Poly[(2,6-(4,8-bis(5-(2-ethyl-hexyl)thiophen-2-yl)-benzo-[1,2-b:4,5-b’]dithiophene))-alt-(5,5-(1’,3’-di-2-thienyl-5’,7’-bis(2-ethyl-hexyl)benzo [1’,2’-c:4’,5’-c’] dithiophene-4,8-dione))] (PBDB-T) and 3,9-bis(2-methylene-(3-(1,1-dicyanomethylene)indanone))-5,5,11,11-tetrakis(4-hexylphenyl)-dithieno[2,3-d:2’,3’-d’]-s-indaceno[1,2-b:5,6-b’]-dithiophene) (ITIC) were purchased from Solarmer Materials Inc. 1,3-propanedithiol (PDT, 97%) and 1,8-octanedithiol (ODT, 98%) were obtained from Alfa Aesar. 1,4-butanedithiol (BDT, ) was from Tokyo Chemical Industry (TCI). The chemical structures of these materials are shown in Fig. 1.

Fig. 1.

	Figure Option ViewDownloadNew Window
	Fig. 1. The chemical structures of the materials used in the active layers.

The ZnO precursor was prepared by dissolving zinc acetate dehydrate (Zn(CH₃COO)₂ 2 H₂O, 219 mg) and ethanolamine (NH₂CH₂CH₂OH, 60 mL) in 2-methoxyethanol (CH₃OCH₂CH₂OH, 2 mL), which is similar to the literature report.^[38] The ZnO precursor was filtered twice by 0.22- PTFE syringe filter before use. The polymer solution was prepared by dissolving PBDB-T and ITIC (1:1 by weight) into chlorobenzene (CB) at a total concentration of 20 mg/mL, which is followed by the addition of solvent additives (such as PDT, BDT, or ODT) at a small fraction.

The indium-tin oxide (ITO) glass substrates were ultrasonically cleaned with detergent, deionized water, acetone, and isopropanol for 20 min each in turn. After drying with N₂ flow, these substrates were placed in ultraviolet ozone environment for 20 min. The filtered ZnO precursor was spin-coated on top of the ITO substrate at a speed of 3000 rpm for 40 s and then annealed at 200 °C for 20 min in air to form a thin ZnO layer. These were transferred into an N₂-filled glove box. Active layers were spin-coated on top of the ZnO layer from the PBDB-T:ITIC solutions (with or without additives) at a speed of 1100 rpm for 60 s with an approximate thickness of 100 nm. These were then annealed at 160 °C for 10 min. A thin MoO₃ layer (6 nm) was thermally evaporated on top of the active layer. Finally, an Ag electrode (∼100 nm) was thermally evaporated at a pressure of 2 × 10⁻⁴ Pa.

The ultraviolet-visible (UV-Vis) absorption spectra of the film specimen were taken using UV-Vis-Near infrared (NIR) scanning spectrometer (Shimadzu UV-3101PC). Steady-state or transient photoluminescence spectra were measured using a Horiba Fluorolog-3 spectrofluorometer or Edinburgh Instruments FLS920 fluorescence spectrometer, respectively. The surface morphologies of films were investigated using an atomic force microscope (AFM, Nanocute, SII NanoTechnology) in a tapping mode and scanning electron microscope (SEM, Hitachi S4800). X-ray diffraction patterns were taken using a Bruker D8 advance diffractometer. The current density–voltage (J–V) characteristics of the devices were measured using a source measure unit (Agilent B2902A) under a standard AM 1.5G solar irradiation illumination with a light intensity of 100 mW/cm² from a solar simulator (Abet Sun 2000). The external quantum efficiencies (EQEs) of the polymer solar cells were detected with a solar cell quantum efficiency measurement system (Zolix Solar Cell Scan 100). All devices were tested in an ambient atmosphere without encapsulation.

