With rapid increase in the efficiencies of polymer solar cells (PSCs) in the last few years, the issue of device stability is taking center stage in organic photovoltaic research. In this work, the effects of oxygen and light on the degradation of charge-transport properties of the bulk polymer active layer are studied over short timescales. It is shown that although different processing techniques produce similar efficiencies for pristine devices, they result in different degradation rates. This variation in degradation rates is primarily due to slightly different morphology parameters, such as molecular packing or disorder in the film. Investigation reveals that the choice of processing for the devices should consider degradation rates as a critical parameter, not just the efficiencies of the pristine devices. It was found that degradation starts with broadening of the effective density of states due to photo-oxidation. Both transient absorption and charge extraction by linearly increasing voltage (CELIV) measurements show increase in disorder in the films with progressive degradation. It is suggested that annealing provides the necessary thermal energy to reduce the trap states by flattening out the energy landscape of the pristine films, improving not only the efficiency, as reported previously, but also slowing the degradation rates.
Due to dwindling sources of organic fuels and the potential opportunity to harvest some of the 84TW of solar power received by planet earth each day, solar cells are at the center of attention as potential candidates to solve the world’s energy crisis. The high cost of inorganic solar cells in spite of their high power conversion efficiency (PCE) has been a major roadblock in their widespread acceptance. This is where the promise of low-cost, light-weight, flexible plastic solar cells comes into the picture. Organic solar photovoltaics (OPVs) have come a long way since the discovery of photo-induced electron transfer.1 Today, state-of-the-art OPV devices have achieved 8%2 (AM1.5G solar spectrum) PCE. As the efficiencies continue to increase, it is time to focus on a critical challenge faced by OPVs: stability.
From a fundamental and technological point of view, degradation is inevitable when subjecting an organic material to strong sunlight (UV-vis), elevated temperatures, electric currents, a reactive metal electrode, oxygen, and moisture. Researchers have worked around these problems (e.g., by extensive passivation), but a fundamental understanding of the decay processes is still much needed. Understanding the mechanisms would allow one to make better polymer solar cells with extended lifetimes.
Jorgensen et al.3 have recently summarized the types of possible degradation routes and methods for studying and elucidating degradation. Oxygen and moisture have been shown to be the main problems for the stability of polymers. Molecular oxygen physically adsorbed to organic semiconductors (accelerated by light) acts as a p-type dopant, increasing charge-carrier concentration and affecting carrier mobility.4–6 Oxygen-doping by exposure to air has been shown to increase hole mobility in poly(3-hexylthiophene) (P3HT) by nearly two orders of magnitude, in contrast to inorganic semiconductors, in which doping generally decreases mobility due to increased carrier-impurity scattering. Oxygen exposure has reportedly increased the photovoltaic performance of devices having active layers formed from other types of organic semiconductors.7, 8 Other reports measure decreased P3HT hole mobility upon exposure to air or high oxygen pressures.4, 6 So far very little is known about the nature of oxygen doping, and the well-established donor/acceptor doping theory of inorganic semiconductors is far from applicable. Additionally, its effect on the electronic properties of the active layers is not well understood.
Degradation in OPVs can primarily occur in three places: 1) in the bulk of the active polymer/organic layers; 2) at the interface of the polymer and the electrode;9, 10 and 3) at the electrode itself. In this paper, we focus on degradation of the bulk photoactive layer under solar illumination. We demonstrate how the starting morphology is critical in determining the rate of degradation of photovoltaic performance, an un-emphasized parameter hitherto.
Morphology has been widely reported as a critical factor for improved OPV device efficiency.11, 12 In particular, polymer regioregularity and film molecular packing are believed to play a crucial role for both the optical and electrical properties of bulk heterojunction films,13 such as the P3HT/methano-fullerene [6,6]-phenyl C61 butyric acid methyl ester (PCBM) blend utilized in this study. From a stability point of view the instability of the morphology (vis-à-vis phase segregation) has been shown to be of some concern. Previous studies on the morphology dependence of degradation mechanisms has revealed the formation of islands of higher efficiency, a characteristic feature of photoinduced and dark-cell degradation. Kroon et al. found a stronger decrease in short-circuit current density (Jsc)and fill factor (FF) by increasing temperature under dark conditions.14 They found that physical changes in the morphology, i.e., phase separation of the donor and acceptor domains in the blend, led to less efficient pathways for charge transport. A combination of thermal stress and continuous illumination of the devices with visible light accelerates their degradation. Yang et al. found that the thermal stability of the morphology was limited.15 Degradation phenomena were especially pronounced at temperatures above the glass-transition temperature (Tg) of the polymer material. Hence, materials with high glass-transition temperatures have been used, resulting in a significant improvement of the thermal stability of the photovoltaic parameters.16 Transmission electron microscopy (TEM) studies have revealed that such improved thermal stability coincides with a more stable active layer morphology. The improvements are attributed to the reduced diffusion movement of the electron acceptor PCBM molecules within the active layer. This results in the fact that the overall PCE, being proportional to Jsc and FF, is much more stable for high Tg polymers. As long as the solar cells are kept below the Tg of the polymer, the matrix is stiff and gives the PCBM molecules less possibility to move freely. When a thermal treatment is applied above Tg, it causes the matrix to become soft, making it easy for PCBM molecules to aggregate.
Work by the Dickson group 17 on substituted poly(phenylene vinylene) (PPV) derivatives indicates that allowing tight folding of the emissive polymers can dramatically suppress photo-bleaching. Tolbert et al. demonstrated that for PPV, confining polymer chains in nanoscale pore spaces leads to significantly slower degradation rates upon irradiation in air. It follows that the initial crystallanity and molecular packing and order have significant effects on the rates of degradation. Different processing techniques can yield active layer films with varying degrees of order. In this work, we correlate the effect of pristine film morphology on the rate of degradation of active polymer films in terms of photovoltaic performance. We study P3HT:PCBM as an example system and show that processing condition of the film can critically affect the degradation rates for a given material system.
Various approaches have been demonstrated for achieving high efficiency OPVs, which include: slow growth/solvent anneal,18 mixed solvent19 or minor solvent additives,20 and thermal annealing.21, 22 All these methods give comparable PCE for P3HT:PCBM bulk heterojunctions (BHJs).23, 24 Here, we systematically investigated devices made from four typical (widely used) methods (see Experimental Section) and study the active layer stability for OPVs. Since the electrode is more active than the bulk active layer, to separate the degradation of the electrode from the bulk film, we used a different approach from that used traditionally. The conventional method to measure degradation in device performance is to make a completed solar cell and then subject it to different external environments. However, this method means the whole device starts degrading; the bulk and electrode degradation happen simultaneously, making it difficult to identify and separate the effects. The device data obtained from such experiments is primarily overshadowed by short-term electrode degradation. Further, the evaporated metal contact covering the photo-active layers in the device can reduce photooxidation to some extent by seving as an encapsulant. Here, we subject only the photoactive spin-coated layers to degradation under ambient air and 1 sun conditions (for details see the Experimental Section) and then deposit electrodes just before measuring device performance. We show that in spite of producing comparable initial efficiencies, the rate of degradation will be different. We also show that during the first few hours of degradation, there is little change in the polymer backbone or structure. At such an early stage of degradation, before such drastic changes take place, there is a significant increase in the energetic disorder in the films leading to broadening of the effective density of states. There is a gradual loss in local order due to introduction of deeper trap states, leading to decreased mobility.
Polymer solar cells were fabricated using four different solvents: a) dichlorobenzene involving a slow growth method (referred to as DCB films/devices in text); b) chlorobenzene involving thermal annealing (referred as CB); c) chloroform, using thermal annealing (referred as CHCl3 films/devices); and d) 1,8-octanedithiol (OT) using a solvent-additive approach24 (referred as OT films/devices). For details regarding device fabrication refer to the Experimental Section. All devices showed a comparable PCE of approximately 3.5–3.8%, except for devices made from OT, which showed slightly lower PCE values of 3%, similar to published values.20 For degradation, the spin-coated active layer was subjected to 1 sun AM1.5 G solar illumination at 27 °C. After degradation, Ca/Al electrodes were deposited to complete the device for further electrical characterization. Basic morphological characterization (UV-vis, atomic force microscopy (AFM), photoluminescence) was carried out on all films. Being optimized high-performance films, they showed results that agreed well with published data.21, 23, 24 All the films show similar packing with comparable d-spacing; 15.6 Å (OT), 16.4 Å (thermally annealed), and 16.1 Å (solvent annealed). The thickness of all the films was checked to ensure that such external variations did not mask our results. The thicknesses were found to be comparable between 150 nm and 220 nm.
2.1. Device Degradation
Figure1a shows the change in normalized (with respect to a pristine non-degraded device) PCE as a function of illumination time for P3HT:PCBM films. These measurements were averaged over 12 devices for each case. The rate of degradation is almost the same for all the polymer films with different morphologies, except for the OT films. OT films showed a much faster reduction in PCE when subjected to the degrading environment. From Figure2 it can be seen that this faster degradation of OT films is not restricted to one particular OPV figure of merit. In fact, all parameters including open-circuit voltage (Voc), Jsc, FF, and series resistance (Rs) of the devices made using OT degrade at a much faster rate than other films. This is explicitly visible in Figure 1b, which shows the external quantum efficiency (EQE) of all the devices after 3 h of solar illumination. (The EQE profile for pristine devices, not shown here, showed a similar shape, with ∼60% peak value). Due to the faster degradation rates of the OT films, they showed negligible EQE after 3 h, indicating no effective charge extraction from these devices. For such short times of exposure there is no change in the UV-vis absorption spectra (data not shown) for the fresh and degraded films. This indicates that on such short timescales no significant photo-bleaching takes place. Additionally, the shoulder peaks and the spectrum shape were well preserved after degradation, indicating that no significant changes in the film morphology or crystallinity occurred. Similarly, no changes were observed in the AFM images of the films during degradation. We monitored the change of the photoluminescence (PL) peak value as a function of degradation time in Figure 1c to study the charge-transfer efficiency for excitons reaching the donor–acceptor (D-A) interface. A significant drop was seen in the PL (P3HT–O2 charge transfer complex formation).4 However, the PL before and after degradation was still of the same magnitude. The rate of decrease of PL was similar for all films. This indicates that no drastic reduction in charge-separation efficiency took place during the photo-oxidation of the bulk film. It has been shown that oxidation degradation yields C=O (1710, 1685 cm−1), C=S=O (1253, 1226 cm−1), and C–O (1154 cm−1) moieties 25 in P3HT upon light illumination. These undesirable polar groups probably act as exciton quenchers and reduce the emission intensity.
This suggests that on the timescale studied no significant changes took place in the active layer morphology or chemical structure. Thus, this initial degradation in the photoactive layer should be the result of other processes. Also, these observations do not explain the faster degradation of OT films. To understand in detail the negative impact of oxygen on the charge transport and recombination properties of the film and the underlying photophysics of degradation in detail transient absorption spectroscopy (TAS) and CELIV were used.
2.2. Charge-Transport Measurements
TAS of films provides an excellent method for investigating recombination dynamics in the active layer of photovoltaic devices:26 films exhibit virtually the same recombination kinetics that complete devices do, giving us a tool to study the degraded films directly without the added complexity of considering the degradation of the electrode. Hence, TAS was used to examine the P3HT polaron photogeneration yields and lifetimes as a result of the degradation process. The yield of dissociated polarons can be estimated approximately from the amplitude of the TAS signal at 1 µs.27 The TAS signal, i.e., ΔOD (change in optical density) evolution on a timescale of microseconds, of the degraded devices was normalized to that of a fresh film as shown in Figure3a. The increase in TAS signal suggests a larger polaron yield after possible photo-oxidation. During the first few hours, when the bulk degradation takes place, there is a multi-fold increase in the number of polarons. This increase in polaron concentration is much higher for OT films as compared to other films. As we do not see any change in the ground state absorption of the films, it is safe to conclude that this increase is not a manifestation of increased absorption from the 510 nm pump beam. We also assume that there is no change in polaron optical cross-section, since no change in morphology is expected in such short times. Enhanced exciton quenching upon degradation can be observed from Figure 1c, however, this does not explain the relaxation in number of polarons seen after approximately 3 h.
