The means of managing momentum: Bridging technological paths and organisational fields

To

Ca

b

ARRAA

KTPPSSI

1

tnngtaesmgsAndasp

TT

j

0h

Research Policy 42 (2013) 1389– 1405

Contents lists available at SciVerse ScienceDirect

Research Policy

jou rn al hom epage: www.elsev ier .com/ locate / respol

he means of managing momentum: Bridging technological paths andrganisational fields

ornelius Schuberta,∗, Jörg Sydowb, Arnold Windelera

Technische Universität Berlin, Department of Sociology, Fraunhoferstr. 33-36, 10587 Berlin, GermanyFreie Universität Berlin, Management Department, Boltzmannstr. 20, 14195 Berlin, Germany

a r t i c l e i n f o

rticle history:eceived 15 September 2010eceived in revised form 3 April 2013ccepted 29 April 2013vailable online 27 May 2013

a b s t r a c t

This paper examines how technological and organisational changes are mediated through different meansof mutually monitoring and collectively coordinating technological developments in the field of semi-conductor manufacturing. As collective practices, both monitoring and coordinating aim at generatingmomentum in order to stabilise or redirect technological paths in organisational fields. The empiricalanalysis of innovation practices in the field of semiconductor manufacturing technology shows that the

eywords:echnological momentumath dependenceath creationtructuration theory

means of managing momentum, above all roadmaps, conferences, and R&D consortia, influence andtransform the development of new technologies as well as the social relations within the organisationalfield. The transformative capacity of these means is elaborated conceptually using Giddens’ theory ofstructuration.

© 2013 Elsevier B.V. All rights reserved.
emiconductor industrynnovation
. Introduction

Technologies drive organisational change as much as organisa-ions influence the development of technologies. Hence it comes aso surprise that the interaction between technological and orga-isational change is a central object of organisation research ineneral and of research on technology and innovation in par-icular. For instance, research on socio-technical systems (Tristnd Bamforth, 1951), strategic management (Chandler, 1977),volutionary economics (Nelson and Winter, 1977), the social con-truction of technological systems (Pinch and Bijker, 1984) and,ore recently, on the structuration of organisational technolo-

ies (Orlikowski, 1992) has highlighted the deep interrelation ofocial and technical components within and among organisations.t the level of industries or organisational fields, research on tech-ical change (Rosenberg, 1963) and, more specifically, on dominantesigns (Abernathy and Utterback, 1978; Dokko et al., 2012; Kaplannd Tripsas, 2008; Tushman and Rosenkopf, 1992) have empha-
ised the fusion of technological and organisational structures in arocess of mutual adaptation.
∗ Corresponding author. Present address: Universität Siegen, DFG Researchraining School “Locating Media”, Am Eichenhang 50, 57076 Siegen, Germany.el.: +49 271 740 3867; fax: +49 271 740 4933.

E-mail addresses: [email protected] (C. Schubert),[email protected] (J. Sydow), [email protected] (A. Windeler).

048-7333/$ – see front matter © 2013 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.respol.2013.04.004

Even though the mutual shaping of technologies and organi-sations has been widely addressed, their relation is still not wellunderstood, particularly in terms of their process dynamics. More-over, on the level of the organisational field – and especially inscience-based industries – technologies and organisations are notonly linked directly through processes of technological innova-tion, but also through intermediate agencies such as collaborativeventures (Barley et al., 1992), cooperative technical organisations(Rosenkopf et al., 2001), and new forms of management (Pisano,2010). In this paper, we will analyse the role of specific means ofcooperation and coordination like technological roadmaps, con-ferences and consortia in the mutual shaping of technologiesand organisations. Such means are, at least in the field of semi-conductors, widely used for managing “technological momentum”(Hughes, 1994). This is because in this field, but also in many otherswhere standards matter, important technologies may constitutea dominant design or even a technological path (Arthur, 1994;David, 1985). By drawing on the empirical case of innovating novelsemiconductor manufacturing technologies (SMT), we will showthat the management of technological momentum is inseparablyrelated to the specific organisational means of cooperative R&D.

High-volume SMT provides an excellent case for studying notonly the constitution of a dominant design, but even more so ofa technological path, since the entire industry seeks to identify
one, and only one, technological option among several compet-ing alternatives to become the global standard for high-volumechip manufacturing in the future (Sydow et al., 2012). Adopting aprocess perspective, the notion of technological paths allows us to
dx.doi.org/10.1016/j.respol.2013.04.004

http://www.sciencedirect.com/science/journal/00487333

http://www.elsevier.com/locate/respol

mailto:[email protected]



dx.doi.org/10.1016/j.respol.2013.04.004

1 ch Pol

aom1GaStpataa

igpnoimbtbgo

m1fettecpfitaoen

f(thmtptmt

iSadutiscoiq

390 C. Schubert et al. / Resear

ccount for the contingent stabilisation of a technological optionver time. To analyse the constitution of technological paths inore detail, we draw on the concepts of path dependence (Arthur,

989; David, 1985) and path creation (Garud and Karnøe, 2001;arud et al., 2010) and adopt a social-constructivist and gradualistpproach towards path constitution (Meyer and Schubert, 2007;ydow et al., 2012; Windeler, 2003), which informs us as to howechnological options become stabilised both through undirectedrocesses of change, as well as the strategic activities of knowledge-ble agents in the field. We look at these dynamics on the level ofhe organisational field (DiMaggio and Powell, 1983) in order toccount for the collective nature of managing technological pathsnd interorganisational change.

The means of managing momentum are one important aspectn the relationship between technologies and organisations moreenerally, and in accounting for the dynamics between them inarticular, because they help to ‘bridge’ the development of tech-ological paths on the one hand with the transformations ofrganisations and organisational fields on the other. By introduc-ng new ways of collectively developing novel technologies, these

eans transform not only the technologies of chip manufacturing,ut also (inter-)organisational structures in the field. Even thoughhese means are employed by powerful field actors, they cannote fully controlled by them; instead, they occasion the contin-ent emergence of interrelated arrangements of technologies andrganisations.

In order to conceptualise the transformative capacity of sucheans, we take up ideas from the pragmatist tradition (Dewey,

958, pp. 121–165) which – in stark contrast to utilitarian orunctionalist notions – understands means not as mere means tonds, but as transformative agencies that enable and constrain, andhereby shape courses of action. We refine Dewey’s emphasis onhe inherently contingent, practical and mediated nature of humanxperience with Giddens’ (1984) theory of structuration by con-eptualising the means of managing momentum to simultaneouslyroduce and re-produce technological paths and organisationalelds – i.e. they constitute and transform the relations in whichhey are embedded. In particular, we will develop our argumentlong three lines of inquiry that bridge technological paths andrganisational fields by combining insights from (a) evolutionaryconomics, (b) organisational research, and (c) science and tech-ology studies.

First, the focus of inquiry in all these fields of study shiftsrom the individual to the collective. Since the days of Schumpeter1912) the number of technological developments driven and con-rolled by individual entrepreneurs or large integrated companiesas steadily declined. This is especially true for the semiconductoranufacturing industry, in which the in-house mode of innova-

ion dominant until the mid 1970s (Mowery and Rosenberg, 1998,p. 124–166), has been rapidly replaced by collaborative R&D sincehe mid 1990s (Ham et al., 1998; Sydow et al., 2012). The means of

anaging momentum are therefore mainly located at the level ofhe organisational field.

Second, the development of technology in science-based fields isnherently complex, uncertain, and dynamic. Current high-volumeMT combines leading-edge applications in physics, chemistry,nd mechanics, and closely connects technological and industrialynamics (cf. Langlois, 2000; Malerba et al., 2008). The complexity,ncertainty, and dynamics of these developments resonate withhe industry’s widely shared understanding that no single companys capable of handling these processes individually. This includesupplying the necessary financial resources as well as technologi-
al knowledge. The very task of aligning the required componentsf this science-based technology is so demanding that the lim-ts of established organisational forms of technology developmentuickly become apparent (Chuma, 2006).
icy 42 (2013) 1389– 1405

Third, and a key focus of our paper, is that technology develop-ment and organisational dynamics in SMT are mediated by differentmeans of monitoring technological progress and coordinatingcollective R&D. The social construction of technologies and thecontroversies surrounding competing technological alternatives(Bijker et al., 1987; Pinch and Bijker, 1984) in this area are notonly characterised by direct struggle between opposing parties, butalso by a collectively mediated process of generating consensus andcommitment (Dokko et al., 2012). In line with Giddens (1984), weargue that the means of managing momentum can be understoodas collectively organised practices for monitoring and developingtechnologies that have, over time, become taken for granted withinthe field and thereby changed the way actors evaluate technologiesand organise collective R&D ventures.

Our empirical data stems from an extensive qualitative studyin the semiconductor manufacturing industry. The main sourcesof information are interviews with 68 representatives from thefield conducted from 2003 to 2010, with a total of 96 interviews(for more detailed information on the sample see Sydow et al.,2012). Selected key informants were interviewed multiple timesand in part on different technological and organisational issues.The interviews were conducted with industry representatives fromcompanies along the SMT supply chain (device makers, tool makers,and component suppliers), test facilities, R&D consortia, universityinstitutes, and government funding agencies. The selected indus-try representatives are typically in charge of the innovation processwithin the companies and across cooperative R&D ventures. Inter-views were conducted in Europe, the US, and Japan in order toaccount for the global nature of the industry. In addition, we inter-viewed selected academic experts and carefully analysed the tradepress, as well as industry and academic publications from the pasttwenty years. The qualitative data allows for an inside perspec-tive on technical and organisational change, which we complementwith quantitative data for tracing how the changes manifest overtime.

Following Giddens’ (1984, p. 284) idea of “double hermeneu-tics”, we study the actors as they create shared frames of meaningand combine necessary resources by collectively organising thetransition from one generation of manufacturing technology to thenext. This gives us the indispensable inside perspective neededto explain the reflexive managing of momentum and the emer-gent stabilisation of technological paths and organisational fields.Thus, we conceive the innovation practices in SMT, like all othersocial practices, as inherently “situated activities” (Giddens, 1979,p. 54), in which social structures are not simply complied to, butalso created. Such emergent phenomena can only be explained ifwe understand innovation practices as “going concerns” (Hughes,1971, p. 52), as continuously made and unmade as the actors strug-gle for extending and creating technological paths (Sydow et al.,2012). Tracing the institutionalisation of novel means of coordi-nating and cooperating into shared innovation practices allowsus to reconstruct not only how such means are used and shapedby actors, but also how they themselves increasingly shape tech-nologies and organisations in the field. In terms of social theory,we conceive the actors in the field as “knowledgeable” (Giddens,1984, p. 30), being much aware of relevant features and practicesand reflexively using this knowledge to monitor and influence thecourse of technological innovation. To this we add the idea thatthe actors in the field do not monitor actual progress directly, butrather through collective means of managing momentum, whichthen shape the expectations and evaluations of technological solu-tions.

This approach is particularly helpful to explain the switchingfrom one technological path to another. Taking a novel manufactur-ing technology from the conceptual level to a proof of principle andfinally into high-volume production requires more than a decade

ch Poli

ofittdmedfianpjsitfsphsaptauto

2

ctoeaarcToaifiti)am

2

rsd–m“wot

C. Schubert et al. / Resear

f substantial investments. Since lithography is the key technologyor transferring the chip design to the material substrate, changesn this technology furthermore imply changes in the industry struc-ure itself. For instance, the increasing complexity and cost ofhe technological components are major reasons for the steadyecrease of companies who can and will partake in the develop-ent of novel lithographic solutions. The intended switch from the

stablished optical lithography procedure to a post-optical can-idate is a process that, until today, has taken the industry overfteen years and cost billions of dollars. Still, it is uncertain whennd even if this switch will actually take place and when and if aew technological path will be created. From our analysis of com-eting technological options, the idea of ‘hard’ criteria by which to

udge the performance of alternative technical options in advanceeems to be a wishful illusion. As our research shows, each options controversially discussed and entails a high degree of “interpre-ative flexibility” (Pinch and Bijker, 1984, p. 40). Thus, its success orailure can be analysed only in retrospect as medium and result ofocial processes collectively brought about by knowledgeable andowerful agents. What we are trying to understand in this case isow the criteria for evaluating a technological option are socially con-tructed in the collective process of managing momentum. We lookt how the criteria are used to organise a delicate decision-makingrocess for the identification of a technologically feasible optionhat is accepted throughout the industry. To do so, we adopted

longitudinal comparative analysis (Barley, 1990) which allowss to analyse the elaboration of the means of managing momen-um as they mediate the development of technical innovations andrganisational forms over time.

. The momentum of technological paths

The central question of managing technological momentumoncerns the extent to which and by which means actors, in par-icular organisations, are able to mindfully influence the coursef technological innovations so that a new technological pathmerges or, more precisely, remains in the process of being cre-ted. Similar to the (re-)production of social institutions (Bergernd Luckmann, 1966), the interactions in innovation processesesult in intended and unintended effects, which over time mayoalesce into “dominant designs” (Anderson and Tushman, 1990).he question of how much influence organisational actors havever this process has been the central issue between conflictingccounts of technological determinism and social voluntarism. Tak-ng the difficult relation of agency and structure as a starting pointor the discussion of technological paths, we will elaborate thedea of “technological momentum” as it was coined by the his-orian of technology Thomas Hughes (1983, pp. 140–174, 1994)n order to conceptualise the intricate relation and mutual (de-stabilisation of technological systems and organisational forms. Inddition, we will also pay close attention to the means of managingomentum.

.1. Technological paths and momentum

Let us first clarify the notion of technological momentum inelation to technological paths. A technological path is under-tood here as the patterned development of a technology that is,ue to increasing returns and other positive feedbacks, difficult

if not impossible – to reverse. Because of positive feedbackechanisms, the technological solution may eventually become

locked-in” (David, 1985) even though the solution is inefficienthen compared with other available alternatives. In the reasoning

f classic path dependence (Arthur, 1989; David, 1985, p. 334),he relation of agency and structure is analogous to ideas of

cy 42 (2013) 1389– 1405 1391

biological evolution, where dominant designs emerge over timefrom complex interactions on multiple levels. This occurs throughthe accumulation of individual actions, but remains beyond thecontrol of individual actors. We call this dynamic the emergence of atechnological path, which – as in Weber’s iron cage metaphor (1988,p. 203) – is transformed from a local practice into a dominantdesign, that is, an institution (cf. Hughes, 1994, p. 113). As David(1985, p. 334) notes, the lock-in is the result of three interrelatedfactors: (1) the tight coupling of material configurations and socialpractices (“technical interrelatedness”), (2) network effects facil-itating the diffusion of a given technological solution (“economiesof scale”) and (3) the rationale of profit-maximising actors seekingto avoid the costs of switching to alternative solutions (“quasiirreversibility of investment”). Later, David (1987) uses the expres-sion of “excess momentum” to point out undesirable bandwagoneffects that can occur after technological choices have beenmade.

This notion closely links momentum to the issues of irreversibil-ity, inertia, and, finally, a nearly inescapable lock-in. Garud andKarnøe (2001), on the other hand, have argued that paths do notsimply emerge, but can be and actually are mindfully created byactors in the field. In line with other studies examining the socialdeterminants of technological change (Bijker et al., 1987; Tushmanand Rosenkopf, 1992), these authors stress the contingenciesinvolved in the creation of a technological path. They demonstrate inparticular how actors mindfully navigate and manoeuvre on differ-ent material and social levels in order to create a technological path.In line with this discussion, we use the idea of managing momen-tum to include both emergent and deliberate aspects of innovatingtechnological paths.

Similar ideas have recently been employed to study institutionalchange. Streeck and Thelen (2005), for instance, have proposedgradual models of endogenous change, more often than not withradical consequences. These models stand in stark contrast notonly to classic path dependence, but also to concepts of punctu-ated equilibrium. The former views agents as merely followingthe path and assumes that a path can only be ‘broken’ by exter-nal shocks; the latter views change as abrupt and radical, mainlydriven by external forces. Nevertheless, emergent and deliberateaspects of change should not simply be equated with incrementaland radical change or aspects of structure and agency. Long-terminstitutional transformations as well as technological paths are con-stituted by the interplay of various aspects of change, i.e. emergentand deliberate as well as incremental and radical. Like Mahoneyand Thelen (2010), we do not assume that only powerful exter-nal shocks can disrupt impending or existing lock-ins. Moreover,very much in line with Garud and Karnøe’s (2001) idea of “mind-ful deviation” from an existing path, endogenous change maycreate gradual as well as radical transformations which may inturn (and for a certain time) lead back to durable arrangements.For the purpose of this paper, we use the notion of technolog-ical momentum to conceptualise this process of change as anincreasing mutual stabilisation of technological and organisationalforms.

Generally, the idea of technological momentum resonates wellwith ideas of path dependence and creation. Whereas the lock-inof classic path dependence reasoning operates mostly at the microlevel of individual action, for instance by linking typists with key-boards and isolated profit-maximising activities, the notion of pathcreation emphasises the power of collective agency (see also Garudet al., 2010; Sydow et al., 2012). Similarly, Hughes’ (1983, 1994)idea of technological momentum is not only situated between vol-
untarism and determinism, but it also specifically targets the levelof the organisational field. Innovating complex technologies is noindividual phenomenon, but rather inherently collective and, as weemphasise here, thoroughly mediated.

1 ch Pol

nsaaiamstav2a“(gotapsufk

totiitioefiiaic

dpdlLatdmotwitcb

cntibdda

produced by the actors in the field. In MacKenzie’s words, the meansof managing momentum are less as ‘camera’ for simply observing


In his study on electrification, Hughes coined the term “tech-ological momentum” in order to characterise the incrementaltabilisation of socio-technical infrastructures. Such infrastructuresre composed of individuals, ideas, institutions, and technologiesnd possess (1) mass, (2) velocity, and (3) direction. Similar to phys-cal momentum, Hughes conceptualises technological momentums the tendency “to resist change in the direction of develop-ent” (Hughes, 1983, p. 140) as socio-technological infrastructure

preads. Momentum therefore characterises the temporal and spa-ial expansion of the infrastructure as an increasing mass whichlso accumulates inertia, the resistance to change, yet is never irre-ersibly locked-in. While a “change-based momentum” (Jansen,004) may unfold intentionally, structural inertia or a lock-in maylso arise. Hughes thus sees momentum as a metaphor whichencompasses both structural features and contingent events”1987, p. 80). Similarly, it is not our objective to advocate analo-ies between physical mechanics and social processes for theirwn sake; rather, we want to enrich the metaphor by using theerms borrowed from physics to denote different aspects of man-ging momentum. If we take the physical definition as a startingoint, technological momentum can be understood as a form oftabilisation that has mass, velocity, and direction. It is the prod-ct of the increasingly irreversible interrelations of organisationalorms, technology systems architecture, computer chip sales mar-ets, R&D consortia, conferences and roadmaps.

In case of technological momentum, mass can be said to refero the active and passive (or voluntary and forced) supportersf a technology; those who share relevant meanings, jointly useest facilities, and rely on specialised human capabilities, etc.,n addition to the material technology itself. The expansion ofnfrastructure and increasing irreversibility of the links betweenechnologies and actors then result in an increased mass. Of course,n our understanding, mass is a relative term, not an absolutene. Managing momentum thus entails managing the inclusion orxclusion of more or less relevant actors and maintaining a suf-cient base of support for a technological option – relative, for

nstance, to an existing technological solution or path. In our case, itlso concerns the control of allocative resources, especially the abil-ty to get SMT up and running under high-volume manufacturingonditions.

The second variable, velocity (or power of movement), is moreifficult to translate from the physical to the social realm. In sim-lified terms, we can conceptualise it as the speed of technologicalevelopments, i.e. the rate at which critical technological prob-

ems, or “reverse salients” (Hughes, 1983, pp. 79–105), are solved.ike mass, velocity is socially produced by the agents who push ordhere to specific technological options. In addition and in contrasto their physical counterparts, mass and velocity are not indepen-ent variables in social processes of technology development, bututually constitutive. Especially for a complex system technol-

gy like SMT, setbacks in one variable can affect the dynamics ofhe other. In addition to the speed of development, the speed athich a large base of economic support is continually built also

nfluences the technological developments. Managing momentumhus entails organising a technology’s development with respect toreating and sustaining support over time, as well as pushing forreakthroughs.

Last but not least, momentum has direction. When looking atompeting technologies, we can argue that the direction of tech-ological momentum comes from decisions and actions relatedo mass and velocity. Over time, increasing stabilisation leads torreversibilities in the choices made along a technological path oretween different technological options; this already implies someegree of direction. Also, the system character of SMT provides
irection, as it guides developments to be in line with many of thelready existing technologies.
icy 42 (2013) 1389– 1405

2.2. Technologies and organisations interacting through means

Both the concepts of technological path and of technologicalmomentum suggest some form of direct relation between tech-nology on the one side and organisations on the other. For thedevelopment of complex technologies, however, a direct relation oftechnologies and organisations does not hold true, since science-based fields are increasingly populated by diverse organisationalagents, even intermediate agencies, all more or less generouslyequipped with capacities for collective action and transformation.In our empirical study, we will focus on two types of mediated col-lective activities: first, on the means of examining and selecting aviable and feasible technological option (see Sections 4.1 and 4.2)and, second, on the means of coordinating cooperative R&D (seeSection 4.3). The latter set of means is mainly aimed at generatingmass and velocity, the former at providing some sort of direction.

To understand at first the transformative capacity of means bet-ter, we need to see them as more than subservient to final ends, apoint which Dewey argued emphatically in contrast to utilitariannotions of ‘mere’ means (1958, pp. 121–165). Especially represen-tational means such as maps (like the technology roadmaps in ourcase) should not be seen as simple copies of reality, but as gen-erating new forms of engagement with the environment. Deweyuses the example of the map in discovering America: “Discoveryof America involved insertion of the newly touched land in a mapof the globe. This insertion, moreover, was not merely additive,but transformative of a prior picture of the world as to its surfacesand their arrangements” (ibid., p. 156). Dewey uses his exampleto make the point that America was not just simply there, eventhough it was still uncharted. For him, a change occurred in boththe map of the world and the world itself. Likewise, developingcomplex technologies such as SMT requires charting and evalu-ating many possible and unknown obstacles, so that technologyroadmaps become indispensable in creating the technologies theypredict. They provide not geographical, but temporal and technicalorientation in a highly complex and dynamic environment. As rep-resentational means, they connect physical objects with systems ofmeaning, whereby transforming both at the same time. FollowingGiddens’ double hermeneutic, the maps, as well as the other meansof managing momentum that we will address, are recursively cre-ated by actors, who thereby create a meaningful environment fordeveloping complex manufacturing technologies.

In the last few decades, the transformative capacities of repre-sentational means have been studied in diverse fields. Managementinstruments such as accounting systems (Power, 2004) or classi-fications on a broader scale (Bowker and Star, 1999) have beencritically examined. Similar ideas have also been prominently fea-tured in the discussion on the performativity of economic theories(Callon, 1998; MacKenzie, 2006). In essence, the argument ofperformativity seeks to elaborate on the mechanisms by which“economics, in the broad sense of the term, performs, shapes andformats the economy, rather than observing how it functions”(Callon, 1998, p. 2). Narrowing down the argument, MacKenzielinks performativity first to the “incorporation of economics intothe infrastructure of markets” (2006, p. 19) and second to the“incorporation into algorithms, procedures, routines, and mate-rial devices” (ibid.). Third, however, performativity should not bemisunderstood in the voluntary sense “that any arbitrary formulafor option prices, if proposed by sufficiently authoritative people,could have “made itself true” by being adopted” (ibid., p. 20). Like-wise, SMT is performed by the means of monitoring technologicalprogress and coordinating collective R&D as they are recursively re-

technological progress, but an ‘engine’ powering and transformingcollective R&D.

ch Poli

sthcoSftotSmophhe1teiawa

2

ammeiwm(

mcaiaeGrao

tettaimmIuoc

ear(


We can use the notion of performativity in order to furtherpecify the transformative capacities of the means – not limitedo technology roadmaps – for managing momentum. First, weave to ask what is actually being performed. For instance, if weonsider how these roadmaps are related to the reality of technol-gy development, the International Technology Roadmap for theemiconductors (ITRS) as the most important R&D-related arte-act in the field does not incorporate assumptions about a specificechnology; however, it does contain assumptions about formsf collective technology development. Thus, the performativity ofhe ITRS is not so much technological, but rather organisational (seeection 4.2). Subsequently, collective technology roadmaps do notake highly favoured technological options come true in the sense

f “self-fulfilling prophecies” (Merton, 1948) – as many highlyromising yet unsuccessful cases in semiconductor manufacturingave shown. As cooperative and often also collaborative means theyave, however, become indispensable in ensuring the continuedxistence of the self-fulfilling prophecy of “Moore’s Law” (Moore,965). Second, the notion of performativity directs our attentionowards intermediate agencies, e.g. the “market devices” (Callont al., 2007). In a general sense, we see that the means of manag-ng momentum perform and shape technologies and organisations,nd we therefore understand them as transformative agencieshich shape and are being shaped by the relations in which they

re embedded.

