A novel optoelectronic device named Solid-State Incandescent LED (SSI-LED) was reported by Kuo's group [1]. Intended to supplant conventional pn-junction LEDs in scenarios requiring ease of fabrication and compatible with Si technology, this device is made from a MOS capacitor with a high-k gate dielectric layer undergoing dielectric breakdown to form a large number of nano-sized conduction paths called Nano-resistors (NRs) [2]. Each NR has a small cross section, e.g., 10 nm diameter, and a high electrical resistance. When a voltage is applied across the device, these structures undergo incandescence, behaving similar to filaments in incandescent bulbs. As a result of thermal excitation, a broad and continuous spectrum of light is emitted, as opposed to narrow band lights from LEDs.Previous simulations of the device using COMSOL were limited by both the number of NRs that could be generated and the sizes of these NRs [3]. SEM images indicate that the SSI-LED should have at least a few hundred NRs per 30 μ2 [4]. In addition, simulation studies have shown that for certain devices, there is a higher density of NRs along the gate circumference [5].In order to extend both the functionality and flexibility of the current simulation to match the above requirements, a framework for SSI-LED geometry generation was made using Python. The current implementation can directly translate both a NR size in tabular format, and a position distribution in the form of concentric bands of limits expressed as rational multiples of the gate diameter, into built geometry within COMSOL. SEM images have also shown that NRs tend to form as aggregated clusters [4], which in the accompanying data were recorded as single large size light dots. In light of this, the code also can generate NRs in clusters by specifying the minimum and maximum 1) spread of clustered, 2) the number of NRs present in each cluster, and 3) the unique size distribution for clustered NRs. The program has the capability of having NRs generated both randomly as singular entities and as clusters in a simple percentage ratio. Additional parameters for finer control include a value to control the minimum space between random NRs and a clearance from the gate circumference edge.The python code removes the limit on number and sizes of NRs in the early light intensity simulations using MATLAB [3]. It extends utility into generating light intensity distribution data for the same geometry used in COMSOL, allowing for a one-to-one correspondence with the simulation model, and takes a number of steps to optimize the calculations for larger numbers of NRs and larger dot sizes.The output is stored as a file and accessed by a COMSOL-MATLAB Livelink, where it is possible to browse to any model through the Livelink interface and update the geometry of any COMSOL simulation model. This allows an already completed simulation model to be updated with the new geometry without having to redevelop the physics and post processing data contained in the simulation file.Figure 1 shows the light emitting patterns of a) a fabricated SSI-LED [6] and b) output from this simulation work, for comparison.[1] Y. Kuo and C.-C. Lin, Appl. Phys. Lett., 102, 031117 (2013).[2] Y. Kuo, IEDM, 4.7.1-4.7.4 (2014).[3] A. Shukla and Y. Kuo, ECS J. Solid State Sci. Technol., 9, 065017 (2020).[4] S. Zhang and Y. Kuo, J. Phys. D: Appl. Phys., 51, 09LT02 (2018).[5] K. Natarajan and Y. Kuo, to be published.[6] S. Zhang and Y. Kuo, ECS Trans., 85(3) 53-58 (2018). Figure 1
Read full abstract