Surface Plasmon? Propagating surface waves can exist between a metal and a dielectric. Although you won’t need to derive this, I will put the full derivation on Keats for those interested - here, I will present the dispersion relationship as we’re interested in the technology. • First of all, what is the physical origin of these surface waves? Surface Plasmon? Well, we know that metals have free charges. Any wave propagating along the surface of a metal must therefore exhibit strong interaction with these charges. Plasmons exist as longitudinal (along the surface) oscillations in free charge density. This is best explained in a figure. Surface Plasmons Surface plasmons We normally consider the two polarisation basis for light at an interface: the transverse magnetic modes (TM-modes, or p-modes) and the transverse electric modes (TEmodes, or s-modes). 𝜀1 𝜀2 y x z For TM modes, only the field components Ex, Ez, and Hy are non zero. Whereas for TE modes, only Hx, Hz and Ey are non zero Surface plasmons 𝜀1 Let’s remind ourselves of the different between TM (p) and TE (s) polarisation. ẑ k 𝜀2 y x k z Surface plasmons Narrator: TE polarised modes are not supported. The dispersion relationship for a surface wave for TM polarised surface wave is given as : 𝑘𝑠𝑝𝑝 = 𝑘0 𝜀1 𝜀2 y 𝜀𝑚 (𝜔)𝜀𝑑 𝜀𝑚 (𝜔) + 𝜀𝑑 This is in the ‘classic’ form for a dispersion relationship 𝑘𝑠𝑝𝑝 = 𝑠𝑝𝑝 𝑛𝑒𝑓𝑓 = 𝜀𝑚 (𝜔)𝜀𝑑 𝜀𝑚 (𝜔) + 𝜀𝑑 𝑠𝑝𝑝 𝑘0 𝑛𝑒𝑓𝑓 x z For derivation, see Keats Surface plasmons So what does it look like? 𝝎𝒔𝒑 𝑘𝑠𝑝𝑝 = 𝑘0 𝜀𝑚 (𝜔)𝜀𝑑 𝜀𝑚 (𝜔) + 𝜀𝑑 Surface plasmons The SPP dispersion is ALWAYS on the “right-hand side” of the light line (higher k): for all values of 𝜔 < 𝜔𝑠𝑝𝑝 , 𝑘𝑠𝑝𝑝 > 𝑐𝑘 This means that (remember 𝒑 = ℏ𝒌) there is a momentum mismatch between incident light in the dielectric and the SPP An SPP CANNOT be excited by direct illumination (without nanostructuring, as we shall see) Surface plasmons Visualising the momentum mismatch in the isofrequency surface in the (𝑘𝑥 , 𝑘𝑦 ) plane Surface plasmons We can have the complete dispersion by adding the volume plasmons we obtained before Within the model of an ideal free electron plasma, the frequency range between surface plasmon frequency and plasma frequency represents a gap where no propagating electromagnetic mode exists. We shall see how to find a simple expression for ωspp in Problem sheet 2 𝜔𝑝 𝜔𝑠𝑝𝑝 Surface plasmons In real metals, damping modifies the SPP dispersion curve and does not allow for unlimited growth of the wavenumber at 𝜔 → 𝜔𝑠𝑝 𝜔𝑝 𝜔𝑠𝑝𝑝 However, 𝜔𝑠𝑝𝑝 remains the upper frequency limit for surface plasmons S. A. Maier. Plasmonics: Fundamentals and Applications The transition between SPP and VP is a region known as the ‘backbending’ region of the dispersion. Surface plasmons Let’s summarise SPPs? Surface waves, confined to the interface between a metal and a dielectric material Commonly called Surface Plasmon Polaritons (SPP) A plasmon is a quasi-particle that represents the quantum of the collective excitation of free electrons in solids. A polariton is a quasi-particle resulting from strong coupling of electromagnetic waves with an electric or magnetic dipole‐carrying excitation: phonon‐polaritons: coupling of photons with optic phonons exciton‐polaritons: coupling of photons with an exciton Surface plasmon‐polaritons: coupling of surface plasmons with light Surface plasmons We can also define an SPP wavelength using the standard equation: 𝑘= 2𝜋 𝜆 Therefore 𝜆𝑆𝑃𝑃 = 2𝜋 The EM field is confined to the interface, and just like a surface wave on a dielectric, it decays exponentially away from the interface. 𝜀𝑚 𝜀𝑑 𝑘0 𝜀𝑚 +𝜀𝑑 Dielectric Interface Metal Surface plasmons Excitations of the conduction electrons of real metals however suffer from damping. Therefore, 𝜀𝑚 (𝜔) is complex, and with it also the SPP propagation constant 𝑘𝑠𝑝𝑝 The intensity decays with the square of the electric field, so at a distance x, the intensity has decreased by a factor of: 𝑒 −2𝐼𝑚 𝑘𝑠𝑝𝑝 𝑥 The factor of 2 is coming from squaring the field The distance over which the SPP intensity has dropped to 𝑒 −1 of its original value is called the propagation length: L = 2 𝐼𝑚 𝑘𝑠𝑝𝑝 −1 Surface plasmons • Of course, in the same way as we did for TIR in a prism in Problem sheet 1, we can determine the penetration depth in either the metal, or the dielectric medium. • The wavevector in the z-direction is given by: 𝜀1 𝜀2 y 𝑘𝑧 = 𝑖 2 𝑘𝑠𝑝𝑝 x 2 − 𝑘0 𝜀𝑑(𝑚) 𝜀𝑚 𝜀𝑑 With 𝑘𝑆𝑃𝑃 = 𝑘0 𝜀𝑚 +𝜀𝑑 • We use either the dielectric or metal permittivity as needed. We get the penetration depth as δd(m)= 2 𝐼𝑚 𝑘𝑧 z −1 The wavevector in z is completely imaginary (it lies to the right of the light line). Surface plasmons For example, for 𝜆0 = 600 nm: For gold-air, we have 𝛿𝑔𝑜𝑙𝑑 = 31 nm and 𝛿𝑎𝑖𝑟 = 280 nm For silver-air, we have 𝛿𝑠𝑖𝑙𝑣𝑒𝑟 = 24 nm and 𝛿𝑎𝑖𝑟 = 390 nm Surface plasmons Recall the SPP dispersion: for 𝜔 < 𝜔𝑠𝑝 , 𝑘𝑠𝑝𝑝 > 𝑐𝑘 𝝎𝒔𝒑 At higher frequencies (lower wavelength), the gap between the light line and SPP dispersion increases. The magnitude of kz is proportional to this gap, as these are what we subtracted on the previous slide! So higher frequencies, larger kz, smaller penetration depth! Surface plasmons Recall the SPP dispersion: for 𝜔 < 𝜔𝑠𝑝 , 𝑘𝑠𝑝𝑝 > 𝑐𝑘 𝝎𝒔𝒑 We cannot excite these modes on the surface of a metal directly without some additional way to increase the momentum of the photons. So how do we excite those modes? It’s also key to remember that the real part of the SPP wavevector (or propagation constant) is along the interface – so for momentum matching, we need a large ‘inplane’ wave-vector. Surface plasmons Optically we need to access large in-plane wavevectors for momentum matching: Recall 𝑘 = 2𝜋 𝜆 = 2𝜋 𝜀𝑟 𝜆0 But, if we increase the dielectric constant next to the metal surface, we also move the SPP dispersion! Surface plasmons Optically we need to access larger wavevectors Recall 𝑘 = 2𝜋 𝜆 = 2𝜋 𝜀𝑟 𝜆0 But, we can still use this if we apply ourselves and think it through a little. We know that if we had a photon in glass, it has enough momentum to couple to an SPP at an air interface. Key point to remember: We need this momentum in the direction of the SPP (along the interface). So the lines on this plot would represent an angle of incidence of 90 degrees! Surface plasmons 𝝎𝒔𝒑 ( 𝜀𝑟 = 1) 𝝎𝒔𝒑 ( 𝜀𝑟 = 1.5) Optically we need to access larger wavevectors Recall 𝑘 = 2𝜋 𝜆 = Air 2𝜋 𝜀𝑟 𝜆0 Metal This is the situation before us on the right. We have put a metal layer on a glass substrate. We now have two SPP modes, one at the metal/air side, and one on the metals. Glass Surface plasmons 𝝎𝒔𝒑 ( 𝜀𝑟 = 1) 𝝎𝒔𝒑 ( 𝜀𝑟 = 1.5) Optically we need to access larger wavevectors Recall 𝑘 = 2𝜋 𝜆 = Air 2𝜋 𝜀𝑟 𝜆0 Metal So, if we shine light from the glass side, we should be able to reach the necessary momentum for the SPP on the air side. However, if we use a planar glass substrate, we still have a problem. Glass Surface plasmons Air Recall: we need the momentum in the plane of the surface. Metal But we know refraction conserves in-plane momentum. The x component of k is the same in the air beneath the glass, as it is in the glass! nglass koSinθglass=nair koSinθ Glass Surface plasmons Optically we need to access larger wavevectors Recall 𝑘 = 2𝜋 𝜆 = 2𝜋 𝜀𝑟 𝜆0 So, if we use a planar substrate, the refraction into the glass from the air can only maintain the in-plane momentum from the air (blue dashed line). Back to square one! Surface plasmons However, if we use a metal layer on a prism (as we showed for TIR) then we solve the refraction problem at the air/glass interface. This idea gave rise to the Kretshmann geometry shown below! A.V. Zayats et al., Physics Reports 408 (2005) 131–314 Surface plasmons Now, thanks to the prism, our incident light has access to the region highlighted in peach below. Here we want to match the in- plane component of our incomping light, 𝒌: 𝑘∥ = 2𝜋 𝜀𝑟 sin 𝜃 𝜆0 We need to tune 𝜃 at a particularly frequency or wavelength, to overlap with 𝑘𝑠𝑝𝑝 𝑘∥ 𝑘∥ Surface plasmons Matching the wavevectors for a given wavelength, we must satisfy the following condition: 𝑘𝑔𝑙𝑎𝑠𝑠 sin 𝜃= 𝑘𝑠𝑝𝑝 𝑘0 𝜀𝑔𝑙𝑎𝑠𝑠 sin 𝜃 = 𝑘𝑠𝑝𝑝 So 2𝜋 𝜀𝑔𝑙𝑎𝑠𝑠 𝜆0 sin 𝜃 = 𝑘𝑠𝑝𝑝 Solving the above equation we can find the angle of incidence required in the glass substrate! Surface plasmons So let us look at an example: silver-air interface 𝑘𝑠𝑝𝑝 𝜆0 So 𝜃𝑆𝑃𝑃 = arcsin 2𝜋 𝜀𝑟3 𝜀𝑚 𝜀𝑑 with 𝑘𝑆𝑃𝑃 = 𝑘0 𝜀𝑚 +𝜀𝑑 Here we only need to consider the real part of 𝑘𝑠𝑝𝑝 𝜀𝑟2 𝜀𝑟1 = 𝜀′ + 𝑖𝜀′′ 𝜀𝑟3 Surface plasmons We consider the following: 𝜆0 = 980 𝑛𝑚 𝜀𝑑1 = 1 (air) 𝜀𝑚 = −48.474 + i 0.55699 𝜀𝑑2 = 1.52 (standard for glass) Notice how much of the light goes to the SPP! Heinz Raether, Surface plasmons on smooth and rough surfaces and on grating, Springer This gives us: 𝑘𝑠𝑝𝑝 = 6.4786 × 106 + 7.8392 × 102 𝑖 Which corresponds to an internal incident angle of 𝜃𝑆𝑃 = 42.3494° Surface plasmons Keeping exactly the same parameters and changing only 𝜀𝑟1 , we obtain: 𝜀𝑑1 = 1, 𝜃𝑆𝑃 = 42.3494° 𝜀𝑑1 = 1.1, 𝜃𝑆𝑃 = 45.0137° Notice that this is a ‘resonant’ condition – and this condition (via Kspp) varies as the dielectric constant of the top interface. Contrast this with a dielectric prism evanescent wave, which is not resonant (exists at all wavelengths at angles above TIR) 𝜀𝑑1 = 1.01, 𝜃𝑆𝑃 = 42.6160° (this is n=1.005!) This is the principle behind SPR (Surface Plasmon Resonance) sensing – which we shall meet later! Surface plasmons - Nanostructuring Rather than using a prism, which is cumbersome, we can use nanostructuring to couple light directly to SPPs. To see how we can do this, let’s consider a diffraction grating and find the grating equation you learned in first year! Just like refraction, this is derived by considering the in- plane components of the wavevector, and the wavevector of the grating! Surface plasmons - Nanostructuring We can describe the wavevector of a grating thus: 2𝜋 𝑘𝑔 = 𝑑𝑔𝑟𝑎𝑡𝑖𝑛𝑔 Where dgrating is the line spacing on the grating. So, considering the conservation of in-plane momentum: 𝑘𝑆𝑖𝑛𝜃𝑖𝑛𝑐𝑖𝑑𝑒𝑛𝑡 ± 𝑚𝑘𝑔 = 𝑘𝑆𝑖𝑛𝜃𝑑𝑖𝑓𝑓𝑟𝑎𝑐𝑡𝑒𝑑 Where m is an integer describing diffraction order (m=0, 1,2 𝑒𝑡𝑐. ) 𝑛𝑘0 𝑆𝑖𝑛𝜃𝑖𝑛𝑐𝑖𝑑𝑒𝑛𝑡 ± 𝑚𝑘𝑔 = 𝑛𝑘0 𝑆𝑖𝑛𝜃𝑑𝑖𝑓𝑓𝑟𝑎𝑐𝑡𝑒𝑑 (𝑛𝑆𝑖𝑛𝜃𝑖𝑛𝑐𝑖𝑑𝑒𝑛𝑡 + n𝑆𝑖𝑛𝜃𝑑𝑖𝑓𝑓𝑟𝑎𝑐𝑡𝑒𝑑 )𝑑𝑔𝑟𝑎𝑡𝑖𝑛𝑔 = ±𝑚𝜆0 At normal incidence and with n=1 (air) we have the familiar equation: 𝑑𝑔𝑟𝑎𝑡𝑖𝑛𝑔 𝑆𝑖𝑛𝜃𝑑𝑖𝑓𝑓𝑟𝑎𝑐𝑡𝑒𝑑 = ±𝑚𝜆0 Surface plasmons - Nanostructuring So we can use a grating to couple to SPPs, i.e. we need to periodically nanostructure the metal surface. 2𝜋 𝑑 𝑛1 𝑛1 𝑘0 sin 𝜃 ± 𝑚 𝑧 = ±𝑘𝑆𝑃𝑃 𝜃 𝑥 𝑘𝑖𝑛 𝑥 𝑘𝑔𝑟𝑎𝑡𝑖𝑛𝑔 𝑘𝑆𝑃𝑃 kgrating adds (and substracts) from kspp This is best illustrated by considering a dispersion plot. SPPs on periodically nanostructured surfaces Original dispersion 𝜔 Bragg scattering of the dispersion 𝜔 We’ve drawn the negative side too, as an SPP can go either direction along the interface k -2p/d -p/d 0 p/d • The grating shifts the SPP dispersion by multiples of kgrating (dashed lines) • Remembering that the wavevector is in-plane for SPPs, so normal incidence corresponds to k=zero. • We can clearly see in the plot (above right) that the curves cross zero. This means that we can excite SPPs at normal incidence, with the grating providing all of the momentum in the plane of the surface. 2p/d k Periodic nanostructures in nature Even periodic dielectric structures in nature show amazing visual effects via diffraction. So do periodic nanostructures on metallic films! We shall see some methods for making these nanostructures later! Periodic nanostructures in nature We shall see some methods for making these nanostructures later! So do periodic nanostructures on metallic films! 10 µm SPPs – Gold Air Interface An example, a grating with a 600 nm period on gold. Au SiO2 Here, the bright colours represent more transmission, the dark colours less. We can also couple to the modes on the glass/Au side – impossible using a prism! Here are some experimental results for a gold film on glass. The transmission is measured through the nanostructured area (right figure) The dark bands correspond to ‘missing light’ which are SPPs! SPPs Summary Light can couple to longitudinal resonant electron oscillations at the interface between a metal and a dielectric, these are termed Surface Plasmon Polaritons. As they are confined to the interface, the lie to the right of the light line in the dispersion plots. This also means that we need to either use a prism or a periodic nanostructure to provide incident light with the necessary momentum to couple to SPPs. For a given metal (and therefore permittivity) higher frequencies have larger wavevectors, this means that the SPP is more confined to the interface. The penetration depth into the dielectric can easily be λ0/4 and higher. However, this will also come at the expense of more losses in the propagation direction (travels shorter distances)! • The SPP will also have higher wavevectors if the dielectric has a higher dielectric constant (slide 29) so light is confined even more tightly to the interface. SPPs Summary SPPs can provide nanoscale electromagnetic fields on metal/dielectric interfaces. As they allow light to be squeezed, they have potential to become optical chip interconnects with high integration density. The challenge is minimising the attenuation of the signal. (prop length) SPPs are resonant, and therefore very sensitive to the dielectric constant close to the surface of the metal film – any variation close to the surface will change the SPP coupling condition (angle required at some wavelength, or the wavelength required at some angle). This means the surface is an incredibly sensitive sensor. Any process which modifies the refractive index close to the surface can be detected. • The incident light can couple very efficiently to the SPP, and as it is squeezed (high wavevector) the local EM field at the surface can be many times higher than the incident field. SPPs Summary This can be incredibly useful for any process that depends on intensity. Fluorescence Photocatalysis Raman Scattering – Inelastic optical scattering which provides a spectra fingerprint of molecules. Optical forces e.g. optical tweezers for manipulating nano-objects, (force proportional to intensity gradients) Nonlinear optics – materials where the refractive index has a contribution proportional to intensity (can make active optical active devices).