Microsystems-Enabled Photovoltaics: a Path to the Widespread Harnessing of Solar Energy
Microsystems-Enabled Photovoltaics: A Path to the Widespread Harnessing of Solar Energy
Vipin Gupta, Jose Luis Cruz-Campa, Murat Okandan, Gregory N. Nielson Sandia National Laboratories
Abstract If solar energy is ever going to become a mainstream power source, the tech- nologies for harnessing sunlight have to become cheaper than all other forms of energy, be easy and quick to install, and work more safely, reliably and durably than present-day grid power. Our research team is striving to make this happen by utilizing microdesign and microfabrication techniques used in the semiconductor, LCD and microsystem industries. In this article, we describe microsystems-enabled photovoltaic (MEPV) costs associated with this module assem- concepts that consist of the fabrication of bly approach in conjunction with optical micro-scale crystalline silicon and GaAs concentration can be well below solar cells, the release of these cells into a $1/Wattpeak while retaining the superior photovoltaic (PV) “ink” solution, and the conversion efficiency and durability of printing of these cells onto a substrate crystalline silicon and III-V materials. using fluidic self-assembly approaches. So far, we have produced 10 percent effi- Our Vision cient crystalline GaAs cells that are 3 µm With the emerging electrification of thick and 14.9 percent efficient crystalline personal transportation, decentralization silicon cells that are 14 µm thick. The of energy generation in places that lack a
| Future Photovoltaics | May 2010 NEW TECHNOLOGIES & MATERIALS
reliable electricity grid, growing concerns low-cost packaging substrates, creating about atmospheric emissions from fossil scalable high-speed assembly processes, fuel use, and ever-increasing global and establishing fast methods to install demand for electricity, there is an insa- and wire the grid-connected circuitry. tiable need for point-of-use technologies Several groups across the world,[1-4] that generate electricity in a clean way. including ours, have developed prototype Our group is striving to make microsys- small and ultrathin photovoltaic cells that tems-enabled photovoltaics (MEPV) the reduce PV material use dramatically. most efficient, versatile and inexpensive These ultrathin cells perform nearly as way to provide that electricity. We envi- well as thick ones[4-6] that are much sion high-concentration, sun-tracking thicker than what is necessary for photo- MEPV for power utilities; low-concentra- electric conversion simply to minimize tion, flat-plate MEPV for horizontal breakage during cell processing, handling rooftops; and flexible, integrated MEPV for and assembly. sloped rooftops, vehicles, gadgets and the Figure 1a depicts the normalized cell human body. efficiency (efficiency/potential efficiency) Many things have to be accomplished vs. the cell thickness. It shows that ultra- before photovoltaics become a ubiquitous thin cells can substantially decrease the technology: reducing the amount of nec- cost associated with the use of silicon essary high-cost semiconductor material, material while retaining 90 percent of the developing highly efficient cells, achieving potential conversion efficiency. This long-term durability and reliability, using graph assumes a single light pass through
1 0.7 0.9 0.6 0.8 current thickness 0.7 0.5 i) microlens array 0.6 most 0.4 0.5 sensible thickness 0.4 0.3 ii) high- efficiency cells 0.3 0.2 0.2 iii) interconnect layer
Efficiency/Potential Efficiency Efficiency/Potential 0.1 0.1 iv) inexpensive and Silicon contribution to cost ($/Watt) 0 0 flexible substrates 0 200 400 Thickness (µm) a) b)
Figure 1 – a) Cost and potential conversion efficiency in silicon solar cells as a function of thickness; b) the microsystems-enabled photovoltaic module concept showing the four distinct layered components
www.futurepv.com | Microsystems-Enabled Photovoltaics: A Path to the Widespread Harnessing of Solar Energy
the cell; adding back reflectors and light thicknesses below 10 µm.[6] trapping structures would further Our method integrates multiple tech- enhance the conversion efficiency and nologies in an effort to produce an inex- make the knee of this curve move to pensive module. Figure 1b shows the four
Step 1: Cells are processed on top Step 2: Cells are detached from the of a handle wafer using only 10-20 µm wafer leaving the handle available of material per batch. to produce more cells.
Step 3: Cells (shown as floating gray Step 4: Cells are embedded in hexagons) are attracted to the desired low-cost substrates with contacts positions (shown as red dots) using and microlenses. self assembly concepts.
