The rapid consumption of fossil fuels has accelerated the global energy crisis and various environmental issues that threaten societal development. Photocatalytic hydrogen production has been recognized as a promising technique based on its direct utilization of clean and sustainable solar energy to realize solar-to-chemical fuel conversion [1-4]. Among various semiconductor photocatalysts, titanium dioxide (TiO₂) has been extensively investigated as one of the most promising candidates for photocatalytic water splitting based on its several advantages, including matched band energy, nontoxicity, low cost, and stability against photocorrosion [5-11]. TiO₂ crystals have three main polymorphs of rutile, anatase, and brookite. It is widely recognized that anatase TiO₂ generally possesses higher photocatalytic activity than that of rutile TiO₂ because of its high reactivity [12]. Recently, nanoscale rutile crystals have also been developed as promising photocatalysts since it was discovered that they play an active role in certain cases [13, 14]. Furthermore, rutile TiO₂ is a more thermodynamically stable state compared to the anatase and brookite phases [15]. However, rutile TiO₂ photocatalysts suffer from the same major issue as anatase photocatalysts, namely high recombination loss of photogenerated electron-hole pairs. Many attempts have been made to improve the separation efficiency of charge carriers to reduce recombination loss, including functional group grafting, construction of surface heterojunctions, and co-catalyst loading [16-25].

Since photocatalytic reactions typically occur on the surface active sites of photocatalysts, surface properties have a significant influence on photocatalytic activity based on their modification of charge transfer pathways [26, 27]. It has been well established that the grafting of inorganic acids, such as phosphoric, sulfuric, boric, and hydrochloric acid, can modify the surface properties of photocatalysts to enhance their photocatalytic activity [28-32]. The influence mechanisms of these inorganic acids have been examined from two main perspectives. One generally accepted mechanism is that acidification enhances the surface protonation of oxygen atoms to promote the formation of hydroxyl groups and oxygen adsorption, thereby improving charge carrier separation to boost reactivity [28-30]. For example, Cao et al. [28] reported that phosphoric acid treatment modifies phosphate groups on the surface of TiO₂ nanocrystals to form –Ti–O–P–OH, which improves photocatalytic degradation efficiency. He et al. [29] treated TiO₂ nanosheets with sulfuric acid to form surface hydroxyl groups and oxygen vacancies/Ti³⁺ species, which extended the lifetime of electron-hole pairs. Wu et al. [30] introduced phosphoric and boric acids to co-modify rutile TiO₂ nanorods using residual chlorine, leading to the formation of –Ti–O–P–OH and –Ti–Cl:B–OH group ends. Their results confirmed that the promotion of adsorption of O₂ facilitates the capture of photogenerated electrons based on acid co-modification. Dhandole et al. [31] proposed a synergistic effect between the OH-rich surfaces of HCl-treated rutile TiO₂ nanorods and metal oxide co-catalysts that contributes to enhanced charge transfer. The second mechanism is based on electrostatic interactions between surface functional groups and reactive species. Li et al. [32] demonstrated that modification using phosphate groups creates a negative field effect on the surface of BiOCl, which effectively traps positive holes to suppress charge recombination. Based on the research above, it can be concluded that inorganic acid treatment has a significant effect on the enhanced photocatalytic degradation of organic pollutants based on the generation of surface hydroxyl groups and inorganic oxyanions. Despite the remarkable merits of such surface modification strategies, their effects on the photocatalytic production of clean solar fuels have not been extensively studied thus far. Therefore, it remains a considerable challenge to develop efficient photocatalysts for hydrogen production using acid treatments.

In this paper, we report the facile synthesis of spindle-like rutile TiO₂ nanobundles via nitric-acid-induced assembly. First, lamellar protonated titanate (LPT) was prepared as a TiO₂ precursor based on our previous research [33, 34]. Subsequently, concentrated nitric acid was introduced to induce the assembly of rutile TiO₂ nanorods. The first three hours of light irradiation revealed an average hydrogen evolution of 402.4 μmol h⁻¹ for the as-prepared TiO₂ photocatalysts, which is 3.1 times higher than that of a Degussa P25-TiO₂ reference. The mechanism for enhanced hydrogen production was determined to be related to the loss and recovery of photocatalytic efficiency. The synthesis strategy and the role of surface-grafted nitrate groups in hydrogen production are illustrated in Scheme 1. The proposed surface modification strategy with inorganic acid was confirmed to be an effective method to improve photocatalytic hydrogen production based on a novel mechanism of pollutant degradation reactions.

