Intro into Antimonides in MOCVD - very challenging topic

MOCVD of Antimonides are notoriously difficult due to a few challenges, that are not present for arsenides and phosphides. They are not so arcane, as totally esoteric MOCVD epitaxy of HgCdTe (nurtured for over ~7 years).
How hard are they? Any, quantifiable metric? Lets see:

  1. Have fame to contaminate reactors, and poison GaAs and InP materials for some time, aftermath require extensive cleanings and bakings.. this scare-off 95% of growers with newer, shiny, metal chamber Planetary and ShowerHead beauties who need to run production or R&D based on GaAs and InP.

  2. Labs which publish some MOCVD works on Antimonides: CQD prof. Razeghi, Tyndall Institute, EPFL, Sheffield University they are politely not interested, and refuse to grow any commercial wafers with specs..

  3. Companies inquired, ask $1M for doing R&D project on Antimonides (btw. it is not Photin ;D ).

  4. Projects with $500k-750k budget, run hundreds of cal runs and show publications, but no real devices ( “Advanced technologies for semiconductor infrared optoelectronics”, PBZ−MNiSW 02/I/2007, ITME Poland)

Technically, problems with Antimonides stems from:

  1. InSb low melting point 525C = cap on the highest possible growth temperatures (pyrolysis of precursors, carbon/oxygen doping issues).

  2. Phase separation and miscibility gap for Quaternary compounds

  3. Lack of semi-insulating GaSb substrate

  4. No viable hydride Sb precursor

  5. Strict, III / V ratio close to 1, any small deviation result in Ga droplets on surface or Sb hillocks

  6. Low temperatures move growths range into kinetically limited regime, where process become very sensitive to multiple parameters

… but we did it, despite nimble budget that would make you laugh, we kept our focus and worked for ~3 years, to get full eSWIR Antimonide detector structure in MOCVD to TRL5-7

!! BIG THANKS TO Jiri Pangrac and Alice Hospodkova !!

Separate big thanks to so many of you who helped us, on this way, and for one reason or other yet, we cannot name you directly!, your help is remembered and appreciated!!!

What has been done:

We not wanted to engage into processing, which is also tricky on antimonides, but as the result of Photon Hub Inverstor Readiness Mentoring Programme, and discussions with Anne Lebreton-Wolf realized that potential investors would expect from us to show devices, not only epi-characterization.. so, gritting our teeth, we delved into searching for partners in the processing challenges.
From ~6 potential places we have started talks, selected 3 most promising, and finally, only one partner delivered what he promised, and got a very special place in our heart!
Voila! here we have our devices and full MVP!!

Two processing schemes were tested: planar (above), and mesa (below).

Discussion of the test detector results, will be subject of separate post.
There are a few improvements, that need to be implemented, but they are daily bread for growers - improving lattice matching, and lowering Al mole fraction in barrier layer.. also more on the epitaxy characterization of MVP later..

For now, lets admire traversing whole path of semiconductor development, from device concept, simulation, growth, characterization, processing and measurements.

Looking back, it is staggeringly amazing, all this happen within 1 year!?!

Technology Readiness Level - TRL

Valley of death. Source: Osawa, Miyazaki, 2006

It was studied and well known, that most of startup projects face “valley-of- death” around TRL5-7. This “disaster” will be addressed in next EU New Horizon Framework, but we have now ~1 year till outcome of this programs will take any real effect..

Lets dive-in, we are still sliding ;-)..

TRL5: Technology validated in relevant environment (industrially relevant environment in the case of key enabling technologies)

TRL6: Technology demonstrated in relevant environment (industrially relevant environment in the case of key enabling technologies)

TRL7: System prototype demonstration in operational environment

Aixtron MOCVD reactors are de-facto standard for industrial production of compound semiconductors, and growth of full structure in such reactor fulfill TRL6 definition.
Processing of full structure, and demonstration of detector operation is weak TRL7, which require a few more cycles of wafer growth and processing, to refine epitaxy parameters, and exclude processing issues affecting measurements.

InGaAsSb / AlGaAsSb barrier detector - IC MOVPE Poster

TBC.. In next posts:

  • why antimonides are 4th generation of semiconductors
  • the best literature results - InGaAsSb eSWIR detectors Chart
  • issues with incumbent extended InGaAs detector technology
  • Comparison of SIMS
  • MVP - what was proven?
  • discussion on valuation of technology

stay tuned…