Markt&Technik | New metrology helps fabs
Originally published in Markt&Technik. (DE)
"We can beat scarcity!"
How can throughput and yield in semiconductor manufacturing be significantly increased? And not by a few percent, but by as much as 40 to 60 percent for certain machines in one fell swoop? Especially in times of scarcity, the answer to this question is likely to be of the utmost interest.
Prakash Murthy, CEO, CTO and founder of Atonarp, has an answer: a sophisticated metrology system based on mass spectrometry that takes place directly in the process chambers of today's cluster tools. For example, over 10 such chambers can be located on a platform for the deposition or etching of specific layers, be they conductive materials or dielectrics. The technical term for on-site measurement: "in-situ" metrology.
Today, however, chip manufacturers usually make do with so-called "in-line" metrology: what actually takes place in each individual process chamber is not measured. Rather, it is determined, for example, how long it takes for a certain process to be completed. Now these durations are monitored "in-line" - with the inclusion of the appropriate safety buffers. This is not very effective and costs time.
"Because of these many shortcomings, we developed a mass spectroscopy-based metrology system from scratch that can be used in-situ," says Prakash Murthy.
It takes two to three years for the additional capacity from new fabs to be felt in the market. With Aston, throughput could improve noticeably after just weeks.
The result is "Aston." The name is a reference to the chemist and physicist Francis William Aston, who invented the mass spectrometer in 1919 and was awarded the Nobel Prize in Chemistry in 1922. Such a naming is an obligation. Prakash Murthy is convinced that he can live up to the high expectations associated with it: "In the case of cleaning process chambers, it can reduce the time by no less than 80 percent." The throughput of individual machines - especially those for deposition - could be increased by over 40 percent with "Aston" in certain cases. "Even a 1 percent improvement in throughput in a fab, which is small at first glance, can result in savings of tens of millions of dollars over the course of a year," says Prakash Murthy. "We can help beat the shortage!"
That's because it takes two to three years for the additional capacity from new fabs to be felt in the market. If new lines are built in existing fabs, then it still takes at least 1 year, assuming there is clean room space. With Aston, throughput could improve noticeably after just weeks.
Risto Puhakka, President of VLSIresearch, is also convinced that Atonarp has made a real breakthrough. "It really is a completely new method. In particular, the throughput of complex etching processes, such as those required in the production of 3D NAND memory ICs, can be significantly increased," he told Markt&Technik in an interview. If the throughput of a NAND fab could be increased by 10 percent, this would already be a significant advancement. For modern fabs, this is a very attractive prospect. The same applies to existing fabs that are to be upgraded for the next process node. After all, with "Aston", existing machines in the field can be upgraded without any problems.
Prakash Murthy, at any rate, is certain that good business awaits Atonarp: 26,000 process chambers are expected to be commissioned in 2021, which would correspond to a sales potential of $1 billion for metrology. Well over 100,000 process chambers would be operating in fabs today, representing another $3 billion or more in potential revenue by 2024. At the moment, Atonarp is primarily targeting CVD and etch process applications which are currently growing at an average of 13 percent per year.
Risto Puhakka, President of VLSIresearch: "Atonarp's approach is really new. In particular, it can significantly increase the throughput of complex etching processes such as those required in the production of 3D NAND memory ICs."