3.1 Characterization of CN, CQD/CN and 1-BCQD/CN
The surface structure and morphology of CN, CQD/CN and 1-BCQD/CN were analyzed using FESEM and TEM. The successful fabrication of an n/n-type g-C3N4 homojunction (CN) was reported in our previous work (Phang et al., 2020a). As shown in Fig. 2(a), the FESEM image of CN features two-dimensional (2D) nanosheets with wide lateral dimensions. The formation of macro-scale g-C3N4 was due to the stacking of 2D polymer molecules while the lamellar structure of g-C3N4 was eventually produced as the planar structure of the molecules undergone distortion. Such distortion was reported to be a result of the enlarged space between g-C3N4 polymer molecules developed during the release of ammonia gas during the calcination step (Song et al., 2020). In addition, the platelet-like structures were also observed to be highly porous with large spatial networks. Figure 2(b) and 2(c) illustrate the FESEM images of CQD/CN and 1-BCQD/CN. From the microscopy images, it is evident that the 2D nanosheet framework of g-C3N4 was generally well-retained. This indicated that the introduction of CQDs and BCQDs had no substantial effects on the surface morphology of g-C3N4, which was consistent with previously reported studies (Liu et al., 2020; Seng et al., 2020).
EDX analysis was also conducted to investigate the elemental composition of each nanocomposite sample. Figure 2(d) presents the EDX spectra of CN, CQD/CN and 1-BCQD/CN, respectively. It can be observed that both CN and CQD/CN contained C, N and O elements while 1-BCQD/CN showed the presence of an additional B peak, which validated the successful doping of B onto pristine CQDs. The presence of C in the EDX spectra could be due to the amorphous carbon of g-C3N4 nanosheets and co-catalyst CQDs. Meanwhile, the existence of N in the spectra could be ascribed to the construction of g-C3N4 using heptazine (C6H3N7) as the fundamental building block of the graphitic carbon framework. Finally, the minor composition of O could be due to the formation of carboxyl groups (─COOH) during the synthesis of BCQDs as well as possible adsorption of water molecules on the photocatalyst surface.
Figure 3(a)-(c) present the TEM images of CN, CQD/CN and 1-BCQD/CN. Similar to the FESEM image, the TEM image of CN depicted in Fig. 3(a) features a highly porous 2D planar sheet with pores of estimated diameters ranging from 22–39 nm. Upon the incorporation of CQD and BCQD, no distinguishable changes were observed with regards to the surface morphology and structure of g-C3N4. The TEM images of CQD/CN and 1-BCQD/CN are depicted in Fig. 3(b) and (c) respectively. Monodispersed quasi-spherical dark spots with an average diameter range of 6.076 and 6.024 nm (see Fig. 3(d)) were observed on the CN sheets for CQD and 1-BCQD respectively. This indicated the successful formation of CQDs and BCQDs on the surface CN. The co-existence of BCQDs and CN with such close contact between them reaffired the successful preparation of the heterostructure nanocomposites. This unique 0D/2D layout is perceived to be beneficial for the effective transfer of photogenerated charge carriers, which will in turn lead to a reduction in electron-hole recombination rate.
The crystalline framework of CN, CQD/CN and 1-BCQD/CN were analyzed with XRD, a non-destructive test method for the identification of plane spacing and crystal structure based on monochromatic beam of X-rays. Figure 4(a) presents the XRD patterns of CN, CQD/CN and 1-BCQD/CN. Two distinctive diffraction peaks were displayed by all the as-prepared photocatalytic samples, which validated the construction of stacked graphitic layers of g-C3N4 (JCPDS, No. 87-1526) (Ran et al., 2019; Orooji et al., 2020). The peak observed at a lower diffraction angle of 13.4° (d = 0.67 nm) corresponded to the (100) plane associated with the in-plane structural packing motif of heptazine units (Tyborski et al., 2013). On the other hand, a more-apparent peak located at approximately 27.4° (d = 0.325 nm) could be indexed to the (002) plane, corresponding to the typical interplanar stacking layers of conjugated aromatic compounds (Elshafie et al., 2020). It should be noted that all samples (CN, CQD/CN and 1-BCQD/CN) exhibited identical XRD trends. This indicated that the hybridization of CN with CQD or BCQD did not alter its fundamental crystalline layout, which was consistent with the microscopy images in Figs. 2 and 3. Interestingly, the characteristic peaks associated with CQD or BCQD were not observed in the XRD profiles. This may ascribe to the low quantity of quantum dots employed in the nanocomposite samples. Another possible reason for this observation could be the overshadowing of amorphous CQD by highly crystalline g-C3N4 polymers. Similar findings were also reported in several published work, where the hybridization of CQDs with different semiconducting materials (e.g. CQD/Fe2O3 and NCQD/TiO2) were studied (Zhang et al., 2011; Martins et al., 2016). Figure 4(b) shows the magnified XRD profiles of CN, CQD/CN and 1-BCQD/CN between 26–29°, where the (002) peaks were respectively located at 27.34°, 27.32° and 27.37° respectively. Generally, a slight right-shift in the XRD peak is an implication of narrower interlayer and intralayer distances (Zhang et al., 2018). The full width at half maximum (FWHM) of the (002) peak was sequenced in an increasing order of CN, CQD/CN and 1-BCQD/CN, which indicated that 1-BCQD/CN possessed the greatest crystallite size in comparison with that of CN and CQD/CN.
The molecular structure and chemical composition of CN, CQD/CN and 1-BCQD/CN were also analyzed via FTIR as shown in Fig. 4(c). For CQD/CN and 1-BCQD/CN nanocomposites, the broad band at 3260 cm− 1 was ascribed to the stretching of -OH groups with the presence of CQD and BCQD (Zhang et al., 2017b). Apart from that, there were no observable changes to the transmittance peaks of the samples. This once again affirmed the well-preserved chemical framework of CN with the coupling of CQD and BCQD. The sharp peaks located in the range of 1200–1650 cm− 1 could be attributed to the stretching modes of -C-N heterocycles, displaying peaks located at 1646 cm− 1, 1578 cm− 1, 1411 cm− 1, 1329 cm− 1, and 1235 cm− 1 (Kumru et al., 2019; Seng et al., 2020). On the other hand, the peak at 1081 cm− 1 could be associated to the oxygenated functional group C-O (Wang et al., 2017b). The typical breathing mode of tri-s-triazine unit of g-C3N4 homojunction was represented by the intense peak at 808 cm− 1 (Zhang et al., 2020a).
To gain a deeper understanding of the chemical interactions, XPS analysis was carried out to identify the elemental composition and bonding on the surface of the photocatalysts. As shown in Fig. 5(a), the survey spectra of CN, CQD/CN and 1-BCQD/CN show that the samples were primarily composed of C, N and O. The atomic concentrations and weight percentages of each element are summarized in Fig. 5(e). Intriguingly, the B element was not detected in the 1-BCQD/CN sample, which could be due to the low quantity of dopant used on the hybrid nanocomposite. Nevertheless, the successful integration of the B dopant was verified through EDX analysis as discussed earlier (see Fig. 2(f)). Figure 5(b) presents the deconvoluted C 1s spectra of 1-BCQD/CN, where four prominent peaks at 288.85, 288.15, 285.70 and 284.72 eV were observed. These peaks corresponded to the C-N-C bonds, sp2-hybridized atomic carbon in N-C = N of the aromatic systems, C-O/C = O and C-C/C = C coordination, respectively (Liu et al., 2011; Qiao et al., 2015; Tan et al., 2015; Qin and Zeng, 2017). Besides that, the N 1s core level spectrum as depicted in Fig. 5(c) displayed a dominant peak centered at 398.57 eV, which could be associated to the pyridine N atom of the s-triazine rings (Sun et al., 2015; Cao et al., 2018). The N peak at 399.64 eV could be attributed to the pyrrolic N atom as part of the π-bonding system while the peak at 400.95 eV was ascribable to the graphitic N atom bonded to three carbon atoms within the aromatic rings (Kundu et al., 2010; Qiao et al., 2015; Fontelles-Carceller et al., 2016). The O 1s spectra, as shown in Fig. 5(d) was deconvoluted to two fitted peaks: 533.09 eV and 531.71 eV, which corresponded to -OH and C-O/C = O bonds respectively. These represented the oxygenic components of the B-doped CQDs which was similarly reported in literature (Qin and Zeng, 2017; Kesarla et al., 2019).
