The Basics of the DRAM Structure and Scaling

A single DRAM chip contains anywhere from hundreds of millions of cells to billions of them, depending on data capacity. Each cell consists of two parts: a capacitor that stores data in the form of an electrical charge, and a transistor that controls access to it. The two parts are collectively referred to as a DRAM cell. The latest DRAM from Samsung is an 8Gb (gigabit) chip, meaning it has more than 8 billion cells.

The DRAM chip is produced from a very thin silicon board called a “wafer.” As the circuit design and process technology—which are applied to a wafer—become more refined, the wafer is able to produce more chips.

A single 10nm-class DRAM wafer developed by Samsung produces more than 1,000 chips, which is 30 percent more than what could be produced on a 20nm DRAM wafer.

Using the new 10nm-class manufacturing technology, 30 percent more chips can be produced from a single wafer than from a 20nm wafer.

3 Core Technologies Behind Samsung’s 10nm-class DRAM Production

Samsung developed three innovative technologies to successfully mass produce 10nm-class DRAM: Samsung’s proprietary cell design technology, QPT (quadruple patterning technology), and ultra-thin dielectric layer deposition.

QPT Achieves Maximum DRAM Scaling Using Currently Available Photolithography Equipment

With the new 10nm-class DRAM, Samsung became the world’s first semiconductor manufacturer to succeed in applying QPT to DRAM mass production. But what exactly is this technology?

In semiconductor engineering, the core of the business is designing and integrating nanometer-scale circuits onto a small, nail-sized chip. The photolithography process refers to printing electric circuit patterns on a wafer in a way that resembles printing a photo. While a basic photolithography process prints a single pattern, multiple patterning technologies, like double patterning technology (DPT) and QPT, print up to two and four patterns respectively. Multiple patterning is widely used for advanced memory products that require a high level of scaling and density.

In case of the QPT used for Samsung’s 10nm-class DRAM, the photolithography process itself is done once and, after that, many steps are added to realize the QPT, as can be seen in the diagram. As a result of these steps, four patterns can be produced in the same surface area. The whole point of multiple patterning is about drawing more circuit patterns in the same space using currently available photolithography technology, thereby maximizing the wafer productivity.

Ultra-Fine Dielectric Layer Deposition That Insulates Electrical Current

Printing more patterns in the same given space is not the end of the story. A DRAM chip does not function properly if even a single cell—out of hundreds of billions—does not function properly. Therefore, ensuring that all cells work well in a sufficient amount of time is a prerequisite to produce a DRAM chip with high performance and power efficiency. To this end, each and every capacitor (where the electric charges representing data are contained) needs to be thin, long, and sturdy. This is where ultra-fine dielectric layer deposition technology comes in.

Capacitors are used to contain electrical charges (data) temporarily. They can be either charged or discharged to represent the two values of a bit: 1 and 0. When a capacitor contains a sufficient amount of electrons, it can quickly determine the digital signals. Since the manufacturing process only gets more refined, capacitors have to be thinner and longer in order to contain enough electrons.

When constructing a capacitor, it must be covered with a thin but solid dielectric material in order to prevent its electrical charge from leaking and the electrical charges in the surrounding capacitors from causing interference (the space between each capacitor is only a few tens of nanometers). The ability to maintain the uniformity of these dielectric layers is the core technology that determines the quality of the manufacturing and product competitiveness.

With the previous technology for 20nm chips, the dielectric layers were thicker at the top than the bottom, making the capacitors look like upside-down cones. This was not much of a problem. However, as the capacitors and the space between them has become thinner, the dielectric layers have also had to get thinner. To address the problem, Samsung developed a new material through the ultra-fine dielectric deposition technology, successfully making the thickness of the dielectric layers uniform to a few angstroms (tenths of a nanometer). This technical breakthrough allowed the birth of the 10nm-class DRAM with high performance and reliability.

Proprietary Circuit Design Technology That Incorporates All Core Technologies and Reduces Energy Consumption

In general, semiconductors consume more power in proportion to the speed at which they operate. Which is why the increase of speed and reduction of power consumption in the new 10nm-class DRAM is so remarkable.

In an idle state, DRAM consumes less energy than when it is operating in an active state. The 10nm-class DRAM is designed to accelerate its performance in controlling and processing data while in an active state and then go back to an idle state as soon as possible. As a result, the 10nm-class DRAM is both faster and more power efficient when processing data.

More specifically, the 10nm-class DRAM operates at 3.2Gbps on PC and server systems (up from the 2.4Gbps of its predecessor), while reducing power consumption by 10 to 20 percent.

Faster speeds can simply be achieved by consuming more power, but energy efficiency is also a very important factor in any computing system, including PC, mobile and server applications. Because of this, achieving higher performance and reducing power consumption at the same time is essential for DRAM products, which are used in all kinds of advanced computing systems today.

Based on its advancements with the new 10nm-class DRAM technology, Samsung expects to also introduce a 10nm-class mobile DRAM solution this year, which is faster and uses less power than currently available mobile DRAM solutions. The new mobile DRAM solution will be able to support battery-dependent, leading-edge mobile devices with high resolution features such as FHD video (which is currently the most common standard) and 4K UHD videos.

All three of these key technologies together enabled the industry’s first 10nm-class DRAM: QPT to print more patterns at a smaller scale, ultra-fine dielectric layer deposition technology to build thinner and stronger capacitors, and Samsung’s proprietary circuit design technology to achieve high performance and energy efficiency. By combining these novel innovations, Samsung came up with the 10nm-class DRAM solution, improving productivity and guaranteeing a stable supply of next-generation memory products for global IT businesses.

The productivity and speed of 10nm-class DRAM has increased by more than 30 percent compared to 20nm DRAM, while the power consumption has reduced by 10 to 20 percent. Let’s look forward to seeing how Samsung will continue to innovate the premium DRAM market, and facilitate the growth of the PC, server and mobile markets.

* Nano (n) is a prefix that means “one billionth.” For measuring distances, a nanometer is nearly inconceivably tiny, much smaller than the eye can see. For instance, a human hair is approximately 80,000 to 100,000 nanometers (nm) wide.

** Scaling refers to the process of shrinking semiconductor cells, a crucial factor in fabricating ever more powerful semiconductor chips. While memory scaling improves manufacturing productivity, it also contributes to the development of high-performance, energy-saving IT and mobile products with longer battery life and better usability for consumers. Therefore, competition is fierce to reduce every possible nanometer, to break the previous limits and help keep the overall industry growing.

As the amount of data services and smartphone devices dramatically increases, more spectrum is required to accommodate fast growing data traffic and help ensure a great user experience. LTE in Unlicensed spectrum includes a key family of technologies that allows aggregation of licensed and unlicensed spectrum bands, such as 5 GHz, to efficiently and timely increase network capacity and speed.

The Samsung LTE-U eFemto cell incorporates a Qualcomm FSM9955 chipset from Qualcomm Technologies. Through aggregation of licensed and unlicensed spectrum, Samsung’s LTE-U eFemto cells can help mobile operators cope with the skyrocketing demand for mobile data by improving network performance in hotspots, venues and enterprise environments including offices, malls, hospitals and other public spaces and facilities.

The Qualcomm FSM9955 chipset incorporates sophisticated enhanced Carrier Sensing Adaptive Transmission (eCSAT) features for coexistence with Wi-Fi per the latest LTE-U Forum specifications released last October, designed to enable fair sharing of unlicensed spectrum in both below and above Wi-Fi’s Energy-Detect (ED) threshold scenarios.

Using the Qualcomm FSM9955 chipset, the Samsung LTE-U eFemto cell can also support LTE Licensed-Assisted Access (LAA) by simple software upgrade. LAA incorporates “Listen-Before-Talk” features required in certain regions for fair co-existence with Wi-Fi, making it a global solution for bringing LTE into unlicensed spectrum. LAA is part of the 3GPP LTE Advanced Pro Release 13 standard to be finalized in early 2016 and then expected to be rapidly adopted by operators around the world.

“We are pleased to collaborate with industry leader Samsung in accelerating small cell innovations and extending the benefits of LTE to unlicensed spectrum,” said Neville Meijers, Vice President of Small Cells, Qualcomm Technologies, Inc. “Making adequate use of both licensed and unlicensed spectrum bands is essential to meet data demand and provide end users with the best possible mobile experience.”

“We are proud to bring this innovative LTE-U technology into our small cell portfolio. Operators and enterprises need a combination of cutting-edge technology and affordability to meet the explosive demand for data, both today with 4G and beyond.” said Kiho Cho, Vice President of Network System Design Lab, Network Business at Samsung Electronics. “Samsung LTE-U eFemto cells will benefit end users by enabling faster data communications, with seamless mobility and reliable performance, while minimizing interference with Wi-Fi.”

The Samsung LTE-U eFemto cell is designed to enable plug-and-play deployments with easier installation and a high capacity for both enterprises and operators. A single small cell unit can support three carriers of 20 MHz each, across both licensed and unlicensed spectrum, offering a peak download throughput up to 450Mbps.

The Samsung LTE-U eFemto cell with Qualcomm FSM9955 chipset will be exhibited at the Qualcomm booth (Exhibit Hall 3 #3E10) and the Samsung booth (Exhibit Hall 2 Stand 2M10) during Mobile World Congress 2016, February 22-25 in Barcelona, Spain.

About Qualcomm Incorporated
Qualcomm Incorporated (NASDAQ: QCOM) is a world leader in 3G, 4G and next-generation wireless technologies. Qualcomm Incorporated includes Qualcomm’s licensing business, QTL, and the vast majority of its patent portfolio. Qualcomm Technologies, Inc., a subsidiary of Qualcomm Incorporated, operates, along with its subsidiaries, substantially all of Qualcomm’s engineering, research and development functions, and substantially all of its products and services businesses, including its semiconductor business, QCT. For more than 30 years, Qualcomm ideas and inventions have driven the evolution of digital communications, linking people everywhere more closely to information, entertainment and each other. For more information, visit Qualcomm’s website, OnQ blog, Twitter and Facebook pages.

“By mass producing next-generation HBM2 DRAM, we can contribute much more to the rapid adoption of next-generation HPC systems by global IT companies,” said Sewon Chun, senior vice president, Memory Marketing, Samsung Electronics. “Also, in using our 3D memory technology here, we can more proactively cope with the multifaceted needs of global IT, while at the same time strengthening the foundation for future growth of the DRAM market.”

The newly introduced 4GB HBM2 DRAM, which uses Samsung’s most efficient 20-nanometer process technology and advanced HBM chip design, satisfies the need for high performance, energy efficiency, reliability and small dimensions making it well suited  for next-generation HPC systems and graphics cards.

Following Samsung’s introduction of a 128GB 3D TSV DDR4 registered dual inline memory module (RDIMM) last October, the new HBM2 DRAM marks the latest milestone in TSV (Through Silicon Via) DRAM technology.

The 4GB HBM2 package is created by stacking a buffer die at the bottom and four 8-gigabit (Gb) core dies on top. These are then vertically interconnected by TSV holes and microbumps. A single 8Gb HBM2 die contains over 5,000 TSV holes, which is more than 36 times that of a 8Gb TSV DDR4 die, offering a dramatic improvement in data transmission performance compared to typical wire-bonding based packages.

Samsung’s new DRAM package features 256GBps of bandwidth, which is double that of a HBM1 DRAM package. This is equivalent to a more than seven-fold increase over the 36GBps bandwidth of a 4Gb GDDR5 DRAM chip, which has the fastest data speed per pin (9Gbps) among currently manufactured DRAM chips. Samsung’s 4GB HBM2 also enables enhanced power efficiency by doubling the bandwidth per watt over a 4Gb-GDDR5-based solution, and embeds ECC (error-correcting code) functionality to offer high reliability.

In addition, Samsung plans to produce an 8GB HBM2 DRAM package within this year. By specifying 8GB HBM2 DRAM in graphics cards, designers will be able to enjoy a space savings of more than 95 percent, compared to using GDDR5 DRAM, offering more optimal solutions for compact devices that require high-level graphics computing capabilities.

The company will steadily increase production volume of its HBM2 DRAM over the remainder of the year to meet anticipated growth in market demand for network systems and servers. Samsung will also expand its line-up of HBM2 DRAM solutions to stay ahead in the high-performance computing market and extend its lead in premium memory production.