- Email: [email protected]
Ozonization, Amination and Photoreduction of Graphene Oxide for Triiodide Reduction Reaction: An Experimental and Theoretical Study
Electrochimica Acta 226 (2017) 10–17
Contents lists available at ScienceDirect
Electrochimica Acta journal homepage: www.elsevier.com/locate/electacta
Ozonization, Amination and Photoreduction of Graphene Oxide for Triiodide Reduction Reaction: An Experimental and Theoretical Study Hongyu Jing, Suzhen Ren* , Yantao Shi, Xuedan Song, Ying Yang, Yanan Guo, Yonglin An, Ce Hao* State Key Laboratory of Fine Chemicals, Dalian University of Technology, Dalian, 116024, Liaoning, China
A R T I C L E I N F O
Article history: Received 28 October 2016 Received in revised form 26 December 2016 Accepted 29 December 2016 Available online 31 December 2016 Keywords: Dye-sensitized solar cells Counter electrode Graphene-based catalysts DFT calculation Photovoltaic performance
A B S T R A C T
This work proposes a mild and environmentally-friendly approach to prepare a highly efficient functional graphene (termed as AGO-hv) using methods of ozone oxidation, solvothermal synthesis, and photoreduction. The use of ozone oxidation in the first step can effectively increase the interlaminar distance between graphite oxide sheets, and create active sites for nucleophilic attack on the epoxy carbon from ammonia. The amino groups were successfully grafted on the surface of graphene as evidenced by the amidation reaction, with a maximum nitrogen content of 10.46 wt% and a C/N molar ratio of 8.46. After further photoreduction of the aminated graphite oxide (AGO), the residual oxygen functionalities, such as C-OH, were effectively removed and the conductivity of the graphene sheet was further recovered. The dye-sensitized solar cell (DSC) exhibited a power conversion efficiency (PCE) of 7.51% based on AGO-hv counter electrode (CE), close to that of Pt counterpart (7.79%). The experimental results indicated that the amidation and photoreduction processes were significantly facilitated by the initial ozonization of graphene oxide, and this process significantly improved the electrochemical activity and the conductivity of graphene oxide. Density functional theory (DFT) calculations revealed that AGOhv had the lowest ionization energy (a better electron-donating ability) and also suitable binding energy with I atoms as well. The combination of ozonization, amination and photoreduction is an efficient route to obtain electrocatalysts with desired compositional distributions and performance for triiodide reduction reaction in DSCs. © 2016 Elsevier Ltd. All rights reserved.
1. Introduction Dye-sensitized solar cells (DSCs) have received much attention due to their low cost, easy fabrication and high efficiency of solar energy conversion [1,2]. These promising photovoltic devices (including a green device of aqueous DSC) are assembled with a TiO2 photoanode, a counter electrode (CE), and a redox couple electrolyte [3–7]. Herein, we focus on the designing of CE component in the liquid I3/I electrolyte using acetonitrile as the solvent. It is well-known that the selected CE in DSC should exhibit high electrical conductivity and electrocatalytic activity for the regeneration of the redox couple. Presently, platinum is one of the most appealing CEs owing to its excellent electrocatalytic activity and high power conversion efficiency (PCE) in DSCs. However, platinum is extremely rare and expensive material, which hinders its large-scale application for DSCs. Alternatives to
* Corresponding authors. E-mail addresses: [email protected] (S. Ren), [email protected] (C. Hao). http://dx.doi.org/10.1016/j.electacta.2016.12.190 0013-4686/© 2016 Elsevier Ltd. All rights reserved.
Pt include carbonaceous materials such as conducting polymers, graphene, and derivatives that are readily available and inexpensive [8–13]. The high specific surface area and excellent conductivity make state-of-the-art carbon materials, such as graphene, promising candidates as an alternative material to platinum [14]. However, pristine graphene sheets are hydrophobic and also exhibit low catalytic activity due to the lack of catalytic active sites for the iodine reduction reaction (IRR) [15]. To address this problem, heteroatoms such as nitrogen, boron, and sulfur have been added to modulate the catalytic activity of graphene. Fang et al. prepared boron-doped graphene by thermal annealing of graphite oxide (GO) with B2O3, which has shown a PCE of 6.73% as a CE for DSCs [16]. Nitrogen-doped graphene (NrG) was obtained by Hou et al. via thermal annealing of a mixture of GO and cyanamide at 900 C in N2 atmosphere, which was subsequently utilized as an alternative to the Pt electrocatalyst for DSCs with the PCE of 5.4% [17]. Song et al. prepared a series of nitrogen-doped porous graphene foams (NPGFs) by hydrothermally treating a mixed
H. Jing et al. / Electrochimica Acta 226 (2017) 10–17
solution of GO and ammonia with optimal PCE of 4.5% for the iodide-based electrolyte [18]. Xue et al., developed threedimensional (3D) N-doped graphene foams (N-GFs) by annealing the freeze-dried graphene oxide foams (GOFs) in ammonia, and found that the resultant DSCs showed a PCE as high as 7.07% [19]. Additionally, other chemical methods to synthesize N-doped graphene have been reported, including chemical vapor deposition, nitrogen plasma treatment, and the reduction of graphene oxide in ammonia or nitrogen gas [20]. Of these, chemical functionalization is considered the most effective method to modify the catalytic activity of graphene [21]. From both theoretical and experimental points of view, we report functionalized graphene, produced by the oxidation of graphene oxide with ozone, amination by inorganic amines of ammonia via solvothermal synthesis, and further photo-reduction by ultraviolet light. In order to enhance the amount of amino groups and prevent the aggregation of the GO, numerous nucleation sites were introduced by ozonization. After reaction with ammonia, the amino groups were grafted onto the preoxidized graphene oxide (GO-O3) via nucleophilic substitution. This modified amination could decrease the damage of the sp2 network of the graphene sheet without sacrificing its conjugative structure. In this case, the amount of oxygen-containing groups in aminated graphite oxide (AGO) were relatively high, thus the residual OH functional groups would be removed via light reduction. The obtained electrocatalyst was denoted as AGO-hv. DFT calculations were used to elucidate the mechanisms of IRR on the synthesized AGO-hv. To the best of our knowledge, the interactions between the functional groups from functional graphene and iodine species have been rarely studied as a critical factor in DSCs. The identification of what type of functional groups serves as the catalytic active site would facilitate the rational design of electrocatalysts for DSCs. Both DFT calculations and experimental results revealed that enhanced performance was partly correlated with the increase in pyridine-like species. The pyridine N and amide moieties introduced may be exploited as the main anchoring site for the IRR. The superior properties of AGO-hv are due to the appropriate binding energy of iodine on the functional group with high conductivity and excellent charge transfer resistance, which improves the PCE (7.51%) to close to that of Pt (7.79%) in the liquid I3/I redox couple electrolyte. 2. Experimental and theoretical methods 2.1. Preparation of catalysts and corresponding CEs The GO-O3 was synthesized by bubbling a certain amount of O3 gas into a GO aqueous suspension by controlling the oxidation time. AGO were synthesized via the modified solvothermal reaction route, using ammonia as nitrogen source, and benzyl alcohol as solvent for the amination of GO-O3 and introduction of amino groups [22]. Then, AGO-hv CEs were prepared by coating
11
AGO films on the FTO glasses and subjected to UV illumination. The preparation of AGO-hv CEs is schematically shown in Scheme 1 (see the Supporting Information for detailed descriptions). The sample of AGO-0 was also prepared based on the above procedures from GO without the addition of O3. Those mentioned samples were also compared to a sample of reduced graphene oxide (rGO) that was produced by a common thermal reduction method described elsewhere [23]. The thickness of all the graphene-based CEs were optimized about 8–10 mm. Additionally, a Pt electrode (200 nm thick) was obtained by drop-casting 0.5 mM H2PtCl6/ethanol solution on a FTO glass substrate. Subsequently, the FTO glass was sintered in a muffle furnace at 450 C for 30 min. 2.2. Structural characterization The obtained samples were characterized by means of X-ray diffraction (XRD), Raman spectroscopy, scanning electron microscopy (SEM), Fourier transform infrared spectroscopy (FTIR), UV–vis absorption spectroscopy, elemental analysis, and X-ray photoelectron spectroscopy (XPS). The details on material characterization are presented in the Supporting Information. 2.3. Electrochemical measurements The current density-voltage (J-V) curves of the DSCs were measured to estimate the photovoltaic performance, which was conducted under simulated AM 1.5 illumination (100 mW cm2, Solar Light Co., INC., USA). The electrochemical properties were investigated by CV, EIS, and Tafel-polarization measurements. The details are presented in the supporting information. 2.4. Computational methods All calculations were performed using Gaussian 09 package based on DFT methods with the B3LYP hybrid functional [24] and 6–31 g (d, p) basis sets [25] for C, H, N and lanl2dz [26] for I atom. The edges of the graphene sheets are terminated by H atoms [27]. The FT-IR spectroscopy (scaled by 0.961) of the models were calculated based on the geometric optimization. Solvation effects were taken into account using the conductor-like polarizable continuum model (CPCM) [28] for all calculations. The ionization energy (IP) is calculated as: IP = EM+ EM, where + EM and EM are the energies of a model with positive charge and a neutral molecule without charges on the ground state [29]. The binding energy (Eb) between I and the binding site is defined as: Eb = E (Site) + 1/2 E (I2) E (I/Site) + E (BSSE). BSSE (basis set superposition error) correction has been taken into account in all the calculations of binding energy [30]. The E (I/Site), E (Site) and E (I2) represent the total energy between I and the site, the energy of AGO-hv, and the energy of the isolated I2 molecule. All the models are shown in the most stable configuration and acetonitrile solvent effects are also taken into account.
Scheme 1. Schematic illustrations of preparation of AGO-hv CE using ozone oxidation, solvothermal synthesis and light irradiation method.
12
H. Jing et al. / Electrochimica Acta 226 (2017) 10–17
We considered the different types and positions of doped nitrogen as follows: NH2 (primary amine group), ¼N H (primary imine group), and CONH2 (carboxylic amide group) on the graphene sheet and pyridine nitrogen at armchair and zigzag edges. Notably, based on the results of XPS in the experiment, we established initial theoretical models of the catalysts. Additionally, by comparing the experimental and calculated results for FT-IR vibrational frequency and UV–vis absorption spectroscopy, we determined the established geometries are reasonable. DFT calculations were also performed by taking iodine and single-layer AGO-hv as the model system to further investigate their interaction. The top view and side view of the optimized structures of graphene-based electrocatalysts are illustrated in Scheme 2. 3. Results and discussion 3.1. Analysis of product characteristics SEM characterization shows that the resultant AGO consists of a crosslinked structure (see Fig. 1a). AGO-hv, in Fig. 1b, displays a wrinkled structure with a somewhat rough surface, indicating that the oxygen-containing groups of AGO were further decreased by photoreduction. The wrinkled structure increases the surface area of nanosheets and reduces their interlayer p-p stacking process, thereby facilitating the formation of porous films. The structural characteristics of the GO-O3 and AGO-hv were determined by XRD. As shown in Fig. S2a, the sharp and strong peak at 10.7 is attributed to the (001) crystal face of GO-O3, indicating that graphite is completely oxidized. For AGO samples subject to different lengths of illumination time, no diffraction peaks from GO-O3 appear due to effective reduction. The crystal face of (001) completely disappears and a new peak at about 25.0 is observed simultaneously, which corresponds to the (002) face of a graphite structure [31]. The stronger the peak, the higher the degree of graphite stacked. As the light reduction time increased, this peak became weaker and broadened, indicating that the structure of photo-reduced AGO is amorphous and the internal arrangement of crystal planes is not as regular as natural graphite which presents a laminated structure. After amination and further light reduction for 1 h, only one weak and broad diffraction peak centered at about 25.0 was detected and the peak at 42.5 diffractions was dramatically weakened and even disappeared, indicating the effective reduction of AGO [32,33]. It is well-known that the weaker the peak of functionalized graphene oxide, the greater the reduction degree of graphene oxide.
Fig. 1. SEM morphology of (a) AGO and (b) AGO-hv.
Fig. S2b displays the Raman spectroscopy of GO-O3, AGO, and AGO-hv (photo-reduction for 1 h, also designed as AGO–1 h). They all exhibited two peaks at 1347 and 1589 cm1, corresponding to the disordered (D) and graphitic (G) band, respectively [34]. The
Scheme 2. The top view and side view of the optimized structures of graphene-based electrocatalysts. In the models, the gray, blue, red, and white balls represent C, N, O, and H atoms, respectively. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
H. Jing et al. / Electrochimica Acta 226 (2017) 10–17
increase in ID/IG intensity ratio from 0.99 for GO-O3 to 1.15 for AGOhv indicates that GO-O3 was reduced due to the restoration of sp2 networks and the removal of oxygen moieties upon UV irradiation [35]. Normally, the ID/IG ratio should decrease in the process of reduction of GO-O3. However, the ID/IG ratio of AGO-hv is higher than that of GO-O3. This change suggests a decrease in the average size of the sp2 domains upon reduction of the GO-O3 [36]. In order to study the effect of the ozonization time on the population of nitrogen and oxygen species, the amount of N contents in AGO was determined by element analysis, and the results are summarized in Table S1. For an oxidation time of 10 min, the nitrogen content reached a maximum of 10.46% and the calculated C/N value was 8.46. A suitable ozonation time is a prerequisite for introducing a high amount of nitrogen-containing moieties onto the surface. Therefore, the optimum time of ozonation was selected as 10 min for the following process. Fig. S3a shows the FT-IR spectroscopy of GO-O3, AGO, and AGOhv. In the spectrum of GO-O3, the peaks at 1726, 1610, 1230, 1045, and 852 cm1 can be ascribed to the vibrations of carboxyl (-COOH) stretching, skeletal ring stretching, hydroxy ( OH) stretching, epoxy (C O C) stretching, and carbon backbone deformation, respectively [37]. Those characteristic peaks are also consistent with the characteristic vibrations of GO [38]. The removal of oxygen-containing groups by amination and photoreduction are evident by FT-IR. After amination, the intensity of these peaks is greatly suppressed and new peaks appear at 1540, 1110–1460 and 3708 cm1 that can be assigned to the antisymmetric C N stretching vibrations coupled with out-of-plane NH2 and NH modes, as well as N-H stretching vibrations [39,40]. The absorption peaks at 540–811 cm1 can be ascribed to the stretching vibrations of C¼C presented in the graphite edge. After photoreduction, however, in the case of AGO-hv, most peaks decreased or vanished in the FT-IR spectrum, proving that most of the oxygen-containing groups were removed by reduction. The scissoring in-plane bending mode of primary NH2 groups is observed at 1662 cm1 by overlapping with the peaks from C C aromatic ring modes at 1627 cm1 [41]. The swing out-of-plane vibration mode of NH2 in CONH2 is also observed at 667 cm1, and the peaks at 1193, and 1553 cm1 can be assigned to carbon backbone deformation. The additional peak at 1386 cm1 is attributed to the swing vibration mode of N-H in C¼NH. The two peaks at 3840 and 3735 cm1 could be ascribed to the antisymmetric and the symmetric N-H stretching vibrations, respectively. Additionally, the absorption peak at about 3448 cm1 reveals the existence of the O-H stretching vibrations from hydroxyl groups on the graphite surface and weakly adsorbed water molecules. The peak at 2368 cm1 may be due to the absorption peak of CO2 in the air during testing. The FT-IR spectrum showed that the peak intensity decreased, indicating that light reduction is an effective method for removing the oxygen-containing functional groups presented in the AGO. Actually, the group of C¼O moves to the low frequency region, indicating the successful introduction of NH2. The types of these amino related species are not affected by further illumination. Note that the absorption peaks of GO-O3, AGO and AGO-hv in the experimental are approximately consistent with the calculated values in Fig. S3b. The UV–vis absorption spectroscopy of GO-O3, AGO and AGOhv are depicted in Fig. S4. From Fig. S4a, the strong absorption peak of GO-O3 is located in 232 nm, which is due to the p–p* transition of C¼C. For the AGO and AGO-hv, their absorption bands gradually shift to 243 nm and 252 nm with a red shift after amination and 1 h of illumination. This may be due to the delocalization of pelectrons in the graphene ring in GO-O3 probably due to both reduction and increasing conjugative interaction, which signifies the restoration of the graphene sp2 structure and the removal of oxygen functionality [42]. It should be noted that the absorption
13
Fig. 2. XPS survey of GO-O3, AGO and AGO-hv samples. The inset table shows the corresponding element compositions and the C, N, and O contents estimated from XPS measurements. The table also summarises the C/O ratios calculated from the atomic concentrations.
peaks of GO-O3, AGO and AGO-hv in the experimental are also approximately consistent with the calculated values in Fig. S4b. XPS measurement was performed to verify the variation of elemental composition and functional groups of the as-prepared samples. Fig. 2 displays the XPS survey spectroscopy for GO-O3, AGO and AGO-hv. The three peaks located at 283 eV, 400 eV and 530 eV are attributed to C1s, N1s and O1s, respectively. The peak intensity of O1s from AGO was significantly weakened and the corresponding peak intensity of AGO-hv was further weakened after light reduction. The emergence of new signal peaks of N 1 s demonstrates that the introduction of nitrogen into AGO and AGOhv relative to GO-O3. Further analysis for element composition shows a C/O ratio of 8.59 and 10.60 for AGO and AGO-hv, respectively, confirming the further removal of oxygen-containing functional groups. From the high-resolution C 1s spectrum of GO-O3 (Fig. S5), the peak centered at 284.4 eV is assigned to C-C/C¼C and the peaks at 285.0 eV, 286.4 eV and 288.5 eV are ascribed to C OH, C O C, and O¼C-OH groups, respectively [43]. As shown in Fig. 3a, new functional groups of C-N, C¼O appear, and the O¼C OH, COC vanish, but C OH still exists for C1s of AGO compared with GO-O3. Note that the four peaks are observed from the C 1s spectrum of AGO, but only three peaks are evident for AGO-hv (see Fig. 3b). As compared to AGO, the peak at 285.0 eV corresponds to C OH disappears, demonstrating the effective removal of OH groups from AGO and partial restoration of the conjugated graphene sheets via light reduction. The presence of amine groups in the AGO and AGO-hv samples is also confirmed by the peak at 285.8285.9 eV for the C N bond in the primary amine [41]. For the deconvoluted N 1s region of AGO and AGO-hv (see Fig. 3c and d), three signal peaks are observed, which demonstrates the existence of C¼N, C-NH2 and pyridine nitrogen, respectively [44–46]. Curve fitting results of C 1s and N 1s are summarized in Table S2 and Table S3. Interestingly, the amount of C-NH2 and C¼N groups decreased from AGO to AGO-hv (from 22.22 to 17.99 at% for C-NH2, and from 45.68 to 33.19 at% for C¼N), but the amount of pyridine N species in AGO-hv increased to a maximum of 48.82 at%, compared with AGO (32.11 at%). The type of C-OH species contained in AGO decreased upon light illumination, and the other types of carbon and nitrogen containing functional groups were well maintained, suggesting the photoreduction method can remove the oxygen-containing functional groups such as C-OH instead of nitrogen types. Taken together, the XPS results show the
14
H. Jing et al. / Electrochimica Acta 226 (2017) 10–17
Fig. 3. High-resolution spectroscopy: (a) and (b): C1s for AGO and AGO-hv; (c) and (d): N1s for AGO and AGO-hv.
successful functionalization of GO-O3 with nitrogen-containing functional groups. Light reduction does not affect nitrogen-doping types, but does change the compositional distributions of nitrogen-containing groups and removes the C-OH. 3.2. Electrochemical performance
Fig. 4. The J-V curves of rGO, Pt, AGO-0, AGO and AGO-hv as CEs for the liquid I3/I redox couple electrolyte.
