A startup founded by an array of experts from Harvard, Yale, University of Maryland, Cadence, and Texas Instruments has come out of stealth to present a breakthrough device for the non-invasive, non-destructive 3D imaging of semiconductors and batteries.
EuQlid’s Qu-MRI platform combines diamond-based quantum sensing, advanced signal processing, and machine learning to scan chips non-destructively and quickly uncover hidden defects, even on the production line. EuQlid’s innovation could “save chip foundries billions of dollars,” the company explained to Tom’s Hardware by way of an emailed press release.
Qu-MRI technology
EuQlid describes its new scanner as a device leveraging quantum magnetometry hardware to provide “buried electrical current maps with high-throughput and nano-amp sensitivity.”
The Qu-MRI relies on EuQlid’s first-gen product, the Quantum Diamond Microscope (QDM), augmented by advanced signal processing and AI-driven analytics to make sense of the complex semiconductor materials and devices this new product is targeting.
"EuQlid's Qu-MRI uses Quantum NV-diamond magnetometry to deliver buried electrical current maps." (Image credit: EuQlid
Potential impacts on industry
EuQlid’s new Qu-MRI offers compelling advantages over established chip-scanning tools. Rival methodologies may offer limited sub-surface imaging abilities, require physical contact with the chip, or deliver limited resolutions. Moreover, some investigation techniques require (time-consuming and destructive) grinding or slicing through chip layers for a full analysis.
Thus, contemporary destructive styles of chip inspection could be consigned to history by Qu-MRI. Moreover, EuQlid claims its solution is speedy and can enable earlier detection of buried defects during design validation, and it can even be deployed for inline production-time inspection for 3D ICs.
"EuQlid's technology provides customers with powerful insights, from spatial analysis of state-dependent power flows in functioning CPUs and GPUs, to the detection and localization of interconnect stacking errors in HBMs." (Image credit: EuQlid
... continue reading