Article | Open | Published:

# Low irradiance multiphoton imaging with alloyed lanthanide nanocrystals

## Abstract

Multiphoton imaging techniques that convert low-energy excitation to higher energy emission are widely used to improve signal over background, reduce scatter, and limit photodamage. Lanthanide-doped upconverting nanoparticles (UCNPs) are among the most efficient multiphoton probes, but even UCNPs with optimized lanthanide dopant levels require laser intensities that may be problematic. Here, we develop protein-sized, alloyed UCNPs (aUCNPs) that can be imaged individually at laser intensities >300-fold lower than needed for comparably sized doped UCNPs. Using single UCNP characterization and kinetic modeling, we find that addition of inert shells changes optimal lanthanide content from Yb3+, Er3+-doped NaYF4 nanocrystals to fully alloyed compositions. At high levels, emitter Er3+ ions can adopt a second role to enhance aUCNP absorption cross-section by desaturating sensitizer Yb3+ or by absorbing photons directly. Core/shell aUCNPs 12 nm in total diameter can be imaged through deep tissue in live mice using a laser intensity of 0.1 W cm−2.

## Introduction

Light microscopy is the primary means of studying complex living systems, enabling real-time analysis with ever-increasing spatial and temporal resolution. Increasingly powerful imaging techniques and lasers have raised concern over light toxicity1,2, which is most acute with high laser intensities at shorter wavelengths in the ultraviolet and visible regions3,4. Near-infrared (NIR) excitation is more benign than these higher energy wavelengths4,5, and nonlinear multiphoton techniques that use NIR excitation have been widely adopted6,7,8,9,10. Both scatter and absorption by cellular components are much lower for NIR light than for visible light8,11,12, and this steep wavelength dependence has been shown in direct comparisons to reduce photodamage using NIR-based techniques4,5,13,14. Multiphoton probes excitable at reduced laser intensities in the NIR would enable powerful high-resolution and deep-tissue imaging techniques in sensitive systems without associated phototoxicity.

Lanthanide-doped upconverting nanoparticles (UCNPs) are phosphors that absorb multiple photons in the NIR and emit at higher energies in the NIR or visible spectral regions. The luminescence efficiencies of UCNPs are orders of magnitude higher than those of the best two-photon fluorophores15,16,17,18, and they exhibit no on-off blinking, no overlap with cellular autofluorescence, and no measurable photobleaching under prolonged single-particle excitation5,19,20. UCNPs make use of energy transfer upconversion between neighboring lanthanide ions (Ln3+), in which sensitizer ions with relatively large absorption cross-sections sequentially transfer absorbed energy to luminescent emitter ions, both of which are doped into a low-phonon-energy nanocrystal host. For many applications, β-phase NaYF4 nanocrystals doped with 20% Yb3+ sensitizer and a low percentage of Er3+ or Tm3+ emitter are most efficient17,18,21. Addition of inert epitaxial shells to these UCNPs has been shown to enhance emission at low excitation powers by reducing Yb3+-mediated energy migration to high-vibrational-frequency modes of surface oleate ligands or solvent19. For UCNPs with high Ln3+ content, this has been attributed to suppression of concentration quenching22,23, an observation that encompasses a number of known as well as unexplored energetic pathways that reduce the quantum yield (QY) of upconverted emission22,24,25,26,27,28.

Here, we use single-nanoparticle characterization and kinetic models of Ln3+ energy transfer to develop antibody-sized (approximately 12–15 nm diameter) alloyed UCNPs that can be imaged at the single-particle level at laser intensities below 300 W cm−2, over 300-fold lower than needed for comparably sized doped UCNPs. Core/shell aUCNPs are brighter than comparably sized doped UCNPs at all laser intensities tested, over a range of four orders of magnitude. Addition of inert epitaxial shells radically changes optimal lanthanide content from Yb3+, Er3+-doped NaYF4 nanocrystals to fully alloyed compositions, and at high levels of the emitter Er3+, these ions can adopt a second role to enhance the effective aUCNP absorption. This leads to a revised UCNP design in which there is no need to dope Ln3+ ions into an inert NaYF4 (or other) host matrix. In live mice, aqueous 12-nm core/shell aUCNPs can be imaged with strong contrast (signal:background >25) through several millimeters of tissue with a laser intensity of just 0.1 W cm−2. aUCNPs open up the possibility of using both low irradiance and low-energy excitation wavelengths for non-destructive bioimaging experiments.

