'iiiHI 6xfrP37
SPBCJAL COUBCTIONS
RELEASE AS SANITIZED
Soviet Microelectronics: Impact of Western Technology Acquisitions
Ab lateffiacMC AWHMI
Soviet Microelectronics: Impact of Western Technology Acquisitions
Kb lettfigaKr Auownt
TUi pepa was pictured by Office of Scientific andonirtbuliom
Com menu andSWR.
Sonet Microelectronics: Impact of Western Tcchnolosy Acquisltiom
Judgment*
M
acquisition of Western technology hu radically advanced (be quality and quantity of Soviet microckarooksilhonl Western technology, (beag is advanced integraled circuil (IQwe estimate to be to nine yean--would be further exaggerated. possibly by an additional decade.Without Western production technology, (be Soviets' annual output of microelectronic devices would be redoced by up toercent for discrete semiconductor devices,ercent for small- to mediam-tole ICs. and possibly more thinercent for large-scale ICs.
The successful and. in moat oases, illicit acquisition of Western technology has enabled the Soviet Union to meet the critical microelectronics needs of the military. We believe thai (he imped for Soviet military systems has been significant In the application of small- to mediam-scale ICs and revolutionary in the application of large-scale ICs. Because the Soviets arc more aggressive in their application of new technology to miliury systems than the United States, they have reduced their microelectronicslag in fielded military systems by approximately threeive yean over the lag they would nave had if they used US design philosophy. This lag reduction has been accomplished because Soviet system designers incorporate new Soviet ICs into major weapon system designs when the ICs reach pilot production in the USSR, while US system designers wail until new US ICs reach full-volume production in the United Slates. We believe thai ICs suchit microprocessors are likely to appear now as embedded components in Soviet major weapon systems, only shortly after (hey appeared in US systems.ynamic random access memories are likely to appear concurrently in both US and Soviet major weapon systems in the.
Soviet ability to produce advanced ICsuch as
We believe that, without microelectronics transfers, the Soviets would have had to delay initiating development of radars comparable toB radar for at least severalop to'ordevelopment of the required ICs
The Soviet practice of placing priority on relatively low-volume miliury microelectronics production xrui high-volume nonmiliury product too haswo-edged sword. This prioritization has enhanced applications to mililnry systems, but probably has delayed overall microelectronics
a
industrial advancement. For example. Western manufacturer* creditproductionigh portion of advances in yield and production technology. Reported Soviet yield* are extremely low compared with Western yield*
i
Theistorical reliance on following many Westerno have negative repercussions on their future microelectronics capabilities. Because the Soviets have opted to rely on the West for innovation, the United States isinimum technology lead of at least two to three yearsinimum production lead of at least three to four years. Using technology transfer to adapt Western circuit designs, production equipment, and productionhe Soviets reduced the US technology and production leads for large-scale ICs to two to three years and three to four years, respectively. This strategy has not yet produced similar progress with very large-scale ICs. and we believe thai toe USSR recently has begun to slip further behind because of technical problems inherent in very large-scale IC production and possibly because of Western multinational export controls. We expect that the currentestern lead of about eight to nine years will steadily increase.
Soviet microelec'ionicsw rne withWest Q
ecause of this, wc estimate that,arge number of production facilities, Soviet output is only aboutercent of US production of discrete devices and aboutercent of US IC production. The low output and the Soviets' continued legal and illegal acquisition of millions of low-level ICs lead us to believe that the Warsaw Pact has an across-the-board shortage of all but the most basic ICs
Since thehe Soviets probably have spent over S2 billion in the West for microelectronics acquisitions. Theieces of production equipment that we know the Soviets have acquired could outfitypical Western IC fabrication areas. In addition, we believe the Soviets have illicitlyignificant amount of Western equipment on which there is no intelligence reporting. We believe the total amount of Western equipment acquired by the Soviets could supply as much as, but probably not more than, one-third of the critical equipment for all Soviet production areas- Because this Western equipment is more capable than its Soviet
counterparts, it probaNy is used primarily on product km Una thai turn oui ihe Soviets* moil advanced ICi and wouM therefore represent an even higher percentage of the equipment uied in thesehe USSR has recently been acquiring *maller numbers of more advanced aad productive equipment concentrated in areas of Soviet technological weakness
We believe thai the Soviets' goals for their microelectronics industry include:
Improvement of their very large-scale integration capabilities.
Development of advanced very large- and uliralarge-scak integration capabilities.
Augmentation of domestic production.
Insertion of state-of-the-art ICs in future military systems.
To meet these goals, needed to improve performance of military systems, the Soviets will initially have to improve their dean room technology, circuit design capabilities, feature resolution, thin-film quality, andtesting equipment They then will aeed to produce higher quality silicon and develop advanced packaging and incubation techniques. Only extensive use of Westerngy will enable Moscow to achieve these goalsimely mannei
esent, we believe that the USSR isarge pan of its technology acquisition projram on US very highspeed integrated circuit (VHS1C) development, both to advance Soviet capabilities and to assess the impact ofn US weapons Aluo. in addition to the likely massive intelligence collection program the Soviets already target against the Strategic Defense Initiative (SDH, they ate likely to focusesources on microelectronics rescaich deriving from the SDI
BLANK)
Sovietf Wrsttn Technology Acquisition
Scapenietsmcm examinesSoviet microelecirooict and measure* (he impact
of Wfjien technoloiy on the Soviet microelectronics industry and. inthe Soviet military. On the basil of this analysis, we havemicroelectronics tails and the rcsultini technologyin the West. This paper is the firsteries of nudies ontargets by the
Medics will address, anvmg other topics, deepwater submersible technology, microelectronici automatic testing equipment, computer-controlled digital switching, hot tsosiatic press technology, and personal computer technology
(REVERSE BLANK)
Contents
Judnmemj
Foundation Technology
Soviet Capabilities and Deficiencies: Overcoming Weaknesses With Strengths
1
Microelectronics Inlrasiructcre: Industrial Centers
Techr-olosy: Wnst the Soviets Can Do 1
Capability
Product Mis
ronka Technology: What the Sonets Cannot Do and How They Compensate
Resulting From Technical Weakness
Resulting From Industrial Practices
Lag
for Technology Lag
Technology Acquisition Strategy: Using thend Fore.gn Trade _
Acquisitions: Imorovine Performance of Miliury Syslcms
Diversion Acquisitions: Increasing Industrial Efficiency
of Soviet Microelectronics Acquisitions; Industrial and Military Gains
on the Microelectronics Industry; Significant Qualitative and Quantitative Gains
Impact on Soviet Microelectronics
impact onoelectronics
on Soviet Weapoo System Effectiveness and Reliability. Closing the Technology Gap
What They Need and Will Therefore Try To Acquire 18
Microelectronics Needs for the Next Decade: Qualitative and
e Acquisitions: Critical Technology Target*
Programs of Special Interest to the Soviets: VHSIC and the SDI
Production: Sand to Circuits
oaadilioB icduaetocj with broad iprOcitKn Difitil microelectronic ctroitt arelaail parpoK in nature ud tictoadystems. These circuits have several Inherentlojic. Uirrlair', speed, low power, mull me. tod ltiehdircdty impact weapon systemsproduction, and operation (seehe United State) and Japan continue to be world leaden In oucrcetcctroaics development andIniSoviet efforts Us those of the Wot. This paper tsrvcyi the Soviet iodtolry. aiicascs ibe impact of Soriel acquisitions on tbe general technical level of tbe industry and oa specific military systems, and projects critical Soviet needs that would prevent tbe tap between US and Soviet microelectronics front
CTOwiof
Sorlet CapablUUrs and DeticieBcles: OterconMnf Wokimta Wllb Strengths
Sonet Mktortretrwaks Ie<ruin*tart. Indinrnal Ceotm
We have located over TO microeleetron^es production plmu in the Soviet Union (tee insci.of an
nt rate of increase, tbe Sonet building coAitiueiion proe-ram for microelectronics production peaked in ibe. alihoaeb pro-dnetion floonpaee has continued to fro* each year We believe that current coosiruction probabK "ille (xodaetion of more advanced IO f
eaoojb production Doorcpace to meet tbeirroals or. leu kidj. that the Soviets are baviac didtcaliy porchuios. bvuVfinf,ern for new facilities.1
The center of Soviet advancedearch, aVvrlccanf at. andate eo ptf in Zelccoirad, abiratJVxrxtcTi from Mnclenonrad is iirtended tonderies-product plaati, an cdacaliooal laatjtuu with pilot-eeodaa bnca. and atVnro acicatihe research iast talcs. Oatside of Zelenenrrad, the dor advanced production planti known are found near Leniner. Kiev, and Minil.
MkTOtoctroafcs Ti rhnalap- What the Soviets Can Da
Tbe Soviets havearteinduiiry. They are able toarte aumbet of ICs of various types, are strong in fundamental research, and turemrtly apply new nucxoclccuonics tMhootoev to anbury systeens (sec inset. "Mieroefce-troalca Development MBntonei")
PtoJwtttm Capability Although 'he USSR produces tarre number of ICi. we believe, on <he ha* nf ^
toaare mcici ofubstantially below what the US industry can achieve. If the US microelectronics industry used the Soviet production Hoonpace and produced the Soviet product mil with IIS sltndanlt and equipment, output would be aboutimes (hat estimated for the Soilcl industry.
:itrbased on our
Soviet production can* of Soviet prorluciioir
slowdown inin ;tcj ii that the Soviets may now have nearly
Secret
modem Easj_ European facilities. Usingandmodel,
wc calculate ihat Soviet yearlytillion discrete semiconductorillion small-scale integration (SSI) and medium-scale(MSI) ICs. ISOLSI) ICs,ew bandred thousand very large-scale integration (VLSI)lthough these produc lion quantities are large, they represent only about IS percent of US production of discrete devices and
ercent of US IC production. Fiirthecrnore, US ICs are generally more sophlstlcaied in the aieai of circuit complexity, operating speed, reliability, and minimum feature size than those produced in the USSR
FtpiuttMirto open sources
_J, the Soviets have developed oversac*>rece*sor types, spreadumber of technologies (TTL. STTU ECU IIU nMOS, pMOS. CMOS) andsrehrtectares. Thesebiiwhich by Waaterne first (foritecond (forit
Growth of an Industry
The world mleroeleetrontes Industry began in eamrsi In theith production la ike Untied Stairs at ike first Integratedcircuitshe Soviets recognised the Immense Impactwould have, particularly on militaryIn part because their Initial research paths soon left them far behind US developments. MoscowIts microelectronics effort and creeled the State Committee for Electronic Technologyhich5 became the Ministry of the Electronics Industry fMEf,
The MEP Is the principal ministry responsibledevelopment, and production ofandcapacitors, resistors, vacuum tubes.devices, optoelectronics, bubblemagnetic cores* The MEP also developsprocessing, engineering instrumentation, andtechnologies. Our analysis oftrends In the industry suggests that theemploys several hundred thousandMEP, as one of ntne defense Industrialdevotes well over half of its output todefense industry uses.
The MEP is not the only Soviet source ofdevices, however. Other Industrial ministries develop and manufacture small Quantities of special-purposeanalogous to the application-specific integrated circuits /ASICs) manufactured in the West. Other ministries that develop and produce ICs Include the Ministry of the Radio Industry: Ministry of Instrumentation. Automation, andSystems; Ministry of the Communications Equipment Industry; Ministry of General Machine Building (space andnd the Ministry of the Defense Industry {armor andhese ministerial production programs probably reflectto avoid excessive dependencies on otherconcents that the MEP will not be able to meet quality-quantity demands on timely schedules, and recognition that the MEP can resist developing and manufacturing prnduett in quantities that itInsltttlfican:
and low-levd third generation (forbit
ul<Future Soviet circuits
probably compile lo high-level iblrd (forbitQQ) or current fourth(forbil1 micro-prccessors produced la the West. The Soviets have radicated their intention* to use these devices inpbinaed for scries production
A higher peiceetage et ihese Sooet ICs than would be ceumderad norusal In the West (SO percent versusercent) are of bit -iliee design. Bn-alioecan be combined lo construct multiple-chip microprocessors with aword length. They arc less advanced than ccrumlcni single-chip mkro-prccessors but can offer comparable performances, although with penalties la ure and power. These factors (variety of production technologies and word lengths plus the large percentage of bit-slice designs) all combine to offer theide variety of microprocessor options, sltbough not optimiied for each application to the Client possible in the United Statesits much larger vsrlety of
In IC memory technology, the Soviets have produced dynamic rsndom access memories (DRAMs) up toevel. As with microprocessors, the Soviets have tcesd iheir memory ICs across several technologies.
.ndieate the Soviets sre attempting
to improve low production yteldiRAMs.hich achieved full-volume production in
that the Soviets have
igMri scries productionXK DRAM
Research. In terms of fundamentalout b) the Academies of Science and technicalbelieve that Soviet efforts sre generally equal to those iu the West. Research into materials, physics, chemistry, and traasiitor or diode structures sometimes surpasses US advances. For example,to US industry asscsMnenit. Soviet research and development on negative-electron-amnliy fNEAlalmost etclutrvely for military night-videmmoving beyond the United
r
Suics in some areas, although the Soviet* lacin production and application of theseAlso. Soviet work on superconducting magnetic flux detectors (used for antisubmarineu> pcrconductor-insula tor-superconductor devices, and Jotephson mixers Is of good quality and hasanticipated Western work. In applied research-feasibilitySoviet* also use<le%lgn practices to bypass production
Application- The Soviets' structured, controlled orga-niiation enables them to apply their limiteddevelopment resources directly toward military production. The Soviets more aggressively apply new microekcttonics technology to military systems than docs the West. For example, new ICs arc designed into developing military systems and
MIcrorUctroelcs Dmlopmem Milestones
Wt divide Soviet microelectronics production Into three development milestones: pilot production.series production, and full-volume production. Pilot production Is achievedesign bureau and It charaderlred by the demonstrationunctioning production process, with Integrated circuit (IQIn the hundreds per year. Production Is then transferredull-scale microelectronicsplant, where the knowledge developed at the design bureau Is used toroduction line and to begin initial talcs production by starting the first wafer through the line. At this stage. IC production averager In the thousands per year. As the production plans refines Its technology, yields Improve until they reach an upper limit dictated by the Quality of the equipment and the workers' expertise. When the yields approach this limit, full-volume production is reached, no matter how low this yield limit might be. In the USSR, full-volume production of ICser year or greater. In the United States, technology is developed In muck the same way. although the production Quantities for similar ICs are In the hundreds for pilot production, tens of thousands for Initial series production,illion or greater for full-volume production. In the United States these milestones may be achieved by annonmerchanl firm such as IBM. even though the circuits are not widely avoilableSrfi* commercial firms such as Intel or Motorola
years, ihey still face chronic problems lesated to quality control. The Sonets have ceodacuon problems primarily in the areas of Uihography. etching, and testing. They also have difficulty In tbe general area of processthe production lines to work as specified We believe that limited Soviet
aided design (CAD) capabilities hinder the development of both new ICs. suchbitand advanced or improved versions of current ICs. suchbit rnlcroprocessors and
Problems Resulting From Industrial Praetlea. The
Soviet practice of placing priority on relatively low-volume military production versus high-volume non-military production haswo-edged sword. It has enhanced applications to military systems, but has probably delayed overall Industrial advancement. For example. Western manufacturers credit volumeby vastigh portion of advances in yield and production technology. Statistical studies of process variations in huge production runs enable problems to be Quickly identified and solved. When the military productionombined with the generally lower quality of Soviet production equipment and process control, we conclude that these factors significantly hamper Sovi-ei eanabilities to increase product yields.
oviet IC production yields are ex-iremely to* compared with Western yields. IC* suchBA Ms are produced in the USSR withpercentage ofercent. In tbe United States these ICs would be oecduced with yields of aboutoercent.
Ibe Sonet* Cannot Do and How They Compensate
Although the USSR Is strong in fundamentaland aggressive application, its greatestlies in preiAw-rinnadvanced ir<
I
1
3
Problems Resulting From Technical Weakness. On the basis of the aforementioned sources and our own exploitation of Soviet ICs. we believe that, although the Soviets have made major improvements in recent
trategy of adapting Western technology, the USSR hasremendous amount of research and development resources that would otherwise have been required to achieve similar results. The Soviets' historical reliance on following many Western IC developments also has some negative repercussions on their microelectronics capabilities. The Soviet policy of relying largely on Western technologyS applied technology lead in pilot production of at least two to three years. This is the
minimum time required for (beS IC and achieve pilot production. Soviet weakness in volume production ensures itutl the United Suiei will remain at lean tbree lo four yean ahead inprcductloo capability.
Ttekmoiotj Lag. To measure ibe relativecal eaptbOtie* of the Uailed State* aad tbe USSR, we compart US Initial series production to Soviet initial series prod.ct.oc. end USvolume lo Sonet full volume. We believe that tbe SorieU msde ihciiloses!S DRAM milestone, in initial series production wilh.K DRAMs, cutting the US lead to two years6or full-volume DRAM production we believe Hot the Soviets made their closestto US milestone* withRAMs, cutting the US lend to three yean8oviet initial scries productionseme limeUS rate of progresso. We believe, however, that they have begun to slip back inK and in the development of nestchips.1
ft
Although the Soviets quickly advanced to tbe LSI level, ibe tranrilioa lo VI| -level prodwetioa ha* been slow. Two factors may have contributed lo this: technical problems Inherent in VLSI IC production,ack of suitable productionnd Western multinational esport controls The trendi (or DRAMs and miciopfocessoes are shown inndndnd tableS industry isbit microprocessors in full volume, and deliveriesmegabit DRAMs are under way. We do not anucrpaie Soviet full-volume prod wet ion of equivaieuithem lossinfy fulland induiuial consumers'theor either type of device We therefore assess the current US lead in volumeat about eight in nine years and increasing.
Tbe preceding argument focases on the quaIkaiive aspect* of the US lead by comparing the dates ai which technological milestones have been or will be reached. The quantluiive aspects of US versus Soviet capabilities are also Important. Even though evidence and analysis indicate that the Sorieu probably achieved fall-volume predictionRAMscroprocesaors. we do not tebcv* the Soviets- prodnokc. meet* then needs for these devices. In ibe West, the time between volume predict-on and lercral ar*uSbffity is sboul one year. Became we stall observe the Soviets trying to buy large volumesRAM*it mieioproces-sors-both of which sre LSI IC* that went Into fall-volume production in the USSR inbelieve it is unlikely that they have adequate supplies of ifcese or more sophisticated ICs- Jodgiag from the contin-
ern few-level SSI aad MSI ICs. we believe that the Warsaw Pact has anoard shortage of all but the meet basic ICs such as standard TTL NAND. AND. OR. and NOR logic gates;lipfiops. It ii difficult lo assess the military impact of ihis shortage because the Soviets saiiify military reqisiremenu first, leaving industrial appbes-oas to fed the pinch. However, tbe inadequate supply of IC* prevent, the Sorieu from fuHy bu ildiog upinch as Industrial process control andwith iodigeoou* produeu.
Campcnuili far TfcAaotogr Lag. Nonetheless, from the military'* view, ibis scenario, while not Ideal, could be sufficient wheo combined wilh the Soviets-aggressive practice of incorporating ne- rrucroelec-trccra itcSuwtocj into -raccc syitcnu on the bauiikanstead gTSuMgHhe .mpact of the Sono pbnosooby ofe appUcatioo is shown inndn which Soviet pilot production milestones are comoarod with US full-volume production milestones.
This impact is apparent wirb As MK DRAMthai
the Sorieu demonstrated pilot prodwetionRAM probablyhilectvined full series proowcooa in
J that the Sorieu demonttrated pilot production ofil microprocessor copy of the0 in
At
Fhvix )ih. ami m-xn a* So-.-
I inMlMand Ult-omt roSniuo"<wmnlopwl It tMM W
oteuIWft* WoA
Wb.OXt- ptflpWKCB* V$ .
i BTtmalllt nam bcoewClAI
iSU ivirW
iitnoa'
USOn* the USSR 1st Coiop.rsbic Mfcrcvproceasoei
(to wo-D
1 Hi 1
Ill
fla.taiiJ
wHa till
im
scao
till -olUKX
I9H
sow
1 ll'
Tan ubkUMi iW
:
-Miut-HuHtmn.fit.wpaaaij<*iii af
tb. MIS* SmtS
*bindSol
figured
DRAM PtodBcBou Milestones, US Venus USSR
* ikon Ac dMci ihc Un-ied Sam .no (He USSR fin*
uBMvcd IsllUI icnei proamiMe and (Uu-iitiidn
various nwi of DRAMileu boi Kn doflntiloni of IIHURAM) lie deltoid bt capiat?.IK Hull to
in of binary daU Moral Th* ipccd-hickaMi OKnic (ifttlt icifii lime) and tfcelr tedtivt alioie menu* ofil(ailon. .iinom"of
OKpkmirluio ijnerJ ihcM bom. Ii ihoolO bo need Owl
Sovkl aoWlrtawilrlr
thai Ac So->ei pan maKhnlha "Mfocmanct ofdH US modd-m
Cel. Stnifll DRAMi Mldom UHOV*loft ens.
ran of USclUaa.
Table I
d Orer tbe USSR In IlRAMs
hile the United Slates achieved lull series prodiKiiosseeausebe United States and Lbc USSR require eight toears toajor weapon system once the embedded IClaRAMs arcppear concurrently as embedded component ia both US and Soriet rnajor weapon system, ia theh microc*ccca*ort arc likely to appear now In Soviet systems, only shortly after theyin US systems. We conclude thai, by using this aggressive application philosophy, Ihe Soviets reduced Ihelr microfxecestor twhiwlogy lag In fielded military systems by approiimatcly five year* over the lag they would have had if they used US design philosophy.
The Soviet* have reduced ibeir DRAM technology lag by three to five years over the lag thevhave had if tbey used US design pm'rosoph?
Sorter Teeaaseiogy Aea-witlon Strategy-Using OstVPIaal Feeetg- Trade
The Ssmet dfort to acquire Westernequipment ha* two overlappiag parts. One la aminaged by lite Soviet Military Industrial Commit*too (VPK) of the Presidium of the Council of Ministers. The otherrade diversion progrnm
I"
-/sr.*
managed by the Mlnbliy or foreign Tradeoth these progranu have beenhrough both nd illegal trade,2 the Sovietsacquired atieces of major micro-elect rotua fabricatloa equipmeat co-ering the entire spectrum of production operations (seee believe the Soviets have spent overillion in the
lectronics acquisitions once the early
VP*nert*f Mllltan Systran*
The VTK proaramprimarily through iaieft-geneeaie-ofa-aiad military and dual-use hard*are, blueprints, productnd test equip ment to improve the technical leveU and performance of Soviet weapons, military equipment, and defense maitufacturing equipment. Most of tbe micro-electronics-related requirements originate from tbe Ministry of the Electronka Industryn addi-tion. other mm.strv* conduct studiea wrth the MEP or issue separate reqairemenu on mkroclcctrc^acs-relat-ed acquisitions, althoogbodi smaller soak. Tbe priraopal rn-nOtrie* in-ah-rf arc the Minuiry of the Radio ladtsstry. the Ministry of the Ceassnsurueaqtwprnent Irrerast-y. the Miussiry of General
Machine Building(space andnd the Minis-tty of ihc Defense Industry (armor and electro-optics) lo thehe MEP atone originated almost one-third of alt VPK requirements (almost all of these MEP requirements were in the microelectronicsemonstrating the high priority the Soviets place on microelectronics.
the Soviets have real-ired many benefits from the VPK program. Mnny Soviet rnkrcelectronics plants have establishedto reverse engineer US production equipment. They have been particularly successful in adapting proven Western designs and incorporating the Soviet adaptations into production lines. Between one-third and one-half of all identified Soviet microprocessors arc known to be adapted from US versions:
Much of the manufacturing equipment acquired is also intended for use in developing Soviet counter-ion Some of tbe significant MEParc shown la Ibe tnset on page 13
Tissue PS inils Aeaniririsnr laaaraunasg ladntrtal FmcSeney
The trade diversion program ii comparable lo the VPK program In scopeharacterised by Illegal and legal acquisitions of relatively large numbers of dual-use products. This program, apparently managed byTs Main Engiitccring and Technicalll Iprobably is less structured than the VPK program, but ii just as rigidly monitored because of the large amounts of bard currency ooces-sary
We celseve. en the basis of oar eanlyiss. -.
Jof reported Sorieihat tbe dual-use microelectronics equipment known to have bean acquired by the Soviets2 would be sufficient in equip uppeoi4 typical Weitern fabrication sreas|each0 square feet of flootspaceJTM* auvmnt representsercent of al( identified noonpscelor Soviet wafer processing. The Soviet acquisitions we have observed, however, are certainlv far below the total actually acquired by the Soviet* tT
Jmt believe that the actual number of acquisitions might provide as much as. but probably not more thad. one-third of the critical equipment for"Trent Soviet microdotsronies fabrication areas,
lieve that, in tbe past, the microelectronics diversion program has concentrated on the acquisition of raw materials, production equipment, and ICs for direct use:
High-purity raw materials such as silicon have primarily originated in the United States. West Germany, and Japan and have been diverted mostly through Europe.
?
MEP Aemmslriems in thend
6 microprocessors and documentmlom.eries microprocessors,2 microprocessors.ubies, tRuble amownri luted here are Sonet nttmates of taring*.)
Texas Instrumentsagnetic bubbleand documents on the production of gadotinl-um-galllum-garnetvblei.
Papersonference on integrated optics,the status and future developments In thatubles. ':
TRW1 analog to digital converter!ubles.
IC testers lSemry-VU and Xlneom4 million rubles.
t
7
Micro0Mes.
ubles.
Research rtpor: on field-effect transistorillion rubles.
Dataeport on poking galliumubles.
Mask making equipmenteport onillion rubles.
i(ito doocl use are vRully acquired by irictaceutral ammria una ibe*uge Kittenloc desnnaioa We etianate thaiim'uta ICs, may behn manner eachrnj oenily notedbe mti ol csaaOOcUd ICsarge Uiperscm at uncontrolled IC*
Recently, however. Soviet aco^risiuori priorities for mkroekaroaie* production equipment mar have
chanced tliihtlv. On the basis aC
ssessbe
overall rale of reported acquisitions from all Western
source* ha* fallen from an average ofiece.
per year22 to about ISO per
year3be specific types of
cquiprneni reported to have been acquired suggtii
that the acquisition rate ha* fallen by almost 50
percent in the areas of material preparation, doping.
and packaging For caidattoo. fatography. and eteh-
intt euu.pment the rate has fallen only slightly. Il has
We doubt that the overall drop in reportedecauserop in US collection eflorta, technology transfer eotatctiori has beengber priority in recent years cornnarvd with that assigned dartnc
3-
These acquisition trends are consistent with ourof Soviet progress in overcoming icchnotoel-oal detkienclea. Although the Soviets still have some problems with material preparation, doping, and packaging. In recent years these problems haveIn cootrasl, reliable sources indicate thatwith aeh-nrsced lithography.-Mteal equipment have continued
Production equipment technology is primarily of US origin Acquisitions, bowevcr. have occurred in lia-rcpe. Japan, and the United State* and have moved mainly through Europeaa coeotne* or ihrcaagbnoo-European transit ports: Kong and Singapore
We believe, however, that the redirection ol acqou. lion effort alone does not eipUm the drop in ibe overall acqiusnion rate. Other factors probably aho at Play are
Western maltiaatscatal eaportA Sonet trend toward aeqasiriog fmaller numbers ore pecdwetrve fabrication
Spent
Average Soviet Acquisition Rale of Mlcrucicrtreuks
Dpit of So.irto.lrri.onki Acquisition: Industrial aad MUlUrr Gala*
Impact en the Micro* lect.oocstatifc aad QuaatftalW Cains
Because of access to Westernhe Soviet miciocleetron>ca hsdnstry bai nude signi&an:and quantitative gains. The Qualitative ienpaet has been tbe acceleration of the develcornent of new ICs either indigenously or throughof Western ICs. The quantitative impact has been the improvement rVtechnology andoutput
QjmIImHh tmpaet on Soritl tiicroetetttonict. To assess the qualitative impact, we haveodel to estimate tbe difference between actual Soviet technological progress and hypotheticalIn which Ihc Soviets would rely solely on indigenous development. The model assumes that, while the Soviets would be aware of. and perhaps
motivated by. Western developments, they would not use technology transferethod of matching or surpassing tbe West. Under this hypothesa. however, we would aspect that the Soviets would have access lo unclassified scientific informs taoa. This methodonlyough approunutionarge uncertainty factor but does give an order of measuredict gainsfrom technology transfer
We have chosen to model the dynamic random access memory (DRAM) because it ba* been the technology-driving engine of the microelectronics industry. DRAM* have been key totechnology because their simple snd repetitive circuit design stlowi procesiing technology to be the main limiting lactor. Instead of circuit design orthe model ha* only two defining
The tin of US progress rcprcicnu"ben ewe"Theate of progress would br leu than ihe
best cade becanae of (he inherent small tin of ibe
industry and iu miliury focus.
We modeled hypothetical Sc-ocl performance fromRAM ih.-oogbRAM level
Western technoloay and added it lo date tbe Sorieu achieved initial scries production of IK DRAMs. On Ibe basis of onr judgment of Soviet IC development, we estimate that the Scfrieu saved at least four years by ibe lime they reached Ihe IK DRAM milestone. Adding this to ibe5 IK DRAM initial series production milestone. we9 as the point when the Sovicu would have achieved series production without Western technology.sing Ibe US progressin figuree est. tinned the best case rate of progress Because of the rrlstr-ely small aire of Ibe Soviet industry snd added time to
from one year for simple orcaits to four years for complicated VLSI circuits. On the basis of our Handing of endemic Soviet difficulties in precisionrocesses, we can mi ledimilartbe time required for the Sorieu lo more from initial aeries production to full-volume production-thai is, we added one or two years, depending on IC sophitticaiion. lo Ibe time required by the Unitedomparison of actual Soviet performance with technology transfer and without, as estimated by this model, is presented in figure I.hows the resulting estimate of the time savings from technology transfer that the Seven reahird at each development
The estimate suggests that byRAM initialmost recent milestone the Soviets haveUSSR saved as much atears in cumulative development lime by following stransfer strategy. Overall, the estimatefrom our model indicates that actual Soviet development with technology transfer hai progressed more rapidly thanvt been possible without technology transfer
mportant lo note, however, that this estimate does not account for potential Soviet reaction lo an increasingly widening technological gulf. For eiam-pte. we do not believe the Soviets would have passivelyycsr US leadKndeed, far smaller US lends have
I Soviet technologyester. ueJmolagy. the reoarpaaoM in keep pace with Western elevrlopcoenu inundamentally important icdmcaogkaJ area would be hkely toore aggressiveeffort, soch ascrash eVrvelepmcol program. For these reasons, il would be incorrect to assume that this estimate represents the effect of perfect eaoort controls aad their perfect enforcement.
Current Soviet efforts, which rely heavily on following Western efforts, lack the innovation and corporate memory on which Indigenous developments build. To achieve ihe US best ease rate of progress without rang iecJ*nogy transfer, the SorieU -coldcres ic the siae of their industry, <
I demandheir emcroeieetrooies pro-
QmtmtilMtiff ImpactSmitt Mirwefrcrniwiea. The quantitative impact of technology transfer on Soviet microelectronics ii more difficult to assess that the qualitative aspecu. The acquisition of Westerntechnology has allowed the SorieU totheir yield snd gross throughput figures. By copying Western equipment designs, followingmanufacturing procedures, and actually using Western equipment on many rnoduction Hues, the Sovteu have been aba* to produceiees than would have othorwiac been poasibtc-
As slated ia the -Prodoctwn Capability- section of lbs paper, we believe thai tbe upper range of Soviet productionoo discreteuTmn SSI/MSI-lcvel ICs, and ISO million LSI ICs Produc lion of discrete devices, however, does not generally require advanced equipment. The only significant advantage technology transfer would offer would be
IS
POOR (mm PAGE
production "rocedures. aRectirig production yields. On the basis of our judgment of Soviet abilities and the relative ease in production of diaciete devices, we believe that this would reduce production by no more thanercent, reducing outputillion. We do not believe that tbe increase ia discrete devicercsuhinat from technology transfer has urnift-caatly unproved Soviet military capabilities
In contrast, we believe that the increase in SSI/MSI prediction resulting from technology transfer has significantly increased Soviet military capabilities. Production of SSI/MSI-level ICs requires some rela-lively advanced equipment and productionFor esample. without Western equipment for direct use or for use as models in developing Soviei
adapts lions, we believe, on the basis of engineering judgment, that the Soviets might have been unable to adequately supply up to SO percent of tbcai current production lines. Furthermore, we believe that the Soviets' prvadacticai yields might have dropped in value by at much asercent of the value now achieved because thoir equipment would not faactMo as well without Western trchrsc4oey. These effects multiply, resulting ia as muchpercentin prod action, reducing output of ICs from today's asiessed annual peodactsonattenillion This reduction would have bad an impact on
i'ir>ireventing ttie niv-rt-ir. nf
theignificant number of military systems
Figure
Impart of Technology Tnutrfn on Soviet DRAM TcchncloQ-
Me moo
miliury impactroduction is even more significant. Production of LSI-level ICs requiresequipment subsuntially different fromfor discrete or SSI/MSMevel circuits. Without Western equipment for direct use or for use as models, wehe basis of engineering judg-raeoljihat the Sovieu might have been unable to supply up toercent of their current production lines. Furthermore, we believe that their production yields might have dropped in value by as much isercent of the value now achieved, if their equipment had not been upgraded with Western technology. These factors would have resulteduge reduction in production, reducing potential outputillion toew million. At this outputCshave been available forew of the highest priority weapon systems. These judgmenu
lead us to believe that tbe increase in LSI production resulting from technology transfer has revolutionized Soviet military capabilities, particularly in areas that require sophisticated, signal processing and computing carabih'ties, such as the next generation of Soviet fi-h'-r, with upgraded lookdown/sbootdown radars.
Imp-el on Soriet Weapon System Effeedmsess and Reliability: Closing (be Technology Cap
Without Western technology the Soviet Union could not have developed, either qualitatively orthe microeleclronies industry at its current
pace Tbcsc Impeosements in ibe Sovietindustry have enhanced Soviet miliury cups-Inline. Is general, tbe use of advanced ICs Insvstcms has several adraBtages.
Computing power msy be included in weapons, imcxoving accuracy and lethality.
Computing power Bayariety of weapon pUtfc-rms. i
au performance.
As more compact and capable ICs are used. Una weight and power are recruited to achieve ihe same 'mictions, which allows weapon payloads, military performances, and reliabilities to be increased
There are tremendous gains to be derived in "foece-multlpHer"greater flesibllily in weapons application as well as increased system survivability
For Ihe Soviets, the abiliiy to produce advanced ICs has promulgated the irrsoact of other acquisition-[J
esample. the Soviets would be unable lo upgrade" editing loobdown/sbMtdown radars on iheir neweai lighter aircraft to field muliimission radatsto iheadar used on8 without low-level LSI microelectronics technology. This type of radar relics on digital electronicschieve extremely fast dau-processing tales lo
oompenaate foe dosing rales upile perllow variable waveform nVxibimy to achieve all-aspect /all-altitude target detection capabilityStore aadrge amount* of data toiter ground mapping Radars that use hard-wirednsteadrogrammable digital signal proceaior wilh large semicamductor memory capaciiy atelt of tbcsc functions, given the volume power,weight limitation*ighter aircraft
Allhouch Ihe Soviet* have noi yelh*S rMv "C
mlcroelectton-c
transfers tbe Sowets woufd have had to delay Initial lag development of comparable radar systems for severalup to eightailing for indigenous ucvekpetent of tbe required ICs. In addition to ksshdcnrn/slsODtdown nsdars, other current or Ionheeem ag Soviet systems or subsystems that probably have benefited MgaitVantly fiom IC
and terminally guided munitions
A* the Soviet Union accelerates the introduction of ICs inio its miliury forces, these types of examples will multiply. On the basis of our analysis of VpK-assigned priorities for acquisition of Westernand ihc stated miliury applications of these aeqeiritions. we believe that the Sorieuigh priority on inserting advanced Soviet-made microesoc-ironseseir avionics, missile guidance, lank and artillery fire corarol. ani-submarioe warfare, aadsystems. Mscrotkctroanea abohasacas-cadiog irr-pacl on Ihe -ol-ame ofumerical control, and heaibk msaufaaur-sng are becoming more Important ins pan com plenty increases' Advances in ibese manufacturing technologies are all heavily dependent on microelectroo.es. Soviet improvomenii in these areas would serve notuch to increase (he speed of produaion of any particular nart but rather to boliier uniform production quality
Out look: What They Need aadTVrefc-re Try fo Acrnrtr*
Soviet Mkreeketraancs Keeda fee the Next Decade:
To improve performance capabilities. Ihc Soviet Union win continue lo insen advancedtechnology inio new or upgraded miliaryems-This effort will be most apparent in major weapon looiysiems. such a*issile guidance. Unk and artillery fire control, aniiiubmarine warfare.
automated command and control, indnvf.itf. To menesired goals in ihesc inu.hai ibe Sorieu must improve iheir LSI production capabihtiei andleast develop moderate level VLSI production capabilities Tbe SovieU re portcdly intend lo develop ibeae types of VLSI cspa-biliiica by the. Ideally, ihe SorieU probably hope lo develop advanced VLSI capabilities and some ultra lane-scale iniestraiion (ULSI)by the, with ULSI capabilitiesat and following tbe turn of the century. In addition, the Soviets sriB need to cupand their praduc-lioa capacity for all types ofcontinue to acquire tbesn in bug*more rnkrceJec-tronics arc introdnccd inlo critical military-related
To further develop iu qualitative microelectronics capabilities In the VLSI area, we believe the USSR will need to radically improve iu dean roomcircuit design capabilities, feature resolution, thin-film quality, and automatic testing equipment To progress lo Ihe advanced VLSI and ULSI levels, the USSR urul also need lo produce higher purity silicon with uniform doping sad to develop advaDced packaging and metabration techniques To develop iu quantitative capabilities. Ihc Soviet Union needsn rod hoc more and belter automated equipment into iu production facilities. Advanced process control equipment is required lo increase productionand production yields Theseof -hieb will require Western technology acquisiti. arc summarised in Uble )
Future Acqaasilaons: Critical Tnrtsntlsgj Targets To improve their clean room technology and thereby increase device yields, (he SorieU need lo develop or acquire high-efficiency particulate air (HEPA) filters and (hequi led to use iheen in sn overall clean loom layout. Proper clean room design also Involves early planning of optimum nonturbulcntpatternseep whatever particulate conUmi nanu that pass through the Slurs away from the wafer procession; area Beyond air filters and clean room design, the Sovietsore difficult rsrobiem of disciplining iheir production workers to folio* through with the annoying. tinK-ccsuumnwrequired to keep their clean room* dean
For the Sorieu. ihe most imporuat reqoirensent forcircuithe use of computer-aided design (CAD) eouipanesH. Soviet cnpsbaUies sre well behind Western standards and are inadequate to meet projected Soviet needs.esult, we believe the USSR willigh eriority on acquiring ibis equipment from the Wert
To Improve their feature resolutioneasure of riicuit density) to the moderate VLSI level, the Sorieu need to acquire boiler tonography and etching eouirjmeot. In particular, the Soviet* probably winprotection alignen snd dry etchers The Soviets' production capacity problem may initiallyem to acquire scanning projection aligners with high throughput and resolutions acceptable up to Ihc DRAM level. For more advancedsuchegaUt DRAMs. however, (he Sovieu probably will concentrate on sieppingalignen and electron-buamay eapotureall with sarmicron resolution capability. Ia etching, we believe the Sovietsseek reactive ion etching) and chemical plasma lysiems forsilicon, nitrides, eaudsa. snd resists, ss -ell as ie* milling system* for etching nsetals
To improve thin-film quality, the Soviets probably will concentrate on acquiring both epitaxial and noa-epitaxial deposition equipment. This equipment will be needed lo fabricate eomples iiructures and thin: high-performance diodes andn epiuuial equipment the USSR win need almost all equipment types, but especially rrraeeulai beam epitaay (MBR) and metal-organic chemical vapor deposition (MOCVDJ. For ncasepitaiaalhe Sovieu wdl ieeh low-pressure and plssma-enbsnccd CVD <Miiirmr.il (LPCVD and| la addition to work in rilicon-based ICs, many of these techniques sre also critical for developing and producing ICs based on compoundnch ss gallium ariemde, that offer signincani apeedndiadon hardness improvement* over silicon
OmtoVLSI .nd
Ibkc uaiocC-tho-in ICi Wo military muim
Improveiwni. Nodod
Oaf rooa lnhnSos. CWcMl
FetiD'r ntotutiaa
Tliin.Bfofl-lr.,
Higl.4aa.ty mo-aridit
.jedfie IO High-tpeer)at
di, tu^iialot con-atm
those mine (be Crochralski method to grow uniformly doped mcnocrystallirtenches in diameter or cream. The purity of this material mast be improved, requiring better polycrystallincIn addition to acquiring better Qualityihe Soviets will also need to expand their silicon production capacity by over ooe-third to meet what we estimate will be their needs ia (he. The USSR might Increase production facilities orat its three known major pdysilicon plants, but optimally wouldew polysilicon plan!in the eastern USSR to disperse its production base for strategiche Soviets wiB also require an expanded capacity to convert this polysili-COB into mooosilkon
, As Soviet ICs more into the advanced VLSI range, the Soviets will encounter packaging and metalimtion problems similar to those now encountered in the West. Advanced VLSI and ULSI ICs will require several hundred terminals for external connections and will dissipate several watts of power. These factors combine to make packagingimiting factor equaleature resolution, thelimitation. As more and more functions arc packed into one IC, metal interconnections will be placed more closely together and will be laid down in (wo or more layers. This is beyond current Soviet capabilities, and (he Soviets will need advanced enuin-ment and know-how to overcome this problem
To overcome slack quality control procedures and normal variations in human performancearticulates that cause device failure, tbe Soviets will emphasize automated equipment for the critical areas lisied above. This emphasis on automated equip menl will also require (he Soviets to acquire new equipment for all process steps, not only those high-lighled above. Acquisition of process control equip menl probably will con cent rale on parametric testers and materials characterisation. This equipment would enable (he Soviets to locate problem areas qukklv when (hey occur on (he wafer processing line
improve circuit yields, especially for advanced VLSI and ULSI circuits, the Soviets need more siliconigher purity. To improve tbeir silicon quality, (he Soviets need automated crystal pullers.
In addition io (be acquisition of advanced production equipment for the production of VLSI-level ICs. we
estimate ibe Soviet* willontinuing aeodcquire finished IC* from (he Weal. These will be sued to meet amsumptitrfi where indigenousvolume dee* nor suffice and so provide siaie-of-Ibe-art ICs that arc beyond Soviet capabilities to prodnee. The moat ngniheam acqatsiuorts will be high-density memory chips, hiea-specd rnicrr^oces-ices. application-specific ICs. and highspeed analog-to-digital and digitalna log converters
US Program of Saecial laueest to the Soviets: VHSJC and the SOI
For the last IS years, US military mkroeiectronics technology has followed civilian advances and has thus lagged consumer application. The Soviet lag between development of new ICa and their application in fielded military systems has been shorter than thattbe United States. The effect has been to reduce the Soviet military technology lag below the bg implied by the relative capabilities of Soviet versus US industryhole. This condition cot Id be changing, however, because of two US militaryVHSIC (very high-speed integrated circuits) and the SOI (Sira tack Dafcnae Initiative I
epartment of Defense program intended to insert (he most advanced VLSI tecfaacaogyfrom US industry into critical new weapon systemi Civilian production technology is currently being used, but ihe circuits developed will be more advanced than those commercially available and will be oplirnired for military uses, instead of beingfrom primarily civilian applications as ispractice. A* the VHSIC program progresses, it will push forward the stale of the art in production technology, much as military programs pushedtechnology in the, and as commercial initiative* have poshed tbe technology since then The ultimate goal ofo apply the most advanced microeacciroauc* availablell appropriate military ayautna If thiseached, the impact will be to advance US military eJeetrccoc* capabrLtres Because of VHSIC'i militarywe believe thearge part of its lecboology acquititton program on VHSICboth to advance Soviet capabilities 1 ssess the impact of VHSIC on US weapons
The SDI currentlyeas direct effect on US military raicroelrxtrtmka than VKS1C. but ii could become more sagnthcaat ia future programs Industry experts believe that roserearlectronic* will be one of lac key eaabitng factors for space-based snissile defensend thai any development resaltmg from the SDI will require massive use of ULSI microelectronic* and it* computer technology derived therefrom This would push the US microelectronics industry to achieve new breakthroughs, well beyond ceraetnnoraryw aacaairaa Scwiot-capabiBa addirint to the likely massive intelligence ccHlcctaon program the Soviets already target against the SDI. they arc likely to focus signifies Dt resource! on microelectronic* rcacarcfa deriving from the SDI.
(REVERSE BLANK)
Appendix
Microckctroiiics Production: StDd to Circuits
Oerrlrw
Microelectronic integrated circuits (ICi) arc used toride variety of electrical functions. These ICsa rnicroscopiccircuits that previously incorporated hundreds or thousands of individual,lectronicsuch as transistors, diodes, capacitors, and resistors. Fabricating these circuitsicroscopic scale Increases system reliability and performance, while decreasing sire and power requirements. In this appendix, we will describe the variety of physical and chemical processes required to fabricate an IC.ICs will require different combinations and permutations of these steps, but the fundamental process remains tbe same
The basic microelectronics production unithin disk of siliconafer, which can contain hundreds of individual ICs. As more ICs are packed onto one wafer, the production process becomes more economical. Microelectronics production begins with making wafers from sand or quartz. Wafers arc then putepetitive procedure in which amask is first patterned on the wafer, and then some physical process is accomplishedhange the electrical properties of the eiposed portions of the wafer. After this repetitive process Is completed, the wafer is diced into hundreds of identical ICs, which are Ihen packaged and tested. These key operations arc known as crystal growth and wafer preparation, lithography, wafer rmxeasing. assembly, and testing (see
Crystal Growth and Wafer Prepsratinn
The bask material for almost all microelectronics is silicon. Silicon in Its raw form is abundant, primarily as silicon dioxide. Before it can be used in mkfoelec-ironies, however, it must be separated and purified in
a procedure known as wafer fabrication. It is first processed9 percent purerslycrystalline silicon, or porysilicon, in which the internal structureil lure of all possible crystal orientation* This procesa I* carried outurnace, and the finished poiyaibeon resemble* rocks These chunks of poryulKDw arc then meltedrystal patter, whscfalewd- of ihe desired crystal onewution into the melt and pulb it out slowly, allowing Use molten tiboon lo solidify on tbe seed. It is PuHcd in tbe formylinderoule. After this pulling process, the internal structure of the bouleniform crystal onentation and is known as rnooocrystalliiM silicon, or monoadicon. The bonk is sliced into disks called wafers, which ire lapped to the desired thickness and pnllihcd lo produce an almost perfectly Aai lorfacc
lithography
Idircuit designeromputer-sided design (CAD) system to design the electrical circuit desired andranslate that idealisedirpresentaiionultilevel physical IChemical layeresist thai i* sensitive to tbe rsdiatioa source to belight, ultraviolet light, electronays, or ionto the wafer. One level of the IC is patterned onto the resist using proaimity aligner* (oidscanning or iieppiag projection sligncrs Irdlectrontems (advancedsy sligncrs (research stageor ion-beam systems (exploratory research).tbe exposed resist or Ihe uiMinoscd tesisi is washed away,afer ptocesalng step to be carried out on Ihe desired poeiloni ol the underlying layer
POOR QUALITY PAGE
Wafer Ptoonilos: Itching. Depoaltiors, anal Uopiaf
Etctnng
Etchingroctaanch portion* of the wafer surface revealed during utaography arc selectively removed The two basic lypes of clctiliig are *ti (acid) etching and dry etching Wo etching tat the cadet lypctill used for relatively simple ICs with line widths greatericrons. Wet etching canoe* be used much below (hat leal ait sice because of its tendency to etch sideways at the same time as it etches downward, causing (he Itoea to spread and merge together To overcome thb drawback, drysed for advanced ICs. The major dry technique* are chemical plasma etching, ion milling, reactive ion etching, and reactive lon-bcam etching-Each has different characicrU(lcs of (hroughput, spread, and material selectivity. Current dry-etching systemi are capable of etching line* down torticrons wide, far beyond what currcnl IC designs require
Deaoairioai
Deposition can be divided into two categories, epjtaxi-al and acasecataxial Fpataiialhe most ahnacuh to achieve, and require* that ibe crystal structure of the wafer be coniinwcd throngs the deposited layer, fhe three basic (ypea of epitaxy arc liquid-phase cpiiaayPI: i. vapor-phase ep*taxynd molecular-bcam epitaxy iMhe oldest teehnxjue and has been overtaken by the popular VPE, which include* metal-organic chemical vapor depositionhed-vaneed technique and produce* the beat result* in terms of (harp doping profile* aid utility for exoticnd II-VI oon.pound semiconductor materials.
silicide* on the wafer to act a* interconnect* between individual device* on each IC. The two methods of PVD are evaporation andaponlib ibe oonvt"ihwiiI method for metal deposition, but ii has been replaced by sputtering ia advancedhe major drawbacksajinra-aiica arc the difficulty in controlling alloy compoiitioo and the nonuniform coverage of steps on (he waferapors tors are classified by (he method offilament, electron beam, flash, and induction.ften the deposition method of choice because of its ability lo produce bigbquality filmsigh rate of growth and at lower temper*lores. The basic types of spattering system* are electron beam, diode, [node, and rnagnctron
Dopiag
Doping is the coo trot led introduction of precise quan-tiiie* of irnpuritiea, or dopants, into certain portion* of the wafer ia order to achieve desired electricalThe conventional dopingiffu-saoo foresee, which relies on beat to spread dopants steadily from the wafer surface into the depths of tbe wafer. The other, more recenton int-pbeiatlon,he direct injection of dopant atom* into (he wafer. One of tbe major advantages of implantation overetter control of doping profiles because of lower processhe two most important characteristic* of implantationaidme the Dumber of sons that reach (he wafer,ontrolled by beam current, and (b) junctiondepth beneath the surface where the ion* stop, which is controlled by beam energy, loo implanters. which arc cbssined by current andhave Ihreeiedium current, highand high energy
Ncmepitaxial chemical vapor deposition (CVD) and physical vapor deposition (PVD) arc leas demanding processes. The four basic types of nonepiuxial CVD are atmospheric, low pressure (LPCVD1 plasma(PECVDk and phcacchcnueal (PCVD) CVD can be used to deposit many matenab, but those generally encountered (other than epitaxial silicon) are pdycryttalline silicon, silicon dioxide, and silicon nitride. PVD is used lo deposit thin layeri of melab or
Assembly
he step following wafer processing. After the individualthe wafer are tawed apart, tbe functional parts (see (eating section) are attachedackage in the die-bonding step. The
2S
three type* of die bonding are eulcctic, preform, and cposy. Die bonding it followed by wire bonding, which is ihe major method usedonnect tbe die electrically to terminals leading outside ihc package. Other techniques include flip-chip, solder bump, beam lead, and film bonding, but none of these has replaced wire bonding. Tbe final step is encapsulation, during which the IC is ecckoed in plastic or ceramic
Testing
Testers can be divided into two categories, wafer probe and packaged IC. The wafer probe tester is used for rapid testing of dice onmall number of tests are made, but not at the circuit's operating speed. The purpose of the test is to mark bad die before time and money arc invested in packaging. Packaged IC testers can be divided into burn-in systems snd functional testers. Bum-inre used to identify quickly the parts that would fail soon after they are first used. Batches of IC* are loadedemperature- and humidity-controlled chamber, powered up. and allowed to sitength of time to simulate longer term normal use. Mostpans fall at this stage, liter easing tbe reliability of the systems that will use those ICs that past the burn-in test. This testing is especially important in military equipment. Functional testers are used to serify thai an IC work* properly at iu intended operating speed. These testers are usually classified by the highest level of integration able to be handled: for example. MSI. LSI. or VLSI. VLSI testersextensive computing power and softwarewhich mutt be updated for each new product development, causing these systems to resemhUtadvanced computers more than teste"
Original document.
Comment about this article, ask questions, or add new information about this topic: