The origin of high PCE in PTB7 based photovoltaics: proper charge neutrality level and free energy of charge separation at PTB7/PC71BM interface

The energy level alignments at donor/acceptor interfaces in organic photovoltaics (OPVs) play a decisive role in device performance. However, little is known about the interfacial energetics in polymer OPVs due to technical issues of the solution process. Here, the frontier ortbial line-ups at the donor/acceptor interface in high performance polymer OPVs, PTB7/PC71BM, were investigated using in situ UPS, XPS and IPES. The evolution of energy levels during PTB7/PC71BM interface formation was investigated using vacuum electrospray deposition, and was compared with that of P3HT/PC61BM. At the PTB7/PC71BM interface, the interface dipole and the band bending were absent due to their identical charge neutrality levels. In contrast, a large interfacial dipole was observed at the P3HT/PC61BM interface. The measured photovoltaic energy gap (EPVG) was 1.10 eV for PTB7/PC71BM and 0.90 eV for P3HT/PC61BM. This difference in the EPVG leads to a larger open-circuit voltage of PTB7/PC71BM than that of P3HT/PC61BM.

However, only inaccurate values of E exc have been reported without comparison between the optical gap and transport gap. We will resolve these issues in the result section below.
In addition, the free energy of charge separation (G CS ) is a very recent issue related to the exciton dissociation at the interfaces. Jonathan et al. suggested that the optimum G CS value exists to maximize the J SC 4,16,17 . However, it still remains as an open problem because no experimental evidence for G CS has been reported at the polymeric junction due to the technical issue discussed below. Collectively, it is highly necessary to measure the energy levels accurately at the polymeric junction to give deep insight into the core OPV-parameters such as E PVG , E exc , G CS and so on.
To determine the true energy level alignments in OPVs, there are two key requirements. (1) Measurements of charge transport level in a solid state: The energy level measurements in a liquid state, e.g. cyclic voltammetry, cannot provide true charge transport levels associated with the solid state device due to ignorance of the screening effect by surrounding molecules/polymers in a solid state. (2) Consideration of an interface dipole and energy level relaxation across the interface: It is well known that the Schottky-Mott limit is not always valid at organic interfaces, and thus the charge injection/extraction barriers in OPVs cannot be estimated accurately from individual measurements of each layer. Therefore, in situ photoelectron spectroscopy (PES) with stepwise deposition-measurement procedures are indispensable 18,19 .
For the interfaces in small molecule OPVs, stepwise measurements are feasible since thermal evaporation enables incremental deposition of organic molecules 5,20 . However, it is challenging to study the energy level alignments at polymeric interfaces prepared by solution process due to technical limitations. It is often impossible to trace the evolution of energy levels from the stepwise formation of polymeric interfaces. This is because the polymer solution used to form an upper layer could dissolve the preformed underlying-layer. Therefore, spin-coating is not an adequate method for film preparation to investigate the interfacial energy level alignments. Due to such limitations, the true electronic structure and its evolution at the polymer donor/acceptor "interface" have not been shown. To overcome these limitations of spin-coating methods, we adopted a vacuum electrospray deposition (VESD) method to form the polymeric thin film in a stepwise manner (see the experimental section and previous report) [21][22][23] . The VESD OPVs have shown almost identical device performance with the spin-coated OPVs, ensuring high film quality with the VESD method for device fabrication 21,24 . By VESD, the HOMO and LUMO energy levels, the interface dipole (eD), and the band bending (V B ) of polymers in bulk heterojunction (BHJ) devices can be studied.
In BHJ OPVs, poly (4,8- [3,4-b]thiophene-4,6-diyl) (PTB7) is a promising donor material recording the remarkable PCE because of its superior optoelectronic properties 25,26 . The low-band gap polymer PTB7 is well-suited to [6,6]-phenyl-C 71 -butyric acid methyl ester (PC 71 BM) which absorbs a broad range of light. However, the true energy level alignments at the PTB7:PC 71 BM interface have not been measured previously due to the inherent problems in spin-coating deposition as aforementioned.
In this study, we determined the complete energy level alignments at the PTB7/PC 71 BM interface and related interfaces. In situ ultraviolet and X-ray photoelectron spectroscopy (UPS and XPS) measurements were performed to determine the valence density of states of PTB7 and PC 71 BM. Inverse photoelectron spectroscopy (IPES) was used to measure their transport band gap (E t ), and thus the true electron transport level (LUMO). Theoretical calculations using density functional theory (DFT) were performed and the calculated results were compared with the measured UPS and IPES spectra. Finally, the PTB7/PC 71 BM interface was compared with a poly(3-hexylthiophene-2,5-diyl) (P3HT)/ [6,6]-phenyl-C 61 -butyric acid methyl ester (PC 61 BM) interface, which is a former popular donor/acceptor junction. From the measured energy level alignments, critical parameters of OPVs such as E PVG , E exc and G cs were accurately evaluated to understand the OPV operation in depth. Finally brief strategy to improve OPV performance will be suggested.

Results and Discussion
The donor/acceptor junction was prepared on PEDOT:PSS since the most widely used OPV structure is ITO/ PEDOT:PSS/Active layer/Cathode [27][28][29] . Yang et al. reported that BHJ films show phase separation and form microscopic junctions of donor and acceptor [30][31][32] . This junction was simulated with the bilayer structure of PTB7/ PC 71 BM to determine the energy level alignments across the junction.
Step-by-step deposition and in situ photoelectron spectroscopy (PES) measurements for PTB7 and PC 71 BM were performed to study the interfacial energy level alignments. Prior to the in situ PES measurements, we compared the morphology and electronic structure of the spin-coated and VESD films to ensure consistent film quality. For that, atomic force microscope (AFM) and PES measurements on each bulk film were conducted (Supplementary Figs S1 and S2). Both films showed quite similar morphology and the same electronic structure, and from this we conclude that interfaces prepared by VESD are identical with the spin-coated films.
To separate the contributions of V B and eD, the thickness dependent evolution of the WF and PTB7 HOMO are depicted in Fig. 2. The V B and eD of PTB7 were evaluated to be 0.35 eV and 0.20 eV, respectively. The high work function of PEDOT:PSS induces the V B and electron transfer from PTB7 to PEDOT:PSS, and thus increases electron carrier in PEDOT:PSS and hole carrier density in PTB7 5,34 . Increased hole carrier density induces the V B , and this bending direction assists the hole extraction from PTB7 to PEDOT:PSS.
XPS core level spectra of polymers were cross-checked with the energy level shifts measured from UPS spectra. To evaluate the peak shift, we deconvoluted each spectrum while keeping system constraints. We first analysed each bulk (pristine) spectrum of S 2p of PTB7 (topmost) and PEDOT:PSS (bottommost). The pristine PTB7 S 2p core level was observed at 163.3 eV and 165.1 eV with a spin-orbit splitting of 1.8 eV 35 . In the bottom of the Fig. 3a, we also deconvoluted the S 2p spectrum with three components (PEDOT, PSSH and PSS − -Na + ) to evaluate the peak shift accurately. We used fitting constraints of PEDOT:PSS from previous reports such as the intensity ratio and energetic distance between each peak (PEDOT, PSSH and PSS − -Na + ) during the fitting process 36,37 . As a result, we observed the 0.3 eV peak shift of PTB7 and PEDOT:PSS in the S 2p orbital, which coincides well with the HOMO level shifts measured from UPS spectra within the margin of error. However, the F 1s peak maintained the same position during PTB7 deposition.  To understand the different core level shifts in S 2p and F 1s, the electronic structures of PTB7 were calculated using DFT. To simulate long polymer chains, we modeled 8-mer of PTB7 unit structure ( Supplementary  Fig. S4). Figure 4 shows the calculated (a) electrostatic potential map, (b) HOMO and (c) LUMO wave function of PTB7 8-mer. The electrostatic potential map shows the charge distribution of PTB7. In Fig. 4a, a low (red) and high (blue) electrostatic potential indicates the electron rich and deficient regions, respectively. The F atom (red-dashed circle) withdraws electrons in the vicinity due to its high electronegativity, which is seen in the electrostatic potential map (a). However, F hardly contributes to the HOMO (b) and LUMO (c) because of the very strong C-F bonding due to the high electronegativity of F. This strong bonding induces localized electronic states which are energetically far below (above) the delocalized HOMO (LUMO) state. Thus, the contribution of F to the HOMO and LUMO is negligible. Since the electron transfer occurs predominantly from the HOMO level of PTB7 to the Fermi level of PEDOT:PSS, the intact F 1s position can be understood with the localized C-F bonding state during the interface charge transfer process.
We measured the interface between PEDOT:PSS and P3HT as a non-push-pull donor counterpart.   0.3 nm-deposition step, the HOMO level of P3HT is observed at 0.35 eV below the Fermi level. As the film thickens, the HOMO of P3HT gradually shifts toward higher binding energies by 0.57 eV. The SEC of PEDOT:PSS is observed at 4.75 eV in kinetic energy and shifts to lower kinetic energy by 1.10 eV, indicating WF reduction. In the same manner, we evaluated V B,P3HT and eD by 0.57 eV and 0.53 eV, respectively. Detailed energy level alignments of each interface will be discussed below.
PTB7/PC 71 BM and P3HT/PC 61 BM interfaces. Another important interface in BHJ is the donor/acceptor interface, PTB7/PC 71 BM and P3HT/PC 61 BM, and we continued PC 71 BM and PC 61 BM deposition on the final PTB7 and P3HT layers explained in section A, respectively. Figure 6 shows the UPS spectra of (a) the normalized SEC region, (b) the background-removed HOMO region and (c) magnified HOMO region of the PTB7 (7 nm)/ PC 71 BM (0.4, 0.6, 0.9, 1.3, 2.5, 5.0 nm) sample. There is no SEC shift during the PC 71 BM deposition, which implies that the WF is intact as 4.25 eV. The bulk HOMO level of PTB7 and PC 71 BM appears at 0.80 and 1.60 eV, respectively. To determine the HOMO onset of PTB7 and PC 71 BM in the middle of PC 71 BM deposition, the HOMO spectra were deconvoluted with two pristine bulk PTB7 (bottommost spectrum) and PC 71 BM spectra (topmost spectrum). The intensity of deconvoluted PTB7 and PC 71 BM spectra was estimated using the electron effective attenuation length (EAL) of PC 71 BM, which was obtained from the attenuation of S 2p of PTB7 (detailed evaluation is in supplementary information). The summation of two bulk spectra reproduced the measured spectra quite well without any additional components. This indicates the absence of molecular orbital hybridization and chemical reactions, which is consistent with the XPS spectra ( Supplementary Fig. S3). Furthermore, both HOMO and SEC positions were maintained intact during the PC 71 BM deposition, meaning the absence of both charge transfer and eD. It concludes that PTB7 and PC 71 BM have the same charge neutrality level 5,38-40 .
The interfacial electronic structure of PTB7/PC 71 BM is compared with P3HT/PC 61 BM as a non-push-pull counterpart. Figure 7 shows the UPS spectra of (a) the SEC region, (b) HOMO region and (c) magnified HOMO region of P3HT (7.0 nm)/PC 61 BM (0.2, 0.5, 1.1 and 2.2 nm). As shown in Fig. 7a, a significant SEC shift is observed from 3.65 eV to 4.10 eV. From the SEC shift and V B , a large eD of 0.35 eV (= 0.45-0.10 eV) exists, which is caused by a large charge-neutrality level difference between P3HT and PC 61 BM. In Fig. 7c, the bulk HOMO level of P3HT and PC 61 BM appears at 0.92 eV and 1.92 eV. We also deconvoluted the HOMO spectra in the same manner. As shown in Fig. 7c, the HOMO of P3HT shifts toward lower binding energies by 0.10 eV, indicating the V B . For the origin of this V B , two scenarios are plausible: (1) electron attraction of PC 61 BM from P3HT and subsequent induced polarization of charge densities or (2) local changes in the inter-ring torsional angle of P3HT at the interface due to the interaction with PC 61 BM 15 . As a result, the P3HT/PC 61 BM interface forms a more complex junction than the PTB7/PC 71 BM interface.
Transport band gap determination. Figure 8 shows the combined UPS and IPES spectra of bulk (a) PTB7 and (c) PC 71 BM and their comparison with calculated spectrum (b) PTB7 and (d) PC 71 BM. The calculated spectrum was rigidly shifted to match the measured HOMO and LUMO position [41][42][43] . In both HOMO and LUMO regions, the DOS of measured and calculated spectra match well, which ensures reliable determination where E t is the transport band gap and E opt is the optical band gap. E opt of each material was determined from previous reports [46][47][48][49] . Ram et al. reported a relatively smaller E exc of PTB7 and thus suggested easy exciton dissociation in PTB7 15 . However, current results show the same E exc of P3HT and PTB7. Therefore, the interfacial electronic structure is more important than the E exc of the bulk donor for efficient exciton dissociation. The measured and evaluated IE, EA, E t and E exc are listed in Table 1.   50 . As shown in Fig. 9a,b, the hole extraction barriers from donor to PEDOT:PSS are 0.45 eV for PTB7 and 0.35 eV for P3HT. PEDOT:PSS strongly pulls electrons from both polymers due to its high WF, which induces band bending of V B,PTB7 = 0.35 eV and V B,P3HT = 0.57 eV. The eD at both interfaces of PEDOT:PSS/PTB7 and PEDOT:PSS/P3HT indicates that the high WF of PEDOT:PSS induces the energy level pinning at both interfaces. The pinning threshold of P3HT (E p+ = 3.9 eV) matches well with previous reports 6,51 . The pinning threshold of PTB7 can be estimated to be less than 4.80 eV based on the eD at the PEDOT:PSS/PTB7 interface. In Fig. 9a, the PTB7/PC 71 BM interface shows a flat band alignment and no interfacial potential change. However, the P3HT/PC 61 BM interface shows a large eD due to Fermi level pinning of PCBM (E p− = 4.3 eV) 6,52 . In the P3HT:PC 61 BM BHJ film, the eD's are randomly distributed in 3-dimensional space, which gives random local electric fields, and thus it does not affect the built-in potential. However, it does gives local singularities in potential distribution and may assist or hinder exciton dissociation and charge transport according to its direction. Therefore it exerts a strong influence on the E PVG at D/A interfaces.
The E PVG of PTB7/PC 71 BM and P3HT/PC 61 BM is evaluated as energetic difference between HOMO donor and LUMO acceptor thus 1.10 eV (0.80 + 0.30 eV) and 0.90 eV (0.82 + 0.08 eV) for P3HT/PC 61 BM, respectively. This accords well with the fact that the PTB7/PC 71 BM OPVs have higher V OC than the P3HT/PC 61 BM OPVs by about 0.1~0.2 eV 53-55 . These findings not only show the direct evidence of correlation between V OC and E PVG but also explain the role of eD in E PVG, i.e. the eD can change the E PVG significantly. Therefore, the eD at the interface should be considered rigorously to determine the E PVG accurately.
Another interesting issue in BHJ OPVs is the Δ G CS during the charge separation process. It is reported that Δ G CS acts as the driving force to dissociate the exciton at the donor/acceptor interfaces 16 . Thus, a high Δ G CS leads to more efficient exciton dissociation and enhances the PCE. However, too large a Δ G CS creates parasitic energy losses and limits the PCE reversely. Generally, the Δ G CS is calculated by the abbreviated Rehm-Weller equation 17 . where the E s is the singlet excited state energy, which is derived from the E opt [46][47][48][49] . Until now, Δ G CS was estimated from the IE Donor and EA acceptor measured from individual film without any consideration of electronic polarization (solid state effect) at the donor-acceptor interface. Therefore, it would under/over-estimate the Δ G CS inherently. In this study, Δ G CS is evaluated as 0.53 eV for PTB7/PC 71 BM and 1.00 eV for P3HT/PC 61 BM including all local interfacial potentials such as V B and eD. Too large a Δ G CS of P3HT/PC 61 BM induces large parasitic energy losses and induces negative effects on exciton dissociation, limiting the PCE significantly. Jonathan et al. reported that an ideal value of Δ G CS and the optical band gap of the donor is about 0.4 eV and 1.6 eV, respectively 16 . Current measurements show that the Δ G CS and optical band gap of PTB7/PC 71 BM system is quite close to the optimum values. Absence of eD, which hinders charge separation in P3HT/PC 61 BM, and proper Δ G CS at the PTB7/ PC 71 BM interface are the origin of high PCE of PTB7 in conjunction with its excellent light absorption. Too large a Δ G CS and singular potential due to eD at the P3HT interfaces have a negative effect on device performance. All these factors are determined from the interfacial electronic structure of donor and acceptor materials. These results demonstrate the role of eD in determination of E PVG and the correlation between V OC and E PVG .
Therefore, we suggest the strategy for the high performance OPVs: (1) proper combination of donor and acceptor materials should be chosen based on their charge neutrality levels. Improper level alignment may cause the potential discontinuity due to the eD, which impedes the charge separation and increases the interface recombination. (2) Δ G CS should be optimized to dissociate exciton efficiently without parasitic energy losses.

Conclusion
In this study, in situ UPS, IPES and XPS measurements were performed to figure out the pristine valence density of states of PEDOT:PSS, PTB7, PC 71 BM and their interfacial energy level alignments. As a result, the E t (E t, PTB7 = 2.25 eV, E t, PC71BM = 1.90 eV), ionization energy (IE PTB7 = 5.05 eV, IE PC71BM = 5.85 eV) and electron affinity (EA PTB7 = 2.80 eV, EA PC71BM = 3.95 eV) were obtained. Large band bending is observed in PTB7 on the PEDOT:PSS layer, which reduces the hole extraction barrier at the PEDOT:PSS/PTB7 interface, and the same is true for PEDOT:PSS/P3HT. Interfacial dipoles and strong band bending are not observed at the PTB7/PC 71 BM interface, meaning negligible charge exchanges between PTB7 and PC 71 BM. In contrast, a large interface dipole is observed between P3HT and PC 61 BM. In addition, the E PVG between a donor and an acceptor (1.10 eV for PTB7/PC 71 BM and 0.90 eV for P3HT/PC 61 BM) were also measured. The Δ G CS at the interface of PTB7/PC 61 BM and P3HT/PC 71 BM were determined as 0.53 eV and 1.00 eV, respectively. A not too large Δ G CS reduces parasitic energy losses and thus enhances the exciton dissociation, which is the reason why the PTB7/PC 71 BM OPV shows high PCE. Current analysis on the energy level alignment with the emphasis on detailed interface dipole and band bending could suggest a proper material combination for high performance BHJ OPVs. and PC 71 BM (Luminescence Technology Corp.) were used as received. PTB7 and PC 71 BM were dissolved in chlorobenzene and each solution was stirred at room temperature for 24 hours before use. The concentration of both solutions for the VESD was 1 mg ml −1 . ITO substrates were sequentially cleaned by ultrasonication with deionized water, detergent, acetone, and methanol and dried with N 2 gas.

Methods
Vacuum electrospray deposition. PTB7, PC 71 BM, P3HT and PC 61 BM were deposited with a bias of 1.5 kV between the capillary out of the syringe and the inlet of the vacuum system. The injection speed of solution was uniformly controlled by a syringe pump (1 μ l s −1 ). The deposition rate was monitored by measuring the ionic current of the injected solute into the vacuum system. All parameters were optimized to obtain a uniform thin film with fine thickness control down to sub-nm scales.
Photoelectron spectroscopy measurement. The electronic structure of ITO/PEDOT:PSS/PTB7/ PC 71 BM was investigated with in situ UPS and XPS. PC 71 BM, PTB7 and PEDOT:PSS were deposited on an ITO substrate in a stepwise manner. The sample was transferred to the analysis chamber at each deposition step under UHV condition below 10 −8 Torr to obtain UPS and XPS spectra. The base pressure of the analysis and preparation chamber was 5.0 × 10 −9 and 2.0 × 10 −8 Torr, respectively. A PHI-5700 spectrometer was used to analyse the kinetic energy of photoelectrons with an Al Kα X-ray (1486.6 eV) and a He І (21.22 eV) ultraviolet source. The spectrometer was calibrated with respect to the binding energy of Au 4f (84.0 eV), using its Fermi step as a reference. The uncertainty of our UPS measurements estimated from the broadening of the Au Fermi step was 0.09 eV. A sample bias of − 10 V was applied during UPS measurements to obtain the true SEC. To determine the HOMO onset accurately, Shirley-type background was removed from the measured HOMO region spectrum. The film thicknesses and deposition rates were monitored by the attenuation ratio of In 3d intensity of the ITO substrate as well as the ionic current measurements during the VESD deposition process.

DFT calculations.
A Becke-style three-parameter exchange and Lee-Yang-Parr correlation hybrid functional (B3LYP) and a split basis set of 6-31G(d) were used in the GAUSSIAN 09 software package. We increased the number of PTB7 monomers to simulate the PTB7 polymer. The effect of the number of a monomer on the frontier orbitals was tested from a monomer to 8-mer unit. The side chains were truncated to a methyl group for simplicity. As shown in supplementary Fig. S4, 8-mer is a reasonable choice and the generated DOS well-matched with the experimental spectrum (Fig. 8b). We also tested the side chain effect by comparing the complete side chain with a truncated side chain, which indicates that this truncation does not affect the calculated DOS near frontier orbitals. This coincides well with previous reports 56-58 . Morphology. Morphological images were obtained by non-contact mode atomic force microscopy measurements (Park system).