Changes of the initial performance of polymer/fullerene solar cells are investigated by continuous illumination with simulated solar light. The fill factor gradually decreases with exposure time, whereas the series resistance gradually increases. These initial performance changes are ascribed to a marginal change in the surface composition of the active layer.
Recently, polymer solar cells have attracted keen interest as sustainable energy conversion media owing to their potential for low-cost manufacturing and advanced features, including their semi-transparency and light-weight, ultrathin, and flexible form factors.1–4 These characteristics are expected to broaden the application of polymer solar cells towards mobile and/or consumer electronics; no longer limited to conventional solar power modules fixed to land installations and/or buildings. So far, the power conversion efficiency (PCE) of polymer solar cells has been gradually improved by controlling the nanomorphology of the active layer as well as by introducing new tailored polymer materials based on bandgap engineering.5–11 Recent reports have forecast that the PCE of polymer solar cells can reach 11–16 % via further bandgap engineering of semiconducting polymers and acceptor molecules.12, 13
However, in addition to enhancing the efficiency, the stability of polymer solar cells is a critical issue for their commercialization. To date, tens of reports on the stability and/or lifetime of polymer solar cells have appeared.14–26 These reports have shown the importance of encapsulation to avoid attack by oxygen, as well-documented in the study of organic light-emitting devices (OLEDs).27 The studies pay particular attention to the time-dependent decay of short-circuit current density (Jsc) and open-circuit voltage (Voc), in the presence of performance oscillation, in polymer solar cells and/or modules prepared using blend films of poly(3-hexylthiophene) (P3HT) and 1-(3-methoxycarbonyl)-propyl-1-phenyl-(6,6)C61 (PCBM).25 The parameters Jsc and Voc quickly degraded after about 200 h at 65 °C, whereas the fill factor (FF) increased up to 600 h. However, the initial quick drop in Jsc and Voc was not clearly addressed although it might be key to understanding the stability issues of polymer solar cells.
Hence, in this work, we attempt to describe the initial performance changes in P3HT:PCBM solar cells during continuous illumination by simulated solar light. To concentrate on the very early stages of illumination, we collected device data every hour, from 0 to 11 h (the longest time here). To understand the performance changes, the blend films were analyzed by optical absorption spectroscopy, X-ray photoelectron spectroscopy (XPS), and phase-contrast atomic force microscopy (AFM) before and after illumination (11 h).
Results and Discussion
P3HT:PCBM solar cells with an Al top electrode (see Figure 1 a) were fabricated and annealed at 140 °C for 2 h (C-1 devices).9 In a parallel experiment, some of the devices were annealed at the same temperature for 30 min (C-2 devices). The annealed devices were mounted onto a sample holder inside a nitrogen-filled glove-box, which was then sealed tightly to secure the inert environment. After placing the sample holder on the stage of a solar simulator at room temperature, current density–voltage (J–V) curves were collected under continuous illumination at intervals of 1 h. The incident light intensity (Pin) was fixed to 85 mW cm−2 for C-1 devices and 100 mW cm−2 for C-2 devices. Considering the inert environment inside the sample holder, it was not expected that P3HT and/or PCBM molecules would undergo oxidative degradation during the optical excitation processes following solar light illumination (Figure 1 b),26 and we thus excluded oxidative degradation of the active layer in this study. This was confirmed by XPS investigation of the samples, which showed no characteristic peaks related to oxidation after illumination for 11 h (data not shown).
As shown in Figure 2 a, the light J–V curves of C-1 devices before and after 11 h of exposure were almost identical. Interestingly, the enlarged J–V curves plotted on a semi-logarithmic scale showed a slight difference between the two curves: after 11 h of exposure the J–V curve shifted marginally towards a higher voltage (Figure 2 c and f). However, this very small shift could not result in a change of the Voc at the present measurement resolution of 10 mV (Table 1). In addition, the shape of the optical absorption spectrum of the P3HT:PCBM blend film after exposure for 11 h (the same condition as for the C-1 device) was very similar to that before exposure, even though the optical density at around 500 nm was increased marginally after exposure (Figure 2 e). In contrast, the J–V curves of C-2 devices showed a relatively larger change, that is, a lowered Voc, after 11 h exposure (Figure 2 b and d), which may be attributed to the imperfect setting of the blend morphology in comparison with the C-1 devices.28, 29
Table 1. Performance summary of P3HT:PCBM solar cells before and after exposure to simulated solar light (AM1.5G).
Jsc [mA cm−2]
[a] Exposure time. [b] 2 h annealing at 85 mW cm−2. [c] 30 min annealing at 100 mW cm−2.
In Figure 3, detailed trends of cell performances as a function of exposure time are given. For C-1 devices (Figure 3 a), the Jsc value initially increased as the exposure time increased up to 2 h, but then showed a gradual decay with up-and-down fluctuations. After 11 h of exposure, the Jsc value was higher than before exposure. This is in good agreement with a result reported in Ref. 25, although the temperature of the sample holder surface during measurement was ca. 38 °C in our experiment versus a measurement temperature of 65 °C in Ref. 25. Similar up-and-down fluctuations of Jsc were observed for the C-2 device (Figure 3 b) but the amplitude of the Jsc change was larger than the C-1 device.
However, for both devices the FF showed a gradual decaying trend with the exposure time (data fluctuations were still observed), as was supported by slightly reduced shunt resistance (Rsh) values after 11 h exposure (Table 1). As a consequence, the continuous exposure resulted in the almost monotonic degradation of the PCE value, with occasional jumps owing to the fluctuations of other parameters. Here, for both C-1 and C-2 devices, we can find an interesting trend in the series resistance (Rs), which gradually increased as the exposure time increased. This indicates that a resistive component affecting charge transport and/or collection might be generated inside the device during continuous illumination. Considering the maintained JSC value of the C-1 device even after exposure, the resistive component generated upon illumination might be not related to the overall charge-generation yield. This implies that for the (optimized) C-1 devices the bulk heterojunction inside the blend film was not noticeably affected by this short-time exposure (11 h). Therefore, we can assign the Rs increase to deformation at the interface between the blend film and the electrodes (bottom and/or top). For the C-2 devices, the same consideration can be applied in the presence of additional changes in the morphology inside the blend film because of the unoptimized device annealing time (only 30 min). It is worth noting that small amounts of oxygen and moisture might affect the initial decay of both devices, but their influence is considered to be minimal because of their extremely low levels (<0.1 ppm).
Hereafter we focus on the optimized C-1 devices for further investigations to understand the performance changes in relationship to the exposure time. We could not investigate the interface between the blend film and the hole-collecting buffer layer [poly(3,4-ethylenedioxythiophene):poly(styrenesulfonate) (PEDOT:PSS)] without causing damage to the film and/or interface. Therefore, we tried to examine the changes in the surface of P3HT:PCBM blend film that contacted the Al (top) electrode in the device. As shown in Figure 4, the photoelectron yield of the blend film was slightly increased after 11 h of exposure for photon energies >4.7 eV. In particular, the increment became larger with increasing photon energy, indicating that zone 2 is more pronounced than zone 1. Considering that the zone 2 corresponds to photoelectrons from PCBM components,30 the increased photoelectron yield in the zone 2 can be attributed to a marginal change of the surface composition of the blend film under illumination. In other words, after 11 h of exposure the fraction of PCBM at the surface of the blend film might be increased marginally, as observed in the case of over-annealing.28, 31
To understand the change in the surface composition of the blend film (annealed at 140 °C for 2 h) after 11 h exposure, AFM measurements were performed on almost the same regions of the blend film surface: the measured surface of the blend film sample was marked before exposure, and the same part was then measured again after exposure. As shown in Figure 5, we attempted to measure the exact same region but the position was slightly shifted (see positions A and B in Figure 5 c and d). Interestingly, after 11 h of exposure, the phase-mode images revealed that the composition of the film surface was slightly changed: the bright reddish spots were formed after the exposure. However, the surface roughness changed very little, from 2.02 nm to 2.06 nm (only 0.04 nm), although the shape of image was changed slightly [note that the dark spot A became slightly bigger after exposure (spot B)]. This result supports the notion that continuous illumination with simulated solar light increases the temperature inside the sample holder (sealed tightly), leading to a minimal phase transition at the surface of the blend film. In the case of metastable domains (phases) caused inevitably by thermal annealing, as often observed from the P3HT:PCBM blend films,28, 29, 31 their phase transition could be easily accelerated by even small (thermal) impacts. Hence, we consider that the surface of the P3HT:PCBM blend film might be affected by heating effects during 11 h exposure. However, this result is only one part of the reasons that are related to the initial performance decay in the P3HT:PCBM solar cells, because the actual interface between the blend film and the Al electrode could not be measured with the presently available nanotechnology.
The initial performance changes of P3HT:PCBM solar cells were investigated by short-time exposure (11 h) with simulated solar light. The Jsc values of C-1 devices annealed at 140 °C for 2 h (optimum annealing conditions) quickly increased in the first 1–2 h of illumination, but gradually decreased slightly as the exposure time increased further (a similar but relatively larger Jsc fluctuation was also observed for the unoptimized C-2 device). However, for both C-1 and C-2 devices the FF value decreased almost linearly (showing oscillations). The Voc value was almost unchanged for the C-1 devices although the light J–V curve was very slightly shifted towards a higher voltage, while it was noticeably changed for the C-2 device. As a result, the PCE value followed a similar decreasing trend with the exposure time. This initial performance change, in the case of C-1 devices, was assigned to the formation of a resistive component when it comes to the linearly decreased series resistance. Photoelectron spectroscopy and AFM measurements suggested that a resistive component could be formed by a change in surface composition of the blend film (annealed at 140 °C for 2 h) during illumination, even though the surface roughness of the blend film was changed only very marginally after exposure. Finally, we note that the results of the present investigation may highlight only one of various reasons affecting the initial decay in the device performance, so that further (in situ) analyses focusing on the actual interface between the blend film and the electrodes (either Al or PEDOT:PSS) are required. In addition, the initial performance change could be different if the geometry of devices is changed, as reported recently.32, 33
P3HT and PCBM were used as-received from Merck Chemicals (Southampton, UK) and Nano-C, respectively. Blend solutions of P3HT and PCBM were prepared using chlorobenzene as a solvent, and stirred vigorously before use. The weight ratio of P3HT/PCBM was 1:1, while the solid concentration was ca. 60 mg mL−1. Indium tin oxide (ITO)-coated glass substrates were scribed and patterned using a photolithography technique, followed by wet (acetone and isopropyl alcohol) and dry (UV-ozone) cleaning processes. To fabricate polymer solar cells, the PEDOT:PSS layer was spin-coated on the ITO-glass substrates, followed by thermal annealing at 230 °C for 15 min. Next, on top of these PEDOT:PSS-coated substrates and/or quartz substrates (for optical absorption measurements), the P3HT:PCBM blend films (ca. 175 nm) were spin-coated and soft-baked at 50 °C for 15 min. One set of these samples was stored inside a nitrogen-filled glove box, while another set was loaded into a vacuum chamber system for the deposition of the Al top electrode. The fabricated devices were then subjected to thermal annealing at 140 °C for 2 h (C-1 devices) and 30 min (C-2 devices) inside the same glove box. Solar cell performance was measured using a home-built system equipped with a solar simulator (with intensity feedback controller, Newport Oriel) and an electrometer (Keithley 2400). The calibrated Pin was controlled to be 85 mW cm−2 for C-1 devices and 100 mW cm−2 for C-2 devices (AM1.5G). (Note: the different Pin densities are due to an upgrade of the measurement system after measuring the C-1 devices.) For measuring the device performance with exposure time, blend films or devices were mounted onto a sample holder that was filled with nitrogen and sealed tightly to avoid any penetration of ambient air. The waveguide effect of unmasked pixels was not corrected for the PCE calculation. The surface temperature of the sample holder (quartz windows) was measured using a portable IR thermometer. The optical absorption spectra of blend films coated onto either the PEDOT:PSS-coated substrates or quartz substrates were measured using a UV-Vis spectrophotometer (Optizen 2120, MECASYS), while the photoelectron spectra were measured using a photoelectron spectrometer (AC-2, Rikken Keikki). The surface morphology of blend films was measured using a multimode scanning probe microscope (Nanoscope IIIa, Digital Instrument): Both height- and phase-mode runs were performed. In order to improve the accuracy of the optical and surface measurements before and after exposure, the same blend film sample was used.
The authors thank Merck Chemicals for supplying P3HT materials and Prof. D. D. C. Bradley, Prof. J. Nelson, Prof. J. R. Durrant, and Prof. I. McCulloch for their help and valuable discussion at the early stages of this work. This work was supported by Korean government grants (Priority Research Center Program: 2009–0093819, Pioneer Research Center Program: 2009–0082820, NRF:R01–2007–000–10836–0, NRF:20090072777, KETEP-2008-N-PV08 J-01–30202008) and partly by British Petroleum International through the OSCER project (Y.K.).