The optical properties of the as-developed photocatalysts were studied using UV-Vis DRS. As presented in Fig. 6(a), the adsorption bands of CN, CQD/CN and 1-BCQD/CN were categorized under the violet-blue regime of the visible spectrum. As observed from Fig. 6(a), the samples presented similar optical absorbances with absorption edges of CN, CQD/CN and 1-BCQD/CN at ca. 460, 453 and 452 nm respectively. As the changes in the absorption edge of bulk and CQDs-based CN was relatively insignificant, it is reasonable to infer that CQDs primarily served as electron transfer conduits to enhance the transportation of photoinduced charge carriers (Seng et al., 2020). The measurements obtained were similar to the typical optical absorption of g-C3N4 published in the previous studies (Liu et al., 2011; Yan et al., 2020).
The optical band gap energy (Eg) of a semiconductor is defined as the energy difference between the highest occupied state in the VB and the lowest unoccupied state in the CB. The Eg can be determined via the construction of a Tauc plot i.e., a graph of [F(R).hv]n vs. photon energy (hv) (Viezbicke et al., 2015). The absorption coefficient, Planck constant and optical frequency are denoted as F(R), h and v respectively whereas the value of exponent n depends on the transition nature of the material (Xu and Gao, 2012; Kumar et al., 2017). As presented in Fig. 6(b), the Eg values of CN, CQD/CN and 1-BCQD/CN were approximately 2.70, 2.76 and 2.78 eV respectively. The widening of Eg upon CQD or BCQD doping could be ascribed to the quantum size effect as well as the presence of structural defects in the hybrid nanocomposites (Liu et al., 2007; Klubnuan et al., 2016).
3.2 Evaluation of photocatalytic activity
The photocatalytic activities of CN, CQD/CN and BCQD/CN (with varying mass loadings of the B dopant) were assessed via the photodegradation of organic dye RhB. Prior to light illumination, the suspension of photocatalyst and dye solution was stirred for 30 min without light to establish an adsorption-desorption equilibrium. The almost-complete saturation at the solid-liquid interface minimizes the effect of physical interaction on the changes in RhB concentration as photocatalytic reaction occurs (Zhang et al., 2016a). The photocatalytic experiments were carried out under the irradiation of an 18 W LED light for a total duration of 4 hrs. The primary objectives of this study are: (i) to analyze the effect of incorporating B-dopant on CQDs-decorated n/n junctioned g-C3N4; and (ii) to investigate the optimum mass loading of B on the hybrid sample. Control experiments were also conducted following the procedures outlined in Sect. 2.4, but without (i) the photocatalyst sample; and (ii) light source (see Supplementary Information – Sect. 1.0). In both conditions, no significant discoloration of RhB was observed, which confirmed the indispensable roles of the photocatalyst and excitation source for the photodegradation process. Figure 7(a) shows the graph of ln (C0/C) against irradiation time of each photocatalyst sample for the evaluation of the k value according to the Langmuir-Hinshelwood model. The best-fitted lines signified the linear relationship between ln (C0/C) and irradiation time, thereby confirming that the photocatalytic degradation of RhB dye was in accord with the first-order kinetics.
Figure 7(b) depicts an overview of the photodegradation efficiencies and kinetic rate constants of CN, CQD/CN and BCQD/CN with varying mass loadings of B (0.5-BCQD/CN, 1-BCQD/CN, 2-BCQD/CN and 3-BCQD/CN). As observed, the photocatalytic performances were in the order of: 1-BCQD/CN > 2-BCQD/CN > 3-BCQD/CN > 0.5-BCQD/CN > CQD/CN > CN. In comparison to pure CN and CQD/CN, all B-modified CQD/CN photocatalysts displayed a pronounced improvement in photoactivities. The optimum 1-BCQD/CN sample achieved a remarkable 96.8% removal of RhB dye in 4 hrs under LED light illumination with an associated k value of 1.39 × 10− 2 min− 1. This translated to an improvement factor of 3.6 and 2.8 over pure CN and CQD/CN, which validated the significance of heteroatom B doping on CQDs for photocatalytic activity enhancement. The discoloration process of RhB in the presence of 1-BCQD/CN is shown in inset of Fig. 7(d). Interestingly, the color of the organic dye was observed to change progressively from its characteristic pink shade to orange and yellow, before turning colourless eventually.
As demonstrated in Fig. 7(b), the mass loading of B on CQD/CN did not impose prominent effects on the photodegradation efficiency of the hybrid material. However, its profound impact on the apparent rate constant was evident. Based on Fig. 7(b), the evaluated rate constant demonstrated an appreciable surge from 1.09 × 10− 2 min− 1 (0.5-BCQD/CN) to 1.46 × 10− 2 min− 1 (1-BCQD/CN). This observation highlighted the significance of precise mass loading of the B-dopant on CQD/CN in terms of expediting interfacial charge transfer and hindering electron-hole recombination. Figure 7(c) shows the absorbance trend exhibited by the RhB solution catalyzed over 1-BCQD/CN sample under visible light irradiation. For a RhB dye, the characteristic peak of the absorbance curve was noted at a wavelength of 554 nm prior to photodegradation. Apart from the gradual decrease in absorbance intensity with time, the peaks shifted gradually towards the blue region (left) of the visible spectrum as the illumination time lengthened. This was in congruent with the reported studies on the photodegradation of RhB catalyzed by g-C3N4/g-C3N4 homojunction systems, Ag3PO4 nanoparticles and Ag3VO4/β-Ag3VO4 nanocomposites (Gao et al., 2017; Xu et al., 2017; Phang et al., 2020a). Besides, the left-shifted peaks signified the generation of de-ethylated intermediates as the chromophores and aromatic rings of the RhB dye undergone deconstruction (Li et al., 2008; Chiu et al., 2019)
3.3 Photoelectrochemical and PL analysis
To further study the charge migration behavior and separation of photo-generated electron-hole pairs across the interface of BCQD/CN photocatalyst, transient photocurrent response and Mott-Schottky plots were also constructed. As presented in Fig. 8(a), the transient photocurrent measured were plotted with respect to time for CN, CQ/CN and 1-BCQD/CN with alternate cycles of on-off visible light irradiation to study their photoelectronic properties. The electric current generated by the photo-excited charge carriers were measured via transient photocurrent as the visible light pulse was turned on and decayed promptly with the light shut off (Xiang et al., 2011; Phang et al., 2020a). It is observed that there was a significant improvement in the transient photocurrent response exhibited by 1-BCQD/CN. The relatively high current density of 1-BCQD/CN is an implication of the boosted effectiveness in electron-hole pairs separation which may be ascribed to the presence of BCQDs. The doping of BCQDs on CN played a significant role as the channels for electron transport (as discussed previously in Sect. 3.2) and this inference was further supported by the generally low Fermi level of CQDs (-0.3 eV) relative to that of CN (-0.61 eV) (Tian et al., 2017; Wang et al., 2017a). Thus, the photoinduced electrons were prompted to shuttle from CN to BCQDs, in which the BCQDs served as electron traps, to effectively retard the recombination of electron-hole pairs (Wang et al., 2018; Di et al., 2020).
Moreover, Mott-Schottky analysis were performed as an effort to analyze the interrelation between applied potential and capacitance space charge region. Figure 8(b) presents the Mott-Schottky plots of the photocatalyst samples and the flat band potential (Efb) was evaluated via the extrapolation of tangent lines at the x-axis intercept whereby C− 2 = 0. From the Mott-Schottky plot, the Efb of CN, CQD/CN and 1-BCQD/CN were determined as -0.70, -0.74 and − 0.77 V vs. Ag/AgCl respectively. The Efb was marginally lowered with the incorporation of CQD and BCQD, indicating a depletion in photo-generated holes on the surface of the hybrid photocatalysts (Rangaraju et al., 2009; Liu et al., 2014). On top of that, Mott-Schottky plots of the samples displayed positive slopes which is the typical characteristic of an n-type semiconductor (Guan et al., 2020). Hence, this verified the successful construction of photocatalytic nanocomposites featuring n/n homojunction via coupling of g-C3N4 derived from urea and thiourea. The measured flat band potentials are with respect to NHE) using the conversion equation as expressed in Eq. (3) whereby pH value was approximately 7.0 and EAgCl = 0.197 V at ambient temperature (Babu et al., 2018; Li et al., 2020a):
Efb(vs.NHE) = Efb(vs.Ag/AgCl)+E(AgCl) + 0.059pH (3)
Upon conversion, the Efb (vs. NHE) of CN, CQD/CN and 1-BCQD/CN were evaluated as -0.09, -0.13 and − 0.16 V respectively. It is well-entrenched that the Ecb of an n-type semiconductor is generally 0.1–0.3 V more negative than its Efb, as governed by the electron effective mass and carrier concentration (Tian et al., 2015). In this case, the difference in voltage between Efb and Ecb was set at 0.3 V. Therefore, the Ecb were calculated as -0.39, -0.43 and − 0.46 V for CN, CQD/CN and 1-BCQD/CN respectively. Subsequently, the valence band potential, Evb were evaluated according to the equation as follows:
The Evb of CN, CQD/CN and 1-BCQD/CN were obtained as 2.31, 2.33 and 2.32 V respectively.
