A molecular nematic liquid crystalline material for high-performance organic photovoltaics

Abstract

Solution-processed organic photovoltaic cells (OPVs) hold great promise to enable roll-to-roll printing of environmentally friendly, mechanically flexible and cost-effective photovoltaic devices. Nevertheless, many high-performing systems show best power conversion efficiencies (PCEs) with a thin active layer (thickness is ~100 nm) that is difficult to translate to roll-to-roll processing with high reproducibility. Here we report a new molecular donor, benzodithiophene terthiophene rhodanine (BTR), which exhibits good processability, nematic liquid crystalline behaviour and excellent optoelectronic properties. A maximum PCE of 9.3% is achieved under AM 1.5G solar irradiation, with fill factor reaching 77%, rarely achieved in solution-processed OPVs. Particularly promising is the fact that BTR-based devices with active layer thicknesses up to 400 nm can still afford high fill factor of ~70% and high PCE of ~8%. Together, the results suggest, with better device architectures for longer device lifetime, BTR is an ideal candidate for mass production of OPVs.

ARTICLE

Received 23 Sep 2014 | Accepted 1 Dec 2014 | Published 14 Jan 2015

Kuan Sun1,2,3,*, Zeyun Xiao1,*, Shirong Lu1,*, Wojciech Zajaczkowski4, Wojciech Pisula4, Eric Hanssen5,

Jonathan M. White1, Rachel M. Williamson6, Jegadesan Subbiah1, Jianyong Ouyang2, Andrew B. Holmes1, Wallace W.H. Wong1 & David J. Jones1

Solution-processed organic photovoltaic cells (OPVs) hold great promise to enable roll-to-roll printing of environmentally friendly, mechanically exible and cost-effective photovoltaic devices. Nevertheless, many high-performing systems show best power conversion efciencies (PCEs) with a thin active layer (thickness is B100 nm) that is difcult to translate to roll-to-roll processing with high reproducibility. Here we report a new molecular donor, benzodithiophene terthiophene rhodanine (BTR), which exhibits good processability, nematic liquid crystalline behaviour and excellent optoelectronic properties. A maximum PCE of 9.3% is achieved under AM 1.5G solar irradiation, with ll factor reaching 77%, rarely achieved in solution-processed OPVs. Particularly promising is the fact that BTR-based devices with active layer thicknesses up to 400 nm can still afford high ll factor of B70% and high PCE of B8%. Together, the results suggest, with better device architectures for longer device lifetime, BTR is an ideal candidate for mass production of OPVs.

1 School of Chemistry, Bio21 Institute, The University of Melbourne, 30 Flemington Road, Parkville, Victoria 3010, Australia. 2 Department of Materials Science and Engineering, National University of Singapore, 7 Engineering Drive 1, Singapore 117574, Singapore. 3 Department of Renewable Energy, School of Power Engineering, Chongqing University, 174 Shazhengjie, Shapingba, Chongqing 400044, China. 4 Max Planck Institute for Polymer Research, Ackermannweg 10, 55128 Mainz, Germany. 5 Advanced Microscopy Facility, Bio21 Institute, The University of Melbourne, 30 Flemington Road, Parkville, Victoria 3010, Australia.

6 MX Beamlines, Australian Synchrotron, 800 Blackburn Road, Clayton, Victoria 3168, Australia. * These authors contributed equally to this work. Correspondence and requests for materials should be addressed to J.O. (email: mailto:[email protected]

Web End [email protected] ) or to W.W.H.W. (email: mailto:[email protected]

Web End [email protected] ) or to D.J.J. (email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 6:6013 | DOI: 10.1038/ncomms7013 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

& 2015 Macmillan Publishers Limited. All rights reserved.

DOI: 10.1038/ncomms7013 OPEN

A molecular nematic liquid crystalline material for high-performance organic photovoltaics

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms7013

Despite recent developments in solid-state photovoltaic devices13, bulk-heterojunction (BHJ) organic photo-voltaics (OPVs)4 continue to be a promising low-cost

renewable energy technology. The reasons for this outlook include the versatility of organic semiconducting materials and simple device architectures that can be constructed from a variety of printing techniques. The development of the BHJ OPVs has been rapid in recent years driven by a combination of organic material design, interface engineering and improvements in device geometry. The reported power conversion efciency (PCE) of single-junction small-area devices is now routinely in the 68% range57.

Published reports of single-junction BHJ OPVs over 9% PCE are still rare. A handful of polymeric electron donor materials and only one molecular donor have been reported in devices that reached this benchmark813. Molecular OPVs are an attractive alternative to polymer-based OPVs. Higher material purity can be achieved with the well-dened discrete structure of molecules and this should ensure greater reproducibility in devices1416. To achieve high photovoltaic conversion efciency, the material should be capable of forming good lms with high molecular order. This can be achieved by smart molecular design and control over crystallization processes. For molecular semiconductors, conjugated at and rigid backbones are preferred for easy packing via pp interactions. Good solubility is conferred through employment of the appropriate number, type and length of side chains, without hindering the packing of the backbones. The desirable donor phase in BHJ OPVs needs well-ordered nanocrystals with sizes comparable to the exciton diffusion length for efcient charge generation. Strategies simultaneously to enhance molecular order and restrict crystal size have been reported, including thermal annealing, solvent additives, solid additives and solvent vapour annealing (SVA)7,1722. Recent reports showed that the rapid SVA treatment was particularly useful in achieving high ll factor (FF) and PCE in molecular OPVs23,24. Solvent selection rules for SVA treatment were identied in our previous study23.

Despite the important progress achieved in small-area OPV devices fabricated in laboratories, the successful commercialization of OPV technology relies on the application of solution-processed roll-to-roll techniques for large-scale printing25. One of the challenges in printing OPV devices is in printing the optimal active layer thickness of 80120 nm that many of the high-performing material systems require, while obtaining pinhole-free thin lms reproducibly at high printing speed. This problem can be relieved by printing thick lms with thickness over 200 nm26. Unfortunately, due to limited charge diffusion length, thick-lm OPV devices often experience severe bimolecular recombination and space charge effect, leading to reduced FF and PCE27. So far only a few studies have achieved high photovoltaic performance on polymer-based OPVs, with active layers above 200 nm26,2830. No report has been found for molecular OPVs.

In this work, a new molecular electron donor material, benzodithiophene terthiophene rhodanine (BTR), with a benzo[1,2-b:4,5-b0]dithiophene (BDT) core and rhodanine peripheral units was developed and used in OPV devices giving PCEs 49%. While its p-conjugated structure is analogous to a high-performance compound reported previously7,31, the strategic placement of the side chains provided BTR with strong intermolecular interactions, as evidenced by its liquid crystalline (LC) behaviour. Such interactions translated successfully into excellent hole transport properties; hole mobilities up to0.1 and 1.6 10 3 cm2 V 1 s 1 were recorded by organic

eld-effect transistor (OFET) and space-charge-limited current (SCLC) methods, respectively. Thus, BTR-based OPVs with thick active layers (300400 nm) could still afford PCEs of over 8%

with high FF of B70%. Normal cell architecture employed in this study showed a moderate device lifetime. With better cell architectures or proper encapsulation for longer device lifetime, it is believed that BTR is a very attractive candidate for roll-to-roll printed OPV modules.

ResultsPhysical properties of BTR. The BTR molecule was synthesized in two steps from known precursors in a good yield (Supplementary Fig. 1). The chemical structure of BTR is shown in Fig. 1a. The backbone consisting of the BDT unit, two terthiophenes and two rhodanine groups formed a coplanar structure. In comparison with analogous structures in the literature7,31, the side chains of BTR were shortened and positioned at the terthiophene building blocks in a regioregular manner to facilitate side-chain interdigitation32,33. In combination with the additional hexyl group on the thienyl-BDT unit, the side chains of BTR imparted LC behaviour (vide infra) that was not observed in previous reports.

BTR shows an excellent solubility of 211 mg ml 1 in chloroform, as derived from concentration and absorption data (Supplementary Fig. 2). BTR in solution displays an absorption maximum (lmax) at 523 nm, with an extinction coefcient (e) of1.10 105 M 1 cm 1 (Fig. 1b; Supplementary Table 1). The high

e is attributed to the planarity of its backbone. The BTR solid lm exhibits a red-shift of the lmax to 572 nm relative to that in solution. Furthermore, an additional absorption peak at 620 nm appears in the absorption spectrum of a thin lm. The red-shift and new absorption peak of the BTR lm suggest the presence of strong intermolecular interaction and aggregation in the solid lm. The absorption onset of the BTR lm is at 681 nm, equivalent to an optical frontier orbital energy gap of 1.82 eV. Determined by cyclic voltammetry (CV) (Supplementary Fig. 3), the highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) of BTR are 5.34 and 3.52 eV, respectively.

The HOMOLUMO gap of BTR is 1.82 eV, which is in good agreement with the optical energy gap. Because the open-circuit voltage (Voc) is largely determined by the HOMOLUMO gap of the donor and acceptor, a deep-lying HOMO of BTR can potentially support a high Voc. In combination with fullerene acceptor [6,6]-phenyl C71 butyric acid methyl ester (PC71BM), whose LUMO level is around 4.0 eV, the LUMO energy offset of

0.48 eV between BTR and PC71BM should provide enough driving force for exciton dissociation34.

The BTR molecule has good thermal stability with a decomposition temperature of 405 C in nitrogen (5% weight loss in thermogravimetric analysis, Supplementary Fig. 4). BTR exhibits a sharp differential scanning calorimetry (DSC) peak at 175 C (Fig. 1c), which is assigned to secondary crystalline phase transition by means of a structural analysis (vide infra). Furthermore, a melting temperature of 186 C into a LC phase and a clearing temperature at 196 C of small enthalpy into an isotropic melt were observed. Upon cooling, three exothermic peaks at 193, 181 and 133 C were recorded. The rst minor transition was attributed to the LC phase transition, while the two major ones were related to the crystallization process of the two crystalline phases. To observe directly the LC transition and precisely assign the phases, BTR powder was sandwiched in between two glass slides, heated and examined under a polarized optical microscope (POM). The BTR molecule was highly crystalline below a stage temperature of 185 C (Fig. 1d). Between 185 and 195 C, the crystalline solid was replaced by a liquid crystal nematic texture (Fig. 1e). The nematic phase suggests that BTR molecules have a rigid rod-like shape, which can maintain a long-range directional order with their long axes parallelly

2 NATURE COMMUNICATIONS | 6:6013 | DOI: 10.1038/ncomms7013 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2015 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms7013 ARTICLE

S

S

S S

S S S

N

O

O N

S

S

S

S

S

S

S

BTR

2

Normalized absorbance (a.u)

1.0 Solution Film

133 C

181 C

Heat flow (W g1, exo up)

1

193 C

0.8

0.6

0

0.4

1

0.2

2

196 C 175 C 186 C

0.0

3

400 500 600 700 50 100 150 200 250

20 m 20 m 20 m

800

0

Wavelength (nm)

Temperature (C)

Figure 1 | BTR chemical structure and physical properties. (a) Chemical structure of BTR. (b) Normalized UVvis absorption spectra of BTR in chloroform (5 mg ml 1) and in a spin-cast lm. (c) DSC thermogram of BTR in nitrogen at a ramp rate of 10 C min 1. The lower trace is from the heating cycle and upper trace from the cooling cycle. (d) BTR thin lm sandwiched in between two glass slides observed under a polarized optical microscope (POM) at a stage temperature of 185 C. (e) The POM image of the same BTR thin lm at the same settings when the stage temperature rises to 195 C. (f) The POM image taken at a stage temperature of 197 C.

aligned. They can thus have high crystallinity in solid state17,3537. The liquid crystal transformed above 196 C into an isotropic melt, leaving no prominent feature under the POM (Fig. 1f). Thereby, the small transition enthalpy determined by DSC is in agreement with the low-ordered nematic phase. The nematic LC behaviour is an important feature of the BTR molecule, implying strong intermolecular interaction resulted from side-chain modications, and potentially high charge carrier mobility due to three-dimensional (3D) charge transport38.

Crystal packing of BTR molecules. To obtain a better understanding of the packing of BTR molecules in the solid, X-ray quality single crystals of BTR were grown from a mixed solution of 2-propanol and dichloromethane by slow evaporation. The single crystal structure was solved using data from the MX2 beamline at the Australian Synchrotron39 (Fig. 2a; Supplementary Figs 58). The X-ray crystal structure of the BTR molecule revealed a coplanar structure of the conjugated backbone, which should facilitate light absorption and also crystal stacking. The crystal packing is dominated by p-stacking between the individual

BTR backbones that arrange themselves into p-stacked centrosymmetric dimers with an average inter-plane separation of ca 3.60 (Fig. 2a). These individual dimers aggregate together

by p-stacking, with an average interplaner separation of 3.62 (Supplementary Fig. 6). This type of packing is consistent with the bathrochromic shifting of the absorption from solution to the solid lm (J-aggregate).

The solid-state structure was also examined using two-dimensional wide-angle X-ray scattering (2D-WAXS) on neat BTR laments. The samples were prepared by lament extrusion40, which imparted bulk orientation on the crystalline material. The 2D-WAXS pattern suggests a crystalline character of BTR in the low-temperature phase as evident by the high number of distinct reections (Fig. 2b). The molecules are organized in a layered structure that is aligned in the direction of the bre axis. An interlayer distance of 18.3 is determined from reections located in the equatorial small-angle range. On the same plane of the pattern, two p-stacking peaks appear that are related to distances of 3.70 and 3.65 of stacked BTR dimers. These values are in the same range as found for the single crystal. Further meridional reections are originated from intramolecular correlations along the extended conjugated BTR backbone. At 179 C, the sample maintains a crystalline phase, however, with a slightly smaller degree of order (Supplementary Fig. 9). The interlayer spacing remains identical at 18.3 , while only one and a little larger stacking distance was observed at 3.76 .

NATURE COMMUNICATIONS | 6:6013 | DOI: 10.1038/ncomms7013 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

& 2015 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms7013

In a thin solid lm, BTR organizes in two different molecular arrangements as indicated by the grazing incidence wide-angle X-ray scattering (GIWAXS) pattern in Fig. 2c. Reections in the meridional plane (along qxy 0 1) in the small- and middle-

range scattering region are related to the formation of a layered structure with an interlayer distance of 18.7 . In addition, 3rd-order reections are visible typical for a long-range order, while their position on the meridional plane of the pattern is characteristic for an edge-on molecular organization. In this

arrangement, the backbone plane is aligned perpendicular to the surface. However, the corresponding equatorial p-stacking peak of the edge-on arranged molecules assembled in the layered structure is too weak to be detected. Instead, the p-stacking reection related to a single distance of 3.70 is located also on the meridional plane, which is typical for a face-on arrangement. These results imply two distinct surface organizations. In the rst phase, the molecules are p-stacked and face-on arranged, but do not organize in a layered structure. In the second phase, the molecules are edge-on aligned with respect to the substrate, but are disordered within the layer organization. Because charge transport in organic semiconductors is mainly via hopping between adjacent molecules, the co-existence of edge-on and face-on orientations can potentially form a 3D network for hopping, thus benecial to charge transport41.

OFET mobility. To study the charge carrier transport, OFETs using different procedures were built. For top-contact devices, BTR was spin-coated from a 4.5-mg ml 1 toluene solutionand subsequently annealed at 179 C. These transistors delivered hole mobilities up to 0.01 cm2 V 1 s 1 (Supplementary Fig. 10).

Bottom-contact OFET devices with the BTR molecules deposited by drop-casting gave mobility values as high as 0.1 cm2 V 1 s 1 (Supplementary Fig. 11). It should be noted that the OFET devices were not intensively optimized. The primary purpose of the OFET experiments was to show the potential of the BTR material as a semiconductor.

Photovoltaic performances. The excellent solubility, strong intermolecular interaction, suitable absorption prole and energy levels, as well as encouraging semiconducting properties prompted us to explore the photovoltaic performance of the BTR

2

1.5

0.5

q z/1

1

0 0.5 1.5

1 2

qxy /1

Figure 2 | Crystal packing resolved by X-ray techniques.(a) Centrosymmetric p-stacked dimers of BTR molecules in its single crystal, the alkyl side chains have been omitted for clarity. (b) 2D-WAXS of BTR lament measured at 30 C. (c) GIWAXS of the as-cast BTR thin lm on silicon wafer via spin coating (p-stacking reection is indicated by an arrow).

4

1,000

2

100

Current density (mA cm2)

0

Dark current

10 As cast

1 THF SVA

Voltage (V)

0.1

Al

Al

2

Current density (mA cm)

0.01

1E3

Ca

4

1E4

1E5

BTR PC71BM

Ca

8

6 1E6 1.0

0.5 0.0

0.5 1.0 1.5

Voltage (V)

10

As cast THF SVA

PEDOT:PSS

12

ITO

14

0.8

0.4 0.0 0.4 0.8 1.2

100

90

4

80

2

70

0

60

2

Jsc = 13.90 mA cm2 Joc = 0.90 V

EQE (%)

50

4

40

6

FF = 74.1% PCE = 9.3%

30

8

20

10

10

12

0

14

300

400 500 600 700 800

Current density (mA cm2)

1.0

0.8

0.6

0.4

0.2

0.0

0.2

0.4

0.6

0.8

1.0

Wavelength (nm)

As cast THF SVA

Voltage (V)

Figure 3 | Device architecture and photovoltaic performances. (a) Schematic diagram of a normal cell architecture used in this study. (b) JV characteristics of BTR:PC71BM BHJ solar cells with or without THF solvent vapour annealing tested in air under 98 mWcm 2 AM1.5G illumination.

Inset: dark current plotted in a semi-log scale of the two solar cells. (c) EQE spectra of optimized BTR-based solar cells with or without THF SVA treatment. (d) JV curve of the most efcient BTR:PC71BM BHJ solar cell after 15 s of THF SVA measured by an independent research institute in nitrogen atmosphere under an illumination of 100 mWcm 2.

4 NATURE COMMUNICATIONS | 6:6013 | DOI: 10.1038/ncomms7013 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2015 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms7013 ARTICLE

molecule. The OPV cells adopted a simple normal architecture, with the BTR:PC71BM blend lm sandwiched between a poly(3,4-ethylenedioxythiophene):poly(styrenesulfonate) (PEDOT:PSS)-coated indium tin oxide (ITO) transparent anode and a Ca/Al back cathode (Fig. 3a). We further treated the active layer with SVA, which has been shown to be effective in enhancing the performance of molecular OPVs2224. The SVA treatment was carried out by exposing the as-cast active layer to solvent vapours. According to the solvent selection rules previously identied23, tetrahydrofuran (THF) was chosen for SVA owing to the moderate solubility of BTR in THF (89 mg ml 1).

The BTR-based OPVs with an optimal active layer thickness of 250 nm were encapsulated and tested in air. The current density (J)voltage (V) curves of the best devices are shown in Fig. 3b, with the photovoltaic parameters summarized in Table 1. Without SVA treatment, the highest performance for the as-cast OPVs showed short-circuit current density (Jsc) 11.64 mA cm 2,

Voc 0.96 V, FF 47% and PCE 5.2%. SVA treatment sig

nicantly enhanced the photovoltaic performance. OPVs with 15 s of THF SVA exhibited Jsc 13.52 mA cm 2, Voc 0.89 V,

FF 73% and PCE 8.7%. Device assembly was reproducible

with around 60 SVA-treated OPV devices having an average PCE of 8.30.2%. Thermal annealing was found to diminish the device performance, due to the overgrowth of the phases (Supplementary Fig. 12).

The causes for the enhanced FF after SVA treatment were investigated by measuring dark currents (inset of Fig. 3b). Compared with an as-cast molecular OPV, the SVA-treated sample displayed notably higher current density under positive bias. In great contrast, the current density was one order of magnitude smaller in reverse bias. To further understand the SVA treatment effect, series resistance (Rs) and shunt resistance (Rsh)

were extracted at 1.5 and 0 V of the dark curves (Table 1). Without SVA treatment, the OPV had a Rs of 14.0 O cm2 and a

Rsh of 5.5 MO cm2. SVA treatment led to a reduction of Rs by six times and a slight increase of Rsh. Together, the results suggest the SVA treatment can suppress leakage current and improve the diode behaviour.

The slight improvement in Jsc after SVA treatment was monitored by external quantum efciency (EQE) measurement (Fig. 3c). A high EQE of over 60% was measured in the visible region from 400 to 650 nm for the non-annealed OPV. The Jsc

calculated by integrating the product of photon ux and EQE at

each wavelength was 11.70 mA cm 2, which was in good agreement with the measured Jsc (11.64 mA cm 2). The SVA treatment lifted the EQE in the entire absorption range. In particular, the EQE stayed above 70% between 400 and 650 nm, and a shoulder was found at 640 nm. As a result, the calculated Jsc

increased to 13.53 mA cm 2. The EQE result clearly indicates SVA treatment plays a positive role in charge generation, transport and/or collection.

Bearing in mind that OPVs with normal cell architecture are not stable in air, we fabricated a batch of 20 devices in Singapore and 8 devices in Australia and tested them under inert atmosphere using the facilities at Solar Energy Research Institute of Singapore and the Commonwealth Scientic and Industrial Research Organisation, respectively. The best BTR-based OPV fabricated in Singapore exhibited a record efciency of 9.3%, with Jsc 13.90 mA cm 2, Voc 0.90 V and FF 74.1% (Fig. 3d;

Table 1). The results were highly reproducible. The same PCE of 9.3% with a Jsc of 13.40 mA cm 2, a Voc of 0.90 V and an extremely high FF of 77.0% was achieved in Australia (Table 1).

This result demonstrates molecular OPVs can achieve comparable efciencies attainable by polymer-based OPVs811. It is worth noting that the FF of 77.0% is among the highest FF value reported in the literature for solution-processed molecular OPVs12,42. The average photovoltaic parameters for the 28 devices were Jsc 13.490.28 mA cm 2, Voc 0.890.01 V,

FF 741% and PCE 8.90.2% (Table 1).

OPVs of a thick active layer. The high FF values suggest that the BTR-based OPVs can accommodate a greater range of active layer thicknesses. This is particularly important in roll-to-roll printing of very thin lms, which are difcult to be precisely controlled, and pinholes are often found in thin-lm devices. We were motivated to explore the thickness-dependent solar cell performance using the BTR molecule. Active layers with different thicknesses ranging from 80 to 400 nm were fabricated by tuning the solution concentrations and spin rates. Figure 4 and Supplementary Table 2 show that BTR-based OPVs maintain a nearly constant Voc

between 0.87 and 0.90 V. The average Jsc increases from B10 to B13 mA cm 2 as the active layer thickness increases from 80 to 250 nm and then it saturates around 13 mA cm 2 when the thickness further increases to 400 nm. Surprisingly, the FF values for BTR-based OPVs remain high and close to 70% even at

Table 1 | Photovoltaic parameters of BTR:PC71BM BHJ solar cells fabricated and tested under different conditions

Jsc (mA cm 2) Voc (V) FF (%) PCE (%) Rs* (X cm2) Rshw (MX cm2) lhz (cm2 V 1 s 1) lez (cm2 V 1 s 1) As-casty11.64 0.96 47 5.2 14.0 5.5 2.2 10 4 3.5 10 4

(11.200.51) (0.960.01) (423) (4.50.4)

THF SVAy13.52 0.89 73 8.7 2.4 42[12.16] [0.90] [76] [8.3] 2.2 6.4(13.170.43) (0.890.01) (712) (8.30.2) 1.6 10 3 9.6 10 3

THF SVA||13.90 0.90 74 9.3 2.7 15[13.40] [0.90] [77] [9.3] 1.9 20(13.610.16) (0.890.01) (741) (8.90.2)

FF, ll factor; PCE, power conversion efciency; SCLC, space-charge-limited current; SVA, solvent vapour annealing; THF, tetrahydrofuran.

Data in brackets show the cell with the highest FF. Data in parentheses are average results out of 40 encapsulated devices tested in air or 28 non-encapsulated devices tested in nitrogen.

*Series resistance of the device. wShunt resistance of the device.

zCharge mobility values obtained from SCLC experiments.yCells were tested in air with encapsulation, illumination intensity was 98 mWcm 2.

||Cells without encapsulation were tested in a glove box lled with dry nitrogen; illumination intensity was 100 mWcm 2.

NATURE COMMUNICATIONS | 6:6013 | DOI: 10.1038/ncomms7013 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

& 2015 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms7013

0.94

100

10

14

95

12

8

0.92

90

Jsc(mA cm2)

10

85

6

8

0.90

Voc(V)

FF (%)

80

PCE (%)

6

75

4

4

0.88

70

2

2

65

Jsc

FF

0 50 100 150 200 250 300 350 400 450

0.86

60 50 100 150 200 250 300 350 400 450

0

Active layer thickness (nm)

V oc

Active layer thickness (nm)

PCE

Figure 4 | Active layer thickness-dependent variation of photovoltaic performances. (a) Plots of Jsc or Voc vs active layer thickness ranging from80 to 400 nm. (b) Plots of FF or PCE against active layer thickness. The results are an average value of 48 devices. The error bars represent the standard deviation from 48 devices.

100 nm

100 nm

250 nm

250 nm

100 nm 2 m

2 m

100 nm

Figure 5 | Solvent vapor annealing induced morphological changes. (a) AFM image shows the topography of an as-cast BTR:PC71BM (1:1 weight

ratio) blend lm. (b) TEM bright-eld image of the as-cast lm taken at a defocusing range of 3 mm. (c) Computer model generated from the TEM tomogram of the as-cast lm. (d) Low-energy HAADF STEM image of the as-cast lm at focus using a beam energy of 15 keV. (e) AFM image of the

BTR:PC71BM blend lm after THF SVA for 15 s. (f) TEM bright-eld image of the SVA-treated lm at a defocusing range of 3 mm. (g) Computer model of the THF SVA lm. (h) HAADF STEM image of the blend lm after SVA treatment.

thicknesses up to 400 nm. This is not commonly observed in thick-lm OPVs, whether it is a molecular OPV or a polymer-based solar cell2629,43. As a result, the overall PCEs formed a at bell curve with a minimum average value of 6.8% and maximum average value of 8.3% at an active layer thickness of 250 nm. The large tolerance for the active layer thickness makes the BTR molecule a strong candidate for printed OPVs.

Solvent vapour annealing. To understand the effect of SVA treatment on the photovoltaic performance of BTR-based OPVs, we carried out studies on active layer morphology and the optoelectronic properties. The surface topography of the active layer was recorded by atomic force microscopy (AFM) operated in the tapping mode. Before the SVA treatment, Fig. 5a depicts a rather smooth surface, with root-mean-square roughness (Rrms) of

0.61 nm. Fine crystal domains co-exist with random pinholes, which are believed to be related with the escaping of processing solvent. After a short THF SVA treatment of 15 s, the active layer exhibits a coarser surface (Fig. 5e). The Rrms value almost doubles to1.04 nm. Transmission electron microscopy (TEM) is able to provide morphological information inside the active layer. The bright-eld TEM images (Fig. 5b,f) suggest THF SVA treatment leads to larger and more well-dened domains. Because of the sharp contrast in the TEM images, we were able to obtain TEM tomograms

and computer models to view the morphological change in 3D (Fig. 5c,g; Supplementary Movies). Both the TEM tomograms and their computer models show that ne-sized domains in the as-cast active layer (Supplementary Movies 1 and 2) evolve into larger domains that are inter-connected to form networks throughout the entire active layer after THF SVA for 15 s (Supplementary Movies 3 and 4). Such networks resemble 3D charge highways that are benecial to fast charge transport. The feature size on TEM images is veried by low-energy high-angle angular dark-eld scanning TEM (HAADF STEM) images (Fig. 5d,h).

The SVA treatment can be monitored by colour change of the active layer. The inset of Supplementary Fig. 13 is a digital image of the active layer before and after the THF SVA treatment. The colour of the lm changed from maroon to purple upon annealing by THF vapour. Such a colour change was reected by the change in absorption prole (Supplementary Fig. 13). There was a slight red-shift of the absorption maximum from 555 to 565 nm. Besides, the shoulder at 620 nm became more prominent, suggesting good alignment of the rod-like molecules. The absorption enhancement at 620 nm directly translated to increased photocurrent, as suggested by the EQE plot (Fig. 3c).

GIWAXS measurements were performed to understand the organization of BTR in the active layer before and after SVA. In comparison with the BTR neat lm, the edge-on layered

6 NATURE COMMUNICATIONS | 6:6013 | DOI: 10.1038/ncomms7013 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2015 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms7013 ARTICLE

organization remains unchanged in the as-cast BTR:PC71BM

blend lm, while the p-stacking distance slightly increases to3.80 and becomes randomly distributed towards the surface, as conrmed by the isotropic intensity of the corresponding peak (Supplementary Fig. 14a). The amorphous halo at q-range of ca.1.25 1 is attributed to PC71BM domains. SVA improves the crystallinity and surface ordering of BTR. The interlayer distance is reduced to 17.75 , while the p-stacking distance decreases to3.60 . The random orientation of p-stacking evolves into the coexistence of both edge-on and face-on arrangements after SVA, evidenced from the p-stacking reections at ca. 1.7 1 in both qxy and qz directions (Supplementary Fig. 14b). Such a molecular arrangement is benecial to 3D charge transport.

SCLC mobilities. The hole mobility was measured using the SCLC method with a cell architecture: ITO/PEDOT:PSS/ BTR:PC71BM/Au. Without SVA treatment, the blend lm exhibited a relatively high hole mobility of 2.2 10 4 cm2

V 1 s 1 (Supplementary Fig. 15; Table 1). The rigid and planar backbone facilitates easy stacking and strong intermolecular interaction due to side-chain modications. The SVA treatment substantially enhanced the mobility by one order of magnitude to 1.6 10 3 cm2 V 1 s 1, which is comparable to or greater

than those reported for high-performing donor:acceptor blend systems7,31,41,44,45. The electron mobility derived from the SCLC method was also improved by one order of magnitude after SVA treatment (Supplementary Fig. 16; Table 1). Such an enhancement can be attributed to larger and more structured domains, as well as better molecular arrangement. The extremely high mobility would partially account for the high FF observed in BTR-based OPVs. However, we do not exclude other possible factors including vertical phase separation or removal of recombination centres and so on33.

Solar cell stability. For practical application, solar cell-stability experiments were carried out in both air and nitrogen environments. Due to the use of active metal-like calcium for the top electrode, the unencapsulated OPVs of the thick active layer (B400 nm) degraded to almost zero efciency within three days of storage in air (Supplementary Fig. 17; Supplementary Table 3). However, a simple encapsulation with ultraviolet (UV)-curable epoxy and thin glass slides could greatly improve the device stability in air. The OPVs degraded three times slower than that without encapsulation (Supplementary Fig. 17; Supplementary Table 4). To minimize the degradation factor due to the oxidation of electrode and further explore the stability of the active layer, one BTR-based solar cell device was stored in a glove box lled with dry nitrogen and was monitored over a time span of 30 days. The cell retained 86% of initial PCE after 7 days, and it exhibited 450% of original PCE after 30 days of storage (Supplementary

Fig. 18a,b; Supplementary Table 5). Further enhancement of device stability could be achieved by improved device architecture. Average of 10 OPV cells of thick active layer of 400 nm and an additional 30-nm-thick silver top electrode/protection layer with or without encapsulation retained 92 and 86% of initial average PCE after 30 days of storage in a glove box (Supplementary Fig. 18c; Supplementary Tables 6 and 7). We believe even better stability can be obtained if the cells were properly encapsulated or inverted cell architecture was employed.

DiscussionIn summary, we present a new molecular donor, BTR, which possesses a rigid and at backbone and a large number of exible side chains that could work synergistically to provide excellent processability, nematic liquid crystal behaviour and

optoelectronic properties. The neat BTR lm exhibited hole mobilities up to 0.1 cm2 V 1 s 1 in OFET devices. The solution-processed single-junction BHJ solar cells based on BTR and

PC71BM demonstrated a reproducible record efciency of 9.3%. The blend lm also supported a high FF of 77% and a high SCLC hole mobility of 1.6 10 3 cm2 V 1 s 1 after SVA with THF.

Thick-lm molecular solar cells with an active layer thickness up to 400 nm were demonstrated, showing a low thickness dependence of photovoltaic performance. Together, the results suggest BTR is an ideal candidate for printed OPVs. Moreover, enhancing the intermolecular interaction through side-chain modication is a viable way further to enhance the efciency of molecular solar cells in excess of 10%.

Methods

Materials. Unless noted, all materials were reagent grade and used as received without further purication. Anhydrous solvents were prepared by drying HPLC-grade solvents using freshly activated molecular sieves.

Synthesis of aldehyde compound 3. Synthetic route could be found in Supplementary Fig. 1. Precursor 1 and 2 were prepared using literature methods46,47. In protection of N2, to a dry 250-ml ask were added compound 1(1.07 g, 1.0 mmol), compound 2 (1.05 g, 2.0 mmol), Pd2(dba)3 (46 mg, 0.04 mmol), P(o-tolyl)3 (97 mg, 0.32 mmol) and 50 ml of toluene. The reaction mixture was reuxed at 125 C for 12 h. The reaction mixture was ltered and the crude product was puried by silica gel column chromatography (petroleum spirit 4060 C: dichloromethane 1:1.5, Rf 0.5) to give the product as yellow solid (830 mg, 53%).

mp 146147 C; infrared (IR) (neat) n 2,955, 2,923, 2,855, 1,656, 1,432, 1,225, 1,058, 820, 786 and 663 cm 1; 1H NMR (d, CDCl3) 9.89 (s, 2 H), 7.72 (d, J 4.0 Hz, 2 H),

7.68 (s, 2 H), 7.23 (d, J 4.0 Hz, 2 H), 7.12 (s, 2 H), 7.01 (s, 2 H), 2.82 (m, 12 H),

2.65 (m, 4 H), 1.601.75 (m, 14 H), 1.261.45 (m, 52 H), 0.88 (m, 30 H); 13C NMR (d, CDCl3) 182.6, 146.1, 142.6, 142.2, 141.2, 139.6, 139.0, 138.6, 137.3, 137.2, 136.9, 136.0, 135.8, 134.8, 130.3, 129.8, 129.5, 129.1, 128.3, 125.9, 123.5, 119.5, 41.7, 32.7,32.3, 31.8, 31.6, 30.8, 30.4, 30.2, 29.8, 29.6, 29.3, 29.2, 29.1, 28.9, 28.4, 26.0, 23.1,22.7, 22.6, 14.2 and 13.1; mass spectrometry (MS) (matrix-assisted laser desorption/ionization (MALDI)) m/z 1,630 [M].

Synthesis of BTR molecule. To the solution of compound 3 (327 mg, 0.2 mmol) in dry chloroform (20 ml) was added rhodanine 4 (434 mg, 2.0 mmol) followed by one drop of 1,8-Diazabicyclo[5.4.0]undec-7-ene (DBU). The reaction mixture was then placed under continuous stirring at room temperature for 3 h. Reaction solvent was removed and the crude product was puried by silica gel column chromatography (petroleum spirit 4060 C: dichloromethane 1:1, Rf 0.8). The desired fractions

were collected and the product was obtained as a purple solid (310 mg, 76%) after washing with acetone. mp 178180 C; IR (neat) n 2,924, 2,855, 1,699, 1,575, 1,423, 1,327, 1,238, 1,180 and 820 cm 1; 1H NMR (d, CDCl3) 7.83 (s, 2 H), 7.65 (s, 2 H),7.35 (d, J 4.0 Hz, 2 H), 7.23 (s, 2 H), 7.20 (d, J 4.0 Hz, 2 H), 7.10 (s, 2 H), 7.00 (s,

2 H), 4.09 (t, J 7.6 Hz, 4 H), 2.81 (m, 12 H), 2.78 (m, 4 H), 1.601.75 (m, 18 H),

1.261.45 (m, 64 H), 0.88 (m, 36 H); 13C NMR (d, CDCl3) 192.2, 167.5, 144.3, 141.9, 141.0, 139.5, 139.0, 138.6, 137.3, 137.2, 137.1, 135.7, 135.6, 134.8, 134.6, 130.4, 129.8, 129.7, 129.0, 128.3, 126.5, 125.0, 123.5, 120.2, 119.5, 44.9, 41.7, 32.7, 32.4, 31.8, 31.7,31.6, 31,3, 30.8, 30.4, 30.3, 29.8, 29.7, 29.3, 29.2, 29.1, 28.9, 28.4, 27.0, 26.4, 26.0, 23.1,22.7, 22.6, 22.5, 14.2, 13.1; MS (MALDI) m/z 2,028.7 [M]. Elemental analyses calcd (%) for C114H152N2O2S14 (BTR): C, 67.41; H, 7.54; N, 1.38; O, 1.58; S, 22.10;

found: C, 67.64; H, 7.65; N, 1.24; O, 1.33; S, 21.87.

Material characterizations. IR spectra were obtained on a Perkin-Elmer Spectrum One Fourier transform infrared spectrometer and UVvisible spectra were recorded using a Cary 50 UVvisible spectrometer. Photoluminescence was measured with a Varian Cary Eclipse uorimeter. Melting points were determined on a Bchi 510 melting point apparatus. 1H NMR and 13C NMR spectra were carried out on a 400-MHz spectrometer. All NMR data were referenced to the chloroform signal and peak multiplicity was reported as follows: s singlet,

d doublet, t triplet, q quartet, p pentet, dd doublets of doublets,

m multiplet and br broad). MALDI-time-of-ight MS was performed on a

Bruker microex instrument, using chloroform as solvent and dithranol as the assisted matrix. Elemental analyses were obtained commercially through Chemical & Analytical Services Pty Ltd. (Australia) an Exeter Analytical CE440 elemental analyzer. Thermal gravimetric analysis experiments were carried out with a Mettler Toledo TGA/SDTA851e, and DSC experiments were performed on a Perkin-Elmer Sapphire DSC. CV experiments were performed at a sweep rate of 100 mV s 1.

CVs were carried out in a three-electrode cell consisting of a glassy carbon working electrode, a platinum wire auxiliary electrode and a Ag/Ag pseudo-reference electrode. The supporting electrolyte was 0.10 M tetrabutylammonium hexauorophosphate (Bu4NPF6) in CH3CN. The solutions were deoxygenated by sparging with argon prior to each scan and blanketed with argon during the scans.

NATURE COMMUNICATIONS | 6:6013 | DOI: 10.1038/ncomms7013 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 7

& 2015 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms7013

The glassy carbon working electrode was prepared by polishing with 5 mm alumina and washed and dried before the polymer was drop-casted on the electrode from chlorobenzene solution to form a lm. Ferrocene/ferrocenium redox couple was used as the internal standard. The HOMO energy level was calculated fromthe onset of the oxidation potential of the polymer using the following:EHOMO (4.8 Eox onset) eV.

Crystal structure determination. Single-crystal data were collected on the MX2 beamline at The Australian Synchrotron at 100 K using a 360 rotation in steps of 1 with 1-s exposure time per step. The beamline was set to the Mo-K-a edge(17.444 keV or 0.71073 ). Crystal data for BTR: C112 H150 N2 O2 S14, M 2,005.17, T 100.0(2) K, l 0.71073 , Triclinic, space group P-1,

a 314.257(3), b 20.519(4), c 21.795(4) , a 114.76(3), b 98.08(3),

g 02.00(3), V 5,474(2) 3, Z 2, Dc 1.217 Mg M 3, m 0.326 mm 1,

F(000) 2,152, crystal size 0.10 0.05 0.05 mm. ymax 27.47, 88,775 reections

measured, 22,889 independent reections (Rint 0.055) the nal R 0.11

(I42s(I), 14,445 data) and wR(F2) 0.3696 (all data) GOOF 1.371.

Thin-lm characterizations. To study the liquid crystal property of BTR, we used a Nikon Eclipse LV100 POM equipped with Linkam LTS350 heating and cooling stage connected to a Linkam TMS 94 temperature programmer. GIWAXS samples were prepared by spin-coating chloroform solution of BTR onto PEDOT:PSS-coated silicon substrates. The measurements were performed by means of a solid anode X-ray tube (Siemens Kristalloex X-ray source, copper anode X-ray tube operated at 35 kV and 40 mA), Osmic confocal MaxFlux optics, X-ray beam with pinhole collimation and a MAR345 image plate detector. The beam size was0.5 0.5 mm and samples were irradiated just below the critical angle for total

reection with respect to the incoming X-ray beam (B0.18). The scattering intensity was detected on a 2D image plate (MAR345) with a pixel size of 100 mm (3,450 3,450 pixels). Data analysis was performed using the Datasqueeze 3.0.0

software. Atomic force microscopy images were acquired using an Asylum Research Cypher scanning probe microscope operated in tapping mode. Samples for electron microscopy were prepared by dissolving the PEDOT:PSS layer using water and transferring the oating active layer to TEM grids. TEM bright-eld images were obtained by a FEI Tecnai TF30 TEM equipped a with beam-blank function. For TEM tomography, tilt series were acquired using the Xplore 3D software (FEI Company). Tomograms were recorded between 65 and 65

degrees at 2 intervals and aligned with IMOD. A 3D model rendering employed 3d mod software. Each model was generated from the aligned tomogram. The BTR donor phase was rendered in pink colour, while the PC71BM acceptor phase was represented by empty space in model movies. The scale bar was 100 nm in both the model movies and the tomogram movies. HAADF STEM was performed at a primary electron energy of 15 keV with a FEI Quanta 3D Microscope equipped with a HAADF STEM detector. To obtain hole and electron mobilities using SCLCs, hole-only devices (ITO/PEDOT:PSS/BTR:PC71BM/Au) and electron-only devices (ITO/Al/ BTR:PC71BM/Al) were constructed. Their dark currents were recorded by a computer-programmed Keithley 2400 source meter and then tted by the MottGurney equation. Film thickness was determined by a Veeco Dektak 150 Surface Proler.

Solvent vapour annealing. SVA was conducted in a glove box lled with dried nitrogen. THF (1 ml) were injected into a 30-mm glass Petri dish. The Petri dish was closed for 1 min to let the vapour saturate the treatment chamber. Then as-cast lms were attached on the backside of the Petri dish lid, which was quickly swapped with the lid covering the solvent-containing Petri dish. The lm was about 1 cm above the solvent level during the SVA. After certain duration, the lm was removed from the treatment chamber. The optimal duration in this study was 15 s.

OFET fabrication and measurement. OFETs were fabricated employing two types of device congurations: bottom-gate, bottom-contact and bottom-gate, top-contact on the 300-nm-thick silicon-dioxide dielectric covering the highly doped silicon that acted as the gate electrode. The sourcedrain channel length (L) to channel width (W) ratio was 1:70 for bottom-contact and 1:33 for top-contact congurations. The substrates were rst cleaned by sonication in acetone and isopropanol for 20 min each, then dried under nitrogen ow and dipped into a freshly prepared piranha solution (7:3 v/v H2SO4/30% H2O2) at 90 C for 60 min.

To remove residues of the piranha solution, substrates were rinsed with copious deionized water and dried under nitrogen ow. The BTR thin lm was deposited via spin-coating (for top-contact OFETs) or drop-casting (bottom-contact OFETs) using 4.5 mg ml 1 toluene with subsequent annealing at 60 C for 60 min to remove residual solvent. In addition, the BTR lms were measured after thermal annealing at 179 and 190 C. All the electrical measurements are performed in a glove box under nitrogen atmosphere by means of a Keithley 4200 SCS.

Solar cell fabrication and characterization. Patterned ITO glasses were washed sequentially by detergent, deionized water, acetone and 2-propanol in an ultrasonication bath and UV/ozone-treated. PEDOT:PSS (Clevios P VP AI 4083) was spin-coated at 6.000 r.p.m. and then baked at 140 C for 10 min in air. The

substrates were transferred to a glove box lled with dried nitrogen, where a chloroform solution of the donoracceptor blend was spin coated at various spin rates on top of the substrate. The best PCE was achieved by spin coating 20 mg BTR and 20 mg PC71BM in 1 ml of chloroform at 1,000 r.p.m. The resulting lm thickness was around 250 nm. Thicker or thinner lms were produced by varying the solution concentration and spin rate from 1.6 to 2.0% and from 600 to 2,000 r.p.m., respectively. To obtain high photovoltaic performance, the lms were solvent vapour annealed by THF. Then they were transferred to a thermal evaporator where 40 nm Ca and 100 nm aluminium were deposited through a shadow mask (active area was 0.11 cm2) at a base pressure of 4 10 7 mbar. The solar

cells were encapsulated under nitrogen by UV-curable epoxy (Epotek OG112-6 by Epoxy Technology Inc.) and cover glass. They were tested in air with a computer-programmed Keithley 2400 source meter under a Newport Oriel class A solar similar, which simulated the AM1.5 solar irradiance with energy density of98 mW cm 2. For accurate measurement and to avoid oxidation during testing, a batch of optimized solar cells were brought to a third-party research institute, Solar

Energy Research Institute of Singapore, where the cells were tested under nitrogen and 1 SUN (100 mW cm 2) condition provided by a 1 kW solar simulator(Sun 2000 Solar Simulator by ABET Technologies) that was calibrated by a silicon reference cell (Fraunhofer ISE) and a Schott visible-colour glass-ltered (KG5 colour ltered) Si diode (Hamamatsu S1133). OPVs made in Australia followed the same fabrication procedures, and were tested at Clayton Laboratories of Commonwealth Scientic and Industrial Research Organisation, where a 1-kW Oriel solar simulator with an AM 1.5G lter was used as the light source.

References

1. Burschka, J. et al. Sequential deposition as a route to high-performance perovskite-sensitized solar cells. Nature 499, 316319 (2013).

2. Liu, M., Johnston, M. B. & Snaith, H. J. Efcient planar heterojunction perovskite solar cells by vapour deposition. Nature 501, 395398 (2013).

3. Zhou, H. et al. Interface engineering of highly efcient perovskite solar cells. Science 345, 542546 (2014).

4. Yu, G., Gao, J., Hummelen, J. C., Wudl, F. & Heeger, A. J. Polymer photovoltaic cells: enhanced efciencies via a network of internal donor-acceptor heterojunctions. Science 270, 17891791 (1995).

5. Chen, H. Y. et al. Polymer solar cells with enhanced open-circuit voltage and efciency. Nat. Photon. 3, 649653 (2009).

6. Small, C. E. et al. High-efciency inverted dithienogermolethienopyrrolodione-based polymer solar cells. Nat. Photon. 6, 115120 (2012).

7. Liu, Y. et al. Solution-processed small-molecule solar cells: breaking the 10% power conversion efciency. Sci. Rep. 3, 3356 (2013).

8. He, Z. et al. Enhanced power-conversion efciency in polymer solar cells using an inverted device structure. Nat. Photon. 6, 591595 (2012).

9. Ye, L., Zhang, S., Zhao, W., Yao, H. & Hou, J. Highly efcient 2D-conjugated benzodithiophene-based photovoltaic polymer with linear alkylthio side chain. Chem. Mater. 26, 36033605 (2014).

10. Liao, S.-H., Jhuo, H.-J., Cheng, Y.-S. & Chen, S.-A. Fullerene derivative-doped zinc oxide nanolm as the cathode of inverted polymer solar cells with lowbandgap polymer (PTB7-Th) for high performance. Adv. Mater. 25, 47664771 (2013).

11. Guo, X. et al. Enhanced photovoltaic performance by modulating surface composition in bulk heterojunction polymer solar cells based on PBDTTT-CT/PC71BM. Adv. Mater. 26, 40434049 (2014).

12. Gupta, V. et al. Barium: an efcient cathode layer for bulk-heterojunction solar cells. Sci. Rep. 3, 1965 (2013).

13. Li, C.-Z. et al. Suppressed charge recombination in inverted organic photovoltaics via enhanced charge extraction by using a conductive fullerene electron transport layer. Adv. Mater. 26, 62626267 (2014).

14. Walker, B., Kim, C. & Nguyen, T.-Q. Small molecule solution-processed bulk heterojunction solar cells. Chem. Mater. 23, 470482 (2010).

15. Mishra, A. & Bauerle, P. Small molecule organic semiconductors on the move: promises for future solar energy technology. Angew. Chem. Int. Ed. 51, 20202067 (2012).

16. Lin, Y., Li, Y. & Zhan, X. Small molecule semiconductors for high-efciency organic photovoltaics. Chem. Soc. Rev. 41, 42454272 (2012).

17. Shin, W., Yasuda, T., Watanabe, G., Yang, Y. S. & Adachi, C. Self-organizing mesomorphic diketopyrrolopyrrole derivatives for efcient solution-processed organic solar cells. Chem. Mater. 25, 25492556 (2013).

18. Graham, K. R. et al. Improved performance of molecular bulk-heterojunction photovoltaic cells through predictable selection of solvent additives. Adv. Funct. Mater. 22, 48014813 (2012).

19. Wang, H. et al. The role of additive in diketopyrrolopyrrole-based small molecular bulk heterojunction solar cells. Adv. Mater. 25, 65196525 (2013).

20. Graham, K. R. et al. Tailor-made additives for morphology control in molecular bulk-heterojunction photovoltaics. ACS Appl. Mater. Interfaces 5, 6371 (2013).

21. Zhou, J. et al. Small molecules based on benzo[1,2-b:4,5-b0]dithiophene unit for high-performance solution-processed organic solar cells. J. Am. Chem. Soc. 134, 1634516351 (2012).

8 NATURE COMMUNICATIONS | 6:6013 | DOI: 10.1038/ncomms7013 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2015 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms7013 ARTICLE

22. Wei, G., Wang, S., Sun, K., Thompson, M. E. & Forrest, S. R. Solvent-annealed crystalline squaraine: PC70BM (1:6) solar cells. Adv. Energy Mater. 1, 184187 (2011).

23. Sun, K. et al. The role of solvent vapor annealing in highly efcient air-processed small molecule solar cells. J. Mater. Chem. A 2, 90489054 (2014).

24. Wessendorf, C. D. et al. Efciency improvement of solution-processed dithienopyrrole-based A-D-A oligothiophene bulk-heterojunction solar cells by solvent vapor annealing. Adv. Energy Mater. 4, 1400266 (2014).

25. Krebs, F. C. Fabrication and processing of polymer solar cells: a review of printing and coating techniques. Sol. Energy Mater. Sol. Cells 93, 394412 (2009).

26. Li, W. et al. Efcient small bandgap polymer solar cells with high ll factors for

300 nm thick lms. Adv. Mater. 25, 31823186 (2013).

27. Li, G., Shrotriya, V., Yao, Y. & Yang, Y. Investigation of annealing effects and lm thickness dependence of polymer solar cells based on poly(3-hexylthiophene). J. Appl. Phys. 98, 043704 (2005).

28. Murphy, L., Hong, W., Aziz, H. & Li, Y. Organic photovoltaics with thick active layers (B800 nm) using a high mobility polymer donor. Sol. Energy Mater. Sol.

Cells 114, 7181 (2013).29. Zeng, L., Tang, C. W. & Chen, S. H. Effects of active layer thickness and thermal annealing on polythiophene: fullerene bulk heterojunction photovoltaic devices. Appl. Phys. Lett. 97, 053305 (2010).

30. Wang, N., Chen, Z., Wei, W. & Jiang, Z. Fluorinated benzothiadiazole-based conjugated polymers for high-performance polymer solar cells without any processing additives or post-treatments. J. Am. Chem. Soc. 135, 1706017068 (2013).

31. Zhou, J. et al. Solution-processed and high-performance organic solar cells using small molecules with a benzodithiophene unit. J. Am. Chem. Soc. 135, 84848487 (2013).

32. Kline, R. J. et al. Critical role of side-chain attachment density on the order and device performance of polythiophenes. Macromolecules 40, 79607965 (2007).

33. Kim, M. S., Kim, B. G. & Kim, J. Effective variables to control the ll factor of organic photovoltaic cells. ACS Appl. Mater. Interfaces 1, 12641269 (2009).

34. Bredas, J. L., Norton, J. E., Cornil, J. & Coropceanu, V. Molecular understanding of organic solar cells: the challenges. Acc. Chem. Res. 42, 16911699 (2009).

35. Schmidt-Mende, L. et al. Self-organized discotic liquid crystals for high-efciency organic photovoltaics. Science 293, 11191122 (2001).

36. Goel, M. & Jayakannan, M. Supramolecular liquid crystalline p-conjugates: the role of aromatic p-stacking and van der Waals forces on the molecular self-assembly of oligophenylenevinylenes. J. Phys. Chem. B 114, 1250812519 (2010).

37. Kim, B.-G. et al. A molecular design principle of lyotropic liquid-crystalline conjugated polymers with directed alignment capability for plastic electronics. Nat. Mater. 12, 659664 (2013).

38. Hanna, J.-i., Ohno, A. & Iino, H. Charge carrier transport in liquid crystals. Thin Solid Films 554, 5863 (2014).

39. McPhillips, T. M. et al. Blu-Ice and the distributed control system: software for data acquisition and instrument control at macromolecular crystallography beamlines. J. Synchrotron Radiat. 9, 401406 (2002).

40. Guo, X., Puniredd, S. R., Baumgarten, M., Pisula, W. & Mllen, K. Benzotrithiophene-based donor-acceptor copolymers with distinct supramolecular organizations. J. Am. Chem. Soc. 134, 84048407 (2012).

41. Guo, X. et al. Polymer solar cells with enhanced ll factors. Nat. Photon. 7, 825833 (2013).

42. Lin, Y. et al. Small-molecule solar cells with ll factors up to 0.75 via a layer-by-layer solution process. Adv. Energy Mater. 4, 1300626 (2014).

43. Murphy, L., Hong, W., Aziz, H. & Li, Y. Inuences of using a high mobility donor polymer on solar cell performance. Org. Electron. 14, 34843492 (2013).

44. Liang, Y. et al. For the bright future: bulk heterojunction polymer solar cells with power conversion efciency of 7.4%. Adv. Mater. 22, E135E138 (2010).

45. Kyaw, A. K. K. et al. Intensity dependence of currentvoltage characteristics and recombination in high-efciency solution-processed small-molecule solar cells. ACS Nano 7, 45694577 (2013).

46. Qin, T. et al. Tailored donoracceptor polymers with an AD1AD2 structure: controlling intermolecular interactions to enable enhanced polymer photovoltaic devices. J. Am. Chem. Soc. 136, 60496055 (2014).

47. Kumar, R. J., MacDonald, J. M., Singh, T. B., Waddington, L. J. & Holmes, A. B. Hierarchical self-assembly of semiconductor functionalized peptide a-helices and optoelectronic properties. J. Am. Chem. Soc. 133, 85648573 (2011).

Acknowledgements

This research was supported by the Victorian Organic Solar Cell Consortium with funding from the Victorian Government (DSDBI) and the Australian Renewable Energy Agency (ARENA) through grants 2-A018, 1-GER001 and the ARENA funded Australian Centre for Advanced Photovoltaics. The research was also supported by the Ministry of Education, Singapore (R-284-000-113-112). W.W.H.W. is supported by an ARC Future Fellowship (FT130100500). We acknowledge the Monash Centre for Electron Microscopy (MCEM) at Monash University, the MX2 beamline at the Australian Synchrotron, CSIRO, Department of Materials Science & Engineering (DMSE) at National University of Singapore (NUS) and Solar Energy Research Institute of Singapore (SERIS) for sharing their facilities with us. We acknowledge Dr Doojin Vak at CSIRO, Dr Xi-Ya Fang at MCEM, Henche Kuan, Yeow Koon Liew, Drs Lei Chen and Pengcheng Li at DMSE NUS and Fang Jeng Lim at SERIS for experiment assistance and in-depth discussions.

Author contributions

K.S. fabricated the solar cell device and characterized the lm and device properties using POM, XRD, AFM, TEM, HAADF STEM and SCLC. Z.X. designed and performed materials synthesis and characterization including NMR, MS, CV and UVvis absorption spectra for solution and thin lm. S.L. scaled up the synthesis, grew single crystals and measured DSC, TGA, PL and solubility. W.Z. and W.P. fabricated OFET, measured and analysed OFET, 2D-XRD and GIWAXS data. E.H. constructed TEM tomograms and computer models. J.M.W. solved the crystal structure using XRD acquired by R.M.W. at Australian Synchrotron. J.S. reproduced the OPV results. K.S., Z.X. and S.L. prepared the manuscript. All authors discussed the results and commented on the manuscript. J.O., W.W.H.W., A.B.H. and D.J.J. supervised the project and revised the manuscript.

Additional information

Accession codes: The crystallographic information le for BTR has been deposited with Cambridge Crystallographic Data Centre, and signed to CCDC code 1029304.

Supplementary Information accompanies this paper at http://www.nature.com/naturecommunications

Web End =http://www.nature.com/ http://www.nature.com/naturecommunications

Web End =naturecommunications

Competing nancial interests: The authors declare no competing nancial interests.

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions

Web End =http://npg.nature.com/ http://npg.nature.com/reprintsandpermissions

Web End =reprintsandpermissions/

How to cite this article: Sun, K. et al. A molecular nematic liquid crystalline material for high-performance organic photovoltaics. Nat. Commun. 6:6013 doi: 10.1038/ ncomms7013 (2015).

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Web End =http://creativecommons.org/licenses/by/4.0/

NATURE COMMUNICATIONS | 6:6013 | DOI: 10.1038/ncomms7013 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 9

& 2015 Macmillan Publishers Limited. All rights reserved.