editor's blog
Subscribe Now

Sticking With their Story: Zeno Demonstrates 1T SRAM at Leading Nodes

Let’s face it: We’re addicted to SRAM. It’s big, it’s power-hungry, but it’s fast. And no matter how much we complain about it, we still use it. Because we don’t have anything better in the mainstream yet.

We’ve looked at attempts to improve conventional SRAM. We’ve looked at completely different memory technologies that might be in line to replace SRAM, like RRAM (first more than five years ago, then last year – and we’re still working on it) and MRAM. And, a couple of years ago, we looked at one attempt to create a new SRAM technology out of conventional transistors – to be clear, just one such transistor (instead of six), plus a possible select transistor.

The company was Zeno, and they were even leveraging this idea as a way to provide greater drive in a transistor – possibly meaning one could stick with older technologies for longer. But, back in the memory story, there was at least one unknown – granted, a “we think so,” but without proof.

Here’s the thing: the original technology was proven on 28-nm planar technology. Could it be leveraged onto more aggressive nodes – ones with FinFETs? While Zeno thought it should work, they hadn’t run the experiments yet to prove it.

And it turns out that there was another nagging question – one of the little details that can sabotage a great new idea. The technology was originally implemented with a buried N-well below the transistor. This wasn’t – and isn’t – part of a standard CMOS flow. Yeah, it’s possible to change flows over time, but it’s super hard, and there better be a dang good payoff from it. Is a 1T SRAM cell enough to motivate that? Perhaps, but the other question to ask – and the one that Zeno asked – is, “Is that modification to the flow really necessary? What happens if we omit the buried N well. Will it still work?”

At this year’s IEDM, they came back and reported on those two items – both in a discussion and in a paper that they presented. And the answer to the two questions is: Yes.

The next figure shows the layout at 14 nm, with a 6T conventional SRAM on the left and their 1T version on the right.

(Image courtesy IEDM/Zeno)

They also showed the “profile” at 14 nm as compared to the original 28-nm planar version. Note both the use of a FinFET and the absence of the underlying n-well on the right. This latest rendition uses the standard foundry CMOS process.

(Image courtesy IEDM/Zeno)

They presented data showing the stability of the bistable element as data is read repeatedly, showing it sticks around – and that the read operation isn’t destructive.

(Image courtesy IEDM/Zeno)

And they presented lots more detail on reading and writing results in their paper. If you have access to the IEDM proceedings, look for, “A Bi-stable 1- /2-Transistor SRAM in 14 nm FinFET Technology for High Density / High Performance Embedded Applications.

 

More info:

Zeno

Leave a Reply

featured blogs
Jan 21, 2020
For many years computer systems have augmented CPUs with special purpose accelerators that are targeted at specialized tasks. Examples of these co-processors include special purpose graphics and digital signal processors. Lately there has been an interest and significant rese...
Jan 21, 2020
My mother is convinced that I'm a prodigy (for some reason she spells it "i-m-b-e-c-i-l-e"), but it may be that innate genius lies dormant in all of us....
Jan 20, 2020
As you probably know, discrete wire component data is quite a bit different than just standard socket and terminal mating relationships. When we look at how Samtec approaches discrete wire products, there are several components involved. We not only sell the assemblies, but w...
Jan 17, 2020
[From the last episode: We saw how virtual memory helps resolve the differences between where a compiler thinks things will go in memory and the real memories in a real system.] We'€™ve talked a lot about memory '€“ different kinds of memory, cache memory, heap memory, vi...

Featured Video

Automotive Trends Driving New SoC Architectures -- Synopsys

Sponsored by Synopsys

Today’s automotive trends are driving new design requirements for automotive SoCs targeting ADAS, gateways, connected cars and infotainment. Find out why it is essential to use pre-designed, pre-verified, reusable automotive-optimized IP to meet such new requirements and accelerate design time.

Click here for more information about DesignWare IP for Automotive