Part of the  

Solid State Technology

  and   

The Confab

  Network

About  |  Contact

Posts Tagged ‘Intel’

Next Page »

Emerging Memory Types Headed for Volumes

Wednesday, July 18th, 2018

thumbnail

By David Lammers

After decades of R&D, two emerging memory types – the phase change memory-based 3D Xpoint, co-developed by Intel and Micron, and the embedded spin-torque transfer magnetic RAM (e-MRAM) from several foundries – are now coming to the market. One point of interest is that neither memory type relies on the charge-based SRAM and DRAM memory technologies that increasingly face difficult scaling challenges. Another is that both have inherent performance advantages that could extend their uses for decades to come.

3D XPoint is a storage class memory (SCM) based on phase-change that fits in between fast DRAM and non-volatile NAND; it is currently available in both SSDs and sampling in a DIMM form factor. David Kanter, an analyst at Real World Technologies (San Francisco) said the Optane SSDs are selling now but the DIMMs are shaping up to be “an early 2019 story” in terms of real adoption. “People are very excited about the DIMMs, including customers, software developers, the whole computer ecosystem. There is a lot of software development going on that is required to take advantage of it, and a lot of system companies are saying they can’t wait. They are telling Intel ‘give me the hardware.’”

“Intel is taking the long view” when it comes to 3D XPoint (the individual devices) and Optane (the SSDs and DIMMs), Kanter said. “This is a new technology and it is not a trivial thing to bring it to the market. It is a testament to Intel that they are taking their time to properly develop the ecosystem.”

However, Kanter said there is not enough public information about 3D XPoint DIMMs, including performance, price, power consumption, and other metrics. Companies that sell enterprise database systems, such as IBM, Microsoft, Oracle, SAP, and others, are willing to pay high prices for a storage-class memory solution that will improve their performance.The Optane DIMMs, according to Intel, are well-suited to “large-capacity in-memory database solutions.”

According to the Intel Web site, Optane DC persistent memory “is sampling today and will ship for revenue to select customers later this year, with broad availability in 2019.” It can be placed on a DDR4 module alongside DRAM, and matched up with next-generation Xeon processors. Intel is offering developers remote access to systems equipped with Optane memory for software development and testing.

Octane DIMM reaches ‘broad availability’ in 2019

Speaking at the Symposium on VLSI Technology in Honolulu, Gary Tressler, a distinguished engineer at IBM Systems, said “the reliability of 3D NAND impacts the enterprise,” and predicted that the Optane storage class memory will serve to improve enterprise-class systems in terms of reliability and performance.

The DRAM scaling picture is not particularly bright. Tressler said “it could be four years before we go beyond the 16-gigabit size in terms of DRAM density.” DRAM companies are eking out scaling improvements of 1nm increments,” an indication of the physical limitations facing the established DRAM makers.

Al Fazio, a senior fellow at Intel who participated in the memory-related evening panel at the VLSI symposia, and said that the early adopters of the Optane technology have seen significant benefits: one IT manager told Fazio that by adding a layer of Optane SSD-based memory he was able to rebuild a database in seconds versus 17 minutes previously. Fazio said he takes particular pride in the fact that, because of Optane, some doctors are now able to immediately read the results of magnetic resonance imaging (MRI) tests.

“An MRI now takes two minutes instead of 40 minutes to render,” Fazio said, adding that a second-generation of 3D Xpoint is being developed which he said draws upon “materials improvements” to enhance performance.

Chris Petti, a senior director of advanced technology at Western Digital, said DRAM pricing has been “flat for the last five to seven years,” making it more expensive to simply add more DRAM to overcome the latency gap between DRAM and flash. “DRAM is not scaling so there are a lot of opportunities for a new technology” such as Optane or the fast NAND technologies, he said. Samsung is working on a single-bit-per-cell form of Fast NAND.

In a Monday short course on emerging memory technologies at the Symposium on VLSI Circuits, Petti said the drawback to phase change memories (PCMs), such as 3D XPoint, is the relatively high write-energy-per-bit, which he estimated at 460 pJ/bit, compared with 250 pJ/bit for standard NAND (based on product spec sheets). In terms of cost, latency, and endurance, Petti judged the PCM memories to be in the “acceptable” range. While the price is five to six times the price-per-bit of standard NAND, Petti noted that the speed improves “because PCM (phase change memory) is inherently faster than charge storage.”

Source: Chris Petti, Western Digital, short course presentation at 2018 Symposium on VLSI Circuits

Phase-change materials, such as Ge2Sb2Te5, change between two different atomic structures, each of which has a different electronic state. A crystalline structure allows electrons to flow while an amorphous structure blocks the flow. The two states are changed by heating the PCM bit electrically.

Philip Wong, a Stanford University professor, said the available literature on PCM materials shows that they can be extremely fast; the latencies at the SSD and DIMM levels are largely governed by “protocols.” In 2016, a team of Stanford researchers said the fundamental properties of phase-change materials could be as much as a thousand times faster than DRAM.

In a keynote speech at the VLSI symposia, Scott DeBoer, executive vice president of technology development at Micron (Boise, Idaho), said “clearly the most successful of the emerging memories is 3D XPoint, where the technology performance has been proven and volume production is underway. 3D XPoint performance and density are midway between DRAM and NAND, which offers opportunities to greatly enhance system-level performance by augmenting existing memory technologies or even directly replacing them in some applications.”

Currently, the 3D XPoint products are made at a fab in Lehigh, Utah. The initial technology stores 128Gb per die across two stacked memory layers. Future generations can either add more memory layers or use lithographic pitch scaling to increase die capacity, according to Micron.

DeBoer noted that “significant system-level enablement is required to exploit the full value of 3D XPoint memory, and this ongoing effort will take time to fully mature.”

eMRAM Race Begins by Major Foundries

Magnetic RAM technology has been under serious development for three decades, resolving significant hurdles along the way with breakthroughs in MgO magnetic materials and device architecture. Everspin Technology has been shipping discrete MRAM devices for nearly a decade, and the three major foundries are readying embedded MRAM for SoCs, automotive ICs, and other products. The initial target is to replace NOR-type flash on devices, largely due to the large charge pumps required to program NOR devices which add multiple mask layers.

GlobalFoundries, which manufactures the Everspin discrete devices, has qualified eMRAM for its 22nm FD-SOI process, called 22FDX. TSMC also has eMRAM plans.

At the Symposium on VLSI Technology, Samsung Foundry (Giheung, Korea) senior manager Yong Kyu Lee described an embedded STT-MRAM in a 28-nm FDSOI logic process, aimed at high-speed industrial MCU and IoT applications.

Interestingly, Lee said compared with the bulk (non-SOI) 28-nm process, the FD-SOI technology “has superior RF performance, low power, and better analog characteristics than 28-nm bulk and 14-nm FinFET CMOS.” Lee indicated that the FD-SOI-based eMRAM would be production-ready later this year.

Samsung ported its STT perpendicular-MTJ (magnetic tunnel junction) eMRAM technology from its 28-nm bulk to its FD-SOI CMOS process. The company offers the eMRAM as a module, complementing an RF module. The “merged embedded STT MRAM and RF-CMOS process is compatible to the existing logic process, enabling reuse of IP,” he said.

Looking forward to the day when MRAM could complement or replace SRAM, Lee said “even though we have not included data in this paper, our MTJ shows a potential for storage working memory due to high endurance (>1E10) and fast writing (<30ns).

Beyond Embedded to Last Level Cache

As foundries and their customers gain confidence in eMRAM’s retention, power consumption, and reliability, it will begin to replace NOR flash at the 40-nm, 28-nm, and smaller nodes. However, future engineering improvements are needed to tackle the SRAM-replacement.

SRAM scaling is proving increasingly difficult, both in terms of the minimum voltages required and the size of the six-transistor-based bits. MRAM researchers are in hot pursuit of the ability to use replace some of the SRAM on processors with Last Level Cache (LLC) iterations of magnetic memory. These LLC MRAMs would be fabricated at the 7nm, 5nm, or beyond nodes.

Mahendra Pakala, senior director of memory and materials at the Applied Materials Advanced Product Technology Development group, said for eMRAM the main challenges now are achieving high yields with less shorting between the magnetic tunnel junctions (MTJs). “The big foundries have been working through those problems, and embedded MRAM is getting closer to reality, ramping up sometime this year,” he said.

For LLC applications, STT-MRAM has approached SRAM and DRAM performance levels for small sample sizes. At the VLSI symposium, researchers from Applied Materials, Qualcomm, Samsung, and TDK-Headway, all presented work on SRAM cache-type MRAM devices with high performance, tight pitches, and relatively low write currents.

Applied’s VLSI symposium presentation was by Lin Xue, who said the LLC-type MRAM performance is largely controlled by the quality of the PVD-deposited layers in the MTJ, while yields are governed by the ability to etch the MTJ pillars efficiently. Etching is extremely challenging for the tight pitches required for SRAM replacement, since the tight-pitch MTJ pillars must be etched without redepositing material on the sidewalls.

Caption: Lin Xue, et al, Applied Materials presentation at 2018 Symposium on VLSI Technology

Deposition is also difficult. The MTJ structures contain multiple stacks of cobalt and platinum, and the thickness of the multilayers must be reduced to meet the 7nm node requirements.  Any roughness in the interfaces creates secondary effects which reduce perpendicular magnetic anisotropy (PMA). “The performance is coming from the interface, essentially. If you don’t make the interface sharp, you don’t end up with the expected improvement in PMA,” Pakala said.

