Multi‐Color Luminescence Transition of Upconversion Nanocrystals via Crystal Phase Control with SiO2 for High Temperature Thermal Labels

Abstract Upconversion nanocrystals (UCNs)‐embedded microarchitectures with luminescence color transition capability and enhanced luminescence intensity under extreme conditions are suitable for developing a robust labeling system in a high‐temperature thermal industrial process. However, most UCNs based labeling systems are limited by the loss of luminescence owing to the destruction of the crystalline phase or by a predetermined luminescence color without color transition capability. Herein, an unusual crystal phase transition of UCNs to a hexagonal apatite phase in the presence of SiO2 nanoparticles is reported with the enhancements of 130‐fold green luminescence and 52‐fold luminance as compared to that of the SiO2‐free counterpart. By rationally combining this strategy with an additive color mixing method using a mask‐less flow lithography technique, single to multiple luminescence color transition, scalable labeling systems with hidden letters‐, and multi‐luminescence colored microparticles are demonstrated for a UCNs luminescence color change‐based high temperature labeling system.

UCNs. Fourier-transform infrared spectroscopy was carried on a Varian spectrophotometer to determine the vibrational frequencies of Si-O-Si bonds in the spectral range of 4,000-50 cm -1 with an attenuated total reflection (ATR) detector. UV-visible-near infrared spectroscopy (solid) was carried out on a Cary-5000 instrument with a photomultiplier tube (PMT) detector in the wavelength range of 300-1,200 nm for reflectance measurement.
Imaging and spectral analysis of UCNs before and after the annealing process. The synthesized UCNs-embedded PUA microparticles were dropped on a slide glass with an imaging solution (10% polyethylene glycol 200 (PEG 200) in ethanol) and then covered with glass. The luminescence of the UCNs-embedded PUA microparticles was measured using a 2 W 980 nm NIR laser with a customized optical setup having a circular irradiation area (Diameter: 550 μm) (MDL-F-980-5W, Dragon laser). A Nikon D-810 camera with a ×20 objective was used for color-imaging, and upconversion luminescence spectra were recorded using a Nikon Ti-E inverted microscope installed with the photoluminescence spectrometer (QEPRO-FL, Ocean Optics). The annealed UCNs-embedded PUA microparticles on a sapphire window were placed on a microscope stage and their upconversion luminescence was measured. A short pass cut-off filter (750 nm cut-off short pass filter, Semrock) was used to block the high-intensity NIR light used for excitation.

Supplementary Note 1. Characterization of β-NaREF 4 (RE = Gd, Y, Yb, Er, and Tm)
The synthesized β-NaYF 4 :Gd 3+ UCNs were characterized by scanning electron microscopy (SEM), X-ray diffraction (XRD), high-resolution transmission electron microscopy (HR-TEM), and energy dispersive spectrometry (EDS) ( Figure S1). The SEM images showed that the hexagonal rod-shaped UCNs are 300 nm in length and 70 nm in width. The XRD patterns revealed that the crystal phase of the synthesized UCNs matched well with that of conventional hexagonal NaYF 4 UCNs (JCPDS 01-072-4799). The HR-TEM images showed the uniform shape and size of the UCNs (300 nm in length and 70 nm in width) and SiO 2 NPs (13 nm in diameter). EDS mapping of the lanthanide ion-doped β-NaYF 4 :Gd 3+ /Yb 3+ /Er 3+ (30/30/2 mol%) indicated a uniform distribution of Y 3+ , Gd 3+ , Yb 3+ , and Er 3+ ions in the nanocrystal.   The contraction of PUA microparticles with annealing temperature was confirmed by Thermogravimetric analysis (TGA). Upon annealing at 300 °C, the volume of the microparticles decreased. After annealing at 500 °C, the system with SiO 2 NPs contained 10% more PUA than that of the system without SiO 2 NPs. This is because the added SiO 2 NPs delay the decomposition of carbon. At a higher PUA content, less of the incident NIR light would be absorbed by UCNs, which results in a slight decrease in the luminescence emission intensity (Figure 2a,b, and Figure S5). Upon increasing the annealing temperature to 900 °C, the volume of the microparticles decreased by 80% as compared to that of the original UCNsembedded microparticles ( Figure S6).     As shown in Figure 2g, when the weight ratio of SiO 2 NPs to Y UCNs in the YBS microparticles was <0.5, the luminescence color changed to red (1 in Figure 2g). However, when the weight ratio of SiO 2 NPs was >0.5, the luminescence color of the YBS microparticles changed to three different greenish-blue colors (2 to 4 in Figure 2g) due to the dominance of the blue transition in the presence of larger amounts of Tm 3+ . Unlike in the case of the YBS microparticles, the GWS microparticles exhibited bluish-green colors as the weight ratio of SiO 2 NPs to G UCNs in the GWS microparticles was increased from 0.5 to 0.75. We believe that the relative amounts of Er 3+ ions from both G and W UCNs are larger than those of Tm 3+ ions, and therefore, green luminescence transition was dominant. Lastly, in the case of BWS microparticles, the luminescence color changed to blue due to the presence of the smallest amount of Er 3+ ; Tm 3+ is mainly responsible for the color transition

