Tag Archives: Top Story Left

May 22, 2012 — Brown University is using supercomputer simulations to better understand the ion bombardment of metal surfaces used in manufacturing nanoelectronics.

The research shows what takes place during ion bombardment in trillionths of a second. It could lead to better prediction and more uses of ion bombardment to make nanoscale structures.

Figure. Three new mechanisms at the nanoscale. A computer-model image of an island of metal atoms formed after bombardment by noble gas ions. Atoms disturbed by the bombardment cluster together under the surface and then glide back up in a matter of 2.1ps. SOURCE: Kim Lab/Brown University.

Ion bombardment is used in semiconductor and optoelectronics manufacturing, but has been limited by the lack of understanding of the underlying physics. Brown University engineers modeled noble gas ion bombardments to provide insights into how ion bombardment works, with the hopes of better predicting what surface patterns and stresses would result.

The research applies to “FCC” metals — copper, silver, gold, nickel, aluminum — crystals made up of cubic arrangements of atoms with one at each corner and one in each cube-face center.

This research builds on previous attempts to model ion bombardment on computers, by modeling more than one bombardment event or isolated point defects in the metal substrate. This work investigated collective behavior of defects during ion bombardments in terms of ion-substrate combinations, said Kyung-Suk Kim, professor of engineering at Brown.

The new model revealed how ion bombardments can set three main mechanisms — “dual layer formation,” “subway-glide mode growth,” and “adatom island eruption” — into motion in a matter of picoseconds. The mechanisms are a consequence of how the incoming ions melt the metal and then how it re-solidifies with the ions occasionally trapped inside.

When ions hit the metal surface, they penetrate it, knocking away nearby atoms in a process that is akin, at the atomic level, to melting. But rather than merely rolling away, the atoms resolidify in a different order.

Some atoms have been shifted out of place. There are some vacancies in the crystal nearer to the surface, and the atoms there pull together across the empty space, that creates a layer with more tension. Beneath that is a layer with more atoms that have been knocked into it. That crowding of atoms creates compression. Hence there are now two layers with different levels of compression and tension. This “dual layer formation” is the precursor to the “subway-glide mode growth” and “adatom island eruption”.

Materials that have been bombarded with ions sometimes produce a pattern of material that seems to have popped up out of the original surface. Scientists theorized that displaced atoms would individually bob back up to the surface, but the team’s models shows that these molecular islands are formed by whole clusters of displaced atoms that bond together and appear to glide back up to the surface, like passengers converging to emerge from a subway car.

Kyung-Suk Kim noted that “predictive design capability for controlling the surface patterns and stresses in nanotechnology products” could lead to flexible electronics, biocompatible surface formation on medical devices, and other new technologies.

“As a next step, I will develop prediction models for nanopattern evolution during ion bombardment which can guide the nanomanufacturing processes,” said Sang-Pil Kim, postdoctoral scholar. “This research will also be expanded to other applications such as soft- or hard-materials under extreme conditions.”

The research will be published May 23 in the Proceedings of the Royal Society A. Sang-Pil Kim is lead author, with Kyung-Suk Kim, Huck Beng Chew, Eric Chason, and Vivek Shenoy.

The research was funded by the Korea Institute of Science and Technology, the U.S. National Science Foundation, and the U.S. Department of Energy. The work used the Extreme Science and Engineering Discovery Environment (XSEDE), which is supported by National Science Foundation grant number OCI-1053575.

Visit the Semiconductors Channel of Solid State Technology!

May 21, 2012 — Stefan Wurm, director of lithography, SEMATECH will present “EUV Lithography Manufacturing Introduction: Infrastructure Readiness” in the session Technology Trends in Semiconductor Manufacturing at The ConFab 2012, June 3-6 in Las Vegas.

Wurm will be joined by An Steegan, IMEC; Thomas Jefferson, G450 Consortium; and Naoya Hayashi, Dai Nippon Printing Co. Ltd.

Extreme ultraviolet lithography (EUVL) has been in development for about a quarter century and is now getting ready for prime time with the first production tools expected to ship by year end 2012 and high volume manufacturing (HVM) ramp-up expected to start in 2014. The EUVL infrastructure to support the ongoing pilot line introduction is largely in place but productivity and cost remain a challenge for the HVM ramp-up. There is still a significant EUV source performance gap that must be closed. Resist materials to support the 22nm half-pitch (hp) requirements are available with acceptable  sensitivity and with some help from post processing also can meet the line width roughness requirements for memory products; contact hole resist performance still needs to improve.

