Upconversion-based nanosystems for fluorescence sensing of pH and H2O2

Hydrogen peroxide (H2O2), a key reactive oxygen species, plays an important role in living organisms, industrial and environmental fields. Here, a non-contact upconversion nanosystem based on the excitation energy attenuation (EEA) effect and a conventional upconversion nanosystem based on the joint effect of EEA and fluorescence resonance energy transfer (FRET) are designed for the fluorescence sensing of H2O2. We show that the upconversion luminescence (UCL) is quenched by MoO3−x nanosheets (NSs) in both systems due to the strong absorbance of MoO3−x NSs in the visible and near-infrared regions. The recovery in UCL emissions upon addition of H2O2 enables quantitative monitoring of H2O2. Benefiting from the non-contact method, hydrophobic OA-NaYF4:Yb,Er can be used as the luminophore directly and ultrahigh quenching efficiency (99.8%) is obtained. Moreover, the non-contact method exhibits high sensitivity toward H2O2 with a detection limit of 0.63 μM, which is lower than that determined by simple spectrophotometry (0.75 μM) and conventional upconversion-based nanocomposites (9.61 μM). As an added benefit, the same strategy can be applied to the sensing of pH, showing a broad pH-responsive property over a range of 2.6 to 8.2. The successful preparation of different upconversion-based nanosystems for H2O2 sensing using the same material as the quencher provides a new design strategy for fluorescence sensing of other analytes.


Introduction
Hydrogen peroxide (H 2 O 2 ), an important bioactive molecule in living systems, plays an essential role in the physiological process including signal transduction, cell proliferation, differentiation, and maintenance. 1,2 Abnormal production or accumulation of H 2 O 2 will lead to severe damage to DNA and proteins, causing a series of serious diseases, [3][4][5][6][7] such as diabetes, Alzheimer's and Parkinson's disease, cardiovascular disorders, and even cancer. Additionally, H 2 O 2 is widely used as a bleaching agent and sterilant in industrial and environmental elds, 8,9 such as food processing, drinking water treatment, packaging, and organic pollutant degradation. However, exposure to high concentrations of H 2 O 2 is a great threat to organisms. 10,11 Therefore, quantitative detection of H 2 O 2 is of great importance for monitoring its potential risk.
Optical methods via uorescence changes have attracted considerable attention, as the uorometric approach is a nondestructive method that can be simply and rapidly performed with high sensitivity and selectivity. 12 In contrast to conventional uorescence probes (such as organic dyes, carbon nanomaterials, and semiconductor quantum dots), upconversion nanoparticles featuring large anti-Stokes shis, excellent chemical-and photo-stability, sharp multicolor emissions, and low toxicity have been regarded as a promising class of luminophores. 13 Up to now, a variety of functional materials including organic dyes, [14][15][16] noble metals, [17][18][19] quantum dots, [20][21][22] carbon nanomaterials, [23][24][25] and two-dimensional materials [26][27][28] has been employed to couple with upconversion nanoparticles to construct uorescence probes, realizing quantitative detection of inorganic ions, [29][30][31] pH, 32-34 small molecules, [35][36][37][38] and nucleic acids. [39][40][41][42] Most of the upconversion-based probes rely on the uorescence resonance energy transfer (FRET) process, in which a very short distance between the upconversion nanoparticles and absorbers is required. Moreover, in order to obtain high-sensitivity detection, high-quality upconversion nanoparticles with strong emission and high upconversion efficiency are employed, which are commonly prepared by applying oleic acid (OA) as the ligand. The oleate-capped upconversion nanoparticles are hydrophobic and prone to disperse in nonpolar solvents, whereas hydrophilic upconversion nanoparticles are required for typical sensing applications of interest. Therefore, the hydrophobic-to-hydrophilic transition of upconversion nanoparticles is essential. 43 Herein, we propose different upconversion nanosystems for H 2 O 2 sensing using MoO 3Àx nanosheets (NSs) as the energy acceptor based on either the excitation energy attenuation (EEA) effect or the joint effect of the EEA and FRET, owing to the strong absorbance of MoO 3Àx NSs in both visible and nearinfrared (NIR) regions. By coupling of MoO 3Àx NSs solution and oleate-capped NaYF 4 :Yb,Er upconversion nanoparticles (abbreviated as OA-UCNPs) solution, a EEA-based upconversion nanosystem for sensing of H 2 O 2 in the non-contact mode is designed, where MoO 3Àx NSs act as the energy acceptor of the incident light for the activation of UCNPs. Additionally, this system can be used for pH sensing as well. Beneting from the non-contact method, hydrophobic OA-UCNPs can be used directly for the sensing and ultrahigh quenching efficiency (99.8%) can be reached. Meanwhile, by the integration of hydrophilic UCNPs and MoO 3Àx NSs, we are able to prepare conventional upconversion-based nanocomposites for H 2 O 2 sensing via the joint effect of the EEA and FRET, where MoO 3Àx NSs act as the energy acceptor of not only the 980 nm exciting light for UCNPs but also uorescence emissions of UCNPs. To the best of our knowledge, this is the rst upconversion-based nanoprobe for the sensing of one analyte by two different systems while using the same material as an energy acceptor.

Characterization
Fourier transform infrared (FT-IR) spectra were recorded in transmission mode on a Thermo Scientic Nicolet iS5 FT-IR spectrometer with the KBr method. X-ray photoelectron spectroscopy (XPS) was measured with a Thermo Fisher Scientic ESCALAB 250Xi instrument. Transmission electron microscopy (TEM) and energy-dispersive X-ray spectroscopy (EDS) were performed on the FEI Tecnai G2 20 S-TWIN with a LaB 6 cathode operated at 200 kV. UV-vis absorption spectra were acquired on a CARY 50 spectrophotometer. Powder X-ray diffraction (XRD) measurements were performed on a Philips X'Pert MPD Pro Xray diffractometer at a scanning rate of 4 min À1 in the 2q range from 10 to 80 (Cu Ka radiation, l ¼ 0.15406 nm). z-Potential measurements were carried out on an Anton Paar Litesizer™ 500 instrument. Upconversion luminescence (UCL) emission spectra were obtained on a ber-coupled spectrometer (Ocean HDX, Ocean Optics) with an external 980 nm continuous-wave (CW) laser (0-5 W, Roithner Lasertechnik GmbH) at room temperature (RT). Quartz cuvettes (0.7 mL, 10 mm Â 2 mm light path) were used for UV-vis absorption and UCL measurements.

Synthesis of MoO 3Àx NSs
MoO 3Àx NSs were prepared according to the previous publication with minor modications. 44,45 In a typical process, 1.5 g bulk MoO 3 powder was ground with 0.3 mL of acetonitrile for 30 min and then added to a water/ethanol solution (25 mL, v/v ¼ 1/1). The dispersion was then probe-sonicated for 2 h at 100 W (Branson Digital Sonier W-250D) at a 5 s ON and 2 s OFF pulse. To avoid overheating of the solvent, the beaker lled with MoO 3 dispersion was immersed in an ice bath during sonication. The light blue supernatant containing a high concentration of MoO 3 NSs (denoted as S-MoO 3 NSs) was collected via centrifugation at 7000g for 30 min. For the preparation of MoO 3Àx NSs, the supernatant dispersion was lled into a quartz glass vial and irradiated with a UV lamp (254 & 365 nm, 15 W) for 5 h, dark blue MoO 3Àx NSs solution was nally obtained, and the MoO 3Àx NSs solution was then diluted to 2 mg mL À1 by water and ethanol (v/v ¼ 1/1) solution, and stored at 4 C for further use.

Synthesis of OA-UCNPs
As previously reported, the synthesis of oleate-capped NaYF 4 : 20 mol% Yb, 2 mol% Er was carried out by employing OA as ligand via a high-temperature coprecipitation method. 46 Briey, in a 100 mL round ask, 3.12 mL of Y(CH 3 COO) 3 (0.2 M), 0.8 mL of Yb(CH 3 COO) 3 (0.2 M) and 0.8 of mL Er(CH 3 COO) 3 (0.02 M) were mixed with 6 mL of OA and 14 mL of ODE at RT. The mixture solution was rst heated to 110 C for 30 min to evaporate the water and then heated to 160 C for 40 min to form lanthanide-oleate complexes, followed by cooling down to 50 C. A methanolic solution (10 mL) containing 3.2 mmol of NH 4 F and 2.0 mmol of NaOH was slowly added and then stirred at 50 C for 30 min. Aer evaporating the methanol, the solution was heated to 310 C at a rate of 10 C min À1 and maintained for 30 min under nitrogen atmosphere. Aer cooling down to RT, OA-UCNPs were precipitated out with the addition of excess ethanol, collected aer washing three times with the ethanol, and nally dissolved in cyclohexane for further use.

