Full shell coating or cation exchange enhances luminescence

Core-shell structure is routinely used for enhancing luminescence of optical nanoparticles, where the luminescent core is passivated by an inert shell. It has been intuitively accepted that the luminescence would gradually enhance with the coverage of inert shell. Here we report an “off-on” effect at the interface of core-shell upconversion nanoparticles, i.e., regardless of the shell coverage, the luminescence is not much enhanced unless the core is completely encapsulated. This effect indicates that full shell coating on the luminescent core is critical to significantly enhance luminescence, which is usually neglected. Inspired by this observation, a cation exchange approach is used to block the energy transfer between core nanoparticle and surface quenchers. We find that the luminescent core exhibits enhanced luminescence after cation exchange creates an effective shell region. These findings are believed to provide a better understanding of the interfacial energy dynamics and subsequent luminescence changes.

Typically, lanthanide-doped UCNPs are doped with two types of ions, sensitizers (e.g., Yb 3+ and Nd 3+ ) and activators (e.g., Er 3+ , Tm 3+ , and Eu 3+ ), of which sensitizers absorb the incident light and transfer excitation energy to activator for luminescence emission. However, for the small-sized UCNPs, the total surface area per unit volume of the particle would increase as the particle size decreases, which may provide more luminescent quenching centers (e.g., surface defects, surface oscillators, etc.) 15 . In light of this issue, various approaches have been proposed in order to improve the luminescence of nanosized UCNPs, including dye-sensitization strategy 16,17 , coupling with surface plasmon 18,19 , local structural engineering [20][21][22] , etc. 23,24 . Nevertheless, construction of core-shell architecture is still considered as one of the widely used methods to enhance luminescence of UCNPs 25 . For example, Zhou et al. recently reported a core-multi-shell design to alleviate crores-relaxation and promote interfacial energy transfer process, which achieved an upconversion luminescence quantum yield (UCQY) of 6.34% (4.5 W cm -2 ) 26 . Homann et al. proposed to use anhydrous rareearth acetates to prepare core-shell UCNPs (45 nm) and reported a UCQY value of 9% (30 W cm -2 ), which was close to that of the bulk counterpart (10.3%) at the same power density 27 .
As for the mechanism of luminescence enhancement via coreshell engineering, Wang et al. firstly reported that coating of a thin inert shell could retain the optical integrity of luminescent core and also effectively reduce emission losses caused by the surface quenching effects 28 . Later, it was proved that the inert shell could decouple the surface and concentration quenching processes for colloidal lanthanide-doped nanocrystals, thus allowing for high-dopant concentrations in the luminescent core 29 . More recently, time-resolved spectroscopic studies indicate that the surface of UCNPs can be gradually passivated with the thickness of inert shell. However, when the UCNP surface is totally passivated, further increasing shell thickness would lead to slight decrease in the emission intensity due to some potential non-radiative quenching pathways from the interface 30 . Based on traditional understanding, it has been intuitively accepted that coating of an inert shell can act as a protective barrier that shields the luminescent UCNP core from its surrounding quenching centers. Nevertheless, the underlying picture of surface quenching dynamics through core-shell interface has not been well elucidated. Despite some discussion of this issue 29 , there are very few studies that directly report a synchronous correlation between a step-wise surface passivation and resulting luminescence change for the core-shell UCNPs. To address these issues, it is necessary to prepare core-shell UCNPs with different shell coverage and to study its relationship with energy transfer and luminescence emission.
In this work, core-shell UCNPs with various shell coverage ratios are demonstrated using silica coated Janus nanoparticles as a template. By increasing the shell coverage, the luminescence intensity of the UCNPs is not much enhanced unless the core nanoparticle is completely covered by the inert shell. In another word, significant enhancement in the luminescence intensity is only observed when a full shell coating is applied. This "off-on" effect in the core-shell UCNPs has not been systematically characterized before and the influence of the shell coverage on the luminescence emission has long been neglected. Furthermore, our time-course study shows that both the emission intensity and color of the highly doped UCNPs vary as the full shell thickness increases. Based on this discovery, a simple post-synthesis approach via cation exchange is developed to eliminate the surface-energy transfer from the core nanoparticle to surface quenching centers, which results in a similar luminescence enhancement effect as the full shell coating. These findings are believed to offer fundamental insights into the mechanisms and process of surface quenching and inspire the design of novel nanostructures with much enhanced luminescence.