The photovoltaic performance of the solar cell devices was next investigated. The current density-voltage (J-V) curves and the corresponding photovoltaic parameters are presented in Fig. 4 and Table 1. Compared to Pt CE, the device made from AGO-hv CE showed a comparable Jsc value, and yielded a similar device efficiency (7.51%) compared to the Pt counterpart (7.79%), consistent with the below electrochemical characterizations. Additionally, the poor performance exhibited by rGO may be attributed to ineffective charge transport and high sheet resistance. EIS and Tafel-polarization measurement were used to further confirm the electrocatalytic activities of the CEs. The Nyquist plots obtained from symmetric cells with different CEs are shown in Fig. 5a. From the equivalent circuit used for EIS fitting (inset in Fig. 5a), the series resistance (Rs) correspondes to high frequency intercept on the real axis. The charge transfer resistance (Rct)
H. Jing et al. / Electrochimica Acta 226 (2017) 10–17
15
Table 1 The photovoltaic parameters of rGO, Pt, AGO-0, AGO and AGO-hv as CEs. CEs
Voc/V
Jsc/mA cm2
FF
PCE/%
rGO AGO-0 AGO AGO-hv Pt
0.76 0.76 0.76 0.77 0.78
12.36 12.92 13.36 14.00 14.20
0.61 0.67 0.71 0.69 0.70
5.70 6.54 7.20 7.51 7.79
indicates the semicircles in the high-frequency range between the CE and electrolyte and is mainly related to the electrocatalytic activity for the reduction of triiodine ions [47]. Smaller Rct values imply lower internal resistance and higher catalytic activity of a CE, which enhances device performance and is shown in the improvement of Jsc. The Jsc values for the rGO and AGO cell are smaller than those for the AGO-hv and Pt cells, in accordance with the results of Rct. The EIS results reveal that the catalytic activities increase in the order of rGO < AGO < AGO-hv < Pt, consistent with the Tafel results described below. Tafel curves were measured to further investigate the catalytic activities. In theory, the Tafel curve can be divided into 3 zones. The curve at the low potential is attributed to the polarization zone, the curve at middle potential with a sharp slope is attributed to the Tafel zone, and the curve at high potential in the horizontal part is attributed to the diffusion zone. In the latter two zones, we can obtain information on the limiting diffusion current density (Jlim) and the exchange current density (J0), closely related to the catalytic activity of the catalysts. Fig. 5b shows the logarithmic current density as a function of voltage. As shown in Fig. 5b, the Tafel plot of AGO-hv sample exhibits a similar slope as that of the Pt electrode, indicating its good electrocatalytic activity. The J0 of Pt is slightly larger than that of AGO-hv sample, which means that Pt is more effective than AGO-hv in catalyzing triiodide reduction. However, AGO-hv has a larger limiting diffusion current density than the Pt electrode. This means AGO-hv can catalyze reducing I3 to I as effectively as Pt. The Tafel-polarization results were consistent with the EIS and J-V experiments. 3.3. DFT calculations The optimized geometric structures of the graphene-based electrocatalysts performed with DFT are shown in Scheme 2.
Fig. 6. (a) NBO charge analysis and (b) six possible binding sites on AGO-hv.
According to DFT calculation, we confirmed that step III (the process of electron transfer for IRR) was the rate-determining step (see Fig. S9 in Supporting Information for details). To rationally design the catalyst of IRR, a lower value of ionization energy (IP) of materials is desired for DSCs. We next calculated the IP value, which represents the ability of catalyst to release electrons (presented in Table S5). The AGO-hv material showed the lowest IP (79.38 kcal mol1) was located on the top of the volcanic plot. This was also lower than the materials tested in our previous work (AFG material, IP = 81.79 kcal mol1 [48]). This result suggests that the AGO-hv catalyst has good electron-donating ability, suggesting potential function as excellent CEs with better catalytic activity in DSCs. The lowest IP value obtained from theoretical calculation agreed with the experimental results. The lowest IP corresponded to the strongest electron-donating ability, due to the decreased oxygen-containing groups after photoreduction. To explore the mechanism of IRR on AGO-hv, it is essential to elucidate the interaction between I2 and AGO-hv. The detailed calculations are as follows. The chemisorption of an I2 molecule on the AGO-hv is considered the first necessary step to initialize the IRR on the catalyst. Firstly, we conducted DFT calculations with natural bond orbital (NBO) charge analysis to identify some active sites of the electrophilic and nucleophilic reactions of small molecules on AGO-hv [49]. We determined that I2 can adsorb to the sites of AGO-hv via six pathways (see Fig. 6a and b). Since I2 is an electron-deficient electrophilic reagent, it can more easily attack the negatively charged position. In the NBO charge analysis, the I2 units were positioned separately on different negatively charged sites at a distance of about 2.5 Å to determine the energetically preferred adsorption. Fig. 6b displays all possible
Fig. 5. Nyquist plots (a) and Tafel-polarization curves (b) of the symmetrical cells fabricated with two identical rGO, Pt, AGO and AGO-hv electrodes using I3/I as redox couple with spacer. Inset in image (a): equivalent circuit diagram.
16
H. Jing et al. / Electrochimica Acta 226 (2017) 10–17
Table 2 The energy of I atom at different sites after geometry optimization, the relative energy values, and the calculated binding energy (Eb) between I and sites (in units of eV). *
Active sites
#
Energy/a.u.
#
Relative values Eb/eV
Site 1
Site 3
Site 4
Site 5
Site 6
2401. 2984 1.4627
2401. 3105 1.4505
2402. 7611 0
-2401. 3059 1.4552
2401. 3101 1.4510
0.34
0.45
0.38
0.58
0.79
*
Note: site 2 shows a molecular adsorption state of I2, proving that the interaction between I2 and site 2 is weak, thus the dissociation of I2 is quite difficult. #The energy of I atom at different sites after geometry optimization was calculated to find the lowest energy point, and the relative values were also presented to determine the preferential adsorption site. From the comparison of relative values, site 4 was found to be energetically preferred.
adsorption of I2 on AGO-hv. The total energy of I atom at different sites, the relative energy values, and the calculated binding energy (in units of eV) are listed in Table 2. From Table 2, the most preferential adsorption site was selected as the model to calculate the binding strength of I atom on the electrode surface, with a binding energy of 0.38 eV between I and site 4 (shown in Table 2). The analysis suggested that the most favorable adsorption site is on the pyridine nitrogen at the armchair edges. Therefore, it is reasonable to speculate that the pyridine nitrogen at armchair edges may be the main active sites for IRR. This result was additionally reflected in the bond length difference. The original bond length of I I in the I2 molecule was 2.86 Å and the bond length of I I was elongated to 3.09 Å as the I2 was attached to the site 4, demonstrating that site 4 can adsorb and activate the I2, a prerequisite for the dissociation of I2. The variation of bond length of I2 at different sites in the dissociation adsorption is also shown in Fig. 7. In Fig. 7, an I2 molecule can preferentially adsorb to sites with an elongated I I bond length, which means I2 dissociation and an I atom can bind to these sites. However, when positioning I2 on the AGO-hv with site 2 to search the molecular adsorption and dissociative adsorption, only the molecular adsorption state can be found (see Fig. 7 and Fig. S10). The distance of the I I bond was still 2.86 Å, indicating that the interaction between I2 and site 2 is weak. Thus, the dissociation of I2 is quite difficult, indicating that AGO-hv with N of CONH2 has almost no catalytic activity for IRR. In addition, negative energies indicate thermodynamically favorable processes and the more
Fig. 8. The possible formation pathway for pyridine N.
negative the value of Eb the more exothermic adsorption is. The larger Eb is, the more strongly I species bind on the surface. The moderate binding strength between site 4 and I leads to the high efficiency of I2 dissociation. DFT calculations revealed that the superior properties of AGO-hv are due to the appropriate binding energy of iodine on the sites of pyridine nitrogen at armchair edges (site 4). As shown in Fig. 8, during the initial amination process, NH3H2O reacts with carboxylic acid species to form intermediate amide or amine-like species through nucleophilic substitution. As the time increases, intramolecular dehydration or decarbonylation will occur to generate pyridine N sites [50]. Herein, ozonation contributes to improving the total nitrogen content, amination plays an important role in the formation of pyridine N, and the subsequent photoreduction makes the pyridine N content increase. For comparison, the binding energies of I for the other sites of AGO-hv were also calculated based on similar calculation conditions (see Table 2). The strong affinity of I to AGO-hv significantly reduced the charge-transfer resistance, as
Fig. 7. Schematic illustration of the I2 adsorbed onto the sites of AGO-hv. Atoms in blue, red, grey and purple colors represent N, O, C and I, respectively. The bond lengths of II are marked in red. The bond lengths are given in Angstrom. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
H. Jing et al. / Electrochimica Acta 226 (2017) 10–17
characterized by the smaller semicircle at high-frequency region in the EIS spectra (Fig. 5a). The sequence of preferential adsorption of I atom for the different sites is: site 4 > site 3 > site 6 > site 5 > site 1 > site 2. Taken together, the synergistic contribution from these functional groups on the graphene framework leads to high catalytic activity and electron-donating ability for excellent performance, due to the synergetic effect between the I atom and various groups. 4. Conclusions In summary, AGO-based electrocatalysts decorated by abundant active sites were prepared via a synthetic approach that included ozonization, amination and photoreduction. The IRR performance and mechanism of the produced materials were experimentally and theoretically investigated. The combination of ozonization and modified amination decreased the graphene aggregation and introduced more nitrogen-containing groups. The method of photoreduction further increased the amount of nitrogen active sites, decreased oxygen content, and increased conductivity. In electrochemical measurements, AGO-hv exhibited an excellent PCE of 7.51%. From DFT calculations, we infer that pyridine nitrogen at the armchair edges of AGO-hv served as the active sites for IRR with an appropriate binding energy of 0.38 eV. This green preparation route for graphene-based materials with high nitrogen content and excellent conductivity under mild conditions may facilitate the development of low-cost electrocatalysts and provide a strategy for the efficient fabrication of highperformance CE in DSCs. Acknowledgments The authors deeply appreciate financial assistance from the National Natural Science Foundation of China (Grant No. 21373042). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j. electacta.2016.12.190. References [1] M. Wu, X. Lin, Y. Wang, L. Wang, W. Guo, D. Qi, X. Peng, A. Hagfeldt, M. Grätzel, T. Ma, J. Am. Chem. Soc. 134 (2012) 3419–3428. [2] A. Fakharuddin, R. Jose, T.M. Brown, F.F. Santiago, J. Bisquert, Energy Environ. Sci. 7 (2014) 3952–3981. [3] F. Bella, C. Gerbaldi, C. Barolo, M. Gratzel, Chem. Soc. Rev. 44 (2015) 3431–3473. [4] W. Xiang, D. Chen, R.A. Caruso, Y.B. Cheng, U. Bach, L. Spiccia, ChemSusChem 8 (2015) 3704–3711. [5] S. Galliano, F. Bella, C. Gerbaldi, M. Falco, G. Viscardi, M. Gratzel, C. Barolo, Energy Technol. 4 (2016) 1–13. [6] F. Bella, S. Galliano, M. Falco, G. Viscardi, C. Barolo, M. Grätzel, C. Gerbaldi, Chem. Sci. 7 (2016) 4880–4890. [7] H. Ellis, R. Jiang, S. Ye, A. Hagfeldt, G. Boschloo, Phys. Chem. Chem. Phys. 18 (2016) 8419–8427. [8] J. Gao, Y. Yang, Z. Zhang, J. Yan, Z. Lin, X. Guo, Nano Energy 26 (2016) 123–130.
17
[9] M. Belekoukia, M.S. Ramasamy, S. Yang, X. Feng, G. Paterakis, V. Dracopoulos, C. Galiotis, P. Lianos, Electrochim. Acta 194 (2016) 110–115. [10] I.A. Sahito, K.C. Sun, A.A. Arbab, M.B. Qadir, Y.S. Choi, S.H. Jeong, J. Power Sources 319 (2016) 90–98. [11] C.T.T. Thuy, J.H. Jung, S. Thogiti, W.S. Jung, K.S. Ahn, J.H. Kim, Electrochim. Acta 205 (2016) 170–177. [12] Z. Gao, L. Wang, J. Chang, X. Liu, D. Wu, F. Xu, Y. Guo, K. Jiang, Electrochim. Acta 188 (2016) 441–449. [13] H. Jing, X. Song, S. Ren, Y. Shi, Y. An, Y. Yang, M. Feng, S. Ma, C. Hao, Electrochim. Acta 213 (2016) 252–259. [14] F. Bonaccorso, L. Colombo, G. Yu, M. Stoller, V. Tozzini, A.C. Ferrari, R.S. Ruoff, V. Pellegrini, Science 347 (2015) 1246501–1246509. [15] V. Georgakilas, M. Otyepka, A.B. Bourlinos, V. Chandra, N. Kim, K.C. Kemp, P. Hobza, R. Zboril, K.S. Kim, Chem. Rev 112 (2012) 6156–6214. [16] H. Fang, C. Yu, T. Ma, J. Qiu, Chem. Commun. 50 (2014) 3328–3330. [17] S. Hou, X. Cai, H. Wu, X. Yu, M. Peng, K. Yan, D. Zou, Energy Environ. Sci. 6 (2013) 3356–3362. [18] L. Song, Q. Luo, F. Zhao, Y. Li, H. Lin, L. Qu, Z. Zhang, Phys. Chem. Chem. Phys. 16 (2014) 21820–21826. [19] Y. Xue, J. Liu, H. Chen, R. Wang, D. Li, J. Qu, L. Dai, Angew. Chem. Int. Ed. 51 (2012) 12124–12127. [20] G. Imamura, K. Saiki, J. Phys. Chem. C. 115 (2011) 10000–10005. [21] Y.X. Xu, H. Bai, G.W. Lu, C. Li, G.Q. Shi, J. Am. Chem. Soc. 130 (2008) 5856–5857. [22] L. Lai, L. Chen, D. Zhan, L. Sun, J. Liu, S.H. Lim, C.K. Poh, Z. Shen, J. Lin, Carbon 49 (2011) 3250–3257. [23] S. Pei, H.M. Cheng, Carbon 50 (2012) 3210–3228. [24] A.D. Becke, J. Chem. Phys. 98 (1993) 5457–5462. [25] G.A. Petersson, A. Bennett, T.G. Tensfeldt, M.A. Al-Laham, W.A. Shirley, J. Mantzaris, J. Chem. Phys. 89 (1988) 2193–2218. [26] T.H. Dunning, P.J. Hay, Mod. Theor. Chem. 3 (1977) 1–28. [27] L. Zhang, Z. Xia, J. Phys. Chem. C 115 (2011) 11170–11176. [28] M. Pastore, S. Fantacci, F.D. Angelis, J. Phys. Chem. C 114 (2010) 22742–22750. [29] X. Meng, C. Yu, X. Song, Y. Liu, S. Liang, Z. Liu, C. Hao, J. Qiu, Adv. Energy Mater. 5 (2015) 150018–150026. [30] R.M. Balabin, J. Chem. Phys. 132 (2010) 231101. [31] Z. Li, C. Lu, Z. Xia, Y. Zhou, Z. Luo, Carbon 45 (2007) 1686–1695. [32] H.F. Yang, Y. Yan, Y. Liu, F.Q. Zhang, R.Y. Zhang, Y.M. Mei Li, S.H. Xie, B. Tu, D.Y. Zhao, J. Phys. Chem. B 108 (2004) 17320–17328. [33] X. Wen, J. Wu, D. Gao, C. Lin, J. Mater. Chem. A 4 (2016) 13482–13487. [34] H.L. Wang, J.T. Robinson, X.L. Li, H.J. Dai, J. Am. Chem. Soc. 131 (2009) 9910– 9911. [35] M. Kumar, K. Singh, S.K. Dhawan, K. Tharanikkarasu, J.S. Chung, B.S. Kong, E.J. Kim, S.H. Hur, Chem. Eng. J. 231 (2013) 397–405. [36] S. Stankovich, D.A. Dikin, R.D. Piner, K.A. Kohlhaas, A. Kleinhammes, Y.Y. Jia, Y. Wu, S.T. Nguyen, R.S. Ruoff, Carbon 45 (2007) 1558–1565. [37] J.W. Liu, Y. Zhang, X.W. Chen, J.H. Wang, ACS Appl. Mater. Interfaces 6 (2014) 10196–10204. [38] R. Chen, T. Zhao, J. Lu, F. Wu, L. Li, J. Chen, G. Tan, Y. Ye, K. Amine, Nano Lett. 13 (2013) 4642–4649. [39] G.C. Li, G.R. Li, S.H. Ye, X.P. Gao, Adv. Energy Mater. 2 (2012) 1238–1245. [40] J. Guo, Z. Yang, Y. Yu, H.D. Abruna, L.A. Archer, J. Am. Chem. Soc. 135 (2013) 763– 767. [41] T. Ramanathan, F.T. Fisher, R.S. Ruoff, L.C. Brinson, Chem. Mater. 17 (2005) 1290–1295. [42] L. Huang, Y. Liu, L.C. Ji, Y.Q. Xie, T. Wang, W.Z. Shi, Carbon 49 (2011) 2431–2436. [43] G. Zhou, L.C. Yin, D.W. Wang, L. Li, S. Pei, I.R. Gentle, F. Li, H.M. Cheng, ACS Nano 7 (2013) 5367–5375. [44] D. Hulicova-Jurcakova, M. Kodama, S. Shiraishi, H. Hatori, Z.H. Zhu, G.Q. Lu, Adv. Funct. Mater. 19 (2009) 1800–1809. [45] L. Lai, G. Huang, X. Wang, J. Weng, Carbon 48 (2010) 3145–3156. [46] A. Gromov, S. Dittmer, J. Svensson, O.A. Nerushev, S.A. Perez-García, L. LiceaJiménez, R. Rychwalski, E.E. Campbell, J. Mater. Chem. 15 (2005) 3334–3339. [47] S.H. Hsu, C.T. Li, H.T. Chien, R.R. Salunkhe, N. Suzuki, Y. Yamauchi, K.C. Ho, K.C. W. Wu, Sci. Rep. 4 (2014) 6983–6988. [48] Y. Jia, Y. Shi, J. Qiu, C. Hao, J. Energy Chem. 25 (2016) 861–867. [49] S.A. Seyed Katouli, M. Sheikhi, D. Sheikh, S.F. Tayyari, Orient J. Chem. 29 (2013) 1121–1128. [50] R. Arrigo, M. Hävecker, S. Wrabetz, R. Blume, M. Lerch, J. McGregor, E.P.J. Parrott, J.A. Zeitler, L.F. Gladden, A.K. Gericke, R. Schlögl, D.S. Su, J. Am. Chem. Soc 132 (2010) 9616–9630.
Contents lists available at ScienceDirect
Electrochimica Acta journal homepage: www.elsevier.com/locate/electacta
Ozonization, Amination and Photoreduction of Graphene Oxide for Triiodide Reduction Reaction: An Experimental and Theoretical Study Hongyu Jing, Suzhen Ren* , Yantao Shi, Xuedan Song, Ying Yang, Yanan Guo, Yonglin An, Ce Hao* State Key Laboratory of Fine Chemicals, Dalian University of Technology, Dalian, 116024, Liaoning, China
A R T I C L E I N F O
Article history: Received 28 October 2016 Received in revised form 26 December 2016 Accepted 29 December 2016 Available online 31 December 2016 Keywords: Dye-sensitized solar cells Counter electrode Graphene-based catalysts DFT calculation Photovoltaic performance
A B S T R A C T
This work proposes a mild and environmentally-friendly approach to prepare a highly efficient functional graphene (termed as AGO-hv) using methods of ozone oxidation, solvothermal synthesis, and photoreduction. The use of ozone oxidation in the first step can effectively increase the interlaminar distance between graphite oxide sheets, and create active sites for nucleophilic attack on the epoxy carbon from ammonia. The amino groups were successfully grafted on the surface of graphene as evidenced by the amidation reaction, with a maximum nitrogen content of 10.46 wt% and a C/N molar ratio of 8.46. After further photoreduction of the aminated graphite oxide (AGO), the residual oxygen functionalities, such as C-OH, were effectively removed and the conductivity of the graphene sheet was further recovered. The dye-sensitized solar cell (DSC) exhibited a power conversion efficiency (PCE) of 7.51% based on AGO-hv counter electrode (CE), close to that of Pt counterpart (7.79%). The experimental results indicated that the amidation and photoreduction processes were significantly facilitated by the initial ozonization of graphene oxide, and this process significantly improved the electrochemical activity and the conductivity of graphene oxide. Density functional theory (DFT) calculations revealed that AGOhv had the lowest ionization energy (a better electron-donating ability) and also suitable binding energy with I atoms as well. The combination of ozonization, amination and photoreduction is an efficient route to obtain electrocatalysts with desired compositional distributions and performance for triiodide reduction reaction in DSCs. © 2016 Elsevier Ltd. All rights reserved.
1. Introduction Dye-sensitized solar cells (DSCs) have received much attention due to their low cost, easy fabrication and high efficiency of solar energy conversion [1,2]. These promising photovoltic devices (including a green device of aqueous DSC) are assembled with a TiO2 photoanode, a counter electrode (CE), and a redox couple electrolyte [3–7]. Herein, we focus on the designing of CE component in the liquid I3/I electrolyte using acetonitrile as the solvent. It is well-known that the selected CE in DSC should exhibit high electrical conductivity and electrocatalytic activity for the regeneration of the redox couple. Presently, platinum is one of the most appealing CEs owing to its excellent electrocatalytic activity and high power conversion efficiency (PCE) in DSCs. However, platinum is extremely rare and expensive material, which hinders its large-scale application for DSCs. Alternatives to
* Corresponding authors. E-mail addresses: [email protected] (S. Ren), [email protected] (C. Hao). http://dx.doi.org/10.1016/j.electacta.2016.12.190 0013-4686/© 2016 Elsevier Ltd. All rights reserved.
Pt include carbonaceous materials such as conducting polymers, graphene, and derivatives that are readily available and inexpensive [8–13]. The high specific surface area and excellent conductivity make state-of-the-art carbon materials, such as graphene, promising candidates as an alternative material to platinum [14]. However, pristine graphene sheets are hydrophobic and also exhibit low catalytic activity due to the lack of catalytic active sites for the iodine reduction reaction (IRR) [15]. To address this problem, heteroatoms such as nitrogen, boron, and sulfur have been added to modulate the catalytic activity of graphene. Fang et al. prepared boron-doped graphene by thermal annealing of graphite oxide (GO) with B2O3, which has shown a PCE of 6.73% as a CE for DSCs [16]. Nitrogen-doped graphene (NrG) was obtained by Hou et al. via thermal annealing of a mixture of GO and cyanamide at 900 C in N2 atmosphere, which was subsequently utilized as an alternative to the Pt electrocatalyst for DSCs with the PCE of 5.4% [17]. Song et al. prepared a series of nitrogen-doped porous graphene foams (NPGFs) by hydrothermally treating a mixed
H. Jing et al. / Electrochimica Acta 226 (2017) 10–17
solution of GO and ammonia with optimal PCE of 4.5% for the iodide-based electrolyte [18]. Xue et al., developed threedimensional (3D) N-doped graphene foams (N-GFs) by annealing the freeze-dried graphene oxide foams (GOFs) in ammonia, and found that the resultant DSCs showed a PCE as high as 7.07% [19]. Additionally, other chemical methods to synthesize N-doped graphene have been reported, including chemical vapor deposition, nitrogen plasma treatment, and the reduction of graphene oxide in ammonia or nitrogen gas [20]. Of these, chemical functionalization is considered the most effective method to modify the catalytic activity of graphene [21]. From both theoretical and experimental points of view, we report functionalized graphene, produced by the oxidation of graphene oxide with ozone, amination by inorganic amines of ammonia via solvothermal synthesis, and further photo-reduction by ultraviolet light. In order to enhance the amount of amino groups and prevent the aggregation of the GO, numerous nucleation sites were introduced by ozonization. After reaction with ammonia, the amino groups were grafted onto the preoxidized graphene oxide (GO-O3) via nucleophilic substitution. This modified amination could decrease the damage of the sp2 network of the graphene sheet without sacrificing its conjugative structure. In this case, the amount of oxygen-containing groups in aminated graphite oxide (AGO) were relatively high, thus the residual OH functional groups would be removed via light reduction. The obtained electrocatalyst was denoted as AGO-hv. DFT calculations were used to elucidate the mechanisms of IRR on the synthesized AGO-hv. To the best of our knowledge, the interactions between the functional groups from functional graphene and iodine species have been rarely studied as a critical factor in DSCs. The identification of what type of functional groups serves as the catalytic active site would facilitate the rational design of electrocatalysts for DSCs. Both DFT calculations and experimental results revealed that enhanced performance was partly correlated with the increase in pyridine-like species. The pyridine N and amide moieties introduced may be exploited as the main anchoring site for the IRR. The superior properties of AGO-hv are due to the appropriate binding energy of iodine on the functional group with high conductivity and excellent charge transfer resistance, which improves the PCE (7.51%) to close to that of Pt (7.79%) in the liquid I3/I redox couple electrolyte. 2. Experimental and theoretical methods 2.1. Preparation of catalysts and corresponding CEs The GO-O3 was synthesized by bubbling a certain amount of O3 gas into a GO aqueous suspension by controlling the oxidation time. AGO were synthesized via the modified solvothermal reaction route, using ammonia as nitrogen source, and benzyl alcohol as solvent for the amination of GO-O3 and introduction of amino groups [22]. Then, AGO-hv CEs were prepared by coating
11
AGO films on the FTO glasses and subjected to UV illumination. The preparation of AGO-hv CEs is schematically shown in Scheme 1 (see the Supporting Information for detailed descriptions). The sample of AGO-0 was also prepared based on the above procedures from GO without the addition of O3. Those mentioned samples were also compared to a sample of reduced graphene oxide (rGO) that was produced by a common thermal reduction method described elsewhere [23]. The thickness of all the graphene-based CEs were optimized about 8–10 mm. Additionally, a Pt electrode (200 nm thick) was obtained by drop-casting 0.5 mM H2PtCl6/ethanol solution on a FTO glass substrate. Subsequently, the FTO glass was sintered in a muffle furnace at 450 C for 30 min. 2.2. Structural characterization The obtained samples were characterized by means of X-ray diffraction (XRD), Raman spectroscopy, scanning electron microscopy (SEM), Fourier transform infrared spectroscopy (FTIR), UV–vis absorption spectroscopy, elemental analysis, and X-ray photoelectron spectroscopy (XPS). The details on material characterization are presented in the Supporting Information. 2.3. Electrochemical measurements The current density-voltage (J-V) curves of the DSCs were measured to estimate the photovoltaic performance, which was conducted under simulated AM 1.5 illumination (100 mW cm2, Solar Light Co., INC., USA). The electrochemical properties were investigated by CV, EIS, and Tafel-polarization measurements. The details are presented in the supporting information. 2.4. Computational methods All calculations were performed using Gaussian 09 package based on DFT methods with the B3LYP hybrid functional [24] and 6–31 g (d, p) basis sets [25] for C, H, N and lanl2dz [26] for I atom. The edges of the graphene sheets are terminated by H atoms [27]. The FT-IR spectroscopy (scaled by 0.961) of the models were calculated based on the geometric optimization. Solvation effects were taken into account using the conductor-like polarizable continuum model (CPCM) [28] for all calculations. The ionization energy (IP) is calculated as: IP = EM+ EM, where + EM and EM are the energies of a model with positive charge and a neutral molecule without charges on the ground state [29]. The binding energy (Eb) between I and the binding site is defined as: Eb = E (Site) + 1/2 E (I2) E (I/Site) + E (BSSE). BSSE (basis set superposition error) correction has been taken into account in all the calculations of binding energy [30]. The E (I/Site), E (Site) and E (I2) represent the total energy between I and the site, the energy of AGO-hv, and the energy of the isolated I2 molecule. All the models are shown in the most stable configuration and acetonitrile solvent effects are also taken into account.
Scheme 1. Schematic illustrations of preparation of AGO-hv CE using ozone oxidation, solvothermal synthesis and light irradiation method.
