III. RESULTS AND DISCUSSION

Urea is known to release NH₃ upon heating, and this reaction helps to incorporate nitrogen in the carbon matrix of graphene.^{23−27,Reference Sun, Wang, Tian, Tan, Xie, Shi, Li and Fu30} In the present study, hydrothermal conditions with high autogenous pressure and temperature or MW irradiation with localized superheating enable the rapid dehydration of glucose in the presence of NH₃ released by the decomposition of urea, to form N-CQDs with high nitrogen content. ^{Reference Walton31} Generally, C–N bond formation occurs in the nucleation step, followed by further carbonization and growth to generate N-CQDs. ^{Reference Ma, Ming, Huang, Liu and Kang19,Reference Tang, Ji, Cao, Lin, Jiang, Li, Teng, Luk, Zeng, Hao and Lau32} Moreover, ammonia acts as both the N-dopant as well as a surface passivator without the addition of any external capping agent. ^{Reference Zhai, Zhang, Liu, Bai, Li, Dai and Liu20} Interestingly, in the absence of urea, carbonization of glucose does not occur under the same reaction conditions, which is also confirmed by controlled PL and TEM studies. Previously, CQDs are formed at 150 °C by the decomposition of carbohydrate after much longer reaction times. ^{Reference De and Karak33} The rapid formation of N-CQDs in the present synthesis may be due to the presence of ammonium and hydroxyl ions from aqueous ammonia, which are known to catalyze the carbonization of carbohydrates. ^{Reference Wang, Qu, Xu, Ren and Qu28,Reference Yu, Cui, Li, Li, Yu, Antonietti and Colfen34}

XPS measurements were carried out to establish the elemental composition of N-CQDs (Fig. 1). The survey scan spectrum indicates a pronounced N 1s peak at ∼399.5 eV, a predominant C 1s peak at ∼284 eV, and an O 1s peak at 532 eV (Fig. 1), which indeed confirms the successful formation of N-CQDs by both hydrothermal reaction and MW irradiation. The N/C atomic ratio was calculated to be ∼18% for hydrothermally produced N-CQDs, whereas for MW synthesis, it is ∼22%, which is higher than that of previously reported N-CQDs. ^{Reference Zhang, Ma, Zhuang, Zhang, Chen, Hong, Yan, K.Yu and Huang17,Reference Ma, Ming, Huang, Liu and Kang19,Reference Li, Zhang, Dai and Li22} The high resolution N 1s spectrum of N-CQDs synthesized by the hydrothermal method reveals the presence of both pyrrolic (399.6 eV) and N–H like (400.3 eV) N atoms [Fig. 1(b)], whereas the N-CQDs formed under MW irradiation show only pyrrolic N at 399.4 eV [Fig. 1(c)]. In addition, the high resolution C 1s spectrum of N-CQDs further confirms the presence of C–C (284.5 eV), C–O (288.3 eV), and most importantly C–N (285.8 eV) bonds (Fig. SI, supporting information). Fourier transform infrared (FTIR) spectra demonstrate the presence stretching vibrations of N–H (from 3500 to 3150 cm⁻¹), C=C (1604 cm⁻¹), and aromatic C–N heterocycles (from 1500 to 1150 cm⁻¹) (Fig. SII, supporting information). ^{Reference Zhang, Ma, Zhuang, Zhang, Chen, Hong, Yan, K.Yu and Huang17} The predominant presence of C compared to N was further confirmed by elemental analysis which shows the N content to be comparable with XPS results (Table SI, supporting information).

FIG. 1.

(a) X-ray photoelectron survey spectra of N-CQDs prepared by hydrothermal and microwave methods. N 1s signal of N-CQDs prepared by (b) hydrothermal and (c) MW methods.

