An epsilon-near-zero-based nonlinear platform for ultrafast re-writable holography

TL;DR

The paper addresses the need for fast, reconfigurable, all-optical holographic surfaces to enable ultrafast optical computation and dynamic light shaping. It demonstrates a nonlinear ENZ platform using a subwavelength thick (310 nm) ITO film, where interference between object and reference beams at ENZ frequencies writes a transient hologram that a separate read beam diffracts from. Key results include ~3% diffraction efficiency across a >$300$ nm bandwidth at telecom wavelengths, and a reported refresh rate up to ~1 THz, enabling ultrafast rewriting and temporal multiplexing. The work enables on-demand, nanofabrication-free reprogrammable diffractive surfaces for all-optical transduction/edge-detection and points to extensions in higher-order differentiation, convolution, and neural-network-like processing.

Abstract

We re-examine real-time holography for all-optical structuring of light and optical computation using a contemporary material: a subwavelength-thick, spatially unstructured film of indium tin oxide (ITO). When excited by spatially structured light at epsilon-near-zero frequencies, the film acts as an efficient and reconfigurable diffractive optical platform for all-optical modulation of light such as spatial structuring and optical computations. We demonstrate a few percent of absolute diffraction efficiency over greater than 300 nm bandwidth around telecom wavelengths using a film four orders of magnitude thinner than and up to six orders of magnitude faster than standard holographic materials. Our findings highlight the potential of using epsilon-near-zero-based nanostructures for efficient modulation of spatially structured light and rapid prototyping without complex nanofabrication processes.

Figures (7)

• Figure 1: Experimental configuration. (a) Interference of the object and the reference beams (orange) imprint a spatially varying hologram (a transient metastructure) onto the ITO film through an intensity-dependent change of refractive index (the example shown in the inset is for an object beam carrying one unit of OAM, that is, $\ell = 1$). A read beam (red) of a distinct wavelength diffracts off the hologram. The diffracted image (conjugate image) beam carries the same (complex conjugate) spatial structure as the original object beam. (b) A camera image taken of the beams exiting the ITO film shows all relevant beams. For clarity, the two image beams were brightened by a factor of 10 (adjusted area indicated by dashed line). See Fig. S2 for the original unprocessed image.
• Figure 2: Transfer of spatial amplitude and phase information. The induced holograms are sensitive to the spatial amplitude and phase of the object beams. To confirm the transfer of phase information, we prepare object beams with different signs and values of OAM ($\ell = \pm1,2,5$) and show the spatial structure of the resulting image beams. By performing cylindrical transformations and comparing the spatial structure and orientation of the transformed image beams with theory, we confirm the transfer of spatial phase information.
• Figure 3: Efficiency of the information transfer process. (a) The maximum efficiencies of various OAM modes, their superpositions, and a few Hermite-Gaussian modes. (b) Dependence of information transfer efficiency on object beam intensity with $\ell=+1$. We set the intensity of the reference beam to be $\rm{300\,GW/cm^2}$ and measure the diffraction efficiency of the image beam as a function of the object beam intensity at the ITO layer. (c) Temporal duration of the induced transient holographic structure. We vary the delay between the induced holographic structures (formed by the reference and object beams) and the read beam while recording the diffraction efficiency of the generated image beam for various spatial modes. We find that the hologram is only induced for the duration of a few hundred femtoseconds, suggesting a refresh rate of up to $\rm{1\,THz}$. (d) Tolerance of read beam wavelength. While the object and reference beams were set to 1260 nm, we varied the read beam wavelength, showing significant image diffraction efficiency across more than 300 nm.
• Figure 4: Holographic 2D spatial first-order differentiation for edge detection. When the object beam carries a topological charge ($\ell=+1$), the two-dimensional spatial differentiation of the image-carrying read beam is generated in the diffracted image beam.
• Figure 5: Permittivity of the ITO film used.