Plasmonics for improved photovoltaic devices

7/28/2019 Plasmonics for improved photovoltaic devices

1/10

nature materials | VOL 9 | MARCH 2010 | www.nature.com/naturematerials 205

review articlePublished online: 19 ebruary 2010 | doi: 10.1038/nmat2629

Photovoltaics, the conversion o sunlight to electricity, isa promising technology that may allow the generation oelectrical power on a very large scale. Worldwide photovoltaic

production was more than 5 GW in 2008, and is expected to riseabove 20 GW by 2015. Photovoltaics could thus make a consider-able contribution to solving the energy problem that our society

aces in the next generation. o make power rom photovoltaicscompetitive with ossil-uel technologies, the cost needs to bereduced by a actor o 25. At present most o the solar-cell marketis based on crystalline silicon waers with thicknesses between180300 m, and most o the price o solar cells is due to the costso silicon materials and processing. Because o this, there is greatinterest in thin-lm solar cells, with lm thicknesses in the range12 m, that can be deposited on cheap module-sized substratessuch as glass, plastic or stainless steel. Tin-lm solar cells are maderom a variety o semiconductors including amorphous and poly-crystalline Si, GaAs, Cde and CuInSe2, as well as organic semi-conductors. A limitation in all thin-lm solar-cell technologies isthat the absorbance o near-bandgap light is small, in particularor the indirect-bandgap semiconductor Si. Tereore, structuring

the thin-lm solar cell so that light is trapped inside to increase theabsorbance is very important. A signicant reduction in thin-lmsolar-cell thickness would also allow the large-scale use o scarcesemiconductor materials such as In and e that are available in theEarths crust in only small quantities.

In conventional thick Si solar cells, light trapping is typicallyachieved using a pyramidal surace texture that causes scattering olight into the solar cell over a large angular range, thereby increasingthe eective path length in the cell13. Such large-scale geometriesare not suitable or thin-lm cells, or geometrical reasons (as thesurace roughness would exceed the lm thickness) and because thegreater surace area increases minority carrier recombination in thesurace and junction regions.

A new method or achieving light trapping in thin-lm solar cells

is the use o metallic nanostructures that support surace plasmons:excitations o the conduction electrons at the interace between ametal and a dielectric. By proper engineering o these metallodi-electric structures, light can be concentrated and olded into a thinsemiconductor layer, thereby increasing the absorption. Both local-ized surace plasmons excited in metal nanoparticles and suraceplasmon polaritons (SPPs) propagating at the metal/semiconductorinterace are o interest.

In the past ew years, the eld o plasmonics has emerged as arapidly expanding new area or materials and device research4. Tis

Pc pv pvc vch a. aw1* a P2*

Th ging d o ponic h yidd thod o giding nd ocizing ight t th nnoc, w bow th c oth wvngth o ight in pc. Now ponic ch tning thi ttntion to photovotic, wh dignppoch bd on ponic cn b d to ipov boption in photovotic dvic, pitting conidb dc-tion in th phyic thickn o o photovotic bob y, nd yiding nw option o o-c dign. In thi viw,w vy cnt dvnc t th intction o ponic nd photovotic nd of n otook on th t o o cbd on th pincip.

is a result o the large array o tools that have become available ornanoscale abrication and nanophotonics characterization, andalso because o the availability o powerul electromagnetic simu-lation methods that are critical to understanding and harnessingplasmon excitations. Studies are just starting to appear illustratingthe coupling o plasmons to optical emitters57; plasmon ocusing8,9;

hybridized plasmonic modes in nanoscale metal shells10; nanoscalewaveguiding1114; nanoscale optical antennas15; plasmonic inte-grated circuits16,17; nanoscale switches18; plasmonic lasers1921; sur-ace-plasmon-enhanced light-emitting diodes22; imaging below thediraction limit23; and materials with negative reractive index 2426.Despite all these exciting opportunities, until recently little sys-tematic thought has been given to the question o how plasmonexcitation and light localization might be used advantageously inhigh-eciency photovoltaics.

Pc pvcConventionally, photovoltaic absorbers must be optically thickto allow near-complete light absorption and photocarrier currentcollection. Figure 1a shows the standard AM1.5 solar spectrum

together with a graph that illustrates what raction o the solar spec-trum is absorbed on a single pass through a 2-m-thick crystallineSi lm. Clearly, a large raction o the solar spectrum, in particularin the intense 6001,100 nm spectral range, is poorly absorbed. Tisis the reason that, or example, conventional waer-based crystallineSi solar cells have a much larger thickness o typically 180300 m.But high-eciency solar cells must have minority carrier diusionlengths several times the material thickness or all photocarriers tobe collected (see Fig. 1b), a requirement that is most easily met orthin cells. Solar-cell design and materials-synthesis considerationsare strongly dictated by these opposing requirements or opticalabsorption thickness and carrier collection length.

Plasmonic structures can oer at least three ways o reducing thephysical thickness o the photovoltaic absorber layers while keep-

ing their optical thickness constant. First, metallic nanoparticlescan be used as subwavelength scattering elements to couple andtrap reely propagating plane waves rom the Sun into an absorbingsemiconductor thin lm, by olding the light into a thin absorberlayer (Fig. 2a). Second, metallic nanoparticles can be used as sub-wavelength antennas in which the plasmonic near-eld is coupledto the semiconductor, increasing its eective absorption cross-sec-tion (Fig. 2b). Tird, a corrugated metallic lm on the back suraceo a thin photovoltaic absorber layer can couple sunlight into SPPmodes supported at the metal/semiconductor interace as well as

1Caltech Center or Sustainable Energy Research and Thomas J. Watson Laboratories o Applied Physics, Caliornia Institute o Technology, Pasadena,

Caliornia 91125, USA. 2Center or Nanophotonics, FOM Institute AMOLF, Science Park 104, 1098 XG Amsterdam, The Netherlands.

*e-mail: haa@caltech.edu; polman@amol.nl

7/28/2019 Plasmonics for improved photovoltaic devices

2/10

206 nature materials | VOL 9 | MARCH 2010 | www.nature.com/naturematerials

review article NaTure maTerIalsdoi: 10.1038/nmat2629

guided modes in the semiconductor slab, whereupon the light isconverted to photocarriers in the semiconductor (Fig. 2c).

As will be discussed in detail in the next section, these threelight-trapping techniques may allow considerable shrinkage(possibly 10- to 100-old) o the photovoltaic layer thickness, whilekeeping the optical absorption (and thus eciency) constant.

Various additional ways o using plasmonic nanostructures toincrease photovoltaic energy conversion are described in the sectionon other plasmonic solar-cell designs.

Pc g ppg -f cLight scattering using particle plasmons. Light scattering roma small metal nanoparticle embedded in a homogeneous mediumis nearly symmetric in the orward and reverse directions27,28. Tissituation changes when the particle is placed close to the interace

between two dielectrics, in which case light will scatter preeren-tially into the dielectric with the larger permittivity29. Te scatteredlight will then acquire an angular spread in the dielectric that eec-tively increases the optical path length (see Fig. 2a). Moreover, lightscattered at an angle beyond the critical angle or reection (16 orthe Si/air interace) will remain trapped in the cell. In addition, ithe cell has a reecting metal back contact, light reected towardsthe surace will couple to the nanoparticles and be partly reradi-ated into the waer by the same scattering mechanism. As a result,the incident light will pass several times through the semiconductorlm, increasing the eective path length.

Te enhanced incoupling o light into semiconductor thin lmsby scattering rom plasmonic nanoparticles was rst recognized byStuart and Hall, who used dense nanoparticle arrays as resonant

scatterers to couple light into Si-on-insulator photodetector struc-tures30,31. Tey observed a roughly 20-old increase in the inraredphotocurrent in such a structure. Tis research eld then remainedrelatively dormant or many years, until applications in thin-lmsolar cells emerged, with papers published on enhanced light cou-pling into single-crystalline Si (re. 32), amorphous Si (res 33,34),Si-on-insulator35, quantum well36 and GaAs (re. 37) solar cells cov-ered with metal nanoparticles.

Although there is now considerable experimental evidence thatlight scattering rom metal nanoparticle arrays increases the photo-current spectral response o thin-lm solar cells, many o the under-lying physical mechanisms and their interplay have not been studiedsystematically. Te ull potential o the particle scattering concept,taking into account integration with optimized anti-reection coat-

ings, is being studied by several research groups. In recent papers38,39

,we reported that both shape and size o metal nanoparticles are keyactors determining the incoupling eciency. Tis is illustratedin Fig. 3a, which shows that smaller particles, with their eectivedipole moment located closer to the semiconductor layer, couple alarger raction o the incident light into the underlying semiconduc-tor because o enhanced near-eld coupling. Indeed, in the limit oa point dipole very near to a silicon substrate, 96% o the incidentlight is scattered into the substrate, demonstrating the power o

a b c

400 600 800 1,400 1,6001,000 1,200

Freespace wavelength (nm)

1.6

1.4

1.2

0.8

0.6

0.4

0.2

0

1.0

Spectralintensity(Wm

2n

m1)

2mthick

Si wafer

AM1.5 solar spectrum

Blue Green Red

Si

p

n

*

*

b

a

Ld

Fig 1 | Optic boption nd ci difion qint in o

c. , AM1.5 solar spectrum, together with a graph that indicates the solar

energy absorbed in a 2-m-thick crystalline Si lm (assuming single-pass

absorption and no reection). Clearly, a large raction o the incident light

in the spectral range 6001,100 nm is not absorbed in a thin crystalline Si

solar cell. b, Schematic indicating carrier difusion rom the region where

photocarriers are generated to the pn junction. Charge carriers generated

ar away (more than the difusion length Ld) rom the pn junction are notefectively collected, owing to bulk recombination (indicated by the asterisk).

Fig 2 | Ponic ight-tpping goti o thin- o c. , Light trapping by scattering rom metal nanoparticles at the surace o the solar

cell. Light is preerentially scattered and trapped into the semiconductor thin lm by multiple and high-angle scattering, causing an increase in the efective

optical path length in the cell. b, Light trapping by the excitation o localized surace plasmons in metal nanoparticles embedded in the semiconductor.

The excited particles near-eld causes the creation o electronhole pairs in the semiconductor. c, Light trapping by the excitation o surace plasmon

polaritons at the metal/semiconductor interace. A corrugated metal back surace couples light to surace plasmon polariton or photonic modes that

propagate in the plane o the semiconductor layer.

7/28/2019 Plasmonics for improved photovoltaic devices

3/10

nature materials | VOL 9 | MARCH 2010 | www.nature.com/naturematerials 207

review articleNaTure maTerIalsdoi: 10.1038/nmat2629

the particle scattering technique. Figure 3b shows the path-lengthenhancement in the solar cell derived rom Fig. 3a using a simplerst-order scattering model. For 100-nm-diameter Ag hemisphereson Si, a 30-old enhancement is ound. Tese light-trapping eectsare most pronounced at the peak o the plasmon resonance spec-trum, which can be tuned by engineering the dielectric constant othe surrounding matrix. For example, small Ag or Au particles in airhave plasmon resonances at 350 nm and 480 nm respectively; theycan be redshifed in a controlled way over the entire 5001,500 nmspectral range by (partially) embedding them in SiO2, Si3N4 or Si(res 4042; Fig. 4a), which are all standard materials in solar-cellmanuacturing. A dielectric spacer layer between the nanoparticles

and the semiconductor also causes electrical isolation, therebyavoiding additional surace recombination owing to the presenceo the metal. Te scattering cross-sections or metal nanoparticlescattering can be as high as ten times the geometrical area28,38, and a~10% coverage o the solar cell would thus suce to capture most othe incident sunlight into plasmon excitations.

Optimization o plasmonic light trapping in a solar cell is abalancing act in which several physical parameters must be takeninto account. First o all, Fig. 3 demonstrates the advantage o usingsmall particles to create the orward scattering anisotropy. But verysmall particles suer rom signicant ohmic losses, which scalewith volume v, whereas scattering scales with v2, so that using largerparticles is advantageous to increase the scattering rate. For example,a 150-nm-diameter Ag particle in air has an albedo (raction o light

emitted as radiation) as large as 95%. Interestingly, the eectivescattering cross-section can be increased by spacing the particlesurther away rom the substrate, as this avoids destructive intererenceeects between the incident and reected elds, albeit at the price oreduced near-eld coupling38,39. For requencies above the plasmonresonance, Fano resonance eects can cause destructive intererencebetween scattered and unscattered light beams, and thus can causereection rather than enhanced incoupling42,43. One way to over-come the latter problem is by using a geometry in which particles areplaced on the rear o the solar cell42,44. In this case, blue and green lightis directly absorbed in the cell, whereas poorly absorbed (inra-)redlight is scattered and trapped using metal nanoparticles. Similarly,light may be coupled in by using arrays o metal strips placed on thesolar-cell surace. Calculations show that using this geometry canenhance the short-circuit current by 45% compared with a solar cell

with a at back contact45. Finally, in designing optimized plasmoniclight-trapping arrays, we must take into account coupling betweenthe nanoparticles, ohmic damping, grating diraction eects46 andthe coupling to waveguide modes4749.

