EUV sources achieve higher power, but need greater collector lifetimes, writes Vivek Bakshi of SEMATECH's technical staff.
According to the International Technology Roadmap for Semiconductors (ITRS), Extreme Ultraviolet Lithography (EUVL) is an excellent option for printing circuits at 45nm or below. In EUVL, invisible light of 13.5nm wavelength is used to print circuits. As light is strongly absorbed at this wavelength, the entire stepper system must be in high vacuum and all optics are reflective, not refractive.
EUV light is generated by heating a fuel material (xenon, tin or lithium) to very high temperatures to create plasma. Such heating is done either by magnetic compression or with laser light. Systems that use magnetic compression are called Discharge- Produced Plasma (DPP) sources, while those that utilise laser light are called Laser-Produced Plasma (LPP) sources.
For EUVL to be cost-effective for High-Volume Manufacturing (HVM) a wafer throughput of at least 100 wafers per hour is needed. This required throughput in turn results in a power requirement of at least 115W at the intermediate focus for the EUV sources.
Many suppliers are currently working to produce EUV sources that meet requirements for high volume manufacturing implementation of EUVL in 2009. Leading EUV source suppliers include Cymer Inc (USA), Philips Extreme (Germany), PLEX LLC (USA), PowerLase (UK), and Xtreme Technologies (Germany), as well as consortia such as EUVA (Japan) and EXULITE (France).
Up to now, source power has been the most difficult requirement to fulfil for EUVL implementation in high volume manufacturing. One should note that not all of the power generated by the source is collectable, and only a fraction of this power is collected at the intermediate focus.
The losses come from limited geometrical collection efficiency of collector mirrors, imperfect reflectivity (<65%) of collectors, loss of EUV light in debris mitigation devices and spectral purity filters, and absorption of light by background gases.
EUV SOURCE STATUS
In November 2004, SEMATECH sponsored its semi-annual EUV Source Workshop in Miyazaki, Japan, in conjunction with the Third International Symposium on EUV Lithography. At the workshop, suppliers reported results showing that EUV source power of up to an estimated 50W at intermediate focus could be achieved from tin DPP sources.
However, the power generated by tin DPP comes with the added problem of tin debris, which coats the collector mirrors and reduces their lifetime. As tin is more efficient (by a factor of two or three over xenon) in producing EUV light, it has been preferred by suppliers (suppliers quickly realised that xenon, although a noble gas and much easier to work with, was only 1% efficient in producing EUV light). Although favoured for its higher conversion efficiency, mitigation of tin debris is a big challenge, since tin is a metal, and because tin plasma cools, it forms a metallic coating on surfaces.
Specifications for collector lifetime call for collectors to last for at least three months for a beta-level tool and up to one year for a production tool. The lifetime of a collector is considered completed when it has lost 10% of its peak reflectivity. To account for wafer throughput, the lifetime of collectors is also measured as the number of pulses of EUV light that a mirror can withstand before it needs to be replaced.
This lifetime requirement is estimated at 10 billion pulses for a beta-level source and 80 billion pulses for a production-level source. The relationship between times and pulses is based on the throughput model for an EUVL stepper, which uses other parameters of an EUVL system that are commonly used to generate the required specifications.
LPP EUV sources have shown progress with xenon-based targets, with suppliers expecting up to 5W of power at the intermediate focus. For such sources, however, the laser power has emerged as the key challenge. At present lasers with 1,500W power have been demonstrated to generate up to 5W of power at the intermediate focus. Today, suppliers are developing lasers with up to 5,000W of power using either Nd:YAG or CO2 lasers or via multiplexing of lasers. More than 30kW of laser power is needed to generate the required EUV power using xenon as the fuel, and much work still needs to be done to demonstrate the feasibility of this level of laser power for EUV sources.
The progress in source power reported at the EUV Source Workshop was so significant that the EUVL Steering Committee, which organises the EUV Symposium, decided to de-emphasise source power as the top-rated EUV issue. The Steering Committee has drawn up a revised list of challenges for EUVL.
Defect-free mask availability is now a high-priority issue for EUV, and lifetime for source components and collectors is now the number two challenge. Source components are defined as electrodes and collectors for DPP EUV sources, and collectors, nozzles and laser diodes for LPP EUV sources.
While the EUV focus is shifting from source power, many challenges for EUV sources remain to be solved in order to meet the requirements for an EUVL stepper for HVM.
Another significant challenge is the thermal load on collector mirrors. Part of the power generated by sources is absorbed by the collector mirrors, making them extremely hot. Technology must therefore be developed to cool the mirrors.
A related issue is Out Of Band (OOB) radiation of EUV sources, which generate light in all wavelengths. However, recent studies have shown that OOB radiation exceeds the recommended specifications, and consequently Spectral Purity Filters (SPFs) will be needed. This creates another problem, as some energy will be lost to SPFs, requiring the generation of additional EUV power to meet EUV power requirements.
In the coming months, the EUV industry needs to monitor the progress of tin debris mitigation, as this could become a 'showstopper' for tin DPP EUV sources. In addition, the industry must also continue to evaluate LPP EUV sources for their potential as EUV sources for HVM. A close eye must also be kept on the development of high-power lasers, as they are the key enablers for the feasibility of LPP EUV sources.