A chance for a Russian lithographer

Russian scientists have developed their own concept for the development of high-performance X-ray lithography. It is based on a number of innovative solutions that will lead to a reduction in the energy of the lithograph, a reduction in its overall dimensions, the cost of equipment and its use by about an order of magnitude

The building of the Institute of Microstructure Physics (IFM) of the Russian Academy of Sciences.
A source: IFM RAS

This article by Nikolai Chkhalo, Head of the Department of the Institute of Microstructure Physics (IFM) of the Russian Academy of Sciences, Doctor of Physico-Mathematical Sciences, is a continuation of the discussion of the development in Russia of the latest EUV photolithograph, the most important installation necessary for the manufacture of microprocessors. Earlier, this topic was touched upon in his interview with our magazine. This time, Nikolai Ivanovich talks about new approaches of IFM to the creation of such installations, which can significantly facilitate their development and manufacture. The article is a summary of the preprint .

The author expresses gratitude to the Director of the IFM RAS, Corresponding Member of the Russian Academy of Sciences Zachary Fishelevich Krasilnik for supporting the topic of X-ray lithography and active efforts to promote this project.

EUV lithography at a wavelength of 13.5 nm, despite its youth (extreme ultraviolet has been used in the industry since the end of 2018), has become one of the key technologies in the production of chips with advanced technological standards. The production of EUV lithographs and related equipment already provides about 50% of the revenue of ASML, the world leader in the production of lithographic equipment and the only manufacturer for EUV lithography. However, their concept of achieving maximum productivity of the lithographic process has led to extremely high cost of equipment and its operation. This has drastically limited the number of companies able to use this technology. Technically, it seems unlikely to repeat the development of ASML, and the use of such equipment for the domestic chip market with its limited volume is impractical.

We propose a new concept of X-ray lithography based on a number of innovative solutions that will reduce the energy of the lithograph, its overall dimensions, the cost of equipment and its use by about an order of magnitude.

Nikolay Chkhalo, Doctor of Physico-Mathematical Sciences, Head of the Department of the Institute of Microstructure Physics (IFM) of the Russian Academy of Sciences.
Source: Photo provided by N. Chkhalo

Limits of progress

The progress of microelectronics is directly related to the reduction of the geometric dimensions of topological elements. The topology and dimensions of the elements are set by lithography. The cost of lithographic equipment and the cost of the lithography process itself can exceed other costs by orders of magnitude in the production of chips. For example, the price of a lithograph with immersion at a wavelength of 193 nm can reach 50 million euros, a lithograph at a wavelength of 13.5 nm — from 300 million euros, and a set of masks can cost up to 10 million dollars or more.

The most advanced chips are mainly produced using lithography at a wavelength of 193 nm. This has become possible due to the continuous development of resolution Enhancement Techniques (RET).

Currently, the use of RET has made it possible to reach the resolution limit of 8 nm ultraviolet lithography. However, due to the high proportion of defects, a multiple drop in productivity and an increase in the cost of the process in the mass production of DUV (Deep Ultra Violet, deep ultraviolet with a wavelength of 248 nm), lithography is not used in the manufacture of critical layers of chips with minimum topological dimensions of 16 nm or less.

It seems that the only way for us is to repeat what ASML has done. I beg to disagree with this. Based on the long-term interaction with ASML and Zeiss on the development of the EUV lithograph, as well as his own experience in creating an experimental sample of the lithograph, the author is convinced that an attempt to copy the lithograph of the ASML company will not lead to success

The resolution of lithography can be improved by switching to a shorter wavelength. Research in the field of EUV lithography at a wavelength of λ = 13.5 nm was started back in the 80s of the last century. And if at first the development of their own X-ray lithograph was carried out in the USA, Japan, the Netherlands and Russia, by now only the Dutch company ASML has remained. Only she was able to integrate the most advanced achievements from all over the world into her product. Other companies and organizations that have achieved success in developing certain nodes and technologies for the EUV lithograph have concentrated on these developments in the interests of ASML.

Despite the relative youth of the technology, it is spreading rapidly, and we can safely say that this is the next generation lithography. ASML expects further growth in the production of EUV lithographs.

After many years of neglect and sometimes aggressive rejection of the projects proposed by Nikolai Nikolaevich Salashchenko and the author of this work, the topic of X-ray lithography has become widely discussed in the Russian Federation. This was largely facilitated by the appearance in 2022 of the roadmap for the development of X-ray lithography in Russia, developed at the Institute of Microstructure Physics of the Russian Academy of Sciences. And most importantly, the government has matured an understanding of the importance of the problem raised.

It seems that the only way for us is to repeat what ASML has done. I beg to disagree with this. Based on the long-term (from the mid-1990s to 2014) interaction with ASML and Zeiss on the development of the EUV lithograph, as well as his own experience in creating an experimental sample of the lithograph, the author is convinced that an attempt to copy the lithograph of ASML will not lead to success.

