Mutual Diffusion of Model Acceptor/Donor Bilayers under Solvent Vapor Annealing as a Novel Route for Organic Solar Cell Fabrication

: The fabrication of bulk heterojunction organic solar cells (OSCs) is primarily based on a phase demixing during solution deposition. This spontaneous process is triggered when, as a result of a decrease in the solvent concentration, interactions between donor and acceptor molecules begin to dominate. Herein, we present that interdiffusion of the same molecules is possible when a bilayers of donors and acceptors are exposed to solvent vapor. Poly(3-hexyl thiophene) (P3HT), and poly[N-9 (cid:48) -heptadecanyl-2,7-carbazole-alt-5,5-(4 (cid:48) ,7 (cid:48) -di-2-thienyl-2 (cid:48) ,1 (cid:48) ,3 (cid:48) -benzothiadiazole) (PCDTBT) were used as donors and two types of fullerene derivatives were used as acceptors: phenyl-C61-butyric acid methyl ester (PC 60 BM) and phenyl-C71-butyric acid methyl ester (PC 70 BM), Secondary ion mass spectrometry depth proﬁling revealed that the interpenetration of donors and acceptors induced by solvent vapor annealing was dependent on solvent vapor and component compatibility. Exposure to chloroform vapor resulted in a complete intermixing of both components. The mutual mixing increased efﬁciency of inverted solar cells prepared by solvent vapor annealing of model donor/acceptor bilayers. These results provide a new means for mixing incompatible components for the fabrication of organic solar cells. bilayer samples prepared by spin-casting of followed by: evaporation, subsequent casting orthogonal PDMS


Introduction
Organic solar cells (OSCs) have attracted significant attention as new sources of renewable energy due to their low cost, light weight and compatibility for large-format solution deposition production technologies. Phenomenal improvement has taken place in recent years, and the power conversion efficiency (PCE) of state-of-the-art devices is comparable with commercially available silicon solar cells [1][2][3][4]. This breakthrough could not have been achieved without new materials [5] and has also been attributed to a better understanding of the factors that affect OSC fabrication and of the relationship between microstructure and charge generation [4,[6][7][8].
The basic principles of organic solar cell operation are similar to those for inorganic devices: excitons created by adsorbed light dissociate and free charge carriers move to external electrodes. Lower degrees of order of organic semiconductors result in higher exciton binding energies, lower charge carrier mobility, and higher absorption coefficients. To promote exciton dissociation, electron acceptors should be placed in the vicinity of exciton formation regions in donor materials with an uninterrupted path to the external electrodes. The simplest structure fulfilling such requirements is a donor and acceptor bilayer sandwiched between electrodes.
This heterojunction concept for organic photovoltaic devices was first investigated by Tang [9], who showed that a layer of acceptor materials placed between cathode and donor layers increased the efficiency of low molecular weight organic semiconductor solar cells. Then, fullerene C 60 with its high ionization potential and high electron mobility was shown to be an excellent candidate as an acceptor material for highly efficient organic solar cells [10]. One of the first fullerene devices had fullerene deposited on top of thin poly(2-methoxy-5-(2 -ethylhexyloxy)-p-phenylene vinylene) film but had no more that 0.04% efficiency. High efficiency light harvesting, exciton dissociation, and charge carrier transport properties have since been developed for these active layers.
A major breakthrough was achieved with the introduction of a bulk heterojunction [11], a structure characterized by an interpenetrating network of large interfaces between continuous donor and acceptor domains. Due to the short lifetimes of excitons in organic semiconductors, their diffusion length is limited to ca. 10 nm [12,13]. This limits the characteristic domain size of nanostructured layers. Fortunately, desired structures spontaneously form due to phase separation phenomena during the solution coating of thin multicomponent films on substrates [14].
Studies on model systems, such as poly(3-hexyl thiophene) (P3HT): phenyl-C61butyric acid methyl ester (PC 60 BM), have shown a number of the factors that influence phase separation allow for the precise control of this spontaneous process to obtain desired layer morphology [8,[15][16][17][18][19][20][21]. A lot of work has been done to optimize the coating conditions of different blends based on solvent composition, constituent blend composition, and film thickness. Additional studies have been conducted to understand the relationships between the obtained structures and solar cell properties. Others have also reported on modifying the properties of these layers. Thermal annealing has been the most often applied post-treatment method, which typically results in increased P3HT:PC 60 BM solar cell efficiency by nearly an order of magnitude [22] due to the enhanced crystallinity of semiconducting materials. The same effect can be achieved by microwave irradiation [23]. Cell efficiency can also be enhanced by reorganizing the nanostructure of the thin films by controlling the drying time, i.e., solvent annealing [24] or solvent vapor annealing [25].
Interactions between solutes and solvents are key parameters that affect the demixing process. Because the number of macromolecule types with the desired electro-optical properties is limited, the choice of solvent for a selected donor and acceptor mixture is crucial for obtaining an optimal OSC structure. Theoretical and experimental studies of phase separation in these ternary systems provide some guidance, but each new donor/acceptor pair would require optimization. This is because spontaneous demixing occurs rapidly and is extremely sensitive to slight changes in fabrication conditions. In their seminal paper, Treat and coworkers [26] revealed thermally induced miscibility in P3HT:PC 60 BM and indicated the possibility of using the mixing process to obtain mutually penetrating domains. Efficient solar cells were fabricated by thermal annealing of P3HT:PC 60 BM bilayers [27,28]. Further studies on thermally induced intermixing in P3HT/PC 60 BM system showed complicated interplay between P3HT crystallization and PC 60 BM interdiffusion [8,21]. The evolution of the P3HT:PC 60 BM solar cell performance is shortly summarized in Table 1. The goal of this study is to apply an intermixing method for the fabrication of bulk heterojunction organic solar cells. For the first time, we show that the interdiffusion of acceptor/donor bilayers can be induced by solvent vapor annealing and is much slower than thermal intermixing. The process can be readily controlled by choice of solvent, which may be more compatible with the first or second component promoting higher mobility. To show this phenomenon can be utilized in fabricating organic solar cells, we prepared a series of devices with P3HT:PC 60 BM and Poly[N-9 -heptadecanyl-2,7-carbazole-alt-5,5-(4 ,7 -di-2-thienyl-2 ,1 ,3 -benzothiadiazole) (PCDTBT): phenyl-C71-butyric acid methyl ester (PC 70 BM) bilayers exposed to chloroform vapor for different periods. Morphology and performance comparisons of devices prepared herein were qualitatively consistent with recently published simulation impact results of vertical composition profiles on OSC operation.
Interdiffusion of bilayers. Silicon wafers (Si-Mat GmbH, Kaufering, Germany) were cut into 10 × 10 mm pieces, cleaned by sonification in toluene for 15 min, rinsed with toluene, and dried with N 2 steam. P3HT and PCDTBT were dissolved in chlorobenzene to form a 15 mg/mL and 5mg/mL solutions, respectively. Before casting, the solutions were heated at 60 • C overnight inside an argon-filled glovebox. The prepared solution was filtered through a 0.45-µm syringe filter and spun-cast onto the silicon pieces at 1000 rpm in the glovebox. Next, the samples were transferred to a vacuum chamber onto which a layer of PC60BM molecules was thermally evaporated. The layer thickness was monitored using a quartz microbalance and was adjusted to 50 nm. The bilayer samples were then transferred back to the glovebox and exposed to different solvent vapors for different time periods.
Solar cell fabrication and characterization. Solar cells were fabricated using a sequential casting routine and thin-film transfer-printing technique [29,30]. Ossila glass substrates (S211) with patterned ITO electrodes were cleaned by sonication in 2-propanol for 15 min and dried under N 2 flux. Substrates were treated with oxygen plasma, and zinc oxide nanoparticles were spun cast onto the substrates at 4000 rpm. The prepared thin films were annealed at 120 • C for 20 min. P3HT was dissolved in 1,2-dichlorobenzene to form a 10 mg/mL solution. The solution was heated at 60 • C overnight inside an argon-filled glovebox and filtered through a 0.45-µm syringe filter prior to spin-casting at 1500 rpm onto prepared ZnO thin films to form an~30 nm layer. PC 60 BM was dissolved in dichloromethane to form an 8 mg/mL solution. Before casting, the solution was heated at 60 • C overnight inside the glovebox. The prepared solution was filtered through a 0.45-µm syringe filter and spun cast at 1500 rpm onto the prepared ZnO/P3HT thin films to form an~90 nm film.
PCDTBT/PC 70 BM bilayers were prepared by thin-film transfer-printing technique as described in ref [29]. First PC 70 BM layer was spun casted from on top of ZnO from 1,2-dichlorobenzene solution (10 mg/mL) filtered through a 0.45 µm syringe filter. PCDTBT layer was prepared on top of silicon wafer covered with thin film of polystyrene sulfonate (PSS) by solution casting from 5 mg/mL solution in chlorobenzene. Next flat elastomeric poly(dimethylsiloxane) (PDMS) stamp was press against the polymer film. After dissolution of the PSS sacrificial layer in a water bath the stamp with grabbed PDCTBT film was placed on top of PC 70 BM layer and gently peeled.
Such prepared bilayers were exposed to chloroform vapor for different durations. Finally, 100 nm silver electrodes were thermally evaporated to form eight 4-mm 2 devices on each substrate. The solar cells were characterized in an inert atmosphere under white light illumination from a solar simulator AM1.5 (Newport, Irvine, CA, USA) at 100 mW/cm 2 ; Thin film characterization. The TOF-SIMS experiments were performed on an ION TOF TOF:SIMS V instrument (Munster, Germany), equipped with a liquid metal ion source and a cesium ion source. Sample depth profile (dSIMS) were obtained in dual beam mode. Briefly, a 500 eV Cs+ ion beam was used to sputter-coat a 500 × 500 µm 2 area and a Bi 3 + 30 keV ion beam was used to analyze a 200 × 200 µm 2 area concentric to the sputtered surface.
An Atomic Force Microscope Agilent 5500 instrument was used to investigate the topography of the free surfaces of the thin films. Set-points and gains were adjusted to each measurement to obtain clear images without noise. Each topography was investigated by collecting images at several randomly chosen areas. The images were processed using Gwyddion 2.39 software.

Results and Discussion
All experiments described herein include bilayers of acceptors on top of polymer donor. Regioregular poly(3-hexyl thiophene) (P3HT) and poly[N-9 -heptadecanyl-2,7-carbazolealt-5,5-(4 ,7 -di-2-thienyl-2 ,1 ,3 -benzothiadiazole) (PCDTBT) were used as model donor while phenyl-C61-butyric acid methyl ester (PC 60 BM) and phenyl-C71-butyric acid methyl ester (PC 70 BM) were used as model acceptor. The polymer layer was spun-coated from dichlorobenzene solution, whereas the top layer was prepared by thermal evaporation [31], subsequent casting from dichloroethane [32] or transfer method [29].The thicknesses of the evaporated PC 60 BM layer were monitored by quartz microbalance. The acceptor/donor bilayers were solvent vapor annealed in a sealed container partially filled with selected solvent. The samples were always kept separate from liquid solvents. Silicon dices were used as substrates when comparing interdiffusion under different solvent vapors; ITO/ZnO transparent conductive electrodes were used for fabricating solar cells. An experimental scheme is depicted in Figure 1. Such prepared bilayers were exposed to chloroform vapor for different durations. Finally, 100 nm silver electrodes were thermally evaporated to form eight 4-mm 2 devices on each substrate. The solar cells were characterized in an inert atmosphere under white light illumination from a solar simulator AM1.5 (Newport, Irvine, CA, USA) at 100 mW/cm 2 ; current-voltage characteristics were acquired by a computer-controlled source meter unit, Keithley 2400 (Keithley Instruments, Solon, OH, USA).
Thin film characterization. The TOF-SIMS experiments were performed on an ION TOF TOF:SIMS V instrument (Munster, Germany), equipped with a liquid metal ion source and a cesium ion source. Sample depth profile (dSIMS) were obtained in dual beam mode. Briefly, a 500 eV Cs+ ion beam was used to sputter-coat a 500 × 500 μm 2 area and a Bi3 + 30 keV ion beam was used to analyze a 200 × 200 μm 2 area concentric to the sputtered surface.
An Atomic Force Microscope Agilent 5500 instrument was used to investigate the topography of the free surfaces of the thin films. Set-points and gains were adjusted to each measurement to obtain clear images without noise. Each topography was investigated by collecting images at several randomly chosen areas. The images were processed using Gwyddion 2.39 software.

Mutual Diffusion of Acceptor/Donor Bilayers
P3HT thin films covered with PC 60 BM layers prepared by thermal evaporation were exposed to three solvent vapors: dichloroethane (DCE), tetrahydrofurane (THF) and chloroform. Compatibility between organic macromolecules and solvents were predicted based on solubility parameters δ. The solubility parameters of model acceptors and donors are: δ = 21-22 MPa 1/2 for PC 60 BM [33], δ = 12.8-13.1 MPa 1/2 for P3HT [34] and δ = 22.8 MPa 1/2 for PCDTBT [35]. The three solvents were selected based on their compatibility with only one component or with both. Dichloroethane, δ = 20.2 MPa 1/2 [36], is a good solvent for fullerene derivatives and does not dissolve P3HT nor PCDTBT and was later used as an orthogonal solvent in sequential casting. Tetrahydrofurane, δ = 19.4 MPa 1/2 [36], dissolves polythiophene well and PC 60 BM only slightly. Chloroform, δ = 19 MPa 1/2 [36], is known to be a good solvent for both PC 60 BM and P3HT and is often used to fabricate bulk heterojunction organic solar cells.
dSIMS profiles of the bilayers prior to and after 1 h exposure to vapors are shown in Figure 2. The depth profiles monitor variation in composition as a function of sputtering time. For all silicon substrate samples, C 4 S − signals (green solid triangles) were used to determine P3HT content, C 2 O − signals (red open circles) for PC 60 BM content, and Si − signals (black solid squares) were used to locate interfaces with substrates. The quantitative conversion of sputtering time to intermixed structure depth of P3HT and PCBM is not straightforward [28]; thus, we presented the profiles as functions of sputtering time.
MPa 1/2 for PCDTBT [35]. The three solvents were selected based on their compatibility with only one component or with both. Dichloroethane, δ = 20.2 MPa 1/2 [36], is a good solvent for fullerene derivatives and does not dissolve P3HT nor PCDTBT and was later used as an orthogonal solvent in sequential casting. Tetrahydrofurane, δ = 19.4 MPa 1/2 [36], dissolves polythiophene well and PC60BM only slightly. Chloroform, δ = 19 MPa 1/2 [36], is known to be a good solvent for both PC60BM and P3HT and is often used to fabricate bulk heterojunction organic solar cells.
dSIMS profiles of the bilayers prior to and after 1 h exposure to vapors are shown in Figure 2. The depth profiles monitor variation in composition as a function of sputtering time. For all silicon substrate samples, C4S -signals (green solid triangles) were used to determine P3HT content, C2O -signals (red open circles) for PC60BM content, and Sisignals (black solid squares) were used to locate interfaces with substrates. The quantitative conversion of sputtering time to intermixed structure depth of P3HT and PCBM is not straightforward [28]; thus, we presented the profiles as functions of sputtering time.   The dSIMS profiles revealed that an initially sharp interface between PC60BM and polythiophene layers (Figure 2a) slightly broadened when exposed to THF (Figure 2b) or DCE (Figure 2c) vapors, whereas a single intermixed layer was formed upon exposure to The dSIMS profiles revealed that an initially sharp interface between PC 60 BM and polythiophene layers (Figure 2a) slightly broadened when exposed to THF (Figure 2b) or DCE (Figure 2c) vapors, whereas a single intermixed layer was formed upon exposure to chloroform vapor (Figure 2d). Sulfur-containing cluster signal characteristics for polythiophene in the top layer increased by an order of magnitude for samples treated with DCE and THF, whereas for chloroform, this value increased by half that value for pure P3HT. All samples were prepared under the same conditions, and the initial thicknesses of both fullerene derivative (44 nm) and polthiophene (55 nm) were the same for all samples. The bilayer depth profiles indicated large differences in PC 60 BM and P3HT sputtering efficiencies. Another factor influencing strongly shape of depth profiles obtained by ion sputtering methods is the topography of analyzed films [37]. In the profile corresponding to the as-prepared bilayer sample, the steep increase in Si signal from the substrate was accompanied by a rapid signal decrease for polythiophene. In contrast, for samples exposed to chloroform vapor, the C 4 S − signal faded long after the silicon signal increased after 1500 s of sputtering (Figure 2d). This indicated that the initially flat surface of bilayer (Figure 2a) was roughened when exposed to chloroform vapor. Qualitatively, a similar relationship between the degree of acceptor and donor interdiffusion and the type of solvent observed for the PC 70 BM/PCDTBT layers (not shown).
Detailed    A systematic broadening of the substrate interface as determined from silicon signals indicated increased free surface corrugation. This observation is supported by studies on the evolution of free surface of bilayers exposed to chloroform vapor. Topographic pictures presented in Figure 5a-d show a systematic wrinkling of PC60BM/P3HT bilayer surfaces with time. Initially, surface undulation amplitude increased (compare Figure 5a,b), followed by the formation of small holes (Figure 5c), and, after 60 min a structure typical of late stage dewetting phenomenon was observed (Figure 5d). For PC60BM/PCDTBT (see Figure S1) high corrugation of the free surface appears after 45 min of chloroform vapor treatment but no structure typical of late stage dewetting were observed within the experiment. Quenching of conjugated polymers (donors) fluorescence caused by interactions with acceptors (PC60MB/PC70BM) allows real-time observation of the intermixing processes using fluorescent microscopy. Presented in Figure S2 fluorescent micro-graphs show the initial formation of round domains (intermixed areas) that grow with time to form finally a structure typical of dewetting. We hypothesize that under solvent vapor annealing domains of saturated donor:acceptor mixtures are formed, and this process may be followed by dewetting of excess of material. A systematic broadening of the substrate interface as determined from silicon signals indicated increased free surface corrugation. This observation is supported by studies on the evolution of free surface of bilayers exposed to chloroform vapor. Topographic pictures presented in Figure 5a-d show a systematic wrinkling of PC 60 BM/P3HT bilayer surfaces with time. Initially, surface undulation amplitude increased (compare Figure 5a,b), followed by the formation of small holes (Figure 5c), and, after 60 min a structure typical of late stage dewetting phenomenon was observed (Figure 5d). For PC 60 BM/PCDTBT (see Figure S1) high corrugation of the free surface appears after 45 min of chloroform vapor treatment but no structure typical of late stage dewetting were observed within the experiment. Quenching of conjugated polymers (donors) fluorescence caused by interactions with acceptors (PC 60 MB/PC 70 BM) allows real-time observation of the intermixing processes using fluorescent microscopy. Presented in Figure S2 fluorescent micro-graphs show the initial formation of round domains (intermixed areas) that grow with time to form finally a structure typical of dewetting. We hypothesize that under solvent vapor annealing domains of saturated donor:acceptor mixtures are formed, and this process may be followed by dewetting of excess of material.

Solar Cells
To study the influence of the interdiffusion in acceptor/donor un annealing on the performance of solar cells, a series of devices on ZnOelectrodes were prepared. First, ZnO ink was cast on top of patterned IT and annealed on a hotplate. Next, PC60BM/P3HT bilayers were prepa casting from orthogonal solvents while PCDTBT/PC70BM bilayers were p fer method [29,30]. Next bilayers were exposed to chloroform vapor in as the scheme depicted in Figure 1. Finally, metallic electrodes (Ag) wer orated through a shadow mask to form eight independent devices on o active area of a single devices was 2 × 2 mm. SIMS profiles and AFM images obtained for samples exposed to for various time periods showed that the changes in the substrate from not significantly affect the interdiffusion process (Figures 6 and S3) (green solid triangles) were used to determine donor profile, the C3Ocircles) were used for acceptors, and the 70 ZnO − signals (black open squ locate the cathode interface. The PC60BM component interpenetrated slightly during casting, and the interface between fullerene derivative a much broader than samples prepared with thermal evaporation (comp 3a). Broadening of the interface between PC70BM and PCDTBT was also ples prepared by transfer method (see Figure S3).

Solar Cells
To study the influence of the interdiffusion in acceptor/donor under solvent vapor annealing on the performance of solar cells, a series of devices on ZnO-covered ITO/glass electrodes were prepared. First, ZnO ink was cast on top of patterned ITO/glass substrates and annealed on a hotplate. Next, PC 60 BM/P3HT bilayers were prepared by sequential casting from orthogonal solvents while PCDTBT/PC 70 BM bilayers were prepared by transfer method [29,30]. Next bilayers were exposed to chloroform vapor in the same manner as the scheme depicted in Figure 1. Finally, metallic electrodes (Ag) were thermally evaporated through a shadow mask to form eight independent devices on one substrate. The active area of a single devices was 2 × 2 mm. SIMS profiles and AFM images obtained for samples exposed to chloroform vapor for various time periods showed that the changes in the substrate from silicon to ZnO did not significantly affect the interdiffusion process (Figures 6 and S3). The C 4 S − signals (green solid triangles) were used to determine donor profile, the C 3 O − signals (red open circles) were used for acceptors, and the 70 ZnO − signals (black open squares) were used to locate the cathode interface. The PC 60 BM component interpenetrated the polymer layer slightly during casting, and the interface between fullerene derivative and the P3HT was much broader than samples prepared with thermal evaporation (compare Figures 2a and 3a). Broadening of the interface between PC 70 BM and PCDTBT was also observed for samples prepared by transfer method (see Figure S3). The dSIMS profiles indicated a complete mixing of both phases after 10 min solvent vapor treatment. AFM topographic images of samples exposed to solvent vapor longer than 10 min showed structures typical of the late stages of the dewetting process. SIMS composition maps (see Figure S4 in Supporting Materials) indicated the surfaces of lower domains were reached in sulfur-containing component, whereas the P3HT contents in higher domains were lower.
Representative current voltage characteristics of solar cells prepared by solvent vapor annealing of the bilayers are presented in Figure 7. To avoid unpredictable effects of solvent vapor annealing on PEDOT:PSS, an organic film typically used as a hole-transport layer, the inverted geometry of solar cells with the ZnO transparent electrode was chosen. Initially, the acceptor layer (PC70BM) was in contact with ZnO, the donor PCDTBT with Ag electrode, and the contact area between donor and acceptor was relatively small. As seen in Figure 7a), the current-voltage curves shifted down the fourth quarter of the coordinate system with the solvent vapor annealing time as the intermixing of PC70BM and PCDTBT proceeded. Significantly lower current values obtained for devices prepared by the layer transfer method may result from the partial degradation of the polymer due to contact with water. The dSIMS profiles indicated a complete mixing of both phases after 10 min solvent vapor treatment. AFM topographic images of samples exposed to solvent vapor longer than 10 min showed structures typical of the late stages of the dewetting process. SIMS composition maps (see Figure S4 in Supporting Materials) indicated the surfaces of lower domains were reached in sulfur-containing component, whereas the P3HT contents in higher domains were lower.
Representative current voltage characteristics of solar cells prepared by solvent vapor annealing of the bilayers are presented in Figure 7. To avoid unpredictable effects of solvent vapor annealing on PEDOT:PSS, an organic film typically used as a hole-transport layer, the inverted geometry of solar cells with the ZnO transparent electrode was chosen. Initially, the acceptor layer (PC 70 BM) was in contact with ZnO, the donor PCDTBT with Ag electrode, and the contact area between donor and acceptor was relatively small. As seen in Figure 7a), the current-voltage curves shifted down the fourth quarter of the coordinate system with the solvent vapor annealing time as the intermixing of PC 70 BM and PCDTBT proceeded. Significantly lower current values obtained for devices prepared by the layer transfer method may result from the partial degradation of the polymer due to contact with water. The initial layer arrangement of P3HT/PC60BM devices was chosen to mimic structures classified as unfavorable for charge transport [7] to better show the effects of bulk heterojunction formation due to the mutual diffusion of acceptor and donor. This was reflected in the shape of the current-voltage characteristics of the as-prepared devices. Large slopes of almost linear I-V profiles and low values of the open circuit voltage VOC were largely due to energy mismatch levels between the lower layer (P3HT) and the cathode (ZnO) and the upper layer (PC60BM) and the anode (Ag).
The I-V curves for devices prepared by solvent vapor annealing evolved with treatment time. The open circuit voltage monotonically increased to reach values close to 0.5 V (PC60BM/P3HT) and 0.65 (PC70BM/PCDTBT) typical for devices prepared on ZnO; this was accompanied by increased short circuit current. The maximum values of short circuit currents and open voltages were observed after 10 min of solvent vapor annealing for PC60BM/P3HT bilayers and after five minutes for PC70BM/PCDTBT. In case of the PC60BM/P3HT it corresponds to the moment when structures typical for late stage of dewettig begins to form. The I-V curves took the shape of a typical solar cell as reflected in an increased fill factor, however for PC70BM/PCDTBT solar cells decrease in FF is observed for longer treatment times. These performance parameters are summarized for each solvent vapor treatment time in Figures 8 and S5 (at least eight per data point) under AM1.5G solar simulator illumination. Monotonous increases were observed for efficiency η (Figure 8a and S5), short circuit current ISC (Figures 8b and S5), open voltage VOC (Figures  8c and S5) and fill factor FF (Figures 8d and S5) for devices annealed up to 10 min. Later, nearly all parameters stabilized, but the spread of their values decreased. Changes in I-V curves are reflected in the shunt resistances Rsh which for both systems increase with solvent vapor annealing time (Tables S1 and S2 and Figure S6). However in the PCDTBT:PC70BM system after initial increase, Rsh drops for annealing times above 1 min. In this system also series resistance Rs is considerably larger than in the P3HT:PC60BM system which can be caused by layer transfer, through water, method which can lead to higher values of Rs in the system. The initial layer arrangement of P3HT/PC 60 BM devices was chosen to mimic structures classified as unfavorable for charge transport [7] to better show the effects of bulk heterojunction formation due to the mutual diffusion of acceptor and donor. This was reflected in the shape of the current-voltage characteristics of the as-prepared devices. Large slopes of almost linear I-V profiles and low values of the open circuit voltage V OC were largely due to energy mismatch levels between the lower layer (P3HT) and the cathode (ZnO) and the upper layer (PC 60 BM) and the anode (Ag).
The I-V curves for devices prepared by solvent vapor annealing evolved with treatment time. The open circuit voltage monotonically increased to reach values close to 0.5 V (PC 60 BM/P3HT) and 0.65 (PC 70 BM/PCDTBT) typical for devices prepared on ZnO; this was accompanied by increased short circuit current. The maximum values of short circuit currents and open voltages were observed after 10 min of solvent vapor annealing for PC 60 BM/P3HT bilayers and after five minutes for PC 70 BM/PCDTBT. In case of the PC 60 BM/P3HT it corresponds to the moment when structures typical for late stage of dewettig begins to form. The I-V curves took the shape of a typical solar cell as reflected in an increased fill factor, however for PC 70 BM/PCDTBT solar cells decrease in FF is observed for longer treatment times. These performance parameters are summarized for each solvent vapor treatment time in Figures 8 and S5 (at least eight per data point) under AM1.5G solar simulator illumination. Monotonous increases were observed for efficiency η (Figures 8a and S5), short circuit current I SC (Figures 8b and S5), open voltage V OC (Figures 8c and S5) and fill factor FF (Figures 8d and S5) for devices annealed up to 10 min. Later, nearly all parameters stabilized, but the spread of their values decreased. Changes in I-V curves are reflected in the shunt resistances R sh which for both systems increase with solvent vapor annealing time (Tables S1 and S2 and Figure S6). However in the PCDTBT:PC 70 BM system after initial increase, R sh drops for annealing times above 1 min. In this system also series resistance R s is considerably larger than in the P3HT:PC 60 BM system which can be caused by layer transfer, through water, method which can lead to higher values of R s in the system. An analysis of the data in Figures 8, S5 and S6 and the corresponding SIMS pro leads to the conclusion that the cell parameters increased when the donor-rich dom and acceptor-rich phases formed large area contacts with the anode and cathode, res tively. This showed that the solvent vapor annealing of PC60BM/P3HT PC70BM/PCDTBT bilayers leads to the formation of bulk heterojunctions after a minutes. Further changes in film morphology under exposure to solvent vapor did lead to improved solar cell efficiency. To exclude a hypothesis that solvent vapor ann ing leads to the dewetting of the acceptor film, P3HT based devices without PC60BM w prepared and analyzed. Their efficiency was much lower (0.0017) than that of the dev prepared by sequential deposition (0.0093). This showed that a donor-acceptor bulk erojunction was formed upon solvent vapor annealing of the bilayers.
Our observations were qualitatively consistent with simulation results of the im of donor-acceptor vertical composition profiles on OSC performance [7]. The structu P3HT/PC60BM devices at early stages of intermixing can be classified (according to al. [7]) as unfavorable for charge transport. The simulations showed that unfavorable tical distributions of donors and acceptors leads to a higher recombination of charge riers resulting in a lower open circuit voltage VOC consistent with our findings. Fo uniform distribution of donors and acceptors in the active layer, in the devices anne for longer durations, the simulation predicted a higher short circuit current ISC amon studied structures but a smaller VOC than configurations favoring charge transport. An analysis of the data in Figures 8, S5 and S6 and the corresponding SIMS profiles leads to the conclusion that the cell parameters increased when the donor-rich domains and acceptor-rich phases formed large area contacts with the anode and cathode, respectively. This showed that the solvent vapor annealing of PC 60 BM/P3HT and PC 70 BM/PCDTBT bilayers leads to the formation of bulk heterojunctions after a few minutes. Further changes in film morphology under exposure to solvent vapor did not lead to improved solar cell efficiency. To exclude a hypothesis that solvent vapor annealing leads to the dewetting of the acceptor film, P3HT based devices without PC 60 BM were prepared and analyzed. Their efficiency was much lower (0.0017) than that of the devices prepared by sequential deposition (0.0093). This showed that a donor-acceptor bulk heterojunction was formed upon solvent vapor annealing of the bilayers.

Conclusions
Our observations were qualitatively consistent with simulation results of the impact of donor-acceptor vertical composition profiles on OSC performance [7]. The structure of P3HT/PC 60 BM devices at early stages of intermixing can be classified (according to Bi et al. [7]) as unfavorable for charge transport. The simulations showed that unfavorable vertical distributions of donors and acceptors leads to a higher recombination of charge carriers resulting in a lower open circuit voltage V OC consistent with our findings. For the uniform distribution of donors and acceptors in the active layer, in the devices annealed for longer durations, the simulation predicted a higher short circuit current I SC among all studied structures but a smaller V OC than configurations favoring charge transport.

Conclusions
We have shown that solvent vapor annealing induced interdiffusion in a model acceptor/donor bilayer structure. This phenomenon was used to fabricate an interpenetrating network of acceptor and donor domains for two pairs of model system, P3HT:PC 60 BM and PCDTBT:PC 70 BM. In contrast to phase separation, which has been used to fabricate bulk organic photovoltaic heterojunction devices, solvent vapor annealing induced interdiffusion can also be used when there is no common solvent for the donor and acceptor. Observed intermixing was much slower and allowed for greater control of the final morphology of the solar cell active layer. This was also slower than thermally induced intermixing and can be used instead to avoid the high-temperature process of the active layer that is crucial for devices prepared on flexible surfaces. Further, the ability to select a solvent gives greater control of the process and potential fine tuning of the final morphology. We demonstrated that the performance of inverted organic solar cells increased up to the moment when the intermixed structure of donor and acceptor formed. The general methodology reported herein provides a new method for fabricating organic solar cells using other recently developed donor and acceptor (macro)molecules, its advantages may play a positive role in the commercialization of organic solar cells.