Smart graphene dispersion stabilized by a CO₂-removable polymer

Abstract

Polymeric dispersants play a pivotal role in improving graphene solubility in common solvents; however, the presence of such a foreign dispersant may exert a negative influence on the intrinsic properties of graphene. Thus, it is particularly important and challenging to remove the dispersant when graphene is used in end applications. Here, we report a smart graphene dispersion by a CO₂-triggered removable diblock polymer – poly(ethylene oxide)-b-poly(N-(3-((3-((4,6-bis((3-(dimethylamino)propyl)amino)-1,3,5-triazin-2-yl)amino)propyl)(methyl)amino)propyl)methacrylamide) (PEA). Using absorption spectroscopy, Raman spectroscopy, XPS spectroscopy and TGA measurements, it was found that PEA can not only strongly interact with graphene to form a stable, concentrated aqueous dispersion, but it can also be removed from the graphene surface upon CO₂ treatment as the tertiary amino groups along the polymer chain can be protonated, thus diminishing the affinity for graphene. This study may offer a general strategy for the design of removable dispersants for nanomaterials.

---

Introduction

Since its discovery in 2004 using micromechanical cleavage on graphite,¹ graphene has attracted much attention due to its remarkable electrical, thermal and mechanical properties.^2–4 The unique characteristics of graphene are mainly associated with its individual or few-layer sheets. However, as-produced graphene sheets are always prone to form irreversible aggregates through strong van der Waals interactions between themselves, rendering poor dispersibility in common solvents.^5–7 As a result, aggregation of graphene not only impedes its production, storage and processing, but also lowers its performance in practical end uses.

To solve this problem, both non-covalent and covalent modifications have been developed to stabilize graphene in solvents,^5,8 especially in water due to its non-toxicity and low cost.^9,10 In comparison with covalent process, non-covalent functionalization is more preferable since it can effectively improve graphene dispersibility without damaging its inherent structure.^11,12 Accordingly, non-covalent modification of graphene with foreign stabilizers, such as surfactants and polymers, has become an overwhelming technique to convert graphene into real-world applications.^13–19 Nevertheless, the presence of foreign stabilizers in end applications is normally undesirable as it may exert obviously negative influence on the thermal conductivity, electronic and mechanical properties of graphene.^7,10,20,21

Generally, stabilizers are adsorbed onto graphene surface via π–π stacking or hydrophobic interaction that are generally invariant, thereby they are difficult to be removed away from graphene.⁵ For example, Han and coworkers²² employed a bifunctional molecule, pyrene-adamantane (Py-Ad), to disperse graphene in dimethyl formamide (DMF) followed by thorough washing, but they found 30 wt% of Py-Ad are still remained on graphene surface. In order to get rid of stabilizers after dispersing of graphene, Adamson et al.²⁰ utilized an equimolar mixture of benzene and hexafluorobenzene as a solvent to disperse graphene sheets, where the benzene-derived molecules form stacks driven by quadrupolar interactions on graphene surface. The boiling point of the mixed solvent is approximately 78 °C, thus it can be easily evaporated at moderate temperatures when the dispersion comes into use. With such a technique, graphene sponge with potential applications for catalysis and electronics could be fabricated. However, both benzene and hexafluorobenzene are expensive and seriously toxic, limiting large-scale production and subsequent use of the dispersion. Therefore, to obtain well-dispersed pure graphene for practical applications still remains a crucial challenge.

In fact, dispersing single-walled carbon nanotubes (SWNTs) also encounters similar problems, but fortunately, removable polymeric dispersants have been developed to overcome such deficiencies. Several conjugated polymers^23–26 were found to be capable of forming a tightly coiled conformation on SWNTs surface via intermolecular interactions, thereby resulting in stable dispersions. More importantly, the polymers can be further removed from the hybrids while external stimuli, including solvent polarity, temperature and light, are introduced to change their conformation. Consequently, relatively pure SWNTs can be separated from the dispersions. Nevertheless, in contrast to SWNTs, as a 2-D nanomaterial, graphene has a huge plane so that conjugated polymers are difficult to be winded and released. On the other hand, Chan-Park et al.²⁷ recently developed an alternative removable dispersant, chondroitin sulfate (CS-A) for SWNTs. They found that CS-A shows weak interaction with nanotubes so that it can be removed by washing with water. The key point of this kind of removable dispersant is to find the good balance between dispersion efficacy and subsequent removability, which depends on the affinity for graphene since too weak affinity cannot afford stable and concentrated dispersion, while too strong affinity is unable to confer the removability. To obtain dispersants with such precise affinity with carbon nanomaterials seems difficult, however, if the interaction between graphene and a dispersant can be switched “on” and “off” upon triggering by external stimuli, the removability of the dispersant from graphene might be achieved.