Light concentration using particle plasmons. An alternativeuse o resonant plasmon excitation in thin-lm solar cells is totake advantage o the strong local eld enhancement around themetal nanoparticles to increase absorption in a surrounding semi-conductor material. Te nanoparticles then act as an eectiveantenna or the incident sunlight that stores the incident energy ina localized surace plasmon mode (Fig. 4b). Tis works particularly

well or small (520 nm diameter) particles or which the albedo islow. Tese antennas are particularly useul in materials where thecarrier diusion lengths are small, and photocarriers must thus begenerated close to the collection junction area. For these antennaenergy conversion eects to be ecient, the absorption rate in thesemiconductor must be larger than the reciprocal o the typicalplasmon decay time (lietime ~1050 s), as otherwise the absorbedenergy is dissipated into ohmic damping in the metal. Such highabsorption rates are achievable in many organic and direct-bandgapinorganic semiconductors.

Several examples o this concept have recently appeared thatdemonstrate enhanced photocurrents owing to the plasmonicnear-eld coupling. Enhanced eciencies have been demonstratedor ultrathin-lm organic solar cells doped with very small (5 nm

diameter) Ag nanoparticles50

. Plasmon-enhanced organic solar cellshave also been demonstrated using electrodeposited Ag (re. 51). Anincrease in eciency by a actor o 1.7 has been shown or organicbulk heterojunction solar cells52,53. Dye-sensitized solar cells can alsobe enhanced by embedding small metal nanoparticles, as reportedelsewhere5456. Also, inorganic solar cells have shown increasedphotocurrents owing to near-eld eects, such as CdSe/Si hetero-structures57. Work on increased light absorption by metal nano-particles embedded in Si has also been reported58,59. More recently,the coupling between plasmons in arrays o metal nanoparticleshas been used to engineer the eld enhancement to overlap witha selected area o the junction53. Te dynamics (and optimization)o the coupling between plasmons, excitons and phonons in metalsemiconductor nanostructures is a rich eld o research that so arhas not received much attention with photovoltaics in mind.

a b

500 0.6

Sphere 150 nm

Sphere 100 nm

Cylinder

Hemisphere

Lambertian

Parallel dipole

1

10

100

10.7 0.8 0.90

1

0.2

0.4

0.6

0.8

600

Wavelength (nm) Fraction scattered into substrate

Maximumpath-lengthenhancement

Fractionscatteredintosubstrate

700

DipoleCylinder

Hemisphere

Sphere 100 nm

Sphere 150 nm

800650 750550

Fig 3 | light ctting nd tpping i vy nitiv to ptic hp. , Fraction o light scattered into the substrate, divided by total scattered

power, or diferent sizes and shapes o Ag particles on Si. Also plotted is the scattered raction or a parallel electric dipole that is 10 nm rom a Si

substrate. b, Maximum path-length enhancement, according to a rst-order geometrical model, or the same geometries as in at a wavelength o

800 nm. Absorption within the particles is neglected or these calculations, and an ideal rear reector is assumed. The line is a guide or the eye. Insets:

(top let) angular distribution o scattered power or a parallel electric dipole that is 10 nm above a Si surace (red) and a Lambertian scatterer (blue);

(lower right) geometry considered or calculating the path-length enhancement. Figure reproduced with permission: 2008 AIP.

7/28/2019 Plasmonics for improved photovoltaic devices

4/10

208 nature materials | VOL 9 | MARCH 2010 | www.nature.com/naturematerials

review article NaTure maTerIalsdoi: 10.1038/nmat2629

Light trapping using SPPs. In a third plasmonic light-trappinggeometry, light is converted into SPPs, which are electromagneticwaves that travel along the interace between a metal back contactand the semiconductor absorber layer (see Fig. 4c)60. Near the plas-

mon resonance requency, the evanescent electromagnetic SPP eldsare conned near the interace at dimensions much smaller than thewavelength. SPPs excited at the metal/semiconductor interace caneciently trap and guide light in the semiconductor layer. In thisgeometry the incident solar ux is eectively turned by 90, andlight is absorbed along the lateral direction o the solar cell, whichhas dimensions that are orders o magnitude larger than the opticalabsorption length. As metal contacts are a standard element in thesolar-cell design, this plasmonic coupling concept can be integratedin a natural way.

At requencies near the plasmon resonance requency (typi-cally in the 350700 nm spectral range, depending on metal anddielectric) SPPs suer rom relatively high losses. Further into theinrared, however, propagation lengths are substantial. For example,or a semi-innite Ag/SiO

2

geometry, SPP propagation lengths range

rom 10 to 100 m in the 8001,500 nm spectral range. By using athin-lm metal geometry the plasmon dispersion can be urtherengineered6164. Increased propagation length comes at the expenseo reduced optical connement and optimum metal-lm design

thus depends on the desired solar-cell geometry. Detailed accountso plasmon dispersion and loss in metaldielectric geometries areound in res 6164.

Te SPP coupling mechanism is benecial or ecient lightabsorption i absorption o the SPP in the semiconductor is strongerthan in the metal. Tis is illustrated in Fig. 5a, which shows a cal-culation o the raction o light that is absorbed in a silicon or GaAslm in contact with either Ag or Al. Te inset shows the mode-elddistribution at the Si/Ag interace at a wavelength o 850 nm. Ascan be seen, or SPPs at a GaAs/Ag interace the semiconductorabsorption raction is high in the spectral range rom the GaAs/AgSPP resonance (600 nm) to the bandgap o GaAs (870 nm). For aSi/Ag interace, with smaller optical absorption in Si owing to theindirect bandgap, plasmon losses dominate over the entire spectralrange, although absorption in the 7001,150 nm spectral range is

a b

Si/Ag

Si/Al

GaAs/Ag

GaAs/Al

pol./Ag

pol./Al

400 600 800800600400

1,000 1,200

Wavelength (nm) Wavelength (nm)

Photonic

SPP

0.0

Fractionabsorbedinsemiconductor

Incouplingcross-section(m)

0.2

0.4

0.6

0.8

1.0

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

1,4001,2001,000

100 nm50 nm

Si

Ag

20

0

20

40

z(

nm)

