Advantage of graphene fluorination instead of oxygenation for restorable adsorption of gaseous ammonia and nitrogen dioxide

Carbon 118 (2017) 225e232

Contents lists available at ScienceDirect

Carbon journal homepage: www.elsevier.com/locate/carbon

Advantage of graphene fluorination instead of oxygenation for restorable adsorption of gaseous ammonia and nitrogen dioxide Vitalii I. Sysoev a, *, Alexander V. Okotrub a, b, Igor P. Asanov a, b, Pavel N. Gevko a, Lyubov G. Bulusheva a, b a b

Nikolaev Institute of Inorganic Chemistry, Siberian Branch of Russian Academy of Sciences, 3 Acad. Lavrentiev Ave., Novosibirsk 630090, Russia Novosibirsk State University, 2 Pirogova Str., Novosibirsk 630090, Russia

a r t i c l e i n f o

a b s t r a c t

Article history: Received 13 September 2016 Received in revised form 30 January 2017 Accepted 9 March 2017 Available online 15 March 2017

An interaction of graphene with gaseous molecules increases substantially with grafting of functional groups to its surface. However, in the efficient sensors, such interaction should not be too strong to provide an easy desorption of molecules. Here, we reveal an influence of fluorine and hydroxyl species on the graphene surface on the restorable adsorption of ammonia and nitrogen dioxide, taken as model gases having a different donor/acceptor property. Conductive films of few-layered fluorinated graphene and oxyfluorinated graphene were produced using a one-step process of the exfoliation and partial reduction of corresponding graphite derivatives. The films showed a similar sensitivity on exposure to NH3 and NO2, while the fluorinated graphene-based sensor had much better recovery after a simple argon purging at room temperature. Density functional theory calculations revealed that NO2 and NH3 molecules are adsorbed on fluorine and oxygen from a hydroxyl group as well as bare carbon atoms located near the functionalized carbon. The strongest adsorption energy was obtained for an oxyfluorinated grapheneeNH3 system due to short N/H(O) contacts. Our results show that fluorinated graphene is more perspective for gas sensing as compared to oxygenated graphene due to its higher chemical stability and weaker interactions with the adsorbed molecules. © 2017 Elsevier Ltd. All rights reserved.

1. Introduction Graphene is an excellent platform for gas sensing due to its unique and outstanding properties such as extremely high surfaceto-volume ratio and sensitivity to charge transfer from/to adsorbed molecules through a change in electrical conductivity [1,2]. Covalent and noncovalent modifications of graphene alter the conductivity in a wide range and create the adsorption sites for gas molecules. That can drastically improve performance of specific devices [3]. The methods of graphene modification include attachment of functional species such as oxygenated groups [4,5] and fluorine atoms [6,7], introduction of structural defects [8,9] or heteroatoms [10,11] in the network, and hybridization with metal/metal oxide nanoparticles [12,13] and polymers [14]. Depending on the type of modification, the energy of interaction between molecules and graphene layer may vary substantially,

* Corresponding author. E-mail address: [email protected] (V.I. Sysoev). http://dx.doi.org/10.1016/j.carbon.2017.03.026 0008-6223/© 2017 Elsevier Ltd. All rights reserved.

which opens the possibility to obtain sensors with fast adsorption/ desorption rates and good selectivity for particular gases [15]. The most intensive attention has been paid to the testing of reduced graphene oxide (rGO), which forms stable dispersions in various solvents and can be used for wafer-scalable thin-film deposition. Oxygen functional groups remaining on the rGO surface act as binding sites for gas molecules and at the same time deteriorate the graphene conductivity. Hence, the number and type of the functional groups should be balanced in order to produce a high-response molecular sensor. Calculations within the density functional theory (DFT) have shown that carbonyl (eC]O) and hydroxyl (eOH) groups enhance the binding energy and charge transfer for the adsorbed nitrogen oxides [16] and ammonia [17] as compared to those expected for non-modified graphene. Rotations of some hydroxyl groups during the adsorption and desorption of NO2 molecules would explain a high reversibility of oxygenated graphene sensor [18]. By combining IR-spectroscopy and DFT calculation data Mattson et al. have proposed that epoxide (CeOeC) is the most reactive functional group [17,19]. This group may dissociate when interacting