Figure 2 – Process to Create Photovoltaic Sheets of Micro Cells From Cell Liftoff to Final Assembly
| Future Photovoltaics | May 2010 NEW TECHNOLOGIES & MATERIALS
main parts of our approach: i) a microlens cells with consistent IV parameters, high array that concentrates the light onto the efficiencies and superior robustness com- cells; ii) the ultrathin, small and highly pared to conventional large-area thin efficient solar cells; iii) an interconnection cells. All of the manufacturing steps (pat- layer; and iv) the entire system encapsu- terning, diffusion, passivation and metal- lated in flexible substrates. ization) are done while the micro cells are still attached to the substrate wafer to Benefits of our Approach facilitate handling. Then, thousands of Our approach uses tiny solar cells that ultrathin micro cells are detached from are 14-20 µm thick and 250 µm to 1 mm in the wafer by means of a sacrificial layer or lateral dimensions. This method enables an anisotropic etch (liftoff technique) and three main ways of reducing the PV mod- released into a solution. Only 15-25 µm of ule costs[7]: by making the cells with a the handle wafer is consumed in each small lateral dimension, by creating ultra- batch and the rest is left available to be thin cells and by using micro-concentra- reused and produce another lot of micro tors. The small lateral size enables better cells. Each substrate wafer (e.g., c-Si, use of the available wafer area (i.e., reduc- GaAs, Ge) thus has the possibility of being tion in edge die loss), increases area usage reused many times rather than being con- through the production of hexagonal cells, sumed in a single fabrication run. and gives semiconductor manufacturers The micro size subsequently requires the flexibility to use any wafer size. novel and workable parallel self-assem- Thinning the cells reduces semiconductor bly methods[8] to place the cells onto material costs, improves carrier collection designated spots. These assembly con- and potentially achieves higher open cir- cepts use energy minimization approach- cuit voltages. The small size also enables es to make arrays of the cells on inexpen- the cells to be processed with IC fabrica- sive substrates. The cells can be embed- tion tools, allowing near-ideal cell per- ded in low-cost flexible materials that formance (>20 percent for c-Si, >40 per- produce virtually flat modules. Figure 2 cent for III-V multijunction cells). Finally, shows the procedure used to create PV micro-optical elements concentrate sun- sheets using this technology. light onto each cell, further reducing the In addition to the benefits obtained at need for high-cost semiconductor materi- the cell level, there are multiple advantages als. All together, this three-pronged at the module and system level due to the approach decreases the need for high-cost scale of the cells. Small cell size allows the materials, reduces overall system cost and use of inexpensive refractive optics[2] increases conversion efficiency per gram instead of Fresnel optics, giving better opti- of utilized PV material. cal efficiencies; reduced thickness and Another advantage of our method is small lateral cell size simplifies the man- that it utilizes mature technology, tools agement of thermal loads on the PV mod- and techniques used in the production of ules (which is particularly important for microelectronics and microsystems. This optical concentrating designs); the micro- enables the production of micro solar scale PV concentrator makes high-accura-
www.futurepv.com | Microsystems-Enabled Photovoltaics: A Path to the Widespread Harnessing of Solar Energy
cy and high-bandwidth tracking possible, racy during windy conditions (due to the thereby reducing tracking cost and com- high-speed response of the tracker)[7,10]; plexity as well as providing pointing accu- and finally, small form factor (i.e., thin) con-
Figure 3 – Pictures of our released solar photovoltaic cells: a) Scanning electron microscope (SEM) image of a 250 µm wide silicon solar cell released through an anisotropic KOH under-etch with lateral interdigitated fingers; b) SEM of 250 µm wide silicon solar cell released using a buried silicon dioxide layer with radial contacts; c) 250 µm GaAs solar cell released using a buried AlAs layer; d) flexible mechanical model with embedded cells and micro optics
| Future Photovoltaics | May 2010 NEW TECHNOLOGIES & MATERIALS
