Single-molecule photoreaction quantitation through intraparticle-surface energy transfer (i-SET) spectroscopy

Quantification of nanoparticle-molecule interaction at a single-molecule level remains a daunting challenge, mainly due to ultra-weak emission from single molecules and the perturbation of the local environment. Here we report the rational design of an intraparticle-surface energy transfer (i-SET) process, analogous to high doping concentration-induced surface quenching effects, to realize single-molecule sensing by nanoparticle probes. This design, based on a Tb3+-activator-rich core-shell upconversion nanoparticle, enables a much-improved spectral response to fluorescent molecules at single-molecule levels through enhanced non-radiative energy transfer with a rate over an order of magnitude faster than conventional counterparts. We demonstrate a quantitative analysis of spectral changes of one to four fluorophores tethered on a single nanoparticle through i-SET spectroscopy. Our results provide opportunities to identify photoreaction kinetics at single-molecule levels and provide direct information for understanding behaviors of individual molecules with unprecedented sensitivity.

for 0.5 h. After the methanol was evaporated at 100 ℃, the reaction solution was then heated to 300 ℃ and kept for 1 h under nitrogen flow to obtain core nanoparticles. Meanwhile, a cyclohexane dispersion containing the required amount of shell precursor nanoparticles (NaLuF4) was added with 1-octadecene (4 mL), followed by bubbling nitrogen to remove cyclohexane. Thereafter, the as-synthesized shell precursor nanoparticles in 1-octadecene (1 mL) were quickly injected and ripened for 12 min to yield core-shell nanoparticles. The injection of the shell precursor was repeated another three times (1 mL each) with a ripening cycle of 12 min. After the final ripening cycle, the solution was cooled to room temperature. The as-prepared nanoparticles were collected by centrifugation, washed with ethanol several times, and finally re-dispersed in cyclohexane (4 mL).

Synthesis of ligand-free nanoparticles.
Ligand-free nanoparticles were obtained according to a previously reported method 7 . In a typical procedure, 50 mg oleate-capped upconversion nanoparticles (UCNPs) were dispersed in a mixture of acetone (5 mL) and hydrochloric acid (0.6 mL, 12 M), followed by ultrasonication for 30 min to remove the oleic acid from the surface of UCNPs. The ligand-free nanoparticles were collected after centrifugation at 4000 rpm for 10 min, washed with acetone three times, and redispersed in methanol (1 mL).

