Integrated frequency comb generator with compensating optical cavity structure

Abstract

Integrated frequency comb generators with a compensating optical cavity structure are disclosed. According to certain embodiments, a collision-pulse mode-locked (CPM) laser is provided. The CPM laser includes an optical cavity including at least two gain regions. A saturable absorber is disposed between adjacent ones of the at least two gain regions. A first one of the at least two gain regions and a second one of the at least two gain regions have equal respective effective lengths. Methods of fabricating the CPM laser are also provided.

Description

[0001] Cross-references to related applications This application claims priority to U.S. Patent Application 19 / 171,991 and U.S. Patent Application 19 / 540,797, the entire contents of which are incorporated herein by reference. Background Technology

[0002] With the increasing use of artificial intelligence and / or machine learning, the amount of information transmitted between large computing resource clusters (such as graphics processing units (GPUs), central processing units (CPUs), data processing units (DPUs), etc.) is also increasing. Wavelength division multiplexing (WDM) can be used to increase the interface density between computing resource clusters. WDM is conventionally implemented using multiple distributed feedback (DFB) lasers. However, this implementation of WDM often has relatively large footprints and requires significant power consumption. Therefore, power-efficient and compact light sources are needed for use in WDM applications. Summary of the Invention

[0003] Data centers rely on fast and robust communication infrastructure. This is achieved through the use of optical interconnects, particularly between different server racks. Each physical link, using a single optical fiber, can include multiple communication channels, which are distinguished by different wavelengths in a wavelength division multiplexing (WDM) system. Such optical interconnects often include transceivers that communicate optically with each other via one or more optical fibers. A transceiver includes one or more optical signal generation components and / or one or more optical signal detection components. For example, a transceiver may include one or more semiconductor devices (e.g., vertical-cavity surface-emitting lasers (VCSELs), distributed feedback (DFB) lasers, light-emitting diodes (LEDs), modulators (such as electroabsorption modulators, intensity and / or phase modulators, and other modulators), electroabsorption modulated lasers (EMLs), etc.) configured to generate optical signals and / or one or more semiconductor devices (e.g., photodiodes) configured to detect optical signals.

[0004] The transmitter in an optical WDM transceiver module is typically based on a discrete single-wavelength laser array, such as a distributed feedback (DFB) laser. However, to reduce power consumption and complexity, these lasers can be replaced with a single-comb laser. A comb laser generates a series of discrete, uniformly spaced frequencies.

[0005] As transceivers increase their line bit rates, currently from 25 Gbit / s to 50 Gbit / s and then to 100 Gbit / s, and upgrade existing techniques such as non-return-to-zero (NRZ) modulation order to pulse amplitude modulation level 4 (PAM-4), the challenge lies in scaling the power consumption of the laser source, as these increments require improving the transmitter's signal-to-noise ratio (SNR).

[0006] For example, when iterating from NRZ to PAM-4 at the same bit rate, the SNR decreases by a factor of 3. Therefore, the power needs to be increased by a factor of 3 to maintain the same bit error rate (BER) count as before.

[0007] When iterating from PAM-4 to PAM-8 (PAM level 8), both operate at the same bit rate, requiring an increase in SNR to ensure that the PAM-8 signal achieves the same BER as PAM-4.

[0008] Supporting a sustainable relationship between SNR and BER typically requires increasing the available power supplied by the laser source, which increases costs. Therefore, the laser needs to generate more light, which in turn leads to a superlinear increase in power consumption, while also introducing heat dissipation pressures and thermal management challenges.

[0009] This is especially important for transceivers that need to be packaged in a standardized pluggable form. These transceivers have limited heat dissipation and small space, so they rely on airflow design, which limits the amount of power they can accept from the mainframe due to form factor standards.

[0010] The frequency comb generator attempts to replace a discrete array of laser sources with a single laser source. The light emitted by the laser is then split into several different beams of different wavelengths, and each of these beams is used to transmit a separate data stream.

[0011] Traditional frequency comb generators are typically built using discrete components, making them suitable for applications that allow for large footprints, such as metrology and sensing. However, such devices are bulky and do not meet the requirements of integrated photonics.

[0012] Various embodiments provide frequency comb generators with active optical cavities. In various embodiments, the frequency comb generator is a monolithic and / or on-chip frequency comb laser. This allows the frequency comb generator to be directly integrated with, for example, photonic integrated circuits. The laser pulses generated by the frequency comb generator comprise discrete and uniformly spaced spectral lines or "tooths," thereby producing a large number of spectral lines with configurable line spacing. In some embodiments, these spectral lines or teeth can be used to perform intensive WDM. The frequency comb generator is capable of producing affordable, efficient, high-power frequency comb sources with ideal mode spacing. Dense, low-power interfaces are crucial for continuing to scale AI / ML systems to interconnect large clusters (GPUs, CPUs, DPUs, etc.). Monolithic frequency comb lasers offer a smaller footprint and lower power consumption compared to multiple DFB laser implementations.

[0013] In some applications, a frequency comb generator can be coupled to an output waveguide (e.g., directly, evanescently, etc.). The output waveguide can be used to optically couple the frequency comb generator to one or more modulators, multiplexers, photonic integrated circuits (PICs), etc. For example, the frequency comb generator can be part of an interconnect used in optical communications.

[0014] One example type of frequency comb laser is the collisional pulse mode-locked (CPM) laser. A CPM laser typically includes an active optical cavity in which gain material is disposed. The gain material is divided into segments or regions, usually separated from each other by saturable absorbers.

[0015] In CPM lasers, two or more backpropagating pulses collide at specific locations within the cavity, typically in the saturable absorber or other segment / segmentation components. These interactions enhance pulse shortening and stability as the pulses mutually reinforce each other's peak intensity at the collision point. When a CPM laser is in harmonic mode-locked mode, multiple pulse pairs collide in a single round trip, effectively increasing the frequency of pulse collisions and thus improving the repetition rate. Multiple pulses are generated in the cavity such that they are evenly spaced along the round-trip path. For example, in second harmonic generation, two pairs of backpropagating pulses collide in each round trip, effectively doubling the repetition rate. However, the propagating pulses within the cavity must meet in the saturable absorber region. Conventional cleaving does not provide sufficient accuracy in matching the effective length of the optical cavity segments and / or regions to ensure efficient harmonic generation.

[0016] Various embodiments offer multiple options for fabricating CPM lasers with precisely matched cavity segments and / or regions. These options include micro-cleaving by etching cantilevered structures beneath the waveguide at desired crystal planes, and micro-cleaving at desired crystal planes using ultrasonic vibrations. This results in high-quality mirrors formed on both sides of the laser facets. Another option is to use photolithography (e.g., by etching mirror surfaces) to precisely set the length of the segments and / or regions. In yet another option, phase transition regions are added to one or both ends of one or more segments or regions. This can be achieved by adding additional electrodes that control a portion of the current in a gain region to induce a phase transition by changing the carrier density, thereby effectively altering the effective length to align it with the other side of the cavity length. One or more phase transition regions can then be used to dynamically adjust the effective length of the corresponding one or more cavity segments and / or regions by applying current to them. Applying a high current to a particular region causes gain saturation, preventing additional gain and altering the phase and effective refractive index.

[0017] In another example, a C-shaped optical cavity is formed, with the collision point / saturable absorber located at the center of the C. The arms of the C are cleaved in the same process, such that the arms are cleaved along the same crystal plane, thereby improving cleavage precision. In another option, the effective length of each region of the cavity is defined by, for example, a grating defined and / or formed using a photolithography process. The resulting CPM laser is an affordable, efficient, high-power frequency comb source that can provide spectral lines with ideal mode spacing. In particular, such a CPM laser is configured to provide region-accurate lengths to ensure that co-propagating pulses meet in the specified saturable absorber for harmonic generation, thereby producing a stable and reliable frequency comb.

[0018] Various embodiments provide optical interconnects, each of which includes a CPM laser as described in the various embodiments. Optical interconnects can be configured to connect between chips or between different communication systems. For example, it can provide optical interconnects for network interface controllers (NICs) to switches, switch-to-switch, and / or chip-to-chip. Optical interconnects can be used in various applications, such as switches, processing units (e.g., graphics processing units (GPUs), etc.). Optical interconnects can include optical links (e.g., optical fibers) for transmitting optical signals. By using the same optical link to transmit optical signals comprising multiple wavelengths, the bandwidth of the optical interconnect can be scaled. To do this, the transmitter is tuned to generate optical signals comprising multiple carrier wavelengths (e.g., possibly via a CPM laser or an array of CPM lasers). Furthermore, each modulator of the modular array can be tuned to receive and modulate a corresponding carrier frequency. Sequencing multiple wavelengths is crucial to ensuring that the transmitter and receiver correctly transmit the data contained in the optical signal.

[0019] In at least one exemplary embodiment, the optical interconnect is part of a data center corresponding to a collection of network devices, such as network switches (e.g., Ethernet switches, IP routers, multi-service platforms, various transport network elements, traditional communication equipment, or any other suitable communication system) connected to a collection of servers or computing nodes. The switch architecture is used to transmit data between switch ports. The switch architecture includes one or more interconnect circuits, which can be arranged in various switch architectures, such as... Cross switches, Banyan, Benes, Omega, Clos, multiplane, STS, TST, shared memory, buffered cross switches, any other suitable blocking or non-blocking architecture or any applicable hybrid architecture thereof. In typical embodiments, the switch architecture is implemented in hardware, which may include field-programmable gate arrays (FPGAs) and / or application-specific integrated circuits (ASICs), and in some implementations also includes bus interconnects. Data centers may follow networking topologies (e.g., hierarchical networking topologies), such as fat-tree topologies, thin-flying topologies, dragon-flying topologies, etc. Traffic is routed between network switches and servers in the data center, and at least one layer of the topology in the data center is coupled to a communication network to allow networking traffic to flow between the data center and one or more network devices.

[0020] Optical interconnects may include a substrate and electro-optical components (CPM lasers, photodiodes, etc.) supported by the substrate and configured to convert electrical signals into optical signals or vice versa. The optical interconnect may further include: a transmission block defined as a receiving surface configured to receive an optical fiber; and a waveguide configured to transmit optical signals between the electro-optical components and the receiving surface, such that in an operational configuration where the receiving surface receives the optical fiber, the electro-optical components and the optical fiber are in optical communication.

[0021] Although this disclosure illustrates and describes optical interconnects without housings or other protective enclosures, those skilled in the art will understand from this disclosure that some or all of the optical interconnects can be supported or enclosed by any housing used in the communication system to protect the components supported therein (e.g., as part of a quad small form factor (QSFP) connector, a small form factor (SFP) connector, etc.). Furthermore, the substrate carrying the electro-optical components can be substantially rectangular in shape and / or can be customized in size (e.g., customized size and shape) for use in any communication system, regardless of geometric constraints (e.g., L-shaped, square, etc.).

[0022] In various embodiments, the optical interconnect for receiving optical fiber can be implemented in a flip-chip configuration. In some embodiments, the optical interconnect is configured as a flip-chip assembly such that the longitudinal axis of a first thermally adiabatic transition profile of the optical interconnect and the longitudinal axis of a second thermally adiabatic transition profile of the optical interconnect can be collinear. In other words, in this embodiment, the orientation of the optical interconnect does not require mirrors or other reflective surfaces to redirect the optical signal between the electro-optic component and the receiving surface. However, as will be apparent to those skilled in the art according to this disclosure, the optical interconnect in a flip-chip configuration may also include one or more mirrors (e.g., reflective surfaces) to accommodate optical fibers received at different angles. In some optical interconnect embodiments, considering that light can still be confined within the waveguide at some bending radii, only the mirrors can be operatively configured to redirect the light. Therefore, embodiments of field-replaceable modular optical interconnect units configured for reception by a main switch system box are described. The field-replaceable modular optical interconnect unit includes: a housing comprising at least a front panel, a rear panel, and a side panel extending between the front and rear panels; a printed circuit board assembly supported within the housing; an optical module supported on the printed circuit board assembly, configured to convert between optical signals and corresponding electrical signals to transmit or receive optical signals via fiber optic cables; a board-to-board connector on the rear panel of the housing, configured to enable the transmission of electrical signals between the printed circuit board assembly and the main switch system box; and an external connector on the front panel of the housing, configured to engage an external optical fiber for transmitting optical signals between the optical module and external components. The field-replaceable modular optical interconnect unit can be configured to be electrically connected to the main switch system box via engagement of the board-to-board connector with a corresponding connector on the main switch system box when the housing is received by the main switch system box.

[0023] In some embodiments, the optical module may be an in-board optical module (MBOM), and / or the field-replaceable modular optical interconnect unit may include multiple external connectors. For example, the external connector may be a first external connector, and the field-replaceable modular optical interconnect unit may also include a second external connector disposed on the front panel of the housing, which is configured to transmit electrical signals between the printed circuit board assembly and external components connected thereto.

[0024] In typical optical interconnects, vertical-cavity surface-emitting laser (VCSEL) diodes serve as optical transmitters, transmitting binary or digital data over optical fibers. However, in principle, the light source can be any suitable source, and the transmitted waveform can be any suitable waveform used to transmit information. Most optical transmitters have a threshold current above which they essentially begin to emit light. Increasing the current driven by the transmitter from zero to above this threshold is time-consuming, so a bias current is typically driven by the light source. The bias current can be set just below, equal to, or above the threshold, but is often set much higher. This bias current is usually programmable, so the same circuit design can be used to drive different optical transmitters and / or for different applications. The additional time-varying current that modulates the emission of the optical transmitter is called the modulation current.

[0025] In some embodiments, the disclosed technology provides an electro-optical (EO) interconnect assembly comprising a pair of pluggable EO transceivers connected at corresponding ends of an optical fiber. EO transceivers are typically used to connect network connectivity devices (e.g., remote client switches, network adapters such as network interface controllers (NICs) and host channel adapters (HCAs)), smart NICs (NICs with embedded CPUs), network-enabled graphics processing units (GPUs), etc.). The terms "network connectivity device" and "network device" are used interchangeably herein. In some embodiments, the optical interconnect includes a substrate, one or more optical waveguides, one or more first microlenses, one or more second microlenses, and first and second mechanical fastening devices.

[0026] This disclosure describes a CPM laser comprising an optical cavity having segments or regions of substantially equal effective length. In some embodiments, the effective lengths of the segments or regions of the optical cavity differ from each other by less than 1%. In some embodiments, the effective lengths of the segments or regions of the optical cavity differ from each other by less than 0.01%. In some embodiments, the CPM laser may be integrated into an optical interconnect (e.g., as an optical signal generation component, an optical signal detection component, etc.).

[0027] According to one aspect, a method for manufacturing a collisional pulse mode-locked (CPM) laser is provided. In an exemplary embodiment, the method includes: etching at least a portion of a first gain region to define an effective length of the first gain region, wherein the first gain region is one of at least two gain regions of the cavity of the CPM laser.

[0028] In an exemplary embodiment, the first gain region is separated from the second gain region of at least two gain regions by a saturable absorber, and the effective length of the first gain region is defined as substantially equal to the effective length of the second gain region. In various embodiments, the effective length of the gain region is less than the operating distance of the optical device coupling the light.

[0029] In an exemplary embodiment, etching is performed on the entire end of the first gain region.

[0030] In an exemplary embodiment, etching is performed to define a cantilever portion at the end of a first gain region, and the method further includes applying ultrasonic vibration to cause the unetched portion of the end of the first gain region to cleave at a crystal plane corresponding to the end of the first gain region.

[0031] In an exemplary embodiment, the CPM laser includes a ridge waveguide comprising a first gain region and a waveguide structure, and the crystal planes are cleaved to define the reflecting ends of the waveguide structure.

[0032] According to another aspect, a method for manufacturing a CPM laser is provided. In an exemplary embodiment, the method includes: forming a saturated gain phase transition region at a first end of a first gain region, wherein the first gain region is one of at least two gain regions of the cavity of the CPM laser.

[0033] In an exemplary embodiment, the saturated gain phase transition region is configured to dynamically tune the effective length of the first gain region.

[0034] In an exemplary embodiment, the method further includes forming a saturable absorber, wherein the saturable absorber separates a first gain region from a second gain region in at least two gain regions.

[0035] According to another aspect, a method for manufacturing a CPM laser is provided. In an exemplary embodiment, the method includes: forming a C-shaped optical cavity including a first gain region and a second gain region, the first gain region and the second gain region being separated from each other by a saturable absorber located at the center of the curve of the C-shaped optical cavity; and cleaving both the first gain region and the second gain region in the same cleaving process.

[0036] In an exemplary embodiment, the method further includes: partially etching through the respective ends of the first gain region and the second gain region before cleaving the first gain region and the second gain region to define the respective cantilever portions at the respective ends of the first gain region and the second gain region, and the cleaving is configured to cleave away the respective cantilever portions.

[0037] In an exemplary embodiment, the first gain region and the second gain region have the same effective length.

[0038] According to another aspect, a method for manufacturing a CPM laser is provided. In an exemplary embodiment, the method further includes: forming a first grating in a first gain region of an optical cavity comprising at least two gain regions, the first grating defining an effective length of the first gain region and being configured to substantially reflect light with wavelengths within a selected wavelength range.

[0039] In an exemplary embodiment, the first grating is formed using at least one photolithography process.

[0040] In an exemplary embodiment, the method further includes forming a second grating in a second gain region of the optical cavity, the second grating defining an effective length of the second gain region and being configured to substantially reflect light with wavelengths within a selected wavelength range.

[0041] In an exemplary embodiment, the first grating includes alternating layers of a first material having a first mode effective refractive index and a second material having a second mode effective refractive index, wherein the first mode effective refractive index is not equal to the second mode effective refractive index, and the difference between the first mode effective refractive index and the second mode effective refractive index defines the width of the selected wavelength range.

[0042] According to another aspect, a CPM laser is provided. In an exemplary embodiment, the CPM laser includes an optical cavity comprising at least two gain regions, wherein a saturable absorber is disposed between adjacent gain regions of the at least two gain regions, and wherein a first gain region and a second gain region of the at least two gain regions have correspondingly equal effective lengths, meaning that the effective optical lengths are equal.

[0043] In an exemplary embodiment, the optical cavity is C-shaped, and a saturable absorber is disposed at the center of the curve of the C-shaped optical cavity.

[0044] In an exemplary embodiment, at least one of the at least two gain regions includes a grating configured to define the effective length of the at least one gain region and configured to substantially reflect light with wavelengths within a selected wavelength range.

[0045] In an exemplary embodiment, at least one of the at least two gain regions is defined at least partially by a saturated gain phase transition region.

[0046] In an exemplary embodiment, at least one of the at least two gain regions is defined at least in part by an end that has been micro-cleaved or etched.

[0047] According to another aspect, a method is provided. The method includes: fabricating a collisional pulse mode-locked (CPM) laser having a first gain region and a second gain region. The first and second gain regions have correspondingly substantially equal effective lengths, such that co-propagating pulses generated respectively within the first and second gain regions collide at the boundary between the first and second gain regions, thereby causing harmonic generation.

[0048] In an exemplary embodiment, manufacturing a CPM laser includes etching at least a portion of a first gain region to define an effective length of the first gain region, wherein the first gain region is one of at least two gain regions of the cavity of the CPM laser.

[0049] In an exemplary embodiment, the boundary between the first gain region and the second gain region is a saturable absorber, and the effective length of the first gain region is defined as substantially equal to the effective length of the second gain region.

[0050] In an exemplary embodiment, the CPM laser further includes a third gain region, the second gain region being located between the first gain region and the third gain region, and the effective length of at least one of the first gain region or the third gain region is determined at least in part based on the cleaving distance of the cleaving process used to define the end of at least one of the first gain region or the third gain region.

[0051] In an exemplary embodiment, the second gain region has an effective length of the middle region, and the effective length of the side gain region of at least one of the first gain region or the third gain region is determined based on the effective length of the middle region and the effective length corresponding to the cleavage distance.

[0052] In an exemplary embodiment, manufacturing a CPM laser includes performing a cleaving process to define an end of at least one of a first gain region or a third gain region.

[0053] In an exemplary embodiment, the method further includes: forming a saturable absorber, wherein forming the saturable absorber includes: electrically isolating the saturable absorber from a first gain region and a second gain region, and defining electrodes configured for injecting current into the saturable absorber, wherein the saturable absorber is a boundary separating the first gain region and the second gain region.

[0054] In an exemplary embodiment, manufacturing a CPM laser includes: forming a C-shaped optical cavity including a first gain region and a second gain region, wherein the boundary between the first gain region and the second gain region is a saturable absorber located at the center of the curve of the C-shaped optical cavity; and cleaving the first gain region and the second gain region in the same cleaving process.

[0055] In an exemplary embodiment, the method further includes: partially etching the corresponding ends of the first gain region and the second gain region before cleaving the first gain region and the second gain region to define corresponding cantilever portions at the ends of the first gain region and the second gain region, and the cleaving is configured to cleave away the cantilever portions.

[0056] In an exemplary embodiment, the first gain region and the second gain region have the same effective length.

[0057] In an exemplary embodiment, manufacturing a CPM laser further includes forming a first grating in a first gain region of an optical cavity, the optical cavity including a first gain region and a second gain region, the first grating defining an effective length of the first gain region and being configured to substantially reflect light with wavelengths within a selected wavelength range.

[0058] In an exemplary embodiment, the first grating is formed using at least one photolithography process.

[0059] In an exemplary embodiment, the method further includes forming a second grating in a second gain region of the optical cavity, the second grating defining an effective length of the second gain region and being configured to substantially reflect light with wavelengths within a selected wavelength range.

[0060] In an exemplary embodiment, the first grating includes alternating layers of a first material having a first mode effective refractive index and a second material having a second mode effective refractive index, wherein the first mode effective refractive index is not equal to the second mode effective refractive index, and the difference between the first mode effective refractive index and the second mode effective refractive index defines the width of the selected wavelength range.

[0061] According to another aspect, a collisional pulse mode-locked (CPM) laser is provided. In an exemplary embodiment, the CPM laser includes an optical cavity comprising at least two gain regions, wherein a saturable absorber is disposed between adjacent gain regions of the at least two gain regions, and wherein a first gain region and a second gain region of the at least two gain regions have correspondingly substantially equal effective lengths.

[0062] In an exemplary embodiment, the optical cavity is a C-shaped optical cavity, and a saturable absorber is disposed at the center of the curve of the C-shaped optical cavity.

[0063] In an exemplary embodiment, at least one of the at least two gain regions is defined at least partially by the cleavage end.

[0064] In an exemplary embodiment, the effective length of the gain region, at least partially defined by the cleaving end, is configured such that the effective length of the gain region, at least partially defined by the cleaving end, corresponds to the cleaving distance of the cleaving process used to form the cleaving end.

[0065] In an exemplary embodiment, the active material of the at least two gain regions includes at least one layer of quantum dots (QDs) or quantum darts (QDashes).

[0066] In an exemplary embodiment, the CPM laser is configured to operate at a repetition rate of at least 100 GHz.

[0067] According to another aspect, an optical transceiver is provided. In an exemplary embodiment, the optical transceiver includes a CPM laser. The CPM laser includes an optical cavity comprising at least two gain regions, wherein a saturable absorber is disposed between adjacent gain regions of the at least two gain regions, and wherein a first gain region and a second gain region of the at least two gain regions have correspondingly equal effective lengths.

[0068] For example, various embodiments of optical transceivers, including the CPM laser disclosed herein, are provided as light sources.

[0069] According to another aspect, an optical interconnect is provided. The optical interconnect includes a first transceiver, a second transceiver, and an optical fiber enabling optical communication between the first transceiver and the second transceiver. At least one of the first or second transceiver includes a CPM laser. The CPM laser includes an optical cavity comprising at least two gain regions, wherein a saturable absorber is disposed between adjacent gain regions of the at least two gain regions, and wherein the first gain region and the second gain region of the at least two gain regions have correspondingly equal effective lengths.

[0070] According to another aspect, a data center network is provided. The data center network includes a plurality of optical interconnects configured to provide optical communication channels between corresponding computing elements (e.g., processing units, switches, etc.) within the data center. Each optical interconnect includes a first transceiver, a second transceiver, and an optical fiber enabling optical communication between the first and second transceivers. At least one of the first or second transceivers includes a CPM laser. The CPM laser includes an optical cavity comprising at least two gain regions, wherein a saturable absorber is disposed between adjacent gain regions, and wherein the first and second gain regions have correspondingly equal effective lengths.

[0071] Various embodiments relate to interconnects (e.g., interconnect topologies) that are scalable and advantageous for networks requiring a large number of all-to-all or point-to-point links between one or more nodes or transceiver pairs. In particular, this document provides silicon photonic interconnects or topologies that can achieve at least moderate bandwidth between many nodes with physical fiber connections. In some implementations, the one or more nodes or transceiver pairs are coupled to optical fibers, allowing a single wavelength to pass through them. In other implementations, multiple wavelengths or sets of wavelengths can be transmitted or received by a node, while simultaneously transmitting multiple wavelengths or sets of wavelengths to other nodes via an optical fiber loop connecting three or more nodes. In some implementations, such interconnects described herein do not rely on or include one or more of the following: wavelength synchronization between transmit and receive pairs, arbitration of one or more fibers, demultiplexers and / or optical cross switches on the receiver side. In some embodiments, the size of the optical interconnect can be designed to fit panel form factor specifications, or as an in-board optical connector or co-packaged optics. In some embodiments, this disclosure provides optical interconnects for high bandwidth density applications, such as switches and GPUs or other processing elements. In one example, the processing elements may include a central processing unit (CPU), a graphics processing unit (GPU), a data processing unit (DPU), a quantum processing unit (QPU), multiple parallel processing units (PPUs), and an application-specific integrated circuit (ASIC). The QPUs are configured to perform one or more operations associated with a quantum algorithm. In some embodiments, each of the one or more QPUs may include multiple qubits, and the one or more QPUs may communicate with each other via quantum channels. In some embodiments, each of the multiple qubits may include local qubits, global qubits, and / or synchronization qubits. In some embodiments, the local qubits of each QPU may be configured to perform one or more operations associated with a quantum algorithm on the QPU associated with the local qubits.

[0072] The term "node" as used herein can refer to any number of computer processing units (CPUs), graphics processing units (GPUs), data processing units (DPUs), or memory media connected to a network switch. A network switch can communicate with other network switches of the same type, and the same processors and memory units can be connected to those other network switches. However, in other implementations, "node" can also refer to a processor capable of communicating with all other nodes in the network or subnet.

[0073] The term "optical fiber" as used herein can refer to a single optical fiber (e.g., comprising a core and cladding) used to provide unidirectional optical communication, a bidirectional optical fiber pair (e.g., each pair comprising a core and cladding) used to provide transmit and receive communication in an optical network, or a multi-core optical fiber such that a single cladding can encapsulate multiple single-mode or multi-mode cores. Optical fibers can extend continuously and uninterruptedly between nodes or transmit / receive pairs (e.g., via a straight-through connection), or comprise two or more optical fibers connected via fiber-to-fiber connections, such that these optical fibers function as a single optical fiber.

[0074] Silicon photonics (SiP) is a technology that enables optical systems to be fabricated using silicon processes with silicon as the optical medium. Various optical components, such as interconnects and signal processing elements, can be fabricated and integrated within a single SiP device. Some SiP devices are fabricated on a silicon substrate or on a silicon layer on a silicon substrate; this technique is often referred to as silicon-on-insulator (SOI). In some optical systems, SiP devices are attached to external devices to facilitate optical communication. However, it is often difficult to precisely align the optical signal on the SiP with the external device receiving the light.

[0075] In some optical systems, SiP devices are attached to external devices to facilitate optical communication. For example, the system includes one or more waveguides that transmit optical signals to and / or from an optical chip. Examples of optical chips that may be included in the device include, but are not limited to, one or more components selected from the group consisting of: facets from which optical signals can enter and / or exit the waveguide; inlets / outlets from above or below the device; multiplexers for combining multiple optical signals onto a single waveguide; multiplexers for separating multiple optical signals so that different optical signals are received on different waveguides; optical couplers; optical switches; lasers serving as optical signal sources; amplifiers for amplifying the intensity of optical signals; attenuators for attenuating the intensity of optical signals; modulators for modulating signals onto optical signals; and modulators for converting optical signals into electrical signals. These components provide an optical path for the optical signals passing through the device. Additionally, the device may optionally include electrical components. For example, the device may include electrical connections for applying a potential or current to the waveguide, controlling active optical components (e.g., modulators), and / or controlling other components on the optics. Attached Figure Description

[0076] The accompanying drawings will now be consulted; these drawings are not necessarily drawn to scale.

[0077] Figure 1AA schematic cross-sectional view of an exemplary collision pulse mode-locked (CPM) laser according to certain embodiments is provided, the cross-section of which is taken from a plane parallel to the surface of the substrate supporting the CPM laser.

[0078] Figure 1B Provided according to certain embodiments Figure 1A The schematic cross-sectional view of the exemplary CPM laser shown is taken from a plane perpendicular to the surface of the substrate and includes the emission direction of the CPM laser.

[0079] Figure 1C A schematic cross-sectional view of another exemplary CPM laser, taken in a plane parallel to the surface of the substrate carrying the CPM laser, is provided according to certain embodiments.

[0080] Figure 1D Provided according to certain embodiments Figure 1C The image shows an enlarged view of one end of the CPM laser.

[0081] Figure 2A Illustrations of manufacturing according to certain embodiments are provided. Figure 1A and Figure 1B The diagram illustrates flowcharts of various processes and / or procedures for the CPM laser.

[0082] Figure 2B Illustrations of manufacturing according to certain embodiments are provided. Figure 1C The diagram illustrates flowcharts of various processes and / or procedures for the CPM laser.

[0083] Figure 3 A schematic cross-sectional view of another exemplary collision pulse mode-locked (CPM) laser according to certain embodiments is provided, the cross-section of which is taken from a plane parallel to the surface of the substrate supporting the CPM laser.

[0084] Figure 4 Illustrations of manufacturing according to certain embodiments are provided. Figure 3 The diagram illustrates flowcharts of various processes and / or procedures for the CPM laser.

[0085] Figure 5 A schematic cross-sectional view of another exemplary collision pulse mode-locked (CPM) laser according to certain embodiments is provided, the cross-section of which is taken from a plane parallel to the surface of the substrate supporting the CPM laser.

[0086] Figure 6 Illustrations of manufacturing according to certain embodiments are provided. Figure 5 The diagram illustrates flowcharts of various processes and / or procedures for the CPM laser.

[0087] Figure 7 A schematic cross-sectional view of another exemplary collision pulse mode-locked (CPM) laser according to certain embodiments is provided, the cross-section of which is taken from a plane parallel to the surface of the substrate supporting the CPM laser.

[0088] Figure 8 Illustrations of manufacturing according to certain embodiments are provided. Figure 5 The diagram illustrates flowcharts of various processes and / or procedures for the CPM laser.

[0089] Figure 9 A block diagram of an optical interconnect including an array of HM devices is provided according to certain embodiments.

[0090] Figure 10 Block diagrams of example systems according to certain embodiments are provided, which may include one or more optical interconnects including HM devices.

[0091] Figure 11 A schematic diagram of an example data center, which may include one or more optical interconnects according to certain embodiments, is provided.

[0092] Figure 12 A block diagram of two example communication devices communicating with each other according to certain embodiments is provided.

[0093] Figure 13A and 13B Top and perspective views of a transceiver module operatively coupled to a network adapter according to embodiments of the present disclosure are shown, respectively.

[0094] Figure 14 Some exemplary scenarios for using optical transceivers according to various embodiments are described.

[0095] Figure 15 A block diagram schematically illustrating a computing system that may include one or more optical interconnects in various embodiments is provided.

[0096] Figure 16 An example computing environment according to at least one embodiment is illustrated.

[0097] Figure 17 The illustration shows another example computer system that may include optical interconnects according to at least one embodiment.

[0098] Figure 18 This is a block diagram illustrating a co-encapsulated network device according to embodiments disclosed herein. Detailed Implementation

[0099] This disclosure describes various embodiments more fully with reference to the accompanying drawings. It should be understood that some, but not all, embodiments are shown and described herein. In fact, these embodiments may take many different forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that this disclosure will satisfy applicable legal requirements. The same reference numerals throughout the drawings refer to the same elements.

[0100] Various embodiments provide collisional pulse mode-locked (CPM) lasers comprising optical cavities of segments or sections having substantially equal effective lengths and / or optical lengths, and methods for fabricating CPM lasers comprising optical cavities of segments or sections having substantially equal effective lengths. As used herein, the effective length is a length L equal to the effective mode refractive index n of the segment multiplied by the physical length d of the segment (e.g., L = nd). Some segments have multiple regions with different effective mode refractive indices. In such cases, the effective length L of the segment is the sum of the products of the effective mode refractive indices of the individual regions of the segment and the physical length of the region (e.g., ...). , where n i It is the effective refractive index of the mode in the i-th region of this segment, while d i This is the physical length of the i-th region of the segment. In some cases, the effective mode refractive index is wavelength-dependent, and therefore the corresponding effective length is also wavelength-dependent. In some embodiments, the effective lengths of the segments or regions of the optical cavity differ from each other by less than 1%. In some embodiments, the CPM laser can be integrated into an optical interconnect (e.g., as an optical signal generation component, an optical signal detection component, etc.).

[0101] In various embodiments, CPM lasers are configured to generate frequency comb laser pulses comprising multiple teeth or spectral lines. The teeth or spectral lines of the frequency comb laser pulses can be used as optical carriers for performing coarse and / or dense wavelength division multiplexing (WDM). One technique for creating comb lasers is the collisional pulse mode-locking (CPM) method, in which two or more counter-propagating pulses collide at specific locations within the optical cavity (typically at a saturable absorber). These interactions can enhance pulse shortening and stability because the pulses amplify each other's peak intensity at the collision point within the optical cavity. In this technique, instead of a single pulse circulating within the optical cavity, multiple pulses are generated within the cavity such that these multiple pulses are uniformly spaced along the round-trip path.

[0102] When a CPM laser is harmonically mode-locked, multiple pulse pairs collide in a single round trip, effectively increasing the frequency of pulse collisions and thus improving the repetition rate. For example, in second harmonic mode-locking, there are two pairs of backpropagating pulses colliding in each round trip, effectively doubling the repetition rate. The higher repetition rate of harmonic mode-locking does not change the fundamental physics of the comb laser; it simply adds more collision events per unit time. However, by introducing harmonics, maintaining system stability becomes more challenging because the cavity setup needs precise alignment to manage the constructive interference of multiple pulses.

[0103] In CPM (Cavity-Based Laser) lasers, the pulses propagating within the cavity must meet at the saturable absorber region. Conventionally, the conventional cleavage of the sample defines the effective length of a segment or region of the CPM cavity. However, conventional cleavage precision is insufficient to ensure efficient harmonic generation. For example, conventional cleavage precision can result in differences in the effective length of different segments or regions of the cavity of up to 5%. Therefore, there is a technical challenge in providing CPM lasers with cavity segments or regions of substantially equal effective length, enabling CPM lasers to be configured to provide frequency comb laser pulses in an energy-efficient manner.

[0104] Various embodiments provide technical solutions to these technical problems. For example, various embodiments provide methods for manufacturing CPM lasers, which include optical cavities having segments or regions whose effective lengths differ from each other by less than 1%. In some embodiments, the effective lengths of the segments or regions of the optical cavity differ from each other by less than 0.01%. These CPM lasers can be integrated into optical interconnects and / or other systems, used as coarse or dense WDM sources, etc. Therefore, various embodiments provide improvements to fields such as CPM lasers and optical communications.

[0105] Exemplary CPM laser and method of manufacturing thereof Figure 1A A schematic cross-sectional view of an exemplary CPM laser 100 is provided, the cross-section of which is taken from a plane passing through the active region 111 of the optical cavity 110 of the CPM laser, the plane being parallel to the surface 106 of the substrate 105 on which the CPM laser 100 is manufactured and / or disposed. Figure 1B A schematic cross-sectional view of a CPM laser 100, taken from a plane including the emission direction 10 and perpendicular to the surface 106 of a substrate 105 on which the CPM laser 100 is manufactured and / or disposed, is provided. For example, Figure 1A The cross-section of the CPM laser 100 shown is along... Figure 1B The AA line shown is the cut-off point, and Figure 1B The cross-section of the CPM laser 200 shown is along... Figure 1A The BB line shown is the cut-off point.

[0106] The CPM laser 100 includes an optical cavity 110 divided into two or more gain segments or regions 112 (e.g., 112A, 112B). In an exemplary embodiment, adjacent gain segments or regions 112A, 112B are spatially separated from each other by a saturable absorber 114. For example, the saturable absorber 114 is disposed between the first gain segment or region 112A and the second gain segment and / or region 112B. In various embodiments, the gain segments or regions 112 include gain or active materials configured to generate light. For example, the gain segments or regions 112 may include active materials and / or semiconductor materials, including quantum wells configured to generate light. In various embodiments, the gain segments or regions 112 include active materials including quantum confined materials selected from the group consisting of quantum wells (QWs), quantum dots (QDs), and quantum short wires (QDashes). For example, for high repetition rate operation (e.g., >100 GHz or >200 GHz), in some embodiments, the gain segment or region 112 includes a quantum dot layer configured to provide a high gain density with a low linewidth enhancement factor, thereby maintaining a stable mode-locking mechanism even with shortened cavity lengths. In some embodiments, the active material includes a combination of QW, QD, and / or QDash (e.g., two or more of these).

[0107] Optical cavity 110 is defined by a first end 116A and a second end 116B. For example, a CPM laser 100 may emit light (e.g., in the form of frequency comb laser pulses) via the second end 116B. In another example, the first end 116A may be reflective in order to retain optical power within optical cavity 110. For example, the first end 116A may be configured to be highly reflective and reflect light incident on the waveguide structure 125 from within the waveguide structure 125 back into the waveguide structure 125 by a proportion greater than 50% (e.g., greater than 75%, greater than 80%, greater than 90%, greater than 95%, greater than 99%). In various embodiments, the second end 116B may be partially reflective.

[0108] like Figure 1B As shown, the optical cavity 110 is a portion of the ridge waveguide 120, which includes the waveguide structure 125. The waveguide structure 125 communicates optically with the active region 111 via its surface 126. For example, the waveguide structure 125 may be a waveguide configured to support one or more guided modes and / or contain / confine light within it. In some embodiments, the ridge waveguide 120 further includes one or more cladding layers 140A, 140B disposed between the active region 111 and the substrate 105 and / or disposed on the surface of the waveguide structure 125, which is opposite to the surface in the waveguide structure 125 that communicates optically with the active region 111.

[0109] The first gain segment or region 112A has a first effective length L1, while the second gain segment or region 112B has a second effective length L2. The first effective length L1 is substantially equal to the second optical length L2 and / or its harmonics. For example, in some embodiments, the first effective length L1 is substantially equal to the second optical length L2 (e.g., L1 = L2). In some embodiments, the optical cavity may include three gain segments or regions (e.g., a left region, a middle region, and a right region). In such embodiments, the physical lengths of the left and right regions may be half the physical length of the middle region, such that a double pass of the pulse in the left and right regions results in all three regions having equal optical lengths. As used herein, the effective lengths of the gain segments or regions are substantially equal to each other when the pulse propagation time is consistent and / or uniform for each gain segment or region. For example, the pulse propagation time in the first gain segment or region is the same as the pulse propagation time in the second gain segment or region.

[0110] Figure 1C A schematic cross-sectional view of another exemplary CPM laser 150 is provided, wherein the cross-section is taken in a plane passing through the active region 161 of the optical cavity 160 of the CPM laser 150, the plane being parallel to the surface of a substrate (not shown) on which the CPM laser 150 is manufactured and / or disposed.

[0111] At least one end of the optical cavity 160 is defined by cleavage. To perform cleavage, a cleavage region 172 is formed and / or defined. In some embodiments, the cleavage region 172 is a BCB-free zone or a polymer-free region through which cleavage is performed. Conventionally, the optical cavity of a semiconductor laser having an optical cavity defined via cleavage is considered to terminate at the interface 176 between the cleavage region 172 and adjacent gain segments or regions 162A, 162C.

[0112] CPM laser 150 includes an optical cavity 160 divided into three gain segments or regions 162 (e.g., 162A, 162B, 162C). Various embodiments may include various numbers of gain segments or regions 162 (e.g., two or more gain segments or regions). In an exemplary embodiment, adjacent gain segments or regions 162A, 162B, 162C are spatially separated from each other by a saturable absorber 164 and a trench 168 configured to electrically isolate the saturable absorber 164 from the gain material of the gain segment or region 162. For example, a saturable absorber 164A is disposed between a first gain segment or region 162A and a second gain segment and / or region 162B, while a saturable absorber 164B is disposed between the second gain segment or region 162B and the third gain segment and / or region 162C. In various embodiments, the gain segments or regions 162 include gain or active material configured to generate light. For example, gain segment or region 162 may include active material and / or semiconductor material, including a quantum well configured to generate light.

[0113] Optical cavity 160 is defined by a first end 166A and a second end 166B. For example, a CPM laser 600 may emit light (e.g., in the form of frequency comb laser pulses) via the second end 166B. In another example, the first end 166A may be reflective in order to retain optical power within optical cavity 160. For example, the first end 166A may be configured to be highly reflective and reflect light incident on the waveguide structure from within the waveguide structure back into the waveguide structure 125 at a rate greater than 50% (e.g., greater than 75%, greater than 80%, greater than 90%, greater than 95%, greater than 99%). In various embodiments, the second end 166B may be partially reflective. For example, the first end 166A may be a cleaved end and may be coated with a reflectivity coating. The second end 166B may be a cleaved end configured to allow partial light transmission.

[0114] In various embodiments, the optical cavity 160 is a ridge waveguide (similar to) including a waveguide structure. Figure 1B (as shown). The waveguide structure communicates optically with the active region 161 via the surface of the waveguide structure. For example, the waveguide structure may be a waveguide configured to support one or more guided modes and / or contain / confine light within it. In some embodiments, the ridge waveguide further includes one or more cladding layers disposed between the active region 161 and the substrate on which the CMP laser 150 is formed / disposed, and / or disposed on the surface of the waveguide structure (the surface opposite to the surface in the waveguide structure that communicates optically with the active region 161).

[0115] The second optical cavity segment (e.g., the second gain segment or region 162B) has an effective length L in the middle segment.M The first and third optical cavity segments (which respectively include gain segments or regions 162A and 162C) are configured to have round-trip times equal to the one-way travel time of the second optical cavity segment (e.g., the second gain segment or region 162B). In other words, the time required for light to travel from the first saturable absorber 164A to the first end 166A and back to the first saturable absorber 164A (approximately and / or substantially) is the same as the time required for light to travel from the second saturable absorber 64B to the second end 166B and back to the second saturable absorber 164B, and (approximately and / or substantially) the same as the time required for light to travel from the first saturable absorber 164B to the second saturable absorber 164A (or vice versa).

[0116] In various embodiments, the end 166 of the optical cavity 160 is defined at least in part by cleavage. Figure 1D The illustration shows a close-up of end 166 of the optical cavity, illustrating the location of the cleavage marking line 170. The cleavage marking line 170 indicates where cleavage should be performed to define end 166. However, due to the precision of the cleavage procedure within the tolerance length L... T Within, the desired cleavage is used to define end 166 at a location within the dashed box 174. In some embodiments, the tolerance distance d T + / - 5 micrometers (the tolerance distance corresponds to the effective tolerance length L) T It is equal to the physical tolerance distance d T The refractive index n of the material multiplied by the cleavage zone 172 C Therefore, cleavage is not performed at the end of the gain segment or region 162. Instead, a cleavage region 172 is added at the end of the gain segment or region 162. In various embodiments, the cleavage region 172 comprises the same material as the gain segment or region 162, but without P-contacts and may not have a cladding or BCB disposed thereon.

[0117] For example, the optical cavity may include: a first optical cavity segment, which includes a cleavage region 172 and a first gain segment or region 162A located between a first end 166A and a first saturable absorber 164A; a second optical cavity segment, which includes a second gain segment or region 162B located between the first saturable absorber 164A and a second saturable absorber 164B; and a third optical cavity segment, which includes a third gain segment or region 162C and a cleavage region 172 located between the second saturable absorber 164B and the second end 166B.

[0118] The cleavage region 172 has a length such that the distance between the end of the gain segment or region 166 and the cleavage mark line 170 is the cleavage length L. C Cleavage distance d cIt is configured to cleave the ridge waveguide. For example, the cleavage region 172 may not include the cladding / BCB / polymer above the active region 160. For example, a CMP laser 150 is formed as a portion of the ridge waveguide having a cladding (e.g., BCB, etc.) disposed between the active region 160 and the outer “top” surface of the ridge waveguide. However, the cladding may be absent in the cleavage region, allowing cleavage to be performed cleaner and more accurately (compared to cleavage performed through the cladding). In some embodiments, the length of the cleavage region 172 is the cleavage length L. C Twice that. In various embodiments, the cleavage distance d C Within the range of 5 to 15 micrometers (e.g., 10 micrometers). It should be understood that the cleavage distance d... C and tolerance distance d T It can vary based on the cleavage procedure used to perform the cleavage that defines end 166.

[0119] When cleaving is performed along cleavage line 170, the physical length of the cleavage region that is partially preserved as part of the optical cavity is the cleavage distance d. C Therefore, the amount of time required for light to propagate from the first saturable absorber 164A to the first end 166A and back to the first saturable absorber 164A depends on the effective cleavage length L. C Its physical distance d C Multiply by the refractive index n of the cleavage zone 172 C (For example, L) C = d C n C For example, light must travel through the effective length L of the side zone. S and effective cleavage length L C From the first saturable absorber 164A to the first end 166A, and then across the effective cleavage distance L C and the effective length L of the lateral zone S To return to the first saturable absorber 164A. Therefore, in some embodiments, the effective length L of the side region S Determined and / or set to make L M It is the effective length of the intermediate gain region - the effective length of the intermediate gain segment or region 162B (e.g., the effective distance between the first saturable absorber 164A and the second saturable absorber 164B). In other words, the effective length L of the side region. S At least in part based on the effective cleavage length L of the cleavage process C and / or physical distance d C To determine.

[0120] It should be understood that the actual cleavage defining end 166 of the optical cavity 160 can be located anywhere within the dashed box 174. In other words, the actual effective length between the first saturable absorber 164A and the first end 166A can, for example, be equal to the ideal effective length L. S +L C The effective tolerance length difference is + / - L T However, due to the physical distance d C Configured to be greater than the physical tolerance distance d T (For example, twice in some embodiments). Therefore, the error introduced by the difference between the actual effective length (e.g., the actual effective length after cleavage) and the ideal effective length (e.g., the planned effective length) between the end 166 of the optical cavity and the adjacent saturable absorber (e.g., the distance between the first end 166A and the first saturable absorber 164A or the distance between the second end 166B and the second saturable absorber 164B) is small.

[0121] Figure 2A Flowcharts illustrating various processes and / or procedures for fabricating a CPM laser are provided. Starting at box 202, a ridge waveguide 120 is fabricated on a substrate 105. For example, a first cladding 140A may be deposited on the substrate 105. An active region 111 may be epitaxially grown on the first cladding 140A. A saturable absorber 114 may be formed in the active region 111 (e.g., via trenching the edges or periphery of the saturable absorber 114 in the active material grown to form the active region 111). A waveguide structure 125 may be fabricated on the active region. For example, material for the waveguide structure 125 may be deposited and / or grown on the active region 111. In some embodiments, a second cladding 140B may be deposited on the waveguide structure 125. In various embodiments, one or more electrical contacts and / or pads may be deposited and / or patterned. The ridge waveguide 120 may be fabricated using various techniques depending on application requirements.

[0122] In various embodiments, forming the saturable absorber 114 includes electrically isolating a portion of the active material in the active region 111 used as the saturable absorber from the gain segment or region 112. For example, a trench (e.g., similar to trench 168) may be formed to electrically isolate the saturable absorber 114 from the gain segment or region 112. Patterned electrical contacts and / or pads may include defining electrodes configured to inject current into the saturable absorber 114. For example, during operation of the CPM100, current may be injected into the saturable absorber 114 such that the active material of the saturable absorber 114 has a high (charge) carrier density (e.g., becomes saturated).

[0123] At frame 204, a first portion 132 of the first end 116A of the optical cavity is defined by etching. For example, photolithographic etching can be performed on the first portion 132 of the first end 116A. For example, at least the first portion 132 of the first end 116A is etched (e.g., through at least a portion of the first gain segment or region 112A) to define a first effective length L1 of the first gain region 112A. For example, during the fabrication of the ridge waveguide 120, a second effective length L2 can be defined by forming a second end 116B. Etching can be performed on the first portion 132 of the first end 116A such that the first effective length L1 is substantially equal to the second effective length L2 (or its harmonics).

[0124] At block 206, in some embodiments, a second portion 134 of the first end 116A of the optical cavity is defined by etching. For example, photolithographic etching can be performed on the second portion 134 of the first end 116A. In some embodiments, the first portion 132 and the second portion 134 of the first end 116A are etched via the same etching process. For example, blocks 204 and 206 can be performed as a continuous process, rather than as two different or separate steps.

[0125] At block 208, in some embodiments, a second portion 134 of the first end 116A of the optical cavity is defined via micro-cleaving. For example, as a result of block 204, a cantilever portion 130 of the ridge waveguide 120 is formed. The cantilever portion 130 can be micro-cleaved to define the second portion 134 of the first end 116A. For example, ultrasonic vibrations can be applied to the cantilever portion 130 and / or the ridge waveguide 120 using an acousto-optic modulator. In various embodiments, the ultrasonic vibrations are characterized by wavelengths that resonate with the cantilever portion 130. As ultrasonic vibrations are applied to the cantilever portion 130 and / or the ridge waveguide 120, the cantilever portion is cleaved from the remainder of the ridge waveguide 120 along the crystal plane exposed by etching the first portion 132 of the first end 116A of the optical cavity 110. For example, applying ultrasonic vibrations to the cantilever portion 130 and / or the ridge waveguide 120 causes an unetched portion (e.g., a second portion 134) of the first end 116A of the optical cavity to cleave at a crystal plane corresponding to the first portion 132 of the first end 116A. The length of the cantilever can be related to the frequency of the ultrasonic waves to achieve efficient cleaving. In various embodiments, the second portion 134 of the first end 116A includes an end or boundary of the waveguide structure 125. In some embodiments, micro-cleaving of the second portion 134 of the first end 116A produces a smooth and / or reflective surface on the second portion 134 of the first end 116A, which is precisely positioned such that a first effective length L1 is substantially equal to a second effective length L2 (or its harmonics).

[0126] In some embodiments, a reflective coating may be applied to the first end 116A to increase its reflectivity.

[0127] Due to the precise placement of the first end 116A (e.g., etching at least a first portion 132 of the first end 116A using a photolithography process), the tolerance for the first effective length L1 to be equal to the second effective length L2 is smaller compared to defining the first end 116A by a conventional cleaving process.

[0128] Figure 2B Flowcharts illustrating various processes and / or procedures for fabricating a CPM laser 150 are provided. Starting at block 222, a ridge waveguide (e.g., ridge waveguide 120) is fabricated on a substrate (e.g., similar to substrate 105). For example, a first cladding (e.g., similar to cladding 140A) may be deposited on the substrate. An active region 161 may be epitaxially grown on the first cladding. A saturable absorber 164 may be formed in the active region 161 (e.g., by etching trenches 168 into the edges or periphery of the saturable absorber 164 in the active material grown for forming the active region 161). A waveguide structure (e.g., similar to waveguide structure 125) may be fabricated on the active region. For example, material for the waveguide structure may be deposited and / or grown on the active region 161. In some embodiments, a second cladding (e.g., similar to cladding 140B) may be deposited on the waveguide structure. In various embodiments, one or more electrical contacts and / or pads may be deposited and / or patterned. Ridge waveguides can be manufactured using various techniques to meet application requirements. The second cladding can be patterned to define a cleavage region 172 between the execution of blocks 232 and 234.

[0129] In various embodiments, forming the saturable absorber 164 includes electrically isolating a portion of the active material in the active region 161, which serves as the saturable absorber, from the gain segment or region 162. For example, a trench 168 may be formed to electrically isolate the saturable absorber 164 from the gain segment or region 162. Patterned electrical contacts and / or pads may include defining electrodes configured to inject current into the saturable absorber 164. For example, during operation of the CPM 150, current may be injected into the saturable absorber 164 such that the active material of the saturable absorber 164 has a high (charge) carrier density (e.g., becomes saturated).

[0130] At box 224, obtain the tolerance distance for the cleavage process to be used. For example, the tolerance distance d T The cleavage distance can be determined based on experience, from one or more data sources about the cleavage process, ad hoc information, etc. At box 226, the cleavage distance of the cleavage process is determined. In various embodiments, the cleavage distance d... C It was determined to be greater than the tolerance distance d T For example, in some embodiments, the cleavage distance d CIt will be determined as the tolerance distance d T twice (e.g., d) C =2d T In some embodiments, the cleavage distance L is determined at block 228. C The corresponding effective length. In an exemplary embodiment, this is related to the cleavage distance L. C The corresponding effective length is equal to the cleavage distance d. C Multiply by the refractive index n of the cleavage zone C .

[0131] At box 230, at least in part based on the cleavage distance L C Determine the effective length L of the lateral zone S In an exemplary embodiment, the effective length L of the side region S Being determined, making L M This is the effective length of the middle section. Then, the effective length L of the side section can be determined. S The corresponding physical distance d S This makes the physical distance d S Multiply by the refractive index n of gain bands or regions 162A, 162B, and 162C. g Equal to the effective lateral zone distance L S (For example, ).

[0132] At least in part based on the effective length L of the lateral region S The corresponding physical distance d S and the cleavage distance d C The location of cleavage marker 170 is determined. Cleavage marker 170 indicates the length for which cleavage is expected to be performed. The location of cleavage region 172 (e.g., the location of interface 176) is also based on the effective length L of the side region. S The corresponding physical distance d S and the cleavage distance d C And thus confirmed.

[0133] In various embodiments, the second cladding formed on the ridge waveguide can be patterned to define the cleavage region 172. For example, any cladding, BCB, polymer, etc., present in the cleavage region 172 can be removed. In some embodiments, the location of the cleavage region 172 is defined prior to the deposition of the second cladding, and the cleavage region 172 is formed via a stripping process. In some embodiments, BCB, polymer, etc., may be present in the cleavage region 172 and removed along the cleavage marker 170 (e.g., the cleavage is expected to be performed along the line). In some embodiments, cleavage can be performed using BCB, polymer, etc.

[0134] At box 234, cleaving is performed. For example, a cleaving process is performed to cleave the ridge waveguide (e.g., cleaving region 172) substantially along cleaving mark 170 (e.g., at a threshold distance d of cleaving mark 170). T Inside).

[0135] In various embodiments, at least a portion of blocks 224-232 is performed by computing entities such as desktop computers, laptop computers, tablets, servers, blades, switches, and / or other computing devices.

[0136] Figure 3 A schematic cross-sectional view of an exemplary CPM laser 300 is provided, the cross-section of which is taken through a plane passing through the active region 311 of the optical cavity 310 of the CPM laser 300, the plane being parallel to a surface 306 on which the substrate 305 on which the CPM laser 300 is fabricated and / or mounted. For example, the active region 311 may be a portion of a ridge waveguide similar to a ridge waveguide 120, the ridge waveguide comprising: a waveguide structure that optically communicates with the active region 311 via a surface of the waveguide structure; and one or more cladding layers (e.g., a first cladding layer disposed between the substrate 305 and the active region 311 and a second cladding layer on a surface of the waveguide structure opposite to the surface on which the waveguide structure optically communicates with the active region 311).

[0137] The CPM laser 300 includes an optical cavity 310 divided into two or more gain segments or regions 312 (e.g., 312A, 312B). In an exemplary embodiment, adjacent gain segments or regions 312A, 312B are spatially separated from each other by a saturable absorber 314. For example, the saturable absorber 314 is disposed between the first gain segment or region 312A and the second gain segment and / or region 312B. In various embodiments, the gain segments or regions 312 include gain or active materials configured to generate light. For example, the gain segments or regions 312 may include active materials and / or semiconductor materials including quantum wells configured to generate light.

[0138] Optical cavity 310 is defined by a first end 316A and a second end 316B. For example, CPM laser 300 may emit light (e.g., in the form of frequency comb laser pulses) via the second end 316B. In another example, the first end 316A may be reflective in order to retain optical power within optical cavity 310. For example, the first end 316A may be configured to be highly reflective and reflect light incident on the waveguide structure of CPM laser 300 at a rate greater than 50% (e.g., greater than 75%, greater than 80%, greater than 90%, greater than 95%, greater than 99%) back into the waveguide structure. In various embodiments, the second end 316B may be partially reflective.

[0139] In various embodiments, a first end 316A of the optical cavity 310 is defined by a saturation gain phase transition region 318. In various embodiments, the saturation gain phase transition region 318 may comprise the same active and / or semiconductor material electrically isolated from the remainder of the first gain segment or region 312A, such that the saturation gain phase transition region can be independently biased. For example, the saturation gain phase transition region 318 may be forward biased relative to the first gain segment or region 312A. In various embodiments, the voltage applied to the saturation gain phase transition region 318 may be tuned to tune a first effective length L1 of the first gain segment or region 312A. For example, during operation of the CPM laser 300, the first effective length L1 may be dynamically adjustable or tunable by controlling the voltage applied to the saturation gain phase transition region 318. For example, the voltage applied to the saturation gain phase transition region 318 controls the amount of phase accumulated by light propagating through the saturation gain phase transition region 318, thereby effectively changing the effective length of the first gain segment or region 312A. In some embodiments, the second gain segment or region 312B additionally or alternatively includes a saturated gain phase transition region 318, such that the second effective length L2 of the second gain segment or region 312B is dynamically adjustable or tunable during operation of the CPM laser 300.

[0140] The first gain segment or region 312A is characterized by a first effective length L1, and the second gain segment or region 312B has a second effective length L2. The first effective length L1 is tunable such that the first effective length L1 is substantially equal to the second optical length L2 and / or its harmonics. For example, in some embodiments, the first effective length L1 is tunable such that the first effective length L1 is substantially equal to the second optical length L2 (e.g., L1 = L2).

[0141] In some embodiments, the optical cavity may include three gain segments or regions (e.g., a left region, a middle region, and a right region). In such embodiments, the physical lengths of the left and right regions may be half the physical length of the middle region, such that a double pass of the pulse in the left and right regions results in all three regions having equal optical lengths. As used herein, when the pulse propagation time is consistent and / or uniform for each gain segment or region, the effective lengths of the gain segments or regions are substantially equal to each other. For example, the pulse propagation time in the first gain segment or region is the same as the pulse propagation time in the second gain segment or region.

[0142] Figure 4 Flowcharts illustrating various processes and / or procedures for fabricating the CPM laser 300 are provided. Starting at box 402, a portion of the ridge waveguide is fabricated on substrate 305. For example, a first cladding layer may be deposited on substrate 305. The active material of the active region 311 may be epitaxially grown on the first cladding layer.

[0143] At frame 404, a saturated gain phase transition region 318 is formed at the first end 316A of the optical cavity 310. For example, trenches can be made in the active material of the active region 311 to electrically isolate the saturated gain phase transition region 318 from the gain region 313 of the first gain segment or region 312A. For example, isolation trenches can be formed or etched to form the saturated gain phase transition region 318. In some embodiments, a reflective coating can be applied to the first end 316A to increase its reflectivity.

[0144] At frame 406, a saturable absorber 314 is formed to define a first gain segment or region 312A and a second gain segment or region 312B. For example, trench etching can be performed on the active material of the active region 311 to electrically isolate the saturable absorber from the first gain segment or region 312A and the second gain segment or region 312B. For example, isolation trenches can be formed or etched in the gain material to define a first edge and a second edge of the saturable absorber 314.

[0145] In various embodiments, block 404 may be executed before block 406. In some embodiments, block 406 may be executed before block 404. In some embodiments, blocks 404 and 406 may be executed simultaneously and / or synchronously. For example, one or more trenches for defining and / or forming the saturable gain phase transition region 318 and trenches for defining and / or forming the saturable absorber 314 may be formed via the same etching process.

[0146] At frame 408, the remaining portion of the ridge waveguide is fabricated. For example, a waveguide structure can be fabricated on active region 311. For example, material for the waveguide structure can be deposited and / or grown on active region 311. In some embodiments, a second cladding layer can be deposited on the waveguide structure. In various embodiments, one or more electrical contacts and / or pads can be deposited and / or patterned. For example, electrical contacts and / or pads can be fabricated for applying corresponding voltages to saturated gain phase transition region 318, saturable absorber 314, and gain segment and / or region 312. The ridge waveguide can be fabricated using various techniques depending on application requirements.

[0147] Figure 5A schematic cross-sectional view of an exemplary CPM laser 500 is provided, the cross-section of which is taken through a plane passing through the active region 511 of the optical cavity 510 of the CPM laser 500, the plane being parallel to a surface of a substrate (not shown) on which the CPM laser 500 is fabricated and / or mounted. For example, the active region 511 may be a portion of a ridge waveguide similar to a ridge waveguide 120, the ridge waveguide comprising: a waveguide structure that optically communicates with the active region 511 via a surface of the waveguide structure; and one or more cladding layers (e.g., a first cladding layer disposed between the substrate and the active region 511 and a second cladding layer on a surface of the waveguide structure opposite to the surface of the waveguide structure that optically communicates with the active region 511).

[0148] The CPM laser 300 includes an optical cavity 510 divided into two or more gain segments or regions 512 (e.g., 512A, 512B). In an exemplary embodiment, adjacent gain segments or regions 512A, 512B are spatially separated from each other by a saturable absorber 514. For example, the saturable absorber 514 is disposed between the first gain segment or region 512A and the second gain segment and / or region 512B. In various embodiments, the gain segments or regions 512 include gain or active materials configured to generate light. For example, the gain segments or regions 512 may include active materials and / or semiconductor materials, including quantum wells configured to generate light.

[0149] The optical cavity 510 is C-shaped. For example, the optical cavity 510 is characterized in that its radius of curvature R relative to a reference point 530 is greater than zero. For example, the optical cavity 510 includes two curved arms 532 (e.g., 532A, 532B), each curved arm defined by a corresponding optical cavity end 516A, 516B, and converging at a saturable absorber 514. For example, the saturable absorber 514 may be located at the center point of the curve of the C-shaped optical cavity 510. The arms 532A, 532B are curved such that the optical cavity ends 516A, 516B are disposed in the same plane (e.g., a plane perpendicular to the surface of the substrate on which the CPM laser 500 is fabricated). In various embodiments, the ends 516A, 516B of the optical cavity 510 are formed during the same cleaving process, such that the cleavage of the ends 516A, 516B occurs at the same crystal plane.

[0150] In various embodiments, the radius of curvature R is configured to minimize the optical loss of the optical cavity 510 while keeping the distance between the two ends 516A, 516B less than a threshold distance. For example, this threshold distance can be selected such that the cleaving process used to simultaneously form and / or define ends 516A, 516B may cleave these two ends along the same crystal plane. In some embodiments, the radius of curvature R is determined and / or configured at least in part based on one or more mode diameters of one or more confined modes of the optical cavity 510.

[0151] For example, the CPM laser 500 may emit light (e.g., in the form of frequency comb laser pulses) via a second end 516B. In various embodiments, the first end 516A may be reflective in order to retain optical power within the optical cavity 510. For example, the first end 516A may be configured to be highly reflective and reflect light incident on the waveguide structure of the CPM laser 500 at a rate greater than 50% (e.g., greater than 75%, greater than 80%, greater than 90%, greater than 95%, greater than 99%) back into the waveguide structure. In various embodiments, the second end 516B may be partially reflective. For example, in various embodiments, the reflectivity of the second end 516B (through which the CPM laser emits laser light) may be in the range of 1% to 99%.

[0152] The first gain segment or region 512A has a first effective length L1, while the second gain segment or region 512B has a second effective length L2. The first effective length L1 is substantially equal to the second optical length L2 and / or its harmonics. For example, in some embodiments, the first effective length L1 is substantially equal to the second optical length L2 (e.g., L1 = L2). In some embodiments, the first effective length L1 is substantially equal to half of the second effective length L2, or vice versa (e.g., L1 = L2 / 2 or L1 / 2 = L2). As used herein, the two values ​​are substantially equal to each other when the difference between them does not exceed 1% of the smaller value. In some embodiments, the effective length of the gain segment or region 512 is determined at least in part based on the effective length corresponding to the cleaving distance of the cleaving process used to define the ends 516A, 516B of the optical cavity 510.

[0153] Figure 6 Flowcharts illustrating various processes and / or procedures for fabricating a CPM laser 500 are provided. Starting at box 602, a curved or C-shaped ridge waveguide portion is fabricated on a substrate. For example, a first cladding layer may be deposited on the substrate. The active material of the curved active region 511 may be epitaxially grown on the first cladding layer. In various embodiments, the active material may be grown or patterned as curved or C-shaped. For example, the active material may be grown or patterned to include two arms that bend away from the center point of the active region 511 with a non-zero radius of curvature R.

[0154] At frame 604, a saturable absorber 514 is formed to define a first gain segment or region 512A and a second gain segment or region 512B. For example, trenches can be made in the active material of the active region 511 to electrically isolate the saturable absorber from the first gain segment or region 512A and the second gain segment or region 512B. In various embodiments, the saturable absorber 514 is defined at the center point within the curved active region 511.

[0155] At block 606, a cleaving process is performed to define a first end 516A and a second end 516B of the optical cavity 510 within the same cleaving process. For example, a single cleaving process can be performed to cleave the first end 516A and the second end 516B along the same crystal plane of the active material. In an exemplary embodiment, the cleaving process is configured to be performed as a micro-cleaving process, wherein a notch is etched (e.g., using photolithography), and then ultrasonic vibration is applied to cleave the active material along the desired crystal plane. In an exemplary embodiment, the cleaving process is a conventional cleaving process. In some embodiments, a reflective coating can be applied to the first end 516A (e.g., to increase its reflectivity).

[0156] In various embodiments, block 604 may be executed before block 606. In some embodiments, block 606 may be executed before block 604.

[0157] At frame 608, the remaining portion of the curved and / or C-shaped ridge waveguide is fabricated. For example, a curved or C-shaped waveguide structure may be fabricated on the active region 511. For example, the material of the waveguide structure may be deposited and / or grown on the active region 511, and may be grown, deposited, and / or patterned to have a curvature with a radius of curvature R. In some embodiments, a second cladding may be deposited on the curved or C-shaped waveguide structure. In various embodiments, one or more electrical contacts and / or pads may be deposited and / or patterned. For example, electrical contacts and / or pads may be fabricated for applying corresponding voltages to the saturable absorber 514 and the gain segment and / or region 512. The ridge waveguide may be fabricated using various techniques depending on application requirements.

[0158] Figure 7A schematic cross-sectional view of an exemplary CPM laser 700 is provided, the cross-section of which is taken through a plane passing through the active region 711 of the optical cavity 710 of the CPM laser 700, the plane being parallel to a surface 706 on which the substrate 705 on which the CPM laser 700 is fabricated and / or mounted. In various embodiments, the active region 711 may be a portion of a ridge waveguide, similar to a ridge waveguide 120, which includes: a waveguide structure that optically communicates with the active region 711 via a surface of the waveguide structure; and one or more cladding layers (e.g., a first cladding layer disposed between the substrate 705 and the active region 711 and a second cladding layer on a surface of the waveguide structure opposite the surface of the waveguide structure that optically communicates with the active region 711).

[0159] The CPM laser 700 includes an optical cavity 710 divided into two or more gain segments or regions 712 (e.g., 712A, 712B). In an exemplary embodiment, adjacent gain segments or regions 712A, 712B are spatially separated from each other by a saturable absorber 714. For example, the saturable absorber 714 is disposed between the first gain segment or region 712A and the second gain segment and / or region 712B. In various embodiments, the gain segments or regions 712 include gain or active materials configured to generate light. For example, the gain segments or regions 712 may include active materials and / or semiconductor materials including quantum wells configured to generate light.

[0160] Optical cavity 710 is defined by a first end 716A and a second end 716B. For example, a CPM laser 700 may emit light (e.g., in the form of frequency comb laser pulses) via the second end 716B. In an exemplary embodiment, at least one of the first end 716A or the second end 716B is defined by a grating 718 (e.g., 718A, 718B). For example, the grating 718 may be a Bragg reflector comprising multiple alternating layers. The alternating layers include first material layers 722 (e.g., 722A, 722B, 722C) having a first mode effective refractive index, which are interleaved between second material layers 724 (e.g., 724A, 724B) having a second mode effective refractive index. The first mode effective refractive index is different from and / or not equal to the second mode effective refractive index. In some embodiments, one of the first material or the second material is the gain material 709 of the active region 711, while the other of the first material or the second material is a different material. In some embodiments, neither the first material nor the second material is the gain material 709.

[0161] In various embodiments, the grating 718 is configured to reflect light with wavelengths within a selected wavelength range. The number of teeth in the grating determines the reflectivity; more teeth result in higher reflectivity. Figure 7In the illustration, the first grating 718A is shown to include five layers 722, 724. However, in various embodiments, the first grating 718A includes a sufficient number of layers to be highly reflective of light with wavelengths within the operating wavelength range of the CPM laser 700. For example, the first grating 718A may be configured to reflect more than 50% (e.g., more than 75%, more than 80%, more than 90%, more than 95%, or more than 99%) of light incident on the grating 718A from within the waveguide structure of the CPM laser 700 back into the waveguide structure.

[0162] In some embodiments, the first grating 718A and the second grating 718B may include the same number of layers 722, 724. In some embodiments, the second grating 718B may include fewer layers 722, 724 than the first grating 718A. For example, the second grating 718B is at least partially reflective of light with wavelengths in a selected wavelength range and is configured to allow frequency comb laser pulses to be emitted through the second grating 715B.

[0163] The alternating layers 722 of the first material and the second material layer 724 of the grating 718 are formed of materials with different refractive indices. The difference between the effective refractive index of the first mode (of the first material) and the effective refractive index of the second mode (of the second material) defines the width of the selected wavelength range. For example, the CPM laser 700 is configured to generate frequency comb laser pulses that include teeth or spectral lines within the selected wavelength range.

[0164] In various embodiments, gratings 718A, 718B define corresponding ends 716A, 716B of the optical cavity 710. For example, gain material 709 may extend beyond gratings 718A, 718B. However, gain material 709 extending beyond the gratings is not part of the active region 711 of the optical cavity 710.

[0165] The ends 716A and 716B of the optical cavity are configured to provide gain segments or regions 712 of equal effective length. For example, the first gain segment or region 712A has a first effective length L1, and the second gain segment or region 712B has a second effective length L2. The first effective length L1 is substantially equal to the second optical length L2 and / or its harmonics. In some embodiments, the first effective length L1 is substantially equal to the second optical length L2 (e.g., L1 = L2).

[0166] In some embodiments, the optical cavity may include three gain segments or regions (e.g., a left region, a middle region, and a right region). In such embodiments, the physical lengths of the left and right regions may be half the physical length of the middle region, such that a double pass of the pulse in the left and right regions results in all three regions having equal optical lengths. As used herein, when the pulse propagation time is consistent and / or uniform for each gain segment or region, the effective lengths of the gain segments or regions are substantially equal to each other. For example, the pulse propagation time in the first gain segment or region is the same as the pulse propagation time in the second gain segment or region.

[0167] For example, since the grating 718 can be defined using a photolithography process, the gain segment or region 712 can be defined with very tight tolerances between the effective lengths of the gain segment or region.

[0168] Figure 8 Flowcharts illustrating various processes and / or procedures for fabricating a CPM laser 700 are provided. Starting at box 802, a portion of a ridge waveguide is fabricated on a substrate 705. For example, a first cladding layer may be deposited on the substrate 705. The active gain material 709 of the active region 711 may be epitaxially grown on the first cladding layer.

[0169] At frame 804, a saturable absorber 714 is formed to separate the first gain segment or region 712A from the second gain segment or region 712B. For example, trenches can be made in the active material of the active region 711 to electrically isolate the saturable absorber 714 from the first gain segment or region 712A and the second gain segment or region 712B.

[0170] At frame 806, a first grating 718A is formed in a first gain segment or region 712A. For example, in some embodiments, the first material is a gain material 709, and forming the first grating 718A includes etching the gain material 709 to form a first material layer 722 and trenches and / or openings in the gain material 709. In some embodiments, a photolithographic etching process can be used to perform the etching. A second material layer 724 can be formed by depositing a second material into the trenches and / or openings in the gain material 709. In some embodiments, the first grating 718A is completed by one or more cleaning steps, chemical mechanical polishing, etc.

[0171] At block 808, a second grating 718B is formed in the second gain segment or region 712B. For example, in some embodiments, the first material is gain material 709, and forming the second grating 718B includes etching the gain material 709 to form a first material layer 722 and trenches and / or openings in the gain material 709. In some embodiments, photolithography etching can be used to perform the etching. The second material layer 724 can be formed by depositing a second material into the trenches and / or openings in the gain material 709. In some embodiments, the second grating 718B is completed by one or more cleaning steps, chemical mechanical polishing, etc.

[0172] In various embodiments, block 804 may be executed before blocks 806 and 808. In some embodiments, blocks 806 and 808 may be executed before block 804. In some embodiments, blocks 804, 806, and 808 are executed at least partially overlapping in time. In some embodiments, blocks 806 and 808 are executed simultaneously and / or synchronously. For example, the first grating 718A and the second grating 718B may be manufactured using the same photolithography etching and deposition process and / or steps.

[0173] At frame 810, the remaining portion of the ridge waveguide is fabricated. For example, a waveguide structure can be fabricated on the active region 711. For example, material for the waveguide structure can be deposited and / or grown on the active region 711. In some embodiments, a second cladding layer can be deposited on the waveguide structure. In various embodiments, one or more electrical contacts and / or pads can be deposited and / or patterned. For example, electrical contacts and / or pads can be fabricated for applying appropriate voltages to the saturable absorber 714 and the gain segment and / or region 712. The ridge waveguide can be fabricated using various techniques depending on application requirements.

[0174] Due to the precise placement of the first end 716A and / or the second end 716B (e.g., using photolithography to define the first grating 718A and / or the first grating 718B), the tolerance for the first effective length L1 to be equal to the second effective length L2 is smaller compared to when the ends of the optical cavity are defined by conventional cleaving processes.

[0175] While examples herein may refer to quantum wells, the active materials for gain segments or regions 112, 162, 312, 512, 712 and saturable absorbers 114, 164, 314, 514, 714 are not limited thereto. In some embodiments, particularly those configured for high repetition rates (e.g., >100 GHz, 200 GHz), the active materials comprise quantum dots (QDs) or quantum dashes (QDashes). In some such embodiments, the active materials also include quantum wells in addition to QDs and / or QDashes. These active materials (e.g., active materials comprising QDs and / or QDashes) provide inhomogeneous broadening of the frequency comb, which is beneficial for a wide spectral bandwidth, and offer higher gain saturation energies compared to conventional quantum wells.

[0176] Exemplary optical interconnect including CPM laser Figure 9 An exemplary optical interconnect 900 including one or more CPM lasers in various embodiments is illustrated. It should be understood that the CPM lasers in the various embodiments can be integrated into various components, assemblies, and / or systems. This document illustrates optical interconnects including one or more CPM lasers in various embodiments to provide examples of such components, assemblies, and / or systems incorporating CPM lasers in various embodiments.

[0177] The optical interconnect 900 includes a first transceiver 905A and a second transceiver 905B. The first transceiver 905A communicates optically with the second transceiver 905B via one or more root optical fibers 920. The first transceiver 905A includes one or more CPM lasers 910. In various embodiments, the one or more CPM lasers 910 are configured to generate frequency comb laser pulses. In an exemplary embodiment, selected teeth or spectral lines of the frequency comb laser pulses can be used to perform coarse or dense WDM.

[0178] The second transceiver 905B includes an optical signal detection component array 915. In various embodiments, the optical signal detection component array 915 is an array of photodiodes, photodetectors, and / or other optical signal detection devices.

[0179] For example, a first transceiver 905A includes one or more CPM lasers configured to generate frequency comb laser pulses. The CPM laser 910 generates frequency comb laser pulses, and one or more teeth or spectral lines of these pulses are modulated to provide an optical signal in which multiple data channels may be encoded. The optical signal is then provided to a second transceiver 905B, for example, via optical fiber 920. One or more optical signal detection components, which may include corresponding photodiodes and / or other photodetectors, can receive and / or detect corresponding portions of the optical signal (e.g., corresponding teeth of the spectral lines of the frequency comb laser pulses), such that information is transmitted from the first transceiver 905A to the second transceiver 905B via one or more optical signals.

[0180] Exemplary Data Center In various embodiments, the system includes a CPM laser configured to generate and provide frequency comb laser pulses. For example, a CPM laser can be used as a coarse or dense WDM source. For example, as... Figure 8 As shown, the CPM lasers in various embodiments can be integrated into an optical interconnect. This optical interconnect can be part of a data center.

[0181] For example, optical interconnects including one or more CPM lasers in various embodiments can be used to enable various components of a data center to communicate with each other. In various embodiments, the system may include: pluggable optical interconnects, chip-to-chip optical interconnects, optical networks including multiple optical interconnects, dense wavelength division multiplexing (WDM) optical links, WDM or DWDM sources, and / or similar components including one or more CPM lasers in various embodiments. For example, according to an exemplary embodiment, a CPM laser may be configured to generate an optical signal and may be used to generate an optical signal transmitted along one or more optical communication paths between two components of a data center.

[0182] Data centers can include multiple network switches within a specific topology (such as fat tree topology, slim fly topology, dragonfly topology, etc.). The specifications and composition of the network switches in the topology affect the overall network performance of the data center (e.g., bandwidth capacity).

[0183] Data centers are the storage and processing hubs of the internet. The massive deployment of cloud applications is causing data center sizes to grow exponentially, spurring the development of faster switches to handle the increasing data traffic within data centers. Currently, the most advanced switches are capable of handling 12.8Tb / s of traffic through electrical switches using application-specific integrated circuits (ASICs) equipped with 256 data channels, each operating at 50Gbps. Such switching ASICs typically consume up to 400W of power, with comparable power consumption for the optical transceiver interfaces attached to each ASIC. To keep pace with increasing business demands, switch capacity roughly doubles every two years. To date, this rapid expansion has been achieved through advancements in manufacturing technologies commonly described by Moore's Law (the observation that the number of transistors in densely packed integrated circuits roughly doubles every two years), such as CMOS technology. However, in recent years, there have been strong indications that Moore's Law is slowing down, raising concerns about maintaining the target expansion rate of switch capacity. Therefore, alternative technologies are being explored.

[0184] Figure 10 A system 1000 according to at least one exemplary embodiment is illustrated. System 1000 includes a data center 1004, a communication network 1008, and one or more network devices 1012. In at least one example embodiment, the data center 1004 corresponds to a set of network devices, such as network switches (e.g., Ethernet switches) connected to a set of servers or a set of compute nodes. The data center 1004 may follow a networking topology (e.g., a hierarchical networking topology), such as a fat tree topology, a slim fly topology, a dragonfly topology, etc. Traffic is routed between the network switches and servers in the data center 1004, and at least one layer of the topology in the data center 1004 is coupled to the communication network 1008 to allow network traffic to flow between the data center 1004 and one or more network devices 1012.

[0185] Examples of communication networks 1008 that can be used to connect data center 1004 and one or more network devices 1012 include Internet Protocol (IP) networks, Ethernet, InfiniBand (TB) networks, Fibre Channel networks, the Internet, cellular communication networks, wireless communication networks, combinations thereof (e.g., Fibre Channel over Ethernet), variations thereof, and / or similar networks.

[0186] The one or more network devices 1012 may include: switches, routers, or network interface controllers (NICs), interconnected using ports, and one or more of personal computers (PCs), laptops, tablets, smartphones, servers, server clusters, and / or any suitable computing devices that transmit and receive signals through the communication network 1008. In at least one exemplary embodiment, the one or more network devices 1012 correspond to another data center similar to or the same as data center 1004.

[0187] As described above, data center 1004 and / or one or more network devices 1012 may include storage devices and / or processing circuitry for performing computing tasks, such as tasks associated with controlling data flow internally and / or via communication network 1008. Such processing circuitry may include software, hardware, or a combination thereof. For example, the processing circuitry may include a memory (containing executable instructions) and a processor (e.g., a microprocessor) that executes the instructions on the memory. The memory may correspond to any suitable type of memory device or collection of memory devices configured to store instructions. Non-limiting examples of suitable memory devices that may be used include flash memory, random access memory (RAM), read-only memory (ROM), variations thereof, combinations thereof, etc. In some embodiments, the memory and processor may be integrated into a common device (e.g., a microprocessor may include integrated memory).

[0188] Additionally or alternatively, the processing circuitry may include hardware such as an application-specific integrated circuit (ASIC). For example, a processor may be or include one or more of an integrated circuit (IC) chip, a microprocessor, a central processing unit (CPU), a graphics processing unit (GPU), a data processing unit (DPU), a field-programmable gate array (FPGA), a network interface controller (NIC), an ASIC, or combinations thereof. The processing circuitry system may include an ASIC and / or any other computing device capable of executing as a CPU, GPU, NIC, DPU, or receiving and / or transmitting data therein.

[0189] Some or all of the processing circuitry may be located on a printed circuit board (PCB) or an assembly of PCBs. It should be understood that any suitable type of electrical component or assembly of electrical components may be suitably included in the processing circuitry.

[0190] Furthermore, although not explicitly shown, it should be understood that data center 1004 and one or more network devices 1012 may include one or more communication interfaces for facilitating wired and / or wireless communication between each other and with other unshown elements of system 1000.

[0191] In related technology systems, fat-tree topologies can use the same electrical switching equipment across all layers (edge, aggregation, core). For example, each switching device could be a 1U switch, where 1U refers to the industry-standard size of rack-mount switches and / or servers. Interconnects between switches at different layers can be achieved using optical links or optical interconnects with active optical cables and optical transceivers implemented in a pluggable form factor specification (also known as "pluggable").

[0192] Optical data center networks rely on the allocation and deallocation of optical paths from the data source to the destination port to ensure that no optical collisions or data loss occur in the fabric. Traditionally, the allocation algorithm is run by a central entity that considers the full demand of source and destination flows and attempts to find the densest mapping of these demands to network resources within a single or multiple time periods.

[0193] Figure 11 An exemplary data center 1100 is illustrated, in which at least one embodiment can be used. In at least one embodiment, the data center 1100 includes a data center infrastructure layer 1110, a framework layer 1120, a software layer 1130, and an application layer 1140.

[0194] In at least one embodiment, such as Figure 11 As shown, the data center infrastructure layer 1110 may include a resource orchestrator 1112, grouped computing resources 1114, and node computing resources (“nodes CR”) 1116(1)-1116(N), where “N” represents a positive integer (which may be an integer “N” different from the integers used in other diagrams). In at least one embodiment, nodes CR 1116(1)-1116(N) may include, but are not limited to, any number of central processing units (“CPUs”) or other processors (including accelerators, field-programmable gate arrays (FPGAs), graphics processors, etc.), memory storage devices 1118(1)-1118(N) (e.g., dynamic read-only memory, solid-state storage devices, or disk drives), network input / output (“NW I / O”) devices, network switches, virtual machines (“VMs”), power modules, and cooling modules, etc. In at least one embodiment, one or more nodes CR 1116(1)-1116(N) may be servers having one or more of the aforementioned computing resources.

[0195] In at least one embodiment, the grouped computing resources 1114 may include separate groups (not shown) of node CRs housed in one or more racks, or separate groups of node CRs housed in multiple racks within data centers (also not shown) in various geographical locations. In at least one embodiment, the separate groups of node CRs within the grouped computing resources 1114 may include computing, networking, memory, or storage resources that can be configured or allocated to support groups of one or more workloads. In at least one embodiment, several node CRs, including CPUs or processors, may be grouped within one or more racks to provide computing resources to support one or more workloads. In at least one embodiment, one or more racks may also include any number of power modules, cooling modules, and network switches, in any combination.

[0196] In at least one embodiment, resource orchestrator 1112 may configure or otherwise control one or more nodes CR1116(1)-1116(N) and / or grouped computing resources 1114. In at least one embodiment, resource orchestrator 1112 may include a Software Design Infrastructure (“SDI”) management entity for data center 1100. In at least one embodiment, resource orchestrator 1112 may include hardware, software, or some combination thereof.

[0197] In at least one embodiment, such as Figure 11As shown, framework layer 1120 includes job scheduler 1122, configuration manager 1124, resource manager 1126, and distributed file system 1128. In at least one embodiment, framework layer 1120 may include a framework of software 1132 supporting software layer 1130 and / or one or more applications 1142 supporting application layer 1140. In at least one embodiment, software 1132 or one or more applications 1142 may respectively include web-based service software or applications, such as service software or applications provided by Amazon Web Services, Google Cloud, and Microsoft Azure. In at least one embodiment, framework layer 1120 may be, but is not limited to, a free and open-source software web application framework, such as Apache Spark™ (hereinafter referred to as "Spark") which can utilize distributed file system 1128 for large-scale data processing (e.g., "big data"). In at least one embodiment, job scheduler 1122 may include Spark driver to facilitate the scheduling of workloads supported by the various layers of data center 1100. In at least one embodiment, configuration manager 1124 may be able to configure different layers, such as software layer 1130 and framework layer 1120 including Spark and distributed file system 1128 for supporting large-scale data processing. In at least one embodiment, resource manager 1126 is able to manage computing resources mapped to or allocated to clusters or groups for supporting distributed file system 1128 and job scheduler 1122. In at least one embodiment, the computing resources of clusters or groups may include group computing resources 1114 on data center infrastructure layer 1110. In at least one embodiment, resource manager 1126 may coordinate with resource orchestrator 1112 to manage these mapped or allocated computing resources.

[0198] In at least one embodiment, the software 1132 included in the software layer 1130 may include software used at least in part by nodes CR1116(1)-1116(N), grouped computing resources 1114, and / or the distributed file system 1128 of the framework layer 1120. In at least one embodiment, one or more types of software may include, but are not limited to, internet webpage search software, email virus scanning software, database software, and streaming video content software.

[0199] In at least one embodiment, one or more applications 1142 included in application layer 1140 may include one or more types of applications used at least in part by nodes CR 1116(1)-1116(N), grouped computing resources 1114, and / or the distributed file system 1128 of framework layer 1120. In at least one embodiment, one or more types of applications may include, but are not limited to, any number of genomics applications, cognitive computing applications, and machine learning applications, including training or inference software, machine learning framework software (e.g., PyTorch, TensorFlow, Caffe, etc.), or other machine learning applications used in conjunction with one or more embodiments.

[0200] In at least one embodiment, any of the configuration manager 1124, resource manager 1126, and resource orchestrator 1112 can implement any number and type of self-modification actions based on any amount and type of data acquired in any technically feasible manner. In at least one embodiment, self-modification actions can protect the data center operator of data center 1100 from making potentially poor configuration decisions and can prevent underutilization and / or poor performance of the data center.

[0201] In at least one embodiment, data center 1100 may include tools, services, software, or other resources to train one or more machine learning models or to use one or more machine learning models to predict or infer information. For example, in at least one embodiment, a machine learning model can be trained by calculating weight parameters based on a neural network architecture using the software and / or computing resources described above with respect to data center 1100. In at least one embodiment, information can be inferred or predicted using trained machine learning models corresponding to one or more neural networks by using weight parameters calculated by one or more training techniques described herein and using the resources associated with data center 1100 described above.

[0202] In at least one embodiment, the data center may use a CPU, application-specific integrated circuit (ASIC), GPU, FPGA, or other hardware to perform training and / or inference using the aforementioned resources. Furthermore, one or more of the aforementioned software and / or hardware resources may be configured as a service to allow a user to train or perform information inference, such as image recognition, speech recognition, or other artificial intelligence services.

[0203] Inference and / or training logic 1115 is used to perform inference and / or training operations associated with one or more embodiments. In at least one embodiment, inference and / or training logic 1115 can be in the system Figure 12In use, at least in part, the operation is inferred or predicted based on weight parameters calculated using neural network training operations, neural network functions and / or architecture, or neural network use cases described herein.

[0204] Figure 12 The illustration depicts a system 1200 including a first communication device 1204A and a second communication device 1204B. Illustratively, but not limitingly, communication device 1204 (e.g., 1204A, 1204B) may correspond to a network device (e.g., network device 1012). Therefore, communication device 1204 may correspond to any type of device that is part of or connected to a communication network (e.g., communication network 1008). Examples of suitable devices that can function or operate like communication device 1204 as described herein include, but are not limited to, one or more of a personal computer (PC), laptop computer, tablet, smartphone, server, server assembly, network interface card, edge router, switch, network interface controller, top-of-rack (ToR) switch, server blade, etc. Communication device 1204 may include a transceiver 1208, a processor 1216, and a memory 1220. The transceiver 1208 may include hardware capable of communicating via the communication channel 1212, while the processor 1216 and memory 1220 may include components that enable the communication device 1204 to provide the required functions or perform certain functions.

[0205] Communication channel 1212 can span a data center or any type of communication network (whether trusted or untrusted). Examples of communication networks that can be used to connect communication devices 1204 and support communication channel 1212 include, but are not limited to, Internet Protocol (IP) networks, Ethernet, InfiniBand (IB) networks, Fibre Channel networks, the Internet, cellular communication networks, wireless communication networks, combinations thereof (e.g., Fibre Channel over Ethernet), and variations thereof. In a particular but non-limiting example, the communication network is capable of transmitting data between communication devices 1204 using optical signals. In this case, communication devices 1204 and the communication network may include waveguides (e.g., optical fibers) carrying optical signals. For example, according to various embodiments, the communication devices and / or the communication network may include one or more CPM lasers 100, 300, 500, 700.

[0206] Figure 13A and 13B Top and perspective views are illustrated respectively of a transceiver module operatively coupled to a network adapter (in this example, a network interface controller (NIC) 1300) according to an embodiment of the present disclosure. Figure 13A and 13BAs shown, the transceiver module may include a first optical module 1301, a second optical module 1303, an adapter 1310, and a dual-port NIC 1320 for the server. Both the first optical module 1301 and the second optical module 1303 can be dual fiber optic transceivers configured for full-duplex communication, allowing bidirectional communication between the source end (e.g., the server) and the destination end (e.g., a leaf switch). The adapter 1310 may be a group of physical components configured to link the first optical module 1301 and the second optical module 1303 to send data to and receive data from the leaf switch.

[0207] In some embodiments, adapter 1310 may be configured to operate in two configurations (e.g., a first configuration and a second configuration). In one aspect, the first configuration may be a default configuration of operation, wherein the first optical module 1301 may be operationally active. The second configuration may be an emergency configuration implemented when the first optical module 1301 experiences an operational failure. Upon detection of such a failure, the second optical module 1303, which was originally inactive or idle, may be engaged to become operationally active and handle all network traffic initially handled by the first optical module 1301.

[0208] In some embodiments, transceiver module 1300 can be configured to operate in a leaf-spine architecture. A leaf-spine architecture is a data center network topology that may include two switching layers: a spine layer and a leaf layer. The leaf layer may include access switches (leaf switches) that aggregate traffic from servers and connect directly to the spine or network core. The spine switches interconnect all leaf switches in a full-mesh topology between the access switches in the leaf layer and the servers that aggregate traffic on those access switches. Therefore, in one embodiment, to ensure reliable downlink operation, transceiver module 1300 may be configured to operate between the servers and the leaf layer. Specifically, as... Figure 13A and 13B As shown, adapter 1310 can be operatively coupled to first optical module 1301 and second optical module 1303, while first optical module 1301 and second optical module 1303 can be operatively coupled to dual-port NIC 1320 of the server.

[0209] In an embodiment, NIC 1300 may include one or more processing circuits as described above; the processing circuits may include a firmware loaded according to the above-described technology.

[0210] Figure 14Some exemplary scenarios for using the optical transceiver 1402 according to some embodiments are depicted. The optical transceiver 1402 can be used in computing systems 1404 (e.g., in a server cluster or within a server computer system), vehicles 1406 (e.g., cars, trucks, trains, or airplanes), and robots 1408 (or robots in a factory), to name just a few. The optical transceiver 1402 can be particularly useful for high-speed communication in environments subject to high levels of electromagnetic interference (EMI).

[0211] Exemplary System Use Cases Figure 15 This is a block diagram illustrating a computing system 1500 (e.g., a data center or high-performance computing (HPC) cluster) according to embodiments described herein. According to at least one embodiment, system 1500 includes multiple subsystems, such as multiple processing devices, multiple network devices, and multiple networks coupled to each other. The computing system 1500 is designed with multiple integrated circuits (referred to as processing devices), each of which may include one or more CPUs and GPUs, forming a powerful and flexible architecture.

[0212] Various processing devices are interconnected via NVLink or other high-speed interconnects to enable high-speed communication between subsystems, and are also connected via NICs or DPUs to ensure efficient data transmission across computing system 1500 and one or more external networks 1530, 1536. In this example, system 1500 includes packet switch 1548 connecting NIC / DPU 1528 to network 1530, and packet switch 1550 connecting NIC / DPU 1532 to network 1536.

[0213] Seamless data exchange and parallel processing are achieved through NVLink-coupled processing devices, thereby improving overall computing performance. The processing devices connect to multiple networks via one or more Network Interface Controllers (NICs) or Data Processing Units (DPUs), enabling the system to handle complex multi-network tasks with high bandwidth and low latency. This configuration is highly suitable for demanding applications requiring massive processing power, such as artificial intelligence (AI), machine learning (ML), and data-intensive computing, while ensuring robust connectivity and scalability across diverse network environments. The integrated circuits of the Computing System 1500 may include one or more CPUs and one or more GPUs.

[0214] Figure 15An exemplary architecture of a multi-GPU architecture is also illustrated. As shown, computing system 1500 includes processing device 1502 with a multi-GPU architecture. Specifically, processing device 1502 may be a system-on-a-chip and includes multiple subsystems, such as CPU 1506, GPU 1508, and GPU 1510. CPU 1506 may be coupled to GPU 1508 via a bare-chip to bare-chip (D2D) or chip-to-chip (C2C) interconnect 1512 (e.g., ground reference signaling interconnect (GRS interconnect)). CPU 1506 may be coupled to GPU 1510 via a D2D or C2C interconnect 1514. CPU 1506 may also be coupled to GPU 1508 and GPU 1510 via a PCIe interconnect.

[0215] The CPU 1506 can be coupled to one or more NICs or DPUs, which in turn are coupled to one or more networks. For example, as Figure 3 As shown, CPU 1506 is coupled to a first NIC / DPU 1526, which is coupled to network 1530. CPU 1506 is also coupled to a second NIC / DPU 1528, which is coupled to network 1530 via switch 1548. NIC / DPU 1526 and NIC / DPU 1528 can be coupled to network 1530 via Ethernet (ETH), NVLINK, or InfiniBand (IB) connections.

[0216] The computing system 1500 also includes a processing device 1504 with a multi-GPU architecture. Specifically, the processing device 1504 includes multiple subsystems, including a CPU 1516, a GPU 1518, and a GPU 1520. The CPU 1516 can be coupled to the GPU 1518 via a D2D or C2C interconnect 1522. The CPU 1516 can be coupled to the GPU 1520 via a D2D or C2C interconnect 1524. The CPU 1516 can also be coupled to the GPU 1518 and GPU 1520 via a PCIe interconnect. The CPU 1516 can be coupled to one or more NICs or DPUs, which in turn are coupled to one or more networks. For example, as... Figure 3As shown, CPU 1516 is coupled to a first NIC / DPU 1534, which is coupled to network 1536. CPU 1516 is also coupled to a second NIC / DPU 1532, which is coupled to network 1536 via switch 1550. NIC / DPU 1532 and NIC / DPU 1534 can be coupled to network 1536 via Ethernet (ETH), NVLINK, or InfiniBand (IB) connections.

[0217] In at least one embodiment, processing device 1502 and processing device 1504 can communicate with each other via NIC / DPU 1538 (e.g., via PCIe interconnect). Processing device 1502 and processing device 1504 can also communicate with each other via high-bandwidth communication interconnect 1540 (e.g., NVLink interconnect or other high-speed interconnect). Figure 3 The packet switch in the diagram can include, for example, an Nvidia Quantum-2 switch. The NIC / DPU in the diagram can include, for example, an Nvidia Bluefield DPU.

[0218] In various embodiments, one or more of interconnects 1512, 1514, 1522, and 1524 may be the optical interconnects disclosed herein. Additionally, in various embodiments, switches 1548 and 1550 may communicate with NIC / DPU 1528 and 1532, respectively, via one or more optical interconnects disclosed herein. In some embodiments, NIC / DPU 1526 and 1534 may communicate with one or more additional components via the optical interconnects disclosed herein. For example, various components of computing system 1500 may communicate with each other via one or more optical interconnects disclosed herein.

[0219] Figure 16An exemplary computing environment 1600 according to at least one embodiment is illustrated, wherein forward pass-off to available memory can be performed. It should be understood that embodiments of this disclosure can also be used with reference to alternative environments, and specific discussions of components are provided in a non-limiting manner and may include equivalents. Furthermore, various features have been omitted for clarity and brevity. Additionally, the systems and methods can be used with a variety of different architectures. The exemplary computing environment 1600 may include a server 1602, which can be used to perform high-performance computing (HPC) workloads, such as AI training or machine learning model training. In embodiments, server 1602 may be an application instance or a compute node. Server 1602 may include a CPU 1610 associated with a switch 1620 (e.g., a Fast Peripheral Component Interconnect (PCIe) switch), which can control at least some data transmissions on communication paths interconnecting various components. In embodiments, CPU 1610 may include a root complex processor.

[0220] PCIe switch 1620 may also be associated with GPU 1630 and DPU 1640, and may transfer data (e.g., via one or more optical interconnects disclosed herein) between at least some of CPU 1610, GPU 1630, DPU 1640, and other components. In one embodiment, PCIe switch 1620 may be associated with more than one GPU or more than one DPU. In another embodiment, PCIe switch 1620 may be located within DPU 1640. PCIe switch 1620 may manage the transfer of at least some of the data between CPU 1610, GPU 1630, and DPU 1640. In another embodiment, the number of GPUs associated with PCIe switch 1620 may be equal to the number of DPUs associated with PCIe switch 1640. In at least one embodiment, server 1602 may include, but is not limited to, any number of CPU 1610, PCIe switch 1620, GPU 1630, and / or DPU 1640, in any combination. For example, in at least one embodiment, server 1602 may include eight, sixteen, thirty-two, and / or more GPUs 1630. In at least one embodiment, for Figure 16 The communication paths that interconnect the various components (including but not limited to CPU 1610, PCIe switch 1620, GPU 1630 and DPU 1640) can be implemented using any suitable protocol, such as PCI (peripheral component interconnect) based protocols (e.g., PCIe), or other bus or point-to-point communication interfaces and / or protocols (e.g., NVLink high-speed interconnect or interconnect protocols).

[0221] The DPU 1640 may include a network interface controller (NIC) 1642, DDR memory 1644, and a non-volatile memory fast (NVMe) device 1646. The NIC 1642 is capable of connecting to a network 1604, which may also connect to additional NVMe devices available to the DPU 1640, for example, via a structure. In one embodiment, the DPU 1640 may not include the NVMe device 1646. In another embodiment, the NVMe device 1646 may reside on a server 1602 instead of the DPU 1640. In yet another embodiment, the computing environment 1600 may include more than one NVMe device 1646, such as a first NVMe device located in the DPU 1640 and a second NVMe device located on the server 1602 and directly associated with the PCIe switch 1620. In another embodiment, the DPU 1640 may not include the DDR memory 1644 and may include a compute storage service (CSS) in place of or as a supplement to the DDR memory 1644. For example, computing environment 1600 may include DPU compute storage (CS) memory 1606, which is part of the CSS and available to DPU 1640. Depending on any suitable interface protocol, such as remote direct memory access over Ethernet, InfiniBand, Fibre Channel (RDMA over Ethernet), etc., network 1604 may be able to interface with DPU CS memory 1606 via NIC 1642.

[0222] The total memory available for data storage in the computing environment 1600 can be expanded by using DPU 1640 on the system nodes. DPU 1640 can access memory pool 1650 already available on server 1602, such as Double Data Rate (DDR) memory, onboard NVMe devices, on-structure NVMe devices, and CS. Memory pool 1650 may include at least one of DDR memory 1644, NVMe 1646, and DPU CS memory 1606. DPU 1640 can also access available memory of other DPUs that are part of pool 1650, and other DPUs may be able to access available memory of DPU 1640, such as pool 1650. This available memory can be accessed and used for data storage without adding computing resources, such as compute nodes, which are necessary for using other solutions. The available pool 1650 accessible to DPU 1640 can be configured to server 1602 to expand the total memory available for data storage, for example, to reduce the data storage load on CPU 1610 or GPU 1630, which can improve the utilization of their memory for processing. For example, during AI training, model states, residual states, activation functions, and checkpoints can be stored on or off a pool 1650 accessible by the DPU 1640.

[0223] Figure 17 A computer system 1700 according to at least one embodiment is illustrated. In at least one embodiment, the computer system 1700 includes at least one optical interconnect of an exemplary embodiment.

[0224] In at least one embodiment, the computer system 1700 includes, but is not limited to, at least one central processing unit (“CPU”) 1702, which is connected to a communication bus 1710 implemented using any suitable protocol, such as PCI (“Peripheral Component Interconnect”), Fast Peripheral Component Interconnect (“PCI-Express”), AGP (“Accelerated Graphics Port”), HyperTransport, or any other bus or one or more point-to-point communication protocols. In at least one embodiment, the computer system 1700 includes, but is not limited to, main memory 1704 and control logic (e.g., implemented in hardware, software, or a combination thereof), and data is stored in main memory 1704, which may be in the form of random access memory (“RAM”). In at least one embodiment, a network interface subsystem (“network interface”) 1722 provides an interface to other computing devices and networks for receiving data from the computer system 1700 and sending data to other systems.

[0225] In at least one embodiment, the computer system 1700 includes, but is not limited to, an input device 1708, a parallel processing system 1712, and a display device 1706, which may be implemented using conventional cathode ray tube (“CRT”), liquid crystal display (“LCD”), light-emitting diode (“LED”), plasma display, or other suitable display technologies. In at least one embodiment, user input is received from the input device 1708, such as a keyboard, mouse, touchpad, microphone, etc. In at least one embodiment, each of the above modules may reside on a single semiconductor platform to form a processing system.

[0226] In at least one embodiment, a computer program in the form of machine-readable executable code or computer control logic algorithms is stored in main memory 1704 and / or secondary storage devices. According to at least one embodiment, if executed by one or more processors, the computer program enables system 1700 to perform various functions. In at least one embodiment, memory 1704, storage devices, and / or any other storage devices are possible examples of computer-readable media. In at least one embodiment, secondary storage devices can refer to any suitable storage device or system, such as hard disk drives and / or removable storage drives, representing floppy disk drives, magnetic tape drives, optical disk drives, digital versatile disc (“DVD”) drives, recording devices, Universal Serial Bus (“USB”) flash memory, etc. In at least one embodiment, the architecture and / or functionality of the various preceding figures are implemented within the context of CPU 1702; parallel processing system 1712; integrated circuits capable of having at least a portion of the capabilities of two CPUs 1702; parallel processing system 1712; chipsets (e.g., groups of integrated circuits designed to operate as units performing related functions and sold); and any suitable combination of one or more integrated circuits.

[0227] In at least one embodiment, the architecture and / or functionality of the various preceding figures are implemented in an environment of a general-purpose computer system, a circuit board system, a game console system dedicated to entertainment purposes, a special-purpose system, etc. In at least one embodiment, the computer system 1700 may take the form of a desktop computer, a laptop computer, a tablet computer, a server, a supercomputer, a smartphone (e.g., a wireless, handheld device), a personal digital assistant (“PDA”), a digital camera, a vehicle, a head-mounted display, a handheld electronic device, a mobile phone device, a television, a workstation, a game console, an embedded system, and / or any other type of logic.

[0228] In at least one embodiment, the parallel processing system 1712 includes, but is not limited to, a plurality of parallel processing units (“PPUs”) 1714 and associated memory 1716. In at least one embodiment, the PPUs 1714 are connected to a host processor or other peripheral device via interconnects 1718 and switches 1720 or multiplexers. In at least one embodiment, the parallel processing system 1712 distributes computational tasks across the parallelizable PPUs 1714, for example, as part of a computational task distribution across multiple graphics processing units (“GPUs”) thread blocks. In at least one embodiment, memory is shared and accessed (e.g., for read and / or write access) among some or all of the PPUs 1714, although such shared memory may incur a performance penalty relative to the use of local memory and registers residing on the PPUs 1714. In at least one embodiment, the operation of the PPUs 1714 is synchronized using commands such as _syncthreads(), wherein all threads in a block (e.g., executing across multiple PPUs 1714) arrive at a certain point of code execution before proceeding.

[0229] According to the exemplary embodiments disclosed herein, PPU 1714 may communicate with each other and / or with switch 1720 via one or more optical interconnects.

[0230] Co-packaging can refer to the tight integration of different electrical and / or optoelectronic chips in the same package. Figure 18This is a block diagram illustrating a co-packaged network device 1800 according to embodiments disclosed herein. Different chips constituting the co-packaged network device are assembled on a single substrate, generally referred to as an MCM assembly 1812. The MCM assembly 1812 may include switching circuitry 1816 surrounded by peripheral or satellite chips 1820. In some embodiments, the switching circuitry 1816 and the surrounding satellite chips 1820 are mounted on a common substrate, although this configuration is not required. The MCM assembly 1812 may be housed within a larger housing of the network device 1812 located behind a front panel 1804. The switching circuitry 1816 may include one or more core digital application-specific integrated circuits (ASICs), CPUs, GPUs, microprocessors, FPGAs, combinations thereof, etc. The switching circuitry 1816 may include several input ports and / or output ports 1828. Input / output (I / O) ports 1824 may include electrical ports and / or optical ports. Furthermore, the switching circuitry 1816 may include a combination of electrical and optical blocks. The electrical blocks of the switching circuit 1816 may include a plurality of electrical switches configured to route signals in the electrical domain. The optical blocks of the switching circuit 1816 may include a plurality of optical components configured to generate, detect, and route signals in the optical domain. In some embodiments, the MCM assembly 1812 may involve or include a plurality of satellite chips 1820 mounted on the same substrate as the switching circuit 1816. In some embodiments, the configuration of one or more optical blocks and one or more electrical blocks depends on (e.g., based on) the number of optical ports in the I / O ports 1824.

[0231] As described above, the optical I / O 1808 (also known as an optical connector) is located at the front panel 1804. As described above, the connectivity between the MCM assembly 1812 and the optical I / O 1808 can be transmitted to the front panel 1804 via optical fiber. This connection can be made directly through the optical I / O 1824 of the switching circuit, or through one or more satellite chips 1820. This connection is typically made with one or more satellite chips 1820 because the satellite chips 1820 may include electro-optical converters and may also include serializers / deserializers (SERDES) to locally support the connection. The satellite chip 1820 may include one or more of a DSP processor, driver, transimpedance amplifier, laser, modulator, photodiode, serializer / deserializer, etc.

[0232] Some embodiments of this disclosure relate to a multi-chip module (MCM) having a centrally positioned main bare core and a plurality of peripherally positioned MCM sockets configured to mechanically receive and electrically connect to a mezzanine package, which may include a co-packaged optics (CPO) package and a co-packaged copper (CPC) package. Each mezzanine package may include a package substrate comprising a connector portion configured to engage the MCM sockets and a main portion extending beyond the periphery of the MCM substrate. The main portion of the mezzanine package may be configured to receive optics and / or integrated circuits, for example, through the mezzanine sockets, to allow connection between the optics and / or integrated circuits / RF copper cable connectors and the main bare core of the MCM. Because the mezzanine package extends beyond the periphery of the MCM substrate, the physical size of the MCM substrate can be kept small to reduce costs and avoid the manufacturing challenges discussed above, while allowing connection to several optics and integrated circuits via the mezzanine package, occupying relatively inexpensive space around the periphery of the MCM substrate. As used herein, the terms "co-packaged optics" (or "CPO") and "co-packaged copper" (or "CPC") can refer to advanced heterogeneous integration of optics with silicon or copper and silicon, where either integration can be implemented on a single package substrate. A CPO can utilize a pluggable optical module including an optical engine (OE) to convert optical signals into electrical signals and vice versa. A CPO can further consist of optical components on a photonics die and electrical components on an electrical die.

[0233] As used herein, a ball grid array (BGA) can be a surface mount package for integrated circuits. A BGA package uses an array of metal conductor balls arranged in a grid to permanently mount devices such as microprocessors onto a PCB. The metal conductor balls can then undergo the reflow process described above, where the metal conductor balls can be preheated and then melted to bond the IC to the substrate to form the IC package.

[0234] As used herein, the term "flip chip (FC)" can refer to a method of interconnecting a bare die (such as a semiconductor device, IC chip, integrated passive device, and microelectromechanical system (MEMS)) to external circuitry via solder bumps deposited on die pads. During final wafer processing, solder bumps may be deposited on die pads on the top side of the wafer. By "flipping" the die, it can be mounted onto external circuitry (such as a circuit board or another chip or wafer) with the top surface of the die facing down and positioned to allow the die pads to align with matching pads on the external circuitry. Solder reflow completes the interconnect.

[0235] in conclusion Many modifications and other embodiments will occur to those skilled in the art to which this disclosure pertains, which have the benefits of the teachings presented in the foregoing description and associated drawings. Therefore, it should be understood that this disclosure is not limited to the specific embodiments disclosed, and that modifications and other embodiments are intended to be included within the scope of the appended claims. Although specific terminology is used herein, it is used only in a general and descriptive sense and not for limiting purposes.

Claims

1. A method, the method comprising: A collision pulse mode-locked (CPM) laser is manufactured, the CPM laser having a first gain region and a second gain region, the first gain region and the second gain region having corresponding substantially equal effective lengths, such that co-propagating pulses generated in the first gain region and the second gain region respectively collide at the boundary between the first gain region and the second gain region, thereby generating harmonics.

2. The method according to claim 1, wherein, Manufacturing the collision pulse mode-locked CPM laser includes etching at least a portion of the first gain region to define the effective length of the first gain region, wherein the first gain region is one of at least two gain regions of the cavity of the collision pulse mode-locked CPM laser.

3. The method according to claim 1, wherein, The boundary between the first gain region and the second gain region is a saturable absorber, and the effective length of the first gain region is defined as substantially equal to the effective length of the second gain region.

4. The method according to claim 1, wherein, The collision pulse mode-locked CPM laser further includes a third gain region, the second gain region being located between the first gain region and the third gain region, and the effective length of at least one of the first gain region or the third gain region is determined at least in part based on the cleaving distance of the cleaving process used to define the end of the first gain region or the at least one of the third gain regions.

5. The method according to claim 4, wherein, The second gain region has an effective length of the middle region, and the effective length of the side gain region of at least one of the first gain region or the third gain region is determined based on the effective length of the middle region and the effective length corresponding to the cleavage distance.

6. The method according to claim 5, wherein, Manufacturing the collision pulse mode-locked CPM laser includes performing the cleavage process to define the end of at least one of the first gain region or the third gain region.

7. The method according to claim 1, further comprising: Forming a saturable absorber, wherein forming the saturable absorber includes: electrically isolating the saturable absorber from the first gain region and the second gain region, and defining an electrode configured to inject current into the saturable absorber, wherein the saturable absorber is the boundary separating the first gain region and the second gain region.

8. The method according to claim 1, wherein, Manufacturing the collision pulse mode-locked CPM laser includes: A C-shaped optical cavity is formed, the C-shaped optical cavity including a first gain region and a second gain region, and the boundary between the first gain region and the second gain region is a saturable absorber located at the center of the curve of the C-shaped optical cavity; and The first gain region and the second gain region are cleaved in the same cleaving process.

9. The method according to claim 8, further comprising: Before cleaving the first gain region and the second gain region, the corresponding ends of the first gain region and the second gain region are partially etched to define corresponding cantilever portions at the ends of the first gain region and the second gain region, and the cleaving is configured to cleave away the cantilever portions.

10. The method according to claim 8, wherein, The first gain region and the second gain region have the same effective length.

11. The method according to claim 1, wherein, Manufacturing the collision pulse mode-locked CPM laser includes: A first grating is formed in the first gain region of an optical cavity that includes the first gain region and the second gain region. The first grating defines the effective length of the first gain region and is configured to reflect light with wavelengths substantially within a selected wavelength range.

12. The method according to claim 11, wherein, The first grating is formed using at least one photolithography process.

13. The method of claim 11, further comprising: A second grating is formed in the second gain region of the optical cavity. The second grating defines the effective length of the second gain region and is configured to substantially reflect light with wavelengths within the selected wavelength range.

14. The method according to claim 11, wherein, The first grating includes alternating layers of a first material having a first mode effective refractive index and a second material having a second mode effective refractive index, wherein the first mode effective refractive index is not equal to the second mode effective refractive index, and the difference between the first mode effective refractive index and the second mode effective refractive index defines the width of the selected wavelength range.

15. A collision pulse mode-locked CPM laser, the collision pulse mode-locked CPM laser comprising: An optical cavity comprising at least two gain regions, wherein a saturable absorber is disposed between adjacent gain regions in the at least two gain regions, and wherein a first gain region and a second gain region in the at least two gain regions have correspondingly substantially equal effective lengths.

16. The collision pulse mode-locked CPM laser according to claim 15, wherein, The optical cavity is a C-shaped optical cavity, and the saturable absorber is disposed at the center of the curve of the C-shaped optical cavity.

17. The collision pulse mode-locked CPM laser according to claim 15, wherein, At least one of the at least two gain regions is defined at least partially by the cleavage end.

18. The collision pulse mode-locked CPM laser according to claim 17, wherein, The effective length of the gain region, at least partially defined by the cleaving end, is configured such that the effective length of the gain region, at least partially defined by the cleaving end, corresponds to the cleaving distance of the cleaving process used to form the cleaving end.

19. The collision pulse mode-locked CPM laser according to claim 15, wherein, The active material of the at least two gain regions includes at least one layer of quantum dot (QD) or quantum short wire (QDash).

20. The collision pulse mode-locked CPM laser according to claim 15, wherein, The collision pulse mode-locked CPM laser is configured to operate at a repetition rate of at least 100 GHz.

Images