Progress Starts with Materials for Chipmaking, Says Entegris
Mitchell Gracie posted on March 07, 2019 |
Jim O’Neill, CTO, elaborates on the necessity of material science in manufacturing.
Entegris is an advanced materials and specialty chemicals manufacturer based in Massachusetts. (Image courtesy of Entegris.)
Entegris is an advanced materials and specialty chemicals manufacturer based in Massachusetts. (Image courtesy of Entegris.)

“Chipmaking is a ‘cascading’ process,” says Entegris. For a company that produces market-leading research into material science for micro-electronics, its words hold a lot of weight.

For years, the industry of chipmaking has been driven by Moore’s Law—the idea that the number of transistors in an integrated circuit doubles about every two years. However, as we move into the realm beyond 10 nm transistors, that old guarantee isn’t so guaranteed anymore. As much as Moore’s Law has led to decades of miniaturization, progress based on that decades-old assurance is slowing down. Luckily, pressure from the Internet of Things (IoT), autonomous vehicles (AVs) and the advent of 5G telecommunication has engineers and scientists finding clever, new ways to push miniaturization forward in spite of any slowdowns to Moore’s Law.

Entegris, the Massachusetts-based specialty materials company focusing in the semiconductor business, is one of the companies of engineers and scientists diligently innovating ways to better devices and the chips embedded within them. Its focus on advanced and specialty materials enables them to provide performance, purity and safety to its products.

“[Moore’s Law] doesn’t really work anymore, given the physics of optical lithography that is used to achieve those historical gains. Today, the majority of the performance and density gains come from the use of new, innovative materials. That is where Entegris comes into play,” says Jim O’Neill, chief technological officer of Entegris.

According to O’Neill, it doesn’t matter if it is “new metallurgies that are used to improve resistance or reliability of the fine wires that connect transistors, new channel materials that allow for faster electron transport in a transistor, or new dielectrics that reduce signal leakage in a device,” Entegris is pushing the envelope in manufacturing chips.

Jim O’Neill, chief technology officer at Entegris. (Image courtesy of Entegris.)
Jim O’Neill, chief technology officer at Entegris. (Image courtesy of Entegris.)

When it comes to the paradigm of massive data streams from billions of sensors around the world, the need for processing, transporting and storing such data requires faster, more reliable hardware. It doesn’t help that the technologies within devices can quickly become outdated, nor does it help that materials or architectures optimal for one implementation may not be optimal for another implementation. According to O’Neill, customers are increasingly relying more and more on materials to improve performance in the face of the slowing down of Moore’s Law. This shift increases complexity in manufacturing, especially in an industry where impurities and defects can be detrimental to the integrity of a chip.

“This is an area of strength for Entegris in terms of the fact that we provide highly clean formulations that allow customers the yields they need to make their manufacturing cost effective,” says O’Neill. That is, avoiding chip-killing defects is mission number one.

Ground-floor for the manufacturing of microelectronics

Regardless of implementation, a chip cannot be made without starting with its materials. Moreover, there is no perfect material that solves every problem. As great as a material with high-performance characteristics can be, if the material cannot be integrated into complex build structures, then it is back to the drawing board.

Once specifically-desired properties of a material are discovered, companies like Entegris must investigate if the material is what O’Neill calls “integrable.” That is, he says, “once you put the material into a device or a stack of other materials, you need to know how it acts among those materials, so it doesn’t lose its performance.”

Further, a material that is integrated and performing as designed in the lab or on the manufacturing floor does not guarantee that the material will be integrated or perform well once it reaches its end user. Quality assurance, O’Neill elaborates, remains in the minds of researchers throughout the whole process.

The globalization of markets adds to the list of hurdles. With customers and manufacturers dispersed around the world, O’Neill says that the supply chain can span several continents and add months in between a manufacturing floor and a customer. Along such a globalized supply chain, delicate materials might not be delivered in their intended state. Focusing on “methodologies, products and capabilities [that ensure] the clean, safe-handling, storage, purification and delivery of those new and innovative materials,” according to O’Neill, is the focus for companies like Entegris.

