Physics Reports 592 (2015) 1–38

Contents lists available at ScienceDirect

Physics Reports
journal homepage: www.elsevier.com/locate/physrep

Transverse and longitudinal angular momenta of light

Konstantin Y. Bliokh a,b,∗ , Franco Nori a,c
a
Center for Emergent Matter Science, RIKEN, Wako-shi, Saitama 351-0198, Japan
b
Nonlinear Physics Centre, RSPhysE, The Australian National University, Canberra, ACT 0200, Australia
c
Physics Department, University of Michigan, Ann Arbor, MI 48109-1040, USA

article info abstract

Article history: We review basic physics and novel types of optical angular momentum. We start with
Accepted 30 June 2015 a theoretical overview of momentum and angular momentum properties of generic
Available online 8 July 2015 optical fields, and discuss methods for their experimental measurements. In particular,
editor: D.K. Campbell
we describe the well-known longitudinal (i.e., aligned with the mean momentum) spin
and orbital angular momenta in polarized vortex beams. Then, we focus on the transverse
(i.e., orthogonal to the mean momentum) spin and orbital angular momenta, which were
recently actively discussed in theory and observed in experiments. First, the recently-
discovered transverse spin angular momenta appear in various structured fields: evanescent
waves, interference fields, and focused beams. We show that there are several kinds of
transverse spin angular momentum, which differ strongly in their origins and physical
properties. We describe extraordinary features of the transverse optical spins and overview
recent experiments. In particular, the helicity-independent transverse spin inherent in
edge evanescent waves offers robust spin–direction coupling at optical interfaces (the
quantum spin Hall effect of light). Second, we overview the transverse orbital angular
momenta of light, which can be both extrinsic and intrinsic. These two types of the
transverse orbital angular momentum are produced by spatial shifts of the optical beams
(e.g., in the spin Hall effect of light) and their Lorentz boosts, respectively. Our review is
underpinned by a unified theory of the angular momentum of light based on the canonical
momentum and spin densities, which avoids complications associated with the separation
of spin and orbital angular momenta in the Poynting picture. It allows us to construct a
comprehensive classification of all known optical angular momenta based on their key
parameters and main physical properties.
© 2015 Elsevier B.V. All rights reserved.

Contents

1. Introduction............................................................................................................................................................................................. 2
1.1. Conventions and notations ........................................................................................................................................................ 3
2. Spin and orbital angular momenta: basic properties ........................................................................................................................... 3
2.1. Particles and waves .................................................................................................................................................................... 3
2.2. Fundamental properties of generic optical fields ..................................................................................................................... 5
2.3. Longitudinal spin and orbital AM in optical beams.................................................................................................................. 7
2.4. Local measurements of the spin and orbital AM ...................................................................................................................... 9

∗ Corresponding author.
E-mail address: [email protected] (K.Y. Bliokh).

http://dx.doi.org/10.1016/j.physrep.2015.06.003
0370-1573/© 2015 Elsevier B.V. All rights reserved.
2 K.Y. Bliokh, F. Nori / Physics Reports 592 (2015) 1–38

3. Transverse spin angular momenta......................................................................................................................................................... 11

3.1. Evanescent waves ....................................................................................................................................................................... 11
3.1.1. Transverse (out-of-plane) helicity-independent spin ............................................................................................... 11
3.1.2. Transverse (in-plane) dual-antisymmetric spin ........................................................................................................ 13
3.2. Propagating waves...................................................................................................................................................................... 15
3.2.1. Two-wave interference ............................................................................................................................................... 15
3.2.2. Integral transverse (in-plane) helicity-dependent spin ............................................................................................ 17
3.2.3. Focused Gaussian beam .............................................................................................................................................. 19
3.3. Measurements and applications ................................................................................................................................................ 20
3.3.1. Reconstructions of 3D fields ....................................................................................................................................... 21
3.3.2. Optomechanical calculations and measurements ..................................................................................................... 21
3.3.3. Probing optical spin using atoms in a magnetic field................................................................................................ 22
3.3.4. Magneto-optical and particle-scattering measurements ......................................................................................... 23
3.3.5. Spin-directional coupling and the quantum spin Hall effect of light ....................................................................... 24
4. Transverse orbital angular momenta..................................................................................................................................................... 26
4.1. Transverse extrinsic orbital AM ................................................................................................................................................. 26
4.1.1. General features........................................................................................................................................................... 26
4.1.2. Extrinsic AM in the spin Hall effect of light ............................................................................................................... 27
4.2. Transverse intrinsic orbital AM.................................................................................................................................................. 30
5. Conclusions.............................................................................................................................................................................................. 33
Acknowledgments .................................................................................................................................................................................. 34
References................................................................................................................................................................................................ 34

1. Introduction

Angular momentum (AM) was recognized as one of the important characteristics of light after the pioneering work by
Poynting [1] and the first experimental evidence by Beth [2]. These works considered the spin AM produced by the circular
polarization of a light beam. In 1992, a paper by Allen et al. [3] started a new era of AM studies in optics. This work described
the orbital AM in so-called vortex beams, which was soon detected experimentally [4]. Since then, the study of optical
angular momentum has grown into a large research field with numerous applications in optical manipulations, quantum
information, photonics, plasmonics, and astrophysics (see the books [5–9] and reviews [10–14]).
Importantly, the spin and orbital AM of light are separately observable properties in optics [5–20]. The separation of AM
into spin and orbital parts is straightforward in paraxial monochromatic beams [3,5–14,21]. At the same time, fundamental
difficulties in quantum electrodynamics and field theory [22–26] result in a number of subtleties for the spin and orbital
AM description in generic non-paraxial or non-monochromatic fields [27–38]. Still, the spin–orbital AM decomposition is
possible and physically meaningful in optics, and we will rely on it in this review.
The spin AM is associated with the polarization of light, so that right-hand and left-hand circular polarizations of a paraxial
beam correspond to the positive and negative helicities σ = ±1 of photons. From paraxial-beam optics [1–18,21] and the
quantum mechanics of photons [22], we know that the spin AM of light is aligned with the direction of propagation of light. If
the mean momentum of the beam (in units of } per photon) can be associated with its mean wave vector ⟨k⟩, then such beam
carries the corresponding spin AM ⟨S⟩ = σ ⟨k⟩/k. Here σ is the helicity parameter, i.e., the degree of circular polarization.
In turn, the orbital AM of a uniformly-polarized paraxial beam is independent of the polarization, and is produced by
the phase and momentum circulation in a scalar wave function. For instance, an optical vortex, i.e., a phase singularity with
a helical phase around it [39–43], generates a circulation of the local momentum density (phase gradient) [14,30] and,
thereby, produces an orbital angular momentum. The orbital AM states of monochromatic light are paraxial vortex beams,
which carry the orbital AM along the beam axis: ⟨L⟩ = ℓ⟨k⟩/k [3–18,21]. Here ℓ is the quantum number (topological charge)
of the optical vortex.
Thus, in the majority of situations considered so far [1–21], the angular momentum of light is aligned with its mean
momentum ⟨k⟩. In other words, it is longitudinal. Recently, there has been a rapidly growing interest in optical fields
with a transverse AM [44–59]. Such angular momentum is orthogonal to the propagation direction (mean momentum) of
light. Importantly, different transverse angular momenta, of either spin or orbital nature, have drastically different physical
properties and origins.
For example, the simplest case of the transverse AM is the orbital AM of a classical point particle with the coordinates r
and momentum p: L = r × p [60]. This AM is extrinsic, i.e., it depends on the choice of the coordinate  origin.
 Similar extrinsic
orbital AM is also carried by a light beam passing at some distance from the coordinate origin: Lext = ⟨r⟩ × ⟨k⟩, where
⟨r⟩ is the mean position, i.e., shift of the beam. Despite its ‘‘trivial’’ character, in 1987 Player and Fedoseyev showed [61,62]
that this extrinsic orbital AM plays a key role in the transverse spin-dependent shifts of light beams reflected or refracted
at planar interfaces. The latter effect is now known as the spin-Hall effectof light or Imbert–Fedorov shift [63–73], and the
extrinsic transverse orbital AM was intensively considered in this context (see [73] for a review).
In contrast to the previous example, there is a more intriguing transverse spin AM, which was first described in 2012 by
Bliokh and Nori [46]. This transverse spin appears locally in structured optical fields, such as evanescent waves, focused
 K.Y. Bliokh, F. Nori / Physics Reports 592 (2015) 1–38 3

beams, and two-wave interference [46,48,50–59]. The transverse spin has extraordinary properties, which are in sharp
contrast to what has been known so far about the spin of photons. Namely, it is independent of the helicity of the wave and can
appear even for linearly polarized waves. (Actually, here we show that there is a family of transverse-spin AM, which have
different physical properties and depend on different field parameters.) Moreover, this polarization-independent transverse
spin strongly depends on the direction of propagation of the wave. Owing to such extraordinary features, the transverse
spin AM in evanescent waves has already found important applications in spin-dependent unidirectional optical interfaces
[52,53,56,57,74–83]. The robust coupling of the transverse spin with the propagation direction in evanescent waves is a
fundamental property of Maxwell equations, which can be associated with the quantum spin Hall effect of light [56].
Here we overview recent theoretical and experimental investigations of different types of AM of light. By exploiting
a unified theory of optical angular momentum based on the canonical (rather than Poynting) momentum and spin
densities [34,37,38,50,54,84], we aim to provide a coherent picture of all the basic kinds of AM, and describe their main
physical features. We particularly focus on the transverse AM, which so far have not been considered in reviews on the
optical AM [5–14]. We describe basic systems, where the transverse AM may appear and play an important role, and discuss
the main experimental works related to the transverse AM (even where its presence was not recognized).
The paper is organized as follows. In Section 2 we provide an overview of the general AM properties in mechanics and
wave physics. We describe canonical momentum, angular-momentum, spin, and helicity densities in generic optical fields,
consider paraxial optical beams carrying longitudinal spin and orbital AM, and discuss measurements of the spin and orbital
AM in optical fields. Section 3 considers various types of transverse spin AM in structured (inhomogeneous) fields. These
include evanescent waves, two-wave interference, and focused beams. In addition to a detailed theoretical analysis, we
review a number of experimental works revealing the transverse spin. The relation of the transverse spin in evanescent
waves to the spin–direction coupling (quantum spin Hall effect of light) is discussed. Section 4 examines two types of
transverse orbital AM: the extrinsic and intrinsic ones. We show that the former (extrinsic) case is related to transverse
shifts of paraxial beams, and it plays an important role in the spin Hall effect of light. The latter (intrinsic) case is realized
in polychromatic spatio-temporal beams, which can be obtained, e.g., via a transverse Lorentz boost to a moving reference
frame. Section 5 summarizes the results and provides a classification of different kinds of AM (Table 1).

1.1. Conventions and notations

Throughout this paper, different types of AM are emphasized by frames around their equations, and their key properties
(spin/orbital, intrinsic/extrinsic, longitudinal/transverse, etc.) are highlighted in special framed equations with Roman
numbers: (I), (II), etc.
It should be emphasized that ‘‘transverse’’ with respect to the mean momentum can be applied to two mutually
orthogonal directions. In many problems, the structured waves are formed by at least two wave-vector or momentum
quantities forming the ‘‘plane of propagation’’ (as, e.g., in an evanescent wave or two-wave interference). In these cases,
we distinguish ‘‘transverse (out-of-plane)’’ and ‘‘transverse (in-plane)’’ quantities, which typically strongly differ in their
properties and origin.
We pay special attention to the time-reversal (T ) and spatial-inversion (P ) symmetries of the parameters determining
different types of angular momentum. Any AM must change its sign under the time-reversal (T ) transformation, but
preserve it under the spatial-inversion (P ) transformation. This enables a clear identification and separation of different
kinds of AM, based on the P - and T -symmetries of the key parameters in their equations.
Overall, the framed equations throughout the paper and the final Table 1 should provide a clear guide to and summary
of the main results.
We also remark on the notations and units used in this review. First, throughout the paper (apart from Section 4.2)
we deal with monochromatic electromagnetic fields of fixed frequency ω, characterized by complex electric and magnetic
field amplitudes E(r) and H(r). The real electric and magnetic fields are given by E (r, t ) = Re[E(r)e−iωt ] and H (r, t ) =
Re[H(r)e−iωt ].
Second, for the sake of simplicity, we mostly consider optical fields in the vacuum and use Gaussian units (to avoid
multiple appearances of the vacuum permittivity and permeability constants). Furthermore, since we do not consider any
truly quantum phenomena, we imply } = 1 units when using the quantum operator formalism for the momentum and AM
of light. In this manner, the energy and momentum of a wave are directly associated with the frequency ω and wave vector k.
Finally, to make the main equations as simple and clear as possible, we often omit inessential factors and use the
proportionality sign ‘‘∝’’ instead of the exact equality sign ‘‘=’’.

2. Spin and orbital angular momenta: basic properties

2.1. Particles and waves

We start with an introductory discussion of the AM properties in mechanics and wave physics. The AM reveals very
different features for particles and waves. First, let us consider the limiting case of a classical point particle. Such particle
4 K.Y. Bliokh, F. Nori / Physics Reports 592 (2015) 1–38

carries orbital AM [60]

L=r×p , (2.1)

where r and p are the position and momentum of the particle, respectively (Fig. 1a). This mechanical AM is extrinsic,
i.e., dependent on the choice of the coordinate origin. Indeed, the translational transformation of the coordinates,
r → r + r0 , (2.2)
changes the AM (2.1) as
L → L + r0 × p. (2.3)
By definition, a point particle cannot have any internal structure and, hence, cannot carry an intrinsic AM. Evidently, the AM
(2.1) is transverse, i.e., orthogonal to the momentum. Since the momentum p is a P -odd and T -odd quantity, whereas the
position r is P -odd and T -even, the AM (2.1) L is P -even and T -odd, as it should be. Thus, summarizing the properties of
the point-particle AM:

Point particle AM: Orbital, Extrinsic, Transverse. Key parameters: p, r. (I)

In contrast to localized particles, waves are extended entities, with the limiting case being a plane wave. Such wave cannot
carry an extrinsic or orbital AM of the type (2.1), because its position is undefined. A circularly-polarized electromagnetic
plane wave, propagating along the z-axis can be written as
x̄ + iσ ȳ
E∝ √ exp(ikz ), H = z̄ × E, (2.4)
2
where x̄, ȳ, and z̄ denote the unit vectors of the corresponding Cartesian axes, the helicity parameter σ = ±1 corresponds to
the right-hand and left-hand circular polarizations (Fig. 1b), and the wave number k = ω/c. The electric (or magnetic)
field (2.4) represents eigenmodes of the z-component of the spin-1 matrix operator Ŝ (generators of the SO(3) vector
rotations) [21,22,31]:

0 0 0 0 0 −1 0 1 0
     
Ŝ = −i 0 0 1 , 0 0 0 , −1 0 0 , Ŝz E = σ E. (2.5)
0 −1 0 1 0 0 0 0 0

Therefore, the plane wave (2.4) carries the spin AM density S defined as the local expectation value of the operator Ŝ with the
wavefunction E or H [21,34] (see also Sections 2.2 and 2.3):

k
S∝σ . (2.6)
k
Here we wrote the spin density in a form valid for an arbitrary propagation direction of the plane wave, k/k.
The spin AM density (2.6) is generated by rotating electric and magnetic fields in a circularly-polarized wave, Fig. 1b.
The total (integral) AM is ill-defined for a plane wave, as it cannot be localized and all integrals diverge. Evidently, the spin
density (2.6) is independent of the radius-vector r, and thus it is an intrinsic quantity. Furthermore, this AM is longitudinal,
i.e., directed along the momentum (wave vector) k. The helicity σ represents the chirality of light [85–89], and, hence, it
is a P -odd and T -even quantity. Since the wave momentum k is P -odd and T -odd, the spin (2.6) has the proper T -odd
symmetry of the AM. Summarizing the properties of the plane-wave AM:

Plane wave AM: Spin, Intrinsic, Longitudinal. Key parameters: σ , k. (II)

Thus, we see that the AM of point particles and plane waves exhibit drastically different properties. These limiting cases
encompass the variety of AM features appearing in structured optical fields.
Since optics deals with waves rather than particles, it is natural to look at the quantum counterpart of the mechanical
orbital AM (2.1). It is described by the canonical quantum operator L̂ = r̂ × p̂ [21,22,31], where r̂ = r and p̂ = −i∇ in the
coordinate representation. Using the cylindrical coordinates (ρ, ϕ, z ), one can write the z-component of this operator and
its eigenmodes as [3–14,21,31]:
∂
L̂z = −i , E ∝ exp(iℓϕ), L̂z E = ℓ E. (2.7)
∂ϕ
These eigenmodes are vortices, i.e., helical-phase waves, where ℓ = 0, ±1, ±2, . . . is the vortex charge [39–43]. The vortex
modes essentially represent structured wave fields (they are produced by the interference of multiple plane waves), and,
hence, combine some properties of particles and waves. We consider the simplest examples of such modes, paraxial vortex
beams, in Section 2.3. Remarkably, in sharp contrast to the mechanical orbital AM (2.1), the orbital AM of the vortex beams
becomes longitudinal and even intrinsic.
 K.Y. Bliokh, F. Nori / Physics Reports 592 (2015) 1–38 5

Fig. 1. Angular momentum of pure particles and waves. (a) A classical point particle carries orbital AM (2.1) and (I) L, which is extrinsic and transverse
(i.e., orthogonal to the particle momentum p). (b) A single electromagnetic plane wave carries spin AM density (2.6) and (II) S, which is intrinsic and
longitudinal (i.e., aligned with the wave vector k). This spin AM is produced by the rotation of the electric and magnetic fields in a circularly-polarized
wave. Here the instantaneous distribution of the real electric and magnetic fields E (z , 0) and H (z , 0) is shown for a right-hand circularly-polarized wave
with helicity σ = 1.

2.2. Fundamental properties of generic optical fields

We now describe the main momentum and angular-momentum properties of generic optical fields. We still consider
monochromatic fields with frequency ω, but they can have arbitrarily inhomogeneous spatial distributions E(r) and H(r),
which fulfill free-space Maxwell equations.
Prior to describing angular momentum, we need to introduce the energy and momentum of light. The time-averaged
energy density of a monochromatic optical field is [90,91]
gω 
|E|2 + |H|2 ,

W = (2.8)
2
where g = (8π ω)−1 in Gaussian units.
It is widely accepted that the momentum density of a free electromagnetic field is described by the Poynting vector
5 = gk Re (E∗ × H) [90,91]. However, it turns out that the more relevant and directly measurable quantity is the so-called
canonical (or orbital) momentum density [14,30,34,50,54,89,92], which appears in canonical Noether conservation laws (in
the Coulomb gauge) in electromagnetic field theory [23,34,37,38]. This momentum density can be written as
g
Im E∗ · (∇)E + H∗ · (∇)H ,
 
P= (2.9)
2
where we use the notation X · (Y)Z ≡

i Xi YZi . Unlike the Poynting vector, the canonical momentum density (2.9) has
an intuitively-clear physical interpretation: it is proportional to the local gradient of the phase of the field, i.e., to the local
wave vector [30] (assuming contributions from both the electric and magnetic fields). Therefore, the canonical momentum
density is independent of the polarization in uniformly-polarized fields and can be equally defined for a scalar wave field
ψ(r). The Poynting vector coincides with the canonical momentum in all cases where one can neglect elliptical polarization
and spin AM phenomena. In all other situations these differ by the so-called ‘‘spin momentum’’ contribution introduced by
Belinfante in field theory and non-measurable in basic light–matter interactions [14,23,25,30,34,50,54].
6 K.Y. Bliokh, F. Nori / Physics Reports 592 (2015) 1–38

The circulation of the canonical momentum immediately yields the orbital AM density in the optical field [14,31,34,37,
50,84]:

L=r×P . (2.10)

This orbital AM density is an extrinsic and transverse quantity, akin to the mechanical AM (2.1)–(2.3). Furthermore, since it
is defined via the momentum density P, it is not an independent property of the field. Thus, the orbital AM (2.10) locally has
properties (I) of the mechanical AM. However, as we show below, the integral orbital AM of the field, ⟨L⟩, can exhibit quite
different properties.
The energy, momentum and orbital AM densities can be equally defined for scalar fields. The specific vector character of
the electromagnetic field manifests itself in the spin AM. The spin AM density is truly intrinsic and can be written as [21,30,
32,34,37,50,84]

g
Im E∗ × E + H∗ × H .
 
