From an industrial perspective, the development trend of automotive lidar laser technology

The past decade has been a decade of vigorous innovation for automotive LiDAR. Numerous laser technologies and system solutions are competing for market share in fierce competition. Recent trends show that solutions based on vertical cavity surface emitting lasers (VCSELs) and anti-reflection vertical cavity surface emitting lasers (AR-VCSELs) are increasingly converging. According to MEMS Consulting, Changzhou Zonghui Xingguang Semiconductor Technology Co., Ltd. (hereinafter referred to as "Zonghui Xingguang") recently published a review article entitled "Evolution of laser technology for automotive LiDAR, an industrial viewpoint" in the journal Nature Communications. Based on an industrial perspective, the article discusses the development trend of commercial automotive LiDAR laser technology, and focuses on the latest applications and future prospects of VCSEL/AR-VCSEL technology.

Introduction to LiDAR

LiDAR was first invented by Hughes in the 1960s. It was originally used in meteorology, ocean sensing, and terrain mapping. In 1971, NASA integrated a LiDAR called the Lunar Laser Ranging Reflector (LRRR) into Apollo 15 to map the lunar surface, and later promoted its use in spacecraft flying to Mars and Mercury. It was not until the 2010s that LiDAR began to be used in commercial vehicles. In the 2020s, automotive LiDAR began to become popular in high-end electric vehicles. LiDAR provides real-time point cloud maps containing object depth and velocity data, and has become an important component for assisted and autonomous driving.

In 2023, driven by fierce competition from Chinese electric vehicle manufacturers, lidar sales surged. Currently, the main suppliers in the global lidar market include Hesai, RoboSense, Seyond and Innoviz in Asia, Luminar and Ouster in North America, and Valeo in Europe.

A typical automotive LiDAR system consists of a scanning laser, a receiver, related optical components, and integrated drive and processor circuits. It works in conjunction with cameras, sensors, and positioning and navigation systems. Functionally, automotive LiDARs can be divided into two categories: primary LiDARs for long-range forward perception, and auxiliary LiDARs for surrounding environment perception. The combination of the two eliminates blind spots and enables 360° all-round perception of the vehicle.

Based on the detection method, LiDAR technology can be divided into two categories, namely frequency modulated continuous wave (FMCW) and time of flight (ToF). FMCW uses mixed reflected light and frequency modulated transmitted light to determine the distance and speed of a moving object. ToF, on the other hand, determines the distance by calculating the time interval between the transmitted pulse and the returned pulse. ToF is also one of the earliest technologies used in LiDAR. Currently, most LiDAR manufacturers prefer ToF technology because it is simple to operate and has a low cost. Therefore, this article mainly discusses ToF and related laser technologies.

Laser Technologies for Commercial LiDAR Applications

There are endless innovations in the combination of laser technology and advanced optical technology. These innovative technologies, especially nanophotonic solutions, have achieved a high degree of integration between lasers and scanning, further miniaturizing lidar systems and bringing hope for their long-term development.

Table 1 Mainstream laser technologies for commercial lidars at different detection distances

VCSELs were first used in short-range LiDAR and 3D sensing for smartphones and consumer devices, and were pioneered by Philips, Lumentum, Coherent (II-V and Finisar), and ams (Princeton and Vixar). Compared with edge-emitting lasers (EELs), VCSELs have many advantages, including: (1) flexible illumination methods, such as 1D/2D addressable arrays; (2) inherent temperature stability (0.07 nm/℃); (3) circular beams, which are conducive to simplifying optical components; (4) easier packaging; (5) replacing single emitters with arrays to increase redundancy reliability; (6) high cost-effectiveness, and 6-inch gallium arsenide (GaAs) foundries are already mature in mass production for smartphone 3D sensing. The only disadvantages of VCSELs are power density and brightness. However, the emergence of multi-junction technology has greatly improved their power density and power conversion efficiency (PCE), overcoming the bottlenecks in the past in medium and long-range LiDAR applications, such as the 5-junction 905 nm VCSEL provided by Lumentum for Hesai Technology AT128.

In the field of LiDAR, a highly competitive AR-VCSEL technology has recently emerged, which has made significant breakthroughs in reducing the divergence angle and increasing the brightness. The introduction of AR-VCSEL further expands the detection distance and resolution of 905 nm and 940 nm LiDARs, covering all the ranges required by automotive LiDARs. AR-VCSEL was successfully developed in 2021. Although it was released not long ago, it has been adopted in commercial long-distance LiDARs.

The collaborative development of laser and scanning technology

Commercial automotive lidar can be divided into three types based on the scanning method: mechanical lidar (involving the movement of lasers, lenses and sensors), hybrid solid-state lidar (in which only the scanning MEMS micromirrors or rotating mirrors move) and all-solid-state lidar (without any mechanical movement and the scanning beam is controlled electronically).

Figure 1 Four VCSEL-based LiDAR scanning solutions and their performance

Hybrid solid-state LiDAR: EEL vs AR-VCSEL

Purely mechanical LiDAR has been almost eliminated in Level 2 and Level 3 advanced driver assistance systems (ADAS) as hybrid solid-state solutions gradually dominate the market.

Hybrid solid-state LiDAR manufacturers initially combined point light sources (such as 1550 nm fiber lasers or 905 nm EELs) with two-dimensional MEMS mirrors or rotating mirrors. A more recent popular solution is that the solid-state light source is electronically scanned in one direction, while a one-dimensional polygon/reflector scans in the other direction. This solid-state light source contains an array of small VCSEL/AR-VCSEL chips (such as Hesai Technology AT128) or a narrow array of VCSEL/AR-VCSELs. This evolution eliminates the need for precise alignment between the laser and MEMS mirrors, solves the field of view (FoV) issues associated with MEMS mirrors (for example, RoboSense M1 requires five EEL modules to achieve a 120° FOV), and reduces the number of motors from two to one.

In the next few years, competition is expected between EEL and AR-VCSEL-based solutions. Low-cost EEL-based lidar can minimize the number of EELs and solve the angular coverage problem through additional lenses. On the other hand, AR-VCSEL has greater room for improvement in power density and brightness while reducing the device area. In terms of performance, the maximum detection distance of EEL-based lidar is generally 200 m, while AR-VSCEL-based lidar, with the current 6-junction technology, has already exceeded this range, and the advancement of 8~10-junction AR-VSCEL technology is expected to further push the detection distance to 300~400 m.

The ultimate goal: all-solid-state lidar

All-solid-state LiDAR completely eliminates moving parts and replaces mechanical scanning with electronic scanning. Commercially viable solutions include VCSEL flash illumination with defocusing lenses and 1D/2D addressable VCSELs with defocusing lenses. Other solutions include VCSEL/EEL with liquid crystal metasurface (LCM) developed by Lumotive, and FMCW EEL for optical phased array (OPA) LiDAR demonstrated by Quanergy, Aeva, LightIC, Scantinel Photonics and other companies. These solutions have not yet been mass-produced, while addressable VCSEL arrays for LiDAR are gradually entering mass production.

Most all-solid-state LiDAR solutions currently under development are first targeted at short- to medium-range applications. Once proven at shorter distances, all-solid-state long-range LiDAR will be on the scene, and AR-VCSEL is likely to play a key role in it. VCSEL solutions for all-solid-state LiDAR have made rapid progress in both technology maturity and cost-effectiveness, and are becoming the most competitive candidate to achieve the ultimate goal.

Key requirements for future LiDAR technology

Power density and PCE

Higher peak power can achieve higher signal-to-noise ratio and longer detection distance. More junctions can ensure higher external quantum efficiency, which is proportional to the number of junctions. Therefore, at the same driving current, the power density is higher and the PCE at the same optical power is larger. Currently, the number of VCSEL-based lidar junctions on the market is 5~6, and it is possible to increase by 2 every 18 months in the future as Moore's Law. For research and development purposes, Zonghui Xingguang has experimentally demonstrated small divergence angle AR-VCSELs with up to 14 junctions. In theory, there is no upper limit to the number of junctions. However, in practical applications, adding more junctions may bring challenges in the following aspects: thick epitaxial growth, high aspect ratio trench/mesa etching and coating manufacturing, and reliability issues under higher power density and material stress.

Figure 2 VCSEL power and efficiency increase with the number of junctions

Beam Parameter Product (BPP)

The BPP is defined as the product of the laser beam divergence angle θ (half angle) and the radius r of the narrowest part of the beam (beam waist).

Figure 3 BPP vs laser power

Although multi-junction EELs offer higher single-emitter power, the BPP in the fast axis decreases significantly as the number of junctions increases from 1 to 5, limiting the resolution at long distances. AR-VCSELs with superior BPP and M2 enable longer distances and higher resolutions than conventional VCSELs. The trend from conventional VCSELs to AR-VCSELs (to the right and downward) is consistent with the expected trajectory of future long-distance LiDAR lasers.