Production of Neutral Fermion in Linear Magnetic Field through Pauli Interaction Hyun Kyu Lee and Yongsung Yoon Department of Physics, Hanyang University, Seoul 133-791, Korea Abstract We calculate the production rate of neutral fermions in linear magnetic fields through the Pauli 6 0 interaction. It is found that the production rate is exponentially decreasing function with respect 0 2 to the inverse of the magnetic field gradient, which shows the non-perturbative characteristics n a analogous to the Schwinger process. It turns out that the production rate density dependson both J 1 the gradient and the strength of magnetic fields in 3+1 dimension. It is quite different from the 1 result in 2+1 dimension, where the production rate depends only on the gradient of the magnetic 2 v fields, not on the strength of the magnetic fields. It is also found that the production of neutral 1 8 fermions through the Pauli interaction is a magnetic effect whereas the production of charged 1 0 1 particles through minimal coupling is an electric effect. 5 0 / PACS numbers: 82.20.Xr,13.40.-f, 12.20.Ds h t - Keywords: production rate, pauli term, neutral fermion, magnetic dipole moment, neutrino p e h : v i X r a 1 I. INTRODUCTION It is well known that the interaction of charged spin-1/2 fermions with the electromag- netic field is described by the minimal coupling in the form of Dirac equation. Pauli [1] suggested a non-minimal coupling of spin 1/2-particle with electromagnetic fields, which can be interpreted as an effective interaction of fermion to describe the anomalous mag- netic moment of fermions. Pauli interaction is particularly interesting for describing the interaction of neutral particle with electromagnetic field provided it has a non-vanishing magnetic dipole moment [2]. One of the immediate possibilities [4] is the electromagnetic interaction(Pauli interaction) of neutrinos, which are recently confirmed to have a non-zero mass with mixing [3]. The presence of the magnetic dipole moment implies that neutrino can directly couple to the electromagnetic field, which leads to a variety of new processes [5, 6]. One of the interesting phenomena with the strong electromagnetic field configuration is the pair creation of particles. The well known example is the Schwinger process with the minimal coupling, in which charged particles are created in pairs [7] under a strong electric field. However it has been demonstrated that no particle creation is possible under the pure magnetic field configuration even with the spatial inhomogeneity [8]. For a neutral particle with the Pauli interaction, the inhomogeneity of the magnetic field coupled directly to the magnetic dipole moment plays an interesting role analogous to the electric field for a charged particle. The non-zero gradient of the magnetic field can exert a force on a magnetic dipole moment such that the neutral fermion can get an energy out of the magnetic field. Hence it will affects the vacuum structure greatly for the strong enough magnetic field as for the case of charged particles in the strong electric field. It is then interesting to see whether the vacuum production of neutral fermion with a non-zero magnetic moment in an inhomogeneous magnetic field is possible on the analogy of the Schwinger process. Interestingly it has been demonstrated in 2+1 dimension that the magnetic dipole coupled to the field gradient induces pair creations in a vacuum [9]. In this work, we will present a realistic calculation in 3+1 dimension for the pair production rate of neutral fermions with non-vanishing magnetic dipole moment. 2 II. PRODUCTION RATE OF NEUTRAL FERMIONS THROUGH PAULI IN- TERACTION The simplest Lagrangian for the neutral fermion, which couples to the external electro- magnetic field, was suggested by Pauli long time ago [1]. The Dirac equation with Pauli term is given by µ = ψ¯(/p+ σµνF m)ψ, (1) µν L 2 − where σµν = i[γµ,γν], g = (+, , , ). µ in the Pauli term measures the magnitude of 2 µν − − − the magnetic dipole moment of fermion. Pauli term can be considered as an effective inter- action which describes anomalous magnetic moment of fermion or as an effective magnetic moment induced by the bulk fermions in a theory with large extra dimensions [10]. The corresponding Pauli Hamiltonian operator is H = α~ (p~ iµβE~)+β(m µ~σ B~), (2) · − − · where σi = 1ǫijkσjk. If the magnetic field is stronger than the critical field, B m/µ, a 2 c ≡ negative energy state of m µB appears. Thus, we will consider magnetic fields weaker − | | than the critical magnetic field, B < B . c In general, the effective potential, V (A), for a background electromagnetic vector po- eff tential, A , can be obtained by integrating out the fermion: µ µ 1 i d4xV (A[x]) = Trln (/p+ σµνF m) , (3) eff µν { 2 − /p m} Z − where F = ∂ A ∂ A . The decay probability of the background magnetic field into the µν µ ν ν µ − neutral fermions is related to the imaginary part of the effective potential V (A), eff P = 1 ei d4xVeff(A[x]) 2 = 1 e−2Im d3xdtVeff(A[x]). (4) −| | − R R That is, the twice of the imaginary part of the effective potential V (A[x]) is the fermion eff production rate per unit volume [11]: w(x) = 2Im(V (A[x])). eff Using the charge conjugation matrix C: Cγ C−1 = γT, CσµνC−1 = σTµν, (5) µ − µ − and the identity a ∞ ds ln = (eis(b+iǫ) eis(a+iǫ)), (6) b s − Z0 3 we can write the effective potential V (A[x]) as follows eff i ∞ ds Veff(A[x]) = e−ism2tr(< x eis(p/+µ2σµνFµν)2 x > < x eisp2 x >). (7) 2 s | | − | | Z0 For an inhomogeneous magnetic field, we consider a static magnetic field configuration of zˆ-direction with a constant gradient along xˆ-direction, B~ = B(x)zˆ, such that B F = B(x) = B +B′x = B′x˜, (x˜ = x +x, x = 0). (8) 12 0 0 0 B′ It is not necessary to consider an infinitely extended ever-increasing magnetic field to meet the linear magnetic field configuration. Because the particle production rate density is a local quantity, it is sufficient to have a uniform gradient magnetic field in the Compton wavelength scale of the particle. For this background magnetic field, we get µ (/p+ σµνF )2 = p2 p2 +(µB′x˜+iγ˜3p +iγ˜0p )2 µB′γ2 , (9) 2 µν − 1 − 2 0 3 − where γ˜3 γ0γ1γ2 and γ˜0 γ1γ2γ3. ≡ ≡ The second term in Eq.(7) is easily calculated to be i v(0) tr < x eisp2 x >= . (10) ≡ | | −4π2s2 Inserting a complete set of momentum eigenstates, the first term of Eq.(7) can be written by v(A) tr < x e−is{p21+p22−(µB′x˜+iγ˜3p0+iγ˜0p3)2+µB′γ2} x > ≡ | | 1 π = (2π)4(is)21tr dp1dp′1dp0dp3ei(p1−p′1)x˜ < p1|e−is(p21−µ2B′2x2)|p′1 > × Z eµB1′(γ˜0p3+γ˜3p0)(p′1−p1)e−isµB′γ2, (11) where we have used [γ˜3,γ2] = 0 = [γ˜0,γ2], [p ,x˜] = i and 1 − p2 +p2 (µB′x˜+iγ˜3p +iγ˜0p )2 = 1 2 − 0 3 e−µB1′(γ˜0p3+γ˜3p0)p1eix0p1(p2 +p2 (µB′x)2)eµB1′(γ˜0p3+γ˜3p0)p1e−ix0p1. (12) 1 2 − Using the properties of gamma matrices γ˜0,γ˜3 = 0, tr(γ˜3γ2) = 0 = tr(γ˜0γ2), (13) { } 4 we get treµB1′(γ˜0p3+γ˜3p0)(p′1−p1)e−isµB′γ2 = 4cosh(sµB′)cos (p′1 −p1)(p2 p2)1/2 . (14) { µB′ 0 − 3 } Thus, v(A) can be written as follows 4 π v(A) = (2π)4(is)21 cosh(sµB′) dp1dp′1dp0dp3ei(p1−p′1)x˜ × Z (p p′) cos{ 1µ−B′ 1 (p20 −p23)1/2} < p1|e−is(p21−µ2B′2x2)|p′1 > . (15) The matrix elements in momentum space in Eq.