Russian Magnetic Nanoelectronics Factory
Researchers have worked on the development of high-speed memory by integrating magnetic elements with semiconductor technologies for several decades. The existing types of memory are at the top of their capacity, so inventors have been looking for other ways and the novel magnetoresistive random access memory (MRAM) appears to be very promising. Currently, this project is at a stage of transition to industrial manufacture of third generation technologies.
Broadly, a magnetoresistive cell is a structure with two magnetic layers and a separating insulator. Storage of information is based on the differential electric resistance of a cell with parallel magnetization layers and anti-parallel magnetization of layers. The speed of magnetization is high. Therefore, the speed of MRAM is comparable to the speed of SRAM and DRAM. However, unlike older technologies, it is capable of storing information without external power supply. In comparison with other types of non-volatile memory, for example, flash memory or EEPROM, MRAM has a much faster recording and reading speed. Another unique feature of MRAM is its higher security because some methods of attacking memory are impossible with it. The only drawback of MRAM is its lesser cell density, and, hence, it has less storing capacity than DRAM and NAND flash have. However, magnetoresistive random access memory is cheaper in production because it is manufactured using ordinary industrial electronic devices in the technological operations and the structure of its cells is simpler than, say, in flash memory.
Crocus Technology was created in 2004 for commercialization of technologies developed by Spintec Laboratory. The head office of the company is in the USA and the research center is in Grenoble, France. Together with IBM, we have developed MLU architecture, which allows production of magnetoresistive elements built in metalized layers on a semiconductor without damaging the structure of the semiconductor. In other words, we have found a solution to produce a magnetoresistive cell and to connect it to the complementary metal oxide semiconductor devices (CMOS). Essentially, we have combined two well-known and highly developed technologies into one process, obtaining a product with new features, which is compatible with conventional elements and can be manufactured on a large scale.
MLU is a universal technology of manufacturing magnetic cells not only for MRAM, but also for magnetic field sensors, as an example. A magnetoresistive cell may also be used as a logical element for comparing stored data with incoming data. In this case, the information is not read from memory, providing the highest data security, which may be used for protected solutions.
Another major feature of our invention is that the magnetic structure is heated before storing data. This technology is called Thermally Assisted Switching (TAS). Furthermore, we are first to manufacture magnetic cells with 90 nm technology nodes.
TAS MRAM does not need energy, and it is marked by high speed, low cost and long life.
In what fields can your magnetoresistive elements be used?
There are three main fields so far: magnetic field sensors, which we produce already, memory discrete modules, and MLU microchips, such as microchips for smart cards, for example, SIM-cards, ID cards, payment cards etc. The prospects are great. MLU is good both for RAM and for ROM, so in the future it may become a universal type of memory, which will significantly simplify the design of electronic devices and computers. So-called RAM databases is another promising field. At present, computers use servers for high-speed database searches, caching large volumes of data in DRAM and consuming too much energy. In the future, when the memory cell density becomes comparable with DRAM, using non-volatile MRAM instead of DRAM will considerably lower the operational cost of such systems.
With further development and improvement of this technology, we are planning to offer new products to the market.
At what phase is the CNE project?
CNE is setting up manufacture of 300-mm wafers with 90 nm technology nodes. It will not be a full cycle production because we buy CMOS-wafers from another company, in particular, from SMIC. In fact, CNE will be the first Russian enterprise processing 300-mm wafers.
The business plan specifies three phases of equipping the manufacture: Phase Zero, Phase A and Phase B. By the end of 2013, the first manufacturing line of covering CMOS-wafers with magnetic layers was launched marking the end of Phase Zero. During Phase A, equipment is installed for lithography and etching magnetoresistive structures, and during Phase B we will complete equipping the manufacture and we will have the capacity for all foreseen operations for production of wafers cut into magnetoresistive products.
We already have all the technological equipment for Phase A, in particular, 193 nm photolithography scanner and an integrated line of etching and depositing magnetic layers, which has been specially configured by Korean company Jusung. By the end of June, a very important event happened, which is ISO 14644-1 cleanroom class 1000 Certification. We have recently started the second phase of equipping the manufacture and creating the full infrastructure of the enterprise. This process must be completed by November. The next phase will be adaptation of the equipment and implementation of the technology. Our partners from IBM will assist us and we expect to have finished this process by the end of February 2015. A comprehensive manufacture test will take about three months, during which the product technical specifications and suitability for packaging will be tested. In May, we are planning to start pilot manufacture and further check of products, which will take about three months. Thus, we expect the commercial production to start in August 2015.
Although the full-scale production is still a future challenge, we have been participating in commercial projects right after Phase Zero was completed, being an intermediary company applying magnetic layers on wafers. With the launch and development of technological lines, the types of works will be diversified until the production reaches 500 wafers per week. The total production area will be 8.2 thousand square meters, including 2.4 thousand square meters of clean premises. If necessary, the output may be increased to 1000 wafers per week.
Why have you chosen Technopolis Moscow as a manufacturing site?
The companies localized in Technopolis Moscow enjoy a simplified construction approval procedure, electric infrastructure, lower income tax and Moscow government subsidies, so we found the conditions here to be the best for implementation of our plans.
What problems have you run into during the implementation of the project?
We had some issues with attraction of supplementary investments because Russian financial institutions were reluctant to invest into innovational projects, but these problems have been resolved.
Are you planning to create a CNE research center and to cooperate with Russian R&D Institutions?
For the time being, CNE will only be a manufacturer. However, our further development plans foresee systemic solutions and modern application of technologies in Russia.
How effective do you find the participation of Rusnano in the project?
Cooperation with Rusnano is highly productive. The Russian founder steadily fulfils all its obligations and provides versatile support to the project.
Thank you for the interesting interview.
The interview was taken by
D.Gudilin and A.Tsaplin