S= (2.11)
2

This quantity also has a clear interpretation. Namely, the spin density (2.11) is proportional to the local ellipticity of the
field polarization, and it is directed along the normal to the polarization ellipse (assuming contributions from both the
electric and magnetic fields). In particular, the spin density (2.11) agrees with Eqs. (2.4)–(2.6) for a circularly-polarized
plane electromagnetic wave. Thus, the spin AM density S is an independent dynamical property of the field, which is related
to the polarization degrees of freedom. Obviously, the spin AM density (2.11) is intrinsic, but its direction with respect to
the wave momentum is not specified in the generic case. This hints that the spin density can have both longitudinal and
transverse components in structured fields.
The total AM density of the field is the sum of the spin and orbital parts: J = S + L. However, the spin and orbital AM
manifest themselves in very different manners in local light–matter interactions [5–20] (see Section 2.4), so that they should
be considered as independent physical properties, corresponding to different degrees of freedom [28–37].
In addition to the energy, momentum, and AM characteristics (2.8)–(2.11), there is one more fundamental quantity, which
is less known. This is the optical helicity (sometimes called optical chirality) density [93,94], which has recently attracted
considerable attention [34,86–89,95–105]. Although in quantum particle physics the helicity is determined by a product of
the spin and momentum, in electromagnetism it is a locally-independent property, which is related to the so-called dual
symmetry between the electric and magnetic field [34,88,95,97,102,105–109]. The helicity density of a monochromatic field
reads

K = −g Im E∗ · H .
 
(2.12)

The energy, momentum, spin, and helicity densities (W , P, S, and K ) are independent quantities, and they form a complete
set of fundamental dynamical properties of light important for our study. (In relativistic problems, one has to add the boost
momentum related to the Lorentz transformations [33,34,110–113].)
Notably, the above dynamical characteristics (2.8)–(2.12) of classical optical fields are perfectly consistent with the
quantum-mechanical approach. Namely, they represent the local expectation values of the corresponding first-quantization
operators, when
  we introduce the wave function proportional to the properly-normalized electric and magnetic fields:
√ E
ψ= g /2 H
[30,34,50,89,114]. In doing so, Eqs. (2.8)–(2.12) can be written as
   
W = ψ Ď · (ω)ψ, P = Re ψ Ď · p̂ ψ , L = Re ψ Ď · L̂ ψ ,
   

   
S = ψ Ď · Ŝ ψ, K = ψ Ď · K̂ ψ. (2.13)

Here different operators act on different degrees of freedom of the wave function ψ . Specifically, the momentum p̂ = −i∇
and orbital AM L̂ = r × p̂ act on the spatial distribution of the wave function ψ(r) (vector in Hilbert space), whereas
 AM operator Ŝ, Eq. (2.5), acts on vector degrees of freedom in the complexified
the spin  3-dimensional
 space (it acts as
0 −1
E∗ · Ŝ E = Im (E∗ × E) on the vectors E and H). In turn, the helicity operator K̂ = −i 1 0
(generator of the dual SO(2)
 
E
transformation) mixes the electric and magnetic degrees of freedom, i.e., it acts on the two-vector H
. In this formalism,
 
the stationary free-space Maxwell equations can be written as K̂ ψ = ψ [89,105,114], which proves the equivalence
Ŝ·p̂
k

of the dual-rotation helicity K̂ and the quantum-particle helicity Ŝ · p̂/k for Maxwell fields.
Alongside the local densities in Eqs. (2.8)–(2.12), we will also use the integral (mean) dynamical characteristics of the
fields, denoted by angular brackets ⟨...⟩. Usually, for quantum particles, this implies 3D integration over the whole space:
  
⟨W ⟩ = W d3 r, ⟨P⟩ = P d3 r, ⟨S⟩ = S d3 r, etc. (2.14)
 K.Y. Bliokh, F. Nori / Physics Reports 592 (2015) 1–38 7

However such integrals are well-defined only for localized fields, such as wave packets. Monochromatic fields (such as optical
beams) cannot be localized in three dimensions, and the spatial integrals (2.14) diverge. Due to this, usually one uses two-
dimensional integrals over the beam cross-section (i.e., the (x, y)-plane for z-propagating beams), which determine the
linear densities of the corresponding quantities per unit z-length of the beam. (It also makes sense to determine the fluxes
of the corresponding quantities through the beam cross-section [37,115].) Therefore, for monochromatic beams, we will
imply integral properties obtained via the integration over the transverse cross-section plane:
  
⟨W ⟩ = W d2 r⊥ , ⟨P⟩ = P d2 r⊥ , ⟨S⟩ = S d2 r⊥ , etc. (2.14a)

Furthermore, for extended periodic field distributions (such as two-wave interference in Section 3.2), the spatial integration
over one period will be implied.

2.3. Longitudinal spin and orbital AM in optical beams

To illustrate the appearance of the above dynamical characteristics in an optical field, let us consider an axially-symmetric
paraxial polarized vortex beam propagating in the z-direction. Its electric and magnetic fields can be written as

x̄ + mȳ
E ≃ A (ρ, z )  exp(ikz + iℓϕ), H ≃ z̄ × E. (2.15)
1 + |m|2
Here m is a complex parameter, which characterizes the polarization [50,54,116], ℓ = 0, ±1, ±2, . . . is the topological
charge of the vortex, and A (ρ, z ) is the complex envelope amplitude of the beam. In the paraxial approximation (2.15),
we neglect the small longitudinal z-components of the electric and magnetic fields (that is why the approximate-equality
symbol ‘‘≃’’ is used). These components become crucial for spin–orbit interaction and transverse-spin phenomena in non-
paraxial fields.
The complex polarization parameter m determines the real normalized Stokes parameters defined as (τ 2 + χ 2 + σ 2 = 1)

1 − |m|2 2 Re m 2 Im m
τ= , χ= , σ = . (2.16)
1 + |m|2 1 + |m|2 1 + |m|2
These three parameters describe the degrees of the x/y linear polarizations, 45°/−45° linear polarizations, and right-
hand/left-hand circular polarizations, respectively (the shapes of the Greek letters τ , χ , and σ resemble horizontal/vertical,
diagonal, and circular polarizations). The third Stokes parameter σ is the helicity parameter, which takes on values ±1 for
circularly-polarized waves, as in Eq. (2.4). Its relation to the helicity density (2.12) is revealed below.
Substituting the field (2.15) into the general equations (2.8)–(2.12), we obtain the energy, momentum, spin AM, orbital
AM, and helicity densities in the paraxial beam:

ℓ
 
W W
W ≃ g |A|2 ω, P≃ k z̄ + ϕ̄ , L≃ (−ρ kϕ̄ + ℓz̄) ,
ω ρ ω

W W
S≃ σ z̄ , K ≃ σ. (2.17)
ω ω
Eqs. (2.17) show a natural and intuitively clear picture of the beam properties, all proportional to the same intensity factor
g |A|2 . First, we note that the beam has an energy density W proportional to the frequency ω and the longitudinal momentum
density Pz proportional to the wave number k = ω/c. Second, the beam carries the longitudinal spin AM density S = σ z̄,
similar to that for the plane wave, Eqs. (2.4)–(2.6) (Fig. 2a). The helicity density K is naturally determined by the helicity
parameter σ , and is in agreement with the longitudinal spin AM and momentum densities: K = S · P/Pz . At the same time,
the momentum has a transverse (with respect to the beam axis) azimuthal component Pϕ produced by the helical phase
of the optical vortex, which generates the longitudinal orbital AM density Lz proportional to ℓ [3–18,21] (Fig. 2b); this is a
signature of the vortex modes, Eq. (2.7).
Integrating the densities (2.17) over the (x, y) cross-section, we determine the integral values (2.14a) of the dynamical
characteristics of the beam:

⟨k⟩
⟨W ⟩ ∝ ω, ⟨P⟩ ∝ ⟨k⟩ , ⟨L⟩ ∝ ℓ ,
k

⟨k⟩
⟨S⟩ ∝ σ , ⟨K ⟩ ∝ σ . (2.18)
k

Here we introduced the mean wave vector, which is ⟨k⟩ ≃ kz̄ in the paraxial approximation, and wrote Eqs. (2.18) in the
form valid for an arbitrary propagation direction of the beam.
8 K.Y. Bliokh, F. Nori / Physics Reports 592 (2015) 1–38

Fig. 2. Longitudinal spin and orbital angular momenta (2.18) and (III), (IV) in a paraxial optical beam (2.15) [1–14]. The beam carries integral momentum
⟨P⟩ determined by its mean wave vector ⟨k⟩ ≃ kz̄. The spin AM ⟨S⟩ is generated by the circular polarization and determined by its helicity parameter
σ , whereas the orbital AM ⟨L⟩ is produced by the helical phase, i.e., optical vortex of charge ℓ. The instantaneous electric and magnetic fields E (z , 0) and
H (z , 0) are shown in (a) for the right-hand circular polarization (parameter m = i, i.e., σ = 1). The constant-phase surface 8 = kz + ℓϕ = 0 is shown in
(b) for the vortex with ℓ = 2.

The transition from the local densities (2.17) to the integral values (2.18) reveals two important peculiarities of the orbital
AM in vortex states. First, the azimuthal momentum Pϕ and AM Lϕ disappear after the integration. As a result, the integral
momentum ⟨P⟩ and the orbital AM ⟨L⟩ become parallel and independent properties! Second, the longitudinal orbital AM
⟨L⟩ becomes intrinsic, i.e., independent of the choice of the coordinate origin: transformation (2.2) does not affect ⟨L⟩ in
Eq. (2.18) [21]. Thus, the momentum-dependent extrinsic transverse local density L = r × P produces an independent intrinsic
longitudinal quantity ⟨L⟩ ∥ ⟨P⟩, Fig. 2. This amazing transmutation of the AM in the transition from its local to integral forms
is a signature of structured fields (vortices in this case), which combine both wave and particle properties.
Summarizing the above properties of the integral spin and orbital AM of a paraxial beam:

Paraxial beam AM: Spin, Intrinsic, Longitudinal. Key parameters: σ , ⟨k⟩. (III)

Paraxial beam AM: Orbital, Intrinsic, Longitudinal. Key parameters: ℓ, ⟨k⟩. (IV)

The vortex charge ℓ has the same P -odd and T -even features as the helicity σ , which ensures the proper T -odd nature of
the intrinsic orbital AM of a vortex beam.
Note that paraxial vortex beams (2.15) with circular polarizations (m = ±i, σ = ±1) are approximate eigenmodes of the
operators p̂z , L̂z , Ŝz , and K̂ with the corresponding eigenvalues k, ℓ, σ , and σ . Such states provide a convenient basis for
performing operations with paraxial photons carrying well-defined momentum, spin, and orbital AM in quantum optics
[11–13].
The proportionality of the spin and orbital AM to the polarization helicity σ and vortex charge ℓ holds true only
in the paraxial approximation. Taking into account the longitudinal field components Ez and Hz in non-paraxial beams
results in spin-to-orbital AM conversion [19,20,31,37,117–123]. This effect originates from the helicity-dependent vortex
structure exp(iσ ϕ) of the longitudinal components of a circularly-polarized beam field, which, in turn, appears due to the
transversality of the Maxwell fields. Introducing the typical convergence angle ϑ of a non-paraxial (focused) vortex beam
 K.Y. Bliokh, F. Nori / Physics Reports 592 (2015) 1–38 9

as cos ϑ = c ⟨Pz ⟩/⟨W ⟩, the integral spin and orbital AM in such beam take the form [31,35,37,121,123]

⟨k⟩ ⟨k⟩
⟨S⟩ ∝ σ cos ϑ , ⟨L⟩ ∝ [ℓ + σ (1 − cos ϑ)] . (2.19)
⟨k⟩ ⟨k⟩
Here we denoted ⟨k⟩ ≡ |⟨k⟩| and implied that the beam is constructed as a superposition of circularly-polarized plane waves
with σ = ±1, and, hence, possesses a well-defined helicity ⟨K ⟩ ∝ σ . Eqs. (2.19) indicate that changing the paraxiality
(focusing the beam) does not change its total AM ⟨J⟩ = ⟨S⟩ + ⟨L⟩ but results in a redistribution between the spin and orbital
parts. Notably, this spin-to-orbital AM conversion can achieve 100% efficiency for ϑ = π /2 [124]. For a detailed analysis
and overview of the spin–orbit interactions in non-paraxial fields we refer the reader to [31,37,123].
We emphasize that the spin and orbital AM (2.19) follow from the same Eqs. (2.9)–(2.11), which are valid for any
monochromatic fields. Furthermore, the canonical expressions (2.9)–(2.11) do not contain the extra surface term, which
affects non-paraxial calculations with the Poynting momentum [26,27,35].
Focused optical beams reveal one more feature of the wave–particle duality. It can be shown from relativistic
considerations that any object carrying intrinsic AM cannot be shrunk to a point and its minimum size is proportional
to its intrinsic AM [112]. The same conclusion occurs for the minimum size of a tightly-focused beam carrying intrinsic
spin and orbital AM [125]. Thus, despite the fact that the integral intrinsic AM is well-defined for optical beams or wave
packets, the finite spatial extension is crucial for it. This also follows from the non-localizability of photons with spin and
the non-commutative character of their covariant coordinates [31,126,127].

2.4. Local measurements of the spin and orbital AM

After we described the main properties of the momentum and angular momenta of free optical fields, it is important
to discuss how these properties can be measured experimentally. Since the integral spin and orbital AM (2.18) in paraxial
beams look very similar, they manifest themselves similarly in experiments with the detector size larger than the beam
cross-section [2,4,11–13,128]. In contrast, local measurements of the spin and orbital AM reveal their densities (2.10) and
(2.11) or (2.17), which have very different properties [16–20,29,34,37,50,92]. This was first demonstrated in remarkable
experiments [16,17], where small probe particles were employed to reveal the local AM properties in paraxial vortex beams,
Fig. 3. It turned out that such a probe particle spins around its axis, proportionally to the local spin AM density Sz (circular
polarization) of the field, and also orbits around the vortex-beam axis proportionally to its orbital AM. The latter motion can
be considered as the local translational motion proportional to the azimuthal canonical momentum Pϕ (2.17) in the vortex
beam. Thus, the spin and orbital degrees of freedom manifest themselves differently in local interactions with matter and
can be separately measured in optical fields. Similar experiments with probe particles were also employed to detect the
spin-to-orbital AM conversion in [19,20].
To understand the results of the experiments [16–20], we now consider the interaction of a small isotropic spherical
particle with a monochromatic optical field. A typical subwavelength (Rayleigh) particle interacts with the field via the
electric-dipole coupling. The particle is characterized by its complex electric polarizability α e , which generates a complex
electric dipole moment π = α e E induced by the electric wave field. Although here we consider the dipole coupling with a
classical particle, similar results take place in quantum interactions with atoms or molecules [24,29,85,129–131].
It is important to note that most of the natural particles interact very weakly with the magnetic field H via the induced
magnetic dipole moment µ = α m H, i.e., have a very small magnetic polarizability α m : |α m | ≪ |α e |. This breaks the dual
(electric–magnetic) symmetry of the free electromagnetic field [105] and makes the probe sensitive to the electric parts of
all quantities [34,89]. Indeed, all the dynamical field properties (2.8)–(2.11), but not the helicity (2.12), naturally represent
sums of the electric and magnetic contributions:
W = W e + W m, P = P e + P m, L = L e + L m, S = S e + S m. (2.20)
Most of the standard laboratory measurements reveal only the electric parts of these quantities. The electric and mag-
netic contributions in Eqs. (2.20) are equivalent in paraxial optical fields, including the beams (2.15)–(2.18), but can dif-
fer significantly in generic structured fields [30,50,54]. It is worth noticing that definitions of the canonical momentum,
spin and orbital AM densities depend on the choice of the Lagrangian of the free electromagnetic field [34]. For instance,
these are
 2‘‘electrically-biased’’,
 P → 2Pe , S → 2Se and L → 2Le , with the standard dual-asymmetric Lagrangian density
2
L ∝ E − H [34,37], while the dual-symmetric definitions (2.9)–(2.11), obtained in the dual-symmetric Lagrangian
formalism [34,111], are the most natural choice for a free field [30,32,34,37].
The energy, momentum, and angular-momentum transfer in light–particle interactions can be quantified by the
absorption rate A, optical force F, and torque T on the particle, respectively. Calculating these quantities in the electric dipole-
coupling approximation (α m ≃ 0), one can derive [50,89,132–134]:
A = g −1 Im α e W e ,
 
(2.21)
 
1
F = g −1 Re α e ∇ W e + Im α e P e ≡ Fgrad + Fpress ,
   
(2.22)
2ω

T = g −1 Im α e Se .
 
(2.23)
10 K.Y. Bliokh, F. Nori / Physics Reports 592 (2015) 1–38

Fig. 3. Local spin AM and transverse-momentum densities and their measurements in the paraxial Bessel beam given by Eq. (2.15) with A (ρ, z ) = J|ℓ| (k⊥ ρ)
and k → kz , where k⊥ = k2 − k2z ≪ k. The right-hand circularly polarized beam (m = i, σ = 1) with a vortex of charge ℓ = 2 is shown via its

energy-density distributions W (ρ) (grayscale plots). (a) The distribution of the transverse circular polarization of the electric field [in cyan] indicates the
longitudinal spin AM density Sze , Eqs. (2.11) and (2.17). (b) The distribution of the transverse (azimuthal) canonical momentum density P⊥e [in orange] from
the helical phase (optical vortex), Eqs. (2.9) and (2.17), produces the corresponding orbital AM density Lez = ρ Pϕe . (c) Experimental measurements from [17].
The spin AM and canonical momentum densities immediately reveal themselves when interacting with a small probe particle. The particle spins due to a
torque proportional to the spin density Sze and orbits (i.e., moves azimuthally) due to the radiation-pressure force proportional to the momentum density
Pϕe , Eqs. (2.22) and (2.23). The radial trapping of the particle at the intensity maximum occurs due to the gradient force in Eq. (2.22). These measurements
clearly show the different intrinsic and extrinsic nature of the spin and orbital AM densities and their separate observability. (For interpretation of the
references to color in this figure legend, the reader is referred to the web version of this article.)

The first and second terms in Eq. (2.22) represent the gradient and radiation-pressure optical forces, respectively.
Eqs. (2.21)–(2.23) show that the absorption rate, radiation-pressure force, and torque on the particle ‘‘measure’’ the energy,
canonical-momentum, and spin-AM densities in the wave field. In particular, Eqs. (2.22) and (2.23) with Eq. (2.17) perfectly
explain the results of the experiments [16,17] (Fig. 3). Namely: (i) the probe particle spins about its center proportionally to
the spin AM density Sze , (ii) it is trapped in the annular field intensity maximum due to the radial gradient force proportional
to ∂ W e /∂ρ , and (iii) it orbits along the circular trajectory due to the radiation-pressure force produced by the azimuthal
canonical momentum Pϕe . Importantly, in local measurements, the orbital AM does not represent an independent degree of
freedom, but it is produced by the canonical momentum density according to Eq. (2.10). Thus, local measurements reveal the
intrinsic nature of the spin AM and locally-extrinsic nature of the orbital AM. The basic optical forces and torques (2.22) and
(2.23) play a key role in numerous problems and applications using mechanical properties of light, such as optical trapping
and manipulation of atoms, molecules, and small particles, as well as in various optomechanical phenomena (see [135–139]
for reviews).
 K.Y. Bliokh, F. Nori / Physics Reports 592 (2015) 1–38 11

It should be emphasized that Eqs. (2.22) and (2.23) support the canonical picture of the optical momentum and
AM, Eqs. (2.9)–(2.11), rather than the kinetic one based on the Poynting vector [34,37,38,50,54,84]. Indeed, the local
momentum transfer in light–matter interactions is determined by the canonical momentum (2.9). For example, the
‘‘supermomentum’’ transfer larger than k = ω/c per photon can be observed near optical-vortex cores and in evanescent
waves with ‘‘superluminal’’ canonical momentum |P|/W > c [30,50,92,131,140,141], while the Poynting vector is always
‘‘subluminal’’: |5| /W < c. Moreover, careful considerations show that a number of other (completely different) methods
of measurements of the local momentum density in optical fields [142–147] results in the same canonical momentum P
(or P e ) rather than the Poynting vector. In particular, measurements of the azimuthal momentum in optical beams reveal
the canonical momentum Pϕ (2.17) due to the optical vortex [142–146], but they cannot measure the azimuthal Poynting
vector Πϕ caused by the circular polarization [14]. In addition, the so-called quantum weak measurements of the photon
momentum density immediately yield the canonical momentum owing to its intimate relation to the momentum operator
p̂ [30,34,92,147].
The helicity density (2.12) does not manifest itself in interaction with spherical particles, but it becomes crucial in the
interaction with chiral particles. The chirality of the particle or a molecule mixes the electric and magnetic interactions, so
that the chiral contributions to the induced electric and magnetic dipole moments can be written as [85,86,89,103]

πch = −i α ch H, µch = i α ch E, (2.24)

where α is the chiral polarizability of the particle. Recently, it was noticed that the absorption-rate term caused by the
ch

chirality of the particle is proportional to the local helicity density (2.12) of the field [86,89]:

Ach = −g −1 ω Im α ch K .
 
