Effective desalination of heavy metal ions from industrial effluents is a challenge mainly due to the existing methods of separation technologies that are energy-intensive, have poor economics of scale, and generate a large amount of sludge. The application of gas hydrate-based technology for the desalination of heavy metals is a promising approach because it generates no sludge and is a relatively green process. In a hydrate-based desalination approach, suitable hydrate-forming guests, a sII hydrate former, interact with water by weak van der Waals forces to produce solid hydrate crystals by excluding the salts and other impurities from an aqueous heavy metal ions solution. As5+, Pb2+, Cd2+, and Cr3+ are common heavy metal ions found in industrial effluents that were individually chosen to prepare a 1000 ppm salt solution. In this work, natural gas was used as the hydrate-forming gas along with cyclopentane (CP) because of its immiscibility in water. The presence of CP also reduces the operating conditions for hydrate formation. CP was used at two different concentrations (6 and 1 mol %), and the kinetics of hydrate formation was further improved by the addition of edible surfactant lecithin to the hydrate-forming solution. The gas uptake kinetics, water to hydrate conversion, and rate of water recovery were studied. Superior kinetics of hydrate growth were observed with 6 mol % CP compared to 1 mol % CP. Also, the addition of a benign additive, lecithin, enhances the kinetics of hydrate formation, resulting in efficient desalination of salt ions. The kinetics of As5+ desalination was the fastest among those of the four selected metal ions.

No abstract

Deep learning (DL) method consisting of Convolutional Neural Network (CNN) was employed to automate the task of microstructural recognition and classification to identify dendritic characteristics in metallic microstructures. The dendrites are an important feature which decide the mechanical properties of an alloy, further the dendritic arm spacing is critical in ascertaining the values of strength and ductility. The current work has been divided into two tasks i.e., classification of microstructures into dendritic and non-dendritic (Task 1) and further classifying the dendritic microstructures into longitudinal and cross-sectional view (Task 2). The data set comprising of micrographs from experimental and online sources covering a broad range of alloy compositions, micrograph magnifications and orientations. The tasks were achieved by employing a 4 layered CNN to yield an accuracy of 97.17±0.64% for Task 1 and 87.86±1.07% for Task 2 independently. The employment of deep learning model for classification of microstructures circumvents the feature extraction step while ensuring high accuracy. This work reduces dependency on skilled and experienced researchers and expedites the material development cycle.

No abstract