Experimental Results from Performance Improvement and Radiation Hardening of Inverted Metamorphic Multi-Junction

Transcription

1 P. Patel, D. Aiken, A. Boca, B. Cho, D. Chumney, M. B. Clevenger, A. Cornfeld, N. Fatemi, Y. Lin, J. McCarty, F. Newman, P. Sharps, J. Spann, M. Stan, J. Steinfeldt, C. Strautin, and T. Varghese EMCORE Photovoltaics, Albuquerque, New Mexico, USA ABSTRACT This paper discusses results from continued development of Inverted Metamorphic Multi-junction (IMM) solar cells with air mass zero (AM) conversion efficiencies greater than 34%. An experimental best four-junction IMM (IMM4J) design is presented. In an effort to improve IMM performance in space radiation environments, 1-MeV electron irradiation studies are conducted on the individual IMM4J subcells. This data is used to engineer an IMM4J structure with beginning of life (BOL) AM conversion efficiency of approximately 34% and an end of life (EOL) remaining factor greater than 82%, where EOL is defined as performance after exposure to 1-MeV electron irradiation at 1E15 e/cm2 fluence. Next generation IMM designs are explored and an avenue toward AM conversion efficiencies of greater than 35% is presented. IMM4J PERFORMANCE IMPROVEMENT The focus of IMM performance advancement has shifted from the three-junction IMM (IMM3J) to the IMM4J device architecture over recent years (see Fig. 1). Performance results for the IMM4J solar cell were presented at the 35th IEEE PVSC [1]. Since that time, continued development has advanced the performance of the IMM4J solar cell by way of improvement of materials and device fabrication processes. Fig. 1 Schematic of IMM4J Cell Structure [1] One area of focus for materials improvement has been the optimization of metamorphic (MM) grading layers between the GaAs, 1-eV InGaAs, and.7-ev InGaAs subcells. These MM grade layers can be a source of high levels of material defects, specifically threading dislocations. Optimization of the MM grade between the GaAs (5.65Å) and 1-eV InGaAs (5.77Å) subcells has resulted in a reduction of threading dislocation densities (TDD) to values below 5x16/cm2 in the 1-eV subcell [2]. Threading dislocation densities were determined by cathodoluminescence (CL) measurements performed at National Renewable Energy Laboratory. Multiple measurements taken across the wafer show uniform TDD which indicate uniformly high material quality (see Fig. 2). 1

2 Substrate Removal Layer Optimization Relative Surface Roughness Relative Voc Fig. 2 CL of Surface of Layer with TDD < 5x1 6 /cm 2. Measured in Four Areas across the Wafer. Another key area for device performance improvement is optimization of the substrate removal process. This consisted of optimizing the epitaxial layers required for substrate removal as well as the etch process required to remove the growth substrate. This growth layer was identified as a source of morphology degradation, as is visible in optical surface roughness characterization. Experiments were conducted to reduce the effect this layer had on subsequently grown active regions. Optimization of the thickness of this layer resulted in a 1% reduction in haze levels. The reduced haze levels had a direct impact on resulting device performance increasing V oc by ~1.5% (see Fig. 3). Four conditions are presented in figure 3, process of record (POR), A, B, & C, the imperfect correlation between surface roughness and V oc is due to countering interactions during the subsequent substrate removal wet etch process..84 POR A B C Experiment 9 Incorporating these new developments into the previously presented IMM4J has resulted in a new experimental best conversion efficiency of 34.5% (1-sun, AM, 28ºC, 1353 mw/cm2) on a 27.6 cm2 size cell. This result was measured under EMCORE s 2-zone AM solar simulator, the resulting I-V curve data is shown in figure 4. Current (A) Fig. 3 Results of Substrate Removal Layer Optimization AM LIV - TS86-3A Voc = V Isc = ma FF = 85% Eff = 34.5% T = 28 o C Area = 27.6 cm Voltage (V) Fig. 4 IMM4J LIV Results The 2-zone simulator spectrum was calibrated using NASA Glenn s Learjet calibrated IMM4J subcells. Since only 2-zones are available for tuning, the spectrum is adjusted such that no subcell is artificially limiting the device. This results in 2 of the IMM4J subcells (GaAs and.7-ev InGaAs) being slightly overfilled, which could inflate 2