centrator modules with integrated in-plane use of a buried silicon dioxide sacrificial trackers can be flat mounted at the point of layer. In the case of GaAs cells, the release use for the generated electricity. sacrificial layer used is AlAs. The designs of our solar cells include: Our Results to Date i) a silicon back-contacted radial pattern; Our technique produced ultrathin ii) a silicon back-contacted interdigitated solar cells with lateral dimensions rang- format; and iii) an ultrathin, single-junc- ing from 250 µm to 1 mm. We have uti- tion GaAs cell with back contacts. Figure 3 lized a variety of materials, approaches shows optical and scanning electron and designs, including single crystal sili- microscope (SEM) images from the fin- con in (111) or (100) crystallographic ori- ished cells using the different materials, entations and single-crystal GaAs. The approaches and designs as well as a approaches differ in the chemistry and mechanical model with the cells embed- technique used to detach the wafers from ded in an inexpensive, flexible substrate the cell: (111) oriented silicon wafers with an integrated microlens array. require a potassium hydroxide (KOH) In order to explore the commercial anisotropic etch that undercuts the cell, potential of MEPV technology, a cost analy- while (100) oriented silicon wafers make sis was carried out assuming a solar flux of
4.00 35 15% efficient cells 3.50 30 )
25% efficient cells 2 3.00 25
2.50 20
2.00 Current $/W prices 15
1.50 10
5 1.00 Current density (mA/cm $/Watt-peak manufactured $/Watt-peak 0.50 0 0 0.2 0.4 0.6 0.8 1 0.00 Voltage (V) 0 25 50 75 100 14.85% efficient silicon solar cell Optical concentration (# suns) 10% efficient GaAs solar cell
a) b)
Figure 4 – a) Cost analysis of our technology: $/Watt vs. concentration for different cell efficiencies assuming a $25/m2 assembly cost; b) experimental J-V measurement curves of our micro cells
www.futurepv.com | Microsystems-Enabled Photovoltaics: A Path to the Widespread Harnessing of Solar Energy
100mW/cm2, a process cost of $150/wafer, a 3 µm thick crystalline GaAs cells along wafer diameter of 200 mm, a $25/m2 with the 14.9 percent efficient, 14 µm assembly cost, and 95 percent overall cell thick crystalline silicon cells developed process yield. This analysis revealed that by our team suggest that there are break- the costs of our technology approach could through possibilities along this pathway. be well below $1/Wattpeak by using moder- Combining these experimental results ate concentration (below 50x) and massive plotted in Figure 4b with the parallel self- parallel assembly. Figure 4a graphically assembly approach illustrated in Figure 2 shows the results of these calculations. as well as the cost analysis shown in So far, our technology has produced Figure 4a, there appears to be a reason- 14.9 percent conversion efficiency for 14 able probability of developing a MEPV µm (13.7 +/- 0.38 µm) thick cells made manufacturing approach that is as low or from crystalline silicon and 10 percent perhaps lower in U.S. dollar per Watt- conversion efficiency for 3 µm thick cells peak cost than all other forms of energy, made of GaAs.[11] This level of perform- including PV thin film technologies. ance is comparable to the conversion effi- Contrary to the popular belief espoused ciency in available commercial silicon- by futurist thinker Tom Friedman, the based modules with PV cells that are 10- world is not flat.[13] It is filled with the var- 15x thicker. Figure 4b shows the current ious shapes and curved contours of natu- density-voltage curves (J-V) for two of our ral terrain, large manmade structures, cells. Both cells were tested with a solar portable electronics, vehicles and the simulator normalized to 1000W/m2 using human body. All of these prospective PV a silicon reference solar cell. hosts are becoming ever more ready to adopt efficient, versatile and inexpensive Conclusions photovoltaic devices for powering impor- In the United States, solar energy gen- tant things. Microsystems-enabled PV can eration, including photovoltaics, present- provide those devices, and our research ly produces 0.1 percent of the energy team is committed to doing our part to used for electricity, heating and trans- develop the technological means of mak- portation. If PV adoption is to grow at a ing it so. compounded annual growth rate of 40 percent per year and thereby become a 40 Acknowledgments percent contributor to the U.S. and global Sandia is a multiprogram laboratory energy portfolio within one human gen- operated by Sandia Corporation, a eration,[12] a series of technology break- Lockheed Martin Company, for the U.S. throughs are needed in PV cell develop- Department of Energy’s NNSA under con- ment, assembly, installation, operation tract DE-AC04-94AL85000. This work was and end-of-life retirement. sponsored by the DOE Solar Energy The microsystems-enabled photo- Technology Program PV Seed Fund. voltaic concepts described in this article The authors would like to thank the show one technology pathway to realiz- following MEPV team members for their ing this vision. The 10 percent efficient, contributions to this research: Catalina
| Future Photovoltaics | May 2010 NEW TECHNOLOGIES & MATERIALS
