Post-Doctoral researcher
이 후보자에게 직접 채용 제안 보내기
As a postdoctoral research associate, I took on the task of data driven prognostics of lithium-ion batteries. Having quick learning abilities, I was able to acquire the needed skills and expertise in various machine learning methods for state estimations (SOH & SOC) and remaining useful life (RUL) prediction of lithium-ion batteries. Data driven Prognostics and Health
Management (PHM) concepts for early state estimations based on sensor data which enables setting up of timely maintenance and replacement schedules were developed. Further, these machine learning models can be used for efficient sorting of spent batteries for echelon utilization and recycling. Machine learning model development involves data preprocessing for feature extractions, data set preparation for training and testing of different models like ANN, GPR, SVM and LSTM.
The accuracy of the models is evaluated based on error values like root mean square error (RMSE), mean absolute deviation (MAD), mean absolute error (MSE) etc.
Having completed my doctoral degree from the Department of Organic Material Science and Engineering, Pusan National University with research focused in renewable energy storage systems (LIBs and NIBs), my primary interest is in finding new alloy electrodes, electrolytes with additives and novel binders and binder mixtures which facilitate efficient energy storage systems which are sustainable and renewable. Issues like low energy densities, huge volume changes and short cycle life are some of the hurdles that need to be addressed. Various approaches are being suggested and studied to mitigate these drawbacks, the most studied and viable approaches are,
bimetallic alloy anodes, incorporation of an intermetallic component provides a buffer phase to accommodate the volume changes. Addition of carbon (TiC, acetylene black) as a conductive support matrix is another efficient approach where the carbon coats the active material surface providing conductive pathways for the Li+/Na+ while providing buffer to accommodate the volume change. FEC (fluoroethylene carbonate) as an electrolyte additive is an effective way to passivate the electrode surface by forming a stable and thin SEI (solid electrolyte interface) layer,
improving the stability and cyclability of the cells. Binders are a crucial component of batteries which keep the electrode from disintegrating and also keep the electrode material adhered to the current collector while accommodating the volume expansions. Functional binders with functional groups form a weak covalent bond with the surface –OH groups of active materials and current collector, keeping the electrode intact and adhered to the current collector.
During Postdoctoral research, various public datasets from NASA PCoE, Stanford, Zenodo etc. were studied and the data from these datasets was preprocessed for training and testing of machine learning models. Feature extraction techniques like PCA, LDA and feature selection techniques like RF, and other filter techniques are used to extract health indicators. Gaussian regression, support vector machines, neural networks and deep learning methods were trained,
tested and the results compared. The effects of data diversity, and size of the training dataset set on the prediction accuracy of the models was studied. Various performance metrics like RMSE,
MSE, MAD and MAPE are used to compare the models predictive capabilities while performing state estimations (SOC, SOH) and RUL predictions.
During my doctoral research, extensive experimental work was conducted with the aim to identify suitable components for NIBs like active material, electrolyte additive, conductive buffer matrix and binders and also to formulate different approaches to enhance the electrochemical performance of NIBs. Through my research I was able to identify the synergistic effect of micro components of battery, ascertain the prominent role of binder in attaining extended cyclabilities and formulate a novel hybrid binder mixture which enhanced the rate capability and cyclability of bimetallic anodes in sodium ion batteries.
Firstly, Sb2Te3 (antimony telluride) bimetallic compound was synthesized with 30 wt% of
TiC (titanium carbide) as the conductive additive and electrochemically evaluated as an anode in
NIBs with PVDF (polyvinylidene fluoride) or PAA (polyacrylic acid) as binders and 0, 2 and 5%
FEC added electrolyte. While the cells with no FEC exhibited unstable kinetics with spikes in the charge plots, the cells with 2% FEC added electrolyte exhibit 363 mAh g-1 after 80 cycles, the cells with 5% FEC exhibit a reversible capacity of 358 mAh g-1 with a capacity retention of 95%
and coulombic efficiency of 98% after 160 cycles. Performance enhancing effect of bimetallic anodes, passivating SEI film forming ability of FEC electrolyte additive, conductive TiC buffer matrix and PAA binder were identified and the synergistic effect of all the battery components in enhancing the stability, capacity and cyclability of the cell was evidently observed.
Secondly, to further identify the optimum combinations of the various battery components like percentage of carbon content and type of binder, SbTe bimetallic compounds were synthesized with 20, 30 and 40% carbon (acetylene black) by HEMM (high energy machine milling) and electrochemically tested for NIBs. While the cells with PVDF binder failed, the cells with PAA as binder and 5% FEC added electrolyte exhibited reversible capacities of 421, 371 and 325 mAh g-
1 for SbTe-C20, SbTe-C30 and SbTe-C40, respectively, retaining 96.0%, 92.9% and 90.0% of their respective capacities after 200 cycles. The role of carbon support matrix in enhancing the cyclability, stability and capacity retentions in NIBs is highly significant. However, considering the tradeoff between capacity and cyclability associated with the increase in carbon content, the role of binder is much more prominent.
Finally, a novel hybrid binder mixture was tailor made to address the performance issues associated with NIBs. PVDF is the binder of choice for LIBs but fails to perform with NIBs due to its de-fluorination mechanism with Na+ and forming of NaF. However, it is observed that the cells with PVDF binder exhibit higher initial capacities than the cells with PAA as binder. Also,
FEC is known to undergo similar decomposition to form NaF, a favorable component of the SEI layer. Keeping these two points in mind, novel hybrid binder mixtures were prepared with different weight ratios of 9:1 (Ver (9:1)), 7:3 (Ver (7:3)) and 5:5 (Ver (5:5)) for PAA: PVDF. The novel hybrid binder mixtures when tested with SbTe bimetallic anodes and 5% FEC added electrolyte exhibit high initial capacities of 364 mAh g-1 and 393 mAh g-1 for Ver (7:3) and Ver (5:5),
respectively, with capacity retentions of 94.7% and 91.3% after 500 cycles at c-rate of 1000 mA g-1.
This approach of enhancing the electrochemical performance of NIBs by hybrid binder mixtures where PVDF binder provides the NaF (de-fluorination) needed for formation of a stable
SEI layer, thereby reducing the FEC consumption/ decomposition and exploring the synergistic effects of various components of a battery in conjunction with binder modifications to sustain key reactions in the cell, can pave the way for more sustainable energy storage solutions in the near future.
