2013石墨烯在太阳能电池应用GrapheneSeestheLight

IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 20, NO. 1, JANUARY/FEBRUARY 2014 4000107The Potential of Grapheneas an ITO ReplacementinOrganic Solar Cells: An Optical PerspectiveWee Shing Koh, Member, IEEE, Choon How Gan, WeeKee Phua, Yuriy A. Akimov, and Ping Bai, Member,IEEEAbstract — Graphene possessesinnate potential to replace in-dium tin oxide (ITO) as the transparent electrode in an organic so-lar cell device. From our transmittance study weighted with the airmass 1.5 global (AM1.5G) solar spectrum, it is found that a highertransparency may be obtained for up to four layers of graphene incomparison to ITO. Our ?ndings suggest that replacing ITO withmonolayer graphene in organic solar cells yields comparable per-formance. Due to the increased optical absorption, organic solarcells with four-layer graphene (with the same sheet resistance asITO at 30Ω / ) are capable of attaining at least 92% of the sameorganic photovoltaic device with an optimized ITO electrode forboth normal and angular AM1.5G illumination.Index Terms— Computational modeling, organic semiconduc-tors, photovoltaic cells, thin ?lm devices.I. INTRODUCTIONG RAPHENE, an isolated graphite plane, is actually crys-tallized carbon in a 2-D form [1]. Since its extraction fromgraphite compounds, its electronic and optical properties havebeen extensively studied [2]. It was found to exhibit high opticaltransparency [3], electrical conductivity [4], as well as chemicaland thermal stabilities among others. These excellent proper-ties make graphene an ideal candidate for the replacement of acommonly used transparent conducting oxide, indium tin oxide(ITO), prevalently used in the electronic displays and solar cellindustries [5], [6]. Moreover, graphene is much more ?exiblethan the brittle ITO; thus, graphene is also viewed as an enablerin ?exible electronics, in particular organic solar cells.Although graphene has been shown to exhibit excellent opti-cal transmittance of more than 90%, organic photovoltaic (OPV)devices which use graphene as the transparent electrode havenot yielded better performance as compared to ITO-based de-vices. For instance, Wu et al. used solution-processed func-tionalized graphene ?lms of 4–7 nm as transparent electrodesin a bilayer CuPc/C 60 organic solar cell, and demonstratedpower conversion ef?ciency (PCE) less than 50% of the ref-erence cell PCE which uses ITO [7]. Similarly, Wang et al.Manuscript received December 21, 2012; revised January 31, 2013; acceptedFebruary 11, 2013.W. S. Koh, W. K. Phua, Y. A. Akimov, and P. Bai are with the Electronicsphuawk@ihpc.a-star.edu.sg; akimov@ihpc.a-star.edu.sg; baiping@ihpc.a-star.edu.sg).C. H. Gan is with with the College of Engineering, Mathematics andPhysical Sciences, University of Exeter, Exeter EX4 4QF, U.K. (e-mail:C.H.Gan@exeter.ac.uk).Color versions of one or more of the ?gures in this paper are available onlineat http://ieeexplore.ieee.org.Digital Object Identi?er 10.1109/JSTQE.2013.2247976have demonstrated that using noncovalent modi?cation of agraphene transparent electrode, the PCE of a typical conju-gated polymer poly(3-hexyl)thiophene:phenyl-C 61 -butyric acidmethyl ester or P3HT:PCBM bulk-heterojunction photovoltaicdevice with graphene transparent anode is about 55% of theITO-based device PCE of 3.1% [8]. A recent improvement ofthe same P3HT:PCBM device by the same group with a four-layer graphene electrode attained 83.3% of the PCE of the samedevice with ITO [9]. For small molecule OPV devices, a chem-ical vapor-deposited graphene electrode has brought the PCEof multilayer graphene-based CuPc:C60 OPV device, which is1.18%, to nearly 93% of the 1.27% PCE of the ITO-based OPVdevice [10]. Besides chemical