The discovery of graphene by Geim and Novoselov,1, 2 reported in 2004, added a previously unknown member to the carbon family and quickly inspired scientists to explore the extraordinary properties of this new material. The unique two-dimensional (2D) hexagonal carbon network of graphene3 has been found to give rise to unprecedented mechanical properties,4, 5 high thermal conductivity (3080–5150 W/mK at room temperature),6–10 as well as interesting optical properties (97.7% transparency).11–13 As a result, graphene has quickly found numerous applications in electronics,14–16 novel composites,17, 18 biosensors,19–21 and a range of biological applications,22 including drug delivery and medical imaging.23, 24 The unusual combination of properties of graphene are solely associated with the single layer two-dimensional carbon lattice,25 which is in contrast to the three-dimensional lamellar structure of graphite.
However, the strong tendency of the single graphene platelets to undergo agglomeration due to strong π–π and Van der Waals interactions26 can have a significant and detrimental effect on final material properties. Numerous strategies have been developed to control the dispersion of graphene, limit reaggregation, and enhance compatibility with a receiving host matrix. These involve either a demanding chemical or physical exfoliation process to initially separate the sheets, followed by surface modification to disrupt the reforming of the π–π and Van der Waals interactions between separated graphene sheets. Alternatively, graphene oxide (GO), a product of graphene oxidation, has found widespread application as it can be easily converted back to graphene and offers several advantages over graphene due to its oxidized surface chemistries. The surface chemistry of GO is characterized by the presence of hydroxyl, carboxylic acid, and epoxy groups,27, 28 allowing GO to be readily dispersed in water and polar solvents.29–31 In addition, covalent modification by, for example, amidation of the carboxylic groups,29, 32 nucleophilic addition to the epoxy groups,33 and diazonium salts coupling,30, 34–36 have further enabled manipulation of the solution properties of GO. As GO can be readily dispersed in aqueous solutions, it can be used to substitute the insoluble graphene in wet-processing techniques. The aqueous dispersibility is greatly beneficial for processing GO into films, sheets, and composite materials.32, 37, 38 Dispersions of GO in organic solvents can also be achieved through surfactant support and ionic interactions39, 40 or appropriate covalent functionalization, for example, with octadecylamine.29
Although there are significant examples of graphene/GO modifications via small molecules by both covalent and noncovalent means, these approaches still suffer many disadvantages. Noncovalent, intermolecular interactions such as hydrogen bonding or electrostatic forces can be exploited to adhere molecular systems to these aromatic supports, but such interactions are difficult to control and quantify due to the inherent instability of the resulting supramolecular system.41 Covalent coupling of preformed small/or macromolecular entities to GO functional groups can be problematic due to steric crowding on a sterically shielded substrate.
An alternative approach is the precise and well-controlled modification of either graphene or GO with polymer chains (brushes) to achieve enhanced dispersion and compatibility with various matrices. This has been found to be particularly advantageous for preparing homogenous nanocomposites/mixtures with remarkable properties.18, 42, 43 Using this approach, it is possible to design and synthesize novel materials with a new array of properties combining those of the polymer component (functionality, biodegradability, temperature responsiveness, etc.) with those of the graphene/GO (electrical-, optical-, and mechanical properties, etc.).
Graphene/GO can be modified with polymer chains via either grafting-from44 or grafting-to45 approaches, which are common general strategies for surface modification. The grafting-from method (also referred to as surface initiated polymerization) entails immobilization of initiator functionalities on the graphene/GO and subsequent polymerization using these moieties as initiators/mediators. Based on the grafting-to method, a presynthesized polymer chain (typically using controlled/living radical polymerization (CLRP)46) is instead attached (covalently or noncovalently) to the (functionalized) graphene/GO surface.
CLRP46 techniques can provide a high degree of control over the functionality, grafting density and thickness of the grafted polymer brush. Integrating the CLRP approaches (mainly nitroxide-mediated radical polymerization (NMP),47, 48 reversible addition-fragmentation chain transfer (RAFT) polymerization,49, 50 atom transfer radical polymerization (ATRP),51–53 and single-electron transfer living radical polymerization (SET-LRP)54–56) with grafting-from and grafting-to methods paves the route for post modification and ultimate improvement of final graphene/GO hybrid structures, expanding the scope for materials applications.
In addition to the use of CLRP to modify graphene/GO with polymer brushes, a range of other techniques have been reported.43 Such techniques, which fall outside the scope of the present Highlight, include grafting-from approaches involving ring-opening57 and Ziegler-Natta58 polymerization, as well as grafting-to approaches using esterification,23 amidation,59 radical coupling,60 and click chemistry.35
This article provides a comprehensive and timely account of the published literature on the functionalization of graphene and GO with polymer brushes. The focus lies on functionalization of graphene/GO using CLRP techniques, whereas a discussion of the physical properties of the obtained materials (in pure form or as part of polymer nanocomposite materials as fillers) goes beyond the scope of this work; the reader is referred to recent reviews on this topic by Kim et al.18 and Kuilla et al.42
Classification of Radical Polymerization Techniques
The process of radical polymerization can be divided into two categories: (i) conventional radical polymerization, and (ii) CLRP.53, 61 In practical terms, the difference between the two categories is that CLRP, unlike conventional radical polymerization, makes it possible to prepare polymer of predetermined molecular weight, narrow molecular weight distribution (MWD; low Mw/Mn), as well as other complex polymer architectures such as stars, comb polymers, block copolymers, and so forth. In CLRP, “livingness” refers to the number fraction of polymer chains that are dormant and can be further chain extended, whereas “control” refers to Mn increasing linearly with conversion and Mw/Mn decreasing with increasing conversion.62 In conventional radical polymerization, the lifetime of a propagating radical is typically of the order of 1 s, and chains are continuously initiated throughout the polymerization,63 making it impossible to control chain growth. All CLRP systems developed to date operate on the same basic principle of propagating radicals alternating between active and dormant states, thereby essentially providing sufficient time for macromolecular engineering. The three most well-known CLRP techniques are NMP,47, 48 ATRP,51–53 and RAFT polymerization.49, 50 To date, grafting-from techniques for graphene/GO modification have been reported using ATRP,64–70 SET-LRP,71, 72 and RAFT.73–75
Atom Transfer Radical Polymerization
The CLRP technique ATRP was developed independently by Matyjaszewski52, 53 and Sawamoto.51 The method is based on a dynamic equilibrium between alkyl halide species (dormant chains) and propagating radicals, which is established via reversible homolytic halogen transfer between a dormant chain and a transition metal (usually Cu) complex in its lower oxidation state, resulting in generation of propagating radicals and the metal complex in its higher oxidation state coordinated with the halogen atom. The transition metals are complexed with ligands (e.g., amines and pyridine based ligands), which serve to increase their solubility in monomer/solvent as well as strongly influence the activation/deactivation rates.
A number of reports have appeared very recently describing the use of ATRP to functionalize GO with polymer chains by use of a grafting-from approach,64–70 which entails modification of GO with covalently attached ATRP-initiating moieties (note that the same initiators can be used for both ATRP and SET-LRP). In most reported cases, the polymerizations have been carried out in the presence of sacrificial (free) ATRP initiator in the continuous phase. Overall, a number of studies have been reported dealing with the use of chemically modified graphene or GO as initiator (i.e., the nanosheets contain covalently attached initiator moieties) for ATRP and SET-LRP. However, to date, the polymerization process itself, as well as the final grafted polymer obtained (as opposed to free polymer originating from sacrificial initiator), have not been carefully studied (e.g., Mn vs. conversion during grafting-from polymerization), and in many instances, uncertainties remain as to what extent control/livingness was actually achieved. This casts some doubt on the structural uniformity of the prepared polymer brushes.
Fang et al.64 functionalized reduced GO sheets (by treatment of GO with hydrazine hydrate) with lateral dimensions of ∼30–150 nm with polystyrene based on the approach outlined in Scheme 1. The reason for functionalizing reduced GO as opposed to GO is that the physical properties (e.g., electrical conductivity and thermal stability) of reduced GO are closer to the extraordinary properties of graphene than GO.4, 6, 13, 76 Reduced GO was treated with (4-aminophenyl)ethanol/isoamyl nitrite to introduce hydroxyl groups, followed by reaction with bromopropionyl bromide to anchor ATRP-initiating moieties. The graphene-OH exhibited improved thermal stability relative to GO due to the (partial) elimination of oxygen-containing labile groups. Polystyrene chains were grown from the functionalized graphene sheets (1,2-dichlorobenzene/CuBr/N,N,N′,N′,N″-pentamethyldiethylenetriamine (PMDETA)/110 °C) in the presence of the sacrificial ATRP initiator methyl-2-bromopropionate. Based on thermogravimetric analysis, the polymer content was 82 wt %. The polymer originating from the sacrificial initiator had Mn = 60,000 g/mol and Mw/Mn = 1.6. The MWD of polymer from sacrificial initiator is in most cases different from that of the grafted polymer,66, 77 but the high value of Mw/Mn nonetheless suggests issues with the level of control/livingness also for the grafted polymer. Successful grafting was supported by Raman spectroscopy, FTIR, as well as AFM analysis, the latter showing that the nanosheets increased in thickness from 0.73 to 3 nm after grafting. As previously observed,23 the nanosheets were of considerably smaller dimensions after the grafting process (20–40 nm) owing to a series of sonication steps during preparation. Some larger sheets exhibited a thickness of 7 nm, which was attributed to limited stacking.
The motivation for functionalization of graphene with polymer chains is often to increase the compatibility between graphene and the polymer matrix in polymeric nanocomposite materials.42, 78 The smaller the graphene sheets, the easier it is to minimize stacking and more effectively disperse the sheets in the polymer matrix. However, the trade-off is that the extraordinary thermal and electrical properties of graphene are gradually compromised as the sheet size is reduced. With this in mind, Fang et al.66 revisited their earlier work,64 but this time used larger GO sheets (0.5–1.5 μm). Because of the lower edge-to-plane ratio (and thus lower content of COOH per mass), the use of a surfactant (sodium dodecylbenzene sulfonate; SDBS) was necessary to achieve an aqueous dispersion for the synthesis of OH-functionalized sheets, and subsequently ATRP initiator-bonded sheets, following their previously reported approach.64 grafting-from ATRP of styrene was carried out as before (described above), using three different ratios of monomer:initiator. The molecular weight of the grafted polymer for one of the samples was measured by GPC after cleavage using acid-catalyzed transesterification, revealing quite good agreement with the polymer formed in solution from the sacrificial initiator, and thus the latter approach was adopted for the other samples. Monomer conversions were 71–85% with Mn = 21,300–81,600 g/mol, and Mw/Mn = 1.59–1.90. The high values of Mw/Mn indicate that the control (and possibly also livingness) of the polymerization was not of the standard expected in CLRP. The Mn of the polymer did increase with decreasing degree of ATRP initiator functionalization of the sheets (as adjusted when grafting the initiator by varying the amount of diazonium salt). The resulting sheet thickness (∼8 nm; sample with Mn = 21,300 g/mol) is consistent with a single layer structure, considering the radius of gyration of the polymer being grafted to both sides of the sheet.
Yang et al.65 attempted surface modification of GO based on a different strategy. Carboxylic acid groups of thermally exfoliated GO nanosheets were selectively coupled with 1,3-diaminopropane aided with N-hydroxysuccinimide and N-(3-(dimethylamino)propyl)-N′-ethylcarbodiimide hydrochloride chemistry, resulting in diaminopropane forming an amide bond with GO carboxylic groups. Tethering of diaminopropane increases the hydroxyl functionality population at the cost of a reduction in electrostatic interaction between GO nanosheets due to carboxylic acid moieties. This led to some restacking of GO platelets, as AFM images revealed 2.7–5.6 nm thickness, representative of three to five layers overlayed GO (Fig. 1). Immobilization of the ATRP initiator 2-bromo-2-methylpropionyl bromide to GO-OH was carried out via reaction with hydroxyl and amine groups. Grafting-from ATRP of 2-(dimethylamino)ethyl methacrylate (DMAEMA) was carried out at 60 °C using PMDETA/CuBr in the absence of solvent. In a control experiment, sacrificial ATRP initiator was used, and analysis of the thus obtained free polymer yielded Mn = 27,000 g/mol and Mw/Mn = 1.72, indicating poor control over the MWD. AFM analysis revealed the polymer domains on the sheets to be ∼4 nm in size. Based on TEM analysis, the sheet size was of the order of several micrometers.
Goncalves et al.68 prepared GO sheets with covalently attached poly(methyl methacrylate) (PMMA) via a similar approach. The carboxylic acid groups of the GO sheets were first converted to acyl chloride groups, then hydroxyl groups via treatment with ethylene glycol, followed by attachment of ATRP initiator moieties by reaction with 2-bromo-2-methylpropionyl bromide. PMMA brushes were subsequently grown (CuBr/PMDETA/65 °C) in DMF as solvent, notably in the absence of sacrificial initiator. The polymer chains were cleaved from the GO sheets using hydrolysis, and subsequent GPC analysis yielded Mn = 1.170 g/mol and Mw/Mn = 1.09. Given the very low degree of polymerization, the control over the MWD (as evidenced by the low Mw/Mn) is surprisingly good (in general, Mw/Mn decreases with increasing degree of polymerization for a given polymerization up to intermediate conversion levels63). In addition to more conventional methods of analysis to confirm successful surface modification of GO (FTIR, TEM, and TGA), it was concluded based on AFM measurements in friction mode that the polymer chains were homogeneously distributed at the GO surface. Based on AFM analysis, the sheet size was of the order of several micrometers.
Wang et al.70 prepared reduced GO sheets (a few hundred nm to 2 μm) and reacted these with the diazonium salt of 2-(4-aminophenyl)ethanol through diazonium addition reaction, followed by attachment of ATRP-initiating moieties using 2-bromoisobutyl bromide. Grafting-from ATRP of 2-ethyl(phenylamino)ethyl methacrylate was conducted in 1,2-dichlorobenzene at 80 °C with CuBr/hexylmethyltriethylenetetramine in the presence of free initiator (methyl bromoisobutyrate), resulting in Mn = 14,800 g/mol with Mw/Mn = 1.33 for the unattached polymer. Finally, the polymer brushes were converted to azo polymer by azo coupling reaction with the diazonium salt of 4-aminobenzonitrile.
