Expand this Topic clickable element to expand a topic
Skip to content
Optica Publishing Group

Optical Fano resonances in a nonconcentric nanoshell

Open Access Open Access

Abstract

The interaction of light with a metal nanoshell with an off-center core generates multipoles of all orders. We show here that the matrix elements used to compute the multipole expansion coefficients can be derived analytically and, with this result, we can show explicitly how the dipole and quadrupole terms in the expansion are coupled and give rise to a Fano resonance. We also show that the off-center core significantly increases the electric field enhancement at the shell surface compared to the concentric case, which can be exploited for surface-enhanced sensing. The multipole solutions are confirmed with finite-element calculations.

© 2016 Optical Society of America

1. INTRODUCTION

In recent years, optically-induced Fano resonances in nanostructures have been proposed as the basis of biochemical sensing with potentially high sensitivity [1]. A Fano resonance arises from the interference between two resonant modes in a system with two or more interacting structures producing a type of resonant scattering phenomenon exhibiting an asymmetric line shape [2,3]. The sensitivity of the distortion or asymmetry in the spectrum to the local environment from this mode coupling has been suggested as the basis for sensing. Our goal in this report is to study a nanostructure in which a Fano resonance arises, but is sufficiently simple to treat analytically, which then helps clarify the key features of the Fano phenomenon. It is well known that a shell with a nonconcentric core generates multipoles of all orders, and the nonconcentric shell–core system has been analyzed using a multipole expansion of the fields generated by the off-center core [4]. In our study, we also employ a multipole analysis of the nonconcentric shell but show how the matrix elements used to compute the multipole expansion coefficients can be derived analytically to all orders with the aid of the solid-harmonic addition theorem. This analytical solution allows us to show explicitly how the two lowest order multipole terms, the dipole and quadrupole contributions, are coupled and display the classic Fano form. As the core moves further off center, the electric field grows in magnitude just outside the shell boundary near the thinner part of the shell. This field enhancement can be clearly seen in profile calculations of the electric field magnitude through the interior of the shell. As a check, we show that these results are in excellent agreement with finite-element calculations. Our calculations illustrate how the electric field enhancement increases with the core offset. The field enhancement outside the shell surface can be exploited in surface-enhanced spectroscopies, such as surface-enhanced Raman scattering (SERS) since the SERS response scales as the fourth power of the electric field magnitude [5]. A key feature that characterizes the Fano effect is the coupling between a radiant mode (or “bright mode”), which is dipolar and can be stimulated from, and observed in, the far field, and a nonradiant mode (or “dark mode”) that is invisible in the far field but couples to the radiant mode and interferes with the latter. It is this interference that gives rise to a distortion, or asymmetry, in the radiant mode and could, in the absence of damping, theoretically quench the radiant mode entirely at some frequency. In the presence of damping the result is a “Fano dip” in the spectrum rather than complete quenching. A second consequence of this interference is a shift in the peak of the radiant mode. It is primarily the sensitivity to the local environment of the location and size of the nanodip and the bright-mode frequency shift that suggests the possibility of exploiting the Fano phenomenon for sensing applications.

The paper is organized as follows. We first carry out the multipole expansion of the electric field for the nonconcentric nanoshell. The analysis shows that the multipole expansion coefficients are the solution to a linear system, which can be expressed in matrix form. One contribution of this report is to show how the matrix elements in this system can be derived analytically with the aid of the solid-harmonic addition theorem. This gives explicit expressions for the first two coupled multipoles, the dipole and quadrupole modes, which play the role, respectively, of the bright and dark modes in a system exhibiting Fano interference. The resulting expression can be placed in the classic Fano form, which shows how the Fano interference arises or, in particular, how the resonance peak of the system is shifted and how partial quenching (the “Fano dip”) occurs. A number of authors have noted that the classical analog of Fano interference can be demonstrated from an analysis of coupled oscillators [2,6,7]. We also show this by comparing our Fano expression for the off-center shell–core system with that of a system of four masses connected by springs. Finally, we compute electric field profiles through the interior of a nonconcentric shell using the full multipole expansion. This shows how the electric field enhancement on the shell surface increases with increasing core offset. As a check, we confirm the electric field calculations by comparing them to finite-element solutions. We conclude with a brief summary.

2. MULTIPOLE ANALYSIS OF THE NONCONCENTRIC SHELL

Figure 1 shows a spherical shell surrounding a spherical core displaced from the shell center by L, where R1 and R2 are the shell and core radii, respectively.

 figure: Fig. 1.

Fig. 1. Core is displaced from the shell center by L.

Download Full Size | PDF

The dielectric constants of the host medium, the shell, and the core are denoted by ε0, ε1, and ε2, respectively. We shall employ the quasi-static approximation, which holds when the shell diameter is small compared to the optical wavelength. In this approximation, the electric field is expressed as the gradient of a potential that obeys Laplace’s equation. We note that when the quasi-static approximation holds, the plasmon frequencies and field enhancements are scale invariant. This follows because the Laplace equation contains no intrinsic length scale. Let ψp(r,θ), ψs(r,θ), ψ1(r,θ), and ψ2(r,θ) denote, respectively, the potential functions associated with the primary (incident) field, the scattered field, and the fields in the shell and core. For the potentials external to and inside the shell, we employ a spherical coordinate system whose origin is at the center of the shell, which we denote by (r1,θ1) or, equivalently, by (r1,u1), where u1cosθ1. For the potential inside the core, we use a coordinate system centered on the core and denoted by (r2,θ2) or (r2,u2), where u2cosθ2. Both θ1 and θ2 are measured away from the z-axis defined along the direction of the core displacement. As shown later, we shall use the solid-harmonic addition theorem to connect these two systems. We assume an incident plane wave of amplitude E0 polarized along the z-axis. The potentials are then given by

ψp(r1,u1)=E0r1P1(u1),
ψs(r1,u1)=n=1anR1n+1r1n+1Pn(u1),
ψ1(r1,u1)=n=1[bnR1n+1r1n+1+cnr1nR1n]Pn(u1),
ψ2(r2,u2)=n=1dnr2nR2nPn(u2).
The factors of R1 and R2 are inserted so that the coefficients an, bn, cn, and dn have identical units. The boundary conditions on the shell and core surfaces are
ψp(R1,u1)+ψs(R1,u1)=ψ1(R1,u1),
ε0r1[ψp(r1,u1)+ψs(r1,u1)]|r1=R1=ε1r1ψ1(r1,u1)|r1=R1,
ψ1(r1,u1)|r2=R2=ψ2(R2,u2),
ε1r2ψ1(r1,u1)|r2=R2=ε2r2ψ2(r2,u2)|r2=R2.
Substituting the potentials gives
E0P1(u1)+n=1anPn(u1)=n=1(bn+cn)Pn(u1),ε0E0P1(u1)ε0n=1(n+1)anPn(u1)=ε1n=1[(n+1)bn+ncn]Pn(u1),n=1[bnR1n+1r1n+1+cnr1nR1n]Pn(u1)|r2=R2=n=1dnPn(u2),ε1n=1r2[bnR1n+1r1n+1+cnr1nR1n]Pn(u1)|r2=R2=ε2n=1nR2dnPn(u2),
where in the first two lines we have set E0E0R1 so that E0 has the units of potential. We next multiply the first two equations by Pm(u1) and the second two equations by Pm(u2) and integrate, respectively, with respect to u1 and u2 between -1 and 1. Using the orthogonality relation
11Pm(u)Pn(u)du=δmnwm,
where wm(2m+1)/2, we obtain
E0δ1m+am=bm+cm,
ε0E0δ1mε0(m+1)am=ε1[(m+1)bm+mcm],
n=1Smnbn+n=1Tmncn=dm,
ε1n=1Umnbn+ε1n=1Vmncn=ε2mdm,
where the matrix elements are given by
Smn=wmR1n+111Pn(u1)r1n+1|r2=R2Pm(u2)du2,
Tmn=wmR1n11r1nPn(u1)|r2=R2Pm(u2)du2,
Umn=wmR2R1n+111r2[Pn(u1)r1n+1]r2=R2Pm(u2)du2,
Vmn=wmR2R1n11r2[r1nPn(u1)]r2=R2Pm(u2)du2.
In Appendix A, we show that these matrix elements can be evaluated analytically using the solid-harmonic addition theorem. We then obtain
Smn=(1)mnm!(mn)!n!R1n+1LmnR2m+1,mn,
Tmn=n!(nm)!m!R2mLnmR1n,nm,
Umn=(1)mnm!(mn)!n!(m+1)R1n+1LmnR2m+1,mn,
Vmn=n!(nm)!m!mR2mLnmR1n,nm,
where Smn=0 for m<n and Tmn=0 for m>n. Note that Umn=(m+1)Smn and Vmn=mTmn. Equations (9)–(12) then become
E0δ1m+am=bm+cm,
ε0E0δ1mε0(m+1)am=ε1[(m+1)bm+mcm],
n=1mSmnbn+n=mNTmncn=dm,
ε1(m+1)n=1mSmnbn+ε1mn=mNTmncn=ε2mdm,
where we have truncated the sums to N terms. This is a system of 4N equations in the 4N coefficients an, bn, cn, and dn, n=1,N, which can be solved using a standard linear system solver. We can, however, easily reduce the system to 2N×2N by eliminating am from the first two equations and dm from the second two equations to obtain equations for bm and cm:
3ε0E0δ1m=(m+1)(ε0ε1)bm+[(m+1)ε0+mε1]cm,
0=[(m+1)ε1+mε2]n=1mSmnbn+m(ε2ε1)n=mTmncn.
After solving for bm and cm, substitution into Eqs. (21) and (24) yields am and dm. If desired, we can further reduce the system to N×N by solving Eq. (25) for bm in terms of cm and substituting into Eq. (26) to yield a system of equations for cm:
pm=n=1NAmncn,m=1,2,,N,
where
Amn[(m+1)ε1+mε2][(n+1)ε0+nε1(n+1)(ε0ε1)]Smn+m(ε2ε1)Tmn
and
pm3E0ε0Sm1(m+1)ε1+mε22(ε0ε1).
In computing Amn from Eq. (28), we must recall that Smn=0 when n>m and Tmn=0 when n<m.

3. DIPOLE AND QUADRUPOLE COUPLING AND THE FANO RESONANCE

Here we demonstrate how the Fano coupling occurs between the dipole and quadrupole modes defined by the two lowest order terms in the expansion of Eqs. (25) and (26) (n=1 and n=2, respectively). Keeping only these terms results in

3ε0E0=2(ε0ε1)b1+(2ε0+ε1)c1,
0=(2ε1+ε2)S11b1+(ε2ε1)(T11c1+T12c2),
0=3(ε0ε1)b2+(3ε0+2ε1)c2,
0=(3ε1+2ε2)(S21b1+S22b2)+2(ε2ε1)T22c2,
where we have noted that S12=T21=0 from Eqs. (17) and (18). For brevity, let tR2/R1 and L˜L/R1; then from Eqs. (17) and (18), S11=1/t2, S21=2L˜/t3, S22=1/t3, T11=t, T12=2L˜t, and T22=t2. We shall assume a simple Drude model for the dielectric constant of the shell of the form
ε1=ε0(1ωp2ω2+iωγ)
and set ε2=ε0 for the dielectric constant of the core, where ωp is the plasma frequency of the shell and γ is the plasma damping parameter. To simplify the algebra, we will temporarily set γ=0 and restore it later (i.e., by replacing ω2 wherever it appears with ω2+iωγ). Substituting Eq. (34) into Eqs. (30)–(33) (with γ=0) results in the following set of coupled equations in matrix form:
f=M1v1+Pv2,
0=M2v2+Qv1,
where we have defined the vectors f(E0ω2,0)T, v1(c1,b1)T, v2(c2,b2)T, and
M1(ω2ω12,ω22,ω12t3,ω2ω22),PL˜(0,0ω22t3,0),
M2(ω2ω32,ω42ω32t5,ω2ω42),Q2L˜(0,00,ω2ω42).
Here ω12ωp2/3, ω222ωp2/3, ω322ωp2/5, and ω423ωp2/5. Note that L˜ plays the role of the coupling parameter. The coupled Eqs. (35) and (36) can be rewritten in a form that more clearly exhibits the Fano interaction between the dipole and quadrupole modes as follows. Recall that the lowest order term in the expansion of Eq. (2) is the dipole term, which contains the coefficient a1. As a result, the far-field scattering amplitude (or radiant mode) will be proportional to a1. Setting m=1 in Eqs. (21) and (22) and solving for a1, we have
a1=1ω2[ω12c1+(ω2ω22)b1]=sTv1,
where we have defined s[ω12/ω2,(ω2ω22)/ω2]T. First, note that for a concentric core (L˜=0) Eqs. (35) and (36) become uncoupled and reduce to
f=M1v1,
0=M2v2.
The determinants of the matrices M1 and M2 are, respectively,
D1=(ω2ω12)(ω2ω22)ω12ω22t3,
D2=(ω2ω32)(ω2ω42)ω32ω42t5,
and the uncoupled resonances are the roots of D1=0 and D2=0. We shall later find it convenient to denote these roots ω¯1, ω¯2, ω¯3, and ω¯4; that is, we write Eqs. (42) and (43) as
D1=(ω2ω¯12)(ω2ω¯22),
D2=(ω2ω¯32)(ω2ω¯42).
If the frequency shift is small (e.g., |ω1ω¯1|2ω12), then to first order in small quantities we have
ω¯12=ω12+ω12ω22t3ω12ω22,
ω¯22=ω22ω12ω22t3ω12ω22,
ω¯32=ω32+ω32ω42t5ω32ω42,
ω¯42=ω42ω32ω42t5ω32ω42.
A sufficient condition for these approximations to hold is t1. If this is not the case, we merely solve the quadratic equations D1=0 and D2=0 for ω2. The frequencies ω¯1 and ω¯2 correspond respectively to the dipole and quadrupole resonances of the concentric shell. The quadrupole mode occurs when the surface charges on the inner and outer boundaries of the shell are of opposite sign. In the quasi-static approximation, these are the only excitable modes for the concentric shell. The frequencies ω¯3 and ω¯4 have no physical meaning for the concentric shell but do play a role for the nonconcentric shell, as shown below. The solution to Eq. (40) is
v1cM11f,
where
M11=1D1(ω2ω22,ω22ω12t3,ω2ω12).
For the concentric core, we then have
a1csTv1c=sTM11f=E0ω12(ω2ω22)(1t3)D1.
Note that in the limit when the shell becomes a solid sphere (t=0), we have D1=(ω2ω12)(ω2ω22), and Eq. (52) reduces to
a1s=E0ω12ω2ω12,
and the plasmon resonance of the solid sphere occurs at ω1=ωp/3.

Now consider a nonconcentric core (L˜>0). Solving Eqs. (35) and (36) for v1 results in

v1=(M1PM21Q)1f,
where the resonances are the roots of the determinant of M1PM21Q. In Eq. (54), we first evaluate PM21Q and obtain
PM21Q=2ω22ω42(ω2ω42)t3L˜2D2(0,00,1),
where D2 is given above by Eq. (43). For brevity, let us define
α2ω22ω42(ω2ω42)t3.
Then
M1PM21Q=(ω2ω12,ω22ω12t3,ω2ω22+αL˜2/D2).
The inverse of this matrix can then be shown to be, after some manipulation,
(M1PM21Q)1=D1D2M11+αL˜2FD1D2+α(ω2ω22)L˜2,
where
F(1,00,0)
and M11 is given by Eq. (51). From Eq. (54), we then have
v1=(M1PM21Q)1f=D1D2M11f+αL˜2FfD1D2+α(ω2ω22)L˜2.
The bright-mode scattering amplitude a1 is then given by a1=sTv1=sT(M1PM21Q)1f. Recalling that a1c=sTM11f is the concentric core scattering amplitude, we have
a1=D1D2a1c+αL˜2sTFfD1D2+α(ω2ω22)L˜2.
After evaluating the product sTFf=E0ω12, we finally obtain
a1=D1D2a1cαE0ω12L˜2D1D2+α(ω2ω22)L˜2.
As expected, this reduces to the concentric core response, a1c, when L˜=0. The roots of the denominator of Eq. (62) define a new set of dipole and quadrupole resonances when L˜>0. Recall that the denominator is given by the expression
DD1D2+α(ω2ω22)L˜2=(ω2ω¯12)(ω2ω¯22)(ω2ω¯32)(ω2ω¯42)+α(ω2ω22)L˜2,
where ω¯1 and ω¯2 are, respectively, the dipole and quadrupole resonances of the concentric shell. As noted earlier, the resonances ω¯3 and ω¯4 are not present in the concentric shell but now appear when the core is off center. Examination of the surface charges on the shell boundaries show that ω¯3 and ω¯4 correspond, respectively, to new dipole and quadrupole resonances.

We next employ an analysis similar to that used by Gallinet and Martin [7] in their demonstration of Fano behavior in a system of coupled oscillators. Suppose we expand Eq. (62) about the resonance ω¯4 corresponding to one of the “dark” modes. In the vicinity of ω¯4, the factors ω2ω¯12, ω2ω¯22, and ω2ω¯32 in D1 and D2 are all slowly varying; we then replace them with ω¯42ω¯12, ω¯42ω¯22, and ω¯42ω¯32 and regard these quantities as constants near ω=ω¯4. We then restore the damping parameter by replacing ω2ω¯42 in D2 with ω2iω¯4γω¯42. To first order in small quantities, we have replaced ωγ with ω¯4γ near the ω¯4 resonance. Defining C(ω¯42ω¯12)(ω¯42ω¯22)(ω¯42ω¯32), Eq. (62) becomes

a1a1c=ω2ω¯42αE0ω12L˜2/(a1cC)+iω¯42γω2ω¯42+α(ω¯42ω22)L˜2/C+iω¯42γ.
From this we can write
|a1a1c|2=(κq)2+1κ2+1,
where κ(ω¯4ω22+Δω2)/Γ, qαL˜2[ω¯43ω22+E0ω12/a1c]/CΓ, Γγω¯4, and Δω2α(ω¯42ω22)L˜2/C. Equation (65) is the classic Fano resonance expression, which displays the frequency shift, Δω, and the asymmetry parameter, q. From Eq. (62), quenching of the radiant mode is theoretically possible in the absence of damping at a frequency for which the numerator vanishes. In particular, when Γ0, quenching occurs for frequencies that satisfy κ=q.

Calculations were carried for |a1| computed from Eq. (62) for the following five values of the core displacement: L=0,1,2,3,4nm assuming shell and core diameters of 30 nm and 20 nm, respectively, using the dielectric parameters for gold. The core offsets are illustrated in Fig. 2, and the normalized plots of |a1| versus frequency are shown in Fig. 3.

 figure: Fig. 2.

Fig. 2. Five core offsets.

Download Full Size | PDF

 figure: Fig. 3.

Fig. 3. Magnitude of the radiant mode amplitude for five core offsets.

Download Full Size | PDF

In Fig. 3 we see the partial quenching of the radiant mode between the two peaks, as well as the frequency shift in the primary resonance peak.

4. MECHANICAL ANALOG

A number of authors have used the analogy of coupled harmonic oscillators to illustrate some of the key features of Fano interference. In a similar fashion, we can show that the equations that describe the response of the nonconcentric shell closely resembles the behavior of four masses connected by three springs, where the masses are confined along a line with small displacements ai, i=1,2,3,4, away from equilibrium with masses mi. Let K denote the spring constant of the springs between masses 1 and 2 and between 3 and 4, while the spring joining masses 2 and 3 has constant kK. It is convenient to write k=εK, where ε1. Thus, masses 1 and 2 are weakly coupled to masses 3 and 4. If we drive mass 1 with a force F=fexp(iωt), we obtain the coupled equations

(ω2ω12)a1+ω12a2=f,[ω2(1+ε2)ω22]a2+ω22a1+ε2ω22a3=0,[ω2(1+ε2)ω32]a3+ε2ω32a2+ω32a4=0,(ω2ω42)a4+ω42a3=0,
where ωi2K/mi. If we define the quantities
D1(ω2ω12)[ω2(1+ε2)ω22]ω12ω22,
D2(ω2ω42)[ω2(1+ε2)ω32]ω32ω42,
and solve the above system for a1, we obtain
a1=fD[(ω2(1+ε2)ω22)D2ε4ω22ω32(ω2ω42)],
where
DD1D2ε4ω22ω32(ω2ω12)(ω2ω42)
is the determinant of the system. If the masses 1 and 2 are uncoupled from masses 3 and 4 (when ε=0), the resultant amplitude of mass 1, denoted by a10, is
a10=f[ω2(1+ε2)ω22]D1.
Substituting this into Eq. (69), we obtain
a1=D1D2a10+ε4fω22ω32(ω2ω42)D1D2ε4ω22ω32(ω2ω12)(ω2ω42),
which has a form similar to that of the bright-mode expression in Eq. (62), where in this case ε plays the role of the coupling parameter.

5. ELECTRIC FIELD PROFILES

Once the potential expansion coefficients are calculated, the electric field is derived from the negative of the gradient of the potentials (E=ψ). Using the above multipole solution, we computed profiles of the electric field magnitude along the z-axis penetrating the particle. In this calculation, the following simple Drude model for the dielectric function of the shell given by Eq. (34) with ωp=8.6eV and γ=0.17eV, the approximate values for gold. In our calculations, we assumed a core dielectric constant of silica (ε2=2.43) and set ε0=1. The calculations were carried out assuming R1=15nm, R2=10nm with five values of the core displacement L=0,1,2,3,4nm. The plots in Fig. 4 show the magnitude of the z-component of the electric field along a line through the particle. We compared the multipole calculations using N=10 terms in the expansion to finite-element calculations obtained with the commercial software package COMSOL Multiphysics [8]. In the figures, the solid and dashed curves are the multipole and COMSOL calculations, respectively, which show close agreement. Each profile calculation was performed at the wavelength corresponding to the peaks in the electric field enhancement at 366, 388, 396, 416, and 468 nm. From the figures, we see that the peak wavelength is redshifted as the core offset increases. In evaluating the multipole expansions for the potentials given by Eqs. (1)–(4), we found that the expansion for the potential ψ1 for points in the interior of the shell converged slowly for large core offsets and when r1 was close to the core boundary. This was a more serious problem for the bn expansion. This problem was solved by using Green’s theorem to compute the interior shell potential by a boundary integration involving the expansion coefficients an and dn only. This works well since the an and dn expansions converge rapidly. The Green’s function analysis is outlined in Appendix B.

 figure: Fig. 4.

Fig. 4. Profiles of the magnitude of the z-component of the electric field through the particle for different core offsets L. Note the change in the vertical scale.

Download Full Size | PDF

6. CONCLUSION

The off-center core generates multipoles of all orders and their amplitudes up to order N were obtained by solving an N by N linear system. We showed how the matrix elements in this system can be analytically derived with the aid of the solid-harmonic addition theorem. For N=10, the multipole calculations of the electric field profile through the shell were shown to be in excellent agreement with finite-element calculations. Moreover, at the shell’s plasmon resonance, the profile calculations show that the electric field enhancement on the thin side of the shell increases with larger core offset and can be significantly higher than the enhancement for the concentric shell. The expansion coefficients of the two lowest order terms, the dipole and quadrupole resonances, were expressed explicitly in analytic form given by Eq. (35) and (36) resulting after some manipulation in the Fano expression in Eq. (65).

APPENDIX A: MATRIX ELEMENT EVALUATION

In this appendix, we evaluate the matrix elements in Eqs. (13)–(16). In spherical coordinates, our problem is radially symmetric in the azimuthal angle ϕ. We then consider the following two solutions to Laplace’s equation:

Rn(r)=rnPn(u),
In(r)=1rn+1Pn(u),
where r=(r,u) with u=cosθ. Addition theorems derived in [9,10] are
Rn(r+r)=k=0nn!k!(nk)!Rk(r)Rnk(r),
In(r+r)=k=0(1)k(n+k)!k!n!Rk(r)In+k(r).
From Fig. 5, r1=R+r2, where r1=(r1,u1), r2=(r2,u2), and R=(L,0). First, setting r=r2 and r=R in Eq. (A3), we find
Rn(r2+R)=Rn(r1)=k=0nn!k!(nk)!Rk(r2)Rnk(R)
or from Eqs. (A1) and (A2)
r1nPn(u1)=k=0nn!k!(nk)!r2kPk(u2)Lnk.
Next, setting r=r2 and r=R in Eq. (A4), we get
In(r2+R)=In(r1)=k=0(1)k(n+k)!k!n!Rk(R)In+k(r2)
or
1r1n+1Pn(u1)=k=0(1)k(n+k)!k!n!Lk1r2n+k+1Pn+k(u2).
Now substituting these results in Eqs. (13)–(16) and using the orthogonality of the Legendre functions, we obtain Eqs. (17)–(20).
 figure: Fig. 5.

Fig. 5. Addition theorem geometry.

Download Full Size | PDF

APPENDIX B: GREEN’S FUNCTION ANALYSIS

The multipole expansion in the interior of the shell for large core offsets was found to converge slowly for values of r1 close to the core surface. This problem can be solved by expressing the potential (and hence the electric field) in terms of the expansion coefficients an of the potential external to the shell and the expansion coefficients dn of the potential in the interior of the core, since both expansions involving an and dn converge rapidly. This is accomplished using Green’s theorem to relate the interior shell potential to an integration of the potential over the shell and core boundaries using Green’s theorem. The boundary potentials can be related to the an and dn expansions using the boundary conditions. Green’s theorem relates a point r inside the shell to a boundary integration given by

ψ(r)=14πS[g(r|r)ϕ(r)ψ(r)g(r|r)]·n^dS,
where the integral is over the surface S enclosing r, in this case the shell and core boundaries, and g(r|r)=1/|rr| is the Green’s function for the Laplacian. Suppose we wish to compute the potential at a point on the z-axis in the shell’s interior. In this case, θ1=θ2=0 (and u1=u2=1) and r2=r1±L, where the plus or minus depends on which side of the core the point r1 resides. The Green’s function for the interior point r1=(r1,1), when the integration is over the outer shell boundary [where here r=(R1,u1) is a boundary point], is
g(r1,1|R1,u1)=1r12+R122r1R1u1.
The Green’s function for the interior point r2=(r2,1), when the integration is over the core surface (where here r=(R2,u2) is a boundary point), is
g(r2,1|R2,u2)=1r22+R222r2R2u2.
In using Green’s theorem, the potentials in the integral in Eq. (B1) are the interior potentials on the boundary, which are related to the exterior potentials using the boundary conditions. On substituting the potentials and the Green’s functions into the integral in Eq. (B1) and using the identity
11Pn(u)dur2+R22rRu={(22n+1)rnRn+1,r<R,(22n+1)Rnrn+1,R<r,
we obtain an expression for the potential on the z-axis at the interior point r1 given by
ψ(r1,1)=E0r13(2+ε0ε1)+(1ε0ε1)n=1Nan(n+12n+1)r1nR1n+(1ε2ε1)n=1Nndn2n+1R2n+1r2n+1,
where in the last term r2=r1±L. This expression only involves the coefficients an and dn and is numerically more stable than the expansion in Eq. (3) for large core offsets. However, as a check, the above formula can be shown to give results identical to the expansion in Eq. (3) for small offsets. The z-component of the electric field is then obtained by computing its derivative with respect to r1.

Funding

Duke Faculty Exploratory Project.

REFERENCES

1. C. Wu, A. B. Khanikaev, R. Adato, N. Arju, A. A. Yanik, H. Altug, and G. Shvets, “Fano-resonant asymmetric metamaterials for ultrasensitive spectroscopy and identification of molecular monolayers,” Nat. Mat. 11, 69–75 (2011). [CrossRef]  

2. A. E. Miroshnichenko, S. Flach, and Y. S. Kivshar, “Fano resonances in nanoscale structures,” Rev. Mod. Phys. 82, 2257–2298 (2010). [CrossRef]  

3. A. M. Satanin and Y. S. Joe, “Fano interference and resonances in open systems,” Phys. Rev. B 71, 205417 (2005). [CrossRef]  

4. Y. Wu and P. Nordlander, “Plasmon hybridization in nanoshells with a nonconcentric core,” J. Chem. Phys. 125, 124708 (2006). [CrossRef]  

5. M. Kerker, ed. Selected Papers on Surface-Enhanced Raman Scattering (SPIE, 1990).

6. Y. S. Joe, A. M. Satanin, and C. S. Kim, “Classical analogy of Fano resonances,” Phys. Scripta 74, 259–266 (2006). [CrossRef]  

7. B. Gallinet and O. J. F. Martin, “Ab initio theory of Fano resonances in plasmonic nanostructures and metamaterials,” Phys. Rev. B 83, 235427 (2011). [CrossRef]  

8. See www.comsol.com for COMSOL Multiphysics.

9. M. J. Caola, “Solid harmonics and their addition theorems,” J. Phys. A 11, L23–L25 (1978). [CrossRef]  

10. J. P. Dahl and M. P. Barnett, “Expansion theorems for solid spherical harmonics,” Mol. Phys. 9, 175–178 (1965). [CrossRef]  

Cited By

Optica participates in Crossref's Cited-By Linking service. Citing articles from Optica Publishing Group journals and other participating publishers are listed here.

Alert me when this article is cited.


Figures (5)

Fig. 1.
Fig. 1. Core is displaced from the shell center by L .
Fig. 2.
Fig. 2. Five core offsets.
Fig. 3.
Fig. 3. Magnitude of the radiant mode amplitude for five core offsets.
Fig. 4.
Fig. 4. Profiles of the magnitude of the z -component of the electric field through the particle for different core offsets L . Note the change in the vertical scale.
Fig. 5.
Fig. 5. Addition theorem geometry.

Equations (87)

Equations on this page are rendered with MathJax. Learn more.

ψ p ( r 1 , u 1 ) = E 0 r 1 P 1 ( u 1 ) ,
ψ s ( r 1 , u 1 ) = n = 1 a n R 1 n + 1 r 1 n + 1 P n ( u 1 ) ,
ψ 1 ( r 1 , u 1 ) = n = 1 [ b n R 1 n + 1 r 1 n + 1 + c n r 1 n R 1 n ] P n ( u 1 ) ,
ψ 2 ( r 2 , u 2 ) = n = 1 d n r 2 n R 2 n P n ( u 2 ) .
ψ p ( R 1 , u 1 ) + ψ s ( R 1 , u 1 ) = ψ 1 ( R 1 , u 1 ) ,
ε 0 r 1 [ ψ p ( r 1 , u 1 ) + ψ s ( r 1 , u 1 ) ] | r 1 = R 1 = ε 1 r 1 ψ 1 ( r 1 , u 1 ) | r 1 = R 1 ,
ψ 1 ( r 1 , u 1 ) | r 2 = R 2 = ψ 2 ( R 2 , u 2 ) ,
ε 1 r 2 ψ 1 ( r 1 , u 1 ) | r 2 = R 2 = ε 2 r 2 ψ 2 ( r 2 , u 2 ) | r 2 = R 2 .
E 0 P 1 ( u 1 ) + n = 1 a n P n ( u 1 ) = n = 1 ( b n + c n ) P n ( u 1 ) , ε 0 E 0 P 1 ( u 1 ) ε 0 n = 1 ( n + 1 ) a n P n ( u 1 ) = ε 1 n = 1 [ ( n + 1 ) b n + n c n ] P n ( u 1 ) , n = 1 [ b n R 1 n + 1 r 1 n + 1 + c n r 1 n R 1 n ] P n ( u 1 ) | r 2 = R 2 = n = 1 d n P n ( u 2 ) , ε 1 n = 1 r 2 [ b n R 1 n + 1 r 1 n + 1 + c n r 1 n R 1 n ] P n ( u 1 ) | r 2 = R 2 = ε 2 n = 1 n R 2 d n P n ( u 2 ) ,
1 1 P m ( u ) P n ( u ) d u = δ m n w m ,
E 0 δ 1 m + a m = b m + c m ,
ε 0 E 0 δ 1 m ε 0 ( m + 1 ) a m = ε 1 [ ( m + 1 ) b m + m c m ] ,
n = 1 S m n b n + n = 1 T m n c n = d m ,
ε 1 n = 1 U m n b n + ε 1 n = 1 V m n c n = ε 2 m d m ,
S m n = w m R 1 n + 1 1 1 P n ( u 1 ) r 1 n + 1 | r 2 = R 2 P m ( u 2 ) d u 2 ,
T m n = w m R 1 n 1 1 r 1 n P n ( u 1 ) | r 2 = R 2 P m ( u 2 ) d u 2 ,
U m n = w m R 2 R 1 n + 1 1 1 r 2 [ P n ( u 1 ) r 1 n + 1 ] r 2 = R 2 P m ( u 2 ) d u 2 ,
V m n = w m R 2 R 1 n 1 1 r 2 [ r 1 n P n ( u 1 ) ] r 2 = R 2 P m ( u 2 ) d u 2 .
S m n = ( 1 ) m n m ! ( m n ) ! n ! R 1 n + 1 L m n R 2 m + 1 , m n ,
T m n = n ! ( n m ) ! m ! R 2 m L n m R 1 n , n m ,
U m n = ( 1 ) m n m ! ( m n ) ! n ! ( m + 1 ) R 1 n + 1 L m n R 2 m + 1 , m n ,
V m n = n ! ( n m ) ! m ! m R 2 m L n m R 1 n , n m ,
E 0 δ 1 m + a m = b m + c m ,
ε 0 E 0 δ 1 m ε 0 ( m + 1 ) a m = ε 1 [ ( m + 1 ) b m + m c m ] ,
n = 1 m S m n b n + n = m N T m n c n = d m ,
ε 1 ( m + 1 ) n = 1 m S m n b n + ε 1 m n = m N T m n c n = ε 2 m d m ,
3 ε 0 E 0 δ 1 m = ( m + 1 ) ( ε 0 ε 1 ) b m + [ ( m + 1 ) ε 0 + m ε 1 ] c m ,
0 = [ ( m + 1 ) ε 1 + m ε 2 ] n = 1 m S m n b n + m ( ε 2 ε 1 ) n = m T m n c n .
p m = n = 1 N A m n c n , m = 1,2 , , N ,
A m n [ ( m + 1 ) ε 1 + m ε 2 ] [ ( n + 1 ) ε 0 + n ε 1 ( n + 1 ) ( ε 0 ε 1 ) ] S m n + m ( ε 2 ε 1 ) T m n
p m 3 E 0 ε 0 S m 1 ( m + 1 ) ε 1 + m ε 2 2 ( ε 0 ε 1 ) .
3 ε 0 E 0 = 2 ( ε 0 ε 1 ) b 1 + ( 2 ε 0 + ε 1 ) c 1 ,
0 = ( 2 ε 1 + ε 2 ) S 11 b 1 + ( ε 2 ε 1 ) ( T 11 c 1 + T 12 c 2 ) ,
0 = 3 ( ε 0 ε 1 ) b 2 + ( 3 ε 0 + 2 ε 1 ) c 2 ,
0 = ( 3 ε 1 + 2 ε 2 ) ( S 21 b 1 + S 22 b 2 ) + 2 ( ε 2 ε 1 ) T 22 c 2 ,
ε 1 = ε 0 ( 1 ω p 2 ω 2 + i ω γ )
f = M 1 v 1 + P v 2 ,
0 = M 2 v 2 + Q v 1 ,
M 1 ( ω 2 ω 1 2 , ω 2 2 , ω 1 2 t 3 , ω 2 ω 2 2 ) , P L ˜ ( 0 , 0 ω 2 2 t 3 , 0 ) ,
M 2 ( ω 2 ω 3 2 , ω 4 2 ω 3 2 t 5 , ω 2 ω 4 2 ) , Q 2 L ˜ ( 0 , 0 0 , ω 2 ω 4 2 ) .
a 1 = 1 ω 2 [ ω 1 2 c 1 + ( ω 2 ω 2 2 ) b 1 ] = s T v 1 ,
f = M 1 v 1 ,
0 = M 2 v 2 .
D 1 = ( ω 2 ω 1 2 ) ( ω 2 ω 2 2 ) ω 1 2 ω 2 2 t 3 ,
D 2 = ( ω 2 ω 3 2 ) ( ω 2 ω 4 2 ) ω 3 2 ω 4 2 t 5 ,
D 1 = ( ω 2 ω ¯ 1 2 ) ( ω 2 ω ¯ 2 2 ) ,
D 2 = ( ω 2 ω ¯ 3 2 ) ( ω 2 ω ¯ 4 2 ) .
ω ¯ 1 2 = ω 1 2 + ω 1 2 ω 2 2 t 3 ω 1 2 ω 2 2 ,
ω ¯ 2 2 = ω 2 2 ω 1 2 ω 2 2 t 3 ω 1 2 ω 2 2 ,
ω ¯ 3 2 = ω 3 2 + ω 3 2 ω 4 2 t 5 ω 3 2 ω 4 2 ,
ω ¯ 4 2 = ω 4 2 ω 3 2 ω 4 2 t 5 ω 3 2 ω 4 2 .
v 1 c M 1 1 f ,
M 1 1 = 1 D 1 ( ω 2 ω 2 2 , ω 2 2 ω 1 2 t 3 , ω 2 ω 1 2 ) .
a 1 c s T v 1 c = s T M 1 1 f = E 0 ω 1 2 ( ω 2 ω 2 2 ) ( 1 t 3 ) D 1 .
a 1 s = E 0 ω 1 2 ω 2 ω 1 2 ,
v 1 = ( M 1 P M 2 1 Q ) 1 f ,
P M 2 1 Q = 2 ω 2 2 ω 4 2 ( ω 2 ω 4 2 ) t 3 L ˜ 2 D 2 ( 0 , 0 0 , 1 ) ,
α 2 ω 2 2 ω 4 2 ( ω 2 ω 4 2 ) t 3 .
M 1 P M 2 1 Q = ( ω 2 ω 1 2 , ω 2 2 ω 1 2 t 3 , ω 2 ω 2 2 + α L ˜ 2 / D 2 ) .
( M 1 P M 2 1 Q ) 1 = D 1 D 2 M 1 1 + α L ˜ 2 F D 1 D 2 + α ( ω 2 ω 2 2 ) L ˜ 2 ,
F ( 1 , 0 0 , 0 )
v 1 = ( M 1 P M 2 1 Q ) 1 f = D 1 D 2 M 1 1 f + α L ˜ 2 F f D 1 D 2 + α ( ω 2 ω 2 2 ) L ˜ 2 .
a 1 = D 1 D 2 a 1 c + α L ˜ 2 s T F f D 1 D 2 + α ( ω 2 ω 2 2 ) L ˜ 2 .
a 1 = D 1 D 2 a 1 c α E 0 ω 1 2 L ˜ 2 D 1 D 2 + α ( ω 2 ω 2 2 ) L ˜ 2 .
D D 1 D 2 + α ( ω 2 ω 2 2 ) L ˜ 2 = ( ω 2 ω ¯ 1 2 ) ( ω 2 ω ¯ 2 2 ) ( ω 2 ω ¯ 3 2 ) ( ω 2 ω ¯ 4 2 ) + α ( ω 2 ω 2 2 ) L ˜ 2 ,
a 1 a 1 c = ω 2 ω ¯ 4 2 α E 0 ω 1 2 L ˜ 2 / ( a 1 c C ) + i ω ¯ 4 2 γ ω 2 ω ¯ 4 2 + α ( ω ¯ 4 2 ω 2 2 ) L ˜ 2 / C + i ω ¯ 4 2 γ .
| a 1 a 1 c | 2 = ( κ q ) 2 + 1 κ 2 + 1 ,
( ω 2 ω 1 2 ) a 1 + ω 1 2 a 2 = f , [ ω 2 ( 1 + ε 2 ) ω 2 2 ] a 2 + ω 2 2 a 1 + ε 2 ω 2 2 a 3 = 0 , [ ω 2 ( 1 + ε 2 ) ω 3 2 ] a 3 + ε 2 ω 3 2 a 2 + ω 3 2 a 4 = 0 , ( ω 2 ω 4 2 ) a 4 + ω 4 2 a 3 = 0 ,
D 1 ( ω 2 ω 1 2 ) [ ω 2 ( 1 + ε 2 ) ω 2 2 ] ω 1 2 ω 2 2 ,
D 2 ( ω 2 ω 4 2 ) [ ω 2 ( 1 + ε 2 ) ω 3 2 ] ω 3 2 ω 4 2 ,
a 1 = f D [ ( ω 2 ( 1 + ε 2 ) ω 2 2 ) D 2 ε 4 ω 2 2 ω 3 2 ( ω 2 ω 4 2 ) ] ,
D D 1 D 2 ε 4 ω 2 2 ω 3 2 ( ω 2 ω 1 2 ) ( ω 2 ω 4 2 )
a 1 0 = f [ ω 2 ( 1 + ε 2 ) ω 2 2 ] D 1 .
a 1 = D 1 D 2 a 1 0 + ε 4 f ω 2 2 ω 3 2 ( ω 2 ω 4 2 ) D 1 D 2 ε 4 ω 2 2 ω 3 2 ( ω 2 ω 1 2 ) ( ω 2 ω 4 2 ) ,
R n ( r ) = r n P n ( u ) ,
I n ( r ) = 1 r n + 1 P n ( u ) ,
R n ( r + r ) = k = 0 n n ! k ! ( n k ) ! R k ( r ) R n k ( r ) ,
I n ( r + r ) = k = 0 ( 1 ) k ( n + k ) ! k ! n ! R k ( r ) I n + k ( r ) .
R n ( r 2 + R ) = R n ( r 1 ) = k = 0 n n ! k ! ( n k ) ! R k ( r 2 ) R n k ( R )
r 1 n P n ( u 1 ) = k = 0 n n ! k ! ( n k ) ! r 2 k P k ( u 2 ) L n k .
I n ( r 2 + R ) = I n ( r 1 ) = k = 0 ( 1 ) k ( n + k ) ! k ! n ! R k ( R ) I n + k ( r 2 )
1 r 1 n + 1 P n ( u 1 ) = k = 0 ( 1 ) k ( n + k ) ! k ! n ! L k 1 r 2 n + k + 1 P n + k ( u 2 ) .
ψ ( r ) = 1 4 π S [ g ( r | r ) ϕ ( r ) ψ ( r ) g ( r | r ) ] · n ^ d S ,
g ( r 1 , 1 | R 1 , u 1 ) = 1 r 1 2 + R 1 2 2 r 1 R 1 u 1 .
g ( r 2 , 1 | R 2 , u 2 ) = 1 r 2 2 + R 2 2 2 r 2 R 2 u 2 .
1 1 P n ( u ) d u r 2 + R 2 2 r R u = { ( 2 2 n + 1 ) r n R n + 1 , r < R , ( 2 2 n + 1 ) R n r n + 1 , R < r ,
ψ ( r 1 , 1 ) = E 0 r 1 3 ( 2 + ε 0 ε 1 ) + ( 1 ε 0 ε 1 ) n = 1 N a n ( n + 1 2 n + 1 ) r 1 n R 1 n + ( 1 ε 2 ε 1 ) n = 1 N n d n 2 n + 1 R 2 n + 1 r 2 n + 1 ,
Select as filters


Select Topics Cancel
© Copyright 2024 | Optica Publishing Group. All rights reserved, including rights for text and data mining and training of artificial technologies or similar technologies.