Wednesday, March 20, 2013

Report on Recent SPIE Activities – Spring 2013

Commentary on SPIE Advanced Lithography IV 2013
SPIE Advanced Lithography IV concluded February 28, a few weeks ago.  I did not attend the conference this year but was in frequent contact with colleagues who were on the scene.  I was able to obtain some interesting perspective (albeit remotely) on EUV events thanks to first person accounts, daily news and video updates on the SPIE web site and from other SEMI industry news media.  My instinct to publish blog commentary immediately after the SPIE conference proceedings was throttled by additional background research in which I rediscovered recent, relevant historic data preceding and supplementing this years SPIE Advanced Lithography event.

ASML made a compelling fifty page presentation at SPIE which outlined their progress in sustaining the evolution of nanometer scale lithography while simultaneously supporting high volume manufacturing (HVM).  The report outlined the status of ASML's current EUV lithography program demonstrating measured incremental progress in EUV power output but underscored its ability to deliver the precision and accuracy necessary for evolving nanometer scale lithography.  Additionally, supportive data reinforced ASML's ability to implement complementary 193nm double patterned strategies as part of its EUV transition program.  Tool to tool accuracy and precision data were presented, ensuring the availability of interim process solution sets concurrent with the evolution of EUV development.  The EUV source power question was deferred to the continuing research and development effort at Cymer where gains have been progressing incrementally.

Following SPIE on March 12, at a recent UBS technology conference in Europe (Source: Seeking,  ASML was asked if the Cymer supplied EUV light source was the only available option for its EUV lithography system.  The ASML response was that they had designed a standard light source interface to the stepper but welcomed customers to optionally supply their own production proven EUV light source. Realistically, other than a few EUV source/systems under evaluation at imec, customers currently have few options and are probably best served by an ASML/Cymer source with an attached performance guarantee.

The SPIE conference again reminded us that the most pressing EUV lithography issues continue to be EUV laser/plasma based source technology, mask fabrication, resist performance and particle contamination concerns prompting a pellicle solution. The complex interaction of these critical issues create a matrix of challenges. The EUV resist component of this matrix was first comprehensively addressed in a 2011 paper: Stochastic Exposure Kinetics of Extreme Ultraviolet Photoresists: Simulation Study, by Chris A. Mack, James W. Thackeray, John J. Biafore and Mark D. Smith.  You might think a paper from 2011 is old news, however, in my opinion it is perhaps the best summary of critical EUV resist issues and remains a valuable work of reference applicable to current EUV process concerns.  Utilizing data derived from then available EUV resist materials a statistical analysis was performed to characterize lithography performance providing recommendations on possible process enhancing resist modifications.  The paper summarizes seven parameters which could be optimized to improve EUV resist performance (itemized below).  For complex structures and contact holes with dose exposure/shot noise sensitivity, the paper proposes that modification of resist chemistry and subsequent performance can be enhanced by:
  • Increase resist absorption of EUV light.
  • Reduce the electron affinity of the matrix polymer (reduce electron energy loss that does not result In PAG (Photo Acid Generator) excitation, thus increasing de and C)
  • Increase the electron affinity (and thus the reaction cross-section) of the PAG.
  • Minimize electron blur by increasing resist atomic density (thus reducing re).
  • Minimize acid diffusivity by using lower activation energy leaving groups and reduced post exposure bake temperatures
  • Reduce the spatial distribution fluctuations of PAG in the resist.
  • Reduce resist sensitivity to out-of-band radiation.
Fast forward two years to SPIE Advanced Lithography 2013.  A few minutes after reading this 2011 paper I discovered James Thackeray's recent You Tube video posted on 2/25/2013 describing how Dow Chemical's recent improvements in EUV resist compounds effectively addressed the issues described in the paper of two years ago.  You might ask, why does it take two years to resolve a photoresist design problem?  The chemistries themselves are complex and comprised of nanoscale molecular components which must react selectively to 13.5nm EUV at 14X below typical exposure levels while maintaining integrity under high vacuum conditions with minimal outgassing.  The search for materials exhibiting these extraordinary physical and performance characteristics is time consuming and requires an exhaustive R&D program.  It's gratifying to observe that those who have effectively identified and communicated a complex problem set are also contributing the solutions.

Chris A. Mack's 2012 SPIE invited paper Line Edge Roughness and the Ultimate Limits of Lithography resets the EUV resist issue yet again.  The recent encouraging news from James Thackeray and Dow Chemical might satisfy interim goals in the pursuit of shrinking nanometer scale geometries but the effort to further scale process technology is never ending (we think).  Chris Mack's recent paper speaks to the continuing struggle with LER.  After thoughtful dissertation on process parameters effecting LER he again suggests ways in which EUV resists might be further improved.  EUV exposure/dose is another parameter which can improve LER if only more power were available to provide viable exposure times demanded by HVM throughput requirements.  An anemic EUV photon count contributes to LER, poor contact hole resolution, shot noise, reduction in throughput and the gamut of related effects.  James Thackeray's  recent You Tube  video employs a popular illustration utilized to depict this problem.  Two adjacent plasma chambers exhibit marked differences in photon density with EUV depicted as 14X less.  Dr. Thackeray comments that with 14X less “information” EUV dosage at desired levels remains a challenge.  As an FCC licensed amateur radio operator, I'll site my own analogy.  Ham operators communicate using radio, mostly by voice but also transmit encoded “information” employing radio teletype, slow and fast scan television.  We sometimes talk to the astronauts on the International Space Station on a special link.  As a gentleman's hobby, the rule is to use minimal power necessary to maintain communications so as not to interfere with other stations or adjacent radio services.  On occasion, path loss between stations, atmospheric conditions and noise levels mandate adjustment to this guide line.  One thousand five hundred watts of effectively radiated peak envelope power usually delivers sufficient RF signal strength to “burn a path” to the receiving station.  Signal to noise problems are mitigated and the message gets through.  It's time for Cymer to crank up the output power.

Although there have been many positive developments reported at this years SPIE Advanced Lithography conference, many of the same critical EUV lithography issues remain on the agenda. Some items have been resolved only to be replaced by newer design and performance challenges.  The engineering funnel feeding Moore's Law flexes to accommodate bumps in the road map as the physical limits of materials and energy science are pressed ever further.  Forward looking HVM road maps project that performance improvements in resist LER (Line Edge Roughness), EUV power output and shot noise will eventually converge to resolve the HVM (High Volume Manufacturing) yield and throughput issues given time.  With continued investment, EUV nanometer scale lithography HVM could well emerge to sustain Moore's Law.  In the interim, parallel research continues in e-beam and direct imprint lithography as well as DSA (Directed Self Assembly) techniques for patterning nanoscale device structures.  ReRAM technology has recently been adapted for key next generation products which are candidates for these alternate manufacturing techniques.

Lawrence Berkeley National Labs CXRO Update
Later this year 2013, the enhanced performance SHARP (SEMATECH High-NA Actinic Reticle review Project) is scheduled to go on line.  The effort will be led by Kenneth Goldberg, Ph. D. at Lawrence Berkeley National Labs CXRO (Center for X-Ray Optics).  The enhanced EUV microscope/inspection station with 0.5 NA will be available for SEMATECH members conducting EUV lithography research. The system has undergone many hardware upgrades and performance enhancements and should prove extremely valuable in the continuing EUV process development agenda.  Patrick Naulleau, Ph.D. is the Director of LBNL CXRO and chaired this years SPIE Advanced Lithography IV program, providing unique insight on the industry wide EUV initiative.   

SPIE at Optics and Photonics: Lighting a Path to the Future
In addition to SPIE Advanced Lithography, another important SPIE co-sponsored initiative took place on February 28, in Washington, DC, titled, “Optics and Photonics: Lighting a Path to the Future.”  Other organizations co-sponsoring the event included the IEEE Photonics Society, Optical Society (OSA), American Physical Society, and the Laser Institute of America.  The event was attended by many government agencies who traditionally sponsor research.  The goal of the conference was to foster better government collaboration with American optics and photonics industries and forge a National Photonics Initiative (NPI).  The value of this proposed initiative is best exemplified by recent semiconductor industry manufacturers' investments in ASML and Cymer.  The collaboration of industry and government in key photonic and optical science endeavors could dramatically distribute R&D expenditures and reduce costs across many disciplines. 

SPIE at CREOL Industrial Affiliates Symposium
On March 8, SPIE members also participated in The CREOL Industrial Affiliates Symposium (College of Optics and Photonics at the University of Central Florida) held each spring (currently numbering 70 members).  Themed “Light in Action” the event was also attended by Dr. Eugene Arthurs and Steve Anderson of SPIE.  The following day Dr. MJ and Cheryl Soileau hosted the annual CREOL “Spring Thing” event, an annual Cajun cuisine cook out at their residence on Lake Jesup.  Dr. MJ Soileau is the current Vice President of Research and Commercialization at UCF.  Having attended the “Spring Thing” event at MJ's in the past, (unfortunately not this year) I can vouch for its Cajun authenticity.  “MJ” goes into rare form, rolls up his sleeves and stirs up a pot of spicy vittles which is always a delectable crowd pleaser.  Unknown to most, MJ also keeps a secret supply of his world famous hot sauce in his desk at UCF.  Those seeking inspiration are encouraged to sample this formula known only to MJ.  I once advised him on its stored energy potential commenting that “While not exactly fusion in a bottle, I suspected Pons and Fleischmann would be proud.”  Thanks MJ for your inspiration and all you do for UCF.

Zplasma Update
My blog entry last month featured an Interview with Henry Berg, CEO of Zplasma who is currently seeking funding for further development of a Xenon based Z-pinch EUV source.  During the SPIE Advanced Lithography conference there was interest in Zplasma EUV technology but no immediate success in securing funding.  Feedback from my recent interview with Henry provides insight he gathered from the conference while providing perspective on possible future opportunities for Zplasma.
Henry Berg – CEO, Zplasma: 
“It was very exciting to bring Zplasma’s SFS technology to SPIE and talk to companies that will benefit from our Stable DPP source.  We are addressing the four key areas where previous DPP sources ran into difficulties:

Electrode Melting: Previous DPP sources used a short, unstable pinch that required high instantaneous power levels that raised electrode surface temperatures above the melting point of tungsten (3695 K).  Zplasma’s SFS pulses are 10-100 times longer, so the required instantaneous power levels are 10 times lower, which keeps electrode surface temperatures below the melting point.

Electrode Pitting and Erosion: The unstable plasma used in previous DPP sources was subject to current interruptions caused by plasma instabilities.  The current interruptions caused high voltage spikes that ripped charge carriers out of the electrodes, causing pitting and erosion.  Zplasma SFS stable plasma eliminates the voltage spikes that were the root cause of the pitting and erosion.

Etendue Match and Collectible EUV Light: Poor etendue match and the use of grazing incidence collectors meant that previous DPP sources could not collect enough light.  SFS emission volume is adjustable for a high etendue match, and SFS geometry supports side-on collection with an ellipsoidal multilayer mirror, so the Zplasma Stable DPP source can deliver high EUV power levels to the stepper IF. 

Dose Uniformity: Previous DPP sources had unpredictable plasma instabilities, which made controlling dose uniformity difficult.  SFS stable plasma allows for EUV emission under control of the power supply to provide dose uniformity control better than the industry target of 0.2% 3-sigma deviation over 50 pulses.

We had some great discussions about the advantages of Zplasma’s source, with particular interest in the simplicity of Stable DPP and the resulting increase in reliability and reduction in cost of ownership.  We are talking to several companies now about supporting the development of our source.  We are also exploring several new secondary applications as a result of meetings at SPIE.”

SPIE Advanced Lithography IV 2013 provided a wealth of information and insight on the current status of EUV lithography and has also set the stage for the continuing development of required complementary technologies.  KLA-Tencor introduced its Spectra Shape 9000 laser/plasma based inspection station and is said to be developing an EUV actinic inspection tool.  As the complexity and sophistication of these instruments increase, so does the cost to manufacturing.  The result is further stratification of the semiconductor equipment industry and the likely continued expansion of the foundry business model.  Interestingly the business models for most of these scenarios are sometimes analyzed on iPhones, but look out Apple, the Samsung S4 will arrive next month.  Competition is a wonderful thing.  But then we all know that.

For those interested in my ham radio comments: Yes hams really can talk to the astronauts on the International Space Station. Most of the astronauts are also licensed hams with their own call signs.  The astronauts have a very busy schedule and operate the ISS ham station in their spare time which is not very often (usually after their dinner).  To contact the ISS it must be overhead for direct path VHF/UHF communications.  Visit the NASA link for ISS ham radio frequencies and transmission modes and the American Radio Relay League for more information on Amateur Radio:

Ham Radio Update 3/28/2013.  Two Hams are scheduled to arrive at the ISS.  NASA TV will televise coverage of vehicle docking with the ISS at 10:30 PM EST Thursday evening 3/28. Check the ARRL link below for more information on the astronauts.

As semiconductor equipment manufacturers expend considerable funding to create exotic plasma systems, the ISS crew is sailing above a sea of plasma easily viewed from their moving vantage point.  I suggest you visit the NASA link below for an incredible view of our planet:

Plasma display courtesy of NASA and the ISS:

Thomas D. Jay
Semiconductor Industry Consultant

The Technology High Ground