Self-Doped Carbon Dots Decorated TiO₂ Nanorods: A Novel Synthesis Route for Enhanced Photoelectrochemical Water Splitting

Abstract

Herein, we have successfully prepared self-doped carbon dots with nitrogen elements (NCD) in a simple one-pot hydrothermal carbonization method, using L-histidine as a new precursor. The effect of as-prepared carbon dots was studied for photoelectrochemical (PEC) water splitting by decorating NCDs upon TiO₂ nanorods systematically by changing the loading time from 2 h to 8 h (TiO₂@NCD2h, TiO₂@NCD4h, TiO₂@NCD6h, and TiO₂@NCD8h). The successful decorating of NCDs on TiO₂ was confirmed by FE-TEM and Raman spectroscopy. The TiO₂@NCD4h has shown a photocurrent density of 2.51 mA.cm⁻², 3.4 times higher than the pristine TiO₂. Moreover, TiO₂@NCD4h exhibited 12% higher applied bias photon-to-current efficiency (ABPE) than the pristine TiO₂. The detailed IPCE, Mott–Schottky, and impedance (EIS) analyses have revealed the enhanced light harvesting property, free carrier concentration, charge separation, and transportation upon introduction of the NCDs on TiO₂. The obtained results clearly portray the key role of NCDs in improving the PEC performance, providing a new insight into the development of highly competent TiO₂ and NCDs based photoanodes for PEC water splitting.

1. Introduction

Rapidly spiking global energy demands and pollution caused by the depletion of fossil fuels necessitated the development of natural and renewable sources of energy [1]. Hydrogen is an excellent contender capable of replacing fossil fuels owing to its both eco-friendly and reusable nature. Photoelectrochemical (PEC) water splitting is the most reliable and popular method employed for converting solar light energy into clean and sustainable chemical fuels, such as hydrogen [2,3]. The initial study on photocatalytic water splitting using TiO₂ was published way back in 1972 [4]. Since then, different types of semiconductor materials including ZnO, [5] BiVO₄, [6] WO₃, [7] Fe₂O₃, [8] SrTiO₃, [9] C₃N₄ [10], and Ta₃N [11] were reported as photoelectrodes for PEC. The TiO₂ material is considered as the most competent semiconductor for investigating PEC devices due to its characteristics such as advantageous band-edge positions, ease of fabrication, abundance, excellent photocorrosion resistance, eco-friendliness, and cost effective nature [12]. However, application of TiO₂ in PEC has been constrained by comparatively greater band gaps for its rutile (3.0 eV) and anatase (3.2 eV) phases [12], severe bulk charge recombination, and slow OER kinetics [13]. As a result, numerous attempts were made to surpass the limitations, such as use of dopants [14], formation of heterojunctions [15], surface modification [16], introduction of defects [17], and quantum dot sensitization [15].

Recently, carbon dots (CDs) have been gaining enormous attention by virtue of their fascinating characteristics such as low cost, simple synthesis, functionalization, superior chemical inertness, and photobleaching resistance. Most essentially, CDs are a viable alternative for heavy metal-based QDs and organic dye, owing to its low toxicity with environmental friendliness [18,19,20]. Since the last decade, astonishing progress has been made in the preparation of CDs either in the top-down or bottom-up route [21,22]. However, new inexpensive, large-scale, and green synthetic approaches of CDs still need to be developed. For instance, a study on the CQDs/BiVO₄ and CQDs/NiFe-LDH/BiVO₄ demonstrated that after the decoration of CDs on their respective semiconductor, negatively shifted onsite potentials and enhanced charge injection rate were observed in PEC water splitting [22,23,24]. In addition, CDs, such as CQDs/TiO₂ [11] CQDs/ZnO [25], CQDs/WO₃ [26], CQDs/BiVO₄ [1], and CQDs/bFe₂O₃ [27], etc., can improve the light harvesting nature of photoanode in ultraviolet region and expand the range of visible region.

The CDs decorated TiO₂ films have been reported earlier from different origin materials by different methods and utilized as photoanode for PEC [23]. Zhou et al. utilized glucose as precursor and alkali-assisted ultrasonic chemical method to prepare CDs; and spin-coated TiO₂ film with CDs solution [15]. Wang et al. employed a hydrothermal method to synthesize CDs from phloroglucinol [24]. Usually, for photo-driven reactions, N-doped carbon dots exhibit improved activity both theoretically and experimentally than CDs, owing to beneficial quantum confinement and were capable of creating defect-rich heterostructures [25,26]. Based on the N-doping source material, light-harvesting ability and energy levels can be modulated [27], while the functionality of NCDs may interpret the interaction with the semi-conducting material [28]. Han et al. described the process of preparation of N-doped CDs (NCDs) anchored to TiO₂ photoanode in electrochemical and hydrothermal methods by using graphite rods and ammonia to obtain a nitrogen-doped CDs (NCDs) solution. This report has demonstrated the enhanced PEC efficiency due to an increased interface charge transfer [12,29]. The report by Wang et al. on NCDs@TiO₂ showed an enhanced photocatalytic property owing to its extended light responses with narrowed bandgap upon introduction of NCDs [30]. However, due to the complexity of NCDs with regards to energy states and chemical structure, the mechanism of NCDs in boosting PEC performance remains unknown [25,31]. Moreover, synthesis of CDs and preparation of photoanode was proceeded in multiple steps, which again increases the preparation cost of the electrode [32,33]. Therefore, it is of critical importance to prepare at low cost, as well as understand the nanostructure of NCDs, their interfacial interactions with semiconductor materials and further developments of NCDs.

In the present study, we report the synthesis of new NCDs decorated TiO₂ film in a simple one-pot hydrothermal method using L-histidine as source material. The effect of NCDs on TiO₂ nanorod film for PEC water splitting has been analyzed systematically by changing the NCDs’ loading time from 2 h–8 h. The prepared photoanodes are named as TiO₂@NCD2h, TiO₂@NCD4h, TiO₂@NCD6h, and TiO₂@NCD8h. NCDs loaded photoanodes showed higher PEC performance than pristine TiO₂, suggesting the contribution of NCDs towards enhancing the performance of PEC. The highest efficiency was found for TiO₂@NCD4h (2.51 mA.cm⁻²), 3.4 times greater than pristine TiO₂ (0.73 mA.cm⁻²). The higher photocurrent for TiO₂@NCDs could be ascribed to the improved light harvesting property, decreased rate of recombination, and increased charge carrier density. The detailed characterization of NCDs and NCD loaded TiO₂ and PEC water splitting performance analysis were performed and discussed.

2. Results and Discussion

2.1. Characterization

FE-SEM and HR-TEM analyses were executed in order to assess the successful loading of NCDs on TiO₂ and their morphology and the obtained images are illustrated in Figure 1 and Figure 2. The FE-SEM analysis of pristine TiO₂ film (Figure 1a) has shown dense nanorod morphology of TiO₂ which have perpendicularly grown on FTO glass showing an average length of ~2.8 μm and width of ~150 nm. Moreover, no obvious changes in the size and morphology of TiO₂ were observed in Figure 1b, even after dipping for 8 h in NCDs solution. Further, HR-TEM (Figure 1c) analysis confirmed the nanorod morphology of TiO₂. Moreover, the observed lattice fringes’ distance in Figure 1d was 0.35 nm, which corresponds to the d-spacings of the rutile TiO₂ (101) planes, which has well-matched with XRD results [12]. Further, HRTEM image of TiO₂@NCD4h (Figure 2) showed that NCDs are uniformly loaded on the TiO₂ nanorods and appeared in a sphere and ellipsoidal morphology with particle size ranging from 4 to 10 nm. In addition, 0.21 nm lattice spacing was observed for NCDs particle, associated with the (100) facet of NCDs (Figure S3) [15,18,34]. Moreover, to further investigate the distribution of Ti, O, C, and N elements, the elemental mapping analysis was executed, and the respective results are displayed in Figure 2b. The obtained results have shown even distribution of C and N elements on TiO₂ nanorods’ surface, suggesting the successful decoration of NCDs on the TiO₂.

The crystalline structure of the as-synthesized TiO₂ and the effect of NCDs loading time on TiO₂ crystallinity (TiO₂@NCD2h, TiO₂@NCD4h, TiO₂@NCD6h, and TiO₂@NCD8h) and the orientation growth were examined using XRD analysis. The obtained XRD peaks are displayed in Figure 3. The diffraction peaks of pristine TiO₂ films appeared at 36.10°, 41.27°, 54.39°, 62.86°, and 69.80° and correspond to the (101), (111), (211), (002), and (112) crystal planes of tetragonal rutile structure [15,18,35,36]. Moreover, regardless of the loading time of NCDs, the peak positions are the same, but the (101) plane intensity has increased with NCDs loading time. The results suggest that the TiO₂ nanorod crystal structure does not get affected by loading NCDs but size of the crystal and preferred orientation directions sparsely get affected. Furthermore, no noticeable diffraction peak of NCDs was observed for TiO₂-NCDs, which could be attributed to the modest load of NCDs, lower than the minimum limitation of XRD detection [15,35].

As seen in Figure 4, the Raman peaks of TiO₂ located at 615.2, 450.5, and 240 cm⁻¹ correspond to (A_1g), (E_g), and multi-photon scattering process, respectively, and represent the TiO₂ rutile phase. The peaks which appeared at 1580 and 1333 cm⁻¹ can be attributed to the D (disordered sp²) band and G band of NCDs, respectively [37,38]. Thus, the Raman spectrum of TiO₂@NCD4h has shown five peaks, indicating successful fabrication of NCDs in TiO₂ nanorods. Furthermore, the enhancement of Raman intensity might be contributed by the increased crystallinity, and it is consistent with the XRD results.

The elemental composition and chemical binding of NCDs decorated on TiO₂ catalyst were determined by the XPS analysis (Figure 5). The survey scans illustrated the elements in two structures, e.g., C, N, Ti, and O elements in TiO₂@NCDs4h, whereas C, Ti, and O elements were in the pristine TiO₂ [12,39] (Figure 5a). An increase in the carbon content and the presence of N1s peak compared with the bare TiO₂ evidently confirm that the NCDs were successfully decorated on TiO₂. Ti2p spectra (Figure 5b) showed two representative peaks at 464.0 and 458.4 eV (difference: 5.6 eV), which correspond to the spin-orbit coupling for Ti2p1/2and Ti2p3/2, respectively, and was identical to those for TiO₂ [15]. Furthermore, Ti2p peaks of the TiO₂@NCDs4h structure have considerably shifted (by 0.2 eV) compared to bare TiO₂ (Figure 5b). This is due to the electronegativity of C/Ns, which increased the binding ability of extra-nuclear electrons, hence raising the binding energy. The C 1s spectra (Figure 5c) is fitted with three peaks corresponding to C-C, C-O & C-N, and C=O & C=N bonds at 284.8, 286.0, and 288.3 eV, respectively [40]. The N1s spectra (Figure 5d) shows three peaks which appeared at 396.7 eV, 401.4 eV, and 403.7 eV, ascribed to the pyridine-N, pyrrole-N, and Graphitic-N, respectively, indicating the carbon dot doped with the N element [41,42]. In Figure S4, the peak in the O1 s region of TiO₂@NCDs@4h was deconvoluted into three peaks at 532.3, 530.1, and 529.6 eV, which are assigned to O-H, C-O, or O-N, and Ti-O bonds in TiO₂@NCDs4h, respectively. In agreement with the above microstructure analysis, XPS results suggest that NCDs were successfully deposited on the TiO₂ [40,42,43].

The absorption properties of the materials are important parameters to estimate the light harvesting nature and energy levels to be used in solar energy conversions. The optical properties of the prepared NCDs in solution and on TiO₂ photoanode with changing time were analyzed systematically. The UV-vis spectra of NCDs in solution and TiO₂@NCDs thin films are shown in Figure 6a,b and c by changing the loading time. The absorption band of pristine TiO₂ ~400 nm represents the rutile TiO₂ band edge [44]. After introduction of the NCDs on TiO₂, the light harvesting property was enhanced with the increased amount of NCDs loading. The results suggest the successful decoration of NCDs and their contribution in improving the light harvesting property. Moreover, the Tauc plot Equation (S1) was employed to calculate the bandgap (Eg) of pristine TiO₂ and TiO₂@NCD photoanodes [36,45,46]. The calculated Eg values of TiO₂, TiO₂@NCD2h, TiO₂@NCD4h, TiO₂@NCD6h, and TiO₂@NCD8h were 3.12, 3.07, 3.03, 3.04, and 3.05 eV, respectively.

2.2. PEC Performance of the Photoanodes

The effect of newly prepared NCDs on photoelectrochemical water oxidation was studied systematically by changing the loading time of NCDs on TiO₂ film and comparing it with the pristine TiO₂. Figure 7a shows linear sweep voltammetry performance of TiO₂, TiO₂@2hNCD, TiO₂@4hNCDs, TiO₂@6h NCDs, and TiO₂@8hNCDs, and their photocurrent data at 1.23 V are illustrated in Table S2. The pristine TiO₂ film displayed a 0.73 mA.cm⁻² photocurrent at 1.89 V vs. RHE. After loading NCDs upon the TiO₂ film, an enhanced photocurrent was observed compared the pristine TiO₂, which instigated to perform optimization studies to improve the NCDs loading and thereby achieve optimum PEC performance using TiO₂ with NCDs. In order to optimize the TiO₂@NCDs photoanodes, the decorated NCDs on TiO₂ was controlled by monitoring the hydrothermal reaction. In Figure 7a, the photocurrents of four NCDs decorated TiO₂ (TiO₂@NCDs)-based photoanodes showed higher photocurrent than pristine TiO₂, indicating the contribution of NCDs in enhancing PEC performance of TiO₂. Significantly, the photoanode corresponding to TiO₂@NCD2h has displayed an improved photocurrent density of 2.33 mA.cm⁻² at 1.89 V vs. RHE, while pristine TiO₂ photoanode has shown 0.73 mA.cm⁻² at 1.89 V vs. RHE. Further, by increasing the loading time from 2 h to 4 h (TiO₂@NCD4h), the photocurrent density has also increased to 2.51 mA.cm⁻² at 1.89 V vs. RHE, which was 3.4 times greater than the pristine TiO₂. Moreover, TiO₂@NCD4h photoanode possesses both enhanced photocurrent density and smaller onset potential than pristine TiO₂. The higher photoresponse of NCD decorated TiO₂ might be due to the addition of NCD which could effectively promote the separation of photogenerated electron-hole, and promote the capture of water molecules and intermediates in the process of water decomposition by electrons and holes at the interfaces [15,47,48]. However, further increasing the NCDs loading by increasing loading time to 6 h (TiO₂@NCD6h) and 8 h (TiO₂@NCD8h) showed declined photocurrent density of 1.99 and 1.85 mA.cm⁻² at 1.89 V vs. RHE, respectively. In addition, onsite potentials also increased compared to the TiO₂@NCD4h-based films, possibly due to the variation in conductivity by decorated NCDs. This phenomenon will be further discussed in electrochemical impedance spectroscopy (EIS) section [29].

Chronoamperometric analysis was performed under chopped illumination for all the prepared electrodes at 1.89 V vs. RHE for 30 s to better understand photo response with time and stability. As depicted in Figure 7b, the photocurrent has rapidly increased immediately after illumination and sharply fell to zero upon stopping the illumination. It confirms that the prepared photoanodes have a well-reproducible photocurrent. Meanwhile, after decorating NCDs on TiO₂, the response speed was boosted compared with the pristine TiO₂, indicating that the presence of NCDs can significantly reduce the charge recombination in the intersection of electrolyte and photoanode surface. The highest photocurrent density was detected for the TiO₂@NCD4h, which is consistent with the observed LSV results. In order to understand the durability of the prepared electrode, 1 h of continuous illuminated chronoamperometric analysis was performed with a high performing photoanode (TiO₂@NCD4h) by comparing with the pristine TiO₂ based photo anode at same experimental condition. As displayed in Figure 7c, after continuous illumination for 1 h, TiO₂@NCD4h has retained 99% of its initial activity, which supports the excellent stability of NCDs decorated TiO₂ photo anode. The current densities have matched well with the LSV data. Both TiO₂ and TiO₂@NCD4h showed excellent stability after 1 h of continuous irradiation without any photocurrent decay.

The incident photon-to-current conversion efficiency (IPCE) was evaluated to analyze the contribution of various photons in obtaining solar photocurrent. The IPCE has been deduced by using the formula (1):

$$\mathrm{IPCE} (\%)=\frac{1240J\left(\lambda \right)}{\lambda P\left(\lambda \right)}\times 100 (\%)$$

(1)

where, P(λ), λ, and J(λ) are the intensity of a specific wavelength, wavelength of incident light, and photocurrent density at specific wavelength, respectively. Figure 8a shows enhanced IPCE after decorating NCDs on TiO₂ with highest IPCE of 29.76% at ~390 nm for TiO₂@NCD4h, which was ~3 times greater than the pristine TiO₂ (9.76%). The IPCE trend was consistent with the obtained photocurrent density and the enhanced IPCE region is in good agreement with the optical absorption properties. The improved IPCE after the introduction of NCDs to the TiO₂ reveals the contribution of NCDs in obtaining enhanced photocurrent density.

Besides, the applied bias photon-to-current efficiency (ABPE) has been calculated by using the Equation (2):

$$\mathrm{ABPE} (\%)=\frac{J\left(1.23-V\right)}{P}\times 100 (\%)$$

(2)

where P, J, and V are the power density of incident light (100 mW cm⁻²), photocurrent density (mA cm⁻²), and the applied bias (V vs. RHE), respectively. As seen in Figure 8b, the pristine TiO₂ reached maximum 0.11% ABPE at 0.88 V vs. RHE, while TiO₂@NCD4h reached 1.37% photo conversion efficiency at 0.36 V vs. RHE, 12 times greater than the pristine TiO₂ ABPE_, suggesting an effective electron-hole pairs separation after the introduction of NCDs [12].

To further comprehend the interfacial charge transfer kinetics at the intersection of photoanode and electrolyte, EIS was employed under illumination and the respective Nyquist plots are displayed in Figure 9a. The decreased radius order of semi-circle was TiO₂ > TiO₂@NCD8h > TiO₂@NCD6h > TiO₂@NCD2h > TiO₂@NCD4h. The smallest arc of the TiO₂@NCD4h compared with its counter parts demonstrates the improved interfacial charge transfer kinetics due to the introduction of NCDs [18]. Furthermore, using Zview software program, the EIS curves have been fitted with an analogous circuit model given in Figure 9a inset, where CPE, Rct, and Rs indicate the constant phase element, charge transfer resistance, and series resistance at the electrolyte/electrode interface, respectively. The observed Rct values of TiO₂ and NCDs decorated TiO₂ films (2 h–8 h) were 376.0 Ω, 273.6 Ω, 248.3 Ω, 277.2 Ω, and 297.4 Ω, respectively. The lowest R_ct value of TiO₂@NCD4h further demonstrates the advantage of NCDs decorated TiO₂ nanorods in enhancing the charge separation and transfer kinetics.

Mott–Schottky analyses have been executed to estimate the energy band position of pristine TiO₂ and TiO₂@NCDs, and the corresponding curves are displayed in Figure 9b and the data are depicted in Table S1. The positive slope of both curves indicates n-type semiconductor of TiO₂ [18]. The flat band (V_FB) potential can be calculated following the Equation (S1). The obtained V_FB values of TiO₂ and TiO₂@NCDs (2 h, 4 h, 6 h, and 8 h) were 0.17, 0.27, 0.34, 0.22, and 0.19 V vs. NHE, respectively, which could be accomplished by the X-axis intercept. Moreover, NCDs decorated TiO₂ films have shown decreased V_FB than pristine TiO₂, suggesting an increased band bending of the photoanode, favorable to enhance the charge transfer between the photoanode interfaces and electrolyte. Eventually, enhanced PEC was observed for NCDs decorated TiO₂ films. As per the available literature [49,50], the bottom of the conduction band (CB) was −0.1 V lower than the V_FB of an n-type semiconductor [50]. Therefore, the CB of TiO₂ and TiO₂@NCDs were estimated to be positioned at less than 0.1 V of their V_FB [33]. Based on the V_FB, the CB edge of TiO₂ and TiO₂@NCDs (2 h, 4 h, 6 h, and 8 h) were determined to be at 0.07, 0.17, 0.24, 0.12, and 0.09 V, respectively. Particularly, doping of NCDs promotes a downward shift in energy levels towards higher potentials and enhances the carrier density in TiO₂ [51].