Results
Synthesis of NaErF 4 /SiO 2 @NaYF 4 Janus nanoparticles. In order to gain insights into the correlation between surface energy dynamics and luminescent properties, we have specifically developed an approach to synthesize core-shell structured NaErF 4 /SiO 2 @NaYF 4 Janus nanoparticles with tunable shell coverage (Fig. 1a). In a typical procedure, monodispersed β-NaErF 4 UCNPs were firstly synthesized using a co-precipitation method (Fig. 1b) 31 . Subsequently, a dense silica layer was coated anisotropically on the NaErF 4 core by a surfactant-templating method, forming a Janus nanoparticle containing a hydrophobic NaErF 4 core and a hydrophilic SiO 2 shell. Afterwards, part of the SiO 2 shell was removed by buffered oxide etchant (BOE), exposing more hydrophobic surface for further growth of the NaYF 4 shell.
As illustrated in Fig. 1a, asymmetric coverage of SiO 2 shell on the NaErF 4 core is a prerequisite for the fabrication of NaErF 4 / SiO 2 @NaYF 4 Janus nanoparticles. In this study, a reverse microemulsion method was used, in which cetyltrimethyl ammonium bromide (CTAB) was employed as the micellar template and tetraethyl orthosilicate (TEOS) as the silicon precursor. This approach is employed because highly concentrated CTAB surfactants (3 mg mL −1 ) in the water-oil mixture can lead to the aggregation of CTAB micelles on one side of the UCNP surface, and therefore offer nucleation sites for the growth of SiO 2 shell through hydrolysis of TEOS precursors. Moreover, the morphology and SiO 2 coverage can be precisely tuned by varying the amount of TEOS precursors added. For example, both the SiO 2 coverage ratio ( Supplementary Fig. 1a−e) and average hydrodynamic diameters ( Supplementary Fig. 1f) can be gradually increased via the addition of TEOS precursors.
Based on the construction of NaErF 4 /SiO 2 Janus nanoparticles, we verified the feasibility of the direct growth of NaYF 4 shell on the NaErF 4 /SiO 2 nanoparticles. Supplementary Fig. 2a and b present the TEM images of NaErF 4 /SiO 2 Janus nanoparticles with SiO 2 coverage ratio of 0% and 100%. As shown in Supplementary  Fig. 2c, the NaYF 4 shell can grow uniformly on the bare NaErF 4 core (i.e., 0% SiO 2 coverage) and form a core-shell structure. However, for the silica-coated NaErF 4 nanoparticle (i.e., 100% SiO 2 coverage), the NaYF 4 phase does not grow on it but instead nucleates separately ( Supplementary Fig. 2d). This is because silica tends to form a thin layer around NaErF 4 nanoparticle at the beginning of hydrolysis process and cover the whole particle surface over time ( Supplementary Fig. 1a−d). Notably, the hydrophilic SiO 2 coating formed on NaErF 4 core would prohibit the growth of hydrophobic NaYF 4 layer. Therefore, the undesired thin portion of the SiO 2 shell should be removed in order to expose some of the NaErF 4 core surface for the growth of NaYF 4 shell.
Although hydrofluoric acid (HF) is a well-known silica removal etchant, it is a highly corrosive liquid and may cause potential risks. Besides, its strong etching ability makes it challenging to precisely control the etching speed. As an alternative, buffered oxide etchant (BOE), a commercial etching mixture of NH 4 F and HF, was used to etch the thin region of SiO 2 shell from the NaErF 4 /SiO 2 Janus nanoparticles at a stable etching speed (800 Å min −1 at 20°C). Supplementary Fig. 3 presents the TEM images of NaErF 4 /SiO 2 Janus nanoparticle with different SiO 2 coverage ratios (25%, 50% and 75%) after BOE etching, clearly indicating the size and morphology of UCNP nanoparticles are not affected after BOE treatment. More importantly, the hydrophobic side of UCNP nanoparticle has been already exposed via BOE etching process (indicated with red arrow in Supplementary Fig. 3) while the other side is still covered by the SiO 2 shell. Besides, the coverage ratio of SiO 2 shell (Supplementary Fig. 4a) and hydrodynamic size ( Supplementary Fig. 4b Fig. 4c), further confirming the partial removal of negatively charged silica. It is worth to mention that no evident change in the upconversion luminescence profiles of UCNPs is observed before and after BOE etching ( Supplementary Fig. 4d), indicating the BOE etching has little influence on the upconversion luminescence.
