nature photonics

Article [Link]

Size-dependent lanthanide energy transfer

amplifies upconversion luminescence
quantum yields

Received: 8 May 2023 Feng Li1,2,8, Langping Tu3,8, Yuqi Zhang1,2, Dingxin Huang1,2, Xingxu Liu1,2,
Xiaorong Zhang1,2, Jiarui Du1,2, Rongwei Fan4, Chunhui Yang1,2, Karl W. Krämer5,
Accepted: 17 January 2024
Jose Marques-Hueso 6,7 & Guanying Chen 1,2
Published online: xx xx xxxx

Check for updates Optical upconversion from lanthanide-doped nanoparticles is promising

for a variety of applications ranging from bioimaging, optogenetics,
nanothermometry, super-resolution nanoscopy and volumetric displays to
solar cells. Despite remarkable progress made in enhancing upconversion
to fuel these applications, achieving luminescence of upconversion
nanoparticles (UCNPs) that is comparable to or higher than the bulk
counterparts has been challenging due to nanoscale-induced quenching
effects. Here we demonstrate a size-dependent lanthanide energy transfer
effect in a conceptual design of hexagonal sodium yttrium fluoride (NaYF4)
core–shell–shell NaYF4@NaYF4:Yb/Tm@NaYF4 UCNPs with depleted
surface quenching. We show that precise control over the domain size (or
the thickness of the middle shell doped with ytterbium (Yb) and thulium
(Tm) from 1.2 to 13 nm) increases the lanthanide energy transfer efficiency
(from 30.2 to 50.4%) and amplifies the upconversion quantum yield
to a high value of 13.0 ± 1.3% in sub-50 nm UCNPs (excitation: 980 nm,
100 W cm−2), which is around fourfold higher than the micrometre-scale
hexagonal NaYF4:Yb/Tm bulk counterparts. Spectroscopic studies and
theoretical microscopic modelling reveal that long-range lanthanide energy
transfer (>9.5 nm) takes place and underlies the observed size-dependent
phenomena. Demonstration of size-dependent lanthanide energy transfer
and upconversion quantum yields at the nanoscale transforms our
long-existing conceptual understanding of lanthanide energy transfer
(size independence), thereby having important implications for
applications of lanthanide nanophotonics and biophotonics.

Lanthanide-doped upconversion nanoparticles (UCNPs) are an and solar cells10. UCNPs typically consist of a low-phonon-energy mate-
emerging class of anti-Stokes emitters1, which have found important rial as the host lattice (such as hexagonal sodium yttrium fluoride
applications in a broad spectrum of prominent fields2, including (β-NaYF4, ~350 cm−1 phonon energy)) that is doped with ytterbium (Yb)
anti-counterfeiting3, biosensing4, microlasers5, deep-tissue bioimag- ions as sensitizers, which absorb infrared radiation and non-radiatively
ing6, super-resolution nanoscopy7, optogenetics8, volumetric displays9 transfer their excitation to doped activator ions such as thulium (Tm)

A full list of affiliations appears at the end of the paper. e-mail: chenguanying@[Link]

Nature Photonics
Article [Link]

or erbium (Er) to produce shorter-wavelength ultraviolet, visible or a c

near-infrared (NIR) upconversion luminescence (UCL)11. The perfor- Inert shell
NaYF4 shell
mance of technological applications crucially hinges on the ability to d thickness ≈ 7 nm
produce bright UCL or high upconversion quantum yields (UCQYs) at
Inert core
low excitation light irradiance. Despite substantial progress made in NaYF4 diameter
enhancing UCL through the core–shell structure design12–14, plasmonic ≈ 5.5 nm

enhancement15–17, organic dye sensitization18–20, deactivation of internal Active shell

defects21,22 and the high laser-irradiance-diminished concentration NaYF4:Yb,Tm shell
thickness (d) varied
quenching of activators23, achieving UCQYs in nanoscale upconverting from 1.2 to 13 nm
materials that are comparable to or higher than the bulk counterparts d Yb
has long been challenging due to nanoscale-induced quenching effects Y
(mainly surface quenching) and has not been realized since the estab- b
lishment of the UCNP concept two decades ago.
Size is a fundamental parameter of nanostructure materials.
Although manipulating the particle size has enabled the production
of size-tunable optical spectra in metals and semiconductors24,25, the
production of optical quantum confinement has been precluded in com- 10 20 30

monly seen UCNPs (size ≥ 20 nm), as 4f electronic wavefunctions are local- Position (nm)
ized at the atomic scale (ångströms) in the lanthanide ions incorporated e
at the lattice sites26. Still, the size effect has been shown to be of essential
importance to UCQYs in UCNPs through size-restricted phonon relaxa-

00 m
tions27,28, size-confined energy migrations29, size-induced distortion of the

(1 3 n
)
5
crystal lattice30 and size-induced surface quenching effects31. All of these

