Dispersion models describing coupled systems in doped crystals. II. Infrared phonon excitations in GaAs single crystals and phonon-plasmon coupling in Zn-doped GaAs

Beata Hroncova; Daniel Franta; Mihai-George Muresan; Stefan Zollner

Dispersion models describing coupled systems in doped crystals. II. Infrared phonon excitations in GaAs single crystals and phonon-plasmon coupling in Zn-doped GaAs

T0 review · 3 major / 7 minor · reviewed 2026-07-30 · grok-4.5

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A generalized IR dispersion model attributes the TO-phonon asymmetry in heavily Zn-doped GaAs to Fano interference with free holes, not a true resonance shift, and yields consistent temperature-dependent optical constants.

desk verdict Solid applied IR metrology paper: usable GaAs optical constants and a credible net Fano picture, with an over-resolved single/multi-phonon decomposition that the data do not uniquely pin down. read the letter →

arxiv 2607.23868 v1 pith:RSJU2TT6 submitted 2026-07-26 physics.optics cond-mat.mtrl-sci

Daniel Franta , Beata Hroncova , Mihai-George Muresan , Stefan Zollner This is my paper · ORCID

classification physics.opticscond-mat.mtrl-sci

keywords GaAsphonon-plasmoncouplinginfrareddispersionmodelFanoresonanceZn-dopedmultiphononabsorptionopticalconstantsspectroscopicellipsometry

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper validates a generalized dispersion framework for coupled phonon and free-carrier excitations on intrinsic and Zn-doped GaAs. Simultaneous fits of reflectance, transmittance, and ellipsometry from 300 to 440 K use Gaussian-broadened multi-peak single-phonon responses and triangle–delta–triangle multiphonon bands, plus two Drude terms for light and heavy holes. The model separates lattice and carrier contributions while preserving sum rules and volume expansion, and it reproduces how phonon–plasmon coupling evolves with doping and temperature. In heavily doped material the TO peak becomes strongly asymmetric; the authors show this is Fano-type interference with the free-hole continuum rather than a genuine shift of the TO resonance energy. The resulting far- and mid-IR optical constants are physically consistent and supply a quantitative tool for thermo-optical modeling of GaAs photonic structures.

What carries the argument

The generalized coupled dispersion model: complex transition strengths (N + iM) on Gaussian-broadened phonon peaks and on triangle–delta–triangle multiphonon bands, linearly interpolated with doping, plus two classical Drude terms for light and heavy holes, all constrained by f-sum rules and a shared Bose–Einstein temperature dependence.

What would settle it

Extend the same simultaneous reflectance/transmittance/ellipsometry fits to a much wider temperature range (or to isotopically pure GaAs); if the extracted single- versus multi-phonon decomposition or the doping-linear coupling parameters change substantially, or if independent low-frequency conductivity disagrees with the optical static conductivity, the Fano attribution and partial strengths are not unique.

Watch

Extended reading notes

Core claim

Simultaneous multi-temperature IR fits with Gaussian-broadened single-phonon distributions, TDT multiphonon bands, and dual-hole Drude terms produce self-consistent GaAs optical constants and demonstrate that the pronounced TO-peak asymmetry in heavily Zn-doped GaAs is Fano interference between lattice vibrations and the free-hole continuum, not a true change of the TO resonance frequency.

Load-bearing premise

The split between single- and multi-phonon strengths, and the linear doping scaling of the coupling amplitudes, are assumed reliable even though the temperature window is only 300–440 K and multiphonon–single-phonon coupling in the undoped crystal is neglected.

Editorial extensions

If this is right

Temperature-dependent far- and mid-IR optical constants of GaAs (300–440 K) become available for thermo-optical device simulation.
Phonon–plasmon coupling in p-GaAs can be modeled without invoking an artificial TO frequency shift.
The same coupled bound–free framework can be applied to other doped zinc-blende semiconductors and crystalline coatings.
Static conductivity emerges from the coupled response rather than being inserted as an independent parameter.
GaAs wafers can be treated as better-characterized IR reference substrates once the model is further refined.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