Following the successful etching process, the NaErF 4 /SiO 2 Janus nanoparticle were obtained and used as seeds for the growth of NaYF 4 shell. Notably, NaYF 4 shell is critical in enhancing the luminescence emission, while SiO 2 shell provides little protection as it can not passivate surface defects of luminescent core 32 . Here, the protection ratio of NaYF 4 shell is defined as follows: Figure 1c−e and Supplementary Fig. 5 present the TEM images of core-shell structured NaErF 4 /SiO 2 @NaYF 4 Janus nanoparticles with different protection ratios of NaYF 4 shell (25%, 75%, and 100% (NaErF 4 @NaYF 4 )), which show that the as-prepared Janus nanoparticle consists of the following parts: the NaErF 4 UCNP core, the SiO 2 shell covering one side of the core, and the NaYF 4 shell on the other side.
Spectroscopic study of core-shell structured NaErF 4 /SiO 2 @-NaYF 4 Janus nanoparticles. Figure 1f compares the upconversion luminescence intensities of core-shell structured NaErF 4 /SiO 2 @-NaYF 4 nanoparticles under 980 nm laser excitation (3 W cm −2 ) with different protection ratios of NaYF 4 shell, i.e., 25% protection (NaErF 4 /SiO 2 @25%NaYF 4 ), 75% protection (NaErF 4 /SiO 2 @75% NaYF 4 ), and 100% protection (NaErF 4 @NaYF 4 ). The bare NaErF 4 core nanoparticle (0% protection) serves as the control group for comparison. Surprisingly, with the protection ratio of NaYF 4 shell increased from 0% to 25%, and to 75%, the upconversion luminescence intensity remains almost unchanged in the spectral range from 500 nm to 700 nm (Fig. 1g). Only when a full NaYF 4 shell (100% protection) is deposited on the NaErF 4 core, the luminescence intensity can be recovered and dramatically improved, by about 2500 times (Supplementary Figure 6). Previous work showed that the inert shell could act as a shield to protect the luminescent core from surface quenchers 28 . In that case, evident enhancement in luminescence intensity should be anticipated when more surface defects are passivated with the protection of NaYF 4 shell. However, our results clearly suggest that the luminescence intensity of UCNP core does not synchronously improve with continuous passivation of exposed surface defects. Instead, an "off-on" effect exists at the core-shell interface, meaning that partial coverage of inert shell to any degree contributes little to the luminescence enhancement, and a full coverage with the inert shell is critically necessary.
In lanthanide-doped nanocrystals, the luminescence emission intensity would decrease when the dopant ion concentration exceeds a threshold, typically referred to as "concentration quenching". It has been well established that an inert shell could reduce the coupling between surface and concentration quenching effects via suppressing energy migration to the surface quenchers 29,33 . Figure 1h and i present the upconversion luminescence decay curves of green emission (542 nm) and red emission (658 nm) for NaErF 4 /SiO 2 @NaYF 4 Janus nanoparticles with different protection ratios of NaYF 4 shell. It can be observed that the upconversion luminescence lifetime for both green and red emissions generally increase with the increased protection ratios of NaYF 4 shell. The enhancement in lifetime does indicate longer time of excitation states remaining in the system when more surface defects are gradually recovered by the coating of NaYF 4 shell. However, as mentioned in Fig. 1f and g, the upconversion luminescence intensity exhibits no evident enhancement with the increase of NaYF 4 shell coverage only until the NaErF 4 core surface is fully protected. Generally, for the NaErF 4 core nanoparticle, some internal defects (e.g., vacancy and interstitial defects) may exist and quench the luminescence (Fig. 1j). Besides, when the size of the particle decreases, the total surface area per volume will increase because of the higher surface-to-volume ratio, which promotes surface-related deactivations due to more exposure to the surface defects (e.g., crystal disorder and non-crystallization) as well as capping ligands and solvent molecules with strong vibrations 34,35 . In this regard, these surface-related quenching processes can affect the luminescence of as-synthesized NaErF 4 / SiO 2 @NaYF 4 Janus nanoparticles via following routes 36 (Fig. 1j): i. Direct deactivation. Photoexcited Er ions located on (or close to) the particle surface are non-radiatively deactivated by the neighboring quenchers. ii. Energy migration induced quenching. This refers to the quenching process, in which the excitation energy of internal photoexcited Er ions randomly migrates to other Er ions around the surface or directly to the surface quenching site.
center). As a result, only when a full inert NaYF 4 shell is coated on the NaErF 4 core, the increased distance between luminescent core and surrounding quenching centers can greatly prevent the quenching of excited Er ions around the surface and also curb the migration of excitation energy to the surrounding quenchers. In addition to shell coverage, we have further investigated the influence of thickness of incomplete shell on the luminescence intensity of NaErF 4 /SiO 2 @NaYF 4 Janus nanoparticles. As shown in Supplementary Fig. 7a, with increasing the thickness of the incomplete NaYF 4 shell (75% coverage) from 0 nm, to 5 nm and to 10 nm, the luminescence intensity still remains almost unchanged. This is a stark contrast to previous results that the luminescence intensity for core-shell UCNPs should generally exhibit a thickness-dependent behavior 29,30,38 . Also, other than NaErF 4 core, we have additionally used the low-doped luminescent core, NaYF 4 :20%Yb, 2%Er, to study the influence of NaYF 4 shell coverage on the luminescence intensity ( Supplementary Fig. 7b). Similar to the case of NaErF 4 /SiO 2 @NaYF 4 , the luminescence intensity of NaYF 4 :20%Yb,2%Er/SiO 2 @NaYF 4 still barely improves with increasing the coverage ratio of NaYF 4 shell from 0 % to 75%. Only when a full NaYF 4 shell is coated on the NaYF 4 :Yb, Er core, the upconversion luminescence intensity can then be recovered and drastically improved. In other words, the well-studied parameters such as shell thickness and dopant concentration in previous studies, have little influence on the luminescent intensity, provided that the luminescent core is not completely covered by the inert shell. Therefore, based on the results in Fig. 1 and Supplementary  Fig. 7, it is rather clear that no matter how to tune the coverage and thickness of the outer shell as well as the dopant concentration, the luminescence does not change accordingly unless the surface of the core nanoparticle is fully encapsulated. The existence of such "offon" effect could be due to the reason that as long as the luminescent core is not completely covered by the inert shell, the exposed surface can still act as weak spots where the majority of excitation energy is dissipated via the migration to the surface quenchers and results in little enhancement in emission intensity.
To further understand the evolution of luminescent profiles with the coating of NaYF 4 shell on the NaErF 4 core, we have performed a time-course study on the growth of core-shell structured NaErF 4 @NaYF 4 UCNPs (Fig. 2a). Our strategy was to start with the bare NaErF 4 core in the solution, and then mix into the shell precursor to initiate the epitaxial growth of NaYF 4 shell. Once the reaction is started by heating the mixture to required temperature, we sampled at different time intervals throughout the synthesis process. In Fig. 2b, upconversion spectra of NaErF 4 core and core-shell NaErF 4 @NaYF 4 UCNPs upon excitation by 980 nm laser light (3 W cm −2 ) are presented for samples obtained at different time points. Over the time course, there is a general increase in the emission intensity of upconversion spectra. In order to make a more direct comparison, the ratio of green to red band (G/R) at each of the time course points are plotted in the inset in Fig. 2b. In terms of the G/R values, there are no significant differences among samples taken at 0 min, and 9 min, but a sharp increase was observed in the subsequent time points. Specifically, the G/R value maximizes (>0.6) at 12 min, and then abruptly drops until 30 min point and the subsequent G/R values remain quite stable over the rest of the time course. Figure 2c presents the lifetime decays of green and red upconversion luminescence of NaErF 4 core and core-shell NaErF 4 @NaYF 4 UCNPs at different time points. The average decay time of the green upconversion luminescence of the NaErF 4 core is calculated to be only 6.1μs, which suggests the excitation energy is nonradiatively lost due to the rapid energy migration to the surface quenchers. On the contrary, the lifetimes of green emission for core-shell NaErF 4 @NaYF 4 UCNPs are significantly increased. This indicates that the nonradioactive process previously occurring on the bare NaErF 4 core is now greatly suppressed by the protection of the inert NaYF 4 shell. Similar to green emission decays, the lifetime of red upconversion luminescence of core-shell NaErF 4 @NaYF 4 UCNPs is also much longer than that of core NaErF 4 . These results reconfirm that the surface quenching dominates the energy loss process. Another interesting observation is that the lifetime for the red emission monochromatically increases over time, while the lifetime for the green emission increases initially but starts to decrease since the time point of 12 min. The decrease of lifetime for the green emission is an indication of the enhancement of the cross-relaxation process between Er 3+ ions over time, i.e., 4 F 7/2 + 4 I 11/2 → 4 F 9/2 + 4 F 9/2 ( Supplementary Fig. 8), since this leads to depopulation of green emitting level 4 F 7/2 and the population of red emitting states 4 F 9/2 .
It should be noted that the Er 3+ ions in the NaErF 4 @NaYF 4 core-shell UCNPs have dual functions, as both sensitizers and activators. On one hand, the Er 3+ ions can directly harvest the incident 980 nm laser light as sensitizers. On the other hand, the excited Er 3+ ions can radiatively transit to the ground states through releasing green and red lights. As shown in Fig. 2d, for the bare NaErF 4 core, its outer surface is exposed to a large amount of quenching centers. Upon illumination by 980 nm laser light, the energy transfer (ET) process from Er 3+ to surface quenchers can be readily established, which significantly lowers the emission intensity due to the substantial energy loss on the surface. However, once the full inert NaYF 4 shell starts to grow on the NaErF 4 core, the Er 3+ ions are protected from the surface quenchers by the NaYF 4 shell. When the thickness of NaYF 4 shell is low (~1 nm) 39 , the surface defects are partly respired, which preferentially promotes the green emission and causes the increase of G/R value. As the thickness of NaYF 4 shell goes up, the energy loss by the ET process from Er 3+ ions to the surface defects is further prohibited, resulting in the prominent preservation of excitation energies. Under such circumstance, the absorbed excitation power is high enough to maintain the active energy transfer between neighboring Er 3+ ions and may lead to adequate populations in the 4 I 11/2 and 4 F 7/2 states. Therefore, the possibility of cross relaxation (CR) process between 4 I 11/2 and 4 F 7/2 states is greatly raised and the number of ions in the 4 F 9/2 states is accordingly increased, which dramatically enhances the red emission ( Supplementary Fig. 8).
Tuning surface energy dynamic via cation exchange. In light of the dominant role of surface-involved energy transfer in controlling luminescent profiles of UCNPs, we moved forward to optimize upconversion intensity by tailoring interfacial energy dynamics via a facile post-treatment. Specifically, a hightemperature ion exchange method is used to enhance the luminescence emission of UCNPs (Fig. 3a). In brief, pre-synthesized NaErF 4 nanoparticles were mixed with YCl 3 solution at elevated temperature (300°C) in order to obtain a new composite via cation exchange process, NaErF 4 @Y. Although surface cations on the nanoparticles are replaced by Y 3+ ions, the hexagonal structure of NaErF 4 phase can still be maintained (Fig. 3b). Besides, the TEM images and the size distribution analysis also reveal that the morphology and size of the NaErF 4 nanoparticles do not significantly change during the ion exchange (Fig. 3c, d; Supplementary Fig. 9). The energy-dispersive X-ray spectroscopy (EDS) spectra ( Fig. 3e and f) also confirm the presence of Y element for cation-exchanged NaErF 4 @Y nanoparticles. Moreover, elemental mapping images of a single NaErF 4 @Y nanoparticle show that Na, Er, and F elements are uniformly distributed on the nanoparticles while Y elements are mainly located on the outmost layer, indicating that ion exchange mainly occurs on the surface (see Fig. 3g). In addition, we have also used X-ray photoelectron spectroscopy (XPS) to analyze the chemical composition of NaErF 4 @Y nanoparticles before and after cation exchange. As shown in Fig. 3h, i, the signals from Y elements can be clearly identified from NaErF 4 @Y. However, no Y signals are observed from the XPS spectra of NaErF 4 nanoparticles ( Supplementary Fig. 10). It is worth mentioning that all the obtained nanoparticles have similar lattice constants compared with NaErF 4 nanoparticles (Supplementary Table 1).
For a bare NaErF 4 nanoparticle, as its surface is exposed to the surrounding quenchers, the upconversion luminescence intensity is severely reduced, due to the substantial transfer of the excitation energy to the surrounding quenchers (Fig. 4a). However, we reason that cations on the surface of NaErF 4 nanoparticles are replaced by inert Y 3+ ions through ion exchange so that the pathway of energy migration to surface quenchers can be blocked, which is expected to enhance upconversion luminescence (Fig. 4b). Figure 4c compares the upconversion luminescence spectra of NaErF 4 and NaErF 4 @Y nanoparticles under 980 nm laser excitation (3 W cm −2 ). As anticipated, the luminescence intensity of NaErF 4 @Y nanoparticles is obviously stronger than that of NaErF 4 nanoparticles, with an enhancement of about 17.2 times. Figure 4d presents the luminescence photos of NaErF 4 and NaErF 4 @Y nanoparticles upon illumination by 980 nm light, which also confirms that the luminescence of NaErF 4 @Y is significantly brighter than that of NaErF 4 . Similar behavior can also be observed for the luminescence decay of Er 3+ emission at 542 nm under 980 nm excitation (Fig. 4e). As expected, the lifetime value is increased from 5.6 µs to 28.6 µs. Notably, the lifetime of Er 3+ emission at 658 nm remains almost unchanged (Fig. 4f), indicating that the exchange of NaErF 4 nanoparticles with Y 3+ ions has a greater influence on the green emitting state.
To further confirm that the cation substitution with Y 3+ ions is the main cause of luminescence enhancement, pre-synthesized NaErF 4 nanoparticles were preheated for 1 h at the same temperature to exclude the temperature influence. As shown in Supplementary Figs. 11 and 12, the reheated NaErF 4 nanoparticles (NaErF 4 @Reheating) can maintain the original shape, size, composition and phase structure. It can also be observed the upconversion emission of NaErF 4 nanoparticles before and after reheating basically remains constant ( Supplementary Fig. 13), indicating that only reheating treatment can not enhance the luminescence intensity NaErF 4 nanoparticles.
In addition, Yb 3+ ions are also used for ion exchange with NaErF 4 nanoparticles to obtain a new composition of nanoparticles (NaErF 4 @Yb) to verify whether it has the same luminescence enchantment effect. The characterization in Supplementary Fig. 14 and Supplementary Fig. 15 confirm that Yb 3+ ions have been successfully incorporated into NaErF 4 nanocrystals. However, unlike NaErF 4 @Y nanoparticles, the upconversion luminescence of NaErF 4 @Yb is barely enhanced compared with that of NaErF 4 nanoparticles ( Supplementary Fig. 16a). A possible explanation is that the energy levels of Yb 3+ ions and Er 3+ ions are greatly matched so that the energy migration of excitation energy to surface quenchers can not be blocked (Supplementary Fig. 16b).

Discussion
For the lanthanide-doped upconversion nanoparticles, a common approach to alleviate surface-related quenching effects is to adopt Energy migration Block Fig. 2 Time-course study on the growth of core-shell NaErF 4 @NaYF 4 UCNPs. a Schematic of the growing process of full NaYF 4 shell (yellow) on the NaErF 4 core (blue). During the full shell coating process, the NaYF 4 shell would grow along both sides (red arrow) of the luminescent NaErF 4 core. b The changes of upconversion emission spectra under 980 nm laser light (3 W cm −2 ) over time. Inset is the variations of green to red ratio over time.
c The lifetime of red and green emissions over time. d The mechanism of the surface protection of NaErF 4 core by NaYF 4 shell with different thickness. Upon excitation by 980 nm light, the excitation energy is rapidly transferred to the surface quenchers for bare NaErF 4 core. When a thin NaYF 4 shell (shaded region) is coated, the surface defects are partly restored that preferentially promotes the green emission. As the thickness of NaYF 4 shell goes up, the energy loss from Er 3+ ions to surface defects is further blocked and the possibility of cross relaxation between Er 3+ ions is greatly raised, which dramatically enhances the red emission.