12
H. Jing et al. / Electrochimica Acta 226 (2017) 10–17
We considered the different types and positions of doped nitrogen as follows: NH2 (primary amine group), ¼N H (primary imine group), and CONH2 (carboxylic amide group) on the graphene sheet and pyridine nitrogen at armchair and zigzag edges. Notably, based on the results of XPS in the experiment, we established initial theoretical models of the catalysts. Additionally, by comparing the experimental and calculated results for FT-IR vibrational frequency and UV–vis absorption spectroscopy, we determined the established geometries are reasonable. DFT calculations were also performed by taking iodine and single-layer AGO-hv as the model system to further investigate their interaction. The top view and side view of the optimized structures of graphene-based electrocatalysts are illustrated in Scheme 2. 3. Results and discussion 3.1. Analysis of product characteristics SEM characterization shows that the resultant AGO consists of a crosslinked structure (see Fig. 1a). AGO-hv, in Fig. 1b, displays a wrinkled structure with a somewhat rough surface, indicating that the oxygen-containing groups of AGO were further decreased by photoreduction. The wrinkled structure increases the surface area of nanosheets and reduces their interlayer p-p stacking process, thereby facilitating the formation of porous films. The structural characteristics of the GO-O3 and AGO-hv were determined by XRD. As shown in Fig. S2a, the sharp and strong peak at 10.7 is attributed to the (001) crystal face of GO-O3, indicating that graphite is completely oxidized. For AGO samples subject to different lengths of illumination time, no diffraction peaks from GO-O3 appear due to effective reduction. The crystal face of (001) completely disappears and a new peak at about 25.0 is observed simultaneously, which corresponds to the (002) face of a graphite structure [31]. The stronger the peak, the higher the degree of graphite stacked. As the light reduction time increased, this peak became weaker and broadened, indicating that the structure of photo-reduced AGO is amorphous and the internal arrangement of crystal planes is not as regular as natural graphite which presents a laminated structure. After amination and further light reduction for 1 h, only one weak and broad diffraction peak centered at about 25.0 was detected and the peak at 42.5 diffractions was dramatically weakened and even disappeared, indicating the effective reduction of AGO [32,33]. It is well-known that the weaker the peak of functionalized graphene oxide, the greater the reduction degree of graphene oxide.
Fig. 1. SEM morphology of (a) AGO and (b) AGO-hv.
Fig. S2b displays the Raman spectroscopy of GO-O3, AGO, and AGO-hv (photo-reduction for 1 h, also designed as AGO–1 h). They all exhibited two peaks at 1347 and 1589 cm1, corresponding to the disordered (D) and graphitic (G) band, respectively [34]. The
Scheme 2. The top view and side view of the optimized structures of graphene-based electrocatalysts. In the models, the gray, blue, red, and white balls represent C, N, O, and H atoms, respectively. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
H. Jing et al. / Electrochimica Acta 226 (2017) 10–17
increase in ID/IG intensity ratio from 0.99 for GO-O3 to 1.15 for AGOhv indicates that GO-O3 was reduced due to the restoration of sp2 networks and the removal of oxygen moieties upon UV irradiation [35]. Normally, the ID/IG ratio should decrease in the process of reduction of GO-O3. However, the ID/IG ratio of AGO-hv is higher than that of GO-O3. This change suggests a decrease in the average size of the sp2 domains upon reduction of the GO-O3 [36]. In order to study the effect of the ozonization time on the population of nitrogen and oxygen species, the amount of N contents in AGO was determined by element analysis, and the results are summarized in Table S1. For an oxidation time of 10 min, the nitrogen content reached a maximum of 10.46% and the calculated C/N value was 8.46. A suitable ozonation time is a prerequisite for introducing a high amount of nitrogen-containing moieties onto the surface. Therefore, the optimum time of ozonation was selected as 10 min for the following process. Fig. S3a shows the FT-IR spectroscopy of GO-O3, AGO, and AGOhv. In the spectrum of GO-O3, the peaks at 1726, 1610, 1230, 1045, and 852 cm1 can be ascribed to the vibrations of carboxyl (-COOH) stretching, skeletal ring stretching, hydroxy ( OH) stretching, epoxy (C O C) stretching, and carbon backbone deformation, respectively [37]. Those characteristic peaks are also consistent with the characteristic vibrations of GO [38]. The removal of oxygen-containing groups by amination and photoreduction are evident by FT-IR. After amination, the intensity of these peaks is greatly suppressed and new peaks appear at 1540, 1110–1460 and 3708 cm1 that can be assigned to the antisymmetric C N stretching vibrations coupled with out-of-plane NH2 and NH modes, as well as N-H stretching vibrations [39,40]. The absorption peaks at 540–811 cm1 can be ascribed to the stretching vibrations of C¼C presented in the graphite edge. After photoreduction, however, in the case of AGO-hv, most peaks decreased or vanished in the FT-IR spectrum, proving that most of the oxygen-containing groups were removed by reduction. The scissoring in-plane bending mode of primary NH2 groups is observed at 1662 cm1 by overlapping with the peaks from C C aromatic ring modes at 1627 cm1 [41]. The swing out-of-plane vibration mode of NH2 in CONH2 is also observed at 667 cm1, and the peaks at 1193, and 1553 cm1 can be assigned to carbon backbone deformation. The additional peak at 1386 cm1 is attributed to the swing vibration mode of N-H in C¼NH. The two peaks at 3840 and 3735 cm1 could be ascribed to the antisymmetric and the symmetric N-H stretching vibrations, respectively. Additionally, the absorption peak at about 3448 cm1 reveals the existence of the O-H stretching vibrations from hydroxyl groups on the graphite surface and weakly adsorbed water molecules. The peak at 2368 cm1 may be due to the absorption peak of CO2 in the air during testing. The FT-IR spectrum showed that the peak intensity decreased, indicating that light reduction is an effective method for removing the oxygen-containing functional groups presented in the AGO. Actually, the group of C¼O moves to the low frequency region, indicating the successful introduction of NH2. The types of these amino related species are not affected by further illumination. Note that the absorption peaks of GO-O3, AGO and AGO-hv in the experimental are approximately consistent with the calculated values in Fig. S3b. The UV–vis absorption spectroscopy of GO-O3, AGO and AGOhv are depicted in Fig. S4. From Fig. S4a, the strong absorption peak of GO-O3 is located in 232 nm, which is due to the p–p* transition of C¼C. For the AGO and AGO-hv, their absorption bands gradually shift to 243 nm and 252 nm with a red shift after amination and 1 h of illumination. This may be due to the delocalization of pelectrons in the graphene ring in GO-O3 probably due to both reduction and increasing conjugative interaction, which signifies the restoration of the graphene sp2 structure and the removal of oxygen functionality [42]. It should be noted that the absorption
13
Fig. 2. XPS survey of GO-O3, AGO and AGO-hv samples. The inset table shows the corresponding element compositions and the C, N, and O contents estimated from XPS measurements. The table also summarises the C/O ratios calculated from the atomic concentrations.
peaks of GO-O3, AGO and AGO-hv in the experimental are also approximately consistent with the calculated values in Fig. S4b. XPS measurement was performed to verify the variation of elemental composition and functional groups of the as-prepared samples. Fig. 2 displays the XPS survey spectroscopy for GO-O3, AGO and AGO-hv. The three peaks located at 283 eV, 400 eV and 530 eV are attributed to C1s, N1s and O1s, respectively. The peak intensity of O1s from AGO was significantly weakened and the corresponding peak intensity of AGO-hv was further weakened after light reduction. The emergence of new signal peaks of N 1 s demonstrates that the introduction of nitrogen into AGO and AGOhv relative to GO-O3. Further analysis for element composition shows a C/O ratio of 8.59 and 10.60 for AGO and AGO-hv, respectively, confirming the further removal of oxygen-containing functional groups. From the high-resolution C 1s spectrum of GO-O3 (Fig. S5), the peak centered at 284.4 eV is assigned to C-C/C¼C and the peaks at 285.0 eV, 286.4 eV and 288.5 eV are ascribed to C OH, C O C, and O¼C-OH groups, respectively [43]. As shown in Fig. 3a, new functional groups of C-N, C¼O appear, and the O¼C OH, COC vanish, but C OH still exists for C1s of AGO compared with GO-O3. Note that the four peaks are observed from the C 1s spectrum of AGO, but only three peaks are evident for AGO-hv (see Fig. 3b). As compared to AGO, the peak at 285.0 eV corresponds to C OH disappears, demonstrating the effective removal of OH groups from AGO and partial restoration of the conjugated graphene sheets via light reduction. The presence of amine groups in the AGO and AGO-hv samples is also confirmed by the peak at 285.8285.9 eV for the C N bond in the primary amine [41]. For the deconvoluted N 1s region of AGO and AGO-hv (see Fig. 3c and d), three signal peaks are observed, which demonstrates the existence of C¼N, C-NH2 and pyridine nitrogen, respectively [44–46]. Curve fitting results of C 1s and N 1s are summarized in Table S2 and Table S3. Interestingly, the amount of C-NH2 and C¼N groups decreased from AGO to AGO-hv (from 22.22 to 17.99 at% for C-NH2, and from 45.68 to 33.19 at% for C¼N), but the amount of pyridine N species in AGO-hv increased to a maximum of 48.82 at%, compared with AGO (32.11 at%). The type of C-OH species contained in AGO decreased upon light illumination, and the other types of carbon and nitrogen containing functional groups were well maintained, suggesting the photoreduction method can remove the oxygen-containing functional groups such as C-OH instead of nitrogen types. Taken together, the XPS results show the
14
H. Jing et al. / Electrochimica Acta 226 (2017) 10–17
Fig. 3. High-resolution spectroscopy: (a) and (b): C1s for AGO and AGO-hv; (c) and (d): N1s for AGO and AGO-hv.
successful functionalization of GO-O3 with nitrogen-containing functional groups. Light reduction does not affect nitrogen-doping types, but does change the compositional distributions of nitrogen-containing groups and removes the C-OH. 3.2. Electrochemical performance
Fig. 4. The J-V curves of rGO, Pt, AGO-0, AGO and AGO-hv as CEs for the liquid I3/I redox couple electrolyte.
The photovoltaic performance of the solar cell devices was next investigated. The current density-voltage (J-V) curves and the corresponding photovoltaic parameters are presented in Fig. 4 and Table 1. Compared to Pt CE, the device made from AGO-hv CE showed a comparable Jsc value, and yielded a similar device efficiency (7.51%) compared to the Pt counterpart (7.79%), consistent with the below electrochemical characterizations. Additionally, the poor performance exhibited by rGO may be attributed to ineffective charge transport and high sheet resistance. EIS and Tafel-polarization measurement were used to further confirm the electrocatalytic activities of the CEs. The Nyquist plots obtained from symmetric cells with different CEs are shown in Fig. 5a. From the equivalent circuit used for EIS fitting (inset in Fig. 5a), the series resistance (Rs) correspondes to high frequency intercept on the real axis. The charge transfer resistance (Rct)
H. Jing et al. / Electrochimica Acta 226 (2017) 10–17
15
Table 1 The photovoltaic parameters of rGO, Pt, AGO-0, AGO and AGO-hv as CEs. CEs
Voc/V
Jsc/mA cm2
FF
PCE/%
rGO AGO-0 AGO AGO-hv Pt
0.76 0.76 0.76 0.77 0.78
12.36 12.92 13.36 14.00 14.20
0.61 0.67 0.71 0.69 0.70
5.70 6.54 7.20 7.51 7.79
indicates the semicircles in the high-frequency range between the CE and electrolyte and is mainly related to the electrocatalytic activity for the reduction of triiodine ions [47]. Smaller Rct values imply lower internal resistance and higher catalytic activity of a CE, which enhances device performance and is shown in the improvement of Jsc. The Jsc values for the rGO and AGO cell are smaller than those for the AGO-hv and Pt cells, in accordance with the results of Rct. The EIS results reveal that the catalytic activities increase in the order of rGO < AGO < AGO-hv < Pt, consistent with the Tafel results described below. Tafel curves were measured to further investigate the catalytic activities. In theory, the Tafel curve can be divided into 3 zones. The curve at the low potential is attributed to the polarization zone, the curve at middle potential with a sharp slope is attributed to the Tafel zone, and the curve at high potential in the horizontal part is attributed to the diffusion zone. In the latter two zones, we can obtain information on the limiting diffusion current density (Jlim) and the exchange current density (J0), closely related to the catalytic activity of the catalysts. Fig. 5b shows the logarithmic current density as a function of voltage. As shown in Fig. 5b, the Tafel plot of AGO-hv sample exhibits a similar slope as that of the Pt electrode, indicating its good electrocatalytic activity. The J0 of Pt is slightly larger than that of AGO-hv sample, which means that Pt is more effective than AGO-hv in catalyzing triiodide reduction. However, AGO-hv has a larger limiting diffusion current density than the Pt electrode. This means AGO-hv can catalyze reducing I3 to I as effectively as Pt. The Tafel-polarization results were consistent with the EIS and J-V experiments. 3.3. DFT calculations The optimized geometric structures of the graphene-based electrocatalysts performed with DFT are shown in Scheme 2.
Fig. 6. (a) NBO charge analysis and (b) six possible binding sites on AGO-hv.
According to DFT calculation, we confirmed that step III (the process of electron transfer for IRR) was the rate-determining step (see Fig. S9 in Supporting Information for details). To rationally design the catalyst of IRR, a lower value of ionization energy (IP) of materials is desired for DSCs. We next calculated the IP value, which represents the ability of catalyst to release electrons (presented in Table S5). The AGO-hv material showed the lowest IP (79.38 kcal mol1) was located on the top of the volcanic plot. This was also lower than the materials tested in our previous work (AFG material, IP = 81.79 kcal mol1 [48]). This result suggests that the AGO-hv catalyst has good electron-donating ability, suggesting potential function as excellent CEs with better catalytic activity in DSCs. The lowest IP value obtained from theoretical calculation agreed with the experimental results. The lowest IP corresponded to the strongest electron-donating ability, due to the decreased oxygen-containing groups after photoreduction. To explore the mechanism of IRR on AGO-hv, it is essential to elucidate the interaction between I2 and AGO-hv. The detailed calculations are as follows. The chemisorption of an I2 molecule on the AGO-hv is considered the first necessary step to initialize the IRR on the catalyst. Firstly, we conducted DFT calculations with natural bond orbital (NBO) charge analysis to identify some active sites of the electrophilic and nucleophilic reactions of small molecules on AGO-hv [49]. We determined that I2 can adsorb to the sites of AGO-hv via six pathways (see Fig. 6a and b). Since I2 is an electron-deficient electrophilic reagent, it can more easily attack the negatively charged position. In the NBO charge analysis, the I2 units were positioned separately on different negatively charged sites at a distance of about 2.5 Å to determine the energetically preferred adsorption. Fig. 6b displays all possible
Fig. 5. Nyquist plots (a) and Tafel-polarization curves (b) of the symmetrical cells fabricated with two identical rGO, Pt, AGO and AGO-hv electrodes using I3/I as redox couple with spacer. Inset in image (a): equivalent circuit diagram.
16
H. Jing et al. / Electrochimica Acta 226 (2017) 10–17
Table 2 The energy of I atom at different sites after geometry optimization, the relative energy values, and the calculated binding energy (Eb) between I and sites (in units of eV). *
Active sites
#
Energy/a.u.
#
Relative values Eb/eV
Site 1
Site 3
Site 4
Site 5
Site 6
2401. 2984 1.4627
2401. 3105 1.4505
2402. 7611 0
-2401. 3059 1.4552
2401. 3101 1.4510
0.34
0.45
0.38
0.58
0.79
*
Note: site 2 shows a molecular adsorption state of I2, proving that the interaction between I2 and site 2 is weak, thus the dissociation of I2 is quite difficult. #The energy of I atom at different sites after geometry optimization was calculated to find the lowest energy point, and the relative values were also presented to determine the preferential adsorption site. From the comparison of relative values, site 4 was found to be energetically preferred.
adsorption of I2 on AGO-hv. The total energy of I atom at different sites, the relative energy values, and the calculated binding energy (in units of eV) are listed in Table 2. From Table 2, the most preferential adsorption site was selected as the model to calculate the binding strength of I atom on the electrode surface, with a binding energy of 0.38 eV between I and site 4 (shown in Table 2). The analysis suggested that the most favorable adsorption site is on the pyridine nitrogen at the armchair edges. Therefore, it is reasonable to speculate that the pyridine nitrogen at armchair edges may be the main active sites for IRR. This result was additionally reflected in the bond length difference. The original bond length of I I in the I2 molecule was 2.86 Å and the bond length of I I was elongated to 3.09 Å as the I2 was attached to the site 4, demonstrating that site 4 can adsorb and activate the I2, a prerequisite for the dissociation of I2. The variation of bond length of I2 at different sites in the dissociation adsorption is also shown in Fig. 7. In Fig. 7, an I2 molecule can preferentially adsorb to sites with an elongated I I bond length, which means I2 dissociation and an I atom can bind to these sites. However, when positioning I2 on the AGO-hv with site 2 to search the molecular adsorption and dissociative adsorption, only the molecular adsorption state can be found (see Fig. 7 and Fig. S10). The distance of the I I bond was still 2.86 Å, indicating that the interaction between I2 and site 2 is weak. Thus, the dissociation of I2 is quite difficult, indicating that AGO-hv with N of CONH2 has almost no catalytic activity for IRR. In addition, negative energies indicate thermodynamically favorable processes and the more
Fig. 8. The possible formation pathway for pyridine N.
