Materials
Thulium(III) acetate hydrate (99.9%), gadolinium(III) acetate hydrate (99.9%), ytterbium(III) acetate hydrate (99.9%), yttrium(III) acetate hydrate (99.9%), terbium(III) acetate hydrate (99.9%), europium acetate hydrate (99.9%), diethylene glycol (DEG, 98%), Poly-L-lysine solution (0.1 wt%), N-(3-dimethylaminopropyl)-Nʹ-ethylcarbodiimide hydrochloride (EDC, ≥ 98%) and N-hydroxysuccinimide (NHS, 98%) were purchased from Sigma-Aldrich. Sodium hydroxide (NaOH, ≥98%), ammonium fluoride (NH4F, ≥ 99.99%), 1-octadecene (ODE, ≥ 90%(GC)), oleic acid (OA, AR), cyclohexane (AR, 99.5%) and ammonia solution (NH3·H2O, 25–28%) were purchased from Aladdin®, China. Methanol (AR), ethanol (AR), ether (AR), hydrochloric acid (HCl, AR, 36–38%) and dimethylsulfoxide (DMSO, AR) were purchased from Sinopharm Chemical Reagent Co., China. Poly(acrylic acid) (PAA, M.W~2000) was purchased from Shanghai Macklin Biochemical Co., Ltd. Compound 2-(N-morpholino)ethanesulfonic acid (MES, 0.1 M, pH = 6.0) was purchased from Leagene Biotechnology co., Beijing. Phosphate-buffered Saline (PBS), ProLong™ Gold Antifade Mountant, Alexa Fluor™ Plus 405 Phalloidin (A30104, Invitrogen™), organic fluorophore Texas Red (>90% single isomer, T6256, Invitrogen™) were purchased from Thermo Fisher Scientific. Amino-phalloidin was purchased from Shanghai Maokang Biotechnology Co. Ltd., China. Paraformaldehyde (PFA, 4%) was purchased from Invitrogen. Triton X-100 was purchased from Amresco. Quick Block blocking buffer, bovine serum albumin (BSA, ≥ 99%, fatty acid & IgG free, BioPremium) and ethanesulfonic acid (HEPES) buffer (0.01 M, pH = 7.3) were purchased from Beyotime, China. HeLa cells source was provided by Sun Yat-Sen University (Guangzhou, China). All the reagents were used as received without any further purification.
Synthesis of 11-nm NaGdF4:Yb/Tm nanoparticles
Lanthanide-doped nanoparticles were synthesized using previously reported protocols with modifications47,48,49. To a 100-mL round-bottom flask, a mixture of 10 mL OA and 15 mL ODE was added, followed by addition of a 5-mL stock solution of Ln(CH3CO2)3 (Ln = Gd/Yb/Tm) (0.2 M). The mixture was heated to 150 °C and kept for 40 min under stirring to form a lanthanide-oleate precursor solution, which was slowly cooled to room temperature before a NaOH-methanol stock solution (1 M, 2.5 mL) and an NH4F-methanol stock solution (0.4 M, 10 mL) were pipetted into the flask. Next, the mixture was kept at 40 °C under stirring for an additional 2 h. Subsequently, the solution was heated to 100 °C for 30 min under a vacuum to evaporate the methanol. After removing methanol, the solution temperature was further elevated and kept at 300 °C under stirring for 1.5 h in an argon atmosphere for nanoparticle growth. After the reaction, the solution was cooled to room temperature naturally, precipitated by adding 15-mL ethanol, and then centrifugated at 7500 r.p.m. (relative centrifuge force (RCF) = 5840 × g) for 5 min. The obtained NaGdF4:Yb/Tm nanoparticles were washed with ethanol and cyclohexane three times and then re-dispersed in 8-mL cyclohexane. NaGdF4:Yb/Tb nanoparticles were synthesized using the same protocol. Note that the mixture volumes of OA and ODE were adjusted to 7.5 and 17.5 mL, respectively, for the NaYF4 host matrix.
Synthesis of 8-nm NaGdF4:Yb/Tm nanoparticles
The synthetic procedure of 8-nm NaGdF4:Yb/Tm is similar to that of 11-nm NaGdF4:Yb/Tm nanoparticles. While keeping the volumes and ratio of Ln(CH3CO2)3 unchanged, the volumes of OA and ODE were changed to 7 and 15 mL. In addition, the reaction temperature was adjusted from 300 to 260 °C, and the reaction time shortened from 1.5 to 1 h.
Synthesis of 15-nm NaGdF4:Yb/Tm nanoparticles
NaGdF4:Yb/Tm nanoparticles of 15 nm in diameter were obtained by growing the same layer onto the as-prepared 11-nm core NaGdF4:Yb/Tm nanoparticles10. First, a stock solution of 10-mL OA, 15-mL ODE and 5-mL Ln(CH3CO2)3 (Ln = Gd/Yb/Tm) (0.2 M) was added into a 100-mL round-bottom flask. The reaction solution was heated to 150 °C and kept for 40 min under stirring to form a clear solution, and then cooled to 80 °C. An 8-mL suspension of the as-prepared 11-nm core NaGdF4:Yb/Tm nanoparticles was pipetted into the flask (the core-to-shell volume ratio, 1:1) and maintained at this temperature for 30 min to remove cyclohexane. Subsequently, the solution was cooled slowly to room temperature, and then 2.5-mL NaOH-methanol stock solution (1.0 M) and 10 mL NH4F-methanol stock solution (0.4 M) were pipetted into the flask. The following steps were the same as those used to synthesize bare core 11-nm NaGdF4:Yb/Tm nanoparticles. The obtained 15-nm NaGdF4:Yb/Tm nanoparticles were re-dispersed in 8-mL cyclohexane.
Synthesis of 17-nm NaGdF4:Yb/Tm@NaYF4 core-shell nanoparticles
To epitaxially grow an inert NaYF4 shell, 7.5-mL OA, 17.5-mL ODE and 5-mL Y(CH3CO2)3 stock solution (0.2 M) were added into a 100-mL round-bottom flask containing 15-nm NaGdF4:Yb/Tm nanoparticles. The core-to-shell volume ratio is 1:1. The obtained 17-nm NaGdF4:Yb/Tm@NaYF4 nanoparticles were re-dispersed in 8-mL cyclohexane.
Synthesis of 28-nm NaGdF4:Yb/Tm@NaYF4 core-shell nanoparticles
The 28-nm NaGdF4:Yb/Tm@NaYF4 core-shell nanoparticles were obtained using the same procedure as 17-nm NaGdF4:Yb/Tm@NaYF4 core-shell nanoparticles, except for the change in the core-to-shell volume ratio from 1:1 to 1:4. A 2-mL suspension of the as-prepared 11-nm NaGdF4:Yb/Tm nanoparticles was added. The obtained 28-nm NaGdF4:Yb/Tm@NaYF4 nanoparticles were re-dispersed in 4-mL cyclohexane.
Synthesis of 38-nm NaGdF4:Yb/Tm@NaYF4 nanoparticles
To further increase NaYF4 shell layer thickness, another NaYF4 layer was epitaxially grown onto the as-prepared 28-nm NaGdF4:Yb/Tm@NaYF4 nanoparticles using the standard core-shell synthesis. The core-to-shell volume ratio is 1:2, with the addition of a 4-mL suspension of 28-nm NaGdF4:Yb/Tm@NaYF4 nanoparticles. The obtained 38-nm NaGdF4:Yb/Tm@NaYF4 nanoparticles were re-dispersed in 4-mL cyclohexane.
Preparation of nanoparticle-coated slides for spectroscopic study
To perform spectroscopic and emission depletion characterizations, the sample cover slides with well-dispersed nanoparticles (10.0 μL, 200.0 μg/mL) were prepared by spin-coating (6000 r.p.m, RCF = 3740 × g). The employed aqueous sample was transferred directly from hydrophobic cyclohexane before spin-coating according to a previously reported hydrochloric acid treatment50. After air-drying, the sample coverglass was placed on a clean glass slide spread with a small drop of embedding medium (ProLong™ Gold Antifade Mountant, Invitrogen) to ensure refractive index matching. Air bubbles were gently extruded and the fringe was properly sealed with nail polish. The sample was kept at room temperature and in the dark for another 10 h to ensure complete drying before measurement.
Sample preparation for single-nanoparticle imaging
To perform single-nanoparticle imaging, sample slides coated with lanthanide-doped nanoparticles were prepared according to literature methods10,24. To prepare a sample slide, a high precision microscope cover glass (thickness No.1.5H, 170 ± 5 μm, Deckgläser, Paul Marienfeld GmbH) was washed with anhydrous ethanol/ultrapure water under ultrasonication and then treated with 60-μL poly-L-lysine solution (0.1% in H2O w/v). After about half an hour, the poly-L-lysine was washed off with ultrapure water, and the cover glass was air-dried. A 20.0-μL nanoparticle solution (0.1 μg/mL in cyclohexane) was added to the treated surface, immediately followed by wash three times with cyclohexane. After air-drying, the cover glass was placed on a clean glass slide spread with a small drop of embedding medium. Air bubbles were extruded and the glasses were sealed with nail polish. The sample was kept at room temperature for another 10 h to ensure complete drying before measurement.
Preparation of gold nanoparticle sample slides
To measure the PSF of the lasers, sample slides with dispersed gold nanoparticles (CTAB-capped, 120 nm in diameter, NS-120-50, NanoSeedz) were prepared. A 20.0-μL water solution of gold nanoparticle (ten times diluted from the original solution) was directly spin-coated onto a cover slide. After air-drying, the cover slide was placed on a clean glass slide spread with a small drop of embedding medium (ProLong™ Gold Antifade Mountant, Invitrogen). Air bubbles were extruded and the glasses were sealed with nail polish. The sample was kept at room temperature for another 10 h to ensure complete drying before measurement.
Preparation of Texas Red-stained slides
According to a literature method51, the sample for single-molecule two-photon microscopic imaging was prepared by spin coating of a DMSO solution of Texas Red molecules (20.0 μL, 2 × 10−7 M) at 6000 r.p.m (RCF = 3740 × g) onto a cover slide. After air-drying, the cover glass was placed on a clean glass slide spread with a small drop of embedding medium (ProLong Gold Antifade Mountant) to relieve the photobleaching of Texas Red. Air bubbles were also avoided and the glasses were sealed with nail polish. The sample was kept at room temperature and in the dark for another 10 h to ensure complete dryness before measurement.
Preparation of PAA-coated lanthanide-doped nanoparticles
PAA-coated nanoparticles of 8-nm in diameter for biomarkers were prepared using a general ligand exchange method52, and the procedures are similar to our previous work25. To a 100-mL round-bottom flask, 20 mL of DEG and 0.2 g of PAA were added. The mixture was heated to 110 °C and kept at this temperature for 1 h under an argon atmosphere. Next, 2 mL of OA-nanoparticles were injected into the solution and maintained at the same temperature for 30 min under vacuum to evaporate cyclohexane. When the cyclohexane was evaporated, the solution was heated to 240 °C and kept at this temperature for 3 h under an argon atmosphere. After the solution was cooled to room temperature naturally, 15 mL of 1% HCL were added and the solution was continued to stir for 20 min. The obtained PAA-coated nanoparticles were precipitated by centrifugation for 30 min (18,000 r.p.m (RCF = 33,660 × g)), then washed with ethanol and ultrapure water (v/v = 1:1) three times, and re-dispersed in 2-mL ultrapure water.
Activation of the carboxyl groups of PAA-coated lanthanide-doped nanoparticles
Following the commonly used methods53,54, the protocol is similar to our previous work25. Firstly, EDC and NHS solids were dissolved in MES solution at a concentration of 0.2 and 0.3 mg μL−1, respectively. Then 1 mL of PAA-coated nanoparticle solution (1 mg mL−1), 20 μL of EDC and 20 μL of NHS were added into a 1.5-mL centrifuge tube and sonicated for 30 s. The mixture was stirred for 2 h at room temperature, then centrifugated for 30 min to remove the redundant EDC and NHS. (18,000 r.p.m (RCF = 33,660 × g)). The precipitate was collected and re-dispersed in 1 mL ultrapure water. The pH value of the solution was adjusted to 7.2–7.5 by adding an ammonia solution (2 M).
For preparing the Phalloidin conjugated 8-nm nanoparticles55, 10 μL amino-Phalloidin DMSO stock solution (100 μM) and 110 μL HEPES buffer were added to 30 μL activated PAA-lanthanide nanoparticles (1.0 mg mL−1). The mixture was stored in a refrigerator at 4 °C for another 4 h. Followed by centrifugation at 18,000 r.p.m (RCF = 33,660 × g). for 20 min to remove the excess of amino-Phalloidin, the obtained precipitate was re-dispersed into 150 μL HEPES buffer.
Labeling of cytoskeleton actin filaments using lanthanide-doped nanoparticles
These procedures followed the commonly used cell labeling method19,53, similar to our previous work25. The HeLa cells were cultured in 96-well glass-bottom plates (thickness 170 ± 5 μm, P96-1.5H-N, Cellvis) with 15,000 cells in each unit, with the temperature set at 37 °C, the humidity at 95%, and the CO2 concentration at 5% overnight (>12 h). The cells were rinsed three times using 200-μL pre-warmed PBS buffer (37 °C, 5 min each) to remove the residual culture media and then fixed using PFA for 15 min at room temperature. Subsequently, the cells were rinsed three times using PBS buffer to remove the residual PFA. Next, the cells were permeabilized in 200-μL 0.2% Triton X-100 for 30 min at room temperature and rinsed three times using HEPES buffer. 200 μL Quick Block blocking buffer was added to block the fixed cells for 20 min at room temperature. When the procedure of blocking was completed, 150 μL of freshly prepared phalloidin-bioconjugated lanthanide-doped nanoparticles (with 1% bovine serum albumin in buffer (200 μg mL−1) were added. The cells were kept at room temperature for 4 h and then rinsed twice with pre-warmed (37 °C) HEPES buffer. Finally, the HEPES buffer was removed and a small drop of embedding medium was dropped into the well. The sample was kept at room temperature and in the dark for 24 h to ensure complete dryness before measurement.
Labeling of cytoskeleton actin filament using Alexa Fluor Plus 405 Phalloidin
To prepare the stock solution of Alexa Fluor Plus 405 Phalloidin, the contents of the vial were dissolved in 150 μL of anhydrous DMSO to yield a 300 × stock solution, which is equivalent to approximately 66 μM. The staining solution was prepared by diluting 0.5 μL of the 300 × stock solution into 150 μL of PBS and adding 1% BSA. When fixation, permeabilization, and blocking were completed, 150 μL of the freshly prepared staining solution was added. The cells were kept at room temperature for 4 h and then rinsed twice with pre-warmed (37 °C) PBS buffer.
Transmission electron microscopy (TEM)
The size and morphology of as-prepared nanoparticles were obtained using a transmission electron microscope JEM-2100HR, JEOL (200 kV). The bioconjugated lanthanide-doped nanoprobes were diluted and dispersed in ultrapure water, whereas other samples were diluted and dispersed in cyclohexane, then dropped onto the surface of copper grids for the test of TEM.
Optical system for spectroscopic study and emission depletion measurements
A lab-made optical system coupled with two laser beams was built to implement spectroscopic measurement, luminescence lifetime measurement, and super-resolution imaging. As the optical system’s layout shown in Supplementary Fig.1, for the spectroscopic study of Tm3+-activated nanoparticles, the excitation 975-nm CW laser beam was generated by a single-mode diode laser (Laser 1, B&A Technology Co. Ltd.). The depletion 730-nm CW laser beam was generated by a Ti:sapphire laser (Laser 2, Mira 900, Coherent). Two band-pass filters (F1, LD01-975/10-12.5, Semrock; F2, FF01-730/39-25, Semrock) were used to clean the spectrum of the two lasers. Two pairs of half-wave plates and polarization beam splitters were used to control the laser beam power. In the optical excitation pathway, the excitation laser beam was firstly expanded by a pair of lenses with 25-mm and 100-mm focus lengths (L1, L2), then optimized by a spatial filter which consisted of a pair of 50-mm focus length lenses (L3, L4) and a 25-μm gold-coated pinhole (PH1). Half-wave plate (HWP2) and quarter-wave plate (QWP1) were used to further regulate the profile of excitation beam. In the optical depletion pathway, the depletion laser beam was firstly optimized by a spatial filter which consisted of a pair of 50-mm focus length lenses (L5, L6) and a 25-μm gold-coated pinhole (PH2), then optimized and shrank by another spatial filter which consisted of a pair of lenses with 50-mm and 25-mm focus lengths (L7, L8) and a 50-μm gold-coated pinhole (PH3). Two diaphragms (D2, D3) were placed in front and behind the second spatial filter, respectively. Half-wave plate (HWP5) and quarter-wave plate (QWP2) were used to further regulate the profile of depletion beam. The positions of lenses L4 and L8 were controlled with three-dimensional displacement stages. During the spectroscopic study, the sizes of Gaussian excitation beam and Gaussian depletion beam were modulated to be the same. The two laser beams were spatially superimposed using a 950-nm short-pass dichroic mirror (DM1, ZT950spxrxt, Chroma), and multiple mirrors in the optical path were used to calibrate the optical system. Then the coupled two beams were further filtered by a 715-nm long-pass emission filter (F3, FF01-715/LP-25, Semrock) and then directed into an commercially available multiphoton laser scanning microscope (FV1000MPE-S with motorized inverted IX81, Olympus), where two laser beams were reflected by a 690-nm short-pass dichroic mirror (DM2), then focused on the samples through an oil-immersed objective (OL, 100 × /NA = 1.45, UPLXAPO100XO, Olympus). These two laser beams were carefully aligned to ensure precise 3D overlap of their PSFs in XYZ directions. The upconversion emission was collected by the same objective and also filtered by a 694-nm short-pass emission filter (F4, FF01-694/sp-25, Semrock). A spectrometer (QE65Pro, Ocean Optics) was used to collect the emission spectra. For the spectroscopic study of Tb3+-activated nanoparticles, The depletion 756−nm CW laser beam was also generated by a Ti:sapphire laser (Laser 2, Mira 900, Coherent) and the filter was changed accordingly. All measurements were performed under ambient, strict light-shielding conditions.
Conventional upconversion laser-scanning imaging and super-resolution imaging
For microscopic imaging, the multiphoton laser scanning microscope (FV1000MPE-S with motorized inverted IX81, Olympus) was configurated with both Galvanometer scanning mirrors unit and non-descanned detection (NDD) module with a high-sensitivity photomultiplier tube (PMT1) employed as a single-photon counting detector for fluorescence imaging (Supplementary Fig.1). In addition, the system was also configured with differential interference contrast (DIC) components for transmission bright field imaging (PMT2), and a single 730 nm beam was used to perform DIC bright-field cell imaging. For laser scanning imaging of Tm3+-activated nanoparticles, in front of the PMT1, a 720-nm short-pass filter (F5, FF01-720/SP-25, Semrock), a 665-nm short-pass filter (F6, FF01-665/SP-25, Semrock), a 458-nm long-pass dichroic mirror (DM3, ZT458rdc, Chroma) and a 480-nm band-pass filter (F7, ET480/20x, Chroma) were used to block the laser light and yield the 475-nm emission of interest. To implement super-resolution microscopy, a 730-nm vortex phase plate (VR1-730-SP, LBTEK) was placed in the depletion optical path to generate the doughnut-shaped point spread function (PSF) for 730 nm beam. A diaphragm (D1) was placed in front of the VPP. HWP5 and QWP2 were precisely modulated to regulate the polarization state for a high-quality zero center. The Gaussian excitation beam and doughnut-shaped depletion beam were aligned to ensure precise 3D overlap of their PSFs in XYZ directions.
For imaging the scattered/reflected light from gold nanoparticles, the 690-nm short-pass dichroic mirror (DM2) was replaced with a 950-nm short-pass dichroic mirror (ZT950spxrxt, Chroma) for the 975-nm excitation beam or a 700-nm short-pass dichroic mirror (T700spxr-UF3, Chroma) for the 730-nm depletion beam, respectively. Filters F4, F5, F6, F7 and dichroic mirror DM3 were removed, instead, a 715-nm long-pass emission filter (FF01-715/LP-25, Semrock) was placed in front of the PMT1. For the two-photon excitation microscopic imaging of Texas Red, the excitation 830-nm femtosecond pulsed laser was generated by the mode-locked Ti:sapphire laser (Laser 2, Mira 900, Coherent). Filters F6, F7, and dichroic mirror DM3 were removed to pass the emission spectrum (590–690 nm) of Texas Red. For the two-photon excitation microscopic imaging of fixed Hela cells stained with Alexa Fluor Plus 405 Phalloidin, the femtosecond pulsed wavelength was shifted to 850 nm. Filters F6, F7, and dichroic mirror DM3 were also removed accordingly.
Luminescence lifetime measurement
To measure the emission lifetime of lanthanide-doped nanoparticles, an optical chopper (SR540, Stanford Research Systems, Inc.) was used to modulate the 975-nm excitation beam (Supplementary Fig.1). The trigger signal from the chopper and the detected photon counts by the PMT1 were synchronized with a time-correlated single-photon counting (TCSPC) system (NanoHarp 200, PicoQuant). The modulation frequency was set to be 1 kHz for the measurements. To minimize the influence of the instrument response function, for example, the falling edge of the mechanically modulated light pulse, the beam was suitably focused (Supplementary Fig.1) and the chopper was placed in the focal plane of L1. In addition, the instrument response function was considered and measured with a Texas Red ensemble sample with a lifetime on the order of nanoseconds.
Power density determination
An optical power meter (PM100D, Thorlabs) with microscope slide power meter sensor head (S170C, Si, Thorlabs) was used to determine the laser power. The power output from the objective lens was directly measured by attaching the sensor head to the front of the oil-immersed objective lens. During the experiment, the depletion power was recorded by the power meter by using an uncoated pellicle beam splitter to reflect a small fraction of the laser beam (~5%). The ratio of the laser powers measured outside the microscope and that output from objective lens were calculated to determine the power during imaging or spectroscopy experiments. To calculate the power density of Gaussian beam, the diameter of the laser focus spot was defined as the full width at 1/e2 (13.5%) of maximum, and the laser intensity was calculated by dividing the power at objective lens by the area of laser spot. Similar to the calculation of Gaussian beam, the area where the intensity exceeds 1/e2 (13.5%) of the maximum was considered as the effective area of doughnut-shaped beam.