Synthesis of BDP-decorated upcovnersion nanoparticles for ensemble characterizations.
In a typical experiment, ligand-free UCNPs (1.4 µM) and BDP TMR (1-80 µM) were mixed in a methanol solution (0.4 mL) and ultrasonicated for 10 min. The mixture was then kept in the dark overnight. After that, the BDP-decorated nanoparticles were collected by centrifugation (20000 rpm, 20 min), washed twice with acetone, and finally redispersed in methanol (0.25 mL). The exact loading concentration of BDP molecules was calculated by comparing the UV-VIS absorption spectra of BDP-loaded nanoparticles to a calibration curve.
Synthesis of BDP-decorated upcovnersion nanoparticles for single-particle spectroscopic characterizations. A cyclohexane solution of oleate-capped, core-shell nanoparticles (0.5 mL 1.1 uM) was mixed with a 2.5-mL THF solution containing BDP (1-40 µM) in a 25-mL flask. The mixture was heated to reflux at 50 ℃ for 2 h with vigorous stirring under nitrogen protection. The conjugates were collected by centrifugation (20000 rpm, 20 min), washed with ethanol twice and redispersed in cyclohexane (0.7 µM, 2.5 mL). The dye loading concentration can be controlled by adjusting the concentration of BDP TMR in the THF solution.
Sample preparation for single-particle TEM and SEM characterization. The TEM and SEM characterizations were carried out to prove monodispersity of BDP-UCNP conjugates. For TEM imaging, a stock solution of BDP-decorated NaYbF4:Tb@NaTbF4 nanoparticles was diluted to 10 pM with cyclohexane. The solution was then dropcasted on a 20-nm thick silicon nitride membrane that contains 9 windows with a size of 100 µm × 100 µm (AR010A, CleanSiN). Note that the same size is used for confocal scanning imaging. For SEM characterization, a diluted nanoparticle solution was dropped on a gold-coated silicon wafer (0.5 cm × 0.5 cm) and then airdried. The specimen was sputtered with Au to form an ultrathin coating to improve conductivity and contrast before being characterized by a field-emission scanning electron microscope (SU-8010).
Sample preparation of BDP-decorated UCNPs for single-nanoparticle microscopic imaging. In a typical procedure, the nanoparticle solution in cyclohexane was diluted to 2.5 pM. The diluted solution (20 uL) was then dropcasted onto a clean cover-glass (2 cm × 2 cm) and carefully rinsed using 20 uL cyclohexane. After cyclohexane evaporation, the as-prepared samples were imaged immediately under a confocal microscope.
Sample preparation of BDP-H2O2 decorated UCNPs for single-particle spectroscopic characterization. The procedure was identical to BDP-decorated UCNPs except using H2O2cyclohexane solution to dilute. In a typical procedure, a 100-µL solution of H2O2 was added into 1.5-mL cyclohexane and then vortexed for 10 s. Let stand for 5 min and extract the upper cyclohexane layer to be used for dilution of BDP-decorated UCNPs. The concentration of H2O2 can be determined by adding different amounts of H2O2-cyclohexane solution while diluting solutions.
Single-nanoparticle microscopy. Single-nanoparticle optical characterization was conducted on a home built confocal microscope optical stage with a Olympus 100X NA 1.30 oil objective and a 980 nm single mode fiber laser. A data acquisition code written in Matlab to form the confocal scanning iamge, and the sample was mounted on a high precision piezoelectric stage with 100 µm × 100 µm scanning area (Physik Instrumente, P-5613CD). Photoluminescence was recorded by the photon counting module (Excelitas SPCM-AQRH-14-FC34229), and emission spectrum were recorded on the spectrometer (Princeton Instruments, ProEM) equipped with a CCD camera (eXcelon3).
Characterization of UCNPs. X-ray diffraction (XDR) data were recorded on a LabX XRD-6000 with an ADDS wide-angle X-ray powder diffractometer (Cu Kα radiation, λ=1.54184 Å). Transmission electron microscopy (TEM) images were obtained from a JEM-2100F transmission electron microscope (JEOL) operating at an acceleration voltage of 200 kV. Energy-dispersive Xray spectroscopy (EDS) was carried out on an FEI Tecnai G2 F20 S-TWIN transmission electron microscope operated at an acceleration voltage of 200 kV. UV-vis absorption spectroscopic measurements were performed by a UV-vis spectrophotometer (UV-2600 Shimadzu). Luminescence spectra were measured by a fluorescence spectrometer (FLSP920, Edinburgh) equipped with a continuous-wave diode laser (980nm). The luminescence decay curves were measured by a phosphorescence lifetime spectrometer (FSP 920, Edinburgh) equipped with a TTL-mode modulated, 980-nm laser diode as the pulse-excitation source. For luminescence decay measurements, the effective lifetimes were determined by where I0 and I(t) represent the maximum luminescence intensity and luminescence intensity at time t after cut-off of the excitation light, respectively. Here, we take the NaYbF4:Tb(40 mol%)@NaTbF4 core-shell nanoparticle as an typical example to calculate the loading concentration of BDP per nanoparticle.

Calculation of UCNPs molar concentration:
In order to estimate the molar concentration of the nanoparticle, we assume that core-shell nanoparticle is a sphere. A hexagonal NaLnF4 unit cell contains 1.5 Na atom, 1.5 Ln atom and 6 F atom, which can be expressed as Na1.5Ln1.5F6. The size of a single nanoparticle can be obtained from TEM images. For NaYbF4:Tb(40 mol%)@NaTbF4 core-shell nanoparticle, the radius of core and core-shell are Rc=7.25 nm and Rcs=11.75 nm, respectively. Therefore, the volume ration of shell to core can be calculated as 3.257 in terms of Then, the single UNCP composition can be redeveloped as unit cell form, which is Herein, the relative molecular mass of Na1.5Yb0.2114Tb1.2886F6 can be calculated as 1 ! @ is Avogadro constant. In addition, the volume of a NaLnF4 unit cell can be obtained from JCPDS database, which is V !"## = 1.05 × 10 $%% cm & , so the density of the NaYbF4:Tb(40 mol%)@NaTbF4 core-shell nanoparticles can be calculated by The mass of a single core-shell nanoparticle can be obtained by To measure the mass concentration of UCNPs in the stock solution, UCNPs solution (V1= 50 µL) was dropcasted on a tared coverslip. After solvent evaporation, the mass of the nanoparticle was weighted. In the typical case, we get m1=1.8 mg. Herein, the molar concentration of the NaYbF4:Tb(40 mol%)@NaTbF4 nanoparticles in stock solution can be calculated by Calculation of loading concentration of BDP per nanoparticle: After conjugation of BDP with the nanoparticles, the absorbance of the BDP-nanoparticle conjugates at 542.5 nm was measured.
The exact loading concentration of BDP on upconversion nanoparticles was calculated by comparing the measured absorbance to a calibration curve function (Supplementary Figure 12).

