734 IEEE TRANSACTIONS ON NANOTECHNOLOGY, Volume 18, 2019

Digital Metasurface Based on Graphene: An

Application to Beam Steering in Terahertz
Plasmonic Antennas
Seyed Ehsan Hosseininejad , Kasra Rouhi, Mohammad Neshat, Albert Cabellos-Aparicio, Sergi Abadal ,
and Eduard Alarcón

Abstract—Metasurfaces, the two-dimensional counterpart of of novel compact devices with unprecedented electromagnetic
metamaterials, have caught great attention thanks to their pow- control. Frequency selectivity, absorption, anomalous reflec-
erful capabilities on manipulation of electromagnetic waves. Re- tion/transmission, polarization conversion, and focusing are
cent times have seen the emergence of a variety of metasurfaces
exhibiting not only countless functionalities, but also a reconfig- among the many electromagnetic functionalities that can be
urable response. Additionally, digital or coding metasurfaces have achieved through the careful design of the metasurfaces [1],
revolutionized the field by describing the device as a matrix of dis- [2]. With such unprecedented control of the response to the im-
crete building block states, thus drawing clear parallelisms with pinging wave, metasurfaces have led to important breakthroughs
information theory and opening new ways to model, compose, and
in electromagnetic cloaking, imaging, as well as in the cre-
(re)program advanced metasurfaces. This paper joins the recon-
figurable and digital approaches, and presents a metasurface that ation of ultra-efficient, miniaturized antennas for sensors and
leverages the tunability of graphene to perform beam steering at implantable communication devices [3]–[8].
terahertz frequencies. A comprehensive design methodology is pre- A metasurface is generally defined as a planar array of peri-
sented encompassing technological, unit cell design, digital meta- odic or quasi-periodic subwavelength elements, whose structure
material synthesis, and programmability aspects. By setting up and coupling determine the electromagnetic function. As long
and dynamically adjusting a phase gradient along the metasurface
plane, the resulting device achieves beam steering at all practical di- as the elements remain subwavelength in size, the working prin-
rections. The proposed design is studied through analytical models ciple of metasurfaces can be applied from microwaves to the
and validated numerically, showing beam widths and steering er- visible range [2]. Between these two extremes lies the terahertz
rors well below 10° and 5% in most cases. Finally, design guidelines (THz) band, for which designs have been reported to manipu-
are extracted through a scalability analysis involving the metasur- late the phase, amplitude or polarization of the waves reflected
face size and number of unit cell states.
or transmitted by the metasurface [9]–[12].
Index Terms—Beam steering, digital metasurfaces, graphene, The main issue of conventional metasurfaces is the lack
plasmonics, terahertz frequencies. of adaptivity and reconfigurability as, in most designs, the
electromagnetic function and its scope are fixed once the
I. INTRODUCTION unit cell is designed. In order to avoid re-designing and re-
fabricating metasurfaces each time a change in frequency or
ETASURFACES, defined as artificial structures with
M subwavelength thickness, have enabled the realization
functionality is required, one can introduce tunable or switch-
able elements in the design of unit cells [13]. The resulting
reconfigurable metasurfaces can be globally or locally tun-
Manuscript received February 27, 2019; revised May 4, 2019; accepted June able depending on the specific design, and better yet through
9, 2019. Date of publication June 25, 2019; date of current version July 29, appropriate control means, they can become programmable
2019. This work was supported in part by the Iran’s National Elites Foundation
(INEF), in part by the Spanish Ministry of Economía y Competitividad under
[14], [15].
Grant PCIN-2015-012, and in part by ICREA under the ICREA Academia pro- Coding metamaterials, sometimes also referred to as digital
gramme. The review of this paper was arranged by Associate Editor Dr. Salvatore metamaterials, are a particular type of programmable meta-
Pontarelli. (Corresponding author: Seyed Ehsan Hosseininejad.)
S. E. Hosseininejad is with the Department of Electrical and Computer Engi-
materials that discretize the number of states of a unit cell
neering, Babol Noshirvani University of Technology, Babol 47148-71167, Iran, [16]–[19]. Each state is represented by a number of bits that are
and also with the School of Electrical and Computer Engineering, University of used to make the actual metasurface. A desired global response
Tehran, Tehran 14174-66191, Iran (e-mail: [email protected]).
K. Rouhi is with the School of Electrical Engineering, Iran University of
is achieved through a medium profile that is not necessarily
Science and Technology, Tehran 13114-16846, Iran (e-mail: kasrarouhi@elec. periodical. Such structure, when built using locally switchable
iust.ac.ir). elements, can be elegantly described as a bit or state matrix
M. Neshat is with the School of Electrical and Computer Engineering, Uni-
versity of Tehran, Tehran 14174-66191, Iran (e-mail: [email protected]).
and digitally controlled through reconfigurable devices such
A. Cabellos-Aparicio, S. Abadal, and E. Alarcón are with the NaNoNet- as Field-Programmable Gate Arrays (FPGAs) [19]. Several
working Center in Catalonia (N3Cat), Universitat Politècnica de Catalunya examples implementing polarization control, focusing control,
08034 Barcelona, Spain (e-mail: [email protected]; [email protected];
[email protected]).
or beam manipulation in the GHz range can be found in the
Digital Object Identifier 10.1109/TNANO.2019.2923727 literature [19]–[21].

1536-125X © 2019 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

Authorized licensed use limited to: Indian Institute Of Technology (IIT) Mandi. Downloaded on April 22,2025 at 13:35:34 UTC from IEEE Xplore. Restrictions apply.
HOSSEININEJAD et al.: DIGITAL METASURFACE BASED ON GRAPHENE: AN APPLICATION TO BEAM STEERING 735

Fig. 1. Sketch of the programmable graphene-based digital metasurface for THz beam steering and its design flow: from the unit cell to the global controller.

