Article

Cite This: ACS Photonics 2018, 5, 3132−3140 pubs.acs.org/journal/apchd5

Full-Stokes Imaging Polarimetry Using Dielectric Metasurfaces

Ehsan Arbabi,† Seyedeh Mahsa Kamali,† Amir Arbabi,‡ and Andrei Faraon*,†
†
T. J. Watson Laboratory of Applied Physics and Kavli Nanoscience Institute, California Institute of Technology, 1200 E. California
Boulevard, Pasadena, California 91125, United States
‡
Department of Electrical and Computer Engineering, University of Massachusetts Amherst, 151 Holdsworth Way, Amherst,
Massachusetts 01003, United States
*
S Supporting Information
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

ABSTRACT: Polarization is a degree of freedom of light carrying important

information that is usually absent in intensity and spectral content. Imaging
polarimetry is the process of determining the polarization state of light, either
partially or fully, over an extended scene. It has found several applications in various
Downloaded via UNIV LAVAL on November 11, 2024 at 19:32:08 (UTC).

fields, from remote sensing to biology. Among different devices for imaging
polarimetry, division of focal plane polarization cameras (DoFP-PCs) are more
compact, less complicated, and less expensive. In general, DoFP-PCs are based on
an array of polarization filters in the focal plane. Here we demonstrate a new
principle and design for DoFP-PCs based on dielectric metasurfaces with the ability
to control polarization and phase. Instead of polarization filtering, the method is
based on splitting and focusing light in three different polarization bases. Therefore,
it enables full Stokes characterization of the state of polarization and overcomes the
50% theoretical efficiency limit of the polarization-filter-based DoFP-PCs.
KEYWORDS: dielectric metasurface, nanophotonics, polarization and phase control, division of focal plane, polarization camera

P olarimetric imaging is the measurement of the polarization

state of light over a scene of interest. While spectral and
hyperspectral imaging techniques provide information about
degree of circular polarization, form-birefringent quarter
waveplates were integrated with linear polarizers in the mid-
IR.23 Recently, liquid crystal retarders have been integrated
the molecular and material composition of a scene,1,2 with linear polarization filters to enable full Stokes polarimetric
polarimetric imaging contains information about the shape imaging by implementing circular24 and elliptical polarization
and texture of reflecting surfaces, the orientation of light filters.25,26 An issue with the previously demonstrated DoFP-
emitters, or the optical activity of various materials.3,4 This PCs is that they all have a theoretical efficiency limit of 50%
additional information has led to many applications for because they use polarization filters3 or spatially block half of
imaging polarimetry ranging from astronomy and remote the aperture.20
sensing to marine biology and medicine.3,5−11 Therefore, Optical metasurfaces are a category of nanofabricated
several methods have been developed over the past decades to diffractive optical elements composed of nanoscatterers on a
enable mapping of the polarization state over an extended surface27−45 that are judiciously designed to control optical
scene.11−18 wavefronts. They have enabled high-efficiency phase and
Generally, polarimetric imaging techniques can be catego- polarization control with large gradients.31,32,37,38,42,46 In
rized into three groups: division of amplitude, division of addition, their compatibility with conventional microfabrica-
aperture, and division of focal plane.3 All of these techniques tion techniques allows for their integration into optical
are based on measuring the intensity in different polarization metasystems.47−50
bases and using them to estimate the full Stokes vector or a
Metasurfaces have previously been used for polarime-
part of it. Division of focal plane polarization cameras (DoFP-
try,51−59 but not polarimetric imaging. An important capability
PCs) are less expensive, more compact, and require less
of high-contrast dielectric metasurfaces is the simultaneous
complicated optics compared to the two other categories of
polarimetric imaging systems.3,16−18 In addition, they require control of polarization and phase.46 Here, we use this capability
much less effort for registering images of different polar- to demonstrate a dielectric metasurface mask for DoFP-PCs
izations, as the registration is automatically achieved in the with the ability to fully measure the Stokes parameters,
fabrication of the polarization-sensitive image sensors. The including the degree of circular polarization and helicity. Since
advances in micro/nanofabrication have increased the quality the mask operates based on polarization splitting and focusing
of DoFP-PCs and reduced their fabrication costs, making them instead of polarization filtering, it overcomes both the 50%
commercially available. DoFP-PCs use either a birefringent
crystal to split polarizations19,20 or thin-film17,21 or wire- Received: March 21, 2018
grid11,16,22 polarization filters. To enable the measurement of Published: July 16, 2018

© 2018 American Chemical Society 3132 DOI: 10.1021/acsphotonics.8b00362

ACS Photonics 2018, 5, 3132−3140
ACS Photonics Article

Figure 1. Concept of a metasurface polarization camera. (a) Top: Schematics of a conventional setup used for polarimetry: a waveplate (quarter or
half) followed by a Wollaston prism and a lens that focuses light on detectors. Bottom: A compact metasurface implements the functionality of all
three components combined and can be directly integrated on an image sensor. WP: waveplate; PD: photodetector. (b) Possible arrangement for a
superpixel of the polarization camera, comprising six image sensor pixels. Three independent polarization bases (H/V, ±45°, and RHCP/LHCP)
are chosen to measure the Stokes parameters at each superpixel. (c) Three-dimensional illustration of a superpixel focusing different polarizations
to different spots. The colors are used only for clarity of the image and bear no wavelength information.

Figure 2. Meta-atom and pixel design. (a) An α-Si nanopost with a rectangular cross section resting on a glass substrate provides full polarization
and phase control. (b) Design graphs used for finding the in-plane dimensions of a nanopost. Given a pair of transmission phases ϕx and ϕy, one
can find the corresponding nanopost dimensions a and b from the two graphs. The nanoposts are 650 nm tall, and the lattice constant is 480 nm.
(c) Schematic illustration of a rotated nanopost, showing the rotation angle and the old and new optical axis sets. (d) Required phase profiles for a
metasurface that does both polarization beam splitting and focusing at two orthogonal polarizations. These can be any set of orthogonal
polarizations, linear or elliptical. The focal distance for these phase profiles is 9.6 μm, equal to the width of the superpixel in the x direction. The
lateral positions of the focal spots are x = ±2.4 μm and y = 0. (e) Scanning electron micrograph of a fabricated superpixel. The polarization basis for
each part is shown with the colored arrows. Scale bar: 1 μm.

theoretical efficiency limit and the one-pixel registration error imaged a custom polarization mask that we fabricated. To the
(resulting from distinct physical areas of the polarization best of our knowledge, this is the first demonstration of a
filters) of the previously demonstrated DoFP-PCs.3 In DoFP-PC mask that measures the polarization state completely
addition, unlike the previously demonstrated full Stokes and is not based on polarization filtering.
DoFP-PCs, the metasurface is fabricated in a single dielectric There are several representations for polarization of light.60
layer and does not require integration of multiple layers Among them, the Stokes vector formalism has some
operating as retarders and polarization filters. The mask is conceptual and experimental advantages since it can be used
designed for an 850 nm center wavelength. The polarization to represent light with various degrees of polarization and can
cross-talk ranges from 10% to 15% for pixel sizes from 7.2 to be directly determined by measuring the power in certain
2.4 μm when using an 850 nm LED as the light source. In polarization bases.60 Therefore, most imaging polarimetry
addition, to demonstrate that the metasurface DoFP-PC can be systems determine the Stokes vector S = [S0, S1, S2, S3],3 where
used to form polarization images over extended scenes, we the components are defined as S0 = I, S1 = Ix − Iy, S2 = I45 −
3133 DOI: 10.1021/acsphotonics.8b00362
ACS Photonics 2018, 5, 3132−3140
ACS Photonics Article

I−45, and S3 = IR − IL. Here I is the total intensity and Ix, Iy, I45, independent control of orthogonal polarizations can be
and I−45 are the intensity of light in linear polarization bases generalized to elliptical and circular polarizations as well
along the x, y, +45-degree, and −45-degree directions, (with the small drawback that the output and input
respectively. IR and IL denote the intensities of the right- polarizations should have opposite handedness). To make
hand and left-hand circularly polarized light. Usually, S1, S2, the design process clear, here we reiterate the results presented
and S3 are normalized to S0 such that their value changes in the Supporting Information of ref 46.
between −1 and +1. To fully characterize the state of The operation of a nanopost can be modeled by a Jones
polarization, all of these intensities should be determined. A matrix relating the input and output electric fields (i.e., Eout =
conventional setup used to measure the full Stokes vector is TEin). For the rotated nanopost shown in Figure 2c, the Jones

ÄÅ É ÄÅ ÉÑ
ÅÅTxx Txy ÑÑÑ
matrix can be written as

ÅÅÅ eiϕx′ 0 ÑÑÑ

shown in Figure 1a: a waveplate (half or quarter), followed by

ÅÅ ÑÑ
T = ÅÅÅ ÑÑ = R(θ )ÅÅÅ ÑÑR( −θ )
a Wollaston prism and a lens that focuses the beams on

ÅÅTyx Tyy ÑÑÑ ÅÅ ÑÑ

ÅÇ 0 eiϕy′ ÑÑÖ
photodetectors. One can determine the four Stokes parame-
ÅÇ ÑÖ
ters60 from the detector signals acquired under three different
conditions: without a waveplate, with a half-waveplate (HWP) (1)
inserted, and with a quarter-waveplate (QWP) inserted in the where R(θ) denotes the rotation matrix by an angle θ in the
setup. An optical metasurface with the ability to fully control counterclockwise direction. Here we have assumed unity
phase and polarization of light46 can perform the same task transmission since the nanoposts are highly transmissive. We
over a much smaller volume and without changing any optical note here that the right-hand side of eq 1 is a unitary and
components. The metasurface can split any two orthogonal symmetric matrix. Using only these two conditions (i.e.,
states of polarization and simultaneously focus them to unitarity and symmetry), we find Txy = Tyx,
different points with high efficiency and on a micrometer-
|Tyx| = 1 − |Txx|2 , and Tyy = −exp(i2∠Tyx)Txx. As one
scale. This is schematically shown in Figure 1a. Such a
metasurface can be directly integrated on an image sensor for would expect, these equations restrict the available number
making a polarization camera. To fully measure the Stokes of controllable parameters to three (|Txx|, ∠Txx, ∠Tyx),
parameters, the projection of the input light on three different corresponding to the three available physical parameters (ϕx′,
polarization basis sets should be measured. A typical choice of ϕy′, and θ). Using these relations to simplify Eout = TEin, we
basis is horizontal/vertical (H/V), ±45° linear, and right-hand- can rewrite it to find the Jones matrix elements in terms of the

ÄÅ É
ÅÅ E out * E out * ÑÑÑÄÅÅT ÉÑÑ ÅÄÅÅ in * ÑÉÑÑ
circular/left-hand-circular (RHCP/LHCP) that can be used to input and output fields:

ÅÅ x ÑÑÅÅ xx ÑÑ ÅÅ Ex ÑÑ
ÅÅ ÑÑÅÅ ÑÑ = ÅÅ ÑÑÑ
directly measure the Stokes parameters. Figure 1b shows a
ÅÅ in in Ñ
ÑÅÅ ÑÑ Å
ÅÅ E Ey ÑÑÑÅÅÇTyx ÑÑÖ ÅÅÅÅÇ Exout ÑÑÑÑÖ
possible configuration where the three metasurface polarization y

