April 1, 2009 / Vol. 34, No. 7 / OPTICS LETTERS 1087

Nanocavity plasmonic device for ultrabroadbandsingle molecule sensing

Ryan M. Gelfand, Lukas Bruderer, and Hooman Mohseni*Electrical Engineering and Computer Science, Northwestern University, 2145 Sheridan Road,

Evanston, Illinois 60208, USA*Corresponding author: hmohseni@northwestern.edu

Received October 17, 2008; revised January 26, 2009; accepted February 27, 2009;posted March 5, 2009 (Doc. ID 102848); published March 30, 2009

We present a new structure that combines a metal–dielectric–metal sandwich with a periodic structure toform a plasmon polariton photonic crystal. Three-dimensional finite-difference time-domain simulationsshow a clear bandgap in the terahertz regime. We exploited this property by adding a defect to the crystal,which produces a cavity with a quality factor of 23.3 at a wavelength of 3.45 �m. Despite the small Q factor,the ultrasmall sensing volume of 15 zeptoliters produces an extremely large Purcell constant of 4.8�106.Compared to photonic crystals with similar Purcell constant, the bandwidth is several orders of magnitudelarger, or about 7 THz, ensuring high tolerances to manufacturing parameters, and environmental changes,as well as a high specificity owing to the possibility of broadband spectral fingerprint detection. © 2009 Op-tical Society of America

OCIS codes: 250.5403, 160.5298, 040.2235, 140.4780.

Label-free single molecule detection is an exciting av-enue of study that will impact many areas of ourlives, from explosives detection [1] and security, [2] tohelping us discover new drugs [3], to biosensing [4],and detecting cancer in it infant stages [5,6]. To ac-complish such a technique, a device must first be fab-ricated that has high sensitivity and accurate speci-ficity. Optical spectroscopy can potentially addressthese issues, as well as provide remote sensing capa-bilities and a fast detection time. Therefore, photonicsensing has become an attractive method for singlemolecule detection [7]. One of the main issues ofsensing larger molecules is that such molecules havea vibrational signature in the mid- and long-IRrange; and to have a strong interaction between themand light with such long wavelengths, the light mustsomehow be compressed much below its diffractionlimit. A promising structure that can squeeze lightsignificantly in one dimension is a metal–dielectric–metal (MDM) sandwich with a very thin dielectriclayer [8,9], although with no interaction point orresonance effect these structures by themselveswould not make for excellent sensors. Photonic crys-tals (PCs) have many attractive properties, includingthe ability to dramatically increase the interactionbetween a small volume and light; however, they relyon extremely high quality �Q� factors to achieve ahigh interaction coefficient. This high Q factor under-scores the need for very narrow spectral linewidth,which in turn reduces the robustness of the deviceand makes it impossible to obtain enough of a mol-ecule’s spectral fingerprint, preventing reliable detec-tion and high specificity. Also, owing to nanometerlimited geometrical tolerances, it is hard to reproducePCs that match the spectral feature of interest. Simi-larly tight tolerances to ambient (temperature, con-centration, humidity, etc.) changes render these de-vices difficult to use in the field.

Besides PCs, another method currently beinglooked at as a contender for single molecule detection

0146-9592/09/071087-3/$15.00 ©

is a functionalized metal nanostructured array [10].These devices are shown to be robust and have a va-riety of ways of being fabricated [11,12]. By squeezinglight into a tight space on the surface of a metal,these plasmonic structures have been used to en-hance Raman spectroscopy. However, like a PC, thesedevices derive their enhanced sensitivity from a tightspectral linewidth. In contrast, our device attainsmoderately high field intensity enhancements of 104

comparable to typical values for plasmonic nano-structured arrays [13], though sophisticated theoret-ical structures can attain values as high as 107 at theexpense of having extremely narrow bandwidth.

The presented device combines MDM structuresand a two-dimensional cavity [see Fig. 1(a)] to pro-duce a practical structure that addresses many is-sues plaguing the aforementioned sensors. We showthat our new structure can squeeze light a milliontimes volumetrically, leading to a high Purcell con-stant comparable to the best reported values by con-ventional PCs [Fig. 1(b)]. However, the small Q factorof the proposed structure leads to a very wide spec-tral bandwidth, producing robustness and betterspecificity as a photonic molecular sensor, while atthe same time we can keep the interaction volumesmall and thus the Purcell factor high. We also showthat the main plasmonic mode of the proposed struc-ture has a good coupling to the far field, and hencecan be excited by conventional optical devices (e.g.,an optical microscope) [Fig. 1(c)].

Before analyzing the nanocavity device, we firsthad to find its fundamental mode and general photo-nic band structure. In designing the device we real-ized that by squeezing light we could increase the ef-fective index of the material, thus slowing it down[14]. However by decreasing the thickness of the di-electric layer, the percent of field energy in the metalincreases, thus increasing the loss and decreasingthe lifetime. Therefore, higher index dielectrics can

increase effective index without a significant reduc-

2009 Optical Society of America

1088 OPTICS LETTERS / Vol. 34, No. 7 / April 1, 2009

tion of the lifetime [see Fig. 2(b)]. Therefore we chosea dielectric thickness of 12 nm, although layers asthin as 3 nm have been shown [15], and a high dielec-tric constant of 9. The gold metal cladding layerswere each 200 nm thick, many times larger than theskin depth, so as to minimize the energy loss to theupper surface.

When two metal dielectric boundaries are broughtclose together the modes on each interface start to in-teract and get coupled. Even at very long wave-lengths (away from the plasma frequency) the modedispersion deviates strongly from the light line forvery small gap thicknesses. Depending on the rela-tive phase of the two waves, a symmetric or antisym-metric mode is formed. This research focuses on theantisymmetric mode because of its superior charac-teristics in propagation distance and dispersion [16].Bloch boundaries in each lateral direction and anantisymmetric boundary in the z direction reduce thefinite-difference time-domain (FDTD) simulationarea. The full band plot is shown in [Fig. 2(a)]. Theregion of the crystal in which we are interested is thebandgap, between 86.6 and 112 THz, or 2.67 to3.46 �m, and because we are working away from theplasmon frequency where the dispersion is flat, eachfrequency component of any mode launched in thecrystal will effectively experience almost the same in-dex [Fig. 2(c)]. This effect is important for sensingmolecules where a wide spectral linewidth is prefer-able.

To expand this design as a sensor one needs a de-tection site, and nanocavities are a natural solution.For our plasmon polariton PC (PPPC) with a defect, a3D FDTD simulation was used to calculate the wave-

Fig. 1. (Color online) (a) Plasmon polariton photonic crys-tal with 200 nm metal cladding layers sandwiching a12 nm dielectric layer pierced with an hexagonal lattice ofholes, 450 nm in diameter, with a period of 500 nm and a40 nm diameter central cavity defect hole. (b) Ez mode pro-file �87 THz� of the center of the cavity at z=0 nm; themode is intensely squeezed in and around the defect cavity.(c) Ez intensity along the longitudinal direction of the crys-tal from the center of the cavity to 200 nm above the crys-tal. (d) Ez intensity at 500 nm above the top of the crystalshows that a small amount of light emanates from the crys-tal through the central column.

length of the resonant modes and the cavity’s Q fac-

tor. We built our cavity by placing a 40 nm defect holein a PPPC with a hexagonal array of 450 nm holesand a period of 500 nm. The time evolution of thefield is analyzed by first plotting the natural log ofthe Ez field versus time [Fig. 3(a)]. The peaks aredetected and a line is fitted with decay time

Fig. 2. (Color online) (a) 3D FDTD band plot simulation ofthe PPPC, hole diameter of 450 nm, period of 500 nm,shows a clear bandgap between 86.6 and 112 THz. (b)Graph of the decay time and effective index at 87 THz forvarious starting permittivities as a function of dielectricthickness. As the thickness of the material decreases the ef-fective index increases, however, the decay time is nearlyindependent of thickness, so we can choose a thin layerhigh dielectric material to optimize our device with mini-mal sacrifice to the lifetime of the photons. (c) Effective in-dex of the MDM slab mode as a function of frequency at athickness of 12 nm. Away from the plasma frequency thedispersion is flat, so the optical properties of the crystal donot significantly change in our area of interest.

Fig. 3. (Color online) (a) Cavity ringdown plot for the fun-damental mode, 87 THz, of the crystal showing both thesource decay and the mode decay. With a decay time of42.6 fs the quality factor for the cavity is 23.3. (b) FFT ofthe fundamental cavity mode showing a spectral line widthat FWHM of 7.4 THz. (c) Flat field profile versus timeshows a gain of �5500 cm−1 in the dielectric layer that canalmost compensate for the total loss, material loss, and

leakage of the structure.

April 1, 2009 / Vol. 34, No. 7 / OPTICS LETTERS 1089

�=−1/slope. The Q factor is the angular frequencymultiplied by the decay time, Q=�*�.

For any optical detection technique to be practicalfor single molecule detection, the interaction time be-tween the light and the target molecule needs to belong and the spectral linewidth needs to be broadenough to allow for detection of the spectral signa-tures of the molecules being detected. By exploitingthe properties of nanoscale cavities, we have beenable to design and simulate a PPPC that can squeezelight in all three dimensions into a small volume fa-cilitating a larger interaction with single molecules.This interaction strength between the light and thecavity can be characterized by Purcell’s constant,

p =3

4�2

Q

V� �3, �1�

where Q is the quality factor of the cavity, V is its vol-ume (1.5�10−23 m3, or a sensing volume of 15 zepto-liters), and � is its resonant mode �3.45�10−6 m�.

A high p, implying an intense interaction betweenthe light inside the cavity and electronic states of themolecules, would be required for any efficient opticalsensor [17]. Furthermore a wide spectral linewidth�f is necessary for the operation and robustness ofany single molecule sensor to ensure detection withinslightly varying conditions such as temperature andcavity geometries. Since �f= f0 /Q, a smaller Q relatesto a wider linewidth. This implies that one needs toincrease Purcell’s constant without relying on ex-tremely high Q factors to build a practical sensor. Tomeet these two conditions, V the volume of the cavitymust be as small as possible.

Our simulation shows a Q factor of 23.3 [Fig. 3(a)]with the fundamental resonant frequency [Fig. 3(b)]of 87 THz. The best PCs at a wavelength of 1.575 �mexperimentally produce p values of 1.9�105 with a Qof 4.5�104 [18]. Theoretical maximums are of the or-der of 106–107 for ultra-high-Q structures with atight linewidth of hundreds of magahertz [19]. Ourdevice at a wavelength of 3.45 �m shows a p value of4.8�106 with a �f of 7.4 THz showing that we cansqueeze a lot of light, with a broad spectral range,into a small volume. It is these two properties thatwould suggest this device to be an excellent candi-date for a single molecule sensor.

Analogously to the sensor this high Purcell con-stant cavity could be used as a laser if the mediuminside can compensate for the high loss introduced bythe proximity of the metal cladding layers to the di-electric layer. Simulation shows that a gain materialwith a gain value of 5500 cm−1 satisfies the lasingcondition, and a coherent stabilization of the electro-magnetic radiation inside the cavity [see Fig. 3(d)].Similar to surface plasmon [20] lasers this devicewould emit light in the terahertz frequency.

Another critical feature of this device is that it canbe easily coupled to from the outside, a property mostuseful for building a sensor or a laser. The cavity not

only exhibits very intense energy density inside, but

it also shows a column of energy emanating from thecenter up through the 40 nm diameter defect holeand stays subdiffraction at least 500 nm above thetop of the crystal [Fig. 1(d)]. So, theoretically it wouldbe possible to optically pump the microcavity by cou-pling back through the column thus enabling energyto build up inside. Since the column runs the fulllength of the device, one could pump from one sideand the cavity could emit through the other. This col-umn also allows the passage of media either gaseousor liquid to be passed through the cavity with its con-tents scanned for certain molecules.

We believe that the presented plasmonic nanocav-ity provides new opportunities to build single mol-ecule detectors and terahertz plasmonic lasers. Theapplications for plasmon-based sensors and lasersare clearly documented and the same tenets wouldapply to any PPPC-based device. Investigating thedevices based on plasmon polaritons that are capableof producing deep subdiffraction photonic integratedcircuits will help give us the versatility we require asdevices shrink ever more into the nanoscale world.

References

1. J. F. Federici, B. Schulkin, F. Huang, D. Gary, R.Barat, F. Oliveira, and D. Zimdars, Semicond. Sci.Technol. 20, S266 (2005).

2. B. Ferguson and X.-C. Zhang, Nature Mater. 1, 26(2002).

3. L. Ho, M. Pepper, and P. F. Taday, Nat. Photonics 20,541 (2008).

4. M. Tonouchi, Nat. Photonics 1, 97 (2007).5. R. M. Woodward, B. E. Cole, V. P. Wallace, R. J. Pye, D.

D. Arnone, E. H. Linfield, and M. Pepper, Phys. Med.Biol. 47, 3853 (2002).

6. S. Nakajima, H. Hoshina, M. Yamashita, and C. Otani,Appl. Phys. Lett. 90, 041102 (2007).

7. M. Dragoman and D. Dragoman, Prog. QuantumElectron. 32, 1 (2008).

8. Y. Kurokawa and H. T. Miyazaki, Phys. Rev. B 75,035411 (2007).

9. G. Veronis and S. Fan, Opt. Express 15, 1211 (2007).10. J. N. Anker, W. P. Hall, O. Lyandres, N. C. Shah, J.

Zhao, and R. P. Van Duyne, Nature Mater. 7, 442(2008).

11. K. Ait-Mansour, A. Buchsbaum, P. Ruffieux, M.Schmid, P. Gröning, P. Varga, R. Fasel, and O. Gröning,Nano Lett. 8, 2035 (2008).

12. J. Henzie, M. H. Lee, and T. W. Odom, Nat.Nanotechnol. 2, 549 (2007).

13. S. Zou and G. C. Schatz, Chem. Phys. Lett. 403, 62(2005).

14. A. P. Hibbins, J. R. Sambles, C. R. Lawrence, and J. R.Brown, Phys. Rev. Lett. 92, 143904 (2004).

15. H. T. Miyazaki and Y. Kurokawa, Phys. Rev. Lett. 96,067401 (2006).

16. J. D. Joannopoulos, S. G. Johnson, J. N. Winn, and R.D. Meade, Photonic Crystals: Molding the Flow of Light(Princeton, 2008).

17. T. F. Krauss, Nature Mater. 2, 777 (2003).18. Y. Akahane, T. Asano, B. Song, and S. Noda, Nature

425, 944 (2003).19. B. Song, S. Noda, T. Asano, and Y. Akahane, Nature

Mater. 4, 207 (2005).

20. N. I. Zheludev, Nat. Photonics 2, 351 (2008).