Research Article Vol. 30, No.

5 / 28 Feb 2022 / Optics Express 7017

Nano-antenna array for high efficiency sunlight

harvesting
M ICHELE M IDRIO, 1 L UCA P IERANTONI , 2 S TEFANO B OSCOLO, 1
DAVIDE T RUCCOLO, 1 AND DAVIDE M ENCARELLI 2,*
1 Università degli Studi di Udine, Departimento Politecnico di Ingegneria e Architettura, Via delle Scienze
206, 33100 Udine, Italy
2 Università Politecnica delle Marche, Department of Information Engineering, Via Brecce Bianche 12,
60131 Ancona, Italy
* [email protected]

Abstract: Solar rectennas are promising devices for energy harvesting. Capability of rectennas
to convert incident light into useful energy depends on the antenna efficiency, that is the ratio
between the power transferred to the load vs the incoming power. In this work, we first emphasize
that for the efficiency to be calculated accurately, antennas need to be treated as receiving
devices, not as transmitting ones. Then, we propose an arrangement of antennas that differs
from those published so far in three respects: (1) the proposed arrangement is formed by an
array of nano-antennas with sub-wavelength inter-element spacing, (2) it comprises a reflecting
mirror, and (3) it allows for dual polarization operation. Through numerical simulations, we
show that the small lattice pitch we use is responsible for frequency flattening of the lattice
impedance over the whole solar spectrum, eventually allowing for excellent matching with the
antennas’ loads. Also, the small pitch allows for a smooth dependence of the receiving efficiency
on the angle of incidence of sunlight. Finally, we show numerically that the reflecting mirror
also allows for an almost complete cancellation of light scattered by the receiving antennas. The
final result is a polarization insensitive receiving theoretical efficiency larger than 70% over the
whole 300-3000 nm spectral range, with a less than 10% energy wasting due to back-scattering
of sunlight.

© 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement

1. Introduction
The increase of renewable energy [1] in the global energy mix [2] and of the energy efficiency
[3] are among the goals that United Nations (UN) have adopted as part of the 2030 Agenda for
Sustainable Development [4].
Energy efficiency is any means for measuring the energy-expenditure required to achieve a
given benefit [3]. Renewable energies are energies obtained from natural and inexhaustible
resources which can continually self-regenerate [1]. Sunlight is one of the major available sources
of renewable energy [5]. Two technologies are currently used to produce electricity from sunlight
[6]: photovoltaics [7] and concentrating solar power [8]. In photovoltaics, sunlight is converted
into electricity through the photo-electric effect [9]. A record lab cell efficiency of 26.7% was
demonstrated for mono-crystalline silicon [10], decreasing to 24.4% for multi-crystalline silicon
wafer-based technology. Using more complex and costly quadruple junctions [11], an efficiency
as large as 44.7% was obtained. A complete list of record-performance obtained with different
technologies can be found in [12].
Concentrating solar power exploits reflectors to focus sunlight on small areas and generate
steam for use in thermal electric plants [8]. A nearly 47% record solar to electric conversion
efficiency was obtained under southern Egypt climate [13]. Photovoltaics exploit a physical
effect related to the quantization of light energy. However, alternative mechanisms for harvesting
of sunlight may be envisaged by taking avantage of the wave nature of sunlight. In this respect,

#450020 https://doi.org/10.1364/OE.450020
Journal © 2022 Received 30 Nov 2021; revised 20 Jan 2022; accepted 20 Jan 2022; published 16 Feb 2022
 Research Article Vol. 30, No. 5 / 28 Feb 2022 / Optics Express 7018

conversion of light into electrical power can be obtained by realizing a so-called rectenna [14],
that is an antenna connected to a rectifying diode. Originally, rectennas were developed to feed
high amplitude atmospheric platforms through microwave beams pointed at the vehicles [15].
Later on, the idea was transferred to much higher frequencies with the aim of extracting DC
power from sunlight [16].
In contrast to photovoltaics that have a theoretical maximum efficiency limit around 33% for
single junction cells [17], raising to a theoretical maximum of 68% in an infinite stack of solar cell
under 1-sun illumination, and further up to 86% for concentrated solar light [18], the efficiency
of rectennas is solely limited by thermodynamic considerations and can be larger than 80% [19].
Such efficiencies have been experimentally demonstrated at microwaves [20]. Also, an efficiency
as large as 58% has been demonstrated at 897 GHz [21], but not yet for visible or infrared light.
For many years, rectification of sunlight was almost exclusively an academic field of research.
Commercial implementation of solar rectifiers clashed because of two main issues. First, typical
dimensions of antennas for solar harvesting are in the order of fractions of the wavelengths
they aim at receiving [22,23]. This in turn requires fabrication processes that can reliably and
reproducibly realize nanometric structures. Second, even the fastest electronic diode cannot
operate at frequencies of visible or near IR light [24,25]. Metal-Insulator-Metal (MIM) diodes
were proposed for rectifying THz waves [26,27]. However, when the capacity of those diodes
is combined with the intrinsic resistance of any receiving antenna, time constant too long to
harvest sunlight [28] is typically obtained. In fact, experimental attempts to rectify visible light
through MIM didoes resulted in rather limited conversion efficiencies [29]. In recent years,
the fast development of nanotechnology and the discovery of graphene [30] have changed the
perspectives of sunlight rectification [31]. Technologies to obtain the extremely tiny dimensions
needed in nanoantennas are at hand: 7-nm node is expected to be fully operative by 2021, and
1.4nm by 2029 [32]. As per rectification, the large mean-free path length of charges in graphene
[33] opened up the way to geometrical diodes [34,35]. Those are diodes where asymmetry comes
from the physical shape of the device. Rectennas working at frequencies as large as 28 THz
have already been demonstrated with this kind of diode [34], though no experimental results
are available yet at visible and IR wavelengths. Whatever the technology for manufacturing
nano-antennas and the choice of the rectifying diode will be, the final efficiency of any rectenna
will consist of two parts [36]: 1) the receiving efficiency by which the power of sunlight is
transferred to the antenna load (that is: to the diode) and 2) the efficiency by which the captured
light is transformed to DC power.
In the present paper, we do not discuss on the second of the above points and we show that if
the impedance of the diode approaches a purely resistive value, the layout of optical antennas
we propose may theoretically provide a remarkable receiving efficiency as large as 71%. As
per the physical manufacturability of a purely resistive diode, we observe that MIM are clearly
not adequate structures. Whereas, geometric or ballistic diodes could be promising candidates.
Indeed, it was shown that in these structures “the electrical contacts are laterally separated
rather than placed on the surface and the back side (substrate), the parasitic capacitance between
contacts is substantially lower than in a conventional vertical device of the same size” (the text is
quoted literally from Ref. [37], see also [38,39])". After having thoroughly discuss the ideal case
of a purely resistive load, we will numerically evaluate to what extent a capacitive part in the
diode impedance reduces the receiving efficiency of the proposed antennas layout.
We have highlighted the term "receiving efficiency", that is the ratio between the optical power
transferred to the load to the power of the incoming sunlight because we believe that receiving
efficiency is the correct way of measuring performance of antennas in the field of sunlight
harvesting. Readers will notice that this way of measuring the antenna performance differs
from those used in previously published papers, where antennas where regarded as transmitting
devices. For instance, Vandenbosch and Ma thoroughly investigated performance of linear dipole
 Research Article Vol. 30, No. 5 / 28 Feb 2022 / Optics Express 7019

