Fabe Th : o t b FroLa e mTh to Integrated Circuits Howar . HufdR f International SEMATECH 2706 Montopolis Drive Austin, TX 78741 Abstract. actio experimentalls nwa y observe Johy db n Bardee Walted nan r Brattai n-typn i e polycrystalline germanium on December 16, 1947 (and subsequently polycrystalline ) as a result of the judicious placement of gold-plated probe tip nearbn si y single crystal polycrystallingraine th f so e material (i.e. point-contace th , t amplifier, often point-contace referreth s a o dt t transistor).The device configuration exploite inversioe dth ne layeth s a r channel through which most of the emitted (minority) carriers were transported from the emitter to the collector. The point-contact transistor was manufactured for ten years starting in 1951 by the Western Electric Division of AT&T. The a priori tuning of the point-contact transistor parameters, however, was not simple inasmuch as the device was dependent detailee onth d surface structure and, therefore, very sensitiv humidito et temperaturd yan s welea s exhibitina l g high noise levels. Accordingly, the devices differed significantly in their characteristics and electrical instabilities leading to "burnout" were not uncommon. With the implementation of crystalline semiconductor materials in the early 1950s, however, p-n junction (bulk) transistors began replacing the point-contact transistor, silicon began replacing germanium antransfee dth transistof o r r accelerated b technologfa e th shale o t W y.b l frola revie e m th historicae wth l rout whicy eb h single crystalline materials were develope accompanyine th d dan g methodologie transistof o s r fabrication, leadine th o gt Integratee onseth f o t d Circuit (1C) era. Finally, highlight earle th y f a wil o s1e year Creviewee er th b l f o s d froe mth 256 bit through the 4M DRAM. Elements of 1C scaling and the role of Moore's Law in setting the parameters by which the 1C industry's growt monitores hwa d wil discussede lb . INTRODUCTION replacing the point-contact transistor, silicon began replacing germanium [5,7,8] and the Transistor action was experimentally transfer of transistor technology from the lab to observed by John Bardeen and Walter Brattain in acceleratedb fa e th . n-type polycrystalline germaniu n Decembemo r We shall briefly review the historical route 16, 1947 (and subsequently polycrystalline by which single crystalline materials were silicon) as a result of the judicious placement of develope e accompanyinth d an d g methodologies gold-plated probe tipn nearbi s y single crystal of bipolar transistor fabrication (i.e., grown grains of the polycrystalline material (i.e., the junction, allo diffused)d yan e oxidTh . e masking point-contact semiconductor amplifier, often and photolithographic technique of Carl Frosch e point-contacreferreth s a o t d t transistor) [1-3]. and Link Derick [9,10] and its embodiment in the e devicTh e configuration exploite e inversioth d n mesa process e utilizatioth , e silicoth f no n oxide e channelayeth s a r l through whiche mosth f o t e passivatiofoth r e silicoth f o n surfacy b e emitted (minority) carriers were presumed to be Mohammed (John) Atall colleagued aan s [11d an ] transported from the emitter to the collector. The the development of the planar silicon transistor point-contact transistor was manufactured for ten by (i.e., the planar process) [12-15] years startin e Wester th n 195i g y b 1 n Electric whereby the SiO2 masking layer, utilized in the Divisio f AT&no prioriTa e [4]Th . tunine th f go fabricatio f diffuseno d silicon transistors s leftwa , point-contact transistor parameters, however, was ie npassivatio th plac n r junctionp- fo e f o n s not simple inasmuch as the device was dependent intersecting the wafer surface set the stage for on the detailed surface structure and, therefore, MOSFET fabrication as well as the utilization of very sensitive to humidity and temperature as the dielectric layer for supporting metallic well as exhibiting high noise levels. Accordingly, conductor overlayers in the (1C) e deviceth s differed significantl n i theiy r era [16]. characteristic electricad an s l instabilities leading The Si-SiO2 diffusion technology, transferred from to "burnout" were not uncommon [5]. With the AT&T's Bell Telephone Laboratories (BTL) to implementation of single crystalline Shockley Semiconductor and, therefore, to materials in the early 1950s [3,6- Semiconductor Corporation led to the phenomenon of 8], however, p-n junction (bulk) transistors began "Silicon e creatioValley e th 1C th d industryf an "no .

This manuscript was published previously in VLSI Process Integration 2003-06V P S III,EC , 15-67 (2003).