Applied has optimized a PVD process for deposition of the 15-plus layers of many different materials required for the magnetic tunnel junctions. Pakala said the PVD technology can sputter more than 10 different materials. The Endura-based system uses a multi-cathode approach, enabling each chamber to have up to five targets. With a system of seven chambers, companies can deposit the required variety of materials and, if desired, increase throughput by doubling up on the targets.

The system would include a metrology capability, and because the materials are easily oxidized, the entire system operates at vacuum levels beyond the normal 10E-8 Torr level. For MRAM deposition, operating at 10 to minus 9 or even 10 to minus 10 Torr levels may be required.

“When we start talking about the 7 and 5 nanometer nodes for SRAM cache replacement, the cell size and distances between the bits becomes very small, less than 100 nm from one MTJ to another. When we get to such small distances, there are etching issues, mainly redepositing on the sidewalls. The challenge is: How do we etch at reduced pitch without shorting?” Pakala said.

“Integrated thermal treatment and metrology to measure the thicknesses, all of which has to be done at extremely low vacuum, are major requirements,” he said.

“At this point it is not a question of the basic physics. For MRAM, it is, as they say, ‘just engineering’ from here on out,” he said.

Applied Materials Fields Cobalt Solution for MOL

Thursday, June 7th, 2018

thumbnail

By Pete Singer, Editor-in-Chief

Applied Materials has introduced a set of processes that enable cobalt to be used instead of tungsten and copper for contacts and middle-of-line interconnects. Higher levels of metal, which typically have wider dimensions, will still employ copper as the material of choice, but at more advanced nodes, cobalt will likely be the best option as linewidths continue to shrink. Tungsten will still be used at the gate contact level.

To enable the use of cobalt, Applied has combined several materials engineering steps – pre-clean, PVD, ALD and CVD – on the Endura® platform. Moreover, Applied has defined an integrated cobalt suite that includes anneal on the Producer® platform, planarization on the Reflexion® LK Prime CMP platform and e-beam inspection on the PROVision™ platform. The process flow is shown in FIGURE 1.

While challenging to integrate, cobalt brings significant benefits to chips and chip making: lower resistance and variability at small dimensions; improved gapfill at very fine dimensions; and improved reliability. The move to cobalt, which is underway at Intel, GlobalFoundries and other semiconductor manufacturing companies, is the first major change in materials used as conductors since copper dual damascene replaced aluminum in 1997. “You don’t see inflections this large very often,” said Jonathan Bakke, global product manager, Metal Deposition Products at Applied Materials. “This is a complete metallization change.”

At IEDM last year, Intel said it would use cobalt for its 10nm logic process for several of the lower metal levels, including a cobalt fill at the trench contacts and cobalt M0 and M1 wiring levels. The result was much-improved resistivity– a 60 percent reduction in line resistance and a 1.5X reduction in contact resistance – and improved reliability.

Today, critical dimensions of contacts and interconnects are about 20 nm, plus or minus a few nanometers depending on the customer and how it’s defined. “As you get smaller – and you typically get about 30% smaller with each node — you’re running out of room for tungsten. Copper is also facing challenges in both gap fill and electromigration,” Bakke said.

As shown in FIGURE 2, cobalt has advantages over copper when dimensions shrink to about 10nm. They are presently at 30 nm. It’s not yet clear when that cross-over point will arrive, but decisions will be based on how much resistivity and electromigration improvement can be gained.

Applied Materials started developing cobalt-based processes in the mid-2000s, and released the Volta CVD Cobalt system in 2013, which was designed to encapsulate copper interconnects in cobalt, which helped improve gap fill and electromigration. “It was shortly thereafter that we started depositing thick CVD cobalt films for metalization. We quickly realized that there’s a lot of challenges with doing this kind of metalization using cobalt because of its unique properties,” Bakke said. Cobalt can be reflowed and recrystallized, which eliminates seams and leads to larger grain sizes, which reduces resistivity. “We started looking at things like interfaces, adhesion and microstructure of the cobalt to make sure that it was an efficient material and it had very low resistance and high yield for in-device manufacturers,” he added. One perfected, it took several years before the processes were fully qualified at customers. “This year is when we start to see proliferation and expect HDM manufacturing of real devices with cobalt,” Bakke said.

Companies Ready Cobalt for MOL, Gate Fill

Thursday, December 21st, 2017

thumbnail

By Dave Lammers

Cobalt for middle-of-the-line and trench contacts emerged at the International Electron Devices Meeting, as Intel, GlobalFoundries, and Applied Materials discussed how to best take advantage of cobalt’s properties.

For its forthcoming 10nm logic process, Intel Corp. used cobalt for several of the lower metal levels, including a cobalt fill at the trench contacts and cobalt M0 and M1 wiring levels. The result was much-improved resistivity and reliability, compared with the traditional metallization at those levels.

Cobalt was used for the local interconnects of the Intel 10nm process, improving line resistance by 60 percent. (Source: Intel)

Chris Auth, director of logic technology at Intel’s Portland Technology Center, said the contacted line resistance “provides a good indicator of the benefits of cobalt versus tungsten,” with a 60 percent reduction in line resistance and a 1.5X reduction in contact resistance.

While cobalt was used for the local interconnects, the upper 10 metal layers were copper, with a cobalt cap used for layers M2-M5 to provide a 50X improvement in electro-migration. Intel continued to use tungsten for the gate fill.

John Pellerin, a vice president at GlobalFoundries who directs global research and development, said GlobalFoundries decided that for its 7nm logic technology, ramping in mid-2018, it would replace tungsten with cobalt at the trench contact level, which is considered the first level of the middle-of-the-line (MOL).

“We are evaluating it for implementation into the next level of contact above that. Cobalt trench level contacts are process of record (POR) for the 7nm technology,” Pellerin said in an interview at the 2017 IEDM, held Dec. 2-6 in San Francisco.

High performance logic often involves four-fin logic cells to drive the maximum amount of current from the largest transistor width. “You have to get that current out of the transistor. That is where the MOL comes into play. Junction and MOL resistance optimization is key to taking advantage of a four-fin footprint, and it takes a multi-front optimization to take advantage of that equation.

Pellerin said the biggest challenge with tungsten trench contacts is that the CVD process tends to leave a seam void. “We are always fighting seam voids. With cobalt deposition we get an intrinsic resistance improvement, and don’t get seam voids by pushing tungsten down in there,” Pellerin said.

Tighter Metal Pitches

Scotten Jones, president of consultancy IC Knowledge (Boston), said semiconductor vendors will introduce cobalt when it makes sense. Because it is a new material, requiring considerable costs prior to insertion, companies will use it when they need it.

“Global has trench contacts, while Intel uses cobalt at three levels. But the reason is that Intel has a 36nm minimum metal pitch with its 10nm process, while Global is at 40nm with its 7nm process. It is only at the point where the line gets narrow enough that cobalt starts to make sense.”

Applied Cobalt Solutions

As cobalt begins to replace tungsten at the smaller-dimension interconnect layers, Applied Materials is readying process flows and E-beam inspection solutions optimized for cobalt.

Namsung Kim, senior director of engineering management at Applied Materials, said cobalt has a bulk resistivity that is similar to tungsten, but the barrier thickness required for tungsten at leading-edge transistors is swinging the advantage to cobalt as dimensions shrink.

Line resistance probability plot of cobalt versus tungsten at 12nm critical dimensions. (Source: Applied Materials)

“Compared with tungsten, cobalt has a very thin barrier thickness, so you can fill up with more material. At our Maydan Technology Center, we’ve developed a reflow process for cobalt that is unique,” Kim said. The cobalt reflow process uses an annealing step to create larger cobalt grain sizes, reducing the resistance. And because there is no source of fluorine in the cobalt deposition steps, a thin barrier layer can suffice.

At IEDM, Naomi Yoshida, a distinguished member of the technical staff at Applied, presented a paper describing Applied’s research using cobalt to fill a 5nm-logic-generation replacement metal gate (RMG). The fill is deposited above the high-k dielectric and work function metals, and at the 5nm node and beyond there is precious little room for the gap fill metal.

Yoshida said modern transistors use multiple layers of work-function metals to control threshold voltages, with high-performance logic requiring low Vt’s and IoT devices requiring relatively high Vt’s. After the different work function layers are deposited, the fill material is deposited.

At the 5nm node, Applied Materials estimates that the contacted poly pitch (CPP) will shrink to about 42nm, while the gate length (Lg) will be less than 12nm. “There is very limited space for the fill materials, so customers need a more conductive metal in a limited space. That is the major challenge,” Yoshida said in an interview at the IEDM.

Work Function Maintained

Naomi Yoshida: room for gate fill disappearing

The Applied R&D work showed that if the barrier layer for a tungsten fill is reduced too much, to a 2nm or 3nm TiN layer for example, the effective work function (eWF) degrades by as much as 500mV eWF and the total gate conductance suffers. With the CVD process used to deposit a tungsten RMG fill, there was “significant fluorine diffusion” into the work function metal layer in the case of a 2nm TiN barrier.

