Accepted for publication in Applied Physics Letters Graphene oxide hole transport layers for large area, high efficiency organic solar cells Chris T. G. Smith1, Rhys W. Rhodes1, Michail J. Beliatis1, K. D. G. Imalka Jayawardena1, Lynn J. Rozanski1, Chris A. Mills1, and S. Ravi P. Silva1,a) 1 Advanced Technology Institute University of Surrey Guildford GU2 7XH, UK Graphene oxide (GO) is becoming increasingly popular for organic electronic applications. We present large active area (0.64 cm2), solution processable, poly[[9-(1-octylnonyl)-9H-carbazole-2,7-diyl]-2,5thiophenediyl-2,1,3-benzothiadiazole-4,7-diyl-2,5-thiophenediyl]:[6,6]-Phenyl C71 butyric acid methyl ester (PCDTBT:PC70BM) organic photovoltaic (OPV) solar cells, incorporating GO hole transport layers (HTL). The power conversion efficiency (PCE) of ~5% is the highest reported for OPV using this architecture. A comparative study of solution-processable devices has been undertaken to benchmark GO OPV performance with poly(3,4-ethylenedioxythiophene) poly(styrenesulfonate) (PEDOT:PSS) HTL devices, confirming the viability of GO devices, with comparable PCEs, suitable as high chemical and thermal stability replacements for PEDOT:PSS in OPV. 1 Accepted for publication in Applied Physics Letters Considerable recent interest in Organic photovoltaics (OPV) can be attributed to their potential for solution-processing, attractive for low-cost, high throughput production. To become price-competitive with the existing solar cell market, novel polymer development has increased device efficiencies, while lowering the manufacturing cost. One recent polymer of interest is Poly[[9-(1-octylnonyl)-9H-carbazole2,7-diyl]-2,5-thiophenediyl-2,1,3-benzothiadiazole-4,7-diyl-2,5-thiophenediyl] (PCDTBT), which combined with the small molecule acceptor [6,6]-Phenyl C71 butyric acid methyl ester (PC70BM), has produced ~6.8% device power conversion efficiencies (PCEs).1 However, PCEs and lifetimes remain limited due to the use of Poly(3,4-ethylenedioxythiophene) poly(styrenesulfonate), (PEDOT:PSS) - the most commonly utilised hole-transport layer (HTL) material to date. PEDOT:PSS, while an excellent material for hole injection, shortens device lifetimes due to its high acidity and hygroscopic nature. 2 As a result, replacement air-stable, long lifetime, solution processable and inert HTL materials are being investigated.3 Many exciting carbon-based materials have been recently developed applicable to large area electronics. Graphene is a highly conducting material that has attracted considerable interest due to its physical, chemical and electrical properties,4 and its potential to be used as a transparent conducting material. Recently, lab-scale graphene devices have been demonstrated via transfer subsequent to chemical vapor deposition,5 or by solution processing, e.g. inkjet printing.6 However, depositing high quality graphene over large areas without conductivity-limiting defects is difficult and requires high temperatures and vacuum conditions for production.5 In contrast, graphene oxide (GO), a graphene derivative, can be solubilized in water and combined in solution with a variety of organic and inorganic materials.7 It retains the high transparency of graphene, and has a work-function similar to PEDOT:PSS, making it a suitable HTL replacement. 8 Chemical modification of GO or combination with different materials also allows for tuning the workfunction, broadening the applicability.9 Here we evaluate the performance of large area (0.64 cm2) PCDTBT:PC70BM solar cells incorporating a GO HTL, and compare the performance with OPV incorporating PEDOT-PSS. 2 Accepted for publication in Applied Physics Letters ITO on glass substrates (15 Ω-1), and BCP were purchased from Luminescence Technology Corp (Taiwan). PEDOT:PSS (Clevios P VP AI 4083, Heraeus GmbH, Germany) was filtered (0.2 µm). GO was produced by the Hummers and Offeman process,10 and exfoliated by sonication. TiOx was produced using a previously reported method.11 Acetone and propan-2-ol (IPA) were purchased from