Preparation of ligand-free UCNPs
Ligand-free UCNPs were prepared using our previously reported method. 47 5 mmol of formic acid was directly added to 2 mL of cyclohexane solution containing 20 mg of OA-UCNPs, ligandfree UCNPs were precipitated out aer shaking for 10 s at 3000 rpm on a vortex mixer. Bare UCNPs were obtained aer centrifugation and washing once with ethanol and three times with water and nally dissolved in water.

Synthesis of UCNPs/MoO 3Àx nanocomposites
To synthesize UCNPs/MoO 3Àx nanocomposites, PEI-capped UCNPs (abbreviated as PEI-UCNPs) was rst prepared. Typically, 4 mL ligand-free UCNPs solution (5 mg mL À1 ) were added to a vial containing 4 mL PEI solution (10 mg mL À1 ), followed by overnight stirring. PEI-UCNPs were collected aer centrifugation at 16 000g for 30 min and washing three times with water, and nally dispersed in water with a concentration of 1 mg mL À1 . UCNPs/MoO 3Àx nanocomposites were prepared by mixing 0.5 mL PEI-UCNPs solution with an appropriate amount of MoO 3Àx NSs solution, the mixture was rst shaken for 3 min (3000 rpm) on a vortex mixer and then ultrasonicated for 5 min. UCNPs/MoO 3Àx nanocomposites were then collected by centrifugation at 7000g for 30 min, washed three times with water, and redispersed in water.

Non-contact uorescence sensing of pH
To detect pH in the non-contact mode, OA-UCNPs dispersed in cyclohexane with a concentration of 1 mg mL À1 were sealed in a quartz cuvette, the cuvette was then aligned with the other cuvette containing 1 mg mL À1 MoO 3Àx NSs solution with different pH. The pH was adjusted by either 50 mM NaOH or 50 mM HCl ethanol/H 2 O (v/v ¼ 1/1) solution. The cuvette containing MoO 3Àx NSs was put in front of the other one containing OA-UCNPs solution, and the UCL spectra were collected under the excitation of a 4 W 980 nm CW laser.

Non-contact uorescence sensing of H 2 O 2
The non-contact sensing procedure for the H 2 O 2 was similar to that of the non-contact pH sensing, except that MoO 3Àx NSs were dissolved in acetate buffer (50 mM, pH 4.

Fluorescence sensing of H 2 O 2 by UCNPs/MoO 3Àx nanoassemblies
To detect H 2 O 2 , 0.5 mg mL À1 of UCNPs/MoO 3Àx aqueous solution (0.35 mg mL À1 MoO 3Àx NSs) and different concentrations of H 2 O 2 (0.4 mL) were added to 0.1 mL acetate buffer (50 mM, pH 4.5, DMF/H 2 O, v/v ¼ 1/1), The mixture was then incubated at RT for 2 h, and the UCL spectra were measured under the excitation of a 4 W 980 nm CW laser.

Design principle of upconversion-based nanosystems for H 2 O 2 and pH
The design strategy of UCNPs/MoO 3Àx nanocomposites for uorescence sensing of H 2 O 2 is based on the modulation of MoO 3Àx NSs-induced reduction in UCL emissions by H 2 O 2 through the joint effect of EEA and FRET. In contrast, the pH and H 2 O 2 dual-responsive upconversion-based nanosystem is realized by the direct adjustment of the excitation energy for UCNPs in the non-contact mode (Fig. 1a).
Without modications, UCNPs give rise to green and red luminescence emissions under 980 nm excitation. Aer the reduction of MoO 3 by UV light, the oxygen-decient MoO 3Àx NSs exhibit strong absorption in both visible and NIR regions, overlapping well with the UCL emissions of UCNPs and the excitation wavelength for UCNPs of 980 nm (Fig. 1b). Owing to the strong NIR absorption of MoO 3Àx NSs attached on UCNPs, the EEA will rst take place in the UCNPs/MoO 3Àx system when activated by the 980 nm light, resulting in a lowered intensity of excitation light arriving at the UCNPs, thus weakening the resulting luminescence emissions. Moreover, the efficient FRET process occurs through the spectral overlap between the absorption of MoO 3Àx NSs and the UCL of UCNPs in the visible region, leading to a further decrease in the intensity of luminescence emissions. Thus, the quenching in UCL of UCNPs is efficiently achieved by the joint effect of the EEA and FRET. However, upon the addition of H 2 O 2 , the oxygen-decient MoO 3Àx NSs can be oxidized back to MoO 3 (denoted as H-MoO 3 ), leading to the decrease of absorption in the visible and NIR regions (Fig. 1b), resulting in the recovery of UCL emissions via the reduction in EEA and FRET. Additionally, XPS was performed to evaluate the valence state of Mo in these nanosheets. As shown in Fig. 1c, the doublet peaks (235.9 eV and 232.8 eV) in the pristine MoO 3 sample are assigned to the binding energies of the 3d 3/2 and 3d 5/2 orbital electrons of Mo 6+ . Aer treatment by tip-sonication, two new peaks at lower binding energies (234.7 eV and 231.6 eV) appear in the obtained S-MoO 3 NSs, which can be assigned to the Mo 5+ oxidation state, and the integral area ratio of Mo 5+ /Mo 6+ is calculated to be 17.1% from the XPS spectrum. This phenomenon indicates that the MoO 3 is slightly reduced during the exfoliation process, showing weak absorption ability of S-MoO 3 NSs in visible and NIR regions (Fig. 1b)  MoO 3Àx enables the ability of UCNPs/MoO 3Àx nanoprobes for H 2 O 2 sensing with high sensitivity. Additionally, the adjustment of pH or addition of H 2 O 2 in the acidic environment will lead to the variation of MoO 3Àx NSs in NIR absorption, and thus uorescence sensing of pH and H 2 O 2 can be achieved through the direct modulation of MoO 3Àx absorption-induced EEA in the non-contact mode.
Characterization of UCNPs, MoO 3Àx NSs, and UCNPs/MoO 3Àx nanocomposites Hydrophobic OA-UCNPs are synthesized by employing OA as the ligand via the high-temperature coprecipitation method. 46 OA-UCNPs present uniform hexagonal shape with a mean diameter of about 28 nm, which is revealed by the TEM measurement (Fig. 2a). The XRD pattern of the obtained OA-UCNPs with well-dened diffraction peaks agrees well with the standard data of hexagonal-phase NaYF 4 (JCPDS no. , demonstrating their high crystallinity (Fig. S1 †). Ligandfree UCNPs are prepared by direct addition of formic acid to the cyclohexane solution containing OA-UCNPs through the vortexing method and sequential modication with PEI to obtain PEI-UCNPs. 47 TEM images demonstrate unchanged morphology and size aer ligand removal and polymer functionalization (Fig. S2 †). The transition of OA-UCNPs to ligand-free UNCPs and further to PEI-UCNPs are conrmed by FT-IR. As shown in Fig. S3, † the transmission bands at 2926 and 2852 cm À1 can be assigned to asymmetric and symmetric methylene (-CH 2 -) stretching, and those at 1561 and 1460 cm À1 can be attributed to the vibrations of the carboxylate groups, indicating the presence of oleate ligand on the surface of OA-UCNPs. However, the disappearance of these characteristic peaks conrms the removal of surface ligand aer treatment by formic acid. When further modied by PEI, new peaks appear at 3396 cm À1 (N-H stretching), 2930 and 2854 cm À1 (asymmetric and symmetric -CH 2 À stretching), and 1545 cm À1 (N-H bending). Accordingly, the FT-IR results verify the success in ligand removal of OA-UCNPs and further attachment of PEI on bare UCNPs. Aer ligand exfoliation and polymer modication, ligand-free UCNPs and PEI-UCNPs are easily dispersed in water, and the z-potentials are measured to be +35.7 mV and +32.8 mV, respectively (Fig. S4 †), indicating the formation of stable colloidal solutions.
To prepare UCNPs/MoO 3Àx nanoassemblies, MoO 3 NSs are rstly prepared by tip sonication of bulk MoO 3 , and oxygen-decient MoO 3Àx NSs are easily obtained by UV irritation. 45 As shown in Fig. 2b, the nanostructure of the MoO 3Àx sample is comprised of NSs with lateral diameters in the range of 20-300 nm. UCNPs/MoO 3Àx nanoassemblies are then constructed by assembling the positive charged PEI-UCNPs and negatively charged MoO 3Àx NSs (Fig. S4 †) via electrostatic interactions, as characterized by TEM (Fig. 2c). Furthermore, the EDS spectrum of UCNPs/MoO 3Àx nanocomposites implies the presence of Na, F, Y, Yb, Er, Mo, and O. These results prove the successful assembling of UCNPs and MoO 3Àx NSs (Fig. 2d).
Next The optical properties of MoO 3Àx NSs solutions (1 mg mL À1 ) at different pH are rst investigated by UV-vis spectroscopy. As represented in Fig. 3a, the absorption intensity in the visible and NIR regions becomes weakened with increasing pH, and the maximum of the absorption peak gradually redshis from 744 to 866 nm. However, no absorption peak is found in the visible and NIR region above pH 7. Moreover, the absorption at 980 nm shows the same trend as well (Fig. S5a †). This phenomenon arises from the reduction of Mo in the reduced state (returning to the Mo VI state) by the addition of OH À to the MoO 3Àx NSs solution, leading to the reduction of free carrier concentration, and thus reducing the absorption in visible and NIR regions. 48,49 Next, the luminescence properties are investigated by placing MoO 3Àx NSs solutions (1 mg mL À1 ) with different pH in front of the OA-UCNPs solution (1 mg mL À1 ) and illuminate it then with the light of 980 nm wavelength at RT, where the 980 nm light rst passes through the MoO 3Àx NSs solution and then reaches OA-UCNPs (Fig. 1a). The luminescence intensity rises generally with increasing pH and remains constant above pH 8.2, as is presented in Fig. 3b. The luminescence intensity at 658 nm grows slowly when pH < 4.4, then increases remarkably in the range of 5.0 to 8.2, and the UCL shows no signicant change aerward. However, the UCL intensity at 658 nm shows a nonlinear relationship with the pH, which is different from typical upconversion sensors based on the FRET process. [32][33][34] Notably, we nd that the logarithm of luminescence intensity at 658 nm exhibits three-separate linear regions with pH, and the linear correlation coefficient of each calibration curve is calculated to be 0.992 (pH 2.6-4.4), 0.988 (pH 5-6), and 0.998 (6.3-8.2), respectively (Fig. 3c). Thus, this upconversion-based sensor shows broad pH responsiveness in the range of 2.6 to 8.2. To investigate the reversibility of this pH sensor, the pH value of MoO 3Àx NSs was adjusted from 8.2 to 2.6 and back to 8.2 by NaOH and HCl solutions for 5 cycles. As shown in Fig. S6, † the uorescence intensity shows good reversibility of the two-way switching processes aer the second cycle of pH adjustment. A slight increase in the uorescence intensity at pH 2.6 was noticed aer the rst pH adjustment from 8.2, which may result from a lower reduction degree of Mo(VI) in the acidic environment than under exposure to the UV light.

Non-contact uorescence sensing of H 2 O 2
The sensing ability of the upconversion-based nanosystem for H 2 O 2 in the non-contact mode is evaluated by the UV-vis absorption and UCL spectroscopy. As can be seen in the absorption spectrum (Fig. 4a) (Fig. 4b). The linear correlation coefficients of these two calibration curves are larger than 0.99, and the limit of detection (LOD) is calculated to be 0.75 mM.  The luminescence properties are then studied using similar procedures as the above-mentioned pH sensing, except that MoO 3Àx solutions (1 mg mL À1 in acetate buffer, pH 4.5) with different added H 2 O 2 concentrations are placed in front of the OA-UCNPs solution. The quenching efficiency (denoted as (F 0 À F)/F 0 , where F and F 0 represent the luminescence intensity at a specic wavelength in the presence and absence of MoO 3Àx NSs, respectively) at 658 nm reaches 99.8% when 1 mg mL À1 MoO 3Àx NSs solution is aligned in front of 1 mg mL À1 OA-UCNPs solution. When H 2 O 2 is added in the range from 0 to 0.8 mM, the absorption intensity of MoO 3Àx NSs solution at 980 nm shows a continuous decrease (Fig. S5b †). As a result, the UCL intensity of OA-UCNPs experiences a gradual uptrend in both red and green regions upon 980 nm excitation with the increasing addition of H 2 O 2 (Fig. 4c). This can be ascribed to the oxidation of MoO 3Àx to MoO 3 by H 2 O 2 , leading to the reduction in excitation energy depletion by MoO 3Àx NSs at 980 nm, and resulting in more excitation energy reached by OA-UCNPs. Similarly, like the above-discussed pH sensing in non-contact mode, the uorescent intensity exhibits a nonlinear relationship with the H 2 O 2 concentration as well. In addition, the logarithm of luminescence intensity at 658 nm is linearly correlated with the H 2 O 2 concentration in the range of 0-200 mM (R 1 2 ¼ 0.993) and 250-500 mM (R 2 2 ¼ 0.997), respectively ( Fig. 4d). According to the 3s rule, the detection of H 2 O 2 can be down to 0.63 mM, providing a lower detection limit than those reported by other upconversion-based nanoprobes (Table 1).
To further estimate the selectivity for H 2 O 2 in the noncontact mode, the uorescence responses of the nanosystem toward various interfering species including cations, anions, and amino acids were investigated. As shown in Fig. 4e, only the addition of H 2 O 2 results in the recovery of the UCL emission, whereas no obvious change in luminescence intensity is observed aer the addition of large excesses of the other interfering species, such as Na + , K + , Ca 2+ , Mg 2+ , Zn 2+ , F À , Cl À , CO 3 2À , NO 3 À , SO 4 2À , cysteine (Cys), glutamine (Gln), glycine (Gly), leucine (Leu), proline (Pro), serine (Ser), threonine (Thr), and valine (Val). Furthermore, competition experiments exhibit the recovery in UCL intensities at 658 nm, performed by adding H 2 O 2 to MoO 3Àx NSs solutions containing other interfering species (Fig. 4f). The results indicate that the sensing of H 2 O 2 is barely affected by these coexistent species. Therefore, this system can serve as an upconversion uorescence nanoprobe for H 2 O 2 with high selectivity in the non-contact mode.