Numerical simulation of the emission depletion in NaGdF4:Yb/Tm with SMED mechanism
According to the energy transfer processes in Yb3+/Tm3+-codoped system, the rate equation set of each energy state can be used to describe the dual-beams excitation/depletion optical process. Theoretical modeling can be developed to elucidate the proposed SMED depletion mechanism in our work. The corresponding energy diagram is shown in Supplementary Fig.6.
$${{{{{{\rm{Tm}}}}}}}^{3+}\Big({}^{3}{{{{{\rm{H}}}}}}_{6}\Big):\frac{d{n}_{0}}{dt}=-{\sum }_{i=1}^{7}\frac{d{n}_{i}}{dt}$$
(1)
$${{{{{{\rm{Tm}}}}}}}^{3+}\Big({}^{3}{{{{{\rm{F}}}}}}_{4}\Big) : \frac{d{n}_{1}}{dt}={\beta }_{2}{n}_{2}+{c}_{1}{n}_{5}{n}_{3}+{c}_{2}{n}_{3}{n}_{5}+2{c}_{3}{n}_{3}{n}_{0} \\+{b}_{51}\frac{{n}_{5}}{{\tau }_{5}}+{b}_{61}\frac{{n}_{6}}{{\tau }_{6}}+{b}_{71}\frac{{n}_{7}}{{\tau }_{7}}-{w}_{2}{n}_{{{{{{\rm{Yb}}}}}}1}{n}_{1}-\frac{{n}_{1}}{{\tau }_{1}}$$
(2)
$${{{{{{\rm{Tm}}}}}}}^{3+}\Big({}^{3}{{{{{\rm{H}}}}}}_{5}\Big):\frac{d{n}_{2}}{dt}={\beta }_{3}{n}_{3}+{w}_{1}{n}_{{{{{{\rm{Yb}}}}}}1}{n}_{0}+{b}_{52}\frac{{n}_{5}}{{\tau }_{5}}+{b}_{62}\frac{{n}_{6}}{{\tau }_{6}}-{\beta }_{2}{n}_{2}$$
(3)
$${{{{{{\rm{Tm}}}}}}}^{3+}\Big({}^{3}{{{{{\rm{H}}}}}}_{4}\Big):\frac{d{n}_{3}}{dt}={\beta }_{4}{n}_{4}+{b}_{73}\frac{{n}_{7}}{{\tau }_{7}}+{c}_{4}{n}_{6}{n}_{0}-{c}_{1}{n}_{5}{n}_{3}-{c}_{2}{n}_{3}{n}_{5}\\ - {c}_{3}{n}_{3}{n}_{0}-{w}_{3}{n}_{{{{{{\rm{Yb}}}}}}1}{n}_{3}-{\beta }_{3}{n}_{3}-\frac{{n}_{3}}{{\tau }_{3}}$$
(4)
$${{{{{{\rm{Tm}}}}}}}^{3+}\Big({}^{3}{{{{{\rm{F}}}}}}_{3}/{}^{3}{{{{{\rm{F}}}}}}_{2}\Big):\frac{d{n}_{4}}{dt}={\beta }_{5}{n}_{5}+{b}_{74}\frac{{n}_{7}}{{\tau }_{7}}+{c}_{4}{n}_{6}{n}_{0}+{w}_{2}{n}_{{{{{{\rm{Yb}}}}}}1}{n}_{1}-{\beta }_{4}{n}_{4}$$
(5)
$${{{{{{\rm{Tm}}}}}}}^{3+}\Big({}^{1}{{{{{\rm{G}}}}}}_{4}\Big):\frac{d{n}_{5}}{dt}={\beta }_{6}{n}_{6}+{b}_{75}\frac{{n}_{7}}{{\tau }_{7}}+{w}_{3}{n}_{{{{{{\rm{Yb}}}}}}1}{n}_{3}-{c}_{1}{n}_{5}{n}_{3}\\ - {c}_{2}{n}_{3}{n}_{5}-{\beta }_{5}{n}_{5}-\frac{{n}_{5}}{{\tau }_{5}}+\frac{{\sigma }_{{{{{{\rm{d}}}}}}}^{se}{I}_{{{{{{\rm{d}}}}}}}}{h{v}_{{{{{{\rm{d}}}}}}}}{n}_{7}-\frac{{\sigma }_{{{{{{\rm{d}}}}}}}^{a}{I}_{{{{{{\rm{d}}}}}}}}{h{v}_{{{{{{\rm{d}}}}}}}}{n}_{5}$$
(6)
$${{{{{{\rm{Tm}}}}}}}^{3+}\Big({}^{1}{{{{{\rm{D}}}}}}_{2}\Big):\frac{d{n}_{6}}{dt}={\beta }_{7}{n}_{7}+{c}_{1}{n}_{5}{n}_{3}+{c}_{2}{n}_{3}{n}_{5}-{c}_{4}{n}_{6}{n}_{0}\\ - {w}_{4}{n}_{{{{{{\rm{Yb}}}}}}1}{n}_{6}-{\beta }_{6}{n}_{6}-\frac{{n}_{6}}{{\tau }_{6}}$$
(7)
$${{{{{{\rm{Tm}}}}}}}^{3+}\Big({}^{1}{{{{{\rm{I}}}}}}_{6}\Big):\frac{d{n}_{7}}{dt}={w}_{4}{n}_{{{{{{\rm{Yb}}}}}}1}{n}_{6}-{\beta }_{7}{n}_{7}-\frac{{n}_{7}}{{\tau }_{7}}-\frac{{\sigma }_{{{{{{\rm{d}}}}}}}^{se}{I}_{{{{{{\rm{d}}}}}}}}{h{v}_{{{{{{\rm{d}}}}}}}}{n}_{7} \\+\frac{{\sigma }_{{{{{{\rm{d}}}}}}}^{a}{I}_{{{{{{\rm{d}}}}}}}}{h{v}_{{{{{{\rm{d}}}}}}}}{n}_{5}-{k}_{{{{{{\rm{et}}}}}}}{\varphi }_{{{{{{\rm{gd}}}}}}}{n}_{{{{{{\rm{defect}}}}}}}{n}_{7}$$
(8)
$${{{{{{\rm{Yb}}}}}}}^{3+}\Big({}^{2}{{{{{\rm{F}}}}}}_{7/2}\Big):\frac{d{n}_{{{{{{\rm{Yb}}}}}}0}}{dt}=-\frac{d{n}_{{{{{{\rm{Yb}}}}}}1}}{dt}$$
(9)
$${{{{{{\rm{Yb}}}}}}}^{3+}\Big({}^{2}{{{{{\rm{F}}}}}}_{5/2}\Big):\frac{d{n}_{{{{{{\rm{Yb}}}}}}1}}{dt}=\frac{{\sigma }_{{{{{{\rm{p}}}}}}}^{a}{I}_{{{{{{\rm{p}}}}}}}}{h{v}_{{{{{{\rm{p}}}}}}}}{n}_{{{{{{\rm{Yb}}}}}}0}-({w}_{1}{n}_{0}+{w}_{2}{n}_{1}+{w}_{3}{n}_{3}+{w}_{4}{n}_{6}){n}_{{{{{{\rm{Yb}}}}}}1}-\frac{{n}_{{{{{{\rm{Yb}}}}}}1}}{{\tau }_{{{{{{\rm{Yb}}}}}}1}}$$