Si

|E|2

Ag

0400 600 800 1,000 1,200

2

4

6

8

10

Wavelength (nm)

Wavelength(nm)

Normalizedcross-section

SiAir

DQD

DQ

a b c

50

50

3

2

1

0 0.00 0.05 0.10 0.1550

50

0

0

2

4

6

8

10

x position (nm)

y

position(nm)

Energy(eV)

= 850 nm

Bound SPP

400

600

800

1,000

2,000

Si3N4 |E|2

20

0

20

40

z

(nm)

kSPP (nm1)

Si

Ag

|E|2

Fig 4 | Chctitic o c pon. , Metal nanoparticles scatter light over a broad spectral range that can be tuned by the surrounding

dielectric. The plots show the scattering cross-section spectrum or a 100-nm-diameter Ag particle embedded in three diferent dielectrics (air, Si3N4

and Si). Dipole (D) and quadrupole (Q) modes are indicated. The cross-section is normalized to the geometrical cross-section o the particle. b, Metal

nanoparticles show an intense near-eld close to the surace. Intensity enhancement around a 25-nm-diameter Au particle embedded in a medium with

index n = 1.5 (plasmon resonance peak at 500 nm). Light with a wavelength= 850 nm is incident with a polarization indicated by the vertical arrow. The

magnitude o the enhanced electric-eld intensity E is indicated by the colour scale. c, SPPs are bound waves at the interace between a semiconductor

and a dielectric. This dispersion diagram, plotting the relationship between requency and wavevector (2/) or SPPs on a Ag /Si interace. The bound

SPP mode occurs at energies below the surace plasmon resonance energy o 2.07 eV (600 nm). The dispersion o light in Si is indicated by the dashed

line. The inset shows a schematic o the SPP mode prole along the Si/Ag interace, at a ree-space wavelength o 785 nm.

Fig 5 | light ctting into sPP nd photonic od in thin icondcto . , Fraction o light absorbed into the semiconductor or SPPs

propagating along interaces between semi-innite layers o GaAs, Si and an organic semiconductor lm made o a PF10TBT:[C60]PCBM polymer

blend (termed pol. in the legend), in contact with either Ag or Al. Graphs are plotted or wavelength down to the surace plasmon resonance. The inset

shows the SPP eld intensity near the interace. b, Two-dimensional calculation o the incoupling cross-section or SPP and photonic modes as a unction

o wavelength or a 200-nm-thick Si slab on an optically thick Ag substrate with a 100-nm-wide, 50-nm-tall Ag ridge, as shown in the inset. Figure

reproduced with permission: b, 2008 ACS.

7/28/2019 Plasmonics for improved photovoltaic devices

5/10

nature materials | VOL 9 | MARCH 2010 | www.nature.com/naturematerials 209

review articleNaTure maTerIalsdoi: 10.1038/nmat2629

still higher than single-pass absorption through a 1-m-thick Silm. In this case, it is more benecial to use a geometry in whicha thin metal lm is embedded in the semiconductor, as in this caseplasmon absorption is much smaller owing to the smaller modeoverlap with the metal63,64. Figure 5a also shows a calculation o theabsorption o an organic semiconductor (PF10B:[C60]PCBM)(re. 65) in contact with Ag or Al. Optical constants or the polymerare taken rom re. 65. Here the absorption eciency is high overthe entire spectral range below 650 nm because the organic semi-

conductor has high absorption, and the low dielectric constant othe organic material causes a small overlap o the modal eld withthe metal and thus lower ohmic losses.

Owing to the momentum mismatch between the incident lightand the in-plane SPPs (see Fig. 4c) a light-incoupling structure mustbe integrated in the metal/dielectric interace. Figure 5b plots theresults o ull-eld electrodynamics simulations (taking into accountin- and outcoupling) o the scattering rom an incident plane waveinto SPPs by a ridge abricated in the Ag/Si interace66. Te Si layeris 200 nm thick and the Ag ridge 50 nm tall. Te simulations showthat light is coupled into an SPP mode as well as a photonic modethat propagates inside the Si waveguide, and that the strength ocoupling to each mode can be controlled by the height o the scatter-ing object66. Te photonic modes are particularly interesting as they

suer rom only very small losses in the metal. Te raction o lightcoupled into both modes increases with increasing wavelength. Tisis mainly because at shorter wavelengths the incoming light beam isdirectly absorbed in the Si layer. Te data demonstrate that light with > 800 nm, which would not be well absorbed by normal incidenceon the Si layer (see Fig. 1a), can now be eciently absorbed byconversion into the in-plane SPP and photonic modes. Althoughthis example shows coupling rom a single, isolated ridge, the shape,height and interparticle arrangements o incoupling structures canall be optimized or preerential coupling to particular modes. In theultimate, ultrathin Si solar cell (thickness 50 GW by 2020, and eventually to the terawatt scale. For thisto happen, the materials used in solar cells must be sucientlyabundant in the Earths crust and amenable to the ormation oecient photovoltaic devices. For single-junction devices, Si hasproved to be a near-ideal photovoltaic material. It is abundant, with

a nearly optimal bandgap, excellent junction ormation character-istics, high minority carrier diusion length and eective methodsor surace passivation to reduce unwanted carrier recombination.Its only shortcoming low spectral absorption o the terrestrialsolar spectrum means that it requires cell thicknesses o over100 m, which gives rise to a relatively high material cost per outputpower. Plasmonic light trapping makes it possible to design crystal-line Si thin-lm cells with spectral quantum eciency approachingthat seen in thick cells, even or absorber layers a ew micrometresin thickness.

Materials resources are a signicant limitation or large-scaleproduction o two o the most common thin-lm solar-cell mate-rials: Cde and CuInSe2. Manuacturing costs or these cells haveallen, and solar-cell production using these semiconductors is

expanding rapidly. able 1 lists the (projected) annual solar-cellproduction per year, as well as the materials eedstock required orthe production o the corresponding solar-cell volume using Si,Cde or CuInSe2. As can be seen, the materials eedstock requiredin 2020 exceeds the present annual world production o e and In,and in the case o In is even close to the total reserve base. I itwere possible to reduce the cell thickness or such compound semi-conductor cells by 10100 times as a result o plasmon-enhancedlight absorption, this could considerably extend the reach o thesecompound semiconductor thin-lm solar cells towards the tera-watt scale. Earth-abundance considerations will also inuenceplasmonic cell designs at large-scale production: although Ag andAu have been the metals o choice in most plasmonic designs andexperiments, they are relatively scarce materials, so scalable designs

will need to ocus on abundant metals such as Al and Cu.Reducing the active-layer thickness by plasmonic light trap-ping not only reduces costs but also improves the electrical char-acteristics o the solar cell78. First o all, reducing the cell thicknessreduces the dark current (Idark), causing the open-circuit voltageVoc to increase, as Voc= (kBT/q) ln(Iphoto/Idark + 1), where kB is theBoltzmann constant, T is temperature, q is the charge and Iphoto isthe photocurrent. Consequently, the cell eciency rises in loga-rithmic proportion to the decrease in thickness, and is ultimately

Tb 1 | Photovotic oc qint: ti by

podction nd v.

y a -c

pc*

m

S (-S) t (cdt) in (cuinS2)2000 0.3 GWp 4 0.03 0.03

2005 1.5 GWp 15 0.15 0.15

2020 50 GWp 150 5 5

W pc

(1,000 p )

1,000 0.3 0.5

rv (1,000 ) Abundant 47 6

*Wp = peak output power under ull solar illumination.

The annual solar-cell production and the required materials eedstock are indicated, assuming cells

are made o Si, CdTe or CuInSe2. The assumed material use or Si is 13 g Wp1 in 2000, 10 g Wp

1 in

2005, 3 g Wp1 in 2020 (projected); it is 0.1 g Wp

1 or Te and In. The two bottom rows indicate the

world reserve base data or Si, Te and In (that is, resources that are economic at present or marginally

economic and some that are at present subeconomic).

Sources: res 93, 94 and G. Willeke (Fraunhoer Institute or Solar Energy Systems),

personal communication.

7/28/2019 Plasmonics for improved photovoltaic devices

6/10

210 nature materials | VOL 9 | MARCH 2010 | www.nature.com/naturematerials

review article NaTure maTerIalsdoi: 10.1038/nmat2629

limited by surace recombination. Second, in a thin-lm geometry,carrier recombination is reduced as carriers need to travel only asmall distance beore being collected at the junction. Tis leads toa higher photocurrent. Greatly reducing the semiconductor layerthickness allows the use o semiconductor materials with lowminority carrier diusion lengths, such as polycrystalline semi-conductors, quantum-dot layers or organic semiconductors. Also,this could render useul abundant and potentially inexpensivesemiconductors with signicant impurity and deect densities,

such as Cu2O, Zn3P2 or SiC, or which the state o electronic materi-als development is not as advanced as it is or Si.

o w pc -c gTe previous section has ocused on the use o plasmonic scatteringand coupling concepts to improve the eciency o single-junctionplanar thin-lm solar cells, but many other cell designs can benetrom the increased light connement and scattering rom metalnanostructures. First o all, plasmonic tandem geometries maybe made, in which semiconductors with dierent bandgaps arestacked on top o each other, separated by a metal contact layer witha plasmonic nanostructure that couples dierent spectral bands inthe solar spectrum into the corresponding semiconductor layer(see Fig. 6a)79. Coupling sunlight into SPPs could also solve the

problem o light absorption in quantum-dot solar cells (see Fig. 6b).Although such cells oer potentially large benets because o theexibility in engineering the semiconductor bandgap by particlesize, eective light absorption requires thick quantum-dot layers,

through which carrier transport is problematic. As we have recentlydemonstrated80, a 20-nm-thick layer o CdSe semiconductorquantum dots deposited on a Ag lm can absorb light connedinto SPPs within a decay length o 1.2 m at an incident photonenergy above the CdSe quantum-dot bandgap at 2.3 eV. Te reversegeometry, in which quantum dots are electrically excited to generateplasmons, has also recently been demonstrated81. We note that theplasmon light-trapping concepts described in the previous sectionrely on scattering using localized modes, and are thus relatively

insensitive to angle o incidence66,82. Tis is an advantage or solar-cell designs made or areas where incident sunlight is mostly diuserather than direct.

In a recent example o nanoscale plasmonic solar-cell engineering,an organic photovoltaic light absorber was integrated in the gapbetween the arms o plasmonic antennas arranged in arrays (seeFig. 6c)83. Other examples o nanoscale antennas are coaxialholes abricated in a metal lm, which show localized plasmonicmodes owing to FabryPerot resonances (see Fig. 6d)8486. Suchnanostructures, with eld enhancements up to a actor o about50, could be used in entirely new solar-cell designs, in which aninexpensive semiconductor with low minority carrier lietime isembedded inside the plasmonic cavity. Similarly, quantum-dotsolar cells based on multiple-exciton generation87, or cells with solar

upconverters or downconverters based on multiphoton absorptioneects, could benet rom such plasmonic eld concentration. Ingeneral, eld concentration in plasmonic nanostructures is likelyto be useul in any type o solar cell where light concentration is

a

c d

b

SPPIncident

light

SPP guiding layer

Quantum-dot

active layer

Top contact3.0 eV

2.0 eV

1.0 eV

SemiconductorMetal

P3HT

Metal

p

n

p

n

p

n

Fig 6 | Nw ponic o-c dign. , Plasmonic tandem solar-cell geometry. Semiconductors with diferent bandgaps are stacked on top

o each other, separated by a metal contact layer with a plasmonic nanostructure that couples diferent spectral bands o the solar spectrum into the

corresponding semiconductor layer. b, Plasmonic quantum-dot solar cell designed or enhanced photoabsorption in ultrathin quantum-dot layers mediated

by coupling to SPP modes propagating in the plane o the interace between Ag and the quantum-dot layer. Semiconductor quantum dots are embedded in

a metal/insulator/metal SPP waveguide. c, Optical antenna array made rom an axial heterostructure o metal and poly(3-hexylthiophene) (P3HT). Light is

concentrated in the nanoscale gap between the two antenna arms, and photocurrent is generated in the P3HT semiconductor83. d, Array o coaxial holes

in a metal lm that support localized FabryPerot plasmon modes. The coaxial holes are lled with an inexpensive semiconductor with low minority carrier

lietime, and carriers are collected by the metal on the inner and outer sides o the coaxial structure. Field enhancements up to a actor o about 50 are

possible and may serve to enhance nonlinear photovoltaic conversion efects85.

