BROWSE
Detail
Journal Article
Artificial neural network based inverse design of organic light emitting diode structures for optimized optical properties
Cited - time in 
Share 
Share
- Authors
- Jun Hee Han
- Issue Date
- 2026-01
- Citation
- Engineering Applications of Artificial Intelligence, v.163, no.2, pp.1-12
- ISSN
- 0952-1976
- Publisher
- Elsevier
- Language
- English
- Type
- Journal Article
- Abstract
- Organic light-emitting diode displays are characterized by their nano thin-film composition, positioning them as advanced display technology. However, this feature poses significant challenges in the optical design of the device. Achieving the targeted optical features in an organic light-emitting diode device requires controlling light interference by adjusting the nanometer-scale thin film thickness. Optical simulations for this process demand expensive computing resources, which increase exponentially with structural complexity. Additionally, the light interference within the device is too complex to be managed effectively through human experience and intuition alone. In this work, we propose a inverse design method using artificial neural networks. The inverse design was executed by varying the target spectrum for six different device structures, resulting in the successful derivation of the layer thicknesses needed to achieve the desired spectrum. We provide a mathematical proof of the operating principle of inverse design using the proposed artificial neural network. Besides, the results obtained through inverse design are verified using finite-difference time-domain simulations, which are commonly employed for optical simulations in the design process. Since the proposed method is based on backpropagation and gradient descent, which are fundamental principles of artificial neural networks, it does not require high-specification computing resources such as a graphics processing unit. Additionally, its intuitive network structure allows application to an unlimited number of organic light-emitting diode structures by adjusting the network configuration.
- Keyword
- Inverse design, Neural networks, Backpropagation, Gradient descent, Organic light emitting diode
- KSP Keywords
- Artificial Neural Network, Computing resources, Design process, Device structure, Display technology, Film composition, Finite-difference Time-domain(FDTD), Gradient Descent, Inverse design method, Light interference, Light-emitting diode device