The Royal Society of 1 , 5-, 2 , 6-and 9 , 10-distyrylanthracenes as luminescent organic semiconductors †

Department of Chemistry and Centre for Se University, 801 Sherbrooke Street West, M dmitrii.perepichka@mcgill.ca Institut national de la recherche scientiq Chemical Structures, 1650 Lionel-Boulet Bo Département de Chimie, Local A-634, Univ Montpetit, Pavillon Roger-Gaudry, Montréa Central Research and Development, Experim DE, 19880-0328, USA † Electronic supplementary informatio calculations; TGA data; details of uores 1,5-DPSAnt lms; a movie of operating 923794 and 923795. For ESI and crystallo format see DOI: 10.1039/c3tc30247d ‡ Current address: Institute of Advanc Technology, 30 Puzhu Road(S), Nanjing, P Cite this: J. Mater. Chem. C, 2013, 1, 2817


Introduction
The semiconducting conjugated organic molecules have received signicant recent interest as they are the basis for a fundamental understanding of charge transport in the solid state and enable innovative technologies such as organic light-emitting diodes (OLEDs), [1][2][3] organic eld-effect transistors (OFETs) 4,5 and organic photovoltaics (OPVs). 6,7The tunability of properties, multifunctionality and low-temperature processing are the three major advantages of molecular semiconductors (MSCs), pushing forward their use in optoelectronics.The combination of tunable light-emitting and charge-transport properties has been the basis of OLEDsthe most mature organic electronic technology.The short electrical channel of OLED devices does not demand high charge mobility in a semiconductor (m $ 10 À3 cm 2 V À1 s À1 ) and a large number of OLED materials with highly efficient luminescence have been synthesized over the last two decades. 8he design of high charge mobility and efficient solid state luminescence in the same molecule is more challenging and only a few materials have been shown to combine both properties.A specic technological application of such materials is organic light-emitting transistors (OLETs) 9 which combine electrical switching and light emission function in a single device.In addition, the high current density achievable in OLETs 10 could potentially lead to not yet realized organic injection lasers. 11Generally, achieving high charge mobility in MSCs requires a highly crystalline structure.However, many if not most semiconducting molecules exhibit strong quenching of the exciton in the crystalline state, empirically ascribed to the "p-aggregation" effect. 12lmost all high mobility MSCs are constructed from fused polycyclic aromatic and/or heteroaromatic (mostly, thiophene) rings. 13Their high charge carrier mobility is usually connected with their at, symmetric and rigid molecular structure that leads to a low reorganization energy associated with charge injection and efficient intermolecular electron delocalization in the crystalline state.A classical example is

Journal of Materials Chemistry C
PAPER pentacene, a benchmark p-type MSC with charge carrier mobility greater than 1 cm 2 V À1 s À1 . 14][17][18] However, most of these polycyclic aromatic MSCs show either no luminescence or very weak luminescence in the solid state.
One of the reasons for luminescence quenching in the solid state is the singlet ssion process that turns the singlets into two non-emissive triplet states. 19The thermodynamics of singlet ssion requires the energy of the triplet state to be at or below half of that of the singlet.This condition is realized in tetracene and longer acenes.For shorter acene molecules, e.g.anthracene, singlet ssion is not possible because of the higher triplet energy; a high PL quantum yield (PLQY, 64%) was reported for anthracene crystals. 20However, to the best of our knowledge, no working thin-lm OFET devices have been realized with anthracene. 21,224][25][26][27][28] Recently we reported a combination of high charge mobility (m + > 2 cm 2 V À1 s À1 ) and strong solid state emission (70%) in crystals of 2-(4-hexylstyryl)anthracene and demonstrated its application in OLET devices. 29Much less is known about anthracene-based MSCs with substituents in other positions, although solution processed OFETs for tetralkoxy-substituted 9,10-distyrylanthracenes have been reported (and showed a relatively low charge mobility of $10 À4 cm 2 V À1 s À1 ). 30In the present work, we perform a detailed comparison of the structural, molecular and solid state optoelectronic properties of a series of three isomeric distyrylanthracenes 2,6-DPSAnt, 1,5-DPSAnt and 9,10-DPSAnt (Scheme 1) and study their performance in thin-lm transistor devices.

General
Starting dibromoanthracenes were purchased from TCI America.2,6-DPSAnt was synthesized as described earlier. 25PLQY in toluene solution were measured relative to 9,10-diphenylanthracene (95% in cyclohexane), for 2,6-and 1,5-DPSAnt), and relative to uorescein (79% in ethanol), for 9,10-DPSAnt.Absolute PLQY of solid state (crystalline powders) samples were measured in the 6" integrating sphere from Labsphere Inc, using a Varian Cary Eclipse uorometer.Fluorescence lifetimes were measured with an Edinburgh Instruments Mini Tau lifetime uorometer exciting with an EPL 405 laser.Cyclic voltammetry measurements were performed in 0.1 M electrolyte solution (Bu 4 NPF 6 ) in CH 2 Cl 2 or PhCl using a Pt disk as a working electrode, Ag/AgCl as a reference electrode and Pt wire as an auxiliary electrode.Ferrocene (Fc) was added to the analyzed solution at the end of the measurement as an internal reference.All redox potentials are given relevant to the Fc/Fc + couple.AFM characterization of molecular thin lms was performed using a Bruker Multimode 8 instrument operated in the tapping mode.DFT calculations of the molecular structure (in the gas phase), molecular orbital energy, orbital splitting in the dimers 32 and hole transfer reorganization energy 35 were performed using the B3LYP functional and the 6-31G(d) basis set as implemented in Gaussian 09. 31 Frequency analysis was performed for all optimized structures.The absorption/emission spectra and the structures of the relaxed excited state were analysed by timedependent DFT at the same level of theory.

4,4,5,5-Tetramethyl-2-[(E)-2-(4-pentylphenyl)ethenyl]-1,3,2dioxaborolane
To a 250 mL Schlenk ask tted with a magnetic bar and a nitrogen inlet were added 1-iodo-4-pentylbenzene (2.0 g, 7.3 mmol) and dry toluene (20 mL).Then, 2-ethenyl-4,4,5,5-tetramethyl-1,3,2-dioxaborolane (2.02 g, 13.1 mmol) and tributylamine (3.5 mL) were added.The resultant solution was degassed using three freeze-pump-thaw cycles.Pd(PPh 3 ) 4 (0.21 g, 0.18 mmol) was added in one portion, and the reaction mixture was heated up to 100 C and kept at this temperature for 4 hours.Aer cooling down, it was diluted with dichloromethane and the resulting solution was washed with 10% HCl.The organic phase was separated, washed with brine and dried over MgSO 4 .Evaporation of the solvent under reduced pressure afforded a brownish oil which was puried by column chromatography (silica, hexane/EtOAc) to afford the pure product as a viscous orange oil (1.4 g, 64%).(6 mL).The resultant solution was degassed by purging with nitrogen for 15 minutes.Then Pd(PPh 3 ) 4 (0.01 g, 0.009 mmol) was added in one portion, and the reaction mixture was heated up to 90 C and kept at this temperature for three days.Aer cooling down to room temperature, the reaction mixture was diluted with dichloromethane (100 mL) and washed with water (3 Â 50 mL).The organic phase was separated and dried over sodium sulphate.The solvent was removed under reduced pressure and the crude product was puried by column chromatography (silica, hexane) to afford a yellow powder (0.16 g, 70%).mp: 180-182 C (from dichloromethane/ethanol). 1  To a 100 mL two neck round-bottom ask tted with a stop-cock adaptor, a magnetic bar, and an argon inlet were added 1,5dibromoanthracene (0.224 g, 0.667 mmol), 4,4,5,5-tetramethyl-2-[(E)-2-(4-pentylphenyl)ethenyl]-1,3,2-dioxaborolane (0.601 g, 2.00 mmol), and toluene (7 mL).The resultant solution was degassed using three freeze-pump-thaw cycles.Then sodium carbonate (0.35 g, 3.3 mmol) solution in water (1.7 mL) and aliquat 336 (0.13 g) were added.The resultant solution was degassed with argon for 15 minutes.Then Pd(PPh 3 ) 4 (0.018 g, 0.015 mmol) was added in one portion, and the reaction mixture was heated up to 90 C and kept at this temperature for 19 hours.Aer cooling down to room temperature, the reaction mixture was diluted with dichloromethane (100 mL) and washed with water (3 Â 50 mL).The organic phase was separated and dried over magnesium sulphate.The solvent was removed under reduced pressure and the crude product was puried by column chromatography (silica, hexane/ethyl acetate) to yield a yellow crystalline product (0.29 g, 83%

Molecular and solid state structure
The effects of the position of substituents on planarity and solid-state packing of distyrylanthracenes were studied by X-ray crystallography and by DFT calculations at the B3LYP/6-31G(d) level of theory.Single crystals of 1,5-DPSAnt and 9,10-DPSAnt were prepared at room temperature by slow evaporation of chloroform and ethanol solutions, respectively.The X-ray structure of 2,6-DPSAnt (for vacuum-grown crystals) was reported previously (CCDC 621869). 25The crystal data and structure renement parameters for all three isomers are summarized in Table 1.
In all three molecules, the vinyl group is rotated out of the plane of aromatic rings (Fig. 1).This twist is small for 2,6-DPSAnt (15 between anthracene and vinylene planes, 11 between phenyl and vinylene planes), and is not expected to signicantly disrupt the conjugation between anthracene and styryl groups.Considering that DFT calculations predict a planar structure in the gas phase, the observed twist must be due to crystal packing forces.In 1,5-DPSAnt, the additional steric interactions with hydrogens in 9,10-positions of the anthracene ring cause a larger twist (31 for anthracene/vinylene planes, 21 for phenyl/vinylene planes).The largest out-of-plane distortion was observed for 9,10-DPSAnt due to interaction with two pairs of peri-hydrogens (1,8  and 4,5); the twist of 60 between the anthracene and vinylene planes (12 between vinylene and phenyl planes) should drastically affect the conjugation with the styryl groups.
Interestingly, in 2,6-DPSAnt, the solid state packing also causes a slight distortion of the anthracene ring: the largest outof-plane shi observed for C-2 and C-6 is 0.04 Å.The anthracene core in the other two isomers is fully planar.
We analyzed the bond length alternation as a rough indicator of the efficiency of p-conjugation.The shortest exocyclic bonds are found for 2,6-DPSAnt (C ant -C vyn ¼ 1.442(10) Å), followed by 1,5-DPSAnt (C ant -C vyn ¼ 1.471(3) Å) and 9,10-DSAnt (C ant -C vyn ¼ 1.481(2) Å).This suggests that the most efficient electronic coupling of substituents occurs in the 2,6-DPSAnt isomer, as expected based on the lowest out-of-plane distortion displayed by this derivative.On the other hand, the conjugation within the anthracene ring appears to be more efficient in 9,10-DSAnt and 1,5-DPSAnt isomers.The largest bond length alternation of the conjugation pathway is 0.031 Å, 0.035 Å, and 0.055 Å for 9,10-, 1,5-and 2,6-disubstituted anthracenes, respectively (Fig. 1).The electronic properties of the three derivatives are likely dened by a competition between the outof-plane rotation of the substituents and different efficiency of conjugation of the 2-6, 1-5 and 9-10 pathways.Accordingly, it is quite difficult to predict the effect of substitution patterns, e.g. on the HOMO-LUMO gap (see below).
The position of substituents in 1,5-DSAnt and 9,10-DPSAnt hinders such herringbone packing and both isomers crystallize in the triclinic space group P 1. Within such a crystal lattice, the molecules form one-dimensional stacks and the anthracene cores of all molecules are co-parallel.However, the large slippage of the molecules within the stacks precludes any p-p interaction between the anthracene moieties.The distance between centroids of anthracene rings is 6.77 Å for 1,5-DSAnt and 6.58 Å for 9,10-DSAnt.The only p-p interactions observed in these crystals occur between the anthracene and styryl fragments (C/C contacts 3.5-3.7 Å).A similar packing arrangement was also observed in other substituted 9,10-(distyryl) anthracenes. 32he overall tighter packing of 2,6-DPSAnt is also apparent from comparison of the calculated density of the isomers (Table 1).The different solid state structures have a profound effect on the electronic communication between the isomeric molecules in the solid state.DFT calculations of the dimeric interactions within the experimentally observed crystal structures show that the orbital splitting 33 in the closest-contact dimers decreases in    2).Together with the calculated hole transfer reorganization energy l + (which increases along with the dihedral angle of the styryl substituents, Table 2) these trends predict that the highest hole mobility should be observed for the 2,6-DSAnt isomer.

Optical properties
The optical absorption/emission properties of distyrylanthracenes were studied in toluene solution and in the solid state (microcrystalline powders and thin lms).In solution 2,6-DPSAnt exhibits an absorption band with a pronounced  33 b 2,6-DPSAnt showed a biexponential decay with a small (6%) contribution of the second component s PL ¼ 18.5 nsec. 35c DFT-calculated hole transfer reorganization energy. 36d Calculated splitting of the HOMO (HOMO À HOMO À 1) for the closest contact molecular dimers observed by X-ray crystallography.   2 and ESI †).This red shi of absorption occurs notwithstanding the signicant increase of twist of the styryl substituent from fully planar for 2,6-DPSAnt, to 34 in 1,5-DPSAnt to 52 in 9,10-DPSAnt (according to gas-phase DFT calculations).We conclude that electronic communication in the anthracene between the 9,10-positions is more efficient than that between 1,5-and, particularly, 2,6-positions, in agreement with the analysis of the bond-length alternation (see above).The same trend and very similar values of the HOMO-LUMO gap were obtained from electrochemical measurements (see above).However, DFT calculations, while giving a very similar value of the HOMO-LUMO gap (3.00 eV for 2,6-DPSAnt, Table 2), fail to reproduce the trend for the other two isomers.This might be due to an overestimation of the twist angle in gas-phase DFT calculations that increases the predicted gap for 1,5-and 9,10-DPSAnt as compared to solution-based experimental values.
The photoluminescence (PL) spectrum of 2,6-DPSAnt in solution shows a vibronically structured band (l max ¼ 409 nm, 428 nm) with a fairly small Stokes shi of 14 nm (0.11 eV).The PL spectra of 1,5-DPSAnt and 9,10-DPSAnt reveal much larger Stokes shi of 49 nm (0.32 eV) and 170 nm (0.81 eV), respectively.Such large reorganization of the excited state must be due to partial or full planarization of these molecules in the excited state.A clear vibronic structure of the emission of 1,5-DPSAnt (l max ¼ 461 nm, 491 nm and 530 nm) suggests a signicant rigidication of the molecule in the excited state, expected for a planarized structure.In contrast, the emission of 9,10-DPSAnt remains featureless (l max ¼ 581 nm), implying that a substantial dihedral twist is maintained in the excited state.
Intermolecular interactions in the solid state have very different effects on the optical properties of the three isomers.The optical band-gap of 2,6-DPSAnt in the solid lms is contracted vs. that in solution by 0.45 eV (to 2.58 eV), while this contraction for 1,5-and 9,10-DPSAnt is only 0.16 and 0.17 eV, respectively.This observation is in agreement with X-ray crystallographic analysis (see above) which showed that only 2,6-DPSAnt has direct p-p interactions between the anthracene cores in the solid state.Accordingly, the exciton delocalization in the solid state of 2,6-DPSAnt leads to a much larger Stokes shi in the solid state (0.54 eV) compared to that in solution (0.11 eV).In contrast, the Stokes shis of 1,5-and 9,10-DPSAnt in the solid state (0.27 eV and 0.65 eV) are actually lower than those in solution since planarization of the excited states is hindered in the solid state.This effectively causes a blueshi of the PL band of 9,10-DPSAnt in the solid state compared to that in solution (Fig. 4).
The absolute PLQYs in the solid state were measured in the integrating sphere (Table 2).PLQY values were somewhat sensitive to the morphology of the sample, as also observed in anthracene. 20However, in all measurements the highest PLQY was observed for 9,10-DPSAnt (18-22%), followed by 2,6-DPSAnt (14%) and 1,5-DPSAnt (4-9%).The powder PLQYs are likely limited by the exciton quenching on surface defects and larger values are expected in single crystal samples. 20,29Strong aggregation-induced emission (AIE) was previously reported for 9,10distyrylanthracene and its dialkoxy derivatives, with PLQY increasing from 0.4-0.7% in THF solution to 30-50% in aqueous THF and in the crystals. 31While the relatively high solid-state PLQY of 9,10-DPSAnt is in line with the AIE report, 31 no net enhancement of PLQY for the solid vs. solution was observed in our case.Furthermore, only weak ($2-fold) emission enhancement was observed upon dilution of THF solution of 9,10-DPSAnt with water (ESI †).

Thin-lm device fabrication and studies
Thin lms of 2,6-, 1,5-and 9,10-DPSAnt were prepared by slow (0.2-0.4 Å s À1 ) vacuum deposition on oxidized Si wafers (190 nm of SiO 2 ) kept at room temperature.The morphologies of the formed lms were analyzed by Atomic Force Microscopy (AFM), both for the early growth stage (sub-monolayers) and for the fully covered samples.
The submonolayer lms of 2,6-DPSAnt consist of islands of uniform height ($3.0 nm) which implies the molecules are deposited in an upright conguration (molecular length 3.46 nm) with some tilt vs. surface normal (Fig. 5a and b).X-ray diffraction analysis of the lms showed a series of (h00) peaks with a characteristic d-spacing of 3.4 nm which corresponds well with the unit cell found in the single crystal analysis (½a). 25 However, the lateral size of these islands is relatively small (up to $0.1 mm, mean area 0.004 mm 2 ) which leads to a large number of grain boundaries in fully covered lms.This is in contrast to monosubstituted 2-(4-hexylstyryl)anthracene which produces very large islands in the same deposition conditions; 29 the difference is likely attributable to higher molecular mobility of a smaller molecule.
The lms of 1,5-DPSAnt also grow in much larger islands (up to mm size, mean area 0.1 mm 2 ) with good interconnection (Fig. 5c and d).At higher coverage, the islands continue to grow laterally to form an almost fully connected monolayer before the second layer starts growing.Such morphology formed at low nominal thickness is benecial for high charge mobility in OFET applications.X-ray analysis of the lms features a strong peak at 3.2 corresponding to a d-spacing of 2.75 nm, which is similar to the molecular length of 1,5-DPSAnt (2.94 nm).However, such a diffraction peak is not expected from the single-crystal X-ray data (see ESI †).Together with AFM data (monolayer thickness of $2.1 nm), this indicates that in vacuum-deposited lms 1,5-DPSAnt adopts a layered structure with molecules standing upright on the surfacea different molecular arrangement from that found in solution-grown crystals.
In contrast, the growth mode of 9,10-DPSAnt is most detrimental for OFET applications.The nucleation of small grains is accompanied by 3D growth which leads to separate islands of large thickness (up to 75 nm) and poor interconnection (Fig. 5e  and f).The full coverage of the surface required depositing lms of very high nominal thicknesses ($100 nm).We could not promote layer-by-layer growth of 9,10-DPSAnt through hexamethyldisilazane (HMDS) treatment of the substrate surface.
Top-contact OFET devices were prepared by evaporating gold electrodes on the $50 nm lms of distyrylanthracenes.The measured output and transfer characteristics of 2,6-DPSAnt are shown in Fig. 6a and b.The devices showed hole mobility of 0.48 AE 0.11 cm 2 V À1 s À1 (m + max 0.75 cm 2 V À1 s À1 ) and on-off ratio 10 5 to 10 7 in the saturated regime.There were no signicant differences between the performance of devices prepared on bare Si/SiO 2 and those on HMDS treated substrates.The relatively large negative threshold voltage V T ¼ À18 AE  deposited at 80 C). 37 Very similar hole mobility (up to 1.3 cm 2 V À1 s À1 ) was also measured for 2-(4-hexylstyryl)anthracene thin lm OFET, 29 which appears to be representative for 2(6)substituted anthracenes with styryl groups.
OFETs fabricated with 1,5-DSAnt under identical conditions yield a somewhat lower but appreciable mobility in the range of 0.11 AE 0.05 cm 2 V À1 s À1 (m + max 0.15 cm 2 V À1 s À1 ) with a threshold voltage of À23 AE 2 and on-off ratio 10 5 to 10 6 .Such mobility was not expected based on the single crystal X-ray analysis data which show no orbital overlap between anthracene cores in 1,5-DSAnt.On the other hand, comparison of the X-ray diffraction pattern of thin lms with that of single crystals clearly shows a different morphology of the former.In contrast, devices fabricated with 9,10-DPSAnt showed no transistor characteristics (i.e., only sub-nA current and no gate effect).This is attributed mainly to the unfavorable packing arrangements found in the crystal structure and poor lm morphology of 9,10-DPSAnt.
The drain-source current ow in the transistors based on 2,6-DSAnt or 1,5-DPSAnt is accompanied by strong green electroluminescence readily observed even in single channel top-contact devices (Fig. 7a and b).Recording the current output and light emission as a function of the drain-source voltage (V ds ) for different gate voltages (V g ) shows gate-modulated electroluminescence in p-channel driven devices (Fig. 7c); thus, the devices function as OLETs.Increasing the V g reduces the turn-on voltage of the electroluminescence.For both materials, the electroluminescence zone is localized in the vicinity of the drain electrode which is consistent with the unipolar (p-type) transport characteristics of the devices.The asymmetric positions of the HOMO and LUMO of distyrylanthracenes (Table 2) with respect to the work function of Au (ca.À5 eV) creates an energy barrier for electron injection into the channel and connes the EL in the vicinity of the electron-injecting electrode.

Conclusions
In summary, we studied the effect of positional isomerism of distyryl-substituted anthracenes on molecular packing in the solid state and their optoelectronic properties.Changing the position of styryl substituents from 2,6-to 1,5-and 9,10-leads to progressive increase of out-of-plane twist of the vinyl groups.Despite this fact, the lowest HOMO-LUMO gap, according to optical and electrochemical measurements, has been observed for the 9,10-DPSAnt which suggests a more efficient electronic communication between these positions.This can be rationalized by lower bond-length alternation in the anthracene core between the 1,5-and 9,10-positions compared to 2,6-positions, as observed by X-ray crystallography.
The semiconducting properties of the three isomers were evaluated in thin-lm transistors.In agreement with the lack of anthracene/anthracene contacts in single crystal structures and poor thin lm morphology, no eld-modulated conductance was observed in 9,10-DPSAnt based devices.However, both 2,6-DPSAnt and 1,5-DPSAnt showed good transistor characteristics with hole mobilities of up to 0.75 and 0.15 cm 2 V À1 s À1 , respectively.Such relatively high charge mobility was not expected for 1,5-DPSAnt based on its packing found by single crystal X-ray analysis.The molecules adopt a different arrangement in vacuum-grown thin lms, which was conrmed by XRD and AFM studies.This emphasizes the role of polymorphism in complicating the predictions of charge mobility in thin lms based on bulk crystallographic analysis.
All three isomers are highly emissive in the solid state and the transistors fabricated with 2,6-DPSAnt and 1,5-DPSAnt showed eld-modulated intense green electroluminescence.The emission zone is localized near the electron-injecting drain electrode, in accordance with the unipolar (hole-transporting) characteristics of the devices.
The work shows that simple modication of the anthracene core allows access to new luminescent semiconducting materials with tunable properties.In particular, high solubility and surprisingly good charge mobility observed in 1,5-DPSAnt as well as enhanced electronic communication achieved through such substitution patterns (as compared to more explored 2,6substition) make the 1,5-disubstituted anthracene an attractive building block for molecular and polymeric optoelectronic materials.
2 V can be ascribed to the low HOMO of 2,6-DPSAnt.These values are consistent with the earlier communication by Meng et al. 25 reporting the hole mobility of 2,6-DPSAnt in top-contact OFETs in the range 0.1-1.28cm 2 V À1 s À1 , depending on the temperature of the substrate during lm deposition.Also, Klauk et al. reported hole mobility of non-alkylated 2,6-(di-b-styryl)-anthracene of up to 1.3 cm 2 V À1 s À1 (in top-contact OFETs for lms
for C 40 H 42 (M + ) 522.3281; found 522.3283.Thin lm OFETs were fabricated in top-contact conguration Au drain and source electrodes thermally evaporated through a shadow mask.Thin lms of molecules were grown at room temperature by vacuum deposition (10 À6 mbar, deposition rate 0.2-0.4Å s À1 ) on HMDS treated SiO 2 /Si.SiO 2 was thermally grown (200 nm thick) on heavily n-doped (Sb) Si (r z 0.01 Ohm cm).The performance of the resulting OFETs was measured in a vacuum and using a Keithley 4200-SCS.A Toshiba IK 1000 charge coupled device (CCD) camera mounted on an optical microscope objective was used for EL imaging.The biasmodulated EL was recorded using a Hamamatsu (3137 QB) Si photodiode.