0.
efforts have shed light on the important impact of size variation on the
upconverted luminescence in lanthanide-doped nanocrystals. Among
them, size-induced surface quenching has long been perceived to be the
100 nm
most pronounced effect that dictates the UCQYs for a given nanostruc-
ture, as the nanoscale size directly exposes a number of surface lanthanide Fig. 1 | Characterization of the as-synthesized core–shell–shell NaYF4@
dopants to surface quenching centres (surface defects, organic ligands, NaYF4:Yb/Tm@NaYF4 nanostructure. a, Schematic illustration of the
designated core–shell–shell nanostructure with colour-coded distinct domains
solvent molecules and so on) or indirectly couples excited lanthanides
in which the core, middle and outermost layers are highlighted in yellow, red and
at the centre of the UCNPs through energy migration to surface quench-
cyan, respectively. b, HAADF-STEM image of NaYF4@NaYF4:99% Yb,1% Tm@
ers32. Coating the UCNPs with an optically inert shell can passivate the
NaYF4 core–shell–shell nanoparticles. c,d, HAADF-STEM image (c) and EDS
surface and spatially isolate the doped lanthanide ions from surface elemental line scan (d) (obtained along the dashed line in c) of a representative
quenchers, thereby enhancing the UCQYs12. For example, shelling 30 nm single core–shell–shell nanoparticle, showing the spatial distribution of the
NaYF4:20% Yb,2% Er UCNPs with an inert NaYF4 layer triples the UCQY Y and Yb elements. e. High-resolution TEM image of a single core–shell–shell
value to 0.3%, which is comparable to that of 100 nm NaYF4:20% Yb,2% Er nanoparticle, with the Fourier-transform diffraction pattern shown in the inset.
UCNPs and is 60-fold higher than that of 10 nm NaYF4:20% Yb,2% Er Scale bar, 10 nm, for c and e.
UCNPs (0.005%) (excitation: 980 nm; power density (PD): 150 W cm−2)33.
Optimizing the thickness of the inert shell has been shown to enable pre-
cise control over the surface quenching effects34, thereby enhancing the the active domain size effect (Fig. 1a), as the controllable synthesis of
UCQYs by up to 10,000-fold (1.7%; excitation: 980 nm; PD: 340 W cm−2) pure core β-NaYF4:Yb/Tm UCNPs with various Yb dopant concentrations
when shelling sub-5 nm UCNPs with an optically inert thick shell of 10 nm but identical nanoparticle sizes is non-trivial (the Yb doping concentra-
(ref. 35). Despite these important advances in understanding the size tion substantially alters the resulting nanoparticle size)29. We chose the
effect as well as the suppression of detrimental surface quenching, UCQYs NaYF4 bare core (fixed diameter: 5.5 nm) as the seed and utilized an epi-
in nanoscale upconversion of varying sizes have typically been reported taxial growth method to successively grow the active middle shell layer
to be orders of magnitude (~10–1000-fold) lower than that of the bulk (NaYF4:Yb/Tm, various thicknesses: ~1.2–13 nm) and then the outermost
materials (several percentages, 3–10%), with rare exceptions close to the inert shell layer (NaYF4, fixed thickness: ~7 nm) (Supplementary Section 1).
bulk but for unidentified reasons21,36. This discrepancy highlights a lack This strategy ensures that the volume (or the shell thickness) of the opti-
of general understanding about the size effect on UCQYs, which prevents cally active domain (the Yb/Tm-doped nanoscale domain) is identical
further improvement of the UCNP brightness to meet the demands of when varying the doped Yb concentration. Atomic-number-contrasted
practical technological applications. high-angle annular dark-field scanning transmission electron micros-
In this work we investigate a fundamental ‘size’ effect in nanoscale copy (HAADF-STEM) imaging confirmed the successful preparation of
lanthanide upconversion that is beyond the well-established size-induced the designated core–shell–shell structure, with clear distinguishment
surface quenching effect in a set of given nanostructures with precisely between the yttrium (Y) atoms (dark) and Yb atoms (bright) spatially
defined lanthanide compositions. We show that the increase in the distributed in a defined way (Fig. 1b). This was substantiated further via
active domain size escalates the lanthanide energy transfer process, the energy dispersive X-ray spectroscopy (EDS) elemental line scanning
underpinned by long-range lanthanide interactions, in a conceptual of a single core–shell–shell nanoparticle in Fig. 1c,d, in which the Y ele-
design of core–shell–shell NaYF4@NaYF4:Yb/Tm@NaYF4 UCNPs with ment is located in the core and the outermost shell layer, while Yb atoms
depleted surface quenching, which amplifies the UCQYs in sub-50 nm are strictly confined in the middle shell layer, in line with measured EDS
UCNPs beyond the micrometre-scale bulk crystal counterparts. elemental mapping results (Supplementary Fig. 1a). Despite variation
of the intermediary layer thickness from thin (1.2 nm) to thick (13 nm),
Results and discussions the apparent core–shell–shell structures were clearly discerned in the
Core–shell–shell nanostructure design HAADF-STEM images (Supplementary Fig. 1b), indicating limited lantha-
We adopted a sandwiched core–shell–shell structure of NaYF4@ nide intermixing at the core/shell interfaces in the designated UCNPs
NaYF4:x% Yb,1% Tm@NaYF4 (x = 20, 30, 50, 70, 89 and 99) to investigate of varying sizes. The high-resolution TEM image (Fig. 1e) confirmed the

Nature Photonics
Article [Link]

a d

3
15 F3
3
H4

Energy (103 cm–1)

2 2
F5/2 F5/2
10 Inert 3
Inert
H5
d core shell
3
F4

800 nm
5

2 2
0 F7/2 3
H6 F7/2
Yb3+ Yb3+ Tm3+ Yb3+ Yb3+

b
d = 1.2 nm
3 3 2 2
d = 2.5 nm H4 H6 F7/2 F5/2
d = 4.5 nm
d = 9.5 nm
Intensity (a.u.)

Absorbance
d = 13 nm ×200
1
1
1

G4
D2
D2

3 3
F3 H6
1

3
I6

3
3

1 3
F4

G4 F4
H6
H6
3
F4

300 375 450 525 600 675 750 825 900 975
Wavelength (nm)

c d 10,000
1
Excitation Emission Cross-relaxation I6 (345 nm)
3
P2 1
D2 (360 nm + 450 nm)
3
P
1 0,1
1
35 G4 (475 nm + 650 nm)
I6
3
1,000 F3 (700 nm)
30 3
H4 (800 nm)
Enhancement factor

1
D2
Abs (975 nm)
Energy (103 cm–1)

25
CR1
1
G4 100
360 nm

20
650 nm

450 nm

345 nm
475 nm

3
15
CR2 F
3 2
F
3 3
H4 10
800 nm

700 nm

10 3
CR3 H5
3
5
F4

3
H6 1
0
Tm3+ 1.2 2.5 4.5 9.5 13.0
d (nm)

Fig. 2 | Optical upconversion in the designated core–shell–shell naked eye at the corresponding wavelengths. All of the measured spectra were
nanostructure with a tunable intermediate shell layer thickness. calibrated in scale by the instrument spectral response. c, Schematic diagram
a, Schematic illustration of simplified lanthanide energy levels as well as the depicting the step-wise upconverting pathways (via energy transfer from Yb
involved upconversion processes in the core–shell–shell nanostructure, with to Tm), the observed radiative transitions and the typical cross-relaxations of
the different domains highlighted in distinct background colours (inert core, Tm activators in its simplified energy levels under 980 nm excitation. The black
light green; active intermediate shell, light yellow; outermost inert shell, light dashed arrows, coloured solid arrows and grey dashed arrows represent the
blue); d represents the thickness of the active intermediate shell layer. b, UCL upconverting (grey background), observed emission (peach background) and
(solid line) and absorbance (line with data-point symbols) spectra of NaYF4@ Tm–Tm cross-relaxation processes (light green background), respectively.
NaYF4:99% Yb,1% Tm(d)@NaYF4 core–shell–shell nanoparticles (dispersed d, The enhancement factors of UCL from the relevant Tm emitting energy levels
in hexane) with various thicknesses (d); the spectra between 300 and 700 nm versus the intermediate layer thickness (compared with that of the sample with
are multiplied 200-fold to facilitate comparison with the NIR UCL at 800 nm. d = 1.2 nm). Excitation: 980 nm, PD: 20 W cm−2. Abs, absorbance.
The coloured band across the top represents the percieved colour as seen by

highly crystalline nature of the formed core–shell–shell UCNPs, with the Effect of the active intermediate layer thickness on the UCQY
lattice fringe spacing corresponding to the (100) planar distance of hex- Next, we investigated the influence of the thickness (d) of the Yb/Tm
agonal phase β-NaYF4 ( JCPDS: 16-0334). Indeed, the X-ray diffraction pat- active intermediate layer on the UCQY of the designated core–shell–
terns give clear evidence that the core, core–shell and core–shell–shell shell NaYF4@NaYF4:99% Yb,1% Tm@NaYF4 UCNPs (Supplementary
nanostructures are all pure hexagonal crystal phase (Supplementary Sections 2–4). To exclude the surface quenching effect caused by
Fig. 2). Moreover, the resulting core–shell–shell UCNPs with differing the nanoscale size, an ultrathick inert shell (thickness: 7 nm) was
Yb concentrations and varying thicknesses in the middle shell were adopted as the outermost layer in all of the investigated core–shell–
all observed to have a uniform size and spherical shape, confirming shell UCNPs. Successful depletion of the surface quenching was veri-
the preparation of high-quality UCNPs with a defined core–shell–shell fied by two facts: (1) a saturated and maximized UCL intensity from
nanostructure (Fig. 1 and Supplementary Figs. 3–8). Tm was observed at 800 nm when the thickness of the outermost shell

