IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 56, NO.

3, MARCH 2008 747

Dyadic Green’s Functions for an Anisotropic,

Non-Local Model of Biased Graphene
George W. Hanson, Senior Member, IEEE

Abstract—Dyadic Green’s functions are presented for an rather then quadratic as in most materials, so that electrons
anisotropic surface conductivity model of biased graphene. The in graphene behave as massless relativistic particles (Dirac
graphene surface can be biased using either a perpendicular fermions) with an energy-independent velocity. The linear
static electric field, or by a static magnetic field via the Hall
energy bands lead to interesting quantum properties, and to a
effect. The graphene is represented by an infinitesimally-thin,
two-sided, non-local anisotropic conductivity surface, and the minimum conductivity even when charge carrier concentrations
field is obtained in terms of Sommerfeld integrals. The role of vanish [4].
spatial dispersion is accessed, and the effect of various static bias In this work, the interaction of an electromagnetic current
fields on electromagnetic field behavior is examined. It is shown source and electrically or magnetically biased graphene at the
that by varying the bias one can exert significant control over interface between two materials is examined. Several situations
graphene’s electromagnetic propagation characteristics, including are discussed: I) the role of spatial dispersion (non-locality),
guided surface wave phenomena, which may be useful for future
electronic and photonic device applications.
which leads to a tensor conductivity; II) effects due to electro-
static bias fields, which result in a scalar conductivity; and III)
Index Terms—Dyadic Green’s functions, electromagnetic effects due to magnetostatic bias fields, which lead to a tensor
theory, nanotechnology.
conductivity. Some simple expressions are obtained for surface
wave dispersion of graphene in a homogeneous medium.
I. INTRODUCTION The paper is organized as follows. In the body of the paper,
dyadic Green’s functions for the structure are presented, along

G RAPHENE, which is a planar atomic layer of carbon

atoms bonded in a hexagonal structure, is a very
promising material in emerging nanoelectronic applications
with associated results. Appendix I provides the formulas
for graphene’s conductivity. In Appendix II, the intraband,
spatially-dispersive conductivity is derived, and Appendix III
[1]. Graphene is the two-dimensional version of graphite, and briefly presents an alternative dyadic Green’s function formu-
is related to carbon nanotubes in that a single-wall carbon lation that was found to be useful. This work follows that in
nanotube can be thought of as a graphene sheet rolled into a [11], where an isotropic model of graphene was considered. All
cylinder [2]. units are in the SI system, and the time variation (suppressed)
Graphene’s band structure, together with it’s extreme thin- is , where is the imaginary unit.
ness, leads to pronounced electric field and Hall effects [3]–[7],
and to the possibility of various electronic devices [8], [9]. In-
II. FORMULATION OF THE MODEL
trinsic graphene is a zero bandgap semiconductor, and its con-
ductivity can be tuned by either electrostatic or magnetostatic
A. Electronic Model of Graphene
gating. Since this is the governing principle behind traditional
semiconductor devices, this effect in graphene is particularly Fig. 1 depicts laterally-infinite graphene lying in the
promising for the development of ultrathin carbon nanoelec- plane at the interface between two different mediums character-
tronic devices. Although both the electric field effect and Hall ized by , for and , for , where all material
effect can occur in atomically thin metal films, these tend to be parameters may be complex-valued.
thermodynamically unstable, and don’t form continuous layers The graphene sheet is modeled as an infinitesimally-thin,
with good transport properties. In contrast, graphene is stable, non-local two-sided surface characterized by a surface con-
and, like its cylindrical carbon nanotube versions, can exhibit ductivity tensor obtained from microscopic, semi-classical and
ballistic transport over at least submicron distances [3]. quantum mechanical considerations. This model of graphene
Only recently has it become possible to fabricate actual follows from the two-sided conductivity surface approach
graphene [3]–[7], although it has long been theoretically developed in [12] for carbon nanotubes, and was applied to
studied as a way to explain properties of graphite [10], and local isotropic graphene in [11]. Here the model is extended to
later, carbon nanotubes [2]. In graphene, the energy-momentum the non-local anisotropic case
relationship for electrons is linear over a wide range of energies,

(1)
Manuscript received July 23, 2007; revised October 7, 2007. where is radian frequency, is the chemical potential [which
The author is with the Department of Electrical Engineering, Uni- can be controlled by an applied electrostatic bias field
versity of Wisconsin-Milwaukee, Milwaukee, WI 53211 USA (e-mail: , or by doping; see the discussion of (51)], is a phe-
[email protected]).
Color versions of one or more of the figures in this paper are available online nomenological electron scattering rate that is assumed to be in-
at http://ieeexplore.ieee.org. dependent of energy, is temperature, and is an ap-
Digital Object Identifier 10.1109/TAP.2008.917005 plied magnetostatic bias field. In the following, the dependence
0018-926X/$25.00 © 2008 IEEE
748 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 56, NO. 3, MARCH 2008

