+ All Categories
Home > Documents > Photophysics and morphology of a polyfluorene donor-acceptor triblock copolymer for solar cells

Photophysics and morphology of a polyfluorene donor-acceptor triblock copolymer for solar cells

Date post: 04-Dec-2023
Category:
Upload: independent
View: 0 times
Download: 0 times
Share this document with a friend
14
Photophysics and Morphology of a Polyfluorene Donor–Acceptor Triblock Copolymer for Solar Cells Chao Yan, 1* Ashley J. Cadby, 2 Andrew J. Parnell, 2 Weihua Tang, 1 Maximilian W. A. Skoda, 3 David Mohamad, 2 Simon P. King, 4 Luke X. Reynolds, 4 Saif A. Haque, 4 Tao Wang, 2 Steven R. Parnell, 5 Andrew B. Holmes, 1 Richard A. L. Jones, 2 David J. Jones 1 1 School of Chemistry, Bio21 Institute, Building 102, 30 Flemington Road, The University of Melbourne, Parkville, Victoria 3010, Australia 2 Department of Physics and Astronomy, The University of Sheffield, Hicks Building, Hounsfield Road, Sheffield S3 7RH, United Kingdom 3 ISIS Pulsed Neutron and Muon Source, Science and Technology Facilities Council, Rutherford Appleton Laboratory, Harwell Science and Innovation Campus, Didcot OX11 0QX, United Kingdom 4 Imperial College London, Department of Chemistry, South Kensington Campus, London SW7 2AZ, United Kingdom 5 Low Energy Neutron Source (LENS) Indiana University, Bloomington, Indiana, Indiana 47408 Correspondence to: D. J. Jones (E - mail: [email protected]) or A. J. Parnell (E - mail: [email protected]) Received 12 July 2013; revised 30 August 2013; accepted 3 September 2013; published online 25 September 2013 DOI: 10.1002/polb.23386 ABSTRACT: We present a study of the optical, structural and device properties of a polyfluorene (PFM)-based (PFM-F8BT- PFM) donor–acceptor triblock copolymer for use in an organic solar cell. Neutron reflectivity is employed to probe the vertical composition profile before and after thermal annealing while the crystallinity was examined using grazing incidence wide- angle X-ray. The absorption spectra and photoluminescence emission for the triblock and analogous blend of PFM with F8BT reveal a greater degree of intermixing in the triblock. However, the triblock copolymer exhibits exciplex emission, which necessitates a geminate polar pair; long-lived exciplex states are detrimental in organic photovoltaic devices. The tri- plet yield in the triblock and the blend is estimated using pho- toinduced absorption, with the triblock copolymer generating a triplet population 20 times that of the blend. This is far from ideal as triplets are wasted states in organic photovoltaic devi- ces and they can also act as scavengers of polarons reducing the efficiency even more. V C 2013 Wiley Periodicals, Inc. J. Polym. Sci., Part B: Polym. Phys. 2013, 51, 1705–1718 KEYWORDS: block copolymers; conjugated polymers; morphol- ogy; organic photovoltaics; photophysics; solar cells INTRODUCTION Organic photovoltaic (OPV) devices have inspired significant academic and industrial excitement due to the possibility of producing large-scale, low-cost renew- able energy. These factors make mass-produced thin film conjugated polymer solar cells highly attractive to tackle the problem of dwindling fossil fuels and the need to de- carbonize the world economy. The power conversion effi- ciency (PCE) of OPV devices has increased steadily in the last decade and the best current single junction polymer device efficiencies range from 7 to 9% and have been achieved using a blend of an alternating copolymer of thieno[3,4-b]thiophene and benzodithophene (known as PTB7) with the soluble fullerene PC 71 BM. 1 Further work has demonstrated that PTB7-based OPV devices can produce inverted devices with an efficiency of 9.2%. 2 The theoretical motivation for using a block copolymer approach in place of a polymer blend is justified, as they would provide very stable structures once in the equilibrium ordered state and so help make their OPV properties stable over the lifetime of the device. The subject of morphology and morphology development has been of intense focus for the entire OPV research community, as the use of different solvents, spin speed, and altering blend ratios can have pro- found effects on the performance of blend devices as the nonequilibrium structure formation mechanisms and the pathways for structure evolution are far from trivial. This makes block copolymer phase separated structures a much simpler way of producing OPV devices, and rules out the intensive optimization processes needed for OPV blends. There would also be intimate contact between the donor Additional Supporting Information may be found in the online version of this article. *Present address: School of Material Science and Engineering, Jiangsu University of Science and Technology, No. 2, Mengxi Rd, Zhenjiang, Jiangsu, People’s Republic of China. V C 2013 Wiley Periodicals, Inc. WWW.MATERIALSVIEWS.COM JOURNAL OF POLYMER SCIENCE, PART B: POLYMER PHYSICS 2013, 51, 1705–1718 1705 JOURNAL OF POLYMER SCIENCE WWW.POLYMERPHYSICS.ORG FULL PAPER
Transcript

Photophysics and Morphology of a Polyfluorene Donor–Acceptor Triblock

Copolymer for Solar Cells

Chao Yan,1* Ashley J. Cadby,2 Andrew J. Parnell,2 Weihua Tang,1 Maximilian W. A. Skoda,3

David Mohamad,2 Simon P. King,4 Luke X. Reynolds,4 Saif A. Haque,4 Tao Wang,2

Steven R. Parnell,5 Andrew B. Holmes,1 Richard A. L. Jones,2 David J. Jones1

1School of Chemistry, Bio21 Institute, Building 102, 30 Flemington Road, The University of Melbourne, Parkville, Victoria 3010,

Australia2Department of Physics and Astronomy, The University of Sheffield, Hicks Building, Hounsfield Road, Sheffield S3 7RH,

United Kingdom3ISIS Pulsed Neutron and Muon Source, Science and Technology Facilities Council, Rutherford Appleton Laboratory,

Harwell Science and Innovation Campus, Didcot OX11 0QX, United Kingdom4Imperial College London, Department of Chemistry, South Kensington Campus, London SW7 2AZ, United Kingdom5Low Energy Neutron Source (LENS) Indiana University, Bloomington, Indiana, Indiana 47408

Correspondence to: D. J. Jones (E-mail: [email protected]) or A. J. Parnell (E -mail: [email protected])

Received 12 July 2013; revised 30 August 2013; accepted 3 September 2013; published online 25 September 2013

DOI: 10.1002/polb.23386

ABSTRACT: We present a study of the optical, structural and

device properties of a polyfluorene (PFM)-based (PFM-F8BT-

PFM) donor–acceptor triblock copolymer for use in an organic

solar cell. Neutron reflectivity is employed to probe the vertical

composition profile before and after thermal annealing while

the crystallinity was examined using grazing incidence wide-

angle X-ray. The absorption spectra and photoluminescence

emission for the triblock and analogous blend of PFM with

F8BT reveal a greater degree of intermixing in the triblock.

However, the triblock copolymer exhibits exciplex emission,

which necessitates a geminate polar pair; long-lived exciplex

states are detrimental in organic photovoltaic devices. The tri-

plet yield in the triblock and the blend is estimated using pho-

toinduced absorption, with the triblock copolymer generating a

triplet population 20 times that of the blend. This is far from

ideal as triplets are wasted states in organic photovoltaic devi-

ces and they can also act as scavengers of polarons reducing

the efficiency even more. VC 2013 Wiley Periodicals, Inc. J.

Polym. Sci., Part B: Polym. Phys. 2013, 51, 1705–1718

KEYWORDS: block copolymers; conjugated polymers; morphol-

ogy; organic photovoltaics; photophysics; solar cells

INTRODUCTION Organic photovoltaic (OPV) devices haveinspired significant academic and industrial excitement dueto the possibility of producing large-scale, low-cost renew-able energy. These factors make mass-produced thin filmconjugated polymer solar cells highly attractive to tackle theproblem of dwindling fossil fuels and the need to de-carbonize the world economy. The power conversion effi-ciency (PCE) of OPV devices has increased steadily in thelast decade and the best current single junction polymerdevice efficiencies range from 7 to 9% and have beenachieved using a blend of an alternating copolymer ofthieno[3,4-b]thiophene and benzodithophene (known asPTB7) with the soluble fullerene PC71BM.1 Further work hasdemonstrated that PTB7-based OPV devices can produceinverted devices with an efficiency of 9.2%.2

The theoretical motivation for using a block copolymerapproach in place of a polymer blend is justified, as theywould provide very stable structures once in the equilibriumordered state and so help make their OPV properties stableover the lifetime of the device. The subject of morphologyand morphology development has been of intense focus forthe entire OPV research community, as the use of differentsolvents, spin speed, and altering blend ratios can have pro-found effects on the performance of blend devices as thenonequilibrium structure formation mechanisms and thepathways for structure evolution are far from trivial. Thismakes block copolymer phase separated structures a muchsimpler way of producing OPV devices, and rules out theintensive optimization processes needed for OPV blends.There would also be intimate contact between the donor

Additional Supporting Information may be found in the online version of this article.

*Present address: School of Material Science and Engineering, Jiangsu University of Science and Technology, No. 2, Mengxi Rd,

Zhenjiang, Jiangsu, People’s Republic of China.

VC 2013 Wiley Periodicals, Inc.

WWW.MATERIALSVIEWS.COM JOURNAL OF POLYMER SCIENCE, PART B: POLYMER PHYSICS 2013, 51, 1705–1718 1705

JOURNAL OFPOLYMER SCIENCE WWW.POLYMERPHYSICS.ORG FULL PAPER

and acceptor groups as block copolymers have a very smallinterfacial region between blocks, which would enable effi-cient exciton separation, and the length scale of the phase sep-aration could be easily controlled by suitable design of theblock copolymer to be less than or comparable to the excitondiffusion length. The block copolymer will also provide abicontinuous network to enable transport of the electronsand holes to the cathode and anode interfaces.

The optimized morphology for a photovoltaic cell requires alarge interfacial area for efficient exciton dissociation, as wellas a continuous network structure to provide suitable con-duction pathways to the electrodes. In OPV devices, the orga-nization of electron donor and acceptor materials is crucialto achieving good device efficiencies; ideal morphologieswith length scales comparable to or smaller than the excitondiffusion length, �10 nm are required.3

Structural and morphology characterization has proved vitalin improving our understanding of device performance inparticular how devices are improved by processing stepsand strategies. Grazing incidence wide-angle X-ray studieshave been used to examine the degree of crystallinity beforeand after annealing as well as the absorption spectra andhave shown that annealing (solvent and thermal) are neces-sary processing step in improving crystallinity of thedomains and red-shifting the maxima in absorption spec-tra.4–7 Furthermore, the vertical component distribution ofan optimized blend of P3HT/PCBM has been studied usingneutron reflectivity (NR).8,9 The annealing stage in optimizedblends was shown to remove the PCBM depleted layer at thecathode interface, a result that partly explained the deviceefficiency improvements after annealing.

Blends based solely on a mixture of two polyfluorenes havealso been examined as potential type 2 heterojunctions.These have typically been inferior to blends based on PCBMand P3HT as PCBM has a high electron affinity and so resultsin efficient photoinduced electron transfer. Optimized puretwo component polymer systems have efficiencies around�2% although they are starting to improve.10 Recent workby the commercial company Polyera has achieved an allpolymer blend OPV device with an efficiency of 6.4%,although the structure of the materials was not disclosedmaking it difficult to comment further on this system.11 Weshall mention some details regarding some of these systemsbut for further information thorough reviews on thesesystems can be found elsewhere.12,13 The most promisingpolyfluorene blends use poly(9,9-dioctylfluorene-co-benzo-thiadiazole) F8BT (an electron acceptor) and triarylaminefunctionalized polyfluorenes (electron donors) such as TFB,PFB, PFMO, and PFM. The triarylamine modification has theeffect of increasing the HOMO level when compared toconventional poly(9,9-octylfluorene).

Recent work on P3HT F8TBT blends has mapped the evolu-tion of the domain size morphology as a function of thermalannealing and showed that a factor of 50 improvement couldbe achieved from the unannealed device to annealing at

180 �C (0.018–0.9%).14 The current best device efficiency(for an openly disclosed) all polymer system stands at 2.7%for a blend of P3HT and poly[2,7-(9,9-didodecylfluorene(PF12TBT)-alt-5,5-(40,70-bis(2-thienyl)-20,10 ,30-benzothiadia-zole)]).15 This polymer is similar to F8BT but with the addi-tion of two thiophene units either side of the BT unit. Thisstudy found large differences in efficiency between differentmolecular weight of PF12BT, with higher molecular weightpolymer giving the highest device efficiency. This was attrib-uted to the polymer maintaining a fine phase separationbetween the two polymers while having good connectivitybetween the ordered P3HT and PF12BT domains.

Block copolymers have the ability to self-assemble into arich variety of morphologies that possess narrow interfacialregions between the different blocks, which are ideal for theefficient charge separation needed for OPVs.16,17 The self-assembled structures of block copolymers offer the possibil-ity of an optimized morphology with a length scale of 10-nm phase separation and also bicontinuous interdigitation(connectivity) of the different donor and acceptor compo-nents. The array of possible morphologies such as lamellarphases, the gyroid phase and vertical rods would be highlyinteresting for photovoltaic devices.18–20 A number ofresearchers have examined the design and synthesis of con-jugated copolymers incorporating donor and acceptorunits21 and in particular rod–coil copolymers.22 However,the relationship between the self-assembled nanostructureof an all-conjugated block copolymer system, its optoelec-tronic properties and device performance along with a fullunderstanding of the lateral and vertical segregation has notyet been made. Here we present a study of a triblock copol-ymer 3a and its deuterated analog 3b to improve the mate-rials properties in comparison with blends of thecorresponding homopolymers (Scheme 1). The ABA triblockcopolymer 3a consists of a central core that is an electronacceptor (F8BT) and the outer blocks are electron donors(PFM). The details of the synthesis will be detailed in a sep-arate publication (submitted for publication). The HOMO–LUMO energy level diagram for the donor and acceptor com-ponents of this system is shown in the electronic SupportingInformation Figure SI1. We have used an array of techniquesto evaluate this novel system. We explored the lateral andvertical component distribution of the self-assembled donorand acceptor blocks, surface morphology, the degree ofordering on the resulting material properties and photo-induced absorption (PA) to evaluate the potential OPVdevice performance. These results highlight the design crite-ria necessary for the single polymer active layer in an all-conjugated triblock copolymer solar cell and will benefitfuture research into achieving the goal of much improveddevice efficiencies.

EXPERIMENTAL

MaterialsThe materials used in the experiment include the donoracceptor copolymer PFM-b-F8BT-b-PFM (3a) and the

FULL PAPER WWW.POLYMERPHYSICS.ORGJOURNAL OF

POLYMER SCIENCE

1706 JOURNAL OF POLYMER SCIENCE, PART B: POLYMER PHYSICS 2013, 51, 1705–1718

deuterated analog 3b (see Scheme 1), and the correspondinghomopolymers PFM and F8BT, The synthesis of 3a and 3bare described in detail in the article by Tang et al. The poly-mers were dissolved in toluene (Aldrich, HPLC) at concentra-tions of 5, 10, 20, and 30 mg cm23. This allowed a series offilm thicknesses to be made. The solutions were heated on ahotplate at 50 �C for 30 min for complete dissolution andthen allowed to cool to room temperature. The solutionswere filtered prior to spin coating (0.45 lm, Millipore,PTFE). The filtered solutions were spin-coated directly ontooxygen plasma-cleaned silicon wafers for AFM and specularNR measurements. The thickness of the active layer was con-trolled by the solution concentration and spin speed. Thelayer thicknesses of the spin-coated films were characterizedusing a J. A. Woolam M2000V spectroscopic ellipsometer(370–1000 nm) with the data fitted using a Cauchy model(in the nonabsorbing region). It is possible to use other tech-niques other than NR to study the vertical profile such as:near-edge X-ray absorption fine structure,23 ellipsometry,24

and dynamic secondary mass ion spectroscopy (DSIMS).25

However, NR in this instance is able to provide excellentneutron contrast between the block copolymer subunits.

To understand the annealing effects on the surface and inter-face structure, thermal annealing studies were carried out.The samples were placed in a vacuum oven. Tapping modeAFM (NanoScope III, Dimension 3100, Digital Instrument,Santa Barbara) was carried out with commercially availabletapping mode tips with a resonance around 275 kHz (Olym-pus). The scanning area was between 10 3 10 and 0.25 3

0.25 lm2. NR measurements reveal the scattering lengthdensity as a function of distance from a substrate. The scat-tering length is a nuclear property, and is directly analogousto electron density for X-ray experiments. The significant dif-ference in neutron scattering length density between hydro-gen and deuterium allows for isotopic substitution and

contrast between the components of the block copolymer.The data presented in this study were obtained using theSURF reflectometer at the ISIS-pulsed neutron source at theRutherford Appleton Laboratory (Oxfordshire, UK). A com-prehensive description of the technique can be found else-where.26 The SURF reflectometer uses a white beam ofneutrons with time-of-flight detection. Three different inci-dent angles were used to obtain a full reflectivity curve (thereflectivity as a function of momentum transfer perpendicu-lar to the sample plane Qz). We used a dq/q resolution of�2.8% at SURF. The NR data were analyzed using our owncustom written routines. The fitting procedure was to fit asingle layer for each sample and depending on the quality ofthe fit another layer was added until the fit was acceptable.The goodness-of-fit was assessed using the normalized v2 forthe fit. When additional layers were introduced, the twocompeting models were compared by an N-sigma analysis,which takes into account the true number of degrees of free-dom.27 Absorption and photoluminescence (PL) spectra weremeasured at room temperature with a UV-visible spectropho-tometer (Shimadzu, UV-1601) and a spectrofluorimeter(Horiba JobinYvon, Fluorolog-3), respectively.

PA allows us to study the subgap-excited states in conju-gated polymers. For PA measurements the sample wasexcited by either the 488-nm line or the 363-nm line of anargon ion laser, the excitation is modulated using squarewave modulation at 135 Hz using an opto-acustic modulatorfor the 488-nm radiation or an optical chopper for the 363-nm radiation. A 100-W tungsten lamp is focused onto thesample to spatially coincide with the laser excitation spot.The transmitted white light probe is collected via a parabolicmirror and reimaged onto the slits of a Spex monochromatorthe intensity of the collected white light is measured using asilicon diode for wavelengths from 400 to 1000 nm and anInSb detector for 800–2500 nm. The absorption caused by

SCHEME 1 Outline of the synthesis of the PFM-b-F8BT-b-PFM triblock copolymer 3a and the corresponding deuterated analog 3b.

JOURNAL OFPOLYMER SCIENCE WWW.POLYMERPHYSICS.ORG FULL PAPER

WWW.MATERIALSVIEWS.COM JOURNAL OF POLYMER SCIENCE, PART B: POLYMER PHYSICS 2013, 51, 1705–1718 1707

excited states within the band gap was recorded using aStanford Research Systems SRS 830 lockin amplifier set tothe same frequency as the laser modulation.

For excited state lifetime measurements the samples wereexcited with the frequency-doubled output of a coherentTi:Saph laser at either 380 nm, in order to excite PFM chro-mophores or 440 nm to excite the F8BT chromophores. Thepulse duration of the laser was significantly shorter than theresolution of the detection system. The PL was collected andfocused onto an Id Quantique ID100 avalanche photodiodeand time-correlated single photon counting (TCSPC) was car-ried out using a Becker and Hickle 830 photon-countingcard with a temporal resolution of 10 ps. Using band passfilters, 530–570 nm for F8BT and 430–470 nm for PFM thePL signals for each chromophore were separated. The rawsignal (presented in Fig. 6) can be reconvoluted using theprompt (�0.25 ns) to obtain a real lifetime. In theory, recon-volution can return accurate lifetimes that are up to 10times shorter than the prompt. The lifetime we obtained iswell below this upper limit. The reconvolution is performedusing software supplied with the Horiba TCSPC machine.28,29

Device FabricationITO-coated glass (Kintec, 15 X/sq) was cleaned by standingin a stirred solution of 5% (v/v) Deconex 12PA detergent at90 �C for 20 min. Each ITO substrate was sonicated sequen-tially for 10 min in distilled water, acetone, and iso-propanol.The substrates were then exposed to a final UV-ozone clean(at room temperature) for 10 min. The PEDOT/PSS (HCStarck, Baytron P AI 4083) was filtered (0.2-lm RC filter)and deposited by spin coating at 5000 rpm for 60 s to give a38-nm layer. The PEDOT/PSS layer was then annealed on ahotplate in a glovebox at 145 �C for 10 min, to remove anyresidual moisture in the layer. The polymer was dissolved inchlorobenzene (Aldrich, anhydrous) in vials with stirring.The solutions were then combined, filtered, and depositedonto the PEDOT/PSS layer by spin coating. Spin speeds wereoptimized and film thicknesses were measured for eachblend. The films were then annealed on a hotplate in a glo-vebox at 140 �C (as measured by a surface thermometer) for10 min. The devices were transferred (without exposure toair) to a vacuum evaporator in an adjacent glovebox. A layerof Ca (20 nm) and then Al (100 nm) was deposited by ther-mal evaporation at pressures below 2 3 1026 mbar. A con-nection point for the ITO electrode was made by manuallyscratching off a small area of the polymer layers. A smallamount of silver paint (Silver Print II, GC electronics, Partno.: 22–023) was then deposited onto all of the connectionpoints, both ITO and Al. The completed devices were thenencapsulated with glass and a UV-cured epoxy (Lens Bondtype J-91) by exposing to 254-nm UV-light inside a glovebox(H2O and O2 levels both <1 ppm) for 10 min. The encapsu-lated devices were then removed from the glovebox andtested in air within 1 h. Electrical connections were madeusing alligator clips. The cells were tested with an Oriel solarsimulator fitted with a 1000-W Xe lamp filtered to give anoutput of 100 mW/cm2 at AM 1.5. The lamp was calibrated

using a standard, filtered Si cell from Peccell Limited. Priorto analysis, the output of the lamp was adjusted to give a JSCof 0.605 mA with the standard device. The devices weretested using a Keithley 2400 Sourcemeter controlled by Lab-view Software. The incident photon collection efficiency(IPCE) data were collected using an Oriel 150-W Xe lampcoupled to a monochromator and an optical fiber. The outputof the optical fiber was focused to give a beam that was con-tained within the area of the device. The IPCE was calibratedwith a standard, unfiltered Si cell.

Thin-Film MorphologyThe spin-coated triblock copolymer thin films show essen-tially featureless and smooth surfaces with a root meansquare (RMS) roughness of 0.5 nm (as shown in Fig. 1, fur-ther images are available in the Supporting Information Fig.SI2). After thermal annealing of the layer, there is phase sep-aration of the two blocks of the copolymer. The diameter ofeach phase is around 10 nanometers with an RMS roughnessof 1.6 nm whereas the annealed blend films have roughnessvalues of 4 nm. The phase separation in the thermallyannealed copolymer is not perfectly ordered, with the small-angle neutron scattering (SANS) data showing weak orderingwith a domain size of 20 nm (Supporting Information Fig.SI2). As regard the phase separation mechanism, for this sys-tem and conventional coil–coil block copolymer systems, theroute is mircophase separation, where the phase equilibriumis determined by the fraction of monomers in the chain andthe product of vN (N is the degree of polymerization and vis the Flory–Huggins interaction parameter that quantifiesthe effective monomer interaction for A and B). A detailedand thorough treatment of which can be found elsewhere.30

For a polymer–polymer blend system enhancing phase sepa-ration normally refers to coalescence and growth in purity ofthe respective polymer domains, these can grow in size tothe micron lengthscale. However, for the case of blockcopolymers, there is an inhibition of this macrophase separa-tion and growth process due to the covalent linkage of therespective chains, meaning that there will be more interfacebetween the two constituents of a phase-separated blockcopolymer. Enhanced interconnectivity will also act in favorof charge percolation through the film and so is promisingfor device performance.

The morphology of the pristine and thermal annealed filmswas thickness independent for films with thicknesses from12 to 150 nm; the morphology of the films with differentthickness is shown in the Supporting Information.

To enable a crude depth profiling through the film, an oxy-gen plasma chamber was used to etch the copolymer and soprofile the in-plane morphology as a function of depth, thetwo polyfluorenes are not expected to be selectively etchedby the O2 plasma due to their similar chemical composition.The thickness of the etched layer can be controlled by thetime in the plasma (the thickness was monitored usingellipsometry), as it is a linear etching process.31 Figure 1(c)shows the morphology of the copolymer film after etching

FULL PAPER WWW.POLYMERPHYSICS.ORGJOURNAL OF

POLYMER SCIENCE

1708 JOURNAL OF POLYMER SCIENCE, PART B: POLYMER PHYSICS 2013, 51, 1705–1718

around 20 nm of the layer. The same or similar nanoscalestructure can be seen with a comparable lengthscale as inFigure 1(b), which means that the nanoscale fine phase sepa-ration exists in the bulk of the film and is not confined justto the surface of the film. AFM phase images of the copoly-mer film with an extended annealing step are shown in theSupporting Information Figures SI3 and SI4. There seems to

be little enhancement in the degree of phase separation forthe thermally annealed sample after 2 days when comparedwith heating the film after 1 h. The blend films (1:1 PFMF8BT blend is thermally annealed; Supporting InformationFig. SI6) show much larger surface features after thermalannealing with characteristic depressions in the film surfaceof �200 nm in size. This clear phase separation morphology

FIGURE 1 The surface morphology of the spin coated films of the conjugated triblock copolymer PFM-b-F8BT-b-PFM 3a, figures

shown on the left are height images and the corresponding phase images are shown on the right. (a) The pristine spin coated thin

film; (b) thermally annealed film at a temperature of 150 �C in a vacuum oven; (c) plasma etched (through the copolymer layer)

thermally annealed film. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

JOURNAL OFPOLYMER SCIENCE WWW.POLYMERPHYSICS.ORG FULL PAPER

WWW.MATERIALSVIEWS.COM JOURNAL OF POLYMER SCIENCE, PART B: POLYMER PHYSICS 2013, 51, 1705–1718 1709

has been seen in the past for F8BT:PFB blends.32 Previousstudies of F8BT:PFB blends were spin coated from p-xylene(boiling point 138 �C) whereas ours were spincoated fromtoluene (boiling point 110 �C), which is a much faster dryingsolvent. This explains the difference in the progression inphase separation between previous F8BT:PFB and ourF8BT:PFM blends. The molecular weight may well play arole in altering the dynamics of this process, along with spincoating speed and solution concentration.

Vertical SegregationSpecular NR was used to quantity the surface and interfacestructure in thin films of deuterated PFM-b-F8BT-b-PFM 3b.Specular neutron reflectivity reveals the scattering length den-sity as a function of distance from a substrate. Due to the iso-tope dependence of the neutron scattering length density thematerial distribution is obtained from the scattering lengthdensity profile, which means the component distribution per-pendicular to the substrate can be measured. The NR datawere analyzed using previously developed routines.33,34

Where a stack of thin recursive layers is used in the model inwhich each layer is characterized by a thickness, roughnessand scattering length density (after the scheme of N�evot andCroce).35 The measured reflectivity data for the pristine asspin cast sample and after thermally annealing step are shownin Figure 2(a) along with the associated fits to the data.

The spin-coated films have a homogeneous component dis-tribution as shown in Figure 2(b), which is what we wouldexpect for a polymer film after spin coating, as this is a non-equilibrium process and phase separation of the componentsis not possible on these short timescales (�20–40 s). Wetried numerous annealing temperatures spanning upto andincluding 200 �C. The vertical morphology from the NRmeasurements showed no change between the differentannealing temperatures. Indeed, we would expect differencesin chain mobility as a function of annealing temperature;however, we were interested in annealing above the glasstransition of F8BT, previously shown to be 135–140 �C inthe bulk, which will be also reduced by the thin film geome-try. The Tg of PFM is also in this region.36 Measurements onthin films of a similar triarylamine of TFB give a Tg from120–140 �C depending on film thickness.37 The small peakin the unannealed samples is due to the native silicon oxide.After thermal annealing however, there is a common featurein the scattering density profiles of an increase in scatteringlength density near the silicon substrate. As the octyl chainsof the PFM blocks in the triblock polymer 3b are deuteratedand so have a larger neutron scattering length density, theresults reveal that PFM has migrated to the substrate inter-face. This behavior was also seen when a PEDOT:PSS layerwas included to replicate the actual device system (Support-ing Information Fig. SI5). It is well documented that whenthe electron donor P3HT is enriched at the anode surfacethe performance is improved in the widely studied bulk het-erojunction polymer solar cell blend of P3HT and PCBM.8,38

Therefore, it is reasonable to suppose that the electrondonor PFM aggregation at the interface of the substrate

would benefit hole extraction at the anode and in doing soimprove the overall performance of the device. This is con-sistent with our preliminary device data, which shows ther-mal annealing dramatically improve the performance of thethermally annealed solar cells devices. It should be noted

FIGURE 2 Neutron reflectivity data and curve fits for thin films

of the spin-coated copolymer 3b before (a) and after (c) ther-

mal annealing at 150 �C; the scattering length density profiles

obtained from the best fits are shown before (b) and after (d)

thermal annealing.

FULL PAPER WWW.POLYMERPHYSICS.ORGJOURNAL OF

POLYMER SCIENCE

1710 JOURNAL OF POLYMER SCIENCE, PART B: POLYMER PHYSICS 2013, 51, 1705–1718

that Friend and coworkers have extensively studied thephase separation of PFB/F8BT and TFB/F8BT in thin filmsused in OLED or photovoltaic devices.39,40 They found thatthe lateral phase separation was on the micron scale and awetted PFB layer exists at the bottom interface, while theF8BT was capped by PFM at the top surface. The phase sep-aration of polymer blends in thin films is ultimately drivenby the different solubility’s of individual polymers in the sol-vent used and their interaction parameters with one another.So the morphology of the polymer blends is stronglydependent on the solvent and preparation route, that is, spincoating or doctorblading. The NR blend data for the as castfilm is shown in the Supporting Information Figure SI7 andagrees with these findings; as the layer has near pure PFMregion at the silicon surface, an intermixed F8BT regioncapped by a PFM layer.

The morphology of the annealed block copolymer filmsshowed similar features regardless of the solvent used, forexample, toluene, chloroform, chlorobenzene, and p-xylene.The composition of the layer after thermal annealing seemsto be unaffected by the solvent used to make the initial film,this is important as it rules out the need for optimization ofthe solvent processing conditions.

GIWAXS and DSCThe differential scanning calorimetry (DSC) trace in Figure3(c) shows the melting behavior of the F8BT homopolymerand PFM-b-F8BT-b-PFM triblock copolymer 3a with a melt-ing peak at around 150 �C for both. The PFM-b-F8BT-b-PFMtriblock copolymer melting behavior is dominated by thecentral F8BT block of the copolymer. This melting tempera-ture is consistent with other measurements of F8BT.41 Graz-ing incidence wide angle X-ray (GIWAXS) was measured forthe thermally annealed and pristine as spin-coated samples.In the thermally annealed sample, a broad diffraction ringcan be seen in 3(a), which may be due to the p–p stackingof the F8BT block. The in plane and out of plane peak datawere fitted to a Lorentzian to extract the peak position, giv-ing in-plane (q51.34 Å21, d54.68 Å) and out of-plane(q51.37 Å21, d54.58 Å). Taking into account that the F8BTblock is the central block in 3a, the structure from the F8BTblock in the center of the copolymer seems to maintain asimilar spacing as has been seen for pure F8BT (4.18 Å),41

but with some degree of swelling possibly due to the PFMblock altering packing distance seen in pure F8BT. It hasbeen postulated that this packing distance is essential forgood devices based on semiconducting polymers.

On the basis of the specular NR data on 3b, and the AFMand GIWAXS results on 3a, it is possible to present a sche-matic model of PFM-F8BT-PFM distribution and orientationin thin films in which the electron donor PFM blocks accu-mulate at the substrate interface, with a thin depletion layerof PFM-rich region; above this is an ordered phase separatedcylinder/rod morphology in the bulk of the film, with afinely phase separated structure having a 10–20-nm lengthscale.

Optical PropertiesThin films of a 2:1 blend of PFM and F8BT and the triblockcopolymer 3a were made by spin casting from solutions of10 mg cm23 of chloroform at 2000 rpm. The resulting blendthin films had thickness between 50 and 60 nm (as meas-ured using ellipsometry) with the exception of the triblockthat had a thickness of 100 nm. The absorption spectra ofthe films were measured and are given in Figure 4. Theabsorption spectra of the PFM and F8BT shown in Figure 4are similar to those reported in other studies.42,43 Theabsorption spectrum of the 2:1 blend of PFM:F8BT and the1:1 triblock copolymer 3a can both be recreated by a simplelinear superposition of the PFM and F8BT absorptionspectra.

The normalized PL spectrum of the thin films studied inabsorption is shown in Figure 5. The films of PFM, the blendand the triblock copolymer 3a were all excited at 375 nmwith the F8BT sample being excited at 440 nm. The PL spec-tra of the blend and triblock copolymer 3a are similar ifexcited at 375 or 440 nm.

The original unnormalized spectrum of the blend of PFMand F8BT is 75 times less intense than that of the pureF8BT sample; as the sample thickness are similar it is fair toassume that the quantum yield of the blend is greatlyreduced, most likely due to charge separation. The blenddoes show some weak emission from the PFM. Annealing ofthe blend reduces but does not remove this feature. The tri-block copolymer 3a when excited at 375 nm shows no emis-sion from the PFM block but does show emission from theF8BT chromophore. This is an indication of greater mixingand that the triblock copolymer has a greater interfacial areathan the F8BT blend. Both the blend and triblock copolymershow very similar PL spectra with a strong red-shifted fea-ture similar to that seen in blends of F8BT with PFB. Thisfeature has previously been attributed to exciplex emission.44

It has been shown that the exciplex emission requires ageminate polaron pair to be formed prior to the formation ofthe exciplex. The formation of long-lived exciplex states isdetrimental to the device performance in a photovoltaicdevice.

To investigate the reduced PL efficiency of the blend and tri-block copolymer, the excited state lifetimes of thin filmswere investigated using TCSPC.

Figure 6(a) shows time-gated PL data for a film of the tri-block copolymer excited at 467 nm and averaged from either0–10 or 30–50 ns after excitation. At early time scales, theemission profile is made up predominantly of short-lived flu-orescence from F8BT singlet excitons, peaking at 540 nm. Atlonger timescales, primary excitons in F8BT have fullydecayed and the emission must be due to longer lived spe-cies. The spectra averaged between 30 and 50 ns peaks at645 nm, a feature that is attributed to exciplex emission.Furthermore, there is an additional shoulder seen at 540nm. In strong agreement with work done by Morteani et al.on binary blends of F8BT and PFB, this additional feature is

JOURNAL OFPOLYMER SCIENCE WWW.POLYMERPHYSICS.ORG FULL PAPER

WWW.MATERIALSVIEWS.COM JOURNAL OF POLYMER SCIENCE, PART B: POLYMER PHYSICS 2013, 51, 1705–1718 1711

the result of thermal regeneration (back-transfer) of second-ary bulk F8BT excitons from the exciplex state.45

Figure 6(b) shows transient PL studies on homopolymerF8BT and PFM-F8BT-PFM block copolymer performed usingTCSPC after excitation at 467 nm. Emission was selectivelymeasured at 540 and 645 nm. The excited state lifetimes ofpure PFM and F8BT were measured at 0.12 and 0.5 ns,respectively. For comparison films of F8BT typically havelifetimes between 1 and 2 ns,33,46 whereas PFB and F8,being similar polymers to PFM have thin film excited statelifetimes of 0.2 ns. In thin films of the triblock copolymer 3a,

the emission from the PFM block is quenched below theresolution of our TCSPC system, while the F8BT block has areduced lifetime of 0.12 ns. The complete quenching of thePFM in films of the triblock is as expected as the PL spectraof triblock films show no emission from the PFM chromo-phore. The 0.12-ns lifetime of the F8BT chromophore in thetriblock copolymer is independent of the excitation wave-length (i.e., it is the same for excitation at 380 and 440 nm)and the reduced lifetime is due to efficient, but not complete,quenching of the F8BT. The PL at 540 nm is predominantlyattributed to excited state relaxation in bulk F8BT. Emissionat 645 nm is mainly attributed to exciplex emissive decay.

FIGURE 3 (a) Two-dimensional GIWAXS image of PFM-b-F8BT-b-PFM 3a along with (b) one-dimensional integrations along the in

plane and out of plane directions. (c) DSC traces of pure F8BT and PFM-b-F8BT-b-PFM 3a. [Color figure can be viewed in the

online issue, which is available at wileyonlinelibrary.com.]

FULL PAPER WWW.POLYMERPHYSICS.ORGJOURNAL OF

POLYMER SCIENCE

1712 JOURNAL OF POLYMER SCIENCE, PART B: POLYMER PHYSICS 2013, 51, 1705–1718

The exciplex decay is longer lived than F8BT excited statefluorescence and is found to have monoexponential decaywith time constant 266 2 ns in the block copolymer (by fit-ting to a single exponential at times after singlet excitonemission has decayed). This value is in good agreement withthe literature value found for the exciplex lifetime in a 1:1blend of F8BT:PFB spin cast from chloroform of 28 ns,although it is noted that this value is highly dependent onfilm morphology.45,47 In addition, confirming our interpreta-tion of long-lived emission at shorter wavelengths as due tosecondary exciton decay, the decay constant for the short-

lived species at 540 nm is found to be 206 3 ns. This is afaster decay, as the emission at this wavelength is a blend ofexciplex and regenerated F8BT exciton emission; the fasterdecaying exciton brings the decay time down.

Photo-Induced Absorption SpectroscopyWe have used steady-state PA spectroscopy48 to study thevarious long-lived photoexcited states in thin films of the tri-block copolymer 3a and its constituent components. The PAspectra of films of PFM, a 2:1 PFM F8BT blend and the tri-block copolymer 3a are shown in Figure 7(a–c), respectively.

All spectra have been corrected for excitation density n0,estimated from the depth averaged pump fluence N0, as

FIGURE 4 The absorption spectra of spin-coated thin films of

pure PFM, F8BT, a 2:1 blend of PFM:F8BT, and the triblock

copolymer 3a.

FIGURE 5 The normalized photoluminescence spectra of thin

films of PFM (black) excited at 375 nm, F8BT (red) excited at

465 nm, a 2:1 PFM:F8BT blend (blue) excited at 375 nm, and

the triblock copolymer 3a (green) excited at 375 nm. Excitation

of the triblock copolymer 3a at 465 nm results in a very similar

spectrum to that obtained by excitation of the triblock copoly-

mer 3a at 375 nm. [Color figure can be viewed in the online

issue, which is available at wileyonlinelibrary.com.]

FIGURE 6 (a) shows the results of normalized, gated PL studies

on the triblock copolymer film. The figure shows an average of

the time-resolved PL between 0–10 ns (black squares), 30–50

ns (gray circles), and steady state (gray line). (b) shows the

results of time correlated single photon counting on films of

F8BT and triblock copolymer at different probe wavelengths:

F8BT probed at 645 nm (black solid squares), block copolymer

probed at 645 nm (gray open circles) and block copolymer

probed at 540 nm (open gray stars). Single photon counting in

each sample was performed for the same length of time. Exci-

tation in all cases was at 467 nm with a pulse width of �250 ps

(FWHM).

JOURNAL OFPOLYMER SCIENCE WWW.POLYMERPHYSICS.ORG FULL PAPER

WWW.MATERIALSVIEWS.COM JOURNAL OF POLYMER SCIENCE, PART B: POLYMER PHYSICS 2013, 51, 1705–1718 1713

n05N0

d5

1

d:P

htA(1)

where d is the penetration depth calculated from the laserattenuation coefficient 1/aL, P is the laser power, A is thecross-sectional area of the laser-beam, and hm is the pumpphoton energy. In part (a), we detect a broad and largely fea-tureless spectrum associated with PFM photoinducedabsorption. Having plotted both the in-phase and out-phasesignals, we find that their ratio is constant across the wave-lengths probed. As the ratio of out-phase to in-phase signalsis directly related to the excited state lifetime, s, of the PAfeature by Y/X5 tanh 5 xs we assert that the PFM spectrumis composed of a contribution from a single excited statepopulation of a unique lifetime which we tentatively assignto the T1–T2 transition (this feature peaks close to 1.25 eV)based on previous PIA studies examining the triarylamine

PFB (1.35 eV) and note that this energy range is typical forconjugated polymers.49 We verify this assignment by repeat-ing the measurement at a pump modulation frequency of 1.6kHz and find that although the PA signal shifts further out ofphase, consistent with driving the system away from steady-state conditions where xs >1, overall, the phase is againconstant as a function of probe energy implying that indeedthe spectrum results from a single PA feature.

The interpretation is rather more straightforward in Figure7(b,c) as the PA spectra of both the triblock copolymer andblend show strong correspondence with published PA spec-tra of pure F8BT and F6BT.49,50 This result implies that, byselectively exciting the F8BT moiety in both the blend andtriblock copolymer with 488-nm laser excitation, we see theformation of a strong triplet band at 1.55 eV and two some-what less intense polaron bands all of which reside on theF8BT. However, the blend shows some PA signal between 0.5and 1 eV that is not present in triblock 3a or pure F8BTspectra and we attribute this contribution to the PFM back-ground. Also plotted in Figure 7(c) is the PA spectrum of thetriblock copolymer under selective excitation of the PFMcomponent at a pump wavelength of 363 nm. It again showsa strong resemblance to pure F8BT. After correction for exci-tation density, this signal appears comparatively weak sug-gesting that singlet–triplet formation is more efficient fromthe F8BT component of the triblock copolymer.

Photo-Induced Triplet YieldsWe can quantify the triplet yields in the blend and triblockusing the fact that the absorption cross section of singletsand triplets are believed to be comparable.51 For a randomlyorientated bulk polymer, the cross section r, wherer 5 amax/N and amax is the maximum absorption coefficient,N is the density of ground states or monomeric segmentsgiven by N5 qNA/mW where q is the density, NA is Avaga-dro’s number, and mW is the molecular weight of the repeatunit. We can thus estimate the triplet absorption cross sec-tion (rt) using

rt �mWamax

qNA(2)

Using values of mW 5 594 g mol21 and q � 1 g cm23 for thetriblock and a value of amax 5 1.9 3 105 cm21 measured inpure F8BT, we find rt 5 1.9 3 10216 cm2; a value that isvery similar to that obtained through density functionaltheory calculations on F8BT.49 The triblock triplet densitycan now be calculated using

nt5DTT

1

drt

� �(3)

with the modulus of DT/T at 1.55 eV being 6.6 3 1024.Using a value of aL as 5.1 3 104 cm21, we estimate that thepenetration depth of the laser as �200 nm. Combining thetriplet cross section, differential transmission and penetra-tion depth, we deduce a triplet density in the triblock thin-

FIGURE 7 PA spectra of various polymer thin films at 77 K and

modulated at 135 Hz. In phase and out-phase, data are plotted

as solid and dashed lines, respectively. Part (a): an unblended

PFM thin film optically excited at 363 nm modulated at 135 Hz

(black) and 1.6 kHz (gray); Part (b): a 2:1 PFM F8BT-blend film

excited at 488 nm; Part (c): a triblock thin film excited at 488

nm (black) and at 363 nm (gray) corrected for excitation

density.

FULL PAPER WWW.POLYMERPHYSICS.ORGJOURNAL OF

POLYMER SCIENCE

1714 JOURNAL OF POLYMER SCIENCE, PART B: POLYMER PHYSICS 2013, 51, 1705–1718

films under our measurement conditions to be 1.7 3 1017

cm23s21.

To better understand the recombination mechanism of theexcited species in the material systems studied, both thedependence of the PA signal on pump intensity and theapplied modulation frequency have been studied as shownin Figure 8(b). We find that both the triblock 3a and the 2:1F8BT/PFM blend are characterized by dispersive dynamics52

resulting from a distribution of excited state lifetimes. Typi-cally, rather than the NI � x22 form expected for the in-phase PA signal in the case of a uniform recombination rate,these systems display a gentle, high-frequency roll-off of theform NI � x2a where 0< a< 1, with lower values of aimplying a more disperse distribution of lifetimes. For thetriblock 3a, we determine a triplet lifetime s 5 0.64ls6 10% with a 5 0.88. Having fit to the modulation data,we can calculate the triplet density that we might expectunder steady state conditions. In particular, we estimate byextrapolation from the PA measurements (performed at achopping frequency of 135 Hz), that the triplet densityexpected at quasi steady-state values will show little changesince for this relatively short lifetime, we are already operat-ing near to steady state conditions, where xs< 1, thus nt

0 �ntf5 135 Hz5 1.7 3 1017 cm23.

We can relate the steady-state triplet density to a generationyield, gt, by solving the rate equation49:

dntdt

5gtaLN012

11cosxtð Þ2 nts

(4)

Note, that this analysis ignores any saturation effects andassumes that the triplets undergo monomolecular recombi-nation with a rate given by nt/s. We can calculate N0 as out-

lined in equation 1. The depth averaged excitation density isthus N0aL, a value that is fixed at 1.6 3 1023 cm23s21 in ourPA experiments. In the steady state, eq 4 can be solved toexpress the triplet yield as

gt5nt

aLN0s(5)

Using our measured parameters, we determine a singlet-exciton to triplet yield of gt 5 2%. Although this value issmaller than expected from the efficient exciton quenchingthat we observe in the triblock, we associate this yield witha long-lived population of trapped photoexcitations that wediscuss later. We note that this value is relatively large incomparison with other studies by List et al.53 where a yieldof 0.08% was measured in a fluorescent conjugated ladder-type polymer; however, in such materials, the direct dissocia-tion of singlet excitons into triplets excitons is anticipated tobe a relatively inefficient process. Perhaps also a planarstructure is unfavorable to triplet formation.

To calculate the triplet yield in the 1:2 F8BT/PFM blend, weassume that the same values for rt, d, aL, and N0 as beforeand from Figure 6(b), we know that DT/T5 5.8 3 1025.Therefore, using eq 3, we estimate that nt 5 1.5 3 1016

cm23s21. We again investigate the dependence of the PA sig-nal on the modulation frequency and have successfully fittedthe data to a dispersive model with parameters s 5 107 lsand a 5 0.74 as shown in 10(a). This lifetime is again suffi-ciently short to assume that at a chop frequency of 135 Hz,we are close to steady state conditions and that nt

0 �nt

f5 135 Hz 5 1.5 3 1016 cm23s21. Using these parameters,we determine a singlet–triplet yield for the F8BT/PFM blendfrom eq 5 of gt 5 0.09%.

Device DataWhen we compare the PCEs of the blend (0.05%) and thetriblock (0.14%), we find a factor of 3 difference betweendevices made of the two materials, with the triblock beingthe more efficient material. It is difficult to compare energyconversion efficiencies arising from the use of blends com-pared with a block copolymer owing to the differences inthe size domains of the blend morphology and the blockcopolymer morphology. One of the major advantages of usinga block copolymer is that we can design the nanostructureand the morphology of the equilibrium structure by tuningthe size of the block copolymer blocks and their respectiveratios. Ideally, we compared blend and triblock when theyboth have similar sized phase separation. The PCE of theblend remained unchanged at 0.05% before and afterannealing the sample, while for the block copolymer devicethe PCE value increased from a value of 0.06–0.14% uponannealing, an improvement factor of 2.5 (see SupportingInformation for J–V curves) due to evolution of the nanoscalephase separation. What is surprising is that the differencebetween the two systems is not greater than it is, as it isobvious that the block copolymer has much more interfacebetween the two components than exists in the blend. Thisis the subject of further work.

FIGURE 8 The frequency dependence of the PA signal for both

the homopolymer blend (part a) and the triblock copolymer 3a

(part b) with frequency-dependent PA corresponding to the 1.5

eV T1 feature are shown with modeled data. The samples were

held at 77 K and excited at 488 nm with fluence 1.8 3 1018

cm22s21. The figure gives the in-phase signal (open circles)

and the quadrature signal (open squares). The data are fitted

to a dispersive recombination model for both in-phase (solid

line) and quadrature signal (broken line).

JOURNAL OFPOLYMER SCIENCE WWW.POLYMERPHYSICS.ORG FULL PAPER

WWW.MATERIALSVIEWS.COM JOURNAL OF POLYMER SCIENCE, PART B: POLYMER PHYSICS 2013, 51, 1705–1718 1715

When we examine the possible reasons why this materialssystem provides low-efficiency devices it is worth examiningthat the absorption spectra of F8BT and PFM are far fromideally matched to the AM 1.5 solar spectrum and this is inpart responsible for the low efficiency of the devices (seeIPCE Supporting Information SI8). Both their absorptionspectra tail off dramatically above 500 nm, while the major-ity of the AM 1.5 solar spectrum is above 500 nm. Therefore,in a practical device, a lower HOMO–LUMO gap polymer isrequired to absorb more of the solar flux.

The material composition change on thermal annealing thetriblock shows an enriched region of PFM near the anodesurface, which would benefit device performance. The sameeffect is also true for the triblock layer when it is thermallyannealed on a PEDOT:PSS layer.

Another possible explanation for the poor performance ofboth the blend devices and the triblock devices could be thelow electron mobility of F8BT. Studies have shown that F8BThas an electron mobility an order of magnitude lower thanPCBM (measured using time of flight.54,55 Optimized polymerblend devices of P3HT and F8BT achieved similar efficienciesranging from 0.02 to 0.13%.

DISCUSSION

It is apparent that triplet formation is roughly 20 timesmore efficient in the triblock copolymer when comparedagainst a representative bulk heterojunction blend of theconstituents: PFM and F8BT. Other studies of polyfluorenephotovoltaic blends have shown charge separation to dependstrongly on donor–acceptor domain size56,57 and in somecases domains larger than the exciton diffusion length some-what confusingly result in the most efficient devices.39 AFMand SANS data for the triblock copolymer indicate that phaseseparation exists after thermal annealing and is on the nano-scale (�20 nm) and commensurate with the exciton diffu-sion length. Furthermore, we see evidence for exciplex statesin cw-PL measurements performed on the triblock copoly-mer that further support the notion that coulombicallybound geminate charge pairs with significant wavefunctionoverlap exist at high concentration. Triplet formation isassumed to proceed via intersystem crossing from theseelectron–hole pairs and has been shown in chloroform castPFB/F8BT thin films, where phase separation is also sup-pressed, to occur with yields as high as 75%.47 Furthermore,externally applied electric fields have been found to reducethe triplet formation rate by dissociating the charge pairbefore the intersystem crossing can take place.49 We believethat similar effects are likely to be the case in this study,since we too measure a high triplet yield. Such a large popu-lation of triplets represents a terminal loss mechanism andmay explain why device efficiencies measured in the triblockcopolymer are not as improved as at first might be thoughtwhen compared to the blend. Further work may aim toaddress the position of the triplet level relative to the inter-facial charge pair state by making the formation of tripletsenergetically unfavorable or perhaps look toward influencing

the morphology of block copolymers to bring about optimaldomain size and affect efficient geminate charge separation.

CONCLUSIONS

A number of techniques have been used to compare themorphology and optical properties of the conjugated triblockcopolymer 3a with blends of the corresponding homopoly-mers that reflect the constituent block components. Thesestudies have shown that the triblock copolymer 3a assumesa nanoscale morphology with a great deal of interface thatshould help promote efficient charge separation of excitonsupon photoexcitation.

The PA spectra for the blend and the triblock show that thesinglet to triplet formation is more efficient in the copolymerthan for the blend. It appears that the large degree of inter-face between the nanosized domains of F8BT and PFM cre-ates a fundamental hurdle in the overall efficiency of theseblock copolymer devices. The absorption of light createsshort-lived exciton states, which diffuse to the interfacebetween F8BT and PFM. Excitons can either be separatedinto an electron and a hole, which diffuse to the electrodes,or can also form a long-lived excpliex state. Previous workon PFB and F8BT blends with similar nanosized domains ofless than 10 nm demonstrated that the majority of photoex-cited states never become separate charges and insteadremain as tightly bound stationary charges at the heterojunc-tion interface.47 These states can then convert to tripletstates via intersystem crossing. Triplets are essentiallywasted states and will reduce the effective performance inan OPV device. Triplet exctions are able to quench singletexcitons and due to the long lifetime of triplets (ms), they areable to do this for a large number of singlets before relaxingback to the ground state. Minimizing intersystem crossingmust be a key consideration in the design of future blockcopolymer OPV materials. This can be accomplished by suita-ble chemical design utilizing lighter atoms that have smallerspin–orbit coupling, planarizing the chemical groups alsoreduces the spin–orbit coupling by reducing overlap of thewavefunctions.58

A strategy to overcome these tightly bound interfacial stateswould be to rapidly transport the charges away from theinterface via high electron and hole mobility domains. Natureuses a series of cascade reactions that prevents reversemechanisms reducing the overall efficiency; this complexityis currently far beyond current OPV systems and need notbe necessary.

Recent work using pump probe has demonstrated thatorganic semiconductors can form short-lived delocalizedband states and that organic materials which enable delocal-ized charge wave functions when coupled with molecularrigidity are the best candidates for improving charge separa-tion efficiencies.59

It has been suggested that block copolymer based OPV activelayers would have many advantages over the current bulk

FULL PAPER WWW.POLYMERPHYSICS.ORGJOURNAL OF

POLYMER SCIENCE

1716 JOURNAL OF POLYMER SCIENCE, PART B: POLYMER PHYSICS 2013, 51, 1705–1718

heterojunction polymer blend OPVs, although in practicethere are a number of shortcomings that need to beaddressed. The most industrially important being the easewith which the block copolymer can be ordered during proc-essing and how compatible this is with existing reel-to-reelprocessing technology. Long thermal annealing steps will notmake these materials viable commercial products.

ACKNOWLEDGMENTS

This work was made possible by support from the AustralianGovernment DIISR ISL grant (CG100059) International Consor-tium in Organic Solar Cells (ICOS), the Victorian Organic SolarCell Consortium (www.vicosc.unimelb.edu.au), the VictorianState Government Department of Business Innovation (Victo-rian Science Agenda) and the Department of Primary Industries(Energy Technology Innovation Strategy). The authorsacknowledge financial support from the Engineering and Physi-cal Science Research Council (EPSRC), A. J. Parnell was fundedby the EPSRC under grant EP/E046215/1. S. A. Haque acknowl-edges support from the Royal Society through award of a Uni-versity Research Fellowship. They also thank the ISIS Neutronfacility (UK) for time on the reflectometer SURF and the Dia-mond Light Source for X-raymeasurements and Nicholas Terrillfrom I22. Construction of LENS was supported in part by theNational Science Foundation grants DMR-0220560 and DMR-0320627, and the operation of LENS is supported by IndianaUniversity.

REFERENCES AND NOTES

1 Y. Liang, Z. Xu, J. Xia, S.-T. Tsai, Y. Wu, G. Li, C. Ray, L. Yu,

Adv. Mater. 2010, 22, E135–E138.

2 Z. He, C. Zhong, S. Su, M. Xu, H. Wu, Y. Cao, Nat. Photon.

2012, 6, 591–595.

3 P. E. Shaw, A. Ruseckas, I. D. W. Samuel, Adv. Mater. 2008,

20, 3516–3520.

4 E. Verploegen, C. E. Miller, M. F. Toney, Synchrot. Radiat.

News 2010, 23, 16–21.

5 T. Wang, A. D. F. Dunbar, P. A. Staniec, A. J. Pearson, P. E.

Hopkinson, J. E. Macdonald, S. Lilliu, C. Pizzey, N. J. Terrill, A.

M. Donald, A. J. Ryan, R. A. L. Jones, D. G. Lidzey, Soft Matter

2010, 6, 4128.

6 M. Y. Chiu, U. S. Jeng, C. H. Su, K. S. Liang, K. H. Wei, Adv.

Mater. 2008, 20, 2573–2578.

7 S. Swaraj, C. Wang, H. Yan, B. Watts, J. L€uning, C. R.

McNeill, H. Ade, Nano Lett. 2010, 10, 2863–2869.

8 A. J. Parnell, A. D. F. Dunbar, A. J. Pearson, P. A. Staniec, A.

J. C. Dennison, H. Hamamatsu, M. W. A. Skoda, D. G. Lidzey,

R. A. L. Jones, Adv. Mater. 2010, 22, 2444–2447.

9 J. W. Kiel, B. J. Kirby, C. F. Majkrzak, B. B. Maranville, M. E.

Mackay, Soft Matter 2010, 6, 641–646.

10 S. C. Veenstra, J. Loos, J. M. Kroon, Prog. Photovolt.: Res.

Appl. 2007, 15, 727.

11 Polyera. Available at: http://www.polyera.com/newsflash/pol-

yera-achieves-6-4-all-polymer-organic-solar-cells, 2013. Last

accessed September 23, 2013.

12 Organic Photovoltaics: Materials, Device Physics, and Manu-

facturing Technologies; Wiley-VCH, 2008.

13 C. R. McNeil, N. C. Greenham, Adv. Mater. 2009, 21, 3840–

3850.

14 H. Yan, B. A. Collins, E. Gann, C. Wang, H. Ade, C. R.

McNeill, ACS Nano 2012, 1, 677–688.

15 D. Mori, H. Benten, H. Ohkita, S. Ito, K. Miyake, ACS Appl.

Mater. Interfaces 2012, 4, 3325–3329.

16 F. S. Bates, G. H. Fredrickson, Phys. Today 1999, 52, 32.

17 R. A. Segalman, B. McCulloch, S. Kirmayer, J. J. Urban,

Macrmolecules 2009, 42, 9205–9216.

18 I. W. Hamley, The Physics of Block Copolymers; Oxford Uni-

versity Press, 1998.

19 P. D. Topham, A. J. Parnell, R. C. Hiorns, J. Polym. Sci.

Polym. Phys. 2011, 49, 1131–1156.

20 S. B. Darling, Energy Environ. Sci. 2009, 2, 1266–1273.

21 I. Botiz, S. B. Darling, Mater. Today 2010, 13, 42–51.

22 A. Cuendias, R. C. Hiorns, E. Cloutet, L. Vignau, H. Cramail,

Polym. Int. 2010, 59, 1452–1476.

23 H. Yan, C. Wang, A. Garcia, S. Swaraj, Z. Gu, C. R. McNeill,

T. Schuettfort, K. E. Sohn, E. J. Kramer, G. C. Bazan, T.

Nguyen, H. Ade, J. Appl. Phys. 2011, 110, 102220.

24 M. Campoy-Quiles, T. Ferenczi, T. Agostinelli, P. G.

Etchegoin, Y. Kim, T. D. Anthopoulos, P. N. Stavrinou, D. D. C.

Bradley, J. Nelson, Nat. Mater. 2008, 7, 158.

25 C. M. Bj€orstr€om, A. Bernasik, J. Rysz, A. Budkowski, S.

Nilsson, M. Svensson, M. R. Andersson, K. O. Magnusson, E.

Moons, J. Phys.: Condens. Matter 2005, 17, L529.

26 T. P. Russell, Phys. B 1996, 221, 267–283.

27 J. Ihringer, J. Appl. Crystallogr. 1995, 28, 618–619.

28 W. Becker, The bh TCSPC Handbook; Becker & Hickl GmbH,

Berlin, Germany. 2nd edition, 2012.

29 D. V. O’Connor, D. Phillips, Time-Correlated Single Photon

Counting; Academic Press, London, 1984.

30 L. Leibler, Macromolecules 1980, 13, 1602–1617.

31 G. Kokkoris, N. Vourdas, E. Gogolides, Plasma Process.

Polym. 2008, 5, 825–833.

32 C. R. McNeill, B. Watts, L. Thomsen, H. Ade, N. C.

Greenham, P. C. Dastoor, Macromolecules 2007, 40, 3263–3270.

33 S. J. Martin, R. A. L. Jones, M. Geoghegan, A. M. Higgins, I.

Grizzi, J. J. M. Halls, S. Kirchmeyer, R. M. Dalgliesh, Phys. Rev.

B 2005, 71, 81408.

34 A. D. F. Dunbar, P. Mokarian-Tabari, A. J. Parnell, S. J.

Martin, M. W. A. Skoda, R. A. L. Jones, Eur. Phys. J. E 2010,

31, 369–375.

35 L. N�evot, P. Croce, Rev. Phys. Appl. 1980, 15, 761.

36 M. Redecker, D. D. C. Bradley, M. Inbasekaran, W. W. Wu, E.

P. Woo, Adv. Mater. 1999, 11, 241–246.

37 D. Liu, R. Osuna Orozco, T. Wang, Phys. Rev. E 2013, 88, 022601.

38 J. W. Kingsley, A. Green, D. G. Lidzey, Proc. SPIE: Int. Soc.

Opt. Eng. 2009, 7416, 7416T.

39 C. R. McNeill, S. Westenhoff, C. Groves, R. H. Friend, N. C.

Greenham, J. Phys. Chem. C 2007, 111, 19153.

40 C. R. McNeill, B. Watts, S. Swaraj, H. Ade, L. Thomsen, W.

J. Belcher, P. C. Dastoor, Nanotechnology 2008, 19.

41 C. L. Donley, J. Zaumseil, J. W. Andreasen, M. M. Nielsen,

H. Sirringhaus, R. H. Friend, J. Kim, J. Am. Chem. Soc. 2005,

127, 12890–12899.

42 L. C. Paililis, D. G. Lidzey, M. Redecker, D. D. C. Bradley, M.

Inbasekaran, E. P. Woo, W. W. Wu, Synth. Met. 2001, 121,

1729–1730.

JOURNAL OFPOLYMER SCIENCE WWW.POLYMERPHYSICS.ORG FULL PAPER

WWW.MATERIALSVIEWS.COM JOURNAL OF POLYMER SCIENCE, PART B: POLYMER PHYSICS 2013, 51, 1705–1718 1717

43 M. A. Stevens, C. Silva, D. M. Russell, R. H. Friend, Phys.

Rev. B 2001, 63.

44 A. C. Morteani, P. Sreearunothai, L. M. Herz, R. H. Friend, C.

Silva, Phys. Rev. Lett. 2004, 92, 247402.

45 A. C. Morteani, R. H. Friend, C. Silva, J. Chem. Phys. 2005,

122, 244906.

46 A. J. Cadby, G. Khalil, A. M. Fox, D. G. Lidzey, J. Appl. Phys.

2008, 103, 093715.

47 S. Westenhoff, I. A. Howard, J. M. Hodgkiss, K. R. Kirov, H.

A. Bronstein, C. K. Williams, N. C. Greenham, R. H. Friend, J.

Am. Chem. Soc. 2008, 130, 13653–13658.

48 A. J. Cadby, P. A. Lane, H. Mellor, S. J. Martin, M. Grell, C.

Giebeler, D. D. C. Bradley, M. Wohlgenannt, C. An, Z. V.

Vardeny, Phys. Rev. B 2000, 62, 15604–15609.

49 T. A. Ford, I. Avilov, D. Beljonne, N. C. Greenham, Phys.

Rev. B 2005, 71, 125212.

50 M. Westerling, C. Vijila, R. €Osterbacka, H. Stubb, Phys. Rev.

B 2003, 67, 195201.

51 E. Zojer, J. Cornil, G. Leising, J. L. Br�edas, Phys. Rev. B

1999, 59, 7957–7968.

52 O. Epshtein, G. Nakhmanovich, Y. Eichen, E. Ehrenfreund,

Phys. Rev. B 2001, 63, 125206.

53 E. J. W. List, C.-H. Kim, A. K. Naik, U. Scherf, G.

Leising, W. Graupner, J. Shinar, Phys. Rev. B 2001, 64,

155204.

54 V. D. Mihailetchi, J. K. J. van Duren, P. W. M. Blom, J. C.

Hummelen, R. A. J. Janssen, J. M. Kroon, M. T. Rispens, W.

J. H. Verhees, M. M. Wienk, Adv. Funct. Mater. 2003, 13,

43–46.

55 Y. Kim, S. Cook, S. A. Choulis, J. Nelson, J. R. Durrant, D.

D. C. Bradley, Chem. Mater. 2004, 16, 4812–4818.

56 M. J. Harding, V.-E. Poplavskyy, V.-E. Choong, A. J.

Campbell, F. So, Org. Electron. 2008, 9, 183–190.

57 R. A. Marsh, C. R. McNeil, A. R. Abrusci, A. R. Campbell, R.

H. Friend, Nano Lett. 2008, 8, 1393–1398.

58 A. J. Cadby, C. Yang, S. Holdcroft, D. D. C. Bradley, P. A.

Lane, Adv. Mater. 2002, 14, 57–60.

59 A. A. Bakulin, A. Rao, V. G. Pavelyev, P. H. M. Loosdrecht,

M. S. Pshenichnikov, D. Niedzialek, J. Cornil, D. Beljonne, R. H.

Friend, Science 2012, 335, 1340–1344.

FULL PAPER WWW.POLYMERPHYSICS.ORGJOURNAL OF

POLYMER SCIENCE

1718 JOURNAL OF POLYMER SCIENCE, PART B: POLYMER PHYSICS 2013, 51, 1705–1718


Recommended