An organic semiconductor as an anode-bu ﬀ er for the improvement of small molecular photovoltaic cells

A hole transport material, 1,3,4,5,6,7-hexaphenyl-2-{3 0 -(9-ethylcarbazolyl)}-isoindole (HPCzI), was used to serve as an e ﬃ cient organic anode-bu ﬀ er for organic photovoltaic cells (OPVs) with a bulk heterojunction structure comprising a copper phthalocyanine (CuPc) and fullerene (C60) mixture. Compared with a CuPc or molybdenum trioxide (MoO 3 ) anode-bu ﬀ er layer, the HPCzI based OPV device exhibits improved performance. Due to its highest occupied molecular orbital (HOMO) energy level being well matched with ITO and its relatively high hole conductivity, HPCzI can facilitate hole-extraction and reduce device resistance, leading to a signi ﬁ cantly improved ﬁ ll factor (FF) in the OPV device. Furthermore, when HPCzI was doped with MoO 3 , additional promotion of the device performance was achieved, which is supposedly attributed to the increase of the hole transport ability in an anode-bu ﬀ er interface and better ohmic contact with ITO.


Introduction
Organic photovoltaic cells (OPVs) are attractive for clean power generation, especially in exible substrates, due to the exibility and solution-processability of organic materials, originating from the weak nature of the van de Waals forces between organic molecules/chains.Other advantages pertaining to OPVs include being inexpensive, low-temperature and easy to fabricate, as well as being compatible with the roll-to-roll method. 1,2ntil now, by the development of new photon-active materials and employing solution processing with lm morphology engineering, the best power conversion efficiencies (PCEs) are reported to be 11-12% for polymer devices 3,4 and 9-10% for small molecule devices, [5][6][7][8] showing their great potential for practical applications.0][11] The interfacial properties can be generally improved by the insertion of buffer layer between electrode and active-layer.It has been reported that besides improving interfacial properties, electrode buffer can also prevent exciton quenching by electrode, 12 facilitate exciton dissociation, 13 improve charge selectivity [14][15][16] and/or induce crystalline phase in active-layer. 17n terms of anode buffer layer (ABL), the commonly used material is poly(3,4-ethylenedioxy-thiophene):poly(styrenesulfonate) (PEDOT:PSS), contributing to its high optical transparency in the visible-NIR region, appropriate work function and good hole transport. 18However, it has been reported that the highly hygroscopic and acidic PEDOT:PSS may interact with the activelayer, 19,20 leading to the chemical instability.Alternatively, various transition metal oxides (TMOs), such as V 2 O 5 , 21 MoO 3 , 22,23 NiO, 24 and WO 3 , 25 have been investigated as ABLs, demonstrating their comparable efficiencies to those using the PEDOT:PSS layer.However, TMOs are limited by the incompatible high deposition temperature, low optical transparency or material toxicity.
Continued research efforts are still being devoted to nd high performance ABL materials.Choi et al. reported polymer based ABLs in OPV.These polymers have strong dipole moment, which decreases the work function of ITO electrode and the surface potential between the active-layer and the ABL, thus hole collection is improved. 26Liu et al. used F 16 CuPc as an ABL in OPV and improved performance was realized. 17They mainly attributed the improvement to the increased crystalline of the CuPc lm when it was deposited on F 16 CuPc lm, which resulted in the increase of the hole mobility and exciton diffusion length of the CuPc lm.Subbiah et al. proposed a thin double-inorganic-organic-interlayer of MoO 3 and poly(9,9-dioctyluorene-co-N-[4-(3-methylpropyl)-diphenylamine]) (TFB), the performance of OPV was constantly improved by the combined effects of electron blocking and enhanced charge extraction from the active-layer to the anode. 27Zhang et al. had demonstrated the improved performance of OPV cells by using a bi-layer of PTFE and MoO 3 as the ABL.The enhanced of holes, effective electron blocking and the suppressed carrier recombination occurred at the interface. 28reviously, our group had designed a hole-transporting material, 1,3,4,5,6,7-hexaphenyl-2-{3 0 -(9-ethylcarbazolyl)}isoindole (HPCzI), which possesses good thermal/chemical stabilities, and it had been successfully utilized in doublelayered OLED devices as hole transport layer. 29In our recent research of mixed small molecules OPV devices, 30 HPCzI was preliminarily used as ABL.To investigate the mechanism of HPCzI as ABL in detail, in this study we prepared a series of OPV devices based on small molecule mixture composing cupper phthalocyanine (CuPc) and fullerene (C60).When HPCzI was used as the anode buffer, 40% and 3% PCE improvements were obtained comparing with devices based on ABL of MoO 3 and CuPc.We further investigated the OPV devices with MoO 3 doped HPCzI as ABL.Through varying doping concentrations, it was found that different doping level of MoO 3 in HPCzI resulted in different OPV performance.When MoO 3 doping concentration was 25%, compared to pure HPCzI ABL, the PCE of OPV device increased from 1.62% to 1.71%.To interpret device physics, cyclic voltammogram, UV-visible absorption spectra, as well as hole-only devices were further studied.

Device fabrication and characterization
Before used in device, indium tin oxide (ITO) glass substrates were cleaned by sonication in a sequence of diluted Hellmanex solution, deionized (DI) water, acetone, and ethyl alcohol baths for 15 min each and then dried at 80 C for 1 h prior to use.Aer drying, the cleaned ITO-coated glasses were treated by oxygen plasma for 50 s before loaded into the vacuum fabrication chamber.All layers were deposited by thermal evaporation in vacuum chamber (3.5 Â 10 À4 Pa).The deposition rates were 0.5 Å s À1 for the organic materials and MoO 3 layer, and 10 Å s À1 for the metal cathode.The thickness of each layer was monitored by quartz crystal monitor.The device active area was 0.1 cm 2 dened by the overlap of anode and cathode.
The current-voltage (J-V) characteristic of the studied devices were measured with a computer-controlled Keithley 2400 source meter under the AM 1.5G illumination (100 mW cm À2 ) of a solar simulator.All the measurements were performed in atmospheric environment without device encapsulation.Absorption spectra of solutions (about 1.0 Â 10 À5 M in DCM (dichloromethane)) were recorded with Shimadzu UV-3600 spectrophotometer from 250 nm to 800 nm.Electrochemical analysis was performed on a Bioanalytical Systems CHI660E operating in cyclic voltammetry (CV) mode.A threeelectrode system consisting of glass carbon and platinum wire as working and auxiliary electrodes, and an Ag/AgCl as reference electrode were used.The scan rate was 100 mV s À1 .Tetrabutylammonium perchlorate (0.1 M) in DCM was used as a supporting electrolyte.
Fig. 2a depicts the J-V characteristics of the best OPV devices.The device performance data deduced from the J-V characteristics are listed in Table 1, with shunt resistance (R sh ) and series resistance (R s ) included. 31Among the three devices with pure material (i.e., CuPc, HPCzI, and MoO 3 ) as ABL, HPCzI device has the highest PCE of 1.62%, showing the potential of HPCzI as ABL in OPV devices.Since V OC is mainly determined by the energy level difference between HOMO of the donor and LUMO of the acceptor; 32 and all our devices are based on the same donor-acceptor system, they show similar V OC .While the slight differences in V OC s still exist and agree well with the values of R s .The relatively large R s of the MoO 3 device could be caused by the unsatised lm formation of MoO 3 , e.g., rough surface, pin hole and overshoot, 33 which may induce bad morphologies of the next active-layer.The decreased V OC of MoO 3 device can be ascribed to a inhomogeneous surface. 34For devices with MoO 3 doped HPCzI as ABL, when doping concentration was 25%, best performance was achieved, with the J sc ¼ 6.63 mA cm À2 , V OC ¼ 0.49 V, FF ¼ 53%, PCE ¼ 1.71%.
Fig. 2b shows IPCE of the 25% MoO 3 doped HPCzI device and the lm absorption spectrum for the active layer (75% CuPc:C60 lm) of this device, as well as the absorption curves for HPCzI, C60 and CuPc lms.In the absorption spectrum of 75% CuPc:C60 blend lm, there are only peaks at 620-693 nm, showing discrepancy with the IPCE curve which presents peaks both at 400-500 nm and 620-693 nm regions.This can be understood by examining the device structure and the absorption of C60 lm.In this device above the active layer, there is an additional layer of C60 (20 nm) functioning as cathode buffer. 30 can be seen from Fig. 2b, the C60 lm presents an absorption peak at 400-500 nm, which contribute mainly to the IPCE response at this region.Based on this IPCE curve, the theoretical J sc can be calculated as 6.09 mA cm À2 , which is slightly smaller than the value of 6.63 mA cm À2 by J-V curve.The difference between integrated photocurrent and measured photocurrent may result from the mismatch between the simulated light and the AM 1.5G spectra.
To investigate the working principles of HPCzI as ABL in OPV device, we studied its electronic structure by cyclic voltammogram (CV) along with UV-visible absorption spectra, and the results are shown in Fig. 3.The CV curve of HPCzI shows a well reversible redox wave for oxidation, which indicates excellent electrochemical stability for hole injection and transport, benecial to its usage as ABL in OPV devices.The oxidation potential (E OX ), determined as the mean value between the anodic current peak voltage and its corresponding cathodic peak, is 0.33 V (vs.Ag/AgCl).Based on this, the HOMO was obtained using the equation: [35][36][37][38][39] E HOMO ¼ À4.72 eV À E OX ¼ À5.05 eV, while LUMOs were obtained via HOMO plus the energy gap of the absorption cutoff from absorption spectra, i.e., E LUMO ¼ E HOMO + 2.90 eV ¼ À2.15 eV.These HOMO and LUMO values agree well with those obtained by UPS in our previous study. 29ccording to the literature, the work function of ITO is 4.70 eV, 40 and the HOMO/LUMO of CuPc and MoO 3 are   To further investigate the mechanism for the improvement of device with HPCzI ABL, a series of hole-only devices were fabricated with the structure of ITO/NPB (10 nm)/X (60 nm)/NPB (100 nm)/Al, where X are pure lms of CuPc, HPCzI, MoO 3 and MoO 3 doped HPCzI lms, respectively.The NPB layer contacting with ITO plays a role of hole injection and circumvents the inuence of the studied lms contacting with ITO directly.The NPB layer before Al blocks electron injection as well as prevents device leak-current.This was especially important for a MoO 3 lm.We found that aer the pure MoO 3 lm deposition, if the following NPB layer was not thick enough, the device oen suffered from short circuit.Fig. 5 shows the J-V characteristics of these hole-only devices.As can be seen, there exist signicant differences in the hole transport properties of these lms, and 50% MoO 3 doped HPCzI lm exhibits the highest values.
The hole mobilities of the lms can be extracted from these J-V curves at large voltage, where the current is limited by the space charge accumulation and can be described by Mott-Gurney's space charge limited current (SCLC) model, here, 3 0 is the free space permittivity having the value of 8.85 Â 10 À12 F m À1 ; 3 r is the relative permittivity of material, with values of 3.4, 3, 9.93 for CuPc, 43 organic materials (e.g., HPCzI) 44 and MoO 3 , 45 respectively.Thickness, d, is 170 nm in our devices.
Taking the square root form of eqn (1), we got, Drawing J 0.5 -V curves, and tting the SCLC regime (shown as red lines in Fig. 6), slopes (K) of each curve can be obtained.Then, based on eqn (2), the hole mobility can be calculated according to eqn (3): Since these hole-only devices contain organic layers other than the concerned ABL, the mobility values cannot exactly  represent those of the corresponding ABL.However, with the similar device structures, for the hole-conduction comparison among these ABLs, these are adequate.These data are shown in Table 2.It was found that the hole mobility of the HPCzI lm was slightly higher than the CuPc lm.However, the OPV device with ABL of CuPc showed smaller R s .This contradictory result may come from the better interfacial property when CuPc was used as ABL for active-layer containing the same component of CuPc.Nevertheless, the HPCzI device gave better PCE, which agrees with the larger FF resulting from its larger R sh (see Table 1).The lm comprising a pure layer of MoO 3 showed an even higher value of hole mobility which turned out to be three or four order of magnitude higher than that of CuPc and HPCzI lms.However, pure MoO 3 based OPV device presented the poorest PCE with high R s and low R sh .This may be due to rough surface morphology of MoO 3 with overshoots, 33 causing bad interface between the MoO 3 and the active-layer, hence leading to high R s and low R sh .Furthermore, when doping MoO 3 into HPCzI, compared with pure HPCzI device, hole mobility signicantly increased.Although the 50% MoO 3 doped HPCzI possesses the highest mobility, its OPV device is not the best.More MoO 3 doped in HPCzI may cause worse lm morphology, hence deteriorating device interface.As a result, the device with HPCzI containing 25% MoO 3 as ABL shows the best performance.

Conclusions
In summary, we fabricated CuPc:C60 OPV devices with CuPc, HPCzI and MoO 3 as ABLs.In the case of pure HPCzI, larger FF than CuPc and higher J sc than MoO 3 were obtained, due to its lower leakage current and improved interface, respectively.Through energy level analysis, the HOMO level of HPCzI provided satisfactory energy alignment, which may lead to efficient hole extraction and reduced charge recombination.For further improvement, devices with MoO 3 doped HPCzI as ABL were fabricated.Study of the hole-only devices of the four ABLs material/composition indicates that the hole mobility of the MoO 3 doped HPCzI devices is three or four order of magnitude higher than that for pure CuPc and pure HPCzI devices.However, the device performance decreased dramatically with the MoO 3 concentration increased to 50%, which might be mainly due to the worse lm morphology caused by more MoO 3 composition.Therefore, the device with 25% MoO 3 doped HPCzI as ABL showed the most highly performance.

Table 1
Photovoltaic characteristics of OPV devices