Polarization insensitive terahertz metamaterial absorberJ. Grant, Y. Ma, S. Saha, L. B. Lok, A. Khalid, and D. R. S. Cumming*

School of Electronics, University of Glasgow, Glasgow, G12 8LT, UK*Corresponding author: david.cumming.2@glasgow.ac.uk

Received November 2, 2010; revised March 3, 2011; accepted March 24, 2011;posted March 30, 2011 (Doc. ID 137604); published April 15, 2011

We present the simulation, implementation, and measurement of a polarization insensitive resonant metamaterialabsorber in the terahertz region. The device consists of a metal/dielectric-spacer/metal structure allowingus to maximize absorption by varying the dielectric material and thickness and, hence, the effective electricalpermittivity and magnetic permeability. Experimental absorption of 77% and 65% at 2:12THz (in the operatingfrequency range of terahertz quantum cascade lasers) is observed for a spacer of polyimide or silicon dioxiderespectively. These metamaterials are promising candidates as absorbing elements for thermally based terahertzimaging. © 2011 Optical Society of AmericaOCIS codes: 160.3918, 040.2235, 250.5403.

Since the first theoretical [1] and experimental demon-stration [2] of the unique properties of metamaterials(MMs), research into the topic has grown rapidly.MMs can provide a highly controllable electromagneticresponse in different frequency bands, and to date MMshave been demonstrated in every technologically rele-vant spectral range, including the optical [3], near-IR[4], mid-IR [5], terahertz [6], millimeter wave [7], micro-wave [8], and radio [9]. MMs have enabled investigationsinto new possibilities, including perfect lenses [10] andinvisibility cloaks [11]. A further aspect, however, ofMMs that is currently provoking wide interest is the topicof so-called MM perfect absorbers. By manipulating theeffective electrical permittivity, ε, and magnetic perme-ability, μ, absorption close to unity is possible [12].The concept of an MM absorber is especially importantat terahertz frequencies, where it is difficult to find strongfrequency selective terahertz absorbers. Such MM absor-bers naturally lend themselves to terahertz detection ap-plications, such as thermal sensors that, if integratedwith suitable terahertz sources (e.g., quantum cascadelasers), could lead to compact, highly sensitive, low-costterahertz imaging systems.MMs typically consist of arrays of subwavelength

metallic elements patterned onto a semi-insulating sub-strate, such as GaAs. Modifying the size and shape ofsubwavelength metallic elements permits frequencyand amplitude tuning of both the electric εðωÞ and mag-netic response μðωÞ of electromagnetic radiation. Theschematic of a single unit cell of our MM absorber isshown in Fig. 1(a) and the layer cross section is shownin Fig. 1(b). It differs from the standard MM configurationin that, instead of the MM structure being patterned in ametallic layer onto a highly resistivity substrate, thereare two metallic elements: one ground plane and across-shaped resonator separated by a dielectric layerof thickness t.The cross-shaped resonator is an example of an elec-

tric ring resonator (ERR) [13,14] and couples stronglyto uniform electric fields, but negligibly to magnetic ones.Earlier work using patterned bar and cross-shapedERRs over a ground plane demonstrated metamaterialresonant absorption [15,16]. In these devices, it washypothesized that Fabry–Perot modes were responsiblefor the strong absorption. Here we explore a device with

a looped cross-shaped geometry. By pairing the ERRwith a ground plane, the magnetic component of the in-cident terahertz wave induces a current in the sections ofthe ERR that are parallel to the direction of the electricfield [see Fig. 1(a)]. An antiparallel image current alsoflows in the region of the ground plane immediately be-low the cross, which results in a resonant response. Thedirection of current flow in the ERR is shown by the bluearrows in the cross in Fig. 1(a). This is a distinctive ab-sorption mechanism that does not rely on Fabry–Perotmodes that would not appear at frequencies of less than10THz in the devices shown. Significantly, the electricand magnetic response can then be tuned independentlyby varying the geometry of the ERR and the distancebetween the two metallic elements.

The optimized MM absorber structure in Fig. 1(a) wasobtained through finite-difference time-domain (FDTD)

Fig. 1. (Color online) (a) Schematic of the ERR of the MMabsorber and (b) cross section of the complete MM absorber.(c) SEM image of the unit cell and (inset) section of the array.(d) Absorption spectra for different incident polarization anglesshowing polarization insensitivity of the MM absorber. Eachsuccessive plot from 0°–90° is offset by one major unit ofthe ordinate axis.

1524 OPTICS LETTERS / Vol. 36, No. 8 / April 15, 2011

0146-9592/11/081524-03$15.00/0 © 2011 Optical Society of America

simulations (Lumerical Inc.). The three-dimensionalsimulations were performed with a plane wave sourceincident in the z direction on the metal/dielectric/metal/substrate unit cell. Periodic boundary conditionswere used for the x–y plane, along with a mesh step sizeof Δx ¼ Δy ¼ 0:2 μm and Δz ¼ 0:05 μm. The metallicsections of the absorber were modeled as Au with afrequency independent conductivity of 4 × 107 Sm−1.Reflection and transmission spectra were recorded atplanes 100 μm above and 100 μm below the ERR. TheFDTD simulations revealed the absorption characteris-tics of the MM absorber were not sensitive to the polar-ization angle of the incident electromagnetic wave [seeFig. 1(d)].Once the optimum MM absorber design had been

established, standard metal evaporation, spin coating,and electron-beam (e-beam) lithography techniques wereused to fabricate devices. First a 20=300 nm Ti/Au metal-lic ground plane was evaporated onto a silicon substrate.The dielectric layer was then deposited onto this metallicfilm either by spin coating a liquid polyimide (HD Micro-systems PI2545) or, in the case of SiO2, by a plasmaenhanced chemical vapor deposition technique. Thepolyimide thickness when spun at 4000 rpm onto a15mm × 15mm substrate was 2:5 μm. The thickness wasmodified by varying the spin speed/duration and by usingmultiple coatings. After spinning the required thicknessof polyimide, the layer was cured in a nitrogen-purgedoven for 2 h at 180 °C. The MM cross-shaped absorberwas defined in a bilayer of polymethyl methacrylate e-beam resist using a Vistec VB6 e-beam tool and a second20=300 nm film of Ti/Au was evaporated and lifted off. AllMM structures presented here had a repeat period of27 μm and dimensions K ¼ 26 μm, L ¼ 20 μm, M ¼10 μm, and N ¼ 5 μm. Scanning electron microscopy(SEM) images of an MM pixel and a section of the arrayare shown in Fig. 1(c).Samples were characterized under vacuum in a Bruker

IFS 66v/S Fourier transform infrared spectrometer intransmission mode at normal incidence and in reflectionmode at 30° incidence. The measured transmission spec-tra were normalized with respect to the signal measuredfrom a 7mm diameter open aperture and the reflectionspectra were normalized to that of a gold mirror. Theresulting absorption, A, was therefore calculated usingAðωÞ ¼ 1 − RðωÞ − TðωÞ, where R is the reflection andT the transmission. Experimental measurements werealso performed on samples with no ERR layer to confirmthat absorption was a consequence of the MM structureand not of the dielectric. The 7:5 μm thick polyimide sam-ple with no ERR structure had a maximum absorption of5% across the frequency range of interest [see Fig. 2(a)]thereby verifying that, at the resonance frequency,absorption was a result of the MM structure.The experimentally obtained transmission and absorp-

tion spectra, as well as the simulated data for a MMabsorber with a 3:1 μm thick polyimide dielectric spacer,are shown in Fig. 2(a). A refractive index of 1:68þ 0:06iwas used for the polyimide. The experimental transmis-sion is zero across the entire frequency range, while thepeak absorption was measured to be 77% at 2:12THz.This result is in excellent agreement with the simulatedabsorption maximum of 81%. Figure 2(b) shows the

experimental data for MM absorbers with polyimidethicknesses ranging from 1 to 7:5 μm and for an absorberwhere the dielectric is 3 μm of SiO2. As the polyimidethickness increases from 1 to 3:1 μm, the peak absorptionincreases, but at polyimide thicknesses greater than3:1 μm, there is a slight reduction in the peak absorptionvalue. A distinct redshift of 0:25THz is observed as thepolyimide thickness increases from 1 to 7:5 μm. Clearlythere is an optimum polyimide thickness where maxi-mum absorption is attained. Absorbers that had SiO2as the dielectric instead of polyimide were also studied.A maximum absorption value of 65% at 1:90THz wasmeasured for such an MM absorber with a 3 μm thickSiO2 dielectric layer.

The effective permittivity and permeability were ex-tracted from the simulated data via inversion of the Sparameters [17]. The retrieved parameters for the simu-lated MM absorber with a 3:1 μm thick polyimide spacerare displayed in Fig. 2(c). As can be observed, the realparts of the optical constants cross close to zero—acondition required for zero reflection, while, wheneverthe real part of the permittivity is positive, the real part

Fig. 2. (Color online) (a) Experimental and simulated data ofan MM absorber with a polyimide thickness of 3:1 μm. Thetransmission is zero across the entire frequency range. (b)Experimental absorption spectra for MMs with differing dielec-tric spacer thickness and type. (c) Extracted optical parametersfrom the simulated 3:1 μm thick polyimide MM absorber.

April 15, 2011 / Vol. 36, No. 8 / OPTICS LETTERS 1525

of the permeability is negative and vice versa—a con-dition required for zero transmission. At the frequencyof maximum absorption, ω0, there is a peak of the ima-ginary component of the permeability, implying highabsorption.The simulated power absorption distributions for the

ERR, dielectric, and the ground plane layers are shownin Figs. 3(a)–3(c), while a cross section of the powerdistribution in the x–z plane at y ¼ 3 μm is shown inFig. 3(d). From these plots, it is clear that the majorityof the energy is dissipated as ohmic loss in the ERR layerand as dielectric loss in the first 500 nm of polyimide be-low this layer. The regions of maximum absorption lossoccur between adjacent unit cells and around the inneredges of the cross. These simulations show concurrencewith our explanation for the operation of the device andthe unlikelihood that Fabry–Perot modes strongly affectbehavior.In conclusion, we have demonstrated a terahertz MM

absorber that is inherently a narrowband device withpotential applications in selective spectral imaging.

Furthermore, the MM absorber is polarization insensi-tive, as a result of the symmetry of the cross structure,making it ideal for several applications as it maximizesabsorption for arbitrarily polarized light. The designmay also be modified to make a wideband absorber usingmultiple unit cells [18] or by incorporating tunable orfrequency agile MM components. The MM absorbersmay be integrated with bolometric sensors, allowingsensitive terahertz detection.

References

1. J. B. Pendry, A. J. Holden, D. J. Robbins, and W. J. Stewart,IEEE Trans. Microwave Theory Tech. 47, 2075 (1999).

2. D. R. Smith, W. J. Padilla, D. C. Vier, S. C. Nemat-Nasser,and S. Schultz, Phys. Rev. Lett. 84, 4184 (2000).

3. G. Dolling, M. Wegener, C. M. Soukoulis, and S. Linden, Opt.Lett. 32, 53 (2007).

4. S. Zhang, W. J. Fan, N. C. Panoiu, K. J. Malloy, R. M. Osgood,and S. R. J. Brueck, Phys. Rev. Lett. 95, 137404 (2005).

5. S. Linden, C. Enkrich, M. Wegener, J. F. Zhou, T. Koschny,and C. M. Soukoulis, Science 306, 1351 (2004).

6. H. Tao, C. M. Bingham, A. C. Strikwerda, D. Pilon,D. Shrekenhamer, N. I. Landy, K. Fan, X. Zhang, W. J.Padilla, and R. D. Averitt, Phys. Rev. B 78, 241103 (2008).

7. M. Gokkavas, K. Guven, I. Bulu, K. Aydin, R. S. Penciu,M. Kafesaki, C. M. Soukoulis, and E. Ozbay, Phys. Rev. B73, 193103 (2006).

8. D. R. Smith and N. Kroll, Phys. Rev. Lett. 85, 2933 (2000).9. M. C. K. Wiltshire, J. B. Pendry, I. R. Young, D. J. Larkman,

D. J. Gilderdale, and J. V. Hajnal, Science 291, 849 (2001).10. N. Fang, H. Lee, C. Sun, and X. Zhang, Science 308,

534 (2005).11. D. Schurig, J. J. Mock, B. J. Justice, S. A. Cummer, J. B.

Pendry, A. F. Starr, and D. R. Smith, Science 314,977 (2006).

12. N. I. Landy, C. M. Bingham, T. Tyler, N. Jokerst, D. R. Smith,and W. J. Padilla, Phys. Rev. B 79, 125104 (2009).

13. D. Schurig, J. J. Mock, B. J. Justice, S. A. Cummer, J. B.Pendry, A. F. Starr, and D. R. Smith, Science 314, 977(2006).

14. W. J. Padilla, M. T. Aronsson, C. Highstrete, M. Lee, A. J.Taylor, and R. D. Averitt, Phys. Rev. B 75, 041102 (2007).

15. H. T. Chen, J. F. Zhou, J. F. O’Hara, F. Chen, A. K. Azad, andA. J. Taylor, Phys. Rev. Lett. 105, 073901 (2010).

16. D. Y. Shchegolkov, A. K. Azad, J. F. O’Hara, and E. I.Simakov, Phys. Rev. B 82, 205117 (2010).

17. D. R. Smith, D. C. Vier, T. Koschny, and C. M. Soukoulis,Phys. Rev. E 71, 036617 (2005).

18. C. M. Bingham, H. Tao, X. L. Liu, R. D. Averitt, X. Zhang, andW. J. Padilla, Opt. Express 16, 18565 (2008).

Fig. 3. (Color online) Energy dissipation in an MM absorberstructure with a 3:1 μm thick polyimide spacer at a frequencyof 2:12THz. Energy dissipation in (a) the ERR layers, (b) thedielectric spacer, (c) the ground plane, and (d) x–z plane aty ¼ 3 μm.

1526 OPTICS LETTERS / Vol. 36, No. 8 / April 15, 2011