Inverted bulk-heterojunction polymer solar cells are fabricated, which have a general device configuration, ITO/ZnO/PBDB-T:ITIC/MoO₃/Ag, as shown in Fig. 2(a) alongside with the energy diagram of the layers in Fig. 2(b). In order to optimize the device performance, a number of organic-solvent additives, PDT, BDT, or ODT, are selected in the current investigation. These are tested individually into the donor-acceptor blended active layers in order to manipulate the device performance. Three different solvent additives with varied volume ratios are used into the PBDB-T:ITIC blended active layers, the J–V curves corresponding to the devices of adding PDT, BDT, and ODT are illustrated in Figs. 2(c), 2(d), and 2(e), respectively, and the detailed photovoltaic parameters are summarized in Table 1. Such optimized fractions for different additives are found to be slightly different, which are 1% for PDT, 0.5% for BDT, and 0.5% for ODT additives. These fractions are then used in the follow-up devices fabrication.

Fig. 2.

	Figure Option ViewDownloadNew Window
	Fig. 2. Wide angle x-ray diffraction patterns of PBDB-T:ITIC blend films with or without additive (the curves are shifted vertically for the sake of clarity).

Table 1.

Table 1.

Table 1. The photovoltaic performance of photovoltaic devices based on PBDB-T:ITIC blend with varying volume ratios of three different solvent additives. . Additive/% VOC/V JSC/(mA/cm2) FF/% Best (ave.) a PCE/% control (0%) 0.86 15.74 60.04 8.13 (8.07) 0.1% PDT 0.85 16.07 62.07 8.48 (8.33) 0.3% PDT 0.85 16.26 62.76 8.67 (8.54) 0.5% PDT 0.85 16.32 63.26 8.77 (8.69) 0.7% PDT 0.86 16.41 63.07 8.90 (8.78) 1% PDT 0.86 16.49 63.74 9.04 (8.92) 2% PDT 0.84 15.79 60.15 7.98 (7.90) 0.1% BDT 0.86 16.29 61.08 8.56 (8.47) 0.3% BDT 0.86 16.70 62.96 9.04 (8.83) 0.5% BDT 0.86 17.06 63.55 9.33 (9.16) 0.7% BDT 0.86 16.75 61.32 8.83 (8.74) 1% BDT 0.86 16.08 61.59 8.52 (8.35) 0.1% ODT 0.86 16.23 62.09 8.67 (8.58) 0.3% ODT 0.86 16.59 63.02 8.99 (8.87) 0.5% ODT 0.86 17.07 64.30 9.44 (9.28) 0.7% ODT 0.86 16.60 62.38 8.91 (8.79) 1% ODT 0.86 16.18 58.89 8.20 (8.11) a The values in brackets are the average PCEs for 10 devices.

Table 1. The photovoltaic performance of photovoltaic devices based on PBDB-T:ITIC blend with varying volume ratios of three different solvent additives. .

It can be seen from Table 1 that the addition of PDT, BDT, or ODT additives into the active layers has no effect on the open-circuit voltage (V_OC) of the devices. However, there are distinct enhancements in short-circuit current density (J_SC) and fill factor (FF) in all devices investigated with an additive in the active layers. It appears that the additive with the longest flexible chain, such as ODT, has the strongest impact on the device performance. The device fabricated with 0.5% ODT in the active layer demonstrates the best performance. In this device, the J_SC reaches 17.07 mA/cm² (8.4% increase from 15.74 mA/cm² of the control device) and FF64.30% (7.1% increase from 60.05% of the control device). The additive optimization leads to an overall 16% enhancement in PCE, from 8.13% of the control to 9.44% of 0.5% ODT device.

The insensitivity of the additives to V_OC of the device can be understood as the fact that the three additives have identical functional groups though the chain lengths are different. Since the functional group determines the polarity of a molecule, it means that the effect on charge distribution of the adjacent molecules, such as the donor/acceptor blend, is similar. Therefore, the V_OC values of the devices are unchanged. In order to understand the increase in J_SC and FF, other factors have to be considered.

Because the boiling temperatures of PDT, BDT, and ODT are quite different, which are 169°, 195°, and 269°, respectively, the additive in the active layers may alter the crystallization process during the film formation and thus the morphology of the films. It is well known that the exciton dissociation, carrier transport and collection are closely related to the morphology of the active layer. The formation of a suitable microstructure with phase seperation and network interpenetrating channels for charge transport will be beneficial to improve the J_SC and FF of the device. The morphology of the active layers is first analyzed using wide-angle x-ray diffraction (XRD). The XRD patterns of PBDB-T:ITIC blended films with or without additives are shown in Fig. 3 and the peak positions are listed in Table 2. The diffraction from each individual material is also obtained for comparison (not shown).