3. Experimental

3.1. Materials

All chemicals were used directly without purifying any further. Hydrochloric acid (35%) was obtained from OCI Company Ltd., titanium butoxide (TBOT, 98%) was purchased from Sigma-Aldrich, L-histidine (98%) was procured from Alfa Aesar, and sodium sulfate anhydrous (99%) was obtained from Duksan. For all the experiments, Milli-Q water (MΩ 18) was used.

3.2. Preparation of Rutile TiO₂ Film (TiO₂):

The FTO coated glasses (1.5 mm × 2.5 mm, 8 Ω/cm²) were cleaned ultrasonically using detergent, milli-Q water, ethanol, and acetone for 1 h, respectively. TiO₂ film was synthesized by following a reported hydrothermal method with certain modifications [18]. Under continuous stirring, 0.33 mL titanium (IV) butoxide was added dropwise to 20 mL equal volumes of HCl (35%) and milli-Q water mixed solution until it turned translucent. The solution was then moved to a 50 mL autoclave lined with Teflon, and the FTO glass was placed against the walls of Teflon vessel, conducted side down, for 12 h and heated to 150 °C. The TiO₂ layer was completely cleaned with milli-Q water and ethanol after cooling to RT, before being sintered in air at 450 °C for 1 h.

3.3. Preparation of TiO₂@NCDs:

NCDs have been prepared using a hydrothermal approach (Scheme 1). First, 0.2 g of L-histidine was included in 20 mL mixture of milli-Q water and HCl in the ratio of 19:1. The solution was shifted to 50 mL autoclave, and two TiO₂ films on FTO glasses were inserted in the Teflon vessel with the TiO₂ side facing down. Then, the hydrothermal reaction was performed at 180 °C for 2, 4, 6, 8, and 10 h. The samples were denoted as TiO₂@NCD2h, TiO₂@NCD4h, TiO₂@NCD6h, TiO₂@NCD8h, respectively (Figures S1 and S2). The TiO₂@NCD films were extensively washed with milli-Q H₂O and ethanol upon cooling to RT. Then, copper wires and as prepared photoanodes were adhered using silver paint. The samples were air dried for 3 h. Finally, the samples were encased by nonconductive epoxy with the illuminated area of 1 cm² and left to rest in air for at least 3 h.

4. Conclusions

In conclusion, the new NCDs were successfully prepared in a simple one-pot hydrothermal synthesis method using L-histidine as an initial precursor and the as-prepared NCDs were decorated on the TiO₂ nanorod-based photoanode. The as-prepared NCDs and NCD decorated TiO₂ nanorods were well characterized using FE-SEM, HR-TEM, EDS elemental mapping, which revealed the nanorod morphology of TiO₂ and uniform distribution of CDs on TiO₂ surface while, XPS and Raman analyses have confirmed the successful self nitrogen element doping, preparation and decorating of NCDs on TiO₂ nanorod. The effect of NCDs decorated TiO₂ was tested for photoelectrochemical water splitting analysis systematically by changing the loading time of NCDs from 2 h to 8 h. The highest efficiency was observed for the TiO₂@NCD4h-based photoanode (2.51 mA.cm⁻²), which was a 3.4 times higher photocurrent density than the pristine TiO₂-based photoanode. It might be attributed to the increased light harvesting property with charge separation and transportation. The observed IPCE of TiO₂@NCD4h has shown 3 times higher quantum yield (29.76%) than pristine TiO₂ (9.76). In addition, the calculated ABPE was 12% higher for TiO₂@NCD4h than the pristine TiO₂, which revealed the enhanced light harvesting property of photoanodes upon loading the NCDs. Moreover, the reduced charge transfer resistance and higher charge carrier density, as observed from EIS and Mott–Schottky analyses, respectively, further support the advantage of newly prepared NCDs in enhancing the PEC performance by promoting effective charge separation and transportation. This study may open up new insights into the rational design and synthesis of highly efficient photoanodes for PEC water splitting.