Nature Photonics
Article [Link]

a 16
b 16
c
99% Yb,1% Tm PD = 100 W cm
−2
UCNPs
6
d = 1.2 nm 5% Yb 10
d = 2.5 nm 20% Yb (CdSe@ZnS)
d = 4.5 nm 40% Yb
12 12

Brightness (M–1 cm–1)

d = 9.5 nm 70% Yb
d = 13 nm 99% Yb 5
10
UCQY (%)

UCQY (%)
(AIEgen)
8 8
4
10 (ICG) (Polymer)

4 4 (AgInSe2)
(MB) (Carbon dots)
3
10 (Ag2S)

0 0
I II III IV
nm

nm
nm

nm

nm
1.2

5
Fluorophores

13
0 20 40 60 80 100 0 3 6 9 12 =

9.
2.

4.
d

=
=
=

d
d
d

d
PD (W cm−2) d (nm)

d e
1–32 reported value

This work
10.00
12

19 1.00
UCQY (%)

UCQY (%)
8
29
30
21 0.10
4 15 25 17 NaYF4@99% Yb,1% Tm@NaYF4
23 11
22 24 Bulk materials (3–5 µm)
0.01
18
10 14 9283 20 2627 7
0 5 12 13 16 8 32 631

0.1 1.0 10.0 100.0 1,000.0 10,000.0 0 20 40 60 80 100

PD (W cm−2) PD (W cm−2)

Fig. 3 | The d-value-related UCQYs. a, Measured PD-dependent UCQYs of error bar. d, Comparison of the highest achieved UCQY with previously
the core–shell–shell NaYF4@NaYF4:99% Yb,1% Tm(d)@NaYF4 nanostructure, reported values in nanoscale lanthanide upconversion at PDs ranging from
with d varied from 1.2 to 13 nm. b, Measured d-dependent UCQYs of the core– 0.3 to 10,000 W cm−2. The numbers 1–32 refer to the number order used in
shell–shell NaYF4@NaYF4:x% Yb,1% Tm@NaYF4 nanostructure (x = 5, 20, 40, Supplementary Table 4, and the unnumbered symbol denotes that achieved
70 and 99) with various Yb concentrations (PD = 100 W cm−2). c, Brightness in this work. The symbols are colour-coded, where green represents UCNPs
comparison of NaYF4@99% Yb,1% Tm(d)@NaYF4 UCNPs (d = 1.2, 2.5, 4.5, 9.5 doped with Er activators under 980 nm excitation, red represents UCNPs
and 13 nm) with four types of commonly seen high-efficiency single-photon doped with Er activators under 800 nm excitation and blue represents
fluorophore: I: dye (ICG (indocyanine green), MB (methylene blue) and UCNPs doped with Tm activators under 980 nm excitation. A detailed
an AIEgen (aggregation-induced emission luminogen)); II: quantum dots comparison is summarized in Supplementary Table 4. e, Comparison of
(core–shell CdSe@ZnS, AgInSe2 and Ag2S); III: polymer; and IV: carbon dots. the UCQY values of micrometre-scale NaYF4:25% Yb,0.3% Tm bulk materials
The symbols are shown as two different colours, in which dark red represents (size ~3–5 µm) and NaYF4@NaYF4:99% Yb,1% Tm(d)@NaYF4 UCNPs (d = 13 nm)
emissions in the NIR region (750–1,000 nm), and blue denotes the emissions under different PD values. For a, b, d and e, the UCQY data points are shown as
in the visible region (400–700 nm). The centre of the symbol denotes the the mean ± standard deviation (n = 3).
brightness mean, with the height of the symbol for UCNPs indicating the

was >5 nm (Supplementary Fig. 9a); and (2) an unchanged Yb lumi- UCQY (integrated from 300 to 850 nm) increases continuously with an
nescence decay lifetime at 1,000 nm was observed on varying the Yb increasing d value. Note that an important size effect opposite to this
concentration in the core–shell–shell nanostructure with an outermost was observed previously29, in which the reduction of the intermediate
shell thickness of >5 nm (Supplementary Fig. 9b). Interestingly, on shell thickness led to an enhancement of ultraviolet upconversion due
increasing the Yb/Tm-doped intermediate layer thickness (d) from to confined energy migration. We believe that the different observa-
1.2 to 13 nm, both the absorbance (at around 980 nm) and the UCL tion lies in the adoption of a different outermost shell thickness, as
spectrum of the core–shell–shell nanostructure were enhanced but the adoption of an ultrathick shell here (7 nm versus 2 nm) enables the
on different scales (Fig. 2b). The UCL enhancement factors for the complete removal of the surface quenching effect (Supplementary
transitions 1I6 → 3F4 (346 nm), 1D2 → 3H6 and 1D2 → 3F4 (360 and 450 nm, Fig. 9).
respectively), 3F3 → 3H6 (700 nm) and 3H4 → 3H6 (800 nm) from different To provide compelling evidence, we also measured the overall
emitting levels of Tm were observed to substantially surpass the absorb- UCQYs of the core–shell–shell NaYF4@NaYF4:99% Yb,1% Tm@NaYF4
ance enhancement factor for d > 1.2 nm. Specifically, for d = 13 nm, the UCNPs dispersed in hexane using an integrating sphere method when
absorbance was increased by 89-fold, but the UCL intensities for the varying the PD from 5 to 100 W cm−2 (Fig. 3a). At each given PD, the
1
I6, 1D2, 3F3 and 3H4 energy levels were improved 3,000-, 1,557-, 264- and measured UCQYs were observed to increase with d, which is consistent
152-fold, respectively (Fig. 2d). An exception is for UCL from the 1G4 state with the observations in Fig. 2d. For d ≤ 2.5 nm, the UCQY values plateau
(for the 1G4 → 3H6/3F4 transitions at 475/650 nm), which was enhanced at PD = 40 W cm−2, whereas for d ≥ 4.5 nm, the UCQYs still increase
58-fold (moss green line), less than the assumed absorbance-induced slowly with a PD of 40–100 W cm−2, achieving the highest UCQY value of
increase (89-fold). It is possible for this to be explained in terms of 13.0 ± 1.3% for d = 13 nm and PD = 100 W cm−2 in sub-50 nm UCNPs (size:
Tm–Tm cross-relaxation (CR) processes, in which the d-enhanced 45.5 nm). Furthermore, the ratio of the UCQY values (for d = 13 nm) to
1
G4 + 1G4 → 1D2 + 3F2 (CR1) and 1G4 + 3F4 → 3F2 + 3H4 (CR2) processes depop- the UCQY values (for d = 1.2 nm) decreases with increasing PD, and
ulate 1G4 at a higher rate than that of the absorption-induced popu- shows a more pronounced size effect at low PD values (Supplemen-
lation enhancement (Fig. 2c). These results indicate that the overall tary Fig. 10). We also investigated the effect of the intermediate layer

Nature Photonics
Article [Link]

a b
Core Shell 1 Shell 2 Shell 3 Yb@Y(d)@4% Tm@Y
300

kYb–Tm (s–1)
100

Long-range interaction

NaYbF4 NaYF4 NaYF4:4% Tm

3+
Yb Tm3+
10
2 4 6 8 10
d (nm)
c d 1.00 e
d = 1.2 nm d = 1.2 nm
20% Yb
d = 9.5 nm d = 4.5 nm
d = 9.5 nm 50 99% Yb
d = 13 nm
log[intensity (a.u.)]
0.10
Y Y
Yb,Y Yb,Y 40
Y Y
Intensity (a.u.)

ηET (%)
0.01 30
1.00 d = 1.2 nm
d = 1.2 nm
d = 4.5 nm d = 4.5 nm
d = 9.5 nm d = 9.5 nm 20
d = 13 nm d = 13 nm
0.10
Y Y
10
Yb,Tm Yb,Tm
Y Y

0.01 0
975 1,050 1,125 0 2,000 4,000 6,000 8,000 1.2 4.5 9.5 13.0
Wavelength (nm) Time (µs) d (nm)

Fig. 4 | Observation of d-dependent lanthanide energy transfer in the NaYF4 nanostructure (d = 1.2 and 9.5 nm) (top) and in the core–shell–shell
designated core–shell–shell nanostructure. a, Schematic illustration of the NaYF4@NaYF4:20% Yb,1% Tm(d)@NaYF4 nanostructure (d = 1.2, 4.5, 9.5 and
core–shell–shell–shell NaYbF4@NaYF4(d)@NaYF4:4% Tm@NaYF4 utilized 13 nm) (bottom). Core: 5.5 nm in diameter; outermost shell: 7 nm in thickness.
to unveil long-range energy transfer from Yb to Tm. Core NaYbF4: 10.5 nm in Excitation: 975 nm, and the spectra are normalized by the Yb absorption
diameter; first NaYF4 shell: various thickness (d) values; second NaYF4 shell: intensity at 975 nm. d, Luminescence decay curves of Yb at 1,020 nm in NaYF4@
4% Tm, of 2.5 nm thickness; outermost NaYF4 shell: 7–8 nm in thickness. NaYF4:20% Yb(d) nanoparticles (top) and NaYF4@NaYF4:20% Yb,1% Tm(d)@
b, The calculated d-dependent Yb–Tm energy transfer rates. c, Yb NaYF4 (bottom) with d = 1.2, 4.5, 9.5 and 13 nm (excitation: 975 nm). e, The
luminescence spectra in the core–shell–shell NaYF4@NaYF4:20%Yb(d)@ calculated d-dependent ηET values from Yb to Tm.

thickness (d) on the UCQY in a set of NaYF4@NaYF4:x% Yb,1% Tm@ B = ε × QY, (1)
NaYF4 UCNPs with various Yb concentrations (x = 5, 20, 40, 70 and
99) (PD = 100 W cm−2). At all given Yb concentrations, the measured where ε is the molar absorption coefficient of the fluorophore at the exci-
UCQY increases with the d value, with a more prominent increase at tation wavelength, QY is the measured luminescence quantum yield value
higher Yb concentrations and levelling off at d = 9.5 nm, implying a and B is the brightness (defined as fluorescence output per mole of fluo-
threshold around this thickness value for amplifying the UCQY (Fig. 3b). rophore). Four different types of single-photon fluorophore are included
The achieved UCQY of 13.0 ± 1.3% compares favourably with reported for comparison: quantum dots (Ag2S, AgInSe2 and core–shell CdSe/
UCQY values in nanoscale lanthanide upconverting materials (of vary- ZnS), a fluorescent polymer (N-{2-[(7-N,N-dimethylaminosulfonyl)-
ing host materials and sizes) in a wide PD range of 0.3–10,000 W cm−2, 2,1,3-benzoxadiazol-4-yl](methyl)amino}ethyl-N-methylacrylamide),
thereby producing considerable upconversion brightness (Fig. 3d carbon dots and small organic dye molecules (indocyanine green,
and Supplementary Tables 1–4). Importantly, the measured methylene blue and an aggregation-induced emission luminogen)
UCQYs in sub-50 nm NaYF4@NaYF4:99% Yb,1% Tm@NaYF4 UCNPs (Fig. 3c). We found that the brightness of the core–shell–shell UCNPs
(d = 13 nm) are apparently higher than those of micrometre-scale with a thickness-defined diameter from 21.8 to 44.3 nm is comparable
(3–5 µm) hexagonal phase NaYF4:Yb/Tm bulk crystals prepared via to commonly used bright single-photon fluorophores with emissions
the well-established high-temperature solid-state reaction with opti- in either the visible or NIR region, highlighting the bright future of
mized lanthanide concentrations for all of the investigated PDs, and these UCNPs for technological applications. We determined that the
have 3.7-fold higher UCQY values at PD = 100 W cm−2 (Fig. 3e)11. Note largest individual nanocrystal contains approximately 193,347 Yb ions
that no noticeable PD-induced temperature effect on the UCQYs was and 1,953 Tm ions. Despite upconversion being a nonlinear process,
observed (Supplementary Section 4). A detailed characterization of the the substantial quantities of both lanthanide sensitizers and emitters
micrometre-scale bulk crystal material is included in Supplementary enable the UCNPs to exhibit a brightness that is comparable to that of
Fig. 11. single-photon fluorophores. Details for calculation of the molecular
Brightness is an important parameter that is directly related to the extinction coefficient and the brightness of the UCNPs are included
performance of luminescent materials in technological applications. in Supplementary Section 5, and the brightness-relevant compari-
We also compared the UCL brightness of the core–shell–shell NaYF4@ son parameters (that is, the number of lanthanide ions per UCNP, the
NaYF4:99% Yb,1% Tm@NaYF4 UCNPs of various active layer thicknesses molecular extinction coefficient, the QY values and the excitation and
(PD = 100 W cm−2) with that of commonly seen single-photon highly emission wavelengths) of the UCNPs and these commonly seen fluo-
efficient fluorophores based on the reported formula37: rophores are compiled in Supplementary Tables 1–3.

