Microelectromechanical Systems or MEMS for brevity and catchiness is an interdisciplinary field dealing with microsystems that have mechanical and electrical components. MEMS devices are slowly becoming the next generation of silicon technology in the world. Typically these devices have sensing and actuating elements that serve various engineering applications but as usual I'm more interested in the science at the micro and nano-scales. Hence the areas that truly interest me here are nano-scale transport and concepts in microfluidics.
When fluid flow systems scale down, surface effects tend to dominate over bulk volume effects. When this happens, our models for macroscale systems start to break down and no longer serve as reasonable estimates for realistic conditions. Below are some projects where I have attempted to model micro and nano-scale fluid systems.
This was a study project that I took in 4th year of my undergraduate degree. The goal of this project was to understand evaporation in the nano-scale under the influence of surface effects. To do this, non-equilibrium MD simulations were used. A thin film of argon was heated and the evaporation properties were observed. Following this, we did the same evaporation experiment under the presence of a platinum slab and we saw that the argon gas preferentially wets the platinum slab. We can quantify this "wetting" with the Hamaker constant, which we obtained from analysing the simulation output.
To conduct the MD simulations, I used Molly.jl, an open-source Julia code, which I have contributed to.
Understanding the physics of the microflows involved is crucial to designing, constructing and using MEMS devices. However, in microfluidics, the surface effects are much more prominent than volumetric effects. A good metric for classifying the flow regime in these cases is the Knudsen number which is defined by Kn = λ/L where λ is the mean free path of the molecules in the system. We would not expect the models for bulk fluid flows to scale for these sort of systems with considerably varied Knudsen numbers. Hence the standard models and assumptions of Navier-Stokes equation and no-slip conditions may not apply. In this project, I have explored the usage of direct simulation monte carlo (DSMC) simulations and compare them with the Navier-Stokes models for microchannels and compared the differences to determine the best model for understanding microflows in these systems.
Electronics cooling is an intricate modern problem and amateurs and professionals alike are keen on solving it. With graphical processing units (GPUs) taking the world by storm either for video gaming at 60 frames per second, graphical designing or running scientific and numerical software in parallel to optimize performance, two massive engineering problems arise. One is powering so many appliances with GPUs and the second is making sure these high power electronics are sufficiently cooled to maintain device performance. As someone who has personally experienced the drastic effects of poor cooling in printed circuit board (PCB) electronics, I was keen to solve the latter problem. My Raspberry Pi model 4 B with 8 GB RAM was left running for 3 days continuously without a proper cooling solution and at the end of the 3 days the main integrated circuit (IC) had heated up so much that the device stopped working. Having seen how the heat effects could damage the IC, the next board I purchased was a Jetson Nano model with a GPU (which increases the heat load on the system) but also a heat sink.
I was scared that a heatsink may not be enough, so I wanted to simulate things to see if they get too hot. For this I found the perfect software : ANSYS Icepak. I tried three different simulations, one with just a board of FR4 (a common PCB material) and let an IC heat it up and saw the damaging effects. The second included a heatsink and under long and intense operating conditions, the heatsink wasn't enough. So then I tried simulating a forced convection heat transport cooling solution and it seemed to do the trick!
The design and mass manufacturing of smart soles in sports shoes is a highly constrained design problem. There is only a 1 cm portion that tapers in size and under 50 grams of mass to work with. With such scarce resources, this study's objective was to design a shock absorber system for the comfort and safety of athletes. For this, we must meet the sensing, actuating, circuitry and power requirements in the room we have to work with. This study uses a grid of cells made of smart materials to implement the shock-absorbing system. Each grid cell comprises a pressure sensor, signal conditioning circuitry and a mechanical actuator. Several options for the individual elements are to be considered and experimented with. The study considers piezoresistive and piezoelectric pressure sensors to compute the extent of stress applied at a local point in the sole. One possible candidate material to introduce damping was electrorheological fluids. The increased viscosity of the fluid under the application of an electric field would allow for one potential damping mechanim. Packaging comes up as one of the primary issues since the sole has to be thin, light, flexible and comfortable. Furthermore, a 3D integration scheme is required. The sensing element needs to be at the bottom of the stack so as to sense the pressure from the impact better. Then comes a layer of circuitry, comprising signal conditioning and drive circuitry, and the corresponding power rails. The signal conditioning circuitry must be atop the sensor and simultaneously wired to it. The drive circuitry must be beneath the actuating smart materials but also connected to it. Lastly, an outer gel covering must be present as present in normal sports footwear. To tackle this challenge, we simulated a single cell of damping material in COMSOL and observed its response to various stresses.