Materials meet architecture, architecture meets materials

According to O’Neill, there exists an “interplay between materials and architecture because all of the most advanced devices have shifted, in some regard, into a third dimension.” In fact, he asserts that “none of those devices could be fabricated without advances in the materials themselves.”

To expand, O’Neill provides the example of the planar process: the primary way by which an overwhelming majority of integrated circuits (ICs) are manufactured. “If you want to build a transistor, you build a switch by stacking materials on top of each other and patterning those films in a way that is relatively straightforward.” However, things change when architecting 3D structures. “Once you’ve configured the transistor in the vertical dimension, you now have to deposit films not just along the top planar surface, but on the sides as well.” The added complexity to manufacturing 3D architectures “brings in requirements of the materials [used] themselves.”

At the magnitude of nanometers, chipmakers are venturing into the territories of highly-precise engineering, necessitating manipulations to the material at almost atomic scales (i.e., atomic-level deposition and atomic-layer etching).

“Without advanced system materials, you wouldn’t be able to engineer these structures,” says O’Neill.

Innovative movements in materials research

One can divide a chip into three pieces during its manufacturing:

1.     The front-end of the line where the device is built of silicon on a wafer.

2.     The back-end of the line where very-fine wires are created to be connected to the transistor.

3.     The middle of the line, which is where the contact between the transistor and the wiring is established.

O’Neill is excited about the work and progress that is being made every day in the back-end of the line.

According to him, as you make the wiring smaller and smaller, there are two roadblocks. First is the resistance of the wire, which—due to Pouillet’s Law—increases as the cross-section of the wire decreases. Second is the reliability of the wire, which degrades as the wire thins.

Fighting against the rigidity of physics, Entegris is looking at novel metallurgies to fill the gaps where O’Neill says copper begins to run into problems.

“As we approach to magnitudes less than 10 nm, the resistivity of copper and its reliability begin to be challenged at the smallest features. So, we’re looking a host of metallurgies that can be used to improve reliability and resistivity overall, such as cobalt, ruthenium or molybdenum.”

Another roadblock is the nature of the dielectric material used as an insulating film. “As you put these very small wires close together, you have to ensure you have a material that can prevent cross-talk between signals in one wire that are only nanometers away from signals in another wire,” says O’Neill.

“As far as the device is concerned, the heart of the device is the gate which switches the channel on and off, so there is a lot of work looking into silicon, germanium and III-V semiconducting materials that improve electron mobility.”

Materials motivating tomorrow’s Internet of Things and autonomy

O’Neill sees an interesting relationship between today’s increasing demand for smarter, better and faster sensors with strides in materials research at the chip level. “Usually the things and their sensors that are connected aren’t all that complex or high-performance.” However, the interesting part, according to O’Neill, is dealing with the enormous amount of data that all of those devices generate in aggregate. “Those data have to be handled, transmitted, stored and processed somewhere,” he elaborates.

“What we are seeing is that with this explosion of data, driven by IoT, is forcing the capabilities in the back-office to improve the higher-density memory, like 3D NAND, higher performance, logic processors or specific application processors (for example, language recognition being able to distinguish between voice and sound, or image recognition that can identify a face from a shadow in surveillance footage).”

Entegris sees its innovation of materials enabling solutions to new demands of specific applications’ needs for advanced performance, higher-density memory or faster communication.

“Historically, a central processor or logic-centric model where you have a core processor in the center, and you surround it with data. Today, however, we are seeing increasingly larger pools of data surrounded by specialized processors that are designed to accomplish a specific task,” O’Neill says. “We’re going to see a more diverse array of designs—still using processing technologies, architectures and materials but in very specific application-driven ways.”

While the enabling of high-performance devices to improve central computation capabilities in the cloud can be great for some applications, progress in low-power devices enables for at least some of that processing to take place locally.

With autonomous vehicles basically being what O’Neill calls “sensorized servers on wheels,” the addition of crucial milliseconds in transit time for data—from device to cloud and back—can be the difference between life and death. “You can now achieve the speed and capability of those kinds of computations and decisions to reside on a vehicle and it begins to enable those applications, like autonomous vehicles,” he says.

“Where people choose to put those capabilities—edge or cloud—will boil down to a cost and functionality point. These problems are relatively new and those are the questions being asked, so stay tuned!”

Recommended For You