Supplementary Note 4. Diffusional interactions between SiO 2 NPs and UCNs during the annealing process
The diffusional interaction increases the coarsening rate through the Ostwald ripening phenomenon, whereby large particles grow at the expense of smaller ones. At the microscopic level, Ostwald ripening occurs due to the atomic migration from small particles to a larger one.
Note that the atomic migration can be explained by the Gibbs-Thomson relation, [1][2][3] which is expressed as follows, where C r is the surface concentration of diffusing atoms of the particle, C ∞ is the surface concentration of atoms in equilibrium with an infinitely large particle, γ is the surface energy of the particle, V is the volume of an adatom, k is the Boltzmann's constant, T is the temperature, and r is the radius of the particle. According to the above equation, small particles have higher C r than the larger particles. Owing to the concentration gradient, the atomic diffusion occurs from the small particle to a large one, resulting in the growth of the larger particle at the expense of the smaller ones. Figure S17. Fourier-transform infrared spectra of an UCNs-and SiO 2 NPs-embedded PUA film. a) UCNs-dispersed PUA film before annealing. b) UCNs-dispersed PUA film after annealing at 900 °C. c) UCNs-and SiO 2 NPs-dispersed PUA film before annealing. d) UCNsand SiO 2 NPs-dispersed PUA film after annealing at 900 °C. Note that the signal of the Si-O-Si bond at 900 cm -1 is more apparent after the annealing process. Figure S18. Variation in the XRD pattern according to the concentration of SiO 2 NPs. The UCNs-and SiO 2 NPs-embedded PUA film was annealed in an air-purged tube furnace at 900 °C for 1 h. Figure S19. XRD patterns revealing the crystal phase transition of UCNs. XRD pattern of a) pristine hexagonal NaREF 4 UCNs (β-NaYF 4 :Gd 3+ /Yb 3+ /Er 3+ (30/30/2 mol%)) and b) cubic NaREF 4 UCNs formed by the annealing of hexagonal NaREF 4 UCNs. c) Hexagonal apatite after the SiO 2 -involved annealing process. d) Transformation of the annealed cubic NaREF 4 UCNs to hexagonal apatite by introducing SiO 2 NPs and annealing. Note that annealing was performed at 900 °C for 1 h under continuous air flow. Figure S20. Comparison of the luminescence spectra of cubic NaREF 4 and hexagonal apatite UCNs. The black curve is the luminescence spectrum of cubic NaREF 4 UCNs obtained by the annealing of hexagonal NaREF 4 UCNs (β-NaYF 4 :Gd 3+ /Yb 3+ /Er 3+ (30/30/2 mol%)) and red curve is the luminescence spectrum of the hexagonal apatite phase UCNs obtained by the annealing of cubic NaREF 4 UCNs in the presence of SiO 2 NPs. Note that annealing was performed at 900 °C for 1 h under continuous air flow.

