Unmasking microchip making for more markets

Miniaturised devices like smart-phones have now reached the mass consumer market, but when first introduced, manufacturing costs for low volumes can keep innovative products from taking off.

EU-funded researchers have developed new technology for microchip manufacture that promises to cut costs and increase design flexibility.

By reducing costs for small-scale production and enabling further miniaturisation in microchips, the results are likely to open the door for new applications and markets. The project’s technology will also help smaller EU microelectronics companies.

Semiconductor lithography is the process by which microchips are printed with the tiny channels and gates which are the heart of transistors that make up the modern integrated circuit. Using a technique called ‘maskless lithography’, researchers in the EU-funded maskless lithography for IC manufacturing (MAGIC) project are starting a new era in small scale IC production that will make the process faster, simpler and cheaper.

Current microchip manufacturing uses a mask-based optical lithography technique, essentially using a template to define the desired circuit pattern. It is a process that has long served the microchip industry well because it is very effective.

But masks are expensive; it can cost up to several hundred thousand euros to produce just one of them. A single chip requires several of these masks. The total mask cost is drastically rising as chip features become smaller and more sophisticated.

For this reason, MAGIC studied the potential for maskless technology to combat the rising costs of semiconductor masks.

Indeed, mask-based lithography becomes more and more complex with each new advance in lithographic technology.

For example, optical proximity correction is a technique that compensates for distortion effects often seen along the edges of circuit channels, while phase-shift masks use interference produced by phase differences in light to improve the resolution of photolithography.

Both these techniques are needed to compensate for problems introduced by pushing lithographic techniques ever further. As lithography becomes more sophisticated, mask manufacture becomes more complex and expensive.

So masks now require higher performing, expensive materials and elaborate, time-consuming production processes.

“There is now a real need for maskless lithography, because the increased costs associated with mask manufacture play a huge role in the overall cost of ownership of this technique,” explains Laurent Pain, coordinator of the MAGIC project.

ML2 technology uses multi-beam principles. Essentially, a large electron beam is focused through an aperture. The aperture can split the beam into thousands, or even millions, of smaller beams. These beams pass through an active micro-electrical-mechanical system (MEMS) element, called a blanker, which can control each beam individually. Some beams are then deflected, or blanked, while those that are allowed to pass create the desired microchip circuit.

Large-scale industrial manufacturers, like Intel and Samsung, focus on massive production of microprocessors or memory chips. For them, the impact of lithography cost increase is less important because the associated costs are spread across huge production runs involving millions of chips.

But companies developing application specific integrated circuits (ASIC) and small and medium-sized enterprises (SME) are struggling with the expense. It is a big problem for Europe because many of the continent’s technology companies are SMEs.

It is a general problem for technology too, because the high cost of small-scale microchip manufacture can kill off innovation; new markets cannot take off if innovative devices are too expensive when manufactured in small numbers.

The concept of a maskless technique is already supported by some key complementary metal oxide semiconductor (CMOS) manufacturers around the world, like TSMC in Taiwan and STMicroelectronics in France.

Even Intel, in recent industry conferences, has highlighted the potential of this technology as a complementary solution to optical lithography. This means the technology has a promising future and makes it an even more attractive R&D path.

MAGIC sought to develop ML2 technologies to ‘alpha-stage’ prototypes for testing.

Two partners, the Netherlands-based MAPPER Lithography and Austria’s IMS Nanofabrication, have already made advances towards high-resolution integrated circuit design using maskless electron lithography, confirming the capability of this technology.

Half MAGIC’s work focused on further development of the tools created by MAPPER and IMS. This provided a wide range of technology, as MAPPER’s tool is based on low voltage electron beams while IMS developed a high voltage beam system.

The project delivered alpha-version prototypes meeting the semiconductor manufacturing requirements of 32 nm half-pitch technology. Half-pitch refers to the size of lines and spaces that separate one element from another on the microchip.

“Over the three years of the project, both platforms have pushed the maturity of their technologies from the proof-of-concept to the pre-alpha level,” notes Dr Pain, adding, “Today, both solutions have achieved significant progress, emphasising the credibility of maskless lithography.”

The second half of MAGIC’s work sought to develop the necessary infrastructure to use these tools in an industrial environment and this infrastructure work was split into three key areas: data treatment, electron beam proximity effects and industrial integration.

MAGIC developed a fast, robust and commercial data preparation platform, compatible with industry standards that can provide full support for a maskless production system.

The platform also compensates for ML2-related electron beam proximity effects. This was a vital development. Overall electron beam interaction on the substrate, as well as potential heating effects, has to be compensated for to ensure the accurate patterning of the microcircuit. The solutions developed within MAGIC eliminate this problem by applying the suitable corrections.

Finally, the project demonstrated, in real-world conditions, that ML2 can be integrated into manufacturing environments using the platforms developed by the partners.

“2010 was an important year for MAGIC,” said Pain. “This last year confirmed the full potential of massively-parallel electron beam lithography. For both developed technologies, operational machines were built that demonstrate all the key functional elements.”

The technique is promising but it has been a long time in development. IMS Nanofabrication first mooted the concept of maskless lithography in the 1980s, but required enabling technologies did not exist at the time.

More recent efforts to support introduction of maskless lithography began in 2005 with another EU-funded project called ‘radical innovation in maskless nanolithography’ and with the MEDEA+ projects.

Cordis Europe

http://cordis.europa.eu/