Video: We presented a future virtual reality platform that is based on a meta-eyepiece and a laser back-illuminated LCD. This video shows that the meta-eyepiece is able to resolve pixels of the display.
Letting a computer make a decision is inherently scary. Would you consider giving a computer free reign to design a metasurface? Metasurfaces, nanostructure-patterned surfaces with tailored functionalities, can spatially engineer the wavefront of electromagnetic waves at will. Their fabrication process is compatible with foundries in the silicon industry. In contrast to traditional bulky compound curved lenses, which rely on phase accumulation through light propagation, metasurfaces engineer light scattering and are compact and lightweight. This makes metasurfaces very suitable for wearable applications. However, designing metasurfaces remains a significant challenge as their size and functionality complexity scale up. On the one hand, it is very difficult to simulate a large-area metasurface due to its multi-scale nature: nanoscale size (tens of nanometers) of meta-atoms and macroscale size (millimeters to centimeters) of devices. It is extremely hard to resolve both length scales in a single simulation. For example, even with the fastest state-of-the-art Maxwell’s solver, simulating a single metasurface of 1 cm diameter in the visible would take months. Computing capabilities hardly keep up with the fast development in fabrication technologies, which now allows larger areas and more complex shapes. On the other hand, optimization goals become more complex where multiple functionalities and design constraints need to be considered simultaneously. For example, correcting chromatic aberration, as in this work, is a complex objective where the scattered intensity needs to be maximized at the same focus for different wavelengths of light.
The conventional intuition-based forward design has been prominent until now, but it is limited in the number of parameters that can be considered simultaneously. Bounded by human working memory capacity, this number is limited to the order of 10. Moreover, it heavily relies on one’s training in physics and even requires a priori knowledge making it inaccessible to the general public. In this work, we developed a computational inverse design framework for large-area 3D complex metasurface design. We taught the knowledge of an approximated model of Maxwell’s equations to our computer program. In return, our program could generate metasurface designs automatically at scale, i.e., designing millions to billions of parameters simultaneously. What is more, the whole process takes less than a day using a single-CPU laptop. In this work, we report metalenses that correct the spherical aberrations and chromatic aberrations for up to 6 wavelengths in the visible. Our metalenses have a large aperture size of up to 1 cm diameter, which equals to 20,000 of the design wavelength. It represents a new record of complex metasurface design at scale.
The large-scale meta-optics are desired for many applications. In this work, we demonstrate a next-generation virtual reality (VR) platform. The current development of VR is bottlenecked by hardware issues due to the mismatch between electronics and optics technology. In contrast to the fast evolution of electronics, which follows Moore’s law, the development of optics is sluggish. Most VR headsets today use refractive lenses as the eyepiece, which are not only uncomfortable to wear but also compromise the viewing experience. Moreover, they are not compatible with the high-end displays that have high resolution and small pixel pitch. Our presented VR platform is based on a meta-eyepiece and a laser back-illuminated micro-LCD. This unique combination offers many desirable features, including compactness, high resolution, wide color gamut, etc. We believe the metasurfaces open a new path to reshape the future of VR.
In the future, we will continue to unlock the potential of complex nanoscale features over a large area to design the next generation of high-performance metasurfaces. Recent progresses in scientific machine learning will enable us to embrace more complicated feature shapes while scaling the metasurface area even further, which may achieve even better performance. We also plan to include new material properties, such as non-linearity, and add other physics in the design framework. This article lays the groundwork and design approach which may influence many real-world devices. Our methods will enable new metasurface designs that can make an impact on virtual or augmented reality, self-driving cars, and machine vision for embarked systems and satellites, where the compact and lightweight properties and advanced wavefront shaping capabilities of metasurfaces are highly desired.