Faces light up over VCSEL prospects

Until 2017, trying to sell a product that relied on shining lasers at customers' faces might have been expected to alarm those customers - and bring on commercial failure. But, as it has often done, Apple has defied that expectation with its latest iPhone and its face identification functionality, giving top-emitting vertical-cavity surface-emitting lasers (VCSELs) a launchpad into consumer markets.

Gaining momentum both from smartphones and from potential use in intelligent and autonomous vehicles, VCSEL companies are therefore upgrading their fabrication capabilities as they position themselves to benefit.

More traditional needs for VCSELs as light sources to carry data communications over optical fiber are also growing, according to Mark Lourie, director of corporate communications at VCSEL producer II-VI (USA). "Rapid growth of services such as cloud computing and video streaming is driving very strong demand for server-to-server communications within large data centers," Lourie says.

These servers communicate up to 300 meters via active optical cables using VCSEL arrays. This VCSEL technology is now finding an application in consumer electronics, in optical HDMI cables. II-VI has recently invested in additional manufacturing capacity to ramp up production lines serving these markets and meet growing demand, Lourie notes.

VCSELs' use as a light source in sensors monitoring their surroundings, for example in combination with detectors that measure reflected or scattered light, isn't new either. II-VI has a long track record of supplying them for sensing in computer mice and other devices, Lourie says, and foresaw the need to scale up to compete in this area. In 2016, II-VI acquired epitaxial growth and device fabrication capabilities on 6-inch gallium arsenide (GaAs) wafer platforms, transferring, developing, scaling and qualifying VCSEL manufacturing processes and products in these new manufacturing lines.

"As a result, II-VI is today the only vertically integrated 6-inch [wafer] VCSEL manufacturer," Lourie says. "In the second half of 2017, II-VI significantly ramped up manufacturing of VCSELs for 3D sensing using these 6-inch manufacturing lines."

Late 2017 also saw three-dimensional sensing - as deployed in the iPhone X - introduced to consumer electronics for the first time. To detect whether there is something in front of the iPhone X, it is believed that the phone uses a "time-of-flight" (TOF) sensor, powered by an LED-based infrared illuminator. TOF itself isn't new, having first been used in smartphones as a rangefinder in the "laser autofocus" of LG's G3 model in 2014. This method illuminates an object repeatedly at a very fast rate, often using VCSELS, and measures the time taken for light to reflect or scatter back to a detector. It is especially useful for measuring distances and speeds.

If the TOF sensor detects an object, it triggers the iPhone X's True Depth camera to take a picture. If that reveals a face, the phone activates its dot projector, shining a single infrared VCSEL through an optical system to create 30,000 spots while its infrared camera captures an image. It sends both regular and spottily illuminated IR face images to an application-processing unit (APU) that can recognize the owner and therefore unlock the phone.

Because the system senses depth, you can't hack it with photographs, says Manuel Tagliavini, principal analyst for MEMS and sensors at research firm IHS Markit (UK). It may be hackable with 3D face reproductions, but previous fingerprint recognition technology could be fooled more easily, Tagliavini says.

Coming soon: to your pocket
IHS Markit has seen differing origins for the iPhone X's "structured light" dot projector. Tagliavini had suggested that the primary supplier was ams (Austria), through its US sensing subsidiary Heptagon, which can produce its own VCSELs.

However, the late-2017 news that Apple had placed guaranteed orders enabling California-based Finisar to scale its VCSEL production confirmed Finisar's key position in the supply chain.


The bar along the top of the front of the iPhone X contains the VCSEL-based dot illuminator, time-offlight sensor, and other components that form its True Depth camera. Credit: Apple.

Tagliavini says that Philips Photonics (Germany) may also have a small share of the iPhone X market, while he thinks that Lumentum (USA), despite persistent links with Apple, seemed not to be involved in the iPhone X.

The analyst notes that Finisar didn't report significant earnings from high-power VCSEL arrays for sensing in the quarter ended July 2017, due to a change in its processes expected to drive production of next-generation sensors. The December announcement that Apple would spend $390 million on Finisar's VCSELs clarified the relationship, and as a result, Finisar will refit a former Texas Instruments silicon semiconductor fab in Texas to GaAs wafer processing.

While Philips Photonics' general manager Joseph Pankert was unable to comment on any direct involvement with Apple, he emphasizes that the firm supplies VCSELs both for sensing and data communications. Philips Photonics also covers a broad spectrum of industrial applications, including sensors, but also combining many VCSELs to produce lasers for materials processing.

Nevertheless, Pankert foresees that the iPhone X's trendsetting will prove especially fruitful. "I expect that all smartphones in the world will eventually adopt sensor technologies based on VCSELs," he predicts. The technology will then spread to tablet computers and wearable devices, consuming a large quantity of VCSELs, Pankert says.

Consequently, Philips Photonics has also been expanding, including adding a second, much larger cleanroom. "We can double or even treble our capacity by 2018," Pankert says, adding that further scale-up could be achieved with foundry services, if needed.

The company is also hiring more engineers to build up its technical capabilities, which Pankert believes will be just as important. "We can accommodate quite different wavelengths, product designs, and different chip sizes," he says. "That level of flexibility is necessary as I believe that VCSELs will become more sophisticated.

"We really have to think about how to integrate VCSELs with a driver, although not necessarily on the same chip. It's very much about design capabilities, and working with customers to make a full system that works optimally," he says.


A Finisar wafer on which VCSEL devices are fabricated. Credit: Finisar/Apple.

Every existing commercial device stays "well below" the laser power limit needed for safety, in particular to avoid damaging people's eyes, Pankert stresses. For mobile devices, optimal function includes minimizing VCSEL power consumption and heat production, he points out.

Yet that's quite a change from VCSEL technology that originated in telecommunications, notes Chuni Ghosh, general manager at Princeton Optronics (USA). For that application, devices didn't need to be optimized for high efficiency. By contrast, Princeton Optronics' technology originates from 10 years of development, funded by the US military, specifically to optimize this parameter. "We have the highest efficiency and highest power density," Ghosh claims. "No one even comes close." As a consequence, ams snapped up Princeton in March 2017, complementing related capabilities the Austrian firm is cultivating.

Long-awaited overnight success
Unlike Finisar, Lumentum, and Philips Photonics, Princeton Optronics operates on a fabless basis, manufacturing its device using semiconductor foundries and device packaging companies in Taiwan. "We're able to deliver many thousands of 6-inch wafers per month," Ghosh says, equating to the hundreds of millions of pieces per year required by cell phone OEMs and taking advantage of existing GaAs wafer capacity to provide power amplifier components in their billions.

Other parameters are also important, depending on the application, and Ghosh cites examples in TOF sensing and lidar. Getting the best depth resolution in both means producing the shortest pulse possible. That in turn means minimizing the time it takes for VCSELs to rise to and fall from their peak output. Ghosh stresses that Princeton Optronics' devices can produce pulses of 50 picoseconds or less.

Similarly, with structured light, depth resolution requires bringing spots as close together as possible, without them overlapping. That demands a very narrow divergence to provide optimal resolution, where Ghosh again claims Princeton Optronics excels.

This year will mark the rise of a major market for VCSELS in smart phones, with Ghosh expecting handsets to include eight or more devices. "Other people think it came up suddenly," he says. "But we've been working with customers now for more than five years. It's not that sudden."

Some companies will also be buying VCSELs from Princeton Optronics for self-driving cars in 2018, Ghosh adds. He suggests autonomous vehicles will become a high-volume market from 2020 onwards. By then as many as 20 million cars could be equipped with lidar facing in all directions, he predicts, adding: "The market could be pretty big, but not as big as cell phones."

If lidar does penetrate the automotive sector in a big way, systems will eventually cost only a few hundred dollars, says Philips Photonics' Pankert, although they're currently much more expensive. VCSEL costs would be only a tiny part of the total, he adds.

Yet switching from smart phones to cars would involve extra considerations, Pankert says. Devices need to work in a specified range of temperatures from -40˚C to 125˚C. Also, in order to provide sensors for self-driving cars, lidar needs a range of at least 200 meters, which Pankert concedes is "pretty challenging" for VCSEL technology.

The market for automotive lidar systems will grow from $290 million in 2016 to $2.7 billion by 2026, according to another IHS market analyst, Akhilesh Kona. Emitter devices will be crucial in enabling this technology, and the most commonly used devices are currently edge-emitting laser diodes, not VCSELs.

Kona says that XenomatiX in Leuven, Belgium, is currently the only lidar system producer that he knows of using VCSELs, although Canada's LeddarTech is also working on it. "The real challenge for VCSELs is that it's a state-of-the-art technology, something being developed right now," said the analyst. In contrast to the three- or four-year lifetimes demanded in smart phones, automotive-grade devices have to last for more than a decade. In addition, getting enough power out of 800-900nm wavelength VCSELs to reach the 200 meter-plus ranges demanded raises eye-safety concerns.

One advantage is that it's easier to create a two-dimensional array of devices with VCSELs than with edge-emitters, Kona says. Another is that thousands of VCSELs can be integrated into a single, two-dimensional package. That could help address the eye-safety concerns, as energy can be spread among them, rather than being concentrated in a single beam.

Gesture recognition VCSELs may also find use in cars before lidar or autonomous vehicles become commonplace, Kona adds. Examples could include recognizing facial expressions or gestures within the car, an application being worked on by TriLumina (USA).


As well as growing demand in sensing, cloud computing and video streaming are improving the prospects for devices like this datacom VCSEL from II-VI. Credit II-VI.

II-VI sounds a note of caution about the prospects offered by lidar for autonomous driving. "We believe that the market potential is still not adequately sized," says Mark Lourie. "It will depend on the laser technology selected if any, [as well as] cost targets, whether autonomous ride-hailing services may have the effect of reducing the total number of cars sold, and whether there will be applications in autonomous vehicles other than autonomous cars."

Instead, VCSELs are an important part of his company's growth strategy in communications and elsewhere in sensing. "II-VI believes that 3D sensing in particular will become a very large market that will eventually be served by a reliable base of suppliers with vertically integrated and scalable manufacturing capabilities," Lourie says.

"In principle, as long as VCSELs from different vendors meet a specified form, fit, and function, they should be interchangeable for that specification. Often times, the challenge is in translating the full set of requirements into specification parameters that can be measured accurately enough or guaranteed by design. Beyond these requirements, II-VI's VCSELs are often differentiated in terms of conversion efficiency and reliability."

The promised growth relies on the ongoing convergence of computing, communications, and consumer electronics, Lourie adds, bringing together big data analytics, cloud computing, broadband wireless, artificial intelligence and sensing.

"This convergence will continue to enable new disruptive applications," he says. "We believe that semiconductor lasers, including VCSELs and edge-emitting lasers will continue to play an important role both in communications and sensing."

IHS Markit's Tagliavini has a more specific vision: one where VCSELs could transform applications related to security, access, and augmented and virtual reality, where an interface between humans and machines is needed. "This is transforming the user interface, since you don't need to touch buttons or screens or to talk with a microphone," he says. Machines can "read" your intentions simply by observing your eyes or reading your lips as they move. "It's really changing your interactions with appliances and devices," the analyst adds.

Pankert says this answers an often-asked question about why the VCSEL market is exploding now, even though the basic device was invented nearly four decades ago. "It feels a little bit like LEDs 10 or 15 years ago," he explains. "LEDs had also been around for many years, and it was only after certain applications had matured that they really took off. The applications that drive VCSEL consumption are only ready now.

"All the sensors that are being built into either mobile devices or cars? Well, they haven't been around for the last 10 years. The same may be true for other applications where there hasn't been time to adopt this technology. It's all changing."

Andy Extance is a freelance science journalist based in the UK. This article was originally published in the April 2018 edition of SPIE Professional magazine.

Related SPIE content:

Short-range links beyond 100Gb/s with vertical-cavity surface-emitting lasers
Effective 100Gb/s single-mode optical transmission can be achieved over 100m of multimode fiber by using advanced modulation schemes and digital-signal processing.

Oxide-confined leaky vertical-cavity surface-emitting lasers for single-mode operation
Lateral photonic integration of oxide-confined leaky vertical-cavity surface-emitting lasers enables their application in data communications and sensing.

Ultrathin and simple design for vertical-cavity surface-emitting lasers
A novel monolithic grating can be designed to reflect light of almost any wavelength and can be used in integrated optoelectronics systems.

Vertical-cavity surface-emitting lasers for optical interconnects
New energy-efficient and temperature-stable gallium arsenide devices operate at 980nm and enable high-speed data transmission at high temperature.