Single-Electron Transfer Living Radical Polymerization
The CLRP technique known as SET-LRP,54–56 also referred to as Cu(0)-mediated radical polymerization, has received significant attention in recent years due to its high polymerization rate at low temperature and exceptionally high end group fidelity (livingness).79 The exact mechanism of SET-LRP remains under debate,46, 80 although it is proposed that a central feature of the process is disproportionation of Cu(I) in suitable solvents resulting in formation of Cu(0) and Cu(II). It has been proposed that activation occurs by reaction of alkyl halide with Cu(0) (as opposed to with Cu(I) in ATRP), whereas deactivation occurs via reaction with Cu(II) as in ATRP. The high end group fidelity of SET-LRP is an important advantage both when synthesizing polymer and when using SET-LRP to grow chains from surfaces such as GO in that it enables one to take the polymerization to complete monomer conversion without significant loss of control/livingness. This is not possible with traditional CLRP techniques such as NMP, ATRP, and RAFT, because the end group fidelity decreases dramatically at very high conversion (beyond ca. 80%). As such, a desired degree of polymerization (molecular weight) can be achieved by simply adjusting the [monomer]/[initiator] ratio, with no consideration given to the conversion at which a given degree of polymerization is reached (as would be the case if the polymerization is stopped at conv. ≪ 80%).79, 81–83 The ability to conduct SET-LRP at low temperature can be beneficial in preserving both pretreated functionalities as well as intrinsic oxygen content of GO. Moreover, it is relatively simple to introduce SET-LRP-initiating moieties on the GO surface (as well as for ATRP), whereas the requirement of multiple synthetic steps can add complexity to the application of RAFT for GO modification.73, 74, 84, 85 There are to date three reports detailing modification of GO by use of SET-LRP using grafting-from techniques.67, 72, 86
Deng et al.86 modified GO (prepared via a modified Hummer's method; lateral dimensions ranging from several hundred nm to tens of μm) with poly[poly(ethylene glycol) ethyl ether methacrylate] (PPEGEEMA) using grafting-from SET-LRP. Epoxide groups of the GO basal plane were reacted with the amine functionality of tris(hydroxymethyl) aminomethane (TRIS), thus increasing the amount of hydroxyl groups. X-ray photoelectron spectroscopy (XPS) and elemental analyses revealed ∼1 TRIS moiety per seven aromatic rings. Alkyl halide initiating groups were subsequently introduced via esterification reaction between hydroxyl groups and α-bromoisobutyryl bromide. Surface-initiated SET-LRP of PEGEEMA using CuBr/tris(2-(dimethylamino)ethyl)amine (Me6TREN) was carried out in water/THF at 40 °C in the presence of the sacrificial initiator methyl 2-bromopropionate. Note that Cu(I) was initially added to the system as opposed to Cu(0) (typically copper wire in SET-LRP). Under the present solvent conditions, Cu(I) would presumably disproportionate rapidly to form Cu(0) and Cu(II) before activation can occur via a Cu(I) ATRP process.87 Covalent linkage of PPEGEEMA to the GO was confirmed by 1H NMR and FTIR. GPC analysis of the PPEGEEMA initiated by the sacrificial initiator resulted in Mn = 8500 g/mol (degree of polymerization = 34) and Mw/Mn = 1.23. Although this approach is routinely used, it has recently been reported that in the majority of cases, the molecular weights (distributions) of polymer originating from surface initiation and sacrificial initiator are significantly different.77 TGA analysis revealed that the PPEGEEMA-GO sheets comprised 45 wt % polymer. The grafting of PPEGEEMA appears as dark spots on the GO sheets as visualized by TEM, in sharp contrast to the original GO sheets (Fig. 2). The dark spots are surprisingly large (∼70 nm in diameter), significantly larger than what can be explained by each spot corresponding to one single chain. While this point was not explored further, AFM analysis revealed that single GO sheets remained isolated (minimal stacking) also after the polymerization, with the polymer causing an increase in sheet thickness from 1 nm to 8–10 nm. The PPEGEEMA-GO sheets could be readily dispersed in both polar (e.g., water and methanol) and nonpolar (e.g., chloroform and toluene) solvents with the aid of sonication.
Chen et al.72 used SET-LRP in a grafting-from approach to prepare reduced GO sheets functionalized with poly(tert-butyl methacrylate (tBMA)). Initially, GO sheets were (partially) reduced to graphene using hydrazine hydrate. Keeping the structure of graphene intact is crucial in regard to benefiting from the exceptional properties of graphene, which are (partially) lost on oxidation. However, removal of the oxygen functionalities (hydroxyl, carboxyl, etc.) from GO imposes severe limitations on the available post modification strategies. In this work,72 this restriction was compensated for by reintroduction of hydroxyl functionalities upon addition of 2-(4-aminophenyl) ethanol and isoamyl nitrite mixture (following diazonium addition). Subsequently, the GO-OH sheets were functionalized with bromopropionyl bromide to incorporate a SET-LRP initiation site. SET-LRP of tBMA in DMSO in the presence of the sacrificial initiator 2-bromopropionate was conducted using Cu(0) wire and Me6TREN at 25 °C for 1 day. Raman and 1H NMR analyses confirmed successful covalent linkage of poly(tBMA) to the GO sheets. The polymer formed in solution from the sacrificial initiator (level of monomer conversion not reported) had Mn = 4500 and Mw/Mn = 1.12. Thermogravimetric analysis revealed that the modified GO sheets comprised 71.7 wt % polymer. TEM analysis revealed polymer as dark spots on the sheets, and AFM analysis indicated that the sheet thickness increased from 0.9 to 8.3 nm on grafting of the polymer chains (Fig. 3). The lateral dimensions of the sheets were of the order of 500 nm before functionalization.
Lee et al.67 used grafting-from SET-LRP (the authors state this is an ATRP mechanism, but the conditions described using Cu(0) fall under the generally accepted SET-LRP concept—as pointed out above, the exact mechanism of SET-LRP remains under debate46, 54, 80) to grow different polymeric brushes (polystyrene, PMMA, and poly(butyl acrylate)) from acid bromide-anchored GO supports. Unlike the above mentioned GO pretreatment strategies before attaching ATRP initiator, these workers subjected hydroxyl functionalities of GO in pristine form to esterification reaction. XPS analysis revealed successful attachment of the Br-containing initiator moieties to the GO surface (Fig. 4). The polymerization was carried out in DMF at 80 °C using Cu(0) and tris(2-aminoethyl)amine. The resulting grafted polymer chains were detached from the GO by saponification with sodium hydroxide in DMF, revealing that Mn was in good accord with the theoretical Mn values (calculated from the various monomer:initiator ratios), and Mw/Mn = 1.36–1.80, indicating controlled/living characteristics.
Reversible Addition-Fragmentation Chain Transfer Polymerization
RAFT polymerization49, 50 is a CLRP technique that is based on degenerative chain transfer as a means of converting dormant chains to active propagating radicals. The method relies on the high chain transfer coefficients of thiocarbonylthio compounds and trithiocarbonates. In contrast with systems based on the persistent radical effect (e.g., ATRP and NMP), degenerative transfer-based systems require the addition of a radical initiator, because the chain activation process does not lead to an overall increase in the number of radicals. RAFT polymerization has been exploited for functionalization of GO with polymer chains using grafting-from approaches,73–75, 88 whereby suitably modified GO sheets operate as macroRAFT agents in a subsequent polymerization step.
Bin et al.73 used RAFT polymerization to grow poly(N-vinylcarbazole) from the surface of GO sheets. Poly(N-vinylcarbazole) was specifically chosen due to its usefulness for fabrication of various optoelectronic devices. GO was first functionalized with the RAFT agent S-1-dodecyl-S′-(α,α′-dimethyl-α″-acetic acid)trithiocarbonate (DDAT) by use of the hydroxyl groups of GO in combination with 1,3-dicyclohexylcarbodiimide (DCC)/DMAP esterification chemistry. Polymerization using the GO-RAFT agent (in the presence of free RAFT agent) was subsequently carried out in THF at 70 °C with azobisisobutyronitrile (AIBN) as initiator. The grafted polymer was detached from the GO by treatment with m-chloroperbenzoic acid, and GPC analysis revealed Mn = 8050 g/mol and Mw/Mn = 1.43, that is, the control over the MWD was relatively poor (level of monomer conversion not reported). RAFT polymerization was also conducted under the same conditions using only free RAFT agent, resulting in Mn = 5420 g/mol and Mw/Mn = 1.16, that is, much better control, highlighting the fact that the surface initiated nature of the polymerization combined with the addition–fragmentation mechanism appears to significantly reduce the control/livingness of the system. This may be related to the fact that the RAFT agent was attached to the GO via the R-group (as opposed to the Z-group50) of the RAFT agent. During polymerization, the RAFT moiety is thus (periodically) detached from the GO surface (if attached via the Z-group, the RAFT moiety is permanently fixed at the surface), and this may lead to loss of control/livingness of the chains tethered to GO.89, 90 Successful grafting was supported by FTIR, Raman spectroscopy and AFM analysis, the latter revealing the initial GO sheets to have a thickness of 0.8 nm (lateral dimensions on micron scale), which increased to ∼20 nm on grafting. This somewhat larger than expected increase in thickness (considering the fairly low molecular weight of the grafted chains) was attributed to stacking of sheets caused by Van der Waals interaction of polymer chains.
Yang et al.75 functionalized reduced GO sheets with poly(N-isopropyl acrylamide (NIPAM)) using click chemistry to immobilize the RAFT agents onto the sheets, followed by grafting-from polymerization. Alkyne-functionalized reduced GO sheets were obtained by modification with the diazonium salt of propargyl p-aminobenzoate, followed by attachment of the azido-terminated RAFT agent DDAT to the sheets via azide-alkyne click chemistry. Grafting-to polymerization was conducted in DMF initiated by AIBN at 60 °C in the presence of a trace amount of free RAFT agent. Molecular weights were only determined for the free polymer (not the polymer grafted to the sheets), resulting in Mn = 19,000 g/mol and Mw/Mn = 1.77 (monomer conversion not reported). Although the grafting process itself was successful (as evidence by FTIR, TGA, and XPS analyses), the high value of Mw/Mn indicates that the control over the MWD was poor. TEM analysis showed relatively evenly distributed dark dots of 2–3 nm in size, representing poly(NIPAM) domains.