For EUV masks, the main technical challenges remain the availability of defect free masks and of critical mask tools such as for defect review and inspection. EUV blank defects have been reduced to a level that can support memory products but further improvements are needed to meet logic defect requirements. Ramping up mask blank volume at yield to support the expected mask demand in 2014/15 is a challenge the industry must address for EUVL HVM introduction to be successful.

Dr. Stefan Wurm is director of lithography at SEMATECH, where he works as a GLOBALFOUNDRIES assignee.  He has more than 20 years of industry and R&D experience and has held technical and management positions in research and development at AMD, Siemens Semiconductor Group, Infineon, Qimonda, and SEMATECH. He was first assigned to SEMATECH in the late 1990s to the International 300 mm Initiative (I300I), where he was responsible for 300 mm metrology tool equipment demonstrations.

Prior to his previous assignment as associate director of lithography at SEMATECH, he served four years as SEMATECH’s extreme ultraviolet (EUV) program manager, where he has been instrumental in shaping and directing the SEMATECH EUV program which provides worldwide EUV infrastructure capabilities and technology learning to SEMATECH members. Dr. Wurm holds a doctorate in natural sciences from the Technische Universität München, Germany.

Visit the Semiconductors Channel of Solid State Technology!

May 21, 2012 — The massive “gray market” for cellphones is propelling sales of lower-cost flash memories like embedded multimedia card (eMMC) and serial peripheral interface (SPI) NOR, as manufacturers of the unregulated phones strive to keep production costs low, says IHS.

Approximately 210 million gray-market cellphones will ship in 2012, or 13% of the global mobile handset business, according to an IHS iSuppli Mobile & Embedded Memory Market Brief. While this is down from the peak year of 2011, when gray market totaled 250 million units, shipments are set to remain high, amounting to 189 million units, or 11 percent of the global cellphone shipments in 2013, as presented in the figure below. The gray market business is dramatically larger than it was just three years ago in 2009, when shipments amounted to 145 million units.

Figure. Worldwide gray handset shipment forecast (Millions of units). SOURCE: IHS iSuppli Research, May 20112.

“Gray-market phones represent a huge segment of the overall mobile handset industry,” said Michael Yang, senior principal analyst for memory & storage research at IHS. “For manufacturers of these phones, keeping up with consumer trends while maintaining low costs will continue to be a concern as they fight to remain viable in a fiercely competitive market. As a result, the adoption of inexpensive flash memories like eMMC and SPI NOR have been a key trend in the segment for gray-market handsets.”

Gray-market handsets are cell phones manufactured in China that are not recognized or licensed by government regulators. Makers of these products generally do not pay China’s value-added taxes and, therefore, profit illegally from their participation in the market.

The gray market for mobile handsets thrives mostly in developing countries where a large underground economy exists and where people are keen to obtain phones at low prices bearing desired functionalities—often those copied from authentic handsets made by name brands.

In China, for instance, gray-market handsets include counterfeits as well as so-called white-box units in which any logo can be slapped on a ready device. Frequently these white-box phones also feature smuggled chips, lack official certification and use fake international mobile equipment identity (IMEI) codes otherwise used to distinguish genuine handsets.

Asia, including China, remained the top opportunity, accounting for 62 percent of the gray market in 2011. Growth sectors like Indonesia and Vietnam are also moving beyond low-cost gray market handsets into gray-market feature phones, which are considered a step above low-cost handsets because of slightly more sophisticated functionalities.

The next two growth segments for the gray market are Latin America, which skews more toward low-end smartphones, as well as the Middle East/Africa (MEA) region, where entire new populations are being exposed to mobile devices for the first time.

As a whole the gray market for handsets is also steadily moving upstream, with smartphones rising from 1.4 percent of gray-market handset shipments in 2011 to 6.0 percent this year. Feature phones, however, will continue to command the lion’s share of gray-market handset shipments this year, at well over 60 percent.

eMMC delivers the high-density storage needed by the latest smartphone operating systems, at the same time also providing time-to-market advantages and less design effort. For its part, SPI NOR can claim cost advantages over traditional parallel NOR as it further reduces the total bill of materials for gray-market handsets.

Overall, the gray market for handsets is expected to retain an important presence in several lucrative geographies that device manufacturers are increasingly cultivating.

Read More in Gray Market Handsets: Flash Memory for the Masses

Visit the Semiconductors Channel of Solid State Technology!

May 18, 2012 — At The ConFab, an invitation-only semiconductor industry event June 3-6 in Las Vegas, TSMC’s BJ Woo, senior director, Graphic/PLD/CPU Business Development Division, will present “Bridging the Fabless-Foundry Gap.”

Woo will speak in Session #3: The Foundry-Fabless Supply Chain, alongside Nick Yu, VP of technology development, Qualcomm; Mike Noonen, senior VP, worldwide sales & marketing, GLOBALFOUNDRIES; and Xin Wu, senior director, silicon technology, Xilinx Inc.

“Bridging the gap between fabless design houses and foundry is the essential factor of the winning strategy,” Woo says. Fabless design houses and foundries need to collaborate closely to deliver successful semiconductor products and to prosper together. She will specify the requirements and challenges for chip designers as the complexity of technology grows in advanced nodes. She will also present solutions to address those challenges, showing how fables companies and foundry can collaborate and succeed.

For fabless design houses, major concerns include competitive technology selection with best performance/watt/cost, first silicon success to enable fast time to market and stable/reliable supply support with decent yield and ample capacity to secure market segment share.

Foundries will provide multiple process options with customization to address the unique requirements for different product applications. Meanwhile, accurate SPICE model, design for manufacturing (DFM), and intellectual property (IP) validation will help improve first silicon success rate and time to market. 

Stable/ reliable supply is one of the keys for fabless companies to gain and secure market segment share. As such, foundry businesses need to provide a more sophisticated manufacturing infrastructure to go after the best manufacturing excellence. Such infrastructure includes fully automated manufacturing systems, Giga-fabs for large scales of economy, speedy productization from tapeout to volume production, steep yield learning curve, nano-scale precision control for yields and technology scaling, product-grade enhancement, and integrated service (mask, wafer, and backend) for coherent quality and delivery.

Fabless and foundry companies need to collaborate closely in design, technology, and manufacturing areas to deliver a successful product with best performance/watt/cost and, in turn, to compete and to win in the marketplace. In addition, sufficient and reliable capacity support with manufacturing excellence will grow customers, growing business.

BJ joined TSMC as Senior Director of Technology Roadmap Division in April 2009.  She is currently in charge of the High Performance Technology definition for Graphics, PLD, CPU and Game Console Business Development Division. Prior to joining TSMC, BJ has spent most of her career in Intel Corp., 24+ years. Because of her excellent work and leadership, she was nominated by Intel and was honored as 2006 inductee Hall of Fame for WITI (women-in-technology-international).

Learn more about speakers for The ConFab 2012 at http://www.theconfab.com/index.html.

Visit the Semiconductors Channel of Solid State Technology!

May 17, 2012 — Barclays Capital’s Asia IT analyst Jones Ku shares details of China’s State Council’s RMB26.5 billion (about US$4.2 billion) subsidy program for household electrical appliances. The program sets aside RMB2.2 billion to promote consumption of light-emitting diodes (LEDs) and “other energy-saving light bulbs.”

The program will be in effect for one year. In addition to LED lighting, it covers energy-efficient vehicles, air conditioners, flat-panel display (FPD) television sets, refrigerators, and more.

There is currently no announcement on when the program will start, though Barclays expects it to be implemented in late June or early July. Ku says it is likely to be similar to the subsidies for energy-saving products that were trialed in Beijing from September 2011-February 2012 (10% of the selling price, with a cap of RMB400), but with the focus expanded to include more cities.

The LED lighting product subsidy has been rumored since November 2011, when China laid out its plans to phase out incandescent bulbs within 5 years. Expectations on the subsidy amount lowered over time (some early estimates were as high as RMB8 billion). The $348M LED subsidy sum discussed may be less than anticipated, though with no details on the per-product subsidies or the format of the cash deployment, estimating the likely impact to China LED demand is difficult. Certain municipal governments may match the subsidies of the central government, lowering the product cost further, Barclays noted in an earlier subsidy assessment.

Barclays looks at a scenario wherein the government subsidizes ~50% of LED bulb cost (similar to the programs deployed on the CFL bulb side several years ago). Assuming an average pre-subsidy price per LED bulb of ~$10, and assuming China accounts for ~20% of the worldwide bulb market, this subsidy would translate to ~69M bulbs or ~4% of China’s bulb demand.

For more information, see Barclays Capital’s report, "U.S. Display & Lighting: Lightfair Highlights Positive End Demand Trends, But Continued ASP Pressure" https://?live.barcap.com/go/publications/content?contentPubID=FC1821997)

Visit the LED Manufacturing Channel on Solid State Technology and subscribe to the LED Manufacturing News monthly e-newsletter!

May 17, 2012 — Barclays Capital’s C.J. Muse looks at the forecast for PC graphics processing units (GPU) based on a new report from Mercury Research. Shipments of PC graphic devices in Q1 2012 came in in-line to slightly below historical trends. Nvidia lost share to Intel and AMD due to tight 28nm chip supply. A structural shift in the fabless/foundry relationship is occurring, moving toward a take-or-pay environment, with which leading-edge fabless players will have to contend going forward, Muse says.

Also read: TSMC, Samsung foundries reconsidering 2012 capex on stronger 28nm demand

GPU shipments declined -1% Q/Q to 123 million units in 1Q12 (from 124.3 million units in 4Q11) compared to typical seasonality of 1.9% Q/Q growth seen over the last 5 years.

Mercury Research’s updated PC Graphics market share forecast for Q1 2012 (“PC Graphics Chip Sets & Technologies,” http://www.mercuryresearch.com/products.shtml) shows that discrete GPUs reversed trends this quarter — growing +3% Q/Q while integrated fell -2% Q/Q. This suggests that PC ODM/OEMs can increase spending on graphics now that hard disk drive (HDD) prices are decreasing with supply restored from the Thailand flooding disaster. Despite the strength in the discrete GPU market, limited availability of Nvidia’s Kepler GPU resulted in 90 bps of share loss in the discrete GPU market, with Intel and AMD taking modest share from Nvidia.

INTC gained 25bps of share in total GPU, AMD added 17bps, while NVDA lost 61bps.

INTC saw +140bps of share gain in the integrated GPU market but this was offset by the modest shift from integrated to discrete in 1Q12.

AMD lost 30bps of share to NVDA in the mobile discrete market but gained 150 bps in the desktop discrete market benefitting from the earlier ramp of 28nm graphics parts as well as from lower shipments by NVDA due to limited availably of its desktop parts. AMD gained 90bps in the discrete market. In the integrated market, AMD gained modest share in desktops but loss share to Intel on the notebook side resulting in AMD’s share in the integrated market slipping to 17.5% from 18% the prior quarter.

Reflecting the wind down of its chipset business (0.6% share in integrated graphics down from 1.7% share the prior quarter) coupled with share loss in the discrete market to ATI, NVDA saw its total GPU share fall to 15.1% in Q1 2012 from 15.7% in Q4 2011. Share dynamics were a bit different in desktop vs. notebook segments, where NVDA outpaced AMD (NVDA had a record Q for notebooks). NVDA’s share loss in desktops (shipments fell 4% Q/Q) came from limited availability of 28nm Kepler GPU parts. Kepler is manufactured using TSMC’s 28nm high performance (HP) process, the foundry’s most advanced 28nm process which uses their first-generation high-k metal gate (HKMG) technology and second-generation silicon germanium (SiGe) straining, Nvidia shares.

NVDA’s management reported that its FY Q1 results were dampened by 28nm chip shortages. NVDA suggests that this is not a yield issue, Muse says, adding that it is likely due to poor planning on NVDA’s part and reluctance on foundry TSMC’s part to add more capacity. In the future, look for Nvidia to regain that share as 28nm shortages lessen, Kepler sees higher attach rates, and Nvidia fulfills Apple orders. Robust design-in momentum on the Ivy Bridge platform, Ultrabooks, as well as Apple products all point to incremental share gains for NVDA going forward.

Overall attach rates were healthy — desktop attach rates were flattish at 33.7% but notebook attach rates increased to 32.8% from 29.7% in Q4 2011.

Barclays analysts expect that attach rates will go up initially on Intel’s Ivy Bridge platform, as seen in prior refresh cycles, but that over time, the market and especially the low-end market will favor integrated GPUs as improvements are made to the graphics performance on integrated chips (Intel’s Ivy Bridge increases graphics performance by 2x). The sector will see an uptick in attach rates over the next couple of quarters, with attach rates declining longer-term.

After two quarters of favoring the desktop market, the GPU market shifted back to notebooks. Notebook GPUs (up 4%, better than 5-year average of +0.5%) outperformed desktops (down 6%, vs a 5-year average of +2.9%), with what Muse sees as a “secular shift back to notebooks once again.” The mix of desktop GPUs declined to 44% in Q1 2012 (back to mid-2011 levels) from 47% in Q4 2011.

Visit the Semiconductors Channel of Solid State Technology!

May 15, 2012 — The market for dynamic random access memory (DRAM) is expected to partially reverse the drastic losses it incurred in 2011 and achieve revenue growth this year, the result of balanced supply and demand following the exit of major manufacturer Elpida Memory Inc., according to an IHS iSuppli DRAM Market Tracker report.

Global DRAM industry revenue this year is forecast to reach $30.6 billion, up 3.3% from $29.6 billion in 2011. Although seemingly small, the revenue expansion for 2012 is a welcome development given the 25% contraction last year. The overall picture will continue to brighten during the next few years, as shown in the figure, with DRAM revenue exceeding $30 billion each year for the next five years and reaching $40.2 billion in 2016 — an unprecedented run scaling unparalleled heights for the market.

Figure. Worldwide DRAM industry revenue forecast. SOURCE: IHS iSuppli.

“This year’s anticipated turnaround comes as somewhat of a surprise, especially as the challenges of 2011 appeared to point to a calamitous 2012,” said Mike Howard, senior principal analyst for DRAM & memory research at IHS. “Weak demand was one of the major challenges last year, when revenue slipped each quarter as prices went from bad to worse. However, the key problem was excess DRAM manufacturing capacity — the same trouble that has bedeviled the industry for much of its history.”

Compounding the difficulties last year were the October floods in Thailand, which depressed PC shipments — a traditional DRAM stronghold. The perceived scarcity of hard disk drives pummeled PC sales and thereby DRAM demand; and the paucity of hard drives meant PC manufacturers were paying more for storage — leaving even fewer dollars to spend on DRAM.

DRAM prospects started looking better, however, after the bankruptcy filing in February of Japan’s Elpida. Elpida was part of the elite echelon of DRAM manufacturers that includes Samsung Electronics Co. Ltd. and Hynix Semiconductor Inc. of South Korea, as well as U.S.-based Micron Technology Inc.

“Elpida’s insolvency will have a massive impact on the industry’s fortunes, primarily because it promises to shift the market from a state of endemic oversupply to sorely needed balance for most of 2012,” Howard said. “As a result of such developments, IHS is now cautiously optimistic that the DRAM industry may actually be through the downturn and headed for improvement.”

Even with the final outcome of Elpida’s bankruptcy uncertain and the disposition of its assets still in negotiation (Micron looks to be the forerunner), the rest of the industry is expected to benefit from Elpida’s exit, with the market lifting on signs of supply rebalance.

Moreover, no significant deterioration is expected within the global economic environment, which should help reverse the paucity of demand that has hobbled the industry of late.

The DRAM space can look forward to continued strong expansion in the next few years because of three powerful growth drivers: ultrathin PCs, smartphones and tablets.

Ultrathin PCs — a category that includes Intel Corp.’s ultrabooks, the MacBook Air from Apple Inc. and ARM-based lightweight PCs — will present plenty of new opportunities for low-power DRAM, especially when ultrathins comprise the majority of shipments by 2016. For high-end ultrabooks in particular, PC manufacturers are projected to have enough margin to afford the installation of low-power double data rate 3 (LPDDR3) DRAM in their products, adding to overall DRAM industry revenue. LPDDR3 will account for as much as 19 percent of the total DRAM market in 2014.

In the case of smartphones, increasing shipments during the next five years coupled with growing memory content per phone suggest rosy prospects as well for DRAM. Average DRAM content in smartphones this year will amount to 5.1 gigabits, up from 3.5 gigabits last year and from 2.3 gigabits in 2010.

Overall, the net increase in DRAM due to smartphones will equate to a compound annual growth rate (CAGR) of 65% from 2011 to 2016.

Very rapid growth is also forecast for DRAM in tablets, with average DRAM content in 2012 for the devices projected to hit 8.0 gigabits, up from 4.5 gigabits last year and 2.5 gigabits in 2010. When tablet shipment increases are combined with DRAM content growth, DRAM use in tablets will be up a substantial 95% CAGR for the same five-year period from 2011 to 2016, accompanying smartphones and ultrathins in boosting total DRAM revenue. Visit the Semiconductors Channel of Solid State Technology!

May 14, 2012 — At the 15th Symposium on Polymers for Microelectronics (May 8-10 in Wilmington, DE), TSMC and Yole Developpement gave plenary presentations on the use of polymeric materials in wafer-level packaging (WLP) from foundry and overall industry perspectives.

The most controversial comment came from TSMC’s Doug Yu, senior director of front-end and back-end technology development, who challenged the current nomenclature and pronounced that the versatile interposer technology should be called

May 14, 2012 — Heraeus introduced its high-reliability composite aluminum/copper (CucorAl) semiconductor bonding wire, which offers strong mechanical and electrical bonds to semiconductor pads, with good thermal properties.

During passive temperature cycling and active power cycling tests, the Al-clad Cu bonding wire showed improved long-term reliability over Al bonding wires.

The aluminum coating is soft, structured around the copper core in a way that enables a reliable bonding window with existing chip metallizations and conventional wirebond tools. The thick wire is 60-70% copper by volume. It is available in 200-500