Ultrawide‐Bandgap Semiconductors: Research Opportunities and Challenges

Ultrawide‐Bandgap Semiconductors: Research Opportunities and Challenges J. Y. Tsao Srabanti Chowdhury M.A. Hollis Debdeep Jena N. M. Johnson K. A. Jones Robert Kaplar Siddharth Rajan Chris G. Van de Walle E. Bellotti C.L. Chua Ramón Collazo Michael E. Coltrin James A. Cooper K. R. Evans Samuel Graham T.A. Grotjohn Eric R. Heller Masataka Higashiwaki M. Saif Islam P Juodawlkis M. Asif Khan Andrew D. Koehler Jacob H. Leach Umesh K. Mishra R. J. Nemanich Robert C. N. Pilawa-Podgurski James B. Shealy Zlatko Sitar Marko J. Tadjer Arthur F. Witulski Michael Wraback J. A. Simmons

Tag layer of the DHSS Meta-Tagging Project. The article text is reproduced from the staged copy named in sourceDesc and is not altered; the tags are held separately in standOff so that overlapping and nested spans are preserved.

Ultrawide‐Bandgap Semiconductors: Research Opportunities and Challenges 2017 Advanced Electronic Materials 10.1002/aelm.201600501 tsao2017-ultrawide-bandgap-semiconductors-r tsao2017-ultrawide-bandgap-semiconductors-r.txt, the project's staged copy 02c891193deecf33079d60a6b76c4be91f674a4a cc-by-nc materials-science

the paper's own words

the tag layer; a sentence written by the tagger about the paper, always anchored to a quotation but not itself a quotation