Etmimi et al.74 used grafting-from RAFT polymerization to grow polystyrene chains from GO sheets in an aqueous miniemulsion system. The RAFT agent, dodecyl isobutyric acid trithiocarbonate (DIBTC), was attached to the GO sheets by esterification between the OH-groups of GO and the COOH-group of DIBTC based on DCC/DMAP chemistry, that is, via the R-group (not the Z-group). A miniemulsion comprising the organic phase (styrene and AIBN) and hexadecane, the latter to prevent Ostwald ripening91) and the aqueous phase (water, GO-RAFT, and the surfactant SDBS) was created by ultrasonication according to a multistep procedure, followed by polymerization at 75 °C. The molecular weights of the grafted polymer (measured after detachment of chains from the GO at the ester bond under basic conditions) increased with decreasing GO-RAFT:monomer ratio as expected, and Mw/Mn was in the range 1.26–1.68, with the highest molecular weight being as high as Mn = 177,400 g/mol. Unlike Bin et al.,73 no free RAFT agent was used. The values of Mw/Mn increased with decreasing amount of GO-RAFT (increasing target Mn), consistent with CLRP theory (in general, it is easier to obtain good control/livingness for shorter chains, because the probability of a given chain undergoing termination or other side reactions increases with an increase in the cumulative time spent in the active state63). In addition, the effect of new chains generated from AIBN throughout the polymerization increases with decreasing GO-RAFT content. The miniemulsion polymerization generated particles of approximate diameters 150–180 nm, and TEM analysis revealed core-shell type morphology with lighter (shell) and darker regions (core) (Fig. 5). The darker regions would be due to the polymer-modified GO (the GO sheet size was significantly smaller than the particle diameter).
Jiang et al.92 modified GO with various polymers (styrene, NIPAM, N,N,-dimethyl acrylamide, methyl acrylate, and t-butyl acrylate) by use of a tandem approach involving “simultaneous” coupling of the RAFT agent S-4-(trimethoxysilyl)benzyl S'-propyltrithiocarbonate via the R-group and RAFT polymerization initiated by AIBN at 60 °C in DMF in a one-pot procedure. Reasonable control/livingness was observed, with brushes of Mn ≈ 4000–12,000 g/mol and Mw/Mn ≈ 1.3. Block copolymers were attached to the GO using the same approach, by replacing the low molecular weight RAFT agent with the corresponding polystyrene RAFT agent. These authors also used the same methodology with the Z-functionalized RAFT agent S-methoxycarbonylphenylmethyl S′-3-(trimethoxysilyl)-propyltrithiocarbonate (i.e., attaching the RAFT agent to GO via the Z-group), which led to lower grafting density, lower Mn and lower Mw/Mn than using the corresponding R-functionalized RAFT agent.
Click Chemistry and Related Approaches
To date, the vast majority of reports dealing with the use of CLRP to introduce polymer brushes onto graphene/GO are concerned with grafting-from techniques. There exists a relatively large body of work on the use of a variety of covalent and noncovalent grafting-to approaches,43 but most reports deal with polymers prepared by means other than CLRP (thus falling outside the scope of the present Highlight).
There are to date only two reports93, 94 on the use of click chemistry95–97 to attach polymer chains prepared by CLRP to GO. Sun et al.93 covalently linked azido-terminated polystyrene (prepared by ATRP; Mn = 4600 g/mol; Mw/Mn = 1.04) to GO sheets using Cu(I)-catalyzed 1,3-dipolar azide-alkyne cycloaddition in DMF at room temperature, after functionalization of GO with propargyl alcohol. Dispersing the polystyrene-modified sheets in THF resulted in a much darker solution than the original GO, indicating partial restoration of the π-network within the carbon structure of the GO during the modification process.98 AFM revealed the GO sheets to have a thickness of 1.4 nm, which increased to 2.6 nm on polymer functionalization. TGA analysis showed that the polystyrene content was 20 wt %, which corresponds to ∼0.07% of the carbon atoms of the GO basal plane being functionalized. Pan et al.94 used a very similar approach to prepare GO functionalized with NIPAM), also relying on synthesis of azido-terminated polymer (initially prepared by ATRP; Mn = 4700 g/mol, Mw/Mn = 1.11).
Deng et al.71 functionalized GO sheets by use of grafting-to of poly(NIPAM) (Mn = 3600 g/mol; Mw/Mn = 1.26) prepared separately by SET-LRP. The grafting-to step was based on atom transfer nitroxide radical coupling,99–101 which involved initially functionalizing the GO sheets with the nitroxide 2,2,6,6-tetramethylpiperidinyl-1-oxy (TEMPO) by acylation reaction between OH-TEMPO and acyl chloride groups at the edges of the GO sheets (obtained by reaction of GO with SOCl2). The introduction of TEMPO groups was accompanied by partial reduction of GO, as evidenced by a change in color from brown to dark, increased thermal stability, and Raman spectroscopy. Activation of poly(NIPAM) with Br ω-end groups by use of CuBr/PMDETA (ATRP activation step) generates polymer radicals that are subsequently trapped by the nitroxide covalently attached to GO, resulting in formation of GO-poly(NIPAM). Dark ball-like structures appearing in GO-poly(NIPAM) TEM images combined with an 8-nm height increase in the corresponding AFM samples were used as evidence for successful immobilisation of poly(NIPAM) chains. The grafting efficiency of poly(NIPAM) was estimated as 34% based on XPS and elemental analyses of GO-TEMPO and GO-poly(NIPAM). The lateral dimensions of GO-TEMPO were ∼800 nm, to be compared with 150–400 nm for GO-poly(NIPAM), probably as a result of sonication. The GO-poly(NIPAM) sheets showed good dispersibility in a wide range of polar and nonpolar solvents (e.g., water, DMF, THF, and n-hexane).
He and Gao102 used nitrene chemistry to functionalize GO according to a grafting-to approach that entailed functionalization with polymer chains and (partial) reduction of GO to graphene in a single step. GO and functional azides were mixed in N-methyl-2-pyrrolidinone at 160 °C, resulting in formation of highly reactive nitrenes (from heating of azides), which undergo cycloaddition with CC bonds of GO to form aziridine rings. Using this approach, a range of functionalized graphene sheets were prepared, the polymeric examples of which comprised polystyrene and poly(ethylene glycol). Br-functionalized sheets were also used for surface initiated ATRP (i.e., grafting-from approach). According to AFM analysis, the lateral dimensions of the graphene sheets were several micrometers, and the polymer-functionalized sheets displayed minimal stacking.
Nitroxide-Mediated Radical Polymerization
NMP has thus far not been frequently used for modification of graphene/GO. Vuluga et al.103 prepared PMMA with Mn = 60,000 and Mw/Mn = 1.3 end-capped with a nitroxide (alkoxyamine chain end) by use of so called in situ NMP,104 which entails the use of a free radical initiator and a nitrone (as opposed to a radical initiator/nitroxide or alkoxyamine47, 48 in “normal” NMP). This polymer was subsequently reacted with GO sheets in a two-phase toluene/water system stirred vigorously at 50 °C. Grafting of the polymer chains was achieved via dissociation of the alkoxyamine CO bonds, thus generating PMMA radicals that were claimed to add to carbon–carbon double bonds of GO. This reaction was carried out in the presence of NaBH4, thus simultaneously achieving reduction of GO to graphene. Shen et al.105 prepared polystyrene-TEMPO by NMP, and subsequently added this species and acrylamide to reduced GO sheets dispersed in DMF/water (9:1) at 125 °C to generate poly(styrene-b-acrylamide) brushes, presumably also via a mechanism similar to that claimed by Vuluga et al.103
To date, there are only two articles reporting the use of CLRP in conjunction with a noncovalent approach to attach polymer chains to graphene or GO. Liu et al. used π–π stacking to noncovalently attach pH84 and temperature85 responsive pyrene-terminated polymers to reduced GO sheets. Pyrene-functionalized RAFT agents (with the pyrene moiety as part of the R-group) were synthesized, and 2-N-dimethylamino ethyl acrylate (DMAEA),84 acrylic acid,84 and N-isopropylacrylamide85 (NIPAM) were polymerized in solution, respectively, to the respective pyrene-terminated polymers with Mw/Mn < 1.2 and Mn = 6800–10,000 g/mol. The polymers were dissolved in an aqueous dispersion of reduced GO sheets followed by sonication and removal of excess polymer. Successful syntheses of hybrid materials were evidenced by ATR-FTIR and XPS analyses. In the case of poly(NIPAM), the obtained sheets (lateral dimensions ranging from a few hundred nm to in excess of 1 μm) were ∼5 nm thick (AFM), consistent with polymer being attached to both sides of the sheet (Scheme 2). This poly(NIPAM) hybrid material also exhibited temperature dependent dispersibility properties in aqueous media. Based on TGA, the polymer content was found to be in excess of 93 wt % in all three cases.
The successful isolation of graphene nanosheets in 20041 has attracted enormous attention due to the material's extraordinary physical properties. One of the limiting factors in regard to exploitation of graphene in a range of applications is that single graphene platelets tend to undergo agglomeration due to strong π–π and Van der Waals interactions. One of the strategies to overcome this problem, and to increase graphene compatibility with a receiving host matrix, is to modify graphene (or GO) with polymer brushes. In regard to polymeric nanocomposite materials, the addition of graphene/GO to polymer matrices has enormous potential for producing materials which uniquely combine the properties of the graphene/GO and the polymer host. However, to fulfil the potential of such nanocomposites, a high degree of dispersion of graphene/GO as individual two-dimensional sheets in the polymer matrix is essential.
As outlined in the present review, numerous approaches exploiting CLRP have been adopted to modify graphene (or GO) with polymer brushes, and these can be grouped into grafting-from and grafting-to techniques. CLRP forms an integral part of the synthetic strategy, either when growing chains from a suitably modified graphene/GO surface (grafting-from) or when attaching chains (that have been preformed by CLRP) to graphene/GO sheets. While in general superior control and grafting architecture can be achieved using grafting-from techniques, grafting-to can be a better alternative in regard to preservation of the intrinsic properties of the target material. Moreover, the grafting-to approach is generally more versatile in regard to the type of polymer than can be attached to the surface, as one is not restricted by the polymerization mechanism (grafting-from is mainly used in connection with radical polymerization, almost always CLRP). As such, condensation polymers as well as functional polymers are most suitably attached to graphene/GO via grafting-to methodology. A major disadvantage of the grafting-to approach is that steric crowding can impose an upper limit on the degree of modification.
The hitherto used methods of functionalizing graphene/GO with polymer brushes can also be categorized based on covalent and noncovalent attachment of the polymer chains. Covalent attachment includes the grafting-from approaches involving CLRP, as well as various other approaches utilizing, for example, click chemistry. Noncovalent approaches are exemplified by the use of, for example, π–π interactions between graphene/GO sheets and appropriately end-functionalized polymers, van der Waals forces, or ionic interactions. Covalent attachment of polymer chains to graphene may result in disruption of the conjugated structure of graphene, which will have a negative impact on physical, optical, and electronic properties. Noncovalent attachment of polymer brushes can instead be plagued by issues with control as it can be challenging to control the inherently relatively unstable assembly.
As outlined in this Highlight, there has been recent significant progress in the field of modification of graphene/GO with polymer brushes exploiting the advantages of CLRP, and there is currently intense research underway to further expand the use of graphene in various applications. The work reported to date has cleared the first hurdle by demonstrating that polymer chemists/material scientists are now equipped with the necessary knowledge and tools to modify graphene/GO nanosheets with a wide range of polymers, thus largely overcoming the challenge set out in the opening paragraph of this Conclusions section. One of the future challenges will be to identify the specific structures of these hybrid materials that are most suitable for a specific application (e.g., in polymer/graphene nanocomposite materials), and hence be able to tailor-make polymer–graphene materials based on detailed knowledge of their complex structure–property relationships.
Per B. Zetterlund is grateful for an ARC Future Fellowship and strategic funding from The University of New South Wales. Albert Badri gratefully acknowledges an Australian Postgraduate Award. The authors thank Stuart Thickett (CAMD and UNSW) for preparing the artwork for the Graphical Abstract.
Albert Badri, born in Tehran, finished his tertiary studies in applied chemistry in Tehran. He started his professional career in the Oil and Gas industry by collecting valuable experiences in process operation and facilities used in oil terminals. After a few years working in industry, he went back to school to pursue his higher education in polymer science and composite materials. He received a master degree from the University of Sheffield with a major project in surface modification of materials assisted with RF plasma-enhanced polymer coatings. Being acquainted with surface science, he developed his skills working with sealants-based composites used in organic solar cell devices. However, thirst for future progress led him to pursue a Ph.D. in the area of surface modification of graphene via novel polymerization techniques for various applications. His interests are in surface modifications, characterization, polymeric nanocomposites, and polymer chemistry.
Michael R. Whittaker received his Ph.D. in polymer chemistry, after spending a number of years in industry, in 2000 from the University of Queensland for work investigating the molecular dynamics in swollen polymer networks using solid state NMR relaxation mechanisms. In 2001, he joined Bio-Layer Pty Ltd. as a Senior Research Scientist where he patented methodologies for the modification, synthesis and assembly of polymer interfaces to preferentially orientate antibodies. During this period, he also held an adjunct lecturer position within the University of New South Wales. In 2004, he joined the Australian Institute for Bioengineering and Nanotechnology (AIBN: http://www.aibn.uq.edu.au/) as Research Fellow. Since October 2008, he is the Research Manager for both the Centre of Advanced Macromolecular Design (CAMD: www.camd.unsw.edu.au) and the Australian Centre for Nanomedicine (ACN: www.acn.unsw.edu.au), University of New South Wales. The IP he has generated during his career has contributed to the formation of two spinout companies, Dendrimed Pty Ltd. (www.dendrimed.com) and MetalloTek Pty Ltd. (www.metalloteck.com). He has coauthored over 60 peer-reviewed research papers including six international patents in a diverse range of areas; toxicology of nanoparticles, synthesis of “smart” hybrid inorganic/organic nanomaterials, new applications of polymerization methods in polymer design, polymers for environmental remediation and novel applications of “click” chemistry in polymer synthesis.
Per B. Zetterlund graduated from The Royal Institute of Technology (KTH, Sweden) in 1994 and obtained his Ph.D. at Leeds University (UK) in 1998. He carried out postdoctoral research at Griffith University (Australia) and was appointed Assistant Professor at Osaka City University (Japan) in 1999. In 2003, he moved to Kobe University (Japan) where he was promoted to Associate Professor in 2005. He joined CAMD at The University of New South Wales (Australia) in 2009. He has published 110 peer-reviewed papers and two book chapters in the area of radical polymerization in homogeneous and heterogeneous systems, with recent research focusing on (controlled/living radical) polymerization in aqueous and carbon dioxide based dispersed systems for synthesis of well-defined polymer and nanoparticles. He is a member of the IUPAC Macromolecular Division (IV) Subcommittee on Modeling of Polymerization Kinetics and Processes and The International Polymer and Colloid Group and was awarded a Future Fellowship from the Australian Research Council in 2011.