A 100X magnification, ± 2.5° field of view micro-concentrating optical system has been developed for a microsystems-enabled photovoltaic (MEPV) prototype module using 250 µm diameter multi-junction “stacked” PV cells.
© 2014 Optical Society of America
1. MEPV introduction
Sandia National Laboratories has been investigating a novel, low cost photovoltaic system architecture that combines the high conversion efficiencies of concentrated photovoltaics (CPV) with the form factor and low system costs of flat panel PV. MEPV marries fully integrated, stacked multi-junction PV cells having lateral dimensions smaller than 1 mm with micro-optic concentrators to reduce the use of expensive semiconductor materials and to increase solar conversion efficiency . While a first generation module demonstrated theviability of the MEPV construct using Si cells with a 720 µm diameter , recent research has developed a second generation prototype to improve device performance and to examine the feasibility of mass producing MEPV at system costs competitive with the DOE SunShot objective of $1 per watt peak for utility scale solar power generation by 2020.
2. Micro-concentrator design
The micro-concentrating optics for a second generation prototype were designed to achieve 100X magnification and greater than 90% optical transmission across a pass band of roughly 400 to 1600 nm. A ± 2.5° field of view was selected to insure compatibility with commercial course sun tracking systems. Environmental considerations for the optics and module design included a 20 year service life; operating ambient temperatures from −40° C to 80° C; and exposure to hail, rain, humidity, dust and UV radiation. Cost analyses provided an over-arching constraint on the final design as improvements in optical performance were constantly evaluated relative to their impact on the total system cost . Figure 1 shows the resulting optical design and ray tracing for the two plano-convex lens, 8th order aspheric design with a 100X magnification and ± 2.5° field of view. Although the design introduces a hot spot with a peak intensity exceeding 700 suns at the cell surface, Fig. 1; the short thermal conduction path for sub-mm sized MEPV cells ensures temperature increases of only a few degrees and minimal degradation in cell performance . Incident rays onto the PV cell were constrained to less than 30° as the front optic entrance aperture is 2.5 mm with an exit aperture onto the PV cell of 0.25 mm. The concentration-acceptance angle product (CAP)  of the final design is 0.57. While this is lower than the 0.060 and 0.69 products achieved for novel Fresnel Kohler  and XR reflective Kohler  designs respectively, the design remains compelling due to its simplicity and expected lower system costs.
The thickness of the lens “sandwich” is only 5.30 mm, a 65% reduction from the first generation prototype , and an even greater reduction from traditional CPV systems. The optics are arranged in a 240 element hexagonal closed packed array across a roughly 40 mm square collection area using a 15 x 16 format with 2.381 mm and 2.058 mm pitch spacing respectively. The error budget for the optic surfaces includes a ± 5 µm tolerance for form accuracy, a 30 nm Ra tolerance for surface finish, a ± 25 µm tolerance for optic to cell planar alignment, and a ± 50 µm tolerance for optic to cell axial placement.
Polycarbonate was selected as the high index (n = 1.59) concentrator lens material due to its low cost and availability for mass production molding. The gap between the two lenses is filled with Sylgard®184 PDMS (n = 1.40) to prevent moisture ingression into the concentrator module, to minimize Fresnel reflections, and to insure high optical transmission without UV degradation. The relatively low elastic modulus of PDMS (2.3 MPa) provides a further advantage in accommodating stresses generated by thermal excursions and CTE mismatches in the optical assembly. Results from thermo-mechanical analyses based on Mooney-Rivlin fit for the PDMS material are shown in Fig. 2 whereby the spacing between the front and rear lens arrays was increased by a factor of five from the initial design. As a result, stress loads inthe PDMS were reduced to levels below the limits for both cohesive and adhesive failure across a temperature change of 40° C.
Figure 3 provides a cross-section of the prototype concentrator design. The lens and MEPV cell arrays are assembled between two Gorilla® glass plates using urethane adhesives for an overall module thickness of only 9.96 mm. Isolation from environmental contamination is achieved with a butyl sealant around the outside perimeter. Assembly and alignment of the front and rear optic arrays is achieved passively using symmetric, over-constrained 45° angled pin-in-slot features that are molded into each part, Fig. 4.Bosses on the pin features set the axial position of the two lens elements to one another. The symmetric geometry of the mating features provides an athermal mounting configuration with expected alignment tolerances better than 25 µm. Alignment and assembly of the cell array is also performed passively using monolithic “wedding cake” features in the rear optic array that mate to holes in the polyimide flex, Fig. 4. AR coatings are included on the front face of the top glass and on the front face ofthe PV cell stack; reducing the air to glass reflective loss from 4% to 1% and the urethane to GaAs cell reflective loss from 20% to 2%. No coatings are used at the polycarbonate to PDMS interfaces since their losses are on the order of only 0.4% per interface.
3. Micro-concentrator fabrication
Fabrication of the second generation prototype optics has investigated techniques amenable to low cost, volume manufacturing. The polycarbonate lens arrays have been injection molded using 6061-T6 aluminum mold inserts that were machined using micro-milling for rough figuring and ultra-precision diamond milling for final finishing. Process development has focused on reducing optic surface finish to improve system efficiency; increasing process throughput to reduce manufacturing costs, and reducing diamond tool wear to minimize performance variations across the lens arrays. The front lens element had the minimum surface radius, 0.677 mm, and maximum lens sag, 1.04 mm; while the rear lens element had the highest surface slope, 89.2°. Constraints implicit from both machining and molding processes were incorporated into the opto-mechanical design process. Rough micro-milling of the insert produced a surface with approximately 20 µm of remaining stock material and a form error of ± 5 µm. It also significantly reduced the overall machining time and diamond tool wear compared to diamond machining the entire insert surface. The insert was then mounted and aligned onto a 4-axis diamond turning machine where a single final finishing pass was performed using diamond milling. Final finishing involved the use of a single diamond tool with a 200 µm nominal radius and a 70° nominal side clearance angle for each mold insert. While metrology of the inserts has not yet been performed due to project schedules, a form accuracy of 1.5 µm and apex surface finish of 30 nm Ra was achieved on a test optic array. Subsequent fabrication of the mold insert for the “wedding cake” features on the rear optic has demonstrated feature dimensional accuracies of ± 1-6 µm with positional accuracies of ± 8 µm.
Figure 5 shows the finished mold insert for the rear optic array and a subsequent molded micro-concentrator array. Metrology of the molded arrays remains incomplete, but data collected to date suggests the concentrator module will meet system requirements. Initial molding experiments demonstrated that the in-plane material shrinkage across the array was less than 0.2% as optic centers were located in X and Y with an accuracy of ± 5 µm, Fig. 6.Out-of-plane distortion of the array, however, approached 100 µm as mold filling of arrays with aspect ratios exceeding 30:1 with minimal part distortion and no flow lines has proven challenging. Design changes were incorporated into the final mold pins to compensate for the 0.2% in-plane material shrinkage, and to reduce mold flow restrictions by changing the mounting feature geometry and increasing the part thickness by 0.6 mm. Surface finish on final molded optic arrays is on the order of 25 nm Ra, as seen for the rear array in Fig. 7, and meets the 30 nm design tolerance. While the ± 11 µm form error in Fig. 7 for the rear array does exceed the design’s ± 5 µm tolerance; inside a radius of 0.71 mm, i.e. across 69% of the surface area, the form error meets specification. A comparable surface finish of 25 nm Ra and form error of ± 12 µm was measured on the front lens arrays.
4. Micro-concentrator performance
Preliminary testing on an optical sub-assembly has demonstrated performance consistent with the optical design. Test samples were assembled consisting of front and rear lens arrays bonded together using the PDMS filler without the cover glass, cell array or AR coatings. Under one sun simulated illumination from a white light source with a 0.5° divergence angle, spot diagrams were generated by re-imaging the output plane corresponding to the location of cells in a complete MEPV assembly onto a camera detector. The beam profile for on-axis illumination is shown in Fig. 8 and demonstrates good agreement with ray-trace simulations under the AM1.5G spectrum from 400 to 2000 nm. Precise intensity and field of view measurements will be performed by subsequent outdoor measurements under the sun. However, beam profile agreement with the design infers that the 100X magnification specification has been realized. Total transmitted optical power emerging from the rear surface of the optical sub-assembly into air has also been measured in a spectrophotometeracross a spectrum from 400 to 2000 nm. The maximum simulated transmission through the sub-assembly is 84% due to Fresnel and absorption losses, which agrees well with the data in Fig. 9.It should be noted that the tested optics have an estimated 5% Fresnel loss at their output from the polycarbonate rear lens array into air. This loss will be essentially eliminated in MEVP assemblies using an index matched adhesive between the rear optic array and the cells. Therefore, it is reasonable to expect system transmission levels approaching the 90% design target.
