Embed Size (px)
Citation preview
COMPOUNDSEMICONDUCTOR
April 2006 Volume 12 Number 3
C O N N E C T I N G T H E C O M P O U N D S E M I C O N D U C T O R C O M M U N I T Y
Going to Vancouver?Find the right boothwith our guide p36
MAPPING OUT MANTECH TECHNOLOGY INTERVIEW
Switched onWarm-white LEDs from Cree and Nichia get into residentiallighting. p5, 32
China crisis?Not according to industryveteran Phil Yin, who is busyreversing the fortunes ofsubstrate vendor AXT. p18
BiFETs
Anadigics’ strategypays off as consumersget the 3G bug
SiC SUBSTR ATES • SiC EPITA X Y • GaN EPITA X Y
©2006 INTRINSIC Semiconductor Corporation
At INTRINSIC, we understand that stability is what guides device manufac-
turers to success. A stable and reliable supply of high quality substrates
and epi sets a course for repeatable downstream processes and reduced
variability. A durable partner with an eye on the long term helps you get
results today, and lets you focus your resources on other critical areas.
If you’re looking for those qualities, you’re looking for INTRINSIC.
Call us today at 1.703.437.4000, or e-mail us at [email protected].
We want to serve as your materials and solutions partner.
Visit www.intrinsicsemi.com to learn more.
The Measure of Quality.
INTRINSIC offers wafers in 2 , 3 , and 4 sizes.
Navigating wide bandgap technology? Chart your course with INTRINSIC.
A P R I L 2 0 0 6 V O L U M E 1 2 N U M B E R 3
C O N N E C T I N G T H E
C O M P O U N D
S E M I C O N D U C T O R
C O M M U N I T Y
Compound Semiconductor April 2006 compoundsemiconductor.net 1
TECHNOLOGY
20 SiC growth rockets with hydrogen chloride addition:The standard growth process for SiC produces high-blocking-voltage devices, but suffers from long processing times. These delays are hindering commercialization, says Francesco La Via, who believes that the problem can be overcome by adding hydrogen chloride into the cell.
23 Materials Update: Crystal growers push nitride envelopeMichael Hatcher reports on a flurry of recent development activity in III-N materials that has seen numerous funding deals and the commercial shipment of AlN substrates.
25 Bespoke MMICs aid radar, phased array and oscilloscope applications: High-performance traveling-wave amplifiers for oscilloscopes, radar and phased-array equipment can be made cheaper and simpler by switching to custom-built designs, claim Mark Kelly and Thorsten Brabetz.
28 Analysis tool helps project managers expose hidden pitfalls:Project management specialist Tal Levanon reveals how a new analysis tool can reduce cleanroom expansion costs at a semiconductor wafer fab by 25%.
30 Equipment Update: Electroluminescence mapping tool accelerates LED development MaxMile Technologies has launched the first non-destructive electroluminescence mapper for unprocessed LED epiwafers.
32 Chip makers and lighting specialists get switched on to new possibilities: High-brightness LED manufacturers andtheir customers in the general lighting industry now seem to be on the same wavelength, discovers Michael Hatcher.
36 CS Mantech 2006: Exhibition Guide.
38 Classified: Incorporating the Product Showcase.
40 Research Review: C-band VCSELs provide cheaperalternative for networks...BeZnO films suggest improved devices...GaN-based amplifier delivers record power.
Customer focusThe traveling-wave amplifier MMICs made at ECIT in Belfast can be fine-tunedto individual requirements. p25
Compound Semiconductor’s circulation figures are audited by BPA International
INDUSTRY
5 Headline News: White LED makers get residential boost...Jump in orders at disciplined Aixtron suggests possible recovery...Sematech goes back to the future with plans for germanium channels.
6 The Month in RFICs: RFMD set for huge volume ramp...Demand for HBTs helps Kopin to record profits.
8 The Month in HB-LEDs: DoE to fill technology gap... Nichia claims lm/W breakthrough...Epistar orders five of Aixtron’s reactors...Osram enhances Power TopLEDs.
10 The Month in Optoelectronics: Yokogawa and Fujitsu hook up on 40G technologies...Optical telecom components: sales up 11% in market recovery...Vertilas eyes expansion with new funds...OFC show goes with a swing...Sony delays launch of PlayStation 3.
15 Behind the Headlines: Anadigics charge hinges on BiFETs Wafer-level integration of InGaP HBT and PHEMTstructures is at the heart of Anadigics’strategy to return to profitability, hears Michael Hatcher.
16 Portfolio: Finisar and Avanex spring into lifeMassive trading volumes and a rapid upturn in the valuations of Finisar and Avanex coincided with the year’s biggest fiber-optic components show.
18 Interview: Changing fortunes in the year of the dogDealing with a legacy of material quality problems, management issues and a plunge in market share doesn’t sound like most people’s idea of fun, but that didn’t stop Phil Yin taking over the reins at substrate supplier AXT.
Switched onHas the time come for LEDs to light up our homes? p5
Sony gets the bluesProblems with the Blu-ray drive havedelayed the launch of Sony’s PlayStation 3until the end of the year. p11
Main cover image: After introducing its BiFET process back in 2003,Anadigics is reaping the rewards with increased fab utilization and ismoving towards profitability. Credit: Anadigics.
compoundsemiconductor.net April 2006 Compound Semiconductor2
E D I T O R I A L
LEDs in the home: is now the time?It’s been called the “holy grail” for the HB-LED and compoundsemiconductor community, and it might be coming sooner thanyou think. As soon as this summer, in fact.
While great strides have been made with white LEDs forgeneral illumination, small chips are still widely regarded asbeing too dim, while the much larger, brighter “power LEDs”
produce a lot of heat that is difficult to dissipate. Both are thought to be justtoo damned expensive for general lighting applications at the moment. As aresult, the received wisdom is that LED-based lighting of the home is still a fairdistance away – anything between two and10 years, depending on who you ask.
These are just some of the reasons why theUS Department of Energy is still ploughingcash into solid-state lighting research,recently approving 16 different R&D projects, including the developmentof quantum-dot structures, photonic crystals, and novel GaN alloysincorporating scandium or yttrium.
But is all this extra development really necessary? Progress Lightingseems to think not, and to feel that HB-LED technology is ready for prime-time right now. This summer the company, which touts itself as thelargest source for residential lighting in the US, is launching what it calls acomplete line of LED-based products under the HI-EFTM brand.
It may not exactly be the snappiest of names, but if the technology catches on with the US public it will be great news for Nichia and Cree, the companies that are supplying the GaN-based chips that lie at the heart of these fixtures.
While the launch of these products could provide a major boost formakers of white LED chips, one note of caution should be urged. If it comestoo soon, it may cause longer-term damage to the nascent SSL industry.Nothing creates bad publicity in the consumer sector quite like productsthat fail to live up to expectations, and that can kill a new technology stonedead. Progress needs to make sure that that doesn’t happen with LEDs.
Michael Hatcher Editor
“The US Department ofEnergy is still ploughingcash into solid-statelighting research.”
Air Products & Chemicals Inc 31AXT 12Bandwidth Semiconductor 8EuroLED 2006 27Freiberger 17Indium Corporation of America 7Instrument Systems 34IntelliEpi 7INTRINSIC Semiconductor IFCISCS 2006 19Jemi UK 34k-Space Associates 22
KLA-Tencor 14LayTec GmbH 9Proton Energy Systems 6Raboutet 11Riber 39SAES Pure Gas 17Shiva Technologies 17Soitec IBCSurface Technology Systems 35Tecdia 4Veeco 3, OBC
Advertisers’ Index
Editor Michael [email protected]: +44 117 930 1013. Fax: +44 117 925 1942
Features editor Richard [email protected]: +44 117 930 1192
Consulting editor Tim [email protected]: +44 117 930 1233
Senior sales executive David [email protected]: +44 117 930 1032. Fax: +44 117 920 0977
Business development manager Molly [email protected]: +1 215 627 0880. Fax: +1 215 627 0879
Circulation manager Jackie [email protected]: +44 117 930 1218. Fax +44 117 930 1178
Publisher Sarah [email protected]: +44 117 930 1020
Senior production editor Ruth LeopoldAd production Joanne Derrick, Mark TrimnellArt director Andrew GiaquintoTechnical illustrator Alison ToveyPublishing director Richard Roe
SubscriptionsAvailable free of charge to qualifying individualsworking at compound semiconductor fabs andfoundries. For further information visitcompoundsemiconductor.net/subscribe. Subscriptionsfor individuals not meeting qualifying criteria:individual £86/$155 US/7125; library £193/$348US/7280. Orders to Compound Semiconductor, WDIS, Units 12 & 13, Cranleigh Gardens IndustrialEstate, Southall, Middlesex UB1 2DB, UK. Tel: +44 208 606 7518; Fax: +44 208 606 7303. General enquiries: [email protected].
9173 average total qualified circulation**December 2005 BPA audit statement
Editorial boardMayank Bulsara Atlas Technology (USA); Andrew Carter Bookham Technology (UK); Jacob TarnEpistar/Gigacomm (Taiwan); Ian Ferguson GeorgiaInstitute of Technology (USA); Toby Strite JDSU(USA); Mark Wilson Motorola (USA); Dwight StreitNorthrop Grumman (USA); Joseph Smart Crystal IS(USA); Colombo Bolognesi Simon Fraser University(Canada); Shuji Nakamura University of California atSanta Barbara (USA)
©2006 IOP Publishing Ltd. All rights reserved.
US mailing information: Compound Semiconductor(ISSN 1096-598X) is published 11 times a year for $148 by Institute of Physics Publishing, Dirac House,Temple Back, Bristol BS1 6BE, UK. Periodicalspostage paid at Middlesex, NJ 08846. POSTMASTER: send address corrections toCompound Semiconductor, c/o PO Box 177,Middlesex, NJ 08846. US agent: Pronto MailersAssociation Inc, 200 Wood Avenue, PO Box 177,Middlesex, NJ 08846.
Advance your MBE research andproduction—even while you sleep
Solutions for a nanoscale world.™
For research and pilot-production, there’s no better system than the Veeco GEN20. Incorporating
industry-leading attributes from the GEN II and the GEN200, this all-new model, in a proven
platform, manufactures wafers up to 4” with multi-chamber processing capabilities. It seamlessly
integrates application-specific design features for both existing and emerging materials, including
nitrides and oxides. Available either in automated cluster tool or manual models, the GEN20
is the best fit for you, providing flexibility to meet a wide range of budgets. For more
information, visit www.veeco.com/gen20 or call 1.888.24.VEECO.
The new Veeco GEN20: application specificdesigns, optional unattended operation.
Compound SemiconductorSolutions
MBE, MOCVD, PVD, IBE AFM and Optical Metrology Process Integration Center
Photo of Veeco Silicon-Style Cluster Tool.
Sematech goes back tothe future with plans forgermanium channels
C O N V E R G E N C E
Compound Semiconductor April 2006 compoundsemiconductor.net 5
I NDUSTRY H E A D L I N E N E W S
LED-based lighting is set for a boost in the US,with leading supplier Progress Lighting gear-ing up for a summer 2006 launch of productsfeaturing chips made by Cree and Nichia.
The two chip companies will supply 40lm/Wwarm-white LED sources to Permlight Prod-ucts, which has developed a low-cost thermalmanagement solution for residential applica-tions that it sells to Progress. Permlight alreadyuses Nichia die for ceiling lights and Cree chipsfor “below-the-eye” applications. Permlightsays that this is because Nichia products arebetter suited to illuminating people’s faces.
“Progress Lighting believes now is the timefor LED systems to illuminate our homes, nottwo to five years from now like most industry
analysts predict,” said Jim Decker, vice-president of brand management at ProgressLighting. “Progress will offer a home lightingfixture solution that incorporates warm-colordimmable LED lighting fixtures that appearand operate like incandescent sources. Progresswill be using Permlight’s breakthrough LEDtechnology as the core of this new product line.”
Permlight CEO Manuel Lynch said that theproduct launch “will be much sooner than later,and a much larger offering than most wouldexpect. We have demonstrated repeatedly thatthe time for LEDs to light our homes is now.This is the light source from now on.” For more information, see the HB-LED/residential lighting feature on p32.
White LED makers get residential boost
Sematech, the collaborative semiconductorresearch and development partnership, haslaunched a project to investigate alternativematerials for metal oxide silicon field effecttransistor (MOSFET) channels, the criticalpathways in silicon-based transistors.
The project will focus on germanium andSiGe primarily, while III-V compounds couldfeature in the longer term. Sematech’s front-end processes (FEP) division will be at theforefront of the investigation and will aim todevelop process technology that increaseschannel mobility beyond silicon’s physicallimits without compromising on reliability.
FEP associate director Larry Larson said:“SiGe and germanium are currently the mostpromising materials for replacing silicon inplanar channels in CMOS manufacturing.”Larson described the initiative as a major neweffort that could have far-reaching implica-tions for the semiconductor industry.
Freescale’s Hsing-Huang Tseng will man-age FEP’s new program. “Continued CMOSscaling means we need to find new materials toreplace silicon in these functions,” he explained.“Germanium, SiGe and potentially III-Vcom-pounds could provide even bigger benefits formobility enhancement.” Freescale already hasplans to develop a GaAs-based MOSFET forcommercial deployment, which it announceddetails of recently, while strained-siliconspecialist Amberwave and Purdue Universityare also teaming up on III-V integration in aproject headed up by Peide Ye at Purdue.
Switched on: Progress Lighting believes that now isthe time for LEDs to be used to light up homes.
After a tough couple of years, a sharp increasein orders may signal a better 2006 for equip-ment vendor Aixtron. The industry’s keysupplier of metal organic chemical vapor depo-sition (MOCVD) machines is hoping to breakeven in 2006, after recording a huge financialloss in 2005.
Aixtron, which believes it has a 60% shareof the MOCVD equipment market, returnedan overall net loss of 753.5million ($64.4mil-lion) in its 2005 financial year. While much ofthat figure can be attributed to balance sheetadjustments that CEO Paul Hyland describedas financial prudence, Aixtron said that its netloss would have been 714.6million if recordedusing standard US accounting methods.
Hyland added that “light”demand from cus-tomers, particularly LED manufacturers, hadmade the last two years among the most diffi-cult in Aixtron’s history – at a time when it hasexperienced the added complication of con-solidating its acquisition of the atomic vapordeposition company Genus.
Overall, Aixtron’s revenue remained flat ataround 7140 million from 2004 to 2005.However, this masked a big decrease in annualMOCVD reactor sales from 7114.8 million to780.9 million during the period.
Revenue from the new Genus business,which was negligible in 2004, accounted for23% of sales in 2005, or 732.1 million.
Aixtron’s disciplined response to the slackdemand for its core products has been to cutstaff numbers, introduce temporary plant clo-sures and more common platforms, as well as
to delay certain research projects. Those mea-sures should have an impact on costs over thenext year, during which Aixtron expects tobreak even on sales of around 7150 million.
Though keen to remain cautious, Hylandpointed to a sudden jump in order intake in thefinal quarter of 2005 as a possible indicationof a better year to come.
At 737.6 million, orders in the final quar-ter were double those of the equivalent three-month period in the previous year. Thosefigures were not influenced by the recent orderfor five MOCVD machines placed by Taiwan-based LED maker Epistar (see p8), which isbeing recorded in the first quarter of 2006.
Hyland and his colleagues will be hopingthat the improving business environment is nota repeat of the “false dawn” witnessed in early2004, when orders reached a similar peakbefore slumping back down.
Jump in orders at disciplinedAixtron suggests possible recovery
COLOR KINETICS
S O L I D - S T A T E L I G H T I N G
On the up? Aixtron CEO Paul Hyland is staying cautiousabout 2006, despite a sudden upturn in orders.
