Novel light management for ultrathin solar cells
Transcription
Novel light management for ultrathin solar cells
Novel light management for ultrathin solar cells Ruud E.I. Schropp Energy research Centre of The Netherlands & TU/e Eindhoven INC10 May 15, 2014 Gaithersburg MD, USA www.ecn.nl 2 Partners in Solliance (www.solliance.eu) • ECN, the leading energy research institute in the Netherlands; thin film research group in Eindhoven • TNO, the leading Dutch institute for applied scientific research • Holst Center, a joint research initiative of imec and TNO (and ECN for PV); roll‐ to‐roll processes, mainly oriented towards OPV on flexible substrates • TU/e, Eindhoven University of Technology, fundamental and technological research in PV • imec, a world‐leading research institute in nanoelectronics, based in Flanders, participates with its OPV and CIGS PV research groups • Forschungszentrum Jülich, is one of the large interdisciplinary research centres in Europe. The Photovoltaics section of the Institute of Energy and Climate Research (IEK‐5) belongs to the world leading research institutions in the range of thin‐film photovoltaics based on amorphous and nanochrystalline silicon. 3 Outline • Key Challenges in Thin Film Photovoltaics • Light trapping (c.q. “concentration”) • Utilization of nanostructure in Photovoltaics: – Conventional “random” textures and designed textures – Nanorod, nanowire structures • Conclusion 4 15‐5‐2014 5 Status of thin film solar cells (laboratory ~ 1 cm2) 2014/05 • Thin film Si: – Single junction: µc‐Si:H 10.7% (EPFL Neuchâtel); a‐Si:H 10.1% (Oerlikon/TEL Solar) – Double junction (tandem): a‐Si:H/µc‐Si:H 12.3% (Kaneka); 12.3% (Oerlikon Solar) – Triple junction: a‐Si:H/µc‐Si:H/µc‐Si:H 13.44% efficiency (LG Electronics) • CIGS: – Single junction: 20.8% (rigid; ZSW); 20.4% (flexible; EMPA) • CdTe: – Single junction: 20.4% (First Solar & GE Global Research partnership) • Perovskite – Single junction 17.9% (KRICT) • OPV: – Double junction 12.0% small molecule cell (Heliatek) – Single junction 11.1% Mitsubishi Chemical – Double junction: 10.6% UCLA‐Sumitomo Chemical 6 Installed PV capacity - world 7 Installed PV capacity - Netherlands 2013: 700 MWp (0.5% of total electricity consumption) 8 PV system prices (Germany) 9 Grid integration Germany explores (and shifts) the frontiers M. Lippert, SAFT Source: Fraunhofer ISE (2013) 10 Commercial module efficiencies (selection) wafer Si IBC wafer Si IBC wafer Si IBC wafer Si HIT wafer Si mono wafer Si multi CIGS tf a/µcSi M.J. de Wild‐Scholten SmartGreenScans (June 2013) tf aSi CdTe Thinner is beautiful 1. (local) concentration of light Better ratio of photocurrent over dark current higher Voc’s* Concentration can be achieved in two ways • Optical concentration outside the cell • Concentration of light inside the cell by internal light trapping structures, such as diffractive and plasmonic structures. In both cases: IL photogenerated current I0 reverse saturation current 13 Thinner is beautiful 1. Better ratio of photocurrent over dark current higher Voc’s* 2. Collection of carriers also improves higher FF’s 3. Better stability and thus higher stabilized efficiency (if Jsc is maintained). 4. Better cycle time (lower cost of ownership) 5. Lower materials consumption – further cost reduction. R.E.I. Schropp Key Challenges • reduce cost (by thinning down) higher efficiency • materials resources • Light trapping Better spectral matching (multijunction, photon manipulation) use only abundant, environmentally benign materials 15 Abundancy table Various random light-scattering morphologies (front TCO) Asahi U Type LPCVD ZnO PVD ZnO+ Wet etch Front TCO Random (bottom‐up) morphologies Haze Challenge: increase the haze at long wavelengths, and scatter light into large angles. M. Zeman et al. , MRS Proc. Ser. (2010) Scalable top down technology: Nanoimprint Lithography Nanoimprint Lithography (NIL) • Nano‐imprinting offers great perspectives to implement sub‐micron light management structures in solar cells in a cost effective manner. • Can periodic textures outperform random textures? Periodic texture ‐ Random texture (ZnO) Solution growth (ZnO) Two examples of periodic textures 20 1. Light trapping by surface plasmons • Design of plasmonic nanoparticles patterns for optimum waveguiding. Fraction scattered into substrate highest for cylinder & hemisphere:strongest near-field coupling Ohmic damping V, scattering V2, Tradeoff: larger size more scattering, but lower coupling Atwater & Polman, Nature Mater. 9, 205 (2010) metal nanostructures controllably couple light to in-plane waveguide modes Substrate Conformal Imprint Lithography PDMS Stamp Thin glass PDMS stamp (6”) is bonded on 200 µm AF-45 glass 1 m Full-wafer soft nano-imprint • Flexible rubber on thin glass • Conform to substrate bow and roughness • No stamp damage due to particles Marc Verschuuren, Hans van Sprang Spring MRS 2007, 1002-N03-05 Electromagnetic Simulation for Generation Rate Calculations simulation FDTD simulations at discrete wavelengths across the spectrum. Assume that one photon absorbed produces one electron: AM1.5g Spectrum for SR: generation rate at each λ for Jsc: solar spectrum weighting around each simulated point V. E. Ferry, et al. Opt. Express, 18, A237-A245 (2010). Vivian Ferry 24 Comparing Random and Designed Surfaces = 500 nm = 670 nm 1016 Nanopatterned 500 nm = 500 nm = 670 nm 1014 1013 Photon Flux (cm-3 sec-1) 1015 1012 Asahi Controlled spatial curvature of metal layer is critical to avoid losses and increase V. E. Ferry, et al. IEEE PVSC Proceedings (2010). photocurrent Vivian Ferry 25 Periodic arrays Penrose Replicated Asahi Pseudo-random 2D FT Designed classes of nanoparticles Design constraints: - 200 nm, 290 nm, 310 nm diameter - Ag particles should not touch - Same packing fraction as the 400 nm periodic pattern Rounded metal nanostructures: Less parasitic absorption Cell Fabrication 90 nm active layer 9.5% ITO a-Si:H ZnO:Al 500 nm Ag SiO2 sol-gel Cells deposit conformally over nanopatterned back contacts: best pseudorandom pattern and periodic pattern have similar efficiencies Vivian Ferry 27 Broadband Light Trapping: Experiment vs. Simulation Experiment: 90 nm i-layers Mie Simulation FDTD waveguide modes - Significant broadband photocurrent enhancement - Good agreement between simulation and experiment Vivian Ferry 28 2. Light trapping by scalable periodic texture (with µc-Si rather than a-Si cells) c-Si nip cells on steel foil with various back contacts: 1. Flat back contact 2. BC with 2D periodic texture 3. BC with random texture (replica of SnO2:F) (EU FP-7 Energy 2009 – 241477 project “Silicon-Light”), Coordinator W. Soppe (ECN). www.silicon-light.eu 29 Light management with R2RR Nano-Imprint Lithography 1. Modeling 2. Mastering 3. Tooling 4. Imprinting 30 Single junction c-Si cells: Jsc Periodic textures can outperform random textures 31 Single junction µc-Si cells (750 nm): spectral response 32 Single junction uc-Si cells: Fill Factor Higher FF for periodic texture! 33 Lower FF on random textures due to shunting R.E.I. Schropp et al. Journal of Crystal Growth 311, 760 (2009) Microcracks 34 Conformal growth without microcrack formation on periodic texture Dark field TEM (one selected diffraction direction) Bright field TEM ITO c Si ZnO Ag HR-TEM by Martial Duchamp (FZ Jűlich) 35 11% efficiency reference cell produced on nanotextured metal foil in a roll-to-roll process 350 nm top cell; 1000 nm bottom cell Best cell on CB‐R Best cell on CB‐WR Jsc top cell Jsc bottom cell FF [%] Efficiency Voc [mV] [mA/cm2] [mA/cm2] [%] 1381 11,84 12,31 67,0 11,0 1366 11,90 13,24 65,7 10,7 ECN labs @ Eindhoven 36 Roughness of which layer is important? 90 nm active layer ITO a-Si:H ZnO:Al 500 nm Ag SiO2 sol-gel Where should the roughness be? AZO 80 or 500 nm AZO Ag Asahi Which roughness is important: that of the ZnO or that of the Ag? Claire van Lare, Albert Polman 15‐5‐2014 Frank Lenzmann 38 Periodic structures - simulations • No structure in AZO poor light trapping Claire van Lare Periodic structures - simulations • No structure in AZO poor light trapping • Flat Ag can be better than structured Ag Claire van Lare Absorption in Ag layer • Structuring Ag leads to increased parasitic absorption Claire van Lare Photonic “roughness” rather than geometric roughness? 15‐5‐2014 43 FLiSS made by polishing down 2D ZnO:Ga grating filled with n-type nc-Si:H Voc with FLiSS Voc with flat reference substrate Chemical mechanical polishing H. Sai et al. Applied Physics Letters 98, 113502 (2011) Plasmonic Flat Scattering Surfaces: Lourens van Dijk Simulation of Absorption enhancement Scattering Design Flat Lourens van Dijk Absorption Si SiO2 Ag Nanocylinder or nanowire Yinghuan Kuang, et al., IOP Reports on Progress in Physics, 2013 (accepted for publication) Avoiding the trade-off between “optically thick” and “electronically thin” Light direction Traditional textured thin film silicon solar cells Current direction Glass TCO Si p-i-n ZnO Ag/Al Trade-off: Optically thick vs Electrically thin Nanostructured three dimensional (nano-3D) solar cells Yinghuan Kuang, et al., IOP Reports on Progress in Physics, 2013 (accepted for publication) R.E.I. Schropp 50 Avoiding the trade-off between “optically thick” and “electronically thin” Light direction Current direction +- SUPER scattering! Nanostructured three dimensional (nano-3D) solar cells Yinghuan Kuang, et al., IOP Reports on Progress in Physics, 2013 (accepted for publication) R.E.I. Schropp 51 Avoiding the trade-off Light direction Elongated nanostructures: “Nanorod solar cell” Current direction Orthogonalize charge carrier path and photon path Super scattering (between nanorods) Anti-reflection texture Yinghuan Kuang, et al., Journal of Non-Crystalline Solids, 358, 17, 2209-2213 (2012) Challenges Conformal coverage (internal shunting paths low FF). Surface and interface recombination lowR.E.I. Voc Schropp Nanostructured three dimensional (nano-3D) solar cells Yinghuan Kuang, et al., IOP Reports on Progress in Physics, 2013 (accepted for publication) 52 Conformal n-i-p coverage by Hot Wire CVD; “nano-3D” cells • • • • • • ITO and grid p‐layer (PECVD, B(CH3)3) i‐layer (HWCVD, 25 nm) n‐layer (PECVD, PH3) ZnO:Al (38 nm) Ag (20 nm) nano‐3D cells - Thicknesses determined by HRSEM - For optimal conformal coverage, essential to use HWCVD Yinghuan Kuang et al., Appl. Phys. Lett. 98 (2011) 113111. doi:10.1063/1.3567527 Synthesis of ZnO nanorods by chemical bath deposition (Zn(CH3COO)2·2H2O) + C6H12N4 Hot Wire CVD Set-up at Utrecht University in Eindhoven Two Ø0.5 mm Ta filaments Substrate to filaments distance: 55 mm. Distance between filaments: 40 mm. Temp. at filaments: ~1750°C Temp. at substrate: ~200°C 5 Introduction: Hot wire CVD for conformal coverage on high aspect ratio structures Qi Wang et al., NREL and Applied Materials Inc., Appl. Phys. Lett. 84 (2004) 338 Makiko Kitazoe, Shuuji Osono, Hiromi Itoh, Shin Asari, Kazuya Saito and Masahiro Hayama ULVAC, Japan, 3rd Hot-Wire CVD Conference, Utrecht Y. Kuang et al., Appl. Phys. Lett. 98, 113111 (2011). 500 nm 4 Cell results • Flat, 75 nm • Asahi texture, 75 nm • Nanorod cell, 25 nm 8.3 mA/cm2 from a 25‐nm layer! Nanorod type solar cell with 25 nm i-layer has higher Jsc than a flat or even a textured solar cell with 75 nm i-layer; slight trade off with Voc. Yinghuan Kuang et al., Appl. Phys. Lett. 98 (2011) 113111. doi:10.1063/1.3567527 ZnO nanorod substrate for thin film solar cells rms:68 Nanorod+Ag+ZnO Shorter rods to prevent Voc loss rms:37 rms:13 Quite different from Asahi‐U Asahi U-type +Ag+ZnO 7/14 Latest results Cell type i-layer Jsc (mA/cm2) thickness (nm) Flat 3D100 Flat 3D200 0.8 9.1 13.3 11.5 15.0 100 100 200 200 Voc (V) FF η (%) 0.89 0.86 0.89 0.85 0.70 0.62 0.63 0.65 5.7 7.1 6.4 8.4 200 nm EQE 0.6 0.4 Flat NR 0.2 0.0 400 500 600 700 800 Wavelength (nm) Yinghuan Kuang et al., J. Non‐Cryst. Solids 358 (17) (2012) 2209‐2213 10/14 Conclusions • New insight in scattering back reflectors for thin film solar cells • Plasmonic and dielectric scatterers have different effects. • Index contrast for geometrically flat scattering. • These concepts are of interest to all thin film solar cells. Large area manufacturing methods need to be developed. •Nanowire/nanorod enhancement schemes can reduce cost of photovoltaically generated kWh’s. • Approaches for non-lithographic random nanorod-type solar cells are very promising. Acknowledgments Yinghuan Kuang, Lourens van Dijk, Pim Veldhuizen, Marcel Di Vece, Jatin K. Rath W. Soppe, Albert Polman M. Dörenkämper, Claire van Lare N.J. Bakker C.H.M. van der Werf 15-5-2014 Chinese 62 Scholarship Yinghuan Kuang Lourens van Dijk Council