Surface plasmon polaritons (SPPs) are surface waves [1–3] propagating along the interface between two media, usually a dielectric and a metal. Their field distribution extends into both media and is evanescent in the direction perpendicular to the interface. In the case of a homogeneous flat metallic film, SPP modes are delocalized over the plane. The control of SPPs flow has so far been mostly based on concepts directly inspired from integrated optics with dielectric waveguides. The most intuitive approach consists of introducing geometrical boundaries by patterning the metal layer. This has for instance enabled subwavelength guiding of SPPs along strips [4, 5], V-grooves and slots [6–8] as well as adiabatic focusing [9, 10].
Alternatively, controlling SPP modes can be done by spatially engineering its effective refractive index by either using a finite dielectric load [11–15] or structuring the metal surface. For instance, Pendry et al. showed in 2004 that one can manipulate surface plasmon ad libitum via homogenization of structured surfaces . This powerful concept has enabled the experimental demonstration of broadband extraordinary transmission in the visible regime by means of plasmonic checkerboards [17, 18].
More recently, the field of metamaterials has inspired a new form of SPP control. In 2008, Smolyaninov et al. achieved a noticeable reduction in the scattering of SPPs incident upon a cloak consisting of polymethylmethacrylate (PMMA) at a wavelength of 532 nm [19, 20]. More recently, Baumeier et al. demonstrated theoretically that it is possible to reduce significantly the scattering of an object by an SPP at a wavelength of 633 nm when it is surrounded by two concentric rings of point scatterers .
The recent proposal of transformation optics (TO) [22–30] has brought a novel way of controlling flows of photons in any desired way. The TO concept has for instance permitted us to achieve unique light control including the concept of invisibility [31–44], lenses [45–48], beam-shifter, splitters and bends [49–54], or more exotic applications such as electromagnetic wormholes [55, 56], hidden pathways [57, 58], or optical black holes [59–66]. Moreover TO can further be used to completely change the appearance of an object thereby creating an illusion [67–70], or to enable subwavelength imaging via hyperlenses that transform the evanescent waves [71–75]. TO often takes advantage of the field of metamaterials [76–78] to implement complex electromagnetic parameters. Metamaterials are artificial materials made up of subwavelength constituents that are designed to effectively mimic a prescribed response to electromagnetic fields and that often make use of plasmonic structures such as plasmonic nanomaterials, multilayer SPPs and so on . On the other hand, we note in passing that plasmonic routes to cloaking based on core shell resonances induced by a coating of low refractive index (scattering cancellation ) and negative refractive index (external cloaking ) have also been proposed.
Whereas at first TO was applied to plane waves, it has been recently shown that TO can also be used to effciently mould the flow of surface plasmon polaritons (SPPs) at metal-dielectric interfaces [82–88]. Using TO to manipulate SPPs is a very general technique that has already led to a wide variety of proposals for plasmonic elements which might even be based on single-atom-thin graphene layers . Moreover, TO has also been applied to analytically handle localized surface plasmons in metallic nanostructures [90–94] including non-local effects .
In the present article we review the state of the art in the exciting field of transformation plasmonics. After presenting how TO can be extended to SPP, we illustrate numerically its capability to fully control the SPP flow by loading the metal surface with a patterned dielectric film. We finally review recent experimental implementations in which transformation plasmonics is applied to broadband plasmonic invisibility and focusing.
2. Formalism of transformation plasmonics
2.1. Transformation optics: coordinate changes in Maxwell’s equations
Conceptually, TO is equivalent to warp space with the aim to control trajectories of light rays. While space can not be transformed in this manner, TO alternatively makes coordinates transformations, which in practice, provide the local optical properties to be fulfilled by the distorted space to achieve the desired light flow. This is a subject which has received steadily more attention from 2006 onwards, thanks to the paradigm of the invisibility cloak. However, this topic got started in the mid 1990s with the original aim to reduce the complexity of computational problems [96–98]. The cornerstone of TO was to map, for instance, an infinite domain on a finite region of space, which would then be considered in the numerical package. The point of view of TO was central in the development of finite elements models for twisted microstructured fibers  as their modeling involves twisted perfectly matched layers.
