Chemical Tuning of Specific Capacitance in Functionalized Fluorographene.

Owing to its high surface area and excellent conductivity, graphene is considered an efficient electrode material for supercapacitors. However, its restacking in electrolytes hampers its broader utilization in this field. Covalent graphene functionalization is a promising strategy for providing more efficient electrode materials. The chemistry of fluorographene is particularly attractive as it allows scalable chemical production of useful graphene derivatives. Nevertheless, the influence of chemical composition on the capacitance of graphene derivatives is a largely unexplored field in nanomaterials science, limiting further development of efficient graphene-based electrode materials. In the present study, we obtained well-defined graphene derivatives differing in chemical composition but with similar morphologies by controlling the reaction time of 5-aminoisophthalic acid with fluorographene. The gravimetric specific capacitance ranged from 271 to 391 F g-1 (in 1 M Na2SO4), with the maximum value achieved by a delicate balance between the amount of covalently grafted functional groups and density of the sp2 carbon network governing the conductivity of the material. Molecular dynamics simulations showed that covalent grafting of functional groups with charged and ionophilic/hydrophilic character significantly enhanced the ionic concentration and hydration due to favorable electrostatic interactions among the charged centers and ions/water molecules. Therefore, conductive and hydrophilic graphitic surfaces are important features of graphene-based supercapacitor electrode materials. These findings provide important insights into the role of chemical composition on capacitance and pave the way toward designing more efficient graphene-based supercapacitor electrode materials.

Introduction

The demand for sustainability in the energy storage/conversion device industry places a high priority on achieving a high capacity/power density, environmental friendliness, and durability of these systems. To meet these requirements, supercapacitors have attracted interest from scientists worldwide, with the aim of combining a high electrical energy storage capacity, efficient energy conversion, and fast charge/discharge rates.[1−5] Supercapacitors store energy via charge accumulation at an electrode–electrolyte interface or via reversible redox processes between an electrode and electrolyte (electrical double-layer (EDL) capacitors and pseudocapacitors, respectively).[6−8] Pseudocapacitive nanomaterials utilize transition-metal compounds or conductive polymers, sometimes integrated into hybrids with carbon materials, such as graphenes, carbon fibers, and carbon nanotubes.[9−12] However, metal oxide nanostructures may easily collapse during charge/discharge processes, affecting their life-cycle duration.[13−15] Furthermore, the cost of transition metals, their limited abundance in natural resources, and the environmental impact of their excessive consumption are drawbacks against their extensive use. On the other hand, conductive polymers may exhibit low mechanical stability and deteriorated performance due to degradation, for example, during repeated charge/discharge processes, or low process ability, for example, due to low solubility.[16−18]

Graphene is a promising material for supercapacitor electrodes owing to its very high conductivity, inherent mechanical strength, and enormous theoretical specific surface area of 2675 m2 g–1, which could afford a specific capacitance of 550 F g–1.[19−22] However, the tendency for restacking of graphene sheets during electrode fabrication lowers the accessible surface area, disrupts the nanoporous structure, and significantly reduces the charge storage capacity compared with a graphene monolayer.[23] Noncovalent and covalent graphene functionalization approaches have been identified as exciting strategies for improving the properties of graphene-based electrode materials. Noncovalent functionalization can improve the porosity and enhance the accessible surface area. However, it does not alter the interface between the hydrophobic graphene surface and polar electrolytes. On the other hand, covalent graphene functionalization based on grafting of suitable organic molecules onto the graphene surface may prevent agglomeration of functionalized graphene single sheets and may broaden the interlayer spacing, allowing easy access and fast diffusion of electrolyte ions.[6,22,24,25] However, covalent functionalization incorporates sp3 carbon atoms into the sp2 graphene lattice, reducing the conductivity of graphene derivatives. Because of this, hydrophilic graphene oxide is not considered a suitable electrode material for supercapacitors and has to be reduced to reinstate its conductivity. This approach was successfully adopted by Caliman and co-workers, who synthesized amine-functionalized graphene oxide by microwave-assisted reactions using aromatic and nonaromatic amines, achieving specific capacitance values up to 290 F g–1 (in 1 M H2SO4, at a current density of 1 mA cm–2 using a three-electrode system).[26] Comparable results were achieved by Shi et al.[27] with aminopyrene-tetraone-modified reduced graphene, which showed specific capacitances of 327 and 77 F g–1 in a three- and two-electrode setup, respectively (in 1 M H2SO4 electrolyte at a current density of 0.5 A g–1) (see Table S1 (Supporting Information) for other systems). These findings suggest that a delicate balance between the degree of functionalization and chemical nature of the functional groups has to be achieved to prepare efficient electrode materials for supercapacitors. In addition, rather facile, upscalable, and cheap chemical approaches have to be adopted for graphene functionalization to keep the cost of the electrode material reasonable.

Covalent functionalization of graphene based on the chemistry of fluorographene (FG) is an attractive option because it provides an elegant method for grafting functional groups onto the graphene surface, that is, toward graphene derivatives. Graphite fluoride (GrF), the pristine material of FG, is a well-known covalent derivative of graphite.[28] The main advantages of FG/GrF are that (i) they have a well-defined structure, that is, a homogeneously decorated hexagonal lattice of sp3 carbons with fluorine atoms; (ii) they are easily accessible on the market because GrF is produced in quantities of tons as an industrial lubricant; and (iii) despite belonging to perfluorocarbons, the C–F bonds are susceptible to substitution reactions.[29,30] Another beneficial feature of FG chemistry is that substitution reactions are accompanied by reductive defluorination, leading to partial reestablishment of the sp2 network inside the honeycomb lattice, which can imprint conductivity into the prepared materials.[31] Although cyanographene and graphene acid prepared via FG chemistry were shown to have low resistances of 18 and 22 Ω, respectively, they displayed only moderate capacitance values of 97 and 86 F g–1, respectively (in 1 M H2SO4 and at 1 A g–1).[32] In contrast, grafting of a zwitterionic network onto the graphene lattice can generate materials with very high specific capacitance values.[33] Zwitterionic molecules have potent water retention that allows extensive hydration due to strong interactions between the charged groups and water molecules. These properties also facilitate the diffusion of electrolyte ions during charging/discharging cycles, boosting the adsorption capability of the functionalized graphene-based material.[34,35] Besides that, the charging mechanism of an electrode includes—almost always—a contribution from ion exchange (swap of ions having the same sign as the electrode with oppositely charged ions).[36] Thus, an enhanced ionic atmosphere around the zwitterionic groups and possible independent structuring of ions on each side of the electrode in the absence of an applied potential (i.e., electrolytes on both sides of the graphene electrode behave as independent entities without any mutual interference)[37] may subsequently result in improved electrochemical performance. However, detailed knowledge concerning the role of chemical functionality on the energy storage properties of graphene derivatives is needed to facilitate the design of more efficient electrode materials.

Rational design of versatile graphene-based materials by tuning the surface properties could tackle some of the major challenges in the future energy storage landscape. Such smart material designs may provide sufficient electroactive interfaces, which in turn could facilitate charge accumulation and high kinetics of electrolyte diffusion.[38] For this purpose, functionalization and extended π-conjugation should be finely balanced since these features critically affect the material’s performance as electrodes in supercapacitors. Functionalization of fluorographene has led to significant improvements regarding the capacitance,[39] but challenges still remain due to low rate capabilities.

Toward this direction, in the present study, a zwitterionic organic moiety was grafted via a one-step and upscalable procedure through the nucleophilic substitution reaction of FG with 5-aminoisophthalic acid (Niso) under mild reaction conditions. Successful covalent grafting of Niso via the amine group onto the graphene lattice was confirmed by a combination of characterization techniques. The functionalization degree (FD), that is, graphene coverage by the functional groups, was controlled by varying the reaction time. The electrochemical properties and specific gravimetric capacitance of all of the prepared few-layer graphene derivatives were thoroughly investigated; thus, the adopted procedure allowed evaluation of the role of chemical functionalization on the specific capacitance. A maximal capacitance of 391 F g–1 was achieved by a delicate balance between substitution of fluorine atoms by the Niso moiety and reductive defluorination. The results demonstrated that promising electrode materials for supercapacitors can be synthesized via the controllable chemistry of FG. The present work provides for the first time new insights into the role of the degree of chemical functionalization of graphene on the specific capacitance of the final graphene derivative. Finally, taking into consideration the upscalable and broad chemistry of fluorographene, as well as the evaluation of the products under real supercapacitor-operating conditions (missing from related studies),[40,41] the present findings may trigger further work in this field toward more efficient and durable electrode materials.

Experimental Section

Reagents and Materials

Graphite fluoride (GrF) (C/–F, 1:1.1), 5-aminoisophthalic acid (Niso) (94%), isophthalic acid (iso) (99%), poly(vinylidene fluoride) (PVDF), 1-methyl-2-pyrrolidone (NMP), methylene blue hydrate (≥97.0%), and boric acid (99.999%) were purchased from Sigma-Aldrich. Acetone (pure), ethanol (absolute), N,N-dimethylformamide (DMF), and triethylamine were purchased from Penta. Sodium hydroxide pearls were purchased from Lach-ner. All reagents were used as received without further purification. All stock solutions were prepared with ultrapure water (18 MΩ cm–1).

Synthesis of Niso Functionalized FG (FG/Niso-xh, x: 1, 3, 6, 24, 36, and 48 h)

First, 100 mg of GrF was dispersed in 5 mL of DMF and subjected to ultrasonication for 10 min (Bandelin Sonoplus, type UW 3200, probe VS70T). The flask containing the dispersion was then sonicated in a sonication bath for 2.5 h (Bandelin Sonorex, DT255H type, frequency 35 kHz, power 640 W, effective power 160 W). Afterward, 0.88 g of Niso was added to the flask, and it was sonicated for a further 1 h. Next, 1.350 mL of triethylamine was added, and the reagents were heated under stirring at 130 °C for x = 1, 3, 6, 24, 36, and 48 h. Finally, a black product was separated by centrifugation and purified by repeated centrifugal washings with DMF, water, and ethanol.

Synthesis of Iso-Treated FG (FG/iso-24h)

First, 100 mg of GrF was dispersed in 5 mL of DMF and subjected to ultrasonication for 10 min (Bandelin Sonoplus, type UW 3200, probe VS70T). The flask with the dispersion was then sonicated in a sonication bath for 2.5 h (Bandelin Sonorex, DT255H type, frequency 35 kHz, power 640 W, effective power 160 W). Afterward, 0.81 g of isophthalic acid (iso) was added to the flask, and it was sonicated for a further 1 h. Next, 1.350 mL of triethylamine was added, and the reagents were heated under stirring at 130 °C for 24 h. Finally, a black product was separated by centrifugation and purified by repeated centrifugal washings with DMF, water, and ethanol.

Synthesis of Reduced FG (rFG)

First, 100 mg of GrF was dispersed in 5 mL of DMF in a flask and subjected to stirring for 4 days, followed by 4 h of sonication in a sonication bath (Bandelin Sonorex, DT255H type, frequency 35 kHz, power 640 W, effective power 160 W), and finally ultrasonication with a probe for 10 min (Bandelin Sonoplus, type UW 3200, probe VS70T). Subsequently, the flask was heated under stirring at 130 °C for 3 days. Finally, a black product was separated by centrifugation and purified by repeated centrifugal washings with DMF, water, and ethanol.

Preparation of Sodium Borate Buffer Solution (pH ∼ 8.5)

Under stirring, 1.55 g of boric acid and 0.26 g of sodium hydroxide were dissolved in 25 mL of distilled water.

Electrochemical Characterization

Three-electrode and two-electrode cell configurations were used to evaluate the electrochemical properties. The three-electrode system was used to collect basic information about the capacitive response. In this system, a modified glassy carbon electrode (GCE) served as the working electrode, a platinum electrode was used as the counter electrode, and a Ag/AgCl electrode was used as the reference electrode. The GCE was modified with the sample by drop-coating: a 10 μL drop of a powder suspension (2 g L–1) was coated onto the GCE surface and allowed to dry at ambient temperature to form a thin film.

For the two-electrode cell configuration, PVDF (0.11 mg) was added to FG/Niso-24h powder (FG/Niso-24h/PVDF = 95:5 in weight, wt %) as a binder. Then, the powder with the binder was mixed into a paste using NMP as a solvent and was cast using a doctor blade onto 18 mm-diameter gold electrodes. Finally, the as-prepared electrodes were dried at 120 °C overnight. The total mass of both electrodes was 2.3 mg, and the surface was 2.2 cm2 per electrode. Two nearly identical (by weight and size) electrodes were assembled in a test cell (El-CELL GmbH, Germany) with an ion-porous separator (Whatman glass microfiber membrane). All electrochemical experiments were carried out with Metrohm Autolab PGSTAT128N (Metrohm Autolab B.V., The Netherlands) driven by NOVA software (version 1.11.2). The value of specific capacitance was calculated from galvanostatic charge–discharge (GCD) measurement, including the mass of the active material and the mass of the binder. The influence of the binder can be neglected due to its small wt %. Electrochemical impedance spectroscopy (EIS) measurements were conducted by applying an AC voltage with 5 mV amplitude over a frequency range 0.01 Hz to 10 kHz at open-circuit potential (OCP). All obtained data were fitted using a modified Randles circuit. Then, 1 M Na2SO4 was used as the electrolyte solution in all experiments.

A higher capacitance in three-electrode setups is expected because of the device architecture. In particular, only one electrode (working electrode) is tested with the materials under study, and it is necessary to consider that the applied voltage, as well as the charge transfer across one electrode, is different from that in the two-electrode setup. Moreover, the potential obtained from other sources (counter electrode) is not controlled/measured in the three-electrode setup, which also influences the overall performance. Finally, in the three-electrode setup, the voltage of the half-cell is measured, and the potential changes of the working electrode are measured independent of the changes that may occur at the counter electrode. Due to this fact, the voltage range in a three-electrode setup is half of that of the two-electrode cell. These differences between two- and three-electrode measurements are, for example, well documented in the review article by Stoller and Ruoff[42] and by others.[43]

Characterization Techniques

Fourier transform infrared (FTIR) spectra were recorded on an iS5 FTIR spectrometer (Thermo Nicolet) using the Smart Orbit ZnSe ATR accessory. Briefly, a droplet of an ethanolic dispersion of the test material was placed on a ZnSe crystal and left to dry and form a film. Spectra were acquired by summing 100 scans, using N2 gas flow through the ATR accessory.

X-ray photoelectron spectroscopy (XPS) was carried out with a PHI VersaProbe II (Physical Electronics) spectrometer using an Al Kα source (15 kV, 50 W). The obtained data were evaluated with the MultiPak (Ulvac-PHI, Inc.) software package.

Dynamic light scattering (DLS) and zeta-potential (ζp) measurements were performed with a Malvern ZetaSizer Nano instrument on aqueous dispersions of ∼0.1 and ∼0.03 mg mL–1 correspondingly.

Raman spectra were collected using a DXR Raman spectroscope (Thermo Scientific) equipped with a laser operating at a wavelength of 633 nm.

Thermogravimetric analysis (TGA) and differential thermogravimetric (DTG) analysis were performed using a SDT 650 thermal analyzer instrument (USA). Measurements were carried out under N2 flow (100 mL min-1). A temperature program from 40 to 1000 °C with a heating rate of 5 °C min–1 was used.

Water droplet contact angle measurements were performed in Drop Shape Analyzer—DSA30S, Krüss, Gmbh.

UV–vis absorption spectra were collected on a SPECORD S600 UV–vis spectrophotometer (Analytikjena) in the range of 250–750 nm. Methylene blue solutions were prepared in concentrations of 10–5–10–3 M. In 5 mL of each solution 10 μL of sodium borate buffer solution was added to adjust the pH at ∼8.5. FG/Niso-24h was weighed in vials (in amounts of 5–7 mg), and 5 mL of the above methylene blue solution was added in each vial. Mixing took place in a Multi bio RS-24 programmable rotator overnight. Centrifugation followed (Sigma 4–16k) at 1000 rpm (116 rcf) for 5 min. The UV–vis absorption spectra of the supernatants were collected. For these measurements, cuvette 100-QS, Suprasil, 10 mm purchased from Fisher Scientific was used.

Surface area and pore size analysis were performed by means of N2 adsorption/desorption measurements at −196 °C on a volumetric gas adsorption analyzer (3Flex, Micromeritics) up to 0.9626 bar. The sample was degassed under high vacuum (7 × 10–2 mbar) for 12 h at 110 °C, whereas high-purity (99.999%) N2 and He gases were used. The Brunauer–Emmett–Teller area (BET) was determined assuming a molecular cross-sectional area of 16.2 Å2 for N2 (−196 °C). The isotherms were further analyzed by means of nonlocal density functional theory slit pore kernels for N2.

The samples were also analyzed by scanning electron microscopy (SEM) on a Hitachi SU6600 instrument with accelerating voltage of 5 kV. Transmission electron microscopy (TEM) images were obtained on a JEOL 2100 instrument equipped with a LaB6-type emission gun operating at 200 kV. Scanning transmission electron microscopy-high-angle annular dark-field imaging analyses for elemental mapping of the products were performed with a FEI Titan HRTEM (high-resolution TEM) microscope operating at 200 kV. For these analyses, a droplet of a dispersion of the material in DMF with concentration of ∼0.1 mg mL–1 was deposited onto a carbon-coated copper grid and dried. Atomic force microscopy (AFM) images were obtained in the amplitude-modulated semicontact mode on an NT-MDT NTegra system equipped with a VIT-P AFM probe using freshly cleaved muscovite mica as a substrate.

Conductivity measurements were performed using an Ossila four-probe system operating in the current range from ±10 nA to ±150 mA (voltage range from ±100 μV to ±10 V). The fluorine-doped tin oxide (FTO) substrate was modified with the sample by the drop-coating method: a 150 μL drop of a powder suspension (5 g L–1) was coated onto the substrate and allowed to dry at ∼40 °C to form a thin film.

Computational Details

Molecular dynamic (MD) simulations were performed using Gromacs 4.5 software, the AMBER parm99 force field, and the SPC/E water model.[44,45] Lennard-Jones parameters for the carbon atoms in graphene were taken from the literature.[46] Three sheets of Niso functionalized graphene (∼40 × 40 Å2; with the same degree of functionalization as in the experiments) were placed into a cubic box (with 90 Å edge length) containing 1 M NaCl using Amber-adapted Aqvist parameters for Na+ and Smith and Dang parameters for Cl–.[47,48] Owing to the acid–base properties of the grafted functional group, three initial protonation states were considered: (i) fully protonated form (mimicking the situation at low pH), (ii) zwitterionic form (with deprotonated carboxylic and protonated amino groups), and (iii) deprotonated form (corresponding to basic conditions, i.e., high pH). Partial charges on the grafted Niso groups were derived according to the RESP procedure for grafting onto pyrene.[49] Periodic boundary conditions were applied in all three dimensions, and a 2 fs time step was used in all simulations. The cutoff radius for van der Waals interactions and the real part of electrostatics was set to 10 Å. Long-range electrostatic interactions were treated using the particle-mesh Ewald method. Bonds involving hydrogens were constrained using the LINCS algorithm.[50] Initially, the system was thermalized from 10 to 300 K and equilibrated at the final temperature (for 5 ns) under NPT ensemble conditions, coupled to the V-rescale thermostat with a 0.1 ps coupling constant and the isotropic Berendsen barostat with a reference pressure of 1 bar and 1.0 ps coupling constant.[51,52] The final production run was carried out in an NVT ensemble for 70 ns. Figures were rendered using PyMOL software.[53]

Results and Discussion

To verify the covalent grafting of Niso on a graphene surface via substitution of the F atoms of FG by amino groups of Niso[54] (Figure a), a complex characterization was performed. FTIR spectra of GrF and FG/Niso-24h are presented in Figure b. The strong band at 1204 cm–1 was attributed to the covalent C–F bond and the one at 1307 cm–1 to peripheral −CF2 groups in GrF.[55,56] In the FG/Niso-24h sample (Figure b), an intense peak was present at 1568 cm–1 corresponding to skeletal vibrations of graphitic regions, suggesting that defluorination and subsequent reduction of GrF as well as establishment of sp2 hybridization were triggered by DMF, as explained in the literature.[57,58] Peaks at 1451, 1388, and ∼1700 cm–1 were attributed to C=C stretching, C–N vibration, and −C=O (carboxyl) stretching, respectively.[30,59,60] The band corresponding to −C=O stretching was particularly intense in the case of FG/Niso-24h (Figure b), indicating the presence of a significant amount of carboxylic groups and successful grafting of Niso onto the FG surface. Furthermore, the presence of bands centered around 762 and 902 cm–1 corresponding to sp2 C–H bending patterns for aromatic compounds was indicative of meta disubstituted benzene (Figure b). Moreover, there was no doublet band in the region ∼3400–3500 cm–1 corresponding to a primary amine (Figure b), suggesting that Niso was covalently grafted onto the graphene surface via the amino group. In this region, only a broad band attributed to −OH and −NH was observed. For further investigation, a control experiment was performed in which isophthalic acid (iso) instead of Niso was employed under the same reaction conditions (see Experimental Section) to prepare FG/iso-24h. In this sample, the band attributed to −C=O stretching was absent, indicating that iso was not grafted to the graphene surface when defluorination of FG occurred (Figure S1, Supporting Information).

Figure 1

(a) Scheme showing FG reaction with 5-aminoisophthalic acid (Niso); hydrogens are not shown. (b) FTIR spectra of GrF and FG/Niso-24h. (c) C 1s XPS spectra of GrF and FG/Niso-24h. (d) N 1s XPS spectra of Niso and FG/Niso-24h.

The time-evolution characteristics of the FG/Niso derivative were monitored with XPS, showing that the content of F atoms decreased with time (Table ) up to 24 h of reaction and that the grafting of Niso maximized after about 36 h. The FTIR spectra were found to be rather time-insensitive (Figure S1, Supporting Information).

Table 1

Elemental Composition of GrF and FG/Niso-xh Samples (x: 1, 3, 6, 24, 36, and 48 h) Derived from XPS Analyses

	element content (atom %)
samples	C	N	O	F	C/F
GrF	44.1		0.2	55.7	0.8
FG/Niso-1h	71.7	4.6	13.0	10.7	6.7
FG/Niso-3h	76.3	5.4	14.4	3.9	19.6
FG/Niso-6h	77.2	5.5	14.5	2.8	27.6
FG/Niso-24h	77.6	5.8	14.7	1.9	40.8
FG/Niso-36h	77.2	5.9	15.0	1.9	40.6
FG/Niso-48h	78.5	5.3	14.3	1.9	41.3

The deconvoluted C 1s XPS spectra of GrF and FG/Niso-24h (Figure c) displayed six and five components, respectively, corresponding to carbon atoms in different functional groups. The peak located at ∼284.81 eV was attributed to graphitic C–C (sp2) bonds, whereas the peaks at 285.92, 286.86, 287.84, 288.39, 289.39, and 291.83 eV were attributed to carbon in C–C (sp3), C–N, C*–C–F, O=C–O, C–F, and C–F2 functional groups, respectively.[31,61,62] C 1s XPS spectra of all samples are presented in Figure S2 (Supporting Information). Increasing the reaction time caused an increase in the atomic percentages of C=C (sp2 C) and decrease of C bound to F atoms (Table ). The content of sp2 C reached a maximum, whereas that of the C–F group reached a minimum at 24 h, and at longer reaction times, the content of sp3 C increased significantly. The presence of C atoms bound in C–N groups and O=C–O groups indicated that the FD reached a maximum at 36 h. Furthermore, the grafting of Niso onto FG via the amine group was confirmed by N 1s XPS spectra (Figure d). These spectra showed that structural changes occurred in the amine group after the nucleophilic substitution reaction. Peaks at around 399.39 and 401.51 eV were assigned to primary amine and ammonium groups of Niso, respectively. In FG/Niso-24h, the shift of the peak from 399.39 to 400.25 eV implies the presence of secondary amines bound to the graphene surface and a dicarboxyl benzene moiety.[63]

Table 2

Atomic Percentage (atom %) of Various Groups in FG/Niso-xh Derivatives and the Parent GrF Obtained from Deconvolution of High-Resolution C 1s Core-Level XPS Spectra

samples	C–C (sp2)	C–C (sp3)	C–N	C*–C–F	O=C–O	C–F	CF2
	∼284.81 eV	∼285.92 eV	∼286.86 eV	∼287.84 eV	∼288.39 eV	∼289.39 eV	∼291.83 eV
GrF	0.7	1.1		3.2	0.8	74.6	19.6
FG/Niso-1h	58.4	15.9	6.0		8.0	11.7
FG/Niso-3h	60.8	17.6	7.0		9.5	5.1
FG/Niso-6h	62.1	17.7	7.2		9.5	3.6
FG/Niso-24h	62.5	17.8	7.5		9.6	2.6
FG/Niso-36h	61.5	18.0	8.0		10.0	2.5
FG/Niso-48h	61.3	20.1	6.7		9.4	2.5

The composition of the FG/Niso-xh samples as well as the covalent FD was investigated by TGA in combination with XPS. TGA and DTG graphs of Niso in N2 showed a weight loss of 4.46% at around 134 °C due to the release of adsorbed water (Figure a,b). At higher temperatures, decarboxylation and condensation reactions occur. The TGA and DTG curves of GrF in a N2 atmosphere (Figure a,b) showed an abrupt weight loss at around ∼612 °C due to defluorination. In the FG/Niso-24h sample (Figure a), there was a weight loss at around 403 °C (Figure b), which could be assigned to partial decarboxylation of the Niso functionalities. TGA and DTG measurements of all of the FG/Niso-xh samples are presented in Figures S3 and S4 (Supporting Information). Combining the XPS and TGA measurements, FDs of the FG/Niso-xh samples were estimated as 1.9, 2.7, 2.8, 3.0, 3.5, and 2.4 for x = 1, 3, 6, 24, 36, and 48 h, respectively. The XPS analysis revealed that this was accompanied by an increase of sp2 C up to 24 h. For longer times, that is, 36 and 48 h, sp3 C was increasing, reaching a maximum at 48h, and FD was significantly reduced at 48 h. In all cases, the notable FDs achieved are indicative of an effective reaction.

Figure 2

(a) TGA and (b) DTG analyses of Niso, the parent GrF, and FG/Niso-24h recorded in a N2 atmosphere. (c) Raman spectrum of FG/Niso-24h. (d) ζp-potential measurements as a function of pH of an aqueous dispersion of FG/Niso-24h and (e, f) water droplet contact angle measurements of GrF and FG/Niso-24h.

The above-mentioned analysis indicated an important FD, as is common for graphene derivatives prepared via FG chemistry.[64] This was also reflected in the Raman spectra, where two intense D and G bands appeared. It should be noted that pristine GrF is a Raman-silent material.[65] The intensity ratio ID/IG ∼1.17 of FG/Niso-24h was rather high (>1) and, alongside the considerable broadening of the bands (Figure c), indicated that the material had a significant degree of graphene functionalization.[66] The ID/IG ratio changed during the course of the reaction, amounting to 1.24, 1.13, 1.27, 1.17, and 1.19 for samples prepared at 1, 3, 6, 24, and 36 h, respectively, and remained 1.19 after 48 h of reaction (Figure S5, Supporting Information), corroborating the previously discussed results. At 1 h, the high ID/IG ratio of 1.24 was mainly due to the high content of sp3 carbons bound to fluorine. In contrast, at 3 h, ID/IG fell to 1.13 since the fluorine content was reduced by 3 times, whereas FD was still far from the maximum value. The highest ID/IG intensity ratio corresponded to 6 h, where the FD was 2.8, and there was still significant fluorine content. At 24 h, where the sp2 C content was at a maximum, ID/IG dropped to 1.17, whereas at 36 and 48 h, the high FD (36 h sample) and the high number of sp3 C atoms (48 h sample) resulted in a slight increase of the ratio to 1.19.

Figure 3

(a) SEM and (b) HRTEM micrographs of an FG/Niso-24h sample. (c) AFM image of a representative flake, and (d) its corresponding height profile. (e) HRTEM image of an FG/Niso-24h flake, and (f) dark-field HRTEM image of the same flake used for energy-dispersive X-ray spectroscopy (EDS) chemical mapping: (g) carbon map, (h) fluorine map, (i) nitrogen map, (j) oxygen map, and (k) overall map of all four elements.

Grafting Niso groups onto a graphene surface should lead to formation of a zwitterionic network. This was evidenced by zeta-potential (ζp) measurements, which indicated that FG/Niso-24h sheets dispersed in water were negatively charged (Figure d) as a result of ionization of the carboxylic groups. The isoelectric point for FG/Niso-24h of ∼3.02 was consistent with the pKa of aromatic carboxylic acids (e.g., pKa of isopthalic acid is 3.46). Below pH 3.02, the ζp values of FG/Niso-24h became positive (up to +16.8 mV) due to protonation of the carboxylic groups, resulting in a prevalence of positively charged nitrogen of the surface-grafted secondary amine groups.[67,68]

Water droplet contact angle measurements reveal the hydrophilicity of the FG/Niso materials. The comparison of the contact angle of the parent material (142.5°), which is highly hydrophobic to that of FG/Niso-24h (∼28.0°), evidences the hydrophilicity and the wettability and thus the effective functionalization of the latter material (Figure e,f). The lowest water contact angle was observed for FG/Niso-36h (∼19.0°) with the maximum FD (Figure S6, Supporting Information).

The surface area characteristics of the FG/Niso samples were investigated with N2 adsorption/desorption isotherms (Figures S7 and S8, Supporting Information) and, particularly for FG/Niso-24h, methylene blue adsorption from an aqueous solution was also performed (Figures S9 and S10 with pertinent calculations in the Supporting Information). In the solid phase (i.e., in N2 isotherms), samples displayed low BET surface areas, with a tendency to decrease as functionalization increased. Therefore, the samples prepared at 1–6 h displayed 48–32 m2 g–1, and those prepared at 24–48 h displayed 18–15 m2 g–1. N2 adsorption/desorption isotherms and related BET specific surface areas for all samples are presented in Figure S7, Supporting Information. The corresponding surface area from methylene blue adsorption experiments was dramatically higher and estimated at ∼267 m2 g–1 (Figures S9 and S10). Overall, these results, combined with the contact angle data (as later discussed), support the increasing hydrophilicity and hydration of the interlayer space of the highly functionalized FG/Niso-24h sample, rendering its inner surface accessible to water and to the aqueous electrolytes.

The morphology of FG/Niso-24h flakes was investigated by SEM (Figure a), which revealed flakes with lateral sizes up to ∼1 μm. SEM images of all of the samples are presented in Figure S11 (Supporting Information). The dynamic light scattering (DLS) footprint of the FG/Niso colloids is shown in Figure S12 (Supporting Information), with a mean hydrodynamic diameter of 330 nm. The DLS results confirm a material with a sheetlike structure and approximate span of flake size from 150 nm to 1 μm. HRTEM images (Figure b) showed the layered characteristics of closely stacked sheets, resulting in almost transparent edges. AFM images (Figure c) and the corresponding height profiles (Figure d) suggested that FG/Niso-24h was a few-layer graphene material, assuming that the thicknesses of an FG/Niso sheet functionalized on both sides and on one side with Niso were 1.07 ± 0.38 and 0.53 ± 0.31 nm, respectively, as estimated by the MD calculations. This multilayered structure does not prevent extensive hydration when it is dispersed in water due to its charged functionalities that strongly interact with water molecules.

Bright-field HRTEM images of an FG/Niso-24h flake and dark-field HRTEM images of the same flake—used for energy-dispersive X-ray spectroscopy (EDS) elemental mapping—confirmed the flaky structure of the material (Figure e,f). EDS density maps (Figure g–k) revealed a dense and homogeneous distribution of grafted functional groups over the surface. However, they showed a lack of preferential edge functionalization, indicating that FG chemistry can be used to synthesize true surface functionalized graphene derivatives.

To evaluate the potential application of FG/Niso as an electrode material for supercapacitors, the electrochemical properties of FG/Niso-24h were studied in a three-electrode setup in 1 M Na2SO4 electrolyte. rFG, prepared by the prolonged defluorination of GrF by DMF without any functionalization, was used as a reference material to investigate the role of functionalization on the electrochemical properties of graphene. Both FG/Niso-24h and rFG exhibited cyclic voltammogram (CV) curves of almost rectangular and symmetric shape, indicating promising capacitive properties (Figure a). However, the area under the CV loop of the FG/Niso-24h sample was significantly higher (about 4 times at V = 0.4 V) than the rFG sample, showing its higher capacitance. The near-rectangular CV curves, even at a high scan rate of 200 mV s–1, indicated a small mass-transfer resistance and good charge propagation behavior of ions (Figure b). We also used other electrolytes to gain a deeper insight into the behavior of the electrode material (Figure c). With Na2SO4, Li2SO4, and H2SO4, we observed similar capacitance behavior. Thus, we concluded that the choice of cation was not critical for the performance of the system. Comparing the results for the salt solutions (pH ∼ 7) with those for a solution of 1 M H2SO4, we deduced that the protonation state of the carboxylic groups also did not significantly affect the capacitance behavior. This observation is also in a good agreement with those previously reported by Oh et al.[69] and Lounasvuori et al.[70] On the other hand, using 2 M KOH led to a drop in the capacitance, which may reflect deprotonation of the secondary amine group and a different ionic distribution at the material surface. We also carried out MD simulations to understand more deeply the described behavior. The zwitterionic groups grafted on graphene enhanced the density of counterions around the material (Figure S13) and its porosity (with respect to unfunctionalized graphene). MD simulations of Niso derivatives with different dissociation forms of the functional groups (mimicking the variation of pH) showed that the packing of the structure strongly depended on the pH of the solution. More closely packed structures were preferred with increasing pH (Figure g), that is, in acidic media, the individual sheets spread out and became surrounded by the electrolyte, providing a greater surface area for effective interaction with ions. On the other hand, the Niso derivatives tended to agglomerate in a basic environment (high pH), which may reduce the nanoporosity of the material created by the voids between the graphene sheets (Figure S8, Supporting Information), lowering the ion mobility, accessibility, and hence capacitance. Use of Na2SO4 helps mitigate the corrosive character of both acid and alkali-based media. Galvanostatic charge–discharge (GCD) measurements of both tested materials at a constant current density of 1 A g–1 (potential window: 0–0.7 V) are shown in Figure d. All of the GCD curves are symmetric, implying a reversible reaction between the alkali cations (Na+) and FG/Niso-24h sample. The almost ideal triangular shape of the GCD curves indicates capacitive behavior (redox-free), in good agreement with the results obtained by cyclic voltammetry. Figure e (top graph) shows the influence of the synthesis time on the capacitive behavior of FG/Niso-xh. Samples prepared at 1, 3, 6, 24, 36, and 48 h exhibited specific capacitances Csp of 271, 297, 340, 391, 359, and 305 F g–1, respectively, at a current density of 0.25 A g–1 in the aqueous 1 M Na2SO4 electrolyte. As can be seen, Csp reached a maximum for the sample prepared at 24 h (FG/Niso-24h). This maximal value of Csp can be explained in terms of an optimum combination of conductivity and FD, as evidenced from both EIS (Figure e, bottom graph) and XPS. Since the capacitive performance depends on both the conductivity and presence of functionalities, a suitable balance between these two key factors has to be achieved. At 36 h, FD was 3.5, the highest value among the samples, but Csp (∼359 F g–1) was not maximal. The maximal Csp value of ∼391 F g–1 was observed after 24 h of reaction. However, further prolongation of the reaction (48 h) led to a drop in the specific capacitance, probably due to the increased content of sp3 defects in the graphene lattice (Table ). It should be noted that Csp of all of the FG/Niso-xh samples was higher than that of rGF (89 F g–1) (Figure f), again emphasizing the critical role of the functional groups for imprinting high capacitance values to covalently functionalized graphene derivatives.

Figure 4

(a) Cyclic voltammograms (CVs) of rFG and FG/Niso-24h recorded at a constant scan rate of 50 mV s–1. (b) CVs of FG/Niso-24h at different potential scan rates ranging from 5 to 200 mV s–1. (c) CVs of FG/Niso-24h recorded in different electrolytes at a constant scan rate of 50 mV s–1. (d) Galvanostatic charge–discharge curves (GCDs) of rFG and FG/Niso-24h recorded at a constant current density of 1 A g–1. (e) Influence of synthesis time of FG/Niso-xh derivatives on their capacitive (top graph) and resistive performance (bottom graph). (f) GCD curves of rFG and FG/Niso-24h recorded at different current densities ranging from 0.25 to 10 A g–1. All measurements were performed in a three-electrode system with a 1 M Na2SO4 aqueous electrolyte unless otherwise specified. (g) Snapshots from MD simulations showing changes in the packing of the structure with respect to increasing pH. Only ions within a 15 Å radius of the functionalized graphenes are shown. Water is omitted for clarity. Coloring scheme: gray, graphene; red spheres, sodium; green spheres, chlorine.

A two-electrode symmetrical supercapacitor cell was used with 1 M Na2SO4 as the electrolyte (the design of the cell is shown in the inset of Figure c) to obtain data under close-to-actual operation conditions of supercapacitor devices. Figure a shows a set of CVs recorded at different scan rates with operating voltage up to 1.2 V. As can be observed, all of the CV curves were rectangular in shape without obvious redox peaks, indicating near-ideal capacitive behavior with good rate performance even at high scan rates (2000 mV s–1), which indicates the high rate-capability of the cell. GCD curves (Figure b) showed good symmetry and nearly linear slopes at current densities of 0.5 A g–1 and above (i.e., at capacitive-relevant time scales), which is indicative of efficient EDL formation. Csp calculated from the discharge curve at a current density of 0.25 A g–1 was found to be 105 F g–1. Although Csp is an important parameter to assess an electrochemical double-layer capacitor material, volumetric capacitance Cvol (Cvol = Csp × ρ, where ρ denotes material density) is another essential metric for evaluating a supercapacitor’s performance.[71] At a current density of 0.25 A g–1, the FG/Niso-24h sample exhibited a Cvol of 21 F cm–3. Then, Figure c shows that both Csp and Cvol remained stable even at high current densities, for example, 5 A g–1 (65 F g–1 and 13 F cm–3). One of the most important factors of supercapacitors is cycle stability. To evaluate this property, the GCD (at 2 A g–1) method was used to characterize the long-term charge–discharge behavior (Figure d). We found that Csp of the cell remained constant up to 5000 cycles, suggesting that FG/Niso-24h would be a suitable material for supercapacitor fabrication.

Figure 5

(a) CVs of FG/Niso-24h recorded at different scan rates. (b) GCDs of FG/Niso-24h at different current densities ranging from 0.25 to 5 A g–1. (c) Change in specific and volumetric capacitance of FG/Niso-24h with increasing current density. (d) Cycle stability of FG/Niso-24h. (e) Nyquist plots of FG/Niso-24h recorded over the frequency range 0.01 Hz to 10 kHz at open-circuit potential (OCP) with 5 mV amplitude. (f) Bode representation of FG/Niso-24h. All measurements were performed in a two-electrode setup with a 1 M Na2SO4 aqueous electrolyte.

To understand the electrochemical performance of FG/Niso-24h, the conductive and diffusive behavior was investigated by EIS. Figure e shows a Nyquist plot of FG/Niso-24h (left-hand plot) with an expanded view of the high-frequency region (right-hand plot). Three distinct regions could be identified: a nearly vertical line at low frequencies, a nearly 45° diagonal line at intermediate frequencies, and a small semicircle at high frequency. The vertical line at low frequencies indicates good diffusive behavior of the electrolyte ions in the system.[72] The slope of the ∼45° line reflects the diffusion of the electrolyte into the electrode interface, corresponding to Warburg impedance.[73] The electron spin resonance (ESR) value of FG/Niso-24h was found to be 0.33 Ω, which shows good transport behavior of the electrolyte to the electrolyte–electrode interface.[74] Moreover, the small value of ESR is also perfectly reflected by the four-probe measurement, giving a high value of electrical conductivity of 8113 S m–1. The Bode phase-angle plot is depicted in Figure f, where the near −90° phase angle at low frequencies proves the capacitive behavior of the FG/Niso-24h sample. In the two-electrode setup, the device exhibited a reasonable energy density of 18.9 Wh kg–1 at a power density of 470 W kg–1 and current density of 0.25 A g–1.

Conclusions

We developed a straightforward, one-step, cost-effective, and environmentally friendly synthesis of a few-layer graphene-based material functionalized with a small zwitterionic organic group, which can be achieved under mild reaction conditions. The functionalized materials were homogeneously decorated with 5-aminoisophthalic acid functionalities and displayed remarkable functionalization degrees that varied with the reaction time. Covalent attachment of functional groups with charged and ionophilic/hydrophilic character onto the graphene surface significantly promoted hydration due to electrostatic interactions between the charged centers and water molecules. The resulting conductive graphitic surfaces also became more accessible to electrolyte ions, enhancing the ionic and electronic transport required for efficient supercapacitor electrode materials. Altering the reaction time enabled tuning of the properties and capacitive performance of the functionalized material. The specific capacitance was found to range from 271 to 391 F g–1 (in 1 M Na2SO4), which, together with the outstanding cycle stability, high energy, and power density, makes the synthesized metal-free, nonpolymeric, lightweight graphene-based material extremely attractive for applications in energy storage devices, such as supercapacitors.