B. Dyadic Green’s Function for a Surface Model of Graphene

In this section, a method for obtaining dyadic Green’s func-
tions for a general two-dimensional anisotropic surface will be
presented, although due to space limitations explicit results will
only be provided for the local Hall effect regime. The method
of obtaining the Green’s dyadics follows [13], adapted to the
surface impedance case, and the resulting Green’s functions are
believed to be new.
For any planarly layered, piecewise-constant medium, the
electric and magnetic fields in region due to an electric
current can be obtained as [14], [15]

(5)
(6)
Fig. 1. (a) Depiction of graphene (top view), where the small circles denote where and are the wavenumber and
carbon atoms, and (b) an anisotropic graphene sheet characterized by tensor
B
conductance  at the interface between two dielectrics (side view). andE electric Hertzian potential in region , respectively. Assuming
denote possible dc bias fields. that the current source is in region 1, then

(7)
on , , and will be suppressed. Three special cases of (1)
(8)
will be considered.
I) Spatial dispersion, but neither electrostatic nor magneto-
static bias . In this case the conductivity
components become operators (9)

where the underscore indicates a dyadic quantity, and where

is the support of the current. With parallel to the interface
normal, the principle Greens dyadic can be written as [14]

(10)
(2)

where , , and are derived from the semi-classical

Boltzmann’s equation in Appendix II (see (96)–(101) for
the final expressions). (11)
II) Electrostatic bias, although no magnetostatic bias nor
spatial dispersion ( , ). In this case the where
conductivity is a scalar
(12)

(3) (13)

where is given by (57) in Appendix I (equivalently, by and where is the unit dyadic.
(54), although this form converges very slowly as The scattered Green’s dyadics can be obtained by enforcing
). the boundary conditions
III) Magnetostatic bias, and possibly electrostatic bias, but no
spatial dispersion ( , and possibly ). We
refer to this as the local Hall effect regime, and in this (14)
case the conductivity tensor can be written as
where (A/m) and (V/m) are electric and magnetic surface
currents on the boundary. In our case, , and .
Introducing the two-dimensional Fourier transform
(4)

where the conductivities and are given by (54) and (15)

(55) in Appendix I.
HANSON: DYADIC GREEN’S FUNCTIONS FOR AN ANISOTROPIC, NON-LOCAL MODEL OF BIASED GRAPHENE 749

where the Sommerfeld integrals are

(16)

(14) becomes
(27)

(17) For an isotropic surface, . The Green’s dyadic for

region 2, , has the same form as for region 1, although
in (27) the replacement
(18)
(28)

must be made.
In the event of general media ( , for and ,
(19) for ) the coefficients are quite complicated
and will be omitted here due to space limitations. The results
for the special case of an isotropic surface conductivity (i.e., if
and spatial dispersion effects are ignored) are given
in [11]. Since spatial dispersion effects are shown to be small
(20) in the microwave regime, here explicit expressions will be pro-
vided for the special case of the graphene surface residing in
where a homogeneous space characterized by and
in the Hall regime with no spatial dispersion
(21) (i.e., using (4) as the conductivity). In this case, the coefficients
in region 1 are given as
In the case of an isotropic surface, a vertical current maintains
only a vertical potential, and a horizontal current induces both (29)
a horizontal and vertical potential [11], which is also the case
for an isotropic layered medium in the absence of surface con-
ductivity [13]. However, in the case of an anisotropic surface, (30)
a vertical current induces both a vertical and horizontal poten-
tial. That is, alone cannot satisfy the boundary conditions, (31)
and, therefore, the pair is relevant for both vertical cur-
rents and -directed horizontal currents, and the pair is (32)
relevant for vertical currents and -directed horizontal currents.
In each case the boundary conditions can be satisfied. The pair (33)
reduce (17)–(20) to
(34)
(22)
(23)

(35)

(24) (36)

where
(25)

with a similar reduction for the pair , where (37)

and .
Enforcing the boundary conditions and following the method
described in [13], the scattered Green’s dyadic is found to have
the form (38)

In the more general case of an arbitrary graphene conductivity

(26) tensor, the results are only moderately more complicated, as-
suming that the graphene resides in a homogeneous space.
750 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 56, NO. 3, MARCH 2008

