betaFeedback

WorksPhysics and Astronomy

A Dirac-material-inspired non-linear electrodynamic model

Abstract

We propose and study the properties of a non-linear electrodynamics that emerges inspired on the physics of Dirac materials. This new electrodynamic model is an extension of the one-loop corrected non-linear effective Lagrangian computed in the work of ref. [3]. In the particular regime of a strong magnetic and a weak electric field, it reduces to the photonic non-linear model worked out by the authors of ref. [3]. We pursue our investigation of the proposed model by analyzing properties of the permittivity and permeability tensors, the energy-momentum tensor and wave propagation effects in presence of a uniform magnetic background. It is shown that the electrodynamics here presented exhibits the vacuum birefringence phenomenon. Subsequently, we calculate the lowest-order modifications to the interaction energy, considering still the presence of a uniform external magnetic field. Our analysis is carried out within the framework of the gauge-invariant but path-dependent variables formalism. The calculation reveals a screened Coulomb-like potential with an effective electric charge that runs with the external magnetic field but, as expected for Dirac-type materials, the screening disappears whenever the external magnetic field is switched off.

Introduction

As well-known, over the last decades, the area of quantum condensed matter physics and quantum field theory have become increasingly interconnected. This success is based on the fact that phenomena appearing in different systems are unified by common physical mechanisms [1, 2]. An illustrative example arises when one considers Weyl and Dirac semimetals, where it was shown in ref. [3] that quantum vacuum non-linear effects contribute to the experimentally observed high field magnetization. These effects have their origin in the well-known result due to Heisenberg and Euler [4].

In this context, it may be recalled that the subject of quantum vacuum non-linearities has been of great interest since the pioneering work by Euler and Heisenberg [4]. In fact, in the case of non-linear electrodynamic models that appear as effective descriptions by summing up electron-positron quantum effects, the polarizable medium is the Quantum Electrodynamics (QED) vacuum, made up of virtual electrons and positrons. That is the case of the Euler-Heisenberg effective model. Incidentally, it is of interest to notice that Schwinger later confirmed this remarkable quantum prediction of light-by-light scattering from QED [5].

In this connection, we call attention to the fact that one of the striking physical effects of the Heisenberg and Euler result has been vacuum birefringence. In other words, the quantum vacuum when is stressed by external electromagnetic fields behaves as if it were a birefringent material medium, which has been emphasized from different perspectives [6, 7, 8, 9, 10, 11]. However, this optical phenomenon has not yet been confirmed. Mention should be made, at this point, to recent results from ATLAS and CMS collaborations at the Large Hadron Collider (LHC), which have reported on the high energy gamma-gamma pair emission from virtual gamma-gamma scattering in ultraperipheral Pb-Pb collisions [12, 13, 14]. It is worthy to emphasize here that, in these results, there is no modification of the optical properties of the vacuum [15]. We further mention that the coming of laser facilities has given rise to various proposals to probe quantum vacuum non-linearities [15, 16, 17].

Inspired by these observations, the purpose of this paper is to further elaborate on the physical content of the non-linear one-loop corrected effective electrodynamics discussed in ref. [3]. Of special interest will be to consider the non-linear electrodynamics for Dirac materials in the familiar language of standard quantum field theory. Along these lines, we establish a connection between condensed matter and quantum field effective theories. Evidently, these connections are of interest in providing unifications among diverse models. We clarify that the model we present in this contribution is not derived as an effective physics stemming from Dirac materials. What we actually do is to write down a Lagrangian density that extends the effective physics worked out in ref. [3] and recovers the latter in the special limit of a strong magnetic and weak electric field. Once the Dirac-material-inspired action is formulated, we obtain the field equations, conservation laws, as well as the properties of the medium due to non-linear contributions. Next, we study the propagation effects of this non-linear electrodynamics under an external and uniform magnetic field. We add the prescription of linearization expanding the Lagrangian density of the theory up to second order in the propagating fields. Thereby, we are able to extract the birefringence phenomenon in some Dirac materials. At last, to make contact with the discussion of the Coulomb screening that occurs in Dirac materials [18], we calculate the interaction energy in Dirac materials under an uniform magnetic field via path-dependent variables formalism, and show that an effective charge comes out that depends on the external magnetic field.

The organization of the paper is as follows: In § 2, we introduce a new non-linear electrodynamics for Dirac materials and analyze aspects such as equations of motion, the permittivity and permeability tensors and the corresponding energy-momentum tensor. In § 3, we study wave propagation effects under a uniform magnetic background. In § 4, we consider the birefringence phenomenon for this new electrodynamics. In § 5, we compute the interaction energy for this electrodynamics under a uniform magnetic field. Our discussion is accomplished using the gauge-invariant but path-dependent variables formalism, which is an alternative to the usual Wilson loop approach. Finally, some concluding comments and considerations are cast in § 6.

Throughout this work, we adopt natural units ℏ=c=1\hbar= c = 1 with 4πϵ0=14\pi\epsilon_0 = 1, and the Minkowski metric ημν=diag⁡(+1,−1,−1,−1)\eta^{\mu\nu} = \operatorname{diag}(+1,-1,-1,-1). We use the conversion 1 m=5×1012 MeV−11\ \mathrm{m} = 5 \times10^{12}\ \mathrm{MeV}^{-1}. The electric and magnetic fields have squared-energy mass dimension, where the conversion of Volt/m and Tesla (T) to the natural system is given by 1 Volt/m=2.27×10−24 GeV21\ \mathrm{Volt/m} = 2.27 \times10^{-24}\ \mathrm{GeV}^2 and 1 T=6.8×10−16 GeV21\ \mathrm{T} = 6.8 \times10^{-16}\ \mathrm{GeV}^2, respectively.

A Non-linear Electrodynamics for Dirac Materials

As was shown in the work [3], the authors computed the non-linear one-loop contribution for a Dirac material, given by the effective Lagrangian density:

LDM≃αD24π2(Δ2evF)2[(e⋅b)2∣b∣+b2ln⁡(∣b∣)](1)\mathcal{L}_{DM} \simeq\frac{\alpha_D}{24\pi^2}\left(\frac{\Delta^2}{e v_F}\right)^2\left[\frac{(\mathbf{e}\cdot\mathbf{b})^2}{|\mathbf{b}|}+\mathbf{b}^2\ln(|\mathbf{b}|)\right] \tag*{(1)}

where vFv_F denotes the Fermi velocity and Δ\Delta is the material gap with energy dimension. This description holds in the regime of a strong magnetic field and a weak electric field. Furthermore, the dimensionless electromagnetic (EM) fields e\mathbf{e} and b\mathbf{b} depend on the fine structure constant αD=e2/vF\alpha_D = e^2/v_F in the Dirac material,

e(αD)=UEE⋆(αD),b(αD)=U−1BB⋆(αD)(2)\mathbf{e}(\alpha_D)=\frac{\mathcal{U}\mathbf{E}}{E_\star(\alpha_D)},\qquad\mathbf{b}(\alpha_D)=\frac{\mathcal{U}^{-1}\mathbf{B}}{B_\star(\alpha_D)} \tag*{(2)}
Δ(meV)\Delta(\mathrm{meV})vF=α/αDv_F = \alpha/\alpha_DE⋆(eV2)E_\star(\mathrm{eV}^2)B⋆(eV2)B_\star(\mathrm{eV}^2)
QED5×1085 \times10^8112.95×10122.95 \times10^{12}3.0×10123.0 \times10^{12}
PbSnTe31.531.50.00170.00176.586.583.8×10−33.8 \times10^{-3}
BiSb7.757.750.00530.00530.130.1324.424.4
TaAs000.00280.00280000

Table 1. The comparison of the parameters for some materials and the QED. The fine structure constant in Dirac material (αD\alpha_D) is defined in relation to QED (α\alpha), i.e., αD/α=vF−1\alpha_D/\alpha= v_F^{-1}.

with E⋆(αD)E_\star(\alpha_D) and B⋆(αD)B_\star(\alpha_D) being the critical Schwinger’s electric and magnetic fields in the material, such that

E⋆2(αD)=vF2B⋆2(αD)=(Δ2evF)2(3)E_\star^2(\alpha_D)=v_F^2 B_\star^2(\alpha_D)=\left(\frac{\Delta^2}{e v_F}\right)^2 \tag*{(3)}

and U\mathcal{U} is a symmetric matrix with det⁡(U)=1\det(\mathcal{U})=1. Thereby, the terms UE\mathcal{U}\mathbf{E} and U−1B\mathcal{U}^{-1}\mathbf{B} in (2) are linear combinations of E\mathbf{E} and B\mathbf{B}, respectively. The Dirac magneton for the materials is μD=evF2/(2Δ)\mu_D = e v_F^2/(2\Delta), that replaces the Bohr magneton in the theory. All these quantities defined previously depend on the nature of the material. Some particular cases are resumed in the table I, where we have used the natural units (to make contact with the usually adopted units in condensed matter physics, we remind the reader that 1 Volt/m=2.27×10−24 GeV21\ \mathrm{Volt/m} = 2.27 \times10^{-24}\ \mathrm{GeV}^2 and 1 T=6.8×10−16 GeV21\ \mathrm{T} = 6.8 \times10^{-16}\ \mathrm{GeV}^2). The Fermi velocity can be determined by the relation vF=α/αDv_F = \alpha/\alpha_D for each material.

