1 Introduction

Nowadays, volatile organic compounds (VOCs) have been becoming the main sources of the indoor air pollution, and its volatility and toxicity harm the human health, and deteriorate ecosystem as well. Among those VOCs, benzene, toluene and formaldehyde are perhaps the most pernicious to human health [1,2,3,4]. Formaldehyde is a highly toxic and suspicious carcinogen at low concentration presented in our living atmosphere. A large number of techniques have been employed to eliminate formaldehyde, such as adsorption [5, 6], plasma oxidation [7, 8], photocatalysis, [9,10,11] and photothermal catalysis [12, 13]. Photocatalysts have found extensive applications in photocatalytic viral disinfection, reduction of CO2 and synthesis of organic compounds, effective contaminant removal [14,15,16]. Researches have found that photocatalysis is one of the most efficient and effective technique for photo-oxidation of formaldehyde, benzene, and toluene, with the advantages of green, pollution-free, mild conditions and reusability. Some commonly used semiconductor catalysts in photocatalytic oxidation of formaldehyde included TiO2, BiVO4, MnOx, graphitic C3N4, [13,14,15,16,17,18,19] and TiO2 has found widely application in VOCs degradation due to its stability, low cost and non-toxicity. However, its photocatalytic oxidation performance under visible light need to be improved further.

The light absorption performance of the catalyst has a great influence on the photocatalytic performance. Through the construction of heterojunction, dye sensitization and element doping, the optical absorption performance can be enhanced to improve the visible light response. As for the element doping, nonmetallic elements (such as F, Cl, Br, I, C and N) dopants can narrow the band gap of titanium dioxide [20,21,22,23,24,25,26,27,28,29]. Research found that fluorine doping makes the absorption edge of titanium dioxide red shift, enhances the visible light response, and makes titanium dioxide exhibiting higher activity in acetone photocatalytic oxidation under ultraviolet light, which means that the doping process is to tune the valence band and conduction band of titanium dioxide. However, those TiO2 photocatalysts doped with fluorine were prepared by sol–gel synthesis, followed by calcination at high temperatures (500–700 °C) [24,25,26,27,28,29]. On the other hand, those photocatalysts are usually in the form of particles. TiO2 nanorod arrays anchored on soft substrate were realized for flexible devices for energy harvesting [30]. Carbon cloth was used as soft substrate due to its high carrier transport performance and large specific surface area. And more importantly, nanorod arrays anchored on flexible substrate realized the immobilized integration catalysts for more conveniently gathering and recycling than powdery ones. Our recent study [31] realized the alkali metal (Na, K) and halogen (Cl) co-doped rutile TiO2 arrays at 550 °C for efficient photocatalytic degradation of formaldehyde. It is therefore important to explore a simple method to prepare the TiO2 arrays anchored on soft carbon cloth at low temperature for applicable utilization in air purifier.

Herein, soft carbon cloth was used as a substrate to anchor the TiO2 nanorods for applicable air purifier, overcoming the shortcoming of powder diffusion by air flow. Rutile TiO2 nanorod arrays were successfully grown on carbon cloth by a hydrothermal reaction of titanium butoxide and hydrochloric acid, and fluorine doped by solid diffusion of NH4F at 300 °C. Fluorine element doping improves the photoelectric properties, increases the generation of oxygen vacancies, and improves the formaldehyde degradation performance.

2 Experimental Details

2.1 Preparation of Rutile TiO2 Nanorod Arrays @ Carbon Cloth (CC)

Carbon cloth (CC) was used as soft substrate, immersed in a strong acid solution (H2SO4:HNO3 = 3:1 by volume ratio) for 24 h, cleaned with deionized water by ultrasonic stirring, and dried at 80 °C. Titanium(IV) chloride (99.5%) and titanium(IV) butoxide (Ti(OC4H9)4, 99%) were used as precursor for TiO2, and ammonium fluoride (99.99%) for fluorine. 220 μL TiCl4 was added into 10 mL 2.4 M hydrochloric acid, and then the dried CC was immersed in the above-mentioned solution for 12 h, then dried at 80 °C. The dried carbon cloth was heated at 400 °C for 30 min, forming TiO2 nanoparticles coated on CC as the seed layer. 600 μL titanium butoxide (TNBT) was added drop by drop into 24 mL 6 M hydrochloric acid solution, stir the solution with ultrasonic stirring until clear solution was obtained. Carbon cloth was immersed into the solution, and transferred into an autoclave heated at 160 °C for 5.5 h. When cooled to room temperature, carbon cloth was taken out, cleaned with deionized water and ethanol, and dried at 80 °C to obtain the rutile TiO2@CC.