By contrast, the cobalt fill maintained the NMOS band-edge eWF with the same 2nm TiN barrier.

Gradually, cobalt will be adopted more widely for the contacts, interconnects, and RMG gate fill steps. “It is time to think about how to achieve more conductance in the gate material. Previously, people said there was a negligible contribution from the gate material, but now with the smaller gates at 5nm, gate fill metal makes a huge contribution to resistance, and barrier thickness reduction is important as well,” Yoshida said.

E-beam Inspection

Nicolas Breil: E-beam void inspection useful for cobalt contacts

Applied Materials also has developed an e-beam inspection solution, ProVision, first introduced in mid-2016, and has optimized it for inspecting cobalt voids. Nicolas Breil, a director in the company’s contact module division, said semiconductor R&D organizations are busy developing cobalt contact solutions, optimizing the deposition, CMP, and other steps. “For such a dense and critical level as the contact, it always needs very careful engineering. They key is to get results as fast as possible, but being fast can be very expensive.”

Amir Wachs, business development manager at Applied’s process diagnostics and control business unit in Rehovat, Israel, said the ProVision e-beam inspection system has a resolution of 1nm, at 10,000-20,000 locations per hour, taking a few hundred measurements on each field of view.

“Voids form when there are adhesion issues between the cobalt and TiN. One of the key issues is the correct engineering of the Ti nitride and PVD cobalt and CVD cobalt. To detect embedded voids requires a TEM inspection, but then customers get very limited statistics. There might be a billion contacts per chip, and with conventional TEM you might get to inspect two.”

The ProVision system speeds up the feedback loop between inspection and co-optimization. “Customers can assess the validity of the optimization. With other inspection methods, co-optimization might take five days to three weeks. With this type of analysis, using ProVision, customers can do tests early in the flow and validate their co-optimization within a few hours,” Wachs said.

Logic Densities Advance at IEDM 2017

Monday, December 18th, 2017

By Dave Lammers

The 63rd International Electron Devices Meeting brought an optimistic slant to transistor density scaling. While some critics have declared the death of Moore’s Law, there was little evidence of that — on the density front at least — at the IEDM, held Dec. 2-6 in San Francisco.

And an Intel engineering manager gave a presentation at IEDM that took a somewhat optimistic view of EUV lithography readiness, auguring further patterning improvements, starting with contacts and vias.

GlobalFoundries, which is skipping the 10nm node, presented its 7nm logic technology, expects to move into manufacturing in mid-2018. John Pellerin, vice president of global R&D, said the foundry has worked closely with its two lead customers, AMD and IBM, to define a high-performance-computing 7nm logic technology that achieves a 2.8X improvement of routed logic density compared with its 14nm technology.

Pellerin said the current 7nm process of record (POR) delivers “the right mix of performance, power, and area (PPA),” adding that GlobalFoundries plans to bring in EUV patterning at an undefined later point in the 7+ generation for further improvements.

Contact Over Active Gate

Chris Auth, director of advanced transistor development at Intel Corp., described a 10nm logic technology that sharply increased the transistor density compared with the 14nm generation, partly due to a contact-over-active-gate (COAG) architecture. The 10nm ring oscillator performance was improved by 20 percent compared with the comparable 14nm test vehicle.

Chris Auth, who presented Intel’s 10nm technology paper at IEDM, was surrounded by questioners following the presentation.

Auth said the COAG approach was a key contributor to Intel’s ability to increase its transistor density by 2.7 times over the company’s previous generation, to 100 million transistors per square millimeter of silicon. While the traditional approach puts the contact via over the isolation area, COAG places the contact via directly over the gate. Auth said the approach does require a second etch stop layer and other process complexities, but contributes “a sizable 10 percent reduction in area.” Elimination of the dummy gate for cell boundary isolation, and the use of cobalt at three layers (see related story), also contributed.

While there has been much hand wringing in the industry over the costs involved with multi-level patterning, Auth didn’t appear phased by it. Intel used a self-aligned quad patterning (SAQP) scheme to create fins with a tight pitch. The SAQP approach required two sacrificial layers, with lithography defining the first large pattern and four additional steps to remove the spacers and create the final lines and spaces.

The Intel 10nm fins are 46nm in height.

The SAQP approach starts by exposing a 130nm line, depositing the two spacers, halving the pattern to 68nm, and again to 34nm. “It is a grating and cut process similar to what we showed at 22nm, except it is SAQP instead of SADP,” using patterning to form a grating of fins, and cutting the ends of the fins with a cut mask.

“There were no additional lithography steps required. The result was fins that are tighter, straighter, and taller, with better drive current and matching” than Intel’s 14nm-generation fins, he said. Intel continued to use self-aligned double patterning (SADP) for M 2-5, and for gate patterning.

GlobalFoundries — which has been in production for 18 months with the 14nm process used by AMD, IBM, and others — plans to ramp its 7nm logic generation starting in mid-2018. The 7nm high-density SRAM cell measures .0269 um2, slightly smaller than TSMC’s published 7nm cell, while Intel reported a .0312 um2 cell size for its 10nm process.

Intel argues that the traditional way of calculating density improvements needs to be replaced with a metric that combines NAND and scan flip-flop densities. (Source: Intel)

GlobalFoundries chief technology officer Gary Patton said, “all of us are in the same zip code” when it comes to SRAM density. What is increasingly important is how the standard cells are designed to minimize the track height and thereby deliver the best logic cell technology to designers, Patton said.

EUV Availability Needs Improvements

Britt Turkot, senior principal engineer at Intel, discussed the readiness of EUV lithography at an IEDM session, giving a cautiously bullish report. With any multi-patterning solution for leading-edge silicon, including etch and CMP steps, placement error is the biggest challenge. With quad patterning, Turkot said multiple masks are involved, creating “compounded alignment errors.”

EUV has its own challenges, including significant secondary ions from the EUV photons. The key challenge for much of the decade, source power, seems to be partially resolved. “We are confident that the 250 Watts of source power needed for volume manufacturing will be ready once the field tools are upgraded,” she said.

Pellicles may be another challenge, with ASML expected to have a polysilicon-based pellicle ready in time for EUV production. However, she said a polysilicon membrane “does give quite a hit to the transmissivity” of the mask. “The transmissivity impact is quite significant,” she acknowledged during the Q&A period following her talk.

Intel has succeeded in repairing some mask defects, Turkot said, and implements pattern shifting so that other defects do not impinge on the patterned wafer.

Asked by a member of the audience about EUV availability or up-time, Turkot said “one day, availability can be great,” and less than good on other days, with “long unscheduled downs.” Intel is predicting 88 percent availability next year, she said in response to a question.

Pellicle Needed for Wiring Layers

Scotten Jones, president of semiconductor cost consultancy IC Knowledge (Boston), said companies may be able to get by without a pellicle for EUV patterning of contacts and via layers late next year. However, a pellicle will be needed for patterning the lower-level wiring layers, absorbing 10-15 percent of the photons and impacting EUV patterning throughput accordingly.

“Companies can do the contacts and vias without a pellicle, but doing the metal layers will required a pellicle and that means that a ton of work still needs to be done. And then at 5nm, the dose you need for the resist goes up dramatically,” Jones said, adding that while it will take some time for ASML to roll out the 250 W source, “they should be able to do it.”

GlobalFoundries will take possession of its second EUV scanner in December 2017, while Intel is believed to own four EUV systems.

Pellerin said GlobalFoundries defined the ground rules for its 7nm process so that the foundry can do a phased implementation of EUV without causing its customers “design discontinuity, bringing a benefit to design costs.”

John Pellerin, v.p. of R&D, said GlobalFoundries plans a phased implementation of EUV without “design discontinuity.”

The foundry will first do the hole levels and then move into the tight-pitch metal levels as mask defectivity improves. “The mask ecosystem needs to evolve,” Pellerin said.

Cost-per-Function on Track

In a keynote speech at IEDM, Lisa Su, the CEO of Advanced Micro Devices, said over the last 10 years the semiconductor industry has succeeded in doubling transistor density every 2-2.4 years. But she said the performance gains have been much smaller. “We are making progress, but it is taking a tremendous amount of work,” said Su, who received a best paper award at the IEDM 25 years earlier.

About 40 percent of the CPU performance improvement now comes from pure process technology, Su said, while the remainder comes from better microarchitectures, power management, and integration of system components such as an on-chip memory controller. While instructions per cycle are increasing at a 7 percent annual clip, Su said “the tricks have run out.”

Overall, the leading semiconductor companies seem to continue to make progress on transistor density. And costs per transistor may also be on track. Kaizad Mistry, co-director of logic technology development at Intel, contends that with its Intel’s 10nm process Intel’s per-transistor costs are actually better than the historical  curve.

Jones said the IC Knowledge cost analysis of TSMC’s processes indicates TSMC also is hewing to historical improvements on the per-transistor cost front. However, the foundries are catching up to Intel.

Intel Cadence Lagging

“What really strikes me is that Intel brought out its 45nm process in 2007, 32nm in 2009, and 22nm in 2011, but then it took three years to do 14nm. We are about to be in the year 2018, and Intel still doesn’t have its 10nm process done. It is a very nice process, but it is not out yet, and TSMC’s 7nm process is ramping right now. By the time Intel gets to 7nm, the foundries may be at 3nm. GlobalFoundries skipped a generation but is ramping its 7nm next year. All will have processes competitive to Intel at the same time, or even earlier,” Jones said.

While foundries such as GlobalFoundries, Samsung, and TSMC may be able to quickly offer advanced logic platforms, the wider semiconductor industry faces design cost challenges, Jones said. “Yes, the cost-per-transistor is going down, and that’s nice, but the cost of a design with finFETs is in the 100-million-dollar range. Intel can do it, but many smaller companies can’t afford to design with FinFETs.”

That is why both GlobalFoundries and Samsung are offering FD-SOI based platforms that use planar transistors, reducing design costs.

“The Internet of Things market is going to be nine million things, at relatively low volumes. IoT companies are finding it hard to justify the cost of a FinFET design, but with the cheaper design costs, SOI gives them an economical path,” Jones said.

Enabling the A.I. Era

Monday, October 23rd, 2017

thumbnail

By Pete Singer, Editor-in-Chief

There’s a strongly held belief now that the way in which semiconductors will be designed and manufactured in the future will be largely determined by a variety of rapidly growing applications, including artificial intelligence/deep learning, virtual and augmented reality, 5G, automotive, the IoT and many other uses, such as bioelectronics and drones.

The key question for most semiconductor manufacturers is how can the benefit from these trends? One of the goals of a recent panel assembled by Applied Materials for an investor day in New York was to answer that question.

Jay Kerley, Praful Krishna, Mukash Khare, Matt Johnson and Christos Georgiopoulos (left to right)

The panel, focused on “enabling the A.I. era,” was moderated by Sundeep Bajikar (former Sellside Analyst, ASIC Design Engineer). The panelists were: Christos Georgiopoulos (former Intel VP, professor), Matt Johnson (SVP in Automotive at NXP), Jay Kerley (CIO of Applied Materials), Mukesh Khare (VP of IBM Research) and Praful Krishna (CEO of Coseer). The panel discussion included three debates: the first one was “Data: Use or Discard”; the second was “Cloud versus Edge”; and the third was “Logic versus Memory.”

“There’s a consensus view that there will be an explosion of data generation across multiple new categories of devices,” said Bajikar, noting that the most important one is the self-driving car.  NXP’s Johnson responded that “when it comes to data generation, automotive is seeing amazing growth.” He noted the megatrends in this space: the autonomy, connectivity, the driver experience, and electrification of the vehicle. “These are changing automotive in huge ways. But if you look underneath that, AI is tied to all of these,” he said.

He said that estimates of data generation by the hour are somewhere from 25 gigabytes per hour on the low end, up to 250 gigabytes or more per hour on the high end. or even more in some estimates. “It’s going to be, by the second, the largest data generator that we’ve seen ever, and it’s really going to have a huge impact on all of us.”

Georgiopoulos agrees that there’s an enormous amount of infrastructure that’s getting built right now. “That infrastructure is consisting of both the ability to generate the data, but also the ability to process the data both on the edge as well as on the cloud,” he said. The good news is that sorting that data may be getting a little easier. “One of the more important things over the last four or five years has been the quality of the data that’s getting generated, which diminishes the need for extreme algorithmic development,” he said. “The better data we get, the more reasonable the AI neural networks can be and the simpler the AI networks can be for us to extract information that we need and turn the data information into dollars.”

Edge computing describes a computing topology in which information processing, and content collection and delivery, are placed closer to the sources of this information. Connectivity and latency challenges, bandwidth constraints and greater functionality embedded at the edge favors distributed models. Jay Kerley (CIO of Applied Materials) addressed the debate of cloud vs edge computing, noting it was a factor of data, then actual value and finally intelligence. “There’s no doubt that with the pervasiveness of the edge and billions of devices, data is going to be generated exponentially. But the true power comes in harnessing that data in the core. Taking it and turning it into actual intelligence. I believe that it’s going to happen in both places, and as a result of that, the edge is not going to only generate data, it’s going to have to consume data, and it’s going to have to make decisions. When you’re talking about problems around latency, maybe problems around security, problems around privacy, that can’t be overcome, the edge is going to have to be able to make decisions,” he said.

Kerley said there used to be a massive push to build data centers, but that’s changed. “You want to shorten the latency to the edge, so that data centers are being deployed in a very pervasive way,” he said. What’s also changing is that cloud providers have a huge opportunity to invest in the edge, to make the edge possible. “If they don’t, they are going to get cut out,” he added. “They’ve got to continue to invest to make access into the cloud as easy, and as frictionless as possible. At the end of the day, with all that data coming into these cloud data centers, the processing of that information, turning it into actual intelligence, turning it into value, is absolutely critical.”

Mukesh Khare (VP of IBM Research) also addressed the value of data. “We all believe that data is our next natural resource. We’re not going to discard it. You’re going to go and figure out how to generate value out of it,” he said.

Khare said that today, most artificial intelligence is too complex. It requires, training, building models and then doing inferencing using those models. “The reason there is good in artificial intelligence is because of the exponential increase in data, and cheap compute. But, keep in mind that, the compute that we are using right now is the old compute. That compute was built to do spreadsheet, databases, the traditional compute.

“Since that compute is cheap and available, we are making use of it. Even with the cheap and available compute in cloud, it takes months to generate those models. So right now, most of the training is still being done in cloud. Whereas, inferencing, making use from that model is done at the edge. However, going forward, it is not possible because the devices at the edge are continuously generating so much data that you cannot send all the data back to the cloud, generate models, and come back on the edge.”

“Eventually, a lot of training needs to move to the edge as well,” Khare said. This will require some innovation so that the compute, which is being done right now in cloud, can be transferred over to edge with low-power devices, cheap devices. Applied Materials’ Kerley added that innovation has to happen not only at the edge, but in the data center and at the network layer, as well as in the software frameworks. “Not only the AI frameworks, but what’s driving compression, de-duplication at the storage layer is absolutely critical as well,” he said.

NXP’s Johnson also weighed in on the edge vs cloud debate with the opinion that both will be required for automotive. “For automotive to do what it needs to, both need to evolve,” he said. “In the classic sense of automotive, the vehicle would be the edge, which needs access to the cloud frequently, or non-stop. I think it’s important to remember that the edge values efficiency. So, efficiency, power, performance and cost are all very important to make this happen,” he added.

Automotive security adds another degree of complexity. “If you think of something that’s always connected, and has the ability to make decisions and control itself, the security risk is very high. And it’s not just to the consumer of the vehicle, but also to the company itself that’s providing these vehicles. It’s actually foundational that the level of safety, security, reliability, that we put into these things is as good as it can be,” Johnson said.

Georgiopoulos said a new compute model is required for A.I. “It’s important to understand that the traditional workloads that we all knew and loved for the last forty years, don’t apply with A.I. They are completely new workloads that require very different type of capabilities from the machines that you build,” he said.  “With these new kind of workloads, you’re going to require not only new architectures, you’re going to require new system level design. And you’re going to require new capabilities like frameworks. He said TensorFlow, which is an open-source software library for machine intelligence originally developed by researchers and engineers working on the Google Brain Team, seems to be the biggest framework right now. “Google made it public for only one very good reason. The TPU that they have created runs TensorFlow better than any other hardware around. Well, guess what? If you write something on TensorFlow, you want to go to the Google backend to run it, because you know you’re going to get great results. These kind of architectures are getting created right now that we’re going to see a lot more of,” he said.

Georgiopoulos said this “architecture war” is by no means over. “There are no standardized ways by which you’re going to do things. There is no one language that everybody’s going to use for these things. It’s going to develop, and it’s going to develop over the next five years. Then we’ll figure out which architecture may be prevalent or not. But right now, it’s an open space,” he said.

IBM’s Khare, weighed in on how transistors and memory will need to evolve to meet the demands of new AI computer architectures, “For artificial intelligence in our world, we have to think very differently. This is an inflection, but this is the kind of inflection that world has not seen for last 60 years.” He said the world has gone from tabulating system era (1900 to 1940) to the programmable system era in 1950s, which we are still using. “We are entering the era of what we call cognitive computing, which we believe started in 2011, when IBM first demonstrated artificial intelligence through our Watson System, which played Jeopardy,” he said.

Khare said “we are still using the technology of programmable systems, such as logic, memory, the traditional way of thinking, and applying it to AI, because that’s the best we’ve got.”

AI needs more innovation at all levels, Khare said. “You have to think about systems level optimization, chip design level optimization, device level optimization, and eventually materials level optimization,” he said.  “The artificial workloads that are coming out are very different. They do not require the traditional way of thinking — they require the way the brain thinks. These are the brain inspired systems that will start to evolve.”

Khare believes analog compute might hold the answer. “Analog compute is where compute started many, many years ago. It was never adopted because the precision was not high enough, so there were a lot of errors. But the brain doesn’t think in 32 bits, our brain thinks analog, right? So we have to bring those technologies to the forefront,” he said. “In research at IBM we can see that there could be several orders of magnitude reduction in power, or improvement in efficiency that’s possible by introducing some of those concepts, which are more brain inspired.”

Lithographic Stochastic Limits on Resolution

Monday, April 3rd, 2017

thumbnail

By Ed Korczynski, Sr. Technical Editor

The physical and economic limits of Moore’s Law are being approached as the commercial IC fab industry continues reducing device features to the atomic-scale. Early signs of such limits are seen when attempting to pattern the smallest possible features using lithography. Stochastic variation in the composition of the photoresist as well as in the number of incident photons combine to destroy determinism for the smallest devices in R&D. The most advanced Extreme Ultra-Violet (EUV) exposure tools from ASML cannot avoid this problem without reducing throughputs, and thereby increasing the cost of manufacturing.

Since the beginning of IC manufacturing over 50 years ago, chip production has been based on deterministic control of fabrication (fab) processes. Variations within process parameters could be controlled with statistics to ensure that all transistors on a chip performed nearly identically. Design rules could be set based on assumed in-fab distributions of CD and misalignment between layers to determine the final performance of transistors.

As the IC fab industry has evolved from micron-scale to nanometer-scale device production, the control of lithographic patterning has evolved to be able to bend-light at 193nm wavelength using Off-Axis Illumination (OAI) of Optical-Proximity Correction (OPC) mask features as part of Reticle Enhancement Technology (RET) to be able to print <40nm half-pitch (HP) line arrays with good definition. The most advanced masks and 193nm-immersion (193i) steppers today are able to focus more photons into each cubic-nanometer of photoresist to improve the contrast between exposed and non-exposed regions in the areal image. To avoid escalating cost and complexity of multi-patterning with 193i, the industry needs Extreme Ultra-Violet Lithography (EUVL) technology.

Figure 1 shows Dr. Britt Turkot, who has been leading Intel’s integration of EUVL since 1996, reassuring a standing-room-only crowd during a 2017 SPIE Advanced Lithography (http://spie.org/conferences-and-exhibitions/advanced-lithography) keynote address that the availability for manufacturing of EUVL steppers has been steadily improving. The new tools are close to 80% available for manufacturing, but they may need to process fewer wafers per hour to ensure high yielding final chips.

Figure 1. Britt Turkot (Intel Corp.) gave a keynote presentation on "EUVL Readiness for High-Volume Manufacturing” during the 2017 SPIE Advanced Lithography conference. (Source: SPIE)

The KLA-Tencor Lithography Users Forum was held in San Jose on February 26 before the start of SPIE-AL; there, Turcot also provided a keynote address that mentioned the inherent stochastic issues associated with patterning 7nm-node device features. We must ensure zero defects within the 10 billion contacts needed in the most advanced ICs. Given 10 billion contacts it is statistically certain that some will be subject to 7-sigma fluctuations, and this leads to problems in controlling the limited number of EUV photons reaching the target area of a resist feature. The volume of resist material available to absorb EUV in a given area is reduced by the need to avoid pattern-collapse when aspect-ratios increase over 2:1; so 15nm half-pitch lines will generally be limited to just 30nm thick resist. “The current state of materials will not gate EUV,” said Turkot, “but we need better stochastics and control of shot-noise so that photoresist will not be a long-term limiter.”

TABLE:  EUVL stochastics due to scaled contact hole size. (Source: Intel Corp.)

CONTACT HOLE DIAMETER 24nm 16nm
INCIDENT EUV PHOTONS 4610 2050
# ABSORBED IN AREAL IMAGE 700 215

From the LithoGuru blog of gentleman scientist Chris Mack (http://www.lithoguru.com/scientist/essays/Tennants_Law.html):

One reason why smaller pixels are harder to control is the stochastic effects of exposure:  as you decrease the number of electrons (or photons) per pixel, the statistical uncertainty in the number of electrons or photons actually used goes up. The uncertainty produces line-width errors, most readily observed as line-width roughness (LWR). To combat the growing uncertainty in smaller pixels, a higher dose is required.

We define a “stochastic” or random process as a collection of random variables (https://en.wikipedia.org/wiki/Stochastic_process), and a Wiener process (https://en.wikipedia.org/wiki/Wiener_process) as a continuous-time stochastic process in honor of Norbert Wiener. Brownian motion and the thermally-driven diffusion of molecules exhibit such “random-walk” behavior. Stochastic phenomena in lithography include the following:

  • Photon count,
  • Photo-acid generator positions,
  • Photon absorption,
  • Photo-acid generation,
  • Polymer position and chain length,
  • Diffusion during post-exposure bake,
  • Dissolution/neutralization, and
  • Etching hard-mask.

Figure 2 shows the stochastics within EUVL start with direct photolysis and include ionization and scattering within a given discrete photoresist volume, as reported by Solid State Technology in 2010.

Figure 2. Discrete acid generation in an EUV resist is based on photolysis as well as ionization and electron scattering; stochastic variations of each must be considered in minimally scaled areal images. (Source: Solid State Technology)

Resist R&D

During SPIE-AL this year, ASML provided an overview of the state of the craft in EUV resist R&D. There has been steady resolution improvement over 10 years with Photo-sensitive Chemically-Amplified Resists (PCAR) from 45nm to 13nm HP; however, 13nm HP needed 58 mJ/cm2, and provided DoF of 99nm with 4.4nm LWR. The recent non-PCAR Metal-Oxide Resist (MOR) from Inpria has been shown to resolve 12nm HP with  4.7 LWR using 38 mJ/cm2, and increasing exposure to 70 mJ/cm2 has produced 10nm HP L/S patterns.

In the EUVL tool with variable pupil control, reducing the pupil fill increases the contrast such that 20nm diameter contact holes with 3nm Local Critical-Dimension Uniformity (LCDU) can be done. The challenge is to get LCDU to <2nm to meet the specification for future chips. ASML’s announced next-generation N.A. >0.5 EUVL stepper will use anamorphic mirrors and masks which will double the illumination intensity per cm2 compared to today’s 0.33 N.A. tools. This will inherently improve the stochastics, when eventually ready after 2020.

The newest generation EUVL steppers use a membrane between the wafer and the optics so that any resist out-gassing cannot contaminate the mirrors, and this allow a much wider range of materials to be used as resists. Regarding MOR, there are 3.5 times more absorbed photons and 8 times more electrons generated per photon compared to PCAR. Metal hard-masks (HM) and other under-layers create reflections that have a significant effect on the LWR, requiring tuning of the materials in resist stacks.

Default R&D hub of the world imec has been testing EUV resists from five different suppliers, targeting 20 mJ/cm2 sensitivity with 30nm thickness for PCAR and 18nm thickness for MOR. All suppliers were able to deliver the requested resolution of 16nm HP line/space (L/S) patterns, yet all resists showed LWR >5nm. In another experiment, the dose to size for imec’s “7nm-node” metal-2 (M2) vias with nominal pitch of 53nm was ~60mJ/cm2. All else equal, three times slower lithography costs three times as much per wafer pass.

—E.K.

Deep Learning Could Boost Yields, Increase Revenues

Thursday, March 23rd, 2017

thumbnail

By Dave Lammers, Contributing Editor

While it is still early days for deep-learning techniques, the semiconductor industry may benefit from the advances in neural networks, according to analysts and industry executives.

First, the design and manufacturing of advanced ICs can become more efficient by deploying neural networks trained to analyze data, though labelling and classifying that data remains a major challenge. Also, demand will be spurred by the inference engines used in smartphones, autos, drones, robots and other systems, while the processors needed to train neural networks will re-energize demand for high-performance systems.

Abel Brown, senior systems architect at Nvidia, said until the 2010-2012 time frame, neural networks “didn’t have enough data.” Then, a “big bang” occurred when computing power multiplied and very large labelled data sets grew at Amazon, Google, and elsewhere. The trifecta was complete with advances in neural network techniques for image, video, and real-time voice recognition, among others.

During the training process, Brown noted, neural networks “figure out the important parts of the data” and then “converge to a set of significant features and parameters.”

Chris Rowen, who recently started Cognite Ventures to advise deep-learning startups, said he is “becoming aware of a lot more interest from the EDA industry” in deep learning techniques, adding that “problems in manufacturing also are very suitable” to the approach.

Chris Rowen, Cognite Ventures

For the semiconductor industry, Rowen said, deep-learning techniques are akin to “a shiny new hammer” that companies are still trying to figure out how to put to good use. But since yield questions are so important, and the causes of defects are often so hard to pinpoint, deep learning is an attractive approach to semiconductor companies.

“When you have masses of data, and you know what the outcome is but have no clear idea of what the causality is, (deep learning) can bring a complex model of causality that is very hard to do with manual methods,” said Rowen, an IEEE fellow who earlier was the CEO of Tensilica Inc.

The magic of deep learning, Rowen said, is that the learning process is highly automated and “doesn’t require a fab expert to look at the particular defect patterns.”

“It really is a rather brute force, naïve method. You don’t really know what the constituent patterns are that lead to these particular failures. But if you have enough examples that relate inputs to outputs, to defects or to failures, then you can use deep learning.”

Juan Rey, senior director of engineering at Mentor Graphics, said Mentor engineers have started investigating deep-learning techniques which could improve models of the lithography process steps, a complex issue that Rey said “is an area where deep neural networks and machine learning seem to be able to help.”

Juan Rey, Mentor Graphics

In the lithography process “we need to create an approximate model of what needs to be analyzed. For example, for photolithography specifically, there is the transition between dark and clear areas, where the slope of intensity for that transition zone plays a very clear role in the physics of the problem being solved. The problem tends to be that the design, the exact formulation, cannot be used in every space, and we are limited by the computational resources. We need to rely on a few discrete measurements, perhaps a few tens of thousands, maybe more, but it still is a discrete data set, and we don’t know if that is enough to cover all the cases when we model the full chip,” he said.

“Where we see an opportunity for deep learning is to try to do an interpretation for that problem, given that an exhaustive analysis is impossible. Using these new types of algorithms, we may be able to move from a problem that is continuous to a problem with a discrete data set.”

Mentor seeks to cooperate with academia and with research consortia such as IMEC. “We want to find the right research projects to sponsor between our research teams and academic teams. We hope that we can get better results with these new types of algorithms, and in the longer term with the new hardware that is being developed,” Rey said.

Many companies are developing specialized processors to run machine-learning algorithms, including non-Von Neumann, asynchronous architectures, which could offer several orders of magnitude less power consumption. “We are paying a lot of attention to the research, and would like to use some of these chips to solve some of the problems that the industry has, problems that are not very well served right now,” Rey said.

While power savings can still be gained with synchronous architectures, Rey said brain-inspired projects such as Qualcomm’s Zeroth processor, or the use of memristors being developed at H-P Labs, may be able to deliver significant power savings. “These are all worth paying attention to. It is my feeling that different architectures may be needed to deal with unstructured data. Otherwise, total power consumption is going through the roof. For unstructured data, these types of problem can be dealt with much better with neuromorphic computers.”

The use of deep learning techniques is moving beyond the biggest players, such as Google, Amazon, and the like. Just as various system integrators package the open source modules of the Hadoop data base technology into a more-secure offering, several system integrators are offering workstations packaged with the appropriate deep-learning tools.

Deep learning has evolved to play a role in speech recognition used in Amazon’s Echo. Source: Amazon

Robert Stober, director of systems engineering at Bright Computing, bundles AI software and tools with hardware based on Nvidia or Intel processors. “Our mission statement is to deploy deep learning packages, infrastructure, and clusters, so there is no more digging around for weeks and weeks by your expensive data scientists,” Stober said.

Deep learning is driving new the need for new types of processors as well as high-speed interconnects. Tim Miller, senior vice president at One Stop Systems, said that training the neural networks used in deep learning is an ideal task for GPUs because they can perform parallel calculations, sharply reducing the training time. However, GPUs often are large and require cooling, which most systems are not equipped to handle.

David Kanter, principal consultant at Real World Technologies, said “as I look at what’s driving the industry, it’s about convolutional neural networks, and using general-purpose hardware to do this is not the most efficient thing.”

However, research efforts focused on using new materials or futuristic architectures may over-complicate the situation for data scientists outside of the research arena. At the International Electron Devices Meeting (IEDM 2017), several research managers discussed using spin torque magnetic (STT-MRAM) technology, or resistive RAMs (ReRAM), to create dense, power-efficient networks of artificial neurons.

While those efforts are worthwhile from a research standpoint, Kanter said “when proving a new technology, you want to minimize the situation, and if you change the software architecture of neural networks, that is asking a lot of programmers, to adopt a different programming method.”

While Nvidia, Intel, and others battle it out at the high end for the processors used in training the neural network, the inference engines which use the results of that training must be less expensive and consume far less power.

Kanter said “today, most inference processing is done on general-purpose CPUs. It does not require a GPU. Most people I know at Google do not use a GPU. Since the (inference processing) workload load looks like the processing of DSP algorithms, it can be done with special-purpose cores from Tensilica (now part of Cadence) or ARC (now part of Synopsys). That is way better than any GPU,” Kanter said.

Rowen was asked if the end-node inference engine will blossom into large volumes. “I would emphatically say, yes, powerful inference engines will be widely deployed” in markets such as imaging, voice processing, language recognition, and modeling.

“There will be some opportunity for stand-alone inference engines, but most IEs will be part of a larger system. Inference doesn’t necessarily need hundreds of square millimeters of silicon. But it will be a major sub-system, widely deployed in a range of SoC platforms,” Rowen said.

Kanter noted that Nvidia has a powerful inference engine processor that has gained traction in the early self-driving cars, and Google has developed an ASIC to process its Tensor deep learning software language.

In many other markets, what is needed are very low power consumption IEs that can be used in security cameras, voice processors, drones, and many other markets. Nvidia CEO Jen Hsung Huang, in a blog post early this year, said that deep learning will spur demand for billions of devices deployed in drones, portable instruments, intelligent cameras, and autonomous vehicles.

“Someday, billions of intelligent devices will take advantage of deep learning to perform seemingly intelligent tasks,” Huang wrote. He envisions a future in which drones will autonomously find an item in a warehouse, for example, while portable medical instruments will use artificial intelligence to diagnose blood samples on-site.

In the long run, that “billions” vision may be correct, Kanter said, adding that the Nvidia CEO, an adept promoter as well as an astute company leader, may be wearing his salesman hat a bit.

“Ten years from now, inference processing will be widespread, and many SoCs will have an inference accelerator on board,” Kanter said.

Vital Control in Fab Materials Supply-Chains – Part 2

Thursday, February 16th, 2017

By Ed Korczynski, Sr. Technical Editor

As detailed in Part 1 of this article published last month by SemiMD, the inaugural Critical Materials Council (CMC) Conference happened May 5-6 in Hillsboro, Oregon. Held just after the yearly private CMC meeting, the public CMC Conference provides a forum for the pre-competitive exchange of information to control the supply-chain of critical materials needed to run high-volume manufacturing (HVM) in IC fabs. The next CMC Conference will happen May 11-12 in Dallas, Texas.

At the end of the 2016 conference, a panel discussion moderated by Ed Korczynski was recorded and transcribed. The following is Part 2 of the conversation between the following industry experts:

  • Jean-Marc Girard, CTO and Director of R&D, Air Liquide Advanced Materials,
  • Jeff Hemphill, Staff Materials R&D Engineer, Intel Corporation,
  • Jonas Sundqvist, Sr. Scientist, Fraunhofer IKTS; and co-chair of ALD Conference, and
  • John Smythe, Distinguished Member of Technical Staff, Micron Technology.

FIGURE 1: 2016 CMC Conference expert panelists (from left to right) John Smyth, Jonas Sundqvist, Jeff Hemphill, and Jean-Marc Girard. (Source: TECHCET CA)

KORCZYNSKI:  We heard from David Thompson [EDITOR’S NOTE:  Director of Process Chemistry, Applied Materials presented on “Agony in New Material Introductions -  Minimizing and Correlating Variabilities”] today on what we must control, and he gave an example of a so-called trace-contaminant that was essential for the process performance of a precursor, where the trace compound helped prevent particles from flaking off chamber walls. Do we need to specify our contaminants?

GIRARD:  Yes. To David’s point this morning, every molecule is different. Some are very tolerant due to the molecular process associated with it, and some are not. I’ll give you an example of a cobalt material that’s been talked about, where it can be run in production at perhaps 95% in terms of assay, provided that one specific contaminant is less than a couple of parts-per-million. So it’s a combination of both, it’s not assay OR a specification of impurities. It’s a matter of specifying the trace components that really matter when you reach the point that the data you gather gives you that understanding, and obviously an assay within control limits.

HEMPHILL:  Talking about whether we’re over-specifying or not, the emphasis is not about putting the right number on known parameters like assay that are obvious to measure, the emphasis is on identifying and understanding what makes up the rest of it and in a sense trying over-specify that. You identify through mass-spectrometry and other techniques that some fraction of a percent is primarily say five different species, it’s finding out how to individually monitor and track and control those as separate parameters. So from a specification point of view what we want is not necessarily the lowest possible numbers, but it’s expanding how many things we’re looking at so that we’re capturing everything that’s there.

KORCZYNSKI:  Is that something that you’re starting to push out to your suppliers?

HEMPHILL:  Yes. It depends on the application we’re talking about, but we go into it with the assumption that just assay will not be enough. Whether a single molecule or a blend of things is supposed to be there, we know that just having those be controlled by specification will not be sufficient. We go under the assumption that we are going to identify what makes up the remaining part of the profile, and those components are going to need to be controlled as well.

KORCZYNSKI:  Is that something that has changed by node? Back when things were simpler say at 45nm and larger, were these aspects of processing that we could safely ignore as ‘noise’ but are now important ‘signals’?

HEMPHILL:  Yes, we certainly didn’t pay as close attention just a couple of generations ago.

KORCZYNSKI:  That seems to lead us to questions about single-sources versus dual-sourcing. There are many good reasons to do both, but not simultaneously. However, it seems that because of all of the challenges we’re heard about over the last day-and-a-half of this conference it creates greater burden on the suppliers, and for critical materials the fabs are moving toward more single-sourcing over time.

SMYTHE:  I think that it comes down to more of a concern over geographic risk. I’ll buy from one entity if that entity has more than one geographic location for the supply, so that I’m not exposed to a single ‘Act of God’ or a ‘random statistical occurrence of global warming.’ So for example I  need to ask if a supplier has a place in the US and a place in France that makes the same thing, so that if something bad happens in one location it can still be sourced? Or do you have an alternate-supply agreement that if you can’t supply it you have an agreement with Company-X to supply it so that you still have control? You can’t come to a Micron and say we want to make sure that we get at minimum 25% no matter what, because what typically happens with second-sourcing is Company-A gets 75% of the business while Company-B gets 25%. There are a lot of reasons that that doesn’t work so well, so people may have an impression that there’s a movement toward single-source but it’s ‘single flexible-source.’

HEMPHILL:  There are a lot of benefits of dual- or multiple-sourcing. The commercial benefits of competition can be positive and we’re for it when it works. The risk is that as things are progressing and we’re getting more sensitive to differences in materials it’s getting harder to maintain that. We have seen situations where historically we were successful with dual-sourcing a raw material coming from two different suppliers or even a single supplier using two different manufacturing lines and everything was fine and qualified and we could alternate sources invisibly. However, as our sensitivity has grown over time we can start to detect differences.

So the concept of being ‘copy-exactly’ that we use in our factories, we really need production lines to do that, and if we’re talking about two different companies producing the same material then we’re not going to get them to be copy-exactly. When that results in enough of a variation in the material that we can detect it in the factory then we cannot rely upon two sources. Our preference would be one company that maintains multiple production sites that are designed to be exactly the same, then we have a high degree of confidence that they will be able to produce the same material.

FIGURE 2: Jean-Marc Girard, Distinguished Member of Technical Staff of Micron Technology, provided the supplier perspective. (Source: SEMI)

GIRARD:  I can give you a supplier perspective on that. We are seeing very different policies from different customers, to the point that we’re seeing an increase in the number of customers doing single-sourcing with us, provided we can show the ability to maintain business continuity in case of a problem. I think that the industry became mature after the tragic earthquake and tsunami in Japan in 2011 with greater understanding of what business continuity means. We have the same discussions with our own suppliers, who may say that they have a dedicated reactor for a certain product with another backup reactor with a certain capacity on the same site, and we ask what happens if the plant goes on strike or there’s a fire there?

A situation where you might think the supply was stable involved silane in the United States. There are two large silane plants in the United States that are very far apart from each other and many Asian manufacturers dependent upon them. When the U.S. harbors went on strike for a long time there was no way that material could ship out of the U.S. customers. So, yes there were two plants but in such an event you wouldn’t have global supply. So there is no one way to manage our supply lines and we need to have conversations with our customers to discuss the risks. How much time would it take to rebuild a supply-chain source with someone else? If you can get that sort of constructive discussion going then customers are usually open to single-sourcing. One regional aspect is that Asian customers tend to favor dual-sourcing more, but that can lead to IP problems.

[DISCLOSURE:  Ed Korczynski is co-chair of the CMC Conference, and Marketing Director of TECHCET CA the advisory services firm that administers the Critical Materials Council (CMC).]

—E.K.

Innovations at 7nm to Keep Moore’s Law Alive

Thursday, January 19th, 2017

thumbnail

By Dave Lammers, Contributing Editor

Despite fears that Moore’s Law improvements are imperiled, the innovations set to come in at the 7nm node this year and next may disprove the naysayers. EUV lithography is likely to gain a toehold at the 7nm node, competing with multi-patterning and, if all goes well, shortening manufacturing cycles. Cobalt may replace tungsten in an effort to reduce resistance-induced delays at the contacts, a major challenge with finFET transistors, experts said.

While the industry did see a slowdown in Moore’s Law cost reductions when double patterning became necessary several years ago, Scotten Jones, who runs a semiconductor consultancy focused on cost analysis, said Intel and the leading foundries are back on track in terms of node-to-node cost improvements.

Speaking at the recent SEMI Industry Strategy Symposium (ISS), Jones said his cost modeling backs up claims made by Intel, GlobalFoundries, and others that their leading-edge processes deliver on die costs. Cost improvements stalled at TSMC for the16nm node due to multi-patterning, Jones said. “That pause at TSMC fooled a lot of people. The reality now may surprise those people who said Moore’s Law was dead. I don’t believe that, and many technologists don’t believe that either,” he said.

As Intel has adopted a roughly 2.5-year cadence for its more-aggressive node scaling, Jones said “the foundries are now neck and neck with Intel on density.” Intel has reached best-ever yield levels with its finFET-based process nodes, and the foundries also report reaching similar yield levels for their FinFET processes. “It is hard, working up the learning curve, but these companies have shown we can get there,” he said.

IC Knowledge cost models show the chip industry is succeeding in scaling density and costs. (Source: Scotten Jones presentation at 2017 SEMI ISS)

TSMC, spurred by its contract with Apple to supply the main iPhone processors, is expected to be first to ship its 7nm products late this year, though its design rules (contacted poly pitch and minimum metal pitch) are somewhat close to Intel’s 10nm node.

While TSMC and GlobalFoundries are expected to start 7nm production using double and quadruple patterning, they may bring in EUV lithography later. TSMC has said publicly it plans to exercise EUV in parallel with 193i manufacturing for the 7nm node. Samsung has put its stake in the ground to use EUV rather than quadruple patterning in 2018 for critical layers of its 7nm process. Jones, president of IC Knowledge LLC, said Intel will have the most aggressive CPP and MPP pitches for its 7nm technology, and is likely to use EUV in 2019-2020 to push its metal pitches to the minimum possible with EUV scanners.

EUV progress at imec

In an interview at the 62nd International Electron Devices Meeting (IEDM) in San Francisco in early December, An Steegen, the senior vice president of process technology at Imec (Leuven, Belgium), said Imec researchers are using an ASML NXE 3300B scanner with 0.3 NA optics and an 80-Watt power supply to pattern about 50 wafers per hour.

“The stability on the tool, the up time, has improved quite a lot, to 55 percent. In the best weeks we go well above 70 percent. That is where we are at today. The next step is a 125-Watt power supply, which should start rolling out in the field, and then 250 Watts.”

Steegen said progress is being made in metal-containing EUV resists, and in development of pellicles “which can withstand hydrogen in the chamber.”

If those challenges can be met, EUV would enable single patterning for vias and several metal layers in the middle of the line (MOL), using cut masks to print the metal line ends. “For six or seven thin wires and vias, at the full (7nm node) 32nm pitch, you can do it with a single exposure by going to EUV. The capability is there,” Steegen said.

TSMC’s 7nm development manager, S.Y. Wu, speaking at IEDM, said quadruple patterning and etch (4P4E) will be required for critical layers until EUV reaches sufficient maturity. “EUV is under development (at TSMC), and we will use 7nm as the test vehicle.”

Huiming Bu was peppered with questions following a presentation of the IBM Alliance 7nm technology at IEDM.

Huiming Bu, who presented the IBM Alliance 7nm paper at IEDM, said “EUV delivers significant depth of field (DoF) improvement” compared with the self-aligned quadruple (SAQP) required for the metal lines with immersion scanners.

A main advantage for EUV compared with multi-patterning is that designs would spend fewer days in the fabs. Speaking at ISS, Gary Patton, the chief technology officer at GlobalFoundries, said EUV could result in 30-day reductions in fab cycle times, compared with multiple patterning with 193nm immersion scanners, based on 1.5 days of cycle time per mask layer.

Moreover, EUV patterns would produce less variation in electrical performance and enable tighter process parameters, Patton said.

Since designers have become accustomed to using several colors to identify multi-patterning layers for the 14nm node, the use of double and quadruple patterning at the 7nm node would not present extraordinary design challenges. Moving from multi-patterning to EUV will be largely transparent to design teams as foundries move from multi-patterning to EUV for critical layers.

Interconnect resistance challenges

As interconnects scale and become more narrow, signals can slow down as electrons get caught up in the metal grain boundaries. Jones estimates that as much as 85 percent of parasitic capacitance is in the contacts.

For the main interconnects, nearly two decades ago, the industry began a switch from aluminum to copper. Tungsten has been used for the contacts, vias, and other metal lines near the transistor, partly out of concerns that copper atoms would “poison” the nearby transistors.

Tungsten worked well, partly because the bi-level liner – tantalum nitride at the interface with the inter-level dielectric (ILD) and tantalum at the metal lines – was successful at protecting against electromigration. The TaN-Ta liner is needed because the fluorine-based CVD processes can attack the silicon. For tungsten contacts, Ti serves to getter oxygen, and TiN – which has high resistance — serves as an oxygen and fluorine barrier.

However, as contacts and MOL lines shrunk, the thickness of the liner began to equal the tungsten metal thicknesses.

Dan Edelstein, an IBM fellow who led development of IBM’s industry-leading copper interconnect process, said a “pinch point” has developed for FinFETs at the point where contacts meet the middle-of-the-line (MOL) interconnects.

“With cobalt, there is no fluorine in the deposition process. There is a little bit of barrier, which can be either electroplated or deposited by CVD, and which can be polished by CMP. Cobalt is fairly inert; it is a known fab-friendly metal,” Edelstein said, due to its longstanding use as a silicide material.

As the industry evaluated cobalt, Edelstein said researchers have found that cobalt “doesn’t present a risk to the device. People have been dropping it in, and while there are still some bugs that need to be worked out, it is not that hard to do. And it gives a big change in performance,” he said.

Annealing advantages to Cobalt

Contacts are a “pinch point” and the industry may switch to cobalt (Source: Applied Materials)

An Applied Materials senior director, Mike Chudzik, writing on the company’s blog, said the annealing step during contact formation also favors cobalt: “It’s not just the deposition step for the bulk fill involved – there is annealing as well. Co has a higher thermal budget making it possible to anneal, which provides a superior, less granular fill with no seams and thus lowers overall resistance and improves yield,” Chudzik explained.

Increasing the volume of material in the contact and getting more current through is critical at the 7nm node. “Pretty much every chipmaker is working aggressively to alleviate this issue. They understand if it’s not resolved then it won’t matter what else is done with the device to try and boost performance,” Chudzik said.

Prof. Koike strikes again

Innovations underway at a Japanese university aim to provide a liner between the cobalt contact fill material and the adjacent materials. At a Sunday short course preceding the IEDM, Reza Arghavani of Lam Research said that by creating an alloy of cobalt and approximately 10 percent titanium, “magical things happen” at the interfaces for the contact, M0 and M1 layers.

The idea for adding titanium arose from Prof. Junichi Koike at Tohoku University, the materials scientist who earlier developed a manganese-copper solution for improved copper interconnects. For contacts and MOL, the Co-Ti liner prevents diffusion into the spacer oxide, Arghavani said. “There is no (resistance) penalty for the liner, and it is thermally stable, up to 400 to 500 degrees C. It is a very promising material, and we are working on it. W (tungsten) is being pushed as far as it can go, but cobalt is being actively pursued,” he said.

Stressor changes ahead

Presentations at the 2016 IEDM by the IBM Alliance (IBM, GlobalFoundries, and Samsung) described the use of a stress relaxed buffer (SRB) layer to induce stress, but that technique requires solutions for the defects introduced in the silicon layer above it. As a result of that learning process, SRB stress techniques may not come into the industry until the 5 nm node, or a second-generation 7nm node.

Technology analyst Dick James, based in Ottawa, said over the past decade companies have pushed silicon-germanium stressors for the PFET transistors about as far as practical.

“The stress mechanisms have changed since Intel started using SiGe at the 90nm node. Now, companies are a bit mysterious, and nobody is saying what they are doing. They can’t do tensile nitride anymore at the NFET; there is precious little room to put linear stress into the channel,” he said.

The SRB technique, James said, is “viable, but it depends on controlling the defects.” He noted that Samsung researchers presented work on defects at the IEDM in December. “That was clearly a research paper, and adding an SRB in production volumes is different than doing it in an R&D lab.”

James noted that scaling by itself helps maintain stress levels, even as the space for the stressor atoms becomes smaller. “If companies shorten the gate length and keep the same stress as before, the stress per nanometer at least maintains itself.”

Huiming Bu, the IBM researcher, was optimistic, saying that the IBM Alliance work succeeded at adding both compressive and tensile strain. The SRB/SSRW approach used by the IBM Alliance was “able to preserve a majority – 75 percent – of the stress on the substrate.”

Jones, the IC Knowledge analyst, said another area of intense interest in research is high-mobility channels, including the use of SiGe channel materials in the PMOS FinFETS.

He also noted that for the NMOS finFETs, “introducing tensile stress in fins is very challenging, with lots of integration issues.” Jones said using an SRB layer is a promising path, but added: “My point here is: Will it be implemented at 7 nm? My guess is no.”

Putting it in a package

Steegen said innovation is increasingly being done by the system vendors, as they figure out how to combine different ICs in new types of packages that improve overall performance.

System companies, faced with rising costs for leading-edge silicon, are figuring out “how to add functionality, by using packaging, SOC partitioning and then putting them together in the package to deliver the logic, cache, and IOs with the right tradeoffs,” she said.

Process Control Deals with Big Data, Busy Engineers

Tuesday, November 22nd, 2016

thumbnail

By Dave Lammers, Contributing Editor

Turning data into insights that will improve fab productivity is one of the semiconductor industry’s biggest opportunities, one that experts say requires a delicate mix between automation and human expertise.

A year ago, after the 2015 Advanced Process Control (APC) conference in Austin, attendees said one of their challenges was that it takes too long to create the fault detection and classification (FDC) models that alert engineers when something is amiss in a process step.

“The industry listened,” said Brad van Eck, APC conference co-chairman. Participants at the 2016 APC in Phoenix heard progress reports from device makers as diverse as Intel, Qorvo, Seagate, and TSMC, as well as from key APC software vendors including Applied Materials, Bistel, and others.

Steve Chadwick, principal engineer for manufacturing IT at Intel, described the challenge in a keynote address. IC manufacturers which have spent billions of dollars on semiconductor equipment are seeking new ways to maximize their investments.

Steve Chadwick

“We all want to increase our quality, make the product in the best time, get the most good die out, and all of that. Time to market can be a game changer. That is universal to the manufacturing space,” Chadwick said.

“Every time we have a new generation of processor, we double the data size. Roughly a gigabyte of information is collected on every wafer, and we sort thousands of wafers a day,” Chadwick said. The result is petabytes of data which needs to be stored, analyzed, and turned into actionable “wisdom.”

Intel has invested in data centers located close their factories, making sure they have the processing power to handle data coming in from roughly 5 billion sensor data points collected each day at a single Intel factory.

“We have to take all of this raw data that we have in a data store and apply some kind of business logic to it. We boil it down to ‘wisdom,’ telling someone something they didn’t know beforehand.”

In a sense, technology is catching up, as Hadoop and several other data search engines are adopted to big data. Also, faster processors allow servers to analyze problems in 15 seconds or less, compared to several hours a few years ago.

Where all of this gets interesting is in figuring out how to relate to busy engineers who don’t want to be bothered with problems that don’t directly concern them. Chadwick detailed the notification problem at Intel fabs, particularly as engineers use smart phones and tablets to receive alarms. “Engineers are busy, and so you only tell them something they need to know. Sometimes engineers will say, ‘Hey, Steve, you just notified my phone of 500 things that I can’t do anything about. Can you cut it out?’”

Notification must be prioritized, and the best option in many cases is to avoid notifiying a person at all, instead sending a notification to an expert system. If that is not an option, the notification has to be tailored to the device the engineer is using. Intel is moving quickly to HTML 5-based data due largely to its portability across multiple devices, he added.

With more than half a million ad hoc jobs per week, Intel’s approach is to keep data and analysis close to the factory, processing whenever possible in the local geography. Instead of shipping data to a distant data center for analysis, the normal procedure is to ship the small analysis code to a very large data set.

False positives decried

Fault detection and classification (FDC) models are difficult to create and oftentimes overly sensitive, resulting in false alarms. These widely used, manually created FDC models can take two weeks or longer to set up. While they take advantage of subject-matter-expert (SME) knowledge and are easy to understand, tool limits tend to be costly to set up and manage, with a high level of false positives and missed alarms.

An Applied Materials presentation — by Parris Hawkins, James Moyne, Jimmy Iskandar, Brad Schulze, and Mike Armacost – detailed work that Applied is doing in cooperation with process control researchers at the University of Cincinnati. The goal is to develop next-generation FDC that leverages Big Data, prediction analytics, and expert engineers to combine automated model development with inputs from human experts.

Fully automated solutions are plagued with significant false positives/negatives, and are “generally not very useful,” said Hawkins. By incorporating metrology and equipment health data, a form of “supervised” model creation can result in more accurate process controls, he said.

The model creation effort first determines which sensors and trace features are relevant, and then optimizes the tool limits and other parameters. The goal is to find the optimum between too-wide limits that fail to alert when faults are existent, and overly tight limits which set off false alarms too often.

Next-generation FDC would leverage Big Data and human expertise. (Source: Applied Materials presentation at APC 2016).

Full-trace FDC

BISTel has developed an approach called Dynamic Full Trace FDC. Tom Ho, president of BISTel USA, presented the work in conjunction with Qorvo engineers, where a beta version of the software is being used.

Tom Ho

Ho said Dynamic Full Trace FDC starts with the notion that the key to manufacturing is repeatability, and in a stable manufacturing environment “anything that differs, isn’t routine, it is an indication of a mis-process and should not be repeatable. Taking that concept, then why not compare a wafer to everything that is supposed to repeat. Based on that, in an individual wafer process, the neighboring wafer becomes the model.”

The full-trace FDC approach has a limited objective: to make an assessment whether the process is good or bad. It doesn’t recommend adjustments, as a run-to-run tool might.

The amount of data involved is small, because it is confined to that unique process recipe. And because the neighboring trace is the model, there is no need for the time-consuming model creation mentioned so often at APC 2016. Compute power can be limited to a personal computer for an individual tool.

Ho took the example of an etch process that might have five recipe steps, starting with pumping down the chamber to the end point where the plasma is turned off. Dynamic full-trace FDC assumes that most wafers will receive a good etch process, and it monitors the full trace to cover the entire process.

“There is no need for a model, because the model is your neighboring trace,” he said. “It definitely saves money in multiple ways. With the rollout of traditional FDC, each tool type can take a few weeks to set up the model and make sure it is running correctly. For multiple tool types that can take a few months. And model maintenance is another big job,” he said.

For the most part, the dynamic full-trace software runs on top of the Bistel FDC platform, though it could be used with another FDC vendor “if the customer has access to the raw trace data,” he said.

Next Page »

Extension Media websites place cookies on your device to give you the best user experience. By using our websites, you agree to placement of these cookies and to our Privacy Policy. Please click here to accept.