IBM's 3D Printer to Revolutionize Chip Prototyping

PORTLAND, Ore. — IBM Research in Zurich today unveiled a microscopic 3D printer capable of writing nanometer resolution patterns into a soft polymer, which can subsequently be transferred to silicon, III-V (gallium arsenide -- GaAs), or graphene substrates. Unlike electron-beam (e-beam) lithography, the patterns can be both written and read for verification in real-time while the engineer watches under a microscope.

"The big difference when compared to e-beam is that you can easily write 3D patterns, which is extremely challenging for e-beams," Colin Rawlings, a scientist at IBM Research told EE Times. "The other big difference is its imaging capability -- we can read as well as write. After creating a 3D pattern we then turn off the heat to the tip and use it like an AFM [atomic force microscope] to measure with sub-nanometer resolution -- allowing us to verify our 3D patterns as well as easily locate structures beneath the polymer layer."

The microscopic 3D printer is being licensed to Zurich startup SwissLitho AG, which calls it the NanoFrazor -- a play on words between the English word razor and the German word for "milling machine," frase. The NanoFrazor, which behaves like a nanometer resolution milling machine, outperforms e-beams in many ways but costs a fraction of the price -- around $500,000, as opposed to to e-beams, which cost from $1.5 million to as much as $30 million.

"The NanoFrazor is great for rapid prototyping of all sorts of applications," Rawlings told EE Times. "It runs open loop in order to achieve scan speeds of millimeters per second and uses a specialized heated tip, mounted on a bendable cantilever, that is 700 nanometers long, but just 10 nanometers in radius at its tip."

Line width accuracy is 10 nm, but 3D depth accuracy is one nm, while reading back the measured depth of patterns has sub-nanometer accuracy. IBM hopes to be prototyping tunneling field-effect transistors (FETs) in III-V and graphene materials by the end of 2014, using a lithographic transfer technique.

"We deposit a polymer layer then a silicon or III-V layer then another polymer layer that we write into," Rawlings told us. "Then after writing we switch to a system that uniformly thins the entire polymer, resulting in holes where the pattern has been made. Then we use standard techniques to etch through the polymer and eventually into the substrate underneath, creating a mask where we can deposit materials through the holes in the pattern."

IBM is also experimenting with using its 3D printing techniques in quantum computing applications where it will create patterns to control and manipulate light on-chip in ways not possible with traditional lithography. It claims one of the unique properties of the system for quantum prototypes is that 3D patterns can be formed to guide light around smooth corners, thereby reducing light scattering problems in lightguides.

Next Page: More uses, where to see it this weekend