The factor of 2 in the denominators of the equations for , siding in a homogeneous medium characterized by , , sur-
[i.e., (30), (33), and (36)] is a result of the com- face wave poles satisfy , where
ponent of current producing all three components of potential
, , and . Because there is no coupling of poten-
tial , one can derive the components
by using each pair and , adding the results,
and dividing by 2 (since the source has essentially been ap- (47)
plied twice). Alternatively, the components can be obtained
using the pair , where is the electric/magnetic In the Hall effect regime (i.e., using (4) as the conductivity),
Hertzian potential [15]. To verify the final electric and mag- simplifies to be in (38), which can be solved
netic fields obtained from the Hertzian potentials, the electric to yield
dyadic Green’s function was derived independent of potentials,
using the field method outlined in [16, Ch. 6], extended to ac-
count for cross-polarization terms (see Appendix III). Numer- (48)
ical tests verified that the final fields were the same in both cases
for the general graphene surface. Although the field method is
a bit more straightforward than the method of potentials pre- where . For both the special case of an isotropic
sented here, there are several reasons why the Hertzian potential sheet characterized by (3), and, since is an odd function of
method is desirable. First, the resulting potential Green’s com- , in the Hall regime when , we have , and in
ponents (e.g., presented above) are simpler than the electric these cases (48) becomes
field Green’s components, and allow for a clear interpretation of
the underlying wave physics. Second, in quantum mechanics the
potential plays a dominant role in governing electron dynamics, (49)
and so a potential formulation may facilitate further work in this
area.
In the lower region, the coefficients are These correspond to the transverse-magnetic (TM, -wave)
case and the transverse-electric (TE, -wave) case, respec-
tively, where transverse is with respect to the radial coordinate.
For isolated graphene characterized by complex surface con-
(39) ductivity , a proper TE surface wave exists if and
only if (associated with interband conductivity; see
(40) the discussion in Appendix I), and a proper TM surface wave
exists for (associated with intraband conductivity).
(41) Surface-wave behavior in the isotropic case for was
considered in [11].
(42)
(43)
III. RESULTS
(44)
In this section, the importance of spatial dispersion will be as-
(45)
sessed, and the effect of dc electric and magnetic bias on elec-
tromagnetic field behavior near a graphene surface in the mi-
(46)
crowave regime will be considered (due to space limitations,
THz and infrared frequencies will be considered elsewhere). In
In the general medium case, both waveparameters all cases, results will be presented at room temperature,
, , 2, lead to branch points at , , and ( ; ); the
and thus the -plane is a four-sheeted Riemann surface. The value of the scattering rate is chosen to be approximately the
standard hyperbolic branch cuts [15] that separate the one same as for electron-acoustic phonon interactions in single-wall
proper sheet (where , such that the radiation carbon nanotubes [17]. For simplicity, and to focus on the ef-
condition as is satisfied) and the three improper fects of the graphene surface, in the following numerical results
sheets (where ) are the same as in the absence of we will assume graphene in a homogeneous medium character-
surface conductivity. As in the usual layered medium theory, ized by , , and having wavenumber , the free-space
branch points (and the associated branch-cut integrals in a wavenumber. Double integrals (27) were computed in rectan-
spectral representation) are associated with radiation into the gular form using a Romberg integration routine. In the case of
surrounding medium. an isotropic surface impedance, it is convenient to convert to
polar form since the angular integral can be performed analyti-
cally.
C. Surface Waves Guided by Graphene
A. Effect of Spatial Dispersion
Pole singularities in the Sommerfeld integrals represent dis-
crete surface waves guided by the medium [14], [15]. For the In the spatial dispersive regime the conductivity has the form
case of graphene having an arbitrary conductivity tensor and re- (2), derived in Appendix II. It is easy to see that the terms asso-
HANSON: DYADIC GREEN’S FUNCTIONS FOR AN ANISOTROPIC, NON-LOCAL MODEL OF BIASED GRAPHENE 751

ciated with and are quite small below THz frequencies by

examining the ratio [see (96)–(101)]

(50)

