Category Archives: Editors Picks

November 2, 2012 – OLED revenues are currently being driven by display applications (e.g. smartphones), but there’s a new battleground slowly emerging: OLEDs for lighting applications where the technology could offer some advantages in design and efficiency for some applications — if panel makers are willing to make some sacrifices, according to a report from Yole Développement.

Conventional LED technology has paved the way in solid-state lighting, and has a large headstart; OLED has to overcome high costs and current lower efficiency, which are hampering market adoption and penetration. The firm sees OLEDs for lighting making initial inroads in specific lighting applications (automotive, general lighting) and in niche specialty and high-end lighting where it can offer some differentiation in design options. To crack more traditional lighting markets (commercial, office buildings, etc.), however, OLED technology will have to advance the technology and expand across different niche markets to achieve economies of scale and will decrease costs. Yole pegs this happening sometime in 2014, with the rise of larger substrates and better process control.

Pars Mukish, technology & market analyst for LED & OLED at Yole, then foresees an astonishing growth projection for OLED lighting panels: from a $2.8M market this year (2012) to nearly $1.7B by 2020, with general lighting applications representing more than 70% of that business.

OLED panels revenue for lighting applications. (Source: Yole Développement)

That won’t come easy, though. There are a number of materials and OLED structures being explored and in production, tweaked to improve performance and lifetime and also decrease manufacturing costs. Polymer materials for OLEDs continue to struggle (vs. small-molecule OLED materials) in demonstrating their capabilities to lower costs and improve performance to production-acceptable levels. Rigid glass is still the go-to substrate for OLED lighting panels, but work continues on other flexible OLED technologies including roll-to-roll processing, ultrathin glass, and encapsulation options.

To have a chance at fulfilling the aforementioned growth expectations for OLED lighting, OLED panel makers have to quickly identify the winning technology approaches and time-to-market strategies. "New business models are mandatory as the traditional lighting industry will be reluctant to integrate new technology as it could eat away at margins — OLED cost directly impacts the cost of OLED-based luminaires," points out Milan Rosina, Yole’s technology & market analyst for OLED & photovoltaics. The kicker: both the new OLED technology and its integration into production are brand-new to panel makers, who are unlikely to sacrifice existing LED lighting sales and complicate production just to deploy a new technology, he notes.

Thus the key to OLED technology’s future in more mainstream lighting applications, the Yole analysts say, boils down to how and when panel makers can establish vertical integration strategies and figure out how to push the new technology through existing distribution channels. And above all, find that "spark" niche market (or markets) that will pave the way to economies-of-scale, which will open up the conversations to convey opportunities and advantages for OLED technology in general consumer lighting applications.

November 1, 2012 – X-Fab Silicon Foundries says it has become the majority shareholder in German MEMS Foundry Itzehoe GmbH (MFI), the latest in a series of recent moves to raise its profile as a top MEMS foundry.

The MFI business, renamed X-Fab MEMS Foundry Itzehoe, complements X-Fab’s capabilities in its MEMS foundry in Erfurt, adding technologies for microsensors, actuators, micro-optical structures and hermetic wafer-level packaging processes. X-Fab originally signed MFI as a contract MEMS manufacturing partner in Feb. 2011, a deal that expanded its capabilities across a range of 200mm MEMS technologies. Its ownership stake in MFI is now 51%, up from 25.5%.

X-Fab MEMS Foundry Itzehoe will continue its long-term cooperation with the Frauhofer Institute for Silicon Technology‘s (ISIT) MEMS Group. MFI was spun out of ISIT in 2009 and is located within the same wafer fabrication facility in Itzehoe/Germany.

"Our customers will benefit from both an even wider spectrum of available MEMS technologies and from direct access to X-Fab’s manufacturing facilities for CMOS-compatible MEMS processes," stated Thomas Hartung, VP of marketing at X-Fab Group. "X-Fab MEMS Foundry Itzehoe will play an important role in the implementation of our MEMS strategy, and brings us closer to our goal of becoming one of the top three pure-play MEMS foundry providers."

"The rich combination of the versatile MEMS-specific technology portfolio at the Itzehoe-based MEMS foundry and the development expertise of Fraunhofer ISIT greatly expands the capabilities of X-Fab’s technology offering," added Peter Merz, managing director of X-Fab MEMS Foundry Itzehoe. "We are delighted to provide the full bandwidth of MEMS technologies including vacuum and optical wafer-level packaging or TSV backed by X-Fab’s existing and well-proven foundry services. This integration brings X-Fab customers bundled and accelerated product development and manufacturing cycles for micro-machined devices such as inertial sensors, micro-mirrors, and piezoelectric transducers."

Barely a month ago X-Fab pledged to invest $50M over the next three years to support projected growth and a goal of "becoming one of the top three worldwide suppliers of MEMS foundry services." (X-Fab placed 10th in Yole Développement’s 2011 MEMS foundry rankings, surging 33% to roughly $16M in revenues, about $31M shy of No.3 Silex Microsystems.) Among X-Fab‘s other recent MEMS accomplishments:

 

October 31, 2012 – Applied Materials has announced two new tools for making ultrahigh-definition displays and high-pixel-density screens for mobile devices. One offers a new design for depositing IGZO films for TFTs; the other handles bigger substrates of low temperature polysilicon (LTPS) films to help lower manufacturing costs.

The Applied AKT-PiVot PVD for metal oxide-based thin-film transistors (TFTs) enables a transition from aluminum to copper interconnect bus lines leading to faster pixel response and lower power consumption in LCD TV panels. It overcomes the problem of "mura effect" that reduce display quality, which the company says has hindered metal-oxide technology’s inroads into mainstream LCDs. The "breakthrough" stability of the IGZO films deposited by the tool offers the promise of metal oxide backplanes for OLEDs which would significantly lower their cost as well, the company adds.

(Source: Applied Materials)

A proprietary rotary cathode design employs unique deposition modulation technology to deposit copper layers and form the transistor channel with uniform grain distribution, low resistivity and high thickness uniformity. The technology enables nearly 3× higher target utilization than competitive systems, according to the company, and its rotary targets have >4× longer lifetimes than conventional planar targets.

(Source: Applied Materials)

The Applied AKT-PX PECVD is an extension of the company’s line of PECVD systems to deposit highly-uniform LTPS films on glass substrates. The new tool extends to larger sheets (1.6-5.7m2, or Gen 5 to Gen 8.5 sizes) to help manufacturers increase production and drive down costs, and accelerate the transition of LTPS technology to larger screen sizes for both mobile devices and TVs, the company points out. AMOLED and advanced TFT-LCD displays are switching to the polysilicon-based transistors, which offer higher electron mobility vs. the amorphous silicon (a-Si) used in conventional LCD displays, leading to smaller and faster pixel-controlling transistors, and displays that are brighter, sharper, and use less power — features most desirable for mobile applications.

(Source: Applied Materials)

"The display industry is undergoing one of the most critical technical transitions in the last 20 years — which is being driven by advances in TFT technology," stated Tom Edman, group VP and GM of Applied’s display business group. He added that "customers have reported excellent results with our systems and we already have received multiple orders from major display manufacturers."

Visit our new Displays Manufacturing Channel on Solid State Technology and subscribe to our Displays Digest e-newsletter!

October 30, 2012 – This year is shaping up to be a historically lousy year for makers of flat-panel display (FPD) manufacturing equipment, but expectations are looking up that demand will catch up to supply in 2013 and balance the market, according to NPD DisplaySearch projections. Spending on FPD equipment is projected to plummet -69% in 2012 to $3.8 billion, making it the worst year in the sector’s history. But even with slow demand growth in maturing markets (TVs and PCs), the firm sees "significantly improved conditions" in 2013, more than doubling to $8.3B.

Most of that spending will be for new low-temperature polysilicon (LTPS) fabs or converting existing amorphous silicon (a-Si) capacity to LTPS, both for use in TFT-LCD and active-matrix OLED (AMOLED) production, explains Charles Annis, VP of manufacturing research at NPD DisplaySearch. "One reason spending is increasing so much is because LTPS fabs cost substantially more than a-Si fabs to build. There are extra process necessitate more than 10 mask steps." LTPS fabs also require higher-priced equipment, particularly high-resolution photolithography tools, he added, but having those technologies does enable production of high-value displays used in smartphones and tablets.



FPD equipment spending forecast. (Source: NPD DisplaySearch)

Such dramatic cutbacks in investment will more quickly rebalance supplies with demand and raise fab utilization rates. Meanwhile, new manufacturing technologies (oxide semiconductors, in-cell touch, flexible AMOLEDs and AMOLED TVs) promise lower costs and higher-value applications. Together that spells improve profitability for panel makers, notes Annis. Even with the cautionary disclaimer that new investments (e.g. AMOLED capacity) can be pushed out or cancelled if performance and cost targets don’t materialize, most of the firm’s indicators project 2013 "to be a much better year than 2012."

October 29, 2012 – It’s been a familiar refrain: with every new Windows operating system that comes out, OEMs to end-users upgrade their systems to handle the new capabilities, including more memory. But not this time with Windows 8. In a departure from past iterations, Windows 8 will not cause any significant rise in DRAM unit shipments, predicts IHS iSuppli.

The firm is projecting only an 8% increase in global DRAM bit shipments in 4Q12 vs. 3Q12, which includes DRAM for PCs, smartphones, and tablets. That’s a departure from the double-digit percentage increases seen with every other major MS OS release. Windows 3.1’s release in the second quarter of 1992 sparked a 29% increase in DRAM shipments, up from 12% in the prior quarter. Similarly, Windows 95 caused DRAM shipments to surge 23% in 4Q95, vs. the 11%-14% seen in the previous four quarters. Three successive Windows versions in the late 1990s and early 2000s all pushed DRAM shipments up by 40% or more in their respective release quarters.

Windows version Release date % DRAM bit growth Q/Q
3.1 1Q92 29%
95 4Q95 23%
98 3Q98 40%
2000 1Q00 49%
XP 3Q01 41%
Vista 1Q07 24%
7 4Q09 18%
8 4Q12 8%

Historical DRAM bit shipment percentage growth, sequential quarterly change in unit shipments. (Source: WSTS, IHS iSuppli)

Credit (or blame?) Windows 8’s "lean hardware requirements," the firm says. "Starting with Windows 7 and continuing with Windows 8, Microsoft has taken a leaner approach with its operating systems, maintaining the same DRAM requirements as before,” stated Clifford Leimbach, analyst for memory demand forecasting at IHS.

And Windows 8 won’t really make a dent in the increase of PCs shipped in 4Q12 when it’s released, either. "Consumers are continuing to eschew new PC purchases in the fourth quarter, with Windows 8 not expected to change this situation,” he added. (And it’s not only consumers either; Gartner thinks 90% of businesses will still be keeping Windows 8 at arm’s length by 2015.)

It’s one more sign of the sea change in computing away from desktops and toward mobile devices. PC share in the DRAM space dipped below 50% for the first time in 2Q12 of this year, IHS iSuppli calculates, handing the mantle to other devices increasingly gobbling up memory such as smartphones and media. In this new era of diversified computing and applications, a new OS just doesn’t have the hardware oomph it used to. (And that’s not entirely a bad thing — the highs of unit shipment growth won’t be as high, but the lows also will be more muted because of the diluted impact of a disparate application base, the firm notes.)

October 26, 2012 – Communications applications are far and away the leading growth market for ICs over the next five years, and PCs and consumer applications have fallen far back in the pack.

IC Insights projects a 7.4% compound annual growth rate for total IC sales from 2011-2016. That’s largely because of good growth in the communications end market (14.1% CAGR) which will nearly double the entire IC market over the period — and nearly double in size to $160B by 2016. The Asia-Pacific region will see the lion’s share (61%) of this year’s comm IC sales, up from 59% in 2011.

Also ahead of the curve is the automotive IC market (9.0% CAGR), pegged to surge to $28B in 2016, 53% bigger than where it was in 2011. Europe has been the big market in that sector, accounting for 37% of sales in 2012 — but by 2016 the Asia-Pacific market will be nearly equal in size, IC Insights projects.

Another emerging sector is medical/industrial applications, within which is an aging global population requiring home healthcare as well as other industrial applications. Analog ICs will continue to dominate here (45% of total in 2012, and still the largest in 2016). The government/military IC market will reach $2.46B by 2016 but still be a tiny slice of the total market (<1%).

