We have designed, fabricated and characterized poly(dimethylsiloxane) (PDMS) arrayed waveguide grating (AWG) with four-channel output for operation in the visible light wavelength range. The PDMS AWG was realized based on the single-mode PDMS rib waveguide. The device was designed for 1 nm channel spacing with the wavelength ranging from 639 to 644 nm. The measured insertion loss is 11.4 dB at the peak transmission spectrum and the adjacent crosstalk is less than −16 dB. The AWG device occupies an area of 7.5 × 15 mm2. PDMS AWG has the potential for integration with microfluidics in a monolithic PDMS lab-on-a-chip device for visible light spectroscopy applications.
©2010 Optical Society of America
The arrayed waveguide grating (AWG) multi/demultiplexer has been developed for wavelength division multiplexing (WDM) photonic networks [1–3]. Since its inception, it has been demonstrated for high wavelength resolution, low insertion loss and high stability . Many groups have reported the realization of AWG in various materials such as Silicon-on-Insulator (SOI) , Silica-on-Silicon  and Silicon-based Polymer Technology  for the infrared wavelength range in telecommunication applications. The attractive characteristics of AWGs have led to new applications other than the telecommunication application such as spectroscopy and sensing of chemical samples . Conventional spectroscopy equipment consists of broad-band light source, dispersive element, slit and detector for high resolution wavelength separation. The highly dispersive array waveguides and focusing slab waveguides in AWG ease the integration of the required elements on a planar substrate for wavelength separation in spectroscopy.
Spectroscopy for biological and chemical samples mostly occurs at the visible light wavelengths (400–700 nm) because a large portion of the spectroscopical signatures lie in this wavelength range. Therefore, spectroscopy requires dedicated integrated photonic devices which are transparent and operable in the visible wavelengths. An AWG-based spectrometer for biochemical spectroscopy has been demonstrated in Silica-on-Silicon technology for the usage in the visible light wavelength range . Silica have initially been chosen because they can be easily produced using the well-established techniques in microelectronics. Nevertheless, AWGs made of silica are more costly than their polymer counterpart. Low cost polymer AWGs enable the realization of disposable biochips for reduced sample contamination. Moreover, polymer-based waveguide devices offer rapid processibility, high yields, high performance such as lower optical loss and smaller birefringence compared to silica as well as compactness owing to the large refractive index contrast .
Poly(dimethylsiloxane) (PDMS) has been widely used for the fabrication of microfluidics and lab-on-a-chip (LOC) devices due to its unique polymer characteristic such as biocompatibility, low cost, and rapid prototyping capability by soft lithography . Most often, LOC devices depend on optical detection for sensing biochemical species  and the high transparency of PDMS in the visible range thus motivates the monolithic integration of optical components with microfluidics in the same material. Recently, we have reported large refractive index contrast single-mode PDMS waveguide operating in the visible light range  which can be used as basic building blocks for complex microphotonics devices. The PDMS single-mode waveguide offers compactness and ease of monolithic integration for both the AWG and microfluidics in a planar configuration which enables the realization of a microspectrometer in a LOC platform.
In this paper, we report a four-channel PDMS AWG based on the single-mode PDMS rib waveguide for operation in the range of 639 to 644 nm. The AWG was designed for 1 nm channel spacing and was fabricated using the inexpensive soft lithography method with two steps curing to form the core and cladding layers of the waveguide. Further simulation studies on the variations of the rib waveguides geometrical and material properties were performed to analyze the fabrication-related performance errors. The modal images from the fabricated AWG output waveguides were observed and the transmission spectrum of the device was characterized for insertion loss and channel crosstalk. The PDMS AWG was demonstrated for wavelength separation operable in the visible light range and is well suited for the implementation of a lab-on-a-chip spectrometric platform.
2. Theoretical design & studies
Figure 1 shows the schematic diagram of an AWG structure which consists of the input & output waveguides, the two concave slab waveguide star couplers (or free propagation region, FPR), connected by an array of waveguides with a constant path length difference, ∆L between adjacent waveguides.
As light enters the center of FPR from the input waveguides, light diffracts as it would be in free space, diverging and coupling into the arrayed waveguides. The arrayed waveguides are positioned along a Rowland arc centered at the input waveguide-slab waveguide junction and with a radius equal to the focal length f of the FPR. Thus, the light arriving at the beginning of each of the waveguide array has the same phase. The arrayed waveguide has been designed such that the constant path length difference ∆L between adjacent waveguides equals an integer multiple of the central wavelength λc. After light propagates through the arrayed waveguides, this central wavelength constructively interferes into one focal point at the center of the second FPR. As the wavelength varies from the center wavelength, λc + ∆λ, a linear phase change occurs across the waveguides of the array due to the constant path length difference between adjacent waveguides. This ultimately results in a tilted wavefront focused in a point shifted away from the center. Thus, by placing the output waveguide at a proper position along the Rowland arc, spatial separation of different wavelengths channels is produced. Detailed explanations and derivation of equations useful for the design and in-depth understanding of the operation of the AWG can be found elsewhere [13–15].
The input and output ports of the PDMS AWG were designed to be perpendicular to one another to circumvent stray light from being detected. Stray light is part of the incident light from the optical fiber which does not couple into the waveguide and propagates unguided in the device to the output optical fiber. In addition, stray light also comes from higher order modes which are produced by fiber’s imperfect coupling to the waveguide and propagate unbounded into the slab of the waveguide.
The first step of designing the PDMS AWG is to produce single-mode PDMS waveguides as the first basic building block necessary for the ultimate realization of the AWG. The theoretical design and simulation of the single-mode PDMS rib waveguide was reported previously  and will not be detailed here. Both the rib height and width of the waveguide are 8 µm and the slab height of the waveguide is 4 µm. For the design of a PDMS AWG operating in the wavelength range of 639 to 644 nm, the refractive indices of core and cladding of the waveguide were chosen as ncore = 1.429 and nclad = 1.412. The effective index of the waveguide, ns, the group effective index, ng of array waveguide and slab waveguide were then determined using Finite Difference Time Domain (FDTD) method simulations of the single-mode waveguide. These results are necessary for the design of the PDMS AWG geometrical parameters. The general procedures for the PDMS AWG design are as follows:
- (1) The number of output channels N was selected as 4 (as a proof-of-concept AWG).
- (2) The wavelength channel spacing ∆λ was selected as 1 nm to demonstrate nanometer resolution sensing capability.
- (3) The free spectral range FSR was calculated using the simple relation FSR = N∆λ and also considering an additional tolerance of ± 0.5 nm which results in a final value of 5 nm for the required FSR.
- (4) The diffraction order m was calculated as m = integer(λc/FSR).
- (5) The length difference ∆L between the adjacent waveguides was calculated as ∆L = (m λc/ncore).
- (6) The output waveguide pitch ∆x and the array waveguide pitch d were selected to be as small as possible to create a compact AWG. The lowest limit of the pitch was constrained by the critical dimensions of the mask used in the fabrication process. On the other hand, the output waveguide pitch must be sufficiently large to provide isolation between neighbouring waveguides.
- (7) The focal length of the FPR Lf can be determined from the dispersion formula:
The parameters of the designed AWG were simulated using the Apollo Photonics Solution Suite (APSS) software. The results of the parameter optimization are shown below in Table 1 .
The simulated insertion losses of the PDMS AWG for both the Transverse Electric (TE) mode and Transverse Magnetic (TM) modes are shown in Fig. 2 . It can be seen that the insertion losses for both the TE and TM modes show similar transmission spectrum peaks and position. The insertion loss is 3.6 dB at the transmission peak located at 641 and 642 nm and the adjacent crosstalk is >34dB. The overall insertion loss of the AWG originates from the combined contribution of several factors, such as the FPR to array waveguides transition, the curvature of the waveguides, and the transition between straight to curved waveguides. The minimum bending radius of our AWG was kept at 4000 µm which has negligible loss (0.0018 dB/µm). The loss due to the transition between straight to bending waveguides was calculated to be 0.2 dB. Thus, the main contribution for the overall insertion loss comes from the FPR to array waveguide transition. An estimated 3 dB power loss was expected at this junction because our 8 µm wide array waveguides were separated laterally only by a gap of 16 µm from one another. Hence, an approximate 3 dB was expected at this junction.
3. Fabrication & characterization
The fabrication of a single-mode PDMS rib waveguide was previously presented in detail elsewhere . Therefore, only a short summary of the key aspects for the fabrication of the PDMS AWG is presented here.
The master mold for the soft lithography was prepared by patterning a 8 µm thick SU-8 2007 (MicroChem, NewtonMA, USA) photoresist spun on 8-inch silicon wafers. The exposure dose and duration for post-exposure bake steps were optimized for acceptable vertical sidewall profile and low surface roughness. The patterned SU-8 master mold was treated with a silane anti-stiction coating in a chemical vapor deposition (CVD) process to avoid sticking of the PDMS onto the SU-8 mold and also to reduce the sidewall roughness. The PDMS precursor mixture (OE-43, Gelest) was prepared using a 1:1 base to curing agent weight ratio. After the mixture was degassed for 15 mins, this PDMS precursor was spun onto the wafer to produce a thin layer of PDMS which made up the core waveguide layer with a slab height of ~4 µm. The spin coated PDMS was cured at 55°C for 4 hours. Thereafter, the second PDMS precursor mixture (Sylgard 184, Dow Corning) was cured at 80°C for 2 hours over the thin layer of PDMS to form the cladding layer. Finally, the cured PDMS structure was released from the master mold.
The PDMS AWG was characterized on a fiber optic alignment stage set. The light source is a solid-state laser (iFLEX2000, PointSource) with a peak wavelength at 641 nm and an emission range of 639 to 644 nm. The laser light source was butt-coupled into the AWG input waveguide through a 9/125 µm single-mode fiber optic. The TE and TM modes measurement was carried out with a polarization maintaining fiber connected to the polarized connector of the laser module. The image of the output end of the waveguide was focused with a 5 × objective lens and captured on a CCD camera (Exwave, Sony, Japan). For the transmission spectrum measurement, the light from each output waveguide was measured using a visible light spectrometer (OSM100, Newport) collected through a multimode optical fiber. The readings were repeated three times and averaged for the spectrum insertion loss measurement.
4. Results & discussions
Figure 3 shows the SEM images of the fabricated four-channel PDMS AWG. The images show that the PDMS AWG has well defined sidewall and corners. Cross-sectional microscope studies showed that the PDMS AWG had a slab height of 4.2 µm, a rib height of 9.1 µm and an 88° negative sloping of the sidewall.
Simulations were further performed to evaluate the outcome of the deviations in the slab height, rib height, refractive index contrast and sidewall slope of the waveguides on the AWG performance due to the fabrication process (Fig. 4 ). The refractive indices of the PDMS may differ from the values used in simulations because of inherent variations in the process, such as the curing temperature and duration. Thus, an estimated change of 5% and 10% was considered for the refractive index contrast ∆n. Figure 4a shows that such ∆n fluctuations have very little impact on the transmission spectrum. As the slab height increases, the insertion loss decreases but the crosstalk deteriorates with the transmission peak shifting to larger wavelengths (Fig. 4b). This observation can be explained as follows: a small slab height produces a small beam diameter with stronger confinement, which helps in the diffraction and spectral separation but at the same time provides a smaller transmission area which increases the insertion loss. The slab height of the waveguide depends on the viscosity of the PDMS precursor mixture and the spin speed during spin coating of the PDMS core waveguide layer. A 4 µm slab height was targeted in fabrication by keeping a constant time between the PDMS precursor mixing step to the spin coating step because PDMS viscosity changed once the curing agent was added to the precursor. The thickness and sidewall profile of the SU-8 photoresist determined the waveguide height and sidewall slope. As shown in Fig. 4c, a negative slope within 2 degrees of a vertical sidewall does not shift the peak nor increases the insertion loss significantly. Figure 4d shows that each 1 µm increment in the rib height produces a peak shift of 0.05 nm and increases insertion loss with 0.1 dB. In the fabricated PDMS AWG, a 9.2 µm thick PDMS AWG with an 88° negative sloping has an estimated peak shift less than 0.1 nm and insertion loss of 4.43 dB. These simulation results show that the PDMS core layer slab height is the most critical parameter in the AWG fabrication for ensuring a practical transmission spectrum very close to the simulated one.
The near field pattern of the four output waveguides separated by a port pitch of 250 µm is shown in Fig. 5 . The plotted profile of the mode field is compared with the simulation results for the 4 peak wavelength transmission at 640, 641, 642 and 643 nm. The mode profile shows excellent transmission of the beam intensity through the AWG. Each of the four captured output mode profiles was fitted with a Gaussian variation that was characterized by a FWHM of 4.92 µm as compared with the value of 4.86 µm indicated by the simulations.
The polarization dependence of the pass wavelength difference of an AWG is largely determined by the birefringence of the waveguide. The polarization-dependent pass wavelength difference is given by 
The birefringence of a rib waveguide structure is mainly caused by stress anisotropy along x and y direction and the geometrical parameters of the rib waveguide, i.e. the width to height ratio and slab height . The calculated polarization-dependent pass wavelength due to geometrical parameters of the rib waveguide is in the order of 0.003 nm, as shown in Fig. 6a . The PDMS rib waveguide is made up of a single polymer material which has negligible stress anisotropy induced by any lattice mismatch. As shown in Fig. 6b, the measured pass wavelength difference is 0.01 nm between the TE and TM polarization. However, the measured crosstalk has deteriorated as compared with the simulated result (Fig. 2) and is mainly attributed to the phase error in the arrayed waveguides region . The phase error is caused by the effective index fluctuation which resulted from the non-uniformities of the width and height of the rib waveguide and the inhomogeneity of the PDMS refractive index. Extra caution was taken in the spin coating and pre-exposure bake steps of the patterning of SU-8 photoresist. The thickness uniformity was measured to be in the range of ± 0.1 µm. Hard contact was employed in the photolithography step to reduce lateral variations due to diffraction effect. The inhomogeneity of the PDMS refractive index is related to the mixing and curing process. This inhomogeneity effect can be alleviated by degassing the PDMS precursor for at least 1 hour before curing and slow ramping-up of the temperature in the curing step.
Figure 7 shows the measured transmission spectrum of the four-channel PDMS AWG. The measured total insertion loss was 11.4 dB for the transmission peak at 641 nm and the insertion loss uniformity of the four channels is 0.5 dB. The non-uniformity insertion loss of the PDMS AWG is being termed as a Gaussian shape spectral response where the center output port has a higher transmission than the peripheral output port. The reason for this spectral response is the Gaussian input electric field being sampled by the arrayed waveguide and reproduced at the output plane . As shown in Fig. 4b, the Gaussian spectral non-uniformity increases as the slab height of the single-mode PDMS waveguide increases from 2 to 6 µm. As the slab height increases, the Gaussian beam mode diameter of the input waveguide increases. At a fixed focal length, the width of the Gaussian far field pattern projected by the input waveguide at the arrayed waveguide is inversely proportional to the beam mode diameter of the input waveguide. For an equal number of arrayed waveguides of identical size and pitch, a broader Gaussian field produces a more uniform insertion loss because it is being sampled only at the center flattened field. In contrast, a narrower Gaussian field will be sampled across the entire field pattern. Nevertheless, the insertion loss for an input waveguide with a smaller slab height increases due to the partial sampling of the Gaussian field as well as a smaller transmission area. In the fabrication of our PDMS AWG, the slab height of the waveguide varies within 4 ± 1 µm. Thus, a variation of the insertion loss uniformity is expected within 0.4 to 0.6 dB (Fig. 4b). The PDMS AWG insertion loss non-uniformity can be improved by modifying the input waveguide with a flat response shape  or by making a sinc-like electric field envelope in the array waveguides .
The adjacent crosstalk of the AWG is >15.1 dB. The insertion loss differs from the simulated value by ~7 dB. The much higher insertion losses observed practically are attributed to the fiber to waveguide coupling loss, measured to be 4.1 dB. The other losses can be attributed to the waveguide bending and propagation loss of the waveguides as well as the variation of the slab height in the fabrication process. The measured transmission spectrum shows channel separation of 1 nm and FSR that spans more than 4 nm.
The face end PDMS waveguide has striation marks due to direct molding or cutting. The waveguide does not allow face end polishing like those utilized in silicon-based waveguide to enhance coupling efficiency. Nonetheless, the fiber-to-waveguide coupling can be improved by incorporating a tapered structure or grating coupler.
The normal operating temperature for PDMS is from –45°C to 200°C and thermal degradation only happens when temperature exceeds 300°C. However, PDMS is a rubbery material which is sensitive to mechanical and temperature fluctuations. Based on the values of dn/dT = –3.4 × 10−4 / °C and coefficient of thermal expansion α = 3.4 × 10−4/°C, the calculated temperature sensitivity of the pass wavelength in the PDMS AWG for visible light is about ∆λ/∆T = 4.8 × 10−2 nm/°C. Hence, the operation of the PDMS AWG for visible light spectroscopy applications should be implemented in a well-controlled environment to prevent thermal shift of the spectrum.
A low diffraction order (m ≤ 2) is needed for the AWG to span a large free spectral range (~300 nm) to cover the whole visible light because FSR ≈λc/m. Such a low diffraction order (i.e. small pathlength difference between adjacent waveguide) causes the waveguides layout in the array to overlap/merge each other and become indistinguishable from one another. In order to eliminate this problem a non-conventional shape has to be adopted for the AWG, such as the gull-wing shape .
A PDMS AWG based on single-mode PDMS rib waveguides has been designed, fabricated and characterized. A channel separation of 1 nm with a FSR that spans more than 4 nm was designed to operate in the wavelength range of 639 to 644 nm. This PDMS AWG was fabricated using rapid prototyping capability by soft lithography with a two step curing process. Further simulation studies on the variations of the rib waveguides geometrical and material properties were performed to assess the fabrication-related performance errors. The measured transmission spectrum agrees well with the simulated transmission spectrum in terms of channel separation and FSR. Such a PDMS AWG has great potential for integration with microfluidics to achieve a monolithic PDMS spectrometer lab-on-a-chip.
References and links
1. M. K. Smit, “New focusing and dispersive planar component based on an optical phased array,” Electron. Lett. 24(7), 385–386 (1988). [CrossRef]
2. H. Takahashi, S. Suzuki, K. Kato, and I. Nishi, “Arrayed-waveguide grating for wavelength division multi/demultiplexer with nanometer resolution,” Electron. Lett. 26(2), 87–88 (1990). [CrossRef]
3. C. Dragone, “An N×N optical multiplexer using a planar arrangement of two star couplers,” IEEE Photon. Technol. Lett. 3(9), 812–815 (1991). [CrossRef]
4. Y. Hibino, “Recent advances in high-density and large-scale AWG multi/demultiplexers with higher index-contrast silica-based PLCs,” IEEE J. Sel. Top. Quantum Electron. 8(6), 1090–1101 (2002). [CrossRef]
5. P. Cheben, J. H. Schmid, A. Delâge, A. Densmore, S. Janz, B. Lamontagne, J. Lapointe, E. Post, P. Waldron, and D. X. Xu, “A high-resolution silicon-on-insulator arrayed waveguide grating microspectrometer with sub-micrometer aperture waveguides,” Opt. Express 15(5), 2299–2306 (2007). [CrossRef] [PubMed]
6. K. Okamoto, K. Moriwaki, and S. Suzuki, “Fabrication of 64 × 64 arrayed waveguide grating multiplexer on Silicon,” Electron. Lett. 31(3), 184–185 (1995). [CrossRef]
7. M. Diemeer, L. Spiekman, R. Ramsamoedj, and M. K. Smit, “Polymeric phased array wavelength multiplexer operating around 1550 nm,” Electron. Lett. 32(12), 1132–1133 (1996). [CrossRef]
8. K. Kodate and Y. Komai, “Compact spectroscopic sensor using an arrayed waveguide grating,” J. Opt. A, Pure Appl. Opt. 10(4), 044011–044018 (2008). [CrossRef]
9. H. Ma, A. K.-Y. Jen, and L. R. Dalton, “Polymer-Based Optical Waveguides: Materials, Processing, and Devices,” Adv. Mater. 14(19), 1339–1365 (2002). [CrossRef]
11. E. Verpoorte, “Chip vision-optics for microchips,” Lab Chip 3(3), 42N–52N (2003).
13. M. K. Smit and C. Van Dam, “PHASAR-Based WDM-Devices: Principles, Design and Applications,” IEEE J. Sel. Top. Quantum Electron. 2(2), 236–250 (1996). [CrossRef]
14. K. Okamoto, Fundamental of Optical Waveguides, (Academic Press, 2006), Chap. 9.
15. P. Cheben, “Wavelength Dispersive Planar Waveguide Devices: Echelle and Arrayed Waveguide Gratings,” in Optical Waveguides: From Theory to Applied Technologies, M. L. Calvo and V. Laksminarayanan, ed.(Taylor & Francis, London, 2007).
16. L. Vivien, S. Laval, B. Dumont, S. Lardenois, A. Koster, and E. Cassan, “Polarization-indepenedent single-mode rib waveguides on silicon-on-insulator for telecommunication wavelengths,” Opt. Commun. 210(1-2), 43–49 (2002). [CrossRef]
17. C. R. Doerr, and K. Okamoto, Optical Fiber Telecommunications V A:Components and Subsystems, (Elsevier Inc., 2008), Chap. 9.
18. K. Okamoto and A. Sugita, “Flat spectral response arrayed-waveguide grating multiplexer with parabolic waveguide horns,” Electron. Lett. 32(18), 1661–1662 (1996). [CrossRef]