clusters nanoelec

8/6/2019 Clusters Nanoelec

1/21

Synthetic presentation of the major clusters in nanoelectronics

Charles Collet, GAEL,October 07


2/21

Introduction

Nanoelectronics can be considered as anexception in nanosciences :its indeed a nanotechnology (producing

devices at the nanoscale), but its top downapproach, as a continuation of themicroelectronics roadmap, doesnt constitue

-yet- a breakthrough as it doesnt bringchanges in physical properties.Its a nanotechnology but not a

nanoscience, yet.


3/21

Comparing & positioning nanoelectronics clustersis a hard task due to heterogeneity of their activities.

Indeed the current split of their business models(continuation of miniaturization, diversification onnew functionalities) makes it for example

irrelevant to compare nb of employees (front vsback end) or investment.

The only criterion that is however common togrowing clusters seems to be the excellence of their applied researchWe can thus present the repartition of the

industry, major clusters in their position on thevalue chain.


4/21

g o a wor : a nee or nnova on versus a push for production relocation.

2 axis in Business Model Change:Digital versus Physical Value Added(design vs production)Modular versus Integral

architecture: miniaturized &standardized chips (More Moore) vsminiaturized & specific chips (More

than Moore)


5/21

Physical valuecreation

Modularproduction

architecture

Integralproduction

architecture

1

2

3

4

Digital value creation


6/21

1) digital & modular: value added is ondesign, & the architecture design is modular

IP providers designing the processcores

or ASICs(like ARM, MIPS)IP providers conceive generic chips coresaround which other firms (fabless or IDMs)

will design some extended architectures for specific applications.coordination costs are low (IP providers justhave to send the digital architecture to other fabless or IDMs), & the IP can beunderstood by any manufacturer


7/21

2) physical & modular : mass production of generic chips (high volume/low unit value)

the pure play foundries Business Model for

chips & memories (TSMC, UMC, Qimonda, etc)VA is indeed physical (chip production that involves a high capital intensity), & the production

model is modular, as these firms produce chips that will be sold to different customers & integrate a large range of products.R&D costs are heavily supported bymanufacturers as their competitive advantagerelies on the ability to keep on with miniaturization(More Moore)


8/21

3) digital & integral : Value on design, specificarchitectures & new functionalities. the Fabless Business Model (Broadcom,Qualcomm, IBM, Freescale, etc)

value added is even more digital & knowledge-intensive, as these design companies conceivespecific chips for specific clients, which will beproduced by foundries.chips are specific & cant be applied to different

products, thus taking place in an integralproduction strategy.


9/21

4) physical & specific : complex systems producedby IDMs & integrative clusters: the Integrated Business Model (TI, Samsung,STM, Grenoble, Dresde, Albany clusters, )etc

VA is on design & also on production of thesespecific chips like NEMS (Nano ElectronicsMechanical Systems), as the complexity of the chipmakes production as knowledge intensive as thedesign.

labs-on-chip, Embedded System on Chip, SecuredSolutions, NEMS, etc.


10/21

Valeur plutt physique

Production de

pucesstandards / modulaires(VA faible, production de

masse)

Production

de pucesspcialisesou intgres(VA forte, prod deligneslimites)

1-IP providers quiconoivent un cur de

process: ARM, MIPS, etc

2- producteurs de masse

More Moore :Dresde(Qimonda,),&Taiwan(fonderies),fabricants de mmoires :Epida, Powerchip, etc.

3- designers fabless :Qualcomm, Broadcom, etc, &

orientation prise par NXP, FSLpour les gnrations < 45nm

4- entreprises intgratrices :ST, TI, Atmel, etc& clusters produisant dessolutions intgres, labos surpuce, etc: Grenoble,Eindhoven Alban .

Valeur plutt digitale


11/21

Physique,architectur

esmodulaires

Digitale,architectur

esmodulaire

s &spcialis

Digitale &architectur

esspcialis

es

Physique,architectu

resmodulaire

s &spcialis

Digitale &architecturesspcialises

Type of Value

Added

Production demasse

Rechercheinternationale &intgration

multidisciplinaire

Rechercheinternation

ale &design

diversifi

Recherchelocale &

production de

masse(processe

urs,mmoires

)

Recherche localeforte &intgrationmultidisciplin

aire(Minatec)

Business Model&

Positionnin

g

76 00050 00025 00020 00020 000Nombre

demploys

ITRICNSEIMECCNTCEA LtiResearch Lab

TSMC, UMC

IBM, Samsung,AMD,

Micron,Infineon,

Chartered.

NXPInfineon, AMDSTMAnchor tenant

firm

TaiwanAlbanyLouvain/EindhovenDresdeGrenoble

Structure of the clusters around a double anchor


12/21

epar on o pro uc oncapacities ownership,nanoelectronics.

distribution de la proprit des capacits deproduction de semiconducteurs en 2005 (%)

0

10

20

30

40

50

USA japon europe taiwan Core duSud

Chine


13/21

epar on o pro uc oncapabilities locations,nanoelectronics.

Distribution gographique des fabs mondiales en2005.

0%

5%

10%

15%

20%

25%

30%

USA japon europe taiwan Core duSud

Chine


14/21

Evolution of the nationalproduction produced inshore

Evolution de la part de production nationaleeffectue sur son sol, 1998 - 2003.

