The School of Natural and Applied Sciences, Maulana Abul Kalam Azad University of Technology (MAKAUT), West Bengal, India organized a Webinar on "Materials Informatics: The Emerging Paradigm of Material Design" on 16.07.2020. The eminent speakers were: Hon'ble Vice Chancellor of MAKAUT Prof. (Dr.) Saikat Maitra, Prof. (Dr) Sukhendu Samajdar, Director, School of Natural and Applied Sciences, MAKAUT, Prof. (Dr) Subhabrata Datta, Research Professor, SRM University, Eng. K. P. Das, Hony. Secretary, Institute of Engineers, India and Ms. Monali Char, Assistant Professor, MAKAUT, WB, coordinated the session.

Prof. (Dr) Sukhendu Samajdar expressed his sincere thanks to the Vice Chancellor, distinguinshed panelists and respected participants. He acknowledged that under the inspiring leadership of MAKAUT's Hon'ble Vice Chancellor it has been possible to encourage and educate the student community. He also said that MAKAUT cares for 150000 students through its in-house courses and 200 affiliated colleges. He then gave a brief but insightful account of materials informatucs. He said, " We nees knowledge to tailor materials for new technologies and to do it Computational Material Science is needed, whose goal is to accelerate the design of the materials. There are many established tools like Density Function Theory, Phase Field , Finite Element Analysis, Computational Thermodynamics, etc. But the emerging field is the Data Driven Models.

We need to have an idea of the things around us, what they are made up of and what else they can be made up with. The interconnection of knowledge from various fields is very important. Links with other disciplines will help to integrate information and realize the importance of Material Science. The inception of Materials Informatics is from the 1990s. A pioneering example in this field is the Ashby Plot designed by Michael Ashby from University of Cambridge. Ashby Plot compressed a lot of plots required for engineering design, that is a lot of useful information is condensed to cater the

needs of engineering materials. Subsequently in 2008 The National Research Council articulated the idea of Integrated Computational materials engineering (ICME), which is the integration of materials information captured in computational tools with engineering product performance analysus and manufacturing-process simulation. The developed committee curated data sets, structure-property models, processing-structure relationships, physical properties, and thermodynamic, kinetic and structural information. In 2011 the

Obama Government in The United States launched the Materials Genome Initiative, with the objective of squeezing the time span of a material from its development to deployment like an accordian. In this context the 3 pillars were: Computatuional tools, Experimental tools and Digital Data running in harmony. We have huge data from variety of sources such as simulation, experiment etc. This data is processed and then made available. for this process there are two approaches to analyze data

• Hard Modeling which is based on statistical learning.
•  Soft Modeling which is based on constitutive equations

To summarize it can be said that materals scientists are constantly striving to advance their ability to understand, predict and improve material properties. Due to high cost of traditional trial and error methods in isolation, material scientists are increasingly relying on simulationa and modeling methods. Materials informatics is a resultng branch of materal science that utilizes high throughput computation to analyze large databases of materials properties to gain unique insights and discover materials in real time. More recenty data driven methods such as machine learning havs been adopted in this field to study the wealth of existing experimental and computational data, leading to a paradigm shift in the way naterial science research is conducted."

Prof. (Dr) Subhabrata Datta expressed his sincere gratitude to MAKAUT family for providing the

opportunity to share his thoughts in this domain of materials informatics. He said, " There are several important properties of materials. The different properties are required for different materials to be engineered. For a particular application more than one property is required to make the material perfect. The performance depends on a combination of properties and the best material is selected by maximizing one or more 'performance indices'. As civilization progressed, the materials being used also changed. In the early ages of civilization metals and alloys were extensively used. Now with the dixcovery of new materials their uses are decreasing day by day.

• Composition
• Processing

This will give rise to the desired properties, which in turn will help the material to perform better.

Again, for designing a material two approaches are followed:

• Conventional approach: This follows the experimental methods by persuing trial-and-error.. This is followed for everal century but it requires a lot of time
• Computational Approach: After the advent of computers we have certain tools with which we can model different properties and design the material computationally and then go for experimental trial to do validation. This saves a lot of time and money and has developed a lot in the last few decades.

In the physics driven approach all the tools simulate the microstructures, after which the properties can be mapped using different methods like property prediction.

In the Data Driven approach only the data is used. Lots of statistical and AI based methods are used to develop correlation from the data which are used to design materials. This approach is called materials informatics.

Materials are combined to form new alloys, composites to achieve particular properties. It the properties are achieved we call it adequate design. This can be achieved by more than one route. Next an optimization technique is applied to this adequate design for getting optimized design, so that we can choose the best one from a number of models. 

Data Analytics leads to materials informatics which in turn leads to design. It has 4 different part and 4 purposes.

• Descriptive Analytics: Visualization of data by various methods.
• Diagnostic Analytics: Seeks the reason behind the behaviour of the material. Data clusturing and feature extraction techniques such as principal component analysis, support vector machine, fuzzy clusteing, rough set etc are used.
• Predictive Analysis: Does model prediction with the help of correlation/prediction tools like linear regression, polynomial regression, PLS, artificial neural network, fuzzy inference system, genetic programming etc.
• Prescriptive Analysis: It prescribes a material which can be perfect for a particular purpose with the help of design and optimization tools like genetic algorithm, multi-objective optimization, multi-criteria decision making etc.

There may be many attributes of a material that may not be revealed, which may lead to uncertainty and imprecessions. This is why data driven models  are required. But it cannot be a competitor to physics based models. It can only support the process of material discovery.

The data can be obtained from :

• Manual Record keeping system
• Data Warehouse
• Published Data
• Experimental Data

The experimantal equipments have improved a lot in the past few years. So generation of data is nit that difficult, if the facility is available. For example, Atom probe tomography(APT) is a new technique. It can map atom in a 3D space for a small specimen like a tip of a needle and generate few terabytes of data.

A few applications based on data-driven approach are: design of nano-bainitic steel, Aluminium Alloys, design of Ti alloys and composite for orthopedic and dental implants, design biodegradable hybrid nano-lubricant etc."

For more useful information visit https://makautwb.ac.in