Rotonic Plasmons: Dual Physical Perspectives

• Rotonic plasmons are quasi-particle excitations exhibiting a dual nature: in molecular systems as hybrid exciton–plasmon polaritons with explicit rotational and vibrational dynamics, and in semiconductors as collective plasmons with a parabolic band edge and finite effective mass.
• In the molecular context, enhanced Rabi splitting and vibrational spectral features reveal the impact of ro-vibrational coupling and pre-alignment on near-field optical responses.
• In semiconductor plasmonic crystals, the parabolic dispersion enables voltage-tunable RF-to-THz frequency conversion through parametric excitation under gate-voltage modulation.

“Rotonic plasmons” denotes two distinct constructions in the supplied arXiv literature. In molecular plasmonics, the term is used for hybrid exciton–plasmon polaritons in which the usual two-level molecular emitter is replaced by a fully quantum ro-vibrational wave packet coupled to surface plasmon-polaritons in a slit-array geometry (Sukharev et al., 2016). In semiconductor plasmonic crystals, closely related language appears as “rotonic plasmons,” referring to collective plasma excitations in periodically gated and ungated field-effect-transistor structures whose band-edge spectrum is parabolic and characterized by a finite effective mass, with dynamics governed by a generalized Mathieu equation under gate-voltage pumping (Aizin et al., 15 Apr 2026). The shared terminology therefore spans two different physical settings: one centered on ro-vibrational strong coupling and spectroscopic structure, the other on plasmonic-band engineering, parametric instability, and RF-to-THz conversion.

1. Terminological scope

The supplied sources use nearly the same label for different objects. One source develops molecular exciton–plasmon materials with ro-vibrational molecular structure and concludes that “Rotonic Plasmons” are hybrid polaritons formed when a 2-level exciton description is replaced by a quantum ro-vibrational wave packet (Sukharev et al., 2016). The other source introduces “rotonic plasmons” for plasmons in plasmonic crystals with a parabolic, roton-like dispersion law and finite effective mass (Aizin et al., 15 Apr 2026).

Usage in supplied literature	Physical platform	Defining feature
“Rotonic Plasmons”	Molecules on a periodic array of slits	Ro-vibrational molecular states under strong coupling with surface plasmon-polaritons
“rotonic plasmons”	Gated/ungated semiconductor plasmonic crystal	Parabolic band-edge plasmon branch with finite effective mass

This dual usage matters because the same label does not imply a common microscopic model. In the molecular setting, the central extension is beyond the two-level-emitter approximation. In the semiconductor setting, the central extension is beyond the linear or square-root dispersion of gated or ungated plasmons.

2. Molecular hybrid-polariton formulation

In the molecular-plasmonic construction, the total Hamiltonian is written as

$$\hat H=\hat H_{\rm EM}+\sum_i \hat H_{\rm mol}^{(i)}+\hat H_{\rm int}.$$

The electromagnetic sector in Coulomb gauge is

$$\hat H_{\rm EM}=\int d^3r\left[\frac12 \varepsilon_0 \hat E^2(r)+\frac12 \mu_0 \hat B^2(r)\right].$$

Each diatomic molecule is modeled by a ro-vibrational Hamiltonian

$$\hat H_{\rm mol}^{(i)} = -\frac{\hbar^2}{2\mu}\frac{\partial^2}{\partial R_i^2} +\frac{\hat J_i^2}{2\mu R_i^2} +V_g(R_i)\,|g\rangle\langle g|_i +V_e(R_i)\,|e\rangle\langle e|_i,$$

where $R_i$ is the internuclear coordinate, $\mu$ is the reduced mass, $\hat J_i^2/(2\mu R_i^2)$ is the rotational kinetic energy in Hund’s case (b), and $V_g(R)$ and $V_e(R)$ are Morse-type potentials. The dipole interaction is

$$\hat H_{\rm int} = -\sum_i \mu(R_i)\cdot \hat E(r_i,t)\left(|e\rangle\langle g|_i+|g\rangle\langle e|_i\right),$$

with the transition dipole assumed parallel to the molecular axis (Sukharev et al., 2016).