Application in real sample analysis
For a practical application of the non-contact upconversionbased sensor, we studied the detection of H 2 O 2 residue in contact lens solution, as H 2 O 2 is usually applied in the contact lens disinfection processes and is harmful to human eyes. The results are summarized in To quantitatively analyze the quenching ability of MoO 3Àx NSs on PEI-UCNPs, a series of MoO 3Àx NSs modied PEI-UCNPs nanocomposites (the concentration of PEI-UCNPs is xed at 0.5 mg mL À1 ) is prepared by changing the MoO 3Àx NSs content (from 0 to 0.4 mg mL À1 ). The overlap integral (J(l)) between the normalized emission spectrum of the donor (UCNPs) and the absorption spectrum of the acceptor (MoO 3Àx NSs) is dened by the equation as follows:  where l is the wavelength in nm, F D is the 980 nm laseractivated UCL spectrum of PEI-UCNPs normalized to an area of 1, 3 A is the extinction coefficient spectrum of MoO 3Àx NSs in units of M À1 cm À1 . The J(l) value for the donor acceptor pair is calculated to be 2.79 Â 10 13 M À1 cm À1 nm 4 . The effect of different MoO 3Àx NSs loading on PEI-UCNPs is evaluated by UCL spectra. As shown in Fig. 5a, the red emission intensity of UCNPs/MoO 3Àx nanoassemblies experiences a signicant decrease with the increasing addition of MoO 3Àx NSs. Additionally, the green emission intensity of upconversion-based nanoassemblies with different loading of MoO 3Àx NSs shows a similar tendency, but with a slower downward trend. As shown in Fig. 5b To elucidate the effect of EEA-induced reduction in the UCL of UCNPs by MoO 3Àx NSs, a non-contact mode is designed. The MoO 3Àx NSs solution (0.35 mg mL À1 ) is sealed in the quartz cuvette, aligning in front of another quartz cuvette containing 0.5 mg mL À1 PEI-UCNPs solution, and the UCL spectra are depicted in Fig. S7. † Upon the excitation by a 980 nm CW laser, the incident light rst passes through the MoO 3Àx NSs solution, and the energy-reduced light then reaches the UCNPs, resulting in the loss of intensity in UCL emissions. Ideally, the UCL intensity at 658 nm reduces by 72.3% compared with the control experiment.
Notably, the EEA effect will affect the intensity in all emissions, and the red-to-green emission ratio (R/G, where the red emission is integrated from 600-700 nm and the green emission is integrated from 500-600 nm) keeps its stability, which is conrmed by the activation of PEI-UCNPs (0.5 mg mL À1 ) with different power of the 980 nm laser. As presented in Fig. S8, † the red and green emission intensities increase with increasing laser power, and the R/G remains stable. However, a gradual decrease in the R/G values is observed for the UCNPs/MoO 3Àx nanocomposites with the increasing loading content of MoO 3Àx NSs (inset of Fig. 5a). This phenomenon can be attributed to the FRET-induced uorescence quenching by MoO 3Àx NSs, where the quenching ability by MoO 3Àx NSs in red emission is more pronounced than in the green region. As discussed above, the uorescence quenching of UCNPs by MoO 3Àx NSs is achieved by the joint effect of EEA and FRET, owing to the strong absorbance ability of MoO 3Àx NSs in both visible and NIR regions.
The sensing performance of UCNPs/MoO 3Àx nanoassemblies toward H 2 O 2 is investigated by UCL emission spectroscopy. As shown in Fig. 5c, the UCL emission intensity in red and green regions increases with the increasing addition of H 2 O 2 solution. As discussed above, the addition of H 2 O 2 leads to the oxidation of MoO 3Àx , resulting in the reduction in the absorption in both visible and NIR regions, and thus inhibiting the EEA effect at 980 nm and FRET process from the UCL of UCNPs to absorption of MoO 3Àx in the visible region, corresponding to the enhancement of UCL emission intensity. However, the uorescence intensity shows no obvious changes if more than 3.0 mM H 2 O 2 are added. The uorescence intensity at 658 nm exhibits a linear correlation to the H 2 O 2 concentration in the range of 0-0.8 mM (R 1 2 ¼ 0.990) and 1.0-2.5 mM (R 2 2 ¼ 0.996), respectively (Fig. 5d)

Conclusions
In summary, we have designed two different methods (i.e., a non-contact method and a conventional method) for upconversion uorescence sensing of H 2 O 2 . The non-contact method relies on the MoO 3Àx NSs absorption-induced EEA effect and operates by placing the MoO 3Àx NSs solution in front of UCNPs solution, whereas the conventional upconversion-based uorescence nanoprobe, based on the joint effect of EEA and FRET, was constructed by the integration of UCNPs and MoO 3Àx NSs via electrostatic interactions. An advantage of the noncontact method is that the valuable sensor particles do not become consumed or contaminated during the measurement and can be reused for a long time. The MoO 3Àx NSs act as the quencher in both nanosystems, owing to the strong absorptive capacity of MoO 3Àx in both visible and NIR regions. However, the addition of H 2 O 2 leads to the oxidation of MoO 3Àx , resulting in the recovery of UCL emissions, and thus enabling the quantitative detection of H 2 O 2 by both methods. Beneting from the non-contact method, hydrophobic OA-UCNPs can be applied as the luminophore directly and ultrahigh uorescence quenching (99.8%) is obtained. Moreover, the non-contact method exhibits high sensitivity toward H 2 O 2 down to 0.63 mM, which is lower than that determined by the spectrophotometry of MoO 3Àx (0.75 mM) and conventional UCNPs/MoO 3Àx nanocomposites (9.61 mM). Additionally, pH sensing can be achieved by employing the non-contact mode as well, which has shown a broad pH-responsive range from 2.6 to 8.2. We believe that these results could provide new insights into the design of upconversion-based nanosystems for uorescence sensing of other analytes.

Conflicts of interest
There are no conicts to declare.