Sigma Aldrich (UK). Solar cells were constructed with architectures of ITO / HTL (GO or PEDOT:PSS) / PCDTBT:PC70BM / ETL (BCP or TiOx) / Al, where ETL = electron transport layer. Aqueous GO was spincast onto ITO substrates (1000 rpm, 1 min) to give 2-3 nm thick films. PEDOT:PSS was sonicated for 5 min. before spincoating (5000 rpm, 1 min) giving a 40 nm film. Both GO and PEDOT:PSS films were annealed (200°C) for 10 min. PCDTBT:PC70BM (1:4, 7mg/mL PCDTBT concentration) were dispersed in dichlorobenzene (DCB) and chlorobenzene (CB) (1:3). This was spincoated (5500 rpm) for a ~70 nm film before annealing (80°C) for 10 min. prior to depositing electron transport layers. For TiO x devices, the TiOx precursor was spincoated on top of the PCDTBT:PC70BM layer, giving a 10 nm film, then annealed in air (80°C) for 10 min. For BCP devices, BCP was evaporated onto to the PCDTBT:PC 70BM giving a 5 nm layer. Finally, Al electrodes (100 nm thick) were deposited by evaporation, producing a 0.90 cm2 active area. The GO was examined using UV-visible and Raman spectroscopy (not reported), FTIR (See supplemental material)12 and compared to the literature,13, 14 to confirm material quality and purity. The electrical properties of the cells were characterized by a four-probe voltage source/ammeter (Keithley 2425). Current-voltage responses were taken using a solar simulator (Model 10500, Abet Technologies) at 1000 Wm-2. During measurements a mask with a 0.64 cm2 aperture was used in order to remove errors associated with optical wave-guiding, shadow effects, etc. SEM images (Figure 1a) show a complete surface coverage of GO flakes on the ITO. AFM imaging of individual flakes (Figure 1b) confirm the presence of single layers of ~1 nm tall. Given the long axis dimensions of the GO flakes (up to ~1 µm), they may provide short-circuits through the active 3 Accepted for publication in Applied Physics Letters layer if not fully orientated parallel to the substrate. However, good GO coverage is assumed from the AFM and SEM images, and supported by similar Rsh and Voc values for GO and PEDOT:PSS. Shorts in the device would cause significantly lower these values. The majority charge carriers in PCDTBT are holes, as with most semiconducting polymers, 15 with PC70BM typically being the electron conductor. The dark current has been examined with respect to electrode limited- (space charge limited current (SCLC),16 Schottky17) and bulk limited- (Fowler Nordheim (FN),18,19 Poole Frenkel (PF)20,21) conduction mechanisms to show whether the GO has comparable charge injection ability to PEDOT:PSS, and to elucidate the limiting charge transport mechanism(s). Figures S2 and S3 (See supplemental material)12 show plots of conduction mechanisms fit to the current-voltage (I-V) data for the ITO / HTL / PCDTBT:PC70BM / BCP / Al and the ITO / HTL / PCDTBT:PC70BM / TiOx / Al OPV cells respectively. The characteristic parameters produced for each conduction mechanism are given in Table 1 and are calculated using the following equations. Schottky: Poole Frenkel: Fowler Nordheim: ( ( ( ) ( ) ( ) ) ) where, ( where, where, ( ( ) ) (1) ) (2) (3) where, I, J and E are the current, current density and electric field respectively; A* is the modified Richardson constant; h, k, e, and m* are Plank’s constant (= 6.626x10-34 C), the Boltzmann constant (= 1.38x10-23 JK-1), the elementary charge (= 1.602x10-19 Js) and the effective mass of a hole (assumed to be equivalent to the mass of an electron = 9.11x10-31 kg); ε0 and εr are the vacuum permittivity (= 8.85x10-12 Fm-1) and relative permittivity; κ is the slope of the Fowler Nordheim plot; and ϕ is the respective barrier height for each conduction model. For Schottky and Poole-Frenkel calculations of εr and φ, data points were only used at sufficiently high fields (<105 Vm-1, calculated using a high relative permittivity for conjugated polymers, εr = 12) to render the pre-factors (A*T2 and E, respectively) insignificant. 