Millikelvin Si-MOSFETs for Quantum Electronics

summary

The gist

Millikelvin Si-MOSFETs for Quantum Electronics has manufactured fully depleted silicon-on-insulator MOSFETs tailored for overcoming the power dissipation barrier towards sub-1 K applications.

In short

Researchers developed fully depleted silicon-on-insulator MOSFETs to solve power dissipation issues in sub-1 K applications. These devices achieve significantly improved switching performance at millikelvin temperatures, enabling ultra-low power operation crucial for large-scale cryogenic CMOS integration and all-cryogenic control of quantum circuits.

Key concepts

Power Dissipation Challenge
Large power consumption in silicon electronics is a major hurdle for scaling up integrated circuits at extremely low temperatures (below 1 K). To scale down power, engineers must reduce both static losses (leakage currents) and dynamic losses (supply voltage), which requires switching performance beyond current CMOS capabilities.
Switching Metric Improvements
The study experimentally demonstrated that Si MOSFETs can achieve a switching speed of approximately 1 mV/dec at millikelvin temperatures. This improvement is quantified by a subthreshold swing (SS) that decreases with temperature, indicating better control over the transistor's on/off state at cryogenic conditions.
Cryogenic Characterization
The devices were tested using a dilution refrigerator cooling system reaching 420 mK. Measurements showed switching speeds of around 5.1 mV/dec for p-MOSFETs and 4.5 mV/dec for n-MOSFETs at this temperature, providing empirical data on the device's performance in the target cryogenic environment.

Terminology used across episodes

This episode discusses

The paper

Millikelvin Si-MOSFETs for Quantum Electronics · Read on arXiv

SemiQon Technologies, University of Jyvaskyla, NanoScience Center

DOI: 10.1063/5.0322860

Transcript

Introduction to the show: ident: Quantum Radio. Generated commentary on the latest quantum physics and condensed matter papers.

Kai: I'm Kai.

Mira: And I'm Mira, here with Lev.

Kai: We're looking at the paper: "Millikelvin Si-MOSFETs for Quantum Electronics".

Mira: Here's the gist of the paper. Millikelvin Si-MOSFETs for Quantum Electronics has manufactured fully depleted silicon-on-insulator MOSFETs tailored for overcoming the power dissipation barrier towards sub-1 K applications.

Kai: This advancement can solve the power dissipation bottleneck, limiting very large-scale integration of CMOS circuits at ultra-low temperatures, and enable all-cryogenic control of quantum circuits, which is a prerequisite for fault-tolerant quantum computation.

Mira: The Power Dissipation Challenge Large power consumption of silicon CMOS electronics is a challenge in very-large-scale integrated circuits and a major roadblock to fault-tolerant quantum computation.

Kai: Matching the power dissipation of Si-MOSFETs to the thermal budget at deep cryogenic temperatures, below 1 K, requires switching performance beyond levels facilitated by currently available CMOS technologies.

Title and authors: Kai: So we're looking at the paper "Millikelvin Si-MOSFETs for Quantum Electronics" and who wrote it. It’s by Nikolai Yurttagül, Markku Kainlauri, Jan Toivonen, Sushan Khadka, Antti Kanniainen, Arvind Kumar, Diego Subero, Juha T. Muhonen Mika Prunnila and Janne Lehtinen >

Mira: It’s a team that spans the necessary expertise—from experimental cryogenics to condensed matter theory to error correction building—which is important because this work touches so many different aspects of quantum hardware development >

Lev: I’m interested in who handled the actual fabrication side. Because when you talk about achieving sub-one Kelvin operation, the physical realization of those MOSFETs is everything > <ref:2410.01077#pg1>

Kai: That's where my focus lies. I worked on superconducting qubits and cryogenics, so seeing how they actually built these fully depleted silicon-on-insulator MOSFETs and what measurements they took is key to me >

Mira: And from a theory standpoint, I want to know the assumptions they made about the physics of those fully depleted channels. For instance, how did they account for things like charge retention times or quantum effects that might shift the threshold voltage >

Lev: From an error correction perspective, I'm concerned with what kind of overhead this hardware will really land on when you try to run real quantum algorithms on it >

Kai: The paper mentions that these devices are tailored specifically for sub-one Kelvin applications, which suggests they’ve made specific design choices for the gate stack fabrication > <ref:2410.01077#pg1>

Mira: And that tailoring is important because it directly addresses the thermal budget issue. Matching silicon CMOS power dissipation to the thermal budget at deep cryogenic temperatures requires performance beyond what's currently available >