Previous studies^28,29 have demonstrated that amino-substituted triazine derivatives show strong affinity with graphite due to the synergetic interaction between graphite and triazine core as well as the amino groups. On the other hand, amines are widely known as CO₂-sensitive compounds that can be reversibly reacted with CO₂ in wet environment,^30–35 and the formed ammonium cations could diminish the affinity between graphite and triazine. In other words, CO₂ gas may be capable of switching the interaction between graphene surface and amino-substituted triazine derivatives. Inspired by this fact, a melamine derivative, N-(3-((3-aminopropyl)(methyl)amino)propyl)-N,N-bis(3-(dimethylamino)propyl)-1,3,5-triazine-2,4,6-triamine (ANME), was prepared and incorporated into poly(ethylene oxide) (PEO) chain to afford a CO₂-switchable diblock copolymer PEO-b-PANME (PEA, Scheme 1). Then, the stabilizing ability of PEA for graphene in water was investigated, and the CO₂-controlled removability of PEA from graphene surface was examined. It was demonstrated that PEA can be adsorbed on graphene surface and facilitates the formation of dispersion. Moreover, it can also be removed upon the CO₂ treatment and leads to the aggregation of graphene.

Scheme 1 General strategy for the synthesis of diblock copolymer PEO-b-PANME.

Experimental

Materials

Cyanuric chloride (99%), 3,3′-diamino-N-methyldipropylamine (96%), 3-(dimethylamino)-1-propylamine (99%), ethyl-2-bromoisobutyrate (98%), poly(ethylene glycol)methyl ether (mPEG-2000, M_n ≈ 2000, flakes), methacrylic acid N-hydroxysuccinimide ester (MASI, 98%), and CuBr (98%) were purchased from Aldrich and used as received. The other reagents and solvents with A.R. grade were obtained from Shanghai Chemical Reagent Co., Ltd., China. The ligand N-(pyridin-2-ylmethylene)propan-1-amine used in polymerization was prepared following a previously-reported procedure.³⁶

Reduced graphene oxide (rGO, purity > 95 wt%; diameter: 0.5–3 μm) was kindly provided by Timesnano (Chengdu, China). CO₂ (≥99.998%) and N₂ (99.998%) were used as received. The deionized water (conductivity, κ = 12.8 μS cm⁻¹) used throughout this study was obtained from an ultrapure water purification system (CDUPT-III type, Chengdu Ultrapure Technology Co., Ltd., China).

Characterization

¹H and ¹³C NMR spectra were recorded at 25 °C on a Bruker AV300 NMR spectrometer at 300 and 75 MHz, respectively. Chemical shifts (δ) were reported in parts per million (ppm) with reference to the internal standard protons of tetramethylsilane (TMS). IR spectra were registered on a Nicolet MX-1E FTIR spectrometer in the scanning range of 4000–400 cm⁻¹ using KBr pellet method. ESI-HRMS spectra were obtained with the Bruker Daltonics Data Analysis 3.2 system.

Molecular weight and molecular weight distribution of the polymers were determined by gel permeation chromatography (GPC) system (THF, HLC-8320) equipped with a refractive index detector, using TSK gel super HZM-M 6.0 × 150 mm and TSK gel SuperHZ3000 6.0 × 150 mm as columns with THF eluent at a flow rate of 0.6 mL min⁻¹ at 40 °C. Monodisperse polystyrene was used as standard to generate the calibration curve.

UV-vis absorption spectra were obtained on a computer-manipulated double-beam UV-vis spectrophotometer (UV-4802, Unico, China) operated at a resolution of 0.5 nm at 25 °C over the wavelength range of 190–700 nm. Raman spectra were registered on LabRAM HR Raman Microscope (HORIBA Scientific, France) using He–Ne laser at 633 nm as the light source.

TEM observation was conducted on a JEM-100CX TEM instrument (JEOL Ltd., Japan) with an accelerating voltage of 80 kV. The specimens for TEM characterization were prepared by placing a drop of graphene dispersion on cooper grids that coated with carbon and dried at room temperature. XPS spectra were recorded on a Kratos XSAM800 XPS system (Kratos Ltd.). The parameters used are as follows: analyser mode, FAT; energy range, X1; exciting source: A1. The CASA XPS program with a Shirley background and Gaussian–Lorentzian mix function was used to analyze the XP spectra quantitatively. TGA measurements were performed on a 299-F1 thermal analysis system (NETZSCH, Germany). Samples were heated in flowing N₂ (50 mL min⁻¹) from room temperature to 800 °C at a heating rate of 10 °C min⁻¹. All the samples were dried in a vacuum oven at 60 °C for 12 h prior to test in order to remove solvents inside.

Synthesis of ANME

In a 100 mL three-neck round bottom flask, 3.68 g (20 mmol) cyanuric chloride was dissolved in 30.0 mL deionized water at 0 °C followed by gradual addition of 4.08 g (40 mmol) 3-(dimethylamino)-1-propylamine in 30 min. The reaction mixture was stirred at 60 °C for 5 h and then was slowly added into 4.35 g (30 mmol) 3,3′-diamino-N-methyldipropylamine at 85 °C. Afterwards, the mixture was stirred for another 5 h and 2.40 g (60 mmol) NaOH were added. One hour later, the reaction was stopped and cooled down to room temperature. Water was removed and 150 mL dichloromethane was added to dissolve the product in twice. Then the solution was collected and dried over anhydrous Na₂SO₄ overnight. The solvent was removed by reduced distillation to give the crude product which was washed with mixture of dichloromethane and toluene (v/v = 1 : 3) for twice and freeze-dried for 24 h to afford pure product in the form of yellow solid (5.13 g, yield: 60%). ¹H NMR (300 MHz, CDCl₃, Fig. S1†), δ/ppm: 1.71 (m, 8H, CH₂CH₂CH₂), 2.19 (s, 15H, NCH₃), 2.33 (t, 8H, NCH₂), 2.72 (t, 2H, NH₂CH₂), 3.34 (t, 6H, NHCH₂). ¹³C NMR (75 MHz, CDCl₃, Fig. S2†), δ/ppm: 27.7, 30.8, 39.1, 40.6, 42.2, 45.5, 55.7, 57.5, 166.0. ESI-HRMS (Fig. S3†): calcd: 425.3823 (M + H⁺); found: m/z = 425.3829. IR (KBr, Fig. S4†): ν = 1360 (s), 1545 (s), 1594 (s), 2790 (s), 2950 (s), 3248 cm⁻¹ (s).