antennas [40]. They showed a 65% transmitting efficiency when averaging over the whole
Planck’s blackbody radiation curve [40]. In a companion paper, the same Authors also showed
that the efficiency is 54% when loss in the antenna and a second order polynomial rectifier
impedance model is included in the model [36]. More recently, Zhao and coauthors claimed an
80% efficiency with a nanospiral antenna [41] (but with limitation in terms of polarization, as it
is remarked later).
Despite the existing literature, there are several reasons why a further study on the antenna
efficiency is worth, at least in our opinion:
1. existing studies considered single isolated antennas only. A real solar panel aiming
at efficient sunlight harvesting will certainly contain an array of antennas instead of a
single isolated one. In addition, antennas in the array will be closely packed for efficient
exploitation of the available space; coupling between neighboring antennas likely changes
the efficiency and should hence be included in the analysis;
2. also, existing studies mostly considered antennas in free space (or, at most, over a dielectric
slab). In our opinion, in any realistic solar panel, a metallic reflector will be placed below
the antennas both to enlarge gain in the broadside direction on top of the antennas, and
to possibly increase mechanical stability of the device while also providing an electrical
contact for collecting rectified currents;
3. the behavior of antennas should be stable with respect to the angle of incidence of the
incoming sunlight; certainly, this is not the case for isolated dipole antennas, nor for spirals;
4. dual polarization antennas will be needed as sunlight is not polarized. In this respect too,
dipoles certainly do not match the requirement. Nor do it spirals, which receive either a
left or a right handed circular polarization, not both.
In the present paper we address all the points above by considering the periodically loaded
surface schematically depicted in the top panel of Fig. 1. An array of optical antennas is arranged
in a square lattice with pitch Λ. Antennas (violet shapes in the figure) are metallic square patches
with lateral dimension LA and thickness HA , placed on top of a dielectric substrate with thickness
Hdiel and relative dielectric constant ϵdiel . A layer of metal beneath the dielectric substrate acts
as a back-reflector. Red rectangles in the figure are loads, possibly diodes allowing for THz
rectification. It is remarked that loads are only seen as resistive loads in the present paper, and
no speculations are made on their physical shape or working principle, nor on their rectifying
capabilities. The bottom panel of the figure shows a top view of the structure.

2. Some distinctive features of the periodically loaded surface in Fig. 1

In this Section we discuss some distinctive features of the periodically loaded surface in Fig. 1.
Coupling between neighboring antennas. Coupling between neighboring antennas in an array
is generally regarded as a detrimental effect that eventually causes decay of performance [42].
Luckily, this is not always the case. Realizing an array of strongly coupled antennas may indeed
allow to obtain performance stability both with respect to frequency and to angle of incidence of
incoming light [43,44,45].
As it was shown by B. A. Munk in Refs. [43,45], the interplay between the presence of
dielectrics and of a back-reflector along with a small inter-element spacing in an array of antennas
can help providing a stable resonant frequency over the angle of incidence. In fact, getting
antennas in the lattice close to each other brings to an almost constant current distribution over
the plane that contains the antennas. This way, the plane becomes similar to an aperture antenna
with a continuous current, which is the ideal situation for a broadband radiator [46]. Also, the
imaginary part of the mutual impedance between antennas in the lattice starts canceling each
 Research Article Vol. 30, No. 5 / 28 Feb 2022 / Optics Express 7020

Fig. 1. Top: schematic diagram of the periodically loaded surface considered in the present
work. Bottom: top view of the unit cell in the squared array.

other as one moves away from the reference element provided that the interelement spacings are
small (<0.4λ), eventually leading to less stored energy between the antennas, and therefore to
smaller imaginary components of the overall antenna impedance. The reader may find a detailed
 Research Article Vol. 30, No. 5 / 28 Feb 2022 / Optics Express 7021

explanation of the bandwidth enhancement that can be realised with small inter-element spacing
arrays in Sections 3, 4, 5 and 12 of Chapter 6 of Ref. [45] as well as in Appendices B and D of
the same book.
Figure 2 confirms the above statement and shows the benefit of using an array of antennas
instead of a single one in terms of antenna impedance in the lattice of Fig. 1. The figure refers to
a lattice where squared aluminum antennas with lateral size LA = 72nm and thickness HA = 90
nm are placed on a lattice with pitch Λ = 140nm. The substrate layer relative permittivity and
thickness are ϵdiel =2.25, and Hdiel = 71 nm, respectively. In these simulations, antennas are made
of aluminum, and the reflector is an 80 nm thick layer of gold. Complex dielectric constants with
parameters taken from Ref. [47] have been considered in these simulations for both aluminum
and gold. In the left panel of the figure, we show the real (black solid line) and imaginary (red
solid line) input impedance between any couple of neighboring antennas in the lattice. On the
right side of the figure, the curves refer to the case of a single antenna, i.e. an antenna formed by
two isolated aluminum squares. Input impedance is the impedance measured at the the antenna
terminals (for a single isolated antenna), or between any couple of neighbouring antennas (for
the lattice) when antennas are used as transmitting devices. The figure confirms that the lattice
stabilizes the antenna impedance. When the antennas are used as receiving devices, this allows
for an easier matching to the load impedance.

Fig. 2. Input impedance (black: real part, read: imaginary part) of aluminum squared
antennas. Left panel: impedance for antennas in a lattice. Right panel: impedance for a
single isolated antenna. Input impedance is the impedance measured at the the antenna
terminals (for a single isolated antenna), or between any couple of neighbouring antennas
(for the lattice) when antennas are used as transmitting devices.

Back-reflector. An even more important feature of the periodically loaded surface in Fig. 1 is
the presence of the back-reflector that allows to increase the amount of power delivered to the
antennas’ loads, and therefore the receiving efficiency. To better clarify the crucial role of the
back-reflector, it is worth recalling two general properties of receiving antennas.