7/28/2019 Plasmonics for improved photovoltaic devices

7/10

nature materials | VOL 9 | MARCH 2010 | www.nature.com/naturematerials 211

review articleNaTure maTerIalsdoi: 10.1038/nmat2629

benecial, such as a high-eciency multijunction solar cells88,89.More details on plasmon light concentration are ound in anotherreview in this issue90.

lg- c pc -c cTe plasmonic coupling eects described in this review all requireintegration o dense arrays o metal nanostructures with controlover dimension tolerances at the nanometre scale. Although sys-tematic laboratory experiments can be perormed using cleanroom

techniques such as electron-beam lithography or ocused ion-beammilling, true application in large-area photovoltaic module produc-tion requires inexpensive and scalable techniques or abricatingmetal nanopatterns in a controlled way. Several research groupshave studied these techniques and have shown their easibility.

Te simplest way to orm metal nanoparticles on a substrate is bythermal evaporation o a thin (1020 nm) metal lm, which is thenheated at a moderate temperature (200300 C), to cause agglom-eration by surace tension o the metal lm into a random array onanoparticles. As has been shown35,42, this leads to the ormation orandom arrays o Ag nanoparticles with a diameter (100150 nm)and hemispherical shape that is well suited or light trapping38. Morecontrol over the Ag nanoparticle size, aspect ratio and density canbe achieved using deposition through a porous alumina template,

as was demonstrated or example in re. 37. Here, the nanoparticleshape was urther tuned by thermal annealing at 200 C, leading tohemispherical particles as well. More recently, substrate conormalimprint lithography91 has been developed, in which a solgel maskis dened by sof lithography using a rubber stamp, ollowed by Agevaporation and lif-o. Tis process leads to

7/28/2019 Plasmonics for improved photovoltaic devices

8/10

212 nature materials | VOL 9 | MARCH 2010 | www.nature.com/naturematerials

review article NaTure maTerIalsdoi: 10.1038/nmat2629

12. Maier, S. A. et al. Local detection o electromagnetic energy transport belowthe diraction limit in metal nanoparticle plasmon waveguides. Nature Mater.2, 229232 (2003).

13. Koenderink, A. F. & Polman, A. Complex response and polariton-likedispersion splitting in periodic metal nanoparticle chains. Phys. Rev. B74, 033402 (2006).

14. Zia, R., Selker, M. D., Catrysse, P. B. & Brongersma, M. L. Geometries andmaterials or subwavelength surace plasmon modes.J. Opt. Soc. Am. A21, 24422446 (2004).

15. Mhlschlegel, P., Eisler, H. J., Martin, O. J. F., Hecht, B. & Pohl, D. W. Resonant

optical antennas. Science308, 16071609 (2005).16. Ditlbacher, H., Krenn, J. R., Schider, G., Leitner, A. & Aussenegg, F. R.wo-dimensional optics with surace plasmonpolaritons.Appl. Phys. Lett.81, 17621764 (2002).

17. Bozhevolnyi, S. I., Volkov, V. S., Devaux, E., Laluet, J. Y. & Ebbesen, . W.Channel plasmon subwavelength waveguide components includingintererometers and ring resonators. Nature 440, 508511 (2006).

18. Krasavin, A. V. & Zheludev, N. I. Active plasmonics: Controlling signals inAu/Ga waveguide using nanoscale structural transormations.Appl. Phys. Lett.84, 14161418 (2004).

19. Colombelli, R. et al. Quantum cascade surace-emitting photonic crystal laser.Science302, 13741377 (2003).

20. Hill, M. . et al. Lasing in metallic-coated nanocavities. Nature Photon.1, 589594 (2007).

21. Oulton, R. F. et al. Plasmon lasers at deep subwavelength scale. Nature461, 629632(2009).

22. Okamoto, K. et al. Surace-plasmon-enhanced light emitters based on InGaN

quantum wells. Nature Mater.3, 601605 (2004).23. Pendry, J. B. Negative reraction makes a perect lens. Phys. Rev. Lett.

85, 39663968 (2000).24. Shelby, R. A., Smith, D. R. & Schultz, S. Experimentalverication o a negative

index o reraction. Science292, 7779 (2001).25. Linden, S. et al. Magnetic response o metamaterials at 100 terahertz. Science

306, 13511353 (2004).26. Fang, N., Lee, H., Sun, C. & Zhang, X. Sub-diraction-limited optical imaging

with a silver superlens. Science308, 534537 (2005).27. Kreibig, U. & Vollmer, M. Optical Properties of Metal Clusters (Springer, 1995).28. Bohren, C. F. & Human, D. R. Absorption and Scattering of Light by Small

Particles (Wiley, 2008).29. Mertz, J. Radiative absorption, uorescence, and scattering o a classical

dipole near a lossless interace: a unied description.J. Opt. Soc. Am. B17, 19061913 (2000).

30. Stuart, H. R. & Hall, D. G. Absorption enhancement in silicon-on-insulatorwaveguides using metal island lms.Appl. Phys. Lett.69, 23272329 (1996).

31. Stuart, H. R. & Hall, D. G. Island size eects in nanoparticle-enhancedphotodetectors.Appl. Phys. Lett.73, 38153817 (1998).

32. Schaadt, D. M., Feng, B. & Yu, E. . Enhanced semiconductor opticalabsorption via surace plasmon excitation in metal nanoparticles.Appl. Phys. Lett.86, 063106 (2005).

33. Derkacs, D., Lim, S. H., Matheu, P., Mar, W. & Yu, E. . Improved perormanceo amorphous silicon solar cells via scattering rom surace plasmon polaritonsin nearby metallic nanoparticles.Appl. Phys. Lett.89, 093103 (2006).

34. Matheu, P., Lim, S. H., Derkacs, D., McPheeters, C. & Yu, E. . Metaland dielectric nanoparticle scattering or improved optical absorption inphotovoltaic devices.Appl. Phys. Lett.93, 113108 (2008).

35. Pillai, S., Catchpole, K. R., rupke, . & Green, M. A. Surace plasmonenhanced silicon solar cells.J. Appl. Phys.101, 093105 (2007).

36. Derkacs, D. et al. Nanoparticle-induced light scattering or improvedperormance o quantum-well solar cells.Appl. Phys. Lett.93, 091107 (2008).

37. Nakayama, K., anabe, K. & Atwater, H. A. Plasmonic nanoparticle enhanced

light absorption in GaAs solar cells.Appl. Phys. Lett.93, 121904 (2008).38. Catchpole, K. R. & Polman, A. Design principles or particle plasmon enhanced

solar cells.Appl. Phys. Lett.93, 191113 (2008).39. Catchpole, K. R. & Polman, A. Plasmonic solar cells. Opt. Express

16, 2179321800(2008).40. Xu, G., azawa, M., Jin, P., Nakao, S. & Yoshimura, K. Wavelength tuning

o surace plasmon resonance using dielectric layers on silver island lms.Appl. Phys. Lett.82, 38113813 (2003).

41. Mertens, H., Verhoeven, J., Polman, A. & ichelaar, F. D. Inrared suraceplasmons in two-dimensional silver nanopartice arrays in silicon.Appl. Phys. Lett.85, 13171319 (2004).

42. Beck, F. J., Polman, A. & Catchpole, K. R. unable light trapping or solar cellsusing localized surace plasmons. J. Appl. Phys.105, 114310 (2009).

43. Lim, S. H., Mar, W., Matheu, P., Derkacs, D. & Yu, E. . Photocurrentspectroscopy o optical absorption enhancement in silicon photodiodes viascattering rom surace plasmon polaritons in gold nanoparticles.J. Appl. Phys.101, 104309 (2007).

44. Beck, F. J., Mokkapati, S., Polman, A. & Catchpole, K. R. Asymmetry inlight-trapping by plasmonic nanoparticle arrays located on the ront or on therear o solar cells. Appl. Phys. Lett. (in the press).

45. Pala, R. A., White, J., Barnard, E., Liu, J. & Brongersma, M. L. Design oplasmonic thin-lm solar cells with broadband absorption enhancements.Adv. Mater.21, 35043509 (2009).

46. Mokkapati, S., Beck, F. J., Polman, A. & Catchpole, K. R. Designing periodicarrays o metal nanoparticles or light trapping applications in solar cells.Appl. Phys. Lett.95, 53115 (2009).

47. Stuart, H. R. & Hall, D. G. Termodynamic limit to light trapping in thin

planar structures.J. Opt. Soc. Am. A 14, 30013008 (1997).48. Stuart, H. R. & Hall, D. G. Enhanced dipoledipole interaction betweenelementary radiators near a surace. Phys. Rev. Lett.80, 56635668 (1998).

49. Catchpole, K. R. & Pillai, S. Absorption enhancement due to scattering bydipoles into silicon waveguides.J. Appl. Phys.100, 044504 (2006).

50. Rand, B. P., Peumans, P. & Forrest, S. R. Long-range absorption enhancementin organic tandem thin-lm solar cells containing silver nanoclusters.J. Appl. Phys.96, 75197526 (2004).

51. Kim, S. S., Na, S.-I., Jo, J., Kim, D. Y. & Nah, Y.-C. Plasmon enhancedperormance o organic solar cells using electrodeposited Ag nanoparticles.Appl. Phys. Lett.93, 073307 (2008).

52. Mora, A. J., Rowlen, K. L., Reilly, . H., Romero, M. J. & Van de Lagemaat, J.Plasmon-enhanced solar energy conversion in organic bulk heterojunctionphotovoltaics. Appl. Phys. Lett. 92, 013504 (2008).

53. Lindquist, N. C., Luhman, W. A., Oh, S. H. & Holmes, R. J. Plasmonicnanocavity arrays or enhanced eciency in organic photovoltaic cells.Appl. Phys. Lett.93, 123308 (2008).

54. Kume, ., Hayashi, S., Ohkuma, H., Yamamoto, K. Enhancement o photoelectricconversion eciency in copper phthalocyanine solar cell: white light excitationo surace plasmon polaritons.Jpn. J. Appl. Phys. 34, 64486451 (1995).

55. Westphalen, M., Kreibig, U., Rostalski, J., Lth, H. & Meissner, D. Metal clusterenhanced organic solar cells. Sol. Energy Mater. Sol. C.61, 97105 (2000).

56. Hgglund, C., Zch, M. & Kasemo, B. Enhanced charge carrier generationin dye sensitized solar cells by nanoparticle plasmons.Appl. Phys. Lett.92, 013113 (2008).

57. Konda, R. B. et al. Surace plasmon excitation via Au nanoparticles in n-CdSe/p-Si heterojunction diodes.Appl. Phys. Lett.91, 191111 (2007).

58. Hgglund, C., Zch, M., Petersson, G. & Kasemo, B. Electromagnetic couplingo light into a silicon solar cell by nanodisk plasmons.Appl. Phys. Lett.92, 053110 (2008).

59. Kirkengena, M., Bergli, J. & Galperin, Y. M. Direct generation o chargecarriers in c-Si solar cells due to embedded nanoparticles. J. Appl. Phys.102, 093713 (2007).

60. Raether, H. Surface Plasmons on Smooth and Rough Surfaces and on Gratings .(Springer racts in Modern Physics III, Springer, 1988).

61. Berini, P. Plasmonpolariton waves guided by thin lossy metal lmso nite width: bound modes o symmetric structures. Phys. Rev. B61, 1048410503 (2000).

62. Berini, P. Plasmonpolariton waves guided by thin lossy metal lms o nitewidth: bound modes o asymmetric structures. Phys. Rev. B 63, 125417 (2001).

63. Dionne, J. A., Sweatlock, L., Atwater, H. A. & Polman, A. Planar plasmon metalwaveguides: requency-dependent dispersion, propagation, localization, andloss beyond the ree electron model. Phys. Rev. B72, 075405 (2005).

64. Dionne, J. A., Sweatlock, L., Atwater, H. A. & Polman, A. Plasmon slotwaveguides: towards chip-scale propagation with subwavelength-scalelocalization. Phys. Rev. B73, 035407 (2006).

65. Sloo, L. H. et al. Determining the internal quantum eciency o highly ecientpolymer solar cells through optical modeling.Appl. Phys. Lett.90, 143506 (2007).

66. Ferry, V. E., Sweatlock, L. A., Pacici, D. & Atwater, H. A. Plasmonicnanostructure design or ecient light coupling into solar cells. Nano Lett.

8, 43914397 (2008).67. Ferry, V. et al. Improved red-response in thin lm a-Si:H solar cells with

nanostructured plasmonic back reectors.Appl. Phys. Lett. 95, 183503 (2009).68. Giannini, V., Zhang, Y., Forcales, M. & Gmez Rivas, J. Long-range

surace plasmon polaritons in ultra-thin lms o silicon. Opt. Express16, 1967419685 (2008).

69. Mapel, J. K., Singh, M., Baldo, M. A. & Celebi, K. Plasmonic excitation oorganic double heterostructure solar cells.Appl. Phys. Lett.90, 121102 (2007).

70. vingstedt, K., Persson, N. K., Ingan, O., Rahachou, A. & Zozoulenko,I. V. Surace plasmon increase absorption in polymer photovoltaic cells.Appl. Phys. Lett.91, 113514 (2007).

