A brief overview ofPlasmonic Nanostructure Design for Efficient Light Coupling into Solar CellsV.E. Ferry, L.A. Sweatlock, D. Pacifici, and H.A. Atwater, Nano Letters, 8 4391Douglas DetertEE235Prof. Connie ChangMarch 2, 2009Subscriber access provided by - Access paid by the | UC Berkeley LibraryNano Letters is published by the American Chemical Society. 1155 SixteenthStreet N.W., Washington, DC 20036LetterPlasmonic Nanostructure Design forEfficient Light Coupling into Solar CellsVivian E. Ferry, Luke A. Sweatlock, Domenico Pacifici, and Harry A. AtwaterNano Lett., 2008, 8 (12), 4391-4397• DOI: 10.1021/nl8022548 • Publication Date (Web): 14 November 2008Downloaded from http://pubs.acs.org on February 23, 2009More About This ArticleAdditional resources and features associated with this article are available within the HTML version:• Supporting Information• Access to high resolution figures• Links to articles and content related to this article• Copyright permission to reproduce figures and/or text from this articleDouglas Detert — EE235 — March 2, 2009Solar Cell Design/Material Considerations• Conventional solar cells (e.g. Silicon) require thick absorption layers for complete absorption• Thin film solar cells (e.g. CdTe, CIGS) decrease bulk recombination effects and allow for higher quality absorber materials• Problem: Thin film cells are limited by decreased absorption, carrier excitation, and photocurrent• Solution: Texture top/bottom surfaces to enhance light absorptionDouglas Detert — EE235 — March 2, 2009Surface Plasmon Polariton Enhanced Solar Cells• Surface Plasmon Polaritons (SPPs) are collective oscillations of free electrons at metal/dielectric boundaries• SPPs are highly localized to interfaces and propagate easily for microns. Energy in SPP modes enhances absorption• Momentum mismatch between incident light and SPPs does not allow for direct excitation of SPPs• Goal: Design a nanostructure back contact that scatters light into SPP modeBarnes. J Opt A-Pure Appl Op 8 S87-S93 (2006)Douglas Detert — EE235 — March 2, 2009Scattering From a Single Groove• Light energy is scattered into two key modes• Photonic (~semiconductor) • SPP (~interface)• Both enhance photoabsorption, but photonic modes are not supported in extremely thin structuresHyDouglas Detert — EE235 — March 2, 2009Results: Scattering From a Single Groove• Finite-difference time-domain (FDTD) simulations paired with modal decomposition analysis• Three physical effects involved in incoupling efficiencies:• Fabry-Pérot resonance of thin film• Photonic mode excitation at SPP resonance wavelength• Polarization resonance of scatterer• Film thickness and scatterer geometry affect above propertiesDouglas Detert — EE235 — March 2, 2009Effect of Groove Dimensions• Groove width: SPP modes break down at large groove sizes, photonic mode flattens out• Groove depth has little effect on incoupling efficiency • Ridge-like structure: enhances photonic modeeffective scattering cross section is proportional to thephysical width w with ratio σ ) aw. This relationship breaksdown for larger grooves, which behave less like quasi-staticdipole scatterers. Above w ) 200 nm the edges of the grooveact independently, and there is no further gain from increas-ing size. The incoupling to SPP modes is strongly affectedby this dipole character; the SPP cross section peaks at w )110 nm then declines for larger grooves. Since the numberof photonic modes depends on the thickness of the absorbinglayer, and for very thin devices only the plasmonic modewill be supported, this confirms that subwavelength scatterersare important for the thinnest devices.Turning our attention from groovelike couplers to ridgelikecouplers of the same aspect ratio, the scattering responsechanges, shown in Figure 4. Although the qualitative spectralshape is similar, with a peak at 680 nm and at 1100 nm, theincoupling cross sections are much higher for the ridge, andenergy is predominantly coupled into photonic modes. Forexample, at λ ) 1100 nm, the total incoupling cross sectionis σgroove) 0.35 µm, with σphotonic) 0.15 µm, while for theridge the total incoupling cross section is σridge) 0.52 µm,with σphotonic) 0.5 µm. Intuitively this is reasonable becausethe ridge is a polarizable dipole-like scatterer located closerto the middle of the waveguide, whereas the groo ve is adipole-like scatter located in the plane of the metal-semi-conductor interface. Although the total incoupling crosssection is higher for the ridge, ultrathin devices that do notsupport photonic modes will be more efficiently served bysubwavelength grooves.Another type of resonance that could affect the incouplingefficiency of the groove is Fabry-Pe´rot cavity modes insidethe groove itself. We probe the effect of this type ofresonance by choosing a small width (w ) 50 nm) andvarying the depth incrementally from 50 to 220 nm at oneincident wavelength (λ ) 800 nm). In Figure 3b we showthe ratio of the incoupling cross sections with respect to areference groove that is 50 nm deep. The bottom panel showsthe ratio of the magnitude of the Ezfield at the groove mouthwith respect to this same reference groove. The firstinteresting detail is that the shallowest groove studied, d )50 nm, is the most efficient incoupler. As the groove getsdeeper, the amplitude of the oscillation decreases, as moreenergy is absorbed within the groove and the field magnitudeis weak at the interface. Most importantly, however, thevariation with depth is quite small, and we conclude thatwhile the Fabry-Pe´rot mode in the groove cavity is present,it is not a critical feature of design.At this point we have found determining factors for thespectral shape of the incoupling cross sections, but we havenot considered whether the energy in each mode is absorbedin the semiconductor or lost to the metal. Light coupled intoa propagating mode can be absorbed in either material, butcell efficiency will only be improved when the absorptionoccurs in the semiconductor, which will be responsible forenhanced carrier generation in the cell. To this end, we usea separate computational method to calculate the absorptionfractions in each material. Our spectral filtering method alsoallows us to determine power absorption in the semiconduc-tor due to standard thin film reflection/transmission processeswithout the influence of the scatteri ng. The averagetime-harmonic
View Full Document
Unlocking...