3 the measured fill factor. By design, the current limiting cell under the AM spectrum is the top subcell (InGaP). This has been confirmed by Quantum Efficiency (QE) measurements integrated with the AM reference spectrum. Due to this design feature the test error due to slight overfilling of the GaAs and.7-ev InGaAs cells is expected to be minimal. This result is approaching the practical efficiency limit of the IMM4J device architecture. To support this assessment, single junction subcells have been fabricated to isolate and characterize performance contributions by subcell, specifically those that are difficult to extract from characterization of a multi-junction structure. The single junction structures are designed such that the surrounding subcell layers are present, but iso-typed so that they are inactive junctions. This ensures that the active device is subjected to the same thermal loads, and strain from surrounding layers, as it would be in the full IMM4J structure. Similar single junction structures were also fabricated to represent the ZTJ structure. These cells will be used as a reference for evaluating the relative subcell qualities. With single junction iso-type cells the contribution of each subcell to the overall open circuit voltage (V oc ) can be extracted. To extract this data, each single junction device is tested under EMCORE s solar simulator, calibrated to the AM spectrum and temperature controlled to 28 degrees Celsius. The conclusion is that the IMM4J subcells are achieving V oc parity with, and in some cases surpassing both the ZTJ subcells and previously best reported results from open literature (see Fig. 5) [3-8]. Included in figure 5 is a collection of reported results for various materials that cover this range of bandgap energies, including but not limited to, germanium, GaSb, InGaAs, silicon, GaAs, AlGaAs, & InGaP. Voc (mv) Voc vs Bandgap Reported Results ZTJ Subcells IMM4J Subcells Egap (ev) Fig. 5 Summary of Reported Voc vs. Bandgap Results Characterization of the Internal Quantum Efficiency (IQE) of the individual subcells is also useful in assessing the overall performance (see Fig. 6). As illustrated in figure 6, the current collection of each IMM4J subcell is approaching 1%. This, along with the V oc data, supports the belief that the IMM4J is approaching the practical efficiency limit of this device architecture. Internal Quantum Efficiency InGaP (In)GaAs 1-eV InGaAs.7-eV InGaAs Wavelength (nm) Reflectance (%) Fig. 6 IMM4J IQE and Reflectance Results 3