71. Heidel, . D., Mapel, J. K., Singh, M., Celebi, K. & Baldo, M. A. Suraceplasmon polariton mediated energy transer in organic photovoltaic devices.Appl. Phys. Lett.91, 093506 (2007).

72. Haug, F. J., Sderstrm, ., Cubero, O., errazzoni-Daudrix, V. & Balli, C.Plasmonic absorption in textured silver back reectors o thin lm solar cells.J. Appl. Phys.104, 064509 (2008).

7/28/2019 Plasmonics for improved photovoltaic devices

9/10

nature materials | VOL 9 | MARCH 2010 | www.nature.com/naturematerials 213

review articleNaTure maTerIalsdoi: 10.1038/nmat2629

73. vingstedt, K., Person, N. K., Ingans, O., Rahachou, A. & Zozoulenko, I. V.Surace plasmon increase absorption in polymer photovoltaic cells.Appl. Phys. Lett.91, 113514 (2007).

74. Franken, R. H. et al. Understanding light trapping by light-scatteringtextured back electrodes in thin-lm nip silicon solar cells. J. Appl. Phys.102, 014503 (2007).

75. Schropp, R. E. I. etal. Nanostructured thin lms or multibandgap silicon triplejunction solar cells. Sol. Energy Mater. Sol. C.93, 11291133 (2009).

76. Rockstuhl, C., Fahr, S. & Lederer, F. Absorption enhancement in solar cells bylocalized plasmon polaritons.J. Appl. Phys.104, 123102 (2008).

77. Keevers, M. J., Young, . L., Schubert, U. & Green, M. A. 10% ecient CSGminimodules. Proc. 22nd Eur. Photovoltaic Solar Energy Conf. Milan, Italy,37 September 2007.

78. Green, M. A., Zhao, J., Wang, A. & Wenham, S. R. Very high eciency siliconsolar cells: science and technology. IEEE Trans. Electr. Dev.46, 19401947 (1999).

79. Fahr, S., Rockstuhl, C. & Lederer, F. Metallic nanoparticles as intermediatereectors in tandem solar cells.Appl. Phys. Lett.95, 121105 (2009).

80. Pacici, D., Lezec, H. & Atwater, H. A. All-optical modulation by plasmonicexcitation o CdSe quantum dots. Nature Photon.1, 402406 (2007).

81. Walters, R. J., van Loon, R. V. A., Brunets, I., Schmitz, J. & Polman, A. Asilicon-based electrical source o surace plasmon polaritons. Nature Mater.9, 2125 (2010).

82. Verhagen, E., Kuipers, L. & Polman, A. Field enhancement in metallicsubwavelength aperture arrays probed by erbium upconversion luminescence.Opt. Express17, 1458614597 (2009).

83. OCarroll, D., Homann, C. E. & Atwater, H. A. Conjugated polymer/metalnanowire heterostructure plasmonic antennas.Adv. Mater. doi:10.1002/

adma.200902024 (2009).84. Labeke, D. V., Gerard, D., Guizal, B., Baida, F. I. & Li, L. An angle-

independent requency selective surace in the optical range. Opt. Express14, 1194511951 (2006).

85. De Waele, R., Burgos, S. P., Polman, A. & Atwater, H. A. Plasmon dispersionin coaxial waveguides rom single-cavity optical transmission measurements.Nano Lett.9, 28322837 (2009).

86. Kroekenstoel, E. J. A., Verhagen, E., Walters, R. J., Kuipers, L. & Polman, A.Enhanced spontaneous emission rate in annular plasmonic nanocavities.Appl. Phys. Lett.95, 263106 (2009).

87. Pijpers, J. J. H. et al. Assessment o carrier-multiplication eciency in bulkPbSe and PbS. Nature Phys.5, 811814 (2009).

88. King, R. R. et al. 40% ecient metamorphic GaInP/GaInAs/Ge multijunctionsolar cells.Appl. Phys. Lett.90, 183516 (2007).

89. Coutts, . J. et al. Critical issues in the design o polycrystalline, thin-lmtandem solar cells. Prog. Photovolt. Res. Appl.11, 359375 (2003).

90. Schuller, J. A. et al. Plasmonics or extreme light concentration and

manipulation. Nature Mater. 9, 193204 (2010).91.Verschuuren, M. A. & van Sprang, H. A. 3D photonic structuresby sol-gel imprint lithography.Mater. Res. Soc. Symp. Proc.1002, 1002-N03-05 (2007).

92. Reilly, ., van de Lagemaat, J., enent, R. C., Mora, A. J. & Rowlen, K. L.Surace-plasmon enhanced transparent electrodes in organic photovoltaics.Appl. Phys. Lett.92, 243304 (2008).

93. US Geological Survey 2004 .94. Green, M. A. Improved estimates or e and Se availability rom Cu anode

slimes and recent price trends. Prog. Phot.14, 743 (2006).

ackwgWe thank M. Bonn, K. Catchpole, V. E. Ferry, J. Gomez Rivas, M. Hebbink, J. N. Munday,

P. Saeta, W. C. Sinke, R. E. I. Schropp, K. anabe, E. Verhagen, M. A. Verschuuren,

R. de Waele and E. . Yu or discussions. Tis work is supported by the Global Climate

and Energy Project. Te FOM portion o this work is part o the research programme

o FOM and o the Joint Solar Panel programme, which are both nancially supportedby the Netherlands Organisation or Scientic Research (NWO). Te Caltech portion

o this work was supported by the Department o Energy under grant number

DOE DE-FG02-07ER46405.

a Te authors declare no competing nancial interests.

7/28/2019 Plasmonics for improved photovoltaic devices

10/10

In Fig. 1a o the version o this Review originally published, the graph labelled 2-m-thick Si waer is that or a 10-m-thick Si waer.Te corrected fgure is shown below. Te original fgure caption and descriptions in the text are correct. Te fgure has been corrected inthe HML and PDF versions o this Review.

Plass f pv phtvlta vs

Harry A. Atwater and Albert Polman

Nature Materials9, 205213 (2010); published online 19 February 2010; corrected online 1 September 2010.

corrigendum

400 600 800 1,400 1,6001,000 1,200

Freespace wavelength (nm)

1.6

1.4

1.2

0.8

0.6

0.4

0.2

0

1.0

Spectralintensity(Wm

2n

m1)

2mthick

Si wafer

AM1.5 solar spectrum

Blue Green Red

Si

a