Philips Quality Reference Handbook 2002.pdf

1

ContentsIntroduction

Part 1 GeneralOrganized for qualityBusiness ExcellenceBusiness excellence policyProduct/Manufacturing centresSeparate units, joint venturesand subcontractorsQuality/Improvement managersPart 2 ReferencesAbbreviationsAcceleration factorsAcceptable Quality Level (AQL)Assembly quality controlBEST (Business Excellence through Speed

and Teamwork)Chemical content of semiconductorsComplaint processing / CHAMPConfidence levelCost of qualityCustomer Notification / CPCN / DODCustomer-specific labelingDrypackElectromagnetic Compatibility (EMC)Electrostatic Discharge (ESD) Environmental careEnvironmental policy and goalsEvolution of qualityExternal standardizationFailure analysisFailure Modes and Effects Analysis (FMEA)Failures In Time Standard (FITS)General quality specificationsGreen Flagship productsHistory of Philips SemiconductorsIdentification labelingISO 9000Marking of ICsMean Time Between Failures (MTBF)Moisture sensitivity level (MSL)Philips Business Excellence (PBE)Philips Business Excellence policyPhilips QualityPhilips Values

PPMPQA-90Product manufacturing codesProduct Quality & Reliability Assurance database (PQRA)QS-9000 / automotive quality standardsQuality Function Deployment (QFD)Quality Improvement Competition (QIC)Quality standards for customersQuality techniques and tools

- PDCA cycle- Problem solving- 8-D method for team-oriented

problem solving- Fact gathering- Checksheets- Brainstorming- Pareto analysis- Fishbone diagram- Histogram- Control chart- Scatter diagram- Paynter chart- Spider graph (Radar chart)

Quality testingRelease of new productsReliabilityReturn shipmentsSampling on the FlySelf-qualificationSemiconductor Assembly Council (SAC)Six SigmaSoftware QualityStatistical Process Control (SPC)Structural similaritySupplier quality systemThermal resistanceTotal Quality Excellence (TQE)Traceability / ROOTSVision & Mission statementWafer level reliabilityWeibullPart 3 KeywordsKeywordsIndex

2

34

111214

41444950535658

5960616466686970727380828488909193959698

103105107108109110112113114

115117120

121123125128130131131131

133134134135136137138139140142143144146148150152154155156158160166167168169171173174176181182193

2

Introduction

This handbook gives detailed information forcustomers and sales- and other employees onquality matters for Philips Semiconductors. Qualityreference subjects are presented in easy-to-understand terms, suitable for the layman andexpert alike. The handbook is organized in threeparts plus an index:

Part 1 General, describing the QualityOrganization of Philips Semiconductors andsetting out Business Excellence. This part alsoincludes details of each Product Centre.

Part 2 References, giving a list of abbreviationsused in the handbook and a series of qualityreference subjects presented in alphabeticalorder.

Part 3 Keywords, listing definitions of qualityreferences not featured in Part 2, presented inalphabetical order.

Index, a general index is provided at the end of the handbook.

3

Part 1

General

4

Product Division Philips Semiconductors (PSC) isone of a group of product divisions reportingdirectly to the Board of Management of RoyalPhilips Electronics.Within Philips Semiconductors there are staffdepartments, such as Financial and LegalDepartments, and nine Business Units (BU).The manager of each Business Unit is responsiblefor the profitability of the allocated business lines,and also formulates worldwide product policy.Each Business Unit consists of several BusinessLines (BL).Many products have common needs for scarceand costly resources, for example in productdevelopment, manufacturing and sales/marketing.Therefore, the business organization of PSCincludes the four shared business functions whichhave the status of Accountable Units (AU): Foundries Assembly & Test International Marketing & Sales Technology.

Organized for quality

International Marketing & Sales consists of fiveGlobal Market Segments and the Global SalesOperations.

The Quality Manager, Sankara Narayan, headsa team of Quality and Reliability specialists,dedicated to ensuring defect-free products.He is also responsible for steering the worldwideProduct Centre Quality Managers, each locatedwithin the relevant product centre, assemblycentre or sales organization. He also leads theBEST improvement managers of BUs and AUs.He reports to the Manager Executive Office.

Organization of the Philips Group

Product DivisionPhilips Semiconductors

Board ofManagement

Other productdivisions

Philips Semiconductors is one of several product divisions reporting directly to the Board of Management.

5

Jan Lobbezoo

CFO

Rick Goeld

Discretes &MultiMarket ICs

Roland Dierick

HRM

Indro Mukerjee

IMS

Frits van Rappard

Legal

Rob Horbach

Executive Office

In addition to staff departments, Philips Semiconductors has nine Business Units covering ICs and discretesemiconductors and four Accountable Units. The managers report directly to the Chief Executive Officer Scott McGregor.

Theo Claasen

CTO

Mario Rivas a.i.

Operations

Mario Rivas

CommunicationsBusinesses

Leon Husson

ConsumerBusinesses

Scott McGregor

CEO

Peter Yates

Foundries

Thierry Laurent

MobileCommunications

John Payne a.i.

DigitalConsumer Systems

Ger Schonk

ATO

Phil Pollok

Networking

Steve Kelley

Display Solutions

Mathieu Clerkx

CIO

Karsten Ottenberg

Identification

Theo Akkermans

MainstreamConsumer Systems

John Payne

Emerging Business

Hans van Otterloo

MobileConsumer Systems

Organization of Philips Semiconductors Nov. 2001

Executive Board

6

Organization of Quality Management Semiconductors

Within Philips Semiconductors, Quality Manager Sankara Narayan heads a team dedicated to ensuringdefect-free products.

Support PD Management in establishing PD Business Improvement Policy

BEST programQuality Improvement CompetitionEducational and Quality Improvement activitiesEnvironmental Management

PD Quality Manager

Sankara Narayan

ExternalStandardization

& QualityImprovementTheo Vedder

QMSData

Management

QualitySupport

Jaap Keyman/Hyok Kwon

PackingCoordination

Gijs Lijbers

CommercialQuality programs

Data manager of qualitydatabaseprograms

QualityAssurancecoordinationAssembly &Diffusion

QualityProcedures &Standards

Contacts withsalesorganizations ongeneral QArelated issues

SecretaryInternationalQualityManagersMeetings

Processing ofgeneral packingchanges

Packing & labelingcoordination

Storesevaluation

Environmentalpacking issues

Secretary

Alice Darquennes

Chairing andparticipationExternalstandardization(NEC, IEC)

Support for IECstandardizationWorking Groupmembers

Support forBEST program

General Qualitypublications

TraceabilitySystemManagement

7

QualityImprovement

Henk Otten

Chairmanreliability council

Wafer levelreliability

Standardizationof test methods

Internationalreliabilitystandardization

Competitionbenchmarking

Contacts withleading-edgecustomers

IC ReliabiltyCoordination

Jaap Bisschop

EnvironmentalCoordination

Leo Klerks

Establishing andprogress controlPDenvironmentalprogramme

ChairmanSemiconductorsplantenvironmentalcoordinatorsmeeting

Environmentalinformation andprocedures

Customerquestionnairesonenvironmentalissues

Driving theimplementationof PBE / BEST,organization ofPBEassessments

Organization ofQuality SystemAudits atsubcontractors

Support ofprocess orientedaudits atassembly / testand waferfactories

8

Contact information for Quality Management Semiconductors is provided below:

Location: Building BE-1, Eindhoven, (Rooms 124-129 and 145-147).

Address: Philips Semiconductors B.V., Building BE-1, P.O. Box 218, 5600 MD, Eindhoven, The Netherlands.

Fax number: (31) 40 272 2415.

Name Telephone number E-mail

Sankara Narayan (31) 40 272 30 12 [email protected] Vedder (31) 40 272 29 01 [email protected] Lijbers (31) 40 272 25 01 [email protected] Otten (31) 40 272 29 77 [email protected] Darquennes (31) 40 272 26 07 [email protected] Klerks (31) 40 272 45 08 [email protected] Keyman (31) 40 272 47 50 [email protected]

For IC Reliability matters contact:

Jaap Bisschop,Philips Semiconductors, Nijmegen, Building BQ 0.070.

Tel: (31) 24 353 39 63Fax: (31) 24 353 36 59E-mail: [email protected]

For IC Assembly Quality matters contact:

Hyok Kwon,Philips Semiconductors Sunnyvale,811 East Arques Avenue,P.O. Box 3409,SUNNYVALE,California 94088-3409, USA.

Tel: (1) 408 991 3265Fax: (1) 408 991 2566E-mail: [email protected]

MobileCommunications

CaenNijmegen

(power management)Nijmegen

(RF modules)San Jose

Zürich

Sales

Eindhoven / ParisSan Jose

TaipeiTokyo

PD Quality Manager

Worldwide Product Centre Quality Managers are coordinated by PD Quality Manager Sankara Narayan.

Identification

Hamburg

Networking

AlbuquerqueNijmegen

Foundries

AlbuquerqueBöblingen

CaenFishkil

HamburgNijmegen (AN)

Nijmegen (MOS-3)Nijmegen (MOS-4)

San Antonio

Display Solutions

SouthamptonSunnyvale

EmergingBusinesses

San JoseHamburg

Mobile ConsumerSystems

HamburgSunnyvale

Assembly & TestOrganization

BangkokCalamba

Kaohsiung

MainstreamConsumerSystems

CaenNijmegen

(mainstream TV)Nijmegen

(analog audio)

Discretes &Multimarket ICs

CabuyaoGuangdong

HamburgHazel GroveHong Kong

Nijmegen (Logic)Nijmegen (automotive)

SerembanStadskanaalSunnyvale

Tempe

Digital ConsumerSystems

SouthamptonSunnyvale

9

10

Quality Managers of Global SalesOperations

Paris, FranceGSO-Europe,Regional Quality Manager: Anne RouchierPhilips Semiconductors,51, rue Carnot,92156 Suresnes,France.Tel: +33 147 286 670Fax: +33 147 286 627

Sunnyvale, USAGSO-America,Regional Quality Manager: Marty MichaelsPhilips Semiconductors Sunnyvale,811 East Arques Avenue, MS 20,P.O. Box 3409,Sunnyvale, CA 94088-3409, USA.Tel: +1 408 474 8082Fax: +1 408 474 8212

Taipei, TaiwanGSO-Asia Pacific,Regional Quality Manager: Betty LiangPhilips Semiconductors,5F, Nr96, Sec. 1, Chien-Kuo N. Rd,P.O. Box 22978,Taipei, Taiwan, R.O.C.Tel: 886 2 21342450Fax: 886 2 21342429E-mail: [email protected]

ISO 9000

QS 9000TQE

ISO 14001PQA-90

P B E

Quality at workPhilips has long recognized the vital importance ofhigh quality in electronic components, and itscrucial effect on the viability and economics offinished equipment. This is especially true forsemiconductors, which often perform critical circuitfunctions, such as handling high frequencies, andtransmitting high-speed digital data, often inhostile environments.To achieve the improvements that have made ourproducts amongst the most reliable available, wefully cooperate with our major customers toimprove products and processes, to refine testmethods to match applications, and to ensurecorrect applications conditions. Feedback of dataon quality levels achieved on customer assemblylines and in service is a vital and continuing part ofthis cooperation, because it measures the qualitythat really matters; the quality experienced by ourcustomers which, in turn, directly influences theirreputation in the market.

From ISO 9000 to BusinessExcellenceISO 9000 certificates for Philips Semiconductorsmanufacturing centres were achieved as early as1991/1992. The sales organizations andheadquarters were certified some years later.With the Ford TQE award and QS-9000certifications the customer requirements in thequality systems were enhanced. ISO 14001shows our dedication to the environment.Subsequently the Philips PQA-90 program set theroad to quality excellence. Almost all eligible unitswere granted the PQA award before 1999.Now, Business Units and Accountable Units arebeing assessed on business excellence to thePBE model in the BEST program.

BEST (Business Excellencethrough Speed and Teamwork)The company-wide BEST program is the Philipsway to Business Excellence. The program waslaunched in July 1999. It is an extension of thePQA-90 program and incorporates the PhilipsBusiness Excellence (PBE) model as the frame ofreference for assessing improvement in overallbusiness performance.

Business Excellence

From ISO 9000 to Business Excellence.

11

12

Business excellence policy

13

As the world becomes increasingly interconnected- through all sorts of new products and services - we foresee substantial changes in the markets served by our customers. In this highly networked and mobile environment, Philips Semiconductors aims to be:

customers first choice in audio, video, communications and combinations thereoffor systems in the home and on the move.

a recognised leader for complete, affordable, and easy-to-use systems-on-silicon, as well as for reliable, cost-effective multi-market ICs and discrete semiconductors.

To achieve this goal, we have adopted the Philipsapproach to total quality management, known asBEST (Business Excellence through Speed andTeamwork). Within the framework of BEST weuse Philips Business Excellence (PBE) model tostructure improvements in our businessprocesses, and in our partnerships with strategiccustomers and suppliers . This process ismanaged at each Business Unit (BU) orAccountable Unit (AU) level. We use balancedscorecards to translate strategy into action, tomonitor our progress, and to reward oursuccesses.

Our approach

Our objective is to maximise shareholder valuethrough consistent, profitable growth-which will beachieved by: Increasing innovation Reducing cycle time Improving quality Honouring commitments

Each BU or AU has defined a set of Key Value Drivers (KVDs) to help the company accomplishthese goals. The KVDs will be reviewed periodically and improvement targets will be set.

Whenever possible, the targets will be validatedthrough customer surveys or benchmarking data.

Periodic self-assessments and management reviews will be done to establish progress. The results, including the PBE score, will be validated through annual PD assessments of all the units. Yearly improvement in the PBE score of >70 points is set as the target.

Recognition through PBE awards of Bronze, Silver and Gold will be given to the units scoring 500, 600 and 700 points, respectively.

Learning from HQ audits will be incorporated in our improvement process

Improvement through Teamwork will be encouraged by means of our Quality Improvement Competition (QIC), which also provides recognition for success.

Arthur van der PoelChief Executive Officer

Note: The Executive Board is re-defining theBusiness Excellence Policy. This new policyis expected to be published end of Q1 2002.Aspects that will be included are the deploymentthrough Business Balanced Scorecards andcontrol of the main processes, leading to improvedservices towards customers and to improvementof Customer Intimacy.

CommunicationsBusinesses

14

Product /Manufacturing centres

Location ofProduct /Manufacturing centre

Europe:France: Caen

Germany: Böblingen

Germany: Hamburg

The Netherlands: Nijmegen

The Netherlands: Stadskanaal

Switzerland: Zürich

UK: Hazel Grove

UK: Southampton

USA:Albuquerque, New Mexico

Fishkill, New York

Sunnyvale, California

San Antonio, Texas

San Jose, California

Asia-Pacific:China: Guangdong

China: Hong Kong

Malaysia: Seremban

Phillipines: Cabuyao

Phillipines: Calamba

Taiwan: Kaohsiung

Taiwan: Taipei

Thailand: Bangkok

Discretes& Multimarket

ICs

ConsumerBusinesses

Business Unit served

15

From its Corporate Centre in Eindhoven, The Netherlands, Philips Semiconductors is responsible forProduct/Manufacturing Centres all over the world, as shown on the above map.

Fran

ce: C

aen

Ger

man

y: B

öblin

gen

Ger

man

y: H

ambu

rgTh

e Ne

ther

land

s: N

ijmeg

enTh

e Ne

ther

land

s: S

tads

kana

alSw

itzer

land

: Zür

ichUK

: Haz

el G

rove

UK: S

outh

ampt

on

Albu

quer

que:

New

Mex

icoSu

nnyv

ale

and

San

Jose

:Ca

liforn

ia

Chin

a: H

ong

Kong

Thai

land

: Ban

gkok

Philip

pines

:Cab

uyao

and

Cala

mba

Taiw

an: K

aohs

iung

Taiw

an: T

aipe

i

Sa

n An

toni

o, Te

xas

China

: G

uang

dong

M

alays

ia:

Sere

mba

n

Fi

shkil

l, Ne

w Yo

rk

16

France: Caen

Built in 1957, Philips Semiconductors factory atCaen, near the D-day landing beaches in Normandy,occupies 27,000 square metres of productionfacilities including over 7000 square metres ofcleanrooms and comprises around 1600 personnel.Philips Semiconductors Caen ensures the design,the manufacturing, the logistics and customersupport of ICs manufactured in a wide range ofprocesses including BiCMOS and CMOS. The Caen site supports a foundry with ultra-modern equipment for the creation of sub-microncomponents. The IC test and finishing areas areorganized as part of the worldwide Assembly andTest Operations (ATO). The site also supports 11Business Lines from Consumer, Communicationsand Multi Market Businesses associated withSystems, Applications, Quality Support, Softwareand RF competences. The factory is certified to ISO 9001 (since 1991),PQA-90 (since 1996), ISO 14001 (since 1997),QS-9000 for the foundry (since 1999) and ECOAudit (since 1999). Plant Quality Manager is Yves Meheust, whoworks closely with the Quality Representativesof the Business Lines and Support. He is alsoresponsible for safety and environmental matters.