Synthesis of PEO₄₅-Br

mPEG-2000 (10.00 g, 5.0 mmol), triethylamine (0.50 g, 5.0 mmol) and 100 mL dichloromethane were added into a 250 mL round bottom flask. Ethyl-2-bromoisobutyrate (1.38 g, 6.25 mmol) dissolved in 80 mL dichloromethane was slowly added into the flask in 2 h at room temperature, then the mixture was stirred for 24 h. After the reaction, the mixture was washed with deionized water for three times to remove triethylamine and then dried over Na₂SO₄ overnight. After removal of inorganic salt by filtration, the solution was concentrated to ∼10 mL and precipitated into 100 mL cold diethyl ether for twice. The product was collected and dried in vacuum oven at room temperature for 24 h, yielding a white solid (7.83 g; yield: 73%). M_n,GPC = 2705, M_n,NMR = 2149, M_w/M_n = 1.03, ¹H NMR (300 MHz, CDCl₃, Fig. S5†), δ/ppm: 1.90 (CCH₃), 3.37 (OCH₃), 3.64 (OCH₂), 4.32 (COOCH₂). IR (KBr, Fig. S6†): ν = 1100 (s), 1261 (w), 1360 (m), 1450 (w), 1630 (w), 2876 (s), 3431 cm⁻¹ (s).

Synthesis of PEO-b-PMASI

To a 25 mL one-neck round bottom flask, macromolecular initiator (PEO-Br, 1.08 g, 0.5 mmol), methacrylic acid N-hydroxysuccinimide ester (MASI, 0.92 g, 5 mmol), dimethylsulfoxide (DMSO, 3.0 mL), N-(pyridin-2-ylmethylene)propan-1-amine (76 mg, 0.5 mmol) and CuBr (70 mg, 0.5 mmol) were added together, followed by three freeze–pump–thaw cycles. The flask was reacted at 70 °C with magnetic stirring for 24 h. Then the resulting solution was immersed to liquid nitrogen in order to stop the radical polymerization. The solution was diluted to 50 mL dichloromethane and passed through a neutral alumina column to remove the copper catalysts. Afterward, the solution was collected and concentrated to ∼5 mL, and then precipitated into cold diethyl ether for twice. The product was collected and dried in vacuum oven at room temperature for 24 h, yielding a white solid (1.06 g, yield: 53%; M_n,GPC = 2802, M_w/M_n = 1.10). ¹H NMR (300 MHz, CDCl₃, Fig. S7†), δ/ppm: 1.22 (CCH₃ and CH₂C(CH₃)), 2.80 (COCH₂CH₂CO), 3.38 (OCH₃), 3.64 (OCH₂), 4.24 (COOCH₂). IR (KBr, Fig. S8†): ν = 1100 (s), 1220 (w), 1273 (w), 1335 (w), 1470 (w), 1730 (S), 2876 (s), 3443 cm⁻¹ (m).

From the ¹H NMR spectrum shown in Fig. S7,† comparing the integrals of the resonance peaks of OCH₂CH₂O of PEO (3.64 ppm) and COCH₂CH₂CO of MASI (2.80 ppm) led to the estimation of 4 units of PMASI, thus the diblock copolymer was denoted as PEO₄₅-b-PMASI₄ corresponding to a NMR-based M_n of 2881 g mol⁻¹.

Synthesis of PEO-b-PANME (PEA)

PEO₄₅-b-PMASI₄ (0.56 g, 0.19 mmol), ANME (0.81 g, 1.9 mmol) and DMSO (3.0 mL) were added into a 25 mL one-neck round bottom flask, followed by three freeze–pump–thaw cycles. The flask was reacted at 70 °C with magnetic stirring for 5 h. Then the resulting solution was cooled down to room temperature and precipitated into cold diethyl ether for twice. The product was collected and dried in vacuum oven at room temperature for 24 h to afford yellow solid (0.72 g; yield: 89%; M_n,GPC = 2798, M_w/M_n = 1.10). ¹H NMR (300 MHz, CDCl₃, Fig. S9†), δ/ppm: 1.05–1.22 (CCH₃ and CH₂C(CH₃)), 1.76 (CH₂CH₂CH₂), 2.21 (NCH₃), 2.36 (NCH₂), 3.40 (OCH₃ and NHCH₂), 3.64 (OCH₂), 4.20 (COOCH₂). IR (KBr, Fig. S10†): ν = 1100 (s), 1236 (w), 1335 (m), 1555 (s), 1666 (s), 2876 (s), 3272 (s), 3370 cm⁻¹ (s).

From ¹H NMR spectra, it was found that the peak at 2.80 ppm assigned to PMASI block was completely disappeared, whereas the peaks attributed to ANME were arising, indicative of all the MASI units have been replaced with ANME, thus the diblock copolymer was PEO₄₅-b-PANME₄ corresponding to an NMR-based M_n of 4185 g mol⁻¹. The difference of molecular weight obtained from ¹H NMR spectrum and GPC was likely owing to the adsorption of tertiary amino groups in PANME block onto the GPC column leading to an increase in retention time and thus lower detected molecular weight.^37–39

Sample preparation

2.0 mg of rGO powder was added to 2.0 mL PEA aqueous solution (10 mg mL⁻¹), then the mixture was sonicated for 90 min (100 W and 40 kHz) in a bath sonicator (KQ-100, Kunshan Ultrasound Instrument Company, China) at room temperature, followed by 10 min of centrifugation at 1500 rpm with a 80-2 centrifuge (Shanghai Medical Instruments Co. Ltd., China) to afford stable PEA–rGO dispersion.

Gas treatment for PEA–rGO dispersion. CO₂ at the flow rate of 3 mL min⁻¹ is bubbled into 5 mL as-prepared dispersion for 10 min, then graphene sheets begin to gradually aggregate and precipitate. N₂ at the flow rate of 10 mL min⁻¹ is streamed to purge CO₂ for 40 min followed by 30 min sonication will lead to homogenous dispersion again.

To prepare PEA–rGO composite for Raman, TGA and XPS analysis, the as-prepared dispersion was filtered through a PTFE micro-porous membrane (0.22 μm). The PEA–rGO hybrids left on the membrane were then washed with deionized water repeatedly and vacuum-dried for 24 h.

When preparing the PEA–rGO–CO₂ composite for Raman, TGA and XPS characterization, CO₂ was bubbled into the as-prepared dispersion at the flow rate of 3 mL min⁻¹ for 10 min at room temperature, and the dispersion was kept for three days. Then the mixture was filtered through the aforementioned membrane, and the residues left on the membrane were then washed with deionized water and vacuum-dried for 24 h to remove the remained water and CO₂.

Results and discussion

Design and synthesis of PEA

PEO is a typical cost-effective, nontoxic and water-soluble polymer, so that conjugated or hydrophobic groups can be incorporated into PEO to disperse nanocarbon materials in water.^19,40 The efficacy of these dispersants is largely dependent on the ratio of adsorption segments and PEO block: while the interaction between adsorption groups and graphene leads to the connection of the whole molecule and graphene, the PEO block forces graphene sheets to separate due to its huge volume. Too few adsorption moieties cannot provide sufficient interaction for the whole molecule to be adsorbed onto graphene surface, whereas too many adsorption groups may give rise to the easy aggregation of graphene sheets since the adsorption groups in one polymer chain may interact with more than one graphene sheet. In an earlier study, Lee and coworker¹⁹ used an aromatic amphiphile consisting of a dendronized PEO and four pyrene units to exfoliate graphite flakes into single- and double-layer graphene sheets in aqueous solution with concentrations up to 1.5 mg mL⁻¹ and stability up to 2 months. Later, Dichtel et al.^18,41 prepared a dispersant with three pyrene moieties, and they found that all the three pyrene groups can make the molecule strongly adsorb onto the basal plane of single-layer graphene. As an adsorption group, ANME behaviors similarly as pyrene, thereby three or four ANME units are anticipated to introduce into PEO chain to form a diblock copolymer stabilizer with good capability of dispersing graphene in water.

The structure of ANME is similar to that of previously reported CO₂-switchable molecule, MET, which can also interact with graphene.³¹ However, MET cannot be incorporated into polymer due to the lack of active group. In contrast, ANME bears a primary amino group, thus can be easily introduced through chemical reaction with other functional group. Thus, our strategy is to synthesize diblock copolymer poly(ethylene oxide)-b-poly(N-hydroxysuccinimide methacrylate) (PEO-b-PMASI) first, and ANME is then utilized to react with the precursor to afford target copolymer PEO-b-PANME.

PEO macro-initiator was prepared from the reaction of poly(ethylene glycol)methyl ether and ethyl-2-bromoisobutyrate, and it was further used for the synthesis of PEO-b-PMASI. Polymerization of 10 eq. of MASI monomer through ATRP initiated by 1 eq. of PEO macro-initiator using the Cu(I)Br/N-(pyridin-2-ylmethylene)propan-1-amine catalyst well introduced 4 MASI units to one PEO chain to afford PEO₄₅-b-PMASI₄, which was confirmed by its ¹H NMR spectrum. Then the target diblock copolymer PEA was obtained by the reaction of PEO₄₅-b-PMASI₄ with ANME in DMSO for 5 h under nitrogen atmosphere. From both ¹H NMR and IR spectra (Fig. S9 and S10, ESI†) of the final polymer, no characteristic peaks of PMASI were observed, but those of ANME did appear, indicative of full conversion of PMASI into PANME.

CO₂-swichability of PEA

As stated above, CO₂ is expected to switch the interaction between PEA and graphene through the reversible reaction of CO₂ and PEA, thereby the CO₂-switchability of PEA is investigated first.

