Introduction

TiO2, one of the most promising photocatalysts, has attracted considerable attention in recent years due to its high oxidative power, abundance, chemical stability, environmentally friendly and low cost1,2,3,4,5. Unfortunately, the wide band gap of TiO2 ca. 3.2 eV has limited its widespread application in industry since it needs to be excited by ultraviolet (UV) light6,7,8. It is well known that the UV light only occupies 5% of solar light, whereas the residual 45% and 50% of solar light are comprised by visible light and near-infrared (NIR) light, respectively. Therefore, there is still a large space for TiO2 to be improved for photocatalysis using solar light9,10.

In order to overcome this issue, several strategies, such as dye sensitizing11,12, heterostructure13,14, ion doping15,16,17, etc. have been developed. For example, Geneviève et al. reported that the visible solar light absorption of titanium dioxide could be significantly enhanced by combining titanium dioxide with Osmium and Ruthenium prolypyridyl complexes18. Zhou et al. presented that the visible light absorption of TiO2 was able to be extended to 650 nm by forming a heterostructure with Ag2O nanoparticles19. Yin et al. demonstrated that different phases of nitrogen doped titania nanocrystals were successfully prepared by a homogeneous precipitation-solvothermal process in TiCl3-hexamethylenetetramine aqueous and alcohol solutions and the as-prepared samples showed excellent visible light absorption due to the nitrogen doping in titania lattice20. However, most of them can only extend the solar light absorption of TiO2 to the visible-light range. As for NIR light absorption, few papers reported about this. Qin group firstly demonstrated NIR-driven photocatalysis of TiO2 by combining it with YF3:Yb, Tm or NaYF4:Yb,Tm9,10. The mechanism for this NIR-driven photocatalysis is as follows. The up-conversion phosphor absorbed the NIR light and then emitted UV light. Finally, this converted UV light would be reabsorbed by TiO2 for photocatalysis. Nevertheless, in this case, only UV and NIR lights could be utilized by these two composites. Furthermore, the light conversion from NIR light to UV light is usually not efficient enough compared with the transformation from NIR light to visible light. Because the light energy conversion would be more efficient when the fewer anti-stokes shift was happened in this process21,22,23. Therefore, the combination of weak visible light induced TiO2 based photocatalyst with NIR-Vis up-conversion phosphor obtaining the lower anti-stokes shift may be the more promising way to effectively use the whole solar light (UV, visible, NIR lights) for photocatalysis.

It is well known that NaYF4 has been widely recognized as the most efficient host for up-conversion phosphor until now due to its low phonon energy and efficient multicolor emissions. Subsequently, the Yb and Er co-doped NaYF4 designated as (Yb,Er)-NaYF4 has become the most promising visible light emitting up-conversion phosphor with two main emission peaks at about 545 and 655 nm under the irradiation of 980 nm24,25,26. In order to use (Yb,Er)-NaYF4 as NIR-Vis conversion matrix achieving the optimal synergetic effect between up-conversion phosphor and photocatalyst, a suitable TiO2 based photocatalyst, which has a good visible absorption covered the 545 and 655 nm range sufficiently, together with excellent photocatalytic activity under very weak light irradiation, is required.

Recently, C has become one of the most studied dopants in TiO227,28 since Sakthivel et al. reported that carbon doped particles presented five times higher activity than nitrogen-doped TiO2 in the degradation of 4-chlorophenol by visible light29. According to the density functional theory calculations within the generalized gradient approximation reported by Valentin et al30. C doped TiO2 presented excellent visible light absorption and could cover the majority of visible solar lights by introducing series of localized occupied states above the valence band gap of TiO2 semiconductor and it has been experimentally confirmed by many researchers27,31. Our groups32 also prepared the C doped TiO2 by a facile solvothermal method, which exhibited fabulous visible light absorption in the range of 380–780 nm and corresponding nice visible light responsive NOx destruction ability superior to that of N-TiO2.

Therefore, in order to improve solar energy utilization by enabling the effective use of not only UV, visible lights but also NIR light for photocatalysis, a novel composite, which consisted of NIR light induced (Yb,Er)-NaYF4 up-conversion phosphor and UV, visible lights driven C-TiO2 photocatalyst, was synthesized by a facile calcination assisted solvothermal method without the addition of any other carbon precursors except for Ti source and ethanol solvent. In addition, the NOx gas destruction activity was employed to test the photocatalytic ability of composite instead of dye solution degradation due to the influence of solution on the emission intensity of up-conversion phosphor and also the week power of LEDs and infrared laser were used as irradiation sources to investigate the high efficiency of photocatalytic performance of the as-prepared samples more efficiently. More importantly, the photocatalytic stability and the effect of (Yb,Er)-NaYF4 up-conversion phosphor on the total photocatalytic activity of (Yb,Er)-NaYF4/C-TiO2 composite under the irradiation of UV, visible, NIR lights, comparing with that of pure C-TiO2 photocatalyst, should be the key points in the consideration of effective utilization of solar light for the enhanced photocatalysis.

Results

Characterization of samples

Figure 1 shows the XRD patterns of samples of C-TiO2, (1:1)@(Yb,Er)-NaYF4/C-TiO2 and (Yb,Er)-NaYF4. All the diffraction peaks of sample C-TiO2 were well indexed to the anatase phase of TiO2 (JCPDS file No. 21-1272) and no impurity peaks appeared. After combining with (Yb,Er)-NaYF4 phosphor by a facile solvothermal method, no impurity peaks appeared in (1:1)@(Yb,Er)-NaYF4/C-TiO2 except for those of anatase TiO2 and (Yb,Er)-NaYF4, indicating that the introduction of (Yb,Er)-NaYF4 in the solution made no visible change on the formation of anatase TiO2 during the solvothermal process. The corresponding XRD patterns for other weight ratios of the composites are shown in supplementary Figure S1 and no difference was exhibited among them except for relative peak intensities.

Figure 1
figure 1

XRD patterns of samples of (a) C-TiO2, (b) (1:1)@(Yb,Er)-NaYF4/C-TiO2, (c) (Yb,Er)-NaYF4.

Full size image

Figure 2 shows the TEM image, HRTEM image, element mapping and EDX spectrum of (1:1)@(Yb,Er)-NaYF4/C-TiO2 composite. It could be clearly seen from element mapping that the element of Na, Y, F, Yb, Er, Ti and O truly existed in the sample. The element mapping of C is not shown because a large amount of C existed in the support film of the TEM grid. However, actually, the existence of C in the TiO2 could be confirmed from the EDX spectrum and the following DRS and XPS analysis. In addition, it could also be found in Figure 2 that the large particle size of (Yb,Er)-NaYF4 (200–400 nm) were successfully coated by small particles of C-TiO2 (7–10 nm) to form core/shell structure. The d-spacing of covered small particle was about 0.351 nm, which agreed with the spacing of (101) of anatase TiO2. The specific surface area of C-TiO2 and (Yb,Er)-NaYF4 were determined as 124.3 and 11.5 m2g−1, respectively. When the C-TiO2 nanoparticles were combined with (Yb,Er)-NaYF4 powders in core/shell structure, the specific surface was changed to 70.1 m2g−1. In addition, the SEM images of the samples C-TiO2, (Yb,Er)-NaYF4 and (1:1)@(Yb,Er)-NaYF4/C-TiO2 were also characterized (see supplementary Figure S2). After coupling small nanosize particle of C-TiO2, the morphology and particle size of (1:1)@(Yb,Er)-NaYF4/C-TiO2 composite presented no big variation compared with those of solely (Yb,Er)-NaYF4 except for a rough surface due to the covering of C-TiO2 nanoparticles.

Figure 2
figure 2

Element mapping, TEM image, HRTEM image and EDX spectrum of the sample (1:1)@(Yb,Er)-NaYF4/C-TiO2.

Full size image

Figure 3 presents the DRS, PL spectra, visual emitting map of C-TiO2, (1:1)@(Yb,Er)-NaYF4/TiO2, (1:1)@(Yb,Er)-NaYF4/C-TiO2 and (Yb,Er)-NaYF4. It was apparent that C-TiO2 exhibited a nice visible light absorption up to about 600 nm due to the C doping in TiO2 by inducing some impurity energy levels of C above the valence band of TiO2 (Figure 3 (a))30,31. (Yb,Er)-NaYF4 displayed only the absorption band from 908 to 1028 nm, corresponding to the 2F7/2 → 2F5/2 transition of Yb3+ in NaYF433,34. When TiO2 was combined with (Yb,Er)-NaYF4, (1:1)@(Yb,Er)-NaYF4/TiO2 represented a strong UV absorption of TiO2 with the NIR light absorption band of (Yb,Er)-NaYF4. Furthermore, (1:1)@(Yb,Er)-NaYF4/C-TiO2 showed the significantly enhanced visible light absorption up to 900 nm, owing to the introduction of C into TiO2 lattice during calcination35,36. In addition, it is interesting that the visible light absorption of (1:1)@(Yb,Er)-NaYF4/C-TiO2 was much stronger than that of pure C-TiO2, which is probably attributed to the increased C doping contents in TiO2 in the presence of (Yb,Er)-NaYF4 crystal. This phenomenon was also observed in our precious works36. The doped C contents in TiO2 would be increased when other impurity ions (Nd3+, etc.) or other phases were existed in the preparation of C-TiO2 by solvothermal method. The advisable reason for this phenomenon is still unclear which will be our next work. The effects of (Yb,Er)-NaYF4 contents on the DRS of (Yb,Er)-NaYF4/C-TiO2 composites are demonstrated in supplementary Figure S3. It was obvious that the visible light absorption of the samples improved with an increase in (Yb,Er)-NaYF4 content. When the four kinds of samples in Figure 3(a) were excited by a NIR diode laser light (980 nm), different intensities of green light were emitted as shown in Figure 3(b). The (Yb,Er)-NaYF4 powders showed the highest emission intensity with four emission peaks at about 408, 525, 540 and 653 nm, which are corresponding to the transitions from 2H9/2, 2H11/2, 4S3/2 and 4F9/2 to 4I15/2 of Er3+, respectively37,38. When the (Yb,Er)-NaYF4 powders were mixed with undoped TiO2 particles to form (1:1)@(Yb,Er)-NaYF4/TiO2, the relative emission intensity of peak at 540 nm decreased to about 1/4, probably due to the shielding of the excitation light by TiO2 nanoparticles on the surface of (Yb,Er)-NaYF4 core. More importantly, the relative emission intensity of sample (1:1)@(Yb,Er)-NaYF4/C-TiO2 declined more significantly. The reason for this dramatic decrement of emission intensity might be explained by Figure 3 (d). It could be explicitly observed that the DRS of (1:1)@(Yb,Er)-NaYF4/C-TiO2 had a good overlap with four emission peaks in PL spectrum, while no overlap could be observed for (1:1)@(Yb,Er)-NaYF4/TiO2, indicating that C-TiO2 could efficiently absorb the visible fluorescence emitted from (Yb,Er)-NaYF4. When the (1:1)@(Yb,Er)-NaYF4/C-TiO2 composite was excited by 980 nm laser, the up-conversion phosphor (Yb,Er)-NaYF4 in the composite would emit green light with four emission peaks at 408, 525, 540 and 653 nm. Meanwhile, a large part of emitted green light would be efficiently absorbed by the C-TiO2 particles coated on the surface of (Yb,Er)-NaYF4, eventually leading to the strong reduction of emission intensity of green light compared with (Yb,Er)-NaYF4 and (1:1)@(Yb,Er)-NaYF4/TiO2. As for sample C-TiO2, no green light was observed when excited by 980 nm laser, since no up-conversion occurred in this sample. The corresponding visual emitting maps for these four samples monitored by 980 nm NIR light irradiation are presented in Figure 3 (c), in which the laser sources and IR light path are shown by computer graphics.

Figure 3
figure 3

DRS(a), PL spectra(b), visual emitting map(c) of C-TiO2, (1:1)@(Yb,Er)-NaYF4/TiO2, (1:1)@(Yb,Er)-NaYF4/C-TiO2 and (Yb,Er)-NaYF4 under the excitation of 980 nm laser and (d) overlap between the DRS and PL spectrum of (1:1)@(Yb,Er)-NaYF4/C-TiO2.

For comparison, the DRS of (1:1)@(Yb,Er)-NaYF4/TiO2 was also indicated in (d).

Full size image

In order to investigate the status of C in the TiO2 lattice, the XPS measurement was employed. From XPS spectra of C 1s for C-TiO2 and (1:1)@(Yb,Er)-NaYF4/C-TiO2 in Figure 4, it could be obviously seen that both samples exhibited three peaks at 282.0 eV, 284.6 eV and 285.6 eV. The peak at 284.6 eV was ascribed to the adventitious carbon species from the XPS measurement and the peak around 285.6 eV is corresponding to the elemental carbon which has the same binding energy as that of carbon in the graphite intercalation compound39,40. More importantly, the small peak at about 282.0 eV should be ascribed to the binding energy of C-Ti bond, indicating that the C had been successfully doped into TiO2 lattice by replacing O site31,41. The relative area of 282.0 eV peak for (1:1)@(Yb,Er)-NaYF4/C-TiO2 was larger than that of C-TiO2 and the corresponding doped C concentration for two samples were about 0.38 at.% and 0.19 at.%, which were well agreements with that of DRS result in Figure 3 (a). It is well known that some impurity C energy levels will be introduced above the valence band of TiO2 when C was doped in O site of TiO2 and then the electrons can be excited from impurity C energy levels instead of the valence band of TiO2 to the conduction band of TiO2, which eventually leads to the strong visible light absorption of C-TiO216,17,30.

Figure 4
figure 4

XPS spectra of C1s for (a) C-TiO2 and (b) (1:1)@(Yb,Er)-NaYF4/C-TiO2.

Full size image

Photocatalytic activities

The photocatalytic activities of samples were investigated by testing the NOx gas destruction ability under the irradiation of UV, visible and NIR lights. Figure 5 illustrates the deNOx activity of various samples using different wavelengths of light sources and all of the samples were kept in dark for 30 min before irradiation to eliminate the influence of absorption of NO species by samples. In Figure 5 (A) and (C), it can be obviously seen that there was no obvious deNOx activity for (Yb,Er)-NaYF4 (sample (c)) regardless of irradiation wavelengths. As for pure C-TiO2 (sample (a)), 0%, 6.2%, 23.2%, 28.3% and 32.9% of NO were decomposed under the excitation of NIR diode laser, red LED, green LED, blue LED and UV LED, respectively. The nice visible light induced photocatalytic activity for this sample was mainly assigned to the excellent visible absorption of TiO2 after C doping, but no NIR light driven deNOx activity was performed, because C-TiO2 could not be excited by 980 nm NIR light. However, when C-TiO2 photocatalyst was combined with (Yb,Er)-NaYF4 up-conversion phosphor, about 8.5% of NOx gas was successfully destructed even when the NIR diode laser was used as the irradiation light source. This NIR light induced activity should be owing to the synergetic effect of C-TiO2 photocatalyst and (Yb,Er)-NaYF4 up-conversion phosphor, since no corresponding performance was observed by pure C-TiO2 and (Yb,Er)-NaYF4 individually. In addition, the visible light induced deNOx performance was significantly enhanced to 19.5%, 26.2% and 30% with the irradiation of 627, 530 and 445 nm LED lights, respectively. This interesting result might be dominantly due to the increased visible light absorption ability by increasing C doping content as shown in Figure 3(a). As for the effect of thermal reaction on the visible-NIR light induced photocatalytic activity, it could be ignored since no visible-NIR light responsive deNOx activity was observed for the pure (Yb,Er)-NaYF4 and no NIR light driven deNOx performance was exhibited for solely C-TiO2. Furthermore, the temperature was fixed at a constant room temperature during the photocatalytic testing.

Figure 5
figure 5

Time dependence of NOx destruction activity (A), (B), corresponding deNOx ability (C) of (a) C-TiO2, (b) (1:1)@(Yb,Er)-NaYF4/C-TiO2, (c) (Yb,Er)-NaYF4, (d) (1:1)@(Yb,Er)-NaYF4/TiO2, (e) (1:1)@(Yb,Er)-NaYF4/N-TiO2, (f) P25 titania under the irradiation of different wavelengths of lights and the effect of (Yb,Er)-NaYF4 content on the deNOx ability of (Yb,Er)-NaYF4/C-TiO2 composites (D).

Full size image

The detailed mechanism for this synergy phenomenon will be explained in the following section. Besides, the deNOx activity of (1:1)@(Yb,Er)-NaYF4/C-TiO2 was also compared with those of (1:1)@(Yb,Er)-NaYF4/TiO2, (1:1)@(Yb,Er)-NaYF4/N-TiO2 and P25 using different wavelengths of lights (Figure 5(B)). When (Yb,Er)-NaYF4 up-conversion phosphor was mixed with pure TiO2, only 27.5% of UV light induced photocatalytic activity was displayed and no distinct visible and NIR lights induced deNOx activity was presented due to the poor visible and NIR light absorption of pure TiO2. As for P25, 26.8% and 9.6% of NOx were destructed when UV and blue LEDs were used as light sources, respectively. The XRD pattern, DRS and photoluminescence spectrum of (1:1)@(Yb,Er)-NaYF4/N-TiO2 are illustrated in supplementary Figure S4. The XRD profie of (1:1)@(Yb,Er)-NaYF4/N-TiO2 was similar to that of (1:1)@(Yb,Er)-NaYF4/C-TiO2, but the visible light absorption was much weaker, indicating only extend the absorption tail to about 560 nm. Consequently, (1:1)@(Yb,Er)-NaYF4/N-TiO2 exhibited only 7.3% and 23% of NOx destruction ability with the visible light irradiation of 530 and 445 nm, respectively. Although the absorption spectrum of (1:1)@(Yb,Er)-NaYF4/N-TiO2 showed a little overlap with its photoluminescence spectrum, it was still not effective enough to induce the obvious deNOx performance under the excitation of 980 nm light. Furthermore, the effect of (Yb,Er)-NaYF4 content on the photocatalytic activity of (Yb,Er)-NaYF4/C-TiO2 composites was also studied using different wavelengths lights (in Figure 5-(D)). (1:1)@(Yb,Er)-NaYF4/C-TiO2 demonstrated a little better deNOx ability than those of other compositions of (Yb,Er)-NaYF4/C-TiO2 composites.

In addition, the deNOx ability of (1:1)@(Yb,Er)-NaYF4/C-TiO2 was also investigated by employing NIR laser, red LED, the combination of NIR laser and red LED as irradiation sources (Figure 6). It was clearly exhibited that about 19% of NOx gas was destructed under the excitation of red LED. While when NIR laser was combined with red LED as the light source, further 4.5% of NOx gas reduction was appeared and then moved back to 19% again as the NIR diode laser was eliminated. This result indicated that the NIR light induced deNOx ability of the C-TiO2 photocatalyst combined with (Yb,Er)-NaYF4 up-conversion phosphor truly had a positive effect on the total photocatalytic activity of photocatalyst when irradiated by UV, visible, NIR lights, therefore, the solar light might be more effectively utilized for photocatalysis compared with that of pure C-TiO2 photocatalyst in the absence of up-conversion phosphor. Meanwhile, the dependence of deNOx ability of (1:1)@(Yb,Er)-NaYF4/C-TiO2 on the irradiation wattages of NIR diode laser is illustrated (see supplementary Figure S5). With an increase in irradiation wattage of NIR laser, the corresponding deNOx ability was increased, which further confirmed that the NIR light induced photocatalytic activity was actually realized and could be affected by irradiation energy power.

Figure 6
figure 6

DeNOx ability of sample (1:1)@(Yb,Er)-NaYF4/C-TiO2 with respect to the irradiation lights.

Full size image

It is well known that the photoluminescence of phosphor is sensitive to the operating condition and the surround atmosphere of the photocatalyst may change after irradiation for some time42,43,44. Therefore, the reproducibility of NIR light induced deNOx ability of (Yb,Er)-NaYF4/C-TiO2 composite was evaluated. The deNOx reaction performance of (1:1)@(Yb,Er)-NaYF4/C-TiO2 with the repeated induction of 980 nm light is shown in Figure 7. It is apparent that the NIR light induced deNOx ability of the sample presented no obvious difference after 4 times running, implying that the (Yb,Er)-NaYF4/C-TiO2 composite had an excellent stability for NOx destruction when irradiated by NIR light.

Figure 7
figure 7

The multi-cycles of deNOx ability of (1:1)@(Yb,Er)-NaYF4/C-TiO2 under the irradiation of 980 nm light.

Full size image

Discussion

Figure 8 illustrates the photoluminescence and UV, visible, NIR lights induced photocatalytic reaction mechanism of the (Yb,Er)-NaYF4/C-TiO2 composite. It is accepted that up-conversion phosphors can convert longer wavelength radiation (NIR light) to shorter wavelength fluorescence (UV or visible light) via a two-photon or multiphoton mechanism and then achieve the anti-stokes shift45,46. When a 980 nm NIR laser was used to excite the (Yb,Er)-NaYF4/C-TiO2, the Yb3+ as a sensitizer ion in (Yb,Er)-NaYF4 phosphor could absorb this light and then an electron was able to be excited from the 2F7/2 to 2F5/2 level. After that, the excited electron would transfer back to the ground state of 2F7/2 and the released energy in this process was dominantly nonradiatively transferred to active ions of Er3+, leading to a population of Er3+ from 4I15/2 to 4I11/2. In addition, a second or more similar 980 nm photons from excited Yb3+ ion could populate to much higher 4F9/2, 4F7/2 and 2H9/2 energetic levels of Er3+ ion. Some of the excited electrons of Er3+ ion then would relax nonradiatively by a fast multiphonon decay process to the 2H11/2, 4S3/2, 4F9/2 etc. energy levels. Finally, these excited electrons would radiatively transfer to the Er3+ ground state of 4I15/2 and present four emissions at 408 nm (2H9/2 → 4I15/2), 521 nm ( 2H11/2 → 4I15/2), 540 nm ( 4S3/2 → 4I15/2) and 653 nm (4F9/2 → 4I15/2)45,47. The (Yb,Er)-NaYF4 eventally emitted green light due to the main peak at 540 nm. In this case, the as-emitted green light from (Yb,Er)-NaYF4 could be reabsorbed by the surface C-TiO2 nanoparticles which had an excellent light absorption in the range of above mentioned four peaks owing to the C doping in TiO2 (shown in Figure 3). At that time, the electron in TiO2 was excited to the conduction band of TiO2 from C doped impurity energy levels, producing a pair of hole and electron in TiO2 lattice. The photogenerated hole would then be trapped by water and adsorbed OH− species in the air to generate hydroxyl radicals ·OH. Besides, the photoinduced electron in the conduction band would form ·O2− in the presence of oxygen. In addition, the detailed destruction mechanism of NO species by photocatalyst has been investigated by some researchers48,49,50,51,52. The characterization system in the present research was quite similar to that of the Japanese Industrial Standard which was established at the beginning of 200454. In this JIS standard, it is recommended that the photocatalytic activity of the photocatalyst should be characterized by measuring the decrease in the concentration of NO at the outlet of a continuous reactor. In the presence of oxygen, the electrons in the conduction band are immediately trapped by the molecular oxygen to form ·O2−, which can then generate active ·OOH radicals20,51,52. The nitrogen monoxide reacts with these reactive oxygen radicals, molecular oxygen and very small amount of water in the air to produce HNO2 or HNO3. It was reported that about 20% of nitrogen monoxide was decomposed to nitrogen and oxygen directly53. As for UV, visible lights induced photocatalytic performance of (Yb,Er)-NaYF4/C-TiO2 composite, it was very similar to that of NIR light induced activity. The only difference was that the irradiation light source for photocatalysis was not transferred from up-conversion phosphor but directly from the UV or visible light sources and then the electron would be excited from the valence band of TiO2 or C doped energy levels to the conduction band of TiO2, eventually inducing the photocatalysis.

Figure 8
figure 8

Schematic illustration of the photoluminescence and photocatalytic mechanism of (Yb,Er)-NaYF4/C-TiO2 composite.

Full size image

In summary, the (Yb,Er)-NaYF4/C-TiO2 composites were successfully prepared by a simple calcination assisted solvothermal method. The results showed that when C-TiO2 photocatalyst was combined with (Yb,Er)-NaYF4 up-conversion phosphor, the visible absorption of C-TiO2 was significantly improved up to 900 nm, which had a nicely overlap with the photoluminescence spectrum of (Yb,Er)-NaYF4 excited by 980 nm. Due to this reason, the (Yb,Er)-NaYF4/C-TiO2 composite presented excellent deNOx ability not only under the irradiation of visible and UV lights but also NIR light, being much superior to those of pure C-TiO2, P25 and even (Yb,Er)-NaYF4/N-TiO2 composite. In addition, the composite exhibited outstanding stability for the photocatalytic activity and truly played a positive effect in the improvement of utilization of total solar light (UV, visible, NIR lights) for photocatalysis. A promising strategy to utilize UV, weak visible lights and infrared light for photocatalysis simultaneously and effectively has been established. Furthermore, it also provides a strong potential to couple various color light-emitting up-conversion phosphors (such as blue, green, yellow, red emitting phosphors) with excellent visible light responsive C doped TiO2 to form novel composites with high photocatalytic activity under the irradiation of UV, visible and infrared lights.

Methods

Sample preparation

A series of (Yb,Er)-NaYF4/C-TiO2 composites, with the (Yb,Er)-NaYF4/C-TiO2 weight ratios of 1:2, 1:1 and 2:1 were prepared by a simple calcination assisted solvothermal method and were denoted as (1:2)@(Yb,Er)-NaYF4/C-TiO2, (1:1)@(Yb,Er)-NaYF4/C-TiO2 and (2:1)@(Yb,Er)-NaYF4/C-TiO2, respectively. The (Yb,Er)-NaYF4 used in this experiment was purchased from Shanghai Huaming Gona Rare Earth New Materials Co., Ltd. (China). In a typical synthesis process of (1:1)@(Yb,Er)-NaYF4/C-TiO2 particle, The appropriate amounts of (Yb,Er)-NaYF4 particles were dispersed in 20 mL ethanol with continuous stirring for 30 min. After that, 0.6 mL titanium tetra-n-butoxide was introduced dropwise to the suspension solution with another 30 min magnetic stirring. Subsequently, 10.5 mL ethanol/water (10:0.5) mixed solution was added dropwise to the solution and stirred for 60 min before transferring into a 100 mL Teflon-lined stainless steel autoclave. Finally, the solution was heated at 190°C for 2 h and the products were separated by centrifugation, washed and calcined at 265°C for 1 h in ambient condition. For comparison, C-TiO2 was prepared by a similar process without the addition of (Yb,Er)-NaYF4 particles and (1:1)@(Yb,Er)-NaYF4/TiO2 without calcination at 265°C.

For comparison, (1:1)@(Yb,Er)-NaYF4/N-TiO2 in which nitrogen doped titania (N-TiO2) was utilized instead of the carbon doped titania (C-TiO2) was also synthesized by a conventional hydrothermal method similar to our previous paper20. A desired amount of (Yb,Er)-NaYF4, HMT (3 g, 99%) and 2 mL of 20 wt% TiCl3 solution were mixed with 30 mL of distilled water. The mixture was placed into a stainless steel autoclave attached to a Teflon tube of internal volume of 100 mL. The autoclave was heated and kept at 90°C for 1 h to realize homogeneous precipitation and then heated at 190°C for 2 h. The powder product was separated by centrifugation, washed with distilled water and acetone three times, then vacuum dried at 80°C overnight. In order to eliminate the effect of organics remained in the sample, the dried particles were finally calcined at 450°C for 2 h.

Characterization

The crystalline phases of the products were identified by X-ray diffraction analysis (XRD, Bruker AXS D2 Phaser) using graphite-monochromized CuKα radiation. The UV–vis diffuse reflectance spectra (DRS) were measured using a UV–vis spectrophotometer (Shimadazu, UV-2450). The specific surface areas were determined by the BET method (Quantachrome Instruments, NOVA4200e). The size, shape and element mapping of the composites were observed by transmission electron microscopy (TEM, JEOLJEM-2010) and high resolution transmission electron microscopy (HRTEM) (FE-TEM, JEM-2100F). The chemical compositions of the samples were determined by the EDX spectrometer (Rayny EDX-800HS, Shimadzu). The surface composition and binding energy of the samples were determined by X-ray photoelectron spectroscopy (XPS, Perkin Elmer PHI 5600). The shift of the binding energy owing to relative surface charging was corrected using the C 1 s level at 284.6 eV as an internal standard and Ar+ sputtering was employed to clean the surface of samples. Photoluminescence (PL) spectra of specimens were measured by a spectrofluorometer (Shimadzu RF-5300P).

Photocatalytic activity tests

The photocatalytic activities of samples were investigated by evaluating the decomposition of NOx (DeNOx) in a flow type reactor under irradiation of different wavelengths of LEDs and NIR laser (see supplementary Figure S6) at room temperature, where the detailed testing parameters were summarized in supplementary Table S1. The sample powder was spread in the hollow (20 mm × 16 mm × 0.5 mm) of a glass plate and then placed at the bottom center of the reactor (373 cm3 of internal volume) in which a 1:1 mixed gas of air and nitrogen balanced 1 ppm of NO was flowed at the rate of 200 cm3 min−1. The sample powder was kept in the dark for 30 min to reach the adsorption and desorption equilibrium of NO gas. After that, the light source was turned on to irradiate the sample for 10 min and then turned off to recover to the original concentration for another 10 min. The measuring time for each sample under each light source was 20 min17. The concentration of NO was monitored by a NOx analyzer (Yanaco, ECL-88A) continuously. In addition, the testing temperature was kept at a constant room temperature by using an electric fan in an air conditioner room.