TiO2@CC was immersed into 25 mL NH4F solution with certain concentration (NH4F: Ti(OC4H9)4 = 0, 2 wt%, 3 wt%, 4 wt%) for 24 h, dried, and heated at 300 °C for 2 h in air. The samples were cleaned with deionized water and ethanol, and dried at 80 °C, denoted as T@CC, 2F-T@CC, 3F-T@CC, 4F-T@CC, respectively. The F doped rutile TiO2 arrays on CC was schematically illustrated in Scheme 1.

Scheme 1
scheme 1

Schematic illustration of the F doped rutile TiO2 arrays on CC

Full size image

2.2 Characterization

The morphology and structure were examined by scanning electron microscopy (SEM, JSM-6610), transmission electron microscopy (TEM, JEM-2100), X-ray diffraction (XRD, Shimadzu 7000) and Raman spectroscopy (LabRAM HR Evolution). X-ray photoelectron spectroscopy (XPS, Thermo Fisher ESCALAB 250Xi), and ultraviolet electron spectroscopy (UPS, He I hν = 21.22 eV) were used to investigate the core level and the valence states of the elements. Photocurrent response and electrochemical measurements were carried out on AutoLab workstation (PGSTAT 302N) using a 500 W Xe lamp (λ ≥ 365 nm) as light source. The working electrode was made of 1 × 1 cm2 sample sandwiched in FTO conductive glass, and the electrolyte was 0.25 M NaSO4. UV–vis diffuse reflectance spectra (UV–vis DRS) were measured on Agilent Cary 5000 UV–vis NIR spectrometer with BaSO4 as reference. Photoluminescence spectra (PL) were collected on Hitachi F-7000 spectrophotometer at room temperature. Free radical trapping experiments were carried out on electron paramagnetic resonance (EPR) spectrometer (Bruke, EMXnano) with 365 nm LED as excitation source.

2.3 Photocatalytic Removal of Formaldehyde

Formaldehyde gas removal was evaluated by a volatile organic compounds (VOCs) purification evaluation system (Suncat POV-18) at room temperature. The formaldehyde concentration was continuously monitored by formaldehyde sensors (DART sensor Co., Ltd.) at the inlet and outlet of the reactor. The 2 cm × 2 cm sample was evenly cut into three equal parts and put into a high transparent quartz tube with diameter of 6 mm. The flow rate of formaldehyde gas is strictly set at 50 mL/min, and the adsorption desorption balance is achieved before the degradation reaction. The formaldehyde concentration at the inlet is set at 0.5 mg/m3. LED lamp (365 nm or 420 nm, 1.5 W) was used to irradiate the catalysts. The degradation rate of formaldehyde can be calculated according to the following equation:

$${\text{Degradation rate }}\eta {\text{ }} = {\text{ }}\left( {{\text{C}}_{0} - {\text{C}}_{{\text{t}}} } \right)/{\text{C}}_{0} \times {\text{1}}00\%$$
(1)

where C0 is the concentration of HCHO at the inlet (CH1) after adsorption–desorption equilibrium, and Ct is the concentration of HCHO at the outlet (CH2) at time of t after irradiation.

3 Results and Discussion

The XRD pattern of T@CC (Fig. 1a) can be indexed with rutile phase TiO2 (PDF no. 87–0710). After doping with fluorine, there appeared two weak diffraction peaks at 2θ of 38.1° and 48.2°, corresponding to the (004) and (200) crystal planes of anatase phase TiO2 (PDF no. 73-1765), and its intensity increases with the increase of NH4F amount during the annealing process. That shows that F doping promotes the growth of anatase TiO2 during annealing at 300 °C, and fractional anatase phase TiO2 may embed at the background of rutile phase. The Raman spectra (Fig. 1b) showed the characteristic peaks of rutile TiO2, in which Eg and A1g are located at 438.7 cm−1 and 610.4 cm−1, respectively. The broad band at 281.5 cm−1 is due to the multi-photon scattering of rutile TiO2, while a week peak at 185.5 cm−1 is attributed to anatase titania. After F doping, the peak of Eg shows slight blue shift and peak broadening, which is related to the formation of F–Ti–O. Some reports demonstrated that the Eg rutile mode showed red shift and attributed to nonstioichiometric effect and oxygen vacancies. However, the Eg mode in this study shows blue shift, and we attribute it to the formation of F–Ti–O bonds. The Eg mode was of the asymmetric bending of the O–Ti–O bonds in the {001} plane, caused by the opposite moving of the O atoms across the O–Ti–O bond [31,32,33]. The stronger F–Ti–O bonds suggested the Hook constant became larger, thus Eg mode shifted to higher wavenumber with F doping. Electron paramagnetic resonance (EPR) spectra (Fig. 1c) exhibit a resonance signal at g = 2.001, ascribed to the unpaired electrons trapped at the surface oxygen vacancies [31,32,33,34]. The signal intensity increased with the increase of fluorine doping, and 3FT@CC has the strongest signal. That strongly suggested that the fluorine doping induces oxygen vacancy in TiO2.

Fig. 1
figure 1

XRD patterns (a), Raman spectroscopy (b) and EPR spectra (c) of T@CC, 2F-T@CC, 3F-T@CC, and 4F-T@CC

Full size image

SEM images (Fig. 2a–d) showed that TiO2 nanorod arrays are densely covered the surface of the carbon fibers. The high magnitude SEM images (Figs. 2a’–d’) showed the randomly aligned TiO2 nanorods are about several micrometers in length and with square cross section with the side length of about 100 nm. The TiO2 nanorod arrays appeared as antler-like rectangular columns spaced several micrometers [30]. The morphologies of TiO2 nanorod arrays remain intact during the annealing process. Furthermore, the density of nanorods can be properly controlled by adjusting the concentration of the titanium (IV) butoxide in the hydrothermal process, providing more air contact area and reaction sites for formaldehyde degradation. The results showed that the nanoscale rutile TiO2 nanorod arrays are successfully grown on carbon fibers.

Fig. 2
figure 2

SEM images of T@CC (a), 2F-T@CC (b), 3F-T@CC (c), and 4F-T@CC (d)

Full size image

TEM (Fig. 3a) image shows that there are folds and holes in the TiO2 nanorods. The TiO2 nanorods are rectangular columns bunches composed of smaller nanowires. For TEM analysis, the specimens were prepared by dissolving the nanorods in ethanol solution by ultrasonic stirring. TEM image shows some thin nanorods, disbundled from the rectangular bunched TiO2. The selected area electron diffraction (SAED) pattern (Fig. 3b) confirms that the TiO2 nanorods are crystalline rutile phase of TiO2, consistent with XRD and Raman measurements. High resolution TEM (Fig. 3c) shows the lattice spacing is 0.325 nm, which corresponds to the rutile (101) crystal plane. The element mappings (Fig. 3d–g) show the homogeneous distribution of Ti, O and F on the TiO2 nanorod arrays on carbon fiber. EDS results gave a semiquantitative element ratio of Ti:O:F is about 41.4%:57.6%:1.0%. The results show that F element is doped into rutile TiO2 nanorods.

Fig. 3
figure 3

TEM image (a), SAED pattern (b) and HRTEM image (c) of 3F-T@CC. SEM image (d) and EDS element mapping of Ti (e), O (f), and F (g) of 3F-T@CC

Full size image

XPS Ti 2p spectra (Fig. 4a) for T@CC showed the doublet peak of 2p3/2 and 2p1/2 located at 457.5 and 463.2 eV, respectively. After fluorine doping, the Ti 2p peaks for 3F-T@CC shifted slightly to the lower binding energy. This can be attributed to the formation of oxygen vacancies by fluorine intercalation. XPS O 1 s spectra (Fig. 4b) can be fitted to Ti–O, O2 and H2O, locating at 531.4, 530.3 and 528.8 eV, respectively. After fluorine doping, the binding energy shifts to low binding energy by 0.3 eV. XPS F 1 s spectrum for 3F-T@CC (Fig. 4c) showed that F 1 s peak appeared at 683.3 eV, which is attributed to F− in F–Ti bond and confirmed that F was successfully doped in TiO2 nanorods. The Ti:O atomic ratio of T@CC is 33.4:66.6; while the Ti:O:F ratio of 3F-T@CC is 32.9%:66.0%:1.1%. The oxygen concentration reduces after fluorine doping. The presence of fluorine in the lattice of TiO2 induces the formation of reduced Ti3+ centers, and oxygen vacancies are likely to coexist with the Ti3+ ions.

Fig. 4
figure 4

Ti 2p (a), O 1s (b), and F 1s (c) core level XPS spectra of T@CC, and 3F-T@CC

Full size image

To evaluate the photocatalytic removal of formaldehyde, the formaldehyde concentration at the inlet (CH1) and outlet (CH2) of reactor was monitored in real time. Under 365 nm LED irradiation with intensity of 1.5 W, the formaldehyde degradation performance is shown in Fig. 5a. The inlet concentration (CH1) is set at 0.5 mg/m3, and the formaldehyde reached the adsorption desorption equilibrium before turning on the light. When light is turned on, the formaldehyde degradation rate increases rapidly to a saturated value, then decreases slowly with the irradiation time. The evaluation of degradation performance should include the saturation value, duration at saturation. All catalysts can reach the saturation of 100% degradation of formaldehyde under 365 nm LED irradiation. The duration of 100% degradation of formaldehyde was 30 min, 30 min, 240 min, 138 min, for T@CC, 2F-C@CC, 3F-T@CC, and 4F-T@CC, respectively. At the radiation time of 240 min, the degradation rate was 74%, 82%, 100%, 96%, for T@CC, 2F-T@CC, 3F-T@CC, 4F-T@CC, respectively. Figure 5b shows stability test for 3F-T@CC. The duration of 100% formaldehyde degradation was up to 58.2 h, which is 291 times of that of T@CC.

Fig. 5
figure 5

HCHO degradation rate (a) of T@CC, 2F-T@CC, 3F-T@CC, and 4F-T@CC

Full size image

The formaldehyde degradation performance under 420 nm LED irradiation with light intensity of 1.5 W was shown in Fig. 6a. When the light was turn on, the formaldehyde degradation rate for T@CC reaches the maximum value of 52% for duration of 100 min, then decreases gradually to 48% at time of 240 min; 2F-T@CC reaches the maximum degradation rate of 68%, then decreases gradually to 46% at time of 240 min; 3F-T@CC reaches the maximum degradation rate of 96% at time of 188.6 min and then decreases slightly to 94% at 240 min; 4F-T@CC reaches the maximum degradation rate of 84% at 14.3 min and then decreases to 76% at 240 min. The results showed that F doping increased the degradation rate for rutile TiO2, and 3F-T@CC exhibits the best performance for formaldehyde degradation among all catalysts. Figure 6b showed the stability test of 3F-T@CC, and the formaldehyde degradation rate reached 100% after 1.7 h irradiation and lasted for 12.1 h. The photocatalytic degradation measurements demonstrated that 3F-T@CC exhibits the best performance under 365 nm and 420 nm LED irradiation.

Fig. 6
figure 6

HCHO degradation rate (a) of T@CC, 2F-T@CC, 3F-T@CC, and 4F-T@CC

Full size image

Electron impendence spectra (EIS) (Fig. 7a) showed the corresponding interface reaction impedances (RCT) are 111.0, 91.9, 79.1 and 97.6 Ω, for T@CC, 2F-T@CC, 3F-T@CC, 4F-T@CC, respectively, indicating that the doping of F element can reduce the interface reaction impedance and facilitate the transmission of photogenerated carriers. The interface reaction impedance for 3FT@CC is the smallest. Photoluminescence spectra (Fig. 7b) showed 3F-T@CC has the lowest peak intensity, which indicates that photogenerated electron hole pairs are more effective than other samples. PL spectra of TiO2 are attributed to three kinds of physical origins: self-trapped excitons, oxygen vacancies (OVs), and surface states. The peak at 390.3 nm is attributed to the oxygen vacancies with two trapped electrons, F center, and peak at 413.8 nm is assigned to the oxygen vacancies with one trapped electrons, F+ center. Peak at 443.2 nm might be the lattice ions surrounding the vacancy. Peak at 466.2 nm may be caused by the defects on the surface of TiO2 nanorods [35,36,37,38].

Fig. 7
figure 7

Electron impendence spectra (a), and PL spectra (b) of T@CC, 2F-T@CC, 3F-T@CC, and 4F-T@CC

Full size image

The UV–vis diffuse reflectance spectra (Fig. 8a) showed that F doping leads to an increase in the adsorption over ultraviolet–visible range with wavelength up to 550 nm, and the adsorption edge exhibits slight red shift. From the Tauc plot in Fig. 8b, the band gap of 3F-T@CC and 4F-T@CC is determined to be 3.01 eV, which is almost identical to that of T@CC. Apparent tail between 2.95 eV and 3.00 eV in the Tauc plot favors the light absorbance. The visible light response mainly originated from the presence of oxygen vacancies created by the fluorine doping. The valence band energy was determined from ultraviolet photoelectron spectroscopy (UPS) by subtracting the width of He I UPS spectra (Fig. 8c) from the excitation energy of 21.22 eV. The valence band energy is 7.49 eV, 7.22 eV, 6.90 eV, and 7.05 eV, for T@CC, 2F-T@CC, 3F-T@CC, and 4F-T@CC, respectively. The valence band in electron volts is converted to electrochemical potential in volts according to the reference standard for which 0 V versus reversible hydrogen electrode (RHE) equals − 4.5 eV versus vacuum level. The valence band related to RHE is 3.0, 2.7, 2.4 and 2.6 V, for T@CC, 2F-T@CC, 3F-T@CC, 4F-T@CC, respectively. Figure 8d shows the energy band structure diagram of T@CC and F doped TiO2 system. The up-shifting the band edges should play a crucial role in the photocatalytic process. The more negative of conduction band make the photogenerated electrons to react with oxygen to generate superoxide ions (O2 + \({e}_{CB}^{-}\)  → \(\cdot {\text{O}}_{2}^{ - }\)). Holes in valence band react with surface adsorbed water to generate hydroxyl radical (H2O + \({h}_{VB}^{+}\)  → ·OH + H+) [39].

Fig. 8
figure 8

UV–vis diffuse reflectance spectra (a) and converted Kubelka–Munk plots (b), UPS spectra (c) and schematic band structure (d) of T@CC, 2F-T@CC, 3F-T@CC, and 4F-T@CC

Full size image

DMPO (5,5-dimethyl-1-pyrrole N-oxide) was used as a free radical trapping agent in EPR analysis to detect free radical intermediates (·OH and \(\cdot {\text{O}}_{2}^{ - }\)) generated under specific potential. Upon 365 nm LED irradiation, the 3F-T@CC in the methanol dispersion system show obvious 1:1:1:1 peaks corresponding to \(\cdot {\text{O}}_{2}^{ - }\), and in the water dispersion system 1:2:2:1 peaks corresponding to ·OH, as shown in Fig. 9. The experimental results proved that the photoinduced electron–hole pairs react with adsorbed H2O and O2 to form ·OH and \(\cdot {\text{O}}_{2}^{ - }\) [39,40,41]. The catalyst 3F-T@CC exhibits stronger signal for ·OH and \(\cdot {\text{O}}_{2}^{ - }\) than T@CC, which confirms that fluorine doping promotes the generation of active species for photocatalytic oxidation of formaldehyde. These two kinds of free radicals have strong oxidation properties and involved in oxidize formaldehyde to CO2 and H2O [17].

Fig. 9
figure 9

EPR spectra in a methanol dispersion for DMPO-\(\cdot {\text{O}}_{2}^{ - }\) (a), and in an aqueous dispersion for DMPO-·OH (b) of 3F-T@CC

Full size image

4 Conclusions

In conclusion, rutile TiO2 nanorod arrays were successfully grown on carbon fibers by seed assisted hydrothermal method, and the carbon fibers facilitated the transfer of photoinduced charges, while the antler-like arrays increase the active sites. The doping of F element induces oxygen vacancies, enhancing light response and tuning energy band structure. 3F-T@CC catalyst has the best photocatalytic degradation performance of formaldehyde. Under 365 nm LED irradiation, the 3F-T@CC catalyst can eliminate formaldehyde completely with duration of 58.2 h, which is 291 times of that of pure TiO2 (0.2 h). Under 420 nm LED, the degradation rate of 3F-T@CC reached to 96% at 240 min, which is about as two times as that of pure T@CC (48%), and the duration of complete degradation reached 12.1 h. Photoinduced electron–hole pairs react with H2O and O2 to form ·OH and \(\cdot {\text{O}}_{2}^{ - }\), which oxidize formaldehyde to CO2 and H2O. This study provide a pathway to fabricate TiO2 arrays attached on carbon cloth at low temperature, ready for the applicable air purifier to remove formaldehyde.