Simulations on the energy transfer process
Simulation of direct energy transfer from low doping NaYF4:Yb,Er(18,2 mol%)@NaYF4 nanoparticles to BDP molecules For energy transfer from upconverison nanoparticles to dye molecules, we consider that each activator is an individual energy center and acts as an energy donor. Therefore, the energy transfer is a sum of interactions of every activator to all of the dye acceptors located on nanoparticle surfaces. When NaYF4:Yb,Er(18,2 mol%)@NaYF4 core-shell nanoparticles are used to transfer energy to BDP molecules, the energy donors are 2% of Er 3+ ions homogenously distributed in the core region. We assume that BDP dye molecules are randomly located on nanoparticle surfaces. In this case, we consider only the direct interaction between Er 3+ and BDP and without Er 3+ -Er 3+ interaction, because of the low doping concentration of the activators. Therefore, the luminescence intensity decay process of the nanoparticle can be expressed as follows: Eq. 1 where I(t) is the emission intensity of the nanoparticle energy donor recorded at time t, kD is the radiative rate of donor emission and kDA is energy transfer rate from donor to acceptor, respectively. Considering FRET for direct energy transfer process from the low doping nanoparticle to BDP, kDA=CDAr -6 , where CDA is the parameter relating to donor-acceptor interaction, r is the distance between donor and acceptor, and k0=1/ 0 ( 0 is the radiative lifetime of energy donor in the absence of dye acceptor).
We assume that the core-shell nanoparticle is a sphere. Herein, the distance distribution of the energy donors and acceptors follows a probability function P(r) 8 , then P(r)e -kt dr can be defined as the luminescence probability of donors locating in the region of [r, r+dr] away from the acceptor recorded at time t. Considering that there are n acceptors conjugated on nanoparticle surfaces, Eq. S1 can be further written as Eq. 2 which can be rewritten as Furthermore, we assume that the average number of acceptor molecules attached to each nanoparticle obeys Poisson distribution 9 and n particles are considered in our simulation. Then the total luminescence intensity can be described by the following equation Moreover, we can get that is Taylor's expansion item of exp(-Qµ). Therefore, Eq. 5 can be written as where µ is the average number of acceptor molecules attaching to each nanoparticle. or the expression of the direct energy transfer efficiency (Eff) by Eq. 7 For direct energy transfer model, we introduce a distance distribution function … Eq. 8 where Rc and R are the core and core-shell radius of the nanoparticles, respectively. By combining Eq. 4, 7, 8, the energy transfer efficiency Eff can be expressed by Eq. 9 For Eq. 6, we assume 1-Q(t) = kt, hence the theoretical simulation curve can be plotted as the blue line, shown in Fig. 2d in the main text. Moreover, the curve fitting shows that the entire energy transfer rate from the Er 3+ -doped nanoparticle to BDP is k≈80 s -1 . The whole process is schematically illustrated in Supplementary Figure 23.
In our simulation, the i-SET process is built on high contents of active dopants in core-shell nanoparticles. These active dopants contribute significantly to energy transfer to the surface with the assistance of fast donor-donor energy migration. Compared with the direct energy transfer model that neglects the effect of energy migration among donors, the derivation of kDA should consider additional donor interactions. Therefore, an average kDA in the form of D DA was approximately derived based on an energy migration and hopping model developed by Burshtein et al 10,11 . Accordingly, we derive where k(r) is the same constant as that in Eq. 4, τ1 denotes the average hopping time of energy migration. Similarly, we can deduce the energy transfer efficiency of the i-SET from Eq. 1 Eq. 11 The energy transfer is considered as Dexter's energy transfer for donor-donor and dipole-diploe interaction for donor-acceptor respectively. We suppose Tb 3+ ions are randomly distributed in the whole particle, thus P(r) can be expressed by 8 Eq.

