Taking whispering gallery-mode single virus detection and sizing to the limitV. R. Dantham, S. Holler, V. Kolchenko, Z. Wan, and S. Arnold Citation: Appl. Phys. Lett. 101, 043704 (2012); doi: 10.1063/1.4739473 View online: http://dx.doi.org/10.1063/1.4739473 View Table of Contents: http://apl.aip.org/resource/1/APPLAB/v101/i4 Published by the American Institute of Physics. Related ArticlesSingle particle demultiplexer based on domain wall conduits Appl. Phys. Lett. 101, 142405 (2012) Studies on CdS nanoparticles prepared in DNA and bovine serum albumin based biotemplates J. Appl. Phys. 112, 064704 (2012) Effect of molecule-particle binding on the reduction in the mixed-frequency alternating current magneticsusceptibility of magnetic bio-reagents J. Appl. Phys. 112, 024704 (2012) Field dependent transition to the non-linear regime in magnetic hyperthermia experiments: Comparison betweenmaghemite, copper, zinc, nickel and cobalt ferrite nanoparticles of similar sizes AIP Advances 2, 032120 (2012) Optofluidics incorporating actively controlled micro- and nano-particles Biomicrofluidics 6, 031501 (2012) Additional information on Appl. Phys. Lett.Journal Homepage: http://apl.aip.org/ Journal Information: http://apl.aip.org/about/about_the_journal Top downloads: http://apl.aip.org/features/most_downloaded Information for Authors: http://apl.aip.org/authors

Downloaded 09 Oct 2012 to 207.237.6.152. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions

Taking whispering gallery-mode single virus detection and sizing to the limit

V. R. Dantham,1 S. Holler,1,2 V. Kolchenko,3 Z. Wan,4 and S. Arnold1,a)

1Microparticle Photophysics Lab, Polytechnic Institute of NYU, Brooklyn, New York 11201, USA2Department of Physics, Fordham University, Bronx, New York 10458, USA3Department of Biological Sciences, NYC College of Technology, Brooklyn, New York 11201, USA4Department of Physics, Hunter College of CUNY, New York, New York 10065, USA

(Received 17 May 2012; accepted 12 July 2012; published online 27 July 2012)

We report the label-free detection and sizing by a microcavity of the smallest individual RNA virus,

MS2, with a mass only �1% of InfluenzaA (6 vs. 512 ag). Although detection of such a small

bio-nano-particle has been beyond the reach of a bare spherical microcavity, it was accomplished

with ease (S/N¼ 8, Q¼ 4� 105) using a single dipole stimulated plasmonic-nanoshell as a

microcavity wavelength shift enhancer, providing an enhancement of �70�, in agreement with

theory. Unique wavelength shift statistics are recorded consistent with an ultra-uniform genetically

programmed substance that is drawn to the plasmonic hot spots by light-forces. VC 2012 AmericanInstitute of Physics. [http://dx.doi.org/10.1063/1.4739473]

Early detection of virus at ultra-low concentration is im-

portant in identification and elimination of pathogens.

Recently, spherical optical microcavities have succeeded in

detecting individual InfluenzaA virions and quantifying their

size and mass [512 ag]. The signal/noise (S/N) ratio in those

experiments would have precluded the detection of single

viruses such as polio (14 ag) or the smallest RNA virus MS2

with a mass only �1% of InfluenzaA (Ref. 1) [6 vs. 512 ag].

Herein, we report the label-free detection and sizing by a

microcavity of MS2.2 Although detection of such a small

bio-nano-particle was beyond the reach of a bare micro-

spherical whispering gallery mode (WGM) biosensor,3 it is

accomplished with ease (S/N¼ 8) using a WGM-nanoplas-

monic-hybrid (WGM-h) composed of a spherical dielectric

microcavity with a nanoplasmonic receptor at the equator

(i.e., center of the WGM ring of light).4 Unlike our earlier

work on plasmonic enhancement that used the quadrupole

mode of a nanoshell at 633 nm to sense a polystyrene (PS)

nanoparticle, and only demonstrated a wavelength shift

enhancement of �4�, the current experiments excite the

plasmonic dipole by moving the excitation into the near

infrared (780 nm), and demonstrate a much larger overall

enhancement of �70�. This allows the MS2 virus to be

detected with a modest hybrid mode Q of only 4� 105. In

what follows we will (1) describe an analytical theory for the

dipole plasmonic enhancement by a nanoshell, (2) present

our experimental results, and (3) interpret these results.

