Architectural Innovation: The Reconfiguration of Existing Product Technologies and the Failure of Established Firms Author(s): Rebecca M. Henderson and Kim B. Clark Source: Administrative Science Quarterly, Vol. 35, No. 1, Special Issue: Technology, Organizations, and Innovation (Mar., 1990), pp. 9-30 Published by: Johnson Graduate School of Management, Cornell University Stable URL: http://www.jstor.org/stable/2393549 Accessed: 24/08/2008 13:09 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=cjohn. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission.
JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that promotes the discovery and use of these resources. For more information about JSTOR, please contact
[email protected].
http://www.jstor.org
ArchitecturalInnovation: The Reconfigurationof ExistingProductTechnologies and the Failure of EstablishedFirms Rebecca M. Henderson Massachusetts Institute of Technology
Kim B. Clark HarvardUniversity
? 1990 by CornellUniversity. .00. 0001-8392/90/3501-0009/$1 This researchwas supportedby the Division of Research,HarvardBusiness School.Theirsupportis gratefullyacknowledged.We would liketo thankDataquest andVLSIResearchIncfor generous permissionto use theirpublisheddata,the staffs at Canon,GCA, Nikon,PerkinElmerand Ultratech,and all those individualsinvolvedwith photolithographicalignmenttechnologywho gave so generouslyof theirtime. We would also liketo thankthe editorsof this journal and three anonymousreviewerswho gave us manyhelpfulcomments.Any errorsor omissions remainentirelyour responsibility.
This paper demonstrates that the traditional categorization of innovation as either incremental or radical is incomplete and potentially misleading and does not account for the sometimes disastrous effects on industry incumbents of seemingly minor improvements in technological products. We examine such innovations more closely and, distinguishing between the components of a product and the ways they are integrated into the system that is the product "architecture,"define them as innovations that change the architecture of a product without changing its components. We show that architectural innovations destroy the usefulness of the architectural knowledge of established firms, and that since architectural knowledge tends to become embedded in the structure and information-processing procedures of established organizations, this destruction is difficult for firms to recognize and hard to correct. Architecturalinnovation therefore presents established organizations with subtle challenges that may have significant competitive implications. We illustrate the concept's explanatory force through an empirical study of the semiconductor photolithographic alignment equipment industry, which has experienced a number of architectural innovations.' The distinctionbetween refiningand improvingan existing design and introducinga new concept that departs in a significantway from past practiceis one of the centralnotions in the existing literatureon technical innovation(Mansfield, 1968; Moch and Morse, 1977; Freeman,1982). Incremental innovationintroducesrelativelyminorchanges to the existing product,exploits the potentialof the established design, and often reinforcesthe dominanceof established firms (Nelson and Winter,1982; Ettlie,Bridges,and O'Keefe, 1984; Dewar and Dutton, 1986; Tushmanand Anderson, 1986). Althoughit draws from no dramaticallynew science, it often calls for considerableskilland ingenuityand, over time, has very significanteconomic consequences (Hollander,1965). Radical innovation,in contrast, is based on a differentset of engineeringand scientific principlesand often opens up whole new marketsand potentialapplications(Dess and Beard, 1984; Ettlie,Bridges,and O'Keefe, 1984; Dewar and Dutton, 1986). Radicalinnovationoften creates great difficultiesfor established firms (Cooperand Schendel, 1976; Daft, 1982; Rothwell,1986; Tushmanand Anderson, 1986) and can be the basis for the successful entry of new firms or even the redefinitionof an industry. Radicaland incrementalinnovationshave such differentcompetitive consequences because they requirequite different organizationalcapabilities.Organizational capabilitiesare difficult to create and costly to adjust (Nelson and Winter,1982; Hannanand Freeman, 1984). Incrementalinnovationreinforces the capabilitiesof established organizations,while radical innovationforces them to ask a new set of questions, to draw on new technicaland commercialskills, and to employ new problem-solvingapproaches (Burnsand Stalker,1966; Hage, 1980; Ettlie,Bridges,and O'Keefe, 1984; Tushmanand Anderson, 1986). 9/AdministrativeScience Quarterly,35 (1990): 9-30
The distinctionbetween radicaland incrementalinnovation has producedimportantinsights, but it is fundamentallyincomplete. There is growingevidence that there are numerous technical innovationsthat involveapparentlymodest changes to the existing technology but that have quite dramaticcompetitive consequences (Clark,1987). The case of Xeroxand small copiers and the case of RCAand the Americanradioreceiver marketare two examples. Xerox,the pioneer of plain-papercopiers, was confrontedin the mid-1970s with competitorsofferingcopiers that were much smaller and more reliablethan the traditionalproduct. The new productsrequiredlittle new scientific or engineering knowledge, but despite the fact that Xeroxhad inventedthe core technologies and had enormous experience in the industry,it took the companyalmost eight years of missteps and false starts to introducea competitive productinto the market.Inthat time Xeroxlost halfof its marketshare and suffered serious financialproblems (Clark,1987). In the mid-1950s engineers at RCA'scorporateresearchand development center developed a prototypeof a portable, transistorizedradioreceiver.The new productused technology in which RCAwas accomplished (transistors,radiocircuits, speakers, tuningdevices), but RCAsaw little reason to pursue such an apparentlyinferiortechnology. Incontrast, Sony, a small, relativelynew company, used the small transistorized radioto gain entry into the U.S. market.Even after Sony's success was apparent,RCAremaineda follower in the marketas Sony introducedsuccessive models with improvedsound qualityand FM capability.The ironyof the situation was not lost on the R&Dengineers: for many years Sony's radioswere producedwith technology licensed from RCA,yet RCAhad great difficultymatchingSony's productin the marketplace(Clark,1987). Existingmodels that rely on the simple distinctionbetween radicaland incrementalinnovationprovidelittle insight into the reasons why such apparentlyminoror straightforwardinnovationsshould have such consequences. Inthis paper,we develop and applya model that grew out of research in the automotive, machine tool, and ceramics industriesthat helps to explainhow minorinnovationscan have great competitive consequences.
