Presents on LPP Source System Development for HVM at SEMICON West 2011
One of the highlights at SEMICON West 2011 was the advanced lithography session that kicked off July 13’s TechXPOT. Headlining the session, Cymer’s David Brandt, senior director of EUV marketing, addressed the crowded audience with an industry update on the company’s latest development of laser-produced plasma (LPP) Extreme Ultraviolet (EUV) technology and source production for the semiconductor industry.
Overall, he said, the outlook remains positive with increased investment and steady traction since the last formal update from Cymer at SPIE 2011. Of note on the EUV source technology front, Cymer reports that two 3100 EUV systems are installed and operational at chipmakers and are realizing improvements in run time, system availability, and stability to enable sub-30nm lithography. The third chipmaker installation is currently in process.
Furthermore, Brandt announced that five of the eight first generation 3100 HVM I sources have shipped to customers and are fully operational. The other three sources are in use in San Diego for testing and validation of designs and upgrades. The Cymer engineering teams completed several long duration runs in order to evaluate EUV power, dose stability and collector protection.
There were comments throughout Brandt’s presentation that grabbed the crowd’s attention, but certainly a standout was when he confirmed the worldwide total EUV energy created by these sources. According to Brandt, “As a result of increased source availability and stability improvements, Cymer’s eight sources have cumulatively produced, since March 2011, approximately 40 megajoules (MJ), energy sufficient to expose nearly 5,000 wafers depending on dose sensitivity.” The takeaway? As productive time increases, Cymer generates more learning cycles.
To track worldwide source availability, Brandt discussed semi productivity parameters on the eight sources, indicating that overall the mean availability was approximately 50%. The other 50% is needed for maintenance primarily on two modules, which are actively being improved with the outlook of 80% achievable by end of the year. Cymer is rapidly ramping up its field support organization to support the HVM I implementation.
In terms of performance, Brandt explained that Cymer’s HVM I source in San Diego has demonstrated <0.2% dose stability over more than 99% of fields for marathon wafer-simulation runs. In similar tests Cymer has demonstrated up to 200 hours (laser-on) of collector protection prior to 50% loss of reflectivity.
At this point in the presentation, Brandt switched gears to discuss the development and production-readiness of Cymer’s second-generation light source. Cymer reported that HVM II source architecture is complete and detailed design is in progress for ASML NXE 3300B scanners. He went on to disclose the guiding drivers of HVM II sources, indicating the second-generation drivers as the following:
- Vessel/collector geometry change
- Increased EUV power
- Reduced installation and service time
With the above in mind, early projections from Brandt place the first shipment of HVM II (3300) sources in Q1 2012. At that point, the power and collector lifetime will be similar to that of the 3100 source, while meeting all other 3300 requirements. Overall, source performance will be increased through a series of upgrades for power and collector protection. Changes include:
- Higher EUV power required
- Increased source orientation angle (increases scanner optical throughput)
- Higher NA collector
- Smaller source vessel
- Designed for improved serviceability
Anticipating the question about forward-looking plans, Brandt projected third generation (HVM III) source development as beginning in 2013.
The formal TechXPOT sessions ran over the allotted presentation time and as a result, a formal Q&A discussion was tabled, but we did catch Brandt for a few additional comments on what he’s seeing at SEMICON West this year—and what today’s issues and trends mean for the industry. Brandt summarized, “Demand for EUV is at an all-time high and all chipmakers are projecting soaring costs associated with 193nm immersion double patterning techniques. Although there are many creative techniques to work around not having an exposure tool with shorter wavelength, those solutions are far more expensive than having an EUV source and scanner to do the same job in a single exposure.”