13
Silicon Photonic Switches

Nadir Ali, Mohammad Faraz Abdullah, and Rajesh Kumar

Department of Physics, Indian Institute of Technology Roorkee, Roorkee, India

13.1 Introduction

A switch is a crucial component of optical networks in long‐haul fiber‐optic communication and data centers. A major function of the switch is routing signals from multiple input ports to multiple output ports. Over the years, the exponential rise in data traffic over optical communication networks has resulted in unsustainable energy usage and complexity in the network. The use of electrical switches in optical networks is energy‐inefficient and a major source of heat generation in data centers. Photonic switches offer numerous advantages over electrical switches, including higher energy efficiency, faster switching response time, larger bandwidth, and minimal heat generation. Silicon photonics has emerged as the preferred technology platform for integrated, reconfigurable, and low‐cost photonic switching devices due to silicon’s excellent material properties and existing mature fabrication process.

In this chapter, we will cover some of the emerging switching technologies based on silicon photonics. The switching technologies covered in this chapter are organized as follows. Section 13.2 lists the crucial performance parameters of switching devices. Section 13.3 presents an introduction and motivation for the silicon photonics platform for switching applications. Section 13.4 describes the associated physical principles for enabling switching functionalities in silicon waveguides. Major photonic switch configurations based on a directional coupler, microring resonator, Mach–Zehnder interferometer, and micro‐electro‐mechanical‐systems are presented in Section 13.5. In Section 13.6, the hybrid switch configurations based on III/V materials, 2D materials, and phase change materials are presented. Section 13.7 presents switch fabrics realized using silicon photonic waveguide components. Finally, the chapter is summarized in Section 13.8.

13.2 Performance Parameters

Switch performance is evaluated using several performance parameters. These parameters depend on the underlying physical principles and material system used for switching operation. A switch with an optimum trade‐off between the different parameters can be selected considering the application requirements. Some of the crucial performance parameters of a photonic switch are presented below [1].

  • Insertion Loss (IL): IL is the fraction of the input power lost due to the placement of switch in the network. The switch IL is calculated as the ratio of the power at its output port (Po) to the power at its input port (Pi) and is defined as
(13.1)upper I upper L left-parenthesis d upper B right-parenthesis equals minus 10 l o g left-parenthesis StartFraction upper P Subscript o Baseline Over upper P Subscript i Baseline EndFraction right-parenthesis

The value of IL should be as low as possible.

  • Extinction Ratio (ER): The switch states can be distinguished either with bar and cross or alternatively with on and off. The ER is the ratio of the output power in the on‐state (Pon) to the output power in the off‐state (Poff) and can be written as
(13.2)upper E upper R left-parenthesis d upper B right-parenthesis equals 10 l o g left-parenthesis StartFraction upper P Subscript o n Baseline Over upper P Subscript o f f Baseline EndFraction right-parenthesis

It defines the contrast between the output states of the switch and should be as high as possible.

  • Cross Talk (CT): This is the ratio of the power at a specific output from the desired input to the power from all other inputs.
  • Switching Time: This is the time required to switch the input signal to the desired output port.

13.3 Silicon Photonic Platform

Silicon photonics has been extensively used for various photonic applications since its inception in the 1980s [2]. The main drive for silicon as an integrating platform is its compatibility with the mature complementary‐metal‐oxide‐semiconductor (CMOS) manufacturing process and other desirable features such as low power consumption, low‐loss propagation of optical wave, and low cost [3]. Silicon photonic platforms are of various material systems such as silicon‐on‐insulator (SOI), SiN, Ge‐on‐Si, silicon‐on‐sapphire, and native CMOS etc. Out of these materials systems, the SOI platform has been widely used for photonic applications including switching devices.

The material system of SOI used for making waveguides for photonic device applications is shown in Figure 13.1. The SOI waveguides are formed by crystalline silicon layers placed on top of the buried silicon dioxide substrate layer. A standard SOI active and passive device is fabricated in semiconductor foundries using multi‐project wafer services, consisting of a 220 nm thick silicon layer on top of a 2 μm buried silicon dioxide layer [4] for devices operating at telecommunication wavelength of 1.55 μm. The SOI offers high refractive index contrast between silicon (n = 3.47) and silica (n = 1.45), enabling high light confinement in nanoscale waveguides. Such waveguides make it possible to integrate photonic components with high density and enable strong light‐matter interaction to realize nonlinear all‐optical effects. The switching of light in silicon waveguide structures is usually achieved by inducing refractive index modulation via various physical mechanisms. Therefore, silicon’s material properties play an important role in determining the characteristics of the switching devices. Although great effort has been put into realizing efficient switching devices using pure silicon, it is a challenging task since silicon lacks some intrinsic material properties needed for switching applications. Various other materials have been brought in to compensate for the drawbacks of silicon and enhance switching performance. Some of these materials include 2D materials, III/V semiconductors, and phase change materials.

Schematic illustration of SOI material system used for photonic waveguides: (a) slant view and (b) cross-section of a waveguide.

Figure 13.1 SOI material system used for photonic waveguides: (a) slant view and (b) cross‐section of a waveguide.

Silicon‐based photonic switches can be broadly divided into two categories: volatile and nonvolatile. The volatile switches are the ones that require a continuous power source to sustain the swicthed state. Examples include switches based on conventional effects such as thermo‐optic, electro‐optic, and carrier modulations. Nonvolatility in photonic switches implies that no holding power or static power is needed to sustain the switch state, i.e., the nonvolatile switch state is self‐sustained. Recently, these switches have been realized by exploiting the self‐holding bi‐stability of phase change materials. Such features further accelerate the potential of the hybrid silicon photonics platform by providing new directions for achieving reconfigurable and low‐power‐consuming photonic devices [5].

13.4 Physical Principles for Operation of Switches

In a switching device, a suitable mechanism is required to induce the refractive index change in the medium to change the path of light. Some of the most common refractive index modulation techniques used in silicon waveguide‐based switching devices are shown in Figure 13.2 and are described in this section.

Schematic illustration of physical mechanisms and techniques implemented for realizing the switching functionalities in SOI.

