Springer Tracts in Modern Physics 252

Alexander Kuznetsov
Nickolay Mikheev

Electroweak
Processes in
External Active
Media
Springer Tracts in Modern Physics

Volume 252

Honorary Editor
G. Höhler, Karlsruhe, Germany

Series Editors
A. Fujimori, Tokyo, Japan
J. H. Kühn, Karlsruhe, Germany
T. Müller, Karlsruhe, Germany
F. Steiner, Ulm, Germany
W. C. Stwalley, Storrs, CT, USA
J. E. Trümper, Garching, Germany
P. Wölfle, Karlsruhe, Germany
U. Woggon, Berlin, Germany

For further volumes:

http://www.springer.com/series/426
Springer Tracts in Modern Physics
Springer Tracts in Modern Physics provides comprehensive and critical reviews of topics of
current interest in physics. The following fields are emphasized: Elementary Particle Physics,
Condensed Matter Physics, Light Matter Interaction, Atomic and Molecular Physics, Complex
Systems, Fundamental Astrophysics.
Suitable reviews of other fields can also be accepted. The Editors encourage prospective
authors to correspond with them in advance of submitting a manuscript. For reviews of
topics belonging to the above mentioned fields, they should address the responsible Editor
as listed below.
Special offer: For all clients with a print standing order we offer free access to the
electronic volumes of the Series published in the current year.

Elementary Particle Physics, Editors Condensed Matter Physics, Editors

Johann H. Kühn Atsushi Fujimori
Institut für Theoretische Teilchenphysik Editor for The Pacific Rim
Karlsruhe Institut für Technologie KIT Department of Physics
Postfach 69 80 University of Tokyo
76049 Karlsruhe, Germany 7-3-1 Hongo, Bunkyo-ku
Phone: +49 (7 21) 6 08 33 72 Tokyo 113-0033, Japan
Fax: +49 (7 21) 37 07 26 Email: [email protected]
Email: [email protected] http://wyvern.phys.s.u-tokyo.ac.jp/
www-ttp.physik.uni-karlsruhe.de/*jk welcome_en.html

Thomas Müller Peter Wölfle

Institut für Experimentelle Kernphysik Institut für Theorie der Kondensierten Materie
Karlsruhe Institut für Technologie KIT Karlsruhe Institut für Technologie KIT
Postfach 69 80 Postfach 69 80
76049 Karlsruhe, Germany 76049 Karlsruhe, Germany
Phone: +49 (7 21) 6 08 35 24 Phone: +49 (7 21) 6 08 35 90
Fax: +49 (7 21) 6 07 26 21 Phone: +49 (7 21) 6 08 77 79
Email: [email protected] Email: [email protected]
www-ekp.physik.uni-karlsruhe.de www-tkm.physik.uni-karlsruhe.de

Fundamental Astrophysics, Editor Complex Systems, Editor

Joachim E. Trümper Frank Steiner
Max-Planck-Institut für Extraterrestrische Institut für Theoretische Physik
Physik Universität Ulm
Postfach 13 12 Albert-Einstein-Allee 11
85741 Garching, Germany 89069 Ulm, Germany
Phone: +49 (89) 30 00 35 59 Phone: +49 (7 31) 5 02 29 10
Fax: +49 (89) 30 00 33 15 Fax: +49 (7 31) 5 02 29 24
Email: [email protected] Email: [email protected]
www.mpe-garching.mpg.de/index.html www.physik.uni-ulm.de/theo/qc/group.html

Solid State and Optical Physics Atomic, Molecular and Optical Physics
Ulrike Woggon William C. Stwalley
Institut für Optik und Atomare Physik University of Connecticut
Technische Universität Berlin Department of Physics
Straße des 17. Juni 135 2152 Hillside Road, U-3046
10623 Berlin, Germany Storrs, CT 06269-3046, USA
Phone: +49 (30) 314 78921 Phone: +1 (860) 486 4924
Fax: +49 (30) 314 21079 Fax: +1 (860) 486 3346
Email: [email protected] Email: [email protected]
www.ioap.tu-berlin.de www-phys.uconn.edu/faculty/stwalley.html
Alexander Kuznetsov Nickolay Mikheev
•

Electroweak Processes
in External Active Media

123
Alexander Kuznetsov
Nickolay Mikheev
Theoretical Physics Department
Yaroslavl State P.G. Demidov University
Yaroslavl
Russia

ISSN 0081-3869 ISSN 1615-0430 (electronic)

ISBN 978-3-642-36225-5 ISBN 978-3-642-36226-2 (eBook)
DOI 10.1007/978-3-642-36226-2
Springer Heidelberg New York Dordrecht London

Library of Congress Control Number: 2013934025

Ó Springer-Verlag Berlin Heidelberg 2013

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of
the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or
information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed. Exempted from this legal reservation are brief
excerpts in connection with reviews or scholarly analysis or material supplied specifically for the
purpose of being entered and executed on a computer system, for exclusive use by the purchaser of the
work. Duplication of this publication or parts thereof is permitted only under the provisions of
the Copyright Law of the Publisher’s location, in its current version, and permission for use must
always be obtained from Springer. Permissions for use may be obtained through RightsLink at the
Copyright Clearance Center. Violations are liable to prosecution under the respective Copyright Law.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt
from the relevant protective laws and regulations and therefore free for general use.
While the advice and information in this book are believed to be true and accurate at the date of
publication, neither the authors nor the editors nor the publisher can accept any legal responsibility for
any errors or omissions that may be made. The publisher makes no warranty, express or implied, with
respect to the material contained herein.

Printed on acid-free paper

Springer is part of Springer Science+Business Media (www.springer.com)

Preface

In the late twentieth and the early twenty-first centuries, the most intensive progress
was observed in the sciences, that develop at the junction of two sciences. The most
interesting among them, seem to be those that combine the branches most distant
from each other. If so, then it is easy to identify the leader of such sciences. It is one
that connects the smallest objects available for research, elementary particles, and
these giant objects like stars, and has the name of Particle Astrophysics. It is not
difficult to identify the major milestones in the life of this relatively young, but very
rapidly developing science. The birth is most likely to be dated to the beginning of
the 1930s. Just then, after the discovery of a neutron by J. Chadwick in 1932, the
concept of a neutron star was proposed by L. D. Landau, and independently by
W. Baade and F. Zwicky. The start of the maturation of this science can be more or
less confidently dated to 1987 when extragalactic neutrinos were registered for the
first time from the supernova SN1987A explosion in the Large Magellanic Cloud, a
satellite galaxy of our Milky Way. For the date of the endpoint of the maturation
period for particle astrophysics, one can propose 2001 when the solar-neutrino
puzzle was solved in a unique experiment at the heavy-water detector installed at
the Sudbury Neutrino Observatory. This experiment confirmed B. Pontecorvo’s key
idea concerning neutrino oscillations and, along with experiments that studied
atmospheric and reactor neutrinos, thereby proved the existence of a nonzero
neutrino mass and the existence of mixing in the lepton sector. The Sun appeared in
this case as a natural laboratory for investigations of neutrino properties.
There exist some books on the topic where the basics of this new science can be
studied. However, new facts and ideas appear so fast that it is necessary for
specialists to follow not only journal papers but also electronic preprints, in order
to keep abreast of the latest developments.
A page of this new science, which on the one hand is rather difficult and on the
other hand is not covered enough by books or reviews, deals with the particle
processes under the extreme conditions of the stellar interior—hot dense plasma
and strong electromagnetic fields. This discipline, which can be called Quantum
Field Theory in an External Active Media, was founded in the 1970s, and now it
continues in motion. As an attempt to set some milestone, the objective of our
previous monograph [1] was to give a systematic description of the methods of
calculation of the quantum processes, both at the tree and loop levels, in external

v
vi Preface

electromagnetic fields. The aim of the present monograph is to consider the

quantum processes under an influence of, along with a magnetic field, one more
external active media which is hot dense plasma.
The review is based in part on the special lecture course given to the second-
year master-course students studying at the Theoretical Physics Department of the
Yaroslavl State University, Yaroslavl, Russia. It can be used by graduate and
postgraduate students specializing in theoretical physics and being familiar with
the basics of the Quantum Field Theory and the Standard Model of the Electro-
weak Interactions. The authors make a great effort to give all the details that will
make this book a valuable text for students. The monograph can be also useful for
specialists in the Quantum Field Theory and particle physics, who are interested in
the problems of physics of quantum phenomena in external active media.
We have obtained a part of the results presented in this monograph in
co-authorship with our colleagues and with our graduate and postgraduate students
at the Department of Theoretical Physics of Yaroslavl State University. We thank
L. A. Vassilevskaya, A. A. Gvozdev, A. Ya. Parkhomenko, M. V. Chistyakov,
I. S. Ognev, E. N. Narynskaya, D. A. Rumyantsev, A. A. Okrugin, R. A. Anikin,
A. M. Shitova, and M. S. Radchenko for collaboration and helpful discussions.
We are grateful to S.I. Blinnikov, V.A. Rubakov, V.B. Semikoz, and M.I.
Vysotsky for many fruitful discussions and to G.G. Raffelt for collaboration in
obtaining the results, concerning the self-energy neutrino operator in a magnetic
field. We are thankful to H.-T. Janka and B. Müller for providing us with detailed
data on radial distributions and time evolution of physical parameters in the
supernova core, obtained in their model of the SN explosion. With a warm feeling
we want to mention the many lively discussions with K.A. Ter-Martirosian, whose
strong support in the 1990s proved to be crucial for our research group.
A part of the results presented in this monograph was obtained in the study
performed within the State Assignment for Yaroslavl University (Project #
2.7508.2013), and supported by the Russian Foundation for Basic Research (Project
# 11-02-00394-a).

Yaroslavl, March 2013 Alexander Kuznetsov

Nickolay Mikheev

Reference

1. A.V. Kuznetsov, N.V. Mikheev, Electroweak Processes in External Electro-

magnetic Fields (Springer, New York, 2003)
Contents

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

2 Solutions of the Dirac Equation in an External

Electromagnetic Field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.1 Magnetic Field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2 The Ground Landau Level . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.3 Crossed Field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.4 Density Matrix of the Plasma Electron in a Magnetic Field
with the Fixed Number of a Landau Level . . . . . . . . . . . . .... 22
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 24

3 Propagators of Charged Particles in External Active Media . . . . . 25

3.1 Propagators of Charged Particles in a Magnetic Field . . . . . . . . 25
3.1.1 Propagators in the Fock Proper-Time Presentation . . . . . 26
3.1.2 A Note on the Noninvariant Phase . . . . . . . . . . . . . . . . 29
3.1.3 Propagators in the Weak-Field Expansion . . . . . . . . . . . 30
3.1.4 Propagators in an Expansion over Landau Levels . . . . . . 31
3.1.5 Electron Propagator in a Strong Magnetic Field . . . . . . . 37
3.2 Propagators of Charged Particles in a Crossed Field . . . . . . . . . 38
3.3 Direct Derivation of the Electron Propagator in a Magnetic
Field as the Sum over Landau Levels on a Basis
of the Dirac Equation Exact Solutions . . . . . . . . . . . . . . . .... 39
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 43

4 Particle Dispersion in External Active Media . . . . . . . . . . . . . ... 45

4.1 Dispersion in Media: Main Definitions. . . . . . . . . . . . . . . . ... 45
4.2 Photon Polarization Operator in an External Magnetic Field . ... 48
4.3 Generalized Two-Point Loop Amplitude j ! f f ! j0
in an External Electromagnetic Field . . . . . . . . . . . . . . . . . . . . 52
4.3.1 Magnetic Field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.3.2 Crossed Field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
4.4 Photon Polarization Operator in Plasma . . . . . . . . . . . . . . . . . . 60

vii
viii Contents

4.5 Neutrino Self-energy Operator in Plasma . . . . . . . . . . . . . . . . . 66

4.5.1 Defnition of the Operator R(p) in Plasma . . . . . . . . . . . 68
4.5.2 Neutrino Additional Energy in Hot Dense Plasma. . . . . . 68
4.5.3 On the Neutrino Radiative Decay in Plasma. . . . . . . . . . 73
4.5.4 Ultra-High Energy Neutrino Dispersion in Plazma . . . . . 78
4.6 Neutrino Self-energy Operator in an External Magnetic Field . . . 89
4.6.1 Defnition of the Operator R(p) in a Magnetic Field . . . . 89
4.6.2 Low Landau Level Contribution
into the Operator R(p) . . . . . . . . . . . . . . . . . . . . . . . . . 93
4.6.3 Calculation of the Operator R(p) in a ‘‘Weak’’ Field . . . . 96
4.6.4 The Case of a Moderately Strong Field . . . . . . . . . . . . . 100
4.6.5 The Neutrino Operator R(p) in a Crossed Field . . . . . . . 106
4.6.6 Field-Induced Neutrino Magnetic Moment . . . . . . . . . . . 109
4.7 Neutrino Self-energy Operator in Magnetized Plasma . . . . . . . . 109
4.7.1 Neutrino Scattering on Magnetized Plasma . . . . . . . . . . 110
4.7.2 Neutrino Additional Energy in Magnetized Plasma . . . . . 114
4.7.3 Asymptotic Expressions for the Neutrino Additional
Energy in Magnetized Plasma . . . . . . . . . . . . . . . . . .. 117
4.7.4 Induced Neutrino Magnetic Moment
in Magnetized Plasma . . . . . . . . . . . . . . . . . . . . . . . .. 119
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 125

5 Electromagnetic Interactions in External Active Media . . . . . . ... 127

5.1 Photon Decay into an Electron–Positron Pair
in a Strong Magnetic Field . . . . . . . . . . . . . . . . . . . . . . . . ... 127
5.1.1 Direct Calculation Based on the Solutions
of the Dirac Equation . . . . . . . . . . . . . . . . . . . . . . ... 128
5.1.2 Calculation Based on the Imaginary Part
of the Loop Amplitude . . . . . . . . . . . . . . . . . . . . . ... 132
5.2 The c ! e eþ Decay in a Crossed Field . . . . . . . . . . . . . . ... 133
5.2.1 Direct Calculation Based on the Solutions
of the Dirac Equation . . . . . . . . . . . . . . . . . . . . . . ... 133
5.2.2 Calculation Based on the Imaginary Part
of the Loop Amplitude . . . . . . . . . . . . . . . . . . . . . ... 139
5.3 Photon Emission by Electron in a Strong Magnetic Field . . . ... 140
5.4 Electromagnetic Interactions of the Dirac Neutrino with
a Magnetic Moment. . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 143
5.4.1 Magnetic Moment of the Dirac Neutrino
and its Astrophysical Manifestations . . . . . . . . . . . . ... 143
5.4.2 Neutrino Interaction with Background . . . . . . . . . . . ... 146
5.4.3 The Rate of Creation of the Right-Handed Neutrino . ... 147
5.4.4 Contributions of Plasma Components
into the Neutrino Scattering Process . . . . . . . . . . . . ... 152
5.4.5 Illustration: Completely Degenerate Plasma at T = 0. ... 154
Contents ix

5.4.6 Uniform Ball Model for the Supernova Core . . . . . .... 157
5.4.7 Models of the Supernova Core with Radial
Distributions of Physical Parameters: Limits
on the Neutrino Magnetic Moment . . . . . . . . . . . . .... 159
5.4.8 Possible Effect of the Neutrino Magnetic Moment:
Shock-Wave Revival in a Supernova Explosion . . . .... 163
5.4.9 Possible Effect of the Neutrino Magnetic Moment:
Neutrino Pulsar. . . . . . . . . . . . . . . . . . . . . . . . . . .... 168
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 171

6 Neutrino-Electron Interactions in External Active Media . . . . . . . 175

6.1 The m ! e W þ Process in a Strong Magnetic Field . . . . . . . . . 175
6.2 The m ! me eþ Process in a Strong Magnetic Field . . . . . . . . . 180
6.2.1 Calculation of the Differential Probability Based
on the Solutions of the Dirac Equation . . . . . . . . . . . . . 180
6.2.2 Calculation Based on the Imaginary Part
of the Loop Amplitude . . . . . . . . . . . . . . . . . . . . . . . . 182
6.2.3 The Total Process Probability . . . . . . . . . . . . . . . . . . . . 183
6.3 The m ! me eþ Process in a Crossed Field. . . . . . . . . . . . . . . . 186
6.3.1 A Historical Overview . . . . . . . . . . . . . . . . . . . . . . . . . 186
6.3.2 Calculation of the Differential Probability Based
on the Imaginary Part of the Loop Amplitude . . . . . . . . 187
6.3.3 The Total Process Probability . . . . . . . . . . . . . . . . . . . . 189
6.4 Possible Astrophysical Manifestations of the m ! me eþ
Process in an External Magnetic Field . . . . . . . . . . . . . . . . . . . 192
6.4.1 Mean Losses of the Neutrino Energy and Momentum . . . 192
6.4.2 Applicability of the Results Obtained in a Pure
Magnetic Field for the Plasma Environment. . . . . . . . . . 194
6.4.3 Possible Astrophysical Manifestations . . . . . . . . . . . . . . 196
6.5 Neutrino in Strongly Magnetized Electron–Positron Plasma . . . . 197
6.5.1 What Do We Mean Under Strongly Magnetized
e eþ Plasma. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198
6.5.2 Neutrino-Electron Processes in Strongly Magnetized
Plasma: A Kinematic Analysis . . . . . . . . . . . . . . . . . . . 199
6.5.3 The Probability of the Process m ! me eþ . . . . . . . . . . . 201
6.5.4 The Total Probability of the Neutrino Interaction
with Magnetized Electron–Positron Plasma . . . . . . . . . . 205
6.5.5 Mean Losses of the Neutrino Energy and Momentum . . . 209
6.5.6 Integral Action of Neutrinos on a Magnetized Plasma . . . 212
6.5.7 Neutrino-Electron Processes Involving the Contributions
of the Excited Landau Levels . . . . . . . . . . . . . . . . . . . . 215
6.5.8 Pulsar Natal Kick Via Neutrino-Triggered
Magnetorotational Asymmetry . . . . . . . . . . . . . . . . . . . 220
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
x Contents

7 Neutrino-Photon Interactions in External Active Media. . . . . . . . . 229

7.1 mmc Interaction in External Active Media . . . . . . . . . . . . . . . . . 229
7.1.1 The Effective Lagrangian of the mmc Interaction . . . . . . . 229
7.1.2 Photon Production by the Massless Neutrino m ! mc . . . . 232
7.1.3 Photon Decay into the Neutrino Pair c ! mm . . . . . . . . . 235
7.1.4 Radiative Neutrino Transition m ! mc in Strongly
Magnetized Plasma . . . . . . . . . . . . . . . . . . . . . . . . ... 239
7.2 Compton-Like Interaction of Neutrinos with Photons. . . . . . ... 244
7.2.1 The Amplitude of the Process cc ! mm in Vacuum . . ... 244
7.2.2 Neutrino Scattering in the Coulomb Field
of a Nucleus. . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 251
7.2.3 The External Field Influence on the Process cc ! mm ... 253
7.2.4 A Conversion cc ! mm in the Left–Right Symmetric
Extension of the Standard Model . . . . . . . . . . . . . . ... 255
7.2.5 Possible Manifestations of the cc ! mm Process
in Astrophysics . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 257
7.3 Neutrino Photoproduction on a Nuclei in a Strong
Magnetic Field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 258
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 264

8 Conclusion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
Chapter 1
Introduction

Astroparticle physics has manifested itself in recent decades as a vigorously growing

and prospective line of investigation at the junction of particle physics, astrophysics,
and cosmology; see e.g. [1–3]. An important stimulus of its development is an under-
standing of the essential role of quantum processes in the dynamics of astrophysical
objects and of the early Universe. On the other hand, extreme physical conditions
existing inside such objects, namely, the presence of hot dense plasma and strong
electromagnetic fields, make an active influence on the run of quantum processes,
thus allowing or enhancing the transitions that are forbidden or strongly suppressed
in a vacuum. In this connection, there exists a stable interest in investigations of
particle interactions in external active media.
This line of research is relevant to at least three from the list of the 30 top problems
of physics and astrophysics for the beginning of the twenty-first century, formulated
by Prof. V.L. Ginzburg in 1999 [4]. They are:
• the behavior of matter in superstrong magnetic fields;
• neutron stars and pulsars, supernova stars;
• neutrino physics and astronomy, neutrino oscillations.
It is known, that matter on the Earth in a natural form is rarely in the plasma
state, or it exists during a very short time. In contrast, most of the baryonic matter
in the Universe as a whole is a plasma in any form. Theoretical and experimental
study of this state of matter has a long history and is still relevant. Nowadays, one
of the priority areas of research is the study of plasma in extreme conditions. Such
states, usually occur either at high temperature or density, or in ultrahigh external
fields. In these conditions, plasma often has a completely new and unusual proper-
ties. The study of them is necessary to describe the behavior of the plasma as well
as objects in which it is present. Appropriate conditions for the emergence of this
plasma could occur in the early stages of the evolution of the Universe when it was
very hot. A similar situation can also be realized in high-power stellar cataclysms
and in compact astrophysical objects having very high density. Extreme values of
physical parameters: temperature, density, magnetic field intensity, component com-
position, arising in Supernova explosions [5, 6], allow to characterize these objects

A. Kuznetsov and N. Mikheev, Electroweak Processes in External Active Media, 1

Springer Tracts in Modern Physics 252, DOI: 10.1007/978-3-642-36226-2_1,
© Springer-Verlag Berlin Heidelberg 2013
2 1 Introduction

as a unique natural laboratories to study the physical properties of the plasma under
conditions that are currently (and may be ever) can not be implemented in terrestrial
experiments [7–9].
The close relationship of the laws of microcosm and macrocosm, which is realized
in the core collapse supernovae [10], where the laws are simultaneously valid both of
the general relativity and nuclear and particle physics, allows to analyze the physical
properties of the plasma in these unique environment, and also to investigate the
effects of hot, dense plasma on the quantum processes, and to determine the funda-
mental characteristics of particles on the basis of astrophysical data, and finally, to
study the impact of microphysics on astrodynamics [2, 3].
The study of plasma at extreme physical parameters which exist in a supernova
explosion is one of the best examples of an interaction of branches of physical science
which seem to be far from each other. The fact is that for a short time, such a plasma
can be obtained in collisions of elementary particles and nuclei in accelerators.
In the recent years, the most significant progress has been made in the experimental
study of the plasma. This is primarily due to the discovery at CERN of a new state
of matter called quark-gluon plasma, which was obtained in collisions of heavy
nuclei [11]. Today, the investigation continues actively at the accelerator of heavy
ions RHIC [12], and the studies have been started at the Large Hadron Collider
(LHC) [13, 14]. It is well-known that the quarks, by reason of the strong interaction,
are associated into colorless objects, hadrons, and can not be observed in the free state.
This phenomenon called quark confinement is sufficiently well studied. However,
at high collision energies plasma can be formed, in which quarks and gluons are
unconnected, then there is a deconfinement. The duration of the quark-gluon stage
is only a small fraction of the evolution time of a system of colliding particles,
though its influence is very essential and can be observed by an increased output
of strange mesons, the decrease in the output of heavy J/ψ mesons, and by an
increased output of photons and lepton-antilepton pairs with high energy [15]. It
should be noted that, despite the fact that the properties of the quark-gluon plasma
significantly different from all known states of matter, it has a lot of similarity with
conventional electromagnetic plasmas [16].
A separate chapter in the physics of hot dense plasma is the research of its impact
on quantum processes, which have a critical influence on the macroscopic char-
acteristics of some astrophysical objects like supernovae and young neutron stars.
The influence of the plasma on the quantum processes is twofold. On the one hand,
under its influence the matrix elements may be modified, which means the change
in the dynamics of the processes. On the other hand, the plasma influence changes
the dispersion properties of particles, i.e., the process kinematics. As the result, the
reactions can be opened or significantly enhanced which are kinematically forbidden
or strongly suppressed in a vacuum. Among the best known processes, the photon
decay into a pair of neutrino and antineutrino, γ → ν ν̄, can be indicated. This
process, being forbidden in vacuum, is possible due to the plasma influence on the
dispersion properties of a photon which acquires an effective mass. As a result, the
decay γ → ν ν̄ is kinematically allowed and may occur in stars [17, 18]. In fact, this
so-called plasma process is the primary mechanism of the neutrino emission by stars
1 Introduction 3

in a wide range of temperatures and densities, including, for example, the physical
conditions inside the white dwarfs and red giants.
Along with the hot dense plasma, a significant effect on the quantum processes
can be provided by another component of active astrophysical environments, which
are strong magnetic fields. However, the magnetic field significantly influences the
quantum processes only in the case when it is strong enough. There exists a natural
scale for the field strength which is the so-called critical value Be = m 2e /e  4.41 ×
1013 G (we use natural units in which c =  = 1).
The fields of such strength are unattainable in a laboratory. However, the astro-
physical objects and processes inside them give us unique possibilities for inves-
tigations of the particle physics, and of the neutrino physics especially under the
extreme conditions of a strong magnetic field. The concept of the astrophysically
strong magnetic field has changed over the years (see Fig. 1.1).
Whereas magnetic fields with strength 109 –1011 G were considered as “very
strong” nearly forty years ago [19], the fields observed at the surface of pulsars
have appeared to be much stronger, of the order of 1012 –1013 G. The physics of pul-
sars, i.e. neutron stars, is described in detail in monographs, see e.g. [20–23]. Now the
fields ∼1012 –1013 G are treated as the so-called “old” magnetic fields [21]. There are
grounds to expect that fields on even larger scale can arise in astrophysical objects.
For example, there exist two classes of stars, the so-called soft gamma-ray repeaters
(SGR) [24, 25] and anomalous X-ray pulsars (AXP) [26, 27] which are believed
to be magnetars [28], neutron stars with magnetic field strength ∼1014 –1015 G. To
the date (March, 2013), the McGill SGR/AXP Online Catalog contains the current
information available on 23 magnetars: 11 SGRs, and 12 AXPs [29].
The fields at the moment of a cataclysm like a supernova explosion, when a neutron
star is born, or a coalescence of neutron stars, could be much greater, ∼1015 –1017 G.

Fig. 1.1 Evolution of the notion “strong magnetic field” in astrophysics

4 1 Introduction

The possible existence of such fields, both of toroidal and poloidal types, is the
subject of wide discussions [30–38].
In the early Universe, in the interval between the stages of the QCD phase transition
(∼10−5 s) and nucleosynthesis (∼10−2 –102 s), very strong magnetic fields, the so
called “primary” fields, in principle, could exist, with an initial strength of the order
1023 G [39] and even more (∼1033 G [40]). Their evolution during the expansion of
the Universe could determine the existence at the present stage of coherent large-scale
(∼100 kpc) magnetic fields with an intensity ∼10−21 G. These fields, in turn, could
be enhanced by the galactic dynamo mechanism to the observed values of galactic
magnetic fields ∼10−6 G. Possible origins of primary strong magnetic fields and
dynamics of their evolution in the expanding Universe are the subject of intense
research (see, for example, the surveys [41, 42] and references cited therein).
Note that, in contrast to the magnetic field, the electric field corresponding to the
critical value m 2e /e is the maximal one, since the generation of an electric field of the
order of the critical value in a macroscopic space region lead to intensive production
of electron–positron pairs from the vacuum, which is equivalent to a short circuit
of a “machine” generating the electric field. On the contrary, the magnetic field
can exceed the critical value Be due to the stability of a vacuum. Furthermore, the
magnetic field plays a stabilizing role, if directed perpendicular to the electric one.
In this configuration, the electric field E can exceed the critical value of Be . The
vacuum stability condition can be written in the invariant form as
 
F μν Fμν = 2 B 2 − E 2  0.

So far, essentially one-dimensional problems have been solved in astrophysical

calculations of processes such as supernova explosions, and analyses of the influ-
ence of the active medium on quantum processes have only contained the plasma
contribution. However, serious arguments have been put forward to suggest that the
physics of supernovas is considerably more complex. In particular, we need to allow
for rotation of the shell and also for the possible existence of a strong magnetic field,
with these two phenomena being interrelated. In fact, the magnetic field generated
during the collapse of a supernova nucleus may reach the critical value ∼1013 G.
The presence of rotation can lead to generation of a toroidal magnetic field with
increasing the field strength by an additional factor of 103 –104 [30, 31].
In astrophysical phenomena such as stellar collapse, the absence of strong mag-
netic fields is an exotic rather than a typical case. It is appropriate to discuss the
following set of questions.
(i) Which can be considered to be the more exotic object: a star possessing a
magnetic field or a star without such a field? As far as we know from astrody-
namics, a star without a magnetic field should be taken as an exotic rather than
a typical case. In exactly the same way the presence of a primary magnetic field
may be considered natural for a presupernova. As we know, a primary magnetic
field of 100 G leads to the generation of a field on the scale of 1012 –1013 G
during the collapse process as a result of the conservation of magnetic flux.
1 Introduction 5

(ii) Which can be considered to be the more typical case: a star possessing rotation
or a star without rotation? Evidently a star without rotation appears to be the
more exotic object.
(iii) Which type of collapse looks more exotic: compression without or with an
angular velocity gradient? Since the velocities at the edge of a compressible
astrophysical object may reach relativistic scales, compression with differential
rotation, i.e., with an angular velocity gradient, seems more probable.
All these factors are required to achieve the Bisnovatyi-Kogan scenario for the
rotational explosion of a supernova [30, 31]. The main component of this scenario
is that the initially poloidal magnetic field lines of a field of 1012 –1013 G are twisted
and compacted as a result of the angular velocity gradient to form an almost toroidal
field of ∼1015 –1017 G. It should be emphasized that a field of the order of 1016 G is
really a rather dense medium with mass density
 2
B2 B g
ρ=  0.4 × 1010 , (1.1)
8π 1016 G cm3

which is comparable with the plasma mass density 1010 –1012 g/cm3 , typical for
the envelope of an exploding supernova. Thus, in detailed studies of astrophysical
processes such as supernova collapse it is absolutely essential to take into account
the influence of the complex active medium including the plasma and the magnetic
field.
A dramatic possibility exists [43, 44], that the topic of an asymmetric supernova
explosion or merger of neutron stars in our galaxy may appear vitally important for
humankind, because of the possible production of a highly beamed gamma ray jet
pointed in our direction, which could devastate life on Earth. The strong magnetic
field is one typical characteristic of the asymmetry in such an astrophysical cataclysm.
Thus, the problem of particle interactions with external active media is of con-
siderable interest for modern physics. At the same time, this problem is not covered
comprehensively in the textbooks on Quantum Field Theory. There are a few clas-
sical books, e.g. [45–48], where the technique of calculations of quantum processes
in external media is partially concerned. A more detailed presentation of this topic
is made in the book [49], and in the reviews [50–53].
It is well known that processes forbidden in a vacuum become possible in
intense external fields (such as the photon decay into an electron–positron pair
γ → e− e+ [54], the photon splitting into two photons γ → γγ [55–65], the neutrino
production of an electron–positron pair, ν → νe− e+ [66–74], the radiative transition
of massless neutrinos, the so-called neutrino Cherenkov process ν → νγ [75–78],
the photon decay into neutrino pair γ → ν ν̄ [75, 76, 79], and the axion decay
a → f f¯ [80]). Apart from this, intense external electromagnetic fields catalyze
some processes allowed in a vacuum, for example, the radiative decay of a massive
neutrino ν → ν  γ [81, 82], and the double-radiative decay of an axion, a → γγ
[83, 84].
6 1 Introduction

The method in which the external field effect is taken into account on the basis
of exact solutions of the field theory equations for a charged particle in an external
electromagnetic field rather than on the basis of perturbation theory, has become an
important tool for studying some fundamental problems of particle interactions with
an electromagnetic field. The extent to which the motion of a particle is influenced
by the field depends on its specific charge, i.e. the ratio of the particle charge to its
mass. The hierarchy of masses of elementary particles existing in Nature leads to the
inverse hierarchy of specific charges. Thus, particles that are the most sensitive to the
external field influence are the lightest charged fermions: the electron is the first one,
and then the muon and the u and d quarks follow. All these particles are described by
the Dirac equation, and its solutions in the presence of an external electromagnetic
field should be used.
In the Quantum Field Theory, the number of cases in which the Dirac equation
can be solved analytically is relatively small. These are the problem of electron
motion in a Coulomb field (hydrogen atom) and the problems of electron motion in
a uniform magnetic field, in the field of a plane electromagnetic wave, and in some
particular combinations of uniform electric and magnetic fields. Specific physical
phenomena are usually calculated on the basis of a diagram technique (which is
in fact the Feynman technique) where the initial and final states feature charged
fermions in an external field, which are described by solutions of the Dirac equation
in this field, and where internal lines for charged fermions represent their propagators
constructed on the basis of the above solutions. This method is advantageous in that it
enables us to analyze processes in high-strength fields—that is, in the case where it is
impossible to treat field effects within perturbation theory. Since the vacuum is stable
in superstrong magnetic fields, one can consider processes in magnetic fields with
the strength significantly exceeding the critical value Be . Thus, these problems form
a separate line of investigation in the Quantum Field Theory having an independent
conceptual interest. On the other hand, as was mentioned above, such fields can exist
near young pulsars; they can also arise in mergers of neutron stars and in supernova
explosions.
The above method has proved to be highly efficient in studying some processes in
intense electromagnetic fields that are important for various applications (among oth-
ers, we mean here beta decay in the field of intense laser radiation and quantum effects
accompanying the propagation of ultrarelativistic particles through monocrystals).
The objective of the present review is to give a systematic description of the
methods of calculation of the quantum processes, both at the tree and loop lev-
els, in external electromagnetic fields. The consideration is accented on the two
limiting cases: (i) the case of a very strong magnetic field when the charged
fermions occupy the ground Landau level; (ii) the case of a crossed field when
all the pure field invariants are equal to zero. These are the cases that allow us
to make the analytical calculations in great detail. The review is based for the
most part on the original results obtained by the authors with their collaborators
[64, 65, 70, 71, 73, 74, 78, 82, 85–113].
The monograph is constructed as follows. In Chap. 2, the solutions of the Dirac
equation for a fermion in an external electromagnetic field are presented for the cases
1 Introduction 7

of a pure magnetic field of arbitrary strength, of a strong magnetic field when fermions
occupy the ground Landau level, and of a crossed field. Propagators of charged
particles in an external electromagnetic field for the same cases are presented in
Chap. 3. Chapter 4 is devoted to an analysis of the dispersion properties of photons and
neutrinos in external active media: magnetic field, plasma, and magnetized plasma.
In Chap. 5, electromagnetic interactions in external active media are analysed. They
are the processes of the photon decay into an electron–positron pair, and of the
photon emission by an electron in magnetic fields, and also the electromagnetic
interactions of the Dirac neutrino with a magnetic moment. Chapters 6 and 7 are
devoted to the analyses of neutrino–electron and neutrino–photon interactions in
external active media. Astrophysical manifestations of the most physical processes
are also analyzed.

Notations

The 4-metrics with the signature (+−−−) and the natural units in which  = 1, c =
1, are used.
e = |e| is the elementary charge.
m e is the electron mass, m f is the fermion mass.
μν is the neutrino magnetic moment, μ̃ν is the chemical potential of the neutrino
gas.
Fαβ is the tensor of the external constant uniform electromagnetic field, F̃αβ =
ε μν is the dual tensor (ε0123 = −ε
0123 = +1).
1
2 αβμν F
ϕαβ = Fαβ /B is the dimensionless tensor of the external magnetic field, ϕ̃αβ =
μν is the dual dimensionless tensor.
2 εαβμν ϕ
1

The tensor indices of four-vectors and tensors standing inside the parentheses are
contracted consecutively, for example:

( p F F p) = p α Fαβ F βδ pδ ;
(F F p)α = Fαβ F βδ pδ ;
(F F) = Fαβ F βα .

αβ = (ϕ̃ϕ̃)αβ are connected by the

The dimensionless tensors Λαβ = (ϕϕ)αβ , Λ

relation Λαβ − Λαβ = gαβ .
In the frame where the magnetic field B is only presented, directed along the
3d axis, the four-vectors with the indices ⊥ and  belong to the Euclidean {1, 2}-
subspace and the Minkowski {0, 3}-subspace, correspondingly. Then

Λαβ = diag(0, 1, 1, 0), αβ = diag(1, 0, 0, −1).

Λ (1.2)
8 1 Introduction

For arbitrary four-vectors pμ , qμ one has

μ μ
p⊥ = (0, p1 , p2 , 0), p = ( p0 , 0, 0, p3 ), (1.3)

( pq)⊥ = ( pΛq) = p1 q1 + p2 q2 ,  = p0 q 0 − p3 q 3 .
( pq) = ( p Λq) (1.4)

The Dirac gamma matrices are used in the standard representation [114]:
     
I 0 0 σ 0I
γ0 = ,γ = , γ5 = iγ 0 γ 1 γ 2 γ 3 = , (1.5)
0 −I −σ 0 I 0

σ are the Pauli matrices.

References

1. M.Yu. Khlopov, Cosmoparticle Physics (World Scientific, Singapore, 1999)

2. H.V. Klapdor-Kleingrothaus, K. Zuber, Particle Astrophysics, 2nd edn. (Institute of Physics
Publishing, Bristol, 2000)
3. G.G. Raffelt, Stars as Laboratories for Fundamental Physics (University of Chicago Press,
Chicago, 1996)
4. V.L. Ginzburg, Usp. Fiz. Nauk 169, 419 (1999). [Physics-Uspekhi 42, 353 (1999)]
5. V.S. Imshennik, D.K. Nadyozhin, Usp. Fiz. Nauk 156, 561 (1988). [Sov. Phys. Usp. 31, 1040
(1988)]
6. D.K. Nadyozhin, V.S. Imshennik, Int. J. Mod. Phys. A 20, 6597 (2005)
7. V.E. Fortov, Usp. Fiz. Nauk 177, 347 (2007). [Physics-Uspekhi 50, 333 (2007)]
8. Ya.N. Istomin, Usp. Fiz. Nauk 178, 880 (2008). [Physics-Uspekhi 51, 844 (2008)]
9. A.M. Bykov, I.N. Toptygin, Usp. Fiz. Nauk 177, 149 (2007). [Physics-Uspekhi 50, 141 (2007)]
10. A. Burrows, Rev. Mod. Phys. 85, 245 (2012).
11. U.W. Heinz, M. Jacob, Evidence for a new state of matter: an assessment of the results from
the CERN lead beam programme. arXiv: nucl-th/0002042 (2000)
12. M. Gyulassy, L. McLerran, Nucl. Phys. A 750, 30 (2005)
13. N.V. Krasnikov, V.A. Matveev, Usp. Fiz. Nauk 174, 697 (2004). [Physics-Uspekhi 47, 643
(2004)]
14. The ALICE Collaboration, J. Instrum. 3, S08002 (2008)
15. I.I. Roizen, E.L. Feinberg, O.D. Chernavskaya, Usp. Fiz. Nauk 174, 473 (2004). [Physics-
Uspekhi 47, 427 (2004)]
16. S. Mrowczynski, M.H. Thoma, Ann. Rev. Nucl. Part. Sci. 57, 61 (2007)
17. J.B. Adams, M.A. Ruderman, C.-H. Woo, Phys. Rev. 129, 1383 (1963)
18. M.H. Zaidi, Nuovo Cim. A 40, 502 (1965)
19. Y.B. Zel’dovich, I.D. Novikov, Theory of Gravitation and Evolution of Stars (Nauka, Moscow,
1971). (in Russian)
20. S.L. Shapiro, S.A. Teukolsky, Black Holes, White Dwarfs, and Neutron Stars (Wiley, New
York, 1985)
21. V.M. Lipunov, Astrophysics of Neutron Stars (Springer, New York, 1992)
22. P. Haensel, A.Y. Potekhin, D.G. Yakovlev, Neutron Stars 1: Equation of State and Structure
(Springer, New York, 2007)
23. B. Kiziltan, Reassessing the Fundamentals: On the Evolution, Ages and Masses of Neutron
Stars (Universal-Publishers, Boca Raton, FL, 2011)
References 9

24. C. Kouveliotou, T. Strohmayer, K. Hurley, J. van Paradijs, M.H. Finger, S. Dieters, P. Woods,
C. Thompson, R.C. Duncan, Astrophys. J. 510, L115 (1999)
25. K. Hurley, T. Cline, E. Mazets, S. Barthelmy, P. Butterworth, F. Marshall, D. Palmer, R. Aptekar,
S. Golenetskii, V. Il’Inskii, D. Frederiks, J. McTiernan, R. Gold, J. Trombka, Nature 397, 41
(1999)
26. X.-D. Li, E.P.J. van den Heuvel, Astrophys. J. 513, L45 (1999)
27. S. Mereghetti, L. Chiarlone, G.L. Israel, L. Stella, in Neutron Stars, Pulsars and Supernova
Remnants, eds. W. Becker et al. Proceedings of 270 WE-Heraeus-Seminar, Bad Honnef, January
2002 (e-print astro-ph/0205122)
28. R.C. Duncan, C. Thompson, Astrophys. J. 392, L9 (1992)
29. McGill Pulsar Group SGR/AXP Online Catalog, http://www.physics.mcgill.ca/~pulsar/
magnetar/main.html
30. G.S. Bisnovatyi-Kogan, Astron. Zh. 47, 813 (1970). [Sov. Astron. 14, 652 (1971)]
31. G.S. Bisnovatyi-Kogan, Stellar Physics 2: Stellar Evolution and Stability (Nauka, Moscow
1989; Springer, New York, 2001)
32. I. Lerche, D.N. Schramm, Astrophys. J. 216, 881 (1977)
33. V.V. Usov, Nature 357, 472 (1992)
34. C. Thompson, R.C. Duncan, Astrophys. J. 408, 194 (1993)
35. C. Thompson, R.C. Duncan, Mon. Not. R. Astron. Soc. 275, 255 (1995)
36. P. Bocquet, S. Bonazzola, E. Gourgoulhon, J. Novak, Astron. Astrophys. 301, 757 (1995)
37. H.C. Spruit, Astron. Astrophys. 341, L1 (1999)
38. C.Y. Cardall, M. Prakash, J.M. Lattimer, Astrophys. J. 554, 322 (2001)
39. T. Vachaspati, Phys. Lett. B 265, 258 (1991)
40. J. Ambjørn, P. Olesen, in The Proceedings of the 4th Hellenic School on Elementary Particle
Physics, Corfu, 1992 (e-print hep-ph/9304220)
41. D. Grasso, H.R. Rubinstein, Phys. Rep. 348, 163 (2001)
42. A. Kandus, K.E. Kunze, C.G. Tsagas, Phys. Rep. 505, 1 (2011)
43. A. Dar, A. Laor, N.J. Shaviv, Phys. Rev. Lett. 80, 5813 (1998)
44. A. Dar, A. De Rújula, in Astrophysics and Gamma Ray Physics in Space, Frascati Physics
Series, Vol. XXIV, ed. by A. Morselli, P. Picozza (INFN, Lab. Naz. di Frascati, Frascati, 2002),
pp. 513–523 (e-print astro-ph/0110162)
45. A.I. Akhiezer, V.B. Berestetskii, Quantum Electrodynamics, 2nd edn. (Wiley, New York, 1965)
46. A.A. Sokolov, I.M. Ternov, Radiation from Relativistic Electrons (American Institute of
Physics, New York, 1986)
47. C. Itzykson, J.-B. Zuber, Quantum Field Theory (McGraw-Hill, New York, 1985)
48. V.B. Berestetskii, E.M. Lifshitz, L.P. Pitaevskii, Quantum Electrodynamics, 2nd edn. (Perga-
mon, Oxford, 1982)
49. I.M. Ternov, V.Ch. Zhukovskii, A.V. Borisov, Quantum Processes in Strong External Field
(Moscow State University, Moscow, 1989). (in Russian)
50. V.I. Ritus, Quantum effects of the interaction of elementary particles with an intense electro-
magnetic field, in Quantum Electrodynamics of Phenomena in an Intense Field, Proceedings of
the P.N. Lebedev Physical Institute, vol. 111 (Nauka, Moscow, 1979), pp. 5–151. (in Russian)
51. V.I. Ritus, Radiative effects and their amplification in an intense electromagnetic field, in Issues
in Intense-Field Quantum Electrodynamics, ed. by V.L. Ginzburg (Nova Science Publishers,
New York, 1989), pp. 180–199
52. V.O. Papanian, V.I. Ritus, Three-photon interaction in an intense field, in Issues in Intense-Field
Quantum Electrodynamics, ed. by V.L. Ginzburg (Nova Science Publishers, New York, 1989),
pp. 153–179
53. A.E. Shabad, in Polarization of the Vacuum and a Quantum Relativistic Gas in an External
Field, ed. by V.L. Ginzburg (Nova Science Publishers, New York, 1992)
54. N.P. Klepikov, Zh. Eksp. Teor. Fiz. 26, 19 (1954)
55. S.L. Adler, J.N. Bahcall, C.G. Callan, M.N. Rosenbluth, Phys. Rev. Lett. 25, 1061 (1970)
56. Z. Bialynicka-Birula, I. Bialynicki-Birula, Phys. Rev. D 2, 2341 (1970)
57. S.L. Adler, Ann. Phys. (N.Y.) 67, 599 (1971)
10 1 Introduction

58. V.O. Papanian, V.I. Ritus, Zh. Eksp. Teor. Fiz. 61, 2231 (1971). [Sov. Phys. JETP 34, 1195
(1972)]
59. V.O. Papanian, V.I. Ritus, Zh. Eksp. Teor. Fiz. 65, 1756 (1973). [Sov. Phys. JETP 38, 879
(1974)]
60. M.G. Baring, Astrophys. J. 440, L69 (1995)
61. S.L. Adler, C. Schubert, Phys. Rev. Lett. 77, 1695 (1996)
62. V.N. Baier, A.I. Milstein, R.Zh Shaisultanov, Zh. Eksp. Teor. Fiz. 111, 52 (1997). [JETP 84,
29 (1997)]
63. C. Wilke, G. Wunner, Phys. Rev. D 55, 997 (1997)
64. M.V. Chistyakov, A.V. Kuznetsov, N.V. Mikheev, Phys. Lett. B 434, 67 (1998)
65. A.V. Kuznetsov, N.V. Mikheev, M.V. Chistyakov, Yad. Fiz. 62, 1638 (1999). [Phys. At. Nucl.
62, 1535 (1999)]
66. E.A. Choban, A.N. Ivanov, Zh. Eksp. Teor. Fiz. 56, 194 (1969). [Sov. Phys. JETP 29, 109
(1969)]
67. A.V. Borisov, V.Ch. Zhukovskii, B.A. Lysov, Izv. Vyssh. Uchebn. Zaved. Fiz. 8, 30 (1983).
[Sov. Phys. J. 26, 701 (1983)]
68. M.Yu. Knizhnikov, A.V. Tatarintsev, Vestn. Mosk. Univ. Fiz. Astron. 25, 26 (1984)
69. A.V. Borisov, A.I. Ternov, V. Ch. Zhukovsky, Phys. Lett. B 318, 489 (1993)
70. A.V. Kuznetsov, N.V. Mikheev, Phys. Lett. B 394, 123 (1997)
71. A.V. Kuznetsov, N.V. Mikheev, Yad. Fiz. 60, 2038 (1997). [Phys. At. Nucl. 60, 1865 (1997)]
72. A.V. Borisov, N.B. Zamorin, Yad. Fiz. 62, 1647 (1999). [Phys. At. Nucl. 62, 1543 (1999)]
73. A.V. Kuznetsov, N.V. Mikheev, D.A. Rumyantsev, Mod. Phys. Lett. A 15, 573 (2000)
74. A.V. Kuznetsov, N.V. Mikheev, D.A. Rumyantsev, Yad. Fiz. 65, 303 (2002). [Phys. At. Nucl.
65, 277 (2002)]
75. D.V. Galtsov, N.S. Nikitina, Zh. Eksp. Teor. Fiz. 62, 2008 (1972). [Sov. Phys. JETP 35, 1047
(1972)]
76. V.V. Skobelev, Zh. Eksp. Teor. Fiz. 71, 1263 (1976). [Sov. Phys. JETP 44, 660 (1976)]
77. A.N. Ioannisian, G.G. Raffelt, Phys. Rev. D 55, 7038 (1997)
78. A.A. Gvozdev, N.V. Mikheev, L.A. Vassilevskaya, Phys. Lett. B 410, 211 (1997)
79. L.L. DeRaad Jr, K.A. Milton, N.D. Hari Dass, Phys. Rev. D 14, 3326 (1976)
80. N.V. Mikheev, L.A. Vassilevskaya, Phys. Lett. B 410, 203 (1997)
81. A.A. Gvozdev, N.V. Mikheev, L.A. Vassilevskaya, Phys. Lett. B 289, 103 (1992)
82. A.A. Gvozdev, N.V. Mikheev, L.A. Vassilevskaya, Phys. Rev. D 54, 5674 (1996)
83. N.V. Mikheev, L.A. Vassilevskaya, Phys. Lett. B 410, 207 (1997)
84. N.V. Mikheev, L.A. Vassilevskaya, A.Ya. Parkhomenko, Phys. Rev. D 60, 035001 (1999)
85. A.V. Kuznetsov, N.V. Mikheev, L.A. Vassilevskaya, Phys. Lett. B 427, 105 (1998)
86. L.A. Vassilevskaya, A.V. Kuznetsov, N.V. Mikheev, Yad. Fiz. 62, 715 (1999). [Phys. At. Nucl.
62, 666 (1999)]
87. M.V. Chistyakov, N.V. Mikheev, Phys. Lett. B 467, 232 (1999)
88. A.V. Kuznetsov, N.V. Mikheev, Mod. Phys. Lett. A 14, 2531 (1999)
89. M.Yu. Borovkov, A.V. Kuznetsov, N.V. Mikheev, Yad. Fiz. 62, 1714 (1999). [Phys. At. Nucl.
62, 1601 (1999)]
90. A.V. Kuznetsov, N.V. Mikheev, Zh. Eksp. Teor. Fiz. 118, 863 (2000). [JETP 91, 748 (2000)]
91. N.V. Mikheev, E.N. Narynskaya, Mod. Phys. Lett. A 15, 1551 (2000)
92. A.V. Kuznetsov, N.V. Mikheev, Mod. Phys. Lett. A 16, 1659 (2001)
93. M.V. Chistyakov, N.V. Mikheev, Pis’ma Zh. Eksp. Teor. Fiz. 73, 726 (2001). [JETP Lett. 73,
642 (2001)]
94. A.V. Kuznetsov, N.V. Mikheev, Pis’ma Zh. Eksp. Teor. Fiz. 75, 531 (2002). [JETP Lett. 75,
441 (2002)]
95. A.V. Kuznetsov, N.V. Mikheev, M.V. Osipov, Mod. Phys. Lett. A 17, 231 (2002)
96. M.V. Chistyakov, N.V. Mikheev, Mod. Phys. Lett. A 17, 2553 (2002)
97. A.V. Kuznetsov, N.V. Mikheev, D.A. Rumyantsev, Yad. Fiz. 66, 319 (2003). [Phys. At. Nucl.
66, 294 (2003)]
98. N.V. Mikheev, E.N. Narynskaya, Cent. Eur. J. Phys. 1, 145 (2003)
References 11

99. A.V. Kuznetsov, N.V. Mikheev, G.G. Raffelt, L.A. Vassilevskaya, Phys. Rev. D 73, 023001
(2006)
100. A.V. Kuznetsov, N.V. Mikheev, Yad. Fiz. 70, 1299 (2007). [Phys. At. Nucl. 70, 1258 (2007)]
101. A.V. Kuznetsov, N.V. Mikheev, Mod. Phys. Lett. A 21, 1769 (2006)
102. A.V. Kuznetsov, N.V. Mikheev, Int. J. Mod. Phys. A 22, 3211 (2007)
103. A.V. Kuznetsov, N.V. Mikheev, J. Cosm. Astropart. Phys. 11, 031 (2007)
104. N.V. Mikheev, E.N. Narynskaya, Int. J. Mod. Phys. A 23, 3747 (2008)
105. A.V. Kuznetsov, N.V. Mikheev, A.A. Okrugin, Int. J. Mod. Phys. A 24, 5977 (2009)
106. A.V. Kuznetsov, N.V. Mikheev, A.A. Okrugin, Zh. Eksp. Teor. Fiz. 138, 80 (2010). [J. Exp.
Theor. Phys. 111, 70 (2010)]
107. M.S. Andreev, N.V. Mikheev, E.N. Narynskaya, Zh. Eksp. Teor. Fiz. 137, 259 (2010). [J. Exp.
Theor. Phys. 110, 227 (2010)]
108. R.A. Anikin, N.V. Mikheev, E.N. Narynskaya, Zh. Eksp. Teor. Fiz. 137, 1115 (2010). [J. Exp.
Theor. Phys. 110, 973 (2010)]
109. A.V. Kuznetsov, N.V. Mikheev, A.V. Serghienko, Phys. Lett. B 690, 386 (2010)
110. R.A. Anikin, A.V. Kuznetsov, N.V. Mikheev, Pis’ma Astron. Zh. 36, 714 (2010). [Astron.
Lett. 36, 680 (2010)]
111. R.A. Anikin, A.V. Kuznetsov, N.V. Mikheev, Yad. Fiz. 73, 2000 (2010). [Phys. At. Nucl. 73,
1948 (2010)]
112. A.V. Kuznetsov, A.A. Okrugin, Int. J. Mod. Phys. A 26, 2725 (2011)
113. A.V. Kuznetsov, N.V. Mikheev, A.M. Shitova, Int. J. Mod. Phys. A 26, 4773 (2011)
114. J.D. Bjorken, S.D. Drell, Relativistic Quantum Mechanics (McGraw-Hill, New York, 1964)
Chapter 2
Solutions of the Dirac Equation in an External
Electromagnetic Field

In this chapter, the solutions of the Dirac equation for a fermion in an external
electromagnetic field are presented for the cases of a pure magnetic field of arbitrary
strength, of a strong magnetic field when fermions occupy the ground Landau level,
and of a crossed field. The density matrix of the plasma electron in a magnetic field
with the fixed number of a Landau level is calculated. In this chapter, we use the
notation for the 4-vectors and their components: X μ = (t, x, y, z).

2.1 Magnetic Field

For calculation of the S matrix elements of quantum processes in external fields, the
standard procedure is applied, which is based on the Feynman diagram technique
using the field operators of charged fermions expanded over the solutions of the Dirac
equation in an external magnetic field
 
(+) (−)
Ψ̂ (X ) = âp,s Ψp,s (X ) + b̂p,s
†
Ψp,s (X ) , (2.1)
p,s

where â is the destruction operator for fermions, b̂† is the creation operator for
antifermions, and Ψ (+) (X ) and Ψ (−) (X ) are the normalized solutions of the Dirac
equation in a magnetic field with positive and negative energy, correspondingly.
There exist several methods of solving the Dirac equation in a magnetic field
which are basically the similar but have some variations in details, see e.g. [1–5].
Here we present the basic points of the procedure which is the most simple and
clear, in our opinion. The description is similar to the one of Ref. [5]. As a charged
fermion, we consider an electron being the particle having the largest specific charge,
i.e. being the most sensitive to the external field influence. More general case for an
arbitrary charged fermion can be found e.g. in [1].

A. Kuznetsov and N. Mikheev, Electroweak Processes in External Active Media, 13

Springer Tracts in Modern Physics 252, DOI: 10.1007/978-3-642-36226-2_2,
© Springer-Verlag Berlin Heidelberg 2013
14 2 Solutions of the Dirac Equation in an External Electromagnetic Field

The Dirac equation for an electron with the mass m e and the charge (−e) in an
external electromagnetic field with the four-potential Aμ = Aμ (X ) has the form
 
i(∂γ) + e(Aγ) − m e Ψ (X ) = 0, (2.2)

where (∂γ) = ∂μ γ μ and (Aγ) = Aμ γ μ . For solving the Eq. (2.2) in a pure magnetic
field B, we take the frame where the field is directed along the z axis, and the Landau
gauge where the four-potential is: Aμ = (0, 0, x B, 0).
To solve the Eq. (2.2), let us rewrite it in the Schrödinger form:

∂
i Ψ (X ) = Ĥ Ψ (X ), (2.3)
∂t
with the Hamiltonian:  
Ĥ = γ0 γ p̂ + eA + m e γ0 . (2.4)

Here, p̂ = −i∇ is the momentum operator.

Since the Hamiltonian does not depend explicitly on time, the problem reduces to
finding the eigenvalues and eigenfunctions of the Schrödinger stationary equation:

Ψ (X ) = e−i p0 t ψ(x, y, z), Ĥ ψ(x, y, z) = p0 ψ(x, y, z). (2.5)

Consider the auxiliary operator, called the longitudinal polarization operator:

1  
T̂ 0 =  p̂ + eA , (2.6)
me

where  is the 3-dimensional double spin operator:

 
σ 0
 = γ0 γγ5 = , (2.7)
0 σ

and σ are the Pauli matrices. It is easy to verify by direct calculation that the operator
T̂ 0 commutes with the Hamiltonian (2.4).
First, we find the eigenvalues and the eigenfunctions of the operator T̂ 0 ,

T̂ 0 ψT (x, y, z) = T 0 ψT (x, y, z). (2.8)

The functions ψT (x, y, z) are also the eigenfunctions of the Hamiltonian (2.4), due
to commutativity of Ĥ and T̂ 0 .
It is convenient to represent the operator T̂ 0 in the form
 
τ̂ 0 0
T̂ 0 = , (2.9)
0 τ̂ 0
2.1 Magnetic Field 15

where
1  
τ̂ 0 = σ p̂ + eA . (2.10)
me

By the structure of the operator T̂ 0 , the system (2.8) of 4 equations splits into two
exactly coinciding equations for the upper and lower spinors forming the bispinor
ψT (x, y, z).
In the chosen gauge, the operator τ̂ 0 has the form:
     
1 ∂ ∂ ∂
τ̂ =
0
σx −i + σ y −i + βx + σz −i , (2.11)
me ∂x ∂y ∂z

where the notation is used: β = eB. Given the operator T̂ 0 not depending explicitly
on the coordinates of y and z, one can write the bispinor ψT (x, y, z) in the form:
   
F(x) f 1 (x)
ψT (x, y, z) = ei( p y y+ pz z) , F(x) = , (2.12)
κ F(x) f 2 (x)

where κ is an arbitrary number. Introducing a new variable

 
py
ξ= β x+ , (2.13)
β

one can transform the equation for the spinor F(x) to the form:
 √ −
   
√ pz + −i 2βa f 1 (ξ) f 1 (ξ)
1
=T 0
, (2.14)
me i 2βa − pz f 2 (ξ) f 2 (ξ)

where the raising and lowering operators of the problem of the quantum harmonic
oscillator arise:
   
+ 1 d − 1 d
a = √ ξ− , a = √ ξ+ . (2.15)
2 dξ 2 dξ

The expression (2.14) is a system of differential equations for the functions f 1 (ξ)
and f 2 (ξ). We obtain:
√
−i 2β
f 1 (ξ) = a − f 2 (ξ),
m e T 0 − pz
 
+ − m 2e (T 0 )2 − pz2
a a − f 2 (ξ) = 0. (2.16)
2β

Multiplying the operators (2.15), one can see that the equation for the function f 2 (ξ)
is reduced to an equation for eigenfunctions of the quantum harmonic oscillator:
16 2 Solutions of the Dirac Equation in an External Electromagnetic Field
 
d2 m 2e (T 0 )2 − pz2
− ξ 2
+ 1 + f 2 (ξ) = 0. (2.17)
dξ 2 β

Hence, we find the eigenvalues T 0 of the operator T̂ 0 :

1  2
T0 = ± pz + 2nβ. (2.18)
me

Here, n = 0, 1, 2, . . . . These numbers, as we shall see below, will determine the

electron energy, i.e., will number the Landau levels. It should be noted that the
eigenvalues T 0 are gauge invariant, being the eigenvalues of the Hermitian operator,
i.e. the physically observable quantities.
The functions f 1 (ξ) and f 2 (ξ) are
√
−i 2nβ
f 1 (ξ) = C Vn−1 (ξ), f 2 (ξ) = C Vn (ξ), (2.19)
m e T 0 − pz

where C is the normalization coefficient, and Vn (ξ) (n = 0, 1, 2, . . . ) are the nor-

malized harmonic oscillator functions, which are expressed in terms of Hermite
polynomials Hn (ξ):

β 1/4 2 d
n
e− ξ /2 Hn (ξ), Hn (ξ) = (−1)n eξ e− ξ ,
2 2
Vn (ξ) = √ n
2 n! π
n dξ
+∞
|Vn (ξ)|2 dx = 1, (2.20)
−∞

and for negative values of the index n the function Vn (ξ) is assumed to be zero.
Returning to the stationary Schrödinger equation (2.5), let us substitute into it
the found function ψT (x, y, z) as an eigenfunction. The Hamiltonian (2.4) can be
expressed in terms of the operator τ̂ 0 :
 
I τ̂ 0
Ĥ = m e . (2.21)
τ̂ 0 −I

In view of (2.14), we obtain the equation:

       
I 0 0 I F(x) F(x)
me +T 0
= p0 , (2.22)
0 −I I 0 κ F(x) κ F(x)

which is transformed to a system of algebraic equations for p0 and κ having two

solutions. Two eigenvalues of the stationary Schrödinger equation (2.5) are:

( p0 )1,2 = ±E n , (2.23)
2.1 Magnetic Field 17

where 
En = m e (T 0 )2 + 1 = pz2 + m 2e + 2nβ. (2.24)

The values of κ corresponding to the two eigenvalues p0 are:

 
En − m e En + m e
κ1 = sign(T ) 0
, κ2 = −sign(T 0 ) . (2.25)
En + m e En − m e

Thus, given the ambiguity of T 0 (2.18), there exist four independent solutions of
the Eq. (2.2).
(i) The eigenvalue p0 = +E n .
The solutions corresponding to the positive eigenvalue p0 = +E n , called the
solutions with positive energy, which differ in sign of T 0 , can be written as:

Ψ (+±) (X ) = A(+±) e−i(E n t− p y y− pz z) u (+±) (ξ). (2.26)

Here, the first of two signs in the superscript refers to p0 , while the second one refers
to T 0 . For the bispinors u (+±) (ξ) we obtain the expressions:
⎛ √ ⎞
−i 2nβ
m e |T 0 |− pz
Vn−1 (ξ)
⎜ ⎟
⎜ Vn (ξ) ⎟
⎜ ⎟
⎜ ⎟
u (++) (ξ) = ⎜  E n −m e √
−i 2nβ ⎟, (2.27)
⎜ E +m V (ξ) ⎟
⎜ n e m e |T 0 |− pz n−1 ⎟
⎝  ⎠
E n −m e
E n +m e n V (ξ)
⎛ √
−i 2nβ
⎞
−m e |T 0 |− pz
Vn−1 (ξ)
⎜ ⎟
⎜ Vn (ξ) ⎟
⎜  ⎟
u (+−) ⎜
(ξ) = ⎜ √ ⎟. (2.28)
E n −m e −i 2nβ ⎟
⎜ − E n +m e −m e |T 0 |− pz Vn−1 (ξ) ⎟
⎝  ⎠
− EE nn −m e
+m e V n (ξ)

The functions (2.26)–(2.28), as well as any of their linear combinations, are the
solutions of the Dirac equation (2.2), corresponding to the eigenvalue p0 = +E n .
As in the analysis of solutions of the Dirac equation in vacuum, the solutions in a
magnetic field are typically used in the form of linear combinations of the functions
(2.26)–(2.28), in which the upper two components of the bispinor correspond to the
states of the electron with the spin projections 1/2 and −1/2 on some direction, in
this case, on the direction of the magnetic field.
Given the normalization

|Ψ (X )|2 dx dy dz = 1, (2.29)
18 2 Solutions of the Dirac Equation in an External Electromagnetic Field

we obtain the final form of the exact solutions of the Dirac equation for an electron
in an external magnetic field on the n-th Landau level:

e− i (E n t− p y y− pz z)
Ψn,(+)p y , pz , s (X ) = (+)
Un, p y , pz , s (ξ), (2.30)
2 E n (E n + m e ) L y L z

where L y and L z are the normalizing sizes along the axes of y and z; the number s =
± 1 is the eigenvalue of the double spin operator σz acting on the spinor composed
of the upper two components of the bispinor.
The bispinor U (+) has different forms for the cases s = +1 and s = −1:
⎛ ⎞
(E n + m e ) Vn−1 (ξ)
⎜ ⎟
(+) ⎜ 0 ⎟
Un, p y , pz , s=+1 (ξ) = ⎜ ⎟, (2.31)
⎝ p V
z n−1 (ξ) ⎠
√
i 2nβ Vn (ξ)
⎛ ⎞
0
⎜ (E n + m e ) Vn (ξ) ⎟
(+) ⎜ √ ⎟
Un, p y , pz , s=−1 (ξ) = ⎜ ⎟. (2.32)
⎝ − i 2nβ Vn−1 (ξ) ⎠
− pz Vn (ξ)

One can see that in each of the bispinors (2.31) and (2.32), the upper two components
form a spinor being the eigenfunction of the operator σz . For the ground Landau level,
n = 0, the solution exists only at s = −1.
Note that the value pz in above expressions is a conserved component of the
electron momentum along the z axis, i.e. along the field, while the value p y is the
generalized momentum, which determines the position of a center of the Gaussian
packet along the x axis by the relation x0 = − p y /β (see (2.13)).
(ii) The eigenvalue p0 = −E n .
The solutions Ψ (−±) (X ) corresponding to this eigenvalue describe the states of an
electron with negative energy in the Dirac sea. To obtain the functions corresponding
to the states of a positron as a physical particle with the energy E n and the momentum
components p y and pz , one should construct the solutions Ψ (−±) (X ) which are
similar to the functions (2.26)–(2.28), in view of (2.25), and then change the signs
of p y and pz . One should also remember that the projection of the spin of a positron,
i.e. of a hole in the sea of negative energies, on any special direction is opposite to
the spin projection of the electron, described by a bispinor.
There are two main variants of constructing the solutions with a negative energy,
with using of different linear combinations of the functions Ψ (−+) (X ) and Ψ (−−) (X ),
which, of course, lead to identical results in calculations of observable quantities. In
the first case, one can simply use the solutions (2.30)–(2.32) and change there the
signs of E n , p y , and pz . The second way is perhaps more physically justified. As
in the analysis of the Dirac equation in vacuum, one can consider the solutions in
which the upper two components of a bispinor are small, if the nonrelativistic limit,
2.1 Magnetic Field 19

pz2  m 2e , and the case of a weak field, β  m 2e , are taken. In this case, the linear
combinations of the functions Ψ (−+) (X ) and Ψ (−−) (X ) should be used, in which
the spinor composed of the two lower components of a bispinor, describes the states
of the electron with the spin projections 1/2 and −1/2 on some direction, in this case,
on the direction of the magnetic field.
The exact solutions of the Dirac equation corresponding to the positron states in
an external magnetic field, on the nth Landau level have the form:

ei (E n t− p y y− pz z)
Ψn,(−)p y , pz , s (X ) = (−)
Un, p y , pz , s (ξ
(−)
), (2.33)
2 E n (E n + m e ) L y L z

where the number s = ± 1 is the eigenvalue of the double electron spin operator σz
acting on the spinor composed of the two lower components of the bispinor,
 
py
ξ (−) = β x− , (2.34)
β
⎛ ⎞
pz Vn−1 (ξ (−) )
⎜ √ ⎟
(−) ⎜ −i 2nβ Vn (ξ (−) ) ⎟
Un, p y , pz , s=+1 (ξ (−) ) = ⎜ ⎟, (2.35)
⎝ (E n + m e ) Vn−1 (ξ (−) ) ⎠
0
⎛ √ ⎞
i 2nβ Vn−1 (ξ (−) )
⎜ ⎟
(−) (−) ⎜ − pz Vn (ξ (−) ) ⎟
Un, p y , pz , s=−1 (ξ ) = ⎜ ⎟. (2.36)
⎝ 0 ⎠
(E n + m e ) Vn (ξ (−) )

For the ground Landau level, n = 0, the solution exists only for the value of the
double spin s = −1 of an electron with negative energy. This corresponds to the
positron state with a value of the double spin s = +1.

2.2 The Ground Landau Level

If some physical process with electrons/positrons, with a typical energy E is realised

in a strong magnetic field, where the field induction B determines the maximum
energy scale of a problem, namely, eB > E 2 , m 2e , electrons/positrons can occupy
only the states that correspond to the ground Landau level, n = 0. Contrary to other
Landau levels with n  1, which are doubly degenerate with respect to spin, the
ground level is not degenerate, i.e. the electron/positron spin is fixed, s = −1/ + 1.
The solution of the Dirac equation for the electron with energy E and momentum
components p y and pz can be presented in this case in the following form
20 2 Solutions of the Dirac Equation in an External Electromagnetic Field

(+) β 1/4 e−i(Et− p y y− pz z)

e−ξ /2 u( p ) ,
2
Ψ0, p y , pz , s=−1 (X ) = √ (2.37)
( π 2 E (E + m e )L y L z )1/2

where p is the energy-momentum vector of an electron in the Minkowski {0,3}

plane. Here, E = pz 2 + m 2e , and ξ is defined by (2.13) and describes the motion
along the x axis.
The bispinor amplitude is given by
⎛ ⎞
0
⎜ E + me ⎟
u( p ) = ⎜
⎝ 0 ⎠.
⎟ (2.38)
− pz

It is interesting to note that the bispinor amplitude (2.38) is exactly the same as the
solution of the free Dirac equation for an electron having a momentum directed along
the z axis. This separation of a bispinor amplitude that does not depend on the spatial
coordinate x is typical for the ground Landau level only.
The calculation technique of electroweak processes in a strong magnetic field,
where electrons occupy the ground Landau level, the so-called two-dimensional
electrodynamics, was developed by Loskutov and Skobelev [6, 7]; for details and
a complete list of references see e.g. [8]. That technique was essentially improved,
with a covariant extension, in our papers; see e.g. [9–14]. For example, the antisym-
metric tensor εαβ (ε30 = −ε03 = 1) in the subspace {0, 3}, used in that technique,
appears to be not a mathematical abstraction, but has a clear physical meaning of the
dimensionless dual magnetic field tensor, εαβ = −ϕ̃αβ . Similarly, all the formulae
can be written in a covariant form with obvious rules of transformation to any frame.

2.3 Crossed Field

There exists a special case of external electromagnetic field, in which the analysis of
quantum processes is essentially simplified. It is the case of a crossed field, where
the vectors of the electric field E and the magnetic field B are orthogonal and their
values are equal, E ⊥ B, E = B. The calculation technique of electromagnetic
processes in the crossed field was developed by Nikishov and Ritus; for details and
the list of references see e.g. [15, 16].
The particular case of a crossed field is in fact more general than it may seem at first
glance. Really, the situation is possible when the so-called field dynamical parameter
χ of the relativistic particle propagating in a relatively weak electromagnetic field,
F < Be (F = E and/or B), could appear rather high. The definition of the dynamical
parameter χ is
e( p F F p)1/2
χ= , (2.39)
m 3e
2.3 Crossed Field 21

where p α is the particle four-momentum, and F αβ is the electromagnetic field tensor.

In this case the field in the particle rest frame can exceed essentially the critical
value and is very close to the crossed field. Even in a magnetic field whose strength
is much greater than the critical value, the result obtained in a crossed field will
correctly describe the leading contribution to the probability of a process in a pure
magnetic field, provided that χ  B/Be . If, in addition, the invariant |e2 ( p F F p)|1/3
for a particle moving in an arbitrary electromagnetic field considerably exceeds the
pure field invariants |e2 (F F)|1/2 and |e2 ( F̃ F)|1/2 , the problem is reducible to a
still simpler calculation, that in a crossed field for which one has (F F) = 0 and
( F̃ F) = 0. Thus, the calculation in a constant crossed field is the relativistic limit of
the calculation in an arbitrary relatively weak smooth field. Consequently, the results
obtained in a crossed field possess a great extent of generality, and acquire interest
by itself.
The crossed field is described by the 4-vector potential Aμ = a μ ϕ, where ϕ =
(k X ), and a μ and k μ are the constant 4-vectors, (kk) = 0, (ak) = 0.
The field tensor in this case is F μν = k μ a ν − k ν a μ , and the contraction of the
two tensors over one index is (F F)μν = −k μ k ν (aa).
The solution of the Dirac equation for an electron in the crossed field can be
found as a particular case of the Dirac equation solution in the field of a plane
electromagnetic wave obtained by Volkov [17, 18], where the above-mentioned linear
dependence of the field vector potential on the phase ϕ, Aμ = a μ ϕ, should be taken.
The solution has the form
 
e(kγ)(aγ) u( p)
Ψ p (X ) = 1 − ϕ √
2(kp) 2E V
 
e(ap) 2 e2 (aa) 3
× exp −i ( p X ) − ϕ − ϕ . (2.40)
2(kp) 6(kp)

where u( p) is the bispinor amplitude of a free electron with the 4-momentum p μ =

(E, p).
The solution with negative energy corresponding to an antiparticle can be obtained
from (2.40) by the change of sign of all the components of the 4-momentum p μ .
The directions of the coordinate frame axes can be taken as follows, without loss
of generality:
k μ = (k0 , k0 , 0, 0), a μ = (0, 0, −a, 0). (2.41)

In this case

ϕ = (k X ) = k0 (t − x), E = (0, E, 0), B = (0, 0, B), E = B = k0 a.

It is worthwhile to introduce also the vector bμ = (0, 0, 0, −a), which can be used
for representing the dual tensor F̃ μν = 21 εμνρσ Fρσ by the following form F̃ μν =
k μ bν − k ν bμ .
22 2 Solutions of the Dirac Equation in an External Electromagnetic Field

2.4 Density Matrix of the Plasma Electron in a Magnetic Field

with the Fixed Number of a Landau Level

When quantum processes in a magnetized plasma are investigated, it is occasionally

necessary to calculate the plasma electron density matrix in the coordinate space,
summed over all quantum states, except the Landau level number. In this Section,
we present the calculations of this matrix which can be defined by the formula:
  d p y d pz

Rn (X, X ) = L y L z f (E n ) Ψn,(+)p y , pz , s (X ) Ψ̄n,(+)p y , pz , s (X  ). (2.42)
s
(2π) 2

(+)
Here, Ψn, p y , pz , s (X ) are the solutions (2.30)–(2.32) of the Dirac equation for an

electron in an external magnetic field, E n = pz2 + m 2e + 2nβ is the energy of the
electron at the nth Landau level, β = eB, and f (E n ) is the electron distribution
function that allows for the presence of a plasma. In the plasma rest frame, it is

f (E) = [e(E−μ)/T + 1]−1 ,

where μ is the chemical potential of plasma and T is its temperature.