LPT was prepared as a TiO₂ precursor based on our previous studies [33, 34]. In a typical procedure, 9 mL of tetrabutyl titanate is first dispersed into 66 mL of absolute ethanol. This solution is then added into 90 mL of deionized water dropwise under vigorous stirring. The resulting suspension is then heated to 70 ℃ for 2 h under mechanical stirring to evaporate most of the solvent. Next, 300 mL of 1-M NaOH aqueous solution is added to the suspension. After continuous stirring for another 12 h, the precipitate is washed with deionized water and ethanol several times to obtain a wet LPT precursor.

The acid treatment process was conducted by directly mixing 6.2 g of wet LPT with 60 mL of 65 wt% concentrated HNO₃ in a 150 mL flask. The mixture was refluxed at 80 ℃ for 4 h under continuous stirring. The final precipitate was collected via centrifugation and washed with ethanol and deionized water, then dried in a vacuum oven at 60 ℃ for 4 h. The as-obtained product was denoted as rutile TiO₂ nanobundles (R-TiO₂).

To clarify the effects of nitrate group grafting, a series of samples were synthesized for comparison. First, LPT was treated with 1-M dilute HNO₃ and the product was labelled as R-TiO₂-1M. Another controlled synthesis was conducted by employing a conventional hydrothermal method to produce anatase TiO₂ without the addition of concentrated HNO₃, which was labelled as A-TiO₂-No [33, 34]. The concentrated HNO₃ treatment was then applied under the same conditions to yield a post-treated sample labelled as A-TiO₂-acid.

X-ray diffraction (XRD) patterns were collected using a powder XRD instrument (SmartLab-SE) operating at 40 kV and 45 mA. The morphologies of the samples were examined using field emission scanning electron microscopy (FESEM, SU8010). Field emission transmission electron microscopy (FTEM) and high-resolution transmission electron microscopy (HRTEM) images were obtained using a Tecnai G² F30 instrument (FEI, Holland). N₂ adsorption-desorption isotherms were acquired at 77 K using a Micrometrics ASAP24600 instrument. Fourier transform infrared (FT-IR) spectra were obtained using a Bruker INVENIO R instrument with the KBr disk method. X-ray photoelectron spectroscopy (XPS) data were analyzed using an EscaLab Xi+ instrument. Ultraviolet-visible (UV-vis) diffuse reflectance spectra (DRS) were obtained using a UV-vis spectrophotometer (UV-3600, Shimadzu, Japan) with BaSO₄ powder as a reflectance standard. Photoluminescence (PL) spectra were measured using a fluorescence spectrometer (RF-5301PC, Shimadzu, Japan) with an excitation wavelength of 360 nm. Photoelectrochemical measurements were carried out in a 1-M Na₂SO₄ solution using an electrochemical workstation (CHI650E, Chenhua Com., China) with a standard three-electrode system. Electrochemical impedance spectroscopy (EIS) was performed in a frequency range of 10⁵–1 Hz with an open potential (AC voltage of 5 mV in amplitude) in a 0.1-M KCl solution containing 5 mM of Fe(CN)₆³⁻/Fe(CN)₆⁴⁻. Zeta potential analysis was performed using a Malvern Zetasizer Nano ZS.

Photocatalytic H₂ evolution reactions were carried out in a closed circulation system using a CEL-HXF300 lamp. The light intensity incident on the photocatalyst surface was measured to be 0.38 W/cm². In a typical photocatalytic test, 100 mg of the as-prepared photocatalyst is uniformly suspended in 100 mL of aqueous solution containing 90 mL of deionized water and 10 mL of methanol in a reaction cell. Next, 1.5 wt% Pt is photodeposited from an H₂PtCl₆·6H₂O aqueous solution. Next, 50 mg of photocatalyst is dispersed into 90 mL of deionized water and 10 mL of triethanolamine (TEOA). The resulting suspension is then bubbled with N₂ for 10 min to remove any residual air. Photocatalytic H₂ evolution is analyzed using an online gas chromatographer (FULI GC9790, TCD) with N₂ as a carrier gas. The apparent quantum yield (AQY) for H₂ production is measured using a 300-W Xe lamp equipped with a 365 ± 15-nm band-pass filter as a monochromatic light source. The average energy density of the proposed photocatalyst was determined to be 15.3 mW/cm². AQY can be calculated using the following equation:

The crystallinities and phase compositions of a series of TiO₂ samples were characterized using XRD. As shown in Fig. 1, the R-TiO₂ and R-TiO₂-1M samples present distinct diffraction peaks at 2θ = 27.5°, 36.1°, 41.2°, 43.8°, 54.3°, 56.4°, 62.9°, and 69.1°, corresponding to the (110), (101), (111), (210), (211), (220), (002), and (301) reflection planes of rutile TiO₂ (JCPDS 21-1276), respectively. In contrast, the A-TiO₂ product obtained from the conventional hydrothermal method is a typical anatase crystal according to the standard JCPDS 21-1272 pattern. The A-TiO₂-acid product also exhibits an anatase structure, indicating that post treatment using concentrated HNO₃ under the same conditions used for the proposed photocatalyst does not induce a transition of the crystal phase. These results indicate that HNO₃ treatment during the formation of TiO₂ crystals induces the transition of LPT into a rutile TiO₂ structure [35, 36]. The surface area and pore distribution of R-TiO₂ nanorods were determined using N₂ adsorption-desorption isotherms (Fig. S1). One can see that the R-TiO₂ photocatalyst possesses a large specific surface area of 221.3 m² g⁻¹ with a pore diameter of approximately 1.4 nm. This large surface area can provide many reaction sites, which are beneficial for reacting with protons to produce H₂ gas. The FESEM images in Fig. 2(a) and (b) reveal the uniform spindle-like morphology of the R-TiO₂ sample. The FTEM images clarify the fine morphology of the spindles as bundles of TiO₂ nanorods (Fig. 2(c)). The R-TiO₂-1M sample treated with a diluted HNO₃ solution also produces rutile TiO₂ nanobundles, but they are smaller than the bundles in the R-TiO₂ sample (Fig. S2(a)). In contrast, the A-TiO₂ sample prepared in the absence of a HNO₃ solution is composed of nanoparticles (Fig. S2(b)). The magnified FTEM image in the inset of Fig. 2(c) and HRTEM image in Fig. 2(d) indicate that each bundle consists of a dozen of nanorods with dimensions of approximately 300 nm in length and 100 nm in width. Regular lattice fringes can be clearly observed in the HRTEM image and the d-spacing of 0.320 nm can be attributed to the (110) crystallographic planes of rutile TiO₂. The lattice fringes are parallel to the axial direction of the nanorods, indicating that nitric acid treatment induces TiO₂ nanorods to grow along the (110) plane and adhere to each other to form bundles.

The chemical structures of the as-prepared TiO₂ samples were investigated using FT-IR spectroscopy, as shown in Fig. 3. All of the spectra exhibit strong and broad bands below 1000 cm⁻¹, which are attributed to the stretching mode of Ti–O bonds in TiO₂ networks. The peak at 1383 cm⁻¹, which corresponds to the bending mode of nitrate groups [37], can only be observed in the FT-IR spectra of the R-TiO₂ and R-TiO₂-1M samples, indicating the grafting of nitrate groups onto the TiO₂ surfaces during the nitric-acid-induced assembly process. One can see that the signal of nitrate groups of R-TiO₂ is much stronger than that of the nitrate groups of R-TiO₂-1M. The peak at approximately 1634 cm⁻¹ originates from the bending vibrations of adsorbed H₂O on the sample surfaces. Another broad peak in the range of 3100–3600 cm⁻¹ corresponds to the stretching vibrations of O–H groups. One can see that the R-TiO₂ sample exhibits the most intense O–H stretching. These results indicate that nitrate group grafting promotes the formation of hydroxyl groups to produce an OH-rich surface, which is consistent with other inorganic acid treatments described in the literatures [28-31]. To clarify the chemical states of the TiO₂ samples, the XPS technique was employed. The results are presented in Fig. 4. All spectra were calibrated using the binding energy of standard C 1s of 284.6 eV. The high-resolution Ti 2p XPS spectrum of R-TiO₂ exhibits two major peaks at 458.5 and 464.3 eV, corresponding to the 2p_3/2 and 2p_1/2 peaks, respectively, of Ti⁴⁺ in the TiO₂ structure (Fig. 4(a)). The binding energy of Ti 2p displays a clear shift to higher values relative to R-TiO₂-1M, indicating electronic interactions between surface Ti atoms and nitrate groups. Accordingly, R-TiO₂ possesses an N content of 0.55 at%, while the surface N of R-TiO₂-1M is undetectable (Fig. 4(b)). The high-resolution N 1s XPS spectrum exhibits a nearly symmetric peak at 399.6 eV, suggesting that N atoms largely exist as surface-bonded N-containing groups, rather than doped states [38]. These results confirm the successful grafting of nitrate groups onto the rutile TiO₂ surface via concentrated nitric acid treatment.

The light absorption levels and band edges of photocatalysts are important properties that affect their performance. Fig. 5(a) presents the UV-vis DRS of the as-obtained samples. One can see that the rutile phase displays a red shift at the absorption edge compared to the anatase phase, indicating a wider light absorption range for rutile TiO₂ compared to that for anatase. The energy gap (E_g) values of the photocatalysts were calculated based on the converted Tauc plot presented in Fig. 5(b). The E_g values of both A-TiO₂ and A-TiO₂-acid were determined to be 3.22 eV, which is consistent with the results in the literature. The results also indicate that post treatment with nitric acid has a negligible influence on band energy. It should be noted that R-TiO₂ has a lower band gap (3.10 eV) compared to that of anatase, and this band gap narrowing can improve light absorption ability over a wider range. The PL spectra are presented in Fig. 5(c) to provide additional evidence for the noted optical properties and charge transfer efficiency. The emission band originates from band-band electronic transitions and excitonic PL from surface vacancies and defects [39, 40]. Similar to the UV-vis DRS results, post treatment with nitric acid has little influence on PL spectra. Compared to anatase TiO₂, an evident decrease in PL intensity can be observed in the rutile TiO₂ catalysts. These results demonstrate that the nitric-acid-induced assembly of rutile TiO₂ nanobundles results in an extended light absorption range and reduced charge recombination loss.

Photocatalytic efficiency in terms of hydrogen production was evaluated in a TEOA aqueous solution under 300-W xenon arc lamp irradiation. Fig. 6(a) presents a comparison of different types of TiO₂ samples with 1.5 wt% of Pt loading. The rutile TiO₂ photocatalyst R-TiO₂-1M exhibits a higher H₂ evolution rate compared to that of A-TiO₂, indicating that the rutile structure can function as an efficient photocatalyst. Post treatment of TiO₂ crystals with concentrated nitric acid increases the average H₂ evolution rate from 52.0 to 92.6 μmol h⁻¹. In contrast, introducing concentrated nitric acid during the formation of TiO₂ crystals dramatically improves the H₂ evolution rate from 110.8 to 402.4 μmol h⁻¹, which is 3.1 times higher than that of the P25-TiO₂ reference photocatalyst. The apparent quantum efficiency of R-TiO₂ for H₂ production was calculated to be 57.5 ± 0.2% at λ = 365 ± 15 nm, which is very high among recently reported results for pure TiO₂ photocatalysts. The influence of acid treatment on the charge carrier separation of the TiO₂ photocatalyst was investigated based on photocurrent measurements. Comparisons of photocurrent signals revealed similar tendencies in terms of the improvement based on acid treatment (Fig. 6(b)). The photocurrent density of R-TiO₂ is higher than that of the comparative samples, which suggests that the most efficient charge separation can be achieved under light irradiation. A smaller semicircle diameter at high frequencies of EIS confirmed reduced charge transfer resistance across the interface of the R-TiO₂ electrode and electrolyte, resulting in reduced charge recombination loss (Fig. 6(c)). As evidenced by the characterizations discussed above, the direct treatment of LPT precursor with concentrated nitric acid can induce a transformation into rutile TiO₂ nanorods and attach them to form discrete nanobundles with nitrate-group-grafted surfaces. Post-acid treatment of TiO₂ crystals cannot initiate such assembly or surface grafting. Therefore, the improved charge separation efficiency and photocatalytic performance of R-TiO₂ can be attributed to surface modification by nitrate groups.

In addition to activity, stability is also an important factor for evaluating the performance of photocatalysts. Fig. 7(a) presents the results of cyclic photocatalytic H₂ production for four consecutive cycles. One can see that the R-TiO₂ photocatalyst exhibits a tendency to gradually decrease in terms of efficiency after each cycle, indicating the instability of rutile TiO₂ nanobundles. To clarify the reason for this instability, changes in structure before and after the photocatalytic reactions were analyzed using XRD and FT-IR characterizations. The comparison in Fig. 7(b) confirms that the crystal phase is well-maintained during the H₂ production reactions. However, the signal of nitrate bending appears to weaken after cyclic testing (Fig. 7(c)), suggesting that proton reduction was accompanied by the consumption of nitrate groups on the R-TiO₂ surface. Because this loss in photocatalytic activity is related to the number of surface nitrate groups, the R-TiO₂ photocatalyst was refreshed after four cycles by repeating the concentrated nitric acid treatment process. As expected, the performance in the fifth cycle partially recovered to match the efficiency of the second cycle. Changes in the amount of adsorbed nitrate groups can be investigated using zeta potential measurements. Fig. 7(d) reveals that the a R-TiO₂ photocatalyst is negatively charged with a zeta potential value of −9.56 mV based on surface modification by nitrate anions. The surface charge gradually decreases after each cycle and nearly all charges are lost after the fourth cycle. It should be noted that the surface becomes negatively charged again after recovery of the used R-TiO₂ photocatalyst. It can be deduced that nitrate group grafting is crucial to photocatalytic H₂ production efficiency. As mentioned previously, in the literature, the enhancement mechanisms of inorganic acids are largely discussed from two perspectives, namely the promotion of hydroxyl group formation and interactions between oppositely charged holes, both of which contribute to oxygenic radical production for pollutant degradation [28-32]. Apparently, the influence mechanism of nitrate group grafting on photocatalytic H₂ production is completely different. Based on the above analysis, a reaction mechanism for this photocatalyst can be proposed (Scheme 1). Because holes are consumed by the sacrificial reagent as TEOA, the negative field effects of nitrate groups facilitate the trapping of positive protons to yield hydrogen gas. Additionally, changes in morphology before and after cyclic testing were studied based on FESEM observations. In contrast to the fresh catalyst in Fig. 2(a) and (b), the crystals are partially aggregated after the first cycle, as shown in Fig. 8(a). After four cycles, the catalyst largely lost the original spindle-like morphology and experiences significant agglomeration, as shown in Fig. 8(b). Fig. 8(c) verifies the recovery of the used R-TiO₂ photocatalyst (product after four cycles) by repeating the acid treatment process, although the original uniform morphology cannot be recovered. The fifth cycling test caused further aggregation of the refreshed crystals, as shown in Fig. 8(d). Morphology changes can also be explained by reduced surface charges. Based on the existence of zeta potential, the electrostatic repulsion resulting from the nitrate-group-grafted surface suppresses the aggregation of crystals. The consumption of nitrate during H₂ production gradually weakens the electrostatic repulsion and causes the structures of the rutile TiO₂ nanobundles to become unstable.

In summary, we developed a facile nitric-acid-induced assembly strategy for the synthesis of spindle-like rutile TiO₂ nanobundles. Structural characterizations demonstrated that the proposed strategy is effective for surface modification using nitrate groups. The R-TiO₂ sample exhibited the highest photocatalytic H₂ evolution rate of 402.4 μmol h⁻¹, which is 3.1 times higher than that of Degussa P25-TiO₂. In contrast, the anatase and rutile TiO₂ photocatalysts without sufficient nitrate group grafting only yielded H₂ evolution rates of 52.0 and 110.8 μmol h⁻¹, respectively. The enhanced photocatalytic efficiency of R-TiO₂ can be attributed to the negative field effects of nitrate groups, which favor the trapping of positive protons on the surface to improve charge carrier separation. Photocatalysts before and after photocatalytic reactions were compared from the perspectives of crystal and chemical structures, morphology, and surface charge properties. It can be deduced that surface charges are crucial for structural stability based on electrostatic repulsion. Overall, this work is of considerable significance for developing highly efficient photocatalysts for H₂ production based on different influence mechanisms of inorganic acids from those reported in the literatures.