Loading summary
Transcript1 lines
- [00:02]
A
For the first few decades of its existence, all lasers were side firing lasers, meaning the beam comes out of the wafer's side horizontally. But in the late 1970s, a new type of semiconductor laser emerged, one that fired out of the wafer surface vertically. Yes, it sounds a bit weird. At first, nobody had any idea what to do with it. But over time, the technology has been adopted into a wide variety of everyday applications. Today, it literally shines into people's faces in today's video, the little vertical lasers that everyone uses. The concept of a laser involves light bouncing between two mirrors facing one another. One of the mirrors is fully reflective, the other just partially. So we call the space between these two mirrors the optical cavity. The laser must also have an active or gain medium. It is made from any material capable of producing and amplifying light, solids, gases, liquids, or semiconductor material. To get the laser working, we pump energy into the gain medium, either using electricity or another light. It generates light, and when light passes through it, it can add yet more light amplification. So the laser light bounces back and forth between the two mirrors and amplifying because of the gain medium. At some point, the light gets strong enough to burst through the partially reflective mirror out towards the target. With a side firing mirror, the light bounces back and forth inside the optical cavity in a horizontal direction, accumulating gain. Then, when the laser fires, it fires out of the wafer's edge. Now we move to Japan. Tokyo Institute of Technology. A graduate student named Kenichi IGA is looking for a topic for a graduation thesis. He applies to do laser research under the legendary professor Yasuharu Suematsu, who would go on to do pioneering work in the field of fiber optics. For his master's thesis project, Suematsu suggested to IGA to rebuild Ted Maimon's first practical laser, which used the ruby crystal as the gain medium energized by a flash lamp. So in the autumn of 1962, IGA replicates the paper using a ruby crystal borrowed from a colleague. It worked, and he fell in love with laser technology. The laser that IGA built was what we call a multimode laser. This means that the laser emits multiple modes or colors. By contrast, a single mode laser would be a laser that emits just one pure color. Because the laser had multiple modes, the beam it shot out was messy and scattered. Igo, with Suematsu's help, intuited that this was because the laser was bouncing light in too many directions inside the ruby crystal medium. So to make it so the laser bounces fewer modes of light. IGA reduced the physical size of one of the mirrors in the laser. This blocked the messy light and let the laser output a much tighter beam. They wrote up the result and it got published, which in those days was still a rare achievement for a Japanese. By the way, before we continue, I want to flag the book the Vicksel Odyssey as a fascinating chronicle of the Vicksel's invention. Written in part by IGA himself, it features interesting anecdotes from the start of the optical revolution a decade later. IGA is now an associate professor at the Institute studying semiconductor lasers. Industry interest in such lasers was growing. In 1966, Charles Gao proposed that extremely pure optical fibers can carry light signals a long way with relatively little loss. Additional work on the nature of glass fibers in 1973 by the American company Corning predicted that wavelengths between 1.4 and 1.7 micrometers would have the fewest losses when sent through silica fibers. With regards to long distance telecom, it was important that the lasers firing signals into these optical fibers be able to consistently fire at a single wavelength. Different light wavelengths travel at different speeds through optical fiber. So for a multimode laser sending signals through a long distance fiber, there was the risk of the different modes arriving at the destination at different times, causing annoying mix ups. Yasuharu Suematsu identified indium gallium arsenide phosphide as a promising material to make these semiconductor single mode lasers. Can such a bulky four element compound be reliable enough for extended use? IGA worked in Suamatsu's laboratory trying to produce the crystal growth system for such a laser. While doing so, he started to ponder whether there was a better way to manufacture these items. As I mentioned, all lasers in those days were edge emitting lasers. So the light inside the optical cavity bounces left and right in a horizontal direction and fires out of the laser's edge. In practice, this optical cavity consists of a small laser chip. Back in those days, you produced those laser chips using a complex multi step process. First, you deposited layers of a desired semiconductor material like indium gallium arsenide phosphide onto a wafer. After two steps called patterning and metallization. We then cleave the wafer along a specific crystal plane to get strips of semiconductor material like as if we're making pasta. In the early days, this cleaving had to be done with a surgical knife and a little nerve, presumably granted by a swig of Suntory whiskey. The cleave strips are then Diced into laser chips of the right size using a diamond saw. IGA was frustrated by this extremely manual production method. There was no way that it can scale to millions of units. Variations in the cleaving make the laser's performance inconsistent, causing the laser to fire additional unwanted modes. Considering how small these chips can get, this was a significant manufacturing challenge for some laser designs and modes. The length of the cavity and thus the size of the laser chips must be as small as 50 micrometers. What sort of laser design can get you single mode and be consistent and reliable with that single mode while also being manufacturable at scale? Existing designs did not seem capable of it. IGA Rice that the idea for a vertical firing laser first came to him in in a midnight dream on March 22, 1977. As soon as the idea appeared, he sketched it down into a research note the next day. He immediately began working on the design idea, recognizing that it fulfilled all the requirements that he wanted to solve. The idea was exceedingly a single mode laser firing vertically from the surface of the wafer rather than its edge. The concept is the same just turned on its side. The gain builds up and down vertically rather than horizontally between two mirrors inside an ultra short cavity. We can manufacture this laser monolithically in a fab by carefully depositing layers of semiconductor materials to create your gain medium and the two mirrors. No more cleaving crystals with a knife by hand. There are other benefits. Side firing lasers have a rectangular exit window, so when the laser beam fires through it, it comes out asymmetrical. An ovalish elliptical shaped beam that can be challenging to focus. This surface firing laser, on the other hand, can fire a beam that is perfectly circular, making it easy to either focus using lenses or couple to an optical fiber. IGA decided to first announce the idea at a conference to gauge people's interest. So in March 1978 he first presented it at the 25th Spring Meeting of the Applied Physics Societies in Japan. Iga recalled people saying that the idea was interesting but the feasibility unlikely, especially in industry. The thing would eventually be called the vertical cavity surface emitting laser, but that name was coined in the late 1980s. For now, Iga called them surface emitting lasers. Iga and his team demonstrated a non practical surface emitting laser concept in 1979. The laser's active medium was gallium indium arsenide phosphide deposited onto a substrate of indium phosphide. The optical Cavity was about 6 microns when first presented. VCSELs were far from the groundbreaking inventions that they eventually turned out to be. And there were a few reasons for this. On the technical side, the VCSEL had higher threshold currents than that of their side firing counterparts. The threshold current is the minimum current required to turn a laser on. A high threshold means that the laser cannot lase unless we pumped a very large current through them, leading to high power consumption and heat generation. The heat was a legit problem as it can cause a semiconductor laser to shut down. That first 1979 VCSEL operated at temperatures of 77 Kelvin and thus did not fire a continuous beam. Another issue concerned the so what? Many side firing lasers were being announced. It was an exciting time in laser research history. Now here comes this funky surface emitting thing with worse performance. People were not sure what they can use it for. Thus the skepticism. In 1979, IGA went to the United States for a year and a half to study at the legendary Bell Labs where he worked on laser concepts unrelated to the surface emitting laser. But upon returning, he dedicated himself to his idea. The key technical hurdle to overcome was the high threshold current, which is responsible for the heat issues. The high current was because the laser's optical cavities leaked too much light. One way to fix that was to shrink the size of the optical cavity, which in turn means shrinking the active medium region between the two mirrors. A smaller active medium means needing to make the mirrors much more reflective than before, something to the order of 99% or more. The mirrors must also be reasonably electrically conductive so that current can pass through to reach the active region. This required acquiring various sophisticated epitaxy and metal organic chemical vapor deposition machines to deposit thin, high quality films of dielectric and metal semiconductor materials. In late 1988, published in 1989, Iga and his team announced the first VCSEL that can generate a continuous laser light whilst operating at room temperature. Shortly after or about the same time, a Bell Labs team led by Jack Jewell presented their own continuous wave room temperature vcsel. What is notable about Joule's VCSEL is that it used a quantum well for the active region and Bragg reflectors for the top and bottom mirrors. Quantum wells are very small semiconductor layers that trap charge carriers and force them together in order to emit a lot of light. Bragg reflectors are mirrors made from many alternating layers of reflecting materials. Combined they can achieve reflectivities of 99% or more. We use them in EUV machines. Two different technical approaches to the same problem. Joule's Wixel was a bit weird in that it fired a laser out of its bottom through the substrate. IgA still suffered moderate heat issues and worked better when bonded to a copper heatsink. But the end results are the same. VCSELs became a practical technology, arriving on the scene at exactly the right time. Researchers in the United States finally picked up on this progress, encouraged by shifts in US government technology spending after the end of the Cold War. Optoelectronics attracted a lot of funding to due to the rise of data communications and the Internet, which are shaping to demand different things from optical fiber technology. Older systems developed by Bell were geared for long distance telephone calls, high performance, high cost lasers and precision optical fiber for sending signals thousands of kilometers. But the dynamics of data communications are different. The distances tend to be shorter, but the data volumes to be transferred are much greater, emphasizing lower cost multimode fiber and laser solutions. For the vcsel, the turning point was gigabit Ethernet, an updated version of the popular Ethernet local area network standard. The target was to send a gigabit of data in both directions and the current LEDs were not good enough. People investigated edge firing lasers then used for CD players for this, but discovered those suffered low reliability. The structures of the VCSELs are a bit more complicated, but for that you get several key benefits, the most significant being cost. We can produce VCSELs at huge volume, oftentimes up to a million lasers on a single wafer, making them very cheap. And since their light comes up from the surface, we can easily test all those lasers before cutting them out of the wafer. Not the case with edge emitting lasers. VCSELs can also be easily packaged like LEDs and coupled to a fiber for digital modulation. And as I said earlier, their beams are round shaped rather than rectangular, which offer optical benefits. In the mid-1990s, big companies like Motorola, Agilent and Honeywell started producing VCSELs. And in 1996 Honeywell broke the plane in publicly presenting production level data and actually starting to offer the VCSEL to the market in volume. Spurred on by the telecom boom, venture capital poured into vcsel, sprouting many startups to develop higher performing lasers. Things got a bit out of hand. Jim Tatum of the VCSEL pioneer finisar Corporation estimated 30 startups during the peak, which was obviously too much. Many went under or were acquired after the bubble popped. The carnage in the post bubble datacom industry led companies to experiment with alternate use cases for VCSEL technologies. One that popped out was the optical mouse, which replaced the old school trackball mice. Optical mice work by illuminating the surface with some light source and then a small CMOS sensor takes pictures of the surface, comparing the results to intuit the distance of travel. The first optical mice used LEDs for that illumination vixels fire a more coherent beam of light, making it easier to discern travel on weird surfaces like metals or glass. In 2004 Agilent integrated the VCSEL into engine modules that allowed vendors to make laser enabled optical mice. Other companies like Philips followed and and today over a billion mice have been made. They have not quite obviated the LED like they did in gigabit Ethernet, but sit comfortably in the high end. After the mice, 3D sensing has been the next major use case. VCLs are ideal for consumer devices because they do not use a lot of energy, do not generate a lot of heat, emit a tight spectrum of light and are physically small. We can mount them right on the circuit board. There are two major methods of doing 3D sensing with structured light and time of flight. This first method, structured light, involves projecting a dot pattern over a surface and observing how that dot pattern deforms over time. We can interpret that data to understand the movement. The first major consumer gesture recognition system was the Xbox Kinect, which gained a lot of hype in 2013 as this new form of gaming. The first version of the Kinect for the Xbox 360 used structured light but suffered issues due to gaps in the dot pattern. Thus the second version of the Kinect switched to using time of flight which was seen as more accurate. Time of flight is where we extract a 3D image by measuring the time it takes for the light to do a round trip from the VCL array to the thingy and back. Despite the hype, the the Xbox Kinect did not quite make the same type splash as the Nintendo Wii. The VCL arrays did find their big breakthrough with facial recognition technology. On the mobile phone. These used structured light to compare the dot pattern of a person's face with that inside the phone face ID. Apple's iPhone X, which hit the market in November 2017, brought this market to the big time and has remained a core technology for mobile phones ever since. VCSELs remain important for data communications, but they seem to have finally hit their limits, especially for transmitting data over medium to long distances. The current hotness thus for V seems to be in lidar for automobiles, an application that has been growing a great deal and seeing a lot of competition thanks to EVs and self driving cars. Today they compete vigorously with traditional edge emitting lasers for dominance in the lidar stack, offering their traditional advantages of reliability, cost and stability. Their downsides being lower efficiency and lower brightness. But vendors continue to work on improving those weaknesses. New structures are continually being experimented on to extend detection range and resolution. The little laser that everyone uses continues to find new use cases. Alright everyone, that's it for tonight. Thanks for watching. Subscribe to the channel, sign up for the Patreon and I'll see you guys next time.