MatSQ Blog > Column

Research for High Strength Adhesion Materials

Viewed : 1249 times,  2021-09-23 02:19:07

This platform implements the correlation between the existing complicated formulations and their properties through machine learning.


In recent years, composite materials, lighter and stronger than steel, have been attracting attention. Their applications cover a wide range of fields, from everyday items such as bicycles, badminton clubs, golf clubs, and tennis rackets to structural materials for ships, rapid electronic railways, and aircraft. In particular, they are recognized as eco-friendly because light materials, when used for the production of vehicles, can reduce carbon emissions. In this article, let me introduce the definition of a composite material and the research methods.


What is a composite material?


First, what is a composite material? It refers to a combination of at least two materials, exhibiting better properties of matter than individual constituent materials. Composite materials largely consist of two materials: the matrix and reinforcement. The former brings together the latter while maintaining the basic shape. The latter adds strength and stiffness to the composite. As a result, composite materials feature higher mechanical strength than metals and alloys but have a density of 2 or less, lower than metals and alloys. In other words, although the volume is the same, the weight can be reduced.


The most recognizable reinforcement is a carbon fiber (CF). Carbon fibers are 5 times lighter but 10 times stronger than steel. They show excellent mechanical properties when existing alone. However, the binding capacity can be weaker when combined into a certain structure. For this reason, the matrix (polymers), such as epoxy, is needed to increase the adhesion of carbon fibers. In short, a composite material with better properties is formed in the process of combining reinforcement with the matrix.


Carbon fibers are 5 times lighter but 10 times stronger than steel.

Figure 1 carbon fiber (CF) / Source =


If used in aircraft design and production, composite materials can basically lower the load and reduce the number of parts because of decreased connections and thus manufacturing costs. The existing aircraft, made of alloys, are susceptible to corrosion, which requires humidity control. However, composite materials such as carbon fiber reinforced polymers can resolve such corrosion problems and have some advantages such as control over on-board temperature and humidity, providing a comfortable environment for passengers.


In fact, recently launched new models including the Boeing 787 Dreamliner (B787) and the Airbus A350 account for more than 50% of the total use of composite materials for aircraft. Because the successful weight lightening of aircraft ensures higher fuel efficiency, more airlines attempt to introduce those models in a rush. In addition, the defense sector has already adopted composite materials more preferentially than the private sector for the purpose of increasing load capacity through the reduced weight of aircraft such as unmanned aerial vehicles and reconnaissance aircraft.


Most of all, composite materials used in the transportation industry such as aviation require high mechanical performance specifications. It is overarching to establish a database (DB) by examining correlations between the formulations, including adhesives and reinforcement (carbon fiber and glass fiber), both of which constitute composite materials, and their properties. I have been pursuing research on the formulations on the basis of atomistic simulations and machine learning to utilize epoxy as a structural adhesive for composite materials.


Dreamliner B787, the first Boeing aircraft to use carbon composite materials for most of the aircraft

Figure 2 Dreamliner B787, the first Boeing aircraft to use carbon composite materials for most of the aircraft / Source = Wikipedia


Trends in the establishment of composite materials research data


This year, a beta version of the “ultrahigh strength adhesive material platform” was developed. This platform is intended to simply model and implement correlations between the existing complicated formulations and their properties through machine learning. It indicates how the properties (shear strength, impact strength, etc.) that influence the bond strength of materials may change depending on the different formulations such as resins, catalysts, softening agents, additives, and curing agents.


At present, this platform is loaded with a formulation model developed through machine learning based on a database shared by researchers who have long conducted research in the corresponding fields. To improve the accuracy of the adhesive formulation model, more databases should be collected from related companies, which, unfortunately, is difficult to achieve because of some issues such as security caused by the leakage of expertise. This hinders the sharing of knowledge, thus preventing the systematization of polymeric materials. In this context, I have utilized atomistic simulations such as molecular dynamics to secure more databases for machine learning.


Nonetheless, a computational simulation of polymers is not easy to model because of multiple variables, so obtaining desired data is quite challenging. Different from metals and ceramic materials with well-defined crystal structures, polymer materials feature highly complex systems because there are numerous variables to be considered in their molecular structures, along with a number of materials to be taken into account in case of the formulations. In addition, the complicated simulation procedures in conjunction with a pretreatment process require significant knowledge and expertise. Therefore, the possibility of acquiring high-quality data that can be used for machine learning depends on the presence of research personnel who specialize in polymer simulations.


As such, the application of atomistic simulations in polymer research demands high computational costs. The more efficient management of research data on polymers such as composite materials and adhesives needs both a reduction in the speed of simulation modeling and the creation of a protocol for the simulations. Probably, professional coding may be a huge barrier for materials researchers to surmount. Therefore, collaboration with machine learning engineers who have a knowledge of materials is essential so that the researchers can freely benefit from the established research data on the ultrahigh strength adhesive material platform.


enlightenedProfessor Lee's Group :



  LEE, Seung Geol | Professor, Pusan National University, Department of Organic Material Science and Engineering