Get 20M+ Full-Text Papers For Less Than $1.50/day. Subscribe now for You or Your Team.

Learn More →

Architectural Innovation: The Reconfiguration of Existing Product Technologies and the Failure of Established Firms

Architectural Innovation: The Reconfiguration of Existing Product Technologies and the Failure of... C9 t. <^^ rt PAPER WORKING SCHOOL OF MANAGEMENT ALFRED P. SLOAN INNOVATION: "GENERATIONAL" SYSTEMS THE RECONFIGURATION OF EXISTING FAILURE OF ESTABLISHED FIRMS AND THE Clark Rebecca M. Henderson and Kim B. May, 1989 WP 3027-89-BPS MASSACHUSETTS INSTITUTE OF TECHNOLOGY 50 MEMORIAL DRIVE CAMBRIDGE. MASSACHUSET" "GENERATIONAL" INNOVATION: THE RECONFIGURATION OF EXISTING SYSTEMS AND THE FAILURE OF ESTABLISHED FIRMS Rebecca M. Henderson and Kim B. Clark WP 3027-89-BPS May, 1989 Rebecca M. Henderson Assistant Professor of Management Massachusetts Institute of Technology Kim B. Clark Professor of Management Harvard University Research, Harvard Business This research was supported by the Division of also like to thank School. Their support is gratefully acknowledged. We would Dataquest and VLSI Research Inc. for generous permission to use their published those data, the staffs at Canon, GCA, Nikon, Perkin Elmer and Ultratech, and all photolithographic alignment technology who gave so individuals involved with generously of their time. 4 1989 I JUL 2 "GFNFRATTnNAl " INNnVATTON: THF RFCnNFTGKRATTON OF FXTSTTNG SYSTFMS AND FATf.I'RF OF FSTABLISHEO FIRMS THF t^ithpr "inrrpnipntal or ThP tiaditinn-Tl catPgori ?:ation nf innovation as " "Generational" innovation "radical" is incomplete and fundanient al ly inisl eadi ng . its elements - i^; that reconfigures a technical system without changing innovation often both incremental and radical innovation and ha< qualitatively different from consequences. important and unexpected organisational and competitive the concept's paper defines generational innovation and illustrates This technical and competitive explanatory force through an empirical study of the the semiconductor photolithographic alignment equipment industry. history of . . TNTBODUCTTON "Wp rippd to know morp abnut differpncps bptwppn rnmpptenrp rtpptrnying advances.. Caboutt what distins'nshps and competence-pnhancjng tpohnologif^al advances. CTushrnan and bPtwpen incrptripntal improvpments and dramatir " Andprson (]^Rf^)^ to and transform organi i^at inns and thf> Tbp power nf npw technology challenge Schnropeter structure of indiistrips has bepn an irfportant thpme nf research si ncp t'TpwkPS, innovation creates great difficulties for e<t.=ih1 i •ihed fivTi.s, "Radical" 19f)9; Cooper and Schendel, 197(^; Rothwpl 1 n^f.; Tush'oan Sawers and St i 1 ] erman , . successful entry and pvpu and Andprson, I^Sf-) and is oftpn the basis for reinforces the rpdpfinition nf thp industry, while "incrempntal" innovation oftPn and incremental innovations havp different dominance of establishpd firms. Radical competitive consequences because they require quite different organizational innovation feeds on and rpinforces the existing problem capabili tips. Incremental while radical innovation forces solving capabilities of established organizations, approaches them to ask a new set of qupstions and to employ new problem solving 196A; Arrow, 1974; Hagp, 1 9R0 ; Dess and Beard, I'^Hh: Fttlie, ^Burns and Stalker, 19c;A'i. Bridges and O'Keefe, 198A; Tushman and Anderson, between radical and incremental innovation has produced The distinction is growing evidpnce important insights, but it is fundamentally incompletp. Thprp "lodest that are numerous technical innovations that involve apparently there to pxisting tpchnology, but that have quite dramatic compptitive changes the 1987). Take, for examplp, thp case of Xerox and small consequences ^ Clark, copiers confrontpd in the mid-1970's Xerox, the pioneer of plain pappr copiprs, was than copiers that were much smaller and more reliable with competitors offering or the traditional product. Thp upw products required no new scientific . rnmp;^iny invpntpd thp cnre enginpprins knowlpdgp. but rlpspitp thp f-^ct th^t thp had (if»l>- vpjnv md hifl piior'Tions pxporipiice in the indiistyy, h i almost t pchnol n^i''-^ that ti^ip yprov pight vp^rs to hring -^ cnmppti t ivp product into thp "laykpt. Tn of it^ markpt -^harp and ctifferpd sprions financHnl prohlp-is iTlirk, lost half ]'^S7'i, modpip that rply on dist i n-^tion?^ bptwppn radical and i ncrpppnt al Fxisting rpasons thesp t ypps of innovation innovation providp littlp insight into thp rnndpl should havp such cnnspqupncps . Tn this pappr wp dpvplop and apply such a parts, Wp prpspnt a fra'-npwork that allows ns to Thp pappr has four first bPtwppn changps in analyzp thp naturp of tPchnical changp and to dPVPlop thp links that thpy prpsent tn an tpchnology and changes in Ihp inforaation procpssing task this organization ^section 2). We dpfinp incrPFiPntal and radical changp within ." framework and charactprizp an important class of innovations as "generat ional We is qualitativply diffprpnt from radical and argue that generational innovation changes the performancp of an incremental innovation. While incremental innovation the elpments of that existing set of components, and radical innovation changes the relationships between them. While radical set, generational innovation changes wholly different system, innovation challenges the established organization with a gpnprational innovation is to learn about a differently the challengp of gained through experience configured version of an pstablished system. Knowledge tpchnologies often rpmains with incremental innovation about particular component betwppn relevant, but "relational knowledge," or knowledge aboiit the interactions performance and user needs is rpndprpd obsolelp. componpnts, systpm we summarize competitive and technical history of In section 3 briefly the a useful the lithographic alignment pquipnipnt industry. The industry offers and to examine its power in explaining context in which to illustrate the model generational innovation. the competitive and organizational consequences of capital equipment used in Photolithographic aligners are sophisticated pieces of improvpd of integrated circuits. Aligner performance has the manufacture hut thp cnrn tpchnnlnpips hn/p flramat irall V over thp last twpnty-fivp ypais. i itirlnstry 'narsinally sinrp thp tpohniqnp wis first invpiitprl. Ypt hp changpd only turbiilpncp. markpt 1 padprshi havp hppn has bppn rharactPii ^pfl by gvpat Chansps in p md incnnhpnts reqiipnt pntry has nrcnrrpd thrnushi>ut thp industry's histniy, f , i ndnrt of dpclinps in markpt sharp fnl lowing thp i nt inn a havp nftpn suffpipd sharp npw ppnpratinn of pquipmptit. thpsp pvpnts arp pxplainpd hy thp intrusion of gpHPrational Wp hplipvp that gpnprationa] innovation rpquirpd littlp npw innovation into thp industry. Whllp rplational VTiowlndgp. scipntific knowlpdgp, it rpquirpd signifirant •shifts in taking thp industry had grpat difficulty rpcogni^ing and FstablishPd firms in a rpsult thptr dpvplopnipnl efforts wptp advantagp of thpsp shifts, and as significantly Ipss pffpctivp than thnsp of pntrants. the with a brief summary and with a discussion of Wp close the pappr organisational rpsponsps to implications of this resparch for undprstanding tpchnological changp. TT Conceptual Frapipwork is the existing litpraturp on tpchnical innovation Thp cpntral notion in the existing design, and introducing a distinction between refining and improving an CMansfield, departs a significant way from past practice. new concept that in innovation introduces iqf>8; Moch and Morsp, 1^77; Frepman, 1982;) "Tncreipental" thp potential of the minor changes to thp existing product, exploiting relatively Dutton, 198^.; design ^Fttlie, Bridges and n'Kpefp, nSA ; Dewar and established that l^Sf^). and Winter, for examplp, have argupd Tushman and Anderson, Nelson trajectories" during incremental innovation occurs along a series of "natural 1*^82). and Winter, "locally" for appropriate solutions ("Nelson which firms search nq82i has ^Kuhn, 19701, Dosi Building on Kuhn's work in the history of science, nnr^irii g'"" snggpstPfl that Inrrpmpntal innn\Mtinn ncrnr's within ?\ "tprhnologif-^ 1 , or thp pptablishpd "..outlnnk, spt of iirnrpdiirps , ripfinitinn of rplpvant prnblprn'^ a,-) to snlntinns," and nf thp spprifir knnwlpdgp rplatpd thpir "Radical" innnvatinri, in contrast, is haspd upon a diffpimit set of principles and oftpn oppns \ip wholp ripw inarkpts and enginppring and scientific fpttlip, I'^'^A; Dewar and Dntton, I'IRA; potpntial applications Rridgps and O'Kepfp, ronroent- Tushnian and Andprson, n8*^>.Thi)s the cotTrnprci al and tpchnical pnvi associated with radical innovation is far n^ore dynamic and imcprtain than that (^Dpss and Beard, T^iRA). associated with incrpinental innovation bnth the snpplv Innovation's impact on compptition depends on its effpcts on the supply side, thp two pattprns of innovation and demand sides of the market. On are associated with very different organizational capabilities. Radical and require not only a difference in knowledge base, but also incremental innovation general it appears that more formal, in organizational structure and process. Tn communication, "hierarchical" organizations that rely on structured patterns of routines and procedures are best equipped to undertake incrempntal problem solving organizations in which the innovation, while more "organic," "entrepreneurial" flow relatively unstructured and which are more responsive to information is innovation (Burns and dramatic change are best equipped to exploit radical 1981; Stalker, 1966; Moch and Morse, 1977; Hage , 1980; Kimberly and F.vanisko, 1986^ Ettlie. Bridges and O'Keefe, 1984; Dewar and Button, Ettlie, 1983; betwppn innovation and organizational capability that It is this connection its ability to transform thp compptitivp environment. gives radical innovation to to adjust (Nelson Organizational capabilities are difficult create, and costly and Freeman, 198A>. An established firm faced with and Winter, 1982; Hannan must makp them vety radical innovation must make very diffprent decisions and innovation differently. The number and magnitude nf the changes that radical stock of knowledge and in the way in which it requires in the firm's existing hp rlfiimting. Thus sevpral stniiips havp "^hnwn that procpssps inforriatinn may radical innovations tpnri In nrivinatf^ outside of estaM i -ht^d finns, and that often have great diffirnlty in making the shift to a new established firms ]'^<^'^] Cooppr and Sch^nd^l 1*^7A; "trajectory" Cjewkes, .Sawers and St i 1 ] erman , , nSA^. Rothwell, 19Sf.; Tushman and Anderson, Differpnces in organizational capability help to explain changps in the also shaped by an innovation's "suppliers" of innovation. But competition is effects on demand. The cnmulative economic effects of incremental innovation arp often broadens a product's appeal often significant, and incremental innovation and expands its market. Radical innovation, in contrast, offers customprs a wholly new set of possibilities, or meets established needs in a wholly different way. concept may bear some relationship to the pstablished design, thp Although the new the new needs it meets are change in price/performance is usually so dramiatlc, and sut)stitntp for thp so different, that the old product becomes at best a very poor innovation is defined as new design. Indeed, in the economics literature, radical Innovation that creates such changes in performance or cost that the old product new competitive prices (^Gilbprt and cannot act as a substitute for the even at Newbery, 1982; Relnganaum, 1983>. suggests, research on Innovation, organization and As this brief summary years has added important insights into competition over the last several characteristics and incremental and radical innovation and the organizational associated with them. However there has been strikingly competitive consequences innovation. little discussion about intermediate classes of break open the "black We believe that this neglect flows from a reluctance to box" that specific technology in order to understand its competitive and is a students of technology and economic organizational implications. Although some n97Ajq82t; Abernathy, n^TRI; history (See for example, work by Rosenberg, (198f.U have described Abernathy and Clark, 09851; Clark, n98S~i; and Sahal , . ' structure to some extent, few studies have technologies and examined their inner of the tt^chnoloey affect the developed an understandine of how thp characteristics process. There is an intuitive sense in the literature, and some innovation different technologies may be empiiical evidence, that innovations in very TMoch different in character because of the underlying structure of the technology Sahal 198A; riark, 1987^ But and Morse, 1977; Kttlie, Bridges and O'Keefe, 1984; , concepts like we have few ways to talk about these differences beyond economic minimum efficient scale, returns to scale and appropr i abi 1 i y framework that allows one to define and analy?e intermediate Developing a to understand an important das*; of classes of innovation may help us understanding of innovation, technological changes, and may also deepen our compptition in general. organization and Framework The f" . we present in this section is summarized in Figure H The framework terms of product technology, In order to focus the discussion, we develop it in starting point the concepts are genpral; they apply to processes as well. Our but - - product the product as a whole the "system" and the is the distinction between - established distinction. in its parts the "components." (This is an old and well the work Marples n9Al) and Alexander (19fi4K^ We conceive of See for example, by a roopi fan. Tts major product as a set of components, (\K Take, for example, guard, a simple components include the blade, the motor that drives it, the blade intuitive mechanical housing. For the moment we will use the control system and a the (1984) have proposed a framework that is related to 1 Saviotti and Metcalfe sets of product one we propose here. They distinguish between three that describe characteristics: those that describe its technical features, those of its the services that it performs and those that describe the method production. "> Figure ( ] : A Basic Framewnrk SYSTEM USER NFFDS rOMPONFNTS PAR^MFTFRS Xi Si^Y- 7i Yi fi^x-) Z2 = S2^Y*> x2i...x^. f.fx*^ Ya Z„ = S«(Y-) . . .X" X^l Y„. f,Jx*) : . and physically distinct notion that each rnniponpnt fills a distinct function is rnmponpnts. from thp othp> parai-ppters" Cv>, which F.ach componpnt is charactpri/pd by a spt of "compnnpnt attribute of thp componpnt, For the fan bladp, for pxamplp, describe some physical wpight of thp bladp and its dinpn'^ions . A the component parameters include the exhaustive description of complpte set of component parameters would constitute an each component characterized by a set of parameters (Y) At the system level, the product is example physical properties of the system as a whole. Tn our fan that describe the weight, the voUimp of air moved the system properties include things like total system property like the per minute, and resistance to impact. Any individual betwppn spvpral moved per minute is determined by the interaction volume of air function the amount of air a fan puts out is a component parameters. For example, the powpr of the motor, and the pfficiency of the size and shape of the bladp, the model by showing with which the motor drives the blade. We represent this in subset of the component parameters. each system parameter as a function of some Thus , we have argument of fj^, x* is an index of the M system parameters, and the , where m is fO is a statement of the component parameters. The function a subset of the system the relationships between the componpnt and physical laws governing from mathematical formulae dprived parameters. Sometimes it can be approximated by empirically. of the physics involvpd or pstimated an understanding parameters. For parameter can affect spvpral system A single component that not only the amouiit of air it example, the dimensions of a fan blade affect restricts the appeal and its portability. This puts out but also its esthetic to dpsign a Tt very difficult, for examplp, available set of system parameters. is to keep a that puts out enough air fan that weighs as little as an alarm clock but 10 hiindiprl people coni . Similarly, rorrponpnt parampters nften ititevart witli earb rlpt a systerri paramPtey . The pffpct of a rhangp in hlade shapp on nther in eimi ni ng power of the fan's motor. thp amount of air moved, for pxamplp, depends on the tlip dpsign of a product is a romplpx process of simultaneously dptermlning Thus componpnt and systpm parametprs. rhangps in the So far thp fraTipwork givp<; us thp basis for talking about of the technology. Rut we also need to understand the internal striicture Wp bpgin by as'^nining connection between the technology and customer needs, customers evaluate the product against some set of "needs," fZ~t, In thp case of needs includp pasp of operation, room fan the relevant criteria or might ('brand image, appearancp"> . The extent reliability, transportability and aesthetics plemput to which product meets a particular need, that is, thp valup of a given parameters of the product. A fan's of Z, dppends on some subset of the system dropped, degree of reliability, for example, depends on its resistance to bping time between failure of the entire system. We hit or twisted and the mean represent these relationships in the model as: the M the N user needs, and the argument of g„n is a subset of where n indexes of the nature of the customer and of system parameters, Y* . gnf^Y*") is a function physical relationships the way in which the product is used as well as of the uses to product's system parameters and the criteria that the customer between the the extent to which a user's need for evaluate the product. Thus, for example, relationships Ce.g, system reliable operation is met depends on a set of physical thp fan's dpsign and reliability, frequency of drops") that are partly inhprpnt in fashionable or on fan is used. Meeting the need for a partly dpppndpnt how the local intpraction betwppn the fan's dPsign, its "attractive" fan dpppuds on the be. what an "attractive" fan would environment and the customer's expectations of underlying Such expectations may be less stable than the physical relationships . rpliability. Thus, gn'^V't capturp;^ bnth thp physicil, anrl the hehavioral and thp nsp nf thp prndnct n^ps social procpssps that dpterminp thp ways in which intn iispr npcds. Tf cnstnmprs diffpr in thp way in thp cnntpyt system paramptprs in which thpy nsp thp product , thpn thp function s^JY""* may diffpr bptwppn customprs a choicp of tradpoffs hptwppn Just as thp dpsign of a product implips cnmponpnt and system parametprs, so product dPsign also embodips either implicit pxplicit tradpoffs between uspr nppds. The users' own preferences across these or "utilitv function" that captures the value tradeoffs can be reprpspnted through a to the customer of some particular set of levels of uspr needs. = (3> U„ hrz^ hCZ"*, Where o indexes the customers with different preferences. The function embody interactions between its like the functions h^fx") and gJY"^ may aesthetic appeal may not enhance the value of a arguments. For example, additional quiet operation fan if It does not meet some minimtim standard of reliability, and highly valued if thp fan Is a relatively small one. may be much more Types of Technological Change "incremental" and "radical" This framework allows us to characterize intermediate innovation and provides us with a useful framework for analyzing "dominant of innovation. In the context of the model, the emergence of a classes 1986^ equivalent to the emergence of a stable design" fAbernathy, 1978; Sahal , is needs. "Tncrerrpnt al" innovation set of components, system parameters and customer the values of one or more improves individual component performance, changing changing the values r>f some of the system component parameters, and thus sets and Z and parameters and user needs. But it leaves the elements of the X,Y g^'^Y*"', unchanged. thp rplat ionships bPtwppn them h./x*!! and 12 occurs when a new design changes the set of components, "Radical" innovation elements sets X, Y, and Z, thereby system parameters and user needs, or the of the relationships between them, or the form of the functions h„(X*) also changing the completely and g„(Y*). In its extreme form, radical innovation introduces a radical" components and a wholly new system. But in general, "more different set of affect more system innovations introduce more new components and consequently parameters and more user needs. generational innovation as innovation that changes the We define a parameters and user needs of the relationships between the components, system needs set of components, system parameters and user technology but that leaves the innovation changes the value of themselves relatively unchanged. VThile incremental innovation changes the elements of the sets X, Y, the sets X, Y, and Z, and radical generational innovation relationships that link them [h„(X*) g„(Y*], and Z and the , the elements largely unchanged. changes the values and relationships, but leaves triggered by changes in a particular component Generational innovation is sometimes the technological but its essence is a fundamental reconfiguring of technology, components, system parameters, and system around an essentially stable set of customer needs. that the model allows fan example can illustrate the distinctions The room air product concept is a large, Suppose, for example, that the established us to draw. with the motor hidden from view and electric powered fan, moimted in the ceiling, mounted on The control system is an on-off switch insulated to dampen the noise. that run inside the wall and connected to the fan through a set of wires the wall and ceiling. blade design and in innovation could involve an improvement in An incremental cooling. The introduction of a power of the motor to achieve a higher rate of the the established would be a radical innovation. Not all of central air conditioner would involve using fans would become obsolete. The new technology knowledge base new electric motors and fan design. But move air and would require knowledge of to . apsnriated with cnmprpssnr?; , re^f rigerantp anii Mipir ac'.cnriai'pfl control.=; cmmporipnts would Tcld whnlp npw tprhniral di '^cipl i pps md new intPi -relationships be used different way and might be sold Furthermore, the product would in a very to wholly different rustomer'^. The between radical and i ncrpiriental innovation is therefore distinction large mounted clear. What of g'^nprat ional innovation"!" For the maker of ceiling room fans, the introduction of a portable fan would be a generational innovation. basic components would be largely the same (f.g. blade, motor, control While the different materials) and thf system!*, the design choices would be different ("e.g. this values of the component parameters would change (e.g. smaller dimensions). Tn from ceiling to portable fan has the character of an in^rpmental sense, the change interactions innovation. Rut there would also be significant changes in the needs the product could meet. The between components and in the menu of user room would smaller size and the co-location of the motor and the blade in the focus attention on new types of interaction between the motor size, the blade fan could generate. Shrinking the size dimensions and the amount of air that the with new properties, as well of the apparatus would probably require new materials could new interactions between performance and weight. On the user side, there as reliability. It is the need to be new tradeoffs between ease of use, safety and that new of interaction and these new sets of tradeoffs explore these patterns sets apart this kind of innovation. knowledge ,-=ind the knowledge Radical innovation obsoletes much of the existing obsolescence is usually processing capabilities of established firms, but this knowledge triggered by immediately evident. The obsolescence of established firm to much harder to observe and may he more difficult generational innovation is -explore their competitive correct. We can best illustrate these differences, and undertake discussion of the knowledge required to significance, through a successful product design. u nf tn rnncpntr.itp nnr attpntinn nn the problern prndnct desipn We hflvp rhnspn reqni tbp intpgntinn nf tprhnologic il since it is one that pxiilicitly rp'^ it the markpt and of rnstorier reqni rements = Moreover knowledge with knowledge of snggp'ited rritiral tn ^nrrpssfnl a problem that several re-^earrhers have is is MSSSi, n9R7>, Freeman nq^?^K organi7ational response to innovation. fClark undprtaking prodiict thp discussion that follows we model the organization Tn individual with limited knowledge and with development as a boundedly rational This abstraction i "^ in the tradition of limited information processing abilities. implications of technical who have studied the competitive a number of researchers and Winter nq82'> and Simon n9^9), and we change, including Arrow (n74^. Nelson problem. The implications of that it provide a useful "first cut" at the believe the way in particular organization will clearly depend upon any innovation for a is managed, but we have abstracted from which knowledge inside the organization differpncps in thp infoi-mation this issue in order to focus clearly upon the types of innovation. tasks presented to the organization by different processing to dpsign a npw product. A designer Consider the information that is required components to bp able to technologies of pach of the must know enough about the CMarples, 19(M Ramstrom and Rhenman generate somp set of component parameters (x"» ; interact with each 1965~». They must understand how these component parameters system parameters, or characterized by a given set of other to produce a product these system and they must understand how something of the relationships hjx"'), of the needs, or understand some portion parameters fill a particular set of user new product successfully a Finally, in order to design a relationships gjY"). differput relative economic value of designer must be able to roughly assess the source "invention," or of thp original 2 We deliberately avoid a discussion of are likely to come from is the new idea. While the problem of where inventions of the implementation of that important and interesting, we believe that a study of at least as important, and new idea, or its translation into a saleable product is processing task that an insight into the nature of the information provides more successful. must undertake in order to be commercially organization 15 sompthing ahnttt thp nistortiprs' clijstprs of user nperls: that is thpy must knnw fnnrtinns V^(7.). utility infor-mat ion about tViP custom.pr and th^ tpchnology Tf a desisn'^r had pprfpct npw product would hp a and was faced with no significant uncprtainty , dpsigning a which trivial exercisp. Thp dpsigner would havp completp knowlpdgp of thp way in x) could bp genpratpd and of thp functions diffprpnt spts of componpnt pararnpters < and V^ = hfZt and could dpvplop an "ontiirial" product. Howpver hni^^^K gn'^Y*'* abilities with only lirnitpd pxppriencp dPsigner of limited infov-mation procpssing will bp in-^omplptp, and will is unlikply to hp in this position. Thpir knowledge the valup of bp function of thp recent history of innovation and of marginal pvolutinn information at different stagps in the product's diffprent typps of 19f.9; (Simon. Npwell and Simon, 1972K and the As a technology pvoIvps both the designer's stock of knowledge new knowledge change. The routines or procedures that they use for acquiring the types design always requires the development of new knowledge, but process of dramatically with the natiire of the of new knowledge that are required differ widespread innovation. During periods of radical innovation there is widely different experimentation in product design. Products are characterised by parameters and user needs. For any single product, sets of components, system parameters and user knowlpdgp of thp rplationships bptwppn componpnts, system - have very needs to bp tacit and incomplptp a designpr is likely to is likely since every new product little knowlpdge of the relationships h^Jx*"* and g„(Y*'>, customprs place embodies a widely different set. Knowledge about the valup? that cost effective way of on diffprpnt configurations of user needs, and of the most thought be The process of radical innovation can be meeting them will also scarce. the "space" in order to of as a process of exploring the market and technology options. The better understand both the customer nepds and thp technological arp most processing capabilities of thp dpsignpr limited information gathering and . oripiitated towards Iparning about new tPchnolosiP!^ '^nrl new rippds , and effpctivply thp rp] hptwppn thpm in a rplativply pxpprimpntal or towards pxploring at ionships about a tacit way. Tn tpnns of onr fraripwork, thp dpsignpY dpvplops knnwlpdgp (\') SPts fXI and ( Z) and thp inf orrnat inn largp nnmbpr of possiblp ir.pmbers of the , , are orientated processing rontinps that they dpvplop to search for new knowledge th^ about potential new elements of the set. Thej r knowledge of towards learning likely to be tacit and universp of possible relationships betwepn thpm is incomplete types of knowledge that incremental innovation changes the The transition to competitive advantage during periods of are most useful to the designer. While products innovation is gained by the introduction of quite different radical technologies, during periods of incorporating quite different component exploitation of a incremental innovation it is gained by the morp pffpctive stable set of set of components and user needs within the context of a limited detailed becomes both possible and valuable to dpvelop a more interactions. It between the existing components, system understanding of the relationships matures this knowledge is likely parameters and customer needs. As the technology likely "cheap" and the desigupr's time is to become widely diffused or relatively particular componpnts and detailed knowledge about the performance of to acquire teois of component/parametpr/customer nppd interactions. Tn about a limited set of detailed knowledge about particular our framework, the designer devplops vpry these elements remain stable, members of the sets ^X) , (Y) and ( Z) and, since relationships h,,,(x*'> and gJY"^. detailed knowledge about some known subset of knowledge that are tightly will strategies for acquiring new Moreover they develop technologies and on a limited set of focused on these particular cornponent critical interactions. particularly we begin to understand the implications of Given this context, can capabilities knowledge processing "generational" innovation for the knowledge and 17 ' 7."^ generational innovation the sets (X) , CYt .md reinain of a flpsigner. nnring obsolescence relatively stable, hiit the relationships between them chini^e. This of the "relational knowledge" of the designer may be just as significant as the with radical innovation. In sonp situations it may obsolescence that is associated to identify. be even more significant because it is more difficult - component The advent of radical innovation of completely different of user needs - is usually unmistakable, and technologies or quite different types knowlpdge a designer runs little risk of assuming that their historical is is more subtle, and it may be much more relevant. But generational innovation become obsolete. The spt of difficult to notice that ones relational knowledge has much of the components, system parameters and user needs remains stable, and the remains relevant. There are fewer signals to alert designer's knowledge designer miay attempt to meet the designer to the nature of the innovation, and a his or her threat of a competitive product with a design that is based upon routines that they have historically derived relational knowledge and the generation gather new knowledge that were appropriate to the previous developed to the risk of producing a significantly of the technology. As a result they run generational innovation. inferior product and "failing" in the face of that is perfoiTned A characterization of the information processing task complex problem. But organization designing a new product is a much more within an single we have described in the case of a to the extent that the dynamics that of the evolution of boundedly rational individual also characterize the dynamics this processing capabilities of an organization, the knowledge and knowledge generational innovation may have framework gives us insight into the reasons that and competitive implications. such dramatic organizational 18 TTT Fmpirical Spctinn the dlscussinn wp now turn to a dP'^rriptinn of As a means tn dppppn our pqnipmpnt indn<^try. Wp have of thp spmiconfliirtor optical photolithographic history innovation it is adpqtiatply characterise technological suggested that in order to technology tn pxainine in some detail npcessarv to break open the "black box" of a customer needs, between its components, system parameters and the relationships spction a detailed analysis of semiconductor Consequently we present in this at the component level. We \isp photolithographic alignment technology focused ^ photolithography and characterize the history of innovation in this analysis to the industry's which is a source of insight into thpn explore the extent to it snggpstlve, but they are presented competitive history. Our empirical results are The of the explanatory power of our framework. here only as an Illustration be hypotheses remains to done. rigorous formulation and testing of appropriate solid state semiconductor Photolithographic aligners are used to manufacture intricate semiconductors requires the transfer of small devices. The production of semiconductor material such as silicon, patterns to the surface of a "wafer" of Figure ^21 schematically transfer Is known as "lithography." and this process of with a surface of the wafer Is coated Illustrates the lithography process. The to be transferrpd to the chemical or "resist." The pattern that is light sensitive as is and the mask is used to block light wafer surface is drawn onto a "mask" '3 of the technical and results of a much larger study Oiir anaiysis draws on the Included the (Henderson, 1988). This study competitive history of the industry for every managerial and sales histories construction of comprehensive technical, These the industry's history since 19f.5. product development project undertaken in individuals, including drew from field interviews with over one hundred histories from intensive study leading engineers and scientists, and product designers and scientific journals. An the trade press and the major of Internal firm documents, process to the use of an iterative validation important element of the work was the research written summaries of ensure its accuracy. At each stage of the and circulated to key individuals results and the preliminary conclusions were follow up interviews. confirmed through <'21: Figiirp Thp r,i thographir Prncpss 1 . Expose Rpslst Mask Resist Wafer 19 resist sn that onlv those portions of the resist definerl hy the falls ontn the , mask are exposed to light, or "exposerl." The light rhprniraliy transforpiS the the areas availahle as a resist so that it can he stripperi away, leaving nnexposeri as basis for further processing.'* The process rnay be repeatfd as many twenty times be during the inanuf actiirer of a semiconductor devii^e. and each layer rriiist located respect to the previous layer. precisely with the mask relative to th^ A photolithographic aligner is used to position expose Figure wafer, to hold the two in place during exposure and to the resist. generic optical photolithographic n> describes the principal components of a stable aligner. The core technologies of photolithographic alignment have remained developed in the middle sixties. Despite this since the technology was first been strikingly turbulent. stability, the competitive history of the industry has sequence of dominant firms have each in turn failed to maintain their position The the Table M) shows the sales histories of the leading firms. in industry. Kulicke and Soffa in llfiS. first commercially successful aligner was introduced by and held nearly 100% of the (very smalH market for They were extremely successful replaced them and held the next nine years, but by 1974 Cobilt and Rasper had market for contact aligners each. In 197A pprkin Elmer approximately half of the the industry. Further entered the market, and immediate became the largest firm in were the leading entry followed in the late 1970s, and by 1981 GCA and Canon while of this writing GCA has also lost its dominant position and players. As player Nikon is probably the largest firm in leading Canon remains an important today. edge photolithographic equipment "positive." Tf a negative resist is used 4 Resist may be either "negative" or processing. If a positive resist is the unexposed areas are stripped away during description of used the exposed areas are stripped away. A more complete (1987). semiconductor lithography is available in Watts and Einspruch ni: Figiirp Thp Principlp rompnnpnt> nf a "Gpneric" nntjral Phntnl i thngraphir Alignpr.' SnURfF, 20 h) Figure ( : Principle elements of photnl i thographir technnlngy. CnMPONF.NT.S r.tlSTnMFR SYSTF.M NEEDS PARAMETERS MTNTNfllM SOURCE FEATURE Wavelength SIZE -Uniformity Col lirriatlon Energy RESOLUTION MECHANICS Contact accuracy -Gap uniformity CRITICAT, Gap accuracy Stage accuracy DIMENSION Stability CONTROI, OPTICS THROUGHPUT Numerical aperture -Wavelength distortion Lens stability Lens ALIGNMENT RELIABILITY Wavelength —Algorithms Accuracy Targets FOOTPRINT CONTROLS SETUP TIME Structure -Distribution MAXTMITM WAFER SIZE ?1 on wafer. nthpr things equal, si7e semicnndnrtnr a All the smaller the of a device, the faster it <~an rnn and the cheaper it is to mannf acture , and since device size heen hy historically has limited the miniirnim feature si;7e capahility of the lithographic process, users have demanded aligners with smaller and smaller mininnim feature capahility. Users are concerned ahont the size also eqnipTnent's throughput, yield, footprint, tTiaximurp wafer size capahility, reliability and f lexibi lity.5 An aligner is an extremely complex piece of eqnipnipnt, and can be those that characterized by many system parairieters . The m'ost critical are support its central function - its ability to accurately and consistently replicate an extremely small pattern on the wafer, or to support particular minirrium feature size capability. Three system parameters are particularly critical in this respect: "resolution," "critical line width control" and "overlay." Figure CS) defines these three parameters and illustrates their relationship to each other. An aligner's minimum feature size capability cannot exceed its resolution since its resolution is the size of the smallest optical image that it can transmit to the image cannot be accurately wafer. But superb resolution is useless if the positioned or reliably reproduced. If the aligner is not accurate that is if its - transfer is overlay characteristics are not very good or if the process of image not reliable that is if the aligner's critical dimension control is not adequate - then the aligner's minimum feature size capability will be less than its minimum resolution. used production has grown from less 5 The Tiiaximum size of the silicon wafers in eight used today, and than one inch in diameter in the sixties to the inch wafer;' users need lithographic equipment that can handle the size that they have chosen, The throughput, aligner all drive its effective cost. yield and footprint of an The faster the throughput and the higher the yield, or the more "good" wafers produced per hour, the cheaper the aligner is to operate. "Footprint" is a measure of the area that aligner requires on the semiconductor production floor. Since the customers prefer aligners to be semiconductor facilities are extremely expensive, as small possible. as Figni-p (S> CRTTTCAL DIMFNSTnN CONTROL Accuracy with which an image can he pnsit-innpd RF.Snr.UTION <^i7P nf the sniallest that can be imaee produced on thp surface of thp wafer. Resolution Actual irfigf OVERLAY posi tion Mpan distancp hetwepn Ideal imTge actual and ideal image position position. Source; Watts and Finspruch 987^ The basic concepts of alignment technology have remained stablp sincp it was developed, the three types of innovation first but the industry has seen all of that wp have identified: radical, incremental and generational, Commprcial production photolithography, which light is used has been dominated by optical in as the exposure source, but radical alternativps that make use of alternativp sources and quite different mask, alignmpnt and image transfer technologies have been explored since the seventies.** offpr customers bptter minimum early Thpy number of unsolved feature size capability, but to date both their cost and a technical problems have prevpntpd them from being widely used beyond research and development. We therpfore focus here on the optical systems that have dominated the industry, Tncrpmental innovation has bppn critical to optical photolithography's nf each component bepn significantly continuing succpss. The technology has ^ Radical alternatives to optical photolithography includp X-ray and Ton Ream aligners that x-rays ion-beams respectively as a source and Direct-write use and electrons to "write" on thf electron beam technology, that uses a beam of focused Wagner, Burggraaf, 1982^ wafer. (Chang et al., 1977; Brown, Venkatesan and 1981; 23 improved: for example modern sources are significantly more powerful -inrl more uniform, alignment systems are much more accurate lenses arp larger and suffer and less from distortion. The industry has also seen significant generational innovation. The key relationships between components and system parameters, and between system parameters and customer nepds have shifted dramatically four times over the course of the industry's history as the industry has shifted from the simple contact and proximity aligners to the more sophisticated optical systems that "scan" the mask relative to the wafer or that "step" it slowly across the wafer surface. One indication the presence innovation visible in figures of of generational is C?) (7'>, its and which show the historical perfonnance of each generation in terms of throughput, yield and minimum feature size. Each arrow reflects a particular technological trajectory, and summari;^es a history of incremental improvement within each generation. The movement between arrows from one technological trajectory to another - reflects innovation in the underlying generational technology. the We can develop more insight into the nature of generational innovation in industry by comparing the relationships between components and system parameters the contact and proximity that underlie the user need for minimum feature size in aligner. Contact and proximity aligners are relatively simple so that their feature size, the most technology can be easily described, and a focus on minimum essence of critical dimension of the aligner's perfonnance, allows us to grasp the the generational innovation without the need to present an exhaustive analysis of determinants of every system parameter. were photolithographic aligners to be used Contact aligners the first the commercially. They use the mask's "shadow" to transfer the mask pattern to the wafer held "in contact" with each other, and wafer surface. The mask and are surface. CFigure light shining through the gaps in the mask falls onto the wafer I f^^i Fisurp : rh?in£PS in th^ B;^1anrp Bptwppri Minimnni F^-itiirp Si7P inrl Yiplri Arrn;^s the Changing r.pnpvatinns of Photnl i thographir Fqnipripnt Minimum Featurp Size 3 (microns') Poor Fair Good Production Yipld^ This table is designed only to give a sense for some general Minimum feature size achieved and production yield vary great relationships. between applications. between customers and and considpred very 2. Prodiiction yields are difficult tn measure atp confidential. Typically a "poor" yield would be a yield of about 20?^ a "good' yield might on the order 60-70%, although these figures would vary with be of measured. the stage in the process at which the yield was Source: Field interviews, Internal firm records. (Henderson, 1989) ^ Figure (7): Changes in the Balance Between Mininnim Feature Si^e and Thrnughnnt Arross the rhanging Genpratinns of Phntn] i thcigraphir Fqnipment Minimum Feature Size 1 Cmicrons) mity 20 40 60 Throughput^ 1. This table is designed only to give a sense for some general relationships. size achieved and throughput vary great between Minimum feature customers and between applications. throughput of the largest wafer size that the 2. Throughput is defined as aligner designed wafers per hour. is to handle in ^Henderson, ISS'I) Source: Field interviews, Internal firm records. 24 (8) presents a schematic diagrarr. nf a contact alignpv.~> Tnntact alignprs arp sinnplp and quick to use but the need to bring thp mask and the wafpr into direct contact can damage the mask or contaminate the wafer. The first proximity aligner was 1973 these problems. proximity aligner the mask is introduced in to solve Tn a held a small distance away from ("in proximity to") the wafer surface. The separation of the rnask and the wafer means that they are less likely to be damaged during exposure, but since the mask and wafer are separated from each other, light coiuing through the mask "spreads out" before it reaches the resist, and the mask's shadow is less well defined that is in the case of a contact aligner. ^Figure it result users (9) presents a schematic diagram of a proximity aligner.) As a switching to proximity aligners were forced to trade off minimum feature size capability for increased yield. advance The introduction of the proximity aligner is clearly not a radical over the contact aligner. The conceptual change involved in going from one to the be aligners. other was minor, and most proximity aligners can also used as contact However, as figures (10) and (11) reveal, the relationships between component parameters, system parameters and minimum feature size are quite different for the introduced generational two technologies. The introduction of proximity alignment innovation into the industry. The minimum feature size capability of a contact aligner is limited by its between the overlay and critical dimension characteristics. As long as contac^t wafer "perfect" - absolutely flat and parallel across the wafer mask and the is yield image the mask is not distorted and the aligner's loss is minimal, the of its resolution. Since the minimum feature size capability is limited only by upon a gap to contact aligner does not rely upon either an optical system or only by the transmit the image of the mask to the wafer, resolution is limited been source wavelength, and until recently the resolution of contact aligners has equipment. Unfortunately in greater than that of any of the other generations of (^8">: Figiiie SrhenLitic Di-igr^m of a rnntact Alignei J L AIJGNMENT SYSTFM MECHANTCA[, SYSTEM MASK m mmwy, WAFFR CONTROr_. SYSTFM (^9): Figure Schematic Diagram of a Proximity Aligner SOURCE J L MECHANICAL SYSTEM ALIGNMENT SYSTEM (Includes gap setting mechanism) MASK WAFFR STAGE CONTROL SYSTFM Figure (10): THF mNTACT PRTNTF.R COMPONENTS SYSTFM CUSTOMER NEEDS PARAMETFRS SOURCE Wavelength -Unifonnity Col 1 imation Energy RESOLUTION MECHANICS Contact accuracy -Gap uniformity Gap accuracy CRITICAL MINIMUM Stage accuracy DIMENSION FEATURE Stabilltv CONTROL SIZE ALIGNMENT Wavelengtli —Algorithms Accuracy Targets Indicates those relationships that are critical and that limit performance of the aligner. the Source: Field interviews, Internal firm records. 'Henderson, 1989) Figiup ni): THE PRnXTMTTY PRTNTFR SYSTFM rnSTOMFR NFFDS cnMPnNFNTS PARAMETFRS SnilRCF, Wavelpngth -ITniformity Cnllimatlon Energy REsnr.uTinN MECHANICS Contact accuracy -Gap uniformity — CRTTTCAL MINIMI TM Gap accuracy FEATURE Stage accuracy DIMENSION SIZE Stability CONTROL ALIGNMENT Wavelengtli —^Algorithms Accuracy Targets that are critical and that Indicates those relationships limit the performance of the aligner. Source: Field interviews. Interna] firm records. ("Henderson, T589) 7'- prndurtion spttings contact is far from perfpct, Tn practice both the mask and often nr wafer are liam-ispd distorted, and minimum featm'p si^e is limited by the overlay and critical dimension control characteristics of the aligner. These are a function of the precision of the contact mechanism and the accuracy of the alignment system. Tn contrast, the minimum feature size capability of a proximity printer is limited by its resolution and by its critical dimension control, and these are functions of quite different component parameters. Resolution is limited not by the wavelength of the source but by the size of the gap between the mask and the wafer. As a result although the minimum feature size capability of a contact aligner is rarely limited by its resolution, resolution is very much a binding constraint in a proximity aligner. The aligner's critical dimension control is limited by the uniformity of the gap between the mask and the wafer, since if the gap is wider in some places than in others the light will have a chance to "spread out further," producing image an of a different size at the wafer. The framework that we have developed thus allows us to make the nature of generational innovation precise. Proximity alignment appears to be a minor innovation. The components, system parameters and user needs met by the aligner remained unchanged. Yet the relationships between the components, system parameters and needs apparpntly small change in a user changed dramatically. An particular component technology had important repercussions in the way in which the entire system operated. We have used this framework to analyze the later stages of the industry's technical evolution CHendprson, iq^qt. Table (2) summarizes our results and the outlines the key changes in technical rpl at ionships th^t nnderl ly gpuprational innovations that have riarked the industry's history. Tn pach case, the existing components introduction of a single new component or of small changes in . Tahle (2): A Summary of Relationa] Change Tn Photolithographic Alignment Terhnology fTenerational Major changes in: Innovation: Technology Critical Customer Needs Relationships PROXIMITY Mask and wafer MFS Mini mum feature Yields much higher, separated during siz»^^ a function of but MFS now much exposure. gap si^e, accuracy. greater constraint. SCANNING Image of mask Rpsolntion driven by MFS constraint PROJECTION projected onto numerical aperture, relaxed, but wafer by scanning wavelength. Overlay, scanner is slower. reflective optics, - a function of lens distortion - is critical constraint for MFS. STEP ^ Image of mask Throughput major Some customers have REPEAT (]) projected through constraint - function large increases in size theoretical refractive lens. of lens field and yield, Image "stepped" source energy. limits to MFS rise, across wafer. But throughput falls significantly. STEP ^ Introduction of Overlay and resolution Dramatic improvement REPEAT C2^ "site by site" constrain MFS. in reliability, lens alignment, larger Resolution driven by field size improves 5x lenses. Numerical aperture, throughput. MFS wavelength. again major reliability constraint Overlay and lens driven by distortion, interactions between lens and mechanical svstem. ("Henderson, T^RR) Source: Field Interviews. Internal Firm Records, 2f. chanp;p(i thp cnrp rel -^tinnshiips thu dptPr-'M iiprl fhp fnnrt inni ng nf thp allgnpr and tradpnffs customer iipprls. stnirtiirpfl upw across thp GpriPratinnal Innovation and Drgani national Capability. of thp Insights into thp tprhnical history indnstry arp intriguing. But our framework also implies that generational innovation shoulrl havp irnportant competitive consequpncps . Wp suggested that genprational innovation ohpolptes thp difficult for them to "relational knowledgp" of thp product designer, making it t it understand the nature of genprational innovation and ^n rpspond n appropriately. Our empirical results suggest that this insight is critical to an of the history of photolithographic understanding competitive the alignment industry, A full analysis of the indtistry's history should obviously considpr more than simply changes in the technology. But in this type of industry, in which thp performance of the product is critical to its acceptance in the market, and in which firms compete primarily on the basis of technical excellpncp, a careful implications of technical change can rpveal patterns that study of thp competitive may be important in a wide range of other industries. HI of sales by generation of equipment for the leading Table presents share been able to pxtend its firms. None of the established firms in the industry has position into the next generation of equipment, despite its experience with the and its ownership of an extensive installed base. technology 1 C3): Table Share nf •^ales hy gener-Ttinn fn» the leading nptiral '^ photolithographic aligniTent f^quipiient 'ictnr'M'^ mannf (%) Contact Proximity "scanners Step and Step .md Repeat ( ) Repeat C Gobi It 28 (ipspitp thp fpicp that thp idpntity of the pstablishprl firm changps in pvpry i nst aiipp,^ Thus traditional pcononic pxpl anat jons of t h*^ fai hue of establisheri firms do not sppp' to hold hpre. Fvplanat ions that )ely on "bureaucratic inertia," or on the difficulties that many encountpr to rpspond any t^ype organizations in attemptine of change are similarlv susppct . sincp pstablishpd f i rm« pxpliritly idpntified thp competitive threat represented by generational innovation, attached great importance to meeting used design teams that weie on average ("omposed of it, and fewer Established firms failed to than twenty five people ("Henderson, l'^88>. experience that they maintain their position into the npxt gpupration hpcause the had gained with incremental innovation in the previous gejieratjon specifically they - very difficult for the relational knowledge that had accumulated made it them to understand the nature of the generational innovations with which they were to faced or shape an appropriate response. to its We illustrate this here by a description of Kasper's failure maintain position in proximity alignment. This failure is particularly intprpsting since firm was relatively and "organic" in structurp and since thp the small appears be similar to the introduction introduction of proximity alignment to so of an incremental innovation. It illustrates graphically the ways in which subtle system parameters and user needs changes in the i nterrplationships of components, can have very significant i^ompetitive implications. T^i^8 and by 1971 was a small but Kasper Instruments was founded in , of the m.arket for contact ^lign'^rs. profitablp firm supplying approximately half with proxir^ity In 1973 Kasper introduced the first contact aligner to be pquipppd Although nearly half of all the aligners that the firTn sold from T17A capability. of the determinants of technical success in the 8 A statistical analysis that established firms were significantly less likely industry's history suggests chance. (^Henderson, to be technically successful than would be predicted by random, 1989) y onwards h.-\d this capability. Kasper alignprs wpip nnlv rarely nspd in prnximity mode, and sales declined steadily until the company left the indnstry in 1*181. The widespread use of proximity aligners did not occur until the introduction and late general adoption of Canon's proximity aligner in the 197ns. Rasper's failure is initially pu77ling given its established position in the market and its depth of experience in photolithography. There were approvimat el five key mechanical and electronic engineers Kasper during the early 197ns. at Several of them were highly skilled and i'T'aginat ive designers, and the group designed a steady stream of contact aligners, each incorporating significant incremental improvements. From 19(^8 to 1973 the minimum feature siT^e capability nf the contact aligner improved from ten to five microns. But Kasper''^ very success in designing contact aligners was a i^iajor contributor their to a that could perform to inability design proximity aligner as successfully as Canon's. Canon's aligner was superficially very similar to Rasper's. Tt incorporated the same components and performed the same functions, but it performed them much more effectively since it incorpoiafed a much more sophisticated understanding of the technical interrelationships that are fundamental to successful proximity alignment. Rasper failed to develop the particular specialist knowledge that would have enabled their' to match Canon's design, but, even more importantly, their experien<:'p with contact aligners left them without routines that would have enabled them to the knowledge gathering they understand the need to acquire it. The relational knowledge that had developed through their experience with contact aligners had the effect of fooising away from the new problems whose solution was critical to their attention the design of a successful proximity aligner. nndified contact aligner, and Kasper conceived of the proximity aligner as a product as the previous it was managed as a rovitlne extension to the line, just setting incremental improvements to the contact aligner had been. The gap 30 tn tn rtiprhanisni th.it v^'^ nspfl in thp rnnt?ict alignpv .^licn tho Pi-isk inri wafpr p^ch nthpr w;is slightly irmrlifipii ind t hp aligiiPV w.is offiTf^'l nii thp markpt. Tn dnins this. Kasppr maclp thp implicit assnmptinn that thp tprhnnlocv that had sustained support pToxinity alignrnpnt^. rontart printing could hp incrprnpntal 1 y pxt pnd^d to But, as thp analysis that wp prpspntpd ahovp rnadp c]par, pioxirrity alignment alignnpnt tpchnology. In particular, the represents a generational shift in intprrplationships between the performance of the gap setting mechanism and the feature si:^e the quite different in a proximity minimum capability of aligopr are and contact aligner. In a contact aligner, the gap SPtting mpchanism is uspd only dui ing alignment. Errors in its setting can be corrected manually, and its accuracv has little influence feature B\it in a proximity aligner over minimum si?^e capability. the quality of the transmitted image is critically dependent \ipon the gap's capability is a dirpct function accuracy and stability, and minimurri feature si?e of its size and accuracy. As rpsult thp succpssful design of a proximity aligner requires both the acquisition of some new knowledge how to build an adequate gap setting mechanism - and an understanding of some new interactions between the alignpr - in particular an component performance and the system parampters of the rpsolution understanding that minimium feature size capability is limited by and critical control thp alignpr, and that in turn resolution and dimpnsion of upon the accuracy and critical dimension control are critically dependent uniformity of the gap betwpen the mask and the wafer. mpchanism is not a trivial Thp successful dpsign of an adequatp gap SPtting betwepn thp mask task. Tn a contact alignpr thp dpsignpr can rply on the contact to the and thp wafpr to pnsurp that the mask is located accurately with respect wafer. Tn aligner, the mask must be located by "dead reckoning" at a proximity point above wafpr, and the designer must rely on the accuracy of some in space the and the wafer is the mpchanical mpchanism to pnsure that thp gap bptween thp mask 11 precise and consistent arrnss the mask surface. This call'^ for significant expertise iri the design of high precision Tiechanical mechanisms, and in hetween the the particular, for a deep understanding of the interactions design of gap setting mechanism and the other elements of the aligner. Kasper the lacked this knowledge, or, in the terTn<^ of onr framework, lacked ability to control the component parameter "gap si7e" with sufficient accuracy. Canon, on the other hand, were ahle to hring to bear a repertoire of skills and knowledge aboiit high precision mer-hanica 1 ass-^mblies deriv^ed from their pxperience related needed tn develop this in industries. The design group at Kasper had not type of knowledge since it was not critical to the performance of the early contact aligners. But this lack of a specific piece of t'echnical knowledge js not the whole story. More critically. continued to make use the problem solving and Kasper of they experience information gathering strategies that had developed during their develop with the contact aligner, and as a result failed tn understand the need to this critical body of knowledge. Tanon invested heavily in building a deep setting understanding of the precise relationships between errors in the gap mechanism, the stability and uniformity of the source, and the performance of the dimension alignment system in the determination of the resolution and critical control of the aligner. This them make informed tradeoffs in the enabled to overall design the to understand, for example, where design effort of system, and They assumed that the could be most fruitfully spent. Kasper did not do this. introduction of a gap between the mask and the wafer during alignment was a minor, incremental change and so put little effort into understanding the new between setting mpchanism and re'^olutlon and interactions that it created the gap critical dimension control. Thus they failed to under'^tand the enormous impact could have on the perfor-^iance of that improvements in its accuracy and precision , thp alisiipr, and the iiroximity alignpi that thpv r|p>;i£npi1 •iuffprpd frnti. an arciirate mechanisi-n untPliahlp and insufficiently gap «ettine to the K.aspey 's failure understand ohsnlesrenre nf their relational knowledge is demonstrated graphirally hy two incidents. The first is the firii^'s interpretation of early complaints about the accuracy of its gap setting the fnask niid the wafer can hp mechanism. In proximity alignment misalignment of caused both hv inaccuracies or instability in the gap setting mechanism and by introduced processing, at-tributpd of the pvr)blems distortions during Kasppr many that usprs of its proximity equipment were eyppripncing to processing error, since they "knew" f rniTi their experience with fhe contact aligner that their gap setting mechanism was adequate to task. As a result they devoted vpry little time to the retrospect improving its performance. In this may sppt' likp a wanton misusp of information, but it represented no more than a continued reliance on an served well The cppond illustration information filter that had them historically. is provided by their response to Canon's initial introduction. The ranon aligner by be a "copy a K.asper was evaluated a team at Kasper and pronounced to of machine." Kasper evaluated it against the criteria that they used for evaluating their own aligners - criteria that had been developed d"ring their expprience with features that made it a significant advance, contact aligners. Thp tpchnical were not observed because they were not particularly the redesigned gap mechanism, considered important. the core design Kasper 's engineers were not incompetent. They assumed that contact aligners could a1«o serve as concepts that underlay the design of their the the basis for a proximity aligner, and that the relational knowledge and knowledge capabilities that they had accumulated dtiring their processing could be transfprrpd to proximity aliguprc. experience with contact aligners from more than this failure in design. Kasper's commercial failure stemriied system of sufficient The company had problems designing an automiatic alignment , . PicrwYHry, and in a hi^h vn1nin(=> "nnnf artnri nj: farility, Thoy il^n :;nffpipr1 >Tianaf,in° ^pvt^ral rapiH of top rr'-^n ^^^''''''^ti^ iliirii!^ ' bp 1 >tp rii'iptpin thrnugh cbanjies thf nh'^nl psrenrp of yplatinnal Vnowlpd^p brnn^ht ihnnt by tbp SPVPnties. But f:\rt-nr i nl" rndurt inn of ^pppratinnal innnv'atinn was a 'ni''i'"a1 in Ibpic dpc-line. A similar study nf thp failnrp of tbp ntbpv dnrrinant fivis in tbp industry suggests tbat a rplianrp ibmit fbp tpcbnnlnpy dpri^'pd frnn thp on Yelational knowlpd^p prpvinus ppnp)'atinn was a sipnifirant factor in pvplaining thp failnrp of all nf tbp pstahlisbpd who to >nppt tbp ^hrpat of npw pntry ''Hpndprson, firms wpfp iinablp 1989) TV SuTTiTnary and Conclusions suggpstpd that the charactpri/at ion of innovation as pitbpr "radical" We bavp or "incrpTTipntal" is incornplptp, and nnist be snppl emented by the concppt of the perfonnance of "genprational" innovation. While incrempntal innovation changps an existing set of components, and radical innovation changes the elements of that gpnerational innovation cbangps set and the rel ationsships betwppn them, performancp and relationships, but Ipaves thp plements largely unchanged. Generational inno\'a^ion 1*= sometirr'ps tjiggprpd by cbangPS in a particular but fundamental reconfiguring of the component technology, its essence is a customer technological system around an essentially stable set of components and Tt qualitatively different from both increrrental and radical innovation needs. is i cat ions and has equally important competitive and organizational ir<pl individual of We used the mode] of a product designer as a boundedly rational limited processing capabilities to highlight thp diffcrpnt information information types innovation pose fov the product procpssing tasks that thp diffprpnt of knowledge development process, and discussed the types of knowlpdgp and types innovation acquisition routines that a designer faced with the different of tn that the degree that this is likely devplop in consequence. Wp suggested to 34 nmrlpl i> indif^ativp of fnrrp>? at work itT^iiic ors^ni zii"inn-^ thf^-n •^.pri'^f- it innal '"^v .'nt nn^-yppc tpil 'Mipippt vf> cnn«"qn<^nr^^x ^iiiri^ it ch^inse h^vi^ in.pmt ^nd i t i nhc;n1ptpp thp rplatinn?i1 knnwlpr!p»^ nf thp nrgani 7at inn . This issue is an important avpnnp for fntiirp ypsparrh. nf t}iP ^nH Wp illnstratpr) on, ftaiiPwork thrmipb a rlp-irri pt inn tprhinral rompptitivp px'olntinn of thp •iprni conr1urto> phnt nl i thop vaphi f' ilipmnpnt i nrlii'^t ry . Wp showpd that the i nt » oHiict pinxi'^iii t ali^nipnt^ an apparpn^ly inn of y into thp "i ncrpffipntal " inno\fat inn , in fact introdiicpd si£nificaiit '"hangp thp rplat ionships hptwppn cnrpponpnts , nid that Kasppr's rnntinopH tp1ian<"p on knnwlpdpp and infoi-Tiation procp'^'^ing capahilitips that thpy h-id dpvploppd thp through thpir pxpprience with contact alisriprs 'ipvprply handicappprl hp firri's ahility to takp adv^antagp of proxirnity ali^naiPnt. of For pxample, wp need Thi-^ resparch oppns up a ntimbpr important qupstions. to to develop metrics for generational innovation and explore its interactions with »^hp othpr economic and organizational forcp'^ that -ihape firms and indnstrie?. R\it we believe that the concept of genprational innovation ha<^ potpntially grpat m^ipptitivp powpr and will pro\'p uspfiil in thp 'Atvidy of thp organizational and implications of technical change. : . . . . REFFRFNCFS and I. Freeman D.Schendel Hannan, M. Cooper, A.C, and Williain Abemathy, "Strurtural Inertia "Stiqtpgic Response tn 1984 197A Productivity 1978 Thp n.-g Change. uiizational Technological Threat-. Baltimorp: Johns Dileniri-'a. 19: hl-f,9 Amerie^n Sociological Business Hori7ons. University Prpss. Hopkins 4^1: 149-1^4. Rpvievj, D. Beard. and Kim Dess.G.G. and Abemathy, William, Henderson, Rebecca M. "Dj-prisinns of 198/i Clark 1^88 "The F^i lure of Task thp Organisational nSS "Innovation: Mappin? ." Pst -^^l i^l^prl Firm.s in Fnvironmenfs Winds of Creative Technical the Face of Administrative Science Destruction." thf 5-^-73. Change: A Stvidy of 29, lA: 3-22. Quarterly, Research Policy, Semiconductor Photolithographic .T.F.mitton, newar, R.n. and Alexander, C. Industry." Radical anr Alignment 198fv "The Adoption of iqA.'i N'otes on the Synthesis of Ph.D. Unpublished Incremental Innovations: an Carobridse: Harvard Form. dissertation, Analyses." Press. Empirical University Massachusetts Management Science, Tnstitutp of 32: U22-1432. Arrow, K. Technology. Organisation. 197A The limits of Norton. Dosi , G. New York: Henderson, Rebecca, M. Paradigms and 1^82 "Technological 1989 "Heterogeneous Firm Trajectories: Technological Brown, W.L., T.Venkatesan and C^pabil ity and the A Suggested Interpretation A.Wagner Established Failure of of the Determinants and Lithography." T^Sl "Ion Beam Paper, Firms." Working of Technical Change. Directions Solid State Technology, Massachusetts -icy, 11: 147-1<^2. Research Pol August of Institute Technology. and W. P. Bridges Fttlie.J.F., Biirggraaf ,P. and R.D.O'Keefe 1983 "X-Ray Lithography: and J., Sawers, D. Strategy and Jewkes, 1984 "Organisational Optica] 's Heir." Stillerman for R. Structural Differences Semiconductor 19AC! The Sources of Incremental Radical versus International, September. Invention M?inagement Innovation." New Macmillan. f.82-f>95 York: Science, 30: G. Stalker Burns, T. and iqfiA of Innovation. The Management and Kimeberly,J.R. F.ttlie.J.F. London: Tavistock. J.Fvanisko Policy imong M. 1983 "Organisational 1981 "Organisational the Food Suppliers to Chang, T.S. et al. Innovation: The Processing Sector." 1977 "Flectron-beam Lithography Influence of Individua Academy of Management Finer fine." draws a and Organisational 2*^: 27-44. Tournal , Electronics, May. factors on Contextual of hospital adoption Freeman, C. Kim B. Clark, technological and Industrial 1982 The Economics of interaction of design 1985 "The administrative Innovation. Cartridge, and market hierarchies innovations." Press. technological MIT concepts in of Management Academv Policv, evolution." Research 24: 689-713. Journal, and D.M.Newbery Gilbert, R..I. 14: 235-251. thf Patenting and 1982 "Preemptive Kuhn, Thomas of Monopoly." Persistence Clark, Kim B. Structure of 1970 The Economic Review, American 1987 "Managing Technology in Scientific 72: 514-95. International Competition." Chicago: Revolutions. Ced.^i Spence and Hazard, In Chicago 'University of International Hage ,1. Press. Organisation 1980 Theories of Competitiveness. New York : Wi 1 ey Camibridge: Bal linger. Ul9 \ . . , . I.W.I.nrsch Sahal I.nwrencp.P.R. anrt l^A? n>-2ani zai" ion aiul Fnvivnt ^rhnolnoicil GnirU niinnjc;: Hornpwnori and Innovation Avpnijp'^." Rpcoirrh Pnlir-y, 1 'i : fy]-^'^. Mansfield, F.dwin T.S. T^AR TndnF^trial Rpsf^arrh and Saviotti,P.P and Mot calf p "\ Tt^'^hniral Innovation. iq!^4 Thporptical Approach to Npw Yo>k: Norton. t-hp Hon^trvictioti of rmtnut indicator- Tpchnological Marples, n.I.. Rpspa-.rh Policy, 11: I'.I-ISI . 116] "Thp Dpcision<= of Fnginperin^ Dosi^n." Simon, Herbert TF.FE Transaction"^ on l'^'^'^ Thp Scipncp? of the Artificial Enginpprin? Managempnt rarpbrid^p: Ma<sa'-h"c-or t-; Timp. Institntp of Tpchnoloov "rpsc Morh,M. and E.V. Morse Tiishman.M.L. and P.Anderson lOSf, 1977 "Si7p^ centralization ind "Tcrhnological nisconti nm t IPS organizational adoption of and Organisational Eiivi roiinipnt' Administrative Soipnce innovations," American AlH-'ti^s Sociolosical Review, njiartprly, 11 : , Octoher. 7K-.-72S. Watts, R.K. and N.G.Einsprnch, (pd.) Nelson, Richard, and Sidney 1987 Lithography for \T,ST. Acade^iic Winter New York: Prps 1982 An Evolutionary Theory of Econoniic Changp. Carnbridge: Harvard Hnivprsitv Prpss, Newell, A., and Herbert Simon 1972 Human Problem Solving. riiffs: Englewood Prpntice Hall. Reinganaum, J.F 1983 "Uncertain Innovation and the persistence of Monopoly." American Economic Review, 73: 7A1-A8. Rosenberg, Nathaniel 197^1 Technology. Perspectives on Cambridgp: Cambridge Univprsity Prpss Rosenberg, Nathaniel 1982 Tnsidp thp Black Rox: Technology and Fconorirs. Cambridge: Cambridge University Press. Rothwell Roy 198<^ "The Role of Small Firms in the Emergence of Npw Tpchnologipj FroiT! Freem'am, C. Ced.) Design, Innovation and Long Cycles in Economic Development, london: Francis Pinter. Date Due ' srsaj NOV 6^9^ I3/i:, *EP- fee 13 199) 199(jt aPR. G 19gB ^6 OCT IS^u PEB. ^^i mo 1 i999 /WAV A7iB?r MAY 2^199 pfB^f.^! FEB N. 1 6 1908 199f' MAR \m* 6 19S9 APR:2 4r9^MAn2 MIT IIBRSRIES QUI Q05b7M3fl M 3 TOflO http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Administrative Science Quarterly Unpaywall

Architectural Innovation: The Reconfiguration of Existing Product Technologies and the Failure of Established Firms

Administrative Science QuarterlyMar 1, 1990

Architectural Innovation: The Reconfiguration of Existing Product Technologies and the Failure of Established Firms

Abstract

C9 t. <^^ rt PAPER WORKING SCHOOL OF MANAGEMENT ALFRED P. SLOAN INNOVATION: "GENERATIONAL" SYSTEMS THE RECONFIGURATION OF EXISTING FAILURE OF ESTABLISHED FIRMS AND THE Clark Rebecca M. Henderson and Kim B. May, 1989 WP 3027-89-BPS MASSACHUSETTS INSTITUTE OF TECHNOLOGY 50 MEMORIAL DRIVE CAMBRIDGE. MASSACHUSET" "GENERATIONAL" INNOVATION: THE RECONFIGURATION OF EXISTING SYSTEMS AND THE FAILURE OF ESTABLISHED FIRMS Rebecca M. Henderson and Kim B. Clark WP 3027-89-BPS May, 1989 Rebecca M. Henderson Assistant Professor of Management Massachusetts Institute of Technology Kim B. Clark Professor of Management Harvard University Research, Harvard Business This research was supported by the Division of also like to thank School. Their support is gratefully acknowledged. We would Dataquest and VLSI Research Inc. for generous permission to use their published those data, the staffs at Canon, GCA, Nikon, Perkin Elmer and Ultratech, and all photolithographic alignment technology who gave so individuals involved with generously of their time. 4 1989 I JUL 2 "GFNFRATTnNAl " INNnVATTON: THF RFCnNFTGKRATTON OF FXTSTTNG SYSTFMS AND FATf.I'RF OF FSTABLISHEO FIRMS THF t^ithpr "inrrpnipntal or ThP tiaditinn-Tl catPgori ?:ation nf innovation as " "Generational" innovation "radical" is incomplete and fundanient al ly inisl eadi ng . its elements - i^; that reconfigures

Loading next page...
 
/lp/unpaywall/architectural-innovation-the-reconfiguration-of-existing-product-Yh6vY0jo3I

References

References for this paper are not available at this time. We will be adding them shortly, thank you for your patience.

Publisher
Unpaywall
ISSN
0001-8392
DOI
10.2307/2393549
Publisher site
See Article on Publisher Site

Abstract

C9 t. <^^ rt PAPER WORKING SCHOOL OF MANAGEMENT ALFRED P. SLOAN INNOVATION: "GENERATIONAL" SYSTEMS THE RECONFIGURATION OF EXISTING FAILURE OF ESTABLISHED FIRMS AND THE Clark Rebecca M. Henderson and Kim B. May, 1989 WP 3027-89-BPS MASSACHUSETTS INSTITUTE OF TECHNOLOGY 50 MEMORIAL DRIVE CAMBRIDGE. MASSACHUSET" "GENERATIONAL" INNOVATION: THE RECONFIGURATION OF EXISTING SYSTEMS AND THE FAILURE OF ESTABLISHED FIRMS Rebecca M. Henderson and Kim B. Clark WP 3027-89-BPS May, 1989 Rebecca M. Henderson Assistant Professor of Management Massachusetts Institute of Technology Kim B. Clark Professor of Management Harvard University Research, Harvard Business This research was supported by the Division of also like to thank School. Their support is gratefully acknowledged. We would Dataquest and VLSI Research Inc. for generous permission to use their published those data, the staffs at Canon, GCA, Nikon, Perkin Elmer and Ultratech, and all photolithographic alignment technology who gave so individuals involved with generously of their time. 4 1989 I JUL 2 "GFNFRATTnNAl " INNnVATTON: THF RFCnNFTGKRATTON OF FXTSTTNG SYSTFMS AND FATf.I'RF OF FSTABLISHEO FIRMS THF t^ithpr "inrrpnipntal or ThP tiaditinn-Tl catPgori ?:ation nf innovation as " "Generational" innovation "radical" is incomplete and fundanient al ly inisl eadi ng . its elements - i^; that reconfigures a technical system without changing innovation often both incremental and radical innovation and ha< qualitatively different from consequences. important and unexpected organisational and competitive the concept's paper defines generational innovation and illustrates This technical and competitive explanatory force through an empirical study of the the semiconductor photolithographic alignment equipment industry. history of . . TNTBODUCTTON "Wp rippd to know morp abnut differpncps bptwppn rnmpptenrp rtpptrnying advances.. Caboutt what distins'nshps and competence-pnhancjng tpohnologif^al advances. CTushrnan and bPtwpen incrptripntal improvpments and dramatir " Andprson (]^Rf^)^ to and transform organi i^at inns and thf> Tbp power nf npw technology challenge Schnropeter structure of indiistrips has bepn an irfportant thpme nf research si ncp t'TpwkPS, innovation creates great difficulties for e<t.=ih1 i •ihed fivTi.s, "Radical" 19f)9; Cooper and Schendel, 197(^; Rothwpl 1 n^f.; Tush'oan Sawers and St i 1 ] erman , . successful entry and pvpu and Andprson, I^Sf-) and is oftpn the basis for reinforces the rpdpfinition nf thp industry, while "incrempntal" innovation oftPn and incremental innovations havp different dominance of establishpd firms. Radical competitive consequences because they require quite different organizational innovation feeds on and rpinforces the existing problem capabili tips. Incremental while radical innovation forces solving capabilities of established organizations, approaches them to ask a new set of qupstions and to employ new problem solving 196A; Arrow, 1974; Hagp, 1 9R0 ; Dess and Beard, I'^Hh: Fttlie, ^Burns and Stalker, 19c;A'i. Bridges and O'Keefe, 198A; Tushman and Anderson, between radical and incremental innovation has produced The distinction is growing evidpnce important insights, but it is fundamentally incompletp. Thprp "lodest that are numerous technical innovations that involve apparently there to pxisting tpchnology, but that have quite dramatic compptitive changes the 1987). Take, for examplp, thp case of Xerox and small consequences ^ Clark, copiers confrontpd in the mid-1970's Xerox, the pioneer of plain pappr copiprs, was than copiers that were much smaller and more reliable with competitors offering or the traditional product. Thp upw products required no new scientific . rnmp;^iny invpntpd thp cnre enginpprins knowlpdgp. but rlpspitp thp f-^ct th^t thp had (if»l>- vpjnv md hifl piior'Tions pxporipiice in the indiistyy, h i almost t pchnol n^i''-^ that ti^ip yprov pight vp^rs to hring -^ cnmppti t ivp product into thp "laykpt. Tn of it^ markpt -^harp and ctifferpd sprions financHnl prohlp-is iTlirk, lost half ]'^S7'i, modpip that rply on dist i n-^tion?^ bptwppn radical and i ncrpppnt al Fxisting rpasons thesp t ypps of innovation innovation providp littlp insight into thp rnndpl should havp such cnnspqupncps . Tn this pappr wp dpvplop and apply such a parts, Wp prpspnt a fra'-npwork that allows ns to Thp pappr has four first bPtwppn changps in analyzp thp naturp of tPchnical changp and to dPVPlop thp links that thpy prpsent tn an tpchnology and changes in Ihp inforaation procpssing task this organization ^section 2). We dpfinp incrPFiPntal and radical changp within ." framework and charactprizp an important class of innovations as "generat ional We is qualitativply diffprpnt from radical and argue that generational innovation changes the performancp of an incremental innovation. While incremental innovation the elpments of that existing set of components, and radical innovation changes the relationships between them. While radical set, generational innovation changes wholly different system, innovation challenges the established organization with a gpnprational innovation is to learn about a differently the challengp of gained through experience configured version of an pstablished system. Knowledge tpchnologies often rpmains with incremental innovation about particular component betwppn relevant, but "relational knowledge," or knowledge aboiit the interactions performance and user needs is rpndprpd obsolelp. componpnts, systpm we summarize competitive and technical history of In section 3 briefly the a useful the lithographic alignment pquipnipnt industry. The industry offers and to examine its power in explaining context in which to illustrate the model generational innovation. the competitive and organizational consequences of capital equipment used in Photolithographic aligners are sophisticated pieces of improvpd of integrated circuits. Aligner performance has the manufacture hut thp cnrn tpchnnlnpips hn/p flramat irall V over thp last twpnty-fivp ypais. i itirlnstry 'narsinally sinrp thp tpohniqnp wis first invpiitprl. Ypt hp changpd only turbiilpncp. markpt 1 padprshi havp hppn has bppn rharactPii ^pfl by gvpat Chansps in p md incnnhpnts reqiipnt pntry has nrcnrrpd thrnushi>ut thp industry's histniy, f , i ndnrt of dpclinps in markpt sharp fnl lowing thp i nt inn a havp nftpn suffpipd sharp npw ppnpratinn of pquipmptit. thpsp pvpnts arp pxplainpd hy thp intrusion of gpHPrational Wp hplipvp that gpnprationa] innovation rpquirpd littlp npw innovation into thp industry. Whllp rplational VTiowlndgp. scipntific knowlpdgp, it rpquirpd signifirant •shifts in taking thp industry had grpat difficulty rpcogni^ing and FstablishPd firms in a rpsult thptr dpvplopnipnl efforts wptp advantagp of thpsp shifts, and as significantly Ipss pffpctivp than thnsp of pntrants. the with a brief summary and with a discussion of Wp close the pappr organisational rpsponsps to implications of this resparch for undprstanding tpchnological changp. TT Conceptual Frapipwork is the existing litpraturp on tpchnical innovation Thp cpntral notion in the existing design, and introducing a distinction between refining and improving an CMansfield, departs a significant way from past practice. new concept that in innovation introduces iqf>8; Moch and Morsp, 1^77; Frepman, 1982;) "Tncreipental" thp potential of the minor changes to thp existing product, exploiting relatively Dutton, 198^.; design ^Fttlie, Bridges and n'Kpefp, nSA ; Dewar and established that l^Sf^). and Winter, for examplp, have argupd Tushman and Anderson, Nelson trajectories" during incremental innovation occurs along a series of "natural 1*^82). and Winter, "locally" for appropriate solutions ("Nelson which firms search nq82i has ^Kuhn, 19701, Dosi Building on Kuhn's work in the history of science, nnr^irii g'"" snggpstPfl that Inrrpmpntal innn\Mtinn ncrnr's within ?\ "tprhnologif-^ 1 , or thp pptablishpd "..outlnnk, spt of iirnrpdiirps , ripfinitinn of rplpvant prnblprn'^ a,-) to snlntinns," and nf thp spprifir knnwlpdgp rplatpd thpir "Radical" innnvatinri, in contrast, is haspd upon a diffpimit set of principles and oftpn oppns \ip wholp ripw inarkpts and enginppring and scientific fpttlip, I'^'^A; Dewar and Dntton, I'IRA; potpntial applications Rridgps and O'Kepfp, ronroent- Tushnian and Andprson, n8*^>.Thi)s the cotTrnprci al and tpchnical pnvi associated with radical innovation is far n^ore dynamic and imcprtain than that (^Dpss and Beard, T^iRA). associated with incrpinental innovation bnth the snpplv Innovation's impact on compptition depends on its effpcts on the supply side, thp two pattprns of innovation and demand sides of the market. On are associated with very different organizational capabilities. Radical and require not only a difference in knowledge base, but also incremental innovation general it appears that more formal, in organizational structure and process. Tn communication, "hierarchical" organizations that rely on structured patterns of routines and procedures are best equipped to undertake incrempntal problem solving organizations in which the innovation, while more "organic," "entrepreneurial" flow relatively unstructured and which are more responsive to information is innovation (Burns and dramatic change are best equipped to exploit radical 1981; Stalker, 1966; Moch and Morse, 1977; Hage , 1980; Kimberly and F.vanisko, 1986^ Ettlie. Bridges and O'Keefe, 1984; Dewar and Button, Ettlie, 1983; betwppn innovation and organizational capability that It is this connection its ability to transform thp compptitivp environment. gives radical innovation to to adjust (Nelson Organizational capabilities are difficult create, and costly and Freeman, 198A>. An established firm faced with and Winter, 1982; Hannan must makp them vety radical innovation must make very diffprent decisions and innovation differently. The number and magnitude nf the changes that radical stock of knowledge and in the way in which it requires in the firm's existing hp rlfiimting. Thus sevpral stniiips havp "^hnwn that procpssps inforriatinn may radical innovations tpnri In nrivinatf^ outside of estaM i -ht^d finns, and that often have great diffirnlty in making the shift to a new established firms ]'^<^'^] Cooppr and Sch^nd^l 1*^7A; "trajectory" Cjewkes, .Sawers and St i 1 ] erman , , nSA^. Rothwell, 19Sf.; Tushman and Anderson, Differpnces in organizational capability help to explain changps in the also shaped by an innovation's "suppliers" of innovation. But competition is effects on demand. The cnmulative economic effects of incremental innovation arp often broadens a product's appeal often significant, and incremental innovation and expands its market. Radical innovation, in contrast, offers customprs a wholly new set of possibilities, or meets established needs in a wholly different way. concept may bear some relationship to the pstablished design, thp Although the new the new needs it meets are change in price/performance is usually so dramiatlc, and sut)stitntp for thp so different, that the old product becomes at best a very poor innovation is defined as new design. Indeed, in the economics literature, radical Innovation that creates such changes in performance or cost that the old product new competitive prices (^Gilbprt and cannot act as a substitute for the even at Newbery, 1982; Relnganaum, 1983>. suggests, research on Innovation, organization and As this brief summary years has added important insights into competition over the last several characteristics and incremental and radical innovation and the organizational associated with them. However there has been strikingly competitive consequences innovation. little discussion about intermediate classes of break open the "black We believe that this neglect flows from a reluctance to box" that specific technology in order to understand its competitive and is a students of technology and economic organizational implications. Although some n97Ajq82t; Abernathy, n^TRI; history (See for example, work by Rosenberg, (198f.U have described Abernathy and Clark, 09851; Clark, n98S~i; and Sahal , . ' structure to some extent, few studies have technologies and examined their inner of the tt^chnoloey affect the developed an understandine of how thp characteristics process. There is an intuitive sense in the literature, and some innovation different technologies may be empiiical evidence, that innovations in very TMoch different in character because of the underlying structure of the technology Sahal 198A; riark, 1987^ But and Morse, 1977; Kttlie, Bridges and O'Keefe, 1984; , concepts like we have few ways to talk about these differences beyond economic minimum efficient scale, returns to scale and appropr i abi 1 i y framework that allows one to define and analy?e intermediate Developing a to understand an important das*; of classes of innovation may help us understanding of innovation, technological changes, and may also deepen our compptition in general. organization and Framework The f" . we present in this section is summarized in Figure H The framework terms of product technology, In order to focus the discussion, we develop it in starting point the concepts are genpral; they apply to processes as well. Our but - - product the product as a whole the "system" and the is the distinction between - established distinction. in its parts the "components." (This is an old and well the work Marples n9Al) and Alexander (19fi4K^ We conceive of See for example, by a roopi fan. Tts major product as a set of components, (\K Take, for example, guard, a simple components include the blade, the motor that drives it, the blade intuitive mechanical housing. For the moment we will use the control system and a the (1984) have proposed a framework that is related to 1 Saviotti and Metcalfe sets of product one we propose here. They distinguish between three that describe characteristics: those that describe its technical features, those of its the services that it performs and those that describe the method production. "> Figure ( ] : A Basic Framewnrk SYSTEM USER NFFDS rOMPONFNTS PAR^MFTFRS Xi Si^Y- 7i Yi fi^x-) Z2 = S2^Y*> x2i...x^. f.fx*^ Ya Z„ = S«(Y-) . . .X" X^l Y„. f,Jx*) : . and physically distinct notion that each rnniponpnt fills a distinct function is rnmponpnts. from thp othp> parai-ppters" Cv>, which F.ach componpnt is charactpri/pd by a spt of "compnnpnt attribute of thp componpnt, For the fan bladp, for pxamplp, describe some physical wpight of thp bladp and its dinpn'^ions . A the component parameters include the exhaustive description of complpte set of component parameters would constitute an each component characterized by a set of parameters (Y) At the system level, the product is example physical properties of the system as a whole. Tn our fan that describe the weight, the voUimp of air moved the system properties include things like total system property like the per minute, and resistance to impact. Any individual betwppn spvpral moved per minute is determined by the interaction volume of air function the amount of air a fan puts out is a component parameters. For example, the powpr of the motor, and the pfficiency of the size and shape of the bladp, the model by showing with which the motor drives the blade. We represent this in subset of the component parameters. each system parameter as a function of some Thus , we have argument of fj^, x* is an index of the M system parameters, and the , where m is fO is a statement of the component parameters. The function a subset of the system the relationships between the componpnt and physical laws governing from mathematical formulae dprived parameters. Sometimes it can be approximated by empirically. of the physics involvpd or pstimated an understanding parameters. For parameter can affect spvpral system A single component that not only the amouiit of air it example, the dimensions of a fan blade affect restricts the appeal and its portability. This puts out but also its esthetic to dpsign a Tt very difficult, for examplp, available set of system parameters. is to keep a that puts out enough air fan that weighs as little as an alarm clock but 10 hiindiprl people coni . Similarly, rorrponpnt parampters nften ititevart witli earb rlpt a systerri paramPtey . The pffpct of a rhangp in hlade shapp on nther in eimi ni ng power of the fan's motor. thp amount of air moved, for pxamplp, depends on the tlip dpsign of a product is a romplpx process of simultaneously dptermlning Thus componpnt and systpm parametprs. rhangps in the So far thp fraTipwork givp<; us thp basis for talking about of the technology. Rut we also need to understand the internal striicture Wp bpgin by as'^nining connection between the technology and customer needs, customers evaluate the product against some set of "needs," fZ~t, In thp case of needs includp pasp of operation, room fan the relevant criteria or might ('brand image, appearancp"> . The extent reliability, transportability and aesthetics plemput to which product meets a particular need, that is, thp valup of a given parameters of the product. A fan's of Z, dppends on some subset of the system dropped, degree of reliability, for example, depends on its resistance to bping time between failure of the entire system. We hit or twisted and the mean represent these relationships in the model as: the M the N user needs, and the argument of g„n is a subset of where n indexes of the nature of the customer and of system parameters, Y* . gnf^Y*") is a function physical relationships the way in which the product is used as well as of the uses to product's system parameters and the criteria that the customer between the the extent to which a user's need for evaluate the product. Thus, for example, relationships Ce.g, system reliable operation is met depends on a set of physical thp fan's dpsign and reliability, frequency of drops") that are partly inhprpnt in fashionable or on fan is used. Meeting the need for a partly dpppndpnt how the local intpraction betwppn the fan's dPsign, its "attractive" fan dpppuds on the be. what an "attractive" fan would environment and the customer's expectations of underlying Such expectations may be less stable than the physical relationships . rpliability. Thus, gn'^V't capturp;^ bnth thp physicil, anrl the hehavioral and thp nsp nf thp prndnct n^ps social procpssps that dpterminp thp ways in which intn iispr npcds. Tf cnstnmprs diffpr in thp way in thp cnntpyt system paramptprs in which thpy nsp thp product , thpn thp function s^JY""* may diffpr bptwppn customprs a choicp of tradpoffs hptwppn Just as thp dpsign of a product implips cnmponpnt and system parametprs, so product dPsign also embodips either implicit pxplicit tradpoffs between uspr nppds. The users' own preferences across these or "utilitv function" that captures the value tradeoffs can be reprpspnted through a to the customer of some particular set of levels of uspr needs. = (3> U„ hrz^ hCZ"*, Where o indexes the customers with different preferences. The function embody interactions between its like the functions h^fx") and gJY"^ may aesthetic appeal may not enhance the value of a arguments. For example, additional quiet operation fan if It does not meet some minimtim standard of reliability, and highly valued if thp fan Is a relatively small one. may be much more Types of Technological Change "incremental" and "radical" This framework allows us to characterize intermediate innovation and provides us with a useful framework for analyzing "dominant of innovation. In the context of the model, the emergence of a classes 1986^ equivalent to the emergence of a stable design" fAbernathy, 1978; Sahal , is needs. "Tncrerrpnt al" innovation set of components, system parameters and customer the values of one or more improves individual component performance, changing changing the values r>f some of the system component parameters, and thus sets and Z and parameters and user needs. But it leaves the elements of the X,Y g^'^Y*"', unchanged. thp rplat ionships bPtwppn them h./x*!! and 12 occurs when a new design changes the set of components, "Radical" innovation elements sets X, Y, and Z, thereby system parameters and user needs, or the of the relationships between them, or the form of the functions h„(X*) also changing the completely and g„(Y*). In its extreme form, radical innovation introduces a radical" components and a wholly new system. But in general, "more different set of affect more system innovations introduce more new components and consequently parameters and more user needs. generational innovation as innovation that changes the We define a parameters and user needs of the relationships between the components, system needs set of components, system parameters and user technology but that leaves the innovation changes the value of themselves relatively unchanged. VThile incremental innovation changes the elements of the sets X, Y, the sets X, Y, and Z, and radical generational innovation relationships that link them [h„(X*) g„(Y*], and Z and the , the elements largely unchanged. changes the values and relationships, but leaves triggered by changes in a particular component Generational innovation is sometimes the technological but its essence is a fundamental reconfiguring of technology, components, system parameters, and system around an essentially stable set of customer needs. that the model allows fan example can illustrate the distinctions The room air product concept is a large, Suppose, for example, that the established us to draw. with the motor hidden from view and electric powered fan, moimted in the ceiling, mounted on The control system is an on-off switch insulated to dampen the noise. that run inside the wall and connected to the fan through a set of wires the wall and ceiling. blade design and in innovation could involve an improvement in An incremental cooling. The introduction of a power of the motor to achieve a higher rate of the the established would be a radical innovation. Not all of central air conditioner would involve using fans would become obsolete. The new technology knowledge base new electric motors and fan design. But move air and would require knowledge of to . apsnriated with cnmprpssnr?; , re^f rigerantp anii Mipir ac'.cnriai'pfl control.=; cmmporipnts would Tcld whnlp npw tprhniral di '^cipl i pps md new intPi -relationships be used different way and might be sold Furthermore, the product would in a very to wholly different rustomer'^. The between radical and i ncrpiriental innovation is therefore distinction large mounted clear. What of g'^nprat ional innovation"!" For the maker of ceiling room fans, the introduction of a portable fan would be a generational innovation. basic components would be largely the same (f.g. blade, motor, control While the different materials) and thf system!*, the design choices would be different ("e.g. this values of the component parameters would change (e.g. smaller dimensions). Tn from ceiling to portable fan has the character of an in^rpmental sense, the change interactions innovation. Rut there would also be significant changes in the needs the product could meet. The between components and in the menu of user room would smaller size and the co-location of the motor and the blade in the focus attention on new types of interaction between the motor size, the blade fan could generate. Shrinking the size dimensions and the amount of air that the with new properties, as well of the apparatus would probably require new materials could new interactions between performance and weight. On the user side, there as reliability. It is the need to be new tradeoffs between ease of use, safety and that new of interaction and these new sets of tradeoffs explore these patterns sets apart this kind of innovation. knowledge ,-=ind the knowledge Radical innovation obsoletes much of the existing obsolescence is usually processing capabilities of established firms, but this knowledge triggered by immediately evident. The obsolescence of established firm to much harder to observe and may he more difficult generational innovation is -explore their competitive correct. We can best illustrate these differences, and undertake discussion of the knowledge required to significance, through a successful product design. u nf tn rnncpntr.itp nnr attpntinn nn the problern prndnct desipn We hflvp rhnspn reqni tbp intpgntinn nf tprhnologic il since it is one that pxiilicitly rp'^ it the markpt and of rnstorier reqni rements = Moreover knowledge with knowledge of snggp'ited rritiral tn ^nrrpssfnl a problem that several re-^earrhers have is is MSSSi, n9R7>, Freeman nq^?^K organi7ational response to innovation. fClark undprtaking prodiict thp discussion that follows we model the organization Tn individual with limited knowledge and with development as a boundedly rational This abstraction i "^ in the tradition of limited information processing abilities. implications of technical who have studied the competitive a number of researchers and Winter nq82'> and Simon n9^9), and we change, including Arrow (n74^. Nelson problem. The implications of that it provide a useful "first cut" at the believe the way in particular organization will clearly depend upon any innovation for a is managed, but we have abstracted from which knowledge inside the organization differpncps in thp infoi-mation this issue in order to focus clearly upon the types of innovation. tasks presented to the organization by different processing to dpsign a npw product. A designer Consider the information that is required components to bp able to technologies of pach of the must know enough about the CMarples, 19(M Ramstrom and Rhenman generate somp set of component parameters (x"» ; interact with each 1965~». They must understand how these component parameters system parameters, or characterized by a given set of other to produce a product these system and they must understand how something of the relationships hjx"'), of the needs, or understand some portion parameters fill a particular set of user new product successfully a Finally, in order to design a relationships gjY"). differput relative economic value of designer must be able to roughly assess the source "invention," or of thp original 2 We deliberately avoid a discussion of are likely to come from is the new idea. While the problem of where inventions of the implementation of that important and interesting, we believe that a study of at least as important, and new idea, or its translation into a saleable product is processing task that an insight into the nature of the information provides more successful. must undertake in order to be commercially organization 15 sompthing ahnttt thp nistortiprs' clijstprs of user nperls: that is thpy must knnw fnnrtinns V^(7.). utility infor-mat ion about tViP custom.pr and th^ tpchnology Tf a desisn'^r had pprfpct npw product would hp a and was faced with no significant uncprtainty , dpsigning a which trivial exercisp. Thp dpsigner would havp completp knowlpdgp of thp way in x) could bp genpratpd and of thp functions diffprpnt spts of componpnt pararnpters < and V^ = hfZt and could dpvplop an "ontiirial" product. Howpver hni^^^K gn'^Y*'* abilities with only lirnitpd pxppriencp dPsigner of limited infov-mation procpssing will bp in-^omplptp, and will is unlikply to hp in this position. Thpir knowledge the valup of bp function of thp recent history of innovation and of marginal pvolutinn information at different stagps in the product's diffprent typps of 19f.9; (Simon. Npwell and Simon, 1972K and the As a technology pvoIvps both the designer's stock of knowledge new knowledge change. The routines or procedures that they use for acquiring the types design always requires the development of new knowledge, but process of dramatically with the natiire of the of new knowledge that are required differ widespread innovation. During periods of radical innovation there is widely different experimentation in product design. Products are characterised by parameters and user needs. For any single product, sets of components, system parameters and user knowlpdgp of thp rplationships bptwppn componpnts, system - have very needs to bp tacit and incomplptp a designpr is likely to is likely since every new product little knowlpdge of the relationships h^Jx*"* and g„(Y*'>, customprs place embodies a widely different set. Knowledge about the valup? that cost effective way of on diffprpnt configurations of user needs, and of the most thought be The process of radical innovation can be meeting them will also scarce. the "space" in order to of as a process of exploring the market and technology options. The better understand both the customer nepds and thp technological arp most processing capabilities of thp dpsignpr limited information gathering and . oripiitated towards Iparning about new tPchnolosiP!^ '^nrl new rippds , and effpctivply thp rp] hptwppn thpm in a rplativply pxpprimpntal or towards pxploring at ionships about a tacit way. Tn tpnns of onr fraripwork, thp dpsignpY dpvplops knnwlpdgp (\') SPts fXI and ( Z) and thp inf orrnat inn largp nnmbpr of possiblp ir.pmbers of the , , are orientated processing rontinps that they dpvplop to search for new knowledge th^ about potential new elements of the set. Thej r knowledge of towards learning likely to be tacit and universp of possible relationships betwepn thpm is incomplete types of knowledge that incremental innovation changes the The transition to competitive advantage during periods of are most useful to the designer. While products innovation is gained by the introduction of quite different radical technologies, during periods of incorporating quite different component exploitation of a incremental innovation it is gained by the morp pffpctive stable set of set of components and user needs within the context of a limited detailed becomes both possible and valuable to dpvelop a more interactions. It between the existing components, system understanding of the relationships matures this knowledge is likely parameters and customer needs. As the technology likely "cheap" and the desigupr's time is to become widely diffused or relatively particular componpnts and detailed knowledge about the performance of to acquire teois of component/parametpr/customer nppd interactions. Tn about a limited set of detailed knowledge about particular our framework, the designer devplops vpry these elements remain stable, members of the sets ^X) , (Y) and ( Z) and, since relationships h,,,(x*'> and gJY"^. detailed knowledge about some known subset of knowledge that are tightly will strategies for acquiring new Moreover they develop technologies and on a limited set of focused on these particular cornponent critical interactions. particularly we begin to understand the implications of Given this context, can capabilities knowledge processing "generational" innovation for the knowledge and 17 ' 7."^ generational innovation the sets (X) , CYt .md reinain of a flpsigner. nnring obsolescence relatively stable, hiit the relationships between them chini^e. This of the "relational knowledge" of the designer may be just as significant as the with radical innovation. In sonp situations it may obsolescence that is associated to identify. be even more significant because it is more difficult - component The advent of radical innovation of completely different of user needs - is usually unmistakable, and technologies or quite different types knowlpdge a designer runs little risk of assuming that their historical is is more subtle, and it may be much more relevant. But generational innovation become obsolete. The spt of difficult to notice that ones relational knowledge has much of the components, system parameters and user needs remains stable, and the remains relevant. There are fewer signals to alert designer's knowledge designer miay attempt to meet the designer to the nature of the innovation, and a his or her threat of a competitive product with a design that is based upon routines that they have historically derived relational knowledge and the generation gather new knowledge that were appropriate to the previous developed to the risk of producing a significantly of the technology. As a result they run generational innovation. inferior product and "failing" in the face of that is perfoiTned A characterization of the information processing task complex problem. But organization designing a new product is a much more within an single we have described in the case of a to the extent that the dynamics that of the evolution of boundedly rational individual also characterize the dynamics this processing capabilities of an organization, the knowledge and knowledge generational innovation may have framework gives us insight into the reasons that and competitive implications. such dramatic organizational 18 TTT Fmpirical Spctinn the dlscussinn wp now turn to a dP'^rriptinn of As a means tn dppppn our pqnipmpnt indn<^try. Wp have of thp spmiconfliirtor optical photolithographic history innovation it is adpqtiatply characterise technological suggested that in order to technology tn pxainine in some detail npcessarv to break open the "black box" of a customer needs, between its components, system parameters and the relationships spction a detailed analysis of semiconductor Consequently we present in this at the component level. We \isp photolithographic alignment technology focused ^ photolithography and characterize the history of innovation in this analysis to the industry's which is a source of insight into thpn explore the extent to it snggpstlve, but they are presented competitive history. Our empirical results are The of the explanatory power of our framework. here only as an Illustration be hypotheses remains to done. rigorous formulation and testing of appropriate solid state semiconductor Photolithographic aligners are used to manufacture intricate semiconductors requires the transfer of small devices. The production of semiconductor material such as silicon, patterns to the surface of a "wafer" of Figure ^21 schematically transfer Is known as "lithography." and this process of with a surface of the wafer Is coated Illustrates the lithography process. The to be transferrpd to the chemical or "resist." The pattern that is light sensitive as is and the mask is used to block light wafer surface is drawn onto a "mask" '3 of the technical and results of a much larger study Oiir anaiysis draws on the Included the (Henderson, 1988). This study competitive history of the industry for every managerial and sales histories construction of comprehensive technical, These the industry's history since 19f.5. product development project undertaken in individuals, including drew from field interviews with over one hundred histories from intensive study leading engineers and scientists, and product designers and scientific journals. An the trade press and the major of Internal firm documents, process to the use of an iterative validation important element of the work was the research written summaries of ensure its accuracy. At each stage of the and circulated to key individuals results and the preliminary conclusions were follow up interviews. confirmed through <'21: Figiirp Thp r,i thographir Prncpss 1 . Expose Rpslst Mask Resist Wafer 19 resist sn that onlv those portions of the resist definerl hy the falls ontn the , mask are exposed to light, or "exposerl." The light rhprniraliy transforpiS the the areas availahle as a resist so that it can he stripperi away, leaving nnexposeri as basis for further processing.'* The process rnay be repeatfd as many twenty times be during the inanuf actiirer of a semiconductor devii^e. and each layer rriiist located respect to the previous layer. precisely with the mask relative to th^ A photolithographic aligner is used to position expose Figure wafer, to hold the two in place during exposure and to the resist. generic optical photolithographic n> describes the principal components of a stable aligner. The core technologies of photolithographic alignment have remained developed in the middle sixties. Despite this since the technology was first been strikingly turbulent. stability, the competitive history of the industry has sequence of dominant firms have each in turn failed to maintain their position The the Table M) shows the sales histories of the leading firms. in industry. Kulicke and Soffa in llfiS. first commercially successful aligner was introduced by and held nearly 100% of the (very smalH market for They were extremely successful replaced them and held the next nine years, but by 1974 Cobilt and Rasper had market for contact aligners each. In 197A pprkin Elmer approximately half of the the industry. Further entered the market, and immediate became the largest firm in were the leading entry followed in the late 1970s, and by 1981 GCA and Canon while of this writing GCA has also lost its dominant position and players. As player Nikon is probably the largest firm in leading Canon remains an important today. edge photolithographic equipment "positive." Tf a negative resist is used 4 Resist may be either "negative" or processing. If a positive resist is the unexposed areas are stripped away during description of used the exposed areas are stripped away. A more complete (1987). semiconductor lithography is available in Watts and Einspruch ni: Figiirp Thp Principlp rompnnpnt> nf a "Gpneric" nntjral Phntnl i thngraphir Alignpr.' SnURfF, 20 h) Figure ( : Principle elements of photnl i thographir technnlngy. CnMPONF.NT.S r.tlSTnMFR SYSTF.M NEEDS PARAMETERS MTNTNfllM SOURCE FEATURE Wavelength SIZE -Uniformity Col lirriatlon Energy RESOLUTION MECHANICS Contact accuracy -Gap uniformity CRITICAT, Gap accuracy Stage accuracy DIMENSION Stability CONTROI, OPTICS THROUGHPUT Numerical aperture -Wavelength distortion Lens stability Lens ALIGNMENT RELIABILITY Wavelength —Algorithms Accuracy Targets FOOTPRINT CONTROLS SETUP TIME Structure -Distribution MAXTMITM WAFER SIZE ?1 on wafer. nthpr things equal, si7e semicnndnrtnr a All the smaller the of a device, the faster it <~an rnn and the cheaper it is to mannf acture , and since device size heen hy historically has limited the miniirnim feature si;7e capahility of the lithographic process, users have demanded aligners with smaller and smaller mininnim feature capahility. Users are concerned ahont the size also eqnipTnent's throughput, yield, footprint, tTiaximurp wafer size capahility, reliability and f lexibi lity.5 An aligner is an extremely complex piece of eqnipnipnt, and can be those that characterized by many system parairieters . The m'ost critical are support its central function - its ability to accurately and consistently replicate an extremely small pattern on the wafer, or to support particular minirrium feature size capability. Three system parameters are particularly critical in this respect: "resolution," "critical line width control" and "overlay." Figure CS) defines these three parameters and illustrates their relationship to each other. An aligner's minimum feature size capability cannot exceed its resolution since its resolution is the size of the smallest optical image that it can transmit to the image cannot be accurately wafer. But superb resolution is useless if the positioned or reliably reproduced. If the aligner is not accurate that is if its - transfer is overlay characteristics are not very good or if the process of image not reliable that is if the aligner's critical dimension control is not adequate - then the aligner's minimum feature size capability will be less than its minimum resolution. used production has grown from less 5 The Tiiaximum size of the silicon wafers in eight used today, and than one inch in diameter in the sixties to the inch wafer;' users need lithographic equipment that can handle the size that they have chosen, The throughput, aligner all drive its effective cost. yield and footprint of an The faster the throughput and the higher the yield, or the more "good" wafers produced per hour, the cheaper the aligner is to operate. "Footprint" is a measure of the area that aligner requires on the semiconductor production floor. Since the customers prefer aligners to be semiconductor facilities are extremely expensive, as small possible. as Figni-p (S> CRTTTCAL DIMFNSTnN CONTROL Accuracy with which an image can he pnsit-innpd RF.Snr.UTION <^i7P nf the sniallest that can be imaee produced on thp surface of thp wafer. Resolution Actual irfigf OVERLAY posi tion Mpan distancp hetwepn Ideal imTge actual and ideal image position position. Source; Watts and Finspruch 987^ The basic concepts of alignment technology have remained stablp sincp it was developed, the three types of innovation first but the industry has seen all of that wp have identified: radical, incremental and generational, Commprcial production photolithography, which light is used has been dominated by optical in as the exposure source, but radical alternativps that make use of alternativp sources and quite different mask, alignmpnt and image transfer technologies have been explored since the seventies.** offpr customers bptter minimum early Thpy number of unsolved feature size capability, but to date both their cost and a technical problems have prevpntpd them from being widely used beyond research and development. We therpfore focus here on the optical systems that have dominated the industry, Tncrpmental innovation has bppn critical to optical photolithography's nf each component bepn significantly continuing succpss. The technology has ^ Radical alternatives to optical photolithography includp X-ray and Ton Ream aligners that x-rays ion-beams respectively as a source and Direct-write use and electrons to "write" on thf electron beam technology, that uses a beam of focused Wagner, Burggraaf, 1982^ wafer. (Chang et al., 1977; Brown, Venkatesan and 1981; 23 improved: for example modern sources are significantly more powerful -inrl more uniform, alignment systems are much more accurate lenses arp larger and suffer and less from distortion. The industry has also seen significant generational innovation. The key relationships between components and system parameters, and between system parameters and customer nepds have shifted dramatically four times over the course of the industry's history as the industry has shifted from the simple contact and proximity aligners to the more sophisticated optical systems that "scan" the mask relative to the wafer or that "step" it slowly across the wafer surface. One indication the presence innovation visible in figures of of generational is C?) (7'>, its and which show the historical perfonnance of each generation in terms of throughput, yield and minimum feature size. Each arrow reflects a particular technological trajectory, and summari;^es a history of incremental improvement within each generation. The movement between arrows from one technological trajectory to another - reflects innovation in the underlying generational technology. the We can develop more insight into the nature of generational innovation in industry by comparing the relationships between components and system parameters the contact and proximity that underlie the user need for minimum feature size in aligner. Contact and proximity aligners are relatively simple so that their feature size, the most technology can be easily described, and a focus on minimum essence of critical dimension of the aligner's perfonnance, allows us to grasp the the generational innovation without the need to present an exhaustive analysis of determinants of every system parameter. were photolithographic aligners to be used Contact aligners the first the commercially. They use the mask's "shadow" to transfer the mask pattern to the wafer held "in contact" with each other, and wafer surface. The mask and are surface. CFigure light shining through the gaps in the mask falls onto the wafer I f^^i Fisurp : rh?in£PS in th^ B;^1anrp Bptwppri Minimnni F^-itiirp Si7P inrl Yiplri Arrn;^s the Changing r.pnpvatinns of Photnl i thographir Fqnipripnt Minimum Featurp Size 3 (microns') Poor Fair Good Production Yipld^ This table is designed only to give a sense for some general Minimum feature size achieved and production yield vary great relationships. between applications. between customers and and considpred very 2. Prodiiction yields are difficult tn measure atp confidential. Typically a "poor" yield would be a yield of about 20?^ a "good' yield might on the order 60-70%, although these figures would vary with be of measured. the stage in the process at which the yield was Source: Field interviews, Internal firm records. (Henderson, 1989) ^ Figure (7): Changes in the Balance Between Mininnim Feature Si^e and Thrnughnnt Arross the rhanging Genpratinns of Phntn] i thcigraphir Fqnipment Minimum Feature Size 1 Cmicrons) mity 20 40 60 Throughput^ 1. This table is designed only to give a sense for some general relationships. size achieved and throughput vary great between Minimum feature customers and between applications. throughput of the largest wafer size that the 2. Throughput is defined as aligner designed wafers per hour. is to handle in ^Henderson, ISS'I) Source: Field interviews, Internal firm records. 24 (8) presents a schematic diagrarr. nf a contact alignpv.~> Tnntact alignprs arp sinnplp and quick to use but the need to bring thp mask and the wafpr into direct contact can damage the mask or contaminate the wafer. The first proximity aligner was 1973 these problems. proximity aligner the mask is introduced in to solve Tn a held a small distance away from ("in proximity to") the wafer surface. The separation of the rnask and the wafer means that they are less likely to be damaged during exposure, but since the mask and wafer are separated from each other, light coiuing through the mask "spreads out" before it reaches the resist, and the mask's shadow is less well defined that is in the case of a contact aligner. ^Figure it result users (9) presents a schematic diagram of a proximity aligner.) As a switching to proximity aligners were forced to trade off minimum feature size capability for increased yield. advance The introduction of the proximity aligner is clearly not a radical over the contact aligner. The conceptual change involved in going from one to the be aligners. other was minor, and most proximity aligners can also used as contact However, as figures (10) and (11) reveal, the relationships between component parameters, system parameters and minimum feature size are quite different for the introduced generational two technologies. The introduction of proximity alignment innovation into the industry. The minimum feature size capability of a contact aligner is limited by its between the overlay and critical dimension characteristics. As long as contac^t wafer "perfect" - absolutely flat and parallel across the wafer mask and the is yield image the mask is not distorted and the aligner's loss is minimal, the of its resolution. Since the minimum feature size capability is limited only by upon a gap to contact aligner does not rely upon either an optical system or only by the transmit the image of the mask to the wafer, resolution is limited been source wavelength, and until recently the resolution of contact aligners has equipment. Unfortunately in greater than that of any of the other generations of (^8">: Figiiie SrhenLitic Di-igr^m of a rnntact Alignei J L AIJGNMENT SYSTFM MECHANTCA[, SYSTEM MASK m mmwy, WAFFR CONTROr_. SYSTFM (^9): Figure Schematic Diagram of a Proximity Aligner SOURCE J L MECHANICAL SYSTEM ALIGNMENT SYSTEM (Includes gap setting mechanism) MASK WAFFR STAGE CONTROL SYSTFM Figure (10): THF mNTACT PRTNTF.R COMPONENTS SYSTFM CUSTOMER NEEDS PARAMETFRS SOURCE Wavelength -Unifonnity Col 1 imation Energy RESOLUTION MECHANICS Contact accuracy -Gap uniformity Gap accuracy CRITICAL MINIMUM Stage accuracy DIMENSION FEATURE Stabilltv CONTROL SIZE ALIGNMENT Wavelengtli —Algorithms Accuracy Targets Indicates those relationships that are critical and that limit performance of the aligner. the Source: Field interviews, Internal firm records. 'Henderson, 1989) Figiup ni): THE PRnXTMTTY PRTNTFR SYSTFM rnSTOMFR NFFDS cnMPnNFNTS PARAMETFRS SnilRCF, Wavelpngth -ITniformity Cnllimatlon Energy REsnr.uTinN MECHANICS Contact accuracy -Gap uniformity — CRTTTCAL MINIMI TM Gap accuracy FEATURE Stage accuracy DIMENSION SIZE Stability CONTROL ALIGNMENT Wavelengtli —^Algorithms Accuracy Targets that are critical and that Indicates those relationships limit the performance of the aligner. Source: Field interviews. Interna] firm records. ("Henderson, T589) 7'- prndurtion spttings contact is far from perfpct, Tn practice both the mask and often nr wafer are liam-ispd distorted, and minimum featm'p si^e is limited by the overlay and critical dimension control characteristics of the aligner. These are a function of the precision of the contact mechanism and the accuracy of the alignment system. Tn contrast, the minimum feature size capability of a proximity printer is limited by its resolution and by its critical dimension control, and these are functions of quite different component parameters. Resolution is limited not by the wavelength of the source but by the size of the gap between the mask and the wafer. As a result although the minimum feature size capability of a contact aligner is rarely limited by its resolution, resolution is very much a binding constraint in a proximity aligner. The aligner's critical dimension control is limited by the uniformity of the gap between the mask and the wafer, since if the gap is wider in some places than in others the light will have a chance to "spread out further," producing image an of a different size at the wafer. The framework that we have developed thus allows us to make the nature of generational innovation precise. Proximity alignment appears to be a minor innovation. The components, system parameters and user needs met by the aligner remained unchanged. Yet the relationships between the components, system parameters and needs apparpntly small change in a user changed dramatically. An particular component technology had important repercussions in the way in which the entire system operated. We have used this framework to analyze the later stages of the industry's technical evolution CHendprson, iq^qt. Table (2) summarizes our results and the outlines the key changes in technical rpl at ionships th^t nnderl ly gpuprational innovations that have riarked the industry's history. Tn pach case, the existing components introduction of a single new component or of small changes in . Tahle (2): A Summary of Relationa] Change Tn Photolithographic Alignment Terhnology fTenerational Major changes in: Innovation: Technology Critical Customer Needs Relationships PROXIMITY Mask and wafer MFS Mini mum feature Yields much higher, separated during siz»^^ a function of but MFS now much exposure. gap si^e, accuracy. greater constraint. SCANNING Image of mask Rpsolntion driven by MFS constraint PROJECTION projected onto numerical aperture, relaxed, but wafer by scanning wavelength. Overlay, scanner is slower. reflective optics, - a function of lens distortion - is critical constraint for MFS. STEP ^ Image of mask Throughput major Some customers have REPEAT (]) projected through constraint - function large increases in size theoretical refractive lens. of lens field and yield, Image "stepped" source energy. limits to MFS rise, across wafer. But throughput falls significantly. STEP ^ Introduction of Overlay and resolution Dramatic improvement REPEAT C2^ "site by site" constrain MFS. in reliability, lens alignment, larger Resolution driven by field size improves 5x lenses. Numerical aperture, throughput. MFS wavelength. again major reliability constraint Overlay and lens driven by distortion, interactions between lens and mechanical svstem. ("Henderson, T^RR) Source: Field Interviews. Internal Firm Records, 2f. chanp;p(i thp cnrp rel -^tinnshiips thu dptPr-'M iiprl fhp fnnrt inni ng nf thp allgnpr and tradpnffs customer iipprls. stnirtiirpfl upw across thp GpriPratinnal Innovation and Drgani national Capability. of thp Insights into thp tprhnical history indnstry arp intriguing. But our framework also implies that generational innovation shoulrl havp irnportant competitive consequpncps . Wp suggested that genprational innovation ohpolptes thp difficult for them to "relational knowledgp" of thp product designer, making it t it understand the nature of genprational innovation and ^n rpspond n appropriately. Our empirical results suggest that this insight is critical to an of the history of photolithographic understanding competitive the alignment industry, A full analysis of the indtistry's history should obviously considpr more than simply changes in the technology. But in this type of industry, in which thp performance of the product is critical to its acceptance in the market, and in which firms compete primarily on the basis of technical excellpncp, a careful implications of technical change can rpveal patterns that study of thp competitive may be important in a wide range of other industries. HI of sales by generation of equipment for the leading Table presents share been able to pxtend its firms. None of the established firms in the industry has position into the next generation of equipment, despite its experience with the and its ownership of an extensive installed base. technology 1 C3): Table Share nf •^ales hy gener-Ttinn fn» the leading nptiral '^ photolithographic aligniTent f^quipiient 'ictnr'M'^ mannf (%) Contact Proximity "scanners Step and Step .md Repeat ( ) Repeat C Gobi It 28 (ipspitp thp fpicp that thp idpntity of the pstablishprl firm changps in pvpry i nst aiipp,^ Thus traditional pcononic pxpl anat jons of t h*^ fai hue of establisheri firms do not sppp' to hold hpre. Fvplanat ions that )ely on "bureaucratic inertia," or on the difficulties that many encountpr to rpspond any t^ype organizations in attemptine of change are similarlv susppct . sincp pstablishpd f i rm« pxpliritly idpntified thp competitive threat represented by generational innovation, attached great importance to meeting used design teams that weie on average ("omposed of it, and fewer Established firms failed to than twenty five people ("Henderson, l'^88>. experience that they maintain their position into the npxt gpupration hpcause the had gained with incremental innovation in the previous gejieratjon specifically they - very difficult for the relational knowledge that had accumulated made it them to understand the nature of the generational innovations with which they were to faced or shape an appropriate response. to its We illustrate this here by a description of Kasper's failure maintain position in proximity alignment. This failure is particularly intprpsting since firm was relatively and "organic" in structurp and since thp the small appears be similar to the introduction introduction of proximity alignment to so of an incremental innovation. It illustrates graphically the ways in which subtle system parameters and user needs changes in the i nterrplationships of components, can have very significant i^ompetitive implications. T^i^8 and by 1971 was a small but Kasper Instruments was founded in , of the m.arket for contact ^lign'^rs. profitablp firm supplying approximately half with proxir^ity In 1973 Kasper introduced the first contact aligner to be pquipppd Although nearly half of all the aligners that the firTn sold from T17A capability. of the determinants of technical success in the 8 A statistical analysis that established firms were significantly less likely industry's history suggests chance. (^Henderson, to be technically successful than would be predicted by random, 1989) y onwards h.-\d this capability. Kasper alignprs wpip nnlv rarely nspd in prnximity mode, and sales declined steadily until the company left the indnstry in 1*181. The widespread use of proximity aligners did not occur until the introduction and late general adoption of Canon's proximity aligner in the 197ns. Rasper's failure is initially pu77ling given its established position in the market and its depth of experience in photolithography. There were approvimat el five key mechanical and electronic engineers Kasper during the early 197ns. at Several of them were highly skilled and i'T'aginat ive designers, and the group designed a steady stream of contact aligners, each incorporating significant incremental improvements. From 19(^8 to 1973 the minimum feature siT^e capability nf the contact aligner improved from ten to five microns. But Kasper''^ very success in designing contact aligners was a i^iajor contributor their to a that could perform to inability design proximity aligner as successfully as Canon's. Canon's aligner was superficially very similar to Rasper's. Tt incorporated the same components and performed the same functions, but it performed them much more effectively since it incorpoiafed a much more sophisticated understanding of the technical interrelationships that are fundamental to successful proximity alignment. Rasper failed to develop the particular specialist knowledge that would have enabled their' to match Canon's design, but, even more importantly, their experien<:'p with contact aligners left them without routines that would have enabled them to the knowledge gathering they understand the need to acquire it. The relational knowledge that had developed through their experience with contact aligners had the effect of fooising away from the new problems whose solution was critical to their attention the design of a successful proximity aligner. nndified contact aligner, and Kasper conceived of the proximity aligner as a product as the previous it was managed as a rovitlne extension to the line, just setting incremental improvements to the contact aligner had been. The gap 30 tn tn rtiprhanisni th.it v^'^ nspfl in thp rnnt?ict alignpv .^licn tho Pi-isk inri wafpr p^ch nthpr w;is slightly irmrlifipii ind t hp aligiiPV w.is offiTf^'l nii thp markpt. Tn dnins this. Kasppr maclp thp implicit assnmptinn that thp tprhnnlocv that had sustained support pToxinity alignrnpnt^. rontart printing could hp incrprnpntal 1 y pxt pnd^d to But, as thp analysis that wp prpspntpd ahovp rnadp c]par, pioxirrity alignment alignnpnt tpchnology. In particular, the represents a generational shift in intprrplationships between the performance of the gap setting mechanism and the feature si:^e the quite different in a proximity minimum capability of aligopr are and contact aligner. In a contact aligner, the gap SPtting mpchanism is uspd only dui ing alignment. Errors in its setting can be corrected manually, and its accuracv has little influence feature B\it in a proximity aligner over minimum si?^e capability. the quality of the transmitted image is critically dependent \ipon the gap's capability is a dirpct function accuracy and stability, and minimurri feature si?e of its size and accuracy. As rpsult thp succpssful design of a proximity aligner requires both the acquisition of some new knowledge how to build an adequate gap setting mechanism - and an understanding of some new interactions between the alignpr - in particular an component performance and the system parampters of the rpsolution understanding that minimium feature size capability is limited by and critical control thp alignpr, and that in turn resolution and dimpnsion of upon the accuracy and critical dimension control are critically dependent uniformity of the gap betwpen the mask and the wafer. mpchanism is not a trivial Thp successful dpsign of an adequatp gap SPtting betwepn thp mask task. Tn a contact alignpr thp dpsignpr can rply on the contact to the and thp wafpr to pnsurp that the mask is located accurately with respect wafer. Tn aligner, the mask must be located by "dead reckoning" at a proximity point above wafpr, and the designer must rely on the accuracy of some in space the and the wafer is the mpchanical mpchanism to pnsure that thp gap bptween thp mask 11 precise and consistent arrnss the mask surface. This call'^ for significant expertise iri the design of high precision Tiechanical mechanisms, and in hetween the the particular, for a deep understanding of the interactions design of gap setting mechanism and the other elements of the aligner. Kasper the lacked this knowledge, or, in the terTn<^ of onr framework, lacked ability to control the component parameter "gap si7e" with sufficient accuracy. Canon, on the other hand, were ahle to hring to bear a repertoire of skills and knowledge aboiit high precision mer-hanica 1 ass-^mblies deriv^ed from their pxperience related needed tn develop this in industries. The design group at Kasper had not type of knowledge since it was not critical to the performance of the early contact aligners. But this lack of a specific piece of t'echnical knowledge js not the whole story. More critically. continued to make use the problem solving and Kasper of they experience information gathering strategies that had developed during their develop with the contact aligner, and as a result failed tn understand the need to this critical body of knowledge. Tanon invested heavily in building a deep setting understanding of the precise relationships between errors in the gap mechanism, the stability and uniformity of the source, and the performance of the dimension alignment system in the determination of the resolution and critical control of the aligner. This them make informed tradeoffs in the enabled to overall design the to understand, for example, where design effort of system, and They assumed that the could be most fruitfully spent. Kasper did not do this. introduction of a gap between the mask and the wafer during alignment was a minor, incremental change and so put little effort into understanding the new between setting mpchanism and re'^olutlon and interactions that it created the gap critical dimension control. Thus they failed to under'^tand the enormous impact could have on the perfor-^iance of that improvements in its accuracy and precision , thp alisiipr, and the iiroximity alignpi that thpv r|p>;i£npi1 •iuffprpd frnti. an arciirate mechanisi-n untPliahlp and insufficiently gap «ettine to the K.aspey 's failure understand ohsnlesrenre nf their relational knowledge is demonstrated graphirally hy two incidents. The first is the firii^'s interpretation of early complaints about the accuracy of its gap setting the fnask niid the wafer can hp mechanism. In proximity alignment misalignment of caused both hv inaccuracies or instability in the gap setting mechanism and by introduced processing, at-tributpd of the pvr)blems distortions during Kasppr many that usprs of its proximity equipment were eyppripncing to processing error, since they "knew" f rniTi their experience with fhe contact aligner that their gap setting mechanism was adequate to task. As a result they devoted vpry little time to the retrospect improving its performance. In this may sppt' likp a wanton misusp of information, but it represented no more than a continued reliance on an served well The cppond illustration information filter that had them historically. is provided by their response to Canon's initial introduction. The ranon aligner by be a "copy a K.asper was evaluated a team at Kasper and pronounced to of machine." Kasper evaluated it against the criteria that they used for evaluating their own aligners - criteria that had been developed d"ring their expprience with features that made it a significant advance, contact aligners. Thp tpchnical were not observed because they were not particularly the redesigned gap mechanism, considered important. the core design Kasper 's engineers were not incompetent. They assumed that contact aligners could a1«o serve as concepts that underlay the design of their the the basis for a proximity aligner, and that the relational knowledge and knowledge capabilities that they had accumulated dtiring their processing could be transfprrpd to proximity aliguprc. experience with contact aligners from more than this failure in design. Kasper's commercial failure stemriied system of sufficient The company had problems designing an automiatic alignment , . PicrwYHry, and in a hi^h vn1nin(=> "nnnf artnri nj: farility, Thoy il^n :;nffpipr1 >Tianaf,in° ^pvt^ral rapiH of top rr'-^n ^^^''''''^ti^ iliirii!^ ' bp 1 >tp rii'iptpin thrnugh cbanjies thf nh'^nl psrenrp of yplatinnal Vnowlpd^p brnn^ht ihnnt by tbp SPVPnties. But f:\rt-nr i nl" rndurt inn of ^pppratinnal innnv'atinn was a 'ni''i'"a1 in Ibpic dpc-line. A similar study nf thp failnrp of tbp ntbpv dnrrinant fivis in tbp industry suggests tbat a rplianrp ibmit fbp tpcbnnlnpy dpri^'pd frnn thp on Yelational knowlpd^p prpvinus ppnp)'atinn was a sipnifirant factor in pvplaining thp failnrp of all nf tbp pstahlisbpd who to >nppt tbp ^hrpat of npw pntry ''Hpndprson, firms wpfp iinablp 1989) TV SuTTiTnary and Conclusions suggpstpd that the charactpri/at ion of innovation as pitbpr "radical" We bavp or "incrpTTipntal" is incornplptp, and nnist be snppl emented by the concppt of the perfonnance of "genprational" innovation. While incrempntal innovation changps an existing set of components, and radical innovation changes the elements of that gpnerational innovation cbangps set and the rel ationsships betwppn them, performancp and relationships, but Ipaves thp plements largely unchanged. Generational inno\'a^ion 1*= sometirr'ps tjiggprpd by cbangPS in a particular but fundamental reconfiguring of the component technology, its essence is a customer technological system around an essentially stable set of components and Tt qualitatively different from both increrrental and radical innovation needs. is i cat ions and has equally important competitive and organizational ir<pl individual of We used the mode] of a product designer as a boundedly rational limited processing capabilities to highlight thp diffcrpnt information information types innovation pose fov the product procpssing tasks that thp diffprpnt of knowledge development process, and discussed the types of knowlpdgp and types innovation acquisition routines that a designer faced with the different of tn that the degree that this is likely devplop in consequence. Wp suggested to 34 nmrlpl i> indif^ativp of fnrrp>? at work itT^iiic ors^ni zii"inn-^ thf^-n •^.pri'^f- it innal '"^v .'nt nn^-yppc tpil 'Mipippt vf> cnn«"qn<^nr^^x ^iiiri^ it ch^inse h^vi^ in.pmt ^nd i t i nhc;n1ptpp thp rplatinn?i1 knnwlpr!p»^ nf thp nrgani 7at inn . This issue is an important avpnnp for fntiirp ypsparrh. nf t}iP ^nH Wp illnstratpr) on, ftaiiPwork thrmipb a rlp-irri pt inn tprhinral rompptitivp px'olntinn of thp •iprni conr1urto> phnt nl i thop vaphi f' ilipmnpnt i nrlii'^t ry . Wp showpd that the i nt » oHiict pinxi'^iii t ali^nipnt^ an apparpn^ly inn of y into thp "i ncrpffipntal " inno\fat inn , in fact introdiicpd si£nificaiit '"hangp thp rplat ionships hptwppn cnrpponpnts , nid that Kasppr's rnntinopH tp1ian<"p on knnwlpdpp and infoi-Tiation procp'^'^ing capahilitips that thpy h-id dpvploppd thp through thpir pxpprience with contact alisriprs 'ipvprply handicappprl hp firri's ahility to takp adv^antagp of proxirnity ali^naiPnt. of For pxample, wp need Thi-^ resparch oppns up a ntimbpr important qupstions. to to develop metrics for generational innovation and explore its interactions with »^hp othpr economic and organizational forcp'^ that -ihape firms and indnstrie?. R\it we believe that the concept of genprational innovation ha<^ potpntially grpat m^ipptitivp powpr and will pro\'p uspfiil in thp 'Atvidy of thp organizational and implications of technical change. : . . . . REFFRFNCFS and I. Freeman D.Schendel Hannan, M. Cooper, A.C, and Williain Abemathy, "Strurtural Inertia "Stiqtpgic Response tn 1984 197A Productivity 1978 Thp n.-g Change. uiizational Technological Threat-. Baltimorp: Johns Dileniri-'a. 19: hl-f,9 Amerie^n Sociological Business Hori7ons. University Prpss. Hopkins 4^1: 149-1^4. Rpvievj, D. Beard. and Kim Dess.G.G. and Abemathy, William, Henderson, Rebecca M. "Dj-prisinns of 198/i Clark 1^88 "The F^i lure of Task thp Organisational nSS "Innovation: Mappin? ." Pst -^^l i^l^prl Firm.s in Fnvironmenfs Winds of Creative Technical the Face of Administrative Science Destruction." thf 5-^-73. Change: A Stvidy of 29, lA: 3-22. Quarterly, Research Policy, Semiconductor Photolithographic .T.F.mitton, newar, R.n. and Alexander, C. Industry." Radical anr Alignment 198fv "The Adoption of iqA.'i N'otes on the Synthesis of Ph.D. Unpublished Incremental Innovations: an Carobridse: Harvard Form. dissertation, Analyses." Press. Empirical University Massachusetts Management Science, Tnstitutp of 32: U22-1432. Arrow, K. Technology. Organisation. 197A The limits of Norton. Dosi , G. New York: Henderson, Rebecca, M. Paradigms and 1^82 "Technological 1989 "Heterogeneous Firm Trajectories: Technological Brown, W.L., T.Venkatesan and C^pabil ity and the A Suggested Interpretation A.Wagner Established Failure of of the Determinants and Lithography." T^Sl "Ion Beam Paper, Firms." Working of Technical Change. Directions Solid State Technology, Massachusetts -icy, 11: 147-1<^2. Research Pol August of Institute Technology. and W. P. Bridges Fttlie.J.F., Biirggraaf ,P. and R.D.O'Keefe 1983 "X-Ray Lithography: and J., Sawers, D. Strategy and Jewkes, 1984 "Organisational Optica] 's Heir." Stillerman for R. Structural Differences Semiconductor 19AC! The Sources of Incremental Radical versus International, September. Invention M?inagement Innovation." New Macmillan. f.82-f>95 York: Science, 30: G. Stalker Burns, T. and iqfiA of Innovation. The Management and Kimeberly,J.R. F.ttlie.J.F. London: Tavistock. J.Fvanisko Policy imong M. 1983 "Organisational 1981 "Organisational the Food Suppliers to Chang, T.S. et al. Innovation: The Processing Sector." 1977 "Flectron-beam Lithography Influence of Individua Academy of Management Finer fine." draws a and Organisational 2*^: 27-44. Tournal , Electronics, May. factors on Contextual of hospital adoption Freeman, C. Kim B. Clark, technological and Industrial 1982 The Economics of interaction of design 1985 "The administrative Innovation. Cartridge, and market hierarchies innovations." Press. technological MIT concepts in of Management Academv Policv, evolution." Research 24: 689-713. Journal, and D.M.Newbery Gilbert, R..I. 14: 235-251. thf Patenting and 1982 "Preemptive Kuhn, Thomas of Monopoly." Persistence Clark, Kim B. Structure of 1970 The Economic Review, American 1987 "Managing Technology in Scientific 72: 514-95. International Competition." Chicago: Revolutions. Ced.^i Spence and Hazard, In Chicago 'University of International Hage ,1. Press. Organisation 1980 Theories of Competitiveness. New York : Wi 1 ey Camibridge: Bal linger. Ul9 \ . . , . I.W.I.nrsch Sahal I.nwrencp.P.R. anrt l^A? n>-2ani zai" ion aiul Fnvivnt ^rhnolnoicil GnirU niinnjc;: Hornpwnori and Innovation Avpnijp'^." Rpcoirrh Pnlir-y, 1 'i : fy]-^'^. Mansfield, F.dwin T.S. T^AR TndnF^trial Rpsf^arrh and Saviotti,P.P and Mot calf p "\ Tt^'^hniral Innovation. iq!^4 Thporptical Approach to Npw Yo>k: Norton. t-hp Hon^trvictioti of rmtnut indicator- Tpchnological Marples, n.I.. Rpspa-.rh Policy, 11: I'.I-ISI . 116] "Thp Dpcision<= of Fnginperin^ Dosi^n." Simon, Herbert TF.FE Transaction"^ on l'^'^'^ Thp Scipncp? of the Artificial Enginpprin? Managempnt rarpbrid^p: Ma<sa'-h"c-or t-; Timp. Institntp of Tpchnoloov "rpsc Morh,M. and E.V. Morse Tiishman.M.L. and P.Anderson lOSf, 1977 "Si7p^ centralization ind "Tcrhnological nisconti nm t IPS organizational adoption of and Organisational Eiivi roiinipnt' Administrative Soipnce innovations," American AlH-'ti^s Sociolosical Review, njiartprly, 11 : , Octoher. 7K-.-72S. Watts, R.K. and N.G.Einsprnch, (pd.) Nelson, Richard, and Sidney 1987 Lithography for \T,ST. Acade^iic Winter New York: Prps 1982 An Evolutionary Theory of Econoniic Changp. Carnbridge: Harvard Hnivprsitv Prpss, Newell, A., and Herbert Simon 1972 Human Problem Solving. riiffs: Englewood Prpntice Hall. Reinganaum, J.F 1983 "Uncertain Innovation and the persistence of Monopoly." American Economic Review, 73: 7A1-A8. Rosenberg, Nathaniel 197^1 Technology. Perspectives on Cambridgp: Cambridge Univprsity Prpss Rosenberg, Nathaniel 1982 Tnsidp thp Black Rox: Technology and Fconorirs. Cambridge: Cambridge University Press. Rothwell Roy 198<^ "The Role of Small Firms in the Emergence of Npw Tpchnologipj FroiT! Freem'am, C. Ced.) Design, Innovation and Long Cycles in Economic Development, london: Francis Pinter. Date Due ' srsaj NOV 6^9^ I3/i:, *EP- fee 13 199) 199(jt aPR. G 19gB ^6 OCT IS^u PEB. ^^i mo 1 i999 /WAV A7iB?r MAY 2^199 pfB^f.^! FEB N. 1 6 1908 199f' MAR \m* 6 19S9 APR:2 4r9^MAn2 MIT IIBRSRIES QUI Q05b7M3fl M 3 TOflO

Journal

Administrative Science QuarterlyUnpaywall

Published: Mar 1, 1990

There are no references for this article.