SMC 2013 Preview:
Looming IC Industry Changes Mean Proliferation of New Materials Needs
By Paula Doe, SEMI Emerging Markets
The onslaught of next generation technologies coming to the semiconductor industry will bring demand for a host of new materials, but also an increasingly fragmented materials market, making strategic investment decisions for materials development even more risky and complex than usual.
As the semiconductor industry transitions to EUV lithography, new memory and logic architectures, 3D packaging, and 450mm wafers, adjacent markets from MEMS to flexible electronics also demand new materials of their own. Navigating all these opportunities will require realistic judgements of market potential and timing, but new approaches — including fundamental materials modeling, more efficient screening methods, and better industry collaboration — could also be key to easing profitable development. Higher-resolution, higher-speed inspection and metrology tools will also be vital to move these new materials into volume production.
“Perfect Storm” of Major Disruptions in Materials
The industry is facing five major disruptions that will impact materials suppliers, as well as the rest of the industry, notes Mark Thirsk, managing partner, Linx Consulting. One is that whatever happens with EUV, the future of lithography will likely remain a complex multistep process using many ancillary layers. “Actual immersion lithography resolution hasn’t progressed for the last six years or so — we’ve just gotten clever using horrendously complex self aligned double patterning and spacers. When EUV does come along, it will probably not be a return to the simplicity of a single resist step,” he suggests. “Directed self assembly, image transfer, tri-patterning — a lot of that will be needed with EUV as well.”
Then next-generation logic’s need for higher speeds will likely mean a move to new, higher mobility channel materials such as germanium and III-Vs within the next few years, notes Thirsk, who will discuss the electronic materials market outlook at SEMI’s upcoming Strategic Materials Conference (www.semi.org/smc) in October. That means different precursors and different materials for cleaning and patterning. Germanium’s common surface passivation form GeO2, for example, is soluble in water, so will need solvent cleans, or else a different material for passivation. III-Vs will also need different processes for cleaning and patterning.
Next generation NAND flash memory is also on the verge of changing over its materials set, with Samsung’s recently announced stacked charge trap memory as one alternative, bringing manufacturing challenges of controlling deposition and etch of these new materials stacks. DRAMs too are running out of space as the current high-aspect ratios for the capacitors are getting difficult to pattern, pushing the sector toward spin torque magnetic junction or MRAM technologies, which will bring challenging new stacks of varied diverse materials, where edge effects become critical and etching of things like cobalt and iron become problematic.
Stacked 2.5D/3D packaging of course means more materials challenges, with its need for low cost interposer material, high aspect ratio via etch and fill, and temporary bonding and debonding.
Finally, the transition to 450mm means new requirements for wafer and film uniformity — but it also likely means a decrease in the growth of materials demand, as materials usage generally scales with wafer area. Resist usage on 300mm wafers is now about the same as it was on 200mm, and is likely to be about the same for 450mm. Wet chemical baths will need somewhat more to cover the 450mm wafer, and CVD will use somewhat more chemicals, but it’s not likely to be more than about 30 percent more.
While total materials growth may slow with larger wafers and more efficient vertical use of wafer area, that volume will likely also be divided among a larger variety of diverse specialty materials for different applications, in smaller volumes each.
Speeding Development will need Better Collaboration, Modeling, and Metrology
The exploding need for all these new materials will require the industry to find more efficient ways to screen materials in development, and new technology to measure and control film quality and uniformity at the few-nanometer scale in high-volume production, and at low cost. “And cost is the hardest part,” says Michael Lercel, senior director of nano defectivity and metrology for SEMATECH, who will also discuss these issues at SMC. Materials suppliers face particular risk for return on development investment as the market increasingly fragments among specialized applications, with a decreasing number of end customers.
One key may be pre-competitive cooperation to narrow down the materials options more quickly. “The quicker you can have some decision on a smaller number of materials, the better,” he says, noting the ten different candidates currently under investigation in the 2D monolayer materials space as a problematic example. “And this will take a collaborative environment with the right partnerships across the supply chain to come to agreement.”
In addition, use of modeling of the fundamental materials properties can better predict the composition and structure for the desired properties. The big drop in prices recently for the heavy duty supercomputer power required now makes such modeling possible, and the life sciences have made considerable progress on these techniques. This first principal's modeling is well suited to university capabilities, Lercel notes, so the industry needs to work with academia to help them focus on key electronic materials issues. More efficient experiment design for screening materials, such as with combinatorial synthesis tools, will also likely be needed to help down-select materials more quickly to reduce development costs and risks.
And then moving the selected materials for these next generation nodes into volume production with the required nanoscale control of purity and uniformity will require new technology for defect inspection and film quality metrology. To control production, the industry needs a solution for wide-area inspection for 10nm defects, either by finding a way to speed up e-beam inspection, or by improving the resolution of optical tools. Lab tools and off-line technologies exist for the needed film quality information, but they are slow or destructive or both. “Doing all this at low cost is the real challenge,” he notes. “Though it’s both a big challenge and also a big opportunity for material supplier partnerships.”
Adjacent Markets Bring New Materials Challenges and Opportunities too
Demand for MEMS materials can expect healthy growth to scale along with MEMS wafer area, as demand for sensors and actuators in and connected to smart phones continues to explode, driving MEMS to a more mainstream volume manufacturing business. Yole Développement expects the annual growth in area shipments for MEMS devices to see compound average annual growth of better than about 9 percent in 8-inch equivalent wafers over the next 5-6 years, from about 2.2 million in 2012 to some 3.8 million in 2018.
Graph shows the timeline for new MEMS processes adoption. The left side of the arrow is showing the starting time for the technology (e.g. DRIE started to be used in 1996 for Bosch inertial MEMS). We expect to see more innovative MEMS processes in the future: TSV, litho stepper, temporary bonding for thin wafers, room temperature bonding. Source: Yole
Piezoelectric materials could be a disruptive material for next generation MEMS. Piezoresistive sensing could potentially enable smaller, higher performance inertial sensors by significantly reducing the larger proof masses needed by current capacitive technology, by technologies. PZT thin film could improve actuators as well, as the PZT has a larger piezo effect than AlN, though it does take higher temperatures and longer times to deposit. “Thin film PZT could find applications in ultrasonic MEMS imagers, ink jet heads and auto focus lenses, for example,” suggests Yole senior analyst Eric Mounier. “Note that the only recent new MEMS equipment startup, Solmates, is offering pulsed laser deposition for PZT.”
Magnetic thin film materials are also a growing opportunity, for compass devices, and also for actuation in applications such as scanning mirrors for pico projectors. While not technically a MEMS device, magnetic sensors are increasingly combined with MEMS inertial sensors, driving demand for more deposition and processing of magnetic thin films. As MEMS makers seek in-house control of all the components in their combination motion sensor units, and as MEMS systems makers seek higher performance and easier integration with inertial sensors, a variety of other magnetic resistive technologies are emerging to challenge the Hall-effect technology.
Reducing the size and cost of multi-die integration is also a major issue for MEMS. That means demand for materials for more direct bonding, bumping or 3D connections between the MEMS and the ASIC or the module substrate or board. Use of 2.5D/3D integration with low density TSVs using a variety of materials technology is already a commercial reality for MEMS, in products from FBAR duplexers to oscillators to accelerometers and gyroscopes, and more such products will follow soon. Room temperature bonding is in development at several suppliers, as a possible alternative for stacking MEMS wafers directly with their ASIC controllers without damaging the CMOS. Use of gold or other metal-to-metal bonding materials could also see growing usage to reduce die size by replacing wider glass frit lines.
Engineered substrates like cavity SOI could start to see wider application, especially if InvenSense’s open platform using that approach attracts more users. SOI suppliers are offering wafers prepared with cavities, trench isolation and TSVs.
Learning from the Experience of other Emerging Materials
For figuring out how to make the right strategic choices for investing in these new materials, it could help to look at the experience of past generations of emerging material technologies, both as business case studies of individual companies, and for materials types in the aggregate.
Lux Research is working on developing a database of advanced materials development in the past that can be used as a strategic predicative tool for forecasting the commercialization trajectories of other emerging material technologies. “If we look at the timeframe it took, say, carbon fiber to get from the first 100 patents to $100 million in revenues, what can that tell us about the commercialization timeline of carbon nanotubes and graphene?” says Lux senior analyst Ross Kozarsky. “Technology scouts and venture capitalists alike looking for the next ‘big thing’ often forget that ‘big’ developments will require more patience — often 15 to 25 years — than their typical metrics allow. Our research seeks to develop new metrics based on quantified data.”
These and other invited experts will discuss the key issues for next generation materials in semiconductors and adjacent markets at the SEMI Strategic Materials Conference 2013, October 16-17in Santa Clara, California. For more information, please visit www.semi.org/smc.
September 3, 2013