As we start with a given set of equations on a given domain, it seems at first sight that we have to map this domain on a new one. Nevertheless, it is the opposite that has to be done, as shown in Figure 1. On the original domain, see Figure 1B, the equations are described in a particular coordinate system (x, y) (taken here by default to be Cartesian coordinates). We want to establish a one-to-one correspondence with a new domain where we use a general (possibly non-orthogonal) coordinate system (u, v), see Figure 1, panels (A) and (D), (which correspond respectively to a rotation and a compression of some region of the original domain). More precisely, consider a map from the curvilinear coordinate system (u, v) to the initial one (x, y) given by the transformation x(u, v) and y(u, v). Here, we must point out that, we are mapping the transformed domain and coordinate system [Figure 1 (A) or (D)] onto the initial one [Figure 1 (B)] with Cartesian coordinates. This change of coordinates is characterized by the transformation of the differentials through the Jacobian matrix J described in Figure 1(C).
Importantly, such a geometric transformation also affects the material properties in the transformed coordinates which are encompassed in the constitutive Maxwell’s equations. More precisely, in electromagnetism, changes of coordinates amount to replacing the different materials (often homogeneous and isotropic, which corresponds to the case of scalar piecewise constant permittivity and permeability) by equivalent inhomogeneous anisotropic materials. Mathematically speaking, the permittivity and permeability are rank-2 anisotropic heterogeneous tensors in the transformed coordinates:
J is the Jacobian matrix of the transformation, see Figure 1, and T is the metric tensor in the transformed coordinates. In Eq. 1, the right hand sides involve matrix products where the matrix associated with the second rank tensor T contains the coeffcients of its representation in the initial Cartesian coordinate system (x, y).
Let us make use of the invisibility cloak to illustrate the concept of TO. One way to conceal a region of space is mapping a disc or radius r<R2 onto a corona R1<r′<R2, as shown in panels (A) and (B) of Figure 2 (note that r is replaced by r′ in the corona, since we have now a stretched radial coordinate). This mapping can be done by means of a linear transformation, r′=R1+r(R2-R1)/R2, as proposed by Pendry et al.  (a transformation formally used in the context of inverse problems in tomography ). The effect of this transformation on the metric of space is shown in Figure 2(C): the metric is compressed in the corona surrounding the concealed region in panel (B). The obtained matrix provides the new coeffcients of the tensor corresponding to the equivalent material in Figure 2 (B). Note that the transformed grid in panel (B) does not have right angles as the one in panel (A): the transformed material is anisotropic, as can be clearly seen in Figure 2 (C). It should be noted that the off-diagonal elements of T are identical and therefore T can be diagonalized. Importantly, all diagonal elements have positive entries (negative values in the off-diagonal elements is just an effect of the rotation required to express T in a diagonal basis), however they either vanish or become extremely large on the inner cloak boundary, which is consistent with the fact that detoured waves, as shown in Figure 2(D) for a source located in the close neighborhood of a two-dimensional cloak, have to be accelerated in the cloak in order to catch up with their wavefront outside the cloak.
2.2. From transformation optics to transformation plasmonics
Let us now apply TO to the control of surface wave trajectories. A few words on the mathematical setup are necessary in order to understand the specificity of the transformation plasmonics design, which involves anisotropic heterogeneous media. To simplify the discussion, we consider a transverse magnetic (p-polarized) SPP propagating in the positive x-direction at a flat interface z=0 between a (isotropic homogeneous) metal (z<0) and a transformed medium (z>0) described by diagonal tensors of relative permittivity and permeability
Note here that Re(kz,1) and Re(kz,2) are strictly positive in order to maintain evanescent fields above and below the interface z=0.
For this field to be solution of Maxwell’s equations, continuity of its tangential components is required across the interface z=0, which translates into mathematics as kx,1=kx,2=kx and this in turn leads us to a local dispersion relation for a p-polarized SPP:
Here c is the speed of light in vacuum, ε1(x, y)>1 (z>0), and ε2 is the usual Drude form in the metal for z<0. Actually, we note that since kz,1 and kz,2 are strictly positive, SPPs can only exist if ε1 and εxx,2 are of opposite signs.
Equation 3 can be recast in a more compact form :
which should be satisfied for SPPs to be able to propagate on a flat interface between a metal substrate and a transformed medium. Indeed, SPPs are bound to the interface, hence, they do not belong to the radiative spectrum.
Invisibility carpets were an early proposal of transformation plasmonics [82–88] based on ground-plane cloaks . As shown in Figure 3, the invisibility carpet flattens the SPPs wavefronts reflected by a curved cylindrical mirror on a flat metal surface. The mapping that makes this possible is given by x′=x[x2(y)-x1(y)]/x2(y)+x1(y), which maps the region 0<x<x2(y) onto x1(y)<x′<x2(y) for a<y<b and z>0. The vertical coordinate z appears trivially in the definition of the carpet as it is of cylindrical geometry, but more complex geometries can be explored, and for these we refer to the next section. The effect of the mapping on the spatial distribution of the metric tensor is shown in panel (B) of this figure, and the corresponding plot of the y-component of the magnetic field for a SPP field incident upon such a plasmonic carpet is shown in panel (C). Interestingly, this metamaterial can be designed with dielectric conical pillars and further tested numerically for different wavelengths such as at λ=700 nm (see Figure 10 E). An experimental demonstration of such theoretical concepts is shown in section IV B.
Now, if a perturbation on the profile of the flat surface is introduced, for instance a bump as in Figure 6, the former dispersion relation Eq. 4 changes and SPPs move to the radiative spectrum and so scattering cannot be avoided. The dispersion relation Eq. 4 becomes slighlty more complex in the case of curved surfaces (see Ref.  for a derivation in the weakly curved configuration), which is the bottom line of transformation plasmonics on curved surfaces, first introduced in Ref.  and : enabling SPPs to propagate free of scattering on curved surfaces, hence on longer distances, is a great challenge. This case will be treated in the next section.
3. Design of plasmonic functionalities based on transformation optics
In this section we exemplify the theory by means of different plasmonic devices that are designed using the TO tools. Here we concentrate on four devices that control the propagation of SPPs at a metal surface: the cylindrical cloak for SPPs [82, 83], the plasmonic right-angle bend , the ground-plane cloak for SPPs [83–86] and the plasmonic Luneburg lens [48, 84]. In order to numerically demonstrate the application of the TO framework to SPPs, we use a finite element method solver .
Since SPPs are surface waves propagating along the interface between two media, its field distribution extends into both media and is evanescent in the direction perpendicular to the interface. Therefore, in order to operate over the whole plasmonic field, the expressions for the electromagnetic (EM) material parameters,
3.1. 3D Cylindrical cloak
As a first illustration of the methodology, let us consider a three-dimensional (3D) cylindrical cloak for SPPs traveling along the interface between a metal and a dielectric. The plasmonic cloak is based on the twodimensional (2D) invisibility cloak that was first proposed within the general TO framework [22, 23] and had an early experimental realization . The purpose of the 2D cloak was to hide to an external observer any object placed inside it, in a circle of radius R1, and the cloak itself [see Figure 4 (A)]. This is done by transforming from a Cartesian space where light propagates in straight lines to a space with a hole of radius R1 surrounded by a compressed region of radius R2. The radial transformation r′(r)=R2/(R2-R1)r+R1 is performed in the cloak region, that corresponds to the shell R1<r′<R2. The EM parameters resulting from the TO approach in a cylindrical basis are
When these parameters are implemented in the cloak, a plane wave traveling through it will follow trajectories that surround the hole and the wave will appear to have traveled through empty space. Note that the required permittivity and permeability are anisotropic, inhomogeneous and take singular values at the inner interface of the cloak.
Let us now discuss the 3D plasmonic cloak, which suppresses the scattering of SPPs propagating along an airgold interface from a metallic cylinder of radius R1. As sketched in Figure 4 (A), the SPP is scattered when it encounters the cylinder on top of the gold surface. Designing the 3D cloak involves extending the TO procedure explained above for the 2D cloak to 3D. In particular, the described 2D transformation should be made in an infinite number of planes parallel to the air-gold interface. In order to operate over the whole plasmonic field, we need to cloak the SPP in the dielectric and metal sides, as illustrated in the top right sketch in Figure 4 (A). In the dielectric side, the cloak is a cylindrical shell of radii R1 and R2 characterized by the EM tensors given by Eq. 5:
Figure 4 shows the results obtained from 3D simulations of the cylindrical cloak for a SPP at λ=800 nm traveling on an air-gold interface. The radius of the metallic cylinder is R1=800 nm while the outer radius of the cloak is R2=2000 nm. Panel (B) depicts the power flow streamlines in a top view for two situations: a metallic cylinder scattering the SPP (left sketch) and the cloak guiding the SPP around the metallic cylinder (right sketch). Panels (C) and (D) show a top and a lateral view, respectively, of the y component of the magnetic field of the SPP when the cylinder is cloaked. These images, together with the 3D view of the simulation [panel (E)] show how the SPP field is smoothly guided around the metallic cylinder avoiding any scattering from it.
3.2. Right angle bend
As a second example of plasmonic devices designed using TO, we present the right-angle bend for SPPs. This device rotates the propagation direction of an incident SPP by 90°, as shown in the sketch in Figure 5 (A). Similarly to the cylindrical cloak, the design of the 3D rightangle bend is based on 2D transformations parallel to the interface plane. In order to bend the SPP propagation direction from the x direction to the y axis, each transformation consists of mapping a squared region of side b in a Cartesian grid to polar grid with a squared crosssection of the same size. This transformation is given by
In order to avoid the singularity of these parameters at r′=0, we limit the radius between r′=a and r′=b. The curvature radius of the bend is defined as ρ=(a+b)/2 and its width by Δ=b-a.
A SPP right-angle bend operating at λ=800 nm is shown in Figure 5 (A). The SPP propagates in the x direction and enters a bend of curvature radius ρ=2 μm. The EM parameters given by Eq. 6, which are anisotropic and inhomogeneous, are implemented in a cylindrical shell comprised between r′=a, r′=b, ϕ′=0, ϕ′=90° in the dielectric side of the interface. From the power flow stream lines (in white) and the shape of the wave-fronts (color scale) it can be interpreted that the SPP field is bended by 90°. The transmittance through the bend, defined as the fraction of the SPP power flow in the exit face compared to the entrance face, is 98%. The lacking 2% corresponds to the field contained in the metal, which is left untouched.
As we have seen for the plasmonic cloak, TO leads to a perfect functionality when we operate over the whole plasmonic field by introducing a transformation medium in the dielectric and metal sides. However, an experimental realization would require a control of both the dielectric and metal EM properties at the nanoscale. This challenge can be circumvented by an approximate approach to TO in which the metal side is left untouched. As we have shown here for the plasmonic bend at 800 nm, manipulating only the dielectric side gives very good results [see Figure 5 (A)]. This model, that is more realistic, has been shown to lead to quasi-perfect performance over broad frequency ranges and for a wide variety of plasmonic devices . The reason for this is that most of the SPP energy is contained in the dielectric since the SPP decay length at this side of the interface is always much larger than in the metal side. Hence, this approximation is more accurate for longer wavelengths, where the skin depth of the metals is almost negligible.
A beam bend with the same functionality as the one we have discussed can also be realized out of isotropic materials. The idea is to use conformal mappings, a specific kind of coordinate transformations that yield grids that are orthogonal, i.e., the angles between coordinate lines are preserved after the mapping. This means that these kind of transformations minimize the anisotropy of the grid. For problems where the two polarizations can be treated separately, such as SPPs, a 2D conformal transformation leads to an isotropic transformation medium. Therefore, a design of the bend based on a conformal transformation replaces the anisotropic EM parameters in Eq. 6 by an isotropic refractive index:
The refractive index for a bend of the same geometrical parameters as before, which is plotted in the inset panel of Figure 5 (B), varies from 4.46 at the inner radius a to 0.67 at the outer radius b. As illustrated in the panel, the plasmonic device is a cylindrical shell comprised between a<r′<b and 0<ϕ′<90°, with a height hd, placed on top of the gold surface and characterized by the n(r) profile. The electric field (color scale) and the power flow (white lines) of a SPP traveling through the bend at λ=800 nm is plotted in Figure 5 (B), showing how the propagation direction of the SPP smoothly rotates by 90°. In this case, the transmittance through the bend is 95%, slightly lower than for the anisotropic bend. The reason for this is that the conformal transformation causes an impedance mismatch at the interfaces of the device (see the inset panel) that results in reflections. This implementation is a more feasible approach for bending the propagation direction of a SPP since it relies on the manipulation of non-magnetic isotropic dielectrics.
3.3. Ground-plane cloak
The TO-based approach to the design of plasmonic devices can also be applied to minimizing the scattering to the far-field of SPPs on irregular surfaces. Uneven surfaces result in strong scattering that lead to radiation of the SPP into free-space, as illustrated in Figure 6(A), where a SPP traveling along an air-gold interface encounters an obstacle. Here we model the irregularities of the metal surface as a bump of length l and height h0, and with a shape given by
The design of the anisotropic cloak involves transforming a rectangular region of a Cartesian space sized l×h into a region of the same shape except that its bottom boundary follows the shape of the bump. The simplest mapping that can be done between these two spaces is a transfinite transformation: x′=x, y′=y and
On the other hand, the same functionality can be achieved by means of an isotropic cloak. In order to design the isotropic ground-plane cloak, the transfinite transformation has to be replaced by a quasiconformal one. This kind of transformation is an approximation to a conformal mapping and it minimizes anisotropy too. It is generated by numerically solving Laplace’s equation with sliding boundary conditions in the region defined by the cloak, which in this case is sized 6 μm×2.5 μm. In contrast to the previous transformation, and as can be seen in panel (C), the new coordinate grid (gray lines) preserves the right-angles when the TO procedure is followed. This fact results in an isotropic refractive index n(x, y) for the cloak illustrated in panel (D). The index profile is smooth, ranging from 0.82 to 1.38, and it recovers the background value n=1 at the outer boundaries of the cloak. The wave fronts of SPPs traveling through the cloak follow the transformed grid while it smoothly goes through the bump without scattering, as illustrated for λ=700 nm in panel (C).
Using the same tools, a ground-plane cloak can also be devised for hiding a 3D bump on a metal surface. Let us consider the bump shown in Figure 6 (E), which is described by
3.4. Plasmonic Luneburg lens
Finally, we present as a last example two designs for a plasmonic Luneburg lens for SPPs propagating on a metal surface. A planar Luneburg lens is a circle of radius R characterized by the following refractive index:
where r is the distance to the center of the lens. This index profile focuses an incident SPP to a point in the perimeter of the lens on the opposite side of the circle. In a first design of the plasmonic Luneburg lens, let us assume a metamaterial that reproduces the refractive index n(r). As sketched in Figure 7 (A), the metamaterial is a cylinder of radius R placed on top of a gold surface with a height that needs to be larger than the SPP vacuum decay length, similar to the previous examples. Panel (B) demonstrates a plasmonic Luneburg lens of radius R=2.5 μm operating at λ=800 nm when the spatial variation of the index n(r) [shown in (C)] is implemented in the metamaterial region. The SPP electric field and power flow lines obtained from 3D simulations show the focusing of SPPs traveling along an air-gold interface at the perimeter of the lens.
A different strategy for designing this plasmonic element that involves the effective mode index of SPPs has been used to experimentally demonstrate the Luneburg lens . Instead of spatially varying the optical properties of the dielectric, the height of a thin dielectric layer on top of the metal is tapered. The index profile given by Eq. 10 is reproduced by the effective mode index of the SPP. The mode index for an air/PMMA/gold structure for SPPs at λ=800 nm is plotted in Figure 7 (D) as a function of the PMMA height, h. In a dielectric/dielectric/metal structure with electric permittivities ε1, ε2 and ε3, respectively, SPPs at frequency ω propagate with an effective mode index neff=k/k0 determined by the parallel wave-vector k. Here, k0=ω2/c2 and c stands for light speed. The SPP dispersion relation is given by:
In our situation, ε1=1, ε2=2.19 and ε3=εAu. For PMMA thicknesses up to 250 nm, we see in panel (B) that the index varies between 1 and 1.5, a range that is enough to reproduce the index required by the Luneburg lens. The plasmonic device, sketched in panel (E), is designed by calculating the PMMA’s height at which the SPP effective mode index satisfies Eq. 10. This results in a variation of the PMMA height between 0 at the perimeter of the lens and 130 nm at the center. A numerical simulation of this lens can be seen in Figure 7 (F). A PMMA disk of 2.5 μm radius with a height modulated according to panel (E) was placed on top of a gold surface. The SPP field is smoothly focused at the perimeter of the lens in a subwavelength spot. This design of the plasmonic lens potentially allows for a broad band performance, as it only involves dielectric materials with a very low dispersion.
These different examples, illustrate the power of transformation plasmonics to fully control SPP and design complex functionalities that would not be achievable otherwise. However, the experimental implementation of such control is conditioned by the possibility to match the required structural parameters. In the next section, we discuss recent experimental advances towards the realization of complex SPP functionalities giving special emphasis on the different fabrication strategies.
4. Transformation plasmonics for real: experimental implementation
4.1. Fabrication and optical characterization methods
The experimental implementation of Transformation Plasmonics at optical frequencies requires engineering on the subwavelength scale the effective mode index neff of the propagating SPPs. In practice, this can be achieved in two different ways:
The first approach is based on a discrete patterning of a dielectric material lying on top of the flat metal film. This leads to an artificial- or meta-medium having a mean refractive index that affects neff of the SPPs as shown by the blue curve in Figure 8. The elementary unit can either be an air hole in a continuous dielectric film as originally proposed in  for optical waveguides or a dielectric protrusion in air . In this case, the local refractive index is spatially adjusted by controlling the units size and/or density [see Figure 9 (C) and (D)]. In practice, this is achieved using standard fabrication tools such as electron-beam lithography and Focused Ion Beam (FIB). The minimum size of the elementary unit as well as the minimum distance between adjacent units determine the smoothness of the actual neff profile. Maximizing the smoothness is crucial as it has direct impact on the back reflection and scattering of the incident SPP. Such parasitic reflection and scattering would for instance directly compromise the effciency of an invisibility cloak or carpet.
Alternatively, the desired neff landscape can be obtained by gradually varying the thickness of a continuous dielectric load  as shown by the red curve in Figure 8. In their recent work, Zentgraf et al.  used 3D electron beam lithography to gradually shape a PMMA load on gold into a patch that fulfills the requirements of a Eaton and Luneburg plasmonic lens [Figure 9 (A–B)].
In both approaches, the accessible range for the effective mode index neff is determined by the dielectric properties of the surrounding media and the dielectric being used. While the use of high refractive index materials a priori gives more flexibility, the propagation length of SPPs ΛSPP=1/(2 Im[neffk0]) decreases for increasing values of neff, thereby adding further constrains in the design.
Because SPPs are surface waves, bound to the metal surface, the optical characterization of plasmonic functionalities requires suitable methods capable to monitor, either directly or indirectly, the interaction of the incident SPP with the metasurface. The simplest approach consists in measuring input and output signals without directly monitoring the interaction region [35, 102]. Such a scheme can for instance be implemented by including input and output grating couplers at strategic locations of the sample. However, it is blind to what actually occurs in the interaction region and does not apply to any functionality. Further insight into the metasurface properties can be gained by employing scanning near-field optical microscopy (SNOM)  which can directly map with a sub-wavelength spatial resolution the optical near-field intensity of plasmon fields at any point of the metasurface. Special attention is though required to prevent significant influence of the scanning probe on the intrinsic properties of the structure under study [103–105]. Alternatively to SNOM, Leakage Radiation Microscopy offers a simpler and faster characterization as long as neff is smaller than the refractive index of the substrate. Under this condition, SPPs are subjected to leakage radiation into the substrate. Although diffraction limited, collecting this leakage radiation offers a wide field, probe-less and quantitative method [106–108] to map the SPP propagation at the metasurface.
4.2. Experimental implementations
In the following we briefly discuss recent experiments that illustrate the power of transformation plasmonics to design complex plasmonic functionalities.
4.2.1. Broadband plasmonic invisibility carpet at visible frequencies
The first example tackles the concept of invisibility carpet and demonstrates experimentally that an object can be concealed from a SPP propagating along a gold/air interface at visible frequencies, as we numerically demonstrated in Figure 3.
In order to meet experimentally the parameters of the simulations, a flat gold surface was structured with TiO2 protrusions. The TiO2 pillars forming the crescentmoon-like carpet were first fabricated on top of a 60-nm thin Au film by combining electron-beam lithography and reactive-ion etching. In a second lithography step, a curved Bragg-type reflector [formed by 15 gold lines (section=150 nm×150 nm) periodically separated by half of the SPP wavelength] was added, acting as the object to be hidden behind the carpet Figure 9 (C) and (D). Each of the small cones of TiO2 are placed at the nodes of a quasi-conformal grid, what allows to minimize the required anisotropy of the metric tensor associated with the carpet shown in Figure 3. This simple trick greatly simplifies the experimental design.
In the optical experiment, the SPPs were launched by focusing the incident beam from a CWTi:Sapph laser on a defect line placed 44 μm away from the reflector. Taking advantage of the asymmetry of the dielectric environment on both side of the gold film, the leakage radiation from the SPPs at the gold/air interface was collected using a high-numerical aperture objective to map the SPP fields. Additionally for the sake of clarity, spatial filtering in the conjugated Fourier-plane was performed to suppress the direct transmitted light from the excitation spot and scattered light in order to isolate the carpet properties.
Leakage radiation microscopy (LRM) was first used to map the distribution of λ=800 nm SPPs propagating at the gold/air interface and interacting with the different structures fabricated at the gold surface. In the case of a bare curved Bragg-reflector, the reflected SPPs propagate into different directions depending on their relative angle to the normal to the mirror lines [see Figure 10 (B)], thus leading to a curved wave front. Conversely, adding the crescent-moon-like TiO2 carpet re-establishes a fringe pattern with a nearly straight wave front [see Figure 10 (D)] similar to the case of a flat Bragg-mirror (A). The remaining small lateral modulations were attributed to imperfections in the manufacturing. Quantifying the modification in the wave front curvature induced by the presence of the crescent-moon-like TiO2 carpet and comparing the area under the numerically averaged curves with the curved obtained with the reference structure without carpet leads to a reduction by a factor 3.7 as shown in Figure 10 (E). Further, the experiment was repeated at different incident wavelengths across the range of the Ti:Sapph laser in order to test the broadband operation of the carpet. The data summarized in Figure 10 (E) show that the structure remains effcient between 750 nm and 850 nm, demonstrating for the first time broadband invisibility in the true visible range.
4.2.2. Luneburg and Eaton plasmonic Lens
Prior experimental studies demonstrated that the change in the effective refractive index neff induced by patterning a metal surface with a dielectric load of constant height can be used to control the propagation of SPPs by exploiting the fact that the SPP propagation at the load boundaries fulfills the 2D equivalent of the Snell-Descartes laws. Through a suitable design of the load geometry, SPPs can thus be guided [12, 13, 15], reflected [109, 110] and focused . Recent experiments by the Zhang’s group  further pushed the analogy between 3D optical and plasmonic elements by using transformation plasmonics. By gradually varying the height of a PMMA disk lying on top of a thin gold film using 3D e-beam lithography (see Figure 9), the authors first built a Luneburg plasmonic lens whose focal point lies on the perimeter of the lens. The fabricated structure was optically tested by combining fluorescence imaging and LRM. Their data, summarized in Figure 11, show a good agreement with their numerical simulation (A), efficient focusing in the near-infrared region of spectrum, between λ=770 and 840 nm (B). To demonstrate the versatility of their approach, the authors also successfully demonstrated the concept of Eaton plasmonic lens that bends the SPP through 90°.
5. Conclusion and prospects
Transformation plasmonics complements the existing tool box available to researchers in the field of plasmonics by providing further control over surface plasmons. Its full determinism potentially enables to tame plasmonic fields at will and this way design functionalities that were not reachable so far. First experimental realizations successfully demonstrated the practical feasibility of this new approach to control the propagation of SPPs at a patterned metal surface. Similarly to what is currently experienced in the field of metamaterials, a full exploitation of transformation plasmonics at optical frequencies remains though hindered by current limitations in nanofabrication. Beyond extending the range of plasmonic functionalities based on SPP and optimizing their effciency, a higher fabrication accuracy should enable to extend TO to the nanoscale light control based on Localized Surface Plasmons supported by 3D metallic nanostructures [93, 94, 111, 112]. Another important future development would consist in combining transformation plasmonics with recent advances in active plasmonics [113–115] by using active materials instead of passive dielectrics in order to introduce a dynamic control through an external electrical or optical signal.
This work was sponsored by the Spanish Ministry of Science under projects CSD2007-046-NanoLight.es and MAT2011-28581-C02-01. P.A.H. acknowledges financial support from the Spanish Ministry of Education through grant no. AP2008-00021. J.R. and R.Q. acknowledge financial support by the european commission through grant ERC-St Plasmolight (259196) and the Fundació Privada CELLEX.
Abelès F, Lopez-Rios T. Surface polaritons at metal surfaces and interfaces. In: Agranovich VM and Mills DL, editors. Surface ploaritions. Amsterdam: North-Holland Publishing Company; 1982, pp. 239–74.Google Scholar
Raether H. Surface plasmons. Berlin: Springer-Verlag; 1988.Google Scholar
Weeber J-C, Krenn JR, Dereux A, Lamprecht B, Lacroute Y, Goudonnet JP. Near-field observation of surface plasmon polariton propagation on thin metal stripes. Phys Rev B 2001;64:045411–1–9.CrossrefGoogle Scholar
Bozhevolnyi SI, Volkov VS, Devaux E, Laluet J-Y, Ebbesen TW. Channel plasmon subwavelength waveguide components including interferometers and ring resonators. Nature 2006;440:508–11.CrossrefGoogle Scholar
Moreno E, Rodrigo SG, Bozhevolnyi SI, Martín-Moreno L, García-Vidal FJ. Guiding and focusing of electromagnetic fields with wedge plasmon polaritons. Phys Rev Lett 2008;100:023901.CrossrefGoogle Scholar
Steinberger B, Hohenau A, Ditlbacher H, Aussenegg FR, Leitner A, Krenn JR. Dielectric stripes on gold as surface plasmon waveguides: bends and directional couplers. Appl Phys Lett 2007;91:081111.CrossrefGoogle Scholar
Grandidier J, Massenot S, Colas des Francs G, Bouhelier A, Weeber J-C, Markey L, Dereux A, Renger J, González MU, Quidant R. Dielectricloaded surface plasmon polariton waveguides: figures of merit and mode characterization by image and Fourier plane leakage microscopy. Phys Rev B 2008;78:245419.CrossrefGoogle Scholar
Ramakrishna SA, Mandal P, Jeyadheepan K, Shukla N, Chakrabarti S, Kadic M, Enoch S, Guenneau S. Plasmonic interaction of visible light with gold nanoscale checkerboards. Phys Rev B 2011;84:245424.CrossrefGoogle Scholar
Kwon D-H, Werner DH. Transformation electromagnetics: an overview of the theory and applications. Antennas and Propagation Magazine, IEEE 2010;52:24–46.Google Scholar
Zhai YB, Ping XW, Jiang WX, Cui TJ. Finite-element analysis of three-dimensional axisymmetrical invisibility cloaks and other metamaterial devices. Commun Comput Phys 2010;8:823–34.Google Scholar
Vasic B, Gajic R, Isic G, Hingerl K. Confined metamaterial structures based on coordinate transformations. Acta Phys Polonica A 2009;116:96.Google Scholar
Kadic M, Dupont G, Guenneau S, Enoch S. Plasmonic wormholes: defeating the early bird. arXiv:1102.2372v1, 2011.Google Scholar
Rahm M, Schurig D, Roberts DA, Cummer SA, Smith DR, Pendry JB. Design of electromagnetic cloaks and concentrators using form-invariant coordinate transformations of maxwells equations. Photonics Nanostruct Fundam Appl 2008;6:87–95.CrossrefGoogle Scholar
Kadic M, Dupont G, Chang T-M, Guenneau S, Enoch S. Curved trajectories on transformed metal surfaces: beam-splitter, invisibility carpet and black hole for surface plasmon polaritons. Photonics Nanostruct Fundam Appl 2011;9:302–7.Google Scholar
Nicolet A, Zolla F, Geuzaine C. Generalised cloaking and optical polyjuice. arXiv:0909.0848v1, 2009.Google Scholar
Engheta N, Ziolkowski RW. Metamaterials: physics and engineering explorations. Wiley-IEEE Press; 2006.Google Scholar
Ramakrishna SA, Grzegorczyk TM. Physics and applications of negative refractive index materials. CRC Press; 2008.Google Scholar
Maier SA. Plasmonics: fundamentals and applications. Springer Verlag; 2007.Google Scholar
Fernández-Domínguez AI, Wiener A, García- Vidal FJ, Maier SA, Pendry JB. Transformation-optics description of nonlocal effects in plasmonic nanostructures. Phys Rev Lett 2012;108:106802.CrossrefGoogle Scholar
Zolla F, Renversez G, Nicolet A, Kuhlmey B, Guenneau S, Felbacq D. Foundations of photonic crystal fibres. Imperial College Press; 2005.Google Scholar
COMSOL. Multiphysics 3.5a. http://www.comsol.com.
Novotny L, Hecht B. Principles of nano-optics. Cambridge University Press; 2006.Google Scholar
Drezet A, Hohenau A, Koller D, Stepanov A, Ditlbacher H, Steinberger B, Aussenegg FR, Leitner A, Krenn JR. Leakage radiation microscopy of surface plasmon polaritons. Mater Sci Eng B 2008;149:220.CrossrefGoogle Scholar
Hohenau A, Krenn JR, Drezet A, Mollet O, Huant S, Genet C, Stein B, Ebbesen TW. Surface plasmon leakage radiation microscopy at the diffraction limit. Opt Express 2011;19:25749–62.CrossrefGoogle Scholar
Krasavin AV, Randhawa S, Bouillard J-S, Renger J, Quidant R, Zayats AV. Optically-programmable nonlinear photonic component for dielectric-loaded plasmonic circuitry. Opt Express 2011;19:25222–9.CrossrefGoogle Scholar
Berini P, De Leon I. Surface plasmonpolariton amplifiers and lasers. Nature Photon 2012;6:16–24.Google Scholar
Seidel J, Grafström S, Eng L. Stimulated emission of surface plasmons at the interface between a silver film and an optically pumped dye solution. Phys Rev Lett 2005;94:177401–1–4.CrossrefGoogle Scholar
Krasavin AV, Vo TP, Dickson W, Bolger PM, Zayats AV. All-plasmonic modulation via stimulated emission of copropagating surface plasmon polaritons on a substrate with gain. Nano Lett 2011;11:2231–5.CrossrefGoogle Scholar
About the article
Published Online: 2012-06-08
Published in Print: 2012-07-01