Open Access
Add to BibSonomy
Abstract
The interlayer distance optimized for low-loss and low-crosstalk double-layer polymer optical waveguides was investigated to enhance their transmission performance. Simulations were conducted to determine the minimal interlayer distances for double-layer optical waveguides with different core sizes. An optimal interlayer distance of 24 µm was identified for a 20 µm × 20 µm double-layer waveguide, which ensured interlayer crosstalk below -30 dB when roughness remained under 80 nm. The double-layer waveguides were fabricated employing ultraviolet lithography combined with the overlay alignment method. Based on experimental optimization, the important fabrication parameters were optimized, such as a plasma treatment time of 10 s, a core exposure dose of 500 mJ/cm2, and a cladding exposure dose of 240 mJ/cm2. Additionally, the fabricated double-layer waveguides, with an interlayer distance of 24.5 µm, exhibited low transmission losses of less than 0.25 dB/cm at 850 nm and 0.40 dB/cm at 1310 nm, respectively. The low interlayer crosstalk values were less than -52 dB at 850 nm and -60 dB at 1310 nm, respectively. The agreement between the experimental results and the simulation findings indicates that this method offers a promising approach for fabricating double-layer waveguides with good performances.
© 2023 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
With the growing demand for data traffic in large-scale data centers and high-performance computers, short-distance optical interconnects have garnered significant interest due to their benefits in transmission data rate, bandwidth, interconnect density, power consumption, and electromagnetic immunity [1–3]. Polymer waveguides have emerged as a promising transmission medium for board-level optical interconnects thanks to their low cost, low optical loss, simple fabrication technology, and compatibility with printed circuit boards (PCBs) and optical fibers [4,5]. Recently, single-layer multi-mode polymer optical waveguides have been proposed and demonstrated exceptional transmission performance at 850 and 1310 nm [6–10]. However, compared with single-layer multi-mode polymer optical waveguides, multi-layer multi-mode polymer optical waveguides can break the geometric space limitation and achieve denser photon integration in the vertical dimension. In optical printed circuit boards (OPCBs), high optical link capacity and optical channel density can be achieved by employing vertically aligned multi-layer waveguides that avoid interlayer coupling [11–13]. Vertical alignment is crucial for precisely aligning multi-layer waveguides with vertical cavity surface emitting laser (VCSEL) arrays, photodiode (PD) arrays [11,12], and mechanical transport (MT) connectors [13]. Therefore, multi-layer multi-mode polymer optical waveguides are expected to meet the demand for high-speed and large-capacity data processing.
Multi-layer polymer optical waveguides have been utilized to generate various three-dimensional (3D) devices, such as 3D polymer directional couplers [14], two-dimensional optical phased arrays [15], fan-in/out devices [16], 3D wavelength demultiplexers [17], and mode multiplexers [18–22]. Among these devices, 3D polymer directional couplers [14], two-dimensional optical phased arrays [15], and mode multiplexers [18–22] operate based on interlayer coupling. Several technologies have been employed for fabrication, such as hot-embossing technology [23], the Mosquito method [14,16], and photolithography [15,17–22,24]. The hot-embossing technique offers advantages such as high precision and batch preparation, but it requires high temperature and pressure, which has limited ability to prepare complex structures. Similarly, the Mosquito method is characterized by its high precision and diverse patterns, but it requires precise control techniques. Compared to the previous two techniques, photolithography offers high resolution and large-area preparation, which makes it possible to achieve complex structures. It can also be combined with the overlay alignment method to produce high-precision multi-layer waveguides and 3D devices. At present, multi-layer waveguides have shown remarkable accomplishments. Multi-layer polymer optical waveguides, exhibiting a transmission loss below 1 dB/cm and inter-waveguide crosstalk below -40 dB at 1550 nm, have been demonstrated using photolithography combined with the overlay alignment method [25]. For multi-layer waveguide structures, the interlayer distance not only dictates the interconnect density but also significantly impacts the transmission characteristics of the multi-layer waveguides, such as loss and crosstalk. However, despite its importance, the existing literature lacks comprehensive studies on the impact of roughness and interlayer distance on waveguide loss and crosstalk. The interlayer distance in previous studies has been expected as 17.32 µm without detailed analysis [25]. While simulation studies on double-layer multi-mode polymer optical waveguides based on two hot-embossing processes have been conducted to prevent coupling between layers [23], these studies do not account for the scattering loss caused by the roughness generated during fabrication [26]. Therefore, comprehensive studies are needed to examine the impact of interlayer distance and roughness on loss and crosstalk for multi-layer waveguides.
