PROCEEDINGS OF SPIE

SPIEDigitalLibrary.org/conference-proceedings-of-spie

Deep-learning enabled modeling tool

Meta-Magus for metadevice
optimization and design

Sadia Noureen, Sumbel Ijaz, Muhammad Zubair,

Muhammad Qasim Mehmood, Yehia Massoud

Sadia Noureen, Sumbel Ijaz, Muhammad Zubair, Muhammad Qasim

Mehmood, Yehia Massoud, "Deep-learning enabled modeling tool Meta-
Magus for metadevice optimization and design," Proc. SPIE 12322,
Nanophotonics, Micro/Nano Optics, and Plasmonics VIII, 1232207 (31
December 2022); doi: 10.1117/12.2643936

Event: SPIE/COS Photonics Asia, 2022, Online Only

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 04 Jan 2023 Terms of Use: https://www.spiedigitallibrary.org/terms-of-use

 Invited Paper

Deep-learning enabled modeling tool Meta-Magus for metadevice

optimization and design
Sadia Noureen1, Sumbel Ijaz1, Muhammad Zubair1, Muhammad Qasim Mehmood1* and Yehia
Massoud2
1
NanoTech Lab, Information Technology University of the Punjab, Ferozepur Road, Lahore 54600,
Pakistan
2
Innovative Technologies Laboratories (ITL), King Abdullah University of Science and Technology
(KAUST), Saudi Arabia

ABSTRACT

Over the past few years, Deep-learning (DL) based modelling solutions have been presented as an alternate to the time-
tedious and computationally draining conventional design and optimization procedure of metasurfaces. While designing a
phase-based transmission meta-device, such as meta-lenses and meta-holograms etc., the most crucial part is to optimize
its unit-cell to ensure maximum electromagnetic (EM) transmission amplitude and full phase coverage (0-2π). Most of the
DL-based solutions have resulted in accurate optimization to provide the desired transmission amplitude. But the abrupt
discontinuities in the phase response, makes it more challenging to map and predict the optimized structure for full phase
coverage. Here, we present a novel DL-based tool named as “Meta-Magus” to design transmission based metaholograms.
Meta-Magus consists of two parts: (i) unit-cell optimization, and (ii) phase mask generation. Here, the first component
takes target transmission amplitude, phase, material properties, and the wavelength aimed as input, process it via regression
based tandem neural network, and provide optimized unit-cell structural parameters as output. Target image whose
hologram is to be generated is fed to the second component which comprises of deep convolutional neural network to
generate the corresponding phase mask as output. A full-wave commercial simulator then maps the optimized unit-cell
onto the generated phase mask and generates the intended meta-hologram. Simulation results of the generated designs
exhibit perfect holography, and validates that the model yields excellent predictions of a complete metasurface design from
scratch within a matter of seconds.
Keywords: Metasurface, Deep-learning (DL), metaholograms, Meta-Magus, U-Net, tandem neural network.

1. INTRODUCTION
Metasurfaces, the most successful and flourishing type of artificially engineered 2D structures, are
comprised of periodic assemblies of nano-resonators exhibiting exceptional powers to control and
maneuver the incident Electromagnetic (EM) waves [1-13]. The geometric dimensions and the spatial
orientation of these nano-resonators generates strong resonances causing sudden phase changes,
providing full control over the phase, amplitude, polarization of the EM waves [14-21]. Due to these
unique features and compactness, Metasurfaces have emerged as the best alternative to the hulking
conventional optical components and a perfect on-chip integrable solution for Photonic Integrated IC’s
[22-26]. Thus they have completely transformed the optics and photonics industry by providing
immensely efficient, smart and compact, on-chip portable devices [27-34]. Owing to these capabilities
they provide many interesting applications such as absorbers [35-38], holograms [39-42], meta-
mirrors [43-45], OAM multiplexing [46], reflectors [47-49], 3D integrated circuits [50-54], on-chip
devices [55, 56], amplifiers [57], self-healing and non-diffracting beams generation [58-60].
Widespread use of these metasurface’s in practical applications is limited by their overpriced
fabrication along with their sluggish and tiresome design procedure. The fundamental dimensional
parameters and the orientation of a meta-atom, that serves as the building block of a metasurface,

Nanophotonics, Micro/Nano Optics, and Plasmonics VIII, edited by Zhiping Zhou,

Kazumi Wada, Limin Tong, Proc. of SPIE Vol. 12322, 1232207
© 2022 SPIE · 0277-786X · doi: 10.1117/12.2643936

Proc. of SPIE Vol. 12322 1232207-1

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 04 Jan 2023
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
 needs to be carefully optimized to achieved the desired response. Typically, the researchers use some
prior knowledge of the domain and perform long, time-inefficient and computationally draining
iterative EM simulations to search through a parameter pace and find optimum set of geometrical
parameters for a meta-atom. In contrast some evolutionary algorithms such as genetic algorithm,
particle swarm optimization (PSO) algorithm are also employed to solve this inverse design problem
of meta-atoms. But the performance of these techniques is significantly degraded as the structural
complexity increases. This entire process gets considerably more difficult in case of dielectric meta-
atoms because they provide far greater efficiencies by exciting both electric and magnetic dipoles,
quadrupoles and other higher order poles. Rigorous work and critical mapping of these resonances on
the meta-atoms physical and mechanical attributes is required to get the desired optical output
amplitude and phase. Some methodical models were put forth in this regard, but their accuracy is
compromised when the size of meta-atom becomes comparable with the operating wave length. Once
the meta-atom’s parameters are optimized, next step is the cautious spatial arrangement of the
optimized meta-atoms to adjust the individual phase of each meta-atom to get the desired phenomenon
from the full wave device simulation.
Recently, artificial intelligence based forward and inverse models have been suggested
learning to develop and enhance metasurface’s by identifying the delicate, non-linear correlations
between the geometry of their meta-atoms and the related EM response. These models have perfectly
handled output amplitude prediction and optimization, but the task of phase prediction is more
complicated to deal with. Due to the sudden discontinuous (0o to 180o phase shifts) in the phase
response, it is highly challenging to infer and optimise a nanostructure to give 360o of phase coverage.
This ineffective phase optimization hampers the ability to deploy phase-based meta-devices such as
lenses, holograms and structured beam generators etc. In the near past, cascaded closed loop neural
networks, generative adversarial networks and tandem neural networks were proposed for the
concurrent estimations of phase and amplitude of dielectric meta-atoms and their inverse design
development [61-63]. But, these approaches require a substantial quantity of training data and their
complexity rises directly with the rise in number of input/output geometrical parameters. Also most
of these approaches overlook the underlying physical aspects and material attributes of the meta-atom
under consideration. In conclusion, to infer the EM amplitude as well as phase response of various
dielectric nanostructures while simultaneously incorporating the physical properties of the material,
the geometrical features of the resonator, and the operating wavelength, we require an easy-to-use,
robust and time-efficient deep learning based forward model. Likewise, we also need robust inverse
design models to predict optimum set of geometrical parameters for any target response while
addressing all the aforementioned issues.
Here, we propose a new time-efficient design and optimization tool for transmission based meta-
devices, named as “Meta-Magus” catering all the challenges discussed above. Meta-Magus comprises
of two fundamental units (i) Meta-atom Optimization unit: This part incorporates de-noising tandem
neural network that consists of two neural networks acting as the mirror image of each other and works
on a dual learning strategy. First, a forward predicting network is trained using the original dataset
collected from software simulations providing extremely high accuracy while taking into account the
material properties and operating frequency. The trained forward network is then used to generate a
larger dataset which is further agitated by adding random noise. This dataset is then used to train the
inverse design neural network based on the predictions of the pre-trained forward neural network.
Thus the trained inverse design network yields the optimum set of geometrical parameters for the

Proc. of SPIE Vol. 12322 1232207-2

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 04 Jan 2023
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
 desired response and as per the material and wavelength specifications. These optimum parameters
are then passed on to the second part of the Meta-Magus i.e., (ii) Phase mask generation unit: it is
composed of U-Net convolutional network and takes the image whose hologram is targeted and
generates the full device phase mask for the corresponding meta-hologram. A full-wave commercial
simulator then maps the optimized unit-cell onto the generated phase mask and generates the intended
meta-hologram. Simulation results of the generated designs exhibit perfect holography, and validates
that the model yields excellent predictions of a complete metasurface design from scratch within a
matter of seconds.

2. METHODOLOGY
A block diagram of the proposed metasurface optimization and design development tool “Meta-
Magus” is shown in the figure 1. A prototype of the cylindrical nano-rod meta atoms under
consideration is also illustrated in this figure. Optical response of each cylindrical meta-atom depends
upon its geometrical dimensions such as its radius (R), height (H), Period (P) and its material
characteristics such as the refractive index (n) and extinction coefficient (k). The multipole resonances
created in the dielectric nano-rod will change if any of these dimensions or the material attributes
changes, which will significantly alter the output EM amplitude and phase. Therefore, the Meta-Magus
takes the target transmission amplitude and phase (for eight distinct radii within the range defined by
Rmin and Rmax), the material properties refractive index (n), extinction coefficient (k) and operating
wavelength (λ) as input, pass it to its first component i.e., the Tandem neural network (TNN) which is
further divided into two parts which are the forward mapping neural network and the inverse design
neural network. Tandemly training these two networks yields the optimized set of dimensions for the
corresponding meta-atom to achieve the target response. The second component of the Meta-Magus
is a convolutional neural network based U-Net that takes the image whose hologram is to be generated
as input and provides the corresponding phase mask for the complete metasurface with any numerical
calculations and iterative Fourier and inverse Fourier transforms. Mapping this phase mask to the
optimized set of dimensions of the nano-rod will yield the final hologram. Here we discuss the two
components or Meta-Magus, their architecture and their functionalities in detail.

Fig.1 Block Diagram of the Proposed Meta-Magus

Proc. of SPIE Vol. 12322 1232207-3

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 04 Jan 2023
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
 Tandem Neural Network:

Our main objective is to harness and generalize the complicated inverse relationship between
the transmitted spectrum of a metasurface and its physical and geometrical characteristics using a deep
neural network. However, due to the inverse design's non-uniqueness problem, training a deep neural
network is extremely difficult. More specifically, numerous distinct metasurface configurations can
provide the same EM field response resulting in some contradictory training samples. This will create
major convergence issues for the inverse neural networks because the training data set will ineluctably
include some instances with different output labels for the same input. To overcome this issue, we
demonstrate a tandem neural network (TNN) which is composed of two sub-networks i.e., the forward
mapping neural network (FMNN) and the inverse optimization neural network (IONN).
First the FMNN is trained using the dataset collected from commercial EM software simulations. For
dataset procurement, a glass substrate and a high indexed dielectric nano-rod; for instance,
hydrogenated amorphous silicon (a-Si:H), titanium dioxide (TiO2) and Gallium nitride (GaN) are
selected for the meta-atom due to their transparency window in the visible regime and substantially
high efficiencies. Since the intrinsic properties of a material are also a function of the operating
wavelength (λ), so λ must also be incorporated in the mapping network. For instance, we select three
visible wavelengths i.e., 404nm, 532nm and 633 nm. An ideal cylindrical meta-atom must provide
maximum possible transmission amplitude and full phase coverage (0 to 360o) when the radius is
increased from a minimum (Rmin) to maximum (Rmax) value. Therefore, we simulate dielectric nano-
cylinders (a-Si:h, TiO2 and GaN) with different combination of Rmin, Rmax, H and P at the selected
visible wavelengths. These simulations are performed in the commercial EM software with periodic
boundary conditions and circular incident light. For each simulation, all the input structural dimension
values are saved as the input of the forward mapping neural network (FMNN) and output of the inverse
optimization network (IONN). The operating wavelength and the material’s n and k are also saved as
the secondary input of both the networks. The output transmission amplitude and phase at 8 discrete
levels of the radius between Rmin to Rmax are saved as the output of the FMNN and input of the
IONN.

Forward mapping neural network (FMNN):

The unpredictable and complex relationship between the output electromagnetic response and
the geometrical as well as material properties of the sub-wavelength meta-atoms is mapped using the
proposed forward mapping neural network (FMNN). The FMNN concurrently includes all the
analytical dimensions, material attributes and the functional wavelength of the meta-atom under
consideration and predicts its transmission amplitude and phase for 8 equally distant discrete radii
between Rmin and Rmax. Primary input of the FMNN is a [4 × 1] vector defined as X FMNN = [Rmin,
Rmax, H, P]. Another input is added at the first hidden layer of the FMNN which is a [3 × 1] vector
defined as [n, k, λ]. Output of the FMNN is a [16 × 1] defined as YFMNN = [AR1, AR2, …, AR8, ΦR1, ΦR2
,…, ΦR8] that corresponds to the output transmission amplitude and phase at 8 discrete levels of the
radius between Rmin to Rmax vector. The hyper-parameters of the network are optimized following
its iterative training and testing using the mentioned inputs to minimize the network's MSE without
any convergence issues. The final architecture of the FMNN contains 10 hidden layers with Leaky
ReLU activation function at each layer. The number of neurons in these layers is given as: 4-50-200-

Proc. of SPIE Vol. 12322 1232207-4

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 04 Jan 2023
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
 500-600-700-600-500-200-50-16. Since we are dealing with a continuous output and a regression
based problem, so use mean squared error (MSE) to access the efficiency of our FMNN, given as:
1
𝑀𝑆𝐸𝐹𝑀𝑁𝑁 = ∑𝑁 ′ 2 ′ 2
𝑖=1[(𝐴𝑖 − 𝐴𝑖 ) + (Φ𝑖 − Φ𝑖 ) ] (1)
𝑁
Here, A and Φ are the ground truth transmission amplitude and phase achieved from simulations,
whereas A’ and Φ’are the FMNN anticipated transmission amplitude and phase, with N being the
total number of dataset samples.

Fig. 2. Tandem Inverse Neural Network

Inverse optimization neural network (IONN):

Fig. 2. Depicts the overall layout of the proposed Tandem neural network. First the FMNN
explained above is trained over the collected dataset of 2000 samples. The Pre-trained FMNN is then
used to generate a larger dataset of 5000 samples for the inverse optimization neural network since it
requires more training samples to converge. This pre-trained FMNN is tandemly connected with the
proposed inverse optimization (IONN) network. The FMNN is kept frozen and rest of the tandem
network is trained to predict the best suitable set of nano-rod’s structural parameters to achieve a
specific target response at a given operating wavelength. The architecture and hyper parameters if the
IONN are exactly the same as the FMNN, but the inputs and outputs are interchanged. The input of
the IONN is a [16 × 1] defined as XIONN = [AR1, AR2, …, AR8, ΦR1, ΦR2 ,…, ΦR8] that corresponds to
the target transmission amplitude and phase at 8 discrete levels of the radius between Rmin to Rmax

Proc. of SPIE Vol. 12322 1232207-5

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 04 Jan 2023
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
 vector. Another input is added at the first hidden layer of the IONN which is a [3 × 1] vector defined
as [n, k, λ]. Output of the IONN is a [4 × 1] vector defined as YIONN = [Rmin, Rmax, H, P]. While
training the IONN and the frozen pre-trained FMNN, dual loss is calculated, one at the intermediate
layer and one at the final output layer of the tandem assembly. This loss is mathematically described
as:
𝑀𝑆𝐸𝑇𝑁𝑁 = 𝐿𝑜𝑠𝑠𝐴Φ + 𝛼 × 𝐿𝑜𝑠𝑠𝐺 (2)
1 𝑁 ′ 2 ′ 2 1 𝑁
𝑀𝑆𝐸𝑇𝑁𝑁 = 𝑁 ∑𝑖=1[(𝐴𝑖 − 𝐴𝑖 ) + (Φ𝑖 − Φ𝑖 ) ] + 𝛼 × 𝑁 ∑𝑖=1(𝐺𝑖 − 𝐺 ′ 𝑖 )2 (3)

Here, LossG is calculated at the intermediate layer, G and G’ represents the set of ground-truth
geometrical parameter’s and predicted geometrical parameters respectively. Rest of all the terms are
the same as that for MSEFMNN.

Phase Mask Generation:

After the optimization of unit cell, next step towards the meta-hologram designing is the
complete phase mask generation for the image whose hologram is required. Conventionally the phase
mask for holograms is generated using Grechberg-Saxton (GS) iterative phase retrieval algorithm.
Spatial periodic distribution of optimized unit cells in the meta-device is dictated by the generated
phase-mask. It exhibits phase angle values from 0-2π, each of these value corresponds to a specific
radius within the optimized range between Rmin and Rmax. So, the nano-rods with different radii’s
are mapped on to the phase mask of the image whose hologram is to be generated. Thus we achieve
the final meta-hologram device which is then simulated to get the desired hologram in the far field.
Here we propose a convolutional neural network (CNN) to replace the iterative GS algorithm. This
CNN takes the image whose hologram is to be generated as the input and predicts the phase mask at
the output in fraction of seconds without any iterative mathematical calculations. The input of the
proposed CNN are 100 × 100 black and white images. For each of these images we have a matrix of
100 × 100 phase values (100 phase values corresponding to 100 pixels of the image) referred as the
“phase-mask”. To reduce the size of the output, we use a dimensionality reduction method ie.,
principle component analysis (PCA). PCA is a widely used technique to reduce the dimensionality of
larger datasets. Here, PCA is applied at the output dataset with 80 principle components. The resulting
100 × 80 matrix is then flattened and the size of pca converted output labels is now [8000 × 1]. Thus,
the output of CNN are [80,000 × 1] pca converted vector of phase masks. The CNN is trained using
1500 dataset samples. After training, the predicted outputs are subjected to inverse PCA transform to
get 100 × 100 predicted phase matrix. The architecture of the proposed CNN consisting multiple
convolutional layers, batch normalization layers, pooling layers, leaky ReLU activation layers, is
depicted in fig. 3. The features extracted by the convolutional layers are then passed onto the fully
connected (FC) layers, which map these features onto the pca applied labels and generates [8000 × 1]
vector at the output which is then converted into 100 × 100 phase matrix via inverse PCA transform.

Proc. of SPIE Vol. 12322 1232207-6

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 04 Jan 2023
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
 Fig. 3. Architecture of the proposed CNN for phase mask generation.

3. RESULTS AND DISCUSSION

FMNN
A forward mapping neural network is trained to minimize the MSEFMNN given by equation 1
and accurately predict the transmission amplitude and phase. Adam optimizer with a learning rate of
10-3 and 1000 epochs is used. The average test MSEFMNN achieved is 3.5 × 10-3. The test dataset shows
an MSE of less than 5.2 × 10-3 in over 95% of the cases, demonstrating the accuracy of the suggested
model. Fig. 4 shows the comparison of the predicted and the simulated (ground-truth) transmission
amplitude and phase. Three test example are chosen to be displayed from each belonging to a different
material and different wavelength to show the model’s accuracy for all the materials and wavelengths.
It is evident from fig. 4 that the predicted amplitude and phase values perfectly matches the simulated
values for a set of radii between Rmin and Rmax. The input geometrical parameters and the material
characteristics for each example is written in the middle in the fig. 4. in red color.

Proc. of SPIE Vol. 12322 1232207-7

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 04 Jan 2023
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
 Fig. 4. Comparison of the simulated and the FMNN predicted transmission amplitude and phase

IONN
A tandem arrangement of inverse optimization neural network and forward mapping neural network
is trained to minimize the MSETNN given by equation 2 and 3 and accurately predict the nano-rod
geometrical parameter. Adam optimizer with a learning rate of 10-3 and 1000 epochs is used. The
average test MSEIONN achieved is 4.3 × 10-3. Fig. 5 shows the comparison of the predicted and the
ground-truth geometrical parameters. Three test example are chosen to be displayed from each
belonging to a different material and different wavelength to show the model’s accuracy for all the
materials and wavelengths. It is evident from fig. 5 that the predicted amplitude and phase values
perfectly matches the simulated values for a set of radii between Rmin and Rmax. The material
characteristics and the operating wavelength for each example is written in the middle in the fig. 5. in
red color.

Fig. 5. Comparison of ground-truth and TNN predicted nano-rod geometrical parameters (nm)

Proc. of SPIE Vol. 12322 1232207-8

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 04 Jan 2023
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
 PHASE Mask prediction
The CNN depicted in fig. 3. Is trained with an Adam optimizer and a learning rates of 10-3.
Average test MSE achieved is 3.4 × 10-3. Fig. 6 shows comparison of the GS algorithm generated
ground-truth phase masks and the predicted phase mask (obtained by performing inverse PCA
transform on the CNN’s output). Left column of the fig. 6 shows the input image whose hologram is
desired. The comparison of the phase masks shows perfect coherence between the label and the
predicted phase values.

Fig. 6. Comparison of the GS algorithm generated ground-truth phase masks and the predicted phase mask (obtained by
performing inverse PCA transform on the CNN’s output)
After getting the optimized unit cell dimension’s and the predicted phase mask, a meta-hologram
device containing 100 × 100 nano-rods is designed in the FDTD Lummerical software, and full-wave
analysis is performed to get the desired hologram. Fig. 7 shows the comparison of the holograms
generated by GS algorithm phase mask and the CNN predicted phase mask. It can be clearly seen that
both the holograms are exactly alike and they reproduce the input image accurately in the metasurfaces
far field. Thus we can replace the conventional time consuming unit cell optimization as well as phase
mask generation’s parts with DL enabled models and automate the whole procedure via trained
networks.

Proc. of SPIE Vol. 12322 1232207-9

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 04 Jan 2023
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
 Fig. 7. Comparison of the holograms generated by GS algorithm phase mask and the CNN predicted phase mask.

4. CONCLUSION
A novel, DL-enabled tool named as “Meta-Magus” is demonstrated to speed up the conventional
design and optimization procedure of metasurfaces. Meta-Magus exhibit two constituent components
i.e., (i) unit-cell optimization and the (ii) phase mask generation components. The first component
consists of a tandem neural network (TNN) formed by the tandem arrangement of a pre-trained
forward mapping neural network (FMNN) and an inverse optimization neural network (IONN). The
FMNN is trained earlier to take all the structural and material characteristics of nano-rods as input and
predict the transmission amplitude as well as phase for a set of radii between R min and Rmax. The
average test MSE achieved for FMNN is 3.5 × 10-3. Tandemly training the FMNN and IONN results
in excellent predictions of the optimum set of geometrical parameters to achieve the target
electromagnetic response. Test MSE achieved for TNN assembly is 5.2 × 10-3. The second component,
which consists of a deep convolutional neural network, receives the target image whose hologram is
to be produced as input and produces the associated phase mask as an output. The dimensionality of
the ground truth phase masks is reduced using principle component analysis and then provided at the
output of the proposed CNN. Once the CNN is trained, it predicts the output with an MSE as low as
3.4 × 10-3, this output is passed through inverse PCA transform to achieve the predicted phase mask.
The optimized unit-cell is then mapped onto the created phase mask using a full-wave commercial
simulator, which creates the desired meta-hologram. A complete metasurface design may be created
from scratch in a matter of seconds, according to simulation findings of the developed designs, which
show perfect holography.

REFERENCES

[1] M. Q. Mehmood, J. Seong, M. A. Naveed et al., “Single‐Cell‐Driven Tri‐Channel Encryption Meta‐Displays,” Advanced
Science, 2203962 (2022).
[2] I. Javed, J. Kim, M. A. Naveed et al., “Broad-Band Polarization-Insensitive Metasurface Holography with a Single-Phase
Map,” ACS Applied Materials & Interfaces, (2022).
[3] M. A. Naveed, J. Kim, I. Javed et al., “Novel Spin‐Decoupling Strategy in Liquid Crystal‐Integrated Metasurfaces for
Interactive Metadisplays,” Advanced Optical Materials, 2200196 (2022).
[4] M. D. Shafqat, N. Mahmood, M. Zubair et al., “Highly Efficient Perfect Vortex Beams Generation Based on All-Dielectric
Metasurface for Ultraviolet Light,” Nanomaterials, 12(19), 3285 (2022).
[5] D. R. Smith, W. J. Padilla, D. Vier et al., “Composite medium with simultaneously negative permeability and permittivity,”
Physical review letters, 84(18), 4184 (2000).
[6] S. Ijaz, A. S. Rana, Z. Ahmad et al., “The Dawn of Metadevices: From Contemporary Designs to Exotic Applications,”
Advanced Devices & Instrumentation, 2022, (2022).

Proc. of SPIE Vol. 12322 1232207-10

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 04 Jan 2023
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
 [7] H. S. Khaliq, J. Kim, T. Naeem et al., “Broadband Chiro‐Optical Effects for Futuristic Meta‐Holographic Displays,” Advanced
Optical Materials, 2201175 (2022).
[8] A. Hosseini, H. Nejati, and Y. Massoud, “Design of a maximally flat optical low pass filter using plasmonic nanostrip
waveguides,” Optics Express, 15(23), 15280-15286 (2007).
[9] S. Eachempati, A. Nieuwoudt, A. Gayasen et al., "Assessing carbon nanotube bundle interconnect for future FPGA
architectures." 1-6.
[10] T. Ragheb, and Y. Massoud, "On the modeling of resistance in graphene nanoribbon (GNR) for future interconnect
applications." 593-597.
[11] M. Alam, and Y. Massoud, “RLC ladder model for scattering in single metallic nanoparticles,” IEEE transactions on
nanotechnology, 5(5), 491-498 (2006).
[12] Z. Duan, C. Guo, and M. Chen, “Enhanced reversed Cherenkov radiation in a waveguide with double-negative metamaterials,”
Optics Express, 19(15), 13825-13830 (2011).
[13] D. Lee, S. So, G. Hu et al., “Hyperbolic metamaterials: fusing artificial structures to natural 2D materials,” ELight, 2(1), 1-23
(2022).
[14] J. Valentine, J. Li, T. Zentgraf et al., “An optical cloak made of dielectrics,” Nature materials, 8(7), 568-571 (2009).
[15] A. Nieuwoudt, and Y. Massoud, “Evaluating the impact of resistance in carbon nanotube bundles for VLSI interconnect using
diameter-dependent modeling techniques,” IEEE Transactions on Electron Devices, 53(10), 2460-2466 (2006).
[16] A. Nieuwoudt, and Y. Massoud, “Understanding the impact of inductance in carbon nanotube bundles for VLSI interconnect
using scalable modeling techniques,” IEEE Transactions on Nanotechnology, 5(6), 758-765 (2006).
[17] A. Nieuwoudt, and Y. Massoud, “On the optimal design, performance, and reliability of future carbon nanotube-based
interconnect solutions,” IEEE transactions on electron devices, 55(8), 2097-2110 (2008).
[18] M. A. Naveed, J. Kim, M. A. Ansari et al., “Single-Step Fabricable Flexible Metadisplays for Sensitive Chemical/Biomedical
Packaging Security and Beyond,” ACS Applied Materials & Interfaces, (2022).
[19] A. Arbabi, and A. Faraon, “Fundamental limits of ultrathin metasurfaces,” Sci Rep, 7, 43722 (2017).
[20] A. Hosseini, A. Nieuwoudt, and Y. Massoud, “Efficient simulation of subwavelength plasmonic waveguides using implicitly
restarted Arnoldi,” Optics Express, 14(16), 7291-7298 (2006).
[21] S. Kirolos, Y. Massoud, and Y. Ismail, "Power-supply-variation-aware timing analysis of synchronous systems." 2418-2421.
[22] Y. Massoud, S. Majors, T. Bustami et al., "Layout techniques for minimizing on-chip interconnect self inductance." 566-571.
[23] Y. Massoud, J. Kawa, D. MacMillen et al., "Modeling and analysis of differential signaling for minimizing inductive cross-
talk." 804-809.
[24] Y. Massoud, and A. Nieuwoudt, “Modeling and design challenges and solutions for carbon nanotube-based interconnect in
future high performance integrated circuits,” ACM Journal on Emerging Technologies in Computing Systems (JETC), 2(3),
155-196 (2006).
[25] G. M. Akselrod, C. Argyropoulos, T. B. Hoang et al., “Probing the mechanisms of large Purcell enhancement in plasmonic
nanoantennas,” Nature Photonics, 8(11), 835-840 (2014).
[26] H. Altug, S.-H. Oh, S. A. Maier et al., “Advances and applications of nanophotonic biosensors,” Nature Nanotechnology,
17(1), 5-16 (2022).
[27] A. Jha, C. Huang, and P. R. Prucnal, “Reconfigurable all-optical nonlinear activation functions for neuromorphic photonics,”
Optics letters, 45(17), 4819-4822 (2020).
[28] A. Nieuwoudt, and Y. Massoud, “On the impact of process variations for carbon nanotube bundles for VLSI interconnect,”
IEEE Transactions on Electron Devices, 54(3), 446-455 (2007).
[29] Y. Massoud, S. Majors, J. Kawa et al., “Managing on-chip inductive effects,” IEEE Transactions on Very Large Scale
Integration (VLSI) Systems, 10(6), 789-798 (2002).
[30] M. Mondal, A. J. Ricketts, S. Kirolos et al., "Thermally robust clocking schemes for 3D integrated circuits." 1-6.
[31] A. Hosseini, T. Ragheb, and Y. Massoud, "A fault-aware dynamic routing algorithm for on-chip networks." 2653-2656.
[32] Y. Massoud, and J. White, “Simulation and modeling of the effect of substrate conductivity on coupling inductance and circuit
crosstalk,” IEEE Transactions on Very Large Scale Integration (VLSI) Systems, 10(3), 286-291 (2002).
[33] J. Han, and R. Chen, “Tunable broadband terahertz absorber based on a single-layer graphene metasurface,” Opt Express,
28(20), 30289-30298 (2020).
[34] X. Huang, W. He, F. Yang et al., “Polarization-independent and angle-insensitive broadband absorber with a target-patterned
graphene layer in the terahertz regime,” Opt Express, 26(20), 25558-25566 (2018).
[35] R. Bilal, M. Naveed, M. Baqir et al., “Design of a wideband terahertz metamaterial absorber based on Pythagorean-tree fractal
geometry,” Optical Materials Express, 10(12), 3007-3020 (2020).
[36] M. A. Abbas, J. Kim, A. S. Rana et al., “Nanostructured chromium-based broadband absorbers and emitters to realize
thermally stable solar thermophotovoltaic systems,” Nanoscale, 14(17), 6425-6436 (2022).
[37] S. Mehrabi, M. H. Rezaei, and A. Zarifkar, “Ultra-broadband solar absorber based on multi-layer TiN/TiO 2 structure with
near-unity absorption,” JOSA B, 36(9), 2602-2609 (2019).
[38] R. M. H. Bilal, M. A. Saeed, M. A. Naveed et al., “Nickel-Based High-Bandwidth Nanostructured Metamaterial Absorber for
Visible and Infrared Spectrum,” Nanomaterials, 12(19), 3356 (2022).

Proc. of SPIE Vol. 12322 1232207-11

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 04 Jan 2023
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
 [39] M. A. Naveed, M. A. Ansari, I. Kim et al., “Optical spin-symmetry breaking for high-efficiency directional helicity-
multiplexed metaholograms,” Microsystems & nanoengineering, 7(1), 1-9 (2021).
[40] B. Groever, W. T. Chen, and F. Capasso, “Meta-lens doublet in the visible region,” Nano letters, 17(8), 4902-4907 (2017).
[41] L. Guo, Z. Hu, R. Wan et al., “Design of aluminum nitride metalens for broadband ultraviolet incidence routing,”
Nanophotonics, 8(1), 171-180 (2019).
[42] Z. Ma, S. Dong, X. Dun et al., “Reconfigurable Metalens with Phase-Change Switching between Beam Acceleration and
Rotation for 3D Depth Imaging,” Micromachines, 13(4), 607 (2022).
[43] M. Pu, P. Chen, Y. Wang et al., “Anisotropic meta-mirror for achromatic electromagnetic polarization manipulation,” Applied
Physics Letters, 102(13), 131906 (2013).
[44] B. Ratni, A. de Lustrac, G.-P. Piau et al., “Reconfigurable meta-mirror for wavefronts control: applications to microwave
antennas,” Optics express, 26(3), 2613-2624 (2018).
[45] X. Xie, M. Pu, K. Liu et al., “High‐Efficiency and Tunable Circular‐Polarization Beam Splitting with a Liquid‐Filled All‐
Metallic Catenary Meta‐Mirror,” Advanced Materials Technologies, 4(7), 1900334 (2019).
[46] Z. Jin, D. Janoschka, J. Deng et al., “Phyllotaxis-inspired nanosieves with multiplexed orbital angular momentum,” ELight,
1(1), 1-11 (2021).
[47] A. Hosseini, and Y. Massoud, “A low-loss metal-insulator-metal plasmonic bragg reflector,” Optics express, 14(23), 11318-
11323 (2006).
[48] A. Hosseini, H. Nejati, and Y. Massoud, “Modeling and design methodology for metal-insulator-metal plasmonic Bragg
reflectors,” Optics express, 16(3), 1475-1480 (2008).
[49] A. Hosseini, and Y. Massoud, “Nanoscale surface plasmon based resonator using rectangular geometry,” Applied Physics
Letters, 90(18), 181102 (2007).
[50] Y. Massoud, and J. White, "Improving the generality of the fictitious magnetic charge approach to computing inductances in
the presence of permeable materials." 552-555.
[51] Y. Massoud, and J. White, "FastMag: a 3-D magnetostatic inductance extraction program for structures with permeable
materials." 478-484.
[52] Y. Massoud, M. Alam, and A. Nieuwoudt, "On the selection of spectral zeros for generating passive reduced order models."
160-164.
[53] A. Nieuwoudt, and Y. Massoud, “Predicting the performance of low-loss on-chip inductors realized using carbon nanotube
bundles,” IEEE Transactions on Electron Devices, 55(1), 298-312 (2007).
[54] M. Alam, and Y. Massoud, “A closed-form analytical model for single nanoshells,” IEEE transactions on nanotechnology,
5(3), 265-272 (2006).
[55] A. Nieuwoudt, and Y. Massoud, "Assessing the implications of process variations on future carbon nanotube bundle
interconnect solutions." 119-126.
[56] A. Nieuwoudt, and Y. Massoud, “Performance implications of inductive effects for carbon-nanotube bundle interconnect,”
IEEE electron device letters, 28(4), 305-307 (2007).
[57] C.-H. Tseng, and C.-L. Chang, “Improvement of return loss bandwidth of balanced amplifier using metamaterial-based
quadrature power splitters,” IEEE Microwave and Wireless Components Letters, 18(4), 269-271 (2008).
[58] L. Huang, X. Song, B. Reineke et al., “Volumetric generation of optical vortices with metasurfaces,” Acs Photonics, 4(2),
338-346 (2017).
[59] F. Z. Shu, F. F. Yu, R. W. Peng et al., “Dynamic plasmonic color generation based on phase transition of vanadium dioxide,”
Advanced Optical Materials, 6(7), 1700939 (2018).
[60] Z. Zhang, T. Li, X. Jiao et al., “High-efficiency all-dielectric metasurfaces for the generation and detection of focused optical
vortex for the ultraviolet domain,” Applied Sciences, 10(16), 5716 (2020).
[61] S. Noureen, M. Q. Mehmood, M. Ali et al., “A unique physics-inspired deep-learning-based platform introducing a
generalized tool for rapid optical-response prediction and parametric-optimization for all-dielectric metasurfaces,” Nanoscale,
(2022).
[62] I. Tanriover, W. Hadibrata, and K. Aydin, “Physics-based approach for a neural networks enabled design of all-dielectric
metasurfaces,” ACS Photonics, 7(8), 1957-1964 (2020).
[63] S. So, and J. Rho, “Designing nanophotonic structures using conditional deep convolutional generative adversarial networks,”
Nanophotonics, 8(7), 1255-1261 (2019).

Proc. of SPIE Vol. 12322 1232207-12

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 04 Jan 2023
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use