EUVL Extreme Ultraviolet Lithography

Transcription

1 EUVL Extreme Ultraviolet Lithography Laying the Foundations for Microchips of the Next Decade Do you remember your first cell phone? And what you could do with it? Exactly: make phone calls. It has only been 10 years. And today? Cell phones have mutated into multifunction devices: now you can make phone calls, send text messages, take and send pictures, surf the Internet, listen to MP3s, play games, navigate and even watch movies. Essentially, cell phones tell the success story of the chip industry like no other technical instrument: even more power in an even smaller space at comparably even lower prices. It is the increasing power of microchips that influence the development of cell phones. Optical lithography is a key driver of this success story. In many aspects it is playing a key role in the manufacture of semiconductors. After all, every microchip is produced using the lithography process. Depending on its complexity, it must undergo this process several dozen times. At the same time, it is primarily the advances in optical lithography that enable the continuously rising integration density of microchips. Mammoth efforts are allowing the semiconductor industry and its suppliers to introduce a new technology today, EUV lithography (EUVL), which will continue to drive progress over the next decade through miniaturization. This article deals with the perspectives of EUVL and the challenges of the technology, and reviews its development. Winfried Kaiser Winfried Kaiser received his Diploma in Physics at the University of Tuebingen. Between 1982 and 1989 he worked at Carl Zeiss in the field of optical design, system engineering in photographic and semiconductor optics. From 1989 and 1991 he hold the position as Project Manager Optical Metrology until he became Manager of the development lab of the Semiconductor Business Unit of Carl Zeiss in Since 2001 Winfried Kaiser serves as Vice President Product Strategy for Lithography Optics at Carl Zeiss SMT. In November 2007 he was designated as Carl Zeiss Fellow. The Authors Winfried Kaiser Carl Zeiss SMT Vice President Product Strategy for Lithography Optics Rudolf-Eber-Str Oberkochen, Germany Tel.: +49 (0) 7364/ info@smt.zeiss.com Peter Kuerz Dr. rer. nat. Peter Kuerz studied physics at the Ludwig Maximilians University of Constance and received his PhD in Between 1994 and 1996 he worked as a scientist at the NTT Basic Research Laboratories, Nippon Telegraph and Telephone Corporation in Atsugi-Shi, Japan. He changed to Carl Zeiss AG in Between 1998 and 1999 he hold the product responsibility for the development of the first 193 nm lens. Since 1999 he serves as Head of the EUV Program and is therefore responsible for the development of EUV technology at Carl Zeiss SMT. In 2006 he received Carl Zeiss Leading Edge Innovation Award. Dr. Peter Kürz Carl Zeiss SMT Senior Manager Systems, EUV Program Rudolf-Eber-Str Oberkochen, Germany Tel.: +49 (0) 7364/ info@smt.zeiss.com Overview Citius altius fortius (faster higher further), the happenings on the chip market strongly recall the Olympic credo. It has undergone a dramatic development since 1958 when Jack Kil presented the first integrated circuit to professional circles at a Texas Instruments laboratory. While there were 2,300 transistors on a chip in 1971 (Intel 4004), now there are 1,720,000,000 (Intel Itanium 2/2006). Almost visionary, Intel founder Gordon Moore described this development in 1965 in what came to be known as Moore s Law: The complexity for minimum component costs has increased at a rate of roughly a factor of two per year Certainly over the short term this rate can be expected to continue, if not to increase. Over the longer term, the rate of increase is a bit more uncertain, although there is no reason to believe it will not remain nearly constant for at least 10 years. That means 1975, the number of components per integrated circuit for minimum cost will be 65,000. I believe that such a large circuit can be built on a single wafer. [1] This prediction has been summarized later to the statement that the number of transistors on a chip will double about every two years. This development was made possible optical lithography in which the chip structures are projected from a mask onto a silicon wafer coated with photo lack. The deci Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 35

2 Figure 1: Assembly of the first EUV Micro Exposure Tool (MET) at Carl Zeiss SMT in On the overhead crane hanging the projection optical system. The company Carls Zeiss SMT Oberkochen, Germany Carl Zeiss SMT and its four divisions focus their procuct portfolio on the needs of R & D and production processes in Nanotechnology namely Semiconductor Technology, Materials Research and Development, Bio- and Life Sciences serving a wide variety of industries and research institutes. As an innovation leader in the field of lithography optics the company generates important momentum for further development in the chip industry and associated organisations. The 100 % subsidiary of Carl Zeiss AG, Germany, employs a total workforce of some 2,500 people and recorded for the latest fiscal year (Oct. 06 September 07) net sales of Mio. sive factor in this process is the ongoing miniaturization of the chip structures which is primarily due to the continually improving resolution of the optics used in lithography. The following formula describes the physical correlations: λ res = k 1 NA k1 is the process factor that can be reduced using complex resolution-enhancing masks, for example. However, this is expensive. The driving force behind the increasing resolution is thus the numerical aperture (NA) in addition to the shortening of the wavelength used during exposure. Although the semiconductor industry initially worked with the g line of mercury vapor lamp at 436 nm, shorter wavelengths have been continually used. Today, the flagships of lithography are systems that work with the 193 nm exposure wavelength generated argon-fluoride lasers (ArF). Immersion technology from microscopy, in which a thin layer of water is placed between the last lens element and the wafer, is used with this wavelength to increase the numerical aperture. Resolution down to 38 nm can thus be achieved in the mass production of microchips, however at the cost of decreasing k 1 factor, meaning highly complex processes and masks using resolution enhancement technologies like optical proximity correction or phase shift features. EUVL Solves Problem This is precisely where EUVL comes into play, for which Carl Zeiss SMT develops optical systems. With a wavelength of 13.5 nm, more than one order of magnitude lower than the current 193 nm, k1 factor can be increased and process windows being relaxed. EUVL is the basis technology favored the semiconductor industry for the next decade. This is primarily attributable to the following benefits: Enormous resolving power: Thanks to the very short wavelength of 13.5 nm, EUV lithography (EUVL) makes it possible to realize structures on computer chips that are considerably smaller than 20 nm. The semiconductor industry is developing a corresponding roadmap of technology nodes that will be achieved every 2 3 years. The 45 nm node (45 nm structures) was reached in nm will still be possible with 193 immersion the then following nodes will probably use a so called double patterning technology, still with 193 immersion. EUVL will be used in serial production for the first time at the beginning of the next decade for the subsequent following 22 nm node. 16 and 11 nm nodes can also be achieved with EUVL: progress in the semiconductor industry is thus secured for the entire coming decade. High productivity: Unlike all other methods (e. g. direct writing with electron beams or nanoimprint technology), EUVL is highly productive. EUV production tools will enable exposure of far more than 100 wafers per hour. The high-power EUV lightsources required for this are currently being developed. Special Features of the Technology The main difference to existing lithography is that there are no transparent materials for the shortwave EUV radiation. Therefore, purely catropic systems i. e. mirror systems can be used for EUVL, both for illumination and for the projection system. Additionally, the entire exposure process must occur in a vacuum as all forms of gas absorb the radiation. These two conditions present manufacturers and users with entirely new challenges: The necessary infrastructure in the application must be created and new concepts are also required for optical production, as well as in assembly and alignment. Ba Steps to Milestones The first publications and patents on EUVL appeared back in the 1980s. Beginning in the mid 1990s, the EUV-LLC industry consortium financed the construction of a prototype with a numerical aperture of 0.1 as part of a collaboration of the American 36 Optik & Photonik June 2008 No Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

3 national laboratories (Lawrence Berkeley National Laboratories, Lawrence Livermore National Laboratories and Sandia National Laboratories). EUVL development at Carl Zeiss also began in the mid 1990s. Carl Zeiss has been working on an EUV micro exposure tool (MET) together with the Lawrence Livermore National Lab since The Lawrence Berkeley National Laboratories received the first of them in 2002 (Figure 1). In the next step, Exitech produced commercial MET systems featuring ZEISS optics which were delivered to Intel and International Sematech (global union of chip manufacturers) in As a result of the small image field of 0.2 x 0.6 mm in the wafer plane, the MET systems are not suitable for the mass production of computer chips. They are deployed for resist tests and basic research on EUVL. A prototype for a full-field wafer stepper (the field size here is 26 x 33 mm) with an NA of 0.25 was finally realized, and thus a key milestone in the introduction of EUVL achieved, with the construction of the Alpha Demo Tool ASML Holding N. V., the global leader for wafer steppers, and its optics partner Carl Zeiss (Figure 2). Optical Concepts Due to the limited reflectivity of the EUV mirror, the number of mirrors in the illumination system and the projection optics must be kept to a minimum. The optical system of the MET (Figure 3) is based on the Cassegrainian principle for telescopes used in astronomy: Two mirrors perpendicular to each other reproduce a mask on the wafer. The optical quality of the mirrors and the positioning accuracy play a major role in this simple concept. Two-mirror systems are no longer enough to generate sufficiently large image fields suitable for the mass production of microchips. Projection optics with six mirrors, such as those developed for the EUVL Alpha Demo Tool (ADT) with an NA of 0.25, provide an optimal compromise between a minimum number of mirrors and perfect imaging. An outer-axial ring field is used to avoid central obscuration in the pupil. This is coupled with the use of outer-axial segments of rotationally symmetric aspheres. Together, they enable the purely catropic imaging of a field width of 26 mm which is standard in microlithography. As with dioptric systems, the key to increased resolution is a higher numerical aperture. In the first step, this can be achieved with larger mirrors combined with even more powerful aspheres. Design studies show that an NA of up to 0.32 can be achieved this way. Higher numerical apertures up to 0.5 can be achieved through the use of two additional mirrors. However, the maximum tolerable loss in transmission is then achieved. For even higher NA without increasing the total number of mirrors the concept of central obscuration of the objective pupil will be adopted. Mirrors with boreholes for an unobstructed beam path become unavoidable. This way, numerical apertures of 0.7 can be achieved which result in resolution down to 11 nanometers. Based on current estimates, these systems will appear in Requirements on the Mirror Surfaces The resolution of a lithography system is scaled to the wavelength of the light used. This also applies to the requirements on the surface quality of the optics. Here, even a quadratic dependence of the surface tolerances depends on the light wavelength. For EUVL, this means that surface defects on mirrors must be corrected on an atomic scale. This is made more difficult as EUV optics, as described above, are made of aspheres whose manufacture and inspection are disproportionately more difficult than would be the case with spherical elements. The three categories of surface defects can be roughly differentiated: 1. Long-wave defects with typical structure sizes between 1 and 500 mm, also referred to as figure, influence the imaging quality of the optical system. 2. Mid spatial frequency roughness (MSFR) with structure sizes between 1 mm and 1 µm lead to stray light and reduced contrast. 3. High spatial frequency roughness (HSFR) with structure sizes between 10 nm and 1 µm reduce the reflectivity of the mirror, thus leading to a loss of the radiation in the system which lowers the productivity of the wafer stepper. It was possible to achieve values of < 0.2 nm rms on the Alpha Demo Tool for these types of defects rms is the root of the mean square surface defect. Values of 50 pm have even been achieved for the surface accuracy. Purpose-built interferometers with never before seen accuracy have been developed and built at Carl Zeiss finally enabling the manufacture of the world s most precise mirror. Additionally Carl Zeiss developed its own polishing technique to achieve this type of precision. It is based on a combination of different local machining methods. In addition to processes such as milling, grinding and pitch polishing, computer Featuring a unique overview of the entire field latest research results in a uniform style over 3,000 color illustrations hands-on expert knowledge Price of each volume if purchased as part of the set: / / US$ Each volume will be invoiced and despatched upon publicaton. Single volume price: Approx / / US$ Set price: / / US$ _bu 6 Volume Set ISBN-10: ISBN-13: Publication dates: Volumes 1 and 2: 2005 Volume 3: December 2006 Volume 4: September 2007 Volume 5: January 2008 Volume 6: October 2008 Wiley Tel.: +49 (0) service@wiley-vch.de Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 37

4 Figure 2: Impression of the EUV pre-production system from ASML. The optical beam path is highlighted in green. Due to the short wavelength of the EUV radiation, the imaging takes place in a vacuum. (Image courtesy of ASML) NA = 0.25 NA = 0.55 NA = 0.70 controlled polishing, ion beam polishing and magnetorheological figuring are used, with which the required accuracy can be achieved in a reproducible process. As the numerical apertures increase and mirror surfaces grow, the machines needed to perform these methods must be adapted accordingly. This also means extraordinarily high investments. Coating Figure 3: The evolution of EUV- Lithography. ASML s Alpha Demo Tool uses a six mirror design (left) with a Numerical Aperture of 0.25 and enables a resolution down to 28 nm. In order to accomplisheven better resolution the NA needs to be increased. For that purpose not only bigger diameters of the individual mirrors are needed, but also two additional mirrors (middle). For even higher NA beyond 0.5 the concept of central obscuration of the objective pupil will be adopted (right). The operating principle of EUV multilayer coatings is based on the constructive interference of light reflected on many interfaces. Ideally, such a multilayer comprises a stack of coatings with alternating high and low refractive indices. The highest reflectivity values for the 13.5 nm wavelength under practically perpendicular incidence are achieved with molybdenum and silicon. EUVL mirror coatings are made up of roughly 100 single layers of the two materials. Thickness of each of these layers is about 3 4 nm with tolerances in the Angström region. One of the main requirements on the multilayer coatings is maximum reflectivity. Zeiss development partners FOM Rijnhuizen in Nieuwegein, The Netherlands, and IWS Dresden have demonstrated multilayer coatings with peak reflectivities of 70.1 % at 13.5 nm and perpendicular incidence. This is close to the theoretical limit of 74 % reflectivity. Assembly and Alignment of EUVL Mirrors EUV technology places highest demands on the positioning accuracy and stability of the mirrors. During wafer exposure, the mirrors must be held in position with sub nm and sub nrad accuracy. To meet these enormous requirements a highly stabile support structure featuring a very high natural frequency has been developed as well as specifically designed positioning systems. By this means it becomes possible to keep a single mirror in the set-point within a tolerance of only 0.3 nm. For comparison: If a highly stabile laser beam is deflected over one of these mirrors and directed at the moon, the laser point would have positioning accuracy of approx. 10 cm. The challenge during assembly of the EUV objective lens can be described as follows: 6 mirrors have to be mounted with positioning accuracy better than 1 mrad to get into the capture area of the above-described positioning system. An entirely new assembly strategy had to be developed. A concept was developed together with employees at Carl Zeiss 3D Coordinate Metrology, in which the mirrors are positioned using a coordinate measuring machine. As result of the above described interaction of highly precise mirror production and coating as well as assembly and positioning of mirrors within the system, the first full field Alpha Demo Tool was successfully aligned at the end of 2005: The achieved wavefront exhibits rms values of about 1 nm deviation from the target value. This enables reproduction of structure sizes < 30 nm. Outlook For Carl Zeiss, the successful implementation of this technology is a key step towards the future. Estimates of global market need call for more than 100 EUV systems annually over the next decade. Several leading chip manufacturers have already placed orders for successor systems to the EUVL Alpha Demo Tool with ASML. Thanks The authors would like to emphasize that the work presented here has been a team effort a large number of people at Carl Zeiss SMT and in various organizations. The following parties are especially acknowledged for their contributions: ASML (Veldhoven), FOM Rijnhuizen, Fraunhofer Institut IWS Dresden, Radiometrielabor der Physikalisch Technischen Bundesanstalt PTB am BESSY 2 Synchrotron in Berlin, TNO Delft and Philips. The authors also would like to express their special gratitude to the German Ministry for Research and Education (BMBF) supporting the projects EXTATIC (13N8088) and EAGLE (13N8837), as well as the European Commission for support of the project More Moore in the 6 th Framework Program. References [1] Gordon E. More: Cramming more components onto integrated circuits Electronics, Volume 38, Number 8, April 19, Optik & Photonik June 2008 No Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

5 4 Issues per Year Photocatalytic Wall Coatings Printed Electronics Nanostructured Textiles Anti-fog Coatings Transferring Nanotechnology into Automotive Series Production Nanotechnologies in the Energy Sector Plasma Deposition Shows Advantages Health Risk Versus Benefit Nanoscaled Materials Nanotoxicity Thermoelectric Thin Film Devices Soot Fine Dust Emissions Nanotechnology Supports Environmental Protection Plasma Functionalisation Technical Textiles for Biomedical Applications Materials for the 21st Century? Chemical Imaging Carbon Nanotubes A Certified Reference Material Deposition with Atmospheric Pressure Plasma Technology Nanoparticles Terminated Gas Condensation Production of Lithography Lenses Production Technology Process Engineering I wish to order for EUR annually ISSN plus VAT inc. postage within Germany. Further information phone ( ) or (aanders@wiley-vch.de) Life Sciences Titania Films Carbon Nanotubes Carbon-based Nanocomposites Nanoelectromechanical Biosensors Textiles Functionalisation Energy Metrology Carbon Nanotubes Nanoelectronics Subscribe now! Order fax: Street/No. Postal Code/City Phone/Fax This subscription is valid initially for It will automatically be renewed a further year if it is not canceled three months before the end of the subscription period. You can cancel this subscription writing to the publisher within 10 days (as per date of postage). I have read this notice and confirm this with my second signature. 1. Signature 2. Signature