E Q U I P M E N T
AIXT
RON
INDUSTRY T H E M O N T H I N R F I C S
RFMD set for huge volume rampLeading GaAs chip manufacturer RF MicroDevices (RFMD) is to increase the wafercapacity of its Greensboro, NC, fab by 40%with an investment of $80 million.
The move comes amid rising demand forRFMD’s power amplifiers, which are widelydeployed in mobile handsets. Sales of ad-vanced handsets, which require multiple poweramplifiers and more GaAs content, are risingparticularly quickly as 3G services becomemore popular.
“RFMD is making this strategic investmentto capitalize on the rapid increase in world-wide demand for GaAs technology,” saidcompany CEO Bob Bruggeworth. “We are theleader in wideband-CDMApower amplifiers,and expect wideband CDMAto be the world’sfastest growing air-interface standard.”
RFMD has a history of making largeinvestments in GaAs capacity, and this is thefifth such expansion in its 15-year history.“RFMD views its continued investment inmanufacturing capacity as key to its revenueand earnings growth,” said the company in a statement.
Having already begun the expansionthrough an investment in additional process-ing equipment at its existing cleanroom facil-ities, RFMD says that volume production withthe new kit will kick off in late 2006. It expectsto add 300 new positions once the expansionis completed.
The company has received a $4.9 millionhandout from the State of North Carolinathrough a job development incentive grant,and local county governments have pledged$1.5 million to the same end.
Industry forecasts suggest that more than900 million mobile handsets will be shippedthis year, representing a 15% increase on the2005 figure. One-hundred million of those areexpected to be 3G handsets, while RFMD saysthat the growing market for wireless LANcompatibility in handsets is increasing GaAsdemand still further.
The capacity expansion is expected toreduce RFMD’s overall cost per wafer, as itshifts to in-house production of the GaAsPHEMTs that are critical to its transmitmodule products.
Kopin Corporation generated a record annualprofit of $11.7 million for fiscal 2005, com-pared with a $13.8 million loss for the previ-ous year. The US-based manufacturer of HBTepiwafers, LEDs and microdisplays postedsales of $24.9 million for the fourth fiscalquarter, and a profit of $3.7 million.
Demand for HBTs increased revenue atKopin’s III-V products division, with salesworth $13.5 million in the recent quarter, up$2.7 million sequentially and $4.9 millionyear-on-year.
According to CEO John Fan, the increase inHBT revenue has been driven by an uptick insales of cell phones with greater functionalitythat require multiple power amplifiers.
“With an 80% share of the merchant HBTmarket, Kopin benefits from this trend becausethe move towards larger and more advancedpower amplifiers increases the amount of ourtransistor content per phone,” added Fan. TheIII-V division is expected to continue toproduce steady sales growth based on thestrength of orders.
Demand for HBTs helpsKopin to record profits
F I N A N C I A L R E S U LT SF A B E X P A N S I O N
INDUSTRY T H E M O N T H I N R F I C S
Compound Semiconductor April 2006 compoundsemiconductor.net 7
…Freescale clarificationIn our March issue we reported that Motorola’ssemiconductor products sector (SPS), nowknown as Freescale, had developed GaAs-on-silicon wafers in 2001. It has since been pointedout that this is not correct. The GaAs-on-silicondevelopment took place at Motorola’s CorporateResearch Lab in Phoenix (now known as theEmbedded Systems and Physical SciencesLaboratory), which was not part of the SPS or theFreescale spin-off. We are happy to clarify this.
…Nokia raises phone forecastLeading cell-phone vendor Nokia has predictedthat well over 900 million handsets will be sold in2006, after revising its growth forecast upwards.CEO Jarmo Ollila said at Nokia’s AGM in Marchthat total shipments in 2006 would be at least15% higher than the 2005 figure of 795 million.
…new center for PeregrineSilicon-on-sapphire RFIC specialist PeregrineSemiconductor has opened its third engineeringdesign center in Nashua, New Hampshire.
…Anadigics raises $53 millionSix-inch GaAs fab owner Anadigics has raised$53.4 million through a public offering of itscommon stock. Most of the proceeds will beused to pay off convertible notes due to maturein November this year. The remaining$6.7 million will be used for general corporatepurposes, including some capital expenditure.Anadigics has enjoyed improved financialresults recently and has seen its share value risesteadily from just less than $2 to more than $7in the past year.
…Sirenza bullish on demandUS-based RF specialist Sirenza Microdevices hasraised its first fiscal quarter 2006 sales guidancefrom $17.5–19 million to $20–21 million. “Ourupdated range represents sequential growth ofup to 8% over our fourth-quarter 2005 recordresults and a potential quarterly revenueincrease of up to 73% year-over-year,” remarkedcompany CEO Robert Van Buskirk.
…WJ and GCS ink supply dealUS foundry Global CommunicationSemiconductors (GCS) has agreed to act as asecond wafer source for RF chip maker WJCommunications. “We’re pleased to add GCS asa foundry for InGaP HBT and GaAs technology,”
remarked Mark Knoch, vice-president of WJ’soperations. “This advantage provides ourcustomers with two qualified sources, therebyensuring a steady supply of wafers.”
…defect-free SiC debutIntrinsic Semiconductor, a privately heldmanufacturer of wide-bandgap materials anddevices, has commenced commercialshipments of SiC wafers completely free ofmicropipes, as well as 100 mm (4 inch)diameter SiC wafers.
…ST Systems bags £1million etch orderProcessing tool supplier ST Systems has bankedan order worth more than £1 million($1.74 million) from a GaAs chip manufacturer.The UK company, which designs and sellsplasma etch and deposition equipment, saysthat it has received multiple orders from the samecomponent manufacturer in North America.
…E-PHEMTs for 3G and WLANAvago Technologies has introduced two new RF products based on its GaAs E-PHEMTtransistors. The modules are designed for 3Gcellular infrastructure in the 1.8–2.2 GHzfrequency range, and dual-band wireless LAN applications.
From our Web pages...visit compoundsemiconductor.netfor daily news updates
compoundsemiconductor.net April 2006 Compound Semiconductor8
I NDUSTRY T H E M O N T H I N H B - L E D S
S O L I D - S T A T E L I G H T I N G
The US Department of Energy (DoE) hasselected 16 applications under a funding pro-gram aimed at developing core technologiesfor solid-state lighting (SSL).
Included in the selection are eight projectsfocused directly on III-V substrates, epitaxyand other processing techniques, with a strongemphasis on filling technology “gaps” such asdeveloping more efficient green high-bright-ness LEDs. Other projects selected include thedevelopment of organic LEDs and quantum-dot technologies.
“The present selections represent the sec-ond round in a series that may span the nextdecade,” said the National Energy TechnologyLaboratory (NETL), the DoE body throughwhich the development is organized. “[These]are expected to represent a significant advance-ment in the SSL technology base.”
III-V laboratories involved in the program
include the Georgia Tech Research Corpor-ation, which will work on a sacrificial substratetechnique to improve GaN LED efficiency,and Sandia National Laboratories, which willdevelop a growth technique for making largebulk GaN substrates in one of three projects(see box for details of all the selections relat-ing to III-V compounds).
The ultimate aim of the NETL program isto develop SSLtechnologies that support a sys-tem efficiency of 50% by 2025. The US uses21% of its total energy consumption on light-ing, a figure that the DoE wants to see halved.
Epitaxy equipment vendor Veeco Instru-ments has joined the solid-state lightingdevelopment center at University of Cali-fornia, Santa Barbara, which includes leadingnitride researchers like Steven DenBaars andShuji Nakamura.
DoE to fill technology gap
Georgia Tech Research Corporation A sacrificial substrate approach that will supporta number of complementary growth and proces-sing techniques to improve the overall quantumefficiency of GaN devices.
Purdue University Growth of hetero-epitaxial device structures onnanoscale substrates.
Rensselaer Polytechnic Institute A project that aims to fill the “green gap” byreducing the high dislocation densities in theactive region and the huge polarization fieldeffects throughout the LED structure. This shouldcut epi production costs and improve deep-greenLED efficiency by a factor of two to three.
Technologies and Devices International Materials research for producing highly dopedp-type GaN materials and device structures.
University of California, Santa Barbara Maximizing white LED efficiency by the use ofphotonic crystals to extract more light from theemitter structure.
Sandia National Laboratories Another project focused on deep-green LEDdevelopment, this time by using strain-relaxedInGaN templates, possibly through epitaxial lateralovergrowth.
A third project focused on deep-green LEDdevelopment, through the use of novel alloys suchas YGaN and ScGaN, which could lead to moreefficient quantum wells emitting throughout theentire visible range.
Development of a novel, scalable, cost-effectivegrowth technique to make low-dislocation-densitybulk GaN substrates, which will improve theinternal quantum efficiency of GaN devices.
Selected NETL projects (III-V related)
Nichia has developed a 6 lm white LED thatcan deliver an efficacy of 100 lm/W at a for-ward current of 20 mA. The company says thatit made a pendant lamp containing more than4000 devices and a spotlight with over 400.
“There wasn’t a big technical jump to reach100 lm/W,” explained Gen-inch Shinomiya,managing director of Nichia’s R&D engi-
neering division. “We improved our technol-ogy incrementally and I believe we can reachthe level of 150 lm/W in the same way.”
The Japanese company plans to begin sam-pling the 100 lm/W chip in June and will startvolume manufacturing later in the year. Thefirst product will be a lamp package measur-ing 5 mm in diameter.
Nichia claims lm/W breakthroughC H I P D E V E L O P M E N T
INDUSTRY T H E M O N T H I N H B - L E D S
Compound Semiconductor April 2006 compoundsemiconductor.net 9
Aixtron has received an order from Epistar forfive multi-wafer reactors designed for GaNLED growth. The sale comprises four AIX2600G3 reactors, along with Aixtron’s newAIX 2800G4 that is claimed to have the largestcapacity of any commercial reactor and canaccommodate up to 42 two-inch wafers.
The recent order will reinforce Epistar’sposition as the leading manufacturer of LEDsin Taiwan in terms of capacity, which wasestablished last year when it merged with UEC.“The additional tools will enable us to rampup production capacity in response to marketdemand with excellent cost-efficiency,” com-
mented Ming-Jiunn Jou, senior vice-presidentof Epistar.
Jou added that the company was lookingforward to being the first customer for theworld’s largest-capacity epitaxy reactor forGaN growth.
According to Aixtron, the AIX 2800G4 fea-tures a redesign of most of the components,including the cell; offers easier operation andhandling; and delivers improved reliability andreproducibility over previous designs.
Meanwhile, Samsung Electro MechanicsCo. (SEMCO) has placed an order withAixtron for another AIX 2600 G3 HTMOCVD system. Since its first reactorpurchase in 1997, SEMCO has continuallyincreased its installed base for the massproduction of a range of semiconductorsincluding GaN-based LEDs for consumeroptoelectronics. The latest order includes anEpiTune II in situ process monitoring tool.
Epistar orders fiveof Aixtron’s reactors
From our Web pages...visit compoundsemiconductor.net for daily news updates
…easy attachment with CreeChipmaker Cree has released its new “EZBright”LED platform. Said to be a major advance inchip technology, EZBright LEDs feature Cree’shighest brightness levels in a new, “easy-to-die-attach” chip. Cree’s brightest chips can now beassembled into packages using industry-standard processes, says the firm.
…JPSA “solves” contamination problemJ P Sercel Associates (JPSA) has begunshipping solid-state UV laser wafer-dicingsystems that are claimed to solve problemsassociated with GaAs and GaP contamination.
The Hollis, NH, firm says that it recentlyshipped multiple IX-300 “ChromaDice” UV laserwafer-processing systems to major GaAs wafer
producers, featuring JPSA’s new “ProtectoLED”Technology – a proprietary, patent-pendingsystem based on a water-soluble protectivecoating that is said to prevent contamination ofthe wafer during the dry-etch process.
…test probe for high throughputSUSS MicroTec and Instrument Systems aredeveloping a high-throughput test system forLED devices at wafer level. The Germancompanies say that it will enable manufacturersto test up to 70,000 LED dice per hour.
Thomas Attenberger, Instrument Systems’ LEDproduct manager, said: “With this innovativesolution, we are leveraging our combined know-how in LED testing to optimize throughput,achieving speeds never seen before.”
Osram Opto Semiconductors has enhancedits Power TopLED line of LEDs. In particular, theaddition of Osram’s advanced thin-film chiptechnology to these TopLED packages makesthem much brighter.
Osram has also refined the offerings in itslensed Power TopLED portfolio. These 30° and60° devices provide matched radiationpatterns in all colors, delivering higher-intensitylight in a more uniform and focused beam forapplications such as information displays.
The 30° Power TopLEDs produce 6300 mcdin super-red, 12,000 mcd in amber and6500 mcd in yellow. The blue and green PowerTopLEDs operate at a forward voltage of 3.4 V.This voltage is lower than that for conventionalInGaN devices, says Osram.O
SR
AM O
S
P R O C E S S I N G
LayTec GmbHHelmholtzstr. 13-14D-10587 Berlin · GermanyTel.: +49 (30) 39 800 80-0Fax: +49 (30) 31 80 82 37Email: [email protected] www.laytec.de
The new EpiCurve TT measures real-time wafer curvature.As the demand for larger compound wafers rises, bowing measurements become indispensable. LayTec´s new in situ monitor EpiCurve TTmeasures wafer curvature online during growth. Optimize your growth and maximise your yield with EpiCurve TT’s unique combina-tion of bowing, true temperatureand growth rate measurement!
Get the bows out
true temperature
bowing
compoundsemiconductor.net April 2006 Compound Semiconductor10
I NDUSTRY T H E M O N T H I N O P T O E L E C T R O N I C S
Yokogawa and Fujitsu hookup on 40G technologiesIII-V component manufacturer YokogawaElectric has signed a strategic agreement withFujitsu over the development of “ultra high-speed” communications networks. The net-works will be based on optical packet switchesthat are set to be made at Yokogawa’s huge newfabrication facility.
The two companies are planning to jointlydevelop the core technologies for faster opti-cal networks, with Yokogawa focused onadvanced compound semiconductor chips andFujitsu working on transmission systems.Yokogawa will supply Fujitsu with 40 Gbit/ssubsystem units.
“By combining Yokogawa’s advanced com-pound semiconductor device technology withFujitsu’s expertise in network systems, thecompanies aim to achieve a faster transmis-sion network infrastructure,” said Yokogawain a statement.
Last year, Yokogawa said that it planned tobuild a $230 million fabrication facility nearTokyo to make optical packet switches andsubsystems in volume. The all-optical switchcan route signals without the need for elec-tronic conversion.
Yokogawa told Compound Semiconductorthat construction of the 28,000 m2, five-storeyfab is going according to plan so far, with ini-tial production runs set for November.“Volume manufacturing is expected to startearly 2007,” said a company insider, addingthat it would have enough capacity to makearound one million modules per year.
The company is aiming to focus on localarea network applications with the new tech-nology. Yokogawa, which already has a globalnetwork comprising 18 manufacturing facili-ties, provides automation and control equip-ment to a wide range of customers includingmany in the power industry.
According to officials at the Japanese com-pany, the compound semiconductor fab underconstruction will also be used to make sensorcomponents that can be exploited in industrialcontrol systems.
But high-speed fiber-optics will be the maindriver. “It was obvious to anyone [at OFC/NFOEC] that the 40G market had finallystarted to pick up,” said a Yokogawa source,adding that the subsystems may also be usedfor transmission speeds of up to 100 Gbit/s.
Sales of optical components used in fiber-optictelecommunications grew by 11% in 2005 toreach $2.92 billion, according to new figuresfrom Ovum-RHK.
The market research company said thatwhile sales in the fourth quarter of 2005 wereslightly weaker than in the equivalent periodin 2004, the annual trend remained positive,driven by network build-out to support IPandvideo services.
“JDSU continues at the top of the heap, withFinisar and Sumitomo in second and third,respectively,” said Ovum-RHK. “Roundingout the top five are Avago and Bookham.”
