Argonne scientists' approach to molecular modeling can accelerate the development of new electronic materials.
Organic electronics allow revolutionary technology with their high cost effectiveness and versatility, compared to the most common non-organic electronics. For example, flexibility can be applied to companies' printing on paper or to electronic garment clothing. However, they do not get a lot of industrial traction because of the difficulty in controlling the electronic structure.
To face this challenge, a member of the Department of Energy of the United States of America, Maria Goeppert Mayer of the Department of National Laboratory (DOE), has developed a faster way to create molecular models through automatic studies. Jackson's models accelerate electronic screening of potential new material materials and can also be useful in other materials research fields.
The internal structure of an organic material affects electrical efficiency. The current processes used to produce these materials are sensitive and the structures are very complex. Scientists find it difficult to predict the latest structure and efficiency of the material in manufacturing conditions. Jackson uses machine learning, the computer is a way of learning to learn a model explicitly to help them make predictions.
Jackson's research focuses on steam deposition as an instrument for assembling organic materials. In this process, scientists evaporate an organic molecule and condense slowly on a surface, producing a film. By manipulating some deposition conditions, scientists synthetically adjust how molecules work in the film.
"It's like a game of Tetris," said Jackson. "Molecules can be oriented in different ways and our aim is to determine how this structure affects the electronic properties of the material."
The film's molecular packaging affects the load mobility of the material, the measurement of the internal load loads. Load mobility plays a role as a material efficiency device. To analyze this relationship and optimize the performance of the device, Jackson's team performs very accurate computer simulations in the vapor deposition process.
"We simulate the magnitude and nanoscopic scale that simulates the behavior of electrons around each molecule," but these models have intensive computation, so they need a long time to run. "
In order to simulate the packaging of complete mechanisms, often with millions of molecules, scientists need to develop cheaper and more flexible complex calculations, which describe the behavior of electrons individually in molecular groups. Buy these models can reduce their spreading time and minute, but the challenge facilitates the prediction of physical models results. Jackson uses algorithms for his machine learning to show the relationship between the exact and thick pattern.
"I dropped my hands and learned the machine to delay the relationship between the serious description and the electronic properties of my system," said Jackson.
Using an artificial neural network and a learning process called backpropagation, he learns to extrapolate machine learning algorithms from bold and more precise models. Using the complex relationship that it finds between models, it predicts the same electronic properties of the material using the model that it envisages itself.
"We are developing cheaper models that play expensive property," said Jackson.
As a result, rugged model scientists can see before the arrangements for anchoring orders of two or three magnitudes. The results of the study of the serious model, then, the experimentalists contribute to the development of high-performance materials.
Shortly after, Jackson began to designate the Chicago University professor and Argonne scientist Juan de Pablo, who accelerated his research machine. He then took advantage of high-performance computing capabilities in the laboratory, in collaboration with Venkatram Vishwanath, Data Sciences and Workflows Team Lead with Argonne Leadership Computing Facility, DOE Office of Science User Facility.
Materials scientists have learned beforehand to find relationships between molecular structure and device performance, but Jackson's only approach is to improve interaction between different lengths and scale models.
"In the physics community, researchers are trying to understand and understand the properties of a system to reduce levels of freedom to simplify levels," said Jackson.
The main objective of this research is to study the electronic accumulated vapors, but it is applicable in many types of polymer research, even in protein science. "He added anything that you're trying to do to make interpolation between thick and bold models.
In addition to broader applications, Jackson's advancements will boost organic electronics for industry's importance.