Nitrogen Graphene: A New and Exciting Generation of Visible Light Driven Photocatalyst and Energy Storage Application.

The synthesis of nitrogen, boron, and nitrogen-boron-codoped graphenes was attained via mixing solutions of GO with urea, boric acid, and a mixture of both, respectively, followed by drying in vacuum and annealing at 900 °C for 10 h. These materials were thoroughly characterized employing XRD, TEM, FTIR, Raman, UV-vis, XPS, IPCE%, and electrical conductivity measurements. The nitrogen-doped graphene (NG) showed an excellent supercapacitor performance with a higher specific capacitance (388 F·g-1 at 1 A·g-1), superior stability, and a higher power density of 0.260 kW kg-1. This was mainly due to the designated N types of doping and most importantly N-O bonds and to lowering charge transfer and equivalent series resistances. The NG also indicated the highest photocatalytic performance for methylene blue (MB 20 ppm, power = 160 W, λ > 420 nm) and phenol (5 ppm) degradation under visible light illumination with rate constants equal 0.013 min-1 and 0.04 min-1, respectively. The photodegradation mechanism was proposed via determining the energy band potentials using the Mott-Schottky measurements. This determined that photoactivity enhancement of the NG is accounted for by acquisition of nitrogen-oxy-carbide phases that shared in inducing a higher IPCE% (60%) and a lower band gap value (1.68 eV) compared to boron and nitrogen-boron-codoped graphenes. The achieved photodegradation mechanism relied on scavengers performance suggesting that •OH and electrons were the main reactive species responsible for the MB photodegradation.

Introduction

Although carbon nanotubes (CNTs) offer high surface area and conductivity, they show high contact resistance at the electrode–current collector and high synthesis charges.[1] This indeed hampered CNT scale up processing for large scale manufacturing and instead helped the emergence of graphene as an alternative to overcome the limitations of carbon-based materials. As a result, graphene is presently the most common 2D material used in energy storage devices due to its chemical stability as well as its fast electronic transportation, thermal stability, and superior mechanical characteristics.[2−4] Boron and nitrogen doping into the graphene lattice generates polarization in the carbon network due to discrepancies in the electronegativities of B (2.05), N (3.04), and C (2.55), thus regulating the electronic, optical, magnetic, and electrochemical properties of graphene. This ternary system constitutes a very rich system with a large extent of miscibility granted by the comparable atomic radii of the components, and thus little perturbation in the 2D hexagonal sheets of graphene is expected.[5] Doping with B and N can form p-type and n-type structures respectively, offering a means to modify the semiconductor properties. The electronic structure (e.g., band gap, charge density, and spin density) and chemical reactivity can also be tuned by forming ternary B–C–N structures of different bonding configurations, with final properties being gradually altered by regulating the elemental compositions as well as their amounts.[6,7] Thus, B,N-codoped graphene holds great potential for applications in catalysis and energy storage. Although much research has been done to synthesize B- or N-doped or B/N-codoped graphene by decomposing B-, C-, and N-rich precursors via thermal and plasma-enhanced chemical vapor deposition (CVD) or incorporating dopants via thermal annealing,[8] it still remains a great challenge to exactly control the doping arrangement at the atomic scale and their concentrations. The approaches based on CVD for the synthesis of dual-doped graphene are found tedious and costly for mass production. A reported chemical method regarding the coannealing of solution-exfoliated graphene oxide in the presence of ammonia and boric acid was straightforward;[8,9] however, it created a covalent boron–carbon–nitride (BCN) rather than B- and N-codoped graphene (B,N-graphene). Such unavoidable production of the hexagonal boron nitride (h-BN) is inevitably obtained, which is chemically inert in view of its catalytic activity. Despite tremendous progress in the synthesis of ternary BCN heteroring nanomaterials,[10] a high phase purity B,N-graphene in which B and N replace C atoms in the graphene framework at well-defined doping sites has not been achieved. This indeed makes the theoretical identification of the function of guest dopants impractical. In this study, a h-BN-free B,N-graphene was prepared by a two-stepwise doping method. This method not only enables the inclusion of heteroatoms at selected graphene sites but also limits the formation of inactive byproducts and thus is expected to induce a synergistic activity enhancement for the B,N-graphene configuration. This catalyst, for example, shows excellent activity and perfect (nearly 100%) selectivity in oxygen reduction reaction (ORR) in an alkaline medium compared with that observed for singly B- or N-doped graphenes.[11]The B,N-graphene also shows a complete tolerance for methanol oxidation and excellent durability, superior to those observed for the commercially used Pt/C catalysts.[12−14] A rare work was done on the codoped N,B-graphene and mainly concentrated on ORR and microwave absorption.[15] However, much work was done on N-doped graphene synthesized via different nitrogen precursors and conditions to be in used in batteries, sensors, and ultracapacitors.[15−17] Also, highly dispersed boron-doped graphene nanoribbons (B-GNRs), synthesized via a simple vacuum activation method, were used for the degradation of Rhodamine B based on their excellent conductivity and photocatalytic activity.[18] However, the latter photocatalyst worked only under UV irradiation to photoexcite electrons that were then consumed to form O2–, while the holes on B-GR were transported to the HOMO of RhB stimulating its degradation. This mechanism has been contradicted by Tang et al.,[19] who put forth a theory where the degradation of RhB stems from its sensitization to form a dye excited state followed by the electron transfer from this LUMO of RhB (−3.08 V vs vacuum) to the more positive Fermi plane of graphene (−4.42 V vs vacuum), where it combines with surface adsorbed O2 to generate reactive oxidative species.