Lev: So, in essence, they are trying to find a way around that fundamental roadblock where silicon CMOS just gets too hot when you try to cool it down deeply >

Kai: That’s right. They have manufactured these transistors and shown they can get the switching performance up enough for this purpose >

The paper's summary: Mira: So, summarizing what the paper says about "Millikelvin Si-MOSFETs for Quantum Electronics," it boils down to overcoming the power dissipation barrier in silicon CMOS electronics for ultra-low temperature work >

Kai: They are focused on matching the power dissipation of these silicon MOSFETs to the thermal budget at temperatures below one Kelvin >

Mira: They achieved this by fabricating fully depleted silicon-on-insulator MOSFETs designed for this purpose, and they found a major milestone with a subthreshold swing of zero point three mV per decade at four hundred twenty millikelvin > <ref:2410.01077#pg1,fully depleted silicon-on-insulator MOSFETs>

Lev: That subthreshold swing number is what really tells you how much better their switching is compared to previous levels, which the paper says is an order of magnitude larger than the thermionic limit at four Kelvin >

Kai: It’s important to remember that they are comparing this to other foundry-fabricated FD-SOI MOSFETs, which haven't shown significant improvement below one hundred millikelvin so far >

Mira: Right. Beyond just switching speed, the paper notes other key metrics improve as you cool down, including reduced wire delays and increased on current due to carrier mobility and lower gate leakage currents >

Lev: Those reduced wire delays are crucial for reducing dynamic power loss in circuit design when you start scaling up the physical size of the chip >

Kai: And they also have this control over the threshold voltage shift, which is something that makes FD-SOI MOSFETs attractive as a technological platform for cryoCMOS because it lets you manage that thermal shift >

Mira: So, in short, the paper shows how these specific device structures can help bypass the power dissipation bottleneck by improving switching performance and controlling device characteristics at cryogenic temperatures >

Lev: The main takeaway is that they’ve found a way to push silicon MOSFETs into a regime where they are useful for all-cryogenic control of quantum circuits >

The paper's improvements: Kai: Now, when we talk about the specific improvements the authors suggest from this work, one major improvement is achieving that subthreshold swing of zero point three mV per decade at four hundred twenty millikelvin > <ref:2410.01077#pg1,subthreshold swing of 0.3 mV>

Mira: That’s a direct result of their custom gate stack fabrication, and it means you can achieve very large-scale integration of cryoCMOS electronics for ultra-low temperature applications >

Lev: And from an engineering standpoint, this improvement allows for the design of inverters where we can use supply voltages that are "the order of tens of millivolts" instead of state-of-the-art values above zero point four volt or one volt >

Kai: That low voltage level leads to a reduction in dynamic power loss by more than a factor of a hundred because the supply voltage is so low >

Mira: And then they suggest that the system can operate with an upper limit on supply voltage defined by the direct cooling power, showing we could incorporate millions of MOSFETs with a common supply voltage of fifty millivolts and switching at ten mega hertz while dissipating less than one hundred nanowatt >

Lev: So they are tying the maximum operating voltage directly to the cooling power, which is a very smart constraint for scaling up hardware >

Kai: And that low power metric is what finally enables the all-cryogenic control of quantum circuits because it means an analog circuit with one hundred transistors only dissipates one pico watt per mega hertz >

Mira: It’s small compared to the effective cooling power available, which gives us enough margin to actually use this for controlling quantum systems >

Lev: So we're looking at a system where the hardware itself is extremely low power and just small enough that it can handle the load of quantum control circuits >

Conclusion: Kai: So, wrapping up on "Millikelvin Si-MOSFETs for Quantum Electronics," this paper shows a clear path forward by demonstrating how to tackle the power dissipation barrier using custom devices tailored for millikelvin operation >

Mira: It’s about reaching that milestone of zero point three mV per decade switching swing at four hundred twenty millikelvin and enabling large-scale integration of cryoCMOS electronics > <ref:2410.01077#pg1>

Lev: The implication for real hardware is that we can start serious design work on inverters running at very low supply voltages, which could cut dynamic power loss by a factor of a hundred >

Kai: And the ultimate goal is to leverage this low power for all-cryogenic control of quantum circuits because it makes the hardware quiet enough to reach that level >

Mira: It's about using this technology to enable fault-tolerant quantum computation by providing the necessary hardware foundation >

Lev: The paper on "Millikelvin Si-MOSFETs for Quantum Electronics" really gives us a concrete, measurable target for what we need to build in the next generation of hardware >

More episodes

← Home