Ahlers, Michael Busse, Craig Carmignani, (2009) 1816-1821. Peggy Clews, Anton Filatov, Jennifer 5. T.H. Wang, T.F. Ciszek, C.R. Granata, Robert Grubbs, Rick Kemp, Schwerdtfeger, H. Moutinho, R. Judith Lavin, Tom Lemp, Tony Lentine, Matson, “Growth of silicon thin layers Kathy Meyers, Jeff Nelson, Mark on cast MGSi from metal solutions for Overberg, David Peters, Tammy Pluym, solar cells,” Solar Energy Materials and Paul Resnick, Carlos Sanchez, Carrie Solar Cells 41-42 (1996) 19-30. Schmidt, Lisa Sena-Henderson, Jerry 6. M.A. Green, “Limiting Efficiency of Simmons, Mike Sinclair, Constantine Bulk and Thin-Film Silicon Solar Cells Stewart, Jason Strauch, Bill Sweatt, in the Presence of Surface Recombi- Benjamin Thurston, George Wang, Mark nation,” Progress in Photovoltaics: Wanlass, and Jonathan Wierer. Research And Applications 7 (1999) 327-330. References 7. G.N. Nielson, M. Okandan, P. Resnick, 1. W.P. Mulligan, A. Terao, D.D. Smith, P.J. J.L. Cruz-Campa, P. Clews, M. Wanlass, Verlinden, R.M. Swanson, “Develop- W. Sweatt, E. Steenbergen, V. Gupta, ment of chip-size silicon solar cells,” “Microscale PV cells for concentrated Conference record on the 28th IEEE PV applications,” Conference record of Photovoltaic Specialist Conference the 24th EU PVSEC (2009) 170-173. (2000) 158-163. 8. S. Shet, R.D. Revero, M.R. Booty, A.T. 2. J. Yoon, A.J. Baca, S.I. Park, P.S. Elvikis, Fiory, M.P. Lepselter, N.M. Ravindra, J.B. Geddes III, L. Li, R.H. Kim, J. Xiao, S. "Microassembly Techniques: A Review," Wang, T.-H. Kim, M.J. Motala, B.Y. Ahn, Conference record on the Materials E.B. Duoss, J.A. Lewis, R.G. Nuzzo, P.M. Science and Technology 1 (2006) 451- Ferreira, Y. Huang, A. Rockett, J.A. 470. Rogers, “Ultrathin silicon solar micro 9. J.L. Cruz-Campa, M. Okandan, M. cells for semitransparent, mechanical- Busse, G.N. Nielson, “Microlens rapid ly flexible, and microconcentrator prototyping technique with capability module designs,” Nature Materials 7 for wide variation in lens diameter and (2008) 907-915. focal length,” Microelectronic Engi- 3. K.J. Weber, A.W. Blakers, M.J. Stocks, J.H. neering (2010) in press. Babaei, V.A. Everett, A.J. Neuendorf, P.J. 10. W.C. Sweatt, B.H. Jared, G.N. Nielson, Verlinden, “A Novel Low-Cost, High- M. Okandan, M.B. Sinclair, A. Filatov, A. Efficiency Micromachined Silicon L. Lentine, “Micro-Optics for High- Solar Cell,” IEEE Electron Device Letters Efficiency Optical Performance and 25 (2004) 37-39. Simplified Tracking for Concentrated 4. G.N. Nielson, M. Okandan, P. Resnick, Photovoltaics (CPV),” OSA International J.L. Cruz-Campa, T. Pluym, P. Clews, E. Optical Design Conference (IODC), in Steenbergen, V. Gupta, "Microscale c-Si press. (c)PV cells for low-cost power," 11.J.L. Cruz-Campa, G.N. Nielson, M. Conference record on the 34th IEEE Okandan, M. Wanlass, C. Sanchez, P. Photovoltaic Specialists Conference Resnick, P. Clews, T. Pluym, V. Gupta,
www.futurepv.com | Microsystems-Enabled Photovoltaics: A Path to the Widespread Harnessing of Solar Energy
“Back contacted and small form factor received 12 awards and scholarships for his GaAs solar cell,” Conference record of academic achievements, and is president of the IEEE 35th Photovoltaic Specialists the MRS chapter at the University of Texas El Conference (2010) in press. Paso. He has presented his research in 12 con- 12. S. Dwyer, “Solar Growth Rate Projec- ferences, on local television, and has authored tions to Make Solar Mainstream in the and co-authored six technical publications. United States,” DOE Tiger Team Technical Memorandum, August 3, Murat Okandan is an electrical microsystems 2009, 1-3. engineer. His work has focused on solid state 13.Friedman, Thomas L., “The World Is device physics, device design, microelectronics Flat: A Brief History of the Twenty-First processing, and sensors. Murat has authored Century,” Farrar, Straus and Giroux and co-authored over 30 technical publica- (2005). I tions, holds 12 patents and has seven patents pending. He received his Ph.D. from Pennsylvania State University in electrical About the Authors engineering.
Vipin Gupta is a systems engineer. His work Gregory N. Nielson is an optical microsys- focuses on system functionality, technology tems engineer. He is known for successfully innovation, solar technical assistance, human demonstrating new techniques for optical factors engineering, group development and MEMS switching that led to world-record dynamic organizational behavior. Vipin earned switching speeds. Since serving as a Truman his Ph.D. in applied physics from Imperial Fellow at Sandia National Laboratories, College London as a U.S. Marshall Scholar. Gregory has explored ways for microsystems to enable new solar technologies. He received Jose Luis Cruz-Campa is a Ph.D. student his Ph.D. from MIT in optical microsystems. intern. His research explores the scaling effect in silicon and thin film PV. Jose Luis has
| Future Photovoltaics | May 2010