It is known that doping of disordered organic semiconductors by charged moieties has two opposing effects: increasing the concentration of charged carriers, and increasing energetic disorder. This suggests that oxidation in OT films is much more severe than for other films, which would explain the faster degradation of OT devices as compared to other devices. Any increase in energetic disorder should result in a change in recombination kinetics. This change would suppress the carrier hopping rate and therefore the mobility. Low carrier mobility due to traps would enhance recombination losses (discussed in more detail later) and increase Rs, accounting for the faster fall in FF for OT devices, as shown in Figure 2.
Polaron decay was measured at 980 nm. It obeyed the power law over the entire microsecond–millisecond time domain and fitted well with a single-power-law component, α. It has been suggested that α is sensitive to film packing and order. Values ranging from 0.25 to 0.65 have been reported, depending on processing conditions. Figure 3b shows α values obtained from fresh and degraded films. α would be expected to equal one for ideal bimolecular recombination. The sub-unity value of α has been discussed previously in terms of polaron trapping in a distribution of energetic traps in polymers.28 The deviation in values away from one indicates progressively more dispersive (trap-dependent) bimolecular charge recombination. It can be observed that degradation causes the slope of α to decrease; indicating that the energetic tail of trap states is longer after degradation. That means increasing degradation converts an increasingly large number of shallow sites into deep traps, broadening the trap distribution within the bandgap. Therefore, smaller values of α indicate that more or deeper trap sites are created on degradation.
Further support for the increasing number of traps in the films with progressive degradation can be found in the dispersive nature of CELIV transients. Disorder model simulations have demonstrated that energetic disorder can manifest itself as broadening of time-of-flight transients.29 Dispersion in a CELIV transient can be characterized by the half-width of the extraction current to time Tmax as T1/2/Tmax. 30 The theoretically calculated value of a nondispersive CELIV transient current is T1/2/Tmax = 1.2. This empirical parameter, which describes the shape of the transient, showing how fast it rises and decays after reaching its maximum value, has been determined for the CELIV transients and is shown as an inset of Figure 3b for CB films (other devices show similar trends). With continued degradation the dispersive nature of the films increases due to the increased presence of trap states in the polymer film. The influence of traps on charge transport depends on their energy relative to the transport states of the material. Photo-induced doping (the small amount of oxygen acting as a more easily ionized molecule) will add energy states to the low-energy tail of the distribution. Since the site energies of the minority dopant overlap the majority DOS (producing displaced, but overlapping, DOS distributions according to the disorder model), there is no drastic change in energy barrier for a carrier leaving the minority site. Therefore, we observe a gradual change in the dispersive nature of the transport, reflecting a smooth increase in the width of the DOS.
Experimentally measured mobility (Figure4) and photo-excited charge-carrier concentration (Figure5) dependences on delay time (Tdelay) also indicate the dispersive character of the transport. Creation of a deeper and longer tail of trap energy states as shown by TAS should also result in slower charge transport, resulting in reduced charge-separation yield (as postulated by Mihailetchi et al.).31 Both the increase in dispersive nature of transport and decrease in carrier mobility can be observed from the CELIV transients. As the devices progressively degrade the charge carrier mobility becomes a negative function of Tdelay. It can be seen that for OT films the changes are much more drastic (almost an order of magnitude change) as compared to other films. The change in absolute values of both the lifetime and the number of carriers show much more significant decrease for films made from OT as a result of degradation. The mobility tends to become a negative function of increasing Tdelay with progressive degradation. However, this transformation is much faster for OT films as compared to other films. With increasing traps as a result of degradation, most of the charge carriers would be trapped. Hence, the slope of extracted number of carriers as a function of Tdelay (dn/dTdelay) would show a more flat feature with increasing degradation. As expected this change in slope (dn/dTdelay) is much faster for films made for OT indicating a higher number of trap generation.
Figure6 shows the field dependence of the RT carrier mobility (extracted from CELIV transients) plotted as a function of the square root of the effective bias field for fresh and degraded CB and OT devices. The negative dependence on field for pristine devices has been explained previously using the Gaussian disorder model (GDM).32 In GDM, it is assumed that charge transport occurs by hopping through a manifold of localized states with superimposed energetic and positional disorder. Transport is an electronic process and no mass displacement is involved. The distributions of hopping site energies and distances are Gaussian and characterized by their width σ (width of the energetic disorder distribution, which arises from the fluctuations of energy levels) and Σ (width of the positional disorder distribution arising from structural or chemical defects), respectively. The variants to yield values of empirical parameters are believed to be indicative of the degree of disorder. In this framework, the mobility is given by:
(μ∞ is the high temperature limit of mobility and C is a parameter obtained from the simulations.)
Hence, mobility decreases when increasing the energetic disorder width σ, and a random distribution of molecules with a permanent dipole moment contributes significantly to this disorder. The photo-doping due to degradation will cause significant increase in disorder. This increase can be understood better when one considers contributions to the energetic disorder. Schein and Tyutnev33 have pointed out that disorder contributes to σ, a dipolar component due to the dipole moments of the dopant molecules in the neighborhood of the charged molecule, σd, a polymer dipole matrix component, σp, and a van der Waals component, σvdw, due to the polarization of neighboring molecules. The argument is made that the electrostatic interaction of the moving charge with a random distribution of dipoles (due to the dopant and the polymer matrix) adds to the local variations of the potentials resulting from van der Waals interactions. It is shown that these contributions add as squares.
Figure 6c and Figure 6d show that the free carrier density increases with increasing field. This is because the applied field decreases the Coulomb well (trap level) binding the incipient carrier to its counterion. However, with increasing degradation, an increased number of deeper trap states will inhibit carrier extraction. The increase in the Coulomb well barrier not only decreases the absolute value of the available free charge carriers but also makes the rate of extraction of carriers (dn/dE) with increasing field smaller and smaller. Here, the electrostatic fluctuations caused by the great majority of charges which remain bound with progressive degradation also need to be considered. These fluctuations will decrease the carrier mobility by attracting mobile carriers into potential wells. This is manifested as the decrease in carrier mobility for a given electric field. The Coulomb interaction with localized dopants raises the effective density of states in the deep tail of the DOS distribution and produces additional deep traps (as can be verified by TAS data) and suppresses the jump rate and hence reduces the mobility.
It should be pointed out that from the TAS data it can be seen that the number of carriers initially increases with degradation time, while the CELIV data shows the number of carriers as decreasing with degradation. These differences arise from the different nature of the charge carriers probed by the two techniques: TAS is sensitive to optically active carriers, while photo-CELIV detects mobile charges able to contribute to an electrical current. (However, the timescale investigated by photo-CELIV does correspond to the microsecond–millisecond decay phase of TAS.) It is important to understand that doping by oxygen produces both charge carriers and deep Coulomb traps. If the activation energy of Coulombic traps is larger than the states that control the mobility in the pristine material, they will trap the majority of the extra carriers produced and the mobility in doped films will be smaller than the carrier mobility in the undoped ones. Most of the extra carriers will be localized within the Coulomb potential wells of ionized dopants and in the deep tail states below the energy levels controlling mobility.
Next, we discuss some of the possible sources of charged defects in degraded π-conjugated RR-P3HT, which was also pointed out previously by Gregg.34 Zhuo et. al. 35 have shown using Fourier-transform (FT)IR spectroscopy that in rr-P3HT O2 interacts with the thiophene core to cause local perturbation and re-orientation of the rings, with possible dilation of the local π–π distance. These chemical photodoped impurities may be charged. The differences between the different P3HT films studied here are more fundamental, morphological defects. They were classified previously as either noncovalent defects, such as stacking faults, in which only low-energy bonds are perturbed, or covalent defects, in which high-energy covalent bonds are distorted.36 The local perturbation and re-orientation as identified by Zhou et. al. can induce considerable mechanical stresses in the conjugated backbones. Any such distortion of an sp2 carbon from its equilibrium planar trigonal geometry can introduce electronic states in the bandgap.37 Due to their non-optimized morphologies, films made from OT start with a much higher concentration of such faults and defects, which makes the diffusion of oxygen and H2O easier producing many rapid changes in transport properties on degradation. It has been shown that annealing provides the necessary thermal energy to reduce the trap states by evening out the energy landscape of the films. Most likely the lack of any thermal annealing treatments on the OT films results in a higher degradation rate. This would also explain why the other devices (all of which have undergone thermal annealing) show very similar rates of degradation. Further, it was noted that if OT films are annealed, the rate of degradation is similar to that observed for other films. This is probably because once the annealing treatment is applied, they are not much different from the other films in terms of morphology.
