Directional enhancement of photoluminescence from phosphor plates with TiO₂ nanoantenna stickers

2.1.1 Fabrication of the hexagonal pattern sticker

The hexagonal pattern TiO₂ nanoantenna sticker was fabricated via nanoimprint lithography, followed by its transfer into the PDMS matrix. The fabrication process is shown in Figure 1a. First, a 750-nm-thick amorphous germanium dioxide (a-GeO₂) layer was deposited onto a commercial SiO₂ glass substrate via mist chemical vapor deposition. This a-GeO₂ layer served as a sacrificial layer for the subsequent transfer process. To improve its water solubility and minimize oxygen vacancies (GeO _x ), the film was annealed in an O₂ atmosphere at 750 °C for 10 min [35]. Then, a thin amorphous TiO₂ film was formed via sputtering, followed by the spin-coating of a resist layer (TU7, Obducat). Then, the designed nanostructure was patterned onto the resist via nanoimprint lithography (Entre3, Obducat). The resist pattern was further transferred to the TiO₂/a-GeO₂ layer by combining an O₂ ashing and neutral loop discharge etching using a CHF₃/C₄F₈/Ar/O₂ gas mixture, yielding TiO₂ NP arrays on the a-GeO₂ surface. For the transfer, a PDMS precursor was mixed with an appropriate curing agent, poured over the patterned surface, and cured at 80 °C for 2 h. Finally, the a-GeO₂ sacrificial layer was dissolved in ultrapure water for 24 h, yielding a free-standing PDMS film incorporating with the hexagonal TiO₂ nanoantenna array. A digital photo of the sticker on the YAG:Ce substrate is shown in Figure 1b. Refer to Supporting Information Figure S1 for the fabrication of square pattern stickers.

Figure 1:

Fabrication and structural schematic of the nanoantenna stikcer–YAG:Ce–DBR system. (a) Fabrication process of the hexagonal pattern sticker (period = 420 nm, nanoantenna area = 6 × 6 mm²). (b) Digital photograph of the hexagonal pattern sticker attached to the YAG:Ce plate. (c) Schematic illustration of the three-layer DBR structure. (d) Cross-sectional scanning electron microscope image of the six-layer DBR. (e) Schematic illustration of the sticker–YAG:Ce–DBR system.

2.1.2 Fabrication of DBR

The DBR was fabricated via sputtering using a multitarget sputtering system (EB1100, Canon Anelva Co., Ltd.). A 90-nm-thick SiO₂ thin film was deposited on the backside of the YAG:Ce substrate, followed by depositing a 180-nm-thick TiO₂ thin film. A schematic and scanning electron microscopy image of the DBR are shown in Figure 1c and d, respectively. The final sample configuration is depicted in Figure 1e.

2.2.1 Optical extinction

We measured the zeroth-order extinction spectrum by illuminating the sample with a collimated beam from a halogen lamp (SLS201L/M, Thorlab). We evaluated the optical extinction (E), defined as E = 1 −T/T ₀, where T represents the zeroth-order transmission through the nanoantenna sticker on the phosphor plate and T ₀ denotes the transmission through the flat phosphor plate. The zeroth-order transmission was collected using a fiber (numerical aperture of 0.22) coupled to the spectrometer (Flame-S, Ocean Insight). The angle of incidence (θ _in) was varied in the z − x plane, with the azimuthal angle from the x-direction being 0°.

2.2.2 Photoluminescence

The nanoantenna sticker was placed on the forward side of the YAG:Ce plate, and the sample was placed on the rotation stage. A diode-pumped solid-state laser (λ = 440.6 nm; Shanghai Dream Lasers Technology) was incident onto the substrate side at the angle that exhibited the strongest emission in each case. The PL that radiated to the opposite side through the sticker was collected by an optical fiber coupled to the spectrometer as a function of the emission angle θ _em defined in the z − x plane. The PL enhancement was defined as the PL intensity obtained with the sticker, I, divided by the value obtained without the sticker, I _ref, i.e., I/I _ref.

2.2.3 Integrating sphere measurements

The quantum yield and the CIE chromaticity diagram were evaluated using a calibrated integrating sphere [diameter = 10 in. (0.254 m)] (RTC-060-SF, Labsphere). An LD (λ = 457 nm) was incident on the sample from the substrate side. The output light was integrated throughout the sphere and collected using a fiber-coupled spectrometer. The spectral range between λ = 419 and 798 nm was used for the chromaticity calculation. The color coordinates were determined using the CIE 1931 XYZ chromaticity system.

2.2.4 Simulation

Finite-difference time-domain (FDTD) simulations were performed using a model consisting of cylinder NPs made of TiO₂ with a diameter of 200 nm and a height of 150 nm. The YAG:Ce substrate and the PDMS superstrate were semi-infinite layers with refractive index, n = 1.820 and 1.415, respectively. A 20 nm PDMS layer was inserted between the nanoantenna and the substrate to reproduce the experimental situation. Periodic boundary conditions were applied in the lateral direction to produce a hexagonal lattice with a periodicity of 420 nm, and perfectly matched layers were used along the z-direction. In all simulations, the incident plane wave was applied from the top (PDMS) side with a linear polarization along the x-direction. The simulations for the square lattice and a single NP were also done.

The DBR transmission and reflection were also simulated using the FDTD method. The DBR model comprised alternating layers of SiO₂ (90 nm) and TiO₂ (180 nm), with their refractive indices set to 1.460 and 2.400, respectively.

3.1.1 Hexagonal pattern

Figure 2 summarizes the PL enhancement spectra of the YAG:Ce plate with the hexagonal pattern sticker. Here, the same sticker was placed on YAG:Ce plates with varied t. In each panel, the PL enhancement was plotted using the bare plate with identical t but without the sticker as the reference. An obvious t dependence was observed: as the substrate became thinner, the enhancement factor increased, accompanied by an improvement in PL directionality. The enhancement factor reached a maximum of 5.64-fold at λ = 548 nm for the 110-μm-thick plate. The sticker can be detached and re-attached repeatedly, and reproducible PL enhancement is obtained each time (see Supporting Information Figure S2).

Figure 2:

PL enhancement spectra of YAG:Ce substrates with TiO₂ nanoantenna stickers (hexagonal pattern, period = 420 nm) detected in p-polarization. Substrate thicknesses (t) are (a) 477, (b) 325, (c) 220, and (d)–(h) 110 μm, the left inset in (a) shows the particle arrangement. The insets in the upper-right corners of panels (a)–(h) indicate the substrate thickness and the presence or absence of the DBR layer. (e)–(f) PL enhancement spectra for t = 110 μm sample with (e) three- and (f) six-layer of DBR. (g)–(h) PL enhancement spectra of the t = 110 μm plate with (g) three- and (h) six-layer DBR, without nanoantenna sticker. (i) PL spectra of the t = 110 μm plate with six-layer DBR and the hexagonal pattern sticker, in comparison to that of the bare 110 μm-thick substrates at θ _em = 0°. (j) Angular profiles of the PL enhancement spectrally integrated over λ = 500–650 nm. (k) Normalized angular profiles of the integrated PL enhancement.

The dotted lines in the panels represent the in-plane diffraction conditions or RA lines, which can be expressed as follows:

where k ⃗ 0 denotes the wavevector in free space, k ⃗ in ( θ in ) denotes the in-plane component of the incident wavevector at an incident angle θ _in, and G ⃗ ( m 1 , m 2 ) denotes the reciprocal lattice vector indexed by integers m ₁ and m ₂. When the incident light has no component along the y-direction, as in the present case, the magnitude of the free-space wavevector k ⃗ 0 is expressed for a hexagonal lattice as

where a represents the lattice constant which is 420 nm in the present sample and m ₁ and m ₂ denote the diffraction orders [26]. RA depends on n of the environment surrounding the nanoantenna, and RAs with n = 1.415 (representing PDMS) and 1.820 (YAG:Ce) are plotted in Figure 2a–f (white dotted lines). Moreover, the main enhancement was confined to the wavelength range between the two RA conditions (n _YAG:Ce and n _PDMS), since there intensified outcoupling is realized [30]. While in other regions, the constraint condition induces part-depressed coupling, and thus results in lower enhancements.

These results mean that the observed PL enhancement is strongly governed by the optical scattering conditions determined by the NP arrangement. To be more precise, the confinement of the enhancement between these scattering condition wavelengths indicates that the directional output is due to coherent scattering of the PL otherwise trapped inside the plate by total internal reflectance or the mismatch of in-plane momentum (k _‖). As shown in the extinction spectra in Supporting Information Figure S3, SLRs are supported in the samples even though they are not very strong. SLR is induced via coupling between the localized Mie resonance of the TiO₂ NPs and the diffractive orders of the periodic array. In Supporting Information Figures S4 and 5, we simulated the field distributions associated with the Mie resonance of the single NP and the SLR of the periodic array, respectively. The results feature a broad Mie resonance of each NP coupling via in-plane diffraction to induce SLR. In the PL outcoupling process, SLRs couple to and redirect PL into the forward direction defined by the pattern and periodicity of the nanoantenna, thereby yielding a notable PL enhancement.

The DBR further enhances forward radiation, as evident by the comparison between Figure 2d–f. Compared with the 110-μm-thick YAG without DBR, where the maximum enhancement is 5.64-fold (Figure 2d), the introduction of three or six layers of DBRs leads to the enhancement up to 9.67- or 11.47-fold, respectively. Because DBRs are designed to transmit blue light and reflect yellow light (see Supporting Information Figures S3b, c and S6 for the spectrum), the DBRs do not reduce the excitation power of blue LDs, and the PL radiated into the backside is reflected into the forward direction. This effect of DBR is better examined in Figure 2g and h, where the PL enhancements from the plates with DBRs are plotted with a bare substrate as a reference. The enhancement is broadband in the 520–600-nm range and dispersionless, consistent with the transmission spectrum in Supporting Information. The enhancements are approximately 1.5- and 2-fold for the three- and six-layer DBRs, respectively. Given the effect of DBR, which recycles the PL radiated to the backside, the maximum value of 2-fold is reasonable. In addition, the enhancement is discontinuously high at θ _em = 0°, because the YAG:Ce plate with DBR acts as a Fabry–Perot cavity and the confinement of light is strongest at θ _em = 0°. This discontinuous enhancement is also evident after placing the sticker (Figure 2e and f).

To closely examine the change in the spectral features, Figure 2i plots the PL spectra at θ _em = 0° for the YAG:Ce plate (t = 110 μm) with the six-layer DBR and nanoantenna sticker, and the bare YAG:Ce plate (t = 110 μm). The reference shows a broadband yellow PL centered at λ = 540 nm, which is typical for YAG:Ce phosphor. In contrast, the PL spectrum from the antenna and DBR plate show significantly higher intensity, with several bumps arising from the spectral variation of the reflectance from the DBR. An evident bump at λ = 625 nm is the edge of the DBR reflectance band, and the reflectance is significantly at longer wavelengths.

The directionality of the PL enhancement is better visualized by the polar plot. In Figure 2j and k, we integrated the PL intensity between λ = 500 and 650 nm and then divided the integrated intensity by that for the bare substrate, to obtain the angular enhancement profiles. Figure 2j shows that the thinner substrate results in stronger enhancement with narrower angular distribution, which is further increased by the DBRs. The increased directionality becomes more evident by normalization (Figure 2k). PL output is a sum of directional components and a diffuse background. Given that the directional output is due to coherent scattering of the PL otherwise trapped inside the plate by total internal reflectance, the thinner plate contains more trapped PL photons in the detection spot. An analysis of the thickness dependence of PL directionality is found in Supporting Information Figure S7.

3.1.2 Square pattern

Subsequently, we investigated the PL enhancement using the square pattern nanoantenna sticker on the t = 110-μm YAG:Ce plate (see Supporting Information for fabrication of the square lattice sticker). In this case, the enhancement increases with the number of DBR layers, as shown in Figure 3a–c. The dotted lines represent the RA lines for the square lattices with a = 390 nm and n = 1.415 and 1.820. For the square lattice, the reciprocal lattice vector is given by

Figure 3:

PL enhancement spectra of 110 μm-thick YAG:Ce substrates with square pattern TiO₂ nanoantenna stickers (period = 390 nm) detected in p-polarization. DBR layer number are (a) 0, (b) 3, and (c) 6. The insets in the upper-right corners of panels (a)–(h) indicate the substrate thickness and the presence or absence of the DBR layer. (d) PL spectra of the t = 110 μm-plate with 6-layer DBR and the square pattern sticker, in comparison to that of the bare 110 μm-thick substrates at θ _em = 0°. (e) Angular profiles of the PL enhancement spectrally integrated over λ = 520–620 nm. (f) Normalized angular profiles of the integrated PL enhancement.

Compared with the hexagonal lattice, the enhancement zone shifts due to differences in both the lattice arrangement and period. Especially, the line associated with (m ₁, m ₂) = (0, ±1) diffraction order lies around 550 nm at normal incidence and shifts to 520 nm at 30°. A notable PL enhancement appears along this RA line in the whole angle range of measurements because the line is inside the spectral region of the maximum PL intensity of YAG:Ce. The effect of (0, ±1) order is evident in the spectrum at θ _em = 0° (Figure 3d), where a dip is observed around λ = 550 nm. The effect is also reflected on the less directional PL profile for the square pattern case (Figure 3e and f). Also the magnitude of PL enhancement is less than that for the hexagonal lattice case. This is because the square lattice is not the close packed lattice and there are less NPs in a unit cell to gather and redirect the emitted PL. We further confirm our analysis using simulation (see Supporting Information Figure S8).

The comparison of the results for hexagonal and square pattern stickers confirms that the enhancement profiles, both spatially and spectrally, can be flexibly tuned by adjusting the lattice symmetry and periodicity of the sticker.