0%

20%

40%

60%

80%

100%

USA Europe Japon AsiePacifique

MoyenneMonde

1998

2003


15/21

Number of dedicated researchers &

operational workers in nanoelectronicsclusters, ex of Grenoble.

0 50000 100000 150000

HsinchuTsukuba

Seoul

Silicon ValleyOregonArizona

New-YorkM assachusset t s

TexasPek inShanghai

GrenobleCataniaDresde

0 5000 10000 15000

Production

Chercheurs

Taille comptitive

Le ple Grenoblois a une taillecomptitive en matire derecherche

il lui faut encore acqurirla dimension de productionindustrielle adquate

0 50000 100000 150000

HsinchuTsukuba

Seoul

Silicon ValleyOregonArizona

New-YorkM assachusset t s

TexasPek inShanghai

GrenobleCataniaDresde

0 5000 10000 15000

Production

Chercheurs

Taille comptitive

Le ple Grenoblois a une taillecomptitive en matire derecherche

il lui faut encore acqurirla dimension de productionindustrielle adquate


16/21

Tsukuba

Hsinchu

Japan

Taiwan

USA

OregonCalifornia

Texas

New-York

Seoul

Korea

Fab 300 mm Fabs

Labo Public

Chinea

Shanghai

Beijing

Portland

AustinDallas

AlbanyEast Fishkill

Competitiors of

European clusters


17/21

From Academia to Market, ex of Grenoble

Academic Research Advanced Research Manufacturing

Process IntegrationModule Development

Advanced Modules

Dan Noble CenterDan Noble Center

IMECIMEC

LETILETI

45nm Device Architecture

45nm Device Architecture

StressorTechnologies

StressorTechnologies

Gate Stack (high-k, M-

Gate)

Gate Stack (high-k, M-

Gate)

BEOL Materials& Integration

BEOL Materials& Integration

ImmersionLithography

ImmersionLithography

Materialsand Advanced

Modules

Materialsand

AdvancedModules

Crolles Narrowed Options

45nm DesignRules Inputs

45nm DesignRules Inputs

E-beam DirectWrite Litho

E-beam DirectWrite Litho


18/21

Conclusion: Successful clusters in high cost regions rely ondifferentiating themselves from their competition

Europe may be loosing the cost war for low value-adding industries,but not the same for high tech & capital-intensive sectors such as

nanoelectronics, as in this high tech industry the cost of equipment(clean rooms, lithography tools) & workforce (engineers) becamesimilar globally.however the differences between territories mainly rely on statesubsidies & tax regimes. Taiwan or China wouldnt have been able tobuild their competitive foundries without a public support being higher than the one granted in Europe & US together for the same timeperiod (about 4 billion $ between 2002-2007). That shows that with asame amount of subsidies & a same tax rate Europeannanoelectronics clusters like Dresden or Grenoble would surely be ascompetitive as Taiwan, all things remaining equal.This statement goes against the common acceptation saying that if their ability to make a good becomes ubiquitous, the competitiveadvantage of European clusters should shift from the production lineto management strategies, innovation, R&D, or marketing (Andersen,2005).But what we do agree with, is that without the same national or environmental conditions the challenge for western clusters is toreinvent themselves in ways that keep a level of local employment, asit will be hard for them to compete on production with places that have


19/21

Most strategies have looked to the universities as

the source of new and distinguishing innovations, &most cluster analyses include rates of patents andpublications as evidence of innovation.But for these indicators Taiwanese cluster again doas good as European ones. The difference betweenAsian & western clusters is the fact that they havebeen built on a longer tradition & already developedsecondary competencies, like biotechnology inGrenoble, Dresden or Albany-Boston, which now

can really become a source of advantage in theconvergence enabled at the nanoscale.Creative Centers, Richard Florida writes, tend tobe the economic winners of our age. These creativecenters have the attributesphysical, diversity, andexperiencesto attract what he defines as thecreative class. & its up to this creative class now todesign complex & integrated systems so as to swiftthe value from low cost to high performance, &

diversify on new functionalities so as to create newvalue.


20/21

Bridgingdisciplines &competencies to

develop new & highvalue addingsolutions

Designing betterchips

Integrative firms &clusters

Miniaturizedarchitecture, lowconsumption &/or

high performance

Designing chipsbetter

Fabless / Designfirms

low cost thanks tomass production,investment inminiaturization

Making chipssmaller & cheaper

Foundries

AdvantageGoalActor

Table: competitive advantage of the micro/nanoelectronics actors, CharlesCollet, 2007.


21/21

Those Golden Days of Scaling:Pricetrend Baseline CMOS

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.10

123

45

67

89

101112

13

1415

Min = 2/3 * Average

Source : Gartner/Dataquest 2004

Q298Q299Q200Q201Q202Q203Q403Q204

a v e r a g e

S i w a f e r

P r i c e

( $ / c m

2 )

Node (um)LTPS : 1-1.5 $ /cm2

Price Trend (time) at fixed Node

@ 0.12 um 200 mm > 300mm +7%

@ 0.5 um : 150 mm > 200mm 20%

Golden Decades of Scaling 2 Functionality + Clockspeed @ equal $/cm2

2ML, 13 masks

7LM, 36 masks

Source: [2005] Carel van der Poel, Philips Research

clusters nanoelec

Documents