A wider temperature campaign would likely tighten the single/multi-phonon separation and reduce the strong parameter correlations noted in the reststrahlen region.
Isotopically engineered GaAs would cleanly test how much of the residual TO asymmetry is mass-disorder versus multiphonon coupling.
The same N+iM coupling construction should transfer directly to n-type GaAs or other III–Vs once light/heavy electron (or hole) Drude terms are adjusted.
Publishing the optical constants in a public refractive-index database would immediately enable thermo-optic simulations of GaAs/AlGaAs supermirrors and pump diodes.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Desk Editor's Note

read the letter

The useful core here is a joint multi-instrument fit (R, T, ellipsometry, plus literature n and α) that delivers temperature-dependent far/mid-IR optical constants for undoped and Zn-doped GaAs from 300–440 K, with sum-rule-normalized responses and a clear demonstration that plain Lorentz/Voigt tails fail once the wafer is partially transparent. That practical constant set, and the honest comparison to Skauli and classical reststrahlen fits, is what most readers will actually use.

What is new is not the existence of multiphonon structure or phonon–plasmon coupling in p-GaAs—those are known—but the application of their coupled Gaussian + TDT + two-Drude framework (Part I / prior Si work) with explicit complex amplitudes, shared thermal factors, and a doping-mix parameter. They make a fair physical case that the heavy-doping TO asymmetry is interference with the hole continuum rather than a real ~1% lattice eigenfrequency shift: only ~10⁻⁴ of sites are Zn, and the total doped-minus-undoped phonon response is well constrained. Static conductivities also land in a sensible range relative to the wafer specs.

The soft spot is real but localized. The channel-resolved story—single-phonon and multiphonon Fano terms that “oppose and partially cancel”—sits on TDT blocks inside the reststrahlen band whose fitted N and M have uncertainties orders of magnitude larger than the values, and on a single/multi split that rests only on a narrow temperature window plus linear doping interpolation. The paper itself flags over-parameterization and opacity-limited errors there. So trust the net optical constants and the qualitative Fano attribution; treat the detailed M_ph vs M_mph decomposition as illustrative, not unique. Free-hole density held T-independent and neglected multiphonon–single-phonon coupling in the undoped crystal are ordinary modeling choices at this stage, not hidden landmines.

Math and citation pattern look fine: Kramers–Kronig structure, f-sum rules, and the usual GaAs/IR literature are in place; self-citation is to their own framework, which is expected for a Part II. Data path to RefractiveIndex.info is the right move; full spectra/fit files would strengthen uniqueness claims.

This is for people who need consistent GaAs IR constants for thermo-optical or photonic modeling, and for anyone tracking practical coupled-dispersion fitting. It deserves a serious referee, not a desk reject. I would engage with the constants and the net coupling result; I would not lean hard on the fine-grained single/multi Fano split without wider T or fewer multiphonon DOF.

Referee Report

3 major / 7 minor

Summary. The manuscript presents a combined ellipsometric/reflectance/transmittance study (300–440 K) of one undoped and two Zn-doped GaAs wafers, analyzed within the authors' previously developed coupled-system dispersion framework. The dielectric function is decomposed into Sellmeier electronic poles, six Gaussian-broadened single-phonon sub-modes with complex (N+iM) coupling amplitudes, 39 triangle–delta–triangle (TDT) multiphonon elements, and two Drude terms for light/heavy holes, all tied together by f-sum-rule normalization and a shared Bose–Einstein temperature scaling. The principal physical claim is that the pronounced asymmetry of the TO-phonon peak in heavily doped GaAs is a Fano-type interference between phonon excitations and the free-hole continuum, not a genuine shift of the TO resonance; single-phonon and multiphonon coupling contributions are said to oppose and partially cancel. A by-product is a temperature-dependent set of GaAs infrared optical constants claimed to be more internally consistent than existing literature data.

Significance. If the results hold, the paper delivers (i) a validated, temperature-dependent set of GaAs infrared optical constants (300–440 K) with external anchors to independent literature n and α data, filling a real gap relative to existing datasets; (ii) a demonstration that Lorentzian-lineshape models fail on transparent-region transmittance where the Gaussian/TDT construction succeeds (Fig. 13); (iii) sum-rule-normalized response functions, making the temperature scaling falsifiable and the f-sum checks (Eqs. 2, 12, 13, 25, 35) internally verifiable; and (iv) a concrete, physically motivated interpretation of the heavily doped TO-peak asymmetry as phonon–free-carrier (Fano) interference, with consistency cross-checks (p_ph^#2 ≈ 5.7% vs N_fc strength ratios 5.47%/6.17%, Table VIII; E_LO from Eq. (36) matching the loss-function maximum). The authors are unusually candid about over-parameterization and ambiguity, and the constants are promised for refractiveindex.info, which raises practical impact. The main caveats are that the channel-resolved decomposition is weakly constrained and the no-shift conclusion is partly built into the parameterization.

major comments (3)

[Sec. IV C, Eqs. (15)–(16), discussion around Eq. (19)] The central interpretive claim — that the apparent TO-peak shift in sample #3 is a Fano-type asymmetry rather than a true resonance shift — is partly enforced by construction. Equations (15)–(16) constrain all three samples to share the same E_ph,j (and Eq. (17) applies one common temperature scaling), so the model cannot represent a genuinely shifted TO mode at all. The authors themselves state that a conventional Drude–Lorentz model with a ~1% shifted TO peak fits the reflectance 'reasonably well,' so the data do not discriminate between the two readings. The argument against the shifted reading (only ~10^-4 of lattice sites substituted, making a 1% eigenfrequency shift implausible) is a reasonable physical prior, but it should be labeled as such, and the no-shift conclusion should be stated as conditional on the shared-energy parameterization. As written, Sec. IV C presents the nois a
[Sec. IV D, Table VII (j=13–18); Figs. 10(b), 11(b)] The channel-resolved claim that single-phonon and multi-phonon Fano contributions 'oppose each other and partially cancel' (Figs. 10(b) vs 11(b)) rests on parameters that are statistically undetermined precisely where the cancellation occurs. In Table VII, the TDT elements located in/at the edge of the reststrahlen band (j=13–18, E ≈ 258–294 cm^-1) have uncertainties orders of magnitude larger than the fitted values — e.g., M_mph,j entries −14941(114566) and 10768(110865)×10^-7 eV² for j=14,15, and N-type entries with errors ~10²–10⁴× the value. The sign and magnitude of the multiphonon coupling term at the TO frequency is therefore not constrained by the fit. The authors partially acknowledge this ('if the results are to be believed') and offer the total interaction (Figs. 11(c), 12(b)) as the robust quantity, which is the right instinct. However, the abstract and Sec. IV D still still
[Secs. III, IV D (over-parameterization discussion); Conclusion; Fig. 4] The model admits to strong over-parameterization (39 TDT blocks, each with N_d,l,r and M, plus six complex-amplitude phonon peaks), and the stated uncertainties are local-curvature regression errors that the authors correctly caution are 'not a full metrological uncertainty analysis.' Given this, the Conclusion's claim of 'a level of accuracy not previously available' is not supported by an uncertainty budget. Fig. 4 shows regression-based uncertainty bands only. The claim should be tempered, or supported by a propagated/comparison-based accuracy estimate in the transparent region (e.g., against the Marple/Hambleton/Skauli anchor data, which the authors estimate at ~10^-3 absolute). This matters because the deliverable most readers will use is the optical-constant set, and its accuracy statement is the load-bearing claim for that use.

minor comments (7)

[Fig. 10 caption] The caption of Fig. 10 labels both panels '(a)'; the second should be (b). The text references Fig. 10(b), so this is presumably a typo.
[Conclusion, first and last paragraphs] Typo: 'optical+ constants'; and 'In this work, it was applied for the first time a recently published approach' is grammatically broken.
[References] Ref. 24 (companion Paper I) has an incomplete citation ('J. Appl. Phys. 139, (2026)') — volume/article number should be completed at proof stage.
[Sec. IV C (native layer assumption); Sec. II] The native surface layer is modeled as 3.31 nm amorphous gallium oxide with identical optical constants for all three samples, including the single-side-polished, heavily doped #3. One sentence justifying that the backside (unpolished) surface of #3 does not require a separate roughness/interface layer would strengthen the ellipsometric analysis.
[Table VII; Eq. (27)] In Table VII several t_j values converge to 0.00 with uncertainties exceeding the physically meaningful range (e.g., t_5=0.00(7), t_13=1.0(6), t_14=0(3)); since t_j is bounded in [0,1] by construction, it would be cleaner to state whether bounds were imposed in the regression and, for the j with t_j indeterminate, whether fixing t_j=0 changes the result.
[Sec. IV C, Eq. (23) vs Sec. IV D, Eq. (36)] The two E_LO values (Eq. (23), 36.003 meV vs Eq. (36), 36.1297 meV) and the explanation of why the total phonon strength must be used in the HO formula are physically reasonable but compressed; a short clarifying remark that Eq. (23) fails because N_ph alone no longer represents the full lattice polarizability in the coupled model would help readers.
[Data Availability] The data-availability statement promises deposition in refractiveindex.info; given that the tabulated optical constants are the most reusable output, a supplementary table of n,k(E,T) at representative temperatures (or a DOI for the deposited dataset) would substantially increase the paper's utility.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for a careful, substantive report and for the recommendation of minor revision. The referee correctly identifies three places where the manuscript's interpretive claims outrun what the data or the uncertainty analysis strictly support: (i) the "no true TO shift" conclusion is conditional on the shared-energy parameterization of Eqs. (15)-(17); (ii) the channel-resolved partial-cancellation statement between single- and multiphonon Fano terms rests on TDT parameters whose regression uncertainties, in the reststrahlen region, vastly exceed their fitted values; (iii) the Conclusion's accuracy claim lacks a comparison-based uncertainty budget. We accept all three points and have revised accordingly: the no-shift statement is now explicitly labeled as conditional on the parameterization and supported by the stated physical prior (10^-4 substitution level) plus the internal cross-checks the referee noted; the partial-cancellation wording is demoted from the abstract and qualified in Sec. IV D, with the total interaction (Figs. 11c, 12b) presented as the robust observable; and the Conclusion's accuracy claim is tempered and tied to the ~10^-3 anchor-based estimate in the transparent region. We respond to each point in detail below.

read point-by-point responses

Referee: [Sec. IV C] The no-shift / Fano-asymmetry claim is partly enforced by construction: Eqs. (15)-(17) constrain all samples to share E_ph,j and one temperature scaling, so a genuinely shifted TO mode cannot be represented. A Drude-Lorentz fit with a ~1% shifted TO peak also fits reasonably well. The 10^-4 substitution argument is a physical prior and should be labeled as such; the no-shift conclusion should be stated as conditional on the shared-energy parameterization.

Authors: The referee is correct, and we have revised Sec. IV C accordingly. The shared-energy constraint is a modeling choice, not a fitting outcome, and we now state explicitly that the no-shift conclusion is conditional on this parameterization and on the physical prior that motivates it. The revised text reads (in essence): within our coupled-system framework, in which the lattice eigenfrequencies are assumed independent of dopant concentration (justified a priori by the ~10^-4 Zn substitution fraction, which makes a 1% eigenfrequency shift implausible), the apparent shift is naturally explained as Fano-type asymmetry. We also note, however, that the data are not entirely mute on the question: (a) the consistency of p_ph^#2 = 5.7% with the Drude strength ratios 5.47%/6.17% (Table VIII) shows the coupling amplitude scales with free-carrier density as the coupled reading predicts, whereas a genuine eigenfrequency shift would have no reason to track carrier density in this way; (b) E_LO from Eq. (36), built from the unshifted E_TO and total phonon strength, reproduces the loss-function maximum. These are consistency checks rather than proof, and the revised text presents them as such. revision: yes
Referee: [Sec. IV D, Table VII j=13-18] The claim that single-phonon and multiphonon Fano contributions oppose and partially cancel rests on TDT parameters that are statistically undetermined exactly where the cancellation occurs (M entries with uncertainties 10^2-10^4 times the values). The total interaction is the right robust quantity, but the abstract and Sec. IV D still assert the channel-resolved cancellation.