I
Inearlierdraftsof this paperwe referred to this type of innovationas "generational."We are indebtedto ProfessorMichaelTushmanfor his suggestion of the term architectural.
CONCEPTUAL FRAMEWORK Component and ArchitecturalKnowledge Inthis paper,we focus on the problemof productdevelopment, takingas the unit of analysis a manufacturedproduct sold to an end user and designed, engineered, and manufactured by a single product-developmentorganization.We define innovationsthat change the way in which the components of a productare linkedtogether, while leaving the core design concepts (andthus the basic knowledge underlyingthe components) untouched,as "architectural"innovation.1This is the kindof innovationthat confrontedXerox and RCA.It destroys the usefulness of a firm'sarchitectural knowledge but preserves the usefulness of its knowledge about the product'scomponents. 1O/ASQ,March 1990
Architectural Innovation
2
We are indebtedto one of the anonymous ASQ reviewersforthe suggestion thatwe use this matrix.
This distinctionbetween the productas a whole-the system-and the productin its parts-the componentshas a long historyin the design literature(Marples,1961; Alexander,1964). Forexample, a room fan's majorcomponents includethe blade, the motorthat drives it, the blade guard,the controlsystem, and the mechanicalhousing.The overallarchitectureof the productlays out how the components will work together. Takentogether, a fan's architecture and its components create a system for movingairin a room. A component is defined here as a physicallydistinctportionof the productthat embodies a core design concept (Clark, 1985) and performsa well-definedfunction. Inthe fan, a particularmotor is a component of the design that delivers power to turnthe fan. There are several design concepts one could use to deliverpower. The choice of one of them-the decision to use an electric motor, for example, establishes a core concept of the design. The actual component-the electric motor-is then a physicalimplementationof this design concept. The distinctionbetween the productas a system and the productas a set of components underscoresthe idea that successful productdevelopment requirestwo types of knowledge. First,it requirescomponent knowledge, or knowledge about each of the core design concepts and the way in which they are implementedin a particularcomponent. Second, it requiresarchitecturalknowledge or knowledge about the ways in which the components are integrated and linkedtogether into a coherent whole. The distinctionbetween architecturaland component knowledge, or between the components themselves and the links between them, is a source of insight into the ways in which innovationsdiffer from each other. Types of Technological Change The notionthat there are differentkindsof innovation,with differentcompetitiveeffects, has been an importanttheme in the literatureon technologicalinnovationsince Schumpeter (1942). FollowingSchumpeter's emphasis on creative destruction,the literaturehas characterizeddifferentkindsof innovations in terms of their impacton the established capabilitiesof the firm.This idea is used in Figure1, which classifies innovationsalong two dimensions. The horizontal dimension captures an innovation'simpact on components, while the verticalcaptures its impacton the linkagesbetween components.2-Thereare, of course, other ways to characterize differentkindsof innovation.But given the focus here on innovationand the development of new products,the frameworkoutlinedin Figure1 is useful because it focuses on the impact of an innovationon the usefulness of the existing architecturaland component knowledge of the firm. Framedin this way, radicaland incrementalinnovationare extreme points along both dimensions. Radicalinnovationestablishes a new dominantdesign and, hence, a new set of core design concepts embodied in components that are linkedtogether in a new architecture.Incrementalinnovation refines and extends an established design. Improvement occurs in individualcomponents, but the underlyingcore design concepts, and the linksbetween them, remainthe same. 11/ASQ, March 1990
Figure 1. A framework for defining innovation. Core Concepts
Unchanged a) cJ a, C
Reinforced
Overturned
Incremental Innovation
Modular Innovation
Architectural Innovation
Radical Innovation
00 CL
3) V D0)
non
(D
~
.'