IC market growth forecast by application, 2011-2016 CAGR. (Source: IC Insights)

The two slowest sectors are the ones which have fallen the farthest from grace. Computer IC sales will drop to 34% of the total market in 2012 vs. nearly 42% in 2011. This year’s -9% drop in computer IC sales is due to a -12% decline in memory sales. And consumer ICs will register just 1.9% CAGR from 2011-2016. Japan was "once the stronghold of the consumer electronics business," points out the analyst firm, but in 2012 it will hold less than half the share of the consumer IC market vs. the Asia Pacific region (22% to 50%).

IC Insights had already been predicting that communications will usurp computers as the biggest application for leading-edge ICs. Comm will be the biggest end-use IC app in every major geographic region starting in 2014. Communications accounted for 31% of worldwide IC sales, vs. 41.7% for computers; the analyst firm sees them swapping to 42.2% and 34.0% by 2016.

(Source: IC Insights)

In fact, telecom apps are already the biggest growth areas for ICs, notes IC Insights. In 2012 the top three growth areas are all telecom, and thanks to application-specific processors used in smartphones and tablet computers. Also note that NAND flash continues to grow, though at much slower rates due to a surge in capacity (and vendors) and subsequently lower pricing pressures. Auto is a top market in 2012 as ell, thanks to needed functions such as self-parking systems, stability control, and collision avoidance.


Top growing IC markets (in revenue growth), 2010-2012F. (Source: IC Insights)

Note that the lines are blurring with every new generation of technology — what counts as a consumer device, or a business application, or communications usage. "It is getting more difficult to keep nice matter-of-fact demarcations between the products," acknowledges IC Insights president Bill McClean. "We are classifying by type of product and not really who is the end user." That puts smartphones in the "communications" category because it’s a type of cellphone, whatever else they are capable of doing and are used for. Consumer gear, then, means systems including TVs, stereo equipment, iPods, etc., he explains. Tablets are a form of PCs so they are included as computers, though some market watchers (particularly for early versions) ranked them as personal media devices.

October 24, 2012 – Semiconductor manufacturers can expect equipment costs to increase about 15% for each new process node, but all the advantages of moving to smaller feature sizes are no longer offsetting those costs — even with the promised advantages of a 450mm wafer transition, say Gartner analysts in a new report.

A 450mm manufacturing transition — which the industry seems to have finally embraced and is pushing ahead on multiple fronts — promises 30% cost reductions vs. 300mm wafer processing, even with the inherent extra challenges in processing the bigger wafers. (Caveat: Gartner’s bearish on the 450mm benefits, saying it might only give 10% savings, or even none at all.) New semiconductor technologies (materials, equipment, architectures) being developed could also slow the rate of cost increases.

Unfortunately, that 30% reduction from a 450mm wafer-size transition "represents about only three or four years of increasing equipment costs, and consequently, delays the inevitable," the analysts say. And even if new advanced technologies pan out as promised and reduce costs — and that’s a big "if," getting semiconductors down through the teens and into the single-digit nodes — that too will be only a temporary reprieve: "The reality is that rising costs are a permanent part of the industry, and the fundamental economics of the industry may start changing sooner rather than later."

Here’s more data they use to back up their case:

  • Equipment costs for leading-edge semiconductor manufacturing are increasing 7%-10% per year, depending on the basic process;

  • By 2016, the minumum capex budget needed to justify building a new fab will be $8B-10B for loic, $3.5B-$4.5B for DRAM, and $6B-$7B for NAND flash;

  • By 2020, current cost trends will push the pricetag for a leading-edge fab investment to a budget-melting $15B-$20B;

  • At current spending rates, only eight chipmakmers will have the financial capability to build new fabs in the next few years.

For all but those chosen few leading-edge chipmakers, here is the Gartner analysts’ advice:

  • To big chip consumers: firmly attach your supply chain to one of those few leading-edge fabs, and make sure any fabless partners link themselves to one of the few foundries remaining on the leading edge.

  • Fabless semiconductor companies: get long-standing contractual arrangements with leading-edge foundries to guarantee access.

  • For the biggest fabless companies: Even if you’re seeing short-term wafer shortages, don’t jump into manufacturing yourself — explore a joint venture with leading foundries to ensure supply.

300mm average wafer fab equipment cost projections. (Source: Gartner)

The real kicker here? Lower costs per transistor aren’t even the real driving factor for semiconductor manufacturers, it’s reducing power consumption while maintaining performance, the Gartner analysts note. That’s the key functionality driving semiconductor content in mobile devices, and that’s what is driving the market now. (It’s a little different for memory which does rely on lower unit/production costs even if it’s a slight improvement.) "Overall, the semiconductor and electronics industry will have to come to grips with the fact that traditional cost reductions with each new node are in jeopardy, and in the future, higher performance may very well come at a higher price," the analysts note.

And one more factor to consider: semiconductor average selling prices (ASPs) have benefitted from the relentless node-shrink cost reductions, but semiconductor manufacturers are unlikely to (nor should they) swallow all these cost increases by themselves for the good of the rest of the supply chain. "What is more likely is that ASP trends will reverse after years of decline, and that reversal will change the supply-and-demand economics of the industry in ways that we don’t really understand today," the Gartner analysts admit.

October 23, 2012 – The semiconductor manufacturing sector continues to spin its wheels in an attempt to find traction amid lingering economic uncertainty. Another example: wafer demand is slowly pulling back onto a growth track, and will set new records in terms of shipments in the coming two years, according to data from SEMI.

Total wafer shipments declined -3% in 2011 after a record surge in 2010, and are expected to stay flat in 2012, according to SEMI’s latest tally. But shipments should swing back into mid-single-digit growth in 2013 and 2014 and surpass 2010’s levels.

"In spite of heightened economic uncertainty, silicon shipments for the first half of 2012 were very promising," stated Denny McGuirk, president and CEO of SEMI. "We expect 2012 silicon shipments to remain essentially flat when compared to 2011 and increase in 2013 and 2014."

Total electronic-grade silicon slices in millions of sq. in. shipped by the wafer
manufacturers to the end-users. Includes virgin test wafers and epitaxial
silicon wafers; does not include non-polished wafers. Shipments are for
semiconductor applications only and do not include solar applications.
(Source: SEMI)

A closer look at wafer demand, courtesy of Semico Research, shows a clearer picture of wafer demand, and the swings for demand of different silicon types. Total wafer demand is increasing at around 11% CAGR over the next five years, but the splits by technology show that wafers for MPUs only make up 3% of total industry wafers — while wafers processed into chips for wireless end use have grown tenfold in the past decade. (Other segments have eroded, like DSPs, as capabilities are integrated into the system-on-chip.)

Silicon shipments vs. wafer demand. (Source: Semico Research, SEMI)

The biggest driver in wafer demand, clearly, is anything tied to mobile devices, and the big winner is NAND flash memory. Demand for NAND wafers tripled from 2005-2010; even with NAND manufacturers increasing density with multilevel cell architectures, NAND wafers have increased their share of total wafer demand from 4% in 2005 to nearly 18% now, according to Semico, and 51 million NAND wafers will ship by 2016. NAND wafer demand in 2011 peaked at 32Gb/45nm, while 2012 NAND demand will peak at 64Gb/22nm. The more mobile devices proliferate with increasing functionality, the stronger the appetite for NAND per system — and as more mobile/remote functionality is pushed into "the cloud" NAND will be making inroads into servers, too, thanks to its better power usage and faster access speeds, Semico notes.

All these different variances in demand funnel down to one very stressed position these days in the semiconductor world. While engineers and visionaries are figuring out how to keep Moore’s Law going with new architectures and materials, purchasing managers are on the hot seat to navigate the variances in wafer usage and inventories. "Product ramps, semiconductor sales, and technology transitions are getting more and more difficult to plan," explains Joanne Itow, managing director at Semico Research. "A purchasing manager’s job is tough in these volatile times."

Keeping it Cool

Back in 2008 we addressed 3D cooling activities [see PFTLE 43, "Keeping it cool in the dog days of summer"] looking a the activities at IBM Zurich, GaTech, and CALCE (U Md) as the groups especially active in this area.

Since then we have looked further at the liquid cooling activities of Bakir at GaTech [see IFTLE 83, "Orange County IEEE CPMT 3DIC Workshop"] and Brunschwiler at IBM Zurich [see "IBM to use water cooling for future 3D IC processors"] and the fact that one of the drivers for 2.5D is that it offers better thermal performance that current 3D stack solutions [ see IFTLE 97, "DATE in Dresden, Synopsys 3D EDA solution"]. For the most part, though, IFTLE has taken the position that thermal would not be the roadblock for 3DIC and that initial products would be ones where the thermal solution was not driving the technology.

Now that we are quickly approaching full commercial production of a number of products, it’s probably a good time to focus more on proposed thermal solutions for the future. To update yourself on where things stand, I suggest Herman Oprins’ article "Modeling and experimental characterization of hot spot dissipation in 3D stacks." He concludes that thermal management issues in these 3D stacks are one of the main challenges for 3D integration since the use of polymer adhesives with low thermal conductivity, the presence of interconnection structures, back end of line (BEOL), redistribution layers (RDL), and through-Si vias (TSVs) increases the complexity of the conductive heat transfer paths in a 3D stack.

Oprins concludes that hot spot power dissipation results in significantly higher temperatures in 3D stacked chips compared to the same power dissipation in single 2D chips. This temperature increase is mainly due to the reduced thermal spreading in the thinned dies on the one hand, and to the use of adhesives with low thermal conductivity for the vertical integration of the chips on the other hand. To limit the temperature increase in 3D-ICs, "too thin chips should be avoided" because the thinner the silicon substrate, the higher the thermal spreading resistance is in the case of hot spots. Simulations show that a minimum die thickness of 50

October 18, 2012 – Worldwide flat-panel display (FPD) revenues will reach a record $120 billion in 2012, up 8% from a challenging year in 2011, and the recovery is entirely on the backs of TFT-LCDs and AMOLED displays, according to NPD DisplaySearch.

TFT-LCD displays still make up the vast share of all display sales (~90%), and so the overall market tracks in-step with this segment, rebounding from a -5% decline in 2011 to an 8% rise in 2012. Note, though, that of all the other display technology slices, AMOLEDs have by far the best growth trajectory — two years ago it was fourth in total market share (1%) behind plasma, passive matrix, and roughly tied with CRT; now it’s the second-most-popular display technology with a 5.4% share and the gap is widening. Credit surging manufacturing capacity and expansion of market players, DisplaySearch says. The only other segment to see any growth is liquid crystal on silicon (LCOS) used for microdisplays. Also note the sharp rise and sharper plummet of active-matrix electrophoretic displays (AMEPD), used in monochrome e-reader devices, which are giving up ground to TFT-LCD tablet PCs.

Worldwide FPD revenues by technology, 2010-2012. (Source: NPD DisplaySearch)

2011 was a tough year for displays due to price erosion in TFT-LCD panels, particularly for TV applications, DisplaySearch notes. The rebound in 2012 has many factors behind it: bigger average sizes and shipments of LCD TVs, higher prices for high-resolution mobile displays, strong unit growth for tablet PCs, expansion of AMOLED shipments and applications, thinner and lighter ultraslim notebook PC panels, the emergence of 4k × 2k LCD TVs, and demand for a number of applications including games, car navigation systems, and digital signage.

"While the industry faces challenges in traditional applications such as plasma TVs and mainstream sizes of LCD TVs and desktop monitors, the addition of new features and lower prices are driving growth of applications such as tablet PCs and smartphones," explained David Hsieh, VP at NPD DisplaySearch.

The outlook for the FPD industry isn’t entirely cloud-free: there’s a lot of saturation in several major markets, Hsieh noted. Nevertheless, the supply chain is figuring out how to "increase the value proposition" of FPDs by emphasizing their technology improvements: higher resolution for mobile devices, bigger screens, thinner and lighter versions for mobile PCs, improved wide-viewing angle, and desirable functionalities like touchscreen. "We expect 2013 to be a good year for the FPD industry, with revenue increasing, as TFT LCD prices recover and AMOLED demand grows," Hsieh said.