Nature Photonics
Article [Link]

a b
Sim1 (long-range)
8
Sim2 (short-range)
Exp
7

UCQY (%)
5

2
Y3+ Pumped Yb3+ Yb3+ Tm3+

Nearest energy transfer Long-range energy transfer 0 2 4 6 8 10 12 14

d (nm)

c d = 2.4 nm d = 4.0 nm d = 10.4 nm

Tm
Probability (%)

20 20 20

0 0 0

Probability (%)
30
10 10 10
2 8 ) 2 8 2 8
)
Pos
4
6
6
(nm Pos
4
6
6
(nm Pos
4
6
6
n m) 15
n(
4
ion
4 4
itio 8
sit
itio 8 ion itio 8
n (n 10
2
Po
n (n
m)
2
sit n (n 2
s itio
m) 10
Po m) 10
Po 0

80 80 80
Yb
Probability (%)

40 40
40

0 80

Probability (%)
0
0
10 10 10
8 8 2 8 ) 50
2 ) 2 )
nm 6 4 6 nm
(nm
6 Pos
Pos 4 6 n( Pos
4
6 4 itio
6 4 n(
itio
4
itio ion
io itio n (n 8 2
n (n 8 2 sit n (n 8 2 sit m) 10
Po
s 20
m) 10 Po m) 10 Po

Fig. 5 | Microscopic simulation of d-dependent UCQYs. a, Scheme depicting and Tm into account, and Sim2 represents the simulated d-dependent UCQYs
two types of basic microscopic energy transfer process. Top: short range considering only next-neighbour energy transfer (traditional energy transfer).
(nearest ion pair). Bottom: long range (spatially isolated ion pair with a distance The experimental (Exp) UCQY data points are shown as the mean ± standard
of several to ten nanometres between them). b, Monte Carlo simulation results deviation (n = 3). c,d, The probability of finding Tm in its excited state (3H4:
for the measured d-dependent UCQY values of the core–shell–shell NaYF4@ 800 nm emission) (c) and Yb in its excited state (2F5/2: 980 nm emission)
NaYF4:99% Yb,1% Tm(d)@NaYF4 nanostructure (red spheres) using two models, (d) plotted against the layer thickness d in the core–shell–shell NaYF4@
with the simulated results shown as solid thick lines. Sim1 represents the NaYF4:20% Yb,1% Tm(d)@NaYF4 nanostructure.
simulated d-dependent UCQYs taking long-range energy transfer between Yb

Mechanistic investigation of the active layer τYb−Tm

ηET = 1 − , (2)
thickness-dependent UCQYs τYb
To understand the underlying mechanism of the d-dependent UCQY
values in the core–shell–shell NaYF4@NaYF4:Yb/Tm@NaYF4 UCNPs, where τYb–Tm and τYb are the measured Yb luminescence decay lifetimes,
we measured the Yb luminescence spectra (Fig. 4c) and decay profiles respectively, with and without the activator Tm. Indeed, increasing
(Fig. 4d) of NaYF4@NaYF4:20% Yb,1% Tm(d)@NaYF4 and utilized the the d value from 1.2 to 13 nm results in an increase in ηET from 24.7%
NaYF4@NaYF4:20% Yb(d)@NaYF4 UCNPs without Tm doping as control to 42.0% (Fig. 4e), which underlies the observed growth trend of the
samples. Compared with the control samples with a d-independent d-dependent UCQY values for the 20% Yb concentration in Fig. 3b (dark
Yb luminescence intensity and decay, both the luminescence inten- red line). Likewise, similar results were observed for the Yb dopant
sity and decay decrease with an increase in the active layer thickness concentration of 99%, in which the ηET increases from 30.2% to 50.4%
(d), clearly indicating the d-enhanced energy transfer efficiency from when increasing d from 1.2 to 13 nm (Fig. 4e and Supplementary Fig. 12).
Yb to Tm. We quantified the energy transfer efficiency (ηET) using The significance of our findings for the d-dependent ηET in a given
equation (2)38: nanostructure with a fixed lanthanide composition is that it breaks an

Nature Photonics
Article [Link]

a d

4
F7/2
20 2
H11/2
4
S3/2

Energy (103 cm–1)

4
F9/2
Inert Inert
d core 4
I9/2 shell
2 2
F5/2 4
I11/2 F5/2
10

4
I13/2

550 nm

655 nm
2 2
0 F7/2 4
I15/2 F7/2
Yb3+ Yb3+ Er3+ Yb3+ Yb3+

b c d
4
S3/2 20% Yb,2% Er 40% Yb,2% Er 20% Yb
d = 1.2 nm d = 1.2 nm 4 40% Yb
4
I15/2 d = 2.5 nm d = 2.5 nm
d = 4.5 nm d = 4.5 nm
d = 9.5 nm d = 9.5 nm 3
Intensity (a.u.)

Intensity (a.u.)

I550 nm/I655 nm
2
2
H11/2
4
F9/2
4
I15/2 1
4
I15/2

0
500 550 600 650 700 500 550 600 650 700 1.2 2.5 4.5 9.5
Wavelength (nm) Wavelength (nm) d (nm)

e f
6 20% Yb,2% Er 6 40% Yb,2% Er
Green Green
5 Red 5 Red
Total Total
4 4
UCQY (%)

UCQY (%)

3 3

2 2

1 1

0 0
2 4 6 8 10 2 4 6 8 10
d (nm) d (nm)

Fig. 6 | The d-dependent UCQYs using Er as the activator. a, Schematic solid green and red lines denote upconverison luminescence. b,c, UCL spectra of
illustration of the adopted core–shell–shell NaYF4@NaYF4:Yb/Er(d)@NaYF4 NaYF4@NaYF4:x% Yb,2% Er(d)@NaYF4, where the concentration of Yb (x) is 20%
nanostructure, simplified Yb and Er energy levels and the upconverting pathways (b) and 40% (c) for various active layer thicknesses. d, The green/red emission
involved. The different domains in the adopted nanostructure are highlighted in intensity ratio (I550 nm/I655 nm) versus the active layer thickness. Each ratio is given
different background colours (inert core, light green; active intermediate shell, as the mean ± standard deviation (n = 3). e,f, Measured overall UCQY values,
pink; outermost inert shell, light blue); d represents the thickness of the active green emission UCQYs and red emission UCQYs as a function of the active layer
intermediate shell layer; the dashed lines and curly arrows represent energy thickness for 20% Yb (e) and 40% Yb (f). Excitation: 980 nm; PD: 60 W cm−2. Each
transfer and non-radiative multiphonon relaxations, respectively, and the thick UCQY data point is shown as the mean ± standard deviation (n = 3).

empirical understanding of the microscopic lanthanide-to-lanthanide lanthanide-to-lanthanide interaction strength, ηET, and the UCQYs
interaction in upconversion. According to the classic theories devel- should be mainly determined by the doped lanthanide concentrations
oped by Förster39 and Dexter40, direct long-range energy transfer (from and be independent of the particle size (given that surface quenching
a few up to ten nanometres) is conceivable as a result of the long-lived has been precluded), in contrast to our experimental observations. In
metastable states of the lanthanide ions. However, the significance of other words, this contradiction implies that direct long-range energy
this process in lanthanide upconversion remains uncharted. In typical transfer, for two lanthanide ions separated by several nanometres
Yb-sensitized UCNPs co-doped with X as the activator (X = Tm, Er or in space, needs to be taken into consideration. To demonstrate the
Ho, for example), it is commonly assumed that upconversion takes existence of direct long-range energy transfer experimentally, we
place through energy migration within the Yb sublattice (covering utilized sodium ytterbium fluoride (NaYbF4) as a core to synthesize a
several nanometres) before transferring the excited energy to the core–shell–shell–shell NaYbF4@NaYF4(d)@NaYF4:4% Tm@NaYF4
nearest activator X41,42. On the basis of this understanding, the average nanostructure with depleted surface quenching (Fig. 4a), in which the

Nature Photonics
Article [Link]

sensitizer Yb and activator Tm are spatially separated into different Fig. 17) demonstrate the active domain size-induced enhancement of
domains (Yb: core; Tm: the second shell layer) and with the Yb–Tm Yb–Er interactions. The decreased green/red emission intensity ratio
distance well defined by the thickness d of the first shell layer (Supple- can possibly be ascribed to the fact that the process of populating the
mentary Fig. 13). Compared with that of the control sample without the red-emitting level (4F9/2) is more sensitive to the change in the Yb–Er
Tm activator, we observed a shortened Yb luminescence lifetime in all interaction strength than that for the green-emitting level (2H11/2/4S3/2)43.
of the samples with various d values from 1.2 to 9.5 nm, which demon- We noted that the NaYF4@NaYF4:Yb/Er@NaYF4 UCNPs exhibit lower
strates the existence of Yb–Tm long-range energy transfer (>9.5 nm). We UCQYs than the NaYF4@NaYF4:Yb/Tm@NaYF4 UCNPs, possibly due to
calculated the donor–acceptor energy transfer rate (kYb–Tm) following more pronounced energy back-transfer from Er to Yb than from Tm to
Förster energy transfer theory (Supplementary Fig. 14), and plotted Yb (refs. 14,42). Moreover, the observed UCQYs are lower than those of
it against the layer thickness d (Fig. 4b). Although a decrease in kYb–Tm Yb/Er-doped β-NaYF4 bulk materials, which can potentially be attrib-
was shown with an increase in d, a considerable kYb–Tm value can still uted to the existence of more quenching sites (internal crystal defects,
be observed at d = 9.5 nm, which indicates a non-negligible long-range OH groups) within the UCNPs than in the bulk materials21,22. Careful
interaction strength between the lanthanide ions. Importantly, we removal of the OH groups during UCNP synthesis, using techniques
need to note that, for a given Yb ion, the number of Yb–Tm ion pairs is such as post-annealing to minimize internal crystal defects, and reduc-
scaled up with the square of the inter-ionic distance (RYb–Tm)2, thereby ing back-energy transfer from the activator to the sensitizer can further
rendering a notable long-range energy transfer process even at low enhance the observed UCQYs in Yb/Er-doped UCNPs and beyond.
kYb–Tm (Supplementary Fig. 15). This conclusion could possibly extend To rule out the possible influence of size-dependent UCQYs on the
to other lanthanide–lanthanide ion interaction, such as Yb–Er ion adopted core–shell–shell nanostructure, we also prepared the most
interaction, in UCNPs shown in Supplementary Figs. 16–20. widely investigated core–shell NaYF4:Yb/Tm(d)@NaYF4 UCNPs with
Next, Monte Carlo simulations were adopted to further cor- different core diameter sizes. As expected, size-dependent UCQYs were
roborate that long-range energy transfer underlies the observed also observed, confirming the fundamental nature of the observed
size-dependent energy transfer and UCQY values. As shown in size effect in nanoscale lanthanide upconversion (Supplementary
our previous studies, the Monte Carlo simulation approach can Figs. 18 and 19).
quasi-quantitatively predict the UCQY values of relevant nanostruc-
tures by treating the upconversion process as a ‘collision’ process Conclusion
among the randomly walked excited states in the sublattice and setting In summary, we carried out the spectroscopic characterization and
the lanthanide-to-lanthanide interaction strengths properly41. As shown microscopic theoretical modelling of the precisely defined upconvert-
in Fig. 5a,b, if just Yb–Tm interactions between nearest-neighbour ion ing core–shell–shell NaYF4@NaYF4:Yb/Tm@NaYF4 nanostructure
pairs are considered (traditional short-range energy transfer), the and revealed a fundamental size effect on nanoscale upconversion,
d-value effect on the UCQY deviates significantly from experimental which is unrelated to the nanoscale surface properties. We found that
observation. When taking long-range Yb–Tm interactions into con- the increase in the active domain size (the intermediate shell thick-
sideration, the simulation results are generally in good agreement ness) is able to increase the Yb–Tm energy transfer efficiency substan-
with the measured UCQYs. Simulation details are included in Supple- tially and thereby produce size-dependent UCQYs. Importantly, the
mentary Figs. 21–27 and Supplementary Tables 5–7. Furthermore, we observed size effect resulted in ultrahigh UCQY values, with the highest
simulated the probability distribution of finding Yb (at 980 nm) and achieved being 13.0 ± 1.3% in sub-50 nm UCNPs (excitation: 980 nm;
Tm (at 800 nm) in their excited state for the core–shell–shell NaYF4@ PD: 100 W cm−2), which is around fourfold higher than well-established
NaYF4:20% Yb,1% Tm(d)@NaYF4 nanostructure. A larger d value results micrometre-scale bulk crystal counterparts. This ultrahigh UCQY value
in an increased population number for Tm in its excited state (that is, enables the preparation of UCNPs (brightness: 200,160 cm−1 M−1, per
the UCQYs increase (Fig. 5c)) but with a decreased population number mole of nanoparticles) that are comparable to or even brighter than a
for Yb in its excited state—this leads to the increased ηET (Fig. 5d), in range of well-established, highly efficient, single-photon fluorophores,
accordance with the experimental observations in Fig. 3b and Fig. 4e. such as semiconductor quantum dots, fluorescent polymers, carbon
dots, small organic molecules and aggregation-induced emission lumi-
The d-dependent UCQYs using different activators and core– nogens. Mechanistic investigations revealed that long-range lantha-
shell structures nide interactions (in the range of several to around ten nanometres)
We further exploited the effect of the layer thickness d using different underpin the observed size-dependent phenomena.
activators in the designated core–shell–shell nanostructure (Fig. 6). Furthermore, our results provide compelling evidence that both
We synthesized a set of NaYF4@NaYF4:x% Yb,2% Er(d)@NaYF4 nano- the lanthanide energy transfer efficiency and the luminescence quan-
structures (x = 20 and 40) with Er as the activator instead of Tm, and tum yield are size-dependent, thereby deepening our basic understand-
with various layer thicknesses (d = 1.2, 2.5, 4.5 and 9.5 nm) (Fig. 6a). ing of lanthanide upconversion as well as lanthanide interactions at the
This nanostructure also uses 5.5-nm-diameter NaYF4 as the inert core nanoscale, which will therefore have profound effects on the design of
and ultrathick inert NaYF4 as the outermost shell (of ~7 nm thickness). all types of lanthanide luminescent material (upconversion, downshift-
Er activators are able to upconvert absorbed 980 nm photons into ing and quantum cutting) for use in a broad spectrum of technological
green (2H11/2/4S3/2 → 4I15/2) and red (4F9/2 → 4I15/2) emissions via two suc- applications that range from deep-tissue biophotonics and photoredox
cessive energy transfers from Yb to Er (Fig. 6a–c)35. As expected, the applications to lighting and solar cells.
overall UCQYs, green emission UCQYs and red emission UCQYs were
observed to depend on d for both the 20% Yb and 40% Yb samples; Online content
the overall UCQYs were increased fivefold when increasing d from 1.2 Any methods, additional references, Nature Portfolio reporting sum-
to 9.5 nm (Fig. 6e,f). maries, source data, extended data, supplementary information,
The overlapping of the Er and Yb luminescence spectra at 1,000 nm acknowledgements, peer review information; details of author contri-
prevents the experimental investigation of d-dependent Yb–Er energy butions and competing interests; and statements of data and code avail-
transfer through direct lifetime measurements, as carried out for Tm ability are available at [Link]
activators. However, the observation of the gradual decrease in the
green/red emission intensity ratio with the increase in d (Fig. 6d), the References
d-dependent UCL and Yb luminescence lifetimes (Supplementary Fig. 16) 1. Auzel, F. Upconversion and anti-Stokes processes with f and d
and the d-enhanced Er luminescence at 1,532 nm (Supplementary ions in solids. Chem. Rev. 104, 139–174 (2004).

Nature Photonics
Article [Link]

2. Chen, G., Ågren, H., Ohulchanskyy, T. Y. & Prasad, P. N. Light 24. García de Arquer, F. P. et al. Semiconductor quantum dots:
upconverting core–shell nanostructures: nanophotonic control technological progress and future challenges. Science 373,
for emerging applications. Chem. Soc. Rev. 44, 1680–1713 (2015). eaaz8541 (2021).
3. Xie, Y. et al. Lanthanide-doped heterostructured nanocomposites 25. Dreaden, E. C., Alkilany, A. M., Huang, X., Murphy, C. J. & El-Sayed,
toward advanced optical anti-counterfeiting and information M. A. The golden age: gold nanoparticles for biomedicine. Chem.
storage. Light Sci. Appl. 11, 150 (2022). Soc. Rev. 41, 2740–2779 (2012).
4. Francés-Soriano, L., Estebanez, N., Pérez-Prieto, J. & Hildebrandt, 26. Liu, G. Advances in the theoretical understanding of photon
N. DNA-coated upconversion nanoparticles for sensitive nucleic upconversion in rare-earth activated nanophosphors. Chem. Soc.
acid FRET biosensing. Adv. Funct. Mater. 32, 2201541 (2022). Rev. 44, 1635–1652 (2015).
5. Fernandez-Bravo, A. et al. Continuous-wave upconverting 27. Macedo, A. G. et al. Effects of phonon confinement on anomalous
nanoparticle microlasers. Nat. Nanotechnol. 13, 572–577 (2018). thermalization, energy transfer, and upconversion in Ln3+-doped
6. Chen, G. et al. (α-NaYbF4:Tm3+)/CaF2 core/shell nanoparticles Gd2O3 nanotubes. Adv. Funct. Mater. 20, 624–634 (2010).
with efficient near-infrared to near-infrared upconversion for 28. Liu, G. K., Zhuang, H. Z. & Chen, X. Y. Restricted phonon relaxation
high-contrast deep tissue bioimaging. ACS Nano 6, 8280–8287 and anomalous thermalization of rare earth ions in nanocrystals.
(2012). Nano Lett. 2, 535–539 (2002).
7. Liu, Y. et al. Amplified stimulated emission in upconversion 29. Chen, X. et al. Confining energy migration in upconversion
nanoparticles for super-resolution nanoscopy. Nature 543, nanoparticles towards deep ultraviolet lasing. Nat. Commun. 7,
229–233 (2017). 10304 (2016).
8. Wu, X. et al. Dye-sensitized core/active shell upconversion 30. Quintanilla, M. et al. Cubic versus hexagonal-phase, size and
nanoparticles for optogenetics and bioimaging applications. ACS morphology effects on the photoluminescence quantum yield
Nano 10, 1060–1066 (2016). of NaGdF4:Er3+/Yb3+ upconverting nanoparticles. Nanoscale 14,
9. Deng, R. et al. Temporal full-colour tuning through 1492–1504 (2022).
non-steady-state upconversion. Nat. Nanotechnol. 10, 237–242 31. Wang, F., Wang, J. & Liu, X. Direct evidence of a surface
(2015). quenching effect on size-dependent luminescence of
10. Richards, B. S., Hudry, D., Busko, D., Turshatov, A. & Howard, I. A. upconversion nanoparticles. Angew. Chem. Int. Ed. 49,
Photon upconversion for photovoltaics and photocatalysis: 7456–7460 (2010).
a critical review. Chem. Rev. 121, 9165–9195 (2021). 32. Johnson, N. J. et al. Direct evidence for coupled surface and
11. Krämer, K. W. et al. Hexagonal sodium yttrium fluoride based concentration quenching dynamics in lanthanide-doped
green and blue emitting upconversion phosphors. Chem. Mater. nanocrystals. J. Am. Chem. Soc. 139, 3275–3282 (2017).
16, 1244–1251 (2004). 33. Boyer, J. C. & van Veggel, F. C. Absolute quantum yield
12. Fan, Y., Liu, L. & Zhang, F. Exploiting lanthanide-doped measurements of colloidal NaYF4: Er3+, Yb3+ upconverting
upconversion nanoparticles with core/shell structures. Nano nanoparticles. Nanoscale 2, 1417–1419 (2010).
Today 25, 68–84 (2019). 34. Fischer, S., Bronstein, N. D., Swabeck, J. K., Chan, E. M. &
13. Zhang, Y., Lei, P., Zhu, X. & Zhang, Y. Full shell coating or cation Alivisatos, A. P. Precise tuning of surface quenching for
exchange enhances luminescence. Nat. Commun. 12, 6178 (2021). luminescence enhancement in core–shell lanthanide-doped
14. Zhou, B. et al. Enhancing multiphoton upconversion through nanocrystals. Nano Lett. 16, 7241–7247 (2016).
interfacial energy transfer in multilayered nanoparticles. Nat. 35. Würth, C., Fischer, S., Grauel, B., Alivisatos, A. P. & Resch-Genger,
Commun. 11, 1174 (2020). U. Quantum yields, surface quenching, and passivation efficiency
15. Feng, W., Sun, L. D. & Yan, C. H. Ag nanowires enhanced for ultrasmall core/shell upconverting nanoparticles. J. Am.
upconversion emission of NaYF4:Yb,Er nanocrystals via a direct Chem. Soc. 140, 4922–4928 (2018).
assembly method. Chem. Commun. 29, 4393–4395 (2009). 36. Jones, C. M. S., Gakamsky, A. & Marques-Hueso, J. The
16. Zhan, Q., Zhang, X., Zhao, Y., Liu, J. & He, S. Tens of thousands-fold upconversion quantum yield (UCQY): a review to standardize
upconversion luminescence enhancement induced by a single the measurement methodology, improve comparability, and
gold nanorod. Laser Photonics Rev. 9, 479–487 (2015). define efficiency standards. Sci. Technol. Adv. Mater. 22, 810–848
17. Meng, Y. J. et al. Bright single-nanocrystal upconversion at sub (2021).
0.5 W cm−2 irradiance via coupling to single nanocavity mode. Nat. 37. Würth, C., Grabolle, M., Pauli, J., Spieles, M. & Resch-Genger, U.
Photonics 17, 73–81 (2023). Relative and absolute determination of fluorescence quantum
18. Chen, G. et al. Energy-cascaded upconversion in an organic yields of transparent samples. Nat. Protoc. 8, 1535–1550
dye-sensitized core/shell fluoride nanocrystal. Nano Lett. 15, (2013).
7400–7407 (2015). 38. Förster, T. 10th Spiers Memorial Lecture. Transfer mechanisms of
19. Garfield, D. J. et al. Enrichment of molecular antenna triplets electronic excitation. Discuss. Faraday Soc. 27, 7–17 (1959).
amplifies upconverting nanoparticle emission. Nat. Photonics 12, 39. Förster, T. Zwischenmolekulare energiewanderung und
402–407 (2018). fluoreszenz. Ann. Phys. 437, 55–75 (1948).
20. Zou, W., Visser, C., Maduro, J. A., Pshenichnikov, M. S. & 40. Dexter, D. L. A theory of sensitized luminescence in solids.
Hummelen, J. C. Broadband dye-sensitized upconversion of J. Chem. Phys. 21, 836–850 (2004).
near-infrared light. Nat. Photoncs 6, 560–564 (2012). 41. Zuo, J. et al. Precisely tailoring upconversion dynamics via energy
21. Homann, C. et al. NaYF4:Yb,Er/NaYF4 core/shell nanocrystals with migration in core–shell nanostructures. Angew. Chem. Int. Ed. 57,
high upconversion luminescence quantum yield. Angew. Chem. 3054–3058 (2018).
Int. Ed. 57, 8765–8769 (2018). 42. Hossan, M. Y. et al. Explaining the nanoscale effect in the
22. Feng, Y. et al. Internal OH− induced cascade quenching of upconversion dynamics of β-NaYF4:Yb3+, Er3+ core and
upconversion luminescence in NaYF4:Yb/Er nanocrystals. Light core–shell nanocrystals. J. Phys. Chem. C 121, 16592–16606
Sci. Appl. 10, 105 (2021). (2017).
23. Zhao, J. et al. Single-nanocrystal sensitivity achieved by 43. Wang, F. & Liu, X. Upconversion multicolor fine-tuning:
enhanced upconversion luminescence. Nat. Nanotechnol. 8, visible to near-infrared emission from lanthanide-doped NaYF4
729–734 (2013). nanoparticles. J. Am. Chem. Soc. 130, 5642–5643 (2008).