Displayed in Fig. 1 is the conductivity and pH variation of PEA aqueous solution at 1 mg mL⁻¹ while sequentially bubbling CO₂ and N₂. When CO₂ was bubbled into the solution for 8 min, the conductivity increased rapidly from 64.3 to 224.7 μS cm⁻¹ accompanied by an evident drop of the pH value from 8.88 to 5.54, suggesting that a great deal of protonated species was generated in the solution. This was further confirmed by the ¹H NMR spectrum of PEA in D₂O (Fig. 2). The peaks of protons from the tertiary amino groups experienced obvious downfield shift upon the treatment of CO₂, indicating the protonation of the tertiary amino groups. Moreover, both the original conductivity and pH values were almost fully recovered after bubbling N₂, ascribing to the opposite deprotonation effect. This is in good agreement with the ¹H NMR result after the treatment of N₂. It is worth noting that such a procedure is still effective over three cycles, implying the good CO₂-switchability of PEA.

Fig. 1 The conductivity (A) and pH (B) of PEA aqueous solution (1 mg mL⁻¹) plotted versus time upon alternating CO₂ and N₂ stimuli.

Fig. 2 Comparison of ¹H NMR spectra of PEA in D₂O upon alternate CO₂ and N₂ treatment.

Next, the protonation degree of the tertiary amino groups of PEA was determined as it is a key parameter to govern the switchable interaction between the polymer and graphene surface. To this end, the pK_aH (pK_a of the protonated species) of PEA was determined by pH titration, and it was found to be 7.0 (see ESI† for details). In our previous study,³⁰ a quantitative relationship between pK_aH value and protonation degree (δ) of amines caused by CO₂ in aqueous solution was established:

δ = −180.3 + 39.6pK_aH (1)

According to eqn (1), the protonation degree of the tertiary amino groups in PEA in aqueous solution is calculated to be 96.9%. That is, almost all the tertiary amino groups are protonated in the presence of CO₂.

Graphene dispersion stabilized by PEA

The capability of PEA to disperse reduced graphene oxide (rGO) in water was then investigated. For the purpose of comparison, rGO dispersibility was first examined in pure PEO (the hydrophilic block of PEA) aqueous solution. As shown in Fig. 3A, rGO cannot be dispersed even with sonication for 2 h. On the contrary, it can be homogeneously dispersed in PEA aqueous solution, and the resultant dispersion appears dark black and can stand stable almost one month without noticeable sediment. TEM observation shows that most of the graphene sheets have been well dispersed without obvious aggregation (Fig. 3B). The thin, layered structure of separated sheet is distinctly revealed by the homogeneous contrast over a large area.

Fig. 3 (A) Snapshot of rGO dispersed in PEO and PEA aqueous solution and (B) a representative TEM image of rGO sheet stabilized by PEA.

To unveil the mechanism of this rGO dispersion, absorption spectroscopy was employed to investigate the interaction between graphene and PEA since it is highly sensitive to their mutual interactions.¹⁹ As exhibited in Fig. 4, PEA shows strong absorption between 200 and 230 nm, which is derived from the melamine ring of ANME moieties.³⁰ The maximum absorbance of the copolymer solution in the absence of rGO is 215 nm; however, it was red-shifted to 216.5 nm when mixed with rGO, implying extended π-conjugation of the ANME groups in the presence of graphene, which distinctly demonstrates that the existence of interaction between PANME segment and graphene.

Fig. 4 UV-vis spectra of PEA aqueous solution in the absence and presence of rGO.

To gain further insight into the mechanism forming graphene dispersion from PEA, Raman spectroscopy was employed to evaluate the noncovalent modification of PEA on graphene surface. As depicted in Fig. 5, both the original rGO and PEA–rGO composite show peaks at ∼1340 and 1596 cm⁻¹. It has been reported that in Raman spectra of graphene-based materials, the peak at ∼1330 cm⁻¹ is assigned to the A_1g breathing mode of the disordered graphite structure (the D band) and the peak at ∼1590 cm⁻¹ is ascribed to the E_2g structural mode of graphite (the G band).^42,43 While the G band reflects the structure of the sp² hybridized carbon atoms, the D band is due to defect sites in the hexagonal framework of the graphite materials. Thus, the intensity ratio of the D to G band (I_D/I_G) can be used to evaluate the extent of defect sites of graphene-based materials.^44,45 The higher I_D/I_G suggests more defect sites inside. If conjugated structures bound to the basal plane of graphene and cover the defect sites, then the intensity of the D band will decrease and leads to a drop of I_D/I_G. The I_D/I_G of untreated rGO is calculated to be 1.05 from the Raman spectrum in Fig. 5, implying the presence of some defect sites in rGO. However, the ratio decreases to 1.01 for PEA–rGO composite. Such an obvious drop of I_D/I_G ratio indicates that the ANME moieties of PEA are adsorbed on rGO surface and shield the defect sites. Hence, both UV-vis and Raman analysis clearly demonstrate that PEA interacts with rGO forming PEA–rGO complex.

Fig. 5 Raman spectra of original rGO, PEA–rGO composites prepared from PEA–rGO dispersion in the presence and absence of CO₂.

Moreover, X-ray photoelectron spectroscopy (XPS), an effective technique for surface analysis, was used to characterize the adsorption of PEA on graphene surface. The results are given in Fig. 6 and Table 1. From the spectra, one can find that the O element was observed at 534.9 eV for the original rGO, suggesting not all the oxygen-containing groups of graphene oxide have been reduced, but the residual groups are not sufficient to confer good dispersibility of the whole rGO sheet in water. The atomic C/O ratio is calculated to be ∼8.5 from the data shown in Table 1. Furthermore, the N element was found at 401.6 eV with relative amount of 5.82% in the PEA–rGO composite, accompanied with decrease of C amount and increase of O amount, ascribing to the higher content of O element in PEA. In short, XPS results provide direct evidence of the presence of PEA on graphene surface.

Fig. 6 XPS spectra of original rGO, and PEA–rGO composites prepared from PEA–rGO dispersion with and without CO₂.

Table 1 XPS results of original rGO, PEA–rGO composites prepared from PEA–rGO dispersion with and without CO₂

Atom	Atomic concentration (%)
	rGO	PEA + rGO	PEA + rGO + CO2
C	89.44%	82.08%	83.24%
O	10.56%	12.10%	13.29%
N	0	5.82%	3.47%

---

To precisely calculate the polymer content of PEA–rGO hybrids, thermogravimetric analysis (TGA) was employed on the complex, and both PEA and untreated rGO were measured as references. From the TGA curves displayed in Fig. 7, it can be seen that PEA became unstable with increasing temperature and almost completely decomposed when temperature reached 800 °C. Its main weight loss was found between 200 and 450 °C in two stages: the weight loss from 200 to 350 °C is assigned to the decomposition of PANME, and that from 350 to 450 °C results from the degradation of PEO. Moreover, the weight loss of PEA–rGO composite was found to be 34.0% when temperature increased to 800 °C, which is higher than that of original rGO (22.5%), owing to the decomposition of PEA. According to the above TGA data, the content of PEA in the composite is calculated to be 14.3%.

Fig. 7 TGA curves for PEA, original rGO, and PEA–rGO composites obtained from PEA–rGO dispersion in the presence and absence of CO₂.

Based on the above results, it is reasonable to speculate that the PANME segment in PEA interacts with graphene, leading to the whole molecule strongly adsorbs onto graphene surface; meanwhile, the PEO block forces the graphene sheets to separate due to its huge volume, thereby affording homogenous dispersion. This particular dispersing mechanism may give concentrated and highly stable graphene dispersion, thus the dispersing ability of PEA for graphene was investigated.

The graphene content in dispersion was determined by UV-vis absorption following a previously-reported procedure.⁴⁶ First of all, the absorption coefficient (α) of PEA–rGO dispersion was obtained to be 3660 mL mg⁻¹ m⁻¹, then loading of graphene stabilized by different concentration of PEA aqueous solution was examined and shown in Fig. 8. One can find that graphene concentration depends on the polymer concentration. The graphene concentration can achieve 0.54 mg mL⁻¹ when polymer concentration is only 1 mg mL⁻¹, but further increasing polymer concentration cannot give a significant rise of graphene concentration. Nevertheless, we observed that the dispersion with higher polymer concentration could keep stable for longer time. For instance, graphene dispersion stabilized by 1 mg mL⁻¹ PEA solution can only stand stable for 5 days, whereas increasing PEA concentration to 20 mg mL⁻¹ will impart sufficient steric hindrance to offset the interactions between graphene sheets. These results show PEA is an effective and efficient dispersant for graphene, which can afford stable, concentrated graphene aqueous dispersion.

Fig. 8 Effect of polymer concentration on graphene concentration. The inset is optical absorbance at 660 nm as a function of graphene content in PEA solution. A Lambert–Beer behaviour is shown, with an absorption coefficient α of 3660 mL mg⁻¹ m⁻¹.

CO₂-induced removal of PEA

As mentioned above, graphene can be stabilized by PEA in aqueous solution due to the interaction between graphene and PANME block in PEA, and PANME can be protonated by CO₂ in the presence of water. Therefore, it is natural to examine whether CO₂ is able to control the interaction between polymer PEA and graphene and then to confer the removability of the polymeric dispersant from graphene surface. UV-vis spectra was employed to determine the interactions between PEA and rGO upon the treatment of CO₂. As displayed in Fig. 9, the polymer showed the same maximum absorbance at 215.5 nm both in the absence and presence of rGO after bubbling CO₂, demonstrating that no interaction between CO₂-protonated PEA and rGO; as a result, the adsorbed polymer will be gradually released from the complex and graphene sheets will slowly aggregate and precipitate. As exhibited in Fig. 10, most of the graphene were observed to precipitate in the bottom of the vial after introduction of CO₂ for 10 days, while the untreated dispersion was still dark black and stable.

Fig. 9 UV-vis spectra of PEA solution and PEA–rGO dispersion in the presence of CO₂.

Fig. 10 Snapshots of PEA–rGO dispersion with and without CO₂.

In order to directly “see” the CO₂-controlled release of the polymer from graphene surface, dialysis experiments were performed (see ESI† for details). Two PEA–rGO dispersions with and without CO₂ were dialyzed against deionized water under the same condition, and the absorbance at 215 nm of the water was recorded at given intervals to monitor the released polymer. As shown in Fig. 11 where the absorbance plotted versus time, one can find that the polymer can be released from both of the two graphene dispersion with high speed at the beginning; however, the release rate of polymer from CO₂-treated dispersion is much higher than that of untreated dispersion with increasing time.

Fig. 11 Dialysis of PEA–rGO dispersion with and without CO₂ against deionized water.

Moreover, the total released amount of the polymer from CO₂-treated dispersion is also higher than that of original dispersion, all of which distinctly shows CO₂ facilitates the removal of the dispersant from the complex.

To examine the removability of PEA from graphene surface upon CO₂ treatment, PEA–rGO–CO₂ composite was prepared and then characterized with XPS, TGA tests and Raman spectroscopy. As depicted in Fig. 6, the peak at 393.6 eV assigned to N atom is still observed but its relative amount is reduced to 3.47%, indicative of the dispersant has not been completely removed. Furthermore, the weight loss of this composite at 800 °C is found to be 27.5% (Fig. 7), between the corresponding weight loss of original rGO and PEA–rGO composite, which corroborates the XPS results. The residual polymer content in PEA–rGO–CO₂ composite is calculated to be 5.2%, suggesting that two thirds of PEA molecules were removed from graphene after the treatment of CO₂. In addition, the I_D/I_G ratio of PEA–rGO–CO₂ is fully recovered to the original value of rGO, again suggesting the CO₂-induced removal of PEA from graphene. Although PEA cannot be completely removed from graphene after the treatment of CO₂, it is the first example of removable polymeric stabilizer for graphene and it is hoped that this approach can offer a general strategy to deign removable dispersants for nanomaterials.

In addition, it is worth noting that PEA can be restored from its CO₂-protonated product while bubbling N₂ to purge CO₂, and it can be re-adsorbed onto graphene surface to form homogenous dispersion again, as shown in Fig. 10. Therefore, the removed polymer can be deprotonated and reused to disperse graphene again, which will largely reduce the cost.

Conclusions

To sum up, we have demonstrated a smart graphene dispersion by a CO₂-controlled removable polymeric dispersant in water. The diblock copolymer, PEA, can not only adsorb onto graphene surface to afford stable, concentrated graphene dispersion, but also can be further removed from graphene upon CO₂ treatment owing to the protonated tertiary amino groups diminish the interaction between the polymer and graphene. The removal of dispersant from carbon nanomaterials after dispersion is particularly important for electronic, thermal and mechanical applications, thus this unique graphene dispersion can be expected to be used in these areas. Moreover, reversible aggregation/dispersion of graphene is accompanied with the removable/adsorption of PEA on graphene surface triggered by “green” gas, CO₂, so it would be desirable for the use of graphene in medical and biological applications, such as diagnostics, sensing, and drug delivery and release.

Acknowledgements

This work was financially supported by the National Natural Science Foundation of China (21273223), and the open funding of State Key Laboratory of Polymer Materials Engineering (sklpme2014-2-06).

Notes and references

1. K. S. Novoselov, A. K. Geim, S. V. Morozov, D. Jiang, Y. Zhang, S. V. Dubonos, I. V. Grigorieva and A. A. Firsov, Science, 2004, 306, 666–669.
2. V. Georgakilas, J. N. Tiwari, K. C. Kemp, J. A. Perman, A. B. Bourlinos, K. S. Kim and R. Zboril, Chem. Rev., 2016, 116, 5464–5519.
3. B. Zhan, C. Li, J. Yang, G. Jenkins, W. Huang and X. Dong, Small, 2014, 10, 4042–4065.
4. P. T. Yin, S. Shah, M. Chhowalla and K. B. Lee, Chem. Rev., 2015, 115, 2483–2531.
5. V. Georgakilas, M. Otyepka, A. B. Bourlinos, V. Chandra, N. Kim, K. C. Kemp, P. Hobza, R. Zboril and K. S. Kim, Chem. Rev., 2012, 112, 6156–6214.
6. A. G. Hsieh, S. Korkut, C. Punckt and I. A. Aksay, Langmuir, 2013, 29, 14831–14838.
7. J. Texter, Curr. Opin. Colloid Interface Sci., 2014, 19, 163–174.
8. J. M. Englert, C. Dotzer, G. A. Yang, M. Schmid, C. Papp, J. M. Gottfried, H. P. Steinruck, E. Spiecker, F. Hauke and A. Hirsch, Nat. Chem., 2011, 3, 279–286.
9. S. Dutta, T. Kar, S. Brahmachari and P. K. Das, J. Mater. Chem., 2012, 22, 6623–6631.
10. M. Yi, Z. Shen, S. Liang, L. Liu, X. Zhang and S. Ma, Chem. Commun., 2013, 49, 11509–11061.
11. J. Q. Liu, J. G. Tang and J. J. Gooding, J. Mater. Chem., 2012, 22, 12435–12452.
12. D. Y. Cai and M. Song, J. Mater. Chem., 2010, 20, 7906–7915.
13. J. W. T. Seo, A. A. Green, A. L. Antaris and M. C. Hersam, J. Phys. Chem. Lett., 2011, 2, 1004–1008.
14. Y. T. Liang and M. C. Hersam, J. Am. Chem. Soc., 2010, 132, 17661–17663.
15. M. Lotya, P. J. King, U. Khan, S. De and J. N. Coleman, ACS Nano, 2010, 4, 3155–3162.
16. D. Ager, V. Arjunan Vasantha, R. Crombez and J. Texter, ACS Nano, 2014, 8, 11191–11205.
17. A. Tchernook, M. Krumova, F. J. Tölle, R. Mülhaupt and S. Mecking, Macromolecules, 2014, 47, 3017–3021.
18. J. A. Mann, J. Rodriguez-Lopez, H. D. Abruna and W. R. Dichtel, J. Am. Chem. Soc., 2011, 133, 17614–17617.
19. D. W. Lee, T. Kim and M. Lee, Chem. Commun., 2011, 47, 8259–8261.
20. A. J. Oyer, J. M. Y. Carrillo, C. C. Hire, H. C. Schniepp, A. D. Asandei, A. V. Dobrynin and D. H. Adamson, J. Am. Chem. Soc., 2012, 134, 5018–5021.
21. Y. C. Yang, W. Rigdon, X. Y. Huang and X. D. Li, Sci. Rep., 2013, 3, 2086.
22. S. Z. Zu, D. Zhou and B. H. Han, J. Nanosci. Nanotechnol., 2013, 13, 946–953.
23. T. Umeyama, K. Kawabata, N. Tezuka, Y. Matano, Y. Miyato, K. Matsushige, M. Tsujimoto, S. Isoda, M. Takano and H. Imahori, Chem. Commun., 2010, 46, 5969–5971.
24. Z. X. Zhang, Y. K. Che, R. A. Smaldone, M. A. Xu, B. R. Bunes, J. S. Moore and L. Zang, J. Am. Chem. Soc., 2010, 132, 14113–14117.
25. F. Lemasson, J. Tittmann, F. Hennrich, N. Sturzl, S. Malik, M. M. Kappes and M. Mayor, Chem. Commun., 2011, 47, 7428–7430.
26. S. Liang, Y. M. Zhao and A. Adronov, J. Am. Chem. Soc., 2014, 136, 970–977.
27. L. Y. Yan, W. F. Li, S. Mesgari, S. S. J. Leong, Y. Chen, L. S. Loo, Y. G. Mu and M. B. Chan-Park, Small, 2011, 7, 2758–2768.
28. J. D. Wuest and A. Rochefort, Chem. Commun., 2010, 46, 2923–2925.
29. V. Leon, M. Quintana, M. A. Herrero, J. L. G. Fierro, A. de la Hoz, M. Prato and E. Vazquez, Chem. Commun., 2011, 47, 10936–10938.
30. H. Yin, Y. Feng, H. Liu, M. Mu and C. Fei, Langmuir, 2014, 30, 9911–9919.
31. H. Yin, H. Liu, W. Wang and Y. Feng, Langmuir, 2015, 31, 12260–12267.
32. H. Yin, A.-L. Bulteau, Y. Feng and L. Billon, Adv. Mater. Interfaces, 2016 DOI:10.1002/admi.201500623.
33. J. Hu, G. Zhang, Z. Ge and S. Liu, Prog. Polym. Sci., 2014, 39, 1096–1143.
34. Q. Yan and Y. Zhao, Chem. Commun., 2014, 50, 11631–11641.
35. P. G. Jessop, S. M. Mercer and D. J. Heldebrant, Energy Environ. Sci., 2012, 5, 7240–7253.
36. D. M. Haddleton, M. C. Crossman, B. H. Dana, D. J. Duncalf, A. M. Heming, D. Kukulj and A. J. Shooter, Macromolecules, 1999, 32, 2110–2119.
37. X. Zhang, J. Xia and K. Matyjaszewski, Macromolecules, 1998, 31, 5167–5169.
38. X. Zhang and K. Matyjaszewski, Macromolecules, 1999, 32, 1763–1766.
39. S. Creutz, P. Teyssié and R. Jérôme, Macromolecules, 1997, 30, 6–9.
40. S. Z. Zu and B. H. Han, J. Phys. Chem. C, 2009, 113, 13651–13657.
41. J. Rodriguez-Lopez, N. L. Ritzert, J. A. Mann, C. Tan, W. R. Dichtel and H. D. Abruna, J. Am. Chem. Soc., 2012, 134, 6224–6236.
42. Y. L. Hsin, K. C. Hwang and C. T. Yeh, J. Am. Chem. Soc., 2007, 129, 9999–10010.
43. L. M. Malard, M. A. Pimenta, G. Dresselhaus and M. S. Dresselhaus, Phys. Rep., 2009, 473, 51–87.
44. J. C. Yinjie Kuang, X. Zheng, X. Zhang and Q. Z. a. C. Lu, Nanotechnology, 2013, 24, 8.
45. M. Zhang, R. R. Parajuli, D. Mastrogiovanni, B. Dai, P. Lo, W. Cheung, R. Brukh, P. L. Chiu, T. Zhou, Z. F. Liu, E. Garfunkel and H. X. He, Small, 2010, 6, 1100–1107.
46. Y. Hernandez, V. Nicolosi, M. Lotya, F. M. Blighe, Z. Y. Sun, S. De, I. T. McGovern, B. Holland, M. Byrne, Y. K. Gun'ko, J. J. Boland, P. Niraj, G. Duesberg, S. Krishnamurthy, R. Goodhue, J. Hutchison, V. Scardaci, A. C. Ferrari and J. N. Coleman, Nat. Nanotechnol., 2008, 3, 563–568.

---

Footnote

† Electronic supplementary information (ESI) available: Details of spectra of compounds and polymers, as well as additional results are displayed. See DOI: 10.1039/c6ra16634b

---