.3. Representational and cooperative means of structuration

Looking at this transformative capacity of the means of man-ging momentum, we can distinguish between two aspects. Theeans of managing momentum consist of both representationaleans (especially conferences, surveys and roadmaps) and coop-

rative means (mainly consortia). Although they often overlap, its helpful to maintain an analytical distinction between the two

ith respect to momentum in the sense that the representationaleans (mainly) provide direction, whereas the cooperative means

mainly) provide mass and velocity.From a structuration perspective, using the means of managing

omentum towards the creation of a new technological path is aontingent process of mutual elaboration and stabilisation. At first,ny new type of means may seem awkward; however, over time,t will perhaps become taken for granted and deeply ingrained in

given field. This is also true for the representational and coop-rative means used for building momentum for SMT. In line withiddens’ (1984, pp. 25–34), we see these means not only as the

esult, but also as the medium of social interaction, as enablingnd constraining, and thereby constituting the mindful activitiesf knowledgeable agents (see also Orlikowski, 1992).

The two types of means-based momentum management men-ioned above are crucial in processes directed at deviating from anxisting technological path and/or creating a new one. As a medium,hey ‘channel’ and ‘form’ social practices depending on how wellhey have been adopted by a given community. As a result, theyre themselves transformed in the process of collectively innovat-ng novel technologies. Spoken bluntly, it is only through these very

eans that knowledgeable and powerful agents are able to reflexivelyonitor their ongoing stream of interactions (Giddens, 1984, p. 30).

n the case of science-based industries with their complex, highlyncertain, and interrelated technologies, this is essentially an issuef organising collectives, of forging alliances and partnerships, andlosely monitoring and rationalising technological progress.

According to Giddens, means can be used to achieve domination,
.g. domination over material production. Means are thus useds allocative resources. Giddens also stresses that “authoritativeesources are every bit as ‘infrastructural’ as allocative resources”1984, p. 258); authoritative resources, thereby, relating to general
cy 42 (2013) 1389– 1405 1393

procedures and techniques in use to utilise knowledge, time andspace and relations among actors. Counter to a conventionalemphasis on the relevance of material resources for technologicalinnovation, he finds it “very important to demonstrate the parallelsignificance of authoritative resources” (ibid., p. 259). In the theoryof structuration, allocative and authoritative resources are distinctyet connected, since “allocative resources cannot be developedwithout transmutations of authoritative resources” (ibid., p. 260)– a crucial aspect in our concept of managing momentum. Thecommand over both allocative and authoritative resources of dom-ination creates different “forms of transformative capacity”, whichare “inherently bound up” with the rules of signification and legit-imation (ibid., p. 33). We would add that this command is neverdirect, but always mediated through a variety of different means.

As the means of commanding authoritative and allocativeresources mediate action by enabling and constraining, the sameis true for the rules of signification and legitimation. In SMT, twodominant structures of signification, namely technological feasi-bility and economic viability, are deeply intertwined (Sydow et al.,2012). However, the economic logic seems to hold the ultimateauthority in the field, as technologically feasible options havebeen dismissed in the past by questioning their economic viabil-ity (e.g. in case of 157 nm lithography, see below). The means ofmanaging momentum therefore include the means for framingprospective technologies in terms of feasibility and viability (e.g.cost-of-ownership calculations, which are used to pit competingtechnologies against each other in terms of apparently weightyindicators such as wafer throughput or component lifetime). Such“calculative practices” (Miller, 2001) are, of course, also a means ofcommanding authoritative resources, e.g. forming alliances to backpromising technological options. In addition, they are importantelements of creating ideas about the manageability of momen-tum itself, generating images of assessment and controllability intechnical and economic terms.

According to the rules of legitimation, specific means, includ-ing calculative practices, have to be accepted as legitimate formsof framing technologies and coordinating research and develop-ment. The history of the semiconductor industry shows that allnewly introduced means were initially confronted with majordoubts. New organisational forms such as R&D consortia were onlyhesitantly adopted in the field 25 years ago; today they are thetaken-for-granted mode of innovating technology (Sydow et al.,2012). As the rules of legitimation changed, so did the means ofcoordination. Up until today, the development process has beenheavily controlled and orchestrated by the device makers, whoset industry goals and are ultimately powerful customers vis-à-vis tool makers and suppliers. Although still very powerful in thefield, they are now finding it necessary to exercise some of thispower through consortia and other agencies. In sum, we can seethat the momentum of technological paths cannot be explainedthrough economic and technical criteria alone. Instead, we need toanalyse the antecedent social processes that produce the sharedcriteria for evaluating technological progress. We therefore ana-lyse the dynamics and locales of collective sensemaking, in whichtechnological promises and challenges are gradually turned intosignificant as well as legitimate claims and we study how thiseffects the cooperation and collaboration within the field.

3. Competing technologies and collaborating competitors

For SMT, the means of managing momentum are important in
at least two regards: first, the representational means for examin-ing and selecting a viable and feasible technological option out of aset of competing alternatives and, second, the means of organisingcooperation among competitors. Competition among technological

1 ch Pol

adnmi

ovtaetthwtnctvawga21amtwtm

3d

toFcratcclidbliaAa2slobgsmo(e


lternatives, and associated competition among firms, is a mainriver behind innovations in SMT as well as in other high tech-ology contexts (Gnyawali and Park, 2011). We focus on advance-ents in microlithography, which has been at the centre of the

ndustry’s progress, as predicted by Moore’s Law, since the 1960s.Lithography is the key technology for transferring circuit design

nto the chip. State-of-the-art microlithography for economicallyiable high-volume manufacturing depends on three technical fac-ors: throughput, performance and expandability. In the search for

technology that guarantees a high hourly wafer throughput, anqually high performance in chip pattern printing and the poten-ial to be extended over multiple chip manufacturing generations,he industry has followed a dual strategy of innovation. On the oneand, it continuously strives to enhance the existing technology,hile on the other hand it searches for promising novel solutions,

hus spurring fierce competition among technologies and compa-ies within the field (Henderson, 1995; Linden et al., 2000). In whatan be seen as a bottom-line consensus, all industry experts agreehat the industry cannot finance more than one solution for high-olume manufacturing. “Betting on the right horse” thus becomes

critical issue in the field. The dual strategy of extending the oldhile inventing the new is matched by increasingly coordinated

lobally networked forms of R&D, especially consortia, which havettracted scholarly attention for some time (Browning and Shetler,000; Carayannis and Alexander, 2004; Fong, 1990; Grindley et al.,996; Ham et al., 1998; Sydow et al., 2012; Thornberry, 2002; Westnd Iansiti, 2003). But how exactly do the means of managingomentum configure both social and technological dynamics in

he case of SMT? Before turning to a discussion of collective means,e will briefly look at some direct connections between competing

echnologies and the techno-economic conditions of high-volumeanufacturing.

.1. The relation of technical and organisationalynamics

Optical lithography has been used for semiconductor manufac-uring since the early 1960s. The dual strategy of enhancing theld while inventing the new can be traced back to the late 1970s.rom a technological perspective, lithography essentially prints thehip designs onto silicon wafers using some form of radiation. Cur-ent optical lithography uses light from the ultraviolet spectrumt 193 nm. The light is emitted by a laser source and projectedhrough a photo mask and a series of lenses. The mask carries thehip design, while the lenses focus and minimise the pattern on thehip. The mask pattern selectively exposes a photosensitive resistacquer, out of which the final chip patterns are developed. Depend-ng on the chip structure, this process is repeated several times withifferent patterns. The different technological components haveeen housed in large exposure tools called wafer steppers since the

ate 1970s. With the advancement of this system technology, thendividual components required an increasingly delicate alignmentnd have become highly interrelated in the current technology.dvances in light sources, for example, need to be matched withdvances in resist and lens production – and vice versa (Levinson,005, p. 142). The components themselves often contain manyophisticated sub-components. The optical system consists of theight source, the lenses and the mask, and the more advanced anyne component, the more complex its associated infrastructureecomes. For instance, the mask itself is the product of a litho-raphic process and involves a highly specialised and sophisticatedupply chain (cf. Wagner and Harned, 2010). Typically, advance-
ents in lithography have mostly been driven by the availability
f new light sources. Early generations using mercury arc lampsfirst emitting visible light and later ultraviolet) were replaced withxcimer lasers (emitting deep ultra violet wavelengths of 248 nm,

icy 42 (2013) 1389– 1405

193 nm, and 157 nm) in the mid 1990s. This triggered related R&Defforts to make the other components, e.g. the lenses, mask, andresist, compatible with the new sources.

The system character of the technology and the high specifica-tions demand that all components and enhancement techniques bealigned with extreme precision. Consequently, technical interde-pendencies among components have increasingly lead to economicentanglements among companies. Any component that falls behindbecomes a “reverse salient” (Hughes, 1983, pp. 79–105), slowingdown or even endangering the progress of the whole technologicalsolution. For instance, in case of 157 nm lithography, the time-consuming process of growing calcium fluoride crystals for theoptical lenses was mentioned in both our interviews and the tradepress (cf. IEEE spectrum, 2004) as one reason why the industry haddiscontinued this path, even though most of the other componentswere ready to use. The second drawback to using calcium fluo-ride crystals is their unwanted intrinsic birefringence, which needsto be reduced through a series of more complex manufacturingsteps (Suzuki and Smith, 2007, pp. 28, 433). In the case of 157 nmlithography Intel, as one of the field’s biggest players, abruptly dis-continued its support and issued a brief press release in early 2003.Even though other actors such as ASML, Zeiss and Infineon contin-ued their support, this was not enough to sustain an industry-wideconsensus – and hence the momentum – for this trajectory. Sub-sequently, 157 nm lithography was removed from the ITRS 2004update, although it was, at least at that time, considered the bestsolution for the 45 nm node and the second best for 65 nm and32 nm nodes on the ITRS 2003 edition. Such expensive dead endsirreversibly tie the success of a source company to that of a lensmanufacturer. Ruinous techno-organisational links are often citedby industry experts to highlight the fact that even though a tech-nology receives substantial support from the main actors, it mightstill fail on the eve of its market launch.

One test facility engineer related this textbook knowledge aboutthe technical interdependencies in microlithography (cf. Suzukiand Smith, 2007, pp. 503–700) to specific timing problems inthe industry and the subsequent organisational transformations incase of 193 nm lithography (cf. Beach et al., 2001; Brainard et al.,2002):

“In fact, 193 nm lithography also suffered from that [technicalinterdependencies] in the early days, that suddenly the scan-ners were ready to go into field but then it turned out thatthere were no resists that were good enough to perform on the193 nm light. So all the pieces of the puzzle have to be in placebefore you have the full technology ready. That’s why it’s veryimportant to communicate and to have these programs, like wehave here, we feel, where we bring together the IC manufactur-ers, but also the resist companies and mask shops and so on”(I 15: 20).

Therefore, the most powerful actors and fierce competitorsin this field, i.e. device makers such as Intel, AMD and Sam-sung, together with the stepper manufacturers, i.e. ASML in theNetherlands and Nikon and Canon in Japan, join forces to coordinatethe supply chains for their products. From an industry perspective,collaborative R&D is essentially aimed at leveraging resources andsharing risks for developing a technology that is increasing com-plex, costly, and uncertain.

Competing technological options thus represent competingalliances of diverse actors (cf. Gomes-Casseres, 1996). However,membership in any one alliance is not exclusive for the compa-nies and, depending on their financial recourses, they might decideto invest in competing options at the same time, hedging their bets
or joining at a later stage. In the early 2000s, device manufacturerslike IBM actually favoured several options. While the firm pushedfor electron projection lithography with Japanese collaborators, it

ch Poli

al

3

ciirratwaaaMt1tc(iiatfaceaiblsttafiiet

sbo(dlcdognpplXrauittb

technologies and organisations by providing legitimate locales ofcooperation. This increased cooperation has led the semiconductorcompanies to label parts of their business as pre-competitive while


lso joined the consortium working to promote extreme ultravioletithography in the United States.

.2. The collective search for alternative solutions

The process of finally selecting a single option that is technologi-ally feasible and economically viable for the entire industry to usen high-volume manufacturing represents one of most demand-ng and complex tasks in SMT today. The associated costs andisks are addressed by a high level of mutual monitoring andationalisation among competitors and along supply chains. As

science-based industry, the means of managing momentum inhe field of SMT have long included international conferences andorkshops, which are organised by both academic institutions

nd private consortia (Müller-Seitz, 2012). In addition, industryssociations were established as early as 1970, when equipmentnd materials suppliers formed the Semiconductor Equipment andaterials International (SEMI) in order to strengthen their posi-

ion with regard to the dominant device makers. In the early980s, the Semiconductor Research Corporation (SRC) was set upo intensify the ties between US universities and the national semi-onductor industry and in the mid 1980s, the first R&D consortiumSEMATECH) was founded. Since the 1990s collaborative test facil-ties like IMEC in Belgium, CEA-LETI in France, SEMATECH Northn the US, and SELETE in Japan have become important localesnd, in consequence, agencies in the field. However, this institu-ionalisation of cooperative R&D is not self-evident. SEMTATECH,or instance, has changed its purpose and performance consider-bly since it was initiated as a US public-private partnership toounter the increased market shares of Japanese chip manufactur-rs in the 1980s. Today, it is a globally oriented, privately fundednd seemingly neutral “catalyst for accelerating the commercial-zation of technology innovations into manufacturing solutions”y “setting global direction, creating opportunities for flexible col-

aboration, and conducting strategic R&D” (SEMTECH website). Asuch agencies increasingly populate the field of SMT, together withest facilities and dedicated R&D consortia, they influence or evenransform the relations of companies within the supply chains andlso across different supply chains. As the companies in the fieldercely compete for market shares, they also compete for positions

n supply chains and are nevertheless – or precisely because of thisxtreme competition – forced to enter cooperative ventures withheir competitors.

For more than 15 years, the search for alternative solutions foremiconductor manufacturing can be described as the struggleetween two main competitors: the enhancement of classicalptical lithography and more or less radically different types ofnew generations of) lithography. Radical alternatives have beeniscussed in the industry for a long time, most notably X-ray

ithography, which was mainly pushed by IBM, in the 1980s. Trulyollaborative R&D for novel manufacturing technologies, however,id not appear until the mid 1990s. From 1997 on, SEMATECHrganised a series of international workshops in order to initiate alobal research effort to identify the most promising candidates forext generation lithography. Several technological options wereut forward and discussed: electron-beam direct-write lithogra-hy (EBDW), ion-projection lithography (IPL), electron-projection

ithography (EPL), extreme ultraviolet lithography (EUV), and also-ray lithography (Linden et al., 2000). Today, as a preliminaryesult, although pending as a real solution, EUV remains the singlelternative to enhanced optical lithography for high-volume man-facturing. In effect, the dual strategy of enhancing the old while

nventing the new is a constant effort to avoid becoming locked-ino a potentially suboptimal technology (Sydow et al., 2012). Inerms of managing momentum, the industry seeks only what cane called an intermittent lock-in by setting standards for each new

cy 42 (2013) 1389– 1405 1395

technological generation and seeking to extend the establishedsystem as far as possible. Our interview partners have oftenhighlighted this as a conservative attitude within the industry.1

The conservative stance towards change in the industry is, dueto the high costs, risks and uncertainties of technology develop-ment in this area as well as experiences with once promisingoptions that nevertheless failed, ultimately rather understandableand widespread. This stance can already be found in early editionsof the National Technology Roadmap for Semiconductors (NTRS).In the 1994 NTRS edition, it is estimated that the semiconductorcompanies invest $4 billion in the two most developed technolog-ical generations. Comparatively, investment in future generationsare limited to $1 billion from federal laboratories, $70 million fromuniversities, $35 million from the Semiconductor Research Cor-poration, and a similar amount from industrial applied research,leading to the conclusion that “competing semiconductor compa-nies channel most of their R&D resources to these two generations[the most developed ones] and concentrate on the improvementand insertion of the chosen technologies” (NTRS, 1994, p. 2). Con-trolling the costs of shrinking feature sizes and evaluations ofeconomic viability continue to be central aspects in the ITRS edi-tion on lithography in 2001, which declares cost control and returnon investment – taken together – as the second most difficult chal-lenge faced by the industry (ITRS, 2001, p. 2). The prominence ofthis criterion still holds today. The 2011 ITRS executive summary,for example, identifies cost-effective manufacturing as the greatestchallenge in the short and long term, next to enhancing technicalperformance (ITRS, 2011, p. 20). This shows how selecting a tech-nological option within the industry’s dual strategy is primarilyguided by economic reasoning. Thus, advancing optical lithogra-phy is no self-reinforcing process largely developing behind thebacks of agents, but one that receives massive support from largeresearch endeavours within the industry (Chiu and Shaw, 1997).Therefore speaking of (active) path extension rather than (passive)path dependence seems more adequate (Meyer and Schubert, 2007;Sydow et al., 2012). In both cases, the momentum of SMT does notaccumulate from undirected individual actions, but is much moremindfully managed and coordinated by its contributors.

Managing momentum in our case thus entails generatingmomentum for new options (path deviation), while at the sametime preserving the momentum of the established solution forextending the present technological path. In addition to the interre-latedness of the technical components, the long-term developmenthorizon of about 15–20 years for each new option closely ties thecompanies to the technologies and the competitors to each other,regardless of whether they mainly pursue optical or post-opticaloptions.

Of course, competitors do not cooperate without friction. Inall interviews it was stated that although the companies wouldprefer to do otherwise, they see no choice but to partake in col-laborative ventures. In very few instances, competitors may shareintellectual property through consortia such as SEMATECH. Forexample, SEMATECH holds patents with Freescale and Infineon(Patent Number: US007709816) or with Intel and National Semi-conductor (Patent Number: US005614444). As means of managingmomentum, consortia articulate the delicate relations between

1 As requested by one of the anonymous referees, all subsequent interview evi-dence has been moved into footnotes. As one interview partner noted: “Now, assoon as optical lithography can go one step further, it ultimately is much cheaperthan any of these post-optical candidates and nobody wants to change and is justcontinuing to use optical” (I 15: 12).

1 ch Pol

fcs

ctmatiabeniicetimbo

tteuSia(viiIbtbptt

4tp

atesmtetl(ulqpatt

presentations and actual technological progress in the laboratories.Therefore, attendance at conferences is not only legitimate, but alsoessential for keeping up to date in an extremely dynamic field.2


encing off core areas of knowledge. The transformative capacity ofooperative means thus reaches deep into the structure of organi-ations willing to take part in collaborative endeavours.

The cooperative nature of R&D does not belie the extremeompetition in the field. In our view, the multiple cooperative ven-ures that characterise the field do not contradict competition, but

ust be seen as an indicator for it: cooperate to compete (seelso Gomes-Casseres, 1996). It is through the collaborative ven-ures that the boundary between competitive and pre-competitivessues is shifted, negotiated and re-created. For instance, the cre-tion of SEMATECH occurred at a time of strong economic rivalryetween the US and Japan. When new forms of coordination arestablished in the field, the lines of tension between the compa-ies may shift, but competition is not resolved. As we noted above,

ncreasing technological demands facilitated by cooperative R&Dncreased the competition among the remaining companies. Hence,ollaborative R&D can hardly be described as a pre-competitivenvironment. Supplier companies actively compete for access tohe device makers in R&D programmes while at the same time try-ng to exclude their competitors from such positions. Collectively

anaging momentum does not denote an absence of competition,ut specific ways of engaging with rivals and reframing the areasf competition.

Let us briefly take up the idea of an intermittent lock-in. Inhe case of semiconductor manufacturing, the actors in the fieldry to create a lock-in, at least for a limited time, in order toxploit its economies of scale and learning curves; this is the oftennderestimated the positive side of path dependence (Martin andunley, 2006). The actors, however, carefully try to avoid becom-ng locked-in completely. This delicate balance requires strategicnd coordinated efforts aimed at creating phases of stability andsome) change in a highly dynamic environment. Of course, indi-idual organisations, most of all companies, have strong interestsn creating intermittent lock-ins which favour their products. Fornstance, mask suppliers are not in favour of maskless lithography.ntermittent lock-ins thus are the product of strategic manoeuvresy relevant actors who try to influence the shape, the beginning andhe end of the lock-in. The different means of managing change-ased momentum, such as the ITRS, can be seen as organisationalractices which help to create and control intermittent lock-ins ando steer the industry’s direction and commit relevant actors, evenhough our empirical cases often reveal the limits of this control.

. The means of managing momentum: formattingechnologies and organisations from a structurationerspective

The means of managing momentum, representational as wells cooperative, are geared less towards initiating momentum thanowards driving, (re-)directing and sustaining it. To achieve thesends, a technologically feasible and economically viable option iselected and collective actions are synchronised. In this respect, theeans of managing momentum have to be delicately tailored to

he technological and organisational conditions in order to becomeffective and unfold their transformative capacity. We must alsoake into account that means – some of them artefacts – are interre-ated with bundles of practices and integrated in multiple contextsLatour, 1986). The ITRS, for instance, is at first glance an instrumentsed to time and synchronise technological developments. A closer

ook, however, reveals its political character: the ITRS is constantlyuestioned and interpreted by actors and companies for their own
urposes. Thus, it organises the development of shared techniquesnd general procedures to signify collective activities as well as,ogether with other cognitive or technological frames and interpre-ative processes (Kaplan and Tripsas, 2008), gives them a shared
icy 42 (2013) 1389– 1405

meaning. In addition, the ITRS sanctions both technological andsocial activities. Since, according to Giddens (1984), the reflexivemonitoring carried out by knowledgeable agents is a constitutiveprocess of social systems, we are now interested in the extent towhich the means of monitoring become part of this process, i.e. inthe shaping of organisational fields through collaborative practicesof evaluating technological progress.

4.1. Conferences, workshops, and surveys

Because of the experimental nature of R&D, one of the mostimportant means of monitoring technological progress in the fieldare scientific conferences. Conferences serve as the prime localesfor assessing the progress of technology at an early stage. They arethe most open occasions for discussing competing technologieswith collaborating competitors – and a broader public. As such,they are important means for shaping the industry, not only as“field-configuring-events” (Lampel and Meyer, 2008) but also asfield-shaping and -reshaping events. In line with Giddens (1984, p.118), we conceptualise the conferences as specific social locales orevents that provide the setting and context of interaction, as well asfacilities to influence the industry by connecting actors and diffus-ing research. A series of conferences constitutes a durable localeof temporal and yet continuous collective coordination of tech-nology development on an industry-wide scale by facilitating “theconstitution of encounters across time and space” (ibid., p. 119).

In the field of lithography, the most important conference is theannual SPIE Advanced Lithography Symposium in California. Inter-national conferences such as SPIE therefore form an important partof the industry-wide praxis of collectively coordinating technologydevelopment. At the conferences, early reports on experiments andtest results are presented by academic scholars as well as indus-try engineers, also often in collaboration. SPIE serves as the mostlegitimate industry-wide locale for signifying technology develop-ment or discussing the significance of research results – and allour interview partners acknowledged this role of the conference.In addition, conferences are seen as the locales for setting up anindustry-wide system of evaluating competing technologies. Con-ferencing can thus be viewed as the generalisable procedure toproduce shared understanding and meaning in the industry, i.e.as a rule of signification. General meetings, workshops, poster ses-sions and informal talks are, as in other industries, more or lessspecifically bound together.

The combination of these modes of knowledge sharing createstheir social and political meaning, i.e. their meaning for managingmomentum. The technical discussions at the SPIE symposium, for-mal in nature at the presentations and informal during conferencebreaks, can be seen as instances of generating momentum by con-necting and reconnecting the promises of a new technology witha number of interested contributors, i.e. creating and directing aninitial mass of momentum. For this reason – and equally for scien-tific pride, as we were told – the presentations at SPIE are driven bythe desire to present the best possible up-to-date results, which inturn sets the pace of research work in the laboratories. As compe-tent actors, the attendees also try to judge the difference betweenthe sometimes exaggerated claims of technical progress made in

2 One interview partner from a large supplier company described the sole purposeof his job as attending conferences and workshops around the world, monitoringtechnological progress, and feeding back his evaluations to his company. A SEMAT-ECH representative related these changes to changes within the companies: “Whenyou think back 20 years, companies largely covered the most aspects of development

C. Schubert et al. / Research Poli

taoowoidrwao

mna1biagSnrg

ctaso

tptccc

most important R&D consortium initiates conferences around theglobe to bring together companies as well as research institutesreflects the particularities of a science-based field, where technol-

3 One SEMATECH representative recalled the consortium’s early activities related

Fig. 1. Contributions to the SPIE conference “emerging lithographic technologies”.

When attending SPIE and other conferences became a necessity,he rules of legitimation were transformed as well. The conferencesre now seen as crucial elements in (de-)selecting a technologicalption and sanctioning the overall development process. This isften done in informal meetings during the conferences, throughhich members of competing companies establish and make use

f personal relations. Over the years and based on their technolog-cal competence, an informal group of actors evolves and meets atifferent occasions around the globe. In these interactions, delicateules of what may or may not be disclosed are created by the actors,ho unanimously insisted in the interviews that these venues are

much better source of information and basis for judgement thanfficial documents or presentations.

The increasing significance of SPIE in the field of lithography isirrored by its increasing popularity as a venue for discussing tech-

ological options. Fig. 1 depicts the contributions of organisationst the SPIE conferences on emerging lithographic technologies in998, 2002 and 2005, grouped by technological options assignedy SPIE. We can see an increase of attending organisations for EPL,

mprint, and most notably for EUV. In contrast, X-ray and e-beamttendance is declining. The contrasting dynamics in mixed andeneral technologies most likely originate in differences amongPIE classifications over the years. More importantly, the overallumber of contributions increases with time, indicating a rise inesearch and organised discussion on the topic of emerging litho-raphic technologies.

Because of its strong reputation in the field, events at SPIE areonsidered significant within the industry and, therefore, orches-rated with a high degree of reflexivity. Even though SPIE symposia
re open to many kinds of technological alternatives, the keynotepeeches at the event are actively used by speakers (and theirrganisations) to indicate the direction of development they are
hemselves and they can’t do that anymore. [. . .] You simply need people in the com-any that evaluate external events, because they don’t do it internally anymore andhat they simply notice developments early enough that were important for theompany. This is what happens at conferences, because there you hear: o.k. thisould be important for my company and then you of course talk to them [companyolleagues]” (I 28: 41).

cy 42 (2013) 1389– 1405 1397

advocating. This, in turn, is valuable information for the audience tocompare their perspective with that of industry “champions” suchas Intel or Samsung. Keynote speeches in particular are used as anauthoritative resource to shape relations in the field. As industrychampions or actors with strong reputations publicly evaluate theprogress and relevance of technologies, they organise sets of actorsto rally behind technologies and – simultaneously – even influencethe survival chances of many organisations. Knowledgeable indus-try actors are aware of the relevance of various contributions –including the fact that they are not all to be taken at face value.Based on the above description, it becomes apparent that SPIEshapes the ways money is spent, technologies are developed, andsets of relations between organisations are established. Confer-ences such as SPIE have indeed turned into resources of domination.

It was, therefore, of high strategic importance that SPIEintroduced a symposium on emerging post-optical lithographictechnologies alongside the established symposium for opticallithography in 1997. Introducing a dedicated locale to discuss thevital issues of competing technologies for what is commonly calledNext Generation Lithography (NGL), SPIE’s central position in thefield was used as a medium for putting the next generation onthe agenda of many companies, while at the same time advanc-ing its own reputation in the field. Innovating NGL was thereby,and despite all the competitive pressures that characterised theindustry, presented as an industry-wide collective effort of primeimportance and not something to be developed in secrecy.

As stated, one purpose of conferences in managing momentumis to provide a legitimate and significant context for an early evalua-tion of technologies as feasible and viable. SPIE is not the only venuethat achieves this end. Other important conferences and workshopsin the field are organised alongside the collaborative R&D venturesby consortia such as SEMATECH. These venues are often focusedon specific issues and allow for a closer, organised monitoring andsteering of R&D activities.3

The fist immersion workshop initiated by SEMATECH in 2002was followed by two successor workshops in 2003 and 2004.Following the workshops, SEMATECH set up six internationalsymposia on immersion lithography from 2004 to 2009. All theseworkshops and symposia were backed and co-organised by otherconsortia, such as SELETE and IMEC, the tool manufacturers ASML,Canon and Nikon, as well as by device makers and supplier com-panies. Due to the recurrent character of these conferences, theynot only initiate developments but also collectively stabilise andtransform them over time. Conferences in Europe and Asia, likeSEMICON or MEMS support this collective monitoring effort. Inaddition, the dedicated workshops concerning individual technolo-gies, like the annual International Workshop on EUV lithographyor the workshops on immersion lithography organised by SEMAT-ECH help to promote individual options. The fact that the industry’s

to immersion lithography: “So SEMATECH organised a workshop December 11th, Ibelieve, of 2002, where we pulled together a hundred people from across the indus-try, mostly suppliers, academic researchers, and end users of chip companies, andbrought everybody together to say what will it take to figure out if this is going to beviable? And we came out of that meeting with a list of 10 critical issues and an actionplan to go and address who was going to figure out the answers. And SEMATECH, atthat point, led a task force with three different working groups where we activelyfunded the research to get this done and in parallel worked with the suppliers whothey themselves, at this point, were now very actively working on it and it wasthrough that period that it then became apparent to the suppliers that this was aninteresting route to proceed with and that really led to the October 2003 announce-ment by ASML that this was something that they were going to really pursue” (I 32:7).

1398 C. Schubert et al. / Research Pol

ogLtoei“lpSatdswontsw

lbs

cahptsab1

Fig. 2. Number of SEMATECH workshops per year.

gy is developed at an experimental level and confronted withreat uncertainty. For instance, SEMATECH organises the biannualitho Forum which, as SEMATECH itself states, is “the semiconduc-or industry’s only global conference on the status and directionf advanced and next generation lithography”. The 2010 confer-nce explicitly links the technological developments to businessnterests by focusing, as the conference is officially advertised,on turning ‘innovation into value,’ the conscious development ofitho technology solutions that incorporate economic feasibility torovide true value and allow the chip industry to grow” (sourceEMATECH). As an additional monitoring measure, the participantsre asked to complete a survey at the end of the conference in ordero evaluate the development status of each technology presenteduring the conference. This survey aggregates the perceived reversealients of each technological option and its components across thehole industry. Assessing technological readiness in more or less

pen forums like conference surveys reflects the industry’s orga-ised search for alternative technological solutions. It also reflectshe – despite or precisely because of competitive pressures – con-ensual character of identifying a single promising alternative asell as the accumulation of momentum behind a certain solution.4

Fig. 2 shows the dominance of SEMATECH workshops on EUV
ithography compared to immersion and other NGL solutionsetween 1999 and 2010. On the one hand, this indicates the mas-ive technical issues which needed to be solved in order to get
4 The SEMATECH representative cited above formulated the difference betweenollective predictions and actual market transactions as follows: “It’s hard to put

specific value on any one part of the process. [. . .] I think first and foremost itelps people within organisations to align their resources to be able to tackle theseroblems. So if you’ve got a champion within the company and by making use ofhe roadmap or making use of these surveys, they can help get the managementupport they need. But apart from that, at the end of the day it’s purchase ordersnd signed contracts that really decide. [. . .] It all flows down and it’s money andilateral agreements, the multilateral piece helps people stay synchronised” (I 32:4).

icy 42 (2013) 1389– 1405

the technology ready for production. On the other hand, it sig-nifies the industry’s willingness to support such a technology inthe first place. As the most difficult EUV challenges are systemat-ically addressed, the technology slowly matures and the numberof SEMATECH workshops begins to decrease since 2005. This doesnot mean that EUV is fully accepted in the industry, nor thatall technological problems have been solved, but rather indicatesthe industry’s collective push towards EUV and the generation ofmomentum which is then carried further in subsequent R&D ven-tures.

As will be shown in the next section, this effort was accompaniedby a prominent positioning of EUV on the ITRS since 1994, where itwas placed as the most promising manufacturing technology about10 years in the future. This timeframe was extended through thefollowing editions of the roadmap, since enhancing 193 nm lithog-raphy further extended the path of classical optical lithography.The workshops also mirror the increased collaborative R&D effortssketched out in Section 4.3, which all together form the collectivepractices of managing momentum in the semiconductor industry.

In sum, conferences, workshops and surveys format technolo-gies and organisations as the medium and result of repeatedcollective activities. They frame technological alternatives as(un)feasible and (not) viable in order to match them with compa-nies’ competitive interests and provide orientation and direction.At the same time, these means of managing momentum placepressure on the companies to partake in collectively organisedactivities. Thereby, managing momentum entails the managementof time, synchronising the activities within the field (as we can alsosee from the quote in footnote 4). We will therefore now take acloser look at the most prominent means of managing momentumthrough timing: the ITRS.

4.2. Technology roadmaps and cost of ownership calculations

Collective roadmapping processes have been used in thesemiconductor industry since the early 1990s. As technologicalforecasting methods, they have been used in-house since the early1970s (Pyke, 1971), but only in recent years has their transfor-mative character drawn attention (Miller and O’Leary, 2007). Theperformativity of the ITRS – in the sense that it increases the prob-able occurrence of its own predictions – rather lies in the generaltransformation of the innovation dynamics according to Moore’sLaw than in accurately predicting a specific solution. Since it isessentially used to drive the miniaturisation of computer chipsaccording to this ‘law’, it is often referred to in the industry as the“shrink roadmap”. We can see that the self-fulfilling prophecy ofMoore’s Law is not a mere idea which becomes true because peoplebelieve in it, but because the material and symbolic infrastructurecontained in the ITRS is actually set up within the industry in orderto make the predictions come true – a fact of which the actors in thefield are well aware and an aspect of how the ITRS not only trans-forms technologies but also collaborative innovation practices atthe same time.

The ITRS itself is relevant in at least two forms, as a printedreport and as an organised process which produces the report. Theprinted ITRS roadmap is a rolling instrument of planning. A new edi-tion is published every two years with interim updates. The reportis used by many agents in this industry and beyond as a meansto influence, for instance, decision processes in individual com-panies and research institutes, as well as to tap into governmentfunding. Thereby, the production process of the ITRS roadmap isat least as important as its outcome, the printed report. And the
production process shares basic similarities with the points dis-cussed in the context of scientific conferences. Looking at the ITRSfrom this angle, it is a specific locale for collectively monitoring andinfluencing technological progress. In contrast to the conferences,

C. Schubert et al. / Research Poli

etoet

acanUTitTmtst2ctfEitogmttwfthgaI

ifbtabar2Doon

choices. It is furthermore a medium and result of actors using theirfinancial resources and knowledge, and controlling authoritativeand allocative resources to collectively define what is going on and

5 One of the early participants in the semiconductor roadmapping process recalledit this way: “And all these three groups [universities, national laboratories and indus-try] used to go with different roadmaps and requirements. So it was very difficultfor the government to understand what to fund. And so, in 1991, there was thisproposal to take these three groups and do the same thing; essentially, get togetherand come out with a single roadmap. [. . .] And so I was involved in the 1994 ver-sion, but since that time – I’m not saying it in a negative way – it was dominatedmore by universities and national labs because they were the ones receiving most ofthe funding. The industry was participating but somewhat almost like a spectator.[. . .] The ’97 [roadmap] already was very different, because there was a much largerindustry component that began changing many of the numbers that at times were

Fig. 3. Technologies in production and as estimated by the NTRS/ITRS.

specially the SPIE conferences, which address the entire indus-ry at single events, the ITRS is segmented into regional groupsf experts (so-called technical working groups) who continuouslylaborate and evaluate the progress of technology development inhe industry.

As a collaborative achievement, the production of the ITRS –s an artefact – reflects the aggregated evaluations of technologi-al feasibility by selected experts, e.g. device makers, tool makersnd suppliers, as regards the goals, solutions, and problems con-ected to all areas of semiconductor manufacturing. Starting as aS project for focussing research efforts in 1992, the first Nationalechnology Roadmap for the Semiconductors (NTRS) was compiledn 1994 with a timeframe of fifteen years. In 1999, the first interna-ional version was created with contributions from the US, Europe,aiwan, Korea, and Japan. The ITRS essentially maps the deviceanufacturers’ demands concerning novel technologies based on

he predictions of Moore’s Law fifteen years into the future. It pre-cribes three-year intervals of scaling down technology nodes (inhe 2010 update: 32 nm by the end of 2009, 22 nm by the end of012, and 16 nm by the end of 2016 for flash memory chips) andontains competing technological options as well as assessments ofheir present technological status. Furthermore, it contains a time-rame for “narrowing down” the options, i.e. for creating a path.ach technology node thus represents a possible intermittent lock-n discussed within the industry, and the ITRS specifies the requiredechnological solutions as well as the duration of use. Because of itspen and public nature, the assessments made in the ITRS are aood indicator of the industry’s direction; however, with regard toass and velocity its predictions are hardly accurate. Again, only

he knowledgeable actors in the field, those with technical exper-ise and an understanding of how technology development is dealtith in the industry, those who have closely monitored the dif-

erent technological developments, are able to competently judgehe assessments involved in the roadmapping process. Fig. 3 showsow the roadmap predictions differ from the actual use of technolo-ies in production. The estimations of the 1994 and 1999 roadmapsre matched with the production spans derived from the NTRS andTRS from 1994 to 2009.

If we look at EUV technology more closely, we can see how its continuously delayed about ten years into the future. EUV wasavoured for 100 nm production in 2007 on the 1994 NTRS andecame the prominent option for 70 nm manufacturing in 2009 inhe 1997 edition. In the 1999 ITRS edition, EUV was pushed downs the most promising solution to the 50 nm node and below toe ready in 2011. In 2001, EUV was favoured for the 45 nm nodend in 2003 for the 32 nm node, to be ready by 2010 and 2013espectively. In 2007 it was set for 22 nm production in 2016. In011 it was favoured for 22 nm manufacturing of processors and
RAM memory in 2015 and for 11 nm production of NAND mem-ry in 2019. All through this time, EUV was constantly challenged byptical enhancements, NGL alternatives or other innovative tech-ologies. Nevertheless, this particular technology has remained the
cy 42 (2013) 1389– 1405 1399

most promising solution projected about 10 years in the future andhas outlasted most of its former rivals (see also Fig. 4 below).

