1 s2.0 S1369702124000105 Main
1 s2.0 S1369702124000105 Main
1 s2.0 S1369702124000105 Main
RESEARCH
efficiency metalenses in the ultraviolet
Joohoon Kim a,1, Yeseul Kim a,1, Wonjoong Kim b,1, Dong Kyo Oh a, Dohyun Kang a,
Junhwa Seong a, Jeong Woo Shin c, Dohyun Go c, Chanwoong Park b, Hyoin Song b,
Jihwan An a,c, Heon Lee b,⇑, Junsuk Rho a,d,e,f,g,⇑
a
Department of Mechanical Engineering, Pohang University of Science and Technology (POSTECH), Pohang 37673, Republic of Korea
b
Department of Materials Science and Engineering, Korea University, Seoul 02841, Republic of Korea
c
Department of Manufacturing Systems and Design Engineering, Seoul Nation University of Science and Technology, Seoul 01811, Republic of Korea
d
Department of Chemical Engineering, Pohang University of Science and Technology (POSTECH), Pohang 37673, Republic of Korea
e
Department of Electrical Engineering, Pohang University of Science and Technology (POSTECH), Pohang 37673, Republic of Korea
f
POSCO-POSTECH-RIST Convergence Research Center for Flat Optics and Metaphotonics, Pohang 37673, Republic of Korea
g
National Institute of Nanomaterials Technology (NINT), Pohang 37673, Republic of Korea
Metalenses have outstanding light-modulating performances, and studies have been conducted on
them to not only replace conventional bulky and heavy refractive lenses but also to expand on them.
However, their operating wavelengths have rarely covered the ultraviolet (UV) regime since UV-
transparent materials are scarce and nanopatterning techniques have a small patterning area, high
cost, and low throughput. These limitations are overcome in this study, and centimeter-scale and
highly efficient UV metalenses are successfully mass-produced. The UV metalens is designed to operate
at a wavelength of 325 nm, with a numerical aperture of 0.2. Argon fluoride photolithography is used
to fabricate an 8-inch master stamp in which 300 metalenses are patterned in an array with a high
resolution. The fabricated master stamp can be duplicated repeatedly using wafer-scale nanoimprint
lithography. To improve efficiency, we developed a zirconium dioxide–polymer hybrid material that is
scalable, easily manufacturable, UV-transparent, and high-index material. The experimental results
confirm that the mass-produced metalenses operate as ideal imaging systems, exhibiting an average
measured efficiency of 45.1 %.
Keywords: Scalable nanomanufacturing; Ultraviolet metasurface; Hybrid material; Zirconium dioxide; ArF photolithography;
Nanoimprint lithography; Atomic layer deposition; UV transparent material
Thus far, UV light modulation has relied on conventional hybrid material with a high refractive index (Fig. 1). We use
reflective or refractive optics. However, they are bulky and heavy high-speed e-beam lithography to fabricate a reticle patterned
and have a low degree of freedom to light modulation, resulting with a single 4 cm UV metalens. An argon fluoride (ArF) scanner
in limited functionality, diversity, manufacturability, and opera- and manufactured reticle are used to fabricate a master mold
tion bandwidth. Metasurfaces have recently emerged as a new with an array pattern of 1 cm UV metalens on an 8-inch silicon
strategy owing to their extraordinary light-modulation ability, wafer. It should be noted that the ArF scanner has a sufficiently
subwavelength resolution, ease of integration, and compact high resolution of 40 nm to pattern the UV metalens. Although
form-factor [10–15]. Their versatile design allows various applica- the ArF scanner is expensive, our method enables extremely low-
tions such as lenses [16–19], holography [20–25], structural color cost and high-throughput fabrication using NIL. ZrO2–polymer
[26–29], and biosensors [30]. However, the operating wave- hybrid material enables the use of an imprinted resin as the final
lengths of previous metasurfaces rarely extend to the UV regime structure. By coating the resin structures with a 15 nm layer of
and remain mostly in the visible and infrared regimes owing to ZrO2, light is strongly confined within the thin ZrO2 film, result-
RESEARCH
the long-standing challenges of the UV metasurface. ing in a high conversion efficiency of 80 % at the target wave-
The main bottlenecks in UV metasurfaces are the scarcity of length of 325 nm. The diffraction-limited focusing and
UV-transparent materials and the high cost and low throughput imaging performance of the fabricated metalens are verified.
of nano-patterning techniques. Conventional dielectric materi- Because both the master stamp and replica mold are reusable,
als, such as silicon (Si)-based materials, titanium dioxide (TiO2), our fabrication method enables the low-cost and high-
and gallium nitride (GaN), have a small bandgap, resulting in throughput fabrication of large-area UV metasurfaces.
large absorption and low transmission in the UV regime [31].
To date, new materials such as silicon nitride (Si3N4) [32], nio- Results
bium pentoxide (Nb2O5) [33], zinc oxide (ZnO) [34], and haf- Design principle for full-phase modulation of UV light
nium oxide (HfO2) [35] have been used as UV metasurfaces; The primary objective of this work is to use the printed resin
however, these UV-transparent materials are not scalable or easy itself as a meta-atom by coating a high-index material, thereby
to manufacture. The fabrication method of previous UV metasur- fabricating a UV metasurface at a low price and with a high
faces involves electron beam lithography for high-resolution pat- throughput. A constituent material of the meta-atom should
terning, resulting in an extremely high cost, a low throughput, have both a sufficiently high refractive index (n) to confine the
and a complicated fabrication process. Thus, large UV metasur- light well and a sufficiently low extinction coefficient (k) to mini-
faces cannot be manufactured using previous approaches, which mize absorption in the UV regime. However, conventional prin-
are far from practical use (Supplementary Note 1). To overcome table materials such as UV and thermal imprint resins have a low
these limitations, nanoimprint lithography (NIL) has been refractive index of approximately 1.5 (Supplementary Note 2). To
actively used to lower the fabrication cost and fasten the increase the effective refractive index, we proposed a high-index
throughput [36–43], and photolithography has been used to fab- atomic layer-polymer hybrid material; however, the previous
ricate large-area metasurfaces on the wafer scale [44,45]. hybrid material is not transparent in the UV region owing to
Recently, visible metalenses are mass-produced using pho- the high UV absorption of TiO2. Therefore, we propose a ZrO2–
tolithography and nanoimprint lithography [46]. However, the polymer hybrid material in which the printed metasurface is
low fabrication compatibility of UV-transparent materials and thinly coated with a high-index material to increase the effective
low patterning resolution still hinder wafer-scale manufacturing refractive index. This high-index material must meet several con-
of UV metasurfaces. ditions: first, the refractive index must be high enough to confine
In this study, we overcome this long-standing challenge by the light well, even with the thin film; second, the bandgap must
introducing a wafer-scale nanoimprintable UV metasurface plat- be wide enough to be UV-transparent; third, the high-index
form and a UV-transparent zirconium dioxide (ZrO2)–polymer material must be deposited using atomic layer deposition (ALD)
FIG. 1
Wafer-scale imprinted UV metasurface platform with high efficiency. (a) Schematic of fabrication of the ZrO2–polymer hybrid material for efficiency
improvement. (b) Photograph of mass-produced 1-cm UV metalenses on 4-inch, 6-inch, and 8-inch wafer.
2
Please cite this article in press as: J. Kim et al., Materials Today (2024), https://doi.org/10.1016/j.mattod.2024.01.010
Materials Today d Volume xxx, Number xx d xxxx 2024 RESEARCH
to form uniform and thin films on all sides of the metasurface. To The transmitted electric field ET can be calculated using T, as
meet all of the above conditions, we choose ZrO2 as a UV high- follows
index material. The optical properties of the ZrO2 films are mea-
sured by ellipsometry (Supplementary Note 3). In the UV regime, tl þ ts 1 t l t s i2h 1
ET ¼ þ e ð3Þ
the ZrO2 film has a high refractive index (n) of over 2.2 and a 2 i 2 i
near-zero extinction coefficient (k) (Fig. 2b). The ZrO2 film has
Transmitted light with converted handedness of polarization
a wide bandgap of 6.195 eV, which makes it transparent to UV
(cross-polarization) has a phase delay of 2h. The amplitude of
light (Fig. 2c, Supplementary Note 4).
the cross-polarized component is defined as the conversion effi-
The finite element method (FEM) is used to simulate the trans-
ciency which refers that the meta-atom transforms into the
mission properties of the thin film-coated meta-atoms. To
desired phase term when passing through it, which a value of
achieve full phase modulation, the Pancharatnam–Berry (PB)
1 representing a perfect attainment of the desired phase. There-
phase, also known as the geometric phase, is used as a meta-
fore, the higher conversion efficiency (which is defined as
RESEARCH
atom. The PB phase uses an anisotropic meta-atom, which is a
‘high-efficiency’) is directly related to the focusing efficiency.
birefringence and can be expressed by the Jones matrix J [47],
To design high-efficiency meta-atoms that operate at a wave-
given by
length of 325 nm, the conversion efficiencies of meta-atoms are
tl 0 calculated by sweeping lengths from 150 to 240 nm and widths
J¼ ð1Þ
0 ts from 40 to 130 nm, with a fixed height of 550 nm, ALD thickness
where tl and ts represent the complex transmission coefficients of of 15 nm, and periodicity of 280 nm (Fig. 2d, Supplementary
light polarized along the long and short axes of the meta-atoms, Note 5). Considering the fabrication constraints, we chose a
respectively. The Jones matrix T for h rotated meta-atoms can meta-atom with a length of 240 nm and width of 60 nm to
be expressed using the rotational matrix R(h) as follows: achieve high efficiency of 80 %. Furthermore, many candidate
structures have a high efficiency near the target structures; thus,
T ¼ RðhÞJRðhÞ ð2Þ
some fabrication errors are acceptable. As expected, the conver-
FIG. 2
Design and simulation of high-efficiency UV meta-atom. (a) Configuration of the hybrid meta-atom structure. Length: 240 nm; width: 60 nm; height: 550 nm;
period: 280 nm; thickness: 15 nm. (b) Measured optical properties of ZrO2 film. Blue line: refractive index; orange line: extinction coefficient. (c) Calculated
optical bandgap of ZrO2 film. (d) Simulated conversion efficiency of the meta-atom at the wavelength of 325 nm with fixed height (550 nm), period (280 nm),
and ALD thickness (15 nm). (e,f) Plot of simulated conversion efficiencies with varying– (e) ALD thickness from 0 nm to 19 nm, and (f) wavelength from
250 nm to 400 nm.
3
Please cite this article in press as: J. Kim et al., Materials Today (2024), https://doi.org/10.1016/j.mattod.2024.01.010
RESEARCH Materials Today d Volume xxx, Number xx d xxxx 2024
sion efficiency increases rapidly as the thickness increases, which e). Next, the fabricated master mold is spin-coated with hard
verifies our approach to achieve high efficiency using the facile polydimethylsiloxane (h-PDMS) solution to replicate the inverse
method (Fig. 2e). The simulation results of the electric-field dis- pattern with a high resolution of 40 nm because of its high mod-
tribution of optimized meta-atom, other sweep results varying ulus (9N/mm2) [48]. Subsequently, polydimethylsiloxane
H and imprint residual are added in Supplementary Note 6 to (PDMS) is spin-coated onto the h-PDMS layer. In both processes,
8. Finally, owing to the broadband property of the PB phase, h-PDMS and PDMS harden by heating to 70℃. The stamp is man-
our designed meta-atom could operate not only at 325 nm but ufactured by detaching the solidified h-PDMS and PDMS from
also throughout the UV regime (Fig. 2f, Supplementary Note 9). the master mold (Fig. 3f). The conventional UV-curable resin
(MINS-311RM) is subsequently spin-coated onto the fabricated
stamp and then covered with a silica substrate. The imprint resin
Facile fabrication method using NIL and ALD for high-efficiency is cured under the correct amount of pressure and UV light. After
UV metasurface the stamp is separated from the cured resin, reactive ion etching
RESEARCH
FIG. 3
Fabrication schematic for wafer-scale manufacturing of highly efficient UV metasurface. Photograph of fabricated (a) reticle, (b,c) master stamp, and (d)
metalenses. Scanning electron microscope (SEM) images of (e) the master stamp, (f) replica mold, and (g) metalens.
4
Please cite this article in press as: J. Kim et al., Materials Today (2024), https://doi.org/10.1016/j.mattod.2024.01.010
Materials Today d Volume xxx, Number xx d xxxx 2024 RESEARCH
Our fabrication method allows large and high-resolution pat- Design and demonstration of UV metalens
terning of UV hybrid metasurfaces at low cost and with high The metasurface must satisfy the lens equation to focus the light
throughput. Our process leverages high-speed e-beam lithogra- at one focal point. Because we used a geometric phase (PB phase),
phy, ArF photolithography, NIL, and ALD systems. High-speed the meta-atoms are rotated to obtain the desired phase at specific
e-beam lithography enables high-resolution patterning of large angles. The phase profile is derived from the set parameters:
areas but is expensive. Therefore, ArF photolithography is used diameter D = 1 cm, focal length fl = 2.4 cm, and numerical aper-
to make an 8-inch master stamp in which the pattern of the sin- ture NA = 0.2. The generated phase map is shown in Supplemen-
gle reticle is continuously exposed by moving the substrate. It tary Note 14. We confirmed the focusing ability of the UV
should be noted that ArF photolithography has a high resolution metalens through both simulations and measurements
of 40 nm, which is sufficient to fabricate the designed UV meta- (Fig. 4a). In the simulation, the angular spectrum method was
atoms. NIL is then used to replicate the master stamp pattern at used to propagate the electric field in the z-direction. Fig. 4f
an extremely low cost and high throughput. Because the residual and 4g are the optical measurement setups for the characteriza-
RESEARCH
layer is adjusted to a minimum through spin coating and process tion of metalens. As shown in the schematic of the focusing mea-
pressure, the imprint resin could be cured in 5 min, despite a surement setup in Fig. 4f, a linear polarizer (LP) and a quarter-
large patterning area. Moreover, the reticle, master stamp, and wave plate (QWP) are used to control the polarization state and
replica mold are reusable, which reduces the cost and increases generate circularly polarized light, and the microscope part
the throughput (Supplementary Note 12). Since conventional including the objective lens (OL), QWP, LP, lens, and sCMOS
nanoimprint resin is used in this work, this method can transfer camera on the stage are used to magnify the image passing
structures as high as 1 lm in height (Supplementary Note 13). through the metalens (abbreviated as ML in the figure to express
Finally, ALD enables printed resin structures to be used as the a brief term) and to scan mechanically along the z-axis. In this
final structure by forming hybrid materials that are scalable part, the pair of LP and QWP on the stage is used as the analyzer
and easily manufacturable. All the fabrication steps are compati- to filter the opposite circularly polarized light compared with the
ble with mature and well-known semiconductor fabrication incident light. To demonstrate the focusing ability of the UV
methods; therefore, our method does not require additional metalens clearly, we cropped the z range from 22 mm to
facilities. 26 mm, and the focal length of the experimental one matched
FIG. 4
Optical characterization of the fabricated UV metalens. (a) The normalized simulated and measured intensity distributions in the x-z planes. (b) The
normalized cross-sectional intensity profiles of the simulated and measured focal points. The grey dashed curve is the simulated data and the purple solid
curve is the measured data. The normalized intensity distribution of the focal point from the experimental measurement is shown on the right top. (c)
Captured images of a negative 1951 USAF resolution target using the metalens for the object in the focal plane. (d) Normalized intensity of vertical cuts
across from elements 1 to 3 of group 7 of the target corresponding to the red dashed line in Fig. 4c. (e) Normalized intensity of vertical cuts across from
elements 4 to 6 of group 7 of the target corresponding to the blue dashed line in Fig. 4c. (f) Schematic of the optical setup to focusing test. (g) Schematic of
the optical setup to image the negative 1951 USAF resolution target.
5
Please cite this article in press as: J. Kim et al., Materials Today (2024), https://doi.org/10.1016/j.mattod.2024.01.010
RESEARCH Materials Today d Volume xxx, Number xx d xxxx 2024
well with the focal length of the theoretical one. Furthermore, we tional imprint resin to create nanostructures. This approach
measure the absolute focusing efficiencies of the manufactured offers advantages in terms of ease of pattern transfer, high yield
metalenses, defined as the amount of focused light versus inci- and suitability for large-area processing. The fabricated UV met-
dent light, to evaluate their performance (Fig. 5). The average alens achieves a focusing efficiency of 48.5 % at a target wave-
focusing efficiency of UV metalenses manufactured on a 4-inch length of 325 nm, and the efficiency can be further improved
wafer is 45.1 % with the standard deviation of 7.5 %. The varia- by minimizing fabrication errors. The measured Airy disk of the
tion in measured efficiency occurs due to the differing etching fabricated metalens is consistent with the ideal Airy disk, and
speeds, hard mask height, and proximity effect at each position the high imaging performance of the fabricated metalens is ver-
within the wafer. Fig. 4b shows the normalized focal spots and ified by imaging a negative 1951 USAF target. We believe that our
line scan plots of the focal plane both theoretically and experi- method will be a fundamental step toward the commercializa-
mentally. The full-width-half-maximum (FWHM) values are tion of UV metalenses and provide a great method to replace
approximately 1.2 lm, and the measured and theoretical Airy conventional UV optics with meta-optics.
RESEARCH
6
Please cite this article in press as: J. Kim et al., Materials Today (2024), https://doi.org/10.1016/j.mattod.2024.01.010
Materials Today d Volume xxx, Number xx d xxxx 2024 RESEARCH
used to produce and filter the circularly polarized light. The focal try & Energy (MOTIE) of the Korean government. J.K.
point was magnified through an OL (Thorlabs LMU-15X-UVB) acknowledges the POSTECH Alchemist fellowship, and the ASAN
and captured using an sCMOS camera (Pco panda 4.2 BI UV). fellowship. Y.K. and D.K.O. acknowledge the Hyundai Motor
We measured 400 images for z-scanning at 0.01 mm intervals Chung Mong-Koo fellowships. Y.K. acknowledges the NRF Ph.D. fel-
and used the motorized translation stage (Thorlabs DDS100). A lowship (NRF-2022R1A6A3A13066251) funded by the Ministry of
negative resolution target (Thorlabs R3L1S4N) was used for the Education (MOE) of the Korean government.
imaging test.
ZrO2 deposition condition: ZrO2 thin film was deposited using a Appendix A. Supplementary data
plasma-enhanced ALD station (NexusBe, Nexus Plaminar series). Supplementary data to this article can be found online at
Tetrakis(dimethylamino)zirconium (TDMAZr) was used as the Zr https://doi.org/10.1016/j.mattod.2024.01.010.
precursor, and oxygen plasma (350 W, 1 torr) was used as the
reactant. The deposition process consisted of 3 s of TDMAZr References
RESEARCH
pulses, 30 s argon purge, 10 s oxygen plasma exposure and [1] N.J. Greenfield, Nat. Protoc. 1 (2006) 2876–2890.
30 s argon purge at 90 °C in the ALD chamber. The growth rate [2] A. Rodger, Far UV Protein Circular Dichroism, in: G. Roberts, A. Watts (Eds.),
Encyclopedia of Biophysics, Springer Berlin Heidelberg, Berlin, Heidelberg,
per cycle was 2.8 Å per cycle. 2018.
[3] S. Divitt et al., Science 364 (2019) 890–894.
CRediT authorship contribution statement [4] A.O. Nwokeoji et al., Anal. Chem. 89 (2017) 13567–13574.
[5] N. Fang et al., Science 308 (2005) 534–537.
Joohoon Kim: Data curation, Formal analysis, Investigation, [6] T. Xu et al., Nature 497 (2013) 470–474.
Methodology, Writing – original draft. Yeseul Kim: Formal [7] M. Campbell et al., Nature 404 (2000) 53–56.
analysis, Investigation, Methodology, Writing – original draft. [8] J. Oh et al., Nat. Mater. 20 (2021) 385–394.
[9] D. Zhao et al., Nanophotonics 10 (2021) 2283–2308.
Wonjoong Kim: Investigation, Methodology. Dong Kyo
[10] A. Tittl et al., Science 360 (2018) 1105–1109.
Oh: Methodology. Dohyun Kang: Methodology. Junhwa [11] W.-J. Joo et al., Science 370 (2020) 459–463.
Seong: Methodology. Jeong Woo Shin: Methodology. [12] Q. Song et al., Science 373 (2021) 1133–1137.
Dohyun Go: Methodology. Chanwoong Park: Methodol- [13] Y. Zhou et al., Nat. Photon. 14 (2020) 316–323.
[14] G. Zheng et al., Nat. Nanotechnol. 10 (2015) 308–312.
ogy. Hyoin Song: Methodology. Jihwan An: Methodology.
[15] H. Ren et al., Nat. Commun. 10 (2019) 2986.
Heon Lee: Conceptualization, Funding acquisition, Project [16] X. Chen et al., Nat. Commun. 3 (2012) 1198.
administration, Supervision. Junsuk Rho: Conceptualization, [17] M. Khorasaninejad et al., Science 352 (2016) 1190–1194.
Funding acquisition, Project administration, Resources, Supervi- [18] G. Yoon et al., Nat. Commun. 11 (2020) 2268.
[19] Y. Wen et al., Adv. Opt. Mater. 11 (2023) 2300127.
sion, Validation, Writing – review & editing.
[20] P. Georgi et al., Sci. Adv. 7 (2021) eabf9718.
[21] K.T.P. Lim et al., Nat. Commun. 10 (2019) 25.
Data availability [22] J. Li et al., Sci. Adv. 4 (2018) eaar6768.
[23] P. Zheng et al., Sci. Adv. 7 (2021) eabg0363.
The data supporting the findings of this study are available
[24] X. Fang et al., Nat. Photon. 14 (2020) 102–108.
from the corresponding author upon reasonable request. [25] S. So et al., Adv. Mater. 35 (2023) 2208520.
[26] M. Song et al., Nat. Nanotechnol. 18 (2023) 71–78.
Declaration of competing interest [27] H. Liu et al., Adv. Mater. 31 (2019) 1807900.
[28] Q. Ruan et al., Adv. Mater. 34 (2022) 2108128.
The authors declare that they have no known competing [29] T. Badloe et al., Light Sci. Appl. 11 (2022) 118.
financial interests or personal relationships that could have [30] F. Yesilkoy et al., Nat. Photon. 13 (2019) 390.
appeared to influence the work reported in this paper. [31] Y. Deng et al., Adv. Mater. 30 (2018) 1802632.
[32] J. Kim et al., ACS Nano 16 (2022) 3546–3553.
[33] K. Huang et al., Laser Photonics Rev. 13 (2019) 1800289.
Acknowledgements [34] M.L. Tseng et al., Sci. Adv. 8 (2022) eabn5644.
This work was financially supported by the POSCO-POSTECH- [35] C. Zhang et al., Light Sci. Appl. 9 (2020) 55.
RIST Convergence Research Center program funded by POSCO, [36] G. Yoon et al., ACS Nano 15 (2021) 698–706.
[37] J. Kim et al., Laser Photonics, Rev. 16 (2022) 2200098.
the Samsung Research Funding & Incubation Center for Future
[38] J. Kim et al., Light Sci. Appl. 12 (2023) 68.
Technology grant (SRFC-IT1901-52) funded by Samsung Electron- [39] V.J. Einck et al., ACS Photonics 8 (2021) 2400–2409.
ics, and the National Research Foundation (NRF) grants (NRF- [40] D.E. Jung et al., ACS Appl. Nano Mater. 6 (2023) 2009–2019.
2022M3C1A3081312, NRF-2022M3H4A1A02074314, NRF- [41] I.R. Howell et al., Opt. Laser Technol. 141 (2021) 107101.
[42] M.R. Beaulieu et al., ACS Photonics 1 (2014) 799–805.
2022M3H4A1A02046445, NRF-2019R1A5A8080290) funded by [43] J. Gong et al., Opt. Express 31 (2023) 8068–8080.
the Ministry of Science and ICT (MSIT) of the Korean government. [44] A. Leitis et al., Adv. Mater. 33 (2021) 2102232.
H.L. acknowledges the NRF grants (RS-2023-00283667, NRF- [45] J. Tao et al., Adv. Mater. 34 (2022) 2106080.
[46] J. Kim et al., Nat. Mater. 22 (2023) 474–481.
2019K1A4A7A02113032, NRF-2018M3D1A1058997) funded by
[47] J. Kim et al., InfoMat 3 (2021) 739–754.
the MSIT of the Korean government, and the Technology Innova- [48] T.W. Odom et al., Langmuir 18 (2002) 5314–5320.
tion Program (20016234) funded by the Ministry of Trade, Indus-
7
Please cite this article in press as: J. Kim et al., Materials Today (2024), https://doi.org/10.1016/j.mattod.2024.01.010