1. Introduction

The properties of microstructured optical fibers (MOFs) can be adjusted by filling or coating of the holes. For example, by injecting high index fluid in the holes, index guiding MOFs can be turned into photonic bandgap guiding fibers, whose dispersion and transmission bands can in turn be tuned over a large range [1–3]. Various tunable all-in-fiber devices using the filling or coating of holes have been demonstrated or suggested, some of which require only selected holes to be filled or coated [4–8].

While in principle these designs promise a new range of applications in sensing, polarization control or even routing, filling only selected holes of a MOF has proved to be experimentally difficult. A few techniques have been suggested in the literature: some rely on the differential surface tension or capillary forces in holes of different size, and therefore can only be used when the diameters of filled and unfilled holes are different [9–13]. Other techniques, such as direct injection using micropipettes [14] or end-face photolithography [15] have been suggested and demonstrated in the context of planar photonic crystals and conventional optical fibers respectively. However, it is unclear how far they can be applied for MOF holes, which by nature have extreme aspect ratios and cross sections of the order of the micrometer.

The combination of plasmonics and photonics is an emerging field that would benefit from improvements in coating techniques. This area is attracting growing interest in the MOF community. Kuhlmey et al [6] have provided a theoretical formalism for metal-coated MOFs and showed that they exhibit plasmonic resonances usable for filtering, sensing or dispersion management. Hassani and Skorobogatiy [7,8] analysed in detail designs for a MOF with metallic coatings for a biosensor application. These applications also require the use of selectively coated holes.

Here we demonstrate a new approach to selectively filling and coating holes of MOFs, and demonstrate what is to our knowledge the first device based on MOFs having selected holes coated with metal.

2. Existing techniques

While most of the early work on MOFs focused on optical properties of the microstructure, there is increasing interest in varying the material properties, and in particular, the role that the microstructure can play in enhancing material effects.

The first fibers in which different matrix materials were used were produced by the stack and draw method. Using this technique it is easy to combine different materials within the structure, provided they can be drawn together. This technique has been used to produce for example, MOF lasers [16] by using a doped core, all solid bandgap structures [17] and hybrid guiding structures [18], in which some of the air holes were filled with doped glass.

Liquids have also been used to infiltrate the holes of MOFs after drawing. Filling all of the holes of the microstructure with a higher index material can convert a regular hexagonal stacked fiber from an index guiding to photonic bandgap guiding structure [1–3]. If a liquid crystal is used, this approach allows the properties of the bandgap fiber to be made switchable [19], although the speed of the response will limit its applicability.

In a recent –and impressive– experiment Sazio et al. [20] showed that various materials including semiconductors could be incorporated into MOFs using high pressure microfluidic chemical deposition. Chemical Vapour Deposition (CVD) may also be possible for materials including silver [21]. While CVD is in principle capable of achieving very high quality films [22], the very high aspect ratios in fibers make this a difficult task.

Selective filling and/or closing of holes at the fiber stage has received quite significant attention, particularly because it is generally needed for index guidance in liquids [23,24]. There are however a number of other applications that require these techniques, including the production of non-uniform tapers (ie, in which the structure varies with length) [25], birefringent fibers [1] and surface plasmon effects [6–8].

The least demanding of these applications are those which require the closure of holes which have a very different size to that of the holes which remain open. Surface tension preferentially closes smaller holes, an approach that has been used extensively, sometimes with the addition of pressure [11, 12]. Capillary action has provided an alternative approach, as larger holes are more quickly filled than smaller ones. Use of curable glue in combination with differential capillary forces allows to generalize this technique to fill any group of similarly sized holes [10]. However, this class of techniques cannot be used to fill selected holes having sizes similar to that of holes having to remain empty.

A more versatile, but experimentally difficult approach is the direct injection of fluid under a microscope [1, 14, 25]. While Intoni et al have demonstrated selective filling of holes in planar photonic crystals using direct microinjection, the experimental setup requires mechanical and optical access to both sides of the sample, so that the technique cannot readily be applied to MOFs [14]. Kerbage et al filled two out of six holes of a MOF [1] with glue, for a fiber having holes of 30µm tapered down to 8µm. Witkowska et al managed to fill selected micrometric holes with glue [25].

3. Experimental method

A two-stage draw process routinely used in mPOF fabrication was used in this work. A primary preform of 70 mm diameter, made from commercially available polymethylmethacrylate rods, was drilled with the required hole pattern using a computer-controlled mill. This was drawn to a 1 cm diameter intermediate perform or “cane” (see Fig. 1), which was then drawn to fibre.

The aim of our work was to selectively coat the holes of a MOF with silver, for the demonstration of potential plasmonic effects. While there are many techniques available for producing metallic coatings, only a few have proved to be suitable for the small scale of microstructured fibers. One possible coating method involves suction and evaporation of metal nanoparticle mixtures into the fiber. Another is chemical deposition via precipitation from a reduction reaction. The chemical deposition method for silver has been widely used in producing of mirrors, and recently has been used to coat hollow waveguides, because of the ease with which the quality of the silver deposition can be controlled by adjusting reaction conditions [26].

The coating method we developed exploits the neck-down region of the fiber preform, as shown in Fig. 1. Industry grade silicone glue was used to selectively block holes on the perform end of the neck-down prior to solution coating. The method is similar to that previously used in MOFs using a fiber taper [1]. However, the holes at the large end of the neck-down have diameters of the order of the millimeter, making the task of selectively glue-plugging holes straightforward. Furthermore, the small end of the neckdown region gives access to the final fiber, with micrometric holes so that as opposed to Kerbage et al’s technique no further tapering is required to fill micrometric holes.

 

Fig. 1. The fiber neck-down region

Download Full Size | PDF

As shown in Fig. 2, the fiber end of the neck-down is connected to a syringe, allowing suction of the reaction mixture through the unblocked holes. The reaction mixture was an aqueous solution of 0.933 M dextrose and 0.675 M silver nitrate. These reactant concentrations were found to be optimal after reaction rate tests with different solute concentrations. At the preform end, the reaction mixture was maintained at 5°C to prevent excess deposition of silver and blockage. The fiber end was heated to increase rate of silver deposition in that area. Suction was maintained for 24 hours. The residual reaction mixture was removed by suction of water then air through the neck-down for a further 24 hours. Details of the physical deposition process are described elsewhere in the literature [26].

Selective coating was performed on a 5-fold symmetric microstructured fiber design with relatively large holes. Figure 3 shows the structure, indicating the coated holes (left) and an example of the coating (right). Elemental analysis confirmed that the coating was highly localized in the desired holes. The diameter of the smallest hole in which selective coating was successfully applied is 5 µm. The silver coatings are granular but form a well-defined, adhering layer on the inner walls of the fiber structure. Lengths of up to 40 cm were coated by this method. While the coatings in the holes are “continuous” in the sense that the deposition is uninterrupted along the length of the fibre, they are rough and granular, as shown in Fig. 3, and not hence necessarily continuous on the nanometric scale. The detailed characterization of the physical, chemical and optical properties of the coatings inside the fibre is a delicate task requiring further work.

 

Fig. 2. Set-up used for silver deposition.

Download Full Size | PDF

 

Fig. 3. A micrograph of the structure, indicating the two coated holes (top). Elemental analysis shows the presence of silver (dots in the image). An SEM of the silver surface of a coated hole is shown on the bottom.

Download Full Size | PDF

4. The MOF absorptive polarizer

4.1 Theory

The structure (shown in Figs. 3–5) breaks the C_5v symmetry. Thus, the degenerate pair forming the fundamental mode splits into two modes, with orthogonal polarisations. These two modes interact differently with the absorptive silver coating, and consequently have different levels of absorption losses.

We calculate the modes and their overall losses (absorptive and geometric) at a wavelength of 623.8nm using the multipole method [6,27,28]. The background index is n_b=1.492, the silver index is taken as n_ag=(0.13461,3.9883). The uncoated holes are air-filled with n_air=1. The other parameters used are: radii of the two holes: r₁=8 µm, and r₂=4.5 µm; distance from the fiber’s centre to the centre of the holes 16µm and 27 µm respectively, wavelength of light 632.8nm. The thickness of the silver coating was assumed to be larger than several skin-depths, so that the coated holes are simulated as bulk silver.

When uncoated, the structure’s fundamental mode forms a degenerate pair, with confinement loss below 4×10^-3dB/km. Higher order modes of the structure are leaky, having confinement loss of 3dB/km for the LP_1,2 mode and over 745dB/km for all other modes.

Figures 4 shows vector plots of the transverse components of the electric field for the x-and y-polarised fundamental modes of our structure. Calculated losses for the x- and y-polarised modes are 7.24dB/m and 1.26dB/m respectively. The metal coating increases the losses of both the fundamental modes. Here, one higher order mode (LP_1,2, x-polarised) has comparable losses (4.3dB/m), but all other higher-order modes have losses at least an order of magnitude larger than the y-polarised fundamental mode. The lower loss for y-polarised modes is attributed to the fact that these have an electric field predominantly parallel to the large silver coated hole, and therefore cannot couple easily to surface plasmonic modes, while the x-polarised modes have an electric field predominantly orthogonal to the silver surface and couple more readily to surface plasmons. This phenomenon appears to be only weakly resonant, with negligible wavelength dependence. While surface plasmon resonances generally have strong wavelength dependence, modelling indicates the presence of multiple resonances which smear out to produce an effect that is fairly constant with wavelength. This effect is currently under further investigation.

 

Fig. 4. x-polarized fundamental mode, with 7.24dB/m losses (left) and y-polarised fundamental mode, with 1.26dB/m losses (right). The arrow’s colour, length and direction reflect the value and direction of the transverse electric field. Red – highest value, blue – lowest value. Holes in red are those coated with silver.

Download Full Size | PDF

4.2 Polarizer Measurements

The polarizing properties of a 2 cm long section of the coated fiber were characterized by the setup shown in Fig. 5. The experiment was designed to have no movable components in front of the fiber to avoid any shift in the beam position which could change the excitation mode distribution. Circularly polarized light from a HeNe laser (632.8nm line) was launched into the coated fiber using a quarter wave plate. At the detection end, a pinhole was used to isolate the core light exiting the fiber. The intensity of core light at different polarizations over 360 degrees was sampled every 10 degrees via the rotation of a polarizer (P2). The optical power measured by photodiode 2 without the sample fiber was used as a control. To ensure attenuation measurements were not affected by drift, both the signal and control readings were normalized to the total power launched into the system, which was recorded by photodiode 1. The fiber sample being only 2cm long and maintained in a straight line, bend losses or torsion effects are assumed to be negligible. The transmission results as a function of the rotation of P2 are shown in Fig. 6.

 

Fig. 5. Polarizer measurement setup.

Download Full Size | PDF

 

Fig. 6. Transmission intensity through the coated fiber, as a function of the rotation of the polarizer P2 (see Fig. 5).

Download Full Size | PDF

The coated fiber section is observed to reduce intensity of core light in one polarization by approximately 50%. The resulting polarization behavior of the fiber reflects both, the differential attenuation of the two polarization states and possibly their different launch conditions. Having optimized launching conditions, light is predominantly coupled to the two polarizations of the fundamental mode, with the LP_1,2 being only weakly excited. From the simulations, if light was coupled solely into the two polarization of the fundamental mode, differential absorption losses of such a short sample would lead to an extinction ratio between the two polarizations of only 0.12dB, so that 50 cm would be required to achieve a 3dB polarization ratio. However, the silver coatings seen in Fig. 3 are clearly rough and are likely to be partially oxidized and impure, so that higher material losses are expected: The presence of impurities in silver reduces its conductivity, and hence increases its loss.

Simulations show that the experimentally measured extinction ratio of 3dB for the two centimeter long sample is consistent with the silver coating having ten times higher material loss than pure silver. With such high material losses, the x-polarized and y-polarized fundamental modes suffer losses of about 175dB/m and 19dB/m respectively, leading to 3.5dB and 0.38dB loss for the 2cm sample, while higher order modes (except the x-polarized LP_1,2 type mode with about 2dB loss) have losses higher than 30dB for the same sample. While more absorptive films increase the differential loss between the x and y polarization states, they do so at the cost of much greater overall absorption.

We note that the roughness of the surface in the fiber will cause coupling into lossy higher order modes, which will also increase the loss.

5. Conclusions

A selective coating technique for MOFs has been demonstrated. Improvements in coating length and quality will be required to make the technique more useful, however it does demonstrate, to our knowledge for the first time, that selective metal coating in MOFs is feasible, which enables a variety of new applications.

A criticism that could be made of our approach is that it is difficult to scale up, particularly as there are generally only two tapered regions per fiber draw (the neck down region and the drop off). There are however at least two possible solutions to this problem. The draw process itself could be modified by periodically changing the parameters so that each draw produced a series of tapered regions, or the fiber ends could be expanded after drawing. This could produce a useful increase in structure size that would facilitate selective hole closure. For polymer MOFs such an expansion is easily produced by localized heating due to relaxation of stress in the drawn fiber (Fig. 7). A similar effect may be feasible in silica MOFs by the use of localized heating and pressure.

 

Fig. 7. End-face of a polymer MOF expanded using heat treatment. Photograph courtesy Thomas Plochberger, OFTC.

Download Full Size | PDF