Image Magnetic Monopoles in Topological Insulators

• Image magnetic monopoles are effective magnetic charges induced by an external electric charge near a magnetoelectric surface, described via axion electrodynamics and classical boundary conditions.
• They provide a unified framework linking topological insulator responses with classical magnetoelectric behavior, as evidenced by spectroscopic splitting in image potential states.
• Experimental realizations use techniques like scanning tunneling spectroscopy and muon spin rotation to detect weak monopolar fields, though challenges remain due to electrostatic interferences.

Searching arXiv for recent and foundational papers on image magnetic monopoles in topological insulators and magnetoelectric media. Image magnetic monopole denotes an effective magnetic charge that appears as the magnetostatic image of an external electric charge near a medium with magnetoelectric response. In ordinary electromagnetism, the absence of true magnetic charge is expressed by $\nabla\!\cdot\!\mathbf B=0$, but in topological or linear magnetoelectrics the boundary conditions can mimic those of a magnetic image charge located beneath an interface. The resulting vacuum magnetic field is monopolar in form, even though no elementary monopole is introduced. This construction has been developed in axion electrodynamics for three-dimensional topological insulators and in classical linear magnetoelectrics such as $\mathrm{Cr_2O_3}$, and has more recently been connected to spectroscopic detection through image potential states on Bi(111) surfaces (Zhan et al., 29 Sep 2025, Meier et al., 2018).

1. Definition and conceptual status

An image magnetic monopole is a boundary-induced effective source of magnetic field generated when an electric charge is brought near a material whose constitutive relations mix electric and magnetic responses. In the method-of-images language, the external charge induces not only an electric image charge but also a magnetic image charge beneath the surface (Meier et al., 2018). In the topological-insulator formulation, the same phenomenon is described as a consequence of the topological magnetoelectric effect arising from a $\theta$-term added to the electromagnetic action (Zhan et al., 29 Sep 2025).

The central conceptual point is that the monopole is “image” rather than elemental. It is fictitious in the same sense as an electrostatic image charge: the external field is exactly or effectively reproduced by a source placed at the mirror point, but the microscopic origin lies in induced polarization, magnetization, or surface Hall currents inside the material (Zhan et al., 29 Sep 2025, Meier et al., 2018). This distinguishes image monopoles from emergent quasiparticles in spin ice, where monopole-like defects are tied to violations of ice rules, and from synthetic monopoles in Bose–Einstein condensates, where a monopolar field appears in a synthetic gauge field rather than in ordinary magnetostatics (Keswani et al., 2020, Ray et al., 2014).

A common misconception is that observation of an image monopole would constitute discovery of a fundamental magnetic charge. The available works do not support that interpretation. They instead show that Maxwell boundary conditions, supplemented by magnetoelectric couplings, can generate an external magnetic field identical in form to that of a monopole source (Meier et al., 2018), or that such a field can be inferred spectroscopically through its Zeeman action on surface-bound electronic states (Zhan et al., 29 Sep 2025).

2. Electromagnetic formulation

The topological-insulator description is formulated by supplementing the ordinary electromagnetic action

$$S_0 = \int d^3x\,dt\,\left[\epsilon_0E^2/2 - B^2/(2\mu_0)\right]$$

with the axion term

$$S_\theta = (\theta \alpha/4\pi)\int d^3x\,dt\,\mathbf E\!\cdot\!\mathbf B,$$

where $\alpha=e^2/(\hbar c)$ is the fine-structure constant (Zhan et al., 29 Sep 2025). For time-reversal-invariant strong topological insulators, $\theta=\pi \;(\mathrm{mod}\;2\pi)$, and the induced responses are

$$\mathbf P = (\theta \alpha/\pi)\,\mathbf B,\qquad \mathbf M = -(\theta \alpha/\pi)\,\mathbf E$$

localized at surfaces or domain walls (Zhan et al., 29 Sep 2025). When a point charge $q$ is placed at height $z_0$ above a semi-infinite topological insulator, the interface boundary conditions are equivalent to those produced by a magnetic image charge

$\mathrm{Cr_2O_3}$0

For $\mathrm{Cr_2O_3}$1, this gives $\mathrm{Cr_2O_3}$2 (Zhan et al., 29 Sep 2025). The corresponding vacuum magnetic field is

$\mathrm{Cr_2O_3}$3

with $\mathrm{Cr_2O_3}$4 measured from the image position beneath the surface (Zhan et al., 29 Sep 2025).