We propose a general non-linear electrodynamics (ED) governed by the Lagrangian density

LDM=−αD24π2Fln⁡[1−2Fβ2+2G2β3−2F](4)\mathcal{L}_{DM}=-\frac{\alpha_D}{24\pi^2}\mathcal{F}\ln\left[1-\frac{2\mathcal{F}}{\beta^2}+\frac{2\mathcal{G}^2}{\beta^3\sqrt{-2\mathcal{F}}}\right] \tag*{(4)}

where the β\beta-parameter is identified as the Schwinger’s electric field in the material, namely, β:=E⋆(αD)\beta:= E_\star(\alpha_D). Moreover, F\mathcal{F} and G\mathcal{G} are the Lorentz and gauge-invariants

F=−14Fμν2=12(E2−B2)(5)\mathcal{F}=-\frac{1}{4}F_{\mu\nu}^2=\frac{1}{2}\left(\mathbf{E}^2-\mathbf{B}^2\right) \tag*{(5)}
G=−14FμνF~μν=E⋅B(6)\mathcal{G}=-\frac{1}{4}F_{\mu\nu}\widetilde{F}^{\mu\nu}=\mathbf{E}\cdot\mathbf{B} \tag*{(6)}

in which FμνF_{\mu\nu} denotes the field strength tensor and F~μν=ϵμναβFαβ/2\widetilde{F}^{\mu\nu}=\epsilon^{\mu\nu\alpha\beta}F_{\alpha\beta}/2 corresponds to the dual tensor, while E\mathbf{E} and B\mathbf{B} are the electric and magnetic fields, respectively. Using the approximation ∣B∣≫∣E∣|\mathbf{B}| \gg|\mathbf{E}| and the relations (3), one can show that the Lagrangian density (4) reduces to the expression (1). If the magnitude of the magnetic field is ∣B∣>∣E∣|\mathbf{B}| > |\mathbf{E}|, the non-linear Lagrangian density (4) is real. Otherwise, if ∣E∣>∣B∣|\mathbf{E}| > |\mathbf{B}|, this theory can present complex terms inside the logarithmic argument. Notice also that eq. (4) goes to zero in the limit ∣E∣→∣B∣|\mathbf{E}| \rightarrow|\mathbf{B}|. It should be mentioned that the effective Lagrangian (1) is valid for Δ≠0\Delta\ne0, which implies into β≠0\beta\ne0 by the relation (3). Thus, the Lagrangian density (4) is valid for β≠0\beta\ne0 with any magnitude of the electromagnetic field, and the non-linear effects disappear when β→∞\beta\rightarrow\infty.

We add the usual Maxwell Lagrangian to the non-linear electrodynamics in eq. (4):

L=F−αD24π2Fln⁡[1−2Fβ2+2G2β3−2F](7)\mathcal{L}=\mathcal{F}-\frac{\alpha_D}{24\pi^2}\mathcal{F}\ln\left[1-\frac{2\mathcal{F}}{\beta^2}+\frac{2\mathcal{G}^2}{\beta^3\sqrt{-2\mathcal{F}}}\right] \tag*{(7)}

in which the usual Maxwell theory is recovered in the limit β→∞\beta\to\infty. Therefore, we have a non-linear ED defined in the Minkowski space-time, whose the Dirac material is spatial 3D. The action principle applied to eq. (7) yields the field equation

∂μ[ξ(F,G)Fμν+χ(F,G)F~μν]=0(8)\partial_\mu\left[\xi(\mathcal{F},\mathcal{G})F^{\mu\nu}+\chi(\mathcal{F},\mathcal{G})\widetilde{F}^{\mu\nu}\right]=0 \tag*{(8)}

where the coefficients ξ(F,G)\xi(\mathcal{F},\mathcal{G}) and χ(F,G)\chi(\mathcal{F},\mathcal{G}) are functions of E\mathbf{E} and B\mathbf{B}, according to

ξ(F,G)=1+αD48π22G2+4β−FF2G2+−Fβ(−2F+β2)−αD24π2ln⁡[1−2Fβ2+2G2β3−2F](9)\xi(\mathcal{F},\mathcal{G})=1+\frac{\alpha_D}{48\pi^2}\frac{\sqrt{2}\mathcal{G}^2+4\beta\sqrt{-\mathcal{F}}\mathcal{F}}{\sqrt{2}\mathcal{G}^2+\sqrt{-\mathcal{F}}\beta(-2\mathcal{F}+\beta^2)} -\frac{\alpha_D}{24\pi^2}\ln\left[1-\frac{2\mathcal{F}}{\beta^2}+\frac{2\mathcal{G}^2}{\beta^3\sqrt{-2\mathcal{F}}}\right] \tag*{(9)}
χ(F,G)=−αD6π2FG2G2+−2Fβ(−F+β2)(10)\chi(\mathcal{F},\mathcal{G})=-\frac{\alpha_D}{6\pi^2}\frac{\mathcal{F}\mathcal{G}}{2\mathcal{G}^2+\sqrt{-2\mathcal{F}}\beta(-\mathcal{F}+\beta^2)} \tag*{(10)}

As in the usual case, the dual tensor F~μν=ϵμναβFαβ/2\widetilde{F}^{\mu\nu}=\epsilon^{\mu\nu\alpha\beta}F_{\alpha\beta}/2 satisfies the Bianchi identity ∂μF~μν=0\partial_\mu\widetilde{F}^{\mu\nu}=0.

The modified Maxwell equations can be written in terms of the constitutive vectors,

∇⋅D=0,∇×E+∂B∂t=0(11)\nabla\cdot\mathbf{D}=0,\qquad\nabla\times\mathbf{E}+\frac{\partial\mathbf{B}}{\partial t}=0 \tag*{(11)}
∇⋅B=0,∇×H−∂D∂t=0(12)\nabla\cdot\mathbf{B}=0,\qquad\nabla\times\mathbf{H}-\frac{\partial\mathbf{D}}{\partial t}=0 \tag*{(12)}

where the components of D\mathbf{D} and H\mathbf{H} vectors are defined by Di=ϵij(E,B)EjD_i=\epsilon_{ij}(\mathbf{E},\mathbf{B})E_j and Hi=(μij)−1(E,B)BjH_i=(\mu_{ij})^{-1}(\mathbf{E},\mathbf{B})B_j, respectively. The permittivity symmetric tensor ϵij\epsilon_{ij} and the inverse of the permeability tensor (μij)−1(\mu_{ij})^{-1} are read below:

ϵij(E,B)={1+αD48π22G2+4β−FF2G2+−Fβ(−2F+β2)−αD24π2ln⁡[1−2Fβ2+2G2β3−2F]}δij−αD6π2FBiBj2G2+−2Fβ(−F+β2)(13)\epsilon_{ij}(\mathbf{E},\mathbf{B})=\left\{1+\frac{\alpha_D}{48\pi^2}\frac{\sqrt{2}\mathcal{G}^2+4\beta\sqrt{-\mathcal{F}}\mathcal{F}}{\sqrt{2}\mathcal{G}^2+\sqrt{-\mathcal{F}}\beta(-2\mathcal{F}+\beta^2)} -\frac{\alpha_D}{24\pi^2}\ln\left[1-\frac{2\mathcal{F}}{\beta^2}+\frac{2\mathcal{G}^2}{\beta^3\sqrt{-2\mathcal{F}}}\right]\right\}\delta_{ij} -\frac{\alpha_D}{6\pi^2}\frac{\mathcal{F}B_iB_j}{2\mathcal{G}^2+\sqrt{-2\mathcal{F}}\beta(-\mathcal{F}+\beta^2)} \tag*{(13)}
(μij)−1(E,B)={1+αD48π22G2+4β−FF2G2+−Fβ(−2F+β2)−αD24π2ln⁡[1−2Fβ2+2G2β3−2F]}δij+αD6π2FEiEj2G2+−2Fβ(−F+β2)(14)(\mu_{ij})^{-1}(\mathbf{E},\mathbf{B})=\left\{1+\frac{\alpha_D}{48\pi^2}\frac{\sqrt{2}\mathcal{G}^2+4\beta\sqrt{-\mathcal{F}}\mathcal{F}}{\sqrt{2}\mathcal{G}^2+\sqrt{-\mathcal{F}}\beta(-2\mathcal{F}+\beta^2)} -\frac{\alpha_D}{24\pi^2}\ln\left[1-\frac{2\mathcal{F}}{\beta^2}+\frac{2\mathcal{G}^2}{\beta^3\sqrt{-2\mathcal{F}}}\right]\right\}\delta_{ij} +\frac{\alpha_D}{6\pi^2}\frac{\mathcal{F}E_iE_j}{2\mathcal{G}^2+\sqrt{-2\mathcal{F}}\beta(-\mathcal{F}+\beta^2)} \tag*{(14)}

The inverse of the matrix (10b) yields the permeability tensor

μij=11+a[δij−bEiEj1+a+bE2](15)\mu_{ij}=\frac{1}{1+a}\left[\delta_{ij}-\frac{bE_iE_j}{1+a+bE^2}\right] \tag*{(15)}

in which, for convenience, we define

a=αD48π22G2+4β−FF2G2+β−F(−2F+β2)−αD24π2ln⁡[1−2Fβ2+2G2β3−2F](16)a=\frac{\alpha_D}{48\pi^2}\frac{\sqrt{2}\mathcal{G}^2+4\beta\sqrt{-\mathcal{F}}\mathcal{F}}{\sqrt{2}\mathcal{G}^2+\beta\sqrt{-\mathcal{F}}(-2\mathcal{F}+\beta^2)} -\frac{\alpha_D}{24\pi^2}\ln\left[1-\frac{2\mathcal{F}}{\beta^2}+\frac{2\mathcal{G}^2}{\beta^3\sqrt{-2\mathcal{F}}}\right] \tag*{(16)}
b=αD6π2F2G2+β−2F(−F+β2)(17)b=\frac{\alpha_D}{6\pi^2}\frac{\mathcal{F}}{2\mathcal{G}^2+\beta\sqrt{-2\mathcal{F}}(-\mathcal{F}+\beta^2)} \tag*{(17)}

The Lagrangian density of eq. (4) yields the electric permittivity tensor presented in eqs. (10a) and eq. (10b). It has two distinct eigenvalues, λ1\lambda_1 and λ2\lambda_2, which read: λ1=1+a\lambda_1=1+a (with degree of degeneracy 2), whereas λ2=1+a−bB2\lambda_2=1+a-b\mathbf{B}^2. By analyzing the positivity of the permittivity eigenvalues, and considering the situation such that the Dirac material shares an interface with some other material, dielectric or metal, we can pursue an investigation as to whether surface plasmons may show up in the interface. This inspection may be relevant if one has in mind to study whether polaritons may arise from the interaction between the (surface) plasmons in the interface and light. This is a potential application of the Lagrangian density (4). Thereby, the positive conditions for the eigenvalues 1+a>01+a>0, 1+a−bB2>01+a-b\mathbf{B}^2>0 and 1+a+bE2>01+a+b\mathbf{E}^2>0 insure that the previous tensors are positive.

The energy-momentum tensor associated with the equations (9a) and (9b) reads as below:

Θμρ=ξ(F,G)FμνFρν+ημρ[χ(F,G)G−L](18)\Theta^{\mu\rho}=\xi(\mathcal{F},\mathcal{G})F^{\mu\nu}F^\rho{}_\nu+\eta^{\mu\rho}\left[\chi(\mathcal{F},\mathcal{G})\mathcal{G}-\mathcal{L}\right] \tag*{(18)}

that satisfies the conservation law ∂μΘμρ=0\partial_\mu\Theta^{\mu\rho}=0. The components of the energy-momentum tensor are read

Θ00=12(E2+B2)+αD48π2(2G2+4β−FF)E22G2+−Fβ(−2F+β2)−αD24π2ln⁡[1−2Fβ2+2G2β3−2F]E2−αD6π2F(E⋅B)22G2+−2Fβ(−F+β2)(19)\Theta^{00}=\frac{1}{2}\left(\mathbf{E}^2+\mathbf{B}^2\right)+\frac{\alpha_D}{48\pi^2}\frac{\left(\sqrt{2}\mathcal{G}^2+4\beta\sqrt{-\mathcal{F}}\mathcal{F}\right)\mathbf{E}^2}{\sqrt{2}\mathcal{G}^2+\sqrt{-\mathcal{F}}\beta(-2\mathcal{F}+\beta^2)} -\frac{\alpha_D}{24\pi^2}\ln\left[1-\frac{2\mathcal{F}}{\beta^2}+\frac{2\mathcal{G}^2}{\beta^3\sqrt{-2\mathcal{F}}}\right]\mathbf{E}^2 -\frac{\alpha_D}{6\pi^2}\frac{\mathcal{F}(\mathbf{E}\cdot\mathbf{B})^2}{2\mathcal{G}^2+\sqrt{-2\mathcal{F}}\beta(-\mathcal{F}+\beta^2)} \tag*{(19)}
Θ0i={1+αD48π22G2+4β−FF2G2+−Fβ(−2F+β2)−αD24π2ln⁡[1−2Fβ2+2G2β3−2F]}(E×B)i(20)\Theta^{0i}=\left\{1+\frac{\alpha_D}{48\pi^2}\frac{\sqrt{2}\mathcal{G}^2+4\beta\sqrt{-\mathcal{F}}\mathcal{F}}{\sqrt{2}\mathcal{G}^2+\sqrt{-\mathcal{F}}\beta(-2\mathcal{F}+\beta^2)} -\frac{\alpha_D}{24\pi^2}\ln\left[1-\frac{2\mathcal{F}}{\beta^2}+\frac{2\mathcal{G}^2}{\beta^3\sqrt{-2\mathcal{F}}}\right]\right\}(\mathbf{E}\times\mathbf{B})^i \tag*{(20)}

The result (14a) requires that ∣B∣>∣E∣|\mathbf{B}|>|\mathbf{E}|, thus the energy density and the Poynting vector are real. The same condition must be imposed for the eigenvalues λ1=1+a\lambda_1=1+a and λ2=1+a−bB2\lambda_2=1+a-b\mathbf{B}^2 to be real. All results of the Maxwell ED are recovered in the limit β→∞\beta\to\infty.

In the electrostatic case, we consider B=0\mathbf{B}=0 in the previous equations. For a point-like charge, ρ(r)=Qδ(r)\rho(r)=Q\delta(r), and the Gauss law yields D=Qr^/r2\mathbf{D}=Q\hat{\mathbf{r}}/r^2. Using the permittivity tensor for B=0\mathbf{B}=0, the electrostatic field (E)(E) satisfies the following equation:

E[1+αD24π2E2β2−E2−αD24π2ln⁡(1−E2β2)]=Qr2(21)E\left[1+\frac{\alpha_D}{24\pi^2}\frac{E^2}{\beta^2-E^2}-\frac{\alpha_D}{24\pi^2}\ln\left(1-\frac{E^2}{\beta^2}\right)\right]=\frac{Q}{r^2} \tag*{(21)}

Notice that the limit β→∞\beta\to\infty reduces (15) to the usual electrostatic field of a point-like charge. We solve numerically the equation (15) to obtain the plot (1). The figure shows the dimensionless variable E/βE/\beta (electric field for point-like charge over the β\beta-parameter) versus x=rβx = r\sqrt{\beta} (radial distance times the squared root of β\beta-parameter) for the QED. In this plot, we use the fundamental charge in natural units Q=e=0.085Q = e = 0.085, and αD=α=1/137=0.0072\alpha_D = \alpha= 1/137 = 0.0072.

Figure 1. The dimensionless variable E/βE/\beta versus x=rβx = r\sqrt{\beta} for the QED.