The PL spectra exhibited by CN, CQD/CN and 1-BCQD/CN were also presented in Fig. 8(c). Hitherto, the governing factor of PL mechanism has yet to be thoroughly elucidated and remains highly debatable as there are multiple argumentations regarding this subject matter. The proposed concepts include quantum confinement effect, functional groups, structural defect, surface passivation and functionalization, etc (Wang et al., 2014; Choi et al., 2016; Das et al., 2017). Nonetheless, it has been widely perceived that the intensity of PL emission is closely associated with the recombination of photo-excited charge carriers. Thus, PL spectra are commonly utilized to interpret the migration behavior, separation and transfer mechanism of electron-hole pairs (Zhu et al., 2019). Based on Fig. 8(c), the PL emission spectra of CN, CQD/CN and 1-BCQD/CN featured a prominent luminescence peak ranging from 420–520 nm. As it is well known, intense fluorescence (FL) emission implies a high rate of electron-hole recombination, whereby photogenerated holes from the lower energy states and electrons from the higher energy states recombine to release energy in the form of light (Wang et al., 2020). It is observed from Fig. 8(c) that CN exhibited the strongest PL emission spectra. On the contrary, relatively low PL emission peaks were demonstrated by CQD/CN and 1-BCQD/CN, reflecting the suppressed electron-hole recombination in the presence of pristine or B-doped CQDs. This also reaffirmed the role of CQDs in the hybridized photocatalytic nanocomposites as an electron transport conduit, considerably boosting the effectiveness of charge carrier transport and separation. Fundamentally, a low charge carriers recombination rate is highly beneficial towards the photocatalytic efficiency of a semiconductor nanomaterial.
3.7 Free radical scavenging test and plausible photocatalytic charge transfer mechanism
A series of scavenging tests were performed to identify the role of each reactive species in the photocatalytic mineralization of RhB and to propose a plausible photocatalytic mechanism over BCQD/CN. The declined photocatalytic performance in the presence of each scavenger implied the importance of the associated reactive species. The scavenging chemicals, TEOA, BZQ and IPA were applied to capture the relevant reactive species, h+, ∙O2− and ∙OH respectively. The changes in concentration of RhB and degradation efficiency of BCQD-CN are summarized in Fig. 9(a). As observed in Fig. 9(b), it is evident that there was a drastic decrease in photoactivity upon the addition of TEOA (scavenger of h+), indicating that h+ species were indispensable for RhB photodegradation. Precisely, h+ reactive species held a crucial role in the generation of radicals including ∙OH and OH−, which actively took part in the mineralization of RhB ions (RhB+) into photodegraded products. In the absence of h+ species, the generation of ∙OH radicals is not viable as the oxidation potential of water (H2O/∙OH = + 2.73 eV) is more positive than the VB potential of the photocatalyst sample (Wang et al., 2017a). Nevertheless, ∙OH radicals could also be generated via a different pathway whereby the reduction of elementary oxygen leads to the formation of ∙O2− species, which are then further reduced to ∙OH radicals through multiple-electron reduction reactions. In addition, the VB potential of BCQD/CN was more positive than the standard redox potential of waterborne hydroxyl radicals (∙OH/OH− = +1.99 eV), rendering the oxidation of OH− feasible (Wu et al., 2015). In short, these reactive species are crucial in the degradation and mineralization of RhB+ under light irradiation.