4 Accepted for publication in Applied Physics Letters The relative permittivity of conducting polymers typically ranges between 2 and 12, 22 and the relative barrier heights between materials can be estimated using the band diagram. The calculated FN barrier heights suggest that FN barrier tunneling is not occurring. On the other hand, the SCLC plots obey Ohm’s law, J  V, at low voltages. At higher voltages, the slope of the graph increases to J  Vn where n > 8 after the trap filled limit, VTFL. If trap-modified SCLC is occurring, the curve will continue on this slope until all the traps are filled, and the current returns to a trap-free situation, i.e. a J  V2 relationship.17 Although there is no evidence of significant portions of the graph displaying a J  V2 character, a trap-modified SCLC mechanism may occur in our diodes, as a number of polymer films, including thiophene based polymers, have been found to have SCLC using ITO substrates as hole injectors.25 Although the barrier heights (ϕS = 0.4-1.0 eV) of the Schottky plots are in the expected range, the relative permittivities (εr = 0.01-0.08) are low, suggesting this mechanism does not accurately represent the measured data. However, the barrier heights (ϕPF = 0.2-0.8 eV) and the relative permittivities (εr = 39) of the majority of the PF plots are in the expected range, suggesting that a PF conduction mechanism is occurring in our devices. Taken together, it is possible that the conduction mechanism is SCLC with a PFfield dependency typical of a hopping mechanism, which is supported by similar PF conduction identified in other polymer blends.26, 27 The presence of trap states in the polymer can also be supported by examining the open circuit voltage, Voc, of the devices. The Voc for PCDTBT:PC70BM solar cells is typically in the region of 0.88 – 0.91 V,28 but the Voc for the GO devices (0.81-0.83 V) is lower. A low Voc indicates a reduction of the flow of charge carriers through the diode (such as would occur with the presence traps), consequently necessitating a higher solar cell operating voltage. However, as the conduction mechanism remains similar for both PEDOT:PSS and GO, the GO is a comparable HTL, and does not hinder OPV charge carrier extraction. 5 Accepted for publication in Applied Physics Letters When illuminated with AM1.5 G solar radiation, (Figure 2) the short circuit current (Jsc) is similar when switching from PEDOT-PSS to GO (Table 2). The fill factor increased for GO devices using TiOx ETLs (from 53.3% to 56.0%), but dropped in devices using BCP ETLs (58.7% to 53.7%), possibly due to incomplete coverage of the BCP over the PCDTBT:PC70BM film. As a result of these differences, there is a negligible decrease in power conversion efficiency (PCE) between PEDOT-PSS and GO for TiOx devices (5.3% vs. 5.1%), but a larger difference for BCP devices (6.0% vs. 5.0%). The reproducibility of device performance can be illustrated by the BCP devices (0.64 cm2), fabricated on different days, which display an average PCE of 4.9 (0.2) % (n=3). Improvements are consistently observed when comparing GO-incorporated devices to those with no HTL, with almost double the PCE with GO. Analysis of the series (Rse) and shunt (Rsh) resistances for all devices (Table I) shows that in all cases Rsh is lower for the GO devices suggesting higher charge carrier losses occurring with GO. These losses can be due to misalignment of the band levels between GO and PCDTBT:PC70BM, short-circuit pathways or charge / mobility imbalances in the device, inducing carrier recombination. Interestingly, the Rse of GO inclusive devices, versus PEDOT-PSS, increases with a BCP ETL, but decreases with TiOx, directly related to the differences observed in the FF and PCE. Analysis of the device physics and morphology will give an insight to the origin of these different values. It should be highlighted that the PCEs seen here (η ~5%) for our GO based devices are the highest reported for this size (0.64 cm2). Comparing with literature, our GO based devices fare well (See supplemental material)12. Typically, comparative devices using different active layers produce a similar relationship between the efficiencies of GO and reference PEDOT:PSS cells. As device efficiencies are known to decrease with increasing active area,30 and reported efficiencies for other OPV incorporating GO were for comparatively smaller active areas, our PCEs are a record for their size, and a milestone in the development of large area OPV. The external quantum efficiency (EQE) of the GO cell (Figure 2) mirrors the shape of the HTLfree device with a maximum at approx. 