Abstract

Mid-infrared (MIR) lasers, covering wavelengths from 2 to 20 µm, have emerged as powerful tools bridging fundamental science and practical technologies. In this tutorial, we will highlight the key spectral characteristics and major applications of this band, and review recent advances in state-of-the-art MIR laser sources, with particular emphasis on the 2–4 µm region where rare-earth and transition-metal ions enable direct MIR generation. Examples include ~2 µm Tm and Ho lasers, 2–3 µm Cr:ZnS/ZnSe lasers, and ~4 µm Fe:ZnSe systems.
The discussion will cover both solid-state and fiber implementations, emphasizing performance milestones such as 100-J pulse energy, kilowatt-class average power, and few-cycle pulse durations, along with an assessment of their respective advantages and limitations. Recent progress in MIR semiconductor lasers, parametric frequency conversion systems, and detection technologies will also be introduced in a concise manner.

Abstract

The realization of a large-scale quantum internet – a network capable of sharing entanglement and quantum information among distant quantum nodes – will open the door to applications such as unconditionally secure communication, distributed quantum computation, and networked quantum sensing.
However, optical losses in fibers make the direct distribution of entanglement and/or quantum states over hundreds of kilometers practically impossible. Quantum repeaters are therefore essential for prolonging the distances. They divide a long channel into shorter segments, generate and store entanglement, and extend it by entanglement swapping and purification in typical quantum repeater schemes. A major challenge in building practical quantum repeaters is to overcome low entanglement generation rate.
This tutorial introduces the concept of quantum internet, principles of quantum repeater, current progress in the experimental implementations of quantum repeater including field-deployed demonstration.
Building on global trends, in the talk, also focused on our laboratory’s integrated platform that connects frequency-multiplexed Pr:YSO memories with cavity two-photon sources (cSPDC) through optical fibers stabilized by optical-frequency-comb (OFC) phase-locking systems. The tutorial explains how such integration can realize a high-rate entanglement link, the fundamental building block of the quantum repeater. One promising solution is multiplexing, which allows many entanglement-generation attempts to occur in parallel across distinct temporal or spectral modes. In particular, frequency-multiplexed quantum repeaters can use a single physical quantum memory crystal and a photon source to support tens or hundreds of frequency channels simultaneously, dramatically increasing the probability that at least one mode succeeds per trial.
Finally, the talk discusses field-deployment efforts toward robust quantum networks. In addition to demonstrations in the other areas of the world, we present data from collaborative tests with Japanese telecom companies, where phase noise and correlation between parallel commercial fibers were characterized in urban area of Osaka city.
In summary, this tutorial connects fundamental quantum-optical principles with cutting-edge engineering toward scalable quantum networks.
Participants will gain:
• a conceptual grasp of quantum repeater architectures,
• insight into strategies for high-rate entanglement distribution, and
• a view for the experimental technologies—quantum memories, photon-pair sources, and optical interfaces—that underpin the future quantum internet.

Abstract

Machine learning and artificial intelligence have been developed rapidly in our information society. However, several issues in deep learning have been pointed out, such as huge learning costs and large power consumption. Consequently, reservoir computing has been paying attention as a new machine learning approach with small learning costs.
Reservoir computing is a novel information processing method utilizing temporal dynamics. One of the characteristics of reservoir computing is the use of the transient response dynamics to the input signal, making it well-suited for time-series prediction. This approach is considered to resemble the information processing in the brain. Another major feature is low learning costs, which are large advantages over deep learning. Furthermore, training can be achieved by learning the output weights only, and learning can be retrained immediately even if the characteristics of the input signal change. This simple and fast learning enables implementation in edge devices with small computational power for applications in edge computing.
Reservoir computing has been realized using various physical devices, including photonic reservoir computing. High-speed and low-power photonic computing can be achieved by leveraging the high speed of light and its temporal, spatial, and wavelength multiplexing capabilities. Furthermore, optical integrated circuits, silicon photonics, and optical communication technologies can be applied for miniaturization and enhanced functionality for optical devices used in reservoir computing.
Photonic reservoir computing can be applied for engineering applications, such as the time-series prediction, speech recognition, and image recognition. Signal compensation in wireless and optical communications can be also achieved using reservoir computing.
Another well-known approach to machine learning is reinforcement learning. In reinforcement learning, an agent learns a policy based on the rewards for its actions. For example, in controlling a walking robot, the robot repeatedly undergoes trial and error to avoid falling over, ultimately learning to walk without falling.
One of the examples of reinforcement learning is the multi-armed bandit problem. The multi-armed bandit problem involves maximizing rewards from multiple slot machines (choices) with unknown hit probabilities. For example, if the hit probabilities of multiple slot machines differ, it is necessary to find which slot machine has a higher probability. This is called “exploration.” On the other hand, once exploration has estimated a slot machine with the highest hit probability, continuing to select that machine increases rewards. This is called “exploitation.” Exploration and exploitation have a trade-off relationship, which is known as the exploration-exploitation dilemma.
Photonic decision making has been proposed for solving the multi-armed bandit problem efficiently. Several methods have been investigated using chaotic dynamics in semiconductor lasers, synchronized laser networks, chaotic mode-hopping dynamics in multimode lasers, and spatio-temporal dynamics using spatial light modulators. This multi-armed bandit problem can be applied to channel selection in wireless and optical communications. It is possible to select the optimal channel using this decision-making approach, even in environments where external traffic dynamically changes.
In this tutorial, we provide comprehensive explanation of recent developments in photonic reservoir computing and photonic decision making.

Abstract

Semiconductor laser technologies have brought remarkable advantages in industrial applications after the realization of current-injection room-temperature continuous laser operation by introducing double-hetero structures in 1970. The major application fields have so far been in optical communication and optical disk technologies, and nowadays we are receiving great benefits by these successful technologies.
It is noteworthy to see that the developments in reliable semiconductor laser technologies are by virtue of the tight cooperation in many scientific and technological fields; these are material engineering, condensed matter physics, electronic engineering, and laser science etc. In present day, semiconductor lasers commonly have ultra-thin layer structures so called quantum-wells inside the devices, and this enabled remarkably higher operational performances.
Operation characteristics of semiconductor lasers also have been investigated both experimentally and theoretically in cooperation with device development activities. These studies have provided us many understandings for carrier dynamics, static and dynamic laser oscillation properties. It should be noted, however, that there are still various unclarified operation features in high-density carrier excitation conditions. This is mainly because of the difficulties in semiconductor laser theoretical analyses, and it is hard to foretell what kinds of operation features appear. Nevertheless, thanks to the recent very robust semiconductor laser devices and advanced high-speed electronics, we have been able to study dynamic operation properties under intensive pulse current excitation. Consequently, we have found some unique and attractive features that are usable for novel applications in contemporary and future photonics technologies.
One is the picosecond optical pulse generation by gain-switching operation; sub-nanosecond strong electric pulse excitation has resulted in the generation of sub-5ps duration optical pulses at arbitrary repetition rate. This technology can replace the mode-locking methods, which require delicate balances in some nonlinear effects in the laser oscillation. A few simplified theoretical analyses have also given us a phenomenological understanding for the essential operation mechanism. Combined with optical amplifier and nonlinear wavelength conversion technologies, successful applications have been also shown in multiphoton-excitation and super-resolution biomedical imaging. Another notable advancement of this gain-switched pulse laser is the mega-watt peak power sub-picosecond optical pulse generation by controlling nonlinear-optic effects in optical amplifiers; this result indicates that semiconductor lasers have potentials of being seed laser sources for very high-peak-power picosecond and femtosecond optical pulse sources.
Another one is the optical injection locking of gain-switched semiconductor lasers. This technology has been intended for generating arbitrary duration smooth shaped optical pulses keeping single longitudinal mode laser oscillation. The main application is assumed to be a master laser pulse source for laser induced material micro-processing. Under certain optical injection power levels, chaotic or quasi-periodic laser oscillation features have appeared, but the expected injection locking situations have taken place by increasing the optical injection power. Semiclassical analyses have also well predicted and supported the experimental results under moderate excitation conditions for a semiconductor laser. However, with intensive excitation conditions, discrepancies between the analyses and experiments have become larger, and the injection locking operation has been found to be very difficult by further increase in the excitation. These discrepancies indicate that the proper theoretical treatments for semiconductor lasers are beyond the conventional laser theory modifications.
It is also to be noted that the second-quantized-sate laser oscillation are often unexpectedly observed in various quantum-well laser devices by intensive electric pulse excitations, although these devices are designed and fabricated for the laser oscillation at first-quantized-states. Several unique operation properties have been verified. The second-quantized-sate laser oscillation is very transient; once the first-quantized-states laser oscillation occurs with a small temporal delay, it tends to suppress the preceding second-quantized-sate laser oscillation. At present, a simplified analytical model gives us the operation mechanisms consistent with the experimental results. If we can optimize the device structures, single optical pulses of 1ps duration or less are to be expected by this mechanism.
Recent semiconductor laser technologies can give us novel applications in various fields including biomedical photonics, high-speed and precise measurement, laser material processing, as well as in quantum information technology. Although semiconductor laser diode devices are already in a very advanced stage, further functional extensions are also expected by introducing contemporary optical excitation schemes.

Abstract

Electronic and photonic computing systems are now facing severe performance bottlenecks caused by the slowdown of Moore’s Law and the limitations of electrical interconnects. As data traffic and computational demands increase exponentially, it has become essential to explore photonic approaches that can process and transmit information with extremely low latency and energy consumption. Among these, microresonator frequency combs, also known as microcombs, have emerged as one of the most promising technologies that bridge optical communication, optical signal processing, and terahertz (THz) wireless communication.

A microresonator frequency comb is a set of discrete, equally spaced optical lines generated through parametric four-wave mixing (FWM) inside a high-Q microcavity. When a continuous-wave (CW) laser pumps the cavity, the optical Kerr nonlinearity induces energy transfer between modes separated by the cavity’s free spectral range (FSR), leading to cascaded sideband generation. Under proper dispersion and pump conditions, these sidebands evolve into dissipative Kerr solitons (DKS), forming a phase-locked frequency comb. The Lugiato-Lefever equation (LLE) accurately models the dynamics of a nonlinear Schrödinger-type mean-field equation that incorporates dispersion, detuning, and cavity losses. Stable soliton formation requires a careful balance between group-velocity dispersion (GVD) and the nonlinear phase shift, which is achieved through precise dispersion engineering in materials such as Si3N4, AlGaAs, and MgF2.

In recent years, dispersion control techniques such as waveguide geometry optimization, coupled-resonator design, and avoided-mode-crossing engineering have enabled the development of broadband, low-noise microcombs with high conversion efficiency. Two distinct regimes, bright soliton combs in anomalous dispersion and dark-pulse combs in normal dispersion, offer complementary benefits. Bright solitons offer clean spectral coherence, whereas dark-pulse combs provide higher pump-to-comb efficiency, which is crucial for on-chip integration. The development of perfect soliton crystals, where multiple solitons are equally spaced within the cavity, has further stabilized the spectral envelope and improved reproducibility across devices.

Microcombs are uniquely suited as multi-wavelength optical sources for massively parallel transmission. Each comb line can act as an individual optical carrier, enabling dense wavelength-division multiplexing (WDM) with a single laser pump. In recent experiments at Keio University’s Yagami and K2 campuses, we employed Si3N4 microcombs as the optical source for low-latency IMDD (Intensity-Modulation Direct-Detection) fiber links. The system simultaneously modulated multiple C-band channels without requiring coherent detection or error correction. By utilizing hundreds of stable wavelengths from a single microcomb, we demonstrated error-free (BER < 10-9) data transmission with drastically reduced latency compared with conventional digital coherent optical links. These results establish a bus-free, analog-style optical link architecture that is particularly promising for intra-data center and short-reach applications, where latency and power consumption are dominant.

In parallel, microcombs are being explored as ultra-low-phase-noise photonic oscillators for terahertz (THz) wireless communications. When adjacent comb lines beat in a photomixer, they generate a stable sub-THz carrier whose frequency corresponds to the comb spacing. By leveraging the optical coherence of the pump laser and the high-Q cavity stability, the resulting THz carrier inherits an exceptionally low phase noise, reaching approximately –90 dBc/Hz at 10 kHz offset for 300 GHz oscillations. This level of stability surpasses that of most electronic oscillators, enabling the realization of high-quality QPSK transmission with clear constellations and a low error vector magnitude (EVM). Such photonic THz sources are ideal for high-speed, short-range wireless links and optical-to-wireless gateways.

To further enhance the modulation capacity, we recently conducted a quantitative analysis of phase-noise tolerance for high-order M-QAM transmission in the 300 GHz band. By reconstructing phase noise from measured oscillator spectra and embedding it into a single-carrier (SC) M-QAM link model, we clarified the relationship between oscillator noise and bit-error rate (BER). Two main distortion mechanisms were identified: common phase error (CPE), which can be removed mainly through phase-locked loop (PLL) carrier recovery, and instantaneous phase jitter, which remains as the dominant residual error. Using the 3σ error criterion derived from the root-mean-square error vector magnitude (RMS-EVM), we quantitatively define the tolerance boundary of phase noise for reliable high-order QAM. The results indicate that 16-, 32-, 64-, and 128-QAM require white phase-noise standard deviations below 105, 68, 42, and 19 mrad, respectively, to achieve BER < 10-9. Significantly, the measured phase noise of our Si3N4 microcomb oscillator falls within this tolerance region, demonstrating its feasibility for future 64-QAM and 128-QAM THz systems.

These findings highlight the synergy between optical communication and wireless photonics. The same microcomb platform can simultaneously serve as a multi-wavelength optical transmitter for fiber links and as a coherent THz oscillator for wireless transceivers. Such integration opens a new direction toward hybrid optical–THz systems that directly bridge fiber and free space, offering seamless conversion between optical computation, data transmission, and wireless broadcasting. Moreover, the low-noise microcomb enables analog photonic processing and photonic tensor cores, where optical convolution or matrix multiplication can be performed and transmitted in real-time via THz carriers. This vision integrates photonic computing and wireless communication within a single, low-latency, and energy-efficient framework.

In the latter part of this tutorial, we will also discuss dispersion design strategies for high-repetition-rate microcombs, Si3N4-Si heterogeneous integration for compact modulators, and noise transfer mechanisms from the optical to the THz domain. The session will conclude with perspectives on scalable photonic architectures that utilize microcombs for parallel signal generation, low-latency analog links, and low-phase-noise THz transmission. By integrating optical and wireless functionalities into a chip-scale platform, microcombs serve as a cornerstone technology for next-generation photonic systems, enabling the seamless bridging of computation, communication, and sensing at unprecedented speeds.

Overall, this tutorial will guide participants from the fundamental physics of microcomb formation to the most recent implementations in optical and THz systems, emphasizing how microcombs can transform both wired and wireless communications through their unique combination of coherence, scalability, and ultra-low noise.

Abstract

Intense laser light can provide various irreversible changes in materials, including combustion, melting, ablation, or modifications of the refractive index. Such phenomena are popularly referred to as laser material processing or simply laser processing. As the pulse duration of the laser light decreases to picoseconds or femtoseconds, the dominant interaction processes between light and matter are significantly altered, leading to distinctive characteristics. For instance, the thermal effects are minimal, resulting in so-called nonthermal processing, which leads to the formation of sharp and well-defined processed edges without melting or burning. This property enables high-precision and high-spatial-resolution material processing at the nano- to micro-scale, which has been widely adopted in various industrial applications. One significant advantage is the ability to process a wide variety of materials—from metals and semiconductors to dielectrics and even biological tissues—due to these unique light–matter interaction processes. In this tutorial, the lecturer will first explain the fundamental mechanisms of optical absorption, followed by nonlinear absorption processes induced by ultrashort laser pulses and the resultant irreversible material modifications. He will also briefly touch on what ultrashort laser pulses are and how they are generated, and finally introduce recent applications and future prospects of ultrashort pulse laser material processing.