Accordingly, in a way of synthesizing a metal-free graphene codoped with N and B with reference to those of individual analogue, two stepwise vacuum-annealing pathways were adopted for improving their photocatalytic qualities to be performed under visible light illumination, despite the acknowledged fact of the nonphotocatalytic action of graphenes, as well as their supercapacitor performances selected for higher power density and cycle efficacy. The proven adsorption capabilities of graphenes were nullified in this work for providing photocatalytic active sites on their surfaces for comprehensive organic oxidation, unlike those in the literature in which graphenes were combined with a metal or a semiconducting material.[20] The structure, surface, optical, and electrical conductivity of the as-synthesized N-doped (NG), B-doped (BG), and NB-codoped graphenes (NBG) was evaluated by a variety of complementary techniques, including transmission electron microscopy (TEM), Raman spectroscopy, FTIR, N2 sorptiometry, X-ray photoelectron spectroscopy (XPS), X-ray diffraction (XRD), UV–vis diffuse reflectance (UV–vis), incident photon to current efficiency (IPCE%), and electrical conductivity measurements. The specific capacitance and the outstanding cycle stability of the electrodes as supercapacitors were estimated via electrochemical studies (CV, charge–discharge, and EIS), together with evaluating methylene blue (MB) and phenol degradation via photocatalysis activities under visible light illumination. This allows elaboration of the effect of the fabrication conditions of N, B, and NB graphenes, their doping percentages, distributions in graphene structures, and optical, and the electronic properties on the consequences of the executed applications.

Results and Discussion

Bulk, Surface, and Morphology Characteristics

In order to determine the interlayer spacing of GO and the corresponding changes thereof following the doping with N, B, and NB, XRD patterns are measured and presented in Figure . The GO pattern shows a prominent crystal plane (001) at a spacing of 0.71 nm exceeding that of graphite (0.336 nm), proposing the involvement of oxygen-containing groups at the edges and on every layer of GO. Additionally, a very weak reflection plane at 2θ = 43° (100) (with spacing about 0.165 nm) is also revealed, probably referring to the incomplete oxidation. Doping at the mentioned conditions stimulated the production of reduced graphene oxide (RGO) exhibiting a shift to the typical diffraction peak attained at 2θ = 24.5°; due to the crystal lattice (002) developed without doping, to 26.41°, 26.63°, and 26.36° for NG, BG, and NBG, respectively. Shifting the peaks to higher angles is attributed to the dislocation and defects introduced following the insertion of N, B, and NB during the annealing process. Decrease of the lattice spacing to 0.337, 0.334, and 0.339 nm in NG, BG, and NBG, respectively, compared to the free doped RGO (0.363 nm) can further support the reduction of RGO at the performed high annealing temperature. Varying the broadness degree of the (002) peak, as well as its intensity ratio, before and after doping could give an idea about not only about the successful adoption of N and B but also on the fair restacking of the graphene nanosheets. Accordingly, the weakened intensity as well as broadened width of the reflection at 2θ = 26.6° of the B-doped graphene compared to the N-graphene (2θ = 26.41°) is taken as a diagnostic for increased defects produced following the boron insertion, destroying the graphene crystal structure. Also, diffusion of the reflection at (100) after doping compared to that in GO gives a hint about decreasing the number of oxygen atoms in this plane. The doping of GO was further supported by the elemental chemical analysis in which the percentage of oxygen was reduced from 20.8% for GO to 17.0% for NG, whereas for NBG, the oxygen was reduced to 12.5%, and the carbon percentage was decreased from 80.2% in NG to 71.2% in NBG. Of particular interest, the nitrogen percentages in NG and NBG were 2.7% and 7%, respectively. Increasing C/O ratio for NBG compared with NG is indicative of increasing reduction in the former material, comparatively.

Figure 1

XRD patterns of GO, RGO, NG, BG, and NBG.

In order to investigate the morphology and the nanoscopic details of the systems, TEM images of NG, BG, and NBG are illustrated in Figure . The TEM image in Figure a shows that NG sheets densely overlay each other forming nodes together and exhibiting apparent wrinkled and folded sheets. The lower magnification as well as the higher one (inset) shows that NG is porous in nature with an average aperture of 38 nm. In the case of BG, few graphene layers are clearly seen (∼5 layers) with a more transparent image (Figure b) than that exhibited by NG. Apart from the smooth and wrinkled morphologies, BG shows spherical aggregates or defects on its surface, which upon magnification (inset) reveal a leafy tree like structure of 350 nm dimension. On the other hand, the TEM image of NBG (Figure c) shows layered structures with higher thickness and many defects, identified as black spots. Magnification of one of the defect areas illustrates that the nanosheets are full of pores with an average diameter of 11 nm. On the other hand, the TEM image of GO shows that it possesses several thin two-dimensional sheets with a large amount of wrinkles (Figure d), configured due to randomly overlaid aggregated sheets.

Figure 2

TEM Images of (a) NG, (b) BG, (c) NBG, and (d) GO together with their corresponding magnified HRTEM images, as insets.

The developed mesopore structure was well established from the N2 adsorption–desorption isotherms (Figure , left). All samples show isotherms of type IV with H2 hysteresis loop indicative of capillary condensation in mesopores. According to the pore size distribution (PSD) curves developed by BJH theory, a hierarchical structure of pore diameter around 1.9 nm was attained (Figure , right). The doping with boron created the highest SBET (110 m2/g), the largest pore volume (0.158 cm3/g), and a pore width of 19.06 Å, comparable to those exhibited via PSD curves for all doped samples. Contrarily, NBG possesses the lowest SBET (37 m2/g) and pore volume (0.07 cm3/g) values, while NG shows median values between the other samples (SBET of 60 m2/g, and pore volume of 0.13 cm3/g). One can suppose that the increase in surface area of BG is due to the high mesoporosity recognized via the hysteresis loop closed at p/po of 0.4, whereas for NG and NBG the hysteresis loops closed at p/po of 0.1, confirming the production of narrower pores. Despite that, NG still remains the highest mesoporous doped graphene sample due to the enhancement of its hysteresis loop area, comparatively.

Figure 3

Adsorption–desorption isotherms (left) of NG, BG, and NBG samples together with their corresponding pore size distribution curves (right).