Address : Philips Semiconductors, 2, rue de la Girafe, B.P. 5120, 14079 Caen Cedex 5, France.

The Quality Managers are:Yves Meheust,Plant/Foundry, Safety and Environment Tel. 33 (0)2 31 45 2140 Fax. 33 (0)2 31 45 2177 E-mail: [email protected] Michel Dubee,Quality Support, Consumer Businesses Tel. 33 (0)2 31 45 2077 Fax. 33 (0)2 31 45 2300 E-mail: [email protected] Jean-Emmanuel Gillet, MC/BL Cordless Tel. 33 (0)2 31 45 6189 Fax. 33 (0)2 31 45 2280 E-mail: [email protected] Gerard Courteille, MaCS/BL Tuners Tel. 33 (0)2 31 45 2224 Fax. 33 (0)2 31 45 3850 E-mail: [email protected] Corinne Taglione, ATO Tel. 33 (0)2 31 45 3017 Fax. 33 (0)2 31 45 2214 E-mail: [email protected]

17

Germany: Böblingen

Situated in Böblingen, southern Germany, PhilipsSemiconductors GmBH Böblingen (PSB), formerlyknown as SMST, has been a 100% Philipssubsidiary since January 1st 1999.The main wafer fab covers 12,180 square metresof which 6584 square metres are cleanroom withclass 1 in the production area. The backgroundcleanroom class is 1000.The company employs around 700 personnel andthe technical manufacturing capacity is 18,000wafers (200 mm) per month.The manufacturing line has a capacity forstructures down to 0.32 micron. Currently thefollowing processes, with derivatives, are qualifiedat PSB: SC 075, 100, 150, 175 High-voltage options SC 175 LCoS Embedded Logic.PSB is the Philips single source for EmbeddedLogic devices, and offers product design and testsupport for Embedded Logic products.

Process development is performed for processderivatives, such as shrink versions and high-voltage applications.PSB has a long history of quality achievement andwas certified according to ISO 9001 in 1995 andQS-9000 as well as ISO 14001 in 1998. As the firstPhilips fab PSB achieved the new TS 16949 in2001.A TQM concept based on empowerment, achieve-ment bonuses, teamwork and individual motivationhas long been established to ensure continuousimprovement. Today it is extended through PBE/EFQM.

The Quality Manager is Roberto Rapp.E-mail: [email protected]

Philips Semiconductors Böblingen,Schickardstrasse 25, 71034 Böblingen.

Tel: (49) 70 31 18 52 51Fax: (49) 70 31 18 54 00

18

Germany: Hamburg

With a wide variety of activities from design and testdevelopment to manufacturing and marketing,Philips Semiconductors GmbH, with locations inHamburg and Böblingen, is the second largestsemiconductors manufacturer in Germany.

The Philips Semiconductors Hamburg site hostsvarious businesses, wafer fabs and test facilitiesfor discrete semiconductors and ICs. With theirheadquarters located in Hamburg, BU Identification,BL Car Infotainment Systems and BL GeneralApplications Products are responsible for theworldwide development, quality assurance, logisticsand marketing of their products. Some other BusinessLines are represented on-site through businesssegments and/or supporting activities. The SystemsTechnology Centre-Hamburg develops and definesnew systems-on-silicon for several applications andsoftware tools and plays a key role in the PDsystems-on-silicon initiative. In the Discrete WaferFab and the IC Foundry Hamburg, discretesemiconductors and ICs for a large variety ofapplications are produced using a wide rangeof technologies.

Products are also tested on-site, with IC TestOperations being organized as a part of theworldwide Assembly Test Operations.

Philips Semiconductors employs around 2100personnel in Hamburg, spread over two locations.The site holds ISO 9001/QS-9000 and ISO 14001registration and meets Fords Total QualityExcellence (TQE) requirements.

Business Line General Application Products(BL GA)The BL GA is a multi-site activity with headquartersin Hamburg. Besides the waferfab, wafertest,development, marketing, logistics and qualitymanagement in Hamburg, the BL is alsoresponsible for the diode activity in Nijmegen andcoope-rates with significant parts of the assemblyactivi-ties in Guangdong-China (PSG), HongKong-China (EDL), Cabuyao-Philippines (PSPI)and Seremban-Malaysia (PSS). The BusinessLine has been assessed and meets the QS-9000requirements.

19

Product portfolio: low-frequency small-signal transistors switching-, Schottky- and zener-diodes semiconductor sensors for temperature

and magnetic-field measurements.

Quality Manager is Michael KüchlerTel: (49) 40 5613 2401Fax: (49) 40 5613 3093E-mail: [email protected]

Business Line Car Infotainment Systems(BL CIS)The Business Line CIS is a fabless multi-siteorganization with headquarters in Hamburg,and additional locations in Nijmegen(The Netherlands) and Tokyo (Japan).The business segments are: car radio car multimedia.

For the BL CIS five people work on quality issuessupported by the Q&R Service Department(30 people) of the Business Unit Mobile ConsumerSystems.Product portfolio: car radio frontend and tuning voltage regulators and audio power amplifiers audio noise reduction telematics & car navigation driver information (trimedia) speech recognition car digital sound processors and digital radio.

Quality Manager is Karl-Heinz Gressmann.Tel: (49) 40 5613 2752Fax: (49) 40 5613 3548E-mail: [email protected]

Business Unit Identification (BU ID)Business Unit Identification has locations inHamburg (Germany), Gratkorn (Austria) andCaen (France). BU ID employs about 350personnel.

The Business Unit consists of four Business Lines: chip cards tags & labels car access & immobilizers readers.

As technology and innovation leader in the area ofidentification, BU ID provides a broad portfolio forchip card-based identification technologies for smartcards, tags, smart labels, read/write terminals andcar immobilization.In addition to ISO 9000, BU ID has to fulfil severalquality and security management requirements,such as: QS-9000, EUROPAY CQM, CommonCriteria (security) and EMV. These requirementsare mandatory for specific market segments suchas banking, e-business, pay-TV and automotive.

Quality Manager is Helmuth MeyerTel: (49) 40 5613 2817Fax: (49) 40 5613 3554E-mail: [email protected]

BL Microcontrollers (MCO) HamburgBL MCO is a multi-site activity with its headquartersin Sunnyvale (USA). Hamburg activities are focusedon operations, applications, marketing, logisticsand quality support.Product portfolio: Controllers for PC/peripherals and industrial.

Quality Manager is Hans-Juergen KuehlTel: (49) 40 5613 3522Fax: (49) 40 5613 3691E-mail: [email protected]

Systems Laboratory Hamburg with:- system and product innovation- reference design / design-in supportfocusing on:- video systems- software and multimedia- sound and identification systems- automotive and industrial electronics.Design Technology Centre (DTC):- computer-aided test tools.Software Service Group:- integration of existing SSG software components

in customer systems- definition and development of new reusable

software components and subsystems- maintenance of existing SSG software components.BL Mainstream TV Solutions SW (MTS SW):- product innovation- product software.Quality Manager is Christian-Frieder OschatzTel: (49) 40 5613 2107Fax: (49) 40 5613 3524E-mail: [email protected]

Philips Semiconductors GmbH Site Lokstedt, Site Hausbruch,Stresemannallee 101, Georg-Heyken-Str.1,D-22529 Hamburg, D-21147 Hamburg,P.O.Box 54 02 40,D-22502 Hamburg Germany.

Site Programme Manager TQM is Lewe PetersenTel: (49) 40 5613 2744Fax: (49) 40 5613 2914E-mail: [email protected]

20

IC Foundry Hamburg (ICFH)The ICFH Waferfab is a high-volume productioncentre to provide Philips Semiconductors withBipolar, BiCMOS and CMOS services forautomotive, monitor, identification, audio and TVapplications. Its waferfab capacity is around 10kwaferstarts (150 mm) a week, with dimensionsdown to 0.6 micron. Cleanroom facilities occupyabout 4000 square metres. The activity rangeincludes: production fab engineering product engineering process development foundry service. Quality Manager is Horst WaschkewitzTel: (49) 40 5613 2662Fax: (49) 40 5613 2686E-mail: [email protected]

Philips IC Test Operations Hamburg (PICTOH)PICTOH belongs to the Assembly & TestOrganization (ATO) with headquarters inSingapore. It is engineering centre for wafer- andpackage-testing as well as for wafer-treatment.Besides testing of high-end-mixed-signal andmultimedia consumer products, PICTOH is theworldwide centre of competence in PhilipsSemiconductors for testing and backend-processing of RF-identification and smart cardtechnologies. The activity range includes: wafer testing wafer treatment package testing.Quality Manager is Roland RadünzTel: (49) 40 5613 1620 Fax: (49) 40 5613 3113E-mail: [email protected].

Systems Technology Centre-Hamburg (STC-H)The STC-H employs 160 personnel and includes

Hamburg, Hausbruch

21

The Netherlands: Nijmegen

Philips Semiconductors produces in Nijmegendiscrete semiconductors and industrial andconsumer ICs. Starting from 5 employees in 1953for the manufacture of diodes and low-frequencytransistors, the Nijmegen factory has grown tobecome the largest in Europe dedicated solely tosolid-state products. The site is certified to ISO9001 and ISO 14001.

Business Line-RF ModulesThe Business Line RF Modules is a multi-siteactivity and has the worldwide businessresponsibility for wideband, RF and microwavepower transistors, CATV, RF and video modules,small-signal FETs and varicap diodes.Development, wafer diffusion and some RF andmicrowave transistor assembly are handledlocally. Development and diffusion of varicapdiodes are located in Hamburg. The developmentgroup and diffusion of small-signal switching andzener diodes under the responsibility of theBusiness Line GA products is located on thisNijmegen site. Main assembly centres are locatedin Cabuyao-Philippines (PSPI), Hong Kong (EDL)and Seremban-Malaysia (SMP). A centre was

opened in 1999 in Mansfield USA for thedevelopment of RF power amplifiers and CATVmodules.The Business Line is organized in four MarketSector Teams who are responsible for the discreteRF business in the sectors BroadbandCommunication, Basestations, PersonalCommunications and Consumer.The Business Line employs about 700 personnelin Nijmegen and is qualified to ISO 9001, QS-9000and ISO 14001 standards. Continuousimprovement is driven by use of the PhilipsBusiness Excellence model.

The discrete semiconductors activity is supportedby central groups including Industrial TechnologyEngineering Centre (ITEC). Operating as a centreof competence for the Philips Group, ITECprocures and develops equipment required forsolid-state manufacturing and testing. One of itsdevelopments is the Phicom computer-controlledwire bonder, 1000 of which have already beensold worldwide.Quality Manager is Jos de Bruijn.

Business Line LogicThe Business Line Logic has worldwide businessresponsibility for the HEF, LV, (A) LVC, HLL,AC/ACT and HC/HCT product ranges, which arediffused in Nijmegen, Albuquerque and Crolles.Business Line Logic, based in Sunnyvale, has animportant manufacturing and design centre inNijmegen. It also has the European marketing andQ&R responsibility for a number of bipolar andBICMOS logic ranges produced in the USA, suchas FAST, ALS, ABT, LVT and Futurebus. Around20% of the diffusion centre load is allocated forother Business Lines, such as cellular, CS andPower-DMOS for Discretes.Nijmegen houses one of the three ProductInnovation Centres (PICs) specifically targetingEuropean customers for new standard productideas. The other PICs are strategically located inSingapore and Albuquerque, for worldwidecoverage.The Business Line employs around 530 personnelin Nijmegen. Assembly takes place in a number ofPhilips Assembly and Test centres worldwide, aswell as at a number of subcontractors premises.The Business Line holds various key-customercertifications, including S&V, Seagate, Ford TQE,Chrysler SQA and IBM Cat 1 and 3. Quality Manager is Joop Willemsen.

Business Line Mainstream TVThe Nijmegen facility includes an importantdevelopment and worldwide production centre foraudio, digital audio, video, power management

and automotive consumer ICs.This centre employs 1100 personnel, of whom 45are in the Quality Department. It has Ford TQE(188 out of 200 points) and Chrysler SQA (96 outof 100 points) certifications. It is also certified forQS-9000 and has the PQA-90 award.Quality Manager is vacant.

Business Line Analog AudioThe Business Line Analog Audio (BL AA) is part ofthe BU Mainstream Consumer. The Business Lineis a multi-site organization with headquarters inNijmegen. Around 65 personnel are involved andorganized in two segments: - data converters- audio amplifiers.The BL AA product range includes: - audio DA and AD converters- audio CODECS- audio amplifiers for multimedia, home

and portable audio.Quality Manager is Paul van Warmerdam.

Business Line Power ManagementThe Business Line Power Management is partof the BU Mobile Communications. PowerManagement aims to be the leading supplierof system solutions and building blocks for high-volume mobile communications, consumer andcomputing products. The BL has about 70employees located in building Meijhorst,and will soon be certified for ISO 9001: 2000.Quality Manager is Kees de Vaal.

22

MOS-3 foundry (photo on previous page)The Nijmegen Submicron wafer fab MOS-3 employsover 500 personnel on a 5-shift operation. Itproduces a large variety of products in submicronCMOS, BiMOS, EPROM and EEPROM processesand many different ROM-codes for variousconsumer and industrial applications. The productsare developed by several Philips Business Linesusing the product introduction line.The Class-1 cleanroom area occupies about 3700square metres with a capability down to 0.5 micronon 150 mm wafers using modern I-line steppers,single-wafer etchers, vertical furnaces and in-linewafer inspection equipment.Wafer fab MOS-3 is certified for ISO 9001, QS-9000 and ISO 14001. It holds the Ford TQE award,the 1995 Fab of the Year award and passed thePQA-90 audit. Quality Manager is Wim Bremer.

MOS4YOU foundry (photo on previous page)The Nijmegen MOS4YOU (Yield-Output-Utilization) 8 inch waferfab is Philips largestwaferfab with approximately 1000 employees.After foundation in 1995 with a planned capacity of200k wafers, MOS4YOU expanded the wafercapacity towards 300k wafers (C100 equivalents)and utilizes CMOS processes in production fromC100 to CMOS18 (feature size 0.5 to 0.18 micron). A large part of the wafer output is in CMOSprocesses with embedded Non-Volatile memories,which are developed in-house.Because of their extensive development in processtechnologies like today's CMOS18S (feature size0.15 micron) MOS4YOU is recognized as a"motherfab" for process transfers to other fabs.Its "Speedline" is marked as "benchmark" andprovides to BLs the fastest "time to market" withrespect to product introductions. MOS4YOU iscertified for ISO 9001, QS-9000 and ISO 14001.The waferfab can be characterized as follows:- Cleanroom: 7519 square metres- Cleanroom concept: mini environments, SMIF

- Environmental conditions: class 0.1 within SMIF- Feature size: 0.5 to 0.18 micron

(0.15 in development)- Technology: CMOS and options- Product mix: logic, mix-mode and Non-Volatile.Quality manager is Gerard de Groot.

Business Line Networking InfrastructureThe Business Line Networking Infrastructuredesigns and manufactures ICs for the fibre opticsinfrastructure that deals with transmitting andreceiving signals over the fibre optics network.The market for this area is strongly driven by theever increasing need for more capacity on theInternet Highway. The products operate with datarates up to 2.5 Gbit/sec. ICs for data rates up to to10 Gbit/sec are in development. This Business Lineis a fast growing group of about 60 personnel. Quality Manager is Hans Buis.

