Park Systems, a world-leading manufacturer of atomic force microscopy (AFM) systems, has integrated robotics and machine learning into its latest atomic force microscope. The AFM features the first-ever fully automated imaging process for research users.
Image Credit: Elizaveta Galitckaia/Shutterstock.com
What is Atomic Force Microscopy?
Atomic Force Microscopy (AFM) has been a backbone of nanoscale materials research for many years, with scientists and engineers utilizing the technique to view materials almost atom by atom. It is a powerful and versatile technique that was invented in 1985 by scientists at IBM and Stanford University.
The technique is favored by researchers because it offers a 3D view of the sample’s surface at sub-nm resolution. Different measurement modalities have also enabled researchers to study material phenomena, including frictional, nanomechanical, electrical, magnetic, and thermal properties.
AFM works with any type of material in ambient conditions, as well as under vacuum and in liquid. It has become a vital tool in all areas of science and technology in the last few decades.
Difficulties with Atomic Force Microscopy
AFM gets its data by scanning a cantilever with an ultra-sharp tip across the sample’s surface. An optical system monitors tiny deflections of the cantilever caused by atomic forces that operate close to the surface.
Users must fit the correct tip, often with tweezers, and then position it close enough to the sample for the cantilever to interact with the surface. Often, this highly delicate procedure is performed by hand.
After this, the optical detection system and viewing optics need to be correctly aligned, and several different imaging parameters must be chosen or tuned.
All of this means that considerable time, patience, and skill are needed for reliable AFM analysis. This has prevented the technique’s wide-scale adoption, as it is still seen as inaccessibly complex to operate and handle. AFM that works more robustly and can be operated with ease would enable more adoption for users’ daily measurement needs.
Park Systems was already working on these complications and has developed solutions to make AFMs easier to use such as pre-mounted tips and a point-and-click software interface in recent years.
Park Systems’ New Fully Automatic Approach
The company’s latest instrument, the Park FX40, brings previously unseen levels of automation and ease of use to research. Robotics and machine learning are integrated throughout the product’s design, resulting in the world’s first fully automated imaging process in an AFM suitable for research.
The Park FX40 automatically completes set up before and during scanning, removing the tedious and time-consuming manual processes required in typical research AFM.
How Does the Park FX40 Work?
The new system automates tip exchange and beam alignment procedures, significantly improving ease of use.
A robotic system is used to automatically change and replace the device’s own tips, with the capacity for up to eight different cantilevers on the cassette. A QR code on each cantilever enables easy identification, and a magnetic mechanism attaches the tip to the AFM head.
Beam alignment is also automated. An optical system is used to locate the tip, ensuring that the laser beam reflects in the right direction from the cantilever and onto the optical detector.
Two Cameras are Better than One
A unique dual-camera design enables the Park FX40 to locate a region of interest with relative ease. The first camera provides an overview of the sample stage, with the ability to monitor four samples at once. This enables users to quickly find and select the target location.
When the stage moves to the selected position, the system automatically switches cameras to use a high-resolution camera for a close view, allowing for any final adjustments to be carried out.
Automatic Fine Tuning
More automation takes place after the system is in the right location. The probe is automatically faced up to the sample’s surface to start measuring. An optical system guides the tip to within a few microns of the surface. Then, mechanical oscillations in the cantilever are measured and input into a machine learning feedback mechanism.
With this feedback, the probe’s approach to the sample surface can be automatically fine-tuned. The instrument adapts to the type of surface under inspection without needing any human operators.
This fine-tuning continues once the scan has started. The system completes a few global overview scans to determine the material’s overall topography, setting up initial feedback parameters with this data. During scanning, the system plans the next lines of the scan, adjusting the feedback parameters to accommodate rapid changes in surface topography.
The result of this automatic fine-tuning is improved scanning times, as the system can adapt scanning speed to work on rougher features slowly and flatter areas more quickly.
Assisting Research with Cutting-Edge AFM
These innovations mean that users do not have to handle or understand the complex inner workings of the AFM to get good quality data from it. This benefits researchers who often need to gather data from a variety of different instruments.
Enabling faster sample analysis will also be key for many future scientific advances. This means that experimentation and iteration can happen faster, often with automated processes guiding the research.
Artificial intelligence (AI) systems are the next step for AFM research and would continue Park Systems’ approach to automation. These kinds of advances can make AFM a much less tedious, laborious, and time-consuming process. This means that more data can be gathered, leading to more scientific breakthroughs in the future.
References and Further Reading
Automated imaging makes AFM experiments faster and easier (2021) PhysicsWorld. [Online] Available at: https://physicsworld.com/a/automated-imaging-makes-afm-experiments-faster-and-easier/.
Huang, B., Z. Li, and J. Li (2018) An artificial intelligence atomic force microscope enabled by machine learning. Nanoscale. https://doi.org/10.1039/C8NR06734A.