Russian lithography, from stealing washing machine chips to the future.

Dateline: September 24, 2023

Elbrus-16s

I really think there is a lot of information about the past of electronics and microelectronics manufacturing in the Soviet Union and its major successor state Russia. I don't want to expand too much here, but let's make an overview.

After the collapse of the previous civilisation, Russia simply did not invest in semiconductor manufacturing capabilities, if during the Cold War they could compete with those of the West, the gap simply got bigger and bigger. There are a variety of explanations for this, ranging from the complete collapse of the economic and social system, loss of talent, lack of funds, the introduction of a liberal "we can buy it from outside" mentality, but the practical reality in the industry is one of neglect, both by the state and by Russian companies seeking these products on the international market. 

In the first decade of the millennium, a recovery began to take place in line with any sector in Russia, theoretical studies and some small enterprises started to be carried out and also began to appear on the international scene, here we would speak mainly of R&D for foreign or domestic companies, although these, with the aforementioned "we can just buy it already made" mentality, decided not to finance R&D in more tangible and advanced stages.

In the international sector, for example, the RAN Institute for Microstructure Physics collaborated with ASML on photomask-free MEMS, where good results were achieved.

The foundations of the X-ray lithography that will be discussed later, in fact, have their origins in the 2008-2011 period. A project was prepared in 2008, in 2011 they already had created a project for the radiation source (llaser and mirrors were made), they proposed to build the rest (engraving, resist, etc.) but Rosatom simply decided not to go ahead.

Indeed, the dynamic of "limited improvement" could be extended not only to the first decade of the millennium but up to 2014 when the first sanctions are imposed on Russia, among other things, on electronics. The key difference with the "collapse" years is that there was work being done. In 2014, due to the need, a new paradigm is created and the work intensifies and takes on a new dynamism.

The work and progress in Russian electronics in the period 2014-2022 is palpable in specialised areas.

GLONASS-K getting ready for launch

In particular, I did a short analysis on 18/05/2022 (https://t. me/the_Right_People/3413) on my Telegram channel about Russian satellites and how their dependence on foreign electronics (in the GLONASS-K series), especially in terms of vacuum resistant circuits, meant a delay of 6 years since the first 2 launches (2011 and 2014) of the K model satellites after the arrival of sanctions, but that these problems have been solved, and circuits suitable for use in space have been developed, something that Russia or the Soviet Union never had, their satellites being heavier and more complex, and with a shorter lifetime due to their pressurisation systems. In 2020 Russia launched again a K-model, which is now the standard model.  

I think this analysis is a good way to lead the talk about the work in the electronics over this years and as a historical review (I think too many people, understand advanced electronics, simply by technology node, but there are more areas of work), at the same time, more than a year has passed since the analysis and I think I was right when I said that the most damage that could be done to Russia in the technological and industrial area was already done in 2014 and that the sanctions would fail here.

Simply put, after the damage was done in 2014, the process of finding additional supply lines as well as localisation began, hardly anyone in the West thought about this in 2022.

2022 Status

It is from the beginning of the Special Military Operation and the anti-Russian sanctions that Russian microelectronics get a big boost; on the one hand the government decides now to allocate resources in the necessary quantities. On the other hand, companies start to invest and demand these Russian products, so the right ecosystem is created for the rapid growth of this sector, as well as other additional measures such as a more adequate legislative framework, etc.

In general we can say that Russia in 2022 can manufacture up to 90nm.

While 100% Russian production is up to 350nm and from 350nm to Mikron's 90nm, there are different contributions from the international market.

Almost from the first days of the SMO and, more importantly, of the sanctions, Mikron reported that the sudden increase in demand for its products was overwhelming its production capacity. Mikron talked about the need to double its production capacity from 3000 wafers per month to 6000 wafers per month, which seems to have already been achieved thanks to the acquisition of a 90nm SSA600/20 lithography machine manufactured by Shanghai Microelectronics Equipment Company.

SSA600/20 lithography machine

As sanctions begin, and access to international fabs is lost, it is clear that the gap cannot be allowed to grow even wider. Plans are drawn up to bring purely domestic production down to more modern standards.