Based on the scavenging test, a plausible step-by-step charge transfer mechanism for RhB photodegradation is presented in Fig. 10. The entire photo-reaction is initiated by the irradiation of visible light, where the electrons from VB are excited and directed towards CB. Upon the departure of electrons, holes are generated at the VB concurrently. Due to the difference in CB and VB levels of g-C3N4 homojunction derived from urea and thiourea, the photo-generated electrons and holes are separated more effectively in comparison to the charge transfer within a pristine g-C3N4 (Liu et al., 2018). Specifically, the flow of photogenerated electrons (e−) is directed from the higher CB level of thiourea to that of urea while majority of the photogenerated holes (h+) are prompted to flow from the lower VB level of urea to that of thiourea. The opposite direction in flow of charge carriers minimizes the recombination possibility for electrons and holes. Furthermore, the doping of CQDs with heteroatoms B enriched the presence of structural defects on the functionalized surface of BCQDs, effectively trapping more photo-excited charge carriers and hampered the recombination of electrons and holes (Peng et al., 2020). Consequently, the photoinduced electrons from VB of CN are readily transferred to the BCQDs due to their lower energy levels (Wang et al., 2017a). The electrons are also captured and accumulated in the BCQDs attributed to their excellent conductivity and large capacity for electron storage (Ong et al., 2017). The electrons accumulated in the BCQDs and remnant of electrons in CN are primarily responsible for the reduction of O2 to generate superoxide anion radicals (∙O2−). The reduction process is feasible as the CB levels of the nanocomposite is more negative than the redox potential of O2/∙O2− at -0.16 V (Krumova and Cosa, 2016). The ∙O2− radicals are indispensable for the mineralization of RhB dye as evidenced by the scavenging test. On the other hand, the photoinduced holes in VB are gradually transported to the surface and participate in the oxidation process of hydroxyl species into hydroxyl radicals (∙OH) as the VB level of BCQD/CN is more positive as compared to the redox potential of OH−/∙OH at + 1.89 V (Armstrong et al., 2015). It is also worth mentioning that the change in redox potential ability upon hybridization is negligible as ratiocinated from the evaluated VB and CB energy levels of pristine CN, CQD- and BCQD-based CN. Therefore, it is reasonable to deduce that the BCQD and/or CQD mainly acted as channels to facilitate electron transport and storage for the photocatalytic enhancement. Another factor that should be taken into consideration for RhB photodegradation is the challenge of possible colloidal instability, competition over active sites and susceptible hole scavenging posed by the inorganic chloride anions (Cl−) (Chong et al., 2010). Nonetheless, the presence of π-π interactions on the exterior of the nanocomposite photocatalyst enhanced the adsorption of organic pollutants, resulting in a boosted photocatalytic efficiency. Ultimately, the mineralization process of RhB pollutants was expedited by reactive species including ∙OH, ∙O2− and h+ in the formation of photo-degraded products.
3.8 Recyclability test
In addition to the initial photocatalytic performance, the reusability of photocatalysts is also critical for practical and long-term applications. For the evaluation of its photo-stability, the optimal sample, 1-BCQD/CN was assessed and recycled for 3 consecutive cycles under identical conditions. The photocatalyst sample was separated from the dye solution via vacuum filtration and washed copiously with distilled water prior to oven-drying for the subsequent cycle. To account for the weight loss in the recovered photocatalyst sample, the amount lost was compensated by a decreased volume of dye solution following a fixed ratio of 0.15 g photocatalyst to 100 mL RhB solution (10 mg/L). The recyclability test of 1-BCQD/CN is presented in Fig. 11(a). It is apparent that the photoactivity of 1-BCQD/CN was successfully retained even after 3 successive cycles without any observable regression in its photocatalytic performance. Precisely, the 1-BCQD/CN photocatalyst sample maintained 98.6 % of its original activity after 3 consecutive runs. Furthermore, there were no significant changes to the molecular structure and chemical composition of the recycled 1-BCQD/CN with reference to its fresh counterpart as evidenced by their respective FTIR spectra presented in Fig. 11(b). This suggests that 1-BCQD/CN is superior in term of its stability and durability, which are highly advantageous from the economic and sustainability perspectives for probable industrial applications in the future.