400 nm and a plateau extending from approx. 440 to 545-580 nm. In comparison, EQE for devices containing PEDOT-PSS show a decrease in EQE to approx. 450 nm for 6 Accepted for publication in Applied Physics Letters BCP as the ETL, and to about 400 nm for TiOx as the ETL, attributed to the absorption maximum for PEDOT-PSS, and remain relatively constant for the rest of the wavelength range. Regardless of the ETL used, EQE is increased for GO devices over those without an HTL, although less enhancement is seen for devices with TiOx. The difference in device efficiencies between PEDOT-PSS and GO incorporated devices can partially be explained by the difference in their EQE maxima- with the decrease seen for GO in the green to red region relative to PEDOT-PSS, equivalent to the maximum absorption range for PCDTBT:PC70BM. In all cases however, the peak EQE approaches 65-72%, with the wavelength range of EQE > 50 % lies typically between 345-620 nm. We have demonstrated that GO is an efficient HTL, producing device performances (PCE ~5%) for PCDTBT:PC70BM solar cells comparable to PEDOT:PSS over large active areas (0.64 cm2). Based on data fitting, a trap modified SCLC conduction mechanism, with PF field dependence, typical of charge hopping can be proposed for both types of devices, indicating that GO is an comparable HTL. Large area, high efficiency GO-based devices were achieved with both BCP and TiOx ETLs with minimal changes in the device performance. Given the comparable performance of a GO HTL, allied with its increased operational stability when compared to PEDOT:PSS,31 it should improve the performance of all solutionprocessable, large area PCDTBT:PC70BM solar cells. ACKNOWLEDGMENTS (This work was supported by the Defense Science and Technology Laboratory (DSTL), via the DSTL UK PhD program (CTGS), the Engineering and Physical Sciences Research Council (EPSRC), through postdoctoral fellowships (to MJB and KDGIJ), and via funds provided through the University of Surrey Knowledge Transfer Account. The authors would like to thank Mr. Ranga Prabhath and Mr. Keyur Gandhi, (ATI, University of Surrey) for their assistance. 7 Accepted for publication in Applied Physics Letters REFERENCES 1 Z. He, C. Zhong, X. Huang, W. Y. Wong, H. Wu, L. Chen, S. Su, Y. Cao, Adv. Mater. 2011, 23, 4636. 2 Y. J Jeon, J. M. Yun, D. Y. Kim, S. I. Na, S. S. Kim, Sol. Energy Mater. Sol. Cells. 2012, 105, 96. 3 G. D. M. R. Dabera, K. D. G. I. Jayawardena, M. R. R. Prabhath, I. Yahya, Y. Y. Tan, N. A. Nismy, H. Shiozawa, M. Sauer, G. Ruiz-Soria, P. Ayala, V. Stolojan, A. A. D. T. Adikaari, P. D. Jarowski, T. Pichler, S. R. P. Silva, ACS Nano. 2013, 7, 556. 4 M. J. Allen, V. C. Tung, R. B. Kaner, Chem. Rev. 2010, 110, 132. 5 Y. Y. Tan, K. D. G. I. Jayawardena, A. A. D. T. Adikaari, L. W. Tan, J. V. Anguita, S. J. Henley, V. Stolojan, J. D. Carey, S. R. P. Silva, Carbon. 2012 50, 668. 6 E. B. Secor, P. L. Prabhumirashi, K. Puntambekar, M. L. Geier, M. C. Hersam, J. Phys. Chem. Lett. 2013, 4, 1347. 7 Y. Zhu, S. Murali, W. Cai, X. Li, J. W. Suk, J. R. Potts, R. S. Ruoff, Adv. Mater. 2010, 22, 3906. 8 S. S. Li, K. H. Tu, C. C. Lin, C. W. Chen, M. Chhowalla, ACS Nano. 2010, 4, 3169. 9 D. R. Dreyer, S. Park, C. W. Bielawski, R. S. Ruoff, Chem. Soc. Rev. 2010, 39, 228. 10 W. S. Hummers, R. E. Offeman, J. Am. Chem. Soc. 1958, 80, 1339. 11 S. Cho, K. Lee, A. J. Heeger, Adv. Mater. 2009, 29, 1941. 12 See supplemental material at [URL will be inserted by AIP] for FTIR, literature comparison and supplementary J-V plots. 13 J. I. Paredes, S. Villar-Rodil, A. Martınez-Alonso, J. M. D. Tascon, Langmuir. 2008, 24, 10560. 14 D. Yang, A. Velamakanni, G. Bozoklu, S. Park, M. Stoller, R. D. Piner, S. Stankovich, I. Jung, D. A. Field, C. A. Ventrice Jr., R. S. Ruoff, Carbon. 2009, 47, 145. 15 A. Babel, S. A. Jenekhe, J. Am. Chem. Soc. 2003, 125, 13656. 16 S. R. P. Silva, G. A. J. Amaratunga, Thin Solid Films. 1994, 253, 146. 17 D. M. Taylor, IEEE Trans. Dielectr. Electr. Insul. 2006, 13, 1063. 18 R. H. Fowler, L. Nordheim Proc. R. Soc. London, Ser. A. 1928, 119, 173. 19 I. D. Parker, J. Appl. Phys. 1994, 74, 1656. 8 Accepted for publication in Applied Physics Letters 20 J. Frenkel Phys. Rev. 1938, 54, 647. 21 S. V. Rakhmanova, E. M. Conwell, Appl. Phys. Lett. 2000, 76, 3822. 22 Q. D. Ling, D. J. Liaw, C. Zhu, D. S. H. Chan, E. T. Kang, K. G. Neoh, Prog. Polym. Sci. 2008, 33, 917. 23 M. D. Bhatt, S. Suzuki, T. Sakurai, K. Akimoto, Appl. Surf. Sci. 2010, 256, 2661. 24 S. H. Park, A. Roy, S. Beaupré, S. Cho, N. Coates, J. S. Moon, A. J. Heeger, Nat. Photonics. 2009, 5, 297. 25 P. W. M. Blom, M. J. M. de Jong, J. J. M. Vleggaar Appl. Phys. Lett. 1996, 68, 3308. 26 A. M. Ballantyne, L. Chen, J. Dane, T. Hammant, F. M. Braun, M. Heeney, W. Duffy, I. McCulloch, D. D. C. Bradley, J. Nelson, Adv. Funct. Mater. 2008, 18, 2373. 27 S. H. Deshmukh, D. K. Burghate, V. P. Akhare, V. S. Deogaonkar, P. T. Deshmukh, M. S. Deshmukh, Bull. Mater. Sci. 2007, 30, 51 28 . T. Y. Chu, S. Alem, P. G. Verly, S. Wakim, J. Lu, Y. Tao, S. Beaupré, M. Leclerc, F. Bélanger, D. Désilets, S. Rodman, D. Waller, R. Gaudiana, Appl. Phys. Lett. 2009, 95, 063304. 29 A. R. B. Mohd Yusoff, H. P. Kim, J. Jang, Org. Electron. 2012, 13, 2379. 30 D. Gupta, M. Bag, K. Narayan, Appl. Phys. Lett. 2008, 93, 163301. 31 I. P. Murray, S. J. Lou, L. J. Cote, S. Loser, C. J. Kadleck, T. Xu, J. M. Szarko, B. S. Rolczynski, J. E. Johns, J. Huang, L. Yu, L. X. Chen, T. J. Marks, M. C. Hersam, J. Phys. Chem. Lett. 2011, 2, 3006. 9 Accepted for publication in Applied Physics Letters FIG. 1. (a) SEM (bar = 500 nm) and (b) AFM (25 μm2, z-scale = 20 nm) images of GO deposited on ITO. Inset in (a) is a SEM (bar = 20 µm) showing the edge of the GO film deposited on ITO. The line scan (Inset in b), corresponding to the white line in (b), shows the height (α) of a single GO flake, ~1 nm. 10 Accepted for publication in Applied Physics Letters FIG. 2. J-V plots (illuminated), band diagrams and EQE spectra for ITO/HTL/ PCDTBT:PC70BM/ETL/Al OPV with BCP (a-c) and TiOx (d-f) as the ETL. Data is given for OPV with no HTL (Black Squares), with PEDOT:PSS (Red Circles) and GO (Blue Triangles). The energy level diagrams of the materials used are shown with data taken from [8, 23, 24]. 11 Accepted for publication in Applied Physics Letters Table I. Approximate relationship between SCLC current and voltage, and slope (-κ), relative permittivity (εr) and corresponding barrier height (ɸ) for Schottky†, Fowler Nordheim and Poole Frenkel‡ conduction mechanisms, calculated from the plots in Figure S2 and S3. SCLC <0.5V >0.5V Schottky† εr Fowler Nordheim ɸ (eV) Slope (=-κ) ɸ (eV) Poole Frenkel‡ εr ɸ (eV) BCP/GO IV IV10 0.0700 0.448 -14.82 0.0054 2.85 0.281 BCP/PP IV IV16 0.0112 0.985 -34.56 2817.3 21.5 0.843 BCP/No HTL IV IV12 0.0245 0.797 -23.51 1.780 8.82 0.622 TiOx/GO IV IV8 0.0716 0.431 -12.43 0.0011 2.65 0.258 TiOx/PP IV IV8 0.0547 0.459 -13.00 0.0016 3.49 0.285 TiOx/No HTL IV IV10 0.0335 0.617 -17.61 0.0349 5.95 0.428 † for data points after the increase in slope at high fields, and above E > 5.50x106 Vm-1 [calculated when εr=12] ‡ for data points after the increase in slope at high fields, and above E > 1.38x106 Vm-1 [calculated when εr=12] 12 Accepted for publication in Applied Physics Letters Table II. Characteristic performance values for ITO/HTL/PCDTBT:PC70BM/ETL/Al OPV. ETL BCP TiOx HTL Voc Jsc FF PCE Rsh Rse (V) (mA.cm-2) (%) (%) (Ω.cm-2) (Ω.cm-2) None 0.45 11.11 51.6 2.6 250.8 5.9 PEDOT:PSS 0.88 11.74 58.7 6.0 342.4 9.3 GO 0.81 11.48 53.7 5.0 245.0 11.8 None 0.46 11.57 50.9 2.7 280.1 6.0 PEDOT:PSS 0.88 11.25 53.3 5.3 348.6 18.0 GO 0.83 11.04 56.0 5.1 288.5 9.9 13