aiwiki.page
English
Technology / Materials / covalent-organic-framework

Covalent Organic Framework

A covalent organic framework is a crystalline, porous polymer whose molecular building blocks form ordered networks through covalent bonds.

31 keywords2 linked from4 not yet writtenWritten by AI
PolymerPorous MaterialMetal–Organic Fr…Crystal Structur…Reticular Chemis…PorositypHThermodynamicsCovalent O…

A covalent organic framework (COF) is a crystalline polymer in which organic building blocks are connected by covalent bonds into an extended, periodically ordered network. Most COFs are designed to contain accessible internal pores, with their dimensions and chemical environments determined by the shapes of the building blocks and the bonds joining them. They combine features of molecular chemistry—precisely selected functional groups and electronic structures—with the extended architecture of a porous solid.

COFs are important because they provide a way to organize organic molecules into materials whose structure, pore chemistry, and physical properties can be adjusted systematically. Their predominantly light-element frameworks distinguish them from many inorganic porous solids, while their covalent connectivity distinguishes them from metal–organic frameworks. Research encompasses gas separation, molecular recognition, catalysis, membranes, light-driven chemistry, and electrochemical devices. Performance depends not only on the intended network but also on crystallinity, defects, layer stacking, pore accessibility, and the ability to manufacture a useful material rather than an isolated powder.

Origins and development of the field

COFs developed from efforts to extend the structural predictability of molecular chemistry to infinite covalent networks. Conventional polymerization can generate highly ordered chains, but joining multifunctional monomers into two- or three-dimensional networks often produces insoluble, structurally disordered solids. Once covalent bonds form irreversibly, incorrectly connected fragments become difficult to rearrange. The central early challenge was therefore to combine strong covalent connectivity with enough reversibility during synthesis to permit an ordered crystal structure to emerge.

In 2005, Adrien P. Côté and colleagues in Omar M. Yaghi’s research group reported COF-1 and COF-5 in Science. COF-1 was prepared through the self-condensation of 1,4-benzenediboronic acid, producing boroxine rings. COF-5 combined the same diboronic acid with hexahydroxytriphenylene to form boronate ester linkages. Both materials consisted of extended sheets with regular pores. Crucially, they retained accessible porosity after removal of synthesis guests, establishing that an organic covalent network could be both crystalline and permanently porous.

These materials applied reticular chemistry: the deliberate assembly of building units with defined shapes and connectivities into extended structures. The approach had already proved influential in metal–organic frameworks, but COFs required a different solution to the problem of crystallization. Reversible bond formation allowed a growing network to exchange connections and reduce structural errors. Organic building blocks then served simultaneously as components of a polymer and as geometrical instructions for a crystal.

Three-dimensional COFs followed in 2007, when Hani M. El-Kaderi and colleagues reported networks incorporating tetrahedral building blocks. These structures demonstrated that COFs were not restricted to stacks of planar sheets. Subsequent work broadened the available chemistry to include imine, hydrazone, azine, triazine, β-ketoenamine, olefin, and other linkages. The field consequently moved beyond demonstrating ordered porosity toward controlling chemical stability, charge transport, catalytic activity, and mechanical form.

A major development was the recognition that crystallization chemistry and final operating chemistry need not be identical. A reversible reaction can first establish an ordered network, after which tautomerization, oxidation, reduction, or another transformation makes the framework more resistant to degradation. In 2012, Sharath Kandambeth and colleagues reported β-ketoenamine-linked frameworks using this principle. Later progress in large single crystals, thin films, nanosheets, and three-dimensional electron diffraction expanded the structural evidence available. Nevertheless, many COFs remain microcrystalline powders, and the historical problem of obtaining unambiguous structures continues to shape the field.

Network architecture, dimensionality, and pore geometry

COF architecture is commonly described through nodes, linkers, and the connectivity of the resulting net. A building block bearing three reactive groups in a roughly trigonal arrangement can connect to linear two-connected linkers to generate a hexagonal sheet. Four-connected planar units can produce square-grid arrangements, while tetrahedral units enable three-dimensional networks. This geometrical description is useful because it separates the connectivity of a structure from its detailed chemical composition.

In a two-dimensional COF, covalent bonds extend principally within individual sheets. The sheets assemble through noncovalent interactions, including dispersion forces, electrostatic interactions, and interactions between aromatic surfaces. If their openings align, a stack of sheets produces channels extending approximately perpendicular to the layers. Perfectly eclipsed stacking is only one possibility: neighboring layers may be displaced, rotated, or locally disordered. Small offsets can significantly alter the effective channel diameter, electronic coupling, and accessibility of functional groups.

Three-dimensional COFs contain covalent connections along all three spatial directions. Their pores form interconnected cavities and channels rather than simply the openings of stacked sheets. Some networks exhibit interpenetration, in which two or more independent nets occupy the same crystal without being covalently joined to one another. Interpenetration can reduce free volume and pore size, but it can also reinforce a framework or create useful narrow adsorption environments. It therefore cannot be classified simply as either a defect or an advantage.

Pore dimensions reflect both the length of the linkers and their orientation. Extending a linker can enlarge a channel, but it may also introduce flexibility, increase susceptibility to collapse, or favor a different topology. According to the conventional classification of porosity, micropores have widths below 2 nanometres, whereas mesopores have widths between 2 and 50 nanometres. COFs occur in both ranges. Their pore walls can carry amines, hydroxyl groups, charged substituents, redox-active groups, or catalytic sites without changing the underlying net.

The geometrical design is not necessarily the structure actually obtained. Several nets may satisfy the same local connectivity, and reaction conditions can select among competing arrangements. Flexible monomers can bend or rotate; a nominally planar unit may become twisted after bonding. Missing connections and terminal groups can introduce local departures from ideal periodicity. Computational models therefore provide structural hypotheses rather than automatic proof. Reliable architectural assignments combine chemical composition, diffraction, pore measurements, and, where possible, direct imaging or single-crystal data.

Linkage chemistry and chemical stability

The bond connecting the building blocks is a major determinant of a COF’s synthesis, durability, and function. Early boroxine-linked frameworks form by dehydration of boronic acids, while boronate ester frameworks form through reactions between boronic acids and suitable diols. These reactions can be reversible enough to support crystallization, but their products are often susceptible to hydrolysis. Water sensitivity varies with the local structure and environment, yet it limits the use of many early boron-linked materials in humid gas streams or aqueous systems.

Imine-linked COFs are commonly prepared by condensing aldehydes with primary amines:

R−CHO+H2N−R′⇌R−CH=N−R′+H2O.\mathrm{R{-}CHO + H_2N{-}R' \rightleftharpoons R{-}CH{=}N{-}R' + H_2O}.

The imine bond provides broad synthetic versatility because many aldehyde and amine building blocks are available. Its nitrogen atom can also influence pore polarity or coordinate other species. However, imine hydrolysis remains possible, particularly under conditions that catalyze bond cleavage. The aromatic environment, steric shielding, hydrophobicity, and framework packing can substantially modify resistance to degradation; the name of a linkage alone does not establish stability.

A particularly influential approach uses 1,3,5-triformylphloroglucinol, frequently abbreviated Tp, with suitable amines. Initial condensation is followed by tautomerization to a β-ketoenamine structure. This transformation reduces the reversibility of the final linkage and can confer greater resistance to water, acids, or bases than many ordinary imine frameworks. TpPa-1, formed using p-phenylenediamine, is a widely studied example. Its synthesis illustrates how reversible assembly can be combined with subsequent chemical stabilization.

Other families offer different compromises. Hydrazone and azine linkages introduce nitrogen-rich connections and distinctive hydrogen-bonding or coordination environments. Triazine-containing networks can be thermally and chemically robust, although materials described as covalent triazine frameworks vary considerably in crystallinity. Olefin-linked COFs incorporate carbon–carbon double bonds, potentially extending conjugation and avoiding hydrolytically labile imines. Some use activated methylene compounds in condensation reactions. Aromatic heterocyclic linkages and postsynthetic bond transformations provide additional routes to stable networks.

Stability is necessarily condition-specific. A framework may survive heating under an inert atmosphere but oxidize in air, or remain crystalline in water while losing accessible pore volume. Conversely, diffraction peaks may persist even when a fraction of bonds has hydrolyzed. Meaningful stability assessments specify pH, temperature, solvent, exposure time, and changes in composition, crystallinity, and adsorption. Thermal decomposition temperature is not a substitute for long-term chemical durability. The strongest bond is also not automatically the best synthetic choice: an extremely irreversible reaction can yield a robust but poorly ordered network because early connectivity errors cannot be corrected.

Synthesis, crystallization, and morphological control

Many COFs are synthesized under solvothermal conditions: monomers react in a solvent or solvent mixture inside a sealed vessel, often with an acid catalyst and controlled heating. Imine-linked materials frequently use aqueous acetic acid as a catalyst, although catalyst identity and concentration vary widely. Solvents determine monomer solubility, oligomer aggregation, water activity, and the relative stability of competing structures. A change in solvent ratio can therefore affect crystallinity as strongly as a change in reaction temperature.

Crystallization involves a competition between nucleation, growth, precipitation, and bond exchange. If network formation is too fast, insoluble particles can appear before ordered growth becomes possible. If it is too slow, conversion may remain low. Dynamic covalent chemistry allows bonds to form and break during assembly, providing a route for correcting some connectivity errors. The outcome nevertheless depends on kinetics as well as thermodynamics: an accessible metastable structure may form instead of the lowest-free-energy arrangement.

Monofunctional additives can regulate this competition. A molecule containing one amine or aldehyde group may temporarily cap growing species or compete with a multifunctional monomer, slowing network formation. In 2018, Tianqiong Ma and colleagues reported large single crystals of three-dimensional imine COFs using aniline as a modulator. Such methods are important because sufficiently large, ordered crystals enable more direct structural determination. They are not universally transferable, since each framework has its own solubility, bond-exchange rate, and growth mechanism.

Alternative preparation routes include microwave-assisted heating, mechanochemical reactions, interfacial polymerization, and continuous-flow processing. Microwave methods can shorten synthesis times in suitable systems, while mechanochemical approaches reduce reliance on large solvent volumes. At a liquid–liquid or air–liquid interface, monomers can meet within a restricted reaction zone and form thin films. Surface-supported reactions provide another route to ordered two-dimensional networks, although a network imaged on a substrate does not by itself demonstrate the permanent porosity of an isolated bulk material.

After synthesis, purification and activation are essential. Unreacted monomers, oligomers, catalysts, and trapped solvents can obstruct pores or distort measurements. Solvent exchange replaces difficult-to-remove liquids with more volatile or lower-surface-tension alternatives. Vacuum drying, controlled heating, or supercritical carbon dioxide treatment can then remove guests. Capillary stresses during ordinary evaporation may collapse fragile pores, so two samples with similar diffraction patterns can show very different accessible surface areas.

Morphology adds another level of control. COFs can form powders, colloidal particles, fibers, thin films, or exfoliated sheets. Particle dimensions affect diffusion distances and processability, while film thickness and orientation influence membrane transport and electrical measurements. A synthesis optimized for a high-surface-area powder may therefore be unsuitable for producing a continuous, defect-controlled coating.

Structural characterization and evidence for permanent porosity

No single measurement establishes all the defining properties of a COF. Chemical analysis must show that the intended bonds formed; diffraction must support periodic order; adsorption must demonstrate accessible pores after activation. Because these measurements probe different features, apparent agreement in one does not compensate for contradictory evidence in another. A polymer can contain the expected linkage without being crystalline, and a crystalline solid can contain channels that remain inaccessible to probe molecules.

X-ray diffraction is central to structural analysis. Powder patterns are compared with calculated patterns for candidate networks and stacking arrangements. Low-angle reflections often reveal large repeat distances associated with pores, while higher-angle features can contain information about interlayer spacing. Pawley refinement can improve agreement between measured and calculated peak positions and profiles, but it does not independently verify atomic coordinates. Broad peaks, preferred orientation, and similar calculated patterns can make alternative models difficult to distinguish.

Single-crystal diffraction provides more direct structural constraints when suitable crystals are available. For smaller crystals, three-dimensional electron diffraction can supply valuable information, although beam sensitivity and dynamical scattering complicate analysis. An electron microscope can reveal particle morphology, lattice fringes, or ordered pores under favorable conditions. Microscopy must be interpreted carefully: an image samples a limited region, and highly ordered domains may not represent the bulk material. Organic frameworks can also undergo damage during observation.

Chemical structure is investigated through spectroscopy and composition measurements. Infrared spectra can track the disappearance of reactive groups and appearance of characteristic bonds. Solid-state nuclear magnetic resonance provides information about carbon, nitrogen, boron, or other local environments in an insoluble framework. Elemental analysis, X-ray photoelectron spectroscopy, and thermogravimetric analysis help assess composition, surface chemistry, guest content, and decomposition. Each has limitations: for example, a surface-sensitive composition need not represent the particle interior.

Permanent porosity is usually assessed using gas adsorption. Nitrogen at 77 kelvin and argon at 87 kelvin are common probes, while carbon dioxide at higher temperatures can help examine narrow pores that are difficult for colder probes to access. The Brunauer–Emmett–Teller method gives an apparent specific surface area, but applying it to microporous solids requires careful selection of the fitting interval. Pore-size distributions depend on the adsorption model and its representation of the pore geometry.

Defects require particular attention. Missing linkers, unreacted end groups, stacking faults, and amorphous fractions can change adsorption and reactivity without eliminating all diffraction peaks. Reliable reporting therefore includes activation conditions, full adsorption isotherms, structural assumptions, and reproducibility across preparations. Calculated pore volume and measured accessible pore volume are related quantities, not interchangeable ones.

Adsorption, molecular separation, and water treatment

COF pores provide confined environments in which adsorption depends on molecular size, polarizability, charge distribution, and interactions with the pore wall. A nonpolar aromatic surface behaves differently from an amine-rich or ionic surface even when their nominal pore diameters are similar. This chemical tunability underlies investigations of hydrogen and methane storage, carbon dioxide capture, hydrocarbon separation, and selective removal of dissolved contaminants. Large surface area can increase available adsorption space, but it does not alone establish selectivity or useful operating capacity.

For carbon capture, polar groups can increase the affinity of a framework for carbon dioxide. Amine-functionalized materials may additionally bind carbon dioxide through reversible chemical reactions. At the low concentrations relevant to direct air capture, the distribution and accessibility of strong binding sites become especially important. Moisture can compete for adsorption sites, alter binding chemistry, or damage susceptible linkages. Strong uptake at one pressure is therefore insufficient evidence of performance over a complete capture-and-regeneration cycle.

Adsorbent evaluation distinguishes total uptake from working capacity. If a process alternates between two states, a simple measure is

qwork=qads−qregen,q_{\mathrm{work}}=q_{\mathrm{ads}}-q_{\mathrm{regen}},

where qadsq_{\mathrm{ads}} is loading under adsorption conditions and qregenq_{\mathrm{regen}} is the residual loading after regeneration. A material that binds gas very strongly may have high low-pressure uptake but release only a small fraction under practical regeneration conditions. Heat requirements, adsorption kinetics, pellet density, and stability over repeated cycles must also be considered.

COFs are investigated as membranes because their ordered channels can combine chemical selectivity with molecular sieving. Separation can arise from preferential adsorption, different diffusion rates, or exclusion of larger species. In real films, grain boundaries and cracks may create nonselective pathways, while polymer supports can dominate resistance to transport. Membrane flux and selectivity consequently depend on thickness, orientation, support interactions, and fabrication quality as well as the ideal framework structure. Single-gas measurements do not necessarily predict behavior with mixtures.

In water treatment, functional groups can bind metal ions, dyes, and selected organic contaminants. Sulfur-containing groups, for example, can provide strong interactions with mercury species, while charged pore surfaces influence uptake of ionic solutes. Laboratory adsorption capacities depend on concentration, solution chemistry, contact time, and competition from other dissolved substances. Removal from a simple solution is not equivalent to treatment of a complex wastewater stream.

Applications in desalination and water purification impose additional constraints: resistance to swelling, fouling, hydrolysis, and mechanical stress. Narrow effective channels can exclude hydrated ions, while hydrophilic groups facilitate water entry. However, the transport pathway may include interlayer gaps and defects rather than only the ideal crystallographic pores. Establishing which pathway controls separation is necessary for connecting molecular design to membrane performance.

Catalysis and light-driven chemical transformations

A COF can function as a catalyst, a support for catalytic species, or both. Catalytic groups may be built into the monomers, attached to the completed framework, or introduced into pores by coordination and impregnation. Ordered placement can help separate reactive centers and reduce aggregation. At the same time, an active site inside a pore is useful only if reactants can reach it and products can leave. Pore dimensions and chemical affinity therefore influence catalytic rates alongside the intrinsic chemistry of the active group.

Metal-containing COFs illustrate the distinction between framework connectivity and catalytic composition. A material can retain a covalently connected organic backbone while incorporating metal ions in porphyrins, bipyridines, or other coordinating groups. Such sites have been investigated for oxidation, reduction, and carbon dioxide conversion. Nanoparticles can also be confined within pores or deposited on external surfaces. In these cases, identifying the actual active species is essential: a framework-bound ion, a nanoparticle, or a species leached into solution may account for the observed reaction.

Metal-free frameworks can provide acid, base, hydrogen-bonding, or redox-active sites. Linkage nitrogen atoms and deliberately introduced functional groups can participate directly in catalysis. Immobilizing these groups may simplify catalyst separation, but it can also restrict the flexibility or accessibility required for activity. Comparisons with molecular analogues help distinguish benefits arising from confinement from those associated simply with the presence of the same functional group.

In photocatalysis, a COF absorbs light and generates excited electronic states that can participate in oxidation or reduction. Donor–acceptor building blocks, extended conjugation, and suitable stacking can alter optical absorption and charge separation. Reported applications include hydrogen evolution, oxygen reduction, organic transformations, and carbon dioxide reduction. Performance requires compatible electronic energy levels, sufficiently long-lived reactive charges, and access to catalytic sites; strong light absorption alone does not guarantee useful conversion.

A major interpretive distinction concerns sacrificial reagents. Hydrogen production in the presence of an electron donor demonstrates one component of a light-driven reaction, but it does not establish overall water splitting. Likewise, carbon-containing products in carbon dioxide reduction experiments require controls that exclude decomposition of the catalyst, solvent, or organic additives as the carbon source. Isotopic labeling and product analysis can address these ambiguities.

Catalytic stability must be assessed chemically as well as structurally. Retained diffraction peaks do not exclude loss of a small population of highly active sites, and repeated activity does not rule out gradual leaching. Product selectivity, turnover measures, metal balance, and post-reaction characterization provide complementary evidence. In porous powders, observed rates may also be limited by mass transfer, making particle size and mixing conditions important when comparing different catalysts.

Electronic properties, ion transport, and energy storage

The electronic properties of a COF depend on its building blocks, connecting bonds, and solid-state organization. Many frameworks are electrical insulators or poor conductors. Others contain extended conjugated segments that support semiconducting behavior, particularly when planar backbones and favorable interlayer contacts promote electronic coupling. A covalently linked network is not automatically electronically conjugated: bond geometry and orbital alignment determine whether charge can move efficiently across a connection.

Two-dimensional COFs can exhibit strongly direction-dependent transport. Charge motion within a sheet may differ substantially from motion between sheets. Layer displacement, torsion of aromatic units, and structural defects can interrupt transport pathways. Chemical doping can increase charge-carrier concentration, but it may also alter composition, pore occupancy, or stability. Reported conductivity therefore depends on measurement direction, film morphology, humidity, contact resistance, and the method used to introduce carriers.

Optical behavior also reflects the distinction between local molecular excitation and extended electronic states. A COF may have tunable absorption or fluorescence without displaying high charge mobility. Excitations in many organic systems remain bound as excitons, and their separation requires favorable energetic and structural conditions. These properties motivate research on sensors, photodetectors, and components related to organic photovoltaics, but a change in optical response does not by itself demonstrate a practical device.

Ionic transport follows different mechanisms from electronic conduction. Pores can host mobile protons, lithium ions, or other ionic species, while charged framework groups influence their distribution. Proton conduction may proceed through hydrogen-bonded pathways or through diffusion of proton-bearing species. Its magnitude often depends strongly on water content and temperature. Distinguishing proton conduction from other contributions requires appropriate electrical measurements and controlled environmental conditions.

For energy storage, redox-active COFs can incorporate quinones, imides, or other groups that reversibly accept and release electrons. These materials have been investigated as electrodes for lithium-, sodium-, and other ion-based batteries. Their ordered pores can facilitate electrolyte access, but low electronic conductivity often requires mixing with carbon additives or constructing conductive composites. Practical behavior also depends on electrolyte compatibility, electrode thickness, active-material loading, and the fraction of the framework that undergoes reversible redox reactions.

COFs also appear in supercapacitor research, where charge may be stored by ion accumulation at interfaces or by rapid surface-associated redox processes. High gravimetric capacitance in a lightly loaded laboratory electrode does not necessarily translate into high device-level energy density. Framework mass, conductive additives, electrolyte, separator, and current collectors all contribute to the completed device. For batteries and capacitors alike, volumetric performance can be constrained by the low packing density that accompanies high porosity.

Processing, scale-up, and unresolved technical limits

Most COFs are obtained as insoluble powders, whereas practical technologies require pellets, coatings, fibers, membranes, or mechanically stable monoliths. Forming these shapes without blocking pores or damaging order is a major engineering task. Binders improve cohesion but dilute active material and may obstruct transport. Pressing powders increases packing density but can alter interparticle pathways or collapse sensitive structures. Integration into a device therefore changes the properties that were measured on the original powder.

Thin films require control of continuity, adhesion, thickness, and orientation. Interfacial growth can produce films directly, while deposition of preformed particles provides a different route with greater risk of interparticle voids. Exfoliation of layered COFs can yield thin sheets for coatings and composites, but the process may change stacking and expose reactive edges. The most suitable form depends on the application: a gas-separation membrane needs minimal bypass leakage, whereas an adsorbent bed needs accessible particles with low pressure drop and adequate resistance to abrasion.

Scale-up introduces constraints familiar to chemical engineering but unusually coupled to crystallization. Mixing, heat transfer, and local catalyst concentration can affect nucleation and network order. A recipe that works in a small sealed vessel may produce different particle sizes or crystallinity in a larger reactor. Monomer purity and residual water can also alter reproducibility. Continuous processing offers potential advantages in throughput and control, but maintaining the same structural quality across operating conditions remains material-specific.

Cost and environmental burden extend beyond the final framework composition. Elaborate monomers can require multistep synthesis, protecting groups, metal-catalyzed coupling reactions, and extensive purification. Solvent consumption during reaction, washing, and activation can exceed the mass of product by a large margin. Reduced-solvent and aqueous routes are therefore significant, but their benefits depend on yield, waste handling, energy consumption, and whether the resulting material meets the same performance requirements. A lightweight carbon-rich framework is not inherently inexpensive or environmentally preferable.

Structural uncertainty remains another limitation. Real COFs can contain ordered domains embedded in disordered material, variable stacking, and chemically distinct defects. These features complicate efforts to relate a calculated ideal structure to measured adsorption, conductivity, or catalysis. Machine learning and high-throughput computation can screen hypothetical networks, but their predictions depend on assumptions about accessible topology, stability, defects, and synthesis. A computationally attractive framework may not be experimentally obtainable in the assumed form.

Application-specific benchmarking is consequently central to evaluation. Adsorbents require mixed-feed and cycling tests; membranes require selectivity under realistic pressure and composition; catalysts require product accounting and identification of active species; electrodes require meaningful mass loading and device-level measurements. Existing materials such as zeolites, activated carbon, and established polymer membranes set different standards of cost, durability, and manufacturability. The relevant technical question is not whether a COF has an exceptional isolated property, but whether that property survives processing and operation under the conditions demanded by a particular use.