• Property 1. The receiving area of an array of antennas with no ground-plane is, at most,
half its physical size [48]. Let us see the practical implications of this property. Suppose
an incident wave impinges on an array of ideally lossless antennas closed on perfectly
matched loads. The above property means that, at best, the array can transfer half of the
incident power to its loads. The remaining half is "lost" to scattering.
We believe it would be natural to say that the harvesting efficiency of such an array is, at
best, 50%. At this stage, we can anticipate a point that, at least in our opinion, is crucial
for realistically assess potentialities of rectennas in the field of solar harvesting.
 Research Article Vol. 30, No. 5 / 28 Feb 2022 / Optics Express 7022

In Refs. [40] and [41], antennas performance was estimated by resorting to the transmitting
efficiency (see Eq. (2) below), namely to the ratio between the power radiated by the
antenna vs. the power that is accepted by the antenna itself. For the ideal case we just
mentioned, i.e. in the absence of ohmic losses in the antenna, such definition would yield
a 100% antenna efficiency. The discrepancy with the intuitive figure of 50% we mentioned
above arises from the difference that exists between the concepts of antenna efficiency for
transmitting and receiving antennas, respectively. We will discuss the point in more details
below.
• Property 2. The receiving area of an array of antennas with ground-plane and matched
loads can equal its physical size [49].
In practical terms, the ensemble of these two properties can be summarized as follows:
only in the presence of a ground-plane all the energy incident upon the array can be
absorbed by matched loads. This result is of paramount importance for the application we
are considering here, where the ultimate goal is the maximization of receiving efficiency
of solar radiation.

Dual polarization operations. As a final remark on the proposed periodically loaded surface,
we highlight that it easily allows for dual polarization since its geometrical shape is invariant by
90 degrees rotation.

3. Estimation of the harvesting efficiency

In this section we discuss the numerical procedure used to compute the harvesting efficiency
of periodically loaded surface in Fig. 1. We report to the figure of merit proposed in Ref. [40],
namely the total harvesting efficiency:
∫∞
dλP(λ)ηA (λ)
ηTOT = 0 ∫ ∞ (1)
0
dλP(λ)

where P(λ) is the Power Spectral Density (PSD) of the incoming sunlight, and ηA (λ) is the
antenna spectral efficiency, respectively. In contrast to Ref. [40], in the present paper:
1. the PSD of real solar spectrum at sea level is considered, as given by the AM1.5 Direct
International standard [50] (see the blue curve in Fig. 3) instead of Planck’s law of
blackbody radiation;
2. most importantly, ηA (λ) is the receiving rather than transmitting antennas’ efficiency. This
is the key point we briefly mentioned above and some detailed comments are now worth.
In Refs. [40] and [41] the following definition of antenna efficiency was used:
Prad (λ)
ηA,TX (λ) = (2)
Prad (λ) + Ploss (λ)
Efficiency was computed as the ratio between the amount of power radiated by the antenna
Prad (λ) vs. the power that is accepted by the antenna itself (Prad (λ) + Ploss (λ)). By contrast, we
use here the following definition:

PLOAD (λ)
ηA,RX (λ) = (3)
PINCIDENT (λ)
According to Eq. (3), the efficiency is defined as the ratio between the power delivered to
the load vs. the incoming power. In the case of a periodic structure as our periodically loaded
 Research Article Vol. 30, No. 5 / 28 Feb 2022 / Optics Express 7023

Fig. 3. Example of the results used to optimize the periodically loaded surface of Fig. 1. The
blue curve is the solar PSD [50]. In the left panel of the Figure, the red curve is the receiving
antenna spectral efficiency ηA,RX (λ). The product between these two curves delimits the
green-filled region. The ratio between the areas of the green-filled region, and the region
below the blue curve gives the total harvesting efficiency. In the right panel of the Figure,
the red curve is the transmitting efficiency.

surface, the incoming power is the amount of power that sunlight carries on the elementary cell
of the periodic lattice (see the dashed line in Fig. 1).
Let us highlight the difference between the two definitions. Equation (2) is the definition of
antenna efficiency that is customarily employed for antennas used as transmitters. Often, it is also
referred to as radiating efficiency (see Ref. [48], Section 11.5) or conduction-dielectric efficiency
(see Ref. [51], Section 2.14). It expresses the ratio of the power radiated to the power input of the
antenna. In equivalent terms, it also expresses the ratio between the antenna gain and directivity
(see Ref. [48], Section 2.15).
In solar harvesting, however, antennas are used as receiving rather than transmitting devices.
For a receiving antenna, the concept of equivalent areas is usually employed to describe the
power capturing characteristics of the antenna when a wave impinges on it (see Ref. [51], Section
2.15.2). Typically, receiving antennas are characterized by their capture area, defined as the
equivalent area, which when multiplied by the incident power density gives the total power
captured, i.e. collected (transferred to the antenna load), scattered, or dissipated by the receiving
antenna. The following general rule applies (Ref. [51], Section 2.15.2):

Capture Area = Effective Area + Scattering Area + Loss Area (4)

Here the effective area is the ratio of the available power at the antenna’s load terminals to the
power flux density of a plane wave incident on the antenna, the scattering area is the equivalent
area, which, when multiplied by the incident power density, is equal to the scattered or re-radiated
power, and the loss area is the equivalent area, which when multiplied by the incident power
density leads to the power dissipated as heat in the antenna itself. It is easily seen that the
definition of antenna efficiency we use in the present paper, Eq. (3), is related to the antenna
capture area as follows:

PLOAD (λ) EffectiveArea(λ)

ηA,RX (λ) = = (5)
PINCIDENT (λ) CaptureArea(λ)
At this stage, a simple and practical comparison between Eq. (2) and Eq. (3) is worth. Let
consider once more the case of the ideally lossless and matched antenna we had introduced above.
In that case, 100% efficiency would be obtained in (2) : all the power entering into the antenna
 Research Article Vol. 30, No. 5 / 28 Feb 2022 / Optics Express 7024

would be radiated. However, the same efficiency is not necessarily obtained in the receiving
case (Eq. (3)). In fact, the scattering area, i.e. the amount of power that is "lost" to scattering
needs to be considered in the receiving case. As we have recalled above, if the antenna has no
back-reflector, the amount of scattered power is at least as large as the power that is transferred to
the load. The two powers equal each other if the load impedance is matched. In that case, the
receiving efficiency amounts to a maximum 50%, the figure we had previously said we would
naturally expect in this case. Only when a back-reflector is added to the antenna the scattered
power can be completely canceled out, and 100% receiving efficiency can be obtained.
We summarize the present session with the following general remark: Eq. (2) over-estimates
the antenna efficiency as it overlooks the power lost to scattering when the antenna is used as a
receiving device. For antennas without back-reflectors, the over-estimation is by a factor of 2
at minimum if the antenna and load impedances are matched, and possibly larger than 2 in the
absence of matching.

4. Optimization of receiving efficiency in the periodically loaded surface of Fig. 1

A thorough numerical optimization of the periodically loaded surface in Fig. 1 has been conducted
numerically. Results have been cross checked by using two commercial frequency domain Finite
Elements Method (FEM) codes (HFSS [52] and Comsol [53]), along with an one in-house
Finite-Difference Time-Domain (FDTD) routine. The single elementary cell of the periodically
loaded surface (see the dashed line in Fig. 1) was enclosed within periodic boundary conditions.
A plane wave was injected from top of the structure. In a first set of simulations, the plane
wave was linearly polarized along either the x or y axis in Fig. 1, and it propagated normally to
periodically loaded surface (that is, with the incidence angle θ=0, see Fig. 1). Later on, both the
azimuthal angle θ and the polar angles ϕ were varied, and a generic elliptic polarization state of
the incoming plane wave was used. Loads (red shapes in Fig. 1) were modeled as purely resistive
components, with resistance RL . Loads were connected at the bottom side of the antenna, i.e.
they laid on the ϵr = 2.25 substrate and they were placed in physical contact with the antenna
patches. Numerically, loads were modeled as lumped ports (Comsol [53], HFSS [52]) or lumped
RLC boundaries (HFSS [52]), or yet as 1nm thick layers of conducting material (Comsol [53]:
transition boundary conditions, HFSS [52]: block material with assigned conductivity). In FDTD
modeling, loads were modeled as conducting blocks (with a minimum grid thickness of 4 nm).
In all cases, calculation of load delivered to the power was done by numerically evaluating the
voltage drop VL at the load terminals, and then computing power as PL = |VL | 2 /(2RL ). When
using the lumped port modeling, the power to the load could also be evaluated by resorting to
the scattering parameter |S21 | between the plane wave port and the resistive lumped port. No
appreciable differences in computed efficiencies were observed when changing the numerical
model for the loads. When we performed the simulations with linearly x− or y−polarized electric
field impinging on the antennas from normal directions, symmetries could be exploited to reduce
the size of the computational domain. This way, the number of elements needed to mesh the
domain was in the order of ∼ 120.000, with minimum element size <5 nm for the top load and
antenna surfaces. Power reflected to the bistatic direction, and the power dissipated in the antenna
metallic patches and the back-reflector were also computed. Two materials for antennas and
reflector were chosen, and different combinations of materials were tested. Following results
from Ref. [40], only aluminum, gold, and silver were considered. Six parameters were then
spanned: the lattice pitch Λ, the height HA and width LA of the antennas, the thickness Hdiel of
the dielectric substrate, the width WL and the resistance RL of the loads.
For any set of spanned parameters, curves like those in Fig. 3 and Fig. 4, as well as contour
plots like in Fig. 5 were computed. In Fig. 3, the blue curve is the AM1.5 solar PSD [50]. In the
left panel of the Figure, the red curve is the percentage of incoming power that is transferred to
the load, i.e. the receiving antenna spectral efficiency ηA,RX (λ) of Eq. (3). The product of the two
 Research Article Vol. 30, No. 5 / 28 Feb 2022 / Optics Express 7025

curves gives the green-filled region. The ratio between the area of the green-filled region and the
area below the blue curve gives the total harvesting efficiency (Eq. (1)) [40]. In the right panel
of the Figure, the red curve is the transmitting efficiency, i.e. the efficiency we have computed
through Eq. (2) by assuming that the loads in Fig. 1 were replaced by lumped ports and that
power were injected into the antennas through those ports.

Fig. 4. Similar to Fig. 3, the figure shows examples of the results used to optimize the
periodically loaded surface of Fig. 1. In both the panels, the blue curve is the solar PSD
[44]. The red curve is the percentage of power that is reflected (left panel) or dissipated
(right panel) by antennas in the lattice. The product between these two curves delimits the
green-filled region. The ratio between the areas of the green-filled region, and the region
below the blue curve gives the total percentage of power that is reflected (left panel) or
absorbed (right panel).

Fig. 5. Left panel. Numerically estimated total harvesting efficiency vs. antenna
length/width and substrate dielectric thickness for a lattice with lattice pitch Λ = 140
nm, load width WL = 40 nm, load resistance RL = 236Ω, antenna thickness HA = 90 nm.
Right panel: Numerically estimated total harvesting efficiency vs. load resistance and
antenna thickness for a lattice with lattice pitch Λ = 140 nm, load width WL = 40 nm, antenna
length/width: LA = 72nm, bottom dielectric thickness Hdiel = 71 nm. In both the panels,
material used for antennas and substrate were aluminum and gold, respectively.

Similar results are shown in Fig. 4. In the left and right panel of this figure the percentage of
power that is reflected or dissipated in the metallic part of the periodically loaded surface are
show, respectively. The green filled regions in both the panel allows to compute the cumulative
 Research Article Vol. 30, No. 5 / 28 Feb 2022 / Optics Express 7026

reflected and dissipated power, as averaged over the AM1.5 Direct International standard [50]
solar PSD.
The following set of parameters was found to provide the largest percentage of power transferred
to the load:

• Material for antenna and substrate: aluminum and gold, respectively

• Lattice pitch: Λ = 140 nm;
• Antenna width: LA = 72 nm;
• Antenna thickness: HA = 90 nm;

• Dielectric substrate thickness: Hdiel = 71 nm;

• Load width: WL = 40 nm;
• Load resistance: RL = 236 Ω;

When averaging over the AM1.5 solar spectral distribution, load, reflected and dissipated
power percentages equal to 71,2%, 8.8% and 20%, respectively, were found in the receiving case.
By contrast, efficiency was ∼ 80% in the transmitting case, leaving a ∼ 20% power loss due to
dissipation in the metallic parts of the structure (antennas and back-bacrflector). These values
reconfirm the statement we made at the end of Section 3. Transmitting efficiency does not allow
to estimate correctly the performance of an antenna aimed at receiving a radiation. In fact, the
transmission efficiency can not distinguish between the amount of radiation that is effectively
transferred to the load from the radiation that is scattered in the receiving process. In fact,
transmission efficiency is larger than receiving efficiency because transmission efficiency is given
by the ratio between radiated power and sum of radiated power plus dissipated power (see Eq. (2)).
By contrast, receiving efficiency Eq. (3) is the ratio between power transferred to the load and
sum of power transferred to the load plus dissipated power plus scattered power. The amount of
power transferred to the load in a receiving antenna closed on a perfectly matched load equals the
transmitted power of a transmitting antenna fed by a matched generator. Dissipated powers also
equal each other in transmitting and receiving antenna under ideal matching conditions. The
main difference between transmitting and receiving efficiency depends therefore on scattered
power, which is accounted for in Eq. (3), and neglected in Eq. (2). As a consequence, Eq. (2)
over-estimates the real receiving efficiency of the antenna. Recall that in the absence of a
back-reflector, the scattered power can be even larger than the power transferred to the load, and
so the over-estimation can be high, especially in absence of matching between the antenna and
load impedances. A few comments are in order.
• First of all: 71.2% is the main result of the present paper. We remark once more that
this is a receiving efficiency, that is percentage of power that is really transferred from
sunlight to the antennas’ loads (under normal illumination). Also, this result was checked
to hold for any state of polarization of the incoming light. This is expected because any
polarization state can always be decomposed into a combination of linear x and y states
(see the Cartesian frame in Fig. 1), possibly with a phase mismatch. Power transfer to
loads, however, is phase independent, and the geometrical shape of loads and antennas
causes the part of power that is not absorbed along the x-axis to be absorbed along the
y-one, and vice-versa. At the best of our knowledge, this is the largest receiving efficiency
ever estimated so far for unpolarized light.
• The left panel of Fig. 4 shows a result analogous to that of Fig. 3, but now for the power that
is reflected, i.e. scattered, by the periodically loaded surface. The red curve in the figure
 Research Article Vol. 30, No. 5 / 28 Feb 2022 / Optics Express 7027

shows that a small amount of power is reflected in the visible spectrum wavelengths, while
severe reflection occurs for wavelengths beyond ∼ 2 µm. However, when the reflection
curve is averaged over the AM1.5 solar spectrum, a very limited overall (8.8%) reflection
is found. Similar curves are finally shown in the right panel of the figure for the power
dissipated in the antennas. When the dissipation curve is averaged over the AM1.5 solar
spectrum, an overall 20% dissipation value is found.
If the computation were restricted to the wavelengths of the visible range, a ∼ 4% reflection
would be obtained. We remark once more that the presence of the back-reflector plays a
crucial role in the reduction of reflection, as it was proved experimentally in a lattice of
gold dipole antennas on top of a gold reflector [54].

• It is worth mentioning that further refinement and optimization of the geometry would
lead to sligthly higher efficiency: shaping properly the patch section with respect the z
direction, and adjusting the z-position of the metal with respect to the dielectric slab, could
lead to an efficiency higher than 75%. In this work, we just avoided to consider such
extra-complicated solutions.
Also, we numerically verified that the efficiency did not change to within a tenth of percent
if the aluminum antenna were coated with a 1nm thin layer of aluminum oxide.
• Finally, we remark some interesting features that can be inferred by looking at the difference
between the left and right panels of Fig. 3, that is between receiving and transmitting
efficiencies. The two curves are almost identical for wavelength smaller than ≃ 1µm, and
remarkably different for wavelengths larger than ≃ 1µm. Recall that the receiving curve
is shown for a resistive load RL = 236 Ω. Whereas, source impedance is assumed to be
perfectly matched to the antenna impedance in the transmitting case. By inspecting the
left panel of Fig. 2, one may recognize the effect of poor load matching in the receiving
case. When wavelengths smaller than ≃ 1µm, the imaginary part of the lattice load is
almost null, and its real part close to 236 Ω. Whereas, lattice resistance starts to reduce for
λ>1 µm, and a reactive component starts to appear, causing poor matching with the load.
Consistently, reflection from the lattice increases (see left panel of Fig. 4).
The contour plots of Fig. 5 show the dependence of the antenna efficiency on variations of
geometrical dimensions around the optimum value. For instance, the efficiency vs. antenna size
50 nm < LA < 90 nm and dielectric substrate thickness 50 nm < Hdiel < 90 nm is shown in the
left panel of the figure. Optimum values listed above were used for lattice pitch, load resistance
and width, and antenna thickness. In a similar fashion, the right panel of the figure shows the
estimated efficiency variation vs the antenna thickness and load resistance. In this case too other
parameters were set to be the optimum ones.
Figure 6 shows the dependence of the efficiency on the load impedance. In the left panel of
the figure, a purely resistive load is considered, and the efficiency vs. load resistance is shown
for a lattice having the optimum parameters listed above. As previously mentioned, the largest
performance leading to ∼ 71% efficiency is observed for load resistances RL ≈ 236Ω, while
efficiency values larger than 50% are still obtained for RL in the range between ∼ 100 to ∼ 1000Ω.
In the right panel of the figure, we show an example of how the efficiency worsens when the load
is no longer purely resistive. Figure refers to the case of a parallel RC load, with R = 50, 236 and
1000 Ω for light blue, red and black curves, respectively. We remark that a precise computation
of the effect of a generic load ZL (λ) on the antenna efficiency cannot be done rigorously because
the field that is back-scattered by the antenna depends on the load itself. In other words, only
a numerical computation could allow to determine exactly how much of the incoming field is
scattered and how much is transferred to the load.
 Research Article Vol. 30, No. 5 / 28 Feb 2022 / Optics Express 7028

Fig. 6. Left panel. Numerically estimated total harvesting efficiency vs. load resistance for
the optimized lattice with the following parameters: lattice pitch Λ = 140nm, load width
WL = 40 nm, antenna length/width: LA = 72 nm, antenna thickness HA = 90 nm, bottom
dielectric thickness Hdiel = 71 nm. Right panel. Numerically estimated efficiency for a
parallel RC load. Light blue, red and black curves refer to load resistance equal to 50, 236
and 1000 Ω, respectively.

An approximate computation can be done as follows. Any receiving antenna can be regarded
as a voltage generator with assigned open circuit voltage V0 (λ) and internal resistance ZA (λ),
closed on a load with impedance ZL (λ). The amount of real power that is transferred to the load is
PLOAD (λ) = |V0 (λ)| 2 ReZL (λ)/|ZL (λ) + ZA (λ)| 2 . Using this expression in Eqs. (3) and (1), along
with the antenna internal impedance ZA (λ) shown in the left panel of Fig. 2 an estimation of the
harvesting efficiency for any generic load ZL (λ) can be obtained.
In both the panels, performance drop with respect to the ideal and optimum case of purely
resistive load having RL = 236 Ω is due to the mismatch between the antenna and load impedances.
In particular, in the right panel of the figure, the load is short-circuited for increasing values of the
RC time constant, and very little real power is transferred to the resistive component of the load.
Finally, Fig. 7 shows the estimated harvesting efficiency as a function of the angle of incidence
of impinging radiation (that is, as a function of the azimuthal angle of a polar coordinate
system centered on the elementary cell of the lattice). In the left panel of the figure, spectral
efficiency is shown for angle of incidence θ i = 0 degrees (black line), θ i = 20 degrees (red line),
θ i = 40 degrees (blue line) (red line), and θ i = 60 degrees (green line), respectively. A reduction
of efficiency for increasing angle of incidence is observed, along with a shift towards longer
wavelengths (where the solar spectrum content is poor). In the right panel of the figure, the
overall receiving efficiency averaged over the AM1.5 solar spectrum is shown (red curve). A
fairly smooth dependence on the angle of incidence is observed. Efficiency remains larger than
60% up to 35 degrees, and larger than 50% up to 50 degrees. Percentages of power dissipated
in the metallic parts of the lattice (black line) and power reflected n the bistatic direction (blue
line) are also shown in the figure. Similar to the case of the receiving efficiency (red curve), also
dissipated and reflected efficiencies are averaged over the AM1.5 solar spectrum.
 Research Article Vol. 30, No. 5 / 28 Feb 2022 / Optics Express 7029

Fig. 7. Left panel. Receiving efficiency curves for light impinging on the array of loaded
resonators with the angle of incidence θ i = 0 degrees (black line), θ i = 20 degrees (red line),
θ i = 40 degrees (blue line) (red line),and θ i = 60 degrees (green line), respectively. Right
panel: Dependence of the total harvesting efficiency (red curve), power dissipated in metallic
parts of the lattice (black curve) and reflected power (light blue curve) for the optimized
lattice vs the angle of incidence of incident light.

5. Estimation of voltage drop at terminals of the antennas’ loads

We previously underlined that antenna loads have been assumed here to be simple resistors, while
rectifying diodes will be required in real applications. We did not speculate in this paper on how
diodes will be fabricated, nor on which physical effect their rectification capabilities will ground.
Yet, we believe it is worth to give an estimate of the magnitude of the voltage drop those diodes
will be presented at their terminals when the sunlight illuminates the antenna array. To give such
estimate, we first evaluate the upper limit to the voltage drop, i.e. the voltage magnitude that
we would find if the sunlight could be completely absorbed by the array. Later on, we compare
this upper limit with realistic values obtained by the lattice with optimum parameters as in the
previous section.
Let us assume initially that the following ideal conditions apply:

1. the loads connecting neighboring antennas (red shapes in Fig. 1) matches perfectly the antenna
internal impedance Ri at all frequencies;
2. the effective area of each antenna equals the geometrical area of the unit cell; that is:
antennas are lossless, their orientation is optimal with respect to the incoming light, and
depolarization factor is unitary so that all the incoming power is delivered to the loads.

Under these conditions, the receiving antenna effective length is [48]:

√︄ √︄ √︃
Ri Aeff Ri Ageom Ri
hRX = 2 =2 = 2Λ (6)
Z0 Z0 Z0

where Z0 = 120π is the wave impedance in vacuum. Keeping in mind that optimal antenna
orientation has been assumed, as well as that the antenna internal impedance is matched to the
 Research Article Vol. 30, No. 5 / 28 Feb 2022 / Optics Express 7030

load, the voltage drop at load terminals is:

√︃
| Ei · hRX | | Ei | · | hRX | Ri
max
Vload = = = ΛEi (7)
2 2 Z0
Here | Ei | is the amplitude of the incoming electric field. Superscript "max" has been added to
remind that this is the largest voltage load that the array could provide under the ideal conditions
listed above.
In the case of solar illumination, the incoming electric field is the superposition of waves with
different frequencies, having the PSD given by the blue curve in Fig. 3 [50].
Computation of the amplitude of the electric field due to the whole solar spectrum can be done
as follows. We divide the PSD into slices sampled over a given set of wavelengths. Then, we
convert the solar PSD:

PSD(f )∆f = PSD(λ)∆λ =⇒ PSD(f ) = PSD(λ)c0 /f 2 (8)

Given that the sunlight is unpolarized, half of the PSD is assigned to each of two orthogonal
polarizations. For each of them, the amplitude of the electric field within each slice of the PSD is
given by
| Ei sun | 2 PSD(fk )
= ∆fk (9)
2Z0 2
with fk and ∆fk the central frequency and the bandwidth of the k-th slice, respectively. Electric
field due to solar radiation therefore is, for each of two orthogonal polarizations:
∑︂ √︁
Etot sun (t) = Z0 PSD(fk )∆fk cos(2πfk t + ϕk ) (10)
k

with ϕk a [0,2π] uniformly distributed random variable accounting for spectral incoherence.
From Eq. (7), the voltage drop at the antenna terminals turns out to be:
√︄
∑︂ | Ei (fk ) · hRX (fk ) | ∑︂ Ri (fk ) ∑︂ √︁
Vload =
max
= sun
| Ei (fk ) | Λ = Λ Ri (fk )PSD(fk )∆fk (11)
k
2 k
Z0 k

where we have used the assumption that under ideal conditions the antenna effective length
is constant (and equal to its maximum attainable value, see Eq. (6)) at all frequencies. Once
more, we remark that the result holds for each orthogonal polarization. Each spectral component
contributes to the total RMS voltage with the term:

1 Λ2 Ri (fk )PSD(fk )∆fk

∫
2
(VRMS,k )max
= Λ2 Ri (fk )PSD(fk )∆fk cos2 (x)dx = (12)
2π 2
For each of two orthogonal polarizations, the total RMS voltage at the load terminals thus is:
√︄ √︄
∑︂
2 Λ ∑︂
max
VRMS = (VRMS,k )max = √ Ri (fk )PSD(fk )∆fk (13)
k 2 k

For the sake of simplicity in the calculation, we assume that the lattice resistance is independent
of frequency, with a value equal to the optimum we found in the preceding section: Ri = 236Ω.
Also, Λ = 140nm, so that VRMS max ≈ 45µV is numerically found from 13 for the AM1.5 solar
spectrum [50].
We now turn to the evaluation of voltage drop at load terminals in a realistic antenna where
both losses of metals and matching between internal antenna impedance and load are taken into
 Research Article Vol. 30, No. 5 / 28 Feb 2022 / Optics Express 7031

Fig. 8. Numerically computed voltage across a RL = 236Ω resistance load for the optimized
antenna array, for an | Ei |=1V/m incident electric field. The dashed blue line is the upper
limit for the load voltage, as computed by assuming that the antenna impedance matches the
load at all frequencies, and that the antenna efficiency is unitary.

account. Figure 8 shows the numerically computed voltage drop VLoad at the terminals of the
array with optimized parameters when the amplitude of the incident field is | Ei |=1V/m at all
frequencies. For comparison, in the figure we also plot with dashed blue line the upper limit
VLoad max = Λ | Ei | (RL /Z0 ) ≈ 111nV for Ei = 1 V/m.
√︁

Similar to computations in Eqs. (10)–(12) above, the total RMS voltage at load terminals due
to the whole AM1.5 spectrum, and for each of two orthogonal polarizations, can be evaluated as
follows: √︄
∑︂ (|Ei sun (fk ) |)
VRMS = ( VLoad (fk ))2 (14)
k
1V/m

Numerically, we found VRMS ≈ 36µV to be comparewd with VRMS max ≈ 45µV. Under normal
illumination, and due due polarization insensitiveness of the periodically loaded surface of Fig. 1,
this is the RMS voltage that is expected to be found at the terminals of all the four diodes that
connect any antenna to neighboring ones.

6. Conclusion
In this paper, a periodically loaded surface has been proposed for harvesting of sunlight. Estimates
of harvesting efficiency have been discussed by several authors in recent years [36,40,41]. Yet,
the present paper distinguishes from previous one in two main respects. First, we proposed
to measure the antenna performance in a different way with respect to previously published
papers. In fact, while in existing literature a transmitting efficiency was used, we referred here
to receiving efficiency. This was made because, on the one hand, antennas for solar harvesting
are receiving devices. On the other hand, as we proved in the paper, transmitting efficiency
may greatly over-estimate the real receiving antenna efficiency due to unavoidable scattering of
incoming radiation. The second novelty of the present paper is the use of a structure comprising
 Research Article Vol. 30, No. 5 / 28 Feb 2022 / Optics Express 7032

an infinite array of closely packed antennas placed over a reflecting mirror instead of a single
isolated antenna in free space or over a dielectric slab. Both the array arrangement and the use of
a reflector represent a more realistic device for harvesting of sunlight than a single antenna. In
fact, close packing of antennas avoids waste of available space. Mirror increases the mechanical
robustness of the structure and also constitutes an electrical contact for extracting converted
energy. Besides, we have proved that both the array arrangement and the reflector have beneficial
effects on efficiency. By using an array with a pitch much smaller than the solar wavelengths,
even those in the visible range, a flat frequency response of the antenna impedance was obtained,
allowing for matching to the load. In addition, the small array pitch also brought to a fair
independence of the harvesting efficiency on the angle of incidence of sunlight. Finally, by
using a square lattice, real dual-polarization operation of the array was made possible. Use of
mirror allowed to enlarge the antenna effective area and to reduce the scattering area at the same
time, finally leading to an extremely efficient power transfer from antennas to loads. We have
underlined in the paper that only in the presence of the mirror a 100 % receiving efficiency is
theoretically possible, while the ultimate upper bound for antennas without mirror can not exceed
50% due to intrinsic scattering of incoming radiation. By carefully optimizing the geometrical
parameters of the proposed periodically loaded surface, a receiving efficiency as high as 71%
was obtained. As a side effect of such a large receiving efficiency, we also proved that when
averaging over the sunlight spectrum, an overall reflection smaller than 10% is found. The
proposed structure behaves as a sort of stealth surface for sunlight, with a 10-fold reduction of the
radar cross section in the 300 to 3000 nm range with respect to the reflection from a flat metallic
surface.
Funding. Horizon 2020 Framework Programme (101006963).
Acknowledgments. This project has received funding from the European Union’s Horizon 2020 research and
innovation program under grant agreement No 101006963 (GreEnergy). This document reflects only the view of the
authors. The funding agency is not responsible for any use that may be made of the information it contains.
Disclosures. The authors declare no conflicts of interest.
Data availability. Data underlying the results presented in this paper are not publicly available at this time but may
be obtained from the authors upon reasonable request.

References
1. B. Sorensen, Renewable Energy (Academic Press, 2017).
2. S. Tagliapietra, Global Energy Fundamentals: Economics, Politics, and Technology (Cambridge University Press,
2020).
3. B. E. D. Martinez and T. Wagner, Energy Efficiency (Elsevier Science, 2019).
4. UN General Assembly, “Transforming our world: the 2030 Agenda for Sustainable Development,” 21 October 2015,
A/RES/70/1, https://www.refworld.org/docid/57b6e3e44.html.
5. U.S. Energy Information Administration, “What is Energy? Sources of energy,” https://www.eia.gov/energyexplained/
what-is-energy/sources-of-energy.php.
6. M. A. Lange, Vulnerability of Energy to Climate (R. A. Pielke, Academic Press, 2013).
7. W. v. S. A. Reinders and P. Verlinden, Front Matter (John Wiley and Sons, Ltd, 2016)., pp. i–xxxv
8. W. S. Keith Lovegrove, Concentrating Solar Power Technology: Principles, Developments and Applications (Keith
Lovegrove, Wes Stein, 2012).
9. M. Green, “Photovoltaic principles,” Phys. E: Low-dimensional Syst. Nanostructures 14(1-2), 11–17 (2002).
10. K. Yoshikawa, H. Kawasaki, W. Yoshida, T. Irie, K. Konishi, K. Nakano, T. Uto, D. Adachi, M. Kanematsu, H.
Uzu, and K. Yamamoto, “Silicon heterojunction solar cell with interdigitated back contacts for a photoconversion
efficiency over 26%,” Nat Energy 2(5), 17032 (2017).
11. UN General Assembly, “World Record Solar Cell with 44.7% Efficiency,” https://www.ise.fraunhofer.de/en/press-
media/press-releases/2013/world-record-solar-cell-with-44-7-efficiency.html.
12. M. A. Green, E. D. Dunlop, J. Hohl-Ebinger, M. Yoshita, N. Kopidakis, and X. Hao, “Solar cell efficiency tables
(version 58),” Prog Photovolt Res Appl. 29, 657-667 (2021).
13. A. E. Elmohlawy, V. F. Ochkov, and B. I. Kazandzhan, “Thermal performance analysis of a concentrated solar power
system (csp) integrated with natural gas combined cycle (ngcc) power plant,” Case Stud. Therm. Eng. 14, 100458
(2019).
14. W. C. Brown and R. H. George, “Rectification of microwave power,” IEEE Spectrum 1(10), 92–97 (1964).
 Research Article Vol. 30, No. 5 / 28 Feb 2022 / Optics Express 7033

15. W. C. Brown, “The history of the development of the rectenna,” in Rectenna session of the SPS microwave system
works, Lyndon B. Johnson Space Center, (1980), pp. 15–18.
16. R. L. Bailey, “A Proposed New Concept for a Solar-Energy Converter,” J. Eng. for Power 94(2), 73–77 (1972).
17. F. H. Alharbi and S. Kais, “Theoretical limits of photovoltaics efficiency and possible improvements by intuitive
approaches learned from photosynthesis and quantum coherence,” Renew. Sustain. Energy Rev. 43, 1073–1089
(2015).
18. A. D. Vos, “Detailed balance limit of the efficiency of tandem solar cells,” J. Phys. D: Appl. Phys. 13(5), 839–846
(1980).
19. R. Corkish, M. Green, and T. Puzzer, “Solar energy collection by antennas,” Sol. Energy 73(6), 395–401 (2002).
20. W. C. Brown, “The receiving antenna and micr0wave power rectification,” J. Microw. Power 5(4), 279–292 (1970).
21. M. Dragoman and M. Aldrigo, “Graphene rectenna for efficient energy harvesting at terahertz frequencies,” Appl.
Phys. Lett. 109(11), 113105 (2016).
22. P. Bharadwaj, B. Deutsch, and L. Novotny, “Optical antennas,” Adv. Opt. Photonics 1(3), 438–483 (2009).
23. L. Novotny, “Effective wavelength scaling for optical antennas,” Phys. Rev. Lett. 98(26), 266802 (2007).
24. A. Khalid, N. J. Pilgrim, G. M. Dunn, M. C. Holland, C. R. Stanley, I. G. Thayne, and D. R. S. Cumming, “A planar
gunn diode operating above 100 ghz,” IEEE Electron Device Lett. 28(10), 849–851 (2007).
25. D. Georgiadou, J. Semple, A. Sagade, H. Forstén, P. Rantakari, Y.-H. Lin, F. Alkhalil, A. Seitkhan, K. Loganathan, H.
Faber, and T. Anthopoulos, “100 ghz zinc oxide schottky diodes processed from solution on a wafer scale,” Nat.
Electron. 3(11), 718–725 (2020).
26. B. Eliasson, “Metal-insulator-metal diodes for solar energy conversion,” PhD thesis, Graduate School of University
of Colorado (2001).
27. B. J. Eliasson and G. Moddel, “Metal-oxide electron tunneling device for solar energy conversion,” US Patent
6534784 (18 March 2003).
28. G. Moddel and S. Grover, Rectenna Solar Cells (Springer, 2013).
29. C. Fumeaux, J. Alda, and G. Boreman, “Lithographic antennas at visible frequencies,” Opt. Lett. 24(22), 1629–1631
(1999).
30. NGI, National Graphene Institute, The University of Manchester, UK, “Discovery of graphene,” https://www.
graphene.manchester.ac.uk/learn/discovery-of-graphene/.
31. European Project 101006963, “GreEnergy,” https://www.greenergy-project.eu/.
32. Anandtech, “Intel’s Manufacturing Roadmap from 2019 to 2029: Back Porting, 7nm, 5nm, 3nm, 2nm, and 1.4 nm,”
https://www.anandtech.com/show/15217/intels-manufacturing-roadmap-from-2019-to-2029.
33. L. Banszerus, M. Schmitz, S. Engels, M. Goldsche, K. Watanabe, T. Taniguchi, B. Beschoten, and C. Stampfer,
“Ballistic transport exceeding 28 µm in cvd grown graphene,” Nano Lett. 16(2), 1387–1391 (2016).
34. Z. Zhu, S. Joshi, S. Grover, and G. Moddel, “Graphene geometric diodes for terahertz rectennas,” J. Phys. D: Appl.
Phys. 46(18), 185101 (2013).
35. S. G. Z. Zhu, S. Joshi, and G. Moddel, Geometric Diodes for Optical Rectennas (Springer NY, 2013), pp. 209–227.
36. Z. Ma and G. A. Vandenbosch, “Optimal solar energy harvesting efficiency of nano-rectenna systems,” Sol. Energy
88, 163–174 (2013).
37. A. M. Song, P. Omling, L. Samuelson, W. Seifert, I. Shorubalko, and H. Zirath, “Operation of ingaas/inp-based
ballistic rectifiers at room temperature and frequencies up to 50 ghz,” Jpn. J. Appl. Phys. 40(Part 2, No. 9A/B),
L909–L911 (2001).
38. G. Auton, J. Zhang, R. K. Kumar, H. Wang, X. Zhang, Q. Wang, E. Hill, and A. Song, “Graphene ballistic
nano-rectifier with very high responsivity,” Nat. Commun. 7(1), 11670 (2016).
39. V. H. Nguyen, D. C. Nguyen, S. Kumar, M. Kim, D. Kang, Y. Lee, N. Nasir, M. A. Rehman, T. P. A. Bach, J. Jung,
and Y. Seo, “Optimum design for the ballistic diode based on graphene field-effect transistors,” npj 2D Mater. Appl.
5(1), 89 (2021).
40. G. A. Vandenbosch and Z. Ma, “Upper bounds for the solar energy harvesting efficiency of nano-antennas,” Nano
Energy 1(3), 494–502 (2012).
41. H. Zhao, H. Gao, T. Cao, and B. Li, “Efficient full-spectrum utilization, reception and conversion of solar energy by
broad-band nanospiral antenna,” Opt. Express 26(2), A178–A191 (2018).
42. V. Agrawal and Y. Lo, “Mutual coupling in phased arrays of randomly spaced antennas,” IEEE Trans. Antennas
Propag. 20(3), 288–295 (1972).
43. B. A. Munk, “Periodic surface for large scan angles,” US Patent 3789404A, (16 October 1968.
44. B. A. Munk, Frequency Selective Surfaces, Theory and Design (John Wiley and Sons, 2000).
45. B. A. Munk, Finite Antenna Arrays and FSS (John Wiley and Sons, 2003).
46. H. A. Wheeler, “Simple relations derived from a phased array antenna made of an infinite current sheet,” IEEE Trans.
Antennas Propag. 13(4), 506–514 (1965).
47. A. D. Rakić, A. B. Djurišić, J. M. Elazar, and M. L. Majewski, “Optical properties of metallic films for vertical-cavity
optoelectronic devices,” Appl. Opt. 37(22), 5271–5283 (1998).
48. J. Kraus, Antennas (New York: McGraw-Hill, 1950).
49. S. A. Schelkunoff and H. T. Friis, Antenna Theory and Practice (John Wiley and Sons, 1952).
50. C. Gueymard, D. Myers, and K. Emery, “Proposed reference irradiance spectra for solar energy systems testing,” Sol.
Energy 73(6), 443–467 (2002).
 Research Article Vol. 30, No. 5 / 28 Feb 2022 / Optics Express 7034

51. C. A. Balanis, Antenna Theory, 3rd Ed. (John Wiley and Sons, 2005).
52. “Ansys HFSS,” https://www.ansys.com/it-it/products/electronics/ansys-hfss.
53. Comsol, Inc., “Comsol - Software for Multiphysics,” https://www.comsol.com/.
54. T. J. Seok, A. Jamshidi, M. Kim, S. Dhuey, A. Lakhani, H. Choo, P. J. Schuck, S. Cabrini, A. M. Schwartzberg, J.
Bokor, E. Yablonovitch, and M. C. Wu, “Radiation engineering of optical antennas for maximum field enhancement,”
Nano Lett. 11(7), 2606–2610 (2011).