c) C
Changed Chne
Figure1 shows two furthertypes of innovation:innovation that changes only the core design concepts of a technology and innovationthat changes only the relationshipsbetween them. The formeris a modularinnovation,such as the replacement of analogwith digitaltelephones. To the degree that one can simply replace an analog dialingdevice with a digitalone, it is an innovationthat changes a core design concept without changingthe product'sarchitecture.Ourconcern, however, is with the last type of innovationshown in the matrix:innovationthat changes a product'sarchitecture but leaves the components, and the core design concepts that they embody, unchanged. The essence of an architecturalinnovationis the reconfiguration of an established system to linktogether existing components in a new way. This does not mean that the components themselves are untouchedby architecturalinnovation.Architecturalinnovationis often triggeredby a change in a component-perhaps size or some other subsidiaryparameterof its design-that creates new interactionsand new linkageswith other components in the established product. The importantpoint is that the core design concept behind each component-and the associated scientific and engineering knowledge- remainthe same. We can illustratethe applicationof this frameworkwith the example of the room airfan. If the established technology is that of large, electricallypowered fans, mounted in the ceiling,with the motor hiddenfrom view and insulatedto dampen the noise, improvementsin blade design or in the power of the motorwould be incrementalinnovations.A move to centralairconditioningwould be a radicalinnovation. New components associated with compressors, refrigerants, and theirassociated controlswould add whole new technical disciplinesand new interrelationships.Forthe makerof large, ceiling-mountedroom fans, however, the introductionof a portablefan would be an architecturalinnovation.While the primarycomponents would be largelythe same (e.g., blade, motor, controlsystem), the architectureof the productwould be quite different.Therewould be significantchanges in the 12/ASQ, March 1990
Architectural Innovation
interactionsbetween components. The smaller size and the co-locationof the motor and the blade in the room would focus attentionon new types of interactionbetween the motor size, the blade dimensions, and the amount of airthat the fan could circulate,while shrinkingthe size of the apparatuswould probablyintroducenew interactionsbetween the performanceof the blade and the weight of the housing. The distinctionsbetween radical,incremental,and architecturalinnovationsare matters of degree. The intentionhere is not to defend the boundariesof a particulardefinition,particularlysince there are several other dimensions on which it may be useful to define radicaland incrementalinnovation. The use of the term architecturalinnovationis designed to draw attentionto innovationsthat use many existing core design concepts in a new architectureand that therefore have a more significantimpacton the relationshipsbetween components than on the technologies of the components themselves. The matrixin Figure1 is designed to suggest that a given innovationmay be less radicalor more architectural,not to suggest that the world can be neatly dividedinto four quadrants. These distinctionsare importantbecause they give us insight into why established firms often have a surprisingdegree of difficultyin adaptingto architecturalinnovation.Incremental innovationtends to reinforcethe competitive positions of establishedfirms, since it buildson their core competencies (Abernathyand Clark,1985) or is "competence enhancing" (Tushmanand Anderson, 1986). Inthe terms of the framework developed here, it buildson the existing architectural and component knowledge of an organization.Incontrast, radicalinnovationcreates unmistakablechallenges for established firms, since it destroys the usefulness of their existing capabilities.In our terms, it destroys the usefulness of both architecturaland component knowledge (Cooperand Schendel, 1976; Daft, 1982; Tushmanand Anderson, 1986). Architecturalinnovationpresents established firms with a more subtle challenge. Much of what the firm knows is useful and needs to be appliedin the new product,but some of what it knows is not only not useful but may actually handicapthe firm. Recognizingwhat is useful and what is not, and acquiringand applyingnew knowledge when necessary, may be quite difficultfor an established firmbecause of the architecturalknowledge- is orway knowledge- particularly ganized and managed. The Evolution of Component and ArchitecturalKnowledge Two concepts are importantto understandingthe ways in which component and architecturalknowledge are managed inside an organization.The first is that of a dominantdesign. Work by Abernathyand Utterback(1978), Rosenberg (1982), Clark(1985), and Sahal (1986) and evidence from studies of several industriesshow that producttechnologies do not emerge fullydeveloped at the outset of theircommerciallives (Mansfield,1977). Technicalevolutionis usuallycharacterized by periods of great experimentationfollowed by the acceptance of a dominantdesign. The second concept is that organizationsbuildknowledge and capabilityaroundthe recurrent tasks that they perform(Cyertand March,1963; Nelson and 13/ASQ, March 1990
Winter,1982). Thus one cannot understandthe development of an organization'sinnovativecapabilityor of its knowledge without understandingthe way in which they are shaped by the organization'sexperience with an evolvingtechnology. The emergence of a new technology is usuallya periodof considerableconfusion. There is littleagreement about what the majorsubsystems of the productshould be or how they should be put together. There is a great deal of experimentation (Burnsand Stalker,1966; Clark,1985). Forexample, in the earlydays of the automobileindustry,cars were builtwith gasoline, electric, or steam engines, with steering wheels or tillers,and with wooden or metal bodies (Abernathy,1978). These periods of experimentationare broughtto an end by the emergence of a dominantdesign (Abernathyand Utterback, 1978; Sahal, 1986). A dominantdesign is characterized both by a set of core design concepts that correspondto the majorfunctions performedby the product(Marples,1961; Alexander,1964; Clark,1985) and that are embodied in components and by a productarchitecturethat defines the ways in which these components are integrated(Clark,1985; Sahal, 1986). It is equivalentto the general acceptance of a particularproductarchitectureand is characteristicof technicalevolutionin a very wide rangeof industries(Clark,1985). A dominantdesign often emerges in response to the opportunityto obtaineconomies of scale or to take advantageof externalities(David,1985; Arthur,1988). Forexample, the dominantdesign for the car encompassed not only the fact that it used a gasoline engine to providemotive force but also that it was connected to the wheels througha transmission and a drivetrainand was mounted on a frame ratherthan on the axles. A dominantdesign incorporatesa range of basic choices about the design that are not revisitedin every subsequent design. Once the dominantautomobiledesign had been accepted, engineers did not reevaluatethe decision to use a gasoline engine each time they developed a new design. Once any dominantdesign is established, the initialset of components is refinedand elaborated,and progress takes the shape of improvementsin the components withinthe frameworkof a stable architecture. This evolutionaryprocess has profoundimplicationsfor the types of knowledge that an organizationdevelopinga new productrequires,since an organization'sknowledge and its information-processing capabilitiesare shaped by the nature of the tasks and the-competitiveenvironmentthat it faces (Lawrenceand Lorsch,1967; Galbraith,1973).3
3
Forsimplicity,we will assume here that organizationscan be assumed to act as boundedlyrationalentities, in the tradition of Arrow(1974)and Nelson and Winter (1982).
In the earlystages of a technology's history,before the emergence of a dominantdesign, organizationscompeting to design successful productsexperimentwith many different technologies. Since success in the marketturns on the synthesis of unfamiliartechnologies in creative new designs, organizationsmust activelydevelop both knowledge about alternatecomponents and knowledge of how these components can be integrated.Withthe emergence of a dominant design, which signals the general acceptance of a single architecture,firms cease to invest in learningabout alternative configurationsof the established set of components. New component knowledge becomes more valuableto a firmthan 14/ASQ, March 1990
ArchitecturalInnovation
new architecturalknowledge because competitionbetween designs revolves aroundrefinements in particularcomponents. Successful organizationstherefore switch their limited attentionfrom learninga littleabout many differentpossible designs to learninga great deal about the dominantdesign. Once gasoline-poweredcars had emerged as the technology of choice, competitive pressures in the industrystronglyencouraged organizationsto learnmore about gasoline-firedengines. Pursuingrefinements in steam- or electric-powered cars became much less attractive.The focus of active problemsolving becomes the elaborationand refinementof knowledge about existing components withina frameworkof stable architecturalknowledge (Dosi, 1982; Clark,1985). Since in an industrycharacterizedby a dominantdesign, architecturalknowledge is stable, it tends to become embedded in the practices and proceduresof the organization. Several authors have noted the importanceof variousinstitutionaldevices likeframeworksand routinesin completingrecurringtasks in an organization(Galbraith,1973; Nelson and Winter,1982; Daft and Weick, 1984). The focus in this paper, however, is on the role of communicationchannels, information filters, and problem-solvingstrategies in managingarchitecturalknowledge. Channels, filters, and strategies. An organization'scommunicationchannels, both those that are implicitin its formalorganization(A reportsto B) and those that are informal("I always call Fredbecause he knows about X"),develop around those interactionswithinthe organizationthat are criticalto its task (Galbraith,1973; Arrow,1974). These are also the interactions that are criticalto effective design. They are the relationships aroundwhich the organizationbuildsarchitectural knowledge. Thus an organization'scommunicationchannels will come to embody its architecturalknowledge of the linkagesbetween components that are criticalto effective design. Forexample, as a dominantdesign for room fans emerges, an effective organizationin the industrywill organize itself aroundits conception of the product'sprimary components, since these are the key subtasks of the organization's design problem(Mintzberg,1979; von Hippel,1990). The organizationmay create a fan-bladegroup, a motor group, and so on. The communicationchannels that are created between these groups will reflect the organization's knowledge of the criticalinteractionsbetween them. The fact that those workingon the motor and the fan blade reportto the same supervisorand meet weekly is an embodiment of the organization'sarchitecturalknowledge about the relationship between the motor and the fan blade. The informationfilters of an organizationalso embody its architecturalknowledge. An organizationis constantlybarraged with information.As the task that it faces stabilizes and becomes less ambiguous,the organizationdevelops filters that allow it to identifyimmediatelywhat is most crucialin its informationstream (Arrow,1974; Daft and Weick, 1984). The emergence of a dominantdesign and its gradualelaboration molds the organization'sfilters so that they come to embody parts of its knowledge of the key relationshipsbetween the components of the technology. Forinstance, the relationships between the designers of motors and controllersfor a room 15/ASQ, March 1990
fan are likelyto change over time as they are able to express the natureof the criticalinteractionbetween the motor and the controllerin an increasinglyprecise way that allows them to ignore irrelevantinformation.The controllerdesigners may discover that they need to know a great deal about the torque and power of the motor but almost nothingabout the materialsfromwhich it is made. They will create informationfilters that reflect this knowledge. As a productevolves, informationfiltersand communication channels develop and help engineers to work efficiently,but the evolutionof the productalso means that engineers face recurringkinds of problems. Over time, engineers acquirea store of knowledge about solutions to the specific kindsof problemsthat have arisen in previousprojects.When confrontedwith such a problem,the engineer does not reexamine all possible alternativesbut, rather,focuses first on those that he or she has found to be helpfulin solving previous problems. In effect, an organization'sproblem-solving strategies summarizewhat it has learnedabout fruitfulways to solve problems in its immediateenvironment(Marchand Simon, 1958; Lyles and Mitroff,1980; Nelson and Winter, 1982). Designers may use strategies of this sort in solving problemswithin components, but problem-solvingstrategies also reflect architecturalknowledge, since they are likelyto express partof an organization'sknowledge about the component linkagesthat are crucialto the solutionof routine problems.An organizationdesigning fans might learnover time that the most effective way to design a quieterfan is to focus on the interactionsbetween the motorand the housing. The strategies designers use, their channels for communication, and their informationfilters emerge in an organizationto help it cope with complexity.They are efficient precisely because they do not have to be activelycreated each time a need for them arises. Further,as they become familiarand effective, using them becomes natural.Likeridinga bicycle, using a strategy, workingin a channel, or employinga filter does not requiredetailed analysis and conscious, deliberate execution. Thus the operationof channels, filters, and strategies may become implicitin the organization. Since architecturalknowledge is stable once a dominantdesign has been accepted, it can be encoded in these forms and thus becomes implicit.Organizationsthat are activelyengaged in incrementalinnovation,which occurs withinthe context of stable architecturalknowledge, are thus likelyto manage much of theirarchitecturalknowledge implicitlyby embedding it in their communicationchannels, information filters, and problem-solvingstrategies. Componentknowledge, in contrast, is more likelyto be managed explicitlybecause it is a constant source of incrementalinnovation. Problems Created by ArchitecturalInnovation Differences in the way in which architecturaland component knowledge are managed within an experienced organization give us insight into why architecturalinnovationoften creates problemsfor established firms. These problems have two sources. First,established organizationsrequiresignificant time (and resources) to identifya particularinnovationas architectural,since architecturalinnovationcan often initiallybe 16/ASQ, March 1990
ArchitecturalInnovation
accommodatedwithin old frameworks.Radicalinnovation tends to be obviously radical-the need for new modes of learningand new skills becomes quicklyapparent.But informationthat might warn the organizationthat a particular innovationis architecturalmay be screened out by the informationfilters and communicationchannels that embody old architecturalknowledge. Since radicalinnovationchanges the core design concepts of the product,it is immediatelyobvious that knowledge about how the old components interact with each other is obsolete. The introductionof new linkages,however, is much harderto spot. Since the core concepts of the design remainuntouched,the organization may mistakenlybelieve that it understandsthe new technology. Inthe case of the fan company,the motorand the fan-bladedesigners will continue to talkto each other. The fact that they may be talkingabout the wrong things may only become apparentafter there are significantfailuresor unexpected problemswith the design. The development of the jet aircraftindustryprovidesan example of the impact of unexpected architecturalinnovation. The jet engine initiallyappearedto have importantbut straightforwardimplicationsfor airframetechnology. Established firms in the industryunderstoodthat they would need to develop jet engine expertise but failed to understandthe ways in which its introductionwould change the interactions between the engine and the rest of the plane in complex and subtle ways (Millerand Sawyers, 1968; Gardiner,1986). This