Open Access
Add to BibSonomy
Abstract
We present an all-silicon transverse-magnetic-pass (TM-pass) polarizer based on anti-symmetric Bragg gratings. We obtain wide operation bandwidth and high polarization extinction ratio (PER) by maximizing the coupling between the forward TE0 mode and the backward TE1 mode through the reduction of the bridge element width. In the meantime, low insertion loss (IL) is acquired with long tapered structures and the exclusion of the center grating part. Experimental results indicate IL below 0.74 dB and PER over 40 dB covering the wavelength ranges of 1275–1360 nm and 1500–1523 nm, while the average IL within these ranges is as low as 0.27 dB. Additionally, simulation results suggest that the performance can be further improved by introducing chirp in the period of Bragg gratings, thus achieving IL < 0.11 dB and PER > 60 dB over a wide range of 280 nm (1290–1570 nm).
© 2023 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Silicon-on-insulator (SOI) holds great promise in the realization of photonic integrated circuits (PICs) due to its numerous advantages [1], including low loss, high integration density, and compatibility with the complementary metal oxide semiconductor (CMOS) process. However, the high refractive index contrast between silicon and silicon dioxide introduces challenges, such as strong light confinement and significant birefringence in waveguides, which can potentially degrade the performance of polarization-sensitive devices. Hence, on-chip polarization management including polarizers, polarization beam splitters (PBSs), polarization splitter and rotators (PSRs), and polarization rotators (PRs) is of high importance [2–8]. Among them, polarizers function as polarization filters which selectively block unwanted polarizations while minimizing polarization crosstalk. So far extensive efforts have been devoted to achieving high-performance polarizers, which are characterized by wide operation bandwidth (BW), low IL, low back reflection (BR), high PER, and compact footprint.
In the past decade, researchers have explored various types of polarizers with different working mechanisms and materials. The three most common mechanisms are absorption, reflection, and leakage of undesired polarization. Over the years, to achieve better integration, polarizers have transited from traditional bulk materials using birefringent bulk crystals [9,10] to on-chip platforms compatible with complementary metal oxide semiconductor (CMOS) technology. Absorption-based polarizers utilize materials with absorptive properties, such as hybrid plasmonic structures [11–13] and graphene [14,15]. These polarizers offer a compact footprint (< 10 µm), but they suffer from high insertion loss due to the inherent metal absorption (IL ∼ 2-5 dB). Although several techniques have been employed to achieve low loss (IL < 0.5 dB) in hybrid plasmonic polarizers, such as directional coupling [16], Bragg effect [17], and mode conversion [18], they often involve complex fabrication processes (additional metal deposition) and have limited bandwidth (BW < 150 nm).
The other two mechanisms are commonly used in all-silicon polarizers. Leakage-based polarizers involve specially designed cross-sections [19–21], double layer tapers combined with shallow etched silicon waveguides [22], bends [23,24], bends assisted by subwavelength gratings (SWGs) [2], SWGs [25], asymmetric directional couplers (ADCs) [26], corner mirrors [27], and segmented waveguide [28]. However, polarizers based on specially designed cross-sections and shallow etched silicon waveguides require a two-step etching process or an additional deposited silicon layer in the substrate, which may not be compatible with the widely used SOI platform with typical silicon layer thicknesses of 220 or 340 nm. Polarizers based on bends and ADCs are wavelength-sensitive and have narrow bandwidths (BW < 100 nm). SWG-type polarizers typically have a minimum feature size of less than 80 nm, making them challenging for fabrication. Polarizers that utilize segmented waveguides operate in the radiation regime, resulting in relatively high levels of insertion losses.
Reflective polarizers exploit Bragg gratings [29–32], photonic crystals (PHCs) [33], tilted SWGs [34], and nanobeams [35,36]. Bragg grating polarizers suffer from high back reflection and require additional isolators or circulators when cascaded directly after the light source. To address the back reflection issue, polarizers based on PHCs, tilted SWGs, and nanobeams convert the input fundamental mode into higher-order modes, which can then be filtered out using single-mode waveguides. However, these approaches either have narrow bandwidths (< 150 nm) or high insertion losses (IL < 1.3 dB).
In the realm of all-silicon polarizers, it is challenging to find devices that simultaneously possess low insertion loss, low back reflection, high polarization extinction ratio, broad bandwidth, and compatibility with the commonly used 220 or 340 nm SOI platform with SiO2 cladding. In this study, we propose and experimentally demonstrate an on-chip 220 nm all-silicon TM-pass polarizer based on anti-symmetric Bragg gratings. Experimental results show competitive performance of the proposed polarizer, indicating IL < 0.74 dB and PER > 40 dB over the wavelength ranges of 1275–1360 nm and 1500–1523 nm with the average IL within these ranges being as low as 0.27 dB (performance within wavelength range of 1360–1500 nm cannot be characterized due to lack of corresponding light source). Additionally, simulation results suggest that the performance can be further improved by introducing chirp in the period of Bragg gratings, thus achieving exceptional performance characteristics of IL < 0.11 dB and PER > 60 dB over a wide range of 280 nm (1290–1570 nm). Overall, our findings demonstrate the outstanding performance of the proposed all-silicon TM polarizer based on anti-symmetric Bragg gratings, offering a compact and efficient solution for on-chip polarization management across a wide range of wavelength.
