Multilayer diffractive optical element material selection method based on transmission, total internal reflection, and thickness

Abstract

The polychromatic integral diffraction efficiency (PIDE) metric is generally used to select the most suitable materials for multilayer diffractive optical elements (MLDOEs). However, this method is based on the thin element approximation, which yields inaccurate results in the case of thick diffractive elements such as MLDOEs. We propose a new material selection approach, to the best of our knowledge, based on three metrics: transmission, total internal reflection, and the optical component’s total thickness. This approach, called “geometric optics material selection method” (GO-MSM), is tested in mid-wave and long-wave infrared bands. Finite-difference time-domain is used to study the optical performance (Strehl ratio) of the “optimal” MLDOE combinations obtained with the PIDE metric and the GO-MSM. Only the proposed method can provide MLDOE designs that perform. This study also shows that an MLDOE gap filled with a low index material (air) strongly degrades the image quality.

Thickness optimization algorithm to improve multilayer diffractive optical elements performance

“…4) are detailed. This EST input selection process is applied to three particular MLDOE combinations, ensuing from the MLDOE material selection process introduced in [13]. The IRG24-air-IRG27, ZnS-air-Ge, and IRG24-IRG27-AgCl combinations have been selected regardless of their potential manufacturing issues.…”

“…The extended scalar theory (EST), introduced in [10] and recently applied to MLDOEs [11], combines the grating equation and Snell's law to optimize the profile of a DOE (reducing the shadowing effect). Finally, ray-tracing and Fourier optics have been combined to estimate MLDOE's optical performance at the focal plane [12] and provide a material selection method [13]. This concept is not new and has been used for the modeling and design of hybrid optical systems [14,15].…”

“…This concept is not new and has been used for the modeling and design of hybrid optical systems [14,15]. The optical performance of many MLDOE examples has been accurately retrieved using the finite-difference time-domain (FDTD) method [12,13]. It has been proven that these MLDOEs, designed with TEA thickness equations, had much lower performance than expected.…”

“…The latter is retrieved using a free space Fourier optics propagator and a reliable scalar approximation of the MLDOE near field: the ray model (Section 3). This algorithm is applied in Section 5 for three MLDOE combinations, ensuing from the MLDOE material selection process introduced in [13]. As a result, a suitable EST input pair ( p, λ opt ) allows for generating an "optimal" MLDOE profile design for each material combination.…”

Review of Integrated Diffractive Deep Neural Networks

“…Though this chromatic system has higher information dimensionality bringing the prospect of multi-channel computing, it weakens the capability of processing complex information because all the computations are based on onefold light intensity. [31,32]…”