CP683, Characterization and Metrology for VLSI Technology: 2003 International Conference, . SeilerDiebold. G C edite. . . ShaffnerJ D A , . y dT b , McDonald. R , . SeculZollner. S ,M . Khosla . P E . a R d , an , © 2003 American Institute of Physics 0-7354-0152-7/03/$20.00 3 Indeed, Gordon Moore has noted "... and you are transistosilicoe th vera r ns fo y rcontroversiawa l matter once again reminded tha tlongeo n thi s i s r jusn a t at that time, althoug e importancth h f high-purito e y industry economin a t bu ,cultura d can l phenomenon, material to achieve a high rectification characteristic cruciaa modere l heare forcth th f t o tea n world" [17]. was understood. Gordon Tea notes lha d [58]: criticae Th l rol f Joheo n Moll's n laboratori L BT t ya 1954 and the development of the oxidation, diffusion, "Bill Shockley was opposed to the work on lithography, aluminum metallizatiod an n germanium single crystals whe nI suggeste , dit thermocompression bonding techniques for the because, as he has publicly stated on several fabricatio e junctioth f o nn transistor d siliconan s - important occasions e thoughh , t that transistor controlled rectifier [18-21] conjunction i , n with Nick science could be elicited from small specimens Holonyak [22], are reviewed. of polycrystalline masses of material." oxidatioe Th n kinetic f silicoo s Brucy nb e Deal and Andy Grove [23] e explicatioth , e chargth f no e Teal, however, was a proponent of the criticality drifd an t mechanism Si-SiC>e th n si 2 syste Deamy b t e l of single-crystal materials for the electronics era, al. [24-30] and the role of Pieter Balk [31,32] in recognizing that Shockley's bipolar junction transistor emphasizing the importance of subsequent hydrogen characteristics [59-62 single-crystan ]i l germanium [63- and nitrogen annealing are briefly discussed. The mesa 66] and silicon [67,68] would be substantially better and and planar processes described abovy e pavewa e dth more reproducible than those of polycrystalline material inventioe foth r 1Jace y Cb th kf n o Kilb 195n yi 8 [33- [6-8,58]. The limited capability for the fabrication of 36] (utilizing the mesa methodology in germanium) single crystal materials further exacerbate situatione dth . Noycb andBo 195n ei 9 [37-39] (utilizin planae gth r Indeed, Shockley later realized the shortcomings in his procedure, e i.e.subsequenn th i silicon, d an ) t previous assessmen usefulnese th f o t necessitd an s f yo microprocessor era [40-42]. The critical differences single crystals [69,70]. o patentbetweetw e sth n (i.e.e interconnectioth , n Teal believed the fundamental property of a methodology e clarifiear ) y Walteb d r Runyad an n crystalline semiconductor, which would s resulit n i t Kenneth Bean [43]. technological importance, was the easily controllable 4 e th earlo e t t Th ybi 1yeare 6 Cth fro25 f so e mth and spatially variable concentration, typmobilitd ean y DRAM e theMar n reviewed [44], buildinb Bo n go of free carriers, whic indees hwa case dth efoune b o dt Dennard's one transistor cell structure [45] and [71-78]. Accordin Teao gt l [6-8,58,79-81]: associated scaling methodology [46-49]. Gordon Moore's remarkably prescient assessment that the I reasone" d that polycrystalline germanium, numbe f memoro r y bits would doubl r yeaepe r (now wits it variationh n i resistivits sit d an y take abous na months)8 1 t , enshrine Moore's da s law, randomly occurring grain boundaries, twins becam e productivitth e y criterio y whicb n e 1Cth h and crystal defects that acte s uncontrolleda d industry grew at about a 25% compound annual resistances, electron or hole emitters and traps growth rate [50-54 s illustratea ] Internationae th n di l would affect transistor operation i n Technology Roadmap for (ITRS) uncontrolled ways. Additionally, it seemed to [55]. More than just monitoring productivity, whether me that use of this material to produce many by staying on the productivity curve or increasing complex units mean identicale b o t , with close manufacturing effectiveness, however, is required. performance tolerances, woul inconsistente db Rather, modeling productivity—the identificatiof o n with high yields and, therefore, also with low new productivity measures—i requirew sno d [56]. costs. Even in developing complex transistor Finally, potential direction r enhancefo s d 1C devices, it seemed to me essential to have a performance ITRe th r S pe [55], briefle ar , y discussed. high-perfection, high-purity controlled These include both carrier transport mechanisme th n si composition semiconducto orden i r achievo rt e channel using variously strained structure enhanco st e a separatio f variouno s available electrod nan the carrie r w MOSFEmobilitne d an Ty device hole conduction processes in order to analyze configurations, including various vertical transistor and understan operatioe dth f thesno e devices configurations [57]. and thus to finally achieve an optimum functiona themf o e .us l Single-Crystal Growth My genera singlle aimth r esfo crystal research Polycrystalline germaniu silicod man n were th e wer s follows eproduca o t ) (1 :conductine a g basic materials used at BTL and elsewhere for transistor medium in which a high degree of lattice research and development in the late 1940's inasmuch perfection, of uniformity of structure and of e utilizatioth s a f singlno e crystal f germaniuo s d man chemical purity is attained; and (2) to build into this highly perfect medium in a controlled crystal diameter. Furthermore s notewa y b dt i , way the required resistivities and electrical Tea 195n i l 2s zea hi thadeveloo t l n i t psingla e boundarie o t giv s a variete f o devicy e crystal transisto 195n ri 0 [79,80,86]: possibilities by control of the chemical composition (i.e., donor and acceptor varietA " f methodyo bees sha n employey db concentration) alon e directioth g f singlo n e various experimentalist o product s e single crystal growth. " crystals of metals, salts, insulators, and semiconductors. It will be apparent from the The successful initial results obtained in a scientific literature and the description that joint program between Tead Johan ln Little, follows that the pulling method of growing begu n Septembei n r 1948, resulte n severai d l single crystals of germanium differs germanium rods with some large single crystaln si materially froe pullinth m g methodf o s the y pullinmb g fro ma mel t [63,79-81]e Th . Czochralski, Kyropoulos, Gomberz, Hoyem technique employe a single-crystad l seed and Tyndall, and others. There are differences (oriented in a <111> or <100> direction), seed materialse inth , designs operatiod an , e th f no rotation and precise temperature control of the crystal batch growing equipment. Also, novel melt-solid interface from whic e crystath hs wa l techniques were developed to produce pulled [6,82-84] (see Figure 1) [84]. These were germanium single crystal n i whics e th h vital factors in the attainment of single crystals in impurity compositio d lattican n e perfection whic e essentiath h l semiconducting properties are controlled throughout the crystal. The use became highly controlled e methoTh . s i d of a pulling method for germanium and the variously referred to as pulling, the Teal-Little employmen w techniquene e b f o o t s process [63,82 , somewhaor ] t inappropriatelys a , described have been vital factors in the e Czochralskth i (CZ) process. Generallys i t i , attainmen f o single-crystat n i whicl e th h simply called the CZ technique, after Czochralski essential semiconducting properties are highly wh n 191oi 8 withdrew thin single-crystal metal controlled. " filaments fro mmela t [85]. Czochralskt no d di i single-crystaa e us l seed, however apparentld an , y In retrospect, it should be noted that during the t recognizdino d e significancth e f directlo e y 1940's the concept, let alone the necessity and controlling the melt temperature to control the usefulnes f o singls e crystal t materialsno s wa ,