We have characterized the N-CQDs by TEM and AFM (Fig. 2). Figure 2(a) and its inset show the TEM images of N-CQDs synthesized by hydrothermal reaction and MW irradiation, respectively. In Fig. 2(b), high-magnification TEM images show a relatively narrow size distribution (3–5 nm) of hydrothermally prepared N-CQDs with an average diameter of ∼4 nm. N-CQDs synthesized by MW irradiation have the average diameter of ∼4.5 nm. Figures 2(c) and 2(d) show typical AFM images of N-CQDs synthesized by hydrothermal and MW irradiation, respectively. The topographic height distributions are in the range of 2–3 and 1–2 nm for N-CQDs synthesized by the hydrothermal and MW treatment, respectively. The respective average heights are ∼2.5 and ∼1.5 nm, which suggest that most of them consist of 1–5 graphene layers.

FIG. 2.

(a, b) TEM images of N-CQDs prepared by the hydrothermal method with size distribution. AFM images of N-CQDs prepared by (c) hydrothermal and (d) MW methods. Insets show height profiles.

We have recorded absorption and PL spectra of the N-doped CQDs prepared by us (Fig. 3). Typical UV-vis spectra of the N-CQDs synthesized by the two methods show a broad absorption band in the 250- to 350-nm region with a tail extending to the visible region, characteristic of a polycyclic aromatic π system [Figs. 3(a) and 3(b)]. ^{Reference Xie, Huang and Zheng35} PL spectra show an excitation wave length dependent intense PL emission [Figs. 3(a) and 3(b)]. The PL emission band shifts from 428 to 456 nm [Fig. 3(a)] and from 432 to 457 nm [Fig. 3(b)] for N-CQDs synthesized by the hydrothermal and MW methods, respectively, on changing the excitation wave lengths in the range of 345–365 nm. Bright blue light is emitted when N-CQDs are irradiated under UV light [photograph in Fig. 3(a)]. N-CQDs also show strong-excitation dependent down-converted PL emission [see Fig. SIII(b) in supporting information]. The exact origin of the intense PL emission and the mechanism behind its excitation wave length-dependent emission is not understood properly, ^{Reference Li, Kang, Liu and Lee3,Reference Baker and Baker4} although the luminescence is suggested to arise from a π–π* transition, due to emissive traps or free zigzag sites and/or edge states. ^{Reference Li, Kang, Liu and Lee3,Reference Baker and Baker4,Reference Cao, Meziani, Sahu and Sun36} We observe a slight blue-shift in the PL band position of N-CQDs compared to that of undoped CQDs. ^{Reference Li, Zhao, Cheng, Hu, Shi, Dai and Qu37} Similar blue-shifts in the PL emission band of N-CQDs have been reported by Li et al. ^{Reference Li, Zhao, Cheng, Hu, Shi, Dai and Qu37} and Zhang et al. ^{Reference Zhang, Ma, Zhuang, Zhang, Chen, Hong, Yan, K.Yu and Huang17} We have also measured the quantum yield of N-CQDs prepared by both the hydrothermal and MW methods using coumarin 102 as the reference (Table SII, supporting information). The quantum yields of N-CQDs prepared by hydrothermal and MW methods are 0.7 and 1% respectively. Absence of any external surface passivating agent in the as-prepared N-CQDs may be the reason of the low quantum yield, though it is comparable with the values reported in the literature. ^{Reference Qiao, Wang, Gao, Li, Dai, Liu and Huo38−Reference Liu, Ye and Mao40} N-CQDs prepared by the hydrothermal reaction and MW irradiation show PL decay (measured at 385-nm excitation) with average lifetimes of 2.6 and 2.5 ns, respectively (Table SIII, supporting information).

FIG. 3.

UV-Vis and PL spectra (recorded at different excitation wavelengths) of N-CQDs prepared by (a) hydrothermal and (b) microwave methods. Inset in (a) shows a photographs of N-CQDs illuminated under UV light (b) PL spectra of undoped CQDs.