Nature Photonics
Article [Link]

Publisher’s note Springer Nature remains neutral with regard the author(s) or other rightsholder(s); author self-archiving of the
to jurisdictional claims in published maps and institutional accepted manuscript version of this article is solely governed by the
affiliations. terms of such publishing agreement and applicable law.

Springer Nature or its licensor (e.g. a society or other partner) holds © The Author(s), under exclusive licence to Springer Nature Limited
exclusive rights to this article under a publishing agreement with 2024

1
MIIT Key Laboratory of Critical Materials Technology for New Energy Conversion and Storage, School of Chemistry and Chemical Engineering, Harbin
Institute of Technology, Harbin, People’s Republic of China. 2Key Laboratory of Micro-systems and Micro-structures, Ministry of Education, Harbin
Institute of Technology, Harbin, People’s Republic of China. 3State Key Laboratory of Luminescence and Applications, Changchun Institute of Optics, Fine
Mechanics and Physics, Chinese Academy of Sciences, Changchun, People’s Republic of China. 4National Key Laboratory of Laser Spatial Information,
Harbin Institute of Technology, Harbin, People’s Republic of China. 5Department of Chemistry, Biochemistry, and Pharmacy, University of Bern, Bern,
Switzerland. 6Institute of Sensors, Signals and Systems, School of Engineering & Physical Sciences, Heriot-Watt University, Edinburgh, UK. 7Institute of
Materials Science, University of Valencia, Valencia, Spain. 8These authors contributed equally: Feng Li, Langping Tu. e-mail: chenguanying@[Link]

Nature Photonics
Article [Link]

Methods Monte Carlo simulations; F.L. collected most of the experimental data,
All experimental details including the sample synthesis, characteriza- with contributions from X.Z., J.D., X.L., D.H. and R.F.; F.L., G.C. and L.T.
tion and optical measurements are provided in the Supplementary discussed and interpreted the collected data; C.Y. provided valuable
Information. discussions and experimental supports; all authors contributed to the
data analysis; the manuscript was written by F.L., L.T. and G.C., with
Data availability comments from J.M.-H. and K.W.K.; and G.C. directed the research.
The data that support the findings of this study are available from
the corresponding author upon reasonable request. Source data are Competing interests
provided with this paper. The authors declare no competing interests.

Code availability Additional information

All codes used in this paper are available from the corresponding author Supplementary information The online version contains supplementary
upon reasonable request. material available at [Link]

Acknowledgements Correspondence and requests for materials should be addressed to

This work is supported by the National Natural Science Foundation Guanying Chen.
of China (52272270 and 51972084 to G.C.) and the Fundamental
Research Funds for the Central Universities, China (AUGA5710052614 Peer review information Nature Photonics thanks Ute Resch-Genger
to G.C.). and the other, anonymous, reviewer(s) for their contribution to the
peer review of this work.
Author contributions
G.C. conceived the research; F.L. and Y.Z. prepared the UCNPs; K.W.K. Reprints and permissions information is available at
synthesized the bulk upconverting materials; L.T. performed the [Link]/reprints.

Nature Photonics