graphit* cructbl« h»at«r therm*! shield

Figur . e1 Schematic illustratio typicaa f no l Czochralski puller wit zonet hho , automatic optica imagd an l e sensing diameter controls and wire reeling system [84]. Reprinted with permission from Elsevier Science. appreciate r semiconductofo d r applications. Teal's density of the solid and liquid phases, respectively, emphasi preparatioe th n o s characterizatiod nan f no and e effectivketh , e distribution coefficiente th s i , single crystal material, however, facilitated e solutth rati f eo o concentratio e crystath n i nl experimental verification of a number of quantum adjacen e melt-crystath o t t l interface relative th o et theoretical concepts developed for electrons and concentration in the bulk liquid [102-104]. The holes in crystalline semiconductors such as effective relationship betwee e effectivth n e distribution mass, drifconductivitd an t y mobility, carrier lifetime coefficient e equilibriuth , d kean , m distribution and tunneling [6,58 clarificatiod ]an numbea f no f o r coefficient, ko, during CZ crystal growth with the phenomena in p-n junctions [87] and, indeed, crystal growth rate f , (later recognizee th e b o t d exhibited significantly improved characteristics microscopic crystal growth rat Kenjy eb i Morizann ei compared to polycrystalline samples. For example, conjunction with Gus Witt and Harry Gatos [105]; Haynes observed in early 1949 that the lifetime of the stagnant boundary layer thickness melte th t -a , 8 , minority carriers in single crystals of germanium was crystal interface e bulth kd melan ; t flow conditions as much as 140 ^is (20 to 300 times greater than influencing the solute diffusion coefficient, D, was observe n i polycrystallind e germaniumd an ) describe e Burton-Prim-Slichteth y b d r theors a y mobilities thre o fout e r timee th s o highert e du , described in equation 2 [106,107]. This equation was greater perfection and purity (45 ohm-cm) of the instrumental in conjunction with equation 1, in single crystals [6,58,88-90]. Teal also reported facilitatin e availabilitth g f o singly e crystalf o s injected carrier lifetimes greater than 200 jas in single germanium and silicon with a specific distribution of crystal germaniu comparems a significantlo dt y lower dopant impurity [7,63,79,80,82,108 discussee b o t ] d carrier lifetimes of 1-5 jas in polycrystalline below n extensivA . e summar e equilibriuth f o y m germanium [58]. By the early 1950's, all distribution coefficient solubilitied an s varieta r sfo y investigator semiconductine th f so g propertie- p d san of element n crystallini s e germaniu d silicoman n n junction studies of germanium [63-66] and silicon were summarize Forresy db t Trumbore [109]. [6,67,68,91,92] preferred to use pulled single crystals. Teal filed for a p-n junction patent in single ke(dl/ds) -1 (D crystal germaniu m195n i firse 0th t[93 d bipolaan ] r

junction transistor (n-p-n) was achieved in single = k0[k0+(l-k0)exp(-f6/D)] (2) crystal germanium (grown-junction techniquey b ) Shockley, Morgan Spark d Tea an n s195i l 1 [66], e rol f microscopiTh o e c fluctuatione th n i s three years after the discovery of transistor action by dopant distribution, both radiall axialld yan y along Bardeen and Brattain [1,2]. crystale th , wer havo et e profound implicationn so e conversioTh f germaniuo n d silicoman n device performance to this day [8]. Pfann and ore o metallurgicat s l grade materia d theian l r colleagues initiated methodologies to rectify this subsequent purification during the 1940's has been situation [110]. reviewed by Frederick Seitz and colleagues [6,94- Finally a rathe, r innocuous observation that 96]. Seitz alse co-editoth o s initiatef wa o r d an d silicon crystals grown by the CZ method contain the venerable Solid State Physics - Advances in part r milliope s n atomic (ppma) concentrationf o s Research and Applications series [97], servicing oxygen [111-113] (due to the dissolution of the e physice needth th f o s s community from 1955 quartz crucibl moltee th y d eb n ha silicon)s ha d an , onwards. Norma . EinsprucnG h later initiated dan extremely important repercussion e presenth o t s t was the editor of the invaluable VLSI Electronics day [114,115] e higheTh . r melting poin f silicoo t n Microstructure Science series servicing the 1C at 1420°C, compare o germaniut d t m937°Ca , community [98]. resulte n i contaminatiod n usin e previousth g , Althoug t receivhno e Teaacclaid th edi l m conventional graphite containers. This accorded Bardeen, Brattain and Shockley, his contamination was overcome by utilizing fused pioneering research and implementation of single- quartz crucible o hol t se liqui th d d silicon rather crystal silicon technolog e 1Ce th basi th f o s a y than graphite, with the unintended consequence microelectronics revolution can hardly be over- thamele th t t became saturated with oxygen. This estimated [7,69,99-101] e descriptioTh . f dopano n t topic and methodologies for the localized removal distribution during single-crystal growth by normal oxygee oth f n fro near-surface mth e regione th f so freezing (see equation 1) was described by William fabricated device and its utilization as effective e relateth Pfann a dvi , zone-refining techniques internal gettering bulk sites beeha s n extensively [102-104], where dopanCe sth (g s )i t concentratiot na discussed [7,8,115,116 e furtheb d t wilan no ]rl fractioe th f crystao n l solidified initiae th , s Qi g l, discussed in this review. concentratio dopanf nliquido e e th th n e i td ,d ar sd an i Bipolar Transistor Fabrication especially after improved technique controo st base th l e width—and thus increase the frequency response, an Grown-Junction Bipolar Transistors initial limitation [128]—were subsequently developed [129]. Specialized techniques to improve the point- The path by which the role of group HI and V contact transistor characteristics of germanium, impurities wer d n-type an deduce - ep dopantss a d , however ,gole sucth ds ha bonde d diode [99] continued. respectively, in silicon and the critical role of metallurgy Nevertheless, small-area silicon diodes replaced was reviewed by Jack Scaff [117,118]. The n- and germanium point-contact diode abouy b s t 1960e Th . p-type impurity dopants such as phosphorus and boron, silicon devices, first reported in 1952 by Gerald Pearson respectively, were show Greiner'y nb s x-ray studief so in conjunction with Sawye Philid utilizey an r pFo e dth the variatioe latticth f eo n constant with dopant alloying technique (see below) to fabricate silicon diode concentratio occupno t y substitutiona lgroup e siteth n si - rectifiers via an aluminum doped (p-type) wire alloyed IVa semiconductors such as germanium and silicon, as to an n-type Si material that could operate up to 300°C reporte n d i [71]inferenc e Th . thas l groual twa e I pII [130,131]. It was clear, however, that the future was and grou V dopantp s behave n thii d s mannen i r wit silicoe hth n bipolar junction transistor [129], which germanium and silicon [117]. The ground states are itself became eclipsed wit e advene silicohth th f o nt typically 40-50 meV from the appropriate band edge for MOSFETinthel970's. silicon [6,119,120] (donors belo conduction-bane wth d A commercially feasible grown-junction silicon edge, acceptors above the valence-band edge) and are transistor, introduced by Teal in 1954 [132], was readily ionized at 300K where the number of free subsequently described by Willis Adcock, in carriers is essentially equal to the dopant density as conjunction with Mort Jones and colleagues [133], determined by neutron activation analysis (NAA) [121]. although an experimental silicon transistor was The deduction of the role of group III and V previously announced in 1950 by BTL [134]. The impurities as p- and n-type dopants, respectively, in silicon transistor raised the power output and doubled germanium conjunction i , n with equation (2)d an , ) (1 s the maximum operating temperature previously attained led to the first grown junction n-p-n transistors, based on by germanium transistors. These results clearly the "double-" technique in 1951 [64]. Since it demonstrated that silico s superiowa n r transistofo r r was easier at the time to make good contact to a p-type performance compare o t germaniud d vastlan m y n-p-n basa n eni transistor rather n-typn thaa o nt e base expanded the types of applications for which transistors in a p-n-p transistor, the former became commercially coul usee db d [6]. Morris Tannenbau co-workerd man s available, subsequently followed by the p-n-p transistor subsequently reported that addition f galliuo s d man usin a gmor e complicated process [122]. Pelletf o s dopants supported the growth of single gallium and antimony alloys of germanium were crystal silicof so n containin fivo t ep gu n-p- n regionf so sequentially (and rapidly) added to the melt during the grown junction silicon transistors [135]. Junction growt n-typn a f ho e germanium crystal [64]. Onle yon transistors wer base t fro th slicee cu em th d layesan r slic f n-p-eo n germanium junction transistors, however, was located and contacted, which was not a trivial coul e fabricateb d y thib ds technique, whics wa h process. subsequently supercede y Roberb d t Hall's "rate- growth" technique, introduce 195n di 2 [123-125]. This Alloy Bipolar Transistors techniqu variatioe bases ei th incorporatio e n dth o f no n of accepto r donoo r r impurities inte solidifyinth o g Concurrently, John Saby developed the alloy germanium semiconductor wit crystae hth l growth rate. transisto d n 195i r. an TrevoJ 2 w d [136di La r s a ] serieA germaniuf so m regions containing numeroun sp- colleagues [137], building on Hall's research [138,139]. junctions (obtaine sliciny db crystale gth ) were grown The alloy process has been described, in retrospect, as withi same nth e crystal [123 n-p-d ]an n transistors with crystal dissolution and regrowth or local liquid phase good yield d performancan s t intermediata e e radio epitaxy (LPE) on both surfaces of silicon or germanium frequencies were also achieved [124,125]. Further [22]r exampleFo . , array f o indius m dots were extension of the utilization of two different impurities in positione oppositn do e surface n-typn a f o se slicf o e various structures withi e samnth e germanium crystal germanium cut from a CZ grown germanium crystal were carried out by Hall [123-125] as well as by (see Figure 2 for a schematic illustration of a p-n-p Bridger Kold san b [126,127]. transistor) [4]. Alloying was accomplished, on Wite subsequenth h t developmene th f o t individua linern a die n ti , atmospher f aboueo t 600°C, microwatt junction transistor in germanium [66], the benefits of larger current handling capability and less during which the recrystallized germanium incorporated noise in the junction, compared to the point-contact, indiume somth f eo , thereby convertin p-typeo gt e Th . e escalatioth e formero th t f transistod o n,le ] [5 r achievemen a precis f d o narrot an e w base width, however, by controlling the alloying temperature was interfac exposo t e e {111th e } surface orientation difficult and the achievement of very high cut-off during alloying [18]. This resulted in the decision frequency performance was also quite difficult [22]. utilizo t e that surface plan r devicefo e fabrication