In this work, we explored the optimization of interlayer distance in double-layer polymer optical waveguides, aiming to achieve low-loss and low-crosstalk performance. First, theoretical simulations were conducted to examine the impact of interlayer distance and roughness on transmission loss and crosstalk for double-layer waveguides of various sizes. Then, the optimal interlayer distance for a 20 µm × 20 µm double-layer waveguide was determined. Subsequently, the double-layer waveguide fabrication process using ultraviolet lithography combined with the overlay alignment method process was developed. The plasma treatment time and the core/cladding exposure dose during waveguide preparation were ascertained by analyzing the roughness and the insertion loss, respectively. Finally, double-layer waveguides with excellent performance were achieved. The approach for determining the optimal interlayer distance was studied and experimentally validated by measuring losses and crosstalk, which showed consistency with the simulation results. This work provides a valuable reference for the fabrication of high-performance double-layer waveguides.
2. Design and simulation
A 3D schematic diagram of a proposed double-layer polymer optical waveguide built on an FR-4 substrate is illustrated in Fig. 1. The waveguide consists of two square cores, denoted as upper-layer and lower-layer, which are aligned centrally along the z-axis and surrounded by cladding material. The core and cladding monomers used are EpoCore_20 and EpoClad_20 (Micro Resist Technology, Germany), respectively. According to the manufacturer's specifications, the refractive indices of the core/cladding are 1.5834/1.5708 at 850 nm and 1.5766/1.5644 at 1310 nm. The waveguide core size is w × h (as shown in the inset of Fig. 1). The interlayer distance d is defined as the height between the bottom of the upper-layer waveguide core and the top of the lower-layer waveguide core.
Fig. 1. 3D schematic diagram of the double-layer polymer optical waveguide (Inset shows the waveguide structure cross-section).
In order to evaluate the performance of the double-layer polymer optical waveguide, numerical simulations are performed using the 3D finite-difference beam propagation method (3D FD-BPM) (Beam-PROP, RSoft). In the simulation, light from a 9-µm core single-mode fiber (SMF) is coupled into a 2-cm long lower-layer waveguide. The transmission characteristics of the double-layer waveguides have been calculated. The insertion loss (IL) represents the total loss occurring in the waveguide, including coupling and transmission losses, which can be expressed as:
where Pin denotes the optical power coupled into the waveguide, and Pout denotes the output optical power of the waveguide. The transmission loss (TL) is defined as the loss of power during the transmission of the optical signal in the waveguide, which can be calculated as: where L denotes the length of the waveguide. On the other hand, the crosstalk (CT) is defined as the amount of light that is coupled from one waveguide channel to another, which is given by: where $P_{\textrm{out}}^\prime$ represents the output signal power of another waveguide channel.The transmission loss TL dependency on the interlayer distance d and the roughness σ for double-layer waveguides with different sizes at the wavelength of 850 nm is shown in Fig. 2. It is found that the TL increases with decreasing d value and increasing σ value. The TL is exceptionally high when d is extremely low (e.g., d = 2 µm), which is due to the distance between the upper-layer waveguide and the lower-layer waveguide being too close, leading to evanescent wave coupling. On the other hand, the insertion loss is less affected by d when d is sufficiently large with the same σ. For instance, with a σ of 0 nm and d larger than 6 µm, the TL of a 10 µm × 10 µm waveguide is less than 0.08 dB/cm and fluctuates less. However, the TL increases significantly when σ increases for a constant d. This is because the increased interaction between the modes and the roughness could cause significant scattering, resulting in transmission loss [27,28]. Moreover, the TL decreases progressively with increasing the waveguide core size. For example, when the σ is 60 nm and the d is 30 µm, the TL of a 20 µm × 20 µm and 50 µm × 50 µm waveguide is 0.11 dB and 0.05 dB, respectively. This dependence can be interpreted as a loss depending on the modal power overlap at the waveguide core and cladding interface at the same operating wavelength. Consequently, narrower waveguides with larger modal power overlap ratios experience higher light scattering losses [29].
Fig. 2. Calculated dependences of the transmission loss on parameters d and σ at 850 nm for the waveguide core size of (a) 5 µm × 5 µm, (b) 10 µm × 10 µm, (c) 20 µm × 20 µm, (d) 30 µm × 30 µm, (e) 40 µm × 40 µm, and (f) 50 µm × 50 µm.
Figure 3 depicts the variations in the crosstalk CT for the double-layer waveguides with different core sizes at various interlayer distances d and roughness σ. The results indicate that the CT increases with increasing σ and decreasing d. As σ increases, waveguides with high scattering loss [28] can scatter more light into the cladding, consequently increasing the CT of another channel. As d decreases, evanescent wave coupling occurs between the upper-layer and lower-layer waveguide cores, which also increases the CT. Furthermore, the CT is relatively low for waveguides with smaller sizes at the same σ when d is sufficiently large. For instance, the CT for 10 µm × 10 µm and 20 µm × 20 µm double-layer waveguides is -35 dB and -32 dB, respectively, with the same σ of 0 nm and d of 24 µm. This can be ascribed to the fact that the mode field distribution in smaller waveguides is relatively concentrated, with energy primarily focused within the waveguide. Consequently, the energy exchange between adjacent waveguides is weak [30]. In general, crosstalk less than -30 dB is considered sufficiently low for board-level optical interconnections [6,31]. As shown in Table 1, the minimal d values for double-layer waveguides with various core sizes and different σ values are listed when the interlayer crosstalk is less than -30 dB.
Fig. 3. Calculated dependence of the interlayer crosstalk on parameters d at 850 nm for the waveguide core size of 10 µm × 10 µm, 20 µm × 20 µm, 30 µm × 30 µm.
Table 1. Simulated minimal interlayer distance d of the double-layer waveguides
The more detailed relationships between the waveguide core sizes, the roughness, and the minimal interlayer distance are shown in Fig. 4. It can be observed that the minimal interlayer distance d increases with increasing the roughness σ, which is attributed to the scattering losses in the waveguide [27]. The increased roughness leads to more scattering losses, resulting in a larger minimal interlayer distance. Additionally, the minimal interlayer distance d fluctuates with increasing the waveguide core sizes for the same roughness. This could be explained by the dual effect caused by the different mode energy distributions of the waveguides with different sizes induced by the roughness, including mode coupling between guided modes and the continuous spectrum of radiation modes and between the guided modes [32–35].
Fig. 4. Calculated dependence of the minimal interlayer distance d on parameters w and σ at 850 nm for the double-layer waveguides.
A double-layer waveguide with a core size of 20 µm × 20 µm is considered a specific case. The calculated dependence of the transmission loss and the interlayer crosstalk on the parameter roughness σ is shown in Fig. 5. Based on previous research, the roughness of the polymer optical waveguide lies within the order of several tens of nanometers [29,36,37]. The optimal interlayer distance of 24 µm is chosen, which is suitable for roughness less than 80 nm when the interlayer crosstalk is less than -30 dB. As shown in the figure, when σ increases, the transmission loss increases at 850 nm and 1310 nm, but it is higher at 850 nm than at 1310 nm. This is because that waveguides with shorter wavelengths are more sensitive to roughness and have larger scattering losses [38]. Furthermore, similar to the transmission loss, the crosstalk also exhibits the same variation pattern. The crosstalk at 850 nm is higher than that at 1310 nm at the same σ, which is due to the fact that the waveguides at 850 nm with higher scattering loss have more light scattered into the cladding.
Fig. 5. Calculated dependence of the transmission loss and the interlayer crosstalk on parameters σ for the double-layer waveguides at 850 nm and 1310 nm.
3. Fabrication
The double-layer polymer optical waveguides were fabricated using ultraviolet (UV) lithography combined with the overlay alignment method. EpoClad_20 and EpoCore_20, a pair of negative optical polymer photoresists, were used for the waveguides with a core size of 20 µm × 20 µm. The complete fabrication process is illustrated in Fig. 6. (a) the metal alignment markers were generated on an FR-4 substrate using magnetron sputtering, (b) the lower-layer cladding, with a thickness of 20 µm, was spin-coated on the substrate, (c) the 12-channel lower-layer waveguide core, with a pitch of 250 µm, was generated via the UV lithography combined with the overlay alignment method, using a maskless lithography machine (MLA100, Heidelberg), (d) the center-cladding was spin-coated to control the d value of 24 µm, (e) the upper-layer waveguide core was prepared by the UV lithography combined with the overlay alignment method, (f) the upper-layer cladding, with a thickness of 50 µm, was spin-coated.
Fig. 6. Fabrication process of the double-layer polymer optical waveguide. (a) the metal alignment markers fabrication, (b) the lower-layer cladding fabrication, (c) the lower-layer waveguide core fabrication, (d) the center-layer cladding fabrication, (e) the upper-layer waveguide core fabrication, (f) the upper-layer cladding fabrication.
Plasma treatment is an essential process for some polymers, which helps to increase wettability and improve adhesion [39]. This process is carried out before spin-coating either the core or the cladding. The 3D images of the top surface and sidewall after air plasma (GD-30, Plasma Cleaner) treatments at different times with a power of 500 W are observed using atomic force microscopy (AFM) in tapping mode (AIST-NT, SmartSPM 1010). The top surface and sidewall roughness without plasma treatment is 1.31 nm and 25.69 nm, respectively, as displayed in Fig. 7(a) and (c). After a 10 s plasma treatment, the roughness increases to 4.82 nm and 33.22 nm, respectively, as shown in Fig. 7(b) and (d). The roughness of several regions after plasma treatment with 0 s, 10 s, 20 s, 30 s, 40 s, and 50 s was measured by AFM, as shown in Fig. 8. The results show that the average top surface roughness and the average sidewall roughness increase with the increase of plasma treatment times. The dependency can be explained by the selective reaction of oxygen plasma with the polymer during the plasma treatment [40,41]. Upon calculation, the roughness increases approximately 23 times and 4 times, respectively, after 50 s plasma treatment. The main reason for this difference is that the ability of oxygen plasma to bombard the sample from top to bottom is much stronger than that of side bombardment. The transmission loss of the waveguide increase with an increasing plasma treatment time, which is illustrated in Fig. 8. Specifically, for a plasma treatment time of 10 s, the measured transmission loss is 0.24 dB/cm, which is consistent with previous reports [42].
Fig. 7. AFM images of the samples: The top surface of the waveguide core (5 µm × 5 µm) with plasma treatment times of (a) 0 s, (b) 10 s; The sidewall of the waveguide core (1.5 µm × 1.5 µm) with plasma treatment times of (c) 0 s, (d) 10 s.
Fig. 8. Measured top surface, sidewall roughness, and transmission loss as a function of the plasma treatment time.
Exposure is a critical process in which the exposure dose determines the performance of the waveguide. When the exposure dose is low, the core/cladding is underexposed and cannot create enough photo-acid to cross-link. Conversely, when the core/cladding is overexposed, a high exposure dose generates excessive photo-acid, leading to cross-linking that should not occur [43]. The exposure dose significantly impacts the photo-acid concentration, subsequently affecting the cross-link density of the polymer, the developed resist profile, the line width, and the refractive index [44]. Therefore, selecting the appropriate exposure dose for the core and the cladding is crucial.
The variation in the waveguide core width and the average insertion loss at 850 nm for a 1-cm waveguide with increasing the core exposure dose is illustrated in Fig. 9(a) when the cladding exposure dose is set to 400 mJ/cm2. As the core exposure dose increases, the waveguide core width gradually increases. When the core exposure dose is less than 500 mJ/cm2, the core is underexposed and cannot create enough photo-acid to cross-link the Epocore_20 molecules. This can cause the developer to diffuse into the underexposed boundary, resulting in a narrower width. Conversely, a high exposure dose could create excessive photo-acid, leading to cross-linking that should not occur, ultimately resulting in a wider width [43]. However, the average insertion loss at 850 nm first decreases and then increases with increasing the core exposure dose. This may be because the sidewall roughness first decreases and then increases with the core exposure dose, then drives the insertion loss fluctuation [45]. It is important to note that the width is close to the designed width of 20 µm, and the insertion loss is relatively low at 0.76 dB when the core exposure dose is set to 500 mJ/cm2. Therefore, the core exposure dose of 500 mJ/cm2 is chosen.
Fig. 9. (a) Variation in the average insertion loss and the waveguide core width with core exposure dose at 850 nm, (b) Variation of the average insertion loss with cladding exposure dose at 850 nm.
In Fig. 9(b), the average insertion losses initially decrease and then increase at 850 nm for a 1-cm waveguide with increasing the cladding exposure dose when the core exposure dose is 500 mJ/cm2. When the cladding exposure dose is low, the cladding is underexposed and unable to create enough photo-acid to fully cross-link the Epoclad_20 molecules, resulting in a soft cladding and increased insertion loss [43]. On the contrary, a high exposure dose can increase the refractive index of the cladding [44], decreasing the refractive index difference between the core and the cladding, then increasing insertion loss induced by roughness. Therefore, the ideal cladding exposure dose is determined to be 240 mJ/cm2, resulting in a low average insertion loss of 0.38 dB.
Based on the experimental results above, the main experimental parameters are summarized in Table 2.
Table 2. Main experimental parameters
4. Measurement and analysis
Figure 10(a) exhibits the through-light diagram of the double-layer polymer waveguide end surfaces measured by a 3D optical profiler (S neox, Sensofar-Tech), where the lateral misalignment between the two waveguide core layers is expected to be less than 2 µm. The parameters w × h of the upper-layer and lower-layer waveguide cores are 23.5 µm × 23.8 µm and 20.7 µm × 23.6 µm, respectively. And the interlayer distance d is 24.5 µm. Figure 10(b) shows the surface and sidewall profiles of the waveguide cores obtained using scanning electron microscopy (SEM) (JSM-7500F, JEOL). It is evident that both the waveguide surface and sidewall are well-shaped.
Fig. 10. (a) Through-light diagram of the double-layer polymer waveguide end surfaces, (b) SEM image of the waveguide cores.
The double-layer polymer optical waveguide, with a length of 4.2 cm, was fabricated on an FR-4 substrate. To measure the loss and crosstalk, an optical signal from a light source (S4FC852, Thorlabs) was coupled into an SMF (SMF28-e, Corning), followed by either the upper-layer or the lower-layer waveguide core. Subsequently, a multi-mode fiber (MMF) with a core diameter of 50 µm connected to a power meter was used to receive the power. The matching oil was dropped on both sides of the waveguide to reduce coupling loss. A six-dimensional adjustment frame (M-562F-XYZ, Newport) was applied to achieve precise alignment between the SMF/MMF and the waveguide.
The insertion loss of the upper-layer and lower-layer waveguides at 850 nm and 1310 nm are shown in Fig. 11(a). It can be observed that the insertion loss for both the upper-layer and lower-layer waveguides exhibits relatively minor fluctuations. The lowest insertion loss is 1.00 dB at 850 nm and 2.14 dB at 1310 nm, respectively. The average insertion loss for all 12 channels is 1.57 dB at 850 nm and 2.42 dB at 1310 nm, respectively. It can be clearly observed that the insertion loss at 850 nm is lower than those at 1310 nm, which is attributed to the different material absorption in the near-infrared band, according to the datasheet from Micro Resist Technology. The insertion loss and the standard deviations are summarized in Table 3.
Fig. 11. (a) 12-channel insertion losses, (b) transmission losses determined by the cut-back method, (c) inter-channel crosstalk (the inset shows the double-layer waveguide end surfaces, with red and blue arrows indicating the direction of the MMF displacement aligned with the lower-layer and upper-layer waveguides, respectively), (d) interlayer crosstalk diagram (inset displays the double-layer waveguide end surfaces, red and blue arrows indicate the direction of upward and downward displacement, respectively) at 850 nm and 1310 nm.
Table 3. Main measurement results of the loss and the crosstalk
Further, the waveguides were characterized by the cut-back method. In practice, the 4.2 cm long waveguide is divided into three sections: 1.05 cm, 1.95 cm, and 2.85 cm, respectively. As indicated in Fig. 11(b), the losses for different waveguide lengths are represented by dots. The solid trend lines (with fitting equations) exhibit the transmission losses (slopes) in units of dB/cm and the coupling loss (intercepts) in units of dB. As listed in Table 3, the transmission losses for the upper-layer waveguides are 0.22 dB/cm at 850 nm and 0.38 dB/cm at 1310 nm, respectively. For the lower-layer waveguides, the transmission losses are 0.25 dB/cm at 850 nm and 0.40 dB/cm at 1310 nm, respectively. The transmission loss is at the same level as previously reported for single-layer waveguides fabricated with the same polymer at the same wavelength [7,42,46,47].
The inter-channel crosstalk between the adjacent channels on the same layer is depicted in Fig. 11(c). The 6-channel upper-layer and lower-layer waveguides are individually aligned with the SMF at the input. The MMF is then scanned horizontally with a step of 50 µm to detect the output power from the other 11 channels. The results indicate that the crosstalk is less than -30 dB at 850 nm and 1310 nm when the MMF is shifted 50 µm toward the 5-channel/7-channel, respectively. Additionally, it can be observed that the crosstalk is less than -36 dB at 850 nm and -40 dB at 1310 nm when the MMF is aligned with the 5-channel/7-channel waveguides, respectively. The waveguides at 850 nm support more high-order modes than at 1310 nm. The percentage fraction of energy obtained by the individual modes at different wavelengths is different, which results in different optical power leakage to adjacent channels [33].
Figure 11(d) illustrates the interlayer crosstalk of the double-layer waveguides. By selecting 1-channel and 6-channel waveguides as examples, which exhibit varying insertion losses between the upper and lower layers, a more representative depiction is provided. The blue (red) arrow in the inset represents that the SMF is aligned with the upper-layer (lower-layer) waveguide core at the input, and then the MMF is moved downward (upward) at the output. Overall, the interlayer crosstalk of the lower-layer waveguides is higher than that of the upper-layer waveguides when the displacement is less than 20 µm. This can be explained by the lower-layer waveguides experiencing longer UV exposure times than the upper-layer waveguides, resulting in overexposure and an increased refractive index. Consequently, more light leaks into the cladding [44]. The measured crosstalk is less than -35 dB at 850 nm and 1310 nm with a displacement of 12.5 µm. When the MMF is aligned with the lower-layer/upper-layer waveguides, the interlayer crosstalk is lower than -52 dB at 850 nm and -60 dB at 1310 nm. The experimental results show that the tendency of the interlayer crosstalk remains in good agreement with the simulated results. This further verifies that the d value of 24 µm is feasible. The interlayer crosstalk, along with the inter-channel crosstalk, is displayed in Table 3.
In future research, double-layer waveguides with various sizes or even multi-layer waveguides will be experimentally fabricated and investigated. The relationship between the interlayer distance and interlayer crosstalk will be determined, providing a reference for good preparation. Moreover, we will further develop 3D polymer waveguide devices deeply based on the interlayer distance optimization method.
5. Conclusion
In this study, we successfully investigated the optimization of interlayer distance for double-layer polymer optical waveguides to achieve low-loss and low-crosstalk performance. The simulation determined the minimal interlayer distances for double-layer waveguides with different core sizes. An optimal interlayer distance of 24 µm was determined for a 20 µm × 20 µm double-layer waveguide, ensuring interlayer crosstalk below -30 dB when the roughness remained under 80 nm. Experimentally, the plasma treatment time, the core exposure dose, and the cladding exposure dose were 10 s, 500 mJ/cm2, and 240 mJ/cm2, respectively. The double-layer waveguides, featuring an interlayer distance of 24.5 µm, were produced with a total length of 4.2 cm. The transmission loss for the upper-layer waveguides was 0.22 dB/cm at 850 nm and 0.38 dB/cm at 1310 nm, while the lower-layer waveguides experienced a transmission loss of 0.25 dB/cm at 850 nm and 0.40 dB/cm at 1310 nm. Furthermore, the interlayer crosstalk was less than -52 dB at 850 nm and -60 dB at 1310 nm. The results demonstrated good agreement between the experiment and the simulation. The investigation of the interlayer distance optimization method provides valuable guidance for fabricating double-layer or multi-layer waveguides and devices, holding significant potential for advancing high-density integration in short-distance optical interconnects.
Funding
National Key Research and Development Program of China (2020YFB1805800); National Natural Science Foundation of China (62027818).
Disclosures
The authors declare that there are no conflicts of interest related to this article.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from authors upon reasonable request.
References
1. Y. Lu and H. Gu, “Flexible and scalable optical interconnects for data centers: trends and challenges,” IEEE Commun. Mag. 57(10), 27–33 (2019). [CrossRef]
2. Q. Hu, M. Chagnon, K. Schuh, F. Buchali, and H. Bülow, “IM/DD beyond bandwidth limitation for data center optical interconnects,” J. Lightwave Technol. 37(19), 4940–4946 (2019). [CrossRef]
3. R. Dangel, J. Hofrichter, F. Horst, D. Jubin, A. La Porta, N. Meier, I. M. Soganci, J. Weiss, and B. J. Offrein, “Polymer waveguides for electro-optical integration in data centers and high-performance computers,” Opt. Express 23(4), 4736–4750 (2015). [CrossRef]
4. R. Dangel, C. Berger, R. Beyeler, L. Dellmann, M. Gmur, R. Hamelin, F. Horst, T. Lamprecht, T. Morf, S. Oggioni, M. Spreafico, and B. J. Offrein, “Polymer-waveguide-based board-level optical interconnect technology for datacom applications,” IEEE Trans. Adv. Packag. 31(4), 759–767 (2008). [CrossRef]
5. X. Zhang, A. Hosseini, X. Lin, H. Subbaraman, and R. T. Chen, “Polymer-based hybrid-integrated photonic devices for silicon on-chip modulation and board-level optical interconnects,” IEEE J. Select. Topics Quantum Electron. 19(6), 196–210 (2013). [CrossRef]
6. N. Bamiedakis, J. Beals, R. V. Penty, I. H. White, J. V. DeGroot, and T. V. Clapp, “Cost-effective multimode polymer waveguides for high-speed on-board optical interconnects,” IEEE J. Quantum Electron. 45(4), 415–424 (2009). [CrossRef]
7. V. Prajzler, M. Neruda, P. Jašek, and P. Nekvindová, “The properties of free-standing epoxy polymer multi-mode optical waveguides,” Microsyst. Technol. 25(1), 257–264 (2019). [CrossRef]
8. N. Bamiedakis, F. Y. Shi, R. V. Penty, I. H. White, and D. P. Chu, “Bend- and twist-insensitive flexible multimode polymer optical interconnects,” J. Lightwave Technol. 38(23), 6561–6568 (2020). [CrossRef]
9. M. O. F. Rasel, A. Yamauchi, and T. Ishigure, “Error-free three-dimensional multimode crossover graded-index polymer waveguides for board-level optical circuitry,” J. Lightwave Technol. 40(19), 1–10 (2022). [CrossRef]
10. W. Wei, X. Lu, F. Pang, H. Wei, L. Zhang, and T. Wang, “Fabrication and characterization of low-loss gaussian-like reversed ridge optical waveguides,” IEEE Photon. Technol. Lett. 34(12), 649–652 (2022). [CrossRef]
11. H. Sang-Pil, C. In-Kui, H. Sung-Hwan, L. Woo-Jin, and A. Seung-Ho, “A high-density two-dimensional parallel optical interconnection module,” IEEE Photon. Technol. Lett. 17(11), 2448–2450 (2005). [CrossRef]
12. S. H. Hwang, M. H. Cho, S.-K. Kang, T.-W. Lee, H.-H. Park, and B. S. Rho, “Two-dimensional optical interconnection based on two-layered optical printed circuit board,” IEEE Photon. Technol. Lett. 19(6), 411–413 (2007). [CrossRef]
13. R. Dangel, F. Horst, D. Jubin, N. Meier, J. Weiss, B. J. Offrein, B. W. Swatowski, C. M. Amb, D. J. DeShazer, and W. K. Weidner, “Development of versatile polymer waveguide flex technology for use in optical interconnects,” J. Lightwave Technol. 31(24), 3915–3926 (2013). [CrossRef]
14. X. Xu, L. Ma, and Z. He, “3D polymer directional coupler for on-board optical interconnects at 1550 nm,” Opt. Express 26(13), 16344–16351 (2018). [CrossRef]
15. A. Raptakis, L. Gounaridis, M. Weigel, M. Kleinert, M. Georgiopoulos, E. Mylonas, P. Groumas, C. Tsokos, N. Keil, H. Avramopoulos, and C. Kouloumentas, “2D optical phased arrays for laser beam steering based on 3D polymer photonic integrated circuits,” J. Lightwave Technol. 39(20), 6509–6523 (2021). [CrossRef]
16. Y. Yamaguchi, S. Yakabe, and T. Ishigure, “Design and fabrication of three-dimensional polymer optical waveguide-based fan-in/out device for multicore fibers,” in 2022 European Conference on Optical Communication (ECOC), (IEEE, 2022), 1–3.
17. X. Jiang, Z. Yang, Z. Liu, Z. Dang, Z. Ding, Q. Chang, and Z. Zhang, “3D integrated wavelength demultiplexer based on a square-core fiber and dual-layer arrayed waveguide gratings,” Opt. Express 29(2), 2090–2098 (2021). [CrossRef]
18. J. Dong, K. S. Chiang, and W. Jin, “Mode multiplexer based on integrated horizontal and vertical polymer waveguide couplers,” Opt. Lett. 40(13), 3125–3128 (2015). [CrossRef]
19. J. Dong, K. S. Chiang, and W. Jin, “Compact three-dimensional polymer waveguide mode multiplexer,” J. Lightwave Technol. 33(22), 4580–4588 (2015). [CrossRef]
20. Y. Wu and K. S. Chiang, “Ultra-broadband mode multiplexers based on three-dimensional asymmetric waveguide branches,” Opt. Lett. 42(3), 407–410 (2017). [CrossRef]
21. W. K. Zhao, K. X. Chen, J. Y. Wu, and K. S. Chiang, “Horizontal directional coupler formed with waveguides of different heights for mode-division multiplexing,” IEEE Photonics J. 9(5), 1–9 (2017). [CrossRef]
22. Q. Huang, Y. Wu, W. Jin, and K. S. Chiang, “Mode multiplexer with cascaded vertical asymmetric waveguide directional couplers,” J. Lightwave Technol. 36(14), 2903–2911 (2018). [CrossRef]
23. J. H. Ryu, T. H. Lee, I.-K. Cho, C.-S. Kim, and M. Y. Jeong, “Simple fabrication of a double-layer multi-channel optical waveguide using passive alignment,” Opt. Express 19(2), 1183–1190 (2011). [CrossRef]
24. Q. Huang, K. S. Chiang, and W. Jin, “Thermo-optically controlled vertical waveguide directional couplers for mode-selective switching,” IEEE Photonics J. 10(6), 1–14 (2018). [CrossRef]
25. Z. Y. Zhang, M. Kleinert, A. Maese-Novo, G. Irmscher, E. Schwartz, C. Zawadzki, and N. Keil, “Multicore polymer waveguides and multistep 45° degrees mirrors for 3D photonic integration,” IEEE Photon. Technol. Lett. 26(19), 1986–1989 (2014). [CrossRef]
26. Y. Ichihashi, P. Henzi, M. Bruendel, J. Mohr, and D. G. Rabus, “Polymer waveguides from alicyclic methacrylate copolymer fabricated by deep-UV exposure,” Opt. Lett. 32(4), 379–381 (2007). [CrossRef]
27. F. P. Payne and J. P. R. Lacey, “A theoretical analysis of scattering loss from planar optical waveguides,” Opt. Quantum Electron. 26(10), 977–986 (1994). [CrossRef]
28. K. K. Lee, D. R. Lim, H.-C. Luan, A. Agarwal, J. Foresi, and L. C. Kimerling, “Effect of size and roughness on light transmission in a Si/SiO2 waveguide: experiments and model,” Appl. Phys. Lett. 77(11), 1617–1619 (2000). [CrossRef]
29. Y. Shi, L. Ma, Y. Zhuang, and Z. He, “Investigation on roughness-induced scattering loss of small-core polymer waveguides for single-mode optical interconnect applications,” Opt. Express 28(26), 38733–38744 (2020). [CrossRef]
30. A. Yariv, “Coupled-mode theory for guided-wave optics,” IEEE J. Quantum Electron. 9(9), 919–933 (1973). [CrossRef]
31. H. H. Hamid, C. E. Rüter, D. V. Thiel, and T. Fickenscher, “Characteristics and crosstalk of optical waveguides fabricated in polymethyl methacrylate polymer circuit board,” Appl. Opt. 55(32), 9017–9021 (2016). [CrossRef]
32. E. A. Marcatili, “Dielectric rectangular waveguide and directional coupler for integrated optics,” Bell Syst. Tech. J. 48(7), 2071–2102 (1969). [CrossRef]
33. C. Deng, T. Wang, J. Chen, C. Mou, X. Zhang, X. Yan, and F. Pang, “Roughness estimation of multimode rectangular optical waveguide based on the scattering loss difference between SMFW and MMFW Structures,” IEEE Photonics J. 9(2), 1–13 (2017). [CrossRef]
34. X. Xu, X. Liu, M. Immonen, L. Ma, and Z. He, “Investigation on mode dispersion and lamination stability of multimode polymer waveguides for an optical backplane,” Opt. Express 30(22), 40505–40514 (2022). [CrossRef]
35. K. Halbe and E. Griese, “A modal approach to model integrated optical waveguides with rough core-cladding-interfaces,” 2006 IEEE Workship on Signal Propagation on Interconnects, 133–136 (2006).
36. S. K. Pani, C. C. Wong, K. Sudharsanam, and V. Lim, “Direct measurement of sidewall roughness of polymeric optical waveguides,” Appl. Surf. Sci. 239(3-4), 445–450 (2005). [CrossRef]
37. S. Pani, C. Wong, and K. Sudharsanam, “Variation of sidewall roughness of polymeric waveguides during reactive ion etching,” Appl. Phys. Lett. 85(7), 1295–1297 (2004). [CrossRef]
38. F. Grillot, L. Vivien, S. Laval, and E. Cassan, “Propagation loss in single-mode ultrasmall square silicon-on-insulator optical waveguides,” J. Lightwave Technol. 24(2), 891–896 (2006). [CrossRef]
39. N. W. Ulrich, J. Andre, J. Williamson, K.-W. Lee, and Z. Chen, “Plasma treatment effect on polymer buried interfacial structure and property,” Phys. Chem. Chem. Phys. 19(19), 12144–12155 (2017). [CrossRef]
40. U. Cvelbar, S. Pejovnik, M. Mozetie, and A. Zalar, “Increased surface roughness by oxygen plasma treatment of graphite/polymer composite,” Appl. Surf. Sci. 210(3-4), 255–261 (2003). [CrossRef]
41. M. Mozetič, A. Zalar, P. Panjan, M. Bele, S. Pejovnik, and R. Grmek, “A method of studying carbon particle distribution in paint films,” Thin Solid Films 376(1-2), 5–8 (2000). [CrossRef]
42. V. Prajzler and V. Chlupatý, “Epoxy polymer optical waveguide for micro-opto-electro-mechanical systems,” Microsyst. Technol. 26(9), 3029–3035 (2020). [CrossRef]
43. Y. Li, Y. Wang, H. Li, B. Jiang, F. Bai, and J. Feng, “Terahertz rectangular waveguides by UV-LIGA with megasonic agitation,” Micromachines 13(10), 1601 (2022). [CrossRef]
44. Y.-S. Sohn, H.-K. Oh, and I. An, “Parameter extraction for 193 nm chemically amplified resist from refractive index change,” J. Vac. Sci. Technol., B: Microelectron. Process. Phenom. 19(6), 2077–2081 (2001). [CrossRef]
45. Y. Zheng, P.-p. Gao, X. Tang, J.-z. Liu, and J.-a. Duan, “Effects of electron beam lithography process parameters on structure of silicon optical waveguide based on SOI,” J. Cent. South Univ. (Engl. Ed.) 29(10), 3335–3345 (2022). [CrossRef]
46. V. Prajzler, M. Neruda, P. Nekvindova, and P. Mikulik, “Properties of multimode optical epoxy polymer waveguides deposited on silicon and TOPAS substrate,” Radioengineering. 26(1), 10–15 (2017). [CrossRef]
47. X. Wang, C. Deng, Y. Huang, F. Zhang, R. Zhang, X. Zhang, and T. Wang, “Spherical concave micro-mirror fabricated using gray-tone optical lithography for vertical coupling,” Opt. Express 29(9), 13288–13301 (2021). [CrossRef]