XPS is used to study the nitrogen and boron configuration after doping in graphene via monitoring the peaks appearing at about 400 and 284 eV corresponding, respectively, to the N 1s and C 1s. The C 1s spectrum of NBG (Figure ) is deconvoluted into four peaks centered at 283.8, 284.9, 286.2, and 288.1 eV composed, respectively, of boron carbide, C–C, C–N, and C–O moieties. The detailed chemical compositions of GO, NG, BG, and BNGs are listed in Table . The percentages of B and N in the as-grown B- and N-graphenes as calculated from the XPS survey spectrum were found to be 8.4% and 2.7%, respectively. The contribution shown at the higher binding energy (288.1 eV) is possibly correlated to the carboxylic acid groups at the edges of GO. In N-graphene, the N 1s spectrum is deconvoluted to several individual peaks that are assigned to pyridinic N (398.1 eV), pyrrolic N (399.6 eV), graphitic N (401.3 eV), and N oxides (402.8 eV) (Figure ). Their amounts are decreasing in the same sequence. The varying peak position of N oxides (402.8 eV) compared to others in the literature may be due to the varying charges of nitrogen and its neighbor atoms. Similarly, the B 1s spectrum (Figure S1) is resolved into three peaks at the binding energies of 190.4, 191.2, and 192.6 eV consistent with the structure of BN, BC3, and partially oxidized B (e.g., BCO2 and BC2O), respectively.[9,19]

Figure 4

XPS spectra: C 1s of NBG and N 1s of NG.

Table 1

Elemental Compositions of NG, BG, and NBG

	C (atom %)	O (atom %)	B (atom %)	N (atom %)
GO	79.0	20.8
NG	80.2	17.0		2.7
BG	75.4	15.2	8.4
NBG	71.2	12.5	8.0	7.0

FTIR and Raman Spectroscopy

The FTIR spectra of GO as well as the doped graphenes are presented in Figure a. The GO spectrum shows small bands at 1720, 1630, 1383, 1222, 1123, and 1044 cm–1, confirming the presence of different oxide moieties on its surface. The bands at 1720 and 1630 cm–1 are, respectively, assigned to the vibrations of C=O stretching[21] and the deformation vibration of OH in absorbed water molecules or the aromatic C=C stretching vibration.[22] The bands at 1383, 1222, 1123, and 1044 cm–1 are assigned to the O–H deformation vibrations of tertiary C–OH,[23] bending absorption of carboxyl group,[24] C–OH stretching,[25] and C–O stretching vibration of the epoxy groups,[26] respectively. The broad band at 3416 cm–1 is indicative of hydrogen bonded OH groups. The chemical structure of NG shows bands similar to those detected for GO but with loss in intensity as seen for the bands at 1630, 1382, and 1032 cm–1, indicating the perseverance of residual OH and C–O groups. Other peaks exist on NG that are not found in GO, such as 1556, 1184, 1120, 786, and 704 cm–1. Simultaneously, the obtained shift in the hydrogen bonded OH group from 3416 cm–1 in GO to 3435 cm–1 in NG raises the probability of associating the latter with the N–H stretching vibrations. The existence of the 1556 cm–1 band correlated to the aromatic C=C skeletal vibrations characterizes the GO reduction.[23−27] It may also be assigned to the C=N stretching together with 1120 and 1184 cm–1 bands used to characterize C–N stretching in benzenoid rings;[28] in accordance with the XPS results. The advent of peaks at 786 and 704 cm–1 is probably correlated to =C–H bending vibrations. The FT-IR spectrum of BG shows strong bands at 1174, 1566, and 550 cm–1 with small shoulders at 1120 and 1048 cm–1. It is interesting that the B–O band (1174 cm–1) and O–B–O band (550 cm–1)[29] appeared in the BG annealed at 900 °C with no probability of forming a B2O3 phase, as XRD analysis cards confirm its absence. The appearance of the broad band in the margin of 1000–1200 cm–1 is caused by combination of B and O atoms to verify the evolution of B–O asymmetric stretching (1120–1048 cm–1) vibrations. The existence of the band at 1566 cm–1 establishes the further facile reduction of GO compared with its counterpart in NG (1556 cm–1). Shift of the band at 3416 cm–1 in GO to 3427 cm–1 in BG, compared to 3435 cm–1 in NG, advocates the presence of residual hydrogen bonded OH groups in the BG sample. In comparison, the NBG spectrum indicates typical bands of NG in the mid range (1628, 1581, and 1380 cm–1), together with shoulders in the range 1250–1032 cm–1, acquiring typical features of the BG spectrum. Additionally, the 3435 cm–1 band is more similar to the NG one, suggesting O–H (N–H) bond formation rather than B–OH ones. B–O bonds are seen also in the NBG spectrum.

Figure 5

(a) FTIR and (b) Raman spectra of GO, NG, BG, and NBG samples.

Raman spectroscopy is another technique to obtain very useful information concerning the predominant features of D, G, and 2D bands in the spectra of N-, B-, and NB-doped graphenes. The latter features are represented by the peaks at 1342–1348, 1590–1596, and 2550–3250 cm–1, respectively (Figure b). The shift in wavenumbers exhibited for the G band (reaching ∼1596 cm–1) following graphene doping with N and B, together with the exhibited loss in intensity compared to that in the nondoped GO (1590 cm–1) is indicative of the replacement of C–C by C–N and C–B bond distances.[30] Increase in the intensity of the D-band when compared to the G-band expresses the extent of deformation exhibited as a result of doping. The ID/IG intensity ratio, which relates to a series of defects, increases slightly for NG (1.37) over BG (1.16) and NBG (1.25) when compared with GO (0.94). Diffusion of the 2D band of doped graphene samples apart from that seen for GO is also a diagnostic for the successful incorporation. No criteria for the B–N bond is formed based on the absence of bands around 1366 cm–1.[31]

Optical, IPCE, and Electrical Conductivity Characteristics

The ultraviolet–visible absorption spectra were collected to investigate the optical properties and the energy band gaps of NG, BG, and NBG samples (Figure a). The absorption spectra of all doped graphenes show absorption edge at 420–460 nm due to π → π* transitions, corresponding to an optical band gap of 2.95–2.69 eV, maximizing the sp2 domains of NG, BG, and NBG. GO indicates a maximum at 245 nm with lower light absorption in the whole margin. In addition, N- and B-doped graphenes exhibit extra shoulders at 540 nm (2.3 eV) and 680 nm (1.8 eV) correlated to the n → π* transitions. They correspond to the electron transition from nonbonding nitrogen and boron states to the π* states. This indeed reflects the heterogeneity of N and B distributions in graphene structures. Interestingly, the NG sample best absorbed light in the UV region and extending into the visible light region to 1100 nm, exceeding in sequence BG and NBG samples. On the basis of the Tauc equation, the measured band gap values were 1.68, 1.9, and 2.04 eV for NG, BG, and NBG, respectively (Figure a), compared with 2.6 eV for GO. Based on the mentioned results, the doped nitrogen at 2.7% maintaining a median particle domain size between BG and NBG samples produced the lowest Eg value. NBG has the highest Eg value compared to NG and BG ones, as it presents the lowest absorption edge in the frequency range of 300–650 nm. This indeed could limit the expected photocatalytic activity of this specific sample as a result of lowering the light harvesting capability.

Figure 6

(a) UV–vis absorption spectra of GO, NG, BG, and NBG, and their corresponding plots of (αhν)2 versus (hν), (b) electrical conductivity of GO, RGO, NG, BG, and NBG, and (c) incident-photon-conversion efficiencies (IPCE%) of NG, BG, and NBG.

Considering that the densities and pressures of all doped graphenes are constant while pellets are manufactured, the electrical conductivity of GO was 8.0 × 10–7 Ω–1cm–1, whereas those of doped graphenes were in the order NG (30.0 × 10–3 Ω–1 cm–1) > BG (23 × 10–3 Ω–1 cm–1) > NBG (20 × 10–3 Ω–1 cm–1), as determined from the σ characteristic curves shown in Figure b. As is known, charge carrier concentration and mobility are the factors affecting the conductivity. The electrical transport studies demonstrate that the predominance of n-type is the most impressive in NG nanosheets whereas in BG p-type action is the most leading. Indeed, doping with nitrogen has substantiated the conductivity via creation of the nonbonding electrons capable of enhancing the electrical properties over those from holes created via boron insertion. Decreasing conductivity of the NBG sample is mostly caused by increase of the boron concentration over nitrogen, as elemental analysis and XPS confirmed, and thus leakage of the latter into the former via the n → p transfer, imparting fast recombination is expected to cause inferior conductivity, that is, electron mobility decrement. A direct relationship between energy gap and electrical conductivity in the present study was observed for the NG, where it was difficult to realize in the case of BG and NBG. In conformity with the conductivity results, IPCE was measured to evaluate how efficiently these materials can convert the incident light into electrical energy at a given wavelength. Figure c shows that NG expresses the highest IPCE% (60%) followed in sequence by BG (32%) and NBG (22%) and thus indicates that NG owns the highest number of carriers created under the full spectrum of light illumination, that is, integration of the area under the curve indicates the highest electrical current density monitored in the visible margin ranges from 450 to 850 nm. The IPCE spectra for all samples have similar shapes with two absorption signatures varied with the type of the doped material; for NG, peaks at 600 and 750 nm are seen, which are red-shifted for rest of the samples. The IPCE curves were in harmony with the corresponding UV–vis spectra, thus verifying the obtained results.

Electrochemical Properties

Given the varied defects and the morphological differences as well as the pore texture variations after functionalization with N, B, and NB, as characterization results confirmed, their beneficial use as credible supercapacitor electrodes are expected to be of wide range. So, the electrochemical characteristics of these materials were evaluated by CVs, charge–discharge, and EISs. Figure a presents the CVs of NG, BG, and NBG electrodes at 100 mV s–1. The NG electrode demonstrates a rapid current response at the initial stages of positive and negative potential scans, therefore displaying quasi-rectangular profiles. This suggests that the capacitive response is triggered by the EDL-feature. The largest loop area within the CV curve of the NG electrode implies the highest capacitance, exceeding in sequence NBG > BG > GO. The GO electrode distorts apparently, accompanied by a broad faradic bump at 0.2 V with a narrowed faradic squeeze within the potential of −0.8 to 0.4 V, reflecting the lower electrons and ion conductivities probably due to lowering the graphitization degree. To further elucidate the electrochemical kinetics at the electrode surface, the CVs of NG at different scan rates were measured using the equation (inset in Figure a). It is shown that the capacitive current increases accordingly with scan rates and the curves can maintain their shape at various scan rates confirming the good rate behavior of a symmetric device. The response currents are basically correlated to the square root of scan rates, showing that the electrochemical charge storage is in part a diffusion controlled process. Thus, the specific capacitance values calculated for all doped graphenes (Table ) indicate higher values at lower scan rates and the best value was for the NG electrode at 10 mV s–1 and equal 374 F·g–1. Maintaining the CV shape at scan rate of 100 mV s–1 without obvious deformation implies good electron and ion mobility benefited by the interconnected mesoporous channels of the conducting NG. This ensures the low resistance of the NG electrode to mass transfer and good charge propagation of ions at the interfaces between the electrolyte and the electrode. The obtained CV shape with slopes at high and low potential was similar to the CV of the ZnFe2O4/NRG composite, which exhibits a specific capacitance of 244 F·g–1 at 0.5 A g–1 with a cycling durability of 83.8% at a scan rate of 100 mV/s after 5000 cycles.[32] Increasing the Csp value for the single element N-doped graphene, over comparable ones recently reported in the literature,[32−37] compared to the other doped graphenes, reflects the efficacy of the generated active sites on the former in enhancing the charge storage capacity. The formation of a uniform porous network structure in NG facilitates the diffusion of Na+ ions where they can be obstructed by BG and NBG, as TEM and N2 sorptiometry results confirmed. Additionally, increasing the electrical conductivity due to the increased charge carrier concentration devoted from the nitrogen doping as well as the defect-like small pores in the basal plane of graphene sheet affected the enhancement of the Csp value.

Figure 7

(a) The CV curves of NG, BG, NBG, and GO (inset) measured in the potential window of −0.8 to 0.65 V, performed in 1.0 M Na2SO4 and at scan rate of 100 mV s–1. Inset in panel a is the corresponding specific capacitances of NG at the scan rates of 10, 20, 50, and 100 mV s–1. (b) The potential vs time curves for NG, BG, NBG, and GO (as inset) performed in 1.0 M Na2SO4 with applied current of 0.5 mA. (c) Electrochemical impedance spectra of NG, BG, NBG, and GO, performed at an open potential in the frequency margin of 0.1–100 kHz with an amplitude equal to 10 mV. Inset is the NG circuit (X2 = 0.45058). (d) Evolution of the relative capacitance as a function of the number of electrochemical cycles at at a constant current density of 0.5 A g–1 in the potential range from −0.78 to 0.6 for 1000 cycles for all samples.

Table 2

Specific Capacitance (Csp) of GO, NG, BG, and NBG Performed in 1.0 M Na2SO4 at Various Scan Rates

	Csp (F·g–1)
vs (mV s–1)	GO	NG	NBG	BG
100	8	122	82	26
50	12	208	120	43
20	15	315	153	51
10	20	374	182	56

The galvanostatic charge/discharge curves of all doped electrodes at a current density of 500 mA g–1 (Figure b) produced nonsymmetric curves and iR drops, exactly as GO shows (inset in Figure b) illustrating the pseudocapacitive nature of the latter electrode, in agreement with its CV results.[37] Based on the discharging behavior in NG, BG, and NBG curves, the specific capacitances determined based on the equation were 425, 400, and 388 F·g–1 at current densities of 0.25, 0.5, and 1.0 A g–1, respectively (Table ). The present values are comparable and even higher than some typical carbon-based supercapacitors cited lately and manufactured via laser reduction of GO films, heteroatom-enriched electrospun carbon nanofiber, and also B-rGO of a specific capacitance value of 200 F·g–1 reported in 6 M KOH at 0.1 A g–1.[33,38−40] The synergistic effects of N,B-codoped graphene are observed in our samples producing higher capacitance values (297 F·g–1) exceeding those in the literature, such as those reported for aligned (239 F·g–1) and nonaligned BCN (167 F·g–1).[33,42] However, notably the capacitance of our NG surpasses those of the latter as well as our synthesized BG and NBG at similar scan rates. Accordingly, the encountered stability of the NG electrode during the capacitance retention cycles can also be correlated to the good contacts between various N–C and N–O bonds, as confirmed from XPS and FTIR results, which are good enough to assist charge transitions via such active sites into the charge collectors. The decreased Csp value of BG in spite of increased SBET compared to NG might be due to greatly increased defects, as well as the particle size increment, which limit the conductivity via slowing down the electrolyte adsorption or transfer processes into the electrode surface.

Table 3

Specific Capacitance (Csp) of GO, NG, BG, and NBG; Performed in 1.0 M Na2SO4 at Various Current Density

	Csp (F·g–1)
current density (A g–1)	GO	NG	NBG	BG
0.25	18	425	297	259
0.5	16	400	237	213
1	13	388	179	188

EISs of different supercapacitors were measured to further elucidate the electrochemical behavior during the charge storage processes. From Figure c, the Nyquist plots of impedances for all devices reveal a straight line in the low frequency range and a semicircle at intermediate and high frequency regions. The intercept onto the Z′ axis at the high frequency end indicates the total series resistance (Rs) including the electrolyte resistance, intrinsic active material resistance, and contact resistance between the active material and collector, and they all show comparable values in the range of 15–18 Ω. The semicircle at intermediate-high frequency arises from the charge transfer resistance (Rct) at the electrode/electrolyte interface, indicating Rct of 7 Ω for NG, apparently lower than those of NBG (30 Ω), BG (76 Ω), and GO (225 Ω). This implies that NG exhibits more rapid adsorption of electrolyte ions onto the electrode surface and on the efficient interface exhibited between N and O species. On the other hand, concerning the Warburg resistance (Zw) of the electrolyte, NG exhibits the shortest projection length, offering a much lower Zws due to the enlarged pore size. The increased slope suggests facile ion penetration and diffusion to the surface of NG through different nitrogen-moieties via the admitted mesopore/micropore interfaces, as indicated from the N2 adsorption data. The above data are consistent with the results obtained with vibrational spectroscopy, electrical conductivity, and CV results. Accordingly, lowering the Rs, Rct, and Zw values of the NG supercapacitor indicate that it is the most preferred electrode material. The encountered higher electrical conductivity of NG together with its unique pore texturing offers higher contact at the electrode/electrolyte interface. The inset circuit of the NG (inset in Figure c) shows the proximity of the circuit with the original data obtained from CVs and EISs and most importantly the existence of double layer capacitance. It shows X2 = 0.45058 μF, which is lower than the other electrodes suggesting lower internal resistance. The calculated time constant, configured for the required time to charge the capacitor, evaluated based on the cutoff frequency fmax, τ = 1/2πfmax (where fmax is the peak frequency) was the lowest for NG (6.22 × 10–5 s) compared with the other electrodes by 8–9 times. The latter observations concerning CVs and impedance measurements indicate the role of the doped N graphene in increasing charge density as well as separation efficiencies.

Figure d shows the stability of the electrodes for 1000 cycles with 100%, 95%, and 91% capacitance retention at current density of 0.5 A g–1 for NG, NBG, and BG, respectively, whereas GO exhibits a retention of 70%. To the best of our knowledge, the capacitance value (388 F·g–1 at 1.0 A g–1) reported here is one of the best data for the SCs reported at different current densities.[31−38] The rate capability and long-term stability of the electrode was retrieved by testing the electrode at higher current densities for longer time. It was seen that when the material was tested at higher current densities, it still kept an excellent capacitance value of 250 F·g–1 at a current density of 3.0 A g–1.

Energy density is a decisive parameter of SCs that restricts their applicability for electric vehicles since it is very important for SCs exhibit high capacitance and extraordinary values of energy density. The Ragone plot (not shown) indicates that NG has much higher energy density values [calculated using the equations Wh/kg and W/kg], exceeding those of BG and NBG. The NG displays 38 Wh/kg at 0.5 A g–1 current density with a good value of power density (260 W/kg) (Table ). A comparison of our results with the recently published values (values are taken for single electrode cell) exhibits that our results show prominently high energy densities without much compromise on their power densities.[33,37,42] Thus, higher values of capacitance and energy density highlight the advantages of the specified interlayer spacing and unique composition of NG, which bring in reality a deep level of EDL action and provide fast transfer of electron and electrolyte ions. The maximum value for E of the NG-based supercapacitor is about 1.2-fold that of NBG, 1.5-fold that of BG and 15.25-fold that of GO. It seems that the remarkable performance of the NG supercapacitance is a result of different existing N-types in graphene that not only improve the wettability of NG in the electrolyte but also diminish the intrinsic resistance of graphene in favor of increasing its mobility.[43] Specifically, pyridinic and pyrrolic N active functional groups participate in creating more active deposition sites of larger binding energies accessible for accommodating a larger number of Na+ ions to play a dominant role in the capacitance enhancement. Also, these N-dopants can impart high charge concentration due to their lone pair of electrons and to the exhibited band gap reduction they cause for carbon. As XPS confirmed, NG exhibited the presence of N–O bonds that indeed boost further the pseudocapacitance performance. Generally, the NG acquires significantly improved capacitance compared with GO and also exhibits good rate and stability performance. Consequently, controlling the appropriate nitrogen configuration, as well as concentration, can greatly promote the capacitance of N-graphenes. On the other hand, the decreasing capacitance of the NBG electrode disfavors the importance of N–B bonds, as well as the BCO ones deconvoluted from XPS B 1s spectra.

Table 4

Best Capacitive Parameters of As Synthesized Electrodes in 1.0 M Na2SO4

samples	Ed (Wh/kg)	Pd (W/kg)	Eeff (%)	ηa (%)
GO	2.5	20	70	105
NG	38	260	100	225
BG	26	184	94	150
NBG	31	216	96	266

The Coulombic efficiency (η) defined as the ratio of discharging time and charging time was calculated by the equation η = (tD/tC) × 100.

Photocatalytic Activities

The MB degradation over doped graphenes performed under visible light irradiation is shown in Figure a. Under dark conditions, it is shown that NG absorbs more MB with a value of 42%, exceeding those of BG (23%) and NBG (19%). Consist with the absorption behavior, NG exhibits the highest MB degradation at 93%, surpassing those of BG (38.5%) and NBG (25%), performed in a 150 min reaction time and at a catalyst loading of 0.5 g L–1. The results in the inset fit to the pseudo-first-order model and give k values in the order; NG 0.0127 min–1 > BG 0.0015 min–1 > NBG 0.0006 min–1. The UV–vis spectra showing the MB concentration dropping in 150 min, was also presented (inset Figure a), to further confirm the first order behavior and the degradation consequences. The photodegradation performance of the NG gives a rate constant as high as 8.4 times that of BG and 21 times that of NBG, in the absence of any oxidants, and maximizes the role of nitrogen-doped types and its amount in achieving such high photoactivity behavior. Indeed, this superior photoactivity is directly correlated to the enhanced adsorption; reflecting the superior contact of MB with NG, as well as its higher conductivity, which facilitates the rapid charge carrier migration between the 2D graphene planes with its nitrogen substituted active sites.[23] In order to be sure that no further adsorption preceded the degradation process, the photocatalyst and the target pollutant were left for 2 h to ensure adsorption–desorption equilibrium was established in the dark before the photocatalytic reaction, and we found that the achieved adsorption equilibrium was still at the value of 42%.

Figure 8

(a) Photocatalytic degradation of MB dye using NG, BG, and NBG catalysts under visible light irradiation (reaction conditions: lamp power = 160 W, filter λ = 420 nm, catalyst load 1.0 g/L, phenol concn 20 ppm). Insets are the kinetic fit for MB degradation via the as-synthesized photocatalysts and UV–vis spectra of MB during degradation using NG under visible illumination. (b) Photocatalytic degradation of phenol using NG photocatalyst under visible light irradiation and under same reaction conditions except that phenol concn was 5 ppm. Insets are the kinetic fit for phenol degradation and UV–vis spectra of phenol during degradation using NG under visible illumination. (c) Effect of reactive scavengers on the degradation activity of the NG photocatalyst toward MB under visible light irradiation and under the mentioned reaction conditions. (d) Mott–Schottky plots of NG, BG, and NBG electrodes, where C is the capacitance and V is the electrode potential at 1 kHz and electrolyte 1.0 M Na2SO4.