n-lyps A Lead

Emitter Lead

p-n-p Alloyed Germanium Junction Transistor

Figure 2. Schematic illustration of the alloy transistor [4], Reproduced by permission of the IEEE, Inc.

For example, the maximum cut-off frequency of to enhance the ability to form a uniform alloy; this an alloy germanium p-n-p transisto typicallwas r y choice of surface orientation was to continue 10 MHz [140] while a germanium point-contact throughout the bipolar junction and bipolar 1C era. transistor exhibited typical values up to 50 MHz The onset of the MOS era in 1968, however, [141]. The fabrication problems were exacerbated quickly shifted e {100th focu o t }s orientation, inasmuch as the emitter-base junction and the which exhibited a reduced concentration of collector-base junction were fabricated from opposite interfac ee Si-SiOth state t a s2 interfacd an e e surfacen-typ th d p-typ f an o e s e silicor o n facilitated better control of the MOS threshold germanium slice for p-n-p and n-p-n transistor voltage [24,143-145] as well as was advantageous fabrication, respectively [22]. The viability of for III-V devices (i.e., lasers). achievin gcontrolleda base widtsilicoa r hfo n n-p-n There is a fundamental difference in the emitter- alloy transisto s quite wa th r e o t difficul e du t basbase-collectod ean r junctions betwee e allonth y variabilit procese th f yo s [22]. Nevertheless, alloying and grown-junction transistors. The alloy junctions (becaus s presumeit f o e d simplicity s readilwa ) y e "stepe abrupth ar f "(o t type) whil e grownth e - implemente a manufacturabl s a d e procesd an s junction gradede sar . Accordingly alloe th , y transistor became widely utilized (compared to the more exhibite a highed r alpha cut-off frequency range uncontrollable base widths for grown-j unction (5-10 MHz) than the grown-junction transistor devices r transistorfo ) germaniun i s m [137,142d an ] (1-10 MHz) due to the emitter-base step junction, silicon for many years [4], although the silicon alloy although the abrupt base-collector junction for the transistor was very difficult to fabricate and never alloy transistor resulte highea n di r capacitancr epe commande significanda t market position [43]. unit area, tending to limit the high-frequency The challenge of upgrading to the GHz range response. An alternate method of transistor was a goal, required to support the extensive range fabrication , a surfacreferre s a eo t d barrier alloy of anticipated electronic applications. Sinces a , transistor [146], was able to achieve cut-off noted earliere s timeasie th o makwa t et a t ri , e frequencies up to 50 MHz by utilizing an good contact to a p-type base in an n-p-n transistor electrochemical fabrication technique [147]e Th . rather than to an n-type base in a p-n-p transistor approach was pioneered by Philco, using a jet etching [122], the latter did not become prominent in the technique whicn i , e germaniuhth ms etchei n a y db marketplace. Finally t shouli , notee db d that there electronically controlled jet of electrolyte [146]. was a strong preference of the melt-crystal Subsequent alloy contact eacn so hthinnee sidth f eo d base material resulted in a higher cut-off frequency, and cut-off frequency wer e0 MHz0.950 d ,8an factoe duth smallen o ete rt r base width, compareo dt respectively [154] y 1959B . , germanium mesa the grown-junction transistor. The mechanical transistors were being fabricated with base widths w productiolo fragilit d an yn yield, however, of 0.2 |im and, in the early sixties, silicon with cut- preclude e surfacth d e barrier transistor from off frequencies approaching 100 z wer0MH e becoming more widely disseminated. Concurrently, fabricate n i double-diffused d planar epitaxial althoug e cut-ofhth f frequenc e point-contacth f o y t structures (see below) [155]. Tannenbaum and germanium transistor had approached 50 MHz [147], Thomas fabricated a diffused base and emitter n-p-n grown-junctioe th s iwa t allod nan y junction devices, mes i transistoS a r wit a bash e widt2 Jim f n i o .h which continue e produceb a masso t dn o d- 1956, with a current amplification factor and cut-off production basis e intlatth oe 1960's [43,147]. frequenc MHz0 12 0.9 f yd o , respectivel7an y [156]. Likewise, the micro-alloy diffused transistor [148] Tannenbau d Thomas'man s f especiaworo s i k l received some attention soos wa n t eclipsebu ,e th y db importance in that it utilized the simultaneous introductio e diffuseth f o nd mesd planaan a r diffusion of both the acceptor (for the base) and transistor due to their lower leakage current and donor (for the emitter) dopants, albeit their greater mechanical stability as described below. concentrations were appropriately differeno t t ensure the desired transistor action. Friedolf Smits Diffused Bipolar Transistors reviewe e spectruth d f o msolid-stat e diffusion techniques now available [157]. The double-doping and rate-growth techniques were critica provinjunctioo e t l th t gou n transistor Mesa and Planar Processes theory in practice. There were, however, inherent limitation theiin s r manufactur regardas e s their mese Th a proces describes swa Aschney db t e r commercial applicability. The junctions, for al. in 1959 [158], who noted it was essential that the example, were physically inside the crystal and the emitter diffusion proceed more rapidly tha base nth e p-type base layer was thicker and less uniform than diffusion to retain control of the base width while the desired e introductioTh . f solid-stato n e diffusion oxide maskin d photolithographian g c technique, procedures, with a key patent issued to Scaff and pioneered by Frosch and Derick [9,10] to be Henry Theuerer in 1951 (filed in 1947) [149] and discussed below s alswa o, utilize o removt d a e implemented by Pearson and Calvin Fuller e [150semiconductoth f ]o p to e rportio th slice f o n , rectified this e situatioin-diffusioth a vi nf o n resulting in the characteristic mesa structure. This impurities in a controlled ambient over the whole geometrical modificatio s essentiawa n n ordei l o t r semiconductore slicth f eo technique Th . e involved junction reducp- o e reduct th ens e a areth e o s a the exposure of the semiconductor slice to a vapor, depletion layer capacitanc achievd an e desiree eth d containing the dopant of sufficient concentration, in high frequency characteristic. Utilization of the a carrier gas to ensure the controlled dopant diffusion (mesa) process, furthermore, openethe d concentratio semiconductoe th t na ra surfact a d an e pathway for the fabrication of devices slightly below sufficiently high temperature to create diffusion planae th r surfacsemiconductoe th f o e r wafer, with rates that would provide precise e controth f o l far-reaching implications. These included narrow, dopant penetration depth in the semiconductor. well-controlled base widths, about a factor ten Diffused layers from a few tenths of a jum to 20 Jim smalle e grown-junctiorth thar fo n d alloan n y were achieved. The initial study of the diffusion of transistors [159], and, thereby, higher frequency donor acceptord an s germaniun i s s publishemwa d operation. Many transistors could be made at one by Fuller [151] followe y b Fulled d an r e durintim G eacn e "batchr o eo gth h i slicS "f o e Ditzenberger's researc f o diffusioh n i silicon n processing with rather similar characteristics, [152] e silicoTh . n diffused junction rectifies wa r especially importan r adjacenfo t t devices with describe y Princb d n 195i e 6 with peak reverse matched device characteristic r high-performancsfo e voltages of 400 V and current ratings of 400 mA circuit applications, and many slices were available [153]. By the mid 1950s, improvements in from each CZ grown crystal. Mesa transistors semiconductor processing facilitate fabricatioe dth n fabricated by the oxide masking and of both n-p-n and p-n-p transistors structures by photolithographic technique were less expensivo t e solid-state diffusion processe a mes n i sa structure fabricate compare grown-junctioo dt n transistorr o s (see Figure 3) [4]. Lee fabricated a p-n-p alloy transistors, although all these fabrication germanium mesa transisto n 195i r 4 wit a bash e methods, to some extent, were prone to excessive width of 1.0 |im by a diffused base and leakage currents. The initiation of the concept of the alloyed Al emitter; the current amplification factor "learning curve," based on the reduction in the cost of producing numerous identical devices througe th h suggestin gcut-ofa f frequenc higs ya 300 s ha 0 MHz. cumulative process experience s enunciatewa , y b d The state-of-the-art cut-off frequency in 1954 of Patrick Haggerty with implication presene th y o st da t s realizea p-n-i-wa r z fo dp MH germaniu5 9 m [160-162]. transistor proposed by Early [163]. Ross noted that With the advent of the diffusion process, the Early "ha e distinctiodth f beinno onle gth y person device frequency limitatio transferreds nwa froe mth other than Shockley to propose a basically new collectoe basth o et r region [4] t appeareI . d that there transistor structure" [4], although one may regard the was a basic, built-in design conflict inasmuch as the innovation more of a design modification. The diffusion process resulte collectoe th n di r havine gth fabrication of such a structure, however, required the highest resistivity, compare emittee th based o dt an r . diffusio r alloo n y proces e incorporateb o t s d from Ross noted that "this led to significant series both surfaces of the semiconductor slice, thereby resistance in the collector, and that, combined with resurrectin e experimentath g l proble f accuratmo e capacitance th collectoe th f eo r junction, limitee dth control of the base width. Nevertheless, Early frequency response" [4]. While the collector significantly improve understandine dth statie th f cgo resistivity could be reduced, this resulted in a higher characteristics of conventional bipolar transistors by capacitanc lowed ean r breakdown voltag basee th t -ea also utilizing a heavily doped, thin base such that the collector junctioe nb inasmuco t e basd th ha es a h space-charge widening in the collector (due to the highly dope o minimizt d e basth e e resistancd an e revers ecollector-base biath t sa e junction) enhanced small width to reduce the transit time. Jim Early had the bipolar transistor's transit time [164,165]. This previously (before the introduction of the diffused e effecth f causino alsd t ha o a ge portio th f o n transistor) proposed a device design solution by the depletion regio spreao nt d ont base oth e th sid f eo

Base Contact emitter Contact

Emltter~Ba*0 Junction

8a«0-CoUeclor / Junction

Collector Terminal Figur . eSchemati3 c cross-sectio earln a f yno mesa transistor mad Fairchily eb d Semiconductor Corporation [122]. Reproduced by permission of the IEEE, Inc.

introductio n intrinsia f o n r vero c y high resistivity junction (even though the base was more highly layer between the base and the collector to create a p- doped e thacollector)th n , thereby reducing, n-i-p structure [163]. This allowe n increasea d d somewhat effective th , e base transpore widtth r hfo t collector doping while retainin capacitancw lo a g e of minority carrier collectore th o st . This resulten di and high breakdown voltag e collectoth f o e r since, an increase of the current gain factor (oc) and a under a reverse bias, the space charge region would decrease in the emitter potential required to ensure a penetrate the total width of the intrinsic region, given emitter current [164,165].

10 e solutioTh achievino nt p-n-i-e gth r n-p-i-po n furthermore s silicon'wa , s oxide, SiO2, whics wa h structur s obtaine fabricatioe ewa th y db lightla f o n y insoluble in water, whereas germanium's oxide was doped laye f silicoo a rheavil n o n y doped silicon water soluble [3,7]. This attribute of silicon substrate, referre epitaxias a o dt l fabrication [166-169]. facilitate s utilizatioit d e planath n i nr procesa s a s Henry Theurerer and colleagues [169] expanded the diffusion mask for p-n junction fabrication as applicability of epitaxial structures by implementing developed by Frosch and Derrick [9,10], the Bernard Murphy's localized, high-concentration sub- fabrication of the planar silicon transistor by Hoerni collector diffusion in a lightly doped silicon substrate [12-15], of the silicon surface and p-n [170-172], before epitaxial deposition, which enhanced junctions intersectin e surfacth g y Atallb ed an a bipolar performanc reducey b e d collector resistance. colleague sdielectria [11d an ] c laye r supportinfo r g The development of high-frequency, small signal metallic conductor overlayer a e [16]1Cer th .n i s devicenewle th a y vi sdevelope d planar process (see Indeed, wite e planaadventh th h f o r t process, the Planar Process section below) was, however, increased breakdown voltage behavior along with limite reverso dt e junction breakdown voltage onlf so y reduced reverse leakage current was achieved [175]. hundrew fe a d volts and, sinc junctioe eth n ares awa All the elements were now available (oxidation, small limiteo t , d power handling capabilitiesa s A . diffusion, , aluminum metallization result, the development of high-power, high-voltage and thermocompression bonding [4,7,20-22e th r fo ]

Tabl . Technologe1 y Evolutio Controller nFo d Base Width Transistors Technology Author Approximate Year Reference

Double-doping Teal 1951 64 Rate-grown Hall 1952 123-125 Alloy Hall 1950 138,139 Electrochemical thinned Tiley and Williams 1953 146 base Diffused base Pearso Fulled nan r 1954 150 P-N-I-P (N-P-I-N) Early 1954 163 Planar process Hoerni 1960 12 Epitaxy Teal, Sangster, Mark, 1954, 1957-1960 166-169 Theuerer

rectifiers and transistors with reverse junction fabricatio f junctioo n n transistor e silicoth d nan s breakdown voltage f severaso l thousand volt s welsa l controlled rectifier (SCR) (alse o th referre s a o t d as thyristors continued to proceed via utilization of the four-layer switc r thrysistor)ho Moll'n i , s laboratory mesa process, although there was, naturally, strong [18,21], in conjunction with Holonyak [22]. The interaction between o thescomplementartw e y SCR, develope y Molb dn i conjunctiol n with approaches (i.e., mesa and planar) so that the Holonyak and colleagues, has a rich history s shara separatio t s migha pno e indicateds b t wa n . [19,20,22,163,174]. Table 1 broadly summarizes the evolving state f devicTh eo ereache w physicno d da sha technological trends to control the base width for sufficiently sophisticated level that Holonyak relates junction transistors. that "Moll asked Holonyak, in conjunction with As noted earlier, silicon rapidly replaced LeLacheur's familiarity with the systems germaniu transistor mfo r fabrication [6-8,16,43a s a ] requirements on switching transistors, to prepare a resul f silicon'o t s larger energ whicp yga h facilitated preliminary design "theoryr ou f o permit "o l al t higher-temperature device operation and lower colleague becomo t s e familiar with what coule db reverse current, effective four-layer n-p-n- r p-n-ppo - silicow ne expecte e n th transistors f o d " [22,176]. n switching devices [19,173,174e th s wela ]s a l Inasmuch as this internal BTL memorandum [176] plentiful availabilit f o singly e crystals wite th h was not subsequently published, an excerpt is noted requisite purity, perfection and controlled electrical below [22]. properties [6-8,86,119]. Germaniu relegates mwa s da a niche material for specialty devices, such as low- e "Somdesigth f o ne variable f diffuseo s d power, requiring performance metric t readilno s y impurity transistor e discussedar s . Design achievable with silicon f especiaO . l importance, compromises between series collector

11 resistanc collectod ean r capacit foune yar o t d commercial silicon controlled rectifier, today's necessarye b relative .Th e advantage lineaf so r thyristor. This later wor alss kowa r baseou n do and circular structures are considered both for 1956 research [19]. base resistanc r collectofo d an er capacity. Parameters, which are expected to affect the In the process of diffusing the p-type substrate frequency behavior considerede ar , , including wafer into an n-p-n configuration for the first emitter depletion layer capacity, collector stage of p-n-p-n construction, particularly in the depletion layer capacit diffusiod yan n transit redistribution "drive-in" e donophasth f ro e time. Finall parametere yth s which mighe b t diffusion at higher temperature in a dry gas

obtainabl comparee ear d with those needer dfo ambient (typically > 1100°C in H2), Frosch a few typical switching applications." would seriously damag r waferseou wafee Th . r surface woul e erodedb pittedd an d r eveo , n The Planar Process totally destroyed. Every time this happenee dth s e apparenlosexpressiowa th s y b tn o n The development of oxide masking by Frosch Frosch' smentiono t face t no , ourn o , s (N.H.). and Derick [9,10] on silicon deserves special We would make some adjustments, get more attention inasmuch as they anticipated planar, oxide- silicon wafers ready, and try again. protected device processing. Silicon is the key ingredien oxids MOSFEr it fo d ey an tpave wa Te dth In the early Spring of 1955, Frosch commented integrated electronics [22]. An account of their to Holonyak t again,i ,d "Weldi "e meaninw l g revolutionary developmen utilizatiod an t f SiOno s 2a e waferth s were again destroyed t thee Bu .nh vitae th l foundatio f today'o n s 1C industr s beeyha n smile displayed dan silicoe dth n wafer - snic e described by Holonyak [22]: and gree n coloi n n furthe(i r r instances also s techniciahi d an pink) ne H Deric. d ha k buildinn "I r variougou s experimental devices, switched from a dry-gas (typically N2 or H2) we were in contact with various groups and impurity diffusio wet-ambiena o nt t (H2O vapor individuals, but above all with Carl Frosch. carrie+ r gas) diffusion a consequenc, n a f o e Frosc consummata s hwa e process chemiso twh accident of the exhaust H2 igniting and flashing- was familiar with many types of processing back int diffusioe oth n chamber (becauss ga f eo procedure d beeha n d workingan s , wits hi h flow fluctuations) and causing H2O to cover, technician Derick n impurito , y diffusion into react with protecd an , silicoe th t n samples with silicon for several years. In spite of his oxide. The "wet" ambient, which was then considerable experience, Frosch, with dry gas immediately evaluated and adopted, created a diffusion procedures utilizing N2 or H2, protective oxide on silicon. It could be regularly reduced many of our silicon wafers to selectively remove r gaseoudfo s diffusion into "cinders, " particularly at higher temperatures the bare regions, which could the resealee nb d with oxide for higher temperature anneals or further diffusion. Many processing sequences Because we had mastered building a diff used- protective th coul f o devisee e db eus r oxidedfo , base alloyed-emitter silicon p-n-p transistor (in which f courseo , , prevented crystal pittind gan r problemspitou f o e s with diffusion)f o e on , erosion. Frosch and Derick quickly found out the p-n-p-n configurations that we could which impurities were blocked from diffusion explore was simply a modification of the p-n-p into silicon by the natural protective oxide (SiO ) transistor coule :W d fabricat diffused-base eth e 2 created in an H2O-vapor ambient and which alloyed-emitter p-n-p on one side of a p-type impurities would permeate the oxide (e.g., Ga). substrate wafer after it first was prepared with It was easy, once the issue of the oxide was an n-type diffused region (symmetrical) on known deviso t , e various scheme diffuso st e into both wafere sideth f o s. Either side coule db bloco t r o k impurity diffusion into silicone Th . chosen to form the p-n-p. The result was a p-n- process was so flexible that planar n-type p-n switch, in fact, the p-n-p-n switch of regions of any desired pattern could be prepared s exampla describe ) (b en i [19]d . (The o p-typna e substrate silicon oppositee th r o , - p , complementary versio f thino s exact structure, on-n diffused regions could be prepared on n- an n-p-n-p with Ga diffused into both sides of type silicon. All other diffusion procedures were an n-type silicon wafe d the an a rAu-Sn b suddenly rendered obsolete. We readily emitter alloyed on one side, was later converted the Frosch-diffused silicon n-p-n into introduce t a Generad l Electrie firsth ts a c a working p-n-p-n switch [19]."

12