## Results

### Characterization of small NaLnF4 core/shell nanoparticles

To better understand how inert epitaxial shells affect UCNP emission, we analyzed emissions of single nanocrystals to compare absolute brightness at different laser intensities. UCNP emission is deeply power-dependent and size-dependent, and single-nanocrystal characterization allows quantitative comparison of non-aggregated nanocrystals under identical environments over four orders of magnitude excitation power density8,13, a range that spans imaging experiments from single-molecule studies to imaging of highly light-sensitive samples. We synthesized a series of 8-nm diameter β-phase NaYF4 cores19, and overcoated them with NaYF4 shells using a layer-by-layer protocol29 (Fig. 1 and Methods). Several NaLnF4 alloys of heavy lanthanides (e.g., Yb3+-Er3+, Yb3+-Tm3+, and Yb3+-Ho3+, as well as NaErF4) have been reported22,25,28,30,31, including sub-20-nm NaYbF4:Tm core/shell nanopoarticles32, although none of these compositions have not been characterized by quantitative single-particle imaging or systematically over a large range of power densities. Characterization of our nanocrystals by electron microscopy (EM) and X-ray diffraction (XRD) showed monodisperse β-phase nanocrystals for both 8-nm core and core/shell UCNPs (Fig. 1e and Supplementary Figs. 13). High-angle annular dark-field scanning transmission EM (STEM), which is sensitive to atomic number Z, confirms the core/shell structure, showing clear boundaries between alloyed cores and 20% Gd3+-doped shells (Fig. 1f).

Upconverted emission spectra of single core/shell UCNPs have similar transitions as seen for unshelled UCNPs (Fig. 1d and Supplementary Fig. 4), but laser power density studies show large increases in emission dependent on both the lanthanide content and laser power densities. At laser intensities above 106 W cm−2, core UCNPs doped with 20% Yb3+ sensitizer show maximum emission with 20% Er3+ emitter (Fig. 2a), while higher Er3+ doping leads to apparent Er3+-Er3+ cross-relaxation or enhanced surface losses18,25. Once shells are added, additional Er3+ increases emission, with fully alloyed UCNPs almost an order of magnitude brighter (Fig. 2b) than the most efficient doped cores18. In contrast to unshelled UCNPs, where power dependence is critical and the brightest compositions at high laser intensities are not even luminescent at low intensities18, here the brightest aUCNP compositions are superior over a 10,000-fold range of excitation intensities. At the highest excitation intensities, shells provide almost a twofold emission increase for 2% Er3+ doping and increase up to 30-fold for aUCNPs (Fig. 2a and Supplementary Fig. 5), suggesting that at laser powers used for single-particle imaging, any quenching typically associated with high Ln3+ content25,26 can be suppressed by the addition of inert shells22. Single UCNP analysis of shell thickness (Fig. 2b and Supplementary Fig. 1) shows larger emission enhancements at lower powers, and with enhancements diminishing above 4-nm shell thickness16. Power series show little difference in single aUCNP brightness with either a majority of Yb3+ sensitizer or Er3+ emitter ions (e.g., NaEr0.6Yb0.4F4 vs. NaEr0.2Yb0.8F4; Supplementary Fig. 5), suggesting more complex behaviors than the traditional roles assigned to these ions.

### Imaging aUCNP nanoparticles at low irradiances

We observed an even greater effect of inert shells at the lowest intensities (Fig. 3a). While the best cores require >105 W cm−2 to image single sub-20-nm nanocrystals at signal-to-background ratios (SBR) above 3 (corresponding here to 100 c.p.s. above background18), most individual core/shell UCNPs can be imaged at irradiances below 1000 W cm−2 (Fig. 3b and Supplementary Fig. 6). We observe that increasing the fraction of Yb3+ (Supplementary Fig. 5) enables single core/shell NaEr0.2Yb0.8F4 aUCNPs to be imaged as low as 290 W cm−2 (Fig. 4a–e and Supplementary Fig. 7), >300-fold lower than the best previous sub-20-nm compositions17,18. For confocal imaging of ensembles of nanoparticles, intensities as low as 4 W cm−2 can be used to image these same aUCNPs loaded into polystyrene beads (Fig. 4f–j and Supplementary Fig. 7), and intensities <0.1 W cm−2 can be used for imaging aUCNPs injected into live mice, as discussed below. From these experiments, a revised UCNP design emerges in which there is no need to dope lanthanides into an inert matrix, as fully alloyed lanthanide UCNPs are brighter at all laser intensities.

### Mechanisms of improved aUCNP brightness

To understand why the brightest core/shell compositions are alloyed, we determined single-nanocrystal absorption cross-sections and QYs (Supplementary Fig 8) and used these data with rate equation models18,33 to uncover how high Ln3+ content alters photophysical processes in UCNPs. At irradiances above the saturation intensity, Is, at least 25% of Yb3+ ions are populated in their 2F5/2 excited state and thus are not able to absorb additional incident 980 nm photons. Calculations of Is based on Yb3+ and Er3+ absorption cross-sections at 980 nm (σabs; Supplementary Fig. 8a and Methods) show that core/shell aUCNPs reach saturation at 3000 W cm−2, and kinetic models of UCNP Ln3+ populations13 find that ≥75% of Yb3+ are in the excited state at intensities above 106 W cm−2. Although the calculated σabs for Yb3+ at 980 nm is 14-fold larger than that of Er3+ (see Methods), at high laser intensities, the contribution of the remaining ground state Yb3+ to total UCNP absorption should be sharply diminished, and even eclipsed by the Er3+ contribution in NaEr0.8Yb0.2F4 aUCNPs. However, with increasing Er3+ levels, Yb3+ ET to nearby Er3+ ions is enhanced, as ET rates scale with R−6 (where R is the distance between ions) and with the product of the ion concentrations. This suggests that nearby Er3+ ions offer new pathways to return Yb3+ to its ground state, and this desaturation would then enhance the effective absorption cross-section of the aUCNP (Fig. 5).

### New roles of Er3+ in enhanced aUCNP absorption

To determine whether desaturation and direct Er3+ absorption lead to the enhanced emission of aUCNPs, we determined single UCNP QYs based on single UCNP emissions (Fig. 2a) relative to a well-established doped UCNP standard13 (Fig. 3c). UCNP absorption cross-sections were calculated (Supplementary Fig. 8) for the saturating intensities of the single UCNP measurements (2 × 106 W cm−2). For UCNPs with 20% Yb3+, QY values rise sharply from 2 to 20% Er3+, but above 20% QY values plateau while relative single UCNP emission continues to rise. This suggests relief from quenching is most important at lower Er3+ content, while Er3+ above 20% leads to increasing emission by enhanced absorption. Comparing the brightest aUCNPs to doped UCNPs with the same Er3+ content, additional Yb3+ leads to unchanged or decreased QY values (e.g., compare 20% Yb3+, 60% Er3+, and NaEr0.6Yb0.4F4 in Fig. 3c). In contrast with previous discussions of high Ln3+ UCNPs, these single UCNP experiments suggest it is not relief from concentration quenching that drives increases in aUCNP emission, but rather concentration enhancement in effective absorption cross-section due to the close proximity and larger ground state populations of of Ln3+ able to absorb incident photons (Fig. 5). Concentration enhancement presents a second role for the emitter ion Er3+, in increasing absorbance directly, or indirectly, by desaturating Yb3+ and enabling it to absorb more photons per unit time. This unexpected role of Er3+ in absorption also answers the question of why aUCNPs are brighter in cases where neither number of absorbing Yb3+ ions nor the QYs change significantly.

To examine whether the higher Ln3+ content of aUCNPs lead to the faster Er3+-Yb3+ ET rates that underlie concentration enhancement, we examined lifetime decays of doped and alloyed UCNPs as a function of excitation intensity (Supplementary Fig. 9). At saturating intensities (Is > 3000 W cm−2), weighted lifetimes of core/shell aUCNP green and red emission decrease with increasing Er3+ fraction, from ~400 μs at 100 W cm−2 to ~10 μs at 106 W cm−2 (Supplementary Fig. 8b, c), without associated quenching manifested as decreases in QY or brightness (Fig. 3c). Above Is, models of time-resolved luminescence18,34 have found strong correlations between decays of emitting Er3+ levels and excited Yb3+ ions, which act as reservoirs that rapidly repopulate Er3+ emitting levels via ET. Shortened aUCNP lifetimes are therefore consistent with the onset of rapid Yb3+ desaturation pathways mediated by close Er3+–Yb3+ pairs. In contrast, at sub-saturation intensities, UCNPs show similar decays regardless of Er3+ content (Supplementary Fig. 9), again suggesting that lifetimes of emitting Er3+ ions are influenced more by the kinetics of the Yb3+ excited state than by other relaxation processes23,25. This unusual combination of brighter emission with shorter radiative lifetimes may be useful for fast-scanning techniques such as confocal imaging, where the long lifetimes of doped UCNPs can lead to blurring35.

