SEMICONDUCTOR INSPECTION
g
frame rates, managing bandwidth and data transfer are becoming important areas of growth. In device inspection, for the most efficient defect detection, cameras need high frame rates and high imaging resolution. Tis creates a huge amount of data. Te challenge is then getting this data into a processing computer or GPU, requiring a high bandwidth interface. “Te big question is ‘how should you
process all this incoming data?’, ” says Alexis Teissie, Senior Product Manager of Lucid Vision Labs. “On the interfacing side, we are looking at a global view of not just pushing everything to the central CPU, but trying to work closely with our customers to scale with the technology.” Teissie says it is about leveraging the
network standard used in high-performance computing clusters, but applying it to machine vision. “And making sure that we can reliably transfer 10Gigabits from multiple cameras with no CPU
‘CEP camera technology is most probably at its peak…it’s all in the software now’
involvement,” he continues. “Because we are working with high-frame-rate cameras and we need the highest reliability – and so, high bandwidth – we’re developing special network transfer technology: a Remote DMA technology.” Remote DMA (RDMA) is a technology
that allows networked computers to exchange information without relying on the operating systems or processors of the computers themselves. Tis technology allows for low latency in the data transfer process, meaning that data packets do not take long to be stored or retrieved. Using
SWIR Vision Systems’ Acuros camera, above, and Emberion’s VS20 Vis-SWIR, right, are both based on sensors utilising quantum dot technology
RDMA, then, data can be transferred quickly without any burden to the CPU of the processors involved. “Tis is a building block to make sure we
have reliable transfer so that we can ensure that whenever a frame is required, it will be delivered with certainty,” says Teissie. “Tere is this move to go to higher speeds
and just getting higher bandwidths – this is just the first step,” Teissie adds. “25GigE is coming.” Lucid’s upcoming Atlas10 camera with RDMA is aimed to be released by the end of Q2 2023, and it notes that “one of the target markets for this is definitely semiconductor wafer inspection”.
Quantum vision One limiting factor surrounding SWIR cameras based on InGaAs sensors is their price point, which can stretch to thousands of euros per sensor chip because of their fabrication process. SWIR sensors utilising quantum dots (QDs) have, in the past few years, provided a potential route to lowering costs. Tese sensors, which are based on thin films, use QDs as a detection mechanism for SWIR radiation. QDs are now disrupting the marketplace
for SWIR cameras, being deployed in products such as SWIR Vision Systems’ Acuros series and Emberion’s VS20 Vis- SWIR camera. Te Acuros cameras use a 15 micron pixel size in their detectors, as well as offering camera cooling technology to reduce the impact of thermal noise.
16 IMAGING AND MACHINE VISION EUROPE APRIL/MAY 2023
Emberion is also developing a version of its VS20 series with a GigE interface, potentially offering a framerate of up to 400fps. “QD-based SWIR cameras are already on
the market, costing around half the price of conventional alternatives,” says Adam Button, spokesman for Quantum Science, a specialist nanomaterials developer that uses a new fabrication technique to push down the price of QD-based SWIR sensors even further. Its in-house-developed INFIQ QDs are produced at scale using a synthetic process that facilitates control and production. Tis enables them, in Button’s words, to manufacture SWIR sensors “at a fraction of the price of existing technologies”. “Traditional formation processes involve
the deposition of 14-16 thin film layers, which introduce multiple opportunities for manufacturing errors to occur. INFIQ QDs can be deposited in a single layer. Tis minimises the risk of defects forming and accelerates the manufacturing process.” Quantum Science’s QD fabrication also
circumvents another barrier to the wide adoption of QDs: the toxicity of heavy metal components used in their manufacture. Te company is already partnering with sensor design companies and large semiconductor businesses with the aim of bringing its QDs to market quickly.
Blue skies ahead Promising advances for potential applications in semiconductor inspection are also happening in academia. Jinyang
@imveurope |
www.imveurope.com
SWIR Vision Systems & Emberion
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32