226

V.I. Sysoev et al. / Carbon 118 (2017) 225e232

with NO2 and NH3 molecules with formation of nitrite ion [19] and eNH2 species [17], respectively. However, other calculations found that for the NH3 molecule particularly, the hydroxyl group is more attractive than the epoxide [20,21]. The differences in the results may be due to a different size and composition of the computed models. Nonetheless, a strong bonding of analyte molecules with a surface of oxygenated graphene revealed by the DFT calculations could explain a low recovery of some rGO sensors at room temperature [22e24]. Another famous member of the family of graphene derivatives is fluorinated graphene (FG). In contrast to the oxygenated graphene, the FG contains mainly a single type of the functional species, namely, the CeF bonds, which are normally directed to the basal graphene plane [25]. These bonds improve adsorption of NH3 molecules that provides higher sensitivity of FG sensors compared to the pristine graphene as have been concluded from the DFT and experimental results [26]. We have previously demonstrated that the surface of a hydrazine-treated graphite fluoride is also able to sense the ammonia gas [27]. Such kind of the chemical treatment removes fluorine atoms primarily from the exposed surface [28] and the active sensor sites are the sp3ehybridized carbon atoms bonded with fluorine atoms located on the backside of graphene layer [27]. Park and co-authors have proposed to modify graphene oxide (GO) by fluorine in order to make ammonia gas sensor, but, all obtained sensors failed to be recovered at room temperature [29]. In the present work, we reveal an influence of oxygen and fluorine functional species on gas sensing performance of chemically modified graphene at comparative study of FG and oxyfluorinated graphene (OFG) thin films. The FG and OFG were produced by a mechanochemical exfoliation resulted in a partial recovering of the parent graphite derivatives and characterized by atomic force microscopy (AFM), Raman spectroscopy, and X-ray photoelectron spectroscopy (XPS). The XPS showed that the obtained samples are good systems for the comparative study, because the former graphene sample contained mostly fluorine as a foreign element, while near equal amounts of oxygen and fluorine were found in the latter sample. The sensor response was studied to NH3 and NO2 gases, which are problematic air pollutants from industry. These molecules are also complementary model analytes because the change in graphene conductivity upon their adsorption has opposite sign that allows identifying NH3 or NO2 in the environment. The tests revealed a superior sensing performance of the FG film with the more balanced sensitivity and recovery behavior. The experimental results were supported by DFT calculations of model structures.

graphite powder in an amount of ~0.5 mg was grinded in an agate mortar for 30 min and then dispersed in 10 mL of toluene under sonication (100 W, 35 kHz) for 10 min. Microscopic aggregates were removed by a mild centrifugation. 5 mL of dispersion was sprayed onto a 3
5 mm2 SiO2/Si substrate using an air gun system. Small drops were delivered onto substrate preheated to ~110  C by argon gas with a purity of 99.95% and inlet pressure of 4 bar. The spraying rate was 0.1 mL/min.

2. Experimental section

the sensor recovery was defined as:

2.1. Material

Recovery ¼

A starting material for chemical modifications was purified natural graphite from Zaval'evsk deposit (Ukraine). The fluorination of graphite was performed at room temperature using a procedure described elsewhere [30]. Graphite crystallites were placed into a Teflon flask containing liquid bromine to prepare a bromineintercalated graphite, which then was transferred into another flask with a solution of BrF3 in Br2 and kept there for several days. A simultaneous attachment of oxygenated groups and fluorine atoms to graphite layers was achieved using a solution of CrO3 in liquid anhydrous HF [31]. After one-month reaction at room temperature, the solution was decanted and solid product was repeatedly washed with concentrated HCl to remove chromium salts. Finally, the samples of chemically modified graphite were dried in a nitrogen flow. For film preparation, fluorinated graphite or oxyfluorinated

2.2. Characterization Samples were structurally characterized by scanning electron microscopy (SEM) on a Carl Zeiss AG e SUPRA 40 microscope and AFM on a Solver Pro microscope (NT-MDT). The AFM measurements were performed in tapping mode using cantilevers NSG10 (NT-MDT) with a tip curvature radius of 6 nm and an average value of the force constant of 11.8 N/m. Raman spectra were obtained on a Spex 1877 triple spectrometer using the 488enm line from an argon laser. XPS spectra were collected with a Phoibos 150 SPECS spectrometer using a monochromatic Al Ka radiation with the energy of 1486.7 eV. The binding energy scale was internally calibrated to the energy 284.4 eV of the 1s line of sp2ehybridized carbon. The C 1s spectra were fitted using a symmetric Gaussian/ Lorentzian product function after subtraction of the background signal by Shirley's method. 2.3. Sensor fabrication and testing A film deposited on SiO2/Si substrate was used for device fabrication. Two silver electrodes of a width 5 mm were formed by a silver glue on the top of the film at a distance of ~1 mm from each other. The device was mounted in a test chamber and investigation of gas sensor properties was carried out under nearly practical conditions (atmospheric pressure and room temperature) against ammonia and nitrogen dioxide diluted by argon. More details about the experimental setup are given elsewhere [27]. A change in electrical resistance of the device was monitored when the sensor was periodically exposed to an analyte gas and pure Ar. Resistance was measured using a Keithly 6485 picoammeter at a DC voltage drop of 1 V. The fractional method [32] was employed to compare the changes in the resistance of different devices. The relative response was calculated as the following:

Relative response ¼

Rg  Ra ; Rg  R0

Rg  R0 ; R0

(1)

(2)

where R0 is the sensor initial resistance (baseline), Rg is the sensor resistance after explosion to the analyte gas, and Ra is the resistance after regenerating the sensor to its original state using pure Ar. 2.4. DFT calculations Calculations were carried out using the three-parameter hybrid functional Becke [33] and LeeeYangeParr correlation functional [34] with a dispersion correction developed by Grimme et al. [35,36] (B3LYP-D3 method) in the framework of the Jaguar program package [37]. Atomic orbitals were described by 6e31G*þ basis set with polarization and diffuse functions for all atoms except hydrogen. Initial graphene fragment had a composition of С80Н22, where hydrogen atoms saturated the dangling bonds of the

V.I. Sysoev et al. / Carbon 118 (2017) 225e232

boundary carbon atoms. Fluorine and/or oxygen functionalities were attached only to the central C]C bond of the fragment. Geometries of models were optimized by an analytical method to the gradient of 5
104 atomic units for the shift of the atomic positions. The absence of imaginary frequencies indicated that the obtained configurations correspond to the local minima on the potential energy surfaces. The adsorption energy of NH3 and NO2 molecules was calculated tot þ Etot  E tot as: Eads ¼ Efrag ; where Ex is a total energy of the mol compl optimized isolated functionalized graphene fragment, the optimized free molecule, and the optimized adsorption complex, respectively. The charges on the adsorbed molecules were obtained as a sum of the atomic charges calculated from Natural Bond Orbital (NBO) analysis.