The sharp power dependence of aUCNP lifetimes suggests different mechanisms of emission enhancement above and below Is (Fig. 5). At higher powers that saturate Yb3+ absorption, enhanced aUCNP emission depends on rapid ET to proximal Er3+ ions, desaturating Yb3+ excited states so that the Yb3+ ions are freed to absorb incident photons. Below Is, UCNPs show similar lifetimes regardless of Er3+ content. At these low excitation intensities, most Yb3+ ions are already in their ground states, enhancement in aUCNPs is driven primarily by increased absorption due to the larger number of absorbing Ln3+ ions per aUCNP. Small increases in total aUCNP σabs at 980 nm (Supplementary Fig. 8a) are significant because of the quadratic dependence of emission on the photon absorption rate. For example, the calculated 980-nm σabs of NaYb0.4Er0.6F4 aUCNPs is twice that of UCNPs doped with 20 Yb3+ and 20% Er3+, which suggests a fourfold higher emission and which aligns well with the experimentally measured 4.2-fold enhancement at 490 W cm−2 (Fig. 3b). Combined with calculations and lifetime decays, quantitative single UCNP measurements across a broad range of Ln3+ content and excitation intensities have allowed us to distinguish the critical mechanisms of aUCNP emission enhancement that dominate at different imaging conditions.

### Deep-tissue imaging of 12-nm aUCNPs

To determine how aUCNPs can be imaged in biological systems at low laser intensities, we transferred 8-nm NaEr0.6Yb0.4F4 aUCNPs with 2-nm shells (12 nm total diameter) to water via polymer encapsulation36 and injected them into mammary fat pads 3–4 mm beneath the skin of 5-week-old mice (Fig. 6 and Methods). Images of green emission acquired with 980-nm excitation at 0.1 W cm−2 all show SBR ≥25, with signal decreasing from 2 to 6 h, likely owing to a slow extravasation of aUCNPs from the mammary glands into draining lymph nodes. By comparison, the maximum permissible exposure for 980 nm continuous-wave lasers to human skin is 0.73 W cm−2 28,37, and previous deep-tissue experiments with doped UCNPs have typically required far larger doped UCNPs to achieve similar SBR values at these low irradiances28,38. These experiments demonstrate that protein-sized aUCNPs can be locally injected and imaged without notable toxicity, and demonstrate a robust aUCNP signal from deep tissue, even with visible emission.

## Discussion

Increasingly complex multi-shell and multi-Ln3+ designs for UCNPs have been successfully deployed for super-resolution imaging, lasing, and sensitized emission21,39,40,41. Here we find a new role for Er3+ in enhancing absorption, leading to simple core/shell designs with just Yb3+ and Er3+ as optically active ions that can yield superior nanoparticles over the entire range of useful UCNP imaging intensities. The addition of inert shells radically changes the brightest compositions from doped UCNPs to alloyed, dispensing with the need for a host matrix. These newly optimized compositions show >300-fold emission enhancements compared to doped UCNPs, and aUCNPs just 12 nm in diameter can be imaged deep in tissue at excitation intensities well below those known to cause physiological damage4,5,28,37. While we have focused on aUCNPs similar in size to antibodies to maintain biocompatibility, increasing the sizes of these same aUCNPs would further decrease the minimum required intensities, as brightness scales with the number of Ln3+, which increase as d3. Larger aUCNPs may limit certain applications in confined or crowded systems, but open up other applications where extremely low intensities of low-energy NIR light are required to minimize phototoxicity, such as with stem cells, embryos, or human tissue.

## Methods

### Synthesis of alloyed β-NaErxYbyF4 nanocrystals

Similar procedures were used for all NaErxYbyF4, NaErF4, and doped NaYF4 cores, except that no oleylamine was required for nanocrystals with <40% Yb. All nanocrystals showed pure β phase by powder XRD and TEM.

### Synthesis of core/shell aUCNPs

Epitaxial NaY0.8Gd0.2F4 shells were grown on NaErxYbyF4, NaErF4, and doped NaYF4 cores using a method modified from that of Li29. The added Gd3+ was found to be necessary for pure β-phase shell growth13. Precursors were prepared by heating YCl3 (0.40 mmol, 78 mg) and GdCl3 (0.10 mmol, 26 mg) to 110 °C in oleic acid (2 mL) and ODE (3 mL) and stirred for 15 min under vacuum. The flask was filled with N2 and heated to 160 °C for 30 min to allow the GdCl3 to dissolve, which was followed by another 15 min at 110 °C under vacuum, to give a 0.10 M solution of 80:20 Y/Gd oleate (Y/Gd-OA). In a separate flask, a Na/F precursor was prepared by dissolving sodium trifluoroacetate (1.20 mmol, 163 mg) in oleic acid (3 mL) and applying vacuum at room temperature for 20 min, giving a 0.40 M Na-TFA-OA solution.

For 4-nm shells grown on 8-nm cores: a stock dispersion of NaErxYbyF4 cores (27 μmol in hexane) was added to 4 mL of oleic acid and 6 mL of ODE. Hexane was removed by applying vacuum at 70 °C for 30 min, N2 was introduced, and the flask heated to 280 °C. Injection cycles of Y/Gd-OA and Na-TFA-OA precursor volumes29 (Supplementary Table 1) were followed 5 min after the reaction reached to 280 °C. Every cycle began with the Y/Gd-OA solution, followed by the Na-TFA-OA solution 15 min later to form single 0.5-nm unit cell layer. After the last injection, the reaction was maintained at 280 °C for 30 min to allow for complete shell growth. Then, the reaction was cooled rapidly, and 15 mL of ethanol and 15 mL of acetone were added when the reaction reached 75 °C. Nanoparticles were precipitated, cleaned, and stored as described for the UCNP cores.

### Nanocrystal characterization

For XRD, 1 mL of a stock solution of the nanoparticles in hexane was precipitated with the addition of 2 mL of EtOH. The nanoparticle slurry was spotted onto a glass coverslip or silicon wafer multiple times until an opaque white film formed, and the sample was allowed to air dry completely. XRD patterns were obtained on a Bruker AXS D8 Discover GADDS X-ray diffractometer system with Co Kα radiation (λ = 1.78897 Å) from 2θ of 15 to 65°.

Nanocrystal size was determined by dynamic light scattering measurements on a Malvern Zetasizer. Samples were prepared from hexane stocks by dilution with hexane to ~50 nM. The diameters of the nanoparticles in each sample were determined based on the fitting by volume according to instrument software.

For EM, UCNPs were precipitated from hexane with EtOH, washed with EtOH, and resuspended in hexane to 10 nM; 7 μL was dropped onto ultrathin carbon film/holey carbon grid, 400 mesh copper (Ted Pella). Images of the nanoparticles were obtained using a Zeiss Gemini Ultra-55 analytical scanning electron microscope. Dark-field images were collected in transmission (STEM) mode with 30 kV beam energy. TEM images were also obtained using a JEOL 2100-F 200 kV field-emission analytical transmission electron microscope. For single-particle determination, samples deposited on silicon nitride windows (Ted Pella) were used and imaged in STEM mode at 30 kV.

### Single UCNP imaging

For emission spectra and power series, UCNPs were diluted in hexane to ~1 fM before dropcasting onto No.1 glass coverslip or silicon nitride grid (Ted Pella). Laser scanning confocal imaging was performed in ambient conditions using a lab-built pinhole confocal18 (Supplementary Fig. 10) with 980-nm continuous-wave laser (Thorlabs TCLDM9, 300 mW diode). Because the diffraction-limited beam spot is larger than individual nanoparticle size, emission single particles was confirmed on SiN TEM-grid samples by correlation with subsequent SEM imaging on a Zeiss Ultra-55, operating in transmission mode. Single-particle determination was also confirmed by intensity histograms compiled from 50–300 individual UCNPs for each composition (Supplementary Fig. 11). From the confocal laser scanned images, the Gaussian distribution peaks were adopted as the single-particle upconversion emission intensity. Samples with ambiguous distributions or ones not correlating with SEM images were discarded. Excitation power series were created by rotating a series of neutral density filters in the path of the laser beam while monitoring power simultaneously.

Excitation power density (PD) was calculated as PD = P/S, where P is the laser power on the sample and S is the laser spot area. Laser powers at the sample were measured in-line by an NIST-traceable power meter (Thorlabs) with S120C sensor, covering 400–1100 nm and 50 nW–50 mW with 1 nW resolution. To calculate power densities (W cm−2), the beam spot area was determined by fitting the UCNP emission profile to a Gaussian distribution, where the diameter is the full-width at half-max in the form FWHM = 2$$\sqrt {2\,{\mathrm{ln}}\,2}$$ σ of 580 nm, giving a laser spot area of 2.64 × 10−9 cm2.

### Absorption cross-sections and QYs

Representative Er3+ absorption cross-sections (σabs) at 980 nm were calculated using Judd–Ofelt theory in conjunction with experimental absorption measurements;33 the Yb3+ σabs value was determined from literature values as in Supplementary Methods. Whole-UCNP σabs values were determined by calculating number of Yb3+ and Er3+ ions per UCNP according to average TEM diameters (Supplementary Fig. 2c), Ln3+ fractions, and hexagonal phase NaLnF4 dimensions. Single UCNP QYs were determined relative to the standard value (QY = 0.49%)13 of core/shell 8-nm 20% Yb3+ 2% Er3+ UCNPs measured at 10 W cm−2, which was converted to a high-intensity value using calculated power dependence curve in Supplementary Fig. 8b. Single UCNP emissions from Fig. 2a were used, and all compositions were taken to be at 75% saturation based on kinetic simulations described in Supplementary Methods, Supplementary Tables 1 and 2, and Supplementary Fig. 12.