The market recovery has had a positiveimpact on the financial health of the remain-ing companies in the ravaged sector. Averagegross margins at nine publicly-traded opticalcomponent suppliers were 29% in the finalquarter of 2005, representing a big improve-
ment compared with the third quarter.In particular, Finisar has been profitable in
its last two fiscal quarters (see Portfolio, p16).And while Bookham and Avanex are yet tobreak even, they have both looked to bolstertheir cash positions recently through publicofferings of stock. Bookham raised around$50million and Avanex aimed to raise $48mil-lion before costs.
“Optical component suppliers are happilysupplying arms to all equipment vendorsengaged in the network build-out,” said DarylInniss, VP of optical components at the mar-ket research company.
Inniss continued: “Growth is coming fromall segments, including carrier metro, long-haul and access networks, and enterprise.Component vendors [that are] best equippedto meet the demand stand to grow and increasemarket share.”
Optical telecom components:sales up 11% in market recovery
...InP targeted for low-cost PVA research collaboration based in the UK haswon £235,000 ($400,000) to develop low-cost photovoltaics based on InP substrates.
The project, which is being funded by theEngineering and Physical Sciences ResearchCouncil and the UK Department of Trade andIndustry, is headed up by an Oxford Universityresearch group. Substrate specialist WaferTechnology and the Centre for IntegratedPhotonics foundry are also involved.
According to principal investigator RobinNicholas, one of the key objectives of theproject is to develop three-terminal devicesbased on the InGaAs/InP material system,which would improve the efficiency of so-calledthermophotovoltaics (TPV).
Current TPV technology employs GaSbsubstrates, which are expensive and cover onlya portion of the useful wavelengths that TPV iscapable of exploiting.
...Advanced Photonix beats tax manUS-based optoelectronics manufacturerAdvanced Photonix is set to receive taxabatement from the City of Ann Arbor inconnection with the consolidation of its facilities.
Advanced Photonix is consolidating itsCalifornia and Wisconsin semiconductormicrofabrication operations into its Ann Arborfacility in Michigan, and is also moving its headoffice to that site. The company stands toreceive up to $94,000 over the five-yearabatement period.
...quantum dots for solar?Ryne Raffaelle and co-workers at theNanoPower Research Labs at RochesterInstitute of Technology have won a contract for$0.85 million from the US Department ofDefense to develop high-efficiency solar cellscontaining quantum dots.
The dots will be added to the central GaAs-based layer of triple junction solar cells, andwill produce an enhancement in short-circuitcurrent and an increase in overall efficiency.
“The use of quantum confinement offered bynanostructured materials provides us with anew means of breaking out of the normaldesign constraints associated with ordinarycrystalline device growth,” said Raffaelle.
...Bookham sells fab, buys AvalonBookham has raised $24 million from the sale and leaseback of its InP fab in Caswell, UK. And in an all-stock deal, the chipmaker has acquired Swiss VCSEL specialistAvalon Photonics.
From our Web pages...visit compoundsemiconductor.netfor daily news update
M A R K E T G R O W T H
S U P P LY D E A L
INDUSTRY T H E M O N T H I N O P T O E L E C T R O N I C S
Compound Semiconductor April 2006 compoundsemiconductor.net 11
German VCSEL specialist Vertilas has won$4.4 million in new venture funding. Includedin the latest round was new investor KfWMittelstandbank, while High Tech PrivateEquity GmbH also added to its previous invest-ment in the company.
Vertilas, which spun out of the WalterSchottky Institute (WSI) at the TechnicalUniversity of Munich back in 2001, says thatit will use the funds to “advance the commer-cialization” of its InP-based products. Thecompany leases an MBE machine that ishoused at WSI, where it makes about half amillion chips per year on 2 inch InP wafers.
The company’s current product rangeincludes singlemode uncooled VCSELs oper-ating at wavelengths of up to 2.3µm. “We havea promising technology with a great growthpotential,” said Vertilas managing directorGerald Vollnhals.
Vollnhals told Compound Semiconductorthat the new funds would be spent mostly onapplications and prototype engineering. Thecompany has no short-term plans to buy itsown manufacturing equipment, but Vollnhals
did say that if business increases as expected,Vertilas may look to set up its own manufac-turing operation around 2009.
The Vertilas VCSELs are used in industrialgas spectroscopy applications, mostly for mon-itoring levels of methane, carbon dioxide andhumidity. The semiconductor, food packag-ing, and petrochemical industries are majorcustomers. These gases all have molecularabsorption peaks in the infrared region cov-ered by the Vertilas VCSELs. Some of the newfunding will be used to improve the detectionlimits of the VCSEL sensors, by reducingbackground noise levels and the effect of inter-ference fringes in the device packaging.
Vertilas is also working to increase chipmanufacturing volumes, partly through a moveto 3 inch production, and partly through reduc-ing the size of its chips from their current500 × 500 µm dimensions.
Long-wavelength VCSELs can also be usedin datacom applications, and Vollnhals saysthat the emergence of fiber-to-the-home couldlead to significant demand for the devices inaround two years’ time.
As had been widely expected, Sony has put back the launch of its PlayStation 3 games consoleuntil November 2006. Problems relating to the GaN-laser-based Blu-ray Disc format are thought tobe at least partly to blame for the delay. The PlayStation 3 console will now be launchedsimultaneously in Japan, the US and Europe.
Analysts at Merrill Lynch commented in a recent research report: “It’s now clear that the [PS3]box is extremely expensive to make. In particular, we think the problem points are the Sony Cellprocessor and the Blu-ray drive.”
Whether or not the problem relates specifically to the fabrication of GaN lasers is unclear. Therehave been problems with GaN laser yields, but the Blu-ray copy protection software has also beena factor in the delayed launch of the technology.
Sony’s first Blu-ray Disc players and PCs featuring the concept are now expected to ship in Julyfor around $1000. In the meantime, Sony will be heavily marketing the signature blue color of thelaser component by installing blue-tinted windows in its Sony Style retail outlets across the US.
The Optical Society of America says that itsOFC/NFOEC 2006 event in early March wasa resounding success. One of the most impor-tant meetings in the fiber-optic community’s
calendar, the Anaheim show attracted 13,000registrants in total. “There was a real sense ofan industry upswing this year and leading ana-lyst reports were optimistic about what 2006has in store,” commented the OSA. “Everyindicator is showing that innovation is aliveand well in the industry again.”
Show goes with a swing
Vertilas eyes expansion with new funds
PHIL
IPS
V E N T U R E F U N D I N G
O F C / N F O E C
GaAs (SI & SC)S U B S T R AT E S
InP (SI & SC)S U B S T R AT E S
GeS U B S T R AT E S
RAW
MATERIALS
USA Headquarters4281 Technology DriveFremont, CA, USA 94538Tel: 510-438-4700Fax: 510-683-5901Email: [email protected]
ChinaBeijing Tongmei XtalTechnology4 Eastern Second StreetIndustrial Development ZoneTongzhou District, Beijing, China 101113Tel: 86-10-6156-2241Fax: 86-10-6156-2245Email: [email protected]
Japan EastMBK Microtek, Inc.Shuwa Shiba Park Bldg A-10F4-1 Shibakoen 2-chomeMinato-ku, Tokyo, Japan 105-0011 Tel: 81-3-5733-0701Fax: 81-3-5733-0702Email: f.nishiyama@
mbk-microtek.co.jp
Japan WestMo Sangyo Co, LTD.201 Ashiya-Farfalla, 6-16 Narihira-choAshiya-Hyogo, Japan 659-0068Tel: 81-797-32-0046Fax: 81-797-32-0304Email: [email protected]
V G F S T R E N G T H . P E R F O R M A N C E . I N N O VAT I O N .
AXT proudly announces the launch of its new corporate look.The change reflects our continuing commitment to valued customers,
ensuring them superior technology, products and customer service.
Please visit axt.com to see why we are the premier source for VGF technology.
(NASDAQ: AXTI )
TaiwanConary Enterprise Co., Ltd.10F No. 28, Sec. 3Nan-King East RoadTaipei, Taiwan, R.O.C.Tel: 886-2-2509-1399Fax: 886-2-2501-6279Email: [email protected]
KoreaIantek CorporationCheongMyung Towntel 6071021-4 YeongTong-Dong,PalDal-GuSuwon-Si, KyungGi-Do, 442-813, KoreaTel: 82-31-204-4221Fax: 82-31-204-4220Email: [email protected]
EuropeGeo Semiconductor Ltd.POB 6262CH 1211 Geneve 6SwitzerlandTel: 33-1-45316284Fax: 33-1-45333943Mobile: 33-680-134-895Email: [email protected]
United KingdomGeo Semiconductor (UK) LtdNewton GrangeKingsley Green, Kingsley RoadFrodsham, Cheshire WA6 6YAUnited KingdomTel/Fax: 44-(0)-1928-735389Mobile: 44-(0)-779-543-8189Email: [email protected]
Think that epi-ready sapphire waferis defect free? Think again.
Detect and classify those critical transparent wafer defects with Candela.™ Our Candela Optical Surface
Analyzers are unique surface inspection systems that automatically detect and classify defects on transparent
materials like sapphire, SiC, LiNbO3, and glass. Faster than operator-dependent microscope inspection,
these systems can detect surface contaminants and defects on wafer substrates that hurt device performance.
Measures scattered light, reflected light, phase shift, and surface topography
Detects and classifies epi and coating defects
Accommodates 50 to 300 mm wafers
Automated cassette-to-cassette operation
For more product information, go to:
©20
05 K
LA-T
enco
r Cor
pora
tion.
www.kla-tencor.com/candela
Accelerating Yield®
Compound Semiconductor April 2006 compoundsemiconductor.net 15
I NDUSTRY B E H I N D T H E H E A D L I N E S
After what has seemed a tortuous gestation period, 3Gwireless is now with us and, it would seem, proving pop-ular. According to leading vendor Nokia and others, thenumber of 3G handsets shipped in 2006 should approach100 million, nearly double the 2005 volume.
At the annual 3GSM World Congress wirelessindustry get-together in Barcelona in February, the 3Gpush was very much in evidence. Ali Khatibzadeh fromInGaP power amplifier (PA) supplier Anadigics wasthere. “As soon as you set foot in Barcelona you sawSamsung billboards advertising the first HSDPAphone,”said the general manager of wireless products at theWarren, NJ, chip maker.
HSDPA stands for high-speed downlink packetaccess, and is a key area for Anadigics. That’s becausewireless industry giant Qualcomm is a big noise whenit comes to this particular technology. “At the moment,if you’re designing and building an HSDPAphone thenyou have to go to Qualcomm,” explained Khatibzadeh.Anadigics now supplies the San Diego company withthe PAs that feature in the reference design for theseadvanced chipsets.
“HSDPAand 3G chipsets are much more tied togetherthan with 2G. It gives us a unique position because thebarrier to entry for other suppliers that are not on thereference design is quite high,” said Khatibzadeh.
Because of the added complexity and resulting cur-rent consumption of 3G cellular, the linearity of the PAis even more critical, and this is where InGaPexcels. Notonly are handset unit shipments rapidly ramping up, sois the III-V content in each handset.
This is where Anadigics’BiFET technology offers acrucial advantage. Most of the chip maker’s new prod-ucts now feature the integrated InGaP HBT/PHEMTstructures, which are manufactured on 6 inch wafers.Rivals like Skyworks Solutions, Triquint Semiconductorand RF Micro Devices are taking a similar approach.Skyworks has its own BiFET process that combinesInGaP HBTs with FETs on the same substrate, whileRFMD’s approach has been to integrate PHEMTs andHBTs manufactured separately.
Skyworks continues to manufacture on 4 inch wafers,but Anadigics converted to the larger platform way backin 1999. Having previously led the PApack in the earlydays of cellular with its GaAs MESFETs, Anadigics wasknocked off its perch when RFMD became Nokia’s keysupplier of GaAs HBTs, as cell phones transitioned to3 V operation in the late 1990s.
The combination of lost market share and the hugeinvestment in its 6 inch GaAs fab has made life toughfor Anadigics since then, but with the arrival of 3G,HSDPA and WiMAX the worm now appears to beturning in the company’s favor.
That much is clear from a strong pick-up in revenue.In its February earnings call, sales for the holiday quarter were up 49% year-on-year, and Anadigics has
predicted an even stronger 2006.Khatibzadeh sees a direct correlation between those
sales increases and the introduction of BiFETs, whichAnadigics transferred into volume production in 2003.“Looking at the strong relationships we have with keyplayers such as Intel and Qualcomm, you could arguethat this [BiFET] technology has been a key perfor-mance differentiator.”
Anadigics already enjoys strong links with Intel onits Centrino Wi-Fi chipset for notebook PCs. With Intelnow pushing the WiMAX equivalent – at its developerforum in March, Intel’s Sean Maloney said that cardsfor mobile WiMAX would begin shipping in late 2006– the new broadband connectivity protocol looks to beanother boon for Anadigics.
“WiMAX is an OFDM system, which means that itreally requires linear performance from the PA, and ingeneral it also means greater power consumption,”explained Khatibzadeh. “So having a good, low-powerconsumption PAwith good linearity is important for 3Gas well as WiMAX – and that means that it favors GaAs.Nobody’s talking about CMOS PAs for 3G or WiMAX.”
The Anadigics strategy now revolves around theBiFETadvantages. “If you look at Skyworks or RFMD,there is more of a focus on 2G and 2.5G, with theirPolaris and Helios platforms,” said Khatibzadeh.“People expect to have 2.5G for the same cost as 2G. Wethink the real story is in 3G. Obviously, 3G featuresGPRS/EDGE as a requirement, but it is one part of theoverall chipset.”
While Anadigics does pick up some 2.5G business,it sees a much greater opportunity for revenue growthand, ultimately, a return to profitability in the moreadvanced platforms. “We go after 2.5G in an oppor-tunistic way. For example, the Research in MotionBlackberrys use Anadigics’PAs, and we’re very proudof that. This kind of market will exist and will grow, butwe’d rather put our investments in 3G.”
Anadigics charge hinges on BiFETsWafer-level integrationof InGaP HBT andPHEMT structures is atthe heart of Anadigics’strategy to return toprofitability, hearsMichael Hatcher.
A D V A N C E D R F I C S
3G big-hitter Qualcomm is using Anadigics’ power amplifiers inits latest HSDPA chipset, which powers the most advanced mobiles on the market. The power amplifiers used feature theInGaP-plus BiFET process, as do most of Anadigics’ new products.
ANAD
IGIC
S
The popularity of Research inMotion’s Blackberry handsets,along with Intel’s highlysuccessful Centrino Wi-Fichipset, has boosted Anadigics’sales, and WiMAX looks set tofollow that trend.
“The [2.5G]market will grow,but we’d rather putour investments in 3G.”Ali KhatibzadehAnadigics
RIM
compoundsemiconductor.net April 2006 Compound Semiconductor16
I NDUSTRY P O R T F O L I O
Finisar and Avanex spring into lifeMassive tradingvolumes and a rapidupturn in thevaluations of Finisarand Avanex coincidedwith the year’s biggest fiber-opticcomponents show.
F I B E R - O P T I C C O M P O N E N T S
The first week of March marked the onset of some earlyspring growth for two of the compound industry’s worst-performing stocks of recent years – Finisar and Avanex.Trading volumes went through the roof, with 80 millionshares in Finisar (around a quarter of the company)changing hands on March 3 and a similar level of activityat Avanex. The market value of both companies doubledin the blink of an eye.
Why the sudden upsurge? Well, global warming mightbe changing some seasonal patterns, but it can’t be heldresponsible for this one. The uptick coincided with theannual OFC/NFOEC trade show in Anaheim, and adeluge of public relations and marketing from fiber-opticcomponent vendors of every kind. But that happens everyyear, and would have surprised nobody.
On the eve of this year’s show, however, Finisar didreport surprisingly good figures for its third quarter, whichended on January 31. Its stock price rocketed up morethan 40% as a result of that, and dragged others, includ-ing Avanex, along with it.
Finisar made a profit of $8.3 million in the quarter.Yes, you read it right – a profit. That hasn’t happened fora while, although it was correctly predicted by the com-pany a couple of quarters back (see Compound Semi-conductor October 2005 p30). Analysts had only beenexpecting a breakeven quarter, hence the big spike.Another profit is expected in the current quarter, and thestock has maintained its value following the upswing.
Finisar’s strength is not in exactly the same areas asother component manufacturers like JDSU, Bookhamand Avanex, but the market seemed to take a wider viewthat all optical components were back in vogue, fuelinga rise in stock prices across the whole sector.
The much higher volume of trading in Finisar sharesmay indicate that the big institutional investors could betaking an interest in fiber-optics again. CFO Steve Work-man must have little idea who some of his biggest invest-ors are, given the rate at which the stock has been changinghands. Workman described the company as “a little dif-ferent” to other firms in the optoelectronics space at thecompany’s recent investor conference call. With prof-itability, it can boast a certain financial uniqueness too.
Vertical strategyThe decision by Finisar to become vertically integratedcame after the telecoms crash, when it acquired theVCSELmanufacturing operations of Honeywell and theFremont, CA, InPfacility formerly belonging to Genoa.This is now a cornerstone of the company’s strategy.
Finisar paid just $3 million for the InP fab, while theprevious residents had sucked up almost $100 million inventure capital. Crucially, the volume of long-wavelengthInAlGaAs laser and detector chips being made inFremont is now ramping. This is one of the big reasonswhy Finisar’s gross and operating margins are soimproved, and why the company has pulled its bottomline out of the red.
According to Joe Young, senior VPand general man-ager of Finisar’s optics division, InP chip volumes inJanuary 2006 were up 174% on January 2005. “Fremontis a big turnaround story for us,” Young said. While vol-umes are currently at the 100,000 chips per week level,Finisar is expecting to make a lot more, he added. “Weare investing in extra volume and increasing capacity.”
The Richardson, TX, Advanced Optical Components(AOC) facility, where Finisar manufactures its 850 nmVCSELs, forms another critical part of the overallstrategy. AOC’s VCSELs are the acknowledged indus-try standard in datacoms, and Finisar is having mixedsuccess in diversifying the applications in which they areused. Although the laser is featuring in advanced com-puter mice, with tens of thousands of integrated laser anddetector chips shipping every week, Finisar admitted thatscaling this volume up to the desired level of 100,000 perweek is proving to be a struggle.
Communications will remain the key business driverfor Finisar, however. Young expects that a strong rampin the transceiver market of around 24% per year will pro-pel company revenue upwards at a similar rate. Excludingpassive optical networks, Young believes that Finisar’sshare of the transceiver market will have increased fromjust under 20% in 2004 to nearly 30% by the end of 2006.
Delivery of video content via the internet is going tobe a critical consumer application that will drive vol-umes: “The big ‘hog’for bandwidth is video,” explainedCEO Jerry Rawls in the investor call. “Thirty minutes ofvideo is equivalent to one year of email.” All that videois going to require a lot of storage too, and luckily enough,products for storage-area networks are a big part of theFinisar business too. Meeting that anticipated extrademand is the challenge now, concluded Rawls: “Thenumber-one topic is capacity, and we plan to increasethis. We know from history that rising markets are gen-erally underforecast by our customers.”
Having control over its own laser and photodetectorsupply will give Finisar an advantage if demand increasesfaster than has been suggested. The company has madethe vertically-integrated model work, but others in thefiber-optic components business still have much to prove.
6
5
4
3
2
1
0January February March
2006
shar
e pr
ice
($)
Finisar
Avanex
Market movers: the value of both Finisar and Avanex more thandoubled in early spring after Finisar posted a profit – as predicted bythe company two quarters previously. Source: Yahoo Finance.
“The number-one topic iscapacity, and we plan toincrease this.”Jerry RawlsFinisar
GaAs Substrates for Micro- andOptoelectronics
Semi-insulating VGF and LEC wafers up to 8“
Semi-conducting VGF and LEC wafers up to 6”/4“
Comprehensive crystal and wafer characterization
Bulk material with extremely low residualimpurity concentration
Excellent surface quality for epitaxial growth
Freiberger Compound Materials
Freiberg, Germany: Phone: +49 (3731) 2800Fax: +49 (3731) 280 106Dayton, OH, USA: Phone: +1 (937) 291 2899 Fax: +1 (937) 291 2893www.fcm-semicon.comIn
all
ianc
e w
ith
the
futu
re
compoundsemiconductor.net April 2006 Compound Semiconductor18
I NDUSTRY I N T E R V I E W
At its former manufacturing site in Fremont, the near-empty AXT car park suggests Mary Celeste compar-isons. While that might be a little unfair, the Californialocation is a shadow of its former self, when around 1000people filled its four large buildings.
“2006 is a big turning point for us,” says CEO PhilYin, confidently predicting a revival in fortunes. Havingrevamped the senior management and dealt with oper-ational problems, Yin believes that AXT’s strong linksin China will prove to be a big competitive advantage asdemand for GaAs ramps up.
In Fremont, that Chinese influence is unmistakable:from the reception staff to the senior management; fromthe new red and gold corporate colors to the boardroomartwork. And China is where the company’s heart nowlies, along with more than 500 employees and five jointventures that supply AXT with its critical materials.
Yin likes a challenge, which is just as well becausewhen the industry veteran took over as its CEO one yearago, AXT was heading for the rocks. Just a few yearsafter it was first to market with 6 inch vertical gradientfreeze (VGF) material and held a market-leading posi-tion in GaAs substrates, things looked bleak.
The hangover that followed the 2000–2001 over-capacity binge hit AXT long and hard. It suffered a bigfall in market share and revenue as rivals followed itspioneering VGF technology. Then came another blow –certain customers found that their AXT-sourced mater-ial suffered from poor surface morphology, causing“post-epi haze”, and went to find alternative suppliers.
Combine that with [as yet, unproved] accusations ofmalpractice by former employees on health and safetygrounds, a merry-go-round in senior management asoperations switched from California to China, and thedistraction of a damaging dispute with rival supplierSumitomo Electric Industries, and Yin’s task must havetaken on nightmarish proportions.
But this is what Yin calls the “old AXT”, one of dam-aged reputation. “We had bad baggage,” he admits.
For the AXT board, enough was enough. Somethinghad to be done. So they called Yin, a veteran of – amongothers – Aixtron, ATMI Epitaxial Services, Crysteco and IBM, to sort things out. Luckily for them, Yin wasup for the challenge. First came the management changes:out went interim CEO Don Tatzin, while company co-founder Morris Young switched from a wide-rangingrole, including CEO of Chinese operations, to a morefocused position as CTO. That move can’t have been easyfor Young, but Yin saw it as a critical one for the futureof AXT. “Morris is back towhat he did best during hisBell Labs days,” Yin said.
Similarly, the appoint-ment of a new operationschief in the form ofMinsheng Lin in July 2005has beefed up the com-pany’s focus on productionissues and quality control.AXT’s material qualitywoes had stemmed directlyfrom some relatively trivialoperational oversights, the primary cause being identi-fied by Lin as the use of a polishing solution that hadbeen made up to the wrong concentration.
Fixing this problem, as well as bringing in automatedcleaning stations, megasonics and spin-rinse driers is allpart of the “new AXT”, along with a revamped corpo-rate image and a customer-first attitude.
Following the setup of the five joint-venture compa-nies to supply AXT with raw materials and equipment,the Chinese links are stronger than ever (see “AXT’sfive joint ventures” box, p19). These ventures, in whichAXTholds substantial shares and board presence, allowthe company control over its supply chain, says Yin: “Weare not susceptible to capacity constraints or swings inraw-material prices.”
Part-ownership of a furnace supplier also allows AXTto add capacity relatively easily. Yin says that the designof these furnaces is critical, and this remains the com-pany’s most closely guarded secret. AXT has neverpatented the design, Yin insisting that this would let thecat out of the bag.
While AXT can pressure the joint ventures on priceto reduce its overheads, this is only one part of thesolution. Above all else AXT needs to take back sub-stantial market share to approach profitability. In recent
Changing fortunes in the year of the dogDealing with a legacy of material quality problems, management issuesand a plunge in market share doesn’t sound like most people’s idea of fun,but that didn’t stop Phil Yin taking over the reins at substrate supplier AXT.
A X T
Phil Yin has made sweeping changes since joining AXT, and expects to reap the benefits in 2006.
AXT
“First came themanagementchanges: out wentinterim CEO DonTatzin, while MorrisYoung was switchedto CTO.”
May 1998: 12 years afterbeing founded, AXT goespublic with an IPO.May 2000: 300% expansionof GaAs production plannedin Fremont and Beijing. DonTatzin becomes CFO.August 2000: substratesales near-double as GaAsdemand ramps.September 2001: sales hitby sudden weak demand.April 2002: 6 inch VGF InPsubstrates released.June 2003: optoelectronicsdivision is discontinued.May 2004: managementreshuffled after substratequality issues hit sales.March 2005: Phil Yinappointed CEO; Tatzinleaves, Morris Youngswitches to CTO.January 2006: suite of raw-materials products launched.February 2006: AXT joinsSEMI organization.
AXT: the fall and rise?
INDUSTRY I N T E R V I E W
quarters there has been little sign of this – while revenuehas begun to creep back up after a long downturn, thecompany continues to post a net loss. But Yin nowexpects to oversee a dramatic ramp-up.
That’s partly because demand is increasing across theboard, particularly as the latest cell-phone handsetsdevour ever more GaAs content. Crucially, AXT’s rivalsuppliers, such as Freiberger, are struggling to meet thevolumes demanded by their customers. As a result, AXTis brimming with orders, and has far better visibility onits customer requirements for the rest of 2006 than it hasseen in recent years.
“Total requirements – in terms of square inches ofmaterial – could be close to double the 2005 figure by theyear-end,” predicts Yin. Aside from the squeeze on GaAscapacity, increasing demand for germanium looks likefavoring AXT. The firm is one of only a handful of ven-dors, along with chief rival Umicore, to sell the germa-nium substrates used by Emcore and Spectrolab to makesolar cells, and by Osram Opto Semiconductors in HB-LED fabrication. Osram’s chip manufacturing volumeswill no doubt continue to soar, while the use of multi-junction solar cells in terrestrial applications is at lastbeginning to take off. Again the Chinese operation is cru-cial – one of AXT’s joint ventures controls a germaniummine in Mongolia, ensuring that material supplies arelargely under control, while 4 inch germanium substratesare in volume production thanks to a large order from theChinese government for two satellite applications.
As you would expect from somebody who spent a lot
of time in the silicon semiconductor business, Yin is keenon standards and roadmaps. As a result, AXT recentlyjoined the SEMI organization as a corporate member.“Compound semiconductor [volumes] will not snow-ball unless we get standard specifications,” he said.
For the Chinese community 2006 is the year of thedog, and as such is said to be one of reflection. If hisexpectations prove correct and market share flows backin AXT’s direction, many will reflect that Yin has beenthe chief architect of AXT’s long-awaited revival.
ISCS 200633rd International Symposium on Compound Semiconductors
August 13 - 17, 2006
University of British Columbia
Vancouver, Canada
Semiconductor Epitaxy:
MBE, MOCVD and related epitaxial technologies;
metamorphic and heteroepitaxial growth;
structural, optical and electrical characterization
of surfaces, interfaces, low dimensional and bulk
materials.
Photonic Structures and Devices:
Theory, fabrication and characterization of
photonic crystals, microcavity structures.
Electronic Materials and Devices:
Electronic properties of compound semi-
conductors and devices, Si-based compounds,
nitrides, arsenides, phosphides, antimonides,
tellurides, RTDs, transistors, RF photonics, high
temperature, high power devices.
Nanostructures:
Quantum dot and nanowire growth; QD/nanowire
lasers and LEDs, sensors and transistors.
Optoelectronics:
Lasers, LEDs, photodetectors, modulators, SOAs
and OEICs.
Semiconductor Spintronics:
Magnetic semiconductor materials and devices,
spin effects.
Materials and Devices for Energy Conversion:
Solar cells, TPV, optical refrigeration, direct
electricity generation and novel TE materials.
www.iscs2006.ca Call for Abstracts Deadline: May 3, 2006
Name Product(s) AXT holdingBeijing Ji Ya 99.99% gallium 46%Nanjing Jin Mei Gallium 99.9999(9)% gallium, B2O3 83%Beijing Bo Yu Manufacturing pBN crucibles, MBE parts 70%Xilingol Tongli Ge raw germanium 25%Emeishan Jia Mei HPM 99.99% and 99.9999% arsenic 25%
AXT’s five joint ventures
AXT
compoundsemiconductor.net April 2006 Compound Semiconductor20
T ECHNOLOGY E P I T A X Y
SiC growth rockets with hyThe standard growth process for silicon carbide produces high-blocking-voltage devices, but suffers from long processingtimes. These delays are hindering commercialization, saysFrancesco La Via, who believes that the problem can beovercome by adding hydrogen chloride into the cell.
Rapid improvements in substrate production andadvances in the growth of high-quality epitaxial filmsby CVD have driven the development of SiC materialthat is ideal for high-power microelectronic devices.This has led to the fabrication of several types of devicewith blocking voltages of at least 10kV, including powerdepletion-type MOSFETs, implanted JFETs, PINdiodes and Schottky barrier diodes. However, to obtainthese breakdown voltages, epitaxial layers 80–100 µmthick are needed, which require processing times of10hours or more for conventional epitaxial growth ratesof 6–8 µm/h. These long growth times equate to highprocessing costs, and this is hampering the commercialdevelopment of high-voltage SiC devices.
CVD using “step-controlled epitaxy” is the standardapproach for homoepitaxial growth of α-SiC, the mostcommon form of the material. With this method thepolytype is controlled through the surface steps on off-axis substrates. Mirror-like surfaces can be producedwith carbon/silicon ratios of 1.4–2.5 at a growth ratealmost independent of the gas mixture. Within thisregime the epitaxial growth rate is directly proportionalto the silane flow rate, but at high silane flows silicondroplets are formed in the gas phase that are thendeposited on the wafer.
Combining these observations with analysis of thegas-phase kinetics in the growth system has led us toconclude that epitaxial growth proceeds through siliconadsorption at atomic steps, followed by carbonizationof hydrocarbon molecules. The main limitations of thisprocess are the low growth rate that results from slowsilicon species diffusion through the stagnant layer, andthe limited silicon/hydrogen-gas ratio, which has toremain below 0.05% to prevent homogeneous nucle-ation of silicon droplets in the gas phase. Nucleation canalso lead to a poor-quality surface through depletion ofthe gas-phase precursors that are needed for deposition.
An improved epitaxial process that overcomes thegrowth-rate problem has recently been developedindependently by both our research team in Italy andscientists at the University of South Florida. Theimprovements include increasing the silane flow andintroducing HCl gas into the deposition chamber, andhave led to much higher growth rates combined withgood surface morphology.
In our work, the 4H-SiC epitaxial films were grownusing silane and ethene precursors in an LPE EpitaxialTechnology ACSiM8 hot-wall reactor that has a highdegree of temperature uniformity and can accommo-date either six 2-inch or three 3-inch substrates.
HCl combines speed with qualityIt turns out that HCl addition can significantly increasethe silane concentration, while avoiding the homoge-neous nucleation of silicon in the gas phase that usuallyoccurs during the standard deposition process. With thislimitation removed, high growth rates of up to 112µm/hthat are directly proportional to the silane flux are pos-sible (figure 1). No homogeneous gas-phase nucleationhas been observed, even at the very high silane concen-tration of 0.6% that was used for the highest growth rate.Even higher growth rates might therefore be possible.
We characterized 100 µm thick epitaxial layersgrown with the HCl-based process using AFM to deter-mine the film’s surface roughness. Measurements from
Our hydrogen chloride-based process can slash the growth times of silicon car
Reactor configuration 7 × 2 inch 7 × 3 inch 6 ×Total useful wafer area (cm2) 92 241 382Growth rate (µm/h) 3 4.9 5cm2 of 100 µm thick epi per day 60.8 247.1 398Wafer throughput 4.6 7.2 6.3Thickness uniformity (σ/mean %) 3 1 3Doping uniformity (σ/mean %) 7 5 9
Reactor type Planetary warm wall
Reactor comparisons
Our research team’s results using our horizontal hot-wall reactor compare favorabMaterials Science Forum 2005 483–485 137), and those presented by SiCED atwhich took place in Pittsburgh, PA. Although our LPE Epitaxial Technology reactor hpossible with the HCl-based process mean that this approach can actually deliver t
Compound Semiconductor April 2006 compoundsemiconductor.net 21
TECHNOLOGY E P I T A X Y
several regions of the wafers show that the averagevalue is independent of growth rate and is typically0.3 nm (figure 2). This roughness is similar to that ofthe standard HCl-free process.
Our potassium hydroxide etching experiments at500 °C have also revealed that there are no major dif-ferences between the dislocation densities of the SiCepilayers produced with and without HCl addition.
We have also measured the minority-carrier lifetimesof both types of epiwafer to establish whether there aredefects within the material that degrade device perfor-mance. These experiments were carried out with a mod-ified microwave photoconductive decay instrumentbuilt by Semilab, which features a 350 nm laser excita-tion source. Figure 3 shows the minority-carrier life-time distribution from a 100 µm thick epitaxial layergrown using HCl addition. The distribution peak showsthat the average carrier lifetime is over 1 µs, which is ahigh value for SiC.
We confirmed the quality of the epitaxial layers
grown with HCl addition using deep-level transientspectroscopy (DLTS) measurements. Although filmsproduced with both types of epitaxial process each con-tain only one defect level – the EH7, which has an asso-ciated energy of 1.5eV– the concentration of this defectis three times lower in the layer grown using HCl addi-tion. The combination of the DLTS and the microwavephotoconductivity decay results suggests that the fasterprocess could also be used to produce high-quality bipo-lar devices and X-ray detectors.
We also know from capacitance-voltage measure-ments that the high growth rate process can producelow background doping levels of less than 1×1014 cm–3.This means that the HCl-based process can yieldintentionally doped layers with low doping concen-trations and good uniformity. Figure 4 (p22) shows thedoping distribution for several wafers grown with thesame process. The average doping concentration is5.6 × 1014 cm–3, with a standard deviation of 6.7%.
The faster process also delivers extremely good thick-
ydrogen chloride addition
bide devices by several hours, consequently reducing their production costs.
100 mm 6 × 2 inch 3 × 3 inch 5 × 3 inch2 78 103 172
112 112 68.6 480.9 635.0 209.9
37.0 18.5 6.11.2 1.6 27 8 9
Horizontal hot wall Planetary hot wall
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
20
40
60
80
100
120
HCl
no HCl
T = 1650 °CC/Si = 1.0
siliconprecipitates
Si/H2 (%)gr
owth
rate
(µm
/h)
T = 1550 °CC/Si = 1.5
Fig. 1. The standard growth process (bluetriangles) is restricted in growth rate to well below 20 µm/h, and at the higher end of thisrange silicon droplets tend to form in the gasphase and contaminate the wafer. The HClprocess (red squares) is capable of far highergrowth rates of up to 112 µm/h.
0 20 40 60 80 100 1200.1
0.2
0.3
0.4
0.5
surfa
ce ro
ughn
ess
(nm
)
growth rate (µm/h)
Fig. 2. Atomic force microscopy measurements of 5 × 5 µm areas of the surfaces of materialproduced by both the standard (blue triangle) and HCl-based (red squares) growth processesreveals that both of the methods produce similarvalues for surface roughness.
bly with those produced at Cree using a planetary warm-wall reactor (see t last year’s International Conference on Silicon Carbide and Related Materials,has a relatively low substrate capacity, the far higher growth rates that are the highest daily throughput for 2 and 3 inch material.
0.50 0.82 1.14 1.46 1.78 2.10
10
5
0
lifetime (µs)
15
coun
ts
Fig. 3. Microwave photoconductivity measurements show that the HCl process produces an averagevalue for the minority-carrier lifetime of just over 1 µs, indicating that this material has very few defects.
About the authorFrancesco La Via is a seniorresearcher at the Institute forMicroelectronics andMicrosystems, National Councilfor Research (CNR-IMM), in Italy,where he leads the department ofSiC growth and epitaxy research.
SiC
ED
TECHNOLOGY E P I T A X Y
ness uniformities. Our Fourier transform infrared spec-troscopy measurements of several epiwafers have shownthat the standard deviation in the film thickness for lay-ers with an average thickness of 56.8µm was just 1.2%.
Very high wafer throughputAlthough the ACSiM8 has a significantly lower capac-ity than the reactors used by Cree or produced by SiCED(see table, p20), its far higher growth rate capabilitymeans that it can deliver a higher wafer throughput for2and 3 inch material. In addition, the epilayer thicknessuniformities produced with the LPE reactor comparevery favorably with those reported elsewhere.
We have also started to investigate the performance
of devices grown with the high growth rate process.We compared Schottky diodes grown on different sub-strates and produced by either the HCl-based processat a growth rate of around 20 µm/h or the standardprocess. Electrical measurements showed little dif-ference between the leakage current distribution ofboth types of diodes – the average leakage current is10–5 A/cm2 at –600Vin both devices. The forward volt-age distribution at 20 A/ cm2 is also very similar forboth processes, indicating that both approaches givegood electron mobility. In the near future we will char-acterize Schottky diodes fabricated with the muchfaster 112 µm/h growth rate.
Our new process, which can routinely produce 50–100 µm thick epilayers with good morphology and lowbackground doping, will provide the SiC industry witha lower cost production technique for high-blocking-voltage devices. This will aid the growing interest inthese devices for a wide range of applications, includ-ing electrical power converters in ships and automo-biles, and voltage transformers for electrical grids.
Further reading D Crippa et al. 2005 Mat. Sci. Forum 483–485 67.M Das et al. 2005 Mat. Sci. Forum 483–485 965.AElasser et al. 2002 Proc. IEEE 90 969.J Zhao et al. 2004 IEEE Elec. Dev. Lett. 25 474.J Zhao et al. 2003 IEEE Elec. Dev. Lett. 24 402.
AcknowledgementsFrancesco La Via thanks StefanoLeone, Marco Mauceri, GiuseppePistone and GiuseppeAbbondanza (ETC) for SiCepitaxial growth, GiuseppaGalvagno (CNR-IMM) for Schottkydiode characterization, LuciaCalcagno (Catania University) forDLTS measurements, GaetanoFoti (Catania University) forphotoluminescencemeasurements, and Gian LucaValenti (LPE), Danilo Crippa (LPE)and Maurizio Masi (MilanUniversity) for helpful discussions regarding theinterpretation of the results.
DIRECT WAFER TEMPERATURE.NO EXCUSES.
To learn more or request a demonstration, please call 734.668.4644 or email us at [email protected]. For worldwide sales and support, visit our Web site at www.k-space.com.
kSA BandiT is the only product on the market that can directly measure GaN film temperature.
Don’t put up with outdated temperaturemonitoring technology when you canget direct, real-time readings fromroom temperature to >1200 °C with thekSA BandiT™ Temperature Monitor.
TE
MP
ER
AT
UR
ES
TR
ES
S G
RO
WT
H R
AT
E R
HE
ED
TE
MP
ER
AT
UR
E
k-Space Associates, Inc.
Actual 2” GaN film temperature profile during MOCVD growth
60
40
20
04×1014 5×1014 6×1014 7×1014 8 ×1014 9×1014
doping (carriers/cm3)
coun
ts (%
)
dopingavg = 5.56×1014 cm3
dopingstd = 6.7%
Fig. 4. The HCl-based process can produce SiC epiwafers with ahigh degree of uniformity, in terms of the doping concentration.
Compound Semiconductor April 2006 compoundsemiconductor.net 23
T ECHNOLOGY M A T E R I A L S U P D A T E
There’s never been any doubt that developers andmanufacturers of nitride-based devices would benefithugely from native nitride substrates on which to growtheir epitaxial structures.
But, compared with silicon, GaAs and, to a lesserextent, InP, nitride substrates have proved far more trou-blesome. The main problem is the extremely high tem-perature and pressure required for molten crystal growth.
Recently, a few developments among a select bandof crystal specialists suggest that makers of GaN, AlGaNand InGaN structures are increasingly set to follow inthe footsteps of the more mature semiconductor mater-ials through the use of native substrates and epiwafers.
One of these companies, Kyma Technologies, fol-lows Cree and Nitronex as a spin-out of the influentialBob Davis research group at North Carolina StateUniversity. Unlike some, Kyma’s aim is to develop aboule growth process similar to that which has provedsuccessful for other semiconductor substrate materials.Although this is more technologically challenging than a faster growth method such as, say, hydride vapor-phase epitaxy, Kyma co-founder Drew Hanser main-tains that in the longer term, the boule growth approachwill prove essential.
Kyma has recently set up a number of developmentprograms with the US Department of Defense. Two ofthese take place under the wing of the Missile DefenseAgency, where the main aim is to develop larger sub-strates that will be essential for cost-effective manufac-turing of relatively large transistors for RF applications,such as radar and X-band communications. Kyma hasalso signed two collaborative R&D agreements, thistime with the US Navy and Air Force, under which it issupplying substrates for FET development and increas-ing its focus on material characterization.
Kyma indicated to Compound Semiconductor that its2 inch GaN wafers would be shipping in substantialvolumes within a year, with the first 3 inch materialsuitable for devices set to follow in mid to late 2007.
Encouragingly, a wide range of devices – includingvarious transistors, Schottky diodes and optoelectronics– have already shown excellent performance on Kymamaterial, even from initial growth runs (e.g. a 600 VSchottky diode and 10 GHz FETs). “What we’ve seenwith these early experimental devices is corroborationof the basic theoretical arguments in favor of a native sub-strate,” said Kyma CEO Keith Evans.
As with any starting material, dislocation density isthe critical attribute where device manufacturers want tosee progress. Typically, Kyma’s epiwafers feature etch-pit densities of around 106 cm–2, although a 10-foldimprovement on that has been seen in Kyma’s best “hero”result. Further improvement will be necessary, however,with laser manufacturers specifying a defect density of104 cm–2. The way to improve quality is to make largerboules, and the nuances of this process are at the root ofanother development that Shuji Nakamura believes could
revolutionize the whole field – non-polar GaN.“Non-polar GaN material looks relatively enticing
on paper,” Evans said, explaining that the dislocationnetworks that develop in boule growth of c-plane GaNhave geometrical aspects that may benefit non-polarsubstrates cut from those boules. Kyma has already pro-duced some low-defect-density non-polar and semi-polar GaN. According to Hanser, semi-polar GaN couldalso ease some of the difficulties with epitaxial growth.
Like Kyma, LED and materials supplier The FoxGroup is focused on developing a true boule growthprocess for AlN, a material that if anything is even moredifficult to produce than GaN. But thanks to its pro-prietary crystal growth technology, based on a tantalumcarbide crucible, The Fox Group is now shipping smallquantities of epi-ready monocrystalline 15 mm diame-ter AlN substrates to a number of different device re-searchers, and is working on increasing wafer size to the2 inch minimum requirement for future commercial usein devices with aluminum-rich active layers, such asultraviolet LEDs and solar-blind detectors.
Barney O’Meara from the company, which grows AlN at its facility in Deer Park, NY, says that whiledislocation density measurements are difficult toascertain with any certainty, the quoted figure for itsmaterial is less than 107 cm–2.
While the recurring theme in native wide-bandgapsubstrate materials is that of defect density reduction,innovation in non-native wafers continues apace.Californian start-up Group4 Labs is focused on improv-ing the operation characteristics and reliability of GaN-based devices through the use of diamond substrates.
Because of the lattice mismatch between the two, theGaN layer is transferred – rather than deposited directly– onto diamond, but company founder Felix Ejeckamsays that the defect density, quoted at around 109 cm–2,is unaffected by this process.
Having announced its presence with a square,nitrogen-facing epiwafer, Group4 has swiftly followedup with gallium-facing material measuring a full twoinches in diameter. “Never before have customers hadaccess to a nitrogen-facing as well as a gallium-facingGaN epiwafer,” claims Group4.
Crystal growers push nitride envelopeMichael Hatcherreports on a flurry ofrecent developmentactivity in III-Nmaterials that hasseen numerousfunding deals and thecommercial shipmentof AlN substrates.
S U B S T R A T E S
GaN 220AlN 200CVD diamond 1200SiC 350–490Silicon 149Sapphire 40InP 68GaAs 46Source: Group4 Labs
Material Thermal conductivity (W/m/K)
Thermal conductivity of typical substrate materials
“Non-polar GaNmaterial looksrelatively enticingon paper.”Keith EvansKyma Technologies
COMPOUND SEMICONDUCTORWEEK 2006
November 12–15, 2006San Antonio,Texas, USA
Event organized by
THE CONFERENCES
IEEE Compound Semiconductor ICSymposium 2006 provides the idealforum to present your latest results in high-speed digital, analog,microwave/millimetre wave, mixedmode and optoelectronic integratedcircuits.
Call for papers - IEEE CSIC Symposium 2006First-time papers concerned with the utilization and applicationof InP, GaAs, SiGe, GaN and other compound semiconductors in military and commercial products are invited. The deadline forelectronic receipt of abstracts is May 15, 2006.
The Key Conference 2006 will feature senior-level experts fromthe industry examining the trends and strategies in compoundsemiconductor markets, technologies and applications, withsessions on GaAs/SiGe/strained Si microelectronics, SiC/powerdevices, alternative III-nitride opportunities, solar cells andlasers/detectors. You can sign up for regular program informationon our website.
THE EXHIBITION
The Compound Semiconductor Week Technology Exhibition,combining visitors from both conferences, will give exhibitorsaccess to buyers who are interested in the entire array ofcompound semiconductor products and services, includingmaterials, manufacturing, device technology, integrated circuitsand commercial and military applications.
Early-bird discountBook your booth by June 15, 2006 for just $1800, saving$200!
Visit www.compoundsemiconductor.net/csweektoday to book your space.
November 12–15, 2006 San Antonio, Texas, USA
COMPOUND SEMICONDUCTORWEEK 2006Conferences and Exhibition
Compound Semiconductor April 2006 compoundsemiconductor.net 25
T ECHNOLOGY G A A S D E V I C E S
Bespoke MMICs aid radar, phasedarray and oscilloscope applications
Traveling-wave amplifiers – also known as distributedamplifiers – are not new products. Their roots can betraced back to 1936 when they were built using valvetechnology. Today they are available as standard prod-ucts featuring amplification up to 40GHz that are beingused in broadband communication networks and piecesof electrical measurement equipment.
However, these off-the-shelf products are not idealfor radar, phased array, or oscilloscope applicationsbecause in these cases the performance specificationsneed to be specifically tailored. Targeting these needs,here at the Institute of Electronics, Communicationsand Information Technology (ECIT), at Queen’s Uni-versity, Belfast, we are designing and building bespoketraveling-wave GaAs amplifiers with very low gainvariations with frequency and very low phase distor-tion. Our designs include amplifiers with a small vari-ation in gain over their frequency range, which are usedfor testing the electromagnetic compatibility of vari-ous RF components, and for addressing the problemsassociated with signal reflection from antennas. Wealso produce amplifiers with low phase distortion thatare suitable for applications that do not need a wide fre-quency response but do require a faithful reproductionof the pulse’s shape. This feature makes these ampli-fiers ideal for use in radar, phased arrays, oscilloscopes,and also in optical receivers, where they can be usedto amplify the photodiode’s signal without distortion.
Traveling-wave amplifier MMICs operate byabsorbing parasitic shunt capacitances – which comefrom the insulator that separates the gate and the con-ducting channel – into artificial transmission lines. Thisremoves the active device’s main cause of frequencydependence, and also ensures a very low phase distor-tion during amplification.
One of our traveling-wave amplifiers is shown inthe “ECIT’s traveling-wave amplifiers” box on p26. It feature a 0.5–18 GHz bandwidth, 12 dB gain withless than ±1 dB ripple and a 17 dBm 1–dB compres-sion point – defined as the output power for which the gain has dropped by 1 dB. The design occupies
3.0 × 1.25 mm of die area and was fabricated using theD01PH power process at OMMIC, the France-basedcompound semiconductor foundry owned by Philips.
Some of the performance characteristics of theseamplifiers can be seen in figures 1 to 3 (p26). Themeasured gain from 10 of our MMICs produced froma single batch is shown in figure 1. The pairs with thehighest and lowest gain were then selected, and their
The $70 million-funded Institute of Electronics, Communications and Information Technology atQueen’s University, Belfast, is located off campus in the Northern Ireland Science Park.
High-performance traveling-waveamplifiers for oscilloscopes, radar andphased-array equipment can be madecheaper and simpler by switching tocustom-built designs, claim Mark Kellyand Thorsten Brabetz.
ECIT – the Institute of Electronics, Communications and Information Technology –combines the expertise of the electronics and computer-science departments of Queen’sUniversity, Belfast, in one building. Since its launch in 2003, the research center hasreceived $70 million of funding from the European Union, Invest Northern Ireland, andthe country’s Department for Employment and Learning.
The center has a high-frequency electronics group that is developing and commercializingwireless front-end technology. The group has various interests, including high-performanceGaAs and sub-micron silicon integrated chips, self-adapting antennas and monolithicpackaging strategies.
The commercial division is focusing on the production, integration and packaging ofmicrowave and millimeter-wave semiconductors to customer specifications, including acustomer-driven project involving the fabrication of traveling-wave amplifier MMICs.
Institute of Electronics, Communications and Information Technology
ECIT
compoundsemiconductor.net April 2006 Compound Semiconductor26
TECHNOLOGY G A A S D E V I C E S
output power recorded as a function of input power(see figure 2). The phase performance of three of ouramplifiers can be seen in figure 3.
The original circuits were designed without any on-chip DC blocking or bias circuitry to improve yieldand minimize performance variation. However, thisapproach requires DC blocking capacitors to be addedoff-chip (see figure 4a), alongside a more complex bias-ing network, and leads to large modules that are noteasy to assemble.
Greater integrationBecause it is planar, a traveling-wave amplifier is well-suited for complete integration as a GaAs MMIC, andwe have integrated all the necessary circuit componentsonto the die. This approach offers many advantages,including less board space, lower module cost, and areduction in the number of solder joints. The increasedintegration also means that no external capacitors arerequired, and the external inductors that are difficult tosource because they require a flat frequency responsefrom 0.5–18GHz can be replaced with cheaper versionsthat just have to block low frequencies. The level of inte-gration has been extended even further by constructingtraveling-wave MMICs with an on-chip bias networkthat do not require an off-chip bias “tee”. This enablesthe first stage to directly feed the second one, eliminat-ing the need for an interstage network (see figure 4b).
We have also built traveling-wave amplifiers thatdo not require a negative gate bias for a non-mobilecommunications provider that did not have a negativevoltage supply. Our MMIC made the off-chip chargepump that was used with the amplifier redundant, andcut the module size by half. The changes also intro-duced some downsides – a higher DC current con-sumption and reduced drain efficiency – but thesedrawbacks were not a major concern because theMMIC was powered by mains electricity.
Building on this existing experience we are nowdeveloping traveling-wave amplifier MMICs with animproved 1 dB compression point. Conventionaldesigns are restricted by artificial transmission linesthat are relatively lossy, which causes input signalstrengths to decrease significantly as the signal travelsalong the gate line, and means that consecutive stagesare driven with declining power. The consequence isthat only the first few stages receive sufficient powerto be driven into gain compression. Although addingfurther stages can increase the saturated output power,it does not improve the 1 dB compression point.
To reduce this effect we have redesigned our circuitsto minimize gate line losses and have developed atraveling-wave amplifier design with a predicted1 dB compression point of 22 dBm. This year ourcommercialization division plans to launch theseimproved devices commercially and extend the
25
20
15
5
–5–10 5 10 15–15–20 0–5
–10
–15
outp
ut p
ower
(dBm
)
10
input power (dBm)
frequency (GHz)
2
1
0
–1
–2ph
ase
(rad)
5 10 15 200
Fig. 2. A plot of output power as a function of input power, which isused to determine the 1–dB compression point, shows that the gainstarts to reduce at an output power of about 17 dBm.
Fig. 3. The saw-tooth profile of three of ECIT’s amplifiers revealstheir low-phase noise performance, which is a requirement forradar, phased-array and oscilloscope applications.
The traveling-wave amplifier MMICs madeat ECIT feature low gain variations withfrequency and low phase distortion,making them strong candidates for radarand phased-array applications, andamplifiers for oscilloscopes andphotodiodes in high-speed opticalnetworks. This amplifier, which wasfabricated using OMMIC’s D01PH process,contains four 6 × 15 µm PHEMTs with0.13 µm double-recess mushroom gates.Artificial transmission lines are formed withinductor/capacitor “tee” circuits thatinterconnect the gate and drain lines. The
series inductors can be created with eitherspiral or distributed inductors from thin,high-impedance transmission lines. Theshunt capacitance that is needed tocomplete each tee cell is provided by thegate-to-source and drain-to-source parasiticcapacitances of the active devices.
ECIT’s traveling-wave amplifiers
15
10
5
00 5 10 15 20 25
frequency (GHz)
gain
(dB)
Fig. 1. Ten of ECIT’s amplifiers that were produced in a single batchprovide gain in excess of 10 dB between 0 and 20 GHz.
ECIT
Compound Semiconductor April 2006 compoundsemiconductor.net 27
TECHNOLOGY G A A S D E V I C E S
frequency range of our existing amplifier designs tomeet customer demand.
Our development and commercialization oftraveling-wave MMIC amplifiers has shown us that forcertain applications these products are a better choicefrom a performance and manufacturing perspectivethan standard products, even if the dies themselves aremore expensive than the off-the-shelf circuits. This isbecause the different characteristics of the distributedamplifier vary in their importance from application to
application, and bespoke designs can target the specificneeds that they serve. It is an approach that is helpedtoday by computer-aided design packages that enablethe characteristics of a traveling-wave amplifier to befine-tuned precisely. These factors lead us to concludethat for relatively low-volume high-performancecircuits, ECIT’s bespoke MMICs not only improve theamplifier’s performance and reduce its size andcomplexity, but create a more cost-effective circuit thatdoes not require expensive hybrid assembly steps.
1st stage 2nd stage
C C
L L
Coutputinput
TWA TWA
(a)L = inductorC = capacitor
Fig. 4. ECIT’s original MMICs did not feature any on-chip DC blocking or bias circuitry (a). This improved yield and minimized performancevariation, but also had its downsides: a large board space; many solder joints; and the need for expensive external inductors and capacitorswith a flat frequency response from 0.5–18GHz. Many of these issues have been addressed by increasing the level of integration (b). Althoughthis circuit still needs external inductors, lower cost versions can be used because in this configuration they only need to block low frequencies.
About the authorsMark Kelly (left) and ThorstenBrabetz (right) can be contactedat: [email protected] [email protected].
TWA TWA
1st stage 2nd stage
input output
L L
(b)
compoundsemiconductor.net April 2006 Compound Semiconductor28
T ECHNOLOGY F A B M A N A G E M E N T
Analysis tool helps projectmanagers expose hidden pitfalls
Delivering a successful project on time and inside budgetis a rewarding experience that can enhance your career.However, for many of us projects rarely turn out likethis, and instead we find ourselves up against deadlines,distracted by minor issues and constantly firefighting.We also fail to foresee problems and then respond in areactive, rather than proactive way, or allocate insuffi-cient time for tasks, which ultimately leads to spiralingcosts and missed deadlines. So, instead of improvingour reputation with good project management, we endup jeopardizing our careers and being labeled asirresponsible managers with poor organizational skills.
However, the need for good project management,whether it is the installation of a new reactor, the designand assembly of a new product, or the merger of twofabs, is now greater than ever. This is because compa-nies today need to provide more accurate estimationsof project costs, face expensive contract penalties fordelays, and often need to commit to new projects safein the knowledge that current ones are not under risk.In addition, many companies form part of a supplychain, so any delays have wider implications, includ-ing the possible loss of future revenue.
Today, many projects kick-off with the creation of aproject network detailing various pieces of informationabout each task, including its duration, cost, theresources needed, and which tasks must precede andsucceed it. All this data is then put into a Gantt chart thatpresents all the tasks in a table and illustrates them graph-ically as bars according to their start and finish date. TheGantt chart contains an enormous amount of data, butextracting valuable information from it is very difficult.
This is because project networks are like icebergs –only around one-ninth is visible and so can be seen andplanned for, but eight-ninths are hidden under the sur-face. In other words, many of the project’s details areburied in its network and only rear up after the projecthas begun to catch us by surprise. So, in order to man-age a project well, the hidden details must be exposed.
Critical and non-critical pathsMany of today’s projects involve analysis of the criticalpath (CP), which identifies the sequence of tasks thattogether last the full duration of the project and presentthe primary risk for the project overrunning. But whatabout all of the other activities, and the other paths inthe network? For example, in a 12 month project, a non-critical path that lasts 11.5 months is theoretically analmost critical path, but for all practical purposes it is acritical path. Until now, no attention has been given to
these “hidden critical paths” (HCPs), and they are onlynoticed when they expose a problem, but such surprisesin real time can prevent a project hitting its deadline.
The new method for project network analysis – thepatent-pending HCP method – identifies both the CPand HCP. It provides the project manager with severaltools to handle the project by exposing the various dif-ficulties that can occur along the way. The first is theHCPscore, which reflects the project’s complexity androbustness, and is calculated from the project network.The score ranges between 1 and 0. A project with justone CPand nothing else is the least robust, and receivesa score of 1. If a project consists of many HCPs, whicheach have a large slack time, then the score will be low,indicating a stable project. So, if your project has a HCPscore of, say, 0.94, you can conclude that it’s high riskand the chances to complete it on time, on budget, andon spec are low.
Although this analysis reveals that delays to varioustasks could prevent your project running on time, atleast you are in a good position to limit the damage. Forexample, you can show the results to your manage-
Project networks are like icebergs. Only around one-ninth can beseen and planned for, but eight-ninths are buried out of sight. Thehidden critical paths method reveals the concealed details thatwould otherwise catch the company and the project team by surprise.
Project managementspecialist TalLevanon revealshow a new analysistool can reducecleanroomexpansion costs at a semiconductorwafer fab by 25%.
Compound Semiconductor April 2006 compoundsemiconductor.net 29
TECHNOLOGY F A B M A N A G E M E N T
ment, convince them that there is a problem, andsuggest taking action that generates additional time,such as delaying customer shipments. You can alsoredistribute your resources more effectively, becausethe analysis can highlight exactly where and when addi-tional resource is required. The HCPapproach can alsodrive a task reorganization, either in terms of runningorder or priority, or the breaking down of lengthy tasksinto more manageable ones.
The HCPmethod can also expose all the hidden crit-ical tasks within the project network that risk its com-pletion. Until now the standard approach to uncoveringa project’s risks consisted of organizing a meetinginvolving everyone in the project, and encouraging eachparticipant to outline their concerns regarding issuesthat could cause the project to miss its deadline. Thiswould then be followed by a risk-control methodology.However, by using this HCPtasks list, there is no needto guess all the points of risk, as they are easily found.Once they are exposed, a risk-control methodology canbe carried out to minimize any potential delays.
In addition, the HCP tasks list allows project man-agers to determine the “red warning lights” for the pro-ject’s hidden critical resources. For example, themethod can uncover a HCP task on a path with fivedays of slack that has a hidden critical resource – canwe replace the person performing this task within fivedays? Or, if this task requires an expensive resource,but management delays its acquisition, the HCPanaly-sis can show that because there are only five days ofslack on the whole path, dragging the decision by morethan that time can risk the whole project!
Afurther advantage of the HCPmethod is that it canlocate days that have no work allocated to them. Byacting on this information, time can be sliced off theproject and money saved. This situation is not uncom-mon in big project networks. It can also identify tasksthat have a relatively long duration. These “long dura-tion” tasks are shown to the project manager, who deter-mines whether they should be left or broken into shortertasks. Long-duration tasks have specific difficultiesassociated with them as they are difficult to monitor.
The HCP method’s primary strength is its ability topredict the likelihood of a project’s success – on time,on spec and on budget. This calculation is based on theproject network and its findings, which were previouslyhidden. This analysis provides upper management andproject entrepreneurs with the ability to rapidly assessall their projects and drill down quickly and simply tothe key problems – both the obvious and the hidden.
Project management companies and their subcon-tractors can use the HCP method to identify all theexpected and unexpected obstacles they will face beforethey occur. Problems can then be addressed shoulder-to-shoulder, and solved ahead of time, rather than get-ting a nasty surprise when it is too late, and apportioningblame. Using these methods helps companies to buildtrust and form good relationships with contractors.
In addition, HCPanalysis provides project managerswith the opportunity to control projects in a new way.Instead of constantly firefighting, events can be con-trolled and problems foreseen. This can save employ-ers a lot of time and money, and help to breed successthrough proficient, organized project management.
MAX International Engineering Group, a globalengineering consultancy for the semiconductorand compound semiconductor industries, isbased in Closter, NJ, and won a factoryexpansion contract for a US silicon waferfabrication facility. The project had three keyparts: the design and building of a 10,000 sq. ft.cleanroom for a new product line; theinstallation of capital equipment and itsqualification; and finally the initial production.
The project contained 165 tasks and 189connections, began on August 6, 2004, and wasscheduled to end by May 11, 2005. The MAXteam onsite requested hidden critical paths(HCPs) analysis on October 8, 2004, two monthsafter the project had begun. This analysis, whichtook three days, uncovered 46 paths with a slackof 0–20% of the project’s total length. Thosepaths included two critical paths, two HCPs withtwo days of slack, two HCPs with 3.2 days ofslack and one HCP with 5.3 days of slack. TheHCP score was 0.72, indicating that the overallproject network was not particularly robust.
The HCP analysis exposed important findingsthat could not be determined by any other form of
analysis. For example, it revealed that while thescheduled project duration was 198.7 days, theactual duration was only 173.1 days. In otherwords, 25.6 work days had no work scheduled.Most of these were lost in the first two months,before HCP analysis was used, but there was alsoone future task that caused an unaccountablefour work-day delay. In addition, the analysis un-covered a future HCP that had 5.3 days of slack.
By acting on the information provided by theHCP analysis, the project was completed three
weeks ahead of time. Financially, it was also agreat success, as the $250,000 contingencybudget was untouched, and the project’s totalcost was $1.8 million, $0.7 million below budget.
Some will argue that a proportion of thebudget was saved due to its early finish, while itcan also be claimed that the success was drivenby a talented project manager. Although bothviews are valid, MAX’s seasoned projectmanager commented that “the HCP is a verypowerful tool for project managers”.
Case study: putting HCP into practice
About the authorTal Levanon is president of the project managementconsultancy Hidden CriticalPaths, and a lecturer in projectmanagement at theEngineering Faculty, Tel AvivUniversity, Israel. E-mail:[email protected].
CIN
T
compoundsemiconductor.net April 2006 Compound Semiconductor30
T ECHNOLOGY E Q U I P M E N T U P D A T E
LED manufacturing usually involves optical measure-ments of epiwafers at various stages of production.Photoluminescence is frequently used after epitaxialgrowth to obtain a quick assessment of the opticalproperties of the material, while electroluminescencecharacterization is usually only carried out after fulldevice fabrication.
This approach has several downsides, though,because these two optical measurements involvedifferent physical mechanisms and produce differentspectra. Photoluminescence is governed by the mater-ial’s optical properties, but electroluminescence alsodepends on the physical structure of the layers, theelectrical properties of their doped regions, and theproperties of the electrical contacts.
The difference between the two spectra has severalconsequences. For example, it means that a measure-ment of the photoluminescence spectrum cannot directlypredict the electroluminescence peak – in fact, for greenLEDs the top contact can produce a significant wave-length shift in the photoluminescence. The differencealso suggests that photoluminescence data should notbe used to optimize device performance, and explainswhy epiwafers with high photoluminescence efficiencymay not actually lead to devices with good electrolu-minescence characteristics.
Addressing the need for a tool that can provideepiwafer measurements is MaxMile Technologies, acharacterization start-up headquartered in Lexington,SC, that has just released the first instrument that canprovide electroluminescence mapping of as-grown LEDepiwafers. This feature differentiates the tool from otherLED wafer testing instruments such as the “BlueRay”system being developed by Suss MicroTec, which canonly map the characteristics of LED die.
MaxMile’s chief research scientist Max Ma revealedthat it is the company’s probe technology that hasenabled it to make electroluminescence measurementson as-grown epiwafers. The instrument operates byusing a metallic probe with a relatively large tip to forman LED in the epiwafer, which avoids any physicaldamage to the material as the tip is not sharp enough toscratch the surface. The probe’s low current of 100 µAalso prevents damage due to device heating, says Ma,although it does result in a small offset compared with the packaged LED’s emission peak wavelength,which is due to the differences in current density andoperating temperature.
“We came to this idea from the research perspective,where we felt there was a need for a tool that can quicklydetermine device performance,” explained Ma. The toolenables fast, non-destructive measurements of the epi-
wafer’s electroluminescence, which can speed devicedevelopment and lower the associated costs.
Volume LED manufacturers can also benefit from therapid acquisition of electroluminescence data becausethey can use this information to optimize growth. Thisis not the only benefit the instrument can bring to the fab,though, says Ma, because it can also be used to stream-line LED production by screening material immediatelyafter it comes out of the reactor. The electrolumines-cence efficiency of a new LED design can even bepredicted once an epiwafer featuring that design hasbeen fully processed, as the data can provide a form ofcalibration. This reference data is required because thedetected emission intensity can be influenced by vari-ous factors, including the type of substrate and whetherit is polished.
The instrument can map various characteristics of 2, 3 and 4 inch epiwafers, and can complete four elec-trical or optical measurements on 100 sampling pointsin 10 minutes. It can determine the external quantumefficiency of the LED formed in the epiwafer, and itspeak wavelength and total emission intensity. Otherforms of LED characterization are also possible, suchas the electroluminescence spectra at specific currentsor voltages, and the device’s current-voltage behaviorand peak emission wavelength at various drive currents.
MaxMile’s first instrument in the range, the EL-100,is available with either an ultraviolet–visible or visible–infrared detector, making it suitable for all forms of LED.In the coming months the company will also launch twoother versions of the instrument, an EL-300, which canprovide photoluminescence spectra, and an EL-500,which is capable of detecting diode currents as low as1 pA, six orders of magnitude below the detectioncapability of the EL-100.
Electroluminescence mapping toolaccelerates LED developmentMaxMile Technologieshas launched the firstnon-destructiveelectroluminescencemapper forunprocessed LEDepiwafers. The tool isclaimed to offer quickfeedback that canspeed LED developmentand improve qualitycontrol. RichardStevenson investigates.
C H A R A C T E R I Z A T I O N
MaxMile’s EL-100 uses a metallic probe to form an LED in theepiwafer. It can record the electroluminescence spectra at specificcurrents and voltages, the current or emission intensity as a functionof voltage, and the output intensity’s dependence on driving current.
MAXM
ILE
“There was aneed for a toolthat can quicklydetermine deviceperformance.”Max MaMaxMile Technologies
Optoelectronic devices like traffic signals
need to work without fail. Which is why
you need to work with Air Products.
We have the products and experts
to help you optimize your III-V process.
Our SolkatronicTM ultrapure gases include
arsine, high purity ammonia, phosphine and
hydrogen selenide. Our systems maintain
strict control of purity and consistency.
So your films and the devices they control can
set standards for reliability. Whether you’re
in LEDs, lasers, solar cells or other emerging
technologies, visit our web site or call
610-706-6000. There’s a lot riding on the
gas you use. Don’t take chances.
Dr. Rajiv Agarwal, George Ryals and
Dr. Suhas Ketkar have developed
proprietary analytical systems and
test protocols to monitor that our
ultrapure gases are consistently
on spec. Their testing and our
GASGUARD® delivery systems,
control gas purity from plant
to point of use.
When it comes to ultrapure gases, reliability doesn’t just save money.
tell me morewww.airproducts.com/compound
©2002 Air Products and Chemicals, Inc.
compoundsemiconductor.net April 2006 Compound Semiconductor32
T ECHNOLOGY H B - L E D A P P L I C A T I O N S
Chip makers and lighting specialistsget switched on to new possibilities
Penetrating applications outside the mobile phonehandset is no longer simply desirable for makers ofhigh-brightness LEDs – it has become absolutely crit-ical to ensure profitable growth.
Any doubts some may have held about that factwould have been dispelled by industry analyst BobSteele at February’s Strategies in Light conference inSan Francisco. In his annual review and forecast of theHB-LED business, Steele declared that the recent rapidgrowth seen in the mobile applications sector hadfinally run out of steam.
In fact, for the first time since the dawn of the cellphone, sales of LEDs into this market sector fell slightlyto $2.1 billion in 2005. A much bigger drop was onlyaverted because many more cell phones were shippedlast year than was originally expected. Since the entireHB-LED market is valued at just under $4 billion, itpresents chip manufacturers, especially makers ofGaN-based white devices, with a bit of a dilemma.While the still-huge mobile sector cannot be ignored,the falling prices of most white LEDs means that it willno longer generate meaningful revenue growth.
It is time to think beyond mobile phones, then. Andthat message was exactly the focus of this year’sStrategies in Light. In the short term, it will likely bebacklighting of large LCD screens that provides a newengine of growth for HB-LEDs of every hue, with auto-motive headlights likely to present a sizeable marketfor high-power white emitters beginning in the secondhalf of next year.
But the fastest growing sector in Steele’s analysiswas illumination, which now represents a $250 mil-lion annual market for packaged HB-LEDs. Thoughdwarfed by the mobile sector, illumination offers muchbetter profit margins for the cutting-edge chip makers,and Steele reckons that its value will quadruple by 2010.
Culture clashOne of the problems in solid-state lighting (SSL) hasbeen that the makers of HB-LEDs and their potentialcustomers in general lighting have not really been onthe same wavelength. However, this communicationproblem is now showing signs of being overcome.
At the San Francisco conference, Niels Haverkornfrom Philips Lighting described the chip and lighting
communities as being at the “shaking hands andmeeting” stage. Leonard Hordyk from TIR Systems,a Canadian lighting company that has focused on solid-state illumination in recent years, described two sepa-rate industries with different priorities: “The SSLindustry is technology-focused, whereas the generallighting industry is application-focused.”
Part of that distinction is the different view of tech-nological innovation, necessary though it is. In the LEDchip world this innovation is regarded as criticallyimportant, as the efficiency of new devices and the costof each emitted lumen is pushed to the limit. In thelighting industry’s view, such innovation can be adouble-edged sword, however: “The semiconductor[innovation] life-cycle is very, very short,” explainedHordyk, contrasting the ‘cutting-edge-to-obsolescence-in-two-years’ world of Moore’s Law with the muchlonger life-cycle inherent to lighting. “Whereas we [thelighting industry] want the same components to beavailable in 10 or 15 years’ time. The challenge is tobridge that gap.”
Where lighting is focused on industry standards,long product cycles, standard color temperatures andlight distribution, the LED industry is more familiarwith proprietary standards, short product life-cyclesand limited color and light distribution. So saysHordyk, although he is confident about future progress:“Over the next five years, we will be surprised by thenumber of applications enabled by SSL.”
Now in full control of Lumileds Lighting, Philips isa critical player with a big foot in each side of the indus-
The manufacturers of high-brightness LEDs and theircustomers in the general lighting industry now seem to beon the same wavelength, discovers Michael Hatcher.
Cree’s XLamp 7090 warm-white LEDs are being used by PermlightProducts in residential lighting products. The chip manufacturerbelieves that its 1 W devices offer the best “sweet spot”combination of total light output, efficacy and heat management.
CREE
Compound Semiconductor April 2006 compoundsemiconductor.net 33
TECHNOLOGY H B - L E D A P P L I C A T I O N S
try. Its current vision for SSL is to exploit the charac-teristics of LEDs in a way that adds extra atmosphereor “ambience” to the normal lighting experience – forexample in the retail sector by responding to the likelymood of shoppers throughout the day by changing thelight accordingly. Philips is also a big player in the med-ical industry, and hospitals are another environmentwhere Haverkorn sees this kind of mood-altering lightas a major application area.
Both of these applications could be seen as specialprojects outside of the more general residential illu-mination sector, and Haverkorn expects five more yearsof this type of activity before SSL moves into generaluse through what he described as the “unprecedentedparadigm shift” offered by LEDs – a development inan industry that is simply not used to such rapid change.
Even when that does happen, the HB-LED industryshould not view its produce as a replacement for thelight bulb. “SSL will have a huge impact, but the lightbulb won’t go away,” commented Haverkorn. “Afterall, people still buy candles over a century after the bulb was invented.”
Haverkorn and Hordyk both see the tunability ofwhite-light illumination as perhaps the key attributethat will allow it to make an impact in the home. AtStrategies in Light, Hordyk demonstrated the powerof tunable white light using TIR’s 1000 lm LEXELSSlighting system. Unlike conventional illumination,where the color temperature of the light source changeswhen the lamp is dimmed, the tunability of TIR’ssystem means that the color temperature can bemaintained independent of brightness, thanks to a feed-back mechanism. Alternatively, the color temperaturecan be tuned according to preference over the courseof the day and night.
Compared with recent years, the Strategies in Lightmeeting was less focused on laboratory-based chipinnovation, with a greater emphasis on productiondevices, standards, test procedures and calibration.Though subtle, the change may indicate a maturationof the SSLbusiness. Undeniably important though thelm/Wmetric is, many other factors come into play withthe subtleties of general lighting, a fact that Haverkornwas keen to stress. His view is that there is no reallm/W “tipping point” at which SSLsuddenly becomesviable, because there are so many other factors thatcontribute to illumination.
While Cree, Nichia and Lumileds have all beenkeenly promoting their latest progress recently, theiremphasis has been on production devices rather thanany “hero” results from the laboratory. Lumiledslaunched its latest high-power (around 3 W) K2 LEDsin late January. The white versions of K2 emit up to140 lm, and Lumileds’ SSL offering is now basedaround these devices.
Meanwhile, Nichia has developed a 100 lm/W effi-cacy lamp based around its 6 lm output devices thatoperate at 20 mA, and plans to begin volume manu-facturing of these small chips later this year. Havingreached the milestone earlier than expected, Nichia hasadjusted its technology roadmap and now plans to pass150 lm/W in 2007.
Hitting the sweet spotNichia’s US rival Cree has made similarly high-efficacydevices in the laboratory, but is focusing on mid-powerdesigns when it comes to general lighting applications.According to Cree’s Mark McClear, it is the 1Wdevicesthat offer the “sweet spot” combination of good lm/Wefficacy, overall brightness and manageable thermal
Restricted mainly to speciallighting projects and coloredinstallations until now, whiteLEDs are beginning to feature insome high-end residentialapplications. This dining room ina Mexico City home featureswarm-white LED lightinginstalled by the US firm ColorKinetics, while (inset) Cree’sXLamp 7090 products are usedin lighting applications such asthis Californian garage byPermlight Products. Permlightalso uses Nichia LEDs inresidential applications,specifically for downlighting.
COLO
R KI
NETI
CS/P
ERM
LIGH
T
“Now is the timefor LED systemsto illuminate our homes.”Jim DeckerProgress Lighting
compoundsemiconductor.net April 2006 Compound Semiconductor34
TECHNOLOGY H B - L E D A P P L I C A T I O N S
effects: “350 mA(1 W) is the optimal drive current forlight output, power dissipation and the overall cost andcomplexity of the design,” said McClear. Rather thanrecord laboratory results for LED chip efficacy, Creehas instead talked up its production XLamp packageddevices, which produce 47 lm/W at 350 mA.
Although Cree’s view of the “sweet spot” was notshared by everyone at the February conference, it doesappear that the company is finding some success withthe approach. For example, its 1 W XLamp productsare now featuring in a range of residential luminairesmade by the lighting specialist Permlight Products.The Permlight “Embryten” fixtures are designed foruse in hallways, stairwells and bathrooms, and comewith a consumer-friendly price tag.
Permlight buys both Nichia and Cree chips for itsluminaires, but because of subtle differences in the whitecolors produced, it uses them for different types of appli-cations in the home. Cree’s XLamps feature in lampsused to illuminate floors, stairs and counters, whereasproducts incorporating Nichia chips are used in ceilinglights because the warmer white color from the Japanesecompany is said to work better with natural skin tones.
Unsurprisingly, California is at the cutting edge ofresidential SSL, with high-end homebuilders now incor-porating LED lighting into new-build houses as a mat-ter of course. Permlight’s Fernando Lynch says that alltypes of LED fixture – interior and exterior – are now
available. He thinks that SSL has passed the “litmustest” already, offering those who install the products apay-back time of just 24 months on reduced energy bills.
Progress Lighting – one of the leading residentiallighting companies in the US – agrees. It is launchinga range of LED-powered products featuring Cree andNichia chips this summer. “Now is the time for LEDsystems to illuminate our homes – not two to five yearsfrom now,” says Jim Decker, vice-president of brandmanagement at Progress.
Some new state laws could also help that penetra-tion. Title 24 in California dictates that up to 50% ofthe lighting in new homes must come from high-efficacy sources such as LEDs and halogen lamps.Even more recently, the title 22 law on hazardous wastehas added further LED-friendly restrictions by out-lawing unauthorized dumping of mercury-containingfluorescent lamps.
With the respective priorities of the lighting and LEDindustries now becoming clearer, the future manufac-turing supply chain of SSL is becoming more defined.Communication between the two is clearly an essentialpart of that process. Bill Kennedy from GaN LEDspecialist Toyoda Gosei is one of many to have issued arallying call for interested parties to join the SSLsectionof the National Electrical Manufacturers Associationindustry group to ensure that the improved communi-cation throughout the supply chain continues.
“The SSL industryis technology-focused, whereasthe generallighting industryis application-focused.”Leonard HordykTIR Systems
Imagination…
MEMS . COMPOUND SEMICONDUCTORS . PHOTONICS . ADVANCED PACKAGING . DATA STORAGE
…engineered
THINK • INNOVATE • CREATE
Whatever your inspiration, STS takes your ideas further. As a keyprovider of plasma etch and deposition technologies for over 20years, STS continues to deliver expert knowledge and advancedthinking to customers based on proven technology.
Whether you are an emerging or established enterprise we offerunrivalled experience, and unparalleled knowledge of our proventechnologies turning your innovations into reality. Above all, we listen to our customers. Their success is all the inspiration we need.To find out more visit: www.stsystems.com
compoundsemiconductor.net April 2006 Compound Semiconductor36
CS MANTECH 2006 E X H I B I T I O N G U I D E
The 2006 International Conference on Compound SemiconductorManufacturing Technology takes place on April 24–27, 2006 in Vancouver,Canada. Visit www.gaasmantech.com for more information.
