With semiconductor structures forecast to shrink to as small as 32 nanometres by 2009, a European consortium (More Moore) of manufacturers, research institutes and universities conclude that Extreme Ultraviolet Lithography (EUVL) is set to be the technique of choice for high-volume semiconductor manufacturing.
Integrated circuits (ICs) today contain structures no larger than 65 billionths of a metre (nanometres or nm), about the distance a fingernail grows every minute. Yet in the search for greater performance and functionality from their chips, semiconductor manufacturers try to cram ever more functions into smaller spaces. 65 nm structures are expected to more than halve in size by 2009, then fall to 22 nm by 2011.
To print such tiny structures on silicon wafers, most electronics manufacturers expect to call on EUVL technology. Operating at a 13.5nm light wavelength, this technology still presents serious technical challenges, even though it has been studied for 12 years. If chip-makers can overcome these challenges, they could work at higher resolutions than is currently possible with 193nm wavelength photolithography.
“From 2010, EUVL will be used for mass manufacturing at the 32nm scale,” says Robert Hartman, coordinator of the IST project More Moore. His confident prediction springs from technical breakthroughs in the project, among them more powerful light sources and better light projection.
In More Moore, research teams from eight different countries focused either on the manufacturing equipment challenges or the environment around them. The equipment side involved preparing new technologies for 32nm EUVL. The environmental side focused on research into advanced light sources, new types of optics, resists and masks.
“EUVL is the sole technology suitable for optical masks,” says Hartman, referring to the devices that shield certain areas of a chip during the manufacturing process. “The only alternative is electron-beam lithography. But that is very time-consuming — producing at these levels of integration at best four wafers an hour. With EUVL, we should eventually produce up to 130 wafers/hour, matching the rate of today's lithography technologies.”
The sophisticated new equipment required for EUVL will probably cost more than current technology. But Hartman remarks that competing lithography technologies, such as immersion and double-exposure, are also very expensive.
“I believe chip manufacturers would be happy to pay up to a third more than they do today, if they were able to double the density of transistors on a wafer,” he says, noting that current chips contain up to a billion transistors. “EUVL is relatively price competitive for mass production — particularly for the flash-memory chips used in goods such as cameras and telephones.”
More Moore has notched up several world-beating achievements on the optics side. For example it addressed the difficulty of making a very bright light by developing a revolutionary new EUV plasma light source, which is heated to 300,000 degrees Celsius. This light source uses tin as fuel for the short-flash discharge.
Because of the need for very clean equipment, researchers developed a special debris-mitigation system. This unique new system calls on a foil trap, gas jet and/or electro-magnetic fields to reduce the stream of tin evaporated during the discharge.
Under the project, the partners also upped the light source's power output from 120 to 800 watts — a significant milestone towards the EUV target of one kilowatt, as required for volume production of semiconductors.
Another notable achievement was the development of improved high-tech mirrors, used for projecting light onto the wafer. “Our mirrors have very high reflectivity, are highly stable and are cleanable for long periods,” notes the coordinator. “They offer an accuracy of 0.1 nm, which is astonishing given that the distance between two silicon atoms is three times larger.” There were several other important project achievements. Among them were improved lifetimes for EUVL optical components, plus the development of new and experimental materials for making a molecular resist (an energy-sensitive material used in lithography).
For the resist, says the coordinator, “We expose molecules rather than a polymer during the lithographic process. Our goal is to achieve sensitivity, line-edge roughness and resolution at the same time.”
More Moore has been extended by three months to March 2007, giving the partners time to carry out further development on an advanced photoemission electron microscope. The team has developed a prototype which is capable of detecting very tiny defects on wafers through smart-lens refocusing. The new instrument would be able to spot 10nm features, giving it the ability to detect dirt on masks without destroying the mask — something a conventional electron microscope cannot do.
As well as promoting its results at events around the world — most recently at the Extreme Ultraviolet Lithography (EUVL) Symposium in October 2006 at Barcelona — the project is also generating numerous European patents in lithography. It has greatly advanced European leadership in the highly competitive sector of electronics-manufacturing technology, Hartman says. “We believe our new plasma light source is a year ahead of any competitors.”
Source: IST Results