Correlating the scattered intensities of P3HT and PCBM to the current densities of polymer solar cellswz

Transcription

1 View Online COMMUNICATION ChemComm Correlating the scattered intensities of P3HT and PCBM to the current densities of polymer solar cellswz Enrique D. Gomez,* a Katherine P. Barteau, a He Wang, ab Michael F. Toney c and Yueh-Lin Loo* a Received 30th July 2010, Accepted 27th August 2010 DOI: /c0cc02927k Downloaded by Princeton University on 27 September 2011 Published on 30 September 2010 on doi: /c0cc02927k Grazing-incidence X-ray diffraction and rocking scans have quantified the structure of poly(3-hexylthiophene) and [6,6]-phenyl-C 61 -butyric acid methyl ester in the active layers of organic solar cells. Our study reveals that the device J SC correlates with the local structural development of pure PCBM and, to second order, the extent of out-of-plane P3HT p-stacking. It is well documented that the processing conditions of organic solar cells comprising poly(3-hexylthiophene), P3HT, can have strong influences on device performance. 1 5 For example, thermal or solvent-vapor annealing of the active layer has led to increases in the short-circuit current densities (J SC )of P3HT/[6,6]-phenyl-C 61 -butyric acid methyl ester, or PCBM, solar cells; this improvement in device performance has been attributed to enhancements of the polymer crystallinity. 6 9 Given that the active layer consists of a blend of a polymer and a small molecule, its phase separation and crystallization characteristics are necessarily complex and interdependent on the aggregation and crystallization of the constituents. Yet, the details of how such processing affect the structuring of PCBM remains unclear. We have examined, for the first time, the structure of the individual constituents in the active layers of bulk heterojunction organic solar cells comprising P3HT and PCBM in relation to the device J SC by quantifying the X-ray scattered intensity acquired during both grazing-incidence X-ray diffraction and rocking scan (or local specular) experiments. We find the structure of PCBM is most important in determining device J SC ; the extent of out-of-plane p-stacking of P3HT is a secondary effect in influencing J SC. These experiments imply that the overall P3HT crystallinity in itself is not the sole factor governing device performance. Figs. 1a and b show two-dimensional grazing-incidence X-ray diffraction (GIXD) patterns acquired on 150 nm thick P3HT/PCBM films deposited on an indium tin oxide (ITO) substrate pre-coated with poly(3,4-ethylenedioxythiophene) that is doped with poly(styrene sulfonate), or PEDOT : PSS. The data shown in Fig. 1a were taken prior to thermal a Department of Chemical and Biological Engineering, Princeton University, Princeton, NJ 08544, USA. edg12@psu.edu, lloo@princeton.edu b Department of Electrical Engineering, Princeton University, Princeton, NJ 08544, USA c Stanford Synchrotron Radiation Lightsource, Menlo Park, CA 94025, USA w This article is part of the Emerging Investigators themed issue for ChemComm. z Electronic supplementary information (ESI) available: Methods and X-ray data and analysis. See DOI: /c0cc02927k Fig. 1 Two-dimensional grazing-incidence X-ray diffraction images of P3HT/PCBM films (a) not annealed and (b) annealed at 165 1C for 240 min. The films were cast on PEDOT : PSS/ITO to resemble the structure of the active layer in solar cells. (c) In-plane radial traces of GIXD data from (a) and (b) taken at q z = 0.1 Å 1. The data from the annealed film are shifted along the y-axis for comparison. (d) Full pole figures of the same P3HT/PCBM films as a function of the polar angle, o. annealing of the active layer while the diffraction pattern in Fig. 1b was acquired on nominally the same film after annealing at 165 1C for 240 min. Reflections that are characteristic of P3HT and PCBM have been labeled for clarity. In Figs. 1a and b, the (100) reflection of P3HT (q E 0.4 A 1 )is most intense at the meridian, indicating that P3HT is preferentially oriented with its (100) plane parallel to the substrate. PCBM reflections are visible near 1.4 A 1. Fig. 1c shows the in-plane traces, selected at q z = 0.1 A 1, of the data shown in Figs. 1a and b. After annealing, we observe a narrowing of the reflections accompanied by a concomitant increase in intensities. Furthermore, three individual reflections for PCBM become visible after annealing. 10 The PCBM reflections remain azimuthally isotropic in intensity, suggesting that after annealing PCBM crystallizes without a preferred orientation. We note that a broad PCBM halo is observed (see Fig. 1a) in the majority of our samples, indicating PCBM is amorphous. Detailed examination of the intensities in this q-range reveals variations in the shape and intensity of the PCBM amorphous halo. Such changes in the amorphous halo can be attributed to local structural changes associated with PCBM, likely stemming from aggregation prior to crystallization. It follows that the PCBM scattered intensity is proportional to the amount of phase separated PCBM dispersed in P3HT. To properly examine the structural development of PCBM in these samples, we have thus tracked the changes in the integrated intensities associated with this PCBM scattering. 436 Chem. Commun., 2011, 47, This journal is c The Royal Society of Chemistry 2011

2 Downloaded by Princeton University on 27 September 2011 Published on 30 September 2010 on doi: /c0cc02927k View Online We can obtain a quantitative estimate of the P3HT crystallinity by constructing full pole figures from a combination of GIXD and X-ray rocking experiments as shown in Fig. 1d.11,12 From the pole figure analysis, we have determined the crystallinity of P3HT to be 3-fold higher in the P3HT/PCBM film that was annealed at 165 1C for 240 min compared to that of the unannealed film. We also estimated the P3HT crystallinity of P3HT/PCBM films prepared under a wide range of conditions using the same pole figure analysis. Films of P3HT and PCBM, for example were thermally annealed for 5, 10, 30 or 240 min at 50, 130 or 165 1C. These films were prepared on ITO, as well as ITO pre-treated with n-tetradecylphosphonic acid (HSAM), 1H,1H,2H,2H-perfluorotetradecylphosphonic acid (FSAM), or pre-coated with PEDOT : PSS to simulate the active layers that are deposited on different anodes in conventional solar cells. Altogether, 19 different processing conditions were explored. Given the strong dependence of the device JSC on the morphology of the active layer,6 9 we have chosen to compare the P3HT and PCBM scattered intensity to the device JSC. In Fig. 2, we have compiled the extracted JSC of all tested devices as a function of the integral of the product of the full pole figure (Fig. 1d) and sin(o) over all orientations. This quantity is proportional to the P3HT crystallinity. The JSC varies between 4 to 10 ma cm 2 across all device processing conditions we explored. Consistent with prior reports,8,9 devices with unannealed active layers having low P3HT crystallinities generally exhibit JSC that are markedly lower than those of devices with annealed active layers. We do not, however, observe a definitive correlation between the P3HT crystallinity and the device JSC that encompasses all of our samples. In particular, Fig. 2 emphasizes that while the P3HT crystallinity is important in determining JSC in many of our samples, it is not the only structural property that influences JSC. In what follows, we have quantified our X-ray data to elucidate how the structural development of PCBM influences device JSC. To fully capture the structural development of PCBM, we should ideally construct full pole figures by stitching the GIXD data with the appropriate rocking scans at the PCBM reflection (q = 1.4 A 1). The acquisition of rocking Fig. 2 Device JSC vs. normalized P3HT crystallinity of films annealed at 50, 130 or 165 1C for 0, 5, 10, 30 or 240 min. The substrate or anode used was either PEDOT : PSS (PEDOT), ITO, HSAM or FSAM. The P3HT crystallinity is normalized by the largest P3HT crystallinity found in our samples. Error bars denote standard deviations obtained from multiple experiments. This journal is c The Royal Society of Chemistry 2011 Fig. 3 Device JSC vs. normalized PCBM scattered intensity obtained from GIXD of P3HT/PCBM films annealed at 50, 130 or 165 1C for 0, 5, 10, 30 or 240 min. The substrate or anode used was either PEDOT : PSS (PEDOT), ITO, HSAM or FSAM. The PCBM scattered intensity is normalized by the largest PCBM scattered intensity in our samples. The lines are guides to the eye. Error bars denote standard deviations obtained from multiple experiments. curves at q = 1.4 A 1, however, requires a relatively large X-ray incident angle of 6.21 (for l = A ). At such angles, the X-rays penetrate deep into the specimen, increasing the background contribution to the X-ray diffraction data, which introduces large uncertainties to the diffracted intensity from the organic thin film. To obviate these problems, we opted to only utilize GIXD data to obtain the PCBM scattered intensity by first normalizing the GIXD intensity of the PCBM reflection at 1.4 A 1 against that of the P3HT (100) reflection acquired during the same scan. This normalization allows us to account for differences in scattered intensities due to small changes in the incident angle during GIXD experiments and enables a meaningful comparison across all samples. For further details, see discussion in the ESIz. Fig. 3 shows the device JSC as a function of the extracted PCBM scattered intensity for organic solar cells processed with HSAM-treated, FSAM-treated and PEDOT PSS-coated anodes, as well as those thermally annealed at different temperatures for varying times. As in Fig. 2, the integrated scattered intensity is normalized by the highest PCBM scattered intensity in our samples for ease in comparison. At normalized PCBM scattered intensity of 0.6 or less, JSC of devices comprising these films is linearly and positively correlated with the PCBM scattered intensity irrespective of processing history. At normalized PCBM scattered intensity 40.6, JSC appears to be independent of PCBM scattered intensity. Interestingly, P3HT/PCBM cast on PEDOT : PSS-coated ITO generally exhibits a higher PCBM scattered intensity, and accordingly, these devices exhibit higher JSC s. The reason for this observation is unclear at this point, though variations in the surface roughness and surface energy between ITO, HSAM- and FSAM-treated ITO or PEDOT : PSS on ITO could lead to differences in the crystallization and aggregation behavior of PCBM (see ESIz for more details). Unlike the operation of organic thin-film transistors in which charge transport occurs along the dielectric surface, charge transport occurs through the thickness of the active layer during organic solar cell operation. Given that charge transport in P3HT is anisotropic and is favored along the p-stacking direction,13,14 the orientation of P3HT crystallites Chem. Commun., 2011, 47,

