TSMC reveals N6 process based on 6nm EUV

onetheycallEric

Posts: 225   +47
Staff
Forward-looking: TSMC and Samsung have been attempting to leapfrog one another in terms of extreme ultraviolet lithography. After years of waiting, it seems the industry is finally making strides with EUV, as TSMC announced it also has a 6nm process shrink set to go into risk production.

TSMC doesn't want to be left out of the conversation as it pertains to semiconductors at 6nm. Trailing news of Samsung's progress in the realm of 5nm and 6nm process nodes, TSMC announced its N6 process, based on 6nm EUV.

TSMC's 6nm manufacturing process is largely a migration from its 7nm N7 process that sees much of the IP and development ecosystem recycled, so to speak. TSMC is able to take what it has learned with EUV at 7nm and port it directly down to 6nm, keeping the same design rules compatible and speeding up the time to market. This also makes it easier for customers to get started on the process.

By keeping the same design rules, previous N7 customers are able to leverage the N6 process at considerable cost savings, while enjoying what TSMC states is around a 15 percent die shrink. Furthermore, the N6 process is supposed to deliver an 18 percent increase in transistor density while maintaining the same performance and power consumption enhancements.

TSMC is expecting the N6 process to be adopted for applications in mobile, consumer applications, AI, networking, 5G infrastructure, GPU, and high-performance computing.

Risk production is scheduled for the first quarter of 2020; volume production likely won't be until 2021. While TSMC's progress with EUV is considerable, it'll be interesting to see what customers might be interested in such an intermediary node.

Permalink to story.

 
This is cool and all, but I'm curious why TSMC needs a 7, 6 and 5 nm process. I'm sure there is a reason, just haven't read about it in this or any other articles yet.
 
This is cool and all, but I'm curious why TSMC needs a 7, 6 and 5 nm process. I'm sure there is a reason, just haven't read about it in this or any other articles yet.

This node is announced for small 'risk' production in early 2020. 7nm is now in full production. There is an opportunity for 7nm customers to possibly transition or sample improved products.

Saying that not everybody needs the most cutting edge manufacturing processes- these are usually the riskiest for production issues and the highest cost for early runs. Also, not every node is optimised for the same type of processor. Some nodes are better for low power, small chips, some better for big high performance ones.

TSMC has various production lines around the world that offer these different node sizes and types at different costs, with finite capacity at each one. The customer can decide what they are building, what exact node they need it to be on, when they need it for, and TSMC can then tell them the cost.

In these cases maybe TSMC offer 6nm capacity if all their lines setup for 7nm are too busy. Maybe it's better to expand on that and that is what customers are telling them they want now instead of a chip that was planned on 7nm, so they increase capacity there.

It's supply and demand in this business, as ever.
 
Back