Because of the manifold delays of the roadmap predictions, theindustry has an ambivalent opinion on the ITRS. On the one hand,all interview partners have stressed that the roadmap is “alwayswrong” with respect to current developments and future fore-casts, some even went so far as to call it a “completely politicalartefact”. On the other hand, the ITRS technical working groupsare legitimate and significant locales for monitoring technologi-cal progress more closely than at conferences. Yan Borodovsky,director of advanced lithography in Intel’s technology and man-ufacturing group, summed up the importance of the ITRS at a SPIEkeynote speech in 2006:

“Moore’s Law is there to set common goals. The power ofMoore’s Law observation and prediction was, is and will befor foreseeable future to provide common, easily understoodquantified metric for everyone in semiconductor and IT econ-omy to synchronize their efforts toward historically based, welldefined, sustainable and mutually rewarding growth goals inthe future. Roadmaps are there to debate the Path. The Pathto achieving those goals (Roadmap) was, is and will be sub-ject to unending debates as it reflects fundamental uncertaintyof assessing risks to schedule and yields of ever more com-plex novel technologies over extending existing “tried and true”approaches beyond its originally defined limits in the absenceof data” (cited from presentation, slide 31).

In addition to creating some form of consensual orientation inthe absence of hard data, the ITRS can be used to control authori-tative and allocative resources. The history of the ITRS shows thatindustry participation was not taken for granted in the early stages,but companies had to be convinced to join. As membership changedover time, the roadmap was transformed from a means of control-ling government funding into a means of coordinating industryresources.5 This shift is also discussed in the reports about theNTRS at that time. Out of all the collaborative efforts, the 1992roadmap workshops were seen as the first main collaborative effortbetween industry, government, and academia in the US (Spencerand Seidel, 1995). However, roadmapping also required the sci-entists to “become more realistic in determining budgets and insetting schedules and priorities for project development” (ibid., p.219). In 2001, industry actors made up 76% of the ITRS processparticipants while actors from consortia, research institutes anduniversities came in at 23%. Government commitment has steadilydeclined since the early 1990s (cf. Schaller, 2004, p. 617).

Today, the ITRS gains its legitimacy as an instrument of commu-nication and coordination from the global participation of relevantactors in processes of signifying technologies and sanctioning

theoretical and not very practical. And in addition I observed another problem. Attimes I would go to Japan there would be a presentation showing how the roadmapwas wrong and they were right. So, the best way if you want to be right, then youtake the people that criticise and make them part of the process” (I 33: 14).

1400 C. Schubert et al. / Research Policy 42 (2013) 1389– 1405

emer

wpImtboba

cpmbwouaosrccllivt

p

lf

tbbann

Fig. 4. Disappearing, continuing and

hich technology to support. As a medium and result of roadmap-ing procedures, we can observe that acceptance of the NTRS and

TRS was not a simple process of adaptation, but one of contestedeanings, connected to the validity of the roadmap predictions and

he competitive positions of the actors. Over the years, the mem-ership structure and focus on manufacturability changed in favourf the industry. Since the predictions remain vague, the most directenefit for the companies is to gain access to a structured discussionbout competing technologies in the working groups.6

For the industry, the ITRS now is the locale for a much more –ompared to conferences – institutionalised controversy on com-eting technologies. In this respect, the ITRS is a major means foranaging momentum with respect to mass, velocity and direction

y setting standards for competing technologies in an industry-ide manner; in addition, it makes different options comparable

n a scientific basis and with respect to technical feasibility in man-facturing. However, a technology’s position on the roadmap canlso be attributed to political manoeuvres among competitors: inrder to generate momentum for a technological option, its inclu-ion in the ITRS is considered essential. Depending on the size andeputation of a company, this can be relatively easy or highly diffi-ult. Assessing the roadmap thus requires both technical and socialompetencies. Therefore, the transformative capacity of the ITRSies precisely in its repeated failure to accurately predict techno-ogical developments, while retaining the capacity to organise thennovation process in the field. This is especially true since indi-idual companies match their roadmaps to the ITRS for strategic
echnology development.7
Roadmapping has not only connected the companies, it has alsoermeated their boundaries and is now an established means for

6 A SEMATECH representative summarised this pointedly: “Look, I think by andarge when you look at it [the ITRS], I think it’s a pretty successful process, apartrom the fact that it’s completely unable to predict anything” (I 32: 19).

7 Referring to a device maker in the industry, one supplier company represen-ative stated: “Of course this changed over time, that the [individual] roadmapsecame clearer and more important and more visible to the customers. I still remem-er in 96/97, I was talking to [the customer] and they had no roadmap whatsoevernd they always asked ‘what are you going to do next?’ – this was at a time whenext generation lithography surfaced. [. . .] There was essentially nothing, they hado strategic considerations. This has changed” (I 24: 15).

ging options on the NTRS and ITRS.

organising strategic technology development in the industry acrossorganisational boundaries. Its transformational capacities nowbecome even clearer: roadmaps are hardly a means of objectivelyforecasting technological futures, i.e. tools of mere observation,but, and more profoundly, have been essential in politically trans-forming the divergent R&D practices of the field into a sharedunderstanding of cooperative strategic technology development.Even though the ITRS is considered to be “always wrong” tothe extent of being a “complete fake”, the individual companyroadmaps are taken very seriously and not disclosed openly. Thisis quite understandable, given their immense implications for theorganisations’, especially the companies’, fortunes.

The dynamics of the roadmap predictions are sketched out inFig. 4. It shows the industry’s switch from one set of competingalternative to another. Until 2003, EUV mainly faced competi-tion from e-beam technologies, ion-projection and X-ray. Thesealternatives were then discontinued in favour of maskless, imprintand other innovative technologies such as directed self assembly.EUV, even though continually postponed, remains prominently onthe ITRS until today.

Similar to conferences and surveys, roadmapping practices asmeans of managing momentum link technological paths and orga-nisational fields in a mutually constitutive process. Thereby, theyexhibit representational as well as cooperative properties. Newforms of organising interests across organisational boundariescoincide with new forms of framing technologies. The instrumentsfor evaluating technological progress trigger (inter)organisationalchange as much as the new forms of organising provide the socialfabric to make sense of the predictions.

Other means for evaluating the economic viability of techno-logical options with the help of calculative practices also figureprominently in the industry. Estimating the costs of purchase andoperation, as well as maintenance, in relation to throughput is amain concern in deciding on the adoption of a technology within theindustry. In the 1980s, SEMATECH prepared an early cost of owner-ship (CoO) calculation model as a simple spreadsheet calculation.This model was subsequently refined and adopted by member
companies and diffused as specialised software within the indus-try in the 1990s. It is interesting to note that in the competitionbetween enhanced optical lithography and EUV, the sophisticatedoptical lithography enhancements now decrease wafer throughput

ch Policy 42 (2013) 1389– 1405 1401

attEpl

nebimaottfnttotrbtetao

4

twlidcticfsm

Sfficcnat

doiwhistc


nd significantly increase the chip manufacturing costs. Especiallyhe enhancement known as “double patterning”, which decreaseshroughput by exposing the wafer twice, approaches the cost ofUV in CoO calculations. The economic reverse salient of through-ut is then directly related to technological reverse salients, e.g.

ight-source power and resist sensitivity.Collectively calculating the economic competitiveness of tech-

ological alternatives in terms of costs of ownership adds economicvaluations to the technical arguments found in the ITRS. Both cane seen as elements of the rules of signification that are used to

dentify reverse salients in the field. As means for signalling andonitoring technological progress, like conferences, workshops

nd surveys, they are likely to have a profound impact on the devel-pment of technology, depending on the outcome and relevance ofhe calculations. Whereas the roadmap incorporates ‘soft’ negotia-ions on technological components, CoOs provide seemingly ‘hard’acts through the use of algorithms, which of course are equally, ifot more, limited when it comes to predicting the future. Together,hese means have increased the future-oriented perspective ofhe industry and are therefore fundamental in the managementf momentum. Today, CoOs have also become established, almostaken-for-granted instruments that are slowly integrated into theules of signification and legitimation within the field until theyecome indispensable elements of institutional life. In the process,hey recursively stabilise these rules as the means are collectivelymployed by a growing number of participants. When used to con-rol authoritative and allocative resources, i.e. in directing fundingnd selecting options, they can even be considered a central meansf managing momentum.

.3. Consortia, networks, and alliances

R&D consortia are typically seen as locales for organising collec-ive ventures, for shaping rules of signification and legitimation, asell as for combining or even pooling resources to drive techno-

ogical progress. But they do not simply drive technology. Rathern the field of SMT, consortia are equally locales for monitoring,iscussing, and evaluating technological progress. This is espe-ially true for consortia like SEMATECH, which are involved inhe diverse monitoring activities discussed above (e.g. organis-ng workshops and publishing CoO calculations). Consortia, otheromplex networks, and several dyadic alliances address particularorms of coordinating interactions and relations between organi-ational agents in time and space and, hence, become means ofanaging momentum.8

The surveys at the end of the Litho Forums organised by theEMATECH consortium mentioned above summarise the criteriaor evaluating technologies. In the 2004 survey, the seven criteriaor the success of each technological solution can be broken downnto four technical issues (minimum feature size, critical dimensionontrol, demo solutions for critical issues and multi-generationalapability), two temporal issues (commercially available when
eeded and infrastructure in place), and the financial issue of beingffordable in terms of cost of ownership. The 2006 survey showshat the technological, temporal, and financial assessments of users
8 A SEMATECH representative highlighted the role of the consortium in evaluatingifferent technological options as lithography progressed over the years: “So I thinkne of the important roles that SEMATECH tried to do was try to – kind of funnyn the way that I’m going to try to describe this – but we tried to standardise the

ay that we looked at competing technologies and tried to put them . . . tried toave a common set of requirements and a common set of critical issues that the

ndustry understood to be the right critical issues, that if these critical issues wereolved by one of the technologies it would be a manufacturable solution. So I thinkhe selection criteria, SEMATECH had a lot to do with instituting what the selectionriteria would be for an individual company” (I 31: 24).

Fig. 5. Duration of NGL research collaborations.

and suppliers may differ quite significantly. For instance, the “gatereadiness” of EUV for 2012 and 2015 is seen much more optimisti-cally by suppliers than by users, the latter being more optimisticin case of 193 nm immersion with possible extensions.9 The stan-dardisation of monitoring practices, in both general and specialisednetworks, has a profound impact on the evaluation of competingtechnologies. What is noteworthy in this context is not whetherSEMATECH’s criteria are or have been better than others, but thata globally organised process of assessing technological develop-ments ultimately transforms all the participants’ perspectives intoshared, institutionalised sets of rules of signification and legitima-tion.

Despite the often competing views held in the general and ded-icated cooperative ventures, we observe the deep interrelationbetween complex system technologies and organisational forms.We can see how the innovation practices become increasinglymindfully organised and how the means of organising, in par-ticular those of consortia like SEMATECH, become increasinglyinstitutionalised and important in the field. The growing number ofconsortia, networks, and alliances around the globe further signifiesthe importance of collaborative means of managing momentum
(e.g. in case of NGL, see Fig. 5). Since the 1990s, several permanentlocales of cooperation have been set up, for instance test centres
9 We must note again that EUV, even though promising and favoured within theindustry, currently is still not ready for high volume manufacturing and of the toolmanufacturers, only ASML is pursuing this option now with a $4.1 billion invest-ment by Intel. Its Japanese competitors have nearly stopped their EUV engagementand industry experts doubt whether EUV will ever meet the necessary technicaland economic requirements (cf. “ASML wins funds for chip technology from Intel”REUTERS, 10.07.2012).

1 ch Pol

lU

mcoehnrmomtcaro

Satleiccttjancw

norllnec(istisws

dta

bfaatmpft2


ike IMEC in Belgium, CEA-LETI in France, SEMATECH North in theS, and SELETE in Japan.

A second important aspect addressed by the consortia isanaging the delicate boundary between pre-competitive and

ompetitive issues of technology development. Competition notnly involves competing technologies, but also the companies andven networks. They compete over technologies as well as overighly qualified engineers and scientists, suppliers, and – last butot least – market shares. Conferences, roadmaps, and surveys areelatively open and concern early stages of technological develop-ents. Even though early instances of competition can be initiated

r spotted at conferences and most certainly within the roadmap,ost players in the field label them as “pre-competitive” and are

hus able to domesticate market forces, at the very least on theognitive level (Porac et al., 1989). More serious competitive issuesrise in locales like consortia and other forms of cooperative R&Delationships, which imply close collaboration and, hence, a morepen exchange of knowledge.

An important consequence of participating in R&D consortia likeEMATECH is the restructuring of competitive and pre-competitivereas within and across companies. Issues addressed in coopera-ion with competitors have to be reframed as pre-competitive, ateast in particular subsystems of the organisation or the network,ven if they were previously considered competitive. Matters ofntellectual property and trade secrets have to be tailored to eachollaborative project as they generally concern cooperation amongompetitors, which remains fierce.10 This is supported by the vir-ual absence of shared patents among the tool manufacturers. Inhe US patent database, there are only two closely related patentsointly held by Nikon and Canon (Patent Number: US007312851nd US007483122) that both concern EUV lithography; there areo patents shared by either of the two firms with ASML. Withinonsortia, competition remains a central issue that has do be dealtith in the programmes which are set up (Thornberry, 2002).11

Even though collaboration among competitors occurs, this iso trivial matter (cf. Yami et al., 2010). Whereas the extension ofptical lithography is largely based on existing buy–sell supplierelationships, the creation of a new technological path requiresarge collaborative investments. This turns technological chal-enges into infrastructural challenges, as new supply agreementseed to be set up. In case of creating the EUV supply structure, Intelstimates that only about 30% of supply agreements are buy–sellompared to about 90% in case of extending 193 nm lithographyGolda and Philippi, 2007). The difficulties of managing the delicatessues of intellectual property and competition are reflected in thetructure and project organisation of the consortia. For instance, ofhe three tool manufacturers, Canon, Nikon and ASML, only ASMLs a member of SEMATECH’s lithography programme. However, for
pecific purposes, SEMTAECH will collaborate in a dyadic allianceith Canon, as in the case of the 2009 partnership with Canon’s
ubsidiary Canon ANELVA in the front end processes programme.

10 One supply company representative summarised it as follows: “I would nevero that [collaborate on intellectual property] with a competitor. [. . .] We work withhree direct competitors to create a common logistic infrastructure. This benefits usll and on the other hand it is not beneficial, if we would not do it.” (I 22: 49).11 One test facility representative pointed out the careful manoeuvrings along theoundary between competitive and pre-competitive issues: “[. . .] because of theact that we are focussing on technology that is not yet commercialised, we can getll these companies also to work together. If it would be too close to what they arelready starting to ramp up in production in their own companies, then of coursehey would not . . . they would refuse to work together here, because it would be too

uch competition between them. This is research and development that is basicallyre-competitive in this phase, so as long as we can do that, all the companies benefitrom all the progress that is made and they do not care about the fact, that there areheir own competitors are working in the same room basically, day in day out” (I7: 15).

icy 42 (2013) 1389– 1405

Similarly, ASML and Canon ANELVA are members of IMEC’s coreCMOS programme, but Canon ANELVA is not a competitor to ASML.The consortia’s research programme structures are thus aimed atavoiding direct competition. IMEC is also continuously adapting itsintellectual property policy in order to afford open “IMEC indus-trial affiliation programmes” and closed “bilateral collaborations”.In 2008, IMEC started a new programme called CMORE, which con-nects IMEC bilaterally with industry partners in the developmentand prototyping stages. In earlier stages, it provides collaborativeprogrammes such as the core CMOS programme.

If we think of cooperative ventures as means of technologydevelopment, their transformative agency lies in the mutual for-matting of technologies and organisations. The consortia thusmanage the boundary between pre-competitive and competitiveissues and they organise the transformation from pre-competitiveto competitive issues of technology development. They also helpto set up shared selection criteria by which to judge technologicaloptions. And, last not least, they are the locales where concrete R&Dis carried out. As technologies become globally coordinated, so dothe activities of companies. We find tightly knit constellations oftechnologies, projects, and programmes, which emerge over timeand are constantly monitored and adjusted. We also find a mul-titude of intermediate agencies at work to connect technologiesand organisations into such constellations. From a structurationperspective, these changes in the relations of technologies andorganisations must be understood as the temporal unfolding ofinterdependent rules and resources organised within the semi-conductor industry. Even though individual cooperative venturesmay only last for a few years, Fig. 5 shows the setup of numer-ous cooperative NGL research programmes around the last turnof the century. Typically, these programmes are mainly concernedwith pre-competitive technological issues, but they also closelyassociate technologies and organisations in the cooperative R&Dpractices which outlast the programmes themselves.

The management of momentum not only involves coping withthe techno-economic conditions, but also mindfully embeddingthem in existing cultural and political practices of innovation, aswell as creating new practices and locales to address emergingambiguities and uncertainties. In general, we can see how thecollective control of allocative resources for SMT is founded onthe organisation of material, e.g. specialised supplies and money,as well as immaterial means, e.g. knowledge and collaborativeresearch ventures. Even though initially contested and still far frombeing unproblematic today, cooperative R&D has become the legit-imate form of innovating novel manufacturing technologies andplays an important role in signifying important developments.

5. Conclusions and directions

In this paper, we have argued that the means of managingmomentum with regard to mass, velocity and direction can beseen as bridging a gap between technological and organisationalchange in science-based industries. In the field of semiconductormanufacturing, technological momentum, i.e. the cumulative pro-cess which leads to the mutual stabilisation of technological pathsand organisational fields, does not emerge out of undirected inter-action between technologies on the one hand and organisations onthe other, but reflects a highly managed and reflexively mediatedprocess. In particular, we have highlighted how, in the field underscrutiny, technological progress is evaluated and pursued at thelevel of the organisational field, and how the representational and
cooperative means for choosing between technological options andfor developing technologies shape the field, while they themselvesare being shaped in return. In particular, we have paid attentionto the emerging dynamics and locales of generating consensus and

C. Schubert et al. / Research Policy 42 (2013) 1389– 1405 1403

n len

cn

orlaiist

itpG2vpWtt

morsttp

nccthtpTfbhcwn

iG

Fig. 6. Means of managing momentum through a structuratio

ommitment for both extending an existing, as well as creating aew technological path.

Let us summarise our main arguments. First, we used the notionf technological momentum to account for the increasing andeciprocal stabilisation of organisational arrangements and techno-ogical systems into durable configurations that are able to directttention and resources towards common goals. In addition, wentroduced the concept of intermittent lock-in to highlight thatndustries like the semiconductor industry strive for the positiveides of path dependence and maybe a temporal lock-in. However,hey are very eager not to get trapped into an irreversible lock-in.

Second, we departed from evolutionary understandings foundn the concepts of technological momentum (Hughes, 1987) andechnological paths (David, 1985) by stressing the collective andurposeful selection of technological options within the field. Likearud and his colleagues (Garud and Karnøe, 2001; Garud et al.,010), we emphasised the interpretative or social-constructivistiew of this process but adopted a more gradual understanding ofath constitution (Meyer and Schubert, 2007; Sydow et al., 2012;indeler, 2003) characterised by the interplay of reproduction and

ransformation in the management of momentum based on theheory of structuration (Giddens, 1984).

Third, we focussed on several concrete means of managingomentum in the constitution of new technological paths and

rganisational arrangements, i.e. on the transformative agency ofepresentational and cooperative means, such as conferences, con-ortia, and roadmaps. Before we revisit the details, Fig. 6 sums uphe reciprocal relations of the means of managing momentum withhe social practices of technology development from a structurationerspective.

Whereas transformations in technology development and orga-isational arrangements were mainly considered to affect theontrol over material and immaterial resources by new means ofooperation (in particular combining or pooling resource), we seehat in the case of science-based fields, and under conditions ofigh uncertainty, the means of mutually monitoring and format-ing the expectations and shaping the progress of technologicalaths within an organisational field become increasingly relevant.his may be true for any technology, but it is even more importantor technological paths that are actively and strategically createdecause the actors in the field are careful not to “bet on the wrongorse” while hoping to “jump on the right train”. For our empiri-al case we can distinguish three transformative aspects of meanshich play a significant role in simultaneously constituting tech-
ological paths and organisational fields:
The first transformative aspect of the means is that they werentroduced to collectively monitor technical progress. In line withiddens, we see the collective actions in the field as being based on

s – the example of the field of semiconductor manufacturing.

the mutual monitoring and collective rationalisation of actions, aswell as the conditions and outcomes of actions. In this respect, wedraw attention to the means of mutually and reflexively monitoringand rationalising technological progress. Especially in science-based fields, such activities must always be seen as some formof collectively organised evaluation, since the technologies are inexperimental stages and their technical and economic value canonly be estimated. At this stage, the shared criteria of selectionwithin the field decide which technology will draw enough invest-ments to be turned into a manufacturable solution. In this way,the means also contribute to the formatting of the technologies interms of technical feasibly and economic viability, in essence fram-ing them according to the rules of signification and legitimation inthe field. In the end, it is the framing and formatting of technolo-gies that enables the management of momentum in the first place,as this makes options comparable and opens a space for mutuallymonitoring the competition and collectively selecting a promisingoption.

The second transformative aspect of the means is that collec-tive means of mutual monitoring and rationalising technologicalprogress format not only technologies, but also the organisationalfield. As we have seen in our case, the new means did not sim-ply diffuse into the field, but were subject to a complex process ofmodification and adaptation. The rules of signification and legit-imation, as well as the resources of domination, slowly changedas the industry converged towards cooperative models of R&Dand as the collective means – conferences, roadmaps, networks –became taken for granted. For the means to be institutionalisedas effective, there must be a shared recognition of their benefits.As the discussion of the ITRS showed, the means need not neces-sarily produce valid results in order to be effective. The value ofthe roadmap lies not in accurate predictions, but in the creationand maintenance of multiple locales of interaction and coopera-tion, which in turn contribute to a manageable level of uncertainty.The means of monitoring technological progress thus transform the(inter)organisational structures of the field itself.

The third transformative aspect of the means concerns the deepinterrelation of technological paths and organisational fields. Asmeans of managing momentum, the conferences, roadmaps, andconsortia facilitate the control of material and immaterial resourcesat the same time. Allocative resources can be used by agents tocontrol authoritative resources and vice versa. Publicly announcedinvestments in a specific technological option on the part of rele-vant actors, for instance, can generate a larger commitment within
the industry. In a similar vein, the prominent placement of a par-ticular technology on the ITRS may result in directing funds (bothpublic and private) towards this technology. The actors in the fieldare aware of the transformative capacity of the means they use

1 ch Pol

aoifaa

ieriosrwotammiabitndlfDqtmsbt

F

s

R

A

A

A

A

B

B

B

B

B

B

B

B


ccordingly. Thus, the ITRS, for instance, sets the pace of devel-pments to which all organisations willing to participate in thennovation of novel technologies have to relate. Even though itsorecasts of technical progress might often be wrong, the ITRS –s a process – is still a powerful means of organising collectivectivities.

As we have shown, the management of momentum happensn many different places and on many different occasions and lev-ls – and should be discussed even more systematically in theseespects than was possible in our study. Careful comparison of thenstitutionalisation of organisational forms in this industry withther science-based fields could lead to a more precise under-tanding of case-specific industry peculiarities, as well as to moreesilient insights – and thus to an understanding of the extent tohich these findings can be generalised. Our particular focus was

n the development of complex manufacturing technologies. Here,he means of managing momentum aim at collectively identifyingnd developing a technologically feasible and economically viableanufacturing solution. Managing momentum thus entails theonitoring and rationalisation of the collective ventures, organ-

sing programmes in order to push certain options while othersre discontinued, as well as carefully regulating the boundaryetween competitive and pre-competitive issues. When compar-

ng our fieldwork context to other science-based fields, we mustherefore pay careful attention to the specific dynamics of tech-ology development and organisational change. We should alsoistinguish between fields which endogenously generate techno-

ogical innovations, like semiconductor manufacturing, and thoseor which innovation is mostly an external event (cf. for instance,olata, 2009). This is a necessary next step, since addressing theuestion of how momentum can be managed collectively and howechnological paths and organisational fields are mediated by the

eans of managing momentum adds new insights to the discus-ion of competing technologies, not only in science-based fields,ut also for a broader analysis of the interdependencies betweenechnical and social dynamics.

unding

The research presented in this paper was funded by the Volk-wagen Foundation under Grant number AZ II/80308.

eferences

bernathy, W.J., Utterback, J.M., 1978. Patterns of innovation in technology. Tech-nology Review 80, 40–47.

nderson, P., Tushman, M.L., 1990. Technological discontinuities and dominantdesigns. A cyclical model of technological change. Administrative Science Quar-terly 35, 604–633.

rthur, W.B., 1989. Competing technologies, increasing returns, and lock-in byhistorical events. Economic Journal 99, 116–131.

rthur, W.B., 1994. Increasing Returns and Path Dependence in the Economy. Uni-versity of Michigan Press, Ann Arbor.

each, J., Satyanarayana, S., Ford, A., Bakshi, V., 2001. Fine-tuning the process. Thestatus of 193 nm technology. Solid State Technology 44, 54–60.

arley, S.R., 1990. Images of imaging. Notes on doing longitudinal field work. Orga-nization Science 1, 220–247.

arley, S.R., Freeman, J., Hybels, R.C., 1992. Strategic alliances in commercialbiotechnology. In: Nohria, N., Eccles, R.G. (Eds.), Networks and Organiza-tions. Structure, Form and Action. Harvard Business School Press, Boston,pp. 311–347.

erger, P.L., Luckmann, T., 1966. The Social Construction of Reality: A Treatise in theSociology of Knowledge. Doubleday, Garden City.

ijker, W.E., Hughes, T.P., Pinch, T.J. (Eds.), 1987. The Social Construction of Tech-nological Systems. MIT Press, Cambridge.

owker, G.C., Star, S.L., 1999. Sorting Things Out. Classification and its Consequences.
MIT Press, Cambridge.
rainard, R.L., Barclay, G.G., Anderson, E.H., Ocol, L.E., 2002. Resists for next gener-ation lithography. Microelectronic Engineering 61/62, 707–715.

rowning, L.D., Shetler, J.C., 2000. SEMATECH. Saving the U.S. Semiconductor Indus-try. Texas A&M University Press, College Station.

icy 42 (2013) 1389– 1405

Callon, M., 1998. Introduction: the embeddedness of economic markets in eco-nomics. In: Callon, M. (Ed.), The Laws of the Market. Blackwell, Oxford, pp.1–57.

Callon, M., Millo, Y., Muniesa, F. (Eds.), 2007. Market Devices. Blackwell, Malden.Carayannis, E.G., Alexander, J., 2004. Strategy, structure and performance issues of

pre-competitive R&D consortia. Insights and lessons learned from SEMATECH.IEEE Transactions on Engineering Management 51, 226–232.

Chandler, A.D., 1977. The Visible Hand. The Managerial Revolution in AmericanBusiness. Harvard University Press, Cambridge.

Chiu, G.L.-T., Shaw, J.M., 1997. Optical lithography. Introduction. IBM Journal ofResearch and Development 1/2, 3–5.

Chuma, H., 2006. Increasing complexity and limits of organization in themicrolithography industry. Implications for science-based industries. ResearchPolicy 35, 394–411.

David, P.A., 1985. Clio and the economics of QWERTY. American Economic Review75, 332–337.

David, P.A., 1987. Some new standards for the economics of standardization in theinformation age. In: Dasgupta, P., Stoneman, P.L. (Eds.), Economic Policy andTechnology Performance. Cambridge University Press, Cambridge, pp. 206–234.

Dewey, J., 1958. Experience and Nature. Mineola, Dover.DiMaggio, P.J., Powell, W.W., 1983. The iron cage revisited: institutional isomor-

phism and collective rationality in organizational fields. American SociologicalReview 48, 147–160.

Dokko, G., Nigam, A., Rosenkopf, L., 2012. Keeping steady as she goes: a negotiatedorder perspective on technological evolution. Organization Studies 33, 681–703.

Dolata, U., 2009. Technological innovations and sectoral change. Transformativecapacity, adaptability, patterns of change: an analytical framework. ResearchPolicy 38, 1066–1076.

Fong, G.R., 1990. State strength, industry structure, and industrial policy. Amer-ican and Japanese experiences in microelectronics. Comparative Politics 22,273–299.

Garud, R., Karnøe, P., 2001. Path creation as a process of mindful deviation. In:Garud, R., Karnøe, P. (Eds.), Path Dependence and Creation. Lawrence ErlbaumAssociates, Mahwah, pp. 1–38.

Garud, R., Kumaraswamy, A., Karnøe, P., 2010. Path dependence or path creation?Journal of Management Studies 47, 760–774.

Giddens, A., 1979. Central Problems in Social Theory. Action, Structure and Contra-diction in Social Analysis. Macmillan, London.

Giddens, A., 1984. The Constitution of Society. Outline of the Theory of Structuration.University of California Press, Berkeley.

Gnyawali, D.R., Park, B.-J.R., 2011. Co-opetition between giants. Collaboration withcompetitors for technological innovation. Research Policy 40, 650–663.

Golda, J., Philippi, C., 2007. Managing new technology risks in the supply chain. IntelTechnology Journal 11, 95–104.

Gomes-Casseres, B., 1996. The Alliance Revolution: The New Shape of BusinessRivalry. Harvard University, Cambridge, MA.

Grindley, P., Mowery, D.C., Silverman, B., 1996. The design of high-technologyconsortia: lessons from SEMATECH. In: Teubal, M., Foray, D., Justman, M.,Zuscovitch, E. (Eds.), Technological Infrastructure Policy: An International Per-spective. Kluwer, Amsterdam, pp. 173–211.

Ham, R.M., Linden, G., Appleyard, M.M., 1998. The evolving role of semiconductorconsortia in the United States and Japan. California Management Review 40,137–163.

Henderson, R., 1995. Of life cycles real and imaginary. The unexpected long old ageof optical lithography. Research Policy 24, 631–643.

Hughes, E.C., 1971. The Sociological Eye. Selected Papers. Aldine, Chicago.Hughes, T.P., 1983. Networks of Power: Electrification in Western Society,

1880–1930. Johns Hopkins University Press, Baltimore.Hughes, T.P., 1987. The evolution of large technological systems. In: Bijker, W.E.,

Hughes, T.P., Pinch, T.J. (Eds.), The Social Construction of Technological Systems.MIT Press, Cambridge, pp. 51–82.

Hughes, T.P., 1994. Technological momentum. In: Smith, M.R., Marx, L. (Eds.), DoesTechnology Drive History? The Dilemma of Technological Determinism. MITPress, Cambridge, pp. 101–113.

IEEE spectrum, 2004. http://spectrum.ieee.org/semiconductors/memory/chip-makings-wet-new-world

ITRS, 2001. Lithography. International Technology Roadmap for Semiconductors,Austin.

ITRS, 2011. Executive Summary. International Technology Roadmap for Semi-conductors, Austin.

Jansen, K.J., 2004. From persistence to pursuit: a longitudinal examination ofmomentum during the early stages of strategic change. Organization Science15 (3), 276–294.

Kaplan, S., Tripsas, M., 2008. Thinking about technology: applying a cognitive lensto technical change. Research Policy 37, 790–805.

Lampel, J., Meyer, A.D., 2008. Field-configuring events as structuring mech-anisms. How conferences, ceremonies, and trade shows constitute newtechnologies, industries, and markets. Journal of Management Studies 45,1025–1035.

Langlois, R.N., 2000. Capabilities and vertical disintegration in process tech-nology. The case of semiconductor fabrication equipment. In: Foss, N.J.,
Robertson, P.L. (Eds.), Resources, Technology, and Strategy. Routledge, London,pp. 199–206.
Latour, B., 1986. The powers of association. In: Law, J. (Ed.), Power, Action and Belief:A New Sociology of Knowledge? Routledge, London, pp. 264–280.

Levinson, H., 2005. Principles of Lithography. SPIE Press, Bellingham.

http://refhub.elsevier.com/S0048-7333(13)00081-4/sbref0005





















































































































































































































































































































































































































































































































































































































































































http://spectrum.ieee.org/semiconductors/memory/chip-makings-wet-new-world

http://spectrum.ieee.org/semiconductors/memory/chip-makings-wet-new-world













































































































































ch Poli

L

M

M

M

M

M

M

M

M

M

M

M

N

N

O

P

P

P


inden, G., Mowery, D.C., Ham Ziedonis, R., 2000. National technology policy inglobal markets: developing next-generation lithography in the semiconductorindustry. Business and Politics 2, 93–113.

acKenzie, D., 2006. An Engine, Not a Camera. How Financial Models Shape Markets.MIT Press, Cambridge.

ahoney, J., Thelen, K., 2010. A theory of gradual institutional change. In: Mahoney,J., Thelen, K. (Eds.), Explaining Institutional Change. Ambiguity, Agency, andPower. Cambridge University Press, Cambridge, pp. 1–38.

alerba, F., Nelson, R.R., Orsenigo, L., Winter, S., 2008. Vertical integration anddisintegration of computer firms. A history-friendly model of the coevolution ofthe computer and semiconductor industries. Industrial and Corporate Change17, 197–231.

artin, R., Sunley, P., 2006. Path dependence and regional economic evolution.Journal of Economic Geography 6 (4), 395–437.

erton, R.K., 1948. The self-fulfilling prophecy. Antioch Review 8 (2),193–210.

iller, P., 2001. Governing by numbers. Why calculative practices matter. SocialResearch 68 (2), 379–396.

eyer, U., Schubert, C., 2007. Integrating path dependency and path creation in ageneral understanding of path constitution. The role of agency and institutionsin the shaping of technological innovations. Science, Technology & InnovationStudies 3, 23–44.

iller, P., O’Leary, T., 2007. Mediating instruments and making markets. Capitalbudgeting, science and the economy. Accounting, Organizations and Society 32,701–734.

oore, G.E., 1965. Cramming more components onto integrated circuits. Electronics38, 114–117.

owery, D.C., Rosenberg, N., 1998. Paths of Innovation. Technological Change in20th-century America. Cambridge University Press, Cambridge.

üller-Seitz, G., 2012. Absorptive and desorptive capacity-related practices at thenetwork level – the case of SEMATECH. R&D Management 42 (1), 90–99.

elson, R.R., Winter, S., 1977. In search of useful theory of innovation. ResearchPolicy 6, 36–76.

TRS, 1994. National Technology Roadmap for Institution. Semiconductor IndustryAssociation, Washington.

rlikowski, W.J., 1992. The duality of technology. Rethinking the concept of tech-nology in organizations. Organization Science 3 (3), 398–427.

inch, T.J., Bijker, W.E., 1984. The social construction of facts and artefacts. Or howthe sociology of science and the sociology of technology might benefit eachother. Social Studies of Science 14, 399–441.

isano, G.P., 2010. The evolution of science-based business. Innovating how weinnovate. Industrial and Corporate Change 19, 465–482.

orac, J.F., Thomas, H., Baden-Fuller, C., 1989. Competitive groups as cognitive com-munities: the case of Scottish knitwear manufacturers. Journal of ManagementStudies 26 (4), 397–416.

cy 42 (2013) 1389– 1405 1405

Power, M., 2004. Counting, control and calculation. Reflections on measuring andmanagement. Human Relations 57, 765–783.

Pyke, D.L., 1971. Mapping – a system concept for displaying alternatives. Techno-logical Forecasting and Social Change 2, 311–320.

Rosenberg, N., 1963. Technological change in the machine tool industry, 1840–1910.The Journal of Economic History 23, 414–443.

Rosenkopf, L., Metiu, A., George, V.P., 2001. From the bottom up? Technical com-mittee activity and alliance formation. Administrative Science Quarterly 46,748–772.

Schaller, R.R., 2004. Technological innovation in the semiconductor industry. A casestudy of the international technology roadmap for semiconductors. Dissertation,George Mason University, Fairfax.

Schumpeter, J.A., 1912. Theorie der wirtschaftlichen Entwicklung. Duncker & Hum-blot, Leipzig.

Spencer, W.J., Seidel, T.E., 1995. National technology roadmaps: the U.S. semi-conductor experience. In: Proceedings of the 4th International Conference onSolid-State and IC Technology, pp. 211–220.

Streeck, W., Thelen, K.A., 2005. Introduction. Institutional change in advancedpolitical economies. In: W. Streeck, W., Thelen, K. (Eds.), Beyond Continuity.Institutional Change in Advanced Political Economies. Oxford University Press,Oxford, pp. 3–39.

Suzuki, K., Smith, B.W. (Eds.), 2007. Microlithography. Science and Technology. CRCPress, Boca Raton.

Sydow, J., Windeler, A., Schubert, C., Möllering, G., 2012. Organizing R&D consor-tia for path creation and extension. The case of semiconductor manufacturingtechnologies. Organization Studies 33, 907–936.

Thornberry, J.B., 2002. Competition and cooperation. A comparative analy-sis of SEMATECH and the VLSI research project. Enterprise and Society 3,657–686.

Trist, E.L., Bamforth, K.W., 1951. Some social and psychological consequences of thelongwall method of coal-getting. Human Relations 4, 3–38.

Tushman, M.L., Rosenkopf, L., 1992. Organizational determinants of technologicalchange. Towards a sociology of technological evolution. Research in Organiza-tional Behavior 14, 311–347.

Wagner, C., Harned, N., 2010. EUV lithography. Lithography gets extreme. NaturePhotonics 4, 24–26.

Weber, M., 1988. Gesammelte Aufsätze zur Religionssoziologie I. Mohr Siebeck,Tübingen.

West, J., Iansiti, M., 2003. Experience, experimentation, and the accumulation ofknowledge. The evolution of R&D in the semiconductor industry. Research Policy
32, 809–825.
Windeler, A., 2003. Kreation technologischer Pfade: ein strukturationstheoretischerAnalyseansatz. Managementforschung 13, 295–328.

Yami, S., Castaldo, S., Dagnino, G.B., Le Roy, F. (Eds.), 2010. Coopetition – WinningStrategies for the 21st Century. Edward Elgar, Cheltenham.




























































































































































































































































































































































































































































































































































































































































Documents

The means of managing momentum: Bridging technological paths and organisational fields