Philips Semiconductors,Gerstweg 2,6534 AE NIJMEGEN,THE NETHERLANDS

Quality ManagersJoop Willemsen, DMI/BL LogicTel: (31) 24 353 2464 Fax: (31) 24 353 3424Jos de Bruijn, MC/BL RFTel: (31) 24 353 4969 Fax: (31) 24 353 3160Vacant, MaCS/BL Mainstream TVTel: (31) 24 353 2547 Fax: (31) 24 353 4638Hans Buis, Net/BL Networking InfrastructureTel: (31) 24 353 2097 Fax: (31) 24 353 3589Paul van Warmerdam, MaCS/BL Analog AudioTel: (31) 24 353 4864 Fax: (31) 24 353 6200Kees de Vaal, BL Power ManagementTel: (31) 24 353 6301 Fax: (31) 24 353 3613Wim Bremer, MOS-3Tel: (31) 24 353 6640 Fax: (31) 24 353 3602Gerard de Groot, MOS4YOUTel: (31) 24 353 3336 Fax: (31) 24 353 3890

23

24

The Netherlands: Stadskanaal

Stadskanaal, in the north of the Netherlands, is theinternational centre for product / processdevelopment, manufacture, testing and qualityassurance of medium-power rectifier diodes. Itemploys around 500 personnel.

Additional activities include: diffusion and assembly of medium-power

rectifiers, switching, efficiency and zener diodes diffusion and assembly of high-voltage rectifier

diodes and stacks.

The factory holds ISO 9001, QS-9000, PQA-90and ISO 14001 certifications.Quality Manager is Rik Elzinga.

Philips Semiconductors,Electronicaweg 1,9503 GA Stadskanaal,The Netherlands.

Tel: (31) 599 632 555Fax: (31) 599 632 505

* Stadskanaal is part of BL Power, headquartered in Hazel Grove, UK.

pager ICs I2C bus ICs clock and watch ICs.

The site holds QS-9000, Ford Q1, ISO 9001and ISO 14001 qualifications.Site quality manager is Daniel Gloor.

Philips Semiconductors,Binzstrasse 44,CH-8045 ZURICH,SWITZERLAND.Tel: (41) 1 465 1314Fax: (41) 1 465 1800

Quality Managers:Daniel Gloor, BU Display SolutionsTel: (41) 1 465 1314Fax: (41) 1 465 1805

Thierry Kieffer, MC/Cellular 3GSMTel: (41) 1 465 1460Fax: (41) 1 465 1806

25

Switzerland: Zürich

Business Lines Mobile Display Drivers, LargeDisplay Solutions, Cellular 3GSM, Cordless andConnectivity all have development activities inPhilips Semiconductors Zürich. Business Unit Display Solutions and its BusinessLines Mobile Display Drivers and Large DisplaySolutions are also headquartered in Zürich.The site specializes in the design, developmentand marketing of advanced electronics for telecomterminals. Philips Semiconductors Zürich is a worldleader in the development of wirelesscommunications products and systems - the ICs,chipsets and software for cellular and cordlesstelephony, display drivers and wired telephony. Philips Semiconductors Zürich employs 400personnel in the creation of ICs, support toolsand software. ICs are manufactured worldwidein CHV40, C175, C100, C075, C050, SAC3and SAC2 processes.Products from Philips Semiconductors Zürichinclude: ICs for GSM, Bluetooth and DECT telephony microcontrollers

26

United Kingdom: Hazel Grove

Occupying 15,000 square metres on the edge ofthe Cheshire countryside, Hazel Grove is thebusiness centre for the power semiconductorproducts and is responsible for the marketing,development and manufacturing activities utilizedby the Business Line Power.The Hazel Grove site employs around 500personnel and is proud of the extensiveinvolvement in quality improvement activitiesgiven by all. The site has both a bipolar andPowerMOS wafer fabrication facility supplyingproducts for applications in the Consumer (TV &Monitor), Automotive (Engine management andsafety applications such as ABS), DomesticAppliance & Personal Care and Industrial marketsectors.Products from the bipolar facility include a fullrange of high-voltage power transistors, thyristors,triacs, and Schottky, zener & breakover diodes.

Products from the PowerMOS facility include a fullrange of standard and logic-level MOSFETS forlow- and medium-voltage applications, togetherwith protected IGBTs and TOPFETs.The plant holds ISO 9001 and QS-9000registration and has an effective TQM processfirmly established to ensure continuousimprovement. Quality Manager is Tim Crispin.

Philips Semiconductors,Bramhall Moor Lane,Hazel Grove,STOCKPORT,CHESHIRE SK7 5BJ,United Kingdom.

Tel: (44) 161 957 5517Fax: (44) 161 957 5395

centre of competence for the development of ICdesign tools for all PS design activities worldwide.The DTC is certified to CMM level 2.

These activities are located in purpose-builtaccommodation and represent one of the singlelargest concentrations of electronics engineeringand software expertise in the UK. The organizationcurrently comprises more than 500 personnel, over50% of whom are graduates or postgraduates,predominantly in electronics engineering andsoftware disciplines. Construction of a new buildingwas completed in the spring of 2000 to provideaccommodation for the extra staff needed to supportthe continuing growth in the sites activities.Total Quality Management is practised throughoutthe site, which is certified to ISO 9001, ISO 14001,PQA-90 and Investors in People, a UK Governmentinitiative in connection with staff development andcare. Additionally the site is adopting the Philips-wideBEST approach to Business Improvement. Quality Manager is Tim Johnson.

Philips Semiconductors Limited, Millbrook Industrial Estate, Southampton, Hampshire SO15 0DJ, United Kingdom. Tel: (44) 2380 316565 Fax: (44) 2380 316305

27

United Kingdom: Southampton

Philips Semiconductors Southampton (PSS) is asite with a long association with IC design andmanufacture. The site started operations in 1956as Mullard. Over the years it has become a worldleader in teletext and CD applications, supplyingproducts to over 40 countries with deliveries nowstanding at over 100 million pieces per year. The Southampton site supports the followingbusiness activities:

Business Line Storage (BL-S): headquartered inSouthampton, BL-S is responsible for marketing,management, quality and logistics support of twomarket sectors:- audio/video: design of ICs for CD audio,

CD recordable, DVD video and DVD recordable- PC applications: design of ICs for CD ROM,

CD rw and DVD rw drives.Business Line Broadband Home Servers(BL- BHS): headquartered in Sunnyvale,California, BL-BHS has its DTV market sectorbased at Southampton, responsible for design ofICs for digital television.Systems Laboratory (SLS): a leading centre ofcompetence for the development of CD, DVD,DTV and navigation systems.Design Technology Centre (DTC): a leading

28

USA: Albuquerque, New Mexico

Philips Semiconductors Albuquerque plantopened in 1982 as one of the world's mostadvanced MOS wafer-fabrication facilities. Fab 22(4 inch/100 mm wafers) reached full production in1983; Fab 23 (6 inch/150 mm wafers) reached fullproduction in 1989. The Albuquerque plantoccupies 43,000 square metres, employs around1000 personnel, and is engaged in the design anddevelopment of integrated circuits; waferfabrication of NMOS, CMOS, Bipolar, BiCMOS(QUBiC) and EPROMs; and wafer test.Total Quality Management (TQM) is practisedthroughout the site, which is certified to ISO 9001,EIA-599, QS-9000, Ford TQE and DESC QMLproducts.All employees are actively engaged in qualitythrough quality teams. On-site qualityrepresentative and Quality Manager contact isJohn Wilson.

Philips Semiconductors Albuquerque,9201 Pan American Frwy., N.E.,Albuquerque,NEW MEXICO 87113,USA.

Quality ManagersRick Bradley, ATO Tel: 1 505 822 7135Chris Merkley, Connectivity Tel: 1 505 822 2878John Wilson, Fab Tel: 1 505 822 7409

Fax: 1 505 822 7752

29

USA: Fishkill, New York

The Philips Semiconductors Fishkill plant (PSF) isan 8 inch wafer fabrication and test facility with thecapability of manufacturing CMOS and QUBICtechnologies with feature sizes down to 0.18 micron.The facility, formerly known as MiCRUS, waspurchased from IBM in July 2000. MiCRUS wasoriginally formed as a joint venture between IBMand Cirrus Logic in 1995. The plant occupiesapproximately 21,000 square metres at the HudsonValley Research Park in Hopewell Junction,New York, and employees 1050 personnel.

In addition to the 0.35 micron and 0.25 micronCMOS technologies inherited from Cirrus Logicand IBM, PSF has received RFP (Release forProduction) on Qubic 3 and C075EE technologies.PSF is working on reaching RFP on Qubic 4 and4G technologies for the future.

PSF achieved ISO 9002 registration in June 1996as recognition that its quality system meetsinternational standards and conforms to the ISO14001 environmental standard.

As MiCRUS, the facility has received numerousaccolades and awards from its customers andthe community, and was named SemiconductorInternationals Fab of the Year in May 1998.The Manager of Quality and CustomerApplications is Bob Incerto.

Philips Semiconductors Fishkill,Hudson Valley Research Park,Box 1279,Hopewell Jct., N.Y. 12533,USA.

Tel: 1 845 894 2550Fax: 1 845 892 2863

30

USA: San Antonio, Texas

The Philips Semiconductors San Antonio plant,recently purchased as part of VLSI Technology,Inc., was opened in 1988. It is an 8 inch wafer fab,producing ICs with its current qualified processesdelivering feature sizes 0.8 to 0.25 micron. Thefab qualified its 0.20 micron process in late 1999.From two to five layers of metal are available forthe advanced ASIC and ASSP products fabricatedin San Antonio. The fab employs about 650personnel with a 4-shift, 7-days a week operation,and occupies 5100 square metres of cleanroomClass 1 and 27,000 square metres ofmanufacturing. The capacity is about 5000 8-inchwafer starts per week. A high percentage of itsworkers have advanced college degrees.

The fab has a reputation for quick turnroundprototyping and a high level of customer serviceon all technologies.The San Antonio fab is certified to ISO 9002and ISO 14001.Quality Manager is Widge Stamback.

Philips Semiconductors San Antonio,9651 Westover Hills Boulevard,San Antonio, Texas,USA 782512701.

Tel: 1 210 522 7400Fax: 1 210 522 7100

31

USA: San Jose, California

Philips Semiconductors San Jose was founded in 1979(as VLSI Technology) and was acquired by PhilipsSemiconductors in June 1999. The trademark onproducts was changed from the VLSI logo to the PhilipsShield in 1999.

The San Jose site occupies 40,000 square metres,employs approximately 500 personnel, houses thesenior staff of BU Emerging Businesses (EB) and BUNetworking (NW) and the Silicon Valley activity of BUMobile Communications (MoC). Specific emphasis isplaced on custom and semi-custom products for targetmarket segments Wireless Communication, Networkingand Advanced Computing. The site also houses thecentral staff of Global Sales Organization (GSO)Americas, supports technology development (CTO -process & library), product design, final test/warehouse(ATO), and laboratories for Reliability Testing (QTS),Failure analysis (QTS) and Package Development(ATO). The ATO lab is ISO 14001 certified. Total QualityManagement (TQM) is practised, led by the SunnyvaleQuality and Technical Services (QTS) organization.

The main buildings within the San Jose site house thefollowing groups: McKay 1 - 1101 McKay Dr: CTO, ATO, QTS McKay 2 - 1109 McKay Dr: Marcom, Finance McKay 3 - 1151 McKay Dr: Cafeteria, Auditorium,

Data Centre/Servers, BU-EB McKay 4 - 1251 McKay Dr: BU-EB, BU-NW,

Labs - Network/Software Dev./Quick Turn McKay 5 - 1240 McKay Dr: HR, IT, PL-IDS,

Library - CTO/IT Ringwood 1 - 1100 Ringwood Ave.: BU MoC Labs Ringwood 2 - 1120 Ringwood Ct.: BU MoC BL and

PL offices Ringwood 3 - 1130 Ringwood Ct.: GSO Americas.

Philips Semiconductors San Jose,1109 McKay Drive, San Jose, California 95131, USA.Tel: (408) 434 3000

See USA: Sunnyvale, California for the primary mailingaddress of Philips Semiconductors Silicon Valley andfor a listing of Silicon Valley Quality Managers.

32

USA: Sunnyvale, California

Philips Semiconductors Sunnyvale was founded in1961 (as Signetics) by a group of scientists andengineers, who were among the pioneers in thedevelopment of integrated circuits. It was the firstcompany in the world to be established for the solepurpose of designing, developing, manufacturingand marketing of ICs. Philips acquired Signetics in1975. The trademark on products was changedfrom the block S Signetics logo to the PhilipsShield in 1994.

The Sunnyvale site occupies 46,000 squaremetres (3 buildings), employs approximately 1200personnel, and houses the senior staff of BusinessUnit Discretes & Multimarket Products (DMI) andPhilips Semiconductors Information Office (IO).Sunnyvale has product design and technologydevelopment (CTO) resources for Business Lines.The site includes Business Lines of BU MobileConsumer (MoC) and BU Digital ConsumerSystems (DCS).ATO has facilities for pre-test (wafers), final testand warehouse (Industrial Distribution Centre -IDC). Quality & Technical Services (QTS) has labsfor Reliability, Calibration and Failure Analysis.

Total Quality Management (TQM) is practisedthroughout the site, which is certified to QS-9000,ISO 9001, ISO 14001, EIA-599 and Ford TQE.

Three main buildings within the Sunnyvale sitehouse the following groups: 811 East Arques Ave: ATO, BL Automotive,

BL Logic, BL MCO, DMI Offices, CTO, General Counsel, QTS.

440 Wolfe Road: CTO Offices/Lab, IO,BL Broadband Home Servers, BL Media Processing.

690 Arques Ave: EDS, CLASS.

The Sunnyvale site in combination with the SanJose site is referred to as the Silicon ValleyCampus. Visit the Silicon Valley IntranetHomepage at http:// www.sv.sc.philips.com/Locations/html/sanjose.html

33

Product Group Quality Manager Phone Fax

ATO Test Steve Christensen 14106 45330BEST/PBE Program Manager Ray Suen 13828 12656MoC/BL Media Processing Anita Chan 14977 13300DCS/BL Broadband Home Servers vacant 14933 15633EB/BL Computing Joy Gandhi (San Jose 408 474 5344)Net/Connectivity Chris Merkley (See Albuquerque)Net/BL Broadband Access Chris Merkley (See Albuquerque)IMS/GSO Americas Marty Michaels (San Jose 408 474 8082) DMI/BL Automotive Paul Autio 13777 13858DMI/BL Logic Joop Willemsen (See Nijmegen)DMI/BL Microcontrollers (MCO) Gary Huffman 12738 13773DMI/BL Standard Analog Dennis Reed (Tempe 480 752 6290) QTS/Quality Services Bill Hobdy 13893 12566MC/BL Cellular Infrastructure Doug Ridley (San Jose 408 474 7667)MC/BL Cellular CDMA Doug Ridley (San Jose 408 474 7667)

The QTS department under Garic Power supports the Business Units, Business Lines and Product Lineslocated in Silicon Valley, Tempe and Albuquerque. In QTS, Bill Hobdy heads Quality Engineering, Bill Fullheads Reliability Engineering Services and Dave Dunlop heads Failure Analysis Services. Silicon ValleyQuality Managers are listed below:

Philips Semiconductors Sunnyvale,811 East Arques Avenue,P.O. Box 3409, Sunnyvale, CA 94088-3409, USA.

Tel: 408 99 13893Fax: 408 99 12566

34

China: Guangdong

Philips Semiconductors (Guangdong) Co., Ltd.(PSG) was founded in 2000, and is anInternational Production Centre for the assemblyand testing of plastic-encapsulated discretesemiconductors including transistors, diodes andsensors.

PSG occupies 23,000 square metres, employsover 600 staff and has a capacity of over 4000million devices per year, produced in leaded andSMD packages.

Quality system implementation and its improvementis always our major objective, and PSG plans toapply for ISO9001: 2000 certification by the endof 2001 and ISO/TS 16949 and ISO 14001certification in 2002.Quality Manager is George Song.

Philips Semiconductors (Guangdong) Co. Ltd.,Tianmei Industrial North District A Section,Huangjiang Town DongGuan City,Guangdong Province, 523750 P. R. China.

