Tuning the Dielectric Properties of Organic Semiconductors via Salt

Sep 27, 2013 - The frequency and phase dependence of the real dielectric function .... with various salt loadings to compare with the mobility and lif...
0 downloads 0 Views 426KB Size
Download PDF
Article pubs.acs.org/JPCB

Tuning the Dielectric Properties of Organic Semiconductors via Salt Doping Xien Liu,†,∥ Kwang S. Jeong,‡,∥ Bryan P. Williams,§ Kiarash Vakhshouri,§ Changhe Guo,§ Kuo Han,† Enrique D. Gomez,§ Qing Wang,† and John B. Asbury*,‡ †

Department of Materials Science and Engineering, The Pennsylvania State University, University Park, Pennsylvania 16802, United States ‡ Department of Chemistry, The Pennsylvania State University, University Park, Pennsylvania 16802, United States § Department of Chemical Engineering and the Materials Research Institute, The Pennsylvania State University, University Park, Pennsylvania 16802, United States S Supporting Information *

ABSTRACT: Enhancing the dielectric permittivity of organic semiconductors may open new opportunities to control charge generation and recombination dynamics in organic solar cells. The potential to tune the dielectric permittivity of organic semiconductors by doping them with redox inactive salts was explored using a combination of organic synthesis, electrical characterization, and time-resolved infrared spectroscopy. The addition of the salt, LiTFSI (lithium bis(trifluoro-methylsulfonyl)imide), to a conjugated polymer specifically designed to incorporate ions into its bulk phase increased the density of holes and enhanced the static dielectric permittivity of the polymer blend by more than an order of magnitude. The frequency and phase dependence of the real dielectric function demonstrates that the increase in dielectric permittivity resulted from dipolar motion of bound ion pairs or clusters of ions. Interestingly, the increases in the hole density and dielectric permittivity were associated with enhancement of the hole mobility by 2 orders of magnitude relative to the undoped polymer. The charge recombination lifetime also increased by an order of magnitude in the blend with a fullerene electron acceptor when ions were added to the polymer. The findings indicate that ion doping enables organic semiconductors with large increases in low frequency dielectric permittivity and that these changes result in improved charge transport and suppressed charge recombination on the microsecond time scale.

■

INTRODUCTION

researchers have tuned the electrical properties of polymeric materials using approaches that modify their dielectric permittivity over a modest range. Blom and co-workers found that increasing the dielectric permittivity of a poly(p-phenylene vinylene) variant by the addition of ethylene oxide side chains resulted in enhancement of the charge separation efficiency in a blend with the fullerene variant, PCB-EH.12 Jen and co-workers explored the inclusion of dipolar side chains onto a series of low bandgap conjugated polymers in organic solar cells.13,14 Ginger and co-workers examined a series of organic and inorganic acceptors that modulated the dielectric properties of the polymer blends.15 Tajima and co-workers tuned the electrical properties of organic solar cells by incorporating perfluorinated side chains onto polythiophene and fullerene moieties that localize at donor/acceptor interfaces.16

The low dielectric constant of organic materials limits their utility as functional materials in a variety of applications. Enhancements in the permittivity are warranted in organic dielectric layers utilized in capacitors to enhance energy storage1 and in transistors to decrease the operating voltage of the gate electrode.2 The low dielectric constants of organic electrolytes limit the ability to dissociate salts, thereby decreasing ionic conductivity.3 Furthermore, organic semiconductors are unable to dissociate electron−hole pairs due to an inability to screen electrostatic forces.4 As a consequence, the low dielectric permittivity of conjugated organic molecules seems to be at odds with their function in organic photovoltaic applications5 in which excitons must be efficiently dissociated to form separated charge carriers.6 Their low dielectric permittivities and associated localization of electronic states give rise to large exciton binding energies 7,8 and significant energetic barriers to charge separation,4,9 depending on the molecular structure of the electron donating and accepting moieties.10,11 A number of © 2013 American Chemical Society

Special Issue: Michael D. Fayer Festschrift Received: August 26, 2013 Revised: September 20, 2013 Published: September 27, 2013 15866

dx.doi.org/10.1021/jp408537p | J. Phys. Chem. B 2013, 117, 15866−15874

The Journal of Physical Chemistry B

Article

Scheme 1. Overview of the Synthetic Route to M-TQ1a

a

Experimental details are provided in the Supporting Information.

In the present work, we investigated whether the dielectric properties of a semiconducting conjugated polymer can be tuned over a wide magnitude and frequency range by the addition of redox inactive ionic species into the matrix of the polymer that was decorated with PEO side chains. Furthermore, we examined the corresponding impact that salt doping has on the optical and electrical properties of films of the conjugated polymers and their corresponding blends with electron accepting fullerene molecules. Electrical characterization methods and time-resolved infrared (TRIR) spectroscopy17 are combined to demonstrate that the charge carrier density and dielectric permittivity of the conjugated polymer can be significantly enhanced by the addition of LiTFSI. The addition of the LiTFSI into the polymer phase increases the conductivity of the polymer blend film both by increasing the density of holes in the films and by increasing their mobility. The charge recombination lifetime of blends of the conjugated polymer with an electron accepting fullerene is also increased by addition of the lithium salt, causing a marked increase of the mobility-lifetime product at moderate LiTFSI loading.

■

acceptor moieties, respectively. The visible absorption spectral features and wavelengths are similar to those of the alkoxy substituted TQ1 conjugated polymer reported by Andersson and co-workers.18 Spectroscopy and Device Characterization. Numerous techniques were combined to determine the influence that ion doping has on the dielectric and charge transport properties of the conjugated polymer, M-TQ1. The techniques, sample preparation, and experimental conditions are described in detail in the Supporting Information and in a previous publication.19 Briefly, field effect thin film transistors (TFTs) were utilized to obtain the hole mobility in pure films of M-TQ1 and in lithiumdoped films of the conjugated polymer. In parallel, TRIR spectroscopy was utilized to measure charge recombination kinetics in polymer blends of M-TQ1 with PCBM at various LiTFSI loadings. Frequency and phase dependent impedance measurements with an LCR meter at room temperature were used to obtain the real and loss dielectric functions of polymer films with various salt loadings to compare with the mobility and lifetime measurements. Transmission electron microscopy (TEM) experiments were performed at the National Center for Electron Microscopy, Lawrence Berkeley National Laboratory, on a Zeiss LIBRA 200MC. Bright field images, thickness maps, and elemental maps were captured. Sulfur and carbon elemental maps were obtained through the standard threewindow method. Prior to use, LiTFSI was heated for 48 h at 150 °C in a glovebox antechamber to remove any residual water. LiTFSI was then taken directly into a N2 glovebox without exposure to ambient atmosphere and dissolved in anhydrous tetrahydrofuran (Sigma-Aldrich) for 12 h at 5 wt % concentration. Solutions of 1:3 by mass M-TQ1 and PCBM (>99.5%, Nano-C) were made with anhydrous 1,2-dichlorobenzene (Sigma-Aldrich). LiTFSI salt was added by introducing the appropriate volume of LiTFSI/THF solution. Solutions containing M-TQ1, PCBM,

EXPERIMENTAL SECTION

Synthesis. The synthesis of the oligo(ethylene oxide) substituted low-band-gap conjugated polymer (M-TQ1) via Stille coupling is outlined in Scheme 1. The detailed synthesis procedures and characterization methods are provided in the Supporting Information. The number-average molecular weight of the polymer measured by gel permeation chromatography (GPC) is 36 000 with a polydispersity index (PDI) of 1.5 using polystyrene as the standard. The polymer is readily soluble in common solvents such as tetrahydrofuran, chloroform, chlorobenzene, and o-dichlorobenzene. The absorption spectra of M-TQ1 films present two absorption bands at 360 and 620 nm attributed to π−π* transition of the conjugated polymer chain and intramolecular charge transfer between donor and 15867

dx.doi.org/10.1021/jp408537p | J. Phys. Chem. B 2013, 117, 15866−15874

The Journal of Physical Chemistry B

Article

and LiTFSI were stirred for at least 1 h at 100 °C prior to use to ensure dissolution. Photovoltaic Device Characterization. Solar cells (device area = 0.162 cm2) were fabricated on indium tin oxide (ITO) coated glass substrates (Kintec, Hong Kong). The substrates were cleaned with Aquet detergent solution and water, followed by 10 min of sonication in acetone, 10 min in isopropanol, and 10 min of UV-ozone treatment. PEDOT:PSS was spun cast in a laminar flow hood at 4000 rpm for 2 min and subsequently dried at 165 °C for 10 min (thickness ca. 70 nm). M-TQ1/PCBM (24 mg/mL, 65 ± 10 nm) active layers were spun cast in a N2 glovebox at 1000 rpm for 1 min on top of the PEDOT:PSS layer. Finally, a 75 nm layer of aluminum was deposited via thermal evaporation at 10−6 Torr. All devices were annealed at 100 °C for 20 min. Electrical characterization of photovoltaic devices in the dark and under AM 1.5G (100 mW/cm2) illumination from a 150W Newport solar simulator was performed using a Keithley 2636A Sourcemeter. At least six devices were averaged for all of the data presented here. All device fabrication and testing took place in a N2 glovebox while never exposing samples to ambient atmosphere.

dielectric function ε′(ω). The in-phase component, C(ω), is related to the loss and/or the conductive current and is used to obtain the loss component, ε″(ω). Sophisticated analyses of the real and loss dielectric functions of materials have been developed for decades.25 We briefly summarize the Havriliak−Negami model26 that is commonly used to analyze the frequency dependence of the real and loss dielectric functions.27−29 In particular, the real dielectric function is described by

RESULTS The structure of the low-band-gap conjugated polymer that was synthesized for the present work is depicted in Scheme 1. The polymer, M-TQ1, bears ethylene oxide side chains designed to chelate small cations permitting dispersal of salt ion pairs or clusters of ion pairs throughout the bulk of the polymer phase. The structure of the conjugated polymer backbone was motivated by recent results of Andersson and co-workers18 who reported organic solar cells with 6% power conversion efficiency using a similar conjugated polymer that bore octyloxy side chains. The LiTFSI salt was chosen as the ionic dopant because of its previous use in the context of solid state dye sensitized solar cells.20,21 The oxidation and reduction potentials of the salt22,23 are far outside of the corresponding potentials of common conjugated polymers such that we do not expect the salt to actively undergo oxidation or reduction chemistry in the polymer layer. LiTFSI was electrochemically inert when it was used to dope hole transport layers in solid state dye sensitized solar cells.24 Motivated by the desire to examine the potential to tune the dielectric permittivity of conjugated polymers by doping them with salts, we undertook an impedance spectroscopy study of a series of polymer films of M-TQ1 with varying LiTFSI concentrations. Films of pure M-TQ1 with varying LiTFSI concentrations were drop cast from chlorobenzene onto electrodes in a N2 glovebox to prevent absorption of water. Care was taken to avoid exposure of the samples to ambient atmosphere throughout the experiments by placing them inside sealed bags. The smallest possible pinholes were formed in the bags to permit electrical contact to the electrodes of the samples in the bags while retaining a dry atmosphere during the impedance spectroscopy measurements. In this experiment, a sinusoidal driving field, V(t) = A sin(ωt) with angular frequency, ω, is applied to a parallel plate capacitor in which M-TQ1 doped with LiTFSI to varying concentrations serves as the dielectric layer. The current associated with application of the driving field, I(t), can be divided into in-phase and out-ofphase components according to I(t) = C(ω) sin(ωt) + D(ω) cos(ωt). The out-of-phase component described by D(ω) is related to the capacitive current and is used to calculate the real

where σ0 is the conductivity of the dielectric material, ε0 is the permittivity of a vacuum, and we have not included the exponent, ni, for clarity because its value was fixed at unity. Equations 1 and 2 produce frequency dependent real dielectric and loss functions displayed in Figure 1 using

ε′(ω) = ε∞ +

∑ i

Δεi (1 + (ω 2τi 2)mi )ni

(1)

where ε∞ is the high frequency dielectric constant, Δεi describes the dielectric strengths of resonances in the material with characteristic time scales, τi, of their response to an applied electric field, and mi and ni are shape parameters that determine the width of the ith resonance. We found that we could adequately fit the experimental data using eq 1 by fixing ni to a value of unity. The corresponding loss function is ε″(ω) =

■

∑ i

Δεi(ωτi)mi 1+

(ω 2τi 2)mi

+

σ0 ε0ω

(2)

Figure 1. Model real (A) and loss (B) dielectric functions calculated from a common set of parameters using eqs 1 and 2, respectively. The model real dielectric function includes a dielectric relaxation process in the 102−103 Hz range with a corresponding absorptive loss term appearing at the same range. Conductivity appears in the loss function but not in the real dielectric function.

dielectric and time constant parameters relevant to the impedance spectroscopy study of M-TQ1. For example, an increase in the real dielectric function such as that represented in Figure 1A in the 102−103 Hz range is associated with an absorptive component in the loss function at the same frequency (Figure 1B, dashed line). Incorporation of conductivity in the sample, due, for example, to the drift of free ions in the applied electric field, does not affect the real 15868

dx.doi.org/10.1021/jp408537p | J. Phys. Chem. B 2013, 117, 15866−15874

The Journal of Physical Chemistry B

Article

films doped with various lithium ion concentrations are represented in Figure 2B. In the films with Li+/O ratios of 0.005, 0.01, and 0.015, an absorptive dielectric loss peak is visible in the 102−103 Hz range that indicates a dipolar dielectric relaxation process. The LiTFSI doped films are somewhat conductive due to the presence of ions in the polymer. The influence of the conductivity of free ions in the films is apparent as the low frequency conductive tail highlighted in Figure 2B. As indicated in eq 2, the conductivity of the film influences the dielectric loss function but does not influence the real dielectric function because the real component corresponds to the capacitive current that is outof-phase with the sinusoidal driving field in the experiment. The 90° phase shift of the in-phase conductive and out-of-phase capacitive currents enables precise discrimination of the real dielectric function from the dielectric loss of the M-TQ1 films. We sum over the amplitudes of all frequency components of the fit functions to obtain the static dielectric constants of the films because the parameters in the fits are coupled and so are not individually significant. Using this approach, we find that films doped with Li+/O ratios of 0, 0.005, 0.010, and 0.015 have static dielectric constants of 5.1 ± 0.3, 135 ± 22, 145 ± 25, and 145 ± 20, respectively. The uncertainty in the static dielectric constants arises from the limited frequency range over which the lowest frequency resonance is observed. The amplitude and frequency of this component are the least constrained parameters of the fit functions. Consequently, we take the amplitude of this component as the confidence interval of the measurements (see Table S1, Supporting Information). Within the observed frequency range, the fits to the real dielectric functions reveal that the static dielectric constants reach a plateau at a doping concentration of 0.005, and addition of LiTFSI beyond this concentration shifts the dipolar resonance to higher frequency. The shift is apparent in both the real and loss dielectric functions (Figure 2). The origin of the shift to higher frequency is unclear at this time but may be related to formation of larger ion clusters at higher concentrations or possibly to plasticization of the polymer matrix. The motion of Li+ ions may be less constrained by interaction with the ethylene oxide side chains at higher doping levels, giving rise to the faster dipolar response. Intrigued by the influence that ion doping has on the dielectric permittivity of M-TQ1, we examined the charge carrier transport and recombination dynamics of the ion-doped polymer films. Bottom-contact, bottom-gate field-effect-transistor (TFT) measurements were used to characterize the influence of ion doping on the mobility of holes in films of MTQ1 doped with various amounts of LiTFSI. Heavily doped ptype Si wafers were used as the gate electrodes with a 300 nm thick thermally grown SiO2 layer as the gate dielectric. Gold source and drain electrodes (with a thickness of ∼100 nm) were deposited by conventional double-layer lithography with channel widths (W) of 220 μm and lengths (L) of 20 μm (Figure 3A). All electrical measurements were performed in a N2 glovebox. Figure 3B displays the source-drain current, ISD, versus gate voltage, Vg, characteristics of TFTs prepared with LiTFSI concentrations such that the Li+/O ratios were 0, 0.005, 0.010, 0.015, and 0.050. The ISD for both negative (on) and positive (off) Vg values are increased upon doping with the salt except at Li+/O ratios of 0.050 and greater. The linear ISD scale represented in Figure 3C demonstrates that application of negative gate voltage significantly modulates the density of holes in the channel even in films with salt loading up to Li+/O

dielectric function but does introduce a frequency dependent conductivity term in the dielectric loss function (Figure 1B, dotted line). The sum of the absorptive and conductive loss terms gives rise to the solid curve in Figure 1B that models loss functions measured experimentally in impedance spectroscopy studies. Figure 2A displays the real dielectric functions of M-TQ1 polymer films with Li+/O ratios of 0, 0.005, 0.010, and 0.015 on

Figure 2. (A) Frequency dependence of the real dielectric function measured in films of M-TQ1 with various LiTFSI concentrations. The data indicate the dielectric constant of undoped M-TQ1 changes little with frequency in the range from 101 to 107 Hz. Fitting the data using standard dielectric relaxation models reveals that the addition of LiTFSI to the polymer layers increases the static dielectric constants of the films more than 20-fold. The inset shows the parallel plate capacitor sample geometry with the polymer film sandwiched between matched planar electrodes. (B) Frequency dependence of the dielectric loss functions of M-TQ1 films with various LiTFSI concentrations. The loss functions display signatures of absorptive loss components corresponding to the large increases of the real dielectric functions of doped polymer films in the 102−103 Hz range. The conductivity of the lithium-doped films gives rise to the conductive tails highlighted in the figure.

