Demonstrating broadband thermal imaging using superoptical technology in a new framework

The research team used a new reverse design framework to demonstrate ultra optical broadband thermal imaging for applications ranging from consumer electronics to thermal sensing and night vision.

The new framework, known as the "Modulation Transfer Function" project, solves the challenges related to broadband metaoptics by determining the functional relationship between image contrast and spatial frequency maintained by the lens.

The University of Washington team manufactured their designed optical devices using a single silicon wafer, which is very promising for future applications involving germanium free longwave infrared imaging systems.

The next generation of optical systems requires lenses not only to be lighter and thinner than ever before, but also to maintain uncompromising image quality. This demand has driven a surge in efforts to develop ultra-thin subwavelength diffractive optical devices. Superoptics, in its simplest form, consists of an array of subwavelength scale nanorods on a plane, each of which introduces local phase shift for passing light. By strategically arranging these pillars, light can be controlled to generate steering and lenses.

The thickness of traditional refractive lenses is close to one centimeter, while the thickness of superoptical elements is about 500 microns, which greatly reduces the overall thickness of optical elements.

However, one challenge of metaoptics is the strong chromatic aberration. That is to say, light of different wavelengths interacts with the structure in different ways, and the result is usually that the lens cannot simultaneously focus light of different wavelengths on the same focal plane. Due to this issue, although superoptical devices have advantages in size and weight reduction, they have not yet completely replaced refractive optical devices.

In particular, compared to visible wavelength superoptics, the field of long wavelength infrared superoptics has relatively not been developed. Given the unique and widespread application of this wavelength range, the potential advantages of superoptics over traditional refractive lenses are significant.

A key innovation in the research team's approach is the use of artificial intelligence to draw maps between the shape and phase of columns. For the reverse design process of large-area optical devices, simulating the interaction between light and each column in each iteration is computationally infeasible. To address this issue, the author simulated a large nanopillar library and used simulated data to train DNN. DNN has achieved rapid mapping between scatterers and phases in optimized circuits, enabling reverse design of large-area optical devices containing millions of micrometer sized pillars.

Another key innovation of this work is the "quality factor". In reverse design, define FoM and optimize the structure or arrangement through calculation to maximize FoM. However, it is often not intuitive to explain why the resulting results are optimal. In this work, the authors utilize their expertise in superoptics to define an intuitive FoM.

Professor Arka Majumdar, who led the project, explained that the quality factor is related to the area under the MTF curve. The idea here is to transmit as much information as possible through the lens, which is captured in the MTF. Then, combined with a lightweight computing backend, we can obtain high-quality images.

The quality factor reflects our intuitive understanding of optics. When all wavelengths perform equally well, this specific FoM is optimized, limiting our optical devices to have uniform performance at specified wavelengths without explicitly defining uniformity as an optimization criterion.
This method combines the intuition of superoptics and lightweight computing backend, significantly improving performance compared to simple superlenses.

Although it is acknowledged that there is still room for improvement in achieving imaging quality comparable to commercial refractive lens systems, this work represents an important step towards this goal.

Source: Laser Net