The first demonstration of grating-coupled long range surface plasmon polaritons in cladded free-standing membrane waveguides is presented. Two different waveguide structures are explored: the first is a gold (Au) stripe embedded in a thin Cytop free-standing membrane, the other being the same structure but with a thin palladium (Pd) over-layer. The waveguides are excited with integrated grating couplers designed for a working wavelength of 1550 nm. The waveguides are characterized by applying a cutback technique with the Au waveguide loss measured as 3.4 dB/mm and the Pd/Au waveguide loss as 57 dB/mm. The wavelength dependency of the weakly reflecting optical cavity is also observed with a free spectral range of ~3.6 nm and a finesse of 2.1.
© 2015 Optical Society of America
Surface Plasmon Polaritons (SPPs) are electromagnetic modes that can propagate along the interface between a dielectric medium and a metal. They exist as a coupled excitation comprised of a TM-polarized electromagnetic wave and a charge density wave in the conductor. A plane interface between a metal and dielectric can allow for guiding of these modes although high attenuation is observed. SPP propagation is strongly sensitive to surface conditions which can be exploited to make highly sensitive sensors. However, large attenuation ultimately limits practical application . The attenuation can be reduced by multiple orders of magnitude by implementing a structure where a thin metal film is symmetrically bounded on both sides by semi-infinite uniform dielectrics . This structure supports what is referred to as a Long Range Surface Plasmon Polariton (LRSPP) which can be used to implement devices having a longer optical interaction with the sensing medium. The structure can be further modified by introducing lateral confinement by limiting the width of the metal film (forming a stripe)  thus allowing for the fabrication of integrated optical plasmonic components that can be structured for various applications .
The main constraint for LRSPP-based devices is the requirement to ensure refractive index symmetry. For any functional implementation, the propagating electric field needs to be able to interact with the analyte of interest so full encasement in the dielectric is impractical. It has been shown that LRSPPs can propagate on a metal stripe supported by an ultrathin free-standing dielectric membrane . This structure, referred to as the membrane waveguide, allows for sensing environment to bound the structure thus ensuring approximate refractive index symmetry. By functionalizing the metal surface the waveguide can be adapted for biological or gas sensing applications [6,7]. The membrane waveguide can be further improved by embedding the metal directly into the thin membrane and functionalizing a portion of the overlying cladding. This structure is referred to as the cladded free-standing membrane waveguide  whose operation and experimental characterization forms the objective of this paper.
2. Device structure
2.1 Cladded membrane waveguide
Figure 1 shows a depiction of the cladded membrane waveguide along with cross sectional views and the desired dimensions labelled. The waveguide is formed by embedding a gold (Au) stripe in a thin, free-standing Cytop membrane. Cytop is chemically resistant amorphous fluoropolymer with low refractive index (~1.335) and has been previously demonstrated with other LRSPP devices [8–12]. Input and output light coupling is achieved through the use of grating couplers which allow for broadside (perpendicular) excitation from focused beams or optical fibers. The device in air can operate as a gas sensor by functionalizing a region on top of the waveguide.
For example, a thin patch of palladium (Pd) can be used, which allows this structure to function as a hydrogen sensor, as Pd absorbs hydrogen selectively. However, the device can be modified to sense other gaseous analytes by replacing the Pd patch with another transduction medium. The device was designed for operation at an operating wavelength of 1550 nm. It was fabricated by first spin-coating and curing the lower portion of the Cytop membrane which is then plasma etched down to the desired thickness. The Au waveguide was formed with a bi-layer lift-off lithography process.
Afterwards the upper Cytop layer was spin coated over the Au and cured. It is then etched to the desired thickness and the Pd deposited, again via lift-off lithography. All devices were fabricated on a 2 inch <100> silicon wafer and a through-wafer wet etch was used to release the membranes using silicon dioxide (SiO2) as a mask. The substrate is entirely for structural purposes and does not contribute to the optical operation. Additional fabrication details can be found in .
2.2 Grating couplers
Input and output light coupling is usually achieved through end-fire excitation from optical fibers. This can allow nearly 100% coupling efficiency [14–16], however this requires precisely cleaved or diced end facets which would be extremely difficult to achieve without compromising the structural integrity of the membrane. This structure therefore uses grating couplers integrated directly onto the waveguides as depicted. Broadside coupling allows for simpler alignment with flat optical fibers with the tradeoff being lower coupling efficiency. The gratings are formed via electron beam lithography and deposited before the Cytop top cladding is applied .
2.3 Cell layout
Figure 2(a) shows a photograph of a fabricated wafer. Figure 2(b) shows a microscope image of a single membrane cell (~100☓1500 μm2) with two parallel Au stripes separated by 30 μm. All gratings are identical (such that the input and output positions can be interchanged) and are fabricated in the positions indicated, defining 4 individual devices. One section is a Au-only reference waveguide while the remaining three devices have identical Pd patches along their paths. The devices with the Pd are considered to have three distinct regions: a Pd/Au waveguide and two surrounding Au-only waveguide sections, which will be individually characterized in Section 4.
3. Experimental setup
The setup for optical characterization of a single device is simple although unconventional. A curved aluminum arm was machined and mounted onto a precision six-axis micro-positioner as shown in Fig. 3(a). A laser diode source at ~1550 nm is connected to a polarization maintaining (PM) single mode fiber that is mounted with the polarization axis orientated along the waveguide direction as indicated in Fig. 2(b). The output fiber is a multimode fiber, which is mounted on another curved arm, and is connected to an optical power meter (EXFO PM-1600). To ensure minimal back reflection, the wafer is not placed on an opaque surface; rather it is mounted on a ring which is attached to a cantilever mounted to a large range of motion positioner as shown in Fig. 3(b).
The fibers are roughly aligned over the gratings under microscope before the laser source is enabled and the fiber positions are finely aligned to obtain maximum output signal power. A dark board was placed underneath the sample to prevent back-reflections from the optical table.
4. Optical characterization
4.1 Au waveguides
There are two distinct waveguide structures to be characterized. The first being the Au stripe in Cytop without a Pd patch (which will be referred to as the Au waveguide) and the second being the region where the Pd overlay is present (referred to as the Pd/Au waveguide). Each of these regions supports different LRSPP modes, having significantly different attenuation that needs to be independently quantified. The Au waveguides are characterized via a cutback technique using a cell as shown in Fig. 4(a) where identical gratings are placed at varying separations. Figures 4(b)-4(d) shows the results of measurements taken from three different cutback cells all from the same wafer. A linear fit is applied to the data and the R2 coefficient of determination is computed showing strong agreement.
The input and output coupling is strongly dependent on distance of the input/output fibers from the wafer. For these experiments the fibers were held relatively far away (~30 μm) to ensure consistent, although non optimized coupling. Thus, all of the cutback measurements show large coupling losses (y-intercept), however the slope which represents the mode power attenuation (MPA) can still be independently extracted. From the experimental data the MPA of the Au-only cladded membrane waveguide is 3.4 dB/mm. This measured attenuation is ~3.5☓ larger than that expected from modelling (0.97 dB/mm) .
4.2 Fabricated waveguide dimensions
The waveguide properties are highly sensitive to the physical dimensions of the metal and membrane and thus the observed discrepancies are likely attributed to geometrical differences between the fabricated devices and the modeled ones. Ideally, the thickness of Cytop above the stripe should be equal to that of the Cytop film beneath (dupper = dlower = 70 nm). Fabrication errors ultimately resulted in a slightly asymmetrical waveguide. The actual thickness of each layer was measured away from the membrane via ellipsometry during the fabrication. For the wafer tested in the previous section they were measured as dupper = 50 nm dlower = 81 nm. However, there may still be small variations in the Cytop thickness across the wafer due to uneven coating and etching.
This thickness asymmetry alone does not fully account for the discrepancy. The devices were examined under atomic force microscope (AFM) to better determine the realized dimensions. Figure 5(a) shows the AFM results for a Au waveguide before the top cladding was applied and Fig. 5(b) shows a scan over the Pd patch of a final device.
From Fig. 5(a) the Au stripe has a thickness of 23.5 nm (20 nm expected), a width of 5.3 μm (5 μm expected) and a root mean squared (RMS) surface roughness of Rq = 1.51 nm. Figure 5(b) shows that the surface profile of the Pd patch is rounded over a bump of height ~44 nm. The Pd film was only intended to be about 5 nm thick which implies that the Cytop upper cladding layer must be causing the topography. Topography is not unexpected since the top layer is applied via spin coating over the Au waveguide without planarization.
Due to the large deviation of the realized dimensions from the intended ones, the original design predictions  need to be updated. Figure 6 shows a comparison of the finite element method (FEM) computed mode profile (y-component of the electric field) between an ideally symmetric Au waveguide structure with exact dimensions and the realized Au waveguide with the previously measured stripe and membrane dimensions. The bump over the waveguide was approximated as a trapezoid.
The computed MPA of the realized structure (2.4 dB/mm) is noticeably higher than that of the ideal device (0.97 dB/mm) and reasonably consistent with the measured value of 3.4 dB/mm. The additional discrepancy can likely be attributed to slight variations in the dimensions from device-to-device. In addition, the surface roughness of the Au, although small, may induce some additional scattering loss.
4.3 Pd/Au waveguides
The Pd/Au waveguides can also be characterized with a cutback technique but a simple cutback structure is not available. Instead, each cell was fabricated with different Pd patch lengths, and individually measured with the results combined into a cutback.
This cutback is less straightforward than in the previous case because the loss of a Pd/Au section needs to be de-embedded from the Au waveguide sections at the input and output. The cutback is performed by measuring the difference in insertion loss between a total waveguide section with Pd (ILtotal) and the reference waveguide (ILREF) as depicted in inset to Fig. 7. The insertion loss of the reference waveguide is expressed in dB as:
The insertion loss of the device with Pd (ILtotal), is expressed as:Eq. (1) from Eq. (2) resulting in:Figure 7 gives the cutback plot using the computed values of ILPd for various Pd lengths. The slope is the isolated MPA of the Pd/Au waveguide, which works out to MPAPd = 57 dB/mm while the intercept represents the total coupling losses, 2CPLPd = 0.4875 dB.
As discussed in Section 4.2, the actual topography of the realized waveguide is rounded. The deposited Pd film was intended to be ~5 nm but with the large underlying topography (~44 nm) it is difficult to accurately isolate the Pd film thickness within the AFM scans. An estimate can be made by simulating various Pd thicknesses on the known membrane topography to attempt to match the measured value. A Pd film of ~3.5 nm would correspond with the measured result but this is not very conclusive as the permittivity in the infra-red of sub-20 nm metal films can deviate substantially from bulk values used in the model .
4.4 Coupling loss
From Eq. (3), the y-intercept in Fig. 7 is the loss contribution due to mode mismatch (2CPLPd) occurring at the input and output of the Pd/Au section. The theoretical coupling loss can be calculated using an overlap integral as described in [8,16]. Using the fabricated dimensions, the total coupling loss has been calculated to be 2CPLPd = 0.35 dB (or 0.175 dB per interface), which is slightly less than the observed ~0.4875 dB. The slight discrepancy can be explained by the previously discussed issues with modeling thin Pd layers as well as the aforementioned device-to-device variations. Although it is quantifiable, the coupling loss contributes an insignificant amount to the total loss
5. Gratings characterization
5.1 Coupling efficiency
The input and output grating coupling efficiency cannot be easily or independently characterized since they cannot be isolated for testing. However, if carefully aligned insertion loss measurements are performed on a Au-only waveguide, the combined input and output coupling efficiency can be estimated by subtracting out the known waveguide loss.
The gratings have been previously modeled  yielding an expected input coupling loss of 5.7 dB and an output coupling loss of 3.8 dB for a total coupling loss of 9.5 dB. Several insertion loss measurements of Au-only waveguides of known length were taken, and the total coupling loss was extracted, resulting in an average measured total coupling loss of 18.4 dB, which is significantly higher than predicted.
However, the design modeling  was performed in 2-D (infinite x-direction) and over-estimates the coupling efficiency. The typical 1/e mode field diameter for a single-mode (1550 nm) PM fiber is about 10 μm, which is significantly wider than the width of the waveguide (5 μm). Therefore without beam focusing, a fair amount of beam power will not be incident on the grating resulting in additional uncoupled power.
Figure 8 shows an AFM image over a section of a grating along with a profile plot along the indicated line. It is clear that the infilling of the space between each Au grating bump is not flat as sketched in Fig. 1(b), but rather forms a rounded valley. For this wafer, the Au bumps should protrude by ~175 nm from the Cytop, whereas the measured peak-to-valley height is 161 nm. These factors explain the deviations between the measured and modelled results.
The AFM scan also clearly shows significant residue along the edges of the protruding Au bumps. This residue did not appear prior to the application of the upper cladding and it is presumed to be residual Cytop. This residue likely contributes to additional coupling loss due to scattering. Despite the increased coupling loss over the predicted, the power is still well within detectable levels, even when using low power input laser sources.
5.2 Wavelength response
The grating wavelength response can be characterized by utilizing a tunable wavelength laser source and a broadband optical power detector. Figure 9 shows the insertion loss of a Au waveguide as the wavelength is varied between 1540 to 1560 nm. An obvious wavelength dependent oscillation is observed. This can be explained by the Fabry-Pérot resonance formed in the optical cavity due to back reflections from the gratings as sketched in inset to Fig. 9. The free spectral range (FSR) of an optical cavity is calculated as:Fig. 9. By considering the peaks at 1546.4 and 1550 nm, a FSR of ~3.6 nm is calculated. The average peak separation between all observable peaks is ~3.8 nm.
The finesse of the cavity is described by the equation
6. Concluding remarks
LRSPP cladded membrane waveguides with grating couplers have been demonstrated for the first time. The waveguides have been characterized with a cutback technique, with the Au waveguides having a measured MPA of 3.4 dB/mm, while the Pd/Au waveguides having a measured MPA of 57 dB/mm. The total coupling losses from the gratings was measured to be approximately 18.4 dB. The performance is quite sensitive to the structural dimensions as fabrication errors resulted in deviations from the simulated behavior. The grating couplers induce a mild Fabry-Pérot resonance.
References and links
1. W. L. Barnes, “Surface plasmon–polariton length scales: a route to sub-wavelength optics,” J. Opt. A, Pure Appl. Opt. 8(4), S87–S93 (2006). [CrossRef]
2. D. Sarid, “Long-range surface-plasma waves on very thin metal films,” Phys. Rev. Lett. 47(26), 1927–1930 (1981). [CrossRef]
3. P. Berini, “Plasmon-polariton waves guided by thin lossy metal films of finite width: bound modes of symmetric structures,” Phys. Rev. B 61(15), 10484–10503 (2000). [CrossRef]
4. P. Berini, “Long-range surface plasmon polaritons,” Adv. Opt. Photon. 1, 484–588 (2009).
6. P. Berini, R. Charbonneau, and N. Lahoud, “Long-range surface plasmons along membrane-supported metal stripes,” IEEE J. Sel. Top. Quantum Electron. 14(6), 1479–1495 (2008). [CrossRef]
7. N. R. Fong, P. Berini, and R. N. Tait, “Modeling and design of hydrogen gas sensors based on a membrane-supported surface plasmon waveguide,” Sensor Actuat. Biol. Chem. 161, 285–296 (2012).
8. N. R. Fong, P. Berini, and R. N. Tait, “Modeling of long range surface plasmon polariton cladded membrane waveguides with integrated grating couplers as hydrogen sensors,” J. Appl. Phys. 117(16), 163108 (2015). [CrossRef]
9. H. Fan, R. Buckley, and P. Berini, “Passive long-range surface plasmon-polariton devices in Cytop,” Appl. Opt. 51(10), 1459–1467 (2012). [PubMed]
10. B. Banan, M. S. Hai, E. Lisicka-Skrzek, P. Berini, and O. Liboiron-Ladouceur, “Multi-channel transmission through a gold strip plasmonic waveguide embedded in Cytop,” IEEE Photon. J. 5(3), 2201811 (2013). [CrossRef]
13. N. R. Fong, P. Berini, and R. N. Tait, “Fabrication of long-range surface plasmon hydrogen sensors on Cytop membranes integrating grating couplers,” J. Vac. Sci. Technol. B 33(2), 021201 (2015). [CrossRef]
14. R. Charbonneau, P. Berini, E. Berolo, and E. Lisicka-Shrzek, “Experimental observation of plasmon polariton waves supported by a thin metal film of finite width,” Opt. Lett. 25(11), 844–846 (2000). [CrossRef] [PubMed]
15. T. Nikolajsen, K. Leosson, I. Salakhutdinov, and S. I. Bozhevolnyi, “Polymer-based surface-plasmon-polariton stripe waveguides at telecommunication wavelengths,” Appl. Phys. Lett. 82(5), 668–670 (2003). [CrossRef]
16. P. Berini, R. Charbonneau, N. Lahoud, and G. Mattiussi, “Characterization of long-range surface plasmon-polariton waveguides,” J. Appl. Phys. 98(4), 043109 (2005). [CrossRef]