Temperature Dependence of the Drift Mobility of Poly(9,9

Nov 10, 2011 - 'INTRODUCTION. Electronic and optoelectronic devices, having thin polymer films (thicknesses of the order of a few hundreds of nanomete...
0 downloads 0 Views 795KB Size
Download PDF
ARTICLE pubs.acs.org/JPCC

Temperature Dependence of the Drift Mobility of Poly(9,90-dioctylfluorene-co-benzothiadiazole)-Based Thin-Film Devices G. C. Faria,*,†,‡ R. M. Faria,† E. R. deAzevedo,† and H. von Seggern‡ † ‡

Instituto de Física de S~ao Carlos, Universidade de S~ao Paulo, Caixa Postal 369, 13560-970 S~ao Carlos, SP, Brazil Institute of Materials Science, Technische Universitaet Darmstadt, Petersenstrasse 23, 64287 Darmstadt, Germany ABSTRACT: Charge transport in spin coated thin films of poly(9,90 -dioctylfluorene-co-benzothiadiazole) (F8BT) was studied using the Photo-Current Extraction by the Linear Increasing Voltage (Photo-CELIV) method in the temperature range of 120 to 420 K. A continuous change in the slope of a lnμ versus T2 plot was detected at 205 K with a very weak temperature dependence of the mobility at lower temperatures. According to the Gaussian Disorder Model (GDM), this behavior can be attributed to a progressive decrease in the dispersiveness of the charge transport. In this sense, the weak temperature dependence of the mobility at low temperatures arises from the nonequilibrium condition between thermal released and trapped charges, so the transport becomes trap controlled at low temperature. Furthermore, another change in the mobility behavior was observed at about 360 K, which was related to the glass transitions of the material. This indicates that even for thin films the drift mobility can be significantly affected by a variation in the dynamics and local packing of the polymer chains, which can be detected by Photo-CELIV experiments.

’ INTRODUCTION Electronic and optoelectronic devices, having thin polymer films (thicknesses of the order of a few hundreds of nanometers) as active layers, have become an important topic in the science and technology of today’s electronic engineering. Organic photovoltaics (OPVs) and organic field-effect transistors (OFETs) have been inspired by the success of organic light-emitting diodes (OLEDs), and as a consequence, a large number of new companies of organic electronics have been created in the past decade. Despite this impressive industrial attention, this technology is still under development and faces many challenges, including fundamental questions that are still not answered, requesting large effort from basic research for the future to come. One of the challenges that still persists is a thorough and detailed understanding of charge carrier transport in electronic polymers. For instance, the relatively low response time of organic optoelectronic devices, compared to that of their inorganic counterparts, is caused by the low values of charge carrier mobility, which is characteristic of highly disordered materials. To overcome this drawback, a great effort in the synthesis of new compounds and in tailoring of the morphologies of thin films made thereof has been carried out.1,2 In this context, the copolymer poly(9,90 -dioctylfluorene-co-benzothiadiazole) (F8BT) is an example of a material that has been used in PLEDs,3 in Light-Emitting Transistors (OLET),4 and in OPVs,5 in which the charge transport is influenced by structural and morphological parameters such as film thickness,6 molecular weight, and packing structure.7 The main technique used to characterize the charge carrier mobility in disordered materials is the time-of-flight (TOF) r 2011 American Chemical Society

method, which has been applied successfully to many conjugated polymers.8 One important feature revealed by this kind of measurement is the existence of two transport regimes, denoted as nondispersive (ND) and dispersive (D).9,10 Many studies also reveal that the dispersiveness of the charge transport may change as a function of temperature, and a transition from nondispersive (ND) to dispersive (D) transport (ND-D transition) has been reported.1113 Borsenberger et al. have performed an extensive study of the ND-D transition making use of Monte Carlo simulations in combination with TOF measurements of p-diethylaminobenzaldehyde diphenylhydrazone (DEH) films of few micrometers in thickness.12 The disorder in those materials was simulated assuming a Gaussian distribution of localized states with a standard deviation (σ). This study indicated a ND to D transition at a given critical temperature Tc, when plotting the logarithm of the mobility against the square of the inverse temperature. It is worth mentioning that in TOF experiments the critical temperature of the ND-D transition is defined as the temperature in which the charge relaxation time becomes longer than the charge extraction time, so thermal equilibrium is not attained during the transit time. Non-equilibrium conditions thereby means, as described in ref 11, that the thermal release of charge is not able to equilibrate trapping of charge, therefore the transport becomes highly dependent on the presence of charged sites and their successive charge release, Received: May 25, 2011 Revised: November 10, 2011 Published: November 10, 2011 25479

dx.doi.org/10.1021/jp204893q | J. Phys. Chem. C 2011, 115, 25479–25483

The Journal of Physical Chemistry C

Figure 1. Schematic illustration of the Photo-CELIV method. Vmax is the top voltage applied; λ is the nanosecond laser pulse; Δj is the extracted transiente current; j(0) is the displacement current; τLVR is the length of the voltage pulse; τdelay is the delay between the laser pulse and the voltage pulse; and τmax is the time at the highest current of Δj. On the right: schematic illustration of the used device and the molecular structure of F8BT.

namely dispersive. This also holds for a process recently described by Tajima and Yasui,14 who performed a temperaturedependent mobility study on P3HT thin films using the PhotoCELIV technique.1518 Their observation was that at low temperatures the photo-generated charge is almost completely captured by traps, and the release of the majority of the charge is not thermally activated, as generally assumed for dispersive transport, but rather activated by the linearly increasing electric field as utilized in CELIV and the related potential barrier lowering due to the Poole-Frenkel effect. The CELIV signal therefore has a significant contribution of trapped charges released by the electric field, which results in a weaker temperature dependence of the mobility, similar to the case of TOF experiments below the ND-D transition.11,12,19 As soon as the thermal equilibrium between release and trapping of charge is achieved, the transport becomes non-dispersive and can be described by a trap-modulated mobility with a stronger temperature dependence. Here, we present a study of the temperature dependence of charge transport in spin-coated F8BT thin films mainly using the photo CELIV technique. This method was developed recently for applications in semiconducting films with relatively high conductivity and has been also applied successfully for organic devices.17,18 To analyze the results, we made use of the Gaussian Disorder Model (GDM)19 together with the idea of the establishment of equilibrium between thermal release of charges and trapping in order to correlate the changes observed in the behavior of the mobility as a function of temperature with dynamic processes in the polymer chains.

’ EXPERIMENTAL PROCEDURES Figure 1 shows the schematics of the Photo-CELIV excitation and response, the chemical structure of the F8BT, and the configuration of the device used in the experiments. The ITO/ F8BT/Al devices were fabricated by spin coating F8BT toluene solution (1 wt %) on cleaned ITO substrates. Purified poly(9,9-dioctylfluoren-2,7-diyl-co-benzothiadiazole) (F8BT) was

ARTICLE

purchased from Fraunhofer Institute of Applied Physics in Golm. The molecular weight and the polydispersiveness were measured by gel permeation chromatography using polystyrene as reference. The measured values were Mn = 206 kg/mol and Mw/Mn = 3. The polymer layer thickness was about 250 nm. As top electrode, a 100 nm thick layer of aluminum was vacuum deposited (107 mbar), providing an active area of 9 mm2. All preparation procedures have been conducted under inert environments. In the Photo-CELIV technique, the incident nanosecond laser pulse I0 penetrates the whole sample through the transparent ITO electrode and thereby generates photocarriers uniformly throughout the bulk. A linear voltage ramp (LVR) is then applied to the electrodes with a delay time τdelay with respect to the laser pulse I0. The photocarriers are then driven out of the sample by the applied electric field, and the transient current Δj is recorded as the difference between two successive voltage sweeps. The measurements were carried out in the temperature range of 120 to 420 K, and according to literature, the mobility can be calculated by the following relationship:20 2d2

μ ¼ 3Aτmax

2

Δj 1 þ 0:36 jð0Þ

!

ð1Þ

where d is the film thickness, A the voltage rate of LVR, and j(0) the displacement current. The experimental parameters used for all Photo-CELIV experiments were τdelay = 2 μs, τLVR = 70 μs, Vmax = 8 V, and A = 14 V/s. The laser power and length were, respectively, 0.5 μJ/cm2 and 10 ns. After the first cycle, in which the transient Photo-CELIV was recorded, the LVR was applied a second time without laser excitation, and j(0) was recorded. The impedance spectroscopy experiments were carried out to investigate the occurrence of the molecular relaxation process in the samples. The measurements were conducted as a function of temperature using Impedance Spectroscopy (Solartron SI 1260) together with the dielectric interface (Solartron SI 1296), at frequency of 1 Hz. The samples were kept under vacuum inside the closed-cycled Helium cryostat. Measurements were conducted in steps of 10 K with a waiting time of 15 min for the sample to reach the thermal equilibrium.

’ RESULTS AND DISCUSSION Figure 2 shows the full response of a Photo-CELIV transient (Δj), superimposed to j(0) obtained at 350 K. The fact that Δj decays asymptotically to almost zero indicates that all photogenerated charge carriers have been extracted by the first LVR. Also, j(0) during the second LVR remains almost constant, which shows that injection of charge carriers by the electrodes is small. Otherwise, one would observe an increasing current j(0). Photo-CELIV profiles (Δj) obtained at different temperatures are shown in Figure 2b. While the maxima of the Δj peak decrease and the respective widths increase for decreasing temperatures, the the total amount of extracted charge carriers remains almost constant. The calculated values for the mobilities in the whole temperature range of 120 to 420 K are displayed in a ln μ versus T 2 curve, as shown in Figure 3a. This curve reveals two characteristic features: a clear change in the slope at 205 K and a deviation from a straight line in the range of 315 to 390 K. Figure 3a also displays the dielectric loss function tan δ = ε00 /ε0 at 1 Hz as a function of temperature measured by impedance spectroscopy. The tan δ versus 1/T2 curve reveals two transitions 25480

dx.doi.org/10.1021/jp204893q |J. Phys. Chem. C 2011, 115, 25479–25483

The Journal of Physical Chemistry C

Figure 2. (a) Photo-CELIV current extraction obtained for the Al/ F8BT/ITO device at 350 K. The laser pulse wavelength and the extraction voltage are indicated. (b) Photo-CELIV current profile for the Al/F8BT/ITO device at several temperatures. Inset: the Δj/j(0) values plotted as a function of temperature.

at approximately 210 K (β  relaxation) and 370 K (glass transition). The coincidence between the characteristic features of the ln μ versus T 2 curve and the peaks in the tan δ versus 1/T2 dependence is remarkable. The dispersiveness of the transport is usually quantified by the ratio t1/2/τmax, where t1/2 is the width at half-maximum of the conduction current profile of Δj.21 Thus, the higher (lower) is t1/2/τmax the more (less) dispersive is the charge transport. Figure 3b shows the dependence of the t1/2/τmax ratio on the square of the inverse temperature. A maximum is observed at 220 ( 20 K, which is coincident with the change of slope of the t1/2 versus T 2 curve and the low temperature peak observed in the tan δ measurement. Above 315 K, a change in the dependence of t1/2/τmax on temperature is clearly observed, suggesting that the charge transport becomes progressively less dispersive as the temperature increases. In fact, both the mobility calculation and the interpretation of t1/2/τmax as a dispersion parameter may be affected by the presence of bimolecular recombination,22 particularly for Δj/j(0) > 0.1. It was shown that τmax decreases, while t1/2/τmax increases as a function Δj/j(0) if bimolecular recombination predominates. However, taking the behavior of Δj/j(0), the inset of Figure 2b and t1/2/τmax of Figure 3b as a function of temperature, one concludes that in our system t1/2/τmax decreases Δj/j(0), which is opposite to that predicted for the bimolecular recombination.22 Moreover, the intensity of the CELIV profiles also increase with temperature, which is the opposite behavior expected for an increasing recombination rate. Therefore, bimolecular recombination is not the main mechanism that dictates the behavior of the CELIV results presented here.

ARTICLE

Figure 3. (a) Mobility calculated from Photo-CELIV transients as a function of temperature together with the dielectric relaxation, tan δ, in order to correlate changes in the mobility with molecular relaxations. (b) Plot of t1/2, tmax, and t1/2/tmax obtained from the CELIV measurements as a function of 1000/T2 and T.

The result exhibited in Figure 3a can be analyzed by the GDM model assuming a Gaussian DOS and hopping transport.19 GDM predicts a temperature dependence of the charge carrier mobility as μ µ exp(cσ^)2 and a ND-D transition at a given temperature Tc, where σ^ = σ/kBT is the energy disorder parameter. This model allows one to derive two expressions for the carrier mobility: one above (nondispersive) and another below (dispersive) Tc:  !  2 ! 2 σ 2 T0 ¼ μ 0 exp  for T > TC μ ðTÞ ¼ μ 0 exp  3 kB T T and



1 σ μðTÞ ¼ μ 0 exp  2 kB T

2 !

 0 2 ! T0 ¼ μ 0 exp  for T < TC T

ð2Þ where T0 and are the slopes of the ln μ versus T 2 curve above and below Tc, respectively, whose ratio T0/T 0 0 = 4/3 = 1.33. Both linear regimes, above and below 205 K (Tc in Figure 3a) were fitted by eq 2, and the calculated ratio between the two slopes was (257/178 = 1.4 ( 0.3), which is in reasonable agreement with the ratio of the slopes predicted by GDM model for an ND-D transition.11,12 Since the GDM model correlates Tc with the ND-D transition, in the framework of this model, the charge transport across the thin film of F8BT is suggested to be nondispersive above Tc and dispersive below it. It is worth pointing out that the σ value obtained in the 200 to 300 K range is only 37 meV, which is considerably lower than those obtained in previous reports.23,24 However, the disorder parameters are strongly dependent on the film morphology, which is, therefore, influenced by sample preparation. This makes it difficult to 2

25481

T 0 02

dx.doi.org/10.1021/jp204893q |J. Phys. Chem. C 2011, 115, 25479–25483

The Journal of Physical Chemistry C compare the DOS function without having deeper insight into the actual morphology.25,26 Indeed, according to ref 27, in complex semicrystalline polymers the DOS can actually be broader due to the superposition of amorphous, crystalline, and interfacial DOS. It is worth mentioning that in TOF experiments the critical temperature of the ND-D transition is defined as the temperature where the charge relaxation time becomes longer than the charge extraction time, so thermal equilibrium is not attained during the transit time. Under these conditions, it was concluded that mobility is no longer a well-defined material property and becomes thickness-dependent.11,12,19 However, despite that, we can still define Tc in the ln μ vs T 2 curve; the change in the mobility with temperature is rather smooth, indicating that the charge transport slowly approaches the equilibrium condition. This is also suggested by the behavior of t1/2/τmax, which varies continuously as a function of temperature. Nevertheless, the kink in the slope of the ln μ vs T 2 curve and a coinciding change in the behavior of the dispersiveness parameter (t1/2/τmax) were observed in the Photo-CELIV data as well, which can be considered as a fingerprint for an ND-D transition. Indeed, a recent work reported a similar behavior for P3HT using photoCELIV experiments.14 The authors measured a disorder parameter of 40 meV in the temperature range of 1.6 to 90 K. They suggested that at low temperatures the charges should be almost completely captured by traps, so the CELIV signal reflects the release of the trapped charges by the electric field under strong nonequilibrium conditions. In this case, the mobility as measured by CELIV can not be considered an intrinsic material property, as in TOF experiments below the ND-D transition. The coincidence between the β-relaxation peak in the tan δ versus T 2 and the change in slope of the ln μ versus T 2 also deserves discussion. The β relaxation is related to the onset of molecular motions in the side groups at about 210 K. Thus, the change in dispersiveness of the charge transport is also likely to be related to a change in the structural disorder or fluctuation of the local conjugation length,28,29 which occurs due to the change in the local packing induced by the molecular agitation above the β transition temperature.30 In this sense, this indicated that the onset of the β-relaxation introduces some thermal disorder in the system which, together with the decrease in the dispersiveness of the transport (approach to equilibrium), explains the change in the slope of the ln μ versus T 2 in this temperature range. Another point for discussion is the connection between the deviation from the straight line starting at 315 K and the occurrence of the glass transition. The faster increase in mobility above Tg (T > 1.2 Tg) can be attributed to the achievement of an equilibrium situation between thermal release and trapping of charges (nondispersive transport), making the mobility strongly temperature-dependent. In this situation, the carriers experience a smoother energy landscape during their motion, so the slope of the ln μ versus T2 curve is increased above 1.2Tg. At temperatures around Tg, the time scale of the collective molecular reorientations is such that the disorder can be considered static. In other words, around Tg one may consider that the energy disorder becomes temperature-dependent, producing a motional narrowing effect in the DOS, so a deviation from the GDM predicted behavior is observed. This is consistent with the nonlinear behavior observed during the glass transition.11 This interpretation has been proposed in ref 11, where a very similar behavior was observed for mobility measurements in vapor

ARTICLE

deposited films of 1-phenyl-3-diethylamino-styryl-5-p-diethylphenylpyrazoline (DEASP) carried out by the TOF technique.11 An equivalent view was proposed by Giro et al.,31 who suggested that below Tg there will be a strong localization of charges due to the geometrical restrictions which do not favor hopping between neighboring sites but rather trapping and detrapping. Following that idea, one would conclude that below the glass transition the conduction is dispersive and that above the glass transition temperature, the trapped charge carriers are liberated by the backbone molecular motion so that the equilibrium situation is achieved leading to a mostly nondispersive transport, with a strong temperature dependence of the mobility above 1.2Tg.

’ CONCLUSIONS In summary, we present a Photo-CELIV study on temperature-dependent carrier mobility in F8BT thin films. It was observed that the temperature dependence of the Photo-CELIV current profiles and the derived mobilities exhibit two main changes associated with the slow approach to the charge transport equilibrium condition, similar to the dispersive to nondispersive transition as reported previously in the literature and the second related to the glass transition of the material. Despite the ND-D transition, it may be explained within a pure geometrical approach that the coincidence of this transition with the onset of the side group motions is a strong hint that the change in the local environment can play a role in this behavior. Regarding the changes observed in the mobility behavior around the glass transition temperature, it is concluded that the molecular motions of the backbone, producing a smoothing of the transport energy landscape and a time-dependent disorder of the DOS during the glass transition process (315 to 390 K) explain the mentioned modification. Despite other effects such as bimolecular recombination and charge generation by exciton split up at the electrodes that may contribute to the mobility behavior, our results show that the general behavior of the mobility as a function of temperature might be related with the molecular relaxations and are in good agreement with the GDM model. ’ AUTHOR INFORMATION Corresponding Author

*Fax: 55-163373 9825. E-mail: [email protected].

’ ACKNOWLEDGMENT G.C.F. acknowledges the fellowship by the Fundac-~ao de Amparo a Pesquisa do Estado de S~ao Paulo (FAPESP) (Procnumber 2008/01935-5) and by the Deutscher Akademischer Austausch Dienst (DAAD) (A/09/72945; ref 415). We are also grateful to the Brazilian funding agencies FAPESP (Proc. number 2009/18354-8 and 2007/08688-0), CNPq, and MCT/INEO for funding. ’ REFERENCES (1) Suzuki, Y.; Shimawaki, M.; Miyazaki, E.; Osaka, I.; Takimiya, K. Chem. Mater. 2011, 23, 795–804. (2) Jung, B. J.; Tremblay, N. J.; Yeh, M.-L.; Katz, H. E. Chem. Mater. 2011, 23, 568–582. (3) Lehnhardt, M.; Riedl, T.; Rabe, T.; Kowalsky, W. Org. Electron. 2011, 12, 486–491. (4) Schidleja, M.; Melzer, C.; von Seggern, H. Frequenz 2008, 62, 100–103. 25482

dx.doi.org/10.1021/jp204893q |J. Phys. Chem. C 2011, 115, 25479–25483

The Journal of Physical Chemistry C

ARTICLE

(5) Lee, C. L.; Hwang, I. W.; Byeon, C. C.; Han, S. H.; Greenham, N. C. J. Nanoelectron. Optoelectron. 2010, 5, 170–174. (6) Campoy-Quiles, M. C.; Sims, M.; Etchegoin, P. G.; Bradley, D. D. C. Macromolecules 2006, 39, 7673–7680. (7) Donley, C. L.; Zaumseil, J.; Andreasen, J. W.; Nielsen, M. M.; Sirringhaus, H.; Friend, R. H.; J-S, K. J. Am. Chem. Soc. 2005, 127, 12890–12899. (8) Hertel, D.; B€assler, H.; Scherf, U.; H€orhold, H. H. J. Chem. Phys. 1999, 110, 9214–9223. (9) Scher, H.; Montroll, E. W. Phys. Rev. B 1975, 12, 2455–2477. (10) Tiedje, T.; Rose, A. Solid State Commun. 1981, 37, 49–52. (11) B€assler, H.; Borsenberger, P. M. Chem. Phys. 1993, 177, 763–771. (12) Borsenberger, P. M.; Pautmeier, L. T.; B€assler, H. Phys. Rev. B 1992, 46, 12145–121153. (13) Santos, L. F.; Faria, R. M.; de Andrade, A. R.; Faria, G. C.; Amorim, C. A.; Mergulh~ao, S. Thin Solid Films 2007, 515, 8034–8039. (14) Tajima, H.; Yasui, M. J. Phys. Soc. Jpn. 2011, 80, 63705–63709. (15) Juska, G.; Arlauskas, K.; Viliunas, M.; Kocka, J. Phys. Rev. Lett. 2000, 84, 4946–4949. € (16) Osterbacka, R.; Pivrikas, A.; Juska, G.; Genevicius, K.; K, A.; Stubb, H. Curr. Appl. Phys. 2004, 4, 534–538. (17) Dennler, G.; Mozer, A. J.; Juska, G.; Pivrikas, A.; Osterbacka, R.; Fuchsbauer, A.; Sariciftci, N. S. Org. Electron. 2006, 7, 229–234. (18) Pivrikas, A.; Ullah, M.; Singh, T. B.; Simbrunner, C.; Matt, G.; Sitter, H.; Sariciftci, N. S. Org. Electron. 2011, 12, 161–168. (19) B€assler, H. Phys. Status Solidi 1993, 175, 15–56. (20) Juska, G.; Arlauskas, K.; Viliunas, M.; Geneviczius, K. Phys. Rev. B. 2000, 62, 16235–16238. (21) Juska, G.; Nekrasas, N.; Genevicius, K.; Stuchlik, J.; Kocka, J. Thin Solid Films 2004, 451452, 290–293. (22) Bange, S.; Schubert, M.; Neher, D. Phys. Rev. B 2010, 81, 35209. (23) Blakesley, J.; Clubb, H.; Greenham, N. Phys. Rev. B. 2010, 81, 45210–45219. (24) Zhang, Y.; Blom, P. Appl. Phys. Lett. 2011, 98, 143504–143507. (25) Kline, R.; M., M. D. J. Macromol. Sci. 2006, 46, 27–45. (26) Yang, Y.; Shi, Y.; Liu, J.; Guo, T.-F. In Electronic and Optical Properties of Conjugated Molecular Systems in Condensed Phases; Hotta, S., Ed.; Research Signport: Kerala, India, 2003; pp 307354. (27) Kaake, L.; Barbara, P.; Zhu, X.-Y. J. Phys. Chem. Lett. 2010, 1, 628–635. (28) Freire, J. A.; Tonezer, C. J. Chem. Phys. 2009, 130, 134901–134908. (29) Tanase, C.; Blom, P. W. M.; de Leeuw, D. M.; Meijer, E. J. In Physics of Organic Semiconductors; Br€utting, W., Ed.; Wiley-VCH: Weinheim, Germany, 2005. (30) Faria, G. C.; Plivelic, T. S.; Cossiello, R. F.; Souza, A. A.; Atvars, T. D. Z.; Torriani, I. L.; deAzevedo, E. R. J. Phys. Chem. B 2009, 113, 11403–11413. (31) Giro, G.; Di Marco, P. G. Chem. Phys. Lett. 1989, 162, 221–226.

25483

dx.doi.org/10.1021/jp204893q |J. Phys. Chem. C 2011, 115, 25479–25483