Author Topic: The future of EPMA innovation  (Read 176 times)

Mike Jercinovic

  • Professor
  • ****
  • Posts: 89
    • UMass Geosciences Microprobe-SEM Facility
The future of EPMA innovation
« on: May 08, 2022, 09:49:18 AM »
There are still a lot of things to say about all this.  I will not reiterate my previous posts concerning the innovations in EPMA that have, in fact, been done in the last 20 years.  I cannot really speak to the market situation in anywhere else in the world, including the growing market in Asia, but the US market has been interesting.  Yes, the cost of ownership is a big factor, and companies that used to buy these machines largely do not do that anymore, more on that in a minute.

Our situation at UMass is probably not uncommon, so this is what I have seen:  NSF made a paradigm switch over the last 10-15 years toward largely investing in large facilities with multi-year support, like IRIS, the UCLA National Ion Microprobe, GeoSoilEnviroCARS, Institute for Rock Magnetism, accelerator PRIME lab, Continental Drilling Office, and a number of others that eat up tens of millions of dollars in multi-year support.  This meant less funding for individual lab acquisitions, much to the dismay of a number of people at NSF!  The NSF EAR IF upper limit went down and down to the point where acquisitions like EPMAs had to go through the MRI program, and just to get an MRI off campus is a difficult, usually several year process.  So NSF shifted EPMA acquisitions mostly into regional labs (with some exceptions) rather than individual campuses.  When I started 25 years ago, NSF EAR IF was really about getting as much equipment into as many labs, on as many campuses as possible, so this was a big change.  Meanwhile, universities have seen unprecedented growth in spending on a rapid expansion of the administration – this has been shocking at UMass, and I know other places.  How does this tie in? Well, spending on campus laboratory and facility support has gone down, with more centralization to minimize staff, making staff more soft money supported, and encouraging self-supporting models for laboratories.  Revenue from academic grants has gone down also, and is highly inconsistent as funding rates decrease, and analytical budgets in grants goes down.  Some manifestations of this include the encouragement by the administration to 1) set up collaborations with commercial entities, like Big Pharma, and 2) figure out ways of getting commercial users into university laboratories.  At UMass, this has included the implementation of a voucher program to lure in commercial clients.  We have moved from cost centers to revenue centers - but we actually call these revenue centers cost centers with cost recovery through revenue generation because we are still, after all, non-profit.

What the companies see is that there is a better way for them to get what they need done by using these university laboratories rather than buy an instrument, hire expert staff, and buy a service agreement.  It is way more cost effective to have a university take on the support and just pay an hourly price (at an approved commercial rate) for analyses.  This means labs may spend a tremendous amount of time running things for commercial customers, and less and less time fulfilling our educational mission.  I recently saw an ad for an XRF manager/technician for a university that just flat out said that this involved two XRF instruments, one for general use, and the other for a commercial machine evidently dedicated to use for their commercial users.  Sadly, the university administration may not care as much about compromising the educational mission of laboratories compared to just the pure revenue you might be able to generate – of course they take a cut out of the budget for “administrative charges”.  And, of course, one big picture aspect of the entire approach by the administration has to do with attracting big corporate or individual donors for whom we can name buildings or schools after.

This laboratory support model does not encourage experimentation or innovation.  It also has meant that fewer and fewer students will dedicate a good portion of their academic careers to learning things like EPMA in a way that is not purely just a service.  For a lot of the campus work we do, it is now much more likely that we will do something here and there as a service arrangement where the students involved will not do anything more than bring samples over.

Twenty years ago, NSF EAR IF was very receptive to ideas involving innovation.  The word that we got was that there were too few risks being taken.  The Ultrachron project was enthusiastically supported because there was risk to innovate, and to develop such innovation through a collaboration between UMass, Cameca and NSF was perfect.  Instead of just buying something off the shelf, what could you do that might be new and of potential benefit to both the scientific community as well as the company?  Seems like this is less much likely to happen now, even though there are still some very good and well-intentioned people at NSF.