### Ensemble UCNP imaging

To image aUCNP ensembles in beads, as described previously13, UCNPs were loaded into 0.5 μm polystyrene beads (Aldrich) by swelling 0.5 mg of beads in 250 μL of a 5% (v v−1) CHCl3 solution in n-BuOH. aUCNPs (2 mg in 15 μL of hexane) were added to the bead suspension and briefly vortexed. After stagnant incubation at 25 °C for 4 h, the beads were washed two times with EtOH, centrifuged at 3000 × g for 4 min, and stored in EtOH. Based on single-bead and single UCNP intensities, we estimate there are 70 UCNPs and 200 aUCNPs per bead in Fig. 4.

### Aqueous passivation of core/shell aUCNPs

Hydrophobic 8-nm NaEr0.6Yb0.4F4 aUCNPs with 2-nm NaY0.8Gd0.2F4 shells were dispersed in hexane with 0.2% (v v−1) oleic acid to 5 μM. For aqueous dispersions36, 6 mg of poly(maleic anhydride-alt-1-octadecene) amphiphilic copolymer (MW 20–25k, Aldrich) was dissolved to 17 μM in 0.5 mL of acetone and 15 mL of CHCl3. aUCNPs (0.5 nmol) in 100 μL of hexane were added with stirring, and the solvents were removed under a gentle stream of N2 overnight. aUCNP/polymer residue was then resuspended in a solution of methoxy-PEG8-amine (Thermo Fisher, 10 μmol) in 10 mL of 100 mM NaHCO3 buffer, pH 8.2, with 1% (v v−1) dimethyl sulfoxide. This suspension was sonicated for 60 min, heated in an 80 °C water bath for 60 min, slowly cooled to room temperature, and then sonicated for 30 min. Excess polymer was removed by extensive spin dialysis (Amicon, 100 kDa MWCO), washing with 7 × 15 mL of 100 mM HEPES (4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid), pH 7.4. The retentate was concentrated to 680 μL and filtered through a 0.2-μm filter into a sterile glass vial. Final concentration (200 nM) was determined using an emission vs. concentration curve measured for the parent hydrophobic aUCNPs.

### Live animal imaging

Animal experiments were conducted according to protocols approved by the UCSF Animal Care and Use Committee. Nude/nude homozygous, female, 5-week-old mice (Taconic Farms) were anesthetized and injected with 25 μL of 200 nM 12-nm core/shell NaEr0.6Yb0.4F4 (8-nm core with 2-nm shell) dispersions into mammary fat pads 3–4 mm below the skin. Mice were imaged with an IVIS Spectrum In Vivo Imaging System (Perkin Elmer) equipped with a 4.5 mW 980-nm continuous-wave laser (Thorlabs) and 780-nm longpass filter (Chroma). The beam was focused to 0.1 W cm−2 and emission was collected from 530–550 nm in the green Er3+ band using 2.5-s integration times at 2, 4, and 6 h after injection. Dark counts were measured in the absence of laser excitation, and background was measured in uninjected areas with laser excitation. SBR values were calculated as the ratio of emission to background with dark counts subtracted from each.

### Statistical analysis

Where indicated, results are presented as mean ± one standard deviation.

### Data availability

All relative data are available from the corresponding authors upon request.

Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

## References

1. 1.

Ji, N., Magee, J. C. & Betzig, E. High-speed, low-photodamage nonlinear imaging using passive pulse splitters. Nat. Methods 5, 197–202 (2008).

2. 2.

Laissue, P. P., Alghamdi, R. A., Tomancak, P., Reynaud, E. G. & Shroff, H. Assessing phototoxicity in live fluorescence imaging. Nat. Methods 14, 657–661 (2017).

3. 3.

Carlton, P. M. et al. Fast live simultaneous multiwavelength four-dimensional optical microscopy. Proc. Natl Acad. Sci. USA 107, 16016–16022 (2010).

4. 4.

Squirrell, J. M., Wokosin, D. L., White, J. G. & Bavister, B. D. Long-term two-photon fluorescence imaging of mammalian embryos without compromising viability. Nat. Biotechnol. 17, 763–767 (1999).

5. 5.

Nam, S. H. et al. Long-term real-time tracking of lanthanide ion doped upconverting nanoparticles in living cells. Angew. Chem. Int. Ed. 50, 6093–6097 (2011).

6. 6.

Denk, W., Strickler, J. H. & Webb, W. W. 2-Photon laser scanning fluorescence microscopy. Science 248, 73–76 (1990).

7. 7.

Zipfel, W. R., Williams, R. M. & Webb, W. W. Nonlinear magic: multiphoton microscopy in the biosciences. Nat. Biotechnol. 21, 1368–1376 (2003).

8. 8.

Horton, N. G. et al. In vivo three-photon microscopy of subcortical structures within an intact mouse brain. Nat. Photonics 7, 205–209 (2013).

9. 9.

Liu, Q., Feng, W., Yang, T. S., Yi, T. & Li, F. Y. Upconversion luminescence imaging of cells and small animals. Nat. Protoc. 8, 2033–2044 (2013).

10. 10.

Sugiyama, N., Sonay, A. Y., Tussiwand, R., Cohen, B. E. & Pantazis, P. Effective labeling of primary somatic stem cells with BaTiO3 nanocrystals for second harmonic generation imaging. Small 14, 1703386 (2018).

11. 11.

Frangioni, J. V. In vivo near-infrared fluorescence imaging. Curr. Opin. Chem. Biol. 7, 626–634 (2003).

12. 12.

Hong, G. et al. Multifunctional in vivo vascular imaging using near-infrared II fluorescence. Nat. Med. 18, 1841–1846 (2012).

13. 13.

Levy, E. S. et al. Energy-looping nanoparticles: harnessing excited-state absorption for deep-tissue imaging. ACS Nano 10, 8423–8433 (2016).

14. 14.

Zhong, M. C., Wei, X. B., Zhou, J. H., Wang, Z. Q. & Li, Y. M. Trapping red blood cells in living animals using optical tweezers. Nat. Commun. 4, 1768 (2013).

15. 15.

Chan, E. M. Combinatorial approaches for developing upconverting nanomaterials: high-throughput screening, modeling, and applications. Chem. Soc. Rev. 44, 1653–1679 (2015).

16. 16.

Fischer, S., Bronstein, N. D., Swabeck, J. K., Chan, E. M. & Alivisatos, A. P. Precise tuning of surface quenching for luminescence enhancement in core-shell lanthanide-doped nanocrystals. Nano Lett. 16, 7241–7247 (2016).

17. 17.

Zhao, J. B. et al. Single-nanocrystal sensitivity achieved by enhanced upconversion luminescence. Nat. Nanotechnol. 8, 729–734 (2013).

18. 18.

Gargas, D. J. et al. Engineering bright sub-10-nm upconverting nanocrystals for single-molecule imaging. Nat. Nanotechnol. 9, 300–305 (2014).

19. 19.

Ostrowski, A. D. et al. Controlled synthesis and single-particle imaging of bright, sub-10 nm lanthanide-doped upconverting nanocrystals. ACS Nano 6, 2686–2692 (2012).

20. 20.

Wu, S. W. et al. Non-blinking and photostable upconverted luminescence from single lanthanide-doped nanocrystals. Proc. Natl Acad. Sci. USA 106, 10917–10921 (2009).

21. 21.

Chan, E. M., Levy, E. S. & Cohen, B. E. Rationally designed energy transfer in upconverting nanoparticles. Adv. Mater. 27, 5752–5761 (2015).

22. 22.

Zuo, J. et al. Employing shells to eliminate concentration quenching in photonic upconversion nanostructure. Nanoscale 9, 7941–7946 (2017).

23. 23.

Wen, S. et al. Advances in highly doped upconversion nanoparticles. Nat. Commun. 9, 2415 (2018).

24. 24.

Hossan, M. Y. H. A., Luu, Q., Smith, S. J., May, P. S. & Berry, M. T. Explaining the nanoscale effect in the upconversion dynamics of β‑NaYF4:Yb3+, Er3+ core and core−shell nanocrystals. J. Phys. Chem. C 121, 16592–16606 (2017).

25. 25.

Johnson, N. J. J. et al. Direct evidence for coupled surface and concentration quenching dynamics in lanthanide-doped nanocrystals. J. Am. Chem. Soc. 139, 3275–3282 (2017).

26. 26.

Ma, C. S. et al. Optimal sensitizer concentration in single upconversion nanocrystals. Nano Lett. 17, 2858–2864 (2017).

27. 27.

Shen, B. et al. Revisiting the optimized doping ratio in core/shell nanostructured upconversion particles. Nanoscale 9, 1964–1971 (2017).

28. 28.

Chen, G. Y., Agren, H., Ohulchanskyy, T. Y. & Prasad, P. N. Light upconverting core-shell nanostructures: nanophotonic control for emerging applications. Chem. Soc. Rev. 44, 1680–1713 (2015).

29. 29.

Li, X. M. et al. Successive layer-by-layer strategy for multi-shell epitaxial growth: shell thickness and doping position dependence in upconverting optical properties. Chem. Mater. 25, 106–112 (2013).

30. 30.

Chen, G. et al. Alpha-NaYbF4:Tm3+/CaF2 core/shell nanoparticles with efficient near-infrared to near-infrared upconversion for high-contrast deep-tissue bioimaging. ACS Nano 6, 8280–8287 (2012).

31. 31.

Chen, B. et al. Amplifying excitation power sensitivity of photon upconversion in a NaYbF4:Ho nanostructure for direct visualization of electromagnetic hotspots. J. Phys. Chem. Lett. 7, 4916–4921 (2016).

32. 32.

Shi, R. et al. Tuning hexagonal NaYbF4 nanocrystals down to sub-10 Nm for enhanced photon upconversion. Nanoscale 9, 13739–13746 (2017).

33. 33.

Chan, E. M. et al. Combinatorial discovery of lanthanide-doped nanocrystals with spectrally pure upconverted emission. Nano Lett. 12, 3839–3845 (2012).

34. 34.

Hossan, M. Y. et al. Explaining the nanoscale effect in the upconversion dynamics of beta-NaYF4:Yb3+, Er3+ core and core-shell nanocrystals. J. Phys. Chem. C 121, 16592–16606 (2017).

35. 35.

Pichaandi, J., Boyer, J.-C., Delaney, K. R. & van Veggel, F. C. J. M. Two-photon upconversion laser (scanning and wide-field) microscopy using Ln3+-doped NaYF4 upconverting nanocrystals: a critical evaluation of their performance and potential in bioimaging. J. Phys. Chem. C 115, 19054–19064 (2011).

36. 36.

Mann, V. R., Powers, A. S., Tilley, D. C., Sack, J. T. & Cohen, B. E. Azide–alkyne click conjugation on quantum dots by selective copper coordination. ACS Nano 12, 4469–4477 (2018).

37. 37.

American National Standards Institute. American National Standard for the Safe Use of Lasers, ANSI Z136.1 (National Standards Institute, New York, 1993).

38. 38.

Zhou, J. et al. Dual-modality in vivo imaging using rare-earth nanocrystals with near-infrared to near-infrared (NIR-to-NIR) upconversion luminescence and magnetic resonance properties. Biomaterials 31, 3287–3295 (2010).

39. 39.

Liu, Y. J. et al. Amplified stimulated emission in upconversion nanoparticles for super-resolution nanoscopy. Nature 543, 229–233 (2017).

40. 40.

Garfield, D. J. et al. Enrichment of molecular antenna triplets amplifies upconverting nanoparticle emission. Nat. Photonics 12, 402–407 (2018).

41. 41.

Fernandez-Bravo, A. Y. et al. Continuous-wave upconverting nanoparticle microlasers. Nat. Nanotechnol. 13, 572–577 (2018).

## Acknowledgements

We thank V. Mann and C. Ajo-Franklin for helpful discussions. B.T. and Y.T. were supported by fellowships from the Chinese Scholarship Council, and A.T. was supported by the Weizmann Institute of Science - National Postdoctoral Award Program for Advancing Women in Science. Work at the Molecular Foundry was supported by the Director, Office of Science, Office of Basic Energy Sciences, Division of Materials Sciences and Engineering, of the U.S. Department of Energy under Contract No. DE-AC02-05CH11231.

## Author information

### Affiliations

1. #### The Molecular Foundry, Lawrence Berkeley National Laboratory, Berkeley, CA, 94720, USA

• Bining Tian
• , Angel Fernandez-Bravo
• , Nicole A. Torquato
• , M. Virginia P. Altoe
• , Ayelet Teitelboim
• , Cheryl A. Tajon
• , Yue Tian
• , Nicholas J. Borys
• , Edward S. Barnard
• , Emory M. Chan
• , P. James Schuck
•  & Bruce E. Cohen
2. #### Department of Electrical Engineering and Computer Sciences, University of California, Berkeley, Berkeley, CA, 94720, USA

• Hossein Najafiaghdam
3. #### Department of Radiation Oncology, University of California, San Francisco, San Francisco, CA, 94158, USA

• Nicole A. Torquato
•  & Mekhail Anwar
4. #### Department of Mechanical Engineering, Columbia University, New York, NY, 10027, USA

• P. James Schuck

### Contributions

B.E.C., P.J.S., and E.M.C. conceived the project and supervised the research. B.T., N.A.T., and C.A.T. synthesized the nanocrystals. B.T., A.F-B., A.T., Y.T., N.J.B., and E.S.B. conducted the optical experiments; M.V.P.A. and B.T. carried out the EM and other characterization; N.A.T., H.N., and M.A. conducted the animal imaging. E.M.C. carried out the calculations and modeling. B.T. and B.E.C. prepared the figures; B.T., E.M.C., P.J.S., and B.E.C. analyzed the results and wrote the manuscript. All authors participated in discussion and editing of the manuscript.

### Competing interests

The authors declare no competing interests.

### Corresponding authors

Correspondence to Emory M. Chan or P. James Schuck or Bruce E. Cohen.