Article pubs.acs.org/Langmuir
Time-Resolved Neutron Reflectometry and Photovoltaic Device Studies on Sequentially Deposited PCDTBT-Fullerene Layers Andrew J. Clulow,† Chen Tao,† Kwan H. Lee,† Marappan Velusamy,† Jake A. McEwan,† Paul E. Shaw,† Norifumi L. Yamada,‡ Michael James,§,∥ Paul L. Burn,*,† Ian R. Gentle,*,† and Paul Meredith† †
Centre for Organic Photonics & Electronics, The University of Queensland, St Lucia, QLD 4072, Australia Neutron Science Division, High Energy Accelerator Research Organization (KEK) - 203-1 Shirakata, Tokai, Naka, Ibaraki 319-1106, Japan § The Bragg Institute, Australian Nuclear Science and Technology Organisation, Locked Bag 2001, Kirrawee DC, NSW 2232, Australia ∥ The Australian Synchrotron, 800 Blackburn Road, Clayton, VIC 3168, Australia ‡
S Supporting Information *
ABSTRACT: We have used steady-state and time-resolved neutron reflectometry to study the diffusion of fullerene derivatives into the narrow optical gap polymer poly[N-9″hepta-decanyl-2,7-carbazole-alt-5,5-(4′,7′-di-2-thienyl-2′,1′,3′benzothiadiazole)] (PCDTBT) to explore the sequential processing of the donor and acceptor for the preparation of efficient organic solar cells. It was found that when [6,6]phenyl-C61-butyric-acid-methyl-ester (60-PCBM) was deposited onto a thin film of PCDTBT from dichloromethane (DCM), a three-layer structure was formed that was stable below the glass-transition temperature of the polymer. When good solvents for the polymer were used in conjunction with DCM, both 60-PCBM and [6,6]-phenyl-C71-butyric-acid-methyl-ester (70-PCBM) were seen to form films that had a thick fullerene layer containing little polymer and a PCDTBT-rich layer near the interface with the substrate. Devices composed of films prepared by sequential deposition of the polymer and fullerene had efficiencies of up to 5.3%, with those based on 60PCBM close to optimized bulk heterojunction (BHJ) cells processed in the conventional manner. Sequential deposition of pure components to form the active layer is attractive for large-area device fabrication, and the results demonstrate that this processing method can give efficient solar cells.
■
butyric-acid-methyl-ester (70-PCBM, Figure 1) combination.6 In this latter report TEM analysis appeared to show that a BHJ film was only formed when good solvents for the polymer were used in conjunction with dichloromethane to dissolve the fullerene followed by postdeposition thermal annealing. However, TEM analysis does not allow the evolution of the film structure to be monitored in real time, and the amount of 70-PCBM mixed into the polymer matrix cannot be determined. The most efficient (exceeding 7%) PCDTBTbased OPV devices comprise a bulk BHJ film structure, which is formed by processing a mixture of the donor and acceptor materials from a common solution.10−22 Thus, understanding the true nature of films formed by sequential deposition is important to enable the method to be more commonly used for the higher-efficiency narrow-optical-gap polymers. In this article we study the formation and evolution of the structure of PCDTBT:fullerene films using neutron reflec-
INTRODUCTION There have recently been reports that bulk heterojunction (BHJ) organic solar cell junctions comprising poly(3-nhexylthiophene) (P3HT) and fullerene acceptor [6,6]-phenylC61-butyric-acid-methyl-ester (60-PCBM, Figure 1) can be formed by sequential deposition of the two materials.1−5 In these reports the BHJ film structure was formed by depositing the fullerene acceptor 60-PCBM onto P3HT from separate solutions, followed by a thermal annealing step to drive the mixing of the active components.1−3,5 Under these conditions the junctions formed gave device performances equivalent to those of BHJ layers deposited from solutions containing both the donor and acceptor. It is important to note that most of the studies on the sequentially deposited films focus only on the ascast and post-thermal or solvent “annealed” films.2,3,6−9 Given the success of the sequential deposition process it is perhaps surprising that there have not been a plethora of studies on other polymer:fullerene combinations. One such study has been on the high-efficiency, narrow-optical-gap poly[N-9″heptadecanyl-2,7-carbazole-alt-5,5-(4′,7′-di-2-thienyl-2′,1′,3′benzothiadiazole)] (PCDTBT, Figure 1)/[6,6]-phenyl-C71© 2014 American Chemical Society
Received: May 29, 2014 Revised: August 28, 2014 Published: September 15, 2014 11474