(15) correspond to the matrix elements of the evolution operator for the simple harmonic oscillator with an imaginary frequency, ω = 2iµB′ and m = 1. Then we get 2 < p1|e−is(p21−µ2B′2x2)|p′1 >= dx′dx′′ < p1|x′′ > U(x′′,s;x′,0) < x′|p′1 >, (16) Z where U(x,s;x′,0) is given by U(x,s;x′,0) < x,s e−is(p21+12ω2x2) x′,0 >= ( ω )1/2eiω(x2+x′24)scinosωωss−2xx′. (17) ≡ | | 4πisinωt Performing the x′ and x′′ integration explicitly, we get < p1|e−is(p21−µ2B′2x2)|p′1 >= (iπα)122αs1inωse−8αi((p11−−cpos′1)ω2s)e8αi((p11++cpos′1)ω2s), (18) where α = ω . −4sinωs Inserting Eq.(18) into Eq.(15), the integration over p and p′ gives 1 1 2 π v(A) = ( )21 coth(sµB′) , (19) (2π)4 is K where, defining γ,a and p as follow − coth(sµB′) 1 γ ≡ 4µB′ , a ≡ µB′(p20 −p23)21, p=p1 −p′1, (20) is given by K 1 ∞ dp dp dp eix˜p−(eiap− +e−iap−)e−iγp2−, − 0 3 K ≡ 2 Z−∞ = 1( π )1/2 dp0dp3(ei(a+4γx˜)2 +ei(a−4γx˜)2) 2 iγ Z = 4π(µB′)2(πγ)12 2(µB)2iπ( π )12 1dξ(1 ξ)eix4˜γ2ξ2. (21) i − iγ − Z0 5 Thus, v(A) is reduced to the following v(A) = 1 [ i (sµB′)coth(sµB′) 23 +2(µB)21 (sµB′)coth(sµB′) 12 1dξ(1 ξ)eix4˜γ2ξ2] −4π2 s2{ } s{ } − Z0 (22) and finally we get the effective potential V as follows eff i ∞ ds V = e−ism2(v(A) v(0)). (23) eff 2 s − Z0 The effective potential for the uniform field configuration can be obtained by putting µB′ = 0 to get (µB)2 ∞ ds 1 i (µB)2s V = i dξ(1 ξ)ei(µB)2ξ2s + e−im2s. (24) eff − 4π2 s2{ − − 2 12 } Z0 Z0 The divergent contributions at s = 0 are removed by adding local counter terms of (µB)2, and (µB)4. This implies the renormalization of the magnetic moment µ to the measured value and the coupling of (µB)4 to zero presumably. The effective potential, Eq.(24), for uniform magnetic fields is found to be real, which implies a stable magnetic background of w = 0. For magnetic fields weaker than the critical field B = m/µ, using a contour c integration in the fourth quadrant, the integration can be done along the negative imaginary axis giving the finite real effective action as (µB)2 ∞ ds 1 (µB)2s 1 V = + dξ(1 ξ)e(µB)2ξ2s e−m2s. (25) eff 4π2 s2{2 12 − − } Z0 Z0 The leading radiative correction term for a weak B field is (µB)6 δV = . (26) eff 240π2m2 For an inhomogeneous field configuration, µB′ = 0, the effective potential is given by 6 V = (µB)2 ∞ ds i µB′scoth(µB′s) 1dξ(1 ξ)ei(µµBB)′2ξ2tanh(µB′s) i + (µB)2s e−im2s eff − 4π2 s2{ − − 2 12 } Z0 q Z0 1 ∞ ds (µB′s)2 + (µB′scoth(µB′s))3/2 1 e−im2s, (27) 8π2 s3{ − − 2 } Z0 where an additional divergent contribution at s = 0 is removed by adding a local counter term of (µB′)2 in the second term. The leading radiative correction terms of the effective potential for a small gradient weak B field are calculated as given by (µB)6 (µB)4(µB′)2 (µB)2(µB′)2 (µB′)4 δV = + . (28) eff 240π2m2 288π2m4 − 48π2m2 − 960π2m4 6 It is found that the effective potential, Eq.(27), has a non-vanishing imaginary part, which implies that the background of inhomogeneous magnetic field configuration is un- stable against the creation of neutral fermions with Pauli interaction. From the imagi- nary part of the effective potential Eq.(27), we obtain the production rate density, w(x) = 2Im(V (A[x])). eff Introducing dimensionless parameters defined as v = sµB′, λ = m2 , κ = m2 , the |µB′| (µB)2 production rate density w(x) in the unit of the fermion mass is finally given by 2m4 ∞ dv λ 1 λv w(x) = √vcothvF( tanhv,λv) cosλv sinλv − 4π2λκ v2{ κ − 2 − 12κ } Z0 m4 ∞ dv v2 (vcothv)3/2 1 sinλv, (29) − 4π2λ2 v3{ − − 2 } Z0 where 1 F(a,b) dξ(1 ξ)cos(aξ2 b) ≡ − − Z0 1 = sin(a b)+sin(b) −2a{ − } π 2a 2a + cos(b)FresnelC( )+sin(b)FresnelS( ) . (30) 2a { s π s π } r Since the scale of inhomogeneity less than Compton wavelength of the fermion is irrelevant to the particle production through this process, we take in this work the spatial gradient of the magnetic field B′ to be smaller than the ratio of field strength, B , to the Compton | | | | wavelength 1, that is, λ > √κ. m The integration Eq.(29) is a finite integration, but has singularities along the imaginary v axis similarly to the Schwinger’s result, where the residue calculation gives the analytic WKB type expression. However, the integration Eq.(29) has essential sigularities along the imaginary axis, so that it seems not possible to get a usual analytic WKB type expression using a contour integration. Therefore we use numerical integrations to investigate the properties of the production rate density given by Eq.(29). For the numerical calculation, we consider the case of κ 1 and λ > 2 as an example. Numerical integrations of the ≥ production rate density show that the second term of Eq.(29) is negligible compared to the first termforλ > 2. The productionrateshows exponential monotonicdecrease forλ > 2√κ and κ 1. The production rate, w(x), is calculated for κ = 1.0 and 2.0 as a function of λ. ≥ The results in the unit of m4 are as shown in [Fig.1-2]. The results of numerical calculations 4π2 are represented by dots in the figures. The numerical integrations of Eq.(29) suffer from 7 large oscillatory fluctuation, which is an unavoidable feature due to the violent oscillations in the integrand as discussed in [12]. To get an analytic expression, we obtain the best fit of the numerical results to the curves in the form of aκe−bκλ for [Fig.1-2]. We can observe that λ the particle creation rate is an exponentially decreasing function with respect to the inverse of the field gradient, w e−constant×m2/|µB′|, (31) ∼ which shows the characteristics of the non-perturbative process. This can be understood as a quantum tunnelling through a potential barrier of height 2m of a particle exposed ∼ to an potential energy µ B′ x due to the inhomogeneous magnetic field coupled to the ∼ | | magnetic dipole moment through Pauli interaction. It is similar to the Schwinger process of electron-positron pair creation in the strong electric field, where the creation rate is decreasing exponentially [11], w e−constant×m2/|eE|. ∼ One can see that the production rate is suppressed very rapidly when the field strength becomes weaker than m/µ as well as the inhomogeneity scale is bigger than Compton ∼ wave length scale. For the inhomogeneity in the Compton wave length scale, the rate per unit time per unit volume is typically of order m4 for the critical field strength. 4π2 However for the realistic estimation of the production rate, more precise information on the mass and magnetic moment of a particle and the strength of the magnetic field as well as the scale of the spatial inhomogeneity of the field in consideration are needed for the observational possibility. As a possible environment, let us consider the pair creation of neutrino with a non-zero magnetic dipole moment in the vicinity of the very strongly magnetized compact objects with B = 1015G as a typical strength [13, 14]. Taking the possible magneticmoment tobeaslargeastheexperimental upper boundµ = 10−11µ and ν B the mass of the neutrino to be m 10−2 eV constrained by the solar neutrino observations, ν ∼ the critical field strength is estimated to be B 1017 G. One can see that the condition for c ∼ κ 1 or B < B assumed in this work is satisfied. Since the the scale of the inhomogeneity c ≥ is naturally about the size of the compact object, R 104 m, which is much larger than ∼ the Compton wave length 10−4m, the production rate is expected to be substantially ∼ suppressed from the typical rate such that it may not provide sufficiently high luminosity for the neutrino detectors. 8 III. DISCUSSION We have examined the vacuum production of neutral fermions in inhomogeneous mag- netic fields through Pauli interaction. The fermions, which are coupled to the background electromagnetic field through Pauli interaction, are integrated out and there appears an imaginary part in the effective action. It turns out that the production rate density depends on both the gradient and the strength of magnetic fields in 3+1 dimension, which is quite different from the result in 2+1 dimension [9], where the production rate depends only on the gradient of the magnetic fields, not on the strength of the magnetic fields. The difference can be attributed to the different nature of spinors in 3+1 and 2+1 dimensions. The vacuum production of fermion with the Pauli interaction is found to be a magnetic effect. Explicit calculations with a linear electric field configuration of xˆ direction with − a constant gradient along xˆ direction such that E = E + E′x shows that the effective x 0 − potentialhasnoimaginarypartwhenthesingularitiesareregularizedproperly. Itcanbealso shown by substituting B iE and B′ iE′ in the effective potential for a pure magnetic → → field Eq.(23) with the s integration along the imaginary axis. Therefore one can see that the pair creation through Pauli interaction is a purely magnetic effect. It is an interesting result whencomparedtothepaircreationofchargedparticlesthroughtheminimalcoupling, which is known to be an electric effect [11]. Although the production rate density in this work has been derived for µB′ = constant, it can be applicable to various types of magnetic fields provided that the magnetic field is linear in the scale of Compton wavelength of the particle considered because the particle production rate density is a local quantity. It may be therefore applicable to a spatially slowly varying µB′(x) as a good approximation if the gradient variation is very small in the Compton wavelength scale. For the realistic calculation of the production rate, more precise information on the mass and magnetic moment of a particle and the strength of the magnetic field as well as the scale of the spatial inhomogeneity of the field in consideration are needed for the observational possibility. This work was supported by Korea Research Foundation Grant(KRF-2004-041-C00085). [1] W. Pauli, Rev. Mod. Phys. 13, 203(1941) 9 0.015 0.01 0.005 4 6 8 10 FIG. 1: κ= 1.0 with varying λ: a = 0.050, b = 0.136 1.0 1.0 0.0012 0.001 0.0008 0.0006 0.0004 0.0002 4 6 8 10 FIG. 2: κ= 2.0 with varying λ: a = 0.013, b = 0.775 2.0 2.0 [2] See, for example, C-L. Ho and P. Roy, Annals. Phys. 312, 161(2004); Q-q Lin, Phys. Rev. A61, 022101(2000) [3] For a recent review see, e.g., J. F. Valle, hep-ph/0508067(2005) [4] K. Fuzikawa and R. Shrock, Phys. Rev. Lett. 45, 963(1980) [5] E. K. Akhmedov, Phys. Lett. B213, 64(1988);C.-S. Lim and J. Marciano, Phys. Rev. D37, 1368(1988) ;A.B. Balantekin and C. Volpe, Phys. Rev. D72, 033008(2005) [6] A. Goyal, Phys. Rev. D64, 013005(2001) [7] J. Schwinger, Phys. Rev. 82, 664(1951) 10