ÅÅÇ x ÑÖ
beam splitters (PBSs) are multiplexed to make a superpixel,
comprising six pixels on the image sensor. Each image sensor (2)
pixel can then be used to measure the power in a single
where * denotes complex conjugation. Equation 2 is
polarization state. A schematic illustration of a superpixel is
important, as it shows how one can find the Jones matrix
shown in Figure 1c. The colors are only used to distinguish required to transform any input field with a given phase and
different parts of the superpixel more easily and do not polarization to any desired output field with a different phase
correspond to actual wavelengths. The blue nanoposts separate and polarization. This is the first application of the birefringent
and focus RHCP/LHCP, the green ones separate ±45°, and meta-atoms, i.e., complete and independent polarization and
the red ones separate H/V. phase control.
The metasurface platform we use here is based on dielectric The Jones matrix is uniquely determined by eq 2, unless the
birefringent nanopost structures.46 As seen in Figure 2a, the determinant of the coefficients matrix on the left-hand side is
metasurface is composed of α-Si nanoposts with rectangular zero. In this case, the matrix rows (i.e., Eout* and Ein) will be
cross sections on a low-index fused silica substrate. With a proportional. Since the Jones matrix is unitary (i.e., the input
proper choice of the α-Si layer thickness and lattice constant and output powers are equal), the proportionality coefficient
(650 and 480 nm, respectively, for an operating wavelength of must have unit amplitude: Eout* = exp(iϕ)Ein, which means
850 nm), the nanoposts can provide full and independent 2π that Eout and Ein have the same polarization ellipse, but an
phase control over x- and y-polarized light, where x and y are opposite handedness. The condition Eout* = exp(iϕ)Ein does
aligned with the axes of the nanopost (see Supporting not uniquely determine the Jones matrix. To uniquely
Information Figure S1).61 Using the phase versus dimension determine the Jones matrix, two conditions for two different
graphs, one can calculate the nanopost dimensions required to sets of input/output fields are necessary and can be chosen as
provide a specific pair of phase values, ϕx and ϕy, as shown in
1 * = exp(iϕ1)E1 and E2 * = exp(iϕ2)E2 . Here we are using
Eout in out in
Figure 2b. This allows for designing a metasurface that controls the numeral subscripts to distinguish between the two input/
x- and y-polarized light independently. With a simple output field sets. If ϕ1 and ϕ2 can be independently controlled,
generalization, the same can be applied to any two orthogonal then Ein1 and Ein2 (as well as Eout out
1 and E2 ) should be orthogonal
linear polarizations using nanoposts that are rotated around to each other from an energy conservation argument. We can

ÄÅ in ÉÑÄ É ÄÅ É ÄÅÅ ÉÑ
their optical axis with the correct angle to match the new linear
out Ñ
write the final equation as
ÅÅ E in Ñ Å Ñ in * Ñ
ÅÅ 1, x E1, y ÑÑÑÅÅÅTxx ÑÑÑ ÅÅÅÅ E1, x ÑÑÑÑ ÅÅÅ exp(iϕ1)E1, x ÑÑÑ
polarizations (e.g., the x′−y′ axis in Figure 2c). An important

ÅÅ ÑÑÅÅ ÑÑ = ÅÅ Å ÑÑ
ÅÅ in in Ñ
ÑÅÅT ÑÑ ÅÅ out ÑÑÑÑ = ÅÅÅ in * Ñ
Ñ
and interesting point demonstrated in ref 46 is that this can be

ÅÅ E Ñ Å Ñ Å Ñ Å Ñ
ÅÅÇ 2, x E2, y ÑÑÑÖÅÇ ÑÖ ÅÅÇ 2, x ÑÑÖ ÅÅÅÇ exp(iϕ2)E2, x ÑÑÑÖ
done in a point-by-point manner, where the polarization basis
is different for each nanopost. This property allows us to easily yx E (3)
design the metasurface PBS for the two linear bases of interest.
Moreover, as demonstrated in ref 46, an even more interesting These equations are the basis for the second important
property of this seemingly simple structure is that the application of the method: given any two orthogonal input
3134 DOI: 10.1021/acsphotonics.8b00362
ACS Photonics 2018, 5, 3132−3140
ACS Photonics Article

Figure 3. Characterization results of the superpixels of the DoFP metasurface mask. (a) Measured average Stokes parameters for different input
polarizations (shown with colored arrows) and (b) the corresponding intensity distributions for a sample superpixel. The Stokes parameters are
averaged over about 120 superpixels (limited by the microscope field of view), and the error bars represent the statistical standard deviations. (c)
Measured average Stokes parameters and the corresponding intensity distribution for the LED light source without any polarization filters in the
setup. All the measurements are performed with an 850 nm LED filtered by a bandpass filter (center: 850 nm, fwhm: 10 nm) as the light source.

polarizations (denoted by Ein1 and Ein2 ), their phase can be pixel size of 4.8 μm, the calculated phase profiles are shown in
independently controlled using the Jones matrix given by eq 3. Figure 2d, where the focal distance is assumed to be 9.6 μm.
For instance, Arbabi et al.46 demonstrated a metasurface that Since each polarization basis covers two image sensor pixels,
focuses RHCP input light to a tight spot and LHCP input light the phases are defined over the area of two pixels. In addition,
to a doughnut shape. The cost is that the output orthogonal the calculated phases are the same for the three different
polarizations have the opposite handedness compared to the polarization bases, and therefore only one basis is shown in
input ones. Once the Jones matrix is calculated from eq 3 (or Figure 2d. Using these phases and knowing the desired
eq 2, depending on the desired type of transformation), the polarization basis at each point, we calculated the rotation
two phases, ϕx′ and ϕy′, and the rotation angle θ can be angles and nanopost dimensions from eqs 3 and 1 along with
calculated from eq 1. Let us emphasize here that since this is a the data shown in Figure 2b.
point-by-point design, all the steps can be repeated The metasurface mask was then fabricated in a process
independently for each nanopost, meaning that the polar- similar to ref 62. A 650 nm thick layer of α-Si was deposited on
ization basis can be changed from one nanopost to the next. a fused silica wafer. The metasurface pattern was defined using
Based on the concept and technique just described, the first electron-beam lithography and transferred to the α-Si layer
design step is identifying the input polarizations at each point. through a lift-off process used to form a hard etch mask,
For the DoFP-PC, three different sets of H/V, ±45°, and followed by dry etching. Figure 2e shows a scanning electron
RHCP/LHCP are chosen, corresponding to the three distinct micrograph of a fabricated superpixel, with the polarization
areas in the superpixel shown in Figure 1b. Then, the required bases denoted by arrows for each section. In addition to the
phase profiles are determined to each split two orthogonal metasurface mask corresponding to a pixel size of 4.8 μm that
polarizations and focus them to the centers of adjacent pixels is mentioned above and shown in Figure 2e, two other masks
on the image sensor, as shown schematically in Figure 1c. For a with pixel sizes of 7.2 and 2.4 μm were also fabricated with
3135 DOI: 10.1021/acsphotonics.8b00362
ACS Photonics 2018, 5, 3132−3140
ACS Photonics Article

Figure 4. Polarimetric imaging. (a) Schematic illustration of target polarization ellipse in different parts of the polarization sample. Stokes
parameters of the polarization sample: (b) the targeted polarization mask, (c) the fabricated mask imaged using conventional polarimetry, and (d)
the same mask imaged using the metasurface polarimetric camera. The scale bars denote 100 μm in the metasurface polarization camera mask
plane.

focal distances of 14.4 and 4.8 μm, respectively. We used these repeated the same measurements with the polarization filter
to study the effect of pixel size on performance. removed from the setup. Figure 3c summarizes the results of
To characterize the metasurface mask, we illuminated it with this measurement that determines the polarization state of light
light from an 850 nm LED filtered by a 10 nm bandpass filter emitted by the LED. The data given in Figure 3a are used to
with different states of polarization. We then imaged the plane estimate the calibration matrix. As expected, the emitted light
corresponding to the image sensor location using a custom- has a low degree of polarization (<0.08). We also characterized
built microscope (see Supporting Information Figure S2 for the polarization state of the emitted LED light using a QWP
measurement details and the setup). Figure 3 summarizes the and a linear polarizer and found the degree of polarization to
superpixel characterization results for the 4.8 μm pixel design. be equal to zero up to the measurement error.
The measured Stokes parameters are plotted in Figure 3a for In addition, we measured the transmission efficiency of the
different input polarizations, showing a <10% cross-talk metasurface mask and found it to be in the range of 60% to
between polarizations and high similarity between different 65% for all pixel size designs and input polarizations. The
superpixels. The measurements were averaged over the field of lower than expected transmission is mainly due to a few
view of the microscope that corresponds to about 120 factors. First, the metasurface has a maximum deflection angle
superpixels. The standard deviations are shown in the graph larger than 50°, which results in lower transmission
as error bars. In addition, the intensity distribution over a efficiency.37,63 Second, the relatively large metasurface lattice
sample superpixel area is shown in Figure 3b. The graphs show constant of 480 nm does not satisfy the Nyquist sampling
the clear ability of the metasurface mask to route light as theorem for the large-deflection-angle transmission masks
desired for various input polarizations. Similar characterization inside the fused silica substrate.64 This results in spurious
results without a bandpass filter, corresponding to a bandwidth diffraction of light inside the substrate. Finally, the mask is
of about 5%, are presented in Supporting Information Figure periodic with a larger-than-wavelength period equal to the
S3. Slight performance degradation is observed with a superpixel dimensions. This results in excitation of the higher
maximum cross-talk of ∼13% since the metasurface efficiency diffraction orders especially inside the substrate that has a
decreases with changing the wavelength. In addition, similar higher refractive index. It is worth noting that the achieved
measurement results for metasurface masks with pixel sizes of ∼65% efficiency is higher than the 50% theoretical limit of a
7.2 and 2.4 μm are presented in Supporting Information polarimetric camera that is based on polarization filtering.
Figures S4 and S5, respectively. The results show a degradation Finally, we show polarimetric imaging using the DoFP
of performance with reducing the pixel size as the cross-talk is metasurface mask. We designed and used a custom dielectric
smaller than 7.5% and 13% for 7.2 μm and 2.4 μm pixels, metasurface polarization mask as the imaging target. We
respectively. To show the ability of the metasurface mask to utilized the polarization-phase control method described above
characterize the polarization state of unpolarized light, we and a fabrication process similar to the DoFP metasurface
3136 DOI: 10.1021/acsphotonics.8b00362
ACS Photonics 2018, 5, 3132−3140
ACS Photonics Article

mask for the imaging target. The mask converts x-polarized performance achieved from the 2.4 μm pixel in this work can
input light to an output polarization state characterized by the be achieved from a ∼1.7 μm pixel if the material between the
polarization ellipses and the Stokes parameters shown in mask and the image sensor has a refractive index of 1.5, which
Figure 4a and b, respectively. Each Stokes parameter is +1 or is the case when the DoFP mask is separated from the image
−1 in an area of the image corresponding to the specific sensor by an oxide or polymer layer, as would be the case in a
polarization. For instance, S3 is +1 in the right half circle, −1 in realistic device. To achieve smaller pixel sizes, better
the left half circle, and 0 elsewhere. Using a second custom- performance, and larger operation bandwidths, one could use
built microscope, the image of the polarization mask was more advanced optimization65 or chromatic-dispersion control
projected on the DoFP metasurface mask (see Supporting techniques,66 especially since the size of a single superpixel is
Information Figure S2 for the measurement setup and the small and allows for a fast simulation of the forward problem.
details). First, we removed the metasurface mask and In addition, a spatial multiplexing scheme67−70 can be used to
performed a conventional polarimetric imaging of the interleave multiple superpixels corresponding to different
projected image using a linear polarizer and a QWP. To this optical bands and therefore make a color-polarization camera.
end, six different images were captured with different rotations Using the polarization-phase control method and the
of the linear polarizer and the QWP to determine the Stokes platform introduced in ref 46, we demonstrated a metasurface
parameters. The results are shown in Figure 4c. Second, we mask for DoFP-PCs. The mask is designed to split and focus
removed the linear polarizer and the QWP and inserted the light to six different pixels on an image sensor for three
DoFP metasurface mask. The Stokes parameters were different polarization bases. This allows for complete character-
extracted from a single image captured at the focal plane of ization of polarization by measuring the four Stokes parameters
the DoFP metasurface mask. The results are shown in Figure over the area of each superpixel, which corresponds to the area
4d and are in good agreement with the results of regular of six pixels on an image sensor. We experimentally
polarimetric imaging. The lower quality of the metasurface demonstrated the ability of the metasurface masks to correctly
polarimetric camera image is mainly due to the limited number measure the polarization state for different input polarizations.
of superpixels that fit inside the single field of view of the In addition, we used the DoFP metasurface mask to form an
microscope. The field of view is limited by the microscope image of a complicated polarization object, showing the ability
magnification and image sensor size, which are ×22 and ∼15 to make a polarization camera. Many of the limitations faced
mm, respectively. This results in a low resolution of 70-by-46 here can be overcome using more advanced optimization
points for the metasurface polarimetric image versus the techniques or better data extraction methods. We anticipate
∼2000-by-2000 point resolution of the regular polarimetric that polarization cameras based on metasurface masks will be
image. In addition, to form the final image, we need to know able to replace the conventional polarization cameras for many
the coordinates of each superpixel a priori. The existing errors applications, as they enable measurement of the full polar-
in estimating these coordinates that result from small tilts in ization state including the degree of circular polarization and
the setup, aberrations of the custom-built microscope, etc., handedness.
cause a degraded performance over some superpixels. In a
polarization camera made using the DoFP metasurface mask,
both of these issues will be resolved, as the resolution can be
■ METHODS
Simulation and Design. To design the DoFP metasurface
much higher and the mask and image sensor are litho- mask, we first calculated the two phase profiles required for the
graphically aligned. two polarization states (Figure 2d). The phase profiles
To extract the polarization information on the image, we correspond to decentered aspheric lenses that focus each
integrated the intensity inside the area of two adjacent image polarization at the center of one image sensor pixel. These
sensor pixels. Then, we calculated the corresponding Stokes phases are then used in eq 3 along with the known input
parameter simply by dividing the difference between the two polarization states to calculate the Jones matrix. To find the
measured intensities by their sum. While straightforward, this nanopost corresponding to each Jones matrix, the matrix is
is not the optimal method to perform this task, as there is non- diagonalized according to eq 1, and the two phases, ϕx′ and ϕy′,
negligible cross-talk between different polarization intensities and the rotation angle θ are then extracted. The dimensions of
measured by the pixels (Figure 3). The issue becomes more the nanopost providing the required pair of phases is then
pronounced moving toward smaller pixel sizes, as seen for the found using the data in Figure 2b.
2.4 μm case (Supporting Information Figure S5). To address The polarization target used for the imaging experiments in
this issue, a better polarization extraction method is to form a Figure 4 was designed in a slightly different manner since in
calibration matrix that relates the actual intensities to the this case only the output polarization is of interest. Assuming
corresponding measured values for a specific DoFP metasur- an x-polarized input light, the output polarization at each point
face mask design (for instance using the data in Figure 3). This on the mask was chosen according to Figure 4a. In the general
allows one to reduce the effect of the cross-talk and measure case, the mask can then be designed using the Jones matrix
the polarization state more precisely. found from eq 2. In this special case, however, the device is a
The designed small focal distances (e.g., 9.6 μm for the 4.8 set of nanoposts acting as quarter- or half-wave plates.
μm pixel) result in an upper limit of 40% for the operation Therefore, we designed the nanoposts in a manner similar to
bandwidth of the device due to diffractive chromatic ref 71 to make it robust to fabrication errors.
dispersion. To get this upper limit, we assumed a constant To find the transmission amplitude and phase for the
phase profile that does not change with wavelength and used nanoposts (Supporting Information Figure S1), we simulated a
the criterion given in ref 47. Therefore, the actual bandwidth of uniform array of nanoposts with rectangular cross sections
the device is limited by the focusing and polarization control under normally incident x- and y-polarized light using the
efficiencies that drop with detuning from the design wave- rigorous coupled wave analysis.61 The resulting complex
length. In addition, it is expected that the same level of transmissions were then used to find the best nanopost that
3137 DOI: 10.1021/acsphotonics.8b00362
ACS Photonics 2018, 5, 3132−3140
ACS Photonics Article

provides each required phase pair through minimizing the Author Contributions
Euclidean distance between [eiϕx, eiϕy] and [tx, ty], where ϕx and E.A., A.A., and A.F. conceived the experiments. E.A., S.M.K.,
ϕy are the desired phase values and tx and ty are complex and A.A. designed and fabricated the devices. E.A. and S.M.K.
nanopost transmissions. The optimized nanopost dimensions performed the measurements and analyzed the data. E.A. and
are plotted in Figure 2b. A.F. wrote the manuscript. All authors discussed the results
Fabrication. The fabrication process is the same for both and the manuscript.
the DoFP metasurface mask and the polarization imaging Notes
target. The fabrication started with deposition of a 650-nm- The authors declare no competing financial interest.

■
thick layer of α-Si on a 500-μm-thick fused silica substrate. The
metasurface pattern is defined in a ∼ 300-nm-thick ZEP-520A ACKNOWLEDGMENTS
positive electron-beam resist using electron-beam lithography.
After development of the resist, a ∼ 70-nm-thick layer of This work was supported by the DARPA EXTREME program.
aluminum oxide is deposited on the sample using electron- We gratefully acknowledge critical support and infrastructure
beam evaporation and lifted off to invert the pattern. The provided for this work by the Kavli Nanoscience Institute at
Caltech.

■
aluminum oxide is then used as a hard mask in the reactive ion
etching of the α-Si layer. Finally, the aluminum oxide mask is
removed in a solution of hydrogen peroxide and ammonium REFERENCES
hydroxide. (1) Chang, C.-I. Hyperspectral Imaging: Techniques for Spectral
Measurement. The measurement setups (including part Detection and Classification; Springer Science & Business Media, 2003;
models) are schematically illustrated in Supporting Informa- Vol. 1.
(2) Sun, D.-W. Hyperspectral Imaging for Food Quality Analysis and
tion Figure S2 for different parts of the characterization Control; Elsevier, 2010.
process. To characterize the DoFP superpixel performance, (3) Tyo, J. S.; Goldstein, D. L.; Chenault, D. B.; Shaw, J. A. Review
light from an LED was passed through a linear polarizer and a of passive imaging polarimetry for remote sensing applications. Appl.
QWP to set the input polarization state. The six different Opt. 2006, 45, 5453−5469.
polarization states (Figure 3a) were generated using this (4) Garcia, N. M.; de Erausquin, I.; Edmiston, C.; Gruev, V. Surface
combination. The intensity distribution patterns at the focal normal reconstruction using circularly polarized light. Opt. Express
plane after the DoFP metasurface mask were then imaged 2015, 23, 14391−14406.
using a custom-built microscope. The data were analyzed by (5) Coffeen, D. L. Polarization and scattering characteristics in the
calculating the Stokes parameters measured by each superpixel atmospheres of Earth, Venus, and Jupiter. J. Opt. Soc. Am. 1979, 69,
and averaging over all the superpixels that fit within the field of 1051−1064.
(6) Walraven, R. Polarization Imagery. Opt. Eng. 1981, 20, 200114.
view. A 10 nm bandwidth filter with a center wavelength of 850
(7) Egan, W. G.; Johnson, W. R.; Whitehead, V. S. Terrestrial
nm was inserted in the path to characterize the narrow-band polarization imagery obtained from the Space Shuttle:characterization
operation and was then removed to acquire the results for a and interpretation. Appl. Opt. 1991, 30, 435−442.
wider-bandwidth source. (8) Liu, Y.; York, T.; Akers, W.; Sudlow, G.; Gruev, V.; Achilefu, S.
The imaging polarimetry experiments were performed in a Complementary fluorescence-polarization microscopy using division-
similar way. For these experiments, the polarization target was of-focal-plane polarization imaging sensor. J. Biomed. Opt. 2012, 17,
illuminated by x-polarized light out of a supercontinuum laser 116001.
source (filtered by the same 10 nm bandwidth filter). The (9) Roberts, N. W.; How, M. J.; Porter, M. L.; Temple, S. E.;
target was imaged onto the DoFP metasurface mask plane Caldwell, R. L.; Powell, S. B.; Gruev, V.; Marshall, N. J.; Cronin, T. W.
using a secondary custom-built microscope (operating as relay Animal Polarization Imaging and Implications for Optical Processing.
Proc. IEEE 2014, 102, 1427−1434.
optics). The intensity distribution at the focal plane after the
(10) Charanya, T.; York, T.; Bloch, S.; Sudlow, G.; Liang, K.; Garcia,
DoFP metasurface mask was then imaged and analyzed to M.; Akers, W. J.; Rubin, D.; Gruev, V.; Achilefu, S. Trimodal color-
generate the polarization images plotted in Figure 4d. For fluorescence-polarization endoscopy aided by a tumor selective
comparison, the DoFP metasurface mask was removed and a molecular probe accurately detects flat lesions in colitis-associated
polarization analyzer (i.e., a QWP and a linear polarizer) was cancer. J. Biomed. Opt. 2014, 19, 14.
inserted into the system to form the reference polarization (11) Garcia, M.; Edmiston, C.; Marinov, R.; Vail, A.; Gruev, V. Bio-
images plotted in Figure 4c. inspired color-polarization imager for real-time in situ imaging. Optica

■ ASSOCIATED CONTENT
* Supporting Information
S
2017, 4, 1263−1271.
(12) Johnson, J. Infrared polarization signature feasibility tests. US
Army Mobility Equipment Research and Development Center, 1974,
TR-EO-74-1 (AD COO1-133).
The Supporting Information is available free of charge on the (13) Chin-Bing, S. Infrared polarization signature analysis. Defense
ACS Publications website at DOI: 10.1021/acsphoto- Technical Information Center, 1976, Rep. ADC008418.
nics.8b00362. (14) Solomon, J. E. Polarization imaging. Appl. Opt. 1981, 20,
Extended simulation data for the metasurface platform; 1537−1544.
measurement setup details; measurement results of the (15) Pezzaniti, J. L.; Chipman, R. A. Mueller matrix imaging
polarimetry. Opt. Eng. 1995, 34, 11.
metasurface masks with other pixel sizes (PDF)

■
(16) Nordin, G. P.; Meier, J. T.; Deguzman, P. C.; Jones, M. W.
Micropolarizer array for infrared imaging polarimetry. J. Opt. Soc. Am.
AUTHOR INFORMATION A 1999, 16, 1168−1174.
Corresponding Author (17) Guo, J.; Brady, D. Fabrication of thin-film micropolarizer arrays
*E-mail: [email protected]. for visible imaging polarimetry. Appl. Opt. 2000, 39, 1486−1492.
(18) Gruev, V.; Spiegel, J. V. d.; Engheta, N. Image sensor with focal
ORCID plane extraction of polarimetric information. 2006 IEEE International
Ehsan Arbabi: 0000-0002-5328-3863 Symposium on Circuits and Systems; 2006.

3138 DOI: 10.1021/acsphotonics.8b00362

ACS Photonics 2018, 5, 3132−3140
ACS Photonics Article

(19) Rust, D. Integrated dual imaging detector. US Patent 5,438,414, et al. Electrically tunable all-dielectric optical metasurfaces based on
1995. liquid crystals. Appl. Phys. Lett. 2017, 110, 071109.
(20) Andreou, A. G.; Kalayjian, Z. K. Polarization imaging: (42) Paniagua-Dominguez, R.; Yu, Y.; Khaidarov, E.; Choi, S.;
principles and integrated polarimeters. IEEE Sens. J. 2002, 2, 566− Leong, V.; Bakker, R. M.; Liang, X.; Fu, Y. H.; Valuckas, V.; Krivitsky,
576. L. A.; et al. A metalens with near-unity numerical aperture. Nano Lett.
(21) Gruev, V.; Ortu, A.; Lazarus, N.; Spiegel, J. V. d.; Engheta, N. 2018, 18, 2124−2132.
Fabrication of a dual-tier thin film micropolarization array. Opt. (43) Zhan, A.; Colburn, S.; Dodson, C. M.; Majumdar, A.
Express 2007, 15, 4994−5007. Metasurface Freeform Nanophotonics. Sci. Rep. 2017, 7, 1673.
(22) Gruev, V.; Perkins, R.; York, T. CCD polarization imaging (44) Thyagarajan, K.; Sokhoyan, R.; Zornberg, L.; Atwater, H. A.
sensor with aluminum nanowire optical filters. Opt. Express 2010, 18, Millivolt Modulation of Plasmonic Metasurface Optical Response via
19087−19094. Ionic Conductance. Adv. Mater. 2017, 29, 1701044.
(23) Deguzman, P. C.; Nordin, G. P. Stacked subwavelength (45) Kamali, S. M.; Arbabi, E.; Arbabi, A.; Horie, Y.; Faraji-Dana,
gratings as circular polarization filters. Appl. Opt. 2001, 40, 5731− M.; Faraon, A. Angle-Multiplexed Metasurfaces: Encoding Independ-
5737. ent Wavefronts in a Single Metasurface under Different Illumination
(24) Myhre, G.; Hsu, W.-L.; Peinado, A.; LaCasse, C.; Brock, N.; Angles. Phys. Rev. X 2017, 7, 041056.
Chipman, R. A.; Pau, S. Liquid crystal polymer full-stokes division of (46) Arbabi, A.; Horie, Y.; Bagheri, M.; Faraon, A. Dielectric
focal plane polarimeter. Opt. Express 2012, 20, 27393−27409. metasurfaces for complete control of phase and polarization with
(25) Hsu, W.-L.; Myhre, G.; Balakrishnan, K.; Brock, N.; Ibn-Elhaj, subwavelength spatial resolution and high transmission. Nat. Nano-
M.; Pau, S. Full-Stokes imaging polarimeter using an array of elliptical technol. 2015, 10, 937−943.
polarizer. Opt. Express 2014, 22, 3063−3074. (47) Arbabi, A.; Arbabi, E.; Kamali, S. M.; Horie, Y.; Han, S.;
(26) Hsu, W.-L.; Davis, J.; Balakrishnan, K.; Ibn-Elhaj, M.; Kroto, S.; Faraon, A. Miniature optical planar camera based on a wide-angle
Brock, N.; Pau, S. Polarization microscope using a near infrared full- metasurface doublet corrected for monochromatic aberrations. Nat.
Stokes imaging polarimeter. Opt. Express 2015, 23, 4357−4368. Commun. 2016, 7, 13682.
(27) Kamali, S. M.; Arbabi, E.; Arbabi, A.; Faraon, A. A review of (48) Arbabi, A.; Arbabi, E.; Horie, Y.; Kamali, S. M.; Faraon, A.
dielectric optical metasurfaces for wavefront control. Nanophotonics Planar metasurface retroreflector. Nat. Photonics 2017, 11, 415−420.
2018, 7, 1041−1068. (49) Groever, B.; Chen, W. T.; Capasso, F. Meta-Lens Doublet in
(28) Kuznetsov, A. I.; Miroshnichenko, A. E.; Brongersma, M. L.;
the Visible Region. Nano Lett. 2017, 17, 4902−4907.
Kivshar, Y. S.; Lukyanchuk, B. Optically resonant dielectric (50) Arbabi, E.; Arbabi, A.; Kamali, S. M.; Horie, Y.; Faraji-Dana,
nanostructures. Science 2016, 354, aag2472. M.; Faraon, A. MEMS-tunable dielectric metasurface lens. Nat.
(29) Staude, I.; Schilling, J. Metamaterial-inspired silicon nano-
Commun. 2018, 9, 812.
photonics. Nat. Photonics 2017, 11, 274−284.
(51) Ellenbogen, T.; Seo, K.; Crozier, K. B. Chromatic Plasmonic
(30) Hsiao, H.-H.; Chu, C. H.; Tsai, D. P. Fundamentals and
Polarizers for Active Visible Color Filtering and Polarimetry. Nano
Applications of Metasurfaces. Small Methods 2017, 1, 1600064.
(31) Astilean, S.; Lalanne, P.; Chavel, P.; Cambril, E.; Launois, H. Lett. 2012, 12, 1026−1031.
(52) Wen, D.; Yue, F.; Kumar, S.; Ma, Y.; Chen, M.; Ren, X.;
High-efficiency subwavelength diffractive element patterned in a high-
Kremer, P. E.; Gerardot, B. D.; Taghizadeh, M. R.; Buller, G. S.; et al.
refractive-index material for 633 nm. Opt. Lett. 1998, 23, 552−554.
(32) Lalanne, P.; Astilean, S.; Chavel, P.; Cambril, E.; Launois, H. Metasurface for characterization of the polarization state of light. Opt.
Blazed binary subwavelength gratings with efficiencies larger than Express 2015, 23, 10272−10281.
those of conventional échelette gratings. Opt. Lett. 1998, 23, 1081− (53) Khorasaninejad, M.; Zhu, W.; Crozier, K. B. Efficient
1083. polarization beam splitter pixels based on a dielectric metasurface.
(33) Fattal, D.; Li, J.; Peng, Z.; Fiorentino, M.; Beausoleil, R. G. Flat Optica 2015, 2, 376−382.
dielectric grating reflectors with focusing abilities. Nat. Photonics (54) Pors, A.; Nielsen, M. G.; Bozhevolnyi, S. I. Plasmonic
2010, 4, 466−470. metagratings for simultaneous determination of Stokes parameters.
(34) Lu, F.; Sedgwick, F. G.; Karagodsky, V.; Chase, C.; Chang- Optica 2015, 2, 716−723.
Hasnain, C. J. Planar high-numerical-aperture low-loss focusing (55) Chen, W. T.; Torok, P.; Foreman, M. R.; Liao, C. Y.; Tsai, W.
reflectors and lenses using subwavelength high contrast gratings. Y.; Wu, P. R.; Tsai, D. P. Integrated plasmonic metasurfaces for
Opt. Express 2010, 18, 12606−12614. spectropolarimetry. Nanotechnology 2016, 27, 224002.
(35) Vo, S.; Fattal, D.; Sorin, W. V.; Zhen, P.; Tho, T.; Fiorentino, (56) Balthasar Mueller, J. P.; Leosson, K.; Capasso, F. Ultracompact
M.; Beausoleil, R. G. Sub-Wavelength Grating Lenses With a Twist. metasurface in-line polarimeter. Optica 2016, 3, 42−47.
IEEE Photonics Technol. Lett. 2014, 26, 1375−1378. (57) Ding, F.; Pors, A.; Chen, Y.; Zenin, V. A.; Bozhevolnyi, S. I.
(36) Lin, D.; Fan, P.; Hasman, E.; Brongersma, M. L. Dielectric Beam-Size-Invariant Spectropolarimeters Using Gap-Plasmon Meta-
gradient metasurface optical elements. Science 2014, 345, 298−302. surfaces. ACS Photonics 2017, 4, 943−949.
(37) Arbabi, A.; Horie, Y.; Ball, A. J.; Bagheri, M.; Faraon, A. (58) Maguid, E.; Yulevich, I.; Yannai, M.; Kleiner, V.; Brongersma,
Subwavelength-thick lenses with high numerical apertures and large M. L.; Hasman, E. Multifunctional interleaved geometric-phase
efficiency based on high-contrast transmitarrays. Nat. Commun. 2015, dielectric metasurfaces. Light: Sci. Appl. 2017, 6, e17027.
6, 7069. (59) Wei, S.; Yang, Z.; Zhao, M. Design of ultracompact
(38) Khorasaninejad, M.; Chen, W. T.; Devlin, R. C.; Oh, J.; Zhu, A. polarimeters based on dielectric metasurfaces. Opt. Lett. 2017, 42,
Y.; Capasso, F. Metalenses at visible wavelengths: Diffraction-limited 1580−1583.
focusing and subwavelength resolution imaging. Science 2016, 352, (60) Huard, S. Polarization of Light; Wiley-VCH, 1997; Vol. 1.
1190−1194. (61) Liu, V.; Fan, S. S4: A free electromagnetic solver for layered
(39) Yin, X.; Steinle, T.; Huang, L.; Taubner, T.; Wuttig, M.; periodic structures. Comput. Phys. Commun. 2012, 183, 2233−2244.
Zentgraf, T.; Giessen, H. Beam switching and bifocal zoom lensing (62) Arbabi, E.; Arbabi, A.; Kamali, S. M.; Horie, Y.; Faraon, A. High
using active plasmonic metasurfaces. Light: Sci. Appl. 2017, 6, e17016. efficiency double-wavelength dielectric metasurface lenses with
(40) Schlickriede, C.; Waterman, N.; Reineke, B.; Georgi, P.; Li, G.; dichroic birefringent meta-atoms. Opt. Express 2016, 24, 18468−
Zhang, S.; Zentgraf, T. Imaging through Nonlinear Metalens Using 18477.
Second Harmonic Generation. Adv. Mater. 2018, 30, 1703843. (63) Arbabi, A.; Arbabi, E.; Kamali, S. M.; Horie, Y.; Han, S.;
(41) Komar, A.; Fang, Z.; Bohn, J.; Sautter, J.; Decker, M.; Faraon, A. Increasing Efficiency of High-NA Metasurface Lenses; SPIE
Miroshnichenko, A.; Pertsch, T.; Brener, I.; Kivshar, Y. S.; Staude, I.; Photon, West, 2017; pp 101130K−1.

3139 DOI: 10.1021/acsphotonics.8b00362

ACS Photonics 2018, 5, 3132−3140
ACS Photonics Article

(64) Kamali, S. M.; Arbabi, E.; Arbabi, A.; Horie, Y.; Faraon, A.
Highly tunable elastic dielectric metasurface lenses. Laser Photon. Rev.
2016, 10, 1062−1062.
(65) Sell, D.; Yang, J.; Doshay, S.; Yang, R.; Fan, J. A. Large-Angle,
Multifunctional Metagratings Based on Freeform Multimode Geo-
metries. Nano Lett. 2017, 17, 3752−3757.
(66) Arbabi, E.; Arbabi, A.; Kamali, S. M.; Horie, Y.; Faraon, A.
Controlling the sign of chromatic dispersion in diffractive optics with
dielectric metasurfaces. Optica 2017, 4, 625−632.
(67) Bayer, B. Color imaging array. US Patent, 3,971,065, 1976.
(68) Arbabi, E.; Arbabi, A.; Kamali, S. M.; Horie, Y.; Faraon, A.
Multiwavelength metasurfaces through spatial multiplexing. Sci. Rep.
2016, 6, 32803.
(69) Lin, D.; Holsteen, A. L.; Maguid, E.; Wetzstein, G.; Kik, P. G.;
Hasman, E.; Brongersma, M. L. Photonic Multitasking Interleaved Si
Nanoantenna Phased Array. Nano Lett. 2016, 16, 7671−7676.
(70) Hu, J.; Liu, C.-H.; Ren, X.; Lauhon, L. J.; Odom, T. W.
Plasmonic Lattice Lenses for Multiwavelength Achromatic Focusing.
ACS Nano 2016, 10, 10275−10282.
(71) Backlund, M. P.; Arbabi, A.; Petrov, P. N.; Arbabi, E.; Saurabh,
S.; Faraon, A.; Moerner, W. E. Removing orientation-induced
localization biases in single-molecule microscopy using a broadband
metasurface mask. Nat. Photonics 2016, 10, 459−462.

3140 DOI: 10.1021/acsphotonics.8b00362

ACS Photonics 2018, 5, 3132−3140