1. Introduction

Polymer optical fibers (POFs) have attracted significant attention the past decade due to their various advantages compared to silica fibers [1]. Although the transmission loss of POFs is significantly higher than their counterparts, polymers have several unique features compared to fused silica, such as high flexibility due to their low Young’s modulus, low processing temperature, biocompatibility, etc., thus making them good candidates for several applications including medical [2] and bio-sensing [3,4]. POFs have been extensively used as sensing elements combined with the Fiber Bragg Grating (FBG) technology [5]. Their significantly large elongation levels, strain and thermal sensitivity allow them to be the most promising solution towards structural health monitoring applications with recently a fully compact polymer optical fiber Bragg grating (POFBG) sensor to be commercially available by SHUTE Sensing solutions A/S [6].

The past 20 years, a wide class of thermoplastic polymers have been used for the fabrication of POFs with poly (methyl-methacrylate) – commonly known as PMMA – to be the first one [7]. Several other polymers were then further investigated for the development of POF sensors such as cyclic-olefin copolymer (known as Topas) [8–10], cyclo-olefin polymer (known as Zeonex) [11,12], polycarbonate (PC) [13,14] and the low loss amorphous fluoropolymer (Cytop) [15–19]. The various intrinsic properties of the aforementioned materials can be used to target specific applications. For example, PMMA has high moisture absorption capability and therefore can be applied for humidity sensing applications while Topas and Zeonex have very low affinity to water which is essential for robust measurements in moisture environments [9,20–22].

While POFs have distinct advantages compared to glass fibers, one of their limitation is that they cannot operate at high temperatures due to the low T_g of the host material [10]. The highest reported stable operation of a FBG-based POF sensor was at maximum temperature of 110°C [10], while it was recently reported an increase in the operation temperature of 15°C by using a PC fiber [14]. However, from industrial perspective, expanding the “thermal operational window” of POFs will also significantly expand their use to different applications such as heavy-industry system monitoring [23].

In this work, we propose a novel approach for the development of a heat-resistant POF by combining the well-known optical Zeonex materials with a completely different family of polymers known as high-performance thermoplastics [24]. The latter – while they are not optically transparent in the visible range – are known mainly for their high heat resistance (>180°C), strength, toughness, durability, biocompatibility, as well as their ability to withstand several cycles and doses of all types of radiation [24]. Specifically in this work, we employed the PSU polymer (T_g=189°C) to fabricate the final multimaterial fiber. This polymer is typically produced from bisphenol A and 4,4-dichlorodiphenylsulfone by a nucleophilic process. So far, this material has been mainly used for the development of medical devices and apart from its heat-resistance properties it also has very good mechanical properties [25]. Table 1 provides an overview of the main mechanical properties of different optical polymers (PMMA, PC, Topas, Zeonex and PSU) compared to silica glass for reference.

Table 1. Comparison of the general mechanical properties of different types of thermoplastics and silica glass. ¹ [26], ² [27], ³ [28], ⁴ [29]

View Table

The core/cladding structure of our fiber was based on two different Zeonex grades E48R and 480R, respectively and it was produced via a co-extrusion method as we describe in section 3. Extrusion is a versatile approach to produce fiber preforms and it was first reported using a single glass billet [30]. Several research groups around the globe have been used this method to produce microstructured polymer optical fibers [31,32], soft-glass preforms [33–35] and multimaterial structures [36,37]. Most of the polymer preforms for either step-index or microstructured profiles are made using a drilling or rod-in-tube method. The main advantage of extrusion compared to drilling or rod-in-tube is that complex preform structures can be obtained in a single automated step while the interface between the two billets can be optically smooth through a thermal-imprinting process we describe later in this article. In addition, non-circular holes, large air-filling fractions and long preforms can be obtained, all of which are not readily achieved using drilling [31,32].

2. Material selection and characterization

We first thermo-mechanically characterized our three materials (Zeonex 480R, E48R and PSU) using a Dynamic Mechanical Thermal Analysis (DMTA) system. The materials are exposed to an oscillatory stress or strain while the temperature is changed and thus the resulting stress or strain response is measured. The obtained profiles are then used to identify the characteristic phase transitions such as the T_g and viscosity of the materials [38]. The measurements of the three samples are performed using a tension mode in a rotational rheometer (TA Discovery Hybrid HR-2). Thin rectangular samples of dimensions 2 mm thick, 10 mm wide and 25 mm in length (inset of Fig. 1(a)) are loaded into the clamps and locked with screws. A constant upward force is applied in the axial direction to analyze the solid samples. An axial force of 1 N with a sensitivity of 0.1 N is chosen. The Zeonex samples are first scanned during a temperature sweep from 25°C to 200°C while PSU sample from 25^°C to 280^°C both at a constant heating rate of 10^°C/min across their glass transition temperatures in order to measure the elastic modulus (E’), loss modulus (E'’) and the tan ((δ)) of the materials. The experiments are performed at a constant frequency of 1 Hz.

 

Fig. 1. a) Elastic modulus (E’) as a function of temperature of two different grades of Zeonex (480R and E48R) and PSU where T_g is identified by the fitted lines through the onset decrease of the curves. Inset: Zeonex sample is deformed after each measurement in the rheometer. b) The logarithmic viscosity curve, as a function of temperature c) Measured refractive index and corresponding numerical aperture (NA) as a function of wavelength for the 480R, E48R grades and PSU.

Download Full Size | PDF

One way to identify the T_g from the obtained data is by analyzing the onset of the decrease of the elastic modulus E’ and by fitting two linear lines through the slope of the curves, the intersection of the two fitted lines corresponds to the T_g, as depicted in Fig. 1(a). We measured T_g=143.1°C and T_g=143.2^°C for the 480R and E48R, respectively, showing similar glass transition temperatures and a T_g=189^°C for the PSU. The viscosity of the materials measured in the rotational rheometer is a complex viscosity and it can be calculated through the complex modulus and angular frequency given in rad/sec [39]. By applying the well-known Cox-Merz rule for linear polymers, the complex viscosity is found to be equivalent to the steady state viscosity, a rule that generally holds for most polymer materials within ±10% [39,40].

The characteristic thermal transitions of 480R, E48R and PSU can be observed through the logarithmic of the viscosity-temperature profile (Fig. 1(b)). At low temperatures, the polymer chains are tightly compressed and locked in place, hence acting as a solid material. The complex viscosity is usually high in this region. As the temperature increases, the polymer expands and the free volume increases. As heating continues, the T_g was reached and the material goes from a solid state to a rubbery state. In this transition region the complex viscosity decreases dramatically and reaches a rubbery plateau region where it tends to decrease much slower. Finally, as the temperature increases further, the viscosity decreases again since chains can slide past each other and the material exhibit viscous flow [32] allowing for extrusion process as well as fiber drawing. Finally, in order to verify the refractive index values provided from the company (Zeon Corporation), we measured the refractive index (RI) profiles (NA is also shown for reference) of the two different Zeonex grades (grade 480R and E48R) and PSU using a commercial ellipsometer (J.A. Woollam) from ∼211nm to ∼ 900 nm wavelength as shown in Fig. 1(c).

3. Co-extrusion and fiber fabrication

The two different grades of Zeonex (grade 480R and E48R) are manufactured by casting commercially available polymer pellets (Zeon Corporation) which are then mechanically cut into smaller pieces of solid disks with dimensions matching the extrusion equipment. PSU is commercially available as solid rods. The bulk samples were stored in a vacuum oven at an annealing temperature below their T_g (∼120^°C) for several days in order to eliminate any internal stress, trapped air and moisture inside the material which can otherwise increase extrinsic loss factors in the fiber preform and consequently the final drawn optical fiber. The core/cladding interface of two Zeonex samples need to be smooth and flat before co-extrusion, to reduce any possible scattering loss that will manifest in the optical fiber. Throughout the years, several methods based on polishing have been presented, such as a rotational diamond cutter [41], sandpaper cutter and chemical-mechanical polishing [42]. Nevertheless, these polishing methods are costly, time consuming and often provide ineffective result for thermoplastic polymers.

Here we demonstrate a so-called thermal-imprint or hot embossing technique to flat and make the samples optically smooth by applying heat and pressure on the bulk disc. This method allows low-cost, time-efficient, and high quality hot embossed surfaces [43,44]. Figure 2(a) shows a schematic of the hot embossing method used prior extrusion. The polymer sample is placed on a clean flat silicon wafer between two cylindrical metal plates. One of the metal plates is attached to a stepper motor controlled from a computer to move downwards and upwards. The other metal plate is connected to a heating element controller by a temperature controller while it remains stationary. When the temperature has reached slightly above the T_g of the material, the metal plate is moved downwards slowly and applies moderate pressure to the sample for a few seconds. The Zeonex samples are heated to a temperature of ∼165^°C with an applied force of around 50 N/cm² for approximately 1 minute and then we gradually decreased the temperature until it gets below the T_g of the polymer. It should be noted that great care has to be taken in apply the optimum pressure, otherwise the material will heavily deform.

 

Fig. 2. a) Schematic of the hot embossing procedure with the resulting facets of the two Zeonex discs. b) Schematic of co-extrusion process of two different grades of Zeonex samples vertically stacked c) Recorded data during co-extrusion showing the die temperature and corresponding force profile versus punch travel d) Measured outer diameter (OD) at intervals along the preform length (extrudate). Inset: light through extrudate and cross-sectional view of preform.

Download Full Size | PDF

All extrusions were done in a custom-made in-house extruder at DTU Fotonik. A schematic of the co-extrusion process can be seen in Fig. 2(b). The extrusion system consists of a stainless steel barrel inside a height-adjustable tube furnace. The samples are placed inside the barrel and stacked vertically so that the cladding material (480R) is placed underneath the core material (E48R).

The samples with diameter 29.5 mm and thickness of 20 and 10 mm, respectively are heated slightly above their softening temperature which in our case was (T_g + 65°C) in order to make sure that the material exhibited viscous flow. The samples are then pushed by a hydraulic punch through a circular die fabricated from stainless steel, which has been found to be a suitable material due to its high mechanical stability and chemical inertness at high temperatures.

The temperature is measured using a thermocouple placed near the die exit. This is because it is not possible to measure the actual temperature of the sample during extrusion without interrupting the polymer flow. Nevertheless, when the extrusion temperature has reached and stabilized, the sample will also reach the same temperature of the die. The heating of the furnace took approximately 3-4 hours to stabilize while the extrusion process took approximately 2 hours to complete, depending on the punch speed (typical range: 0.35–0.4 mm/min), extrusion temperature and initial thickness of the samples. The die temperature (±1°C uncertainty) and the punch force were recorded as a function of position and time. The force profile based on the change of the punch force during extrusion is demonstrated in Fig. 2(c). Initially, at zero position the sample is resting on top of the extrusion die where the punch is located at the top surface of the bobbin. The punch force rapidly increases when the polymer sample starts filling the die and flows against the die. When the force reaches a region where it does not change significantly with position, the polymer starts to emerge from the die exit. This process occurs when the viscosity of the polymer has decreased and the sample is soft enough to flow out of the die. The punch force increases significantly at the end of extrusion due to the remaining material being forced through the die punch reaching the top of the die.

The outer diameter (OD) at intervals along the preform length (extrudate) is measured using a micrometer-precision caliper. Slight decrease (<1 mm) of the total diameter can be observed along the preform length due to a slight increase of the die temperature (see Fig. 2(c)). The length of the extrudate can be primarily controlled based on the thickness of the core/cladding disks and the extrusion die diameter. The diameter of the extrudate depends mostly on the die diameter as well as control of the temperature and punch speed. In addition, light was shined through the co-extruded preform to observe the core-cladding interface along the preform length, as depicted in Fig. 2(d). The core material was forced through the cladding material and exhibited a parabolic shape, where the flow diverged throughout a large section of the preform as it has been observed elsewhere [35,36]. In this region, the preform was cut and the cross-sectional core-cladding interface could be observed clearly seen in Fig. 2(d). A cylindrical PSU with diameter of 30 mm and a central hole diameter of 8 mm was used to insert (rod-in-tube) the extruded preform and form the final fiber preform, as depicted in Fig. 3(a) and (b). PSU and co-extruded Zeonex preform are then drawn together as a single concentrically structured optical fiber. The temperature of the middle zone of the furnace during drawing was stabilized to ∼260°C. As depicted in Fig. 3(b), the core, cladding and the total fiber diameters of the final fiber are around 70 µm, 13 µm and 300 µm, respectively.

 

Fig. 3. a) Schematic of thermal drawing of two different grades of Zeonex (grade 480R and E48R) and PSU b) Rod-in-tube approach and final optical microscope images of the cross-sectional facet of multimaterial fiber and the fiber tightly coiled.

Download Full Size | PDF

4. Optical characterization and thermal stability

The experimental setup used for the transmission, loss as well as thermal characterization of the fiber is illustrated in Fig. 4(a). In our experiments, we used a supercontinuum laser (SuperK Versa - NKT Photonics) that spans from visible (400 nm) up to near-infrared (2000 nm) spectral range. The output from the SuperK was coupled to a SuperK Split module in order to use only the visible part of the spectrum (up to 900 nm wavelength) that is then butt-coupled to the POF and the output light from the fiber is collimated by an objective lens L1 (10x, NA=0.25). The collimated light is directed to a 50/50 beam splitter where part of the light is directed into a camera that captures near-field images of the output light from the optical fiber. An iris is used to block the light from the cladding so that only light from the core is recorded. The light is then directed to an objective lens L2 (20x, NA=0.4) which focuses light into a 550 µm multimode fiber (NA=0.22) that is connected to an optical spectrum analyzer (OSA ANDO AQ-6315A) which records the transmission spectra of the optical fiber as seen in Fig. 4(b). The transmission loss of the fiber is measured using a cut-back method where the fiber was cut from ∼2 m down to 70 cm, recording the transmission spectrum at 5 different lengths. The fiber was placed on a flat surface and cleaved mechanically using a sharp blade. The transmission and loss profile of the fiber from 500 nm up to 900 nm wavelength is shown in Fig. 4(b) and (c), respectively. The loss peaks (dips in the transmission) observed at around 750 nm and 900 nm are linked with the C-H vibrational overtones absorption bands of the material [45]. The minimum loss was found to be ∼13.9 dB/m at 800 nm which is higher than the previous reported [12]. This can be attributed to the i) multimode nature of the fiber that is known to have higher loss than a single-mode fiber [46] and ii) thin 480R cladding with a high RI PSU over-cladding which further increases the absorption and scattering. The total loss of the fiber can be further minimized by reducing the core diameter and increasing the 480R cladding of the fiber.

 

Fig. 4. a) Schematic of experimental setup used for optical characterization and thermal stability measurements b) Transmission spectra of a 70 cm multimaterial fiber c) Optical loss of the multimaterial fiber measured using a cut-back method with minimum loss of 13.92 dB/m

Download Full Size | PDF

The thermal performance was characterized by placing the fiber inside a 4 × 4×100 mm custom-made furnace which is heated by a heating plate (Thermo Scientific). A thermocouple is also placed inside the furnace to monitor the temperature using a digital thermometer. A thermal power meter (Thorlabs S401C) is placed after the iris to continuously record the power of the output light using the Thorlabs power monitor software. In order to directly compare the thermal stability of the presented fiber, a step-index single-mode pure Zeonex fiber with length of 73 cm, fiber diameter of ∼125 µm and core size of approximately 4.8 µm was also characterized [12]. In addition, we repeated the same experiment using a commercially available low-loss Cytop fiber (GigaPOF-50SR). At each step, we left the temperature to stabilize for at least 10 minutes [10,13,21]. The normalized output power as a function of time at elevated temperatures for each different fiber can be seen in Fig. 5(a).

 

Fig. 5. a) Normalized output power as a function of time for different POFs b) Normalized output power as a function of time for the multimaterial POF at 180^°C for 5 hours and at 160^°C including cooling/heating of the fiber.

Download Full Size | PDF

The multimaterial POF after stabilizing at 180^°C maintained a stable output power for several hours while the step-index Zeonex and Cytop fiber were fully degraded when they reached 145^°C and 120^°C, respectively, as shown in Fig. 5(a). It should be noted that none of the fibers recovered after cooling them to 25^°C. Furthermore, the output power of the multimaterial POF was recorded at 180^°C for over 5 hours to make sure it will not degrade over time. The measurements were repeated over different full cycles (cooling/heating) up to 160^°C for more than 4 hours as shown in Fig. 5(b). The decrease in power with increasing temperatures can be attributed to the change in the refractive index with temperature (thermo-optic effect) that can increase the propagation loss of the fiber as it has been previously discussed in [47]. Furthermore, we anticipate that PSU due to its high T_g, the core/cladding step-index structure maintains its geometry even if the Zeonex materials are in a “soft” state at 180^°C.

5. Conclusion

In conclusion, we have presented for the first time to the best of our knowledge the fabrication of an all-polymer multimaterial optical fiber based on two different grades of Zeonex (E48R and 480R) and PSU with a T_g=189°C. The initial preform was fabricated using a co-extrusion method followed by a rod-in-tube process and it was thermally drawn to a fiber with ∼70 µm and ∼300 µm core and total diameter, respectively. By adding an extra fabrication step (rod-in-tube of the extruded preform to a 480R tube), the core diameter could be further scaled down to ∼5 µm becoming single-mode in the visible range [12]. The loss of the fiber was measured to be ∼13.9 dB/m using a cut-back method and it is anticipated that the loss can be further reduced by increasing the cladding size of the 480R material while decreasing the core size. Finally, the proposed POF was thermally compared with a standard single-mode step-index pure Zeonex and a commercially available Cytop fiber. The proposed multimaterial POF was able to withstand temperatures up to 180°C over several hours paving the way towards applications where stable operation above 170°C is a crucial factor.