3 View Online Downloaded by Princeton University on 27 September 2011 Published on 30 September 2010 on doi: /c0cc02927k Fig. 4 Device J SC vs. normalized P3HT rocking curve intensity at o = 901, proportional to the extent of out-of-plane p-stacking of P3HT, from P3HT/PCBM films annealed at various temperatures. The P3HT intensity is normalized by the largest P3HT intensity in our samples. The samples included in this figure have PCBM scattered intensities The line is a guide to the eye. Error bars denote standard deviations obtained from multiple experiments. must affect the J SC we measure during device operation. To quantify this structure function relationship, we examined the extent of out-of-plane p-stacking of P3HT within active layers of our organic solar cells. We estimate the extent of out-of-plane P3HT p-stacking from the intensity of our P3HT rocking curves near o =901. The P3HT intensity is then normalized by the highest value in our samples for comparison. To determine the factors governing device J SC at normalized PCBM scattered intensity 4 0.6, we plot J SC as a function of the normalized P3HT rocking curve intensity near o =901in Fig. 4. We observe a linear correlation between the P3HT rocking curve intensity near o = 901, which is related to the extent of out-of-plane p-stacking, and J SC. In the ESIz, we also obtain an estimate of the extent of out-of-plane p-stacking from GIXD alone. Similar to the results shown in Fig. 4, we obtain a linear trend between the estimated out-ofplane P3HT p-stacking and J SC. The operation of polymer solar cells requires efficient hole and electron conduction; one should thus only expect optimal performance when electron and hole transport through the active layer is balanced. Our results in Fig. 3 suggest that establishing local organization of PCBM in the active layer, as evident from the high PCBM scattering intensity, is critical for high-performance P3HT/PCBM devices. On the other hand, when the PCBM local organization in the active layer is sufficiently high (B0.6 in Fig. 3), and we postulate when percolation of PCBM domains throughout the active layer occurs, the device J SC ceases to be sensitive to the local organization of PCBM. Instead, intermolecular hole transport along the p-stacking direction of P3HT 13,14 influences device J SC. Careful examination of the GIXD patterns shown in Fig. 1 does in fact suggest the presence of multiple populations of P3HT crystals, from crystals oriented with their (100) planes parallel to the substrate to crystals oriented with their (010) planes parallel to the substrate. Phase separation of P3HT and PCBM in the active layer can have a strong effect on device performance. 8,15 18 The existence of percolating networks with a domain size near the exciton diffusion length (ca. 10 nm) 19 is important to prevent recombination losses and promote efficient charge transport. 20 Phase separation is notoriously difficult to quantify; while this work does not directly address this phenomenon in P3HT/PCBM active layers, it highlights the structural changes in the active layer that do have strong influences on device performance. MRSEC funding through Princeton s NSF-sponsored Princeton Center for Complex Materials is acknowledged. Funding from the Sloan Foundation and the Photovoltaics Program at the Office of Naval Research (N ) are also acknowledged. Portions of this research were carried out at the Stanford Synchrotron Radiation Lightsource, a national user facility operated by Stanford University on behalf of the U.S. Department of Energy, Office of Basic Energy Sciences. We gratefully acknowledge George Malliaras and Yee-Fun Lim for advice with the initial set up of our solar cell testing facilities, Brad Olsen and Shong Yin for providing Igor routines and codes for X-ray data reduction, and Jeffrey Schwartz for providing the phosphonic acid molecules. Notes and references 1 B. C. Thompson and J. M. J. Frechet, Angew. Chem., Int. Ed., 2008, 47, R. Kroon, M. Lenes, J. C. Hummelen, P. W. M. Blom and B. de Boer, Polym. Rev., 2008, 48, M. C. Scharber, D. Wuhlbacher, M. Koppe, P. Denk, C. Waldauf, A. J. Heeger and C. L. Brabec, Adv. Mater., 2006, 18, H. Hoppe and N. S. Sariciftci, Adv. Polym. Sci., 2008, 214, G. Li, V. Shrotriya, Y. Yao, J. S. Huang and Y. Yang, J. Mater. Chem., 2007, 17, G. Li, V. Shrotriya, J. S. Huang, Y. Yao, T. Moriarty, K. Emery and Y. Yang, Nat. Mater., 2005, 4, H.-Y. Chen, H. Yang, G. Yang, S. Sista, R. Zadoyan, G. Li and Y. Yang, J. Phys. Chem. C, 2009, 113, C. W. Chu, H. C. Yang, W. J. Hou, J. S. Huang, G. Li and Y. Yang, Appl. Phys. Lett., 2008, 92, T. Erb, U. Zhokhavets, G. Gobsch, S. Raleva, B. Stuhn, P. Schilinsky, C. Waldauf and C. J. Brabec, Adv. Funct. Mater., 2005, 15, J. B. Kim, S. S. Lee, M. F. Toney, Z. Chen, A. Facchetti, Y. S. Kim and Y. L. Loo, Chem. Mater., 2010, 22, J. L. Baker, L. H. Jimison, S. Mannsfeld, S. Volkman, S. Yin, V. Subramanian, A. Salleo, A. P. Alivisatos and M. F. Toney, Langmuir, 2010, 26, L. H. Jimison, A. Salleo, M. L. Chabinyc, D. P. Bernstein and M. F. Toney, Phys. Rev. B: Condens. Matter Mater. Phys., 2008, 78, H. Sirringhaus, P. J. Brown, R. H. Friend, M. M. Nielsen, K. Bechgaard, B. M. W. Langeveld-Voss, A. J. H. Spiering, R. A. J. Janssen, E. W. Meijer, P. Herwig and D. M. de Leeuw, Nature, 1999, 401, L. H. Jimison, M. F. Toney, I. McCulloch, M. Heeney and A. Salleo, Adv. Mater., 2009, 21, C. R. McNeill, S. Westenhoff, C. Groves, R. H. Friend and N. C. Greenham, J. Phys. Chem. C, 2007, 111, V. D. Mihailetchi, H. X. Xie, B. de Boer, L. M. Popescu, J. C. Hummelen, P. W. M. Blom and L. J. A. Koster, Appl. Phys. Lett., 2006, 89, Y. Yao, J. H. Hou, Z. Xu, G. Li and Y. Yang, Adv. Funct. Mater., 2008, 18, J. Peet, J. Y. Kim, N. E. Coates, W. L. Ma, D. Moses, A. J. Heeger and G. C. Bazan, Nat. Mater., 2007, 6, P. Peumans, A. Yakimov and S. R. Forrest, J. Appl. Phys., 2003, 93, P. K. Watkins, A. B. Walker and G. L. B. Verschoor, Nano Lett., 2005, 5, Chem. Commun., 2011, 47, This journal is c The Royal Society of Chemistry 2011

4 Supplementary Information for Correlating the scattered intensities of P3HT and PCBM to the current densities of polymer solar cells Enrique D. Gomez*, Katherine P. Barteau, He Wang, Michael F. Toney, and Yueh-Lin Loo* 1. Methods 1.1. Fabrication of devices and samples for X-ray scattering experiments Indium tin oxide (ITO) on glass (Colorado Concept Coatings, nm thickness, 9-15 Ω/sq.) was used for this study. Substrates were modified by either depositing a molecular layer of 1H,1H,2H,2H-perfluorotetradecylphosphonic acid, FSAM (Specific Polymers), n- tetradecylphosphonic acid, HSAM, (Alfa Aesar) or 100 nm of poly(3,4-ethylenedioxythiophene) poly(styrenesulfonate), PEDOT:PSS, (Clevios P, H.C. Starck) on ITO. Deposition of HSAM or FSAM were carried out at ambient conditions by first dissolving the molecule at 0.2 mm concentration in anhydrous THF and then submerging the pre-cleaned ITO/glass substrates in the solution for 16 hrs. Afterwards, the substrates were rinsed extensively with THF and methanol. The substrates were then placed in an oven for 24 hrs at 130 o C before they were rinsed again with isopropanol or methanol. As-purchased PEDOT:PSS dispersions were spin coated onto ITO at 2000 rpm for 120 s. The PEDOT:PSS film, approximately 100 nm thick, was annealed at 170 o C for 20 min. All substrate processing was carried out at ambient conditions. Deposition of the active layer on ITO and modified ITO substrates was carried out inside a nitrogen glove box (<1ppm O 2 and H 2 O) equipped with an integrated spin coater and thermal evaporator. P3HT (50 kg/mol M w, 96% H-T regioregular, Merck) and PCBM (Nano-C) were codissolved in chlorobenzene at a 1:0.8 P3HT to PCBM mass ratio and a total concentration of 1