failurewas one of the factors that led to Boeing's rise to leadershipin the industry. [his effect is analogous to the tendency of individualsto continue to relyon beliefs about the world that a rationalevaluation of new informationshould lead them to discard (Kahneman,Slovic, and Tversky,1982). Researchershave commented extensively on the ways in which organizations facing threats may continue to rely on their old frameworks -or in our terms on their old architecturalknowledge-and hence misunderstandthe natureof a threat.They shoehorn the bad news, or the unexpected new information,back into the patternswith which they are familiar(Lylesand Mitroff, 1980; Duttonand Jackson, 1987; Jackson and Dutton, 1988). Once an organizationhas recognizedthe natureof an architecturalinnovation,it faces a second majorsource of problems: the need to buildand to applynew architectural knowledge effectively. Simplyrecognizingthat a new technology is architecturalin characterdoes not give an established organizationthe architecturalknowledge that it needs. It must first switch to a new mode of learningand then invest time and resources in learningabout the new architecture (Louisand Sutton, 1989). It is handicappedin its attempts to do this, both by the difficultyall organizationsexperience in switching from one mode of learningto anotherand by the fact that it must buildnew architecturalknowledge in a context in which some of its old architecturalknowledge may be relevant. An established organizationsetting out to buildnew architecturalknowledge must change its orientationfrom one of refinement withina stable architectureto one of active search 17/ASQ, March 1990
for new solutions withina constantlychangingcontext. As long as the dominantdesign remainsstable, an organization can segment and specialize its knowledge and relyon standardoperatingproceduresto design and develop products. Architecturalinnovation,in contrast, places a premiumon explorationin design and the assimilationof new knowledge. Manyorganizationsencounter difficultiesin theirattempts to make this type of transition(Argyrisand Schdn, 1978; Weick, 1979; Hedberg, 1981; Louisand Sutton, 1989). New entrants, with smaller commitments to older ways of learning about the environmentand organizingtheir knowledge, often find it easier to buildthe organizationalflexibilitythat abandoningold architecturalknowledge and buildingnew requires. Once an organizationhas succeeded in reorientatingitself, the buildingof new architecturalknowledge still takes time and resources. This learningmay be quite subtle and difficult. New entrantsto the industrymust also buildthe architectural knowledge necessary to exploit an architecturalinnovation, but since they have no existing assets, they can optimize their organizationand information-processing structuresto exploit the potentialof a new design. Establishedfirms are faced with an awkwardproblem.Because theirarchitectural knowledge is embedded in channels, filters, and strategies, the discovery process and the process of creatingnew information(androotingout the old) usuallytakes time. The organizationmay be tempted to modifythe channels, filters, and strategies that alreadyexist ratherthan to incurthe significant fixed costs and considerableorganizationalfrictionrequiredto buildnew sets from scratch (Arrow,1974). But it may be difficultto identifypreciselywhich filters, channels, and problem-solvingstrategies need to be modified,and the attempt to builda new productwith old (albeitmodified)organizationaltools can create significantproblems. The problems created by an architecturalinnovationare evident in the introductionof high-strength-low-alloy (HSLA) steel in automobilebodies in the 1970s. The new materials allowed body panels to be thinnerand lighterbut opened up a whole new set of interactionsthat were not containedin existing channels and strategies. One automaker'sbody-engineeringgroup, using traditionalmethods, designed an HSLA hood for the engine compartment.The hoods, however, resonated and oscillatedwith engine vibrationsduringtesting. On furtherinvestigation,it became apparentthat the traditional methods for designing hoods workedjust fine with traditionalmaterials,althoughno one knew quite why.,The knowledge embedded in established problem-solvingstrategies and communicationchannels was sufficientto achieve effective designs with established materials,but the new materialcreated new interactionsand requiredthe engineers to buildnew knowledge about them. Architecturalinnovationmay thus have very significantcompetitive implications.Establishedorganizationsmay invest heavilyin the new innovation,interpretingit as an incremental extension of the existing technology or underestimatingits impacton their embedded architecturalknowledge. But new entrantsto the industrymay exploit its potentialmuch more effectively, since they are not handicappedby a legacy of embedded and partiallyirrelevantarchitecturalknowledge. l18/ASQ,March 1990
ArchitecturalInnovation
We explore the validityof our frameworkthrougha brief summaryof the competitive and technical historyof the semiconductorphotolithographic alignmentequipment industry. Photolithographic alignersare sophisticatedpieces of capitalequipment used in the manufactureof integratedcircuits. Theirperformancehas improveddramaticallyover the last twenty-fiveyears, and althoughthe core technologies have changed only marginallysince the technique was first invented, the industryhas been characterizedby great turbulence. Changes in marketleadershiphave been frequent, the entry of new firms has occurredthroughoutthe industry's history,and incumbentshave often suffered sharpdeclines in marketshare followingthe introductionof equipment incorporatingseemingly minorinnovation.We believe that these events are explainedby the intrusionof architecturalinnovation into the industry,and we use three episodes in the industry's history-particularlyCanon'sintroductionof the proximityalignerand Kasper'sresponse to it-to illustrate this idea in detail. IN PHOTOLITHOGRAPHIC INNOVATION ALIGNMENT EQUIPMENT Data The data were collected duringa two-year,field-basedstudy of the photolithographic alignmentequipment industry.The study was initiallydesigned to serve as an explorationof the validityof the concept of architecturalinnovation,a concept originallydeveloped by one of the authorsduringthe course of his experience with the automobileand ceramics industry (Clark,1987). The core of the data is a panel data set consisting of research and development costs and sales revenue by productfor every productdevelopment projectconducted between 1962, when work on the first commercialproductbegan, and 1986. This data is supplemented by a detailed managerialand technicalhistoryof each project.The data were collected through research in both primaryand secondary sources. The secondarysources, includingtradejournals,scientificjournals, and consultingreports,were used to identifythe companies that had been active in the industryand the productsthat they had introducedand to buildup a preliminarypictureof the industry'stechnical history. Datawere then collected about each product-development projectby contactingdirectlyat least one of the members of the product-developmentteam and requestingan interview. Interviewswere conducted over a fourteen-monthperiod, from March1987 to May 1988. Duringthe course of the research, over a hundredpeople were interviewed.As far as possible, the interviewees includedthe senior design engineer for each projectand a senior marketingexecutive from each firm.Otherindustryobservers and participants,including chief executives, universityscientists, skilleddesign engineers, and service managerswere also interviewed.Interview data were supplemented whenever possible through the use of internalfirmrecords.The majorityof the interviews were semistructuredand lasted about two hours. Respondents were asked to describe the technical,commercial,and 19/ASQ, March 1990
managerialhistoryof the product-developmentprojectswith which they were familiarand to discuss the technicaland commercialsuccess of the productsthat grew out of them. In orderto validatethe data that were collected duringthis process, a brief historyof productdevelopment for each equipmentvendor was circulatedto all the individualswho had been interviewedand to others who knew a firm's history well, and the accuracyof this account was discussed over the telephone in supplementaryinterviews.The same validationprocedurewas followed in the constructionof the technical historyof the industry.A technicalhistorywas constructed using interviewdata, publishedproductliterature, and the scientific press. This historywas circulatedto key individualswho had a detailed knowledge of the technical history of the industry,who corrected it as appropriate. We chose to study the semiconductorphotolithographic alignmentequipment industryfor two reasons. The first is that it is very differentfrom the industriesin which our frameworkwas first formulated,since it is characterizedby much smallerfirms and a much faster rate of technological innovation.The second is that it providesseveral examples of the impactof architecturalinnovationon the competitive position of established firms. Photolithographic equipment has been shaken by four waves of architecturalinnovation,each of which resulted in a new entrantcapturingthe leadershipof the industry.Inorderto groundthe discussion of architectural innovationwe providea briefdescriptionof photolithographic technology. The Technology Photolithographic alignersare used to manufacturesolid-state semiconductordevices. The productionof semiconductors requiresthe transferof small, intricatepatternsto the surface of a wafer of semiconductormaterialsuch as silicon, and this process of transferis known as lithography.The surface of the wafer is coated with a light-sensitivechemical,or "resist." The patternthat is to be transferredto the wafer surface is drawnonto a mask and the mask is used to block light as it falls onto the resist, so that only those portionsof the resist defined by the mask are exposed to light.The light chemicallytransformsthe resist so that it can be stripped away. The resultingpatternis then used as the basis for either the deposition of material-onto the wafer surface or for the etching of the existing materialon the surface of the wafer. The process may be repeated as many as twenty times duringthe manufactureof a semiconductordevice, and each layermust be located preciselywith respect to the previous layer(Wattsand Einspruch,1987). Figure2 gives a very simplifiedrepresentationof this complex process. A photolithographicaligneris used to positionthe mask relative to the wafer, to hold the two in place duringexposure, and to expose the resist. Figure3 shows a schematic diagram of a contact aligner,the first generationof alignmentequipment developed. Improvementin alignmenttechnology has meant improvementin minimumfeature size, the size of the smallest patternthat can be producedon the wafer surface, yield, the percentage of wafers successfully processed, and 20/ASQ, March 1990
Architectural Innovation Figure 2. Schematic representation
of the lithographic process.
STEPS 1. Expose Resist Light
Mask Resist Wafer
2. Develop Resist Resist Wafer
3. Deposit Material
Material
Resist Wafer
4. Remove Remaining Resist
Pattern formed on wafer
throughput, the number of wafers the aligner can handle in a given time. Contact aligners were the first photolithographic aligners to be used commercially. They use the mask's shadow to transfer the mask pattern to the wafer surface. The mask and the wafer are held in contact with each other, and light shining through the gaps in the mask falls onto the wafer surface. Contact aligners are simple and quick to use, but the need to bring the mask and the wafer into direct contact can
damage the mask or contaminatethe water. The first proximityalignerwas introducedin 1913 to solve these problems. 21/ASQ, March 1990
Figure 3. Schematic diagram of a contact aligner. Source
Mechanical System A
Alignment System
Mask
Wafer Stage
C
ControlSystem
In a proximity aligner the mask is held a small distance away from (in proximity to) the wafer surface, as shown in the simplified drawing in Figure 4. The separation of the mask 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 coming through the mask spreads out before it reaches the resist, and the mask's shadow is less well defined than it is in the case of a contact aligner. As a result, users switching to proximity aligners traded off some minimum feature size capability for increased yield. The basic set of core design concepts that underlie optical photolithography-the use of a visible light source to transmit the image of the mask to the wafer, a lens or other device to focus the image of the mask on the wafer, an alignment system that uses visible light, and a mechanical system that holds the mask and the wafer in place-have remained unchanged since the technology was first developed, although aligner performance has improved dramatically. The minimum-feature-size capability of the first aligners was about fifteen to twenty microns. Modern aligners are sometimes specified to have minimum feature sizes of less than half a micron.
Figure 4. Schematic diagram of a proximity aligner. Source
Mechanical System (Includes gapsetting mechanism)
A
Mask I________________________
_I
Wafer
Stage
C
ControlSystem
22/ASQ, March 1990
Alignment System
ArchitecturalInnovation
Radicalalternatives,makinguse of quite differentcore concepts, have been exploredin the laboratorybut have yet to be widely introducedinto full-scaleproduction.Alignersusing xrays and ion beams as sources have been developed, as have direct-writeelectron beam aligners, in which a focused beam of electrons is used to write directlyon the wafer (Changet al., 1977; Brown,Venkatesan,and Wagner, 1981; Burggraaf, 1983). These technologies are clearlyradical.They rely not only on quite differentcore concepts for the source, but they also use quite differentmask, alignment,and lens technologies.
A constant stream of incrementalinnovationhas been critical to opticalphotolithography'scontinuingsuccess. The technology of each component has been significantlyimproved. Modernlightsources are significantlymore powerfuland more uniform,and modernalignmentsystems are much more accurate. In addition,the technology has seen four waves of architecturalinnovation:the move from contact to proximityalignment,from proximityto scanning projection alignment,and from scanners to first-and then secondgeneration"steppers." Table 1 summarizesthe changes in the technology introducedby each generation. In each case the core technologies of opticallithographyremainedlargely untouched,and much of the technical knowledge gained in buildinga previousgenerationcould be transferredto the next. Yet, in each case, the industryleaderwas unableto make the transition. Table 1 A Summary of Architectural Innovation in Photolithographic
Alignment Technology
Major Changes Technology
Critical relationships between components
Proximity aligner
Mask and wafer separated during exposure.
Accuracy and stability of gap is a function of links between gap-setting mechanism and other components.
Scanning projection
Image of mask projected onto wafer by scanning reflective optics.
Interactions beween lens and other components is critical to successful performance.
First-generation stepper
Image of mask projected through refractive lens. Image "stepped" across wafer.
Relationship between lens field size and source energy becomes significant determinant of throughput. Depth of focus characteristics-driven by relationship between source wavelength and lens numerical critical. Interactions between stage aperture-become and alignment system are critical.
Second-generation stepper
Introduction of "site-by-site" alignment, larger 5 x lenses.
Throughput now driven by calibration and stepper stability. Relationship between lens and mechanical system becomes crucial means of controlling distortion.
Equipment
Source: Field interviews, internal firm records (Henderson, 1988).
Table2 shows share of deflated cumulativesales, 1962-1986, by generationof equipmentfor the leadingfirms. The first commerciallysuccessful alignerwas introducedby Kulickeand Soffa in 1965. They were extremely successful and held nearly100 percent of the (verysmall) marketfor the next nine years, but by 1974 Cobiltand Kasperhad replaced entered the marketwith the them. In 1974 Perkin-Elmer 23/ASQ, March 1990
Table2 Share of Deflated Cumulative Sales (%)1962-1986, by Generation, for the Leading Optical Photolithographic Alignment Equipment Manufacturers*
Firm Cobilt Kasper Canon Perkin-Elmer GCA Nikon Total *
Contact 44 17
61
Alignment Equipment Step and Proximity Scanners repeat (1)
Step and repeat (2)