in the Fourier transform domain, where the term ,

, arises from the spatial derivatives in (2), and electron ve-
locity is described in Appendix I.
The value of the wavenumbers and govern the size of
, and to access upper limits for the wavenumbers it is worth-
while to represent the spectral integrals (27) as a sum of branch
cuts and residues [14], [15], [18]. For propagating space-wave
radiation arising from the branch cuts, all of the spectral content Fig. 2. The relationship between electric bias field E and chemical potential
occurs for . If we also assume that surface-wave prop-  obtained from (53), with " = ".
agation constants from (47) are such that , then below
the THz regime where , ,
such that at 10 GHz (and obviously decreases as B. Effect of Electric Bias
frequency is lowered). Thus, in this case the derivative terms
associated with spatial dispersion in (2) can be ignored, and the In the electric bias regime the conductivity has the form (3).
conductivity can be modeled as isotropic in the absence of a The influence of the electric bias is incorporated into the
magnetic bias field. conductivity expressions via the chemical potential (through the
The assumption is reasonable at GHz frequencies Fermi-Dirac distribution; see Appendix I), and so a relation be-
and below, where the intraband conductivity having tween chemical potential and electrostatic bias must be estab-
strongly dominates over the interband conductivity having lished. The normal displacement vector on either side of
(see Appendix I). Thus, at these frequencies only TM surface a charged sheet in a homogeneous dielectric characterized by
waves may propagate, the conductivity is relatively large, and is , where is the two-dimensional
from, e.g., (49), . Thus, spatial dispersion effects seem surface-charge density and is the electronic charge. For an
to be unimportant below THz frequencies. The presence of a
isolated graphene sheet the chemical potential is determined
magnetic bias may change this conclusion, although even in the
by the extra carrier density as [19], [20]
case of a magnetic-bias induced TE surface wave (see Fig. 8),
, and so it is reasonable that the same conclusions
hold. (51)
In order to gauge the effect of spatial dispersion at THz fre-
quencies and above, a procedure different than that presented
in Appendix II is required, since in that treatment only the in- where
traband conductivity is considered. This conductivity becomes
comparable to, or smaller than the interband term above a few (52)
THz.
Assuming for the moment that the spatial dispersion analysis is the Fermi-Dirac distribution, is energy, and is Boltz-
holds above THz frequencies, it is easily seen that extremely
mann’s constant (for the undoped, ungated case, ).
slow surfaces waves may necessitate the inclusion of spatial dis-
persion. For the isotropic graphene surface, it is shown in [11] Thus, for a given
that in the near-infrared, in some cases only a TE surface wave
with will propagate, whereas in other situations only (53)
a TM surface wave with will exist. With ,
since , , where is the speed of light in
the medium. Assuming , the speed of light in a vacuum, can be solved for the chemical potential using a numer-
then , such that if (TE case), , and ical root search, or, more simply, can be directly obtained
if (a typical value in the TM case), . Obvi- for a given chemical potential (the carrier density can also be
ously, when the condition is violated, spatial dispersion controlled by chemical doping). The relationship between bias
effects will be important, and, since , the same com-
field and chemical potential in graphene obtained from (53)
ments apply to . Thus, the existence of very slow surface waves
is shown in Fig. 2.
may necessitate the inclusion of spatial dispersion.
In summary, below the THz frequency range spatial disper- It can be seen that the chemical potential can be easily tuned
sion effects seem to be unimportant. Spatial dispersion may be from 1 eV to 1 eV by typical values of bias fields that have
important for surface waves with , but this situation been used in experiments [3], [4]. In these measurements, the
seems to occur only for moderately high THz frequencies and graphene surface was supported by a dielectric substrate on a
above, for which the development presented in Appendix II is ground plane, with a voltage applied across the structure. In
not strictly valid. this case, the factor is replaced by , where is the
752 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 56, NO. 3, MARCH 2008

Fig. 3. Conductivity (57) as a function of applied bias field E at 10 GHz. For Fig. 4. Vertical electric field due to a vertical current (E ) and horizontal
E = = 0, = = 200:84 0 j 37:85. electric field due to a horizontal current (E ) from (5), as functions of position
along the graphene surface for two different bias fields E . The source is located
at (x ; y ; z ) = (0; 0; 1:67), and the observation point is (x; 0; 0).

substrate permittivity (typically ) and is the sub-

strate thickness (typically ). Thus, for the typical ex-
perimental parameters of in the range 0–100 V, is
in the range 0–1 V/nF, and so bias fields on the order of sev-
eral volts per nanometer are easily obtainable, and larger fields
can by achieved by using thinner substrates or higher voltages.
In this work, we do not have a grounded structure, and so a bias
field is assumed rather than a gate voltage, but in practice
this would be a gated structure where the bias voltage source
provides the excess charge carriers.
The isotropic conductivity (57) (i.e., (54) with ) as
a function of electric bias field is shown in Fig. 3, where
(S). It can be seen that con-
ductivity is significantly changed using moderate bias fields.
Fig. 4 shows the time-harmonic vertical electric field due to
a vertical current and the horizontal electric field due to a
horizontal current from (5), as functions of position along Fig. 5. Conductivity (4) as a function of magnetic bias B for E =
the graphene surface for two different bias fields . The source 0:0886 V=nm ( = 0:1 eV , solid and dashed lines) and for E = 0
( = 0, symbols; upper curve is  , lower curve is  , and  = 0). The
is located at , and the observation small filled circles at B = 0 are the values of  obtained from the simple
point is . From (53), for , , and for formula (58), valid in the absence of magnetic bias.
, . It can be seen that the bias has a
strong effect on the fields. The field behavior can be understood
by considering that as the bias increases, more charge exists on obtained from the simple formula (58), valid in the absence of
the surface by (53), and the surface conductivity increases. Thus, magnetic bias. It can be seen that the simple, low magnetic bias
as the surface becomes a better conductor, the horizontal field conductivities (58) and (62) are only valid for very small values
is shorted out, and the vertical field increases (obviously, in the of magnetic bias, since (58) is independent of and (62) is
limit , ). linear in . From the figure, it is clear that these will only hold
for .
C. Effect of Magnetic Bias The field components and corresponding to the con-
ductivity in Fig. 5 are shown in Fig. 6 for , 0.08 T,
In the magnetic bias (Hall effect) regime where , the and 1 T, where the electric bias is
conductivity has the form (4), with explicit expressions given by . As increases the conductivity tends to decrease in
(54) and (55). In Fig. 5 the conductivity is shown as a function magnitude, in which case increases and decreases.
of magnetic bias for two different values of electric bias, An interesting situation occurs for and .
and . The Since graphene is a zero-gap semiconductor, in the intrinsic case
conductivity is an odd function of and , and therefore , as in any intrinsic semiconductor, the Hall electron
as and for regardless of the current is equal and opposite to the Hall hole current, and so
value of . The small circles at are the values of the conductivity is a scalar, , given by (54). Fig. 7 shows the
HANSON: DYADIC GREEN’S FUNCTIONS FOR AN ANISOTROPIC, NON-LOCAL MODEL OF BIASED GRAPHENE 753

Fig. 8. Surface wave propagation constant for E =  = 0 ( = 0) at

10 GHz. For B < 0:0286 T,  < 0 (shown as the thin line), such that only
Fig. 6. Electric field components E and E corresponding to the conduc- a proper TM surface wave exists, and for B > 0:0286 T,  > 0, such that
=
tivity in Fig. 5 for B 0, B = 0:08 T, and B = 1 T, with E = only a proper TE surface wave exists.
0:0886 V=nm ( = 0:1 eV).

Fig. 9. Surface wave propagation constant at 10 GHz for various values of

chemical potential.
Fig. 7. Field components E and E for B = 0 and B = 1 T, when
E =  = 0 (scalar conductivity in the Hall regime).
surface wave is on the proper sheet. However,
at , and so there is a gap in the wavenumber at
difference in the and components for and this critical value of bias field.
when . A small electrostatic bias can change this behavior. In Fig. 9,
Surface wave propagation for the electrostatic bias case the result is shown, along with several values of chem-
was discussed in [11] and [21]. For the case of ical potential corresponding to small electrostatic bias values
magnetostatic bias, the surface wave propagation constant can ( for ,
be significantly affected by the bias, depending on the level of for , etc.). For the cases , conductivity is not
the chemical potential. In the case at a scalar , and the gap in the spectrum tends to disap-
10 GHz, the normalized surface wave propagation constant is pear as the chemical potential is increased.
shown in Fig. 8. In particular, referring to the conductivity in
Fig. 5 for (symbols), for , , IV. CONCLUSION
such that only a proper TM surface wave exists, and for Dyadic Green’s functions have been presented in terms of
, , such that only a proper TE surface Sommerfeld integrals for a non-local, anisotropic, two-sided
wave exists (the imaginary part of conductivity is also surface conductivity model of biased graphene. Either electric
shown in Fig. 8 as a guide). There is a gap in the spectrum or magnetic bias can be applied, and used to tune the conduc-
at the point where . This can be understood from the tivity of the graphene surface. The formulation is presented for
dispersion behavior of (49). For , an arbitrary conductivity tensor, and explicit expressions are
and , so that the TM surface wave is on the proper provided for the special case of graphene in a homogeneous
sheet, whereas the TE surface wave is on the improper sheet. medium in the Hall effect regime. It is shown that the effect
For , and , so that of both magnetostatic and electrostatic bias on field and surface
the TM surface wave is on the improper sheet, whereas the TE wave behavior in the microwave regime is substantial. This may
754 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 56, NO. 3, MARCH 2008

be useful for advanced applications where electronic control of sheet, and electronic interactions with surrounding dielectrics
electromagnetic properties is desirable. are ignored. Electromagnetic interactions with surrounding di-
electrics are included in the derived Green’s functions.
APPENDIX I
As a special case, in the low magnetic field limit for
CONDUCTIVITY COMPONENTS
[23, Eq. (13)]
The conductivities and in (3) and (4) can be determined
from the Kubo formalism [22], and explicit expressions from
[23, Eqs. (11) and (12)] (see also [24]–[29]) are

(57)

where the first term in (57) is due to intraband contributions, and

the second term to interband contributions. (57) is particularity
useful when , since in this case (54) converges very
slowly.
The intraband term in (57) can be evaluated as

(54)
(58)
and With , it can be seen that and
. The sign of the imaginary part of conductivity
plays an important role in the propagation of surface waves
guided by the graphene sheet [21].
The interband conductivity is on the order of [28], and at
room temperatures and for frequencies below the THz regime,
the interband conductivity is very small compared to the intra-
band term, and usually can be ignored. In general, the interband
term must be evaluated numerically, but it can be approximated
for as [32]

(55) (59)

where such that for and , with

. For and , is complex-
(56) valued, with

and where is the charge of an electron, is the re- (60)

duced Planck’s constant, is the Fermi-Dirac distribution (52),
is the electron’s energy-independent velocity
and for .
(due to graphene’s linear bands) [4], and is an excitonic en-
The Hall conductivity is [23, Eq. (14)]
ergy gap. This gap is opened due to electron interactions in the
presence of a magnetic field [30], [31], and can play a signifi-
cant role at low temperatures [25]. However, at higher temper-
ature the increased carrier density enhances screening effects,
which diminishes the gap [31]. Moreover, for the room tem-
perature results considered here, energy gaps less than several
hundred Kelvin (the expected range of values) are small com-
pared to thermal energy, and their numerical inclusion resulted (61)
in no observable change in the conductivity. Therefore, all nu-
merical results for which were computed using (54)
and (55) with . Furthermore, it should be noted that the As in (57), the first term is due to intraband contributions, and
conductivity expressions were derived for an isolated graphene the second term to interband contributions, with the interband
HANSON: DYADIC GREEN’S FUNCTIONS FOR AN ANISOTROPIC, NON-LOCAL MODEL OF BIASED GRAPHENE 755

term very small compared to the intraband term at room temper- The current density in the Fourier transform domain is
ature and for frequencies below the THz regime. The intraband
contribution can be evaluated as

(62) (70)

APPENDIX II since no current flows from the equilibrium distribution. The

CONDUCTIVITY TENSOR IN THE PRESENCE OF hole current is obtained in the same manner, and so the total
SPATIAL DISPERSION conductivity dyadic is
In this appendix the spatially-dispersive conductivity tensor
(2) is derived. Although the procedure follows a well-known (71)
general method [22, pp. 106-111] that has been applied to
carbon nanotubes in [12], the expression (2) is believed new
and, thus
and so it’s derivation is outlined here. The starting point is
Boltzmann’s equation
(72)
(63)
with . For the special case (i.e., neglecting
where is the electron distribution function (i.e., spatial dispersion), since and are odd in
the probability of finding an electron at position at time , and and , respectively [see (88)], and the limits of integration
having quasi-momentum ), and where is the are symmetric. In the presence of spatial dispersion, the material
equilibrium Fermi distribution (52) evaluated at energy (for response is anisotropic, as is well known [33].
undoped, unbiased graphene, ), which is position and To evaluate (72), we assume the long wavelength case and
time-independent perform a power series expansion in terms of the small param-
eter
(64)

is the electron velocity, and

(73)
(65)
The expansion will certainly be valid if ,
is the force on the electron. such that . Since the main contribution to
Following the standard perturbation procedure, assume the electron velocity is the Fermi velocity ,
where is the change in the distribution func- if the phase velocity of the electromagnetic wave is , then
tion due to the small, time-varying force . We assume that the this relation is always satisfied. If, e.g., , then we need
time and spatial variation of the perturbation follows that of .
the applied field. Upon keeping the first non-zero terms and making the re-
Using the Fourier transform (15), noting the time-dependence placements ,
and that
(74)
(66)
(67) (75)
(76)
and keeping only first-order terms, it is easy to show that
(77)
(68)
where
The requirement for to be small results from the requirement
for to be small. Ignoring spatial dispersion comes from ig- (78)
noring the term
(79)
(69)
(80)
which implies that the perturbation from equilibrium of the elec-
tron’s distribution function due to the applied field is indepen-
dent of position. (81)
756 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 56, NO. 3, MARCH 2008

(82) and therefore

(83) (96)

To evaluate the integrals, note that (97)

(98)
(84)
(99)

and where

(85)
(100)

where is quasi-momentum. The specific aspects of

(101)
graphene are incorporated via the energy function . For
graphene in the vicinity of the Dirac points (which supply the
main contribution to the integral) [10] It can be seen that (100) agrees with (58) for .

(86)
APPENDIX III
so that ALTERNATE METHOD FOR OBTAINING THE
DYADIC GREEN’S FUNCTIONS
(87) The reasons for using the Hertzian potential method for de-
riving the dyadic Green’s functions are discussed in the paper.
Upon defining a coordinate system centered at (at which In this appendix an alternative field method is provided, since
point ), noting that the radial coordinate is it is somewhat more straightforward, and may therefore be of
interest. The method is detailed in [16] for isotropic media, and
, with and applied to the case of a one-sided anisotropic impedance sur-
face in [34]. Here we merely briefly outline the method to the
(88) two-sided anisotropic impedance surface case.
The fields in each region are
and
(102)
(89)

the first integral is easily evaluated as (103)

(90) where, assuming a source in region 1

(104)
The factor of 2 comes from having two in-equivalent Dirac
points, and (105)

(91) (106)

The Green’s dyadics can be written as

In a similar manner

(92)

(107)
(93)

(94)

(95)
(108)
HANSON: DYADIC GREEN’S FUNCTIONS FOR AN ANISOTROPIC, NON-LOCAL MODEL OF BIASED GRAPHENE 757

[15] A. Ishimaru, Electromagnetic Wave Propagation, Radiation, and Scat-

tering. Englwood Cliffs, NJ: Prentice Hall, 1991.
[16] S. K. Chu, Electromagnetic Scattering. New York: Springer-Verlag,
1990.
[17] R. A. Jishi, M. S. Dresselhaus, and G. Dresselhaus, “Electron-phonon
(109) coupling and the electrical conductivity of fullerene nanotubules,”
Phys. Rev. B, vol. 48, pp. 11385–11389, 1993.
[18] G. W. Hanson, “On the applicability of the surface impedance integral
where in (107) the upper sign is for , and the lower sign equation for optical and near infrared copper dipole antennas,” IEEE
for . In these expressions Trans. Antennas Propag., vol. 54, pp. 3677–3685, Dec. 2006.
[19] V. Ryzhii, A. Satou, and T. Otsuji, “Plasma waves in two-dimensional
(110) electron-hole system in gated graphene heterostructures,” J. Appl.
Phys., vol. 101, p. 024509 (1-5), 2007.
[20] L. A. Falkovsky, “Unusual field and temperature dependence of the
(111) Hall effect in graphene,” Phys. Rev. B, vol. 75, p. 033409 (1-4), 2007.
[21] S. A. Mikhailov and K. Ziegler, “A new electromagnetic mode in
graphene,” Phys. Rev. Lett., vol. 99, p. 016803 (1-4), 2007.
[22] M. Dressel and G. Grüner, Electrodynamics of Solids. Cambridge,
where, since , ( in U.K.: Cambridge Univ. Press, 2002.
the notation in (12)). The boundary conditions are [23] V. P. Gusynin, S. G. Sharapov, and J. P. Carbotte, “Magneto-optical
conductivity in graphene,” J. Phy. Condens. Matter, vol. 19, p. 026222
(112) (1-25), 2007.
[24] V. P. Gusynin and S. G. Sharapov, “Transport of Dirac quasiparticles
in graphene: Hall and optical conductivities,” Phy. Rev. B., vol. 73, p.
245411 (1-18), 2006.
[25] V. P. Gusynin, S. G. Sharapov, and J. P. Carbotte, “Unusual microwave
(113) response of Dirac quasiparticles in graphene,” Phys. Rev. Lett., vol. 96,
p. 256802 (1-4), 2006.
at , leading to eight equations in the eight unknowns , [26] N. M. R. Peres, F. Guinea, and A. H. Castro Neto, “Electronic prop-
. erties of disordered two-dimensional carbon,” Phys. Rev. B, vol. 73, p.
125411 (1-23), 2006.
REFERENCES [27] N. M. R. Peres, A. H. Castro Neto, and F. Guinea, “Conductance quan-
tization in mesoscopic graphene,” Phys. Rev. B, vol. 73, p. 195411
[1] A. K. Geim and K. S. Novoselov, “The rise of graphene,” Nature Ma- (1-8), 2006.
terials, vol. 6, pp. 183–191, 2007. [28] K. Ziegler, “Minimal conductivity of graphene: Nonuniversal values
[2] R. Saito, G. Dresselhaus, and M. S. Dresselhaus, Physical Properties from the Kubo formula,” Phys. Rev. B, vol. 75, p. 233407 (1-4), 2007.
of Carbon Nanotubes. London, U.K.: Imperial College Press, 2003. [29] L. A. Falkovsky and A. A. Varlamov, “Space-time dispersion of
[3] K. S. Novoselov, A. K. Geim, S. V. Morozov, D. Jiang, Y. Zhang, S. graphene conductivity,” Eur. Phys. J. B, vol. 56, pp. 281–284, 2007.
V. Dubonos, I. V. Grigorieva, and A. A. Firsov, “Electric field effect [30] D. V. Khveshchenko, “Magnetic-field-induced insulating behavior in
in atomically thin carbon films,” Science, vol. 306, pp. 666–669, 2004. highly oriented pyrolitic graphite,” Phys. Rev. Letts., vol. 87, p. 206401
[4] K. S. Novoselov, A. K. Geim, S. V. Morozov, D. Jiang, M. I. Kat- (1-4), 2001.
snelson, I. V. Grigorieva, S. V. Dubonos, and A. A. Firsov, “Two-di- [31] E. V. Gorbar, V. P. Gusynin, V. A. Miransky, and I. A. Shovkovy,
mensional gas of massless Dirac fermions in graphene,” Nature, vol. “Magnetic field driven metal-insulator phase transition in planar sys-
438, pp. 197–200, 2005. tems,” Phys. Rev. B, vol. 66, p. 045108 (1-22), 2002.
[5] Y. Zhang, J. P. Small, W. V. Pontius, and P. Kim, “Fabrication [32] V. P. Gusynin, S. G. Sharapov, and J. P. Carbotte, “Sum rules for the
and electric-field-dependent transport measurements of mesoscopic optical and Hall conductivity in graphene,” Phys. Rev. B, vol. 75, p.
graphite devices,” Appl. Phys. Lett., vol. 86, p. 073104 (1-3), 2005. 165407 (1-12), 2007.
[6] C. Berger, Z. Song, X. Li, X. Wu, N. Brown, C. Naud, D. Mayou, T. [33] V. M. Agranovich, Spatial Dispersion in Crystal Optics and the Theory
Li, J. Hass, A. N. Marchenkov, E. H. Conrad, P. N. First, and W. A. of Excitons. New York: Wiley, 1966.
de Heer, “Electronic confinement and coherence in patterned epitaxial [34] A. Lakhtakia, “Green’s functions and Brewster condition for a halfs-
graphene,” Science, vol. 312, pp. 1191–1196, 2006. pace bounded by an anisotropic impedance plane,” Int. J. Infrared Mil-
[7] Y. Zhang, Y.-W. Tan, H. L. Stormer, and P. Kim, “Experimental ob- limeter Waves, vol. 13, pp. 161–170, 1992.
servation of the quantum Hall effect and Berry’s phase in graphene,”
Nature, vol. 438, pp. 201–204, 2005.
[8] C. Berger, Z. Song, T. Li, X. Li, A. Y. Ogbazghi, R. Feng, Z. Dai, A. George W. Hanson (S’85–M’91–SM’98) was born
N. Marchenkov, E. H. Conrad, P. N. First, and W. A. de Heer, “Ul- in Glen Ridge, NJ, in 1963. He received the B.S.E.E.
trathin epitaxial graphite: 2D electron gas properties and a route to- degree from Lehigh University, Bethlehem, PA, the
ward graphene-based nanoelectronics,” J. Phys. Chem., vol. 108, pp. M.S.E.E. degree from Southern Methodist Univer-
19912–19916, 2004. sity, Dallas, TX, and the Ph.D. degree from Michigan
[9] Y. Ouyang, Y. Yoon, J. K. Fodor, and J. Guo, “Comparison of perfor- State University, East Lansing, in 1986, 1988, and
mance limits for carbon nanoribbon and carbon nanotube transistors,” 1991, respectively.
Appl. Phys. Lett., vol. 89, p. 203107 (1-3), 2006. From 1986 to 1988, he was a Development Engi-
[10] P. R. Wallace, “The band theory of graphite,” Phys. Rev., vol. 71, pp. neer with General Dynamics, Fort Worth, TX, where
622–634, 1947. he worked on radar simulators. From 1988 to 1991,
[11] G. W. Hanson, “Dyadic green’s functions and guided surface waves he was a Research and Teaching Assistant in the De-
for a surface conductivity model of graphene,” J. Appl. Phys., to be partment of Electrical Engineering, Michigan State University. He is currently
published. an Associate Professor of electrical engineering and computer science at the
[12] G. Y. Slepyan, S. A. Maksimenko, A. Lakhtakia, O. Yevtushenko, and University of Wisconsin in Milwaukee. His research interests include nanoelec-
A. V. Gusakov, “Electrodynamics of carbon nanotubes: Dynamic con- tromagnetics, mathematical methods in electromagnetics, electromagnetic wave
ductivity, impedance boundary conditions, and surface wave propaga- phenomena in layered media, integrated transmission lines, waveguides, and an-
tion,” Phys. Rev. B, vol. 60, pp. 17136–17149, 1999. tennas, and leaky wave phenomena.
[13] J. S. Bagby and D. P. Nyquist, “Dyadic Green’s functions for integrated Dr. Hanson is a member of URSI Commission B, Sigma Xi, and Eta
electronic and optical circuits,” IEEE Trans. Microwave Theory Tech., Kappa Nu. He was an Associate Editor for the IEEE TRANSACTIONS ON
vol. 35, pp. 207–210, 1987. ANTENNAS AND PROPAGATION from 2002 to 2007. In 2006, he received the
[14] W. C. Chew, Waves and Fields in Inhomogeneous Media. Piscat- S. A. Schelkunoff Best Paper Award from the IEEE Antennas and Propagation
away, NJ: IEEE Press, 1999. Society.