Dr. Nigel Farrar, Vice President of EUV Strategic Marketing at Cymer, Inc. talks to AZoOptics about Lithography Light Source Technology.
As a leader in the development of light sources, can you provide me with an overview on Cymer and the key technology that is developed by this company?
Cymer was founded in 1986 to develop and manufacture light sources for the semiconductor industry. This was at a time when the industry was moving away from mercury lamp light sources to excimer laser light sources – a transition that was delayed and did not start to ramp until 1996, when Cymer grew rapidly and went public.
Cymer initially provided krypton fluoride (248nm) light sources to the industry and then developed argon fluoride (193nm) light sources to support both argon fluoride dry and immersion lithography technology .
The next wavelength step the industry is planning is the move to extreme ultraviolet (EUV), which is a big wavelength jump to 13.5 nm. This involves a number of big changes to both source and exposure tool design because at this wavelength most materials do not transmit light.
Regular transmissive lenses cannot be used; reflective optics are required. The systems also have to operate a vacuum because air will also absorb at this wavelength.
Cymer already has a number of EUV sources operating in the field and we are continuing to work on the next generation source.
What key fields of application do you focus on?
Cymer manufactures semiconductor process equipment. Photolithography is the key enabler of the entire semiconductor manufacturing process.
Chips are made up of multiple process layers where the circuit or device design is fabricated on a wafer – many processes are involved in building up the device (i.e., first, the transistor is fabricated, then the wiring to interconnect the transistors).
There can be 20 to 50 layers required to build a chip and each layer requires a patterning step which requires an exposure tool.
These exposure tools are made by ASML, Nikon and Canon, and Cymer provides the light sources that are used in these tools to expose the circuit pattern onto the wafer.
EUV Lithography Source 3000. Image courtesy of Cymer.
One of the current developments with Cymer is the industry’s transition to extreme ultraviolet (EUV) lithography with the aim to develop smaller and faster chips. How does Cymer plan on facing this challenge to help maintain the transition?
Resolution depends on wavelength and the very short EUV wavelength enables very high resolution exposure tools. There are a number of challenges with EUVL technology. First, the system has to operate in a vacuum; using air would absorb too much of the light at this wavelength.
Both the exposure tool and sources have vessels that are evacuated inside, so that there is little absorption by the gas molecules of the light passing through this vessel.
How will Cymer make the transition from 3100 pilot sources to their first 3300 production sources? What will be the development processes involved?
The 3100 sources are out in the field today (we have a total of five of these sources at chipmaker development fabs). These sources are important for chip makers to learn about the technology and develop their processes.
The 3300 source is designed to be a high volume production system. Both sources are based on ASML names for the exposure tools they are integrated with. The 3300 is a higher resolution exposure tool. In terms of the source, the key difference is the amount of power that is provided.
The 3300 source will have more power output than the 3100 and the reason why power is important is because higher power allows for higher throughput and higher throughput means lower cost of operation. The objective of shrinking transistors is to get more on a wafer, and reduce the cost per transistor.
So it is important that the cost of processing the wafer does not increase significantly otherwise the next generation of transistors would not achieve the cost reduction that the industry is targeting.
When considering deep ultraviolet (DUV) lithography, there is a high demand for smaller features. Can you explain why DUV lithography has reached its limitations in the face of such demands?
As mentioned, resolution depends on wavelength, as defined in the well-known Rayleigh resolution equation. Higher resolution means that smaller features can be printed on the wafer. This depends directly on the wavelength and inversely with the numerical aperture of the exposure lens.
So there are two ways to improve resolution: 1) to reduce the wavelength, 2) to increase numerical aperture. At DUV wavelengths, the industry has reached the maximum numerical aperture that is practical even when using immersion lithography (i.e., using water between the lens and the wafer to allow numerical apertures greater than 1).
The limitations to increasing NA are mostly in the materials; there are paths to achieve slightly higher numerical aperture, but these require new materials for the lens, the immersion liquid and on the wafer, which creates a complexity and cost that would not offset the benefit.
The other approach of making an incremental step in DUV wavelengths was also considered – 157 nm wavelength tools were investigated for a number of years, and also had many materials challenges.
So DUV has reached the limits of scaling wavelength and numerical aperture, and resolution has been extended using other techniques that increase resolution such as double patterning. This process puts stringent demands on overlay for pattern placement and other process budget items.
The industry would rather go back to a single exposure technique which EUV offers because of the large reduction in wavelength.
What new technology will allow chip makers to increase the resolution of chip features?
There are a number of alternative technologies to EUVL, such as the electron beam (Multiple E-beam Direct Write) technology and imprint technology, which both have the ability to produce small features.
E-beam technology has been around for a long time and inherently has high resolution, but is a very slow process which reduces its cost competitiveness. Imprint technology is where a relief pattern is created on a mask and is imprinted directly into material on the wafer.
One of the biggest problems with this approach is that touching the wafer with the mask can introduce defects and if there are too many that can limit transistor performance or device yield.
There is also another technique called directed self-assembly where block copolymers can be designed to self-assemble into features on the wafer. This technology still needs a template to be printed on the wafer by a conventional exposure tool before the self-assembly process.
EUV lithography is a great example of technological advancements for the semiconductor industry. How does this technology compare to prior lithography techniques?
EUVL uses reflective optics instead of transmissive optics, which is the biggest difference.
Conventional glass material does not transmit at this wavelength so we have to use reflective mirrors that are at a maximum 70% reflective, so if there are a lot of mirrors in the system, a lot of light is lost at each reflection. This puts a greater demand on source power than in the past, which is why the source is such an important part of an EUVL exposure tool.
Integrated laser produced plasma (LPP) EUV light source. Image courtesy of Cymer.
Can you provide an overview on the International Technology Roadmap for Semiconductors (ITRS) and how this is driving the semiconductor industry? How does this relate to “Moore’s Law”?
Both are closely related. Moore’s Law is based on the observation that the number of transistors on the chip increases from generation to generation.
The ITRS provides the details of what needs to be done in lithography, deposition, device technology etc. and provides a time frame for these requirements in order for the industry to stay on Moore’s Law (i.e., the roadmap details out how Moore’s Law drives specifications and performance to stay on track in a timely manner).
Cymer is currently developing new crystallization tools for the flat panel display industry. Can you explain the inspiration behind this new development and the challenges that may be involved?
Cymer recently announced that it was going to stop developing products for this industry. We were developing and manufacturing a polysilicon recrystallization tool. However, for the application that we were targeting, the industry appears to be moving to an alternative process that does not use polysilicon.
Although the polysilicon provides a better quality device, it is more expensive, and after seeing the industry away from this approach, we decided to discontinue our product development for the flat panel industry. We will continue to support products already installed at customer facilities.
From a global perspective and if thinking about fields of application, where is there the most demand for your technology?
Cymer supplies its products to every chip maker worldwide. The leading Chip manufacturing regions are Korea, Taiwan, Japan, China, Singapore, USA, and a number of countries in Europe and we sell into all of these countries. Right now, the largest demand for semiconductor equipment is in Asia.
As a key company to leading the light generation, how do you see this industry progressing over the next decade and how will you keep up with these changes?
There is continued demand for new technology, especially in the consumer area. Products such as smartphones and tablets are becoming more powerful with including more and more features.
A key factor in the demand growth for these devices is portability (i.e., computing power is important, but battery life must be sufficient), and so there needs to be transistor devices that offer performance without consuming too much power; therefore, shrinking the devices using lithography such as EUV allows chipmakers to pack more capabilities and more features into these portable devices and to achieve this without consuming too much power.
Consumers today want these devices to constantly be connected wirelessly to the internet so that they can pull in information and process this data – and that requires a lot of power. So the key is to make sure these devices are low power for the user, which is achievable when you can make these devices much smaller.
EUVL will be introduced and then extended to provide the capability to produce multiple generations of smaller chips, and this is why there is such strong industry demand for this technology.
About Dr. Nigel Farrar
Nigel Farrar, PhD serves as vice president of EUV Strategic Marketing at Cymer, Inc., and is a member of its Scientific Advisory Board. He joined Cymer in 1999 and has led activities to identify the critical performance factors for the light source in the microlithography process.
Prior to joining Cymer, he spent fifteen years at Hewlett Packard working on advanced microlithography technologies. He has authored over 40 technical papers in the field of microlithography process technology. He holds a Bachelor of Science degree and Ph.D. in physics from the University of Bristol in England.
Disclaimer: The views expressed here are those of the interviewee and do not necessarily represent the views of AZoM.com Limited (T/A) AZoNetwork, the owner and operator of this website. This disclaimer forms part of the Terms and Conditions of use of this website.