dx.doi.org/10.1021/la5020779 | Langmuir 2014, 30, 11474−11484
Langmuir
Article
NR Sample Preparation. In the following descriptions PCBM is used to denote both 60- and 70-PCBM where preparations were identical. All film deposition was performed under a nitrogen atmosphere in a glovebox within a class 1000 clean room. NR samples were prepared on 50-mm-diameter silicon wafers. Prior to film deposition the substrates were cleaned in piranha solution (a 2:1 v/v mixture of 98% sulfuric acid and 30% aq. hydrogen peroxide) and then ultrasonicated in acetone and 2-propanol for 5 min each prior to film deposition. The following procedures were used to deposit the photoactive layers for both the NR measurements and device characterization unless otherwise stated. Dry PCDTBT/PCBM Films Prepared Using Orthogonal Solvents. Solutions of PCDTBT (7 mg mL−1) in 1,2-dichlorobenzene (DCB) and PCBM (10 mg mL−1) in dichloromethane (DCM) were prepared and both solutions were filtered. Initially a film of PCDTBT was spin coated onto the desired substrate at 700 rpm for 60 s, and the film was allowed to dry at ambient temperature for 30 min (henceforth referred to as dry). A solution of PCBM was then spin coated onto the PCDTBT film at 4000 rpm for 10 s. Wet PCDTBT/PCBM Films Prepared Using Orthogonal Solvents. Solutions of PCDTBT (7 mg mL−1) in DCB and PCBM (10 mg mL−1) in DCM were prepared and both solutions were filtered. Initially a film of PCDTBT was spin coated onto the desired substrate at 700 rpm for 60 s. A solution of PCBM was then immediately spin coated onto the PCDTBT film at 4000 rpm for 10 s. PCDTBT/PCBM Films Prepared Using a Cosolvent Mixture. A solution of PCDTBT (7 mg mL−1) in DCB was prepared and filtered. A solution of PCBM (28 mg mL−1) in a 79:15.75:5.25 mixture of DCM, DCB, and chlorobenzene (CB) by volume was prepared and filtered. Initially a film of PCDTBT was spin coated onto the desired substrate at 700 rpm for 60 s, and the film was allowed to dry at ambient temperature for 30 min. A solution of PCBM was then spin coated onto the PCDTBT film at 4000 rpm for 10 s. Neutron Reflectometry (NR) Measurements and Analysis. NR measurements on the PCDTBT/PCBM films were performed using the BL16 SOFIA time-of-flight neutron reflectometer at JPARC/MLF, Japan.31 25 Hz neutron pulses were generated with a wavelength band of 2.0 Å < λ < 8.8 Å and detected using a twodimensional ZnS/6LiF scintillation detector. Steady-state reflected beam profiles were recorded at reflection angles of 0.3, 0.7, and 1.8° to give a Q range of 0.007−0.197 Å−1. The spectra were allowed to accumulate until integrated counts at the detector of 100 000 (0.3°), 40 000 (0.7°), and 80 000 (1.8°) were reached. Time-resolved reflectivity profiles were recorded in double-frame mode (neutron wavelength band 2.0 Å < λ < 17.8 Å) at a reflection angle of 0.7° to give a Q range of 0.009−0.077 Å−1. Experiments were performed under a coarse vacuum produced by a dry scroll pump. A custom-built experimental cell with in situ annealing capabilities was used for the NR measurements. For time-resolved NR thermal annealing measurements an aluminum block heating stage, isolated from the neutron cell by a ceramic stand, was heated (2 °C min−1 ramping rate) with two cartridge heaters, with the temperature controlled by a Watlow series 988 temperature controller. Ex situ annealing was performed under a nitrogen atmosphere in a glovebox with an As One HP-2SA digital hot plate. Analysis of the reflectivity profiles was performed using the Motofit reflectometry analysis program.32 All of the NR fits were modeled with a 5 to 6 Å thick oxide layer on the surface of the silicon substrate with a scattering length density (SLD) of 3.47 × 10−6 Å−2. The SLDs of silicon and air were taken to be 2.07 × 10−6 and 0.00 × 10−6 Å−2, respectively. The weight percentages and volume fractions of PCDTBT and PCBM were determined from the optimized fitting models following a previously reported method.27 For a description of the analysis and the origin of the errors see the Supporting Information for this article. OPV Device Fabrication and Measurements. Prior to device fabrication prepatterned ITO-coated glass substrates were ultrasonicated in detergent (Alconox), deionized water, acetone, and 2propanol for 5 min each. The substrates were blown dry with nitrogen before PEDOT:PSS was spin coated on top at 5000 rpm for 60 s. The
Figure 1. Chemical structures and molecular formulae of PCDTBT and the fullerene acceptors.
tometry (NR). NR has been used to probe the location and migration of the components within P3HT:60-PCBM films2,4,5,23,24 and, more recently, the out-of-plane phase distribution in blended PCDTBT:fullerene BHJ films deposited by spin- or spray-coating.25−27 These latter reports suggest that some accumulation of PCDTBT occurs at the interface when the BHJ is deposited onto poly(3,4ethylenedioxythiophene):poly(styrenesulfonate) (PEDOT:PSS) and that thermal annealing of the films causes interdiffusion of the polymer and fullerene phases. However, these studies again utilize only steady-state NR measurements of as-cast and annealed films. We have examined PCDTBT/60- and /70-PCBM films in which the polymer and fullerenes were deposited sequentially. Initially steady-state NR was used to probe the composition of the as-cast films and the changes in their structure upon thermal annealing. Time-resolved NR was then used to observe the interdiffusion of the active components in real time. The changes observed in the physical structure of the films were then related to the performance of operational organic solar cells with active layers deposited under conditions identical to those for the samples studied by NR. An important result from the study was that when cosolvents were used for the deposition of the fullerene the BHJ proposed in the original study6 was not formed. Instead, a thin polymer layer at the interface and bulk layer consisting of mainly fullerene was created.28−30 Nevertheless, devices composed of PCDTBT and 70-PCBM deposited using the cosolvent approach had efficiencies of up to 5.3% in spite of the small amount of polymer in the bulk of the film.
■
EXPERIMENTAL DETAILS
Materials and General Methods. The PCDTBT used in this work was prepared in-house following Suzuki cross-coupling protocols previously described.10 GPC (140 °C, 1,2,4-trichlorobenzene) [M̅ n = 17.9 kDa, M̅ w = 47.3 kDa], 60-PCBM, and 70-PCBM were purchased from Nano-C and used as received. Poly(3,4ethylenedioxythiopehene):poly(styrenesulfonate) (PEDOT:PSS, Baytron P VP AI 4083) was used as received. The glass-transition temperature (Tg) of the PCDTBT was recorded on a PerkinElmer diamond DSC differential scanning calorimeter calibrated to an indium standard. Thermograms were recorded between 0 and 200 °C with a heating and cooling rate of 50 °C min−1, and the temperature was held constant for 1 min prior to the start of each heating/cooling period. Surface contact profilometry was performed on a Veeco Dektak 150. 11475
dx.doi.org/10.1021/la5020779 | Langmuir 2014, 30, 11474−11484
Langmuir
Article
Figure 2. (A, C) Neutron reflectivity profiles for the wet and dry films with 60-PCBM cast from DCM. Individual points indicate experimental data, solid lines indicate model fits, and traces are offset for clarity. (B, D) Corresponding SLD versus thickness plots. PEDOT:PSS-coated substrates were then baked on a hot plate at 170 °C for 20 min in air before being transferred to a nitrogen-filled glovebox. Sequential deposition of the active layers onto the PEDOT:PSS film was performed in the same way as for the corresponding NR experiments. Deposition of traditional BHJ active layers from a common solution was performed as described previously.27 The optimized PCDTBT:60-PCBM blend BHJ device was not annealed after film deposition and the optimized PCDTBT:70-PCBM blend BHJ device was annealed at 70 °C for 20 min after film deposition. The samples were then transferred into an evaporation chamber for electrode deposition. The electrodes were thermally evaporated through a shadow mask to create devices with a total area of 0.2 cm2. Approximately 20 nm of calcium (Ca) and 60 nm of aluminum (Al) were thermally deposited in sequence under a vacuum of 1 × 10−6 mbar to complete the cathode as measured by a calibrated quartz crystal monitor for the sequentially deposited devices, while approximately 1 nm of samarium (Sm) and 60 nm of Al were deposited to create the cathode for the traditional BHJ devices. In our experience Ca/Al electrodes work well when the device has a high PCBM content at the cathode interface. The J−V characteristics of the devices were measured by a computer-controlled Keithley 2400 source meter in a glovebox filled with nitrogen (O2 < 0.1 ppm and H2O < 0.1 ppm). The simulated Air Mass 1.5 Global (AM1.5G) irradiance was provided by an Abet Sun 2000 solar simulator. At least six devices of each type were measured using a four-wire connection configuration without a shadow mask under an irradiation intensity of ∼100 mW cm−2 as determined by an NREL-certified silicon solar cell. Within the illumination area of 5 cm × 5 cm, the variation of the light intensity was below 3%. The EQE was measured with a PV Systems model QEX7 setup. Care was taken to ensure that in all cases the integrated EQE (IPCE) agreed with the white light short circuit current to within 10%. However, in the
sequentially deposited devices of PCDTBT:60-PCBM the integrated short circuit current density was 20% higher than that measured under AM1.5G. We believe the reason for this apparent discrepancy is increased bimolecular recombination because of the relatively high weight fraction of 60-PCBM.