Graphene is reported in the literature to be a good PL quencher for fluorescent dyes, conjugated polymers, and semiconductor dots. ^{Reference Chen, Berciaud, Nuckolls, Heinz and Brus41−Reference Matte, Subrahmanyam, Rao, George and Rao44} The origin of quenching is considered as resonance energy transfer, electron transfer, or dipole–dipole coupling in between graphene and the fluorescent species. ^{Reference Chen, Berciaud, Nuckolls, Heinz and Brus41−Reference Matte, Subrahmanyam, Rao, George and Rao44} Recently, fluorescence resonance energy transfer from amine-functionalized CQDs to reduced graphene oxide has been studied to sense metal ions. ^{Reference Wei, Xu, Ren, Xu and Qu45} We have observed quenching of PL emission of both N-CQDs and pristine CQDs by graphene (Fig. 4). Small size CQDs can easily adsorb on the basal plane of graphene through noncovalent π–π interaction. There can be photoinduced energy transfer from N-doped or undoped CQDs to graphene through the mutual overlap of their π cloud. Besides, the defect sites, emissive traps, dangling bonds, and functional moieties present in the CQDs and graphene surface may be expected to play a role as the PL quencher of CQDs. A linear Stern–Volmer plot has been observed in the case of pristine CQDs upon interaction with graphene unlike the case of N-CQDs which shows positive deviation from linearity [Fig. 4(b)]. The linear Stern–Volmer plot in the case of CQDs indicates dynamic quenching, whereas the nonlinear curve observed in the case of N-CQDs suggests that the quenching process follows both static and dynamic energy transfer. ^{Reference Airinei, Tigoianu, Rusu and Dorohoi46,Reference Lakowicz47} The quenching rate constants (k _q) for both CQDs and N-CQDs are similar (Table SIV, supporting information).

FIG. 4.

(a) PL emission spectra (365-nm excitation) and (b) Stern–Volmer plots for quenching of luminescence of N-CQDs and undoped CQDs in ethanol in the presence of exfoliated graphene.

We have examined the quenching of PL of the CQDs in the presence of N,N-dimethyl aniline, which is known to be an excellent fluorescence quencher (Fig. SIV, supporting information). The k _q values obtained from the linear Stern–Volmer plots are in the range of 10⁹ M⁻¹ s⁻¹, which agrees well with the literature reported values. ^{Reference Wang, Cao, Lu, Meziani, Heting, Gang, Bing, Harruff, Kermarrec and Sun48} The exact mechanism of the interaction of all these quenchers with undoped and N-doped CQDs is not fully understood and needs further investigation.

Tetrathiafulvalene (TTF) and tetracyanoethylene (TCNE) are known for their charge transfer interaction with aromatics including graphene. ^{Reference Voggu, Das, Rout and Rao49,Reference Rao and Voggu50} In the case of undoped and N-doped CQDs, electron-donating TTF shows significant quenching of luminescence [Fig. 5(a)]. We observe a nonlinear Stern–Volmer plot [Fig. 5(b)] with a quenching constant of 6.1 × 10¹² and 2.6 × 10¹² M⁻¹ s⁻¹ for N-doped and undoped CQDs, respectively (Table SIV, supporting information). The high values of quenching constant indicate a possible binding interaction between the fluorophore and the quencher. ^{Reference Lakowicz47} Unlike TTF, electron-accepting TCNE interacts differently with N-doped and undoped CQDs [Figs. 5(c) and 5(d)]. Undoped CQDs show a gradual decrease in luminescence on increasing the TCNE concentration, whereas N-doped CQDs show enhancement in the PL intensity up to a moderate concentration of TCNE [Fig. 5(c)]. A decrease in the full-width-half-maximum (FWHM) of the PL band occurs on addition of TCNE to N-CQDs. To understand this behavior, we have deconvoluted the broad PL emission spectrum into multiple Gaussian functions, which indicate the presence of more than one emissive site (Fig. SV, supporting information). ^{Reference Yu, Wen, Toh and Tang51} The initial addition of TCNE (1 mM) may quench surface-related emission (bands 2 and 3 in Fig. SV) of N-CQDs resulting in a decrease in the FWHM of the PL band. The quenching of surface-related emission increases the core-related emission intensity [Fig. 5(c)]. We speculate that the electron acceptor TCNE occupies some of the surface-emissive sites containing N-atoms in the CQD lattice. At high concentrations of TCNE, sharp quenching in PL emission along with the red shift in PL suggests agglomeration of N-CQDs or chemical reaction. ^{Reference Dey, Das, Voggu, Nag, Sarma and Rao52} We note that specific interaction of TCNE with N-CQDs is to be expected since these CQDs can act as electron donors while TCNE is an electron acceptor.