2. Device design
2.1 Device structure and operating principle
Figure 1(a) illustrates the configuration of the proposed TM polarizer. It consists of two tapered anti-symmetric Bragg gratings, with the upper half part intentionally misaligned by half the grating pitch compared to the lower half. As shown in Fig. 1(b), the width of the input and output waveguide ${\textrm{W}_1}$ is set at a fixed value of 350 nm to ensure compliance with the single-mode condition for the desired broadband operation. The width of the grating segment increases linearly from ${\textrm{W}_1}$ to ${\textrm{W}_3}$ over N periods and then symmetrically decreases back to ${\textrm{W}_1}$. Simultaneously, the pitch of the grating segment follows a similar pattern, increasing from ${\mathrm{\Lambda }_{\textrm{min}}}$ to ${\mathrm{\Lambda }_{\textrm{max}}}$ and then decreasing back to ${\mathrm{\Lambda }_{\textrm{min}}}$. A bridge element spans the entire length of the Bragg gratings, facilitating the adiabatic transition of the TM mode. The width of the bridge narrows down linearly from W1 to W2 and then symmetrically widens back to ${\textrm{W}_1}$. The duty cycle, denoted as $\textrm{DC} = \textrm{a}/\mathrm{\Lambda }$ in the inset of Fig. 1(b), is chosen as $\textrm{DC} = \textrm{}0.5$ to achieve a larger minimum feature size and stronger coupling between the forward TE0 mode and the backward TE1 mode. In this study, the device is based on a 220 nm silicon-on-insulator (SOI) platform, which includes a 3 µm-thick SiO2 buried layer and cladding.
Fig. 1. (a) Schematic of the proposed TM-polarizer. For TE polarization, the polarizer acts as a Bragg reflector converting forward TE0 mode into backward TE1 mode whereas for TM polarization, the polarizer behaves like a homogeneous metamaterial supporting the transmission with negligible loss. (b) Essential parameters of the polarizer. Inset: Detailed parameters of grating unit cell. For clarity, only a small number of periods are depicted.
Anti-symmetric waveguide Bragg gratings were originally proposed as narrowband filters [37,38]. However, in order to achieve broadband reflective polarizers, it is necessary to maximize the bandgap induced by anti-symmetric perturbation. In this paper, we investigate the impact of the width of the bridge element on the band diagram. The band structure of Floquet–Bloch modes is calculated using a 3D Finite Difference Time Domain (FDTD) method. For the grating unit cell depicted in Fig. 2(a), a uniform mesh grid is employed with $\mathrm{\Delta x\;\ } = \mathrm{\;\ \Delta y\;\ } = \mathrm{\;\ \Delta z\;\ } = \textrm{}10\textrm{nm}$. Bloch boundary conditions are applied to the two surfaces perpendicular to the propagation direction, while perfectly matched layer (PML) absorbing boundary conditions are used on the four surfaces parallel to the propagation direction. In our demonstration, we select a grating unit with a pitch Λ of 400 nm and a waveguide width ${\textrm{W}_{\textrm{unit}}}$ of 1.1 µm to analyze the effect induced by the width of the bridge element. The band diagram plots the normalized wavenumbers $\mathrm{k\Lambda }/({2\mathrm{\pi }} )$ of the Floquet–Bloch modes against the wavelength $\mathrm{\lambda }$, where $\textrm{k} = \textrm{}{\textrm{n}_{\textrm{eff}}}2\mathrm{\pi }/\mathrm{\lambda }$ and ${\textrm{n}_{\textrm{eff}}}$ represents the effective index of the Floquet–Bloch mode supported by the structure. Figures 2(c) and 2(d) illustrate the band diagram of TE-polarized Floquet–Bloch modes and TM-polarized Floquet–Bloch modes, respectively, in the absence of a bridge element. In the TE band structure, a wide bandgap is generated by the anti-crossing of the fundamental TE mode and the first-order TE mode (denoted by the gray shaded area). When the wavelength of the forward TE0 mode falls within this bandgap, it is reflected into the backward TE1 mode, while the conversion between two modes with the same symmetry is prohibited. A similar behavior can be observed in the TM band diagram, although it is comparatively weaker due to the gentler confinement of TM modes. The influence of the bridge element width is depicted in Fig. 2(b), which displays the upper and lower boundaries of the TE-polarized bandgap, as well as the threshold wavelength of the TM0 mode, as the width of the bridge element gradually increases from 0 to 0.8 µm. It can be observed that when the bridge is narrow, the TM threshold is lower than the upper boundary of the TE bandgap, resulting in a limited useful operation window between the TM threshold and the lower boundary of the TE bandgap. As the bridge width linearly increases, the TE bandgap narrows accordingly, resembling the behavior of narrowband filters that utilize waveguide Bragg gratings or side-wall Bragg gratings. In this study, we choose ${\textrm{W}_2}$ to be 0.1 µm, striking a balance between fabrication feasibility and achieving the maximum useful bandgap.
Fig. 2. (a) Layout of grating unit cell with bridge element, (b) The upper and lower boundary of the TE-polarized bandgap (represented by blue triangles) and the cutoff wavelength of TM polarization (represented by purple dots) versus width of bridge element ${\textrm{W}_{\textrm{bridge}}}$, (c) Band diagram of TE polarization when the width of bridge element is set as ${\textrm{W}_{\textrm{bridge}}} = 0{\;\ \mathrm{\mu} \mathrm{m}}$, (d) Band diagram of TM polarization when the width of bridge element is set as ${\textrm{W}_{\textrm{bridge}}} = 0{\;\ \mathrm{\mu} \mathrm{m}}$.
2.2 Optimization of the geometries
Considering the characteristics of the TE0 and TE1 mode interaction, maximizing the bandgap of the grating unit cell is crucial for achieving broadband polarizers. There are two parameters that have influences on the position and width of the bandgap: Waveguide width ${\textrm{W}_{\textrm{unit}}}$ and pitch $\mathrm{\Lambda }$. We swept various values of waveguide width ${\textrm{W}_{\textrm{unit}}}$ and pitch $\mathrm{\Lambda }$ in the absence of bridge element. The range for pitch sweeping was set from 330 nm to 420 nm (only represented from 330 nm to 380 nm), with intervals of 10 nm, to satisfy the Bragg condition ${\mathrm{\Lambda }_{\textrm{Bragg}}} = \textrm{}{\mathrm{\lambda }_0}/({2{\textrm{n}_{\textrm{eff}}}} )$, where ${\mathrm{\lambda }_0}$ represents the operating wavelength in vacuum. Additionally, the range for waveguide width sweeping was chosen from 0.7 µm to 1.7 µm, with intervals of 0.1 µm. That’s because multimode Bragg gratings are desired, and wider waveguide widths would require longer tapers, resulting in higher losses. The results of the sweep are presented in Fig. 3. To demonstrate the changes in bandwidth, we plot the upper and lower boundaries of the TE polarization bandgap (represented by blue triangles) and the upper threshold of the fundamental TM0 mode (represented by purple dots). For the TE0 mode, wavelengths within the bandgap of TE polarization (between the upper and lower boundaries) will be reflected. For the TM0 mode, wavelengths longer than the cutoff wavelength will be supported. As depicted in Fig. 3, there is a trade-off between the width, position of the TE bandgap, and the cutoff wavelength of the TM0 mode. With an increase in pitch or waveguide width, the bandgap shifts downward (red shift) and becomes broader. To cover the commonly used optical communication bands (represented by the shadowed area, 1260 - 1565 nm), we selected a grating unit cell with parameters of ${\textrm{W}_{\textrm{unit}}} = 1.5{\;\ \mathrm{\mu} \mathrm{m}}$ and Λ = 360 nm (indicated by the red dashed box in Fig. 3(d)). This choice allows for coverage of the O-band and C-band while achieving a relatively broad bandgap (1260 - 1580 nm). Therefore, ${\textrm{W}_3}$ is set to be 1.5 µm and ${\mathrm{\Lambda }_{\textrm{max}}}$ is fixed at 360 nm. It is worth noting that if we neglect the impact of the bridge element and set the condition of ${\mathrm{\Lambda }_{\textrm{min}}} = \textrm{}{\mathrm{\Lambda }_{\textrm{max}}}$, as the width of the grating unit cell decreases, the device can reflect shorter wavelength, thereby widening the reflection band of TE polarization.
Fig. 3. The upper and lower boundary of the TE-polarized bandgap (represented by blue triangles) and the cutoff wavelength of TM polarization (represented by purple dots) versus waveguide width W in the absence of bridge element under different pitch $\mathrm{\Lambda }$: (a) $\mathrm{\Lambda } = 330\textrm{nm}$, (b) $\mathrm{\Lambda } = 340\textrm{nm}$, (c) $\mathrm{\Lambda } = 350\textrm{nm}$, (d) $\mathrm{\Lambda } = 360\textrm{nm}$, (e) $\mathrm{\Lambda } = 370\textrm{nm}$, (f) $\mathrm{\Lambda } = 380\textrm{nm}$.
It is important to highlight that our device differs significantly from the polarizers presented in [28–30,32] due to the exclusion of the center grating part. This exclusion is based on two inextricable reasons. Firstly, we consider the issue of loss. The effective index disparity between the mode in conventional straight waveguides and the Floquet–Bloch mode in Bragg gratings is substantial. Consequently, long tapered structures are required to achieve an adiabatic transition with low loss. Secondly, the long taper already exhibits excellent performance in blocking TE polarization, providing a high polarization extinction ratio. Therefore, there is no need to incorporate an additional center grating part, which would only result in redundant device footprint.
To quantify the performance of the whole device, the insertion loss is defined as:
The polarization extinction ratio is defined as:
The back-reflection to the fundamental and the first order TE mode are defined as:
where ${\textrm{T}_{\textrm{TXn} \pm }}$ refers to the transmission of modes as TX denotes TE/TM polarization, n indicates the order of modes and subscript ${\pm} $ represents the forward and backward propagation direction.We analyze the overall performance of the polarizer by varying the number of periods (N) to be set at 10, 20, 30, and 40. These values correspond to the device lengths of 7.2 µm, 14.4 µm, 21.6 µm, and 28.8 µm, respectively. The other conditions, namely ${\textrm{W}_1} = 0.35{\;\ \mathrm{\mu} \mathrm{m}}$, ${\textrm{W}_2} = 0.1{\;\ \mathrm{\mu} \mathrm{m}}$, ${\textrm{W}_3} = 1.5{\;\ \mathrm{\mu} \mathrm{m}}$ and ${\mathrm{\Lambda }_{\textrm{min}}} = \textrm{}{\mathrm{\Lambda }_{\textrm{max}}} = 360\textrm{nm}$, remain unchanged. The results, as shown in Fig. 4, demonstrate notable improvements in terms of IL, PER, BR0, and BR1 as N increases from 10 to 30. However, only minor enhancements are observed when N increases from 30 to 40. In particular, when it comes to the performance of the polarization extinction ratio, N = 30 exhibited a higher PER but with a slightly narrower bandwidth. Hence, we select N = 30, resulting in a bandwidth for PER > 60 dB of 284 nm. It is important to note that due to the presence of the bridge element, the operational windows of the polarizers are red-shifted compared to the predicted values. Additionally, there are significant oscillations in insertion loss as the wavelength approaches the TM threshold, which can be detrimental to the overall bandwidth.
Fig. 4. Simulation performances in the conditions of ${\textrm{W}_1} = 0.35{\;\ \mathrm{\mu} \mathrm{m}}$, ${\textrm{W}_2} = 0.1{\;\ \mathrm{\mu} \mathrm{m}}$, ${\textrm{W}_3} = 1.5{\;\ \mathrm{\mu} \mathrm{m}}$ and ${\mathrm{\Lambda }_{\textrm{min}}} = \textrm{}{\mathrm{\Lambda }_{\textrm{max}}} = 360\textrm{nm}$ with varying numbers of periods N set at 10, 20, 30, and 40 in terms of: (a) Insertion loss, (b) polarization extinction ratio, (c) back reflection into TE0, (d) back reflection into TE1.
2.3 Further improvement by introducing chirped period
To mitigate the adverse ripples near the TM threshold wavelength and achieve the desired operation window, we introduce chirped anti-symmetric Bragg gratings, where the pitches increase linearly from ${\mathrm{\Lambda }_{\textrm{min}}}$ to ${\mathrm{\Lambda }_{\textrm{max}}}$ over the number of periods. In this case, we set ${\textrm{W}_1} = 0.35{\;\ \mathrm{\mu} \mathrm{m}}$, ${\textrm{W}_2} = 0.1{\;\ \mathrm{\mu} \mathrm{m}}$, ${\textrm{W}_3} = 1.5{\;\ \mathrm{\mu} \mathrm{m}}$, ${\mathrm{\Lambda }_{\textrm{max}}} = 360\textrm{nm}$ and $\textrm{N} = \textrm{}30$, and vary ${\mathrm{\Lambda }_{\textrm{min}}}$ from 280 nm to 340 nm with an interval of 20 nm. The results are presented in Fig. 5. As observed, with an increase in the chirp rate, the entire operation window shifts to the anticipated range. In terms of insertion loss, even a small chirp rate can significantly improve the undesired oscillations near the TM0 cutoff wavelength without introducing excess loss. For the polarization extinction ratio, the introduced chirp may slightly degrade the PER, although it still remains at a high level (PER > 60 dB). Therefore, for our final design, we select ${\mathrm{\Lambda }_{\textrm{min}}} = 340\textrm{nm}$. By incorporating the chirped anti-symmetric Bragg gratings with ${\mathrm{\Lambda }_{\textrm{min}}} = 340\textrm{nm}$, we effectively eliminate the adverse ripples near the TM threshold wavelength, achieve the desired operation window, and maintain excellent performance in terms of insertion loss and polarization extinction ratio.
Fig. 5. Simulation performances in the conditions of ${\textrm{W}_1} = 0.35{\;\ \mathrm{\mu} \mathrm{m}}$, ${\textrm{W}_2} = 0.1{\;\ \mathrm{\mu} \mathrm{m}}$, ${\textrm{W}_3} = 1.5{\;\ \mathrm{\mu} \mathrm{m}}$, ${\mathrm{\Lambda }_{\textrm{max}}} = 360\textrm{nm}$ and $\textrm{N} = \textrm{}30$ with varying minimum pitches (${\mathrm{\Lambda }_{\textrm{min}}}$) set at 280 nm, 300 nm, 320 nm, 340 nm and 360 nm in terms of: (a) Insertion loss, (b) Polarization extinction ratio, (c) Back reflection into TE0, (d) Back reflection into TE1.
The performance of the polarizer corresponding to the specified parameters is shown in Fig. 6(a). The gray-shaded regions, alternating with the blank regions, represent all the optical communication bands (O-, E-, S-, C-, L-, U-bands). As observed, the polarization extinction ratio exceeds 60 dB over a wide bandwidth of 280 nm (1290 - 1570 nm), while the insertion loss remains below 0.11 dB. This achievement represents the highest PER attained on the SOI platform to date. The majority of TE-polarized light is reflected into the TE1 mode (BR1 > -1.9 dB), while the reflection into the TE0 mode is significantly low (BR0 < -18 dB). If we consider a moderately easy condition of PER > 20 dB, IL < 0.11 dB, BR0 < -17 dB, a broader bandwidth can be obtained (BW: 325 nm, 1285–1610 nm). Our device has a total length of 21 µm. To eliminate the reflected TE1 mode, a simple single-mode curve can be utilized, which is commonly used for on-chip routing purposes. Figure 6(b) illustrates the electric distribution profiles for TE and TM polarizations at wavelengths of 1310 nm and 1550 nm. As predicted, when the TE0 mode is excited at the input, the polarizer predominantly reflects the TE0 mode into the TE1 mode with minimal transmission. On the other hand, when the TM0 mode is launched, it can propagate through the device with negligible loss. Overall, the demonstrated polarizer exhibits exceptional performance in terms of polarization extinction ratio, insertion loss, and back reflection. In the meantime, it covers a wide bandwidth suitable for most optical communication bands and can be effectively utilized for on-chip polarization division multiplexing (PDM) systems.
Fig. 6. (a) Simulation performance in the conditions of ${\textrm{W}_1} = 0.35\textrm{um}$, ${\textrm{W}_2} = 0.1\textrm{um}$, ${\textrm{W}_3} = 1.5\textrm{um}$, ${\mathrm{\Lambda }_{\textrm{min}}} = 340\textrm{nm}$, ${\mathrm{\Lambda }_{\textrm{max}}} = 360\textrm{nm}$ and $\textrm{N} = \textrm{}30$, (b) Calculated transmission behavior at wavelength of 1310 nm and 1550 nm.
2.4 Fabrication tolerance
Regarding fabrication tolerance, we analyze the impact of the most critical parameters of our polarizer: the duty cycle ($\textrm{DC} = \textrm{a}/\mathrm{\Lambda }$) of the grating structure and the width of the bridge element. We assume physical variations of ${\pm} 20\textrm{nm}$ in the segment width a (as depicted in Fig. 1(b)) and the bridge width ${\textrm{W}_{\textrm{bridge}}}$ (as depicted in Fig. 2(a)). The 3D FDTD simulation results are illustrated in Fig. 7. From the results, it can be observed that fabrication errors have minimal impact on the insertion loss and back reflections into TE0 and TE1 modes, as IL remains below 0.1 dB and BR0 remains below -17 dB. The polarization extinction ratio is affected to some extent. In the case of $\mathrm{\Delta a\;\ } = \textrm{} - 20\textrm{nm}$, PER is marginally increased but blue-shifts by around 30 nm, resulting in a narrower bandwidth (291 nm for PER > 20 dB). Conversely, in the case of $\mathrm{\Delta a\;\ } = \textrm{} + 20\textrm{nm}$, PER is slightly decreased but still maintains at a high level, and the bandwidth remains almost unchanged. However, the TE blocking window red-shifts by around 20 nm. In the cases of ${\textrm{W}_{\textrm{bridge}}} = \textrm{} \pm 20\textrm{ nm}$, it seems that these kinds of variations do not have significant impact on the performance of the whole device, as PER slightly changes but bandwidth remains at the same level. In terms of the influence brought by the waveguide width ${\textrm{W}_3}$, we can infer from Fig. 3 that any narrower width ${\textrm{W}_3}$ will lead to blue shift of operation band, especially for the right boundary of the TE blocking window, which will result in narrower bandwidth. It is important to note that our device exhibits a higher level of fabrication-friendliness compared to the polarizer in [34]. This is because our grating segments do not have minor structures such as triangle shapes, which can lead to incomplete filling of the gaps with the top cladding, causing potential fabrication challenges.
Fig. 7. Simulation fabrication tolerance tests considering grating segment width variations of (a) $\mathrm{\Delta a\;\ } = \textrm{} - 20\textrm{nm}$ and (b) $\mathrm{\Delta a\;\ } = \textrm{} + 20\textrm{nm}$, as well as bridge width variations of (c) ${\textrm{W}_{\textrm{bridge}}} = \textrm{} - 20\textrm{nm}$ and (d) ${\textrm{W}_{\textrm{bridge}}} = \textrm{} + 20\textrm{nm}$.
3. Fabrication and measurement results
The proposed device is fabricated on the SOI platform with a top silicon thickness of 220 nm and a buried oxide layer thickness of 3 µm. The device layout is patterned through 100 keV electron-beam lithography fabrication process provided by Advanced Electronics Materials and Devices, Shanghai Jiao Tong University. A two-step process is used, due to the fact that some of the grating couplers (GCs) are designed with two layers. Then a 3-µm thick silicon dioxide film is deposited on the top by plasma-enhanced chemical vapor deposition (PECVD).
Since single GC has a limited operation bandwidth, we fabricate four types of CGs to meet the needs of different wavelength ranges as well as different polarizations, with each type of CG being designed to cover a wavelength range of 100 nm. As shown in Fig. 8(a), in order to measure the low ILs, ten TM-pass polarizers are cascaded for the band range of 1260–1360 nm because of the relatively high IL shown by simulations whereas thirty polarizers are cascaded for the band range of 1500–1600 nm. Single-mode S-bends are utilized to filter out the reflected TE1 mode thus the need for isolator after light source is eliminated. Straight single-mode waveguides are also fabricated on the same chip for normalization. The scanning electron microscopy (SEM) image of the device is displayed in Fig. 8(b), and a zoom-in picture with more details is depicted in Fig. 8(c).
Fig. 8. (a) Microscopic image of fabricated devices for measurement. (b) Scanning electron microscopy (SEM) image of TM-pass polarizer. (c) More detailed information about the actual parameters.
For test purposes, the vertical coupling test system (Apico Technology MCR-P8) is utilized to input and output light. The test system also includes the Santec TSL-500 O-band and C-band tunable semiconductor lasers, the Santec MPM-210 H multi-channel power meter, and the Santec PCU-100 polarization control unit. Limited by the wavelength range of the tunable lasers, we obtain the transmission spectra of both TE and TM modes within the ranges of 1260–1360 nm and 1500–1600 nm. The coupling angle of all GCs is around 12° during the experimental measurement.
The measured IL and PER spectra of our device are shown in Fig. 9. It should be emphasized that the test results are for the design before further optimization, which means that the period is constant in the actual device. Therefore, we can see the oscillations as the wavelength approaches the TM threshold in Fig. 9(c), which are consistent with the simulation results shown in Fig. 4(a). Furthermore, from the measured information in Fig. 8(c) we know that our device is over-etched, with $\mathrm{\Delta a} \approx{-} 13\textrm{nm}$ and ${\textrm{W}_3} \approx{-} 60\textrm{nm}$. Thus, we anticipate a blue shift of about 40 nm in the actual product, which is also in accordance with the test results. Despite these fabrication errors, our device still shows great potential of achieving high PER while maintaining low IL. As reported in Fig. 9(c) and (d), for wavelength within range 1275–1360 nm and 1500–1600 nm, the maximum measured IL is 0.74 dB with the average IL within these ranges being as low as 0.27 dB. The two unexpected notches are caused by the fact that the cascade of devices forms a series of Fabry-Perot cavities [30]. Therefore, one can estimate the actual single-device IL more optimistically around those maxima. As for the measured PER, it can be seen that the wavelength ranges for PER > 20 dB are 1275–1360 nm and 1500–1535 nm, while the wavelength ranges for PER > 40 dB are 1275–1360 nm and 1500–1523 nm. As a matter of fact, the results do not show PER as high as 70 dB, because the ILs of the grating couplers used are relatively high and the power meter lacks the ability to detect light less than -80 dB. Still, among all TM-pass polarizers based on reflection mechanism, our polarizer stands out with competitive IL and PER. We believe the bandwidth can be further improved in the actual devices by introducing chirped period as our simulation results predict.
Fig. 9. Measured PER spectra from (a) 1260–1360 nm, (b) 1500–1600 nm, and measured IL spectra from (c) 1260–1360 nm, (d) 1500–1600 nm.
4. Conclusion
A performance comparison of all-silicon TM-pass polarizers is given in Table 1. Despite the missing test wavelength range of 1360–1500 nm, experimental results still show competitive performance of the proposed polarizer. Measurement results indicate IL < 0.74 dB and PER > 40 dB over the wavelength range of 1275–1360 nm and 1500–1523 nm with the average IL falling in these ranges as low as 0.27 dB. Additionally, simulation results suggest that the performance can be further improved by introducing chirp in the period of Bragg gratings, thus achieving IL < 0.11 dB and PER > 60 dB over a wide range of 280 nm (1290–1570 nm). In conclusion, we have successfully proposed and experimentally demonstrated a broadband, low insertion loss, and ultra-high polarization extinction ratio on-chip TM-pass polarizer utilizing anti-symmetric Bragg gratings. Our device stands out from other reflection-based polarizers by maximizing the coupling between the forward TE0 mode and the backward TE1 mode through the reduction of the bridge element width, resulting in high PER and a wide reflection band. Additionally, the optimized tapered structure of the device ensures low loss performance. We believe that our proposed TM-pass polarizer will find significant applications in various areas, including optical communication and optical sensing.
Table 1. Performance comparison of reported SOI-based TM-pass polarizers
Funding
Strategic Priority Research Program of Chinese Academy of Sciences (XDB43000000); National Natural Science Foundation of China (61975198); National Science Fund for Distinguished Young Scholars (61825504); National Key Research and Development Program of China (2022YFB2804202).
Disclosures
The authors declare no conflicts of interest.
Data availability
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.
References
1. D. Dai and J. E. Bowers, “Silicon-based on-chip multiplexing technologies and devices for Peta-bit optical interconnects,” Nanophotonics 3(4-5), 283–311 (2014). [CrossRef]
2. H. Xu, D. Dai, and Y. Shi, “Anisotropic metamaterial-assisted all-silicon polarizer with 415-nm bandwidth,” Photonics Res. 7(12), 1432–1439 (2019). [CrossRef]
3. W. Liu, S. Zhao, H. Li, et al., “Ultra-high performance all-silicon TM polarizer covering O-U optical communication bands,” J. Lightwave Technol. 40(22), 7326–7332 (2022). [CrossRef]
4. H. Xu, Y. Qin, G. Hu, et al., “Meta-structured silicon nanophotonic polarization beam splitter with an optical bandwidth of 415 nm,” Laser Photonics Rev. 17(4), 2200550 (2023). [CrossRef]
5. S. Yin, H. Qiu, Z. Wang, et al., “On-chip silicon switchable polarization beam splitter,” Opt. Lett. 47(4), 961–964 (2022). [CrossRef]
6. H.-C. Chung, G.-X. Lu, and S.-Y. Tseng, “Shortcut to adiabaticity in a silicon polarization splitter rotator using multi-wavelength adiabaticity engineering,” Opt. Express 30(5), 8115–8125 (2022). [CrossRef]
7. M. Borhan Mia, N. Jaidye, I. Ahmed, et al., “Broadband integrated polarization splitter and rotator using subwavelength grating claddings,” Opt. Express 31(3), 4140–4151 (2023). [CrossRef]
8. H. Xu and Y. Shi, “Subwavelength-grating-assisted silicon polarization rotator covering all optical communication bands,” Opt. Express 27(4), 5588–5597 (2019). [CrossRef]
9. M. Gilo and K. Rabinovitch, “Design parameters of thin-film cubic-type polarizers for high-power lasers,” Appl. Opt. 26(13), 2518–2521 (1987). [CrossRef]
10. J. Wang, Y. Zhao, I. Agha, et al., “SU-8 nanoimprint fabrication of wire-grid polarizers using deep-UV interference lithography,” Opt. Lett. 40(19), 4396–4399 (2015). [CrossRef]
11. M. Z. Alam, J. S. Aitchison, and M. Mojahedi, “Compact and silicon-on-insulator-compatible hybrid plasmonic TE-pass polarizer,” Opt. Lett. 37(1), 55–57 (2012). [CrossRef]
12. T. K. Ng, M. Z. M. Khan, A. Al-Jabr, et al., “Analysis of CMOS compatible Cu-based TM-pass optical polarizer,” IEEE Photon. Technol. Lett. 24(9), 724–726 (2012). [CrossRef]
13. Y. Huang, S. Zhu, H. Zhang, et al., “CMOS compatible horizontal nanoplasmonic slot waveguides TE-pass polarizer on silicon-on-insulator platform,” Opt. Express 21(10), 12790–12796 (2013). [CrossRef]
14. X. Hu and J. Wang, “Ultrabroadband Compact Graphene-Silicon TM-Pass Polarizer,” IEEE Photonics J. 9(2), 1–10 (2017). [CrossRef]
15. W. S. Chong, S. X. Gan, C. K. Lai, et al., “Configurable TE- and TM-pass Graphene oxide-coated waveguide polarizer,” IEEE Photon. Technol. Lett. 32(11), 627–630 (2020). [CrossRef]
16. X. Sun, M. Mojahedi, and J. S. Aitchison, “Hybrid plasmonic waveguide-based ultra-low insertion loss transverse electric-pass polarizer,” Opt. Lett. 41(17), 4020–4023 (2016). [CrossRef]
17. B. Bai, L. Liu, R. Chen, et al., “Low Loss, Compact TM-pass polarizer based on hybrid plasmonic grating,” IEEE Photon. Technol. Lett. 29(7), 607–610 (2017). [CrossRef]
18. R. Chen, B. Bai, and Z. Zhou, “Low-loss hybrid plasmonic TM-pass polarizer using polarization-dependent mode conversion,” Photon. Res. 8(7), 1197–1202 (2020). [CrossRef]
19. D. Dai, Z. Wang, N. Julian, et al., “Compact broadband polarizer based on shallowly-etched silicon-on-insulator ridge optical waveguides,” Opt. Express 18(26), 27404–27415 (2010). [CrossRef]
20. S. I. H. Azzam, M. F. O. Hameed, N. F. F. Areed, et al., “Proposal of an ultracompact CMOS-compatible TE-/TM-pass polarizer based on SOI Platform,” IEEE Photon. Technol. Lett. 26(16), 1633–1636 (2014). [CrossRef]
21. S. I. Azzam and S. S. A. Obayya, “Ultra-compact resonant tunneling-based TE-pass and TM-pass polarizers for SOI platform,” Opt. Lett. 40(6), 1061–1064 (2015). [CrossRef]
22. W. Liu, D. Dai, and Y. Shi, “High-performance all-silicon polarizer with 415 nm bandwidth,” Opt. Lett. 46(6), 1321–1324 (2021). [CrossRef]
23. B. Paredes, H. Zafar, M. S. Dahlem, et al., “Silicon Photonic TE Polarizer Using Adiabatic Waveguide Bends,” in 2016 21st OptoElectronics and Communications Conference (OECC) held jointly with 2016 international conference on Photonics in Switching (PS), (IEEE, 2016), pp. 1–3.
24. H. Zafar, P. Moreira, A. M. Taha, et al., “Compact silicon TE-pass polarizer using adiabatically-bent fully-etched waveguides,” Opt. Express 26(24), 31850–31860 (2018). [CrossRef]
25. Z. Lin, X. Li, and S. He, “High-performance silicon TE-pass polarizer assisted by anisotropic metamaterials,” Opt. Express 30(14), 24841–24851 (2022). [CrossRef]
26. H. Xu and Y. Shi, “On-chip silicon TE-pass polarizer based on asymmetrical directional couplers,” IEEE Photon. Technol. Lett. 29(11), 861–864 (2017). [CrossRef]
27. H. N. Xu, D. X. Dai, and Y. C. Shi, “Fabrication-friendly on-chip silicon polarizer based on polarization-selective corner mirrors,” IEEE Photon. Technol. Lett. 33(13), 652–655 (2021). [CrossRef]
28. H. Zafar, M. Odeh, A. Khilo, et al., “Low-Loss Broadband Silicon TM-Pass Polarizer Based on Periodically Structured Waveguides,” IEEE Photon. Technol. Lett. 32(17), 1029–1032 (2020). [CrossRef]
29. X. Guan, P. Chen, S. Chen, et al., “Low-loss ultracompact transverse-magnetic-pass polarizer with a silicon subwavelength grating waveguide,” Opt. Lett. 39(15), 4514–4517 (2014). [CrossRef]
30. S. Wu, Z. Guo, T. Feng, et al., “Compact and ultra-broadband all-silicon TM-pass and TE-reflected polarizer using grating based weakly coupled nanowires,” Opt. Express 30(17), 29844–29855 (2022). [CrossRef]
31. J. Zhang, L. Xu, D. Mao, et al., “All-silicon multi-band TM-pass polarizer on a 220 nm SOI enabled by multiplexing grating regimes,” Opt. Express 30(1), 326–335 (2022). [CrossRef]
32. Y. Dong, Y. Liu, Y. Xu, et al., “An Ultra-Broadband Design of TM-Pass/TE-Stop Polarizer Based on Multistage Bragg Gratings,” Photonics 9(6), 409 (2022). [CrossRef]
33. D. W. Kim, M. H. Lee, Y. Kim, et al., “Ultracompact transverse magnetic mode-pass filter based on one-dimensional photonic crystals with subwavelength structures,” Opt. Express 24(19), 21560–21565 (2016). [CrossRef]
34. M. Barona-Ruiz, C. Perez-Armenta, A. Ortega-Monux, et al., “Broadband and low-loss TM-pass polarizer using tilted subwavelength structures,” Opt. Express 30(21), 38930–38937 (2022). [CrossRef]
35. R. Cheng, W. Wang, R. Sun, et al., “Broadband, compact and reflection-less silicon polarizer and polarization beam splitter using chirped anti-symmetric multimode nanobeams,” Opt. Express 31(12), 19347–19361 (2023). [CrossRef]
36. S. Wu, J. Hao, Z. Zhao, et al., “Low loss and high extinction ratio all-silicon TM-pass polarizer with reflection removal enabled by contra-mode conversion Bragg-gratings,” Opt. Express 29(17), 27640–27652 (2021). [CrossRef]
37. H. Qiu, J. Jiang, T. Hu, et al., “Silicon add-drop filter based on multimode Bragg sidewall gratings and adiabatic couplers,” J. Lightwave Technol. 35(9), 1705–1709 (2017). [CrossRef]
38. D. Oser, F. Mazeas, X. Le Roux, et al., “Coherency-Broken Bragg Filters: Overcoming On-Chip Rejection Limitations,” Laser Photonics Rev. 13(8), 1800226 (2019). [CrossRef]