negative the value of Eb the more exothermic adsorption is. The larger Eb is, the more strongly I species bind on the surface. The moderate binding strength between site 4 and I leads to the high efficiency of I2 dissociation. DFT calculations revealed that the superior properties of AGO-hv are due to the appropriate binding energy of iodine on the sites of pyridine nitrogen at armchair edges (site 4). As shown in Fig. 8, during the initial amination process, NH3H2O reacts with carboxylic acid species to form intermediate amide or amine-like species through nucleophilic substitution. As the time increases, intramolecular dehydration or decarbonylation will occur to generate pyridine N sites [50]. Herein, ozonation contributes to improving the total nitrogen content, amination plays an important role in the formation of pyridine N, and the subsequent photoreduction makes the pyridine N content increase. For comparison, the binding energies of I for the other sites of AGO-hv were also calculated based on similar calculation conditions (see Table 2). The strong affinity of I to AGO-hv significantly reduced the charge-transfer resistance, as
Fig. 7. Schematic illustration of the I2 adsorbed onto the sites of AGO-hv. Atoms in blue, red, grey and purple colors represent N, O, C and I, respectively. The bond lengths of II are marked in red. The bond lengths are given in Angstrom. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
H. Jing et al. / Electrochimica Acta 226 (2017) 10–17
characterized by the smaller semicircle at high-frequency region in the EIS spectra (Fig. 5a). The sequence of preferential adsorption of I atom for the different sites is: site 4 > site 3 > site 6 > site 5 > site 1 > site 2. Taken together, the synergistic contribution from these functional groups on the graphene framework leads to high catalytic activity and electron-donating ability for excellent performance, due to the synergetic effect between the I atom and various groups. 4. Conclusions In summary, AGO-based electrocatalysts decorated by abundant active sites were prepared via a synthetic approach that included ozonization, amination and photoreduction. The IRR performance and mechanism of the produced materials were experimentally and theoretically investigated. The combination of ozonization and modified amination decreased the graphene aggregation and introduced more nitrogen-containing groups. The method of photoreduction further increased the amount of nitrogen active sites, decreased oxygen content, and increased conductivity. In electrochemical measurements, AGO-hv exhibited an excellent PCE of 7.51%. From DFT calculations, we infer that pyridine nitrogen at the armchair edges of AGO-hv served as the active sites for IRR with an appropriate binding energy of 0.38 eV. This green preparation route for graphene-based materials with high nitrogen content and excellent conductivity under mild conditions may facilitate the development of low-cost electrocatalysts and provide a strategy for the efficient fabrication of highperformance CE in DSCs. Acknowledgments The authors deeply appreciate financial assistance from the National Natural Science Foundation of China (Grant No. 21373042). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j. electacta.2016.12.190. References [1] M. Wu, X. Lin, Y. Wang, L. Wang, W. Guo, D. Qi, X. Peng, A. Hagfeldt, M. Grätzel, T. Ma, J. Am. Chem. Soc. 134 (2012) 3419–3428. [2] A. Fakharuddin, R. Jose, T.M. Brown, F.F. Santiago, J. Bisquert, Energy Environ. Sci. 7 (2014) 3952–3981. [3] F. Bella, C. Gerbaldi, C. Barolo, M. Gratzel, Chem. Soc. Rev. 44 (2015) 3431–3473. [4] W. Xiang, D. Chen, R.A. Caruso, Y.B. Cheng, U. Bach, L. Spiccia, ChemSusChem 8 (2015) 3704–3711. [5] S. Galliano, F. Bella, C. Gerbaldi, M. Falco, G. Viscardi, M. Gratzel, C. Barolo, Energy Technol. 4 (2016) 1–13. [6] F. Bella, S. Galliano, M. Falco, G. Viscardi, C. Barolo, M. Grätzel, C. Gerbaldi, Chem. Sci. 7 (2016) 4880–4890. [7] H. Ellis, R. Jiang, S. Ye, A. Hagfeldt, G. Boschloo, Phys. Chem. Chem. Phys. 18 (2016) 8419–8427. [8] J. Gao, Y. Yang, Z. Zhang, J. Yan, Z. Lin, X. Guo, Nano Energy 26 (2016) 123–130.
17
[9] M. Belekoukia, M.S. Ramasamy, S. Yang, X. Feng, G. Paterakis, V. Dracopoulos, C. Galiotis, P. Lianos, Electrochim. Acta 194 (2016) 110–115. [10] I.A. Sahito, K.C. Sun, A.A. Arbab, M.B. Qadir, Y.S. Choi, S.H. Jeong, J. Power Sources 319 (2016) 90–98. [11] C.T.T. Thuy, J.H. Jung, S. Thogiti, W.S. Jung, K.S. Ahn, J.H. Kim, Electrochim. Acta 205 (2016) 170–177. [12] Z. Gao, L. Wang, J. Chang, X. Liu, D. Wu, F. Xu, Y. Guo, K. Jiang, Electrochim. Acta 188 (2016) 441–449. [13] H. Jing, X. Song, S. Ren, Y. Shi, Y. An, Y. Yang, M. Feng, S. Ma, C. Hao, Electrochim. Acta 213 (2016) 252–259. [14] F. Bonaccorso, L. Colombo, G. Yu, M. Stoller, V. Tozzini, A.C. Ferrari, R.S. Ruoff, V. Pellegrini, Science 347 (2015) 1246501–1246509. [15] V. Georgakilas, M. Otyepka, A.B. Bourlinos, V. Chandra, N. Kim, K.C. Kemp, P. Hobza, R. Zboril, K.S. Kim, Chem. Rev 112 (2012) 6156–6214. [16] H. Fang, C. Yu, T. Ma, J. Qiu, Chem. Commun. 50 (2014) 3328–3330. [17] S. Hou, X. Cai, H. Wu, X. Yu, M. Peng, K. Yan, D. Zou, Energy Environ. Sci. 6 (2013) 3356–3362. [18] L. Song, Q. Luo, F. Zhao, Y. Li, H. Lin, L. Qu, Z. Zhang, Phys. Chem. Chem. Phys. 16 (2014) 21820–21826. [19] Y. Xue, J. Liu, H. Chen, R. Wang, D. Li, J. Qu, L. Dai, Angew. Chem. Int. Ed. 51 (2012) 12124–12127. [20] G. Imamura, K. Saiki, J. Phys. Chem. C. 115 (2011) 10000–10005. [21] Y.X. Xu, H. Bai, G.W. Lu, C. Li, G.Q. Shi, J. Am. Chem. Soc. 130 (2008) 5856–5857. [22] L. Lai, L. Chen, D. Zhan, L. Sun, J. Liu, S.H. Lim, C.K. Poh, Z. Shen, J. Lin, Carbon 49 (2011) 3250–3257. [23] S. Pei, H.M. Cheng, Carbon 50 (2012) 3210–3228. [24] A.D. Becke, J. Chem. Phys. 98 (1993) 5457–5462. [25] G.A. Petersson, A. Bennett, T.G. Tensfeldt, M.A. Al-Laham, W.A. Shirley, J. Mantzaris, J. Chem. Phys. 89 (1988) 2193–2218. [26] T.H. Dunning, P.J. Hay, Mod. Theor. Chem. 3 (1977) 1–28. [27] L. Zhang, Z. Xia, J. Phys. Chem. C 115 (2011) 11170–11176. [28] M. Pastore, S. Fantacci, F.D. Angelis, J. Phys. Chem. C 114 (2010) 22742–22750. [29] X. Meng, C. Yu, X. Song, Y. Liu, S. Liang, Z. Liu, C. Hao, J. Qiu, Adv. Energy Mater. 5 (2015) 150018–150026. [30] R.M. Balabin, J. Chem. Phys. 132 (2010) 231101. [31] Z. Li, C. Lu, Z. Xia, Y. Zhou, Z. Luo, Carbon 45 (2007) 1686–1695. [32] H.F. Yang, Y. Yan, Y. Liu, F.Q. Zhang, R.Y. Zhang, Y.M. Mei Li, S.H. Xie, B. Tu, D.Y. Zhao, J. Phys. Chem. B 108 (2004) 17320–17328. [33] X. Wen, J. Wu, D. Gao, C. Lin, J. Mater. Chem. A 4 (2016) 13482–13487. [34] H.L. Wang, J.T. Robinson, X.L. Li, H.J. Dai, J. Am. Chem. Soc. 131 (2009) 9910– 9911. [35] M. Kumar, K. Singh, S.K. Dhawan, K. Tharanikkarasu, J.S. Chung, B.S. Kong, E.J. Kim, S.H. Hur, Chem. Eng. J. 231 (2013) 397–405. [36] S. Stankovich, D.A. Dikin, R.D. Piner, K.A. Kohlhaas, A. Kleinhammes, Y.Y. Jia, Y. Wu, S.T. Nguyen, R.S. Ruoff, Carbon 45 (2007) 1558–1565. [37] J.W. Liu, Y. Zhang, X.W. Chen, J.H. Wang, ACS Appl. Mater. Interfaces 6 (2014) 10196–10204. [38] R. Chen, T. Zhao, J. Lu, F. Wu, L. Li, J. Chen, G. Tan, Y. Ye, K. Amine, Nano Lett. 13 (2013) 4642–4649. [39] G.C. Li, G.R. Li, S.H. Ye, X.P. Gao, Adv. Energy Mater. 2 (2012) 1238–1245. [40] J. Guo, Z. Yang, Y. Yu, H.D. Abruna, L.A. Archer, J. Am. Chem. Soc. 135 (2013) 763– 767. [41] T. Ramanathan, F.T. Fisher, R.S. Ruoff, L.C. Brinson, Chem. Mater. 17 (2005) 1290–1295. [42] L. Huang, Y. Liu, L.C. Ji, Y.Q. Xie, T. Wang, W.Z. Shi, Carbon 49 (2011) 2431–2436. [43] G. Zhou, L.C. Yin, D.W. Wang, L. Li, S. Pei, I.R. Gentle, F. Li, H.M. Cheng, ACS Nano 7 (2013) 5367–5375. [44] D. Hulicova-Jurcakova, M. Kodama, S. Shiraishi, H. Hatori, Z.H. Zhu, G.Q. Lu, Adv. Funct. Mater. 19 (2009) 1800–1809. [45] L. Lai, G. Huang, X. Wang, J. Weng, Carbon 48 (2010) 3145–3156. [46] A. Gromov, S. Dittmer, J. Svensson, O.A. Nerushev, S.A. Perez-García, L. LiceaJiménez, R. Rychwalski, E.E. Campbell, J. Mater. Chem. 15 (2005) 3334–3339. [47] S.H. Hsu, C.T. Li, H.T. Chien, R.R. Salunkhe, N. Suzuki, Y. Yamauchi, K.C. Ho, K.C. W. Wu, Sci. Rep. 4 (2014) 6983–6988. [48] Y. Jia, Y. Shi, J. Qiu, C. Hao, J. Energy Chem. 25 (2016) 861–867. [49] S.A. Seyed Katouli, M. Sheikhi, D. Sheikh, S.F. Tayyari, Orient J. Chem. 29 (2013) 1121–1128. [50] R. Arrigo, M. Hävecker, S. Wrabetz, R. Blume, M. Lerch, J. McGregor, E.P.J. Parrott, J.A. Zeitler, L.F. Gladden, A.K. Gericke, R. Schlögl, D.S. Su, J. Am. Chem. Soc 132 (2010) 9616–9630.