Paci
fic B
allro
om
54 53
55 52
56 51
57 50
58 49
44 43
45 42
46 41
47 40
48 39
34 33
35 32
36 31
3738 29
2122 23 24
1825
1726
1627
1528
11 12 13 14
68 69 70626364
345
6667 59
6061
26
78
910
7475
7372
71
30
1920
65Vanc
ouve
r Isl
and
Room
registration
phon
es
cybe
rca
fe
elev
ator
s
bar
banq
uet
banquetbar
bar
foye
rlo
adin
g
1
stag
e
The
floor
pla
n an
d ex
hibi
tor l
ist
is c
orre
ct a
t tim
e of
goi
ng to
pre
ss.
Compound Semiconductor April 2006 compoundsemiconductor.net 37
CS MANTECH 2006 E X H I B I T I O N G U I D E
Accent Optical Technologies 24AIXTRON AG 44AROTEC/Nikko Materials 57Asahi Glass Co Ltd 51Aviza Technology Inc 31AXT 34BMR Technology Corporation 8Brewer Science 54Centrotherm GmbH + Co KG 23Corning Tropel Corporation 59Cree Inc 3 & 4Diamond Wire Technology 46Disco Hi-Tec America 66Doe & Ingalls of North Carolina Inc 67Emcore Corporation 6Engis Corporation 5EpiWorks 55Evans Analytical Group 53GE Advanced Materials 16Gold Canyon Resources Inc 42Hitachi Cable Ltd 26II-VI Incorporated – Wide Bandgap Materials Group 25
III-Vs Review 63Insaco Inc 7Intelligent Epitaxy Technology Inc 22IQE Inc 68Johnson Matthey 21Kopin 40LayTec 29Lehighton Electronics Inc. 62MAX International Engineering Group 18MicroChem Corp 2Neosemitech 45NOVASiC 56Reedholm Instruments 32Rubicon Technology Inc 69SAES Pure Gas Inc 38SAMCO International Inc 41Spectrum Processing Equipment 47Sumika Electronic Materials Inc 60Sumitomo Electric Semiconductor Materials Inc 70
Surface Technology Systems plc 15SVTAssociates Inc 61SYNOVASA 48TDI Inc 39TecHarmonic Inc 11TECHNOS International Inc 27Tegal Corporation 35Unaxis Wafer Processing 33Vacuum Engineering & Materials Co Inc 52Veeco Instruments 36Vistec Electron Beam GmbH 37Visual Photonics Epitaxy Company 50Wafer World Inc 14
BOC Edwards 49Tel: +1 978 658 5410 Fax: +1 978 658 7969E-mail: [email protected]: www.bocedwards.com/compsemi BOC Edwards offers a broad range of equipment,materials and expertise for compound semi-conductor manufacturing. We have an impressiveinstall base in compound semiconductorapplications, as well as a presence in every siliconsemiconductor fab in the world, supporting theproduction of devices such as LEDs, LDs,HEMTs, HBTs, MEMs and optical wave guides.
Freiberger Compound Materials GmbH 17Tel: +49 3731 280 0 Fax: +49 3731 280 106E-mail: [email protected]: www.fcm-germany.comFreiberger offers GaAs substrates for a widerange of applications. The product spectrumincludes both LEC and VGF grown material forsubstrates with diameters ranging from 2 inch to200 mm. Freiberger can provide superiorcharacterization data for all products.
MBE Technology Pte Ltd 10Tel: +65 67735211 Fax: +65 67735068E-mail: [email protected]: www.mbetech.com Contact: Jiang JianMBE Technology Pte Ltd manufactures MBEepi-wafers mainly for wireless communications.Equipped with multi-wafers MBE reactors.MBE Technology offers heterostructure waferson GaAs and InP substrates for a wide variety ofdevice structures such as MESFET, PHEMT,MHEMT, etc.
Riber 58Tel: +33 (0)1 4708 9250 Fax: +33 (0)1 4708 3239E-mail: [email protected]: www.riber.comRiber, with an installed base of 480 machines, isthe industry leader of MBE. Riber offers largeproduct range for high volume production,including MBE49, MBE6000 and MBE7000 formulti-4 and 6 inch. Riber systems are settingindustry standards for high wafer and deviceyields with low COO.
KLA-Tencor 43Tel: +1 408 875 2000 Fax: +1 408 875 3030E-mail: [email protected]: www.kla-tencor.comOptical surface analyzers (OSA) from KLA-Tencor automatically detect and classify surfacedefects on optoelectronic and semiconductorwafers, even transparent wafers such as sapphireand glass. OSAsystems combine multipletechnologies to simultaneously measure reflectivity and topographic variations on thesurface, enabling the detection of particles, stains,scratches, pits and bumps.
Optical Reference Systems Ltd 64Tel: +44 (0)1745 535188 Fax: +44 (0)1745 535101E-mail: [email protected]: www.ors-ltd.comORS take control – unparalleled knowledge andexpertise at the forefront of today’s cutting-edgetechnology. The world’s most advancedintelligent thin-film monitoring systems, withreal-time analysis. Accurate, stable, flexible. Fullycustomisable, time-saving and cost-effective.
Picogiga International 28ASoitec Group CompanyTel: +33 (0)1 6931 6130 Fax: +33 (0)1 6931 6179E-mail: [email protected]: www.picogiga.comExplore new frontiers in III-V performance.Rely on compound material solutions fromPicogiga International. Only Picogiga combines20 years of MBE expertise with the SoitecGroup’s global resources and industry-leadingSmart Cut™ technology. Picogiga – yourpartner at the leading edge of innovation.
Compound Semiconductor Magazine 9Tel: +44 117 929 7481 Fax: +44 117 930 1178E-mail: [email protected]: compoundsemiconductor.net BPAaudited and published monthly, CompoundSemiconductor delivers unrivalled coverage ofGaAs and III-V integrated circuits, LEDs,telecoms components and wide-bandgapsemiconductors, and it is a must-readpublication for anyone working in this industry.FREE subscriptions are available to qualifyingindividuals.
compoundsemiconductor.net April 2006 Compound Semiconductor38
PRODUCT SHOWCASE / CLASS I F I ED
Chemical delayering and planarization systemLogitech LimitedThe CDP Chemical Delayering & Planarization System from LogitechLimited (Semicon Europa, booth A1-010) offers the ideal CMP solution for III-V wafers, devices or ICs. With a low cost of ownership, the CDP isan ideal research and test facility for those requiring to delayer both single or multiple die, whilst the Logitech range of custom-developedtemplates ensure accurate, repeatable results. Optional end point detection
prevents overpolishing, whilst theprogrammable control menu allows process recipes to be storedfor future use.
Contact: Logitech Limited, Erskine Ferry Road, Old Kilpatrick,Glasgow G60 5EU, Scotland, UKTel: +44 (0)1389 875 444Fax: +44 (0)1389 890 956E-mail: [email protected]: www.logitech.uk.com
Probe stationsLake Shore Cryotronics, IncLake Shore offers both cryogenic and superconducting magnet-based probestations. They provide a platform for the measurement of magneto-transport,electrical, electro-optical, parametric, high Z, DC, RF, and microwave (up to
67GHz) properties of materials and test devices.Typical materials measured include nanoscaleelectronics, quantum wires and dots, semiconductors,superconductors and spintronic devices. Featuresinclude temperatures from 1.5K to 475K, vertical orhorizontal field superconducting magnets, up to 6 ultra-stable micro-manipulated probe arms, and up to 4 inchwafer probe capabilities. Awide selection of optionsand accessories make it possible to configure a probestation to meet your specific measurement applications.
Contact: Lake Shore CryotronicsTel: +1 614 891 2244E-mail: [email protected]: www.lakeshore.com
EngisOptimized Surface PreparationHigher levels of precision control are now achievable in the planarization,delayering and polishing of advanced materials and substrates forcompound semiconductors, thanks to the new MPC surface preparationsystem from Engis-Microtech. Incorporated into the MPC’s integralmachine design is a 6-Sigma capability for material removal, flatness andsurface finish generation. The menu-driven microprocessor controls enableoperators to fully optimize any polishing, delayering or planarization
process. Plus, the MPC’s oscillationsystem provides 3-axis motion fortrue CMP kinematics.
Contact: Engis Corporation, 105 W HINTZ Rd, Wheeling, IL 60090, USATel: +1 847 808 9400Fax: +1 847 808 9430E-mail: [email protected]: www.engismicrotech.com
Magneto-transport measurementsLake Shore Cryotronics, IncHall effect measurements on dilute magnetic semiconductors andcompound semiconductors. Resistance ranges from 10 µΩ to 200 GΩ,fields to 9 T, and temperatures from 2 to 800 K. Quantitative MobilitySpectrum Analysis software resolves individual carrier mobilities anddensities for multicarrier devices and compound semiconductors.Anomalous Hall Effect measurements for spintronics and an AC current option measures resistance down to 10 µΩ, and increasesresolution to 10 ppm at 2 mΩ.
Contact: Lake ShoreCryotronicsTel: +1 614 891 2244E-mail: [email protected]: www.lakeshore.com
Looking for further information on newproducts and jobs?Why not visit compoundsemiconductor.netand view the Buyer’s Guide too?
For more information go tohttp://compoundsemiconductor.net/buyers/ Be part of your community
To advertise in the NEW classified section of Compound Semiconductor, contact David Iddon on tel: +44 117 930 1032, or e-mail: [email protected].
compoundsemiconductor.net April 2006 Compound Semiconductor40
T ECHNOLOGY R E S E A R C H R E V I E W
Researchers from Stanford University claimto have produced the first electrically pumpeddilute-nitride VCSEL emitting in the C-band(1530–1560 nm).
The team says that its lasers could benefitlocal and metro-area networks that have beenhampered by a lack of inexpensive lasers. Thedilute-nitride lasers are cheaper to fabricatethan their InP-based counterparts deployed inexisting networks, because they are grown onlower cost GaAs substrates.
The researchers also held the previousrecord for long-wavelength electrically in-jected VCSELs of 1460 nm. The emissionwavelength was extended, explained team-
member Mark Wistey, by increasing the nitro-gen content in the active region while ensur-ing that the material did not phase-segregate.
The Stanford team’s laser was grown byMBE on an n-doped substrate and featuredthree 7.5 nm thick Ga0.62In0.38N0.03As0.94Sb0.03quantum wells surrounded by 21 nm thickGaInN0.04As0.96 barriers and a 14 µm diameteraperture. The VCSEL emitted at 1534 nmunder operating conditions of –48 °C, a drivecurrent of 110 mAand a 0.67% duty cycle.
At room temperature the spontaneous emis-sion from the device peaked at 1585nm, whichdidn’t match the cavity dimensions that weredesigned for lasing at 1540 nm. The redshift
was unintentional, but consistent with othersamples grown on the same day.
Cooling of the VCSEL was seen to reducethe emission wavelength. Lasing started at–25 °C and was most intense at –50 °C, thelimit of the cooling apparatus.
Wistey says that the group’s GaInNAsSbVCSELs should soon be delivering continuous-wave emission at room temperature, based onthe low threshold currents and relative temp-erature insensitivity of broad-area lasers pro-duced at Stanford and elsewhere.
Stanford’s C-band VCSELs providecheaper alternative for networks
Journal referenceM Wistey et al. 2006 Electron. Lett. 42 282.
High-efficiency ZnO device development has been hampered by a lack of suitablewider bandgap materials, which are needed inorder to form the barriers for quantum wellsand superlattices.
But that could be set to change thanks to aUS–Korea collaboration that has grown thefirst BeZnO films. The team says that thismaterial is better suited to ZnO devices thanthe previous candidate, MgZnO, becauseBeZnO does not phase-segregate into twoseparate alloys.
The collaboration comprised MOXtronicsand the University of Missouri in the US, and Seoul National University, JeonbukNational University, Changwon NationalUniversity and Moxtronix (a separate com-
pany) in Korea. They fabricated 0.3–0.5 µmthick films of BeZnO on sapphire substratesusing a hybrid beam-deposition techniqueinvolving polycrystalline ZnO, radical oxygen
and beryllium sources.Using a spectrometer to measure their
alloy’s transmission characteristics at roomtemperature, the researchers deduced that theBeZnO bandgap can be tuned from 3.3 to10.6 eV. X-ray diffraction measurements ofvarious films showed that the alloy does notphase-segregate into ZnO and BeO.
Researcher Yungryel Ryu from MOXtronics(the US firm) said that the team has notmeasured BeZnO’s surface roughness, butsuggested that mirror-like smoothness wouldhave a surface roughness of less than 1% of thefilm’s thickness.
Ryu also revealed that the collaboration hasmade recent progress in incorporating theBeZnO films into devices. These includevarious ZnO/BeZnO heterostructures featur-ing single or multiple quantum wells, includ-ing ultraviolet LEDs.
Journal referenceY Ryu 2006 Appl. Phys. Lett. 88 052103.
BeZnO films suggestimproved devices
W I D E - B A N D G A P A L L O Y S12
4.5 5.04.03.53.02.5lattice constant (Å)
10
8
6
4
2
ener
gy g
ap (e
V)
ZnO-hexagonal
CdSe-hexagonal
CdS-hexagonalCdO-cubic
MgO-cubic
BeO-hexagonal
latticematch line
BeZnO has a wider bandgap than ZnO, and is suitable forthe barriers in ZnO-based quantum wells and superlattices.
Next-generation mobile and satellite com-munication is set to benefit from the recordoutput power of a solid-state amplifier built byNEC Corporation that features high-efficiencyGaN field-effect transistors (FETs).
The high-power amplifier, which was fab-ricated at NEC’s research labs in Japan, is suit-able for fixed point-to-point access systems
and can produce a continuous-wave output of100 W and a pulsed output of 155 W.
The amplifier is formed from AlGaN/GaNFETs that feature field plates and a recessed gatestructure, and have a gate length and a gate–drain spacing of 0.5 and 3.5 µm, respectively.
To generate the high-power performance,
sets of 10 FETs are arranged together to forma “cell” with an input port and an output port.Twelve of these cells are connected togetherto produce the amplifier, which has a total gatewidth of 24 mm.
The record continuous-wave output wasdelivered with a power-added efficiency of31% at a 56 V drain bias. Aslightly lower gatebias of 50 V was used to produce the 155 Wpulsed output using a 10 µs pulse width and a1% duty cycle.
Amplifier deliversrecord power at 5 GHz
G A N F E T s
Journal referenceY Okamoto 2006 Electron. Lett. 42 283.
NEC’s record power-output amplifier is formed from12 cells containing 10 GaN FETs built on SiC substrates.
T E L E C O M L A S E R S
NEC
www.picogiga.com
Explore new frontiers in III-V performance. Rely on compound material
solutions from Picogiga International. Only Picogiga combines 20 years
of MBE expertise with the Soitec Group’s global resources and industry-
leading Smart Cut™ technology. Picogiga—your partner at the leading
edge of innovation. Just imagine the possibilities.
Picogiga International. A division of the Soitec Group.
B R E A K A W A Y
Solutions for a nanoscale world.™
For high-volume production of GaN-based HB LEDs, GaNzilla™ II sets the standard. GaNzilla II, Veeco’s next generation TurboDisc® reactor, improves LED yield and brightness. Plus, its unique loadlock design enables continuous operation for the highest throughput available today. Visit www.veeco.com/gan or call 1.888.24.VEECO to learn how GaNzilla II can improve your premium LED output.
Want more premium LEDsfrom your MOCVD system?
Introducing GaNzilla II: higher throughput,superior uniformity, brighter LEDs.
GaNzilla™ II MOCVD System
Compound Semiconductor SolutionsMBE, MOCVD, PVD, IBD, IBE,AFM and Optical MetrologyProcess Integration Center
