Documents

The Limits to Organic Solar Cells

Description
Organic solar cells
Categories
Published
of 5
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Transcript
  28MRS BULLETIN ã VOLUME 30 ã JANUARY 2005 The Energy Conversion Process To determine the limits to the power con-version efficiency of organic solar cells, we begin by referring to the optical-to-electricalconversion process detailed in Figure1.The internal quantum efficiency,  IQE , is theproduct of four efficiencies, 1 each corre-sponding to a step in the charge generationprocess:  IQE   A  ED  CT  CC ,(1)where  A is the absorption efficiency of lightwithin the active region of the solar cell;  ED is the exciton diffusion efficiency to a dis-sociation site;  CT is the charge transferefficiency, which is the efficiency for disso-ciation of an exciton into a free electron andhole pair at that site; and  CC is the chargecollection efficiency. Taking into considera-tion the optical losses that occur on cou-pling light in the device active region, wearrive at the external quantum efficiency:  EQE  (1 – R )  IQE ,(2)where R is the reflectivity of the substrate–air interface. Finally, the power conversionefficiency of the cell is given by(3)  P  V  OC  J  SC FFP inc ,where FF is the fill factor, V  OC is the open-circuit voltage,  J  SC is the short-circuit cur-rent density, and P inc is the incident powerdensity.Organic photovoltaic (PV) cells confrontseveral limitations that are apparent byexamining Equations1–3. First, there is aninherent tradeoff between the absorptionand the exciton diffusion efficiencies. Thatis, the exciton diffusion length, L D , is typi-cally much less than the optical absorptionlength, 1/  . For an absorbing organic layerof thickness d , we have  A  (1 – e  d ),(4)whereas, assuming that the arrival of exci-tons at a dissociation site is independentof electric fields or other extrinsic condi-tions, the diffusion efficiency is given by(5)In general, it has been found that thecharge collection and charge transfer effi-ciencies at organic donor/acceptor (DA)interfaces commonly used in thin-filmmolecular organic semiconductor PV cellsapproach 100%, in which case the internalquantum efficiency is determined by theproduct  A  ED . This is shown in Figure2 asa function of d / L D for  L D  0.1, 0.2, and 1.0,  ED  e  d / L D .the former two values being more charac-teristic of organic semiconductors eithercast from solution (in the case of polymers)or grown from the vapor phase (as in thecase of small-molecular-weight materials).Under the conditions of  L D  1, thethickness of the active organic regionshould be such that d  L D . However, as L D decreases, this leads to a decreasingabsorption efficiency, ultimately leadingto  IQE  0.1 for materials commonlyemployed in thin-film bilayer cells basedon molecular organic semiconductors. In-deed,  IQE is very close to the value obtained by C.W. Tang 2 in the earliest demonstrationof such a cell based on the DAmaterialscombination of copper phthalocyanine(CuPc) and 3,4,9,10-perylenetetracarboxylic- bis-imidazaole (PTCBI), where L D  5 nmand 1/   200nm (corresponding to  L D   0.025), consistent with the results in Figure2. The situation corresponding to L D    1/  , which is characteristic of al-most all organic materials used in PV cells,forms an exciton diffusion bottleneck  , wherebyphotogenerated excitons cannot reach aDAinterface prior to dissociation into freecarriers, ultimately limiting cell efficiency.This limitation on  IQE results in a powerconversion efficiency of only 1% for bilayercells—a value that was not convincinglyexceeded in either polymer or molecularorganic cells until the 21st century. Recently,the exciton diffusion bottleneck has beenreduced, thereby increasing organic PV cellefficiency, by applying several strategiesthat have influenced the device architec-ture. The most effective strategies include:1.Employing a double heterostructure, 3 thereby increasing both 1/  and L D ;2.Employing materials with long-rangeorder, 4 thereby increasing L D ;3.Employing a bulk 5,6 or mixed hetero- junction 7,8  between the donor and accep-tor materials to increase 1/  withoutdecreasing L D ; and4.Employing light-trapping schemes 3 ormulti-heterojunction (tandem) cells 9,10 toincrease the optical path length within thethin film, thereby increasing  A .For the remainder of this article, we will briefly discuss each of these schemes, par-ticularly as they have been applied tosmall-molecular-weight organic thinfilms. We will show that a combination of strategies has led to  P  5.7% (under AM1.5 G simulated solar radiation at 1 sun in-tensity*) 11 in only the last few years, sug- T he Limits to OrganicPhotovoltaic CellEfficiency Stephen R.Forrest Abstract We consider the fundamental limits to organic solar cell efficiency, and the schemesthat have been used to overcome many of these limitations.In particular, the use ofdouble and bulk heterojunctions, as well as tandem cells employing materials with highexciton diffusion lengths, is discussed.We show that in the last few years, a combinationof strategies has led to a power conversion efficiency of  P  5.7% (under AM 1.5 Gsimulated solar radiation at 1 sun intensity) for tandem cells based on small-molecular-weight materials, suggesting that even higher efficiencies are possible.We conclude byconsidering the ultimate power conversion efficiency that is expected from organic thin-film solar cells. Keywords:  organic photovoltaics, solar cell efficiency. *Note that efficiency measurements based onspectra obtained from laboratory solar simulatorscan differ from those that adhere more closelyto the terrestrial solar spectrum. See, for ex-ample, the introductory article by Shaheen etal.in this issue. www.mrs.org/publications/bulletin  The Limits to Organic Photovoltaic Cell Efficiency MRS BULLETIN ã VOLUME 30 ã JANUARY 200529 gesting that even higher efficiencies arepossible. This article will conclude by con-sidering the ultimate power conversionefficiency that is expected from organicthin-film solar cells. Double Heterojunctions andIncreasing the Exciton DiffusionLength In addition to the exciton diffusion bottle-neck, bilayer PV cells suffer from two otherproblems that significantly limit theirpower conversion efficiencies. First, deposi-tion of the cathode metal onto the acceptorlayer introduces damage (typically in theform of deep trap levels), thereby reducingthe exciton diffusion length in that layer.Second, the incident optical electric fieldvanishes at the surface of the highly con-ducting electrode. Since the active layerthicknesses must typically be d    /4 n  100nm, where  is the incident lightwavelength and n is the spatially weightedaverage refractive index of the thin-filmcell, the optical field intensity is less thanits maximum value at the heterojunctionwhere photoinduced charge transfer ismost efficient.To circumvent these problems, the dou- ble heterostructure organic PV cell was in-troduced, 3 as illustrated schematically inFigure3(inset). Here, a transparent exci-ton blocking layer (EBL) is interposed between the acceptor layer and the metalcathode. This EBLdisplaces the active DAinterface such that it is centered near wherethe optical field intensity is at a maximum,thereby maximizing absorption. Further-more, as this particular layer is photoelec-trically “inert” (in that it does not absorblight), it can absorb the damage introducedduring cathode metal deposition withoutthat damage resulting in exciton quench-ing. Due to the larger energy gap (and hencetransparency) of the EBLas compared withthat of the acceptor layer, the electrons areconfined close to the DAheterointerface,thus preventing them from migrating tothe damaged region near the cathode. Onthe contrary, the damage to bathocuproine(BCP) EBLs has been shown to result inthe sufficiently high electron conductivityrequired to extract separated charges fromthe DAinterface. 1 Employing a double heterostructure al-lows for the use of very thin active regionswithout a reduction in  EQE . In the firstdemonstration based on the archetypeCuPc/PTCBI system, an  8-nm-thickBCPlayer was used as the EBLwhile theactive DAlayers were each only  10nmthick. Remarkably,  EQE slightly increased(see Figure3), even though  A was reduced by more than a factor of 2.5, providing evi-dence for exciton confinement in the DAinterface region by the EBL. Thinning theactive layers has the additional benefit of substantially reducing cell series resistance.This increases the fill factor, as well as al-lowing for efficient response even at veryhigh incident light intensities (  10 suns),thereby allowing these thin-film cells to beused with solar concentrators and otherlight-trapping schemes. 3 Finally, combining the double hetero-structure architecture with materials withlong diffusion lengths has resulted in asignificant increase in efficiency 4 over thatfirst reported by Tang. 2 For example, re-placing the acceptor, PTCBI, with C 60 leadsto an increase in the exciton diffusion lengthfrom 3nm to 40nm. Many of the advanta-geous properties of C 60 arise from itsspherical symmetry: it can pack tightly toform highly conductive films with excellentorbital overlap between adjacent mole-cules, thereby improving both the electronand exciton diffusion efficiencies, and theintersystem crossing resulting from thelarge orbital angular momentum inherentin the π -electron system converts all excitedstates to triplets with their correspondinglylong diffusion lengths. Using C 60 in an Figure1.Steps in the photocurrent generation process.The horizontal lines to the right and left of each illustration correspond to the Fermi energies (  E F   ) of the cathode and anode contacts, respectively.The boxes correspond to the acceptor (left) and donor (right) energy gaps, respectively.Here, LUMO is the lowest unoccupied molecular orbital, and HOMO is the highest occupied molecular orbital of the organic film.Red dots are electrons and red circles are holes, with dashed lines drawn between them to represent excitons.Also, the dip in the energy levels in the vicinity of the exciton qualitatively depicts its binding energy (   0.5–1eV), placing this quasi-particle at an energy somewhat below that of the HOMO– LUMO gap energy.   A is the absorption efficiency of light within the active region of the solar cell,   ED  is the exciton diffusion efficiency to a dissociation site,  CC  is the charge collection efficiency, and   CT  is the charge transfer efficiency.Figure2.The product of adsorption efficiency (    A  ) and the exciton diffusion efficiency (    ED   ) illustrating the depend- ence of the “exciton diffusion bottleneck” on the ratio of layer thickness (  d  )to exciton diffusion length, L D  .Here,  is the optical absorption length within a particular layer.  30MRS BULLETIN ã VOLUME 30 ã JANUARY 2005 The Limits to Organic Photovoltaic Cell Efficiency otherwise conventional double hetero- junction cell has resulted in efficiencies 4 of   P  3.6%, increasing to 4.2% at  1 sunillumination intensity 12 (simulated AM1.5G spectrum) in low-series-resistancestructures.The calculated relationship between in-ternal quantum efficiency and active layerthickness for different values of diffusionlength, along with values obtained for rep-resentative small-molecular-weight bilayerPV cells, is shown in Figure4, along withseveral experimental results. Clearly, em-ploying materials with a large L D results ina peak in  IQE at larger layer thicknesses,with the C 60 -based device correspondingto L D  20nm. Bulk and Mixed Heterojunctions An alternative approach to overcomingthe exciton diffusion bottleneck is to forma heterojunction (HJ) between the donorand acceptor materials with a very largesurface area, as shown in Figure5a. In thiscase, by entangling the regions containingthe two constituents, a region is formedwhereby photons can be absorbed over avery long distance, creating all excitonswithin a diffusion length of a DAinterfacewhere photoinduced charge transfer canoccur. The first demonstration of such a“bulk” heterojunction employed a blendof a polymer and a fullerene that formedan entangled network extending acrossthe device. This approach has resulted in asignificant improvement in power conver-sion efficiency over that obtained using asimple planar structure. 6,13 To achieve a bulk HJ based on small-molecular-weight materials is more diffi-cult, since the phase separation must occurin the solid rather than the liquid phase, asin the case of polymers. Simply mixingmaterials by coevaporation of the donorand acceptor source molecules onto a sub-strate can result in a significant decrease incharge carrier mobility, as has been ob-served by the co-deposition of the arche-type DApair, CuPc and PTCBI. 5 Thereduction in carrier mobility (and, hence,increase in cell series resistance) resultsfrom the planar shape of the molecules. Ina pure thin film, both CuPc and PTCBI formorderly stacks, allowing charge transportalong the overlapping  -electron systemsin the stacking direction. When depositedinto a mixture, the stacks are disrupted, in-troducing charge trapping and scatteringduring transport to the electrodes. Hence,mixed-layer devices typically have verylow (  0.1%) power conversion efficiencies.The mixed thin film, however, is not anequilibrium structure. Thus, phase segre-gation into an entangled HJ can be achieved by annealing the mixture at high tempera-tures. For example, Peumans and co-workers 5 have shown that annealingCuPc/PTCBI mixed-layer double HJ cellsat  2000  C for  10 min can result in in-ternal crystallization of the film in the bulkHJ structure shown in Figure5a. In thiscase, the solar power conversion efficiencyrises from 1% for a simple planar HJ, tonearly 1.5% under 1 sun, AM 1.5 G simu-lated solar radiation. However, the fill fac-tor of the annealed cell is only FF  0.31,which is somewhat lower than a planarHJ, where FF  0.52. This reduction indi-cates that the phase separation has resultedin resistive bottlenecks and cul-de-sacswhere the free charge is trapped prior tocollection at the electrodes, which in manycases may present a significant limitationto the efficiency achievable using thisapproach.Aconsiderably different situation appliesfor CuPc/C 60 mixtures. In this case, thespherically symmetric C 60 apparently doesnot disrupt the CuPc stacking to the sameextent as in similar mixtures with the pla-nar PTCBI molecule. Hence, CuPc/C 60 mixed cells do not exhibit an improve-ment on annealing. On the contrary, theas-deposited mixed cell itself has a high Figure3.Quantum efficiency versus layer thickness for a bilayer CuPc/ PTCBI cell with and without an exciton blocking layer (EBL).Both the external (    EQE   ) and internal (    IQE   ) quantum efficiencies are shown.Inset shows the energy-level scheme of a double heterojunction photovoltaic cell.The dashed line indicates defect levels that allow for conduction of electrons (solid circle) to the cathode (Ag).The shaded circles represent Ag atoms or clusters that may have diffused into the wide- energy-gap (and therefore transparent)EBL during the deposition process.Here, D and A represent the donor and acceptor materials, respectively, and ITO is the transparent indium tin oxide anode.Figure4.Calculated internal quantum efficiency versus donor and acceptor layer thicknesses.The solid curves are for several different exciton diffusion lengths, L D  , and the dotted curves (labeled 20 Å, 40 Å, 100 Å, 200 Å, and 400 Å) are L D  values for simple exciton diffusion, neglecting the optical boundary conditions.Experimental data points for several different cell configurations are also shown.Figure5.(a)A bulk heterojunction consisting of an entangled region of donor and acceptor materials, shown as open and shaded areas.Excitons are indicated by a close pairing of an electron (solid circle) and hole (open circle) enclosed with a dotted line.(b)A mixed heterojunction shown on the microscopic scale, with percolating conducting pathways formed by stacks of the planar CuPc molecules, as well as by the close-packed spherical C  60  molecules.  The Limits to Organic Photovoltaic Cell Efficiency MRS BULLETIN ã VOLUME 30 ã JANUARY 200531 efficiency, equal to that of a planar cellcontaining the same materials. 8 Coupledwith the fact that “annealing” theCuPc/C 60 mixed PV cell has no significanteffect on the device efficiency, this sug-gests that the deposited mixture is anequilibrium phase, where percolating“bulk heterojunction-like” conductionpaths are already formed during deposi-tion, as shown in Figure5b.Further benefits can be realized using acombination of a planar and a mixed layerwithin the same double heterostructure,known as the hybrid planar-mixed hetero- junction (PM-HJ). This structure takesmaximum advantage of both the chargeconduction properties of homogenouslayers of an organic film, and the excitonseparation properties of films consistingof a mixture of the donor and acceptormolecular species. 14 Here, in a single de-vice structure, the homogeneous donorand acceptor layers form a sandwich, withthe mixed layer as the filling. Analysis of carrier transport and charge separation inthe PM-HJ cell has shown 14 that the thick-nesses of the homogeneous layers are op-timized when they are on the scale of L D ,whereas the mixed-layer thickness should be on the order of the carrier collectionlength ( L c )—i.e., the distance a free carriercan diffuse within a layer prior to recom- bination with its opposite charge. Thetotal thickness of the device then becomesthe sum of the thicknesses of these threecharacteristic lengths, resulting in increasedoptical absorption over that of a somewhatthinner bilayer heterojunction.Arecent demonstration of a PM-HJ de-vice consisted of a combination of the donor,CuPc, and the acceptor, C 60 . Since the dif-fusion lengths as well as the carrier collec-tion lengths are all in the range of 20–40nm,the device active region thickness increasedto  50–60nm. The efficiency 14 measuredunder 1 sun illumination and AM 1.5 Gsimulated solar radiation was 14  P  5.0%with a high fill factor (  0.6), indicating theclear benefits of the PM-HJ structure. Tandem Cells One significant limitation of almost allorganic compounds is their relatively nar-row absorption spectra. Hence, it is notfeasible to absorb the entire solar spectrumusing a single DApair forming the organicPV cell. Furthermore, the open-circuitvoltage that is produced by most organicDAheterojunction cells is small (typicallyranging between 0.4V and 0.8V). Oneparticularly powerful means of circum-venting these shortcomings is the use of amulticell series stack of PV cells, each op-timized to absorb in a different region of the solar spectrum. In this so-called tandemarchitecture (Figure6), currents generateddue to absorption within each subcell flowin series to the opposing contacts. In a seriesconfiguration, the photocurrent flowing inthe tandem cell is limited by the smallestcurrent generated by a particular subcellin the stack. Furthermore, the open-circuitvoltage is equal to the sum of the open-circuit voltages of the subcells. 9,10 The example tandem cell shown in Fig-ure6 consists of two PM-HJ cells in series. 11 To allow the charges that flow betweenthe subcells (as opposed to those that flowto the electrodes) to recombine within thedevice interior, the tandem cell requiressites that serve to attract carriers of oppo-site charge. One means to generate suchsites is to evaporate ultrasmall (  5Å dia-meter) metal particles onto the surface of the first cell prior to depositing the layersfor the second subcell. This nanoclusterlayer is sufficiently thin to efficiently pro-vide recombination centers, yet not thickenough to absorb light on its way to the back cell, nearest the reflecting cathode.With the plane of polarization of the in-cident light oriented along the particleaxis, separation of free charge on the metalsurface results in an instantaneous oscil-lating dipole, or “surface plasmon.” 1 Thisexcitation re-radiates the light, whose peakintensity lies  10nm from the particle,thereby localizing the radiation almostideally at the nearby DAinterface. 1 Thisconcentrates the light within L D of the ac-tive heterojunction, circumventing the ex-citon diffusion bottleneck. Indeed, in thefirst successful CuPc/PTCBI tandem cells,the Ag nanocluster layer introduced a  20% increase in efficiency beyond thatexpected due to simple absorption in themultilayer cell, achieving  P  2.5% for adouble-layer tandem 1,9 (versus  P  1.9%,as expected due to the increased cell thick-ness and, hence, increased absorption).In the case of the cell in Figure6, thefront cell was designed to preferentiallyabsorb red light by making it rich in CuPc,whereas the back cell preferentially ab-sorbed blue solar radiation due to thelarger total thickness of C 60 in that region. 11 By this means, the efficiency of the tandemcell is optimized to respond to the broadestpossible span of the solar spectrum usingonly the standard materials of CuPc, C 60 ,and PTCBI. It was found that an efficiencyof  P  5.7% (1 sun illumination and AM1.5 G simulated solar radiation) 11 was ob-tained with a fill factor equal to that of asingle junction cell (  0.6) and V  OC  1.02V,corresponding to a doubling of V  OC com-pared with that of a single-element CuPc/C 60 cell.Analysis of the tandem architectureusing this particular materials combina-tion has indicated that cells with efficien-cies of 6.5% are possible, 11 although thishas not yet been achieved. Further im-provements can be expected using im-proved coupling of light into the cell activeregion, and extending the materials com- binations to reach into the infrared part of the solar spectrum, perhaps by employinga third subcell in the stack. The Outer Limits Figure 7illustrates the recent progressmade in improving the efficiency of organicthin-film photovoltaic cells. Following along period since the demonstration of thefirst DAheterojunction by Tang 2 in 1986,there has recently been a rapid increase incell efficiency. Many of these latest develop-ments are due to lessons learned fromimproving the performance of organiclight-emitting devices that often employmaterials and device structures that havedirect application to light detection. As of this writing, the maximum reported effi-ciency obtained for an organic PV cell in adual tandem structure is 5.7%. It is reason-able, therefore, to ask what the limits tocell efficiencies might be.Based on simple physical considerations,the ultimate limitation to the cell efficien-cies is the charge separation process at the Figure6.A bilayer tandem photovoltaic cell consisting of a stack of planar mixed double heterojunction subcells.The cell nearest the Ag cathode (the back cell)is rich in C  60  , which absorbs in the blue spectral region, whereas the front cell is rich in CuPc, which absorbs preferentially in the red and yellow spectral regions.Hence, the cell is designed to have a maximum overlap between a given layer and the part of the solar spectrum which it optimally absorbs.Such a tandem cell,including the Ag nanocluster layer that enhances absorption via surface plasmon effects, has demonstrated power conversion efficiencies of 5.7% under 1 sun, AM 1.5 G simulated illumination.(See Reference11.)
Search
Similar documents
View more...
Tags
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks