High performance polymer solar cells with as-prepared zirconium acetylacetonate film as cathode buffer layer

Low-work-function active metals are commonly used as cathode in polymer solar cells (PSCs), but sensitivity of the active metals towards moisture and oxygen results in poor stability of the devices. Therefore, solution-proceessable and stable cathode buffer layer is of great importance for the application of PSCs. Here we demonstrate high performance PSCs by employing as-prepared zirconium acetylacetonate (a-ZrAcac) film spin-cast from its ethanol solution as cathode buffer layer. The PSCs based on a low bandgap polymer PBDTBDD as donor and PC60BM as acceptor with a-ZrAcac/Al cathode demonstrated an average power conversion efficiency (PCE) of 8.75% which is significantly improved than that of the devices with traditional Ca/Al cathode. The improved photovoltaic performance is benefitted from the decreased series resistance and enhanced light harvest of the PSCs with the a-ZrAcac/Al cathode. The results indicate that a-ZrAcac is a promising high performance cathode buffer layer for fabricating large area flexible PSCs.

S ince the first reports of photo-induced electron transfer from a conjugated polymer to fullerene 1 , polymer solar cells (PSCs) have attracted exclusive attention due to low-cost, light-weight, and mechanical flexibility with compatibility to future large-area roll-to-roll printing production. The performance of PSCs depends not only on the electronic energy levels, absorption and carrier mobility of the conjugated polymer donor and fullerene derivative acceptor photovoltaic materials [2][3][4][5] , but also on the effective charge extraction of both holes and electrons from the photoactive layer and then collection to the respective anode and cathode [6][7][8][9] . Therefore, the electrode materials or the electrode buffer layers play a key role in improving photovoltaic performance of the PSCs 7,8 . High workfunction anode buffer layers and low workfunction cathode buffer layers are pursued in selecting the electrode buffer layers.

Results
The a-ZrAcac layer is highly transparent in the visible wavelength range as shown in Figure S1 in Supporting Information (SI) and bearing an amorphous structure confirmed by XRD ( Figure S2 in SI). The characteristic absorption peak of acetylacetonate located at 300 nm attributes to the n-p* and p-p* intra-ligand electronic transitions 37,38 . The transparence of the a-ZrAcac layer will benefit the transmission and reflection on the back Al electrode for the transmitted light through the active layer, which will potentially increase the light harvest of the photoactive layer and thus enhance the photocurrent of the PSCs. a-ZrAcac possesses a high stability constant 39 and is thermally stable at lower than 150uC as shown in the thermogravimetric analysis (TGA) and differential thermal analysis (DTA) plots in Figure S3 in SI. Our result is in good agreement with previous study 40 .
The effects of a-ZrAcac CBL on the photovoltaic performance of PSCs were examined by constructing the traditional structured devices of ITO/PEDOT5PSS/Active layer/a-ZrAcac/Al, as shown in Figure 1(b). In the PSCs, the active layer is P3HT5PC 60 BM, P3HT5PC 70 BM, PBT15PC 70 BM, PBDTTT-C-T5PC 70 BM or PBDTBDD5PC 60 BM blend film, PEDOT5PSS is poly(3,4-ethylenedioxythiophene)5poly(styrenesulfonate). Figure 1(c) shows the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO) energy levels of P3HT 41 , PBDTBDD 42 , PBT1 43 , PBDTTT-C-T 44 , PC 60 BM 45 , PC 70 BM 46 and a-ZrAcac, where the energy levels of ZrAcac were measured by ultraviolet photoelectron spectroscopy (UPS). Figure 1(d) illustrates the UPS spectra taken from the ZrAcac coated on ITO substrate. The work function can be deduced from the energy of the secondary cutoff, which is 2.37 eV. The HOMO level can be calculated as 4.32 eV from the difference between the incident light energy (He I, 21.22 eV) and the energy of the onset (19.27 eV) as well as the work function. The LUMO level of 1.22 eV was obtained from the band gap of 3.1 eV (determined from onset of the absorption (400 nm) (see Figure S1 in SI)) and the HOMO level. The energy level alignment at the photoactive blend layer/a-ZrAcac interface is very important for high performance PSCs. The extremely low work function of ZrAcac (2.37 eV) lies above the LUMO energy levels of PC 60 BM (3.90 eV) and PC 70 BM (3.91 eV), inducing the interfacial dipole formation across this interface, which will facilitate ''barrierfree'' electron extraction from the LUMOs of PC 60 BM and   47 . The good energy level matching of ZrAcac can expect excellent electron extraction from photoactive layer and enhanced performance of the PSCs.
