Phase-change materials open up a new frontier for storage chips.

Memory is one of the most critical technologies in integrated circuits and a key indicator of their core competitiveness. However, as the world’s largest consumer of ICs, China still lags behind major overseas semiconductor companies such as Samsung and Intel in terms of both memory technology and product capabilities. Consequently, it has become imperative to accelerate research and development into memory chip materials, which has brought phase-change memory (PCM) into the spotlight. Unlike traditional non-volatile memory based on NAND, PCM devices can support virtually unlimited write cycles. Moreover, PCM boasts several other advantages, including short access response times, byte-addressability, and random read/write capabilities. For this reason, PCM is often hailed as one of the storage technologies poised to “reshape the future.” At the same time, phase-change materials have emerged as a key focus in the study of memory chip materials.

        One of the storage technologies that “changes the future”
In recent years, the rapid advancement of integrated circuit technology has placed increasingly stringent demands on various performance metrics of memory chips, including power consumption, lifespan, size, and endurance. Scientists around the world are stepping up their efforts to tackle key challenges in the development of new storage materials. It is reported that phase-change memory is a high-performance, non-volatile memory technology whose phase-change materials are based on chalcogenide glass compounds. These compounds possess a crucial property: when they transition from one phase to another, their electrical resistance can be altered. The crystalline phase of these materials exhibits low resistance, whereas the amorphous phase has high resistance. The phase transition is achieved by applying or removing an electric current.
In 2017, Dr. Song Zhitang, Director of the Nanomaterials and Devices Laboratory at the Shanghai Institute of Microsystem and Information Technology of the Chinese Academy of Sciences, led a research team to achieve a major breakthrough in the study of novel phase-change storage materials. They innovatively proposed a design concept for high-speed phase-change materials: by reducing the randomness of nucleation within the amorphous phase-change thin films, they succeeded in achieving rapid crystallization of these materials.
Song Zhitang told a reporter from China Electronics News that the phase-change storage material currently in widespread international use is “germanium-antimony-tellurium” (Ge-Sb-Te), and many chip manufacturers are already conducting related research. For example, memory chip maker SK Hynix began producing 3D crosspoint memory based on phase-change materials in 2018. The 3D crosspoint storage cells used for SCM are made from sulfur-based phase-change materials. Moreover, IBM Research has shown that by using analog chips based on phase-change memory, machine-learning capabilities can be accelerated by a factor of a thousand. IBM has also revealed that it is establishing a research center to develop next-generation AI hardware and explore the potential applications of phase-change memory in the field of artificial intelligence.

        Phase-change materials enable storage devices to be compatible with new CMOS technology.
Memory plays a pivotal role in the semiconductor industry. According to data from China Industry Information Network, in 2018, the global semiconductor market was valued at US$478 billion, with the memory market accounting for US$165 billion—representing 35% of the global semiconductor market size. Today, the memory industry has evolved into three relatively independent markets: DRAM chips, NAND Flash chips, and specialized memory devices. However, as Moore’s Law continues to extend, technological demands are steadily increasing, and the shortcomings of traditional memory chips are gradually becoming more apparent.
“As chip technology nodes approach their physical limits, the number of electrons in capacitors decreases, making DRAM memory more susceptible to external charge influences. Flash memory, meanwhile, faces severe crosstalk issues during operation, which can shorten its service life. SRAM also encounters certain challenges in terms of signal-to-noise ratio and soft errors. Moreover, these problems become even more pronounced when chip manufacturing processes fall below 28nm,” Song Zhitang told a reporter from China Electronics News.
In addition, Song Zhitang pointed out that previous storage technologies, such as DRAM and flash memory, were incompatible with the new application CMOS technology that employs high-k dielectrics, metal gates (MG), and fin structures. Consequently, around the world, research and development efforts are underway to create non-volatile storage technologies that are compatible with these emerging CMOS technologies and possess excellent scalability, three-dimensional integration capabilities, fast computing performance, low power consumption, and long service life. Phase-change memory is one such technology.
Song Zhitang introduced that a new type of phase-change material, which uses stable octahedra as nucleation centers to reduce the randomness of amorphous nucleation and thereby achieve rapid crystallization of the phase-change material, was developed under the guidance of an original theory on self-organized octahedral building blocks and metastable face-centered cubic structures. This approach innovatively proposes a research and development strategy in which two octahedral lattices are matched with their electronic structures. Through first-principles theoretical calculations and molecular dynamics simulations, among numerous transition-metal elements, scandium and iridium (Sc and Y) were selected as dopant elements. By conducting storage-cell performance tests—particularly high-speed erase/write tests on the storage cells—an ultra-high-speed, low-power, long-lifetime, and highly stable “scandium-antimony-tellurium” (Sc-Sb-Te) phase-change material was invented. Phase-change memory devices based on the Sc-Sb-Te material, fabricated using a 0.13-μm CMOS process, demonstrated a high-speed reversible write/erase operation time of just 700 picoseconds and a cycle life exceeding 10^7 cycles.
Compared to conventional Ge-Sb-Te-based phase-change memory devices, this new type reduces operating power consumption by 90% while maintaining comparable data retention over a decade. By further optimizing the materials and shrinking device dimensions, the overall performance of Sc-Sb-Te-based PCRAM will be further enhanced. Song Zhitang explained that the stable octahedral structure of Sc-Te serves as the nucleation and growth core, which is the primary reason for the device’s high speed and low power consumption. The matching of lattice and electronic structures is the main factor behind its long lifespan, and the stabilization of the octahedral structure also inhibits the transformation from face-centered cubic (FCC) to hexagonal close-packed (HEX), contributing to both high speed and low power consumption.

        Many hands make light work.
The semiconductor electronics industry market has now become one of the primary application areas for phase-change materials. According to QYResearch data, in 2018, phase-change materials accounted for 17.69% of the semiconductor electronics industry market share. In 2019, the total market value of phase-change materials reached 5.4 billion yuan, and is projected to grow to 12.1 billion yuan by 2026, with an annual growth rate of 12.2%.
Song Zhitang believes that as the applications of storage devices continue to expand, phase-change storage materials will also see increasing use in the semiconductor industry in the future. However, for phase-change materials to move beyond the initial stage of basic innovation in the semiconductor field, the next step is to move out of the laboratory and be integrated into a wider range of products.
“Currently, phase-change materials are still in the early stages of innovation, and there are limited references available for guidance. However, many hands make light work—if we hope that phase-change materials can be more widely adopted in the semiconductor industry, the entire industrial chain needs to pay close attention and collaborate closely for joint development. Only through the close integration of industry, academia, and research can we truly achieve a qualitative leap,” said Song Zhitang.
Regarding the future technological development needs of phase-change materials in the semiconductor industry, Song Zhitang believes there are three main points. “First, high purity. For semiconductor materials, the requirement for purity is extremely stringent; all semiconductor materials necessitate the purification of raw materials, and low purity can directly affect device performance. Second, high reproducibility. Given that the semiconductor industry is a high-tech sector with long R&D cycles and high R&D costs, semiconductor materials must exhibit excellent reusability. Third, high reliability and stability. In the semiconductor field, every single step is critically important, so utmost rigor is required—otherwise, even a small mistake can snowball into a cascade of errors. This holds true for upstream materials as well,” said Song Zhitang.
(Interview Location: 2020 China Semiconductor Materials Innovation and Development Conference)