The optical absorption of the NG extends to the red region, exceeding BG and NBG samples, and exhibits the lowest Eg value facilitating the charge transfer process, as has also been assessed via CV and EIS results. Increasing the NG catalyst amount into 1.0 g/L produces a further increase in the MB dark adsorption to 50% (Figure S2), thus maximizing the photocatalytic degradation of the MB completing in only 15 min, enhancing the rate constant 16-fold (0.197 min–1) over that measured at 0.5 g/L (0.0127 min–1). This is attributed to increasing the number of the active sites and the consequences thereof concerning the increase in the amount of photons absorbed as a result of intensifying the light harvesting capabilities. This increase in photocatalytic activity is also a consequence of prolonging the charge lifetime, leading to a low recombination of electrons and holes. This result was in harmony with the incident photon to current efficiency that elucidates that NG response was 2 times higher than that of BG and ∼3 times higher than that of NBG. This supreme photocurrent recorded for the NG (60%) is indicative of increased migration rate for the photogenerated carriers and of higher separation efficiency, facilitating the MB photodegradation. It seems also that the encountered BG defects signified via TEM images restrict the charge transfer, which affects by its turn the photocatalytic efficiency. Interestingly, this excellent visible light driven photocatalytic activity of the NG exceeded some of graphene based semiconductors and multilayered graphene quantum dots, as well as the Au/pg-C3N4/GR composite.[42−44]

To eliminate the effect of dye self-photosensitization, a colorless phenol was adopted to evaluate the sensitivity and the efficacy of the NG toward its photosensitivity. Figure b shows magnificent activity for NG toward phenol complete degradation with a rate constant of 0.04 min–1 performed in only 60 min under visible light illumination. This indeed nullifies but does not exclude the electron transfer expected to take place from MB to NG upon illumination. It seems that the post-treatment of GO specifically with nitrogen causes reconstruction of the defects formed via severe oxidation, by repairing such vacant-type defects.[45]

The left curve slope at nearly time 0 is almost same as the right one for both MB and phenol during their degradation in the presence of the NG photocatalyst (Figure a,b). In contrast, BG and NBG show a normal behavior. In order to obviate any doubt concerning the degradation of MB and phenol on NG surfaces, TOC was measured for both of the pollutants. After ca. 150 min of visible light irradiation, the MB aqueous (20 ppm) solution in NG that turned completely colorless was accompanied by approximately 75% TOC removal (Figure S3). This indicates that some colorless intermediates were still dissolved in the bulk solution. The increase in TOC removal to 95% was clearly shown after 250 min showing that the phenothiazine structure and the methyl groups of the MB cations were removed and converted ultimately to mineralized inorganic products. The TOC for phenol (5 ppm) reached 4.0% TOC removal in 150 min (Figure S3). The stability test of the NG over five photocatalytic degradation cycles for MB was evaluated (Figure S4). It is established that the NG photocatalyst can work at least five cycles without noticeable loss in the catalytic activity retaining 85% of the catalyst activity during reuse for 750 min, demonstrating long-term durability of this photocatalyst.

To determine the reactive species involved on NG surfaces for MB degradation, trapping experiments using p-benzoquinone (•O2–scavenger), triethanol amine (TEOA; hole scavenger), isopropanol (IPA; •OH), and CCl4 (electron scavenger) were performed (Figure c). It is apparent that the abundant reactive oxygen species •OH and electrons are the most active initiators for the photocatalytic degradation of the MB contaminant. Meanwhile, holes also shared in the photoactivity but at much lower amounts than •OH and electrons. It is further confirmed through the trapping experiments that benzoquinone induces very small changes suggesting either a marginal effect or a decreased lifetime of •O2– species during the photocatalytic performance of the NG nanoparticles. Based on the above results, it was concluded that NG was the most photoactive component; thus, a tentative mechanism for its photocatalytic action is proposed and shown in Scheme . As a decrease in the oxygen content on the NG and the consequences of the formation of sp3-hybridized carbon atoms, an upward VB shift with no change of the CB is likely.[46] Thus, the exhibited small band gap can allow NG to work as a photosensitizer capable of expanding the visible light absorption range. Due to the declared semiconductor nature of the NG, it excites more electrons from the VB, which hardly can react with O2– as the reactive study confirms but indeed can be trapped by carbon atoms in GO/G due to their p-type character.[46] This indeed leads to a low recombination rate and lengthening of the lifetime of the charge carriers, as has been confirmed before using IPCE, conductivity, and EIS results. The photogenerated holes on the other hand can react with OH–/H2O to form •OH ready for the MB degradation. Simultaneously, if •O2– is formed and according to its decreased lifetime, as evidenced from the scavengers study, it can react with 2H+ (from N–H or O–H groups) to generate •OH as well. It seems that the appropriate defects appearing on NG surfaces, as revealed from Raman and XRD results, will decrease the electron transfer and thus facilitate the separation between e––h+ pairs to enhance the photocatalytic efficiency.

Scheme 1

The electrochemical Mott–Schottky experiment was carried out to clarify the relative positions of CB and VB edges (Figure d) for all doped graphenes via using the equation to have a complete picture and to support the proposed mechanism (Scheme ). The plots reveal that NG is a standard n-type semiconductor, whereas BG is a p-type owing, respectively, to the evident positive and negative slopes. In the same context, the NBG plot revealed the combination of n- and p-types, reflecting that nitrogen doping resulted in the donation of electrons that compensated the electron deficiency initiated by the boron doping. Of particular interest, GO shows a negative slope corresponding to the p-type conductivity manifesting the presence of high oxygen vacancy defects that have been compensated following the N-doping. Importantly, the flat band potentials (Efb) of NG and BG were determined to be 0.31 and 0.87 V versus MSE whereas those of NBG (n- and p-types) almost coincided around the potential of 0.32 V. Combining the band gaps determined from UV–vis curves with the determined Efb, the VB levels of NG and BG were calculated to be 1.27 and 1.04 V, respectively (Scheme ). The NG electrode showed a positive shift of the flat-band potential as compared to the BG electrode, and the slope of the NG line showed a lower value, compared with BG and NBG lines, testifying clearly the high donor density of NG, as confirmed previously using IPCE and electrical conductivity data, than its comparable one in NBG and indeed exceeding the density of acceptors devoted from the BG. Based on the previous information, a band diagram is constructed to comprehend that NG has a significant overpotential toward the photocatalytic water oxidation over NBG and BG ones, facilitating the MB oxidation consequences.

Conclusion

The formation of various nitrogenous moieties in NG, synthesized using urea–GO mixture and annealed at 900 °C in Ar atmosphere, with a lower band gap and a wide photoresponse margin together with a separation of the photogenerated charge carriers, has produced outstanding properties as a photocatalyst for the degradation of various organic moieties (dye and phenol) under visible light illumination compared with BG and NBG samples. It has been shown via Mott–Schottky measurements that the valence band maximum rises obviously in the NG, and thus elaborated with both optical and electrical measurements, the oxidation mechanism involves a limited role played by holes in the valence band, as well as an appreciable role played by the electrons. The NG has shown marvelous morphological, optical, electrical, and texturing properties, which all share in boosting the supercapacitance performance via the EDL mechanism. It is believed that N–O as well as pyridinic N and pyrrolic N function as active sites for improving the supercapacitive performance. This study proves the importance of dopant atoms in graphene, rendering doping as a cost-effective and ecofriendly synthetic strategy to heighten the supercapacitance, with superior cycling stability caused by different characteristics such as ultrafine size, controllable chemical compositions composed of conductive nitrogen–oxygen–carbons, and the hierarchical porous structures, unlike B- and BN-graphene. The improved electrical conductivity due to the increased charge carrier concentration derived from nitrogen doping and defect-like small pores in the basal plane of the graphene sheet are considered as paramount factors for the improvement in the specific capacitance and photocatalysis.

Materials and Methods

Synthesis of NG, BG, and NBG Catalysts

Graphene oxide (GO) nanosheets were prepared by a modified Hummers method.[47] The obtained GO (0.2 g) was systematically mixed with a solution of boric acid (0.03 g) and urea (0.36 g) in deionized water (50 mL). The mixture was then warmed to 80 °C to form a thick slippery liquid, which was then dried at the same temperature for 36 h in vacuum. The dried mixture was then annealed at 900 °C for 10 h in Ar atmosphere to give nitrogen–boron-codoped graphene (NBG). The monodoped graphenes were similarly synthesized under typical conditions and concentrations of urea and boric acid to form nitrogen-doped graphene (NG) and boron-doped graphene (BG), respectively.

Characterization

The phase structure of the products was measured by powder X-ray diffraction (XRD) using a RigaKu D/max-RB diffractometer with Ni-filtered graphite-monochromatized Cu Kα radiation (λ = 1.54056 nm) at a scan step of 0.02° from 5° to 70°. Transmission electron microscopy (TEM) studies were performed using a transmission electron microscope (TEM, FEI) with an accelerating voltage of 200 kV. The specific surface areas and pore structures were estimated by the Brunauer–Emmet–Teller (BET) method based on nitrogen absorption–desorption (Micromeritics ASAP 2020) characteristics. X-ray photoelectron spectrometry (XPS, X-ray monochromatization, Thermon Scientific) was carried out with Al Kα as the excitation source; the collected binding energies in the XPS analysis were calibrated against the C 1s peak at 285.0 eV to analyze the surface composition. FTIR analysis was performed using a Nicolet 6700 spectrometer in the range of 4000–400 cm–1. Raman measurements were performed with a Raman laser spectrometer (DXR) with an excitation line of 632.8 nm. The optical characterizations of the samples were performed using UV–vis spectroscopy (Cary 5000). The dc conductivity, σdc, of as-synthesized materials determined at 25 °C was calculated using the equation σdc = (l/As)(1/Rdc) where Rdc is the sample resistance, l is the length of the sample, and A is the cross-sectional area, collected employing a programmable automatic LCR bridge (HIOKI 3532-50) and an electrometer (model 6517, Keithley), voltameter (Keithley, 2182), and 5 kV dc power supply.

Photocatalytic Reaction

The photocatalytic degradation of methylene blue (MB) dye and phenol was performed under visible light irradiation using a 160 W lamp with a wavelength of 420 nm. Doped graphenes (100 mg) and 100 mL of the pollutant solution (5–20 mg/L) were added to a reactor. This solution was stirred at 25 °C in the dark for 60 min until equilibration. Then, 3 mL of the reaction mixture was collected and immediately centrifuged to detect the adsorbed amount of either MB or phenol on the catalyst surface. Clear sample absorbance was tested using UV–vis jenway-6800 spectrophotometer at 664 and 267 nm, respectively. The removal efficiency was calculated using the following equation: removal (%) = (C0 – Ce)/C0 × 100), where C0 and C were initial and equilibrium concentrations, respectively, of MB or phenol (mg/L). For exploring the reactive species that might be produced in the photocatalytic reaction,[47] we used different scavengers including isopropanol (a quencher of •OH), p-benzoquinone (a quencher of •O2–), Na2EDTA (a quencher of h+), and carbon tetrachloride (a quencher of e–) at a concentration of 1.0 mM.

Electrochemical Measurement

The electrochemical measurements of as-synthesized samples were thoroughly examined at room temperature in three- and two electrode cell systems named PGSTAT204 associated with Nova 1.11 software for data calculation. The working electrodes were made-up by mixing the synthesized catalysts (85 wt %) with 10 wt % acetylene black and 5 wt % PVDF binder. The FTO conductive glass sheets of dimensions 1 cm2 and resistance 15 Ω cm–2 were well cleaned and dried before depositing the doped graphenes. The doped graphene electrodes were heated to 200 °C for 3 h then left to cool down to room temperature. The latter deposited working electrodes were transferred to the electrochemical cell containing an electrolyte (Na2SO4 of 1.0 M concn), a platinum electrode as the counter electrode, and a saturated Hg/Hg2SO4 electrode as a reference electrode. The CVs were performed between the potential of −2 and +1 V and at scan rates 5.0, 10.0, 20.0, and 50.0 mV/s. Electrochemical impedance spectroscopy (EIS) measurements were conducted using the same mentioned apparatus with conditions of current ranging from 10 μA to 100 mA, a frequency margin of 0.1 Hz to 100 kHz, and at a constant potential of 10 mV. The justification of the impedance spectra was determined based on the Kramers–Kronig transformation. Specific capacitance was also measured by galvanostatic charge–discharge via using chronopotentiometry using Digi-Ivy 2116 B-USA, equipped with DY2100B software for data calculation. The prepared working electrodes were measured at step current of 0.05, 0.2, 0.5, 1, and 2 mA.