Tel: (86) 769 3632838-408Fax: (86) 769 3632818

35

China: Hong Kong

Philips Semiconductors Electronic Devices Limited(EDL) is an International Production Centre for theassembly, testing, packaging and distribution ofplastic-encapsulated discrete semiconductorsincluding: small-signal transistors/diodes junction FETs, MOSFETs, VDMOS

PowerMOS and TrenchMOS wide-band transistors thyristors and triacs.

EDL occupies 26,000 square metres and employsover 1700 staff. It has a capacity of over 10,000million devices per year, produced in 15 leadedand SMD packages where SOT 23 is the highestrunner. SMDs are produced on flowlines withstate-of-the-art production facilities.

Quality has always been imperative in the EDLculture. EDL is certified to ISO 9002, QS-9000and ISO 14001. It also holds Ford TQE and AT&TQLP Gold preferred supplier status. EDL receivedthe PQA-90 award in 1997 and is pursuing thePhilips Business Excellence Programme. Quality Manager is Kam Leung Luk.

Electronic Devices Ltd. (EDL),10th fl., General Garment Building,100110 Kwai Cheong Road,Kwai Chung, New Territories,(PO Box 122, Texaco Road Post Office),HONG KONG.

Tel: (852) 2424 4024 / (852) 2480 7800Fax: (852) 2480 0602

36

Philippines: Cabuyao (PSPI)

Philips Semiconductors (Phils.) Inc. (PSPI) is anInternational Production Centre for the assembly,testing, packaging and distribution of discretesemiconductors and hybrid modules which include: PowerMOS, high-voltage transistors, thyristors,

triacs and power diodes CATV, base-station devices, optical receivers,

RF microwave transistors and circulators /isolators

glass and ceramic diodes, magnetic sensors RF modules.

The modern plant occupies 27,000 square metresand employs over 3500 staff. Continuous qualityand productivity improvement is rigorously practisedvia the Philips Business Excellence (PBE) program.Advanced quality tools and statistical techniquesare widely applied. PSPI is certified to ISO 9001,QS-9000 and is aggressively pursuing customer-driven programmes such as QS-9000 and Ford TQE.The Plant Quality Manager is Ms. Emma R.Tomelden and the Quality Leaders for each

product line are: Power: Mr. Dong Valdeavilla General Applications: Mr. Edd Lincuna RF Infrastructure: Mr. Ronnie Rivera RF PCC: Ms. Mariver Limosinero.

Mailing Address:Philips Semiconductors Philippines Inc.,P.O. Box 7051,Domestic Airport Post Office,1300 Pasay City, Manila,PHILIPPINES.

Plant Address:Philips Semiconductors (Phils.) Inc.,Light Industries and Science Park,Philips Avenue, Barrio Diezmo,Cabuyao, Laguna,Philippines.

Tel: 63-2-844 5139Fax: 63-2-844 5248

37

Philippines: Calamba (PSC)

Philips Semiconductors Calamba (PSC) is thethird IC Assembly and Test Centre within theAssembly & Test Organization (ATO) of PhilipsSemiconductors.

Located some 50 km south of Manila, amanufacturing facility is being built in two phaseson 85,000 square metres of land. The first phase,with 18,000 square metres of cleanroomproduction area, started production in January1999. Construction of the second phase, with anadditional 25,000 square metres of cleanroomproduction space, started in 2000 and has nowbeen completed.

The first packages assembled and tested at PSCwere plastic QFPs, with pincounts up to 100.In 2000, the range of packages was extended withmore complex QFPs, as well as (T)SSOPpackages. BGA package qualifications wereconducted in 2001. By the end of 2001, the planthad approximately 1700 employees. Continuous Quality Improvement and TotalCustomer Satisfaction are focal points of the TQMprogram. Advanced quality tools and statisticaltechniques are part of the standard training

programme for all employees.PSC successfully passed third-party certificationto the ISO 9002: QS-9000 standard within its firstyear of operation, while ISO 14001 certificationwas achieved in March 2000.The Quality Manager is Jojo Fabia.

Philips Semiconductors Calamba,9 Mountain Drive, LISP-ll,Bgy. La Mesa, Calamba, Laguna, Philippines 4027.

Tel: (63) 49 545 7800Fax: (63) 49 545 0681

38

Taiwan: Kaohsiung

Philips Semiconductors Kaohsiung (PSK) wasestablished in Kaohsiung in 1996. Today itemploys 2720 personnel housed in three buildingsoccupying about 53,000 square metres, of whichover 8400 square metres are cleanroom facilities.Producing ICs for all IC Business Units, its annualcapacity is 1050 million pieces. Main packages areDIL 300 and 600, shrink-DILs, SILs, QFP, L/TQFPand small/large SO, VSO, SSOP and BGA.Quality is paramount in the Kaohsiung plant, whichhas ISO 9001, ISO 14001 and QS-9000certification for IC assembly, testing and packagedevelopment. It also received the prestigiousJapanese Deming award and the JapaneseQuality Medal for quality excellence.Quality Manager is Adam Lai.

Philips Semiconductors Kaohsiung (PSK),10 Chin 5th Road N.E.P.Z.,P.O. Box 35-48,KAOHSIUNG,TAIWAN (R.O.C.).

Tel: 886-7-36 12 51 1Fax: 886-7-36 12 49 3

39

Taiwan: Taipei

Consumer Systems Taipei (CST) was establishedin 1985 as a part of the development of FaselecTelecom IC MDP-team. In 1989 this design houseturned to be an MDP-team and is now one of theproduct centres of Philips Semiconductors. Theorganization currently comprises around 100 staffin marketing, development, quality, factoryinterface, testing, systems, application, softwareand logistics, and offers a broad range of ICs.The market segment of Consumer Systems Taipeiis focusing on microcontrollers and peripherals forconsumer applications: TV micros and standalone OSD Multiscan autosync Monitor micros and

standalone OSD Remote control micros and dedicated ICs (Portable) Radio/Audio micros Graphic User Interfaces and Speech

Recognition Monitor system ICs Software Micros for VCD.

Technologies employed are SAC2, C15xx andC100, manufactured by world-class IC fabricationat TSMC (Hsinchu), MOS3, MOS4 and PSK, PSTassembly foundries.Total Quality Management (TQM) is practisedcontinuously by CST, and ISO 9001 has beencertified by KEMA since December 1995.Quality Manager is P. S. Yen.

Philips Taiwan Ltd. (CS),23FB, No.66, Chung Hsiao W Rd., Sec. 1,P.O. Box 22978,TAIPEI,TAIWAN (R.O.C.)

Tel: 886 7 367 8862Fax: 886 2 382 4777

40

Thailand: Bangkok

Philips Semiconductors Thailand (PST), inBangkok, assembles and tests ICs for worldwidemarkets. Bangkok supports Business Unitsheadquartered in both Europe and the USA.Major product lines include standard bipolar logic,standard CMOS logic, programmable logic, analogindustrial, bipolar and identification devices.Bangkok also handles testing and control ofhermetic products built offshore by a subcontractor.Located 5 km from Bangkoks international airport,the Bangkok facility recently completed a majorexpansion at its 63,000 square metre site, of which45,000 square metres is allocated formanufacturing. The facility employs 4200personnel, with production capability to exceed3500 million devices per year.Its world-class manufacturing capability is driven by continuous quality improvement. The plant holds DESC, ISO 14001, ISO 9001 and QS-9000certifications. Quality Department manager is Apichai Lertapiruk.

Philips Semiconductors Thailand,303 Moo 3 Chaengwattana Road,Laksi,BANGKOK 10210,THAILAND.

Tel: 66-2-5511052-62Fax: 66-2-5511063-64

41

Separate units, joint ventures and subcontractors

Austria, Gratkorn, PhilipsSemiconductorsPhilips Semiconductors Gratkorn is a companyspecialized in contactless RFID and Smart CardSystems design, development, production andmarketing.The Gratkorn site is part of the Business LineIdentification (BLI) in Hamburg and is ISO 9001certified. Quality Manager is Helmuth Meyer in Hamburg.

Helmuth Meyer, Business Unit Identification (BUI)Tel: (49) 40 5613 2817Fax: (49) 40 5613 3554E-mail: [email protected]:Philips Semiconductors Gratkorn,Micron-Weg 1, A-8101 Gratkorn, Austria.

China, Shanghai, ASMCAdvanced Semiconductor Manufacturing Corp. ofShanghai (ASMC) is a joint venture with a Philipsshare of about 38%. ASMC has a 125 mm waferfab for bipolar and SACMOS processes and a150 mm wafer fab for CMOS processes. The siteemploys about 450 personnel and is ISO 9002certified. The quality contact is Shi Gang.Address: 385 Hong Cao Road, Shanghai, 200233 ChinaTel. : (86) 21 6485 1900Fax : (86) 21 6485 1056

France, Limeil, PMLPhilips Microwave Limeil (PML) is a wafer foundryfor gallium arsenide (GaAs). Started as part of thelaboratory LEP, it has 20 years experience indesign and manufacture in GaAs.The main product focus is on RF MMIC for mobilecommunications. The foundry is part of theBusiness Line Telecom in Zürich, and employsabout 100 personnel. PML is ISO 9000 certified

and is actively involved in Philips PQA-90. Quality Manager is Jean-Louis Deviller.Address:22 Avenue Descartes, BP 15,94453 Limeil-Brévannes Cedex, France.

France, Sophia AntipolisWith more than 200 employees from about 15countries, Philips Semiconductors SophiaAntipolis comprises a truly international and multicultural working environment.At the heart of the Telecom Valley, around 75% ofthe employees are involved in product and testprogram development; the remainder in quality,marketing, product engineering and logisticsactivities. Sophia Antipolis is a main centre forPhilips Semiconductors telecom activities and hasfull business responsibility. Moreover, it is a keysupplier of advanced ICs to virtually every majortelecommunications company in the world.The Sophia Antipolis innovation centreconcentrates its resources on providing systemsolutions (standard products and ASICs) in thefollowing key areas: Cellular baseband technology (GSM, 2G+ and

3G) Bluetooth baseband and RF CMOS technology Embedded Cores (DSP, ARM) Communication Platform Development.The site has been certified to ISO 9001 since1994. The Quality System Manager is Claudia Salmen.Address:Philips Semiconductors Sophia,505, route des Lucioles, Sophia Antipolis,06560 Valbonne, FranceTel. : 33 4 92 96 1100Fax : 33 4 92 96 1101

42

India, Bangalore, BSCPhilips Semiconductors Bangalore SoftwareCentre (BSC) is part of the Philips SoftwareCentre, which also operates for other productdivisions.Started in 1996, BSC has a headcount of 180. It isa CTO resource organizationally under STG, butincluding departmental groups of RTG (Re-useTechnology Group), DTG (Design TechnologyGroup) and LTG (Library Technology Group). BLSoftware development is facilitated through adedicated group working on software projects forBU-TT (GSM), BU-CS (BL-Digital Media) and BU-EB (BL Networking). Recently added activities are a System lab and aTC-General.BSC is certified to ISO 9001 and CMM level 2.Quality Manager is Gopal Balu.Address :3 Sterling Square, Madras Bank Road,Bangalore 560 001, India.Tel. : (91) 80 222 92 00Fax : (91) 80 222 97 00

Malaysia, Seremban, PSSPhilips Semiconductors Serambam Snd, Bhd(PSS) is a 50/50 joint venture between Philips andMotorola for the manufacture of discrete surface-mount products: SOT23, SC59, SC70 and SC90,mainly targeted at the Japanese market. Thefacility was set up in 1994 and now employs morethan 500 personnel. PSS is certified to ISO 9002. The Quality Manager is P. Sivakumar.Address:Pt. No 12687, Tuanku Jaafar Industrial Park,71450 Seremban, Malaysia.Tel. : (60) 6677 28 53Fax : (60) 6677 30 99

Singapore, ATOATO is one of the Business Units within PhilipsSemiconductors, handling all the assembly andtesting business of the ICs for all the businesslines. ATO has three in-house plants: PSK inKaohsiung, Taiwan; PST in Bangkok, Thailandand PSC in Calamba, Philippines. In addition,ATO handles all subcontracting business, andalso assumes responsibility for the local test sitesat Nijmegen, Hamburg, Caen, Sunnyvale,Albuquerque, San Jose and San Antonio.ATO Innovation group is located at Nijmegen,PSK and PST.ATO has been headquartered in Singapore sinceJanuary 1997, in the Philips Asia-Pacific regionaloffice.The quality coordinator is Kim Tey.Address:Philips Electronics Singapore Pte Ltd.,620A, Lorong 1 Toa Payoh, TP2, Level 2,Singapore 319 762.Tel. : (65) 799 52 00Fax : (65) 799 51 99

Singapore, SSMCSystems on Silicon Manufacturing Co. Pte Ltd(SSMC) is a joint venture with TSMC and the localorganization EBI, with a 48% ownership byPhilips. SSMC has an 8 inch wafer fab for 30 kouts per month (60% going to Philips), with acapability to below 0.15 micron.The fab covers 71,760 square metres on threelevels, of which 14,028 square metres arecleanroom facilities with class 1 mini-environments. Services include Esort, FailureAnalysis and Reliability Testing. The site employsabout 600 personnel (planned to grow to 1000 by2002) and plans certification to ISO 9001 by June2001 and to QS-9000 before June 2002. TheQuality Systems Manager is Chia Beng Chye.

43

Address:70 Pasir Ris Drive 1, Singapore 519527.Tel: (65) 780 8707Fax: (65) 786 9529

Taiwan, Hsinchu, TSMCTaiwan Semiconductors Manufacturing CompanyLtd (TSMC) is a joint venture with a 30% share byPhilips. It contains two wafer fabs for 150 mm, andthree for 200 mm. TSMC employs about 3400 staff and is certified to ISO 9001.The Quality and Reliability Director is Dr. Y. W.Yau.Address:121 Park Ave 3, Science-based Industrial Park,Hsinchu, Taiwan (ROC).Tel. : (886) 35 780221Fax : (886) 35 781546

The Netherlands, EindhovenThe Eindhoven site Beatrix houses, among others, the headquarters of Philips Semiconductors and the System Lab Eindhoven.Headquarters contains three Business Units and all staff departments of the PD, including QMS.Situated on other sites in Eindhoven are: SalesEurope (HKK), Information Office (TAM),International Marketing & Sales (HVG) and theASIC Service Group on the Nat.Lab. site (WAY). The total headcount for PD Semiconductorsin Eindhoven is about 600.There are separate ISO certificates forheadquarters, System Lab and RSO Europe.

The Netherlands, Eindhoven (SLE)Philips Semiconductors Systems LaboratoryEindhoven (SLE) is located at the Beatrix site inEindhoven. It employs about 275 personnel whowork on hardware and software systems innovation and applications in a wide spectrum of business areas, ranging from Digital Video andAudio Broadcast to TV, Telephony, Paging, Radio,

Monitors and Microprocessors. The lab is theguardian of system architectures and is thecompetence centre for platform-based systems,software concepts and platform roadmaps. It develops reference boards for systemimplementation based on platforms. The laboratoryis ISO 9001-certified as a separate unit. TheQuality System Manager is Bram van den Berge.Address:Philips Semiconductors B.V.,Building BE / Hurksestraat 19,P.O. Box 218,5600 MD Eindhoven, The Netherlands.Tel. : (31) 40 2723148Fax : (31) 40 2722764

USA, TempePhilips Semiconductors Tempe was originally thedesign facility for VLSI Technology. The facility isthe headquarters for the Standard Analog BusinessLine. Additionally the facility houses a broad arrayof functional organizations including ConsumerSystems, Networking, Emerging Businesses, CTO(DTG, LTG, EDP, EDP/ARM), Field Sales,Components and Finance.Located in the Arizona State University ResearchPark, Philips Semiconductors Tempe currentlyhas 238 employees. The facility size for the fourbuildings is a combined 12,800 square metres.The facility is currently engaged in activitiesrelated to ISO 9000 registration, scheduled for2002, and imple-mentation of Product CreationProcesses. The Tempe facility is ISO 14001registered. The Quality Manager for StandardAnalog, Dennis Reed, is based in Tempe.Tel: 1 480 752 6290

Philips Semiconductors Tempe,8375 South River Parkway,Tempe,Arizona 85284,U.S.A.

44

Daniel GloorDisplay Systems

Zürich

Karl-Heinz GressmannCar Infotainment

Hamburg

Quality/Improvement managers

Gopal BaluBSC

Bangalore

Jaap BisschopQMS

Nijmegen

Steven BraderBEST-ATOSingapore

Wim BremerMOS-3

Nijmegen

Hans BuisNetworking Infrastructure

Nijmegen

Gerard CourteilleTunersCaen

Tim CrispinHazel Grove

Alice DarquennesQMS

Eindhoven

Michel DubeeConsumer

Caen

Rik ElzingaStadskanaal

Jojo FabiaPSC

Calamba

Bill FullQTS-Reliability

Sunnyvale

Joy GandhiComputingSan Jose

Jean-Emmanuel GilletCordless

Caen

45

Gerard de GrootMOS4YOUNijmegen

Maureen HeuerBEST-foundries

Eindhoven

Bill HobdyQTS-QualitySunnyvale

Jan Holsbrinkfab-AN

Nijmegen

Roger InnesBEST-DMINijmegen

Tim JohnsonStorage Systems

Southampton

Gary HuffmanMicrocontrollers

Sunnyvale

Jaap KeymanQMS

Eindhoven

Thierry KiefferCellular 3GSM

Zürich

Bob Incerto

Fishkill

Leo KlerksQMS

Eindhoven

Michael KüchlerGA Products

Hamburg

Ton van de KraatsBEST-CTOEindhoven

Betty LiangSales Asia

Taipei

Apichai LertapirukPST

Bangkok

Adam LaiPSK

Kaohsiung

46

Gijs LijbersQMS

Eindhoven

Yves MeheustPlantCaen

Chris MerkleyConnectivityAlbuquerque

Helmuth MeyerIdentification/BEST-ID

Hamburg

Marty MichaelsSales NASan Jose

Sankara NarayanQMS

Eindhoven

Henk OttenQMS

Eindhoven

Lewe PetersenSite Programs

Hamburg

Roland RadünzPICTOHHamburg

Dennis ReedStandard Analog

Tempe

Roberto RappFoundry

Böblingen

Doug RidleyCellular

Sunnyvale

Anne RouchierSales Europe

Paris

George Song

Guangdong

Claudia SalmenBEST-MC

Sophia Antipolis

Kam Leung LukEDL

Hong Kong

47

Ray SuenSite/BEST-Net

Sunnyvale/San Jose

Emma TomeldenPSPI

Cabuyao

Kees de VaalPower Management

Nijmegen

Paul van WarmerdamAnalog Audio

Nijmegen

Theo VedderQMS

Eindhoven

Horst WaschkewitzFoundryHamburg

Masaaki SuganumaSales Japan

Tokyo

Widge StambackFoundry

San Antonio

Joop WillemsenLogic

Nijmegen

P.S. YenConsumer Systems

Taipei

Andrew WhittardBEST-IMS

London

49

Part 2

References

50

Abbreviations

ACLAOQAQLASICATDATOATSBESTBLBUCHAMPCLIPCMMCMOSCPCNCpkCQBCQSCSCWQIDCSDMIDODDOEDSEBEDIEDPEFQMEMCESDFETFITSFMEAGQSGSOICIDIECIMSIPMMIQMM

acceptance control limitaverage outgoing qualityacceptable quality levelapplication specific ICacceptance for type developmentassembly & test organizationacceptance for type studybusiness excellence through speed and teamworkbusiness linebusiness unitcomplaint handling and management programconfirmed line item performancecapability maturity model (for software)complementary MOScustomer product/process change notificationprocess capability indexcorporate quality bureaucustomer qualification samplesconsumer systemscompany-wide quality improvement(BU) Digital Computer Systems(BU) Discretes & Multimarket ICsdiscontinuation of deliverydesign of experiments(BU) Display Solutionsexecutive board / (BU) Emerging Businesseselectronic data interchangeelectronic data processingeuropean foundation for quality managementelectromagnetic compatibilityelectrostatic dischargefield effect transistorfailures in time standardfailure modes and effects analysisgeneral quality specificationGlobal Sales Operationsintegrated circuit(BU) Identificationinternational electrotechnical commissioninternational marketing and salesinternational product marketing managerinternational quality managers meeting

51

ISOISRJEDECJITLCLLSLMaCSMCMISDMoCMOSMSLMTBFMTTFNetNOOEMPBEPDPPMPQPQA-90PQRAPSCQAQCQDSQFDQICQITQLPQMLQMSQOSQPLQPMQ&RQSPRFSRLIPROOTSSAC

international organization for standardizationinitial sample releasejoint electron device engineering counciljust-in-timelower control limitlower specification limit(BU) Mainstream Consumer Systemsmanagement council / (BU) Mobile Communicationsmanufacturing instructions and standardization department(BU) Mobile Consumer Systemsmetal-oxide semiconductormoisture sensitivity levelmean time between failuresmean time to failure(BU) Networkingnational organizationoriginal equipment manufacturerPhilips business excellenceproduct divisionparts per millionpacking quantityPhilips quality award (for the nineties)Product quality and reliability assurance (database)Philips Semiconductorsquality assurancequality control (or quality circle)quality description sheetquality function deploymentquality improvement competitionquality improvement teamquality leadership program (AT & T)qualified manufacturers listquality management semiconductorsquality operating system (Ford)qualified products listquality policy meetingquality and reliabilityquality supplier plan (Chrysler)release for supplyrequested line item performancerapid on-line overalltraceability systemsemiconductor assembly council

52

SEMSMARTSMDSOARSODSOTSPCSPQSTSTOTOPSTQCTQETQMTQSTTLUCLUSLZD

scanning electron microscopespecific, measurable, ambitious, realistic, time-phasedsurface-mounted devicesafe operating areastandard outline diodestandard outline transistorstatistical process controlsmallest packing quantityship-to-stocktransistor outlineteam-oriented problem solvingtotal quality controltotal quality excellence (Ford)total quality managementtotal quality systemtransistor-transistor logicupper control limitupper specification limitzero defects

We make such investigations into newsemiconductor structures, where entirely newprocesses or materials are involved; as a resultactivation energies for all commonly-usedmethods of construction are well documented, andsome are given below.

Activation energies for commonfailure mechanismsThe activation energies for some of the majorsemiconductor failure mechanisms are given inthe Table below. These are generalized estimatestaken from published literature and internalreliability studies. In cases where a specific failuremechanism has been more thoroughlycharacterized for its thermal acceleration, thatestimate of EA should be used.If no failure analysis data is available, and areliability estimate is required, an activationenergy of 0.7 eV should be used.

Activation energies for common failuremechanisms

Failure mechanism

Mechanical wireshortsDiffusion and bulk defectsOxide defectsTop-to-bottom metal shortElectromigrationCharge trappingElectrolytic corrosionGold-aluminium intermetallicsGold-aluminium bond degradationIonic contaminationAlloy pitting

53

Acceleration factors

EA 1 1

k T1 T2 (1)

A major factor in determining the reliability ofsemiconductors is the total stress applied by theapplication. The total stress will be the sum ofseveral components, both electrical andenvironmental. Principal environmental stressesare vibration and humidity; electrical stresses arevoltage and current. Operating temperature,resulting from ambient temperature and heat dueto power dissipation, is, however, the mostimportant applied operating stress where asemiconductor is otherwise generally operatedwithin its ratings.

Arrhenius' equationSwedish chemist S. Arrhenius' expression for theeffect of temperature on the velocity of a chemicalreaction is now widely used to predict the effect oftemperature on electronic-component failure rate.It is generally used to derive an acceleration factorA, the ratio of the expected failure rate atoperating temperature T1 to the known failure rateat test temperature T2 :

A = exp

Quantity EA is the activation energy for theexpected failure mechanism.T is absolute temperature (K).k is Boltzmann's constant (8.6 x 10-5 eV/K).

Defining EA is obviously critical to the use of theexpression. Its value depends on deviceconstruction: the materials used and theprocesses used to combine them. Determining thevalue of EA requires a long series of life tests atdifferent temperatures with analysis of failures toidentify failure mechanisms. Weibull charts areoften used to assist in the processing of the datafrom such tests. The whole procedure involvesmany thousands of hours of testing.

Activationenergy (eV)

0.3-0.40.3-0.40.3-0.40.50.4-1.20.060.8 -10.8 -21-2.21.021.77

54

HumidityHumidity can have a significant effect on thereliability of some semiconductors. This isespecially true where aluminium metallization andplastic encapsulants are used, although thesilicon-nitride passivation used on the majority ofPhilips semiconductors greatly reduces the effect.The Peck model is used to predict theacceleration factor (A) due to the combined actionof temperature (T) and humidity (H):

A = ·exp ( · )

Here, H1 and H2 are the humidities associatedwith temperatures T1 and T2, respectively.The values for the other parameters are n = 3,Ea = 0.9 eV, k (Boltzmanns constant) = 8.6 x 10-5 eV/K.

ExamplesFigures 1 and 2 of acceleration factors calculatedusing Eqs (1) and (2) show the reduction in failurerate expected at various operating temperaturescompared with those observed during life testing.Note that the temperatures are junction (die)temperatures in all cases.

Thermal CyclingThermal cycling induces stresses due todifferences in expansion coefficients of thedifferent materials in semiconductor components.The cyclic behaviour of the stresses can causefatigue effects, leading ultimately to failuremechanisms such as cracking and shift ofpassivation and metal layers or wire break and

bond lift. The degradation due to temperaturecycling can be described by the simplified Coffin-Manson equation for low-cycle fatigue effects:

A =

∆Tstress and ∆Tuse are the temperatureexcursions during the test and use conditions,respectively. A is the acceleration factor, whichrelates the number of cycles with ∆Tstress to thenumber of cycles with ∆Tuse. The exponent mdepends on the failure mechanism (see tablebelow).

Failure mechanism Coffin-Mansonexponent m

A1 wire bond failure 3.5intermetallic bond fracture 4.0PQFP delamination/bond fail 4.2Au wire bond heel crack 5.1interlayer dielectric cracking 5.5chip-out bond failure 7.1thin film cracking 8.4die-attach Rth degradation 9.35

The model can be used to calculate life times forknown failure mechanisms at use conditions, andto compare different stress conditions. Thetemperature ranges must be corrected for thestress-free temperature range.

Ea 1 1

k T1 T2

H 2

H 1

∆Tstress

∆Tuse

m

n

(2)

55

Fig. 2 The effect of two levels of operating humidity on deceleration

Fig. 1 The effect of reducing temperature on failure rates determined at 125 °C and 150 °C.

operating (junction) temperature (°C)

0.5

0.2

0.1

0.05

0.02

0.01

0.005

0.002

0.001

1

0.1

0.01

50 100 150

0 20 40

operating (junction) temperature (°C)

125 150 175test temperature (°C)

150 175test temperature (°C)

test humidity 85%operating humidity

EA = 0.7 eV

1

50%85%

125

150 175125

60 80 100 120 140 160 180 200

(A)-1

(A)-1

56

Acceptable Quality Level (AQL)

AQL is 'the maximum percentage defective that, forpurposes of sampling inspection, can beconsidered satisfactory as a process average'.AQL is not a licence to ship rejects. It's the basis ofsampling systems, which seek to assure aconsistent level of quality to purchasers ofelectronic components.However, when the quality level of a semiconductorproduction process is very high, a sampling systemto prove this lot-by-lot becomes fairly inadequateand inefficient.

AQL values for acceptance testsOur standard internal AQLs for Group A tests havebeen regularly reduced in the past, but we'rereluctant to reduce them significantly from theirpresent 0.1% level because: sample size increases rapidly for AQLs less

than 0.1%. For quantity production, a fixedsampling plan, and level ll inspection, thesample size for a 0.1% AQL is 125. However,sample size increases to 315 for a 0.04% AQL.The larger the sample the higher are theadministration and handling costs. This isparticularly true for visual and mechanical tests.

our existing average process quality level formany types is already so high that loweringAQLs at acceptance testing will hardlycontribute to further improvement.

For these reasons the AQL-system is only used todefine the size of the sample used for final checkbefore delivery. No lot leaves our factories if anydefect is found in this final check.Thus, AQL methods are only used to provideassurance that processing errors have notoccurred (to avoid rogue lots), and to confirm thatour in-process controls are effective and providedata which allows measurement of the PPM level.

Zero defects and PPMThe acceptable defect level can only be zero. Lotswith defects are not accepted for delivery and willbe 100% retested. The real quality level is notdetermined by AQL but by PPM (see section onPPM).

57

AQL

BATC

H SI

ZE A

ND IN

SPEC

TIO

N LE

VEL

AQL

ACCE

PTAN

CE/R

EJEC

T NU

MBE

R (N

orm

al in

spec

tion)

COM

BINE

D SI

NGLE

(S) A

ND D

OUB

LE (D

) SAM

PLIN

G P

LANS

FO

R NO

RMAL

INSP

ECTI

ON

(see

MIL

-STD

-105

0, IE

C 41

0, IS

O 2

859)

SAM

PLE

SIZE

SING

LEDO

UBLE

SD

DD

DD

DD

DD

DD

SS

SS

SS

SS

SS

CODE LETTER

S4S4

S3S3

S3S3

S3S3

S3

S2S2

S2

S4

S4S4

S4

S4S4

S4II

II

II

II

II

III

III

III

III

III

III

III

III

III

III

III

III

II

II

II

II

II

III

I

I

I

I

I

I

I

IE F G H J K L M N P Q R

281

to 500

151

to 280

91 to 150

51 to 90

26 to 50

501

to

1200

1201 to

3200

3201 to

1000

0

1000

1

to

3500

0

3500

1

to

1500

00

1500

01

to

5000

00

5000

01

to

Ove

r

0.01

00.

015

0.02

50.

040

0.06

50.

100.

150.

250.

400.

651.

0

13 20 32 50 80 125

200

315

500

800

1250

2000

8 16 13 26 20 46 32 64 50 100

80 160

125

250

200

400

315

630

500

1000

800

1600

1250

2500

0/1

0/1

0/1

0/1

0/1

0/1

0/1

0/1

0/1

0/1

0/1

0/2

0/2

0/2

0/2

0/2

0/2

0/2

0/2

0/2

1/2

1/2

1/2

1/2

1/2

1/2

1/2

1/2

1/2

1/2

1/2

1/2

1/2

1/2

1/2

1/2

2/3

3/4

3/4

3/4

3/4

3/4

3/4

3/4

3/4

3/4

3/4

3/4

3/4

3/4

3/4

1/4

1/4

1/4

1/4

1/4

1/4

1/4

4/5

4/5

4/5

4/5

4/5

4/5

4/5

5/6

5/6

5/6

5/6

5/6

5/6

2/5

2/5

2/5

2/5

2/5

2/5

6/7

6/7

6/7

6/7

6/7

6/7

7/8

7/8

7/8

7/8

7/8

3/7

3/7

3/7

3/7

3/7

8/9

8/9

8/9

8/9

8/9

5/9

5/9

5/9

5/9

10/1

1

10/1

1

10/1

1

10/1

1

7/11

7/11

7/11

14/1

5

14/1

5

14/1

512

/13

12/1

3

12/1

3

12/1

3

18/1

9

18/1

9

18/1

9

11/1

6

11/1

621

/22

21/2

226

/27

26/2

7

3/4

2/3

2/3

2/3

2/3

2/3

2/3

2/3

0/3

0/3

0/3

0/3

0/3

0/3

0/3

0/3

1/2

1/2

*

*

*

*

*

*

*

*

*

*

*

Acco

rdin

g to

MIL

-STD

-105

DUs

e fir

st s

ampl

ing

plan

bel

ow a

rrow.

If

sam

ple

size

equa

ls or

exc

eeds

lot

or b

atch

size

, do

100%

insp

ectio

n.

Use

first

sam

plin

g pl

an a

bove

arro

w.0/

1 m

eans

0 =

acc

epta

nce

num

ber/1

= re

ject

ion

num

ber e

tc.

* use

cor

resp

ondi

ng s

ingl

e sa

mpl

ing

plan

.

58

Assembly quality control

unsawn wafers100% pretested

sawn wafers100% pretested

lead frame, glue

die inspection

sawing & mounting

reject lots

die bonding

curing

wire bonding

preseal lotinspection

reject lots

product moulding

die coating

preseal visual 100%

100%

post curing

solder plating

laser marking

Ink marking andbaking

cutting and bending

open/short test

visual/mechanicalinspection

visual/mechanicalacceptance

gold wire

mould compound

marking ink

production process

in-line inspection

incoming material inspection

incoming material

quality department lot acceptance

packing box,tube, stop

final testing DCand function

reject lots

reject lots

reject lots

packing

STORE

acceptance testingAC, DC and function

shipment inspection

59

BEST

BEST (Business Excellence through Speed andTeamwork) is the Philips way to BusinessExcellence.The BEST program was launched on July 6th1999 by Mr. C. Boonstra. It is a managementprocess that drives the company to world-classperformance levels through improvement andalignment of all business processes.Philips Business Excellence (PBE) assessmentand recognition are key elements in the BESTprogram. The main tools of BEST are shown inthe PDCA cycle (Fig.1) as:

BBS (Business Balanced Scorecards)A Balanced Scorecard translates the businessplan into the vital few goals that mark the path toexcellence. The power of a scorecard lies in itsability to align everyone in the organization on thechosen business priorities and to make thenetwork of cause and effect relationship visible inBusiness Reviews.

PST (Process Survey Tools)Process Surveys establish the relative maturity ofa process and indicate what should be done toreach the next step. Evaluation criteria and stages

of maturity of the main business processes aredescribed in process-specific terms.

HQA (Headquarter Audits)A Headquarter Audit is a means to accelerateimprovement by shared learning and assessingthe quality leadership provided by theheadquarters of Business, Divisions and theCompany.

PBE (Philips Business Excellence)PBE assessments are an essential part of PhilipsBusiness Excellence and identify areas forimprovement using a reference model forexcellence that puts enablers and results inperspective. Enabler criteria deal with how resultsare achieved, while the results criteria focus onwhat the organization has achieved. The scoreagainst the PBE model is an indicator for theprogress towards Business Excellence.

KM (Knowledge Management)Applying the lessons learned is the completion ofthe PDCA cycle. Knowledge Management meansto expand Applying the lessons learned toSharing the lessons learned and learning fromthe lessons others learned.

KM BBS

HQAPBE PST

Fig.1. Plan-Do-Check-Act (PDCA) cycle.

Chemical content SOT319substance mass

Device part (mg)leadframe Cu-alloy* 320

SnPb15 plated 10active device doped Si 20bond wire Au 3chip coating silicone gel** 10encapsulation partially brominated 1360

epoxySiO2 ≈ 80%

Sb2O3 < 2% as SbTBBA < 0.6% as Br

Packing material (tray pack/dry packed)substance mass

Device part (g)box cardboard 113JEDEC trays copolymer, 850

carbon loadedbag aluminium laminated 47

polyethylenedrying agent alumina silicate 88humidity ind. Paper + CoCl2 1.2strap polypropylene 2.2labels paper 2.6seal acrylate 0.2* optional: NiFe leadframe ** not used in all packages

In addition to the brochure, a series ofenvironmental certificates is available, based onthe EACEM and VDA list of hazardous substances.

Information per package type is available on theinternet, for direct access on:http://www.philips.semiconductors.com/profile/env/information/

60

Chemical content of semiconductors

Maximum benefit – minimum impactPhilips new technologies result in shared benefitsfor our customers, balancing maximumtechnological benefits with minimum environmentalimpact. This starts with innovative chip designs,which result in more compact applications withfewer external components. These chips,moreover, also reduce the consumption of rawmaterials, reduce packaging sizes and saveenergy. And as more and more hazardousmaterials are eliminated from our products, costsof disposal at the end of their useful life get lower,again minimizing environmental impact.

All this is embodied in the concept of eco-design.Eco-designed products are those designed toimprove functionality and minimize the environmentalimpact of a product through all stages of its lifetime,from source material, through the manufacturingprocess and working life to the end of its useful life.By 2002, 75% of Philips Semiconductors productswill be eco-designed.

Maintaining openness on environmental issuesPhilips Semiconductors chemical content brochure(9397 750 04906) is intended to foster anatmosphere of openness on environmental issues.It does this by providing clear and comprehensiveinformation on the composition of representativeexamples of every product manufactured by PhilipsSemiconductors. Typical examples are shown inthe following tables.

Fig. 1 Flowchart for complaint processing.

61

Complaint processing / CHAMP

Communication is vitalAs quality improves, and as our own testingreveals fewer defects, complaints become a veryimportant source of data for corrective action.Moreover, data from complaints providesfeedback on quality where it really matters in theapplication. It is therefore essential thatcomplaints are correctly routed and handled.Particularly, reject circumstances must becorrectly documented, rejects must be analyzed indetail, and the results must be communicated toeveryone involved in a short period of time.

Routeing of complaints and rejects(see Fig. 1)Complaints from customers are normally receivedby the customercontact in the local Sales Office,who will usually send the complaint directly to theresponsible Quality Centre. For ICs, this is thediffusion centre; for discrete semiconductors itsthe regional customer support centre. The rejects are sent from the Sales Office to theresponsible Quality Centre by high-speed courier.Where a PPM cooperation exists with thecustomer, the rejects may be sent directly by thecustomer.

Information needed for complaintprocessingFor fast processing of the complaint, thedocumentation supplied with the rejects shouldinclude the following information: customer name customer-contact Philips part number lot size quantity tested quantity rejected quantity returned date code inspection and test reference number purchase order reference test temperature test equipment location of failure (receiving inspection,

qualification testing, assembly or field).

Customer

Sales Office

responsible Quality CentreQuality Management

rejects/complaint documentation

responsiveness data

responsiveness data

confirmation/final report

62

Follow-up of the complaintAfter verification of the defect, the products arereturned to the customer (see Return shipments).It is also possible that the rejects are withinspecification. In that case the products are notreturned, the customer is informed and applicationadvice is offered or a correlation exercise isstarted.

Analysis and corrective actionsAfter verification of the defect, the defectiveproducts are analyzed to establish the root cause.Each Quality Laboratory is equipped withsophisticated facilities for failure analysis, andstaffed by specialists for identifying failures andfailure mechanisms. After the root cause has beenestablished, the 8-D method (see Qualitytechniques and tools) is used to solve theproblem. Step 3 of this method assurescontainment actions to halt the delivery of furtherproducts which may have the identified failure.Steps 6 and 7 ensure that permanent correctiveactions are taken to prevent repetition of theproblem.

ResponsivenessThe complaint procedure determines theinformation flow and time limits to process thecomplaint and communicate to the customer.Receipt of the complaint at the responsible QualityCentre is confirmed within 3 working days. Thetime to the final report varies between 8 and 15calendar days, depending on complexity. Theelapsed times for the various steps in processingthe complaint are recorded and reported asresponsiveness indicators to QualityManagement.

CHAMPTraditionally, PD Semiconductors has utilizedmany different complaint-handling systems, whichwere independent of each other. Thus one

Business Line did not have access to anotherssystem, sharing lessons learned was notpossible and there was no one face to thecustomer. Also Management Overviews (numberof complaints, average throughput time etc) weredifficult to obtain. To improve this situation, inMarch 1998 a Project Team was set up to realizeCHAMP, which stands for Complaint HandlingAnd Management Program. CHAMP was madeoperational at the end of 1999.The purpose of CHAMP is to offer a worldwideaccessible common system for the handling ofCustomer Complaints. Worldwide means thatall Business Lines, Sales Organizations andManufacturing Centres will use CHAMP.Common means that all information will be inone central database, from the smallest detail tothe Final Report which is sent to the customer.With appropriate search functionality a user willbe offered the possibility to search for similar rootcauses and corrective actions.CHAMP offers the following complaint processflow (Fig. 2) to the users: Registration of a complaint Routeing the complaint to the responsible

Business Line or Manufacturing Centre Verification of the Complaint Initiating Containment Actions Detailed (Electrical/Physical) Analysis Initiating Corrective Actions Writing the Final Customer Report and closing

the complaint.

It is expected that around 1000 users worldwidewill have a direct access to CHAMP. In the longerterm, it is planned that selected customers will getaccess to CHAMP via the Internet to registercomplaints, to view the progress and to retrievethe Final Customer Report (of course only for theirown complaints).

63

Further information on the progress of CHAMPcan be obtained from the CHAMP web site:http://pww.sc.philips.com/qms/champ/index.htm

1

ComplaintEntryand

Routeing

2

Verification

3

Analysis

6

FinalCustomer

Report

7

ClosureComplaint

Fig. 2 CHAMP complaint process flow.

4

ContainmentActions

5

CorrectiveActions

Conformity confidence levelsMost of the Quality Control testing of oursemiconductors uses relatively small samplessince such testing is expensive in time, skilledpeople, and elaborate equipment. However, it'sobviously vital that the results should reflect thosethat would be obtained from testing largersamples, or whole lots. Statistical methods enableus to set probable limits to the results that wouldbe obtained from testing a whole lot, on the basisof the results obtained from tests performed on asample. The most common of these methodsinvolves confidence limits.As with most statistical predictions, it is firstnecessary to assume a distribution of the testresults. For confidence-level calculations, however,this is not the distribution of the results of just onetest, but the distribution of the results of a wholeseries of identical tests (a population). Once the distribution is known, it is possible tocalculate a range of values that, to apredetermined probability, contains the true value(that would be obtained from the testing of a wholelot). This range of values is termed the confidenceinterval, and the probability of it containing the truevalue is the confidence level, Fig. 1.

Reliability confidence levelsDuring evaluation of the results of life (endurance)tests, a correction is used that increases theactual (observed) number of failures to the valueat the upper end of the confidence interval. Thiscorrected result, known as the assessed value, isthen described as being to a given upperconfidence level (UCL): the actual confidencelevel used being the probability that the true valueis not less than the assessed value. The usualconfidence level applied to life test results is 60%,although 90% or even 95% confidence levels arealso used. (A useful rule of thumb is: to correct agiven life-test result to 60% UCL, just add one tothe observed number of failures. Thus 0 failuresbecomes 1. Similarly, 1 failure becomes 2, etc.Actual values are 0.92 and 2.02. The method isacceptable up to 10 failures (11.52 at 60% UCL)).The statistical calculations used to determineconfidence levels are similar to those used toderive process averages from Cpk values instatistical process control.The differences between observed life test resultsand assessed values to upper confidence limits of60% and 90% (UCL) are shown in Figs 2 to 4.

Fig. 1. A typical distribution curve (here Gaussian) witha confidence level (1 - α). Note that quantities α and(1 - α) refer to the areas under the curve. This exampleis a 'two-tailed' confidence interval, symmetrical aboutthe median. Single-sided confidence levels are alsoused, especially with reliability data, but withasymmetrical distributions, where α is collected to oneside of the confidence interval. Where α is at the highend of the distribution, the associated confidence level istermed the 'upper confidence level UCL'.

64

Confidence level

Median

Confidence interval1 - α

α/2 α/2

Figure 2 shows that correcting an observednumber of failures to a given confidence levelshifts the derived failure rate by a constantamount irrespectiive of the number of device-hours testing. On the other hand, Fig. 3 showsthat, as the observed number of failures increasesin a fixed number of device-hours testing, so thecalculated failure rates tend to converge to theobserved value. Finally, Fig. 4 shows thecalculated 60% and 90% UCL failure rates thatcould be obtained from tests of various durationson samples from a lot of known failure rate inthis case 10x10-6/h.These examples show that, to obtain a failure rateclose to the real value, the total device-hourstesting and the test conditions must be sufficientto generate a significant number of failures. Thetests that we use in Group C are carried out underAbsolute Maximum Rating (accelerated stressconditions) to maximize failure rates. Even so, it isapparent that, even when no failures areobserved, the accumulated results of 2 to 3 years'Group C life testing are required in order todemonstrate that we have achieved our currenttarget failure rate (at Absolute Maximum Ratings)of 10-6/h to a UCL of 60%.

Fig. 2. Where the number of observed failures C isconstant, the effect on failure rate λ of correction to agiven (upper) confidence level is independent of thedevice-hours of testing (nt).

Although 60% may seem a low value to use forsemiconductors, the actual confidence level for agroup of devices (in a circuit) increases rapidlywith the number of devices. Even for four devices,whose failure rate was calculated individually at aUCL of 60%, the combined confidence level isabout 90%, due to the sum of the appliedcorrections.

Fig. 3. As the number of observed failures increases, sofailure rate λ at various confidence levels converges.

Fig. 4. Here, the true failure rate for a wholesemiconductor lot is assumed to be 10-6/h. Even if thesample (here, n = 200) is assumed to be representativeof the lot, many device-hours of testing are requiredbefore the UCL values approach the real value. Wherethe correction required for a given UCL results in a largechange in failure rate, the confidence band is said to bewide.

65

1 2 5 10

10

λx10-6

(h-1)

1

0.1

nt x10-6(h-1)

90% UCL

C, constant = 2

60% UCLobserved

1000

λx10-6

(h-1)

100

10102 103 104

t(h)

90% UCL60% UCL

n, constant = 200(λ, constant = 10 x 10-6/h)

100

λx10-6

(h-1)

10

11 2 5 10

c

90% UCL

60% UCL

observed

nt, constant = 106h

The cost of quality can be a highly significantelement of the profit-and-loss statement,particularly in the long term. It is the cost ofachieving quality goals. Quality cost reportingprovides a means for evaluating effectiveness,and establishing the basis for internalimprovement programmes. It's important thatquality costs are regularly reported and monitoredby management. They should be related to othercost measures such as sales, turnover and addedvalue, and should:

evaluate the adequacy and effectiveness of thequality management system

identify additional areas requiring attention establish quality and cost objectives.

How is cost of quality defined?In real terms, the cost of quality is the summationof the cost of conformance and the cost of non-conformance, where the cost of conformance is(prevention costs + appraisal costs) and the costof non-conformance is (internal failure costs +external failure costs).So: cost of quality =

(prevention costs + appraisal costs)+

(internal failure costs + external failure costs).

These costs are defined below:

Prevention costs(the costs of trying to ensure that we do thework right first time)These are the costs incurred in preventing thefuture recurrence of non-conformances. They aredirected towards the satisfaction of the customer'squality, reliability and safety requirements in alloperations with the first and all succeeding units ofproduct produced.Typical prevention costs are:

quality planning in design, manufacturing andquality systems

process optimization quality training developing and implementing reliability

measurement and calculation methods, qualityanalysis methods, and quality informationsystems

evaluation of vendors, and satisfying customerrequirements.

Appraisal costs(the costs of checking to make sure we did thework right first time)These are the costs incurred in measuring,evaluating and controlling current production toassure conformance to requirements, includingcertain costs of related equipment and services.

Typical appraisal costs are: planned inspections laboratory testing process control quality audits destructive testing maintenance and calibration of test,

measurement and inspection equipment.

Internal failure costs(the costs we incur when we discover wedidn't do the work right first time)These are the costs generated before a product isshipped, as a result on non-conformance torequirements.

Typical internal failure costs are: scrap of products and materials for quality

reasons rework or repair downgrading fault-finding of quality problems in production

66

Cost of quality

67

Fig. 1 Quality costs model.

Econ

omic

bala

nce

Cost

s pe

r uni

t of p

rodu

ct Total costs

Failure costs

Prevention andappraisal costs

Measure of quality (quality level)

tracing and repair of non-conforming productsand materials

re-inspection and re-test on non-conformingproducts and materials

losses caused by downtime damage caused by internal transport or storage costs of extra handling or storage.

External failure costs(the costs we incur when the customerdiscovers we didn't do the work right firsttime, and demands replacement orcompensation)These are the costs generated after a product isshipped, as a result of non-conformance torequirements.

Typical external failure costs are: complaints investigation returns after-sales costs (free replacements,

guarantees, compensation etc)

damage due to transport (if covered by deliveryterms)

claims because of non-conforming materials orproducts delivered.

Balancing of quality costsAn important aspect of quality costs is thepossibility of reducing internal and external failurecosts by investing in prevention and appraisal.As a result, the total quality costs reduce while thequality level improves.In the quality costs model (Fig. 1) this leads to aneconomic balance, where the optimum in qualitycost is reached.

This optimum is not fixed in time. By continuousimprovement (investment in prevention), themanufacturing process improves in time, resultingin a higher quality level at the same cost. In thequality costs model the graph for prevention andappraisal costs should thus get lower in time.

68

Customer Notification / CPCN / DOD

When Philips Semiconductors products are to bechanged or withdrawn from the market, a CustomerNotification is sent in advance to the customer.This Notification is sent by the sales represen-tative, either in the form of a Customer Product/Process Change Notification or a DiscontinuationOf Delivery Notification.

Customer Product/ProcessChange Notification (CPCN)When manufacturing processes are to bechanged, customers are notified 90 days beforethe change, by a Customer Product/ProcessChange Notification.Notifiable changes are those affecting form, fit,function or reliability of the product. Whencustomer agreement or comment is invited, thereare two stages of Customer Notification: theadvance notification and the final notification,respectively 90 days and 30 days beforeimplementation of the change.The CPCN handling process is described inquality standard SNW-SQ-650. Generation,tracking and sign-off of (advanced) CPCNmessages are handled by a web-based tool:the CPCN database (only for internal use).

Discontinuation Of Deliverynotification (DOD notification)When a product is being withdrawn from themarket, customers are notified in advance by aDiscontinuation Of Delivery notification, whichinvites the customer to place last time orders.Where available, the DOD notification will alsoindicate the replacement type.For single-source products the customer is given9 months prior notification; for multiple-sourceproducts the time period is 6 months.DOD notifications are only sent to customers twicea year, and contain products from all ProductGroups within Philips Semiconductors.Discontinuation of Delivery is also called pruning.The DOD process is described in quality standardSNW-SQ-651.

69

Customer-specific labeling

Many customers use barcode readers to checkthe semiconductor product-type at their receivingpoint and sometimes during assembly. To assistour customers we can supply products withcustomer-specific labels containing, for example,customer part number, supplier code and/orcustomer order number. Such labels are appliedby regional sales operations, under the control ofthe regional sales label coordinator.If a customer requests a specific label, our salesrepresentative will discuss the possibilities withthe regional sales label coordinator.For most requests a label design will already exist,and the regional sales label coordinator willarrange a sample of a suitable label design forcustomer agreement.Once the label design has been agreed by thecustomer, the design is implemented in thecustomer-specific label system. The customer partnumbers and supplier codes are stored (and mustbe maintained) in cross-reference lists in thecomputer system for handling customer orders.Varying information, such as customer ordernumber, will be supplied in the order lineinformation. A label program in the warehouseuses this data with the label design code, toproduce the label as requested by the customer.The regional sales label coordinators are:

Europe: Ruud van Leeuwen,E-mail: [email protected]

USA: Michael Beckstrand, E-mail: [email protected]

Asia: I. Chuang, E-mail: [email protected]

The customer-specific labeling process isdescribed in quality standard SNW-SQ-407.The worldwide labeling coordinator is Gijs Lijbers.E-mail: [email protected]

70

Drypack

0.40

0.35

0.30

0.25

0.20

0.15

0.10

0.05

00 4 8 12 16 20 24

drying time (h)

Fig. 1 Moisture content as a function of drying time for ICs in a selection of large surface-mount packages.

PLCC68QFP120QFP64QFP48

moi

stur

e le

vel (

%)

If infrared or vapour-phase soldering is used tosurface-mount an IC in a large plastic package(QFP or PLCC), vapourization of the small amountof moisture absorbed by the package duringstorage can increase the internal pressure to suchan extent that the plastic cracks.To provide an immediate solution to this problemof moisture-cracking, these ICs are packed in aresealable moisture-resistant plastic packet calleda Drypack.A Drypack is a laminated plastic packet thatmaintains the moisture content of the packages ofthe ICs it contains below 0.1% by weight for up toa year.It must be stored at a temperature below 40 °C inan atmosphere of less than 90% relative humidity(RH).The Drypack contains a desiccant and a humidityindicator which allows the moisture content to bechecked when the bag is opened.

Using ICs from a DrypackBefore using ICs from a Drypack, it is essential tocheck the humidity indicator. If it shows RH of less

than 30% (the colour of the 30% dot has notchanged from blue to pink), the ICs it contains areready for use. However, to prevent themabsorbing moisture after the Drypack is opened,the ICs must be soldered onto a PCB within theperiod specified in Fig. 2. The times indicated inFig. 2 apply to ICs awaiting soldering at atemperature of not more than 30 °C in a RH ofless than 60%.If the humidity indicator in a Drypack shows RHof more than 30% (the colour of the 30% dot haschanged from blue to pink), the Drypack has beendamaged, opened, or stored under too severeclimatic conditions. In this case, to eliminate anypossibility of moisture-cracking, the ICs containedin the Drypack must be dried (baked) beforesoldering.Figure 1 shows the reduction of moisture content as a function of time for some large plastic ICpackages. Figure 2 shows a typical Caution labelwhich is on the bag. Table 1 gives therecommended drying times for reducing themoisture content of large plastic IC packages froman initial level of 0.3% by weight to less than0.05% by weight.

Note: longer dryingtimes will have anegative influence onthe solderability.

71

Resealing a DrypackIf any ICs from a Drypack are not used, thedesiccant and humidity detector should bereinserted and the Drypack resealed within half anhour of opening by using commercially availableheat-sealing equipment.

Table 1Drying times if a Drypack indicates more than30% RH

Package

QFP44QFP48QFP64, 80, 100QFP120, 128, 160PLCC44, 68, 84

Temp. (°C)

125125125125125

Drying time (h)

86101212

Fig. 2 Typical CAUTION label.

72

Electromagnetic Compatibility (EMC)

The EMC directive itself actually excludeselectronic components from the scope but formost products these will be the cause or victimof interference problems. As a major supplier ofelectronic components we are aware of this andwe are therefore actively engaged in developingEMC-friendly components to anticipate ourcustomers needs for the future. These EMC-friendly components will help our customer toarrive at an economic application which will meetthe European (and other international) EMCnorms.

To benefit from these EMC-friendly componentsit will be evident that a whole system must bedeveloped according to this philosophy. EMCguidelines will come together with thesecomponents (in application notes) which will putsome constraints on our customers application.It must be emphasised that EMC-friendlycomponents will still require correctly designedprinted circuit boards, filtering and cablingbecause the final product will be as weak as itsweakest link.

To ensure EMC empowerment the following stepsare taken:1. EMC is generally defined in the specification of

new components.2. EMC is taken into account during the product

type-approval process.3. A network of EMC specialists is established

throughout the organization.

Both in design and production centres as well asin product concept and application laboratories(PCALs) dedicated support can be given on EMCproblem finding and solutions.Weve committed ourselves to help our customersin meeting the EMC requirements at minimumcost and with shortened design-in cycles.

All electrical and electronic products, apparatus,appliances, equipment and installations marketedin European countries must comply with a strictEMC directive. The EMC directive itself does notcontain any technical requirements or limits butmakes reference to generic or product-specificEMC requirements which will apply for both RFemission as well as immunity.

Its particularly important that electronic productsdo not cause interference to sensitive receivers,for example lap-top computers in an airplane orcordless telecommunications systems in a carwith ABS, air-bags and engine managementsystems.

The directive means that all our customers(equipment manufacturers, setmakers),independent of the product type they produce:telecom, automotive, consumer, multimedia, etc.,must give a declaration of conformity based ontype-testing results, either performed bythemselves or carried out by a testhouse (thirdparty). In addition, a quality assurance system(ISO 9001) is required for the production centre toguarantee reproducibility referred to the approvedsample(s).

Active involvement in the EMC standardizationprocess enables us to be aware of the EMCrequirements which will apply to our customersproducts within a few years time. We cant waituntil one of our customers requests certain EMCspecifications for our products, simply because wewill be too late by the time the component isrequired for production.

The scope of the EMC directive is very broad andwill have profound effects on the manufacturingindustry, and as a result, on their suppliers.

73

Electrostatic Discharge (ESD)

Fig. 1. ESD hazards are everywhere in a normal working environment: especially where modern synthetic materials arewidely used. Static charges are invisible, ubiquitous, and destroy semiconductors without warning. Rigorous observation ofa few basis precautions can have a dramatic effect on quality levels.

Damage to semiconductors from electrostaticdischarge (ESD) is a major cause of rejects now,and will become an even-greater hazard as devicegeometries shrink.Main sources of ESD are dry, clean workingconditions, coupled with the universal use ofplastics for containers, clothing and work surfaces.Only rigid observance of good working practiceseverywhere semiconductors are handled bothindividually and on boards (including rejects foranalysis) will combat it. ESD prevention isVITAL to the achievement of low reject levelseverywhere semiconductors are used: and to thepreservation of the valuable quality-improvementdata in rejects.Experiments indicate that the worst-case electricalmodel for a person sitting on a chair (the human-body model) is a 100 pF capacitor in series with a1500 Ω body resistance. Human static potentialscan certainly reach 10 kV; under extremeconditions they can exceed 30 kV.

The associated energy level (1/2 cv2) may thus beof the order of millijoules, whereas MOS devices,even with protective networks, can only dissipate20 microjoule pulses.Energy pulses due to excessive static chargespunch fine holes in the glass layers separatingmetal film interconnects on semiconductorsurfaces. These holes may be lined with metal orsilicon vaporized during the discharge, and soprovide short-circuits.

ESD IN PRACTICEIn a typical electronics-industry workingenvironment, charges may be generated bymachinery operating, plastic storage bins, jobinstructions stored in plastic envelopes, airblowing over machinery or table tops, and byhuman motion (especially in some artificial-fibreoveralls), Fig. 1.

This is adangeroussituation!

Plastic storage bins

Plastic table top

Nylon overall

Air blowing over table top

Plastic trays

Plastic envelopes

Nylon carpet or plastic flooring

74

Table 1Circuit values for ESD sensitivity tests

(see Fig. 2)

Philips test method

Note: for machine-model test results to be comparable,residual resistances and stray inductances must be thesame.

ESD sensitivitySince glass insulation layers are used in bothdiscrete and integrated semiconductor devices, allsuch devices may be expected to be ESD-sensitive to an extent that depends on geometryand glass properties. Test results indicate thefollowing general range of sensitivities to ESD:

unprotected MOS is most sensitive CMOS and bipolar analog devices are equally

sensitive Schottky TTL and some linear ICs, are as

sensitive as CMOS low-power Schottky TTL is almost as sensitive

as Schottky TTL and CMOS standard TTL may be degraded by ESD at 2.5

kV (e.g. from handling personnel) some low-power and small-signal discrete

semiconductors are ESD-sensitive.

Tests for ESD sensitivityAll tests for ESD sensitivity are based on thedischarge of a capacitor through a resistor into thedevice under test, Fig. 2.Table 1 gives the circuit values for two testmethods: machine model (low impedance)

see SNW-FQ-302B human-body model (high impedance)

see SNW-FQ-302A.

component machinemodel

human-bodymodel

units

CRV

20025

>200

1001500

>2000

pFΩV

Deviceunder test

R

CV

Fig. 2. Circuit for testing the sensitivity of semiconductordevices. Capacitor C is charged to the specified voltageand then discharged into the device under test throughresistor R. Values for both versions of the test are givenin Table 1.

75

ESD PRECAUTIONS

The ESD work stationEssential features of a work station for handlingESD devices are shown in Fig. 3. Adaptations forinspection, assembly, repair and other purposesshould respect these guidelines: conductive work-surface sheet resistance

10 kΩ to 1 MΩ per square metre resistor for grounding wrist strap between 0.9

and 5 MΩ. Maximum ground current 2 mA:enough for operator to feel a fault but wellbelow danger level.

all test equipment grounded switching transients suppressed

all metal table trim, support frames andbrackets grounded

cotton working garments static-safe rails, bags, foam pads and shorting

clips available, if needed.

ESD precautions in semiconductordesignOur semiconductors generally have either intrinsicprotection networks (resulting from active junctions)or added protection networks. Protection is,inevitably, a trade-off between degradedperformance (clamping diodes limit the operating

Fig. 3. Essential features of an ESD work station. Variations to suit inspection, assembly, repair and packing should followthe same principles.

Conductivecompartment

trays Electrostaticvoltage sensor

Cotton overall

Conductive bootsor heel groundig

protectors

Conductive stool

Strap (resistancebetween 0.9 and 5 MΩ)

Conductive floor mat

1 MΩ

1 MΩ

1 MΩ

Ground

Commonreference

point

Supplychain

Safetyisolation

transformer

Distributionsupply box

76

voltage input range, the added parasiticcapacitance reduces speed) and increasedsecurity against ESD.

Circuit layout precautionsDesigning of a circuit board for ESD-sensitivedevices should allow for handling by personsunaware of the ESD hazard. Observe thefollowing precautions: Tracks to and from ESD-sensitive devices

should not pass board edges, to minimize therisk of their being touched in handling.

Where possible, connect a resistor of about 1 MΩ between conductors from ESD-sensitivedevices and board inputs and outputs.

Avoid long signal lines; they increase the risk ofinduced large-signal pick-up.

Observe the maximum rated values for supplyturn-on and turn-off transients. Suppress powersupply turn-on and turn-off transients, powersupply ripple or regulation and ground noise, toavoid exceeding the Absolute Maximumratings. Fast zener protection diodes are usefulhere.

Label the board with an ESD warning.

Make sure that the service documentation callsattention to the use of ESD-sensitive devices andthe precautions to be taken with them.

Marking of ESD-sensitive devicesIEC 417 and MIL-STD-1686 recommend that thesymbol shown in Fig. 4 is used to mark ESD-sensitive devices. The symbol should besupplemented by the notice 'ATTENTION observe precautions for handlingELECTROSTATIC SENSITIVE DEVICES'. Wherespace is restricted, the simplified symbol shown inFig. 4(b) may be used. Symbol and letteringshould be in black on a yellow ground.

(a)

(b)

Fig. 4(a) Warning symbol for ESD sensitivity according toIEC 417.

(b) Simplified version for use where space is restricted.

USER PRECAUTIONSAs a general rule, ESD-sensitive devices shouldalways be handled at an ESD station conforming toFig. 3. Pay particular attention to stores andinspection areas where personnel may not be fullyaware of ESD hazards.

Packing and storageESD-sensitive devices are packed in antistatic orconductive boxes or rails. Intimate (tube, tape, bagetc.) and proximity (level 1 box) packing is markedwith the Fig. 4 symbol. ESD-sensitive devices notsupplied in antistatic packing should be returned tothe supplier. ESD-sensitive devices should bestored in their original packing, preferably in a coolplace set aside for the purpose. Do not unpackthem until they are required for incoming inspectionor use in production.

77

Receiving inspectionDo not put ESD-sensitive devices where staticdischarges can occur, even if they have protectivepacking. In their immediate vicinity avoid thepresence of: materials which can develop static charges

(see Table 2) electrical switching equipment and tools.

These precautions also apply to assemblies thatincorporate ESD-sensitive devices.

Unpack and handle the devices at an ESD workstation generally conforming to Fig. 3. Take carethat the devices are not exposed to the voltagepulses that can occur when switching the powersupply on and off. Increase the supply voltageslowly to its normal value before applying testsignals, to avoid the latching effect that occurswhen the signal voltage exceeds the supplyvoltage. During testing, and especially when goingfrom one test to another, ensure that all suppliedvoltages are under control.If possible, ground all unused inputs during tests.Do not allow a signal to remain on an input whenthe power supply is switched off. If necessary,connect a buffer stage between the signal sourceand the input in such a way that it automaticallyswitches off the signal when the power supply isswitched off.After testing, repack the devices in their originalanti-static packing; keeping the warning labelintact. Repack at an ESD workstation.

Table 2Triboelectric series of some common materials

airhandscatskinglassmicanylonwoolleadsilkaluminiumpapercottonsteelwoodhard rubbernickel/copperbrass/silvergold/platinumsulphurRayonpolyesterOrlonSaranpolyurethanepolyethylenepolypropylenepolyvinylchloridesiliconTeflon

+positive

negative

78

Repair and maintenanceprecautionsSwitch off the equipment in which the board isincorporated before removing a board containingESD-sensitive devices.For repair and maintenance use an ESD workstation arranged as shown in Fig. 3. Place theboard on an antistatic foam pad. Observing the'Assembly precautions', remove and replace thefaulty device. After testing, replace the board in theequipment.