The molecular wave function is expanded as

$$\Psi(r,R,t)=\chi_g(R,t)\Phi_g(r;R)+\chi_e(R,t)\Phi_e(r;R),$$

and the nuclear amplitudes are further expanded in a truncated basis of rotational Wigner $$\hat H_{\rm EM}=\int d^3r\left[\frac12 \varepsilon_0 \hat E^2(r)+\frac12 \mu_0 \hat B^2(r)\right].$$0-matrices, for example

$$\hat H_{\rm EM}=\int d^3r\left[\frac12 \varepsilon_0 \hat E^2(r)+\frac12 \mu_0 \hat B^2(r)\right].$$1

The coupled nuclear wave packets are propagated on the two potential-energy surfaces via a short-time split-operator scheme. Pure dephasing $$\hat H_{\rm EM}=\int d^3r\left[\frac12 \varepsilon_0 \hat E^2(r)+\frac12 \mu_0 \hat B^2(r)\right].$$2 and nonradiative population decay $$\hat H_{\rm EM}=\int d^3r\left[\frac12 \varepsilon_0 \hat E^2(r)+\frac12 \mu_0 \hat B^2(r)\right].$$3 are included as time-dependent imaginary potentials in a non-Hermitian Schrödinger treatment which, in the weak-field limit, is formally equivalent to the Liouville-von Neumann approach while propagating amplitudes rather than a density matrix (Sukharev et al., 2016).

The physical system used to demonstrate the model is a thin layer of molecules placed on top of a periodic array of slits. The model therefore extends conventional exciton–plasmon materials by incorporating rotational alignment and vibrational dynamics explicitly, rather than representing each molecule as a single transition frequency.

3. Spectral manifestations and alignment sensitivity

At normal incidence and with period fixed at $$\hat H_{\rm EM}=\int d^3r\left[\frac12 \varepsilon_0 \hat E^2(r)+\frac12 \mu_0 \hat B^2(r)\right].$$4, the calculated transmission spectra distinguish three molecular descriptions. For two-level molecules, the Rabi splitting is $$\hat H_{\rm EM}=\int d^3r\left[\frac12 \varepsilon_0 \hat E^2(r)+\frac12 \mu_0 \hat B^2(r)\right].$$5. For the bound–bound ro-vibrational model it is $$\hat H_{\rm EM}=\int d^3r\left[\frac12 \varepsilon_0 \hat E^2(r)+\frac12 \mu_0 \hat B^2(r)\right].$$6, and vibrational peaks appear. For the bound–continuum model with a dissociative excited state, it is $$\hat H_{\rm EM}=\int d^3r\left[\frac12 \varepsilon_0 \hat E^2(r)+\frac12 \mu_0 \hat B^2(r)\right].$$7. The increased splitting in the bound–continuum case is attributed to larger Franck–Condon overlap integrated over the dissociative continuum (Sukharev et al., 2016).

The bound–bound model exhibits a clear series of narrow dips and peaks between $$\hat H_{\rm EM}=\int d^3r\left[\frac12 \varepsilon_0 \hat E^2(r)+\frac12 \mu_0 \hat B^2(r)\right].$$8 and $$\hat H_{\rm EM}=\int d^3r\left[\frac12 \varepsilon_0 \hat E^2(r)+\frac12 \mu_0 \hat B^2(r)\right].$$9 in transmission. These directly map the vibrational eigenlevels of $$\hat H_{\rm mol}^{(i)} = -\frac{\hbar^2}{2\mu}\frac{\partial^2}{\partial R_i^2} +\frac{\hat J_i^2}{2\mu R_i^2} +V_g(R_i)\,|g\rangle\langle g|_i +V_e(R_i)\,|e\rangle\langle e|_i,$$0. The same set of vibrational oscillations appears in reflection and absorption, superimposed on the upper and lower polariton branches. By contrast, the bound–continuum model shows a smoother continuum band in place of discrete vibrational peaks, but still displays enhanced splitting and broadened polaritonic resonances. The visibility of the vibrational ladder requires dephasing $$\hat H_{\rm mol}^{(i)} = -\frac{\hbar^2}{2\mu}\frac{\partial^2}{\partial R_i^2} +\frac{\hat J_i^2}{2\mu R_i^2} +V_g(R_i)\,|g\rangle\langle g|_i +V_e(R_i)\,|e\rangle\langle e|_i,$$1, equivalently $$\hat H_{\rm mol}^{(i)} = -\frac{\hbar^2}{2\mu}\frac{\partial^2}{\partial R_i^2} +\frac{\hat J_i^2}{2\mu R_i^2} +V_g(R_i)\,|g\rangle\langle g|_i +V_e(R_i)\,|e\rangle\langle e|_i,$$2, consistent with high-$$\hat H_{\rm mol}^{(i)} = -\frac{\hbar^2}{2\mu}\frac{\partial^2}{\partial R_i^2} +\frac{\hat J_i^2}{2\mu R_i^2} +V_g(R_i)\,|g\rangle\langle g|_i +V_e(R_i)\,|e\rangle\langle e|_i,$$3 plasmonic modes with $$\hat H_{\rm mol}^{(i)} = -\frac{\hbar^2}{2\mu}\frac{\partial^2}{\partial R_i^2} +\frac{\hat J_i^2}{2\mu R_i^2} +V_g(R_i)\,|g\rangle\langle g|_i +V_e(R_i)\,|e\rangle\langle e|_i,$$4.

Initial molecular alignment is introduced through an angular distribution

$$\hat H_{\rm mol}^{(i)} = -\frac{\hbar^2}{2\mu}\frac{\partial^2}{\partial R_i^2} +\frac{\hat J_i^2}{2\mu R_i^2} +V_g(R_i)\,|g\rangle\langle g|_i +V_e(R_i)\,|e\rangle\langle e|_i,$$5

where $$\hat H_{\rm mol}^{(i)} = -\frac{\hbar^2}{2\mu}\frac{\partial^2}{\partial R_i^2} +\frac{\hat J_i^2}{2\mu R_i^2} +V_g(R_i)\,|g\rangle\langle g|_i +V_e(R_i)\,|e\rangle\langle e|_i,$$6 is measured relative to the $$\hat H_{\rm mol}^{(i)} = -\frac{\hbar^2}{2\mu}\frac{\partial^2}{\partial R_i^2} +\frac{\hat J_i^2}{2\mu R_i^2} +V_g(R_i)\,|g\rangle\langle g|_i +V_e(R_i)\,|e\rangle\langle e|_i,$$7-axis. The case $$\hat H_{\rm mol}^{(i)} = -\frac{\hbar^2}{2\mu}\frac{\partial^2}{\partial R_i^2} +\frac{\hat J_i^2}{2\mu R_i^2} +V_g(R_i)\,|g\rangle\langle g|_i +V_e(R_i)\,|e\rangle\langle e|_i,$$8 is isotropic, whereas $$\hat H_{\rm mol}^{(i)} = -\frac{\hbar^2}{2\mu}\frac{\partial^2}{\partial R_i^2} +\frac{\hat J_i^2}{2\mu R_i^2} +V_g(R_i)\,|g\rangle\langle g|_i +V_e(R_i)\,|e\rangle\langle e|_i,$$9 gives strong alignment. For period $R_i$0 and for both bound–bound and bound–continuum models, alignment along $R_i$1 produces slightly deeper transmission minima at the polariton resonances, indicating stronger coupling. Alignment along $R_i$2 shifts the resonances to higher energy and deepens the minima further. This behavior is traced to the local polarization of the SPP near field: on the input side, $R_i$3, whereas inside the slits, $R_i$4 (Sukharev et al., 2016).

These calculations support the use of pre-aligned molecules as a sub-diffraction near-field probe. By comparing spectra for $R_i$5-aligned and $R_i$6-aligned molecules, one can infer the local polarization direction and amplitude of the SPP near field at a metal interface. The supplied summary describes this as a “rotational ruler,” with the aligned diatom acting as a local polarization sensor that maps $R_i$7 with molecular-scale resolution.

4. Plasmonic-crystal derivation of the rotonic branch

In the semiconductor-plasmonic-crystal construction, the system is a one-dimensional sequence of gated and ungated regions of total period $R_i$8, supporting a two-dimensional electron gas. Small fluctuations of electron density $R_i$9 and drift velocity $\mu$0 are described hydrodynamically by

$\mu$1

$\mu$2

In the quasi-static approximation,

$\mu$3

with

$\mu$4

for gated and ungated sections respectively, under the condition $\mu$5 (Aizin et al., 15 Apr 2026).

Plane-wave solutions in each uniform section give

$\mu$6

for gated regions and

$\mu$7

for ungated regions. Continuity of current and potential at interfaces, together with the Bloch condition $\mu$8, produces the plasmonic-crystal dispersion relation. The contrast parameter is

$\mu$9

Near a plasmonic band edge, such as $\hat J_i^2/(2\mu R_i^2)$0 at the bottom of the first band or $\hat J_i^2/(2\mu R_i^2)$1 at the top, the spectrum becomes quadratic in the deviation from the band edge:

$\hat J_i^2/(2\mu R_i^2)$2

The defining claim of the paper is that these collective modes therefore differ fundamentally from conventional plasmons in isolated gated or ungated regions: instead of purely linear or square-root dispersion, they display a parabolic branch with a finite effective mass. This is the basis for the term “rotonic plasmons,” used to emphasize an analogy to roton-like excitations.

The paper further states that closed-form expressions can be obtained for the band-edge frequency $\hat J_i^2/(2\mu R_i^2)$3, the band-edge wavevector $\hat J_i^2/(2\mu R_i^2)$4, and the effective mass $\hat J_i^2/(2\mu R_i^2)$5 in terms of $\hat J_i^2/(2\mu R_i^2)$6, $\hat J_i^2/(2\mu R_i^2)$7, $\hat J_i^2/(2\mu R_i^2)$8, $\hat J_i^2/(2\mu R_i^2)$9, $V_g(R)$0, and $V_g(R)$1. For the first band, $V_g(R)$2, $V_g(R)$3, and $V_g(R)$4 is the bottom-of-band frequency.

5. Parametric excitation and generalized Mathieu dynamics

A central feature of the plasmonic-crystal usage is electrical pumping through gate-voltage modulation,

$V_g(R)$5

This modulation changes the gated-region density $V_g(R)$6 and therefore the band-edge frequency. To leading order,

$V_g(R)$7

where $V_g(R)$8 is the threshold above which the gated region depletes. For a single rotonic mode amplitude $V_g(R)$9 near the band edge, including damping $V_e(R)$0, the dynamics reduce to the damped Mathieu equation

$V_e(R)$1

(Aizin et al., 15 Apr 2026).

The principal parametric resonance occurs when $V_e(R)$2, equivalently $V_e(R)$3. The instability threshold is

$V_e(R)$4

and near threshold the growth rate is

$V_e(R)$5

For $V_e(R)$6, the mode amplitude grows as $V_e(R)$7. The same analysis yields a small-signal conversion efficiency for the $V_e(R)$8-th harmonic that scales as

$V_e(R)$9

with a peak when $$\hat H_{\rm int} = -\sum_i \mu(R_i)\cdot \hat E(r_i,t)\left(|e\rangle\langle g|_i+|g\rangle\langle e|_i\right),$$0.

This framework is the basis for RF-to-THz frequency conversion. The supplied summary states that rotonic plasmons in a grating-gate FET can be driven by an RF pump in the $$\hat H_{\rm int} = -\sum_i \mu(R_i)\cdot \hat E(r_i,t)\left(|e\rangle\langle g|_i+|g\rangle\langle e|_i\right),$$1–$$\hat H_{\rm int} = -\sum_i \mu(R_i)\cdot \hat E(r_i,t)\left(|e\rangle\langle g|_i+|g\rangle\langle e|_i\right),$$2 range into a THz plasmon at $$\hat H_{\rm int} = -\sum_i \mu(R_i)\cdot \hat E(r_i,t)\left(|e\rangle\langle g|_i+|g\rangle\langle e|_i\right),$$3. Because the pumping is through the gate voltage rather than source-drain excitation, the same gate-voltage swing can be applied over large-area transistors or transistor arrays, and the method avoids the spatial nonuniformities and electron-drift-velocity saturation effects associated with current-driven excitation.

6. Scaling, representative material systems, and device implications

The THz output power is described as

$$\hat H_{\rm int} = -\sum_i \mu(R_i)\cdot \hat E(r_i,t)\left(|e\rangle\langle g|_i+|g\rangle\langle e|_i\right),$$4

where $$\hat H_{\rm int} = -\sum_i \mu(R_i)\cdot \hat E(r_i,t)\left(|e\rangle\langle g|_i+|g\rangle\langle e|_i\right),$$5 is the total area of the plasmonic crystal and $$\hat H_{\rm int} = -\sum_i \mu(R_i)\cdot \hat E(r_i,t)\left(|e\rangle\langle g|_i+|g\rangle\langle e|_i\right),$$6 is the radiation-loss rate, typically much smaller than $$\hat H_{\rm int} = -\sum_i \mu(R_i)\cdot \hat E(r_i,t)\left(|e\rangle\langle g|_i+|g\rangle\langle e|_i\right),$$7. Larger area and lower damping therefore increase the output power (Aizin et al., 15 Apr 2026).

The band-edge frequency obeys the scaling $$\hat H_{\rm int} = -\sum_i \mu(R_i)\cdot \hat E(r_i,t)\left(|e\rangle\langle g|_i+|g\rangle\langle e|_i\right),$$8. The supplied summary therefore states that the roton minimum can be tuned from sub-THz to multi-THz by adjusting $$\hat H_{\rm int} = -\sum_i \mu(R_i)\cdot \hat E(r_i,t)\left(|e\rangle\langle g|_i+|g\rangle\langle e|_i\right),$$9 in the range $$\Psi(r,R,t)=\chi_g(R,t)\Phi_g(r;R)+\chi_e(R,t)\Phi_e(r;R),$$0–$$\Psi(r,R,t)=\chi_g(R,t)\Phi_g(r;R)+\chi_e(R,t)\Phi_e(r;R),$$1 and by varying the gate bias. Representative numerical examples are given for two materials:

Material	$$\Psi(r,R,t)=\chi_g(R,t)\Phi_g(r;R)+\chi_e(R,t)\Phi_e(r;R),$$2 (ps)	Fundamental $$\Psi(r,R,t)=\chi_g(R,t)\Phi_g(r;R)+\chi_e(R,t)\Phi_e(r;R),$$3
AlGaAs/GaAs (77 K)	3.8	0.75 THz
AlGaN/GaN (77 K)	1.0	2.44 THz

For these examples, the table in the supplied summary also specifies $$\Psi(r,R,t)=\chi_g(R,t)\Phi_g(r;R)+\chi_e(R,t)\Phi_e(r;R),$$4, $$\Psi(r,R,t)=\chi_g(R,t)\Phi_g(r;R)+\chi_e(R,t)\Phi_e(r;R),$$5, $$\Psi(r,R,t)=\chi_g(R,t)\Phi_g(r;R)+\chi_e(R,t)\Phi_e(r;R),$$6, and $$\Psi(r,R,t)=\chi_g(R,t)\Phi_g(r;R)+\chi_e(R,t)\Phi_e(r;R),$$7. At room temperature, $$\Psi(r,R,t)=\chi_g(R,t)\Phi_g(r;R)+\chi_e(R,t)\Phi_e(r;R),$$8 drops by approximately $$\Psi(r,R,t)=\chi_g(R,t)\Phi_g(r;R)+\chi_e(R,t)\Phi_e(r;R),$$9, which raises $$\hat H_{\rm EM}=\int d^3r\left[\frac12 \varepsilon_0 \hat E^2(r)+\frac12 \mu_0 \hat B^2(r)\right].$$00, but the summary states that parametric gain remains achievable for $$\hat H_{\rm EM}=\int d^3r\left[\frac12 \varepsilon_0 \hat E^2(r)+\frac12 \mu_0 \hat B^2(r)\right].$$01 and pump amplitudes $$\hat H_{\rm EM}=\int d^3r\left[\frac12 \varepsilon_0 \hat E^2(r)+\frac12 \mu_0 \hat B^2(r)\right].$$02.

The device-level implication drawn in the source is that electrically tunable plasmonic crystals can operate as compact, voltage-controlled RF-to-THz multipliers or oscillators. Because the same source also attributes parabolic dispersion, narrowband behavior, and a high group index near $$\hat H_{\rm EM}=\int d^3r\left[\frac12 \varepsilon_0 \hat E^2(r)+\frac12 \mu_0 \hat B^2(r)\right].$$03 to these modes, it identifies applications in tunable THz sources and detectors for 6G communications and sensing, including imaging, spectroscopy, security, biomedical, and chemical sensing.

7. Conceptual distinctions and recurrent misconceptions

A recurring source of confusion is to treat the molecular and semiconductor usages as if they described the same quasiparticle. They do not. In the molecular work, the essential departure from standard exciton–plasmon theory is the replacement of conventional two-level emitters by molecules with explicit rotational and vibrational structure, propagated as wave packets on electronic potential-energy surfaces (Sukharev et al., 2016). In the semiconductor work, the essential departure is the emergence of a parabolic plasmonic band edge with finite effective mass in a periodic gated/ungated crystal, together with nonlinear parametric dynamics under gate-voltage pumping (Aizin et al., 15 Apr 2026).

A second misconception is to read the semiconductor term “rotonic” as an assertion of literal superfluid-roton physics. The paper instead states that the name emphasizes an analogy to roton-like excitations. The operative content is the parabolic dispersion law and the associated effective mass near the band edge, not the transfer of the full phenomenology of superfluid helium.

A third misconception concerns the molecular case: the ro-vibrational extension is not a minor perturbation of a two-level model. The reported consequences include significantly higher values of the Rabi splitting, vibrational patterns clearly seen in transmission, reflection, and absorption, and a strong dependence of the optical response on initial molecular pre-alignment. This suggests that the molecular usage is best understood as a qualitatively richer strong-coupling theory, rather than as a small correction to Maxwell–Bloch dynamics.

Taken together, the two usages show that “Rotonic Plasmons” functions as a term of art rather than a uniquely standardized designation. In one setting it names ro-vibrationally structured exciton–plasmon polaritons with alignment-sensitive spectroscopy and sub-diffraction near-field probing. In the other it names plasmonic-crystal modes with a parabolic band-edge spectrum, finite effective mass, parametric instability, and RF-to-THz frequency conversion.