vapor deposition, attempts to p-doped graphene with AuCl 3 in nitromethane have been shownto produce graphene with much smaller sheetresistance. TheirPCE of a three-layer graphene-based CuPc:C 60 OPV device is1.63% ascompared to 1.77% PCE of the similar ITO device [11].In this paper, we evaluate the potential of doped graphene asan alternative to the typical transparent electrode made of ITOin terms of its transmittance and optical absorption. Some ofthe questions which we will attempt to answer include: 1) Is itpossible to improve the performance of organic solar cells byreplacing ITO with graphene? 2) Is the improvement or reduc-tion in performance valid for different OPV systems with activelayers that absorb in different ranges of optical wavelengths? 3)Can we improve the angular absorption of the active layers byreplacing ITO with graphene?II. GRAPHENE /ITO ON GLASSTo determine the impact of replacing ITO with graphene,we ?rst look at a simple glass(1.1 mm)/ITO/air or glass(1.1 mm)/graphene/air con?guration (see the inset in Fig. 1).Normally incident monochromatic plane waves illuminate thebilayer structure (i.e., light passes through glass ?rst and thengraphene or ITO as in a typical organic solar cell).The integral transmittance Tint through the structure for λ =350– 800nm is determined by weighting the transmission coef-?cient to account for the different intensities of the air mass 1.5global (AM1.5G) solar illumination (from λ = 350– 800nm)and normalized by the incident power of the spectrum given asTint =λ =800 nmλ =350 nm T (λ )S(λ )dλλ =800 nmλ =350 nm S(λ )dλ(1)where S(λ ) is the AM1.5G solar spectrum, T( λ ) = P b o t t o m (λ )I 0is the normalized wavelength-dependent transmittance ofglass(1.1 nm)/ITO or graphene/air con?guration at each λ andis computed by solving Maxwell ’s equations using the transfer1077-260X/$31.00 ? 2013 IEEE4000107 IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 20, NO. 1, JANUARY/FEBRUARY 2014Fig. 1. Integral transmittance Tint of light (λ = 350– 800nm) illuminating1.1-mm-thick glass and graphene (ITO) as a function of number of graphenelayers (ITO thickness).matrix method [12], and Pbottom (λ ) = E × H is the Poyntingvector (i.e., E is the electric ?eld vector and H is the magnetic?eld vector) taken below the graphene(ITO) layer, after the lightpassesthrough it. Here, I 0 taken as 1 W/cm 2 is the referenceincident light intensity. In the computation of T (λ ), the relativepermittivity of glass is assumed to be 2.2201 and the complexrelative permittivity of ITO is taken from [13]. The frequency-dependent optical conductivity σ (ω ) of graphene is computedtaking the carrier mobility to be μ m = 9000 cm2 /Vs, which istypical of measured values [14], [15]. Here, ω = 2π c/λ is thean-gular frequency with c being the speedof light in vacuum. Let ustake the carrier density to be nc = 6 × 1012 cm - 2 , correspond-ing to a sheet resistance ρ = 1/ (n c eμ m ) ≈ 120 Ω / for dopedgrapheneproduced by roll-to-roll technology [16]. We may pro-ceed to estimate the chemical potential μ c ≈ vF √ n cπ ～ 0.3eV (μ c K B τ ～ 26 meV, temperature τ taken to be 300 K),where vF ≈ 108 cm/s is the Fermi velocity. As the sheet re-sistance directly affects the performance of a solar cell, wehave employed analytical calculations for σ (ω ) so that one mayquantify changes in the optical conductivity of graphene withrespect to variations in the sheet resistanceor carrier density. Forsmall wave vectors k (i.e., |k |vF ω ) and at high frequencieswhere ω τ - 1e (τ e being a phenomenological electron relax-ation time in the order of 10- 13 s) [17], the optical conductivityσ (ω ) = σ inter ( ω ) + σ intra (ω ) of graphene may be expressedas [18], [19]σ inter (ω ) = ie2ωπ 2∞0F ( - ε ) - F (ε )ω 2 - 4(ε/ )2 dε= e24 1 +iπ lnω - 2μ cω + 2μ c, ( μ c K B τ ) (2)where σ inter (ω ) takes into account interband electron transi-tions, andσ intra (ω ) = e2iπ 2(ω + iτ - 1e )∞0ε ?F (ε )?ε - ?F (- ε )?ε dε= e2 μ cτ e(1 + i ωτ e)π 2(1 + ω 2 τ 2e ) , (μ c K B τ ) (3)where σ intra ( ω ) takes into account intraband electron – photonscattering, and F ( ε ) = (1 + exp[( ε - μ c )/K B τ ]) - 1 is theFermi – Diracdistribution. For the range of frequencies consid-ered in (1), ω 2τ 2e 1, (σ intra ) ≈ 0, andσ ( ω ) ≈ e24 1 +iπ lnω - 2μ cω + 2μ c+ 4μ cω. (4)The conductivity for few-layer graphene (FLG) may be esti-mated as Nσ (ω ) , where N is the number of layers (N 90% over the 350– 800nm wavelength range, whichresults in better integral transmittance Tint than 130-nm-thickITO with Tint = 87.1%. However, as the number of graphenelayers increases, the integral transmittance decreasesby 1.68%to 1.48% for each additional layer of graphene. Note that withgraphene on glass, the reduction of the transmittance (after scal-ing with the AM1.5G solar spectrum) is no longer a function ofthe ?ne structure constant where we would expect a reduction ofthe transmittance by 2.3% [3], [22] for each additional layer ofgraphene. This decoupling of the reduction in the transmittancefrom the expected 2.3% reduction per additional graphene layercanbe attributed primarily to 1) the weighting of the wavelength-dependenttransmittance T (λ ) basedon the AM1.5G solar spec-trum and 2) to a lesser extent the presence of the glass substrate.For transmittance Tint > 85%, the thickness of ITO has to beeither less than 20 nm or greater than 100 nm. In the commercialuse of ITO, the typical thickness usually adopted is more than100 nm to ensure that the electrical conductivity is not com-promised [7], [10]. Therefore, Fig. 1 suggests that it is possibleto achieve comparable or even higher transmittance than pro-vided by ITO with four or less layers of graphene. In terms ofthe sheet resistance, N = 4 layers of graphene (sheet resistance～ 30 Ω / ) is roughly equivalent to ITO on glass (sheet resis-tance ～ 20– 30Ω / ). The linear decrease in the transmittancegave an indication that absorption may be a signi?cant factorfor graphene as the number of layers is increased. On the otherhand, with ITO asthe transparent electrode, optical interferenceseems to have a much more important effect since it is mostlytransparent in the wavelengths of interest for organic solar cells.III. GRAPHENE AS A T RANSPARENT ELECTRODE IN AP3HT:PCBM ORGANIC SOLAR CELLNext, we consider the use of graphene to replace ITOtransparent electrode in a well-studied conjugated-polymer-based organic bulk-heterojunction solar cell with the followingstructure as shown in Fig. 2(a): glass(1.1 mm)/ITO(130 nm)/PEDOT:PSS(45 nm)/P3HT:PCBM(75 nm)/Ca(10 nm)/Ag(100nm). The wavelength-dependent complex relative permittivitiesof the hole-transporting poly(3,4-ethylenedioxythiophene)poly(styrenesulfonate) (PEDOT:PSS) layer, poly(3-hexylthio-phene):methanofullerene-phenyl C61 -butryic acid methyl ester(P3HT:PCBM in 1:0.8 weight ratio) active layer, calcium (Ca)KOH et al. : THE POTENTIAL OF GRAPHENE AS AN ITO REPLACEMENT IN ORGANIC SOLAR CELLS: AN OPTICAL PERSPECTIVE 4000107Fig. 2. (a) Structure of a P3HT:PCBM organic bulk-heterojunction solar cellwith graphene asthe transparent electrode. Normalized absorbed power of lightin (b) P3HT:PCBM active layer and (c) graphene and ITO, as a function of thenumber of graphene layers (and ITO thickness) under AM1.5G solar illumina-tion from λ = 350– 800 nm. (d) Spectral absorption rate in P3HT:PCBM (solidlines), graphene and ITO (dashed lines) for the OPV device with monolayergraphene, four-layer graphene, and 130-nm-thick ITO.electron