From 350nm to 90nm

Work in this segment is well advanced. On 21 February 2023, a very remarkable interview with the general director of the Zelenograd Nanotechnology Centre, Anatoly Kovalev, was published. 

It is known that his company, in collaboration with the Belarusian plant Planar (Planar machines reached a minimum topology level of 500 nm), is developing photolithography for printing VLSI according to 350 nm and 130 nm design standards.

According to the information given. We would be talking about mass production around Q2-Q4 2024 for 350nm, followed a year and a half to two years later for 130nm. Interestingly, the company already this year started to collect applications from potential buyers.  The preparation of 3 production plants for this company has already started, the reconstructed Proton plant, another plant in St. Petersburg and another one in the Npp "Istok" area.

Therefore we have to expect a large volume production, in the interview, an extensive export intention to South-East Asia was mentioned.

From 90nm to 28nm

Here we must understand that this development is not carried out in series with the previous one, but in parallel. In this case, the development is carried out by the Sarov National Centre for Physics and Mathematics, in partnership with many others, in general all under Rosatom.

Thus, the Ministry of Industry and Trade and the Ministry of Economic Development were instructed to study Rosatom's proposals for several projects, including the project to create a national X-ray lithography and production technologies for its use. The deadline for the execution of orders was 30 August 2023.

August has passed, Rosatom's proposals have obviously been drafted, the go-ahead has been given, and Rosatom has begun to finance the project with funds from its investment programmes. As stated by the scientific director of the National Centre for Physics and Mathematics, Academician of the Russian Academy of Sciences Alexander Sergeev, the project is also actively supported by the Russian government.

By the end of 2025, a prototype of the facility will be manufactured for 90 nm, and by the end of 2028, an industrial prototype will be manufactured for 32 nm. After that, the standards will continue to gradually decrease due to a more precise tuning of the equipment.

X-ray lithography

The first thing to say is that Russian "X-ray" lithography will finally have masks, despite what was said two years ago.

On 10 March 2023, a seminar was held in Sarov at which Nikolai Ivanovich Chkhalo, head of the Department of Multilayer X-ray Optics at the Institute of Physics of Microstructures (IPM) of the Russian Academy of Sciences, presented a report "EUV lithography: principles, status and roadmap for development in Russia".

Responding to the question about the promotion of classical EUV lithography and the information silence about Maskless lithography, Nikolai Chkhalo said that within the framework of the research it was not possible to create one of the two key technologies that did not exist in the world, that is MEMS mirrors of the necessary characteristics.

The second key technology that needed to be created for the first time in the world was a four-hundred-layer X-ray lens. It was possible to create it.

As we said, in 2008 ASML commissioned IPM to create new MEMS based on Imec's MEMS and they achieved 40% reflectivity, but lost their functionality due to short circuits after sputtering. 

Despite the temporary setbacks with MEMS and that this time and despite a decade of studies, it has not been possible to develop the technology satisfactorily enough. Nikolai Ivanovich Chkhalo considers maskless lithography to be a very promising technology and hopes that everything will work out in the future. It is just a question of MEMS of the required quality.

I also think it is important to note that the machine will work at a wavelength of 11.2 nm and not 13.5 nm, like ASML. Technically, it would be closer to the X-spectrum, but it still does not touch 10nm in international nomenclature. On the other hand, the Russian tradition puts the beginning of the X-spectrum actually at 13.5nm, although nowadays the international standard is applied. ...

Many different organisations will work on the project, here are some of the main ones:

Phases of the machine

Alpha machine

This stage is designed for 2 years and has already started. A 3D model of the vacuum chamber has been developed and production has started. A 3D model of the X-ray optical system has been developed.

The optics of the machine will provide a resolution of less than 28 nm, but due to the still imperfect positioning system (scanning and alignment), the technology standards at this stage are expected to be around 90 nm.

The theoretical productivity limit of this prototype in the case of automatic feeding of wafers with a diameter of 100 mm is 40 wafers per hour. In fact, the feeding will be manual and the real productivity will be in the order of 3 wafers per hour.

Characteristics of the alpha machine

Thus, the prototype promises to be built by the end of 2025. Its task is to confirm the accuracy of the chosen technology. Verify that all pieces of technology work together and not separately. These are an X-ray source, X-ray optics, which correctly projects the image of the microprocessor layer onto the photoresist, and a positioning system with nanometre precision.

1st generation industrial machine

The duration of the work on this project is estimated at 3 years. In the beta machine, the topological standard is expected to be raised to 32 nm.

The 2.4 kW disc laser is expected to be ready. The size of the plates will be increased to 200 mm. Productivity is up to 66 wafers per hour, which is very good according to Micron technologists, who, according to them, need 40 wafers per hour.

This option can already be considered industrial and can be supplied to companies in the microelectronics industry. It is promised to be built by the end of 2028.

Characteristics of the beta machine

Later-generation industrial machines

After studying Rosatom's proposals, nothing was said about later generations previously included in the revision plans. Moreover, the previous stage was not previously called industrial. At that time it was only a beta machine. Perhaps, after discussing the issue with the Ministry of Industry and Trade and the Ministry of Economic Development, it was decided to speed up the process of supplying equipment to enterprises and make the beta version in the form of a serial copy.

However, this does not negate the fact that work on lithography will continue and the next generations of machines will appear. Previously also 3 years were allotted for the emergence of new generations. It was planned to gradually improve all features of the machine. The technological standards were expected to increase to 28 nm , and then to 16 and 12 nm . A 3.6 kW disc laser is expected to be available at that time, but in theory the machine could also use several parallel lasers of lower power. Thus, by 2032 we will have our own equipment for at least one 28 nm process technology, and the maximum predicted by scientists is 12 nm.

Characteristics of the industrial machine (according to the previous revision)

Conclusion

Of course, one can be both optimistic and sceptical that we will be able to build X-ray lithography and develop production technologies for its use, create other equipment for lithography lines, chemistry of adequate purity, etc. in 2028-2031.

By 2030 we will have a 28 nm process technology with a fair degree of certainty if the funds continue to be properly allocated and the project is not abandoned.

It may be possible to produce a 28 nm machine in 2028 in the first industrial generation. This is an optimistic forecast. 

Worst case scenario 32 nm by 2030 and 28 nm by 2032. 

Of course, this is only one part of creating a "fab", we should add computer-aided design systems, libraries, revision machines, the whole industry in terms of extraction and purification of materials, etc.

Many of these tasks are already being done in Russia, others, such as inspection machines, are less of a problem to acquire on the market, so it is possible that it will be decided to shorten the path there.

In general, for the first time in the 30 years of the Russian Federation's existence and since the fall of the previous civilisation we see a well-planned and systematic work, supported by the government and industry, on this whole issue, and over the course of this decade we will see a revival of Russian electronics.

On the other hand, this work is geared towards enhancing sovereignty in industrial and military matters, as we see the plan is not simply to run after the leaders in order to participate in the civilian market in processors, that is simply not realistic.

As of today, American semiconductors in military machinery work at 24-28 nanometres, so achieving this refinement by 2028 would therefore be interesting.

In the civilian sphere, we will see Elbrus and Baikal continue to work in fabless mode, but instead of Taiwan as a fab, these processes are known to be moving to friendly countries, mainly China. Shortly before the sanctions, the 32-core Elbrus processor was taking shape. This would have the refinement to start being used in servers and other more civilian machines, over the decade we will see this and others come into play in the civilian market, albeit manufactured outside Russia as happened in the 2010-2020s.

Finally, it is known as I said that Russia does not have the means, nor the domestic market to simply run after the only two countries with the means to compete in silicon processes (USA and China), therefore and referring to the Soviet aircraft designer Robert Bartini "We do not have to run after them to catch up technologically, but to cross them".

In other words, it is well known that Russia is not going to catch up with the silicon leaders, but sooner or later silicon will give way to another technology to further advance computing. One of the main candidates is photonics, where Russia is in fact quite well positioned in its development.

This mentality of "catching up with the world's leaders in the next paradigm shift" is present both in the industry, as reflected in the words of Mikron CEO Gulnara Khasyanova at this week's Kazan Digital Week, as well as in the government, in the aforementioned plans outlined on 30 August 2023, a strong commitment to photonics is included.

As always, I have tried to keep the text short but informative. I hope you like it and don't forget to subscribe to my telegram channel: https://t.me/the_Right_People