The electric field is finite at the origin (r→0r \to0), and it is given by the critical electric field of the material, i. e., E(r=0)=β=2.95×1012 eV2=2.95 MeV2E(r = 0) = \beta= 2.95 \times10^{12}\ \mathrm{eV}^{2} = 2.95\ \mathrm{MeV}^{2} (in the case of QED). The energy density in the electrostatic case is

uE=12E2[1+αD12π2E2β2−E2−αD12π2ln⁡(1−E2β2)](22)u_E = \frac{1}{2}\mathbf{E}^{2}\left[1+\frac{\alpha_D}{12\pi^{2}}\frac{\mathbf{E}^{2}}{\beta^{2}-\mathbf{E}^{2}}-\frac{\alpha_D}{12\pi^{2}}\ln\left(1-\frac{\mathbf{E}^{2}}{\beta^{2}}\right)\right] \tag*{(22)}

that is positive if β>∣E∣\beta> |\mathbf{E}|, and the electrostatic energy stored in the electron is

U=2π∫λc∞r2 dr×E2[1+αD12π2E2β2−E2−αD12π2ln⁡(1−E2β2)](23)U = 2\pi\int_{\lambda_c}^{\infty}r^{2}\,dr\times\mathbf{E}^{2}\left[1+\frac{\alpha_D}{12\pi^{2}}\frac{\mathbf{E}^{2}}{\beta^{2}-\mathbf{E}^{2}}-\frac{\alpha_D}{12\pi^{2}}\ln\left(1-\frac{\mathbf{E}^{2}}{\beta^{2}}\right)\right] \tag*{(23)}

where λc=3.86×10−13 m=1.93 MeV−1\lambda_c = 3.86 \times10^{-13}\ \mathrm{m} = 1.93\ \mathrm{MeV}^{-1} is the electron’s Compton wavelength. Using the expression of the electric field for a like-point charge in (15), and that αD=α=1/137\alpha_D = \alpha= 1/137 for the QED, we solve numerically the integral (17) to obtain the finite electron’s self-energy

U=0.023 MeV.(24)U = 0.023\ \mathrm{MeV}. \tag*{(24)}

Wave propagation properties in a uniform magnetic background

We now examine wave propagation effects under a uniform magnetic field for the non-linear ED (6). Following a similar procedure of ref. [19], we start off our analysis by introducing a prescription for the external and uniform magnetic field through the gauge 4-potential Aμ=aμ+ABμA_\mu= a_\mu+ A_{B\mu}, where aμa_\mu is interpreted as the propagation 4-potential, and ABμA_{B\mu} is the potential associated with the external and uniform magnetic field B0\mathbf{B}_0. Consequently, the EM field-strength tensor is also decomposed as Fμν=fμν+FBμνF_{\mu\nu} = f_{\mu\nu} + F_{B\mu\nu}, in which fμν=∂μaν−∂νaμ=(−ei,−ϵijkbk)f^{\mu\nu} = \partial^\mu a^\nu- \partial^\nu a^\mu= (-e^i,-\epsilon^{ijk}b^k) denotes the EM field strength tensor of the propagating fields, whereas FBμν=∂μABν−∂νABμ=(0,−ϵijkB0k)F_B^{\mu\nu} = \partial^\mu A_B^\nu- \partial^\nu A_B^\mu= (0,-\epsilon^{ijk}B_0^k) corresponds to the field strength of the magnetic background. Using this prescription, the Lagrangian density (6) up to second order in the propagation gauge field leads to

Lmodel(2)=−14c1fμν2−14c2fμνf~μν+18QBμνκλfμνfκλ(25)\mathcal{L}^{(2)}_{\mathrm{model}}=-\frac{1}{4}c_1 f_{\mu\nu}^{2}-\frac{1}{4}c_2 f_{\mu\nu}\widetilde{f}^{\mu\nu}+\frac{1}{8}Q_{B\mu\nu\kappa\lambda}f^{\mu\nu}f^{\kappa\lambda} \tag*{(25)}

where f~μν=ϵμναβfαβ/2\widetilde{f}_{\mu\nu}=\epsilon_{\mu\nu\alpha\beta}f^{\alpha\beta}/2 is the propagation field strength dual tensor. In addition, we have defined the tensor QBμνκλQ_{B\mu\nu\kappa\lambda} at the background fields (BB) as follows

QBμνκλ=d1FBμνFBκλ+d2F~BμνF~Bκλ+d3FBμνF~Bκλ+d3F~BμνFBκλ(26)Q_{B\mu\nu\kappa\lambda}=d_1F_{B\mu\nu}F_{B\kappa\lambda}+d_2\widetilde{F}_{B\mu\nu}\widetilde{F}_{B\kappa\lambda}+d_3F_{B\mu\nu}\widetilde{F}_{B\kappa\lambda}+d_3\widetilde{F}_{B\mu\nu}F_{B\kappa\lambda} \tag*{(26)}

with F~Bμν=ϵμναβFBαβ/2\widetilde{F}_{B\mu\nu}=\epsilon_{\mu\nu\alpha\beta}F_B^{\alpha\beta}/2 being the dual tensor of the background field. The coefficients of the expansion cic_i (i=1,2i=1,2) and did_i (i=1,2,3i=1,2,3) are given by

c1=∂L∂F∣B0,c2=∂L∂G∣B0,d1=∂2L∂F2∣B0,c_1=\left.\frac{\partial\mathcal{L}}{\partial\mathcal{F}}\right|_{B_0},\qquad c_2=\left.\frac{\partial\mathcal{L}}{\partial\mathcal{G}}\right|_{B_0},\qquad d_1=\left.\frac{\partial^2\mathcal{L}}{\partial\mathcal{F}^2}\right|_{B_0},
d2=∂2L∂G2∣B0,d3=∂2L∂F∂G∣B0(27)d_2=\left.\frac{\partial^2\mathcal{L}}{\partial\mathcal{G}^2}\right|_{B_0},\qquad d_3=\left.\frac{\partial^2\mathcal{L}}{\partial\mathcal{F}\partial\mathcal{G}}\right|_{B_0} \tag*{(27)}

For the particular model in eq. (6), we obtain the coefficients in a magnetic background:

c1=1−αD24π2B02β2+B02−αD24π2ln⁡(1+B02β2),c_1=1-\frac{\alpha_D}{24\pi^2}\frac{\mathbf{B}_0^2}{\beta^2+\mathbf{B}_0^2}-\frac{\alpha_D}{24\pi^2}\ln\left(1+\frac{\mathbf{B}_0^2}{\beta^2}\right),
c2=0,d1=αD12π2B02+2β2(B02+β2)2,c_2=0,\qquad d_1=\frac{\alpha_D}{12\pi^2}\frac{\mathbf{B}_0^2+2\beta^2}{(\mathbf{B}_0^2+\beta^2)^2},
d2=αD24π2∣B0∣/βB02+β2,d3=0.(28)d_2=\frac{\alpha_D}{24\pi^2}\frac{|\mathbf{B}_0|/\beta}{\mathbf{B}_0^2+\beta^2},\qquad d_3=0. \tag*{(28)}

Taking into account the non-null coefficients, the Lagrangian density (19) can be recast as

L(2)=−14c1fμν2+d18(FBμνfμν)2+d28(F~Bμνfμν)2.(29)\mathcal{L}^{(2)}=-\frac{1}{4}c_1f_{\mu\nu}^{2}+\frac{d_1}{8}\left(F_{B\mu\nu}f^{\mu\nu}\right)^2+\frac{d_2}{8}\left(\widetilde{F}_{B\mu\nu}f^{\mu\nu}\right)^2. \tag*{(29)}

The action principle applied to this Lagrangian density yields the following field equation

∂μ[c1fμν−d12FBμν(FBκλfκλ)−d22F~Bμν(F~Bκλfκλ)]=0,(30)\partial^\mu\left[c_1f_{\mu\nu}-\frac{d_1}{2}F_{B\mu\nu}\left(F_{B\kappa\lambda}f^{\kappa\lambda}\right)-\frac{d_2}{2}\widetilde{F}_{B\mu\nu}\left(\widetilde{F}_{B\kappa\lambda}f^{\kappa\lambda}\right)\right]=0, \tag*{(30)}

and the dual tensor f~μν\widetilde{f}^{\mu\nu} satisfies the Bianchi identity ∂μf~μν=0\partial_\mu\widetilde{f}^{\mu\nu}=0. For the analysis of the propagation effects, we consider the Lagrangian density (23). The correspondent field equation are similar to eqs. (9a) and (9b), just exchanging (E,B)(\mathbf{E},\mathbf{B}) by (e,b)(\mathbf{e},\mathbf{b}), where D=ϵij(B0)ej\mathbf{D}=\epsilon_{ij}(\mathbf{B}_0)e_j and H=(μij)−1(B0)bj\mathbf{H}=(\mu_{ij})^{-1}(\mathbf{B}_0)b_j, in which the permittivity and the permeability tensors are functions of the magnetic background components:

ϵij(B0)=δij+d2c1B0iB0j,(31)\epsilon_{ij}(\mathbf{B}_0)=\delta_{ij}+\frac{d_2}{c_1}B_{0i}B_{0j}, \tag*{(31)}
μij(B0)=δij+d1B0iB0jc1−d1B02.(32)\mu_{ij}(\mathbf{B}_0)=\delta_{ij}+\frac{d_1B_{0i}B_{0j}}{c_1-d_1\mathbf{B}_0^2}. \tag*{(32)}

The eigenvalues of the permittivity and permeability matrices are given by

λ1ϵ=λ2ϵ=1,λ3ϵ(B0)=1+d2c1B02,(33)\lambda_{1\epsilon}=\lambda_{2\epsilon}=1,\quad\lambda_{3\epsilon}(\mathbf{B}_0)=1+\frac{d_2}{c_1}\mathbf{B}_0^2, \tag*{(33)}
λ1μ=λ2μ=1,λ3μ(B0)=c1c1−d1B02.(34)\lambda_{1\mu}=\lambda_{2\mu}=1,\quad\lambda_{3\mu}(\mathbf{B}_0)=\frac{c_1}{c_1-d_1\mathbf{B}_0^2}. \tag*{(34)}

In the limit β→∞\beta\to\infty, the matrices (25a) and (25b) reduce to identity matrix, and the eigenvalues go to one. Note that these limits are also recovered by turning off the magnetic background field. The permittivity matrix is positive for any background magnitude, while the permeability matrix is positive if we impose the conditions c1>0c_1>0 and c1>d1B02c_1>d_1\mathbf{B}_0^2. Under these conditions, the material medium behaves like a paramagnetic material. Using the relations ϵij=δij+χEij\epsilon_{ij}=\delta_{ij}+\chi_{Eij} and μij=δij+χMij\mu_{ij}=\delta_{ij}+\chi_{Mij}, the electric and magnetic susceptibility tensors are, respectively, read below:

χEij(B0)=d2c1B0iB0j,(35)\chi_{Eij}(\mathbf{B}_0)=\frac{d_2}{c_1}B_{0i}B_{0j}, \tag*{(35)}
χMij(B0)=d1B0iB0jc1−d1B02.(36)\chi_{Mij}(\mathbf{B}_0)=\frac{d_1B_{0i}B_{0j}}{c_1-d_1\mathbf{B}_0^2}. \tag*{(36)}

In addition, the eigenvalues of the susceptibility matrices are given by

χ1E=χ2E=0,χ3E(B0)=d2c1B02,(37)\chi_{1E}=\chi_{2E}=0,\quad\chi_{3E}(\mathbf{B}_0)=\frac{d_2}{c_1}\mathbf{B}_0^2, \tag*{(37)}
χ1M=χ2M=0,χ3M(B0)=1c1−d1B02.(38)\chi_{1M}=\chi_{2M}=0,\quad\chi_{3M}(\mathbf{B}_0)=\frac{1}{c_1-d_1\mathbf{B}_0^2}. \tag*{(38)}

where the magnetic susceptibility is positive if we also impose the condition c1>d1B02c_1>d_1\mathbf{B}_0^2. Contracting the field eq. (24) with fναf_{\nu\alpha} and using the Bianchi identity, we obtain the conservation law

∂μΘ αμ=0,(39)\partial_\mu\Theta^\mu_{\ \alpha}=0, \tag*{(39)}

with the following energy-momentum tensor

Θ αμ=c1fμνfνα−d12(FBμνfνα)(FBκλfκλ)−d22(F~Bμνfνα)(F~Bκλfκλ)−δ αμL(2).(40)\Theta^\mu_{\ \alpha}=c_1 f^{\mu\nu}f_{\nu\alpha}-\frac{d_1}{2}\left(F_B^{\mu\nu}f_{\nu\alpha}\right)\left(F_{B\kappa\lambda}f^{\kappa\lambda}\right)-\frac{d_2}{2}\left(\widetilde{F}_B^{\mu\nu}f_{\nu\alpha}\right)\left(\widetilde{F}_{B\kappa\lambda}f^{\kappa\lambda}\right)-\delta^\mu_{\ \alpha}\mathcal{L}^{(2)}. \tag*{(40)}

Notice that the eq. (29) is due to the uniform magnetic background field. The conserved components of the energy-momentum tensor are Θμ0=(Θ00,Θi0)\Theta^{\mu0}=(\Theta^{00},\Theta^{i0}) in which the energy density and the Poynting vector components can be written as

Θ00=12ϵij(B0)eiej+12(μij)−1(B0)bibj,(41)\Theta^{00}=\frac{1}{2}\epsilon_{ij}(\mathbf{B}_0)e_i e_j+\frac{1}{2}(\mu_{ij})^{-1}(\mathbf{B}_0)b_i b_j, \tag*{(41)}
Θi0=c1(e×b)i+d1(e⋅E)(e×B)i−d1(b⋅B)(e×B)i−d2(e⋅B)(e×E)i−d2(b⋅E)(e×E)i.(42)\Theta^{i0}=c_1(\mathbf{e}\times\mathbf{b})^i+d_1(\mathbf{e}\cdot\mathbf{E})(\mathbf{e}\times\mathbf{B})^i-d_1(\mathbf{b}\cdot\mathbf{B})(\mathbf{e}\times\mathbf{B})^i-d_2(\mathbf{e}\cdot\mathbf{B})(\mathbf{e}\times\mathbf{E})^i-d_2(\mathbf{b}\cdot\mathbf{E})(\mathbf{e}\times\mathbf{E})^i. \tag*{(42)}

Thereby, the energy density is so positive if it also satisfies the condition c1>d1B02c_1>d_1\mathbf{B}_0^2.

The plane wave solutions for the propagating fields e\mathbf{e} and b\mathbf{b} are given in the usual form:

e(r,t)=e0ei(k⋅r−ωt)andb(r,t)=b0ei(k⋅r−ωt),(43)\mathbf{e}(\mathbf{r},t)=\mathbf{e}_0e^{i(\mathbf{k}\cdot\mathbf{r}-\omega t)}\quad\text{and}\quad\mathbf{b}(\mathbf{r},t)=\mathbf{b}_0e^{i(\mathbf{k}\cdot\mathbf{r}-\omega t)}, \tag*{(43)}

in which the electric e0\mathbf{e}_0 and magnetic b0\mathbf{b}_0 amplitudes are related by the Faraday’s law as b0=(k×e0)/ω\mathbf{b}_0=(\mathbf{k}\times\mathbf{e}_0)/\omega, where k\mathbf{k} denotes the wave vector, and ω\omega is the frequency. Substituting these solutions in the field equations, the wave equation of the electric amplitude is

Mije0j=0,(44)M_{ij}e_{0j}=0, \tag*{(44)}

in which the matrix elements of MijM_{ij} are written as

Mij=[1−n2+d1c1(n×B0)2]δij+(1−d1c1B02)ninj+(d2c1−d1c1n2)B0iB0j+d1c1(B0⋅n)(B0inj+B0jni).(45)\begin{aligned} M_{ij}={}&\left[1-\mathbf{n}^2+\frac{d_1}{c_1}(\mathbf{n}\times\mathbf{B}_0)^2\right]\delta_{ij}\\ &+\left(1-\frac{d_1}{c_1}\mathbf{B}_0^2\right)n_i n_j+\left(\frac{d_2}{c_1}-\frac{d_1}{c_1}\mathbf{n}^2\right)B_{0i}B_{0j}\\ &+\frac{d_1}{c_1}(\mathbf{B}_0\cdot\mathbf{n})(B_{0i}n_j+B_{0j}n_i). \tag*{(45)} \end{aligned}

where we have defined n=k/ω\mathbf{n}=\mathbf{k}/\omega. The null determinant det⁡(M)=0\det(M)=0 yields the n\mathbf{n}-polynomial equation