a semilogarithmic plot that were measured using the sandwich cell sample geometry represented in the inset. The dielectric function of undoped M-TQ1 varies little over the 101−107 Hz range represented in the figure. The dielectric functions of the LiTFSI doped M-TQ1 polymer layers were fit using eq 1. Parameters from best fits to the data are included in Table S1 (Supporting Information). The fit functions are not distinguishable from the experimental data. The fit procedure reveals that the static dielectric permittivity of pure M-TQ1 without the addition of LiTFSI is around 5.1, which is slightly greater than the relative dielectric constant of P3HT.30 This increase in dielectric permittivity probably results from the tethering of polar ethylene glycol side chains to the conjugated polymer backbone as has been observed in other systems.12 Fitting the LiTFSI doped M-TQ1 dielectric functions indicates the formation of a new dielectric relaxation process in the 102− 103 Hz range that is not present in the undoped material. To ascertain the origin of the new dielectric relaxation process, the corresponding dielectric loss curves of the M-TQ1 15869

dx.doi.org/10.1021/jp408537p | J. Phys. Chem. B 2013, 117, 15866−15874

The Journal of Physical Chemistry B

Article

In the analysis described by Sze,31 the field effect mobility depends on the change in source-drain current with gate voltage. Therefore, differential mobility analysis can be used to determine the field effect mobility of holes in the films doped at all Li+/O ratios independent of magnitudes of the “off” currents at positive gate voltage because the gate modulates the density of holes in the films and therefore the conductivity of the channels in all cases. The differential mobilities of the iondoped films at saturation (Vg = −50 V) reveal that the hole mobility increases more than 2 orders of magnitude with increasing Li+/O ratio up to 0.015. The sudden drop in conductivity of the film with Li+/O ratio of 0.05 suggests that the salt is not soluble in the polymer phase at this concentration and therefore accumulates in the channel, obstructing charge transport that occurs within only a few nanometers of the gate dielectric interface. Charge recombination kinetics of M-TQ1 blended with PCBM containing various concentrations of LiTFSI were measured using TRIR spectroscopy17 in an effort to assess the influence that ion doping has on the process. In the experiment, visible pump pulses of 10 ns duration and 532 nm wavelength were used to photoexcite the polymer blend, leading to the formation of positive polarons (holes) in the polymer phase (Figure 4A).32 Kinetics traces measured near the maximum of the polaron absorption of 1:3 (by mass) blends of M-TQ1 with PCBM are represented in the main panel of Figure 4B for polymer blend films containing Li+/O ratios of 0, 0.005, 0.010, and 0.015, respectively. The kinetics traces have been Figure 3. (A) Schematic of the thin film transistor electrode and sample geometry. (B) Source-drain current, ISD, versus gate voltage, V g, measured in thin films of M-TQ1 doped with various concentrations of LiTFSI. Both the on currents at negative gate voltage and the off currents at positive gate voltage increase markedly with lithium ion doping except at a Li+/O ratio of 0.05 where deposition of the salt in the channel disrupts current flow. (C) Plot of data from panel B on a linear ISD scale. The gate modulates ISD in the Li+/O = 0.015 film permitting analysis of the transfer curve to obtain the hole mobility. (D) Mobilities versus gate voltage calculated from the ISD−Vg curves appearing in panel B for various M-TQ1 TFT films. The mobilities near Vg = −50 V, when the differential mobilities are well-behaved, are taken as estimates of the hole mobilities for the samples. The data presented here correspond to one device. The mobilities reported in Figure 5 are averages from several devices.

= 0.015. Figure 3D represents the mobilities (μ) as a function of gate voltage of the films calculated from the ISD transfer curves versus Vg appearing in panel B. The ISD of a field effect transistor depends on the device parameters and gate voltage according to ISD =

μCgW 2L

(Vg − Vth)2

(3)

Figure 4. (A) Schematic diagram of the time-resolved infrared spectrometer and sample geometry. (B) Charge recombination kinetics traces of 1:3 by mass blends of M-TQ1 with PCBM doped with various concentrations of LiTFSI. The transient signals result from the absorption of polarons in the polymer phase measured at 1700 cm−1. The shaded region within the inset represents the frequency range within the polaron absorption at which the recombination kinetics were measured in the polymer blends. The gray curve in the main panel represents the instrument response function used to deconvolve the kinetic traces. The data indicate slower charge recombination dynamics with increasing LiTFSI concentration.

where Cg and Vth are the capacitance of the gate dielectric and the threshold voltage at which sufficient charge density accumulates in the channel to allow conduction, respectively. The field effect mobility of charges in the channel can then be calculated from the derivative of the ISD versus Vg curves according to ⎡ d(I 1/2) ⎤2 2L μ = ⎢ SD ⎥ ⎢⎣ dVg ⎥⎦ WCg

(4) 15870

dx.doi.org/10.1021/jp408537p | J. Phys. Chem. B 2013, 117, 15866−15874

The Journal of Physical Chemistry B

Article

O ratios, respectively. An exponential model was adopted to describe the longest decay components because the signals do not decay to zero within the measured time window. Therefore, the indicated time constants are considered lower bounds to the actual decays. The data indicate significant increases in the charge recombination lifetimes of the polymer blends with increasing salt concentration. The simultaneous increase in both the charge recombination lifetime and hole mobility reported here is similar to results reported by Grätzel and coworkers in their studies of solid state dye sensitized solar cells.20,21,24

normalized for ease of comparison. We examined charge recombination kinetics in the 1:3 blends in an effort to measure the recombination of electrons and holes in materials that have bearing on photovoltaic materials. Photoexcitation of films of neat M-TQ1 with no PCBM content did not result in measurable transient absorption signals on the microsecond time scale because this time scale is long in comparison to the exciton lifetime. Efficient photogeneration of electrons and holes in organic semiconductors generally requires mixing of electron donating and accepting materials.7,8,33,34 By examining blends of M-TQ1 with PCBM, long-lived electrons and holes were generated that were not generated in either pure material alone. Because the generation of electrons and holes is unique to the polymer blend, we are able to interpret the time dependence of TRIR signals measured in the blends in terms of charge recombination. The samples for these experiments were spin coated and loaded into a cryostat in a N2 glovebox to avoid exposure of the films to water and oxygen from ambient atmosphere. The inset in Figure 4B displays the polaron absorption covering the mid-infrared region32 from which the charge recombination kinetics were measured. The shaded area reflects the specific spectral region that was integrated to obtain the charge recombination kinetics displayed in the main panel of Figure 4B. The shaded region overlaps the carbonyl absorption of PCBM. Because we integrated all wavelengths around the vibrational feature to obtain the kinetics, the frequency evolution of the feature reported previously has no influence on the measured kinetics.35 Furthermore, it was demonstrated that the amplitude of the vibrational feature varies proportionally with the polaron absorption and thus does not interfere with measurement of the recombination kinetics.17 Because the polaron absorption arises from the formation of holes in the conjugated polymer (and possibly electrons in the PCBM phase), the loss of this signal can be traced to the loss of electron and hole populations in the polymer blend.17 Therefore, the decay traces of the polaron absorption signals represented in Figure 4B are measures of the charge recombination kinetics in the blends. Overlaid on the experimental data in Figure 4B are multiexponential fit functions used to quantify the time dependence of the charge recombination kinetics. The fit functions were convolved with the instrument response function (gray line near the time origin), enabling accurate description of the recombination kinetics on the 200 ns and longer time scales. Average charge recombination lifetime values were obtained by convolving multicomponent fit functions with the instrument response function and comparing to the time-resolved kinetics traces represented in Figure 4B to minimize the sum of the squares of the residuals. The best fit functions, F(t), were then integrated according to ⟨τ⟩ = ∫ t F(t) dt/∫ F(t) dt to obtain the average decay time, ⟨τ⟩. This analysis scheme was selected because we desired a method to quantify the dynamics that does not require the adoption of a particular kinetic model. Although a multicomponent fit function was used in the deconvolution procedure, the average lifetime obtained from the analysis is not sensitive to the choice of functional form as long as the fit functions describe the data accurately. The fit functions in Figure 4B are indistinguishable from the data, indicating the fidelity of the fits. From the fitting procedure, average recombination time constants of 16, 101, 119, and 133 μs were obtained for 1:3 MTQ1:PCBM blends containing 0, 0.005, 0.010, and 0.015 Li+/

■

DISCUSSION In the following, we discuss the observations reported above in an effort to understand the influence that lithium ion doping has on the real dielectric function, charge carrier density, transport, and recombination properties of the M-TQ1/PCBM polymer blends. Figure 5 compiles the variation of the hole

Figure 5. Summary plot of the increase in hole mobility (right axis) and charge recombination lifetime (left axis) as a function of lithium ion concentration. These increases are associated with a more than 20fold enhancement of the static dielectric constant of the Li-doped polymer, suggesting that ion doping provides a means to improve charge transport in organic semiconductors.

mobility (right axis) and charge recombination lifetime (left axis) as a function of lithium ion concentration. The values were extracted from the data represented in Figures 3D and 4B, respectively. Analysis of mobility versus gate voltage was used to extract the hole mobility from the ISD versus Vg curves. From this analysis applied to multiple devices, the mobility varied from just under 10−5 cm2/(V s) to about 10−3 cm2/(V s) as the Li+/O ratio was increased from 0 to 0.015. A similar increase in hole mobility with ion doping was reported by Snaith and coworkers in spiro-MeOTAD films.21 There is also a pronounced increase in the “off” current at positive gate voltage that exceeds the increase in hole mobility with lithium-ion doping. This observation suggests that both the mobility and density of holes in the ion-doped M-TQ1 layer were increased by the addition of LiTFSI. Other investigators have observed the influence of doping on the free carrier density of organic semiconductors. For example, Kahn and co-workers observed an increase of the charge carrier density in a variety of organic semiconductors when they were doped with a 19-electron transition metal complex.36,37 The complex, Rh(Cp)2 where Cp = cyclopentadienyl, introduced excess electrons into the organic semiconductor resulting in 6 orders of magnitude increase in the conductivity of TIPSpentacene and other organic semiconductors when the films were doped at 2% by weight. Importantly, the investigators observed a decrease in the activation energy for charge transport in the organic semiconductors with increased doping 15871

dx.doi.org/10.1021/jp408537p | J. Phys. Chem. B 2013, 117, 15866−15874

The Journal of Physical Chemistry B

Article

Figure 6. (A) Diagram of test structures used to evaluate the photovoltaic properties of M-TQ1:PCBM polymer blends. Summaries of photovoltaic device characteristics as a function of salt concentration, r (Li/EO), in the active layer are represented: (B) power conversion efficiency; (C) shortcircuit current (JSC); (D) open circuit voltage (VOC); (E) fill factor. The error bars also denote the standard deviation from multiple measurements. The lines connecting the data are guides to the eye.

function of the lithium-doped M-TQ1 films in the 102−103 Hz range arises from a dielectric relaxation process not present in undoped M-TQ1 films. Analysis of the phase and frequency dependence of the impedance spectroscopy data reveals that the increases of the real dielectric functions of lithium-doped films do not result from conductivity of the films. The influence of conductivity appears in the corresponding dielectric loss functions in Figure 2B. Instead, the dielectric relaxation process present in the lithium-doped M-TQ1 films results from bound ion pairs or clusters of ion pairs reorienting in response to the applied electric field, giving rise to a macroscopic dipole. It is important to be clear that orientational motion of ion pairs does not result from conduction of free ions in the films. The conduction of free ions gives rise to current that is in-phase with the driving field and contributes to conductive loss in the film (see eq 2 and Figure 1B). Orientational motion of ion pairs or clusters is one of many dipolar response processes that determine the frequency dependence of the real dielectric function of a material.25 The increase of hole mobility with increasing LiTFSI concentration up to R = 0.015 indicates that the addition of the salt to M-TQ1 does not create deep hole traps in the polymer. We investigated whether addition of LiTFSI might create deep electron traps by examining the electrical properties

of the films. The decrease in activation energy was traced to filling of deep midgap trap states by electrons donated by the transition metal complexes. Conduction in the doped films therefore occurred in states closer to the mobility edge, giving rise to lower activation energy for charge transport. We believe a similar mechanism gives rise to the increase in hole mobility and hole density with lithium ion doping of the M-TQ1 films reported here. An interesting difference between the present study and that of Kahn and co-workers is that the LiTFSI salt is not expected to be oxidized or reduced by M-TQ1, unlike 19-electron transition metal complexes. Gregg and co-workers examined ntype doping of organic semiconductors by the addition of redox-inactive charged species and found that the doping density increased with greater concentrations of ionic species.38,39 The investigators observed a superlinear increase of the conductivity with doping density that they rationalized in terms of an increase in the free carrier density as a result of enhanced dielectric permittivity of the organic semiconductor.38 The conclusion by Gregg and co-workers that doping organic semiconductors with redox-inactive charged species can increase their dielectric permittivity has direct bearing on the present study. In particular, the increase of the real dielectric 15872

dx.doi.org/10.1021/jp408537p | J. Phys. Chem. B 2013, 117, 15866−15874

The Journal of Physical Chemistry B

Article

M-TQ1 and PCBM-rich domains suggests that the short-circuit current of the photovoltaic test structures may be limited by exciton splitting to form charge carriers rather than by transport of carriers to the electrodes. Because the presence of LiTFSI has little influence on the high frequency dielectric permittivity of the polymer, addition of the salt should have little influence on the diffusion or exciton dissociation5 yield in the bulk of the polymer and therefore on the short-circuit current of the photovoltaic test structures.

of organic photovoltaic test structures fabricated from polymer blends of M-TQ1 with PCBM and doped with varying amounts of LiTFSI. We reasoned that the generation of photocurrent in photovoltaic test structures requires the transport of both electrons and holes in the active layer. Therefore, if the addition of LiTFSI introduces electron traps to M-TQ1 blends with PCBM, then we would expect dramatic reduction of the short circuit currents and fill factors of photovoltaic test structures with increasing salt concentration. Figure 6 represents the variation of electrical properties (power conversion efficiency, open-circuit voltage, short-circuit current, and fill factor) of a series of photovoltaic test structures (panel A) fabricated from M-TQ1:PCBM blends doped with varying concentrations of LiTFSI. The detailed device fabrication procedures and current−voltage curves are provided in the Supporting Information. The reduction of the power conversion efficiency and short-circuit current at a salt concentration of R = 0.05 corresponds to the concentration at which the salt interfered with the TFT mobility measurements. However, at lower salt concentrations, the electrical properties of the test structures are not strongly affected by the addition of the salt, indicating that the addition of LiTFSI at concentrations of R = 0.015 or less does not cause a high density of electron (or hole) traps in the polymer blend. On the other hand, the enhanced hole mobility and charge carrier lifetime observed with increasing salt concentration up to R = 0.015 does not result in improved photovoltaic device performance. To investigate possible reasons for this behavior, we examined the phase separated morphology of the polymer blends of M-TQ1 with PCBM. Figure 7 represents a carbon

■

CONCLUSION We investigated the possibility to tune the low frequency dielectric permittivity of an organic semiconductor over a wide range by the addition of redox inactive salts. A conjugated polymer with ethylene oxide side chains, M-TQ1, was synthesized to chelate alkali metal cations enabling the salt, LiTFSI, to be incorporated into the bulk of the polymer phase. We report an increase of the low frequency dielectric permittivity of the conjugated polymer by a factor of more than 20-fold in comparison to the dielectric permittivity of the undoped polymer. The addition of LiTFSI to the conjugated polymer increases the density of holes that gives rise to substantial increases of the hole mobility. The charge recombination lifetime of films of the polymer blended with the electron acceptor, PCBM, increases by nearly an order of magnitude with addition of the salt to the polymer phase. Photovoltaic device studies indicate that the addition of LiTFSI does not result in the formation of a high density of electron or hole traps. The findings suggest that doping conjugated polymers with redox inactive salts provides a means to tune their dielectric permittivity over a significant range. We suggest that development of ion-doped conjugated polymers capable of finer phase separated morphologies in their blends with fullerenes is a pathway to favorably influence organic photovoltaic device performance. We also believe that modifications of the ions or side chain properties of conjugated polymers that permit increases of the dielectric permittivity at higher frequency will be important for improving charge collection in functional organic solar cells.

■

ASSOCIATED CONTENT

S Supporting Information *

Detailed experimental procedures, tabulated dielectric fitting parameters, and current/voltage characteristics of photovoltaic devices. This material is available free of charge via the Internet at http://pubs.acs.org.

Figure 7. Energy filtered TEM image of a 1:3 by mass polymer blend of M-TQ1 with PCBM. Because the image represents a carbon map, the lighter regions correspond to the PCBM-rich phase. The micrograph indicates that M-TQ1 and PCBM undergo macroscopic phase separation, which limits the photocurrent of the corresponding photovoltaic devices due to the finite exciton diffusion length in the conjugated polymer.

■

AUTHOR INFORMATION

Author Contributions ∥

These authors contributed equally to the work.

Notes

The authors declare no competing financial interest.

map obtained from energy filtered TEM images of a 1:3 (by mass) M-TQ1:PCBM polymer blend that has been annealed at 100 °C for 10 min. Details of how the carbon maps are generated are provided in the Supporting Information. It should be noted that thermal annealing does not result in significant changes of the polymer blend morphology. The spheroidal objects in the carbon map consist mostly of M-TQ1 domains that are embedded in the PCBM-rich phase (the lighter regions are the denser carbon-rich fullerene phase). Since the exciton diffusion length is ∼10 nm in many polymer systems,33 the ∼500 nm length scale of phase separation of the

■

ACKNOWLEDGMENTS X.L., K.S.J., K.H., Q.W., and J.B.A. gratefully acknowledge support for this research from the Office of Naval Research under Grant No. N00014-11-1-0239, and K.S.J. and J.B.A. acknowledge partial support from the National Science Foundation under Grant No. CHE-0846241. B.P.W., K.V., C.G., and E.D.G. acknowledge financial support from NSF under Award DMR-1056199. The authors also acknowledge support from the National Center for Electron Microscopy, 15873

dx.doi.org/10.1021/jp408537p | J. Phys. Chem. B 2013, 117, 15866−15874

The Journal of Physical Chemistry B

Article

Interface Charge Recombination. Appl. Phys. Lett. 2001, 79, 2085− 2087. (21) Snaith, H. J.; Gratzel, M. Enhanced Charge Mobility in a Molecular Hole Transporter via Addition of Redox Inactive Ionic Dopants: Implication to Dye-Sensitized Solar Cells. Appl. Phys. Lett. 2006, 89, 262114(3). (22) Wibowo, R.; Aldous, L.; Ward Jones, S.; Compton, R. G. The Group I Alkali Metals in Ionic Liquids: Electrodeposition and Determination of Their Kinetic and Thermodynamic Properties. ECS Trans. 2010, 33, 523−535. (23) Belhocine, T.; Forsyth, S. A.; Nimal Gunaratne, H. Q.; Nieuwenhuyzen, M.; Puga, A. V.; Seddon, K. R.; Srinivasan, G.; Whiston, K. New Ionic Liquids from Asepane and 3-Methylpiperidine Exhibiting Wide Electrochemical Windows. Green Chem. 2011, 13, 59−63. (24) Snaith, H. J.; Schmidt-Mende, L.; Gratzel, M. Light Intensity, Temperature, and Thickness Dependence of the Open-Circuit Voltage in Solid-State Dye-Sensitized Solar Cells. Phys. Rev. B 2006, 74, 045306(6). (25) Frohlich, H. Theory of Dielectrics; Clarendon Press: Oxford, U.K., 1949. (26) Havriliak, S.; Negami, S. A Complex Plane Analysis of alphaDispersions in Some Polymer Systems. J. Polym. Sci., Polym. Symp. 1966, 14, 99−117. (27) Kremer, F. Dielectric Spectroscopy - Yesterday, Today, and Tomorrow. J. Non-Cryst. Solids 2002, 305, 1−9. (28) Klein, R. J.; Zhang, S.; Dou, S.; Jones, B. H.; Colby, R. H.; Runt, J. Modeling Electrode Polarization in Dielectric Spectroscopy: Ion Mobility and Mobile Ion Concentration of Single-Ion Polymer Electrolytes. J. Chem. Phys. 2006, 124, 144903(8). (29) Atorngitjawat, P.; Runt, J. Ion Conduction and Polymer Dynamics of Poly(2-vinylpyridine)-Lithium Perchlorate Mixtures. J. Phys. Chem. B 2007, 111, 13483−13490. (30) Knipper, M.; Parisi, J.; Coakley, K. M.; Waldauf, C.; Brabec, C. J.; Dyakonov, V. Impedance Spectroscopy on Polymer-Fullerene Solar Cells. Z. Naturforsch. 2007, 62, 490−494. (31) Sze, S. M. Semiconductor Devices: Physics and Technology; John Wiley and Sons, Inc.: New York, 2002. (32) Sheng, C.-X.; Tong, M.; Singh, S.; Vardeny, Z. V. Experimental Determination of the Charge/Neutral Braning Ration in the Photoexcitation of pi-Conjugated Polymers by Broadband Ultrafast spectroscopy. Phys. Rev. B 2007, 75, 085206. (33) Peet, J.; Heeger, A. J.; Bazan, G. “Plastic” Solar Cells: SelfAssembly of Bulk Heterojunction Nanomaterials by Spontaneous Phase Separation. Acc. Chem. Res. 2009, 42, 1700−1708. (34) Dennler, G.; Scharber, M.; Brabec, C. J. Polymer-Fullerene Bulk-Heterojunction Solar Cells. Adv. Mater. 2009, 21, 1323−1338. (35) Barbour, L. W.; Hegadorn, M.; Asbury, J. B. Watching Electrons Move in Real Time: Ultrafast Infrared Spectroscopy of a Polymer Blend Photovoltaic Material. J. Am. Chem. Soc. 2007, 129, 15884− 15894. (36) Qi, Y.; Mohapatra, S. K.; Kim, S., B.; Barlow, S.; Marder, S. A.; Kahn, A. Solution Doping of Organic Semiconductors Using AirStable n-Dopants. Appl. Phys. Lett. 2012, 100, 083305(3). (37) Guo, S.; Kim, S. B.; Mohapatra, S. K.; Qi, Y.; Sajoto, T.; Kahn, A.; Marder, S. A.; Barlow, S. n-Doping of Organic Electronic Materials using Air-Stable Organometallics. Adv. Mater. 2012, 24, 699−703. (38) Gregg, B. A.; Chen, S.-G.; Branz, H. M. On the Superlinear Increase in Conductivity with Dopant Concentration in Excitonic Semiconductors. Appl. Phys. Lett. 2004, 84, 1707−1709. (39) Gregg, B. A.; Chen, S.-G.; Cormier, R. A. Coulomb Forces and Doping in Organic Semiconductors. Chem. Mater. 2004, 16, 4586− 4599.

Lawrence Berkeley National Laboratory, which is supported by the U.S. Department of Energy under Contract No. DE-AC0205CH11231. We are grateful to Noel C. Giebink and Thomas N. Jackson for helpful discussions.

■

REFERENCES

(1) Ling, Q. D.; Liaw, D. J.; Zhu, C. X.; Chan, D. S. H.; Kang, E. T.; Neoh, K. G. Polymer Electronic Memories: Materials Devices and Mechanisms. Prog. Polym. Sci. 2008, 33, 917−978. (2) Sirringhaus, H. Device Physics of Solution-Processed Organic Field-Effect Transistors. Adv. Mater. 2005, 17, 2411−2425. (3) MacCallum, J. R.; Vincent, C. A. Polymer Electrolyte Reviews-1; Elsevier Applied Science: London, 1987. (4) Clarke, T. M.; Durrant, J. R. Charge Photogeneration in Organic Solar Cells. Chem. Rev. 2010, 110, 6736−6767. (5) Koster, L. J. A.; Shaheen, S. E.; Hummelen, J. C. Pathways to a New Efficiency Regime for Organic Solar Cells. Adv. Energy Mater. 2012, 2, 1246−1253. (6) Gregg, B. A. The Photoconversion Mechanism of Excitonic Solar Cells. Mater. Res. Soc. Bull. 2005, 30, 20−22. (7) Thompson, B. C.; Frechet, J. M. J. Polymer-Fullerene Composite Solar Cells. Angew. Chem., Int. Ed. 2008, 47, 58−77. (8) Gunes, S.; Neugebauer, H.; Sariciftci, N. S. Conjugated PolymerBased Organic Solar Cells. Chem. Rev. 2007, 107, 1324−1338. (9) Bredas, J.-L.; Norton, J. E.; Cornil, J.; Coropceanu, V. Molecular Understanding of Organic Solar Cells: The Challenges. Acc. Chem. Res. 2009, 42, 1691−1699. (10) Pensack, R. D.; Guo, C.; Vakhshouri, K.; Gomez, E. D.; Asbury, J. B. Influence of Acceptor Structure on Barriers to Charge Separation in Organic Photovoltaic Materials. J. Phys. Chem. C 2012, 116, 4824− 4831. (11) Yi, Y.; Coropceanu, V.; Bredas, J.-L. A Comparative Theoretical Study of Exciton-Dissociation and Charge-Recombination Processes in Oligothiophene/Fullerene and Oligothiophene/Perylenediimide Complexes for Organic Solar Cells. J. Mater. Chem. 2011, 21, 1479− 1486. (12) lenes, M.; Kooistra, F. B.; Hummelen, J. C.; Van Severen, I.; lutsen, L.; Vanderzande, D.; Cleij, T. J.; Blom, P. W. M. Charge Dissociation in Polymer:Fullerene Bulk Heterojunction Solar Cells with Enhanced Permittivity. J. Appl. Phys. 2008, 104, 114517(4). (13) Zeigler, D. F.; Chen, K.-S.; Yip, H.-L.; Zhang, Y.; Jen, A. K.-Y. Tunable Light-Harvesting Polymers Containing Embedded Dipolar Chromophores for Polymer Solar Cell Applications. J. Polym. Sci., Part A 2012, 50, 1362−1373. (14) Sun, Y.; Chien, S.; Yip, H.-L.; Chen, K.-S.; Zhang, Y.; Davies, J. A.; Chen, F.-C.; Lin, B.; Jen, A. K.-Y. Improved Thin Film Morphology and Bulk-Heterojunction Solar Cell Performance Through Systematic Tuning of the Surface Energy of Conjugated Polymers. J. Mater. Chem. 2012, 22, 5587−5595. (15) Noone, K. M.; Subramanian, S.; Zhang, Q.; Cao, G.; Jenekhe, S. A.; Ginger, D. S. Photoinduced Charge Transfer and Polaron Dynamics in Polymer and Hybrid Photovoltaic Thin Films: Organic vs Inorganic Acceptors. J. Phys. Chem. C 2011, 115, 24403−24410. (16) Tada, A.; Geng, Y.; Wei, Q.; Hashimoto, K.; Tajima, K. Tailoring Organic Heterojunction Interfaces in Bilayer Polymer Photovoltaic Devices. Nat. Mater. 2011, 10, 450−455. (17) Pensack, R. D.; Banyas, K. M.; Asbury, J. B. Charge Trapping in Organic Photovoltaic Materials Examined with Time Resolved Vibrational Spectroscopy. J. Phys. Chem. C 2010, 114, 5344−5350. (18) Wang, E.; Hou, L.; Wang, Z.; Hellstrom, S.; Zhang, F.; Inganas, O.; Andersson, M. R. An Easily Synthesized Blue Polymer for HighPerformance Polymer Solar Cells. Adv. Mater. 2010, 22, 5240−5244. (19) Vakhshouri, K.; Kozub, D. R.; Wang, C.; Salleo, A.; Gomez, E. D. Effect of Miscibility and Percolation on Electron Transport in Amorphous Poly(3-Hexylthiophene)/Phenyl-C61-Butyric Acid Methyl Ester Blends. Phys. Rev. Lett. 2012, 108 (2), 026601. (20) Kruger, J.; Plass, R.; Cevey, L.; Piccirelli, M.; Gratzel, M. High Efficiency Solid-State Photovoltaic Device Due to Inhibition of 15874

dx.doi.org/10.1021/jp408537p | J. Phys. Chem. B 2013, 117, 15866−15874