The classical magnetoelectric treatment starts from the static Maxwell equations with constitutive relations inside the medium

$\mathrm{Cr_2O_3}$5

while outside the medium one has the vacuum relations $\mathrm{Cr_2O_3}$6 and $\mathrm{Cr_2O_3}$7 (Meier et al., 2018). For an isotropic magnetoelectric with a point charge $\mathrm{Cr_2O_3}$8 at $\mathrm{Cr_2O_3}$9 above the plane $\theta$0, the image monopole strength is

$\theta$1

and the external field is

$\theta$2

with $\theta$3 (Meier et al., 2018).

For uniaxial media such as $\theta$4, the response tensors are diagonal but anisotropic,

$\theta$5

yet the field above the surface remains purely monopolar:

$\theta$6

A symmetry argument shows that only the sum $\theta$7 enters the external monopole strength (Meier et al., 2018).

3. Realizations in topological insulators and magnetoelectrics

Two material realizations are explicitly documented in the supplied literature: three-dimensional topological insulators and linear magnetoelectric surfaces.

In the topological-insulator case, Zhan et al. study Bi(111) thin films of thickness $\theta$8 on Si(111), probed by scanning tunneling microscopy with a negatively biased tungsten tip at $\theta$9 (Zhan et al., 29 Sep 2025). The STM junction produces both a perpendicular electric field that binds image potential states and a radial in-plane electric field $$S_0 = \int d^3x\,dt\,\left[\epsilon_0E^2/2 - B^2/(2\mu_0)\right]$$0. Through the topological magnetoelectric effect, this radial field drives a surface Hall current

$$S_0 = \int d^3x\,dt\,\left[\epsilon_0E^2/2 - B^2/(2\mu_0)\right]$$1

which generates the image-monopole magnetic field in the vacuum gap (Zhan et al., 29 Sep 2025). In this formulation, the monopole is not introduced as a separate degree of freedom but emerges from the axion response of the topological surface.

In the classical magnetoelectric realization, the prototypical material is $$S_0 = \int d^3x\,dt\,\left[\epsilon_0E^2/2 - B^2/(2\mu_0)\right]$$2 (Meier et al., 2018). Using low-temperature parameters $$S_0 = \int d^3x\,dt\,\left[\epsilon_0E^2/2 - B^2/(2\mu_0)\right]$$3, $$S_0 = \int d^3x\,dt\,\left[\epsilon_0E^2/2 - B^2/(2\mu_0)\right]$$4, $$S_0 = \int d^3x\,dt\,\left[\epsilon_0E^2/2 - B^2/(2\mu_0)\right]$$5, $$S_0 = \int d^3x\,dt\,\left[\epsilon_0E^2/2 - B^2/(2\mu_0)\right]$$6, $$S_0 = \int d^3x\,dt\,\left[\epsilon_0E^2/2 - B^2/(2\mu_0)\right]$$7, and $$S_0 = \int d^3x\,dt\,\left[\epsilon_0E^2/2 - B^2/(2\mu_0)\right]$$8, the full uniaxial solution gives

$$S_0 = \int d^3x\,dt\,\left[\epsilon_0E^2/2 - B^2/(2\mu_0)\right]$$9

for a charge $$S_\theta = (\theta \alpha/4\pi)\int d^3x\,dt\,\mathbf E\!\cdot\!\mathbf B,$$0 at $$S_\theta = (\theta \alpha/4\pi)\int d^3x\,dt\,\mathbf E\!\cdot\!\mathbf B,$$1, corresponding to

$$S_\theta = (\theta \alpha/4\pi)\int d^3x\,dt\,\mathbf E\!\cdot\!\mathbf B,$$2

The predicted field at the surface is approximately $$S_\theta = (\theta \alpha/4\pi)\int d^3x\,dt\,\mathbf E\!\cdot\!\mathbf B,$$3, decreasing to about $$S_\theta = (\theta \alpha/4\pi)\int d^3x\,dt\,\mathbf E\!\cdot\!\mathbf B,$$4 at $$S_\theta = (\theta \alpha/4\pi)\int d^3x\,dt\,\mathbf E\!\cdot\!\mathbf B,$$5 and about $$S_\theta = (\theta \alpha/4\pi)\int d^3x\,dt\,\mathbf E\!\cdot\!\mathbf B,$$6 at $$S_\theta = (\theta \alpha/4\pi)\int d^3x\,dt\,\mathbf E\!\cdot\!\mathbf B,$$7 (Meier et al., 2018).

These two realizations differ in formal language and experimental access. The topological-insulator work emphasizes axion electrodynamics and spectroscopic signatures in electronic bound states (Zhan et al., 29 Sep 2025), whereas the magnetoelectric work emphasizes exact classical boundary-value solutions and direct field magnitudes above a surface (Meier et al., 2018). A plausible implication is that the image-monopole concept functions as a unifying effective description across both topological and non-topological magnetoelectric media.

4. Detection by image potential states

The most explicit claimed detection mechanism in the supplied literature uses image potential states (IPS) on Bi(111) (Zhan et al., 29 Sep 2025). IPS are vacuum-bound electronic states formed when the STM tip bias exceeds the sample work function, so that electrons experience an attractive Coulomb potential $$S_\theta = (\theta \alpha/4\pi)\int d^3x\,dt\,\mathbf E\!\cdot\!\mathbf B,$$8 just below the vacuum level. Their energies are approximately

$$S_\theta = (\theta \alpha/4\pi)\int d^3x\,dt\,\mathbf E\!\cdot\!\mathbf B,$$9

with an effective Rydberg $\alpha=e^2/(\hbar c)$0–$\alpha=e^2/(\hbar c)$1 depending on effective mass and geometry (Zhan et al., 29 Sep 2025).

The key detection idea is that the image monopole field couples to the orbital magnetic moments of these IPS through the Zeeman interaction. Each IPS orbital carries

$\alpha=e^2/(\hbar c)$2

and the corresponding energy shift is

$\alpha=e^2/(\hbar c)$3

Because the monopole field is radial and the system retains $\alpha=e^2/(\hbar c)$4 symmetry, the magnetic quantum number $\alpha=e^2/(\hbar c)$5 remains degenerate, while different $\alpha=e^2/(\hbar c)$6 sublevels split (Zhan et al., 29 Sep 2025). For principal quantum number $\alpha=e^2/(\hbar c)$7, the total splitting between the $\alpha=e^2/(\hbar c)$8 and $\alpha=e^2/(\hbar c)$9 sublevels behaves as

$\theta=\pi \;(\mathrm{mod}\;2\pi)$0

which is the form fitted to the measured data in Eq. 10 of the paper (Zhan et al., 29 Sep 2025).

Experimentally, scanning tunneling spectroscopy on 3-Bi and Bi(110) yields unsplit Gaussian IPS peaks for $\theta=\pi \;(\mathrm{mod}\;2\pi)$1, whereas on Bi(111) the peaks with $\theta=\pi \;(\mathrm{mod}\;2\pi)$2 exhibit sub-peak structure (Zhan et al., 29 Sep 2025). The $\theta=\pi \;(\mathrm{mod}\;2\pi)$3 IPS splits into two resolvable peaks with measured splitting $\theta=\pi \;(\mathrm{mod}\;2\pi)$4, compared with a theory value of approximately $\theta=\pi \;(\mathrm{mod}\;2\pi)$5 in Supplement S3 (Zhan et al., 29 Sep 2025). Higher levels $\theta=\pi \;(\mathrm{mod}\;2\pi)$6 show broader asymmetric line shapes consistent with multiple $\theta=\pi \;(\mathrm{mod}\;2\pi)$7 sublevels. By varying the STM setpoint current $\theta=\pi \;(\mathrm{mod}\;2\pi)$8 from $\theta=\pi \;(\mathrm{mod}\;2\pi)$9 to $$\mathbf P = (\theta \alpha/\pi)\,\mathbf B,\qquad \mathbf M = -(\theta \alpha/\pi)\,\mathbf E$$0, the tip–sample distance and hence the electric field are tuned; empirically the splitting follows a Fowler–Nordheim–like form

$$\mathbf P = (\theta \alpha/\pi)\,\mathbf B,\qquad \mathbf M = -(\theta \alpha/\pi)\,\mathbf E$$1

which is presented as evidence that $$\mathbf P = (\theta \alpha/\pi)\,\mathbf B,\qquad \mathbf M = -(\theta \alpha/\pi)\,\mathbf E$$2 and thus that the splitting tracks the TME-induced monopole field (Zhan et al., 29 Sep 2025).

This interpretation is explicitly strong in the source: the work describes the measurement as the first direct detection of an image magnetic monopole field in a solid-state experiment and the first spectroscopic confirmation of axion electrodynamics with $$\mathbf P = (\theta \alpha/\pi)\,\mathbf B,\qquad \mathbf M = -(\theta \alpha/\pi)\,\mathbf E$$3 in a 3D topological insulator (Zhan et al., 29 Sep 2025). At the same time, the data block does not provide an independent exclusion analysis for all alternative line-broadening or splitting mechanisms, so any broader methodological assessment must remain cautious.

5. Direct field probes and experimental constraints

A complementary route is to detect the external monopolar field directly rather than spectroscopically. The most detailed classical study is the search at a magnetoelectric $$\mathbf P = (\theta \alpha/\pi)\,\mathbf B,\qquad \mathbf M = -(\theta \alpha/\pi)\,\mathbf E$$4 surface (Meier et al., 2018).

The principal experiment there uses low-energy muon spin rotation. The sample is a $$\mathbf P = (\theta \alpha/\pi)\,\mathbf B,\qquad \mathbf M = -(\theta \alpha/\pi)\,\mathbf E$$5 $$\mathbf P = (\theta \alpha/\pi)\,\mathbf B,\qquad \mathbf M = -(\theta \alpha/\pi)\,\mathbf E$$6 film on $$\mathbf P = (\theta \alpha/\pi)\,\mathbf B,\qquad \mathbf M = -(\theta \alpha/\pi)\,\mathbf E$$7, magnetoelectrically annealed into a single antiferromagnetic domain under $$\mathbf P = (\theta \alpha/\pi)\,\mathbf B,\qquad \mathbf M = -(\theta \alpha/\pi)\,\mathbf E$$8 poling at $$\mathbf P = (\theta \alpha/\pi)\,\mathbf B,\qquad \mathbf M = -(\theta \alpha/\pi)\,\mathbf E$$9 and $q$0 (Meier et al., 2018). A $q$1 solid $q$2 overlayer serves as an insulating muon stopping region. Fully polarized $q$3 with tunable energy from $q$4 to $q$5 are implanted into the $q$6 layer at mean depths of $q$7–$q$8, with about $q$9 remaining as $z_0$0 rather than forming muonium (Meier et al., 2018). A transverse bias field $z_0$1 is applied, and the local field is inferred from the muon precession signal. Muons stopping in $z_0$2 depolarize rapidly and do not contribute, whereas those in $z_0$3 detect the bias field plus the monopolar leakage field from the image monopole. The measured frequency shifts are reported to be of order a few $z_0$4, increasing as the mean muon–surface distance decreases, and consistent in sign, magnitude, and $z_0$5 trend with the theoretical prediction (Meier et al., 2018).

The same work analyzes magnetic force microscopy as a possible direct probe. The proposed idea is to use a charged magnetic AFM tip at about $z_0$6 both to create the effective point charge and to sense the field. The predicted field at the tip, $z_0$7–$z_0$8, lies above an MFM sensitivity of about $z_0$9 (Meier et al., 2018). However, no domain-dependent MFM contrast was observed because surface roughness of approximately $\mathrm{Cr_2O_3}$00 and charging-induced electrostatic forces overwhelm the much smaller magnetostatic signal (Meier et al., 2018). This result is methodologically important because it shows that detectability of the field amplitude alone is insufficient; charge control and suppression of electrostatic backgrounds are decisive.

The paper also discusses scanning SQUID magnetometry and near-surface spin probes such as NV centers as possible alternatives. For a SQUID loop of radius $\mathrm{Cr_2O_3}$01 above the surface, the collected flux is

$\mathrm{Cr_2O_3}$02

which approaches $\mathrm{Cr_2O_3}$03 for $\mathrm{Cr_2O_3}$04 and is estimated to be comparable to a few flux quanta for $\mathrm{Cr_2O_3}$05 (Meier et al., 2018). The authors note that near-surface single spins in diamond could in principle map the $\mathrm{Cr_2O_3}$06 field with sub-$\mathrm{Cr_2O_3}$07 depth resolution, though the same surface-charging issues remain (Meier et al., 2018).

6. Relation to monopole analogues in other platforms

The image magnetic monopole belongs to a broader family of monopole analogues, but its physical mechanism is distinct from those in artificial spin ice, spinor condensates, photonic nanoantennas, and polariton spin ice.

In square artificial spin ice, monopole-like excitations are charged vertices defined by the dumbbell approximation and the charge sum

$\mathrm{Cr_2O_3}$08

with isolated $\mathrm{Cr_2O_3}$09-out/$\mathrm{Cr_2O_3}$10-in excitations carrying $\mathrm{Cr_2O_3}$11 (Keswani et al., 2020). Keswani et al. demonstrate a controlled stabilization of a robust isolated emergent monopole in an open-edged square ASI vertex under in-plane field, with charge neutrality maintained by boundary charges rather than by a partner antimonopole on the same plaquette (Keswani et al., 2020). In a related rectangular two-dimensional ASI geometry, MFM measurements show monopole creation, transport, and annihilation without visible strings in the deconfined regime near aspect ratio $\mathrm{Cr_2O_3}$12 (Duarte et al., 2022). These are emergent quasiparticles in frustrated dipolar arrays, not image sources produced by boundary conditions in a magnetoelectric medium.

In spinor Bose–Einstein condensates, Ray et al. engineer a synthetic monopole field through the spin texture of a ferromagnetic spin-1 condensate, obtaining

$\mathrm{Cr_2O_3}$13

together with a terminating vortex line identified as the Dirac string (Ray et al., 2014). The field is synthetic and tied to the condensate order parameter. It is therefore closer to a gauge-field monopole than to an image monopole.

In photonics, a half-nanoslit carved in a semi-infinite gold film behaves as an effective oscillating magnetic charge under optical excitation. Reynier et al. formulate this using

$\mathrm{Cr_2O_3}$14

and show numerically that the half slit produces a single magnetic hot spot and radiates with the characteristic $\mathrm{Cr_2O_3}$15 pattern of an oscillating monopolar source (Reynier et al., 2023). Again, the mechanism is geometric boundary engineering rather than magnetoelectric image formation.

The comparison clarifies a useful taxonomy. Image magnetic monopoles are interface-induced effective sources determined by constitutive relations and boundary conditions (Meier et al., 2018, Zhan et al., 29 Sep 2025). Emergent monopoles in spin ice are vertex defects in frustrated many-body systems (Keswani et al., 2020, Duarte et al., 2022). Synthetic monopoles in quantum fluids are singularities in effective gauge fields (Ray et al., 2014). Photonic monopole antennas are boundary-condition analogues in driven electromagnetic nanostructures (Reynier et al., 2023). This suggests that “magnetic monopole” functions less as a single ontological category than as a recurring field pattern realized by different microscopic mechanisms.

7. Significance, limitations, and open directions

The principal significance of the image-monopole concept is that it provides a controlled route to realizing a monopolar magnetic field in vacuum without requiring a fundamental magnetic charge. In topological-insulator language, it makes axion electrodynamics experimentally accessible through local probes and spectroscopy (Zhan et al., 29 Sep 2025). In classical magnetoelectrics, it reduces a coupled electro-magnetostatic boundary problem to an analytically tractable and potentially measurable monopolar field outside the sample (Meier et al., 2018).

Several limitations are explicit in the available literature. First, the monopole is effective and interface-bound in origin; its existence depends on material response, surface quality, and the localization of the inducing electric charge (Meier et al., 2018, Zhan et al., 29 Sep 2025). Second, the field strengths are small on the scale of typical near-surface experimental backgrounds, with the $\mathrm{Cr_2O_3}$16 case reaching the $\mathrm{Cr_2O_3}$17 range at nanometric distances and rapidly decaying as $\mathrm{Cr_2O_3}$18 (Meier et al., 2018). Third, direct scanning probes must contend with electrostatic artifacts severe enough to mask the signal even when nominal magnetic sensitivity is adequate (Meier et al., 2018). Fourth, the STM-IPS detection scheme infers the monopole field through spectral splitting rather than direct vector-field imaging (Zhan et al., 29 Sep 2025).

The literature also points to several research directions. Zhan et al. suggest extension to other topological insulators and axion insulators, with possible relevance to monopole–electron dyon excitations and quantum information settings involving topological magnetoelectric coupling (Zhan et al., 29 Sep 2025). The magnetoelectric-surface work identifies scanning SQUID and single-spin probes as promising alternatives to MFM, provided surface-charge control can be improved (Meier et al., 2018). A plausible implication is that progress will depend less on the formal monopole solution itself, which is already well established, than on nanoscale charge engineering and high-dynamic-range field discrimination at interfaces.

Taken together, the image magnetic monopole has become a technically precise and experimentally active concept linking classical magnetoelectric boundary problems, axion electrodynamics in topological matter, and nanoscale spectroscopy. Its importance lies not in resolving the existence of fundamental magnetic charge, but in providing a concrete realization of monopolar magnetic fields in condensed-matter and surface-physics settings (Meier et al., 2018, Zhan et al., 29 Sep 2025).