(10)
Here, ni (i = 0 to 7) represents the population of Tm3+ ions on the 3H6, 3F4, 3H5, 3H4, 3F3/3F2, 1G4, 1D2 and 1I6 states, respectively. nYb0 and nYb1 represent the population of Yb3+ ions on the 2F7/2 and 2F5/2 states. βi (i = 2 to 7) represents the non-radiative decay rates of the 3H5, 3H4, 3F3/3F2, 1G4, 1D2 and 1I6 states of Tm3+ ions. τi (i = 1, 3, 5, 6, 7) represents the radiative lifetimes of the 3F4, 3H4, 1G4, 1D2 and 1I6 states of Tm3+ ions, respectively, while \(\frac{1}{{\tau }_{i}}\) represents the radiative decay rates of the corresponding states. ci (i = 1, 2, 3, 4) denotes the coefficients of cross-relaxation processes, 1G4 + 3H4 → 1D2 + 3F4, 3H4 + 1G4 → 1D2 + 3F4, 3H4 + 3H6 → 3F4 + 3F4 and 1D2 + 3H6 → 3F2 + 3H5. wi (i = 1 to 4) denotes the energy transfer upconversion coefficients from Yb3+ to the 3H6, 3F4, 3H4 and 1D2 states of Tm3+, respectively. bij is the branching ratios for the radiative transitions from the initial state i to the terminal state j of Tm3+ (j < i). \({\sigma }_{{{{{{\rm{p}}}}}}}^{a}\) is the absorption cross-section of Yb3+ at 975 nm. \({\sigma }_{{{{{{\rm{d}}}}}}}^{a}\) and \({\sigma }_{{{{{{\rm{d}}}}}}}^{{se}}\) represent the excited state absorption cross-section and stimulated emission cross-section, respectively, for the energy transfer process between 1G4 and 1I6. vp and vd denote the frequencies of 975-nm and 730-nm beams, respectively. Ip and Id denote the intensities of 975-nm and 730-nm beams, respectively. h is Planck’s constant. ket is the coefficient of energy transfer from the 1I6 state of Tm3+ to the 6P7/2 state of Gd3+. φgd represents the efficiency of the energy migration process through the Gd3+ sublattice. ndefect represents the effective population of surface quenchers. Here, similar to the energy transfer upconversion process, the SMED process can be regarded as the energy transfer from 1I6 state to surface quenchers through the Gd3+ sublattice, and written as the term \({k}_{{{{{{\rm{et}}}}}}}{\varphi }_{{{{{{\rm{gd}}}}}}}{n}_{{{{{{\rm{defect}}}}}}}{n}_{7}\). We defined the product of ket, φgd and ndefect as MSEMD (s−1), which represents the equivalent energy dissipation rate of the SMED mechanism. Here, by referring to the experimental results, we estimated the value of MSEMD for the cases of bare-core, inert shell and NaYF4-based Tm3+-activated systems. To achieve a high depletion efficiency of 95% for the bare-core nanoparticles, the value of MSEMD is estimated to be at least 2.0 × 106 s−1. As for the inert-shell protected nanoparticles, the value of MSEMD turns out to be 5.0 × 103 s−1, which is only 1/400 of the counterpart in bare-core nanoparticles. For the NaYF4-based Tm3+-doped systems, the value of MSEMD is set to 0 s−1. The values of the parameters used in the numerical simulation were listed in Supplementary Table2.
Based on the above rate equations, an increased decline in the population of the 1G4 state with increasing depletion intensity was obtained in our simulation, which accorded with our experimental results. Moreover, nanocomposites and nanostructures of particles would affect the depletion efficiency (e.g., inert shell, particle size and concentration of Gd3+ ions) since they can affect the value of MSEMD. Increasing MSEMD enhanced the optical depletion efficiency of emission from the 1G4 state, and the representative results were shown in Fig.3f and Supplementary Fig.6.
Theoretical analysis of the SMED mechanism
To better understand the proposed SMED depletion mechanism, a simplified four-level system (Supplementary Fig.7) is used to investigate the relationship between optical depletion efficiency and the SMED process via steady-state rate equations:
$$\frac{d{n}_{s0}}{{dt}}=-\frac{d{n}_{{{{{{\rm{s}}}}}}1}}{{dt}}$$
(11)
$$\frac{d{n}_{{{{{{\rm{s}}}}}}1}}{{dt}}={\alpha }_{{{{{{\rm{p}}}}}}}{n}_{{{{{{\rm{s}}}}}}0}-({w}_{1}{n}_{0}+{w}_{2}{n}_{1}+{w}_{3}{n}_{2}){n}_{{{{{{\rm{s}}}}}}1}-\frac{{n}_{{{{{{\rm{s}}}}}}1}}{{\tau }_{{{{{{\rm{s}}}}}}1}}$$
(12)
$$\frac{d{n}_{0}}{{dt}}={\beta }_{1}{n}_{1}+\frac{{n}_{1}}{{\tau }_{1}}+\frac{{n}_{2}}{{\tau }_{2}}+\frac{{n}_{3}}{{\tau }_{3}}+{k}_{{{{{{\rm{et}}}}}}}{\varphi }_{{{{{{\rm{net}}}}}}}{n}_{{{{{{\rm{defect}}}}}}}{n}_{3}-{w}_{1}{n}_{{{{{{\rm{s}}}}}}1}{n}_{0}$$
(13)
$$\frac{d{n}_{1}}{{dt}}={\beta }_{2}{n}_{2}+{w}_{1}{n}_{{{{{{\rm{s}}}}}}1}{n}_{0}-{\beta }_{1}{n}_{1}-{w}_{2}{n}_{{{{{{\rm{s}}}}}}1}{n}_{1}-\frac{{n}_{1}}{{\tau }_{1}}-{\alpha }_{{{{{{\rm{d}}}}}}}{n}_{1}$$
(14)
$$\frac{d{n}_{2}}{{dt}}={\beta }_{3}{n}_{3}+{w}_{2}{n}_{{{{{{\rm{s}}}}}}1}{n}_{1}-{\beta }_{2}{n}_{2}-{w}_{3}{n}_{{{{{{\rm{s}}}}}}1}{n}_{2}-\frac{{n}_{2}}{{\tau }_{2}}$$
(15)
$$\frac{d{n}_{3}}{{dt}}={w}_{3}{n}_{{{{{{\rm{s}}}}}}1}{n}_{2}+{\alpha }_{{{{{{\rm{d}}}}}}}{n}_{1}-\frac{{n}_{3}}{{\tau }_{3}}-{\beta }_{3}{n}_{3}-{k}_{{{{{{\rm{et}}}}}}}{\varphi }_{{{{{{\rm{net}}}}}}}{n}_{{{{{{\rm{defect}}}}}}}{n}_{3}$$
(16)
In these equations, \({n}_{i}\) (i = 0 to 3) represents the population of activator ions at different energy states. Particularly, n1 is the population of ions at the emitting state of interest. ns0 and ns1 represent the population of sensitizer ions on the ground and excited states. βi (i = 1 to 3) and τi (i = 1 to 3) represent the non-radiative decay rates and the radiative lifetimes of different energy states in activator ions. wi (i = 1 to 3) denotes the energy transfer upconversion coefficients from sensitizers to activators, respectively. αp and αd represent the absorption rates for the excitation beam and depletion beam, respectively. ket is the coefficient of energy transfer from the activators to the migrators. φnet represents the energy migration efficiency of the sublattice network. ndefect represents the effective population of surface quenchers. Note that for a better understanding of the proposed mechanism, the excitation process has been simplified with the condition that the sensitizers existing in the luminescence system would not affect the calculation results.
Equations (14)–(16) can be rewritten as
$${n}_{1}=\frac{{\beta }_{2}{n}_{2}+{w}_{1}{n}_{{{{{{\rm{s}}}}}}1}{n}_{0}}{{w}_{2}{n}_{{{{{{\rm{s}}}}}}1}+\frac{1}{{\tau }_{1}}+{\beta }_{1}+{\alpha }_{{{{{{\rm{d}}}}}}}}$$
(17)
$${n}_{2}=\frac{{\beta }_{3}{n}_{3}+{w}_{2}{n}_{{{{{{\rm{s}}}}}}1}{n}_{1}}{{w}_{3}{n}_{{{{{{\rm{s}}}}}}1}+\frac{1}{{\tau }_{2}}+{\beta }_{2}}$$
(18)
$${n}_{3}=\frac{{\alpha }_{{{{{{\rm{d}}}}}}}{n}_{1}+{w}_{3}{n}_{{{{{{\rm{s}}}}}}1}{n}_{2}}{\frac{1}{{\tau }_{3}}+{\beta }_{3}+{k}_{{{{{{\rm{et}}}}}}}{\varphi }_{{{{{{\rm{net}}}}}}}{n}_{{{{{{\rm{defect}}}}}}}}$$
(19)
In a feasible SMED system, the upconversion processes from the emitting state to further higher-lying states should be weak. For example, in Tm3+-doped NaGdF4 system, the cross-relaxation processes 1G4 + 3H4 → 1D2 + 3F4 and 3H4 + 1G4 → 1D2 + 3F4 are too weak to transfer electrons from 1G4 state to 1D2 state; in Tb3+-doped NaGdF4 system, the excitation beam cannot even pump the electrons from the emitting state 5D4 to any higher-lying states. Therefore, the upconversion terms w2 ns1 n1 and w3 ns1 n1 can be neglected here.
Combining Eqs. (17) and (18), we have:
$${n}_{1}=\frac{\frac{{{\beta }_{2}\beta }_{3}{n}_{3}}{\frac{1}{{\tau }_{2}}+{\beta }_{2}}+{w}_{1}{n}_{{{{{{\rm{s}}}}}}1}{n}_{0}}{\frac{1}{{\tau }_{1}}+{\beta }_{1}+{\alpha }_{{{{{{\rm{d}}}}}}}}$$
(20)
With the combination of Eqs. (19) and (20), we could also get:
$${n}_{1}=\frac{\frac{{{\beta }_{2}\beta }_{3}{\alpha }_{{{{{{\rm{d}}}}}}}{n}_{1}}{\left(\frac{1}{{\tau }_{3}}+{\beta }_{3}+{k}_{{{{{{\rm{et}}}}}}}{\varphi }_{{{{{{\rm{net}}}}}}}{n}_{{{{{{\rm{defect}}}}}}}\right)\left(\frac{1}{{\tau }_{2}}+{\beta }_{2}\right)}+{w}_{1}{n}_{{{{{{\rm{s}}}}}}1}{n}_{0}}{\frac{1}{{\tau }_{1}}+{\beta }_{1}+{\alpha }_{{{{{{\rm{d}}}}}}}}$$
(21)
By rearranging Eq. (21), we have:
$${n}_{1}=\frac{{w}_{1}{n}_{{{{{{\rm{s}}}}}}1}{n}_{0}}{\frac{1}{{\tau }_{1}}+{\beta }_{1}+{\alpha }_{{{{{{\rm{d}}}}}}}\left(1-\frac{{{\beta }_{2}\beta }_{3}}{\left(\frac{1}{{\tau }_{3}}+{\beta }_{3}+{k}_{{{{{{\rm{et}}}}}}}{\varphi }_{{{{{{\rm{net}}}}}}}{n}_{{{{{{\rm{defect}}}}}}}\right)\left(\frac{1}{{\tau }_{2}}+{\beta }_{2}\right)}\right)}$$
(22)
Equation (22) has expressed the correlation of the population n1 and the absorption rate of the depletion beam αd, which follows a simple functional relationship \(y=a/(b+{cx})\). Here, y denotes the population of the emitting state. x denotes the intensity of the depletion laser beam. The variable a mainly represents the factor of energy excitation processes involved in the emitting state. The variable b represents the factor of intrinsic energy loss pathways involved in the emitting state. The variable c mainly represents the factor of energy states which participate in the depletion processes, and typically the state accepts the losing energy from the emitting state. According to this equation, if the intensity of the depletion beam increases, the population of the emitting state will be effectively reduced. Moreover, increasing the value of c would relieve the required depletion intensity for a specific depletion efficiency. Here, the parameters ket, φnet and ndefect can substantially contribute to the value of c. In other words, the higher the energy transfer rate is from activators to migrators, the more efficient the energy migration process occurs. As a result, more surface quenchers would help lower the saturation intensity significantly, which is in accord with the experimental observation. Hence, the SMED mechanism is a powerful strategy for achieving a highly efficient emission depletion process.