Surface morphology of a-ZrAcac layer spin-coated on clean ITO glass surface was investigated by tapping-mode atomic force microscopy (AFM), as shown in Figure S4 in SI. The AFM image shows a root-mean-square (rms) roughness of 3.4 nm, which is a little lower than that (4.6 nm) of bare ITO surface as shown in Figure S5. The surface morphologies of the photoactive layer with and without a-ZrAcac CBL were also investigated by AFM, as illustrated in Figure 2. The rms roughness of the blend layer of P3HT5PC 60 BM is 16.9 nm (Figure 2(a)), while the surface covered by the a-ZrAcac layer becomes more rough with a rms of 19.6 nm (Figure 2(b)). The same roughness changing trend but smaller roughness was observed for the blend layers of PBDTBDD5PC 60 BM. The rms roughness of the blend layers of PBDTBDD5PC 60 BM with and without a-ZrAcac CBL is 5.5 nm (Figure 2(d)) and 2.7 nm (Figure 2(c)), respectively. The increased roughness for the active layers with a-ZrAcac CBL could increase the contact area with the Al electrode deposited on it 48 .
The chemical component of ZrAcac powder and as prepared Zracac layer (a-ZrAcac, 40 nm) on glass slide was analyzed by Xray photoelectron spectroscopy (XPS) as shown in Figure 3. The binding energies (BE) obtained in the XPS analysis are corrected for specimen charge by referencing the C 1 s peak to 284.8 eV. As shown in Figure 3(a), the survey scan for both powder and thin film samples performed in the 0-1000 eV BE range show characteristic peaks of the elements Zr, O, and C. As shown in Figure 3(b), the XPS spectrum of the core level of Zr 3d for the powder sample has a strong spin-orbit doublet due to Zr 3d 5/2 at 182.3 eV and Zr 3d 3/2 at 184.7 eV 49,50 . The XPS spectrum is constrained by the Zr 3d 5/2 -Zr 3d 3/2 spin-orbit separation being 2.4 eV, and the area ratio of the two peaks of each doublet being 352. These features are characteristic of Zr 41 ions in full oxidation states 49,50 . In the XPS spectrum of the a-ZrAcac thin film sample, the peaks corresponding to Zr 3d5/2 and Zr 3d3/2 are unchanged at 182.3 eV and 184.7 eV, respectively, with an Zr 3d5/2-Zr 3d3/2 spin-orbit separation of 2.4 eV and an intensity ratio of 352. The results indicate that the zirconium in the thin film sample is also in the Zr 41 oxidation state. The O 1 s XPS spectra for the powder and thin film samples are shown in Figure 3(c). It can be observed that the original powder sample presents nonsymmetric O 1 s peak centered at 531.3 eV with a shoulder around 530 eV. The peak at 531.3 eV corresponds to the oxygen in the acetylacetonic ligand cycle 30 , while the shoulder at 530 eV should attribute to the oxygen bonded with zirconium 49,50 , which means there is a small portion (17% from the peak fitting) of Zr 41 ions coordinate less than four acetylacetonic ligands in the original power sample and this is a common case in coordination complexes 51 . The thin film sample shows similar O 1 s peak, and the portion of non-four-coordination is 15%, which means there is no decomposition occurred. As shown in Figure 3(d), the C 1 s peaks of ZrAcac powder are located at 284.7 eV, which is assigned to the carbon in C-H/C-C group 52 , the peak at 286.7 eV corresponds to the carbon in delocalized C5O of ligand cycle 52 , and the weak shoulder at 288.3 eV attributes to the absorbed CO 2 from ambient atmosphere 52 . The atomic ratio of C 1 s (286.7 eV) and C 1 s (284.7 eV) peaks is 153.50, which is very close to the ratio of 153.54 for thin film sample. The XPS results confirm that there is no structure change between the powder and the obtained thin film.  The device performance parameters are given in Table 1, which are average values of 20 individual devices. For example, the V oc , J sc , FF and PCE for PBDTBDD5PC 60 BM based on 20 devices are listed in Table S1 in SI. As shown in the inset of Figure 4(a), the J-V curves measured in the dark for the PSCs based on P3HT5PC 60 BM show quite different charge injection and rectification behavior for the devices with different cathode. At positive bias of 1.5 V, the injection current of the device with Al cathode is 72.3 mA/cm 2 , with a rectification ratio of 8.03 3 10 3 at 61.5 V. While that of the device with Ca/Al as cathode increased to 250.2 mA/cm 2 with a rectification ratio of 1.39 3 10 4 . Interestingly, when utilizing a-ZrAcac/Al as cathode, the injection current is dramatically increased to 388.8 mA/cm 2 with a rectification ratio of 3.09 3 10 4 , which is even much higher than that of the Ca/Al based device. The ZrAcac CBL works very well in PBDTBDD5PC 60 BM based devices, the injection current at 1.5 V and rectification ratio at 61.5 V for device with ZrAcac buffer layer (Figure 4(b) inset) reach 436.0 mA/cm 2 and 9.91 3 10 3 , respectively, much higher than that of the PSCs with Ca/Al (275.9 mA/cm 2 , 2.39 3 10 3 ) or Al (59.7 mA/cm 2 , 1.90 3 10 3 ) as cathode. The enhanced charge injection and improved rectification ratio also can be seen in PC 70 BM based P3HT (Figure 4(c)) and low bandgap PBT1 (Figure 4(d)) and PBDTTT-C-T (Figure 4(e)) devices with a-ZrAcac CBL.
The PSCs based on P3HT5PC 60 BM with only Al cathode gives a PCE of 2.65%, with an open-circuit voltage (V oc ) of 0.56 V, a short-circuit current density (J sc ) of 9.55 mA/cm 2 , and a fill factor (FF) of 49.6%. In contrast, the four parameters of V oc , J sc , FF, and PCE for the device with a-ZrAcac/Al cathode are all enhanced to 0.60 V, 10.66 mA/cm 2 , 66.1%, and 4.23%, respectively. The pho-tovoltaic performance of the device with a-ZrAcac/Al cathode is also significantly improved than that of the device with traditional Ca/Al cathode with PCE increased by 12.8%. Furthermore, the a-ZrAcac CBL shows excellent performance in the PSCs based on low bandgap D-A copolymer PBDTBDD5PC 60 BM, as shown in Figure 4(b) and Table 1. The control device with only Al cathode shows a PCE of 5.72%, with a V oc of 0.78 V, a J sc of 12.10 mA/cm 2 , and an FF of 60.6%. With Ca/Al cathode, a similar J sc is obtained, while both the V oc and the FF are improved to 0.86 V and 70.8%, respectively, achieving an enhanced PCE of 7.34%. Surprisingly, the V oc , J sc and PCE of the PSCs with a-ZrAcac/Al cathode are all greatly enhanced to 0.88 V, 14.28 mA/cm 2 and 8.75%, respectively. The best device shows a PCE of 9.23%, a V oc of 0.89 V, a J sc of 14.25 mA/cm 2 , and a FF of 72.7%, among the highest values reported in the literature so far for PSCs. For the significantly increased V oc of the devices with Ca/Al or a-ZrAcac/Al cathode in comparison with that with Al cathode, it should be ascribed to the lower work function of Ca and ZrAcac compared with that of Al. In comparing the a-ZrAcac/Al based devices with the Ca/Al based devices, the improved PCE is mainly attributed to the enhanced J sc and FF. The increased FF for the devices with a-ZrAcac/Al cathode could result from the enhanced charge extraction, increased light harvest and greatly decreased series resistance 44 (see Table 1). The series resistance (R s ) of the PSCs based on P3HT5PC 60 BM is decreased from 13.0 ohm for Al cathode to 3.5 ohm for Ca/Al cathode and to 2.3 ohm for the a-ZrAcac/Al cathode, and that of the PSCs based on PBDTBDD5PC 60 BM is decreased from 11.5 ohm for Al cathode to 2.0 ohm for Ca/Al cathode and to 1.2 ohm for the a-ZrAcac/Al cathode. The a-ZrAcac CBL is of great compatibility with a variety of photoactive layers and shows good photovoltaic performance. As shown in Figure 4(c-e), with modified by a-ZrAcac layer, the PCE of the devices based on P3HT5PC 70 BM, PBT15PC 70 BM, and PBDTTT-C-T5PC 70 BM reaches 4.01%, 7.06%, and 7.55%, respectively, increased by 42%, 80%, and 66%, in comparison with the devices with bare Al cathode. It's also higher than that (3.88%, 6.34% and 7.22%, respectively) of the devices with Ca/Al electrodes. Comparing the performance of PC 70 BM based devices, the enhanced PCE for a-ZrAcac buffered devices can mainly attributed to the enhanced V oc , since the J sc and FF is similar to the devices with Ca/Al cathodes. The improved V oc should attribute to the interfacial dipoles 47 formed by inserting a-ZrAcac layer as shown in Figure 1(c).
The J-V curves of PBDTBDD5PC 60 BM and P3HT5PC 60 BM based devices with varied a-ZrAcac thickness are shown in Figure 4(f) and Figure S6, respectively, and the parameters of the devices are summarized in Table S2 and Table S3 in SI, respectively. The J sc is insensitive to a-ZrAcac layer thickness in a wide thickness range, while the overall performance critically depends on its thickness, because a too thick interfacial layer will induce a high series resistance, while a too-thin layer could not provide an ohmic contact for electron extraction.

Discussion
In order to further elucidate the increased J sc for the devices with a-ZrAcac CBL, we compared the external quantum efficiency (EQE) spectra of the PSCs with Al, Ca/Al and a-ZrAcac/Al cathodes, as given in Figure 5(a) (PBDTBDD5PC 60 BM based devices) and Figure S7 (P3HT5PC 60 BM based devices) in SI. As shown in Figure 5(a), in comparison with the devices with Al and Ca/Al cathode, the EQE spectra of PSCs based on PBDTBDD5PCBM with a-ZrAcac CBL exhibits enhanced light response in the whole absorption band of 300-700 nm, giving a notable enhancement in wavelength range from 380 nm to 500 nm, where it is a deep valley for the devices with Al and Ca/Al cathode due to the limited absorption of the polymer as shown in Figure S8 in SI.
To further clarify the different light response behavior, we measured the reflectance spectra of the devices with Al and a-ZrAcac/Al cathode. As shown in Figure 5(b), the reflectance of the device with a-ZrAcac/Al cathode is lower than that of the device with only Al cathode in the wavelength range of 400-650 nm. Actually, the highly transparent characteristic and appropriate thickness of the a-ZrAcac layer could be its advantage on the reflectance over Ca or other organic CBL. To investigate the relationship between the enhanced light harvest (reduced light reflectance) and the increased EQE values (photocurrent response), we derived the additional absorption, Da Abs , and the additional EQE, Da EQE , from the reflectance and the EQE spectra, respectively, where Da Abs 5 2ln(Ref Zracac/Al / Ref Al ), and Da EQE 5 2ln(EQE Zracac/Al /EQE Al . Figure 5(c) shows the Da Abs and Da EQE curves as the function of wavelength. It can be seen that the shape of Da EQE curve is very similar to that of the Da Abs curve, indicating that the enhanced photocurrent in the device with a-ZrAcac/Al cathode can be mainly ascribed to the additional absorption of the photoactive layer in the device.
To gain further insight into the origin of the enhanced J sc in the device, one-dimensional transfer matrix formalism 53 based on optical modelling calculations was conducted, where experimentally measured refractive indices and extinction coefficients were used. The exciton generation rate versus position in the PBDTBDD: PC 60 BM active layer (100 nm) is shown in Figure 6(a) for devices with Ca/Al and a-ZrAcac/Al cathode, verifying that the device with a-ZrAcac/Al cathode can generate more carriers than that of the Ca/ Al based device, which accounts in part for the higher J sc of the device with a-ZrAcac/Al cathode. The calculated exciton generation rate is in good agreement with EQE measurement as shown in Figure 5(a). The normalized modulus of the optical electric field is shown in Figure 6(b) for the PSCs with an active layer thickness of 100 nm and with Ca/Al and a-ZrAcac/Al cathode for a wavelength of 430 nm. This wavelength has been chosen as the difference in EQE, and absorption between the two device stacks is most prominent at 430 nm. Modified with a-ZrAcac causes little shift of the maximum of the electric field but greatly increases the overall modulus throughout the whole active layer. The calculated results of enhanced light distribution within the active layer for a-ZrAcac based device is in agreement with tested reflection spectra as shown in Figure 5(b).
The long term stability of non-encapsulated P3HT5PC 60 BM based PSCs with a-ZrAcac/Al and Ca/Al cathodes were investigated by testing the evolution of the characteristic parameters (V oc , J sc , FF, and PCE) with time in a nitrogen filled glove-box (the concentration of H 2 O and O 2 is less than 20 ppm) as shown in Figure 7. After 720 hours (30 days), the PCEs of the devices with Ca/Al drop to about 56% of the initial PCEs. The observed decrease in the PCE is mainly due to a deterioration of both J sc and FF, with a little drop (5%) of V oc . On the contrary, the devices with a-ZrAcac/Al cathodes were found to show dramatically improved stability and after 30 days retained 93% of its initial PCE under N 2 . This is in excellent agreement with the stability of PSCs using solution processed PFN layers reported previously 32 . These results indicate that the a-ZrAcac interfacial layer can effectively improve the long term stability of PSCs.
In summary, we successfully demonstrate high performance PSCs by employing as-prepared ZrAcac film spin-cast from its ethanol solution as CBL. The devices with a-ZrAcac/Al cathode show decreased series resistance and enhanced photocurrent due to the remarkable improvement of the charge extraction and light harvesting. The PCE of the device based on P3HT5PC 60 BM with a-ZrAcac CBL reaches 4.23% under the illumination of AM1.5G 100 mW/cm 2 . When choosing D-A copolymer PBDTBDD as electron donor, an average PCE of 8.75% and a best PCE of 9.23% was achieved, which is greatly improved (19% enhancement) in comparison with that (7.34%) of the device with traditional Ca/Al cathode. Our findings indicate that a-ZrAcac is a promising CBL for the fabrication of large area flexible PSCs.

Methods
Device fabrication and characterization. The ITO glass (sheet resistance:10 V/sq, CSG Holding, China) was sequential ultrasonic cleaned in detergent, deionized water, acetone, and isopropanol, and then treated in an ultraviolet-ozone chamber (Jelight, USA) for 20 min. PEDOT:PSS (Clevious P VP AI 4083, H. C. Stark) aqueous solution was filtered through a 0.45 mm filter and spin-coated at 2000 rpm for 60 s on the pretreated ITO substrate, followed by 150uC baking for 10 min in air, achieving a 30 nm thick PEDOT:PSS layer. Subsequently, the PEDOT:PSS modified substrate was transferred to a nitrogen-filled glove-box for photoactive layer and cathode buffer layer preparation. The P3HT (Rieke Metals) and PC 60 BM or PC 70 BM (Nano-C) blend layer was prepared by spin-coating (800 rpm) the 1, 2-dichlorobenzene (ODCB) solution of P3HT and PC 60 BM or PC 70 BM (151 w/w, polymer concentration of 20 mg/mL) on the PEDOT:PSS modified ITO electrode for 20 s, followed by solvent-annealing 54 in covered glass Petri dish, obtaining a 240 nm thick Instrumentation. The surface morphologies of the P3HT5PC 60 BM and PBDTBDD5PC 60 BM photoactive layers with and without ZrAcac layer were analyzed using a VEECO DICP-II atomic force microscope (AFM) operated in the tapping mode under ambient atmosphere at room temperature. The reflectance spectra were measured by LAMBDA 950 UV/Vis/NIR Spectrophotometer with the device structures of ITO/PEDOT5PSS/PBDTBDD5PC 60 BM/Al and ITO/ PEDOT5PSS/PBDTBDD5PC 60 BM/a-ZrAcac/Al. The thickness of the photoactive layer was measured by Ambios Technology XP-2 surface profilometer. An ESCA Lab220i-XL electron spectrometer from VG Scientific using 300 W Al K a radiation operated at a base pressure of 3 3 10 29 mbar was used to obtain XPS data. The