12
R is the radius of the particle. Furthermore, we assume that exchange interaction (Dexter`s energy transfer) dominates the energy transfer between the donors (Tb 3+ ). Herein, 1/ 1 can be obtained as Eq. 13 Eq. 14 where CDD, CDA and LDD are parameters denoting donor-donor interaction, donor-acceptor interaction and the spatial-overlapping degree of donor-donor wave functions. Combining Eqs. S10, S12, S13, S14, D DA can be written as follow: Eq.15 As illustrated in Eq. 15, ND and µ are the only variables for any given donor-acceptor pairs. To verify the nature of donor-donor and donor-acceptor interactions in the i-SET process, we need to fit the Eq. 15 to the experimental results as a function of ND (while keeping the particle radium R and µ as constants). Hence, we conducted measurements for a series of NaYbF4:40% Tb@NaLuF4:X% Tb (X=10, 30, 40, 60, 70, 100) core-shell nanoparticles with diameter 23.64 ± 1.04 nm and BDP loading concentration of µ≈3.5. The upconversion luminescence lifetime of Tb 3+ at 547 nm was measured for these nanoparticles. The experimental kDA and Eff values can be obtained as Eq. 17 where DA and D are the Tb 3+ (donor) lifetime at 547 nm with and without BDP, respectively.
As the experimental results of D DA match with Eq. S15 approximately, we can derive best-fit parameters as follows: LDD=0.03nm, CDA=3.03×10 -50 m 6 s -1 , CDD=2.47×10 17 s -1 . Combining Eq. S11 and S15, the energy transfer efficiency Eff can be expressed as Eq. 18 By taking the above-fitted parameters into the Eq.18, we can derive energy transfer efficiency as a function of the number of BDP molecules per nanoparticle (µ) or particle radius (R) (see Fig. 2d black line). Since we have determined that donor-donor energy migration is through exchange interaction, then the migration rate can be calculated by the following equation

Supplementary
Supplementary Figure 17. Optical characterization of 24-nm NaYbF4:Tb(40 mol%)@NaTbF4 core-shell nanoparticles conjugated with BDP molecules at different concentrations. a, Emission spectra of the core-shell nanoparticles conjugated with varied concentrations of BDP. b, Corresponding Tb 3+ -emission (547 nm) decay curves of the samples shown in a. Figure 18. Normalized UV-vis absorption (Abs.) and emission (Em.) spectra of BDP molecules and upconversion luminescence spectrum of NaYF4:Yb,Er(18,2 mol%)@NaYF4 (Yb/Er@Y) core-shell nanoparticles. The substantial overlapping between the emission spectrum of Er 3+ and the absorption spectrum of BDP TMR supports the analysis of the energy transfer mechanism in BDP-conjugated Y/Er@Y systems. Figure 19. Luminescence decay dynamics showing the change in upconversion luminescence lifetime as a function of the number of BDP molecules per particle. a, Luminescence decay curves of Tb 3+ emission at 547 nm recorded for 24 nm NaYbF4:Tb@NaTbF4 nanoparticles. b, Luminescence decay curves of Er 3+ emission at 541 nm, recorded for 24 nm NaYF4:Yb,Er(18,2 mol%)@NaYF4 core-shell nanoparticles.

Supplementary
Supplementary Figure 20. a-d, TEM images and e-h, corresponding size distribution of NaYbF4:Tb core and NaYbF4:Tb@NaTbF4 core-shell nanoparticles with different sizes. The size distributions of the nanocrystals were calculated by counting > 200 particles recorded in TEM images. Figure 22. Comparison of measured luminescence decay curves of nanoparticles with different sizes loaded with the same amount of BDP molecules. a, Upconversion luminescence decay curves of Tb 3+ emission at 547 nm. b, Upconversion luminescence decay curves of Er 3+ emission at 541 nm. All the nanoparticles have the identical dye loading number of 3 dye molecules per particle. Figure 25. Excitation power-dependent upconversion luminescence intensity measured for a, a single NaYbF4:Tb(40 mol%)@NaTbF4 core-shell nanoparticle under confocal microscopy, and b, BDP-modified NaYbF4:Tb(40 mol%)@NaTbF4 nanoparticles dispersed in methanol. Figure 26. A typical confocal upconverted luminescence image of pristine NaYbF4:Tb(40 mol%)@NaTbF4 core-shell nanoparticles under 980 nm excitation. Figure 27. Time-course plot of luminescence intensity recorded from a single NaYbF4:Tb(40 mol%)@NaTbF4 nanoparticle, exhibiting exceptional photostability. Figure 33. Single-particle luminescence spectra of NaYbF4:Tb(40 mol%)@NaTbF4 nanoparticles loaded with average 52 BDP per nanoparticle. The spectra (a-l) were recorded from randomly picked luminescence spots by confocal scanning imaging. The dashed, red circles highlight the emission spectra of BDP dye molecules.

Supplementary
Supplementary Figure 34. Single-particle luminescence spectra of NaYbF4:Tb(40 mol%)@NaTbF4 nanoparticles loaded with average 92 BDP per nanoparticle. The spectra (a-l) were recorded from randomly picked luminescence spots by confocal scanning imaging. The dashed, red circles highlight the emission spectra of BDP dye molecules. Figure 35. Upconversion luminescence characterizations of discrete NaYF4:Yb,Er(18,2 mol%)@NaYF4 nanoparticles. a, Confocal luminescence imaging of a nanoparticle-dispersed sample specimen under 980 nm excitation. b, Upconversion luminescence spectra of 9 individual nanoparticles, corresponding to luminescent spots shown in a. Note that spots 1 to 8 show similar intensities both in the confocal scanning image and the emission spectra. The intensity of spot 9 is two times higher than other spots, indicating the existence of a particle dimer at spot 9. Figure 36. Microscopy imaging of discrete BDP-decorated NaYF4:Yb,Er(18,2 mol%)@NaYF4 nanoparticles. a, Confocal upconversion luminescence imaging of the sample under 980 nm excitation. b, Corresponding upconversion emission spectra of 6 BDP-nanoparticle conjugates shown in a. Note that no sensitized dye emission is detected in these nanoparticles, albeit the high loading number of dye molecules (95 BDP per particle). Figure 37. Tracking single-molecule photobleaching by upconversion nanoprobes. Time-correlated, single-particle emission spectra of BDPdecorated NaYbF4:Tb(40 mol%)@NaTbF4 nanoparticles (~0.9 BDP per particle) recorded from individual nanoparticles, showing distinguishable stepwise photobleaching of BDP emission containing (a-c) 3 BDP molecules, (d-e) 2 BDP molecules and (f) 1 BDP molecule. Figure 38. Tracking single-molecule photobleaching by upconversion nanoprobes. Time-correlated single-particle emission spectra of BDP-decorated NaYbF4:Tb(40 mol%)@NaTbF4 nanoparticles (~0.9 BDP per particle) from 9 randomly picked nanoparticles, showing distinguishable stepwise photobleaching from individual nanoparticles containing (a-c) 1 BDP molecule, (d-e) 2 BDP molecules, (fg) 3 BDP molecules, and (h-i) 4 BDP molecules. Note that 450 nM H2O2 was added to the stock solution of BDP-decorated NaYbF4:Tb@NaTbF4 nanoparticles before dropcasting the sample. time channels, and the histogram represents the dynamic fluorescence change as a function of time (Supplementary Figure 39a).

Supplementary
In our experiments, we first measured the single-molecule photobleaching of individual nanoparticles by examining the stepwise quenching of BDP luminescence under continuous laser irradiation, as illustrated in Supplementary Figure 37 and 38. For every one minute, we captured a luminescence spectrum to monitor the intensity dropping of BDP. We recorded the events of the stepwise quenching for an individual nanoparticle until the emission of BDP was completely quenched. The measurement cycle was repeated on several randomly picked single nanoparticles until enough number of quenching events were recorded. By gathering all of the quenching events from a large number of individual measurements, we can count the number of survived dye molecules versus irradiation time. This allows us to generate a time-dependent histogram reflecting the reaction kinetics of single-molecule photobleaching on upconversion nanoparticles (Supplementary Figure 39b).