Figure 13.2 Physical mechanisms and techniques implemented for realizing the switching functionalities in SOI: (a) carrier extraction; (b) carrier injection; (c) carrier accumulation; (d) thermo‐optic effect; (e) refractive index modulation via hybridization of a silicon waveguide with other materials such as PCM.

13.4.1 Electro‐optic Effect

The electro‐optic effect is the change in the material’s refractive index under the application of an electric field. The linear electro‐optic effect or Pockels effect is the most commonly used physical mechanism in conventional switches based on LiNbO3. It is found in the materials having zincblende or wurtzite lattice symmetry. However, silicon does not exhibit the Pockels effect due to its centrosymmetric lattice structure. Although it is possible to achieve the refractive index modulation in strained silicon using Pockels effect, the change in the refractive index is minimal [6]. The quadratic effect (Kerr effect) is weak in silicon and is not usually employed in silicon since it induces small refractive index change, consumes high power, and requires a large interaction length to achieve a sizeable effect. The electro‐optic effect has the advantage of high speed (GHz range), but it is energy‐inefficient.

13.4.2 Carrier Injection/Extraction

The carrier injection/extraction mechanisms have been widely used for the refractive index modulation in silicon. In these effects, the carrier concentration of the semiconductor material is manipulated to alter the effective refractive index of the material. The carrier depletion using the PN diode is the most common mechanism to modulate the refractive index of the silicon waveguides. The PN semiconductor structure used for the carrier depletion is shown in Figure 13.2(a). A reverse‐biased PN junction alters the depletion region characteristics, changing the carrier concentration inside the junction leading to the refractive index modulation. The carrier concentration of the semiconductor can be further increased through the carrier injection mechanisms by electron‐hole injection by PIN structure, as shown in Figure 13.2(b). In carrier injection, forward‐biased PIN diodes are used. Electrons and holes are injected into the intrinsic region of silicon, leading to the change in the refractive index near the active region. The accumulation of electrons and holes at two silicon interfaces separated by a thin SiO2 film in a semiconductor‐insulator‐semiconductor (SIS) type structure is shown in Figure 13.2(c). In case of carrier accumulation, a silicon‐insulator‐silicon type structure is used to form the capacitor, which induces the opposite charges on the two sides of the insulator on applying a bias voltage and changes the carrier concentration of material. This kind of structure is difficult to fabricate and introduces large losses due to metal electrodes.

The tuning of the refractive index (Δn) and the absorption coefficient (Δα) in silicon due to the change in the carrier density (ΔNe or ΔNh) can be estimated for the 1550 nm using the equations [7]

(13.3)upper Delta n equals minus left-parenthesis 8.8 times 10 Superscript negative 22 Baseline times upper Delta upper N Subscript e Baseline plus 8.5 times 10 Superscript negative 18 Baseline times upper Delta upper N Subscript h Superscript 0.8 Baseline right-parenthesis
(13.4)upper Delta alpha equals 8.5 times 10 Superscript negative 18 Baseline times upper Delta upper N Subscript e Baseline plus 6.0 times 10 Superscript negative 18 Baseline times upper Delta upper N Subscript h Baseline period

The free holes are more effective in tuning the refractive index and less effective in absorption modulation as compared to the free electrons. These effects are fast on the scale of ps, but they require large active lengths, which leads to high insertion losses [8]. The devices based on these effects have good performance as stand‐alone devices when used for applications as switches.

13.4.3 Thermo‐optic Effect

The thermo‐optic effect is the variation in the material refractive index (Δn) due to the change in temperature (ΔT) of the material itself. The response of a material is dependent upon the thermo‐optic coefficient, dn/dT, where n and T are the refractive index and temperature, respectively. The phase shift (Δϕ) of the light traveling through the heated waveguide of length L, induced by a temperature change of ΔT, can be calculated using the equation

(13.5)upper Delta phi equals StartFraction 2 pi Over lamda EndFraction left-parenthesis StartFraction d n Subscript e f f Baseline Over d upper T EndFraction right-parenthesis upper Delta upper T upper L

where dneff/dT is the change of waveguide mode effective index with temperature and λ is the vacuum wavelength of guided light. The thermo‐optic effect has been widely used in silicon guided wave components since silicon exhibits a relatively high thermo‐optic coefficient (1.86 × 10−4 K−1) at 1.55 μm wavelength.

In practical applications, the thermo‐optic effect is utilized by heating the waveguide through a metal heater mostly placed on top of the waveguide or at the side of the waveguide as shown in Figure 13.2(d). The metal heaters are usually composed of materials such as Al, NiCr, etc. Position optimization of heater is the main task in thermo‐optic effect‐based switches, since the placement of the heater can greatly affect the optical as well as the thermal performance of the device. The thermo‐optic switches exhibit switching time of the order of few microseconds, which is adequate for some switching applications. However, it is unsuitable for modern telecommunication applications. The thermo‐optic switches enable the dense integration of the switches due to their low insertion losses. The main drawbacks of the thermo‐optic effect are high driving power and high power dissipation.

13.4.4 All‐optical Effect

All‐optical effect implies the control of one light beam by another and can be used for realizing switching devices. In all‐optical effect‐based switches, one beam causes the change in effective index while another one is used as a probe. For all‐optical switching, the interaction between the photons is realized efficiently using the nonlinear optical media, hence to a great extent, the performance of the switching device is determined by the nonlinear optical response of the optical medium. A material can exhibit various orders of non‐linearity. It is the third‐order optical nonlinearity that is exploited for the all‐optical switching in photonic devices. Therefore, for the realization of high‐performing all‐optical switches, it is essential that the optical material should possess high third‐order susceptibility and ultra‐fast response time. However, traditional semiconductor materials have small third non‐linear susceptibilities. The enhancement of the non‐linearity can be achieved using the micro‐structures, including micro‐cavities and the slow light effect.

Another issue limiting the performance of the integrated all‐optic photonic switching is the methods of material excitation. The vertical triggered method, in which the laser pulses are applied from vertically above the waveguide, is not practically applicable in the integrated photonic devices. Another method involves the on‐chip excitation, where the switching is triggered by the controlled light passing through the waveguides. This method is suitable for the integration purpose, but it suffers from the faint intensity of the propagating light and requires strong third‐order nonlinear susceptibility. Considering these, it is relatively difficult to enable the all‐optical switching with high response time, low power consumption, and device structure suitable for the high‐density integration [9]. In silicon photonics, different nonlinear effects such as the two‐photon absorption (TPA), Kerr effect, free carrier absorption, and thermo‐optic effect have been demonstrated to enable the all‐optical switching with diverse properties. In order to further improve the performance of the all‐optical switching devices, other novel designs and materials with high refractive index responses are used in addition to silicon. Some of the widely used nonlinear photonic structures are photonic crystals and nanocavities [10, 11], nonlinear dielectric microring resonators [12], and nonlinear plasmonic nanostructures [13]. A silicon microring resonator has been used to demonstrate the all‐optical switching via thermal refractive index modulation with a switching time of the order of microseconds [14]. The high switching speed of 500 ps was experimentally demonstrated using undoped crystalline silicon microrings [15]. Utilizing faster electron‐hole recombination and a shorter carrier lifetime provided by grain boundaries, further reduction in switching time to 135 ps in the polycrystalline silicon microring was demonstrated with a high extinction ratio of 10 dB [16].

13.5 Major Configurations

13.5.1 Directional Coupler

A directional coupler (DC) consists of two waveguides placed in closed proximity (see Figure 13.3). The light launched from one of the waveguides is coupled into the other waveguide when the separation between the two waveguides is small enough to enable evanescent interaction. In the case of two identical waveguides, all the optical power launched into the input waveguide is transferred to the other waveguide after traveling a distance called the coupling length. The coupled power returns to the original waveguide after traveling a distance equal to twice the coupling length. In this manner, the light can be transferred between the waveguides in a periodic manner. If length of the waveguides is fixed equal to one coupling length, the light launched from the input waveguide will exit from the other waveguide representing the cross state of the switch while it will exit from the same waveguide representing the bar state of the switch if the length of the waveguide is equal to twice the coupling length. The bar and cross states are represented by the schematic drawing in Figure 13.3(b) and (c). By inducing the refractive index change (and hence the optical phase) in one of the waveguides, light can be made to exit either from the same waveguide or the other. The operation principle of the directional coupler can be understood using the concept of normal modes. The normal modes are the two lowest‐order modes of a single system consisting of two coupled waveguides. As these two modes propagate through the waveguides, the interference between them results in the power transfer from one waveguide to the other. When the optical power is launched into the input waveguide, the two normal modes, symmetric and antisymmetric, are excited in the two‐waveguide system.

Schematic illustration of (a) Silicon photonic switch configuration based on the directional coupler, (b) Bar state, and (c) Cross state.

Figure 13.3 (a) Silicon photonic switch configuration based on the directional coupler, (b) Bar state, and (c) Cross state.

After traveling a length equal to the one coupling length, the phase of the two modes differ by π, and the modes interfere destructively in the same waveguide while interfering constructively in the other waveguide, giving the cross state of the switch. The power for the two output ports of the coupler is represented by the following equations:

(13.6)upper P Subscript 1 minus o u t Baseline equals 1 minus upper F sine squared left-parenthesis gamma z right-parenthesis
(13.7)upper P Subscript 2 minus o u t Baseline equals upper F sine squared left-parenthesis gamma z right-parenthesis
(13.8)upper L Subscript c Baseline equals StartFraction pi Over 2 gamma EndFraction
(13.9)gamma squared equals kappa squared plus upper Delta squared
(13.10)upper F equals StartFraction kappa squared Over gamma squared EndFraction equals StartFraction 1 Over 1 plus left-parenthesis StartFraction upper Delta Over kappa EndFraction right-parenthesis squared EndFraction

where κ is the coupling coefficient, Δ is the phase mismatch between the two supermodes, and F denotes the coupling efficiency. The power transfers maximally from one waveguide to the other at each Lc distance. Therefore, Lc is called cross‐coupling length or beat length. The 3‐dB coupling length L3−dB of the directional coupler, at which the transfer of the power is 50%, can be estimated as:

(13.11)upper L Subscript 3 d upper B Baseline equals StartFraction upper L Subscript c Baseline Over 2 EndFraction period

13.5.2 Microring Resonator

Microring resonators (MRRs) are good candidates for photonic switching applications owing to their compact footprint, strong resonance field enhancement, and narrowband wavelength selectivity. MRR‐based switches have a compact active length and relatively low tuning power consumption [17]. Due to their resonant behavior, MRRs can be utilized to tune the resonance wavelength by a few nm without the need for a long active waveguide to achieve a π phase shift. A MRR typically consists of a microring formed by looping around a single silicon waveguide into a ring structure and two bus waveguides placed in close proximity to the microring. When there are two bus waveguides placed closely to MRR, the resulting configuration is called add‐drop configuration and this is shown in Figure 13.4. The light that enters from the input port can leave from either drop port or through port, cf. Figure 13.4(b) and (c), depending upon whether the resonance condition is satisfied or not.

When the resonance condition given by (2πRneff = mλ, m = 1, 2, 3…) is satisfied, the light of a particular wavelength is coupled to the microring and will exit from the drop port. Otherwise, the light will exit from the through port. The transmission at the through (Tthrough) port and drop (Tdrop) port can be calculated using the following expressions [17]:

(13.12)upper T Subscript t h r o u g h Baseline equals StartFraction r 2 squared a squared minus 2 r 1 r 2 a cosine phi plus r 1 squared Over 1 minus 2 r 1 r 2 a cosine phi plus left-parenthesis r 1 r 2 a right-parenthesis squared EndFraction comma
Schematic illustration of (a) add-drop configuration of a microring resonator that can be used for switching purpose. Representation of light leaving at (b) Through port and (c) Drop port.

Figure 13.4 (a) Add‐drop configuration of a microring resonator that can be used for switching purpose. Representation of light leaving at (b) Through port and (c) Drop port.

(13.13)upper T Subscript d r o p Baseline equals StartFraction left-parenthesis 1 minus r 1 squared right-parenthesis left-parenthesis 1 minus r 2 squared right-parenthesis a Over 1 minus 2 r 1 r 2 a cosine phi plus left-parenthesis r 1 r 2 a right-parenthesis squared EndFraction comma

where r1 and r2 are the self coupling coefficients and a = exp (−α2πR) is the round trip loss, with α and R representing the absorption coefficient and radius of microring, respectively. The free spectral range (FSR) measures the separation between the two resonance wavelengths. The cavity resonances exhibit a finite linewidth called the 3‐dB bandwidth or the full‐width‐half‐maximum (FWHM). A measure of sharpness of resonances relative to their spacing (FSR) is given by the parameter finesse (ℱ):

(13.14)script upper F equals StartFraction upper F upper S upper R Over upper F upper W upper H upper M EndFraction period

Another parameter of interest is the quality factor (Q), which is the measure of sharpness of resonances relative to their central frequency.

(13.15)upper Q equals StartFraction lamda Subscript r e s o n a n c e Baseline Over upper F upper W upper H upper M EndFraction comma

By using electro‐optic, thermo‐optic, or hybridizing microring with other materials, one can achieve the switching functionalities in MRR. A large‐scale switch fabric can also be created by cascading multiple MRR [18].

13.5.3 Mach–Zehnder Interferometer

The Mach–Zehnder Interferometer (MZI) is an interferometric device consisting of two 3‐dB couplers/splitters, and a phase tuning section is often placed in one of the arms as shown in Figure 13.5. The input 3‐dB coupler splits the light into two waveguide arms while the output 3‐dB coupler acts as a combiner. The tuning section of the MZI modifies the phase of propagating light in such a way that it gives either constructive or destructive interference with light in the coupler arms, and the output signal exits from either the bar port or the cross port based on phase‐matching condition, cf. Figure 13.5(b) and (c). The output of the MZI depends on the critical 3‐dB ratio of the couplers and losses inside the two arms of the MZI. The phase change introduced by one of the arms with tuning section of length (L) can be written as Δφ = ΔβL, where Δβ is the change in the propagation constant of light due to the tuning section. The ratio of the output power (Pout) to input power (Pin) for the bar and cross states are given by the equations

Schematic illustration of (a) Silicon photonic switch configuration based on Mach–Zehnder interferometer, (b) Bar state, and (c) Cros sstate.

Figure 13.5 (a) Silicon photonic switch configuration based on Mach–Zehnder interferometer, (b) Bar state, and (c) Cross state.

(13.16)StartFraction upper P Subscript 2 minus o u t Baseline Over upper P Subscript i n Baseline EndFraction equals c o s squared left-parenthesis upper Delta beta StartFraction upper L Over 2 EndFraction right-parenthesis left-parenthesis c r o s s right-parenthesis
(13.17)StartFraction upper P Subscript 1 minus o u t Baseline Over upper P Subscript i n Baseline EndFraction equals s i n squared left-parenthesis upper Delta beta StartFraction upper L Over 2 EndFraction right-parenthesis normal left-parenthesis b a r right-parenthesis period

The phase shifter can be constructed in silicon waveguides by utilizing the thermo‐optic effect by using a heater or exploiting the carrier effects using the PIN structure. By inducing the phase difference between the two arms as 0 or π, the output can be switched from the bar state to the cross state and vice versa.

13.5.4 Micro‐Electro‐Mechanical System

Micro‐electro‐mechanical system (MEMS)‐based switches utilize the mechanical tuning method and have a stronger effect than the conventional material properties effects. They mechanically move or deform the waveguides to switch the optical signal between different output ports. MEMS‐based switches have several advantages over the switches based on the conventional methods including low power consumption, high scalability, and the fabrication process being compatible with semiconductors [19]. The switching time of MEMS‐based switches ranges from sub‐microseconds to microseconds [19]. The four principles used in the MEMS‐based switches for the manipulation of light are shown in Figure 13.6 [20].

In the first principle, the coupling between the stationary waveguides is changed using the MEMS‐actuated mirrors in the gap between the waveguides as depicted in Figure 13.6(a). In the second method, as shown in Figure 13.6(b), the direction of the waveguide itself is changed to channel the light from one waveguide to another. The third method, depicted in Figure 13.6(c), is to change the propagation of the light with the help of an external element by interacting light with it. In the fourth method, the refractive index of the waveguide itself is changed by inducing the longitudinal strain, as shown in Figure 13.6(d). Using various photonic structures, MEMS‐based switches have been realized in the past. In coupled waveguide configuration, the extinction ratio of 17 dB has been experimentally demonstrated for 1550 nm wavelength [21]. In [22], vertically coupled rib waveguides are used to construct a unit cell to reduce the device footprint and insertion loss. Using the multiple unit cells, a 64 × 64 on‐chip switch with a high extinction ratio of 60 dB and fast switching time of about 0.91 μs with a low insertion loss of about 3.7 dB was demonstrated [22].

Schematic illustration of MEMS elements used to realize reconfigurable components.

Figure 13.6 MEMS elements used to realize reconfigurable components. The reconfigurability can be achieved using principles such as (a) MEMS‐actuated mirror, (b) changing the direction of the waveguide, (c) evanescent coupling of light to an external element, and (d) inducing strain in the waveguide. Modified from [20].

13.6 Hybrid Silicon Photonic Switches

13.6.1 III‐V Materials

The III‐V materials offer the opportunity to realize compact, low‐cost, energy‐efficient switch fabrics using a hybrid III‐V/Si platform. The IIIV/Si material system leverages the low‐loss guidance of silicon photonics while keeping the low power consumption and small footprint. The hybrid integration of III‐V and silicon has been possible due to the advancement of wafer bonding techniques [23]. The bonding of III‐V materials on silicon wafers allows the fabrication with a high‐throughput process. This hybrid platform effectively combines the strengths of both platforms. With the help of III‐V material, the electro‐optic effect can be enhanced while keeping most of the light guided in the low‐loss silicon waveguides. In addition, the hybrid III‐V/Si photonic components can be fabricated using a CMOS‐compatible process and this scales easily in terms of both wafer size and integration. By using a hybrid III‐V/Si platform, the enhancement of the electro‐optic effect has been demonstrated for phase manipulation [24]. Hybrid semiconductor‐optical‐amplifier (SOA) element was demonstrated by wafer bonding of III‐V on a silicon waveguide where optical mode is mostly confined in the silicon waveguide and evanescently interacts with the III‐V layer [25]. Such a configuration restricts the modal gain since only the evanescent tail of the propagating mode is influenced by the III‐V layer. An MZI‐based switch was demonstrated on the hybrid silicon platform with a power consumption of about 1 mW [24]. However, this switch requires optical amplifiers for the large‐scale switch fabrics to compensate for the optical loss of each element.

13.6.2 2D Materials

An efficient switching device would require a simple heating mechanism providing efficient heating and be scalable at a large scale for the switching application. Recently 2D materials have attracted great attention in this regard for photonic applications due to their excellent portfolio of material properties. 2D materials have novel electronic and optical properties distinctively different from the bulk parental materials. Various emerging 2D materials include molybdenum disulfide (MoS2), tungsten diselenide (WSe2), hexagonal boron nitride (hBN), and graphene [26]. Among these, graphene has attracted great attention for integrated photonic devices and complex nanostructures due to its high scalability [26–29]. The surface of the 2D materials is naturally passivated and does not show the lattice mismatch since it has no dangling bonds. Therefore, it is easy to integrate or place 2D materials on silicon waveguides. The room temperature values of the thermal conductivity of graphene can go up to 5.30 × 103 W/mK. The ultra‐high thermal conductivity and low loss of the graphene is beneficial for the thermal management of the thermo‐optic devices [30]. Graphene as a heat conductor has been used to tune one of the arms of the MZI thermally. The graphene acts as a heat conductor and transports heat from the metal heater to the MZI arm. An spectral shift of 7 nm was shown for an applied heating power of 110 mW [27]. A microring resonator‐based device by coating the microring with the graphene has been demonstrated to tune the resonance wavelength. By electrically heating the graphene, the thermal energy generated can shift the resonance wavelength by 2.9 nm with an electrical power consumption of 28 mW. It also exhibited a modulation depth of 7 dB with switching time of 750 ns[28]. Using a graphene heater placed at a small (240 nm) distance from the silicon microring, without any light absorption, demonstrated a tuning power of 22 mW per spectral range with a response time of 3 μs [29]. By optimizing the coverage length of graphene on a microring resonator, an on‐off microring switch exhibiting an extinction ratio of 12.8 dB with a voltage of 8.8 V has been demonstrated [31].

13.6.3 Phase Change Materials

Photonic switches based on conventional methods have limitations in terms of speed, optical loss, and footprint. The electro‐optic effect‐based switches can enable high‐speed but suffer from high power consumption and high insertion losses. The thermo‐optic effect‐based switches have a slow response time of the microsecond order, making them unsuitable for modern telecommunication and data center‐related applications. In addition, these conventional effects need a continuous static power bias to sustain the device state, adding heavily to overall power consumption of the device.

To surmount the above‐mentioned limitations, high‐performance, reconfigurable switches having low static power consumption, high optical contrast, and ultrafast response time are required. Integration of phase change materials (PCMs) represents a promising way for realizing nonvolatile photonic switching davices on a SOI platform. Apart from the nonvolatility, there are many advantages of using PCM material for photonic switching applications [32]. First, PCMs exhibit a high refractive index contrast when switched from one physical state to another. Second, the transition between the phases proceeds with a fast switching time ranging from sub‐nanosecond to a few picoseconds. Third, switching between the physical states can be achieved by thermal, electrical, or optical means. Fourth, PCM can be easily deposited, with the silicon material using the CMOS‐compatible fabrication process, scaled down to the nanoscale, making it possible to realize miniaturized photonic switching devices. Fifth, a high cyclability of the devices up to 1010 has been demonstrated in the switching devices [33]. Therefore, PCMs are an ideal candidate for realizing reconfigurable hybrid PCM‐silicon photonic switching devices. Until now, extensive efforts have been made to realize non‐volatile, ultra‐compact, CMOS‐compatible photonic switching devices by employing Ge2Sb2Te5 or GST. The GST has given the most promising results and is extensively used in various photonic devices. In most devices, hybrid GST‐Si waveguides are formed by placing GST on top of the silicon waveguide, which manipulates the absorption and phase of the propagating light through evanescent interaction. The optical power is absorbed in the hybrid waveguide region and depending on the structural phase of the GST layer, the output differs. In the amorphous phase with low optical loss, the light passes through waveguides without much appreciable loss. This gives the high or the ON state of the switch. On the other hand, the crystalline phase with high optical loss absorbs most of the light and produces a low or OFF state of the switch. The switch state can be changed by switching the phase of the GST layer from amorphous to crystalline and vice versa. The crystalline state is obtained by raising the GST temperature above the crystallization temperature (∼ 150 °C) but keeping the maximum temperature below the amorphization temperature (∼ 600 °C) [34]. For amorphization, the GST temperature should be raised above the melting temperature and then rapidly quenched down to form amorphous phase.

Optical and electrical heating mechanisms have been used to induce the phase change behavior in GST‐based photonic switches. Free space laser heating was demonstrated in a hybrid GST‐Si waveguide device [35]. Improvement in the insertion losses and switching time was achieved using a silicon multimode interference waveguide with optical light focused on a circular GST cell placed on top of the silicon waveguide [36]. This heating approach was also used in the silicon microring ring resonator integrated with a GST thickness of 20 nm and 12 dB extinction ratio was obtained [37]. However, this approach is not viable for realizing practical integrated photonic swiching devices. Many researchers have used the on‐chip optical pumping technique to heat the GST‐Si waveguides e.g. [38, 39]. In this case, a pump is coupled to the waveguide (by an end‐fire or grating coupler scheme) and is made to pass through the hybrid waveguide region. In [39], a switching contrast of 12% with energy consumption of 9.5 nJ and response time of 3.8 μs was achieved using a 4 μm‐long hybrid GST cell placed on top of a rib waveguide. This on‐chip pumping approach is energy‐inefficient since only the evanescent wave of the propagating mode interacts with the GST layer placed on top of the waveguide.

Another approach [40–42], in which GST can be placed in the trench created by partially etching the silicon waveguide, is a better alternate and enhances the light‐matter interaction, enabling high optical readout contrast with a compact active volume. In a 1 × 1 switch operating at 1.55 μm, high extinction ratio of 43 dB with a moderately low insertion loss of 2.7 dB was achieved [40]. This switch was able to maintain an extinction ratio above 30 dB for a wavelength span of 1500–1600 nm. This high extinction ratio was obtained with a very compact active volume of 400 × 180 × 450 nm3 (length × height × width). The phase change analysis of the embedded GST confirmed that during the phase change process, the phase change occured in most of the GST regions. The switch states can be altered with 1.6 mW and 7.2 mW of power for the process of crystallization (ON to OFF) and amorphization (OFF to ON), respectively [40].

Electrically induced phase change in the hybrid GST‐Si waveguide has been demonstrated using Joule heating. Low‐loss indium‐tin‐oxide (ITO) material as an electrode allows for electrical heating with negligible optical losses and provided localized heating of the active region. In electrical heating, a single electrical pulse is able to induce crystallization and amorphization. The heating of GST placed on top of a silicon waveguide was demonstrated [43] with an extinction ratio of 1.2 dB. The low extinction ratio is due to the weak interaction with the optical mode and small active volume.

Taking the embedded approach 1 × 1 waveguide switch and 1 × 2 directional coupler switch for operation at 2.1 μm were also investigated. The GST phase was transformed by applying electrical pulses through ITO electrodes, which alters the device state [41]. Both amorphization and crystallization processes were studied to confirm whether the phase transition occurs in the complete GST region. In a 1 × 1 waveguide switch, an extinction ratio of 33.79 dB was obtained with 0.52 dB insertion loss for an optimized GST length of only 0.92 μm. The crystallization was induced by a 5 V pulse which corresponds to the energy consumption of 0.9 nJ, while amorphization was achieved with a 7.5 V pulse which increases the GST temperature above the melting point, and energy consumption of 62.21 nJ.

In a 1 × 2 directional coupler switch, reversible switching was achieved with an extinction ratio of 10.33 dB and 5.23 dB in the cross and bar state, respectively. The optimized coupler has an active length of only 52 μm with a coupling gap of 100 nm. The crystallization and amorphization of GST was achieved with 6 V and 7.5 V electrical pulse, respectively. The phase transition of the GST was achieved with an energy of 44.67 nJ per cycle.

Design and analysis of a 1 × 2 tunable switch based on a hybrid GST‐silicon microring resonator as shown in Figure 13.7 [42] was carried out. The overview of the devices is shown in Figure 13.7(a). The hybrid part of the microring consists of 20 nm thick GST layer placed on top of the partially etched silicon waveguide. The ITO layers of 50 nm thickness are placed on top of the GST and the side of the hybrid waveguide for electrically induced Joule heating as shown in Figure 13.7(b). The output of the microring is switched between through and drop port by electrically inducing the phase change in the embedded GST layer. Through port transmission spectrum of the device, covering one of the resonance wavelengths, is shown in Figure 13.7(c). The switch exhibited high extinction ratio of 18.75 dB at the through port as shown in Figure 13.7(c), while for the drop port (not shown here), the extinction ratio was 25.57 dB. The switch states were interchangeable by applying 5 V and 8 V pulses, and energy consumption for one switching cycle of the switch was 108.11 mW. High value of amorphous GST thermo‐optic coefficient (1.1 × 10−3 K−1) was exploited for tuning of the resonance wavelength of microring switch. Tuning curve of resonance wavelength shift as a function of applied power is shown in Figure 13.7(d). The switch exhibits a wavelength tuning efficiency of 1.16 nm/mW.For a GST length of 7 μm, we obtained a maximum wavelength tuning range of 4.63 nm [42].

Image described by caption.

Figure 13.7 (a) A 1 × 2 tunable switch designed using the phase change material embedded microring resonator. (b) Side‐view cross‐section of the microring showing the hybrid waveguide region. (c) Transmission spectrum at through port for the amorphous and crystalline phase of GST. (d) Tuning curve of switch in the amorphous phase of GST. Reprinted with permission from [42] © The Optical Society.

Schematic illustration of representation of (a) MRR- and (b) MZI-based switch fabrics.

Figure 13.8 Representation of (a) MRR‐ and (b) MZI‐based switch fabrics.

13.7 Switch Fabrics Using MRR and MZI

Several large‐scale silicon switch fabrics based on the switching building blocks have been demonstrated. The MZI and MRR are more suitable for large‐scale switch fabrics as they are highly scalable. A representative MRR‐based switch fabric is shown in Figure 13.8(a). Various types of MRR based switch fabrics with different port counts have been reported. These switch fabrics include a 5 × 5 electro‐optic switch fabric [44], a modular 8 × 8 switch fabric [45], and a 48 × 8 thermo‐optic switch fabric [46]. The MZI‐based switch fabrics are represented in Figure 13.8(b). Large‐scale switch matrices based on the MZI elements have been demonstrated in recent years. Examples include 16 × 16 electro‐optic MZI switch fabric [47], 32 × 32 electro‐optic switch fabric [48], 64 × 64 thermo‐optic switch fabric [49], and 32 × 32 thermo‐optic switch fabric [50].

13.8 Summary

The ever‐increasing demand of interconnects in data centers and telecommunication networks motivates the deployment of high‐performance switching technologies. Photonic switching has received great attention to potentially address the challenges regarding bandwidth, cost, and power consumption in data centers. In this chapter, photonics‐based switching relying on various physical effects such as electro‐optic, thermo‐optic, free carrier injection/extraction, and all‐optical effects has been presented. Major configurations of the switching devices viz. MRR, DC, MZI, and MEMS technologies are covered. The need for hybrid silicon photonics with materials such as III‐V/Si, 2D materials, and phase change materials has been presented. Furthermore, the emerging nonvolatile switches based on the phase change material are covered and discussed in detail.

Bibliography

  1. 1 G.I. Papadimitriou, C. Papazoglou, and A.S. Pomportsis. Optical switching: switch fabrics, techniques, and architectures. Journal of Lightwave Technology, 21(2):384–405, 2003.
  2. 2 R. Soref and J. Larenzo. All‐silicon active and passive guided‐wave components for λ= 1.3 and 1.6 μm. IEEE Journal of Quantum Electronics, 22(6):873–879, 1986.
  3. 3 B. Jalali and S. Fathpour. Silicon photonics. Journal of Lightwave Technology, 24(12):4600–4615, 2006.
  4. 4 D. Inniss and R. Rubenstein. Silicon Photonics: Fueling the Next Information Revolution. Morgan Kaufmann, Cambridge, MA; Singapore, 2017.
  5. 5 R. Kumar, N. Ali, and S. Singh. High performance and cmos compatible photonic switches based on phase change materials. In Optoelectronic Devices and Integration VIII, volume 11184, page 111840C. International Society for Optics and Photonics, 2019.
  6. 6 B. Chmielak, M. Waldow, C. Matheisen, C. Ripperda, J. Bolten, T. Wahlbrink, M. Nagel, F. Merget, and H. Kurz. Pockels effect based fully integrated, strained silicon electro‐optic modulator. Optics Express, 19(18):17212–17219, 2011.
  7. 7 R. Soref and B. Bennett. Electrooptical effects in silicon. IEEE Journal of Quantum Electronics, 23(1):123–129, 1987.
  8. 8 X. Xiao, H. Xu, X. Li, Z. Li, T. Chu, Y. Yu, and J. Yu. High‐speed, low‐loss silicon mach–zehnder modulators with doping optimization. Optics Express, 21(4): 4116–4125, 2013.
  9. 9 Z. Chai, X. Hu, F. Wang, X. Niu, J. Xie, and Q. Gong. Ultrafast all‐optical switching. Advanced Optical Materials, 5(7):1600665, 2017.
  10. 10 S. Fan. Sharp asymmetric line shapes in side‐coupled waveguide‐cavity systems. Applied Physics Letters, 80(6):908–910, 2002.
  11. 11 M.F. Yanik, S. Fan, and M. Soljačić. High‐contrast all‐optical bistable switching in photonic crystal microcavities. Applied Physics Letters, 83(14):2739–2741, 2003.
  12. 12 O. Wada. Femtosecond all‐optical devices for ultrafast communication and signal processing. New Journal of Physics, 6(1):183, 2004.
  13. 13 A. Pasquazi, S. Stivala, G. Assanto, V. Amendola, M. Meneghetti, M. Cucini, and D. Comoretto. In situ tuning of a photonic band gap with laser pulses. Applied Physics Letters, 93(9):091111, 2008.
  14. 14 V.R. Almeida and M. Lipson. Optical bistability on a silicon chip. Optics Letters, 29(20):2387–2389, 2004.
  15. 15 V.R. Almeida, C.A. Barrios, R.R. Panepucci, and M. Lipson. All‐optical control of light on a silicon chip. Nature, 431(7012):1081–1084, 2004.
  16. 16 K. Preston, P. Dong, B. Schmidt, and M. Lipson. High‐speed all‐optical modulation using polycrystalline silicon microring resonators. Applied Physics Letters, 92(15):151104, 2008.
  17. 17 W. Bogaerts, P. De Heyn, T. Van Vaerenbergh, K. De Vos, S.K. Selvaraja, T. Claes, P. Dumon, P. Bienstman, D. Van Thourhout, and R. Baets. Silicon microring resonators. Laser & Photonics Reviews, 6(1):47–73, 2012.
  18. 18 H. Jia, Y. Zhao, L. Zhang, Q. Chen, J. Ding, X. Fu, and L. Yang. Five‐port optical router based on silicon microring optical switches for photonic networks‐on‐chip. IEEE Photonics Technology Letters, 28(9):947–950, 2016.
  19. 19 X. Tu, C. Song, T. Huang, Z. Chen, and H. Fu. State of the art and perspectives on silicon photonic switches. Micromachines, 10(1):51, 2019.
  20. 20 F. Chollet. Devices based on co‐integrated mems actuators and optical waveguide: A review. Micromachines, 7(2):18, 2016.
  21. 21 Y. Akihama, Y. Kanamori, and K. Hane. Ultra‐small silicon waveguide coupler switch using gap‐variable mechanism. Optics Express, 19(24):23658–23663, 2011.
  22. 22 T.J. Seok, N. Quack, S. Han, R.S. Muller, and M.C. Wu. Large‐scale broadband digital silicon photonic switches with vertical adiabatic couplers. Optica, 3(1):64–70, 2016.
  23. 23 M.J.R. Heck, J.F. Bauters, M.L. Davenport, J.K. Doylend, S. Jain, G. Kurczveil, S. Srinivasan, Y. Tang, and J.E. Bowers. Hybrid silicon photonic integrated circuit technology. IEEE Journal of Selected Topics in Quantum Electronics, 19(4):6100117, 2012.
  24. 24 L. Chen, E. Hall, L. Theogarajan, and J. Bowers. Photonic switching for data center applications. IEEE Photonics Journal, 3(5):834–844, 2011.
  25. 25 H. Park, A.W. Fang, O. Cohen, R. Jones, M.J. Paniccia, and J.E. Bowers. A hybrid algainas–silicon evanescent amplifier. IEEE Photonics Technology Letters, 19(4):230–232, 2007.
  26. 26 F. Xia, H. Wang, D. Xiao, M. Dubey, and A. Ramasubramaniam. Two‐dimensional material nanophotonics. Nature Photonics, 8(12):899–907, 2014.
  27. 27 L. Yu, D. Dai, and S. He. Graphene‐based transparent flexible heat conductor for thermally tuning nanophotonic integrated devices. Applied Physics Letters, 105(25):251104, 2014.
  28. 28 S. Gan, C. Cheng, Y. Zhan, B. Huang, X. Gan, S. Li, S. Lin, X. Li, J. Zhao, H. Chen, et al. A highly efficient thermo‐optic microring modulator assisted by graphene. Nanoscale, 7(47):20249–20255, 2015.
  29. 29 D. Schall, M. Mohsin, A.A. Sagade, M. Otto, B. Chmielak, S. Suckow, A.L. Giesecke, D. Neumaier, and H. Kurz. Infrared transparent graphene heater for silicon photonic integrated circuits. Optics Express, 24(8):7871–7878, 2016.
  30. 30 A.A. Balandin, S. Ghosh, W. Bao, I. Calizo, D. Teweldebrhan, F. Miao, and C.N. Lau. Superior thermal conductivity of single‐layer graphene. Nano Letters, 8(3):902–907, 2008.
  31. 31 Y. Ding, X. Zhu, S. Xiao, H. Hu, L.H. Frandsen, N.A. Mortensen, and K. Yvind. Effective electro‐optical modulation with high extinction ratio by a graphene–silicon microring resonator. Nano Letters, 15(7):4393–4400, 2015.
  32. 32 M. Wuttig, H. Bhaskaran, and T. Taubner. Phase‐change materials for non‐volatile photonic applications. Nature Photonics, 11(8):465–476, 2017.
  33. 33 S.B. Kim, G.W. Burr, W. Kim, and S.‐W. Nam. Phase‐change memory cycling endurance. MRS Bulletin, 44(9):710–714, 2019.
  34. 34 P. Guo, A.M. Sarangan, and I. Agha. A review of germanium‐antimony‐telluride phase change materials for non‐volatile memories and optical modulators. Applied Sciences, 9(3):530, 2019.
  35. 35 Y. Ikuma, Y. Shoji, M. Kuwahara, X. Wang, K. Kintaka, H. Kawashima, D. Tanaka, and H. Tsuda. Small‐sized optical gate switch using Ge2Sb2Te5 phase‐change material integrated with silicon waveguide. Electronics Letters, 46(5):368–369, 2010.
  36. 36 D. Tanaka, Y. Shoji, M. Kuwahara, X. Wang, K. Kintaka, H. Kawashima, T. Toyosaki, Y. Ikuma, and H. Tsuda. Ultra‐small, self‐holding, optical gate switch using Ge2Sb2Te5 with a multi‐mode si waveguide. Optics Express, 20(9):10283–10294, 2012.
  37. 37 M. Rudé, J. Pello, R.E. Simpson, J. Osmond, G. Roelkens, J.J.G.M. van der Tol, and V. Pruneri. Optical switching at 1.55 μ m in silicon racetrack resonators using phase change materials. Applied Physics Letters, 103(14):141119, 2013.
  38. 38 H. Zhang, L. Zhou, J. Xu, L. Lu, J. Chen, and B.M.A. Rahman. All‐optical non‐volatile tuning of an amzi‐coupled ring resonator with gst phase‐change material. Optics Letters, 43(22):5539–5542, 2018.
  39. 39 X. Li, N. Youngblood, Z. Cheng, S.G.‐C. Carrillo, E. Gemo, W.H.P. Pernice, C.D. Wright, and H. Bhaskaran. Experimental investigation of silicon and silicon nitride platforms for phase‐change photonic in‐memory computing. Optica, 7(3):218–225, 2020.
  40. 40 N. Ali and R. Kumar. Design of a novel nanoscale high‐performance phase‐change silicon photonic switch. Photonics and Nanostructures‐Fundamentals and Applications, 32:81–85, 2018.
  41. 41 N. Ali and R. Kumar. Mid‐infrared non‐volatile silicon photonic switches using nanoscale Ge2Sb2Te5 embedded in silicon‐on‐insulator waveguides. Nanotechnology, 31(11): 115207, 2020.
  42. 42 N. Ali, R.R. Panepucci, Y. Xie, D. Dai, and R. Kumar. Electrically controlled 1× 2 tunable switch using a phase change material embedded silicon microring. Applied Optics, 60(13):3559–3568, 2021.
  43. 43 K. Kato, M. Kuwahara, H. Kawashima, T. Tsuruoka, and H. Tsuda. Current‐driven phase‐change optical gate switch using indium–tin‐oxide heater. Applied Physics Express, 10(7):072201, 2017.
  44. 44 A.W. Poon, X. Luo, F. Xu, and H. Chen. Cascaded microresonator‐based matrix switch for silicon on‐chip optical interconnection. Proceedings of the IEEE, 97(7):1216–1238, 2009.
  45. 45 D. Nikolova, D.M. Calhoun, Y. Liu, S. Rumley, A. Novack, T. Baehr‐Jones, M. Hochberg, and K. Bergman. Modular architecture for fully non‐blocking silicon photonic switch fabric. Microsystems & Nanoengineering, 3(1):1–9, 2017.
  46. 46 F. Testa, C.J. Oton, C. Kopp, J.‐M. Lee, R. Ortuno, R. Enne, S. Tondini, G. Chiaretti, A. Bianchi, P. Pintus, et al. Design and implementation of an integrated reconfigurable silicon photonics switch matrix in IRIS project. IEEE Journal of Selected Topics in Quantum Electronics, 22(6):155–168, 2016.
  47. 47 L. Lu, S. Zhao, L. Zhou, D. Li, Z. Li, M. Wang, X. Li, and J. Chen. 16× 16 non‐blocking silicon optical switch based on electro‐optic mach‐zehnder interferometers. Optics Express, 24(9):9295–9307, 2016.
  48. 48 L. Qiao, W. Tang, and T. Chu. 32× 32 silicon electro‐optic switch with built‐in monitors and balanced‐status units. Scientific Reports, 7(1):1–7, 2017.
  49. 49 L. Qiao, W. Tang, and T. Chu. Ultra‐large‐scale silicon optical switches. In 2016 IEEE 13th International Conference on Group IV Photonics (GFP), pages 1–2. IEEE, 2016.
  50. 50 P. Dumais, D.J. Goodwill, D. Celo, J. Jiang, C. Zhang, F. Zhao, X. Tu, C. Zhang, S. Yan, J. He, et al. Silicon photonic switch subsystem with 900 monolithically integrated calibration photodiodes and 64‐fiber package. Journal of Lightwave Technology, 36(2):233–238, 2017.
..................Content has been hidden....................

You can't read the all page of ebook, please click here login for view all page.
Reset
35.170.81.33