Caltech camera chip yields superfine 3-D resolution

Imagine you need to have an almost exact copy of an object. Now imagine that you can just pull out your smartphone, take a snapshot with its integrated 3-D imager, send it to your 3-D printer, and within minutes you have reproduced a replica accurate to within microns of the original object. This feat may soon be possible because of a new, tiny high-resolution 3-D imager developed in the laboratory of Ali Hajimiri, Professor of Electrical Engineering at Caltech. This design is described in the February 2015 issue of Optics Express.

A cheap, compact yet highly accurate new imaging device known as a nanophotonic coherent imager (NCI), promises to produce a high-resolution scan. Each pixel in an image created by the NCI provides both the distance and intensity. “Each pixel on the chip is an independent interferometer which detects the phase and frequency of the signal in addition to the intensity,” says Hajimiri. 

Three dimensional map of the hills and valleys on a U.S. penny.  Photo: Ali Hajimiri/Caltech

The new chip uses LIDAR, where a target object is illuminated with scanning laser beams. This illumination is with coherent light. The chip uses grating couplers that serve as “pixels”. On the NCI chip, the phase, frequency, and intensity of the reflected light from different points on the object are detected and used to determine the exact distance of the target point.

Because the coherent light has a consistent wavelength, it is used as a reference with which to measure the differences in the reflected light. The incorporation of coherent light not only allows 3-D imaging with the highest level of depth-measurement accuracy ever achieved in silicon photonics, and lets the device be very small. “By coupling, confining, and processing the reflected light in small pipes on a silicon chip, we were able to scale each LIDAR element down to just a couple of hundred microns – small enough that we can form an array of 16 of these coherent detectors on an active area of 300 microns by 300 microns,” Hajimiri says.

The proof of concept of the NCI has only 16 coherent pixels. However, the researchers developed a method to image a four-pixel-by-four-pixel section, then move the object in four-pixel increments to image the next section. With this method, the team used the device to scan and create a 3-D image of the “hills and valleys” on the front face of a U.S. penny—with micron-level resolution—from half a meter away.

In addition to Hajimiri, other Caltech co-authors include former postdoctoral scholar Firooz Aflatouni, graduate student Behrooz Abiri, and Angad Rekhi (BS '14). www.caltech.edu