the core-shell strategy that passivates the luminescent core with an optically inert shell 28 . Due to its great effectiveness and easy fabrication, core-shell engineering has drawn extensive attentions. For example, B. Richards et al. have systematically investigated the crystallographic structure at the interface of core-shell and core-multi-shell nanoparticles, demonstrating that the cation inter-diffusion occurs between core and shell materials during the synthesis 15,40 . Besides, they have recently proposed an energymigration strategy that controls the spatial redistribution of harvested energy from one sensitizing zone to two different emitting zones, which allows for the color-tuning of small-sized upconversion nanoparticles (<20 nm) with a high quantum yield efficiency up to 3.5% (60 W cm −2 ) 41 . Moreover, other factors have also been verified to have pronounced influence on the luminescence intensity, such as shell thickness 30,38 , doping concentration 29,33 , refractive index of surrounding solvents 37,42 , synthetic routes 27,43 , etc. Despite these great efforts, it should be noted that full-shell coating is generally utilized for constructing core-shell structures in these studies. In this context, the impact of this work lies in solving a long-neglected issue in core-shell nanoparticles-that is to probe the relationship between step-wise surface passivation and resulting luminescence change. Only by precisely controlling how shell grows on the luminescent core can we draw accurate conclusions about the passivation-luminescence relationships. In this work, using the surface-templating strategy, the growth of NaYF 4 shell on the NaErF 4 core can be finely controlled, as shown in Fig. 1a-e. Besides, this strategy can also be employed to fabricate other Janus nanoparticles with dual surface hydrophobicity/hydrophilicity for many potential applications, including nanomotor, multi-modal imaging, cancer therapy, etc. On the successful construction of NaErF 4 /SiO 2 @NaYF 4 Janus nanoparticles, we have demonstrated an "off-on" effect at the coreshell interface: regardless of the coverage ratio and thickness of the inert shell as well as the doping concentration, the luminescence is not much changed unless the core surface is completely encapsulated. Therefore, this means that the key factor that influences most on the surface passivation of an individual UCNP nanoparticle is the minimum thickness of inert shell instead of the average value. In other words, if the core nanoparticle is not fully coated, it seems that the core surface is little passivated. Fig. 3 Structural characterization of NaErF 4 nanoparticles before and after ion exchange. a Schematic of the NaErF4 core after ion exchange. The Y 3+ ions (yellow) would replace some Er 3+ ions (blue) on the surface and block the energy transfer to the quenching site. b X-ray diffraction (XRD) pattern of the NaErF 4 nanoparticles, obtained before and after exchange with Y 3+ ions (abbreviated as NaErF 4 @Y). c, d TEM image of the NaErF 4 nanoparticles and NaErF 4 @Y nanoparticles, respectively. Scale bar: 50 nm. e, f The EDS line-scan profile of NaErF 4 and NaErF4@Y nanoparticles, respectively. g HAADF-STEM (high angle annular dark field scanning TEM) of single NaErF 4 @Y nanoparticle and its corresponding elemental mapping images, Scale bar: 10 nm. h, i Full survey XPS spectrum and detailed spectra (Na 1 s, Er 4d, F 1 s, and Y 3d) of NaErF 4 @Y nanoparticles.
Furthermore, it is believed that such "off-on" effect is not only applicable to core-shell UCNPs, but to many other luminescent counterparts, which would also be rather helpful to better understand the interfacial energy dynamics of core-shell nanosystems.
In summary, the long neglected relationship between interfacial quenching dynamics and upconversion luminescence in core-shell UCNPs is elucidated through step-wise passivation of surface defects via a specially designed Janus architecture. Our results clearly indicate that full shell encapsulation on the luminescent core nanoparticle plays decisive roles in enhancing upconversion luminescence intensity. The time-course studies further show that both the upconversion luminescence intensity and color output vary with the thickness of full shell coverage. In light of this understanding, a non-core-shell strategy based cation exchange method is utilized to enhance upconversion luminescence of highly doped UCNPs, and demonstrate similar functionality as the core-shell approach. It is believed that our results offer important insights into the interfacial energy dynamics and may inspire new design of UCNP-based nanocomposites with optimized performance and broader application potentials.

Methods
We synthesized upconversion nanoparticles using the method described in ref. 31 . Additional experimental details are provided in the Supplementary Note.
Synthesis of NaErF 4 /SiO 2 Janus nanoparticles. Firstly, 3 mg mL −1 CTAB solution (10 mL) was mixed with 5 mg/mL NaErF 4 in cyclohexane (1 mL), followed by 45 s of sonication using the Sonic Material Vibra-Cell Ultrasonic Processor (SciMed (Asia) Pte Ltd) until a milk-like emulsion formed. Then, the emulsion was transferred to a 25 mL one-neck round bottom flask and vigorously stirred at 1400 rpm and heated to 70°C for 3 h until full removal of the cyclohexane. The solution would turn back to a clear state. After the solution was cooled down to room temperature, the NaErF 4 @CTAB solution (0.5 mg mL −1 , mg mL −1 ) was transferred to another 25 mL one-neck round bottom flask. Then, 200 μL ammonium hydroxide solution was added into the flask and the solution was kept stirring under 1000 rpm. After 30 min, different amount of TEOS was added dropwise into the flask, during which the stirring speed was kept as 1400 rpm. After 3 h of stirring, the Janus nanoparticles were precipitated down by centrifugation and dispersed in 10 mL DI water and stored for further use. Notably, for the tuning of silica coverage, different amount of TEOS (15,20,25,30, and 35 μL) were added into NaErF 4 @CTAB solution.
