Japanese chemists say they have captured video of single molecules in motion at 1600 frames per second – 100 times faster than previous experiments of this nature.

The difficulty with promoting this achievement is that, as the video sample below attests, it’s difficult for the untrained eye to actually make out the molecules in motion.

Credit: CC-0

But the team led by Eiichi Nakamura from the University of Tokyo is confident that what it has accomplished by combining a powerful electron microscope with a highly sensitive camera and advanced image processing will aid many areas of nanoscale research.

If video is captured at high fps (frames per second) then shown at lower fps, the smooth slowing of motion allows you to perceive otherwise inaccessible details.

Special microscopes and cameras now allow researchers to capture atomic-scale events at about 16 fps, Nakamura says, but he and colleagues have upped the ante.

The key was dealing with the visual “noise”.

“To capture high fps, you need an imaging sensor with high sensitivity, and greater sensitivity brings with it a high degree of visual noise. This is an unavoidable fact of electronic engineering,” says Koji Harano, co-author of a paper in the Bulletin of the Chemical Society of Japan.

“To compensate for this noise and achieve greater clarity, we used an image-processing technique called Chambolle total variation denoising. You may not realise, but you have probably seen this algorithm in action as it is widely used to improve image quality of web videos.”

The researchers tested their setup by imaging vibrating carbon nanotubes which housed fullerene (C60) molecules resembling faceted soccer balls made from carbon atoms. The imaging setup captured some mechanical behaviour never seen before on the nanoscale.

Like a pebble in a shaken maraca, they say, the oscillating motion of the C60 molecule is coupled with the oscillation of the carbon nanotube container. This is only visible at high frame rates.

“We were pleasantly surprised that this denoising and image processing revealed the unseen motion of fullerene molecules,” said Harano.

“However, we still have a serious problem in that the processing takes place after the video is captured; this means the visual feedback from the experiment under the microscope is not yet real-time.

“But with high-performance computation this might be possible before too long.”