Supplementary Note 5. Construction of the energy level diagram of the Er 3+ ion
In this study, the energy level diagram of Er 3+ ion, which was doped in the UCNs with three types of crystal structures, was calculated under the crystal field interaction. To build the energy level diagram of the Er 3+ ion considering the crystal field effect, the effective-operator Hamiltonian model (Eq. 1) was used. [ where F k and ς f indicate the Coulomb and spin orbit interactions, respectively. f k and A SO are the angular parts of F k and ς f , respectively. α, β, and γ are the parameters associated with the two body correction terms. L is the total orbital angular momentum. G(G 2 ) and G(R 7 ) are Casimir's operators for groups G 2 and R 7 . T i and t i are the three-body interactions and operators, respectively. M h and P k are the correlated interactions of magnetic and electrostatic fields, respectively. m h and p k are the effective operators of M h and P k .
For the crystal field interaction, the Hamiltonian model is as follows: where B q k and C q (k) are the crystal field parameters and the spherical operators, respectively.
For the C q (k) term, the Wybourne notation is utilized in this calculation.
For the calculation of the Hamiltonian models, the SPECTRA program [5] was used. Based on the peak wavelengths observed in the UV-Vis-NIR absorbance spectra ( Figure S21), the parameters of free ion and crystal field interactions were optimized for the three crystal phases (Table S5). For the crystal field parameters, the site symmetries of Er 3+ ion, which are D 3h , O h , and C 3 point-group symmetries for hexagonal NaYF 4 :Er 3+ , cubic NaYF 4 :Er 3+ , and hexagonal Na 2 Y 8 (SiO 4 ) 6 F 2 :Er 3+ systems, respectively, were considered. Notably, the C 3 pointgroup symmetry was replaced by C 3v symmetry using the descent-of-symmetry method. [4] Based on the optimized parameters, the energy level diagrams of the Er 3+ ion corresponding to the three crystal phases were constructed ( Figure S22).   functional, [6] norm-conserving pseudopotential, and a plane-wave basis set were employed in all calculations. The kinetic energy cutoff was set to 780 eV and the Brillouin zone was sampled by 2 × 2 × 2 k-point grid with the Monkhorst-Pack scheme. [7] The self-consistent field calculation was performed with electron smearing of 0.5 eV, until the convergence criterion of 2 × 10 -6 eV/atom was satisfied. The ab initio molecular dynamics (AIMD) simulation was performed in an isothermal-isobaric ensemble (i.e., NPT ensemble), where temperature and pressure were controlled by Nose thermostat [8] and Andersen barostat, [9] respectively. Each system was simulated at 298.15 K and 1 atm for 26 ps with a time step of 1 fs. For geometry optimization, the convergence criteria were set to 2 × 10 -5 eV/atom for energy change, 0.05 eV/Å for maximum force, 0.1 GPa for maximum stress, and 0.002 Å for maximum displacement, respectively. Moreover, the phonon density of states of Er 3+ ion was calculated by the finite displacement method (supercell defined by a cutoff radius of 5 Å) and 4 × 4 × 3 (2 × 2 × 3) k-point for the Er 3+ -doped hexagonal or cubic NaREF 4 (hexagonal apatite) phase. Lastly, for calculating the electronic structures, Heyd-Scuseria-Ernzerhof (HSE) 06 hybrid functional [10] was used for obtaining accurate band structure and density of states.

B. Model systems 1. Crystal structure of the host nanocrystals
The crystal structures of NaYF 4 with P6 3 /m and Fm 3 m space groups and Na 2 Y 8 (SiO 4 ) 6 F 2 with P6 3 /m space group were used to construct model systems for hexagonal NaREF 4 , cubic NaREF 4 , and hexagonal apatite phases of the upconversion nanocrystals, respectively. For the cubic NaREF 4 phase, we employed the unit cell of NaYF 4 , where two out of four 4a Wyckoff positions were occupied by Y 3+ and the other two 4a positions were occupied by Na + ( Figure   S23a). For the hexagonal NaREF 4 phase, we used the 1 × 1 × 2 supercell of the NaYF 4 unit cell, where two 4e Wyckoff positions were occupied by Na + and four 2d Wyckoff positions were occupied by one Na + and three Y 3+ ( Figure S23b). For the hexagonal apatite phase, 4f Wyckoff positions were occupied by Y 3+ , 6h Wyckoff positions were occupied by Y 3+ and Na + , and 2a Wyckoff positions were occupied by F − (Figure S23c).

Crystal structure of UCNs doped with Er 3+
The crystal structures of hexagonal NaREF 4 , cubic NaREF 4 , and hexagonal apatite systems doped with Er 3+ were modeled by replacing one of the Y 3+ atoms by Er 3+ in the previously constructed model system ( Figure S23). All possible doping sites were investigated and the most stable Er 3+ -doped systems were adopted for the analysis of the phonon density of states and electronic structure ( Figure S24).

Figure S23.
Model systems for the crystal structure of upconversion nanocrystal. Three types of crystal structures, a) cubic NaREF 4 , b) hexagonal NaREF 4 , and c) hexagonal apatite phase structures, are considered. Note that for the hexagonal apatite phase, all possible configurations were investigated where Na + and Y 3+ ions were placed in 6h Wyckoff positions, and the most stable configuration was adopted. Figure S24. Model systems for the crystal structure of UCNs doped with Er 3+ . Three types of crystal structures with doped Er 3+ , a) cubic NaREF 4 , b) hexagonal NaREF 4 , and c) hexagonal apatite phase structures, are considered. Figure S25. Phonon density of states of Er 3+ in three types of crystal phases. Phonon density of states of Er 3+ in Er 3+ -doped hexagonal NaREF 4 , cubic NaREF 4 , and hexagonal apatite phases. Note that although the imaginary frequency (negative frequency), which indicates structural instability, appeared in the cubic NaREF 4 and hexagonal apatite phases, the contribution to the instability of the entire structure was marginal because the intensity was very weak.

Supplementary Note 7. Upconversion luminescence transition of Tm 3+ -doped UCNs
The blue UCNs (i.e., Tm 3+ -doped UCNs) showed phase transition from the hexagonal NaREF 4 phase UCNs to hexagonal apatite or cubic NaREF 4 phase in the presence or absence of SiO 2 NPs, respectively, when annealed at 900 °C ( Figure S26). This result indicates that, in our method, regardless of the type of the dopant (e.g., Er 3+ , Tm 3+ ), the crystal phase of the host matrix of the UCNs plays a crucial role in the tuning of the luminescence of UCNs when subjected to annealing. Based on these results, we could predict that the enhanced luminescence of the system with blue UCNs and SiO 2 NPs after the annealing process is induced by the phase transition of the UCNs (hexagonal NaREF 4 phase  hexagonal apatite phase). The difference between blue UCNs and yellow UCNs is the type of emitting ion, which determines the luminescence color. The luminescence color of blue UCNs is determined by Tm 3+ rather than Er 3+ . As with the case of the Er 3+ -doped UCNs, the upconversion luminescence of Tm 3+ -doped UCNs is influenced by the characteristic of the crystal phase of UCNs, cubic or hexagonal apatite phase ( Figure S27). In the cubic phase, processes such as cross relaxation ( Figure S27a), less excitation of ions by photons (left side of Figure S27b), and strong nonradiative relaxation ( Figure S27c) occur due to the shorter interionic distance of Tm 3+ ions, higher energy level of Tm 3+ , and higher phonon energy of Tm 3+ , respectively, as compared to those in hexagonal apatite phase. The hexagonal apatite phase showed higher photon excitation ( Figure S27b, right) and weak nonradiative relaxation ( Figure S27c). Based on the energy transfer mechanism of Tm 3+ , [11,12] we speculated that the weak intensity of the upconversion luminescence of the cubic UCNs is induced by the above-   We further expended the utility of our luminescence color transition mechanism to realize binary luminescence color-hidden microparticles ( Figure S28). First, as shown in Figure 5a (top), two laminar streams-both streams contained the same colored UCNs (yellow or green) but only one laminar stream contained SiO 2 NPs-were generated in the PDMS channel and photo-crosslinking was carried out in a continuous manner. However, two luminescence colors are distinguished upon increasing the temperature. When the temperature reached 900 °C, the color of the area containing SiO 2 NPs changed to bright green while that of the counterpart changed to red, indicating the generation of binary spectral colored microparticles from singly yellow or green microparticles (Figure 5a (bottom)). Thus, hidden spectral colors could be generated by simply increasing the number of multi-color emitting UCNs in the laminar flow streams and adjusting the presence or absence of SiO 2 NPs.

B. Decryption of hidden pattern and letters.
Microstructures with hidden pattern and letters were fabricated by controlling the photocrosslinking location of SiO 2 NPs. UCNs with/without SiO 2 NPs dispersed in a photocurable PUA resin (PUA and photo-initiator at 9:1 ratio) were coated on an acrylated substrate, followed by microparticle fabrication through selective DMD-patterned UV irradiation using LabVIEW program and an inverted microscope (Nikon Ti-E). The unpolymerized monomer was rinsed with ethanol. Note that microstructure fabrication was first performed in the presence of SiO 2 NPs (triangle of Figure 5b, background of Figure 5c) and then in the absence of SiO 2 NPs (circle of Figure 5b and letters (UNIST) in Figure 5c). After heat process, the area containing SiO 2 NPs in the yellow micro-post appeared as a green and micro-posts without SiO 2 NPs appeared red under 980 nm NIR laser illumination. Upon following a similar protocol, the blue micro-post appeared as a triangle with a distinguishable emission intensity difference during the high temperature thermal process (Figure 5b).
C. Glass forming for extreme thermal process monitoring system.
The UCNs embedded yellow luminescence micro-post array with/without SiO 2 NPs was fabricated on the acrylated substrate. For glass forming process, spreading of borosilicate glass powder that is typically used for laboratory glass equipment on the micro-post array, followed by annealing process.

D. Ceramic glazing for extreme thermal process monitoring system.
UCNs-embedded microparticles with/without SiO 2 NPs were fabricated by a stop-flow lithography technique. UCNs with/without SiO 2 NPs dispersed photocurable resin were flown in a microfluidic channel and polymerized by patterned UV irradiation. Figure S28. Schematic representation of the fabrication of multi spectral colored microparticles. Figure S29. Schematic representation of the fabrication of microstructure array for a hidden pattern labeling system.

Supplementary Note 9. Quantum yield measurement
Quantum yields (QYs) of hexagonal NaREF 4 and hexagonal apatite UCNs were measured using an integrating sphere setup at a laser power density of 20 W cm -2 , [13] as shown in Figure   S29.
The quantum yield is calculated as follows: [14] () The QY of an upconverting material is highly dependent on the power density of the NIR laser source because of the nonlinear process of upconversion. [15] Here, QY was measured at a power density of 20 W cm -2 of the NIR laser source. As a result, the QY of hexagonal NaREF 4 and hexagonal apatite phase UCNs were determined to be ~3.08% and 0.91%, respectively. The QY of hexagonal NaREF 4 is consistent with the value reported previously for this material. [14] Figure S30. Schematic representation of the integrating sphere setup used for quantum yield measurements.