron fluence of 1.5 × 10 15 e/cm 2 , and electrical and spectral parameters have been measured before and after irradiation. The remaining factors of I sc , V oc , and P max for each cell have been listed in Table II. Fig. 6(a) shows the degradation of spectra response of UMM cell with two different GaInP thicknesses before and after irradiation. The initial I sc of top GaInP subcell increases from 18.55 mA/cm 2 to 18.98 mA/cm 2 and the initial I sc of middle GaInAs subcell decreases from 19.39 mA/cm 2 to 18.70 mA/cm 2 with the thickness increases from 400 nm to 450 nm before electron irradiation, respectively. The degradation of top cell and middle cell I sc after irradiation are shown in the inset of Fig. 6(a) as a remaining factor normalized to their initial values, while the remaining factor of I sc , V oc , and P max for UMM cell are also plotted in Fig. 6(b) for comparison. The result indicates that thickness of top GaInP cells has less influence on the degradation of spectra response of UMM cells, and the degradation of subcell I sc is equal to 2% and 10% for GaInP top subcell and GaInAs middle subcells, respectively. Therefore, the degradation of I sc of UMM cells should be 10% considering the middle subcell current limiting property after electron irradiation. This is consistent well with degradation of I sc of UMM, the remaining factor of around 0.91, shown in Fig. 6(b). The remaining factor of V oc , decreased from 0.900 to 0.894 when the top InGaP subcell thickness increased from 400 nm to 500 nm, while the corresponding degradation of P max is from 0.826 to 0.809. These results indicate that a thinner base layer is very important to reduce P max degradation of top subcell, even GaInP material has good radiation tolerance property.
The result of another group cells, with different GaInAs middle subcell thickness, are plotted in the same way in Fig. 6(c) and 6(d). The I sc of top GaInP subcells is unchanged and the I sc of middle GaInAs subcells increases a little from 19.84 mA/cm 2 to 20.02 mA/cm 2 with the increase of GaInAs middle subcell thickness from 1.5 µm to 1.7 µm. It indicates that middle GaInAs thickness has less influence on the spectra response of UMM cells, and this can be attributed to the DBR shown in Fig. 1. The variation of GaInAs subcell thickness has no influence on the degradation of spectra response of top GaInP subcell, and the degradation of I sc is equal to 2%. However, it has significant influence on the degradation of spectra response and I sc of middle GaInAs subcell, and degradation scale of I sc is consistent with the value shown in Fig. 6(d). The remaining factors of all I sc , V oc , P max decrease as the GaInAs thickness increases from 1.5 µm to 1.8 µm. The degradation of I sc and V oc indicate that GaInAs has relatively weak radiation resistance compared to GaInP, and it is consistent well with the report of Imazuri et al 30 The optimized cell with an initial conversion efficiency of 30.5% with subcell active layer thickness of 470 nm (top)/1.5 µm (middle) is fabricated based on above study.

IV. CONCLUSION
In this paper, the electrical and spectral response of 1 MeV electron beam irradiated MOCVD grown GaInP/GaInAs/Ge UMM triple junction solar cells have been studied. Non-ionizing energy loss (NIEL) approach and MULASSIS simulation were applied for analyzing the effects of irradiation induced displacement damage on output parameters of solar cell. The result shows that both electrical and spectral parameters of UMM cell were degraded along with the increase of electron fluence, V oc , I sc , and P max decreased by 11.8%, 9.5% and 23.6%, respectively, when the electron fluence reached 1.5 × 10 15 e/cm 2 . The main reason for these degradations is the displacement damage in solar cell active layers induced by the irradiation. EQE spectral response of top GaInP and middle GaInAs subcells mainly happened in long wavelength due to the bigger displacement damage density in base layer of subcell compared to that of in emitter layer. Radiation resistance of subcells have been studied by different base layer thickness of top and middle subcells, respectively. It indicated that a thinner base layer results in better resistance of GaInP top subcell, and, increase of GaInAs middle subcell base layer thickness decrease the radiation resistance of middle subcell significantly.