Thursday, April 15, 2021

NSF Workshop on Quantum Engineering Infrastructure

 I spent three afternoons this week attending a NSF workshop on Quantum Engineering Infrastructure.  This was based in part on the perceived critical need for shared infrastructure (materials growth, lithographic patterning, deposition, etching, characterization) across large swaths of experimental quantum information sciences, and the fact that the NSF already runs the NNCI, which was the successor of the NNIN.  There will end up being a report generated as a result of the workshop, hopefully steering future efforts.  (I was invited because of this post.)

The workshop was very informative, touching on platforms including superconducting qubits, trapped ions, photonic devices including color centers in diamond/SiC, topological materials, and spin qubits in semiconductors.  Some key themes emerged:

  • There are many possible platforms out there for quantum information science, and all of them will require very serious materials development to be ready for prime time.  People forget that our command of silicon comes after thousands of person-years worth of research and process development.  Essentially every platform is in its infancy compared to that.  
  • There is clearly a tension between the need for exploratory research, trying new processes at the onesy-twosy level, and the requirements for work at larger scale, which needs dedicated process expertise and control at a level not typically possible in a shared university facility.  Everyone also knows that progress is automatically slow if people have to travel off-site to some user facility to do part of their processing.  Some places are well situated - MIT, for example, has an exploratory fab facility here, and a dedicated 200 mm substrate superconducting circuit fab at Lincoln Labs.  Life is extra complicated when running an unusual process in some tool like a PECVD system or an etcher can "season" the gadget, leaving an imprint on subsequent process runs.
  • Whoever really figures out how to do wafer-scale heteroepitaxy of single-crystal diamond will either become incredibly rich or will be assassinated by DeBeers.  
  • Fostering a healthy relationship between industrial materials growers and academic researchers would be very important.  Industrial expertise can be fantastic, but there is not necessarily much economic incentive to work closely with academia compared with large-scale commercial pressures.  There may be a key role for government encouragement or subsidy.  
  • It's going to be increasingly challenging for new faculty to get started in some research topics at universities - the detailed process knowhow and the need to buildup expertise can be expensive and slow to acquire compared to the timescale of, e.g., promotion to tenure.  An improved network that supports, curates, and communicates process development expertise might be extremely helpful.


Don Monroe said...

Sounds like they are going to need to understand two-level systems in amorphous materials better. (Not actually kidding)

Anonymous said...

Cannot emphasize the need for collaboration between industry and academic sample growers enough. But I really don't see an easy solution there, commercial interests trump all, and I really don't think external funding from the NSF can overcome that.

Paul Barclay said...

Hi Doug - Are there any public documents available from this workshop? I am involved with ramping up some fab facility efforts and am always trying to convey the multidisciplinary nature of the problems we face to Dean-type figures.

And yes, I am interested in purchasing some single crystal diamond on silicon wafers.