3. Results and discussion 3.1. Materials structure and composition A sonication of fluorinated graphite and oxyfluorinated graphite powders in toluene for 60 min has produced colored dispersions (Fig. 1(a)), which were stable a couple of hours. The preliminary grinding of the powders caused an exfoliation of functionalized graphite crystallites, which provided a few days living of the dispersions after 10 min of sonication only. A darkening of these dispersions as compared to those obtained without the grinding step (Fig. 1(a)) indicates a partial removal of functional groups from the graphene surface. SEM study of a sample produced on SiO2/Si substrate for gassensor device fabrication showed formation of a continuous film from overlapping thin flakes (Fig. 1(b)). A thickness of the obtained films was ~300 nm and an average roughness of the surface was ~90 nm as determined by AFM. The flakes deposited on the

227

substrate by spraying of small amount of the dispersion obtained from the fluorinated graphite were examined by AFM (Fig. 1(c)). An average lateral size of the flakes was 0.4 mm, the height measured for an isolated flake was about 2.9 nm (the insert in Fig. 1(c)). Since the thickness of fluorinated graphene monolayer measured by AFM is approximately 0.78e0.87 nm [38], we may conclude that the flakes from the dispersions consist of 3e5 functionalized graphene layers. Because the grinding of samples followed by short-time sonication produced few-layered flakes with partially recovered graphene layers, this procedure can be considered as a mechanochemical exfoliation. In further discussion, the samples obtained by mechanochemical exfoliation of fluorinated graphite and oxyfluorinated graphite will be referred as rFG and rOFG, respectively. Raman spectroscopy confirmed that structures of fluorinated graphite and oxyfluorinated graphite were changed after the treatment (Fig. 1(d)). The Raman spectra of the initial samples showed two main bands D and G located around 1354 and 1586 cm1, respectively. Additional band W detected at 1430 cm1 in the spectrum of fluorinated graphite is related to the lattice distortions at the boundaries between flat carbon areas and fluorinated carbon regions within a layer [39]. The D band corresponds to scatterings on structural defects of the hexagonal graphite lattice. In case of the investigated graphite samples, these defects are introduced by the attachment of fluorine and oxygenated groups. A ratio of the intensities of D and G bands is often used to estimate the defect density in carbon materials. It is interestingly, that the ratio increased in the Raman spectra of the rFG and rOFG films as compared to the parent bulk samples. Moreover, the W band was totally vanished in the rFG spectrum. These spectral changes could be explained by creation of new sp2ehybridized carbon areas that are smaller than those present in the layers of initial functionalized graphites, but more numerous in number [40]. Also an increase of the D/G intensity ratio may be associated with the edge states produced during exfoliation process [41].

Fig. 1. (A) Fresh toluene dispersions of fluorinated graphite (FG), reduced fluorinated graphite (rFG), oxyfluorinated graphite (OFG) and reduced oxyfluorinated graphite (rOFG) samples. (B) SEM image of rFG film and (C) AFM image of isolated rFG flakes, deposited on SiO2/Si substrate. Inset shows height profile of the flake along white line. (D) Raman spectra of the samples after subtraction of fluorescence background and normalization to the G-band height. (A colour version of this figure can be viewed online.)

228

V.I. Sysoev et al. / Carbon 118 (2017) 225e232

Chemical compositions of the samples were evaluated by XPS. The XPS C 1s spectrum of fluorinated graphite showed two main peaks located at about 286.5 and 288.9 eV (Fig. 2(a)) and corresponding to nearest neighbor carbon atoms at CFegroups and carbon covalently bonded with fluorine, respectively [42]. Weak components at 284.5 and 290.2 eV are attributed to sp2 carbon areas remained in the sample after fluorination and CF2 species located at graphene layer edges. The spectrum of oxyfluorinated graphite was fitted by five components (Fig. 2(b)). A component at 288.5 eV corresponds to the CeF bonding and an intense component at 287.5 eV could be attributed to carbon bonded with hydroxyl groups. Note, that in the case of oxygenated graphite samples the component at this binding energy is related with the epoxy groups, while the CeOH species appear at a lower energy [43,44]. We cannot do such assignment, because the oxygenated groups may interact with HF [45] and the epoxy group is more reactive than the hydroxyl one. The bare carbon atoms linked with carbon atoms functionalized by fluorine and oxygenated groups contribute to the component at 285.6 eV. Since the position of this component is also shifted as compared to the middle peak in the C 1s spectrum of fluorinated graphite, we assume that adjacency of CeF and CeOH species affects binding energies of core electrons. A high-energy component at 289.6 eV is attributed to carboxyl groups. Mechanochemical exfoliation of functionalized graphite samples causes a partial removal of fluorine and oxygenated groups (Fig. 2 (c, d)). This appears as a strong increase of the relative intensity of the sp2 carbon component, which indicates a restoring of conjugative p-system, and a suppression of the C*eO and C*eF components. The suppression is more significant for the oxyfluorinated graphite. Hence, the fluorinated graphite has a higher stability. An appearance of the C*eO component in the rFG spectrum (Fig. 2 (c)) is connected with hydrolysis of a part of CF groups during the film sample preparation.

The composition of rFG and rOFG films was estimated from the deconvolution of the XPS C 1s spectra to be CF0.25O0.04 and CF0.05O0.06, respectively. Since the former sample is greatly enriched with fluorine, while the latter sample contains almost equal amounts of fluorine and oxygen, these systems are good for a comparative study with the aim to differentiate an influence of fluorine and oxygenated species on the efficiency of interaction of functionalized graphene with gaseous molecules. 3.2. Sensor response study Relative sensor responses of rFG and rOFG films against a low concentration of ammonia or nitrogen dioxide in argon are compared in Fig. 3. An opposite change of the relative resistances of the sensors with respect to ammonia and nitrogen dioxide gases is consistent with the charge transfer mechanism proposed for pdoped graphene [46]. The adsorbed NH3 molecules donate electron density to graphene layer that reduces the concentration of charge carriers and increases the sensor resistance (Fig. 3 (a, c)). The NO2 molecules accept electrons from graphene, thereby decreasing the resistance (Fig. 3 (b, d)). A decrease in the relative response during the run-to-run test indicates incomplete recovery of sensors after argon purging for 20 min. Performances of the gas sensors were evaluated from the changes in resistance in the third cycle using Eqs. (1) and (2). Initial resistances of the rFG and rOFG films were 2.93 and 1.38 MU, correspondingly. The rFG sensor showed a greater response than the rOFG sensor and its response amplitude was higher for the NH3 sensing (Fig. 3). As relative to nitrogen dioxide, the run-to-run recoveries of the rFG and rOFG sensors had a similar behavior (Fig. 3(b, d)). They were minimal after the first cycle and reached ~60e70% after the third cycle. In the case of ammonia, the rFG sensor showed a weak recovery at the first cycle (Fig. 3 (a)) and ~84% recovery at the third one, while the rOFG sensor possessed a

Fig. 2. XPS C 1s spectra of fluorinated graphite (A) and oxyfluorinated graphite (B) crystallites and rFG (C) and rOFG (D) films obtained by mechanochemical exfoliation of starting materials. (A colour version of this figure can be viewed online.)

V.I. Sysoev et al. / Carbon 118 (2017) 225e232

229

Fig. 3. Run-to-run response of rFG sensor (A, B) and rOFG sensor (C, D) for 100 ppm ammonia (A, C) and nitrogen dioxide (B, D). The percentages are the relative response (shaded part) and relative recovery (white part) of sensor in corresponding cycle (Roman numeral). (A colour version of this figure can be viewed online.)

poor reversibility during all three cycles (Fig. 3 (c)). To study the adsorption kinetics, we used an approach of the Langmuir theory, which assumes that all adsorption sites are identical, one site binds only one molecule, and the molecules do not interact with each other [47]. In this case, the adsorption of analyte molecules is described as:

dq ¼ ka $p$ð1  qÞ  kd $q; dt

(3)

where q represents the fraction of functionalized graphene surface actually covered by the molecules, p e partial pressure of analyte molecules and ka and kd are the specific rate constants for adsorption and desorption, respectively. Integration of Eq. (3) gives:

qðtÞ ¼


$ 1  eðka þkd Þ$t :

1

(4)

d 1 þ kk$p a

As can be seen from Eq. (4), the qðtÞ is dimensionless quantity, the constants ka and kd have units of an inverse time. The quantity 1 can be related with the characteristic time of adsorption t . In a ka þkd our experiments, the fraction of the coverage of the adsorption sites can be written as a relative response multiplied by a proportionality factor a. Then, Eq. (4) becomes:

0

0

1

1

t C t C B B $@1  eta A ¼ A$@1  eta A

1

ResponseðtÞ ¼ a$ d 1 þ kk$p

(5)

a

By using the same approach for desorption process, the surface recovery is expressed as: tt

RecoveryðtÞ ¼ b$qa ð0Þ$ekd $t ¼ B$e

d

(6)

where qa ð0Þ is the fraction of coverage before the recovery starts, td

is the characteristic time of desorption and b is the proportionality factor relative to the experimental curve. Experimental curves for all systems are well fitted by Eqs. (5) and (6) except for the first cycles. The fitting details are presented in Supplementary Information, Fig. S1. The derived characteristic times of adsorption ta and desorption td are shown in Fig. 3. These values are larger for the rOFG sensor especially in the case to ammonia exposure. From the XPS analysis, the surface of rFG film is mainly functionalized by fluorine, while the rOFG film contains fluorine and hydroxyl species in equal quantities. Therefore, the stronger interaction of ammonia with rOFG sensors is likely owing to the latter functional groups. This point will be investigated further using the DFT calculations. In practical terms, functional groups present on the graphene surface promote better adsorption of molecules, which has a positive effect on the response magnitude. On the other hand, the adsorption energy should be enough low to ensure a quick response and acceptable recovery values of sensor by simply purging argon (air). Since the time of characteristic adsorption of analyte was shortest in the rFG-NO2 system (Fig. 3), we examined this sensor in more detail. The r-FG film was exposed to 100 ppm of NO2 for 3 min that is close to the ta value and in such conditions the sensor showed a shorter desorption time ~510 s and a good repeatability during 7 cycles at least (Fig. 4 (a)). On the other hand, when the exposure time was significantly longer than the ta, the sensor had unsaturated response and a sensitivity enhanced by 3e4 times (Fig. 4(b)). Regarding the changes in the relative response of the sensor with a duration to analyte exposure, the experimental curve can be divided into two parts. The first part is a kinetic region corresponding to analyte adsorption on the film surface, which results in a rapid change in the response. The second part is due to analyte diffusion in film volume, which causes a slow change in the sample resistance. Therefore, the differences in the obtained times of desorption are connected with processes of extraction of

230

V.I. Sysoev et al. / Carbon 118 (2017) 225e232

Fig. 4. Response of the rFG sensor to nitrogen dioxide. Run-to-run variation for 100 ppm NO2 with an exposure time of 3 min (A). Dynamic behavior, where the first shaded area corresponds to kinetics of gas adsorption and the second part is due to diffusion of gas into the sensor volume (B). Response for different concentrations of NO2. Insert shows dependence of relative response on the NO2 concentration (C). (A colour version of this figure can be viewed online.)

molecules from the film, which may require more efforts than a simple argon purging of the sensor. A test of the rFG sensor to different concentrations of NO2 revealed a linear response dropped from 16.5 to 3.5% in a range of 100 to 10 ppm (Fig. 4(c)). A sensitivity level of the sensor in these conditions is ~0.2% ppm1 as determined from a dependence of the relative response versus analyte concentration (inset in Fig. 4 (c)). Taking into account the trend of resistance during the cycling with shorter duration of analyte exposure (Fig. S2) we expect the sensor sensitivity can be increased by a factor of 1.3. Sensor performances of the studied rFG and rOFG films were compared with published data for GO and rGO samples, ethylenediamine-modified graphene (EDA-G), sulfonated graphene (S-G), fluorinated graphene (F-G) and epitaxial and CVD graphenes (Table 1). Both our materials have the sensitivity level comparable with the values determined for epitaxial and CVD-graphene samples, F-G and rGO. However, the rFG sensor showed better response and recovery times than the rGO and CVD-graphene sensors. The

lower detection limit for the epitaxial graphene as compared to our sensor is due to the higher signal-to-noise ratio provided by good graphene crystallinity. Modification of graphene layers by EDA and sulfur species led to superior sensitivity, however the recovery time for those sensors was larger than that for our rFG sensor. Hence, we can conclude that the rFG-based sensor has a reasonable performance for NO2 detection at room temperature without any additional procedures for sensor recovery that makes it beneficial for practical applications. Moreover, it has been shown that design of graphene-like material morphology allows achieving a high specific surface area and accessible pore structure [48,49] possessing large numbers of adsorption sites, which can increase the sensitivity of sensor our device. 3.3. DFT modelling Based on the XPS data, the models of rFG and rOFG were constructed through an attachment of correspondingly two fluorine

Table 1 Characteristics of rFG and rOFG sensors to ammonia and nitrogen dioxide in comparison with other graphene-based sensors. Material

Gas

Relative response, %

Response time, s

Recovery time, s

Reference

rFG

NH3, 100 ppm NO2, 100 ppm NH3, 100 ppm NO2, 100 ppm NO2, 1 ppm NO2, 1.25 ppm NH3, 100 ppm NO2, 20 ppm NO2, 20 ppm NH3, 100 ppm NO2, 18 ppm NH3, 200 ppm

42 11a 20 20 20 10 20 76 90 5 9 60

330 230a 690 320 3600 2100 600 300 600 220 300 7200

1100 510a 2500 2000 3600 2700 200 1600 2000 500 400 7200

Current work

rOFG GO rGO rGO EDA-G S-G F-G Epitaxial graphene CVD-graphene a

Sensor performance for response time equal to adsorption time.

Current work [52] [53] [54] [55] [55] [26] [50] [51]

V.I. Sysoev et al. / Carbon 118 (2017) 225e232

231

Fig. 5. Reduced density-gradient-based isosurfaces for noncovalent interactions between NH3 (A, C) or NO2 (B, D) molecule and fluorinated graphene (A, B) or oxyfluorinated graphene (C, D) model. The color scales below represent the sign and strength of the interaction. Red indicates a strong attractive interaction, and blue indicates a strong nonbonding overlap. (A colour version of this figure can be viewed online.)

atoms and fluorine atom and hydroxyl group to the С80Н22 graphene fragment. Location of neighboring functional groups at opposite sides of graphene less deforms the lattice and this was done for the rFG model. In the rOFG model, the fluorine and eOH species were purposely placed at one graphene side to reveal their combined impact on the adsorption of analyte molecules. The found stable configurations of NH3 and NO2 molecules at the surfaces of models are shown in Fig. 5, the geometrical and energetic values are collected in Table 2. The NBO charges on the adsorbed molecules were used for evaluation of charge transfer in the considered systems. The calculations showed that independently on the functional composition of graphene, the NH3 molecule is a donor and the NO2 molecule is an acceptor of electron density thus reproducing the observed behaviors of sensors (Fig. 3). Noncovalent interactions in the complexes determined by an approach based on the electron density and its derivatives [56] were visualized using the gradient isosurfaces, which were colored in accordance with the sign of the Laplacian and the strength of interaction. Large negative values (in red) correspond to strong attractions, and large positive values (in blue) indicate that the interactions are nonbonding. Particularly, the areas of nonbonded overlaps are located in the center of each carbon hexagon. Values near zero indicate very weak, van der Waals interactions. The NH3 molecule interacts with fluorine and oxygen through a hydrogen atom (Fig. 5 (a, c)), the NO2 molecule is oriented by nitrogen to a functional group (Fig. 5 (b, d)). These contacts are in a range of 1.84e2.97 Å (Table 2). In addition, the molecules have a bonding with graphene atoms located near the functionalized carbon atoms. These interactions are weaker than the previous ones due to a quite large distance between a molecule and graphene layer. Among the considered complexes, the rOFG model with NH3, where the molecule is able strongly interacting with fluorine and hydroxyl group, has a markedly larger adsorption energy. This is in agreement with experimental data, which indicated the lower recovery ability of the rOFG sensor (Fig. 3). 4. Conclusions We have developed a simple wafer scalable one-pot preparation process of exfoliation and reduction of fluorinated and oxifluorinated graphite. Obtained materials are sensitive to low concentration of NO2 and NH3 diluted in argon. The sensor response/ recovery data indicates that type of chemical modification plays

