Manufacturing Bits: Aug. 20

Making carbon nanotubes with AI
Russia’s Skolkovo Institute of Science and Technology (Skoltech) has developed a method to monitor the growth of carbon nanotubes using an artificial intelligence (AI) technology called machine learning.

Skoltech used AI to predict the performance of the synthesis of single-walled carbon nanotubes using a chemical vapor deposition (CVD) process.

The technology could accelerate the development of logic and memory devices based on carbon nanotubes. For years, the industry has been talking about these types of devices. Carbon nanotubes are tube-shaped materials, which are 100,000 times smaller than the diameter of human hair. These structures have good electrical, chemical, thermal and mechanical properties.

For years, IBM and others have been exploring the use of carbon nanotube FETs for logic devices. Nantero has been developing carbon nanotube RAMs for memory applications. More recently, MIT, Stanford and SkyWater are developing monolithic 3D devices that stack ReRAM on top of carbon nanotube logic. ReRAM is based on the electronic switching of a resistor element.

In the fab, carbon nanotubes are formed using a deposition process. The challenge is that the nanotubes are prone to variations and misalignments during the process, making carbon nanotubes a difficult technology to put into mass production.

To address some of these issues, Skoltech has developed machine learning methods for use in making carbon nanotubes. Machine learning involves an artificial neural network (ANN). In ANNs, a system crunches data and identifies patterns. It matches certain patterns and learns which of those attributes are important.

In one example, chipmakers use machine learning techniques for defect detection in chips. For this, a chipmaker inputs different defect images into an ANN system in the fab. Then, in production, an inspection system looks at a wafer. The system compares the data of the wafer with the ANN to determine if there is a matching defect.

Skoltech, meanwhile, used different data, such as temperature, gas pressure and the flow rate, to monitor the properties of carbon nanotube films. Researchers used ANNs to process the experimental data and to predict the performance of the synthesis of carbon nanotubes.

“A major hindrance to unlocking the vast potential of nanotubes is their multiphase manufacturing process which is extremely difficult to manage. We have suggested using artificial neural networks to analyze experimental data and predict the efficiency of single-walled carbon nanotubes synthesis,” said Dmitry Krasnikov, a researcher at Skoltech.

“The development of human civilization and the advancement of the materials manufacturing and application technologies are closely interlinked in the era of information and technology, with both materials and computational algorithms and their applications shaping our day-to-day life. This is equally true for ANN that have evolved into an indispensable tool for dealing with multi-parameter tasks, which run the gamut from object recognition to medical diagnosis. Over the last 25 years, little headway was made in the development of electronics based on carbon nanotubes due to the complex nature of the nanotube growth process. We believe that our method will help create an effective carbon nanotube production framework and open new horizons for their real-life applications,” added Albert Nasibulin, a professor at Skoltech.

In 2011, Skoltech was founded by a collaboration of nine Russian institutions and organizations. In the same year, Skoltech signed an agreement with the Massachusetts Institute of Technology (MIT) and the Skolkovo Foundation to further advance R&D.

Flexible tubes
In a separate development, the Skoltech Center for Photonics and Quantum Materials (CPQM) has developed a way to tune the optoelectrical properties of single-walled carbon nanotubes (SWCNT)—a technology that could pave the way for foldable LCDs.

In the lab, researchers applied an aerosolized dopant solution on the surface of the carbon nanotubes. “This method is based on aerosolization of a dopant solution (HAuCl4 in ethanol) and time-controlled deposition of uniform aerosol particles on the nanotube film surface,” according to researchers in The Journal of Physical Chemistry Letters.

This in turn enables better transparent and flexible conductive films (TCF) made using SWCNT. “Our method allows easy tuning of SWCNT film parameters thanks to time-controlled deposition of doping aerosol particles,” said Skoltech PhD student Alexey Tsapenko.

Printing CNTs
Using a 3D printer, the University of Delaware has developed various structures based on carbon nanotubes.

Researchers printed tiny replicas of the Eiffel Tower and the Great Wall of China, a small pig’s face, and a honeycomb. Over time, the technology could be used to print materials for batteries and electronics, water purification and desalination technologies, medical implants and other structures.