transporting layer, and silver (Ag) back re?ector aretaken from [13]. We have to emphasize that the choice of thedimensions and structure of the P3HT:PCBM OPV device issuch that the actual optimized experimental devices were shownto exhibit PCEs of more than 3.5% as shown in [13] and [23].To quantify and compare the performance of theP3HT:PCBMOPV device, we have computed the normalized absorbed powerin the active layer and also the transparent electrode (i.e.,graphene and ITO). The normalized absorbed power of the layerof interest is given byA =λ =800 nmλ =350 nm [Ptop (λ ) - Pbottom (λ )] S(λ ) dλλ =800 nmλ =350 nm S(λ ) dλ(5)where Ptop (λ ) and Pbottom (λ ) are the Poynting vectors of thetop and bottom interface of the layer of interest.Fig. 2(b) illustrates the normalized absorbed power in theP3HT:PCBM active layer with respect to the integrated incidentpower of the AM1.5G solar spectrum, over λ = 350– 800nm,Sint = 575.35 W/m 2 at normal incidence. We observed that thenormalized power absorbed by the P3HT:PCBM photoactivelayer decreaseswith number of graphenelayers (from 57.6% forthe monolayer to 45.0% for ten layers of graphene), while withup to 200-nm-thick ITO, it is about 58.9 – 57.8%.This showsthat even with > 90% transmittance for monolayer grapheneon glass, the performance of the P3HT:PCBM device is onlycomparable to that of an ITO-based device, measured in termsof the power absorbed in P3HT:PCBM. With four layers ofgraphene (i.e., sheetresistance ～ 30 Ω / ), the amount of opticalpower absorbed in P3HT:PCBM is 53.9% or 92.3% of the sameOPV device with a 130-nm-thick ITO transparent electrode.The difference in the power absorbed in the P3HT:PCBM activelayer with four-layer graphene or 130-nm-thick ITO illustratesthat while the graphene electrode allows comparable amountof light to enter the active layer, the amount of light absorbedFig. 3. Normalized absorbed power of the light in the P3HT:PCBM activelayer for monolayer (black-solid line) and four-layer graphene (red-dashed line)transparent electrodes with respect to the absorbed power in the same activelayer with a 160-nm-thick ITO electrode (blue-dotted line) under AM1.5Gsolar illumination from λ = 350– 800nm at different angles of incident light.by the P3HT:PCBM after accounting for multiple passes andoptical interference is lesser compared to the casewhen the ITOelectrode is used.The normalized absorbed power for both graphene and ITOin the organic solar cell con?guration is plotted in Fig. 2(c).It is evident that the absorption in the thin graphene electrode(～ 1.93% for a monolayer of graphene) is much higher thanthat of ITO (0.0006% for 130-nm-thick ITO) even for a mono-layer graphene. Therefore, it is obvious that the absorption ingraphene is a limiting mechanism to replace ITO as a trans-parent electrode. Although graphene absorption losses are somuch higher than ITO, it is instructive to look at the normalizedspectral absorption rate of P3HT:PCBM for both graphene andITO electrodes. To understand why the normalized absorbedpower in P3HT:PCBM is just slightly smaller by 4% betweenthe monolayer graphene case and the 130-nm-thick ITO ref-erence case even though the absorption of graphene is almostten times higher (～ 1.93% for the monolayer graphene caseand0.0006% for the 130-nm ITO reference case), we plot the nor-malized spectral absorption pro?le of P3HT:PCBM for both thegraphene and ITO electrodes.Fig. 2(d) shows the normalized spectral absorption inP3HT:PCBM for monolayer graphene, four-layer graphene, and130-nm ITO before weighting with the AM1.5G solar spec-trum for λ = 350– 800nm. We observed that compared to ITO,graphene facilitates broadband absorption of light for the wholerange of wavelengths where P3HT:PC