Table 2 Adsorption energy (Eads), NBO charge, and shortest contacts for NH3 and NO2 molecules interacted with the models of reduced fluorinated and oxyfluorinated graphites. rFG model

Eads (eV) Charge (e) Contact (Å)

rOFG model

NH3

NO2

NH3

NO2

0.257 þ0.002 2.25 (H/F) 3.17 (N/C)

0.264 0.001 2.97 (N/F) 3.12 (O/C) 3.15 (O/C)

0.515 þ0.059 1.84 (N/H(O)) 2.43 (H/F) 3.31 (N/C)

0.254 0.036 2.77 (N/O) 2.90 (N/C)

important role in the kinetics of analyte adsorption on the graphene surface. Fluorinated graphene can spontaneously recover to their initial states by argon purging, without heating or vacuum pumping, while desorption process was sufficiently slower in the presence of hydroxyl groups. Moreover, DFT calculations on the interaction of NO2 and NH3 with modified graphene surface explain observed differences in the recovery. For the NH3 adsorbate particularly, the calculations show weak van der Waals interactions of the molecule with the fluorinated graphene surface and the formation of stronger hydrogen bonds with oxygen from hydroxyl groups. In our experiments, the best sensor performance was achieved for the fluorinated graphene film with respect to NO2 gas. The sensitivity was determined from the relative response versus analyte concentration to be 0.2% ppm1 and based on the literature data [52,57], we believe that sensor response of such material can be substantially improved by employing the interdigitated electrodes in device fabrication. Acknowledgment This research was supported by the Russian Science Foundation (Grant #14-13-00813) in the part of sensor preparation and characterization, the Russian Foundation for Basic Research (Grant #1629-06144) in the part of DFT calculations, and the FP7-PEOPLE2013-IRSES #612577 (NanoCF) grant. Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.carbon.2017.03.026.

232

V.I. Sysoev et al. / Carbon 118 (2017) 225e232

References [1] A.H. Castro Neto, F. Guinea, N.M.R. Peres, K.S. Novoselov, A.K. Geim, The electronic properties of graphene, Rev. Mod. Phys. 81 (2009) 109e162. [2] F. Schedin, A.K. Geim, S.V. Morozov, E.W. Hill, P. Blake, M.I. Katsnelson, et al., Detection of individual gas molecules adsorbed on graphene, Nat. Mater. 6 (2007) 652e655. [3] J. Liu, J. Tanga, J.J. Gooding, Strategies for chemical modification of graphene and applications of chemically modified graphene, J. Mater. Chem. 22 (2012) 12435e12452. [4] J.D. Fowler, M.J. Allen, V.C. Tung, Y. Yang, R.B. Kaner, B.H. Weiller, Practical chemical sensors from chemically derived graphene, ACS Nano. 3 (2009) 301e306. [5] M.G. Chung, D.H. Kim, H.M. Lee, T. Kim, J.H. Choi, D.K. Seo, et al., Highly sensitive NO2 gas sensor based on ozone treated graphene, Sens. Actuators B 166e167 (2012) 172e176. [6] A.V. Okotrub, K.S. Babin, A.V. Gusel'nikov, I.P. Asanov, L.G. Bulusheva, Interaction of NH3 with the reduced surface of graphite fluoride C2F, Phys. Status Solidi B 247 (2010) 3039e3042. [7] V.I. Sysoev, A.V. Gusel'nikov, M.V. Katkov, I.P. Asanov, L.G. Bulusheva, A.V. Okotrub, Sensor properties of electron beam irradiated fluorinated graphite, J. Nanophot. 10 (2016), 012512-1e012512-6. [8] T.H. Han, Y.-K. Huang, A.T.L. Tan, V.P. Dravid, J. Huang, Steam etched porous graphene oxide network for chemical sensing, J. Am. Chem. Soc. 133 (2011) 15264e15267. [9] S. Zhang, D. Zhang, V.I. Sysoev, O.V. Sedelnikova, I.P. Asanov, M.V. Katkov, et al., Wrinkled reduced graphene oxide nanosheets for highly sensitive and easy recoverable NH3 gas detector, RCS Adv. 4 (2014) 46930e46933. [10] J. Dai, J. Yuan, P. Giannozzi, Gas adsorption on graphene doped with B, N, Al, and S: a theoretical study, Appl. Phys. Lett. 95 (2009), 232105-1e232105-3. [11] Z.M. Ao, J. Yang, S. Li, Q. Jiang, Enhancement of CO detection in Al doped graphene, Chem. Phys. Lett. 461 (2008) 276e279. [12] H. Vedala, D.C. Sorescu, G.P. Kotchey, A. Star, Chemical sensitivity of graphene edges decorated with metal nanoparticles, Nano Lett. 11 (2011) 2342e2347. [13] G. Singh, A. Choudhary, D. Haranath, A.G. Joshi, N. Singh, S. Singh, et al., ZnO decorated luminescent graphene as a potential gas sensor at room temperature, Carbon 50 (2012) 385e394. [14] H. Bai, K. Sheng, P. Zhang, C. Li, G. Shi, Graphene oxide/conducting polymer composite hydrogels, J. Mater. Chem. 21 (2011) 18653e18658. [15] S.S. Varghese, S. Lonkar, K.K. Singh, S. Swaminathan, A. Abdala, Recent advances in graphene based gas sensors, Sens. Actuators B 218 (2015) 160e183. [16] S. Tang, Z. Cao, Adsorption of nitrogen oxides on graphene and graphene oxides: insights from density functional calculations, J. Chem. Phys. 134 (2011), 044710-1e044710-14. [17] E.C. Mattson, K. Pande, M. Unger, S. Cui, G. Lu, M. Gajdardziska-Josifovska, et al., Exploring adsorption and reactivity of NH3 on reduced graphene oxide, J. Phys. Chem. C 117 (2013) 10698e10707. [18] Y.R. Choi, Y.-G. Yoon, K.S. Choi, J.H. Kang, Y.-S. Shim, Y.H. Kim, et al., Role of oxygen functional groups in graphene oxide for reversible room-temperature NO2 sensing, Carbon 91 (2015) 178e187. [19] E.C. Mattson, K. Pande, S. Cui, M. Weinert, J.H. Chen, C.J. Hirschmugl, Investigation of NO2 adsorption on reduced graphene oxide, Chem. Phys. Lett. 622 (2015) 86e91. [20] S. Tang, Z. Cao, Adsorption and dissociation of ammonia on graphene oxides: a first-principles study, J. Phys. Chem. C 116 (2012) 8778e8791. [21] Y. Peng, J.H. Li, Ammonia adsorption on graphene and graphene oxide: a firstprinciples study, Front. Environ. Sci. Eng. 7 (2013) 403e411. [22] G. Lu, L.E. Ocola, J. Chen, Reduced graphene oxide for room-temperature gas sensors, Nanotechnology 20 (2009), 445502-1e445502-9. [23] R. Ghosh, A. Midya, S. Santra, S.K. Ray, P.K. Guha, Chemically reduced graphene oxide for ammonia detection at room temperature, ACS Appl. Mater. Interfaces 5 (2013) 7599e7603. [24] J.T. Robinson, F.K. Perkins, E.S. Snow, Z. Wei, P.E. Sheehan, Reduced graphene oxide molecular sensors, Nano Lett. 8 (2008) 3137e3140. [25] A.V. Okotrub, N.F. Yudanov, I.P. Asanov, D.V. Vyalikh, L.G. Bulusheva, Anisotropy of chemical bonding in semifluorinated graphite C2F revealed with angle-resolved X-ray absorption spectroscopy, ACS Nano 7 (2013) 65e74. [26] H. Zhang, L. Fan, H. Dong, P. Zhang, K. Nie, J. Zhong, et al., Spectroscopic investigation of plasma-fluorinated monolayer graphene and application for gas sensing, ACS Appl. Mater. Interfaces 8 (2016) 8652e8661. [27] M.V. Katkov, V.I. Sysoev, A.V. Gusel'nikov, I.P. Asanov, L.G. Bulusheva, A.V. Okotrub, Backside fluorine-functionalized graphene for ammonia detection, Phys. Chem. Chem. Phys. 17 (2015) 444e450. [28] A.V. Okotrub, I.P. Asanov, N.F. Yudanov, K.S. Babin, A.V. Gusel'nikov, T.I. Nedoseikina, et al., Development of graphene layers by reduction of graphite fluoride C2F surface, Phys. Status Solidi B 246 (2009) 2545e2548. [29] M.-S. Park, K.H. Kim, M.-J. Kim, Y.-S. Lee, NH3 gas sensing properties of a gas sensor based on fluorinated graphene oxide, Colloids Surf. A Physicochem.

Eng. Asp. 490 (2016) 104e109.
, [30] L.G. Bulusheva, V.A. Tur, E.O. Fedorovskaya, I.P. Asanov, D. Pontiroli, M. Ricco et al., Structure and supercapacitor performance of graphene materials obtained from brominated and fluorinated graphites, Carbon 78 (2014) 137e146. [31] V.A. Tur, A.V. Okotrub, M.M. Shmakov, E.O. Fedorovskaya, I.P. Asanov, L.G. Bulusheva, Functional composition and supercapacitor properties of graphite oxide reduced with hot sulfuric acid, Phys. Status Solidi B 250 (2013) 2747e2752. [32] T.S. Pearce, S.S. Schiffman, H.T. Nagle, J.W. Gardner, Handbook of Machine Olfaction, Wiley-VCH, Weinheim, 2003. [33] A.D. Becke, Density-functional thermochemistry. III. The role of exact exchange, J. Chem. Phys. 98 (1993) 5648e5652. [34] C. Lee, W. Yang, R.G. Parr, Development of the Colle-Salvetti correlation-energy formula into a functional of the electron-density, Phys. Rev. B 37 (1988) 785e789. [35] S. Grimme, J. Antony, S. Ehrlich, H. Krieg, A consistent and accurate ab initio parametrization of density functional dispersion correction (DFT-D) for the 94 elements H-Pu, J. Chem. Phys. 132 (2010), 154104-1e154104-19. [36] L. Goerigk, S. Grimme, A thorough benchmark of density functional methods for general main group thermochemistry, kinetics, and noncovalent interactions, Phys. Chem. Chem. Phys. 11 (2011) 6670e6688. €dinger, Inc., New York, NY, 2015. [37] Jaguar, Version 8.9, Schro [38] R. Zboril, F. Karlický, A.B. Bourlinos, T.A. Sterioris, A.K. Stubos, V. Georgakilas, et al., Graphene fluoride: a stable stoichiometric graphene derivative and its chemical conversion to graphene, Small 6 (2010) 2885e2891. [39] I.P. Asanov, A.V. Okotrub, A.V. Gusel'nikov, I.V. Yushina, D.V. Vyalikh, L.G. Bulusheva, Charge-induced formation of thin conducting layers on fluorinated graphite surface, Carbon 82 (2015) 446e458. [40] S. Stankovich, D.A. Dikin, R.D. Piner, K.A. Kohlhaas, A. Kleinhammes, Y. Jia, et al., Synthesis of graphene-based nanosheets via chemical reduction of exfoliated graphite oxide, Carbon 45 (2007) 1558e1565. [41] U. Khan, A. O'Neill, M. Lotya, S. De, J.N. Coleman, High-concentration solvent exfoliation of graphene, Small 7 (2010) 864e871. [42] I.P. Asanov, L.G. Bulusheva, M. Dubois, N.F. Yudanov, A.V. Alexeev, T.L. Makarova, et al., Graphene nanochains and nanoislands in the layers of room-temperature fluorinated graphite, Carbon 59 (2013) 518e529. [43] H.-K. Jeong, Y.P. Lee, R.J.W.E. Lahaye, M.-H. Park, K.H. An, I.J. Kim, et al., Evidence of graphitic AB stacking order of graphite oxides, J. Am. Chem. Soc. 130 (2008) 1362e1366. [44] A. Hunt, D.A. Dikin, E.Z. Kurmaev, T.D. Boyko, P. Bazylewski, G.S. Chang, et al., Epoxide speciation and functional group distribution in graphene oxide paper-like materials, Adv. Funct. Mater. 22 (2012) 3950e3957. [45] Z. Wang, J. Wang, Z. Li, P. Gong, X. Liu, L. Zhang, et al., Synthesis of fluorinated graphene with tunable degree of fluorination, Carbon 50 (2012) 5403e5410. [46] F. Yavari, E. Castillo, H. Gullapalli, P.M. Ajayan, N. Koratkar, High sensitivity detection of NO2 and NH3 in air using chemical vapor deposition grown graphene, Appl. Phys. Lett. 100 (2012), 203120-1e203120-4. [47] I. Langmuir, The adsorption of gases on plane surfaces of glass, mica and platinum, J. Am. Chem. Soc. 40 (1918) 1361e1403. [48] F. Yavari, Z. Chen, A.V. Thomas, W. Ren, H.-M. Cheng, N. Koratkar, et al., High sensitivity gas detection using a macroscopic three-dimensional graphene foam network, Sci. Rep. 1 (2011), 166-1e166-5. [49] G. Yang, C. Lee, J. Kim, Three-dimensional graphene network-based chemical sensors on paper substrate, J. Electrochem. Soc. 160 (2013) 160e163. [50] Md.W.K. Nomani, R. Shishir, M. Qazi, D. Diwan, V.B. Shields, et al., Highly sensitive and selective detection of NO2 using epitaxial graphene on 6H-SiC, Sens. Actuators, B 150 (2010) 301e307. [51] F. Yavari, E. Castillo, H. Gullapalli, P.M. Ajayan, N. Koratkar, High sensitivity detection of NO2 and NH3 in air using chemical vapor deposition grown graphene, Appl. Phys. Lett. 100 (2012) 203120. [52] S. Prezioso, F. Perrozzi, L. Giancaterini, C. Cantalini, E. Treossi, V. Palermo, et al., Graphene oxide as a practical solution to high sensitivity gas sensing, J. Phys. Chem. C 117 (2013) 10683e10690. [53] Y. Wang, L. Zhang, N. Hu, Y. Wang, Y. Zhang, et al., Ammonia gas sensors based on chemically reduced graphene oxide sheets self-assembled on Au electrodes, Nanoscale Res. Lett. 9 (2014) 251. [54] Y.J. Yun, W.G. Hong, N.-J. Choi, B.H. Kim, Y. Jun, et al., Ultrasensitive and highly selective graphene-based single yarn for use in wearable gas sensor, Sci. Rep. 5 (2015) 10904. [55] W. Yuan, A. Liu, L. Huang, C. Li, G. Shi, High-performance NO2 sensors based on chemically modified graphene, Adv. Mater. 25 (2013) 766e771. nchez, J. Contreras-García, A.J. Cohen, [56] E.R. Johnson, S. Keinan, P. Mori-Sa W. Jang, Revealing noncovalent interactions, J. Am. Chem. Soc. 132 (2010) 6498e6506. [57] M. Donarelli, S. Prezioso, F. Perrozzi, L. Giancaterini, C. Cantalini, E. Treossi, et al., Graphene oxide for gas detection under standard humidity conditions, 2D Mater. 2 (2015), 035018-1e035018-9.