Authors: We accept this point. The individual N and M parameters of TDT elements j=13-18 are effectively unconstrained by the fit, so neither the sign nor the magnitude of the multiphonon coupling term at the TO frequency is data-determined, and a quantitative 'partial cancellation' statement cannot be defended at the channel level. In revision we have (i) removed the channel-resolved cancellation claim from the abstract; (ii) reworded the relevant passage in Sec. IV D so that the opposition of the two contributions is presented as a qualitative feature of the model decomposition, explicitly qualified by the large uncertainties and by the intrinsic ambiguity of the temperature-based separation (which the manuscript already acknowledges); and (iii) made Figs. 11(c) and 12(b) — the total phonon-free-carrier interaction, which is robust because the sum is insensitive to the correlated redistribution among TDT elements — the basis of the phonon-plasmon coupling interpretation. We also added a sentence noting that extending the temperature range is the concrete route to constraining this decomposition, as it changes the single-/multiphonon statistical factors differentially. revision: yes
Referee: [Secs. III, IV D, Conclusion, Fig. 4] Uncertainties are local-curvature regression errors, not a metrological budget; Fig. 4 shows regression bands only. The Conclusion's claim of 'a level of accuracy not previously available' is unsupported. Temper it or provide a comparison-based accuracy estimate in the transparent region against the Marple/Hambleton/Skauli anchors (~10^-3).

Authors: We agree and have tempered the claim. The quoted parameter errors are local-curvature regression uncertainties, as stated in Sec. III, and Fig. 4 reflects only those; we do not have a full metrological uncertainty budget, and we now say so explicitly in the Conclusion. The revised Conclusion replaces the unqualified accuracy statement with a comparison-based one: in the transparent region the model reproduces the Marple MDM, Hambleton, and Skauli anchor data to within approximately 10^-3 absolute in n (consistent with the declared accuracy of those datasets, Fig. 2(b)), and we state this level of agreement as the accuracy estimate for the optical constants in the transparency range. We also note that the strongest external evidence for improvement over existing datasets is precisely the internal consistency across the far-IR phonon region and the transparent region — a combination existing tabulations do not provide — rather than a claim of superior absolute accuracy everywhere. In the reststrahlen region, where the samples are opaque and the fit is least constraining, we now explicitly caution that the constants carry substantially larger uncertainty (as Fig. 4 shows). revision: yes

Circularity Check

2 steps flagged · score 3.0 of 10

Ordinary inverse modeling of IR data; mild self-definitional circularity only in the shared-E_ph construction that forces the 'no TO shift' conclusion.

self definitional [Sec. IV C, Eqs. (15)–(16), Table VI; accompanying text on Figs. 8–10]
"coupling with free carriers does not change the resonance frequency, but alters the phase with which the vibrational states respond to the mean field, which manifests as asymmetry in the frequency-domain absorption peak. ... all three absorption peaks correspond to the same excitation energies (resonance frequencies). ... M#2_ph,j = (1 − p#2_ph) M#1_ph,j + p#2_ph M#3_ph,j ... χ̂#2_ph(E,T) = (1 − p#2_ph) χ̂#1_ph(E,T) + p#2_ph χ̂#3_ph(E,T)."

N_ph,j and E_ph,j are shared across undoped and doped samples; only the coupling amplitudes M_ph,j are allowed to vary with Zn concentration via a single scalar p#2_ph. Inside this parameterization a true doping-induced shift of the TO eigenfrequency is unrepresentable, so the conclusion that coupling leaves resonance frequencies unchanged is guaranteed by construction. The paper's physical argument (only ~10^{-4} sites substituted) is independent but does not remove the structural exclusion of the alternative; the authors also concede an uncoupled Drude–Lorentz fit with an apparent ~1% red-shift works 'reasonably well.'
self citation load bearing [Introduction / Sec. IV opening; Refs. 24, 40 (and 23) for the coupled-response and TDT formalism]
"This work builds on our previous development... We present a temperature-dependent infrared study of intrinsic and Zn-doped GaAs analyzed within a recently developed dispersion framework for coupled phonon–carrier systems. ... As shown previously, this model can produce an unphysical response function... To address this deficiency, constraints... With these constraints, the Kukharskii model corresponds to the response of classical coupled particles.24"