5. Summary and future work
Sandia National Laboratories has been developing microsystems-enabled photovoltaics as a novel, low cost photovoltaic architecture which combines the high conversion efficiencies of concentrated photovoltaics (CPV) with the form factor and low system costs of flat panel PV. Work has described the design, fabrication and testing of a 100X magnification, ± 2.5° field of view micro-concentrating optical array for a second generation prototype using 250 µm diameter multi-junction PV cells. Prototype system assembly and testing remains underway, as design efforts are focused on the next generation prototype for further reductions in system cost and complexity, for improved cell efficiency and performance, and for a demonstration of cost effective manufacturing methodologies.
Sandia National Laboratories is a multi-program laboratory managed and operated by Sandia Corporation, a wholly owned subsidiary of Lockheed Martin Corporation, for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-AC04-94AL85000. This document has been reviewed and approved for unclassified, unlimited release under SAND2014-0304J.
References and links
1. V. Gupta, G. Nielson, M. Okandan, and J. Cruz-Campa, “Microsystems-enabled photovoltaics: A path to the widespread harnessing of solar energy,” Fut. Photovoltaics (May 2010).
2. G. Nielson, M. Okandan, J. Cruz-Campa, A. Lentine, W. Sweatt, B. Jared, P. Resnick, B. Kim, B. Anderson, V. Gupta, A. Tauke-Pedretti, J. Cederberg, T. Gu, M. Haney, S. Paap, C. Sanchez, C. Nordquist, M. Saavedra, M. Ballance, J. Nguyen, C. Alford, and J. Nelson, “Cell microconcentrator module with moderate concentration, ±4° acceptance angle, and 13.3 mm focal length,” in Proceedings of PVSC (IEEE, Tampa Bay, FL, 2013).
3. S. Paap, V. Gupta, J. Cruz-Campa, M. Okandan, W. Sweatt, B. Jared, B. Anderson, G. Nielson, A. Tauke-Pedretti, and J. Nelson, “Cost analysis for flat-plate concentrators employing microscale photovoltaic cells,” in Proceedings of PVSC (IEEE, Tampa Bay, FL, 2013).
4. A. Tauke-Pedretti, J. Cederberg, G. Nielson, J. Cruz-Campa, C. Sanchez, C. Alford, M. Okandan, E. Skogen, and A. Lentine, “Resistance considerations for stacked small multi-junction photovoltaic cells,” in Proceedings of PVSC (IEEE, Tampa Bay, FL, 2013).
5. R. Winston, J. Miñano, and P. Benítez, Non-imaging optics, (Elsevier, 2005).
6. P. Benítez, J. Miñano, P. Zamora, R. Mohedano, A. Cvetkovic, M. Buljan, J. Chaves, and M. Hernández, “High performance Fresnel based photovoltaic concentrator,” Opt. Express 18(S1), A25–A40 (2010). [CrossRef]
7. M. Hernández, P. Benítez, J. Miñano, A. Cvetkovic, R. Mohedano, O. Dross, R. Jones, D. Whelan, G. Kinsey, and R. Alvarez, “XR: A high-performance PV concentrator,” Proc. SPIE 6649, 664904 (2007). [CrossRef]
8. B. Jared, M. Saavedra, B. Anderson, W. Sweatt, G. Nielson, and M. Okandan, “Micro-optic fabrication for microsystems-enabled photovoltaics”, in Proceedings of American Society of Precision Engineering, (ASPE, St. Paul, MN, 2013).