Sorry for the long Sunday diatribe, but to me, if this is the demise of EPMA, it has been a community effort that involves a lot more than just the perceived shortcomings of Cameca and JEOL.  In the US anyway, a lot of us using this forum have to remember that WE ARE NSF.  As ad-hoc proposal reviewers or panelists, we participate in the process.  It will only change when we make it change.

Probeman

  • Emeritus
  • *****
  • Posts: 2384
  • Never sleeps...
    • John Donovan
Re: The future of EPMA innovation
« Reply #1 on: May 08, 2022, 07:43:07 PM »
I will add my 2 cents to Mikes comments.

The "administrationism" of academia is a pernicious trend that has and will continue to affect the quality of research at universities.

When I arrived in the geology department at UC Berkeley in the late 1980s we had 2 front office people and about 8 technical staff (2 in the machine shop, a petrographic technician, an XRF technican, an electronics technician, a field and support person and helper and a museum/collections person). One of those shop people was me, who eventually started working on the microprobe!

By the time I left in the early 2000s we were down to 2 technical staff and there were around 8 people in the front office.

When I went to Oregon we deliberately designed the CAMCOR facility to maximize the ratio of technical staff to office people, and we had 1 front office person and about 9 technical staff.  But we could only support those technical staff (salaries and benefits) by charging about 20 to 30% of our instrument hours at commercial rates (which are roughly 10x academic rates). In effect the commercial work subsidizes the academic work, which remains about 70 to 80% of our chargeable instrument hours.

Mike makes a good point about the effects of commercial work versus pure scientific research, but I maintain that by creating a large highly professional technical staff we can create an environment which produces excellent science on well maintained instruments (we generally don't have service contracts, instead we have our own instrument engineer with occasional OEM support).

In addition we have a large number of student interns who work in the lab along with regular grad students, all taking instrument classes and getting hands on experience with the instruments. And yes, most of whom eventually usually go into industry, but that's where the jobs are, if we are being honest.

It's definitely a trade off, but certainly a viable model for supporting academic research labs.

https://camcor.uoregon.edu/
« Last Edit: May 08, 2022, 09:40:27 PM by Probeman »
The only stupid question is the one not asked!

Probeman

  • Emeritus
  • *****
  • Posts: 2384
  • Never sleeps...
    • John Donovan
Re: The future of EPMA innovation
« Reply #2 on: May 09, 2022, 10:08:19 AM »
I need to add to my comments above and mention that although CAMCOR started with one front office worker, university paperwork requirements demands more person hours that just that, so we also hired a buyer and an accountant.

But most importantly, just as with the technical staff, we made them responsible for generating enough income to pay for their positions. So how does a buyer and a buyer or accountant accomplish this?  By offering their services to other departments at the university!

I don't know the exact number but our front office pool provides services to half a dozen other academic departments. It makes financial sense, especially for the smaller departments.

As to innovation in academia, that is a tough one.  I personally think innovation, particularly in EPMA science has been difficult for several reasons. One, because the companies no longer are run by scientists, but rather MBAs and second, they don't collaborate as actively with academic or government scientists as was common in the 1980s and 1990s. And it's difficult to justify scientists on a company payroll when the product has such a limited market.

Basically JEOL subsidizes their EPMA preoduct throiugh SEM/TEM sales, I had hoped that Cameca could do the same through their SIMs/APT market.

I was really excited about 5 years ago when Cameca approached me and a number of our colleagues, to work with Tom Kelly on a new EPMA design, but Tom was fired and that project shutdown by the bean counters.

The other problem in EPMA innovation is that it's just hard to find time in one's day when running a lab and dealing with students/customers to set aside time for research and development. I really tried at Oregon, but eventually I had to go half time to get enough time (and hurting my back didn't help!) to devote to improving the technique. And I was lucky because I had a half time lab manager.

I sometimes wonder if there could be some provision for technique development through NSF, but I suspect they would say that should be done at the companies. At least that was the response to our NSF TEPNA proposal (see attached).

Today I would modify this proposal extensively, for instance replace the sample biasing stage with an electrostatic decel lens (now that the Zeiss patent has expired). A decel lens would reduce landing energy *and* reduce the beam diameter.

In fact I wish JEOL would offer a decel lens on their EPMA instrument, which would really help with sub micron thin film micro chemistry and even nano-geology.
« Last Edit: May 09, 2022, 10:18:51 AM by Probeman »
The only stupid question is the one not asked!

NicholasRitchie

  • Professor
  • ****
  • Posts: 106
    • NIST DTSA-II
Re: The future of EPMA innovation
« Reply #3 on: May 10, 2022, 07:32:56 AM »
My earlier comment about "incentives for innovation" had less to do with innovation driven by academic researchers and more to do with the innovation by instrument companies.
  • Electron beam instruments are complex.  The only people with the engineering staff necessary to design and innovate them are the companies.
  • There is a lot of money in certain ebeam applications - think semiconductors
  • There is almost no money in other ebeam domains - think academic geology, forensics
  • Companies are increasingly run by bean-counters - not scientists
  • Bean counters are going to focus limited resources on those fields with money
  • Other fields may benefit from this indirectly
EPMA is in a catch-22.  The instrument is hard to use because of all the fiddly setup required and all the ways that a measurement can go wrong.  This limits the number of people who can/will perform EPMA.  Basically, each microprobe requires a PhD to run it.  The limited number of users provide little incentive to bean-counters to invest more money to improve the instrument (or, it would seem, continue to sell a pre-existing instrument).

On the other hand, if they did see past the current limitations and invest more money to make EPMA easier to use, they could potentially attract more users who currently decide that EPMA is too complex.  Changes like replacing the counters with solid stated detectors or replacing the analog pulse processors with digital pulse processors would go a long way to making EPMA less error prone.  More assistive software could too. (Probe for EPMA is amazing software for the expert but overwhelming for the novice.)

Could we double the number of users if EPMA was half as complex?  There are tens-of-thousands of SEM/EDS users.  If we could convince a small fraction of these people that an EPMA is a better alternative, we could easily double the market for microprobes.  Is ten times possible?  With ten times the number of users, companies might see an incentive to invest more money in microprobe development leading to a "virtuous cycle" of more customers => more development => more customers.

How do we get from the "catch-22" to a "virtuous cycle" of innovation?
« Last Edit: May 10, 2022, 08:30:11 AM by John Donovan »
"Do what you can, with what you have, where you are"
  - Teddy Roosevelt

Jacob

  • Professor
  • ****
  • Posts: 26
Re: The future of EPMA innovation
« Reply #4 on: May 10, 2022, 09:27:21 AM »
I suspect it may be about time to give up on EPMA conceptually as single instrument and explore ways to make SEMs more EPMA-like.  To me that means solving the working distance problem by offering some kind of off-axis light pathway, maybe with a tilted focal field or interferometer or $cleverIdea.  Faraday cups could be pushed into existing column designs, which helps everyone (EDS users and lithography users included).

Then instead of a microprobe, a company like Cameca could sell WDS kits like Bruxford-Fisher but with a really nice proven spectrometer design.  WDS and optical paths could also be integrated into all sorts of fascinating configurations, like into automated mineralogy systems.  People could finally build that mythical beast, the instrument with the Cameca spectrometers and a JEOL/Hitachi/xxx column.  EDS packages could have enough retractable faraday cups in the wild to make probe current measurement API call implementation worth their while.

Probeman

  • Emeritus
  • *****
  • Posts: 2384
  • Never sleeps...
    • John Donovan
Re: The future of EPMA innovation
« Reply #5 on: May 10, 2022, 10:17:06 AM »
I don't know if modifying an SEM to be more effective with WDS makes sense or not, but it's funny considering how I started out on an ARL SEMQ instrument. The SEMQ stood for "Scanning Electron Microscope Quantometer". The idea being that it could be used as an SEM with a long working distance by lowering the stage, or as a WDS microprobe at short working distances.

The JEOL instrument still allows lowering the stage for SEM imaging I think...
« Last Edit: May 10, 2022, 09:52:33 PM by Probeman »
The only stupid question is the one not asked!

sem-geologist

  • Professor
  • ****
  • Posts: 162
Re: The future of EPMA innovation
« Reply #6 on: May 11, 2022, 07:56:16 AM »
I suspect it may be about time to give up on EPMA conceptually as single instrument and explore ways to make SEMs more EPMA-like.  To me that means solving the working distance problem by offering some kind of off-axis light pathway, maybe with a tilted focal field or interferometer or $cleverIdea.  Faraday cups could be pushed into existing column designs, which helps everyone (EDS users and lithography users included).

This is going backwards to go forward (back to current point). EPMA design differs from SEM in few crucial places. Making SEM more like EPMA is effort to achieve something what was already achieved previously. It is not enough just to stick few devices which are available on EPMA to SEM to make it EPMA-like. Even if instruments looks very similar at outside, the design principles are quite different and devil is in details. SEM is designed and constructed to be versatile and cheep (in production, not selling) - it does lots of compromises especially in stability vs universality (favoring later than former). SEM workflows does not require beam HV and current to be stable within +/- 0.05% of set value and thus electronics are most commonly much more simpler on SEM (read it: "worse stability"). For Quantitative work we need to not only know what beam current is before and after analysis, but also to be sure that beam is very stable during the analysis. Even looking at cover it is easy to spot that EPMA has more electronics cabins than most of SEM (often all primary electronics of SEM fits into the space under the column). Most of space/volume in EPMA electronics is occupied by separate PSU sections for every sub-system, and voltage-current stabilization for control and powering of different sub-systems. Show me a SEM with stable 200, 500 or 900nA focused beam? EPMA has integrated current real-time stabilisation to achieve 0.05%/h stability. It is not just simple device attached to column - it is few systems tightly integrated (HV regulation, column coil regulation, picoamperometer...). The column design in EPMA is focused on the beam stability - imaging comes as secondary outcome, where SEM is focused on imaging and beam stability (energetic and positional) is not primary important. The faraday cup inside the column is only half of requirement  (which is better than on-stage FC not due to convenience but for repeatability as the resistance is constant, where measurement will depend on stage from positoin FC is mounted at). EPMA has precise picomaperometer, which is more sophisticated than what SEM normally (if at all) contains, as EPMA has more dynamic range of beam currents and picoamperometer is divided to segments for different current ranges, so that it would scale linearly from very small to very huge beam currents, and thus [cps/nA] units are unversal.

Then instead of a microprobe, a company like Cameca could sell WDS kits like Bruxford-Fisher but with a really nice proven spectrometer design.  WDS and optical paths could also be integrated into all sorts of fascinating configurations, like into automated mineralogy systems.  People could finally build that mythical beast, the instrument with the Cameca spectrometers and a JEOL/Hitachi/xxx column.  EDS packages could have enough retractable faraday cups in the wild to make probe current measurement API call implementation worth their while.


What is so wrong with Cameca column? I see not a first time someone wants JEOL, Hitachi, Zeiss or etc... Cameca column is designed buttom-up for EPMA, where those SEM columns are not or (in case of JEOL) compromised between imaging and stability.
One more thing about Cameca WDS - it is not physical WDS You see attached to the column - WDS is highly integrated with whole system through VME (where it is coupled with stage). You can't detach it from EPMA and attach to SEM without a complete major redesign how it works. Electronics is part of the spectrometer. Also most of SEM has 35 degree ports, where EPMA is 40.