Substituting the explicit form of the electron wave functions (2.30)–(2.32) into
Eq. (2.42) for the density matrix, we can reduce it to the form
 
Rn (X, X  ) = eiΦ(X,X ) Rn ((X − X  ) ) Rns⊥ ((X − X  )⊥ ). (2.43)
s

Here, the following functions are introduced:

β
Φ(X, X  ) = − (x + x  ) (y − y  ), (2.44)
2
+∞
d pz
Rn (X  ) = f (E n ) e−i( p X ) , (2.45)
E n (E n + m e )
−∞
√ +∞ √ √
β
Rns⊥ (X ⊥ ) = dξ Us (ξ) Ūs (ξ − βx) e−i β( βx y/2−ξ y)
, (2.46)
8π 2
−∞

where we changed the integration variable from p y to ξ, see (2.13). In Eq. (2.46),
(+)
we have omitted all the indexes except s of the bispinors Un, p y , pz , s . However, one
should keep in mind that there is not a simple product of the functions Rn and Rns⊥
stands in Eq. (2.43), because Rns⊥ depends on pz .
The function Rns⊥ (X ⊥ ) as a function of two variables x and y can be expanded
into a Fourier integral:
2.4 Density Matrix of the Plasma Electron in a Magnetic Field 23

d2 p⊥ i( p X )⊥
Rns⊥ (X ⊥ ) = e Rns⊥ ( p⊥ ), (2.47)
(2π)2

Rns⊥ ( p⊥ ) = d2 X ⊥ e−i( p X )⊥ Rns⊥ (X ⊥ ) . (2.48)

Integrating the function Rns⊥ ( p⊥ ) over the coordinates x and y and substituting
the result into (2.43) yields
 
eiΦ(X,X ) d3 p f (E n ) −i( p(X −X  )) 2i px p y /β
Rn (X, X  ) = √ e e
(2π) β
3 E n (E n + m e )
+∞ √ 
× dξ e−2i px ξ/ β Us (ξ) Ūs (ξ  ), (2.49)
−∞ s

√
where ξ  = 2 p y / β − ξ.
After simple but slightly cumbersome calculations, including the summation over
the spin states of the initial and final electrons that occupy the same Landau level n,
the product of the bispinor amplitudes can be reduced to

Us (ξ) Ūs (ξ  ) = (E n + m e )
s
× {(( pγ) + m e ) [+ Vn−1 (ξ)Vn−1 (ξ  ) + − Vn (ξ)Vn (ξ  )]
− 2nβ [+ γ 2 Vn−1 (ξ)Vn (ξ  ) + − γ 2 Vn (ξ)Vn−1 (ξ  )]}. (2.50)

Here, the projection operators are introduced:

1
± = (I ± i γ 1 γ 2 ) , ± ± = ± , ± ∓ = 0. (2.51)
2
The integral over the variable ξ in Eq. (2.49) can be calculated using the formula

+∞
eiab/2
Jn,n  = √ dξ e−ia ξ Vn (ξ)Vn  (b − ξ)
β
−∞
n  −i(n−n  )ϕ
= (−1) e Fn  ,n (u), n  n , (2.52)

where

a a 2 + b2
tan ϕ = , u= ,
b 4

n!  
Fn  ,n (u) = (2u)(n−n )/2 e−u L n−n
n  (2u),
n!
24 2 Solutions of the Dirac Equation in an External Electromagnetic Field

and the associated Laguerre polynomials L sk (x) are defined as

1 x −s dk −x k+s
L sk (x) = e x (e x ) . (2.53)
k! dx k
Finally, the electron density matrix can be reduced to a triple integral convenient
for the subsequent use:

 iΦ(X,X  ) d3 p f (E n ) −u −i p(X −X  )
Rn (X, X ) = e (−1) n
e e (2.54)
(2π)3 E n
× {(( pγ) + m e )[L n (2u)− − L n−1 (2u)+ ] + 2( pγ)⊥ L 1n−1 (2u)},

where u = p⊥ 2 /β. Equation (2.54) can be used to investigate quantum processes in

a plasma in the presence of a magnetic field with an arbitrary strength.

References

1. A.V. Kuznetsov, N.V. Mikheev, Electroweak Processes in External Electromagnetic Fields

(Springer, New York, 2003)
2. A.I. Akhiezer, V.B. Berestetskii, Quantum Electrodynamics, 2nd edn. (Wiley, New York, 1965)
3. A.A. Sokolov, I.M. Ternov, Radiation from Relativistic Electrons (American Institute of
Physics, New York, 1986)
4. K. Bhattacharya, P.B. Pal, Pramana. J. Phys. 62, 1041 (2004)
5. I.A. Balantsev, Yu.V. Popov, A.I. Studenikin, J. Phys. A 44, 255301 (2011)
6. V.V. Skobelev, Izv. Vyssh. Uchebn. Zaved., Fiz. No. 10, 142 (1975)
7. Yu.M. Loskutov, V.V. Skobelev, Phys. Lett. A 56, 151 (1976)
8. V.V. Skobelev, Zh. Eksp. Teor. Fiz. 120, 786 (2001) [JETP 93, 685 (2001)]
9. M.V. Chistyakov, A.V. Kuznetsov, N.V. Mikheev, Phys. Lett. B 434, 67 (1998)
10. A.V. Kuznetsov, N.V. Mikheev, Yad. Fiz. 60, 2038 (1997) [Phys. At. Nucl. 60, 1865 (1997)]
11. A.A. Gvozdev, N.V. Mikheev, L.A. Vassilevskaya, Phys. Rev. D 54, 5674 (1996)
12. M.Y. Borovkov, A.V. Kuznetsov, N.V. Mikheev, Yad. Fiz. 62, 1714 (1999) [Phys. At. Nucl. 62,
1601 (1999)]
13. A.V. Kuznetsov, N.V. Mikheev, G.G. Raffelt, L.A. Vassilevskaya, Phys. Rev. D 73, 023001
(2006)
14. A.V. Kuznetsov, N.V. Mikheev, Yad. Fiz. 70, 1299 (2007) [Phys. At. Nucl. 70, 1258 (2007)]
15. V.I. Ritus, Quantum effects of the interaction of elementary particles with an intense electro-
magnetic field (in Russian). in Quantum Electrodynamics of Phenomena in an Intense Field
Proceeding, P.N. Lebedev Physical Institute, vol. 111 (Nauka, Moscow, 1979) pp. 5–151
16. V.I. Ritus, Radiative Effects and Their Amplification in an Intense Electromagnetic Field,
in Issues in Intense-Field Quantum Electrodynamics, ed. by V.L. Ginzburg (Nova Science
Publishers, New York, 1989), pp. 180–199
17. D.M. Volkov, Z. Phys. 94, 250 (1935)
18. D.M. Volkov, Zh Eksp, Teor. Fiz. 7, 1286 (1937)
Chapter 3
Propagators of Charged Particles in External
Active Media

In this chapter, we give different representations of the charged particle propagators

in an external active environment that will be needed for the analysis of quantum
processes. The transformation from one representation to another are provided which
can be useful from a methodological point of view. The exact propagator for an
electron in a constant uniform magnetic field as the sum over Landau levels is obtained
by the direct derivation by standard methods of quantum field theory from exact
solutions of the Dirac equation in the magnetic field. In this chapter, we use the
notation for the 4-vectors and their components: X μ = (t, x, y, z); digital indices
are used to enumerate the various 4-vectors. Throughout the chapter, all the masses
squared are assumed to have small negative imaginary parts, m 2 → m 2 − i.

3.1 Propagators of Charged Particles in a Magnetic Field

The magnetic field influence on the particle properties is determined by the specific
charge, i. e. by the particle charge and mass ratio. Hence, the charged fermion which
is the most sensitive to the external field influence is the electron. The calculations
of specific physical phenomena in strong external field are based on the application
of Feynman diagram technique generalization. It consists in the following proce-
dure: in initial and final states the electron is described by the exact solution of the
Dirac equation in the external field, and internal electron lines in quantum processes
correspond to exact propagators that are constructed on the basis of these solutions.
The expression for the exact electron propagator in the constant uniform magnetic
field was obtained by J. Schwinger [1] in the Fock proper-time formalism [2]; see
e.g. [3]. There are another propagator representations given in a number of works.
Thus, in Refs. [4, 5] the case was considered of superstrong field and the contribution
of the ground Landau level to the electron propagator was obtained. In Ref. [6], see
also Ref. [7], the propagator was transformed from the form of Ref. [1] into the sum
over Landau levels. Also in Ref. [7] the electron propagator decomposition over the
power series of the magnetic field strength was given.

A. Kuznetsov and N. Mikheev, Electroweak Processes in External Active Media, 25

Springer Tracts in Modern Physics 252, DOI: 10.1007/978-3-642-36226-2_3,
© Springer-Verlag Berlin Heidelberg 2013
26 3 Propagators of Charged Particles in External Active Media

In our opinion, it is quite important to know different representations of the

electron propagator in the external magnetic field and the conditions of their
applicability. There were some examples where misunderstanding of such condi-
tions has led to erroneous papers. Thus, in Refs. [8, 9] the self-energy operator of a
neutrino in the magnetic field was calculated by the analysis of the one-loop diagram
ν → e− W + → ν. The authors of the paper restricted themselves by consideration
of the ground Landau level contribution to the electron propagator. As was shown in
Ref. [10], because of large virtuality of an electron, q 2 ∼ m 2W , the ground Landau
level contribution was not dominant and the next levels gave the contributions of
the same order of magnitude. Ignoring that fact led the authors [8, 9] to erroneous
results. Another example of this kind was an attempt to re-analyse the probability
of the ultrahigh-energy neutrino decay ν → e− W + in the external magnetic field,
which was calculated through the imaginary part of the one-loop amplitude of the
above-mentioned transition ν → e− W + → ν. Initially, this probability was calcu-
lated in Ref. [11]. Another authors [12] performed a new calculation, insisting on
a different result. The third calculation, we carried out [13], confirmed the result of
Ref. [11]. The most likely reason for the error in the calculation [12] was that the
authors used the propagator of the W -boson in an external field in the decomposition
over the tensor F μν , and they limited themselves with only linear terms, while the
quadratic terms were also significant in that case.
Among papers devoted to the study of the particle propagators in an external
field, an article [14] should be highlighted, where the computation carried out of
the neutrino self-energy operator in a magnetic field in an arbitrary ξ-gauge. It was
demonstrated that, although the self-energy operator depended on the gauge para-
meter ξ, the neutrino observable characteristics arising from its dispersion law, as
expected, were gauge-invariant.

3.1.1 Propagators in the Fock Proper-Time Presentation

The electron propagator in the constant uniform magnetic field in the Fock proper-
time formalism can be presented in the form

S (e) (X 1 , X 2 ) = eiΦ(X 1 ,X 2 ) S(X 1 − X 2 ). (3.1)

Here, S(X ) is the translational and gauge invariant part of the propagator
∞   
iβ ds 1
S(X ) = − cos(βs)(X ϕ̃ϕ̃γ) − i sin(βs)(X ϕ̃γ)γ5
2(4π)2 s sin(βs) s
0

β
− (X ϕϕγ) + m e [2 cos(βs) − sin(βs)(γϕγ)]
sin(βs)
  
(X ϕ̃ϕ̃X ) β
× exp −i m 2e s + − (X ϕϕX ) , (3.2)
4s 4 tan(βs)
3.1 Propagators of Charged Particles in a Magnetic Field 27

where β = eB, e being the elementary charge, m e is the electron mass, X μ =

(X 1 − X 2 )μ , the s variable is the Fock proper time, ϕαβ being the dimensionless
strength tensor of the external field, ϕαβ = Fαβ /B; and ϕ̃αβ = 21 εαβρσ ϕρσ is the
dual tensor.
Integration over the variable s in (3.2) should be correctly defined because the
integrand has the poles in the points s = πk/β, where k = 0, 1, 2 . . . . The integration
is supposed to be performed in the complex plane s along a contour starting from
the point s = 0 and underlying the real axis. The contour can be also turned down
to the negative imaginary axis; see Sect. 3.1.5 below.
The phase Φ(X 1 , X 2 ) is the translational and gauge noninvariant value, and can
be defined in terms of an integral along an arbitrary contour as

X 2
Φ(X 1 , X 2 ) = −e dX μ K μ (X ), (3.3)
X1

μ μ 1 μν
K (X ) = A (X ) + F (X − X 2 )ν . (3.4)
2
The integration path from X 1 to X 2 in (3.3) is arbitrary due to the relation ∂ μ K ν −
∂ ν K μ = 0. For more details on the noninvariant phase see below, Sect. 3.1.2.
Similarly to Eq. (3.1), one can define the propagators of the W boson and the
charged scalar  boson in a magnetic field (we consider negative charged W − and
− bosons as particles):

G (W )
μν (X 1 , X 2 ) = e
iΦ(X 1 ,X 2 )
G μν (X 1 − X 2 ), (3.5)
() iΦ(X 1 ,X 2 )
D (X 1 , X 2 ) = e D(X 1 − X 2 ), (3.6)

where the phase Φ(X 1 , X 2 ) is defined by the same Eqs. (3.3), (3.4).
It can be convenient to use the Fourier transforms of the translational invariant
parts of the propagators:

d4 q
S(X ) = S(q) e−iq X , (3.7)
(2π)4

d4 q
G μν (X ) = G μν (q) e−iq X , (3.8)
(2π)4

d4 q
D(X ) = D(q) e−iq X . (3.9)
(2π)4

From Eqs. (3.2) and (3.7), one can obtain the Fourier transform of the electron
propagator in the form
28 3 Propagators of Charged Particles in External Active Media

∞   
ds 2 tan(βs)
S(q) = exp −is m e − q + q⊥
2 2
cos(βs) βs
0
   
(γϕγ) (qγ)⊥
× (qγ) + m e cos(βs) − sin(βs) − . (3.10)
2 cos(βs)

The Fourier transforms of the W boson propagator (3.5), (3.8) and of the charged
scalar  boson propagator (3.6), (3.9) are gauge dependent. In an arbitrary ξ-gauge
they have the forms [14]:

∞   
ds 2 tan(βs)
G μν (q) = − exp is q − q⊥
2
cos(βs) βs
0
  
× e−ism W gμν + (ϕϕ)μν (1 − cos(2βs)) − ϕμν sin(2βs)
2

  
− qμ + (ϕq)μ tan(βs) qν + (qϕ)ν tan(βs)
 
β
+i ϕμν − (ϕϕ)μν tan(βs)
2

1
× 2 e−ism W − e−is ξ m W ,
2 2
(3.11)
mW
∞   
ds 2 tan(βs)
D(q) = exp −is ξ m 2W − q2 + q⊥ . (3.12)
cos(βs) βs
0

In the Feynman gauge, when ξ = 1, the Fourier transform of the W boson prop-
agator is essentially simplified [15]:

∞   
ds 2 tan(βs)
G μν (q) = − exp −is m 2W − q2 + q⊥
cos(βs) βs
0
 
× gμν + (ϕϕ)μν (1 − cos(2βs)) − ϕμν sin(2βs) . (3.13)

Finally, the Fourier transform of the charged scalar  boson in the Feynman gauge
has the form
∞   
ds 2 tan(βs)
D(q) = exp −is m 2W − q2 + q⊥ . (3.14)
cos(βs) βs
0
3.1 Propagators of Charged Particles in a Magnetic Field 29

3.1.2 A Note on the Noninvariant Phase

At first glance, the expression for the translational and gauge noninvariant phase
Φ(X 1 , X 2 ) written in the covariant form (3.3), (3.4) is rather cumbersome. Some
authors prefer to fix the gauge by the choice of the 4-potential of an external field as
Aμ (X ) = (0, 0, x B, 0), to write the phase in a more compact form:

eB
Φ(X, X  ) = − (x + x  )(y − y  ). (3.15)
2
However, just the covariant form (3.3), (3.4) of a phase is much more convenient in
the analysis of closed loops containing multiple propagators of charged particles.
In a case of the two-vertex loop, the sum of the phases, arising in the amplitude,
is zero:
Φ(X 1 , X 2 ) + Φ(X 2 , X 1 ) = 0. (3.16)

In the case of three or more vertices in the loop, the total phase of all propaga-
tors is translational- and gauge-invariant. It can be easily shown by presenting the
4-potential of the constant uniform external field in an arbitrary gauge in the form:

1
Aμ (X ) = X ν F νμ + ∂ μ χ(X ), (3.17)
2
where χ(X ) is an arbitrary function. With (3.17), one automatically has ∂ μ Aν −
∂ ν Aμ = F μν . Integrating (3.3) with (3.17) one obtains:
e
Φ(X 1 , X 2 ) = − (X 1 F X 2 ) − e [χ(X 2 ) − χ(X 1 )]. (3.18)
2
It is seen from Eq. (3.18) that the terms with the function χ(X ) totally cancel each
other in the sum of phases inside a closed loop, providing the gauge invariance. It
is easy to check that the sum of phases (3.18) inside a closed loop is translational
invariant also. For example, the total phases of three and four propagators of charged
particles in the loop are the following:
e
Φ(X 1 , X 2 ) + Φ(X 2 , X 3 ) + Φ(X 3 , X 1 ) = − (X 1 − X 2 )μ F μν (X 2 − X 3 )ν , (3.19)
2
e
Φ(X 1 , X 2 ) + Φ(X 2 , X 3 ) + Φ(X 3 , X 4 ) + Φ(X 4 , X 1 ) = − (X 1 − X 3 )μ F μν (X 2 − X 4 )ν .
2
(3.20)

In a general case of the sum of n phases one has:


e
n
 e
n−1 l−1
Φtot =− (X i F X i+1 ) =− (Z k F Z l ), (3.21)
2 X n+1 ≡X 1 2
i=1 l=2 k=1
30 3 Propagators of Charged Particles in External Active Media

where
Z i = X i − X i+1 .

3.1.3 Propagators in the Weak-Field Expansion

Manipulations with the exact expressions (3.10) and (3.13) are extremely cumber-
some. On the other hand, magnetic fields existing in Nature, except the early Universe,
are always weak compared with the critical field for the W boson, m 2W /e 1024 G.
Therefore, the propagators of the W boson and the charged scalar  boson can be
expanded in powers of β = eB as a small parameter. We find up to second order,
using the Feynman gauge:

gμν 2 ϕμν
G μν (q) = −i −β 2
q2− mW2 (q − m 2W )2
  
2 q⊥ 2
1
+ i β gμν
2
+ 2
(q 2 − m 2W )3 (q − m 2W )4

1
+ 4 (ϕϕ)μν 2 + O(β 3 ). (3.22)
(q − m 2W )3

It is not difficult to find the similar expansion for the propagator of the W boson in an
arbitrary ξ-gauge, however the resulting expression appear to be rather cumbersome,
and we do not present it here.
Comparing Eqs. (3.13) and (3.14), one can easily see that the  boson propagator
D(q) differs only in sign from the coefficient at the term gμν in the expansion of the
propagator G μν (q) over the three independent tensor structures. One obtains
 
2
2 q⊥
i 1
D(q) = 2 − i β2 + + O(β 3 ). (3.23)
q − m 2W (q 2 − m 2W )3 (q 2 − m 2W )4

Likewise, the asymptotic expression for the electron propagator S(q) is realised
when the field strength is the smallest dimensional parameter, β m 2e m 2W .
In this “weak field approximation” the charged-lepton propagator can be expanded
as [7]

(qγ) + m e (qγ) + m e
S(q) = i +β (γϕγ)
q − me
2 2 2(q 2 − m 2e )2
 
2i (q2 − m 2e )(qγ)⊥ − q⊥ 2 ((qγ) + m )
 e
+ β2 + O(β 3 ). (3.24)
(q 2 − m 2e )4
3.1 Propagators of Charged Particles in a Magnetic Field 31

One can see from this expansion that the contribution of the region of small virtual
momenta q 2 ∼ m 2e m 2W is enhanced in each succeeding term. If the propagator
is used for a moderate field, m 2e β m 2W , the expansion (3.24) is not applicable
and the exact propagator Eq. (3.10) must be used.

3.1.4 Propagators in an Expansion over Landau Levels

3.1.4.1 Electron Propagator

If a magnetic field is sufficiently large, B  Be = m 2e /e, it is convenient to use the

expression for the electron propagator in an expansion over the Landau levels. We
present here the procedure for obtaining such an expression, following to Ref. [6] (see
also [7]). It should be noted, that there was an error in expression for the propagator
in Ref. [6], namely, the term in the second line of Eq. (4.33) should contain the extra
factor (−i). This error was corrected in Ref. [7], Eqs. (39) and (40), and also in
Ref. [16], Eqs. (13) and (14), but without any comments.
Let us rewrite the Fourier transform of the translationally and gauge invariant part
of the electron propagator (3.10) by introducing a new integration variable v = βs,
to obtain:
∞ 
1
S(q) = dv exp(−iρv) (qγ) + m e f 1 (v)
β
0

(γϕγ)
− (qγ) + m e f 2 (v) − (qγ)⊥ f 3 (v) , (3.25)
2

where the following notations are used:

f 1 (v) = exp(−iα tan v),

f 2 (v) = tan v exp(−iα tan v),
1
f 3 (v) = exp(−iα tan v), (3.26)
cos2 v

and ρ = (m 2e − q2 )/β, α = q⊥ 2 /β. Since the functions f (v) ( j = 1, 2, 3) are

j
periodic, f j (v) = f j (v + nπ), let us divide the integration domain (0, ∞) into
intervals (0, π), (π, 2π), . . . (nπ, (n + 1)π) . . . . Making in each segment the change
of variable, v → v + nπ, we can write:

∞ ∞ π
dv exp(−iρv) f j (v) = exp(−iρnπ) dv exp(−iρv) f j (v)
0 n=0 0
32 3 Propagators of Charged Particles in External Active Media

1
= Aj, (3.27)
1 − exp(−iρπ)

where
π
Aj = dv exp(−iρv) f j (v). (3.28)
0

It suffices to compute the integral A1 , because the other two integrals can then be
found by the formulas:

∂
A2 = i A1 ,
∂α
i ρ
A3 = − 1 − e−iρπ − A1 . (3.29)
α α
The validity of the last relation is easily seen by representing the integral A3 in the
form:
π
i d
A3 = dv exp(−i ρv) exp(−iα tan v) (3.30)
α dv
0

and integrating by parts.

To calculate A1 , we write f 1 (v) in the form
 
−e−2iv + 1
f 1 (v) = exp(−iα tan v) = exp α −2iv . (3.31)
−e −1

The right-hand side of this equation can be expressed through the Laguerre polyno-
mials:
1 dn  
L n (x) = e x n x n e−x . (3.32)
n! dx
The generating function for Laguerre polynomials is determined by:
  ∞
1 xt
exp − = L n (x)t n (3.33)
1−t 1−t
n=0

for |t| < 1, where you can get

  ∞
xt
exp − = L n (x) − L n−1 (x) t n . (3.34)
1−t
n=0

with a completion of L −1 (x) ≡ 0. Denoting −e−2iv = t and 2α = x in the right-hand

side of Eq. (3.31) and using the identity
3.1 Propagators of Charged Particles in a Magnetic Field 33
   
x t +1 xt x
exp = exp − exp − , (3.35)
2t −1 1−t 2

we transform the expression for A1 to the form:

π ∞
A1 = dv e−α L n (2α) − L n−1 (2α) (−1)n exp(−2inv) exp(−iρv)
0 n=0

∞ π
= e−α (−1)n L n (2α) − L n−1 (2α) dv exp[−i(ρ + 2n)v]
n=0 0
∞
(−1)n
= −ie−α 1 − e−iρπ L n (2α) − L n−1 (2α) . (3.36)
ρ + 2n
n=0

Finally, using Eqs. (3.25), (3.26), (3.28), (3.29) and (3.36), we write the Fourier trans-
form of the translationally and gauge invariant part of the electron propagator in the
form:
∞   
i i 
S(q) = (qγ) + m e dn (α) − (γϕγ) dn (α)
n=0
q2 − m 2e − 2nβ 2

dn (α)
− (qγ)⊥ 2n , (3.37)
α

where α = q⊥
2 /β, and the functions are introduced:

dn (α) = (−1)n e−α [L n (2α) − L n−1 (2α)]. (3.38)

3.1.4.2 Propagators of the W and  Bosons

Similarly to the electron propagator, the propagators of the W and  bosons can
also be represented as an expansions over the Landau levels. As it was noted in
the Introduction, magnetic fields could exist in the early Universe of the scale of
the critical field value for a W -boson, BW = m 2W /e 1024 gauss. In this case, a
knowledge of the vector-boson propagator expanded over the Landau levels can be
helpful for investigations of processes in the early Universe.
The Fourier transform of the translationally invariant part of the W boson propa-
gator (3.8) in the ξ-gauge is presented in Eq. (3.11). Similarly to the transformations
of the electron propagator, let us rewrite (3.11) in a more convenient form:
34 3 Propagators of Charged Particles in External Active Media

∞  
1 −iρv
G μν (q) = − dv e (ϕ̃ϕ̃)μν f 4 (v) − (ϕϕ)μν f 5 (v) − ϕμν f 6 (v)
β
0
∞  
1 −i ρv −i ρξ v β
+ dv e −e qμ qν + i ϕμν f 4 (v) (3.39)
β m 2W 2
0
  
β
+ (ϕq)μ qν + qμ (qϕ)ν − i (ϕϕ)μν f 7 (v) + (ϕq)μ (qϕ)ν f 8 (v) ,
2

where the functions are introduced:

1
f 4 (v) = exp(−iα tan v),
cos v
cos(2v)
f 5 (v) = exp(−iα tan v),
cos v
sin(2v)
f 6 (v) = exp(−iα tan v),
cos v
tan v ∂
f 7 (v) = exp(−iα tan v) = i f 4 (v),
cos v ∂α
tan2 v ∂2
f 8 (v) = exp(−iα tan v) = − 2 f 4 (v), (3.40)
cos v ∂α

and ρ = (m 2W − q2 )/β, ρξ = (ξ m 2W − q2 )/β, α = q⊥ 2 /β.

Through the same procedure as in the case of the fermion propagator and noting
that f j (v + π n) = (−1)n f j (v) ( j = 4, 5, 6, 7, 8), we can write

∞
1
dv exp(−iρv) f j (v) = Aj, (3.41)
1 + exp(−iρπ)
0

where the integrals similar to Eq. (3.28) are introduced:

π
Aj = dv exp(−iρv) f j (v) ( j = 4, 5, 6, 7, 8). (3.42)
0

It is worthwhile to introduce the auxiliary integrals:

π
C(α) = dv exp(−iρv) exp(−iα tan v) cos v, (3.43)
0
π
S(α) = dv exp(−iρv) exp(−iα tan v) sin v, (3.44)
0
3.1 Propagators of Charged Particles in a Magnetic Field 35

E(±) (α) = C(α) ± iS(α)

π
= dv exp −i(ρ ∓ 1)v exp(−iα tan v). (3.45)
0

The integral A4 can be represented in a form

π
i d
A4 = dv exp(−iρv) cos v exp(−iα tan v) (3.46)
α dv
0

and further, integrating by parts, we write:

 
i
A4 = −1 − exp(−iρπ) + iρ C(α) + S(α) . (3.47)
α

The integrals A5 and A6 are expressed in terms of A4 , C(α) and S(α):

A5 = 2 C(α) − A4 , (3.48)
A6 = 2 S(α). (3.49)

To find the integrals C(α) and S(α), let us compute E(±) (α) and apply the relations:

1  (+) 
C(α) = E (α) + E (−) (α) , (3.50)
2
1  (+) 
S(α) = E (α) − E (−) (α) . (3.51)
2i

The integral E(±) (α) is computed similarly to the integral A1 for the fermion prop-
agator and is
∞
dn (α)
E(±) (α) = −i 1 + exp(−iρπ) . (3.52)
ρ+2n ∓1
n=0

Here, as before, the functions dn (v) are defined by the expression (3.38). We obtain
the integrals C(α) and S(α) as
∞
i dn (α) + dn−1 (α)
C(α) = − 1 + exp(−iρπ) , (3.53)
2 ρ+2n −1
n=0
∞
1 dn (α) − dn−1 (α)
S(α) = − 1 + exp(−iρπ) . (3.54)
2 ρ+2n −1
n=0
36 3 Propagators of Charged Particles in External Active Media

∞ To obtain the final expressions for A4 , A5 , A6 , the relation should be used

n=0 dn (α) = 1. The result can be written in a more compact form, being expressed
in terms of the functions:
(n + 1) dn+1 (α) + n dn (α)
n (α) = = (−1)n e−α L n (2α). (3.55)
2α
One obtains:
∞
n−1 (α)
A4 = −2i 1 + e−iρπ , (3.56)
ρ+ 2n −1
n=0
∞
n (α) + n−2 (α)
A5 = −i 1 + e−iρπ , (3.57)
ρ+2n −1
n=0
∞
n (α) − n−2 (α)
A6 = − 1 + e−iρπ , (3.58)
ρ+2n −1
n=0
∞
n−1 (α)
A7 = 2 1 + e−iρπ , (3.59)
ρ+2n −1
n=0
∞
n−1 (α)
A8 = 2i 1 + e−iρπ . (3.60)
ρ+2n −1
n=0

Substituting the integrals (3.56)–(3.60) into the expression for the propagator
(3.39), we find:
∞ 
−i
G μν (q) = 2(ϕ̃ϕ̃)μν n−1 (α)
q 2 − m 2W − β(2 n − 1)
n=0 
   
− (ϕϕ)μν n (α) + n−2 (α) + iϕμν n (α) − n−2 (α)
 
ξ−1
+ 2 2qμ qν + iβ ϕμν n−1 (α)
q − ξ m 2W − β(2 n − 1)
 
+ i 2(ϕq)μ qν + 2qμ (qϕ)ν − iβ(ϕϕ)μν n−1 (α)


− 2(ϕq)μ (qϕ)ν n−1 (α) . (3.61)

It is worth noting a singularity that the contribution of the ground level, n = 0,

into the W boson propagator contains, in contrast to the contribution of the ground
Landau level into the electron propagator (3.37). For the W boson, this contribution
has the gauge independent form
3.1 Propagators of Charged Particles in a Magnetic Field 37

−i
G (0) e−q⊥ /β −(ϕ ϕ)μν + iϕμν ,
2
μν (q) = (3.62)
q2 − mW + β
2

that is, it contains a pole at q2 = m 2W − β. Thus, if the magnetic field approaches
the critical value of the field for the W boson, BW = m 2W /e 1024 G, the so-called
instability arises of the perturbation theory for a W boson vacuum (see, e.g., [17]).
The propagator of the  boson in the ξ gauge, D(q), as in the case of a weak
field, is reconstructed from (3.61) in the form
∞
2i n−1 (α)
D(q) = . (3.63)
q2
n=0 
− ξ m 2W − β(2 n − 1)

It should be noted, that the summation over n in Eq. (3.61) formally starts from
n = 0, but in fact it starts from n = 1, because −1 (α) = 0 by definition. This means
that the propagator of the  boson, as one could expect, does not contain a pole at
q2 = ξ m 2W − β.

3.1.5 Electron Propagator in a Strong Magnetic Field

Translationally invariant part of the electron propagator S(X ) has also other repre-
sentations. For example, to analyze the processes in a strong magnetic field, it is
worthwhile to use the asymptotic expression for the propagator. To obtain this, let us
perform the rotation of the contour of integration in the complex plane of the variable
s in the integral (3.2) onto the negative imaginary axis, s = −iτ , and perform a par-
tial decomposition into the Fourier integral over the coordinates t = X 0 and z = X 3
(the magnetic field is directed along the third axis):

∞  
i dτ d 2 q
S(X ) = − [(qγ) + m e ]− (1 + tanh τ )
4π tanh τ (2π)2
0

iβ
+ [(qγ) + m e ]+ (1 − tanh τ ) − (X γ)⊥ (1 − tanh2 τ )
2 tanh τ
 
β X⊥ 2 τ (m e − q )
2 2
× exp − − − i(q X ) , (3.64)
4 tanh τ β

Here, γα are the Dirac matrices in the standard representation, ± are the projection
operators (2.51),
d2 q = dq0 dq3 , [± , (aγ) ] = 0.
38 3 Propagators of Charged Particles in External Active Media

The asymptotic expression for the propagator in a strong magnetic field can be
obtained from Eq. (3.64) by an approximate estimate of the integral over τ in the
limit β/|m 2e − q2 | 1. In this case, the main contribution into the integral over
τ comes from the region τ ∼ β/|m 2e − q2 |. Considering that tanh τ 1 − 2e−2τ
for τ 1, we obtain the following asymptotic expressions for the translationally
invariant part of the electron propagator in a strong magnetic field:
 
iβ β X⊥2
d2 q (qγ) + m e
S(X ) exp − − e−i(q X ) , (3.65)
2π 4 (2π)2 q2 − m 2e

which was first obtained in Refs. [4, 5]. It is easy to see that the expression (3.65)
coincides with the contribution of the ground Landau level. Indeed, substituting the
term with n = 0 from (3.37) into (3.7) and integrating over d2 q⊥ = dq1 dq2 , we
reproduce the formula (3.65).

3.2 Propagators of Charged Particles in a Crossed Field

In the case of a crossed field, the electron propagator in the Fock proper-time for-
malism has the same form of (3.1), where the translational and gauge invariant part
S(X ) can be obtained from (3.2) by the limiting transition when the field invari-
ant β ∼ [−(F F)]1/2 is made to tend to zero in such a way that the field tensor
Fαβ ∼ βϕαβ remains finite. Thus one obtains

∞ 
i ds 1 ie se2
S(X ) = − (X γ) − (X F̃γ)γ5 − (X F Fγ) + m e
16π 2 s 2 2s 2 3
0
   
sm e e X2 se2
− (γ Fγ) exp −i m e s +
2
+ (X F F X ) , (3.66)
2 4s 12

where Fμν and F̃μν are the strength tensor and the dual strength tensor for the external
crossed field.
The Fourier transforms of the translational invariant parts of the propagators has
the form:
∞ 
S(q) = ds e−iΩe (qγ) + ise(q F̃γ)γ5 − s 2 e2 (q F Fγ)
0

1
+ m e − sm e e(γ Fγ) , (3.67)
2
3.2 Propagators of Charged Particles in a Crossed Field 39

∞  
G (W )
μν (q) =− ds e −iΩW
gμν + 2s e (F F)μν − 2seFμν ,
2 2
(3.68)
0
∞
D () (q) = ds e−iΩW , (3.69)
0

where the Feynman gauge is taken for the W and -bosons, and the notation is used
( j = e, W ):
s3
Ω j = s(m 2j − q 2 ) + e2 (q F Fq). (3.70)
3

3.3 Direct Derivation of the Electron Propagator in a Magnetic

Field as the Sum over Landau Levels on a Basis of the Dirac
Equation Exact Solutions

In this section, we explore such a methodologically important issue as a direct deriva-

tion by the standard quantum field theory methods of the exact electron propagator
in the external magnetic field in the form of the sum over Landau levels from the
exact solutions of the Dirac equation in a magnetic field. The presentation is based
on the paper [18].
To calculate the electron propagator, the standard method is applied based on
using the field operators which include the Dirac equation solutions in a magnetic
field:

Ψ̂ (X ) = an, p y , pz , s Ψn,(+)p y , pz , s (X ) + bn,

† (−)
p y , pz , s Ψn, p y , pz , s (X ) . (3.71)
n, p y , pz , s

Here, a is the destruction operator of the electron, b † is the creation operator of the
positron, Ψ (+) and Ψ (−) are the normalized solutions of the Dirac equation (2.2)
in a magnetic field with positive and negative energy correspondingly, presented in
Sect. 2.1.
The propagator is defined as the difference of time-ordered and normal-ordered
productions of the field operators (3.71):

S (e) (X, X  ) = T Ψ̂ (X )Ψ̂ (X  ) − N Ψ̂ (X )Ψ̂ (X  ) . (3.72)

Using anticommutation relations for the creation and destruction operators, we

obtain, that the propagator at t > t  and at t < t  is expressed in terms of the
solutions with positive energy (2.30)–(2.32) and negative energy correspondingly:
40 3 Propagators of Charged Particles in External Active Media


S (e) (X, X  ) =± Ψn,(±)p y , pz , s (X )Ψ n,
(±) 
p y , pz , s (X ). (3.73)
t ≷t 
n, p y , pz , s

Thus, the propagator is divided into the sum over Landau levels:
∞
S (e) (X, X  ) = Sn(e) (X, X  ). (3.74)
n=0

Further we will find the nth Landau level contribution into the propagator (3.73).
It is convenient to come from the summation over the momenta p y and pz to the
integration, by the substitution

1 d p y d pz
→ . (3.75)
L y Lz p y , pz
(2π)2

For the nth level contribution we found:

 
 d p y d pz
Sn(e) (X, X  )  =
t ≷t (2π) 2 E n (±E n + m)
2
 
× exp i ∓E n (t − t  ) ± p y (y − y  ) ± pz (z − z  )
(±) (±) (±)
× Un, p y , pz , s (ξ ) U n, p y , pz , s (ξ (±) ). (3.76)
s=±1

After simple but quite cumbersome transformations one can reduce the matrices in
Eq. (3.76), which are constructed from the bispinors (2.31), (2.32) and the corre-
sponding bispinors of the solution with negative energy, to:

1 (±) (±) (±) (±)

Un, p y , pz , s (ξ ) U n, p y , pz , s (ξ )
±E n + m
s=±1
  
1 β 1 (±) 2 1 (±) 2
= exp − (ξ ) − (ξ ) ±E n γ0 ∓ pz γ 3 + m
2n n! π 2 2
 
× − Hn (ξ (±) ) Hn (ξ (±) ) + + 2n Hn−1 (ξ (±) ) Hn−1 (ξ (±) ) (3.77)
 1 
+ i2n βγ − Hn−1 (ξ (±) ) Hn (ξ (±) ) − + Hn (ξ (±) ) Hn−1 (ξ (±) ) ,

where ± are the projection operators (2.51). One can see, that after changing the
signs of integration variables p y → − p y and pz → − pz in the expression (3.76)
at t < t  , the ± sign at t > t  and t < t  still remains just in the sign at E n . It is
appropriate to use the following relation, where the expression for energy (2.24) is
taken into account:
3.3 Direct Derivation of the Electron Propagator in a Magnetic Field 41

 +∞
f (±E n ) ∓ i E n (t−t  ) 

i d p0 f ( p0 ) e−i p0 (t−t )
e  = , (3.78)
2 En t ≷t 2π −∞ p2 − m 2 − 2βn + i

where p2 = p02 − pz2 .

Using the relation (3.78) we add to the expression (3.76) an integration over the
zero momentum component. As a result the propagator can be written at t > t  and
at t < t  identically. Renaming the variables ξ (+) = ξ, ξ (+) = ξ  , we reduce (3.76)
with taking into account (3.77) and (3.78) to the form
  
i β x2 + x 2 d p0 d p y d p z
Sn(e) (X, 
X)= exp −β
2n n! π 2 (2π)3
 
e−i( p(X −X ))

p 2y  
× 2 exp − − p y x + x − i(y − y )
p − m 2 − 2 βn + i  β

× ( pγ) + m − Hn (ξ) Hn (ξ  ) + + 2n Hn−1 (ξ) Hn−1 (ξ  )
 1 
 
+ i 2n β γ − Hn−1 (ξ) Hn (ξ ) − + Hn (ξ)Hn−1 (ξ ) .
(3.79)

It is worthwhile to note that the expression (3.74) with (3.79) for the electron
propagator in a constant uniform magnetic field as the sum over Landau levels in the
x-space has its own significance. In some cases, this form of the propagator can be
more convenient than other representations.
One can make an integration over p y in the propagator (3.79) by introducing a
new variable √
py β
u=√ + x + x  − i(y − y  ) ,
β 2

and using the well-known integrals being expressed via the Laguerre polynomi-
als [19]:
 ∞ √
e−u Hn (u + a) Hn (u + b) du = 2n n!
2
π L n (−2 a b),
−∞
 ∞
e−u Hn (u + a) Hn−1 (u + b) du
2

−∞
√ 1
= 2n−1 n! π L n (−2 a b) − L n−1 (−2 a b) . (3.80)
b
As a result, the nth Landau level contribution into the electron propagator in a
magnetic field can be presented in the form:

Sn(e) (X, X  ) = ei(X, X ) Sn (X − X  ) , (3.81)
42 3 Propagators of Charged Particles in External Active Media

where (X, X  ) is the translational and gauge non-invariant phase, which is equal
for all Landau levels:
β
(X, X  ) = − (x + x  )(y − y  ).
2
For more details about properties of the phase, see, e. g., Sect. 3.1.2. Sn (Z ) is the
gauge and translational invariant part of the propagator (Z = X − X  ), represented
in the form of the double integral over p :
  2
iβ β 2 d p e−i( p Z )
Sn (Z ) = exp − Z ⊥
2π 4 (2π)2 p2 − m 2 − 2βn + i
     
β 2 β 2
× ( pγ) + m − L n Z ⊥ + + L n−1 Z⊥
2 2
    
(Z γ)⊥ β 2 β 2
+2in Ln Z ⊥ − L n−1 Z . (3.82)
Z⊥2 2 2 ⊥

Let us compare the obtained expression (3.82) with the electron propagator (3.7)
expanded over Landau levels (3.37). To ensure that the expressions for the propagator
are consistent, it is enough to perform in Eqs. (3.7), (3.37) the integration over the
momentum components px , p y , which are transverse to the field. Thus, the nth
Landau level contribution to the propagator is expressed via three different integrals
I1, 2, 3 (Z ⊥ ) in the Euclidean plane ( px , p y ):

d 2 p ie−i( p Z )
Sn (z) =
(2π)2 p2 − m 2 − 2βn + i
   
i
× ( pγ) + m I1 (Z ⊥ ) − (γϕγ)I2 (Z ⊥ ) − 2 nI3 (Z ⊥ ) . (3.83)
2

An integration over the polar angle leads to the Bessel integral:

2π
ei (ξ cos ϕ−n ϕ) dϕ = 2 π in Jn (ξ), (3.84)
0

where Jn (ξ) is the Bessel function. As a result, the integrals I1, 2, 3 (Z ⊥ ) take the
form:
 ∞ 
d 2 p⊥ β √
I1 (Z ⊥ ) = dn (v) ei ( p Z )⊥ = dv J0 β Z⊥ v dn (v),
(2π)2 4π
0
3.3 Direct Derivation of the Electron Propagator in a Magnetic Field 43

 ∞ 
d 2 p⊥  β √
I2 (Z ⊥ ) = dn (v) ei ( p Z )⊥ = dv J0 β Z⊥ v dn (v),
(2π) 2 4π
0

d2 p⊥ dn (v) i ( p Z )⊥
I3 (Z ⊥ ) = e ( p γ)⊥
(2π)2 v
∞ 
β 3/2 (Z γ)⊥ √ dn (v)
=i dv J1 β Z⊥ v √ ,
4π Z⊥ v
0
 
where Z ⊥ = 2 =
Z⊥ (x − x  )2 + (y − y  )2 . Calculating the integrals [19]:
     
β β 2 β 2 β 2
I1 (Z ⊥ ) = exp − Z ⊥ L n Z + L n−1 Z ,
4π 4 2 ⊥ 2 ⊥
     
β β 2 β 2 β 2
I2 (Z ⊥ ) = − exp − Z ⊥ Ln Z ⊥ − L n−1 Z ,
4π 4 2 2 ⊥
     
β (Z γ)⊥ β 2 β 2 β 2
I3 (Z ⊥ ) = − i exp − Z ⊥ Ln Z − L n−1 Z ,
2π Z ⊥ 2 4 2 ⊥ 2 ⊥

and substituting them into (3.83), one finally obtains the expression, which coincides
with (3.82).

References

1. J. Schwinger, Phys. Rev. 82, 664 (1951)

2. V.A. Fock, Phyzik. Z. Sowjetunion 12, 404 (1937)
3. C. Itzykson, J.-B. Zuber, Quantum Field Theory (McGraw-Hill, New York, 1985)
4. V.V. Skobelev, Izv. Vyssh. Uchebn. Zaved. Fiz. No. 10, 142 (1975)
5. Yu.M. Loskutov, V.V. Skobelev, Phys. Lett. A 56, 151 (1976)
6. A. Chodos, K. Everding, D.A. Owen, Phys. Rev. D 42, 2881 (1990)
7. T.-K. Chyi, C.-W. Hwang, W.F. Kao, G.-L. Lin, K.-W. Ng, J.-J. Tseng, Phys. Rev. D 62, 105014
(2000)
8. E. Elizalde, E.J. Ferrer, V. de la Incera, Ann. Phys. 295, 33 (2002)
9. E. Elizalde, E.J. Ferrer, V. de la Incera, Phys. Rev. D 70, 043012 (2004)
10. A.V. Kuznetsov, N.V. Mikheev, G.G. Raffelt, L.A. Vassilevskaya, Phys. Rev. D 73, 023001
(2006)
11. A. Erdas, M. Lissia, Phys. Rev. D 67, 033001 (2003)
12. K. Bhattacharya, S. Sahu, Eur. Phys. J. C 62, 481 (2009)
13. A.V. Kuznetsov, N.V. Mikheev, A.V. Serghienko, Phys. Lett. B 690, 386 (2010)
14. A. Erdas, C. Isola, Phys. Lett. B 494, 262 (2000)
15. A. Erdas, G. Feldman, Nucl. Phys. B 343, 597 (1990)
16. V.P. Gusynin, A.V. Smilga, Phys. Lett. B 450, 267 (1999)
17. V.V. Skalozub, Yad. Fiz. 45, 1708 (1987). [Sov. J. Nucl. Phys. 45, 1058 (1987)]
18. A.V. Kuznetsov, A.A. Okrugin, Int. J. Mod. Phys. A 26, 2725 (2011)
19. A.P. Prudnikov, Yu.A. Brychkov, O.I. Marichev, Integrals and Series: Special Fuctions
(Gordon-Breach, New York, 1990)
Chapter 4
Particle Dispersion in External Active
Media

This chapter is devoted to an analysis of the dispersion properties of photons and

neutrinos in external active media: magnetic field, plasma, and magnetized plasma.
Possible astrophysical manifestations of particle processes influenced by external
active media are also considered.

4.1 Dispersion in Media: Main Definitions

Dispersion effects in the medium significantly affect the propagation of particles with
small masses (photons, neutrinos), while other particles remain almost insensitive to
the influence of the environment (e.g., axions and other Nambu–Goldstone bosons).
The direct way to investigate the dispersion relations of photons and neutrinos is to
analyze the link between forward scattering and refractive index.
In accordance with the general concepts of quantum field theory, particles are
quantized excitations of the corresponding fields: the electromagnetic field produces
photons, the electron-positron field produces the electrons, and so on. Usually, it
is convenient to describe these fields by means of plane waves, characterized by a
frequency ω and wavevector k. Then the excitations of these modes have the time and
spatial dependence, which is described by a factor exp [−i (ωt − kx)]. With a wave
vector given, the frequency is determined by the dispersion relation. Since (ω, k) is a
4-vector, basing on Lorentz invariance we find that in vacuum the value ω 2 −k2 = m2
is the same for all frequencies and m is the particle mass. One consequence of the
covariant dispersion relation is that the decay of the form 1 → 2 + 3 is possible
only if m1 > m2 + m3 , so that the particle 1 in its rest frame had enough energy for
production of the final state.
In a medium, dispersion relations are changed, as a rule, by the coherent interaction
with the background. In the simplest case, a particle acquires an effective mass
caused by the presence of a medium. For example, dispersion relation for photons
in a nonrelativistic plasma is of the form ω 2 = ωP2 + k2 , where ωP is the so-called
plasma frequency, defined by the expression

A. Kuznetsov and N. Mikheev, Electroweak Processes in External Active Media, 45

Springer Tracts in Modern Physics 252, DOI: 10.1007/978-3-642-36226-2_4,
© Springer-Verlag Berlin Heidelberg 2013
46 4 Particle Dispersion in External Active Media

4π α Ne
ωP2 = , (4.1)
me

where Ne is the electron density. For a process in an environment which induces

effective masses of particles, the kinematic condition for the process realization
should be considered more carefully. For example, the kinematic condition for the
decay 1 → 2 + 3, instead of the simplified vacuum relation m1 > m2 + m3 , is
expressed in its original form through the squares of masses:
   2
m14 − 2 m12 m22 + m32 + m22 − m32 > 0 . (4.2)

In this form, the kinematic condition for the possibility of the process is applicable
for the case of the negative effective mass squared. The appearance of the photon
effective mass in a medium leads to the fact that if ωP2 > 4 mν2 , the decay γ → ν ν̄
becomes kinematically allowed, which can occur in stars. In fact, this so-called
plasma process is the main mechanism of the neutrino emission in a wide range of
temperatures and densities, including, for example, the physical conditions inside
the white dwarfs and red giants.
Note that the dispersion relation can be such that the 4-momentum Pμ = (E, p) be
a space-like, P2 = E 2 − p2 < 0. This means an appearance of the negative effective
mass squared, P2 = meff 2 < 0. No physical problems with such a “tachyon” would

arise, because the speed of propagation is determined by the group velocity, which
is always less than the speed of light. The dispersion relation in a homogeneous
medium is often written in terms of the refractive index n as k = |k| = n ω. Space-
like excitations correspond to the condition n > 1; an example of such kind is a
photon in water or in air. In this case, the well-known process of the Cherenkov
radiation which can be treated as the “decay” e → eγ, is kinematically allowed
for a sufficiently fast moving electrons. Similarly, the neutrino Cherenkov process
ν → νγ is possible for a massless neutrino in an external magnetic field, where the
photon 4-momentum can be space-like.
Neutrinos can participate in non-standard electromagnetic processes, for example,
due to the intrinsic magnetic moments. This can lead to plasma processes of the
creation of sterile neutrinos, and thus, to the cooling of stars. Limits on an anomalous
cooling rate derived from the observations of white dwarfs and stars of the globular
star clusters, have allowed to establish the most stringent limits on the electromagnetic
interactions of neutrinos.
In the standard model, all fermions are initially massless. They acquire effective
masses due to interaction with the Higgs scalar field. Its vacuum expectation value
Φ0 is the main factor determining the values of the masses. Therefore, even the
vacuum masses can be interpreted as a phenomenon of refraction. Since the scalar
Φ0 is a Lorentz-invariant, the dispersion relation is thus derived from the standard
formula E 2 − p2 = m2 . In general, the active medium changes this relation, and the
dependence E(p) is usually more complicated function than (m2 + p2 )1/2 .
4.1 Dispersion in Media: Main Definitions 47

The dispersion relation may also depend on the polarization of the radiation.
In the optically active medium, the left- and right-polarized photons have different
refractive indices. In this sense, the entire medium is optically active for neutrinos,
since only left-handed neutrinos are involved in interactions, and the right-handed
neutrinos are sterile.
The interaction of the muon and tau neutrinos, νμ and ντ , with an ordinary
astrophysical environments, that is, containing no thermal muons and tau leptons
is different from the interaction of the electron neutrinos νe due to the contribution
of the charged current (νe e− ) to the scattering amplitude. Therefore, this environ-
ment is a birefringent medium with respect to the flavor of neutrino, in the sense that
the environment induces a variety of dispersion relations for neutrinos of different
flavors. The importance of this effect for neutrino oscillations, which are actually
determined by the relation of phases in the propagation of neutrinos of different
flavors, is extremely high.
Considering different quantum processes in active media, one should take into
account that all the particles have non-trivial dispersion properties, while it depends
on the circumstances, whether an effect of refraction is significant or not. For example,
the statement appeared in the literature that in a sufficiently dense plasma, where
ωP > 2me , photons decay with a pair creation, γ → e+ e− . However, this is not
true, because the effective masses induced by plasma, which the charged leptons
also acquire, are so large that such decays do not occur [1], see also the discussion
below in Sect. 4.5.3.
In addition to the modification of the particle dispersion relations, the presence
of medium can lead to an appearance of entirely new excitations. The well-known
example is the longitudinally polarized state of the electromagnetic field that exists
in plasma in addition to the normal state with transverse polarization. These objects,
usually called the longitudinal plasmons, were first discussed in 1926 by Langmuir. In
many cases these quantized collective excitations play a role similar to that of ordinary
particles. For example, both the usual states with the transverse polarization, called
the transverse plasmons or simply photons, and longitudinal plasmons can decay
into neutrino pairs and thus contribute to the plasma neutrino emission processes.
While the dispersion relations and particle interactions in a plasma are formally
best described in terms of field theory at finite temperatures and densities, most of the
important results of elementary particle physics in stars have been obtained before
the development of this formalism by using simpler tools of kinetic theory. Indeed,
for many problems in describing the dispersion properties of particles and collective
effects, the kinetic approach often seems more physically transparent, leading to
identical results. Further discussion is entirely based on the kinetic theory.
To obtain the dispersion relation in a plasma for a given particle with known
properties, it is usually sufficient to use the simplest approximation, calculating
the forward scattering amplitude off the corresponding field excitations, being the
components of the plasma.
Along with the hot dense plasma, another component of active astrophysical
environment, a strong magnetic field could have a significant influence on the dis-
persion properties of particles. However, this effect of the field is significant only in a
48 4 Particle Dispersion in External Active Media

case of the sufficiently high field intensity. There exists a natural scale of the magnetic
field, the so-called critical value Be = me2 /e  4.41×1013 gauss. A detailed analysis
of the magnetic field influence on the photon and neutrino dispersion properties is
presented below in Sects. 4.2 and 4.6.

4.2 Photon Polarization Operator in an External

Magnetic Field

The dispersion properties of photons in a magnetic field are determined by the

polarization operator, which can be obtained from the amplitude of the photon to
photon transition, Mγ→γ :

Mγ→γ = −ε∗α Παβ εβ , (4.3)

described by the Feynman diagram shown in Fig. 4.1. In this case, the dominant role
is played by the electron as a particle with a maximal specific charge, e/me , which
is the most sensitive to the influence of an external field. The photon polarization
operator in an external field was studied in a number of papers, see, e.g., [2–6]. It is
convenient to represent the polarization operator in the form:

3 b(λ) b(λ)
 α β
Παβ = Π (λ) (q), (4.4)
(b(λ) )2
λ=1

(λ)
where Π (λ) are the eigenvalues of the polarization operator, bα are the eigenvectors
of the orthogonal basis:

bα(1) = (qϕ)α , bα(2) = (qϕ̃)α ,

bα(3) = q2 (qϕϕ)α − qα (qϕϕq), bα(4) = qα . (4.5)

The functions Π (λ) (q) obtained in Ref. [6] can be written as

Fig. 4.1 Photon polarization operator in a strong magnetic field: the double line in the loop
corresponds to the exact propagator of a charged fermion in a magnetic field
4.2 Photon Polarization Operator in an External Magnetic Field 49

1 ∞  
(λ) α dt βt (λ) −iΩ 1 − u2 −iΩ0
Π (q) = − du e − q2 e + Π (vac) (q2 ),
π t sin βt 2
0 0

(1)
q2 u sin βtu 2 cos βtu − cos βt
= cos βtu − − q⊥ ,
2 tan βt sin2 βt

(2) 1 − u2 q2 u sin βtu
= q2 cos βt − ⊥ cos βtu − ,
2 2 tan βt

(3) q2 u sin βtu
= cos βtu − , (4.6)
2 tan βt

where
2 
q⊥ cos βtu − cos βt 1 − u2
Ω = Ω0 + − t ,
2 β sin βt 2

1 − u2
Ω0 = t me2 − q2 .
4

In Eq. (4.6), the subtraction is made of the vacuum polarization operator, resulting in
a convergence of the integral over t, and then the renormalized vacuum polarization
operator was added. The function Π (vac) (q2 ) describes the vacuum polarization in
the absence of a field and has the form, see, e.g., [7]:
α 2
Π (vac) (q2 ) = q v(q2 ) , (4.7)
2π
1 
q2
v(q ) = du (1 − u ) ln 1 −
2 2
(1 − u2 ) . (4.8)
4me2
0

The dispersion equations for a real photon in a magnetic field has the form:

q2 − Π (λ) (q) = 0 (λ = 1, 2, 3) . (4.9)

An analysis of Eq. (4.9) shows that only two transverse polarizations, λ = 1, 2, are
physical, while the third photon polarization, λ = 3, is unphysical. Indeed, substi-
tuting the expression for Π (3) (q) into Eq. (4.9), we see that it has a unique solution
q2 = 0. As it follows from (4.5), in this case the basis vector bα(3) is proportional to
the photon four-momentum qα , i.e., the corresponding operator of the electromag-
netic field is proportional to the total divergence and can be removed by a gauge
transformation.
The polarization vectors of photons with the certain dispersion laws are
proportional to the eigenvectors bα(1,2) :
50 4 Particle Dispersion in External Active Media

(qϕ)α (qϕ̃)α
ε(1)
α = Z1 , ε(2)
α = Z2 . (4.10)
2
q⊥ q2

√
The factors Zλ are caused by renormalization of the photon wave function

∂Π (λ)
Zλ−1 = 1 − . (4.11)
∂q2

These renormalizations are especially significant near the values of q2 corresponding
to the so-called cyclotron resonances:
 2
q2 = me2 + 2neB + me2 + 2n eB , (4.12)

where the functions Π (λ) (q) have the square-root singularity.

There exists some discordance of terms for these polarization vectors (4.10). In
the classical paper by S. Adler [8] they were called as “longitudinal”  and the
()
“transversal” ⊥ photon modes, ε(1) (2) (⊥)
α = εα , εα = εα . These notations were based
on the position of the magnetic field vector of the photon electromagnetic wave with
respect to the plane formed by the vectors of external magnetic field, B, and of the
photon momentum, q. Later on, some authors decided that it was more natural to
consider the position of the electric field vector of the photon wave with respect to that
plane, and they used the opposite notations; see e.g. [9], and [10]. As a result, some
authors–see e.g. [11]–confused these notations, using the ones of [9] while referring
to [8]. Sometimes attempts were also made to introduce another notations for these
two photon polarizations, B and C, I and II–see e.g. [12]–or σ and π polarizations (to
the gauge transformation); see e.g. [13]. In our previous book [14] we used the terms
“ordinary” and “extraordinary” for the photon 1 and 2 polarizations in a magnetic
(1) (O) (2) (E)
field (4.10): εα = εα and εα = εα . Introducing such notations, we based on
the properties of these modes with respect to the CP transformation. Here, we use
(1,2)
the notation εα (see (4.10)).
In the limit of strong fields, in the kinematic region q2 eB, the expressions for
the functions Π (λ) (q) are simplified and can be written as

(1) α 2 α 2 B 3
Π (q) = − q⊥ + q ln − C − γE + v(q2 )
3π 3π Be 2

1
+O , (4.13)
eB

(2) 2α q2 α 2 B 3
Π (q) = − eB H + q ln − C − γE + v(q2 )
π 4me2 3π Be 2
4.2 Photon Polarization Operator in an External Magnetic Field 51

1
+O , (4.14)
eB
 
α 2 B 3 1
Π (3) (q2 ) = q ln − C − γE + v(q2 ) + O , (4.15)
3π Be 2 eB

where γE = 0.577 . . . is the Euler constant, C  1.2147 is the numerical value of

the integral
∞ 
1 dz 1 + 3z 3 3
C= + − . (4.16)
2 z 1+z z tanh z tanh2 z
0

The function H(z) introduced in Eq. (4.14) is

1
du
H(z) = − 1. (4.17)
1 − z(1 − u2 ) − i0
0

In different areas of the argument the function takes the form:

√ √
1 1 − z + −z
H(z) = √ ln √ √ − 1, z < 0,
2 −z(1 − z) 1 − z − −z

1 z
H(z) = √ arctan − 1, 0 < z < 1, (4.18)
z(1 − z) 1−z
√ √
1 z+ z−1 iπ
H(z) = − √ ln √ √ −1+ √ , z > 1.
2 z(z − 1) z− z−1 2 z(z − 1)

The function has the asymptotics:

2 8 2 16 3
H(z)  z+ z + z , |z| 1, (4.19)
3 15 35
1 iπ
H(z)  −1 − ln 4|z| + Θ(z) , |z| 1, (4.20)
2z 2z

where Θ(z) is the step function.

It should be noted that in real calculations, the terms with q2 contained in
Eqs. (4.13) and (4.14) are inessential, because they determine the corrections of
the α order, in accordance with the dispersion Eq. (4.9).
The solutions of the dispersion Eq. (4.9) for photons of the 1st and 2nd modes
defined by Eqs. (4.13) and (4.14) are shown in Fig. 4.2. The dotted line corresponds to
the vacuum dispersion q2 = 0. In the region above this line, the square of the “photon
mass” Re Π (2) has the positive sign, while below the line the sign is negative. The
vertical distance from the given point of the dispersion curve to the line q2 = 0 is
52 4 Particle Dispersion in External Active Media

•
•
•
•
•
•
•
•
•
•
2
q •
•
•
•
mode 1
•

mode 2 •
•
•
•
•
•
•
•
•
•
2
4m •
•
•
mode 2
•
•
•
•
•
•
•

0 •

2
q⊥

Fig. 4.2 The dispersion in a strong magnetic field of the first and second photon modes; the
dispersion curve for the mode 2 photon above the line q2 = 4me2 is a real part of the function Π (2)
(see Eq. (4.14)); the dotted line corresponds to the vacuum dispersion at q2 = 0

|q2 |. The line q2 = 0 and the horizontal line q2 = 4me2 divide the plane into regions
corresponding to the physical processes with essentially different kinematics.
The solution of Eq. (4.9) for a photon of the 1st mode, as seen from the expression
for the function (4.13), in the considered kinematic area is a straight line, slightly
deviating from the vacuum line q2 = 0 into the region of negative q2 .

4.3 Generalized Two-Point Loop Amplitude j → f f̄ → j

in an External Electromagnetic Field

The result obtained for the photon polarization operator in external magnetic field,
can be easily generalized by performing the one-loop calculation of the two-point
amplitude of the transition j → f f¯ → j in a constant uniform magnetic field for
various combinations of scalar, pseudoscalar, vector and pseudovector currents j and
j interacting with charged fermions. By the currents j and j , we mean generalized
4.3 Generalized Two-Point Loop Amplitude 53

local quantum-field objects that can be currents, as such, or the wave functions of the
corresponding particles. In this section, we present the basic points of such calculation
in a magnetic field and in a crossed field, and give the results in the cases of the vector
and pseudovector currents j and j . As a charged fermion, we consider an electron
as a particle with a maximal specific charge, e/me , which is the most sensitive to
the influence of an external field. Both a more detailed calculation procedure and the
results for the other combinations of currents are presented in Refs. [14, 15].
The field-induced one-loop contributions to the amplitude for the transition j →
f f¯ → j , presented here, can be used in the investigations of both tree-level and
loop-level quantum processes in external electromagnetic fields. The field effects
are taken into account exactly, because exact solutions of the Dirac equation are
used. Owing to this, the expression obtained here for the amplitude is quite general.
The amplitude ΔMVV defines, for example, the field-induced part of the photon
polarization operator. Upon the substitutions

GF GF
jV α → √ CV jα(ν) , jAα → √ CA jα(ν) , jV α → eεα , (4.21)
2 2

the sum of ΔMVV and ΔMVA describes the process amplitude for the radiative
transition of massless neutrino ν → νγ. In (4.21), CV and CA are, respectively, the
vector and axial-vector coupling constants in the effective Lagrangian for neutrino
interaction with electrons in the Standard Model; jα(ν) is the neutrino current; and
εα is the photon polarization vector. Similarly, combining the amplitudes ΔMVV ,
ΔMAA and ΔMVA where the neutrino currents (4.21) are substituted, one can also
analyze the process ν ν̄ → e− e+ by using the imaginary parts of the amplitudes.

4.3.1 Magnetic Field

The generalized amplitude of the transition j → f f¯ → j will be analyzed by using

the effective Lagrangian for the interaction of the current j with electrons in the form
  
L(X) = jn (X) Ψ̂ (X)Γn Ψ̂ (X) , (4.22)
n

where Ψ̂ (X) is the field operator (2.1), the generic index n = S, P, V , A numbers the
matrices
Γn = 1, γ5 , γα , γ5 γα , (4.23)

while jn (X) is the generalized current including the coupling constant.

The one-loop amplitude for the transition j → j is described by the Feynman
diagram in the Fig. 4.3, and has the form
54 4 Particle Dispersion in External Active Media

Fig. 4.3 Feynman diagram for the transition j → j . Double lines indicate that the effect of an
external electromagnetic field is taken exactly into account in the propagators of virtual fermions


Mnn = − i jn jn d4 Z Tr [S(−Z)Γn S(Z)Γn ] e−iqZ . (4.24)

Here, S(Z) is the translational invariant part of the fermion propagator in a magnetic
field (3.2), Z = X − X , jn and jn are the Fourier transforms of the corresponding
currents, while q is the momentum transfer. From expression (3.2) for the propagator,
it can be seen that the amplitude in (4.24) diverges at the lower limit of integration
with respect to the proper time. This divergence, an ultraviolet one, as a matter of fact,
is due the use of a local limit in the Lagrangian (4.22). Below, only the field-induced
part of the amplitude will be analyzed,


ΔMnn = Mnn − Mnn  . (4.25)
B=0

As can be deduced from the corresponding analysis, the difference in (4.25) is free
from ultraviolet divergences.
Given the bilinear dependence of the phase of the translational invariant part
S(Z) (3.2) of the fermion propagator on the Z variable, the integration with respect
to Z in the expression for the amplitude (4.24) is reduced to the calculation of the
generalized Gaussian integrals of the scalar, vector, and tensor types. The scalar
integral has the form
   
1
Φ= d Z exp −i (Zp) + (ZGZ) ,
4
(4.26)
4

where
v + s μν sin(β(v + s))
Gμν = Λ̃ − β Λμν .
vs sin(βv) sin(βs)

Here, β = eB, the variables v and s are the Fock proper-times in electron propagators.
The matrices Λμν and Λ̃μν are defined in (1.2). The vector and tensor integrals can
be defined from the scalar one by taking the derivatives of Φ with respect to the
momentum p:
4.3 Generalized Two-Point Loop Amplitude 55
   
1 ∂Φ
Φμ = d4 Z Zμ exp −i (Zp) + (ZGZ) = i , (4.27)
4 ∂pμ
   
1 ∂2Φ
Φμν = d Z Zμ Zν exp −i (Zp) + (ZGZ) = −
4
. (4.28)
4 ∂pμ ∂pν

Performing integrations over the spacetime variable Z one obtains

 
Φ = −(4π)2 (det G)−1/2 exp i(pG−1 p) ,
Φμ = −2(pG−1 )μ Φ , (4.29)
 
Φμν = 2 2(pG−1 )μ (pG−1 )ν − iG−1
μν Φ ,

where the inverse matrix G−1 is

vs sin(βv) sin(βs)
G−1
μν = Λ̃μν − Λμν ,
v+s β sin(β(v + s))

and the determinant of the G matrix is

 2
(v + s) β sin(β(v + s))
det G = − . (4.30)
sv sin(βv) sin(βs)

After performing integrations over Z, the generalized amplitude can be expressed in

the form of a double integral.
Here, we present a complete set of expressions for the amplitudes ΔMnn in the
magnetic field (n, n = V , A).
If one of the currents is a vector one (jn ≡ jV α , Γn ≡ γα ), it can be shown
by a direct calculation that this currents appears in the amplitude only through the
combination fαβ = qα jV β − qβ jV α . If, in addition, the current jV appears to be the
photon polarization vector, the tensor fαβ has the meaning of the strength tensor of
the photon electromagnetic field. This corresponds to the gauge invariance of the
amplitude for the processes being considered.
Thus, the vector–vector amplitude (n, n = V ) is described in terms of the tensors
fαβ and fαβ ; that is,

1 (f ϕ)(f ∗ ϕ) (1) (f ϕ̃)(f ∗ ϕ̃) (2)
ΔMVV = 2
YVV + YVV
4π 2 4q⊥ 4q2

(qϕϕfq)(qϕϕf ∗ q) (3)
+ YVV , (4.31)
q2 q2 q⊥
2
56 4 Particle Dispersion in External Active Media

where

1 ∞  
(i) dt βt (i) −iΩ 21−u
2
−iΩ0
YVV = du y e −q e ,
t sin βt VV 2
0 0

(1) q2 u sin βtu 2 cos βtu − cos βt
yVV = cos βtu − − q⊥ ,
2 tan βt sin2 βt
2 
21−u u sin βtu
2 q⊥
(2)
yVV = q cos βt − cos βtu − ,
2 2 tan βt

(3) q2 u sin βtu
yVV = cos βtu − ,
2 tan βt
fαβ = qα jV β − qβ jV α , fαβ = qα jV β − qβ jV α .

In the above expressions, the as-yet-undefined quantities are given by


1 − u2
Ω0 = t me2 − q2 ,
4
2 
q⊥ cos βtu − cos βt 1 − u2
Ω = Ω0 + − t ,
2 β sin βt 2
2
q⊥ = (qϕϕq) = qμ ϕμν ϕνρ qρ ,
q2 = (qϕ̃ϕ̃q), q2 − q⊥
2
= q2 .

The amplitude for transitions between axial-vector currents (n, n = A) has the
form

1 (jA ϕq)(jA∗ ϕq) (1) (jA ϕ̃q)(jA∗ ϕ̃q) (2)
ΔMAA = 2
YAA + YAA
4π 2 q⊥ q2
q2 (3)
+ 2 2
(jA ϕϕq)(jA∗ ϕϕq)YAA (4.32)
q q⊥

(jA ϕϕq)(jA∗ q) + (jA∗ ϕϕq)(jA q) (4) (jA q)(jA∗ q) (5)
− YAA + YAA ,
q2 q2

where

1 ∞   
(i) dt βt (i) −iΩ 1 − u2
YAA = du yAA e + 2me2 − q2 e−iΩ0 ,
t sin βt 2
0 0
i = 1, 2, 3, 4,
4.3 Generalized Two-Point Loop Amplitude 57

1 ∞   
(5) dt βt (5) −iΩ 2 1−u
2
YAA = du yAA e − 2me2 + q⊥ e−iΩ0 .
t sin βt 2
0 0

In the above expressions, the following notations are used:


(1) q2 u sin βtu 2 cos βtu − cos βt
yAA = cos βtu − − q⊥ − 2me2 cos βtu,
2 tan βt sin2 βt
2 
21−u u sin βtu
2 q⊥
(2)
yAA = q cos βt − cos βtu − − 2me2 cos βt,
2 2 tan βt

(3) q2 u sin βtu 2m2  2 
yAA = cos βtu − + 2e q⊥ cos βt − q2 cos βtu ,
2 tan βt q

(4) q2 u sin βtu
yAA = cos βtu − − 2me2 cos βt,
2 tan βt
2 
(5) q⊥ u sin βtu
yAA = cos βtu − + 2me2 cos βt.
2 tan βt

In the case of the vector and axial-vector vertices (Γn ≡ γα , Γn ≡ γ5 γβ ), the

field-induced part of the amplitude is given by

1 (f ϕ̃)(jA∗ ϕϕq) (1) (jA∗ ϕ̃q)(qf ϕϕq) (2)
ΔMVA = − β YVA + YVA
4π 2 2q2 q2 q⊥
2

(f ϕ̃)(jA∗ q) (3)
+ Y VA , (4.33)
2q2

where

1 ∞ 2 + q2
q⊥ q2 1 − u2 q2
(1) 
YVA =i du dt 2
me2 − q2 2
e−iΩ − 2
,
q⊥ q⊥ 4 q⊥
0 0
1 ∞ 
(2) 1 − u2 q2
YVA =i du dt me2 − q2 e−iΩ − ,
4 q2
0 0
1 ∞ 
(3) 1 − u2
YVA =i du dt me2 + q2 e−iΩ .
4
0 0

It should be emphasized that, in using our results to calculate the amplitudes of

processes featuring axial-vector currents, care should be taken in dealing with terms
linear in an external field in diagrams of the type shown in Fig. 4.3. The point is
that such terms may prove incorrect because of the Adler triangle anomaly. Strictly
58 4 Particle Dispersion in External Active Media

speaking, it is therefore necessary to specify a procedure for subtracting terms linear

in the field, which must then be recovered. Thus, the correct expression for the
field-induced part of the amplitude must have the form
  
 ∂M 
ΔM = M − M
 −B (1) ,
+M (4.34)
B=0 ∂B B=0

where the expression in parentheses is free from the Adler anomaly. A scheme for
recovering the correct form of the term M (1) linear in the field is determined by
a specific type of process and by the origin of the triangle anomaly. An example
of recovering the linear term for the vector—axial-vector part of the amplitude of
the neutrino Cherenkov process in a strong magnetic field, ν → ν + γ [10, 16],
is presented below in Sect. 7.1.1. In this case, the origin of the triangle anomaly is
connected with the transition to the local limit of weak interaction.

4.3.2 Crossed Field

The amplitude for the transition j→j in a crossed field can be derived by performing
once again the calculations outlined in the previous section, but the fermion propa-
gator in a crossed field (3.66) should be used now.
The field-induced parts of the amplitudes ΔMnn (n, n = V , A) can be written
as follows.
The vector—vector amplitude is:

1 (fF)(f ∗ F) (1) (f F̃)(f ∗ F̃) (2)
ΔMVV = Y + Y
4π 2 4(qFFq) VV 4(qFFq) VV

(qFFfq)(qFFf ∗ q) (3)
+ YVV , (4.35)
q2 (qFFq)2

where

1   1/3 
(1) 1 4 df (x) 21−u
2
YVV =− du me2 χ2/3
q (3 + u )
2
−q f1 (x) ,
6 1 − u2 dx 2
0
1   1/3 
(2) 1 2 2/3 4 df (x) 1 − u2
YVV =− du m χ (3 − u2 ) − q2 f1 (x) ,
3 e q 1 − u2 dx 2
0
1
(3) q2
YVV = du(1 − u2 )f1 (x).
2
0
4.3 Generalized Two-Point Loop Amplitude 59

Here, the notations are used:

e2 (qFFq)
χ2q = ,
me6
 2/3 
4 q2
x= 1− (1 − u2 ) ,
χq (1 − u )
2 4me2
∞
t3
f (x) = i dt e−i(tx+ 3 ) , (4.36)
0
∞ 
dt t3
f1 (x) = e−i(tx+ 3 ) − e−itx
t
0
x
1 2 iπ
=− f (z)dz + ln x + ln 3 + γE + , (4.37)
3 3 3
0

f (x) being the Hardy—Stokes function, γE = 0.577 . . . being the Euler constant.
The axial vector—axial vector amplitude is:

1 (jA Fq)(jA∗ Fq) (1) (jA F̃q)(jA∗ F̃q) (2)
ΔMAA = YAA + YAA
4π 2 (qFFq) (qFFq)
(jA FFjA∗ ) (3)
+ q2 Y (4.38)
(qFFq) AA

(jA FFq)(jA∗ q) + (jA∗ FFq)(jA q) (4) (5)
− YAA + (jA q)(jA∗ q)YAA ,
(qFFq)

where

1
(1,2) (1,2)
YAA = YVV − 2me2 duf1 (x),
0
1   1/3
(3) m2 4 df (x)
YAA =− du 4 2e me2 χ2/3
q
q 1 − u2 dx
0
 
21−u
2
+ 2me − q
2
f1 (x) ,
2
1 
(4) 1 − u2
YAA = − du 2me2 − q2 f1 (x),
2
0
60 4 Particle Dispersion in External Active Media

1
(5) 1 − u2
YAA = du f1 (x).
2
0

The vector—axial vector amplitude is:


e (f F̃)(jA∗ FFq) (1) (jA∗ F̃q)(qfFFq) (2)
ΔMVA = − YVA + YVA
4π 2 2(qFFq) q2 (qFFq)

(f F̃)(jA∗ q) (3)
+ YVA , (4.39)
2q2

where

1  2/3 
(1) 1 4 1 − u2
YVA =− du 2me2 − q2 f (x) + 1,
me2 χq
2/3 1 − u2 4
0
(2) (3)
YVA = −YVA + 1,
1  1/3
(3) 1 1 − u2
YVA =− 2/3
q 2
du f (x).
me2 χq 4
0

It should be noted that, in general, the expression for the amplitude ΔMVA in-
volves indefinite forms associated with the Adler anomaly. The procedure for re-
moving them is described above in (4.34).
The expressions obtained for the amplitudes in a crossed field can be used to test
the correctness of a more cumbersome calculation in the presence of a magnetic
field. If, in the amplitudes calculated in the previous section, the field invariant
β ∼ [−(FF)]1/2 is made to tend to zero in such a way that the field tensor eFαβ =
βϕαβ remains finite, the required amplitudes in a crossed field can be obtained from
the resulting expressions.

4.4 Photon Polarization Operator in Plasma

In describing the electromagnetic processes with virtual photons in plasma, the

principal point is to use the photon propagator Gαβ (Q) with the plasma polariza-
tion effects taken into account. We use the straightforward way of taking account of
these effects by summation of the Feynman diagrams of the forward photon scatter-
ing off plasma particles. Similarly to the vacuum case, this summation leads to the
Dyson equation which provides a correct result for the photon propagator in plasma
in the region where the photon polarization operator is real, in the form:
4.4 Photon Polarization Operator in Plasma 61

(t) ( )
i ραβ i ραβ
Gαβ (Q) = + , (4.40)
Q 2 − Πt Q2 − Π

where Πt, are the eigenvalues of the photon polarization tensor Παβ for the trans-
verse and longitudinal plasmon,

(t) ( )
Παβ = −Πt ραβ − Π ραβ , (4.41)

and ρ(t, )
αβ are the corresponding density matrices

(t) Qα Qβ Lα Lβ
ραβ = − gαβ − 2
− , (4.42)
Q L2
( ) Lα Lβ
ραβ = − 2 , (4.43)
L
Lα = Qα (u Q) − uα Q2 , (4.44)

(λ)
uα is the four-vector of the plasma velocity. The density matrices ραβ with λ = t,
have properties of the projection operators:

μ(λ ) (λ)
ρ(λ)
αμ ρβ = −δλλ ραβ . (4.45)

In the region where the eigenvalues Πt, of the photon polarization tensor develop
imaginary parts, they can be written as:

Πλ = Rλ + i Iλ , (4.46)

where Rλ and Iλ are the real and imaginary parts, containing the contributions of
all components of the active medium. For extracting the imaginary parts It, , it will
suffice to make an analytical extension q0 → q0 + i  corresponding to the retarded
polarization operator.
The eigenvalues Πt, of the photon polarization tensor are presented below both
in the general form and in some particular cases.
The expressions for the contributions of a charged fermion into the polarization
functions Πt, in the hard thermal loop approximation can be found e.g. in [17] and
have the form
∞
4α dPP 2  
Πt = fF (E) + f¯F (E)
π E
0

q02 q2 − q2 q0 q0 + vq
× − 0 2 ln , (4.47)
q 2 q 2vq q0 − vq
62 4 Particle Dispersion in External Active Media

∞
4α q02 − q2 dP P 2  
Π = fF (E) + f¯F (E)
π q2 E
0

q0 q0 + vq q2 − q2
× ln − 20 2 2 − 1 , (4.48)
vq q0 − vq q0 − v q

where E = P 2 + mf2 , v = P/E, mf is the effective fermion mass in plasma, and

the Fermi–Dirac distribution functions for the fermions and anti-fermions are
1 1
fF (E) = , f¯F (E) = , (4.49)
e(E −μ)/T +1 e(E +μ)/T +1

μ is the fermion chemical potential.

For the supernova core conditions, the main contribution comes from the plasma
electrons and protons:

(e) (p) (e) (p)

Rt,  Rt, + Rt, , It,  It, + It, . (4.50)

In these conditions, there is a good approximation to consider the electron fraction

as the relativistic plasma (μe , T me ).
The real and imaginary parts (4.50) of the electron contributions into the photon
polarization functions take the following form:
   
(e) x 1 − x2 1 + x
Rt = mγ2 x + 2 
ln   , (4.51)
2 1 − x
π  
It(e) = − mγ2 x 1 − x 2 , (4.52)
2  
  x  1 + x 
(e)
R = 2 mγ2 1 − x 2 1 − ln  , (4.53)
2 1 − x
 
(e)
I = π mγ2 x 1 − x 2 , (4.54)

where x = q0 /q, |x| < 1, mγ is the so-called photon thermal mass,


2α π2 T 2
mγ2 = μ2e + . (4.55)
π 3

For the proton contributions, the situation appears to be more complicated. For
the real and imaginary parts of the proton contribution into the polarization func-
tions (4.47), (4.48), for the conditions μp T , where μp is the proton chemical
potential, one obtains:
4.4 Photon Polarization Operator in Plasma 63

∞    
(p) 4α dP P 2 x 1 − x2 x + v
Rt =   x + 2 
ln   , (4.56)
π E e(E −μp )/T +1 2v x − v
0
  ∞ dP P mp |x|
(p)
It = −2α x 1 − x 2
, Pmin = √ , (4.57)
e(E −μp )/T +1 1 − x2
Pmin

4α   ∞ dP P 2
(p)
R = 1 − x2  (E −μ )/T 
π E e p +1
0
  
1 − x2 x  x + v 
× 1+ 2 − ln  , (4.58)
v − x2 v x − v
  −1
(p) (p) mp μp
I = −2 It + 2α mp2 x exp √ − +1 , (4.59)
T 1 − x2 T

where mp is the effective proton mass in plasma [18]. For example, at the nuclear
density 3 × 1014 g/cm3 , one has mp  700 MeV.
The proton chemical potential μp is defined from the equation

∞
μ3 1 dP P 2
Np  Ne  e2 = 2 . (4.60)
3π π e(E −μp )/T + 1
0

As the analysis of Eq. (4.60) shows, the difference μp − mp (the so-called non-
relativistic proton chemical potential) appears to be of the positive sign at the
temperatures T  30 − 60 MeV, and of the same order of magnitude, as the tempera-
ture. Thus, in the supernova core conditions both the approximations of the degenerate
Fermi gas and of the classical Boltzmann gas should be, in general, hardly applicable
for protons. However, as it will be shown later in Sect. 5.4, the observables such as
the neutrino luminosity appear to be rather stable with respect to the choice of the
approximation for the proton distribution function.
In the Figs. 4.4, 4.5, 4.6, and 4.7, we present for the sake of illustration the electron
and proton contributions into the eigenvalues Π ,t for the longitudinal and transverse
plasmon. It is seen that the electron and proton contributions are of the same order
of magnitude.
Together with electrons and protons, in general, a small fraction Yi of the free ions
could also present in plasma, Yi = Ni /NB , NB is the barion density. This fraction
can be considered with a good accuracy as the classical Boltzmann gas. The real and
imaginary parts of the corresponding polarization functions have the form:
64 4 Particle Dispersion in External Active Media

2000

1500

1000

2
Rl (x)/ MeV 500

500

1000
1 0.5 0 0.5 1
x

Fig. 4.4 Electron contribution (dotted line) and proton contribution (dashed line) at T = 30 MeV
to the real part of Π

2000

1000

2
Il (x)/ MeV 0

1000

2000
1 0.5 0 0.5 1
x

Fig. 4.5 Electron contribution (dotted line) and proton contribution (dashed line) at T = 30 MeV
to the imaginary part of Π
4.4 Photon Polarization Operator in Plasma 65

500

400

2
Rt (x)/ MeV 300

200

100

0
1 0.5 0 0.5 1
x

Fig. 4.6 Electron contribution (dotted line) and proton contribution (dashed line) at T = 30 MeV
to the real part of Πt

300

200

100

2 0
I t(x)/ MeV

100

200

300
1 0.5 0 0.5 1
x

Fig. 4.7 Electron contribution (dotted line) and proton contribution (dashed line) at T = 30 MeV
to the imaginary part of Πt
66 4 Particle Dispersion in External Active Media
  
(i) Z 2 Ni x
R = 4πα i 1−φ ,
T x0
 2
(i) 1 q0 q x2
I = 8 π 3/2 α Zi2 Ni sinh exp exp − , (4.61)
x0 q 2T 8 mi T x02
√
where x0 = 2 T /mi , and the function is introduced:

∞  
2  1 + u  −y2 u2
φ(y) = √ |y|3 
u ln  e du . (4.62)
π 1 − u
0

As is seen from Eq. (4.61), the function I (i) differs

√ from zero only in the narrow area
of the variable x = q0 /q, namely, x  x0 ∼ T /mi 1.
The functions Rt(i) and It(i) for the transversal plasmon are of the order α Zi2 Ni /mi
and thus are suppressed by the large mass of the ion in the denominator. Thus, the
contribution of the neutrino scattering off free ions via the longitudinal plasmon
(λ = ) is only essential.
The ion contribution (4.61) comes with the factor Zi2 Yi , and it is negligibly small
in the supernova core conditions, because of the smallness of Yi . However, it could
be essential in the upper layers of the supernova envelope, which are believed to be
rich in elements of the iron group.

4.5 Neutrino Self-energy Operator in Plasma

The most important event in neutrino physics of the last decades was the solving
of the Solar neutrino problem, made in the unique experiment on the heavy-water
detector at the Sudbury Neutrino Observatory [19–21]. This experiment, together
with the atmospheric and the reactor neutrino experiments [22–25] has confirmed
the key idea by B. Pontecorvo on neutrino oscillations [26, 27]. The existence of
non-zero neutrino mass and lepton mixing is thereby established. On the one hand,
the Sun appeared in this case as a natural laboratory for investigations of neutrino
properties. On the other hand, the process of solving of the Solar neutrino problem
significantly stimulated the progress of the Solar physics in different aspects [28] and
of several sciences investigating microscopic matter properties: physics of nuclear
reactions, radiochemistry, etc.
Another direction of neutrino astrophysics, which also interact with several
branches of physical science, is the registration of neutrinos from a supernova explo-
sion. At the moment, there is only one registered neutrino signal from the supernova
SN1987A in the Large Magellanic Cloud, where four underground neutrino detec-
tors, Kamiokande 2, IMB, LSD and Baksan scintillation telescope, for the first time
registered electron antineutrinos in the reaction ν̄e + p → n + e+ .
4.5 Neutrino Self-energy Operator in Plasma 67

Supernova explosions can be called unique natural laboratories for studying the
fundamental properties of matter under extreme physical conditions. At the same
time, one of the most important factors almost completely determining the energetics
of the process is the presence of giant neutrino fluxes. This means that the presence
of microscopic characteristics of the neutrino, determined by its dispersion in the
active medium, could have a critical impact on macroscopic properties of these
astrophysical events.
In real astrophysical conditions, the external active medium is usually represented
by two components: a strong magnetic field and the hot dense plasma. Therefore,
the investigation of the neutrino dispersion properties in a medium containing both
plasma and field is of the most interest. However, due to the large computational com-
plexity of such studies, the analyses were initially carried out, where the dominance
of one of the two indicated components of the active medium, or strong magnetic
field, or the hot dense plasma was supposed.
The calculation of the neutrino self-energy operator in a hot dense plasma without
a magnetic field was carried out in Refs. [29–31]. The contribution of the external
magnetic field into the neutrino self-energy operator, without taking into account the
plasma has been studied in a series of papers [32–37]. The series of papers [38–41]
has been devoted to the analysis of the operator Σ(p) with taking into account both
components of the environment, both field and plasma, with the dominance of the
influence of the latter, that is, the contribution of the field has been taken into account
in the form of small corrections. Finally, in the papers [42, 43] the calculation of the
operator Σ(p) in a magnetized plasma is carried out over a wide range of magnetic
field intensity.
The early Universe can be treated as another natural laboratory for fundamental
physics, where the role of neutrinos is also high. Thus, there has been a steady growth
of interest in neutrino physics in the external active media.
Investigation of the active media influence on the neutrino dispersion is based
on the analysis of the neutrino self-energy operator Σ(p). Knowing of the operator
Σ(p) can solve at least three important tasks:
(i) From the neutrino self-energy operator, an additional energy can be easily
determined acquired by neutrinos in a medium. The astrophysical medium is
asymmetric with respect to lepton flavors: it contains electrons and positrons,
but no muons and tau leptons. Due to this, neutrinos of different flavors acquire
a variety of additional energy, which is the determining factor in the influence
of environment on the neutrino flavor oscillation.
(ii) The importance of calculating the self-energy operator is supported by the fact
that you can extract from it the neutrino anomalous magnetic moment.
(iii) The imaginary part of the neutrino self-energy in the medium determines the
probability of the neutrino decay into the W + -boson and the charged lepton,
ν → −W +.
Further we discuss each of these tasks.
68 4 Particle Dispersion in External Active Media

4.5.1 Definition of the Operator Σ(p) in Plasma

The neutrino self-energy operator Σ(p) can be defined from the invariant amplitude
of the transition ν → ν by the relation
   
M(ν → ν) = − ν̄(p) Σ(p) ν(p) = −Tr Σ(p) ρ(p) , (4.63)

where pα = (E, p) is the neutrino 4-momentum, ρ(p) = ν(p)ν̄(p) is the density

matrix of neutrinos. An additional energy ΔE, acquired by neutrinos in the external
active medium is determined by the invariant amplitude (4.63) as follows:

1 1  
ΔE = − M(ν → ν) = Tr Σ(p) ρ(p) . (4.64)
2E 2E
It is convenient to represent the operator Σ(p) in plasma in a general form of an
expansion over the linearly independent covariant structures:
 
Σ(p) = AL (pγ) + BL (uγ) γL
 
+ AR (pγ) + BR (uγ) γR + K1 mν . (4.65)

Here, γL = (1 − γ5 )/2 and γR = (1 + γ5 )/2 are, respectively, the left-handed and the
right-handed chiral projection operators, uα is the 4-velocity vector of the medium.
Note that the coefficients AL , AR and K1 in Eq. (4.65) contain an ultraviolet
divergence. But it does not have an independent meaning, since it does not contribute
into the real energy of neutrinos in the external media at the one-loop level, taking
into account the renormalization of the vacuum wave function and the mass of a
neutrino.

4.5.2 Neutrino Additional Energy in Hot Dense Plasma

As was first shown by L. Wolfenstein [44], studying the propagation of neutrinos

in a medium one must take into account the effect of coherent forward scattering.
In astrophysical conditions, the influence of a medium on the neutrino properties
is primarily due to the additional energy, which only a left-handed neutrino (with
the spin oriented opposite to the direction of motion) acquires. For illustration, we
present here a detailed calculation of the contribution into the neutrino additional
energy from the electron-positron plasma component in accordance with Eq. (4.64).
Note that in the approximation of the massless left-handed neutrino, there are only
two linearly independent covariant structures present in the expression (4.65) for the
operator Σ(p) with the coefficients AL and BL .
Let us consider the process of a coherent neutrino forward scattering on electrons
and positrons of the plasma. To begin with, we consider the local limit of the weak
4.5 Neutrino Self-energy Operator in Plasma 69

Fig. 4.8 Feynman diagrams illustrating the transition to an effective ν − e interaction in the local
limit

interaction of the left-handed neutrinos with electrons, when the propagators of the
intermediate W and Z bosons “shrink” to the point, as shown in Fig. 4.8. The effects
of non-locality of the weak interaction which must be taken into account in a case
of high neutrino energies, will be considered below in Sect. 4.5.4.
The effective local Lagrangian of the neutrino—electron interaction can be written
in the form1
GF  ( ) ( )  
L = − √ ēγα (CV − CA γ5 )e ν̄γ α (1 − γ5 )ν , (4.66)
2
( ) ( )
where the constants CV and CA are different in two cases:
• if the neutrinos in the Lagrangian (4.66) are of the electron type, ν = νe , a
contribution from the exchange of Z and W boson appears, and we have:

(e) 1 (e) 1
CV = + + 2 sin2 θW , CA = + , (4.67)
2 2

where θW is the Weinberg angle, sin2 θW  0.231;

• if we consider the muon and tau neutrinos, ν = νμ , ντ , only the Z boson contributes
in this case, and we have:

(μ) 1 (μ) 1
CV = − + 2 sin2 θW , CA = − . (4.68)
2 2
Let us start with the scattering by electrons. We write the S matrix element of the
process in the standard form:

i(2π)4 δ (4) (p + k − p − k)
S (e−) = √ M{ν(p)+e− (k) → ν(p )+e− (k )} . (4.69)
2EV 2εV 2E V 2ε V

1 Note that the sign of the effective Lagrangian is significant in this case, since the additional
neutrino energy is the linear in GF effect. In the calculation of probabilities and cross sections of
weak processes, which are proportional to G2F , the sign of the effective Lagrangian does not appear.
70 4 Particle Dispersion in External Active Media

Here, pα = (E, p) and k α = (ε, k) are, respectively, the 4-momenta of the initial
neutrino and electron, p and k are the 4-momenta of the final neutrino and electron,
M is the invariant amplitude:

GF  
M{ν(p) + e− (k) → ν(p ) + e− (k )} = − √ ē(k )γα (CV(e) − CA(e) γ5 )e(k)
2
 
× ν̄(p )γ α (1 − γ5 )ν(p) . (4.70)

Given the process is a forward scattering, we need to put p = p and k = k in the S

matrix element (4.69). At the same time

4 (4)
(2π) δ (0) = d4 x ei0 = V T , (4.71)

where V is the total volume of the interaction region, T is the total time of interaction.
The S matrix element of the forward neutrino scattering off the electrons takes the
form
(e−) iV T
Sforw = M{ν(p) + e− (k) → ν(p) + e− (k)} . (4.72)
2EV 2εV
Since this is a coherent process, the total scattering amplitude is obtained by summing
the scattering amplitudes for all electrons of the medium:

 
(e−) (e−) d3 k V (e−)
Stot = Sforw = 2 fe (k) Sforw , (4.73)
(2π)3
k,s

where the coefficient 2 takes into account two electronic spin states s, fe (k) is the
distribution function of the electrons of medium. We assume this distribution to be
in an equilibrium and consider the reference frame where the medium moves as a
whole with the 4-velocity vector u. The Fermi—Dirac distribution function is written
as  −1
(ku) − μe
fe (k) = exp +1 , (4.74)
T

where μe is the chemical potential of the electron-positron plasma, T is the plasma

temperature.
We can now determine the contribution to the invariant transition amplitude
M(e−) (ν → ν) caused by the coherent forward scattering on the electron fraction
of plasma, from the expression

(e−) iV T
Stot = S (e−) (ν → ν) = M(e−) (ν → ν) . (4.75)
2EV

From Eqs. (4.70)–(4.75), we obtain

4.5 Neutrino Self-energy Operator in Plasma 71

√ 
(e−) (e)  α
 d3 k kα
M (ν → ν) = − 2 GF CV ν̄(p)γ γL ν(p) 2 fe (k) . (4.76)
(2π)3 ε

The arising integral is a relativistic covariant and can be easily calculated:


d3 k kα
2 fe (k) = uα Ne , (4.77)
(2π)3 ε

where Ne is the electron density.

The contribution from the coherent forward scattering on the positron fraction of
plasma can be calculated quite similarly. As a result, taking into account Eq. (4.63),
we finally obtain the neutrino self-energy operator in the electron-positron plasma
in the form √  
(e)
Σ(p) = 2 GF CV (uγ)γL Ne − N̄e , (4.78)

where Ne and N̄e are the densities of electrons and positrons. Comparing (4.78)
√
with (4.65), one can see that only one structure with a coefficient BL = 2 GF CV(e)
 
Ne − N̄e presents in the operator Σ(p) in this case.
According to Eq. (4.64), for the additional neutrino energy in electron-positron
plasma we obtain
√ (e) (pu)  
ΔE = 2 GF CV Ne − N̄e . (4.79)
E
In the transition from an arbitrary reference frame to the plasma rest frame one should
put (pu) = E.
In the analysis of the neutrino dispersion properties in the active astrophysical
media one should generally take into account, along with the electron-positron
plasma, the presence of other components. The contribution of protons and neu-
trons can be found similarly to the previous analysis, with the effective Lagrangian
caused only by the exchange of Z boson (see Fig. 4.8). In a dense plasma of the
supernova core, the contribution of the neutrino gas which can be regarded to be in
approximate equilibrium, could also be significant. The general expression for the
additional energy of the electron, muon and tau neutrinos, i = e, μ, τ , is given by

√ 1  
ΔEi = 2 GF δie − + 2 sin2 θW Ne − N̄e
2

1  
+ − 2 sin2 θW Np − N̄p (4.80)
2
⎤
1    
− Nn − N̄n + (1 + δi ) Nν − N̄ν ⎦ ,
2
=e,μ,τ

where Ne , Np , Nn , and Nν are the densitis of electrons, protons, neutrons and

neutrinos, N̄e , N̄p , N̄n , and N̄ν are the densitis of the corresponding antiparticles.
72 4 Particle Dispersion in External Active Media

To find the additional energy of an antineutrino in plasma, one should change the
overall sign in the right-hand side of Eq. (4.80).
The formula (4.80) obtained in the local limit of weak interaction, is not sufficient
for the case when the plasma is nearly charge-symmetric, for example, in the early
Universe. In this case the value of ΔEi in Eq. (4.80) tends to zero, and the contribution
into the neutrino energy becomes significant caused by the nonlocality of the weak
interaction. This non-local contribution was investigated in Refs. [29, 39, 45] in the
form of the next terms in the expansion of the W – and Z–boson propagators by the
−2
inverse powers of their masses mW ,Z . The result can be presented as follows:

16GF E Eν Nν + Eν̄ N̄ν

Δ(nloc) E ν = − √
3 2 mZ2
Ee Ne + Eē N̄e
+ δ e 2
. (4.81)
mW

Here, Eν , Eν̄l , Ee , Eē  are the average energies of plasma neutrinos, antineu-
trinos, electrons and positrons respectively. In a particular case of a charge symmetric
hot plasma, the expression (4.81) reproduces the result of Refs. [29, 39]:
√
7 2 π 2 GF T 4 1 2δ e
Δ(nloc) E ν = − + 2 E. (4.82)
45 mZ2 mW

However, the correction of the type of Eq. (4.81) can be insufficient in the case of
ultra-high neutrino or antineutrino energies. The neutrino self-energy operator with
using the exact dependence of the propagators of gauge bosons on the momentum
transferred was investigated in Ref. [46], see Sect. 4.5.4 below.
For a typical astrophysical plasma, with the exception of the early Universe and
supernova core, we have N̄e  N̄p  N̄n  Nν  N̄ν  0 and Np  Ne =
Ye NB , Nn  (1 − Ye ) NB , where NB is the density of baryons. If the neutrino energy
is not extremely high, for the additional energy of neutrinos of different flavors, we
obtain
GF NB
ΔEe = √ (3 Ye − 1) , (4.83)
2
GF NB
ΔEμ,τ = − √ (1 − Ye ) . (4.84)
2

Since Ye < 1, the additional energy of the left-handed muon and tau neutrinos is
always negative. At the same time, the additional energy of the left-handed electron
neutrinos is positive for Ye > 1/3. Conversely, the additional energy of the electron
antineutrinos is positive for Ye < 1/3, while it is always positive for the muon
and tau antineutrinos. In turn, the right-handed neutrino, with the spin oriented in
4.5 Neutrino Self-energy Operator in Plasma 73

the direction of movement, and its antiparticle, the left-handed antineutrino, being
sterile with respect to weak interaction, do not acquire the additional energy.

4.5.3 On the Neutrino Radiative Decay in Plasma

It should be noted that the history of studies of the neutrino dispersion modifications
by plasma has not been without its oddities. In this section, we illustrate how a consid-
eration of the plasma influence on the neutrino dispersion, with ignoring the photon
dispersion in plasma, has led the authors [47], for a comprehensive list of references
see [48], to a detailed discussion of an effect, which is physically impossible, strictly
speaking.
It is known that the effect of plasma on the particle properties may open new
possibilities for the realization of processes, forbidden in vacuum by conservation
laws. However, it is necessary to consider the plasma impact on all components of
the process, and it can complicate the kinematics essentially.
The additional energy ΔE, defined by the expression (4.83), results in the appear-
ance of the effective mass square mL2 for the left-handed electron neutrinos:

mL2 = P 2 = (E + ΔE)2 − p2 , (4.85)

where P is the 4-momentum of the neutrino in a plasma in its rest frame, while the
4-vector (E, p) would be a 4-momentum of the neutrino in vacuum, E = p2 + mν2 .
Given the neutrino magnetic moment interaction with a photon, which leads to
the neutrino helicity-flip, the appearance of an additional energy for left-handed
neutrinos in plasma would open new kinematic possibilities for the neutrino radiative
transition:
νL → νR + γ . (4.86)

It can be considered as the radiative decay of the left-handed neutrino which becomes
heavier in plasma, into lighter right-handed neutrino.
At the same time it should be obvious that it is necessary to take into account the
influence of plasma on the dispersion of the photon ω = |k|/n, where n = 1 is the
index of refraction.
First of all, the plasma influence can provide the condition n > 1 to be satisfied
(the square of the effective photon mass is negative, mγ2 ≡ q2 = ω 2 − k2 < 0),
which corresponds to a well-known effect of the neutrino Cherenkov radiation
[10, 49, 50]. In this situation, a change of the neutrino dispersion properties under
the plasma influence could be neglected at all. Really, while the neutrino dispersion
is defined by a weak interaction, the change of the photon dispersion depends on its
much more intense electromagnetic interaction with plasma.
74 4 Particle Dispersion in External Active Media

Theoretically, one can consider another situation: if the photon dispersion in

plasma was the same as in vacuum,2 the process would occur of the neutrino radiative
transition ν → νγ, caused only by the neutrino dispersion. Since the effect of the
plasma changes the dispersion properties of the only left-handed neutrino, the tran-
sitions (4.86) would be possible due to the photon interaction with the neutrino
magnetic moment. Such an imaginary effect, called “neutrino spin light” (SLν),
has been proposed and studied in detail in an extensive series of papers, see [48].
However, in analyzing this effect the authors have not considered such an important
phenomenon as the above-mentioned plasma influence on the photon dispersion. As
was shown in [51, 52], this phenomenon makes the SLν effect forbidden for all real
astrophysical situations.
Following the papers [51, 52], we analyze here the process νL → νR γ, taking into
account the dispersion properties of both the neutrino and photon in astrophysical
plasmas.
To analyze the kinematics of the process, it is worthwhile to estimate the scales
of the values of additional neutrino energy ΔE and the effective mass of the photon
(plasmon) mγ .
From the expression (4.83) for the electron antineutrino, we obtain

NB
ΔE  6 eV (1 − 3 Ye ) , (4.87)
1038 cm−3

where the scale of the baryon density is taken, which is typical e.g. for the interior
of a neutron star.
In turn, a plasmon acquires in medium an effective mass mγ , which is approx-
imately constant at high energies. For the transverse plasmon, the value of mγ2 is
always positive and is determined by the so-called plasma frequency ωP . For a non-
relativistic classical plasma (e.g. in the Sun), we obtain
  1/2
4π α Ne Ne
mγ ≡ ωP =  4 × 102 eV . (4.88)
me 10 cm−3
26

For the ultra-relativistic dense matter one has:


2α π2 2
mγ =
2
μ2e + T , (4.89)
π 3

where μe is the chemical potential of plasma electrons. For the case of cold degenerate
plasma one obtains from Eq. (4.89):

2 Strictly speaking, a particle that interacts with the magnetic moment of neutrinos, and at the same

time, is sterile with respect to the interactions with electrically charged plasma particles, should not
be called a photon.
4.5 Neutrino Self-energy Operator in Plasma 75
  1/2  1/3  1/3
3 2α Ne
mγ = ωpl = 3 π Ne
2
 10 eV
7
. (4.90)
2 π 10 cm−3
37

In the case of hot plasma, where its temperature is the largest physical parameter,
the effective mass of the plasmon is
 
2πα T
mγ = T  1.2 × 107 eV . (4.91)
3 100 MeV

Comparison of the scales of mγ (4.88)–(4.91) with the scale of ΔE (4.87) should

indicate that neglecting the mass of the plasmon, made in the consideration of the SLν
effect [48], was obviously incorrect. At the same time another physical parameter, a
great attention was paid to in the SLν analysis, was the neutrino vacuum mass mν . As
the scale of neutrino vacuum mass could not exceed essentially a few electron-volts,
which is much less than typical plasmon mass scales for real astrophysical situations,
see Eqs. (4.88)–(4.91), it is reasonable to neglect mν in our analysis.
Thus, in accordance with (4.85), a simple condition for the kinematic opening of
the process νL → νR γ is:
mL2  2 E ΔE > mγ2 . (4.92)

This means that the process becomes kinematically opened when the neutrino energy
exceeds the threshold value,
mγ2
E > E0 = . (4.93)
2 ΔE
The appearance of the threshold energy of neutrinos can be demonstrated by
considering the range of integration over the energies and momenta of the photon
(plasmon) in the νL → νR γ process, taking into account the dispersion properties
of both neutrinos and photons in astrophysical plasmas. In Fig. 4.9, the line of the
photon dispersion in vacuum, q0 = k, lies inside the allowed kinematical region
(left panel), whereas the line of the photon dispersion, modified by plasma, may be
outside this area if the neutrino energy is not large enough (right panel). In this case
the phase volume, and hence the process probability is zero.
For fixed plasma parameters, the value ωP remains constant. The value ΔE remains
constant also, if we disregard the contribution to the neutrino energy from the non-
locality of the weak interaction. Therefore, in order to obtain the non-zero phase
volume and the process probability or, in other words, in order to put a part of the
plasmon dispersion curve into the integration region, it is necessary to increase the
neutrino energy E, i.e., the width of the oblique rectangle in Fig. 4.9. It should be
clear, that there is the minimum energy E0 for the integration region to exist. This is
just the threshold energy (4.93).
Let us estimate these threshold energies for various astrophysical situations.
In the approximation of nonrelativistic classical plasma, one obtains from Eqs.
(4.87) and (4.88) that the threshold neutrino energy does not depend on density, and
do depend on the chemical composition only:
76 4 Particle Dispersion in External Active Media

k k
q0 = k

q0 = q (k)
0

0 ΔE q
0 0 ΔE ωP q
0

Fig. 4.9 The integration region for the calculation of the probability of the process νL → νR γ at
fixed energy E of the initial neutrino (inside the oblique rectangle drawn by dashed lines) and the
line of the photon dispersion (thick line) in a vacuum (left panel) and in plasma (right)

Ye m2
E0  4 sin2 θW W . (4.94)
3Ye − 1 me

For the solar interior Ye  0.6, and the threshold neutrino energy is

E0  1010 MeV , (4.95)

to be compared with the upper bound ∼20 MeV for the solar neutrino energies.
For the interior of a neutron star, where Ye 1, the additional energy for neutri-
nos (4.83), (4.84) is negative, and the process νL → νR + γ is closed. On the other
hand, there exists a possibility for opening the antineutrino decay. Taking for the
estimation Ye  0.1, one obtains from (4.87) and (4.89) the threshold value

E0  107 MeV , (4.96)

to be compared with the typical energy ∼1—0.1 MeV of neutrinos emitted via the
direct or modified Urca processes [53].
For the conditions of a supernova core, the additional energy of left-handed elec-
tron neutrinos can be obtained from Eq. (4.80) in the form:

GF NB  
ΔEe = √ 3 Ye + 4 Yνe − 1 , (4.97)
2

where Yνe describes the fraction of the trapped electron neutrinos in the supernova
core, Nνe = Yνe NB . Using the typical parameters of a supernova core, we obtain

E0  107 MeV , (4.98)

to be compared with the averaged energy ∼102 MeV of trapped neutrinos.

4.5 Neutrino Self-energy Operator in Plasma 77

In the early Universe, when the plasma was almost charge symmetric, the
formula (4.80), which gives a null result must be supplemented by the non-local
contribution (4.82), which is the same for neutrinos and antineutrinos. The minus
sign in (4.82) unambiguously shows that in the early Universe, in contrast to the
neutron star interior, the process of the radiative spin-flip transition is forbidden both
for neutrinos and antineutrinos regardless of their energy.
An analysis of the sum of the local and non-local weak contributions (4.80)
and (4.81) in a case if the neutrino energy is not ultra-high, shows that adding of the
non-local term leads in general to the decreasing of the additional neutrino energy in
plasma, i.e. to the increasing of the threshold energy (4.93). Strictly speaking, one
has to perform an analysis of the kinematical inequality (4.92), which leads to the
solving of the quadratic equation. As a result, there arises the window in the neutrino
energies for the process to be kinematically opened, E0 <E<Emax , where E0 and Emax
are the lower and the upper limits connected with the roots of the above-mentioned
quadratic equation, if they exist. For example, in the solar interior there is no window
for the process with electron neutrinos at all, i.e. the transition νeL → νeR + γ is
forbidden kinematically.
Thus, the above analysis shows that the nice effect of the “spin light of neutrino”,
unfortunately, has no place in real astrophysical conditions if the dispersion properties
of neutrinos and photons are properly taken into account. The sole possibility for
the discussed process νL → νR + γ to be theoretically possible, could be connected
only with the situation when an ultra-high energy neutrino threads a star. Obviously,
this task can only have a purely methodological sense. In the papers [51, 52], the
mean free path L of the ultra-high energy neutrino with respect to the radiative decay
process was correctly calculated in the situation where a neutrino arrived from outside
penetrates a neutron star.
Based on the typical neutron star parameters NB  1038 cm−3 , Ye  0.05, the
mean free path was obtained:
 2
10−12 μB
L  1019 cm × , (4.99)
μν

where μν is the neutrino magnetic moment, μB is the Bohr magneton. This mean free
path should be compared with the radius of the neutron star ∼106 cm, to illustrate
the extremely low probability of the process.
It is interesting to note that it was not the first case when the plasma influence was
taken into account for one participant of the physical process while it was not taken
for other participant. As E. Braaten wrote in Ref. [1]:
“In Ref. [54], it was argued that their calculation for the emissivities from photon
and plasmon decay would break down at temperatures large enough that mγ > 2 me ,
since the decay γ → e+ e− is then kinematically allowed. This statement, which has
been repeated in subsequent papers, [55–58] is simply untrue. The plasma effects
which generate the photon mass mγ also generate corrections to the electron mass
such that the decay γ → e+ e− is always kinematically forbidden.”
78 4 Particle Dispersion in External Active Media

Thus, a history repeated itself. The authors [48] made the same mistake when they
considered the plasma-induced additional neutrino energy and ignored the effective
photon mass mγ arising by the same reason.
The only question remained open whether this effect was possible in the case of
ultra-high neutrino energies [59]. This gap was eliminated in Ref. [46]. In the next
section, we reproduce that analysis.

4.5.4 Ultra-High Energy Neutrino Dispersion in Plazma

As it was already mentioned, the accounting of the non-local contribution to the

neutrino additional energy made by the retention of the next term in the expan-
sion of the W – and Z–boson propagators in the inverse powers of their masses
[29, 39, 45, 52] would be irrelevant in the limit of the ultra-high neutrino energies.
Therefore, it is necessary to use the exact expressions for the W – and Z–boson propa-
gators. Analysis of the neutrino additional energy in a plasma in the limit of ultra-high
energies, with taking account of the nonlocality of the weak interaction was made
in a series of papers, Refs. [60–62], with respect to the neutrino oscillations. In this
section, we consider the neutrino self-energy operator in medium similarly to the
procedure described in Sect. 4.5.2 but with taking into account the dependence of
the W and Z–boson propagators on the momentum transferred, and we reanalyse its
effects on the neutrino radiative conversion (4.86). The presentation is based mainly
on Ref. [46].
We first consider the electron neutrino scattering on the electron-positron compo-
nent of plasma. For the channel of the νe e scattering through the W –boson exchange,
the Lagrangian of the interaction is:
g g
L = √ (ē γα (1 − γ5 ) νe ) Wα + √ (ν̄e γα (1 − γ5 ) e ) Wα† . (4.100)
2 2 2 2

It leads to the invariant amplitude of the process:

GF  
Mνe e− →νe e− = − √ ē(k )γα (1 − γ5 )e(k)
2
  m2
× ν̄e (p )γ α (1 − γ5 )νe (p) 2 W 2 , (4.101)
mW − q1

where we use the notation q1 = k −p for the W − –boson momentum (see Fig. 4.10a).
Here, the Fiertz transformation is performed, and the term in the W –boson propagator
leading to the small term of the order of (me /mW )2 is neglected.
The amplitude of the neutrino-positron scattering process can be written in the
similar form (see Fig. 4.10b):
4.5 Neutrino Self-energy Operator in Plasma 79

Fig. 4.10 The Feynman

diagrams for the neutrino ν e (p) e −(k ) ν e (p) ν e (p )
W −(q2 )
scattering through W –boson:
←
a on plasma electrons; b on
↑W − (q1 )
plasma positrons

e−(k ) ν e (p ) e+(k ) e+(k )

(a) (b)

GF  
Mνe e+ →νe e+ = √ ē(−k)γα (1 − γ5 )e(−k )
2
  m2
× ν̄e (p )γ α (1 − γ5 )νe (p) ∗ 2 W 2 , (4.102)
mW − q2

where W − –boson momentum is q2 = −p − k. Note that the amplitudes (4.101)

and (4.102), described by the diagrams in Figs. 4.10a,b differ essentially. Namely,
in the s-channel process of the neutrino scattering off positrons, Fig. 4.10b, we have
q22 > 0, i.e. a resonance behavior of the W –boson propagator manifests itself. On the
contrary, in the u-channel process of the neutrino scattering off electrons, Fig. 4.10a,
we have q12 < 0, and no resonance arises. Taking account of this type of resonance
is made by introducing a complex mass of W –boson, mW ∗ = m − 1 i Γ , where
W 2 W
ΓW is the total decay width of W –boson, ΓW  2.1 GeV, see e.g.[63].
Because of the t –channel behavior of the neutrino-electron and neutrino-positron
scattering diagrams for neutrinos of all flavors through Z–boson, see Fig. 4.11, and
keeping in mind that the forward scattering is considered, i.e. the scattering with
zero-momentum transfer, one concludes that the contribution to the energy from
these subprocesses is described by the local limit of the weak interaction.
The total contribution to the –flavor neutrino self-energy operator from the scat-
tering processes on plasma electrons and positrons can be found by the same way as
Eq. (4.78) and be represented in the form:
√ 
ν (μ)
Σ(e − e+ ) (p) = 2GF CV (uγ)γL (Ne − N̄e )
− 
+ δ e γ α γL mW2
jα − jα+ , (4.103)


where Ne , N̄e = 2(2π)−3 d3 k (exp ((ε ∓ μe )/T ) + 1)−1 are the electron and
positron densities respectively, and we use the notation
  −1  −1
d3 k kα ε ∓ μe
jα∓ =2 exp +1 2
mW ± 2(kp) . (4.104)
(2π)3 ε T

(μ)
The constant CV in Eq. (4.103) comes from the electron Z–current and is the same
for = e, μ, τ , see Eq. (4.68). For taking account of the resonance behavior in the
denominator of the integral jα+ , mW should be replaced by mW∗ .
80 4 Particle Dispersion in External Active Media

Fig. 4.11 The Feynman

diagrams for the neutrino ν (p) ν (p ) ν (p) ν (p )
scattering on plasma elec-
trons and positrons through
Z–boson Z Z

e −( k ) e −( k ) e+( k ) e +( k )

(a) (b)

In accordance with Eq. (4.64), the neutrino ν additional energy in the electron
and positron medium takes the form:
√ 
ν (μ)
ΔE(e − e+ ) = 2GF CV (Ne − N̄e )
 ∗

+ δ e F1 (μe , mW ) − F2 (−μe , mW ) , (4.105)

where we introduce the functions

  −1
2 m2 d3 k ε−μ (pk)
F1,2 (μ, m) = exp +1 . (4.106)
(2π)3 E ε T m2 ± 2(pk)

In order to obtain the antineutrino additional energy in the same medium, one has
to make the replacement μe → −μe in the right-hand side of Eq. (4.105). In the
first term with the difference of the electron and positron densities it simply means
a change of sign.
In the analysis of the neutrino dispersion in active astrophysical medium in a
general case, the presence of the other plasma components, protons and neutrons,
must be considered. In a dense plasma of the supernova core the donation from
thermal neutrinos that can be considered to be approximately in equilibrium, can
also be significant. The corresponding Feynman diagrams are shown in Fig. 4.12.
The two Feynman diagrams, Fig. 4.12c,d contain a contribution from the non-locality
of weak interaction.
A complete formula for the ν neutrino and ν̄ antineutrino additional energy can
be written in the following way:

ν ,ν̄
√ 1
ΔE = 2GF ∓ (Nn − N̄n ) ± (Nνe − N̄νe )
2
± (Nνμ − N̄νμ ) ± (Nντ − N̄ντ )
 ∗

+ δ e F1 (±μe , mW ) − F2 (∓μe , mW )

1 ∗

+ F1 (±μ̃ν , mZ ) − F2 (∓μ̃ν , mZ ) . (4.107)
2
4.5 Neutrino Self-energy Operator in Plasma 81

Fig. 4.12 The Feynman

diagrams for the neutrino ν ( p) ν (p ) ν ( p) ν (p )
scattering on neutrinos and
antineutrinos of the supernova
core dense plasma Z Z

ν (k ) ν (k ) ν (k ) ν (k )

(a) (b)

ν ( p) ν (p )
ν ( p) ν (k )
Z
Z

ν (k ) ν (p ) ν (k ) ν (k )

(c) (d)

In this expression, Nn , Nν are the neutron and neutrino densities and N̄n , N̄ν
are the densities of the corresponding antiparticles. The proton contribution in
Eq. (4.107) is cancelled by the electron contribution, because of plasma electroneu-
trality. Note that in both functions F2 there exists the mentioned above resonance
behavior, which is accounted by the introduction of complex masses of W and Z
∗
,Z = mW ,Z − 2 i ΓW ,Z , where the total decay width of the Z– boson is
1
bosons, mW
ΓZ  2.5 GeV.
Tending formally mW and mZ in Eq. (4.107) to infinity, one obtains the neutrino
additional energy in the local limit of weak interaction, the so-called Wolfenstein
energy [44]. Taking the next terms in the expansion of the W – and Z–boson prop-
−2
agators by the inverse powers of their masses, mW ,Z , i.e. retaining the first term in
−2
the expansion of the functions F1,2 by m , one obtains the first non-local correc-
tion (4.81) to the Wolfenstein energy.
Further we consider the kinematical possibilities of the ultra-high-energy neutrino
radiative conversion (4.86) for different astrophysical situations.

4.5.4.1 Nonrelativistic Cold Plasma: The Sun and Red Giants

Let us consider first the limit of “cold” plasma, T → 0. In this case, the electron
gas is completely degenerate, and there are no positrons in a medium. Calculation
of the additional neutrino energy reduces in this case to a simplified calculation of
the function F1 (μe , mW ), taking the form in the limit T → 0 :
82 4 Particle Dispersion in External Active Media

μe 
m2 2
mW m2 + 2E ε(1 + v)
F1 (μe , mW ) = W v ε dε 1 − ln W , (4.108)
2π 2 E 4Ev ε mW2 + 2E ε(1 − v)
me

where ε is the energy of a plasma electron, and v = 1 − me2 /ε2 is its velocity.
Similarly, calculation of the additional antineutrino energy reduces to the calcu-
lation of the integral:

μe
∗ m2
F2 (μe , mW )=− W vεdε
2π 2 E
me
  2 2
2
mW mW − 2Eε(1 − v) + mW
2 Γ2
W
× 1− ln   2
(4.109)
8Evε mW − 2Eε(1 + v) + mW ΓW2
2 2

2
i mW m2 − 2E ε(1 − v) m2 − 2E ε(1 + v)
− arctan W − arctan W .
4Evε mW ΓW mW ΓW

Here, we have neglected the terms of order ΓW /mW compared to unity wherever
it does not cause problems. Thus, the imaginary part of the additional antineutrino
energy, in general, differs from zero. The presence of the imaginary part in the self-
energy of a particle indicates its instability, that is, an electron antineutrino is unstable
with respect to the process ν̄e + e− → W − on the plasma electrons. The width of
this process can be found, using the formula:

w = −2 Im ΔE . (4.110)

In the case of non-relativistic cold plasma, the integral (4.108), with taking account
of the smallness of the Fermi momentum, pF = μ2e − me2 me , can be obtained
in the form:

(nr) p3F Ye NB
F1 (μe , mW ) =  = , (4.111)
3π 2 1 + 2me E(mW ) −2 1 + 2me E(mW )−2

where NB is the baryon density, Ye = Ne /NB is the fraction of electrons.

Let us consider the high-energy neutrino propagation through the “cold” plasma
of the Sun or of red giants, where the temperature is T ∼ (107 − 108 ) K ∼ (10−3 −
10−2 ) me , and the electron density is Ne √
∼ 1026 cm−3 . The effective plasmon mass in
these conditions takes the form: mγ = 4παNe /me . The stellar substance is trans-
parent for the neutrino radiation, thus the contribution into the neutrino additional
energy from thermal neutrinos can be neglected.
As a result, the additional energy of a neutrino ν in the nonrelativistic cold plasma
becomes:
4.5 Neutrino Self-energy Operator in Plasma 83

√ δ e Ye 1
ΔE ν = 2GF NB − (1 − Ye ) . (4.112)
1 + 2me E(mW )−2 2

Accordingly, the additional energy of an antineutrino ν̄ in the same conditions can

be written as:

√ −δ e Ye 1
ΔE ν̄ = 2GF NB + (1 − Ye ) . (4.113)
1 − 2me E(mW )−2 − iΓW (mW )−1 2

The analysis of the threshold inequality (4.92) for the electron neutrino reduces,
in view of (4.112), to the investigation of the positiveness of the square trinomial
with respect to the energy E. Assuming that Ye  0.6 inside the Sun, we conclude
that the inequality (4.92) is not satisfied for any neutrino energies. One can see
that taking account of the non-locality of the weak interaction dramatically changes
the conclusion on a possibility of the electron neutrino radiative conversion in the
nonrelativistic cold plasma. Really, in the earlier papers [51, 52] where the local
limit of the weak interaction was used, it was concluded that the neutrino radiative
conversion in the considered conditions was possible for neutrino energies E greater
than threshold energy E0  107 GeV. However, in reality the effect for νe is totally
closed.
Consider now the possibilities for a trueness of the inequality (4.92) in the same
conditions for other neutrino flavors. Note that the question on any observational
realization of this process remains open.
The analysis of the inequality (4.92) for the electron antineutrino, in view
of (4.113), where a real part of ΔE should be taken, shows that the radiative neutrino
conversion is possible for antineutrino energies greater than the threshold energy
value, E > E0  0.6 × 107 GeV.
As it was already mentioned, the imaginary part of ΔE ν̄e causes the instability
of the electron antineutrino with respect to the process ν̄e + e− → W − on plasma
electrons. Using the formula (4.110) one obtains from Eq. (4.113) the width of the
process:
√ ΓW E0 /mW
w(ν̄e + e− → W − ) = 2 2 GF Ne E0 , (4.114)
(E − E0 )2 + (ΓW E0 /mW )2

where E0 = mW 2 /(2m ). Evaluation of a mean free path with respect to this process,
e
λ = 1/w, for Ne ∼ 1026 cm−3 , E ∼ 107 GeV provides λ ∼ 100 km, while the mini-
mum value is reached at E = E0 , to be: λ ∼ 200 m. It is obvious, that the process
ν̄e + e− → W − dominates the radiative neutrino conversion, see e.g. Eq. (4.99). If
one formally takes the limit ΓW → 0 in Eq. (4.114) to obtain:
√
w(ν̄e + e− → W − ) = 2 2 π GF Ne E0 δ(E − E0 ) . (4.115)
84 4 Particle Dispersion in External Active Media

It coinsides with the result of a direct calculation of the W –boson production by ν̄e
scattered off nonrelativistic electron gas, without taking account of the instability of
the W –boson.
The interaction of the μ- and τ -neutrinos with medium occurs only through the
Z–boson exchange with the zero momentum transfer and, as it was pointed above, it
is completely described by the local limit of the weak interaction. As it can be seen
from Eq. (4.112), the additional energy of νμ and ντ is negative, consequently, the
neutrino radiative conversion process is closed for these neutrino flavors.
In turn, the additional energy (4.113) of antineutrinos ν̄μ and ν̄τ is positive. To
estimate the border of the kinematically possible region for the SLν process in this
case one can use a simple inequality:

2
Ye mW
E > E0 = 4 sin2 θW . (4.116)
1 − Ye me

For Ye  0.6, the process is kinematically opened for μ – and τ –antineutrino energies
greater than E0  2 × 107 GeV.

4.5.4.2 Relativistic Cold Plasma: Neutron Stars

The substance of a neutron star is transparent for the neutrino radiation, as in the
previous case. Electrons in extremely dense neutron stars are ultra-relativistic, there-
fore μe  pF  120 (Ne /(0.05 N0 ))1/3 MeV, where pF is the electron Fermi mo-
mentum, and N0 = 0.16 Fm−3 is the typical nuclear density [64]. Due to the modern
estimations, the temperature inside neutron stars does not exceed a part of MeV, so
the electron gas can be considered to be degenerate and the approximation of the
zero temperature can be used. In this case the electron density is Ne = μ3e /(3π 2 )
and the square effective plasmon mass is mγ2 = 2αμ2e /π.
The functions F1,2 (μ, m) for ultrarelativistic electrons can be obtained from
Eqs. (4.108) and (4.109) by taking the limit me → 0.
The additional energy for a neutrino ν under conditions being considered takes
the following form:

√ 1 δe
ΔE ν = 2GF − (1 − Ye ) NB + 2 A(E, μe ) , (4.117)
2 2π

1
A(E, μe ) = 2
4EmW μe (mW
2
+ 2Eμe )
16E 3

4Eμe
− (mW + 4Eμe mW ) ln 1 + 2
6 4
. (4.118)
mW
4.5 Neutrino Self-energy Operator in Plasma 85

The analysis of the threshold inequality (4.92) with taking account of Eqs. (4.117),
(4.118) indicates that the SLν process for the electron neutrino is forbidden in the
conditions of a neutron star.
The similar analysis can be held for an antineutrino ν̄ . The additional energy in
this case is 
√ 1 δe
ΔE ν̄ = 2GF (1 − Ye ) NB − 2 Ā(E, μe ) , (4.119)
2 2π

μe 1
(1 − x)dx
Ā(E, μe ) = 2
k dk . (4.120)
1 − 2E(1 − x)k(mW )−2 − iΓW (mW )−1
0 −1

This integral can be easily calculated analytically but the final expression is too
cumbersome. From the analysis of the kinematically possible region (4.92), where
a real part of ΔE should be taken, we can conclude that the radiative conversion
process (4.86) is permitted for the electron antineutrino for energies greater than the
threshold value E0  8 × 104 GeV, for Ye  0.1 and NB  1037 cm−3 .
A comparison of these conclusions with the results of Refs. [51, 52] shows that tak-
ing account of the non-locality of the weak interaction does not lead to any qualitative
changes of the conclusions on kinematical possibilities of the radiative conversion
for the electron neutrino and antineutrino in the conditions of a neutron star.
Again, as in the considered case of nonrelativistic cold plasma, the imaginary part
of ΔE ν̄e means an instability of the electron antineutrino with respect to the process
ν̄e + e− → W − on plasma electrons. A width of the process can be obtained from
Eqs. (4.110), (4.119), and (4.120), but in a general case the expression is rather cum-
bersome. It is esssentially simplified for high neutrino energies, E mW ΓW /μe ,
taking the form:

GF m4 μe 2
mW 2
mW
w(ν̄e + e− → W − ) = √ W 1− Θ E− . (4.121)
2 2 π E2 4μe E 4μe

Evaluation of a mean free path with respect to this process for μe  120 MeV,
E  5×104 GeV provides λ ∼ 10−5 cm. Domination of the process ν̄e +e− → W −
over the radiative neutrino conversion in the neutron star conditions is undoubted,
see Eq. (4.99).
For μ –, τ –neutrino and antineutrino, as well as in the case of nonrelativistic cold
plasma, it is correct to use the local limit of the weak interaction. Substituting the
additional energy for = μ, τ

GF
ΔE ν ,ν̄
= ∓ √ (1 − Ye ) NB , (4.122)
2

and the plasmon mass in the case of a cold degenerate plasma

86 4 Particle Dispersion in External Active Media

 1/2  1/3
2α
mγ = 3 π 2 Ye NB (4.123)
π

into the threshold inequality (4.92), we come to the conclusion that for νμ , ντ the
radiative conversion process (4.86) is forbidden. For ν̄μ , ν̄τ the process is kinemati-
cally permitted for the energies greater than
 2/3
2 sin2 θW 3 Ye 2
mW
E > E0 = . (4.124)
1 − Ye π 1/3
NB

Using for estimation the values Ye  0.1, NB  1037 cm−3 , we obtain E0  2 × 104
GeV.

4.5.4.3 Hot Plasma of a Supernova Core

In this case one needs to use the general expression for the additional energy (4.107)
of the neutrino ν and antineutrino ν̄ with taking account of the scattering on all
plasma components. The additional energy can be written as:

ν ,ν̄
√ 1
ΔE = 2GF ∓ (Nn − N̄n ) ± (Nνe − N̄νe )
2
±(Nνμ − N̄νμ ) ± (Nντ − N̄ντ )

T3  ∗

+ 2 δ e B(±μe , mW , T ) − B(±μe , mW , −T )
2π

1 
+ B(±μ̃ν , mZ , T ) − B(±μ̃ν , mZ∗ , −T ) , (4.125)
2

where we use the notation

   ∞  
m2 μ/T dy y
B(μ, m, T ) = − Li2 −e +b ln 1 + .
ET exp (y − μ/T ) + 1 b
0
(4.126)
Here, Li2 (z) is the Euler dilogarithm, and b is the dimensionless parameter, b =
m2 /4ET . The complex masses are introduced in the functions B(±μe , mW ∗ , −T )
∗
and B(±μ̃ν , mZ , −T ) of Eq. (4.125) for proper taking account of imaginary parts,
similarly to Eqs. (4.109) and (4.120).
In the limit mW2 4ET , that is b 1, assuming that plasma is not degenerate
(μe ∼ T ), the integral in Eq. (4.126) can be represented as the series expansion that
can be calculated analytically:
4.5 Neutrino Self-energy Operator in Plasma 87

∞ 
dy y
ln 1 +
e−μ/T ey + 1 b
0
∞ ∞
1 ydy 1 y2 dy
= −μ/T
−
b e e + 1 2 b2
y e−μ/T ey + 1
0 0
∞
1 y3 dy
+ − ··· (4.127)
3 b3 e−μ/T ey +1
0

Taking into account that the arising Fermi integrals are expressed in terms of
polylogarithms:
∞  
yn dy
−μ/T
= −n! Lin+1 −eμ/T , (4.128)
e e +1
y
0

and using the recurrent connections between the polylogarithms Lin (x) and Lin (x −1 )
[65], one obtains the following expression:
  
√ (e) μe
ΔE νe = 2GF CV μ 2
+ π 2 2
T
3π 2 e

2 E 7π 4 4
− 2 2 μ4e + 2π 2 μ2e T 2 + T
3π mW 15

8 E 2 μe 10π 2 2 2 7π 4 4
+ 2 4 μ4e + μ T + T
5π mW 3 e 3
 
64 E 3 31 6 6
− 6
μe + 5π μe T + 7μe π T + π T + · · · . (4.129)
6 2 4 2 2 4 4
15π 2 mW 21

As it is illustrated in Fig. 4.13, taking account of only few terms in the series (4.129)
for the additional electron neutrino energy ΔE as a function of the initial neutrino
energy E, leads to an overestimation or understatement of the additional energy.
For a numerical estimation of the borders of the kinemetically possible region for
the SLν process in a general case with using of Eq. (4.125), let us take μe  160 MeV,
μ̃ν  μe /4  40 MeV, and T  30 MeV, see e.g. Refs. [66] and [67]. The analysis
displays that the process is forbidden for neutrinos of all flavors. For all types of
antineutrinos the effect becomes possible for energies greater than 2 × 104 GeV.
As in the considered cases of nonrelativistic cold plasma and of the neutron
star interior, for electron neutrinos and antineutrinos the processes of the W –boson
production on plasma electrons and positrons, νe + e+ → W + and ν̄e + e− → W − ,
are dominating. Using Eqs. (4.107), (4.110), one obtains the width of the process
in the conditions of a hot dense plasma, μe ∼ T me , for high neutrino energies,
E mW ΓW /μe :
88 4 Particle Dispersion in External Active Media

Fig. 4.13 Additional electron neutrino energy in the electron-positron medium (μe  160 MeV,
T  30 MeV) as an expansion into the series by initial neutrino energy: 0 is the local contribution;
1, 2 and 3—with consecutive adding of non-local terms ∼ E, ∼ E 2 and ∼ E 3 ; 4 is the exact function
(Figure reprinted from [46] with the World Scientific Publishing Company’s permission.)
 
−
4 T
GF mW
− 4μe E − mW
2
w(ν̄e + e → W ) = √ ln 1 + exp . (4.130)
2 2 π E2 4ET

Taking here the limit of cold plasma, T → 0, one readily comes to Eq. (4.121). The
width of the W + production by νe on positrons can be obtained from Eq. (4.130) by
the replacement μe → −μe .
Since in a dense plasma of the supernova core thermal neutrinos and antineutri-
nos of all flavors present, the processes of the Z–boson production should be also
considered for the sake of completeness. Using Eqs. (4.107), (4.110), one obtains
the width of the process where a high-energy antineutrino of the flavor scatters off
a thermal ν :
 
GF mZ4 T 4μ̃ν E − mZ2
w(ν̄ + ν → Z) = √ ln 1 + exp . (4.131)
4 2 π E2 4ET

The width of the process with a high-energy neutrino and a thermal antineutrino can
be obtained from Eq. (4.131) by the replacement μ̃ν → −μ̃ν . It should be noted
that in the supernova core conditions, μ̃ν  0 for = μ, τ .
Thus, the analysis of a possibility of the neutrino radiative conversion effect νL →
νR + γ (“spin light of neutrino”, SLν) based on the additional neutrino energy in
plasma in the case of ultra-high neutrino energies [59] should be performed only
with taking into account the dependence of the W and Z–boson propagators on the
4.5 Neutrino Self-energy Operator in Plasma 89

momentum transferred. It should be noted that the question about any observational
realization of the studied process requires a separate consideration. For high energy
neutrinos and antineutrinos, the processes of the W – and Z–boson production on
plasma, νe + e+ → W + , ν̄e + e− → W − and ν̄ + ν → Z, are dominating.

4.6 Neutrino Self-energy Operator in an External

Magnetic Field

4.6.1 Definition of the Operator Σ(p) in a Magnetic Field

As it was noted above, the analysis of the influence, along with plasma, of another
component of external active astrophysical environment, which is strong magnetic
field, onto the properties of neutrinos, in particular onto the neutrino oscillation
mechanism, is of considerable interest. However, this effect of the field could be
significant only in a case of its sufficiently high intensity. As already noted, there is a
natural scale of the magnetic field, called the critical value, Be = me2 /e  4.41×1013
gauss. There are arguments in favor of the field of such and larger scales to be
generated in astrophysical processes, such as supernova explosions and mergings of
neutron stars, which are characterized also by giant neutrino fluxes.
It should be noted that the study of the self-energy operator of a neutrino in a
magnetic field has a 30-year-old history [32–34, 36, 37, 68].
A general Lorentz structure of the self-energy neutrino operator Σ(p) in a mag-
netic field can be presented in a form similar to the expression (4.65), in terms of
linearly independent covariant structures:
    
Σ(p) = AL (pγ) + B̄L e2 pF̃ F̃γ + C¯L e pF̃γ γL
    
+ AR (pγ) + B̄R e2 pF̃ F̃γ + C¯R e pF̃γ γR
+ mν [K1 + i K2 e (γFγ)] . (4.132)

Similarly to Eq. (4.65), if the approximation is used of the massless left-handed

neutrino, only three terms with the coefficients AL , B̄L , and C¯L present in the operator
Σ(p).
The analysis shows that the results of calculations of the invariant coefficients B̄L ,
and C¯L in Eq. (4.132), obtained by different authors, are not consistent. In Table 4.1,
we give the values of these coefficients obtained in previous studies, and the results
of our calculations, which are discussed in detail below. The field B is called “weak”
at eB m2 and “moderate” at m2 eB mW2 .

Below, it will be demonstrated in detail that for massive neutrino the coefficient
C¯L defines its anomalous magnetic moment. As a validation of the calculations of
the coefficient C¯L , it should be its agreement with the known result for the neutrino
anomalous magnetic moment [69, 70]:
90 4 Particle Dispersion in External Active Media

Table 4.1 The coefficients in the formula (4.132) for the self-energy neutrino operator Σ(p) in an
external magnetic field. *)
√ 2 √ 2
2π 2π
Authors Field B̄L × ¯
CL ×
GF GF
McKeon [32] – 0 +3
mW2
1 3
Erdas et al. [34] Mod. − 2 ln 2 + 0
3mW m 4
1 1
Elizalde et al. [35] Mod. + −
2eB 2
1 −p2 /(2eB) 1
− e−p⊥ /(2eB)
2
Elizalde et al. [41] Mod. + e ⊥
4eB 4
1 m2 3 3
Our result [36] Weak − 2 ln W2 + +
3mW m 4 4
2
mW
1 3
Our result [36] Mod. − 2
ln + 2.542 +
3mW eB 4
*) It is indicated in Ref. [41] that their result is valid in the region of the neutrino momenta 0 <
p2⊥ eB. Our result is valid in the region 0 < p2⊥ mW 4 /β.

e mν C¯L 3e GF mν
μν  = √ . (4.133)
2 8π 2 2

The comparison shows that in Ref. [32] the coefficient C¯L was overstated by 4 times,
while in Refs. [35] and [41] it contains the extra factors: −2/3 and −1/3, respectively.
In addition, in Ref. [32], a non-zero value is declared for the coefficient at the structure
of the form (pFγ), defining the electric dipole moment of the neutrino. However,
this contribution to the neutrino self-energy operator can differ from zero only in
the presence of the electromagnetic field with a nonzero CP-odd field invariant
(F F̃) = 4 (EB). But even in this case, it is strongly suppressed (see [33]). One
should conclude that the result for the neutrino electric dipole moment obtained in
Ref. [32], where a purely magnetic field was considered, was erroneous.
The differences in the results for the coefficient of B̄L are the most significant. In
Ref. [32], it was not calculated as negligible. Computation of the B̄L , carried out in
Ref. [34], led to the magnitude scale GF /mW 2 . If compared with this value, the result

for B̄L obtained in Refs. [35, 41], has a huge amplification factor of mW 2 /eB. Being

correct, that result would lead to important consequences for the physics of neutrinos
in medium (see [71]) because the field contribution to the additional neutrino energy
would exceed the plasma contribution.
Earlier calculations of the plasma contribution to the operator Σ(p) both excluding
and including the magnetic field, were performed in a number of papers (see, e.g., [30,
38, 39]).
In Ref. [39], the neutrino dispersion properties were studied in the approximations
me T mW and B  T 2 for the sake of applying the results to the early Universe.
In particular, for a charge-symmetric plasma it is possible to extract from Ref. [39]
4.6 Neutrino Self-energy Operator in an External Magnetic Field 91

the difference of the neutrino self-energies which is the same for neutrinos and
antineutrinos,3 ΔE = Eνe − Eνi (i = μ, τ ), in the form:

GF T 4 GF T 2
ΔE (T , B)  −6.0 2
|p | + 0.47 2
e (B p ) , (4.134)
mW mW

where p is the momentum of a neutrino or antineutrino. The first term is the domi-
nating contribution of pure plasma, and the second term is caused by the collective
influence of the plasma and magnetic field.
The pure field contribution to the neutrino self-energy was not considered by the
authors [39] as insignificant. In contrast, the authors [35, 41] argue that just the
field contribution is dominant. The result of Ref. [41] for the pure field contribution
to the difference of the neutrino self-energies, which is the same for neutrinos and
antineutrinos, can be written as

GF eB
ΔE (B)  √ |p | sin2 φ , (4.135)
4 2π 2

where φ is the angle between B and p. A comparison of the formulas (4.134)

and (4.135) shows that the pure field contribution obtained by the authors [35, 41]
may significantly exceed the plasma contribution (4.134). Indeed, for the ratio of the
contributions one obtains
 
 ΔE (B)  2
  −3 eB mW
R=   1.5 × 10 , (4.136)
 ΔE (T , B)  T2 T2

where an averaging over the angle φ is performed in the value ΔE (B) , and only
the leading term is taken in the value ΔE (T , B) . Since the temperature during the
considered stage of the evolution of the Universe T mW , the ratio R can appear
significantly greater than one due to a large factor (mW /T )2 .
Thus, since the question was of fundamental importance, whether the contribution
of the external magnetic field into the neutrino energy was negligible or dominant, the
necessity of its independent calculation was obvious. This calculation was performed
in Ref. [36]. Here we reproduce the calculation of the neutrino self-energy operator
in a constant uniform magnetic field which is weaker than the critical field for a W
boson, eB mW 2 .

The S-matrix element for the transition ν→ν corresponds to the Feynman dia-
grams in Fig. 4.14.
Similarly to the procedure described in Sect. 4.5.2, the self-energy operator of a
neutrino in a magnetic field can be found to be

3 The sign ± at the linear in the field term in Eq. (13) of Ref. [39] came from poorly chosen
notations: the neutrino momentum in this article was k, while the antineutrino momentum was −k
(G. Raffelt, private communication).
92 4 Particle Dispersion in External Active Media

Fig. 4.14 The Feynman diagrams describing the magnetic field-induced contribution to the neutrino
self-energy operator in the Feynman gauge: the double lines correspond to the exact propagators of
the charged lepton, W -boson and the unphysical charged scalar Φ-boson in an external magnetic
field


i g2 α (W )
Σ(p) = − γ γL Jαβ (p) γ β γL (4.137)
2

1 (Φ)
+ 2
(m γR − m γ
ν L ) J (p) (m γL − mν R ,
γ )
mW

where g is the constant of the electroweak standard model. The integrals introduced
in Eq. (4.137), have the form

(W ) d4 q (W )
Jαβ (p) = S(q) Gβα (q − p) , (4.138)
(2π)4

(Φ) d4 q
J (p) = S(q) D(Φ) (q − p) , (4.139)
(2π)4
(W )
where S(q), Gβα (q−p) and D(Φ) (q−p) are the Fourier transforms of the translation-
ally invariant parts of the propagators for a charged lepton, W − -boson and charged
scalar Φ-boson respectively, see Eqs. (3.10), (3.13) and (3.14). We emphasize that
mν in Eq. (4.137) is generally a non-diagonal Dirac neutrino mass matrix caused by
the mixing in the lepton sector.
It should be noted that the coefficients AR , B̄R , C¯R , and K1,2 in Eq. (4.132)
originated from the Feynman diagram with a scalar Φ-boson and are suppressed by
the square of the ratio of the lepton mass to the mass of the W -boson, while the
coefficients AL , B̄L , and C¯L contain the contributions from both diagrams.
Further, we calculate the contribution to the neutrino self-energy operator from
the nth Landau level in the propagator of the charged lepton in combination with the
exact W -propagator. It is shown that the contribution of the ground Landau level is
not dominant and the higher levels give contributions of the same order, contrary to
the assumption used in Refs. [35, 41]. Then we present a detailed calculation of the
neutrino self-energy operator in a magnetic field in two limiting cases, of a relatively
weak field, eB m2 , and moderately strong field, m2 eB mW 2 . The additional

energy acquired by a neutrino in an external magnetic field is calculated, and possible

cosmological and astrophysical implications are analyzed.
4.6 Neutrino Self-energy Operator in an External Magnetic Field 93

4.6.2 Low Landau Level Contribution into the Operator Σ(p)

As we have already mentioned, our results for the coefficients of the neutrino
self-energy operator (4.132) strongly disagree with that of Refs. [35, 41]. We think
that the disagreement arises because these authors used only one lowest Landau level
contribution in the charged-lepton propagator in the case of moderate field strengths
which they call “strong fields.” However, the contributions of the next Landau levels
can be of the same order as the ground-level contribution because in the integration
over the virtual lepton four-momentum in the loop the region q2 ∼ mW 2 β appears
to be essential.
To substantiate this point we calculate the contribution to the neutrino self-energy
operator from the nth charged-lepton Landau level in conjunction with the exact
W -propagator in the limit p2⊥ /mW2 mW2 /β. Substituting the exact W -propagator

(3.13) and the nth Landau level contribution to the charged-lepton propagator from
Eq. (3.37) into Eq. (4.138) we find
   
(n) d4 q i i
Jσρ (p) =− (qγ) dn (v) − (γϕγ) dn (v)
(2π)4 q2 − m2 − 2nβ 2

dn (v)
− (qγ)⊥ 2n
v
 ∞   
ds tan(βs)
× exp −is mW2
− (q − p)2 + (q − p)2⊥
cos(βs) βs
0 
× (ϕ̃ϕ̃)ρσ − (ϕϕ)ρσ cos(2βs) − ϕρσ sin(2βs) . (4.140)

The terms with even numbers of γ matrices were omitted because they are removed
by the chiral structure of the operator Eq. (4.132). Next we perform a clockwise
rotation in the complex plane s = −iτ and use the identity
 ∞   
1
=− dτ exp −τ m2 + 2nβ − q2 . (4.141)
q − m2 − 2nβ
2
0

These manipulations allow us to rewrite the integral Eq. (4.140) as

    
(n) d4 q i dn (v)
Jσρ (p) = (qγ) dn (v) − (γϕγ) dn (v) − (qγ)⊥ 2n
(2π)4 2 v
 ∞  
dτ dτ
× (ϕ̃ϕ̃)ρσ − (ϕϕ)ρσ cosh(2βτ ) + i ϕρσ sinh(2βτ )
0 cosh(βτ )
    
× exp −τ m2 + 2nβ − q2 − τ mW 2
− (q − p)2

tanh(βτ )
− (q − p)⊥ .
2
(4.142)
β
94 4 Particle Dispersion in External Active Media

In the integration over d4 q = d2 q d2 q⊥ , the integrals over d2 q can be easily

calculated because they are of Gaussian form. As a result we find
 ∞  
(n) i dx dy xy
Jσρ (p) = exp −x + ξ − y (2nη + λ)
16π 3 mW2
0 (x + y) cosh(ηx) x+y
 
× (ϕ̃ϕ̃)ρσ − (ϕϕ)ρσ cosh(2ηx) + i ϕρσ sinh(2ηx)
  
tanh(ηx)
× d q⊥ exp −
2
(q − p)⊥
2
β
   
x i dn (v)
× (pγ) dn (v) − (γϕγ) dn (v) − (qγ)⊥ 2n , (4.143)
x+y 2 v

where the dimensionless variables x = mW 2 τ and y = m2 τ have been introduced

W
as well as the parameters η = β/mW , ξ = p2 /mW
2 2  p2 /m2 and λ = m2 /m2 .
⊥ W W
It follows from Eq. (4.143) that the essential region of the x variable is x ∼ 1 due
to the exponential e−x . Given the condition η 1, the argument of the hyperbolic
functions is small, ηx 1, leading to an obvious simplification. One should also take
into account the condition q⊥ 2 ∼ β caused by the functions d (v), see Eq. (3.38),
n
−v
containing the exponential e . For a wide range of the numbers n the exponential
in the integral over d2 q⊥ is simplified, with the only restriction n 1/η = mW2 /β:

 
tanh(ηx) p2
exp − (q − p)2⊥  exp −x ⊥
β 2
mW
2 − 2(qp)
q⊥ ⊥
× exp −x 2
. (4.144)
mW

Here, the first exponential is equal to e−ξx . We consider the value p2⊥ to vary in a very
wide range, 0 < p2⊥ mW 4 /β. The second exponential is equal to unity with a good

accuracy, because q⊥ 2 ∼β mW2 and (qp) mW2 . With these approximations, the
⊥
2
integration over d q⊥ can be easily performed,
 
d2 q⊥ dn (v) = π β (2 − δn0 ) , d2 q⊥ dn (v) = −π β δn0 , (4.145)

dn (v)
d2 q⊥ (qγ)⊥ = 0. (4.146)
v

Let us return to the expression (4.144) and make additional comments, to prevent
possible misunderstanding. At first glance it might seem that the replacement of the
value (p−q)2⊥ by p2⊥ , which actually occurred in the expression (4.144), meant that an
additional condition had been taken: p2⊥ 2 ∼ eB, which significantly narrowed
q⊥
the area of the values of p⊥ under consideration. We show by direct calculation that
this is not so and that the result is valid in the entire range 0 < p2⊥ mW4 /β.
4.6 Neutrino Self-energy Operator in an External Magnetic Field 95

Let us save the second exponential term in Eq. (4.144) and substitute it into the
first integral (4.146) denoting it as j(n) :
 2 2 − 2(qp)
q⊥ q⊥ ⊥
j(n) = d2 q⊥ dn exp −x . (4.147)
β 2
mW

Consider first the case n = 0, with d0 (v) = exp(−v). Making a transition in the plane
q⊥ to polar coordinates {q⊥ , φ}, while (qp)⊥ = q⊥ p⊥ cos φ, and using the known
integral
2π
dφ eb cos φ = 2π I0 (b) , (4.148)
0

where I0 (b) is the modified Bessel function of zero order, we obtain

∞  
(0) 1 x 2x q⊥ p⊥
j = 2π q⊥ dq⊥ exp −q⊥
2
+ 2 I0 . (4.149)
β mW mW2
0

We emphasize that no approximation has not been done yet. Using another well-
known integral
∞
√
dy e−y I0 (2z y) = ez ,
2
(4.150)
0

we finally obtain

πβ x 2 p2⊥ β
j(0) = exp . (4.151)
1 + xβ/mW
2 mW (1 + xβ/mW
4 2 )

Recalling that x  1, β mW 2 and p2

⊥ mW 4 /β, we exactly reproduce from (4.151)

the first integral of Eq. (4.145) with n = 0:

j(0)  πβ . (4.152)

The similar calculation for n = 1, when d1 (v) = 2v exp(−v), leads to one more
well-known integral:

∞
√
y dy e−y I0 (2z y) = (1 + z2 ) ez .
2
(4.153)
0
96 4 Particle Dispersion in External Active Media

In the approximation used, this gives

j(1)  2πβ . (4.154)

The similar analysis can be performed for any n. Thus, the approximation used in
the calculation of the integrals (4.145)–(4.146), is justified.
Given the relations (4.145)–(4.146), the integral (4.143) acquires the form
   
(n) iβ i
Jσρ (p) = 2
(pγ) gρσ 2 − 1 − (γϕγ) δn0
16π 2 mW 2
 ∞  
x dx dy x2
× exp −x − ξ − y (2nη + λ) . (4.155)
0 (x + y)2 x+y

Taking into account the smallness of the parameters η and λ, one finally obtains for
n mW 2 /β

   
(n) iβ p2⊥ i
Jσρ (p) = ln 1 + (pγ) gρσ 2 − 1 − (γϕγ) δn0 . (4.156)
16π 2 p2⊥ mW2 2

Substituting Eq. (4.156) into Eq. (4.137) we finally find the contribution of the nth
Landau level of the lepton propagator to the neutrino self-energy operator

GF eB mW 2
p2⊥  
Σ (n) (p) = − √ ln 1 + (2 − δn0 ) (pγ) − δn0 (pϕ̃γ) γL .
2 2π 2 p2⊥ mW2

(4.157)
We conclude from Eq. (4.157) that, contrary to the treatment of Refs. [35, 41], the
lowest Landau level does not dominate.
2 /β, the calculation is more cumbersome. There-
For higher Landau levels, n  mW
fore, using the lepton propagator expanded in terms of the Landau levels, with a
further summation, is extremely inconvenient. It is much simpler to take the exact
lepton propagator in the form of Eq. (3.10). This approach is used in Sect. 4.6.4
below.

4.6.3 Calculation of the Operator Σ(p) in a “Weak” Field

Because of the discrepancy of our results with the results of Refs. [35, 41], we
present here our calculations of the operator Σ(p) in detail. We start the analysis
with the simpler case of a relatively weak field, when the field strength is the smallest
dimensional parameter of the problem, eB m2 mW 2 . In this case, for the Fourier

transforms of the propagators both the W -boson and lepton one can use the field
decompositions (3.22) and (3.24) and evaluate the integral (4.138) as a series in
powers of β/mW 2 :
4.6 Neutrino Self-energy Operator in an External Magnetic Field 97

d4 q  (0) (1) (2)

Jαβ (p) = S (q) + S (q) + S (q) + · · ·
(2π)4
 
× G(0)
βα (q − p) + G(1)
βα (q − p) + G(2)
βα (q − p) + · · ·
= Δ0 Jαβ (p) + Δ1 Jαβ (p) + Δ2 Jαβ (p) + · · · (4.158)

It is easy to show that the fieldless term Δ0 Jαβ (p) containing an ultraviolet diver-
gence, has the structure of gαβ (pγ) and contributes only to the coefficient AL of the
operator Σ(p), see (4.132), which is absorbed to the renormalization of the neutrino
wave function.
The first-order term consists of two parts:

( W ) ( W )
Δ1 Jαβ (p) = Jαβ0 1 (p) + Jαβ1 0 (p) (4.159)
 4  4
d q (0) (1) d q (1) (0)
= S (q) Gβα (q − p) + S (q) Gβα (q − p) .
(2π) 4 (2π)4

The part containing the zero-order term of the lepton propagator and the first-order
term of the W -propagator has the form

( W ) d4 q (qγ) + m 1
Jαβ0 1 (p) = 2i β ϕαβ . (4.160)
(2π)4 q2 − m2 [(q − p)2 − mW
2 ]2

Due to the chiral structure of the operator (4.137), only the terms with an odd number
of γ-matrices should be taken. Using the expansion

1 1 2n (qp)
 2 + 2 , (4.161)
[(q − p)2 − mW
2 ]n (q − mW
2 )n (q − mW 2 )n+1

where we have neglected the neutrino mass p2 = mν2 , we obtain in the approximation
m2 mW 2 :

( W ) 1 β
Jαβ0 1 (p)  2 2
ϕαβ (pγ) . (4.162)
16π mW

The part containing the first-order term of the lepton propagator and the zero-order
term of the W -propagator is calculated similarly:

( 1 W0 ) i d4 q (qγ) + m 1
Jαβ (p) = − β gαβ (γϕγ)
2 (2π)4 (q2 − m2 )2 (q − p)2 − mW
2

i β
− 2
g (pϕ̃γ) γ5 ,
2 αβ
(4.163)
32π mW

where we use the identity

98 4 Particle Dispersion in External Active Media

(pγ) (γϕγ) = 2i (pϕ̃γ) γ5 . (4.164)

Contribution of the second order into the integral Jαβ (p) consists of three parts:

( W ) ( W ) ( W )
Δ2 Jαβ (p) = Jαβ0 2 (p) + Jαβ1 1 (p) + Jαβ2 0 (p)
 
d4 q (0) (2) d4 q (1) (1)
= S (q) Gβα (q − p) + S (q) Gβα (q − p)
(2π)4 (2π)4

d4 q (2) (0)
+ S (q) Gβα (q − p) . (4.165)
(2π)4

In the approximation considered, as well as for p2⊥  p2 mW 2 , using the expan-

sion (4.161), one can obtain the part containing the zero-order term of the lepton
propagator and the second-order term of the W -propagator in the form
  
( 0 W2 ) d4 q (qγ) + m 1
Jαβ (p) = −β 2 gαβ
(2π)4 q2 − m2 [(q − p)2− mW
2 ]3

2 (q − p)2⊥ 1
+ + 4 (ϕϕ)αβ
[(q − p)2 − mW
2 ]4 [(q
− mW
2 ]3 − p)2
2 
i β 1
 2
gαβ (pγ) − (ϕϕ)αβ (pγ) + · · · , (4.166)
16π 2 mW 18

where dots mean the term of the form gαβ (pγ), which contributes only to the coef-
ficient AL .
The part containing the first-order term of the lepton propagator and the first-order
term of the W -propagator is found to be

( W1 ) d4 q (qγ) + m 1
Jαβ1 (p) = β 2 ϕαβ (γϕγ)
(2π)4 (q2 − m2 )2 [(q − p)2 − mW
2 ]2

2
1 β
− 2
ϕαβ (pϕ̃γ) γ5 . (4.167)
16π 2 mW

The combination of the second-order term of the lepton propagator and the zero-order
term of the W -propagator is

d4 q (q − m ) (qγ)⊥ − q⊥ [(qγ) + m ]
2 2 2
( 2 W0 )
Jαβ (p) = 2 β 2 gαβ
(2π)4 (q2 − m2 )4
1
× . (4.168)
(q − p)2 − mW
2
4.6 Neutrino Self-energy Operator in an External Magnetic Field 99

As it was already noted, this part contains the increased contribution of the region of
the relatively small virtual momenta, q2 ∼ m2 mW 2 . Making the Wick rotation in

the complex plane q0 , q0 = iq4 , q2 = q02 − q32 = −(q32 + q42 ), and after the standard
transformations we rewrite the integral (4.168) as

2
( W ) i β
Jαβ2 0 (p) = 2
gαβ (pγ) (2 I1 + I2 ) . (4.169)
8π 2 mW

Here, the following integrals are introduced, using the notations x = −q2 /mW
2 ,

y = q⊥ /mW , λ = m /mW 1:
2 2 2 2

∞
xdx ydy
I1 = ,
(x + y + λ)4 (x + y + 1)2
0
∞
dx ydy
I2 = λ , (4.170)
(x + y + λ)4 (x + y + 1)2
0

which, due to the smallness of the parameter λ, can be easily calculated:

1 17 1
I1  − ln λ − , I2  . (4.171)
6 36 6
Finally, for the contribution (4.168), we obtain

2
( W ) i β 2
mW 7
Jαβ2 0 (p)  2
gαβ (pγ) ln − . (4.172)
24π 2 mW m2 3

Collecting the calculated contributions to the expression (4.158), we find the

integral Jαβ (p) as the following expansion in powers of the field strength:
  
1 eB i
Jαβ (p)  2
− gαβ (pϕ̃γ)γ5 + ϕαβ (pγ) (4.173)
16π 2 mW 2
2 ⎫
eB 2 mW 2
3 ⎬
+i g αβ (pγ)  ln − + iϕ αβ (pϕ̃γ)γ5 − (ϕϕ) αβ (pγ) .
mW2 3 m2 2 ⎭

Here, the terms are omitted which have the structure of gαβ (pγ) and are totally
absorbed by the renormalization of the neutrino wave function, as well as the terms
of the even- number of γ matrices, which are removed due to the chiral structure of
the operator (4.137).
100 4 Particle Dispersion in External Active Media

Substituting the result (4.173) to Eq. (4.137), we finally obtain the neutrino
self-energy operator in a relatively weak field in the form:
 
2
GF eB eB 4 mW
Σ(p) = √ 3(pϕ̃γ) − 2 ln 2 + 1 (pγ) γL . (4.174)
2 4π 2 mW 3 m

Comparing (4.174) with Eqs. (4.132) and (4.133), one can conclude that the
coefficient C¯L is in agreement with the well-known results for the neutrino anom-
alous magnetic moment [69, 70], and with the results of Refs. [72, 73], where the
non-diagonal transitions νi ↔ νj (i = j) in an external electromagnetic field were
investigated.
In turn, the coefficient B̄L coincides with the result of Ref. [34], but not in the
case of a moderately strong field eB mW 2 , as stated in that paper, but only in the

considered weak field limit.

There is another criterion for the correctness of the presented calculation of the
coefficient B̄L . Really, an effective Lagrangian of the ννγγ interaction constructed on
the basis of the corresponding term of the amplitude (4.63) with (4.132) and (4.174),
and with the replacement of the external field by the photon field operators, is in
agreement, to the definitions, with the result of Ref. [74].

4.6.4 The Case of a Moderately Strong Field

In the case of a moderately strong field, m2 eB mW 2 , as was noted above, the

expansion of the lepton propagator in powers of the field (3.24) is inapplicable.

Using the exact expression (3.10) and the expansion (3.22) for the propagator of the
W -boson, we represent the integral Jαβ (p) as
  
d4 q (0) (1) (2)
Jαβ (p) = S(q) Gβα (q − p) + Gβα (q − p) + Gβα (q − p) + · · ·
(2π)4
( W0 ) ( W1 ) ( W2 )
= JαβE (p) + JαβE (p) + JαβE (p) + · · · (4.175)

For the first of the integrals (4.175) we get


( W0 ) d4 q 1
JαβE (p) = −i gαβ
(2π)4 (q − p)2 − mW2

∞   
ds tan(βs) 2
× exp −is m2 − q2 + q⊥ (4.176)
cos(βs) βs
0
   
1 (qγ)⊥
× [(qγ) + m ] cos(βs) − (γϕγ) sin(βs) − .
2 cos(βs)
4.6 Neutrino Self-energy Operator in an External Magnetic Field 101

Given the decomposition of the W -propagator (4.161), making a rotation in the

complex plane s, s = −iτ , and omitting the terms with an even number of the γ
matrices, we obtain

 ∞  
2
( W ) d4 q q⊥
JαβE 0 (p) = −2gαβ dτ exp −τ (m 2
− q2 ) − tanh(βτ )
(2π)4 β
0
   
i (qγ)⊥ (qp) − (qp)⊥
× (qγ) 1 + (γϕγ) tanh(βτ ) − . (4.177)
2 cosh (βτ )
2 (q2 − mW2 )2

Making the Wick rotation in the complex plane q0 , q0 = iq4 , integrating over the
angles in the Euclidean planes {q1 , q2 } and {q3 , q4 }, passing to the dimensionless
variables u = −q2 /mW2 , v = q2 /β, x = m2 τ and introducing the dimensionless
⊥ W
small parameter η = β/mW 2 1, we can rewrite the integral (4.177) in the form

∞ ∞ ∞
( W ) i gαβ e−ux−v tanh(ηx)
JαβE 0 (p) = η dx du dv (4.178)
16π 2 (1 + u + ηv)2
0 0 0
   
i ηv
× (pγ) u 1 + (γϕγ) tanh(ηx) − (pγ)⊥ .
2 cosh2 (ηx)

The integral over x requires a careful handling both at the lower and upper limits.
Using the smallness of the parameter η, it is advisable to choose the intermediate
scale A for the x variable, such that A 1, but ηA 1. The region of integration
over x is then divided into two parts, 0 < x < A and A < x < ∞:

( W0 ) (0A) (A∞)
JαβE (p) = Jαβ (p) + Jαβ (p) . (4.179)

In the region 0 < x < A, the argument of the hyperbolic functions is small, ηx 1,
and the first of the integrals (4.179) is essentially simplified with the change of the
variable ηv = w:

A ∞ 
(0A) i gαβ du dw 1
Jαβ (p) = dx 1 + η 2 x 3 w e−x(u+w)
16π 2 (1 + u + w)2 3
0 0
    
i
× (pγ) u 1 + (γϕγ) ηx − (pγ)⊥ w 1 − η 2 x 2 . (4.180)
2

Passing from the variables {u, w} to the new variables {z, ξ}:

∞ ∞ 1
1+ξ 1−ξ 1
u=z , w=z , du dw = z dz dξ , (4.181)
2 2 2
0 0 −1
102 4 Particle Dispersion in External Active Media

we can integrate over ξ. Omitting as before the terms of the form gαβ (pγ), let us
rewrite the integral (4.180) with the identity (4.164) in account, as
  
(0A) i gαβ 1
Jαβ (p) = − (pϕ̃γ) γ5 η I3 + (pγ) η I4 − I5 2
. (4.182)
32π 2 9

Here, the following integrals are introduced:

A ∞ A ∞
z2 dz −xz z2 dz −xz
I3 = x dx e , I4 = 2
x dx e ,
(1 + z)2 (1 + z)2
0 0 0 0
A ∞
z3 dz −xz
I5 = x 3 dx e . (4.183)
(1 + z)2
0 0

Given that A is the large parameter, we obtain up to the terms of O(1/A):

I3 = 1 , I4 = 2 ln A − 5 + 2 γE , I5 = 6 ln A − 17 + 6 γE , (4.184)

where γE = 0.577 . . . is the Euler constant. As a result, we have for the integral
(0A)
Jαβ (p):
  
(0A) i gαβ 4 7
Jαβ (p) = − (pϕ̃γ) γ5 η + (pγ) η 2 ln A − + γE . (4.185)
32π 2 3 3

The second of the integrals (4.179) can also be simplified. As one can see from
Eq. (4.178), the exponential in the numerator provides for A < x < ∞ that the region
of integration is only significant where the terms u and ηv in the denominator are
small if compared with unity. It is worthwhile to move to the new variables z = ηx,
y = u/η to obtain

∞ ∞ ∞
(A∞) i gαβ 2
Jαβ (p) = η dz dy dv e−yz e−v tanh z
16π 2
ηA 0 0
   
i v
× (pγ) y 1 + (γϕγ) tanh z − (pγ)⊥
2 cosh2 z
1
× . (4.186)
(1 + ηy + ηv)2

Replacing the last fraction by 1, we see that the integrals over y and v are easily
calculated. Neglecting the term O(1/A), we get
4.6 Neutrino Self-energy Operator in an External Magnetic Field 103
⎧ ⎫
⎪
⎨ ∞ ∞ ⎪
⎬
(A∞) i gαβ 2 dz dz
Jαβ (p) = η (pγ) − (pγ)⊥ . (4.187)
16π 2 ⎪ ⎩ z2 tanh z z sinh2 z ⎪
⎭
ηA ηA

Here, the first integral can be converted to the second one using the integration by
parts:

∞ ∞  
dz dz 1 1
=− + + + O (ηA) 2
. (4.188)
2
z tanh z z sinh2 z (ηA)2 3
ηA ηA

Given that ηA 1, the remaining integral can be rewritten as

∞ ∞  ∞
dz dz z2 3 dz
= − +3 . (4.189)
z sinh2 z z3 sinh z 3 + z2
2 z3 (3 + z2 )
ηA ηA ηA

Here, the added and subtracted term is chosen in such a way that, on the one hand, it
provided a convergence of the first integral at both the lower and upper limits, and on
the other hand, it was easily calculable. So, the first of the integrals (4.189) is finite,
if we tend the lower limit to zero. Its numerical value is
∞ 
dz z2 3
C= −  −0.055 . (4.190)
z3 sinh z 3 + z2
2
0

For the integral (4.189) we obtain

∞
dz 1 1 1 1
2
= 2
+ ln A + ln η − ln 3 + C (4.191)
z sinh z 2(ηA) 3 3 6
ηA

up to terms of higher order. Collecting the calculated contributions and omitting, as

before, the terms of the form gαβ (pγ), for the integral in the region A < x < ∞ we
have

(A∞) i gαβ 1 1 1
Jαβ (p) = (pγ) η − ln A + ln + + ln 3 − 3C .
2
(4.192)
24π 2 η 2 2

The final expression for the integral (4.179), as expected, does not contain the
intermediate scale A:
104 4 Particle Dispersion in External Active Media

( E W0 ) i gαβ
Jαβ (p) = − (pϕ̃γ) γ5 η
32π 2
 
4 1 11 1
+ (pγ) η 2 ln − + ln 3 + γE − 3C . (4.193)
3 η 6 2

The presence of the term η 2 ln η (η = β/mW 2 ) shows again that the expansion of the

lepton propagator over β as the small parameter is impossible in this case.

The similar analysis of the second and the third terms of the expansion (4.175)
shows that the “dangerous” contribution with the logarithm of β appears here in the
next orders of the η parameter, so it is possible to use the field expansion of the lepton
propagator, writing the integrals as

( E W1 ) ( 0 W1 ) ( 1 W1 )
Jαβ (p) = Jαβ (p) + Jαβ (p) ,
( W2 ) ( W2 )
JαβE (p) = Jαβ0 (p) . (4.194)

Summing (4.193), (4.162), (4.166), and (4.167) we find

  
1 eB i
Jαβ (p) = 2
− g αβ (p ϕ̃γ)γ5 + ϕαβ (pγ) 
16π 2 mW 2
2 2
eB 2 mW 7 1 2
+i 2
g αβ (pγ)  ln − + ln 3 + γE − 2C
mW 3 eB 6 3 3

+i ϕαβ (pϕ̃γ)γ5 − (ϕϕ)αβ (pγ) . (4.195)

For the neutrino self-energy operator in the case of a moderately strong field,
m2 eB mW 2 , we finally obtain

 
2
GF eB eB 4 mW
Σ(p) = √ 3(pϕ̃γ) − 2 ln + 3.389 (pγ) γL . (4.196)
2 4π 2 mW 3 eB

At first glance, the second terms in Eqs. (4.174) and (4.196), which contain the
small extra factor eB/mW 2 , can be neglected. However, as we will show below, just

these terms give the dominant contribution to the neutrino additional energy in an
external magnetic field.
In Ref. [35], the authors made an attempt to test the correctness of their analytical
calculations by producing a numerical evaluation of the coefficients B̄L and C¯L of the
operator Σ(p), being written in the form of a double integrals (see Eqs. (89) and (90)
of Ref. [35]). As we show below, this numerical calculation is also incorrect. The main
reason for the error is, probably, in the attempt of the authors to calculate numerically
the integral of the difference between the two, in fact, infinite quantities. The analysis
4.6 Neutrino Self-energy Operator in an External Magnetic Field 105

shows that the integral is finite and has an order of the value eB/mW 2  10−6 for the

corresponding field strength, but not of the order of unity, as the authors [35] claim.
Similarly to Ref. [35], let us represent the expressions for the coefficients of the
operator Σ(p) in the form of double integrals. We substitute the exact expressions
for the propagators (3.10) and (3.13), where it is convenient to make a turn in the
complex plane s, s = −iτ , in the integral (4.138). In this case, the integrals over
the 4-momentum d4 q = d2 q d2 q⊥ can be rather easily calculated. Substituting the
result into Eq. (4.137) and comparing it with the definition of the self-energy operator,
Eq. (4.132), one can present the coefficients AL , B̄L , and C¯L as follows:

∞
g2 η dx dy sinh(ηx)
AL = − exp[−Φ(x, y, λ, p, mW )] , (4.197)
16π 2 (x + y) sinh2 [η(x + y)]
0
∞  
g2 η dx dy sinh(ηx) x cosh[η(2x + y)]
B̄L = −
16π 2 (x + y) sinh[η(x + y)] sinh[η(x + y)] x+y
0
× exp[−Φ(x, y, λ, p, mW )] , (4.198)
2 η  x dx dy sinh[η(2x + y)]
∞
g
C¯L = exp[−Φ(x, y, λ, p, mW )] , (4.199)
16π 2 (x + y)2 sinh[η(x + y)]
0

where

Φ(x, y, λ, p, mW ) = x + λy

x y p2 xy sinh(ηx) sinh(ηy) p2⊥
− − − , (4.200)
x + y mW 2 x+y η sinh[η(x + y)] 2
mW

and the notation are also introduced: η = β/mW 2 , λ = m2 /m2 .

W
It is easy to see that the integral for the coefficient AL is divergent. As has been
noted, this coefficient is absorbed by the renormalization of the neutrino wave func-
tion.
We note that the expressions for the coefficients of B̄L and C¯L are in agreement
with Eqs. (89) and (90) of Ref. [35] up to an obvious error in the sign of Eq. (90).
However, as one can see from Eq. (4.198), the coefficient B̄L is an even function of
the η parameter, therefore, the linear dependence of B̄L on η declared in Ref. [35] is
an obvious error.
To verify the correctness of our analytical calculations let us consider the limiting
case: m2 mW 2 and p2  p2 mW2 . Moving to a new variable z = x + y, we can
 ⊥
simplify the integrals (4.198) and (4.199) as:
106 4 Particle Dispersion in External Active Media

∞ z  
g2 η dz −x sinh(ηx) x cosh[η(z + x)]
B̄L = dx e − ,
16π 2 z sinh(ηz) sinh(ηz) z
0 0
(4.201)
∞ z
g2 η dz
C¯L = xdx e−x sinh[η(z + x)] . (4.202)
16π 2 z2 sinh(ηz)
0 0

The results of numerical calculation of B̄L and C¯L as the functions of the η pa-
rameter demonstrate a good agreement with the previous approximate formulas,
especially for small values of η.

4.6.5 The Neutrino Operator Σ(p) in a Crossed Field

In addition to the limiting cases of a weak (eB m2 ) and a moderately strong

(m2 eB mW 2 ) field, which were considered in Ref. [36], there is yet another

region of values of the physical parameters that requires a dedicated analysis. We

mean here the situation where the neutrino transverse momentum p⊥ with respect to
the magnetic field is rather high — for example, p⊥  mW or p⊥ mW . This region
of parameter values is of importance in connection with problems of the physics
of ultrahigh-energy cosmic rays. In particular, the possibility of detecting cosmic
neutrinos of ultrahigh energy (Eν ∼ 107÷17 GeV) is widely discussed (see, for ex-
ample, Ref. [75] and references therein). Apparently, the propagation of neutrinos
having such energies cannot be described adequately without taking into account
their interaction with magnetic fields of astrophysical nature.
The above region of parameter values corresponds to the crossed-field approxima-
tion, where the Fourier transforms of the translation-invariant parts of the propagators
from expressions (4.138) and (4.139), are presented in Eqs. (3.67), (3.68), and (3.69).
The general Lorentz structure of the operator Σ(p) in the presence of a magnetic
field is represented in Eq. (4.132).
In the crossed field approximation, the coefficients in Eq. (4.132) we are interested
in, are expressed in terms of the integrals containing the Hardy—Stokes function f (u),
see Eq. (4.36), and its derivative:

1
GF dv v [2(1 + v)(2 + v) + λ (1 − v)(2 − v)]
B̄L = √
12 2 π 2 mW
2 [v + λ (1 − v)]2
0
df (u)
× u2 , (4.203)
du
4.6 Neutrino Self-energy Operator in an External Magnetic Field 107

1
GF dv v [2(1 + v) − λ (1 − v)]
C¯L = √ u f (u) , (4.204)
4 2 π2 v + λ (1 − v)
0
1
GF mν2 dv v (1 − v)(2 − v) 2 df (u)
B̄R = √ u , (4.205)
12 2 π 2 mW
2 2
mW [v + λ (1 − v)]2 du
0
1
GF mν2 dv v (1 − v)
C¯R = √ u f (u) , (4.206)
2
4 2 π 2 mW v + λ (1 − v)
0
1
GF λ dv (1 − v)
K2 = √ u f (u) , (4.207)
8 2 π2 v + λ (1 − v)
0

where λ = m2 /mW
2 . The argument of the function f (u) in Eqs. (4.203)–(4.207) has

the form
v + λ (1 − v)
u= , (4.208)
[χ v(1 − v)]2/3

where χ is the dynamical field parameter, χ2 = e2 (pFFp)/mW 6 .

For the dynamical parameter χ, there are three regions of values where one can
obtain simple approximate analytic expressions for the integrals in Eqs. (4.203)–
(4.207).

(i) Region where χ is the smallest parameter in the problem, χ2 λ, or eB p⊥ 2 .

m mW
In this region, we have
 √ √ 
GF 1 3 3λ 3λ
B̄L − √ ln + + iπ exp − , (4.209)
3 2 π 2 mW2 λ 4 χ χ
  
3GF 2 1 1 4 2 1
C¯L  √ 1 − λ ln − + χ ln − 3
4 2π 2 3 λ 4 3 λ
√ 
4πλ 3λ
+i exp − , (4.210)
3 χ

GF mν2 1 9
B̄R − √ ln − , (4.211)
6 2 π 2 mW2 m2
W
λ 4
GF mν2
C¯R  √ 2
, (4.212)
8 2 π2 mW
  
GF 1 2
K2  √ λ ln − 1 + χ2 . (4.213)
8 2 π2 λ 3
108 4 Particle Dispersion in External Active Media

(ii) Region of intermediate values of the dynamical parameter, λ χ2 1. This

region is likely to be of greatest interest. We recall that λ = me2 /mW
2  4×

10−11 . Representing the parameter χ in the form

 2 2
B E
χ2  3 × 10−3 , (4.214)
Be 1020 eV

we can see that, over very broad ranges of magnetic-field strengths and neutrino
energies, the parameter χ falls within this intermediate region. Here, we have

GF 1 5
B̄L  − √ − + ln 3 + 2 γE + iπ ,
2 ln (4.215)
3 2 π 2 mW
2 χ 4
  
3GF 4 2 1 17
C¯L  √ 1+ χ 2 ln − + ln 3 + 2 γE + iπ , (4.216)
4 2 π2 3 χ 3

GF mν2 1 17
B̄R  − √ 2 ln − + ln 3 + 2 γE + iπ , (4.217)
6 2 π 2 mW
2 mW2 χ 4
GF mν2
C¯R  √ 2
, (4.218)
12 2 π 2 mW

GF m2 1
K2  √ 2 ln − 1 + ln 3 + 2 γE + iπ . (4.219)
8 2π 2 mW
2 χ

(iii) Region of large values of the dynamical parameter χ 1. Our results in this
region are:
√
3 GF
B̄L  −i √ 2 χ
, (4.220)
2 2 π mW
 √ 
π GF 1 + i 3
C¯L  √ , (4.221)
5 2 31/6 Γ 4 (2/3) χ2/3
 √ 
37/6 Γ 4 (2/3) GF 1 − i 3 m2
ν
B̄R  √ 4
, (4.222)
32 2 π 3 χ4/3 mW
 √ 
π GF 1 + i 3 mν2
C¯R  √ 2
, (4.223)
90 2 31/6 Γ 4 (2/3) χ2/3 mW
 √ 
π GF 1 + i 3 m2
K2  √ 2
, (4.224)
36 2 31/6 Γ 4 (2/3) χ2/3 mW

where Γ (x) is a gamma function, Γ (2/3) = 1.354...

4.6 Neutrino Self-energy Operator in an External Magnetic Field 109

4.6.6 Field-Induced Neutrino Magnetic Moment

The field-induced correction to the magnetic moment μν of the neutrino ν is yet

another quantity that can be extracted from the neutrino self-energy operator. The
anomalous neutrino magnetic moment is expressed in terms of the coefficients in the
self-energy operator (4.132), for the details see Sect. 4.7.4 below, as
e mν  ¯   
μν = CL − C¯R − B̄L − B̄R e(Bv) + 4 K2 , (4.225)
2
where v = p/E is the neutrino velocity.
In the limiting case of χ2 λ = m2 /mW 2 , the neutrino magnetic moment becomes

  
1 4 1 1
μν  μ(0)
ν 1 − λ + χ2 ln − 3 + , (4.226)
2 3 λ 3
3e GF mν
μ(0)
ν = √ ,
8π 2 2

where μ(0)
ν is the neutrino magnetic moment in a vacuum [69, 70]. In the field-induced
corrections in Eq. (4.226), the leading term of order ∼χ2 , which involves a large
logarithm, coincides with the result presented in Ref. [33], where the postlogarithmic
terms were disregarded. The last term in the field-induced correction in Eq. (4.226)
originates from the Φ–boson contribution. One can see that it is relatively small but
does not involve a parametric suppression.

4.7 Neutrino Self-energy Operator in Magnetized Plasma

One of the topical questions of the physics of elementary particles in an external

medium is the question of how the external active medium influences the neutrino
dispersion properties.
An analysis of the neutrino self-energy operator Σ(p) in a magnetized plasma,
defining, in particular, the neutrino dispersion law, was performed in a number of
papers (see, e.g., [29, 39–41]).
In Refs. [29, 39, 40], the neutrino dispersion was investigated in a charge-
symmetric, weakly magnetized plasma under physical conditions
2
mW T2 eB me2 . (4.227)

The additional energy of electron neutrinos was written in the form

110 4 Particle Dispersion in External Active Media
√
ΔE 7 2 GF π 2 T 4 1 2
=− + 2
|p| 45 mZ2 mW
√ √ 
2 GF T 2 eB 2 GF (eB)2 T
+ cos φ + ln sin2 φ. (4.228)
3mW2 3 π mW
2 2 me

Here, E and p are the neutrino energy and momentum, respectively; T is the plasma
temperature; and φ is the angle between the direction of the magnetic field B and the
momentum vector p. The first term in Eq. (4.228) describes the additional neutrino
energy in a plasma without a magnetic field [29], while the second [39] and the
third [40] terms are attributable to the simultaneous presence of a plasma and a
magnetic field. As we see from Eq. (4.228), the term proportional to the square of the
magnetic field strength contains amplification by the logarithmic factor ln(T /me ),
which, in general, raises doubts under the indicated physical conditions (4.227).
Indeed, under such conditions the contribution to the neutrino energy is determined
by the plasma electrons and positrons that populate the highest Landau levels. The
energy of these electrons and positrons at the nth Landau level is given by the formula

ωn = me2 + k32 + 2eBn, n 1. (4.229)

Since the electron mass under the presumed conditions is the smallest parameter of
the problem, it can be neglected in Eq. (4.229) for the energy. Therefore, it is unlikely
that the electron mass could present in the final result in the principal approximation.
Thus, an independent calculation of the neutrino dispersion in a magnetized plasma
was of considerable interest.
In this section, we present, following the papers [76, 77], the results of our analysis
of the charge-symmetric magnetized plasma influence on the neutrino dispersion in
the presence of an external magnetic field. A general expression for the neutrino
self-energy operator Σ(p) is derived. The neutrino dispersion under the physical
conditions of weakly, moderately, and strongly magnetized plasmas is analyzed in
detail.

4.7.1 Neutrino Scattering on Magnetized Plasma

Similarly to consideration performed in Sect. 4.5.1, an expression for the neutrino

self-energy operator Σ(p) is defined via the amplitude of the neutrino forward scat-
tering (4.63). The additional neutrino energy due to the neutrino forward scattering
in the medium is expressed in terms of the amplitude of this process,

1 1
ΔE = − M(ν → ν) = Tr {((pγ) + mν ) (1 − (sγ) γ5 ) Σ(p)} , (4.230)
2E 4E
4.7 Neutrino Self-energy Operator in Magnetized Plasma 111

where E = p2 + mν2 is the neutrino energy in a vacuum, mν is the neutrino

mass, and sμ is the neutrino doubled spin 4-vector.
A detailed description of calculation of the neutrino self-energy operator Σ(p)
in a magnetized plasma can be found, for example, in Ref. [77]. The operator Σ(p)
is represented there in a general form of an expansion over the linearly independent
covariant structures:
 
Σ(p) = AL (pγ) + BL (uγ) + CL e (pF̃γ) γL
 
+ AR (pγ) + BR (uγ) + CR e (pF̃γ) γR
+ mν [K1 + i K2 e (γFγ)] , (4.231)

where uα is the four-vector of the plasma velocity. Comparing this formula with
Eq. (4.132) for Σ(p) in a magnetic field, one can see that a replacement is made here
of the structure (pF̃ F̃γ) to the structure (uγ). Such a replacement is possible, due to
the relation
(pu)(pF̃ F̃γ) = (pF̃ F̃p)(uγ) + (pF̃u)(pF̃γ). (4.232)

This equality holds if the spatial part of the plasma velocity four-vector uα
is directed along the magnetic field. It should be kept in mind that under the
term “magnetized plasma” we mean a situation where in the plasma rest frame,
uα = (1, 0), the electromagnetic field is reduced to a purely magnetic. The covari-
ance of the operator Σ(p) means in this case that there are many reference frames
moving parallel to the magnetic field in which the operator (4.231) retains its shape.
Given the relation (4.232), one can connect the coefficients of Eqs. (4.132)
and (4.231) as follows:

e2 (pF̃ F̃p) e(pF̃u)

BL = B̄L , CL = C¯L + B̄L , (4.233)
(pu) (pu)

and similarly for BR , CR coefficients.

Using Eqs. (4.231) and (4.230), one can represent the additional neutrino energy
ΔE in a magnetized plasma in the form

1 1
ΔE = BL [1 − (sv)] + BR [1 + (sv)]
2 2  
e mν mν
− (CL − CR + 4K2 ) (sBt ) + (sB )
2 E
mν2
+ (AL + AR + 2K1 ) , (4.234)
2E
where v = p/E is the neutrino velocity vector, s is the average neutrino doubled spin
vector, Bt, are the transversal and longitudinal components of a magnetic field B
with respect to the neutrino momentum, B = Bt + B .
112 4 Particle Dispersion in External Active Media

As it was mentioned above, the coefficients AR , BR , CR , K1,2 are supressed by the

square of the lepton and W boson mass ratio, and can be neglected. For ultrarelativistic
neutrinos, we obtain from (4.234):

(1 − (ns)) emν
ΔE  BL − CL (B[ n × [s × n]]), (4.235)
2 2
where n is a unit vector in the direction of the neutrino momentum. The terms
proportional to the square of the neutrino mass were omitted in Eq. (4.235).
Thus, finding the additional energy acquired by the neutrinos during their forward
scattering in a magnetized plasma is reduced to calculating the parameters BL and CL .
The term in Eq. (4.235) proportional to the first power of the neutrino mass
corresponds to the additional neutrino energy attributable to the neutrino magnetic
moment and will be further analysed in detail. The additional neutrino energy in the
medium for left-handed massless neutrinos is defined only by the parameter BL :

ΔE = BL .

Since the additional neutrino energy, being a physical quantity, is gauge-invariant,

we will perform our calculations in a unitary gauge, which is convenient in that it
contains no contribution from scalar bosons. In this gauge, the amplitude of the ν → ν
scattering in a magnetized plasma can be represented as the sum of two terms:

M(ν→ν) = M(ν→ν)
W
+ M(ν→ν)
Z
, (4.236)

where the first term corresponds to the amplitude of the neutrino forward scattering
by plasma electrons and positrons of the medium via a W boson (see Fig. 4.15) and the
second term is attributable to the ν → ν transition via a Z boson (see Fig. 4.12 (c,d)
where the forward scattering case p = p and k = k should be taken). The neutrino
scattering by plasma neutrinos shown in Fig. 4.12 is insensitive to the presence of
an external magnetic field; its contribution to the additional neutrino energy was
investigated previously and was calculated in Ref. [29], see Eq. (4.82):
√
ΔE Z 7 2 GF π 2 T 4
=− . (4.237)
|p| 45 mZ2

Note that we do not consider the diagrams of Figs. 4.11 and 4.12a,b where the
4-momentum of the intermediate Z boson is zero in the forward scattering. This is
because such diagrams give only a local contribution, which is zero in a charge-
symmetric plasma. Thus, our problem is reduced to calculating the contribution of a
magnetized plasma to the additional neutrino energy from the W boson exchange.
The scattering process that corresponds to the diagrams in Fig. 4.15 is described by
the Lagrangian (4.100). The corresponding S-matrix element of the neutrino forward
scattering by plasma electrons is:
4.7 Neutrino Self-energy Operator in Magnetized Plasma 113

e (p) e (k ) e (p) e (p)

W
W

e (k ) e (p) e+(k ) e +(k )

Fig. 4.15 The Feynman diagrams for the neutrino forward scattering on plasma electrons and
positrons through W –boson. Double lines correspond to charged particles influenced by an external
magnetic field


g2  d4 x d4 x
W
Sνe − →νe− = √ √ ei(px −p x) (4.238)
2 n 2EV 2E V
(W )
× ν̄(p )γα γL Rn (x, x )γβ γL ν(p)Gβα (x , x),

where V = L1 L2 L3 is the normalization volume, pμ = (E, p) and p μ = (E , p )

are the 4-momenta of the initial and final neutrinos, ν(p) is the bispinor neutrino
(W )
amplitude, γL = (1 − γ5 )/2, Gβα (x , x) is the W boson propagator in a magnetic
field (3.5), Rn (x, x ) is the density matrix of the plasma electron with a fixed Landau
level number n:
  dk2 dk3
Rn (x, x ) = L2 L3 f (ωn ) ψe (x) ψ̄e (x ). (4.239)
s
(2π)2

Here, ψe (x) are the solutions (2.30) of the Dirac equation in the external magnetic
field,4 ωn = k32 + 2eBn + me2 is the energy of the electron at the nth Landau level,
k3 is the kinetic momentum along the third axis, k2 is a generalized momentum that
defines the position x0 = −k2 /eB of the Gaussian packet center on the first axis, and
f (ωn ) is the electron distribution function, which describes the presence of plasma.
In the plasma rest frame, it is

f (ω) = [e(ωn −μe )/T + 1]−1 ,

where μe is the chemical potential of the plasma and T is its temperature.

A detailed calculation of the function Rn (x, x ) is presented in Sect. 2.4. In a
constant uniform magnetic field, it can be reduced to the form

d3 k f (ωn ) −u −ik(x−x )
Rn (x, x ) = e−iΦ(x ,x) (−1)n e e (4.240)
(2π)3 ωn
× {[(kγ) + me ][Ln (2u)Π− − Ln−1 (2u)Π+ ] + 2(kγ)⊥ Ln−1
1
(2u)},

4 We perform our calculations in the gauge Aμ = (0, 0, Bx, 0); the magnetic field is directed along
the third axis B = (0, 0, B).
114 4 Particle Dispersion in External Active Media

eB
Φ(x , x) = (x1 + x1 ) (x2 − x2 ), (4.241)
2
2 /eB. The associated Laguerre polynomials are defined as follows:
where u = k⊥

1 x −s dn  n+s − x 
Lns (x) = e x x e . (4.242)
n! dx n
Equation (4.240) can be used to investigate quantum processes in a plasma in the
presence of a magnetic field with an arbitrary strength.

4.7.2 Neutrino Additional Energy in Magnetized Plasma

After the substitution of the function Rn (x, x ) in the form (4.240) and the W boson
propagator (3.5) into Eq. (4.238) and the integration over the 4-coordinates, the
S-matrix element of the νe− → νe− process can be reduced to the form

g 2 (2π)4 δ 4 (p − p )  d3 k f (ωn ) −u W
− →νe− = (−1)n e Gβα (k − p)
W
Sνe √ √
2 2EV 2E V n
(2π)3 ωn
'
× ν̄(p) γα (kγ) [Ln (2u)Π− − Ln−1 (2u)Π+ ]
(
+ 2 (kγ)⊥ Ln−1 1
(2u) γβ γL ν(p) . (4.243)

The four-dimensional δ function corresponding to the energy and momentum

conservation law has been separated out in the S-matrix element. Therefore, we can
use the standard relation between the S-matrix element and the invariant transition
amplitude,
i(2π)4 δ (4) (q − q )
Sif = Mif (4.244)
2ωV
and separate out the invariant amplitude of the neutrino scattering by plasma electrons
in the form

i g2  d3 k f (ωn ) −u W
− →νe− = − (−1)n e Gβα (k − p)
W
Mνe
2 n (2π)3 ωn
'
× ν̄(p) γα (kγ) [Ln (2u)Π− − Ln−1 (2u)Π+ ]
(
+ 2 (kγ)⊥ Ln−1
1
(2u) γβ γL ν(p). (4.245)

W
Calculating the amplitude Mνe + →νe+ of the neutrino scattering by plasma
W
positrons is identical to calculating the amplitude Mνe − →νe− . The result for the
W
transition amplitude Mνe+ →νe+ in a charge-symmetric plasma turned out to differ
4.7 Neutrino Self-energy Operator in Magnetized Plasma 115

from (4.245) by the general sign and by the substitution kμ → −kμ in the argument
of the W boson propagator. The amplitude of the coherent νe → νe scattering by all
plasma electrons and positrons is

ig 2 
W
Mνe→νe = Mνe
W
− →νe− + Mνe+ →νe+ = −
W
(−1)n (4.246)
2 n

d3 k f (ωn ) −u W
× e (Gβα (k − p) − GWβα (−k − p))
(2π)3 ωn
'
× ν̄(p) γα (kγ) [Ln (2u) Π− − Ln−1 (2u) Π+ ]
(
+ 2 (kγ)⊥ Ln−1
1
(2u) γβ γL ν(p) .

Using Eq. (4.63) for the amplitude, we find the contribution of plasma electrons
and positrons to the neutrino self-energy operator

i g2  d3 k f (ωn ) −u W
Σ W (p) = (−1)n e [Gβα (k − p) − GW
βα (−k − p)]
2 n (2π)3 ωn
(4.247)
) (
× γα (kγ) [Ln (2u)Π− − Ln−1 (2u)Π+ ] + 2(kγ)⊥ Ln−1 (2u) γβ γL .
1

The Fourier transform of the W boson propagator in a unitary gauge is fairly

cumbersome. For the physical conditions under consideration, where the W boson
mass is the largest physical parameter of the problem, the Fourier transform of the
propagator can be represented as an expansion in terms of inverse powers of the W
boson mass:

i gβα 3e Fβα i qβ qα i q2 gβα 1

βα (q) 
GW 2
− 4
− 4
+ 4
+O 6
. (4.248)
mW 2mW mW mW mW

Here, the first and the second momentum-independent terms give a contribution in
the local limit that is zero in a charge-symmetric plasma [29], as is clearly seen from
Eq. (4.246). The third and the fourth terms allow for the nonlocality of the interaction.
As our analysis shows, the third term in (4.248) contributes only to the parameter
AL,R and, hence, does not contribute to the additional neutrino energy.
Substituting the W boson propagator in the form (4.248) into Eq. (4.247) and
discarding the terms that do not contribute to the additional neutrino energy,
we obtain

2g 2 gαβ  d3 k f (ωn )
Σ (p) =
W
(−1) n
(pk) e−u
mW4
n
(2π)3 ωn
) (
× γα (kγ) [Ln (2u)Π− − Ln−1 (2u)Π+ ] + 2(kγ)⊥ Ln−1
1
(2u) γβ γL .
116 4 Particle Dispersion in External Active Media

Passing to integration over the variable u = k⊥

2 /eB in this expression,

+∞ +∞ ∞
d k = πeB
3
dk3 du
−∞ −∞ 0

and performing the integration over the variable u using the relations

∞
du e−u Ln (2u) = (−1)n ,
0

1
2u Ln−1 (2u) = n [Ln−1 (2u) − Ln (2u)],

we finally obtain

+∞
g 2 eB  dk3 f (ωn )
Σ (p) =
W
2 π mW n=0
2 4 ωn
−∞
 
(k ϕ̃p) 2
× (pγ)eBn − (pϕ̃γ) (k − eBn) − δn0 ωn2 (4.249)
ωn E 3
*
p23 2
− (uγ) E (ωn2 + eBn + (k − eBn)) − δno p3 (k32 + ωn2 ) γL .
E2 3

Here, δn0 is the Kronecker symbol, which is nonzero only for the ground Landau
level; the sum over the Landau levels (with a prime) is defined as

  ∞
1
F(n) = F(n = 0) + F(n).
2
n=0 n=1

Finally we find the contribution to the additional neutrino energy from the neutrino
forward scattering by electrons and positrons of a magnetized plasma:

√ +∞
2 2 GF eBE  dk3 f (ωn )
B W
=− (4.250)
π mW
2 2 ωn
n=0 −∞
× [ωn2 + eBn + cos φ (k32 − eBn) − δn0 cos φ (k32 + ωn2 )],
2

where φ is the angle between the magnetic field direction and the neutrino
momentum vector.
Below, we will consider some limiting cases that can be of interest from the
standpoint of possible astrophysical applications.
4.7 Neutrino Self-energy Operator in Magnetized Plasma 117

4.7.3 Asymptotic Expressions for the Neutrino Additional

Energy in Magnetized Plasma

A general expression for the additional neutrino energy in a charge-symmetric

magnetized plasma can be obtained by summing the contributions from the processes
attributable to the Z boson and W boson exchange and can be represented in the form:

√ √ +∞
ΔE 7 2 GF π 2 T 4 2 2 GF eB  dk3 f (ωn )
=− − (4.251)
|p| 2
45 mZ π mW n=0
2 2 ωn
−∞
 
× ωn2 + eBn + cos2 φ (k32 − eBn) − δn0 cos φ (k32 + ωn2 ) .

It should be noted that Eq. (4.251) describes the partial contribution of a

magnetized plasma to the additional neutrino energy. To obtain a complete expres-
sion for the neutrino energy in a magnetized plasma, the purely field contribution
calculated in Ref. [36] must be added to the result (4.251).
The integral in Eq. (4.251) can be calculated in some limiting cases considered
below.

(i) The limit of a weak magnetic field, when the magnetic field strength is the
smallest physical parameter of the problem,

T2 me2 eB. (4.252)

The additional neutrino energy in such a weakly magnetized plasma can be

reduced to the form
√ 
ΔE 2 GF 7 π2 T 4 2
mW
= − 2 + + T 2 eB cos φ +
|p | 3mW2 15 mZ2
  
(eB)2 T2
+ sin φ ln 2 + 0.635 − 1 .
2
(4.253)
2 π2 me

Equation (4.253) contains a logarithmic factor with the electron mass me . How-
ever, the electron mass is not the smallest parameter for the physical conditions
(4.252) under consideration and, hence, the additional neutrino energy (4.253)
cannot be investigated in the limit me → 0.
(ii) The limit of a moderate magnetic field, when the field strength is small on the
scale of physical parameters of the medium, but, at the same time, it is much
larger than the critical field strength for the electron:

T2 eB me2 . (4.254)
118 4 Particle Dispersion in External Active Media

Such a physical situation could take place, for example, in a supernova core
after its collapse, where the plasma temperature T ∼ 70 me . Substituting this
value into the conditions (4.254) yields

T2 B
∼ 5 × 103 1. (4.255)
me2 Be

Thus, we see that even the magnetic fields with strengths up to B ∼ 1015 − 1016
G satisfy the conditions (4.254) and, hence, may be considered as ‘relatively
weak’.
A large number of Landau levels are excited under the physical conditions
(4.254). In this limit, we find the additional neutrino energy to be
√ 
ΔE 2 GF 7 π2 T 4 2
mW
= − 2 + + T 2 eB cos φ
|p | 3mW2 15 mZ2
  
(eB)2 T2
+ sin φ ln
2
+ 2.93 − 1 . (4.256)
2 π2 eB

As one can see from Eq. (4.256), in contrast to the result of Ref. [40], the
additional neutrino energy under the physical conditions (4.254) contains no
infrared divergence in the limit me → 0.
(iii) The limit of a strong magnetic field, which corresponds to a physical situation
where the magnetic field strength is the largest of all the physical parameters
that characterize a magnetized plasma:

eB T 2 , me2 . (4.257)

Under the conditions (4.257), the plasma electrons and positrons occupy mostly
the ground Landau level.
In the limit of a strongly magnetized plasma, the additional neutrino energy is
√ 
ΔE 2 GF 7 π 2 T 4 mW
2
T 2 eB
=− + (1 − cos φ)2 (4.258)
|p | 3mW 2 15 mZ2 2
 
3/2  2 1/4 √
2 T
+ 3 (eB)2 (3 − cos2 φ) e− 2eB/T .
π 2eB

Here, the second term is attributable to the contribution from the ground Landau
level, and the third term containing the exponential suppression is caused by
the first Landau level.
4.7 Neutrino Self-energy Operator in Magnetized Plasma 119

4.7.4 Induced Neutrino Magnetic Moment in Magnetized

Plasma

As already noted, the additional interest in calculating the neutrino self-energy

operator stems from the possibility of obtaining the data on the anomalous mag-
netic moment of a neutrino. However, there is some doubt concerning the validity
of the data presented in the literature so far on the magnetic moment of the neutrino
in a magnetized plasma, since the results imply that the magnetic moment of the
neutrino is either independent of its mass mν [78] or exhibits a giant enhancement
by a factor of 1/mν . As was reasonably pointed out [39], these results confuse the
situation with the magnetic moment of the neutrino, instead of elucidating it. An
independent calculation of the neutrino magnetic moment in a magnetized plasma
was carried out in Ref. [77]. Here we reproduce a general scheme of the analysis.
Let us find a contribution to the energy of a neutrino, which is related to the
presence of its magnetic moment μν . This energy correction can be determined
using the Lagrangian expressed as

iμν
ΔLint = − ( Ψ̄ σμν Ψ ) F μν , (4.259)
2
where Ψ is the fermion field and σμν = (γμ γν − γν γμ )/2.
Substituting this formula into an expression for the additional energy defined as

(μ)
ΔE =− dV ΔLint  , (4.260)

we eventually obtain the following formula:

 mν 
ΔE (μ) = −μ (s Bt ) + (s B ) , (4.261)
E
where s is the average twice spin vector of a fermion, Bt, are the transverse and
longitudinal components of the magnetic field B with respect to the momentum of a
fermion, introduced in Eq. (4.234).
Thus, in expression (4.234) for the additional energy of the neutrino in a mag-
netized plasma, the magnetic moment only enters into the structure, which is
proportional to the following sum:
mν
(s Bt ) + (s B ) .
E
We should conclude that determination of the magnetic moment of the neutrino in
Ref. [78] was incorrect because it was assumed that the entire additional energy of the
neutrino (related to its dependence on the spin and magnetic field) in a magnetized
plasma contributes to the induced magnetic moment. However, as was shown above,
120 4 Particle Dispersion in External Active Media

Fig. 4.16 The Feynman

diagrams for the neutrino e (p) e (k) e (p) e (p)
scattering on plasma electrons Φ
and positrons through the Φ
charged scalar Φ–boson
e (k) e (p) e+(k) e+(k)

only one term in the expression for the additional energy of the neutrino refers to its
magnetic moment.
A comparison of expression (4.234) to formula (4.261) for the additional energy
of a neutrino shows that, in order to determine the magnetic moment of the neutrino
in a magnetized plasma, it is sufficient to find the coefficients CL , CR and K2 , see
Eq. (4.231), by which the magnetic moment is expressed as follows:
e mν
μν = (CL − CR + 4 K2 ) . (4.262)
2
Further we calculate the terms of the neutrino self-energy operator Σ(p), which
contribute to the magnetic moment of a neutrino. For variety, the calculation will be
given in the Feynman gauge.
In a magnetized plasma, this magnetic moment consists of two parts: the purely
field contribution and the plasma contribution. The field contribution to the mag-
netic moment of the neutrino was calculated in a number of papers (see, e.g., Refs.
[33, 37, 43]). An expression for the magnetic moment of the neutrino in a broad
range of its energies and of magnetic fields strengths, such that

m2 /mW
2
(eB)2 p2⊥ /mW
6
1,

can be written as follows [37]:

  
4 χ2 1 17
μν  μ0ν 1+ ln − + ln 3 + 2 γE + iπ . (4.263)
3 χ 3

Here, μ0ν is the neutrino magnetic moment in vacuum [69, 70]:

3 e GF mν
μ0ν = √ , (4.264)
8 2 π2

mν is the neutrino mass, p⊥ is the neutrino transverse momentum with respect to the
magnetic field direction, χ2 = (eB)2 p2⊥ /mW 6 , λ = m2 /m2 , γ = 0, 577 . . . is the
W E
Euler constant. The imaginary part of the magnetic moment (4.263) corresponds to
the neutrino instability in the external electromagnetic field with respect to the decay
ν → W.
4.7 Neutrino Self-energy Operator in Magnetized Plasma 121

Then, in order to calculate the self-energy operator Σ(p), it is sufficient to deter-

mine the amplitude (4.63) of the forward scattering of the neutrino in a magnetized
plasma. Under real astrophysical conditions, the main contribution to the ampli-
tude that accounts for the magnetic moment of the neutrino is due to scattering on
plasma electrons and positrons. The amplitude of the ν → ν scattering process in a
magnetized plasma (and, hence, the neutrino self-energy operator) can be represented
as the sum of three terms that correspond to the diagrams depicted in Figs. 4.11
and 4.15 and also the diagrams with the charged scalar Φ–boson, Fig. 4.16, which
appear in the Feynman gauge:

Σ(p) = Σ W (p) + Σ Φ (p) + Σ Z (p). (4.265)

The calculation of Σ(p) is similar to the one performed in Sect. 4.7.1. The
contribution to Σ(p) due to the scattering with W -boson exchange is:
∞ 
i g2  d3 k e−u
Σ W (p) = (−1)n (4.266)
2 (2π)3 ωn
n=0
 
× f (ωn ) GW ¯
βα (p − k) − f (ωn ) Gβα (p + k)
W

× γα [ (kγ) (Ln (2u) Π− − Ln−1 (2u) Π+ ) + 2(kγ)⊥ Ln−1

1
(2u) ] γβ γL ,

where g is the electroweak interaction constant in the standard model, Π± are the
projection operators (2.51), Gβα (q) is the Fourier transform of the translationally
invariant part of the W -boson propagator (3.13), f (ωn ) and f¯ (ωn ) are the distribution
functions of electrons and positrons, respectively. In the plasma rest frame, the latter
functions have the following form:

f (ωn ) = [e(ωn −μe )/T + 1]−1 , f¯ (ωn ) = [e(ωn +μe )/T + 1]−1 ,

where μe and T are the chemical potential and temperature of the plasma, respectively,
and ωn is the electron (positron) energy on the nth Landau level.
Similarly, the contribution from the process of neutrino scattering with scalar
Φ-boson exchange is as follows:
∞ 
i g2  d3 k e−u  
Σ Φ (p) = − (−1)n f (ωn )D(p − k) − f¯ (ωn )D(p + k)
2 (2π) ωn
3
n=0

me2 mν
× 2
(Ln (2u) Π− − Ln−1 (2u) Π+ ) [(kγ) (Ln (2u) Π− − Ln−1 (2u) Π+ )
mW
*
m 2 m 2
+ 2(kγ)⊥ Ln−1
1
(2u) ] e
2
γL − 2ν γR . (4.267)
mW mW
122 4 Particle Dispersion in External Active Media

Here, D(q) is the Fourier transform of the translationally invariant part of the Φ-
boson propagator (3.14).
Note that the contributions considered above refer only to the electron neutrino,
since contributions with exchange by charged bosons for neutrinos of other types
(νμ , ντ ) vanish (Σ W (p) = Σ Φ (p) = 0).
Below, we consider a realistic physical situation where the W -boson mass (mW )
is the largest parameter of the problem. This implies that parameters characterizing
a magnetized plasma obey the following condition:

me2 , μ2 , T 2 , eB 2
mW . (4.268)

If the plasma is charge-asymmetric, it suffices to retain only the main contributions

in the W and Φ boson propagators expanded in inverse powers of mW 2 , that is,

i gβα i
Gβα (q)  2
, D(q)  − 2
. (4.269)
mW mW

Having accomplished simple calculations, we can write the two contributions to

the neutrino self-energy as follows:
 0
√ n − n̄e0
Σ (p)  2 GF − e
W
(p F̃γ) + · · · γL , (4.270)
EB

Φ GF ne0 − n̄e0 me2 mν2
Σ (p)  √ (p F̃γ) 2
γ L + 2
γR (4.271)
2 EB mW mW
⎫
∞ ⎬
ie me2 dk
− mν (γ Fγ) (f (ω0 ) − f¯ (ω0 )) + · · · ,
4π 2 2
mW ω0 ⎭
0

where the dots corresponds to the terms not contributing to the magnetic moment
of the neutrino; ω0 is the electron (positron) energy on the ground (n = 0) Landau
level; and ne0 and n̄e0 are the electron and positron densities, respectively, on this level.
The difference of these densities is given by the following integral:

∞
eB
ne0 − n̄e0 = dk (f (ω0 ) − f¯ (ω0 )). (4.272)
2π 2
0

Comparing expressions (4.270) and (4.271) to the parametrization in Eq. (4.231),

we obtain the following formulas for the coefficients CL , CR , and K2 :
4.7 Neutrino Self-energy Operator in Magnetized Plasma 123

∞
e GF
CLW =−√ dk (f (ω0 ) − f¯ (ω0 )), (4.273)
2 π2 E
0
CRW = K2W = 0, (4.274)
me2 mν2
CLΦ = − 2
CLW , CRΦ = − 2
CLW , (4.275)
2mW 2mW
∞
e GF me2 dk
K2Φ =− √ (f (ω0 ) − f¯ (ω0 )). (4.276)
2
4 2π 2 mW ω0
0

Note that, as could be expected, the contributions from charged scalar exchange
are suppressed by the small factors mν2 /mW
2 and m2 /m2 .
e W
The third term, Σ , in the neutrino self-energy operator (4.265), which accounts
Z

for the contribution from neutrino scattering on charged fermions with Z-boson
exchange, is readily calculated as follows:

√ f
T3 0
ΣfZ = 2 GF − (n − n̄f0 ) (p F̃ γ) + · · · γL . (4.277)
BE f

where nf0 , and n̄f0 are the densities of charged fermions and antifermions, respectively,
f
on the ground Landau level; T3 is the third component of the weak isospin of a
charged fermion; and the dots correspond to terms not contributing to the magnetic
moment of neutrino. Taking into account that the maximum density of particles
on the ground Landau level corresponds to electron and positrons, we obtain from
expression (4.277) the following formulas for the coefficients CL , CR and K2 :

∞
e GF
CLZ =− √ dk (f (ω0 ) − f¯ (ω0 )), (4.278)
2 2 π2 E
0

CRZ = K2Z = 0. (4.279)

Thus, the neutrino magnetic moment induced by a charge-asymmetric plasma is

expressed in terms of the coefficients CLZ and CLW , with an addition of the purely field
contribution (4.263). The final formula for CL is:
⎛ ⎞
∞
3eG 2
dk (f (ω0 ) − f¯ (ω0 ))⎠ ,
F ⎝
CL  CLW + CLZ  √ 1∓ (4.280)
4 2 π2 3E
0
124 4 Particle Dispersion in External Active Media

where the upper sign refers to the electron neutrino (νe ) and the lower sign, to the
muon and tau neutrinos (νμ , ντ ). Contributions proportional to 1/mW 4 , 1/m6 , etc.,
W
were neglected.
The integral in expression (4.280) is easily calculated for an ultrarelativistic
plasma. In this case, the magnetic moment of the neutrino is given by the following
simple formula: 
CL mν 3e GF mν 2 μe
μν =  √ 1∓ . (4.281)
2 8 2π 2 3 E

where μe is the chemical potential of electrons in the ultrarelativistic plasma. For

weakly magnetized plasma, it reduces to

μe  (3π 2 (ne − n̄e ))1/3 , (4.282)

For strongly magnetized plasma, in which case magnetic field rather than the plasma
is the dominant component of the active medium and plasma electrons occupy the
ground Landau level, the chemical potential of electrons is

(ne − n̄e )
μe  2π 2 , (4.283)
eB
where ne and n̄e are the total electron and positron densities, respectively.
Another situation for which analytical calculation of the neutrino magnetic
moment can be performed refers to physical conditions of a charge-symmetric
electron-positron plasma. In this case, the contribution from the diagram of neu-
trino scattering with Z-boson exchange vanishes and, hence, the νμ and ντ -type
neutrinos possess no additional magnetic moment induced by a magnetized plasma.
For the electron neutrino in a charge-symmetric e− e+ plasma, the magnetic mo-
ment is determined by the following expression:

3e GF mν 4π 2 T 2
μνe  √ 1+ 2
. (4.284)
8 2 π2 9 mW

As one can see, under real astrophysical conditions, where T mW , the plasma
contribution to the neutrino magnetic moment is suppressed.
Thus, we have shown that the presence of a plasma does not lead to an enhancement
of the neutrino magnetic moment, in contrast to the statement of Ref. [78]. The
plasma-induced part of the magnetic moment is suppressed by the neutrino mass
mν ; in a charge-symmetric plasma, it is also suppressed by a factor of T 2 /mW
2 1.
References 125

References

1. E. Braaten, Phys. Rev. Lett. 66, 1655 (1991)

2. I.A. Batalin, A.E. Shabad, Zh. Eksp. Teor. Fiz. 60, 894 (1971). [Sov. Phys. JETP 33, 483
(1971)]
3. W.-Y. Tsai, Phys. Rev. D 10, 2699 (1974)
4. A.E. Shabad, Ann. Phys. (N.Y.) 90, 166 (1975)
5. V.V. Skobelev, Izv. Vyssh. Uchebn. Zaved. Fiz. 10, 142 (1975)
6. A.E. Shabad, in Polarization of the Vacuum and a Quantum Relativistic Gas in an External
Field, ed. by V.L. Ginzburg (Nova Science Publishers, New York, 1992)
7. V.B. Berestetskii, E.M. Lifshitz, L.P. Pitaevskii, Quantum Electrodynamics, 2nd edn. (Perga-
mon, Oxford, 1982)
8. S.L. Adler, Ann. Phys. (N.Y.) 67, 599 (1971)
9. R.J. Stoneham, J. Phys. A 12, 2187 (1979)
10. A.N. Ioannisian, G.G. Raffelt, Phys. Rev. D 55, 7038 (1997)
11. V.N. Baier, A.I. Milstein, R.Zh. Shaisultanov, Phys. Rev. Lett. 77, 1691 (1996)
12. V.N. Baier, A.I. Milstein, R.Zh. Shaisultanov, Zh. Eksp. Teor. Fiz. 90, 1141 (1986). [Sov. Phys.
JETP 63, 665 (1986)]
13. A.A. Sokolov, I.M. Ternov, V.Ch. Zhukovskii, A.V. Borisov, Quantum Electrodynamics (Mir
Publishers, Moscow, 1988)
14. A.V. Kuznetsov, N.V. Mikheev, Electroweak Processes in External Electromagnetic Fields
(Springer, New York, 2003)
15. M.Yu. Borovkov, A.V. Kuznetsov, N.V. Mikheev, Yad. Fiz. 62, 1714 (1999). [Phys. At. Nucl.
62, 1601 (1999)]
16. A.A. Gvozdev, N.V. Mikheev, L.A. Vassilevskaya, Phys. Lett. B 410, 211 (1997)
17. E. Braaten, D. Segel, Phys. Rev. D 48, 1478 (1993)
18. C.J. Horowitz, B.D. Serot, Nucl. Phys. A 464, 613 (1987)
19. Q.R. Ahmad et al., SNO Collab. Phys. Rev. Lett. 87, 071301 (2001)
20. Q.R. Ahmad et al., SNO Collab. Phys. Rev. Lett. 89, 011301 (2002)
21. Q.R. Ahmad et al., SNO Collab. Phys. Rev. Lett. 89, 011302 (2002)
22. Y. Fukuda et al., Super-Kamiokande Collab. Phys. Lett. B 433, 9 (1998)
23. Y. Fukuda et al., Super-Kamiokande Collab. Phys. Lett. B 436, 33 (1998)
24. Y. Fukuda et al., Super-Kamiokande Collab. Phys. Rev. Lett. 82, 2644 (1999)
25. K. Eguchi et al., KamLAND Collab. Phys. Rev. Lett. 90, 021802 (2003)
26. B. Pontecorvo, Zh. Eksp. Teor. Fiz. 33, 549 (1957). [Sov. Phys. JETP 6, 429 (1957)]
27. B. Pontecorvo, Zh. Eksp. Teor. Fiz. 34, 247 (1958). [Sov. Phys. JETP 7, 172 (1958)]
28. J.N. Bahcall, in XXIII Physics in Collisions Conference (PIC03) (Zeuthen, Germany, 2003).
(e-print astro-ph/0310030)
29. D. Nötzold, G. Raffelt, Nucl. Phys. B 307, 924 (1988)
30. J.C. D’Olivo, J.F. Nieves, P.B. Pal, Phys. Rev. D 40, 3679 (1989)
31. J.C. D’Olivo, J.F. Nieves, M. Torres, Phys. Rev. D 46, 1172 (1992)
32. G. McKeon, Phys. Rev. D 24, 2744 (1981)
33. A.V. Borisov, V.Ch. Zhukovskii, A.V. Kurilin, A.I. Ternov, Yad. Fiz. 41, 743 (1985). [Sov. J.
Nucl. Phys. 41, 473 (1985)]
34. A. Erdas, G. Feldman, Nucl. Phys. B 343, 597 (1990)
35. E. Elizalde, E.J. Ferrer, V. de la Incera, Ann. Phys. 295, 33 (2002)
36. A.V. Kuznetsov, N.V. Mikheev, G.G. Raffelt, L.A. Vassilevskaya, Phys. Rev. D 73, 023001
(2006)
37. A.V. Kuznetsov, N.V. Mikheev, Yad. Fiz. 70, 1299 (2007). [Phys. At. Nucl. 70, 1258 (2007)]
38. V.B. Semikoz, J.W.F. Valle, Nucl. Phys. B 425 651 (1994); Erratum: ibid. 485, 545 (1997)
39. P. Elmfors, D. Grasso, G. Raffelt, Nucl. Phys. B 479, 3 (1996)
40. A. Erdas, G.W. Kim, T.H. Lee, Phys. Rev. D 58, 085016 (1998)
41. E. Elizalde, E.J. Ferrer, V. de la Incera, Phys. Rev. D 70, 043012 (2004)
126 4 Particle Dispersion in External Active Media

42. A. Bravo Garcia, K. Bhattacharya, S. Sahu, Mod. Phys. Lett. A 23, 2771 (2008)
43. A. Erdas, Phys. Rev. D 80, 113004 (2009)
44. L. Wolfenstein, Phys. Rev. D 17, 2369 (1978)
45. P. Langacker, J. Liu, Phys. Rev. D 46, 4140 (1992)
46. A.V. Kuznetsov, N.V. Mikheev, A.M. Shitova, Int. J. Mod. Phys. A 26, 4773 (2011)
47. A. Lobanov, A. Studenikin, Phys. Lett. B 564, 27 (2003)
48. A. Studenikin, J. Phys. A 39, 6769 (2006)
49. W. Grimus, H. Neufeld, Phys. Lett. B 315, 129 (1993)
50. J.C. D’Olivo, J.F. Nieves, P.B. Pal, Phys. Lett. B 365, 178 (1996)
51. A.V. Kuznetsov, N.V. Mikheev, Mod. Phys. Lett. A 21, 1769 (2006)
52. A.V. Kuznetsov, N.V. Mikheev, Int. J. Mod. Phys. A 22, 3211 (2007)
53. D.G. Yakovlev, K.P. Levenfish, Yu.A. Shibanov, Usp. Fiz. Nauk 169, 825 (1999). [Physics-
Uspekhi 42, 737 (1999)]
54. G. Beaudet, V. Petrosian, E.E. Salpeter, Astrophys. J. 150, 979 (1967)
55. D.A. Dicus, Phys. Rev. D 6, 941 (1972)
56. H. Munakata, Y. Kohiyama, N. Itoh, Astrophys. J. 296, 197 (1985)
57. P.J. Schinder et al., Astrophys. J. 313, 531 (1987)
58. N. Itoh et al., Astrophys. J. 339, 354 (1989)
59. A. Studenikin, J. Phys. A 41, 164047 (2008)
60. C. Lunardini, A. Yu Smirnov, Nucl. Phys. B 583, 260 (2000)
61. C. Lunardini, A. Yu Smirnov, Phys. Rev. D 64, 073006 (2001)
62. S. Sahu, W.-Y.P. Hwang, Eur. Phys. J. C 58, 609 (2008)
63. E.D. Commins, P.H. Bucksbaum, Weak Interactions of Leptons and Quarks (Cambridge Uni-
versity Press, Cambridge, 1983)
64. A.Yu. Potekhin, Usp. Fiz. Nauk 180, 1279 (2010). [Physics-Uspekhi 53, 1235 (2010)]
65. L. Lewin, Polylogarithms and Associated Functions (North Holland, New York, 1981)
66. H.-Th. Janka, K. Langanke, A. Marek, G. Martínez-Pinedo, B. Müller, Phys. Rept. 442, 38
(2007)
67. F.S. Kitaura, H.-Th. Janka, W. Hillebrandt, Astron. Astrophys. 450, 345 (2006)
68. A.V. Kuznetsov, N.V. Mikheev, A.V. Serghienko, Phys. Lett. B 690, 386 (2010)
69. B.W. Lee, R.E. Shrock, Phys. Rev. D 16, 1444 (1977)
70. K. Fujikawa, R.E. Shrock, Phys. Rev. Lett. 45, 963 (1980)
71. E.J. Ferrer, V. de la Incera, Int. J. Mod. Phys. A 19, 5385 (2004)
72. A.V. Borisov, I.M. Ternov, L.A. Vassilevskaya, Phys. Lett. B 273, 163 (1991)
73. A.V. Borisov, L.A. Vassilevskaya, I.M. Ternov, Yad. Fiz. 54, 1384 (1991). [Sov. J. Nucl. Phys.
54, 845 (1991)]
74. D.A. Dicus, K. Kovner, W.W. Repko, Phys. Rev. D 62, 053013 (2000)
75. A. Ringwald, Int. J. Mod. Phys. A 21(Suppl. 1), 12 (2006)
76. M.S. Andreev, N.V. Mikheev, E.N. Narynskaya, Zh. Eksp. Teor. Fiz. 137, 259 (2010). [J. Exp.
Theor. Phys. 110, 227 (2010)]
77. R.A. Anikin, N.V. Mikheev, E.N. Narynskaya, Zh. Eksp. Teor. Fiz. 137, 1115 (2010). [J. Exp.
Theor. Phys. 110, 973 (2010)]
78. V.Ch. Zhukovskii, T.L. Shoniya, P.A. Aminov, Zh. Eksp. Teor. Fiz. 104, 3269 (1993). [Sov.
Phys. JETP 77, 539 (1993)]
Chapter 5
Electromagnetic Interactions in External
Active Media

In this chapter, we present in great detail the technique of calculations of the

electromagnetic processes in external active media. We consider mainly the two
processes. The first one, which is forbidden in a vacuum but is possible in an intense
external electromagnetic field, is the photon decay into the electron–positron pair
γ → e− e+ . We calculate the probability of this process in an external field, in the
two limiting cases where the detailed analytical calculations are possible. These are:
(i) the case of a very strong magnetic field when electrons and positrons occupy the
ground Landau level; (ii) the case of a relatively weak external field when the energy
of the initial particle is the main physical parameter of a problem; this case can be
analyzed in the crossed field approximation. Calculations are performed by the two
methods: (i) using the exact solutions of the Dirac equation; (ii) via the imaginary
part of the loop amplitude. We analyse also the process of the photon emission by
an electron in magnetic fields which is the crossed process to the γ → e− e+ decay.
The second process considered is the electromagnetic interaction of the Dirac neu-
trino having a magnetic moment, with plasma. The plasma influence on the virtual
photon, and contributions of plasma components into the neutrino scattering process
are taken into account. The upper bound on the neutrino magnetic moment using the
data on supernova S N 1987A is established. Possible effects of the neutrino magnetic
moment: shock-wave revival in a supernova explosion and the time evolution of the
neutrino signal (neutrino pulsar) are analysed.

5.1 Photon Decay into an Electron–Positron Pair in a Strong

Magnetic Field

The γ → e+ e− process is kinematically forbidden in a vacuum. The magnetic field

changes the kinematics of charged particles, electrons, and positrons, allowing the
production of an electron–positron pair in the kinematic region q2 = q02 −qz2  4m 2e ,
where q0 is the photon energy (the z axis is directed along the magnetic field).

A. Kuznetsov and N. Mikheev, Electroweak Processes in External Active Media, 127

Springer Tracts in Modern Physics 252, DOI: 10.1007/978-3-642-36226-2_5,
© Springer-Verlag Berlin Heidelberg 2013
128 5 Electromagnetic Interactions in External Active Media

In 1954, Klepikov [1] examined the production of an electron–positron pair by a

photon in a magnetic field and obtained the amplitude and width of the γ → e+ e−
decay in the semiclassical approximation. Later, the authors of Refs. [2–7] considered
this process in the context of its astrophysical applications. It was pointed out in
Ref. [6, 7] that the use of the expression derived in Ref. [1] for the width considerably
overestimates the result in the strong magnetic field limit. In this case, one should use
an exact expression for the width of one-photon production of a pair when electrons
and positrons occupy only the ground Landau level. This calculation is demonstrated
in the following section.

5.1.1 Direct Calculation Based on the Solutions of the Dirac

Equation

Photon decay into the electron–positron pair γ(q) → e− ( p  ) + e+ ( p) in a magnetic

field is described by the Lagrangian of the electromagnetic interaction
 
Lem = e Ψ (x) Â(x)Ψ (x) (5.1)

and is depicted by the Feynman diagram presented in Fig. 5.1.

In the first order of the perturbation theory with the interaction (5.1), one obtains
the following expression for the matrix element Si f

Si f = i e  f | N (Ψ ÂΨ )d4 x | i, (5.2)

where Aα is the electromagnetic field operator,

 1  
Aα = √ cλ ε(λ)
α e
−iq x
+ cλ+ εα(λ)∗ eiq x ,
q,λ
2ωV

Ψ is the operator of the electron–positron field,

Fig. 5.1 The Feynman diagram for the process γ → e− e+ in a magnetic field. Double lines indicate
that the effect of an external field is taken exactly into account in the wave functions of the electron
and the positron
5.1 Photon Decay into an Electron–Positron Pair in a Strong Magnetic Field 129
 
Ψ = a p,s Ψ (+) + b+
p,s Ψ
(−)
,
p,s,n

Ψ (+) is the normalized solution of the Dirac equation in a magnetic field, with
positive energy (2.30)–(2.32), and Ψ (−) is the corresponding solution with negative
energy (2.33)–(2.36). In a strong magnetic field, the electron and the positron can be
produced only in the states that correspond to the ground Landau level (2.37), (2.38),
which are described by the wave functions:

(eB)1/4    2
Ψ (+) = √ 
e−i(E t− p y y− pz z) e−ξ /2 u p , (5.3)
( π2E L y L z ) 1/2

(eB)1/4
Ψ (−) = √ ei(Et− p y y− pz z) e−ξ /2 u − p ,
2
(5.4)
( π2E L y L z )1/2

where
 
E= pz2 + m 2e , E =
pz2 + m 2e ,
 
√  py  
√ p y
ξ = eB x − , ξ = eB x + ,
eB eB
⎞ ⎛ ⎛⎞
0 0
1 ⎜ E  + me ⎟ 1 ⎜ E − me ⎟
u p =√  ⎜ ⎟, u− p =√ ⎜ ⎟. (5.5)
E + me ⎝ 0 ⎠ E − me ⎝ 0 ⎠
− pz − pz

Substituting the wave functions of the final state (5.3) and (5.4) into the expres-
sion (5.2) and integrating over dt dy dz, one obtains

ie(2π)3 δ 3 ( p + p  − q) 2 /2
(ū p ε̂(λ) u − p ) eiqx x e−ξ
2 /2
Si f = √ e−ξ dx,
2L y L z 2ωV E E 

where δ 3 ( p + p  − q) = δ(E + E  − ω) δ( p y + p y − q y ) δ( pz + pz − qz ). It is

convenient to perform the integration over x in the frame where the initial photon
momentum has a form q = (qx , 0, qz ), and therefore p y = − p y , ξ = ξ  . As u( p  )
and u(− p) do not depend on x, the integral of the Gaussian type with respect to x
can be easily calculated. Taking the strong field limit into account, we assume that
exp(−qx2 /2eB) 1. The S matrix element squared can be written as follows

e2 (2π)3 T
|Si f |2 = |ū p ε̂(λ) u − p |2 δ 3 ( p + p  − q), (5.6)
8L y L z V ω E E 
130 5 Electromagnetic Interactions in External Active Media

where T is the total interaction time. The expression |ū p ε̂(λ) u − p |2 can be rewritten
in terms of the trace calculation

|ū p ε̂(λ) u − p |2 = Tr[ρ( p  )ε̂(λ) ρ(− p)ε̂(λ) ],

where the density matrix of polarized electrons is

ρ( p  ) = ( pˆ + m e )− , (5.7)

and for polarized positrons

ρ(− p) = ( p̂ − m e )− . (5.8)

− is the projecting operator (2.51) corresponding to the electron or positron state on

the ground Landau level, where the electron spin direction is opposite to the external
field direction, while the positron spin is directed along the field.
To simplify further calculations, we make the Lorentz transformation along the
field direction, as q is the timelike vector,

q2 = ω 2 − qz2 = (E + E  )2 − ( pz + pz )2 > 0,

to the frame where qz = 0. In this frame pz = − pz , E = E  . Further we perform

separate calculations for the definite photon polarizations, using the explicit form of
the polarization vectors ε(λ) (λ = 1, 2). The vectors describing the physical states
of a photon in a magnetic field (for details see Sect. 4.2) are

(qϕ)α (q ϕ̃)α
ε(1)
α =  , ε(2)
α =  . (5.9)
2
q⊥ q2

Substituting the polarization vector of the 1st mode photon, one obtains

Tr[ρ( p  )ε̂(1) ρ(− p)ε̂(1) ] = 0.

By this means the 1st mode photon cannot decay into the electron–positron pair with
both electron and positron being produced in the ground Landau level. Performing
the similar calculation for the 2nd mode photon one obtains

Tr[ρ( p  )ε̂(2) ρ(− p)ε̂(2) ] = 4m 2e .

The resulting S matrix element squared for the decay of the 2-mode photon takes the
form
e2 (2π)3 m 2e T 3
|Si f |2 = δ ( p + p  − q), (5.10)
2L y L z V ω E E 
5.1 Photon Decay into an Electron–Positron Pair in a Strong Magnetic Field 131

where δ 3 ( p + p  − q) = δ(2E − ω) δ( p y + p y − q y ) δ( pz + pz ).

To find the decay probability, one should perform the integration over the phase
space of final electrons and positrons:

|Si f |2
dW = dn f , (5.11)
T
where

d p y d pz d p y d pz
dn f = L 2y L 2z . (5.12)
(2π)4

The δ function for energies can be transformed into the following form

δ(2E − ω) 1
= [δ( pz − p ∗ ) + δ( pz + p ∗ )] Θ(ω 2 − 4m 2e ),
ω 4| p ∗ |

where p ∗ = ± 21 ω 2 − 4m 2e , Θ(x) is the step function.
The integration over d p y d pz d pz removes the δ functions. It can easily be
seen that the integrand is independent on p y ; hence, integration with respect to
p y actually determines the degeneracy multiplicity of the electron state at a given
energy:

 Lx /2
Ly eB L y eB L x L y
NE = d py = dx0 = , (5.13)
2π 2π 2π
−L x /2

where x0 = p y /eB determines the center of the Gaussian packet on the x axis;
see (5.5). As a result, for the decay probability of the 2-mode photon one obtains

4αeBm 2e
W (2) =  Θ(ω 2 − 4m 2e ). (5.14)
ω ω − 4m e
2 2 2

The Θ function is seen to define the threshold of the photon decay into the e− e+ pair.
Making the inverse Lorentz transformation, in view of the invariance of the product
ωW , one can rewrite the probability (5.14) in an arbitrary frame

4αeBm 2e
W (2) =  Θ(ω 2 sin2 θ − 4m 2e ), (5.15)
ω 2 sin θ ω 2 sin θ − 4m 2e
2

where θ is the angle between the photon momentum and the magnetic field direction.
The formula obtained shows that the photon decay process has a resonant
character. It is enhanced essentially when the angle θ is close to θres = arcsin(2m e /ω).
132 5 Electromagnetic Interactions in External Active Media

5.1.2 Calculation Based on the Imaginary Part of the Loop

Amplitude

There exists another way to calculate the probability of the photon decay in a magneic
field, which is based on an application of the unitarity relation (see e.g. [8])

1
W (γ → e− e+ ) = Im M(γ → γ), (5.16)
ω
where ω is the photon energy. The amplitude of the transition M(γ → γ) in a
magnetic field can be obtained from (4.31), where the vector currents should be
replaced as follows, jV α → eε(λ) (λ)
α , here εα (λ = 1, 2) are the polarization vectors
(4.10); the condition q = 0 should be set also. By this means, we obtain from (4.31)
2

  α
(λ)
ΔM(λ) ≡ ΔM γ (λ) → γ (λ) = YV V , λ = 1, 2. (5.17)
π
(λ)
To take the strong field limit in the functions YV V , it is worthwhile to make the Wick
rotation of the integration contour in the complex plane t (see Chap. 3), replacing
it on the negative imaginary axis, t = −iτ , where τ is a real variable. In this case
sin βt = −i sinh βτ and cos βt = cosh βτ . Let us analyze first the amplitude (5.17)
for the 2nd mode photon. We obtain:

1 ∞  
α dτ βτ 1 − u2
e−τ [m e −q (1−u )/4] q2
2 2 2
(2)
ΔM = du cosh βτ
π τ sinh βτ 2
0 0
(5.18)
   
q2 u sinh βτ u 1 − u 2 −τ [m 2e −q 2 (1−u 2 )/4]
− ⊥ cosh βτ u − − q2 e .
2 tanh βτ 2

Taking the strong field limit we assume that the field parameter β = eB is the maximal
dimensional parameter of our problem, β q2 , q⊥
2 , m 2 . It is seen from the integrand
e
in (5.18) that the region τ ∼ 1/m 2e , 1/q2 1/β gives the main contribution. In this
region one can assume
1 βτ
cosh βτ sinh βτ e .
2
In the strong field limit, the field-induced part of the amplitude dominates and actually
it defines the total amplitude of the transition γ (2) → γ (2) , M(2) ΔM(2) . The
integral with respect to τ in (5.18) can be easily calculated to give
 
(2) 2αβ q2
M H , (5.19)
π 4m 2e
5.1 Photon Decay into an Electron–Positron Pair in a Strong Magnetic Field 133

where the function H (z) is defined in (4.17). Using (4.18), one obtains the following
expression for the imaginary part of the amplitude,

4αβm 2e
Im M(2) =  Θ(q2 − 4m 2e ). (5.20)
q (q − 4m e )
2 2 2

Substituting (5.20) into (5.16), in view of β = eB and q2 = ω 2 − q32 = ω 2 sin2 θ,

we obtain the result for the probability that coincides with (5.15).
The similar analysis of the amplitude for the transition γ (1) → γ (1) shows that in
(1)
the strong field limit, the integral YV V does not have an interval where the enhancing
(2)
factor β could arise, as it was for YV V . But it is more essential that the amplitude
M(1) does not have an imaginary part in the strong magnetic field limit β q2 .
− +
Thus, only the 2nd mode photon can decay into the e e pair in the strong field
limit.

5.2 The γ → e− e+ Decay in a Crossed Field

As was already mentioned, the case of a relatively weak external field when the
photon energy is the largest physical parameter, corresponds to the crossed field
approximation. We perform the calculations by the two ways, first by using the exact
solution of the Dirac equation (2.40), and second via the imaginary part of the loop
amplitude MV V (4.35) for the transition γ → e− e+ → γ.

5.2.1 Direct Calculation Based on the Solutions of the Dirac

Equation

Substituting the solutions in a crossed field (2.40) for the electron and the positron
into the S matrix element (5.2), one obtains
   
ie 1 3
Si f =√ d x exp −i (Qx) − r κ (ϕ0 ϕ + ϕ )
4 3 3 2
2ωV 2E V 2E  V 3
     
eâ k̂ ek̂ â
× ū( p) 1 − ϕ ε̂ 1 + ϕ u(− p  ) , (5.21)
2(kp) 2(kp  )

where the following notations are used: Q = q − p − p  , and

134 5 Electromagnetic Interactions in External Active Media

 1/3
χ e2 (aa) e(q F p)
r= , κ2 = − , ϕ0 = − 4 ,
2χ1 χ2 me 2 m e κχ
 1/2
e2 (q F Fq) κ(qk)
χ= = ,
m 6e m 2e
 1/2
e2 ( p F F p) κ( pk)
χ1 = = ,
m 6e m 2e
 1/2
e2 ( p  F F p  ) κ( p  k)
χ2 = = . (5.22)
m 6e m 2e

Taking the frame (2.41) and keeping in mind that ϕ = (kx) = k0 (t − x), we can
write
Qx
(Qx) = (Q 0 − Q x )t − Q y y − Q z z + sϕ, s = .
k0

The integrals with respect to y and z give the two-dimensional δ function:


dy dz ei(Q y y+Q z z) = (2π)2 δ 2 (Q⊥ ).

Changing the variables t, x to t, ϕ

 
1
dt dx = dt dϕ
k0

and integrating with respect to t:


κ
dt e−i(Q 0 −Q x )t = 2πδ(Q 0 − Q x ) = 2πk0 δ(k Q) = 2πk0 δ(χ − χ1 − χ2 ),
m 2e

we transform the S matrix element (5.21) to the form

∞  
ie(2π)3 δ 2 (Q⊥ )δ(k Q)
Si f = √ dϕ ū( p)γμ L μν εν u(− p  )
2ωV 2E V 2E  V
−∞
  
1
× exp −i sϕ − r 3 κ 3 (ϕ0 ϕ2 + ϕ3 ) , (5.23)
3

where

e2 κ 2
L μν = g μν + κ− F μν ϕ − i κ+ γ5 F̃ μν ϕ − (F F)μν ϕ2 , (5.24)
2m 4e χ1 χ2
 
eκ 1 1
κ± = ± . (5.25)
2m 2e χ1 χ2
5.2 The γ → e− e+ Decay in a Crossed Field 135

It is worthwhile to perform further calculations for a photon of the definite

polarization. The polarization vectors (4.10) in a crossed field can be presented in
the form
(Fq)α ( F̃q)α
ε(1)
α = √ , ε(2)
α = √ . (5.26)
(q F Fq) (q F Fq)

We obtain
   
(1) 1
Lε =√ (Fq)μ + κ− (F Fq)μ (ϕ − ϕ0 ) ,
μ (q F Fq)
   
(2) 1
Lε =√ ( F̃q)μ − iκ+ γ5 (F Fq)μ (ϕ − ϕ0 ) . (5.27)
μ (q F Fq)

Making a shift in the integral with respect to the variable ϕ, ϕ → ϕ − ϕ0 , we

can remove the terms in the exponent which are proportional to ϕ2 , to obtain
 
1 1
s ϕ − r 3 κ 3 ϕ0 ϕ2 + ϕ3 → s̄ϕ − r 3 κ 3 ϕ3 + A,
3 3

where
s̄ = s + r 3 κ 3 ϕ20 .

The value A not depending on ϕ is inessential; it leads to the appearance of a constant

phase factor in the S matrix element. Given the symmetry of the integral limits, this
shift on ϕ allows to express the result in terms of the Airy function

∞  
1 z3
Ai(y) = dz cos yz + , (5.28)
π 3
0

satisfying the equation

Ai (y) − yAi(y) = 0 . (5.29)

Thus, the integrals over ϕ can be rewritten as follows:

+∞  
1 2π
dϕ exp −i(s̄ϕ − r 3 κ 3 ϕ3 ) = Ai(y) , (5.30)
3 rκ
−∞
+∞  
1 3 3 3 2πi
dϕ ϕ exp −i(s̄ϕ − r κ ϕ ) = − 2 2 Ai (y) , (5.31)
3 r κ
−∞
+∞  
1 2π
dϕ ϕ2 exp −i(s̄ϕ − r 3 κ 3 ϕ3 ) = − 3 3 Ai (y) , (5.32)
3 r κ
−∞
136 5 Electromagnetic Interactions in External Active Media

where
s̄
y=− . (5.33)
rκ
The S matrix elements for the decays of photons with definite polarizations have
the form
 

−i A e(2π) δ (Q⊥ )δ(k Q) ū( p)γμ u(− p )
4 2
(1)
Si f = i e √ √
2ωV 2E V 2E  V r κ (q F Fq)
  
i 
× (Fq)μ Ai(y) + κ− (F Fq)μ − Ai (y) − ϕ0 Ai(y) , (5.34)
rκ

(2) e(2π)4 δ 2 (Q⊥ )δ(k Q) 1  
Si f = i e−i A √ √ ū( p)γμ u(− p  ) ( F̃q)μ Ai(y)
2ωV 2E V 2E V r κ (q F Fq)

 
 
 i 
+ i ū( p)γ5 γμ u(− p ) κ+ (F Fq)μ − Ai (y) − ϕ0 Ai(y) . (5.35)
rκ

The photon decay probability is defined as


1 d3 p V d3 p  V
W = |Si f |2 . (5.36)
T (2π)3 (2π)3

Substituting the matrix element, one should take into account, that, as usual,

L y Lz T
δ 2 (Q⊥ = 0) = , δ(k Q = 0) = ,
(2π) 2 2πk0

where L x , L y , L z are the typical scales along the axes O X, OY , and O Z , and T is
the total interaction time.
Integration over the positron momenta with the δ functions yields

d3 p  2 κ

δ (Q⊥ )δ(k Q){. . . } = 2 { p  → q − p − sk; χ2 → χ − χ1 }.
E m e χ2

For the integration over the electron momenta it is convenient to insert the variables
τ and u as follows

e(q F̃ p) χ1
τ= , u =1−2 . (5.37)
m 4e χ χ

In this case
1−u 1+u
χ1 = χ, χ2 = χ, (5.38)
2 2
5.2 The γ → e− e+ Decay in a Crossed Field 137

and we can write

 1 ∞ 
d3 p 1 2m 2e κ du
= dτ dϕ0 .
E χ2 χ 1 − u2
−1 −∞

However, as the calculation shows, the integrand does not depend on ϕ0 . If the
connection between ϕ and x is recalled, we can conclude that the integral with respect
to ϕ0 represents an arbitrariness of the choice of the zero point for the x coordinate.
Analyzing a problem within the finite quantization volume V = L x L y L z , we should
obviously take the integration region over ϕ0 to be finite and equal to k0 L x , i.e.
 
dϕ0 = k0 dx0 = k0 L x .

The argument (5.33) of the Airy function in the notations (5.37) has the form
 1/3
2
y = r 2 (τ 2 + 1), r= . (5.39)
χ(1 − u 2 )

The result of calculation of the decay probabilities for the photons of both polar-
izations (5.26) can be represented in the form

1 ∞  
(1,2) e2 m 2 χ1/3 du 1 + u2 1 − u2   2
W = 1/3e dτ ∓ Ai (y)
2 πω (1 − u 2 )2/3 2 2
0 −∞
 2/3    
2 1+u 1 − u2
2 2
+ 1+τ ± [Ai(y)] . (5.40)
2
χ(1 − u 2 ) 2 2

This result coincides, to the notations, with the result of [9], where the polarizations
 and ⊥ correspond to our 1 and 2.
To calculate the integrals with respect to the τ variable, which are involved in
(5.40),

∞ ∞ ∞
 2
I1 = dτ [Ai(y)] , 2
I2 = dτ τ [Ai(y)] ,
2 2
I3 = dτ Ai (y) ,
−∞ −∞ −∞
(5.41)
we use the known relations for the Airy function; see [9]:

 2 1 d2
y [Ai(y)]2 + Ai (y) = [Ai(y)]2 , (5.42)
2 dy 2
138 5 Electromagnetic Interactions in External Active Media

∞ ∞
dt 1
√ [Ai(t + a)]2 = dyAi(y), (5.43)
t 2
0 22/3 a
∞   ∞
σ σ d2
dtt [Ai(t + a)] = 2
− 4a dtt σ−1 [Ai(t + a)]2 , σ > 0.
2(2σ + 1) da 2
0 0
(5.44)

For the integrals (5.41) we obtain

1 21/3  
I1 = Bi(z), I2 = −Ai (z) − zBi(z) ,
2r 8r 3
21/3  
I3 = −3Ai (z) − zBi(z) , (5.45)
8r
where
∞  2/3
4
Bi(z) = dyAi(y), z = 22/3 r 2 = . (5.46)
χ(1 − u 2 )
z

Inserting the integrals (5.45) and turning to a new variable v = 1/(1−u 2 ), we present
the probability (5.40) in the form
 ∞  
αm 2e dv 4v − 2 ∓ 1 
W (1,2) = √ Bi(z) − Ai (z) , (5.47)
2ω 1 v v(v − 1) z

where z = (4v/χ)2/3 . The expression (5.47) can be further simplified by using the
Eq. (5.29) for the Airy function. We obtain

∞
(1,2) αm 2e χ dz 8v + 1 ∓ 3 χz 3/2
W =− √ √ Ai (z), v= . (5.48)
16 ω z v v(v − 1) 4
(4/χ)2/3

The formulae for the probability are simplified significantly in the two limiting cases:
for small values of the dynamical parameter χ

(1,2) 3 (3 ∓ 1)αm 2e
W (χ) = χ e−8/3χ , χ  1, (5.49)
2 16ω

and for large dynamical parameter

3(5 ∓ 1)Γ 4 (2/3)αm 2e

W (1,2) (χ) = (3χ)2/3 , χ 1. (5.50)
28π 2 ω
5.2 The γ → e− e+ Decay in a Crossed Field 139

Here Γ (z) is the gamma function, Γ (2/3) = 1.354 . . . . The presented expressions
for the probabilities coincide, to the notations, with corresponding formulas of [9].

5.2.2 Calculation Based on the Imaginary Part of the Loop

Amplitude

Similarly to Sect. 5.1.2, the decay probability in a crossed field can be calculated via
the unitarity relation. For this purpose the expression (4.35) should be substituted as
the amplitude M(γ → γ) into (5.16), replacing the vector current by the photon
(1,2)
polarization vectors (5.26), jV α → eεα , and setting q 2 = 0. We obtain
α (1)
M(γ (1) → γ (1) ) = Y
π VV
1  1/3
αm 2e χ2/3 4 d f (z)
=− du (3 + u 2 ) ,
6π 1 − u2 dz
0
(2) (2) α (2)
M(γ → γ ) = YV V
π
1  1/3
αm 2e χ2/3 4 d f (z)
=− du (3 − u 2 ) , (5.51)
3π 1−u 2 dz
0
 2/3
4
z= . (5.52)
χ(1 − u 2 )

Keeping in mind that the imaginary part of the Hardy–Stokes function is expressed
via the Airy function,
Im f (z) = π Ai(z), (5.53)

and changing the variable u to z in the integral

1 ∞
3 dz 1
du = √ ,
4 z v(v − 1)
0 (4/χ)2/3

where v = χz 3/2 /4, we readily obtain the formula (5.48).

140 5 Electromagnetic Interactions in External Active Media

5.3 Photon Emission by Electron in a Strong

Magnetic Field

Photon emission by an electron in an external electromagnetic field, e → e + γ, is

the crossed process to the photon decay into the pair e− e+ . Therefore, it is described
by the same diagram, Fig. 5.1, with the replacement p → − p, q → −q.
The S matrix element (5.2) for this process can be written in the form

ie
Si f = √ (Ψ ε̂(λ) Ψ )eiq x d4 x, (5.54)
2ωV

where Ψ and Ψ̄ correspond to the solutions of the Dirac equation in a magnetic field
with positive energy (2.30)–(2.32), ω is the photon energy.
It should be noted that the photon emission process is impossible when the initial
electron occupy the ground Landau level. To see this, it is enough to make the Lorentz
transformation to the rest frame of the initial electron ( pz = 0) where its energy is
equal to its mass. In another case, when both initial and final electrons occupy the
 where pz = 0, the energy conservation
first Landau level, and in the same frame,

law taking the form 2eB + m 2e = 2eB + m 2e + pz2 + ω, obviously cannot be
valid for the nonzero energy of the photon. Only the process is possible where the
electron emitting the photon, passes from the first Landau level into the ground one.
In a general case, only the processes could be realized where the electron passes into
a lower Landau level.
Let us consider the case when the field is strong enough and the electrons, which
are relativistic, can occupy only the ground and the first Landau levels. It is just the
case when the electron emitting the photon passes from the first Landau level into the
ground one. The energy of the relativistic electron in a magnetic field is (see (2.24)):

En pz2 + 2nβ.

The first Landau level (n = 1) is doubly degenerate because two spin states exist,
s = −1 and s = +1.
It is convenient for further calculations to take the frame where the pz component
√
of the initial electron momentum is equal to zero. In this frame pz = 0, E 2eB,
and the wave functions describing the state of relativistic electrons that occupy the
first Landau level, takes the following form, according to (2.30)–(2.32):
 1/4
(+) eB u p,s=+1 −ξ 2 /2 −i(Et− p y y)
Ψs=+1 =  e e . (5.55)
π 2L y L z
 1/4
(+) eB u p,s=−1 −ξ 2 /2 −i(Et− p y y)
Ψs=−1 =  e e . (5.56)
π 2L y L z
5.3 Photon Emission by Electron in a Strong Magnetic Field 141
⎛ ⎞ ⎛ ⎞
1 √0
⎜ 0 ⎟ ⎜ 2ξ ⎟
u p,s=+1 =⎜
⎝ 0 ⎠,
⎟ u p,s=−1 =⎜ ⎟
⎝ −i ⎠ .
√
i 2ξ 0

Substituting the wave functions of the initial state (5.55) and (5.56) and of the final
state (5.53) into the expression (5.54), we obtain the matrix elements Si f correspond-
ing to the two projections of the initial electron spin on the field direction,

ie(eB/π)1/2 2 /2
(ū p ε̂(λ) u p,s=±1 )e−ξ e−ξ
2 /2
Si f,s=±1 = √
2L y L z 2ωV
  
× eiq x e−i(Et− p y y) ei(E t− p y y− pz z) d4 x, (5.57)

where  
√  py   √ p y
ξ= eB x + , ξ = eB x + .
eB eB

By choosing the coordinate axes in such a manner that the vector of the photon
momentum would have the form q = (qx , 0, qz ), the integration with respect to x
in the expression Si f can be easily performed. In this frame we have p y = p y and
ξ = ξ  , and the matrix element Si f is transformed to the form

ie(eB/π)1/2
Si f,s=±1 = √ (2π)3 δ 3 (q + p  − p)
2L y L z 2ωV

× (ū p ε̂(λ) u p,s=±1 )e−iqx x e−ξ dx,
2
(5.58)

where δ 3 (q + p  − p) = δ(ω + E  − E) δ( p y − p y ) δ(qz + pz ) and the integration

over dt dy dz is taken.
Calculating the values (ū p ε̂(λ) u p ) for the initial electron with s = +1 and for
the photon 1- and 2-modes (4.10), we obtain
√
(1) iηqx (2) i 2ξ
ū p ε̂ u p,s=+1 = , ū p ε̂ u p,s=+1 =  (ηqz − ω), (5.59)
2
q⊥ q2

where η = pz /| pz |. Note that in this frame q⊥

2 = q 2 . For the initial electron with the
x
spin projection s = −1, we obtain
√
(1) qx (2) 2ξ
ū ε̂ u p,s=−1
p = − , ū ε̂
p u p,s=−1 = −  (qz − ηω). (5.60)
q⊥2 q2
142 5 Electromagnetic Interactions in External Active Media

The remaining integration with respect to x in (5.58) is reduced to the Gaussian

integral, and the calculation of the S matrix elements yields

(1) ie(2π)3 δ 3 (q + p  − p) iηqx −qx2 /4eB iqx p y /eB

Si f,s=+1 = √  e e ,
2L y L z 2ωV 2
q⊥

(2) ie(2π)3 δ 3 (q + p  − p) (ηqz − ω) (−iqx ) −qx2 /4eB iqx p y /eB

Si f,s=+1 = √  √ e e ,
2L y L z 2ωV q2 2eB

(1,2) (1,2)
Si f,s=−1 = iηSi f,s=+1 . (5.61)

Returning into a more general frame where q = (qx , q y , qz ), let us write the
matrix elements squared
 2  2 e2 (2π)3 T 3
 (1)   (1) 
δ (q + p  − p)e−q⊥ /2eB ,
2
 Si f,s=+1  =  Si f,s=−1  = (5.62)
8L y L z ωV
 2  2 e2 (2π)3 T (ηqz − ω)2 q⊥ 2
 (2)   (2) 
Si f,s=+1  = Si f,s=−1  =
8L y L z ωV 2eB q2
× δ 3 (q + p  − p)e−q⊥ /2eB ,
2
(5.63)

where δ 3 (q + p  − p) = δ(E − E  − ω)δ( pz + qz )δ(q y + p y − p y ). Thus, the

probability of the photon emission is seen to be independent on the polarization of
the initial electron. To find the total probability of the photon emission, the integration
over the phase space of final particles should be performed:

|Si f |2 d3 q V d p y d pz L y L z
W = . (5.64)
T (2π)3 (2π)2

Upon integrating (5.64) over d p y d pz with (5.62) and (5.63) taken into account
we obtain that the emission probabilities of the photons of the two modes, λ = 1, 2,
coincide at q 2 = 0:

α d2 q⊥ dqz −q 2 /2eB
W (1) = W (2) ≡ W = e ⊥ δ(E − |qz | − ω). (5.65)
8π ω

The δ function for energies can be presented as follows

δ(E − |qz | − ω) δ(qz + q ∗ ) + δ(qz − q ∗ )

= √ ,
ω 2eB
√
where q ∗ = (2eB − q⊥ 2 )/(2 2eB) defines the absolute value of q . From the
z
condition q ∗ > 0 we find the integration limits over the q⊥
2 variable,
5.3 Photon Emission by Electron in a Strong Magnetic Field 143

0 < q⊥
2
< 2eB.

Finally, we obtain

2eB
α α√
e−q⊥ /2eB dq⊥ 2eB(1 − e−1 ).
2
W = √ 2
= (5.66)
4 2eB 4
0

The total probability of the process

e− → e− + γ

averaged over the polarizations of the initial electron and summarized over polariza-
tions of the final photon, in the frame where pz = 0, is
α√
W = 2eB(1 − e−1 ). (5.67)
2
Taking account of the Lorentz invariance of the product of the probability by the
initial electron energy, we can rewrite the expression (5.67) to the arbitrary frame,
to obtain
αeB
W = (1 − e−1 ). (5.68)
pz + 2eB
2

5.4 Electromagnetic Interactions of the Dirac Neutrino

with a Magnetic Moment

Throughout this section, we use the notation μν for the magnetic moment of a neu-
trino, and the notation μ̃ν for a chemical potential of the neutrino gas.

5.4.1 Magnetic Moment of the Dirac Neutrino

and its Astrophysical Manifestations

Nonvanishing neutrino magnetic moment leads to various chirality-flipping processes

where the left-handed neutrinos produced in the stellar interior become the right-
handed ones, i.e. sterile with respect to the weak interaction, and this can be impor-
tant e.g. for the stellar energy-loss. In the standard model extended to include the
neutrino mass m ν , the well-known result for the neutrino magnetic moment is
[10, 11]:
3e G F m ν  m 
ν
μ(S
ν
M)
= √ = 3.20 × 10 −19
μB , (5.69)
8π 2 2 1 eV
144 5 Electromagnetic Interactions in External Active Media

where μB = e/2m e is the Bohr magneton. Thus, it is unobservably small given the
known limits on neutrino masses. On the other hand, nontrivial extensions of the
standard model such as left-right symmetry [12–19] can lead to more significant
values for the neutrino magnetic moment [20–22].
First attempts of exploiting the mechanism of the neutrino chirality flipping were
connected with the solar neutrino problem, and two different scenarios were analysed.
The first one, based on the neutrino magnetic moment rotation in a stellar magnetic
field, was investigated in the papers [23–25]. In the second scenario, a neutrino
changed the chirality due to the electromagnetic interaction of its magnetic moment
with plasma [26, 27]. For a more extended list of references see, e.g., [28]. In all
these cases the effect appeared to be small to have an essential impact on the solar
neutrino problem, if μν < 10−10 μB .
More stringent constraints on μν are provided by other stars. For example, the
cores of low-mass red giants are about 104 times denser than the Sun, and nonstandard
neutrino losses would have a more essential effect there, delaying the ignition of
heluim. Thus, the limit was obtained [29, 30]:

μν < 0.3 × 10−11 μB . (5.70)

An independent constraint on the magnetic moment of a neutrino was also

obtained from the Early Universe [31, 32]:

μν < 6.2 × 10−11 μB , (5.71)

where spin-flip collisions would populate the sterile Dirac components in the era
before the decoupling of the neutrinos. Thus, it doubles the effective number of
thermally excited neutrino degrees of freedom and increases the expansion rate of
the Universe, causing the overabundance of helium.
Interest in possible astrophysical and cosmological manifestations of the neu-
trino magnetic moment stimulated experiments on its measurement in laboratory
conditions. The best constraint was obtained in the GEMMA experiment to study
the scattering of antineutrinos by electrons carried out at the Kalinin nuclear power
station by the collaboration of the Institute of Theoretical and Experimental Physics
(Moscow) and the Joint Institute for Nuclear Research (Dubna). The upper bound
for the neutrino magnetic moment was [33]:

μν < 3.2 × 10−11 μB . (5.72)

A considerable interest to the neutrino magnetic moment arised after the great
event of SN1987A, in connection with the modelling of a supernova explosion,
where gigantic neutrino fluxes define in fact the process energetics. It means that
such a microscopic neutrino characteristic, as the neutrino magnetic moment, would
have a critical influence on macroscopic properties of these astrophysical events.
Namely, the left-handed neutrinos produced inside the supernova core during the
collapse, could convert into the right-handed neutrinos due to the magnetic moment
5.4 Electromagnetic Interactions of the Dirac Neutrino with a Magnetic Moment 145

interaction with a virtual plasmon γ ∗ that can be both produced and absorbed:

νL → ν R + γ∗, νL + γ∗ → ν R . (5.73)

These sterile neutrinos would escape from the core leaving no energy to explain
the observed neutrino luminosity of the supernova. Thus, the upper bound on the
neutrino magnetic moment can be established.
This matter was investigated by many authors in different aspects [34–38]. The
authors [36] considered the neutrino spin-flip via both ν L e− → ν R e− and ν L p →
ν R p scattering processes in the inner core of a supernova immediately after the
collapse. Imposing for the ν R luminosity Q ν R the upper limit of 1053 ergs/s, the
authors obtained the upper bound on the neutrino magnetic moment:

μν < (0.2 − 0.8) × 10−11 μB . (5.74)

However, the essential plasma polarization effects in the photon propagator were
not considered in Ref. [36], and the photon dispersion was taken in a phenomenolical
way, by inserting an ad hoc thermal mass into the vacuum photon propagator. A
detailed investigation of this question was performed in Refs. [39, 40], where the
formalism was used of the thermal field theory to take into account the influence of
hot dense astrophysical plasma on the photon propagator. The upper bound on the
neutrino magnetic moment compared with the result of the paper [36] was improved
in Refs. [39, 40] by the factor of 2:

μν < (0.1 − 0.4) × 10−11 μB . (5.75)

However, looking at the intermediate analytical results of the authors [39, 40], one
can see that only the contribution of plasma electrons was taken into account there,
while the proton fraction was omitted. This is despite the fact that the electron and
proton contributions to the neutrino spin flip process were evaluated in Ref. [36]
to be of the same order. It should be mentioned also that the improvement of the
bound (5.75) with respect to the bound (5.74) was based in part on the enhancement
by the factor of 2 of the supernova core volume made in Refs. [39, 40] if compared
with Ref. [36], while the density was taken to be the same, ρc 8 × 1014 g cm−3 .
This means that the core mass appeared to be in Ref. [39, 40] of the order of 3 M ,
which is nearly twice the mass of the supernova remnant believed to be typical.
The neutrino spin flip processes in the supernova core was reconsider more atten-
tively in Refs. [41–43]. It was shown in part, that the proton contribution into the pho-
ton propagator was not less essential, than the electron contribution. In this section,
we reproduce that analysis. We consider the Dirac neutrinos only, because in this
case the neutrino magnetic moment interaction (both diagonal and non-diagonal)
with a photon transforms the active left-handed neutrinos into the right-handed neu-
trinos which are sterile with respect to the weak interaction. We do not consider the
Majorana neutrinos, because the produced right-handed antineutrino states are not
sterile in this case.
146 5 Electromagnetic Interactions in External Active Media

The amplitude of the helicity flip through the scattering by plasma components
is calculated. A general expression for the creation probability of right-handed neu-
trinos with a fixed energy is derived. We estimate the core luminosity with respect
to the emission of neutrinos ν R and obtain an upper limit on the neutrino magnetic
moment by taking into account the radial distributions and time evolution of physical
parameters.

5.4.2 Neutrino Interaction with Background

The neutrino chirality flip is caused by the scattering via the intermediate photon
(plasmon) off the plasma electromagnetic current presented by electrons, ν L e− →
ν R e− , protons, ν L p → ν R p, etc. The total process Lagrangian consists of two parts,
ij
the first one is the interaction of a neutrino having a magnetic moment μν (both
diagonal and transition) with photons, while the second part describes the plasma
interaction with photons:

i  ij  
L=− μν ν̄ j σαβ νi F αβ − e Jα Aα , (5.76)
2
i, j

where σαβ = (1/2) (γα γβ − γβ γα ), F αβ is the tensor of the photon electromag-

netic field, Jα = − (ēγα e) + ( p̄γα p) + · · · is an electromagnetic current in the
general sense, formed by different components of the medium, i.e. free electrons and
positrons, protons, free ions, etc.
ij
The neutrino magnetic moment is generally a matrix μνi ν j ≡ μν that contains
both diagonal and transition magnetic moments, where νi and ν j are the states of a
neutrino with a specific mass. The neutrino states ν with specific flavors  = e, μ, τ
being produced in weak processes are superpositions of states νi :

ν = Ui∗ νi , (5.77)
i

where Ui is the unitary Pontecorvo–Maki–Nakagawa–Sakata leptonic mixing matrix

[44–47]. Below, for simplicity, we will consider the diagonal neutrino magnetic
moment μν . The extension to the general case of the matrix of magnetic moments
ij
μν presents no difficulty and consists in the following: the magnetic moment in all of
the succeeding expressions should be considered as an effective value. For example,
for the processes with initial electron neutrinos, by μν we should mean
⎛  2 ⎞1/2
 
⎜  i j  ⎟

μν → μνe ≡⎝  μν ej  ⎠
U . (5.78)
i  j 
5.4 Electromagnetic Interactions of the Dirac Neutrino with a Magnetic Moment 147

Fig. 5.2 The Feynman dia- L R

gram for the neutrino spin-flip
scattering via the intermediate
plasmon γ ∗ on the plasma
electromagnetic current J
∗

and a similar quantity for initial muon and tau neutrinos.

With the Lagrangian (5.76), the process is described by the Feynman diagram
shown in Fig. 5.2.
The technics of calculations of the neutrino spin-flip rate is rather standard. The
invariant amplitude for the process of the neutrino scattering off the k-th plasma
component can be written in the form

β
M(k) = −i e μν j(ν)
α
G αβ (Q) J(k) , (5.79)

α is the Fourier transform of the neutrino magnetic moment current,

where j(ν)

α
 
j(ν) = ν̄ R ( p  ) σ μα ν L ( p) Q μ ,

β
J(k) is the Fourier transform of the k-th plasma component electromagnetic current,
and Q = (q0 , q) is the four-momentum transferred. The only principal point is to
use the photon propagator G αβ (Q) with the plasma polarization effects taken into
account, see Sect. 4.4.

5.4.3 The Rate of Creation of the Right-Handed Neutrino

The value of physical interest is the rate of creation of the right-handed neutrino ν R ,
Γν R (E  ), with the fixed energy E  by all the left-handed neutrinos. This function
can be obtained by integration of the amplitude (5.79) squared over the states of
the initial left-handed neutrinos and over the states of the initial and final plasma
β
particles forming the electromagnetic current J(k)
148 5 Electromagnetic Interactions in External Active Media

Γν R (E  ) = Γν(k)
R
(E  ) , (5.80)
k
 
1
Γν(k) (E  ) = |M(k) |2 δ (4) ( p  + P  − p − P)
R
16 (2π)5 E 
s,s 
d3 P d P 
3  d3 p
× f k (E) 1 ∓ f k (E  ) f ν (E) . (5.81)
E E E

Here, p α = (E, p) and p α = (E  , p ) are the four-momenta of the initial and

final neutrinos, P α = (E, P) and α  
 P = (E , P ) are the four-momenta of the ini-
tial and final plasma particles; s,s  means the summation over the spins of these
particles, the index k = e, p, i, . . . corresponds to the type of the plasma parti-
cles (electrons, protons, free ions, etc.) with
 the distribution
 function f k (E), which
can be both fermions (the upper sign in 1 ∓ f k (E  ) ) and bosons (the lower sign);
 −1
f ν (E) = e(E−μ̃ν )/T + 1 is the Fermi—Dirac distribution function for the initial
left-handed neutrinos in the plasma restframe.
It is convenient to pass in Eq. (5.81) from integration over the initial neutrino
momentum p to the integration over the virtual plasmon momentum p − p  = Q =
(q0 , q), |q| ≡ q, using the relation:

d3 p 2π
f ν (E) =  q dq dq0 θ(−Q 2 ) θ(2E  + q0 − q) f ν (E  + q0 ) .
E E
Substituting the amplitude (5.79) squared into Eq. (5.81), one obtains

∞ 2E +q0
 μ2ν 
Γν R (E ) = dq0 q dq f ν (E  + q0 ) j(ν)
α α∗
j(ν)
8 π 2 E 2
−E  |q0 |

 ραβ (λ) ρα β  (λ ) 
× T ββ , (5.82)
(Q 2 − λ ) (Q 2 − ∗λ )
λ,λ

where the following tensor integral is introduced:


e2   α β∗
T αβ = J(k) J(k) d Φ , (5.83)
32 π 2  k s,s
d3 P d3 P  
dΦ = 
f k (P) 1 ∓ f k (P  ) δ (4) (P  − P − Q) .
EE

Further, we present the detailed calculation of the tensor T αβ . To use the covariant
properties of this tensor, one should write the distribution functions f k (P) in the
arbitrary frame
5.4 Electromagnetic Interactions of the Dirac Neutrino with a Magnetic Moment 149

 −1
(Pu) − μ̃
f k (P) = exp ±1 , (5.84)
T

where u α is the four-vector of the plasma velocity. This vector and the four-vector
Q α are the building bricks for constructing the tensor T αβ . This tensor is symmet-
α is real. The tensor is also orthogonal
ric because the electromagnetic current J(k)
to the four-vector Q α because of the electromagnetic current conservation. There
exist only two independent structures having these properties, which are the density
matrices (4.42) and (4.43), and thus one can write:

T αβ = A(t) ραβ (t) + A() ραβ () . (5.85)

Because of orthogonality of the tensors ραβ (t) and ραβ () , see Eq. (4.45), one
obtains

1 αβ e2 
α β∗
A(t) = T ραβ (t) = ραβ
(t)
J(k) J(k) d Φ , (5.86)
2 64 π 2
k s,s 
e2 
α β∗
A() αβ ()
= T ραβ = ραβ ()
J(k) J(k) d Φ . (5.87)
32 π 2  k s,s

As we show below, just these integrals (5.86) and (5.87) define the widths of
absorption (at q0 > 0) and creation (at q0 < 0) of a plasmon by the plasma particles.
Really, let us consider for definiteness the width of absorption of the transversal
β
plasmon by plasma particles forming the electromagnetic current J(k) . The amplitude
of the process has the form

M(k)(t) = −e εα (t) J(k)

α
. (5.88)

where εα (t) is the unit polarization four-vector. Performing standard calculations, one
obtains for the width of the plasmon absorption by all the components of plasma:

1 1
Γ(t)
abs
= |M(k)(t) |2 d Φ , (5.89)
32 π 2 q0 2 τ  k s,s

where the summation is made both over the kth types of the plasma particles and
over the polarizations of all particles participating in the process, τ for a plasmon
and s, s  for plasma particles.
Substituting the amplitude (5.88) into (5.89),

e2 
(t) α β∗
Γ(t)
abs
= ραβ J(k) J(k) d Φ , (5.90)
64 π 2 q0 
k s,s
150 5 Electromagnetic Interactions in External Active Media


2
τ (t) τ (t)
where ραβ (t) = εα εβ , and comparing it with Eq. (5.86), one can find the
τ =1
value
A(t) = q0 Γ(t)
abs
. (5.91)

Using the known relation [48] between the width of absorption of the transversal
plasmon and the imaginary part It of the eigenvalue t of the photon polarization
tensor αβ ,  
It (q0 ) = −q0 1 − e−q0 /T Γ(t)
abs
, (5.92)

we express the value A(t) in terms of It :

It  
A(t) = − −q /T
= −It 1 + f γ (q0 ) , (5.93)
1−e 0
 −1
where f γ (q0 ) = eq0 /T − 1 is the Bose–Einstein distribution function for a pho-
ton. This relation obtained in the case q0 > 0 is also correct for the case q0 < 0,
which corresponds to the transversal plasmon creation with the energy ω = −q0 > 0.
The connection should be used here between the imaginary part It and the width of
creation of the transversal plasmon:
 
It (ω) = −ω eω/T − 1 Γ(t)
cr
. (5.94)

It is essential also that the function It is odd:

It (−q0 ) = −It (q0 ) , (5.95)

and this is the feature of the retarded polarization operator.

Performing the similar calculations, one can see that the relation of the form (5.93)
is valid for the longitudinal plasmon also. It is necessary to remember that ραβ () =
−εα () εβ () , and  
I (q0 ) = q0 1 − e−q0 /T Γ()
abs
. (5.96)

Finally, we obtain the tensor T αβ in the form of decomposition over the density
matrices (4.42), (4.43):
  
T αβ = −It ραβ(t) − I ραβ() 1 + f γ (q0 ) , (5.97)

where It,  are the imaginary parts of the eigenvalues t,  of the photon polarization
tensor; f γ (q0 ) is the Bose–Einstein distribution function for a photon.
Substituting (5.97) into (5.82), using the orthogonality of the tensors ραβ(t) and
ρ αβ() , see Eq. (4.45), and taking into account the expressions for the contractions of
5.4 Electromagnetic Interactions of the Dirac Neutrino with a Magnetic Moment 151

the neutrino current with these tensors:

 
α β∗ (2E  + q0 )2
j(ν) j(ν) ραβ (t) = Q 4 − 1 ,
q2

α β∗ (2E  + q0 )2
j(ν) j(ν) ραβ () = −Q 4 ,
q2

one finally obtains for the rate of creation of the right-handed neutrino:

∞ 2E +q0
μ2ν
Γν R (E  ) = dq0 q 3 dq f ν (E  + q0 ) (2E  + q0 )2
16 π 2 E 2
−E  |q0 |
 2   
q02   q2
× 1− 1 + f γ (q0 ) 1− t −  . (5.98)
q2 (2E  + q0 )2

Here, the plasmon spectral densities are introduced:

−2 Iλ
λ = , (5.99)
(Q 2 − Rλ )2 + Iλ2

which are defined by the eigenvalues (4.46) of the photon polarization tensor (4.41).
The formula (5.98) is in agreement, to the notations, with the rate obtained in
Ref. [32] from the retarded self-energy operator of the right-handed neutrino. How-
ever, extracting from our general expression the electron contribution only, we obtain
the result which is larger by the factor of 2 than the corresponding formula in the
papers [39, 40]. It can be seen that an error was made there just in the first formula
defining the production rate Γ of a right-handed neutrino.
The formula (5.98) being obtained for the process of the neutrino interaction with
virtual photons, has in fact a more general sense, and can be used for neutrino-photon
processes in any optically active medium. We only need to identify the photon spectral
density functions λ . For example, in the medium where It → 0 in the space-like
region Q 2 < 0 corresponding to the refractive index values n > 1, the spectral
density function is transformed to δ-function, and we can reproduce the result of the
paper [49] devoted to the study of the Cherenkov radiation of transversal photons by
neutrinos.
If one formally takes the limit I → 0, the result obtained in Ref. [50] can be
reproduced, namely, as the authors believed, it would be the width of the Cherenkov
radiation and absorption of longitudinal photons by neutrinos in the space-like region
Q 2 < 0. However, the limit I → 0 itself is irrelevant for Q 2 < 0 in the real
astrophysical plasma conditions considered by those authors. As it was mentioned
in Refs. [39, 40], see also Fig. 4.5, the space-like branch of the longitudinal photon
152 5 Electromagnetic Interactions in External Active Media

mode developped a large imaginary part in the supernova core conditions. Thus,
taking the limit I → 0 leads to the strong overestimation of a result.

5.4.4 Contributions of Plasma Components into the Neutrino

Scattering Process

As it was mentioned above, an analysis of the neutrino chirality flip process has
to be performed with taking account of the neutrino scattering off various plasma
components: electrons, protons, free ions, etc. For the first step we consider the
contribution of the neutrino scattering off electrons into the right-handed neutrino
production rate. This means that we take into account the electron contribution only
into the function Iλ in the numerator of Eq. (5.99). It should be stressed however, that
the functions Rλ and Iλ in the denominator of Eq. (5.99) contain the contributions
of all plasma components. At this point our result for the neutrino scattering off
electrons differs from the result of Ref. [39, 40], where the electron contribution
only was taken both in the numerator and in the denominator of the plasmon spectral
densities.
As the analysis shows, see Sect. 4.4, the electron and proton contributions into
the imaginary parts Iλ of the eigenvalues λ of the photon polarization tensor are of
the same order of magnitude and have the same sign both for λ = t and for λ = ,
see Figs. 4.5 and 4.7. This fact itself should lead to a decreasing of the electron
contribution into the function Γν R (E  ). On the other hand, it is seen from Fig. 4.4,
that the electron and proton contributions into the real part R of the eigenvalue 
are of the same order of magnitude but have the opposite signs in the region where
the imaginary part of the electron contribution into the numerator of Eq. (5.99) is
relatively large. As a result, the contribution of the neutrino scattering off electrons
into the right-handed neutrino production rate, obtained by us, appears to be close to
the result of Ref. [39, 40], besides the above-mentioned factor of 2.
It is possible to consider similarly the contribution of the neutrino scattering off
protons into the right-handed neutrino production rate. In this case, we take the proton
contribution into the functions Iλ (4.57), (4.59) in the numerator of Eq. (5.99).
The results of our numerical analysis of the separate contributions of the neutrino
scattering off electrons and protons, as well as the total ν R production rate in the
typical conditions of the supernova core are presented in Fig. 5.3.
The plotted dimensionless creation width R(E  ) is defined by the expression

Γν R (E  ) = μ2ν μ3e T 3 R(E  ) . (5.100)

For comparison, the result of Ref. [40] is also shown in Fig. 5.3, illustrating a strong
underestimation of the neutrino chirality flip rate made by those authors.
We consider also the contribution of the neutrino scattering off free ions into
the ν R production rate. While the ions are believed to be absent in the supernova
5.4 Electromagnetic Interactions of the Dirac Neutrino with a Magnetic Moment 153

4
(E' ) 104
3

0
0 50 100 150 200
E' [MeV]

Fig. 5.3 Contributions from electrons (dashed line) and protons (dashdotted line) to the
dimensionless creation width R(E  ) of a right-handed neutrino and total width (solid line) for
plasma temperature T = 25 MeV and chemical potentials of electrons μe = 250 MeV and neutrinos
μ̃νe = 100 MeV. The dotted line indicates the result of Ref. [40]

core, a significant fraction of them could be presented e.g. in the upper layers of
the supernova envelope. It should be mentioned that longitudinal virtual plasmons
give the main contribution into the ν R production rate in this case. As is seen from
(i)
Eq. (4.61), the function I differs √from zero only in the narrow area Δx of the
variable x = q0 /q, namely, Δx ∼ T /m i  1, where m i is the ion mass. This
allows to perform calculations of the ion contribution into the ν R production rate
analitically, to obtain:
 
4E 2 + m 2D 4E 2
Δ Γν(i) (E  ) = μ2ν α Z i2 n i 
f ν (E ) ln − , (5.101)
R
m 2D 4E 2 + m 2D

where α is the fine structure constant, eZ i and n i are the charge and the density of ions,
 (k)
m D has a meaning of the Debye screening radius inversed, m 2D = k R (q0 = 0).
We remind that the summation is performed over all plasma components.
It is interesting to note that Eq. (5.101) obtained in the approximation of heavy
ions, describes rather satisfactory the proton contribution.
Given the function Γν R (E  ), one can calculate the total number of right-handed
neutrinos emitted per 1 MeV per unit time from the unit volume, i.e. the right-handed
neutrino energy spectrum:

dn ν R E 2

= Γν (E  ) . (5.102)
dE 2 π2 R
154 5 Electromagnetic Interactions in External Active Media

One can see from Eq. (5.102), that very narrow peak of the function Γν R (E  ) at
small neutrino energy, which was analysed in detail in Ref. [40], does not provide
a huge number of soft right-handed neutrino production, as it was declared in [40],
because of the factor E 2 .
The right-handed neutrino energy spectrum (5.102) can be useful for investi-
gations of possible mechanisms of the energy transfer from these neutrinos to the
outer layers of the supernova envelope. For example, a process is possible of the
inverse conversion of a part of right-handed neutrinos into left-handed ones, with
their subsequent absorption. Just these processes were proposed [51] and then inves-
tigated [52–54] as a possible mechanism for the stalled shock wave revival in the
supernova explosion. A consistent analysis of such scenario would be doubtful with-
out knowing the ν R energy spectrum (5.102). We discuss this question below in
Sect. 5.4.8.
The function Γν R (E  ) provides also the calculation of the spectral density of the
supernova core luminosity via right-handed neutrinos as follows:

dL ν R dn ν R  E 3
= V E = V Γν (E  ) . (5.103)
dE  dE  2 π2 R

Here, V is the volume of the neutrino-emitting region, V 4 × 1018 cm3 [55]. The
value dL ν R /dE  is presented in Fig. 5.4 for several values of the plasma temperature.

5.4.5 Illustration: Completely Degenerate Plasma at T = 0

In this section, we give a clear illustration of the fact that neutrino scattering by
protons dominates over their scattering by plasma electrons, basing on an analysis
of a simplified case of the completely degenerate plasma, T = 0.
The comparison of the typical parameters of the supernova core, where the tem-
perature is believed to be of order T 15–30 MeV, while the electron and neutrino
chemical potentials are μe 200–250 MeV and μ̃νe 100 MeV, respectively, shows
that the temperature is the smallest physical parameter.1 Thus, the limiting case of
the completely degenerate plasma, T = 0, seems to give a reasonable estimate. It is
remarkable that for the zero temperature limit the contributions from neutrino scat-
tering by protons and electrons to the neutrino creation probability can be evaluated
analytically using Eqs. (5.98) and (5.99) and the corresponding formulas of Sect. 4.4.
It is appropriate to analyse the function Γν R (E) defining the energy spectrum of
right-handed neutrinos (5.102).
The contribution of ultrarelativistic electrons to the function Γν R (E) in the case
T = 0 can be obtained from Eqs. (5.98) and (5.99) in the simple form:

1 Hereafter we consider neutrinos as a quasiequilibrium gas described by the distribution functions:

f ν (T, μ̃νe ) for the electron neutrinos, and f ν (T, 0) for the muon and tau neutrinos. This is believed
to be a rather good approximation inside the SN core during a few seconds after the collapse.
5.4 Electromagnetic Interactions of the Dirac Neutrino with a Magnetic Moment 155

5.0

' [erg/ (s . MeV)] 10–48

4.0

3.0

2.0
R /dE

1.0
dL

0
0 100 200 300 400
E' [MeV]

Fig. 5.4 Energy distributions of the right-handed neutrino luminosity for plasma temperatures
T = 35 MeV (solid line), T = 25 MeV (dashed line), T = 15 MeV (dashdotted line), T = 5 MeV
(dotted line) and for neutrino magnetic moment μν = 3 × 10−13 μB

μ2νe m 2γ
Γν(e) (E) = (μ̃νe − E) Θ(μ̃νe − E) , (5.104)
R
2π
where E is the right-handed neutrino energy, μνe is the effective electron neutrino
magnetic moment (5.78), μ̃νe is the electron neutrino chemical potential, m 2γ =
2 α μ2e /π is the squared mass of a transverse plasmon at T = 0, and Θ(x) is the step
function.
The analytical expression describing the proton contribution turns out to be more
complicated since it depends also on the proton mass. The plasma charge neu-
trality condition for T = 0 takes the form n p = n e− and ensures that the elec-
(e) ( p)
tron and proton Fermi momenta are equal: kF = kF . Then, the proton  chem-
( p)
ical potential coinciding with the Fermi energy is μ p = E F = m 2p + μ2e
and the proton contribution is expressed
 in terms of the proton Fermi velocity
( p) ( p)
vF = kF /E F = μe /μ p = μe / m p + μ2e . As a result, the proton contribution
2

can be expressed in the form:

μ2νe m 2γ μ̃νe E
Γν(Rp) (E) = ϕ p (y) , y= , 0  y  1. (5.105)
2π μ̃νe

Here, the function ϕ p (y) has different forms in two intervals: it is

1 + vF /3
ϕ p (y) = y, (5.106)
1 − vF

for 0  y  (1 − vF )/(1 + vF ), and

156 5 Electromagnetic Interactions in External Active Media

ϕ p (y)

Fig. 5.5 Plots of the function ϕ p (y) for various vF values. The dependence ϕe (y) = (1 − y) for
the electron contribution is reproduced for vF = 1 (dashed line). The value vF = 0.394 (solid
curve) corresponds to the effective proton mass m p 700 MeV. The case vF = 0 (dotted line)
corresponds to the limit of infinitely large proton mass (Figs. 5.5–5.12 reprinted from [42] with the
World Scientific Publishing Company’s permission.)

 
1−y (1 − vF )2
ϕ p (y) = 1− (1 − y) (1 + 2 y) , (5.107)
vF 12 y 2 vF

for (1 − vF )/(1 + vF )  y  1.
Note that the formal turn to the limit m p → 0, i.e. vF → 1, in Eqs. (5.105)–
(5.107) yields ϕ p (y) → ϕe (y) = (1 − y) θ(1 − y), where the function ϕe (y) can be
introduced in Eq. (5.104) in complete analogy with Eq. (5.105). Thus, as expected,
Eq. (5.104) for the electron contribution is reproduced.
In Fig. 5.5, the plots are shown of the function ϕ p (y) for vF = 1, vF = 0.394, and
vF = 0. The value vF = 0.394 corresponds to the effective proton mass m p 700
MeV in a plasma with a nuclear density 3 × 1014 g cm−3 (see Ref. [55], p. 152). The
value vF = 0 corresponds to the formal limit m p → ∞ for which this function is
also significantly simplified: ϕ p (y) → ϕ∞ (y) = y θ(1 − y).
The function Γν R (E) defined in Eq. (5.102) determines as well the right-handed
neutrino emissivity of a supernova core, i.e. the energy passed away by right-handed
neutrinos per 1 MeV of the neutrino energy spectrum per unit time from unit volume:

dn ν R E3
QνR = E = Γν (E) . (5.108)
dE 2 π2 R
According to Eqs. (5.102) and (5.108), the right-handed neutrino emissivity is
given by the formula

μ2νe m 2γ μ̃4νe  
QνR = y 3 ϕe (y) + ϕ p (y) . (5.109)
4 π3
5.4 Electromagnetic Interactions of the Dirac Neutrino with a Magnetic Moment 157

1.0

0.8

0.6
y3 ϕ (y )

0.4

0.2

0.0
0.00 0.2 0.2 0.6 0.8 1.0
y

Fig. 5.6 The function y 3 ϕ(y) defining the contributions from electrons (dashed line) and protons
with m p 700 MeV (solid line) and m p → ∞ (dotted line) to the right-handed neutrino emissivity
at T = 0

The difference between the electron and proton contributions to the quantity given
by Eq. (5.109) is illustrated in Fig. 5.6. It is clearly seen that the factor y 3 causes the
increasing of the proton contribution to the emissivity.

5.4.6 Uniform Ball Model for the Supernova Core

The spectral density of the supernova core luminosity via right-handed neutrinos is
defined as follows, see Eq. (5.103):

dL ν R dn ν R μ2ν m 2γ μ̃4ν
=V E = V e 3 e y 3 ϕ(num) (y, T ) . (5.110)
dE dE 4π
Here, m γ is the mass of a transverse plasmon,
 
2α π2 T 2
m 2γ = μe 2 + . (5.111)
π 3

The function ϕ(num) (y, T ) introduced in Eq. (5.110) similarly to Eqs. (5.105) and
(5.109) can be extracted from Ref. [41]. The function y 3 ϕ(num) (y, T ) is plotted in
Fig. 5.7 for two values of the averaged temperature and for the electron and electron-
neutrino chemical potentials μe 300 MeV and μ̃νe 160 MeV. We neglected
in our analysis [41] the contributions of the processes with the initial muon and
tau neutrinos. However, as will be shown below, these contributions appear to be
essential.
158 5 Electromagnetic Interactions in External Active Media

3.0

2.5
( y)
2.0
(num)

1.5
y 3ϕ

1.0

0.5

0
0 1 2 3 4
y

Fig. 5.7 The function y 3 ϕ(num) (y, T ) representing the result of the numerical calculation of the
right-handed neutrino emissivity at T = 30 MeV (dashed line) and T = 60 MeV (solid line)

A comparison of Figs. 5.6 and 5.7 shows that taking of a nonzero temperature
leads to a shift of the maximum of the energy distribution of the luminosity towards
higher energies of right-handed neutrinos. This additionally enhances the proton
contribution.
As a result, using the data on supernova S N 1987A, a new astrophysical limit was
imposed [41] on the electron-neutrino magnetic moment:

μν < (0.7 − 1.5) × 10−12 μB . (5.112)

This is a factor of two better than the previous constraint [39, 40]. We have to remind,
however, that both the previous and this improved bound on the electron-neutrino
magnetic moment were based on a very simplified model of the supernova core as
the uniform ball with some averaged values of physical parameters. In addition, the
parameter values were set too high. For example, the upper limit 1.5 × 10−12 μB in
Eq. (5.112) corresponds to the SN core temperature 30 MeV, while the limit 0.7 ×
10−12 μB corresponds to the temperature 60 MeV. As is seen from Fig. 5.7, the right-
handed neutrino emissivity grows with temperature very rapidly. However, according
to recent simulations of the SN explosion, the temperature values inside the SN core
are believed not to exceed 40 MeV, see e.g. Fig. 5.8. Anyway, taking account of the
radial distribution of physical parameters inside the SN core would give more solid
results.
5.4 Electromagnetic Interactions of the Dirac Neutrino with a Magnetic Moment 159

40

T [MeV] 30

20

10

0
0 5 10 15 20
R [km]

Fig. 5.8 The radial distribution for the temperature within the SN core at the moment t = 1.0 s
after the bounce, Ref. [56]

5.4.7 Models of the Supernova Core with Radial Distributions

of Physical Parameters: Limits on the Neutrino
Magnetic Moment

In this section we make the estimation of the upper bound on the Dirac neutrino
magnetic moment by a more reliable way, with taking account of radial distributions
and time dependences of physical parameters from realistic models of the SN core.
Here we consider the models in the inverse chronology.

5.4.7.1 The Model of the O-Ne-Mg Core Collapse SN

This model was developed by H.-Th. Janka with collaborators who presented us the
results of their simulations [56] of the O-Ne-Mg core collapse supernovae which
were a continuation of their model simulations [57, 58]. The successful explosion
results for this case were independently confirmed by the Arizona/Princeton SN
modelling group [59, 60], which found very similar results. So we were provided
with a model whose explosion behavior was comparatively well understood and
generally accepted.
We should stress that this O-Ne-Mg core collapse model (for the initial stellar
mass of 8.8 M ) is not applicable directly to S N 1987A which was 15 − 20 M
prior to collapse and according to the evolution theory it had a collapsing core which
consisted of iron-peak elements.
We redefine Eq. (5.103), where, instead of multiplying by the volume of the
neutrino-emitting region V , we integrate over this volume to obtain the spectral
density of the energy luminosity of a supernova core via right-handed neutrinos:
160 5 Electromagnetic Interactions in External Active Media

250

200

150
μ [M eV]

100

50

0
0 5 10 15 20
R [km]

Fig. 5.9 The radial distributions for the chemical potentials of electrons (solid line) and electron
neutrinos (dashed line) within the SN core at the moment t = 1.0 s after the bounce


dL ν R E3
= dV Γν (E) . (5.113)
dE 2 π2 R

Here, taking the values defined in Eqs. (5.98) and (5.99) and the corresponding
formulas of Sect. 4.4, we take account of their dependence on the radius R and time t.
A comprehensive set of parameter distributions used in our estimation includes the
profiles [56] of the density ρ, the temperature T , the electron fraction Ye , the fractions
of electron neutrinos Yνe , electron anti-neutrinos Yν̄e , and the fractions Yνx for one
kind of heavy-lepton neutrino or antineutrino (νx = νμ,τ , ν̄μ,τ ), which are treated
identically. The time evolution of the parameter distributions is calculated [56] within
the interval until ∼ 2 s after the bounce. For the sake of illustration, we present
in Figs. 5.8, 5.9 and 5.10 the radial distributions within the SN core, from 0 to
20 km, at the moment t = 1.0 s after the bounce. The plots are presented for the
temperature [56], for the chemical potentials of electrons μe and electron neutrinos
μ̃νe (calculated on the base of the data of Ref. [56]), and for the proton nonrelativistic
chemical potential μ∗p = μ p − m ∗N defining the degeneracy of protons (calculated
on the base of the data of Ref. [56] and the effective nucleon mass m ∗N in plasma, see
Ref. [55], p. 152).
To analyse the influence of the right-handed neutrino emission on the SN energy
loss, we also used the time evolution of the total luminosity of all species of
left-handed neutrinos [56], presented in Fig. 5.11.
Integrating Eq. (5.113) over the neutrino energy, one obtains the time evolution
of the right-handed neutrino luminosity:

 ∞
1
L ν R (t) = dV dE E 3 Γν R (E) . (5.114)
2 π2
0
5.4 Electromagnetic Interactions of the Dirac Neutrino with a Magnetic Moment 161

100

50
μ p − m ∗N [M eV]

−50

−100
0 5 10 15 20
R [km]

Fig. 5.10 The radial distribution for the proton nonrelativistic chemical potential μ∗p = μ p − m ∗N
within the SN core at the moment t = 1.0 s after the bounce

2.0
L totL [erg/sec] × 10 −53

1.5

1.0

0.5

0
0.0 0.5 1.0 1.5 2.0
t [sec]

Fig. 5.11 The time evolution of the total luminosity of all active neutrino species, Ref. [56]

The right-handed neutrino is a novel cooling agent which would have to compete
with the energy-loss via active neutrino species in order to affect the total cooling
time scale significantly. Therefore, the observed S N 1987A signal duration indicates
that a novel energy-loss via right-handed neutrinos is bounded by

L ν R < L νL , (5.115)

and we believe this estimation to be applicable also to the considered O-Ne-Mg core
collapse model. Within the considered time interval until 2 s after the bounce, one
obtains from Eqs. (5.114), (5.115) the time-dependent upper bound on the combi-
nation of the effective magnetic moments of the electron, muon and tau neutrinos.
Assuming for simplicity that these effective magnetic moments are equal, one obtains
162 5 Electromagnetic Interactions in External Active Media

5.0

4.5

µ−12 [µ B ] 4.0

3.5

3.0

2.5

2.0
0.0 0.5 1.0 1.5 2.0
t [sec]

Fig. 5.12 The time evolution of the upper bound on the neutrino magnetic moment within the time
interval until 2 s after the bounce (in assumption that the effective magnetic moments of electron,
muon and tau neutrinos are equal)

the time evolution of the upper bound on some flavor-averaged neutrino magnetic
moment μ̄ν shown in Fig. 5.12, where μ̄12 = μ̄ν /(10−12 μB ).
As is seen from Fig. 5.12, the averaged upper bound tends to some value, providing
the limit
μ̄ν < 2.4 × 10−12 μB . (5.116)

In a general case the combined limit on the effective magnetic moments of the
electron, muon and tau neutrinos is
   1/2
μ2νe + 0.71 μ2νμ + μ2ντ < 3.7 × 10−12 μB , (5.117)

where the effective magnetic moments are defined according to Eq. (5.78). This limit
is less stringent than the bound (5.112) obtained in the frame of the uniform ball
model for the SN core, but it is surely more reliable. Additionally, the upper bound
on the effective magnetic moments of muon and tau neutrinos is established.

5.4.7.2 Earlier Models of the SN Explosion

The similar procedure of evaluation was performed with using of the data of the
model [61] of the two-dimensional hydrodynamic core-collapse supernova simula-
tion for a 15 M star. Namely, the radial distributions of parameters at the moments
t = 0.2, 0.4, 0.6, 0.8 s after the bounce in the model s15Gio_32.a were taken from
Fig. 40 of Ref. [61]. Additionally, the fraction of electron neutrinos was evaluated as
Yνe (1/5) Ye . Calculating the right-handed neutrino luminosity with those para-
meters and putting the limit (5.115), where the total luminosity via active neutrino
5.4 Electromagnetic Interactions of the Dirac Neutrino with a Magnetic Moment 163

species L ν L in that model can be taken from Fig. 42 of Ref. [61], one obtains that
the upper bound on the flavor-averaged neutrino magnetic moment μ̄ν also varies in
time as in the previous case. The time-averaged upper bound on μ̄ν corresponding
to the interval 0.4–0.8 s, is:

μ̄ν < 2.7 × 10−12 μB , (5.118)

to be compared with the limit (5.116).

Using the results of Ref. [62] where the thermal and chemical evolution during
the Kelvin-Helmholtz phase of the birth of a neutron star was studied, taking the data
from Figs. 9 and 14, we have obtained the time-averaged upper bound on μ̄ν for the
time interval 1–10 s of the post-bounce evolution in the form:

μ̄ν < 1.2 × 10−12 μB . (5.119)

We also used the results of Ref. [63] where the numerical simulations were per-
formed of the neutrino-driven deleptonization and cooling of newly formed, hot,
lepton-rich neutron star. Using the data presented in Figs. 3–9 on the SBH model
(of the hot star with a “small” baryonic mass), we have evaluated the time-averaged
upper bound on μ̄ν for the time interval 0.5–5 s after the bounce in the form:

μ̄ν < 1.1 × 10−12 μB . (5.120)

One can summarize that the upper bound on the flavor- and time-averaged neutrino
magnetic moment at the Kelvin-Helmholtz phase of the supernova explosion occurs
to be
μ̄ν < (1.1 − 2.7) × 10−12 μB , (5.121)

depending on the explosion model.

5.4.8 Possible Effect of the Neutrino Magnetic Moment:

Shock-Wave Revival in a Supernova Explosion

Two basic problems arise in numerical simulation of a supernova explosion

[55, 61, 64–66]. First, the mechanism of damped-shock-wave stimulation has not yet
been developed conclusively, but, without this mechanism, an explosion can hardly
occur. We recall that the damping of a shock wave is due largely to the loss of energy
spent on the dissociation of nuclei. Second, the energy deposition even in the case of
a “successful” theoretical supernova explosion proves to be substantially less than
the observed envelope kinetic energy of about ∼ 1051 erg [so-called FOE (ten to the
Fifty-One Ergs) problem]. In order to describe self-consistently explosion dynamics,
164 5 Electromagnetic Interactions in External Active Media

it is therefore necessary that, via some mechanism, the neutrino flux going from a
supernova central part transfer an energy on the order ∼ 1051 erg to the envelope.
The mechanism proposed by Dar [51] and based on the assumption that the neu-
trino magnetic moment is not overly small is one of the possible means for solving
the above problems. We note that this mechanism is operative only for Dirac but not
for Majorana neutrinos. Left-handed electron neutrinos produced abundantly in the
collapsing supernova core form a degenerate neutrino gas such that typical values
of its chemical potential fall within the range μ̃νe ∼ 150–200 MeV [55]. Since the
values of μ̃νe are much higher than typical temperature values of T ∼ 30 MeV, the
density of electron neutrinos in the supernova core exceeds substantially the densities
of neutrinos belonging to any other flavor. Part of left-handed electron neutrinos are
converted into right-handed neutrinos via magnetic-moment interaction with plasma
electrons and protons. In turn, right-handed neutrinos, which are sterile with respect
to weak interaction, escape freely from the supernova central part if the neutrino mag-
netic moment lies in the range μν < 10−11 μB . Some of these neutrinos may again
transform into left-handed neutrinos owing to magnetic-moment interaction with a
magnetic field in the envelope of the supernova core. According to currently preva-
lent ideas, the magnetic-field strength there may reach high values, on the scale of the
critical value of Be = m 2e /e 4.41 × 1013 G, or even higher [67–69]. Newly pro-
duced left-handed neutrinos can transfer additional energy to the supernova envelope
upon undergoing absorption in the course of νe n → e− p beta processes.
A sufficient motivation for reconsidering the Dar mechanism has appeared after
publication of the papers [41, 42] where it was shown that the flux and luminosity of
right-handed neutrinos from the supernova central part were strongly underestimated
in previous studies. In this section, we analyze the ν L → ν R → ν L double conversion
of the neutrino helicity under supernova conditions and consider the possibility of
stimulating a damped shock wave via this process.
At typical values of the supernova-core parameters (a temperature of T 30 MeV;
electron and electron- neutrino chemical potentials of μe 300 MeV and μ̃νe 160
MeV, respectively; and a volume of V 4×1018 cm3 [55]), the integrated luminosity
of right-handed neutrinos is
 2
erg μν
L ν R = 4 × 10 51
. (5.122)
s 3 × 10−13 μB

The energies of right-handed neutrinos that escaped from the core are on the same
order of magnitude as the chemical potential of left-handed neutrinos captured in the
core, E ν ∼ 100–200 MeV.
For the sake of definiteness, we henceforth set the neutrino magnetic moment to
μν = 3 × 10−13 μB . On one hand, this value is sufficiently small for the dynamics of
the supernova core to remain undistorted; on the other hand, it ensures the required
level of luminosity in (5.122).
If it were possible to convert the energy of right- handed neutrinos into the energy
of left-handed neutrinos, for example, via the well-known mechanism of spin oscil-
5.4 Electromagnetic Interactions of the Dirac Neutrino with a Magnetic Moment 165

lations, then, within the typical shock-wave-stagnation time of about a few tenths of
a second, an additional energy of about 1051 erg would be injected into the supernova
envelope. We recall that we deal here with electron neutrinos, whose absorption in
the envelope is due to beta processes.
We consider the part of the supernova envelope between the neutrinosphere
(of radius Rν ) and the shock-wave-stagnation region (of radius Rs ). According to
currently prevalent ideas, typical values of Rν and Rs change only slightly within
the stagnation time, amounting to Rν ∼ 20–50 km and Rs ∼ 100–200 km (see, for
example, [61]). If a rather strong magnetic field is present in this region, neutrino
spin oscillations, which, under certain conditions, may have a resonance character,
occur.
The effect of a magnetic field on a neutrino that has a magnetic moment can be
the most conveniently illustrated in terms of the equation that describes neutrino-
spin evolution in a uniform external magnetic field. With allowance for the addi-
tional energy ΔE L(e) that a left-handed electron neutrino acquires in a medium, see
Eqs. (4.80) and (4.83), the spin-evolution equation can be represented in the form
[24, 25, 53, 70, 71]
      
∂ νR 0 μν B⊥ νR
i = Ê 0 + (e) , (5.123)
∂t νL μν B⊥ ΔE L νL

where  
(e) 3 GF ρ 4 1
ΔE L = √ Ye + Yνe − , (5.124)
2 mN 3 3

Here, ρ/m N = n B is the nucleon density; Ye = n e /n B = n p /n B ; Yνe = n νe /n B ,

with n e, p,νe standing for the electron, proton, and neutrino densities; and B⊥ is the
transverse component of the magnetic field with respect to the direction of neutrino
motion.
It should be explained why use is made here of expression (5.124) for the additional
energy of left-handed electron neutrinos in an unpolarized medium, even though, in
general, electrons must at least be partly polarized in a field on the scale of Be . The
following considerations prove the validity of the unpolarized-medium approxima-
tion in this case. As is well known, electron states in a magnetic field that correspond
to all Landau levels, with the exception of the ground one, are doubly degenerate in
the spin projection onto the magnetic-field direction and, hence, do not contribute
to medium polarization. In order to assess the degree of polarization, it is therefore
necessary to estimate the fraction of electrons that populate the ground Landau level
and whose spins are not compensated. Under conditions typical of the supernova-
envelope region being considered, we have μe 5–10 MeV (see, for example, [61]);
taking the ratio of the concentration of electrons populating the ground Landau
level, n 0 eBμe /(2π 2 ) (see, for example, [72]), to the total electron concentration,
n μ3e /(3π 2 ), we estimate the degree of medium polarization at
166 5 Electromagnetic Interactions in External Active Media

n0 eB B
P∼  2 ∼ 10−2 . (5.125)
n μe Be

Thus, the use of the unpolarized-medium approximation is legitimate at magnetic-

field strengths around B ∼ Be , which are used here. A more rigorous condition of
weak plasma magnetization under which the influence of the magnetic field on the
polarization of the medium can be neglected is formulated as

(3π 2 n e )2/3  ne 2/3

B 0.6 × 1016 G . (5.126)
e 1033 cm−3

Expression (5.124) for the additional energy ΔE L(e) of left-handed electron neu-
trinos deserves a more detailed analysis. It is noteworthy that the discussed energy
can appear to be exactly zero in the supernova-envelope region of our interest, and
this is in turn the condition of the ν R → ν L resonance transition. Since the neutrino
density is rather low in the supernova envelope, the quantity Yνe in expression (5.124)
can be disregarded, in which case the resonance condition is written as Ye = 1/3.
We note that, in the supernova envelope, Ye takes values characteristic of collaps-
ing matter, Ye ∼ 0.4–0.5. However, a shock wave causing the dissociation of heavy
nuclei renders matter more transparent to neutrinos, thus leading to a so-called short
neutrino burst and, hence, to a considerable deleptonization of matter in this region.
According to existing estimates, the radial distribution of Ye develops a characteristic
dip, where Ye may decrease to values of about 0.1 (see, for example, [61, 65]). Thus,
a point where Ye acquires a value of 1/3 does inevitably exist. It is noteworthy that
there is only one such point where dYe /dr > 0 (see [61, 65]).
We emphasize that expression (5.124) refers only to the electron neutrino, in
which case the amplitude for its scattering on medium electrons features channels of
exchange of both a neutral Z boson and a charged W boson. For the muon neutrino
and for the tau neutrino, which are scattered on electrons only via the exchange of a
neutral Z boson, the additional energy has the form (4.84), or:

(μ,τ ) GF ρ
ΔE L =−√ (1 − Ye ) , (5.127)
2 mN

that is, it does not vanish anywhere, so that the above resonance transition is
impossible.
A qualitative character of the dependence Ye (r ) according to [61] is depicted in
Fig. 5.13.
We note that the condition Ye = 1/3 is necessary for the resonance conversion
of right-handed neutrinos into left-handed ones, but it is not sufficient. In addition,
fulfillment of the so-called adiabaticity condition is required. Its meaning is the
following: upon moving off the resonance point by a distance of about one oscillation
(e)
wavelength, the diagonal element ΔE L in Eq. (5.123) at least must not exceed the
off-diagonal element μν B⊥ . This leads to the condition [52]
5.4 Electromagnetic Interactions of the Dirac Neutrino with a Magnetic Moment 167

 (e)
1/2  1/2
dΔE L 3 G F ρ dYe
μν B⊥  √ . (5.128)
dr 2 m N dr

In the region being considered, typical parameter values are the following (see
[61, 65]):
dYe
∼ 10−8 cm−1 , ρ ∼ 1010 g cm−3 . (5.129)
dr
For the magnetic-field strength ensuring fulfillment of the resonance condition, we
obtain
  1/2
10−13 μB ρ
B⊥  2.6 × 10 G 13
μν 1010 g cm−3
 1/2
dYe
× × 108 cm . (5.130)
dr

Thus, our analysis has revealed that, if the neutrino has a magnetic moment in the
range
10−13 μB < μν < 10−12 μB (5.131)

and if a magnetic field of strength about 1013 G exists in the region Rν < R < Rs ,
the mechanism of the double conversion of the neutrino helicity, ν L → ν R → ν L ,
according to Dar’s scenario is operative. At energies estimated at E ν ∼ 100–200 MeV,
the neutrino mean free path with respect to beta processes is
 2
1 150 MeV
λ 800 m . (5.132)
1 − Ye Eν

Fig. 5.13 Qualitative charac-

ter of the radial distribution 0.5
of Ye (r ) approximately after
0.1–0.2 s from the generation
0.4
of a shock wave featuring a dip
caused by a “short” neutrino
burst (see, for example [61]). 0.3
The dashed line corresponds Ye
to the value of Ye = 1/3
0.2

0.1

10 100 1000
r [km]
168 5 Electromagnetic Interactions in External Active Media

Therefore, the additional energy

ΔE L ν R Δt ∼ 1051 erg (5.133)

is injected into the supernova-envelope region within the time of shock-wave

stagnation, Δt ∼ 0.2–0.4 s. This solves the FOE problem.

5.4.9 Possible Effect of the Neutrino Magnetic Moment:

Neutrino Pulsar

If the neutrino magnetic moment is less than the values of the range (5.131), the
conversion of sterile neutrinos produced in the supernova core into active ones
through the scattering mechanism of the Dirac neutrino magnetic moment with the
microscopic electromagnetic field of a virtual plasmon (5.73), did not influence
the supernova explosion dynamics. In this case, the process of the neutrino helic-
ity flip in a strong magnetic field of the supernova envelope can lead to interesting
observational consequences when the expected neutrino signal from an imminent
supernova explosion is studied in detail [73, 74].
According to existing views, during the explosion of a Galactic supernova at
a distance up to 10 kpc, the expected number of neutrino events in the Super-
Kamiokande detector will be ∼104 . This will allow the time evolution of the neutrino
flux to be recorded with a good accuracy.
In the presence of a sufficiently strong magnetic field in the supernova envelope,
not only the above-mentioned conversion of right-handed neutrinos into left-handed
ones, ν R → ν L [51, 52], but also the conversion of active electron neutrinos and
antineutrinos of the main neutrino flux into a form sterile with respect to weak
interactions, ν L → ν R , ν̄ R → ν̄ L , is possible.
Numerical analysis of Eq. (5.123) shows that after its passage through the
resonance region (Ye = 1/3), the flux of left-handed neutrinos is attenuated as a
result of the above conversion by the factor W L L , which has the meaning of the
survival probability of left-handed neutrinos, νeL → νeL , or, in other words, the
transparency. Figure 5.14 shows the characteristic variation in W L L when passing
through the resonance point (placed here at the coordinate origin) for various mag-
netic field strengths. We see that the supernova envelope in the presence of a suf-
ficiently strong magnetic field is virtually opaque to active electron neutrinos and
antineutrinos, which can cause the expected neutrino signal from the supernova to
be attenuated.
A more detailed analysis of the numerical solution of Eq. (5.123) allows us to
establish a relationship between the magnetic field strength and parameters of the
medium in the supernova envelope, on the one hand, and the survival probability of
active neutrinos W L L , on the other hand. Using typical scales of parameters in the
region under consideration [61, 65]
5.4 Electromagnetic Interactions of the Dirac Neutrino with a Magnetic Moment 169

WLL
1.0

a
0.8

b
0.6

0.4

0.2 c

x
–10 −5 5 10 15 20

Fig. 5.14 Pattern of variations in W L L , the survival probability of left-handed neutrinos, νeL →
νeL , (transparency), with distance x (in arbitrary units) when passing through the resonance point
placed at the coordinate origin for several magnetic field strengths: B = 0.2 Be (a), B = 0.5 Be
(b), B = Be (c). To be specific, the neutrino magnetic moment is assumed to be 10−13 μB , the
density is 1010 g cm−3 , and the gradient of the electron fraction is dYe /dr 10−7 cm−1

dYe
∼ 10−7 cm−1 , ρ ∼ 1010 g cm−3 , (5.134)
dr
we find an approximation formula,
 −13 
B⊥ (t) 10 μB
= f (W L L )
Be μν
 1/2  1/2
ρ(t) dYe
× (t) × 10 cm
7
. (5.135)
1010 g cm−3 dr

Here, the factor

(1 − W L L )0.62
f (W L L ) = 0.88 (5.136)
(W L L )0.13

characterizes the degree of adiabaticity of the conversion process. The literal adi-
abaticity corresponds to the limit f → ∞ when W L L → 0; in this case, the
left-handed neutrinos are completely converted into right-handed ones, W L R =
(1 − W L L ) → 1.
The conservative value of 10−13 μB introduced in Eq. (5.135) as the scale for the
neutrino magnetic moment was chosen in order not to distort the supernova explosion
dynamics. Thus, we can use the parameters of the explosion model without allowance
for the influence of the neutrino magnetic moment.
170 5 Electromagnetic Interactions in External Active Media

Our analysis based on detailed data on the radial distributions and time evolution of
physical properties in a supernova core obtained in the specific model of a successful
explosion [56] showed that the gradient of the electron fraction dYe /dr in Eq. (5.135)
grows fairly rapidly with time at point Ye = 1/3 and, thus, the envelope becomes
more transparent to active neutrinos at a fixed magnetic field strength. This means
that the neutrino signal from the supernova can be attenuated within some limited
time interval after its explosion.
Thus, if the Dirac neutrino had a magnetic moment and if the magnetic field
in the supernova envelope were sufficiently strong, then the characteristic effect of
a significant attenuation of the initial neutrino signal intensity peak predicted by
supernova models could take place. For example, there would be a tenfold reduction
in the neutrino signal (W L L = 0.1) for typical parameters of the medium at a magnetic
field strength
 
10−13 μB
B⊥ = 4.9 × 10 G 13
μν
 1/2  1/2
ρ dYe
× × 10 7
cm . (5.137)
1010 g cm−3 dr

Note that the possible strengths of a magnetic field generated in a supernova envelope
are believed to reach 1016 G [67, 68, 75–78].
Note another possible interesting manifestation of the neutrino magnetic moment.
If a magnetar with a poloidal magnetic field of 1014 − 1015 G is formed during a
supernova explosion, then, given that Eqs. (5.123) and (5.135) contain the transverse
magnetic field component B⊥ , the neutrinos can avoid the conversion of their helicity
only in a narrow region near the poles. When the nascent magnetar rotates around
dN dt

Fig. 5.15 Illustration of the pulsating behavior of the neutrino signal from a nascent magnetar
rotating around an axis that does not coincide with its magnetic moment, a neutrino pulsar
5.4 Electromagnetic Interactions of the Dirac Neutrino with a Magnetic Moment 171

an axis that does not coincide with its magnetic moment and if we are lucky with the
orientation of the rotation axis, the neutrino signal will have a pulsating behavior, as
is illustrated in Fig. 5.15, i.e., a kind of a neutrino pulsar can be observed.
It should be noted that, strictly speaking, the described influence of a strong
magnetic field when the neutrino has a magnetic moment on the time evolution of
the neutrino signal is incomplete without allowance for the effects of neutrino flavor
oscillations (see, e.g., [79]). The combined action of these effects on the neutrino
flux requires a special study.

References

1. N.P. Klepikov, Zh Eksp, Teor. Fiz. 26, 19 (1954)

2. P.A. Sturrock, Astrophys. J. 164, 529 (1971)
3. E. Tadermaru, Astrophys. J. 183, 625 (1973)
4. A.E. Shabad, Ann. Phys. (N.Y.) 90, 166 (1975)
5. M.A. Ruderman, P.S. Sutherland, Astrophys. J. 196, 51 (1975)
6. V.S. Beskin, Astrofizika 18, 439 (1982) [Astrophysics 18, 266 (1983)]
7. J.K. Daugherty, A.K. Harding, Astrophys. J. 273, 761 (1983)
8. V.B. Berestetskii, E.M. Lifshitz, L.P. Pitaevskii, Quantum Electrodynamics, 2nd edn. (Perga-
mon, Oxford, 1982)
9. V.I. Ritus, Quantum Effects of the Interaction of Elementary Particles with an Intense Electro-
magnetic Field (in Russian). in Quantum Electrodynamics of Phenomena in an Intense Field,
Proceedings of the P.N. Lebedev Physical Institute, vol. 111 (Nauka, Moscow, 1979) pp. 5–151
10. B.W. Lee, R.E. Shrock, Phys. Rev. D 16, 1444 (1977)
11. K. Fujikawa, R.E. Shrock, Phys. Rev. Lett. 45, 963 (1980)
12. E.M. Lipmanov, Yad. Fiz. 6, 541 (1967) [Sov. J. Nucl. Phys. 6, 395 (1968)]
13. E.M. Lipmanov, N.V. Mikheev, Pis’ma. Zh. Eksp. Teor. Fiz. 7, 139 (1968) [JETP Lett. 7, 107
(1968)]
14. E.M. Lipmanov, Zh. Eksp. Teor. Fiz. 55, 2245 (1968) [Sov. Phys. JETP 28, 1191 (1969)]
15. J.C. Pati, A. Salam, Phys. Rev. D 10, 275 (1974)
16. M.A.B. Bég, R. Budny, R.N. Mohapatra, A. Sirlin, Phys. Rev. Lett. 38, 1252 (1977)
17. R.N. Mohapatra, J.C. Pati, Phys. Rev. D 11, 566 (1975)
18. R.N. Mohapatra, J.C. Pati, Phys. Rev. D 11, 2558 (1975)
19. G. Senjanović, R.N. Mohapatra, Phys. Rev. D 12, 1502 (1975)
20. J.E. Kim, Phys. Rev. D 14, 3000 (1976)
21. W.J. Marciano, A.I. Sanda, Phys. Lett. B 67, 303 (1977)
22. M.A.B. Bég, W.J. Marciano, M. Ruderman, Phys. Rev. D 17, 1395 (1978)
23. A. Cisneros, Astrophys. & Space Sci. 10, 87 (1971)
24. M.B. Voloshin, M.I. Vysotsky, L.B. Okun, Yad. Fiz. 44, 677 (1986) [Sov. J. Nucl. Phys. 44,
440 (1986)]
25. M.B. Voloshin, M.I. Vysotsky, L.B. Okun, Zh. Eksp. Teor. Fiz. 91, 754 (1986) [Sov. Phys.
JETP 64, 446 (1986)]; Erratum: Zh. Eksp. Teor. Fiz. 92, 368 (1987) [Sov. Phys. JETP 65, 209
(1987)]
26. R.B. Clark, R.D. Pedigo, Phys. Rev. D 8, 2261 (1973)
27. M. Radomski, Phys. Rev. D 12, 2208 (1975)
28. G.G. Raffelt, Phys. Rept. 320, 319 (1999)
29. G.G. Raffelt, Phys. Rev. Lett. 64, 2856 (1990)
30. G.G. Raffelt, Astrophys. J. 365, 559
31. M. Fukugita, S. Yazaki, Phys. Rev. D 36, 3817 (1987)
172 5 Electromagnetic Interactions in External Active Media

32. P. Elmfors, K. Enqvist, G. Raffelt, G. Sigl, Nucl. Phys. B 503, 3 (1997)

33. A.G. Beda, V.B. Brudanin, V.G. Egorov, D.V. Medvedev, M.V. Shirchenko, A.S. Starostin,
(GEMMA Collab.). Phys. Part. Nucl. Lett. 7, 406 (2010)
34. I. Goldman, Y. Aharonov, G. Alexander, S. Nussinov, Phys. Rev. Lett. 60, 1789 (1988)
35. J.M. Lattimer, J. Cooperstein, Phys. Rev. Lett. 61, 23 (1988)
36. R. Barbieri, R.N. Mohapatra, Phys. Rev. Lett. 61, 27 (1988)
37. D. Nötzold, Phys. Rev. D 38, 1658 (1988)
38. A. Goyal, S. Dutta, S.R. Choudhury, Phys. Lett. B 346, 312 (1995)
39. A. Ayala, J.C. D’Olivo, M. Torres, Phys. Rev. D 59, 111901 (1999)
40. A. Ayala, J.C. D’Olivo, M. Torres, Nucl. Phys. B 564, 204 (2000)
41. A.V. Kuznetsov, N.V. Mikheev, J. Cosm, Astropart. Phys. 11, 031 (2007)
42. A.V. Kuznetsov, N.V. Mikheev, A.A. Okrugin, Int. J. Mod. Phys. A 24, 5977 (2009)
43. A.V. Kuznetsov, N.V. Mikheev, A.A. Okrugin, Zh. Eksp. Teor. Fiz. 138, 80 (2010) [J. Exp.
Theor. Phys. 111, 70 (2010)]
44. B. Pontecorvo, Zh. Eksp. Teor. Fiz. 33, 549 (1957) [Sov. Phys. JETP 6, 429 (1957)]
45. B. Pontecorvo, Zh. Eksp. Teor. Fiz. 34, 247 (1958) [Sov. Phys. JETP 7, 172 (1958)]
46. Z. Maki, M. Nakagawa, S. Sakata, Prog. Theor. Phys. 28, 870 (1962)
47. B. Pontecorvo, Zh. Eksp. Teor. Fiz. 53, 1717 (1967) [Sov. Phys. JETP 26, 984 (1968)]
48. H.A. Weldon, Phys. Rev. D 28, 2007 (1983)
49. W. Grimus, H. Neufeld, Phys. Lett. B 315, 129 (1993)
50. S. Mohanty, S. Sahu, Neutrino helicity flip by Cerenkov emission and absorption of plasmons
in supernova, arXiv: hep-ph/9710404 (1997)
51. A. Dar, Neutrino magnetic moment may solve the supernovae problem, Princeton, Institute
Advanced Study Preprint-87-0178 (unpublished) (1987)
52. M.B. Voloshin, Phys. Lett. B 209, 360 (1988)
53. L.B. Okun, Yad. Fiz. 48, 1519 (1988) [Sov. J. Nucl. Phys. 48, 967 (1988)]
54. S.I. Blinnikov, L.B. Okun, Pis’ma Astron. Zh. 14, 867 (1988) [Sov. Astron. Lett. 14, 368
(1988)]
55. G.G. Raffelt, Stars as Laboratories for Fundamental Physics (Univ. Chicago Press, Chicago,
1996)
56. B. Müller, H.-Th. Janka, Unpublished results (private communication)
57. F.S. Kitaura, H.-Th. Janka, W. Hillebrandt. Astron. Astrophys. 450, 345 (2006)
58. H.-Th. Janka, B. Müller, F.S. Kitaura, R. Buras. Astron. Astrophys. 485, 199 (2008)
59. L. Dessart, A. Burrows, C.D. Ott, E. Livne, S.-C. Yoon, N. Langer, Astrophys. J. 644, 1063
(2006)
60. A. Burrows, L. Dessart, C.D. Ott, E. Livne, Phys. Rep. 442, 23 (2007)
61. R. Buras, M. Rampp, H.-Th. Janka, K. Kifonidis. Astron. Astrophys. 447, 1049 (2006)
62. J.A. Pons, S. Reddy, M. Prakash, J.M. Lattimer, J.A. Miralles, Astrophys. J. 513, 780 (1999)
63. W. Keil, H.-Th. Janka. Astron. Astrophys. 296, 145 (1995)
64. V.S. Imshennik, D.K. Nadyozhin, Usp. Fiz. Nauk 156, 561 (1988) [Sov. Phys. Usp. 31, 1040
(1988)]
65. H.A. Bethe, Rev. Mod. Phys. 62, 801 (1990)
66. H.-Th. Janka, K. Langanke, A. Marek, G. Martínez-Pinedo, B. Müller, Phys. Rept. 442, 38
(2007)
67. G.S. Bisnovatyi-Kogan, Astron. Zh. 47, 813 (1970) [Sov. Astron. 14, 652 (1971)]
68. N.V. Ardeljan, G.S. Bisnovatyi-Kogan, S.G. Moiseenko, Mon. Not. Roy. Astron. Soc. 359, 333
(2005)
69. S. Akiyama, J.C. Wheeler, D.L. Meier, I. Lichtenstadt, Astrophys. J. 584, 954 (2003)
70. L.B. Okun, Yad. Fiz. 44, 847 (1986) [Sov. J. Nucl. Phys. 44, 546 (1986)]
71. M.B. Voloshin, M.I. Vysotskii, Yad. Fiz. 44, 845 (1986) [Sov. J. Nucl. Phys. 44, 544 (1986)]
72. L.D. Landau, E.M. Lifshitz, Course of Theoretical Physics, Vol. 5: Statistical Physics, Part 2
(Nauka, Moscow, 1976; Pergamon, Oxford, 1980)
73. R.A. Anikin, A.V. Kuznetsov, N.V. Mikheev, Pis’ma Astron. Zh. 36, 714 (2010) [Astron. Lett.
36, 680 (2010)]
References 173

74. R.A. Anikin, A.V. Kuznetsov, N.V. Mikheev, Yad. Fiz. 73, 2000 (2010) [Phys. At. Nucl. 73,
1948 (2010)]
75. G.S. Bisnovatyi-Kogan, Stellar Physics 2: Stellar Evolution and Stability (Nauka, Moscow
1989 (Springer, New York, 2001)
76. P. Bocquet, S. Bonazzola, E. Gourgoulhon, J. Novak, Astron. Astrophys. 301, 757 (1995)
77. H.C. Spruit, Astron. Astrophys. 341, L1 (1999)
78. C.Y. Cardall, M. Prakash, J.M. Lattimer, Astrophys. J. 554, 322 (2001)
79. J.P. Kneller, G.C. McLaughlin, J. Brockman, Phys. Rev. D 77, 045023 (2008)
Chapter 6
Neutrino-Electron Interactions in External
Active Media

As it was mentioned above, an intense electromagnetic field makes possible the

processes which are forbidden in a vacuum such as the neutrino decay into the W +
boson and a charged lepton, ν → − W + ( = e, μ, τ ) and the production of a lepton
pair by neutrino, ν → ν− + . In this chapter, we present in details the technique of
calculations of the neutrino-electron processes in external active media. We consider
mainly the two processes. The first one, which is possible in an intense external
electromagnetic field and in the case of sufficiently high neutrino energy, is the
decay ν → e− W + . The second process is the electron–positron pair production by
a neutrino ν → ν + e− + e+ . We present the procedure of calculation of the process
probability in the case of a strong magnetic field, when an electron and a positron
are created in the ground Landau level, and in the crossed field limit. We calculate
also the four-vector of the mean values of the neutrino energy and momentum losses
due to the process ν → ν + e− + e+ , which could be essential in astrophysical
applications. The process of the electron–positron pair production by neutrino in a
strong magnetic field, if one more component of the external active medium which
is dense plasma is taken into account, should be suppressed by the Fermi—Dirac
statistical factors. In this chapter, we also consider the electron–positron plasma
influence on the process ν → ν + e− + e+ , and take into consideration the crossed
neutrino-electron processes. We also try to apply the results obtained to the well-
known problem of large kick velocities of pulsars born in supernova explosions.

6.1 The ν → e− W + Process in a Strong Magnetic Field

The probability of the neutrino decay ν → e− W + in an external electromagnetic

field is one of the most interesting results that can be extracted from the neutrino
self-energy operator. This probability can be expressed in terms of the imaginary
part of the amplitude (4.63) with the neutrino self-energy operator (4.132).

A. Kuznetsov and N. Mikheev, Electroweak Processes in External Active Media, 175

Springer Tracts in Modern Physics 252, DOI: 10.1007/978-3-642-36226-2_6,
© Springer-Verlag Berlin Heidelberg 2013
176 6 Neutrino-Electron Interactions in External Active Media

For simplicity, hereafter we neglect the neutrino mass m ν , taking the density
matrix of the left-handed neutrino as ρ( p) = γ L ( pγ). One obtains:

1
w(ν → e− W + ) = Im M(νe → νe )
E
1 p2
= − Im Tr [Σ( p) γ L ( pγ)] = −2 ⊥ Im B̄ L . (6.1)
E E

An analysis of the neutrino decay ν → e− W + in an external field is of interest

only at ultrahigh neutrino energies.
In the papers [1–4], the neutrino decay width in an external electromagnetic
field was calculated in the crossed field approximation, in which case the width
is expressed in terms of the dynamical field parameter χ and the lepton mass para-
meter λ:
e( p F F p)1/2 m2
χ= 3
, λ = 2e . (6.2)
mW mW

In the frame where the field is pure magnetic one, the dynamical field parameter
takes the form:
eB p⊥
χ= . (6.3)
m 3W

The decay width is expressed via the parameters (6.2) as follows, see Eqs. (4.203)
and (5.53):
√
− + 2 G F m 4W χ2/3
w(ν → e W ) =
12π E
1  
dv v [2(1 + v)(2 + v) + λ (1 − v)(2 − v)] dAi(u)
× − , (6.4)
[v(1 − v)]4/3 du
0

where Ai(u) is the Airy function (5.28) with the argument:

v + λ (1 − v)
u= . (6.5)
[χ v(1 − v)]2/3

The derivative of the Airy function is expressed via the modified Bessel function
K ν (x)  
dAi(u) u 2 3/2
− =√ K 2/3 u . (6.6)
du 3π 3

Taking in Eq. (6.4) the limit χ, λ  1, one obtains the result which can be written
in terms of the only modified dynamical field parameter
6.1 The ν → e− W + Process in a Strong Magnetic Field 177

χ eB p⊥
ξ=√ = . (6.7)
λ m e m 2W
√
√ range for the ξ parameter appears to be rather large, 0 < ξ  1/ λ, while
The
1/ λ  1. Taking account of the exponential decrease of the modified Bessel
function K ν (x) at large argument value, one can see that the region of small v gives
the main contribution into the integral Eq. (6.4) at small χ. Changing the variable
v = λ x, one can transform the decay width to the form
√
− + 2 G F (eB p⊥ )2
w(ν → e W ) = F(ξ) , (6.8)
3π m 2W E

where
∞  
1 1+x 2 (1 + x)3/2
F(ξ) = √ dx K 2/3 . (6.9)
3 π ξ2 x 3 ξx
0

We remind that these formulas are valid in the approximation ξ  m W /m e . The

range being very wide for the electron, ξ  1.6 × 105 , is not too wide for the τ
lepton, ξ  45.
The integration in Eq. (6.9) can be performed exactly to give
√   √ 
3 3
F(ξ) = 1 + exp − . (6.10)
ξ ξ

The formulas (6.8)–(6.10) should be compared with the results of Refs. [1, 2, 4]. It
should be mentioned that the decay width w defined in Refs. [1, 3] is the same, in the
natural system of units, than the absorption coefficient α [2] and the damping rate of
the neutrino γ [4]. One can see that the absorption coefficient α presented in Eq. (25)
of Ref. [2] looks very similar to our Eqs. (6.8) and (6.10). However, the angular
dependence in our formulas is quite different: instead of the factor p⊥ 2 /E = E sin2 θ

standing in our Eq. (6.8), there is the factor p⊥ = E sin θ in Eq. (25) of Ref. [2].
On the other hand, one can see that our result (6.8)–(6.10) surely contradicts the
Eq. (58) of Ref. [4], where an attempt was made of reinvestigation of the process
ν → e− W + in the crossed field approximation. The difference is the most essential
at small values of ξ, where the result of Ref. [4] appears to be strongly underestimated.
In the earlier paper by Borisov et al. [1] the calculations of the process ν → e− W +
width were performed in the two limiting cases of the small and large values of the
parameter χ. In the limit χ2  λ their result can be presented in the form
√  
2 GF √ m e m 2W
w= √ m e eB sin θ exp − 3 , (6.11)
3π eB p⊥

and can be reproduced from the general formulas (6.8)–(6.10).

178 6 Neutrino-Electron Interactions in External Active Media
√
On the other hand, in the limit χ  1 (ξ  1/ λ) the result of [1] can be written
as √
3 GF
w= √ m W eB sin θ , (6.12)
2π

and can be reproduced from our more general formula (6.4), or by an easier way
from Eqs. (6.1) and (4.220).
A problem of the decay ν → e− W + has a physical meaning only in the fields of
the pulsar type, where the field strength is of order of the critical value ∼1013 G. The
above formulas for the probability except for Eq. (6.12) are applicable for relatively
weak fields only, B  1013 G. It is interesting to consider the process ν → e− W +
in strong magnetic fields of magnetars, of the order of ∼1014 − 1015 G, where the
crossed-field approximation is inapplicable.
Thus, we will use the following hierarchy of the physical parameters: p⊥ 2 
− +
m W  eB  m e . A general expression for the process ν → e W probability can
2 2

be obtained by the substitution of Eq. (4.137) into Eq. (6.1) with taking account of
Eqs. (3.10)–(3.14). After calculations which are not difficult but rather cumbersome,
the process width can be presented in the form

G F (eB)3/2 p⊥
w(ν → e− W + ) = √ Φ(η) , (6.13)
π 2π E

where Φ(η) is the function depending on the one parameter η only:

2
4 eBp⊥
η= , (6.14)
m 4W
∞
1 dy (tanh y)1/2 (sinh y)2 − y tanh y
Φ(η) =
η y 1/2 (sinh y)2 (y − tanh y)3/2
0
 
y tanh y
× exp − . (6.15)
η(y − tanh y)

We stress that we have obtained this formula neglecting the electron mass as the
smallest parameter in the hierarchy used.
The formulas (6.13), (6.15) are valid in a wide region of the parameter η values,
0 < η  m 2W /(eB). The function Φ(η) is essentially simplified at large and small
values of the argument.
In the limit η  1, one obtains:

1
Φ(η  1) π(η − 0.3) , (6.16)
3
and the error is less than 1 % for η > 10.
6.1 The ν → e− W + Process in a Strong Magnetic Field 179

The formulas (6.13), (6.16) reproduce the probability (6.8), (6.10), where the limit
ξ  1 should be taken, and F(ξ  1) 1.
In the other limit η  1 one obtains
  
1 1 3 2
Φ(η  1) exp − 1− η+ η (6.17)
η 2 4

and the error is less than 1 % for η < 0.5.

The formulas obtained allow to establish an upper limit on the energy spectrum
of neutrinos propagating in a strong magnetic field. Let us take the typical size R of
the region with the strong magnetic field as R ∼ 10 km. If the neutrino mean free
path λ = 1/w is much less than the field size, λ  R, all the neutrinos are decaying
inside such the field. For λ = 1 km  R, we can find the cutoff energies E c for the
neutrino spectrum, depending on the magnetic field strength, as follows:
(i) for relatively weak field, B 0.1Be 4 × 1012 G, the neutrino mean free path
can be obtained from Eq. (6.11):
 
4.9 m 219
λ exp , (6.18)
B0.1 sin θ B0.1 E 15 sin θ

where B0.1 = B/(0.1Be ), E 15 = E/(1015 eV), and the cutoff energy corre-
sponding to λ = 1 km, at B0.1 = 1, θ = π/2, is

Ec 0.4 × 1017 eV; (6.19)

(ii) for relatively strong field, B 10Be 4 × 1014 G, the neutrino mean free path
can be obtained from Eqs. (6.13), (6.17):
 
3.2 cm 4.0
λ 3/2
exp 2 sin2 θ
, (6.20)
B10 sin θ B10 E 15

where B10 = B/(10Be ), and the cutoff energy corresponding to λ = 1 km, at

B10 = 1, θ = π/2, is
E c 0.6 × 1015 eV. (6.21)

The results obtained show an essential influence of the intense magnetic field on
the process ν → e− W + width. Despite the exponential character of suppression
of the width in a strong field, Eqs. (6.13), (6.17), as well as in a weak field,
Eq. (6.11), the decay width in a strong field is greater in orders of magnitude
than the one in a weak field, for the same neutrino energy.
180 6 Neutrino-Electron Interactions in External Active Media

6.2 The ν → νe− e+ Process in a Strong Magnetic Field

The process of the electron–positron pair production by a neutrino

ν(P) → ν(P ) + e− ( p) + e+ ( p )

for the relatively small momentum transfers |q 2 |  m 2W , where m W is the W

boson mass, is described by the effective local Lagrangian of the neutrino-electron
interaction (4.66), when the propagators of intermediate W and Z bosons are shrunk
into a point, as is shown in Fig. 4.8.

6.2.1 Calculation of the Differential Probability Based

on the Solutions of the Dirac Equation

The total amplitude for neutrino-electron processes is obtained directly from the
Lagrangian (4.66) where known solutions of the Dirac equation in a magnetic
field (2.1) must be used. As was already mentioned in Chap. 2, in a strong mag-
netic field, eB  E 2 , the electron and the positron can be produced only in the
states that correspond to the ground Landau level (2.37).
Using the Lagrangian (4.66) and the wave functions (2.37), we write the S matrix
element of the process ν → νe− e+ in the following form

G F (2π)3 δ(ε + ε − q0 ) δ( p y + p y − q y ) δ( pz + pz − qz )
S=i √
2 2E V 2E V 2ε(ε + m e )L y L z 2ε (ε − m e )L y L z
× e−q⊥ /4eB−iqx ( p y − p y )/2eB [ū( p ) jˆ(C V − C A γ5 ) u(− p )],
2
(6.22)

where q = P − P = p + p is the change of the four-vector of the neutrino

momentum equal to the four-momentum of the e− e+ pair, ε and ε are the electron
and positron energies, q⊥ is the projection of the vector q on the plane perpendicular
to the vector B = (0, 0, B), q⊥ 2 = q 2 + q 2 , and j = ν̄(P )γ (1 − γ )ν(P)
x y α α 5
is the Fourier transform of the current of the left-handed neutrinos. Note that in
this approximation where the field strength is the largest physical parameter of the
problem, the exponential factor e−q⊥ /4eB in the amplitude (6.22) slightly differs from
2

unity and may be omitted. Direct calculations taking into account the conservation
laws in (6.22) give

me q 2 − 4m 2e q
z
[ū( p ) jˆ(C V − C A γ5 ) u(− p )] = [C V ( j ϕ̃q) + C A ( j ϕ̃ϕ̃q)].
q2 |qz |
(6.23)
6.2 The ν → νe− e+ Process in a Strong Magnetic Field 181

The further calculations will be performed for the case when the electron mass
is the smallest parameter of the problem, i.e. for the following hierarchy: eB 
E 2  m 2e . In this case the expression (6.23) and thus the total amplitude (6.22)
contain the suppression associated with the relative smallness of the electron mass.
This suppression is not random and reflects the angular momentum conservation
law. For example, in the crossed process ν ν̄ → e− e+ being described by the same
amplitude (6.22), the total spin of a neutrino–antineutrino pair in the center-of-inertia
system is one, whereas the total spin of an electron–positron pair in the ground Landau
level is zero. Consequently, the amplitude of the process would be zero for massless
particles and contain the suppression in the relativistic limit under study. However, an
analysis shows that when integration is performed over the phase volume, the main
contribution arises from the kinematic region where q 2 ∼ m e , and this suppression
disappears.
For the probability of the process per unit time we obtain

1 d3 P V
W = |S|2 dn e− dn e+ , (6.24)
T (2π)3

where T is the total interaction time, and the elements of the phase volume are
introduced for the electron and the positron occupying the ground Landau level:

d2 p L y L z d2 p L y L z
dn e− = , dn e+ = . (6.25)
(2π)2 (2π)2

Substituting (6.22) into (6.24) and integrating using δ functions with respect to d2 p
[where, as is usually the case δ 3 (0) = T L y L z /(2π)3 ], we obtain for the total prob-
ability per unit time
 3
G 2F 1 d P d p y d pz
W = δ(ε + ε − q0 )
4
32(2π) E L x E ε(ε + m e )ε (ε − m e )
× |ū( p ) jˆ (C V − C A γ5 ) u(− p )|2 , (6.26)

where we need to substitute ε = m 2e + (qz − pz )2 and pz = qz − pz . The inte-

grand in (6.26) does not depend on p y , and the integration should be performed
in accordance with (5.13). Upon integrating in (6.26) with respect to the electron
momentum we obtain the probability of the ν → νe− e+ process in the form of the
following integral over the final neutrino momentum:

G 2F eBm 2e d3 P |C V ( j ϕ̃q) + C A ( j ϕ̃ϕ̃q)|2
W = Θ(q0 − qz2 + 4m 2e ) .
32π 4 E E (q 2 )3/2 (q 2 − 4m 2e )1/2
(6.27)
182 6 Neutrino-Electron Interactions in External Active Media

6.2.2 Calculation Based on the Imaginary Part of the Loop

Amplitude

As for the photon decay, we present here another method of calculation of the
probability (6.27) based on the unitarity relation. The crossed process for the pair
production by a neutrino ν → νe− e+ is the reaction of the conversion of the
neutrino–antineutrino pair into the electron–positron pair ν ν̄ → e− e+ . It is well
known that the cross section for this reaction is related to the imaginary part of the
transition ν ν̄ → ν ν̄ via the electron loop (see Fig. 6.1) by the unitarity condition

1
σ(ν ν̄ → e− e+ ) = Im M(ν ν̄ → ν ν̄), (6.28)
q2

where q α is the four-momentum of the neutrino–antineutrino pair. It can easily be

seen that the relation (6.28) makes it possible to find the probability of the process
ν → νe− e+ , if we integrate this relation over the phase volume of the final neutrino.
We have  3
1 d P
w(ν → νe− e+ )E = 3
Im M(ν ν̄ → ν ν̄). (6.29)
16π E

Remember that P α = (E, P ) and P α = (E , P ) are the four-momenta of the initial

and final neutrinos, and q = P − P .
The magnetic-field-induced part of the process amplitude, Fig. 6.1, can be easily
constructed from the generalized amplitude (4.24) of the vector—vector type (4.31),
the axial-vector—axial-vector type (4.32), and the vector—axial-vector type (4.33),
with the corresponding substitutions of the generalized currents:

GF GF
jV α → √ C V jα , j Aα → √ C A jα , (6.30)
2 2

where jα is the neutrino current. It should be noted also that ( jq) = 0 and β = eB.
We obtain

Fig. 6.1 The Feynman diagram for the process ν ν̄ → ν ν̄. The double line corresponds to the exact
propagator of an electron in a magnetic field
6.2 The ν → νe− e+ Process in a Strong Magnetic Field 183

G 2F (1) (1) |qϕ j|2

ΔM j→ j = C V2 YV V + C 2A Y A A 2
8π 2 q⊥
|q ϕ̃ j|2 q 2 |qϕϕ j|2
+ C V2 YV(2)V + C 2A Y A(2)A
2
+ C V2 YV(3)V + C 2A Y A(3)A 2 q2
q q⊥
  
(1) q 2 (2) Re[(q ϕ̃ j)(qϕϕ j ∗ )]
+ 2C V C A eB YV A + 2 YV A . (6.31)
q⊥ q2

Turning to the strong field limit, as was done in Sect. 5.1.2, one can show that
only the following functions have the imaginary parts, of all the functions Y included
in (6.31)
 
(2) q2 (3) (1) q 2 (2)
ImYV V = 2 ImY A A = eB Im YV A + 2 YV A
q⊥ q⊥
4πeBm 2e
= Θ(q 2 − 4m 2e ). (6.32)
q 2 (q 2 − 4m 2e )

Substituting (6.31) into (6.29) and taking account of (6.32), we immediately obtain
the expression (6.27) for the probability of the process ν → νe− e+ .

6.2.3 The Total Process Probability

It is convenient to perform the further integration over the final neutrino momentum,
without loss of generality, not in the arbitrary frame (referred to as K ), but in the
special frame K 0 , where the initial neutrino momentum is perpendicular to the mag-
netic field direction, Pz = 0. In the case of a pure magnetic field we can then return
from the frame K 0 to K by the Lorentz transformation along the field direction (we
recall that the field is invariant with respect to this transformation). Really, the value
E W defined by (6.27) is seen to contain the invariants only, including the sign of the
Θ function argument.
It is worthwhile to introduce in (6.27) the dimensionless cylindrical coordinates
in the space of the final neutrino momentum vector P ,

ρ= Px2 + Py2 /E ⊥ , tan φ = Py /Px , ζ = Pz /E ⊥ .

Here, E ⊥ is the initial neutrino energy in the K 0 frame, connected with its energy E
in the arbitrary frame K by the relation E ⊥ = E sin θ, where θ is the angle between
the initial neutrino momentum and the field direction in the K frame.
Representing the expression (6.27) in the form of the integral over the ρ, φ and ζ
variables one obtains:
184 6 Neutrino-Electron Interactions in External Active Media

2π 
1−λ
G 2 m 2 eB E 2 dφ
EW = F e 3 ⊥ dρ ρ e−ε(1−2ρ cos φ+ρ )/2
2

2π 2π
0 0
ζm
dζ
×
γ ρ2 + ζ 2 (1 − 2
ρ2 + ζ 2 + ρ2 )2
−ζm
 
× (C V2 + C 2A ) (1 + ρ2 ) ρ2 + ζ 2 − 2ρ2 − 2C V C A (1 − ρ2 )ζ

− (C V2 − C 2A ) ρ (1 − 2 ρ2 + ζ 2 + ρ2 ) cos φ , (6.33)

where
 
4m 2 λ2
γ= 1 − 2e = 1− ,
q 1 − 2 ρ2 + ζ 2 + ρ2
2m e 1  2
λ= , ζm = 1 + ρ2 − λ2 − 4ρ2 .
E⊥ 2

Note that the integrand in (6.33) has an enhancement that completely compensates
for the suppression by the smallness of the electron mass. The main contribution
then comes from the region near the upper limit of the integral over ρ corresponding
to the relation q 2 ∼ m e .
The term in (6.33) with the C V C A product is caused by the interference of the
vector and axial-vector electron currents. It determines the asymmetry of the electron
emission with respect to the magnetic field, and obviously this term does not con-
tribute to the probability. However, it could be important in calculating the asymmetry
of the averaged neutrino momentum loss, see Sect. 6.4.1.
Neutrino energies in the region E  m e are typical for the above-mentioned
astrophysical processes. It should be noted that expressions (6.27) and (6.33), which
were obtained for the ground Landau level, have the physical meaning of the total
probability of the process only for eB > E 2 /2, in which case the contribution of
other Landau levels is completely suppressed. For the sake of completeness, we nev-
ertheless present here the asymptotic expressions for both strong (eB  m 2e ) and
relatively weak (eB  E 2 ) fields in order to estimate below the relative contribu-
tion of the ground Landau level to the probability of the process. The cumbersome
expression (6.33) is then replaced by simple formulas whose applicability ranges
partially overlap.
(i) For eB  m 2e , we have

G 2F (C V2 + C 2A )
W = eB E 3 sin4 θ f 1 (ε) , (6.34)
16π 3
6.2 The ν → νe− e+ Process in a Strong Magnetic Field 185

where
1
dρ ρ(1 − ρ2 ) e−ε(1+ρ
2 )/2
f 1 (ε) = 4 I0 (ερ)
0
2 5 2 7 3
=1 − ε+ ε − ε + ··· , (6.35)
3 16 60
2 /eB, and I (x) is the modified Bessel function of the zeroth order. For
ε = E⊥ 0
eB  E ⊥
2 , the formula (6.34) takes the simple form

G 2F (C V2 + C 2A )
W = eB E 3 sin4 θ. (6.36)
16π 3
In this region, the result determines precisely the total probability of the process.
It can be seen that the probability grows with neutrino energy in proportion to
E 3 , but it will be shown below that, at higher neutrino energies, higher Landau
levels come into play. As a result, this type of behavior changes to a linear growth,
which persists up to energies corresponding to the boundary of the applicability
range of the effective local Lagrangian (4.66).
In the case of relatively weak fields (m 2e  eB  E ⊥ 2 ), it follows from (6.34)

that the contribution of the ground Landau level is given by

G 2F (C V2 + C 2A )
W = (eB)5/2 sin θ. (6.37)
23/2 π 7/2

(ii) For eB  E ⊥
2 , the general expression (6.33) yields

 ∞ √ 2 
21/2 G 2F m 2e (eB)3/2 −2u 2 /η u −1
W = (C 2
V + C 2
A ) du u e E
π 7/2 u
1
∞ √ 2 
du −2u 2 /η u −1
− C V2 e K sin θ , (6.38)
u u
1

where η = eB/m 2e = B/Be is the field intensity parameter and K(k) and E(k)
are the complete elliptic integrals of the first and second type, respectively [5].
It should be noted that the applicability ranges of formulas (6.34) and (6.38)
partially overlap, and the region of overlap is m 2e  eB  E ⊥ 2 . Indeed, if we

go over to the extreme case of η  1 in (6.38), formula (6.37) is recovered, as

might have been expected. In weak fields, eB  m 2e , the result is exponentially
small, as is usually the case; specifically, we have

G 2F C 2A
W = (eB)5/2 e−2/η sin θ . (6.39)
(2π)5/2
186 6 Neutrino-Electron Interactions in External Active Media

6.3 The ν → νe− e+ Process in a Crossed Field

6.3.1 A Historical Overview

Theoretical study of the process of the electron–positron pair production by a neu-

trino in the crossed field limit has a rather long history [6–14]. The correct type of
dependence of the probability on the dynamical parameter χ:

e2 (P F F P)
χ2 =
m 6e

in the leading log approximation, namely, ∼χ2 ln χ, was found in the paper [6], where
the numerical coefficient was incorrect, however. In succeeding papers, attempts were
made to adjust this coefficient and to find the next postlogarithm terms, which could
appear quite essential when ln χ is not very large.
According to the definition of the problem in the crossed field approximation, one
should consider the ultrarelativistic neutrino only, which exists as the left-handed
one due to the chiral type of its interaction in the frame of the Standard Model,
even if the neutrino mass is nonzero. This remains true if we admit the existence
of exotic properties of the neutrino, which could lead in certain physical conditions
to the depolarizing effects, which were not observed yet. Lack of understanding
that unpolarized ultrarelativistic neutrino fluxes do not exist in Nature, often caused
erroneous extra factors of 1/2 in formulas for the process probabilities with a neutrino
in the initial state because of the non-physical averaging on its polarizations (see,
e.g., [12, 15]).
There are significant differences in the results for the probability of the process
ν → νe− e+ in the crossed field, obtained in the listed papers. In Ref. [12], dedicated
to the study of the decay of a massive neutrino νi → ν j e− e+ (m i > m j + 2m e ) in
an external field, the different formulas for the probability of the process were also
compared, and a conclusion on the mutual agreement of the results was made. In our
opinion, such an agreement is absent.
Indeed, the probability of the process in the limit χ  1 can be presented as
follows:
 
− + 1
W (ν → νe e ) = K W0 χ ln χ − ln 3 − γ E + Δ ,
2
(6.40)
2

where
G 2F (C V2 + C 2A ) m 6e
W0 = , (6.41)
27π 3 E
γ E = 0.577 . . . is the Euler constant, E is the energy of the initial neutrino. The
constants K and Δ entering the expression (6.40), which were obtained by different
authors are given in Table 6.1. It should be noted that in Refs. [6, 12], calculations
6.3 The ν → νe− e+ Process in a Crossed Field 187

Table 6.1 The constants K and Δ of the expression (6.40), obtained in different studies
Authors K Δ
29
Choban and Ivanov (1969) [6] –
1024π
389 9 C V2 − C 2A
Borisov e a. (1983) [7] 1 −2 ln 2 − +
384 128 C V2 + C 2A
9 E
Knizhnikov e a. (1984) [8] –
16 m e
1 5
Borisov e a. (1993) [9] +
2 4
29
Our result (1997) [10] 1 −
24
1 29
Borisov and Zamorin (1999) [12] −
2 24

were performed with taking account of the neutrino-electron interaction through the
W boson only. For comparison of the result (6.40) with these studies, one should put
in Eq. (6.40), respectively, C V = C A = 1 [6] and C V = C A = |Uei Ue3 | [12]. The
loss of the factor m e /E in formulas of [8] for the probability is not a numerical but
a physical errors, since it leads to a loss of relativistic invariance of the value E W .
As it was already noted, the formula (6.40) for the probability describes rather
special case of ln χ  1. There exist a number of physical tasks where the situation is
realized when the dynamical parameter takes moderately high values, so that χ  1,
but ln χ ∼ 1. In this case, the crossed field approximation is applicable, but the
above condition ln χ  1 is not satisfied, so that the formula (6.40) is not enough.
For example, one would wish to consider the next terms in the expansion in inverse
powers of the large parameter χ. On the other hand, the formulas for the probability
for arbitrary values of the χ parameter presented in some of the listed papers, have a
very cumbersome form of multiple integrals, and are inconvenient for the analysis.
The final point in the analysis of the process ν → νe− e+ in the crossed field
approximation was put, as we believe, in our papers [13, 14]. Here we present the
calculation in some detail.

6.3.2 Calculation of the Differential Probability Based

on the Imaginary Part of the Loop Amplitude

Because of differences in the results for the probability of the process ν → νe− e+
in the crossed field, see Table 6.1, a reliable analysis was necessary. For this sake,
we performed the calculation of the differential probability of the process using the
two different methods. The first one was based on the exact solutions of the Dirac
equation in the crossed field (2.40). In the second method, the imaginary part of the
loop amplitude of the transition ν ν̄ → e− e+ → ν ν̄, see Fig. 6.1, was used. The
188 6 Neutrino-Electron Interactions in External Active Media

calculation based on the solutions of the Dirac equation is similar in many details to
the one performed in Sect. 5.2.1 for the photon decay γ → e− e+ in a crossed field.
We do not present here this analysis and refer the reader to Sect. 5.6.1 of our previous
book [16]. In this section, we focus on exploiting the loop amplitude which allows
to find the probability via the unitarity relation.
Note that the results for the process ν → νe− e+ are trivially generalized to other
neutrino-lepton processes. For example, the probability of the process νe → νe e− e+
with replacing m e → m μ and the corresponding change of the constants C V , C A
gives the probability of the process νμ → νμ μ− μ+ , etc.
As in Sect. 6.2.2, we use the relation (6.29) where the field-induced amplitude is
constructed from the amplitudes of the vector—vector type (4.35), the axial-vector—
axial-vector type (4.38), and the vector—axial-vector type (4.39), with the corre-
sponding substitutions of the generalized currents (6.30). Similarly to Eq. (6.31) we
obtain:

G 2F |q F j|2
ΔM j→ j = C V2 YV(1)V + C 2A Y A(1)A
8π 2 (q F Fq)
|q F̃ j|2 q 2( j F F j ∗)
+ C V2 YV(2)V + C 2A Y A(2)A + C V2 YV(3)V + C 2A Y A(3)A
(q F Fq) (q F Fq)
∗ 
(1) (2) Re[(q F̃ j)(q F F j )]
+ 2eC V C A YV A + YV A , (6.42)
(q F Fq)

where the functions Y for the crossed field from Eqs. (4.35), (4.38), and (4.39) should
be substituted. We remind that the imaginary part of the Hardy—Stokes function is
expressed via the Airy function, Im f (z) = π Ai(z).
The resulting probability of the process takes the form of the following integral
over the final neutrino momentum

 1 
G 2 (C 2 + C 2 )m 2 d3 P 2 3 + u2 χq 2/3
W = F V7 4 A e du i 0 Ai (U )
2 π E E 3 (1 − u 2 )1/3 4
0
  4/3
q 2 (1 − u 2 ) q2 3 + u2 4
− Bi(U ) + i 1 Ai (U )
2m e 2 24m e (1 − u )
2 2 1/3 χq
 4/3
1 4
− i 2 (1 − u ) 2 2/3
Ai (U )
8 χq
  4/3 
C 2A 1 4
+ 2 i 0 2Bi(U ) − i 1 Ai (U )
C V + C 2A (1 − u 2 )1/3 χq
 2 
CV C A 4
− i 3 U Ai(U ) . (6.43)
4(C V2 + C 2A ) χq
6.3 The ν → νe− e+ Process in a Crossed Field 189

where the invariants are introduced that are constructed of the neutrino current and
the field tensor:

e2 ( j F F j ∗ ) e2 (q F̃ j)(q F̃ j ∗ )
i 0 = ( j j ∗ ), i 1 = , i 2 = ,
m 4e m 6e
e3 Re[(q F̃ j)(q F F j ∗ )]
i3 = , (6.44)
m 8e

Ai(U ) is the Airy function (5.28) while Bi(U ) is the integral:

∞
Bi(U ) = dy Ai(y), (6.45)
U

and the argument of the Airy function is

 2/3  
4 q 2 (1 − u 2 )
U= 1− . (6.46)
χq (1 − u 2 ) 4m 2e

q = P − P is the four-momentum lost by a neutrino. Hereafter, we denote the

dynamical parameter constructed of the initial neutrino momentum P as χ, and the
dynamical parameter constructed of the q momentum as χq :
 1/2  1/2
e2 (P F F P) e2 (q F Fq)
χ= , χq = . (6.47)
m 6e m 6e

6.3.3 The Total Process Probability

To integrate the expression (6.43) with respect to the final neutrino momentum, let us
introduce new variables κ, ξ, and φ, which are the relativistic invariants, as follows

q2 (q F Fq)
κ=− 2 , ξ= ,
[4e (P F F P)]1/3 (P F F P)
(P F̃ P )
cos φ = . (6.48)
(P F̃ P )2 + (P F P )2

In the frame where the initial neutrino momentum P is perpendicular to the mag-
netic field vector B, the angle φ has a meaning of the azimuthal angle in the plane
perpendicular to the vector P, between the magnetic field and the projection of the
vector P on this plane. With these variables, the invariants (6.44) take the form
190 6 Neutrino-Electron Interactions in External Active Media

i 0 = −8 m 2e (2χ)2/3 κ ,
i 1 = 16 m 2e χ2 (1 − ξ) ,
 
i 2 = m 2e (2χ)8/3 κ ξ 2 + 4(1 − ξ) sin2 φ ,

i 3 = m 2e (2χ)10/3 ξ (2 − ξ) κ(1 − ξ) cos φ . (6.49)

The integral over the final neutrino momentum can be written as

 1 ∞ 2π
d3 P χ 2/3 dφ
Θ(χq ) = 4πm 2e dξ dκ . (6.50)
E 4 2π
0 0 0

As was already mentioned, the interference term in (6.43), which is proportional to

the product C V C A , does not contribute to the probability, but it could be important
in calculating the averaged neutrino momentum loss.
Upon integrating over φ the expression (6.43) takes the form

1 1 ∞
G 2 (C 2 + C 2 )m 6 χ2
W = F V 3A e du dξ dκ − 2κ2 (1 − u 2 )Bi(U )
16π E
0 0 0
2 − 2ξ + ξ2 9 − u2
−κ Ai (U ) (6.51)
3ξ 4/3 (1 − u 2 )1/3
    
2C 2 4 2/3 1−ξ 1
− 2 A 2 κ Bi(U ) + 4 4/3 Ai (U ) ,
CV + C A χ ξ (1 − u 2 )1/3

where  2/3
(1 − u 2 )1/3 4
U =κ + .
ξ 2/3 χ(1 − u 2 )ξ

Performing integration over the variable κ, one obtains

1 1
G 2 (C 2 + C 2A ) m 6e χ2
W = F V du xdx z Ai(z)
27π 3 E
0 0

3 + x2 3 9 C 2A
× + (1 − 3x) + (5 + x) , (6.52)
(1 − u 2 )(1 − x) 8 4 C V2 + C 2A

where  2/3
4
z= .
χ(1 − u 2 )(1 − x)
6.3 The ν → νe− e+ Process in a Crossed Field 191

Finally, performing one more cumbersome integration, we present the result for
the probability in a form of the single integral containing the Airy function:

1
G 2F (C V2 + C 2A ) m 6e χ2
W = u 2 du t Ai(t)
27π 3 E
0
 
4 29 15 47
× 2L(u) − − L(u) −
1 − u2 24 2 48
 
1 47
+ 1 + (1 − u )L(u) 33 − (1 − u )
2 2
(6.53)
8 4
9 2
CA  
+ 48L(u) + 2 − 1 + (1 − u 2 )L(u) 28 − 3(1 − u ) 2
.
16 C V + C 2A
2

Here  2/3
4 1 1+u
t = , L(u) = ln . (6.54)
χ(1 − u 2 ) 2u 1 − u

In the case χ  1, one obtains from Eq. (6.53) the formula for the probability which
demonstrates the well-known exponential suppression, in agreement with Ref. [9]:
√  
3 6 G 2F m 6e 8
W (χ  1) (3 C 2
V + 13 C 2
A ) χ 4
exp − . (6.55)
(16π)3 E 3χ

In the case χ  1 (more exactly, in the case ln χ  1) we obtain from Eq. (6.53)
the formula (6.40) where K = 1 and Δ = −29/24, in agreement with [10, 11]:
 
G 2F (C V2 + C 2A ) m 6e χ2 1 29
W (χ  1) ln χ − ln 3 − γE − . (6.56)
27π 3 E 2 24

As the dynamical parameter χ is proportional to the neutrino energy, the probabil-

ity (6.56) is seen to grow with energy as E ln E instead of the growth W ∼ E 3 in the
strong field limit, cf. (6.36). Comparing also (6.56) with (6.37), one can see that
the contribution of the ground
√ Landau level into the probability is relatively small in
the limit E 2  eB (∼ eB/E  1).
It is not difficult to find from (6.53) the next term of expansion over the inversed
powers of the parameter χ, to obtain

G 2F (C V2 + C 2A ) m 6e χ2 1 29
W (χ  1) ln χ −
ln 3 − γE −
27π 3 E 2 24

1 9 31/3 π 2 19 C V2 − 63 C 2A
− ,
56 Γ  2 4
(6.57)
χ2/3 C V2 + C 2A
3
192 6 Neutrino-Electron Interactions in External Active Media

where Γ (x) is the gamma function, Γ (2/3) = 1.354 . . . .

As is seen from (6.57), the correcting term ∼ χ−2/3 is not universal with respect to
the neutrino flavor. It is relatively small and negative for the process νe → νe e− e+ ,
while for the process νμ → νμ e− e+ the correction term is positive and rather large.
The dependence of the probability of the process νe → νe e− e+ on the dynamical
parameter χ in the region where its value is moderately large, was analysed numeri-
cally in [13], see also [16]. It appears that the correction term ∼ χ−2/3 is more likely
to worsen than to improve the presentation of the probability in this region. A possi-
ble explanation of this could be that the next term of expansion over the parameter
χ inversed has the form ∼ χ−4/3 ln χ to be rather large. However, it appears to be a
difficult problem to extract this term. On the other hand, it is unnecessary because
the exact formula (6.53) can be used in a detailed analysis of the probability of the
e− e+ pair production by a neutrino propagating in an external electromagnetic field,
when the value of the dynamical parameter χ is moderately large.

6.4 Possible Astrophysical Manifestations of the ν → νe− e+

Process in an External Magnetic Field

6.4.1 Mean Losses of the Neutrino Energy and Momentum

The probability of the ν → νe− e+ process defines its partial contribution into the
neutrino opacity of the medium. The estimation, e.g. of the electron neutrino mean
free path with respect to this process, obtained from the probability (6.36) yields:
 3  
− + 1 10 Be 10 MeV 3
λ(ν → νe e ) = ∼ 4400 km . (6.58)
W B E

It is too large compared with the typical size of a compact astrophysical object, e.g. the
supernova remnant, where a strong magnetic field could exist. However, a mean free
path does not exhaust the neutrino physics in a medium. In astrophysical applications,
we could consider the values that probably are more essential, namely, the mean
values of the neutrino energy and momentum loss and especially the asymmetry of
the momentum loss, caused by the influence of an external magnetic field. These
values can be described by the four-vector of losses Q α ,

Qα = E q α dW = −E (I, F) . (6.59)

where q is the difference of the momenta of the initial and final neutrinos, q = P − P ,
dW is the total differential probability of the process. The zeroth component of Q α
is connected with the mean energy lost by a neutrino per unit time due to the process
considered, I = dE/dt. The space components of the four-vector (6.59) are similarly
6.4 Possible Astrophysical Manifestations of the ν → νe− e+ Process 193

connected with the mean neutrino momentum loss per unit time, F = dP/dt. We
present here the results of our calculation of the four-vector Q α in the two limiting
cases considered above.
(i) In the case eB  E ⊥
2 one obtains:

G 2F eB(PϕϕP)2 (C V2 + C 2A ) α
Qα = [P f 2 (ε) − 2(ϕϕP)α f 3 (ε)
48π 3
2C V C A
+ 2 (ϕ̃P)α f 2 (ε)], (6.60)
C V + C 2A
1
dρ ρ(1 − ρ2 )2 e−ε(1+ρ
2 )/2
f 2 (ε) = 6 I0 (ερ)
0
5 21 2 7 3
=1 − ε+ ε − ε + ··· ,
8 80 80
1
dρ ρ(1 − ρ2 ) e−ε(1+ρ
2 )/2
f 3 (ε) = 3 [(1 + ρ2 )I0 (ερ) − 2ρI1 (ερ)]
0
15 21 2 7 3
=1 − ε+ ε − ε + ··· ,
16 40 32

where ε = E ⊥ 2 /eB, I (x) and I (x) are the modified Bessel functions. In the
0 1
strong field limit, eB  E ⊥
2 , one obtains for the neutrino energy and momentum

loss,
 
1 2C V C A
Ė = E W 1 + 2 cos θ , (6.61)
3 C V + C 2A
 
1 2C V C A
Fz = E W cos θ + 2 , F⊥ = E W sin θ, (6.62)
3 C V + C 2A

where the O Z axis is directed along the field, the vector F⊥ orthogonal to the
field direction belongs to the plane of the vectors B and p. The probability W
should be taken from (6.36).
(ii) In the limiting case eB  E 2 sin2 θ corresponding to the crossed field limit we
have obtained the following result for the four-vector Q α of the neutrino energy
and momentum losses due to the process ν → νe− e+ :

α 7G 2F (C V2 + C 2A )m 6e χ2 √ η2
Q = P α (ln χ − 1.888) − 3 (ϕϕP)α
432π 3 χ

CV C A η
+ 7.465 (ϕ̃P)α . (6.63)
C V2 + C 2A χ2/3
194 6 Neutrino-Electron Interactions in External Active Media

We recall that η = eB/m 2e = B/Be is the field intensity parameter.

In the limiting case of very large dynamical parameter ln χ  1, the expression
for the four-vector is simplified significantly:

7
Qα E W P α, (6.64)
16
where the probability W should be taken from (6.56).

6.4.2 Applicability of the Results Obtained in a Pure Magnetic

Field for the Plasma Environment

Note that the formulas obtained are valid also in the presence of dense plasma with
an electron density of about 1033 cm−3 . This is due to the specificity of the ultrarela-
tivistic electron gas statistics in a magnetic field, see Ref. [17]. Given the degeneracy
with respect to the transverse momentum, see Eq. (5.13), the connection of the den-
sity of the ultrarelativistic electron–positron gas with the chemical potential μe and
the temperature T is described by the sum over the Landau levels:

∞    −1
eB p − μe
n e = n e− − n e+ = dp exp +1
2π 2 T
0
∞
   −1 
 p 2 + 2keB − μe
+2 exp +1 − (μe → −μe ) . (6.65)
T
k=1
√
In a strong field, under the condition eB − μe  T , when practically the main
Landau level is only occupied, the temperature dependence in Eq. (6.65) disappears
and the chemical potential depends only on the plasma density and the field intensity:
 
2π 2 n e ne 1016 G
μe = 2.6 MeV . (6.66)
eB 10 cm−3
33 B

Thus, the chemical potential can be significantly less than in the absence of the field,
μe (3π 2 n e )1/3 for the same values of the density. However, it is clear that the
chemical potential increases with the density much faster than in the absence of the
field. For the density values
 3/2
−3 B
n > 3.5 × 10 cm 33
1016 G

the next Landau levels become to be occupied and the connection between the chem-
ical potential and the density is given by
6.4 Possible Astrophysical Manifestations of the ν → νe− e+ Process 195

100

10
µ̃ e ( MeV)

0.1

0.01
10 31 10 32 10 33 10 34 10 35 10 36 10 37
3
n e (cm )

Fig. 6.2 The dependence of the chemical potential on the density of the relativistic electron gas: 1
in the absence of the field; 2,3,4 in a strong magnetic field (see Eq. (6.67)) for the values of the field
strength 1015 , 1016 , 1017 G, respectively; breaks of the curves correspond to successive opening of
the Landau levels

 k
   
eBμe max
2eB μ2e
ne 1+2 1−k 2 , kmax = , (6.67)
2π 2 μe 2eB
k=1

where [x] is the integer part of x. The dependence of the chemical potential on the
density of the relativistic electron gas is shown in Fig. 6.2.
It can be seen that the chemical potential almost coincides with its value in the
absence of the field, when several lower Landau levels are excited. Thus, for the
typical energies of the electrons and positrons produced by neutrinos with ener-
gies ∼20 MeV, when the plasma parameters correspond to the supernova envelope
conditions, n ∼ 1033 cm−3 and T ∼ 1 MeV, the suppressing statistical factors are
unimportant. √
In the other limiting case of very high temperatures T  eB, μe , taking account
of the medium influence leads to the appearance of the constant statistical factors
196 6 Neutrino-Electron Interactions in External Active Media

equal to 1/2 both for electrons and positrons. It reduces the process probability in
4 times.

6.4.3 Possible Astrophysical Manifestations

To illustrate the formulae obtained we consider the astrophysical process of the birth
of a magnetized neutron star (i.e. pulsar) in a supernova explosion. Let us suppose
that a very strong magnetic field of the order of 1016 –1018 G [18–22] arises in the
cataclysm in the vicinity of a neutrinosphere. The electron density in this region will
be considered to be not too high, so a creation of the e− e+ pairs is not suppressed
by statistical factors. In this case the neutrino propagating through the magnetic field
would lose energy and momentum in accordance with the above formulae. A part of
the total energy lost by neutrinos in the strong magnetic field due to the process of
the e− e+ pair creation could be estimated from Eq. (6.61):
  3  
ΔE B Ē Δ
∼ 0.6 × 10−2 . (6.68)
Etot 1017 G 10 MeV 10 km

Here, Δ is the characteristic size of the region where the field strength varies insignif-
icantly, Etot is the total energy carried off by neutrinos in a supernova explosion, and
Ē is the neutrino energy averaged over the neutrino spectrum. Here we take the
energy scales that are believed to be typical for supernova explosions [23, 24]. One
can see from (6.68) that the effect could manifest itself at a level of about one per-
cent. In principle, it could be essential in a detailed theoretical description of the
process of a supernova explosion. Namely, if the magnetic field is strong enough, the
well-known FOE problem could be solved due to the process of the production of
electron–positron pairs by neutrinos, ν → νe− e+ . The meaning of the FOE prob-
lem is the following: for the self-consisted description of a supernova explosion, it
is necessary to find any mechanism of transferring the energy ∼1051 erg (ten to the
Fifty One Erg) from the neutrino outflow to the supernova envelope i.e. near 1 % of
the total energy ∼1053 erg produced in the explosion.
One more interesting effect is an asymmetry of outgoing neutrinos:
 
 
 pi 
 
i
A=  . (6.69)
|pi |
i

In the same limit of the strong field we obtain

  3  
−3 B Ē Δ
A ∼ 3 × 10 . (6.70)
1016 G 20 MeV 20 km
6.4 Possible Astrophysical Manifestations of the ν → νe− e+ Process 197

Let us note that an origin of the asymmetry of the neutrino momentum loss with
respect to the magnetic field direction is a manifestation of the parity violation in weak
interaction, because the Fz value contains the term proportional to the product of the
constants C V and C A . This asymmetry could result in the recoil “kick” velocity of the
rest of the cataclysm. The long-standing problem of the observed high space velocities
of pulsars is discussed in more detail below in Sect. 6.5.8. For the parameters used,
the asymmetry due to the process ν → νe− e+ (6.70) would provide a “kick” velocity
on the order of 150 km/s for a pulsar with a mass on the order of the solar mass.
It is important for astrophysical manifestations that all expressions obtained for
the process ν → νe− e+ are also applicable for the process with antineutrino ν̄ →
ν̄e− e+ due to the C P-invariance of the weak interaction.
In the limiting case eB  E 2 sin2 θ corresponding to the crossed field limit,
for the total energy loss via the production of electron–positron pairs by neutrinos
ν → νe− e+ one obtains from (6.64):
 2   
ΔE −6 B Ē Δ
∼ 10
Etot 1015 G 20 MeV 10 km
  
B Ē
× 4.7 + ln , (6.71)
1015 G 20 MeV

which is much less than (6.68). The asymmetry is suppressed in this case and has no
practical interest.

6.5 Neutrino in Strongly Magnetized Electron–Positron Plasma

The process of the electron–positron pair production by neutrino in a strong magnetic

field, if one more component of the external active medium which is dense plasma is
taken into account, should be suppressed by the Fermi—Dirac statistical factors. In
Sect. 6.4.2, the conditions are defined when such a suppression is inessential. These
conditions could be realized, for example, in the process of the neutron star merging.
For higher plasma densities corresponding to the conditions of a supernova explosion,
the effect of plasma must be considered. At the same time, along with the above-
mentioned suppression of the e− e+ pair birth, new channels of the neutrino-electron
interaction arise.
In this section, the full set of the neutrino-electron processes in a magnetized
plasma is considered according to Refs. [25, 26]. Besides the canonical scattering
and annihilation reactions νe∓ → νe∓ and ν ν̄ → e− e+ , which are possible in the
absence of the field, the processes are also analysed of “synchrotron” emission and
absorption of a neutrino pair e ↔ eν ν̄ and of the electron–positron pair production by
neutrino ν → νe− e+ , which are possible only in a magnetic field. Finally, an “exotic”
process of the plasma electron–positron pair capture by a neutrino, νe− e+ → ν, is
also considered. This process is allowed only in the presence of both a magnetic field
and hot plasma.
198 6 Neutrino-Electron Interactions in External Active Media

6.5.1 What Do We Mean Under Strongly Magnetized e− e+ Plasma

Here we discuss the conditions where, among all the physical parameters character-
izing an electron–positron plasma, the field parameter is the dominant one. These
conditions can be characterized simply by the relationship: eB  μ2e , T 2 , where μe
is the chemical potential and T is the temperature of the electron–positron plasma.
In order to find a better substantiated relationship we compare the energy densities
of the magnetic field B 2 /8π and the electron–positron plasma.
As we know, a magnetic field changes the statistical properties of an electron–
positron gas [17]. Taking into account degeneracy of the transverse momentum, the
dependencies of the concentration and energy density of an electron–positron gas
on the chemical potential and temperature are described by the following sums over
Landau levels:
∞
eB
n = n e− − n e+ = d p [Φ( p, μe , T ) − Φ( p, −μe , T )] , (6.72)
2π 2
0
∞
eB
E = Ee− + Ee+ = p d p [Φ( p, μe , T ) + Φ( p, −μe , T )] , (6.73)
2π 2
0
   −1
p−μ
Φ( p, μ, T ) = exp +1
T
∞
   −1
 p 2 + 2keB − μ
+2 exp +1 . (6.74)
T
k=1

Here we used the approximation of an ultrarelativistic electron–positron gas since

astrophysical processes are characterized by fairly high neutrino and plasma electron
energies E  m e . Thus, we shall neglect the electron mass wherever this causes no
misunderstandings. √
In a strong field and specifically, when the condition eB − μe  T is satisfied,
in practice only the ground Landau level is occupied. From (6.72) and (6.73) we then
obtain
eBμe
n= , (6.75)
2π 2
eBμ2e eBT 2
E= + . (6.76)
4π 2 12
Thus, a more exact condition that the electron–positron plasma is strongly magnetized
may be written in the form

B2 π 2 n 2e eBT 2
 + . (6.77)
8π eB 12
6.5 Neutrino in Strongly Magnetized Electron–Positron Plasma 199

Selecting values of the physical parameters typical for a supernova envelope as scales
in the relationship (6.77), we rewrite this in the form

ρ212 Y0.1
2
erg
0.8 × 1032 B32  1.7 × 1030 + 1.1 × 1027 B3 T52 , (6.78)
B3 cm3

where
B ρ Ye T
B3 = , ρ12 = 12 , Y0.1 = , T5 = , (6.79)
103 Be 10 g cm−3 0.1 5 MeV

ρ is the total plasma mass density in the envelope, and Ye is the ratio of the number
of electrons to the number of baryons. It can be seen that the plasma magnetization
condition is definitely satisfied.

6.5.2 Neutrino-Electron Processes in Strongly Magnetized

Plasma: A Kinematic Analysis

In this section, calculations are similar to the ones performed in Sect. 6.2. When the
processes in strongly magnetized plasma are studied, the additional conditions of
applicability of the Lagrangian (4.66) should be taken: eBT, eBμe  m 3W .
All neutrino-electron processes determined by the Lagrangian (4.66) can be
divided into two groups.
(i) Processes in which a neutrino presents in both the initial and final states: νe∓ →
νe∓ , ν → νe− e+ , νe− e+ → ν, and the similar antineutrino processes.
(ii) Processes involving creation or absorption of a neutrino-antineutrino pair:
e− e+ → ν ν̄, ν ν̄ → e− e+ , e → eν ν̄, eν ν̄ → e.
It can be seen from Eq. (6.23) that the square of the amplitude of each neutrino-
electron process contains the factor m 2e /q 2 . However, the value of q 2 = q02 − qz2
differs fundamentally for processes of the first and second types. For processes with
a neutrino-antineutrino pair we have q = P + P (P and P are the four-momenta of
a neutrino and an antineutrino, respectively), and consequently q 2 > 0. Since q 2 =
q 2 + q⊥2 , where both terms are positive, the value of q 2 can only be small when both
2
q , and q⊥ 2 are small which is only possible in a small region of a phase space. This

implies that almost everywhere in the phase space one has q2 ∼ E ∼ T  me.
This leads to reduction of the probability by a factor m 2e /T 2  1.
At the same time, we have q = P − P for processes involving neutrinos in the
initial and final states and consequently q 2 < 0 and the value of q 2 may be small over
a fairly wide region of phase space. Calculations confirm that kinematic amplification
is achieved for these processes, leading to the disappearance of the factor m 2e /T 2 in
the probabilities.
200 6 Neutrino-Electron Interactions in External Active Media

Fig. 6.3 Kinematic regions in the momentum space of a final neutrino: I for the pair creation
process ν → νe− e+ ; II for the scattering channels νe− → νe− , νe+ → νe+ ; III for the pair
capture process νe− e+ → ν; the lines correspond to the condition q 2 = 0

Hence, neutrino interaction with a strongly magnetized electron–positron plasma

is determined by the processes νe− → νe− , νe+ → νe+ , ν → νe− e+ , νe− e+ → ν.
In Fig. 6.3, kinematic region in the momentum space of a finite neutrino is shown
for the processes listed above in a convenient reference frame where the momentum
of the initial neutrino is perpendicular to the magnetic field. The main contribution
to the probability comes from regions near the parabola q 2 = 0 where this kinematic
amplification takes place.
It is interesting to analyze the kinematics of the processes ν → νe− e+ , νe− e+ →
ν in the Minkowski plane {0, 3}. The energies of the electron and positron (see
Eq. (2.24)) on the ground Landau level are ε = pz2 + m 2e and ε = pz2 + m 2e .
Under the condition ε ∼ ε ∼ T  m e , electrons and positrons can be treated as
almost massless particles. In turn, in the energies of the initial and final neutrinos,
6.5 Neutrino in Strongly Magnetized Electron–Positron Plasma 201

E= Pz2 + P⊥2 and E = Pz 2 + P⊥2 , the transverse momentum components can

play the role of effective masses in the plane {0, 3}, namely, P⊥2 = (m ν )2eff and P⊥2 =
(m ν )2eff . Thus, the process ν → νe− e+ is open at the condition (m ν )eff > (m ν )eff .
In the plane {0, 3}, it looks as the decay of “heavier” neutrino into the “lighter” one
plus the “massless” electron and positron. Accordingly, the process of the capture of
a pair, νe− e+ → ν, is open at (m ν )eff < (m ν )eff .

6.5.3 The Probability of the Process ν → νe− e+

The S matrix element of the process

ν(P) → ν(P ) + e− ( p) + e+ ( p )

has the form (6.22), and the amplitudes of the other neutrino-electron processes are
then obtained by crossing transformations.
We express the probability of the creation of an e− e+ pair by neutrino per unit
time in the following form:

1
− +
W (ν → νe e ) = |S|2 dΓe− (1− f e− ) dΓe+ (1− f e+ ) dΓν (1− f ν ), (6.80)
T

where T is the total interaction time, and dΓ is an element of the particle phase
volume,

d2 p L y L z d2 p L y L z d3 P V
dΓe− = , dΓe+ = , dΓν = . (6.81)
(2π) 2 (2π) 2 (2π)3

The electron and positron distribution functions

1 1
f e− = , f e+ = (6.82)
e(ε−μe )/T + 1 e(ε +μe )/T + 1

allow for the presence of a plasma; here μe and T are the chemical potential and
temperature of the electron–positron gas. To be general, we also allowed for the
possible presence of a quasiequilibrium neutrino gas described by the distribution
function f ν . In general, the question of the accuracy of the description of the state of
a neutrino gas under conditions of stellar collapse or another astrophysical process
using an equilibrium distribution function and also the determination of this function
is a complex astrophysical problem (see, e.g., [27]). Quite clearly, the approximation
of an equilibrium neutrino Fermi gas using the distribution function

1
f ν(eq) = , (6.83)
e(E−μ̃ν )/Tν + 1
202 6 Neutrino-Electron Interactions in External Active Media

where μ̃ν and Tν are the chemical potential and the temperature of the neutrino gas,
should give satisfactory results inside the neutrinosphere. Outside the neutrinosphere,
where an outgoing neutrino flux is formed and the neutrino momenta become asym-
metric, a factorization of the local distribution is usually assumed

Φ(ϑ, R)
fν = , (6.84)
e(E−μ̃ν )/Tν + 1

where the energy distribution is assumed to be approximately equilibrium, the func-

tion Φ(ϑ, R) determines the neutrino angular distribution, ϑ = cos α, α is the angle
between the neutrino momentum and the radial direction in the star, and R is the
distance from the center of the star. An analysis shows [27], that in the vicinity
of the neutrinosphere the function Φ(ϑ, R) differs negligibly from unity. In order
to calculate the probability we shall use the neutrino distribution function in the
form (6.83), neglecting the asymmetry. Later in Sect. 6.5.6 when analyzing possible
astrophysical manifestations of these neutrino-electron processes, we shall also allow
for asymmetry in the distribution function (6.84) for the initial and final neutrinos.
Substituting the S matrix element (6.22) into (6.80) and integrating using δ-
functions over d2 p [where, as is usually the case δ 3 (0) = T L y L z /(2π)3 ], we
obtain
 3
G 2F 1 d P d p y d pz
W = (1 − f ν ) δ(ε + ε − q0 )
32(2π)4 E L x E ε(ε + m e )ε (ε − m e )
× (1 − f e− ) (1 − f e+ ) |ū( p ) jˆ(C V − C A γ5 ) u(− p )|2 , (6.85)

where we need to substitute ε = m 2e + (qz − pz )2 , pz = qz − pz . It is easy

to see that the expression in the integrand in Eq. (6.85) does not depend on k y and
consequently integration over k y essentially determines the degree of degeneracy of
an electron having a given energy.
Integrating over the electron momentum in Eq. (6.85) with taking account of
Eq. (5.13) we obtain the probability of the ν → νe− e+ process in the form of the
following integral over the final neutrino momentum:

G 2F eBm 2e d3 P 1
W = Θ(q0 − qz2 + 4m 2e )
64π 4 E E (q 2 )3/2 (q 2 − 4m 2e )1/2
× |C V ( j ϕ̃q) + C A ( j ϕ̃ϕ̃q)|2 (1 − f ν ) [(1 − f e− )(1 − f e+ ) + (qz → −qz )].
(6.86)

In this expression the electron and positron energies ε and ε appearing in the distri-
bution functions f e∓ are determined by the conservation law ε + ε − q0 = 0 and
are given by
6.5 Neutrino in Strongly Magnetized Electron–Positron Plasma 203
     
1 4m 2e 1 4m 2e
ε= q0 + q z 1 − 2 , ε = q0 − q z 1 − 2 . (6.87)
2 q 2 q

Expression (6.86) is a generalization of the formula (6.27), where we investigated

the neutrino-electron process ν → νe− e+ in a high-intensity purely magnetic field,
to the case where electron–positron and neutrino gases are present.
Further integration over the final neutrino momentum can be conveniently per-
formed as in Sect. 6.2 in a reference frame where the initial neutrino momentum is
perpendicular to the magnetic field, Pz = 0. For the case of a purely magnetic field
we could convert to this frame without any loss of generality by performing a Lorentz
transformation parallel to the field. In fact, we can see that in addition to statistical
Fermi factors the value of E W determined from Eq. (6.86) only contains invariants
with respect to this transformation (including the sign of the argument of the Θ
function). However, we now have a special reference frame, namely, the plasma rest
frame, in which the distribution functions (6.82) and (6.83) are formulated. In order
to convert to a frame where Pz = 0 we express these functions in the invariant form:

1 1
f e− = , f e+ = ,
e(( pv)−μe )/T +1 e(( p v)+μe )/T +1
1
fν = . (6.88)
e((P v)−μ̃ν )/Tν +1

Here we introduce the four-vector of the plasma velocity v α , (v 2 = 1) which in its

rest frame is v α = (1, 0) and the distribution functions (6.88) are exactly the same
as the functions (6.82) and (6.83). In the frame Pz = 0 we have

v α = (v0 , 0, 0, vz ) , v0 = 1/ sin θ , vz = − cos θ/ sin θ ,

where θ is the angle between the vectors of the initial neutrino momentum and the
magnetic field in the plasma rest frame.
In formula (6.86) it is convenient to use the dimensionless cylindrical coordinates
in the space of the final neutrino momentum vector P :

ρ= Px2 + Py2 /E ⊥ , tan φ = Py /Px , ζ = Pz /E ⊥ .

Here E ⊥ is the energy of the initial neutrino in the frame Pz = 0 which is related
to its energy E in the plasma rest frame by E ⊥ = E sin θ. In terms of the variables
ρ, ζ, Eq. (6.86) is rewritten in the form


1−λ ζm
G 2 m 2 eB E 2 dζ
EW = F e 3 ⊥ dρ ρ
4π β ρ2 + ζ 2 (1 − 2 ρ2 + ζ 2 + ρ2 )2
0 −ζm
204 6 Neutrino-Electron Interactions in External Active Media
  
× (C V2 + C 2A ) (1 + ρ2 ) ρ2 + ζ 2 − 2ρ2 − 2C V C A (1 − ρ2 )ζ
 
1 1 1 
× 
1 + e−(P v)/Tν −ην 1 + e−( pv)/T +η 1 + e−( p v)/T −η σ=+1
 
1 1 
+  , (6.89)
1 + e−( pv)/T +η 1 + e−( p v)/T −η σ=−1

where we need to substitute in the distribution functions (6.88)

E⊥  
( pv) = 1 − ρ2 + ζ 2 (1 + σβ cos θ) − ζ(cos θ + σβ) ,
2 sin θ
E⊥  
( p v) = 1 − ρ2 + ζ 2 (1 − σβ cos θ) − ζ(cos θ − σβ) ,
2 sin θ
E⊥
(P v) = ρ2 + ζ 2 + ζ cos θ ,
sin θ

and also introduce the notations η = μe /T , ην = μ̃ν /Tν ,

 
4m 2 λ2
β= 1 − 2e = 1− ,
q 1 − 2 ρ2 + ζ 2 + ρ2
2m e 1  2
λ= , ζm = 1 + ρ2 − λ2 − 4ρ2 .
E⊥ 2

Note that the expression in the integrand in (6.89) exhibits an enhancement which
completely compensates for the suppression by the smallness of the electron mass.
The main contribution then comes from the region near the upper limits of the
integrals over ρ, ζ corresponding to the values q 2 ∼ m e . Converting to the new
integration variables β and x = E ⊥ (1 − ρ2 )/4T sin θ in Eq. (6.89) and extracting
2 /m 2 , we transform the expression for the probability
the leading contribution ∼ E ⊥ e
to the form
τ /4
2 T 2 sin2 θ  1
G 2F eB E ⊥ (C V + C A )2
EW = xdx dβ
2π 3 1 + e−+2x(1+u)/τ +ην
0 0
 
× f(β, u, η) f(−β, u, −η) + f(β, u, −η) f(−β, u, η)

(C V − C A )2
+ f(β, −u, η) f(−β, −u, −η)
1 + e−+2x(1−u)/τ +ην

+ f(β, −u, −η) f(−β, −u, η) , (6.90)

where  = E ⊥ /(Tν sin θ), u = cos θ, τ = Tν /T ,

6.5 Neutrino in Strongly Magnetized Electron–Positron Plasma 205

1
f(β, u, η) = .
1 + e−x(1+β)(1+u)+η

Integrating (6.90) over the variable β with using the relation

1
dβ f(β, u, η) f(−β, u, −η)
0
 
1 1 + e−2a+η 1 + ea+η
= ln , (6.91)
a(1 − e−2a ) 1 + eη 1 + e−a+η

where a = x(1 + u) and converting to the plasma rest frame, we finally obtain

G 2F eBT 2 E
W (ν → νe− e+ ) = (C V + C A )2 (1 − u)2
4π 3
τ 1+u
 2
dξ cosh ξ + cosh η
× ln +
(1 − e−ξ )(1 + e−+ξ/τ +ην ) 1 + cosh η
0

+ (C A → −C A ; u → −u) , (6.92)

where  = E/Tν . The dependence of the probability (6.92) on the electron–

positron gas density n = n e− − n e+ is defined in terms of its chemical potential
[see (6.75)]. Note that the formula for the probability (6.92) holds for hot (μe  T )
and cold (μe  T ) plasmas. For low-density electron–positron and neutrino gases
(T, μe , Tν , μν → 0) formula (6.92) reproduces the result (6.36) for the probability
of the process ν → νe− e+ in the strong magnetic field limit, eB  E 2 sin2 θ,
without a plasma.
In the absence of a neutrino gas, Tν , μν → 0, the expression for the probabil-
ity (6.92) for a hot electron–positron plasma (T → ∞) becomes equal to 1/4 of the
probability in a pure magnetic field (6.36) as we indicated in Sect. 6.4.2 since the
statistical factors for an electron and positron in this limit are 1/2.

6.5.4 The Total Probability of the Neutrino Interaction

with Magnetized Electron–Positron Plasma

A correct analysis of the neutrino propagation process in a hot dense plasma in

the presence of a strong magnetic field requires to consider the complete set of
neutrino-electron processes. Specifically, in addition to the νe∓ → νe∓ scattering
reactions which also take place in the absence of a field, and the ν → νe− e+ pair
creation process which is only possible in a magnetic field, we also need to take
206 6 Neutrino-Electron Interactions in External Active Media

into account the “exotic” process when a neutrino captures an electron–positron pair
from the plasma: νe− e+ → ν. This process is only allowed when both a magnetic
field and a plasma are present. Then only the probability of the process summed over
all initial states of the plasma electrons and positrons is physically meaningful. The
probability of the νe∓ → νe∓ scattering channels is defined similarly as the sum
over all e− or e+ initial states. The total probability of the neutrino interaction with an
electron–positron plasma in a magnetic field is made up of the probabilities of these
processes. Thus, the probabilities of the scattering processes should be defined as

1
W (νe∓ → νe∓ ) = |S|2 dΓe∓ f e∓ dΓe∓ (1 − f e∓ ) dΓν (1 − f ν ) , (6.93)
T

where dΓ and f functions are defined in Eqs. (6.81)–(6.83). Similarly, the probability
for the pair capture process is:

1
W (νe− e+ → ν) = |S|2 dΓe− f e− dΓe+ f e+ dΓν (1 − f ν ). (6.94)
T

It can be seen from Fig. 6.3 that the scattering and pair capture processes correspond
to infinite kinematic regions since the initial electrons and positrons can formally have
any energy. Convergence of the integrals is provided by the distribution functions.
The expressions (6.93) and (6.94) are integrated by the same scheme as that
described above for the ν → νe− e+ pair creation process. An important factor
for the integration will be that the energy imparted from the neutrino to the active
medium q0 = E − E is not positive-definite. For the probability (per unit time) of
the neutrino scattering on magnetized plasma electrons we have

G 2F eBT 2 E
W (νe− → νe− ) =
4π 3
τ 1+u

2
dξ 1 + eη
× (C V + C A )2 (1 − u)2 ln
(1 − e−ξ )(1 + e−+ξ/τ +ην ) 1 + e−ξ+η
0
1−u
τ
 2
dξ 1 + eη
+ (C V − C A ) (1 + u)
2 2
ln
(1 − e−ξ )(1 + e−+ξ/τ +ην ) 1 + e−ξ+η
0
+ [(C V + C A )2 (1 − u)2 + (C V − C A )2 (1 + u)2 ]
∞ 
dξ 1 + eη
× ln . (6.95)
(eξ − 1)(1 + e−−ξ/τ +ην ) 1 + e−ξ+η
0

Taking into account the distribution functions (6.88), the probability of scattering on
positrons is obtained from Eq. (6.95) by substituting η → −η. For the pair capture
channel we have
6.5 Neutrino in Strongly Magnetized Electron–Positron Plasma 207

G 2F eBT 2 E 
W (νe− e+ → ν) = (C V + C A )2 (1 − u)2
4π 3
(6.96)
∞
 dξ cosh ξ + cosh η
+ (C V − C A )2 (1 + u)2 ln .
(eξ − 1)(1 + e−−ξ/τ +ην ) 1 + cosh η
0

As we have already noted, only the total probability of neutrino interaction with
an electron–positron plasma is physically meaningful:

W (ν → ν) = W (ν → νe− e+ ) + W (νe− e+ → ν)
+ W (νe− → νe− ) + W (νe+ → νe+ ). (6.97)

It was found that this quantity had a substantially simpler form:

G 2F eBT 2 E
W (ν → ν) = (C V + C A )2 (1 − u)2
4π 3
(6.98)
    
τ (1 + u)
× F1 − F1 (−∞) + (C A → −C A ; u → −u) ,
2

where F1 (z) is one of the set of functions defined as

z
ξ k dξ
Fk (z) = . (6.99)
(1 − e−ξ )(1 + e−+ην +ξ/τ )
0

It is interesting that the dependence on the chemical potential of the electron–

positron gas μ which was present in the probabilities of the various processes, was
cancelled in the total probability.
At first glance, this result seems unusual. Indeed, the chemical potential of a
strongly magnetized plasma according to Eq. (6.75) is proportional to its density,
while the total probability of the neutrino interaction with the plasma appears to
be independent on the density. Turning to individual channels, one can see that for
the process of the pair production ν → νe− e+ , the increase of the plasma density
should lead to a decrease of the probability by reducing the number of free energy
levels for electrons and positrons. In the process of the pair capture, νe− e+ → ν,
the probability, respectively, increases with the density. Both of these mechanisms
are important for the scattering channels. However, the found effect of the exact
cancellation of these mechanisms in the total probability, apparently, could not be
predicted in advance, without specific calculations.
208 6 Neutrino-Electron Interactions in External Active Media

For a rarefied neutrino gas the probability (6.98) is expressed in terms of the Euler
dilogarithm Li2 (x):

G 2F eB T 2 E 2 E sin θ
2 4
W (ν → ν) = (C 2
V + C A )
4π 3 4T 2

+ (C V + C A )2 (1 − cos θ)2 Li2 1 − e−E(1+cos θ)/2T

+ (C V − C A )2 (1 + cos θ)2 Li2 1 − e−E(1−cos θ)/2T

π2  2 
+ (C V + C 2A )(1 + cos2 θ) − 4 C V C A cos θ . (6.100)
3

We remind that the nth-order polylogarithm Lin (x) is defined as

∞
 xk
Lin (x) = . (6.101)
kn
k=1

The relative contributions of the plasma and the magnetic field to the process of
neutrino interaction with the active medium are illustrated in Fig. 6.4 which gives the
ratio of the probabilities of neutrino interaction with a magnetized plasma and a pure
magnetic field, Rw = W B+ pl /W B , for the angle θ = π/2 as a function of the ratio
of the neutrino energy to the plasma temperature. It can be seen that the interaction
probability increases with the temperature increase.
The probability (6.98) determines the partial contribution of these processes to
the opacity for neutrino propagation in a medium. An estimate of the mean free path
associated with neutrino-electron processes gives
   3
1 103 Be 5 MeV
λe = 170 km . (6.102)
W B T

This should be compared with the neutrino mean free path as a result of interaction
with nucleons, which is of the order of a kilometer at the density ρ ∼ 1012 g cm−3 .
At first glance the influence of the neutrino-electron reactions on the neutrino prop-
agation process is negligible. However, the mean free path does not exhaust the
neutrino physics in a medium. Other important quantities in astrophysical appli-
cations are the neutrino energy and momentum losses. Of particular importance
is the asymmetry of the neutrino momentum loss caused by the influence of an
external magnetic field. Many attempts have been made to calculate these asym-
metries caused by neutrino-nucleon processes associated with the problem of the
high proper velocities of pulsars (see Sect. 6.5.8). As we shall show, despite the
relatively low probability of the neutrino-electron processes, their contribution to
the asymmetry may be comparable to the contributions of the neutrino-nucleon
processes.
6.5 Neutrino in Strongly Magnetized Electron–Positron Plasma 209

Fig. 6.4 Ratio of the probabilities of neutrino interaction with a magnetized plasma and a pure
magnetic field, Rw = W B+ pl /W B , for θ = π/2 as a function of the ratio of the neutrino energy to
the plasma temperature

6.5.5 Mean Losses of the Neutrino Energy and Momentum

In studies of these neutrino-electron interactions in a magnetic field and/or plasma

[6–9, 28], the analysis has usually been confined to calculation of the probabilities
and cross sections of processes. As we have noted, not only the probabilities of the
processes are of practical interest for astrophysics but also the average loss of neutrino
energy and momentum in the medium1 which can be determined by the four-vector
Q α , see Eq. (6.59). The zeroth component Q 0 is associated with the average energy

1In general a neutrino can lose and acquire energy and momentum so that we shall subsequently
understand “loss” of energy and momentum in the algebraic sense.
210 6 Neutrino-Electron Interactions in External Active Media

lost by a single neutrino per unit time and the spatial components Q are associated
with the loss of the neutrino momentum per unit time.
For a purely magnetic field the four-vector of the losses Q α was calculated in
Sect. 6.4.1. In that case, the losses were caused by the only possible process in the
absence of plasma, the pair creation during the motion of a neutrino in a strong
magnetic field ν → νe− e+ . In the strong magnetic field limit for the zeroth and
z-components of the vector Q α we obtained (the field is directed along z)

(B) G 2F eB E 5 sin4 θ
Q 0,z = 3
(6.103)
 48π 
× C V2 + C 2A + 2C V C A cos θ, (C V2 + C 2A ) cos θ + 2C V C A .

It can be seen from Eq. (6.103) in particular that even for an isotropic neutrino
momentum distribution the average momentum loss will be nonzero (proportional
to C V C A ) because of parity nonconservation in weak interaction. As it was shown in
6.4.3, in fields of ∼103 Be the integral asymmetry of the neutrino
 emission
 caused by
the component Q z and determined by the expression A = | P|/ |P| could only
reach the scale of ∼1 % required to explain the observed pulsar proper velocities as
a result of the ν → νe− e+ process only.
In the presence of a magnetized plasma our calculations yield the following result
for the same components of the loss four-vector:

G 2F eBT 3 E 2
Q 0,z = (C V + C A )2 (1 − u)2
4π 3
(6.104)
    
τ (1 + u)
× F2 − F2 (−∞) ± (C A → −C A ; u → −u) ,
2

where the function F2 (z) was determined in Eq. (6.99), and the upper or lower
signs correspond to the zeroth and z components. Our result for the loss four-vector
obtained for the case of a purely magnetic field (6.103) is reproduced from Eq. (6.104)
in the low-density plasma limit (T, Tν , μν → 0).
In order to illustrate the relationship between the contributions of the plasma and
the magnetic field to the four-vector of the neutrino energy and momentum losses in
an active medium we shall consider the simpler situation of a low-density neutrino
gas and rewrite Eq. (6.104) for the angle θ = π/2 in the following form:
 
G 2 eB E 5 E
Q 0,z (θ = π/2) = F 3 C V2 + C 2A , 2C V C A F , (6.105)
48π T

where
6 24 48
F(x) = 1 + ln 1 − e−x/2 − 2 Li2 e−x/2 − 3 Li3 e−x/2 . (6.106)
x x x
6.5 Neutrino in Strongly Magnetized Electron–Positron Plasma 211

0
F(E / T)

-2

-4

-6

-8
0 2 4 6 8

E/T

Fig. 6.5 The function F (E/T ) introduced in Eq. (6.105) and determining the dependence of the
components of the four-vector of the neutrino energy and momentum losses in a magnetized plasma
on the ratio of the neutrino energy to the plasma temperature

It can be seen from a comparison of (6.105) with Eq. (6.103) for θ = π/2 that
the function F(E/T ) is the ratio of the components of the loss vector in a magne-
tized plasma and in a purely magnetic field. Figure 6.5 gives a graph of the function
F(E/T ). It can be seen that at E = E 0 3.4 T there is a unique “window of
transparency” when a neutrino does not exchange energy and momentum with a
magnetized plasma. The negative values of the function F(E/T ) at lower energies
imply that the neutrino captures energy from the plasma and acquires momentum in
the opposite direction to the magnetic field. At energies higher than E 0 the neutrino
imparts energy to the plasma and also momentum in the direction of the field. This
may have extremely interesting astrophysical consequences.
212 6 Neutrino-Electron Interactions in External Active Media

6.5.6 Integral Action of Neutrinos on a Magnetized Plasma

As an illustration of the application of our results to astrophysical conditions we

estimate the volume density of the energy lost by a neutrino per unit time Ė and the
component Fz (parallel to the field) of the volume density of the force acting on the
plasma from neutrinos 
1
(Ė, Fz ) = dn ν Q 0,z , (6.107)
E

where dn ν is the initial neutrino density:

d3 P Φ(ϑ, R)
dn ν = (E−μ
. (6.108)
(2π) e
3 ν )/Tν + 1

Here, the angular distribution of the initial neutrinos is taken into account in the
function Φ(ϑ, R), ϑ = cos α, α is the angle between the neutrino momentum and
the radial direction in the star, and R is the distance from the center of the star. At the
same time, the similar function Φ(ϑ , R) should be introduced in the statistical factor
(1− f ν ) when integrating over the momenta of the final neutrino. In a supernova shell,
the neutrino angular distribution is close to isotropic [27] so that in the expansion
of the function Φ in terms of ϑ, we can confine ourselves to the lowest Legendre
polynomials P (ϑ) and this function can be expressed in terms of the average values
ϑ and ϑ2  (which depend on R) as follows:

Φ(ϑ, R) 1 + 3P1 (ϑ)P1 (ϑ) + 5P2 (ϑ)P2 (ϑ). (6.109)

Neutrinos leaving the central region of a star at high temperature enter the periph-
eral region where a strong magnetic field is generated and the temperature of the
electron–positron gas is lower. In this case the spectral temperatures for different
types of neutrino differ [23, 27]:

Tνe 4 MeV, Tν̄e 5 MeV, Tνμ,τ Tν̄μ,τ 8 MeV . (6.110)

The action of a neutrino on a plasma leads to the establishment of thermal equilib-

rium, Ėtot = 0. When analyzing this equilibrium we need to take into account the
contributions to Ėtot made by all processes of neutrino interaction with the medium.
As we have noted, the probability of the β processes νe +n ↔ e− + p is substantially
higher than that for neutrino-electron processes so that these dominate in the energy
balance. The energy transferred per unit time per unit plasma volume as a result of
these processes involving only electron neutrinos may be expressed in the form

Tνe − T
Ė(β) B . (6.111)
T
6.5 Neutrino in Strongly Magnetized Electron–Positron Plasma 213

From this it follows that as a result of neutrino heating the plasma temperature
should be very close to the spectral temperature of the electron neutrinos (T Tνe ).
However, the contribution to Ė made by other types of neutrino whose spectral
temperatures exceed Tνe , has the result that the plasma temperature is slightly higher
(T  Tνe ). It is therefore meaningful to make separate estimates of the contributions
to (Ė, Fz ) made by neutrino-electron processes involving νe and all other neutrinos
and antineutrinos.
We stress that the appearance of the force density Fz in Eq. (6.107) is caused
by interference between the vector and axial-vector couplings in the effective
Lagrangian (4.66) and is a macroscopic manifestation of parity nonconservation
in weak interactions. At first glance, the main contribution to Fz should be made
by electron neutrinos since C V (νe )  C V (νμ,τ ). However, as we shall show below,
the main contributions are made by μ and τ neutrinos and antineutrinos (as a result
of the conservation of C P parity neutrinos and antineutrinos push the plasma in the
same direction). This is because in the vicinity of the νe neutrinosphere the spec-
tral temperatures of the other types of neutrinos differ substantially from the plasma
temperature T Tνe .

6.5.6.1 Processes Involving Electron Neutrinos

We obtained the following expression for the volume density of the neutrino energy
losses and the force density (6.107):

G 2F eBT 7
(Ė, Fz )νe = C V2 + C 2A , 2C V C A
3π 5
∞ 3 ∞
x dx y 3 dy
× (τe − 1)
ex − 1 (1 + e−x−y+ην )(1 + e y−ην )
0 0
  ∞ 3 ∞ 
27 1 x dx y 3 (3y − x)dy
+ ϑ  −
2
, (6.112)
8 3 ex − 1 (x + y)2 (1 + e y−ην )
0 0

where τe = Tνe /T . This formula is written assuming a small deviation from thermal
equilibrium between the neutrino gas and the electron–positron plasma (τe −1)  1,
and relatively weak asymmetry of the neutrino distribution, (ϑ2  − 1/3)  1, is
also assumed.
A numerical estimate gives
   7
30 erg 20 dyne B T
(Ė, Fz )νe 2.0 × 10 , 0.57 × 10
cm3 s cm3 1016 G 4 MeV
  
1
× eην (τe − 1) + 0.53 ϑ2  − . (6.113)
3
214 6 Neutrino-Electron Interactions in External Active Media

6.5.6.2 Processes Involving ν̄e , νμ,τ , ν̄μ,τ

In this case (Tν /T − 1) cannot be considered as a small parameter. However, the

relative contribution of the asymmetry of the neutrino distribution is small [27] and
can be neglected.
For numerical estimates we can conveniently express the values Ė and Fz (6.107)
in the following form:

(Ė, F)νi A (C V2 + C 2A , 2C V C A ) ϕ(ηi ) ψ(τi ), (6.114)

where
  7  erg
12G 2F eBT 7 B T 1.6 × 1030 cm3 s
,
A= = ×
π5 1016 G 4 MeV 0.55 × 1020 dyne ,
cm3
ηi4 π 2 ηi2 7π 4 7π 4
ϕ(ηi ) = + + + Li4 (−e−ηi ), ϕ(0) = 0.947,
24 12 360 720
∞
τ7 y 2 dy  (τi −1)y 
ψ(τi ) = i e − 1 ,
6 eτi y − 1
0
 π4
ψ(τi )τ →1 (τi − 1) . (6.115)
i 90
Formulas (6.112)–(6.115) demonstrate in particular that the action of each indi-
vidual neutrino fraction on an electron–positron plasma would go to zero when
thermodynamic equilibrium is established between this fraction and the plasma
τi = 1, ϑ = 0, ϑ2  = 1/3.
We show that the main contribution to the neutrino action on the plasma is made
by μ and τ neutrinos and antineutrinos. In fact the function ψ(τi ) (6.115) increases
rapidly as the difference between the spectral temperature of the neutrinos and
the plasma temperature increases. For example, at temperatures (6.110) we have
ψ(1.25) 0.824 for electron antineutrinos and ψ(2) 38.47 for μ and τ neutrinos
and antineutrinos. This factor leads to compensation for the smallness of the con-
stant C V (νμ,τ ) and makes the νμ,τ , ν̄μ,τ contribution not only comparable with the
contribution of the electron neutrinos and antineutrinos but even dominant.
As we have noted, the contribution of neutrino-electron processes to the energy
action of a neutrino on the plasma is small compared with the contribution of β
processes and leads to a small departure from equilibrium between electron neutrinos
and the plasma so that the total contribution of β processes and all νe processes to
the value of Ė is zero.
For the force action of a neutrino on the plasma parallel to the magnetic field
described by Fz in formulas (6.112)–(6.115) the total contribution of all types of
neutrinos is given by
6.5 Neutrino in Strongly Magnetized Electron–Positron Plasma 215

  7
dyne B T
Fz 3.6 × 10 20
. (6.116)
cm3 1016 G 4 MeV

Here we assumed for estimates that the chemical potentials of the neutrinos are
zero [23]. Note that the value (6.116) was independent of the chemical potential of
an electron–positron plasma.
The force density (6.116) should be compared with the result for a similar force
caused by β-processes [29, 30]. Under the same physical conditions our value of the
force as a result of neutrino-electron processes is of the same order of magnitude
and, which is particularly important, of the same sign as the result of [29, 30]. Thus,
the role of neutrino-electron processes in a high-intensity magnetic field may be
significant in addition to the contribution of β processes.
The force density (6.116) could lead to a very interesting consequences if a strong
toroidal magnetic field [18, 19] is generated in the supernova envelope. This possi-
bility is analyzed in detail below in Sect. 6.5.8.
As we know, in existing systems for numerical modeling of astrophysical cat-
aclysms such as supernova explosions and coalescing of neutron stars, where the
physical conditions being studied can be achieved in principle, the neutrino-electron
interaction effects studied by us were neglected. However, in detailed analyses of
these astrophysical processes it may be important to take into account the influence
of an active medium such as a magnetized e− e+ plasma, on quantum processes
involving neutrinos.

6.5.7 Neutrino-Electron Processes Involving the Contributions

of the Excited Landau Levels

In the case when an active medium consists of a magnetic field and very dense plasma,
such that the condition is valid: μ2  2eB, the plasma electrons could occupy the
excited Landau levels. The full set of neutrino-electron processes in such physical
conditions of dense magnetized plasma was analyzed in Refs. [31, 32]. In these
papers, in contrast to the above-considered situation, the physical conditions were
analyzed when the magnetic field was not so strong, whereas the density of plasma
was large. Thus, the chemical potential of electrons μe was the dominant parameter:

μ2e >2eB  T 2 , E 2  m 2e . (6.117)

Here, T is the plasma temperature, E is the typical neutrino energy. Under the con-
ditions (6.117), plasma electrons occupy the excited Landau levels. At the same time
it is assumed that the magnetic field strength being relatively weak, see Eq. (6.117),
is strong enough, so that the following condition is satisfied:

eB  μe E. (6.118)
216 6 Neutrino-Electron Interactions in External Active Media

In the present astrophysical context, the conditions (6.117) and (6.118) could be
realized, for example, in a supernova envelope, where the electron chemical potential
is assumed to be μe ∼ 15 MeV, and the plasma temperature T ∼ 3 MeV. The magnetic
field could be as high as B ∼ 1015 − 1016 G. Under the conditions considered,
the approximation of ultrarelativistic plasma is a good one, so we shall neglect the
electron mass wherever this causes no complications.
As it was shown in [31], the total set of neutrino-electron processes reduces under
the conditions (6.117) and (6.118) to the process of neutrino scattering on plasma
electrons. Moreover, both initial and final electrons occupy the same Landau level.
The neutrino-electron scattering in dense magnetized plasma was investigated in
[28]. Numerical calculations of the differential cross-section of this process in the
limit of a weak magnetic field (eB < μe E) were performed. The purpose of this study
based on [32], is to calculate analytically not only the probability of the neutrino-
electron scattering process, but also the volume density of the neutrino energy and
momentum losses under the conditions (6.117) and (6.118).

6.5.7.1 Neutrino-Electron Scattering Probability

The probability of the neutrino-electron scattering per unit time can be obtain by
integration over the final and the initial electron states:
n
max  
1
W (νe− → νe− ) = |S|2 dΓe− f e− (εn ) dΓe − [1 − f e − (εn )]
T
n=0 s,s
× dΓν [1 − f ν (E )] . (6.119)

Here, n max corresponds to the maximal possible Landau level number, which is
defined as the integer part of the ratio μ2e /(2eB)  1, εn pz2 + 2eBn is the
energy of an ultrarelativistic plasma electron occupying the nth Landau level, E
is the final neutrino energy, μ̃ν and Tν are the effective chemical potential and the
spectral temperature of the neutrino gas respectively. In a general case the neutrino
spectral temperature Tν can differ from the plasma temperature T (we do not assume
an equilibrium between neutrino gas and plasma).
The details of integration over the phase space of particles can be found in [31].
The result of the calculation of the probability (6.119) can be presented in a relatively
simple form:

G 2F (C V2 + C 2A ) eB T 2 E 
n max
1
W (νe− → νe− ) =
4π 3 z2
n=0
b
× ((1 + z )(1 + u ) − 4uz)
2 2
Φ(ξ)dξ (6.120)
−a
6.5 Neutrino in Strongly Magnetized Electron–Positron Plasma 217

b 
1 2
+ (z − 1)(z − u) ξΦ(ξ)dξ + (u → − u) ,
zr τ
−a

where z = 1 − 2eBn/μ2e , Φ(ξ) = ξ[(eξ − 1)(eην −r −ξ/τ + 1)]−1 , a = r τ z(1 +

u)/(1 + z) and b = r τ z(1 − u)/(1 − z), r = E/Tν , τ = Tν /T , ην = μ̃ν /Tν ,
u = cos θ, θ is the angle between the initial neutrino momentum k and the magnetic
field direction. The variable ξ defines the spectrum of the probability (6.120) in terms
of the final neutrino energy, ξ = (E − E)/T .
In the limit of a very dense plasma (μ2e  eB), when a great number of Landau
levels are occupied by plasma electrons, one can transform the summation over n to
an integration over z:

[μ2e /(2eB)]
 1
μ2e
F(z(n)) F(z) zdz . (6.121)
eB
n=0 0

In this case, the contribution from the lowest Landau levels turns out to be negligi-
bly small, so the main contribution to the probability arises from the highest Landau
levels. In this limit, the probability (6.120) can be rewritten in the following form:

1
− G 2 (C 2 + C 2A ) μ2e T 2 E
− dz
W (νe → νe ) = F V
4 π3 z
0
b
× ((1 + z )(1 + u ) − 4uz)
2 2
Φ(ξ)dξ (6.122)
−a
b 
1 2
+ (z − 1)(z − u) ξΦ(ξ)dξ + (u → − u).
zr τ
−a

As one can see, the probability (6.120) does not depend on the value of the
magnetic field strength, but is not isotropic. The dependence on the angle θ manifests
this anisotropy of the neutrino-electron process in the presence of a magnetic field.
In the limit of a rare neutrino gas when f ν (E )  1, the result has a more simple
form:

G 2F (C V2 + C 2A ) μ2e E 3
W (νe− → νe− ) I (u) , (6.123)
12 π 3
1
zdz
I (u) = u 4 (3z 2 + 2z + 1) − 12 u 2 z + z 2 + 2z + 3 .
(1 + z)2
0
218 6 Neutrino-Electron Interactions in External Active Media

For purposes of comparison, we present here the probability of the neutrino-

electron scattering in the absence of field in the same limit of the rare neutrino gas:

G 2F (C V2 + C 2A ) μ2e E 3
Wvac = . (6.124)
15 π 3
The numerical estimate of the ratio of the probabilities (6.123) and (6.124) is
presented in Fig. 6.6. It is seen that the probability in a magnetized plasma exceeds
the vacuum probability in the vicinity of the point θ = π/2 only.

6.5.7.2 Integral Neutrino Action on a Magnetized Plasma

In this section, we calculate the volume density of the neutrino energy and momentum
losses per unit time in a medium. According to Eqs. (6.59), (6.107), and (6.108), we
can write:

1 (q0 , q) d3 P
(Ė, F ) = dW , (6.125)
(2π)3 e(E−μ̃ν )/Tν + 1

where qα is the difference between the momenta of the initial and final neutrinos,
qα = Pα − Pα . The zeroth component, Ė, determines the neutrino energy loss from
unit volume per unit time. In general, a neutrino propagating through plasma can both
lose and capture energy. So, we mean the “loss” of energy in the algebraic sense.
For the neutrino energy loss from unit volume per unit time due to the scattering
νe− → νe− in the limit of a very dense plasma we obtain the following result:

Fig. 6.6 The relative probability of the neutrino-electron scattering in a magnetized plasma as a
function of the angle between the initial neutrino momentum and the magnetic field direction. Wvac
is the probability in a non-magnetized plasma
6.5 Neutrino in Strongly Magnetized Electron–Positron Plasma 219

G 2F (C V2 + C 2A ) 2 4
Ė = μe T n ν J B (τ ), (6.126)
π3
1 ∞
τ4 dz
J B (τ ) = dy y 2 [y(1 − z 2 ) + 4 z (1 + z 2 ) ]
2 z2
0 0
1 − e y(1−τ ) −y(1+z)/2z
× e , (6.127)
1 − e−yτ

where n ν is the density of initial neutrinos, the parameter τ has a meaning of a relative
neutrino spectral temperature, τ = Tν /T . It is interesting to compare this result with
the one in a non-magnetized plasma which can be presented in a similar form:

G 2F (C V2 + C 2A ) 2 4
Ė B=0 = μe T n ν J B=0 (τ ), (6.128)
π3
∞
eξ(τ −1) − 1
J B=0 (τ ) = 4τ 4
dξ ξ 2 . (6.129)
eξτ − 1
0

The functions J B (τ ) and J B=0 (τ ) define the dependence of the neutrino energy
losses on the relative neutrino spectral temperature in a magnetized plasma and in a
plasma without field respectively. In the limit of a sufficiently large neutrino spectral
temperature τ  1 (Tν  T ) the functions take the form:

J B (τ ) 4.35 τ 4 , J B=0 (τ ) 8 τ 4.

The graphs of the functions J B (τ ) and J B=0 (τ ) are presented in Fig. 6.7.
As one would expect, at neutrino spectral temperature smaller than the plasma
temperature Tν < T (τ < 1) the functions J B (τ ) and J B=0 (τ ) are negative. It
implies that a neutrino propagating in a medium picks up energy from the plasma.
When Tν > T (τ > 1), the neutrino gives energy to the plasma. When τ = 1
there is a thermal equilibrium when there is no energy exchange between neutrino
and electron–positron plasma. It can be seen that the neutrino energy loss in a mag-
netized plasma is less than the one in a non-magnetized plasma. Hence, under the
conditions (6.117) and (6.118), the magnetized plasma becomes more transparent
for neutrinos than in the case of plasma without field.
As for the vector F in Eq. (6.125), it is associated with the volume density of the
neutrino momentum loss per unit time, and therefore it defines the neutrino force
acting on plasma. Because of the isotropy of plasma in the absence of a magnetic field,
one would expect that in the presence of a magnetic field the neutrino force action
would be directed along the magnetic field only. However, as it was shown before, the
probability of the neutrino-electron scattering (6.122) is symmetric with respect to
the substitution u → −u (or θ → π − θ). This means that the neutrino scattering on
excited electrons does not give a contribution to the neutrino force acting on plasma
220 6 Neutrino-Electron Interactions in External Active Media

Fig. 6.7 The functions J B (τ ) (solid line) and J B=0 (τ ) (dashed line) versus the relative spectral
neutrino temperature

along the magnetic field. Thus, under the conditions (6.117) and (6.118), there is
no neutrino force on plasma at all. Therefore, this force is caused by a contribution
of neutrino interactions with ground Landau level electrons only, and the results
presented in Eqs. (6.112)–(6.116) have in fact a more general applicability. It may
be used even in the limit of dense plasma when chemical potential is considerably
greater than the magnetic field strength (μ2e  eB).

6.5.8 Pulsar Natal Kick Via Neutrino-Triggered

Magnetorotational Asymmetry

In this subsection, we will try to apply the results presented above to the well-known
problem of large kick velocities of pulsars born in supernova explosions.

6.5.8.1 Pulsar Natal Kick

This problem has been discussed for more than 40 years. The total list of publications
with observational data is fairly long. Here, we will point out only the first papers [33,
34], where this problem was formulated and the papers where the data on a sample
of 99 pulsars [35] and a sample of 233 pulsars [36] were summarized. In the latter
paper, the mean velocity for the sample of 233 pulsars was estimated to be 400 km/s,
with more than 15 % of the pulsars having velocities greater than 1000 km/s. The
velocities of the two fastest pulsars PSRs B2011+38 and B2224+64 were estimated
to be ∼1600 km/s.
6.5 Neutrino in Strongly Magnetized Electron–Positron Plasma 221

It is important that a correlation was established between the pulsar velocity direc-
tions and rotation axes. Initially, having analyzed a sample of 29 pulsars, Deshpande
et al. [37] concluded that the mechanisms predicting a correlation between the pulsar
velocity and rotation axis were ruled out. Subsequently, however, Johnston et al. [38]
presented strong observational evidence for a relationship between the direction of a
pulsar’s motion and its rotation axis. A sample of 25 pulsars younger than those used
in [37] was analyzed. In particular, for 10 pulsars detected in [38] an offset between
the velocity vector and the rotation axis, which is either less than 10◦ or more than
80◦ , a fraction that is very unlikely by random chance.
Obviously, an asymmetry in a supernova explosion is responsible for the initial
kick, but its nature has not yet been revealed. Various explanations of this asymmetry
have been offered in a number of papers.
The attempts to describe the effect only by the hydrodynamics of a supernova
explosion without invoking other physical factors could not explain the large veloci-
ties. Three-dimensional simulations of the explosion with the assumption of an initial
asymmetry in the supernova core before its collapse, which increases during its col-
lapse, lead to a pulsar velocity of no more than 200 km/s [39]. Multidimensional
simulations by H.-T. Janka et al. [40], where the explosion anisotropy develops chaot-
ically, yielded a possible pulsar velocity of 103 km/s. However, the established cor-
relation between the pulsar velocity direction and rotation axis [38] is not explained
in this approach.
In addition to the hydrodynamic approach, there are also other ideas of explain-
ing the pulsar velocities. For example, the pulsar escape was considered during the
decay of a close binary system [41]. Another example was the pulsar acceleration
within several months after the explosion due to asymmetric electromagnetic radia-
tion caused by the inclination of the magnetic moment with respect to the rotation
axis and its displacement relative to the stellar center [42]. However, both of these
scenario lead to velocities of the scale of 100 km/s.
In our view, the mechanisms involving neutrinos appear most interesting. Neu-
trinos are known to carry away about 99 % of the total emitted supernova energy
E ∼ 3 × 1053 erg. If there is an asymmetry in the neutrino escape of ∼3 %, then
they would carry away a momentum of ∼ 0.03 E/c. The compact explosion rem-
nant, i.e., a neutron star with a mass of ∼ 1.4M , would get the same momentum.
In this case, its velocity can be easily estimated to be ∼1000 km/s.
An asymmetric neutrino (antineutrino) radiation during a collapse via Urca
processes in a strong magnetic field of 1014 −1015 G in a supernova core was consid-
ered [43–48] as a reactive force expelling the neutron star. However, as was subse-
quently shown [49–52], the neutrinos produced in electroweak processes have small
mean free paths in the matter of the central part of a supernova and cannot provide
high pulsar velocities.
An interesting mechanism of asymmetry in neutrino radiation during a supernova
explosion was considered in Refs. [53–55]. Here, the neutrino flux asymmetry results
not from parity violation [43–48], but from an asymmetry in the distribution of the
toroidal magnetic field developing during the collapse.
222 6 Neutrino-Electron Interactions in External Active Media

A lively debate was generated by the idea [56], according to which the asymmetry
in the neutrino flux from a protoneutron star appears due to neutrino oscillations
in matter and an intense magnetic field. The neutrinosphere for ντ lies within the
neutrinosphere for νe , and the resonance transition νe → ντ is possible under certain
conditions in the region between the neutrinospheres, where νe are ‘entangled’ in
the medium, while ντ are ‘free’ to escape. Therefore, the surface of the resonance
transition becomes an effective neutrinosphere for ντ . In the presence of a magnetic
field, this sphere is deformed along the field. Since the temperature depends on the
radius, the neutrinosphere deformation results in an anisotropy of the energy flux
carried away by neutrinos. This should impart a kick to the nascent neutron star.
However, the idea of an initial pulsar kick due to a deformed neutrinosphere [56]
came under serious criticism [57]: after the neutrinosphere deformation, the sur-
faces of constant temperature will also be deformed, because precisely the neutrinos
provide a thermal equilibrium. However, the main problem of this model was soon
revealed: it required the existence of neutrinos with a mass of ∼100 eV. The estab-
lished constraint on the neutrino mass, m ν < 2 eV, ‘closed’ the model.
Attempts were also made to explain the large space velocities of young pulsars
using some possible nonstandard properties of neutrinos. For example, a mechanism
was proposed [58] based on the resonant spin-flavor precession of neutrinos with a
transition magnetic moment in the magnetic field of a supernova. It was assumed
that the asymmetric neutrino radiation could be caused by a distortion of the res-
onance surface due to matter polarization effects in the supernova magnetic field.
The authors [58] argued that the necessary field strength should be 1016 G, with the
neutrino parameters at the level of existing experimental bounds. However, as was
pointed out in [57], the magnetic fields required in the model [58] should actually
be more than an order of magnitude stronger.

6.5.8.2 The Initial Pulsar Kick and Sterile Neutrinos

Sterile neutrinos appeared on stage in the paper [59] (see also the review [60] for
details). Here, as in [56], the deformation of the neutrinosphere by a magnetic field
was discussed, but instead of the oscillations νμ,τ ↔ νe , the transitions into ‘heavy’
sterile neutrinos νμ,τ ↔ νs were considered. The model was attractive in that the
heavy sterile neutrinos (with a mass scale of a few keV) could simultaneously solve
two problems: providing an initial velocity of pulsars, they could also play the role
of dark matter.
However, when we reproduced the calculations performed in Refs. [59, 60], we
found that the asymmetry was overestimated in [60] by a factor of 15. In other words,
the necessary magnetic field strength for the declared asymmetry should be a factor
of 15 larger: not ∼3 × 1016 but ∼4.6 × 1017 G.
Another scenario for using sterile neutrinos to explain the pulsar kick, based on
off-resonance transitions was developed in [61]. In this scenario, the fact was used that
sterile neutrinos could be produced in beta processes through neutrino mixing, with
6.5 Neutrino in Strongly Magnetized Electron–Positron Plasma 223

this process being suppressed due to the smallness of the mixing angle. Nevertheless,
they could carry away a significant amount of energy due to two factors:
(1) the neutrinos in the supernova core had energies, ∼150 MeV, much greater than
those of the active neutrinos, ∼20 MeV, emitted from the neutrinosphere;
(2) the emission here originated from the volume, not from the surface.
In the presence of a magnetic field, the neutrinos were emitted asymmetrically
and this asymmetry was retained, because the sterile neutrinos were not absorbed but
escaped freely, as distinct from the situation considered in Refs. [43–47]. However, as
our analysis shows, the authors [61] overestimated the asymmetry at least by a factor
of 40. In other words, the magnetic field strengths should be a factor of 40 larger to
achieve the asymmetry declared by these authors: not ∼1016 but ∼4×1017 G. In our
view, a mistake was made in calculating k0 defined in Eq. (9) and presented in Fig.
2 of [61]. Note that the authors call k0 the fraction of electrons in the lowest Landau
level, while actually this is the fraction of the electron energy squared in the lowest
Landau level. It is this quantity that defines the asymmetry of the neutrino-electron
interaction in beta processes. It can be shown that the result [61] is erroneous, both
by direct numerical calculations and analytically. Indeed, using Eqs. (9) and (10)
from the paper under consideration, the expression for k0 can be transformed with a
good accuracy to
eB J2 (μe /T )
k0 , (6.130)
2T 2 J4 (μe /T )

where B is the magnetic field strength, μe and T are the chemical potential and
temperature of the electrons, and Jn (η) are the Fermi integrals:

∞
x n dx
Jn (η) = . (6.131)
e x−η+1
0

Depending on the electron chemical potential and the magnetic field strength, k0 was
overestimated in Fig. 2 of [61] by a factor from 40 to 90.
In the paper [62], a detailed numerical analysis presented of the transformation
of active neutrinos to sterile ones through an MSW-like resonance in a protoneutron
star to explain the initial pulsar kick. However, the magnetic field strength needed to
achieve the desirable effect should be 1017−18 G.

6.5.8.3 Back to Standard Neutrinos?

The following question arises: if we actually need such strong magnetic fields to
provide a natal neutron star kick from sterile neutrinos, is it possible to manage with
standard neutrinos?
224 6 Neutrino-Electron Interactions in External Active Media

As has already been noted, see Eqs. (6.69) and (6.70), the asymmetry in the emis-
sion of standard neutrinos in a strong poloidal magnetic field at the scale of 1016 G
was not enough to provide the observable neutron star kick.
Note that the mechanism of a significant enhancement in the magnetic field
strength during a supernova explosion is known. This is the magnetorotational model
for the generation of a toroidal magnetic field in a supernova explosion [18, 19, 63].
A poloidal magnetic field being enhanced during supernova core collapse and frozen
in plasma produces a strong toroidal magnetic field due to the differential rotation,
which can be greater than the poloidal field by an order of magnitude.
A possible integral effect of neutrinos on a magnetized plasma was evaluated in
Sect. 6.5.6, and the combined force action of all types of neutrinos interacting with
an electron–positron plasma was obtained, see Eq. (6.116).
The contribution from the neutrino-nucleon processes was estimated in Refs. [29,
30]. For supernova envelope parameters Ye 0.2 and ρ 1011−12 g cm−3 , one can
obtain (‘ν N ’ means both Urca processes and ν N scattering)
 
B dyn
F B(ν N ) 2.4 × 1020 . (6.132)
1016 G cm3

It is important that the contributions from both neutrino-electron and neutrino-

nucleon processes have the same sign. The total neutrino force density is
 
(total) B dyn
FB 0.6 × 1021 . (6.133)
1016 G cm3

Note that the force density (6.133) is approximately five orders of magnitude
lower than the gravitational force density in the same part of the supernova and, con-
sequently, its influence on the radial dynamics of the supernova envelope is negligible.
However, when a toroidal magnetic field is generated in the envelope [18, 19, 63],
the force (6.133) directed along the field is in no way compensated. It can fairly
rapidly (in a time of the order of a second2 ) lead to a significant redistribution of the
tangential plasma velocity. In two toroids in which the magnetic fields have opposite
directions, the tangential plasma acceleration under the neutrino flux will then have
different signs with respect to the direction of rotational plasma motion. This effect
can lead to a significant redistribution of the magnetic field lines, concentrating them
predominantly in one of the toroids. A similar field configuration was considered in
the papers cited above [53–55], where the presence of an initial toroidal field was
needed for its appearance. The resulting considerable asymmetry of the magnetic
field energy in the two hemispheres can lead to an asymmetry of the supernova
explosion and, in particular, can explain the phenomenon of high intrinsic pulsar
velocities being discussed. In our view, it would be very interesting to model the

2 The cooling of a supernova envelope, the so-called Kelvin–Helmholz stage, is known to last for
about 10 s.
6.5 Neutrino in Strongly Magnetized Electron–Positron Plasma 225

toroidal magnetic field generation mechanism by taking into account the neutrino
force action on plasma via both neutrino-nucleon and neutrino-electron processes.

6.5.8.4 Neutrino-Triggered Magnetorotational Pulsar Natal Kick

The neutrino processes in a toroidal magnetic field frozen in plasma under consid-
eration impart an angular acceleration to a plasma element at distance R from the
rotation axis:  
F B 1
Ω̇ = 1.2 × 10 3
. (6.134)
ρR 10 G s2
16

This means that the increase in angular velocity in a time of ∼1 s will be

 
B 1
ΔΩ ∼ 10 3
. (6.135)
1016 G s

In one hemisphere the angular acceleration coincides with the direction of initial
rotation, while in the other hemisphere they are opposite. Pushing the plasma, the
neutrino flux curls the toroids in different directions.
Thus, three stages of a pulsar kick can be identified:
(i) the presupernova core collapses with rotation during 0.1 s with the generation
of a strong toroidal magnetic field due to the differential rotation;
(ii) pushing the plasma by the tangential force directed along the toroidal magnetic
field frozen in plasma, the neutrino outburst leads to a magnetic field asymmetry:
the field strength increases in one hemisphere and decreases in the other one,
during ∼1 s;
(iii) the pressure difference arising in the two hemispheres pushes the core.
According to the momentum conservation law, an energetic plasma jet can be
formed in a direction opposite to the pulsar velocity. Such plasma jets being formed
in supernova explosions could be gamma-ray burst sources [64]. Of course, a detailed
multidimensional numerical simulation of the process is needed. Let us make an
order-of-magnitude estimate of the effect that may be expected.
The pressure difference arising in the two hemispheres can be estimated as

B2 (eB)2
Δp = , (6.136)
8π 8πα
where α = 1/137 is the fine-structure constant. The magnetic field pressure causes
the compact supernova core, a protoneutron star of mass M, to accelerate:
 2  2  
dVkick B R 1.4 M km
1.6 × 10 5
sin 2θ Δθ , (6.137)
dt 1016 G 20 km M s2
226 6 Neutrino-Electron Interactions in External Active Media

Fig. 6.8 The region of a

strong toroidal magnetic field Δθ
in section by a meridional
half-plane. The symbols ⊗
and  denote the magnetic B
field directed away from and θ
toward us, respectively. If the
magnetic field in the upper
hemisphere exceeds that in the
lower one by a factor of ∼2.3,
as is shown in the figure, then R
the magnetic field pressure in
the upper hemisphere will be
a factor of ∼5.4 larger than
that in the lower one
B

where R, θ and Δθ are the parameters that characterize the region of a strong toroidal
magnetic field (see Fig. 6.8).
Taking Δθ ∼ 15◦ ∼ 41 and θ ∼ 45◦ for our estimation, we obtain
 2  2  
dVkick B R 1.4 M km
4 × 104 . (6.138)
dt 1016 G 20 km M s2

Actually the acceleration is not constant, because the expansion of the magnetic field
volume, which reduces the field strength, should be taken into account. From the
magnetic flux conservation we have p V 2 = const.
In the same geometry, for the initial pulsar kick velocity we obtain
   1/2  1/2
B0 R Δz 1.4 M km
Vkick 600 , (6.139)
1016 G 20 km 5 km M s

where B0 is the maximum toroidal field strength, and Δ z is the distance traveled
by the compact explosion remnant during the acceleration. It is natural to expect
that the field remained after the explosion will be much smaller than the maximum
strength B0 .
We emphasize that in our analysis we use the toroidal magnetic fields, which can
be greater than the poloidal fields used in other approaches by an order of magnitude.
In our view, a detailed multidimensional numerical simulation of the described
mechanism is needed. We hope that it will confirm this effect.
References 227

References

1. A.V. Borisov, V.Ch. Zhukovskii, A.V. Kurilin, A.I. Ternov, Yad. Fiz. 41, 743 (1985) [Sov. J.
Nucl. Phys. 41, 473 (1985)]
2. A. Erdas, M. Lissia, Phys. Rev. D 67, 033001 (2003)
3. A.V. Kuznetsov, N.V. Mikheev, Yad. Fiz. 70, 1299 (2007) [Phys. At. Nucl. 70, 1258 (2007)]
4. K. Bhattacharya, S. Sahu, Eur. Phys. J. C 62, 481 (2009)
5. M. Abramowitz, I.A. Stegun (eds.), Handbook of Mathematical Functions (Dover, New York,
1965)
6. E.A. Choban, A.N. Ivanov, Zh. Eksp. Teor. Fiz. 56, 194 (1969) [Sov. Phys. JETP 29, 109
(1969)]
7. A.V. Borisov, V.Ch. Zhukovskii, B.A. Lysov, Izv. Vyssh. Uchebn. Zaved. Fiz. No. 8, 30 (1983)
[Sov. Phys. J. 26, 701 (1983)]
8. M.Yu. Knizhnikov, A.V. Tatarintsev, Vestn. Mosk. Univ. Fiz. Astron. 25, 26 (1984)
9. A.V. Borisov, A.I. Ternov, V.Ch. Zhukovsky, Phys. Lett. B 318, 489 (1993)
10. A.V. Kuznetsov, N.V. Mikheev, Phys. Lett. B 394, 123 (1997)
11. A.V. Kuznetsov, N.V. Mikheev, Yad. Fiz. 60, 2038 (1997) [Phys. At. Nucl. 60, 1865 (1997)]
12. A.V. Borisov, N.B. Zamorin, Yad. Fiz. 62, 1647 (1999) [Phys. At. Nucl. 62, 1543 (1999)]
13. A.V. Kuznetsov, N.V. Mikheev, D.A. Rumyantsev, Mod. Phys. Lett. A 15, 573 (2000)
14. A.V. Kuznetsov, N.V. Mikheev, D.A. Rumyantsev, Yad. Fiz. 65, 303 (2002) [Phys. At. Nucl.
65, 277 (2002)]
15. V.V. Skobelev, Zh. Eksp. Teor. Fiz. 108, 3 (1995) [JETP 81, 1 (1995)]
16. A.V. Kuznetsov, N.V. Mikheev, Electroweak Processes in External Electromagnetic Fields
(Springer, New York, 2003)
17. L.D. Landau, E.M. Lifshitz, Course of theoretical physics, vol. 5, Statistical Physics, Part 2
(Nauka, Moscow, 1976; Pergamon, Oxford, 1980)
18. G.S. Bisnovatyi-Kogan, Astron. Zh. 47, 813 (1970) [Sov. Astron. 14, 652 (1971)]
19. G.S. Bisnovatyi-Kogan, Stellar Physics 2: Stellar Evolution and Stability (Nauka, Moscow
1989) (Springer, New York, 2001)
20. R.C. Duncan, C. Thompson, Astrophys. J. 392, L9 (1992)
21. P. Bocquet, S. Bonazzola, E. Gourgoulhon, J. Novak, Astron. Astrophys. 301, 757 (1995)
22. C.Y. Cardall, M. Prakash, J.M. Lattimer, Astrophys. J. 554, 322 (2001)
23. V.S. Imshennik, D.K. Nadyozhin, Usp. Fiz. Nauk 156, 561 (1988) [Sov. Phys. Usp. 31, 1040
(1988)]
24. D.K. Nadyozhin, in: Proceedings of the Baksan International School on Particles and Cosmol-
ogy, ed. by V.A. Matveev et al. (World Scientific, Singapore, 1992), pp. 153–190
25. A.V. Kuznetsov, N.V. Mikheev, Mod. Phys. Lett. A 14, 2531 (1999)
26. A.V. Kuznetsov, N.V. Mikheev, Zh. Eksp. Teor. Fiz. 118, 863 (2000) [JETP 91, 748 (2000)]
27. S. Yamada, H.-Th. Janka, H. Suzuki, Astron. Astrophys. 344, 533 (1999)
28. V.G. Bezchastnov, P. Haensel, Phys. Rev. D 54, 3706 (1996)
29. A.A. Gvozdev, I.S. Ognev, Pis’ma v Zh. Eksp. Teor. Fiz. 69, 337 (1999) [JETP Lett. 69, 365
(1999)]
30. A.A. Gvozdev, I.S. Ognev, Zh. Eksp. Teor. Fiz. 121, 1219 (2002) [JETP 94, 1043 (2002)]
31. N.V. Mikheev, E.N. Narynskaya, Mod. Phys. Lett. A 15, 1551 (2000)
32. N.V. Mikheev, E.N. Narynskaya, Cent. Eur. J. Phys. 1, 145 (2003)
33. I.S. Shklovskii, Astron. Zh. 46, 715 (1969) [Sov. Astron. 13, 562 (1969)]
34. J.E. Gunn, J.P. Ostriker, Astrophys. J. 160, 979 (1970)
35. A.G. Lyne, D.R. Lorimer, Nature 369, 127 (1994)
36. G. Hobbs, D.R. Lorimer, A.G. Lyne, M. Kramer, Mon. Not. R. Astron. Soc. 360, 974 (2005)
37. A.A. Deshpande, R. Ramachandran, V. Radhakrishnan, Astron. Astrophys. 351, 195 (1999)
38. S. Johnston, G. Hobbs, S. Vigeland et al., Mon. Not. R. Astron. Soc. 364, 1397 (2005)
39. C.L. Fryer, Astrophys. J. 601, L175 (2004)
40. L. Scheck, K. Kifonidis, H.-Th. Janka, E. Müller, Astron. Astrophys. 457, 963 (2006)
228 6 Neutrino-Electron Interactions in External Active Media

41. J.R. Gott, J.E. Gunn, J.P. Ostriker, Astrophys. J. 160, L91 (1970)
42. E.R. Harrison, E. Tademaru, Astrophys. J. 201, 447 (1975)
43. N.N. Chugai, Pis’ma Astron. Zh. 10, 210 (1984) [Sov. Astron. Lett. 10, 87 (1984)]
44. Yu.M. Loskutov, Pis’ma v Zh. Eksp. Teor. Fiz. 39, 438 (1984) [JETP Lett. 39, 31 (1984)]
45. Yu.M. Loskutov, Teor. Mat. Fiz. 65, 141 (1985)
46. O.F. Dorofeev, V.N. Rodionov, I.M. Ternov, Pis’ma v Zh. Eksp. Teor. Fiz. 40, 159 (1984) [JETP
Lett. 40, 917 (1984)]
47. O.F. Dorofeev, V.N. Rodionov, I.M. Ternov, Pis’ma Astron. Zh. 11, 302 (1985) [Sov. Astron.
Lett. 11, 123 (1985)]
48. Yu.P. Pskovsky, O.F. Dorofeev, Nature 340, 701 (1989)
49. A. Vilenkin, Astrophys. J. 451, 700 (1995)
50. D. Lai, Y.-Z. Qian, Astrophys. J. 505, 844 (1998)
51. P. Arras, D. Lai, Astrophys. J. 519, 745 (1999)
52. P. Arras, D. Lai, Phys. Rev. D 60, 043001 (1999)
53. G.S. Bisnovatyi-Kogan, S.G. Moiseenko, Astron. Zh. 69, 563 (1992) [Sov. Astron. 36, 285
(1992)]
54. G.S. Bisnovatyi-Kogan, Astron. Astrophys. Trans. 3, 287 (1993)
55. G.S. Bisnovatyi-Kogan, AIP Conf. Proc. 366, 38 (1996)
56. A. Kusenko, G. Segrè, Phys. Rev. Lett. 77, 4872 (1996)
57. H.-Th. Janka, G.G. Raffelt, Phys. Rev. D 59, 023005 (1998)
58. E.Kh. Akhmedov, A. Lanza, D.W. Sciama, Phys. Rev. D 56, 6117 (1997)
59. A. Kusenko, G. Segrè, Phys. Lett. B 396, 197 (1997)
60. A. Kusenko, Phys. Rep. 481, 1 (2009)
61. G.M. Fuller, A. Kusenko, I. Mocioiu, S. Pascoli, Phys. Rev. D 68, 103002 (2003)
62. C. Kishimoto, Pulsar kicks from active-sterile neutrino transformation in supernovae,
arXiv:1101.1304 (2011)
63. S.G. Moiseenko, G.S. Bisnovatyi-Kogan, N.V. Ardeljan, in: Supernovae: One Millennium After
SN1006, 26th Meeting of IAU, Prague, Czech Rep., JD09, No. 33 (2006); astro-ph/0603789
64. K.A. Postnov, Usp. Fiz. Nauk 169, 545 (1999) [Phys. Usp. 42, 469 (1999)]
Chapter 7
Neutrino-Photon Interactions
in External Active Media

A strong magnetic field influences essentially on the properties of particles. Firstly,

it can induce new interactions between particles—for example, an interaction arising
at the one-loop level between electrically neutral neutrinos and photons. Secondly,
magnetic fields dramatically change particle kinematics, opening new channels that
are forbidden in a vacuum by conservation laws. Among these processes, the radiative
transition of a massless neutrino ν → νγ, which is also called the neutrino Cherenkov
process, has been of great interest for a long time. In this chapter, we analyse this
process in external active media: in a magnetic field, and in magnetized plasma.
We also consider other neutrino-photon processes, when neutrino interacts with two
photons (Compton-like process) and with three photons, under an influence of a
magnetic field.

7.1 ννγ Interaction in External Active Media

7.1.1 The Effective Lagrangian of the ννγ Interaction

In this section, we present a calculation of the amplitude of the neutrino–photon

process due to the ννγ interaction induced by a magnetic field, for a case when the
particles involved are, in general, off mass-shell. In other words it means that the
effective Lagrangian for the ννγ interaction in a momentum space will be obtained.
The calculation is performed within the Standard Model with a possible mixing in
the lepton sector. The result is applicable for a magnetic field of any strength when
the local limit of the weak interaction is valid.
The effective local Lagrangian of the neutrino–electron interaction (4.66) with a
possible lepton mixing taken into account can be rewritten to the form

GF   
Lνe = − √ ēγα (C V − C A γ5 )e ν̄ j γ α (1 − γ5 )νi , (7.1)
2

A. Kuznetsov and N. Mikheev, Electroweak Processes in External Active Media, 229

Springer Tracts in Modern Physics 252, DOI: 10.1007/978-3-642-36226-2_7,
© Springer-Verlag Berlin Heidelberg 2013
230 7 Neutrino-Photon Interactions in External Active Media

Fig. 7.1 The Feynman

diagram describing the vertex
ννγ in the local limit of the
weak interaction

where C V , C A are the vector and axial-vector electroweak constants:

1
C V = Uie U ∗je − δi j (1 − 4 sin2 θW ),
2
1
C A = Uie U ∗je − δi j .
2
Here, the subscripts i and j label neutrino mass eigenstates, and the matrix elements
Uie describe the mixing in the lepton sector. The Feynman diagram describing the
vertex ννγ is presented in Fig. 7.1.
It should be recalled that a subtraction procedure is required in calculating the
effective Lagrangian of ννγ interaction induced by an external magnetic field. This
is because the use of the local limit of weak interaction causes two problems: the
amplitude acquires both the ultraviolet divergence and the triangle axial anomaly.
It can be readily seen by the expansion of the amplitude of the process ν → νγ in
terms of the external magnetic field, as is shown in Fig. 7.2.
The zero term in this expansion,

L(0) = L(B = 0),

involves an ultraviolet divergence, while the term linear in the field,


(1) dL 
L =B ,
d B  B=0

Fig. 7.2 The expansion of the amplitude of the process ν → νγ in terms of the external magnetic
field. The double line corresponds to the exact propagator of an electron in a magnetic field; the
dashed lines correspond to the external field
7.1 ννγ Interaction in External Active Media 231

involves the well-known Adler anomaly, because of the presence of the axial-vector
interaction in the effective weak Lagrangian. Strictly speaking, both these terms
cannot be properly calculated in the local limit, and the correct expression for the
effective Lagrangian ΔLννγ induced by an external field can be defined as follows
 
ΔLννγ = L − L(0) − L(1) + L̃(1) , (7.2)

where the correct term L̃(1) linear in the field should be calculated in the electroweak
theory without going to the local limit, and with taking into account the contri-
bution from all virtual charged fermions. The expression for L̃(1) can be deduced,
for example, from the amplitude for the Compton-like process ν( p1 ) + γ ∗ (q1 ) →
ν( p2 ) + γ ∗ (q2 ) [1, 2] (in general, the photons γ ∗ (q1 ) and γ ∗ (q2 ) are off the mass
shell, and the amplitude has the meaning of an effective Lagrangian in the momentum
space) by replacing the field-strength tensor for one of the photons by the strength
tensor for a constant uniform magnetic field; that is,

q1α → 0, f 1αβ → iFαβ , q2α → qα , f 2αβ → f αβ ,

where f αβ = qα εβ −qβ εα is the Fourier transform of the photon field-strength tensor,

while Fαβ is the strength tensor for an external field. Upon some transformations,
the expression for L̃(1) can be recast into the form

(1) e GF eB ( f ϕ̃)(qϕϕ j) (qϕϕ f q)(q ϕ̃ j)
L̃ = √ CA 2 −
2 4π 2q2 q2 q 2
2
( f ϕ̃)( jq) q + q
2
+ I(q 2 ), (7.3)
2q2 q2

where j α = ν̄ j γ α (1 − γ5 )νi is the neutrino current,

1 ∞  
q2 1 − u2
I(q ) = i
2
du (1 − u )
2
dt exp −it m 2e − q 2 −i .
4 4
0 0

The effective Lagrangian L associated with the diagram in Fig. 7.1 is calculated
on the basis of conventional Feynman rules, with using the electron propagator in an
external constant magnetic field (3.1). We have

eG F
L = − i √ jα ε∗β (q) d4 Z Tr S(−Z )γ β S(Z ) γ α (C V − C A γ5 ) e−iq Z .
2
(7.4)
Thus the field-induced part of this Lagrangian can be constructed as the sum of the
vector–vector and the vector–axial-vector amplitudes (4.24), ΔMV V and ΔMV A ,
with the following substitutions of the currents
232 7 Neutrino-Photon Interactions in External Active Media

GF GF
jV β → eε∗β (q), jV α → √ C V jα , j Aα

→ √ C A jα ,
2 2

and with the further subtraction and restoration of the term linear in the field, as is
described above, see (7.2).
The resulting expression for the field-induced effective Lagrangian of the ννγ
interaction takes the form
 
e GF ( f ϕ) (qϕ j) (1) ( f ϕ̃) (q ϕ̃ j) (2)
ΔLννγ = − √ CV 2
YV V + YV V
8π 2 2 q⊥ q2
  
(qϕϕ f q) (qϕϕ j) ( jq) (3)
+2 − 2 YV V
q2 2
q⊥ q

( f ϕ̃) (qϕϕ j)  (1) 
+ C A eB 2
YV A − 1
q
 
(qϕϕ f q) (q ϕ̃ j) (2) q⊥2
+2 2 q2
YV A + 2
q⊥ 
q
( f ϕ̃)( jq)   
(3)
+ YV A − 1 + 2 I(q ) 2
, (7.5)
q2

where the functions YV(i)V and YV(i)A are defined in (4.31) and (4.33).
The effective Lagrangian (7.5) obtained is manifestly gauge invariant, and is valid
for photon and neutrino off-shell. Consequently, it can be used in an analysis of the
neutrino electroweak processes in a magnetic field, as the external-field-induced
vertex of the ννγ interaction.
However, the kinematics of the processes with photons in a strong magnetic field
essentially depends on the photon dispersion properties which were analyzed in
Sect. 4.2. A big difference of the 2nd mode photon dispersion properties below and
above the threshold q2 = 4m 2e , which is seen in Fig. 4.2, leads to different neutrino
processes being possible in the regions of the plot (q⊥ 2 , q 2 ), as is shown in Fig. 7.3.

A small region depicted by the rectangle where the photon dispersion slightly deviates
from the vacuum one, corresponds to the radiative decay of the massive neutrino
νi → ν j γ.

7.1.2 Photon Production by the Massless Neutrino ν → νγ

The process ν → νγ in a magnetic field was investigated in the cases of a relatively

weak field [3], a strong field [4], and an arbitrary field [5]. In these papers, only
the region of relatively small neutrino energies, E < 2m e , was considered. For
the case of larger neutrino energies, E  2m e , which is interesting in the light of
possible astrophysical applications, large radiative corrections become significant,
7.1 ννγ Interaction in External Active Media 233

Fig. 7.3 The set of the neutrino processes being kinematically allowed, depending on the 2nd mode
photon dispersion properties in a strong magnetic field

which are reduced to the photon wave function renormalization (4.10). One more
essential factor is the significant deviation of the 2nd mode photon dispersion from
the vacuum one; see Fig. 4.2. Both these factors were taken into account in [6].
The general expression for the effective ννγ vertex is represented in (7.5). We note
that the vertex is enhanced substantially in the vicinity of the cyclotron resonances
(4.12) as it took place for the photon dispersion operator in a field. The amplitude of
the transition ν( p) → ν  ( p  )+γ(q) is simplified essentially in a case of high neutrino
energies, E m e , and in the strong field limit where the strength is the maximum
physical parameter, eB > E 2 . The field-induced amplitudes of the processes of ννγ
interactions where real photons participate with the polarization vectors defined in
(4.10), take the form
√
(1) eG F Z1 (1)
M(ννγ ) = − √  C V (qϕ j)YV V ,
2
4π 2 q 2
⊥
√  
eG F Z2 (2) (1)
M(ννγ (2) ) = − √  C V (q ϕ̃ j)YV V + C A eB(qϕϕ j) YV A − 1 ,
4π 2 2 q 2

(7.6)

where Z1 , Z2 are the renormalization factors defined in (4.11), and jα is the neutrino
current. The amplitudes (7.6) describe both the photon emission in the neutrino
234 7 Neutrino-Photon Interactions in External Active Media

process ν → ν  γ (it can be either the radiative decay of massive neutrino or the
radiative transition of massless neutrino), and the photon decay into a neutrino pair
γ → ν ν̄.
As was mentioned above, the dispersion of the 1st mode photon slightly deviates
from the vacuum law even in a strong field. It means that the collinear kinematics is
realized in the process ν → νγ (1) :

jα ∼ qα ∼ pα ∼ pα . (7.7)

Consequently, the amplitude M(ννγ (1) ) has an additional suppression because

(qϕ j) (qϕq) = 0. On the other hand, the kinematics is far from collinearity in
the transition where the 2nd mode photon participates, especially near the cyclotron
resonance where q2 tends to 4m 2e from below.
We note, that the amplitude M(ννγ (2) ) would have the square root singularity in
the point q2 = 4m 2e without taking the renormalization of the photon wave function
√
into account. With the renormalization accounted (the factor Z2 ) the amplitude
becomes finite:
eG F eB  
M(ννγ (2) ) −  C V (q ϕ̃ j) + C A (qϕϕ j) . (7.8)
4π 2
q⊥

The calculation of the process probability is performed in the conventional way for a
two-particle decay. In the integration over the phase volume of the final photon, one
should keep in mind its dispersion law: ω |q3 |.
The result for the probability of the process ν → νγ (2) is rather simple in the
case eB E 2 sin2 θ,

αG 2F 2
W (ν → νγ (2) ) (C + C 2A )e2 B 2 E sin2 θ, (7.9)
8π 2 V
where E is the energy of the initial neutrino, and θ is the angle between the momentum
of the initial neutrino and the magnetic field direction.
The probability of the process ν → νγ (2) is also nonzero in the region of Fig. 4.2
which is above the threshold of the e− e+ pair creation, q2 > 4m 2e . This is due
to an existence of the imaginary part of the polarization operator which causes an
uncertainty of the photon dispersion in a magnetic field in this kinematic region.
However, the tree-level channel ν → νe− e+ considered earlier dominates in this
region.
For the four-vector Q α (6.59) of the neutrino energy and momentum loss in the
considered strong field limit, eB E 2 sin2 θ, one obtains for the process ν → νγ:
 
1 2C V C A
I = EW 1+ 2 cos θ , (7.10)
4 C V + C 2A
7.1 ννγ Interaction in External Active Media 235
 
1 2C V C A 1
Fz = E W cos θ + 2 , F⊥ = E W sin θ, (7.11)
4 C V + C 2A 2

where the probability W should be taken from (7.9).

The asymmetry (6.69) due to the process ν → νγ differs from its value (6.70)
obtained for the process ν → νe− e+ by the factor ∼ α(eB/E 2 ), or more exactly:

eB (ee)
A(γ) ∼ 2πα A , (7.12)
E2

where A(ee) is the value defined in (6.70). It is seen that the contributions of the
processes ν → νγ and ν → νe− e+ into the asymmetry could be comparable in the
strong magnetic field despite the suppressing factor α in (7.12).

7.1.3 Photon Decay into the Neutrino Pair γ → ν ν̄

The process γ → ν ν̄ is kinematically allowed (q 2 > 0) in a magnetic field owing

to specific features of photon dispersion. This is so in the region q2 > 4m 2e for the
(2) 
photon polarization εα and in the region q2 > (m e + m 2e + 2eB)2 for the photon
(1)
polarization εα .
An analysis reveals that, in the considered region (q 2 > 0), the photon “mass”
induced by a magnetic field is much less than the photon energy ω: q 2 ω2 .
This implies that the photon decay γ → ν ν̄ occurs under the condition of collinear
kinematics (7.7), so that the neutrino current can be represented as

jα 4 x(1 − x) qα , (7.13)

where x = E/ω and 1−x = E  /ω are the energy fractions carried by the antineutrino
and the neutrino, respectively.
From this and from (7.6), it follows that, in the collinear limit, the amplitude for
the decay of the 1st mode photon vanishes and that the expression for the amplitude
describing the decay of the 2nd mode photon becomes considerably simpler; that is,

M(ννγ (1) ) 0,
2 e GF C A 
M(ννγ (2) ) √ x(1 − x) [e2 (q F Fq)]1/2 J (q2 ), (7.14)
2π 2

where we took into account that Z2 1. The dimensionless field form factor J (q2 )
has the form
236 7 Neutrino-Photon Interactions in External Active Media

1  (1)

J (q2 ) = 1 − YV A
2
1 ∞  
1 − u2
1−i m 2e du dt exp − i t m 2e − q2 (7.15)
4
0 0

q2
cos βut − cos βt
+ .
2β sin βt

The process under consideration involves three particles, but its amplitude is not a
constant, in contrast to one that occurs in a vacuum. The reason is that the amplitude
now depends not only on the 4-momenta of the particles involved but also on the
strength tensor of the external field. Therefore, the probability of this process is not
merely the product of the amplitude squared and the phase-space volume, but is given
by
x2
 2
(2) 1  
W (γ → ν j ν̄i ) = dx M(ννγ (2) ) . (7.16)
16πω
x1

The limits of the integration in (7.16), x1 and x2 , are defined by the ratios of the
neutrino masses to the photon “mass”, μi2 = m i2 /q 2 , i = 1, 2, and can be represented
as
1
x1,2 = (ε ± p), ε = 1 + μi2 − μ2j ,
2

p = [1 − (μi + μ j )2 ][1 − (μi − μ j )2 ].

Here, ε and p are,

 respectively, the energy and the momentum of the ith antineutrino
in the ratio to q 2 /2 defined in the reference frame comoving with the decaying
photon. Substituting expression (7.14) for the amplitude M(ννγ (2) ) into (7.16), we
arrive at
 
α G 2F C 2A 2
W (γ (2) → ν j ν̄i ) = e (q F Fq) |J (q  )|
2 2
1 + μ2
i + μ2
j − 2(μ2
i − μ2 2
j )
12π 4 ω

× [1 − (μi + μ j )2 ][1 − (μi − μ j )2 ]. (7.17)

The integral J depends on the variable q2 . The physical meaning of q2 is seen from
the relation
q2 q⊥ 2
ω 2 sin2 θ, (7.18)

where θ is the angle between the momentum q of the decaying photon and the
direction of the magnetic field B.
The expression (7.17) for the probability describes only one channel, the decay of a
photon into a neutrino of the type j and an antineutrino of the type i, but only the total
7.1 ννγ Interaction in External Active Media 237

decay probability representing the sum over all allowed modes (μi + μ j < 1) is the
quantity of physical interest. Assuming a hierarchy in the neutrino mass spectrum—
that is, m i2 q 2 for i  N L and m i2 > q 2 for i > N L (thus, N L is the number of
the “light” neutrino species)—we obtain the total probability of the photon decay in
the form


NL
α G 2F C 2A 2
W = W (γ (2) → ν j ν̄i ) = e (q F Fq) |J (q2 )|2 , (7.19)
12π 4 ω
i, j=1

where


NL
1
C 2A = C 2A = N L − U 2 (1 − U 2 ),
4
i, j=1


NL
U2 = |Uie |2  1.
i=1

If all three neutrino species are “light”, m i2 q 2 (N L = 3, U 2 = 1), we have

C 2A = 3/4, and the probability of the decay γ (2) → ν ν̄ is independent of the
parameters of mixing in the lepton sector.
The function J (q2 ) is simplified in the two limiting cases.

(i) If the magnetic field is the largest parameter in the problem (eB q2 ), we
obtain
1 − v2 1+v
J (q2 ) ln − iπ + 1, (7.20)
2v 1−v

where v = 1 − 4m 2e /q2 .
(ii) In the opposite case of eB q2 , we arrive at

J (q2 ) 1. (7.21)

At first glance, it may seem that, in view of relation (7.20), the decay probability
(7.19) in a strong field has a pole singularity at q2 → 4m 2e . However, a more accurate
solution of the dispersion equation for a photon in this limit shows that

|q2 − 4m 2e |min = ω Γγ→e− e+ . (7.22)

An apparent singularity like this, but of the square-root type, is known [7] to be
encountered in dealing with the photon decay into an electron–positron pair in a
magnetic field, γ → e+ e− . By taking into account the dispersion of the photon in
238 7 Neutrino-Photon Interactions in External Active Media

the process γ → e+ e− , it was shown in [8] that the decay width is everywhere finite
and that, at q2 4m 2e , it reaches a maximum value of
√
  3 2 α eB 2/3
m 2e
Γγ→e− e+ max = . (7.23)
2 m 2e ω

By virtue of relations (7.22) and (7.23), the probability of the decay process
γ → ν ν̄ is also finite, and its maximum is
1/3
1 2 α eB eB
Wmax (γ (2) → ν ν̄) = √ (G F m 2e )2 C 2A . (7.24)
3 3π 2 m 2e ω

The probability (7.24) of the electroweak process γ → ν ν̄ is much less than

the probability (7.23) of the process γ → e+ e− by the factor (G F m 2e )2 ∼ 10−23 .
However, the former may play the role of an additional source of neutrino cooling
in astrophysics.
Let us estimate the energy carried away by neutrinos from a unit volume of the
photon gas per unit time. This quantity, referred to as neutrino emissivity, is given
by
d3 k 1
Q = dNγ ω W = ω/T
ω W. (7.25)
(2π) e3 −1

Here, we considered that only the 2nd mode photons in (4.10) contribute to the
emissivity. In our estimate, we assume that all neutrino species are light: m i2 q 2,
C 2A = 3/4.
Substituting the probability given by (7.19) into (7.25), we can recast the expres-
sion for the emissivity into the form

2
α (G F eB)2 5 erg B
Q= m e F(T ) 0.96 × 1018 F(T ), (7.26)
8π 4 s cm3 Be

where
1 ∞
8 x 4 dx
F(T ) = 2 du (1 − u )2
|J (q2 )|2 . (7.27)
π e x/τ − 1
0 x0

Here, τ = T /2m e , the variables of integration are given by u = cos θ, x = ω/2m e ,

and the argument of the function J is q2 = 4m 2e x 2 (1−u 2 ), x0 is defined from (7.22).
As the analysis shows, the function F(T ) slightly depends on the field strength
and has the following form in a wide temperatire region, with the only restriction
T  me:
4ζ(5) T 5
F(T ) , (7.28)
π2 me
7.1 ννγ Interaction in External Active Media 239

where ζ(5) 1.037 (ζ is the Riemann zeta function).

At low temperatures, T 2m e , the function F(T ) is exponentially small,
F(T ) ∼ exp(−2m e /T ).
Finally, we estimate the contribution of the field-induced photon-decay process
γ → ν ν̄ to the neutrino emissivity under the conditions of a supernova explosion.
We assume that, in the central region of the explosion of a size of a few hundred
kilometers, a strong magnetic field of toroidal type is generated [9, 10]. We then
have
2 5 3
dE erg B T R
∼ 1045 . (7.29)
dt s 1015 G 2 MeV 100 km

Recall that the estimated value for the total neutrino emissivity of a supernova is about
1052 erg/s. We note that the contribution of the process γ → ν ν̄ is independent of
the neutrino flavors. It can be significant in the low-energy region of the neutrino
spectrum.

7.1.4 Radiative Neutrino Transition ν → νγ in Strongly

Magnetized Plasma

The process of the radiative massless neutrino transition ν → νγ (neutrino

Cherenkov process) is forbidden in vacuum, and it becomes allowed in the presence
of plasma and/or magnetic field. There exist several papers where this transition was
studied in plasma or magnetic field separately. In plasma the process ν → νγ was
firstly investigated in Ref. [11] and later in Refs. [12, 13]. In a pure magnetic field
the radiative neutrino transition ν → νγ was studied in the papers [3–6]. In the
framework of four-fermion theory the amplitude and the probability of the process
were calculated in Refs. [3] and [4] in the crossed field and strong magnetic field
respectively. In the Standard Model the amplitude of the neutrino transition ν → νγ
was found in [5, 6] for the arbitrary magnetic field strength. In the paper [5] the case
of the moderate neutrino energies, E < 2m e was studied in the kinematical region
where the final photon dispersion law was closed to the vacuum one, q 2 = 0. The
limit of the large neutrino energies and strong magnetic field was investigated in
Ref. [6]. There is that case which could be realized at the Kelvin-Helmholz stage of
supernova remnant cooling, when the energies of the neutrino are E 10−20 MeV
and the magnetic field strength could be as high as 1016 −1017 G [14, 15]. It was
shown in [6] that the main contribution into the probability of the neutrino transition
ν → νγ was determined from the vicinity of the lowest cyclotron resonance, when
the amplitude of the process and photon polarization operator contained simultane-
ously the square-root singularity.
The purpose of this section is to study the influence of the electron-positron plasma
on the process of the radiative massless neutrino transition ν → νγ in a strong
magnetic field. The presentation is based on Ref. [16]. This process is considered
240 7 Neutrino-Photon Interactions in External Active Media

in the framework of the Standard Model using the effective local Lagrangian of
the neutrino-electron interaction (4.66). We investigate the limit of ultrarelativistic
strongly magnetized plasma, when the magnetic field strength is the largest physical
parameter
eB > E 2 , μ2e , T 2 m 2e . (7.30)

Here, E is the initial neutrino energy, μe is the electron chemical potential, T is the
temperature of plasma. Under these conditions electrons and positrons in plasma
occupy dominantly the lowest Landau level.
Notice that the amplitude and the probability of the process ν → νγ depend
essentially on the polarization of the final photon. In a general case there exist three
eigenmodes of the photon polarization operator. The corresponding eigenvectors can
be written in the following form:

(qϕ)μ (q ϕ̃)μ q 2 (qϕϕ)μ − qμ (qϕϕq)

ε(1)
μ =  ; ε(2)
μ =  ; ε(3)
μ =  . (7.31)
2
q⊥ q2 q 2 q2 q⊥
2

Only two of these modes, ε(1) (2)

μ and εμ are the physical one in the pure magnetic field.
As the analysis shows, the presence of the strongly magnetized plasma doesn’t modify
the eigenvectors (7.31) but modifies the eigenvalue corresponding to the vector ε(2) μ
only. This is due to the fact that the interaction of the two other eigenmodes with the
electrons and positrons which occupy the lowest Landau level is strongly suppressed
(2)
under the condition (7.30). Hence, only the photon with eigenvector εμ can be
created in the process under consideration, as it takes place in the pure magnetic
field, see Sect. 7.1.2.
The process of the radiative neutrino transition is depicted by the Feynman dia-
gram, see Fig. 7.1, where the double line corresponds to the propagator of an electron
in the presence of a magnetic field and plasma. Several methods are known in liter-
ature describing the process in the background plasma. Here, we use the real-time
formalism. The general expression of the real-time propagator in an external field
can be found in the paper [17]. In the limit of a strong magnetic field the electron
propagator in plasma can be presented in the form:

d4 p
S(x, y) = eiΦ(x,y) S( p) e−i p(x−y) , (7.32)
(2π)4

where

1
2((γ p) + m e )Π− e− p⊥ /eB
2
S( p) − 2iπ f F ( p0 ) δ( p2 − m 2e ) ,
− m 2e + i
p2
(7.33)
1
f F ( p0 ) = f − ( p0 )Θ( p0 ) + f + (− p0 )Θ(− p0 ), Π− = (1 − iγ1 γ2 ) .
2
7.1 ννγ Interaction in External Active Media 241

Here, f ∓ ( p0 ) are the distribution functions of electrons and positrons in plasma

1
f ∓ ( p0 ) = .
e( p0 ∓μe )/T + 1

As it was noticed, in the case of two-point function the noninvariant phase factors
Φ(x, y) were cancelled: Φ(x, y) + Φ(y, x) = 0. With using the propagator (7.33),
the amplitude of the process can be presented in the form:

M = M B + M pl , (7.34)

where M B is the amplitude of the process ν → νγ corresponding to the pure

magnetic field contribution (T = μe = 0). Following Ref. [6], it can be expressed
in the form
 2 
eG F eB q
MB = √  {C V ( j ϕ̃q) + C A ( jq) } H , (7.35)
2
2π 2 q 2 4m 2e


where the function H (z) is defined in Eq. (4.18). It should be noted that M B is the
(2)
amplitude with the definite photon polarization corresponding to the mode εμ from
Eq. (7.31).
The second term in Eq. (7.34), M pl , is induced by the coherent neutrino scattering
on plasma electrons and positrons with photon radiation. For M pl we find

eG F 
M pl = − √ eB m 2e q2 {C V ( j ϕ̃q) + C A ( jq) }
π2 2
d pz f − (ε) + f + (ε)
× . (7.36)
ε 4( pq)2 − (q2 )2

As was mentioned above, the amplitude M B contains the square-root singularity

which is connected with the cyclotron resonance on the lowest Landau level. In the
vicinity of the resonance point q2 = 4m 2e it becomes:

eG F eB
MB √  {C V ( j ϕ̃q) + C A ( jq) }. (7.37)
4π 2 4m 2 − q 2
e 

It is particularly remarkable that the amplitude M pl contains the singularity of the

same type. In the limit q2 → 4m 2e the total amplitude (7.34) can be presented in the
following form:
|q0 |
M MB F , (7.38)
2T
242 7 Neutrino-Photon Interactions in External Active Media

where
sinh x μe
F(x) = , η= .
cosh x + cosh η T

It should be stressed that not only the amplitude M has the singular behaviour
but the photon polarization Π (2) as well. It can be obtained from Eq. (7.38) by the
following replacements

GF
Π (2) = −M √ C V → e, C A → 0, jα → ε(2)
α .
2

For Π (2) one has:

2αeBm e |q0 |
Π (2) − F . (7.39)
4m 2e − q2 2T

A large value of Π (2) near the resonance requires taking account of large radiative
corrections which reduce to a renormalization of the photon wave function:

 ∂Π (2)
ε(2) (2)
α → εα Z2 , Z2−1 = 1 − . (7.40)
∂q2

Using the formula (7.40) for the amplitude we find

 eG F eB |q0 |
M→ Z2 M  {C V ( j ϕ̃q) + C A ( jq) }F . (7.41)
4π 2
q⊥ 2T

Thus, the photon wave-function renormalization corrects the singular behaviour of

the amplitude.
The probability of the process ν → νγ can be obtained by integration of
the amplitude over the phase space with taking account of the photon dispersion
2 = Π (2) .
q2 − q⊥

  2 d3 P 
1   1
EW = M Z2  δ(E − E  − q0 ) , (7.42)
32π 2 1 − e−q0 /T E  q0

where the non-trivial photon dispersion law q0 = q0 (q) should be taken into account.
We assume that the neutrino distribution is closed to the Boltzmann one, so one can
neglect the deviation of the neutrino statistical factor from the unity. The probability
(7.42) is rather complicated in the general case. Here we present the results of our
calculation in two limiting cases of the cold plasma, μe T , and hot plasma,
T μe . Notice that in the vicinity of the cyclotron resonance, which gives the
main contribution
 to the probability, the photon dispersion has a rather simple form
q0 q32 + 4m 2e .
7.1 ννγ Interaction in External Active Media 243

In the limit of the low temperature, μe T , for the probability we obtain:

  
α(G F eB)2 E 4μe 4μe
WL T 2
(C V − C A ) 1 − u −
2 2
(1 + u) Θ 1 − u −
16π E E
  
4μ e 4μ e
+ (C V + C A )2 1 − u 2 − (1 − u) Θ 1 + u − . (7.43)
E E

Here, u = cos θ, θ is the angle between the initial neutrino momentum P and the
magnetic field direction.
In the opposite limit of high temperature, T μe , the result for the probability
of the process ν → νγ is:

α(G F eB)2 T E(1 − u)
WH T 2
(C V − C A )2 (1 + u) F1
4π 8T

E(1 + u)
+ (C V + C A )2 (1 − u) F1 , (7.44)
8T
1
F1 (x) = x + ln(cosh x) − tanh2 x − tanh x .
4
In the limit of the rarefied plasma both expressions (7.43) and (7.44) provide:

α(G F eB)2 2
WB (C V + C 2A ) E(1 − u 2 ) . (7.45)
8π 2
This result reproduces the formula (7.9) for the radiative neutrino transition proba-
bility in the pure strong magnetic field.
Keeping in mind possible applications of our results in astrophysics we calculate
the mean values of the neutrino energy and momentum losses. These values were
defined earlier by the four-vector Q α , see Eq. (6.59):
For the zero and third components of Q α we obtain the following expression in
the limit of cold plasma, T μe :

α(G F eB)2 3
Q 0,3 E (1 − u 2 )
64π 2
  
16μ2e 4μe
× (C V + C A ) 1 + u −
2
Θ 1+u−
E (1 + u)
2 E
 2  
16μ e 4μ e
± (C V − C A )2 1 − u − 2 Θ 1−u− . (7.46)
E (1 − u) E

In the opposite case, when μe T we find


α(G F eB)2 E(1 + u)
Q 0,3 E T 2
(C V + C A )2 (1 − u) F2
2π 2 4T
244 7 Neutrino-Photon Interactions in External Active Media

E(1 − u)
± (C V + C A )2 (1 + u) F2 , (7.47)
4T
1 x xe x (1 + 2e x )
F2 (x) = tanh − + (2 + x) ln(1 + e x )
2 2 (1 + e x )2
π2
+ Li2 (−e x ) − ln 4 + ,
12
where Li2 (x) is the polylogarythm function. Notice that in the limit T → 0, μe → 0,
both expressions (7.46) and (7.47) reproduce the formula for the four-vector of losses
via the process ν → νγ in the pure strong magnetic field, see Eqs. (7.10) and (7.11):

α(G F eB)2 3
Q 0,3 = 2
E (1 − u 2 )
 64π 
× (C V + C A )2 (1 + u) ± (C V − C A )2 (1 − u) . (7.48)

We note that electron-positron plasma and photon gas make an opposite influence
on the process under consideration. On one hand, the electron-positron background
decreases the amplitude of the process (F(q0 ) < 1). On the other hand, the prob-
ability and the mean value of the neutrino energy and momentum loss increases
due to the effect of the stimulated photon emission. The numerical analysis, for de-
tails see Ref. [16], shows that the combined effect of electron-positron plasma and
photon gas leads to the decreasing of the probability in comparison to the result in
the strong magnetic field, see Eq. (7.45). The similar supressing plasma influence
on four-vector of neutrino energy and momentum losses takes place. Therefore the
complex medium plasma + strong magnetic field is more transparent to neutrino
with regard to the process ν → νγ, than the pure magnetic field.

7.2 Compton-Like Interaction of Neutrinos with Photons

7.2.1 The Amplitude of the Process γγ → ν ν̄ in Vacuum

7.2.1.1 Standard Weak Interaction

Historically the reaction γγ → ν ν̄ was one of the first photon–neutrino processes

considered in the context of its astrophysical application. In 1959, Pontecorvo sug-
gested that (eν)(eν) coupling could induce reactions leading to energy loss in
stars [18]. One of these processes, γγ → ν ν̄, caused by this coupling was compared
in [19] with other neutrino reactions and a rough estimation of the neutrino energy
loss rate was obtained. In both papers the authors used the four-fermion (V − A) Fermi
model. The process of conversion of the photon pair into the neutrino-antineutrino
pair is described by the two Feynman diagrams with a virtual fermion in the loop
and with the photon interchange, see Fig. 7.4.
7.2 Compton-Like Interaction of Neutrinos with Photons 245

Fig. 7.4 Feynman diagram

for the process γγ → ν ν̄;
a large circle represents the
effective weak interaction of
a fermion with a neutrino, see
Fig. 4.8

Given the gauge invariance of the electromagnetic interaction, the amplitude of

the process can be written in the most general form:

α GF   αβ μν
M= √ ν̄ ( p1 ) Tαβμν ν (− p2 ) f 1 f 2 , (7.49)
π 2

where the index  defines the neutrino flavors,  = e, μ, τ ,1 f αβ = k α εβ − k β εα is

the tensor of the photon electromagnetic field in the momentum space. The tensor
Tαβμν which is the Dirac matrix, has the dimension of an inverse mass and must be
built from the available covariants.
Apparently, the very first correct conclusion about the amplitude was done in
Ref. [20]. It is the Gell-Mann theorem: in the case of massless neutrinos, real photons,
and in the local limit of the standard weak interaction, the amplitude is exactly zero.
Qualitatively, this can be seen from the following argument. In the center-of-mass
frame, the left-handed neutrino and the right-handed antineutrino carry the total
angular momentum equal to a unit. However, as it was shown by Landau [21] and
Yang [22], the system of two photons can not exist in a state with a unit angular
momentum (Landau–Yang theorem). In terms of the tensor analysis, this means that
with chirality of massless neutrinos and Bose symmetry of photons, there are no
covariants to construct the tensor Tαβμν .
The nonzero amplitude (7.49) arises if any of the Gell-Mann theorem conditions
is broken. It may be non-zero neutrino mass, non-locality of the standard weak
interaction, non-standard neutrino interaction, or off-shell photons. In the case of
massive neutrinos, the process becomes allowed [23, 24] due to the change of a
neutrino chirality, with the amplitude being proportional to a neutrino mass. To
illustrate the Lorentz structure, we present here an expression for the tensor Tαβμν
in the case of low-energy photons (ω m e ), where the electron loop gives the
maximal contribution to the amplitude:

i 1 m ν
Tαβμν = δe − γ5 εαβμν . (7.50)
12 2 m 2e

When the non-locality of the weak interaction via the W boson is taken into account,
the momenta of a neutrino and an antineutrino can enter the amplitude not just as

1 The expression (7.49) can be easily generalized to take into account the lepton mixing.
246 7 Neutrino-Photon Interactions in External Active Media

a sum but separately, providing the following structure [25–27]:

16i mW 3 1
Tαβμν = ln +
3 me 8 m 2W
 
× γα gβμ ( p1 − p2 )ν + γμ gνα ( p1 − p2 )β (1 − γ5 ) . (7.51)

We see that in both cases, the amplitude has a strong suppression either through the
small neutrino mass in the numerator or the large W boson mass in the denominator.

7.2.1.2 Model with a Broken Left-Right Symmetry

Another deviation from the conditions of the Gell-Mann theorem, in which the
process γγ → ν ν̄ is also possible, is realized when the neutrino changes its chirality
in the effective Lagrangian of the lepton-neutrino interaction. When writing the
Lagrangian in the form of the neutral current coupling, the neutrino chirality change
is provided if currents are scalar or pseudoscalar. This case considered in Ref. [28]
takes place in a model with a broken left-right symmetry [29–36] and with the mixing
of vector bosons interacting with the left-handed and right-handed charged weak cur-
rents [33]. In this model, the Lagrangian of the νeW interaction can be represented
as

g  
L= √ [ēγα (1 − γ5 ) νe ] W1α cos ζ + W2α sin ζ
2 2

 
+ [ēγα (1 + γ5 ) νe ] −W1α sin ζ + W2α cos ζ + h.c. , (7.52)

where W1,2 are the charged vector W bosons with a definite mass, and ζ is the
mixing angle. The existing restrictions on the parameters of the model are obtained
in low-energy accelerator experiments, and have the form [37]

MW2 > 715 GeV , ζ < 0.013 . (7.53)

Due to the smallness of the mixing angle, the state W2 almost coincides with the
right-handed boson W R .
There also exists a stronger limit on the model parameters, obtained from as-
trophysical data, namely, from the analysis of neutrino events from the supernova
SN1987A. In combination with accelerator data, the limits were obtained [38]:

MW R > 23 TeV , ζ < 10−5 . (7.54)

For realization of the process γγ → ν ν̄, a part of the effective ννee interaction
Lagrangian is important, providing a non-standard neutrino or antineutrino chirality.
This is possible due to the mixing of the gauge bosons, when the left-handed and
7.2 Compton-Like Interaction of Neutrinos with Photons 247

right-handed currents from Eq. (7.52) are multiplied in the effective Lagrangian.
Given the smallness of the mixing angle and the mass ratio MW1 /MW2 , we can write
the Lagrangian of the ννee interaction in the form

GF
Leff −4 ζ √ [(ēe) (ν̄e νe ) − (ēγ5 e) (ν̄e γ5 νe )] . (7.55)
2

There exist two new channels for the conversion of the photon pair into the neutrino-
antineutrino pair, if compared with the standard model, namely:

γγ → (νe ) L (ν̄e ) L , γγ → (νe ) R (ν̄e ) R . (7.56)

Here, (νe ) R and (ν̄e ) L are the states which are sterile with respect to the standard
weak interaction. The total spin of a neutrino pair in both processes (7.56) in the
center of mass is zero, and the process γγ → ν ν̄ is open.
Representing the amplitude of the process caused by the effective Lagrangian
(7.55) in the form of (7.49), we have the following expression for the tensor Tαβμν :
 
4ζm e 1
Tαβμν = 1 + (1 − 4τ )I (τ ) gαν gβμ .
(k1 k2 ) 2

i
+ I (τ )γ5 εαβμν , (7.57)
4

where
1 1−x
m e2 1
τ= , I (τ ) = dx dy . (7.58)
2(k1 k2 ) τ − xy − i
0 0

Note that our result (7.57), coinciding in terms of the tensor structure with the one,
which can be extracted from Ref. [28], differs from it in numerical coefficients.
The amplitude of the process γγ → ν ν̄ in this model has also the suppression
due to the smallness of the mixing angle ζ.

7.2.1.3 The Case of Virtual Photons

Another case of a non-zero amplitude is realized if one of the photons [39] or both
photons [40] are off-shell. In this case, kμ f μν = 0 and the photon momenta can
participate in the construction of the tensor Tαβμν .
Let us calculate the total amplitude of the process νγ ∗ → νγ ∗ in the standard
model in the case of virtual photons, with non-zero neutrino mass, and with a possible
mixing in the lepton sector [1, 2].
As the analysis shows, the neutrino (V − A) current is factorized in this amplitude
which can be presented in the following general form:
248 7 Neutrino-Photon Interactions in External Active Media

α G F (ν) ∗
M= √ j e1α e2β L αβρ (k1 , k2 ), (7.59)
π 2 ρ

(ν)
where jρ = ν¯j ( p2 )γρ (1 − γ5 )νi ( p1 ), the indices i and j (generally, i = j) label
the neutrino states with definite masses; e1,2 are the 4-vectors of polarization and
k1,2 are the 4-momenta of the photons. As it follows from the above, the tensor L αβρ
could contain only two independent momenta k1 and k2 .
Let us consider in more detail the contribution to the amplitude from the Z boson
exchange. For its obtaining, it is necessary to summarize over all fundamental charged
fermions f , both leptons and quarks, in the loop. The L αβρ tensor takes the form:

 (f)
L αβρ = T3 f Q 2f L αβρ , (7.60)
f

where Q f is the electric charge of a fermion in units of the elementary charge e, T3 f

is the third component of the weak isospin. For the contribution of a single fermion
we obtain the following expression:

1 1−x 
(f) dy
L αβρ = i ελμβρ dx gλα k1μ [ k12 x(1 − 2x) + k22 y(1 − 2y)
af
0 0
− 4(k1 k2 ) x y] + 2 gλα k2μ k12 x (7.61)

+ 4 k1λ k2μ x[k2α y − k1α (1 − x)] + (k1 ↔ −k2 , α ↔ β),

where the notation is used:

a f = m 2f + 2(k1 k2 )x y − k12 x(1 − x) − k22 y(1 − y). (7.62)

In the formula (7.61), the terms are omitted that do 2not depend on the mass of a
fermion, since, due to the known relation f T3 f Q f = 0 (for each generation),
they do not contribute to the amplitude. The expression (7.61) can be rewritten in
such form that the amplitude becomes manifestly gauge invariant. For this, we use
the photon electromagnetic field tensor in the momentum space

f μν = kμ eν − kν eμ , (7.63)

and also the dual tensor

1
f˜μν = εμναβ f αβ . (7.64)
2
Introducing the notation
(f)
Rρ( f ) = e1α e2β
∗
L αβρ , (7.65)
7.2 Compton-Like Interaction of Neutrinos with Photons 249

(f)
we find the expression for the vector Rρ representing the amplitude in an explicitly
gauge invariant form:

 1 1−x
dy
Rρ( f ) = 4i f˜2ρμ f 1μν xdx [k1ν (1 − x) − k2ν y]
af
0 0
1 1−x 
ydy
+ f˜1ρμ f 2μν dx [ k1ν x − k2ν (1 − y)] . (7.66)
af
0 0

In the transition from Eq. (7.61) to (7.66), the following identity was used:

1 1−x
dy 2
dx k1 x(1 − 2x) − k22 y(1 − 2y) ≡ 0. (7.67)
af
0 0

An analysis shows that the contribution to the amplitude of the diagram with a
virtual W boson is also expressed through the vector (7.66), where a charged lepton
only appears as a virtual fermion. The total amplitude of the process νi γ ∗ → ν j γ ∗
can be represented as
⎛ ⎞
α G F (ν) ⎝ 
M = √ jρ Ui U ∗j Rρ() + δi j T3 f Q 2f Rρ( f ) ⎠ , (7.68)
π 2
 f

where Ui is a unitary matrix of the lepton mixing,  = e, μ, τ . The amplitude must
satisfy the requirements of the Gell-Mann theorem [20], but in the expression (7.66)
it is not obvious yet. Using the following relation for the tensors (7.63) and (7.64):

1
f˜1ρμ f 2μσ + f˜2ρμ f 1μσ = f 1μν f˜2νμ gρσ , (7.69)
2
(f)
we write the vector Rρ in the final form:

1
Rρ( f ) = −4 i ( f 1 f˜2 )(k2 − k1 )ρ A(m f , k1 , k2 ) (7.70)
2

− ( f˜2 f 1 k1 )ρ B(m f , k1 , k2 ) + ( f˜1 f 2 k2 )ρ B(m f , k2 , k1 ) ,
250 7 Neutrino-Photon Interactions in External Active Media

where the functions are introduced:

1 1−x
ydy
A(m f , k1 , k2 ) = xdx ,
m 2f + 2(k1 k2 )x y − k12 x(1 − x) − k22 y(1 − y)
0 0
(7.71)
1 1−x
(1 − x − y)dy
B(m f , k1 , k2 ) = xdx .
m 2f + 2(k1 k2 )x y − k12 x(1 − x) − k22 y(1 − y)
0 0

Thus, for the amplitude in the form of (7.68) and (7.70), the Gell-Mann theorem is
satisfied obviously.
The obtained amplitude in special cases coincides with the known results [23, 24,
39, 40]. Thus, the first term in Eq. (7.70) being substituted into Eq. (7.68), gives the
divergence of the neutrino current, i.e. it is proportional to the neutrino mass. For
photons on mass shell at low energies, ω m e , imposing  = f = e and excluding
the lepton mixing, i = j =  , Uk = δk , one reproduces from the amplitude of
Eqs. (7.68), (7.70) and (7.71) the expression for the tensor (7.50). In another case,
2  = 0, the amplitude can be transformed in the case
when both photons are virtual, k1,2
of massless neutrinos to the form which coincides with the result of Ref. [40]. We
emphasize that the authors [40] introduced an artificial dependence of the amplitude
on the neutrino momenta. It is clear, however, that in this approximation (in fact in
the local limit of the weak interaction), the amplitude of the process νγ ∗ → νγ ∗ can
explicitly depend only on the photon momenta.
In this case at low photon energies, ω m e , the tensor Tαβμν introduced in
Eq. (7.49) has the form:

i 1  
Tαβμν = 2
Uie U ∗je − δi j γ ρ (1 − γ5 ) εραμν k1β + ερμαβ k2ν . (7.72)
12m e 2

It should be noted that the total amplitude (7.68), (7.70) allows in particular to
obtain the first terms of the expansion over the external field of the amplitudes of
the radiative neutrino decay νi → ν j γ and of the non-radiative transition νi → ν j
in the electromagnetic field of an arbitrary configuration. It is enough to replace in
Eq. (7.70) the electromagnetic field tensor of the one or both photons to the external
electromagnetic field tensor.
Let us apply the obtained amplitude of the process νγ ∗ → νγ ∗ to calculate
the probability of the massive neutrino radiative decay νi → ν j γ in an external
field [41, 42], in the case of relatively weak field. The field tensor of one of the
photons is replaced to the tensor of the constant uniform magnetic field:

q1α → 0, f 1αβ → iFαβ , q2α → qα , f 2αβ → f αβ . (7.73)

7.2 Compton-Like Interaction of Neutrinos with Photons 251

Taking into account that the main contribution comes from the electron loop, and that
the photon dispersion in a weak field does not differ from the dispersion in vacuum
(q 2 = 0), we obtain the amplitude of the process νi → ν j γ:

eG F C A B
M= √ (ϕ f˜∗ ) ( j (ν) q), (7.74)
48 2π 2 Be

where C A = Uie U ∗je − 21 δi j . The expression (7.74) coincides with the linear in the
field term of the amplitude presented in Eq. (4) of Ref. [41].
Assuming for simplicity the finite neutrino to be massless, we find the probability
of the decay νi → ν j γ in the rest frame of the initial neutrino:

2
α G 2F C 2A 5 B
W = m . (7.75)
18π 192π 3 νi Be

The probability (7.75) agrees with Eq. (5) of Ref. [41], but it is 4 times less than the
probability obtained from Eq. (32) of Ref. [42] in the weak field limit.

7.2.2 Neutrino Scattering in the Coulomb Field of a Nucleus

As one more illustration of the application of the formula (7.68), we consider the
scattering of a high-energy neutrino on a nucleus with the photon radiation. In the
cited papers [23, 24, 39], only astrophysical manifestations of the process νγ →
νγ were studied . Our aim is to explore the possibility to detect this reaction in
the laboratory experiment with high-energy neutrinos from the accelerator. From
the observational point of view, this process would appear as a bremsstrahlung in the
neutrino scattering in the Coulomb field of a nucleus

νi + nucleus → ν j + γ + nucleus. (7.76)

The experimental evidence of the reaction should be the detection of a single hard
photon without any escort.
The reaction (7.76) amplitude can be obtained from Eqs. (7.68) and (7.70) taking
one of the photons (e.g. γ2 ) to be real. In this case one has f 2μν k2ν = 0. We shall
regard m ν = 0 and neglect the lepton mixing. Then the amplitude will be defined
by the second term in Eq. (7.70). Inserting (Z e/k12 )Jμ instead of e1μ , where Jμ and
Z e are the electromagnetic current and the charge of the nucleus, k1μ and e1μ are the
momentum and the polarization vector of the virtual photon, one obtains
252 7 Neutrino-Photon Interactions in External Active Media

Z eα G F
M = 4i √ ερμαβ jρ(ν) Jμ k2α e2β ∗
B(m  , k1 , k2 )
π 2
 
+ T3 f Q 2f B(m f , k1 , k2 ) . (7.77)
f

Here, m  is the mass of the charged lepton which is the partner of the neutrino taking
part in the reaction. Let us examine the case of small transmitted momenta when the
nucleus is still nearly motionless. The momentum modulo |k1 | is restricted then by
the value of km which can be estimated as the inverted nucleus radius 1/r ∼ km
200 × A−1/3 MeV. As the analysis shows, at high energies of the neutrino all the
charged fermions contribute to the amplitude (7.77) except t-quark (we still presume
( pk1 ) m 2W < m 2t ). In the leading log approximation, the mass of a fermion in the
integral B(m f , k1 , k2 ) defined by Eq. (7.71) can be neglected. We get the following
expression for the spectrum of radiated photons:
 
α Zα 2
G 2F km
2
dω ω 1  ω 2 2ω
dσ = 1− + ln3 , (7.78)
54π π π ω E 2 E km

where ω is the photon energy, E is the initial neutrino energy, km is the maximal
momentum of the nucleus recoil. For the high energy neutrinos, within the leading
log approximation the total cross-section of the process is

 α 3 Z 2 G 2 k 2 2E
F m
σ ln4 . (7.79)
2π 27 π km

For example, for a neutrino energy E = 100 GeV we have

Z2
σ∼ 10−46 cm2 . (7.80)
A2/3
This small value of the cross-section makes it difficult to observe the
bremsstrahlung in the neutrino scattering by the coulombian field of the nucleus.
This is true even if one takes into account the distinctive signature of the reaction
as the production of a high energy photon without any accompanying particles. It
must be noted that the same signature in the neutrino reaction may correspond to
the coherent production of photons by nucleons of the nucleus [43, 44]. However,
the process we √consider has a narrower angular distribution of photons, θ < km /E
instead of θ < km /E [43, 44]. Moreover, it is necessary to distinguish in the neu-
trino experiment between the electromagnetic showers produced by photons and by
recoiled electrons in the process νe → νe which has a cross-section 104 times larger
than (7.79).
Nevertheless, we hope that the experimental difficulties we have pointed out can
be overcome in the future. Then the process νγ ∗ → νγ we have discussed could
be accessible to observation. This process (one-loop at the minimum) could be one
7.2 Compton-Like Interaction of Neutrinos with Photons 253

of the few tests for the validity of higher-order perturbation theory in the standard
model of electroweak interaction.

7.2.3 The External Field Influence on the Process γγ → ν ν̄

As it was mentioned already, a strong magnetic field could enhance this process.
Since the electromagnetic tensor field Fμν arises, it opens up a new opportunity to
build a tensor Tαβμν in the amplitude (7.49). In fact, the field comes into the amplitude
in the form of the dimensionless tensor eFμν /m 2e , providing an extra enhancement
if the value of the field exceeds a critical value Be = m 2e /e.
The process γγ → ν ν̄ was investigated in Ref. [45] in the framework of the
standard model in a relatively weak magnetic field B Be , in the lowest-order
expansion over B/Be , and for the case of low photon energies, ω m e . Just in this
approximation it is appropriate to use the effective Lagrangian obtained in Ref. [46]
from the amplitude of the process γγ → γν ν̄ and used in Ref. [45]. It follows from
Ref. [45], that the amplitude of the process depends linearly on the field. As we show
below, this growth takes place only at B Be , but in a strong field B Be , the
amplitude becomes a constant in the case of the standard weak interaction.
The process γγ → ν ν̄ and the crossed channels were also studied in Refs.
[47, 48] in a weak magnetic field, and in a wide region of the photon energy, namely,
for ω < m W . In the limit ω m e , the amplitude obtained in Ref. [48] is consistent
with the result of Ref. [45]. Unfortunately, the amplitude is written in Ref. [48] in
a very cumbersome form, and just the gauge invariance test is extremely difficult to
conduct.
In an earlier paper [49], the process γγ → ν ν̄ was investigated in a strong
magnetic field B Be , for low-energy photons, ω m e , and without taking into
account the contribution of the Z boson.
A general analysis of the three-vertex loop process γγ → ν ν̄ in a strong magnetic
field, based on the asymptotic form of the electron propagator in the field, for arbitrary
kinematic conditions was first performed in Ref. [50].
Consider the general case of a three-vertex loop process in a strong magnetic field,
which is described by the Feynman diagram shown in Fig. 7.5.
In the process of transformation of the photon pair into a pair of neutrino and
antineutrino γγ → ν ν̄, two vertices are vectors, e.g. Γ1 = Γ2 = V , and the third
one can be of the vector and axial-vector type in the standard model, Γ3 = V, A,
and can also be of the scalar and pseudoscalar type when going beyond the standard
model, Γ3 = S, P. In the case Γ3 = V , the diagram of Fig. 7.5 describes also the
photon splitting γ → γγ.
We will use the propagator of the electron in a magnetic field (see Sect. 3.1). The
invariant amplitude of the process described by the Feynman diagram in Fig. 7.5,
with Eqs. (3.1) and (3.19) in account, has the form
254 7 Neutrino-Photon Interactions in External Active Media

M = e2 g3 d4 X d4 Y Tr{( j3 Γ3 )S(Y )(ε2 γ)S(−X − Y )(ε1 γ)S(X )}

× e−ie (X FY )/2 ei(k1 X −k2 Y ) + (γ1 ↔ γ2 ), (7.81)

where X = z − x, Y = x − y, Γ3 is the matrix corresponding to S, P, V or A vertex,

g3 is the coupling constant, j3 is the corresponding part of the neutrino current in a
momentum space, ε1 , k1 and ε2 , k2 are the polarization vectors and the 4-momenta
of the initial photons.
Using the propagator of the form (3.1) and (3.2) in the three-vertex loop leads, in
general, to very cumbersome expressions. The relatively simple results were obtained
only for the process of photon splitting in two cases: in the weak field limit [51] and
in a strong field in the approximation of collinear kinematics [52, 53].
To analyze the amplitude of the process (7.81) in a strong field it is advisable
to use the asymptotic expression for the electron propagator (3.65). Substituting the
propagator into the amplitude, one obtains that two parts of it which differ by the
photon interchange, are proportional to the field strength B:
 
iαg3 eB k1⊥ 2 + k2⊥
2 + (k k )
1 2 ⊥ (k1 ϕk2 )
M − exp − exp −i
(4π)2 2eB 2eB

× d2 p Tr{( j3 Γ3 )S ( p + k2 )(ε2 γ)S ( p)(ε1 γ)S ( p − k1 )}

+ (γ1 ↔ γ2 ), (7.82)

where S ( p) = 2Π− (( pγ) + m e )/( p2 − m 2e ). It should be noted that the projection
operator Π− selects in the amplitude (7.82) only photons of the one polarization
from the two possible, namely, the second mode (see Eq. (4.10)),

Fαβ kβ Fαβ kβ
ε(1)
α = √ , ε(2)
α =  . (7.83)
(k F Fk) Fk)
(k F 

Fig. 7.5 Feynman diagram

for a three-vertex loop process
in a strong magnetic field:
double lines correspond to
the electron propagators
constructed on the base of
the exact solutions of the
Dirac equation in the external
magnetic field
7.2 Compton-Like Interaction of Neutrinos with Photons 255

Using the standard procedure, we can transform the trace in the second term of
Eq. (7.82) with the interchanged photons to the trace in the first term. This proceeds
with the change of sign for Γ3 = P, V, A (and the factor sin[(k1 ϕk2 )/2eB] arises
in the resulting amplitude) and without change of sign for Γ3 = S (and the factor
cos[(k1 ϕk2 )/2eB] appears after summation).
So, when the magnetic field strength is the maximal physical parameter,
eB 2 , k 2 , only the amplitude with the scalar vertex grows linearly with the
k⊥ 
field.

7.2.4 A Conversion γγ → ν ν̄ in the Left–Right Symmetric

Extension of the Standard Model

Using the effective Lagrangian of the ννee√ interaction with the scalar coupling
(7.55), substituting Γ3 = 1, g3 = −4 ζ G F / 2 and j3 = [ν̄e ( p1 )νe (− p2 )] into the
amplitude (7.82) and integrating over the virtual momenta in the strong field limit,
we obtain
1 1−x
8α G F ζ B (2) (2) dy
M= √ [ν̄e ( p1 ) νe (− p2 )] ε1α ε2β dx
π 2 m e Be a2
0 0

× k12
x(1 − 2x) + k2 2
y(1 − 2y) − (k1 k2 ) (1 − 4x y) αβ

α β α β
− (1 − 2x)(1 − 2y) k1 k2 + (1 − 4x y) k2 k1

α β α β
− 2x(1 − 2x) k1 k1 − 2y(1 − 2y) k2 k2 , (7.84)
 
q2 2
k1 2
k2
a = 1 − 2 x y − (1 − x − y) x+ 2 y , (7.85)
me m 2e me

where q = k1 + k2 . The amplitude (7.84) can be rewritten in the explicitly gauge
invariant form (7.49):

α GF   (2)αβ (2)μν
M= √ ν̄e ( p1 ) Tαβμν νe (− p2 ) f 1 f2 , (7.86)
π 2

where the photon field tensors of the 2nd polarization only enter:

(2) (2)
f αβ = kα εβ − kβ ε(2)
α .
256 7 Neutrino-Photon Interactions in External Active Media

The tensor Tαβμν in this case takes the form

1 1−x 
4ζ B dy
Tαβμν = dx αν 
(1 − 4x y)  βμ
m e Be a2
0 0

1
+ 4(1 − x − y)(1 − 2x − 2y) k 

1α βμ 2ν .
k (7.87)
q2

To transform the amplitude to the form (7.86), the following non-trivial integral
identities were used:
1 1−x
Sx(1 − 2x) − T y(1 − 2y)
dx dy ≡ 0, (7.88)
AN
0 0

1 1−x
Z y(1 − 2y) + S(1 − x − y)(1 − 2x − 2y)
dx dy ≡ 0, (7.89)
AN
0 0

A = 1 − Z x y − (1 − x − y)(Sx + T y), (7.90)

where Z , S, T are the arbitrary parameters, N is an arbitrary integer; in this case,

N = 2. The identity (7.88) has been already used earlier for the case of N = 1 (see
Eq. (4.105)).
The expression for the amplitude can be simplified for the two limiting cases
(2)
(here, the polarization vectors ε1,2 are substituted already, see Eq. (7.83)):
(i) at low photon energies, ω  m e :

8α G F ζ B 
M √ [ν̄e ( p1 ) νe (− p2 )] k1
2 k2 ;
2 (7.91)
3π 2 m e Be

(ii) for high-energy photons, ω m e , in the leading log approximation:


2 k2
k1
16α G F B 3 1 2
M √ ζ m e [ν̄e ( p1 ) νe (− p2 )]  ln . (7.92)
π 2 Be 2 2
k1 k2 m 2e

Calculating the cross-sections of the two processes, γγ → (νe ) L (ν̄e ) L and

γγ → (νe ) R (ν̄e ) R , by the standard way, we find that they are equal, σ L L = σ R R ≡ σ.
In the two limiting cases, the expression for the cross-section takes the form
7.2 Compton-Like Interaction of Neutrinos with Photons 257

2 2 k2
2 α2 G 2F ζ 2 B k1 2
σ(ω  m e ) , (7.93)
9π 3 Be m 2e
2 α2 G 2F ζ 2 B 2
m 6e k2 k2
2 1 2
σ(ω me) ln . (7.94)
π3 Be 2 k2
k1 2
m 4e

7.2.5 Possible Manifestations of the γγ → ν ν̄ Process in

Astrophysics

As the observable value in astrophysics, it is considered the stellar energy-loss from

unit volume per unit time due to the neutrino escape (neutrino emissivity). For the
process γγ → ν ν̄ enhanced by a magnetic field, considered in the previous section
it can be written in the form

1 d3 k1 1 d3 k2 1
Q= ω /T
2 (2π) e 1 − 1
3 (2π)3 eω2 /T − 1
(k1 k2 )
× (ω1 + ω2 ) σ (γγ → ν ν̄), (7.95)
ω1 ω2

where T is the temperature of the photon gas. It is taken into account in Eq. (7.95) that
photons of only one polarization are involved in this process. Since only “sterile”
(anti) neutrino of a pair (see Eq. (7.56)) freely departs from hot and dense stellar
medium (other neutrino participating in the standard interaction, has a small free
path and is trapped) the cross-section should be written as (σ L L + σ R R )/2 = σ.
(i) The case of low temperatures, T  m e
In this case, substituting Eq. (7.93) into Eq. (7.95), we obtain

2 2 11
erg ζ B T
Q (B) 2.5 × 1013 . (7.96)
s cm3 0.013 Be me

Let us compare this value with the contributions to the neutrino emissivity
through other mechanisms in the γγ → ν ν̄ process, discussed in this chapter.
For example, for the contribution due to the non-zero mass of neutrinos, it was
obtained in Ref. [24]:

erg  m ν 2 T 11
Q (m ν ) 0.4 × 105 . (7.97)
s cm3 1 eV me

Substituting the cross-section calculated with taking account of the non-locality

of the weak interaction [26] into the expression (7.95), one obtains
258 7 Neutrino-Photon Interactions in External Active Media

13
erg T
Q (nloc) 10 . (7.98)
s cm3 me

It is seen that for B  Be , and for mixing at the level of ζ ∼ 10−5 , the field-
induced mechanism of the reaction γγ → ν ν̄ strongly dominates all the other
indicated mechanisms.
(ii) The case of high temperatures, T me
In the case of high temperatures, substituting Eq. (7.94) into Eq. (7.95), we obtain

2 2 3 5
erg ζ B T T
Q (B) 0.4 × 1012 ln . (7.99)
s cm3 0.013 Be me me

In order to make a numerical estimation, let us consider the Supernova explosion

with generation of very strong magnetic field B ∼ 103 Be , see e.g. [9, 54–56],
with the temperature T ∼ 35 MeV which is believed to be typical for the
Supernova core [57], and V ∼ 1018 cm3 . For the contribution of the considered
field-enhanced process γγ → ν ν̄ into the neutrino luminosity we obtain

2
erg ζ
L ∼ 1045 . (7.100)
s 0.013

It is too small if compared with the typical Supernova neutrino luminosity

1052 erg/sec. Nevertheless, for the field strength B  Be this mechanism could
dominate other discussed in the literature mechanisms of the process γγ → ν ν̄
in the neutrino emissivity of magnetized stars.
As it was noted above, the amplitude (7.82) with the vector and axial-vector
vertices, Γ3 = V, A, does not contain in the strong field limit, eB 2 , k2,
k⊥ 
a part linearly increasing with the field growth. This means that the amplitudes
both for the process γγ → ν ν̄ with the standard ννee interaction, and for the
photon splitting γ → γγ do not depend on the field in this limit. It should be also
noted that neutrinos produced in the standard interaction, do intensely absorbed
by hot and dense stellar environment due to the same interaction. Thus, they can
not provide any significant contribution to the cooling of the central region of
the exploding supernova.

7.3 Neutrino Photoproduction on a Nuclei in a Strong

Magnetic Field

As it was mentioned above, the loop quantum processes whose initial and final
states involve only electrically neutral particles such as neutrinos and photons are of
special interest. The action of an external field on these processes is caused, first, by
the sensitivity of charged virtual fermions to the field. In this case, an electron as a
particle with the maximum specific charge e/m e plays the dominant role. Second,
7.3 Neutrino Photoproduction on a Nuclei in a Strong Magnetic Field 259

Fig. 7.6 Feynman diagram

for the γ + γ + γ → ν + ν̄
process

a strong magnetic field gives rise to a considerable change in the dispersion properties
of photons and, therefore, in their kinematics.
The contribution of the loop process of neutrino-pair photoproduction on a nucleus

γ + Z e → Z e + γ + ν + ν̄ (7.101)

in a strong external magnetic field to the cooling of stars was studied in the paper [58]
and it was stated that this contribution can compete with the contribution from Urca
processes. Therefore, the process (7.101), as one more channel of neutrino energy
loss, would be taken into account when describing the cooling of strongly magnetized
neutron stars. However, the photon dispersion in the field was ignored in Ref. [58].
In this section, the process of photoproduction of a neutrino pair on a nu-
cleus (7.101) is investigated in a strong magnetic field, with taking account of the
photon dispersion in a strong field. The presentation is based on Ref. [59].
The amplitude of neutrino pair photoproduction on a nucleus, Eq. (7.101), can be
derived from the amplitude of the interaction between three photons and a neutrino
pair, e.g.,
γ + γ + γ → ν + ν̄, (7.102)

whose Feynman diagram is shown in Fig. 7.6. As is known (see, e.g., [60]), three-
photon processes (7.102) in a strong magnetic field are more intense than the corre-
sponding two-photon processes, because the amplitude of processes (7.102) with the
vector-axial neutrino current increases linearly with the field, whereas the amplitude
of the γγ → ν ν̄ processes with such a neutrino current is independent of the field.
The amplitude of the process (7.102) in a strong magnetic field can be represented
in the covariant form [60]

8e3 G F eBm 2e
M=− √ (ε1 ϕ̃k1 ) (ε2 ϕ̃k2 ) (ε3 ϕ̃k3 ) [C V ( j ϕ̃k4 ) + C A ( j ϕ̃ϕ̃k4 )]
2π 2
× I (k1 , k2 , k3 ), (7.103)

Here, C V and C A are the vector and axial-vector constants of the effective ννee
Lagrangian (4.66); ε1,2,3 and k1,2,3 are the polarization 4-vectors and photon
4-momenta, respectively; jα = [ν̄(q1 )γα (1 − γ5 )ν(−q2 )] is the Fourier transform
of the neutrino current; k4 = q1 + q2 is the 4-momentum of a neutrino pair.
The form factor I (k1 , k2 , k3 ) has the form of the following triple integral with
respect to the Feynman variables:
260 7 Neutrino-Photon Interactions in External Active Media

1 x y 
1 a(k1 , k2 , k3 )
I (k1 , k2 , k3 ) = dx dy dz
D [m 2e − b(k1 , k2 , k3 )]3
0 0 0

+ {k1 ↔ k2 } + {k2 ↔ k3 } . (7.104)

Here,

D = k12 (k2 k3 ) + k22 (k1 k3 ) + k32 (k1 k2 ) + 2(k1 k2 )(k1 k3 )

+ 2(k1 k2 )(k2 k3 ) + 2(k1 k3 )(k2 k3 ), (7.105)
a(k1 , k2 , k3 ) = k12 (1 − x) 2
− k22 y(1 − y) + k32 z 2 + (k1 k2 ) (1 − 2x)(1 − y)
+ (k1 k3 ) [1 − x − z(1 − 2x)] − (k2 k3 ) y(1 − 2z), (7.106)
b(k1 , k2 , k3 ) = x(1 −
k12 x) + k22
y(1 − y) + k32
z(1 − z) + 2(k1 k2 ) (1 − x)y
+ 2(k1 k3 ) (1 − x)z + 2(k2 k3 ) (1 − y)z. (7.107)

where the scalar products (ki k j ) are the contractions (ki ϕ̃ϕ̃k j ).
For low photon energies, i.e., for ω1,2,3 m e , the integral (7.104) is easily
calculated to give
1
I (k1 , k2 , k3 ) . (7.108)
60 m 8e

In this case, the amplitude (7.103), in view of Eq. (7.108), corresponds to the effective
local γγγν ν̄ Lagrangian

e3 G F eB ∂ Aα 3
Le f f = − √ ϕ̃αβ
45 2π 2 m 6e ∂xβ
∂
× [ν̄γ ρ (1 − γ5 )ν] [C V ϕ̃ρσ + C A (ϕ̃ϕ̃)ρσ ]. (7.109)
∂xσ

The γγγν ν̄ interaction at low energies was previously studied in Ref. [60], where
the Lagrangian was overestimated by a factor of two.
An analysis of the dimensionality of the amplitude (7.103) for the limiting val-
ues of the characteristic photon energy |k1 | ∼ |k2 | ∼ |k3 | ∼ ω indicates that the
amplitude increases as ∼ ω 5 at low energies and decreases as ∼ ω −3 at high energies.
When calculating the amplitude of the process (7.101) on a nucleus in the local
limit of the effective γγγν ν̄ interaction (7.109), it is necessary to take into account
the effect of a strong magnetic field on the dispersion properties of real and virtual
photons. We will demonstrate that this effect is of crucial importance. We recall
that the process (7.101) in a strong magnetic field involves photons of only the 2nd
polarizations.
For a virtual photon, it is necessary to use, instead of the vacuum propagator
∼ q −2 , the propagator including the photon polarization tensor eigenvalue Π (2) (q2 )
7.3 Neutrino Photoproduction on a Nuclei in a Strong Magnetic Field 261

in a magnetic field:
1
D (B) (q2 , q⊥
2
) = , (7.110)
q 2 − Π (2) (q2 )

where, q2 = q02 − qz2 , q⊥

2 = q 2 + q 2 , q 2 = q 2 − q 2 (the magnetic field is directed
x y  ⊥
along the z axis). For the strong field B Be and in the approximation |q2 | m 2e ,
this operator takes the simple form [8]

α B 2
Π (2) (q2 ) − q . (7.111)
3π Be 

It is convenient to introduce the following dimensionless parameter that specifies the

field effect in all subsequent expressions:

α B
β= . (7.112)
3π Be

The parameter β is equal to 0.77 and 7.7 for fields 103 Be and 104 Be , respectively;
i.e., it is not small. Taking into account Eqs. (7.111) and (7.112) and that q0 = 0 for
the virtual photon connected with a fixed nucleus, we can represent the propagator
(7.110) in the form
1
D (B) − 2 . (7.113)
q⊥ + (1 + β)qz2

At the same time, the strong magnetic field also acts on the real photons involved
in process (7.101) and, hence, renormalizes the wave functions:

εα −→ Z2 εα , (7.114)

In view of Eq. (7.111), the renormalization factor Z2 takes the form

 −1
∂Π (q2 ) 1
Z2 = 1 − = . (7.115)
∂q2 1+β

In addition, the kinematic properties of photons change substantially. Taking into

account Eqs. (7.111) and (7.112), one can represent the photon dispersion relation
k 2 −Π (k2 ) = 0 as ω 2 = k2 (1+β cos2 θ)/(1+β) and the element of the momentum
space in the form
 
d3 k = (1 + β) ω 2 dω dy dϕ, y = cos θ 1 + β 1 + β cos2 θ,

where θ and ϕ are the polar and azimuthal angles, respectively.

Using effective Lagrangian (7.109), taking into account the effect of the magnetic
field on photon properties (7.110)–(7.115), and substituting the polarization vectors
262 7 Neutrino-Photon Interactions in External Active Media

of real photons
(ϕ̃k)α
ε(2)
α =  , (7.116)
k2

we represent the amplitude of the process (7.101) in the form


2 2
32παZ G F β 2m N qz k1 k2
M= √ [C V ( j ϕ̃k4 ) + C A ( j ϕ̃ϕ̃k4 )], (7.117)
5 2 m 4e 1 + β q⊥
2 + (1 + β)q 2
z

where m N is the nuclear mass, q α = (0, q) is the momentum transfer to the nucleus.
This expression for the amplitude differs considerably from that obtained in Ref. [58],
where the effect of a strong magnetic field on the dispersion properties of photons
was not considered.
The energy carried away by neutrinos from the stellar unit volume per unit time
is an important quantity in astrophysical applications. It is defined in terms of the
amplitude of the process (7.101) as

(2π)4 n N d3 k1
Qν = |M|2 (ε1 + ε2 ) δ 4 (k1 − k2 − q1 − q2 − q) f (ω1 )
2m N (2π)3 2ω1
d3 k2 d 3 q1 d 3 q2 d3 q
× [1 + f (ω 2 )] , (7.118)
(2π)3 2ω2 (2π)3 2ε1 (2π)3 2ε2 (2π)3 2m N

where n N is the nuclear density, ε1 and ε2 are the energies of neutrino and antineu-
trino, respectively, and f (ω) = [exp(ω/T ) − 1]−1 is the distribution function for
the equilibrium photon gas at the temperature T .
Substitution of the amplitude (7.117) into Eq. (7.118) leads to the following
expression for the neutrino emissivity:

14
8 (2π)9 2 2 2 6 T
Qν = Z α GF me n N J (β) . (7.119)
225 me

The dependence on the field parameter (7.112) is determined by the integral

1 1 1 1

J (β) = β 2
du (1 − u )2
dv (1 − v )
2
ds s (1 − s)
3 8
dr r 2
−1 −1 0 0
1

× dx[u − sv − (1 − s)r x]2 (1 − r 2 x 2 ) C V2 (1 − r 2 ) + C 2A r 2 (1 − x 2 )

−1
2π 2π
dϕ1 dϕ2 1
× , (7.120)
2π 2π [F(β)]2
0 0
7.3 Neutrino Photoproduction on a Nuclei in a Strong Magnetic Field 263

Fig. 7.7 1 Function J (β) [Eq. (7.120)] versus field parameter β; 2 asymptotic behavior of J (β) →
8 × 10−5 at large β; 3 dependence ∼ β 2 obtained disregarding the magnetic field effect on the
photon dispersion

where
  
F(β) = (1 + β) 1 − u 2 + s 2 (1 − v 2 ) − 2s 1 − u 2 1 − v 2 cos ϕ1
  
+ [u − sv − (1 − s)r x]2 − 2 1 + β(1 − s)r 1 − x 2 (7.121)
 
× 1 − u 2 cos ϕ2 − s 1 − v 2 cos(ϕ2 − ϕ1 ) + (1 − s 2 )r 2 (1 − x 2 ),

and the constants C V2 = 0.93 and C 2A = 0.75 are obtained by summing over all
neutrino production channels for the νe , νμ and ντ neutrinos.
The numerically calculated integral (7.120) is shown in Fig. 7.7. It is seen that
taking account of the effect of a strong magnetic field on the photon dispersion
changes fundamentally the dependence of the neutrino energy loss on the field mag-
nitude: the quadratic dependence turns into a constant. Taking this behavior into
account, we obtain an upper limit for Q ν in the asymptotically strong field:

14  
T Z2 ρ erg
Q ν  2.3 × 10 27
, (7.122)
me A ρ0 cm3 s

where Z and A are the charge and mass numbers of the nucleus, the averaging goes
over all nuclei, ρ0 = 2.8 × 1014 g/cm3 is the characteristic nuclear mass density,
and ρ is the average mass density of the star.
264 7 Neutrino-Photon Interactions in External Active Media

The result (7.122) should be compared with the power of neutrino energy loss
through the standard channel of the modified Urca process [61, 62]:

8 2/3
T ρ erg
Q ν (Urca) ∼ 1027 . (7.123)
me ρ0 cm3 s

At first glance, the values (7.122) and (7.123) are of the same order of magnitude.
However, a more careful analysis of Eq. (7.122) indicates that the conclusion made
in Ref. [58] about the competition of the process (7.101) with the Urca processes
at magnetic fields B ∼ 103 Be − 104 Be is erroneous. The cause is that the large
numerical factor arising in Eq. (7.119) and similar formulas in Ref. [58] originates
from the integral over the energy ω1 (x = ω1 /T ) of the initial photon:
∞
x 13 dx (2π)14
= 13! ζ(14) = 6.2 × 109 . (7.124)
ex − 1 24
0

The main contribution to the integral (7.124) comes from x ∼ 10 ÷ 20 (ω1 ∼

(10 ÷ 20) T ). Therefore, since the amplitude (7.117) of the process is obtained in
the approximation ω  m e , the corresponding expression for the neutrino energy loss
power is valid for the photon gas temperatures T  (1/10) m e and is inapplicable at
temperatures T ∼ m e . Thus, the assumption made in [58] that the factor (T /m e )14
can be taken to be on the order of unity is erroneous. Taking into account the above
applicability range, we obtain (T /m e )14  10−14 .
Thus, the catalyzing effect of a strong magnetic field on the process of the neutrino
pair photoproduction on a nucleus decreases considerably if the photon dispersion
in the field is taken into account. Therefore, at any field magnitude, neutrino photo-
production cannot compete with the Urca processes.

References

1. A.V. Kuznetsov, N.V. Mikheev, Phys. Lett. B 299, 367 (1993)

2. A.V. Kuznetsov, N.V. Mikheev, Yad. Fiz. 56(6), 108 (1993). [Phys. At. Nucl. 56, 773 (1993)]
3. D.V. Galtsov, N.S. Nikitina, Zh. Eksp. Teor. Fiz. 62, 2008 (1972). [Sov. Phys. JETP 35, 1047
(1972)]
4. V.V. Skobelev, Zh. Eksp. Teor. Fiz. 71, 1263 (1976). [Sov. Phys. JETP 44, 660 (1976)]
5. A.N. Ioannisian, G.G. Raffelt, Phys. Rev. D 55, 7038 (1997)
6. A.A. Gvozdev, N.V. Mikheev, L.A. Vassilevskaya, Phys. Lett. B 410, 211 (1997)
7. N.P. Klepikov, Zh. Eksp. Teor. Fiz. 26, 19 (1954)
8. A.E. Shabad, in Polarization of the Vacuum and a Quantum Relativistic Gas in an External
Field, ed. by V.L. Ginzburg (Nova Science Publishers, New York, 1992)
9. G.S. Bisnovatyi-Kogan, Astron. Zh. 47, 813 (1970). [Sov. Astron. 14, 652 (1971)]
10. G.S. Bisnovatyi-Kogan, in Stellar Physics 2: Stellar Evolution and Stability (Nauka, Moscow
1989; Springer, New York 2001)
11. V.N. Tsytovich, Zh. Eksp. Teor. Fiz. 45, 1183 (1963). [Sov. Phys. JETP 18, 816 (1964)]
References 265

12. J.C. D’Olivo, J.F. Nieves, P.B. Pal, Phys. Lett. B 365, 178 (1996)
13. S.J. Hardy, D.B. Melrose, Publ. Astron. Soc. Aus. 13, 144 (1996)
14. G.S. Bisnovatyi-Kogan, S.G. Moiseenko, Astron. Zh. 69, 563 (1992). [Sov. Astron. 36, 285
(1992)]
15. G. S. Bisnovatyi-Kogan, Astron. Astrophys. Trans. 3, 287 (1993)
16. M.V. Chistyakov, N.V. Mikheev, Phys. Lett. B 467, 232 (1999)
17. P. Elmfors, D. Persson, B. Skagerstam, Nucl. Phys. B 464, 153 (1996)
18. B. Pontecorvo, Zh. Eksp. Teor. Fiz. 9, 1615 (1959). [Sov. Phys. JETP 9, 1148 (1959)]
19. H.-E. Chiu, P. Morrison, Phys. Rev. Lett. 5, 573 (1960)
20. M. Gell-Mann, Phys. Rev. Lett. 6, 70 (1961)
21. L.D. Landau, Dokl. Akad. Nauk SSSR 60, 207 (1948)
22. C.N. Yang, Phys. Rev. 77, 242 (1950)
23. R.J. Crewther, J. Finjord, P. Minkowski, Nucl. Phys. B 207, 269 (1982)
24. S. Dodelson, G. Feinberg, Phys. Rev. D 43, 913 (1991)
25. M.J. Levine, Nuovo Cim. A 48, 67 (1967)
26. D.A. Dicus, W.W. Repko, Phys. Rev. D 48, 5106 (1993)
27. D.A. Dicus, K. Kovner, W.W. Repko, Phys. Rev. D 62, 053013 (2000)
28. J. Liu, Phys. Rev. D 44, 2879 (1991)
29. E.M. Lipmanov, Yad. Fiz. 6, 541 (1967). [Sov. J. Nucl. Phys. 6, 395 (1968)]
30. E.M. Lipmanov, N.V. Mikheev, Pis’ma Zh. Eksp. Teor. Fiz. 7, 139 (1968). [JETP Lett. 7, 107
(1968)]
31. E.M. Lipmanov, Zh. Eksp. Teor. Fiz. 55, 2245 (1968). [Sov. Phys. JETP 28, 1191 (1969)]
32. J.C. Pati, A. Salam, Phys. Rev. D 10, 275 (1974)
33. M.A.B. Bég, R. Budny, R.N. Mohapatra, A. Sirlin, Phys. Rev. Lett. 38, 1252 (1977)
34. R.N. Mohapatra, J.C. Pati, Phys. Rev. D 11, 566 (1975)
35. R.N. Mohapatra, J.C. Pati, Phys. Rev. D 11, 2558 (1975)
36. G. Senjanović, R.N. Mohapatra, Phys. Rev. D 12, 1502 (1975)
37. M. Czakon, J. Gluza, M. Zralek, Phys. Lett. B 458, 355 (1999)
38. R. Barbieri, R.N. Mohapatra, Phys. Rev. D 39, 1229 (1989)
39. L. Rosenberg, Phys. Rev. 129, 2786 (1963)
40. V.K. Cung, M. Yoshimura, Nuovo Cim. 29 A, 557 (1975)
41. L.A. Vassilevskaya, A.A. Gvozdev, N.V. Mikheev, Yad. Fiz. 57, 124 (1994). [Phys. At. Nucl.
57, 117 (1994)]
42. V.Ch. Zhukovsky, P.A. Eminov, A.E. Grigoruk, Mod. Phys. Lett. A 11, 3119 (1996)
43. D. Rein, L.M. Sehgal, Phys. Lett. B 104, 394 (1981). [Erratum: ibid. B 106, 513 (1981)]
44. S.S. Gershtein, Yu.Ya. Komachenko, M.Yu. Khlopov, Yad. Fiz. 33, 1597 (1981). [Sov. J. Nucl.
Phys. 33, 860 (1981)]
45. R. Shaisultanov, Phys. Rev. Lett. 80, 1586 (1998)
46. D.A. Dicus, W.W. Repko, Phys. Rev. Lett. 79, 569 (1997)
47. T.-K. Chyi, C.-W. Hwang, W.F. Kao, G.-L. Lin, K.-W. Ng, J.-J. Tseng, Phys. Lett. B 466, 274
(1999)
48. T.-K. Chyi, C.-W. Hwang, W.F. Kao, G.-L. Lin, K.-W. Ng, J.-J. Tseng, Phys. Rev. D 62, 105014
(2000)
49. Yu.M. Loskutov, V.V. Skobelev, Vestn. Mosk. Univ. Fiz. Astron. 22, 10 (1981)
50. A.V. Kuznetsov, N.V. Mikheev, Mod. Phys. Lett. A 16, 1659 (2001)
51. S.L. Adler, Ann. Phys. (N.Y.) 67, 599 (1971)
52. V.N. Baier, A.I. Milstein, R.Zh. Shaisultanov, Phys. Rev. Lett. 77, 1691 (1996)
53. V.N. Baier, A.I. Milstein, R.Zh. Shaisultanov, Zh. Eksp. Teor. Fiz. 111, 52 (1997). [JETP 8 4,
29 (1997)]
54. R.C. Duncan, C. Thompson, Astrophys. J. 392, L 9 (1992)
55. P. Bocquet, S. Bonazzola, E. Gourgoulhon, J. Novak, Astron. Astrophys. 301, 757 (1995)
56. C.Y. Cardall, M. Prakash, J.M. Lattimer, Astrophys. J. 554, 322 (2001)
57. G.G. Raffelt, in Stars as Laboratories for Fundamental Physics (University Chicago Press,
Chicago 1996)
266 7 Neutrino-Photon Interactions in External Active Media

58. V.V. Skobelev, Zh. Eksp. Teor. Fiz. 120, 786 (2001). [JETP 93, 685 (2001)]
59. A.V. Kuznetsov, N.V. Mikheev, Pis’ma Zh. Eksp. Teor. Fiz. 75, 531 (2002). [JETP Lett. 75,
441 (2002)]
60. Yu.M. Loskutov, V.V. Skobelev, Teor. Mat. Fiz. 70, 303 (1987). [Theor. Math. Phys. 70, 215
(1987)]
61. B.L. Friman, O.V. Maxwell, Astrophys. J. 232, 541 (1979)
62. D.G. Yakovlev, K.P. Levenfish, Astron. Astrophys. 297, 717 (1995)
Chapter 8
Conclusion

The questions raised in this book, refer to the actual scientific direction, lying at
the junction of plasma physics, high magnetic fields, quantum field theory, particle
physics and astrophysics. Analysis of problems in the physics of hot dense mag-
netized plasma, resulting in a detailed quantitative description of the core collapse
supernova, definitely points to the need for development of new physics that may
be associated with the equation of state of nuclear and subnuclear plasma and weak
interactions in the subnuclear regime, as well as the need for further research on
the fundamental properties of neutrinos and mechanisms of neutrino interactions in
hot dense strongly magnetized plasma, or on the need for the consideration of other,
hypothetical, weakly interacting elementary particles.
This branch of science intensively developing for about 40 years, is, of course, far
from being complete. There are high expectations both in the further development
of a theory, and for new experimental results.
As for the development of a theory, it is impossible to predict an emergence of new
productive ideas. However, in the framework of the already developed theoretical
apparatus, comprehensive studies surely will continue of hot dense plasma consisting
of electron–positron, proton and nucleon components at extreme physical parameters.
These are the physical conditions which are realized in the central part of massive
stars. At the same time, these conditions are relevant to the characteristics of nuclear
and subnuclear matter. Among the factors affecting the astrophysical plasma, which
need to be considered, an important role is played by a strong magnetic field and an
intensive neutrino flux. In particular, the following questions should be examined:
1. Effect of plasma and magnetic field on the physical characteristics of a neutrino.
2. Neutrino absorption and emission by plasma in a magnetic field.
3. The joint effect of the plasma, magnetic field and the neutrino flux on electro-
magnetic radiation and its inverse effect on the plasma.
4. Mechanisms of generation of electron-positron plasma by a flux of high-energy
photons and electrons.
As for the experimental studies related to the field, they could be divided into two
directions. The first group is formed by the ground-based experiments, in particular,

A. Kuznetsov and N. Mikheev, Electroweak Processes in External Active Media, 267

Springer Tracts in Modern Physics 252, DOI: 10.1007/978-3-642-36226-2_8,
© Springer-Verlag Berlin Heidelberg 2013
268 8 Conclusion

neutrino beam experiments. The principal expectations are apparently associated

with long-baseline neutrino oscillation experiments, aimed at clarifying the mixing
parameters and other physical characteristics of a neutrino.
The second area is a group of experiments that can be described as cosmic-
terrestrial. It is connected with observations of astrophysical objects, primarily, the
remnants of supernova explosions, in a wide range of electromagnetic and other types
of radiation. A special class of experiments is formed by neutrino telescopes with the
underground, underwater and under the Antarctic ice location, which are focused on
the expected explosion of a galactic supernova. An interesting possibility also exists
of registration the neutrino signal from the core collapse of a massive star that occurs
without disruption of an envelope, that is, without formation of a supernova. Finally,
high expectations are associated with gravitational wave detectors, which, according
to experts, are very close to the level of sensitivity to the optimistic outlook for the
border of the astrophysical intensity and probable frequency of gravitational wave
signals expected from the collapse of massive stars.
We still expect a lot of discoveries.
Index

A in a strong field, 194

Adler triangle anomaly, 57, 58, 60, 230, 231 density matrix, 130
Airy function, 19, 135, 137, 139, 176, 188, 189 Elliptic integrals, 185
derivative, 176 Euclidean {1, 2}-subspace, 7
Angular momentum Euler constant, 51, 186
conservation law, 181 Euler dilogarithm, 86, 208
Axion decay, 5

F
B Fock proper-time formalism, 25, 38
Bessel function, 42 FOE problem, 196
modified, 95, 176, 185, 193
Bessel integral, 42
G
Gamma function, 108, 139, 192
C Gaussian integral, 129, 142
Cyclotron resonance, 50, 234 generalized, 54
Gaussian packet, 131
Gell-Mann theorem, 245
D
Dirac equation, 13
negative energy solution, 13, 21, 129 H
positive energy solution, 13, 129, 140 Hardy—Stokes function, 59, 106, 139, 188
Dirac gamma matrices, 8 Harmonic oscillator
Distribution functions, 16
Bose–Einstein, 150 lowering operator, 15
Fermi—Dirac, 62, 70, 148 raising operator, 15
Hermite polynomials, 16

E
Electromagnetic field L
crossed, 6, 20, 38, 58, 133 Lagrangian
dynamical parameter, 20, 186 effective, 53
intensity parameter, 185, 194 of neutrino–electron interaction, 229
invariants, 6, 21, 60 effective local, 69
Electron electromagnetic interaction, 128
chemical potential, 194 magnetic moment interaction, 119

A. Kuznetsov and N. Mikheev, Electroweak Processes in External Active Media, 269

Springer Tracts in Modern Physics 252, DOI: 10.1007/978-3-642-36226-2,
Ó Springer-Verlag Berlin Heidelberg 2013
270 Index

Laguerre polynomials, 32 Neutrinosphere, 196

associated, 24, 114 Neutron stars, 3, 196
generating function, 32 merger of, 3, 6
Landau gauge, 14
Landau level
first, 140 P
ground, 6, 19, 129, 130, 140, 180, 181 Pauli matrices, 8, 14
Landau levels, 16 Photon
higher, 19 1st polarization, 234, 235
Landau–Yang theorem, 245 2nd polarization, 232, 234, 235
Lepton mixing matrix, 229 decay into neutrino pair, 5, 234, 235
Life devastation on Earth, 5 dispersion in plasma, 76
Longitudinal polarization operator, 14 renormalization of the wave function, 50,
233
Photon pair conversion into neutrino pair, 244
M Photon splitting, 5
Magnetar, 3 Plasma frequency, 45
Magnetic field Plasmon
critical, 3, 6, 21 effective mass, 74, 75, 82–85
W-boson, 33 Polylogarithm
galactic, 4 nth-order, 208
poloidal, 4 Polynomial
primary, 4 Hermite, 16
toroidal, 4, 215, 239 Laguerre, 32
Minkowski {0, 3}-subspace, 7, 20 associated, 24, 114
generating function, 32
Positron
N density matrix, 130
Neutrino Projecting operator, 130
‘‘spin light’’ of, 74 Pulsar, 3, 6, 196, 197
‘‘window of transparency’’ in magnetized anomalous X-ray, 3
plasma, 211 space velocity, 197, 200
additional energy, 68, 71, 74, 80, 86, 88,
109–111, 117–119
Cherenkov process, 5 R
Compton-like process, 231 Riemann zeta function, 239
emissivity, 238, 262
energy loss, 192, 234
mass eigenstate, 230 S
massive Schrödinger stationary equation, 14
radiative decay of, 5, 234 Soft gamma-ray repeater, 3
massless Specific charge, 6
radiative transition of, 5, 53, 234 Supernova
mean free path, 192, 208 envelope, 5, 196
momentum loss, 192, 234 explosion, 3, 6, 196, 239
self-energy operator, 68, 71, 79, 89–91, remnant, 192
100, 104, 111 Symmetry
spectral temperatures, 212 broken left-right, 246
Index 271

T W
Two-dimensional electrodynamics, 20 Weinberg angle, 69
covariant extension, 20

U
Ultraviolet divergence, 54, 230
Unitarity relation, 132, 182