Silica etching. 10 mL NaErF 4 /SiO 2 solution (0.5 mg mL −1 ) were mixed with specific amounts of BOE ranging from 5 to 12.5 μL to obtain NaErF 4 /SiO 2 Janus nanoparticles with different SiO 2 coverage ratio. The BOE solution was firstly diluted by 1 mL DI water to reduce the etching speed and achieve uniform etching morphology. The diluted BOE solution was added dropwise by the control a micro syringe pump at a speed of 1 mL h −1 . Notably, the solution was kept for stirring under 1000 rpm. Then, the solution was centrifuged at 15000 rpm for 10 min and the obtained precipitates were dispersed in 10 mL ethanol.
Preparation of core-shell structured NaErF 4 /SiO 2 @NaYF 4 nanoparticles. Typically, based on the core-shell ratio, for every 10 mL ethanol containing etched NaErF 4 /SiO 2 Janus nanoparticles, the aqueous solution of 0.03 mmol YCl 3 was added into a 100 mL three-neck flask. After removing the water under 110°C, the residuals were further dissolved in the mixture of 6 mL oleic acid and 15 mL 1-octadecene and the whole system was maintained at 156°C for 10 min to completely form the RE-oleate complexes (RE: rare elements). Then, the mixture was cooled down to room temperature. The etched Janus NaErF 4 /SiO 2 nanoparticles dispersed in 10 mL ethanol were added into the solution and the resulting mixture was heated to 120°C to evaporate the ethanol. Upon cooling down to room temperature, followed was the addition of 150 μL methanol solution containing 0.12 mmol NH 4 F and 0.075 mmol NaOH. To remove the methanol, the temperature of the whole mixture was raised to 120°C for 10 min. After that, to remove the oxygen and residual methanol, the system was degassed and filled by argon alternatively three times. Subsequently, the whole reaction was raised to 300°C and kept for 1 h under the argon environment. The resulting solution was precipitated by adding the equal volume of acetone and centrifuged at 10000 rpm for 10 min. The precipitates were washed with acetone twice and finally dispersed in 10 mL cyclohexane and stored for further use.
Synthesis of hexagonal NaErF 4 @Y UCNPs. NaErF 4 @Y UCNPs were obtained by ion exchange method. In a typical process, 1 mmol YCl 3 aqueous solution was added to a 100 mL flask. The solution was heated at 120°C to fully remove the water. Then, 6 mL of oleic acid and 15 mL of 1-octadecene were added to the flask and the mixture was heated to 155°C until the solid powder is completely dissolved. The resulting solution was cooled down to room temperature. Subsequently, the pre-synthesized NaErF 4 nanoparticles dispersed in cyclohexane (20 mL) were added. The resulting mixture was heated at 120°C for 30 min to evaporate the cyclohexane. Then, the solution was heated to 300°C under argon for 1 h and cooled down to room temperature. The products were precipitated by addition of acetone, collected by centrifugation, and washed with acetone several Fig. 4 Spectroscopic study of NaErF 4 before and after ion exchange. a Schematic diagram of luminescence quenching caused by energy transfer to the surface. The excitation energy would migrate among Er ions (blue) and travel to the quenching site (dark) on the surface. b Proposed mechanism of luminescence enhancement after ion exchange. The Y 3+ ions replaced some Er 3+ ions on the surface and created an effective shell region (shaded area) such that the energy migration to the surface quenching site was blocked. c, d Upconversion emission spectra and corresponding luminescence photographs of the NaErF 4 and NaErF 4 @Y nanoparticles upon 980 nm laser excitation (3 W cm −2 ). e, f Lifetime decay curve of Er 3+ emission at 542 nm and 658 nm from the NaErF 4 and NaErF 4 @Y nanoparticles upon 980 nm laser excitation (3 W cm −2 ).
times. NaErF 4 @Yb UCNPs were synthesized by using an identical procedure, except for the use of YbCl 3 in the synthesis.
Synthesis of hexagonal NaErF 4 @Reheating UCNPs. In a typical process, the assynthesized NaErF 4 nanoparticles dispersed in cyclohexane (20 mL) were added. The resulting mixture was heated at 120°C for 30 min to evaporate the cyclohexane. Then, the solution was heated to 300°C under argon for 1 h and then cooled down to room temperature. The products were precipitated by addition of acetone, collected by centrifugation, and washed with acetone several times.

Data availability
The data that support the findings of this study are available from the corresponding author upon reasonable request. Source data are provided with this paper.