Edge Magnetoplasmons (EMP): Theory & Applications

• Edge magnetoplasmons (EMPs) are chiral collective charge-density waves confined to the edges of 2D electron systems, characterized by unidirectional propagation and sensitivity to Hall conductivity.
• They are probed via microwave transmission, time-of-flight measurements, and photovoltage detection techniques, which reveal resonant frequencies, group velocities, and dissipation behaviors.
• EMPs are integral to non-reciprocal circuit design and topological device engineering, with tunable dynamics influenced by gate screening, Coulomb interactions, and edge confinement.

Edge magnetoplasmons (EMPs) are chiral collective charge-density waves bound to the boundary of a two-dimensional electron system in a perpendicular magnetic field, or, in quantum anomalous Hall settings, to a chiral zero-field edge channel. They are low-frequency, gapless edge excitations distinct from gapped bulk magnetoplasmons, and their propagation is controlled by Hall response, Coulomb interaction, edge confinement, carrier drift, and the electrostatic environment. Across GaAs/AlGaAs heterostructures, graphene, two-dimensional hole systems, magnetic topological insulators, and Chern insulators, EMPs serve both as probes of edge electrodynamics and as microwave carriers in non-reciprocal circuit architectures (Endo et al., 2018, Petkovic et al., 2012, Mahoney et al., 2017).

1. Physical definition and characteristic signatures

In the conventional quantum Hall setting, EMPs are collective density oscillations confined to the sample edge and propagating only in the direction selected by the sign of the magnetic field and the Hall conductivity. Graphene work describes them as one-dimensional, gapless collective charge excitations and, in the edge-channel Tomonaga–Luttinger liquid description, as the bosonic modes of the chiral edge channel. In magnetic topological-insulator and quantum anomalous Hall devices, the same collective object appears on top of a chiral edge state that exists even at zero applied magnetic field (Petkovic et al., 2012, Kumada et al., 2014, Mahoney et al., 2017).

Several experimental signatures recur across platforms. In gated GaAs/AlGaAs quantum Hall edges, microwave transmission shows a fundamental resonance at $f_0$ and harmonics at $f=i f_0$, with $i=2,3,\dots$, which is the expected structure of a resonant collective mode (Endo et al., 2018). In disk geometries, the mode wavelength satisfies

$\lambda=\frac{P}{j},$

with perimeter $P$ and harmonic index $j$, so the phase velocity follows $v_p=fP/j$ (Kumada et al., 2014). In time-domain measurements, chirality appears directly as one-way propagation: reversing the magnetic field reverses the propagation direction, while no counter-propagating mode is seen under the same conditions (Petkovic et al., 2012).

A common misconception is that EMPs are merely single-particle edge currents observed at microwave frequencies. The cited works instead treat them as self-consistent collective excitations of the edge charge distribution. This distinction matters because the measured velocity generally exceeds or differs from the microscopic drift velocity, depends on the electrostatic kernel, and can exhibit resonant, multimode, and dissipative behavior not reducible to single-particle transport (Petkovic et al., 2012, Sokolik et al., 2023, Martinez et al., 2023).

2. Dispersion, screening, and velocity control

The central theoretical structure of EMP physics is the coupling between Hall current and edge electrostatics. In a local-capacitance approximation for a gated quantum Hall edge, the fundamental resonance obeys

$$f_0=\frac{1}{L}\frac{\sigma_{xy}}{C}, \qquad \omega=k_y\frac{\sigma_{xy}}{C},$$

where $L$ is the propagation length, $\sigma_{xy}$ the Hall conductivity, and $f=i f_0$0 the capacitance per unit length between the edge state and the gate. In that description, increasing the lateral separation between the edge and the metallic gate reduces $f=i f_0$1, weakens screening, and raises the EMP velocity and resonance frequency (Endo et al., 2018).

This gate-screening mechanism was isolated directly in time-of-flight measurements on a gate-defined edge channel. There the metallic gate damps the in-plane electric field and lowers the group velocity, while making the gate voltage more negative moves the edge channel farther from the metal and reduces screening. In the strongly screened limit $f=i f_0$2, the quoted result is

$f=i f_0$3

with $f=i f_0$4 the 2DEG depth and $f=i f_0$5 the edge-channel width. Experimentally, the group velocity was tuned from $f=i f_0$6 to $f=i f_0$7 at $f=i f_0$8 (Kamata et al., 2010).

Graphene work expresses the same physics in a form that separates Coulomb and drift contributions: $f=i f_0$9 with the group velocity given by a Hall-conductivity term plus a drift term $i=2,3,\dots$0. In graphene, the drift contribution can be a substantial fraction of the total velocity because the edge is abrupt rather than depletion-defined (Petkovic et al., 2012). A later quantum treatment recast that drift term as a property of magnetic edge-channel response, showing that $i=2,3,\dots$1 is the harmonic mean of the group velocities of the edge channels crossing the Fermi level. In the long-wavelength limit, the total EMP velocity acquires an additive $i=2,3,\dots$2 correction, consistent with the classical skipping-orbit picture (Sokolik et al., 2023).