0 spans located by searching the stored text for the quote's exact words, allowing any whitespace between them the ultrawide-bandgap semiconductors - AlGaN/AlN, diamond, Ga2O3 and cubic BN - and what stands between their theoretical advantage and a working device Quotation withheld because this record is not cleared for redistribution. which materials, physics and device problems must be solved before ultrawide-bandgap semiconductors can deliver the performance their bandgaps imply Quotation withheld because this record is not cleared for redistribution. 26 numbered research opportunities and challenges, a re-derived table of reference material properties, and an explicit maturity assessment of each material system Quotation withheld because this record is not cleared for redistribution. critique Quotation withheld because this record is not cleared for redistribution. a nonlinear payoff: because the critical field scales as the square of the bandgap and the Baliga figure of merit as its cube, device performance scales as roughly the sixth power of bandgap Quotation withheld because this record is not cleared for redistribution. review review Quotation withheld because this record is not cleared for redistribution. why a modest bandgap increase is worth a hard materials problem: the critical field goes as the square of the bandgap and the Baliga FOM as its cube, so GaN to AlN is 1.8x in bandgap and about 34x in figure of merit sixth-power scaling Quotation withheld because this record is not cleared for redistribution. why doping gets harder as the gap widens - Quotation withheld, so free carriers first rise with dopant concentration and then fall compensation rising with bandgap Quotation withheld because this record is not cleared for redistribution. a second doping obstacle: localized valence-band states, notably from first-row elements such as oxygen, let a hole trap itself in its own lattice distortion as a small polaron, suppressing mobility even where a shallow acceptor exists self-trapping Quotation withheld because this record is not cleared for redistribution. the fundamental obstacle to gate insulators: a dielectric with a large bandgap tends to have a low dielectric constant and vice versa, so the high-k insulator a UWBG MOSFET wants may not exist the bandgap-permittivity trade-off Quotation withheld because this record is not cleared for redistribution. new physics that only appears here - when the bandgap exceeds the electron affinity, a carrier must have kinetic energy above the affinity to impact-ionise, so it momentarily occupies vacuum states before doing so breakdown through vacuum states Quotation withheld because this record is not cleared for redistribution. why carrier confinement gets harder with a wider gap: the number of semiconductors and insulators with large enough band offsets DECREASES, and for AlN, BN and diamond few dielectrics are predicted to confine electrons at all confiners run out Quotation withheld because this record is not cleared for redistribution. every one of the three leading UWBG materials is poor on at least one of the three metrics that matter: AlN and Ga2O3 cannot be usefully p-doped, and diamond substrates are small or defective Quotation withheld because this record is not cleared for redistribution. the critical breakdown fields of AlN, BN, Ga2O3 and diamond - the property the whole case rests on - are yet to be accurately determined Quotation withheld because this record is not cleared for redistribution. Ga2O3's thermal conductivity is one to two orders of magnitude below the other UWBG semiconductors, which the paper calls perhaps its single most serious weakness Quotation withheld because this record is not cleared for redistribution. and p-type doping of Ga2O3 has never been reported, with hole self-trapping predicted from first principles Quotation withheld because this record is not cleared for redistribution. AlN has no known acceptors at all, and only one postulated shallow donor Quotation withheld because this record is not cleared for redistribution. c-BN is the least mature of them, with substrates available only by the high-pressure high-temperature route Quotation withheld because this record is not cleared for redistribution. the survey covers only the KNOWN applications, and the authors say plainly that the technology will create applications that do not yet exist Quotation withheld because this record is not cleared for redistribution. extracting a critical field from a working device is unreliable because edge terminations cause premature breakdown - and the higher the voltage the more likely that is, so the measurement gets harder exactly where it matters most Quotation withheld because this record is not cleared for redistribution. deep-level transient spectroscopy stops working on these materials because the traps are too deep and emission times too long - a standard characterisation technique defeated by the property under study Quotation withheld because this record is not cleared for redistribution. and as devices shrink, optical thermometry becomes diffraction-limited, so a point-temperature measurement silently becomes a weighted average over a volume Quotation withheld because this record is not cleared for redistribution. TCAD device-modelling tools and physical models developed for silicon, applied to ultrawide-bandgap semiconductors Quotation withheld because this record is not cleared for redistribution. re-surveying the room-temperature material properties from the primary literature rather than reproducing the values in common use eliminates: a comparison table built on numbers measured decades ago on material that would now be called highly defective - which is the stated reason the re-survey was necessary · outcome: held Quotation withheld because this record is not cleared for redistribution. projecting every material's critical field from GaN's measured value THROUGH THE SAME assumed device - a non-punched-through p+/n diode at a stated donor concentration eliminates: materials being ranked on critical fields measured in different device geometries at different dopings, which the authors call ensuring a level playing field · outcome: held Quotation withheld because this record is not cleared for redistribution. choosing the Hall-derived bulk drift mobility over values fitted to current-voltage characteristics, on the stated ground that the fits carry more assumptions eliminates: a mobility chosen for convenience - AND THE AUTHORS BOUND THE STAKES rather than just declaring the choice: the critical field varies as the cube root of mobility, so the difference is small · outcome: held Quotation withheld because this record is not cleared for redistribution. flagging every table entry where the reported figure is a Hall mobility because the conversion factor to drift mobility is unknown for that material eliminates: two different quantities being read down one column as if they were the same measurement · outcome: held Quotation withheld because this record is not cleared for redistribution. labelling the maturity verdict INFORMAL and printing its three sub-criteria as separate table rows eliminates: a Mature/Immature label being read as a measurement - the inputs are beside it and a reader can disagree with the aggregation · outcome: held Quotation withheld because this record is not cleared for redistribution. compiling all 26 research opportunities/challenges into a single numbered list in Section 5, having stated each where it arose eliminates: an agenda that exists only as prose scattered through 42,000 words - the list is what makes the paper's claims about the field's gaps countable · outcome: held Quotation withheld because this record is not cleared for redistribution. warning against importing default physical mechanisms from silicon TCAD, or extrapolating them past the range the legacy materials support eliminates: a model that runs and converges on UWBG devices while computing silicon's physics - the paper then gives four specific ways it goes wrong · outcome: held Quotation withheld because this record is not cleared for redistribution. stating that standard characterisation assumptions, such as complete ionisation of dopants, may simply be untrue in these materials eliminates: experimental results being interpreted through assumptions that hold for narrower-gap semiconductors and not here · outcome: held Quotation withheld because this record is not cleared for redistribution. bandgap bandgap Quotation withheld because this record is not cleared for redistribution. critical electric field critical electric field Quotation withheld because this record is not cleared for redistribution. figures-of-merit figures-of-merit Quotation withheld because this record is not cleared for redistribution. Johnson figure of merit Johnson figure of merit Quotation withheld because this record is not cleared for redistribution. Baliga figure of merit Baliga figure of merit Quotation withheld because this record is not cleared for redistribution. maturity maturity Quotation withheld because this record is not cleared for redistribution. permittivity permittivity Quotation withheld because this record is not cleared for redistribution. compensation compensation Quotation withheld because this record is not cleared for redistribution. self-trapping self-trapping Quotation withheld because this record is not cleared for redistribution. dielectric constant dielectric constant Quotation withheld because this record is not cleared for redistribution. AlN AlN Quotation withheld because this record is not cleared for redistribution. diamond diamond Quotation withheld because this record is not cleared for redistribution. Ga2O3 Ga2O3 Quotation withheld because this record is not cleared for redistribution. GaN GaN Quotation withheld because this record is not cleared for redistribution. thermal conductivity thermal conductivity Quotation withheld because this record is not cleared for redistribution. acceptors acceptors Quotation withheld because this record is not cleared for redistribution. Hall mobilities Hall mobilities Quotation withheld because this record is not cleared for redistribution. bandgap → dielectric constant AN INVERSE TRADE-OFF ACROSS MATERIALS, NOT A CAUSAL EFFECT WITHIN ONE. 'any dielectric…'. Nothing is being done to a bandgap to lower a permittivity; it is a regularity over the space of available materials, and it is the reason a high-k ultrawide-bandgap gate insulator may not exist. 'causes ... reduces' would assert a mechanism the paper does not claim, and there is no 'trades-off-against' or 'correlates-with' type Quotation withheld because this record is not cleared for redistribution. AlN → acceptors A MATERIAL DEFINED HERE BY WHAT IT LACKS. 'no known acceptors…' - this is the single fact that keeps AlN out of bipolar devices, and every one of the eight relation types asserts that something HOLDS. The nearest available edge, 'causes ... eliminates', says AlN does something to acceptors; what is true is that none has been found Quotation withheld because this record is not cleared for redistribution. maturity → Ga2O3 A JUDGEMENT ABOUT A THING, MADE BY THE PAPER, AND LABELLED AS ONE. 'an informal assessment of the maturity level', resolving to Mature or Immature over three named sub-criteria. This is the relation type this corpus most needs and does not have - an ASSESSMENT edge, carrying who judged, on what criteria, and how firmly. Every one of our eight types says something about the world; none says something about an appraisal of the world Quotation withheld because this record is not cleared for redistribution. maturity assessment - THE PAPER CALLS ITS OWN JUDGEMENT LAYER INFORMAL, IN THE SENTENCE THAT INTRODUCES IT. It is a Mature/Immature verdict resting on three named sub-criteria - doping, volume/cost, substrate size and availability - each reported separately in its own table row. A score whose inputs are printed beside it is a score a reader can disagree with Quotation withheld because this record is not cleared for redistribution. figure of merit - THERE IS NO SINGLE ONE, AND WHICH APPLIES DEPENDS ON THE APPLICATION. The Baliga FOM is for low-frequency unipolar vertical power switches; for high-frequency high-power work the Johnson FOM is `more appropriate`. A material's rank changes with the question, so a table of FOMs is a table of conditional rankings Quotation withheld because this record is not cleared for redistribution. critical electric field, in the properties table - PROJECTED, NOT MEASURED. The values for AlN, Ga2O3, c-BN and diamond are calculated from GaN's experimentally derived field, and to compare them fairly the authors assume the SAME hypothetical device for all: a non-punched-through p+/n diode at a stated donor concentration. A number in a properties column that is a model output with a device baked into it Quotation withheld because this record is not cleared for redistribution. intrinsic carrier concentration, for an ultrawide-bandgap material - A NUMBER FERMI-DIRAC STATISTICS PRODUCE THAT IS NOT A QUANTITY. Around 10^-10 cm^-3, which the authors say `has no direct physical meaning`, because the statistics assume carriers come from the high-energy tail of a thermal distribution and that assumption fails several electron-volts from a band edge. It still enters TCAD models and destabilises them Quotation withheld because this record is not cleared for redistribution. immature - THE PAPER'S VERDICT ON EVERY UWBG MATERIAL INCLUDING THE ONES IT CHAMPIONS, and it is a comparative claim rather than a criticism: the field is placed where GaN and SiC were in the 1980s. Each of the three leading materials is poor on at least one of the three metrics, and the paper says which Quotation withheld because this record is not cleared for redistribution. material property, of the kind tabulated in reviews - NOT NECESSARILY STABLE. This paper re-derived them because values in common use were measured decades ago on material that would now be called highly defective. GaN turned out to have at least FIVE different permittivity sets in circulation, and AlN's bandgap was moved from 6.2 to 6.0 eV in 2010 by a better measurement. A constant in a table has a date and a method behind it Quotation withheld because this record is not cleared for redistribution. critiques → the material property values in common use across the semiconductor literature, including for GaN, many measured decades ago on highly defective material Quotation withheld because this record is not cleared for redistribution. distinguishes-from → the five different sets of GaN permittivity values circulating in the field note: the paper does not adjudicate between the five on theoretical grounds - it re-searches the literature and takes new AlN measurements, and calls establishing the best values an important outcome OF THIS STUDY rather than a side note Quotation withheld because this record is not cleared for redistribution. contradicts → the historical 6.2 eV bandgap of AlN, based on older absorption measurements note: not this paper's own finding - it reports the 2010 redefinition by high-resolution photoluminescence and adopts it. The relation is to the superseded value Quotation withheld because this record is not cleared for redistribution. qualifies → the Baliga figure of merit as the general measure of a power semiconductor's promise Quotation withheld because this record is not cleared for redistribution. builds-on → the community of UWBG researchers who met at the second Technical Exchange in April 2016, whose identified challenges this article articulates Quotation withheld because this record is not cleared for redistribution. bandgap —causes→ critical electric field bandgap critical electric field direction on the measure: increases · strength of the relation: 0.95 · note: and the exponent is what makes the whole field worth the trouble - squared here, cubed again into the Baliga FOM, sixth power overall Quotation withheld because this record is not cleared for redistribution. critical electric field —causes→ Baliga figure of merit critical electric field Baliga figure of merit direction on the measure: increases · strength of the relation: 0.95 Quotation withheld because this record is not cleared for redistribution. bandgap —causes→ compensation bandgap compensation direction on the measure: increases · strength of the relation: 0.9 · note: THE SAME VARIABLE THAT PAYS AT THE SIXTH POWER ALSO CAUSES THE PROBLEM. Two signed edges out of 'bandgap' in opposite directions of usefulness, from two sentences in the same paper Audit 2026-09-28: DOUBTFUL - read against its span and kept, but the span describes a tendency to compensation in UWBG semiconductors; 'bandgap increases' is loose. Quotation withheld because this record is not cleared for redistribution. Baliga figure of merit —contrasted-with→ Johnson figure of merit Baliga figure of merit Johnson figure of merit strength of the relation: 1 · note: and the contrast is conditional on the application, not on the material - which is why a single ranking of these semiconductors cannot be right Quotation withheld because this record is not cleared for redistribution. Ga2O3 —distinct-from→ diamond Ga2O3 diamond strength of the relation: 0.9 · note: distinguished here by WHICH metric each fails - Ga2O3 cannot be p-doped, diamond's substrates are small or defective Audit 2026-09-28: DOUBTFUL - read against its full span and kept, but the span lists different weaknesses of the UWBG materials; distinct-from is loose. Quotation withheld because this record is not cleared for redistribution. Especially important research opportunities and challenges are called out throughout the paper and compiled in Section 5. Quotation withheld because this record is not cleared for redistribution. UWBG semiconductors are at a development stage similar to GaN and SiC in the 1980s. Quotation withheld because this record is not cleared for redistribution. Each of the three UWBG materials is poor on at least one of three metrics: AlN and Ga2O3 cannot or can only poorly be p-doped, and diamond substrates are small or of poor quality. Quotation withheld because this record is not cleared for redistribution. GaN has at least FIVE different sets of permittivity values in common use, and establishing the best ones is claimed as an outcome of this study Quotation withheld because this record is not cleared for redistribution. AlN's bandgap was redefined from a historical 6.2 eV to about 6.0 eV in 2010 by high-resolution photoluminescence Quotation withheld because this record is not cleared for redistribution. The critical breakdown fields of AlN, BN, Ga2O3 and diamond are yet to be accurately determined. Quotation withheld because this record is not cleared for redistribution. ultrawide-bandgap semiconductor = one whose bandgap is significantly wider than GaN's 3.4 eV UWBG semiconductor Quotation withheld because this record is not cleared for redistribution. Baliga figure of merit = breakdown voltage squared over specific on-resistance, for a low-frequency unipolar vertical power switch Baliga figure of merit Quotation withheld because this record is not cleared for redistribution. and the Johnson figure of merit is the appropriate one instead for high-frequency high-power applications Johnson figure of merit Quotation withheld because this record is not cleared for redistribution. maturity assessment = an informal verdict of Mature or Immature over three named criteria: doping, volume/cost, and substrate size and availability maturity assessment Quotation withheld because this record is not cleared for redistribution. moving from GaN to AlN raises the bandgap by 1.8x and the Baliga figure of merit by about 34x - the sixth power of that ratio Quotation withheld because this record is not cleared for redistribution. breakdown voltage scales as the 4th power of bandgap, since the critical field goes as its square Quotation withheld because this record is not cleared for redistribution. the choice between two mobility values matters little because the critical field varies only as the CUBE ROOT of mobility - a sensitivity stated rather than assumed Quotation withheld because this record is not cleared for redistribution. Fermi-Dirac statistics give these materials an intrinsic carrier concentration around 10^-10 per cubic centimetre, which the authors say has no direct physical meaning Quotation withheld because this record is not cleared for redistribution. replacing a 96%-efficient silicon power converter with a 98%-efficient wide-bandgap one halves the waste heat Quotation withheld because this record is not cleared for redistribution.

The full text of this article is not redistributable, so it is not included here. The evidence sentences are withheld with it; the tag layer and its grounding status remain. The article itself is at https://doi.org/10.1002/aelm.201600501.