Catalytic reduction of organic pollutants using supported metal nanoparticles

Metal nanoparticle catalysts have in the last decades been extensively researched for their enhanced performance compared to their bulk counterpart. Properties of nanoparticles can be controlled by modifying their size and shape as well as adding a support and stabilizing agent. In this study, preformed colloidal gold nanoparticles supported on activated carbon were tested on the reduction of 4-nitrophenol by NaBH4, a model reaction for evaluating catalytic activity of metal nanoparticles and one with high significance in the remediation of industrial wastewaters. Methods of wastewater remediation are reviewed, with case studies from literature on two major reactions, ozonation and reduction, displaying the synergistic effects observed with bimetallic and trimetallic catalysts, as well as the effects of differences in metal and support. Several methods of preparation of nanoparticles are discussed, in particular, the sol immobilization technique, which was used to prepare the supported nanoparticles in this study. Different characterization techniques used in this study to evaluate the materials and spectroscopic techniques to analyze catalytic activities of the catalyst are reviewed: ultraviolet-visible (UV-Vis) spectroscopy, dynamic light scattering (DLS) analysis, X-ray diffraction (XRD) analysis and transmission electron microscopy (TEM) imaging. Optimization of catalytic parameters was carried out through modifications in the reaction setup. The effects of the molar ratio of reactants, stirring, type and amount of stabilizing agent are explored. Another important factor of an effective catalyst is its reusability and long-term stability, which was examined with suggestions for further studies. Lastly, a biochar support was newly tested for its potential as a replacement for activated carbon.

A deep learning model via thermodynamics and condensed matter physics

Deep Learning architectures give brilliant results in a large variety of fields, but a comprehensive theoretical description of their inner functioning is still lacking. In this work, we try to understand the behavior of neural networks by modelling in the frameworks of Thermodynamics and Condensed Matter Physics. We approach neural networks as in a real laboratory and we measure the frequency spectrum and the entropy of the weights of the trained model. The stochasticity of the training occupies a central role in the dynamics of the weights and makes it difficult to assimilate neural networks to simple physical systems. However, the analogy with Thermodynamics and the introduction of a well defined temperature leads us to an interesting result: if we eliminate from a CNN the "hottest" filters, the performance of the model remains the same, whereas, if we eliminate the "coldest" ones, the performance gets drastically worst. This result could be exploited in the realization of a training loop which eliminates the filters that do not contribute to loss reduction. In this way, the computational cost of the training will be lightened and more importantly this would be done by following a physical model. In any case, beside important practical applications, our analysis proves that a new and improved modeling of Deep Learning systems can pave the way to new and more efficient algorithms.