Utilize machine learning in designing soft materials

(Nanowerk Highlights) Soft materials such as polymers, rubber and hydrogels play an important role in your daily life. From your car tires to the elastic in your favorite loungewear, from the latest flexible electronics to wearable technology like electronic tattoos — these versatile materials are everywhere. However, to use these materials effectively in a variety of applications, their mechanical properties need to be suitably adapted.

For example, wearable sensors such as electronic skins and tattoos need to be gentle enough — in scientific terms, they require a relatively low Young’s modulus — to match our skin’s natural tension. They also need moderate flexibility to move with our bodies without causing discomfort. When used in soft robotics, these materials require various properties for biocompatibility or to mimic biological designs. Additionally, when used on artificial leather, the material must be strong, durable and tough.

The trick to achieving these unique properties lies in carefully manipulating aspects such as the polymer chain, monomer composition, and intermolecular hydrogen bonds. However, sewing these materials is not trivial. This requires a deep understanding of materials chemistry and some experimental experiments. This process can be a significant hurdle for end users who require materials with specific mechanical properties for their applications. Therefore, this new approach to the design process is very important. Soft material design system and design process demonstration. a) The soft material design process uses a design system. The system takes the desired mechanical properties as input and returns the experimental conditions in the form (X1, X2, and X3). b) Examples of elastomers with high Young’s modulus (b-1) and elastomers with high strain at break (ie high tensile strength) (b-2). (Reprinted with permission by Wiley-VCH Verlag)

The Rise of Machine Learning in Material Design

This is where the power of machine learning and materials informatics comes into play. These advances have significantly accelerated the materials discovery process. Machine learning algorithms can pick up on subtle patterns in data sets that would be difficult to identify through human intuition alone. This capability enables reverse material design. That means using a set of desired material properties to determine the experimental parameters, which significantly speeds up the design process.

However, using machine learning models in experimental studies has its challenges. Collecting the large amounts of high-quality experimental data needed for model training can be time-consuming and labor intensive. Fortunately, innovative strategies have emerged to collect high-quality data with minimal effort, such as utilizing archived laboratory notebooks or applying experimental design techniques such as experimental design (DoE).

Data Driven Approach to Soft Material Design

One of the most interesting developments in the field is the application of a data-driven approach to adjusting the mechanical properties of soft materials. In a recent study on Advanced Functional Materials(“A Data-Based Approach to Adjusting the Mechanical Properties of Soft Materials”), researchers from Stanford University demonstrated this approach using polyurethane (PU) elastomers, a common type of soft material.

The team adjusted the mechanical properties of the PU elastomer by changing the mixing ratio of its components. They collected data on the mechanical properties of materials, such as Young’s modulus, strain at break, maximum strength, and toughness. Using this data, they trained a machine learning model to predict this property based on the blending ratio. Five samples were prepared for experimental verification. The mechanical properties of interest are labeled as “Predicted” in the graph, and the mechanical properties measured are presented: Young’s modulus c), strain at break d), maximum strength e), and toughness f). The “measured” data point error bars show a standard deviation of 1 calculated from 3 measured data. (Reprinted with permission by Wiley-VCH Verlag)

The beauty of this method is that it can do a ‘reverse design’. You input the mechanical properties you want, and the model outputs the synthetic recipe for achieving those properties. The researchers tested this by preparing elastomeric samples using this recipe and found that the resulting mechanical properties closely matched those of the input.

The researchers concluded that this data-driven approach to soft material design, leveraging machine learning, can accurately predict and adapt the mechanical properties of these materials to a very small data set. By focusing on macroscopic structural information controlled by synthetic recipes, this approach can provide soft materials with properties close to those desired.

The success of this research may prompt further discussion between materials research and the artificial intelligence research community. It could also spur the development of new algorithms specifically designed for small data sets, a common challenge in the field. By leveraging data-driven approaches and machine learning, we can more efficiently explore various soft material systems and design processes, bringing us one step closer to laboratory automation.

By

Michael
Berger

– Michael is the author of three books by the Royal Society of Chemistry:
Nano-Society: Pushing the Boundaries of Technology,
Nanotechnology: A Small Future And
Nanoengineering: Skills and Tools for Making Technology Invisible
Copyright ©




Nanowerk LLC