Morphological, Chemical, and Electronic Changes of the Conjugated Polymer PTB7 with Thermal Annealing

Summary

There is considerable interest in improving the performance of organic optoelectronic devices through processing techniques. Here, we study the effect of high-temperature annealing on the properties of the semiconducting polymer PTB7 and PTB7:fullerene blends, of interest as efficient organic photovoltaic (OPV) devices. Annealing to moderate temperature improves the PTB7 morphology and optoelectronic properties. High-temperature annealing also improves morphology but results in poorer optoelectronic properties. This is a result of side chain cleavage that creates by-products that act as trap states, increasing electronic disorder and decreasing mobility. We further observe changes to the PTB7 chemical structure after thermal cleavage that are similar to those following solar irradiation. This implies that side chain cleavage is an important mechanism in device photodegradation, which is a major “burn-in” loss mechanism in OPV. These results lend insight into side chain cleavage as a method of improving optoelectronic properties and suggest strategies for improvement in device photostability.

Graphical Abstract

Highlights

• •Annealing to 260°C improves morphology and hole mobility in PTB7-based thin films
• •Annealing to 290°C induces side-chain cleavage and closer packing of PTB7
• •Thermally cleaved PTB7 films resemble irradiated films, suggesting a burn-in mechanism
• •Release of trapped reaction by-products could result in improved device performance

Energy Materials; Polymers; Spectroscopy

Introduction

Organic semiconducting films can be used for making efficient lighting (organic light-emitting diodes, or OLEDs), electronics with low production energy cost (organic field effect transistors, or OFETs [Kymissis, 2009]), and next-generation solar cells (organic photovoltaics, or OPVs). OPVs can be used to generate energy with near-zero CO₂ production and have many attractive features such as their potentially low cost, flexible form factor, and light weight. These desirable properties are especially important to bring solar energy to developing nations, where energy demand is growing fastest (BP, 2016) but the capital to invest in large-scale energy plants is not readily available.

Although these applications of organic semiconductors have different materials engineering goals, a fundamental understanding of the electronic and morphological properties of polymers and small organic molecules is crucial for the optimization of organic semiconductors. In particular, increasing charge carrier mobility is crucial for increasing the performance of organic electronic devices, and this requires an in-depth understanding of the relationship between microstructure and mobility. Polymers and small molecules for solution-processed devices typically consist of a semiconducting backbone and solubilizing, electrically insulating alkyl side chains (Vogelbaum and Sauvé, 2017). Charges are delocalized along the backbone, making intra-molecular charge transport effective, so average mobility can be limited by inter-molecular charge transport (hopping between chains). For polymers, it is often believed that inter-molecular hopping rate is inversely related to the distance between hopping sites and the amount of molecular disorder (Coropceanu et al., 2007). Thus, decreasing inter-molecular distance along the π-π stacking (hopping) direction is often considered a design rule for increasing mobility in conjugated polymers.

In this paper, we explore the effects of thermal annealing on the morphology and mobility of PTB7 (poly[[4,8-bis[(2-ethylhexyl)oxy]benzo[1,2-b:4,5-b']dithiophene-2,6-diyl][3-fluoro-2-[(2-ethylhexyl)carbonyl]thieno[3,4-b]thiophenediyl]]), a polymer with chemical structure shown in Figure 1A. The PTB7:PC₇₁BM ([6,6]-phenyl C71 butyric acid methyl ester) bulk heterojunction (BHJ) is of interest because it is among the highest-performing OPV devices, achieving a power conversion efficiency of up to 9.15% (He et al., 2015). Thermal annealing to moderate temperatures has been shown to allow polymer chains to move into a more ordered configuration while fullerenes diffuse out of mixed regions, which can have a favorable effect on device properties (Verploegen et al., 2010). This process has been shown to be suppressed to some extent in PTB7:PC₇₁BM BHJs (Dkhil et al., 2017). Previous work has often focused on annealing of a BHJ blend, making it difficult to separate the effects of material nanophase segregation from the effects of polymer crystallization (Hedley et al., 2017). Thus we first focus on the properties of the neat film of PTB7 during and after thermal annealing, and then extend the results to the BHJ system.

