Mirjam
Dogru
and
Thomas
Bein
*
Department of Chemistry and Center for NanoScience (CeNS), University of Munich (LMU), Butenandtstr. 5-13 (E), 81377 Munich, Germany. E-mail: bein@lmu.de; Fax: +49-(0)89-2180-77622; Tel: +49-(0)89-2180-77621
First published on 21st November 2013
Covalent organic frameworks (COFs) are a novel class of porous crystalline organic materials assembled from molecular building blocks. The construction principles of these materials allow for the design of precisely controllable structures since their chemical and physical properties can be easily tuned through the selection of the building blocks and the linkage motif. Their extraordinary and versatile properties impart functionality that is of great interest in areas such as gas storage, separation, catalysis and optoelectronics. This feature article discusses key aspects of the design of covalent organic frameworks with a focus on electroactive COFs for potential optoelectronic and photovoltaic applications.
Nanoporous frameworks such as zeolites, aluminophosphates (AlPOs) or metal–organic frameworks (MOFs) are attractive due to their well-defined crystalline nature combined with a great structural and compositional variety. Crystallinity implies atomically precise definition of pore sizes, pore shapes, pore topology, and internal distribution of reactive sites, just to name a few. Thus, storage, separation and catalytic conversion of molecules are among the most prominent applications of these materials. For many concepts addressing efficient energy storage or energy conversion, accessible high internal surfaces with well-defined and tunable wall properties are often desirable for charge transfer and transport. Global warming and the need for a sustainable energy supply illustrate the importance of discovering and developing new materials and technologies in the field of clean or renewable energy, including gas storage, photocatalysis or photovoltaics. For example, solar cells have great potential as emission-free energy sources, however this technology is still too expensive to compete with fossil fuel-based power.8 In this context, the extensively investigated field of organic semiconductors provides several attractive features with respect to conventional inorganic solar cells. Many organic compounds can be efficiently produced and processed on a large scale and can be easily modified by the versatile tools of organic chemistry.9,10 The ability to design the desired material on a molecular basis permits fine tuning of the energy gap and the light absorbance of the semiconductor. In addition, due to large absorption coefficients even small amounts of the photoactive species offer the potential for efficient energy conversion. This allows for the fabrication of very thin devices. However, important issues still need to be addressed, for example, the devices often exhibit low stability and efficiencies that are lower than expected. The latter is often caused by low charge carrier mobilities and recombination, due to inefficient stacking of the conducting polymers or due to insufficient and disordered donor–acceptor interfaces. Therefore it would be desirable to have synthetic access to conducting materials with total control over their nanoscale structure and orientation.
The first donor–acceptor system (D–A system), a bilayer planar heterojunction, was introduced by Tang in 1979.11 The concept of combining a donor and acceptor layer revolutionized the field of organic solar cells and paved the way towards creating more efficient donor–acceptor architectures. The first 1% solar cell was achieved by Tang with a system that contained a derivative of phthalocyanine as donor material and a perylene derivative serving as acceptor material. The main advantage of a D–A-system is the enhanced charge separation compared to a single layer device.8 The introduction of C60 fullerene and its derivatives, such as [6,6]-phenyl C61 butyric acid methyl ester ([60]PCBM),12 as acceptor molecules helped to enhance efficiency. The most intensively studied systems today are polymer–fullerene combinations. In 1992 the first electron transfer between a fullerene and a polymer was independently demonstrated by Heeger et al. and Yoshino et al.13,14 Only one year later the first polymer–fullerene photovoltaic device was reported.15 A major breakthrough towards higher efficiencies was realized by mixing the donor and acceptor phases to form interpenetrated donor–acceptor systems; these so-called bulk heterojunctions were introduced by Hiramoto et al. through the co-evaporation of donor and acceptor molecules under high-vacuum conditions. The first organic photovoltaic devices based on mixed phases of polymer–fullerene and polymer–polymer were independently reported by Heeger and Friend.16,17 The widely postulated ideal structure for an efficient organic solar cell is a bicontinuous and interpenetrated network of the donor and acceptor phases in a bulk heterojunction (Fig. 1).10
Fig. 1 Evolution of the organic solar cell devices from bilayer to blended bulk heterojunctions and targeted ordered interpenetrated donor–acceptor systems. |
In these bulk heterojunctions the active layer forms a diffuse interface between the donor and acceptor, thus the interface is significantly larger and more charges can be separated at the donor acceptor interface, compared to planar bilayers. The overall efficiency for exciton dissociation could be improved and long exciton lifetimes in the bulk materials were no longer necessary. However, we note that for efficient charge transport and collection of photogenerated electrons and holes efficient percolation pathways are required. Clearly, the percolation pathways will strongly depend on the morphology of each semiconductor phase.18,19 Numerous studies show how the morphology of the donor–acceptor blends can be influenced, the most exhaustively studied D–A systems being poly(3-hexylthiophene) (P3HT) and [6,6]-phenyl C61 butyric acid methyl ester ([60]PCBM). It was demonstrated that the performance of the active layer is extremely sensitive towards synthesis parameters, such as the ratio of the polymer to fullerene, solvent or solvent mixtures, additives, and thermal or solvent annealing.20–23 However, while in donor–acceptor blends the (disordered) morphology of both materials can be influenced to a certain degree, it cannot be predicted and control over the structural parameters is lacking. In the following, different concepts aimed at controlling the morphology of the active layer at the nano or even atomic scale will be discussed. The first attempt was to covalently attach the acceptor fullerene moieties to the donor polymer backbone (Fig. 2a). The first so-called double cable polymers with a polythiophene backbone were reported in 1996 by Benincori et al.24 and one year later by Yassar et al.25 These compounds were intensively investigated in organic photovoltaics, however, until now only rather low efficiencies could be reported.26–28 This was attributed to increased recombination, due to too short distances between the donor and acceptor on the one hand and non-ideal packing and low concentration of fullerene on the other hand. Increasing the amount of fullerene and the distance between D and A was expected to lead to higher efficiencies. It became obvious that the domain size and morphology of donor and acceptor phases at the nanoscale are crucial for the device performance. Generally, using self-assembly processes of di-block copolymers can permit morphological control of the donor–acceptor phases on the nanometer scale.29,30 The concept of using diblock copolymers in organic solar cells was presented by Sun et al.31 The resulting structures can be fine-tuned by changing synthetic parameters, the domain size and the phase separation behavior can be adjusted by choosing the type of the copolymer, and the overall nano-morphology can be shaped by the volume fraction of the components (Fig. 2b).32,33 With these systems the elusive goal of elucidating the direct impact of nano-morphology on the photovoltaic efficiency seemed to be within reach. However, while the resulting D–A interfaces served well to separate charges (as shown by efficient luminescence quenching) unfortunately these systems faced severe recombination losses and low efficiencies.34–36 It was postulated that recombination might be reduced by inserting spacers between the donor and acceptor branches of the polymer.37 Furthermore control over the orientation relative to the substrate was also viewed as a way to increase the efficiency.37 Another approach was introduced by Müllen and co-workers, relying on a self-assembly process induced by differences in solubility of hexabenzocoronene and perylenebisdicarboximide (Fig. 2c). The donor and acceptor phases assemble into a bilayered structure with large interfacial area.38 In addition to ‘bottom up’ strategies based on molecular assembly, ‘top down’ methods were also developed with a view of creating ordered nanomorphologies for photovoltaics. For example, nanoimprint lithography is capable of creating interpenetrated organic photovoltaic devices (Fig. 2d). Thereby a pattern is imprinted from a mold onto a substrate that is coated with an organic film; the structure of the mold determines the structure of the film. The complimentary semiconductor can be deposited on the imprinted film by spin coating a solution. Hereby the penetration of the solution into the nanostructured film is crucial to obtain an intimate contact and a large-area donor acceptor interface.39
Fig. 2 Examples of ordered organic donor–acceptor systems. (a) Illustration of double cable polymers, adapted with permission from ref. 33, copyright 2010 American Chemical Society, (b) di-block copolymer morphologies as a function of molar fraction and schematic representation of a di-block copolymer, reproduced from ref. 40; (c) self-assembly of hexabenzocoronene and perylenediimide, adapted with permission from ref. 33, copyright 2010 American Chemical Society, (d) nanoimprint process flow, mold pressed onto a polymer and SEM images of (left) Si nanolined mold and (right) nanoimprinted P3HT, adapted with permission from ref. 39, copyright 2012 American Chemical Society. |
In addition to organic and inorganic solar cells, hybrid organic–inorganic systems have also attracted much attention. Typical hybrid solar cells comprise nanostructures (e.g., arrays of nanorods or nanotubes) of zinc oxide, titanium oxide or cadmium selenide that function as acceptors for interpenetrating polymer donors.37,41,42 While such designs often take advantage of inorganic solid nanostructures to control nanomorphologies to a certain degree, they are not within the scope of this review.
In this review we will explore the potential of a new class of materials, covalent organic frameworks (COFs), as promising candidates for organic solar cells. Based on the powerful toolbox of organic chemistry, these crystalline systems are proposed as an intriguing family of highly defined model systems that provide the possibility to investigate the impact of nanomorphology on the behavior of photovoltaic systems in great detail. After a short introduction to this new class of materials we will focus on electroactive covalent organic frameworks and their potential in organic solar cells.
Fig. 3 Reversible reactions of boronic acids: (a) self-condensation forms a boroxine ring resulting in a staggered COF-1, and (b) co-condensation with a vicinal diol (HHTP) forms a boronate ester resulting in COF-5. Coloring scheme: C, gray; H, white, B, orange, O, red. Adapted with permission from ref. 43. Copyright 2005 AAAS. |
Yaghi postulated that the reaction has to be carried out in a closed reaction system with solvents in which diboronic acid is not completely soluble. These conditions allow a slow condensation of BDBA. The sparing solubility of BDBA in this system controls the diffusion of the building blocks into solution and facilitates the nucleation of a crystalline material. The use of a closed reaction system sustains the availability of H2O for maintaining slightly reversible conditions conducive to crystallite growth.43 The co-condensation reaction of diboronic acid with the trigonal building block hexahydroxytriphenylene (HHTP) produces a five-membered BO2C2 ring (Fig. 3). In order to build up a 2D COF the organic linkers have to meet certain criteria. Connectivity of the organic building blocks is one criterion. Until now organic building block combinations with either linear–trigonal, trigonal–trigonal or linear–tetragonal connectivity gave a large variety of COF structures.4,5
In 2006 Lavigne reported a reverse assembly concept, i.e. co-condensation of a linear alcohol (tetraol) with a triangular triboronic acid giving COF-18 Å.46 In 2007 Yaghi and co-workers extended the scope of 2D hexagonal COFs. COF-6, -8 and 10 were synthesized by linking the trigonal HHTP with trigonal 1,3,5-benzenetriboronic acid (BTBA), trigonal 1,3,5-benzenetris(4-phenylboronic acid) (BTPA) or linear 4,4′-biphenyldiboronic acid (BPDA), respectively.47 Due to the reticular assembly of planar aromatic building blocks of different size, COFs with varying pore sizes from 0.9 to 4.7 nm were constructed.
For host–guest interactions with large molecules, pores with several nm in diameter are desired. Recently we introduced the BTP-COF with fully accessible open pores of 4 nm, synthesized by co-condensation of 1,3,5-benzenetris(4-phenylboronic acid) (BTPA) and the polyol 2,3,6,7-tetrahydroxy-9,10-dimethylanthracene (THDMA).48 After degassing the material for 12 h at 150 °C a large surface area of 2000 m2 g−1 was obtained; this value agrees well with the calculated Connolly surface. The largest pore size reported until now was achieved by condensing 2,3,6,7,10,11-hexahydroxytriphenylene (HHTP) and 4,4′-diphenylbutadiyne bis(boronic acid) (DPBA).49 The porosity of the HTTP-DPB COF was evaluated by nitrogen sorption measurements. The obtained BET surface of 930 m2 g−1 was compared to a simulated isotherm using Monte Carlo simulations (2670 m2 g−1) and to the calculated Connolly surface (2640 m2 g−1), thus achieving access to about 40% of the theoretical porosity of the COF. The interlayer stacking of this material was evaluated by molecular dynamics and DFT calculations, predicting a offset of about 0.17 nm between the adjacent layers of the hexagonal sheets.
The first non-hexagonal 2D COF was reported by Spitler et al., made by condensing a tetragonal octaol, or more precisely a tetragonal phthalocyanine tetra(acetonide) with a linear diboronic acid and addition of catalytic amounts of a Lewis acid (BF3*OEt2) to form a tetragonal lattice.50 Following the concept of reticular chemistry, several tetragonal COF structures have been made to date, ranging from the microporous to the mesoporous regime with nominal pore sizes up to 4.4 nm.51–55 The condensation of zinc porphyrin tetraboronic acid with tetrahydroxybenzene was investigated using different solvent ratios and reaction times. Due to the solubility of the porphyrin species in dioxane the solvent ratio had to be adjusted to a higher mesitylene content allowing a reversible self-healing and slow crystallization. After 15 days at 120 °C cube-shaped crystals sized about 0.5 μm were obtained.52 In Fig. 4, building blocks with different connectivity and resulting COF structures are depicted.
Banerjee and co-workers recently reported a new route to acid- and base-stable COFs using a combination of a reversible and an irreversible organic reaction. In the first step the COFs were synthesized by the reversible Schiff base reactions of 1,3,5-triformylphloroglucinol (Tp) with p-phenylenediamine (Pa-1) and 2,5-dimethyl-p-phenylenediamine (Pa-2), respectively. The expected enol–imine (OH) form underwent irreversible proton tautomerism, and only the keto–enamine form was obtained, which resists boiling water, strong acids and bases.59 The formation of three isoreticular covalent organic frameworks (COFs) via a room-temperature and solvent-free mechanochemical synthesis route was recently reported by Biswal et al.60 However, while such methods would allow for large-scale COF production, these mechanochemically synthesized COFs show rather poor crystallinity and porosity.
Fig. 5 Modeled crystalline structures of COF-102 (a), COF-105 (b), and COF-108 (c), reprinted with permission from ref. 44, copyright 2007 AAAS. Condensation of tert-butylsilane triol with tetra(4-dihydroxyboryl-phenyl)methane as boronic acid forms a borosilicate cage and gives a 3D crystalline framework with the crystalline structure depicted in (d). Color code: carbon is grey, boron is illustrated as orange and oxygen atoms are represented as red. Adapted with permission from ref. 61. Copyright 2008 American Chemical Society. |
These structures are among the most porous frameworks, exhibiting fully accessible porosity with surface areas as high as 4210 m2 g−1 (COF-103), comparable to the surface areas of MOFs such as MOF-177 (4500 m2 g−1)62 and MIL-101.63 Furthermore, COF-108 is one of the most porous organic materials with the lowest density (0.17 g cm−3) reported.44 Therefore these materials are extensively investigated for gas storage applications. Expanding the idea of reticular chemistry, Hunt et al. employed the condensation of borosilicate clusters, known from the borosilicate glass Pyrex, to build thermally and chemically stable 3D COFs. COF-202 is formed by B–O–Si linkage of tert-butylsilane and the tetrahedral tetra(4-dihydroxyboryl-phenyl)methane (Fig. 5). The tert-butyl groups are facing the pores of the network providing a chemically transformable site with sufficient space for further reactions.61 While the gas storage properties of COFs are of great present interest, they are not within the scope of this review. The interested reader is referred to pertinent articles and reviews.4,5,64–70
Fig. 7 (a) Co-condensation reaction of triboronic acid and alkyl-functionalized tetraol resulting in COF-18 Å to 11 Å, reproduced with permission from ref. 77, copyright 2008 by John Wiley Sons, Inc., (b) general strategy for the pore surface engineering of COF-5 by a co-condensation of an azide bearing phenylboronic acid followed by click reaction. Reproduced with permission from ref. 78. Copyright 2011 Nature Publishing Group. |
The first internal functionalization of 3D COFs was achieved by Bunck et al. using a monomer truncation strategy.79 The modification of COF-10244 was realized by replacing one of the four boronic acid groups with a dodecyl-functionalization and a vinyl-functionalization (Fig. 8a). Co-condensation of the tetrahedral building block and the functionalized trigonal building block results in a modified COF-102-C12/-allyl.
Fig. 8 (a) Co-condensation of a truncated linker and a tetraboronic acid results in the functionalized 3D COF-102, reproduced with permission from ref. 79, copyright 2012 by John Wiley Sons, Inc., (b) thiolene click reaction with COF-102/allyl results in COF-102-SPr, reproduced from ref. 80 with permission from The Royal Society of Chemistry. |
The functionalization was proven by IR spectroscopy of the powder and 1H-NMR after digestion in CD3CN/D2O. The crystallinity of COF-102-C12/-allyl was confirmed with X-ray diffraction. After staining the COF-102-allyl with OsO4 vapor, transmission electron microscope images indicate a distribution of the allyl groups throughout the crystallites. These results show a homogeneous incorporation of the functionalized monomer into the lattice of the framework. In a subsequent study the truncated allyl groups inside the pores were used as anchor points for further organic post-modification reactions. The authors show that the framework can be successfully functionalized using a thiolene click reaction (Fig. 8b). The crystallinity and porosity of the framework remained intact after the modification reaction, illustrating the potential versatility of this reaction for other boronate ester-based COFs.80
Fig. 9 Schematic representation of the different stacking types: AA, AB, serrated and inclined. Reproduced from ref. 81. Copyright 2010 Lukose et al.; licensee Beilstein-Institut. |
Clancy and Dichtel used complementary density functional theory and molecular dynamics calculations to determine the interlayer distance in the diphenylbutadiyne (DPB)-based COF, DBP-HHTP COF. They suggest that in this material the adjacent layers are horizontally offset by about 0.17–0.18 nm rather than showing an eclipsed stacking, which agrees with the values reported by Heine.81 Recently they extended the study to other boronate ester-based covalent organic frameworks. In all 33 cases of already synthesized structures and hypothetical structures an offset of 0.14–0.28 nm is predicted, thus the exact eclipsed stacking type for boronate-based COFs does not seem to be a realistic model.82 This calculated offset in the stacking behavior might have a substantial impact on the charge carrier transport properties of COFs.83,84
The electronic characteristics of planar covalent organic frameworks on graphene were investigated by means of dispersion-corrected density functional theory.85 The oriented growth of several hexagonal and square lattice COFs had already been shown by Dichtel and coworkers (see above).86 Recently Gunasinghe et al. examined the self-assembly of two different COFs, the NiPc-PBBA and the HHTP-PBBA COF (PBBA is also called 1,4-benzenediboronic acid (BDBA) or phenyldiboronic acid (PDBA)), on graphene and the associated charge transfer at the interface (Fig. 10). Apparently, the self-assembly of COFs on graphene is aided by the strong delocalization of electrons from the graphene layer into the HOMO–LUMO levels of the COF, resulting in strong interactions between the orbitals.
Fig. 10 Ball and stick models of (a) and (c) COF-5, and (b) and (d) NiPc-COF. Carbon atoms of COF are colored purple and of graphene green and hydrogen white, oxygen red, nitrogen blue, nickel light blue and boron pink, respectively. Reproduced with permission from ref. 85. Copyright 2012 American Chemical Society. |
Siebbeles et al. introduced a new computational method to calculate the extent of delocalization of charges in triphenylene-based COFs with varied intermolecular distances and twist angles.84 The authors found that the electronic coupling between the sheets is strongly dependent on the intermolecular distance and the twist angle, and suggested that the best electronic coupling will occur in ecliptically stacked COF sheets. Thus delocalized charges should be able to move via a temperature dependent band-like mechanism along the stacked columns and COF materials are expected to exhibit very high charge carrier mobilities. Experimental mobility studies at different temperatures will elucidate the charge transport mechanism. These calculations strongly motivate experimental investigations of the optoelectronic properties of these materials, and suggest that promising candidates for organic solar cells might be found in this family of materials.
Fig. 11 (a) Ppy-COF synthesized by self-condensation of pyrene diboronic acid, (b) TP-COF synthesized by a co-condensation reaction of pyrenediboronic acid and hexahydroxytriphenylene. Tetragonal COFs with reported high charge carrier mobility: (c) nickel phthalocyanine-based COF, reproduced with permission from ref. 51, copyright 2011 by John Wiley Sons, Inc., (d) porphyrin-based tetragonal COFs synthesized by imine condensation and (e) by boronate ester condensation reactions. Reproduced with permission from ref. 54. Copyright 2011 American Chemical Society. |
The electrical conductivity of TP-COF was measured using the two-probe method. For semiconducting materials a linear I–V curve is expected, this was observed for TP-COF under standard conditions, with a current of 4.3 nA at a bias voltage of 2 V between a 10 μm gap. The COF was dispersed in acetone and drop-cast on the electrode, the second electrode was then thermally evaporated on the COF film. Upon doping the framework with iodine a significant rise in current up to 20 nA was observed, thus suggesting a p-type semiconducting character of TP-COF. Self-condensation of pyrenediboronic acid (Fig. 11a) also results in an optoelectronically active framework with a pore size of 1.73 nm and a specific surface area of 923 m2 g−1.
Ppy-COF generates a moderate photocurrent of 5 nA upon illumination with a xenon lamp, which was attributed to the defined conduction paths for charge migration provided through the highly ordered structure.90 For the device fabrication for photoconductivity measurements, the COF was dispersed in PMMA (Ppy-COF/PMMA = 50/50 wt%), the resulting film thickness was 100 μm. Another photoconductive COF with a tetragonal metallophthalocyanine building block, NiPc-COF (Fig. 11c), was reported to have high charge carrier mobilities.51 Incorporating phthalocyanine into the framework broadens the optical absorption profile of the COF. The eclipsed stacking of the phthalocyanine and arene boronic acid leads to an open framework with a surface area of 624 m2 g−1 and a micropore diameter of 1.9 nm. Upon irradiation with a xenon lamp the current at 1 V bias increased from 20 nA to 3 μA, furthermore panchromatic photoresponse was very fast and reproducible without deterioration. The intrinsic charge carrier mobility of NiPc-COF and the following structures shown in this review were obtained with laser flash-photolysis time-resolved microwave conductivity measurements (FP-TRMC), which give the product ϕ*Σμ (ϕ is the photocarrier generation yield and Σμ is the sum of the charge carrier mobilities). The yield of photocarrier generation ϕ was determined by integration of the time-of-flight (TOF) transients at different bias voltages. The number of charge carriers was extrapolated to a bias of 0 V, leading to the charge carrier generation yield ϕ, expressed as the number of charge carriers per photon. Yaghi extended the scope of macrocyclic COFs exhibiting semiconductor properties with two new structures based on the condensation reactions of porphyrin derivatives (Fig. 11d and e), through either boronate ester formation with tetrahydroxyanthracene (COF-66) or imine bond formation with terephthalaldehyde (COF-366), respectively.54 With time-resolved microwave conductivity measurements of a 1.5 μm thick COF/poly(methylmethacrylate) (60:40 wt%) film sandwiched between indium tin oxide and aluminum electrodes, the hole mobilities of COF-66 and COF-366 were reported to be as high as 3.0 cm2 V−1 s−1 and 8.1 cm2 V−1 s−1, respectively.
The first electron-transporting COF was created by substituting the benzene groups at the edges of the NiPc-COF with electron-deficient benzothiadiazole (BTDA) building blocks (Fig. 12a). The resulting n-channel NiPc-BTDA COF was reported to exhibit a high electron mobility of 0.6 cm2 V−1 s−1. The absorbance of this COF is very broad and ranges up to 1000 nm, resulting in a panchromatic photoresponse with high sensitivity to near-infrared photons. Upon excitation with white light photocurrent at 1 V bias was enhanced from 250 nA to 15 μA (Fig. 12a).55
Fig. 12 (a) Schematic synthesis pathway for NiPc-BTDA with absorbance spectra of 2D-NiPc-BTDA COF (red) and (MeO)8PcNi (black), and on–off switching of photocurrent of 2D-NiPc-BTDA COF under the bias voltage of 1.0 V at different wavelengths, reproduced with permission from ref. 55, copyright 2011 American Chemical Society, (b) D–A-COF co-condensed from benzothiadiazole diboronic acid and HHTP shows no current in the dark (left, black) and a linear photoresponse upon irradiation (red) with visible light from a xenon light source, and exhibits a high on–off ratio photoconductive response (right), reproduced with permission from ref. 92, copyright 2012 by John Wiley Sons, Inc., (c) schematic representation of the structure of ZnPc-NDI COF with the absorbance spectrum, reproduced with permission from ref. 93. Copyright 2013 by John Wiley Sons, Inc. |
Semiconducting COFs with tunable charge carrier transport behavior were created with 2D porphyrin COFs having different central metals (MP-COFs).91 Insertion of a copper central metal leads to electron transport along the framework, whereas free base porphyrin COF is hole conducting. The zincated material ZnP-COF exhibits ambipolar charge transport. Flash-photolysis time-resolved microwave conductivity (FP/TRMC) methods under argon atmosphere gave the total carrier mobilities (electrons and holes), while in measurements under an SF6 atmosphere the electrons are trapped, hence giving only hole mobilities. Another example of ambipolar charge transport, a 2D D–A-COF, was formed by a co-condensation reaction of a benzothiadiazole bearing boronic acid with the triangular building block HHTP.92 The resulting COF with a pore size of about 2.8 nm was reported to exhibit high electron and hole mobilities of 0.04 and 0.01 cm2 V−1 s−1, respectively. In the case of D–A-COF it was suggested that the electrons can be conducted along the columns of the benzothiadiazole unit and the holes along the HHTP columns (Fig. 12b).
Control over morphology and the creation of separated domains of donor and acceptor phases is one of the challenges in creating efficient organic solar cells. One possibility to gain control over the location of the donor and the acceptor at an atomistic level is the incorporation of D and A molecules into the crystalline structure of a covalent organic framework. Jiang and coworkers reported that by co-condensing the electron-accepting benzothiadiazole benzenediboronic acid and the donor hexahydroxytriphenylene, a D–A-heterojunction with atomistically separated periodic D columns and A columns can be synthesized.92 This ‘ambipolar’ COF shows a high on–off ratio photoconductive response and high charge carrier mobilities. The charge carrier dynamics of these D–A-COFs were elucidated with time-resolved spectroscopy. The generation of free charges upon light absorption in the D–A columns is achieved within 2 ps. The π-stacking of the COF allows for delocalization of the charges and suppresses recombination. Expanding on the intracrystalline D–A theme, a COF based on zinc phthalocyanine serving as the donor and naphthalene diimide (NDI) as the acceptor shows light absorption over a wide range of the visible spectrum and into the near IR up to 1100 nm (see Fig. 12c).53,93 Differential pulse voltammetric measurements of the built-in D–A system showed oxidation of the ZnPc at 0.42 eV and a reduction of the NDI at −0.52 V, which the authors interpreted as a large exothermic driving force for electron transfer. The electrochemical measurements were recorded with a dispersion of ZnPc-NDI-COF in benzonitrile, with respect to a Ag/AgNO3 reference electrode. The ultrafast charge separation and the long retention of the charge separated state, due to the delocalization along the π-columns in the D–A-COF, illustrate the high potential of these systems for photovoltaic applications.
The familiy of semiconducting COFs is constantly growing. Recently Bertrand et al. reported three new thiophene-based covalent organic frameworks (Fig. 13a). In general the boronic acid equipped thiophene building blocks allow COF formation, following the concepts of reticular chemistry.94 However, bent monomers such as the thiophene diboronic acid tend to form defects in the structure, due to their nonlinear binding motif. Combining two thiophene molecules in an antiparallel fashion leads to a linear linkage motif and COF structures can be readily obtained with high crystallinity. The authors could also show the formation of a charge transfer complex between the thiophene diboronic acid building block and tetracyanoquinodimethane (TCNQ), while electron transfer from the COF to the TCNQ was not observed. Other oxidizing agents such as 2,3-dichloro-5,6-dicyano-1,4-benzoquinone (DDQ), chloranyl or I2 caused degradation of the framework. These results illustrate the importance of a careful search for suitable redox partners to achieve full electron transfer without disrupting the crystalline framework.
Fig. 13 (a) Schematic representation of T-COF 1 (left) and T-COF-3 based on thiophene-based building blocks, reproduced with permission from ref. 94, copyright 2013 by the National Academy of Sciences of the United States of America, (b) thienothiophene-based TT-COF showing charge transfer to included PCBM as indicated by complete photoluminescence quenching, right: I–V-profile of the first COF-based photovoltaic device, reproduced with permission from ref. 95. Copyright 2013 by John Wiley Sons, Inc. |
We could recently show that it is possible to integrate covalent organic frameworks in a bulk-heterojunction photovoltaic cell.95 The thienothiophene-based TT-COF was synthesized in a co-condensation reaction of thienothiophene diboronic acid with HHTP. Upon irradiation with light the system showed significant photocurrent. The large surface area of 1800 m2 g−1 and the 3 nm open pores of the hole-transporting TT-COF enable the uptake of large electron transporting materials, such as the fullerene derivative PCBM (see Fig. 13b). Inclusion of PCBM leads to a novel periodic interpenetrated donor–acceptor system, which shows the spectroscopic signatures of efficient charge transfer at the nanoscale and can act as a photovoltaic cell.
One important aspect in designing novel 2D COFs relates to the exact stacking motif of the building blocks; control of the stacking is expected to have a profound effect on optoelectronic properties. The interplay between theoretical investigations and synthetic realization of stacking motifs with subtle structural differences is anticipated to shed more light on this important feature of COFs. Moreover, the degree of lateral conjugation and electronic coupling in 2D COFs, together with control over domain size and crystal morphology will be additional means to modify and optimize the behavior of excitons and charge carriers in these systems. A further extension of these concepts may be realized by chemical functionalization of the walls of the COFs. This approach as well as refilling the pore system with complementary donor or acceptor phases rests on the ability to clean the pores from residues resulting from the synthesis – careful optimization of extraction techniques will be important.
There will also be a need to enhance the chemical stability of COFs for some applications. This can be achieved by using new building blocks and new linkage motifs. As discussed above, one novel approach is the formation of imine-linked networks that can undergo keto–enol tautomerism to yield COFs that are stable in boiling water, acids and bases.
With judicious use and further development of the above synthetic tools, the physical and chemical properties, such as thermal stability, absorbance spectrum and conductivity of COFs can be tailored by the choice of the appropriate building blocks. The well-defined crystalline structure and the large accessible internal surface area and porosity, which can be filled with sizeable functional molecules, point to the great potential of these materials as excellent model systems for investigating ordered and interpenetrated networks of donor–acceptor systems at the nanoscale.
This journal is © The Royal Society of Chemistry 2014 |