Abstract
Wavelength-selective thermal emitters (WS-EMs) are of high interest due to the lack of cost-effective, narrow-band light sources in the mid- to long-wave infrared. Cost-effective WS-EMs can be realized via Tamm plasmon polariton (TPP) structures supported by distributed Bragg reflectors (DBRs) on metal layers, however, optimizing TPP-WS-EMs is challenging because of the large number of parameters to optimize. To address this challenge, we use stochastic gradient descent (SGD) to optimize TPP-WS-EMs composed of an aperiodic DBR deposited on doped cadmium oxide (CdO) plasmonic films. While the SGD-optimized, aperiodic DBR offers extensive spectral control, the material choice, i.e., plasma-frequency-tunable doped CdO, enables the design capabilities not accessible with noble metals. Here, the individual layer thickness and carrier density of CdO are optimized by our SGD inverse design strategy. The resultant experimental designs demonstrate TPP-WS-EMs exhibiting isolated, high-Q (narrow bandwidth), and structures featuring multiple emission bands for applications such as free-space communications and gas sensing. Furthermore, we illustrate the deterministic design capability within the infrared, such as user-designated Q-factors (28 − 10,127) at a desired frequency, multi-band emitters with user-defined Q, and the ability to directly match arbitrary chemical absorption spectra. Thus, the combination of our SGD inverse design and the broadly tunable plasma frequency of CdO enables lithography-free, CMOS-compatible, and wafer-scale solutions for WS-EMs with unprecedented spectral control.
Introduction
The mid- to long-wave infrared (MWIR/LWIR) spectrum, often referred to as the "fingerprint region," is critical for chemical sensing as most molecules exhibit unique absorption signatures in this range. Detecting substances like CO₂ or environmental pollutants requires narrow-band light sources. However, traditional nanophotonic emitters face a significant "trilemma": high fabrication costs due to lithography, limited tunability of noble metals like gold, and the computational complexity of optimizing multi-layer structures. Designing a simple 9-layer stack using brute-force parameter sweeps could take thousands of hours.
This research introduces a "deterministic" approach to infrared light source design, allowing researchers to effectively "paint" a desired emission spectrum. By combining an efficient inverse-design algorithm with a tunable material platform, high-performance thermal emitters can be produced without expensive lithography.
Methods: SGD Inverse Design and Tunable Materials
The core of this approach lies in pairing a novel material platform with modern optimization. Instead of fixed-property metals, the researchers utilize n-doped Cadmium Oxide (CdO). The optical response of CdO can be precisely tuned during growth by adjusting its carrier density.
The device architecture utilizes Tamm Plasmon Polariton (TPP) structures—aperiodic Distributed Bragg Reflectors (DBRs) placed atop a conducting CdO film. This configuration traps light at the interface, generating sharp emission peaks without the need for etched patterns.
To optimize these structures, the team treated the design challenge as a machine learning task. Using Stochastic Gradient Descent (SGD) in conjunction with the Transfer Matrix Method (TMM), they defined target spectra and allowed the algorithm to automatically adjust layer thicknesses and CdO carrier density to minimize error. This method reduces optimization time from weeks to mere minutes on a standard desktop computer.
Key Results and Experimental Validation
The inverse-design strategy unlocked several key capabilities:
- Aperiodic Precision: Unlike traditional periodic stacks, aperiodic designs allow for the suppression of unwanted "spurious" peaks and higher spectral purity.
- Record Selectivity: The team demonstrated theoretical Q-factors (selectivity) ranging from 28 to over 10,000.
- Multi-Gas Sensing: A single device was designed and fabricated with two distinct emission peaks tailored to detect CO₂ and SO₂ simultaneously.
- Complex Matching: The algorithm successfully matched the complex absorption spectrum of DMMP (a nerve agent simulant), a task where traditional gold-based emitters fail.
Experimental validation on 2-inch wafer-scale samples showed excellent agreement with simulations, confirming the deterministic nature of the design process.
Conclusion
By treating material properties as design variables and utilizing SGD for rapid optimization, this work provides a CMOS-compatible, wafer-scale solution for infrared emitters. While minor spectral shifts occurred due to fabrication tolerances in the research setting, commercial-grade deposition systems are expected to resolve these discrepancies. Future work will focus on integrating these emitters into portable, on-chip sensing systems for real-world environmental monitoring.
Summary updated