1. Introduction

Industrial applications of surface plasmon polaritons (SPPs) are expected to begin based on the recent interesting scientific results [1–10]. SPPs can be used in sensing, the enhancement of light emission or detection, lithography, and optical interconnection applications.

One potential application to optical interconnections is the interconnection between the main body and the display section using an optical waveguide. Recently, we demonstrated the possibility of using optical interconnects in hand-held devices such as mobile phones using a long-range surface plasmon polariton (LRSPP) waveguide [11]. We expect this type of waveguides [1–10] to be used in optical interconnection in hand-held devices if a flexible LRSPP waveguide is fabricated with a low optical loss. The optical loss includes propagation and coupling losses as well as bending loss. Recently, propagation losses less than 2.0 dB/cm were reported with LRSPP waveguides [7, 12]. We also recently reported a propagation loss of sub-dB/cm, which is comparable with that of dielectric waveguides [13].

LRSPP waveguides embedded within a polymeric membrane have been reported [14, 15], showing the possibility of the flexible LRSPP waveguide, and there has been a theoretical study on vertical bending losses in LRSPP waveguides [3], but no experimental report for the vertical bending loss of the flexible LRSPP waveguide to our knowledge. The vertical bending loss in a conventional LRSPP waveguide with a single layer of cladding is known to be too severe to be used even with a slow bending of several tens of millimeters in radius [3], even though interconnections in hand-held devices require a bending to a radius of less than 3 mm.

In these regards, we fabricate and measure the vertical bending loss of flexible LRSPP waveguides, for the first time, and reduce optical bending loss by using multilayer polymer claddings. The flexible LRSPP waveguide is 68 mm long and composed of an 8 nm-thick Ag strip embedded in a free-standing multilayer polymer film. The polymer film is composed of a 10 μm-thick inner cladding with a refractive index of 1.524, and a pair of 20 μm thick outer claddings having a refractive index of 1.514, resulting in a total thickness of 50 μm.

2. Experiment and results

Figure 1(a) shows an LRSPP waveguide with a single layer of cladding. An Ag strip with a thickness of 8 nm is embedded as a core in a 60 μm-thick polymer cladding. The thickness of the silver strip was calibrated with an atomic force microscope (AFM). Figure 1(b) shows an AFM scan of the silver strip which has a width of 4 μm and a thickness of 8 nm. The surface roughness of the silver strip was measured as less than 1 nm. The polymer cladding named as FOWG is an ultra-violet (UV) cured film based on an ethylene oxide acrylate blend including fluorinated styrene, and is supplied by ChemOptics, Inc. (www.chemoptics.co.kr). The refractive index of FOWG can be controlled with the mixing ratio of the components (between 1.506 and 1.547), and a material with a refractive index of 1.514 is used as the single layer cladding. The detailed fabrication process of the LRSPP waveguide on a Si substrate is similar to that in our previous reports [7–10].

A flexible LRSPP waveguide is obtained by detaching the polymeric film from the substrate after fabrication is complete. We provide a 30-nm thick gold layer on the central area of the substrate at the first step of the fabrication for an effective and selective detachment. We use the poor adhesion of the gold layer to the silicon substrate. The central section of the polymer film fabricated on the gold layer is partially detachable from the substrate after the input and output fiber is pigtailed at the ends of the waveguide.

 

Fig. 1. (a) Schematic structure of an LRSPP waveguide with a single layer of cladding, (b) AFM image and thickness profile of a 4 μm-wide silver strip, (c) a picture of a fiber-pigtailed flexible LRSPP waveguide fabricated in this experiment, (d) a picture of a flexible LRSPP waveguide inserted within a slot with bending radius of 5 mm to measure the insertion loss induced by the bending. Bending loss can be measured for a 180° turn with a bending radius between 1 mm and 12 mm, and for smaller angles with bending radii of 20, 25, and 30 mm with two sets of mechanical components.

Download Full Size | PDF

Figure 1(c) shows a flexible LRSPP waveguide fabricated in this experiment with the cross-sectional structure shown in Fig. 1(a). The length of the LRSPP waveguide was 68 mm, and it was detached and pigtailed to a pair of single-mode fibers (SMF) as in Fig. 1(c). The detached film was flexible enough to be mechanically bent down to a radius of 0.5 mm, though degradation in optical insertion loss was induced by the bending. The optical bending loss of the flexible LRSPP waveguide was measured by inserting the fiber-pigtailed film into a slot with bending radiuses from 1 to 30 mm (Fig. 1(d)). Bending loss was measured for a 180° turn with bending radii from 1 mm to 12 mm, and for smaller angles with bending radii of 20, 25, and 30 mm by inserting the flexible LRSPP waveguide into the slot of two sets of mechanical components shown in Fig. 1(d).

Figure 2 shows the measured bending loss for the fiber-pigtailed flexible LRSPP waveguide with a single layer of cladding, compared to the bending loss before the detachment of the film from the substrate. The insertion loss of the SMF-pigtailed 68 mm-long LRSPP waveguide with a 4 μm-wide Ag strip was measured to be 10 dB before detachment, and is marked as ∞ (straight) in Fig. 2. The insertion loss of the waveguide detached from the substrate increased rapidly over 25 dB (Fig. 2) when the waveguide was bent to 180° with a bending radius less than 25 mm. The insertion loss was not stable even at a fixed bending radius, but was sensitive to the shape of the film which changed slightly. The bending loss in Fig. 2 is the minimal measured loss. Figure 2 shows that a flexible LRSPP waveguide with a single layer of cladding is too sensitive in vertical bending to be used, even at a slight bending radius of 30 mm.

 

Fig. 2. Measured bending loss of the flexible LRSPP waveguide with a single layer of cladding compared to the insertion loss of the waveguide before detachment from the substrate, marked as ∞ (straight).

Download Full Size | PDF

 

Fig. 3. (a) Schematic structure of a flexible LRSPP waveguide built to overcome optical bending loss with 8 nm-thick Ag strip embedded in a high-index inner cladding within low-index outer claddings. (b) Simulated bending loss of a dielectric slab which is composed of a 10 μm-thick inner slab with a refractive index of 1.524 without a metal strip and outer claddings with a refractive index of 1.514.

Download Full Size | PDF

The severe bending loss of the LRSPP waveguide with a single layer of cladding was predicted by simulation [3], and is a big problem that must be overcome for useful application of LRSPP waveguides.

For this reason, we fabricated an LRSPP waveguide with multilayer cladding as in Fig. 3 (a). LRSPP waveguides with hybrid cladding structures have been studied with interesting results for the propagation loss, lateral bending loss and fiber-coupling loss [1, 6, 12, 15–17]. Here, we fabricate a flexible hybrid LRSPP waveguide targeting the reduction of the vertical bending loss, based on the previous results. An 8 nm-thick Ag strip is embedded in a freestanding multilayer FOWG film which is composed of a 10 μm-thick inner cladding having a refractive index of 1.524 and a pair of 20 μm thick outer claddings having a refractive index of 1.514. This results in a total thickness of 50 μm (Fig. 3(a)). We designed the dielectric slab structure to enhance the vertical confinement of the guided LRSPP mode to overcome radiation loss due to a vertical bending, based on a simulated bending loss of a dielectric slab without a metal strip as in Fig. 3(b). The simulated bending loss of the dielectric slab without a metal strip was dramatically reduced when the bending radius was larger than 2 mm in Fig. 3(b). Beam propagation method (BPM) simulation tool BeamPROP of Rsoft was used in calculating the vertical bending loss of the slab waveguide in Fig. 3(b).

Figure 4 shows the measured bending loss for the fiber-pigtailed flexible LRSPP waveguide with multilayered cladding compared to the insertion loss measured before the detachment of the film from the substrate. We compare the bending loss with two different output coupling conditions in Fig. 4. An SMF is pigtailed to the output in Fig. 4(a), while a multimode fiber (MMF) with a core diameter of 50 um is pigtailed to the output in Fig. 4(b). A polarization-maintaining SMF (PMF) was pigtailed to the input in both cases.

 

Fig. 4. Measured bending loss of the flexible LRSPP waveguide with multilayer claddings (a) pigtailed from PMF to SMF, and (b) pigtailed from PMF to MMF.

Download Full Size | PDF

Before the detachment from the substrate, we measured the insertion loss of the 68 mm long LRSPP waveguide with a 3.5 μm wide Ag strip within the multilayer claddings to be 13 dB when it was pigtailed from PMF to SMF (Fig. 4(a)), and 11 dB when it was pigtailed from PMF to MMF (Fig. 4(b)). The insertion loss of the waveguide detached from the substrate didn’t change significantly when the waveguide was bent to a radius of 2 mm, but began to increase when it was bent down to a radius of less than 1.5 mm. The results are comparable to the dielectric slab in Fig. 3(b), even though it is not a slab but a 3.5 μm wide Ag strip waveguide within the dielectric slab in Fig. 4.

 

Fig. 5. Pictures of the experimental setup used to measure the output beam profile of the flexible LRSPP waveguide bent at various radii in (a) through (d), and twisted to 90° in (e). The beam profiles measured with the setup in (f) to (l), matched the pictures on the left. The right half of the beam profile simulated for a straight waveguide is in (m) for comparison.

Download Full Size | PDF

These results show that the vertical bending loss of the flexible LRSPP waveguide is dramatically reduced using the multilayer-cladding structure which allowed the device to be bent down to a radius of 2 mm without showing a noticeable bending loss. The bending loss in Fig. 4(b) is less than 0.3 dB for the bending radius of 2 mm.

The insertion loss shown in Fig. 4(a) was not maintained with a fixed bending radius, but changed by ±0.7 dB as the shape of the film altered slightly. Insertion loss was stably maintained when the MMF is used for the output (Fig. 4(b)). These results show that the origin of the slight change was the condition of the output coupling to the fiber. The origin of this change is not clear but it is regarded to come from the interference between the guided LRSPP mode and cladding modes within the dielectric slab.

It is very interesting that the characteristics of the bending loss for the hybrid LRSPP waveguide in Fig. 4(b) is almost in accord with the dielectric slab waveguide in Fig. 3(b). Before the experiments, we were not sure whether the hybrid LRSPP mode could be maintained against the tight bending and whether the hybrid LRSPP mode to evolve toward a pure dielectric slab mode with the tight bending. The experimental results of low bending losses, however, lead us to believe that the hybrid LRSPP mode is maintained through the tight bending without evolution toward the pure dielectric slab mode. There should be a critical bending loss if the hybrid LRSPP mode evolves toward the pure dielectric slab mode because the slab mode is not laterally confined. And the guided mode is regarded to be maintained without a periodic exchange of energy between the hybrid LRSPP mode and the dielectric slab mode, considering the non-oscillatory increase of the bending loss as the radius is reduced. A rigorous theoretical analysis is necessary to fully understand the behavior of the hybrid LRSPP mode through the bent flexible waveguide, and we are investigating on it as another research.

Observation of the beam profile though the bent flexible waveguide can be an important clue to understand the characteristics of the hybrid LRSPP mode. Figures 5(f)-5(l) show the variation in the output beam profiles of a flexible LRSPP waveguide with multilayer cladding when vertically bent as shown in the pictures in Figs. 5(a)-5(d). The measured beam profile in Fig. 5(f) is comparable to the beam profile simulated, with FemSIM of Rsoft in Fig. 5(m), for the LRSPP waveguide with the same multilayered structure. The pictures in Figs. 5(f)-5(i) show that the beam profiles were maintained as the film macroscopically bent from a straight configuration to a radius of about 5 mm, but Fig. 5(k) shows that the beam profile acquired a lateral tail when the film changed slightly in shape without a change in macroscopic appearance. This difference in the beam profile between Figs. 6(k) and 6(f) explains the small changes that occurred in the insertion loss of the flexible LRSPP waveguide coupled to the SMF output (Fig. 4(a)) in comparison with the stable insertion loss of the flexible LRSPP waveguide coupled to the MMF output (Fig. 4(b)). The modal size of the beam profile in Fig. 5(f) is 23 μm in the lateral direction and 5 μm in the vertical direction.

We also measured the variation of the beam profile when the flexible LRSPP waveguide was twisted by 90° as in Fig. 5(e). Interestingly, the beam profile was maintained without significant degradation (Fig. 5(j)), even though there was the same minor change in the beam profile that occurred with a slight change in the shape of the film (Fig. 5(l)). The measured additional insertion loss by 90° twist was 2 dB including the delicate changes for the SMF output but it was less than 0.3 dB for the MMF output. These results lead us to believe that the polarization of LRSPP mode is maintained when the plane of polarization changes as the film is slowly twisted to 90°. These results show that the flexible LRSPP waveguide shows promise for use in future applications requiring low bending and twisting loss.

3. Conclusion

We fabricated flexible long-range surface plasmon-polariton (LRSPP) waveguides and reduced vertical bending loss by using multilayer polymer claddings. The flexible LSRSPP waveguide fabricated in this experiment was 68 mm long, and was composed of an 8 nm-thick Ag strip embedded in a free-standing multilayer polymer film. The polymer film was composed of a 10 μm-thick inner cladding (refractive index = 1.524), and a pair of 20 μm-thick outer claddings (refractive index = 1.514), resulting in a total thickness of 50 μm. We showed that the LRSPP waveguide can be bent down to a radius of 2 mm, and can be twisted to 90° without a big increase in insertion loss at a wavelength of 1310 nm. These results show that the flexible LRSPP waveguide shows promise for use in future applications requiring low bending and twisting loss.