Theory and Modelling of Optical Waveguide Sensors Utilising Surface Plasmon Resonance
Theory and Modelling of Optical Waveguide Sensors Utilising Surface Plasmon Resonance
Theory and Modelling of Optical Waveguide Sensors Utilising Surface Plasmon Resonance
Abstract
A theoretical analysis of the phenomenon of excitation of surface plasma waves in integrated-optical waveguide structures is
carried out. Rigorous approach to analysis of light propagation through a waveguide structure with a thin metal overlayer
supporting surface plasma waves is formulated using a bi-directional mode expansion and propagation method. It is demonstrated
that because the back-reflections in the structure are very weak and most of optical power is transmitted by only a limited number
of modes of the sensing structure, the method can be considerably simplified. 1999 Elsevier Science S.A. All rights reserved.
0925-4005/99/$ - see front matter 1999 Elsevier Science S.A. All rights reserved.
PII: S 0 9 2 5 - 4 0 0 5 ( 9 8 ) 0 0 3 2 8 - 1
J. C& tyroky et al. / Sensors and Actuators B 54 (1999) 6673 67
Fig. 1. Effective refractive indices and attenuation constants of SPWs supported by a thin gold layer. Na and Ns or ba and bs correspond to
symmetric and antisymmetric SPWs, respectively.
td =
1
Im {
om + od}, tm =
1 ) !
Im
om +od ")
k0od k0 om
(2)
into the dielectric and metal, respectively. It is this
strong localisation of energy of the SPW that makes its
application for optical sensing attractive.
Solution of the eigenmode equation for a thin metal Fig. 2. Magnetic field distribution of an antisymmetric and symmet-
ric surface plasmons propagating along a gold film 50 nm thick
layer surrounded by two dielectric materials (see inset
surrounded with dielectrics of refractive indices ng =1.478 and na =
in Fig. 1) leads to two guided eigenmodes [10]. If the 1.40. (a) Real part (b) imaginary part (note the difference in vertical
metal layer thickness is large compared to tm, these scales). The origin of the x co-ordinate x =0 lies at the bottom
eigenmodes are basically the two SPWs localised at the dielectric-metal interface (see inset in Fig. 1).
68 J. C& tyroky et al. / Sensors and Actuators B 54 (1999) 6673
Fig. 5. Field distribution of the lowest-order TM normal modes of the planar waveguide structure supporting SPWs. The parameters are given
in the graphs. The TM0 mode is shown only for na = 1.30 as its character is only very weakly dependent of na. Zero point of the x-co-ordinate
lies at the waveguide metal interface.
Our calculations for both model structures concern show that three or two guided TM-polarised modes
the performance of the SPR integrated optical systems exist in the metal-loaded waveguide structure, depend-
with various values of the thickness of the metal layer, ing on its parameters. The lowest mode (denoted as
namely 30, 40, 50 and 60 nm, and the refractive indices TM0) corresponds to the antisymmetric SPW with a
of the analyte na varying in the range of 1.3 1.45. In very high modal attenuation, and with a considerably
this range of parameters, the propagation constants and higher effective refractive index than the other guided
field distributions of guided modes of both waveguide modes. The next two higher order modes represent
structures in Figs. 3 and 4 will be shown to be nearly hybrid modes exhibiting both features of guided mode
identical. of a dielectric waveguide and surface plasma waves. It
We made a rigorous numerical analysis of wave should be noted that more symmetric structures and
propagation in the multilayer sensing waveguide section those with thicker metal layer support two such hybrid
using the method of transversal resonance [11 14], and modes while more asymmetric structures and those with
its results are presented in Figs. 5 and 6. The results thinner metal support only one. More specifically, for
70 J. C& tyroky et al. / Sensors and Actuators B 54 (1999) 6673
Fig. 6. Effective refractive index and propagation constant of four lowest-order modes of the sensing section S of the waveguide structure in Fig.
4 (dot-dot-dash, solid, dash, dot-dash, and short-dash lines). Squares and circles correspond to guided modes calculated for the open structure in
Fig. 3.
J. C& tyroky et al. / Sensors and Actuators B 54 (1999) 6673 71
the lowest refractive index of an analyte of 1.30, only maintain their identity and the interaction with SPW is
the next higher mode TM1 is also guided (i.e. localised reflected by resonantly enhanced losses of the modes.
near the guiding layer, see Fig. 5a), for higher refractive The refractive index of analyte at which the modal
index of analyte na of 1.37 (Fig. 5b) and a thicker gold transformation or resonant attenuation occur depends
layer (Fig. 5df) also the third mode TM2 is guided, for on the metal layer thickness. and is the smaller the
na = 1.40 the TM2 mode is guided for both lower and higher is the metal thickness falling within between 1.39
higher gold thicknesses (Fig. 5c and f). With increasing for 30 nm thick Au layer and 1.36 for 60 nm thick Au
na the character of the TM1 mode changes from that layer. This is a consequence of the fact that the propa-
similar to the guided mode of the waveguide to that gation constant of the symmetric SPW increases with
resembling more the symmetric SPW. increasing metal layer thickness and decreases with
The dependence of the effective refractive indices and decreasing refractive index of analyte so that these two
loss coefficients of four lowest-order modes of the effects may cancel out each other.
sensing section S of the enclosed waveguide structure
(Fig. 4) on the refractive index of the analyte na are
plotted in Fig. 6. It is apparent that while the lowest-or- 3. Modelling transmission through a surface plasmon
der modethe antisymmetric surface plasmon TM0 resonance waveguide sensor
remains unaffected by the presence of the waveguide, it
is not the case of the higher-order modes. We observe For the design of amplitude waveguide based SPR
two different types of behaviour of the guided modes sensing structures, the dependence of the total optical
when varying the refractive index of analyte. For rather power transfer from the input waveguide to the output
thin metal layers (see plots for 30 and 40 nm thick gold waveguide on the refractive index of analyte needs to be
layers in Fig. 6ad), as a result of the excitation of determined. In our modelling we assumed that both the
SPW, the waveguide modes undergo modal transforma- input and output waveguide sections (denoted as F1
tion and are reduced to higher order modes when and F2 in Figs. 3 and 4) are single-mode and the input
decreasing the refractive index of analyte. In structures section is excited by its eigenmode. The task is to
with rather thick metal layers (Fig. 6e h) the modes calculate the dependence of the output power carried
Fig. 7. Dependence of the relative power transfer of the waveguide SPR sensor on the refractive index of the analyte, na, as calculated by four
laboratories.
72 J. C& tyroky et al. / Sensors and Actuators B 54 (1999) 6673
by the mode of the output waveguide on the refractive 4. Discussion and conclusions
index of the analyte, na, for values of the metal (gold)
thickness of 30, 40, 50 and 60 nm. The length of the It may be concluded from Fig. 7 that all the methods
sensing section is 2 mm, the permittivity of gold is give essentially identical results. Apparently the method
considered to be om = 11.836107 + 1.357724i at the used by IREE relying on considering a larger number
wavelength of 633 nm. The task was solved indepen- of modes (of an enclosed structure) differs from the
dently at four research laboratories (UT, ORC, IREE, other methods in the resonant regions where the power
and GPI). All the groups used, in principle, the same transfer of about and B 60 dB was determined. This
eigenmode expansion and propagation approach in the difference can be explained by a very small power
approximation of which the multiple reflections in the carried by higher-order (radiation) modes that are ne-
structure are neglected. In order to check the validity of glected in other methods. It may be objected that
this approximation, IREE used the rigorous BEP residual reflections from the magnetic walls in the en-
method [8,14] to calculate the total back-reflection in closed structure can increase the power transfer at the
the enclosed waveguide structure. It was found that the resonance significantly. To investigate this, we repeated
back reflected power was always below 30 dB, typi- the calculations for several different positions of the
cally in the range of 40 to 60 dB, without observ- walls to allow the reflected waves to combine with
able interferences while slightly changing na and thus different phases. It lead to changes in the order of a few
the propagation constants of modes. dB in the detailed shapes of the curves only at the
Neglecting multiple reflections, the power transfer power levels below : 60 dB. We can thus conclude
through the waveguide structure can be expressed by a that the results presented in Fig. 7 are accurate up to
) )
simple expression: this level.
2 That the results of UT do not differ significantly
p (L)= a0 (L)/a0 (0) 2 = % c0p exp [ibpL] (3) from others, means that the approximation of neglect-
p
ing the power transfer to the short range SPW (TM0) is
where a0(0) and a0(L) are the complex amplitudes of justifiable. This is the consequence both of its very
the fundamental mode of the input and output wave- strong attenuation and of its very low overlap with the
& &
guide (F =F1 or F2), respectively: guided mode of the input and output waveguide sec-
tions.
E Fx,0 H Sy,p dx E Sx,pH Fy,0 dx We can thus conclude that in the systems described
c0p =
& E F
x,0 H F
y,0
&
dx E S
H
x,p
S
y,p dx
(4) here the power transfer to the asymmetrical mode and
to radiation modes can be neglected in a very good
approximation, and sensor performance can be calcu-
are the coupling coefficients of the fundamental mode lated by taking into account guided modes of the sensor
of the input and output section F and the p-th mode section only. Furthermore, some explicit conclusions
TMp of the sensing section S. Here, E Fx,0, H Fy,0 and can be drawn for the performance of the device. In all
E Sx,p, H Sy,p are the transversal components of the mode sensors there are two regions where the dynamic range
fields in the waveguide sections F and S, respectively. If of the output power is : 50 dB for a refractive index
all modes of the section S are taken into account, it change of about 0.02. Variation of the metal layer
holds: thickness causes small shifts of these regions, while also
the resolution changes somewhat. Hence the thickness
% c0p =1 (5) of the metal layer is one of the parameters suited for
p
tailoring the device to specific applications.
ORC and GPI used this approach for the simulation
of the open structure (Fig. 3) with all (i.e. either two or
three) localised modes of the sensing section S. In the
calculations performed at the UT the contribution of
the asymmetrical surface plasmon mode to the power Acknowledgements
transfer has been deliberately neglected because they
estimated it to be relatively small. Also they did not The authors would like to acknowledge the support
calculate attenuation over 60 dB because they con- of EU Environment and Climate Programme project
sidered it to be out of the scope of practical measure- BIOPTICAS (contract No. EV5V-CT92-0067, R.D.H.
ments. IREE used 15 lowest order modes of the and J.S.W.), and of the Grant Agency of the Czech
enclosed waveguide (Fig. 4). It was verified that further Republic (contracts No. 102/96/1561 and 303/96/1358,
increase in the number of modes did not lead to notable J.H. and J.C& .). The bi-directional mode expansion
changes in the results. The calculated results are plotted propagation method was developed in the framework
in Fig. 7. of the European Action COST 240.
J. C& tyroky et al. / Sensors and Actuators B 54 (1999) 6673 73