Thank you for visiting nature.com. The browser version you are using has limited CSS support. For the best experience, we recommend using the latest browser version (or disabling compatibility mode in Internet Explorer). Additionally, to ensure continued support, this site will be free of styles and JavaScript.
The production of optical fibers with microstructured cross-sections is based on a multi-step process that includes the production of an optical fiber preform and the drawing of the preform into an optical fiber. These processes require specialized and advanced equipment, including a fiber draw tower. This paper shows how to produce optical fibers with complex microstructures in a single step, from the optical material pellets directly to the finished fiber, using a commercial, low-cost benchtop filament extruder. This process eliminates the need for a fiber draw tower and saves time, energy, and floor space. Fibers with different geometries (hexagonal solid core, suspended core, and hollow core) have been successfully produced and tested. The figure shows the air flow path in a hollow core fiber in the wavelength range where the fiber material is opaque.
Fiber optics have revolutionized the way we communicate, carrying the vast majority of the world’s data traffic. There are hundreds of millions of kilometers of fiber optics laid around the world. The volume of data transmitted doubles every two years, meaning it has increased 1,000-fold in just 20 years.
Optical fiber technology made significant progress in the late 1990s when structures with internal microstructured cross-sections were proposed and developed. The development of photonic crystal fibers (PCFs) or microstructured fibers (MSFs), pioneered by Philip Russell and his research group at the University of Bath in the UK, has expanded and revolutionized the field of fiber optics.1,2,3 The availability of wavelength-scale structures with high refractive index contrast (fiber material versus air) opens up the possibility of extensive control over the optical properties of the fiber. For example, properties such as dispersion, modal area, cladding attenuation field, birefringence, and nonlinearity are highly dependent on the specific distribution of holes—size, shape, and location.1,2 Conventional optical fibers, on the other hand, exhibit low core/cladding refractive index contrast, typically less than 1%.
While solid-core optical fibers with a low-refractive index holey cladding transmit light via total internal reflection like conventional optical fibers, hollow-core fibers (HCFs) implement new transmission mechanisms. The complex cladding structure enables light transmission across photonic band gaps. Simpler structures provide low-loss transmission due to suppressed coupling4 or antiresonance5.
Although most conventional optical fibers and metal-organic fibers (MOFs) are made from silica due to its exceptional optical and physical properties, optical fibers can also be made from polymers and non-silica. The development of microstructured polymer optical fibers in the early 2000s6 has expanded the application range of conventional polymer optical fibers.
In all cases, optical fibers are typically drawn in a multi-step process, the main step of which is to create a larger version of the fiber, known as a preform. Various approaches have been used to produce macroscopic preforms. Standard optical fibers use vapor deposition to produce low-loss preforms. MOF preforms, on the other hand, with their characteristic arrays of air holes, are produced using different technologies. Silicon MOFs are typically produced by a stacking process1, in which millimeter-thick capillaries are manually stacked on top of each other to form the desired structure. This is a convenient and versatile process when tubes are widely available, such as for quartz glass and some borosilicate glasses such as Duran7. However, stacking is time-consuming. Soft glass MOFs can also be produced by this process, but it is more complex due to the initial need to make the tubes8.
Polymer MOFs are made by drilling holes directly into plastic rods6, which is also applied to glass9,10. Similar to the stacking method, drilling is limited to circular holes. It is also limited to short blanks. Another approach is to cast the fibrous material into a pre-designed mold, which is also used to produce plastic11 and glass fibers12.
Soft glass and polymer preforms can also be produced by preform extrusion, a simple and convenient method for producing carefully designed structures. Preform extrusion involves preparing a preform of the chosen optical material, heating it to reduce the viscosity (typically 108–1010 dPa⋅s13), and then extruding the material through a die with a given pattern14 using a punch. The die consists of an initial section with holes for feeding the extruded material and a subsequent section with solid elements to block the material flow in given areas, which allows preforms with a porous pattern to be extruded. This method has been shown to be successfully applied to the preparation of high-quality MOFs from soft glasses14 (e.g., lead silicates, tellurites, bismuth, fluorides, chalcogenides, phosphates) and polymers (e.g., PMMA15,16).
For soft glass, the parison extrusion equipment can be combined with the fiber drawing equipment, positioning it directly above the fiber drawing tower. In this case, the extruded parison is heated in the tower furnace and drawn to a reduced diameter. 17 Multi-component parisons can also be extruded, where the parison consists of several different materials stacked on top of each other. 18
Extrusion dies are typically manufactured on CNC machines, and the most common mold material is stainless steel. Recent studies have shown that 3D printed Cr-Co-Mo and titanium molds can withstand the high temperatures (560–600 °C) and high forces (20 kN) encountered during the extrusion of commercial lead silicate glass (19) without any mechanical damage to the 3D printed parts. This provides unprecedented freedom in mold design with 3D printing. 20 Recently, 3D printed titanium molds were used to fabricate multi-strand MOFs for imaging. 19 In that study, optical fiber preforms were extruded through a mold and subsequently drawn into optical fibers, while in this study, 20 four-strand fiber rods were extruded, stacked into a 100-strand structure, and finally drawn into optical fibers. Notably, 3D-printed molds have higher surface roughness than machined molds, which may affect the surface quality of both the extruded preform and the optical fiber. This may particularly negatively affect the scattering loss in microstructured optical fibers, which depends on factors such as the refractive index of the fiber material and the core size. However, studies have shown that this issue can be addressed by simply polishing the last few millimeters of the inner surface of the spinneret exit orifice (accessible to the nozzle),19 resulting in extruded optical samples with a surface quality similar to those produced using a machined spinneret.
It is important to note that the extrusion process allows all cross-sectional elements to be formed simultaneously, unlike the sequential drawing and drilling processes, where holes are formed sequentially. However, the standard approach is to first extrude the macroscopic preform and then draw it to the fiber formation stage, which means a multi-step process that requires complex equipment.
Standard all-polymer fibers can be extruded in a single process where two materials are fed simultaneously to form a fiber composed of two different materials/compositions, thus forming a core-sheath interface. In this case, the feedstock can be polymer particles or purified monomers.
On the other hand, extrusion of glass or polymer MOF structures uses bulk feedstock rather than pellets. The preforms are usually cut from larger blocks of material or produced by fully fusing the feedstock/pellets to ensure high optical quality.
In the last few years, a new method for producing optical fiber preforms has been developed based on additive manufacturing. This method starts with 3D printing the preform itself, which is then drawn into an optical fiber using a specialized draw tower. Using commercially available polymer filaments, both hollow-core21,22 and solid-core23 optical fibers can be produced. Guidance methods in the visible and infrared ranges have been demonstrated. Recently, this technology has been extended to printing glass samples such as borosilicate24, silicon25, and chalcogenide26. Step-index optical fibers27 have also been produced using multi-step additive manufacturing.
Another breakthrough in this area was the idea of simplifying the MOF production process by using acrylonitrile butadiene styrene (ABS) filaments and a special micromechanical nozzle28 to extrude the MOF directly from a 3D printer. Although suspended core filaments were successfully produced, their lengths were short and the pores tended to collapse. Although this method combines extrusion and drawing in a continuous process, the starting material, the polymer filament, must be produced from pellets in a separate process using different equipment.
In this study, a single continuous process from pellets to finished MOF was realized by simultaneously extruding pellets and drawing the extruded material using a compact benchtop horizontal extruder originally designed for producing 3D printer filaments. This process is fundamentally different from existing preform extrusion methods for MOF production, which involve multiple steps and equipment, including using glass/polymer melting capabilities to form preforms, extruding the preforms into preforms using a ram extruder, and finally drawing the preforms into optical fibers using a draw tower. In contrast, our new single continuous process technology requires only a basic set of equipment to produce MOFs directly from pellets. Moreover, unlike using a 3D printer extruder as a tool, where MOFs are produced from filaments through simultaneous extrusion and drawing, our simultaneous extrusion and drawing process uses pellets directly as a starting material, eliminating the need for extrusion to produce filaments. In addition, using a compact extruder to produce filaments directly from pellets improves process control and stability, as these extruders are already designed to produce fiber-like structures (i.e., 3D printing filaments). Our process enables the production of complex fiber geometries while saving cost, time, energy, and space. Metal 3D printing has proven to be an efficient nozzle fabrication method for MOF production. To demonstrate the applicability of our new technology to the fabrication of various MOF structures, three well-established MOF geometries were selected. The extruder flow parameters were determined and the filament geometries were characterized. Analysis of optical guidance inside the hollow fiber showed that guidance occurs inside the air core in the spectral range where the fiber material is opaque.
While PMMA is widely used in the production of standard optical fibers and optical components, cyclic olefin copolymer (COC)29 is increasingly used in the production of optical waveguides, including specialty optical fibers. The most common COCs for optical fibers are Zeonex and Topas. Their main advantages include low water absorption, reduced brittleness, easier forming, reduced birefringence, improved chemical and thermal stability30, and high transmittance in certain frequency ranges (such as THz30,31,32). Zeonex is also thermally compatible with PMMA, allowing optical fibers to be drawn from both materials33. Commercially available Zeonex 480 R particles34 were used in this study. This material has a glass transition temperature (Tg) of 138 °C, water absorption of less than 0.01%, and a refractive index (at 589 nm) of 1.525.
ABS was also used to demonstrate the feasibility of fabricating doped MOFs directly from functionalized polymer materials. ABS is widely used for injection molding and extrusion in additive manufacturing. Due to its stress cracking properties, it whitens under stress and recovers upon heating, a property that can be exploited in fiber optic sensor applications. 35 ABS is also highly soluble in acetone, allowing the preparation of doped ABS powder by adding the dopant to a dilute ABS solution and then evaporating the solvent. As a proof of concept, a luminescent MOF was prepared in this study using rhodamine dye as the dopant. The details of the preparation of doped ABS powder are described in the Methods section. Note that PMMA can be subjected to exactly the same doping procedure since it also has high solubility in acetone.
To demonstrate the versatility of the methodology developed in this study, we selected three types of MOFs with different cross-sections and geometric characteristics.
The photonic crystal fiber (PCF) geometry is based on the well-known and widely studied triangular air-hole array with a missing hole (defect) in the center forming the core. The optical properties of the PCF, such as modal area, confinement loss, dispersion, and birefringence, can be tuned over a wide range by adjusting the ratio of the aperture diameter (d) to the pitch (Λ). 36 For example, if the fiber hole is small enough (d/Λ < 0.43), a ring-shaped single-mode fiber can be obtained that transmits only the fundamental mode. 3 In this study, a PCF with a target d/Λ ratio of 0.5 was chosen.
The suspended core fiber (SCF) geometry is a widely studied and applied type of metal–organic framework (MOF) because it combines the simple structure of a single-component fiber37 with a large number of air holes surrounding the core (in contrast to, for example, the triangular lattice PCF geometry). This makes SCF an ideal platform for devices based on the evanescent field interaction of an optical mode with a target liquid or film9,38. In this study, the three-air-hole SCF geometry was chosen.
Hollow-core fibers (HCFs) enable the guidance of light within a low-refractive-index core via either photonic bandgap or suppressed coupling/antiresonant guidance. These structures have provided significant advances in optical field control and opened up a number of exciting possibilities in fundamental and applied science. Over the past two decades, fiber-based components have been extensively investigated, ranging from broadband supercontinuum sources to pulse compressors and high-power all-optical transmission channels. Gas-filled hollow-core fibers are an excellent platform for gas-filled nonlinear optical devices.39,40 Novel fiber designs are widely investigated to extend the transmission window, reduce losses, or simplify the waveguide cross-sectional geometry. Recently, an HCF geometry based on a single-ring hexagonal core connected to the fiber cladding via thin struts has been proposed and demonstrated in lead-silicate glass. This fiber was fabricated using a preform extrusion method and demonstrated Raman scattering (RS) measurement performance.41,42 A similar geometry was subsequently demonstrated in silicon43,44 with a slightly more complex cladding structure fabricated by capillary layup. In this study, the HCF geometry chosen was a single-ring hexagonal core.
Three different MOFs were fabricated using a commercial horizontal filament extruder (Filabot EX-2). This extruder, originally designed to produce polymer filaments for layer-by-layer 3D printers, works by feeding polymer powder or pellets into a nozzle with a round opening. By using a custom-made nozzle and pulling the material out, we were able to produce MOFs with different geometries, from pellets to fibers, in a single continuous process.
The commercial horizontal filament extruder shown in Figure 1a consists of a rotating screw that is heated as the extruded polymer is fed. The extruded material can be in the form of granules or powders. The temperature of the heater at the end of the rotating screw, connected to the nozzle, must be adjusted to match the processing temperature of the polymer. The shear forces generated by the rotation of the screw also heat the polymer. 45 The feed rate of the material is controlled by the rotation speed of the screw. Two pulleys of the drum system, originally designed to draw and wind the extruded filaments, are now used to draw the MOF directly from the material flowing from the nozzle. The pulley speed can be adjusted to control the drawing speed and the fiber diameter.
(a) Schematic of the fiber extrusion setup including a commercial filament extruder, filament pulling system, and a 3D printed nozzle. The inset shows an image of the fabricated 3D printed nozzle with a dual-ring photonic crystal fiber, a hollow core fiber, and a suspended core fiber arranged from left to right. (b) The cross-section shows the polymer flow through the feed chamber (FC) and the fusion chamber (WC) (transparent gray arrows).
The design of the internal structure of the nozzle is based on the previously developed concept of the extrusion head.14 The design of the nozzle consists of two sections arranged along the direction of the material flow. The first section, the so-called feed chamber, contains a number of longitudinal holes through which the material transported by the rotating screw is fed into the second section, the so-called welding chamber, which contains a number of solid longitudinal pins blocking the flow of material. The obstruction of the flow in the welding chamber leads to the formation of pores in the material exiting the welding chamber into the free space at the exit plane of the nozzle. The pulling of the material as it exits the nozzle ultimately leads to the formation of pores in the fiber. The manufacturing process of the nozzle is described in the Methods section.
The inset of Fig. 1a shows a photograph of the three nozzles used to fabricate the three MOFs in this study: a solid-core PCF with two hole circles, an HCF, and an SCF with a suspended core. For clarity, these nozzle types will be referred to as “PCF nozzle” (nozzle exit diameter D = 13 mm), “HCF nozzle” (D = 12.2 mm), and “SCF nozzle” (D = 9 mm), respectively. Nozzles with more complex cross-sections, such as PCF, are made wider to avoid excessively thin pins.
For the three types of nozzles (PCF, HCF and SCF), the cross-sectional area AF of the flow channel at the nozzle outlet plane (where the material enters the free space) is 124, 66 and 41 mm², respectively. The flow area fraction, defined as the ratio of the flow channel area AF to the total area AT at the nozzle outlet plane (π/4×D²), is 93%, 56% and 35%, respectively. AF determines the material feed rate as described below. The difference between the flow channel area fraction and 100% is the flow obstruction area fraction AO/AT, where AF + AO = AT. The AO/AT values for the three types of nozzles are 7%, 44% and 65%, respectively. Note that the solid area Asolid and the air-filled area Aair of the cross-section of the material at the nozzle outlet are equal to the flow channel area (Asolid = AF) and the flow obstruction area (Aair = AO), respectively, at the nozzle outlet.
Pellet extrusion occurs at a molten material viscosity (approximately 100 dPa s). Drawing the extruded material (i.e. the material exiting the nozzle) into a fibrous form reduces the cross-sectional area of the material and prevents the pores from collapsing. This is fundamentally different from bulk extrusion, which occurs at a higher viscosity where the material simply softens (approximately 108 dPa s), eliminating the need for tension to maintain the pores.
The extrusion mass flow rate (μfeed = mass/time) from the PCF and HCF nozzles was measured to calculate the expected fiber diameter at a given drawing speed. Zeonex polymer pellets were free extruded without drawing, with the screw speed adjusted using a knob on the extruder barrel. Five samples were extruded at each knob position and their mass was measured. The extrusion time at each position (20 to 90 seconds) was chosen to produce a sample mass of approximately 1.5 to 3.5 grams. The mass flow rate at a given knob position is the arithmetic mean of the mass flows of all samples at that knob position.
Figure 2a shows that for Zeonex products extruded at 208 °C using PCF and HCF nozzles, the mass flow rate increases linearly with handle position (slope of fit = 0.0374, screw speed index) over the range of 0.02 to 0.13 g/s. No significant differences were observed at extrusion temperatures of 215 °C and 222 °C, indicating that screw speed is the dominant factor in mass flow rate.
(a) Extruder mass flow rate measured with Zeonex with “PCF nozzle” and “HCF nozzle”; (b) Calculation of feed flow rate for different flow area (AF) at the nozzle exit plane; (c) Calculation of the expected final fiber diameter for a nozzle ID of 10 mm, a mass flow rate of 0.03 and 0.12 g/s and a nozzle spinneret exit area of 35 and 70 mm2; (d) Calculation of the expected fiber diameter for two different feed flow rates (0.285 and 0.74 mm/s) using a “PCF nozzle” (D = 13 mm, A = 124 mm2) and using the experimentally measured diameter at a feed flow rate of 0.74 mm/s.
Using the experimentally obtained mass flow rate (μfeed = 0.02-0.13 g/s), the known polymer density (ρ = 1.01 g/cm3 for Zeonex) and the known nozzle outlet area (AF), the polymer feed rate (vfeed) can be calculated using equation (1):
For different handle positions (0 to 3) and typical values of nozzle cross-sectional area (AF = 35, 70 and 120 mm²), the calculated feed rates range from vfeed = 0.2 to 3.0 mm/s. Figure 2b shows that for the same mass flow rate (i.e. the same screw speed), a higher AF value results in a lower feed rate.
The polymer material extruded from the nozzle can be pulled manually or with a puller as shown in Figure 1. Assuming that the air fill fraction of the fiber cross section remains constant relative to the air fill fraction at the nozzle outlet as the material is pulled, the law of conservation of mass allows the fiber diameter (d) to be calculated from known values of the nozzle outlet diameter (D), feed rate (vfeed), and fiber pull rate (vpull).
For typical values of D (10 mm), μfeed (0.03 and 0.12 g/s), AF (35 and 70 mm²) and ρ from Zeonex, Figure 2c shows that the calculated fiber diameter is <2 mm for drawing speeds of 10–100 mm/s (i.e. 0.6–6 m/min) and feed speeds of 0.4–3.4 mm/s (i.e. 25–204 mm/min). For a fixed drawing speed, the fiber diameter decreases with decreasing mass flow rate and increasing nozzle exit area. In other words, finer fibers are easier to obtain at lower feed speeds, higher solids loading fraction (i.e. lower air loading fraction), smaller nozzle exit diameter and higher fiber drawing speed.
It is worth noting that when drawing optical fiber using a pellet extruder (described in this paper), typical feed rates (approximately 10-100 mm/min) are 1-2 orders of magnitude higher than those of a standard fiber drawing tower (approximately mm/min). This is primarily due to the fact that to produce a fiber of the same diameter, a higher extruder feed rate requires a higher drawing speed compared to a drawing tower. This is primarily due to the fact that extruder manufacturers initially developed the screw speed control system for high-throughput filament production, which requires a higher mass flow rate.
Figure 2d shows the calculated fiber diameter as a function of the drawing speed for the PCF nozzle design (D = 13 mm, AF = 124 mm²) at two different feed rates: vfeed = 0.285 mm/s (knob position 0.5, mass flow rate = 0.035 g/s) and 0.74 mm/s (knob position 2.0, mass flow rate = 0.09 g/s). The fibers were produced at a feed rate of 0.74 mm/s and their diameters were measured manually with a caliper. The calculated and measured fiber diameters were in good agreement (Figure 2d).
All manufactured optical fibers were manually cleaved using new blades without any process optimization. It is worth noting that heating the optical fiber and/or blade 46, as well as polishing the end of the optical fiber 47 after cleavage can improve the smoothness of the cross section.
Using the nozzle shown in the inset of Fig. 1a (left), a solid Zeonex photonic crystal fiber (PCF) with two circles of holes was fabricated. The solder joint of the 3D-printed nozzle contains 18 pins that form a pattern of holes. The pins are 0.8 mm thick and spaced 1.55 mm apart, resulting in a d/Λ ratio of 0.52. The overall inner diameter is 13 mm.
Figure 3a shows the cross-sections of fibers extruded at two different temperatures (208 and 215 °C) and two different feed rates (0.45 and 0.75 mm/s) with drawing speeds of approximately 20–50 mm/s. At higher temperatures and feed rates, the cladding air fill fraction (d/Λ) decreases significantly due to the partial collapse of the cross-section of the fiber drawn at low viscosities. Figure 3b shows an example of a fiber with an outer diameter of 1450 μm, an average pore size of 200 μm, and an average pore spacing of 270 μm (d/Λ ~ 0.75).
(a) Optical images of Zeonex solid-core polymer fibers produced at different temperatures and feed rates. All fibers were drawn at speeds ranging from 20 to 50 mm/s and had outer diameters ranging from 1.5 to 2.0 mm. (b) The fiber diameter is 1450 μm and the core diameter is 385 μm. The white scale bar represents 200 μm.
Since the solid PCF geometry has a larger flow area (AF) in the nozzle design and the printing nozzle has a larger diameter (D), further reduction of the fiber diameter requires some modification of the extruder. In particular, a higher draw speed will allow for thinner fibers to be produced with this geometry.
Two 3D printed nozzles for producing pendant cores, each with different support rod thicknesses (0.5 mm or 0.8 mm), were used with transparent ABS pellets to produce polymer pendant cores. To optimize the production parameters, the extruder screw speed was adjusted to control the feed rate (from 1.5 to 3 mm/s). The results are shown in Figure 4. The drawing speed was adjusted to produce fibers of uniform diameter. In order to be able to vary the feed rate while maintaining a constant fiber diameter and not exceeding the maximum drawing speed, fibers with a diameter of 2.0 mm were produced.
Optical images of SCF ABS resin extruded at 160 and 165°C using different feed rates and two SCF nozzles with different rod thicknesses (0.5 mm and 0.8 mm). The cracks in the fibers are only visible on the surface and are the result of a suboptimal cutting process. The white scale bar in the lower right image is 200 μm and is the same for all nine images.
The results show that the thickness of the fiber columns increases with increasing feed rate. Moreover, as expected, nozzles with thicker columns (middle and bottom rows) also produce thicker fiber columns.
To minimize retention losses, fibers with thinner support structures are required. The results presented in Figure 4 show that the ideal conditions are a low feed rate, a temperature of about 160 °C, and a nozzle with a support structure width of 0.5 mm.
MOF manufacturing typically involves controlling the pore size (and shape) by applying positive or negative pressure to the preform during the drawing process. In this study, we explored this concept during the extrusion of fiber pellets. We designed a customized mold similar to the one shown in the inset (right) of Figure 1a, but with air channels in the flow blocking region (center image in Figure 5). As a proof-of-concept experiment, one of the pores was subjected to pressure during the extrusion of Zeonex pellets. The pressure was increased during fiber drawing to demonstrate the expansion of one of the pores and, as a result, the displacement of the fiber core from the center to the outer solid region. The temperature was kept constant (234 °C) throughout the process. Figure 5 shows optical images of a 1500 μm thick fiber as the pressure was increased from 0 mbar (Figure 5a) to 20, 40, and 80 mbar (Figures 5b, c, d). As expected, the struts surrounding the expanded pore became thinner.
The center of the image shows a specially designed nozzle used to control the pressure inside the orifice during extrusion. The optical images show the production of a Zeonex self-adhesive film with an outer diameter of 1500 µm without applying pressure (a) and with increased pressure (b, d). The orifice with positive pressure is indicated by the yellow dot. The white scale corresponds to 200 µm.
Post time: Sep-05-2025