We report the design, fabrication and characterization of 3D large core (1 mm) multimode fiber splitters using a low-cost stereolithography 3D printer. The results were accomplished for symmetrical 1 × 2 and 1 × 4 splitters, where the angle between the output arms was varied from 10 to 180°, showing good uniformity between the splitting ratios. Asymmetrical 1 × 2 splitters were also studied to achieve different splitting ratios. This was done by fixing one arm at a specific angle, while varying the other. Results were quite satisfactory, paving the way for simple and customizable manufacturing of passive optical elements.
© 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
Polymer optical fibers (POFs) have unique characteristics, such as high flexibility, low weight, immunity to electromagnetic interference and easy installation, due to their large core diameter (1 mm), allowing the use of simple plug connectors as well as inexpensive LEDs. Because all of these characteristics, POFs have been employed in a variety of applications. Among them are their use as sensors , lighting or decoration , automotive applications , namely for multimedia data transmission, such as multimedia oriented system transport (MOST), IDB 1394, and FlexRay communication system [3,4], short-range data communications [5–7], among others. For most of these technologies, it is sometimes required the use of an essential component capable to split or combine optical signals into multiple input or output ports, respectively. Furthermore, the optical power can be equally or unequally distributed among the different arms. Optical splitters/combiners are a passive component capable to execute this task with minimum insertion loss. Nowadays, it is possible to find a variety of methods to produce this passive component, such as the side polishing method , the hot embossing method [9,10], the injection molding in a metallic molding mask , the use of ink-jet printers , UV laser direct writing on a curable polymer  and Light-Induced-Self-Written (LISW) waveguide technique together with WDM filters . Despite the wide range of techniques, they are sometimes difficult to implement and in some of them, be hard to apply in mass production. Due to the wide availability of techniques, different manufacturers have attempted to enter the field during the last decades, however, none of them have been able to become well established. Nowadays, those components can still be purchased, however, a certain minimum amount is required to make an order. Finally, splitter customization is difficult or expensive to implement, restricting the consumer to specific on-the-shelf products.
Nowadays, 3D printing has been widely used for the production of a variety of optical components. Examples of that are lenses , ultracompact multi-lens systems , drawing polymer [17,18] and silica fibers  from 3D printed preforms, direct drawing of polymer fibers , direct-write assembly of optical waveguides , and in sensor applications, such as: Fabry-Pérot cavities at the tip of fibers, to work as humidity sensors  and long period gratings in 3D printed grooved plates to measure torsion and shear strain , or to be used as pressure sensors .
The popularity of 3D printing is related to its intrinsic advantages, such as, fit-for-purpose, simplicity, flexibility and fast design and production. With this technology, the objects are built layer by layer, allowing to form complex structures, that could be too expensive to create with a mold, or time consuming to mill. The most widely known 3D printing technique is the fused deposition modelling (FDM). However, “vat polymerization” techniques have grown in popularity in recent years due to their better resolutions. Thus, they have been widely used in the fabrication of optical components. Examples of “vat polymerization” technique are the stereolithography (SLA), the digital light processing (DLP) and the two photon polymerization (TPP) . Due the higher resolution of TPP compared to other 3D printing technologies, (i.e. 200 nm, compared to 50 µm for the DLP or 1 µm for the SLA), they have already been proposed for the realization of splitters projected for photonics interconnects and neuronal networks [26,27], intended to operate with single mode behavior at the C-band region, were most of the telecom devices operate. TPP is of course the reference choice for such nanoscale devices, however, the cons about the technology are related to the market price for such printers, which can reach a few hundred thousand dollars. Furthermore, the TPP technology is intended for micro or nanoscale usage and thus, the production of large core splitters needed for instance in applications requiring the use of large core MM-POFs, or also in hard polymer cladd fibers (HPCF), where the waveguide cross section can be larger than 1 mm, while the lengths can reach centimeter scale values, is not the correct choice. For such devices it is more suitable the use of low-cost vat polymerization technologies, such as the ones found for the DLP technology, where the price market is around $150, allowing their usage for a broad audience.
In this work, a low cost DLP 3D printer will be used to fabricate three dimensional splitters intended for large core MM fibers. The work will comprise both numerical and experimental results, regarding symmetrical 1 × 2 and 1 x4 splitter configurations as well as asymmetrical 1 × 2 configuration. The influence of the branching separation angle on the output power will be analyzed, allowing to extrapolate the results that best fit the required needs.
2. Materials and methods
2.1. 3D drawing and theoretical modeling
The splitter elements presented in this paper were designed and optimized using the ray optics module from Comsol Multiphysics software. The ray optics module includes the geometrical optics interface, which does not account for certain optical effects such as diffraction and interference, allowing structures to be modeled with a coarser mesh size, that is suitable to reduce the computational time, namely for large scale structures, such as the ones proposed in this work, that are at the millimeter scale. The arms of the splitter were designed with 0.5 mm radius, to match the standard diameter of the MM-POF cables (i.e. diameters of ∼980 µm). The length of each arm was set to 10 mm. To draw the first arm of the splitter (input arm), a circumference was drawn at the origin of the xy plan, and extruded in the zz direction with 10 mm in length. Next, the same procedure was used for the two output arms, being their origin located on top of the input arm (i.e. origin of the xy plan and at zz = 10 mm). A parametric sweep is created in order to adjust the separation angle of the output arms. The rotation angle is related to the longitudinal axis of the first arm (zz axis). For the 1 × 2 splitter, the two output arms are considered in the same working plan (i.e. yz plan). However, for the 1 × 4 splitter, the two remaining arms were drawn orthogonal to this plane (i.e. drawn at the xz plane), allowing to keep the symmetry of the structure. A 2D geometrical representation of a 1 × 2 splitter at the yz plane is shown in Fig. 1.
In Fig. 1, the angle of rotation for each of the output arms (2nd and 3rd arm), are defined as α and β. When α = β, the splitter is said symmetrical and the output power is equally distributed among the output arms. On the other hand, for asymmetrical splitters, the symmetry needs to be broken, and thus, α ≠ β. In this way, the output power is unequally distributed. The first case (i.e. symmetrical), was explored in this work for both 1 × 2 and 1 × 4 splitters, with angles α = β ranging from 5 to 90°, in steps of 5°. Concerning the asymmetrical configuration, we tested it only for the 1 × 2 splitter, where α was kept constant with a small value to avoid excess loss . The chosen value was α = 30°, being β swept from 0 to 90° in steps of 5°.
A total of 2000 rays were launched at the origin of the first arm with direction parallel to the zz axis. They were uniformly distributed in a conical shape with π/10 angle. The simulation was performed at the wavelength of 627 nm, which is at the low loss window for polymers, i.e. 650 nm. The refractive index of the material that composes the splitter was defined as 1.54, that is close to the value of the UV resin used in this work. The mesh was user controlled being the element size defined as extra fine. Furthermore, the mesh was refined at the boundaries where the rays will be deposited (i.e. where the power will be computed), namely at the end face of each of the output ports. For that, the minimum element size was defined as 0.02 mm while the curvature factor was set to 0.02.
After performing the simulations, each splitter design was exported to a stereolithography file. The assembly of all those files in a single sliced file (i.e. one for each splitter type (1 × 2 and 1 × 4)), was performed in order to 3D print all the splitters at once. The 3D view of the corresponding splitters is shown in Fig. 2(a) and 2(b) for the 1 × 2 and 1 × 4 splitters, respectively.
2.2. 3D printed MM splitters – fabrication
The splitters were fabricated with a low-cost (∼$150 ), consumer grade DLP 3D printer (Photon from Anycubic). This printer has a liquid crystal display (LCD) with resolution of 47 µm in both xy directions and 1 µm resolution in the printing direction (zz direction). It is based on the LCD shallow masking using a 405 nm LED.
The printer resin vat was filled with a low-cost UV photopolymerizable resin (Clear Resin material (3DR3582C, from Anycubic (∼$30 per liter)) . The splitters were printed in layers of 50 µm using an exposure time of 10 seconds per layer. An additional exposure time of 150 seconds is used to print two bottom layers needed to firmly attach the model to the building platform. The splitters were printed vertically as is shown in Fig. 2. For α = β < 80°, the gravity force isn’t an issue (due to the small dimensions), and thus, there is no requirement to add supports. For α = β > 80°, supports would be required. However, we didn’t worry about those splitters since they would induce extremely high attenuations as we will see in the theoretical results. Printing the splitters in directions other than the one mentioned in this work would always require the use of supports which would inevitably decrease the light transmission due to the scattering caused at the contact points. Taking into account that the tallest splitter has ∼2 cm, when α ≈ β ≈ 0°, the required fabrication time was approximately one hour and half.
At the end of the printing process, the splitters attached to the building platform were fully immersed in the same liquid resin and the platform was hanged down for about two minutes, allowing to drain the excess resin by gravity forces. The liquid resin on top of the splitters is then hardened through UV exposure. This fabrication step is crucial since it allows the liquid resin to fill and remove the “stair-like effect” that commonly appears in 3D printing, associated to the layering process and printer resolution (pixel size in the case of DLP 3D printers) [16,23,24], (e.g. see Fig. 5(c) of Ref.  or the inset of Fig. 5(a) in Ref. ). The layer thickness used in our work was set to 50 µm, and the pixel size is also 50 µm. Thus, the surface roughness of the printed objects is of the same size Ideally, the surface roughness should be removed in order to avoid scattering effects at the surface . This was performed by the 1st dip coating process, allowing to fill the voids with the liquid resin, creating a smooth surface. An example of the visual effect before and after the dip coating process may be seen in Fig. 3(a) and 3(b), respectively, for a 1 cm cube, printed with the same parameters used for the splitters.
After the “smoothing” process, the splitters are dip coated in another liquid resin (NOA 85 from Norland Products, Inc.) and hanged down again for two minutes, following the hardening process through UV radiation. This hardening process allows the creation of a thin cladding layer that will allow light confinement between the core (ncore (@589 nm) 1.54) and cladding (ncladd (@589 nm) 1.46). The splitters are finally washed in an isopropyl alcohol bath and cured, using an all-in-one curing and washing machine from Anycubic. The resultant 3D printed splitters may be seen in Fig. 4.
To have a better understanding of all the fabrication steps involved, together with their effect, we decide to take microscope images of the tip of the splitters at different moments. Thus, we show in Fig. 5(a)-(c) and 5(f), the top view and in 5(d) and 5(e), the side view images of the tip of a splitter.
The brittle nature of the hardened photopolymerizable resin that composes the 3D printed splitters, makes it very prone to fracture propagation. Considering the fabricated 3D printed splitters, we used a sharp razor blade (Aeterna 191100, from Svenska Rakbladsfabriken), to make a small indentation perpendicular to the fiber length. Then, the fracture propagates, giving a flat tip surface. The microscope images shown in Fig. 5(a)-(c), 5(e), and 5(f), were prepared through this method.
Figure 5(a) was collected after the 3D printing process and Fig. 5(b) after the dip coating process, done with the same resin used for the 3D printing. The resolution of the printer, given by the size of the pixel (∼50 µm), gives a stair-like contour as is shown on the inset of Fig. 5(a). This roughness gives rise to scattering effects, increasing the insertion losses of the splitter. However, the dip coating process performed after the 3D printing, fills the voids associated to the roughness, producing a smooth surface as seen on the inset of Fig. 5(b). The second dip coating, now performed with NOA85, allows to create a refractive index layer with a lower refractive index (cladding layer), needed for total internal reflection. The image of the tip of the splitter right after the 2nd dip coating is shown in Fig. 5(d), and the one after the cleaving process is shown in Fig. 5(c), 5(e) and 5(f). For the later image, it is possible to observe red light (627 nm) propagating at the inner-most region.
2.3. 3D printed MM splitters – characterization
To estimate the attenuation of the 3D printed structures, we printed a rod with 1 mm (diameter) x 6 cm (length) and did the same dip coating process performed to the splitters. In order to achieve high quality POF end face terminals needed to reduce the coupling losses, to the 3D printed structure, we used our polishing technology developed and published previously .
To estimate the 3D printed fiber loss, we used the cut-back method, in which light is coupled into a given length of fiber and the transmitted power is measured as the fibre is shortened. For that, light from a supercontinuum source (Fianium WL-SC-400) was launched via silica fiber into a 1-meter standard multimode step-index POF (HFBR-RUS100Z, from Broadcom). This fiber is normally used in data links and sensing applications at the visible region, since it provides low loss transmission at around 650 nm, achieving values of 0.2 dB/m. It is composed of a 980 µm diameter PMMA core and a thin 20 µm fluorinated cladding material. The refractive index of the fiber core and cladding at D-line (589 nm) are 1.492 and 1.417, respectively, giving a numerical aperture (NA) of ∼0.47.
The POF terminal is coupled to the 3D printed fiber using a small portion (∼1 cm) of the protective cover of the POF (i.e. polyethylene jacket), in order to act as a fiber optic mating sleeve. The other terminal of the 3D printed fiber is coupled in a similar way to another POF that is connected to an optical spectrum analyzer (OSA), either the Yokogawa AQ6373B for the wavelength range 400–1200 nm as well as the Yokogawa AQ6370 for the 1200–2400 nm. Proper signal normalization was performed, allowing to just get the waveguide loss and get rid of external sources of attenuation, such as the ones from POFs. The fiber was cut-back five times to determine the attenuation. The experimental setup used to estimate the attenuation of the waveguide is shown in Fig. 6(a), while the transmission spectrum is shown in Fig. 6(b).
The cut-back method is best suited for long fiber lengths, allowing to achieve an equilibrium mode distribution. In our case, the fiber length was too short, and the probability to have different cleave end face terminals as we cut the fiber can induce some uncertainty on the measurement. Even though, the spectra obtained for each fiber length had the same signature allowing us to have some confidence about the result. The OSAs trace shown in Fig. 6(b) reveal that the minimum loss is obtained for the visible region, achieving values close to 1.3 dB/cm at 700 nm. This is a high value and possible reasons for that could be related to poor transmissibility of the core polymer UV resin and possible contamination with small debris. The dispersion of light at the layer interfaces is also a possible cause. Furthermore, the coupling losses, associated to imperfection at the fiber end face terminals, fiber to fiber misalignments, NA mismatch, and also due to the Fresnel reflection have also to be taken into account. The defects created during the fiber terminals preparation, such as scratches and bad cleavage angle could also be present. Furthermore, the fiber to fiber alignment is performed with the fiber polyethylene jacket, leaving the possibility to have small misalignments due the soft nature of this material. The two last sources of losses, (i.e. NA mismatch and Fresnel reflection), are related to the refractive index of the materials. The refractive index of the cured resin at 589 nm was 1.5366 (experimentally measured with an Abbemat refractometer from Anton Paar), and the one from the core POF at the same wavelength is 1.4920. Considering normal incidence, this gives a Fresnel reflection loss of 0.08% (considering two splice regions). This value is residual and can be neglected. Regarding the NA mismatch, the 3D printed fiber propagates light between the core material and the cured cladding layer (ncladd (589 nm) 1.46), and thus, its NA is ∼0.48, that is similar to the one of the POF fiber (NA ∼ 0.47). Because of that, this will induce non-relevant losses.
A similar measurement scheme described before to estimate the 3D printed fiber loss was used to characterize the split ratio of the MM splitters, it was used the same measurement scheme described before to estimate the 3D printed fiber loss. Yet, the supercontinuum source and the OSA were replaced by an LED source (LLS-627 from Ocean Optics) with central wavelength of 627 nm and 20 nm bandwidth, and a power meter (PM20A from Thorlabs Inc.), respectively. The signal was normalized in order to avoid the loss contributions of the emitting and receiving fiber and the output power at each of the arms was measured by inserting the splitter between the emitting and receiving POFs.
3. Results and discussion
3.1. Simulation results
The simulation results regarding the symmetric splitters (1 × 2 and 1 × 4) revealed negligible excess loss when α = β ≤ 30°. Yet, for angles above 30°, the excess loss starts to appear. For the asymmetric 1 × 2 splitter, one of the output arms is kept fixed (α = 30°) and 0 < β <90 and the results indicate that the excess loss become to take effect only when α >30°. The ray trajectories observed for the MM splitters when they behave with negligible excess loss, may be seen in Fig. 7(a) and 7(c) for the symmetric 1 × 2 and 1 × 4 splitters (α = β = 20°), respectively and in Fig. 7(b) for the asymmetric 1 × 2 splitter, when α = 30° and β =20°.
As is shown in Fig. 7(a)–7(c), the ray trajectories along the splitters are unaffected by the angle (α = β = 20°), allowing the rays to propagate along the splitter without refraction, (see the inset of Fig. 7(a)). However, as the arm separation angle is increased (namely for angles ≥ 30°), as is observed in Fig. 7(d)–7(f), some of the rays are refracted and become lost, giving rise to excess loss. The numerical quantification of the excess loss as function of the angle is later reproduced in Fig. 11, for comparison purposes with the experimental results.
3.2. Microscope visual inspection of the 3D printed MM splitters
A visual inspection of the printed MM-splitters was made by taking photos through a microscope (Leica DM750 M). A set of images taken for the symmetric MM-splitter may be seen in Fig. 8(a)-8(i).
The pictures presented in Fig. 8(a)-8(i), show that the angle between the output arms reproduces quite well the expected value. The surface seems to be smooth thanks to the dip coating process done after the 3D printing. One important feature to consider from the microscope images is the “stair-like effect” related to the 3D printing process. This is inherent to the fabrication process, since for each layer, the region close to the LCD receives more power than the one at the back, due to the UV attenuation, inducing a gradient refractive index. Furthermore, the polymerization at each printing layer, induces material shrinkage on the top most region of the previous printed layer, changing its refractive index properties due to the strain optic effect. Both effects are summarized in Fig. 9, for better clarification.
The “stair-like effect” observed in Fig. 8, and explained in Fig. 9, will influence the splitters light guiding conditions, and, potentially induce losses. One possible way to reduce this effect could be done by reducing the layer thickness, minimizing the effects of the UV attenuation across the layer, at a cost of increasing fabrication time. Furthermore, the exposure time should be just the minimum required for polymerization, to minimize the material shrinkage effects. Yet, these approaches need to be investigated in a future work.
3.3. Light guidance performance of the MM splitters
After performing the qualitative analysis of the MM splitters, their light guidance performance was visualized by naked eye. This was done by injecting red light from the 627 nm LED source (LLS-627 nm) in a MM POF, being its output connected to each of the ports of the 3D printed splitter. The results of this inspection for the symmetric (α = β = 15°) splitters, may be seen in Visualization 1. From those observations, we can clearly see that the splitters transmit light in both directions. The propagation of light in these splitters was also observed for two other wavelength sources, namely for LEDs operating at 532 nm (green) and 445 nm (blue), and the pictures of them when light is injected at the In/Out port (1st arm), may be seen in Fig. 10(a) and 10(c) for the 1 × 2 splitter and in Fig. 10(b) and 10(d) for the 1 × 4 splitter, either for the side captures Fig. 10(a) and 10(b), and for the far-field observations in a white screen board, as shown in Fig. 10(c) and 10(d).
The results concerning the far-field captures observed in Fig. 10(c) and 10(d), show that the light field is well distributed for the 1 × 2 splitters while the 1 × 4 splitters showed a higher degree of scattering around the bright illumination spots.
The results concerning the output power measured in each output arm for the α and β angles studied in this work after signal normalization, are plotted together with the numerical results in Fig. 11(a) and 11(b) for the symmetric splitters (1 x2 (a) and 1 x4 (b)), and in Fig. 11(c) for the asymmetrical splitter (α = 30°, 0° ≤ β ≤90°).
Regarding the experimental data shown for the 1 × 2 symmetric splitter (Fig. 11(a)), it is possible to notice that the output power is similarly distributed, having just small discrepancies between each other (i.e. splitting ratio of 45:47%, 42:43%, 27:28%, etc., for α = β = 5°, 10° and 15°, respectively). The same also occurs for the symmetric 1 × 4 splitter seen on Fig. 11(b), meaning that the power is similarly distributed. Yet, one can easily see that in both symmetric splitters (either 1 × 2 and 1 × 4), the excess loss becomes pronounced for angles α = β > 10°, i.e. when α = β = 5°, 10°, 15° and 20°, the excess loss reaches values of 9%, 15%, 46% and 63%, for the 1 × 2 splitter, and 7%, 10%, 26% and 35% for the 1 × 4 splitter, respectively. These findings are in contradiction with the numerical results shown by the dark line in Fig. 11(a) and 11(b), where it could be seen that the excess loss should only appear for angles α, β > 30°. Yet, for α, β = 30°, experimental results show that almost 80% of the input signal is lost. One possible reason for such discrepancies between experimental and theoretical results may arise from the rays’ incidence angle at each printing layer. When the 2nd and 3rd arm increase their separation angles, the light incidence angle at each of the layers that compose the multilayer structure (see Fig. 12(a)), will change, and this enables refraction to occur at each layer interface (see Fig. 12(b)), and thus, more probability to refraction at the core cladding interface, reducing the power transmitted in the splitter. This multilayer structure (see the description found at the end of the previous section), has not been taken into account in the simulations, because the layer refractive index gradient is unknown. Thus, the discrepancies found between experimental and simulation results could be related with this.
Despite the excessive excess loss for larger splitting angles, we could still observe acceptable performance for small splitting angles (α, β ≤ 20°).
Regarding the experimental results collected for the asymmetric 1 × 2 splitter, seen in Fig. 11(c), where the 1st arm is kept fixed (i.e. α = 30°), while the 2nd arm (β) is swept from 0° up to 90°, it is possible to see that the splitting ratio for α = 30° and β = 0°, performs as theoretical predicted (see black lines), reaching splitting ratio values of 8:85%, which were similar to the theoretical values of 12:88%. Yet, the excess loss was also present with a value of ∼7%. Additionally, when β increases, the power measured in the first arm (i.e. α = 30°) is constant, reaching splitting ratios around 8 to 9% for 0 ≤ β ≤ 60°, which is contradictory to the theoretical results. Furthermore, the drop of power for the second arm is very abrupt when compared to the simulated one. While it is possible to reach different splitting ratios when β > 0°, the excess loss becomes very noticeable. Examples of excess loss values are 26%, 49% and 59% for β = 10°, 20°, 30°, respectively. Preferably, the results taken for the asymmetric splitter should keep a fixed β < 10°, allowing to reach lower excess loss as verified for the experimental results performed for the symmetric splitters. To finalize, it is worth to mention that the loss measurements could still present some degree of uncertainty due to the mode distribution as well as the quality of the splitter terminals.
In the overall we have shown that simple and cost effective solutions based on 3D printing technology could be used to produce MM-splitters with good similarities in the splitting ratio and with acceptable excess loss when the angle between the output arms is less than 10°. The excess loss discrepancies between the numerical and experimental results are attributed to the “stair-like effect” associated to the 3D printing layering process, which can be controlled by adjusting the layer thickness and UV exposure time. Furthermore, the material losses play also an important role in this type of devices. Thus transparent resins with more suited optical properties such as the ones from Ormocer GmbH, Norland Products Inc., Henkel Corp., etc., could benefit the results shown in this work.
This work showed that the easiness of use and cost effectiveness of 3D printing technologies could be an opportunity to fabricate customized splitters commonly used in automobile industry, fiber to the home communications, sensors applications, lightning, etc. The technology used in this work provides a faster and reliable solution to produce customized passive optical components, such as fiber splitters, and still maintains the fabrication simplicity. On top of that the ability to customize the MM-splitters for a specific application, make this methodology very competitive over standardized technologies. Furthermore, the characteristics presented by the 3D printing technology are well suited for mass production, allowing economic valorization. that the new technique presented in this work pave the way for the fabrication of large core MM fiber devices based on light based 3D printed technology.
European Regional Development Fund; Fundação para a Ciência e a Tecnologia (AQUATICsens (POCI-01-0145-FEDER-032057), FOPEComSens (PTDC/EEI-TEL/1511/20), UIDB/50008/2020-UIDP/50008/2020).
This work was funded by FCT-Fundação para a Ciência e Tecnologia and by the European Regional Development Fund (FEDER), through the Competitiveness and Internationalization Operational Programme (COMPETE 2020) of the Portugal 2020 framework and by national funds under the projects AQUATICsens (POCI-01-0145-FEDER-032057), FOPEComSens (PTDC/EEI-TEL/1511/20) and UIDB/50008/2020-UIDP/50008/2020.
The authors declare no conflicts of interest.
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
1. R. Oliveira, F. Sequeira, L. Bilro, and R. Nogueira, “Polymer Optical Fiber Sensors and Devices,” in Handbook of Optical Fibers, G.-D. Peng, ed., 1st ed. (SpringerSingapore, 2019), pp. 1957–1996.
2. E. N. Udofia and W. Zhou, “3D printed optics with a soft and stretchable optical material,” Addit. Manuf. 31, 100912 (2020). [CrossRef]
3. T. Kibler, S. Poferl, G. Böck, H.-P. Huber, and E. Zeeb, “Optical data buses for automotive applications,” J. Lightwave Technol. 22(9), 2184–2199 (2004). [CrossRef]
4. A. Grzemba, “Physical Layer,” in MOST – The Automotive Multimedia Network (Franzis Verlag, 2012), pp. 121–147.
5. T. Almeida, R. Oliveira, A. M. Rocha, L. Bilro, and R. Nogueira, “WDM Transmission over Graded-Index Plastic Optical Fiber,” in 26th International Conference on Plastic Optical Fibres (2017), p. Paper 89.
6. F. M. A. Al-Zubaidi, D. S. Montero, and C. Vazquez, “SI-POF Supporting Power-Over-Fiber in Multi-Gbit/s Transmission for In-Home Networks,” J. Lightwave Technol. 39(1), 112–121 (2021). [CrossRef]
7. R. Gaudino, D. Cardenas, M. Bellec, B. Charbonnier, N. Evanno, P. Guignard, S. Meyer, A. Pizzinat, I. Mollers, and D. Jager, “Perspective in next-generation home networks: Toward optical solutions?” IEEE Commun. Mag. 48(2), 39–47 (2010). [CrossRef]
8. “Die Mount GmbH,” https://www.diemount.com.
9. M. Rezem, A. Gunther, B. Roth, E. Reithmeier, and M. Rahlves, “Low-cost fabrication of all-polymer components for integrated photonics,” J. Lightwave Technol. 35(2), 299–308 (2017). [CrossRef]
10. H. Mizuno, O. Sugihara, S. Jordan, N. Okamoto, M. Ohama, and T. Kaino, “Replicated Polymeric Optical Waveguide Devices With Large Core Connectable to Plastic Optical Fiber Using Thermo-Plastic and Thermo-Curable Resins,” J. Lightwave Technol. 24(2), 919–926 (2006). [CrossRef]
11. H.-J. Park, K.-S. L. Lim, and H. S. Kang, “Low-cost 1×2 plastic optical beam splitter using a V-type angle polymer waveguide for the automotive network,” Opt. Eng. 50(7), 075002 (2011). [CrossRef]
12. W. R. Cox, T. Chen, and D. J. Hayes, “Micro-Optics Fabrication by Ink-Jet Printers,” Opt. Photonics News 12(6), 32–35 (2001). [CrossRef]
13. S. Li, Q. Lin, L. Chen, and X. Wu, “Design and fabrication of polymeric multimode power splitter with secondary asymmetric branches,” Chin. Opt. Lett. 9(8), 81302–81305 (2011). [CrossRef]
14. T. Yamashita, A. Kawasaki, M. Kagami, T. Yasuda, and H. Goto, “Polymeric multi/demultiplexers using light-induced self-written waveguides for cost-effective optical interconnection,” in 2010 IEEE CPMT Symposium Japan (IEEE, 2010), (11744735).
15. G. D. Berglund and T. S. Tkaczyk, “Fabrication of optical components using a consumer-grade lithographic printer,” Opt. Express 27(21), 30405–30420 (2019). [CrossRef]
16. T. Gissibl, S. Thiele, A. Herkommer, and H. Giessen, “Two-photon direct laser writing of ultracompact multi-lens objectives,” Nat. Photonics 10(8), 554–560 (2016). [CrossRef]
17. M. Ghebrebrhan, G. Z. J. Loke, and Y. Fink, “Fabrication and measurement of 3D printed retroreflective fibers,” Opt. Mater. Express 9(8), 3432–3438 (2019). [CrossRef]
18. K. Cook, J. Canning, S. Leon-Saval, Z. Reid, M. A. Hossain, J.-E. Comatti, Y. Luo, and G.-D. Peng, “Air-structured optical fiber drawn from a 3D-printed preform,” Opt. Lett. 40(17), 3966–3969 (2015). [CrossRef]
19. Y. Chu, X. Fu, Y. Luo, J. Canning, Y. Tian, K. Cook, J. Zhang, and G.-D. Peng, “Silica optical fiber drawn from 3D printed preforms,” Opt. Lett. 44(21), 5358–5361 (2019). [CrossRef]
20. J. Canning, M. A. Hossain, C. Han, L. Chartier, K. Cook, and T. Athanaze, “Drawing optical fibers from three-dimensional printers,” Opt. Lett. 41(23), 5551–5554 (2016). [CrossRef]
21. D. J. Lorang, D. Tanaka, C. M. Spadaccini, K. A. Rose, N. J. Cherepy, and J. A. Lewis, “Photocurable liquid core-fugitive shell printing of optical waveguides,” Adv. Mater. 23(43), 5055–5058 (2011). [CrossRef]
22. M. Chen, Y. Zhao, H. Wei, C. Zhu, and S. Krishnaswamy, “3D printed castle style Fabry-Perot microcavity on optical fiber tip as a highly sensitive humidity sensor,” Sensors and Actuators B: Chemical 328, 128981 (2021). [CrossRef]
23. R. Oliveira, R. Nogueira, and L. Bilro, “3D printed long period gratings and their applications as high sensitivity shear-strain and torsion sensors,” Opt. Express 29(12), 17795–17814 (2021). [CrossRef]
24. R. Oliveira, L. M. Sousa, A. M. Rocha, R. Nogueira, and L. Bilro, “UV Inscription and Pressure Induced Long-Period Gratings through 3D Printed Amplitude Masks,” Sensors 21(6), 1977 (2021). [CrossRef]
25. T. J. Wallin, J. Pikul, and R. F. Shepherd, “3D printing of soft robotic systems,” Nat. Rev. Mater. 3(6), 84–100 (2018). [CrossRef]
26. J. Moughames, X. Porte, M. Thiel, G. Ulliac, L. Larger, M. Jacquot, M. Kadic, and D. Brunner, “Three-dimensional waveguide interconnects for scalable integration of photonic neural networks,” Optica 7(6), 640–646 (2020). [CrossRef]
27. J. Moughames, X. Porte, L. Larger, M. Jacquot, M. Kadic, and D. Brunner, “3D printed multimode-splitters for photonic interconnects,” Opt. Mater. Express 10(11), 2952–2960 (2020). [CrossRef]
28. J. H. Kuang, P. C. Chen, and Y. C. Chen, “Optical performance of symmetrical and asymmetrical Y-branch couplers for plastic optical fibers,” Appl. Opt. 51(19), 4448–4456 (2012). [CrossRef]
29. “Anycubic 3D printing,” https://www.anycubic.com.
30. M. S. Ab-Rahman, F. S. Ater, and R. Mohamed, “Effects of random sidewall roughness on optical power splitter,” Opt. Eng. 54(5), 055103 (2015). [CrossRef]
31. R. Oliveira, L. Bilro, and R. Nogueira, “Smooth end face termination of microstructured, graded-index, and step-index polymer optical fibers,” Appl. Opt. 54(18), 5629–5633 (2015). [CrossRef]