You are on page 1of 11

Project no: Project acronym: Project title: Instrument: Thematic Priority:

NMP-CT-2003-505699 SPP Surface Plasmon Photonics STREP NMP

D5 Report providing the specification of computational techniques appropriate for modelling fields associated with surface plasmons

Due date of deliverable: Actual submission date: Start date of project:

Month 12 Month 12 January 1st 2004 Duration: 3 Years UAM

Organisation name of lead contractor for this deliverable:

Project co-funded by the European Commission within the Sixth Framework Programme (2002-2006)
PU PP RE CO Dissemination Level Public Restricted to other programme participants (including the Commission Services) Restricted to a group specified by the consortium (including the Commission Services) Confidential, only for members of the consortium (including the Commission Services)

_____________________________________________________________________________________________ Deliverable 5 Surface Plasmon Photonics (SPP) NMP-CT-2003-505699 Page 1 of 11

Surface Plasmon Photonics (SPP) NMP-CT-2003-505699


Deliverable 5: Report providing specification of computational techniques appropriate for modelling fields associated with SPs. (Part of WP3 Surface Plasmon Devices). Due: Month 12. Partners: Contents. 1 Introduction/Overview.................................................................................................................3 2 Background...................................................................................................................................4 3 Different Techniques that can be employed................................................................................5 4 Summary.......................................................................................................................................9 5 Table 1 Comparison of numerical techniques............................................................................10 6 References...................................................................................................................................11 UAM, UZ, IC

_____________________________________________________________________________________________ Deliverable 5 Surface Plasmon Photonics (SPP) NMP-CT-2003-505699 Page 2 of 11

Deliverable 5 Report providing specification of computational techniques appropriate for modelling fields associated with SPs.
_____________________________________________________________________________

Introduction/Overview

Exploiting the field enhancements associated with the SP modes of metallic nanostructures is one of the keys to obtaining a functional photonics based on SPs. Elsewhere in the project we are investigating such enhancements experimentally. However both for building understanding and for developing design tools it is important that we are able to numerically model the fields associated with SP modes on metallic nanostructures. Task 3.1 and part of Task 2.2 were to look at the best technique for this purpose this report critically examines these different approaches. This work is also of direct relevance to Task 3.2. Three techniques for computing the fields associated with SPs have been compared. The first two techniques, transfer matrix (TM) and multiple multi-poles (MMP), have been used before whilst the third, order-N, is a newer technique that may offer advantages for modelling the electromagnetic fields in a variety of candidate SP device structures. After testing all three methodologies we have established that the order-N technique is the most appropriate technique for modelling the fields associate with SP modes of nanostructured metals. This is an important conclusion in that it differs from what was expected at the outset of the project (Annex 1, Task 2.2). Despite being different it in no way upsets our programme of work, rather it is an advantage of the work carried out so far we have established a better technique that the one originally envisaged.

_____________________________________________________________________________________________ Deliverable 5 Surface Plasmon Photonics (SPP) NMP-CT-2003-505699 Page 3 of 11

2 Background. When modelling the electromagnetic (EM) fields associated with the excitation of surface plasmons supported by structured metal surfaces, we have to deal with Maxwells four equations which, in the absence of sources, take the form, D B (1) E = H = D = 0 B = 0 t t where D is the electric displacement vector related to the electric field vector E by the equation D = 0E (where is the local dielectric function of the medium) . In the set of equations (1), H is the magnetic field intensity related to the magnetic flux density B by B = 0 H (where is the magnetic permeability of the material). In the last fifteen years, theorists have been very active in developing methodologies to solve Maxwells equations in periodic dielectric media, mainly in connection with the emerging optical properties of the photonic crystals [1]. This has been a particular challenge owing to the vector nature of the electromagnetic field. The two main methodologies used to solve this type of problems are linked to the two classical models of electronic band structure: nearly-free-electron approximation and the tightbinding model. These two models start from completely different standpoints: the former assumes that plane waves are a good approximation to the electron wave function, the latter that electrons interact strongly with the atoms. The first calculations of photonic band structure in periodic dielectric media were based on plane-wave expansions that are the counterparts of the nearly-free-electron model. However, for the case of metals in which the dielectric constant depends strongly on frequency, the plane-wave algorithms are much more complicated and therefore more difficult to handle. For metallic materials an analogue of the tight binding approach is better; in this approach one represents the photon wave fields on a discrete lattice of points in real space. This approach was first proposed by Pendry and MacKinnon [2] for the case of periodic dielectric media. The main problem in representing differential equations on such a lattice is in finding suitable approximations for the field derivatives. For Maxwells equations we have another problem: they have the property that all longitudinal modes are dead modes, appearing only as zero-frequency solutions. In the Transfer Matrix framework this fundamental property is preserved [2], those techniques that do not accomplish this are prone to yielding false solutions.
_____________________________________________________________________________________________ Deliverable 5 Surface Plasmon Photonics (SPP) NMP-CT-2003-505699 Page 4 of 11

Different Techniques that can be employed

The Transfer Matrix (TM) technique is very well suited for analyzing the EM fields present at metal surfaces [2, 3]. In the past few years we have analysed the optical properties of 1Dstructured metal surfaces (reflection gratings [4] and arrays of subwavelength slits [5]) with the help of this theoretical framework. However, there is an additional problem when dealing with structured metal surfaces. Metals with finite conductivity have two length scales: the wavelength of radiation outside the metal (this is usually of the order of the period of the structure), and the much shorter skin depth inside the metal, these two lengths often differing by orders of magnitude. Dealing with this requires using a very dense mesh in order to describe the rapid variations of the EM fields at the interface between the (structured) metal surface and vacuum (or dielectric). When handling very large and complex metal structures in 3D (2Dstructured metal surfaces, for example), we have discovered that TM technique is not capable to giving sufficiently accurate results. This is because the computational time for the TM technique scales with the square of the number of discrete mesh points, i.e. goes as N resulting in a prohibitive time requirement for most calculations of interest. An alternative strategy we have been working on during this the first year of the project is the multiple multipole method (MMP). In this methodology, and in contrast to the plane wave/transfer matrix method, the basis functions are spherical functions (cylindrical for 2D problems). Details of this methodology applied to metallic photonic crystals can be found in Moreno et al. [6]. In essence the desired fields are built up using fictional sources. Within this framework we have analyzed different 1D structures. As an example of the effectiveness of this technique, in figure 1 we show the Poynting vector field at the resonant frequency in which the optical transmission for s-polarized light through an array of slits is greatly enhanced). Calculations such as this will help us explore where the energy is carried, i.e. to what extent energy is transported in the metal as against the dielectric. This is a key part of Task 2.2 since such information is vital in assessing which modes will exhibit the lower loss required for applications purposes.
2

(in

practice it is even worse than this). An accurate result requires using a very large mesh points

_____________________________________________________________________________________________ Deliverable 5 Surface Plasmon Photonics (SPP) NMP-CT-2003-505699 Page 5 of 11

Figure 1. Extraordinary optical transmission for the s-polarization case. A plane wave impinges (from above) on a structure made of a metallic film pierced by an array of slits, on top of a dielectric film. The plot shows the time-averaged Poynting vector field at the resonant frequency. The surface electromagnetic wave supported by the dielectric film mediates this resonant effect. This calculation is based on the multiple multi-pole technique.

However, as in the case of the transfer matrix method, we have found that in order to treat complex 2D structures on a metallic film, the number of multipole functions we have to introduce in our numerical code in order to obtain accurate results for the E-fields is so large that the time taken to perform the calculations is prohibitive. Another approach is to make use of a technique that has been known for some time in electrical engineering community as the finite difference time domain (FDTD) method. During this year, we have been working along this line by updating the FDTD based order-N code developed at Imperial College by Ward Pendry [7] to treat metallic systems. In the standard version of this numerical code, the local dielectric constant is assumed to be independent of the frequency of the radiation, an approximation that fails when trying to describe the optical response of a metal. To overcome this problem, we have followed the method of Luebbers et al. [8]. Basically, if the optical response of the metal is assumed to be described by the Drude formula:
2 p 2

( ) = 1

(2)

then the resulting equations allow to one to update the electric and magnetic fields in the time domain from a knowledge of the immediate past only; it is not necessary to store the entire previous history of the calculation in order to update the fields, as it is the case for a general
_____________________________________________________________________________________________ Deliverable 5 Surface Plasmon Photonics (SPP) NMP-CT-2003-505699 Page 6 of 11

( ) . Details of this procedure can be found in Arriaga et al. [9] for the case in which the
photonic band structure is required. For the calculation of reflection and transmission coefficients and the corresponding EM fields (the goal of this deliverable), the procedure is quite similar. During this year we have also been working along a completely different approach. Looking towards the future we envisage devices based on SPs. Then a theoretical description of the EM coupling between an incident wave (plane-wave, Gaussian beam, etc) and a finite collection of features on a metallic surface (rather than an infinite periodic array) will be essential. If one tries to solve this problem by using order-N (or FDTD) methodologies, there is a strong limitation on the number of indentations that can be described. During year 1 we have developed a formalism capable of treating the optical properties of even thousands of features (with any shape and place arbitrarily) in metal films [10]. This framework is a nontrivial extension of a simpler one developed in recent years for sets of 1D indentations [11, 12]. Here we describe the basic ingredients of this methodology. The only approximation in the formalism is that the metal is treated as a perfect conductor. For analyzing the properties of metal surfaces in the optical regime (the ultimate goal of the STREP project), we plan to extend this framework to include a more realistic dielectric function by using the surface impedance boundary conditions, thus fitting in well with Tasks 2.2, 2.3, 3.1 and 3.2. From our experience, the perfect conductor approach is an excellent starting point that has even semi-quantitative value in the optical regime for good conductors such as silver. Within this approach, the EMfields are first expanded in terms of plane waves in vacuum regions: | E ( z ) >=|k 0 0 > + rk | k > (Reflection region)
k

| E ( z ) >= t k | k >
k

(3) (Transmission region)

whereas the EM fields inside the indentations can be written in terms of the expansion coefficients ( A and B ): | E ( z ) >= | > A e iqz z + B e iq z z

(4)

By matching the EM fields appropriately on all interfaces (the equations related to the continuity of the E-field are projected onto the EM-eigenmodes of the indentations whereas the corresponding equations linked to the continuity of the H-field are projected onto plane waves),
_____________________________________________________________________________________________ Deliverable 5 Surface Plasmon Photonics (SPP) NMP-CT-2003-505699 Page 7 of 11

we end up with a set of linear equations for the expansion coefficients, A and B . By defining the quantities, E = A + B (modal amplitude of the E-field at the input surface of indentation
iq h iq h ' ) and E = [ A e z + B e z ] (modal amplitude of the E-field at the output surface of ' indentation ), it is possible to write down the set of linear equations for E , E as:

V ' [G ]E + G E G E = I

[G ]E' + G E' GV E = 0

(5)

The physical interpretation of the different terms appearing in this set of equations can be found elsewhere [10]. Next we concentrate on the results we can obtain by using this theoretical (and numerical) framework. For example, it is possible to analyze the evolution of the transmission of light through a 2D hole array as a function of the number of holes present in the structure. In the next figure we plot this evolution for holes of radius 135nm, the period of the arrays is 600nm and the thickness of the metal film is 320nm (these geometrical parameters correspond to typical values used by the ULP group in Strasbourg.

Figure 2. Calculated transmittance of a finite array of holes for a variety of array sizes, 5 x 5 holes, 11 x 11, 21 x 21 and 31 x 31 holes. The metal film thickness was 320 nm, the period 600 nm and the hole diameter 270 nm.

Also within this framework it is possible to reconstruct the EM-fields in real space. In the next figure we plot the intensity of the E-field at the exit plane of the 21x21 hole array for a
_____________________________________________________________________________________________ Deliverable 5 Surface Plasmon Photonics (SPP) NMP-CT-2003-505699 Page 8 of 11

wavelength equal to 630nm, the wavelength that corresponds to the resonant wavelength in the transmission spectrum:

Figure 3. Calculated field distribution of a finite array of holes for a variety of array sizes, 5 x 5 holes, 11 x 11, 21 x 21 and 31 x 31 holes. The metal film thickness was 320 nm, the period 600 nm and the hole diameter 270 nm.

4 Summary During this last year, as part of Tasks 2.2 and 3.1, we have checked the viability of three different numerical methodologies (TM, MMP and order-N) in order to describe the EM-fields associated to the excitation of SPs in structured metal surfaces. Our conclusion is that order-N is the best suited for achieving this goal. On the other hand, we have developed a new theoretical framework based on the modal expansion of the EM fields on structured perfect metal surfaces. Using this new methodology, we have been able to describe the optical properties of finite arrays of subwavelength holes perforated on metallic films. One goal of our future work will be to extend this formalism to treat real metals in the optical regime. If this can be accomplished then a new technique will have been developed that would be extremely important and useful in order to describe the huge concentration of light associated with the excitation of SPs. The different approaches are compared in the table below.

_____________________________________________________________________________________________ Deliverable 5 Surface Plasmon Photonics (SPP) NMP-CT-2003-505699 Page 9 of 11

Table 1

Comparison of numerical techniques

Technique Transfer Matrix (TM)


Is exact

Advantages
Solves 1D problems easily. Good for systems periodic

Disadvantages
3D problems are computationally very demanding High computational time Limited to very simple systems Very difficult to handle dispersive media, such as metals Requires care in setting up perfect absorbing boundary layers each new system studied has to be checked Relies (so far) on approximation of perfect metals

Multiple Multi-Poles Solves 1D problems exactly (MMP) Can treat finite systems with any
dielectric function

Order-N

Good, fast Can treat non-linear response Time evolution of fields allows entire frequency response to be calculated via Fourier transform Can handle complex structures

Finite Array

Very flexible as to structures considered Can handle large structures Good speed

_____________________________________________________________________________________________ Deliverable 5 Surface Plasmon Photonics (SPP) NMP-CT-2003-505699 Page 10 of 11

6 1 2 3

References "Photonic Crystals" J. Joannopoulos, R. Meade and J. Winn Princeton University Press, (1995). "Calculation of photon dispersion relations" J. B. Pendry and A. MacKinnon Physical Review Letters, (1992), 69, pp 2772-2775 "A program for calculating photonic band structures and transmission coefficients of complex structures" P. M. Bell, J. B. Pendry, L. Martn-Moreno and A. J. Ward Computer Physics Communications, (1995), 85, pp 306-322 "Surface shape resonances in lamellar metallic gratings" T. Lpez-Rios, D. Menoza, F. J. Garca-Vidal, J. Snchez-Dehesa and B. Pannetier Physical Review Letters, (1998), 81, pp 665-668 "Transmission resonances on metallic gratings with very narrow slits" J. A. Porto, F. J. Garca-Vidal and J. B. Pendry Physical Review Letters, (1999), 83, pp 2845-2848 "Band structure computations of metallic photonic crystals with the multiple multipole method" E. Moreno, D. Erni and C. Hafner Physical Review B, (2002), 65, pp 155120 "A program for calculating photonic band structures, Greens functions and transmission/reflection coefficients using a non-orthogonal FDTD method" A. J. Ward and J. B. Pendry Computer Physics Communications, (2000), 128, pp 590-621 "A frequency-dependent finite-difference time-domain formulation for transient propagation in plasma" R. J. Luebbers, F. Hunsberger and K. S. Kunz IEEE Transactions on Antennas and Propagation, (1991), 39, pp 29-34 "Order N photonic band structures for metals and other dispersive materials" J. Arriaga, A. J. Ward and J. B. Pendry Physical Review B, (1999), 59, pp 1474-1477 "Resonant transmission of light through finite chains of subwavelength holes" J. Bravo-Abad, F. J. Garca-Vidal and L. Martn-Moreno Physical Review Letters, (2004), 93, pp 227401 "Theory of highly directional emission from a single sub-wavelength aperture surrounded by surface corrugations" L. Martn-Moreno, F. J. Garcia-Vidal, H. J. Lezec, A. Degiron and T. W. Ebbesen Physical Review Letters, (2003), 90, pp 167401 "Multiple paths to enhance optical transmission through a single subwavelength slit" F. J. Garca-Vidal, H. J. Lezec, T. W. Ebbesen and L. Martn-Moreno Physical Review Letters, (2003), 90, pp 213901

4 5 6

9 10 11

12

_____________________________________________________________________________________________ Deliverable 5 Surface Plasmon Photonics (SPP) NMP-CT-2003-505699 Page 11 of 11

You might also like