Graphene, with its outstanding optoelectrical properties, has the unit cell to the metasurface controller (Figure 1). The pro-
been recently introduced as a key enabler of a myriad of posed methodology is then applied to develop a metasurface for
applications in countless domains [22]–[25]. It is well known fine-grained beam steering at terahertz frequencies. The meta-
that graphene naturally supports Surface Plasmon Polaritons surface acts as a reflectarray that forms dynamically reconfig-
(SPP) in the terahertz band, and therefore, becomes an excellent urable phase gradients in the X and Y directions, through which
option for the implementation of terahertz sources [26] and an- the reflected beam can be driven to any desired direction. The
tennas [27], among others. The plasmonic nature of graphene at unit cells of the reflectarray are based on a graphene-insulator-
terahertz frequencies leads to miniaturized devices [28], whereas graphene stack that achieves wide phase tuning via electrostatic
its inherent tunability has been leveraged in frequency-agile or biasing of the graphene patches. With two bits per unit cell and
reconfigurable concepts [29]–[31]. Some of such designs are the appropriate controller, the proposed metasurface achieves a
array-based, and similar to programmable metasurfaces, they very wide steering range with low beam width.
achieve reconfigurability by switching the state of its elements, The proposed metasurface is particularly suitable for wireless
i.e. tuning them in or out [32]–[34]. communication applications. In this context, the use of the lower
The above-mentioned properties turn graphene into a unique part of the THz spectrum (our design operates at f = 2 THz)
material for the implementation of terahertz reconfigurable becomes extremely attractive due to the abundance of band-
metasurfaces. First explorations in this regard considered width that allows to satisfy the extreme data rate demands of 5G
graphene reflectarrays and studied their amplitude-phase re- networks and beyond [45]. Communication in the THz band,
sponses when tuning the chemical potential of the graphene however, requires overcoming high path losses mainly through
[35]. By means of local tuning of the graphene elements through directive antennas with very narrow beams and through the use
electrostatic biasing, the scattering profile of the reflectarray can of smart programmable reflectors [34], [46]–[49]. It is thus fun-
be modified to achieve beam steering [36], focusing [37], dif- damental that these devices be capable of steering the THz beam
fusive scattering [38], cloaking [39] or wave vorticity control with high precision to track the users and avoid interrupting com-
[40]. These functionalities have been achieved in the microwave munication.
regime thanks to the use of phase change materials (PCMs) [41], In summary, the main contributions of this paper are:
semiconductor diodes [42], or microelectromechanical systems r The development of a comprehensive methodology for the
(MEMS) [43]. However, as we approach to terahertz frequen- design of graphene-based programmable terahertz meta-
cies, diodes and MEMS become lossy and too large to be in- surfaces for beam steering, from the unit cell up to the
tegrated within individual unit cells. With graphene, the design global controller.
can be greatly simplified and the device can be reconfigured r The use of the proposed methodology to design and eval-
much faster. uate a 2-bit coding metasurface for beam steering. Wide
Although the natural switchability of graphene in the terahertz steering range with a sharp reflected beam and low over-
band matches perfectly with the coding metamaterial paradigm, heads are demonstrated. The chosen frequency of operation
which has been explored in [44] for the first time, the lack of is f = 2 THz, within the range expected for THz wireless
literature to present a clear methodology for the unit cell design communication applications.
based on graphene and the coding of the metasurface seems ev- r A scalability analysis illustrating the relation between the
ident. To bridge this gap, this paper presents a comprehensive different design parameters and performance metrics, and
methodology for the design of programmable metasurfaces from uncovering several co-design opportunities.

Authorized licensed use limited to: Indian Institute Of Technology (IIT) Mandi. Downloaded on April 22,2025 at 13:35:34 UTC from IEEE Xplore. Restrictions apply.
736 IEEE TRANSACTIONS ON NANOTECHNOLOGY, Volume 18, 2019

The rest of this paper is organized as shown schematically sheets is on the order of 10,000 cm2 V−1 s−1 , which is achiev-
in Fig. 1. Section II presents a design space exploration of able with current fabrication and encapsulation techniques [52].
graphene-based unit cells from the perspectives of size, chemical Thus, the amplitude requirement can be met as well.
potential, and number of states. Section III formulates a design
flow for beam steering coding metasurfaces, which is then tested
by showing the effective steering of the antenna beam in several B. Unit Cell Design
directions. Section IV discusses and evaluates the implementa- Figure 2 shows a schematic representation of the proposed
tion of the scheme that actually controls and (re)programs of unit cells together with their approximate equivalent circuit
the metasurface. Finally, Section V outlines the main scalabil- models. We numerically simulate different designs in CST Mi-
ity trends and co-design opportunities of the proposed design. crowave Studio [53] to obtain the amplitude and phase re-
Section VI concludes the paper. sponses. Then, through the equivalent circuit models, we verify
the results of the numerical approach, and reason about the be-
II. GRAPHENE-BASED UNIT CELL havior of different unit cells. In all cases, we assume a lateral
size of du = 20 µm. This value is around λ0 /8 for the targeted
The design of any metasurface starts with its most basic build-
frequency of operation (2 THz), enough to provide the subwave-
ing block, namely, the unit cell. For beam manipulation, we need
length behavior required in the metasurface. The relaxation time
to provide a unit cell with the ability of controlling the phase
of graphene is assumed to be τ = 0.6 ps, which is compatible
response over a wide range of values [12]. Moreover, since the
with the carrier mobility requirements mentioned above.
proposed device acts as a reflectarray, the unit cell needs to yield
The first unit cell (Fig. 2(a)) consists of a fully covered layer
a high reflection amplitude at all times. To enable dynamic re-
of graphene on top of a silicon substrate with refractive index
configurability, it is necessary that both objectives can be met
nSi = 3.45 and thickness dSi = 10 µm along with a metallic
without physically changing any geometry. In this paper, re-
ground plane on the backside. In such a unit cell, graphene is a
configurability is achieved at THz frequencies by means of the
lossy medium that can be modeled through an RL circuit. Fig. 3
electrostatic tuning of graphene.
plots the amplitude and phase responses of reflection coefficient
for such unit cell versus frequency and the chemical potential. It
A. Graphene Modeling
is observed that the amplitude response is within an acceptable
We analyze different unit cells that leverage the tunability of range, whereas the phase range is not wide enough –around 135°
graphene to achieve the desired phase variation with reasonable at 2 THz assuming a maximum chemical potential range of 1 eV.
losses and without the need of changing any geometry. To drive A very good agreement is obtained between the numerical results
the design and to perform an accurate evaluation of different pro- and the equivalent circuit model.
posals, we model graphene as an infinitesimally thin sheet with The second unit cell (Fig. 2(b)) consists of a graphene patch
surface impedance Z = 1/σ(ω), where σ(ω) is the frequency- that partially covers the unit cell. The substrate and ground plane
dependent conductivity of graphene. The complex conductivity remain unchanged. In this case, a capacitance is introduced to
is given by model the coupling effects generated between the edges of ad-
   jacent graphene patches. As shown in Fig. 4, the size of the
2e2 kB T μc i
σ (ω) = ln 2 cosh , (1) graphene patch provides an extra degree of freedom to deliver
π  2kB T ω + iτ −1
the target amplitude and phase responses. Exploring the {chem-
where e,  and kB are constants corresponding to the charge ical potential, patch size} design space, we observe that there
of an electron, the reduced Planck constant and the Boltzmann is a tradeoff between the amplitude and phase variation. For a
constant, respectively [50]. Variables T , τ and μc correspond patch size of 6 µm, the phase range covers almost 360° with less
to the temperature, the relaxation time and the chemical po- than 0.8 eV chemical potential variation. However, the ampli-
tential of the graphene layer. Note that this expression neglects tude response also has a very large variation, which discourages
the edge effects of the graphene and considers that the Drude- the choice of such design point. A patch size of 8 µm or larger
like intraband contribution dominates, which are experimentally provides a better amplitude response with a reasonable phase
validated assumptions at the sizes and frequencies considered in change variation.
this work [51]. The third unit cell (Fig. 2(c)) is composed of a graphene-
On the one hand, the phase control in the graphene metasur- insulator-graphene stack placed over the substrate along with
face is achieved via changes in its complex conductivity when a ground plane on the backside. High-density polyethylene
biased – the effect that can be modeled through the chemical (HDPE) is chosen as the insulator due to its particularly low
potential value μc . The chemical potential can be controlled losses in the terahertz band [54]. The refractive index of the in-
through electrostatic biasing, and therefore, we can meet the sulator is nHDP E = 1.54 and its thickness is dHDP E = 4 µm.
phase change requirement. On the other hand, the amplitude re- The equivalent circuit model of this structure consists of two
sponse depends on the losses within graphene, which are mostly parallel RLC cells representing each of the graphene sheets. As
influenced by the relaxation time value τ . Note that the relax- shown in Fig. 5, this unit cell achieves a much wider phase vari-
ation time is proportional to the carrier mobility, which depends ation, and by addressing each graphene patch independently, it
on the quality of the material. For the purpose of this work, losses provides an extra degree of freedom to choose the states of the
will be affordable as long as the carrier mobility of the graphene metasurface.

Authorized licensed use limited to: Indian Institute Of Technology (IIT) Mandi. Downloaded on April 22,2025 at 13:35:34 UTC from IEEE Xplore. Restrictions apply.
HOSSEININEJAD et al.: DIGITAL METASURFACE BASED ON GRAPHENE: AN APPLICATION TO BEAM STEERING 737

Fig. 2. A schematic representation of the graphene unit cells with their respective equivalent circuit models.

Fig. 3. a) Amplitude and b) phase responses of reflection coefficient for the 1L unit cell. Effect of chemical potential variation in c) amplitude and d) phase for
various frequencies. Unless noted, EF = 0.7 eV, τ = 0.6 ps, and f = 2 THz.

Fig. 4. a) Amplitude and b) phase responses of reflection coefficient for the 1G unit cell. Unless noted, EF = 0.2 eV, τ = 0.6 ps, dG = 16 µm, and f = 2 THz.
Effect of chemical potential and patch size variation in c) amplitude and d) phase for a constant frequency f = 2 THz.

Fig. 5. a) Amplitude and b) phase responses of reflection coefficient for the 2G unit cell. In the frequency response figures, EF = 0.2 eV, τ = 0.6 ps, dG = 16 µm.
Effect of top and bottom layers chemical potential in c) amplitude and d) phase for a constant frequency (the four chosen coding states used in the simulation are
marked by different signs). In these figures, τ = 0.6 ps, dG = 12 µm and f = 2 THz.

Authorized licensed use limited to: Indian Institute Of Technology (IIT) Mandi. Downloaded on April 22,2025 at 13:35:34 UTC from IEEE Xplore. Restrictions apply.
738 IEEE TRANSACTIONS ON NANOTECHNOLOGY, Volume 18, 2019

Fig. 6. a) Amplitude and b) phase of reflection coefficient for 1-bit digital Fig. 7. a) Amplitude and b) phase of reflection coefficient for 2-bit digital
metasurface using a 1G unit cell. metasurface using a 2G unit cell.

According to the formulation presented in [55], the initial a constant reflection coefficient of around 0.7 and deliver the
values for RLC model parameters are estimated. For the dual targeted 180° phase shift.
patch unit cell, due to the coupling effect between two graphene Two-bit coding leads to a phase shift resolution of 90° and
layers, small changes can occur in the RLC parameters which are would improve the beam steering accuracy substantially. The
optimized by genetic algoritm. The parameters for three unit cell 1G unit cell, however, barely meets the amplitude and phase
are calculated respectively as (R, L) = (20.196 Ω, 0.012 pH), shift requirements with a 90° resolution. With dG = 8 µm, there
(R, L, C) = (192.855 Ω, 63.49 pH, 1.368 fF), and (R1 , L1 , C1 ; is no combination of chemical potentials capable of avoiding
R2 , L2 , C2 ) = (20.32 Ω, 0.013 pH, 0.107 fF; 35.97 Ω, 26.26 the region of low amplitude around 0.9 eV. For larger patch
pH, 3.38 fF). sizes, the phase response is not wide enough to accommodate
Regarding fabrication feasibility of the proposed low-profile two bits. For three or more bits, this unit cell would not be
structure (10 µm substrate), there are advanced Silicon substrate suitable for beam steering, at least for the relaxation time values
thinning techniques that can be used to achieve an ultra-thinning and geometry considered in this work.
down to 4 µm without damage occurred due to thinning pro- Alternatively, the 2G unit cell offers much more freedom
cesses [56]. and is capable of accommodating two or more bits. Address-
ing the 2G unit cell with two bits, one can find suitable
C. Unit Cell Discrete States design points with dG = 12 µm. Using the design space explo-
ration from Fig. 5, good performance is obtained for the follow-
The results above show the response of the metasurface for
ing up-layer and down-layer chemical potentials, respectively:
a continuous range of chemical potentials. However, in order to
μc,1 = {0.6, 1.3, 0.1, 0.4} eV and μc,2 = {0, 0.6, 0.1, 0.1} eV
design a bit-programmable metasurface, we need to discretize
corresponding to the bit combinations B = {00, 01, 10, 11}. It
the potentials to obtain a finite set of addressable states.
is observed in Fig. 7 that these states consistently achieve a re-
The first decision concerns the number of target states, and
flection coefficient around 0.7 and a phase difference of 90°
thus, the number of bits required to address a unit cell. Here, the
covering the whole phase space. In Section IV, we discuss how
number of states will determine the phase difference between
to electronically achieve these states.
consecutive unit cell states. For the application at hand, there is a
relation between the phase difference and the steering resolution,
III. CODING METASURFACE TERAHERTZ ANTENNA
i.e., the angle difference between consecutive achievable beam
directions (see Section V for details). Therefore, the number of To illustrate the design approach of a terahertz metasurface
bits must be chosen carefully. for beam steering application, a metasurface including M × N
Let us now exemplify this process by deriving the states re- controllable unit cells is considered. Our design allows to intro-
quired for the metasurface to work at 2 THz. We start by ad- duce a phase gradient by smartly changing the chemical potential
dressing the 1G unit cell with a single bit. From the design μc of the graphene sheets from one unit cell to another. In this
space exploration shown in Fig. 4, we choose design points that case, we need to use the generalized reflection law to evaluate
have high amplitude and a phase difference of approximately the response of the metasurface [57].
180°. A good choice is dG = 8 µm with μc = {0.6, 1.28} eV In the following, we first derive the conditions required
corresponding to a bit combination of B = {0, 1}. The result- to achieve beam steering to a desired direction {θr , φr } in
ing amplitude and phase responses, illustrated in Fig. 6, provide Section III-A. Then, we define the design and configuration flow

Authorized licensed use limited to: Indian Institute Of Technology (IIT) Mandi. Downloaded on April 22,2025 at 13:35:34 UTC from IEEE Xplore. Restrictions apply.
HOSSEININEJAD et al.: DIGITAL METASURFACE BASED ON GRAPHENE: AN APPLICATION TO BEAM STEERING 739

By mathematically simplifying the above equations as shown

in the Appendix, the reflected elevation angle θr and azimuth
angle φr are obtained as
θr = arcsin

dΦ 2 dΦ 2
(ki sin θi cos φi + dx ) + (ki sin θi sin φi + dy )
kr

ki sin θi sin φi + dΦ
dy
φr = arctan (7)
ki sin θi cos φi + dΦ
dx

When the metasurface is illuminated by a normally incident

Fig. 8. Coordinate system used in the formulations of generalized reflection wave (θi = φi = 0), and assuming air as the medium of the
law. incident and reflected wave, we can simplify the formulas as

(∇x Φ)2 + (∇y Φ)2
to achieve such the desired direction by our proposed design in θr = arcsin
k0
Section III-B. Finally, we evaluate the performance of the pro-
posed metasurface in Section III-C. ∇y Φ
φr = arctan , (8)
∇x Φ
A. Generalized Reflection Law Formulation which relates the phase gradient in the metasurface to the direc-
Consider a reflective metasurface under illumination of an tion of the reflected wave.
incident plane wave at elevation angle θi and azimuth angle φi
according to the coordinate system shown in Fig. 8. The incident B. Design Flow
wave vector ki can be written as Using (8), the reflected angles for all phase profiles of the
ki = kix x̂ + kiy ŷ + kiz ẑ (2) metasurface can be calculated. The design of the metasurface
can be then thought as an inversion process. We need to estimate
where {kix , kiy , kiz } are the wave vector coordinates, given by the necessary phase profile to achieve the desired elevation and
kix = ki sin θi cos φi = k0 ni sin θi cos φi azimuth angles of the reflected wave. To this end, again assuming
a normal incident plane wave and air as the medium, we can
kiy = ki sin θi sin φi = k0 ni sin θi sin φi rearrange (6) as

kiz = ki cos θi = k0 ni cos θi (3) λ0 dΦ

dx =
2π cos φr sin θr
The same formulation can be applied to the reflected wave vector
kr given the elevation angle θr and azimuth angle φr of the λ0 dΦ
dy = (9)
reflected wave. 2π sin φr sin θr
Assuming that the metasurface imposes the phase profile
where dΦ describes the phase difference between adjacent unit
Φ(x, y), we assign it the virtual wave vector kΦ so that
cell states. Starting from here, the design methodology requires
dΦ dΦ the knowledge of the unit cell dimensions and the number of
kΦ = kΦx x̂ + kΦy ŷ = x̂ + ŷ = ∇Φx x̂ + ∇Φy ŷ (4)
dx dy states. The design flow is as follows,
1) Obtaining the cluster size (in μm): Assume that the cod-
where ∇x Φ = dΦ dx and ∇y Φ = dy are the phase gradients along
dΦ
ing metasurface can choose among 2n states for each unit cell
the x and y directions, respectively. (referred to as n-bit coding). In this case, the granularity of the
Applying the boundary conditions of the tangential compo- gradient is dΦ = π/2n−1 . Therefore, the lateral dimensions of
nents of the electromagnetic fields, the momentum conservation the required cluster of unit cells (dcx and dcy ), as shown in Fig. 1,
law for wave vectors can be expressed as are obtained by substituting dΦ in Eq. (9),
kix + kΦx = krx λ0
dcx =
kiy + kΦy = kry (5) 2n cos φr sin θr

and substituting (3) and (4) in (5) yields λ0

dcy = . (10)
2n sin φr sin θr
dΦ
ki sin θi cos φi + = kr sin θr cos φr The value of φr determines the difference between dcx and dcy ,
dx
which impacts on the shape of the reflected beam. The larger
dΦ difference results in the more elliptical shape of the reflected
ki sin θi sin φi + = kr sin θr sin φr (6)
dy beam.

Authorized licensed use limited to: Indian Institute Of Technology (IIT) Mandi. Downloaded on April 22,2025 at 13:35:34 UTC from IEEE Xplore. Restrictions apply.
740 IEEE TRANSACTIONS ON NANOTECHNOLOGY, Volume 18, 2019

where θ and φ are the elevation and azimuth angles of an arbi-

trary direction, respectively, and fE (θ, φ) and fA (θ, φ) are the
element factor (pattern function of unit cell) and array factor
(pattern function of unit cell arrangement), respectively. Here,
the unit cells are assumed to be isotropic, and therefore the scat-
tering pattern depends only on the array factor


M 
N
F (θ, φ) = exp{−j[Φ(p, q)
p=1 q=1

+ kdu (p − 1/2) sin θ cos φ

Fig. 9. Absolute value of cx as a function of the desired direction of reflection.
+ kdu (q − 1/2) sin θ sin φ]}. (14)

2) Obtaining the cluster size (in number of unit cells): The We first evaluate the metasurface when configured to steer
nature of a metasurface, consisting an array of unit cells, dictates the beam at {φr = 30◦ , θr = 45◦ } with M = N = 100. Fol-
the discretization of space. Therefore, the values of dcx and dcy lowing the design flow from Section III-B and assuming 20-µm
need to be approximated to an integer number of unit cells. For unit cells and 2-bit coding, we obtain that dcx = 61.24 µm and
this purpose, we consider that the number of unit cells in the x dcy = 106.07 µm, which leads to a cluster of 3 × 5. The super
and y directions, designated by cx ∈ Z and cy ∈ Z, respectively, unit cell thus extends for 12 × 20 unit cells. Figure 10 shows the
are rounded as far field pattern of the resulting metasurface, which confirms
    that there is a good agreement between the numerical and ana-
dcx dcy
cx = , cy = . (11) lytical solutions. The reasons for the small differences between
du du
the numerical and theoretical results in terms of sidelobe level
Figure 9 shows the absolute value of cx as a function of the could be the marginal unit cells in the clusters, super unit cells,
target direction for a normally incident plane wave. It is ob- and the whole structure. We obtain the amplitude and phase of
served that cx becomes arbitrarily large as the reflected angle the reflection coefficient of the proposed unit cells while taking
approaches θr = 0 (white areas of the figure). This is consis- into consideration mutual coupling between the same adjacent
tent with the fact that such direction implies specular reflection, unit cells by assuming a periodic boundary condition. However,
which can only be realized with a homogeneous surface, i.e. zero in the real metasurface structure for beam steering, there are
gradient. For directions approaching φ = π/2, 3π/2, cx also be- some marginal unit cells which break the periodic layout. Con-
comes large because the gradient is only needed in the y axis. sequently, the periodicity condition is broken. In the simulation
On the contrary, cx approaches zero in the co-planar directions, of the full structure, the correct coupling of marginal unit cells
where an infinite gradient would be required. The black area in is considered while in the theoretical analysis of entire metasur-
Fig. 9 denotes cx < 1, which is unfeasible. face structure, it is ignored perforce. It is seen that the reflected
3) Obtaining the size of the super unit cell: To calculate the beam indeed points to the target direction. The steering error,
size of the super unit cell, designated by sx ∈ Z and sy ∈ Z, in evaluated as the difference between the target and achieved an-
number of unit cells, one needs to apply gles, is 2.5% and 3.3% in φ and θ, respectively. The 3-dB width
of the beam is approximately 5° in both cases.
sx = 2n cx , sy = 2n cy . (12) To further verify the validity of the approach, we reconfigure
the metasurface to operate at three different steering directions.
C. Evaluation Figure 11 shows how the proposed metasurface design is capa-
In this section, we evaluate the proposed metasurface both ble of achieving the desired responses and Table I summarizes
numerically and analytically. We numerically model the four the characteristics and performance of the resulting configura-
states of the 2G unit cell in CST [53], and apply the formulation tions. A wide range of reflected angles is achieved with clusters
developed above to assign the states to different unit cells of of 2–8 unit cells, achieving in all cases beam widths below 11°
an M × N metasurface. Then, we obtain the response of the (minimum 4°) with steering error below 7% (minimum 0.11%).
metasurface in the form of the far-field pattern produced by a Note that the error and beam width generally increase when
normally incident plane wave. We assume that the beam covers approaching forbidden areas in the design space, where the gra-
the whole metasurface. dient tends to zero or infinity. Also, the reflected beam for the
In the analytical approach that is used to verify the numerical cases {30°, 45°} and {340°, 60°} (especially the latter) tend to
results, the reflection phase Φ(p, q) of each unit cell of size du be elliptical due to the larger difference between dcx and dcy ,
is assumed to be exactly either 0, π/2, π, or 3π/2. Assuming as hinted in Section III-B. In addition, to achieve a continuous
a designed phase distribution assigned to the unit cells, we can beam scanning ability of coding metasurface with minimum an-
express the far-field scattering pattern F (θ, φ) as gle variation, the convolution approach can be leveraged to steer
the far-field pattern to a predetermined direction [58]. Regarding
F (θ, φ) = fE (θ, φ) × fA (θ, φ) (13) the Fourier relation between the field distribution on the coding

Authorized licensed use limited to: Indian Institute Of Technology (IIT) Mandi. Downloaded on April 22,2025 at 13:35:34 UTC from IEEE Xplore. Restrictions apply.
HOSSEININEJAD et al.: DIGITAL METASURFACE BASED ON GRAPHENE: AN APPLICATION TO BEAM STEERING 741

Fig. 10. Radiation pattern of the metasurface structure at 2 THz with the main reflected beam pointed at {φr = 30◦ , θr = 45◦ } by a) numerical approach and
b) analytical approach. The incident wave is normal to the metasurface.

Fig. 11. Radiation pattern of the metasurface structure at 2 THz with the main reflected beam pointed at the directions (numerical approach): a) {φr = 130◦ , θr =
30◦ }, b) {φr = 230◦ , θr = 20◦ }, and c) {φr = 340◦ , θr = 60◦ }. The incident wave is normal to the metasurface.

TABLE I
DESIGN AND PERFORMANCE RESULTS

metasurface and the resultant scattering pattern in the far-field, appropriate voltage (chemical potential). Finally, we review
one can shift the reflected pencil beam by adding the two cal- source considerations in Section IV-C.
culated phase gradient coding patterns so that the total phase
gradient deflection angle is equal to the desired angle.
A. Controller Design
To achieve programmability, it is necessary to attach the meta-
IV. PROGRAMMABILITY AND IMPLEMENTATION ISSUES surface to a digital device capable of translating the beam steer-
The final steps in the design of our beam steering device relate ing requirements into the global metasurface state. Algorithm 1
to the elements that control and excite the metasurface. More pre- shows a pseudocode that exemplifies this function. The process
cisely, we need to conceive a setup that takes the target reflected starts by calculating the size of the unit cell clusters cx × cy as
angle as input and modifies the metasurface accordingly. To a function of the bit number per cell n, and the dimension of the
this end, in Section IV-A we propose a controller that automati- unit cell du . Then, the gradient can be built easily by assigning
cally converts the target reflected angle into a bit matrix defining consecutive states to adjacent clusters of unit cells. As already
the states of each unit cell. Then, in Section IV-B we discuss mentioned in Section III-B, (10) and (11) can produce negative
the biasing scheme required to address each unit cell with the values, in which case the order of states is reversed.

Authorized licensed use limited to: Indian Institute Of Technology (IIT) Mandi. Downloaded on April 22,2025 at 13:35:34 UTC from IEEE Xplore. Restrictions apply.
742 IEEE TRANSACTIONS ON NANOTECHNOLOGY, Volume 18, 2019

Algorithm 1: Algorithm for Clustered Gradient Formation

Inputs: du, phiR, thetaR, n, f
/* CALCULATION OF THE CLUSTER SIZES */
lambda = 3e8/f;
dcx = lambda/(2∧ n*cos(phiR)*sin(thetaR));
dcy = lambda/(2∧ n*sin(phiR)*sin(thetaR));
cx = round(dcx/du); // MAY BE NEGATIVE
cy = round(dcy/du); // MAY BE NEGATIVE

/* CALCULATION OF THE STATE MATRIX */

for (i = 1; i < M; i++) {
for (j = 1; j < N; j++) {
B (i, j) = (round(i/cx) + round(j/cy)) mod(2∧ n);
}
}
Fig. 12. Sample implementation of the actuator for the metasurface based on
the 2G unit cell and 2-bit coding.
a point where the potential reaches the maximum desired point.
The resulting chemical potential μc relates to the change of volt-
Algorithm 1 assumes that all unit cells are addressed by a cen- age Δvg as
tralized device, probably an FPGA. However, since the meta- eμ2c t
surface implements a discretized gradient, it would be rela- Δvg = , (15)
π vF2 ε0 εr
2
tively straightforward to come up with an algorithm that can
calculate the required state in a distributed way, only relying on where e is the elementary charge,  is the reduced Planck con-
the state of the immediate neighbour. Such simplified scheme stant, vF ≈ 106 m/s is the Fermi velocity, ε0 is the vacuum
would be suitable for the rising Software-Defined Metamate- permittivity, whereas εr and t are the permittivity and thickness
rial (SDM) paradigm [59], [60], which aims to provide pro- of the material below graphene [62].
grammable metamaterials that can be reconfigured via an inte- Figure 13 illustrates the voltage ranges required to achieve
grated network of controllers that drive unit cells individually. a certain target chemical potential range. As directly implied
In that case, an external entity called gateway would receive the by (15), the voltage requirements increase quadratically with
command of changing the direction of the beam. The gateway the target chemical potential range. To limit the requirements,
would compute cx and cy , then rely them to the first controller one can either minimize the space between the gate and the
together with n. The first controller would be initialized and pass graphene layer or use materials with high dielectric constant.
its state along with cx , cy , and n to their neighbours, which would However, the former is determined by technological constraints,
repeat the process until the whole metasurface is programmed. and the latter needs to take into consideration the cost and other
characteristics of the material.
B. Actuator Design The chemical potential range required by our metasurface
can be obtained easily once the unit cell states are defined. In
The actuator is a circuit that translates the state matrix [B]
the present design, Δμc = 1.3 eV. Assuming a Al2 O3 layer
provided by the controller into the matrix of appropriate volt-
(εr = 9.1) with thickness t = 10 nm, achievable with current
ages [VG ] that, in turn, leads to the required chemical potentials
technologies [32], the resulting voltage range is 24.9 V.
in each graphene patch of the metasurface. As shown in Fig. 12,
a set of voltage level shifters and a matrix of multiplexers would
be enough for this purpose. Note that several independent sets C. Towards an Experimental Setup
of multiplexers (two in our case) may be required to drive the Figure 14 illustrates a possible measurement setup for the
graphene patches of individual unit cells. It is also worth noting experimental validation of the metasurface. The testbed would
that only five distinct voltages are needed in our case, because mainly consist of a fiber-coupled time domain spectroscopy
several states share the same target chemical potentials accord- (THz-TDS) system with a fixed source (generally harder to cal-
ing to the calculations made in Section II-C. ibrate) and a movable receiver placed on a rotatory platform.
The actual voltages required at the output of the level shifters The source is based on a photoconductive antenna coupled to
mainly depend on the graphene biasing structure and the required a focusing or collimating lens that minimizes spreading losses.
chemical potential [32], [61]. The configuration assumed in this Additional optics such as parabolic reflectors can be incorpo-
paper is similar to that used in [32], which couples graphene rated to meet the receiver sensitivity as well as the source–
capacitively with a back gate through a thin Al2 O3 layer. Es- metasurface–receiver distance requirements. It is worth noting
sentially, this scheme shifts the operation of graphene between that such a scheme has been used successfully in other works
the Dirac point, where the chemical potential is minimum, and proving anomalous reflection in the THz band [11], [17]. A very

Authorized licensed use limited to: Indian Institute Of Technology (IIT) Mandi. Downloaded on April 22,2025 at 13:35:34 UTC from IEEE Xplore. Restrictions apply.
HOSSEININEJAD et al.: DIGITAL METASURFACE BASED ON GRAPHENE: AN APPLICATION TO BEAM STEERING 743

Fig. 13. Voltage range required at the level shifting stage to achieve a given target chemical potential range.

with fixed size, but increasing number of bits. The 3-bit in-
stance greatly reduces the side lobes and has its maximum
at θr = 5.26°, whereas the beam moves away from the
desired direction for the 2-bit and 1-bit cases (4.92° and
4.22°, respectively). Note, however, that the gain in accu-
racy comes at the cost of a substantially higher complexity
at the controller and the actuator.
r In arrays, adding more antennas allows to reduce the beam
width. The same principle should apply in our design, as
Fig. 14. Sketch of a potential measurement setup for the proposed device. exemplified in Section III-C with the array factor formu-
lation. We verified such hypothesis by fixing the gradi-
ent and doubling the number of supercells once and twice
similar system has been built in [63] for the measurement of the (from M = N = 40 to M = N = 160). The resulting far
reflection coefficient of surfaces in the terahertz band entirely field patterns, shown in Fig. 15(b), clearly demonstrate that
with commercial solutions. the beam is sharpened without significantly changing the
direction of maximum energy. In fact, the beam width is
reduced by a factor proportional to the increase in meta-
V. DISCUSSION surface size, i.e. from 50° and 5° to 12° and 1.3° in the θ
In this section, we qualitatively discuss several cross-cutting and φ angles, respectively.
issues related to the design of the metasurface. Co-design opportunities: Understanding the design flow
Scalability analysis: To accommodate the proposed design from the point of view of the unit cell, the metasurface, or the
flow to different beam steering specifications, it is crucial to device as a whole helps to identify possible co-design oppor-
understand which are the key design parameters and what per- tunities. For instance, the unit cell configuration and the cod-
formance metrics do they affect. Here, we highlight some: ing determine the complexity in terms of number of required
r The size of the unit cell presents an interesting tradeoff. voltage levels, as well as the quantity and size of the multiplex-
While it may be difficult to achieve a wide phase range if ers. For an n-bit coding with g independently biased graphene
the unit cell is too small with respect to the wavelength (see patches, we may require up to g · 2n levels and g multiplex-
Figs. 3 and 4), reducing its dimensions leads to a raise in ers with 2n inputs. However, advanced design exploration tech-
the maximum achievable phase gradient. This is useful to niques may allow to find design points that reduce the voltage
achieve better control at the end-fire directions (θr → 90°) range, the number of required levels, and the multiplexer inputs
of the metasurface, as exemplified by Fig. 15(a). For in- while gracefully degrading the performance of the system. For
stance, we can achieve beam steering at θr > 70° only for instance, in our 2-bit implementation with the 2G unit cell, we
du < 5 µm. For angles even closer to θr = 90°, a design could take μc,1 = μc,2 = {0.1, 0.6} and still achieve reasonable
converting the incident wave into a surface wave may be performance, but with a 4× reduction of voltage range and num-
required [64]. In any case, note that such fine-grained con- ber of levels and greatly simplfying the multiplexing circuits.
trol at THz frequencies can only be achieved with graphene, Adaptive clusterization: The strength of the proposed de-
thanks to its support of plasmonic slow-wave propagation sign flow is its simplicity. By fixing the cluster size first and
in this frequency band. then statically building the super cells, the state matrix can be
r Increasing the number of bits provides better control on calculated easily. However, the rounding operation used in the
the phase as it allows to draw the phase gradient more number of unit cells per cluster (Equation (11)) introduces an
accurately, with more clusters and less unit cells per clus- error, specially for large unit cells, that is later amplified by
ter. This is of special importance in directions close to the the static building of super cells. Both issues can be allevi-
boresight (θr → 0°), where a subtle gradient is required. ated by simply inverting the design flow, i.e. obtaining the size
Fig. 15(b) exemplifies this for a design targeting θr = 5.37° of the super cell first, and then breaking it down into unequal
Authorized licensed use limited to: Indian Institute Of Technology (IIT) Mandi. Downloaded on April 22,2025 at 13:35:34 UTC from IEEE Xplore. Restrictions apply.
744 IEEE TRANSACTIONS ON NANOTECHNOLOGY, Volume 18, 2019

Fig. 15. Performance scalability analysis of the metasurface from the unit cell, coding, and complete device perspectives.

clusters. For instance, a super cell composed by 18 unit cells can To extract θr from (6), we square and sum both expressions:
be coded with 4, 5, 4, and 5 unit cells per state; otherwise, the
super cell would be statically coded to either 4 or 5 unit cells kr2 sin θr 2 cos φr 2 + kr2 sin θr 2 sin φr 2
per state, leading to a significant error. In a similar approach, the 2 2
dΦ dΦ
coding algorithm could dynamically adapt the number of states, = ki sin θi cos φi + + ki sin θi sin φi + ,
using fewer bits in those directions that require a very large dx dy
gradient. (18)
which, after applying basic trigonometry, becomes
VI. CONCLUSION
kr2 sin θr 2
This paper has presented the complete design, from the unit
2 2
cell up to the programming algorithm, of a reconfigurable digital dΦ dΦ
= ki sin θi cos φi + + ki sin θi sin φi + .
metamaterial for beam steering in the terahertz band. The tun- dx dy
ability of graphene is exploited at the unit cell level to provide a (19)
phase range close to 2π, whereas the generalized Snell’s law of
Isolating, we obtain
reflection has been used to derive the phase gradients required to
target the beam to the desired direction. The results confirm the θr = arcsin
validity of the approach, which for normal incidence achieves a 
dΦ 2 dΦ 2
very broad reflection range with angle-dependent beam widths (ki sin θi cos φi + dx ) + (ki sin θi sin φi + dy )
and steering errors. Considering normal incidence, the analyt- . (20)
kr
ical formulation also models forbidden (and unreasonable) re-
flection directions effectively as infinite gradients. Finally, the ACKNOWLEDGMENT
scalability analysis confirms that the beam width depends on
the size of the metasurface, the reflection range depends on the The authors would like to thank C. Süßmeier and the anony-
size of the unit cells, and the steering error and side lobe levels mous reviewers for their invaluable feedback.
depend on the number of phases that the graphene-based unit
cells can implement. Future works could leverage the compre- REFERENCES
hensive methodology developed herein to optimize the unit cell [1] H. T. Chen, A. J. Taylor, and N. Yu, “A review of metasurfaces: Physics and
design and phase gradient formation to reduce the overhead of applications,” Rep. Progress Phys., vol. 79, no. 7, 2016, Art. no. 076401.
the solution and further improve the beam steering performance. [2] S. B. Glybovski, S. A. Tretyakov, P. A. Belov, Y. S. Kivshar, and C.
R. Simovski, “Metasurfaces: From microwaves to visible,” Phys. Rep.,
vol. 634, pp. 1–72, 2016.
APPENDIX [3] P. Y. Chen and A. Alù, “Mantle cloaking using thin patterned metasur-
faces,” Phys. Rev. B Condensed Matter Mater. Phys., vol. 84, no. 20,
To extract φr from (6), we divide both expressions and apply pp. 1–13, 2011.
[4] X. Chen, L. Huang, H. Mühlenbernd, G. Li, and B. Bai, “Dual-polarity
basic trigonometry to obtain plasmonic metalens for visible light,” Nature Commun., no. 3, 2012, Art.
no. 1198.
ki sin θi sin φi + dΦ
dy [5] L. Li et al., “Electromagnetic reprogrammable coding-metasurface holo-
tan φr = , (16) grams,” Nature Commun., vol. 8, no. 1, pp. 1–7, 2017.
ki sin θi cos φi + dΦ
dx [6] S. Vellucci, A. Monti, M. Barbuto, A. Toscano, and F. Bilotti, “Satellite
applications of electromagnetic cloaking,” IEEE Trans. Antennas Propag.,
which yields vol. 65, no. 9, pp. 4931–4934, Sep. 2017.
[7] A. C. Tasolamprou, T. Koschny, M. Kafesaki, and C. M. Soukoulis,
ki sin θi sin φi + dΦ
dy “Near-infrared and optical beam steering and frequency splitting in air-
φr = arctan . (17) holes-in-silicon inverse photonic crystals,” ACS Photon., vol. 4, no. 11,
ki sin θi cos φi + dΦ
dx pp. 2782–2788, 2017.

Authorized licensed use limited to: Indian Institute Of Technology (IIT) Mandi. Downloaded on April 22,2025 at 13:35:34 UTC from IEEE Xplore. Restrictions apply.
HOSSEININEJAD et al.: DIGITAL METASURFACE BASED ON GRAPHENE: AN APPLICATION TO BEAM STEERING 745

[8] O. Tsilipakos, A. C. Tasolamprou, T. Koschny, M. Kafesaki, E. N. [34] S. E. Hosseininejad et al., “MAC-Oriented programmable terahertz PHY
Economou, and C. M. Soukoulis, “Pairing toroidal and magnetic dipole via graphene-based yagi-uda antennas,” in Proc. IEEE Wireless Commun.
resonances in elliptic dielectric rod metasurfaces for reconfigurable wave- Netw. Conf., 2018, pp. 1–6.
front manipulation in reflection,” Adv. Opt. Mater., vol. 6, no. 22, 2018, [35] E. Carrasco, M. Tamagnone, and J. Perruisseau-Carrier, “Tunable
Art. no. 1800633. graphene reflective cells for THz reflectarrays and generalized law of re-
[9] S. W. Qu et al., “Controlling dispersion characteristics of terahertz meta- flection,” Appl. Phys. Lett., vol. 102, no. 10, 2013, Art. no. 104103.
surface,” Sci. Rep., vol. 5, 2015, Art. no. 9367. [36] B. Orazbayev, M. Beruete, and I. Khromova, “Ultrafast beam steering
[10] Y. Zhang et al., “Broadband diffuse terahertz wave scattering by flexible based on graphene metamaterial,” in Proc. 11th Eur. Conf. Antennas
metasurface with randomized phase distribution,” Sci. Rep., vol. 6, 2016, Propag., 2017, pp. 3896–3899.
Art. no. 26875. [37] S. E. Hosseininejad et al., “Reprogrammable graphene-based metasurface
[11] S. Liu et al., “Anisotropic coding metamaterials and their powerful ma- mirror with adaptive focal point for THz imaging,” Sci. Rep., vol. 9, 2019,
nipulation of differently polarized terahertz waves,” Light, Sci. Appl., Art. no. 2868.
vol. 5, no. 5, 2016, Art. no. e16 076. [38] K. Rouhi, H. Rajabalipanah, and A. Abdolali, “Real-time and broadband
[12] S. W. Qu, H. Yi, B. J. Chen, K. B. Ng, and C. H. Chan, “Terahertz terahertz wave scattering manipulation via polarization-insensitive confor-
reflecting and transmitting metasurfaces,” Proc. IEEE, vol. 105, no. 6, mal graphene-based coding metasurfaces,” Annalen der Physik, vol. 530,
pp. 1166–1184, Jun. 2017. 2018, Art. no. 1700310.
[13] G. Oliveri, D. Werner, and A. Massa, “Reconfigurable electromagnet- [39] S. R. Biswas, C. E. Gutiérrez, A. Nemilentsau, I.-H. Lee, S.-H. Oh, P.
ics through metamaterials—A Review,” Proc. IEEE, vol. 103, no. 7, Avouris, and T. Low, “Tunable graphene metasurface reflectarray for
pp. 1034–1056, Jul. 2015. cloaking, illusion and focusing,” Phys. Rev. Appl., vol. 9, no. 3, 2018,
[14] F. Liu et al., “Programmable metasurfaces: State of the art and prospects,” Art. no. 034021.
in Proc. IEEE Int. Symp. Circuits Syst., 2018, pp. 1–5. [40] Z. Chang, B. You, L. S. Wu, M. Tang, Y. P. Zhang, and J. F. Mao, “A
[15] F. Liu et al., “Intelligent metasurfaces with continuously tunable local reconfigurable graphene reflectarray for generation of vortex THz waves,”
surface impedance for multiple reconfigurable functions,” Phys. Rev. Appl., IEEE Antennas Wireless Propag. Lett., vol. 15, pp. 1537–1540, 2016.
vol. 11, no. 4, 2019, Art. no. 044204. [41] Y. Chen et al., “Engineering the phase front of light with phase-
[16] S. Liu and T. J. Cui, “Concepts, working principles, and applications of change material based planar lenses,” Sci. Rep., vol. 5, 2015, Art.
coding and programmable metamaterials,” Adv. Opt. Mater., vol. 5, no. 22, no. 8660.
2017, Art. no. 1700624. [42] J. Perruisseau-Carrier, F. Bongard, R. Golubovic-Niciforovic, R. Torres-
[17] L. Liang et al., “Anomalous terahertz reflection and scattering by flexible Sanchez, and J. R. Mosig, “Contributions to the modeling and design
and conformal coding metamaterials,” Adv. Opt. Mater., vol. 3, no. 10, of reconfigurable reflecting cells embedding discrete control elements,”
pp. 1374–1380, 2015. IEEE Trans. Microw. Theory Techn., vol. 58, no. 6, pp. 1621–1628, Jun.
[18] C. Della Giovampaola and N. Engheta, “Digital metamaterials,” Nature 2010.
Mater., vol. 13, no. 12, pp. 1115–1121, 2014. [43] F. S. Ma, Y. S. Lin, X. H. Zhang, and C. Lee, “Tunable multiband terahertz
[19] T. J. Cui, M. Q. Qi, X. Wan, J. Zhao, and Q. Cheng, “Coding metamaterials, metamaterials using a reconfigurable electric split-ring resonator array,”
digital metamaterials and programming metamaterials,” Light Sci. Appl., Light Sci. Appl., vol. 3, 2014, Art. no. e171.
vol. 3, 2014, Art. no. e218. [44] J. Wang, W. B. Lu, J. L. Liu, and T. J. Cui, “Digital metamaterials using
[20] H. Yang et al., “A programmable metasurface with dynamic polarization, graphene,” Plasmonics, vol. 10, no. 5, pp. 1141–1145, 2015.
scattering and focusing control,” Sci. Rep., vol. 6, 2016, Art. no. 35692. [45] I. Akyildiz, J. Jornet, and C. Han, “TeraNets: Ultra-broadband communi-
[21] C. Huang, B. Sun, W. Pan, J. Cui, X. Wu, and X. Luo, “Dynamical beam cation networks in the terahertz band,” IEEE Wireless Commun., vol. 21,
manipulation based on 2-bit digitally-controlled coding metasurface,” Sci. no. 4, pp. 130–135, Aug. 2014.
Rep., vol. 7, pp. 1–8, Jan. 2017. [46] I. F. Akyildiz, C. Han, and S. Nie, “Combating the distance problem in the
[22] K. S. Novoselov, V. I. Fal’Ko, L. Colombo, P. R. Gellert, M. G. Schwab, millimeter wave and terahertz frequency bands,” IEEE Commun. Mag.,
and K. Kim, “A roadmap for graphene,” Nature, vol. 490, no. 7419, vol. 56, no. 6, pp. 102–108, Jun. 2018.
pp. 192–200, 2012. [47] I. F. Akyildiz and J. M. Jornet, “Realizing ultra-massive MIMO (1024×
[23] Y. Wu et al., “State-of-the-art graphene high-frequency electronics,” Nano 1024) communication in the (0.06–10) terahertz band,” Nano Commun.
Lett., vol. 12, no. 6, pp. 3062–3067, Jun. 2012. Netw., vol. 8, pp. 46–54, 2016.
[24] T. Low and P. Avouris, “Graphene plasmonics for terahertz to mid-infrared [48] X. Tan, Z. Sun, D. Koutsonikolas, and J. M. Jornet, “Enabling indoor
applications,” ACS Nano, vol. 8, no. 2, pp. 1086–1101, 2014. mobile millimeter-wave networks based on smart reflect-arrays,” in Proc.
[25] S. E. Hosseininejad et al., “Study of hybrid and pure plasmonic tera- INFOCOM, 2018, pp. 270–278.
hertz antennas based on graphene guided-wave structures,” Nano Com- [49] C. Liaskos, A. Tsioliaridou, A. Pitsillides, S. Ioannidis, and I. Akyildiz,
mun. Netw., vol. 12, pp. 34–42, 2017. “Using any surface to realize a new paradigm for wireless communica-
[26] J. M. Jornet and I. F. Akyildiz, “Graphene-based plasmonic nano- tions,” Commun. ACM, vol. 61, no. 11, pp. 30–33, 2018.
transceiver for terahertz band communication,” in Proc. Eur. Conf. An- [50] G. W. Hanson, “Dyadic green’s functions for an anisotropic , non-local
tennas Propag., 2014, pp. 492–496. model of biased graphene,” IEEE Trans. Antennas Propag., vol. 56, no. 3,
[27] D. Correas-Serrano and J. S. Gomez-Diaz, “Graphene-based antennas pp. 747–757, Mar. 2008.
for terahertz systems: A Review,” FERMAT, 2017. [Online]. Available: [51] S. Abadal, I. Llatser, A. Mestres, H. Lee, E. Alarcon, and A. Cabellos-
http://arxiv.org/abs/1704.00371 Aparicio, “Time-domain analysis of graphene-based miniaturized anten-
[28] I. Llatser, C. Kremers, A. Cabellos-Aparicio, E. Alarcón, and D. N. Chi- nas for ultra-short-range impulse radio communications,” IEEE Trans.
grin, “Comparison of the resonant frequency in graphene and metallic Commun., vol. 63, no. 4, pp. 1470–1482, Apr. 2015.
nano-antennas,” AIP Conf. Proc., vol. 1475, pp. 143–145, 2012. [52] L. Banszerus, M. Schmitz, S. Engels, J. Dauber, M. Oellers, and P. Gr,
[29] M. Tamagnone, J. Gomez-Diaz, J. R. Mosig, and J. Perruisseau-Carrier, “Ultra-high mobility graphene devices from chemical vapor deposition on
“Reconfigurable terahertz plasmonic antenna concept using a graphene reusable copper,” Sci. Adv., vol. 1, no. 6, 2015, Art. no. e1500222.
stack,” Appl. Phys. Lett., vol. 101, no. 21, 2012, Art. no. 214102. [53] “CST Microwave Studio,” 2018. [Online]. Available: http://www.cst.com
[30] S. Hosseininejad and N. Komjani, “Comparative analysis of graphene- [54] X. Zhou, T. Zhang, L. Chen, W. Hong, and X. Li, “A graphene-based
integrated slab waveguides for terahertz plasmonics,” Photon. hybrid plasmonic waveguide with ultra-deep subwavelength confinement,”
Nanostructures—Fundamentals Appl., vol. 20, no. Jul., pp. 59–67, J. Lightw. Technol., vol. 32, no. 21, pp. 4199–4203, Nov. 2014.
2016. [55] X. Huang, X. Zhang, Z. Hu, M. Aqeeli, and A. Alburaikan, “Design of
[31] S. E. Hosseininejad et al., “Terahertz dielectric resonator antenna coupled broadband and tunable terahertz absorbers based on graphene metasur-
to graphene plasmonic dipole,” in Proc. Eur. Conf. Antennas Propag., face: Equivalent circuit model approach,” IET Microw., Antennas Propag.,
2018, pp. 1–5. vol. 9, no. 4, pp. 307–312, Mar. 2014.
[32] Y. Huang, L. Wu, M. Tang, and J. Mao, “Design of a beam reconfigurable [56] Y. Kim et al., “Ultra thinning down to 4-μm using 300-mm wafer proven
THz antenna with graphene-based switchable high-impedance surface,” by 40-nm node 2gb dram for 3d multi-stack wow applications,” in Proc.
IEEE Trans. Nanotechnol., vol. 11, no. 4, pp. 836–842, Jul. 2012. IEEE Symp. VLSI Technol, Digest Tech. Papers 2014, pp. 1–2.
[33] Z. Xu, X. Dong, and J. Bornemann, “Design of a reconfigurable MIMO [57] N. Yu et al., “Light propagation with phase discontinuities: Generalized
system for THz communications based on graphene antennas,” IEEE laws of reflection and refraction,” Science, vol. 334, no. 6054, pp. 333–337,
Trans. Terahertz Sci. Technol., vol. 4, no. 5, pp. 609–617, Sep. 2014. 2011.

Authorized licensed use limited to: Indian Institute Of Technology (IIT) Mandi. Downloaded on April 22,2025 at 13:35:34 UTC from IEEE Xplore. Restrictions apply.
746 IEEE TRANSACTIONS ON NANOTECHNOLOGY, Volume 18, 2019

[58] S. Liu et al., “Convolution operations on coding metasurface to reach [63] J. Kokkoniemi, V. Petrov, D. Moltchanov, J. Lehtomaki, Y. Koucheryavy,
flexible and continuous controls of terahertz beams,” Adv. Sci., vol. 3, and M. Juntti, “Wideband terahertz band reflection and diffuse scattering
no. 10, 2016, Art. no. 1600156. measurements for beyond 5G indoor wireless networks,” in Proc. 22th
[59] C. Liaskos et al., “Design and development of software defined meta- Eur. Wireless Conf. Eur. Wireless, 2016, pp. 1–6.
materials for nanonetworks,” IEEE Circuits Syst. Mag., vol. 15, no. 4, [64] S. N. Tcvetkova, D. H. Kwon, A. Diaz-Rubio, and S. A. Tretyakov, “Nearly
pp. 12–25, Fourth Quarter 2015. perfect conversion of a propagating wave into a surface wave,” in Proc.
[60] S. Abadal et al., “Computing and communications for the software- Int. Conf. Electromagn. Adv. Appl., 2017, pp. 1426–1428.
defined metamaterial paradigm: A context analysis,” IEEE Access,
vol. 5, pp. 6225–6235, 2017.
[61] J. S. Gómez-Díaz et al., “Self-biased reconfigurable graphene stacks for
terahertz plasmonics,” Nature Commun., vol. 6, no. 6334, pp. 1–8, 2015.
[62] Y.-J. Yu, Y. Zhao, S. Ryu, L. E. Brus, K. S. Kim, and P. Kim, “Tuning the
graphene work function by electric field effect,” Nano Lett., vol. 9, no. 10,
pp. 3430–3434, 2009. Authors’ photographs and biographies not available at the time of publication.

Authorized licensed use limited to: Indian Institute Of Technology (IIT) Mandi. Downloaded on April 22,2025 at 13:35:34 UTC from IEEE Xplore. Restrictions apply.