The entire bound–free coupling construction (complex transition strengths, sum-rule-safe asymptotics, Gaussian-broadened multi-peak phonons, TDT multiphonon bands, interaction-term extraction χ_C = χ_doped − χ_undoped) is taken from the authors' own Paper I and related prior papers. That is normal framework reuse and is not uniqueness-forced against external mathematics, but the interpretive claim that the observed asymmetry 'is' Fano-type phonon–plasmon coupling in their sense does lean on that self-developed apparatus rather than on an independent external derivation. External optical data still fix the total response, so this is only partially load-bearing.

full rationale

The paper's primary products—temperature-dependent optical constants and a net Fano-type TO asymmetry in heavily Zn-doped GaAs—are obtained by simultaneous nonlinear least-squares fitting of reflectance, transmittance, and ellipsometry plus external anchors (Novikova/Glazov α, Marple/Hambleton/Skauli n, manufacturer doping/resistivity ranges). That is standard inverse modeling, not definitional circularity: the data constrain the total dielectric function, and static conductivities and doping ratios provide independent cross-checks. The coupling formalism, Gaussian complex amplitudes, TDT multiphonon bands, and sum-rule structure are imported from the authors' Paper I and related prior work; this is self-citation of a modeling framework, not a uniqueness theorem that forbids alternatives or that alone produces the GaAs numbers. The one clear circular step is structural: Eqs. (15)–(16) and Table VI force all three samples to share the same E_ph,j and N_ph,j, varying only the coupling amplitudes M_ph,j with doping, so the claim that 'coupling does not change the resonance frequency' is true by construction inside the model (a genuine eigenfrequency shift cannot be represented). The paper itself notes that an uncoupled Drude–Lorentz model with an apparent ~1% TO red-shift also fits 'reasonably well,' and then prefers the no-shift reading on physical grounds (only ~10^{-4} lattice sites substituted). That physical prior is independent but not airtight; the model parameterization still excludes the alternative a priori. Channel-resolved single- vs multiphonon Fano decomposition (Figs. 10b vs 11b) is underconstrained in the reststrahlen region (Table VII, j≈13–18) and relies on a limited 300–440 K lever arm—the paper admits the separation is 'intrinsically ambiguous'—but that is overparameterization/weak identification, not a derivation that reduces to its inputs by identity. Overall circularity is therefore mild and localized; score 3.

Assumptions & free parameters 8 free parameters · 7 assumptions · 2 invented entities

The central claim rests on a large empirical dispersion ansatz (many Gaussian TO peaks, 39 TDT multiphonon units, two Drude holes, Bose–Einstein thermal factors, linear doping mix of coupling) fitted to the authors’ spectra plus selected literature n and α. Physics inputs (zinc-blende one IR-active TO, light/heavy holes, f-sum rules, Fano interference picture) are standard; the TDT multiphonon representation and the specific coupling parameterization are framework choices from the authors’ series.

free parameters (8)

Six single-phonon N_ph,j, E_ph,j, B_ph,j and M_ph,j (per doping endpoint) = N_ph total ≈ 3.2848e-3 eV^2; E_TO mean ≈ 33.225 meV at 300 K
Shape and asymmetry of the TO response are carried by these fitted amplitudes/widths/energies (Table VI).
p_ph^#2 doping-mix parameter for phonon coupling = 0.057(7)
Linear interpolation of coupling between undoped and heavily doped endpoints (Eqs. 15–16).
39 multiphonon TDT blocks (N_d,l,r, E_j, t_j, M_j) plus B_mph, E_40 = N_mph^300K ≈ 0.1559e-3 eV^2; B_mph ≈ 3.63 cm^{-1}
Ad hoc multiphonon band shape and two- vs three-phonon mix (Table VII); highly correlated.
p_mph^#2 multiphonon doping-mix parameter = 0.0243(12)
Analogous linear mix for multiphonon–carrier coupling (Eq. 34).
Two Drude terms per doped sample (N_fc,j, Γ_fc,j) and β_fc = e.g. #3: N≈99.8e-3 and 28.0e-3 eV^2
Light- and heavy-hole continuum and T-dependent scattering (Table VIII).
Three Sellmeier electronic poles N_el,j, E_el,j, ε0K_el,j and shared Θ = ε1≈10.873 at RT in HO reduction
Transparent-region index and far-IR ε_∞ background (Table IV).
Three-term thermal expansion e_j, Θ_j = Θ_j ≈ 92, 346, 2203 K
Volume factor E(T) multiplies all susceptibilities (Table III, Eq. 5).
Wafer thickness, thickness variation, native-oxide thickness, spectral resolution = d_s,1=0.628061(14) mm; oxide 3.31(2) nm
Instrument/sample nuisance parameters in the joint fit.

assumptions (7)

domain assumption Dielectric function decomposes into electronic + single-phonon + multiphonon + free-carrier susceptibilities multiplied by a scalar volume expansion factor E(T).
Eq. (1); standard additive response plus thermal-density correction used throughout.
domain assumption f-sum rules constrain integrated transition strengths; single-phonon N_ph is T-independent aside from E(T); multiphonon strengths follow Bose factors.
Eqs. (2), (13), (26)–(30); from the authors’ prior dispersion program and semiconductor textbook statistics.
ad hoc to paper Coupling between subsystems is represented by complex (N+iM) amplitudes / antisymmetric interaction spectra without shifting bare resonance frequencies.
Core of Part I framework applied in Eqs. (10), (18), (32); justifies Fano asymmetry without true TO shift.
domain assumption Free-carrier density is constant over 300–440 K; only Drude broadenings vary with a shared Bose-like factor; light and heavy holes are uncoupled Drude terms.
Sec. IV E, Eq. (38); stated as adequate only in the measured window.
domain assumption Electronic interband contribution is independent of Zn doping at the studied concentrations.
Sec. IV B; dopant density ≪ host atomic density.
ad hoc to paper Native surface layer may be modeled as 3.3 nm amorphous gallium oxide with fixed optical constants for all samples.
Sec. IV C; affects IR R/T at the percent level.
standard math Kramers–Kronig consistency via Dawson/Faddeeva/error-function constructions for Gaussian and TDT broadenings.
Appendices A–B; standard KK partners of the chosen line shapes.

invented entities (2)

Triangle–delta–triangle (TDT) multiphonon dipole distribution elements
purpose: Flexible, sum-rule-normalizable band pieces for multiphonon absorption without a full van Hove critical-point model.
Introduced in the authors’ recent framework and detailed in Appendix B; 39 elements are phenomenological, not derived from GaAs phonon DOS.
Six discrete Gaussian-broadened TO sub-modes with isotope/coupling-motivated complex amplitudes
purpose: Reproduce slight TO asymmetry and non-Lorentzian wings in undoped and doped GaAs.
Not mapped one-to-one to measured isotope modes; binomial isotope attempt failed (Sec. IV C).

how reviews work

0 comments

Cite this review

Pith. "Pith review of Dispersion models describing coupled systems in doped crystals. II. Infrared phonon excitations in GaAs single crystals and phonon-plasmon coupling in Zn-doped GaAs." pith.science (2026). https://pith.science/paper/RSJU2TT6

@misc{pith2026260723868,
  author       = {Pith},
  title        = {Pith review of: Dispersion models describing coupled systems in doped crystals. II. Infrared phonon excitations in GaAs single crystals and phonon-plasmon coupling in Zn-doped GaAs},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RSJU2TT6}},
  note         = {Machine review of arXiv:2607.23868}
}

read the original abstract

We present a temperature-dependent infrared study of intrinsic and Zn-doped GaAs within a generalized dispersion framework describing coupled phonon and free-carrier excitations. Reflectance, transmittance, and ellipsometric data measured from 300 to 440 K are fitted simultaneously using Gaussian-broadened distributions for single-phonon modes and triangle-delta-triangle representations for multiphonon absorption bands. The formalism provides a self-consistent description of lattice and carrier contributions to the dielectric response and reproduces the evolution of phonon-plasmon coupling with doping and temperature. In heavily doped samples, a pronounced Fano-type asymmetry of the TO-phonon resonance is observed and attributed to interference between phonon excitations and the Drude background of light and heavy holes. The resulting optical constants remain physically consistent across the far- and mid-infrared range and demonstrate the applicability of the framework to coupled bound and free excitations in doped semiconductors. The approach provides a quantitative tool for modeling infrared optical properties of GaAs and related photonic materials.

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Pith tools

Reviewed July 30, 2026 · model on record in the stance chip above.