Category Archives: LED Manufacturing

December 28, 2012 – Researchers in Japan have devised a microelectromechanical system (MEMS) fabrication technology using printing and injection molding, fabrication of large-area devices with low capital investment, without a vacuum process, and lower production costs. Thus, MEMS devices can be made and applied for fields where manufacturing cost has been an issue, such as lighting.

The team from the Research Center for Ubiquitous MEMS and Micro Engineering of the National Institute of Advanced Industrial Science and Technology (AIST) integrated microfabrication technology and MEMS design evaluation technology, and combined it with Design Tech Co. Ltd.’s signal processing technology to fabricate a lighting device.

Conventional commercial MEMS devices use fabrication techniques with semiconductor manufacturing systems used to produce integrated circuits, including vacuum processes. Resins could be used to form patterns onto moving microstructures but production costs are high due to vacuum-based processes. Also, it has proven difficult to form and thin MEMS structures such as springs and cantilevers because resins harden immediately after mold injection.

AIST researchers now say they have realized low-cost printing and transferred the structure using injection molding, and improved the mold structure to fill thin moving structures. A film for transferring the MEMS functional laser is formed, and the release layer and MEMS functional layer are printed onto the film with a screen or gravure printer. The printed film is aligned and put into an injection mold, into which is injected a molten resin that is cooled and solidified into the MEMS structure. The mold is then opened and the MEMS structure is separated from the film; the ink layers printed on the film are transferred to the MEMS structure.

Figure 1: MEMS fabrication processes by printing and injection molding.

The printed MEMS functional layers can be changed according to the desired purpose of the MEMS device — from acceleration sensors and gas sensors to power generation devices. This enables low-cost MEMS fabrication in fields where costs are currently too high. One example the AIST highlights is in light distribution control of LED lighting. MEMS mirrors produced with semiconductor manufacturing processes are based on costs determined by devices per wafer; so large-area mirrors are costly, while more cost-friendly micromirrors necessitate a more complex optical system. This new MEMS fabrication technology, though, could produce low-cost large MEMS devices (larger than several mm across), which opens the door for MEMS-based active light distribution control devices. Future work will seek to improve the symmetry of the MEMS mirror synchronization with the LED timing, and expand the range of the light distribution by improving the arrangement of the optical system, the signal processing, and the control circuit.

Figure 2: MEMS mirrors for active light distribution fabricated by using only printing and injection molding (left), and examples of the resulting light distribution patterns (right).

Injection molding can be used easily to form complex 3D objects such as spheres; the researchers expect MEMS devices will be formed on the surface of, or inside, 3D objects. Moreover, injection molding processes are commonly available in Japan, and systems cost less than semiconductor manufacturing systems. AIST projects its work will lead to MEMS fabrication coming out of non-semiconductor industries, such as plastics molding — and participation from these other sectors into MEMS manufacturing will help develop new applications for MEMS devices.

Figure 3: Examples of MEMS devices fabricated with the AIST technology. Top & middle: A reflective mirror and a mirror displacement sensor incorporated into a MEMS mirror device for lighting. A mirror ink for the reflective mirror, a conductive ink for the strain sensor, and a magnetic ink for driving the mirror are printed on the film, and then the printed ink patterns are transferred to the MEMS structure by injection molding. The MEMS mirror device for lighting did not break after more than 100 million operations driven by an external coil. Bottom: A MEMS device array can be fabricated using an arrayed MEMS pattern mold.

Canon U.S.A., Inc. recently launched the FPA-3030i5+ i-line stepper, designed for the manufacturing of LEDs, MEMS and power semiconductors. The FPA-3030 platform is an upgrade to earlier Canon “FPA-3000 platform” steppers.  The FPA-3030i5+ features an overhauled software structure and electrical control system that allow application of optional advanced hardware (e.g., projection lens, wafer stage, and alignment system) that is not compatible with traditional FPA-3000 platform steppers.

The FPA-3030i5+ is capable of providing imaging resolution below 0.35mm, while maintaining overlay accuracy of less than or equal to 40nm and throughput equal to or in excess of 104 wafers per hour. 

The FPA-3030 platform allows the use of optional equipment designed for the processing of silicon (Si), sapphire (Al2O3), silicon carbide (SiC) and a wide variety of wafer materials used in  environmentally conscious device manufacturing. Optional equipment for the FPA-3030i5+ includes warped-wafer handling systems to allow processing of distorted substrates, and advanced image processing systems for clear substrates.

With the purchase of the optional Multi-Size Wafer Kit, the FPA-3030i5+ stepper can also be configured to process multiple wafer sizes, and can be equipped with other optional equipment to help improve productivity and efficiency.

December 11, 2012 – Taiwan makers of light-emitting diode (LED) products are unlikely to see a recovery in 2013 LED equipment spending due to a lack of investments and an alarming sector-wide cash crunch, warns one industry analyst.

Maxim analyst Aaron Chew, in a recent research note, lowered his outlook for MOCVD tools in 2013 to 240 tools, down from previous expectations of 300 tools. Despite investments drying up due to LED chip overcapacity and underutilization, Taiwan had been seen as "a beacon of hope for a recovery in MOCVD spending," but widespread losses and rising debt are choking off the money needed to support those reinvestments, he cautions.

The top eight Taiwan LED chip producers (Epistar, Lextar Electronics, Formosa Epitaxy, Genesis Photonics, Huga Optatech, Tekcore, Epileds, Arima Optoelelectronics) have generated around $67M in operating losses through almost all of 2012 YTD despite "burning through $1 billion in cash since 1Q10," and they are "burdened by $1B in debt with cash down (21%) YTD to $900M" — meaning they don’t have the financial muscle to support a market rebound, he writes. Taiwan’s flagship LED maker Epistar — which represents over half of the total $2.5B market cap of Taiwan’s top eight LED chipmakers — is still generating 14% gross margins and eking out 3% operating margins, but even it has generated massive losses in free cash flow ($236M) over the past two years.

Swimming beneath these murky financial waters is the shadow of consolidation, threatening to undo the LED "arms race" that blossomed in 2010-2011, Chew says. Epistar has acquired smaller competitor Huga Optatech and taken stakes in Tekcore and Na Ya Photonics, but also making inroads is China’s Sanan Optoelectronics, with a 20% stake in Taiwan’s ForEpi. As capacity continues to concentrate with consolidation, look for capital spending to rationalize, Chew says.

Taiwan LED makers have an installed base of roughly 1000 MOCVD tools, representing more than a third (35%-40%) of all MOCVD investments since 2010, Chew notes. But with total capex down (61%) to $273M, and the only ones really spending are Epistar and ForEpi, Chew sees MOCVD spending holding flat in 2013. With overcapacity/underutilization rampant among China and Korea LED makers, and Taiwan unable to float a LED recovery by itself, 2013 doesn’t look as rosy for LED equipment demand.

Capex at the Top 8 Taiwan LED chipmakers. Applies average quarterly NT$ / US$ exchange rates ranging from NT$29 to NT$30 for each period. (Source: Maxim Group, citing company data)

In an IC fab, cycle time is the time interval between when a lot is started and when it is completed. The benefits of shorter cycle time during volume production are well known: reduced capital costs associated with having less work in progress (WIP); reduced number of finished goods required as safety stock; reduced number of wafers affected by engineering change notices (ECNs); reduced inventory costs in case of a drop in demand; more flexibility to accept orders, including short turnaround orders; and shorter response time to customer demands. Additionally, during development and ramp, shorter cycle times accelerate end-of-line learning and can result in faster time to market for the first lots out the door.

Given all the benefits of reducing cycle time, it’s useful to consider how wafer defect inspection contributes to the situation. To begin with, the majority of lots do not accrue any cycle time associated with the inspection, since usually less than 25 percent of lots go through any given inspection point. For those that are inspected, cycle time is accrued by sending a lot over to the inspection tool, waiting until it’s available, inspecting the lot and then dispositioning the wafers. On the other hand, defect inspection can decrease variability in the lot arrival rate—thereby reducing cycle time.

Three of the most important factors used in calculating fab cycle time are variability, availability, and utilization. Of these, variability is by far the most important. If lots arrive at process tools at a constant rate, exactly equal to the processing time, then no lot will ever have to wait and the queue time will be identically zero. Other sources of variability affect cycle time, such as maintenance schedules and variability in processing time, but variability in the lot arrival rate tends to have the biggest impact on cycle time.

In the real world lots don’t arrive at a constant rate and one of the biggest sources of variability in the lot arrival rate is the dreaded WIP bubble—a huge bulge in inventory that moves slowly through the line like an over-fed snake. In the middle of a WIP bubble every lot just sits there, accruing cycle time, waiting for the next process tool to become available. Then it moves to the next process step where the same thing happens again until eventually the bubble dissipates. Sometimes WIP bubbles are a result of the natural ebb and flow of material as it moves through the line, but often they are the result of a temporary restriction in capacity at a particular process step (e.g., a long “tool down”).

When a defect excursion is discovered at a given inspection step, a fab may put down every process tool that the offending lot encountered, from the last inspection point where the defect count was known to be in control, to the current inspection step.  Each down process tool is then re-qualified until, through a process of elimination, the offending process tool is identified.

If the inspection points are close together, then there will be relatively few process tools put down and the WIP bubble will be small.  However, if the inspection points are far apart, not only will more tools be down, but each tool will be down for a longer period of time because it will take longer to find the problem.  The resulting WIP bubble can persist for weeks, as it often acts like a wave that reverberates back and forth through the line creating abnormally high cycle times for an extended period of time. 

Consider the two situations depicted in Figure 1 (below). The chart on the top represents a fab where the cycle time is relatively constant. In this case, increasing the number of wafer inspection steps in the process flow probably won’t help.  However, in the second situation (bottom), the cycle time is highly variable. Often this type of pattern is indicative of WIP bubbles.  Having more wafer inspection steps in the process flow both reduces the number of lots at risk, and may also help reduce the cycle time by smoothing out the lot arrival rate.

 

Because of its rich benefits, reducing cycle time is nearly always a value-added activity. However, reducing cycle time by eliminating inspection steps may be a short-sighted approach for three important reasons. First, only a small percentage of lots actually go through inspection points, so the cycle time improvement may be minimal. Second, the potential yield loss that results from having fewer inspection points typically has a much greater financial impact than that realized by shorter cycle time. Third, reducing the number of inspection points often increases the number and size of WIP bubbles. 

For further discussions on this topic, please explore the references listed at the end of the article, or contact the first author.

Doug Sutherland, Ph.D., is a principal scientist and Rebecca Howland, Ph.D., is a senior director in the corporate group at KLA-Tencor.

Check out other Process Watch articles: “The Dangerous Disappearing Defect,” “Skewing the Defect Pareto,” “Bigger and Better Wafers,” “Taming the Overlay Beast,” “A Clean, Well-Lighted Reticle,” “Breaking Parametric Correlation,” “Cycle Time’s Paradoxical Relationship to Yield,” and “The Gleam of Well-Polished Sapphire.”

References

1.       David W. Price and Doug Sutherland, “The Impact of Wafer Inspection on Fab Cycle Time,” Future Technology and Challenges Forum, SEMICON West, 2007.

2.       Peter Gaboury, “Equipment Process Time Variability: Cycle Time Impacts,” Future Fab International. Volume 11 (6/29/2001).  

3.       Fab-Time, Inc.  “Cycle Time Management for Wafer Fabs:  Technical Library and Tutorial.”

4.       W.J. Hopp and M.L. Spearman, “Factory Physics,” McGraw-Hill, 2001, p 325.

December 6, 2012 – KLA-Tencor says its new fourth-generation LED wafer inspection system achieves greater flexibility, increased throughput, and improved efficiency for inspecting defects and performing 2D metrology in LED applications, as well as MEMS and semiconductor wafers (up to 200mm).

The ICOS WI-2280, built on the company’s WI-22xx platform, supports handling of whole wafers in carriers and diced wafers in hoop ring or film frame carriers, to accommodate multiple media with minimal equipment changeover. An enhanced rule-based binning defect classification and recipe qualification engine enable faster yield learning during production ramps, and improved process control and process tool monitoring strategies. Highly flexible advanced optical modules with dedicated image processing enable high defect capture rate and recipe robustness against varying process background. A frontend-to-backend-of-line connectivity analysis capability — working in conjunction with the company’s Candela LED unpatterned wafer inspection system and Klarity LED automated analysis and defect data management system — delivers a single platform for defect source analysis.

"Increasingly, LED manufacturers are demanding improved detection and classification of yield relevant defects of interest, which enables them to take faster corrective actions to improve their yields at higher inspection throughput. There is also a growing need to boost productivity by enabling faster production recipe creation," stated Jeff Donnelly, group VP for growth and emerging markets at KLA-Tencor. The ICOS WI-2280 "ultimately enabl[es] LED manufacturers to achieve better lumens per watt and lumens per dollar performance."

In addition to LED manufacturing, the system can work in MEMS, semiconductor and compound semiconductor, and power device applications (wafers spanning 2-8 in.), the company says: backend-of-line and post-dicing outgoing quality control or binning; frontend-of-line patterned wafer inspection for baseline yield improvement, rework, excursion control, or overlay; and 2D surface inspection and metrology.

EV Group has completed its expanded cleanroom IV facility at its corporate headquarters in Austria, which doubled its cleanroom space for process development and pilot production services.

As part of the company’s long-term growth strategy to address high-volume tool orders and speed time to market, EV Group, a supplier of wafer bonding and lithography equipment, also increased the size of its application labs, added new R&D facilities for internal tool development and testing, and opened a new customer and employee training center.

The customer and employee training center provides several new rooms for instructional training courses, as well as a large number of manual and automated EVG tools for training.

While manufacturing and product development are centralized at EV Group’s corporate headquarters, technology and process development teams in Austria work closely with the company’s subsidiaries in Tempe, AZ; Albany, NY; Yokohama and Fukuoka, Japan; Seoul, South Korea; and Chung-Li, Taiwan, where additional, state-of-the-art application labs and cleanroom facilities are available.

Earlier this year, the addition of an ultra-modern manufacturing facility that doubled the production floor space marked the completion of the first phase of EVG’s long term expansion plans. Already positively contributing to EVG’s growth from the beginning of 2012, the company increased its order intake in FY12 (ended September 30) by 5 percent over fiscal 2011, and increased its revenue by 20 percent within the same period.

November 27, 2012 – LEDs have struggled to gain a foothold in the marketplace for indoor lighting applications, but technology improvements and supportive legislation are gathering momentum to help push LED adoption for residential buildings — the largest lighting application sector.

Global sales of LEDs for lighting applications totaled $3.57B in 2011, and should surge to $23.24B by 2018, calculates Frost & Sullivan. Behind that swell is "legislation that will essentially phase out incandescent lighting and other inefficient lighting technologies," as well as declining prices for LEDs that will boost demand and penetration of LED technology across multiple lighting applications, explains Frost & Sullivan industry analyst Hammam Ahmed.

The European Union has been an early adopter of legislation supporting a shift away from both manufacturing and sales of incandescent lighting; this legislation, though coming in multiple phases, has been echoed with similar policies sprouting up and implemented in various other countries (US, Switzerland, Canada, Australia). In Asia, Japan, China, Taiwan, and Korea are adopting LED-supportive legislation including financial incentives for both consumers and manufacturers.

Total global LED lighting market (2011), percent LED revenue by region (left)
and application (right). All figures are rounded. (Source: Frost & Sullivan)

Key factors limiting LED penetration into general lighting applications are pricing and technology improvements, but sharp and continued price declines should speed up the tipping point of price parity with other lighting technologies by the end of this decade, Frost & Sullivan says.

On the other side of that coin, manufacturers continue to improve lumens/dollar by pushing R&D and improvements in brightness, design, and quality of components, Hammam notes — though he admits "it remains to be seen how customers receive these new product developments." Additionally, those same relentless price declines are forcing manufacturers to come up with sustainable, long-term growth plans. "Participants from Eastern Asia, who have the ability to compete on prices, need to address quality issues to expand into the more developed markets of North America and Europe," he noted, while current market leaders "need to offer high-quality products and explore avenues for reducing cost of production."

November 23, 2012 – Growth in the industrial electronics semiconductor market is set to fall short of previous expectations in 2012 as the business is buffeted by weakening global economic conditions, with the LED market the sole bright spot, says IHS iSuppli in an updated report.

In general, revenue for industrial semiconductors — used in a wide array of application markets from home automation to aeronautics and military purposes — is projected to rise just 3% in 2012 to $31.4 billion — that’s less than half than the 7.7% growth forecast back in July. It’s also a meager expansion compared with 2011’s solid 9% increase and the exuberant 35% surge in 2010 immediately after the recession. For the next four years, revenue is set to rise in a range from 7-12% each year, reaching $44.8B by 2016.

Worldwide industrial electronics semiconductor revenue, in US $B. (Source: IHS iSuppli)

"Economic headwinds" started intensifying in 2Q12 and undercut chip revenue forecasts, affecting top semiconductor suppliers and OEMs of industrial electronics, explained Jacobo Carrasco-Heres, industrial electronics analyst at IHS. "And when hoped-for growth did not pan out as expected, and sales eventually came out lower, the market was downgraded to reflect the changed circumstances."

One segment that seems to have remained untouched this year is the robust LED market, thanks to the LED lighting boom that has taken hold in many parts of the world, noted Robbie Galoso, principal analyst for electronics at IHS. Philips enjoyed a 37% climb in LED sales in 2Q12 vs. a year-ago, and other LED lamp suppliers like Cree, LG Innotek, and Samsung LED also enjoyed solid 2Q results.

Industrial semiconductors are used in energy generation and distribution; military and civil aerospace; building and home control; medical electronics; manufacturing and process automation; and the test and measurement segment.

November 19, 2012 – Researchers from Rice U. say they have developed a micron-scale spatial light modulator (SLM) built on SOI that runs orders-of-magnitude faster than its siblings used in sensing and imaging devices. The "antenna-on-a-chip for light modulation," developed with backing from the Air Force Office of Scientific Research, is described in Nature‘s Scientific Reports.

While light processing has found use in consumer electronics (CDs and DVDs), communications (fiber optics), of course lighting applications (LEDs) and even industrial materials processing (lasers for cutting, welding, etc.), photonics for computing applications are still being explored, and reliant upon waveguides in 2D space. So-called "free space" spatial light modulators (SLM), however, could tap into "the massive multiplexing capability of optics," in that "multiple light beams can propagate in the same space without affecting each other," explains researcher Qianfan Xu.

To demonstrate, the Rice team built SLM chips with nanoscale ribs of crystalline silicon surrounded by SiO2 claddings, forming a cavity between positively and negatively dopes Si connected to metallic electrodes. The positions of the ribs are subject to nanoscale "perturbations" and tune the resonating cavity to couple with incident light outside. This coupling pulls incident light into the cavity; infrared light passes through silicon but is captured by the SML and can be manipulated to the chip on the other side, with electrodes’ field switched on/off at very high speeds.

In the paper they go into more detail on the structure of the device:

SLMs are fabricated in a CMOS photonics foundry at the Institute of Microelectronics of Singapore. The fabrication starts on an SOI wafer with a 220nm-thick silicon layer and a 3μm-thick buried oxide layer. To construct the 1D PhC cavities, silicon ribs with the height of 170nm are patterned on a silicon slab with the thickness of 50nm using 248nm deep-UV lithography and inductively-coupled plasma etching. Following the etching, the p-i-n junctions are formed by patterned ion implantations with a dosage of 5 × 1014 cm-2 for both the p+ and n+ doping regions. A 2.1μm-thick SiO2 layer is then deposited onto the wafer using plasma-enhanced chemical vapor deposition (PECVD). Finally, vias are opened on the ion-implanted areas and a 1.5μm-thick aluminum layer is sputtered and etched to form the electric connections. The serial resistance of the diode is measured to be 105 Ω. After the fabrication process, the contact pads connecting to the p-i-n junction are wire-bonded to a SMA connector with a 50-ohms terminal resistor for impedance match.

The 3D FDTD simulations are done with commercial software Lumermical FDTD. A non-uniform grid is used which has a spatial resolution ~30nm around the resonator. Even though perturbation we introduced is much smaller than the grid size, the software is capable of incorporate that in the simulation. When a dielectric interface (Si/SiO2) lies between two grid points, the program modifies the dielectric constant at the neighboring grid points according to the position of interface. This way, the small shift of the dielectric interface due to the width perturbation is taken into account in the simulation.

Conventional integrated photonics incorporate an array of pixels whose transmission can be manipulated at very high speed, explains Xu; adding an optical beam can change the intensity or phase of the exiting light. In LED screens and micromirror arrays in projectors (both of which are SLMs) where each pixel changes the intensity of light which generates an image, some switching speeds can get down to microseconds, but that’s far too slow for moving data around in a computing application. The new Rice device can "potentially modulate a signal at more than 10 gigabits per second."

Another key to their device is that it is silicon-based and can be fabricated at volume in a CMOS fab, which can scale up the capabilities to build very large arrays with high yield, he adds. For example, Rice researchers are separately creating a single-pixel camera, which initially took eight hours to process an image; this new SLM chip could enable it to handle real-time video. Alternatively, a million-pixel array could mean "a million channels of data throughput in your system, with all this signal processing in parallel" and at gigahertz levels, he said.

Xu is careful to note that the new SLM antenna-on-a-chip is not for general computing, but more for optical processing comparable in power to supercomputers. Optical information processing is " not fast-developing right now like plasmonics, nanophotonics, those areas," he admits, "but I hope our device can put some excitement back into that field."

Left: An illustration showing the design of Rice University researchers’ antenna-on-a-chip for spatial light modulation. The chip couples with incident light and makes possible the manipulation of infrared light at very high speeds for signal processing and other optical applications. Right: Crystalline silicon sits between two electrodes in the antenna-on-a-chip.  (Credit: Xu Group/Rice University)

Visit the MEMS Channel of Solid State Technology, and subscribe to our MEMS Direct e-newsletter!

November 16, 2012 – The latest monthly numbers are in for semiconductor manufacturing equipment demand, and they’re not pretty: lows in both orders and sales not seen since the last major downcycle three years ago, and the short-term comparisons continue to widen.

North America-based manufacturers of semiconductor manufacturing equipment reported bookings (orders) of just $743.2M in October, down -18% from September and roughly -20% from a year ago. Billings (sales) came in at $986.5M, off by -15% M/M and nearly -22% Y/Y. (Both are three-month moving averages.) SEMI also revised downward its September data: Bookings lowered to $912.8M (they had been $952.9M), and billings down to $1164.4M (vs. $1177.4M). The book-to-bill (B:B) came in at an anemic 0.75, meaning that $75 worth of orders came in for every $100 shipped out. (A B:B above 1.0 would indicate a good sign of more business coming in; a number below 1.0 means the opposite, and a number substantially below 1.0 and sinking for a while, well…)

Here are some chilling metrics to illustrate just how sour the market for chip tools has become as we head to the finish line of 2012. (All data is compiled from SEMI’s historical tallies dating from Jan. 1991)

  • Bookings are at their lowest point since October 2009. Billings haven’t been this low since January 2010. Since peaking in May, equipment bookings have been slashed by half (-54%) and sales are off by more than a third (-36%).

  • For the ten months through October, equipment orders were tracking down -8.5% from the same period in 2011 to $12.6B, and sales were down -15.7% at $13.3B.

  • Bookings have declined by double-digits for five consecutive months (-11% to -18%), which hasn’t happened since the grand old days of December 2000-April 2001. Except for a single month of mathematically zero growth (April), bookings have declined Y/Y for 16 out of the past 17 months. (This might say more about the industry’s reliably brutal cyclicality than current malaise; May 2011 was the end of a 19-month period in the black, which was preceded by a 29-month trip through the doldrums.)

  • The B:B ratio has been in freefall since April when it was well above the parity level (1.12) — that’s six straight months of decline, which according to SEMI’s data hasn’t happened since late 2010. (We’ve had several five-month slides in the past two years.)

Denny McGuirk, president and CEO of SEMI, labeled the environment for semiconductor industry investments as "muted" entering the final quarter of 2012, though he stated that "investments in leading-edge technologies will continue to drive spending in the near-term." The outlook for 2013 will clear up shortly as chipmakers crystallize their 2013 capex plans, he added. (Note that with about six weeks remaining, any lack of clarity into 2013 planning doesn’t exactly inspire confidence.)

SEMI will present its updated consensus forecast in conjunction with SEMICON Japan on Tuesday Dec 4 (technically it’ll be 11am local time, which is the wee hours late Monday/early Tuesday morning here in the US). One can reasonably expect some drastically different numbers from its current official forecast, issued at SEMICON West in July, which predicted an overall -2.6% decline for the year in global frontend + backend equipment. Hopefully there will be some improved clarity in these coming weeks.