These results collectively show that EMP velocity is not fixed by Hall conductivity alone. Nearby metal, depletion length, edge width, confining potential, and the microscopic edge-channel spectrum all enter at leading order.

3. Experimental access and observables

EMPs have been measured in both frequency and time domains, with each method emphasizing a different aspect of edge dynamics. In a gated coplanar waveguide (CPW) on a GaAs/AlGaAs 2DEG, a negative bias on the central electrode depletes the 2DEG underneath and creates an electrostatically defined edge in the slot region. Microwave transmission $i=2,3,\dots$3 and thermoelectric voltage $i=2,3,\dots$4 both show the EMP resonance series, providing simultaneous electromagnetic and hot-electron probes of the same edge mode (Endo et al., 2018).

Time-of-flight methods access the propagating wavepacket directly. In GaAs, a short voltage pulse applied to an Ohmic source launches an EMP, while a quantum point contact detects the local electrochemical potential through a delayed gate pulse. The detector current is treated as a correlation between the QPC conductance modulation and the arriving EMP potential, and the measured delay time grows linearly with added path length, confirming ballistic propagation along the edge (Kamata et al., 2010). In graphene, picosecond step pulses and narrow edge contacts allow direct extraction of propagation times around unequal edge paths, with timing calibrated to within about $i=2,3,\dots$5 by three independent procedures (Petkovic et al., 2012).

Other probes emphasize complementary observables. In a p-type two-dimensional hole system, chopped-microwave photovoltage reveals both a giant bulk magnetoplasmon resonance and weaker $i=2,3,\dots$6-periodic oscillations attributed to EMPs, with $i=2,3,\dots$7 approximately (Mi et al., 2016). Microwave impedance microscopy of Chern insulators probes the edge density response rather than a local scalar conductivity; in the EMP interpretation, the tip-sample admittance is resonantly enhanced by collective edge modes whose allowed momenta are quantized by the sample perimeter (Wang et al., 2023).

The coexistence of resonant spectroscopy, time-domain transport, photovoltage detection, and near-field microwave microscopy has made EMPs one of the few quantum Hall edge phenomena that can be followed across both circuit-level and microscopic descriptions.

4. Realizations in GaAs, graphene, and two-dimensional hole systems

In GaAs/AlGaAs integer quantum Hall devices, EMPs have been observed over a broad range of filling factors. In the gated CPW geometry, peaks attributable to EMP excitation were seen from approximately $i=2,3,\dots$8 to $i=2,3,\dots$9; the fundamental frequency $\lambda=\frac{P}{j},$0 increases as $\lambda=\frac{P}{j},$1 becomes more negative, as $\lambda=\frac{P}{j},$2 decreases within a plateau, and with higher integer filling factor $\lambda=\frac{P}{j},$3. Notably, the peaks persist, though less sharply, even in the regime $\lambda=\frac{P}{j},$4, where the 2DEG still remains under the gate, and the interpretation links this to screening changes as the gated region crosses localized and extended integer-quantum-Hall states of its own (Endo et al., 2018).

Graphene introduced several nonstandard EMP regimes. Picosecond time-of-flight measurements established chiral propagation with low attenuation, an attenuation length of $\lambda=\frac{P}{j},$5, and a relaxation time of $\lambda=\frac{P}{j},$6. The velocity is quantized on Hall plateaus, and the extracted drift contribution at $\lambda=\frac{P}{j},$7,

$\lambda=\frac{P}{j},$8

is slightly less than the graphene Fermi velocity, as expected for an abrupt edge. The same analysis extracted a characteristic Coulomb length $\lambda=\frac{P}{j},$9, much larger than the magnetic length but smaller than in soft-edge systems (Petkovic et al., 2012).

Graphene edge structure can also change the number and chirality of the fundamental EMPs. In a wide armchair ribbon at $P$0, when the Fermi level intersects four degenerate states of the zero Landau level at one location and two degenerate states at another, two fundamental counter-propagating EMPs exist with opposite chirality and substantial spatial overlap; when the Fermi level is sufficiently high, only one fundamental EMP remains (Balev et al., 2010). Adding a weak smooth superlattice potential enlarges the number of EMP branches, produces finite frequency gaps and zero-group-velocity points, and leaves only two branches gapless as $P$1. The predicted response is a strong resonance at the frequency where the relevant branch has zero group velocity (Balev et al., 2012).

In high-mobility p-type AlGaAs/GaAs quantum wells, EMPs were identified in photovoltage as $P$2-periodic oscillations coexisting with a giant bulk magnetoplasmon resonance. The heavy-hole effective mass, $P$3, places the system in an intermediate regime that is neither the low-frequency nor the high-frequency limit of standard EMP theory, which explains why the observed $P$4 law acquires a nonzero intercept (Mi et al., 2016).

5. Dissipation, multimode structure, and circuit descriptions

Although EMPs are chiral, they are not generically dissipationless. In disk-shaped graphene, frequency- and time-domain measurements separated two mechanisms. The first is intrinsic decay from nonlinear dispersion: because the dispersion contains a logarithmic interaction term, a wavepacket spreads and deforms even without energy loss. The second is extrinsic dissipation through localized states in the bulk graphene, with a decay rate modeled as

$P$5

where the $P$6 term is associated with capacitive coupling and $P$7 follows a variable-range hopping form. The measured quality factors were $P$8 at $P$9 and $j$0 at $j$1 (Kumada et al., 2014).

A distinct dissipation regime appears in not-too-low-temperature GaAs theory. For $j$2 and very strong dissipation, a weakly damped symmetric mode survives while other EMP modes are strongly damped; this mode is termed the helical edge magnetoplasmon. Its existence depends on nonperturbative Coulomb renormalization of the transverse charge profile. The same framework also shows that, at $j$3, the fundamental EMPs associated with the $j$4 and $j$5 Landau levels are strongly renormalized by Coulomb coupling (Silva et al., 2010).

EMP transport is intrinsically multimode and circuit-like. A chiral distributed-element (CDE) description models a quantum Hall edge as a unidirectional transmission line with electrochemical capacitance $j$6 and characteristic velocity

$j$7

Within this language, inter-edge capacitance produces charge fractionalization, a pinched-off quantum point contact remains an rf plasmonic coupler through capacitive transmission, and single- and double-cavity spectra can be fitted quantitatively to extract local electrostatic parameters (Hashisaka et al., 2013). Extending the same logic to large networks, “edge magnetoplasmon crystals” composed of chains, ladders, and honeycomb arrangements exhibit a linear dispersive band from extended chiral propagation modes and a flat band from standing waves localized in coupled regions (Sasaki, 2021).

6. Topological, zero-field, and engineered extensions

The EMP concept has been carried beyond conventional quantum Hall edges into zero-field topological systems. In a magnetized disk of Cr-$j$8 in the quantum anomalous Hall regime, contactless microwave measurements revealed resonance-like features attributed to chiral EMPs, including a fundamental mode near $j$9, a velocity of approximately $v_p=fP/j$0, and about $v_p=fP/j$1 non-reciprocity at zero field. The signal was interpreted as interference between a fast parasitic capacitive path and a slow EMP path around the disk, producing a Fano-like resonance (Mahoney et al., 2017).

Frequency- and time-domain measurements on QAH disks refined this picture. Using five disk radii, the EMP dispersion was fitted with an edge-width parameter $v_p=fP/j$2 and drift velocity $v_p=fP/j$3, while time-of-flight measurements gave an edge plasmon packet velocity $v_p=fP/j$4. The decay rate again followed $v_p=fP/j$5, with microwave power enhancing dissipative hopping through bulk puddles. The same study reported that low microwave power and disk diameter $v_p=fP/j$6 can significantly reduce decay, and achieved $v_p=fP/j$7 non-reciprocity at $v_p=fP/j$8 in an $v_p=fP/j$9 device (Martinez et al., 2023).

Microwave impedance microscopy provides a complementary topological viewpoint. In the linear-response formulation for a Chern insulator, the enhanced MIM edge signal is attributed to collective EMP poles of the edge susceptibility rather than to a static edge conductivity. The predicted resonance frequencies depend on the Chern number and the sample circumference, making the response explicitly nonlocal and topological (Wang et al., 2023).

Several works also delineate the boundary between conventional EMPs and related Hall-driven modes. Domain-boundary magnetoplasmons in an edgeless 2DEG arise at sign-changing magnetic-field interfaces rather than at a physical density edge, and are protected by the difference of gap Chern numbers across the magnetic domains (Jin et al., 2018). A radially inhomogeneous but cylindrically symmetric electron fluid on liquid helium supports a delocalized low-frequency magnetoplasmon that is gapless, chiral, and Hall-driven like an EMP, but is not edge-localized (Kostylev et al., 2024). Magnetic-edge magnetoplasmons in a 2DES with a semi-infinite gate and ferromagnetic film provide yet another variant, with unusual spatial dispersion and fast branches of opposite chirality (Balev et al., 2016). These developments suggest a broader class of low-frequency Hall magnetoplasmons, of which the conventional EMP remains the edge-localized prototype.