While the scope of this paper is limited to the P3HT:PCBM system and comparing the degradation rates of morphologies produced using the solvents studied, it has far-reaching consequences, since very high efficiency polymer photovoltaics have been reported recently using the solvent additive fabrication approach.2, 20 As a general rule, it is probably true that annealing of films would give rise to more stable and defect-free morphologies. However, it would be dangerous to make the generalization that solvent addition leads to morphologically less-stable devices for all polymer systems. Carrier transport in organic semiconductors is too vast a topic to be adequately covered here, however it is worth pointing out that whatever the processing technique, the depth and number of initial defects will play a significant role in determining the interaction of oxygen with such disordered systems, and hence, ultimately, affect the rates of degradation. This does explain earlier conflicting reports on the effect of oxygen on charge mobility. Small changes in crystal packing can significantly affect the degradation rates of different active layers. The fact that the physical/chemical nature of these defects is not completely understood makes this topic all the more challenging. An understanding of the nature of this disorder itself (which is lacking in OPVs) would help develop a better understanding of how processing conditions and the produced defects affect the degradation rates.
In summary, we have shown the effect of degradation on charge transport of P3HT:PCBM-based polymeric solar cells. The contribution of the interfaces and electrodes was eliminated to elucidate the degradation of the active layer. The initial stage of degradation starts from a broadening of the effective DOS distribution due to the Coulomb interaction of the defects with carriers localized in randomly distributed intrinsic hopping sites. It has been shown that although morphologically similar, films processed using different processing techniques might have significant difference in degradation rates. In the present study it was found that oxidation degradation in OT films is much more severe than for other films. It was shown that films made from OT start with a higher concentration of such faults and defects, which makes them more susceptible to degradation, producing many rapid changes in transport properties on degradation. Increase in energetic disorder suppresses the carrier hopping rate and therefore the mobility more for OT films as compared to others. Low carrier mobility due to traps enhances recombination losses and increases Rs, which accounts for the faster fall in FF (and PCE) for OT devices.
This study brings out the importance of studying the effect of initial film morphology on the degradation rates of polymers. Morphology-determining parameters, such as polymer regioregularity, molecular packing and order should have significant effect on the degradation rates of the active layer. This is all the more important considering the fact that recently there have been concentrated efforts to go beyond spin-coating to achieve large-area devices. These include but are not limited to inkjet printing,38 doctor blading, gravure,39 slot-die coating,40 flexographic printing,41 and other roll-to-roll processing techniques. Independent of the efficiency considerations, it is important to look into the degradation rate of films produced using these different techniques.
5. Experimental Section
Solar Cell Fabrication and Degradation: RR-P3HT (purchased from Rieke Metals and used as received) and PCBM (purchased from Nano-C and used as received) were used to make the active layers. Four different kinds of processing techniques, published previously, were used to fabricate P3HT: PCBM active layer. For device fabrication, a poly(ethylenedioxythiophene):polystyrene sulfonate (PEDOT:PSS) (Baytron P VP Al 4083) layer was coated on precleaned UV-ozone treated ITO substrates. The active layer was then deposited using the four methods given below. The device was then subjected to AM1.5G 1sun environment for various lengths of time. The temperature of the room was 22 °C and relative humidity was between 40 and 60%. After fixed times, the films were taken for TAS measurements. For efficiency and CELIV measurements, a bilayer cathode consisting of a calcium layer (20 nm) and an aluminum layer (100 nm) was deposited by thermal evaporation under vacuum (1 × 10−6 Torr). The current–voltage (I–V) measurements were performed under a nitrogen atmosphere using a Keithley 2400 source meter unit. The photocurrent was measured under an AM 1.5G illumination at 100 mW cm−2 using an Oriel 96000 150 W solar simulator source. The light intensity was determined using a silicon detector (with KG-5 visible color filter) calibrated by the National Renewable Energy Laboratory (NREL, Golden, CO). The details of the active layer fabrication are as follows:
1.Films made using OT: P3HT and PCBM with 1:1 weight ratio (20 mg/mL for each) were first dissolved in chlorobenzene (CB; purchased from Sigma Aldrich and used as received) and well mixed. 7.5 μL OT (purchased from Sigma Aldrich and used as received) was added to 250 μL of CB solutions. The solutions with OT were then stirred for more than 10 min. All solutions were then spin-coated at 3000 rpm for 70 s to make the active layer. No further thermal annealing treatments were given.