Specialized beryllium laboratory (two technological installations for four and six magnetrons).
A source: IFM RAS

The ASML concept

Back in 2012, it was shown on experimental lithographs (β-tools) that, if you do not take into account the short life cycle of lithograph elements and the cost of repair and downtime of equipment, starting from the topological norms of 32 nm, the cost of the lithographic process on the EUV machine becomes less than on the DUV. As the topological dimensions decrease, this effect only increases.

But along the way, the developers of the X-ray lithograph faced a number of intractable problems with extremely expensive solutions. The fact is that the performance of the lithograph is determined by the efficiency of reflective multilayer optics and the conversion of energy supplied to the source into EUV radiation at the operating wavelength, in a spectral band equal to the bandwidth of the optical system of the lithograph.

If we calculate the efficiency of the 12-mirror optical system of the lithograph, consisting of a collector, four mirrors of the illuminator, a mask and six mirrors of the projection lens, and also take into account the transmission coefficients of the filter protecting the mask from contamination (pellicle) and the filter separating the volumes of the projection lens and the table with the exposed GDL plate (Gas Dynamic Lock), then the efficiency of the system will be less than 0.9%. In practice, this value is even lower due to the low collector reflection coefficient at 41%, the presence of protective layers on the surface of Mo/Si mirrors and polarization effects. Therefore, in order to achieve high performance, the main efforts of the developers are aimed at increasing the power of the laser plasma source (LPI) radiation by 13.5 nm.

This source, which has the highest conversion rate in the region of 13.5 nm, is the main problem of the EUV lithograph from ASML. Firstly, it is the huge size of the installation, the laser system occupies an entire floor. Secondly, due to the use of a gas-discharge CO2 laser, which is unique in terms of parameters, the installation exceeded the megawatt level in power. Thirdly, the most serious problems entail the use of tin in the lithographic process, which must be removed from the installation, since even nanometer contamination of mirrors with tin vapors will lead to a loss of reflectivity. The solution to protect optics from tin vapor pollution was the use of hydrogen. The use of hydrogen in a lithograph, and even activated by ionization due to irradiation with EUV and DUV radiation, imposes strict requirements on construction materials, severely limiting their nomenclature or shortening their service life.

From an economic point of view, repeating the ASML project is pointless for Russia. The argument that the price for special equipment is not so fundamental, in our opinion, is incorrect, since even with classical DUV lithography in small-scale production, it can exceed the cost of special equipment for which this chip was produced

Collectively, due to the described problems, less than 1.2% of the laser radiation power reaches the intermediate focus from 6% of the conversion efficiency of the source. These losses are the result of measures protecting optics and other elements of the lithograph from vapors and high—energy tin ions, as well as from radiation loads due to laser radiation scattering.

The very short lifetime of the collector and mask, the cost of each in the region of one million euros, and the huge consumption of electrical energy make the cost of using this type of equipment extremely high.

According to various sources, the price of the currently produced lithographs of the NXE:3400C and NXE:3600D series exceeds 300 million euros, and the new generation EXE:5000 is many times more. Nevertheless, top managers of TSMC, Samsung and Intel confirm that, despite all these costs, EUV lithography is cost-effective. However, it must be borne in mind that this efficiency is due to the huge chip market occupied by these companies, which are essentially monopolists. With a shrinking market, this efficiency will plummet. This is indirectly confirmed by the fact that apart from these giants, as well as the American Micron Technology and the Korean SK Hynix Korea, which are among the top 5 chip manufacturers in the world, no one else has purchased such equipment and, according to ASML forecasts, does not plan to in the near future.

Based on the volume of the market, it can be concluded that from an economic point of view, it is pointless to repeat the ASML project for Russia. The argument that the price for special equipment is not so fundamental, in our opinion, is incorrect, since even with classical DUV lithography, the cost of a chip varies by five orders of magnitude depending on the serial production, and with small-scale production it may exceed the cost of special equipment for which this chip was produced.

The incredible technological difficulties of creating an EUV lithograph led to the fact that even the USA and Japan, having started in this race first, could not bring their EUV programs to a competitive product and limited themselves to only individual components for ASML. The reason for their failure and, in contrast, the success of ASML, in our opinion, is that ASML was able to integrate the best world achievements in all major components into its product. They achieved this through the unprecedented openness of the project.

THE DIAGRAM OF THE LITHOGRAPH.
A source: IFM RAS

Our response is ASML

The latest achievements of the IFM RAS in the field of multilayer X-ray optics and a xenon-based laser plasma source with a wavelength of 11.2 nm have allowed us to take a fresh look at the development of X-ray lithography from the point of view of reducing the cost and further operation of the lithograph while maintaining minimum technological standards at the ASML level at the cost of some reduction in its performance. Achieving this goal involves a multiple reduction in the energy of the laser-plasma source. If the lithograph from ASML is a megawatt level installation, then the concept from IFM RAS is about 0.1–0.2 MW. Overall dimensions will be drastically reduced, the life cycle of the laser, collector and other elements of the lithograph will be increased. It will be possible to increase the efficiency of the X-ray optical scheme and simplify the projection scheme.

The proposed concept is based on the following innovations.

1. Reducing the operating wavelength from 13.5 to 11.2 nm, which will increase the resolution of the installation lens by 20%. This will allow to reduce its numerical aperture to achieve the same resolution*. The consequence of this will be a reduction in overall dimensions and a significant simplification of the manufacture of mirrors. Therefore, we can expect a reduction in overall dimensions and a noticeable reduction in the cost of lens production. And since multilayer Ru/Be mirrors are used at this wavelength instead of Mo/Si, our research shows that the efficiency of the optical system will significantly increase.

The latest achievements of the IFM RAS have allowed us to take a fresh look at the development of X-ray lithography from the point of view of cost reduction while maintaining minimum technological standards at the ASML level at the cost of some reduction in its performance

2. Replacing a tin laser-plasma source with a xenon one reduces contamination of optical elements by orders of magnitude with the products of the source material expansion. After all, xenon is an inert gas, and it cannot pollute optics. As a result, the lifetime of expensive collectors and masks will increase many times. All this reduces the cost of both the manufacture of vacuum elements and systems, as well as the lithograph as a whole, and the cost of operation.

In addition, instead of a large-sized gas-discharge CO2 laser, the proposed LPI uses a reliable small-sized and energy-saving solid-state laser with diode pumping and a disk amplifier. It is important to note that Russia currently does not even have an experimental prototype of a pulsed CO2 laser with parameters close to ASML, while there are advanced developments in the field of high-power solid-state hybrid lasers.

3. The transition to a wavelength of 11.2 nm opens up the possibility of using silicon-based photoresists. It can be expected that an increase in the proportion of silicon in the resistor will lead to a noticeable increase in the efficiency of the resistor at a wavelength of 11.2 nm compared to 13.5 nm.

The table below compares the main parameters of the TWINSCAN NXE:3600D lithograph with the expected parameters of the lithograph developed at the IFM RAS. A number of "internal" parameters of the lithograph from ASML were restored by the author from the analysis of various sources, but the main ones were taken from the ASML website. Conservative estimates were made when calculating the productivity of the lithographic process from the IFM RAS.

Comparison of the main parameters of lithographs.
Source: Nikolay Chkhalo

As can be seen from the table, fairly conservative estimates show that with an average laser power of 2.7 kW, the expected performance at a wavelength of 11.2 nm will be about 2.9 times less than that of the ASML lithograph. But for factories whose product market is smaller than that of the top 5 companies, this value is quite enough, given that of all the layers on the chip, X-ray lithography is used only when forming several critical layers.

The successful implementation of this concept, based on technologies from the 2020s rather than the 2000s, will achieve the goals of increasing the availability of X-ray lithography for users without compromising resolution.

Why are we confident of success?

For the successful implementation of the project, the IFM RAS has created a world-class scientific and technological reserve. A two-stage shaping technology has been developed for the manufacture of X-ray optics, including aspherical optics. At the first stage, flat or spherical blanks are produced using the classical deep grinding-polishing method using unique polishing compounds.

At the next stage, finishing polishing, aspherization and correction of local errors are carried out by ion beam etching.

In particular, the IFM RAS has developed a technology for spraying Ru/Be mirrors with reflection coefficients at a wavelength of 11.4 nm up to 72.2%, which is noticeably higher than the record 70.15% of Mo/Si mirrors used in the ASML installation.

Research on the xenon X-ray source was started almost ten years ago. Their results suggest that the efficiency of a xenon source will be no lower than that of a tin source.

For the successful implementation of the project, the IFM RAS has created a world-class scientific and technological reserve. A two-stage shaping technology has been developed for the manufacture of X-ray optics. At the first stage, flat or spherical blanks are made using the classical deep grinding-polishing method using unique polishing compounds

For lithographic purposes, a pulsed periodic laser with a pulse energy of tens to hundreds of millijoules, a duration of several nanoseconds and an average power of a kilowatt is needed. An experimental sample of such a laser has been developed at the IPF RAS.

Significant results were also obtained jointly in the field of resists for 13.5 nm by the IFM RAS and the Institute of Chemistry of the N. I. Lobachevsky National Research University. When switching to silicon-based resistors, one can expect an increase in sensitivity by their greater absorption at a wavelength of 11.2 nm compared to 13.5 nm.

The IFM RAS has significant achievements in the field of masks for EUV lithography and free-hanging multilayer films to protect masks (pellicle) and optics from contamination by decomposition products of resists (GDL), spectral purification filters (SPF). In particular, ASML's experimental lithographs were equipped with this free-hanging optics. About 20 joint patents with ASML have been obtained. Currently, this optics is in demand, in particular, in the countries of Southeast Asia. And finally, the IFM RAS has competencies in the field of scanning systems and autofocus, which previously made it possible to create the first Russian model of a lithograph at a wavelength of 13.5 nm.

Nikolai Chkhalo