FIG. 5.

PL emission spectra (365-nm excitation) of N-CQDs in ethanol in the presence of (a) TTF and (c) TCNE and of (d) undoped CQDs in the presence of TCNE. (b) Stern–Volmer plots for the quenching of luminescence in the presence of TTF in ethanol.

Defects in ZnO give rise to a characteristic broad emission in the green–yellow region. A combination of blue-light-emitting materials with ZnO or graphene oxide (GO) to generate white light is reported in the literature. ^{Reference Kumar, Panchakarla, Bhat, Maitra, Subrahmanyam and Rao53−Reference Son, Kwon, Park, Seo, Yi, Angadi, Lee and Choi55} There are reports where CQDs or N-CQDs have been used as color converters to emit white light from blue LEDs. ^{Reference Walton31,Reference Kwon, Do, Lee, Hwang, Kim and Rhee56} We find that admixing N-CQDs with ZnO in proper proportions gives rise to white light emission (Fig. 6). In Fig. 6(a), we show typical PL spectra of ZnO with different amounts of N-CQDs. ZnO shows a broad defect-related emission in the visible region centered around 550 nm. As the amount of N-CQDs increases, the PL emission spectrum covers the entire visible region. The PL spectra corresponding to 27%, 30%, and 33% of N-CQDs mixed with ZnO (v/v) have CIE coordinates of (0.32, 0.39), (0.30, 0.36), and (0.29, 0.34) respectively, which are in the white region of the CIE (Commission Internationale de L’Eclairaged) 1931 diagram [Fig. 6(b), light] (excitation 325 nm). In Fig. 6(c), we show the proper admixture of yellow-emitting filled balls marked by a circle. ⁵⁷ In the inset of Fig. 6(b), we have presented the photograph of emitted white graphene oxide (GO) with blue-emitting N-CQDs and also shows white light emission [inset of Fig. 6(c)] with CIE coordinates of (0.29, 0.35) and (0.27, 0.33) [Fig. 6(d)].

FIG. 6.

PL spectra of (a) ZnO + N-CQDs mixture and (c) graphene oxide + N-CQDs mixture in water (excitation 325 and 350 nm, respectively). CIE 1931 color diagrams for the admixture of N-CQDs with (b) ZnO and (d) GO. Filled dots indicate the compositions close to white light emission. Inset shows the photograph of the white light emitted.

We have studied the catalytic activity of the CQDs for the photodegradation of methylene blue (MB) under visible light irradiation (Fig. 7). Figure 7(a) shows the time-dependent changes in the absorbance of MB under visible light irradiation. We observe that in the presence of N-CQDs almost complete degradation (∼88%) of MB is achieved upon irradiation for 30 min [Fig. 7(b)], whereas in the presence of pristine CQD, MB degrades only up to 32% [Fig. 7(b)]. MB undergoes negligible self-degradation (24%) under visible light (control experiment). The rate constants (k) of degradation of MB in the presence of N-CQDs and undoped CQDs were calculated using the Langmuir–Hinshelwood rate equation (supporting information). The values of k for N-doped and undoped CQDs are 0.048 and 0.009 min⁻¹ respectively. Clearly, N-CQDs are more effective in the photodegradation of MB than undoped CQDs. Such superior performance in the photodegradation of methyl orange has been noted in the case of N-CQDs. ^{Reference Ma, Ming, Huang, Liu and Kang19} N-doping can generate local disordered structure and can affect the local electronic structure of CQDs, which favors the formation of transient intermediates by enhancing the binding of ions present in solution and results in the formation of active OH·, which is responsible for photodegredation of the dye. ^{Reference Li, Zhao, Cheng, Hu, Shi, Dai and Qu37,58−61} We should note that N-CQDs would be n-type semiconductors unlike undoped CQDs.

FIG. 7.

(a) Absorption spectra of methylene blue on interaction with N-CQDs at different irradiation times. (b) Photocatalytic degradation of MB (normalized concentration) with the irradiation time in the presence of undoped and N-doped CQDs.