5 24 mg/ml. The solution was annealed for 24 hrs at room temperature, and then heated to 90 o C for 10 s on a hot plate before it was spin coated at 500 rpm to form a uniform film of 150 nm. All thermal annealing took place inside the glove box on a digital hot plate equipped with an independent surface thermometer prior to top electrode deposition. Aluminum was evaporated at 10-6 torr at Å/s for the first 20 nm and 1 Å /s for the last 80 nm, for a total metal thickness of 100 nm. Samples for X-ray diffraction studies were prepared in an analogous fashion, but aluminum was not deposited on the photoactive layer surface Device testing After solar cell fabrication, devices were transferred into an attached glove box for testing without exposure to the atmosphere. A 300 W Xe solar simulator (Newport) was used with an AM1.5G filter and metallic fused-silica neutral density filters (Melles Griot) to match the solar spectrum at the standard 100 mw/cm 2 illumination intensity. J-V characteristics were obtained by making contact through indium-coated probes connected to a Keithley 2600 series source meter controlled through an in-house Labview program X-ray scattering experiments X-ray diffraction experiments took place on beamline 11-3 at the Stanford Synchrotron Radiation Laboratory. The X-ray wavelength was /Å. Rocking curves were obtained by rocking the sample (1.55 to 1.8 o ) during data acquisition. Grazing incidence diffraction experiments were carried out at an incident angle of 0.11 o. Five sequential scans were performed for grazing-incidence experiments and were averaged together. The data were corrected for scattering from air, scattering from the substrate and for the polarization of incident X-rays (98%, in-plane). Scattering intensities were normalized by the incident photon flux and acquisition time. 2

6 1.4. X-ray data analysis For the rocking curves, X-rays scattered for a polar angles, ω, of o (ω = 0 o at the substrate normal) are blocked by the substrate. We have thus opted to stitch together our GIXD data with that acquired during rocking experiments to construct full pole figures of the P3HT (100) reflection. The stitching process is done by scaling the grazing-incidence data to match the rocking data near ω = 55 o, where ω is the polar angle (see Figure 1 of the main text) 1. Rocking data is used for ω < 55 o, while grazing-incident data is used for ω > 55 o. The construction of these full pole figures enables quantitative comparison of the extent of crystallinity of P3HT across different samples. Figure 1d of the main text shows the full pole figures for the same P3HT/PCBM films on which the GIXD data in Figures 1a and b were acquired. The sharp downturn in the intensity near ω = - 90 o is an artifact of the experimental setup as the beam stop blocks any scattered intensities in this range. To compute the P3HT crystallinity, we integrate the product of the pole figure and sin(ω) over all ω. Sin(ω) is a necessary correction for the thin film geometry. Absolute quantification of the extent of crystallinity of P3HT is notoriously difficult and would require, for example, experimentation with a sample of known crystallinity. We compute the PCBM scattered intensity by first normalizing the GIXD intensity of the PCBM reflection at 1.4 1/Å against that of the P3HT (100) reflection acquired during the same scan. This normalization allows us to account for differences in scattered intensities due to small changes in the incident angle during GIXD experiments and enables a meaningful comparison across all samples. It follows that this ratio is directly proportional to the structuring of PCBM after it is normalized against the crystallinity of P3HT, as extracted from the full pole figure analysis outlined above: 3

7 I PCBM,corrected / I P3HT,crystallinity = I PCBM,GIXD / I P3HT,GIXD (S1) here I PCBM,corrected is a quantity that is directly proportional to the PCBM scattered intensity; I P3HT,crystallinity is the extracted intensity from the full pole figure analysis of the P3HT (100) reflection and is therefore proportional to the crystallinity of P3HT; I PCBM,GIXD and I P3HT,GIXD represent the integrated intensities of the PCBM and P3HT reflections from GIXD data, respectively. Due to the experimental geometry, the GIXD data does not include scattering from crystals oriented within ~ 2 degrees of the surface normal. In calculating I P3HT,GIXD, we correct for this effect using the rocking curve data near ω ~ 0 o. Furthermore, equation 1 assumes that the orientation of P3HT does not change significantly between samples, an assumption that is valid given that we do not observe large changes in the P3HT(100) pole figures across our samples. The integrated intensity of the PCBM reflections is obtained through deconvolution of the PCBM peaks as described in Section 3 below. Applying equation S1 to the extracted intensities of the GIXD traces thus allows quantitative comparisons of the PCBM scattered intensity among all samples. We take the measured PCBM scattered intensity to be directly related to the amount of PCBM in PCBM/P3HT films which adopts the same local structure of PCBM-only films. Thus, we hypothesize that the changes in our scattered intensity are due to changes in the aggregation characteristics of PCBM in the active layers. 2. Grazing-incidence X-ray diffraction (GIXD) images of PCBM films Figure S1 shows 2D GIXD images of PCBM films (15 mg/ml, chlorobenzene) deposited via spin coating on ITO (Figures S1a and S1c) and on PEDOT:PSS-coated ITO substrates (Figures S1b and S1d). Data taken on unannealed films are shown in Figures S1a and S1b, while data on films annealed for 30 min at 165 o C are shown in Figures S1c and S1d. We were unable 4

8 Figure S1. Two-dimensional grazing-incidence X-ray diffraction images of PCBM films (a), (b) not thermally annealed and (c), (d) annealed for o C. Images were obtained from films cast on ITO (Figures S1a and S1c) and PEDOT:PSS (Figures S1b and S1d). Note the highly crystalline structure evident from the large number of spots in Figure S1d. to spin coat PCBM on HSAM- or FSAM-modified ITO substrates, most likely due to the low surface energy of these substrates. The GIXD images in Figures S1a through S1c exhibit a reflection that is characteristics of PCBM at 1.4 1/Å; the intensity of this reflection appears to be uniform azimuthally. The GIXD image in Figure S1d, on the other hand, appears to be different from the other GIXD images. Specifically, the GIXD image in Figure S1d shows a large number of spots, showing that annealing PCBM on PEDOT:PSS leads to crystallization of the PCBM with preferentially oriented crystals. Although we have no concrete explanation for this phenomenon at this time, we hypothesize that a combination of surface energetics and surface roughness can play a role in promoting the crystallization of PCBM. 3. Deconvolution of PCBM and P3HT reflections In order to deconvolute the constituent P3HT and PCBM peaks in P3HT/PCBM GIXD data, we first analyze data from PCBM- and P3HT-only films. Figure S2a shows the azimuthally-averaged intensity as function of q for a well-ordered PCBM film (Figure S1d). The data can be deconvoluted into four Lorentzians: three individual peaks corresponding to 5

9 crystalline PCBM reflections at /Å, /Å and /Å, and a fourth broad peak at /Å, most likely corresponding to an amorphous phase. Assignment of the crystalline PCBM peaks is beyond the scope of this paper. A similar deconvolution procedure can be performed for P3HT-only films, as shown in Figure S2b. Using the peak positions and the peak width of the amorphous peaks obtained in Figure S2 for both P3HT and PCBM, we can now deconvolute the individual contributions to the scattered intensity for data generated from P3HT/PCBM films. Figure S3 shows azimuthallyaveraged GIXD data for P3HT/PCBM films and the resulting constituent peaks. The PCBM scattered intensity is computed as the sum of the area of all the constituent PCBM peaks. The aggregation of PCBM in P3HT/PCBM mixtures prior to PCBM crystallization contributes to an increase of the PCBM scattered intensity through an increase in the average number of adjacent PCBM molecules or increase in the extent of local organization. Thus, we take the PCBM scattered intensity as a measure of the extent of aggregation and order of PCBM, even when distinct crystalline reflections are not present. I (a.u.) Amorphous Peak 1 Peak 2 Peak 3 Total GIXD data (a) I (a.u.) Amorphous (300) (010) Total GIXD data (b) q (1/Å) q (1/Å) Figure S2. (a) GIXD intensity (pink), constituent peaks and sum of peaks (light blue, Total) for a PCBM-only film. (b) GIXD intensity (light blue), constituent peaks and sum of peaks (blue, Total) for a P3HT-only film. 6

10 I (a.u.) 10 P3HT Amorphous P3HT (300) P3HT (010) PCBM Amorphous PCBM Peak 1 PCBM Peak 2 PCBM Peak 3 Total GIXD data (a) I (a.u.) 10 5 P3HT Amorphous P3HT (300) P3HT (010) PCBM Amorphous PCBM Peak 1 PCBM Peak 2 PCBM Peak 3 Total GIXD data (b) q (1/A) q (1/Å) Figure S3. GIXD intensity (dark blue), constituent peaks and sum of peaks (gold, total) for P3HT/PCBM films (a) annealed for 240 min at 130 o C and (b) not annealed. Alternatively, we can compute the PCBM scattered intensity by integrating between /Å and applying equation 1 of the main text with a linear background subtraction between 1.21 and /Å. As shown in Figure S4, the results are similar to extraction of PCBM scattered intensities through deconvolution (compare with Figure 3 of the main text). In both methodologies for quantifying the PCBM scattered intensity, our experimental geometry does not include contributions from crystals oriented within ~ 6 o of the surface normal. Given that PCBM is always isotropic in our P3HT/PCBM samples, we do not expect this experimental limitation to affect our results. 4. Extent of out-of-plane π-stacking from rocking curve intensities near ω = 90 o Figure S5 shows X-ray rocking curves for P3HT/PCBM films as a function of polar angle ω. As shown in Figure 4 of the main text, we estimate the extent of out-of-plane P3HT π- stacking in P3HT/PCBM films by integrating the rocking curve intensity between ω = 75 o - 80 o. We chose to use 75 o - 80 o to avoid the intensity enhancement near ω = 90 o due to the scattering geometry in Figure S5. As shown in Figure S6, the rocking curve intensity integrated between ω 7

11 J SC (ma/cm 2 ) Normalized PCBM scattered intensity unannealed (PEDOT) unannealed (ITO) unannealed (HSAM) unannealed (FSAM) 165 o C (PEDOT) 165 o C (ITO) 165 o C (HSAM) 165 o C (FSAM) 130 o C (PEDOT) 50 o C (PEDOT) Figure S4. J SC vs normalized PCBM scattered intensity where the scattered intensity is computed using a linear background subtraction and eqn 1 of the main text. The PCBM scattered intensity is normalized to the largest PCBM scattered intensity. = 80 o - 90 o is quantitatively the same as the intensity integrated between 75 o - 80 o, confirming that our results are the same with either choice of integration range. In estimating the extent of out-of-plane P3HT π-stacking from rocking curves we ignore contributions to the rocking curve intensity near ω = 90 o from P3HT crystals oriented with both the (100) and (010) planes parallel to the substrate normal. We expect the population of these o C Unannealed I/I ω ( o ) Figure S5. Rocking curves of the P3HT/PCBM films as a function of the polar angle, ω. A linear background was calculated using scattered intensities away from the Bragg reflections (q ~ 0.3 and /Å) and subtracted from the data. For the data shown in Figure 4 of the main text, we integrate the rocking curve intensity between ω = 75 o and 80 o. 8

12 Rocking curve intensity integrated from ω = 80 o to 90 o Rocking curve intensity integrated from ω = 75 o to 80 o Figure S6. Rocking curve intensity integrated from ω = 80 o to 90 o as a function of the rocking curve intensity integrated from ω = 75 o to 80 o. Both intensities are normalized to the largest intensity found in our samples. crystal orientations to be negligibly small because P3HT generally does not orient with its chain axis perpendicular to the substrate Alternative methodology for quantifying the extent of out-of-plane π-stacking As an alternative to quantifying the extent of out-of-plane π-stacking from rocking curves at ω ~ 90 o, we can use the P3HT (010) reflection in the q z direction from our GIXD images. We again turn to equation 1 to estimate the extent of out-of-plane P3HT π-stacking, I P3HT(010),corrected, by integrating the (010) intensity ( /Å) in the z-direction over 0.051/Å (q xy ) after subtracting a linear contribution we attribute to an isotropic background. Note that I PCBM,corrected is replaced by I P3HT(010),corrected and I PCBM,GIXD is replaced by I P3HT(010),GIXD in equation 1. Due to the grazing-incidence geometry, however, the scattering vector is tilted by ~ 7 o with respect to the substrate normal. Crystals with their (010) planes tilted less than 7 o thus do not contribute to the GIXD out-of-plane (010) intensity. Nevertheless, our results, shown in Figure S7, indicate a linear correlation between the device J SC and the out-of-plane P3HT (010) intensity. The intensity is normalized to the largest value found in our samples. This is consistent with the observed trend of Figure 4 of the main text, confirming that the structure- 9

13 function correlations reported herein are real, and not artifacts of data processing. Despite the large error bars, both Figure S7 and Figure 4 implicate a non-negligible correlation between the device J SC and the out-of-plane P3HT π-stacking. J SC (ma/cm 2 ) 10 unannealed 165 o C 130 o C o C Normalized extent of out-of-plane P3HT π-stacking Figure S7. Device J SC vs normalized out-of-plane P3HT (010) intensity obtained from GIXD images and equation 1 of the main text of P3HT/PCBM films annealed at various temperatures. The out-of-plane intensity is normalized by the largest intensity in our samples. The samples included in this figure have PCBM scattered intensity > P3HT and PCBM scattered intensities obtained from X-ray scattering experiments In this study, we varied the annealing time and temperature of the active layer of P3HT/PCBM solar cells. Using rocking scans, GIXD data and eqn S1 we determined the P3HT crystallinity and PCBM scattered intensity. A summary of our results can be found in Table S1. Note that all values are normalized to the highest value measured in our samples. 10

14 Table S1. Summary of the normalized* P3HT crystallinity and PCBM scattered intensity obtained for all of the samples in this study. Anode Annealing temperature ( o C) Annealing time (min) P3HT crystallinity PCBM intensity (100) P3HT intensity (ω = 90 o ) PEDOT:PSS ITO FSAM/ITO HSAM/ITO *All samples are normalized to the largest value obtained in our study These samples show clear evidence of PCBM crystallization 7. References 1. J. L. Baker, L. H. Jimison, S. Mannsfeld, S. Volkman, S. Yin, V. Subramanian, A. Salleo, A. P. Alivisatos and M. F. Toney, Langmuir, 2010, 26, C. H. Woo, B. C. Thompson, B. J. Kim, M. F. Toney and M. J. Frechet, Journal of the American Chemical Society, 2008, 130, C. W. Chu, H. C. Yang, W. J. Hou, J. S. Huang, G. Li and Y. Yang, Applied Physics Letters, 2008, 92, H. C. Yang, T. J. Shin, L. Yang, K. Cho, C. Y. Ryu and Z. N. Bao, Advanced Functional Materials, 2005, 15,