(2.25)

In combination with the natural dual asymmetry of molecules, this has resulted in a new method for the enhancement of
the optical circular dichroism and enhanced optical enantioselectivity [86,98–101,148,149].
While the natural circular dichroism due to the particle chirality is coupled to the helicity density of the field, the magnetic
circular dichroism (MCD), induced by the presence of an external constant magnetic field H0 , turns out to be coupled to the
spin AM density (2.11) [150,151]. According to [150] the H0 -induced term in the electric dipole polarizability of an isotropic
particle can be written as

πMCD = i α MCD (E × H0 ) . (2.26)

= ω2 Im πMCD · E∗ , and using Eqs. (2.11) and

 
Calculating the corresponding MCD contribution to the absorption rate, AMCD
(2.20), we arrive at

AMCD = −g −1 ω Im α MCD Se · H0 .
  
(2.27)

A similar coupling of the electric spin density to an external magnetic field also occurs in atomic transitions in light–atom
interactions [152–154].
Thus, the natural circular dichroism and MCD provide efficient methods of non-mechanical local measurements of the
helicity and spin AM densities in optical fields. Note that such correspondence looks natural from spatial-inversion and
time-reversal symmetry arguments. Indeed, the chirality and helicity are both P -odd and T -even scalar properties, while
the magnetic field and spin AM are both P -even and T -odd vector quantities [85,89].

3. Transverse spin angular momenta

In this Section we consider a new kind of angular momentum of light: the transverse spin AM. This angular momentum
exhibits features which are in sharp contrast to the usual longitudinal spin AM in Eqs. (2.6), (2.17), (2.18), and to what is
known about the spin of photons. As we show below, there are different types of transverse spin, which can have properties
dramatically different from (II) and (III). Throughout this Section we do not consider the orbital AM (2.10) because in all
examples here it does not represent an independent degree of freedom.

3.1. Evanescent waves

3.1.1. Transverse (out-of-plane) helicity-independent spin

The first example of the transverse spin in evanescent waves was described by Bliokh and Nori [46,50] and a bit later by
Kim et al. [48,59]. A single evanescent wave is one of the simplest solutions of free-space Maxwell equations [155]. Assuming
that the wave propagates along the z-axis and decays in the x > 0 half-space, it can be represented as a plane wave with
the complex wave vector

k = kz z̄ + iκ x̄, k2 = k2z − κ 2 . (3.1)

12 K.Y. Bliokh, F. Nori / Physics Reports 592 (2015) 1–38

Fig. 4. Evanescent wave (3.2) generated by the total internal reflection of a plane wave and the transverse (out-of-plane) spin AM density (3.5) and (V)
in the evanescent field [46,50]. The evanescent wave is characterized by the complex wave vector (3.1) k and is defined in the x > 0 half-space. The
inset shows the z-evolution of the instantaneous electric and magnetic fields E (r, 0) and H (r, 0) for the simplest linear x-polarization (TM mode with
m = 0, τ = 1). The cycloidal rotation of the electric field within the propagation (x, z )-plane generates the helicity-independent transverse spin AM S⊥ ,
Eqs. (3.4)–(3.7).

The electric and magnetic fields of the evanescent wave with an arbitrary polarization can be written as [50]
κ
 
A0 k k
E=  x̄ + m ȳ − i z̄ exp(ikz z − κ x), H= × E, (3.2)
1 + |m|2 kz kz k
where A0 is a constant field amplitude.
Even though the electric wave field (3.2) has all three components, its polarization degrees of freedom are described
by the same complex polarization parameter m and the corresponding Stokes parameters as in Eqs. (2.15) and (2.16).
Therefore, we will call the m = 0 and m = ∞ waves with τ = ±1 as ‘‘linearly polarized’’, the m = ±1 waves with
χ = ±1 as ‘‘diagonally polarized’’, and the m = ±i waves with σ = ±1 as ‘‘circularly polarized’’, even though the actual 3D
polarization vector can be rather nontrivial [50]. The evanescent wave (3.2) propagates along the z-axis with ‘‘anomalous’’
wave momentum kz > k, and it decays exponentially in the positive x-direction. Such waves can be generated, e.g., in
total internal reflection at the glass–air interface x = 0 [155], as shown in Fig. 4. Linearly-polarized evanescent waves with
m = 0 (TM mode) or m = ∞ (TE mode) are also parts of surface electromagnetic waves at the x = 0 interface between two
optical media [56,156]: e.g., surface plasmon-polaritons at the metal–air interface [157]. Thus, the wave (3.2) is physically
meaningful and well-defined only in the half-space x > 0.
Most importantly, the imaginary component of the wave vector (3.1) and the plane-wave transversality condition
∇ · E = k · E = 0 result in the ‘‘imaginary’’ longitudinal component Ez ∝ −i κ/kz in the field (3.2). This component,
together with the ‘‘real’’ component Ex , produces a rotation of the electric field (i.e., effective elliptical polarization) in the
(x, z ) plane of the wave propagation [46,50]. Since the field also propagates along the z-axis, the electric field follows a
cycloidal trajectory, like a point on a moving and spinning wheel, see Fig. 4 [49,50,158]. This is in contrast to the helical
trajectory in the case of the usual circular polarization and the longitudinal spin, Figs. 1b and 2a. Nonetheless, according
to Eq. (2.11), the elliptical polarization in the (x, z )-plane implies the spin AM density directed along the transverse y-axis,
i.e., orthogonally to the momentum and wave vector of the evanescent field, Figs. 4 and 5.
To quantify the main properties of the evanescent wave, we substitute the field (3.2) into the general
equations (2.8)–(2.12). Thus, we obtain the energy, momentum, spin AM, and helicity densities:
W
W = g |A0 |2 ω exp(−2κ x), P= kz z̄, (3.3)
ω
 
W k κ W
S= σ z̄ + ȳ , K = σ. (3.4)
ω kz kz ω
The energy and momentum densities (3.3) show intuitively-clear values proportional to ω and Re k = kz z̄, and decaying with
x together with the wave intensity, Fig. 5b. The ‘‘superluminal’’ canonical momentum (3.3) with kz > ω/c is in contrast to
the ‘‘subluminal’’ Poynting vector [46,92], and it was measured in light–atom interactions in [140,141]. The spin AM density
 K.Y. Bliokh, F. Nori / Physics Reports 592 (2015) 1–38 13

Fig. 5. The energy, momentum, and transverse spin AM densities (3.3)–(3.7) in the evanescent wave (3.2) with kz /k = 1.1 and linear TM polarization
(m = 0, τ = 1, see Fig. 4). The energy density W (x) is shown as the grayscale background distribution. (a) Akin to Fig. 3, the transverse spin AM density
(3.6) Sye [normalized by W (x)] is shown in the form of a polarization-ellipse distribution in the propagation (x, z )-plane. It exerts the transverse optical
torque (2.23) Ty on a probe particle [50,51,160]. (b) The longitudinal momentum density is naturally proportional to the wave number kz , and it exerts the
radiation-pressure force (2.22) Fz .

in Eq. (3.4) contains the usual longitudinal component Sz , which is similar to the spin AM of propagating waves (2.6) and
(2.17). It is directed along the wave momentum Re k and is proportional to the helicity parameter σ . In addition, the helicity
density K in Eq. (3.4) is also similar to that in a propagating plane wave, Eq. (2.17).
The only extraordinary term in the above equations is the transverse spin AM density Sy (shown in the red frame), Figs. 4
and 5a. We write it separately in the form valid for an arbitrary propagation direction of the evanescent wave:

W Re k × Im k
S⊥ = . (3.5)
ω (Re k)2
In contrast to the usual longitudinal spin considered in Section 2, the transverse spin (3.5) is completely independent of the
helicity and other polarization parameters. Instead, it is solely determined by the complex wave vector of the evanescent
wave and is directed out of the plane formed by its real and imaginary parts. Obviously, as any spin AM, it has an intrinsic
nature. Importantly, the imaginary part of the wave vector, Im k, is a P -odd but T -even quantity (because the time-reversal
transformation implies complex conjugation for all complex quantities [89,159]), so that the transverse spin (3.5) has the
proper T -odd and P -even symmetry. Summarizing the unusual properties of the transverse spin AM (3.5):

Evanescent wave: Spin, Transverse (out-of-plane), Key parameters: Re k, Im k. (V)

Remarkably, the crucial dependence of the transverse spin (3.5) on the wave-vector (momentum) is intimately related to
the fundamental topological properties of Maxwell equations [56], and it has been employed for transverse spin–direction
interfaces with evanescent waves [52,53,57,74–83] (see Section 3.3.5). According to the general Eqs. (2.22) and (2.23), the
longitudinal momentum and transverse spin AM in evanescent waves can be directly measured via the local interaction
with a probe particle. In particular, the transverse spin in Eq. (3.4) exerts the corresponding helicity-independent torque
Ty ∝ Sye [50,51]. This is schematically shown in Fig. 5, and in Section 3.3 we consider the main experiments involving the
transverse spin.
The fact that the transverse spin AM of evanescent waves was not discussed before 2012 is a remarkable example of
preconception and the lack of communication between different physical communities. On the one hand, the imaginary
longitudinal field component and the (x, z )-plane rotation of the electric field in evanescent waves were known for
decades in optics and plasmonics (see, e.g., [155,157,161]). Furthermore, the in-plane elliptical polarization was measured
in experiments [153,162]. On the other hand, knowing the direct connection between the elliptical polarization and spin
AM since Poynting and Beth [1,2], the optical AM community has never considered linearly-polarized evanescent waves
(e.g., m = 0 TM-modes in the case of surface plasmon-polaritons) as possible candidates for spin AM studies. There were
even numerical simulations showing the transverse helicity-independent torque on a probe particle in the evanescent
field [160], but this torque was not connected to the effective elliptical in-plane polarization, and the key conclusion about
the presence of the transverse spin AM was not made until [46,50].

3.1.2. Transverse (in-plane) dual-antisymmetric spin

In contrast to the usual longitudinal spin AM in circularly-polarized paraxial waves, the transverse spin has asymmetric
electric and magnetic properties. For instance, in the TM-polarized wave (3.2) with m = 0, τ = 1, only the electric field rotates
14 K.Y. Bliokh, F. Nori / Physics Reports 592 (2015) 1–38

Fig. 6. The dual-antisymmetric transverse (in-plane) spin AM, Eqs. (3.6), (3.7), and (VI), in a diagonally-polarized evanescent wave (3.2) [50], cf., Fig. 4. A
diagonal polarization with m = −1 and χ = −1 is shown here, and the inset displays the z-evolution of the instantaneous electric and magnetic fields
E (r, 0) and H (r, 0). In addition to the transverse (out-of-plane) spin AM Sy shown in Fig. 4, the diagonal polarization produces χ -dependent anti-parallel
vertical (in-plane) electric and magnetic spin AM Sxe = −Sxm , Eqs. (3.6) and (3.7). These oppositely-directed electric and magnetic spins are generated by
opposite (y, z )-plane rotations of the electric and magnetic fields in a diagonally-polarized evanescent wave.

in the (x, z )-plane, while the magnetic field has only the y-component, as shown in Fig. 4. In this case, the transverse spin
AM has a purely electric origin: S = Se , Sm = 0, see Eqs. (2.11) and (2.20). If the wave has a linear y (TE) polarization with
m = ∞, τ = −1, then the situation becomes the opposite: the magnetic field rotates in the (x, z )-plane, the electric field
has only a y-component, and the transverse spin is purely magnetic: S = Sm , Se = 0 [50]. The latter feature was mistakenly
interpreted in [48] as the absence of the transverse spin AM in the TE-mode.
For a generic polarization, the electric and magnetic parts of the spin AM density (2.11) and (2.20) in the evanescent wave
(3.2) are [50]:
 
W k κ κk
Se,m = σ z̄ + (1 ± τ ) ȳ ±χ x̄ . (3.6)
2ω kz kz k2z

Here the transverse components Sye,m show the above electric–magnetic asymmetry, which is controlled by the first Stokes
parameter τ . Moreover, Eq. (3.6) reveals a new, vertical term Sxe,m (shown in the green frame), which is directed along the
wave inhomogeneity, Im k, see Fig. 6. It is also transverse with respect to the wave propagation direction Re k, but lies in
plane with the complex wave vector k. This ‘‘dual-antisymmetric transverse (in-plane) spin AM’’ has the opposite electric and
magnetic contributions: Sxe = −Sxm , so that Sx = 0 [50]. Nonetheless, the vertical electric spin Sxe naturally contributes to
light–matter interactions sensitive to the electric rather than magnetic parts of the fundamental quantities, see Section 2.4.
The vertical spin AM Sxe,m is caused by the opposite rotations of the electric and magnetic fields projected on the (y, z ) plane
for the waves polarized diagonally at ±45° (Fig. 6). Therefore, it is proportional to the second Stokes parameter χ .
Representing Eq. (3.6) in the general vector form based on the wave vector (3.1) yields
 
e,m W k Re k Re k × Im k k Im k
S = σ + (1 ± τ ) ±χ . (3.7)
2ω (Re k)2 (Re k)2 (Re k)2
Here all the three terms must have the same T -odd and P -even character of the AM. Therefore, we conclude that the Stokes
parameter τ is a P -even and T -even quantity, while the Stokes parameter χ is a P -odd and T -odd quantity. Summarizing
the properties of the ‘‘dual-antisymmetric transverse spin AM’’ in Eq. (3.7):

Evanescent wave AM: Spin, Transverse (in-plane), Anti-dual. Key parameters: χ , Im k. (VI)
In this Subsection we analyzed only local momentum and AM densities. The reason for this is that the evanescent wave
itself is defined only in a half-space. Since the wave (3.2) is homogeneous in the y and z directions, it only makes sense to
calculate integral quantities (2.14) with the integration over the x > 0 semi-axis. This integration preserves all features of the
x-dependent densities and can be symbolically written as
1 1 1 1
⟨W ⟩+ = W (0) , ⟨P⟩+ = P (0) , ⟨S⟩+ = S (0) , ⟨K ⟩+ = K (0) , etc. (3.8)
2κ 2κ 2κ 2κ
 K.Y. Bliokh, F. Nori / Physics Reports 592 (2015) 1–38 15

Here the ‘‘+’’ superscript stands for the positive-x integration and the ‘‘0’’ arguments indicate the values at x = 0.

3.2. Propagating waves

The transverse spin AM is not an exclusive feature of evanescent waves. A very similar transverse spin AM density also
appears in basic configurations with freely propagating fields. The necessary condition of its appearance is the transverse
inhomogeneity of the field intensity: in all cases the transverse-spin density is accompanied by transverse intensity gradients.
It is these gradients that generate the ‘‘imaginary’’ longitudinal field components.

3.2.1. Two-wave interference

The simplest propagating-wave configuration with inhomogeneity is the interference of two plane waves propagating at
an angle 2γ between their wave vectors. Below, we basically follow the analysis of the recent work [54]. It is convenient to
choose the z-axis along the mean direction of propagation, and the (x, z )-plane as the propagation plane formed by the two
wave vectors (see Fig. 7):
k1,2 = k cos γ z̄ ± k sin γ x̄ ≡ kz z̄ ± kx x̄. (3.9)
Then, the electric and magnetic fields of the two plane waves can be written as [cf., Eqs. (2.4) and (2.15)]
 
A0 kz kx k1,2
E 1 ,2 =  x̄ + m1,2 ȳ ∓ z̄ exp(i81,2 ), H1,2 = × E 1 ,2 , (3.10)
 2 k k k
1 + m1,2 

where 81,2 = kz z ± kx x are the wave phases, and we assume that the two waves have equal electric-field ampli-
tudes A0 .
The resulting interference field is E = E1 + E2 and H = H1 + H2 , and the interference picture is characterized by the
relative x-dependent phase δ 8 = 81 − 82 = 2kx x. For the sake of simplicity, we first consider equal polarizations of the
two waves and the corresponding Stokes parameters, Eq. (2.16):
m1 = m2 ≡ m, (τ1 , χ1 , σ1 ) = (τ2 , χ2 , σ2 ) ≡ (τ , χ , σ ) . (3.11)
In this case the total wave electric field becomes
δ8 δ8 δ8
 
2A0 kz kx
E=  cos x̄ + m cos ȳ − i sin z̄ exp(ikz z ). (3.12)
 2 k 2 2 k 2
1 + m1,2 

Importantly, this field has the ‘‘imaginary’’ longitudinal component Ez ∝ −i kkx sin δ28 , which is quite similar to the longitudinal
component in the evanescent wave (3.2), but here it oscillates and changes sign across the interference pattern.
Substituting Eqs. (3.9)–(3.12) into the general equations (2.8)–(2.12) and (2.20), we obtain the energy, momentum, spin
AM, and helicity densities in the two-wave interference field:

k2z
 
W
W = 2g |A0 |2 ω 1 + cos δ 8 , P= kz z̄, (3.13)
k2 ω
 
kz kx kz
S = 2g |A0 | 2
σ (1 + cos δ 8) z̄ + sin δ 8 ȳ , (3.14a)
k k2
 
kz kx kz kx
Se,m = g |A0 |2 σ (1 + cos δ 8) z̄ + (1 ± τ ) sin δ 8 ȳ ∓χ sin δ 8 x̄ , (3.14b)
k k2 k
W
K = σ. (3.15)
ω
Here, akin to Eq. (3.6), we separated the electric and magnetic contributions (2.20) in the spin AM density (3.14), S = Se + Sm ,
because they show the electric–magnetic asymmetry of the transverse spin.
Eqs. (3.13)–(3.15) exhibit basic features similar to those of an evanescent wave, Eqs. (3.3), (3.4), and (3.6). Namely, the
canonical momentum density P is proportional to the energy density W and the mean wave vector ⟨k⟩ = (k1 + k2 ) /2 = kz z̄
(Fig. 8b). The spin AM density (3.14) contains the usual longitudinal term Sz proportional to the helicity parameter σ and
varying together with the energy density W and the helicity density K , Eq. (3.15). In addition, there are also two transverse
terms in Eq. (3.14), which are orthogonal to the field momentum P.
First, this is the transverse (out-of-plane) helicity-independent spin AM density Sy (shown in the red frame), which shares all
the main features of its evanescent-wave counterpart in Eqs. (3.4)–(3.7) (see Figs. 7 and 8a). Namely, it is orthogonal to the
momentum and the wave vectors k1,2 , independent of the helicity and polarization, and is determined solely by the wave-
vector (momentum) parameters. Second, akin to Eq. (3.6), the spin AM density (3.14) exhibits strong electric–magnetic
16 K.Y. Bliokh, F. Nori / Physics Reports 592 (2015) 1–38

Fig. 7. Interference of two propagating plane waves (3.9)–(3.12) and the transverse (out-of-plane) spin AM density (3.14) or (3.16) [54], cf., Fig. 4. The
simplest case of linear in-plane polarization m = 0, τ = 1 is depicted. The inset panels display the z-evolutions of the instantaneous electric and magnetic
fields E (r, 0) and H (r, 0) for different x-points of the interference pattern (marked by the phase difference δ 8 between the waves). The cycloidal rotations
of the electric field in the propagation (x, z )-plane generate the helicity-independent transverse spin AM density (3.14) and (3.16) S⊥ . It has the properties
listed in (VII) and varies sinusoidally across the interference pattern.

asymmetry, and the ‘‘dual-antisymmetric transverse (in-plane) spin AM’’ densities Sxe = −Sxm appear (shown in the green
frame). These densities are directed along the wave inhomogeneity and are controlled by the second Stokes parameter χ ,
exactly as described in Section 3.1.2 for evanescent waves.
The main difference between the transverse spin AM densities in the evanescent wave (3.6) and in the two-wave
interference (3.14) is that the latter transverse spin oscillates and changes its sign across the interference pattern: Sxe,,ym ∝
e,m
sin δ 8 = sin (2kx x). Therefore, the integral (i.e., the δ 8-averaged) values of these transverse spins vanish: S⊥
 
= 0.
Nonetheless, locally they exist, and appear from the similar rotations of the field as in the evanescent wave. Fig. 7 shows the
 K.Y. Bliokh, F. Nori / Physics Reports 592 (2015) 1–38 17

Fig. 8. The energy, momentum, and transverse spin AM densities (3.13) and (3.14) in the two-wave interference field (3.9)–(3.12) with γ = 30° and
linear-polarization (m = 0, τ = 1, see Fig. 7). The energy density W (x) is shown as the grayscale background distribution. (a) The transverse spin AM
density (3.14) Sye (x) [normalized by W (x)] is shown in the form of a polarization-ellipse distribution in the propagation (x, z )-plane. It exerts the local
transverse torque (2.23) Ty on a probe particle [54]. (b) The longitudinal momentum density P is naturally proportional to the mean wave vector ⟨k⟩ = kz z̄
and exerts the radiation-pressure force (2.22) Fz .

instantaneous distributions of the real electric and magnetic fields in two-wave interference with the simplest linear TM
polarization: m = 0, τ = 1. One can see that the transverse (out-of plane) spin Sy appears on the slopes of the interference
picture due to the cycloidal rotation of the electric field in the propagation (x, z )-plane, see Eq. (3.12). Fig. 8 shows the
(x, z )-plane distributions of the electric-field polarization underlying the transverse spin density Sy = Sye , as well as the
distribution of the canonical momentum P. According to the general equations in Section 2.4, these quantities can be directly
measured using local light–matter interactions: e.g., via the optical torque (2.23) and radiation-pressure force (2.22) on a
probe particle.
Using the mean wave vector ⟨k⟩ and the wave-vector difference δ k = k1 − k2 = 2kx x̄, we write the spin AM density
(3.14) in the general vector form:
 
⟨k⟩ ⟨k⟩ × δ k
S = 2g |A0 | 2
σ (1 + cos δ 8) + sin δ 8 , (3.16a)
k 2k2
 
⟨k⟩ ⟨k⟩ × δ k δk
Se,m = g |A0 |2 σ (1 + cos δ 8) + (1 ± τ ) sin δ 8 ∓χ sin δ 8 . (3.16b)
k 2k2 2k

One can note the remarkable similarity between Eqs. (3.16) and (3.5), (3.7). This indicates that both types of the transverse
spin mentioned above are rather generic phenomena. However, instead of the imaginary wave vector Im k in the evanescent-
wave spin (3.5) and (3.7), the transverse terms in the two-wave spin (3.16) contain the real wave-vector difference δ k and
the phase difference δ 8. Since the phase is a P -even and T -odd quantity, the product δ 8δ k has the same P -odd and
T -even properties as the imaginary momentum Im k in the evanescent field. This ensures that all the terms in Eq. (3.16)
have the proper AM symmetries.
Akin to the properties (V) and (VI), we now list the main features of the transverse spin AM densities in the interference
field:

Interfering waves AM: Spin, Transverse (out-of-plane) , Local. Key parameters: ⟨k⟩, δ k, δ 8. (VII)

Interfering waves AM: Spin, Transverse (in-plane), Local, Anti-dual. Key parameters: χ , δ k, δ 8. (VIII)

3.2.2. Integral transverse (in-plane) helicity-dependent spin

Alongside the local densities (3.13)–(3.16), we examine the integral dynamical properties of the two-wave interference
field. Here, by integral we imply the natural averaging of the densities over the interference pattern, i.e.: ⟨...⟩ ≡
 2π
1
2π 0
... d (δ 8). We now abandon the condition (3.11) and consider the generic case of different polarizations m1 and
m2 with their corresponding Stokes parameters (2.16). Performing straightforward calculations with the fields (3.10) and
18 K.Y. Bliokh, F. Nori / Physics Reports 592 (2015) 1–38

Fig. 9. Interference of two plane waves (3.9) and (3.10) with opposite circular polarizations (m1 = −m2 = i, σ1 = −σ2 = 1 here) and the transverse
(in-plane) helicity-dependent spin AM (3.18), (3.20), and (IX) [49,54]. The integral (δ 8-averaged) momentum ⟨P⟩ and spin AM ⟨S⟩ are obtained by simple
summation of the single-wave momenta k1 and k2 and spins S1 = σ1 k1 /k and S2 = σ2 k2 /k. The z-evolutions of the instantaneous electric and magnetic
fields E (r, 0) and H (r, 0) in different x-points (marked by the δ 8 values) exhibit the cycloidal rotations of both fields in the (y, z )-plane (cf., Fig. 7).

general equations (2.8)–(2.12), we derive (omitting the common g |A0 |2 factor):

⟨W ⟩ ∝ 2ω, ⟨P⟩ ∝ 2kz z̄ = k1 + k2 , (3.17)
 
kx kz k1 k2
⟨S⟩ ∝ (σ1 − σ2 ) x̄ + (σ1 + σ2 ) z̄ = σ1 + σ2 , (3.18)
k k k k

⟨K ⟩ ∝ σ1 + σ2 . (3.19)
These equations reveal several interesting features. First, we emphasize that the electric and magnetic contributions to
Eqs. (3.17) and (3.18) are equivalent: ⟨W e ⟩ = ⟨W m ⟩, ⟨P e ⟩ = ⟨P m ⟩ , and ⟨Se ⟩ = ⟨Sm ⟩. In other words, the integral
quantities become dual-symmetric in propagating fields [32]. Second, the transverse spin AM densities of Eqs. (3.14) and
(3.16) disappear in the integral spin (3.18): ⟨S⊥ ⟩ = 0 for σ1 = σ2 = σ . Finally, one can notice that the values (3.17)–(3.19)
precisely correspond to the sum of two single-photon energies ω, momenta k1 and k2 , spin AM S1 = σ1 k1 and S2 = σ2 k2 , as
k k

well as helicities σ1 and σ2 . Thus, the δ 8-averaging eliminates the fine interference features in the dynamical properties of
the field and makes them additive.
Despite the simplicity of Eqs. (3.17)–(3.19), they offer another type of transverse spin AM in Eq. (3.18) (shown in the blue
frame). Introducing the mean helicity ⟨σ ⟩ = (σ1 + σ2 ) /2 and the helicity difference δσ = σ1 − σ2 , we represent Eq. (3.18)
as

δk ⟨k⟩
⟨S⟩ ∝ δσ + 2 ⟨σ ⟩ . (3.20)
2k k
In the case of opposite helicities of the two interfering waves, σ1 = −σ2 ≡ σ , the mean helicity vanishes, ⟨K ⟩ = ⟨σ ⟩ = 0,
and the integral spin AM (3.20) becomes purely transverse with respect to the momentum ⟨P⟩ ∝ ⟨k⟩, but in-plane with
respect to the two wave vectors: ⟨S⟩ ∝ σ δ k (see Fig. 9). The idea of the transverse AM according to Eqs. (3.18) and (3.20)
was first suggested by Banzer et al. in [49]. Fig. 9 illustrates the appearance of the integral in-plane transverse spin AM (3.18)
and (3.20) in the interference of circularly-polarized waves with σ1 = −σ2 = 1. One can see that this spin AM is generated
by the cycloidal rotation of both the electric and magnetic fields in the (y, z ) plane.
We emphasize that the integral transverse (in-plane) spin AM (3.18) and (3.20) is crucially determined by the helicities of
the interfering waves. Unlike the previous transverse spins in Sections 3.1 and 3.2.1, it cannot appear in the interference
of linearly-polarized waves with σ1 = σ2 = 0. Rather it represents a transversely-directed perturbation of the usual
longitudinal spin AM (2.6) and (2.18), where σ and k are substituted by their variations. Summarizing the properties of
this third type of the transverse spin:

Interfering waves AM: Spin, Transverse (in-plane), Integral. Key parameters: δσ , δ k. (IX)
 K.Y. Bliokh, F. Nori / Physics Reports 592 (2015) 1–38 19

Fig. 10. Focused Gaussian beam (3.23) and the transverse (out-of-plane) spin AM density (3.25) and (3.26) near its focus [55,150,151]. The case of linear
x-polarization (m = 0, τ = 1) is shown. The transverse spin AM density S⊥ near the focus originates from the interference of plane waves forming the
beam, cf., Fig. 7. The insets show the z-evolutions of the instantaneous electric and magnetic fields, E (r, 0) and H (r, 0), in the x > 0 and x < 0 halves
of the beam near the beam focus. The opposite cycloidal rotations of the electric field generate the opposite-sign transverse spin AM density (3.25) Sye in
the upper and lower halves of the beam. The total (electric plus magnetic) transverse spin density (3.26) is independent of the beam polarization and is
directed azimuthally: S⊥ ∝ ϕ̄.

Here we have emphasized its non-vanishing integral character; but, of course, this spin also appears in local spin AM
densities [54].
The Eqs. (V) and (VII), (VI) and (VIII), as well as (IX), and the corresponding red-framed, green-framed, and blue-framed
terms in the other equations describe three distinct types of the transverse spin AM in optical fields. These types differ in their
nature, direction with respect to the wave vectors, and dependences on the polarization and wave-vector parameters. Their
simple forms and appearance in very basic optical fields suggest that these three types of the transverse spin AM have a
universal and robust nature. Therefore, it is natural to expect the presence of similar transverse spin densities in a variety
of more complicated structured fields.

3.2.3. Focused Gaussian beam

The interference between two plane waves can serve as a toy planar model for focused (non-paraxial) beams, which consist
of multiple plane waves propagating in different directions. We consider the simplest case of a focused polarized Gaussian
beam. The transverse electric and magnetic fields of such beam, E⊥ and H⊥ , can be taken from Eqs. (2.15) for paraxial beams,
with ℓ = 0 and the Gaussian envelope

ρ2
 
zR
A (ρ, z ) = A0 exp ik . (3.21)
q (z ) 2q (z )
Here q (z ) = z − i zR is the complex beam parameter, zR = kw02 /2 is the Rayleigh diffraction length, and w0 is the beam
waist [163]. In contrast to the paraxial approximation (2.15), we need to take into account non-zero longitudinal fields
Ez and Hz . In the first post-paraxial approximation (assuming kzR ≫ 1) these components can be determined from the
transversality conditions ∇ · E = ∇ · H = 0 as
ρ ρ
Ez ≃ ik−1 ∇⊥ · E⊥ = − Eρ , Hz ≃ − Hρ , (3.22)
q (z ) q (z )
where Eρ and Hρ are the radial field components.
Combining the transverse field components (2.15) with the longitudinal fields (3.22), we obtain the full 3D field of the
focused Gaussian beam:
x+my y−mx
x̄ + mȳ − q(z )
z̄ ȳ − mx̄ − q(z )
z̄
E=  A (ρ, z ) eikz , H=  A (ρ, z ) eikz . (3.23)
1 + |m|2 1 + |m|2
In the focal plane z = 0, q = −i zR , and fields (3.23) acquire ‘‘imaginary’’ longitudinal components, which generate in-plane
cycloidal rotations of the fields and the transverse spin AM density [55,150,151], Figs. 10 and 11.
20 K.Y. Bliokh, F. Nori / Physics Reports 592 (2015) 1–38

Fig. 11. The energy, momentum, and transverse spin AM densities in the focused Gaussian beam (3.23) with kzR = 5 and linear x-polarization (m = 0,
see Fig. 10). The energy density W (x) is shown as the grayscale background distribution. (a) The transverse spin AM density (3.25) Sye (x) [normalized by
W (x)] is shown in the form of the electric-field polarization distribution in the (x, z )-plane [55,150,151]. It exerts the local transverse torque (2.23) Ty on
a probe particle. (b) The longitudinal momentum density (2.9) P has the main longitudinal component proportional to k, a small radial component due to
the diffraction, and it exerts the radiation-pressure force (2.22) Fz .

Akin to the two-wave interference in Figs. 7 and 8, we consider the simplest x-linear polarization with m = 0, τ = 1.
The small longitudinal field components Ez and Hz make only second-order contributions to the energy, momentum, and
helicity densities in the beam. Therefore, these are described by the paraxial equations (2.17):
W W
W ≃ g |A (ρ, z )|2 ω, P≃ k z̄, K ≃ σ = 0. (3.24)
ω ω
In contrast, the spin AM density (2.11) involves the first-order products between the longitudinal and transverse field
components, which generate the transverse spin AM densities of both electric and magnetic origins:

W x zR W y zR
Se = ȳ , Sm = − x̄ . (3.25)
ω z2 + zR2 ω z2 + zR2
These helicity-independent transverse spin AM densities are entirely similar to those considered in the two-wave
interference [the red-framed terms in Eqs. (3.14) and (3.16)]. For instance, considering the y = 0 cross-section of the beam
(Fig. 10), we see that the transverse (out-of-plane) spin AM density Sy = Sye is of purely electric origin and appears due to
the interference of plane waves forming the beam in the (x, z ) plane. This spin density is maximal in the focal plane z = 0
and has opposite signs in the x > 0 and x < 0 halves of the beam, exactly like the opposite spin densities on the two
slopes of an interference fringe in Figs. 7 and 8. The transverse spin AM (3.25) is characterized by properties (VII) of the
transverse (out-of-plane) spin in interfering waves. Indeed, the inverse Rayleigh-length parameter w0 /zR is proportional to
the variations in the transverse wave-vector components in the beam spectrum, δ kx,y , whereas the dimensionless x/w0 and
y/w0 distances from the beam axis underpin the transverse phases δ 8x,y in the wave interference (constructive interference
on the beam axis and a destructive one at infinity). Fig. 11 shows the (x, z ) distributions of the in-plane polarization ellipses
(with their ellipticity proportional to the transverse spin AM density Sye ) and momentum density P in the focused Gaussian
beam (3.23). The presence of the in-plane elliptical polarizations near the focus of such beams was recently emphasized by
Yang and Cohen [150] and measured in [55,151] (see Section 3.3.4).
Note that the total (electric plus magnetic) transverse spin (3.25) is directed azimuthally, i.e., orthogonally to the radial
wave-vector distribution. This polarization-independent (for uniformly-polarized beams) spin AM can be written as

W ρ zR
S⊥ = ϕ̄ . (3.26)
ω z 2 + zR2
Thus, a focused beam exhibits a transverse-spin vortex in its focal plane (Fig. 10). Its direction is determined solely by the radial
inhomogeneity and propagation direction of the beam (cf., the edge transverse spin in Section 3.3.5 and [56]). Obviously,
the transverse spin (3.26) does not contribute to the integral spin AM: ⟨S⊥ ⟩ = 0.

3.3. Measurements and applications

We are now in a position to describe the main experimental measurements involving the transverse spin AM. The
transverse spin densities were recently detected in both evanescent and propagating fields. Moreover, due to its unusual
 K.Y. Bliokh, F. Nori / Physics Reports 592 (2015) 1–38 21

Fig. 12. (a) Reconstruction of the (x, z )-polarization of the electric field E(r) [shown by arrows] of a standing evanescent wave via probe-particle
scattering [162]. The background color-scale plot depicts the electric energy density W e (r). The rotation of the linear polarization with z is a signature
of elliptical polarizations in the counter-propagating evanescent waves forming the standing wave, cf. Figs. 4 and 5a. (b) Reconstruction, via a near-field
probe, of the (y, z )-components of the instantaneous magnetic wave field H (r, 0) [shown by arrows] in a z-propagating azimuthally-polarized Bessel
beam [166]. The background color-scale plot shows the x-component of the instantaneous electric field E (r, 0). Note the opposite rotations of the magnetic
field in the y > 0 and y < 0 halves of the beam, which indicate the transverse magnetic spin AM density Sxm ∝ y (a radial polarization would result in the
analogous electric spin AM density, cf. Figs. 10, 11a and Fig. 14).

properties (V), the transverse (out-of-plane) spin AM of evanescent waves, Eq. (3.5), has been employed for the robust spin-
dependent transport of light.

3.3.1. Reconstructions of 3D fields

First of all, the elliptical in-plane polarization in evanescent waves has been known for a long time in the literature; it
can be found in textbooks and reviews on evanescent waves and plasmonics (see, e.g., [155,157,161]). In the past decade,
rapid progress in nano-optics stimulated the development of several methods allowing the probing and reconstruction of
full 3D polarization distributions in structured optical fields [162,164–167]. The full 3D polarization at a given point of a
structured optical field can be reconstructed using local subwavelength probes: e.g., a small nanoparticle scatterer [162,
167] or a near-field tip [164–166]. In this manner, the elliptical in-plane polarization in evanescent waves was measured
in [162], Fig. 12a, as well as similar polarizations in non-paraxial (focused) beams in [166], Fig. 12b. Note that the rotating
in-plane magnetic field in Fig. 12b corresponds to the transverse magnetic spin AM, Sxm ∝ +y or S⊥ ∝ −ρ ϕ̄, i.e., of opposite
sign as compared with Eqs. (3.25) and (3.26). This is because of the non-uniform azimuthal polarization in the beam, which
results in an additional π phase in the interference of pairs of plane waves in the beam (cf., Figs. 7 and 10). Obviously, such
full reconstructions of the electric and magnetic wave fields allow retrieving all the properties of these fields (including
momentum, spin, etc.), so that probing the 3D field can be used as an indirect measurement of the dynamical properties of
the field.

3.3.2. Optomechanical calculations and measurements

A direct detection of the dynamical field characteristics involves optomechanical methods. The straightforward
mechanical detection of the spin AM density employs small probe particles and an optical torque acting on these (see
Fig. 3) [2,16,17,50,54,134]. Interestingly, already in 1998 Chang and Lee [160] calculated optical torques on a spherical
particle in an evanescent field and found an unusual transverse torque Ty , which was independent of the wave helicity.
However, they interpreted this torque as originating from the vertical x-gradient of the longitudinal z-directed optical
pressure. In fact, this was the torque (2.23) originating from the transverse spin AM density Sye , Eqs. (3.4)–(3.7). One can
show that for small subwavelength particles of radius a, ka ≪ 1, the gradient-pressure torque (suggested by Chang and Lee)
is much smaller than the transverse-spin torque. Complete calculations of all the three components of the optical torque
T acting on a spherical particle in an evanescent wave (3.2) were made in [50], see Figs. 13a, b. In perfect agreement with
Eqs. (2.23) and (3.6) or (3.7), these calculations clearly show: (i) the longitudinal σ -dependent torque Tz due to the usual
spin AM Sze ; (ii) the τ -dependent (but helicity-independent) transverse torque Ty , originating from the electric transverse
(out-of plane) spin Sye ; and (iii) the vertical χ -dependent torque Tx produced by the transverse (in-plane) electric spin AM
density Sxe (despite the vanishing total vertical spin (3.4) Sx = 0). Importantly, the dependences of these torques on the
polarization Stokes parameters hold true exactly even for larger Mie particles with ka > 1, i.e., beyond the range of validity
of the dipole-approximation Eq. (2.23). This confirms that the spin AM explanation of the mechanical action of light is robust
and fundamental.
22 K.Y. Bliokh, F. Nori / Physics Reports 592 (2015) 1–38

Fig. 13. Optical torques for a particle immersed in an evanescent field. (a) Schematics of an experiment with the evanescent field (3.2) generated by the
total internal reflection and three components of the torque T indicated by rotational directions. (b) Numerical calculations [50] of the three components
of the torque T on a gold spherical particle of radius a for six basic polarizations of the evanescent wave, which are described by the Stokes parameters
(τ , χ, σ ) = ±1. The σ -dependent longitudinal torque Tz , τ -dependent transverse torque Ty , and the χ -dependent vertical torque Tx perfectly correspond
to the three terms in the electric spin AM density Se , Eqs. (3.6) and (3.7), even beyond the dipole-approximation (ka ≪ 1) of Eq. (2.23). (c). Experimental
observation of the transverse (out-of-plane) torque Ty via rotation of a double Mie particle [168]. Due to the complex macroscopic shape of the particle,
the torque here can also be exerted by the vertical x-gradient of the longitudinal (z) radiation-pressure force (2.22) (indicated by white arrows), as it was
interpreted in [160,168]. In the limit of small subwavelength particles, only the torque (2.23) from the spin AM density survives.

In 1999, Song et al. [168] reported the first experimental observation of the transverse in-plane rotation of probe Mie
particles in evanescent waves, Fig. 13c, again interpreted as the vertical gradient of the longitudinal radiation pressure. In
that experiment such mechanism was indeed possible (in addition to the transverse spin), because large clusters of Mie
particles with ka ≫ 1 were used. Therefore, although the experiment [168] can be considered as the first optomechanical
confirmation of the transverse (out-of-plane) spin AM in evanescent waves (without recognizing it), additional experiments
with nanoparticles would be desirable. Finally, optomechanical attempts to detect the transverse (in-plane) χ -dependent
electric spin AM (3.6) and (3.7), as well as the transverse spin AM densities in propagating fields (Section 3.2.1) are still to
be done.

3.3.3. Probing optical spin using atoms in a magnetic field

The light–atom interaction can be used as a quantum-mechanical analogue of classical particles probing light. In this
manner, transitions between different atomic angular-momentum states correspond to the torque on a classical particle
[29,130]. These internal angular-momentum atomic states are characterized by the m quantum number (electron vortex
number) and are defined with respect to a certain quantization axis. To fix the quantization axis and distinguish different
m-states, usually an external static magnetic field H0 is introduced, which generates a fine Zeeman splitting of the atomic
m-sublevels [152–154]. Thus, atomic transitions involving different m-levels with respect to the corresponding axis can
probe the local polarization and spin state of an optical field.
In this manner, the first attempt to perform the atomic Zeeman spectroscopy of the evanescent-wave polarization was
made in [153], where transitions to m = ±1 and m = 0 levels for different directions of an external magnetic field were
measured. These measurements clearly indicated the longitudinal z-component in the evanescent-field polarization for
 K.Y. Bliokh, F. Nori / Physics Reports 592 (2015) 1–38 23

Fig. 14. Recent measurements of the transverse spin AM density (3.25) and (3.26) in a focused linearly-polarized Gaussian beam (cf. Figs. 10 and 11).
(a) Schematics of the experiments [151] and [55], using magnetic circular dichroism (MCD) with the transverse static magnetic field H0 = H0 ȳ and probe-
particle scattering, respectively. (b) Experimentally measured MCD signal in the x > 0 half-beam (vs. the reference MCD signal shown in black) for the
x- and y-polarizations (τ = ±1) [151]. The presence of the τ = 1 signal and absence of the τ = −1 signal correspond to the presence of the electric
spin AM density (3.25) Sye in the first case and the magnetic spin AM density Sym in the second case. The latter does not induce the MCD response, see
Eq. (2.27). (c) Transverse electric spin AM distribution in the focal plane of a radially-polarized Gaussian beam, which was measured in [55] using nontrivial
scattering-particle probing. In this case, the electric spin AM density has an azimuthal distribution (3.26): S⊥ = Se⊥ ∝ −ϕ̄ (cf., the azimuthal distribution
of the magnetic spin AM density S⊥ = Sm ⊥ ∝ −ϕ̄ in the azimuthally-polarized beam in Fig. 12b).

H0 = H0 z̄ (see Fig. 8c in [153]), but could not properly show the ellipticity of the polarization in the (x, z )-plane for H0 = H0 ȳ
(see Fig. 8a in [153]).
Recently, a series of atomic measurements by Mitsch et al. [52,57,154] perfectly confirmed the presence of the
elliptical polarization and transverse spin in evanescent waves. Generating evanescent waves with opposite directions of
propagations Re k ∝ ± z̄, or opposite decay directions Im k ∝ ± x̄, experiments [52,57,154] indicated the presence of the
corresponding transverse spin AM (3.5) S⊥ ∝ ± ȳ. In particular, two types of experiments were realized in [154]. In both
experiments, a homogeneous external magnetic field H0 = H0 ȳ was applied. First, in this setting, a resonant evanescent
optical field drives ∆m = +1 (∆m = −1) transitions when its transverse spin is parallel (antiparallel) to the external
magnetic field. This was confirmed by optically pumping cold atoms to the m = +4 and m = −4 Zeeman states for the
two signs of the transverse spin Sye . Second, using an off-resonant interaction, where the spin AM of light acts as a fictitious
 
magnetic field Hfict
0 ∝ Se , the Zeeman splitting of levels proportional to H0 + Hfict
0 was detected, which clearly indicated
the transverse spin Sye in evanescent waves.

3.3.4. Magneto-optical and particle-scattering measurements

Two remarkable measurements of the transverse spin AM density in focused Gaussian beams were recently realized
in [55,151]. These works used completely different methods and confirmed the same results of Figs. 10 and 11 and
Eqs. (3.25), (3.26).
First, as we indicated in Section 2.4 and Eq. (2.27), the spin AM density Se is naturally coupled to an external magnetic
field H0 in magnetic circular dichroism (MCD) [150]. The experiment by Mathevet and Rikken [151] used x- and y-polarized
focused Gaussian beams, i.e., m = 0, τ = 1 and m = ∞, τ = −1, respectively. The external magnetic field H0 = H0 ȳ and
MCD crystal probe were used to measure the differential MCD signal between the upper x > 0 and lower x < 0 halves of the
beam (Figs. 14a, b). As a result, a clear MCD response was detected for the x-polarized but not the y-polarized beam. Thus,
these measurements registered the presence of a purely electric transverse spin AM density (3.25) Sy = Sye (with opposite
signs in the x > 0 and x < 0 halves of the beam) for the x-polarization, while these were insensitive to the transverse
24 K.Y. Bliokh, F. Nori / Physics Reports 592 (2015) 1–38

Fig. 15. The quantum spin Hall effect of light manifests itself as a strong transverse spin–momentum coupling (locking) in evanescent surface waves [56],
see Eqs. (3.5) and (V). Any interface supporting surface waves with evanescent free-space tails (3.2) has counterpropagating modes [in orange and cyan]
with opposite transverse spins [in red and blue]. This is observed experimentally via the excitation of unidirectional surface-evanescent modes with kz > 0
or kz < 0, depending on the helicity σinc = 1 or σinc = −1 and the corresponding y-directed longitudinal spin AM (2.17) of the transversely-incident
light [in green] (see [52,53,74–83] and Fig. 16). The incident propagating light is coupled to the surface-evanescent modes via a scatterer [in yellow]: e.g., a
nanoparticle or an atom. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

magnetic spin AM density Sy = Sym which takes place for the y-polarization. This is an important evidence of the dual
(electric–magnetic) asymmetry of the transverse helicity-independent spin AM.
Second, the experiment by Neugebauer et al. [55] extended their previous works [167,169] developing field scattering by
a probe nanoparticle for detection of nontrivial properties of an optical field. Namely, they showed that asymmetries in the
intensities of the particle-scattered near-fields are directly proportional to the transverse electric spin AM densities in the
field. Employing this observation and scanning the focal plane of the beam with a probe nanoparticle, the experiment [55]
provided a direct mapping of the transverse electric spin AM densities in linearly x-polarized and radially-polarized focused
Gaussian beams, Fig. 14c. Those results revealed the transverse y-directed spin AM density Sye ∝ +x, Eq. (3.25), for the
x-polarized beam, as well as the transverse azimuthal spin S⊥ = Se⊥ ∝ −ρ ϕ̄ in the radially-polarized beam. Note that the
azimuthal spin AM density has a purely electric (magnetic) origin for radially-(azimuthally-) polarized beams (cf. Figs. 14c
and 12b). In addition, as we remarked in Section 3.3.1, the transverse spin in such non-uniformly polarized beams has the
sign opposite to the uniformly-polarized case (3.26).

3.3.5. Spin-directional coupling and the quantum spin Hall effect of light
Perhaps the most remarkable application of the transverse spin AM is its ability to provide spin-controlled unidirectional
propagation of light [52,53,74–83]. This is based on the property (V), which strongly couples the transverse spin AM of an
evanescent wave, Eq. (3.5) and Fig. 4, with its direction of propagation. Indeed, evanescent waves propagating in opposite
directions along the same interface (i.e., having opposite Re k but the same Im k) have opposite transverse spins (3.5),
as shown in Fig. 15. In a similar manner, evanescent waves propagating in the same direction on opposite sides of the
same sample (i.e., having the same Re k but opposite Im k) also have opposite transverse spins. Notably, these features
are independent of the nature of the interface, and are valid for any interfaces supporting evanescent waves (3.2). These
could be metallic surfaces with plasmon-polariton modes [53,74,75], optical nano-fibers [52,57,77], or photonic-crystal
waveguides [78,80,81]. In all cases, 2D or 3D samples supporting edge or surface modes with evanescent tails (3.2) have
counter-propagating modes with opposite transverse spins (3.5), Fig. 15. This is a fundamental property of free-space Maxwell
equations that the transverse spin is strongly coupled (locked) with the direction of propagation of light. Recently it was
shown that this property can be associated with the intrinsic quantum spin Hall effect of light, which originates from the
spin–orbit coupling and topological properties of photons [56]. Analogous topological phenomena for electrons in solids
recently gave rise to a new class of materials: topological insulators [170,171].
Transverse spin–momentum locking is attracting rapidly-growing attention, and a number of prominent experiments
has been reported by different groups [52,53,57,74–80]. Fig. 16 shows examples from several experiments using quite
 K.Y. Bliokh, F. Nori / Physics Reports 592 (2015) 1–38 25

Fig. 16. Examples of recent experiments [53,77,78] demonstrating spin-controlled unidirectional excitation of modes with evanescent tails. Despite very
different systems [surface plasmon-polaritons in (a), nanofiber in (b), and photonic-crystal waveguide (c)], all these experiments show the robust quantum
spin Hall effect properties of light with the transverse spin–momentum locking in evanescent waves, see [56] and Fig. 15.

different setups and interfaces, but all based on the same transverse spin–direction coupling. In all these cases external
y-propagating light (kinc = k ȳ) with usual longitudinal spin AM (2.17) from the circular polarization (σinc = ±1) was
coupled to surface-evanescent waves propagating along the z-axis via some scatterer (nanoparticle, atom, etc.). Since the
spin AM of the incident wave has to match the y-directed transverse spin AM in evanescent waves, opposite helicities
(σinc = 1 and σinc = −1) of the incident light generated unidirectional evanescent waves propagating in opposite
z-directions (kz > 0 and kz < 0), respectively. Thus, such a strong and robust spin–direction locking offers perfect
‘‘chiral unidirectional interfaces’’, potentially important for many applications, such as quantum information [77,78],
topological photonics [56,172], and chiral spin networks [83]. Note that the spin–direction coupling is reversible, i.e., the
transverse emission of propagating light from oppositely-propagating evanescent waves has opposite circular polarizations
(spins) [53]. Furthermore, combining transverse spin–direction locking with spin-dependent magnetooptical scattering or
absorption [52,154] (Section 3.3.3), results in an efficient ‘‘optical diode’’, i.e., non-reciprocal transmission of light [57].
26 K.Y. Bliokh, F. Nori / Physics Reports 592 (2015) 1–38

To conclude this Section 3, various examples of the transverse spin AM densities in basic optical fields reveal themselves
in a variety of experiments involving light–matter interactions. Remarkably, in many cases, the presence and key role of the
transverse spin were not properly realized. Now, having the theoretical considerations and classification provided in this
review, as well as an analysis of the experiments, one can properly appreciate the important role of the transverse spin AM
as one of the inherent dynamical properties of light.

4. Transverse orbital angular momenta

In Section 3 we only considered the spin AM of light, which is determined by the intrinsic (polarization) degrees of
freedom. We now consider the transverse orbital AM, determined by the spatial (phase) degrees of freedom.

4.1. Transverse extrinsic orbital AM

4.1.1. General features

In Section 2 we described the main properties of the longitudinal orbital AM of light (IV), which is produced in paraxial
vortex beams and is widely used in modern optics [3–14]. According to Eqs. (2.10), (2.17) and (2.18), such orbital AM is
extrinsic locally, but becomes intrinsic integrally. (Recall that we use ‘‘intrinsic’’ and ‘‘extrinsic’’ to distinguish the coordinate
origin-independent and origin-dependent quantities, respectively.) Therefore, we regard such integral longitudinal vortex-
dependent AM as intrinsic orbital AM: ⟨L⟩ = Lint ∝ ℓ⟨k⟩/k.
 
∂
The z-propagating vortex beams (2.15) represent eigenmodes of the quantum AM operator L̂z = −i ∂ϕ . Thus, they reveal
the wave (intrinsic) features of the orbital AM. But what about the particle (extrinsic) aspects of the orbital AM, Eq. (2.1)?
The paraxial-beam analysis in Section 2.3 was done based on an assumption that the z-axis coincides with the beam axis,
i.e., the beam passes through the coordinate origin (Fig. 2). To unveil the particle-like AM of light, we have to consider the
same paraxial beam (2.15) but now shifted away from the origin. Without loss of generality, we assume that the beam center
is displaced by the vector r0 = ∆ ȳ along the y-axis, and the beam still propagates parallel to the z-axis, Fig. 17.
The beam field is given by Eq. (2.15) with the transformation (2.2):
r → r + r0 . (4.1)
This transformation trivially translates the energy (2.8), momentum (2.9), spin AM (2.11), and helicity (2.12) distributions.
Only the orbital AM density (2.10) explicitly involves the radius vector r, and, thus, transforms as in the point-particle
Eq. (2.3):
L → L + r0 × P. (4.2)
From this equation we immediately obtain the transformation of the integral orbital AM induced by the beam shift (4.1):

⟨L⟩ → ⟨L⟩ + r0 × ⟨P⟩ . (4.3)

Importantly, the second term in Eq. (4.3) (shown in the magenta frame) is transverse with respect to the mean momentum
⟨P⟩, and also extrinsic because it explicitly involves position of the beam with respect to the coordinate origin. For the paraxial
vortex beam (2.15), transformations (4.2) and (4.3) yield [cf. Eqs. (2.17) and (2.18)]:

W W
(−ρ kϕ̄ + ℓ z̄) + k∆ x̄, ⟨L⟩ ∝ ℓz̄ + k∆ x̄ ≡ Lint + Lext .
   
L≃ (4.4)
ω ω
Here the ℓ-dependent terms describe the longitudinal intrinsic vortex-dependent orbital AM L , which remains
 int

unchanged as in Eqs. (2.17) and (2.18). At the same time, the ∆-dependent terms in Eq. (4.4) originate from the cross-product
(4.3) between the transverse coordinate of the beam  center, r0 , and the longitudinal momentum of the beam, ⟨P⟩ ∝ kz̄, see
Fig. 17. This is the extrinsic orbital AM of light Lext = r0 × ⟨P⟩, which is always transverse by its definition (4.3). This AM is
independent of the vortex or polarization and is determined by the most basic particle properties of a light beam: its position
and direction of propagation, i.e., trajectory.
Since the orbital AM density (2.10) is extrinsic by its definition, the separation between the intrinsic and extrinsic orbital
AM appears only in integral values. For the generic localized optical field, one can calculate the mean (expectation) values
of the coordinate (weighted with the energy density (2.8) or another relevant optical density), ⟨r⟩, momentum, ⟨P⟩ ∝ ⟨k⟩,
and orbital AM as ⟨L⟩ = ⟨r × P⟩. Then, the intrinsic and extrinsic parts of the orbital AM can be separated as [31]

Lext = ⟨r⟩ × ⟨P⟩ , Lint = ⟨L⟩ − Lext .

     
(4.5)

This definition coincides with Eq. (4.4) in the case of paraxial beams. Since both the local coordinates r and the expectation
values ⟨r⟩ are equally transformed upon  translations
  (4.1),
  Eqs. (4.5) guarantee that the intrinsic and extrinsic parts of the
orbital AM ⟨L⟩ are transformed as Lint → Lint and Lext → Lext + r0 × ⟨P⟩. One can say that the intrinsic orbital AM
is the orbital AM calculated with respect to the centroid of the field, i.e., when ⟨r⟩ = 0. It should be remarked, however,
 K.Y. Bliokh, F. Nori / Physics Reports 592 (2015) 1–38 27

Fig. 17. The extrinsic orbital AM of light, Eqs. (4.1)–(4.5). The paraxial optical beam is shifted away from the coordinate
 origin
 by the r0 vector. It still carries
the same momentum and intrinsic longitudinal AM as in Fig. 2, but also acquires the extrinsic orbital contribution Lext = r0 × ⟨P⟩, which is transverse
with respect to the mean momentum ⟨P⟩. The extrinsic orbital AM takes into account the mechanical particle-like properties of light: its coordinate and
momentum, i.e., its trajectory (cf., Fig. 1a).

that the definition (4.5) depends on the particular definition of the field centroid (mean coordinates) ⟨r⟩. The coordinates
weighted with either the energy density, or the energy-flux density, or the photon-number density may result in different
values of ⟨r⟩, see examples in [44,112,113]. Nonetheless, they all coincide in paraxial monochromatic beams (2.15). Note that
the above equations can be applied to optical beams localized in two transverse dimensions (⟨r⟩ = ⟨r⊥ ⟩, the longitudinal
coordinate does not contribute to the extrinsic AM), as well as to wave packets localized in three dimensions.
Naturally, the main properties of the extrinsic orbital AM are similar to (I) for mechanical particles and are in sharp
contrast to the intrinsic orbital AM (IV):

Shifted beam AM: Orbital, Extrinsic, Transverse (out-of-plane). Key parameters: ⟨r⟩ , ⟨k⟩. (X)
 ext

Here we indicated the ‘‘out-of-plane’’ geometry, because the AM Lx = ⟨y⟩ ⟨Pz ⟩ (as shown in Fig. 17) is defined for the
y-localized and z-propagating beams formed by wave vectors with different ky , kz components [64].
 

4.1.2. Extrinsic AM in the spin Hall effect of light

It might seem at first that the extrinsic orbital AM does not make a physical difference, and cannot produce any observable
effects. Indeed, the coordinate origin is an abstract object, and an extrinsic quantity cannot play any role in free space. How-
ever, any interaction with matter (which singles out some specific coordinate frame attached to it) can involve the extrinsic
orbital AM in observable phenomena. The most remarkable example is the spin-Hall and orbital-Hall effects of light [63–73].
This group of phenomena stems from the spin–orbit interactions of light [31,173–175] and appears as transverse spin- and
vortex-dependent shifts of light (see also [122,176–185]). These effects attracted rapidly growing attention during the past
decade (see [174,175] for reviews), and here we only consider the simplest model example of the optical spin Hall effect.
We examine the total reflection of a paraxial optical beam (2.15) (for simplicity, without a vortex: ℓ = 0) at a planar
interface between free space and an isotropic non-absorbing medium, Fig. 18. The natural laboratory coordinates (X , y, Z )
are attached to the interface Z = 0, such that the medium occupies the Z > 0 half-space. The beam impinges the interface
at an angle θ with respect to the Z -axis, and its axis lies in the (X , Z ) plane, as shown in Fig. 18. From the law of reflection,
the propagation directions of the incident and reflected beams can be characterized by the following unit vectors:

z̄ = Z̄ cos θ + X̄ sin θ and z̄′ = −Z̄ cos θ + X̄ sin θ . (4.6)

Hereafter we use primes to indicate quantities related to the reflected beam.
According to Eq. (2.18), the incident and reflected beams carry longitudinal momenta and spin AM given by

⟨P⟩ ∝ k z̄, ⟨S⟩ ∝ σ z̄ and P ∝ k z̄′ , S ∝ σ ′ z̄′ .

 ′  ′
(4.7)
Here σ and σ ′ are the polarization helicities of the two beams, and we omit inessential factors, which do not affect the
considerations below. The incident-beam helicity σ is given by the initial conditions, whereas the helicity of the reflected
beam depends on the properties of the particular interface: e.g., σ ′ = −σ for a reflection from an ideal metal and σ ′ ≃ σ for
the total internal reflection at a dielectric interface near the critical incidence. Below we use only general arguments, which
are independent of the particular medium.
Note that the medium under consideration and the corresponding Maxwell equations describing light are rotationally
symmetric with respect to the Z -axis. Hence, according to the Noether theorem, the Z -component of the total angular
28 K.Y. Bliokh, F. Nori / Physics Reports 592 (2015) 1–38

Fig. 18. Schematics of the spin Hall effect of light (transverse beam shift) at a planar interface. The reflection of a circularly-polarized light beam at a planar
interface produces a transverse spin-dependent beam shift (4.10) [63,64,66,68,73]. In the simplest case of total reflection and isotropic non-absorbing
media, this shift can be derived solely from the balance of the z-component of the total AM in the system [61–64,73], Eq. (4.9). The spin-Hall shift generates
a transverse extrinsic orbital AM in the reflected beam (Fig. 17), which ensures the conservation of the z-component of the total AM between the incident
and reflected light.

momentum should be conserved in time. Furthermore, the medium plays the role of an external reflecting scalar potential in
Maxwell equations, and it does not acquire any AM. Therefore, the Z -component of the total AM of light must be conserved:
⟨JZ ⟩ = const. Next, we can substitute the continuous reflection of an infinitely-long stationary beam with the reflection of
an arbitrarily long but finite wave packet. Then, only the incident (reflected) beam-packet exists at time t = −∞ (t = +∞),
and we conclude that the Z -component of the total AM of the incident beam-packet must be equal to the Z -component of the
total AM of the reflected beam-packet: ⟨JZ ⟩ = JZ′ . Obviously, the dynamical characteristics of arbitrary long paraxial wave
packets are the same as for the beams (where the integral values (2.18) are calculated per unit propagation length).
However, projecting the longitudinal spin AM of the incident and reflected beams, Eqs. (4.6) and (4.7) on the Z -axis, we
have: ⟨SZ ⟩ ∝ σ cos θ and SZ′ ∝ −σ ′ cos θ ̸= ⟨SZ ⟩. This seemingly contradicts the AM conservation law in the problem. In

fact, this difference between the Z -components of the spin AM of the incident and reflected light can be compensated from
only one source: the extrinsic orbital AM (4.3) and (4.4). To produce this AM, the reflected beam must be shifted along the y
axis by some distance r0 = ∆ ȳ, as shown in Fig. 18. Then, this beam will possess the orbital AM
 
′
∝ r0 × P′ .
 ′
L ≡ Lext
 
(4.8)

∝ −∆ k sin θ . Now we can satisfy the above AM

 ext ′

Projection onto the Z -axis using Eqs. (4.6) and (4.7) yields LZ
conservation law:
⟨SZ ⟩ = SZ′ + Lext ,
   ′
Z (4.9)
when
 cot θ
∆ = − σ + σ′ .

(4.10)
k
Eq. (4.10) describes the simplest case of the spin Hall effect of light, also known as the Imbert–Fedorov transverse shift
[63,64,66,68] (see [73] for a review). It means that a beam of light carrying intrinsic longitudinal spin AM experiences a spin-
dependent transverse shift (4.10) after reflection (or refraction) at a plane interface. This shift was predicted by Fedorov in
1955 and detected for the first time by Imbert in 1972 [186,187]. However, the original theoretical explanation of this effect
was misleading, and the 50-years-long studies of this fine phenomenon were full of controversies. Only recently the effect
was properly explained in terms of the spin–orbit interactions of light [63,64,68,73] and measured with a great accuracy
using the ‘‘quantum weak measurement’’ technique [66,72,188–193] (see Fig. 19 for examples of experiments measuring
the spin Hall effect of light). Although the shift (4.10) is small (a fraction of the wavelength), it is important both because of
its fundamental nature and considerable contribution at the nano-scales of modern optics [72,122,175,182].
The role of the AM conservation (4.9) and the extrinsic orbital AM in the transverse beam shift was first revealed by Player
and Fedoseyev in 1987 [61,62], and later confirmed in [63,64,69,70]. This could be regarded as the first example indicating
 K.Y. Bliokh, F. Nori / Physics Reports 592 (2015) 1–38 29

Fig. 19. Examples of experiments [66,67,72] demonstrating the transverse spin-dependent shifts (spin Hall effect) of light, which are related to the AM
conservation as in Eqs. (4.9) and (4.10). (a) A tiny subwavelength shift in the beam refraction was amplified in [66] using the ‘‘quantum weak measurements’’
technique [188–193] and, as a result, measured with extraordinary Ångstrom accuracy. (b) The smooth propagation of light along a curvilinear (helical
here) trajectory also brings about a transverse spin-dependent deflection [67], which ensures the conservation of the total AM of light (4.11). (c) Spin-Hall
effect with surface-plasmon beams generated at a single slit on the metal surface [72]. An amplification by the ‘‘quantum weak measurement’’ method
makes the beam shifts visible, in both momentum [directions shown by orange arrows] and coordinates [shown by green arrows]. (For interpretation of
the references to color in this figure legend, the reader is referred to the web version of this article.)

the importance of the transverse extrinsic orbital AM of light for observable effects. Note that here we considered the
simplest total-reflection case and derived the exact expression for the shift (4.10) from heuristic AM-balance considerations.
At the same time, the proper solution of the generic beam-reflection/refraction problem requires rather delicate calculations
involving the Fourier spectra of the beams [64,68,73] or the corresponding three-dimensional real-space fields [194]. But
in any case the exact solutions of the beam reflection/refraction problems at planar isotropic interfaces satisfy the total AM
conservation law [70,73].
We also briefly mention two important extensions of the transverse spin-dependent beam shift considered above. First,
a quite similar vortex-dependent shift
 and orbital Hall effect takes place for paraxial beams carrying intrinsic longitudinal
orbital angular momentum (2.18) Lint ∝ ℓ ⟨k⟩ /k. In this case, the balance between the intrinsic and extrinsic parts of the
orbital AM results in observable beam-shift effects akin to Eqs. (4.9) and (4.10) [69–71,73,179,181]. Second, instead of sharp
interfaces with jumps in the propagation direction of light, one can consider the smooth propagation of light in a gradient-
index dielectric medium. The evolution of light in such medium is described by the Hamiltonian equations of motion for
coordinates ⟨r⟩ and momentum ⟨P⟩, and a smooth trajectory, exactly as for a particle in classical mechanics [195]. Taking into
30 K.Y. Bliokh, F. Nori / Physics Reports 592 (2015) 1–38

account the spin–orbit interaction in Maxwell equations with ‘‘semi-classical’’ wavelength-order corrections [196], one can
derive the helicity-dependent corrections to the traditional geometrical-optics equations for the trajectory of light [63,65,67,
173,180]. The resulting equations have fundamental importance for the evolution of various spinning particles [197–199].
Importantly, these equations of motion possess a new integral of motion—the total angular momentum of light including
the intrinsic spin and extrinsic orbital contributions [63,67,196]:
⟨P⟩
Lext + ⟨S⟩ = ⟨r⟩ × ⟨P⟩ + σ = const.
 
(4.11)
k
Fig. 19 shows examples of experimental observations of the spin Hall effect of light and the corresponding beam shifts
in various physical situations: beam refraction at a dielectric interface [66], propagation of light along a smooth curvilinear
trajectory [67], and refraction into surface-plasmon beams at a light–plasmon interface [72].
An interesting extension of the spin Hall effect of light was suggested by Aiello et al. in [44], where they considered the
usual paraxial beam (2.15) carrying a longitudinal spin AM (2.18): ⟨S⟩ ∝ σ ⟨k⟩/k. Instead of real changes in the direction of
propagation of the beam (as in refraction or reflection), Aiello et al. considered ‘‘virtual’’ rotation of the coordinate frame by
the angle θ in the (x, z )-plane in free space. In the new (X , y, Z ) coordinates, the longitudinal spin AM of the beam acquires
a ‘‘transverse’’ component, ⟨SX ⟩ = − ⟨Sz ⟩ sin θ ̸= 0, while the centroid of the Poynting-momentum flux through a tilted
(X , y) plane becomes shifted: ⟨y⟩Πz = σ tan θ /2k [cf. Eq. (4.10)]. This effect was called the ‘‘geometric spin-Hall effect of
light’’. Naturally, the rotation of a coordinate frame cannot change any beam properties in free space, and this shift is also
‘‘virtual’’. It is related to the definition of the beam centroid via a flux of the tilted Poynting vector. At the same time, the
beam centroid defined via the energy density remains non-shifted: ⟨y⟩W = 0 [44]. The angular momentum of the beam also
remains purely longitudinal, as in the frame-independent equation (2.18). Nonetheless, the non-zero shift ⟨y⟩Πz can play a
role in the interaction of light with oblique detectors sensitive to the momentum density [200,201].

4.2. Transverse intrinsic orbital AM

In paraxial monochromatic vortex beams (2.15) the intrinsic orbital AM is always longitudinal, Eq. (2.18): Lint ∝ ℓ⟨P⟩/k.
 
 int

Moreover, one can show that the general definition (4.5) results in L ∥ z̄ when ⟨P⟩ ∥ z̄ and the beam centroid is
determined as the transverse centroid of the longitudinal momentum density Pz : ⟨r⟩ = ⟨r⊥ ⟩Pz . However, for the generic
optical field and more natural energy-centroid definition ⟨r⟩ = ⟨r⟩W , the intrinsic orbital AM (4.5) is
 not
 restricted to be
purely longitudinal. Can there be optical fields with non-collinear intrinsic orbital AM and momentum, Lint ∦ ⟨P⟩, i.e., a non-
 
zero transverse component Lint ⊥ ? This question was recently examined by Bliokh and Nori [47], and it was concluded that
such transverse intrinsic orbital AM can naturally occur in polychromatic fields with non-stationary intensity distributions.
Since the intrinsic orbital AM of paraxial beams is associated with optical vortices inside the beam, one can expect
the transverse intrinsic orbital AM in fields with transverse vortices. In contrast to the screw dislocation of phase fronts
in longitudinal vortices (Fig. 2b), such transverse or skew vortices represent edge or mixed edge-screw dislocations in phase
fronts [39,41,43]. In other words, the vortex singularity line, which is the nodal intensity line, is skew or orthogonal to the
propagation direction of the wave. Note that transverse vortex-like circulations of the Poynting vector occur in nonparaxial
light fields in many basic interference and diffraction problems [202–206]. However, in all these problems, transverse
vortices appear either in vortex–antivortex pairs or in non-localized fields where the integral AM is ill-defined. Furthermore,
previous studies [202–206] found such transverse circulations in the Poynting vector, and no proper AM analysis based
on the canonical momentum (2.9) has been made. Thus, so far no transverse intrinsic orbital AM has been found in
monochromatic optical fields. At the same time, skew vortices and intrinsic orbital AM are ubiquitous in polychromatic
fields with transversely-moving intensity distributions [47].
Since in this Section 4.2 we discuss only orbital AM, for simplicity we will ignore polarization  (spin)
 degrees of freedom
and consider scalar waves. To construct non-collinear momentum and intrinsic orbital AM, Lint ∦ ⟨P⟩, we look for non-
stationary wave fields with transversely moving vortices. Such solutions are very easy to find in free space. It is sufficient to
observe the usual stationary vortex beam (2.15) in a transversely-moving reference frame. Indeed, let the observer moves
with velocity v = v x̄ along
 the x axis. The vortex line (and, hence, the intrinsic orbital AM) will keep its direction along
the z-axis: Lint = Lint z̄, i.e., orthogonal to the observer motion. At the same time, the beam will become non-stationary
and moving with velocity −v along the x-axis. Thus, the beam will acquire a momentum component along the x-axis, ⟨Px ⟩,
which is transverse with respect to the intrinsic orbital AM.
To quantify this consideration, we now consider the Lorentz transformation of the momentum and angular momentum of
a particle in relativistic mechanics. In the original reference frame (t , r), let the particle have energy E, momentum p = p z̄,
and AM L = L z̄. Then, the  Lorentz boost with v = v x̄ results in the following energy, momentum, and AM values in the
moving reference frame t ′ , r′ [112,207,208]:


E
E′ = γ E, p′ = p − γ v, L′ = γ L, (4.12)
c2
where γ = 1/ 1 − v 2 /c 2 is the Lorentz factor. Thus, the momentum and AM become non-parallel to each other due

to the transverse momentum acquired by the particle. The difference in transformations of the momentum and AM in
 K.Y. Bliokh, F. Nori / Physics Reports 592 (2015) 1–38 31

Fig. 20. Transverse Lorentz boost of the paraxial vortex beam carrying an intrinsic orbital
  AM [47,112]. (a) Monochromatic vortex beam  (as in Fig. 2b), in
the motionless reference frame (t , r), with the longitudinal intrinsic orbital AM (2.18) Lint . (b) The same beam in the reference frame t ′ , r′ moving with
the transverse velocity v = v x̄. The beam becomes non-stationary (polychromatic) and acquires the transverse momentum component Px′ = −γ k vc ,
 
 
  ′
Eq. (4.15). Thus, the beam momentum P′ and intrinsic orbital AM Lint , Eq. (4.17), become non-collinear to each other, and the beam carries a transverse
intrinsic orbital AM component (4.19) and (XI). The tilted phase fronts of the vortex signify the mixed edge-screw phase dislocation in the transversely-
moving vortex, i.e., spatio-temporal vortex [47,213] (see Fig. 21).

relativity is explained by the fact that the momentum is part of the energy–momentum four-vector, while the AM is part of
the antisymmetric rank-2 AM tensor [207,208].
Let us now consider the same Lorentz boost for a scalar paraxial wave beam similar to Eq. (2.15) (see Figs. 20 and 21):
ψ(r, t ) ≃ A (ρ, z ) exp(ikz + iℓϕ − iωt ). (4.13)
The beam is stationary in the original reference frame (t , r) and is characterized by the energy, momentum, and intrinsic
orbital AM similar to Eq. (2.18):
⟨W ⟩ ⟨W ⟩
⟨W ⟩ ∝ ω, ⟨P⟩ = k z̄, ⟨L⟩ = ℓ z̄. (4.14)
ω ω
Performing the Lorentz transformation of the field (4.13) and calculating the same quantities in the moving reference frame
t ′ , r′ , one can obtain [112]


 ′  ⟨W ⟩  ′ ω   ′  ⟨W ⟩
W ∝ γ ω, k z̄ − γ 2 v , γ ℓ z̄′ ,
 ′
P = L = (4.15)
ω c ω
which is in agreement with Eq. (4.12).  
There is a subtle issue, however, and the orbital AM L′ is not purely intrinsic anymore. Careful consideration shows that
the beam acquires a transverse AM-dependent shift [112,209]:
v × ⟨L⟩
r′0 = − − v t ′, (4.16)
2 ⟨W ⟩
32 K.Y. Bliokh, F. Nori / Physics Reports 592 (2015) 1–38

Fig. 21. Transverse Lorentz boost of the scalar Bessel beam (4.13) with A (ρ, z ) = J|ℓ| (k⊥ ρ) and k → kz , where k⊥ = k2 − k2z . Here k⊥ /k = 0.5 and

ℓ = 2, while the Lorentz transformation is characterized by v/c = 0.8. (a) The original beam’s Fourier spectrum (in k-space), its transverse intensity and
momentum-density distributions (as in Fig. 3b), as well as the transverse phase distribution showing a screw phase dislocation (vortex). (b) The same beam
observed in a transversely-moving reference frame with velocity v = v x̄. The beam spectrum is now shifted away from the iso-frequency sphere, i.e., the
beam becomes polychromatic. The transverse motion of the vortex (clearly seen in the transverse-momentum distribution) makes its momentum P′ and
 
orbital AM L′ tilted with respect to each other, Eq. (4.19). The transverse phase distribution exhibits a mixed edge-screw dislocation, which signifies the
spatio-temporal character of the vortex [47,213]. The transverse intensity distribution reveals the x-squeezing of the beam due to the Lorentz contraction
and also a small transverse y-deformation due to the transverse shift of beam’s centroid (4.16), i.e., the relativistic Hall effect [112,113].

which is defined here as a centroid of the photon-number density (the transverse displacement of the energy-density
centroid being twice as large). This observer- and AM-dependent transverse shift can be called ‘‘relativistic Hall effect’’
[112,113,210–212] (Fig. 21). Notably, the analogous spin-AM-dependent shift for relativistic particles is related to the
conservation of the total AM, similar to the spin-Hall effect in Section 4.1.2 [210]. Thus, the orbital AM (4.15) consists of
the following intrinsic and extrinsic contributions:
 ⟨W ⟩ γ + γ −1 ′
′
 
′

Lint ℓ z̄ , Lext = r′0 × P′ ,
 
= (4.17)
ω 2
 int ′   ext ′   ′ 
which ensure that Lz + Lz = Lz , in agreement with Eq. (4.15).
 
 ′ ′
Eqs. (4.15) and (4.17) reveal the non-collinear momentum P and intrinsic orbital AM Lint of the beam, Figs. 20 and
21. Let us introduce the coordinate frame (X , y, Z ) rotated by the angle θ = arctan γ vc in the (x, z )-plane, such that the
 
beam momentum is aligned with the Z -axis [113]:

z̄′ = Z̄ cos θ + X̄ sin θ and x̄′ = −Z̄ sin θ + X̄ cos θ . (4.18)

v
Using sin θ = c
, cos θ = γ −1
, and ω = kc, we obtain the momentum (4.15) and intrinsic orbital AM (4.17) in these
coordinates:

 ′
 γ + γ −1 γ + γ −1 v
P ∝ γ k Z̄, γ −1 ℓ Z̄ + ℓ X̄ .
 ′
Lint ∝ (4.19)
2 2 c
 
Here, the last term in the orange frame is the transverse intrinsic orbital AM Lint ′
⊥ , induced by the transverse Lorentz boost
of the beam. Note that this AM lies in the plane formed by ⟨P⟩ and v, and it appears in linear order in v/c. Summarizing the
 K.Y. Bliokh, F. Nori / Physics Reports 592 (2015) 1–38 33

properties of the transverse AM in Eq. (4.19):

Lorentz-boosted beam AM: Orbital, Intrinsic, Transverse (in-plane). Key parameters: ℓ, v. (XI)

Note that here the boost velocity v substituted the beam momentum ⟨k⟩ in the longitudinal intrinsic orbital AM (IV). Since
the velocity and momentum share the same P -odd and T -odd nature, Eq. (4.19) have the proper T -odd symmetry of the
AM.
It is worth remarking that the transversely-moving beam carries a spatio-temporal vortex [47,213]. It is characterized by
a mixed edge-screw dislocation  in phase [39,41,43] (see Figs. 20b and
 21b), and forms a vortex and circulating four-current
not only in the (x, y) or kx , ky planes but also in the (t , y) or ω, ky planes [47]. Fig. 21 shows details of the transformation

of the scalar vortex Bessel beam (4.13) observed in a transversely Lorentz-boosted reference frame. Transversely-moving
spatio-temporal vortices carrying the transverse intrinsic orbital AM component (XI) can appear in two-dimensional surface
waves (e.g., surface plasmon-polaritons), nonlinear media [213], or in interference of three or more plane waves with
different frequencies (cf. [214]). In addition, such vortex states of light can naturally be generated from moving sources:
e.g., astrophysical ones [215–217].

5. Conclusions

In this review we aimed to provide a unified theory of the angular momentum of light (Section 2) and a comprehensive
picture of various types of spin and orbital AM in structured optical fields (Sections 3 and 4). We also provided an overview
of the main experiments revealing the traditional and novel types of optical AM.
In contrast to previous reviews on the optical AM [5–13], we considered the canonical picture of the optical AM, which
is based on the canonical momentum and spin densities rather than the Poynting vector. Such description is in perfect
agreement with experiments, free of difficulties appearing with the Poynting momentum and AM, and it naturally admits
the separation of the orbital and spin degrees of freedom of light.
Most importantly, this general theory enabled us to thoroughly investigate and classify novel types of spin and orbital
AM, which are attracting rapidly-growing attention. In addition to the most common longitudinal spin and orbital AM in
paraxial beams (previously explored in detail in many books and reviews [5–13]), we described a number of transverse AM
of different nature. The rather unusual and contrasting properties of different AM kinds are summarized in Eqs. (I)–(XI)
throughout the review and compiled here in Table 1. One can see that various kinds of optical AM reveal dependences on
a variety of key parameters. This is in contrast to the traditional picture with the longitudinal spin AM determined by the
wave helicity and the longitudinal orbital AM determined by the vortex charge.
Apparently, the most intriguing new kind of AM are the transverse spin AM (Section 3). We have revealed at least three
basic types of transverse spin:

• The transverse (out-of-plane) helicity-independent spin AM density in evanescent and interfering propagating waves.
This spin AM is determined solely by the wave-vector and phase parameters of the waves, and it is orthogonal to the
wave vectors in the field.
• The transverse (in-plane) ‘‘dual-antisymmetric’’ spin AM density in evanescent and interfering propagating waves. This
spin AM has opposite ‘‘electric’’ and ‘‘magnetic’’ parts, so that the total spin vanishes, but these parts can appear in dual-
asymmetric light–matter interactions. This spin lies in the plane with the wave vectors of the field but orthogonally to
the main momentum. Furthermore, it is proportional to the second Stokes parameter of the polarization rather than the
helicity.
• The integral transverse (in-plane) spin AM, which appears in the interference of propagating waves with different wave
vectors and helicities. This integral spin is generated by the vector sum of the usual longitudinal spins of the interfering
waves, and, hence, depends on their helicities.

The first of the above, the transverse helicity-independent spin AM, was recently measured in several experiments in both
propagating and evanescent fields. Most importantly, it promises remarkable applications in spin-controlled unidirectional
interfaces with evanescent waves. This is because such transverse spin in evanescent waves is intimately related to
fundamental topological properties of Maxwell equations in free space.
We have also described the transverse orbital AM of light (Section 4). Here one should distinguish the intrinsic and
extrinsic types of the integral orbital AM:

• The extrinsic orbital AM is the most common and trivial type of AM, which is entirely analogous to that of a classical
point particle. Nonetheless, it can play a crucial role in observable optical effects involving spin–orbit interactions and
the conservation of the total AM. We have shown that the transverse extrinsic orbital AM is responsible for the spin-Hall
effect of light refracted or reflected in inhomogeneous media.
• The transverse part of the intrinsic orbital AM (from the optical vortex) can appear in polychromatic fields. In particular,
it naturally appears in Lorentz-boosted beams observed in a transversely-moving reference frame.
34 K.Y. Bliokh, F. Nori / Physics Reports 592 (2015) 1–38

Table 1
Different types of the AM (I)–(XI), their basic features, and key parameters. Previous reviews [5–13] focused mostly on the paraxial monochromatic beams.
System and AM type Density Integral Ext./Int. Key parameters

Point particle: Transverse orbital AM L=r×p ext. r, p

Particle/wave
k
Plane wave: Longitudinal spin AM S∝σ int. σ,k
k
P∥ ⟨P⟩
Paraxial Polarized beam: Longitudinal spin AM S≃σ ⟨S⟩ ≃ σ int. σ , ⟨k⟩
monochromatic k k
beams ⟨P⟩
Vortex beam: Transv./Long. orbital AM L=r×P ⟨L⟩ ≃ ℓ ext./int. ℓ, ⟨k⟩
k

Re k × Im k Re k × Im k
Other Evanescent wave: Transverse S⊥ ∝ ⟨S⊥ ⟩+ ∝ int. Re k, Im k
structured fields (out-of-plane) σ -indep. spin AM (Re k)2 (Re k)2
e,m k Im k e,m k Im k
±χ χ, Im k
 
Evanescent wave: Transverse S⊥ ∝ ±χ S⊥ int.
(in-plane) anti-dual spin AM (Re k)2 (Re k)2
⟨k⟩ × δ k
Interfering plane waves: Transverse S⊥ ∝ sin δ 8 ⟨S⊥ ⟩ = 0 int. ⟨k⟩ , δ k, δ 8
(out-of-plane) σ -indep. spin AM k2
e,m δk e,m
sin δ 8 χ, δ k, δ 8
 
Interfering plane waves: Transverse S⊥ ∝ ∓χ S⊥ =0 int.
(in-plane) anti-dual spin AM k
δk
Interfering plane waves: Transverse ··· ⟨S⊥ ⟩ = δσ int. δσ , δ k
(in-plane) σ -dep. spin AM k

⟨r⟩ , ⟨k⟩
 
Modified Shifted beam: Transverse extrinsic Lext = ⟨r⟩ × ⟨P⟩ ext.
paraxial beams orbital AM
v⊥
L⊥ ∝ ℓ , ℓ, v
 int

Lorentz-boosted beam: Skew int.
intrinsic orbital AM c

Notably, the transverse extrinsic orbital AM and the transverse helicity-independent spin in evanescent waves are
closely related to different manifestations of the spin–orbit interactions of light. Namely, they underpin the spin Hall effect
(i.e., small spin-dependent transverse shift) and quantum spin Hall effect of light (i.e., spin-dependent unidirectional edge
modes), respectively. The spin–orbit interactions of light originate from the transversality of electromagnetic waves and the
corresponding longitudinal field components in the non-plane-wave fields. As we have seen in Section 3, these longitudinal
field components are also responsible for the generation of the transverse spin AM.
To summarize, we have shown how a very concise and fundamental theory of the spin and orbital AM of light in generic
optical fields describes a rich variety of AM forms with extraordinary features in specific light configurations. All of these
angular momenta appear from the interplay between the particle (localized) and wave (extended) features of structured light
fields. Modern nano-optics and photonics tend to explore and employ new degrees of freedom of structured light, which
are absent in simple plane waves or particle-like Gaussian wave packets. Therefore, these frontier areas of research offer
extraordinary opportunities for exploiting longitudinal and transverse angular momenta of light in various fundamental
studies and applications.

Acknowledgments

We acknowledge fruitful discussions and correspondence with A.Y. Bekshaev, A. Aiello, P. Banzer, P. Schneeweiss,
E.A. Ostrovskaya, R. Mathevet, G.L.J.A. Rikken, and T. Kawalec. This work was partially supported by the RIKEN iTHES Project,
MURI Center for Dynamic Magneto-Optics (AFOSR grant no. FA9550-14-1-0040), JSPS-RFBR (contract no. 12-02-92100), a
Grant-in-Aid for Scientific Research (A), and the Australian Research Council.

References

[1] J.H. Poynting, The wave-motion of a revolving shaft, and a suggestion as to the angular momentum in a beam of circularly-polarized light, Proc. R.
Soc. Lond. Ser. A 82 (1909) 560–567.
[2] R.A. Beth, Mechanical detection and measurement of the angular momentum of light, Phys. Rev. 50 (1936) 115–125.
[3] L. Allen, M.W. Beijersbergen, R.J.C. Spreeuw, J.P. Woerdman, Orbital angular-momentum of light and the transformation of Laguerre–Gaussian laser
modes, Phys. Rev. A 45 (1992) 8185–8189.
[4] H. He, M.E.J. Friese, N.R. Heckenberg, H. Rubinsztein-Dunlop, Direct observation of transfer of angular-momentum to absorptive particles from a
laser beam with a phase singularity, Phys. Rev. Lett. 75 (1995) 826–829.
[5] L. Allen, S.M. Barnett, M.J. Padgett (Eds.), Optical Angular Momentum, IoP Publishing, Bristol, 2003.
[6] A. Bekshaev, M. Soskin, M. Vasnetsov, Paraxial Light Beams with Angular Momentum, Nova Science Publishers, New York, 2008.
[7] D.L. Andrews (Ed.), Structured Light and its Applications, Academic Press, Amsterdam, 2008.
[8] J.P. Torres, L. Torner (Eds.), Twisted Photons, Wiley-VCH, Weinheim, 2011.
[9] D.L. Andrews, M. Babiker (Eds.), The Angular Momentum of Light, Cambridge University Press, Cambridge, 2013.
[10] L. Allen, M.J. Padgett, M. Babiker, The orbital angular momentum of light, Prog. Opt. 39 (1999) 291–372.
 K.Y. Bliokh, F. Nori / Physics Reports 592 (2015) 1–38 35

[11] G. Molina-Terriza, J.P. Torres, L. Torner, Twisted photons, Nat. Phys. 3 (2007) 305–310.
[12] S. Franke-Arnold, L. Allen, M.J. Padgett, Advances in optical angular momentum, Laser & Photon. Rev 2 (2008) 299–313.
[13] A.M. Yao, M.J. Padgett, Optical angular momentum: origins, behavior, and applications, Adv. Opt. Photon. 3 (2011) 161–204.
[14] A. Bekshaev, K.Y. Bliokh, M. Soskin, Internal flows and energy circulation in light beams, J. Opt. 13 (2011) 053001.
[15] M.E.J. Friese, T.A. Nieminen, N.R. Heckenberg, H. Rubinsztein-Dunlop, Optical alignment and spinning of laser-trapped microscopic particles, Nature
394 (1995) 348–350.
[16] A.T. O’Neil, I. MacVicar, L. Allen, M.J. Padgett, Intrinsic and extrinsic nature of the orbital angular momentum of a light beam, Phys. Rev. Lett. 16
(2002) 053601.
[17] V. Garcés-Chavéz, D. McGloin, M.J. Padgett, W. Dultz, H. Schmitzer, K. Dholakia, Observation of the transfer of the local angular momentum density
of a multiringed light beam to an optically trapped particle, Phys. Rev. Lett. 91 (2003) 093602.
[18] J.E. Curtis, D.G. Grier, Structure of optical vortices, Phys. Rev. Lett. 90 (2003) 133901.
[19] H. Adachi, S. Akahoshi, K. Miyakawa, Orbital motion of spherical microparticles trapped in diffraction patterns of circularly polarized light, Phys. Rev.
A 75 (2007) 063409.
[20] Y. Zhao, J.S. Edgar, G.D.M. Jeffries, D. McGloin, D.T. Chiu, Spin-to-orbital angular momentum conversion in a strongly focused optical beam, Phys.
Rev. Lett. 99 (2007) 073901.
[21] M.V. Berry, Paraxial beams of spinning light, Proc. SPIE 3487 (1998) 6–11.
[22] A.I. Akhiezer, V.B. Berestetskii, Quantum Electrodynamics, Interscience Publishers, New York, 1965.
[23] D.E. Soper, Classical Field Theory, Wiley, New York, 1976.
[24] C. Cohen-Tannoudji, J. Dupont-Roc, G. Grynberg, Photons and Atoms, Wiley-Interscience, New York, 1997.
[25] F.J. Belinfante, On the current and the density of the electric charge, the energy, the linear momentum and the angular momentum of arbitrary fields,
Physica 7 (1940) 449–474.
[26] J. Humblet, Sur le moment d’impulsion d’une onde électromagnétique, Physica 10 (1943) 585–603.
[27] S.M. Barnett, L. Allen, Orbital angular-momentum and nonparaxial light-beams, Opt. Commun. 110 (1994) 670–678.
[28] S.J. van Enk, G. Nienhuis, Spin and orbital angular-momentum of photons, Europhys. Lett. 25 (1994) 497–501.
[29] S.J. van Enk, G. Nienhuis, Commutation rules and eigenvalues of spin and orbital angular momentum of radiation fields, J. Modern Opt. 41 (1994)
963–977.
[30] M.V. Berry, Optical currents, J. Opt. A: Pure Appl. Opt. 11 (2009) 094001.
[31] K.Y. Bliokh, M.A. Alonso, E.A. Ostrovskaya, A. Aiello, Angular momenta and spin–orbit interaction of nonparaxial light in free space, Phys. Rev A 82
(2010) 063825.
[32] S.M. Barnett, Rotation of electromagnetic fields and the nature of optical angular momentum, J. Modern Opt. 57 (2010) 1339–1343.
[33] I. Bialynicki-Birula, Z. Bialynicki-Birula, Canonical separation of angular momentum of light into its orbital and spin parts, J. Opt. 13 (2011) 064014.
[34] K.Y. Bliokh, A.Y. Bekshaev, F. Nori, Dual electromagnetism: helicity, spin, momentum, and angular momentum, New J. Phys. 15 (2013) 033026.
[35] M. Ornigotti, A. Aiello, Surface angular momentum of light beams, Opt. Express 22 (2014) 6586–6596.
[36] I. Fernandez-Corbaton, X. Zambrana-Puyalto, G. Molina-Terriza, On the transformations generated by the electromagnetic spin and orbital angular
momentum operators, J. Opt. Soc. Amer. B 31 (2014) 2136–2141.
[37] K.Y. Bliokh, J. Dressel, F. Nori, Conservation of the spin and orbital angular momenta in electromagnetism, New J. Phys. 16 (2014) 093037.
[38] E. Leader, C. Lorcé, The angular momentum controversy: What’s it all about and does it matter?, Phys. Rep. 541 (2014) 163–248.
[39] J.F. Nye, M.V. Berry, Dislocations in wave trains, Proc. R. Soc. Lond. Ser. A 336 (1974) 165–190.
[40] V.Y. Bazhenov, M.V. Vasnetsov, M.S. Soskin, Laser beams with screw dislocations in their wavefronts, JETP Lett. 52 (1990) 429–431.
[41] M.S. Soskin, M.V. Vasnetsov, Singular optics, Prog. Opt. 42 (2001) 219–276.
[42] A.S. Desyatnikov, Y.S. Kivshar, L. Torner, Optical vortices and vortex solitons, Prog. Opt. 47 (2005) 291–391.
[43] M.R. Dennis, K. O’Holleran, M.J. Padgett, Singular optics: optical vortices and polarization singularities, Prog. Opt. 53 (2009) 293–363.
[44] A. Aiello, N. Lindlein, C. Marquardt, G. Leuchs, Transverse angular momentum and geometric spin Hall effect of light, Phys. Rev Lett. 103 (2009)
100401.
[45] A. Aiello, C. Marquardt, G. Leuchs, Transverse angular momentum of photons, Phys. Rev A 81 (2010) 053838.
[46] K.Y. Bliokh, F. Nori, Transverse spin of a surface polariton, Phys. Rev. A 85 (2012) 061801(R).
[47] K.Y. Bliokh, F. Nori, Spatiotemporal vortex beams and angular momentum, Phys. Rev. A 86 (2012) 033824.
[48] K.-Y. Kim, I.-M. Lee, J. Kim, J. Jung, B. Lee, Time reversal and the spin angular momentum of transverse-electric and transverse-magnetic surface
modes, Phys. Rev. A 86 (2012) 063805.
[49] P. Banzer, M. Neugebauer, A. Aiello, C. Marquardt, N. Lindlein, T. Bauer, G. Leuchs, The photonic wheel – demonstration of a state of light with purely
transverse angular momentum, J. Europ. Opt. Soc. Rap. Public. 8 (2013) 13032.
[50] K.Y. Bliokh, A.Y. Bekshaev, F. Nori, Extraordinary momentum and spin in evanescent waves, Nature Commun. 5 (2014) 3300.
[51] A. Canaguier-Durand, C. Genet, Transverse spinning of a sphere in plasmonic field, Phys. Rev. A 89 (2014) 033841.
[52] R. Mitsch, C. Sayrin, B. Albrecht, P. Schneeweiss, A. Rauschenbeutel, Quantum state-controlled directional spontaneous emission of photons into a
nanophotonic waveguide, Nature Commun. 5 (2014) 5713.
[53] D. O’Connor, P. Ginzburg, F.J.F.J. Rodríguez-Fortuño, G.A. Wurtz, A.V. Zayats, Spin–orbit coupling in surface plasmon scattering by nanostructures,
Nature Commun. 5 (2014) 5327.
[54] A.Y. Bekshaev, K.Y. Bliokh, F. Nori, Transverse spin and momentum in two-wave interference, Phys. Rev. X 5 (2015) 011039.
[55] M. Neugebauer, T. Bauer, A. Aiello, P. Banzer, Measuring the transverse spin density of light, Phys. Rev. Lett. 114 (2015) 063901.
[56] K.Y. Bliokh, D. Smirnova, F. Nori, Quantum spin Hall effect of light, Science 348 (2015) 1448–1451.
[57] C. Sayrin, C. Junge, R. Mitsch, B. Albrecht, D. O’Shea, P. Schneeweiss, J. Volz, A. Rauschenbeutel, Optical diode based on the chirality of guided photons,
arXiv:1502.01549v1.
[58] A. Aiello, P. Banzer, Transverse spin of light for all wavefields, arXiv:1502.05350v1.
[59] K.-Y. Kim, A.X. Wang, Spin angular momentum of surface modes from the perspective of optical power flow, Opt. Lett. 40 (2015) 2929–2932.
[60] L.D. Landau, E.M. Lifshitz, Mechanics, third ed., Butterworth-Heinmann, Oxford, 1976.
[61] M.A. Player, Angular momentum balance and transverse shift on reflection of light, J. Phys. A: Math. Gen. 20 (1987) 3667–3678.
[62] V.G. Fedoseyev, Conservation laws and transverse motion of energy on reflection and transmission of electromagnetic waves, J. Phys. A: Math. Gen.
21 (1988) 2045–2059.
[63] M. Onoda, S. Murakami, N. Nagaosa, Hall effect of light, Phys. Rev. Lett. 93 (2004) 083901.
[64] K.Y. Bliokh, Y.P. Bliokh, Conservation of angular momentum, transverse shift, and spin Hall effect in reflection and refraction of an electromagnetic
wave packet, Phys. Rev. Lett. 96 (2006) 073903.
[65] K.Y. Bliokh, Geometrical optics of beams with vortices: Berry phase and orbital angular momentum Hall effect, Phys. Rev. Lett. 97 (2006) 043901.
[66] O. Hosten, P. Kwiat, Observation of the spin Hall effect of light via weak measurements, Science 319 (2008) 787–790.
[67] K.Y. Bliokh, A. Niv, V. Kleiner, E. Hasman, Geometrodynamics of spinning light, Nature Photon. 2 (2008) 748–753.
[68] A. Aiello, J.P. Woerdman, Role of beam propagation in Goos–Hänchen and Imbert–Fedorov shifts, Opt. Lett. 33 (2008) 1437–1439.
[69] V.G. Fedoseyev, Transformation of the orbital angular momentum at the reflection and transmission of a light beam on a plane interface, J. Phys. A:
Math. Theor. 41 (2008) 505202.
[70] K.Y. Bliokh, I.V. Shadrivov, Y.S. Kivshar, Goos–Hanchen and Imbert–Fedorov shifts of polarized vortex beams, Opt. Lett. 34 (2009) 389–391.
[71] M. Merano, N. Hermosa, J.P. Woerdman, A. Aiello, How orbital angular momentum affects beam shifts in optical reflection, Phys. Rev. A 82 (2010)
023817.
36 K.Y. Bliokh, F. Nori / Physics Reports 592 (2015) 1–38

[72] Y. Gorodetski, K.Y. Bliokh, B. Stein, C. Genet, N. Shitrit, V. Kleiner, E. Hasman, T.W. Ebbesen, Weak measurements of light chirality with a plasmonics
slit, Phys. Rev. Lett. 109 (2012) 013901.
[73] K.Y. Bliokh, A. Aiello, Goos–Hanchen and Imbert–Fedorov beam shifts: an overview, J. Opt. 15 (2013) 014001.
[74] S.Y. Lee, I.-M. Lee, J. Park, S. Oh, W. Lee, K.-Y. Kim, B. Lee, Role of magnetic induction currents in nanoslit excitation of surface plasmon polaritons,
Phys. Rev. Lett. 108 (2012) 213907.
[75] F.J. Rodríguez-Fortuño, G. Marino, P. Ginzburg, D. O’Connor, A. Martinez, G.A. Wurtz, A.V. Zayats, Near-field interference for the unidirectional
excitation of electromagnetic guided modes, Science 340 (2013) 328–330.
[76] A.E. Miroshnichenko, Y.S. Kivshar, Polarization traffic control for surface plasmons, Science 340 (2013) 283–284.
[77] J. Petersen, J. Volz, A. Rauschenbeutel, Chiral nanophotonic waveguide interface based on spin–orbit interaction of light, Science 346 (2014) 67–71.
[78] B. le Feber, N. Rotenberg, L. Kuipers, Nanophotonic control of circular dipole emission: towards a scalable solid-state to flying-qubits interface, Nature
Commun. 6 (2015) 6695.
[79] L. Marrucci, Spin gives direction, Nature Photon. 11 (2015) 9–10.
[80] I. Söllner, S. Mahmoodian, S.L. Hansen, L. Midolo, A. Javadi, G. Kirsanske, T. Pregnolato, H. El-Ella, E.H. Lee, J.D. Song, S. Stobbe, P. Lodahl, Deterministic
photon-emitter coupling in chiral photonic circuits, Nature Nanotechnol. http://dx.doi.org/10.1038/NNANO.2015.159.
[81] A.B. Young, A. Thijssen, D.M. Beggs, L. Kuipers, J.G. Rarity, R. Oulton, Polarization engineering in photonic crystal waveguides for spin-photon
entanglers, arXiv:1406.0714v1.
[82] Y. Lefier, T. Grosjean, Unidirectional sub-diffraction waveguiding based on optical spin–orbit coupling in subwavelength plasmonic waveguides, Opt.
Lett. 40 (2015) 2890–2893.
[83] H. Pichler, T. Ramos, A.J. Daley, P. Zoller, Quantum optics of chiral spin networks, Phys. Rev. A 91 (2015) 042116.
[84] A.A. Izmest’ev, Classical theory of wave beams, Sov. Phys. J. 14 (1971) 77–80. Translated from A.A. Izmestev, Izvestiya Vysshikh. Uchebnykh Zavedenii
Fizika 1, 101–105 (1971).
[85] L.D. Barron, Molecular Light Scattering and Optical Activity, second ed., Cambridge University Press, Cambridge, 2004.
[86] Y. Tang, A.E. Cohen, Optical chirality and its interaction with matter, Phys. Rev. Lett. 104 (2010) 163901.
[87] K.Y. Bliokh, F. Nori, Characterizing optical chirality, Phys. Rev. A 83 (2011) 021803(R).
[88] R.P. Cameron, S.M. Barnett, A.M. Yao, Optical helicity, optical spin, and related quantities in electromagnetic theory, New J. Phys. 14 (2012) 053050.
[89] K.Y. Bliokh, Y.S. Kivshar, F. Nori, Magnetoelectric effects in local light–matter interactions, Phys. Rev. Lett. 113 (2014) 033601.
[90] J.D. Jackson, Classical Electrodynamics, third ed., Wiley, 1999.
[91] M. Born, E. Wolf, Principles of Optics, seventh ed., Pergamon, London, 2005.
[92] K.Y. Bliokh, A.Y. Bekshaev, A.G. Kofman, F. Nori, Photon trajectories, anomalous velocities, and weak measurements: a classical interpretation, New
J. Phys. 15 (2013) 073022.
[93] G.N. Afanasiev, Y.P. Stepanovsky, The helicity of the free electromagnetic field and its physical meaning, Nuovo Cimento A 109 (1996) 271–279.
[94] J.L. Trueba, A.F. Rañada, The electromagnetic helicity, Eur. J. Phys. 17 (1996) 141–144.
[95] S.M. Barnett, R.P. Cameron, A.M. Yao, Duplex symmetry and its relation to the conservation of optical helicity, Phys. Rev. A 86 (2012) 013845.
[96] D.L. Andrews, M.M. Coles, Measures of chirality and angular momentum in the electromagnetic field, Opt. Lett. 37 (2012) 3009–3011.
[97] I. Fernandez-Corbaton, X. Zambrana-Puyalto, G. Molina-Terriza, Helicity and angular momentum: A symmetry-based framework for the study of
light–matter interactions, Phys. Rev. A 86 (2012) 042103.
[98] E. Hendry, R.V. Mikhaylovskiy, L.D. Barron, M. Kadodwala, T.J. Davis, Chiral electromagnetic fields generated by arrays of nanoslits, Nano Lett. 12
(2012) 3640–3644.
[99] M. Schäferling, D. Dregely, M. Hentschel, H. Giessen, Tailoring enhanced optical chirality: design principles for chiral plasmonic nanostructures, Phys.
Rev. X 2 (2012) 031010.
[100] C. Rosales-Guzmán, K. Volke-Sepulveda, J.P. Torres, Opt. Lett. 37 (2012) 3486–3488.
[101] J.S. Choi, M. Cho, Limitations of a superchiral field, Phys. Rev. A 86 (2012) 063834.
[102] I. Fernandez-Corbaton, X. Zambrana-Puyalto, N. Tischler, X. Vidal, M.L. Juan, G. Molina-Terriza, Electromagnetic duality symmetry and helicity
conservation for the macroscopic Maxwell’s equations, Phys. Rev. Lett. 111 (2013) 060401.
[103] A. Canaguier-Durand, J.A. Hutchison, C. Genet, T.W. Ebbesen, Mechanical separation of chiral dipoles by chiral light, New J. Phys. 15 (2013) 123037.
[104] S.J. van Enk, The covariant description of electric and magnetic field lines of null fields: application to Hopf-Rañada solutions, J. Phys. A: Math. Theor.
46 (2013) 175204.
[105] J. Dressel, K.Y. Bliokh, F. Nori, Spacetime algebra as a powerful tool for electromagnetism, Phys. Rep. 589 (2015) 1–71.
[106] M.G. Calkin, An invariance property of the free electromagnetic field, Amer. J. Phys. 33 (1965) 958–960.
[107] D. Zwanziger, Quantum field theory of particles with both electric and magnetic charges, Phys. Rev. 176 (1968) 1489–1495.
[108] S. Deser, C. Teitelboim, Duality transformations of Abelian and non-Abelian gauge fields, Phys. Rev. 13 (1976) 1592–1597.
[109] M.K. Gaillard, B. Zumino, Duality rotations for interacting fields, Nuclear Phys. B 193 (1981) 221–244.
[110] S.M. Barnett, On the six components of optical angular momentum, J. Opt. 13 (2011) 064010.
[111] R.P. Cameron, S.M. Barnett, Electric-magnetic symmetry and Noether’s theorem, New J. Phys. 14 (2012) 123019.
[112] K.Y. Bliokh, F. Nori, Relativistic Hall effect, Phys. Rev. Lett. 108 (2012) 120403.
[113] M. Stone, V. Dwivedi, T. Zhou, Wigner translations and the observer-dependence of the position of massless spinning particles, Phys. Rev. Lett. 114
(2015) 210402.
[114] I. Bialynicki-Birula, Photon wave function, Prog. Opt. 36 (1996) 245–294.
[115] S.M. Barnett, Optical angular-momentum flux, J. Opt. B: Quantum Semiclass. Opt 4 (2002) S7–S16.
[116] R.M.A. Azzam, N.M. Bashara, Ellipsometry and Polarized Light, North-Holland, 1977.
[117] C. Schwartz, A. Dogariu, Conservation of angular momentum of light in single scattering, Opt. Express 14 (2006) 8425–8433.
[118] Z. Bomzon, M. Gu, Space-variant geometrical phases in focused cylindrical light beams, Opt. Lett. 32 (2007) 3017–3019.
[119] T.A. Nieminen, A.B. Stilgoe, N.R. Heckenberg, H. Rubinsztein-Dunlop, Angular momentum of a strongly focused Gaussian beam, J. Opt. A: Pure Appl.
Opt. 10 (2008) 115005.
[120] P.B. Monteiro, P.A.M. Neto, H.M. Nussenzveig, Angular momentum of focused beams: Beyond the paraxial approximation, Phys. Rev. A 120 (2009)
033830.
[121] C.-F. Li, Spin and orbital angular momentum of a class of nonparaxial light beams having a globally defined polarization, Phys. Rev. A 80 (2009)
063814.
[122] O.G. Rodríguez-Herrera, D. Lara, K.Y. Bliokh, E.A. Ostrovskaya, C. Dainty, Optical nanoprobing via spin–orbit interaction of light, Phys. Rev. Lett. 104
(2010) 253601.
[123] K.Y. Bliokh, E.A. Ostrovskaya, M.A. Alonso, O. Rodríguez-Herrera, D. Lara, C. Dainty, Spin-to-orbital angular momentum conversion in focusing,
scattering, and imaging systems, Opt. Express 19 (2011) 26132–26149.
[124] Y. Gorodetski, A. Niv, V. Kleiner, E. Hasman, Observation of the spin-based plasmonic effect in nanoscale structures, Phys. Rev. Lett. 101 (2008)
043903.
[125] M.A. Alonso, The effect of orbital angular momentum and helicity in the uncertainty-type relations between focal spot size and angular spread,
J. Opt. 13 (2011) 064016.
[126] H. Bacry, Localizability and Space in Quantum Physics, Springer, Berlin, 1988.
[127] B.-S. Skagerstam, Lozalization of massless spinning particles and the Berry phase, arXiv:9210054v1.
[128] N.B. Simpson, K. Dholakia, L. Allen, M.J. Padgett, Mechanical equivalence of spin and orbital angular momentum of light: an optical spanner, Opt.
Lett. 22 (1997) 52–54.
[129] A. Afanasev, C.E. Carlson, A. Mukherjee, Off-axis excitation of hydrogenlike atoms by twisted photons, Phys. Rev. A 88 (2013) 033841.
 K.Y. Bliokh, F. Nori / Physics Reports 592 (2015) 1–38 37

[130] A. Afanasev, C.E. Carlson, A. Mukherjee, Two properties of twisted-light absorption, J. Opt. Soc. Amer. B 31 (2014) 2721–2727.
[131] S.M. Barnett, M.V. Berry, Superweak momentum transfer near optical vortices, J. Opt. 15 (2013) 125701.
[132] A. Ashkin, J.P. Gordon, Stability of radiation-pressure particle traps: an optical Earnshaw theorem, Opt. Lett. 8 (1983) 511–513.
[133] P.C. Chaumet, M. Nieto-Vesperinas, Time-averaged total force on a dipolar sphere in an electromagnetic field, Opt. Lett. 25 (2000) 1065–1067.
[134] A. Canaguier-Durand, A. Cuche, C. Genet, T.W. Ebbesen, Force and torque on an electric dipole by spinning light fields, Phys. Rev. A 88 (2013) 033831.
[135] A. Ashkin, Optical Trapping and Manipulation of Neutral Particles Using Lasers, World Scientific, Singapore, 2006.
[136] D.G. Grier, A revolution in optical manipulation, Nature 424 (2003) 810–816.
[137] M. Dienerowitz, M. Mazilu, K. Dholakia, Optical manipulation of nanoparticles: a review, J. Nanophoton. 2 (2008) 021875.
[138] R.W. Bowman, M.J. Padgett, Optical trapping and binding, Rep. Progr. Phys. 76 (2013) 026401.
[139] M. Aspelmeyer, T.J. Kippenberg, F. Marquardt, Cavity optomechanics, Rev. Modern Phys. 86 (2014) 1391–1452.
[140] S. Huard, C. Imbert, Measurement of exchanged momentum during interaction between surface-wave and moving atom, Opt. Commun. 24 (1978)
185–189.
[141] T. Matsudo, Y. Takahara, H. Hori, T. Sakurai, Pseudomomentum transfer from evanescent waves to atoms measured by saturated absorption
spectroscopy, Opt. Commun. 145 (1998) 64–68.
[142] J. Arlt, Handedness and azimuthal energy flow of optical vortex beams, J. Modern Opt. 50 (2006) 1573–1580.
[143] J. Leach, S. Keen, M.J. Padgett, C. Saunter, G.D. Love, Direct measurements of the skew angle of the Poynting vector in a helically phased beam, Opt.
Express 14 (2006) 11919–11924.
[144] F.A. Starikov, G.G. Kochemasov, S.M. Kulikov, et al., Wavefront reconstruction of an optical vortex by a Hartmann-Shack sensor, Opt. Lett. 32 (2007)
2291–2293.
[145] R.A. Terborg, K. Volke-Sepúlveda, Quantitative characterization of the energy circulation in helical beams by means of near-field diffraction, Opt.
Express 21 (2013) 3379–3387.
[146] M. Yeganeh, S. Rasouli, M. Dashti, S. Slussarenko, E. Santamato, E. Karimi, Reconstruction the Poynting vector skew angle and wavefront of optical
vortex beams via two-channel moire deflectometery, Opt. Lett. 38 (2013) 887–889.
[147] S. Kocsis, B. Braverman, S. Ravets, M.J. Stevens, R.P. Mirin, L.K. Shalm, A.M. Steinberg, Observing the average trajectories of single photons in a two-slit
interferometer, Science 332 (2011) 1170–1173.
[148] Y. Tang, A.E. Cohen, Enhanced enantioselectivity in excitation of chiral molecules by superchiral light, Science 332 (2011) 333–336.
[149] E. Hendry, T. Carpy, J. Johnston, M. Popland, R.V. Mikhaylovskiy, A.J. Lapthorn, S.M. Kelly, L.D. Barron, N. Gadegaard, M. Kadodwala, Ultrasensitive
detection and characterization of biomolecules using superchiral fields, Nature Nanotechnol. 5 (2010) 783–787.
[150] N. Yang, A.E. Cohen, Local geometry of electromagnetic fields and its role in molecular multipole transitions, J. Phys. Chem. B 115 (2011) 5304–5311.
[151] R. Mathevet, G.L.J.A. Rikken, Magnetic circular dichroism as a local probe of the polarization of a focused Gaussian beam, Opt. Materials Express 4
(2014) 2574–2585.
[152] C. Cohen-Tannoudji, Atoms in Electromagnetic Fields, World Scientific, Singapore, 2004.
[153] T. Kawalec, L. Jozefowski, J. Fiutowski, M.J. Kasprowicz, T. Dohnalik, Spectroscopic measurements of the evanescent wave polarization state, Opt.
Commun. 274 (2007) 341–346.
[154] R. Mitsch, C. Sayrin, B. Albrecht, P. Schneeweiss, A. Rauschenbeutel, Exploiting the local polarization of strongly confined light for sub-micrometer-
resolution internal state preparation and manipulation of cold atoms, Phys. Rev. A 89 (2014) 063829.
[155] F. de Fornel, Evanescent Waves: From Newtonian Optics to Atomic Physics, Springer, 2001.
[156] I.V. Shadrivov, A.A. Sukhorukov, Y.S. Kivshar, A.A. Zharov, A.D. Boardman, P. Egan, Nonlinear surface waves in left-handed materials, Phys. Rev. E 69
(2004) 016617.
[157] A.V. Zayats, I.I. Smolyaninov, A.A. Maradudin, Nano-optics of surface plasmon polaritons, Phys. Rep. 408 (2005) 131–314.
[158] C. Junge, D. O’Shea, J. Volz, A. Rauschenbeutel, Strong coupling between single atoms and nontransversal photons, Phys. Rev. Lett. 110 (2013) 213604.
[159] C.M. Bender, Making sense of non-Hermitian Hamiltonians, Rep. Progr. Phys. 70 (2007) 947–1018.
[160] S. Chang, S.S. Lee, Optical torque exerted on a sphere in the evanescent field of a circularly-polarized Gaussian laser beam, Opt. Commun. 151 (1998)
286–296.
[161] Wikipedia: Surface plasmon polariton, http://en.wikipedia.org/wiki/Surface_plasmon_polariton.
[162] K.G. Lee, H.W. Kihm, J.E. Kihm, et al., Vector field macroscopic imaging of light, Nature Photon. 1 (2007) 53–56.
[163] A.E. Siegman, Lasers, University Science Books, Mill Valey, 1986.
[164] M. Burresi, R.J.P. Engelen, A. Opheij, D. van Oosten, D. Mori, T. Baba, L. Kuipers, Observation of polarization singularities at the nanoscale, Phys. Rev.
Lett. 102 (2009) 033902.
[165] M. Schnell, A. Garcia-Etxarri, J. Alkorta, J. Aizpurua, R. Hillenbrand, Phase-resolved mapping of the near-field vector and polarization state in nanoscale
antenna gaps, Nano Lett. 10 (2010) 3524–3528.
[166] T. Grosjean, I.A. Ibrahim, M.A. Suarez, G.W. Burr, M. Mivelle, D. Charraut, Full vectorial imaging of electromagnetic light at subwavelength scale, Opt.
Express 18 (2010) 5809–5824.
[167] T. Bauer, S. Orlov, U. Peschel, P. Banzer, G. Leuchs, Nanointerferometric amplitude and phase reconstruction of tightly focused vector beams, Nature
Photon. 8 (2014) 23–27.
[168] Y.G. Song, S. Chang, J.H. Jo, Optically induced rotation of combined Mie particles within an evanescent field of a Gaussian beam, Japan. J. Appl. Phys.
38 (1999) L380–L383.
[169] M. Neugebauer, T. Bauer, P. Banzer, G. Leuchs, Polarization tailored light driven directional optical nanobeacon, Nano Lett. 14 (2014) 2546–2551.
[170] M.Z. Hasan, C.L. Kane, Topological insulators, Rev. Modern Phys. 82 (2010) 3045–3067.
[171] X.-L. Qi, S.-C. Zhang, Topological insulators and superconductors, Rev. Modern Phys. 83 (2011) 1057–1110.
[172] L. Lu, J.D. Joannopoulos, M. Soljacic, Topological photonics, Nature Photon. 8 (2011) 821–829.
[173] V.S. Liberman, B.Y. Zel’dovich, Spin–orbit interaction of a photon in an inhomogeneous medium, Phys. Rev. A 46 (1992) 5199–5207.
[174] K.Y. Bliokh, A. Aiello, M.A. Alonso, Spin–orbit interactions of light in isotropic media, in: D.L. Andrews, M. Babiker (Eds.), The Angular Momentum of
Light, Cambridge University Press, 2013.
[175] K.Y. Bliokh, F.J. Rodríguez-Fortuño, F. Nori, A.V. Zayats, Spin–orbit interactions of light, arXiv:1505.02864.
[176] A.V. Dooghin, N.D. Kundikova, V.S. Liberman, B.Y. Zel’dovich, Optical Magnus effect, Phys. Rev. A 45 (1992) 8204–8208.
[177] N.B. Baranova, A.Y. Savchenko, B.Y. Zel’dovich, Transverse shift of a focal spot due to switching of the sign of circular-polarization, JETP Lett. 59 (1994)
232–234.
[178] B.Y. Zel’dovich, N.D. Kundikova, L.F. Rogacheva, Observed transverse shift of a focal spot upon a change in the sign of circular polarization, JETP Lett.
59 (1994) 766–769.
[179] V.G. Fedoseyev, Spin-independent transverse shift of the centre of gravity of a reflected and of a refracted light beam, Opt. Commun. 193 (2001)
9–18.
[180] K.Y. Bliokh, Y.P. Bliokh, Topological spin transport of photons: the optical Magnus effect and Berry phase, Phys. Lett. A 333 (2004) 181–186.
[181] R. Dasgupta, P.K. Gupta, Experimental observation of spin-independent transverse shift of the centre of gravity of a reflected Laguerre–Gaussian light
beam, Opt. Commun. 257 (2006) 91–96.
[182] K.Y. Bliokh, Y. Gorodetski, V. Kleiner, E. Hasman, Coriolis effect in optics: Unified geometric phase and spin-Hall effect, Phys. Rev. Lett. 101 (2008)
030404.
[183] D. Haefner, S. Sukhov, A. Dogariu, Spin Hall Effect of light in spherical geometry, Phys. Rev. Lett. 102 (2009) 123903.
[184] V. Garbin, G. Volpe, E. Ferrari, M. Versluis, D. Cojoc, D. Petrov, Mie scattering distinguishes the topological charge of an optical vortex: a homage to
Gustav Mie, New J. Phys. 11 (2009) 013046.
[185] N. Shitrit, I. Bretner, Y. Gorodetski, V. Kleiner, E. Hasman, Optical spin Hall effects in plasmonic chains, Nano Lett. 11 (2011) 2038–2042.
38 K.Y. Bliokh, F. Nori / Physics Reports 592 (2015) 1–38

[186] F.I. Fedorov, To the theory of total reflection, Doklady Akademii Nauk SSSR 105 (1955) 465–468. translated and reprinted in J. Opt. 15, 014002 (2013)].
[187] C. Imbert, Calculation and experimental proof of the transverse shift induced by total internal reflection of a circularly polarized light beam, Phys.
Rev. D 5 (1972) 787–796.
[188] M.R. Dennis, J.B. Götte, The analogy between optical beam shifts and quantum weak measurements, New J. Phys. 14 (2012) 073013.
[189] Y. Aharonov, D.Z. Albert, L. Vaidman, How the results of a measurement of a component of the spin of a spin-1/2 particle can turn out to be 100,
Phys. Rev. Lett. 11 (1988) 1351–1354.
[190] I.M. Duck, P.M. Stevenson, E.C.G. Sudarshan, The sense in which a weak measurement of a spin-1/2 particle’s spin component yields a values 100,
Phys. Rev. D 40 (1989) 2112–2117.
[191] Y. Aharonov, S. Popescu, J. Tollaksen, A time-symmetric formulation of quantum mechanics, Phys. Today (2010) 27–32. November.
[192] A.G. Kofman, S. Ashhab, F. Nori, Nonperturbative theory of weak pre- and post-selected measurements, Phys. Rep. 520 (2010) 43–133.
[193] J. Dressel, M. Malik, F.M. Miatto, A.N. Jordan, R.W. Boyd, Understanding quantum weak values: Basics and applications, Rev. Modern Phys. 86 (2014)
307–316.
[194] A.Y. Bekshaev, Polarization-dependent transformation of a paraxial beam upon reflection and refraction: A real-space approach, Phys. Rev. A 85
(2012) 023842.
[195] Y.A Kravtsov, Y.I. Orlov, Geometrical Optics of Inhomogeneous Media, Springer, Berlin, 1990.
[196] K.Y. Bliokh, Geometrodynamics of polarized light: Berry phase and spin Hall effect in a gradient-index medium, J. Opt. A: Pure Appl. Opt. 11 (2009)
094009.
[197] S. Murakami, N. Nagaosa, S.-C. Zhang, Dissipationless quantum spin current at room temperature, Science 301 (2003) 1348–1351.
[198] A. Bérard, H. Mohrbach, Spin Hall effect and Berry phase of spinning particles, Phys. Lett. A 352 (2006) 190–195.
[199] C. Duval, Z. Horváth, P.A. Horváthy, Fermat principle for spinning light, Phys. Rev. D 74 (2006) 021701(R).
[200] A.Y. Bekshaev, Oblique section of a paraxial light beam: criteria for azimuthal energy flow and orbital angular momentum, J. Opt. A: Pure Appl. Opt.
11 (2009) 094003.
[201] J. Korger, et al., Observation of the geometric spin Hall effect of light, Phys. Rev. Lett. 112 (2014) 113902.
[202] H. Wolter, Zur frage des lichtweges bei totalreflexion, Z. Naturf. 5 (1950) 276–283. translated to English and reprinted as H. Wolter, Concerning the
path of light upon total reflection, J. Opt. A: Pure Appl. Opt. 11, 090401 (2009).
[203] W. Braunbek, G. Laukien, Einzelheiten zur halbebenen-beugung, Optik 9 (1952) 174–179.
[204] A. Boivin, J. Dow, E. Wolf, Energy flow in the neighborhood of the focus of a coherent beam, J. Opt. Soc. Amer. 57 (1967) 1171–1175.
[205] V.A. Pas’ko, M.S. Soskin, M.V. Vasnetsov, Transverse optical vortex, Opt. Commun. 198 (2001) 49–56.
[206] M.R. Dennis, J.B. Götte, Beam shifts for pairs of plane waves, J. Opt. 15 (2013) 014015.
[207] L.D. Landau, E.M. Lifshitz, Classical Theory of Fields, fourth ed., Butterworth-Heinemann, Amsterdam, 1994.
[208] W. Rindler, Introduction to Special Relativity, Clarendon Press, Oxford, 1982.
[209] R.A. Muller, Thomas precession: Where is the torque?, Amer. J. Phys. 60 (1992) 313–317.
[210] J.-Y. Chen, D.T. Son, M.A. Stephanov, H.-U. Yee, Y. Yin, Lorentz invariance in chiral kinetic theory, Phys. Rev. Lett. 113 (2014) 182302.
[211] C. Duval, P.A. Horváthy, Chiral fermions as classical massless particles, Phys. Rev. D 91 (2015) 045013.
[212] C. Duval, M. Elbistan, P.A. Horváthy, P.-M. Zhang, Wigner-Souriau translations and Lorentz symmetry of chiral fermions, Phys. Lett. B 742 (2015)
322–326.
[213] A.P. Sukhorukov, V.V. Yangirova, Spatio-temporal vortices: properties, generation and recording, Proc. SPIE 5949 (2005) 594906.
[214] J. Masajada, B. Dubik, Optical vortex generation by three plane wave interference, Opt. Commun. 198 (2001) 21–27.
[215] M. Harwit, Photon orbital angular momentum in atsrophysics, Astrophys. J. 597 (2003) 1266–1270.
[216] G.C.G. Berkhout, M.W. Beijersbergen, Method for probing the orbital angular momentum of optical vortices in electromagnetic waves from
astronomical objects, Phys. Rev. Lett. 216 (2008) 100801.
[217] F. Tamburini, B. Thidé, G. Molina-Terriza, G. Anzolin, Twisting of light around rotating black holes, Nat. Phys. 7 (2011) 195–197.