2.Films made using slow growth from DCB: P3HT and PCBM with 1:1 weight ratio (20 mg/mL for each) were first dissolved in dichlorobenzene (DCB, purchased from Sigma Aldrich and used as received). Films were made by spin coating the solution at 700 rpm for 40 s following by slow drying in a covered Petri dish as detailed in elsewhere.18 After drying the films were annealed at 110 °C for 10 min in N2 glove-box.
3.Films made using CB: P3HT and PCBM with 1:1 weight ratio (20 mg/mL for each) are first dissolved in chlorobenzene (CB, purchased from Sigma Aldrich and used as received). Films were made by spin-coating the solution at 1000 rpm for 60 s following by annealing on a hotplate at 150 °C for 20 min in a N2 glove-box.
4.Films made using CHCl3: P3HT and PCBM with 1:1 weight ratio (20 mg/mL for each) were first dissolved in chloroform (purchased from Sigma Aldrich and used as received). Films were made by spin-coating the solution at 5000 rpm for 60 s following by annealing on a hotplate at 150 °C for 30 min in a N2 glove-box.
Transient-Absorption Microscopy: Micro-millisecond transient absorption of the polymer film devices was done in the transmission mode. A 510 nm dye laser (LSI DUO-220) pumped by a nitrogen laser (LSI VSL-337ND-S) was used as the excitation source. The pulse energy and pulse width were about 3 μJ/cm2 and 4 ns, respectively. The probe beam was supplied by a 980 nm laser diode. The wavelength corresponds to the absorption of P3HT+ polarons. The change in absorption of the polymer film was measured using a Thorlabs DET110 silicon detector with a 980 nm band pass filter (Thorlabs FB980–10). Collimating and focusing lenses were used to reduce stray light from entering the photodiode. The current of the photodiode was first amplified by a current amplifier (Femto DHPCA-100) and then a preamplifier (SRS SR445A) and recorded by a digital oscilloscope (Tektronix DPO 4104). Each measurement was averaged over 4096 times to improve the signal-to-noise ratio. The response time of the measurement circuit was estimated to be 350 ns. Low-intensity excitation conditions were employed to ensure that the densities of photo-excited charge carriers are comparable to those generated under solar radiation. The obtained decay was as reported earlier in literature.42 A mono-exponential decay function (ΔOD = At−α, where A is the pre-exponential function, t the time and α the power law component) was used to fit the decay.
Photo-charge Extraction by Linearly Increasing Voltage (Photo-CELIV): The experimental setup for CELIV is described elsewhere.43, 44 A 510 nm dye laser (LSI DUO-220) pumped by a nitrogen laser (LSI VSL-337ND-S) was used as the excitation source. The pulse energy and pulse width were about 3 μJ/cm2 and 4 ns, respectively. A variable pulse generator Wavetek Datron 195 and a memory oscilloscope Tektronix TDS 430A were used to recode the extraction currents. The current of the photodiode was amplified by a current amplifier (Femto DHPCA-100). Appropriate negative bias was applied to compensate for the built-in-voltage. The mobility and the electric field were calculated as by Juska et al.45
A. Kumar and Y. Yang thank Dr. Gang Li, Dr. W.-L. Kwan and Dr. Zhengqing Gan for fruitful scientific discussions.
This work is supported by the Air Force Office of Scientific Research (grant # FA9550–09-1–0610, Program manager-Dr. Charles Lee).