Veeco Puts Faith In 'future-proof' Tool
The fundamental reason is that as the LED market grows, volume is going up, so you have to come up with more productive tools that reduce the cost of ownership. Another reason is tighter technology specifications, such as higher brightness. Every time the LED cost-versus-brightness curve moves, we have to come out with a new version of MOCVD tools.
So what is different about the K300 and K465?
The key thing is the modular platform, which is upgradeable over multiple technologies and gener-ations. The K300 is a bridge tool for people who need immediate capacity. It uses exactly the same process recipes as our previous GaNzilla II E300, but its design makes it easier to configure gases and materials.
Owners of the K300 will be able to upgrade to the K465 in six to nine months if they see the need for higher production, but people can also buy the K465 directly. In the K465 the biggest change is the reactor. We are going from a 21 × 2 inch wafer capacity [which is used in the GaNzilla II and K300] to a 45 × 2 inch capacity.
Is this larger capacity behind the claim that the K465 offers 50% more throughput than any other MOCVD tool currently available?
The K465 has two big advantages. The first one is wafer capacity, which is 8–40% higher than the nearest two competitors.
The second is unique to the Veeco platform. We have an automatic load-lock transfer mechanism that allows for continuous operation of the MOCVD system. While a platter of LEDs is being grown inside the reactor you load another wafer carrier in parallel into the vacuum load-lock. Normally after a run you let the reactor cool down before taking wafers out. But in our machine you don t need to cool it because it s under vacuum. Instead, the robot picks up the finished wafer carrier with 45 wafers, transfers it to a hub inside, and then takes the one that s loaded and puts it back in. It s all done within about 10 min, and without lowering the temperature.
Opening the reactor, lowering the temperature and then ramping up the temperature can take at least 1 h on other competitive tools. So for a typical 5 h LED process, you get an extra run each day. The combination of wafer capacity and more runs per day yields a huge number of LEDs. If you do the cost-of-ownership calculation, with yield [from different types of reactors] being equal, we offer at the very least a 50% advantage to the customer.
Do you include gas and bubbler usage in this calculation?
Yes. The K-series reactor has several significant changes. The flow flange – through which the gas is injected from the top and circulated inside the reactor – has been completely redesigned. It cuts down the use of ammonia, one of the biggest costs for LED manufacture, by 25%. We were a little bit behind [the competition], but this has been narrowed significantly.
Why is upgradeability such a big deal?
It is very simple from a customer s perspective: they are protecting their initial capital investment, and can continue to extend their platform over multiple generations.
In addition, there is a point when the footprint becomes a major issue. You can t just keep adding equipment, because what are you going to do with the existing tool? They say: "I just bought a tool a year ago. Now you re coming in with a much better tool, and I ve got to go and buy it. Why can t I balance my manufacturing line and upgrade this?"
Clearly there are benefits to customers. But is it a good strategy for equipment makers?
I know that some people think that it s going to cannibalize existing tool sales. But usually it doesn t. It actually protects and grows your market share in the long run.
Can we expect to hear about any other additions to the K-series in the next few months?
Probably not. We expect that this is going to be the major introduction on the GaN series for this year.