Fig. 3.

	Figure Option ViewDownloadNew Window
	Fig. 3. (a) The structure, (b) the energy level diagram of each layer, and the J–V characteristics of the PBDB-T:ITIC-based photovoltaic devices with additives (c) PDT, (d) BDT, and (e) ODT.

Table 2.

Table 2.

Table 2. Peak positions, d-spacings, and the calculated correlation lengths from wide angle x-ray diffraction of the blend films with different additives. . Additives /°) d1-spacing/Å /°) d2-spacing/Å /Å a Control (0%) 4.70 18.8 7.39 12.0 197 1% PDT 4.91 18.0 7.41 11.9 239 0.5% BDT 4.89 18.1 7.39 12.0 221 0.5% ODT 4.62 19.1 7.41 11.9 246 a Calculated using the diffraction peak at .

Table 2. Peak positions, d-spacings, and the calculated correlation lengths from wide angle x-ray diffraction of the blend films with different additives. .

From Fig. 3, three distinct diffractions are observed. The one at low angle q₁ = 0.33Å˙⁻¹ (2 corresponding to a d-spacing of 18.8 Å is the diffraction from PBDB-T lamellar structure,^[39] while the peak at q₂ = 0.53 Å⁻¹ is the diffraction from ITIC lamellar structure.^[40] Unlike PBDB-T:PBDTTPD blend in a previous work where an alloy was developed,^[39] the lamellar structures of the PBDB-T:ITIC blend display distinct structural features of individual components. These means that both PBDB-T or ITIC are in separated phases rather than in an alloy. The donor/acceptor is blended in a form mixing crystalline domains of each individual component. In addition, the dimension of the PBDB-T lamellar is smaller in the PBDB-T:ITIC blend than that in the PBDB-T:PBDTTPD blend as the x-ray diffraction peak of the PBDB-T component occurs at a slightly higher angle in PBDB-T:ITIC blend.^[39] Diffraction at a relatively high angle, q₃=1.55 Å⁻¹ (2 , is superimposed on top of an amorphous diffraction and related to the feature in the out-plane direction.

A distinct feature can be observed where the diffraction peaks at and at become narrower when adding the solvent additives into the active layers. This is an indication of improvement in structural order in the crystallization. The changes in structural order can be estimated using correlation length ( ), which is defined as / , where the is the full width at half maximum intensity of the peak, and q is the scattering factor defined by .^[41–43] The values calculated based on the diffraction at q₂ of diffraction representing ITIC component from specimens prepared using different additives are listed in Table 2. It is found that increases significantly after the addition of additives into PBDB-T:ITIC blended active layer. Longer correlation length of diffraction at q₂ represents an extended order structure in ITIC component, whereas there is almost no change in the correlation length of diffraction at q₁ representing PBDB-T component. This indicates that the additives would have a stronger impact on the morphology of the molecular acceptor ITIC than that of donor PBDB-T. A relative high degree of order in molecular organization of ITIC would benefit the electron transport in the actively layer and balance the hole/electron transport, which in turn improves the device performance. This work demonstrates that simple solvent additives may manipulate structural orders and improve the device performance.

Solvent additives have been widely used in device fabrication. However, the majority are focused on phase separation between the donor and acceptor and/or crystallization in polymer donor. Rogers et al. reported the use of ODT as a solvent additive, and found that the PCEs of the devices were boosted from 2.8% to 5.5% for a fullerene PCPDTBT:PC₇₁BM system due to subtly increased crystalline correlation lengths of the donor’s features.^[44] Another early work reported by Jenekhe’s group in 2011 demonstrated that a solvent additive (1,8-diiodooctane, DIO) was added into the processing solvent (dichloromethane) in a non-fullerene P3HT:NDI-3TH blend. The PCE of this device was improved by a factor of 10. It was found that DIO was able to regulate the aggregation and the domain size of the acceptor (NDI-3TH) phase in the BHJ blends.^[45] On the whole, only limited investigations reported the tuning order of the crystallization in acceptor’s domain in the blends.

The surface morphology of PBDB-T:ITIC blended films with and without solvent additives is observed using AFM as displayed in Fig. 4. The control film without additive shows a root-mean-square (RMS) surface roughness of 6.18 nm, where the average RMS surface roughness of the films with additive PDT (1%), BDT (0.5%), or ODT (0.5%) slightly decreases to 5.92 nm, 5.02 nm, or 4.95 nm, respectively. In comparison with the films without additives, the surface of the active layer becomes smoother by adding the solvent additives. At the same time, it can be seen from the phase images that the films with additives show more distinct phase separations between the donor and accepter materials; in addition, nano-scale fiber structure appears when 0.5% ODT is added. This is beneficial to the dissociation of excitons and the transport of the charges.

Fig. 4.

	Figure Option ViewDownloadNew Window
	Fig. 4. (a)–(d) AFM height images and (e)–(h) phase images of PBDB-T:ITIC blend films; (a) and (e) without additive, and with additive of: (b) and (f) 1% PDT, (c) and (g) 0.5% BDT, and (d) and (h) 0.5% ODT.

Further SEM investigation (see Fig. 5) shows clearer images on surface morphology of the active layers. It can be seen that the surface of the films consists of similar interpenetrating network structure, but the grain size becomes more uniform in the active layers with additives.

Fig. 5.

	Figure Option ViewDownloadNew Window
	Fig. 5. SEM images of PBDB-T:ITIC blend films (a) without additive and with additive of (b) 1% PDT, (c) 0.5% BDT, and (d) 0.5% ODT.

In order to understand device performance, photophysical properties of the active layer are investigated. The absorption spectra of PBDB-T:ITIC blend films with or without additive are shown in Fig. 6(a). It can be seen that the absorption spectra from specimens with or without additives exhibit a rather similar profile, which has multiple peaks in the range from 500 nm to 740 nm, a maximum at 634 nm, and two shoulders at 580 nm and 700 nm. According to our early work, the absorption in the range from 500 nm to 630 nm was mainly contributed by donor PBDB-T, whereas the absorption from 600 nm to 740 nm was from acceptor ITIC.^[39] The side absorption at 580 nm therefore mainly belongs to PBDB-T and the absorption at the other side at 700 nm is due to ITIC absorption. The maximum absorption at 634 nm is an overlap region between the absorptions of the two components. The EQE spectra of the corresponding devices are measured as shown in Fig. 6(b). It covers a similar spectral range from 400 nm to 740 nm compared to the absorption. It is interesting to see that the device with additives in the active layers demonstrates significant enhancement in EQE at all spectral range. The EQE spectrum from the control device maximized around 70.7% is sitting at the bottom, while the EQE curve from the device with 0.5% BDT additive (EQE max. ∼79.7%) almost coincides with that from the device with 0.5% ODT additive (EQE max. ∼80.1%) lying on the top. The measured J_SC value from the J–V curve is consistent with the integral of EQE response obtained.

Fig. 6.

	Figure Option ViewDownloadNew Window
	Fig. 6. (a) UV-Vis absorption; (b) EQE spectra; (c) and (d) PL spectra of neat PBDB-T, neat ITIC, and PBDB-T:ITIC blended films with or without solvent additives as indicated; (e) the current density measured under dark condition and (f) Jph–Veff curves of PBDB-T:ITIC-based photovoltaic devices with or without solvent additives as indicated.

In order to understand the excitation and charge transfer status in the active layers, photoluminescence (PL) spectra are measured as illustrated in Figs. 6(c) and 6(d), and excited at 560 nm where mainly absorbed by PBDB-T. Unblended PBDB-T specimen exhibits a broad, multiple PL emission with a peak at 725 nm, accompanying with shoulders around 685 nm and 820 nm. Pure ITIC specimen shows a double PL emission with a main peak at 820 nm and a shoulder roughly around 780 nm. However, the PL emission of PBDB-T:ITIC blended film exhibits two peaks located at 680 nm and 800 nm with a substantially reduced intensity (see Fig. 6(c)). In addition, the two emission peaks shift away from the emissions of the individual components (PBDB-T or ITIC). Because the range of emission from the blend completely overlaps with the absorption of the ITIC component,^[39] an inner filter effect is unavoidable. The emission peak at 680 nm appears to be the residue emission from PBDB-T, while the emission around 800 nm may also be the residue PBDB-T emission. A slightly high fraction in red-end emission may be caused by energy transfer from PBDB-T to ITIC, i.e., the PBDB-T emission is partially absorbed by ITIC and then is re-emitted. Nevertheless, the overall reduction or quenching in emission from the PBDB-T:ITIC blend means that the majority of energy absorbed by ITIC does not generate PL emission. There is an additional quenching channel that is opened. This is rather similar to the investigation where photo-induced electron transfer was found.^[39] The emission quenching in donor-acceptor pair is often related to the electron transfer, energy transfer, charge transfer, or exciton dissociation process at the D/A interface.^[39,46–48]

It can be seen from Fig. 6(d) that the addition of solvent additives further leads to an overall reduction in PL emission from the PBDB-T:ITIC blend. At the same time, the emission profiles in the long wavelength range from 750 nm to 830 nm are also altered. This agrees with morphology investigation in the last section that additives are able to re-organize order structure of ITIC crystalline domain, extending the lamellar structure. A better donor/acceptor microcrystal organization will further quench the emission since the more efficient charge transfer channel is opened.

The impact of solvent additives to the excitation status in active layer and the photocurrent in the device can be analyzed via charge generation and recombination process of the devices.^[39] The saturation current density (J_sat), maximum exciton generation rate (G_max), and charge dissociation probabilities P(E,T) can be evaluated from the photocurrent of the devices. The photocurrent density (J_ph) is defined as J_ph=J_L–J_D, where J_L and J_D are the current densities measured under illumination and in the dark. Figure 6(e) shows the J_D–V curves, which demonstrates rectified characteristics and a substantially reduced dark current. A relationship between J_ph and the effective voltage (V_eff) is plotted in Fig. 6(f) on a double logarithmic scale. V_eff is given by V_eff=V₀–V, where V is the applied voltage and V₀ is the compensation voltage obtained from the J_ph–V characteristics at J_ph=0. Assuming the recombination of free charge carriers and space-charge effects can be neglected, the photocurrent through the external circuit is J_ph=eGL, where e is the elementary charge, G is the generation rate of charge carriers, and L is the thickness of the active layer. As the applied voltage is increased, the photocurrent of the device reaches saturation (J_sat). At these voltages, all bound e–h pairs are separated and the maximum photocurrent J_sat=eG_max L is reached, where G_max is the maximum exciton generation rate.^[49,50] In this work, the saturation current density J_sat is obtained at (Fig. 6(f)). Thereby, G_max can be calculated through the above formula. The charge dissociation probabilities, , can also be obtained from photocurrent of the device.^[50] The G_max and P(E,T) obtained at short-current and maximum output power conditions are listed in Table 3. It can be seen from Table 3 that the maximum exciton generation rates G_max of devices with solvent additives are all increased due to a general rise in charge dissociation probabilities, which in turn enhances the photocurrent and J_sat in the device when having solvent additives (see Fig. 6(f)).

Table 3.

Table 3.

Table 3. The evaluated values of maximum exciton generation rate and charge dissociation probability. . Additives Gmax/ /% a /% b Control (0%) 1.304 × 1028 94.71 76.41 1% PDT 1.355 × 1028 95.48 80.31 0.5% BDT 1.399 × 1028 95.68 82.84 0.5% ODT 1.397 × 1028 96.28 82.35 a At short circuit condition. b At maximum output power condition.

Table 3. The evaluated values of maximum exciton generation rate and charge dissociation probability. .

In combination with the morphology result, this can be understood that the solvent additives effectively re-organize the structure in acceptor crystalline domain which promotes the exciton generation by minimizing bimolecular recombination, in which the latter is supported by the observation in the reduction in PL emission. The charge dissociation is enhanced. The investigation well explains the increase in J_SC of the devices and the general increase in the EQE spectra.