The resonance wavelength shift of a dielectric WGM

resonator due to nanoparticle entering the evanescent field

Dkr was described earlier.5 It comes down to a simple result

of 1st order perturbation theory—the fractional wavelength

shift is equal to the cycle-averaged work required for the

local field to polarize the nanoparticle Wp divided by the

cycle averaged energy within the cavity Wc; Dkr=kr

¼ Wp=Wc. This so called “reactive sensing principle (RSP)”

has been tested extensively and correctly predicts the wave-

length shift for a nanoparticle of a given polarizability

adsorbing to a microcavity’s equator. For a Rayleigh particle,

Wp may be written as Wp ¼ ð1=4ÞRe½aex� jE0ðrvÞj2, where

aex is the polarizability in excess of the surrounding medium,

and E0ðrvÞ is the amplitude of the evanescent field at the posi-

tion of the analyte. The wavelength shift generated by the

RSP is proportional to polarizability which in turn is propor-

tional to mass. Our interest is principally in detecting individ-

ual virus with masses so small that a bare cavity produces a

wavelength shift below the noise.3 For a microspherical cav-

ity, the smallest virus currently detected is InfluenzaA with a

mass of 512 ag.6 The S/N ratio for these experiments was 3,

implying that the limit of detection would be �170 ag. The

detection of MS2 with a mass of only 6 ag will require an

enhancement of >30�. The RSP indicates that enhancing the

intensity jE0ðrvÞj2 at the adsorption site relative to overall

energy in the cavity should generate a proportional enhance-

ment in wavelength shift. A dipolar plasmonic mode of a

nanoshell receptor with an inner silica core (radius r1 and

dielectric constant Es) and an outer shell of gold extending to

radius r2 has a peculiar but useful properties in this regard.

These properties are most easily revealed by using Rayleigh

theory, where simple analytical results point to the spectral

region in which the enhancement is optimized.

A Rayleigh sized gold nanoshell excited into a dipole

plasmon resonance develops an intensity at its hot spots

jEhsj2 in excess of the intensity jE0j2 that stimulates it. For

the lowest order transverse electric (TE), WGM interacting

with a Rayleigh sized nanoshell on the equator the evanes-

cent wave polarizes the shell allowing the field at the hotspot

to be written as the sum of the driving field E0 (modeled as a

plane wave) and an induced field Eind from the nanoshell.

The maximum intensity enhancement on one of the hotspots

north or south of the equator RE;max is easily evaluated from

a quasi-electrostatic model7

RE;max ¼jEhsj2

E20

¼ E0 þ EindðrhsÞE0

��������2

¼ 3 Eg

Eg þ 2Eeg

��������2

; (1)

where Eg and Ee are the dielectric constants of the gold shell

and environment (i.e., water), respectively. The easiest way

a)Author to whom correspondence should be addressed. Electronic mail:

sarnold935@aol.com.

0003-6951/2012/101(4)/043704/4/$30.00 VC 2012 American Institute of Physics101, 043704-1

APPLIED PHYSICS LETTERS 101, 043704 (2012)

Downloaded 09 Oct 2012 to 207.237.6.152. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions

to understand this enhancement is to examine the denomina-

tor in the rightmost expression of Eq. (1). The plasmonic

dipole resonance occurs when the real part of this denomina-

tor goes to zero

Re½Eg� ¼ �2Ee Re½g�; (2)

where

Re½g� ¼ ReEsPþ Egð3� PÞ

Esð3� 2PÞ þ 2EgP

� �with P ¼ 1� r1

r2

� �3

:

For a solid gold particle for which r1¼ 0, Re[g]¼ 1, provid-

ing the well known resonance condition Re½Eg� ¼ �2Ee for

which resonance occurs in water at �540 nm and the

enhancement is j3Eg=Im½Eg�j2 �37, limited by the Im½Eg�.4At longer wavelengths such as 770 nm, Im[Eg] is much

smaller, but unfortunately the solid gold nanosphere’s reso-

nance cannot reach this wavelength. This is where a core

shell structure “out shines” the solid gold nanosphere. With

ðr1=r2Þ¼ 0.85 and for a silica core, Eq. (2) gives

Re[g]¼ 4.9, for which the dipole resonance shifts to 770 nm,

and enhancement grows to�340. Fig. 1 shows enhancement

spectra calculated from Eqs. (1) and (2) for a 50 nm diameter

core shell structures having a variety of shell thicknesses.

Although the wavelength shift enhancement of the hybrid

mode will be depleted somewhat by spatial decay of the

local field across the adsorbed virus, the interaction with a

dipole mode minimizes this effect by having an intensity

which drops off considerably slower than all other plasmonic

multipoles. As a result, the wavelength shift enhancement

for MS2 is found to be diminished by about a factor of 3

from the surface intensity enhancement, which easily

exceeds our goal of 30�.

All experiments were carried out in a poly-dimethyl silox-

ane (PDMS)-glass microfluidic cell3 held at 25.00 6 0.01 �C.

Contained within were a silica microsphere (radius R� 45 lm)

and a tapered fiber for evanescently coupling light into a

WGM of the microsphere as shown in Fig. 2. The laser driving

the guided wave in the fiber was an isolated distributed feed-

back laser (DFB) oscillating near 780 nm and tuned in wave-

length using a saw tooth current supply. The wavelength of a

WGM signature dip in transmission through the fiber at kr was

detected by a photodiode and measured using a LabVIEW run

multi-point parabolic fit. In what follows we describe the

attachment of the plasmonic structure to the microsphere reso-

nator, and the preparation of the MS2 virus.

Assembling the WGM-h was accomplished by using

carousel light forces8 generated by the WGM of the bare

microsphere (inset in Fig. 2).4 In particular, the gradient

force draws a gold nanoshell (r1� 60 nm, t� 12 nm from

Nanospectra Biosciences) to the equator of a silica micro-

sphere.4 Since both the silica microsphere and gold nano-

shells have negative charge on their surfaces, nanoshells are

repelled from the microsphere’s surface unless there are

strong WGM gradient forces.8 To aid these forces salt is

added (NaCl at 20 mM) to increase the conductivity of the

solution and thereby reduce the range of the electrostatic

fields emanating from the silica-water and gold-water inter-

faces. Attachment of a single nanoshell, likely at a silica

defect site,9 was verified for all of the WGM-h assemblies

through light scattering and a simultaneous shift in resonance

wavelength. This was followed within seconds by washing

for 5 min with DI water to eliminate all suspended nanoshells

from solution. During this period, the laser was turned off.

Attachment of only a single shell was facilitated by working

with a small shell concentration of 1.3 fM and carefully tim-

ing the washing. In each assembly, the nanoshell was veri-

fied to remain attached in the same apparent place without

the aid of the optical gradient force. Next, we will describe

the preparation of the MS2 virus and its injection into the

microfluidic cell.

MS2 was prepared by inoculating E. coli bacteria with

viable virus stock [American Type Culture Collection

(ATCC), Manassas, VA], harvesting the amplified number

of viruses, and dispersing them in saline solution. The result-

ing solution was tested by dynamic light scattering and found

to show no evidence of clustering. Since MS2 is genetically

programmed, it is perfectly uniform in size compared with

artificial nanoparticles such as PS or gold.

After identifying a mode of the WGM-h from a dip in

transmission through the fiber, MS2 viruses were then

injected so that their resulting concentration in the cell was

330 fM, with the cell brought to a salt concentration of

60 mM; again to aid adsorption.

FIG. 1. Enhancement in the WGM-hybrid wavelength shift calculated

quasi-statically for an infinitesimally small particle binding at T on a gold

nanoshell 50 nm in diameter having thicknesses ranging from the full radius

(i.e., solid gold) down to 3.75 nm.

FIG. 2. Microfludic WGM biosensor with the image of an assembled

WGM-h resonator.

043704-2 Dantham et al. Appl. Phys. Lett. 101, 043704 (2012)

Downloaded 09 Oct 2012 to 207.237.6.152. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions

Before recording dip trace characteristics of the WGM-h,

we adsorbed virus onto a bare resonator. A wavelength shift

binding curve was recorded with characteristic of nonspecific

adsorption with no detectable steps.3 Next, we carried out the

same experiment with the bare resonator replaced by a WGM-

h. A portion of a typical dip trace is displayed by the upper

trace (black) in Fig. 3(a). In this figure, we can see clear steps

due to adsorption of individual MS2 virus on the nanoshell.

The underlying slope matched that of adsorption by a bare

silica resonator at the same time following injection within

the experimental noise. This clearly shows the ease with

which one can distinguish adsorption on plasmonic particle

from its silica substrate. The plasmonic enhancement is

entirely responsible for this contrast; adsorption of individual

MS2 virus at the equator of the bare resonator of the same

radius produces a theoretical shift �0.25 fm, which is well

below the r.m.s. background noise of 2 fm (lower trace). The

16 fm wavelength shift step shown near the center of the upper

trace represents an enhancement of �64�. In this particular

experiment, 28 total steps were recorded after the first over an

interval of 3200 s. The largest of these was 17 fm for an

enhancement of 68�. The experiment was repeated with dif-

ferent WGM-h assemblies with similar results.

In Fig. 3(b), the 28 events are represented by a histo-

gram. This histogram bares little resemblance to experiments

with artificial nanoparticles or InfluenzaA virus. When light

forces draw nanoparticles to the equator of a bare cavity we

expect the distribution of wavelength shifts to reflect the dis-

tribution of mass, which is typically Gaussian (i.e., bell

shaped) for artificial nanoparticles. This is also the case for

InfluenzaA, since the cell membrane coat on the virus’ cap-

sid can vary from one virus to another. The histogram in Fig.

3(b) simply shows more events as the wavelength shift is

reduced followed by non-symmetric termination for smaller

resonance shifts. We believe this is because MS2 is pro-

grammed by its viral genome to be essentially uniform, and

because all of the 28 events are on the same nanoshell. The

increase in the number of events having smaller shifts occurs

because only one virus can occupy each of the two hot spot

centers (e.g., T in Fig. 1), and as the polar angle h increases

along the plasmonic sphere the geodesic parallels increase in

circumference as the sinðhÞ, thereby enabling more particles

to fit at larger angles.

In addition, increasing polar angle for a dipole mode

leads to a lower surface intensity and therefore a smaller

shift. Where the light force becomes too small to compete

for adsorption events the distribution terminates. The differ-

ence between this behavior and a Gaussian is clear, distinct,

and associated with light induced adsorption10 of an ultra-

uniform virus on a single spherical nanoshell.

A key question is whether the size of the virus can be

extracted from the measured signal. The answer will be af-

firmative for the largest wavelength shift. However, there are

some limitations to obtaining a simple analytic expression

imposed by using commercial nanoshells. Unfortunately, a

shell having an outer circumference to wavelength ratio of

2pr2=k ¼ 2pð71:5 nmÞ=780 nm ¼ 0:58 can be treated as a

Rayleigh particle only in a gross approximation. This led us

to simulate the intensity using the finite element method

(FEM, COMSOL). Fig. 4(a) shows the result at 780 nm for a

shell with a core radius of 60 nm and a shell thickness of

11.5 nm having a virus (refractive index¼ 1.5, typical of vi-

rus) of radius a¼ 12.5 nm on its point of highest intensity.

The dipole pattern is apparent along with some distortion

caused by the electromagnetic wave changing phase across

the structure. This causes the highest intensity point to be

slightly in the forward direction. The enhancement we are

interested in is the wavelength shift for the virus seated on

the gold hot spot to that on the silica equator Dkg=Dks which

from the RSP is the polarization energy of the virus in con-

tact with the nanoshell in comparison with that on the silica

equator. This ratio requires integration over the body of the

virus.

We have modeled this enhancement as a function of the

virus radius a using FEM and found an analytical expression

using the shell parameters in Fig. 4(a)

FIG. 3. (a) Shift of resonance wavelength above 780.674 nm of a WGM res-

onator R¼ 45 lm having a gold nanoshell attached at its equator due to the

adsorption of MS2 viruses (upper trace). The lower trace shows the back-

ground without MS2 or the gold nanoshell (r.m.s. noise 2 fm). Insets show

the recorded spectrum SD for the hybrid resonator (Q� 4� 105) and an illus-

tration of MS2 virus (radius a� 13.6 nm). (b) Step number statistics for all

of the steps recorded over 3000 s.

043704-3 Dantham et al. Appl. Phys. Lett. 101, 043704 (2012)

Downloaded 09 Oct 2012 to 207.237.6.152. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions

Dkg=Dks ¼ A½r2=ðr2 þ faÞ�6; (3)

with A¼ 155.56 and f¼ 0.76. The wavelength shift ratio

reaches 155.56 for an infinitesimally small adsorbate and

falls rapidly as the adsorbate increases in size [Fig. 4(b)] due

to the attenuation of the near field with range. Eq. (3) is the

key to finding the size of the virus from the experimental

wavelength shift Dkg but requires working out Dks. Fortu-

nately, the latter has been worked out earlier and is propor-

tional to the adsorbate volume; Dks / a3.6 With the

appropriate expression from Ref. 6 incorporated into Eq. (3),

we arrived at a solution for the virus size

a � G

1� 2Gfr2

� � ; where G ¼ R5=6

D1=3A1=3k1=6r

ðDkgÞ1=3; (4)

R is the microsphere radius, and the dimensionless parameter

D which includes the dielectric properties of the virus,

microsphere and water has a value6 of 1.50. For the largest

measured shift of 17 fm, we obtain a viral radius of 13.3 nm

which is in excellent agreement with a neutron diffraction

analysis of MS2 for which the reported outer radius is

13.6 6 1.0 nm.2

In this paper, we have demonstrated single particle

detection of the smallest RNA virus by using a microcavity-

hybrid having a modest Q factor (4� 105), and used the

RSP to determine its size from the largest wavelength shift

signal. Detection was possible because of an amplified wave-

length shift signal (17 fm) due to a nanoshell plasmonic

dipole enhancement of �70�. This amplification allowed

detection in the presence of r.m.s. noise of 2 fm. As a conse-

quence of this noise level, our limit of detection from Eq. (4)

is a � 5:7 nm (0.4 ag), below the size of all known viruses.

By contrast with our wavelength shift of 17 fm for MS2,

the detection of higher refractive index polystyrene particles

of comparable size on a ultra-high Q microtoroid (�108)

only produced a signal of 0.4 fm, and required a noise reduc-

ing reference interferometer to lower the r.m.s. noise to

0.2 fm.11 Combining the two advances of plasmon enhance-

ment and reference interferometry should prove very power-

ful for the next challenge, label-free single protein detection.

We have estimated the limit of detection using Eq. (4) for

our experimental setup by adding a reference interferometer

of our own and it is found to be a� 2 nm, a dimension below

that of a typical protein.

Further sensitivity can be gained by using plasmonic

structures with additional lightning rod enhancements, such

as nano-rods.12 We have merely scratched the surface of

what is likely to be possible.

The authors thank the NSF for supporting this work

(Grant No. CBET 0933531). We also wish to thank N. L.

Goddard of Hunter College for making her student Z. Wan

available for preparing the MS2 virus.

1C. B. Reimer, R. S. Baker, T. E. Newlin, and M. L. Havens, Science 152,

1379 (1966).2D. A. Kuzmanovic, I. Elashvili, C. Wick, C. O’Connell, and S. Krueger,

Structure 11, 1339 (2003).3S. Arnold, R. Ramjit, D. Keng, V. Kolchenko, and I. Teraoka, Faraday

Discuss. 137, 65 (2008).4S. I. Shopova, R. Rajmangal, S. Holler, and S. Arnold, Appl. Phys. Lett.

98, 243104 (2011).5S. Arnold, M. Khoshsima, I. Teraoka, S. Holler, and F. Vollmer, Opt. Lett.

28, 272 (2003).6F. Vollmer, S. Arnold, and D. Keng, Proc. Natl. Acad. Sci. U.S.A. 105,

20701 (2008).7R. D. Averitt, S. L. Westcott, and N. J. Halas, J. Opt. Soc. Am. B 16, 1824

(1999).8S. Arnold, D. Keng, S. I. Shopova, S. Holler, W. Zurawsky, and F.

Vollmer, Opt. Express 17, 6230 (2009).9U. Martinez, J. F. Jerratsch, N. Nilius, L. Giordano, G. Pacchioni, and

H.-J. Freund, Phys. Rev. Lett. 103, 056801 (2009).10L. Novotny, R. X. Bian, and X. S. Xie, Phys. Rev. Lett. 79, 645 (1997).11T. Lu, H. Lee, T. Chen, S. Herchak, J.-H. Kim, S. E. Fraser, R. C. Flagan,

and K. Vahala, Proc. Natl. Acad. Sci. U.S.A. 108, 5976 (2011).12J. D. Swaim, J. Knittel, and W. P. Bowen, Appl. Phys. Lett. 99, 243109

(2011).

FIG. 4. (a) FEM simulation of the park-

ing of MS2 virus at one of the dipole

lobes of a plasmonic nanoshell with

60 nm inner core radius of silica and

11.5 nm shell thickness of Au. The field

intensity at the point of contact grows to

just over 252� the field of the 780 nm

TE mode as indicated by the rainbow

scale on the left. (b) shows the enhance-

ment in the wavelength shift calculated

from FEM for different a values, fitted

with an analytical expression (Eq. (3)).

043704-4 Dantham et al. Appl. Phys. Lett. 101, 043704 (2012)

Downloaded 09 Oct 2012 to 207.237.6.152. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions