Terahertz metamaterials perfect absorbers for sensing and imaging

David S. Wilbert, Mohammad P. Hokmabadi, Joshua Martinez, Patrick Kung, Seongsin M. Kim*
Department of Electrical and Computer Engineering, University of Alabama, Tuscaloosa, AL,
35487, USA
*[email protected]; phone 1 205 348-5755; fax 1 205 348-6959

ABSTRACT

Devices operating at THz frequencies have been continuously expanded in many areas of application and major research
field, which requires materials with suitable electromagnetic responses at THz frequency ranges. Unlike most naturally
occurring materials, novel THz metamaterials have proven to be well suited for use in various devices due to narrow and
tunable operating ranges. In this work, we present the results of two THz metamaterial absorber structures aiming two
important device aspects; polarization sensitivity and broad band absorption. The absorbers were simulated by finite
element method and fabricated through the combination of standard lift-off photolithography and electron beam metal
deposition. The fabricated devices were characterized by reflection mode THz time domain spectroscopy. The narrow
band absorber structures exhibit up to 95% absorption with a bandwidth of 0.1 THz to 0.15 THz.

Keywords: Terahertz, metamaterial, absorber, polarization insensitive, resonance, sensing

1. INTRODUCTION
Research focused on the terahertz (THz) gap of the electromagnetic spectrum from 0.1 to 10 THz has been experienced
rapid growth recently due to numerous possible applications such as remote sensing, material optical and electrical
property probing, astronomical radiation detection, and biomedical imaging.1-5 Because terahertz radiation is non-
ionizing, terahertz devices have become attractive for use in a wide variety of detection and imaging applications. Also,
non-ionizing terahertz radiation offers a safe alternative method to characterize material and identify material
composition. However, manipulation of the electromagnetic response of conventional materials in the terahertz regime is
very difficult, and therefore development of new optics and devices which are capable of unique operation in this regime
has been strongly demanded. Devices based on THz Metamaterials stem from the possibility to engineer optical
properties such as refractive index, electric permittivity, and magnetic permeability individually. Metamaterials have
been used to achieve negative refractive index devices such as invisibility cloaks6 and superlenses7. It has recently been
shown that careful manipulation of effective ε(ω) and µ(ω), near-perfect, narrow-band, electromagnetic absorbers may
be realized8-9. These electromagnetic absorbers are spectrally selective and can easily be scaled to operate in the
terahertz range. Terahertz absorbers could potentially be used as fundamental elements in many detection, sensing,
filtering, and imaging applications.

The work presented here discusses the design and experimental demonstration of a novel polarization insensitive
terahertz metamaterial absorber to exhibit near perfect narrow band absorption at THz frequencies. This absorber is
proposed to exhibit near-perfect absorption and insensitive to incident field polarization. First, computer simulations
were performed to develop and optimize the structure by studying the scattered field. Arrays of the absorber structure
were then fabricated using standard photolithography and electron beam deposition methods. Both designs consist of a
conducting Cu ground plane, a polyimide dielectric spacer layer with arrays of the periodic and patterned Cu resonator
structures on top. The fabricated devices were characterized by reflection mode THz time domain spectroscopy. The
narrow band absorber structures exhibit up to 95% absorption with a bandwidth of 0.1 THz to 0.15 THz.

Terahertz and Ultrashort Electromagnetic Pulses for Biomedical Applications, edited by Gerald J. Wilmink, Bennett L. Ibey,
Proc. of SPIE Vol. 8585, 85850Y · © 2013 SPIE · CCC code: 1605-7422/13/$18 · doi: 10.1117/12.2005709

Proc. of SPIE Vol. 8585 85850Y-1

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 06/18/2015 Terms of Use: http://spiedl.org/terms

 2. EXPERIMENTAL PROCEDURE
The THz metamaterial absorber devices designed and studied in this work consists of two metallic layers, an array of
electric ring resonators (ERR) and a continuous ground plane, which are separated by a dielectric layer8. In order to
develop functional absorbing elements for use in detection, sensing, and imaging applications, it is necessary that the
absorber exhibit near-perfect absorption. Near-unity absorption occurs when transmission through and reflection from
the device are simultaneously minimized. This absorption stems from losses in the dielectric which are characterized by
the imaginary component of the electric permittivity, ε’’(ω). Such structure forms 3D planar absorbers, which have been
intensely studied recently due to the fact that their electromagnetic response can be easily engineered along with simple,
inexpensive fabrication possibilities.

Each unit cell of the top metallic layer (ERR array) was designed to contain a square closed ring with four plates
arranged 90º from one another inside the ring. These plates form 45º gaps in order to generate an electric response at any
incident field polarization10. The ground plane provides freedom to place the absorber device on any desired surface
since no transmission is allowed beyond the metallic plane. The magnetic response generated by the absorber structure
occurs between the two metallic layers. As a result the magnetic response, and thus the absorption, is highly dependent
on the dielectric layer thickness. The optimization process, therefore includes first the selection of the operation
frequency by determining the in-plane dimension, then it follows by the process of optimization of the thickness of the
polyimide dielectric spacing layer to obtain the highest absorption amplitude. The various set of structures with varying
in size of the ring and thicknesses of the dielectric layer were designed by finite element simulation. This analysis was
carried out using commercially available COMSOL Multiphysics platform. The procedure begins by dividing the
structure of interest into many (104 – 105) smaller elements forming a “mesh” where Maxwell’s equations must be
solved. Finite element simulation is able to obtain solutions for very complex structures. In this work, the structure under
study was divided into approximately 76,000 elements. Both metallic layers were modeled as copper with an electrical
conductivity σ = 6.7 x 10-16 S/m and a constant frequency-independent permittivity. From scattered field numerical
solutions, the reflected power was calculated, followed by the reflection (R) and absorption (A = 1 - R).

The devices were fabricated using standard optical ultraviolet photolithography on two-inch silicon wafers. First, the
ground plane (bottom metallic layer) was deposited onto the silicon wafer in the form of a 200 nm thick Cu film. Next, a
dielectric spacer layer (polyimide) was spin coated on top of the ground plane. The thickness of polyimide applied was
varied according to simulation results such that the absorption can be optimized. The polyimide layer was then cured in a
vacuum oven. After curing, photolithography was performed in order to realize the ERR metal pattern. A positive
photoresist layer, approximately 0.8 µm thick was spin-coated on the structure and then soft baked. The resulting wafer
containing a photoresist coating was then placed into a mask aligner system with a dark-field photomask etched with the
ERR designs. The masked system was then exposed to UV radiation where the regions of photoresist not requiring metal
were shielded from exposure. Next, the resulting patterns were developed to remove all areas where the photoresist was
exposed to UV radiation. A 200 nm thick Cu layer was then deposited and the photoresist lifted off using sonication in
acetone. The resulting absorber structure was examined with an optical microscope after the lift-off process in order to
ensure that the ERR pattern integrity was maintained.
Characterization was accomplished by THz time domain spectroscopy (TDS) with a reflection mode setup as shown in
Figure 1. Linearly polarized, broadband THz emission (0.2 to 3 THz) was achieved by pumping a photoconductive
antenna which was biased at 130V using high voltage modulator with an ultrafast pulse with peak wavelength at 775 nm
and average power at 120 mW. The emitted THz beam was then passed through a high conductivity Si beam splitter and
focused with a high density polyethylene lens to a spot size of about 500 µm. A three axis motion controlled stage was
used to position the absorber arrays at the focal point for measurement. The motion controlled stage was also used to
raster scan the absorber arrays through the focal point while collecting spectroscopic data at set coordinates thereby
imaging the arrays. The entire system was maintained under a clean, dry air purge to minimize the THz beam interaction
with water vapor. An electro-optic sampling method of detection was used to measure the reflected THz pulse amplitude
which was then fast Fourier transformed (FFT) to obtain frequency dependent amplitude and phase data.

Proc. of SPIE Vol. 8585 85850Y-2

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 06/18/2015 Terms of Use: http://spiedl.org/terms

 Reflection Mode

Figure 1. Schematic diagram of reflection mode of THz TDS system used in this experiment.

3. RESULTS
Figure 2 shows the simulation results and represents the simulated electric field vector and induced surface currents. The
structures were functioning as resonators at the desired frequency in the presence of the THz radiation to be absorbed. At
the resonance frequency, we observed that the structure acts as an electric dipole with the induced electric field oriented
in the vertical direction, and the electric field was locally concentrated and maximized across the 45° gaps in the inner
loop. The surface currents, however, circulate around the four separate, intermediate loops formed by the arms, and the
maximum currents were seen near the corners giving rise to a strong magnetic response at those locations.

Figure 2. Simulated electric field vectors (left) and surface current vectors (right) for a typical absorber structure unit cell under
resonant frequency photoexcitation.

According to the simulations, the optimized parameters for the absorber arrays were expected to require polyimide
thicknesses of 5.8 µm and 7.1 µm, and yield near perfect absorption (between 98.7% to 99.6%) at the frequencies 0.84
THz and 0.95 THz.
Initial measurement was done on the three different absorber arrays and found to be in disagreement with expected
values in terms of both frequency and amplitude. Figure 3 (a) shows a typical absorber element from an array under an
optical microscope through an object. A black outline has been superimposed over the picture showing the originally
simulated dimensions. In the picture there are small, but clearly visible differences in feature sizes, especially at the 45°
gaps at the corners of the inner loop. To further understand the difference in the measurement and simulation and to
improve the device performance, we attempted incorporate the small differences in feature sizes of the actual fabricated
arrays. It was also noted that there were a discrepancy in the conductivity of Cu metal when the size falls in
subwavelength regime compared to that of bulk. A second set of identical devices was fabricated with the appropriate
10.7 µm polyimide thickness. A comparison of the absorption of similar devices with the different dielectric thicknesses
is shown in Figure 3 (b) The slight differences in feature sizes, along with the change in dielectric thickness led to a shift

Proc. of SPIE Vol. 8585 85850Y-3

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 06/18/2015 Terms of Use: http://spiedl.org/terms

 in resonance frequency from 924 GHz to 849 GHz as well as an increase in absorption amplitude from 31% to 96% at
10.7 µ m dielectric thickness.

1.0
(b) - 10.7 um
(a) -7.1 um
0.8 - 6.8 um

i1 á p 06

o
N á0.a

1 0.2

0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
Frequency (THz)

Figure 3. (a) Optical microscope picture of typical absorber array element with a superimposed outline of the initial
simulated dimensions, and (b) the frequency dependent absorption of similar structures with different dielectric thicknesses.

It was previously stated that the absorber structures were setup in different series designed to allow observation of the
effects of varying specified dimensions. Figure 4 shows the absorption spectra of these three different arrays, all with
10.7 µm polyimide thickness. The variation in the dimensions of the unit cell structures include only a single dimension
change with A being the common denominator. An increase in the diameter of the inner loop from 18 µm to 24 µ m
leads to only a slight 25 GHz change in absorption frequency from array A to B, whereas a more noticeable shift of 750
GHz occurs due to changing the length of the outer walls going from array A to C . This demonstrates the dependency of
the resonant frequency on certain dimensions more than others. In particular, the most important dimensions on the
absorption frequency are the length of the outside metal square and the closest distance between both adjacent T-shaped
structures and opposite inside T-shaped structures.

1.0
-A
-B
0.8 -C
E 0.6
:o
e-
0
ÿ .4
an
Q
02

0.0 '
02 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2D
Frequency (THz)

Figure 4. Absorption of three different arrays with single size dimension variation .

We observed a significant variations in the time domain reflected pulse between different arrays as shown in Figure 5.
Interestingly, very little difference, about 5%, is observed in the actual pulse peak amplitude, but the ringing that occurs
after the pulse is affected with the first two ringing peaks from array B being reduced from 0.44 mV and 0.4 mV to both
being only 0.32 mV. As it would be expected based on the previous absorption frequency shift from array A to C, the
time domain variation between these two arrays is even more drastic with the ringing signal peaks due to array C shifted
sooner by 0.18 ps, 0.23 ps, and 0.36 ps, respectively. It is often difficult to determine the optimal placement of the array
in the THz beam path when using the pulse peak position to as a polling position, which is what is traditionally done.

Proc. of SPIE Vol. 8585 85850Y-4

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 06/18/2015 Terms of Use: http://spiedl.org/terms

 This results in the necessity to perform an entire spectroscopic scan every time a movement is made. By polling a
position in the ringing signal it is possible to find the optimum measurement location much more quickly.

B 0.4 B
E E T
c
4; 2.0
-ÿ T3 0.2 _

il8q o
'"'SN
N
1.6
1W
1y.63 1U.0 1U.1
I

2
V 11 11 13 14 19
16 18 20 22 24
Time Delay [ps] Time Delay [ps]
Time Delay [ps]

Figure 5. Reflected time domain THz pulses the same from three absorber arrays (center) with zoomed in views of both the
main pulse peak (left) and following ringing signal (right).

It was also found that the absorber structures were indeed polarization insensitive. This measurement was performed by
rotating the absorber array in-plane about the THz beam propagation vector. There is negligible variation in the actual
absorption frequency, but the amplitude varies between 0.78 and 0.94—a change of 17%. Due to the symmetry of the
structure, it is expected that any trend seen through the rotation from 0° to 45°, should be mirrored though the rotation
from 45° to 90°, but that was not seen to occur. Upon further inspection it was seen that the rotation stage used for the
measurement was not perfectly concentric resulting in a variation of up to 1 mm in the location inside of the array that
was being measured. Small inhomogeneity in the individual elements in the array, resulting from small differences in
structure due to the fabrication process, led to the variation in absorption amplitude.
It can often be difficult to accurately place a sample at the focal point of the THz beam not only because the beam is
invisible, but also due to the fact that noise in the time domain signal makes it difficult to discern increasing or
decreasing pulse amplitude and delay. A simple solution to this problem is imaging. By using a three dimensional
motion controlled stage to raster scan a specimen in the THz beam path and taking a spectroscopic scan at set intervals,
an image of the specimen can be reconstructed. Another method that simply sets the delay stage at a constant position to
detect the THz pulse peak is much faster because it requires no actual scanning time per pixel. This fast scanning
technique was used to find, and subsequently place samples at, the focal point before measurement. This is especially
important for use before full spectroscopic imaging to ensure proper placement of the sample and proper selection of the
length and width of the imaging window. This is also invaluable for positioning samples such as the absorbers, because
they produce only small variation to the time domain signal.
TD Amplitude FD Power
-Nwmouliir'I

I
4 4 16 it 1: is
cod

Figure 6. Reconstructed images based on the time domain signal (left), specifically from the ringing after the
pulse, and the transmitted power spectrum (right) of an absorber array.

Proc. of SPIE Vol. 8585 85850Y-5

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 06/18/2015 Terms of Use: http://spiedl.org/terms

 Figure 6 shows two reconstructed spectroscopic images of an absorber array. The image (left) reconstructed from the
time domain signal was only ascertainable at a specific delay and would have been nearly impossible to find without the
imaging process. According to the image (right) reconstructed from the frequency dependent power spectrum, the array
is seems to exhibit stronger absorption near its edges indicating that the quality of the individual elements around the
outside of the array is better than that of the elements more toward the center of the array.

4. CONCLUSION
While THz frequency range optics and devices still have a long way to go before reaching maturity, the list of possible
applications continues its rapid growth. Towards that end, the metamaterial absorber devices designed and fabricated for
this work hold promise for use in the development of THz frequency range detection applications. The absorbers have
proven to exhibit a readily tunable frequency response with near perfect absorption.

ACKNOWLEDGMENT

We would like to acknowledge NSF (EECS-0824452, CAREER, and IIT-1127831) and DoD for support of this work.

REFERENCES

[1] Federici, J. F., Schulkin, B., Huang, F., Gary, D., Barat, R., Oliveira, F., and Zimdars, D., “THz imaging and
sensing for security applications—explosives, weapons, and drugs,” Semiconductor Science and Technology,
vol. 20, S266-S280 (2005).
[2] Fekete, L., Kužel, P., Nĕmec, H., Kadlec, F., Dejneka, A., Stuchlfk, J., and Fejfar, A., "Ultrafast carrier
dynamics in microcrystalline silicon probed by time-resolved terahertz spectroscopy," Physical Review B, 79,
115306 (2009).
[3] Taylor, Z. D., Singh, R. S., Bennett, D. B., Tewari, P., Kealey, C. P., Bajwa, N., Culjat, M. O., Stojadinovic, A.,
Hua, L., Hubschman, J. P., Brown, E. R., and Grundfest, W. S., "THz medical imaging: in vivo hydration
sensing," IEEE Transactions On Terahertz Science and Technology, vol. 1, no. 1, 201-219 (2011).
[4] Kippenberg, T. J., Holzwarth, R., and Diddams, S. A., “Microresonator-based optical frequency combs,”
Science, vol. 332, no. 6029, 555-559 (2011).
[5] Komiyama, S., “Single-photon detectors in the terahertz range,” IEEE Journal of Selected Topics in Quantum
Electronics, vol. 17, no. 1, 54-66 (2011).
[6] Cummer, S. A., Popa, B. I., Schurig, D., Smith, D. R., and Pendry, J. B., “Full-wave simulations of
electromagnetic cloaking structures,” Physical Review E, vol. 74, 036621, (2006).
[7] Fang, N., Lee, H., Sun, C., and Zhang, X., “Sub-diffraction-limited optical imaging with a silver superlens,”
Science, vol. 308, 534-537, (2005).
[8] Tao, H., Landy, N. I., Bingham, C. M., Zhang, X., Averitt, R. D., and Padilla, W. J., “A metamaterial absorber
for the terahertz regime: design, fabrication, and characterization,” Optics Express, vol. 16, 7181-7188, (2008).
[9] Tao, H., Kadlec, E. A., Strikwerda, A. C., Fan, K., Padilla, W. J., Averitt, R. D., Shaner, E. A., and Zhang, X.,
“Microwave and terahertz wave sensing with metamaterials,” Optics Express, vol. 19, no. 22, 21620-21626,
(2011).
[10] Butler, L., Wilbert, D., Baughman E., Balci, S., Kung, P., Kim, S.M., Heimbeck, M.S., and Everitt H.O.,
“Design, Simulation, and Characterization of THz Metamaterial Absorber,” in Terahertz Physics, Devices, and
Systems VI: Advance Applications in Industry and Defense, eds SPIE Proceedings Series, vol. 8363
83630J.(2012)

Proc. of SPIE Vol. 8585 85850Y-6

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 06/18/2015 Terms of Use: http://spiedl.org/terms