Correlating photovoltaic properties of PTB7-Th:PC71BM blend to photophysics and microstructure as a function of thermal annealing

Please note that technical editing may introduce minor changes to the text and/or graphics, which may alter content. The journal’s standard Terms & Conditions and the ethical guidelines, outlined in our author and reviewer resource centre, still apply. In no event shall the Royal Society of Chemistry be held responsible for any errors or omissions in this Accepted Manuscript or any consequences arising from the use of any information it contains. Accepted Manuscript


Introduction
Organic solar cells are a very promising and versatile photovoltaic technology because of their simple fabrication, flexible and thin form, tunablility of absorption properties and robust performance under a wide range of lighting conditions [1][2][3][4] . All these attributes make them suitable for a range of applications, such as being placed on irregularly shaped products; hence they are ideal energy sources for portable battery powered consumer electronics. Over a decade now, the focus of research for organic solar cells has been on increasing the power conversion efficiency (PCE). Single junction polymer solar cells with power conversion efficiencies of over 10 % have now consistently been achieved over the last two years -a longstanding target for commercialisation [5][6][7][8][9][10][11][12] . This has been made possible by directed development, research and systematic optimisation of low bandgap conjugated polymers, processing methods and interface engineering. The recently developed promising low bandgap conjugated polymers in this regard are PTB7-Th 8 , PffBT4T-2OD 9 , PBDB-T 10 , blended with PC 71 BM to realise a near ideal nanoscale phase separation for efficient photovoltaic properties. OPV device architectures incorporating plasmonic nanostructures to maximise light absorption 13 and doping of the charge transporting layers 11 have further enhanced the power conversion efficiency of these blends. However, as discussed in the Unification Challenge 14 -which discusses realizing the objective of achieving stable, efficient, low cost polymer solar cells -the stability of organic solar cells should also be on par with power conversion efficiency and processability, otherwise the progress towards the application of the technology will remain slow. Though the simple solution processing and the possibility of roll to roll fabrication can ensure the cost effectiveness of this photovoltaic technology, stability is a great concern for organic solar cells. Accordingly, investigation of the stability of organic photovoltaic (OPV) blends is recently receiving more attention and standard stability testing protocols are now being implemented [15][16][17][18][19] . Among the highly efficient narrow bandgap donor polymers, the benzodithiophene (BDT) derivative based PTB7-Th mixed with fullerene PC 71 BM acceptor has reproducibly demonstrated power conversion efficiency greater than 10 % by different research groups 7,10,11,20 and hence is an interesting candidate for further thermal processing and stability investigation.
The stability of organic solar cells is often studied by measuring the effect of chemical/photo/thermal processes on the power conversion efficiency. Chemical degradation of organic solar cells, due to water and/or oxygen from ambient air reacting with the active layer, is reported for systems like P3HT:PCBM 21,22 and now needs to be understood and extended to OPV blends which have demonstrated lab scale PCE of 10 % and above. The inherent acidity of PEDOT:PSS, the most commonly used hole transport layer material in standard OPV architecture (causes it to react with ITO and the low work function of electrodes such as Al (making it react with ambient oxygen) are two important factors contributing towards the OPV performance degradation 14 . Inverted device architectures with thin oxide layers functioning as charge transport layers and high work function metal electrodes as anodes have been developed to improve stability and performance 14 . The photostability of PTB7-Th:PC 71 BM blends has been reported very recently and has shown that incomplete removal of DIO additive accelerates the photo-degradation of the OPV devices 23 . The structural ageing or degradation of nanoscale morphology of P3HT:PC 61 BM and PCDTBT:PC 71 BM organic solar cells was previously reported by Schaffer et al under in-operando conditions for a prolonged period of ~ 18 h. 24,25 To analyse the role of thermal effects on OPV blends, two aspects should be taken into consideration: the first is thermal annealing as a processing method for improving the photovoltaic performance, and, the second is the use of thermal annealing as a stressing method to probe the operational stability of an OPV blend. Considering the former aspect: thermal annealing can modify the nanoscale phase separation and crystallization of OPV blends, which impacts the efficiency of charge generation, extraction and operational stability. The optimised molecular packing and morphology of both the acceptor and donor is essential for efficient exciton transport (diffusion) and maximum charge carrier mobility to the electrodes. This has been demonstrated as an efficient method for systems such as P3HT:PC 61 BM, where polymer P3HT is semi-crystalline. 26,27 In the case of poorly ordered polymers such as PTB7-Th, characterised by weak intermolecular interaction, the suitability or applicability of such post processing techniques on the nano-scale morphology, phase behaviour, optical and electrical properties of PTB7-Th:PC 71 BM blend thin films has not been reported. In order to facilitate further increase in efficiency and applicability, it is vital that the factors governing the performance of these high efficiency OPV blends are identified and understood. A deeper understanding of thermal annealing effects on photovoltaic performance parameters of fill factor (FF), short-circuit current density (Jsc), open circuit voltage (Voc), and correlating these parameters with polymer orientation, packing structure, photophysics and phase separation of these blends, will help accelerate the understanding of how to manipulate the molecular arrangement of BHJ systems to achieve improved stability without compromising the power conversion efficiency. Though the focus of our work is thermal processing induced changes in the photovoltaic properties of PTB7-Th:PC 71 BM blends, the changes following the heating are also relevant to for long term thermal stability.
In this work, we have investigated the photovoltaic properties of PTB7-Th:PC 71 BM blend as a function of thermal annealing over a range of temperatures (from room temperature to 150 o C) and how the morphological, structural, charge transport and photophysical properties are contributing to the observed effect. An inverted device architecture was employed to avoid degradation of the ITO/PEDOT:PSS or Ca/Al contacts. Ex-situ thermal annealing process was applied only to the active layer of the PTB7-Th:PC 71 BM blend films, and not to the full organic solar cells. This was to exclude any influence of degradation of charge selective layers or metal contacts on the photovoltaic properties of PTB7-Th:PC 71 BM blends during the thermal annealing process. With increase in thermal annealing temperature from room temperature to 150 o C, the power conversion efficiency of the PTB7-Th: PC 71 BM blend based OPV devices decreases from 9.1 % to 6.85 %, mainly due to a drop in Jsc and FF. However, the open circuit voltage slightly improves from 0.77 V to 0.78 V. Since the active layer consists of a blend of a conjugated polymer and a small molecule, the phase separation and crystallization processes are complex and hence a range of characterisation methods has been employed to probe the possible reasons for the efficiency drop. The developed understanding will contribute towards enhancing the operational stability of OPVs based on this high efficiency blend system. Preliminary results on improved thermal stability of PTB7-Th:ITIC blend system, where ITIC is used as a non-fullerene acceptor, are also demonstrated.

Experimental section Preparation of ZnO electron transporting layer
The electron transporting layer was an amorphous ZnO (a-ZnO) thin film of ~ 25 nm thickness 28  mbar base pressure) in the glove box for thermally evaporating the hole transporting layer of MoO x (7 nm) and anode Ag (100 nm) using a shadow mask. The active area of the devices was 0.08 cm 2 (4 mm × 2 mm).

Characterisation of solar cells
After the electrode deposition, the devices were encapsulated with a UV optical adhesive and a glass coverslip. The current-voltage characteristics were determined under an illumination intensity of 100 mW∕cm 2 in air using an air mass 1.5 global (AM 1.5G) Sciencetech solar simulator and a Keithley 2400 source-measure unit. The illumination intensity was verified with a calibrated monosilicon detector and a KG-5 filter. The external quantum efficiency (EQE) measurements were performed at zero bias by illuminating the device with monochromatic light supplied from a Xenon arc lamp in combination with a dual-grating monochromator. The number of photons incident on the sample was calculated for each wavelength by using a silicon photodiode calibrated by national physical laboratory (NPL).

Please do not adjust margins
Please do not adjust margins

Characterisation of PTB7-Th:PC 71 BM active layer blend
The surface morphology of the of PTB7-Th:PC 71 BM films was characterised using atomic force microscopy (AFM). AFM images were obtained with a Bruker MultiMode 8 instrument in the tapping mode. NANOSENSORS™ PPP-NCSTR Si cantilever tips with force constant of 6-7 Nm −1 were used as AFM probes. Steady state absorption spectra of the blend films were recorded using a Varian Cary 300 spectrophotometer for the wavelength range of 300-800 nm. For this the active layer blend was deposited under identical conditions to those used for OPV devices, but on a fused silica disc. Time resolved photoluminescence (TRPL) spectra in the pico-nano seconds regimes were acquired with a synchroscan streak camera (C6860 from Hamamatsu). The excitation wavelength was from the frequency-doubled output of a Ti:sapphire laser, giving 400 nm, 100 fs full-width half-maximum pulses at 80 MHz. Samples for streak camera measurements were held under an active vacuum of ~10 −5 mbar. The excitation wavelength used for the time -resolved PL measurements was 515 nm.
Grazing incidence wide angle X-ray scattering measurements were performed at Stanford Synchrotron Radiation Lightsource beamline 11-3. 12.735 keV photons were used with 2D scattering patterns recorded on a MARCCD detector. The beam center and sample-to-detector distance was calibrated using a Lanthanum Hexaboride standard. Scattering patterns were recorded as a function of the X-ray angle of incidence, with an angle of incidence of 0.14° or 0.2°. The former angle was chosen close to the critical angle to maximise intensity, whereas the latter angle was chosen well above the critical angle to minimise the effect of alignment error. Data acquisition times of 60 s were used, and 3 to 10 exposures were averaged to improve SNR. X-ray diffraction data were expressed as a function of the scattering vector, q, which is defined as (4π/λ) sin θ, where 2θ is the scattering angle and λ is the wavelength of the incident radiation. From Bragg's Law, q=2 π/d, where d is a crystal repeat length (or in this case, intermolecular spacing).
Hole-only or electron-only diodes were fabricated using the architectures ITO/PEDOT:PSS/PTB7-Th: PC 71 BM/MoO 3 /Ag for holes and Al/ PTB7-Th: PC 71 BM/LiF/ Al for electrons. The mobility was extracted by fitting the current density-voltage curves using the space charge limited current method (SCLC).

Photovoltaic properties of the blend films
The photovoltaic properties of the PTB7-Th:PC 71 BM solar cells annealed at different temperatures are shown in Figure 1. The illuminated J-V curve (dark J-V curves in inset) in Figure 1a shows a clear trend of decreasing short circuit current density (Jsc) and fill factor (FF) resulting in an overall drop in power conversion efficiency from 9.1 to 6.85 % with increase in thermal annealing temperature from room temperature to 150 o C. In contrast, the open circuit voltage (Voc) shows a small increase. The FF drops from 71 to 62% with a corresponding decrease in Jsc from 16 to 13 mA/cm 2 . Previous reports on thermal annealing of other efficient OPV blend systems of PCDTBT:PC 71 BM, and PBDTTPD:PC 71 BM, also displayed a decreasing PCE with thermal annealing 29,30 . However, for the same temperature range, the drop in photovoltaic parameters (FF, J sc and V oc ) was different from that observed for PTB7-Th:PC 71 BM. This indicates the need to study the thermal processing effects and hence thermal stability of each efficient OPV blend separately. The shunt resistance decreases monotonically with increase in thermal annealing of PTB7-Th:PC 71 BM blend, and can account for the increased leakage current seen in dark J-V characteristics of high temperature thermally annealed blends [ Table 1 and inset of Figure 1a]. Possible reasons for the reduction in photocurrent density and FF can be reduced exciton dissociation, poor charge separation efficiency, or decreased charge extraction efficiency and the increased recombination losses of the photogenerated charge carriers. In order to investigate the cause of the reduction in these photovoltaic properties, detailed microstructural and photophysical characterisation was performed as detailed below.  For each annealing temperature, the first row shows average photovoltaic parameters of 18 devices (and the standard deviation) and the second row shows the best device.
To identify the potential loss mechanisms that limit the overall PCE of OPV devices as a function of thermal annealing, light intensity (I) dependent J-V curves were measured and analysed. The relation between short circuit current and light intensity follows a power law relation

FF (%)
Light Intensity Figure 2(b) shows V oc variation vs logarithm of light intensity for PTB7-Th:PC 71 BM blends as a function of thermal annealing. It is interesting to note that up to 100 °C, the values remain close to unity, then starts to deviate from unity (1.16) for the blend annealed at 120 o C and is highest for 150 °C annealed sample where the slope gives equal to 2.15. Thus, the main non-geminate recombination loss is bimolecular for the blends annealed up to a temperature of 120 o C and the contribution of trap assisted recombination is dominating for 150 °C annealed blend. Any nongeminate recombination event is effectively eliminating charge carriers that could otherwise contribute to a photocurrent. Hence the drop in J sc with increase in thermal annealing, especially at higher temperatures can be attributed to the loss of photogenerated e-h pairs due to non-geminate recombination.
In Figure 2c, the variation of FF as a function of light intensity is shown. For the blend films annealed up to 120 o C, the FF increases with lowering of light intensity, and for the blends annealed at 150 o C, the FF follows a monotonic decrease with decrease in light intensity. The drop in FF at low light intensities suggests that carrier loss by recombination is significant in PTB7-Th:PC 71 BM blends annealed at a temperature of 150 o C. The factors that lead to nongeminate recombination are inconsistent pathways of donor/acceptor islands due to phase separation, electron/hole mobility mismatch and increased energetic disorder. Since these factors are correlated to each other, a careful and systematic investigation is carried out to elucidate the contribution of each factor to the observed PCE drop.

Surface morphology of the blends by atomic force microscopy (AFM)
To identify the morphological stability of the PTB7-Th:PC 71 BM blends to thermal annealing, morphological features were characterised as a function of different annealing temperatures using atomic force microscopy. Time resolved photoluminescence quenching was used as a complementary probe for estimating the domain size 33,34 and is described at the end of the next section. The morphology of the donor/acceptor network is crucial in maximising efficiency as the carrier lifetime is largely controlled by nanoscale phase morphology between the donor and acceptor materials. Figure S1 and Figure 3(a)-(d) show the AFM height and phase images of PTB7-Th:PC 71 BM films annealed at different temperatures respectively. The RMS surface roughness [ Figure 3e], estimated from statistical distribution of surface height from an area of 10 x 10 µm 2 , shows an overall increasing trend. At room temperature, the OPV blend has a surface roughness of ~ 1.8 nm, which increases to ~ 3.2 nm for 150 °C thermally annealed blend.
During the intermediate temperature range of 75 • C to 120 • C, the surface roughness remains more or less the same. This could be related to the phase separation of the donor/acceptor blend as the annealing temperature is raised from room temperature to 150 °C , increasing the molecular mobility.
As shown in the AFM phase images (Figure 3), the contrast as well as the size of the bright spots (which correspond to PCBM aggregates due to its high elastic modulus) is increased with thermal annealing temperature indicating that there is an enhancement of PCBM aggregation/phase separation length scale. The phase images of the blends annealed at 75 and 120 °C shows high-contrast interpenetrating channels , while that of 100 and 150 °C show a more gradual and random structure with interspersed  Figure S2. As seen, the average domain size increases from 16 to 20 nm and the FWHM of normal distribution increases from 8 to 10.4 nm as the annealing temperature is increased to 150 o C. The slight increase in FWHM indicates the existence of domains with a varied range of sizes due to thermal annealing at high temperatures. This clearly demonstrates that, donor and/or acceptor molecular motion happens with thermal annealing and corresponding morphological changes are occurring.
Large-scale, unfavorable phase separation in BHJ blends would reduce the probability of photo-generated excitons to arrive at a donor-acceptor interface for their dissociation prior to recombining compared to the system in which donor and acceptor phases are homogeneously distributed. This would reduce the photogenerated current (and hence the J sc ). Moreover when free photogenerated electrons and holes transport through the isolated domains of donor and/or acceptor, the probability of recombination with trapped carriers within the hopping/tunneling distance also increases. Thus the reduced J sc and FF of the OPV blends with increase in thermal annealing temperature can be attributed to the reduced exciton dissociation and increased recombination losses due to inconsistent and unfavorable phase separation. To validate this inference drawn from morphological measurements, detailed photo-physical characterisations of the blend films were performed as described below.

Photophysics of the PTB7-Th:PC 71 BM blend as a function of thermal annealing
In order to get insight on how the thermal annealing influences the optical properties and exciton/charge dynamics of the PTB7-Th:PC 71 BM blend, various spectroscopic techniques such as UV-Vis absorption, steady state and time resolved PL spectroscopy are applied. The UV-Vis absorption spectrum shown in Figure 4(a) reveals that due to thermal annealing, PTB7-Th:PC 71 BM blend does not exhibit any strong blue shift or redshift of the absorption maxima. This observation clearly reveals that the photocurrent drop as a function of thermal annealing is not due to a change in absorption properties of the blend films. To investigate the photoinduced charge transfer between the donor and acceptor molecules, static and dynamic PL spectra of pristine and blend films are recorded. The steady state photoluminescence from pristine PTB7-Th, PC 71 BM and their blends as a function of thermal annealing are shown in Figure 4(b). The PL emission from PTB7-Th is mainly peaked at 760 nm and from PC 71 BM at 720 nm, matching well with the previous reports on PL emission 35,36 for these materials. The PL spectrum of the blend is constituted by combination from both PC 71 BM and PTB7-Th and no additional peak due to a CT state is observed in the long wavelength region for the analyzed spectral range. This is in contrast to PCDTBT:PC 71 BM blend where a CT state emission peak is clearly observed. The absence of CT state emission indicates that the CT intermediate state density is too low or that the energy is too close to the PL peak. PL quenching provides direct evidence for exciton dissociation, and the degree of PL quenching reflects the efficiency of the exciton charge separation which in turn influences the J sc value of the organic solar cells. 37 A comparison of the PL intensity of the PTB7-Th:PC 71 BM blend prepared at room temperature and 150 °C is given in Figure S3(a). A reduced exciton quenching efficiency for the blend films annealed at 150 o C is observed in comparison to the room temperature processed blend films. To further confirm this, time resolved PL measurements of the neat and blend films were performed. The PL decay dynamics for the neat PTB7-Th and neat PC 71 BM are shown in Figure S3 (b). For the case of neat PC 71 BM, the PL decay is mono-exponential with lifetime of ~ 640 ps (corresponding fit to PL decay is given in Figure S3 (b)), whereas the PL decay of pristine PTB7-Th is bi-exponential with a 1/e lifetime of ~ 280 ps. As shown in Figure 4 (b), both PTB7-Th and PC 71 BM contributed to PL of blend in contrast to previous studies of PTB7 and PC 71 BM blends where only emission from PC 71 BM was observed 33 . In order to separate the PTB7-Th and PC 71 BM emissions, we measured the PL excitation spectra of PTB7-Th:PC 71 BM blend film at different emission wavelengths (see Figure  S4) and observed the contributions from both donor and acceptor molecules. We found that for λ em > 815 nm, emission was mainly from PTB7-Th and for λ em < 720 nm, only PC 71 BM was emitting. In between these wavelengths, both materials emit. Therefore for our analysis we selected PL decays between 670 and 720 nm for the PC 71 BM and between 820 and 900 nm for PTB7-Th.
The corresponding PL decays as a function of thermal annealing for the case of PC 71 BM and PTB7-Th are shown in Figure 4c and 4d respectively. The PL decays of blend films in both cases are faster compared to decays of their corresponding neat films and this is likely due to charge transfer or resonant energy transfer between the well intermixed donor and acceptor molecules in the blend. We extracted quenching efficiency using the previously reported fitting approach 33,38 ࣘ ൌ െ ‫‬ ‫‬ where and are PL intensity decays of blend and neat films respectively. The obtained values are given in Table S1. The reduced exciton quenching (both in steady state and time resolved PL) of the blend films annealed at 150 °C is accompanied by a reduction in J sc from the thermally annealed blends. The reduced exciton quenching can be due to the increase in length scale of phase separation of the donor-acceptor in the blend. This observation is in agreement with the findings from AFM phase and height images shown in Figure 3 and Figure S1. Since all blend films have bi-exponential decays (very fast decay at an early time and slower decay at a later time), the lifetimes were estimated from PL decay when they fall to 1/e of their initial value and are given as a function of thermal annealing in Table S2. The corresponding fitted PL decay curves are given in Figure S5 (a) and (b). The 1/e lifetime in both cases increased systematically with increase in thermal annealing and indicates decreasing exciton dissociation efficiency at the polymer: fullerene interfaces after the photoexcitation. This is likely due to an increased domain size resulting in fewer singlet excitons reaching a hetero-interface during their migration. This interpretation is supported by the Please do not adjust margins Please do not adjust margins increase in feature sizes and surface roughness in AFM seen in Figures 3 and S1. 33,35 In order to get quantitative information about the size of PC 71 BM domains, we used a similar approach to Hedley et al 33 and used the relation, Where is the diffusion coefficient, is the ratio of the PL of the blend (detected between 670 nm and 720 nm) and the PL of the neat PC 71 BM film, is radius of the PCBM domain, and is time. For our analysis we used the previously reported value of ൌ . ൈ ି / for PC 71 BM. 33 At the earliest times, fast decay is due to direct energy transfer. Therefore we fitted the ratio of PL decays of blend and pristine film in the region > 8 ps (where PL decay is mainly due to exciton diffusion) and found that the size of pure domains of PC 71 BM increased with annealing temperature (Table S2 and Figure S6). Compared to the domain size estimated from the AFM height images ( Figure S2), the domain size estimated from the PL decay studies is smaller and this difference in size estimation could be due to the difference in the assumptions and resolutions achievable with the two different methods.
Thus both steady state and time resolved PL spectroscopic measurements imply a reduced exciton dissociation efficiency due to increased domain size in the blend films with increase in thermal annealing temperature. This can be due to the coarse phaseseparation of the donor-acceptor blend films with increase in thermal annealing temperature, reducing the interfacial area between the donor PTB7-Th and acceptor PC 71 BM leading to a suppression of density of initially photo-generated charge carriers. Thus the decrease in short circuit density of the PTB7-Th:PC 71 BM solar cells with increase in thermal annealing temperatures can be due to the increased domain size of the components in the blend films. Previously Guo et al 39 have reported similar observation of reduced interfacial boundaries in thermally annealed PTB1:PCBM films thereby dropping the solar cell PCE by ~ 60%. The main factor that contributed to the drop in PCE was the lowered Jsc because of the reduced exciton splitting in annealed blend films due to the larger PCBM domains.
So far the thermal annealing induced morphological and photophysical properties of PTB7-Th:PC 71 BM blend films and their correlation to the observed photovoltaic performance, with a focus on short circuit current density, were discussed. In the following section, the molecular packing and the crystallinity of the blend films due to thermal annealing and its correspondence to other photovoltaic properties of Voc and FF are discussed.

Structural properties by GIWAXS data analysis
To identify how the thermal annealing changes the molecular packing of both the donor and the acceptor, GIWAXS studies were performed on thermally annealed blend films. Figure 5 shows the GIWAXS 2D patterns of the PTB7-Th:PC 71 BM blend as a function of thermal annealing. After applying necessary geometry corrections, we can identify three peaks typical of PC 71 BM (at Q=0.7, 1.36, and 1.96 Å -1 ) and peaks associated with the polymer alkyl stacking peak (Q=0.32 Å -1 ) [corresponding to (100) Bragg reflection] and the polymer π-stacking peak (Q=1.6 Å -1 ) 30,40 [corresponding to (001) Bragg reflection]. The alkyl stacking peak is more intense in the inplane direction, and the π-stacking peak is prominent in the out-ofplane direction, showing a preferential face-on orientation. With increase in thermal annealing, the alkyl peak becomes more intense relative to the PCBM diffraction, suggesting a slight increase in ordering of PTB7-Th donors. No orientation change of the donor molecules with increase in thermal annealing is observed. A previous study on X-ray scattering of P3HT:PCBM blend and photocurrent has revealed that J sc depends more on the local ordering of the PCBM than on the π-stacking of the donor molecule 41 . Even though the degree of crystallinity of the PTB7-Th donors increases with thermal annealing, in contrast to P3HT which shows greatly improved ordering on annealing, the J sc of the devices drops systematically with increase in annealing temperature. 42 The PC 71 BM scattering is observed as broad diffusive rings (at 0.6 and 1.4 A°-1 ), indicating a disordered amorphous or nanocrystalline state. The PCBM ring intensity increases with temperature indicating increase in aggregation of PCBM. This observation is in agreement with AFM height and phase imaging in Figure 3 and Figure S1, and photophysical analysis shown in Figure 4, where the increase in phase purity and domain size with increase in thermal annealing temperature is clearly visible. Normalising the GIWAXS data by the second PCBM scattering peak reveals that the polymer alkyl diffraction grows faster than the PCBM aggregates [ Figure S7]. This could be due to PCBM diffusion out of mixed regions, forming weak PCBM aggregates and allowing the polymer to crystallize strongly. The improved photovoltage observed for the thermally annealed PTB7-Th:PC 71 BM blend can be related to the improved crystallinity of the PTB7-Th donor molecule which would reduce the energetic disorder/defect states, resulting in downshifting of the hole fermi level 43 . Recently, Gonzalez et al 44 have reported direct evidence for how the increased crystallinity of the donor polymer (P3HT) in the active layer can contribute to enhanced photovoltage in a BHJ OPV by using an in-operando combination of GIWAXS and I−V tracking.

Space charge limited current (SCLC) mobility measurements
The photovoltaic properties of the polymer blend films are highly correlated to the charge transport properties. In order to obtain high fill factor for OPVs, balanced and high charge carrier mobilities are required for holes and electrons. To study how the thermal annealing is affecting the charge transporting properties of the PTB7-Th:PC 71 BM blend, the hole and electron mobility in the blend films are estimated using the space charge limited current (SCLC) method. The structure of hole only devices are ITO/PEDOT:PSS/PTB7-Th:PC 71 BM/MoO 3 /Ag and Al/PTB7-Th:PC 71 BM/LiF/Al respectively ( Figure S8). The J-V curves of the devices made with blend films annealed at different temperatures are shown in Figure 6. The mobility of holes and electrons are estimated using the Mott-Murgatroyd relation 45 : where is the steady state current density function of applied voltage ; is the film thickness, is the dielectric constant of the organic semiconductors (set as 3.5), and represents the mobility in the limit of zero electric field, the built in voltage and ࢽ the electric field dependent factor. As given in Table 2, with increase in thermal annealing there is no considerable variation in hole/electron mobility. However, for the blend films annealed at 150 °C, the charge mobility imbalance is highest. This charge mobility imbalance can cause space charge effects in the active layer blend impeding the charge extraction processes and promoting the trap-assisted recombination losses. 46 This is in accordance with the inference drawn from the recombination losses studies shown in Figure 2b, where the 150 °C annealed blend films showed the highest recombination losses. Thus the drop in Jsc and FF observed for the PTB7-Th:PC 71 BM blends with increase in thermal annealing temperatures can be summarized as due to reduced exciton dissociation due to large length scale phase separation and increased recombination losses.   24,25 where the authors employed GISAXS to probe the microstructural evolution of the BHJ active layer. Though solar cell performance loss is attributed to morphological instability, depending on the active layer blend, different responses in photovoltaic performance parameters and microstructure evolution with time were observed. In the case of P3HT : PCBM solar cells, growth of the domain size of the active layer components was observed whilst in PCDTBT:PC 71 BM solar cells, domain size shrank. Hence, for P3HT:PCBM solar cells, the highest drop was observed in Jsc, (due to reduced exciton splitting) with smaller changes in Voc and FF. In contrast, for PCDTBT : PC 71 BM, the shrinking leads to a loss of connectivity in the interpenetrating network and hence a drop in FF played the largest role in solar cell performance degradation. In the present case of thermally annealed PTB7-Th: PC 71 BM blend based solar cells, the increased domain size caused a reduction in exciton splitting as discussed in the photophysics section, and hence a higher drop is observed in Jsc (16 %) than in FF (11 %). The increased crystallinity of the PTB7-Th and small effects of thermal annealing on electron/hole mobility are possible reasons for the relatively low drop in FF. Thus, while discussing the performance degradation of OPVs due to morphology instability, these two different morphological aspects should be taken into consideration and based on the response of the PV performance parameters, structural changes can be inferred.

Thermally annealed blends using non-fullerene acceptors
Since, morphological instability was the main reason for the adverse photovoltaic properties of the PTB7-Th:PC 71 BM blends annealed at higher temperatures, we were motivated to test the photovoltaic properties of a non-fullerene acceptor based blend system at similar conditions. The recently reported non-fullerene small molecule acceptor ITIC was selected to investigate the thermal stability of the blend 47,48 . ITIC is based on a bulky sevenring fused core (indacenodithieno[3,2-b]thiophene, IT), with 2-(3oxo-2,3-dihydroinden-1-ylidene)malononitrile (INCN) groups endcapped to the ring, and with four 4-hexylphenyl groups substituted on it [ Figure 8a]. These four rigid 4-hexylphenyl substituents are reported to restrict molecular planarity, aggregation, and large phase separation in non-fullerene BHJ blend films 47 . The OPV device performance as a function of thermal annealing is given in Table 3 and Figure 8. It should be noted that the PCE of this blend system is not as high as the PC 71 BM-based system and more optimisation process are underway. The PCE only drops by 9% of its as-spun value for an increase in thermal annealing from room temperature to 150 °C, while the PC 71 BM based fullerene acceptor blend showed a PCE drop of 30 %. Similarly to PCBM based systems, the EQE spectra (Figure 8(c)) of the PTB7-Th: ITIC OPVs do not show considerable difference as a function of thermal annealing. To compare how the thermal annealing affects the morphology of the PTB7-Th: ITIC blend, atomic force microscopy images of the blend films at RT and annealed at 150 o C were measured and are shown in Figure 9. The AFM height images are shown in Figure 9(a) and (b), whereas the corresponding phase images are shown in Figure 9(c) and 9(d). Homogeneous mixing of the donor and acceptor molecules are evident from these images. No considerable difference in domain size or length scale of phase separation is observed either from the AFM height or phase images due to thermal annealing at 150 o C. This is in accordance with the observed RMS roughness values for the RT and thermally annealed blends, which are respectively 2.6 nm and 2.5 nm. This is in contrary to PTB7-Th:PC 71 BM blend where thermally annealing process increases the RMS roughness [ Figure 3(e)]. The retention of the photovoltaic properties of the PTB7-Th: ITIC blend films even after annealing at 150 o C can be attributed to its morphological stability. Table 3: Photovoltaic performance of PTB7-Th:ITIC blend films as a function of thermal annealing. For each annealing temperature, the first row shows average photovoltaic parameters of 7 devices (and the standard deviation) and second row shows the best device.

Conclusion
The effect of thermal annealing on the photovoltaic properties of a highly efficient BHJ system comprised of PTB7-Th:PC 71 BM is investigated. The decrease in power conversion efficiency, reflected mainly through a drop in Jsc and FF, is identified as due to the suppression of exciton dissociation and increased recombination losses induced by morphological instability with increase in thermal annealing temperature. However, the improved ordering of the PTB7-Th donor molecule with thermal annealing resulted in an enhanced open circuit voltage. By replacing the fullerene PC 71 BM acceptor by non-fullerene acceptor molecule of ITIC, improved morphological stability is observed and hence better retention of the photovoltaic performance as a function of thermal annealing is obtained. This result strengthens the need to develop BHJ systems with morphological stability and stable acceptor molecules along with the development of highly efficient, narrow band gap donor molecules.