Figure 1.

Polymer Structure

(A) Chemical structure of PTB7.

(B) Illustration of face-on packing of three PTB7 strands on a substrate. Thick lines represent polymer backbones and thin lines represent alkyl side chains. The π-π stacking distance is labeled “d.”

Furthermore, high-temperature annealing steps have been shown to induce cleavage of alkyl side chains in many conjugated polymers. Although alkyl side chains are necessary to solubilize conjugated polymers, their insulating properties make them unfavorable for charge transport. Because of this, there has been considerable interest in eliminating these side chains through thermal cleavage (Sun et al., 2012). Thermo-cleavable side chains have been shown to have several advantages: increase in backbone planarity (Guo et al., 2015), increase in efficiency in some devices, improvement in device stability (Manceau et al., 2010, Helgesen et al., 2011), suppression of PCBM aggregation in blends (Vahdani et al., 2016), and decrease in solubility of the film, allowing multilayer processing (Kuhn et al., 2015). In many cases, cleavage occurs at the ester group, leaving a carboxyl group (Manceau et al., 2010, Helgesen et al., 2011) or removing the ester group entirely (Guo et al., 2015, Kuhn et al., 2015, Höfle et al., 2017, Hillebrandt et al., 2016). In polymers based on the poly(thieno[3,4-b]thiophene-benzo[1,2-b:4,5-b′]dithiophene) (PTB) backbone, cleavage occurs at the ether group, yielding a hydroxyl group (Vahdani et al., 2016). Although the annealing temperatures used in this work are high, the literature has shown that the cleavage temperature can be reduced through inclusion of a catalyst, as in the case of Vahdani et al. (2016), or through engineering of side chains (Hillebrandt et al., 2016).

In this work, we show that moderate-temperature (260°C) annealing decreases the disorder in the PTB7 in neat and blend films with a corresponding increase in mobility in neat PTB7. Unexpectedly, annealing the polymer at temperatures above 260°C produces a previously unseen morphology with a smaller inter-molecular distance between PTB7 backbones in both neat and blend films. Our data suggest that this temperature results in a side chain cleavage reaction and this causes the smaller inter-molecular distance. In contrast to expectations, the mobility of this emerged polymorph is lower despite a decrease in morphological disorder, decrease in π-π stacking distance, and partial removal of insulating side chains. Furthermore, the decay of time-resolved photoluminescence becomes significantly faster after annealing to 290°C. We speculate that the presence of reaction by-products as trap states is responsible for these changes.

Previous work on photoinduced degradation of neat PTB7 yields similar FTIR and UV-Vis spectral changes to our 290°C annealed films (Son et al., 2011). Thus we suggest that formation of trap states is partially responsible for this burn-in performance loss in PTB7 and possibly in similar polymers as well (Upama et al., 2016, Mateker and McGehee, 2017). This indicates that side chain cleavage may be an important mechanism behind photoinduced degradation in OPV and that by-product removal may be effective to regain pre-burn-in efficiency.

Results and Discussion

Mass and Volume Loss Caused by 290°C Annealing is Consistent with Side Chain Cleavage

Thermogravimetric analysis (TGA) was performed on PTB7 powder to test thermal reactions occurring at high temperature. Results are shown in Figures 2A and 2B and in Figure S1A. Two temperature ramps were performed: the first used a continuous ramp of 10°C/min; the second ramped to 290°C at 20°C/min and held the powder at 290°C for 15 min before continuing to ramp at 10°C/min. The first experiment indicated two mass loss transitions in TGA, evidenced by a peak and a shoulder in the dW/dt curve (Figure 2A). The onset of the first transition was at about 260°C. Since no additional mass loss transitions are observed up to a temperature of 500°C, the second mass loss must represent a final degradation stage wherein the backbone breaks down. In the second ramp, the first mass loss transition (the shoulder) was removed without significantly affecting the second mass loss peak. This indicates that the first mass loss mechanism completes in 15 min at 290°C without initiating the backbone degradation associated with the second mass loss peak. The total mass loss upon completion of the first transition is 10%.

Figure 2.

Evidence of a Side Chain Cleavage Reaction at 290°C

(A) Comparison of two TGA experiments: first with a linear ramp and second with temperature held at 290°C for 15 min. The derivative of mass loss is plotted to make mass loss mechanisms easier to distinguish. The first mass loss mechanism (shoulder) marked with an arrow disappears in the second experiment.

(B) Mass loss and temperature over time during the second TGA experiment. The mass loss plateaus at about 10% when held at 290°C.

(C) FTIR of powder pellet, as received and after 290°C anneal (full spectra shown in Figure S1D). The annealed sample has been shifted and normalized to ease comparison. The frequencies discussed in the text are marked with arrows.

See also Figure S1 for (A–C) and Figure S2 for (C).

The double transition in TGA is consistent with the previous literature on side-chain-cleavable materials, in which the first TGA mass loss is attributed to side chain cleavage (Guo et al., 2015, Helgesen et al., 2011, Vahdani et al., 2016, Kuhn et al., 2015). Furthermore, the solubility of PTB7 powder dropped significantly after annealing at 290°C, with visible precipitates remaining in a 10-mg/mL solution in chlorobenzene after stirring for several hours at 60°C. Therefore the most likely explanation for the first mass loss mechanism is a side chain cleave, and the plateauing of the mass loss peak after the 290°C anneal indicates that the polymer backbone is not degraded in this process.

Profilometry on annealed films showed a thickness loss of 15% (±10%) after the 290°C anneal (see Figure S1C). This consistency of mass and volume loss indicates that the resultant film does not have a significantly altered density or porosity due to side chain outgassing, which is speculated in the previous literature (Hillebrandt et al., 2016). The high annealing temperature likely allows reorganization of backbones, which closes any pores that might have formed.

The measured mass loss is less than that expected by side chain cleavage. Cleavage on the thiophene ester group would result in a 15% or 21% mass loss depending on the remaining functional group (carboxyl or none). Cleavage of the ether groups on the benzodithiophene result in a mass loss of up to 30% if both side chains are cleaved. The discrepancy in the measured and the expected mass loss, 10% vs 15%–30%, suggests cleavage by-products remaining in the film and/or incomplete side chain cleavage. Since TGA shows completion of the cleavage reaction with these annealing conditions, the mass loss of only 10% indicates that some by-products remain in the film.

FTIR and UV-Vis Changes Support a Cleavage Reaction and Are Similar between Annealing and Irradiation

FTIR spectra were collected on as-received and annealed PTB7 powder to study the changes in chemical structure, and the results are shown in Figure 2C. The most significant change in the FTIR spectra was the increased absorption at 1,656 and 1,427 cm⁻¹ after annealing. To elucidate the source of these changes, we studied FTIR spectra of a series of potential side-chain-removal intermediate compounds from NIST Chemistry WebBook, Standard Reference Database Number 69 (Linstrom and Mallard, 2017). The FTIR of the annealed PTB7 was compared with the FTIR spectra of simple thiophene and benzodithiophene derivatives (reproduced in Figure S2 for ease of comparison).

Peaks at or near 1,656 and 1,427 cm⁻¹ are present in FTIR spectra of 5-methyl-2-thiophenecarboxylic acid and 2-thiophenecarboxaldehyde (CAS registry numbers 1918-79-2 and 98-03-3) but not in the FTIR spectrum of 2-thiophenemethanol (636-72-6) or thiophene (110-02-1). In addition, the 1,656 cm⁻¹ peak is not present in 2-thiophenecarboxylic acid, ethyl ester (2810-04-0).

This indicates that the structural changes may involve modification or removal of the ester side chain of the thieno[3,4-b]thiophene, resulting in the formation of either a carboxylic acid or an aldehyde group. A similar analysis of a side chain removal reaction on the benzodithiophene indicates that formation of benzoquinone could be responsible for the peak at 1,656 cm⁻¹; however, this would result in significant changes to the UV-Vis absorbance spectrum (Wilke et al., 2013), which is not observed (see below).

Alternatively, a strong peak at 1,656 cm⁻¹ is observed in the FTIR for 2-ethyl-1-hexene (CAS 1632-16-2), which is a potential by-product of a side chain cleavage reaction. This is consistent with side chain cleavage with some by-product trapping.

Thus the most probable cause of the changes in FTIR after annealing is cleavage at the thiophene, removing an 8-carbon segment, with or without a hydroxyl group. It follows that the composition of any by-product of this reaction must be some configuration of C₈H₁₇OH or fragments thereof.

The optical properties were analyzed using UV-Vis spectroscopy, shown in Figure 4C. The spectra are shown as absorption coefficient (see Supplemental Information for details). The main features in the spectra are overlapping absorptions at approximately 628 and 676 nm, which we assign to the 0-1 and 0-0 levels of the π-π* transition, respectively.

Figure 4.

Changes to Neat PTB7 Films with Different Annealing Temperatures

(A) In-plane integration of GIWAXS patterns in Figure 3 between χ = 5°–20° (and 160°–175°), after χ-correction (see Supplemental Information for details).

(B) Out-of-plane integration of GIWAXS between χ = 75°–105°.

(C) UV-Vis absorption spectra of neat PTB7 films.

(D–F) d-Spacing of the π-π stacking peak during in situ heating to 255°C (D), 260°C (E), and 290°C (F). For clarity, measurements during heating and cooling are shown in red and blue, respectively. It should be noted that the sample in (D) was measured with an older PTB7 batch, GIWAXS setup, and calibration procedure than (E) and (F), which explains the slightly different initial d-spacing.

See also Figures S4 and S5 for (A and B), Figure S3 for (C), and Figure S6 for (D–F).

Annealing the film to 120°C and 260°C showed a gradual redshift in absorption, which can be attributed to an increase in the planarity of the polymer backbone, leading to an increase in the extent of electron delocalization, or conjugation length (Li et al., 2005). The shape of the absorption was approximately unchanged on annealing to 120°C and 260°C.

Annealing at 290°C resulted in a noticeable blueshift in the absorption position relative to the films annealed at 260°C, accompanied by a reduction in the absorption primarily at higher wavelengths (i.e., 0-0 level) and an increase in the width. These changes suggest that the film becomes more disordered and that the conjugation length decreases (Li et al., 2005, Kim and Swager, 2001, Bencheikh et al., 2015, Barford and Marcus, 2017). Previous work shows that blueshift in UV-Vis spectra can be associated with decreasing side chain bulkiness (Naik et al., 2012), so this also supports a side chain cleavage reaction. What is not observed is the drastic change in the absorbance spectra that would result from decomposition of the conjugated π-system (backbone).

Both FTIR and UV-Vis changes seen here are remarkably similar to changes observed in PTB7 films after UV irradiation in air (Son et al., 2011). Here, a photochemical reaction on the thiophene, which does not destroy the conjugation, is responsible for the loss in device efficiency. This similarity of changes in FTIR and UV-Vis indicates that the same reaction that we observe under high-temperature annealing may also occur under irradiation in air. Namely, side chain cleavage may be an important mechanism in photoinduced degradation in PTB7, with reaction by-products leading to efficiency loss. A comparison of FTIR and UV-Vis from as-cast, annealed, and irradiated samples is shown in Figures S1B and S3, respectively, and confirms that the opto-electronic changes during annealing (at 290°C) are similar to those arising from solar irradiation in air, suggesting that high-temperature annealing and irradiation result in the same decomposition product produced via side chain cleavage. Furthermore, the film that has been annealed at 290°C and then irradiated shows a UV-Vis spectrum nearly identical to that of the annealed sample (see Figure S3), suggesting that annealing stabilizes the film toward further degradation. This is consistent with increased stability after side chain cleavage seen in the literature (Manceau et al., 2010, Helgesen et al., 2011).

Annealing Results in a Decrease in Morphological Disorder and a Closer Packing Motif

Grazing Incidence Wide Angle X-ray Scattering (GIWAXS) was measured on neat PTB7 films to analyze the morphological changes occurring after increasing the polymer chain mobility (heating up to 260°C) and after cleavage of the side chains (heating to 290°C). The GIWAXS detector images for four neat PTB7 films with different annealing temperatures—as-cast, 120°C, 260°C, and 290°C—are shown in Figures 3A–3D, and the integrated 1D spectra are shown in Figure 4A (in-plane) and Figure 4B (out-of-plane). The as-cast image shows scattering peaks at approximately Q = 0.33 Å⁻¹ primarily in the in-plane direction (along Q_xy) and at Q = 1.7 Å⁻¹ in the out-of-plane direction (along Q_z). These peaks correspond to an alkyl side chain spacing of 19 Å and a π-π stacking distance of 3.7 Å, respectively (Prosa et al., 1992). The orientation of these peaks shows a face-on configuration, illustrated in Figure 1B, consistent with previous measurements of PTB7 (Hammond et al., 2011, Tang and McNeill, 2013, Das et al., 2015).

Figure 3.

2D Scattering Pattern for Neat PTB7 Films

(A–D) (A) as-cast, (B) 120°C anneal, (C) 260°C anneal, and (D) 290°C anneal. Images are plotted with intensity (detector counts) on a logarithmic scale, using the same color scaling for all images. Each image represents a 10-min exposure. The yellow spots at Q_xy≈1.9, Q_z≈1.2 Å⁻¹ are diffuse scattering from the silicon substrate.

See also Figures S4 and S5.

Data from the sample annealed to 120°C shows good agreement with those from the as-cast sample. In contrast, the 260°C annealed sample shows a decrease in peak width along Q from the as-cast sample. After the 260°C anneal, the peak at Q = 0.33 Å⁻¹ splits into two overlapping peaks (or, a peak plus a shoulder). The broad alkyl peak in the as-cast sample is at an intermediate position between the overlapping peaks in the annealed sample, so it is reasonable to assume that both peaks are present but not resolvable in the as-cast material. The stronger of these peaks can be attributed to alkyl stacking, but the source of the second peak is not immediately clear. It is possible that this peak is due to the spacing along the polymer backbone. The peak position corresponds to a spacing of about 16 Å (see Figure S5), which is close to the expected monomer length in PTB7.

In addition to the decrease in peak width, a decrease in the angular width of both the alkyl stacking peak and the π-π stacking peak after annealing to 260°C is noticeable in the pole figures (Figure S4). These observations show a significant decrease in disorder and increase in the degree of orientation in the polymer, which is consistent with the effects of annealing found for other polymers (Ruderer and Muller-Buschbaum, 2011). The decrease in disorder is also consistent with the redshift seen in the UV-Vis spectra.

Annealing to 290°C gave an in-plane scattering pattern similar to that of the 260°C annealed sample. Remarkably, however, the 290°C annealed sample showed a shift in the π-π stacking peak from Q = 1.69 Å⁻¹ to Q = 1.80 Å⁻¹, or from d = 3.73 Å to d = 3.49 Å (Table S1). This 6% decrease in the π-π stacking distance represents a closer packing structure previously unseen in PTB7. This finding is consistent with the removal of side chains, as steric interactions between side chains can push polymer backbones apart to larger π-π stacking distances, especially for branched side chains (Lei et al., 2014, Oosterhout et al., 2017).

The UV-Vis showed an increase in opto-electronic disorder when the annealing temperature was increased from 260°C to 290°C, whereas GIWAXS showed similar peak intensities and orientations for the 260°C and the 290°C annealed samples. Thus changes in the opto-electronic properties are not correlated directly with changes in morphology as observed with X-ray diffraction. This suggests that by-products in the film created by side chain cleavage cause the observed changes in opto-electronic structure (e.g., form traps) without affecting the polymer packing.

An in situ annealing study was performed to track the dynamics of morphological changes. In-plane and out-of-plane integrated curves for in situ GIWAXS experiments are shown in Figure S6. The first as-cast sample was gradually heated to 255°C and cooled to room temperature while measuring GIWAXS in a 4-hr procedure. The π-π stacking peak shifted to higher d-spacing upon heating and reversibly shifted back during cooling as shown in Figure 4D, which is due to thermal expansion and contraction.

The second sample was gradually heated to 260°C and cooled during a 2-hr procedure (Figure 4E), and the third sample was gradually heated to 290°C and cooled during a 4-hr procedure (Figure 4F). The π-π stacking distance decreased irreversibly while the temperature was above an onset temperature of about 260°C and otherwise followed thermal expansion and contraction trends. Post 290°C anneal, a decrease in π-π stacking distance of 11% was observed. These data show that the decrease in the π-π stacking distance can be associated with the side chain cleavage that occurs above 260°C. The larger change in the distance of the in situ annealed films (11% for in situ versus 6% for ex situ) is likely due to the film being held at high temperatures for a longer time period during in situ annealing, allowing a more complete reaction.

The linear coefficient of thermal expansion (CTE) was calculated to be 3.6 × 10⁻⁴/°C for the sample annealed to 255°C, 2.9 and 2.5 × 10⁻⁴/°C for the sample annealed to 260°C (for heating and cooling, respectively), and 3.2 and 2.7 × 10⁻⁴/°C for the sample annealed to 290°C. These values are consistent to within a factor of two of CTE estimated for the lamella spacing of a thin film of P3HT (∼4.8 × 10⁻⁴/°C) (Verploegen et al., 2010).

In situ annealed samples were cooled slowly, unlike ex situ annealed samples, which were quenched (details in Supplemental Information). The previous literature for P3HT:PCBM has shown that rapid quenching can result in non-equilibrium morphological states (Wang et al., 2011). The consistency of the morphology change in quenched and gradually cooled samples indicates that the closer packing motif is not a kinetically trapped state but is nearly energetically stable.

Mobility Decreases due to Side Chain Cleavage

The as-cast and annealed neat PTB7 films were further analyzed using space charge limited current (SCLC) measurements. This technique measures charge transport mobility in the out-of-plane direction, which is most influenced by hopping in the π-π stacking direction of this face-on polymer. The device schematic used is shown in Figure S8A, and resulting J-V curves are shown in Figure S8B. Hole mobility was calculated from SCLC measurements using the Mott-Gurney equation:

$$\text{J=}\frac{\mathrm{9}}{\mathrm{8}}{\mathrm{\varepsilon }}_0{\mathrm{\varepsilon }}_{\mathrm{r}}\mathrm{\mu }\frac{{\mathrm{V}}^{\mathrm{2}}}{{\mathrm{L}}^{\mathrm{3}}}$$

where J is the current density, V is the applied effective voltage, L is the thickness of the active layer, ɛ₀ is the absolute permittivity, ɛ_r is the relative permittivity (taken as 3), and μ is the hole mobility (Lampert and Mark, 1970, Blakesley et al., 2014). To confirm the reliability of the SCLC measurement, the thickness dependence of J (J∝L⁻³) was shown to be consistent with the Mott-Gurney equation (Figure S8C).

The hole mobility of the PTB7 measured as a function of annealing temperature is shown in Table 1. Each result shown is from an average of eight devices. The room temperature mobility of as-cast PTB7 was within the mobility range found in the literature using SCLC (Foster et al., 2014, Liang et al., 2010, Zhou et al., 2014). The mobility increased significantly from the as-cast device to the 120°C annealed device and from the 120°C annealed device to the 260°C annealed device (∼2–3 times). Thus an increase in mobility accompanies the increase in conjugation length and decrease in morphological disorder, and this can be achieved through moderate-temperature annealing. This supports the picture in semiconducting polymers that increased order within the same polymer leads to higher mobility.

Table 1.

Hole Mobility of Neat PTB7 Films, Measured Using SCLC

Processing Temp (°C)	Hole Mobility: (cm2/Vs)
As-cast	(2.85 ± 1.3) × 10−4
120°C	(4.7 ± 0.9) × 10−4
260°C	(10.5 ± 1) × 10−4
290°C	(0.82 ± 0.55) × 10−4

Each value is the average mobility for eight devices and ± error is the standard deviation. See also Figure S8.

Conversely, the mobility fell by over an order of magnitude after annealing to 290°C, consistent with the higher electronic disorder seen in UV-Vis. Previous work has also reported that a decrease in π-π spacing does not always correlate with increased mobility; however, this work instigates an unfavorable change in polymer crystallinity (Dou et al., 2014), which we do not observe in GIWAXS. Our result is significant, as it shows that an improved morphology (decrease in disorder and decrease in π-π spacing) and favorable removal of insulating side chains may come with complex adverse side effects, leading to an overall decrease in electronic performance.

We speculate that two factors may contribute. The first is that by-products of the side chain cleavage reaction cause an increase in electronic disorder that leads to a decrease in mobility. The second is the possibility of carrier trapping at the post-cleavage functional group such as proton transfer at a carboxylic acid.

The Effect of Annealing on PTB7:PC₇₁BM Blend Films is Consistent with that on Neat PTB7

We have shown that neat PTB7 films become better ordered with annealing and undergo a side chain cleavage reaction above 260°C, which decreases intermolecular spacing. In this section, we show that these morphological changes are also observed in PTB7:PC₇₁BM blends. Furthermore, the mobility of neat PTB7 decreases significantly post cleavage, which we attribute to the inclusion of reaction by-products. Here, we give evidence for the existence of trapped by-products by observing faster decay of time-resolved photoluminescence in PTB7:PC₇₁BM blends. This shows that our findings can be applied to BHJ-based devices. BHJ films were spin cast from a PTB7:PC₇₁BM 1:1.5 ratio in chlorobenzene with a total solids concentration of 10 mg/mL.

First, we examine GIWAXS results for blend films, shown in Figure S7. Accurate quantitative analysis of blend data is complicated by the overlap between PC₇₁BM peaks at ∼1.3 and 1.9 Å⁻¹ and the PTB7 π-π stacking peak at ∼1.6 Å⁻¹. However, changes to the GIWAXS patterns are clearly visible in the out-of-plane integrated image in Figure 5. Here, annealing to 120°C gives a pattern similar to that of the as-cast film, whereas annealing to 260°C gives a distinct sharpening of the PTB7 π-π stacking peak. This behavior is identical to that observed in neat PTB7 films. Furthermore, annealing the BHJ to 290°C causes the PTB7 π-π stacking peak to shift to a higher Q, with the peak shifting from approximately Q = 1.57 to Q = 1.75 Å⁻¹ (d = 4.00 to 3.59 Å). This is a change of about 11%, consistent with the neat film. However, the PTB7 π-π stacking peak overlaps strongly with the PC₇₁BM peaks at 1.3 and 1.9 Å⁻¹, making determination of the π-π peak position inaccurate and explaining the slight difference with Figure 4 (neat film data). These results show that PTB7 morphology behaves similarly in blend films as in neat films and that mixing with a fullerene does not hinder the side chain cleavage reaction presented here.

Figure 5.

Changes in GIWAXS Patterns of Blend PTB7:PC₇₁BM Films with Different Annealing Temperatures

Integrated out-of-plane cake slice (75°–105°) from Figure S7 in the region of the π-π stacking peak, with arrows showing positions of PC₇₁BM and PTB7 peaks. Arbitrary normalization.

See also Figure S7.

With GIWAXS results indicating a closer packing arrangement due to side chain cleavage, we next examine how cleavage by-products affect the blend films compared with neat films. Side chain cleavage by-products may act as recombination sites, resulting in faster exciton quenching in PC₇₁BM blends. This was studied using time-resolved photoluminescence (TR-PL), with a 515 nm excitation and emission in the range 670–720 nm. In previous work with a 400 nm excitation, photoluminescence from PTB7 was quenched very quickly in this blend, so only PC₇₁BM emission was detected using this method (Hedley et al., 2017). In the present work the detection wavelength range selects for emission from PC₇₁BM. The emission had two components: a fast component, which occurs on a timescale of several picoseconds, and a slow component (>∼10 picoseconds). Previous work indicates that fast quenching may be due to hole transfer or resonant energy transfer from the fullerene to the polymer in intimately mixed polymer:fullerene regions, in which the sizes of PC₇₁BM domains are less than 10 nm. Thus the degree of fast quenching (proportional to the change in PL in the first few picoseconds after excitation) loosely corresponds to the amount of small PC₇₁BM domains, and domain growth will lead to less quenching on a fast timescale.

The slow component of quenching results from quenching at the polymer:fullerene interface after exciton diffusion from within larger PC₇₁BM domains. The diffusion distance increases with increasing domain size and increasing domain purity. Including even a small amount of impurity in a fullerene domain (e.g., 0.2 weight % PTB7) has been shown to significantly increase the rate of quenching, corresponding to a faster time constant in the TR-PL curve (Hedley et al., 2017).

TR-PL data from PTB7:PC₇₁BM films annealed to various temperatures are shown in Figure 6. From the ratio of maximum luminescence to luminescence after 10 ps, we can see a decrease in the fast-timescale quenching from the as-cast film to 120°C, and from 120°C to 260°C. This is consistent with the increased phase separation in PTB7-Th:PC₆₀BM observed in previous work: the level of mixing decreases and domain sizes increase with annealing (Hsieh et al., 2017). Notably, the 260°C- and 290°C-annealed BHJ films show similar amounts of fast-timescale quenching, which may be due to the film reaching a stable morphology. A saturation of thermally induced intermixing has been observed in PTB7:PC₆₀BM previously (Liu et al., 2014), and so it is not unexpected that an equilibrium may be reached.

Figure 6.

TR-PL Data from Blend Films As-cast and Annealed to 120°C, 260°C, and 290°C

Data have been normalized by the maximum luminescence at 0 ps and smoothed for visibility.

The slow-timescale quenching, in contrast, is fairly similar between as-cast, 120°C, and 260°C annealed films, but becomes substantially faster after annealing at 290°C. Because the fast-timescale quenching is similar between 260°C and 290°C, this is most likely due to the existence of trap states in PC₇₁BM domains that are created by the cleavage reaction. This would arise if some of the cleavage products, which have low molecular weight and are therefore mobile in a hot film, diffuse into the fullerene domains and quench the PL. It supports our conjecture that cleavage by-products remain in the film, disrupting mobility and exciton diffusion by acting as trap states.

GIWAXS and TR-PL on BHJ films show that annealing effects are similar in neat and blend films. Up to 260°C, PTB7 becomes better ordered and more segregated from PC₇₁BM. Above 260°C, a side chain cleavage reaction occurs, which decreases the π-π stacking distance of PTB7 and results in by-product trap states in both the neat PTB7 and in PC₇₁BM-rich domains of the blend.

Conclusions

In this work, we have shown that moderate-temperature annealing leads to an improvement in film morphology and a 2–3x increase in hole mobility. High-temperature annealing leads to a side chain cleavage reaction in PTB7, which coincides with a decrease in intermolecular distance, both in neat and PC₇₁BM blend films. These results are important for organic electronics because the removal of insulating side chains and decrease in intermolecular distance are expected to improve device efficiency. However, the presence of reaction by-products, likely composed of C₈H₁₇OH and/or fragments thereof, leads instead to a decrease in mobility and faster decay of the PL. The similarity between thermally cleaved films and irradiated films leads us to suggest that side chain cleavage is a mechanism responsible for burn-in.

The results presented here show that thermally induced side chain cleavage is possible in high-performing materials without additional structural modification and that this cleavage process can lead to favorable changes in molecular packing. Although the cleavage temperature used here is too high to be a viable processing step for flexible substrates used in OPV, previous work suggests that reduction of cleavage temperature can be achieved through the inclusion of a catalyst (Vahdani et al., 2016) or by appropriately engineered side chains (Hillebrandt et al., 2016). Additional processing steps may be developed to target and remove trap states formed in this reaction, and the decrease of solubility of the polymer backbones without side chains can aid this process. For example, the polymer matrix may be swelled using an appropriate solvent wash, allowing the release of volatile hydrocarbon by-products. The correlation of side chain cleavage with photoinduced degradation indicates that these same washing procedures may be used to remove the damaging sub-gap trap states that form during burn-in.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.

Published: March 22, 2018