Density functional theory (DFT) calculations
We performed first-principles calculations based on density functional theory (DFT) using the Vienna ab initio simulation package with the projector augmented wave method36,56. The exchange-correlation interactions were approximated by the Perdew–Burke–Ernzerh generalized gradient approximation (GGA-PBE)57. To precisely map the position of the localized 4 f orbital of gadolinium ions, the screened-exchange hybrid density functional HSE06 with 5% Hartree–Fock (HF) exchange interaction was employed58,59. The kinetic energy cut-off of the plane wave was set to 520 eV, the energy convergence criterion was 1 × 10−4, and the maximum force on each relaxed atom was less than 0.02 eV/Å. For the surface model, a 15-Å vacuum spacer was introduced to separate the 9-layer YF-terminated (0001) surface. Note that we chose NaYF4 lattice as model systems instead of NaGdF4 because quantum calculation involving a large quantity of 4 f electrons could be highly resource-demanding. Given the same hexagonal space group, the electronic structure of Gd dopants in the NaYF4 lattice should resemble the host Gd ions in the NaGdF4 lattice.
We studied the influence of defects on the electronic structure of Gd-doped NaYF4 nanocrystal using both bulk and slab models that represent the interior area and the surface of a given nanoparticle, respectively. The defects of interest were fluorine vacancy (VF), oxygen substitution (OF and O2F), hydroxyl substitution (OHF), and water adsorbed on Gd (H2O-Gd), as well as OH adsorbed on Gd (OH-Gd), which could be formed during nanoparticle synthesis. The calculated density of states (DOS) showed that the NaYF4 host has a clean bandgap of 7.74 eV and the Gd dopant barely affects the electronic structure of the system except for 4f-based midgap states. Here, we used the energy bandgap between occupied and empty 4 f orbitals to mimic the energy difference between the Gd’s ground state (3S7/2) and its first excited state (6P7/2). Note that the calculated 4f–4 f gap is overestimated compared to the 3S7/2−6P7/2 gap because DFT calculation can only access the ground-state electronic structure of the given system and the spin-orbit coupling was not considered. Despite the quantitative discrepancy between simulated and experimentally measured 3S7/2−6P7/2 gap, it is rational to use the theoretically captured electronic changes in NaYF4-based systems containing different defects for the mechanistic explanation.
As the total and projected density of states of NaYF4:Gd and NaYF4:Gd+O2F shown in Supplementary Fig.8a, b, considering that the values of these energy gaps are close to the 4f-4f gap of the Gd3+ dopant, Förster resonance energy transfer from excited Gd3+ ions to defect sites can occur via dipole-dipole coupling. The occupied state and empty states mainly originate from O’s p orbital and Y’s 4d orbital (Supplementary Fig.8c, d), respectively, suggesting dipole-allowed electronic transitions between the states of high oscillation strength. Additionally, these impurity-state-involved transitions should have broadband absorption that ensures a large spectral overlap with the Gd3+ emission. Moreover, our calculations suggested that these lattice defects prefer proximity to Gd3+ dopant, indicating a short Gd-defect distance. This implies that energy transfer from Gd3+ ions to defect quenchers can occur at a high transfer rate, in agreement with the MSMED calculation in the aforementioned rateequation modeling. On the other hand, as the total and projected density of states of the hydroxyl- and water-adsorbed NaYF4:Gd (0001) surface shown in Supplementary Fig.8e, f, although adsorption of OH− or H2O introduced midgap states, the energy differences between these states and host states do not match the 3S7/2−6P7/2 gap, suggesting that electronic deactivation of excited Gd3+ ions via absorbed moieties is trivial. These adsorbates likely deactivate excited states of Gd3+ or Tm3+ directly through inefficient multiphonon coupling. Supplementary Fig.8g summarizes the optically active and inert defects for the depletion of Gd3+ ions.
Software
Software Olympus FV10 ASW ver. 4.0a was used to control laser scanning microscopy imaging and acquire images. Software Ocean Optics SpectraSuite was used to control spectrometer and acquire spectra. No custom algorithms or software that are central to the data collection. Matlab (v2019a) was used for the analysis of spectra, the analysis of microscopic images and the calculation of numerical simulations. FRC analysis was performed with the code provided by previous work60. The first principles calculations based on DFT were performed using the Vienna ab initio simulation package with the projector augmented wave method36,56.
Reporting summary
Further information on research design is available in theNature Research Reporting Summary linked to this article.