[1−n2+d1c1(n×B0)2]×[(1+d2c1B02)(1−n2)+d2c1(n×B0)2]=0,(46)\left[1-\mathbf{n}^2+\frac{d_1}{c_1}(\mathbf{n}\times\mathbf{B}_0)^2\right]\times\left[\left(1+\frac{d_2}{c_1}\mathbf{B}_0^2\right)(1-\mathbf{n}^2)+\frac{d_2}{c_1}(\mathbf{n}\times\mathbf{B}_0)^2\right]=0, \tag*{(46)}

whose solutions provide two possible refractive indices

n1=c1c1−d1(B0×k^)2,(47)n_1=\sqrt{\frac{c_1}{c_1-d_1(\mathbf{B}_0\times\hat{\mathbf{k}})^2}}, \tag*{(47)}
n2=c1+d2B02c1+d2(B0⋅k^)2.(48)n_2=\sqrt{\frac{c_1+d_2\mathbf{B}_0^2}{c_1+d_2(\mathbf{B}_0\cdot\hat{\mathbf{k}})^2}}. \tag*{(48)}

Using the coefficients from eq. (22), the refractive indices are functions of the angle θ\theta between the magnetic background field and the wave propagation direction k^\hat{\mathbf{k}}. In the limit β→∞\beta\to\infty, both the refractive indices go to one. Moreover, in the regime of a strong magnetic field (∣B0∣≫β)(|\mathbf{B}_0|\gg\beta), the first solution is n1≃1n_1\simeq1, and the second refractive index depends only on the θ\theta-angle, namely,

n2≃∣sec⁡θ∣.(49)n_2\simeq|\sec\theta|. \tag*{(49)}

For the TaAs Dirac material listed in the Table I, β→0\beta\to0, and the refractive index have the same results, n1=1n_1=1 and n2=∣sec⁡θ∣n_2=|\sec\theta|, respectively. In the case of the QED, we consider αD=α=0.0073\alpha_D=\alpha=0.0073, and in this approximation, the refractive index are given by

n1≃1+α24π2B02+2β2(B02+β2)2(k^×B0)2,(50)n_1 \simeq1+\frac{\alpha}{24\pi^2}\frac{B_0^2+2\beta^2}{\left(B_0^2+\beta^2\right)^2}\left(\hat{\mathbf{k}}\times\mathbf{B}_0\right)^2, \tag*{(50)}
n2≃1+α24π2∣B0∣β(k^×B0)2B02+β2.(51)n_2 \simeq1+\frac{\alpha}{24\pi^2}\frac{|\mathbf{B}_0|}{\beta}\frac{\left(\hat{\mathbf{k}}\times\mathbf{B}_0\right)^2}{B_0^2+\beta^2}. \tag*{(51)}

In the regime of a strong magnetic field, the refraction index in the QED are reduced to the results:

n1≃1+α24π2sin⁡2θ,(52)n_1 \simeq1+\frac{\alpha}{24\pi^2}\sin^2\theta, \tag*{(52)}
n2≃1+α24π2∣B0∣βsin⁡2θ.(53)n_2 \simeq1+\frac{\alpha}{24\pi^2}\frac{|\mathbf{B}_0|}{\beta}\sin^2\theta. \tag*{(53)}

The first solution does not depend on the magnitude of the magnetic background, and it is function of the θ\theta-angle. This conclusion shows that the correspondent dispersion relations does not depend on the magnitude of the magnetic background field in the regime of a strong background, and it depends only on the θ\theta-angle.

Birefringence phenomenon in a uniform magnetic background

We now consider the birefringence phenomenon, which rises from the wave propagation associated with different amplitudes. In this case, we consider the magnetic background on the zz-direction, i.e., B0=B0z^\mathbf{B}_0=B_0\hat{\mathbf{z}}, and in the first situation, the plane wave solution for the electric field has an amplitude parallel to the magnetic background, where e(r,t)=e03z^ei(k⋅r−ωt)\mathbf{e}(\mathbf{r},t)=e_{03}\hat{\mathbf{z}}e^{i(\mathbf{k}\cdot\mathbf{r}-\omega t)}. Using this solution, the wave equation can be written as

[k2ω2−ϵ33(B0)μ22(B0)]e03=0,(54)\left[\frac{k^2}{\omega^2}-\epsilon_{33}(B_0)\mu_{22}(B_0)\right]e_{03}=0, \tag*{(54)}

in which the correspondent refractive index is

n∥(B0)=μ22(B0)ϵ33(B0)=1+d2c1B02.(55)n_{\parallel}(B_0)=\sqrt{\mu_{22}(B_0)\epsilon_{33}(B_0)}=\sqrt{1+\frac{d_2}{c_1}B_0^2}. \tag*{(55)}

The second situation happens when the plane wave has amplitude perpendicular to the magnetic background, such that e(r,t)=e02y^ei(k⋅r−ωt)\mathbf{e}(\mathbf{r},t)=e_{02}\hat{\mathbf{y}}e^{i(\mathbf{k}\cdot\mathbf{r}-\omega t)}. In this case, the wave equation is

[k2ω2−ϵ22(B0)μ33(B0)]e02=0,(56)\left[\frac{k^2}{\omega^2}-\epsilon_{22}(B_0)\mu_{33}(B_0)\right]e_{02}=0, \tag*{(56)}

with the refractive index

n⊥(B0)=μ33(B0)ϵ22(B0)=c1c1−d1B02.(57)n_{\perp}(B_0)=\sqrt{\mu_{33}(B_0)\epsilon_{22}(B_0)}=\sqrt{\frac{c_1}{c_1-d_1B_0^2}}. \tag*{(57)}

The birefringence phenomena rises from the difference

Δn(B0)=n∥(B0)−n⊥(B0)≠0.(58)\Delta n(B_0)=n_{\parallel}(B_0)-n_{\perp}(B_0)\ne0. \tag*{(58)}

Therefore, the birefringence does not occur if we constraint the condition

μ22(B0)ϵ33(B0)=ϵ22(B0)μ33(B0),(59)\mu_{22}(B_0)\epsilon_{33}(B_0)=\epsilon_{22}(B_0)\mu_{33}(B_0), \tag*{(59)}

and using the 22 and 33 components from eqs. (25a) and (25b), we obtain the relation:

d2−d1=d1d2c1B02.(60)d_2-d_1=\frac{d_1d_2}{c_1}B_0^2. \tag*{(60)}

In the limit β→∞\beta\to\infty, the birefringence disappears in eq. (58). For β≫B0\beta\gg B_0, this difference is residual in (B0/β)2(B_0/\beta)^2,

Δn(B0)≃−αD12π2(B0β)2.(61)\Delta n(B_0)\simeq-\frac{\alpha_D}{12\pi^2}\left(\frac{B_0}{\beta}\right)^2. \tag*{(61)}

The plot of Δn\Delta n versus the dimensionless variable x=B0/βx=B_0/\beta is illustrated in the Figure 2, for the Dirac materials from the table I: BiSb (blue line) and PbSnTe (red line). The lines intercept the horizontal axis where the birefringence is null in accord with the condition (46). Using the coefficients (22) and the physical parameters in the table I, we obtain the magnitude of the magnetic background field for the QED and the Dirac materials in which the birefringence is null:

Figure 2. The variation of the refractive index (Δn\Delta n) versus the dimensionless variable x=B0/βx=B_0/\beta for the Dirac materials BiSb (blue line) and PbSnTe (red line).

B0=4.0 MeV2=5.6 GTfor QED,(62)B_0=4.0\ \mathrm{MeV}^2=5.6\ \mathrm{GT}\quad\text{for QED}, \tag*{(62)}
B0=8.90 eV2=12 mTfor PbSnTe,(63)B_0=8.90\ \mathrm{eV}^2=12\ \mathrm{mT}\quad\text{for PbSnTe}, \tag*{(63)}
B0=0.18 eV2=0.25 mTfor BiSb.(64)B_0=0.18\ \mathrm{eV}^2=0.25\ \mathrm{mT}\quad\text{for BiSb}. \tag*{(64)}

Otherwise, the condition (44) is kept for values of B0B_0 different from the results (48a)-(48c).

Interaction Energy for an Electrodynamics in Dirac Materials under a Uniform Magnetic Field

As already expressed, here we calculate the interaction energy between static point-like sources for an electrodynamics in Dirac materials under a uniform magnetic field, along the lines of refs. [20, 21, 22, 23, 24]. To do that, we compute the expectation value of the energy operator HH in the physical state ∣Φ⟩\lvert\Phi\rangle, which we write as ⟨H⟩Φ\langle H\rangle_{\Phi}.

In this perspective, our aim here is to evaluate to the lowest-order the interaction energy. For this purpose, we linearize the effective theory (6) in a similar manner to that which led to the eq. (29). Furthermore, in what follows, we confine our attention to the case of a pure and uniform magnetic background (E=0\mathbf{E}=0), as in the section III. In such a case, we back to the effective Lagrangian density in eq. (29).

We now carry out a Hamiltonian analysis of this theory. The canonical momenta following from eq. (29) are

Πμ=−c1f0μ+d12FBρσFρσFB0μ+d22F~BρσfρσF~B0μ,(65)\Pi^\mu=-c_1f^{0\mu}+\frac{d_1}{2}F_B^{\rho\sigma}F_{\rho\sigma}F_B^{0\mu}+\frac{d_2}{2}\widetilde{F}_B^{\rho\sigma}f_{\rho\sigma}\widetilde{F}_B^{0\mu}, \tag*{(65)}

which produces the usual primary constraint Π0=0\Pi^0=0, and

Πi=c1(δij+d2c1B0iB0j)ej.\Pi_i=c_1\left(\delta_{ij}+\frac{d_2}{c_1}B_{0i}B_{0j}\right)e_j.

Making use of this last equation, we readily verify that the electric field associated to the aμa^\mu-field takes the form

ei=1c1det⁡D(δijdet⁡D−d2c1B0iB0j)Πj.(66)e_i=\frac{1}{c_1\det D}\left(\delta_{ij}\det D-\frac{d_2}{c_1}B_{0i}B_{0j}\right)\Pi_j. \tag*{(66)}

Here det⁡D=1+d2c1B02\det D=1+\frac{d_2}{c_1}\mathbf{B}_0^2, where B0\mathbf{B}_0 represents the external (background) magnetic field around which the aμa^\mu-field fluctuates.

The canonical Hamiltonian is now obtained in the usual way by a Legendre transformation to obtain

HC=∫d3x[Πi∂ia0+Π22c1+c12b2−d12(B0⋅b)2]−∫d3xd22c12det⁡D(B0⋅Π)2.(67)H_C=\int d^3x\left[\Pi_i\partial^i a_0+\frac{\boldsymbol{\Pi}^2}{2c_1}+\frac{c_1}{2}\mathbf{b}^2-\frac{d_1}{2}(\mathbf{B}_0\cdot\mathbf{b})^2\right]-\int d^3x\frac{d_2}{2c_1^2\det D}(\mathbf{B}_0\cdot\boldsymbol{\Pi})^2. \tag*{(67)}

It can be quickly checked that the preservation in time of the primary constraint, Π0\Pi_0, leads to the secondary constraint Γ1≡∂iΠi=0\Gamma_1\equiv\partial_i\Pi^i=0 (Gauss constraint). In this case, there are two constraints, which are first-class ones. According to the general theory, we obtain the extended Hamiltonian, which generates translations in time, by adding all the first-class constraints with arbitrary coefficients to the canonical Hamiltonian. Accordingly, we write

H=HC+∫d3x[c0(x)Π0(x)+c1(x)Γ1(x)],(68)H=H_C+\int d^3x\left[c_0(x)\Pi_0(x)+c_1(x)\Gamma_1(x)\right], \tag*{(68)}

where c0(x)c_0(x) and c1(x)c_1(x) are arbitrary functions of the space-time coordinates. Since Π0=0\Pi^0=0 always, neither a0a^0 nor Π0\Pi^0 are of interest in describing the system and may be excluded from the theory. Thus the variables a0a^0 and Π0\Pi^0 play no role in the theory. If a new arbitrary coefficient c(x)=c1(x)−a0(x)c(x)=c_1(x)-a_0(x) is introduced the Hamiltonian may be rewritten in the form

H=∫d3x[c(x)∂iΠi+Π22c1+c12b2−d12(B0⋅b)2]−∫d3xd22c12det⁡D(B0⋅Π)2.(69)H=\int d^3x\left[c(x)\partial^i\Pi_i+\frac{\boldsymbol{\Pi}^2}{2c_1}+\frac{c_1}{2}\mathbf{b}^2-\frac{d_1}{2}(\mathbf{B}_0\cdot\mathbf{b})^2\right]-\int d^3x\frac{d_2}{2c_1^2\det D}(\mathbf{B}_0\cdot\boldsymbol{\Pi})^2. \tag*{(69)}

Since there is one first constraint, Γ1\Gamma_1, we choose a gauge-fixing condition that will make the full set of constraints to become second class. To do this, we adopt the gauge discussed previously [20], that is,

Γ2(x)≡∫Cζdzνaν(z)=∫01dλ xiai(λx)=0.(70)\Gamma_2(x)\equiv\int_{C_{\zeta}}d z^\nu a_\nu(z)=\int_0^1d\lambda\,x^i a_i(\lambda x)=0. \tag*{(70)}

Here λ\lambda (0≤λ≤10\leq\lambda\leq1) is the parameter describing the space-like straight path xi=ζi+λ(x−ζ)ix^i=\zeta^i+\lambda(x-\zeta)^i, and ζi\zeta^i is a fixed point (reference point). There is no essential loss of generality if we restrict our considerations to ζi=0\zeta^i=0. With such a choice, the fundamental Dirac bracket is given by

{ai(x),Πj(y)}∗=δijδ(3)(x−y)−∂ix∫01dλ xjδ(3)(λx−y).(71)\{a_i(x),\Pi^j(y)\}^{*}=\delta_i^j\delta^{(3)}(x-y)-\partial_i^x\int_0^1d\lambda\,x^j\delta^{(3)}(\lambda x-y). \tag*{(71)}

We now recall that the physical states ∣Φ⟩\lvert\Phi\rangle are gauge-invariant. In this case, we may therefore write the stringy gauge-invariant state

∣Φ⟩≡∣Ψ‾(y)Ψ(y′)⟩=Ψ‾(y)exp⁡[iq∫y′ydziai(z)]Ψ(y′)∣0⟩,(72)\lvert\Phi\rangle\equiv\lvert\overline{\Psi}(y)\Psi(y')\rangle =\overline{\Psi}(y)\exp\left[iq\int_{y'}^{y}dz^i a_i(z)\right]\Psi(y')\lvert0\rangle, \tag*{(72)}

where the line integral is along a space-like path on a fixed time slice, qq is the fermion charge and ∣0⟩\lvert0\rangle is the physical vacuum state.

At this point, we have all the elements necessary for the calculation of the expectation value ⟨H⟩Φ\langle H\rangle_{\Phi}. From the foregoing discussion we readily see that ⟨H⟩Φ\langle H\rangle_{\Phi} can therefore be written as follows

⟨H⟩Φ=⟨H⟩0+⟨H⟩Φ(1),(73)\langle H\rangle_{\Phi}=\langle H\rangle_0+\langle H\rangle_{\Phi}^{(1)}, \tag*{(73)}

where ⟨H⟩0=⟨0∣H∣0⟩\langle H\rangle_0=\langle0\lvert H\lvert0\rangle, whereas the ⟨H⟩0(1)\langle H\rangle_0^{(1)} term is given by

⟨H⟩Φ(1)=⟨Φ∣∫d3x[Π22c1−d22c12det⁡D(B0⋅Π)2]∣Φ⟩,(74)\langle H\rangle_{\Phi}^{(1)}=\langle\Phi\lvert\int d^3x\left[\frac{\boldsymbol{\Pi}^2}{2c_1}-\frac{d_2}{2c_1^2\det D}(\mathbf{B}_0\cdot\boldsymbol{\Pi})^2\right]\rvert\Phi\rangle, \tag*{(74)}

Following our earlier procedure [23, 24], the static potential profile for two opposite charges located at yy and y′y' then reads

V(L)=−qeff2(B0)4π1L,(75)V(L)=-\frac{q_{\mathrm{eff}}^2(\mathbf{B}_0)}{4\pi}\frac{1}{L}, \tag*{(75)}

where the effective charge is

qeff(B0)=∣q∣c1+d2B02(76)q_{ef f}(\mathbf{B}_0) = \frac{|q|}{\sqrt{c_1+d_2\mathbf{B}_0^2}} \tag*{(76)}

and L≡∣y−y′∣L \equiv|\mathbf{y}-\mathbf{y}'| is the distance that separates the two charges. We see, therefore, that at the lowest-order in our effective model, the introduction of the non-linearities induce redefinition in the charges. Whenever ∣B0∣→0|\mathbf{B}_0| \to0, the effective charge qeffq_{ef f} reduces to the usual qq-charge. This result agrees with the absence of screening in the Coulomb potential that is a characteristic of Dirac materials. Under an intense magnetic background, qeffq_{ef f} goes to zero, and consequently, the potential (59) signals a situation of free particles. The dimensionless quantity q~=qeff/∣q∣\tilde{q}=q_{ef f}/|q| is plotted as function of B0/βB_0/\beta in Figure 3, for the QED case (black line) and for the Dirac materials BiSb (blue line) and PbSnTe (red line).

Figure 3. The dimensionless effective charge q~\tilde{q} as function of the ratio B0/βB_0/\beta (magnetic background field over critical field) for the QED (black line), and for the Dirac materials BiSb (blue line) and PbSnTe (red line).

The QED curve is approximately horizontal since the contribution of the fine structure constant is very weak in this case. When the fine structure constant increases (that are the cases of the BiSb and PbSnTe), there arises a peak showed in the blue and red lines, respectively. In the case of the PbSnTe, the peak is localized at magnetic background B0=4.70 eV2B_0=4.70\ \mathrm{eV}^2, where the effective charge is maximum at qeff=0.3035q_{ef f}=0.3035 for two electrons interacting under the potential (59).

Concluding Comments and Considerations

In this contribution, a non-linear electrodynamics (ED) similar to a non-linear logarithmic model is proposed as a possible formulation from which there follows the effective Lagrangian density in ref. [3], obtained from the one-loop radiative corrections for Dirac materials. We highlight that, in the regime of a strong magnetic

field related to electric field, the non-linear ED under consideration reduces to the situation presented in ref. [3]. The non-linear model introduces a critical electric field as a parameter that depends on the characteristics of the particular Dirac material. The usual Maxwell electrodynamics is recovered in the limit in which the critical electric field (β\beta-parameter) goes to infinity. Using the non-linear Lagrangian density, we obtain the non-linear field equations, the permittivity and permeability tensors, and the correspondent energy-momentum tensor.

Afterwards, we study the aspects of the model when the system is submitted to an external (and uniform) magnetic field. We expand the non-linear Lagrangian up to second order in the propagating electromagnetic field. Thereby, the permittivity and permeability tensors are obtained in terms of the magnetic background field components. The refractive indices are calculated as function of the β\beta-parameter and magnetic background field. We observe that it also depends on the angle (θ\theta) in that the propagation direction does with magnetic background field. In the regime of a strong magnetic background, one of the solutions for the refractive index depends only on the θ\theta-angle. Furthermore, we investigate the birefringence phenomenon for the wave electric field propagating in this non-linear theory under a uniform magnetic background. We calculate numerically the magnetic background values for which birefringence does not take place; to do that, we must use physical characteristics of the particular Dirac material.

In the paper of Ref. [25], the authors inspect the interesting question of the propagation of an undamped transverse electric mode due to charge density fluctuations in a type of material known as tilted Dirac-cone material (TDM). Charge-density excitations referred to as plasmons are actually present in Dirac semimetals; in the work cited as Ref. [26], the authors discuss possible practical applications of plasmon excitations with frequency range in the THz region of the spectrum. Stimulated by the results of the references quoted above and by the interest in studying surface plasmons in Dirac semimetals, we have in project to re-assess these issues by applying our developments based on the Lagrangian density of eq. (4) to specific Dirac-type materials. The permittivity tensor presented in eq. (10a) plays a key role in our endeavor.

In Section IV, we devote efforts to investigate the phenomenon of birefringence associated to the model described by the Lagrangian density of eq. (4). By considering the graph depicted in Figure 2, we may propose, as a possible experimental observation, the measurement of the birefringence as given by Δn(B0)\Delta n(B_0), according to eq. (58). As the graph illustrates, for laboratory magnetic fields, corresponding to values of xx between 2.0 and 2.5, for example, Δn\Delta n presents accessible values for the BiSb and PbSnTe materials. This could provide a way of testing our results on birefringence.

To conclude, we apply the path-dependent formalism to the linearized Lagrangian to obtain the interaction energy between two static point-like charges. There results a Coulomb-like potential with an effective charge that runs with the magnetic background field. For the PbSnTe Dirac material, the effective charge has a smooth peak at a magnetic field of 4.70 eV24.70\ \mathrm{eV}^{2}, as illustrated in Figure 3.

Taking a forward-looking perspective, as a possible path to further exploit the effective photonic model in discussion here, we believe that a natural follow-up would be the inspection of the regime of vortex formation and the consequent study of the properties of the magnetic vortices governed by the set of the non-linear electromagnetic field equations. As a consequence of the non-linearity, charged vortices might naturally arise and investigating the dynamics of fermions in presence of such vortices may be a research question of interest in the context of Dirac materials.

Acknowledgments

L.P.R. Ospedal is grateful to FAPERJ (grant number E-26/203.997/2022) for his post-doctoral fellowship. P. Gaete was partially supported by ANID PIA/APOYO AFB220004.

References

  1. [1]N. Nagaosa, Quantum Field Theory in Condensed Matter Physics, Springer, 1999.
  2. [2]A. M. Tsvelik, Quantum Field Theory in Condensed Matter Physics, Cambridge, 2003.
  3. [3]Aydin C. Keser, Yuli Lyanda-Geller and Oleg P. Sushkov, Phys. Rev. Lett. 128, 066402, 2022.
  4. [4]H. Euler and W. Heisenberg, Z. Phys. 98, 714 (1936).
  5. [5]J. Schwinger, Phys. Rev. 82, 664 (1951).
  6. [6]S.L. Adler, Ann. Phys. (N.Y.) 67, 599 (1971).
  7. [7]C. Costantini, B. De Tollis, G. Pistoni, Nuovo Cimento A 2, 733 (1973).
  8. [8]R. Ruffini, G. Vereshchagin, S.-S. Xue, Phys. Rep. 487, 1-140 (2010).
  9. [9]G.V. Dunne, Int. J. Mod. Phys. Conf. Ser. 14, 42 (2012).
  10. [10]R.Battesti, C.Rizzo, Rep.Prog.Phys. 76, 016401 (2013).
  11. [11]X. Sarazin, F. Couchot, A. Djannati-Atai, O. Guilbaud, S. Kazamias, M. Pittman, M. Urban, Eur. Phys. J. D 70, 13 (2016).
  12. [12]ATLAS Collaboration, Nat. Physical. 13, 852 (2017).
  13. [13]CMS Collaboration, Phys. Lett. B 797, 134826 (2019).
  14. [14]L. Schoeffel, C. Baldenegro, H. Hamdaoui, S. Hassani, C. Royon and M. Saimpert, Prog. Part. Nucl. Phys. 120, 103889 (2021).
  15. [15]S. Robertson, A. Mailliet, X. Sarazin, F. Couchot, E. Baynard, J. Demailly, M. Pittman, A. Djannati-Atai, S. Kazamias, and M. Urban, Phys. Rev. A 103, 023524 (2021).
  16. [16]R. Battesti et al, Phys. Rept. 765-766, 1-39, (2018).
  17. [17]S. Ataman, Phys. Rev. A 97, 063811 (2018).
  18. [18]T. O. Wehling, A. M. Black-Schaffer and A. V. Balatsky, Dirac materials, Adv. Phys. 76 (2014) 1.DOI
  19. [19]M. J. Neves, Jorge B. de Oliveira, L. P. R. Ospedal and J. A. Helayël-Neto, Phys. Rev. D 104 (2021) 015006.
  20. [20]P. Gaete, Z. Phys. C 76 355 (1997).
  21. [21]P. Gaete and J. Helayël-Neto, Eur. Phys. J. C 74, 3182 (2014).
  22. [22]P. Gaete and J. A. Helayël-Neto, Eur. Phys. J. C 81, 899 (2021).
  23. [23]P. Gaete, J.A. Helayël-Neto and L.P.R. Ospedal, Adv. High Energy Phys. 2021, 6621975 (2021).
  24. [24]M. J. Neves, L. P. R. Ospedal, J. A. Helayël-Neto and P. Gaete, Eur. Phys. J. C 82, no.4, 327 (2022).
  25. [25]Z. Jalali-Mola and S. A. Jafari, Phys. Rev. B 102 (2020) 245148.
  26. [26]D. E. Kharzeev, R. D. Pisarski and H.-U. Yee, Phys. Rev. Lett. 115 (2015) 236402.

Paper details

vlnc:W_74zsc6pg
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents