Page 58 - EE Times Europe November 2021 final
P. 58

58 EE|Times EUROPE

           Semiconductor Lasers: Innovations, Applications, and Directions


           particularly for datacom interconnects and
           computing applications, as well as for LiDAR
           used in future autonomous electric vehicles.
           Therefore, we see a high demand for dense
           integration of photonics with electron-
           ics. Integrated optoelectronics using III-V
           compound semiconductor technology on
           low-cost silicon substrates has shown exciting
           possibilities. One key driver for using silicon
           for photonics is the advantages of scaling up
           to 300-mm wafer CMOS processes; this drives
           the cost down significantly — up to 60%. A
           further advantage is in the low-loss silicon
           waveguides that connect various photonic
           elements, including lasers, modulators, pho-
           todetectors, and other passive devices.
             Improvements in the yield of silicon
           photonics-based devices coupled with the
           application-driven requirements enable the
           integration of a larger number of devices on
           a single silicon photonic integrated circuit
           [PIC]. We expect that heterogeneous integra-  are increasingly gaining share within the   EETE: What are the trends in high-power
           tion [direct wafer bonding] will be preferred in   materials-processing market because of the   laser applications?
           the coming years.                   greater precision, processing speeds, and   Yole: Currently, we observe more dynamics
                                               flexibility this technology enables. EELs for   in additive manufacturing, semiconductor
           EETE: How will lasers contribute to the   kilowatt material processing are projected to   manufacturing, and medical applications,
           enhancement of AI and ML applications?  grow from US$409 million in 2020 to    especially in dermatology and surgery.
           Yole: Artificial-intelligence and   US$468 million in 2026 at a compound annual   The additive manufacturing industry
           machine-learning applications primarily   growth rate for the period of 2%. This moder-  is going through extraordinary times. 3D
           use silicon chips in deep neural networks. AI   ate growth will continue to be driven by new   printing is enabling digital supply chains
           chips [also called AI accelerators] are specially   applications as well as the automated-   and establishing itself as a viable technology
           designed accelerators for applications based   manufacturing trend.     for on-demand manufacturing. In consumer
           on artificial neural networks [ANNs]. Most                              electronics, laser micromachining is widely
           commercial ANN applications are deep-   EETE: What future innovations do you   used for high-volume semiconductor man-
           learning applications. The main system trends   see for edge-emitting lasers?  ufacturing to achieve the desired precision,
           in data centers are disaggregation, scale-up,   Yole: It is important to emphasize that   quality, throughput, and cost per machined
           heterogeneous acceleration, and the conver-  edge emitters make up a core technology of   part. For devices such as smartphones, tablets,
           gence of high-performance computing [HPC]   multiple laser systems targeting many end   and wearables, laser micromachining is used
           and the cloud. The goal is to bring technol-  markets. Generally, edge-emitter technology   in the fabrication of semiconductor chips and
           ogies to the market that can contribute to   has been driven by multiple laser applications’   their packaging, electronic components, and
           building more efficient systems for workloads   demands on intrinsic laser diode parameters   the circuitry connecting these components in
           in HPC and AI training/inference. As HPC   for different end markets.   the form of PCBs.
           moves into the cloud, the systems will require   For example, the traffic in data centers is   The manufacture of high-resolution
           a low-latency and high-bandwidth commu-  growing exponentially. Therefore, for optical   touchscreen displays also relies on many
           nication fabric for AI accelerators. Integrated   communication, it is essential to increase the   laser-micromachining processes. The camera
           lasers will serve for optical interconnection   bandwidth and speed of the optical mod-  windows, the sensors, and even the external
           of HPC and AI/ML workloads. This solution is   ules at all levels of DC infrastructure while   enclosure utilize laser micromachining to
           very new and is still in R&D.       simultaneously decreasing the footprint and   achieve the required machined results.
                                               power consumption. In terms of discrete   Lasers for medical applications reflect
           EETE: Do you think the laser materials-  laser diodes, innovations stem from the   recent research on the laser’s principles,
           processing market is a potential area    DFB [distributed-feedback] laser design by   technologies, and applications in diagnos-
           of growth?                          enhancing the multi-quantum well of EA   tics, therapy, and surgery. A wide variety of
           Yole: The laser-based materials-processing   [electro-absorption] modulators for high-  lasers serve medical applications, depending
           market is a traditional laser market. Laser   speed operations and reducing the inductance   on such factors as the optical wavelength,
           materials processing has been used for over   of the wires by bonding short wires for high   output power, and pulse format required. In
           four decades and plays a vital role in modern   bandwidth. Future innovations will likely   many cases, the laser wavelength is chosen
           manufacturing and the economy. While   result in the integration of PICs, also known   such that certain specific substances, such
           laser cutting, welding, marking, and drilling   as silicon photonics. Integrating complex   as pigments in tattoos or caries in teeth,
           processes have reached maturity and wide   electro-optical circuits onto a single silicon   will absorb light more strongly than the
           industrial acceptance, new developments   chip enables the creation of new form factors   surrounding tissue so that they can be more
           in recent years in additive manufacturing   while increasing the bandwidth and distance,   precisely targeted. ■
           and micro/nanofabrication have enabled   significantly improving power efficiency and
           new capabilities that lasers can bring to the   density in ways that simply could never be   Maurizio Di Paolo Emilio is editor-in-chief of
           manufacturing industry. Laser-based systems   addressed with discrete devices.  Power Electronics News and EEWeb.

           NOVEMBER 2021 | www.eetimes.eu
   53   54   55   56   57   58   59   60   61   62   63