■
RESULTS AND DISCUSSION The aim of the study was to investigate the out-of-plane film structure of PCDTBT:PCBM films prepared by different methods and the effect on the related organic solar cell performance. In a NR experiment the scattering length density (SLD) for neutrons of each of the layers within a structure is determined from a fitting model (eq 1 in the Supporting Information) derived from the experimental observations. The ability to discern the proportion of PCDTBT and PCBM at a point within a film is dependent on their individual neutron SLDs. The molecular scattering lengths (given by the summation term in eq 1) for the PCDTBT repeat unit, 60PCBM, and 70-PCBM with natural isotopic abundances are 1.47 × 10−3, 4.38 × 10−3, and 5.05 × 10−3 Å, respectively. Because of the relatively low number of hydrogen nuclei (bH = −3.74 × 10−5 Å)33 in the fullerene molecules (Figure 1), their molecular scattering lengths are considerably greater than that of PCDTBT which bears a branched C17H35 chain in the repeat unit. The SLDs of the fullerenes are further enhanced by their relatively high mass densities, recently reported to be 1.61 ± 0.08 g cm−3 for clusters within a PCDTBT film.27 In the same report, the SLD of PCDTBT with a similar molecular weight to that used here was found to be (1.45 ± 0.07) × 10−6 Å−2. The SLDs of 60-PCBM and 70-PCBM within the PCDTBT matrix 11476
dx.doi.org/10.1021/la5020779 | Langmuir 2014, 30, 11474−11484
Langmuir
Article
were (4.66 ± 0.23) × 10−6 and (4.74 ± 0.24) × 10−6 Å−2, respectively, which are similar to values found for layers formed by small organic molecules containing deuterium (bD = 6.67 ×10−5 Å)33 that have been used for studying layered thin film architectures.34 Thus, selective deuteration was not necessary to obtain satisfactory neutron contrast between the polymer and fullerene phases, and this allowed the behavior under thermal annealing to be analyzed without concerns regarding the effect of deuteration on the thermodynamics of the system. Sequentially Deposited Films Spin Coated from Orthogonal Solvents. We first studied PCDTBT:PCBM films formed by the sequential deposition of PCDTBT from 1,2-dichlorobenzene (DCB) followed by dichloromethane (DCM) solutions of either 60-PCBM or 70-PCBM. Two types of sequentially deposited film were prepared in which the solvent in the PCDTBT layer was either allowed to evaporate at ambient temperature for 30 min prior to deposition of the PCBM layer (denoted “dry”) or the PCBM layer was deposited immediately on top of the PCDTBT layer (denoted “wet”). Both wet and dry films were prepared in duplicate with each fullerene, and one film of each type was initially examined by steady-state NR on the SOFIA reflectometer.31 It was found that under steady-state conditions the films containing 60PCBM gave stronger specular neutron reflection than their 70PCBM counterparts. Strong specular reflection from the PCDTBT:70-PCBM films was observed only at the lowest angle of incidence (0.3°) and no fringes were observed within the corresponding Q range (0.008−0.033 Å −1 ). The momentum transfer at the critical edges of around 0.010 Å−1 for the 70-PCBM dry and wet films was consistent with reflection from the underlying silicon substrate. Contact profilometry showed that the films had high root-mean-square (rms) surface roughness values of around 100 Å, which was consistent with the substantial dampening of the Kiessig fringes in the NR measurements, and it was concluded that DCM is a poor solvent for 70-PCBM. In contrast, the films formed by depositing DCM solutions of 60-PCBM onto the PCDTBT layer gave smoother films with surface roughness values
