Introduction

A rapid industrial development driven by unsustainable technology advances can cause plenty of industrial sewage, spreading chemical hazards into water resources. As a result, water pollution has emerged as one of the most serious environmental issues worldwide1,2,3,4. Photocatalytic oxidation technology has shown great prospects in removing the toxic and harmful contaminants in aqueous environment5,6,7. Semiconductors (e.g. TiO2, ZnO, SnO2) have been widely researched for organic pollutant degradation, however, the large band gap hinders their practical applications8,9,10,11,12. For example, TiO2 with band gap of 3.2 eV can only absorb the ultra-violet light, accounting for only 4–5% of entire solar spectrum13. Therefore, various visible light sensitive photocatalysts has also been widely explored, such as g-C3N4, BiVO4, CdSe, Bi2WO614,15,16,17,18,19. On the other hand, TiO2 is recognized as one of the excellent materials owning to its good inertness, eco-friendly, low cost, strong oxidizing power, and long-term stability against photo and chemical corrosion9,13,20,21,22. Thus, plenty of works have been made to extend the absorption spectrum of TiO2 to visible light so to make a full use of solar spectrum. Several different approaches can be employed, including doping23,24,25,26 and coupling with small band gap semiconductors or metals27,28,29,30.

Small band gap semiconductors not only increase the absorption of visible light but also inhibit photo-generated electrons-holes recombination when constructed as a semiconductor/semiconductor heterojunction structure, thus improving the photocatalytic performance dramatically31. Therefore, various TiO2 based heterojunction photocatalysts have been proposed for visible light photocatalysis, including NiO/TiO232,33, Ag2O/TiO234, CdTe/TiO235, C3N4/TiO236, Bi2O3/TiO237, Cu2O/TiO238, Fe2O3/TiO239, etc. For Fe2O3/TiO2 heterojunction photocatalysts, a variety of composites have been investigated, such as Fe2O3 nanoparticles on TiO2 nanotube40, Fe2O3/TiO2 nanoparticles41, TiO2 coated cubic Fe2O342, Fe2O3 nanosheet/TiO2 hollow sphere39, and Fe2O3 coated TiO243. For instance, Lin et al. demonstrated that Fe2O3 coating can effectively enhance the visible light photocatalytic activity of TiO243. Various fabrication methods were applied, including hydrothermal or solvothermal process and sol–gel, to prepare the heterojunction photocatalysts44,45,46. Nevertheless, precise control of the interface between Fe2O3 and TiO2 at atomic level by conventional methods remain challenges.

Atomic layer deposition (ALD) is a unique and promising thin film deposition technique based on self limited and saturated surface chemisorption reactions. It can deposit ultrathin, conformal, and uniform layers at sub-nanometer scale, which has attracted great attentions in surface engineering of nanostructures over the years47,48,49. In catalysts design, ALD enables a conformal layer with precise thickness control and tunable film composition onto another nanostructures with high aspect ratio50. The ALD coating can work as photo-active materials51,52 or surface protection layer53,54. Herein, we modified the commercial anatase TiO2 powders with ultrathin Fe2O3 surface coating by ALD. The photocatalytic performance was investigated by visible light degradation of methyl orange (MO). The ultrathin Fe2O3 coating can enhance the absorption of TiO2 supports for visible light. Fe2O3 modified TiO2 powders show much better visible light photocatalytic degradation of MO than pristine TiO2. A possible mechanism for improved photocatalytic performance is proposed. In addition, an ultrathin ALD Al2O3 (~ 2 nm) was used to promote the long-term durability of TiO2@Fe2O3 catalyst.

Methods

ALD deposition on TiO2 powders

Commercial TiO2 powders with anatase phase (Nanjing Haitai nano materials Co.) were used as supports in this work. Ferrocene (Fe(Cp)2, Suzhou Fornano Corporation Ltd., 99.99%) and ozone were adopted as Fe and oxygen precursors for ALD Fe2O3 deposition. Fe(Cp)2 was vaporized at 85 °C. High purity nitrogen gas (N2, 99.999%) was used as carrier gas at a total flow rate of 750 sccm and a pressure of 6 hPa in our ALD system (Picosun SUNALE™ R-150B). A particular container with porous mesh was used for ALD coating on powders, which has been reported elsewhere24,55, as shown in Fig. 1. Herein, precursors can flow through the TiO2 powders to achieve great conformality. X cycles of ALD Fe2O3 (X = 200, 400, 600, and 800) were coated on TiO2 powder at 300 °C, the samples are marked as TiO2@X-Fe2O3. One cycle of ALD Fe2O3 contains the following four steps, 5 s Fe(Cp)2 injection, 20 s N2 purge, 5 s O3 injection, and 20 s N2 purge. At the same system, 20 cycles of Al2O3 were deposited on TiO2@400-Fe2O3 at 300 °C, where one ALD cycle of Al2O3 is consisted of 5 s trimethylaluminum dose, 20 s N2 purging, 5 s H2O dose, and 20 s N2 purging.

Figure 1
figure 1

The schematic diagram of coating TiO2 powders by ALD Fe2O3.

Full size image

Materials characterizations

X-ray diffraction (XRD) using a Rigaku-D/MAX 2000 system was used for crystallinity and phase structure analysis. Scanning electron microscopy (SEM) images were taken using ZEISS Gemini SEM 500 instrument operated at 2 kV. The high-resolution transmission electron microscopy (HRTEM) was performed on a FEI Tecnai F20 S-Twin to observe the microstructures, where TiO2 powders were loaded on the ultra-thin carbon coated copper grids. The surface chemical features and valence band spectra were explored by X-ray photoelectron spectroscopy (XPS) using Thermo Fisher K-Alpha. The adventitious carbon signal (C 1 s = 284.6 eV) was adopted to calibrate the binding energies. UV–visible absorption spectra were conducted on a UV–vis-NIR spectrophotometer (UV-3600, Shimadzu, Japan). Photoluminescence (PL) spectra were collected on a Horiba Jobin Yvon HR800 spectrometer.

Photocatalytic degradation

The photocatalytic performance of Fe2O3 coated TiO2 catalysts was investigated by visible light degradation of MO. 100 ml MO solution (4 mg L−1) with 100 mg photocatalysts were loaded into a glass reactor, which was magnetically stirred at 500 rpm. In order to establish the adsorption/desorption equilibrium between MO and catalysts before irradiation, the MO solutions with catalysts were magnetically stirred for 30 min in darkness. Then, the suspension was irradiated under a 300 W Xe lamp (MircoSolar300, Perfect Light). A 420 nm filter was adopted to cut off UV light. The lamp was placed at 15 cm above the suspension, whose average visible light intensity is around 80 mW cm−2. Water cooling was applied throughout the experiment to maintain the temperature at 25 ± 1 °C. 3 mL solution was collected after each 15 min irradiation. The photocatalysts were removed by centrifugal separation. The residual MO concentration was determined using the absorption at 464 nm by UV–Vis–NIR spectrophotometer. The recycled usage experiment was performed for three times to explore the long-term stability of photocatalysts, e.g. after each photocatalytic experiment, the photocatalysts powders were gathered and rinsed by ethanol and water, then baked for 12 h at 100 °C. At last, a new MO solution was used to evaluate the photocatalytic activity of collected photocatalysts.

Photoelectrochemical measurements

Photoelectrochemical measurements were performed in a three-electrode electrochemical cell at room temperature using 1 M Na2SO4 as the electrolyte. The TiO2 or TiO2@400-Fe2O3 on FTO were used as the working electrode. A Pt wire and Ag/AgCl were used as the counter electrode and the reference electrode, respectively. Photocurrent densities were collected by an electrochemistry workstation (CHI660E, Shanghai) using a potentiostatic method at 0.50 V. Light was chopped on and off cyclically. A solar simulator (300 W Xe lamp, MircoSolar300, Perfect Light) with a 420 nm cut-off filter provides the visible-light irradiation.

Results

Figure 2 depicts the XRD patterns of pristine TiO2 and Fe2O3 coated TiO2 powders. All the samples show the similar characteristic diffraction peaks, in accord with anatase TiO2 (JCPDS No. 21–1,272). This result indicates that ultrathin ALD Fe2O3 would not affect the crystal structure of TiO2, consistent with our previous finding24,55. In addition, signals related to Fe2O3 were absent.

Figure 2
figure 2

XRD patterns of pristine TiO2 and Fe2O3 coated TiO2.

Full size image

SEM images of TiO2 powders without and with 400 cycles ALD Fe2O3 deposition are shown in Fig. 3a,b. It can be seen that the pristine TiO2 powder exhibits a disk-like morphology with a diameter of approximately 40 nm and a thickness of approximately 10 nm, with severe aggregation. After ALD Fe2O3 deposition, it was observed that Fe2O3 coated TiO2 exhibited almost identical morphology, indicating that ultra-thin Fe2O3 coating did not have significant influence on particle size and morphology of TiO2. HRTEM was further conducted to observe the microstructure of TiO2@400-Fe2O3 and pristine TiO2 powders as shown in Fig. 3c,d. While pristine TiO2 shows a sharp well-ordered surface with good crystallinity (Fig. 3c), an amorphous layer of ~ 2.6 nm formed on the TiO2 surface was observed which was contributed to ultra-thin Fe2O3 layer (400 cycles) formed by ALD deposition. Moreover, both samples exhibit the lattice spacing of 0.35 nm, corresponding to (101) planes of anatase TiO2. Based on above XRD and TEM data, it was speculated that an ultra-thin amorphous Fe2O3 is coated on TiO2 nanoparticles surface without significantly modifying the morphology of the catalyst supports.

Figure 3
figure 3

SEM images of (a) pristine TiO2 and (b) TiO2@400-Fe2O3. TEM images of (c) pristine TiO2 and (d) TiO2@400-Fe2O3.

Full size image

To further determine the successful deposition of ALD Fe2O3, XPS was performed to characterize the surface chemical features of ALD Fe2O3 coated TiO2 powders. Figure 4a shows the Ti 2p spectra of TiO2@400-Fe2O3 and pristine TiO2 powders. The spectra can be fitted into two peaks at 464.3 eV and 458.5 eV, which can be assigned to Ti 2p1/2 and Ti 2p3/2 peaks28. In O 1 s spectra (Fig. 4b), both samples present the main peak at 529.9 eV related to Ti–O bonding from TiO2. In addition, there is a peak at 531.7 eV can be ascribed to the surface -OH56. Figure 4c exhibits the Fe 2p spectrum of TiO2@400-Fe2O3, Fe 2p1/2 and Fe 2p3/2 peaks locate at 723.8 eV and 710.5 eV, in accord with Fe–O bonding value in Fe2O357. Based on XPS data, the Fe atom ratio (Fe/Fe + Ti) is determined to be 1.1% in TiO2@400-Fe2O3. It is anticipated that an ultra-thin amorphous Fe2O3 is coated on TiO2 nanoparticles surface successfully. Due to the low content of Fe2O3, Fe 2p spectrum shows a bad signal to noise ratio. Therefore, the Fe 2p spectrum of ALD Fe2O3 film is also presented for reference, as shown in Fig. 4d which can exhibit much better signal to noise ratio.

Figure 4
figure 4

XPS spectra of (a) Ti 2p and (b) O 1s for TiO2 and TiO2@400-Fe2O3, Fe 2p XPS spectra for (c) TiO2@400-Fe2O3 and (d) Fe2O3 film.

Full size image

The influence of ultra-thin Fe2O3 coating on absorption of TiO2 powders in visible light was explored using UV–Vis diffuse reflectance spectra, as illustrated in Fig. 5a. The spectrum of pristine TiO2 powders is also plotted for comparison. Pristine TiO2 powders exhibit the absorption edge of around 371 nm without noticeable visible light absorption. However, noticeable absorption in the visible light region from 390 to 750 nm can be observed after ultra-thin Fe2O3 surface modification. The relationship of the absorption edge with the photon energy (hν) for the indirect bandgap semiconductor is shown in the following formula: (αhν)1/2 = A(hν-Eg), where α and A are the absorption coefficient and absorption constant, respectively. Since the absorption coefficient α is determined by the scattering and reflectance spectra based on Kubelka–Munk theory, therefore, the bandgap values can be determined by the intercept of the tangent lines. As depicted from Fig. 5b, only one tangent line can be extrapolated for pristine TiO2 powders showing a bandgap of 3.25 eV, while two bandgap values can be obtained from the plots for TiO2@400-Fe2O3 powders, attributing to Fe2O3 coating with a bandgap value of ~ 2.20 eV and TiO2 supports with a bandgap value of ~ 3.08 eV. It can be concluded that ultra-thin Fe2O3 coating results in a smaller bandgap which can increase the absorption of TiO2 powder support in visible light.

Figure 5
figure 5

(a) UV–Vis diffuse reflectance spectra and (b) Tauc plot of TiO2 with and without 400 cycles Fe2O3 coating.

Full size image

The visible light photocatalytic activity of TiO2 and Fe2O3 coated TiO2 catalysts was compared by degrading MO. All samples exhibit low adsorption capacity of MO molecules. As reported in our previous work, MO is selected here for its stability under visible light irradiation in the absence of catalysts24. Figure 6a shows the evolution of UV–vis absorption spectra of MO solution in the presence of pristine TiO2 under visible light irradiation. It can been seen that the absorption peaks at 464 nm decreases slightly after 90 min, exhibiting very poor photocatalytic activity due to its large bandgap. In contrast, the peak intensity at 464 nm fades rapidly for TiO2@400-Fe2O3 with the irradiation time extending, as shown in Fig. 6b. In addition, the orange MO solution turns into colorless after 90 min, as presented in the insert of Fig. 6b, indicating the degradation of MO. Figure 6c compares visible light photocatalytic activity of Fe2O3 coated TiO2 catalysts. It can be found that a much-improved photocatalytic degradation efficiency of ~ 86.2% can be achieved with only 200 cycles of ALD Fe2O3 modification. And the TiO2@400-Fe2O3 photocatalysts display the highest photocatalytic degradation efficiency of 97.4%. In comparison with reported Fe2O3-TiO2 heterojunction catalysts for photodegradation of organic pollutants and antibiotics, ALD Fe2O3 coated TiO2 (TiO2@400-Fe2O3) in this work exhibit excellent removal efficiency for MO degradation, as summarized in Table 1. The photocatalytic degradation efficiency decreases to 95.8% and 90.4% for TiO2@600-Fe2O3 and TiO2@800-Fe2O3 samples, respectively, along with further increasing the thickness of ALD Fe2O3 coating. The reduced efficiency can be ascribed to the fact that more Fe2O3 coating would introduce more recombination sites for photoinduced electron–hole pairs42, diminishing the photocatalytic efficiency.

Figure 6
figure 6

UV–vis absorption spectra of MO exposed to different irradiation time in the presence of (a) pristine TiO2 and (b) TiO2@400-Fe2O3 catalysts under visible light irradiation. The inserts are the photos of MO solution before and after irradiation. (c) Visible light photocatalytic degradation curves of MO and (d) −In(C/C0) vs. time curves by using TiO2 with and without Fe2O3 coating as catalysts.

Full size image
Table 1 Comparison of photocatalytic activity for degradation of organic pollutants using TiO2–Fe2O3 based catalysts.
Full size table

The degradation data were also fitted by the pseudo-first-order kinetics. The rate constant k can be determined by In(Ct/C0) = −kt at low initial pollutant concentration. Herein, C0 is the initial MO concentration, while the Ct is the MO concentration after irradiation time of t. k is the first-order rate constant (min−1). The −In(Ct/C0) vs. t curves are plotted in Fig. 6d. It can be seen that −In(Ct/C0) has a linear relationship with t, indicating the photocatalytic degradation of MO by Fe2O3 modified TiO2 catalysts obeys the first-order kinetics. The first-order rate constant (k) is determined to be 3.8 × 10−2 min−1 for TiO2@400-Fe2O3, which is much larger than pristine TiO2 of 1.3 × 10–3 min−1. The result indicates that ALD Fe2O3 modification can effectively enhance the visible light photocatalytic performance of TiO2 supports.

PL and photocurrent response measurements were conducted to explore the recombination rate of photo-generated electron–hole pairs. The lower the PL intensity, the higher separation efficiency of electron–hole pairs in the catalysts. Figure 7a shows the PL spectra of pristine TiO2 and various cycles of ALD Fe2O3 coated TiO2. It can be seen that all the Fe2O3 coated TiO2 exhibit lower intensity than pristine TiO2, indicating that the coupling of TiO2 and Fe2O3 can effectively inhibit the recombination of the photo-generated electron–hole pairs. Moreover, it can be seen that the 400 cycles Fe2O3 coating results in the lowest intensity, thicker Fe2O3 coating would increase the recombination of the photo-generated electron–hole pairs. The PL data are consistent well with the results of photocatalytic degradation of MO. Figure 7b presents the photocurrent response curves of TiO2 and TiO2@400-Fe2O3, it can be seen that TiO2@400-Fe2O3 exhibits a much higher photocurrent density than pristine TiO2, indicating a more efficient separation of the photoexcited electron–hole pairs.

Figure 7
figure 7

(a) PL spectra and (b) photocurrent response curves of TiO2 and Fe2O3 coated TiO2.

Full size image

The band alignment for Fe2O3 coated TiO2 was determined by measuring the valence band offset ∆Ev (VBO) using XPS. Figure 8a shows VB spectra of TiO2 and Fe2O3–TiO2 determined by linear extrapolation method, respectively. The VB of TiO2 and Fe2O3–TiO2 are found to be 3.09 eV and 0.84 eV, respectively. The VB of Fe2O3 is higher than that of TiO2, and ∆Ev at the interface of Fe2O3-TiO2 is estimated to be 2.25 eV. The optical bandgaps of TiO2 and Fe2O3 have been determined to be 3.25 and 2.20 eV, respectively, in Fig. 5. Therefore, the conduction band offset ∆Ec (CBO) at the interface of Fe2O3-TiO2 is estimated to be 1.20 eV. Considering the band structure of TiO2 vs. standard hydrogen electrode (NHE)62,63,64, the energy band structure of Fe2O3 coated TiO2 can be depicted in Fig. 8b. Under visible light, TiO2 shows no photo-electronic response due to its large band gap. Only Fe2O3 can be excited, yielding photo-generated electron from its VB to CB. Due to the aligned equilibrium of Fermi level at the interface of TiO2 and Fe2O3, as shown in Fig. 8b, the photogenerated electrons can transfer from CB of Fe2O3 to that of TiO2 driven by the built-in electric field and the concentration gradient of electrons, while remaining the holes in VB of Fe2O360,62,63,64. Therefore, the separation efficiency of photoinduced electron–hole pairs can be improved, which has been demonstrated by PL and photocurrent response results in Fig. 7. There are a large number of literatures focusing on the photocatalytic activity of Fe2O3-TiO2 composites catalyst60,62,63,64. It is widely accepted that OH. radicals can be formed via the reaction of water and photogenerated holes in VB of Fe2O3. And electrons in CB of TiO2 can react with oxygen to form O2-. These radicals with high activities can degrade organic molecules into harmless substances.

Figure 8
figure 8

(a) Valence band spectra of TiO2 and Fe2O3–TiO2. (b) Schematic illustration of energy band structure of Fe2O3 coated TiO2 and proposed charge transfer mechanism during visible light irradiation.

Full size image

The stability of photocatalysts is one of the significant factors for practical applications, therefore, TiO2@400-Fe2O3 and TiO2@800-Fe2O3 catalysts were tested in recycling experiments of MO photodegradation. As shown in Fig. 9a, compared to the first usage, both TiO2@400-Fe2O3 and TiO2@800-Fe2O3 catalysts exhibit a declining degradation efficiency. The degradation efficiency retention compared to the first usage is only 59.6% and 66.7% for TiO2@400-Fe2O3 and TiO2@800-Fe2O3, respectively. It may be ascribed to the fact that stability of Fe2O3 during photocatalytic reactions is reduced due to photo-corrosion42,65. ALD coatings have been widely used as surface protection layer to protect active materials from photo-corrosion53,54. Therefore, an ultrathin Al2O3 (~ 2 nm) protective layer was deposited on catalysts surface by ALD in the same system to improve the stability of Fe2O3 coated TiO2 catalysts. Figure 9b shows the photocatalytic degradation of MO using both TiO2@400-Fe2O3 with Al2O3 passivation. The photocatalytic degradation efficiency decreases to 47.8% in 90 min, which is ascribed to the fact that Al2O3 coating with large band gap would also hinder the photogenerated carriers form migrating to the surface of electrode62,66. But it is still much better than that of pristine TiO2. More importantly, the stability of catalysts is persistent, as there is limited decline of degradation efficiency after three usage as shown in Fig. 8a, retaining 86.8% compared to the first usage. The results indicate that thin Al2O3 can act as a physical shell to protect Fe2O3 from photo-corrosion. This can be ascribed to that Al2O3 can prevent the direct contact between solution and Fe2O3.

Figure 9
figure 9

(a) Three cycles of MO degradation for TiO2@400-Fe2O3, TiO2@800-Fe2O3 and 20 cycles Al2O3 coated TiO2@400-Fe2O3 in 90 min. (b) Comparison of MO degradation curves for pristine TiO2, TiO2@400-Fe2O3, and 20 cycles Al2O3 coated TiO2@400-Fe2O3.

Full size image

Conclusions

In summary, commercial anatase TiO2 powders were modified using ultrathin ALD Fe2O3 surface coating. The ultrathin Fe2O3 layer with small bandgap of ~ 2.20 V can increase the absorption of TiO2 supports in visible light. In addition, Fe2O3/TiO2 heterojunction can suppress the photoinduced electron–hole pairs recombination. The above results indicate excellent visible light photocatalytic activity for the Fe2O3 modified TiO2 powders. 400 cycles Fe2O3 (~ 2.6 nm) coated TiO2 photocatalysts show excellent degradation efficiency of 97.4% in 90 min, far above the performance of pristine TiO2 powders with only 12.5%. Moreover, an ultrathin Al2O3 (~ 2 nm) can improve the recycling usage performance of Fe2O3 coated TiO2 catalyst effectively. As conclusions, ALD surface modification with ultrathin film is a promising route for improving the visible light activity and long-term stability of photocatalysts.