Semiconductor exposure equipment are, in a nutshell, monster cameras. They use a massive lens unit of several hundred kilograms to expose circuit patterns on semiconductor wafers, and so really are like monster cameras. These devices are the biggest products that Canon makes, and the most expensive too (costing several hundred to several thousand million yen), but because they tend to be located deep within the bowels of semiconductor factories, not that many people have ever seen one. And these devices go on evolving there, hidden from sight, one generation replaced by another in quick succession. That's why I sometimes feel genuine envy for other occupations, such as architecture, for example, where what you create remains for at least a while and gets to be seen by others.

The optics of a semiconductor exposure equipment

In the ongoing effort to improve the performance, semiconductors are evolving with their electronic circuit pattern widths being rapidly reduced. The narrower the linewidths, the faster they become. This also enables to downsize semiconductor chips with advanced functions.

As such, it's the resolution which depends on the capability of semiconductor exposure equipment defines the performance of the chips. And the resolution is set by the wavelength of the light source, and a figure referred as NA that represents the performance of the projection lens.

Because light source wavelength and NA evolve in short cycles, exposure equipment technology is faced by an endless stream of challenges. Wavelength has been changed every seven years— g-line, i-line, KrF to ArF- going through several generations of ICs. Each time, it has been reduced by about 30%.

NA on the other hand has become bigger and bigger. I was always getting told off by my boss. I would design to the absolute limits of the existing technology at that time, and tell my boss, "This lens's NA is the best I can do, so let's go with this." Then a year later, a bigger NA becomes able to be achieved , but when I tell my boss, he'd turn on me with "What? You told me last year that you couldn't do any better!" (laughs). In that respect, you could say that development of exposure equipment involves constant bluffing.

Of course, there are sound technological reasons for the increase in NA. You work under some constraints at the time, but going beyond that limit automatically becomes the next target. Surpassing the existing limit is impossible without some kind of new innovation, and you're feverishly doing this and that to prepare for the deployment of the next generation technology by challenging the constraints. And then that new technology becomes a springboard to yet an even more advanced stage.

I think it's probably because exposure equipment technology is profound enough to absorb one new technology after another that I've been fascinated with it throughout the years.

Equation: exposure equipment resolution

Resolution of exposure equipment is calculated with the following equation:

In short, resolution can be improved by shortening wavelength of the light source, and increasing NA.

Table: Evolution of exposure equipment light sources and their wavelengths

In a frontier field like this, existing data is often not much help, and you just have to go ahead and do all the verification yourself.

For example, unlike the excimer lasers used in the previous generation of exposure equipment, the 157 nm light that an F2 laser produces can't pass through quartz. As a result, out of the limited number of optical materials available, the only one that works as lens material for F2 devices is fluorite.

However, even if you know that theoretically fluorite should let F2 laser light through, past data like the Chronological Scientific Tables tells you that it can't transmit 157 nm light. This is actually due to impurities – fluorite pure enough to let such light through just didn't exist. Consequently the first step in developing the next generation of exposure equipment was developing a pure enough fluorite, and we managed to raise purity in time to a level of over 99.5% transmittance per centimeter. Having to consider everything from the basic data of raw materials afresh like this makes development pretty difficult at times, but it's fascinating. It's this kind of interesting challenge that I'd like more people to know about.

What makes development in frontier fields so special

Canon began developing exposure equipment with the launch of what it called its "Micron Project" in 1970. I joined Canon in 1973, and have been involved in this field from the development of devices called PLAs – parallel light mask aligners that use contact/proximity exposure, with a mask and a wafer in close contact or proximity. After PLAs, I worked on MPAs, mirror projection aligners that use a unit magnification mirror system to guide light to a wafer. In relation to this work, I obtained a patent for mirror image formation theory in 1976, and wrote an internal paper that later won the Optical Paper Award for excellent papers. I went on to work on actual device development and won the Prize of Japan Society for the Promotion of Machine Industry too.

Mirror projection is used even nowadays in exposure equipments for producing LCDs for flat panel displays. Of course these equipment and their mirror units are a lot larger now than they used to be at that time, but the fact remains that basically the same optics system has been used now for close to 30 years, which is a very long time in this day and age – and that, for me, is very gratifying.

I was also involved in the development of the automatic-alignment system used in our PLAs and MPAs. It was the first automatic-alignment system of its kind worldwide, in effect an ultra-high precision robot that uses light to make measurements and adjust position accordingly. I was responsible for the optics, and whenever a hitch arose in development, the electronics chief would complain that it was the optics that was at fault. We would check it out, and discover that it was the electrics that were to blame. This kind of exchange took place on a daily basis, and through it we'd iron out one problem after another, until finally, the day before we were due to exhibit the system at a trade show, everything clicked. I'll never forget the thrill of finally getting the equipment worked as it was supposed to.

I also worked on automatic-alignment systems for steppers. What I remember most vividly about stepper development is the U lens (Ultra-high resolution lens). In the first stages of development, our stepper projection lenses were unable to deliver the kind of quality that our customers were looking for. It was an aberration problem. I took a look at the design specifications, and realized that there was a flaw in the development system. This was caused by the fact that we were using a development system borrowed from cameras at that time, and it failed to address certain areas of stepper lens design. I immediately brought our lens design system software development people in, and we rushed out a new lens development system. Thanks to the wholehearted cooperation of our plant, we were able to go into production quickly and manufacture good lenses. I think it's a reflection on Canon's strengths that we were able to do this in such a short time, and thanks to such efforts, our steppers went on to develop a high reputation.

It's such aspects of development that make it really fascinating for an engineer. Things don't lie. If you mess up in some way, it'll show up in the results, and then what's important is to put things right as much as best you can. To do so, you need to take a logical, step-by-step approach – an "OK, if we try such and such, it should give us such and such" kind of approach.

For example, this was very much the case with the quadrupole illumination I developed for our i-line steppers. It was around 1990, when I was writing up my doctoral thesis on ultra-high precision optical systems for semiconductor exposure equipment. To write my resume, I was checking through my past work, and came across something which made me sit up and think, "Wait a minute, what if ...," and that was when I clicked to this new technique, which involved illuminating a reticle from four specific directions. When I looked into the possibilities, it seemed feasible, so I broached the idea to the whole team, and they were all enthusiastic, like "Let's do it!," as a result of which a new system was born.

The timeline of exposure equipment development at Canon

I'm currently working on ArF immersion technology (a hot area at the moment), measurement technology, and EUV, which is destined to become the next generation exposure technology. EUV has a very short wavelength, only 13.5 nm. Exposure equipment using ArF immersion technology will probably be our mainstay in the second half of the 2000s, with EUV taking over in the 2010s. We're always facing a string of new challenges.

I've been with Canon now for over 30 years. I've spent more of that time in product development group than on the pure research side at a laboratory. I think of myself as an engineer rather than a researcher. Exposure equipments are the kind of thing in which research and development go hand in hand. There are some products for which development is more just a matter of design, but for leading-edge exposure equipment, you have to have a good grasp of both design and theory. That's why I think of myself as an engineer – a combination of researcher and engineer.

Creating things as an engineer is really fascinating – the rush of excitement you get, for example, when an idea suddenly comes to you, and you realize that new possibilities have just opened up, and then the fun of nursing the idea through development and seeing it take shape as a product, and the kick you get out of seeing people use that product. The thrill of R&D is something I'd like as many people as possible to experience.

Learn more about Semiconductor exposure equipments