Sustainability Research @SEEEL
My work as a research intern at the Georgia Tech Social Equity & Environmental Engineering Lab (SEEEL) explores the fabrication parameters of magnetite nanoparticles in wastewater treatment, emphasizing environmentally sustainable fabrication methods. I am currently leading this project, and working under the mentorship and guidance of Mr. Ahmed Ibrahim Yunus, a current PhD student. Additionally, I attend biweekly lab meetings led by Dr. Joe Bozeman III, the lab's Principal Investigator. Find more about the SEEEL lab's mission and initiatives below.
Research Objective
I am currently conducting a meta-analysis research project based on research papers relating to the fabrication and wastewater pollutant removal efficiency and performance of magnetite nanoparticles (MNP) . Specifically, I am pulling a wide range of 135 parameters including residence time, crystallite size, and kinetic models. The extracted data is compiled in a Microsoft Excel spreadsheet and will be used to train categorical machine learning algorithms through an inverse design framework. This design aims to predict the most environmentally friendly and effective fabrication parameters for magnetite nanoparticles in wastewater treatment. Through this project, I hope to ultimately improve experimental efficiency, reducing both time and resource consumption in labs across the world.
Reviewed Literature
Journals
Scientific Reports, Environmental Science & Technology, Materials Chemistry and Physics, Environmental Science: Nano, Separation and Purification Technology, Toxicology Reports, Heliyon, Bioresource Technology, Methods, Chemical Physics Letters
Algorithms
Because the dataset is highly nuanced and involves many changing variables, traditional algorithmic models such as linear and logarithmic regression are less effective for this application. In turn, Random Forest, CatBoost, and XGBoost will be used to train the extracted data. They will be employed as classification algorithms to categorize fabrication conditions based on their environmental performance and pollutant removal efficiency. Together, these models analyze complex, nonlinear relationships amongst a variety of parameters, determine which variables most strongly influence outcomes, and accurately classify each parameter set into performance categories.
Findings
(Currently in progress)
Source: https://www.nature.com/articles/s41598-024-69790-w/figures/1
This example illustrates one of the many fabrication processes of magnetite nanoparticles. The particles are produced using the coprecipitation method by combining Fe2(SO4)3 and FeSO4 in their respective stoichiometric proportions in distilled water. Sodium hydroxide is then added gradually under constant stirring until the pH reaches 11.0. The mixture is maintained under continuous stirring and then heated to 80°C. After synthesis, the nanoparticles are separated, washed with distilled water until a pH of 7 was achieved. After, the nanoparticles are dried in an electric oven at 105°C.
Here's a look into some of the key parameters I've been extracting:
Residence Time
Residence time is the duration that the precursor chemicals stay in a reactor at a set temperature during synthesis. This is a key component of the fabrication process, as it directly affects the size, shape, crystallinity, and magnetic properties that are vital to the structure of the produced magnetite nanoparticle. When structure and physiochemical properties vary, so does the pollutant removal efficiency.
XRD Phase
The XRD Phase (X-ray Diffraction Phase) is another key parameter. When trying to produce magnetite nanoparticles, other phases like maghemite or hematite may form, showing impurities within the produced nanoparticle. When reading the graph, there are various things to look for. The altitude of the peaks are measured in arbitrary units and compared to standard databases to ensure that the crystal phase is really magnetite. Crystallite size can also be gauged from the XRD graph. The width of the XRD peaks have an inverse relationship with the size of the crystallite. So, broader peaks indicate smaller crystals and vice versa. For our project, smaller crystallites are favorable for pollutant removal, as smaller crystals typically have a larger surface-to-volume ratio, allowing for greater pollutant removal.
Kinetic Models
Kinetic models offer crucial insights into the distinct adsorption processes. The value of k ,given by the different ways to calculate it on the y axis, tells us how fast the reaction proceeds. While various processes happen simultaneously during the pollutant removal phase, kinetic models such as pseudo kinetics allow us to assume what is happening at at the atomic level. First order psuedo kinetics tell us adsorption rate depends linearly on pollutant concentration and typically represents physical adsorption . Second order psuedo kinetics depends on the square of the number of available adsorption sites, with chemical bonding and electron exchange being the primary processes. Beyond that lies psuedo third kinetics. If the reaction fits neither the first or second order, then it is usually classified as third order, where multiple adsorption processes occurring. Therefore, if experimental data align more closely with a particular kinetic model, it indicates which underlying processes dominate the reaction, allowing us to target and optimize those mechanisms