The secret to nanoscale chip wiring is using capillary action to pull heat-softened copper down into tiny structures.
By Richard Lewington
If you were to slice up a microchip and take a look (you’d need a really powerful microscope, I’m afraid) you would see what looks like a nanoscale layer cake.
All the active circuit elements—transistors, memory cells, etc.—are on the bottom. The other 90% of the chip is a maze of tiny copper wires, which we call interconnects.
The history of chip development is all about shrinking circuit features. When the transistors shrink, so must the interconnects. Today, the smallest interconnects can be fewer than 200 atoms across.
Interconnects are made by filling molds of insulating material with copper. At these miniscule dimensions, completely filling these features becomes very difficult. If it’s not done just right, bubbles of vacuum, called voids, can be formed, which can create a short circuit. In the most advanced device designs, even one void can render the entire chip useless.
Applied Materials’ Endura Amber physical vapor deposition system aims to banish interconnect voids forever. The new Amber system takes advantage of capillary action to pull heat-softened copper down into even the tiniest structures.
In this video, I take a quick look at how the interconnect fabrication process is done and then demonstrate how our revolutionary copper reflow technology works.
AI-enabled systems are being designed to process more data locally as device scaling benefits decline.
Foundry forms ASIC subsidiary as it focuses on 14nm/12nm and above.
Growing volume of data and limited improvements in performance create new opportunities for approaches that never got off the ground.
Why auto tech companies are so concerned about interactions with humans.
Company’s push into deep learning opens door to a variety of new architectures, including tiles, advanced packaging and more customized solutions.
Costs of developing a complex chip could run as high as $1.5B, while power/performance benefits are likely to decrease.
AI-enabled systems are being designed to process more data locally as device scaling benefits decline.
Lines blur as processors are added into traditional FPGAs, and programmability is added into ASICs.
Foundry forms ASIC subsidiary as it focuses on 14nm/12nm and above.
Technology has the potential to reshape processing everywhere, starting with limited scientific and commercial applications.
Market overcrowding, more efficient manufacturing, and growing list of scaling issues create a challenging competitive landscape.
Using the Von Neumann architecture for artificial intelligence applications is inefficient. What will replace it?
Experts at the Table, part 1: Why machine learning works in some cases and not in others.
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
Leave a Reply