Study of Packaging and Assembly Materials for Thermal Performance Enhancement of Optoelectronic Devices with Small Form Factor
- Author(s): Huang, Linjuan
- Advisor(s): Shi, Frank G.
- et al.
Recently, there is an evident trend of ever-thinner and intricate opto-electronic package and devices, which brings about severer thermal issues as well as unprecedented challenges for the thermal design. Not only thickness of opto-electronic package itself will raise the peak temperature and pose a potential risk to electronic devices, but also the limited-space and interacted opto-electro-thermo-mechanical properties restrict the use of traditional active thermal management means and precise estimation of cooling performance. This dissertation aims to numerically and experimentally analyze thermal behaviors of various LED package configurations while thinning it as well as apply novel radiation coating and packaging materials to cooling electronic devices with limited-space. What’s more, an opto-thermal coupled numerical method in the application of LED was discussed. Last but not the least, much faster Machine Learning (ML) algorithms were adopted to predict patterns for cooling performance of our thermal radiation coating in Li-ion battery system and acceleration stage before the onset of thermal runaway, which is the key to cut-off the battery before explosion.
It’s found that: (1) LED packages have different thermal behaviors for various configurations while thinning it. There can be a 5-10 ℃ of temperature change for single-chip LED. (2) For white LEDs and light bars in direct view LED backlight units (DLED-BLUs), our thermal conductive Die Attach Adhesive (DAA) and thicker encapsulant are able to increase the power level by up to 1.5 times. (3) Our thermal radiation can effectively decrease the peak temperature by 14.5 oC for linear LED modules as well as boost the uniformity of temperature distribution. (4) The thermal performance of LED package is different with and without considering optical effect. So optical effect should not be neglected while conducting thermal simulation for opto-electronic devices. (5) Machine learning algorithms can shorten prediction time of thermal performance of passive radiation material from 1 day to less than 0.5 min, compared with finite element method. A practical system design to effectively prevent explosion of Li-ion battery system is provided, too.