4 IMM4J RADIATION HARDNESS 4J Subcell 1MeV Radiation Response A key requirement of solar cells intended for space applications is the ability to withstand exposure to electron and proton particle radiation. The first two IMM4J subcells consist of lattice matched (LM) InGaP and (In)GaAs. The electron and proton irradiation effects on these materials have been studied extensively in the EMCORE state of the art InGaP/InGaAs/Ge rev Z triple-junction (ZTJ) solar cells, currently in volume production at EMCORE [9]. The ZTJ solar cell can be considered the benchmark for radiation resistance, with EOL remaining factor of approximately 85%. The second two IMM4J subcells consist of InGaAs. Previous electron radiation studies conducted on InGaAs solar cells have demonstrated lower radiation resistance relative to InGaP and GaAs [1]. Thus incorporating an InGaAs subcell, while also maintaining radiation hardness of the incumbent triple-junctions, is challenging. EQE InGaP GaAs 1eV InGaAs.7eV InGaAs BOL EQE 1 MeV, 5e14 e/cm2 1 MeV, 1e15 e/cm Wavelength Fig. 7 Subcell EQE pre-post Irradiation Subcell Jsc Remaining Factor In an effort to engineer a radiation resistant IMM4J structure, single junction cells were fabricated to represent each IMM4J subcell. Surrounding subcell materials were iso-typed to produce the same device heat loads during growth as well as absorption characteristics for irradiation. These devices were then exposed to 1-MeV electron radiation at fluences of 5E14 and 1E15 e/cm2. The post irradiation QE results showed an increased degradation in the InGaAs subcells as expected (see Figs. 7 and 8). One unexpected result from this study was that the.7-ev InGaAs subcell demonstrates a higher radiation resistance than the 1-eV InGaAs subcell. The nominal active structures are very similar between these two subcells including targeted base and emitter doping profiles & thicknesses. Since no unique engineering of the radiation performance of the individual MM subcell structures had been performed to this point, it was expected that the higher InAs content in the.7-ev cell would result in a higher degree of degradation relative to the lower InAs content 1-eV subcell. Jsc Remaining Factor InGaP.88 GaAs.86 1eV InGaAs.7eV InGaAs.84 4.E+14 5.E+14 6.E+14 7.E+14 8.E+14 9.E+14 1.E+15 1.E+15 1 MeV Electron Fluence (e/cm2) Fig. 8 Subcell Jsc Remaining Factor as a Function of 1-MeV Electron Fluence One theory that may explain this result is that the BOL diffusion length of the higher InAs content cell (.7-eV InGaAs) is much longer than that of the lower InAs content cell (1-eV InGaAs). This would result from the higher minority carrier mobility in InAs relative to GaAs. Once subjected to electron radiation, the change in diffusion length of the.7-ev InGaAs may be larger than that of 1-eV InGaAs, but the net EOL diffusion length is still longer in the.7-ev subcell. This theory will be investigated further. 4

5 The subcells J sc remaining factors, determined in this study, have been used to assemble a radiation resistant IMM4J structure. This is accomplished by adjusting the BOL current mismatch as required to produce a current matched IMM4J structure at EOL (see Fig. 9). Optimization of antireflective coatings (ARC), subcell thickness, and subcell bandgap are used to produce the desired BOL current mismatch. The additional current needed in the GaAs cell is achieved by thinning the InGaP subcell so that it transmits more light to the GaAs subcell. The additional current needed in the InGaAs subcells is generated by a slight reduction in bandgap, thus increasing the absorption band. The ARC design is then re-optimized for the resulting device. Applying these changes to the present IMM4J structure will result in a net reduction in BOL performance. However the reduction in performance is less than previously expected. Modeled current generation will be the same as the present IMM4J device due to countering effects of thinning the InGaP subcell and improved ARC design. Modeled Voc will be reduced by ~1% due to decreased InGaAs subcell bandgaps. The projected IMM4J device still results in ~34% BOL AM conversion efficiency and will have an EOL remaining factor greater than 82%. This device is currently being assembled and will be subjected to electron irradiation testing for confirmation. Relative Jsc Subcell Jsc / InGap Top Cell Jsc InGaP GaAs 1eV InGaAs.7eV InGaAs NEXT GENERATION IMM DESIGN Beyond incremental improvement of the IMM4J device, a major change in device architecture is required to realize AM conversion efficiencies >35%. A series of bandgap models have been assembled to evaluate possible next generation IMM structures. Included in these models are known practical limits of readily available materials such as ARC materials and III/V bulk semiconductors. Multiple cell architectures were evaluated including further bandgap optimization of the IMM4J structure. As a result of this study, it is believed that the most promising next generation device is the IMM5J architecture (see Fig. 1). The total spectral range absorbed in the IMM5J structure is less than that of the IMM4J as seen from the higher bandgap of the bottom subcell. The benefit comes from significantly higher voltage increase than current loss, since the spectrum that is being absorbed will be converted more efficiently. By dividing the spectrum into smaller segments, thermalization losses will be significantly improved. Practical efficiency modeling of this device architecture predicts an AM conversion efficiency of 35.8% compared to a practical upper limit of 34.6% for the current IMM4J architecture. This structure does not represent the optimal 5J device structure one could theoretically assemble but it has significant practical advantageous, including maintaining lattice match in three of the highest power producing subcells. The biggest challenges in realizing the proposed device are development of a high quality, high bandgap top subcell and an optically transparent tunnel diode (TD) to interconnect the top two subcells. 8 BOL - Jsc / TC Jsc EOL - Jsc / TC Jsc Fig. 9 Subcell Jsc Relative to InGaP TC Jsc 5

6 Lattice Constant -ev 1.1-eV 1.4-eV 1.7-eV 2.1-eV Growth Template Voc (mv) Voc vs Bandgap Reported Results IMM4J Subcells IMM5J Subcells Egap (ev) Fig. 11 Summary of Reported Voc vs. Bandgap Results Fig. 1 IMM5J Structure Single junction devices have been produced to evaluate the performance of the new subcells required to assemble the IMM5J. These include the high bandgap 2.5-eV and intermediate 1.74-eV LM subcells as well as the 1.13-eV and.89-ev MM subcells. Metalorganic vapor phase epitaxy (MOVPE) growth conditions, for the high bandgap materials, have been explored in order to minimize unintentional background doping levels. Optimized conditions have produced background p-type doping levels of less than 2E16 /cm 3, which should provide adequate material quality for these subcells. Lessons learned during optimization of the MM grading buffers in the IMM4J structure are used to form the necessary grades for the new MM subcell bandgaps. As was the case in the performance study of the IMM4J, these single junctions will be characterized to better understand their individual contributions to the IMM5J device. Extracting V oc from these devices and comparing to the IMM4J subcells provides a good illustration of their quality (see Fig. 11). By comparison to historical results, the high bandgap materials exhibit good V oc characteristics. The new subcells have been characterized by QE to determine their current collection properties. This information is needed to further asses the quality of the new subcells, as well as properly current match the IMM5J subcells. An absorption model is used to determine the initial bandgap and absorption thickness combinations but there are too many unknown variables to set the exact current match based on this modeling. The Internal Quantum Efficiency (IQE) results for the high bandgap subcells shows current collection that is on par with expectations from the absorption modeling (see Fig.12). These results suggest the largest opportunity for continued improvement exists in the 1.74-eV subcell. IQE High Bandgap Subcell IQE eV SJ eV SJ.4 2-eV IQE Model eV IQE Model Wavelength (nm) Fig. 12 IQE Model and Results for 2.5-eV and 1.74-eV Subcells 6

7 One of the keys to realizing the projected conversion efficiency of 35.8% is developing an optically transparent TD that interconnects the 2.5-eV and 1.74-eV subcells. As currently assembled, the bandgap of the TD layers is approximately 1.9-eV. Since this is lower than that of the top subcell a portion of the photon energy that should be transmitted through to the 1.74-eV cell is being absorbed in the TD (see Fig. 13). Electron-hole pairs that are generated in this region recombine and are lost. Integrating the QE of the two models shows a relative current gain of 11% with the advent of a transparent TD. IQE TD Absorption Effects on 1.74eV Subcell 1.91eV TD Transparent TD Wavelength (nm) Following successful demonstration of the IMM5J device, the next logical step for performance improvement is to recoup the reduction in total absorption relative to the IMM4J. This could be accomplished by adding a third metamorphic subcell with a bandgap on the order of.7-ev which creates a 6- junction IMM structure (IMM6J) (see Fig. 14). Practical efficiency modeling of this device architecture predicts an AM conversion efficiency of 37.8% compared to a practical upper limit of 34.6% for the current IMM4J architecture. Considering that the only requirement to realizing this device over the IMM5J device is the addition of a.7ev subcell, this is viewed to be a practical path toward achieving an AM Solar Cell with conversion efficiency approaching, and possibly exceeding, 37%. Lattice Constant.7-eV -ev 1.1-eV 1.4-eV 1.7-eV 2.1-eV Growth Template Fig. 14 IMM6J Structure SUMMARY Results have been presented, demonstrating further advancement of IMM4J AM conversion efficiency. Implementation of recent materials and fabrication process improvements has resulted in a new experimental best AM efficiency of 34.2%. A performance investigation of the IMM4J subcells was presented. The results of this study suggest that the IMM4J is approaching the practical efficiency limit for this device architecture. A study was performed to analyze the radiation resistance of the IMM4J subcells. This data has been used to design an IMM4J device with projected BOL efficiency of 34% and EOL remaining factor greater than 82%. Next generation IMM devices were discussed, including the IMM5J and IMM6J devices. Performance data for the IMM5J subcells was presented. Overall, the IMM5J subcell results provide confidence that this device architecture is a viable path toward improving upon the IMM4J performance. A practical path toward a 37% efficient AM IMM6J was presented. This device is believed to have the most potential for major advances beyond the IMM4J that was presented. EMCORE has patents pending for the aforementioned IMM4J, IMM5J, & IMM6J devices. 7

8 ACKNOWLEDGEMENTS The authors would like to recognize the Air Force Research Laboratory, Space Vehicles Directorate (AFRL/RVSV), specifically Alex Howard, Henry Yoo, John Merrill, and David Wilt for their continued support of the IMM cell development. REFERENCES [9] M. Stan, D. Aiken, B. Cho, A. Cornfeld, J. Diaz, A. Korostyshevsky, V. Ley, P. Patel, P. Sharps, and T. Varghese, Evolution of the High Efficiency Triple Junction Solar Cell for Space Power, Proc. 33 rd IEEE PVSC, San Diego, CA, 28 [1] M. Yamaguchi, Radiation Resistance of Compound Semiconductor Solar Cells, J. Appl. Phys. 78, 1995, pp [1] A. Cornfeld, D. Aiken, B. Cho, V. Ley, P. Sharps, M. Stan, and T. Varghese, Development of a Four Sub-Cell, Inverted Metamorphic Multi-Junction (IMM) Highly Efficient AM Solar Cell, Proc. 35 th IEEE PVSC, Honolulu, HI, 21 [2] M. Stan, D. Aiken, B. Cho, A. Cornfeld, V.Ley, P. Patel, P. Sharps, T. Varghese, High-efficiency quadruple junction solar cells using OMVPE with inverted metamorphic device structures, J. Cryst. Growth 312 (21) [3] L. Fraas, J.E. Avery, V.S. Sundaram, V.T. Dinh, T.M. Davenport, J.W. Yerkes, J.M. Gee, and K.A. Emery, Over 35% Efficient GaAs/GaSb Stacked Concentrator Cell Assemblies for Terrestrial Applications, Proceedings of the 21st IEEE PVSC 199, p [4] J.C. Schultz, M.E. Klausmeier-Brown, M. Ladle Ristow, M.M. Al-Jassim, High Efficiency 1.-eV GaInAs Bottom Solar Cell for 3-Junction Monolithic Stack, 21st IEEE PVSC 199, p [5] M. A. Green, K. Emery, Y. Hishikawa, W. Warta, E. D. Dunlop, Solar Cell Efficiency Tables (version 39), Progress in Photovoltaics Volume 2 (211), p.12-2 [6] C.J. Keavney, V.E. Haven, and S.M. Vernon, "Emitter Structures in MOCVD InP Solar Cells", 21st PVSC 199, p [7] R. P. Gale, J. C. C. Fan, G. W. Turner, and R. L. Chapman,, A New high-efficiency GaAs solar cell structure using a heterostructure back-surface field, Proc. 17th IEEE PVSC 1984, pp [8] T. Takamoto, E. Ikeda, H. Kurita, and M. Ohmori, High Efficiency InGaP Solar Cells for InGaP/GaAs Tandem Cell Application, 1st WCPEC (1994), p