Electro-Optic Modulation in Polycrystalline Barium Titanate Metasurfaces Enhanced by Poling

Abstract

Electrically tunable metasurfaces leveraging the strong Pockel's effect in barium titanate (BaTiO$_3$ or BTO) are a promising platform for reconfigurable free-space optical devices. However, the high cost, limited scalability, and restricted substrate compatibility of epitaxial BTO films hinder its exploitation. Here, we demonstrate free-space optical modulators based on imprinted BTO metasurfaces with targeted designs for optical and electric field confinement within the active material. With resonances exhibiting high quality factors of up to 200, we demonstrate improved transmission modulation at sub-volt driving amplitudes and frequencies up to 5 MHz. Additional enhancement is achieved via ferroelectric domain alignment, resulting in up to 25 % higher modulation strength compared to the unbiased case and up to 75 % compared to previous demonstrations. This enhanced EO response, arising from the effective permittivity engineering and domain orientation in these polycrystalline metasurfaces, holds significant potential for scalable and efficient EO modulators and active metasurfaces.

Figures (4)

• Figure 1: Structural configurations of barium titanate (BTO) metasurfaces. (a) Schematic of the device used to demonstrate electro-optic modulation of the transmitted signal under an applied AC voltage across the sandwiched ITO electrodes. The BTO pillars are buried in a SiO2 layer and a conformal insulating $\mathrm{AlO_x}$ film is deposited on top of the capping SiO2 layer. (b) Scanning electron micrograph (SEM) (tilted at 30$^\circ$) showing the top view of the imprinted metasurface consisting of arrays of BTO pillars with a periodicity of p=500 nm and a diameter of d=200 nm. (c) Sketch of the embedded configuration, where SiO2 is etched to the height of the pillars. (d) Cross-sectional SEM of the embedded device. (e) Schematic cross-section of the conformal configuration, where $\mathrm{AlO_x}$ and ITO are deposited conformally over the nanostructures. (f) Cross-sectional SEM of the conformal device. All the cross-sectional SEM images were prepared using focused ion beam (FIB) milling. The devices shown in (d) and (f) are clone devices fabricated on silicon substrate without a ground ITO electrode below the metasurface for better imaging quality.
• Figure 2: Electric and optical field simulations for the two electro-optic device configurations. Panels (a) and (b) show finite-element simulations of the electric and optical fields, respectively, for the embedded structure at 783.6 nm. Panels (c) and (d) correspond to the electric and optical field distributions, respectively, of the conformal structure at 768.6 nm. The electric field lines are overlaid in white in panels (a) and (c). The color scales for the electric field ((a), (c)) and the optical field ((b), (d)) are shown on the right for both device configurations. For the electric field, the color scale is saturated at 5 MV/m, with a maximum field strength of 120 MV/m in $\mathrm{AlO_x}$ and SiO2 (in green). The scale bar and coordinate axes shown at the bottom right of panel (d) are global and common for all simulations.
• Figure 3: Experimental and simulated results for the embedded and conformal device configurations. Panels (a) and (b) illustrate the measured transmission spectrum (in gray) and electro-optic (EO) modulation (in red) of the embedded and conformal structures, respectively, when an electric field with an amplitude of 1.5 V and a driving frequency of 400 kHz are applied. The transmission spectra were obtained by sweeping the laser wavelength and acquire the DC component of the demodulated signal ($T_{V_{DC}=0}$). The transmission data are normalized by the transmission of an unstructured region. The EO modulation is defined as the ratio $\frac{\Delta T}{T} = \frac{T_{V=V_{pp}} - T_{V=0}}{T_{V=0}}$, where $V_{pp}$ is the peak-to-peak amplitude of the sinusoidal wavefunction applied at a specific frequency. It is noted that the negative values reported for the EO efficiency stem from a sign change in the transmission signal relative to the unbiased signal. Therefore, only the absolute value of the EO modulation should be considered. The vertical dotted lines in the plots mark the different wavelengths at which the EO response was further examined in Figures S4(a)-(d) in the SI section 4. Panels (c) and (d) present the finite-element simulated transmission spectra (in gray) and EO modulations (in red) of the embedded and conformal architectures, respectively.
• Figure 4: Experimental electro-optic performance of the embedded and conformal devices. (a) Relative modulation of both devices as a function of the applied AC driving frequency for a fixed amplitude of 1.5 V, measured at the optimal operating wavelengths of each device. (b) Temporal evolution of the relative modulation of the conformal device under a DC bias of an absolute value of 10 V (33 MV/m) applied for 100 minutes, during which ferroelectric domains alignment enhances the electro-optic modulation. After the bias is removed, the modulation remains stable during a two hours relaxation period.