NOTES ON STATIC ELECTRICITY

Electrostatic charge generationIn neutral material, the net charge of protons(positive charge) and electrons (negative charge)is zero. When the surface of one material is rubbedalong that of another, local (frictional) heating cantransfer energy to the electrons near the surface inexcess of the Coulomb binding energy. Suchelectrons may leave their outer valence orbit andbe trapped in an outer valence orbit in the othermaterial. Thus two ions will be formed: positive, for electron-donor material negative, for electron-acceptor material.

Friction between any two surfaces involving atleast one non-conductive material is a potentialgenerator of electrostatic (triboelectric) charge; themagnitude and polarity of the charge depends on: the materials involved. Charge magnitude and

polarity depends on the sum of the separationsfrom the neutral boundary of the two materialsin the triboelectric series (Table 2)

frictional heat, which depends on speed andapplied force

surface conductivity. Part of the charge may bedrained off during and after rubbing, inhibitingbuild up of maximum possible voltage, but thisis true only for surface conductivities below109 Ω per surface square.

Assembly precautionsESD-sensitive devices should be the lastcomponents to be inserted in a circuit board orsystem.

Manual insertion : Use an ESD work station.

Automatic insertion : Ground insertion equipmentand machinery. Use only tools of conductive orantistatic material.Use grounded component tongs to remove ESD-sensitive devices from their antistatic packing. Donot remove more components at a time than areimmediately required.

Soldering : Attach short-circuit clips to ESD-sensitive devices before soldering them; makesure that the clip short-circuits all leads. Removethe short-circuit clips only after soldering, cleaningand drying.Ground the soldering iron or bath. Do not solder tocircuits that are connected to a switched-on powersupply.Ensure that every work surface on which a circuitboard may be placed is provided with aconductive or anti-static sheet big enough toreceive the whole board.

Handle boards that contain ESD-sensitive devicesas single components. Pack them in antistatic orconductive packing. Label them with an ESDwarning. Ground all handling personnel.

Measurement precautionsPlace the board, soldered side down, on aconductive or antistatic foam pad to discharge anystatic electricity. Remove short-circuit clips.Handle the board only by its edges, remove itfrom the foam pad for testing. After testing,replace it on the foam pad for transport.

79

A grounded operator cannot drain charge from anon-conductive object. Thus, an operator'sclothing may be charged even though his body isgrounded by a conductive wrist strap. Similarly,charged plastic boxes or trays will not bedischarged by a grounded operator or bench top.

InductionStatic charges can be transferred by induction;that is, without direct contact. Objects which cantransfer charges by induction include the plasticboxes, trays and covers used extensively inproduction lines. An ESD-sensitive devicecharged by induction can be damaged if touchedby a grounded operator.Removing static charges from insulating materialscan only be achieved by use of ionizers.

Limitations of anti-static agentsAnti-static agents conductive sprays arecommonly used to protect against ESD. Althoughthey do protect against charging by friction, theydo not form an effective shield and therefore giveno protection against charge induction. The onlysure protection against charge induction is aFaraday cage shielding the protected object fromall possible sources of induced charge.

Static detection and preventionequipmentA wide range of commercial products is availableto help detect static electricity, equip work stationsand prevent ESD. They range from conductivebags, gloves, mats, foam, wrist straps and boxes,through to static voltmeters, ionizers and ESDsimulators. Careful use of available products canhelp locate and prevent ESD hazards, and soimprove quality wherever semiconductors areused.

80

Environmental care

Environmental care is an integral part of thebusiness policy of Philips Semiconductors. It isbased on four principles: sustainable development development of

products and processes that have minimumeffect on the quality of the environment todayand in the future

prevention is better than cure the total effect on the environment counts

embodied in the development of productswhose production (including energy use),operation and disposal at end-of-life haveminimum adverse effects on the environment

open contact with the authorities andcustomers.

Commitment to these four principles by PhilipsSemiconductors leads to a specific programme ofobjectives and targets and to the allocation ofcapacity. Programmes and progress are reviewedannually.

Environmental planningThe plants develop an annual environmentalimprovement plan, based on evaluation of theenvironmental effects of their activities andservices under normal and abnormal operatingconditions, and on the corporate and divisionalprogrammes. Philips Semiconductors long-termprogramme is based on Philips Ecovisionprogramme, and consists of targets formanufacturing processes and for products. Fromthese long-term targets the annual PD goals arederived and set.The plants report annually on the progress of theirenvironmental improvement plan, and the PDsummarizes the plant progress in a Philips

Semiconductors environmental report.The chemical content of Philips Semiconductorsproducts is registered in a database. This data ismaintained by the local MISD groups based oninformation supplied by the developmentdepartments at RFS (release for supply).

Legal requirementsThe Corporate Environmental & Energy Officearranges the collection and recording of applicableregulatory and European legislative requirementspertaining to environmental aspects.The local laws and regulations are recorded byeach National Organization and communicated byits Environmental Coordinator to the plants withinhis organization.

PD Semiconductors targets 19982002Process improvement energy saving: efficiency improvement 35% waste reduction: 50% less to

landfill/incineration water consumption: 50% reduction of intake emissions to air/water:

cat 1 98% cat 2 50% cat 3 20%.

(reference year 1994)

Product impact green product marketing products eco-designed based on green focal

areas: 10% in 1999 improving to 7.5% in 2002 packing reduction: 15% (reference year 1994).

81

Certification of the Environmental Management SystemAll Philips Semiconductor plants have anEnvironmental Management System which hasbeen certified according to ISO 14001.

Eco-designBy adopting Eco-design principles, PhilipsSemiconductors constantly strives to developproducts that have minimum impact on theenvironment. This means designing for recyclingand the economic use of materials and energy inproduction.

Reducing hazardous materialsOne of our current focuses is on replacing nickelleadframes by leadframes of copper to reduce thequantity of nickel passing into the environment both in production and at end-of-life disposal. Andwere working to eliminate lead from the die-bonding process by replacing soldering by anepoxy gluing process.Research is also going on to replace antimonyand bromine, currently used as flame retardants insemiconductor encapsulations, by moreenvironmentally benign additives, and to replacetoxic beryllia in semiconductors by, for example,aluminium nitride.So by looking ahead when developing newproducts, we believe that its possible to virtuallyeliminate their toxic impact on the environment.

Reduced environmental impact of packingWeve also taken steps to reduce the impact ofused packing material on the environment. Topromote recycling, we have switched to monomaterial (for example, moving from aluminium-lined boxes to carbon-coated boxes).Other measures include switching from two-piece

to one-piece boxes to reduce weight, changingfrom boxes of white cardboard (which requirebleach in manufacture) to brown recycledcardboard, and using water-based inks (withoutheavy metals) for marking. All parts are markedwith recycling symbols.Whats more, we actively promote re-use of reelsand trays used for discrete semiconductors andintegrated circuits. The boxes are also markedwith labels giving an address to contact to arrangecollection of used reels and trays.

ODC-freeIn the elimination of ozone-depleting chemicalsfrom its production processes, PhilipsSemiconductors can claim major successes. Asearly as May 1993, all plants had eliminated CFCs(chloro-fluorocarbons) from their manufacturingprocesses.This led the way to a complete phasing out of allClass I and Class II ODCs (listed in the 1986Montreal Protocol) from our products andmanufacturing processes in compliance with theUS Clean Air Act.

Involving partnersSuppliers and subcontractors, too, form a crucialelement of our EcoDesign programme. We requirethem to be environmentally responsible, to havetheir own environmental policy and improvementplans and to record environmental information onall raw materials supplied to us. A company-widesystem to communicate our environmentalrequirements to them is now being installed.Future preferred suppliers will also be required tohave ISO 14001 certification.

82

Environmental policy and goals

Evolution of quality

The start of quality controlThe concept of Quality Control originated inWorld War l. The term 'control of quality' wasprobably first used in a paper published in 1917(Ref. 1), with the first book on the subjectappearing in 1922 (Ref. 2).Quality developed slowly, and largely in isolationbetween the Wars: statistical methods wereintroduced to results analysis; sampling methods,already used for census applications, began to beadapted to quality control. The control chart(fundamental to Statistical Process Control) wasinvented by Shewart in 1924, and described in1931, laying stress on the cost-effectiveness ofthe method! (Ref. 3). Thus, by the end of the1930s, many of the fundamental tools of qualitycontrol were available, if not widely appreciated.As with many other disciplines, quality controlderived enormous impetus from World War ll. By1945, Statistical Process Control had reached anadvanced state of development, and samplingmethods had largely been standardized (MIL-STD-105 was published in 1950). Since itsrelationship to cost was, as yet, not appreciated,quality control was still largely regarded as anecessary evil; the people practicing it had littlestatus (except when things went wrong).

Quality in declineAs defence production ran down, some of thepractices introduced for munitions production,especially that of the independence of the QualityDepartment, remained – but mainly to satisfyinspection requirements for Governmentcontracts. (Independence soon disappeared,though, when quality practices ran contrary toshort-term commercial expediency!). Had it notbeen for the requirements of Governmentagencies, quality disciplines might well havedeclined even further.Fascination with statistical methods in the Westlargely eclipsed the original concept of quality

control (as a means of defect prevention) for thenext 20 years or so. In industry, perhaps the onlyreadily-demonstrable commercial advantage ofstatistical quality methods lay in the AQL systemwhich reduced inspection (the examination of100% of items) effort. Since the statisticalmethods employed allowed customers to verifysuppliers' claims, AQL became the basis forquality requirements in purchase contracts of allkinds, both governmental and commercial.During the period of rapid technologicaldevelopment of the 1960s and early 1970s, theQA Department existed in isolation: its activitiesseemed to have little relevance to the excitingdevelopments in the products themselves, andless to the process of marketing them. In otherwords, quality became the problem solely of theQuality Department.

The great AQL raceDuring the 1950s, AQLs around 1% were usual forthe majority of electronic components. Asequipment became more complex, with largercomponent counts, OEMs became more andmore concerned to reduce AQLs. It's easy to seewhy: in TV set production using 500 componentsper set supplied to an AQL of 1%, there would bean average of 5 defects per set, or 99.3% of setswould be defective. With AQLs of 0.1%, however,the average number of defects would fall to 0.5per set, and the percentage of defective setswould be down to 40%, with a consequent savingof rectification costs. (Since early failures arerelated to conformity, there would be fewerproblems during the guarantee period, too.)By the 1960s, with the increasing reliance oncomplex electronics for defence purposes, theAQL problem was becoming acute. Successfuluse of automatic assembly techniques alsorequired higher component quality. Moreover, thecost consequences of on defective componentsbeing found in finished equipment tends to

84

85

counter the savings due to automatic assembly.As a result, during the early 1970s, there waspressure on component suppliers to reduce AQLsstill further. However, lower AQLs mean largersamples to be tested, which increases costs forboth supplier and user.Some manufacturers even offered a choice ofAQL levels. Lower values were achieved bydouble inspection or screening to eliminatedefective products in other words, inspectingquality in. As we can now appreciate, this did notimprove the real quality of the product: weakproducts still crept through, to fail on the assemblyline or early in the guarantee period. Moreover,the associated costs became rapidlyunacceptable.

Wise men in the East...One of the men active in quality proceduredevelopment in the USA during World War ll wasDr. W. Edwards Deming. During 1942 and 1943he published several papers supporting SPC andgave short courses on quality methods at StanfordUniversity. About this time, Deming joined theNational Bureau of Census.In 1948, the Military Government of Japan, thenheaded by General Douglas MacArthur, carriedout a population census, for which purposeMacArthur employed Deming. By this time, theJapanese were already looking for a means ofimproving their product reputation. In 1950,Deming presented courses in statistical methodsof quality control. (The Deming Prize is today themost prestigious quality award in Japaneseindustry). The Japanese Union of Scientists andEngineers, founded in 1949, organized follow-upcourses, employed other top American qualityexperts (who were glad of a receptive audience),including J. M. Juran in 1954, and founded afaculty of specialized teachers. By the time alicence for the production of transistors had beenobtained from Bell Labs in 1955 (on which the

Sony Corporation was founded), quality was anaccepted management tool.Looking back, the arrival in Japan of solid-statetechnology just when quality management wasbecoming accepted seems almost fateful. By theend of the 1960s, the quality of Japanese goodswas already apparent to the consumer. By the mid1970s, even Western industry was beginning totake notice. By 1980, when there were some200,000 Japanese managers and engineerstrained in SPC, around 40% of workers trained inquality appreciation and methods, and schoolstaught statistics for one or two years, thesuperiority of Japanese electronic products wasobvious to all.

Awakening in the WestFrom the mid 1970s, the electronics industry inboth Europe and the USA was losing marketshare to the Japanese so rapidly that it wasevident that something had to be done. Thevarious patent protections that had held the linefor a while were due to expire, in any case, andthis would make matters even worse.It cannot be said, though, that the development ofquality methods had entirely stagnated in theWest. Defence requirements, again, stimulatedsome attack on the growing quality problems. Theconcept of 'Zero Defects' was introduced (MartinCompany, 1961-62), and was widely publicized(Ref. 4, for example); and Philip Crosby originatedthe 'Do It Right First Time' principle while at ITT.

How bad was it?Those outside the small circle preaching themessage of quality during the 1970s probablynever really knew how great the difference wasbetween Japanese and Western quality. TVproduction provides a stark example: in aEuropean TV factory in 1976, the fall-off rateaveraged 200%: 2 faults per TV set produced(these were major faults: if minor faults, such as

dry joints, were taken into account, the true figurewould have been nearer 1500). Japanesefactories had fall-off rates around 1%. (Reliability,although perceived by the consumer as the majorproblem, only ever differed by a factor of four orfive, except for design faults). The cost differencelay mainly in repairing all those faults. To quotethe Quality Manager of one offshore Japanese TVfactory (in 1979!): 'You can't sell junk, so whymake junk?'.Many of the problems stemmed from theincompatibility of the goals of production and QA.Production strove to manufacture at minimum costand on time: quality was not part of their brief;neither was it part of the brief of the purchasingdepartment, whose goals were minimum initialcost and prompt delivery. QA had no responsibilityfor cost or schedules: only for quality. Inevitably,there was always conflict, with QA blamed forholding up production, and production devisingways of deceiving QA. A favourite ploy was to mixgood and reject batches after Acceptance Testingto just meet the AQL (which was a licence to shiprejects in practice).

Burying the mythsInitial reactions to Japanese success were reallyexcuses for doing nothing about it (remember?):'they live on a handful of rice a day'; theirs is agroup culture, our workers could never performlike that'; 'wait until they expect a decent standardof living' . . . Eventually, it became obvious that thecost and quality advantage of Japanese goodswas due to fundamental differences in philosophyof their design and the circumstances of theirmanufacture. Delegations from Westernorganizations started visiting Japan to see forthemselves. These visits were of limited value, inpractice: the gap in attitude and methodology wastoo great. Impressions gained could even bemisleading: Western obsession with QualityCircles probably delayed real quality improvement

by around two years. Many managers were, inany case, still reluctant to believe that qualitybegins at the top.

Quality begins with managementReally convincing demonstrations of theimportance of quality-oriented management camewhen Japanese companies started movingoffshore; when a small team of key people movedin down the road, or took over a failing business,and, within six months, using the local labour thatlocal management had blamed for their problems,started producing goods with a quality comparablewith that of the factory back home.

Meanwhile, at Philips . . .How does Philips semiconductor production fit intothis picture? Well, we were ahead of our localcompetition in formulating, implementing and (asfar as we could, honestly) publicizing our QualityImprovement. Serious attempts to improve qualitystarted in the late 1970s, with Signetics a littleahead of the European operations. Early quality-improvement efforts were AQL-oriented, in linewith those of our competition, and therequirements of customers.Attempts to introduce PPM-based qualitycooperation relationships with customers met withconsiderable opposition, mainly due tomisunderstanding of the principles andrequirements involved. We are, however,fortunate that our internal customers – such asPhilips Consumer Electronics – also realized thevital importance of quality improvement, andprovided valuable experience in developing in-depth customer-oriented quality-improvementprocedures (as Delco did for Signetics). It wasevident that education was required at all levels,an activity which has been in progress ever since.CONIM started in 1982, and the 14-step quality-improvement programme with its associatedQuality College in 1984. Four years later, staff

86

87

from all centres had received quality-awarenesstraining in some form.During this time Statistical Process Control wasintroduced and the CWQI programme started.ISO 9000 was implemented for all manufacturingcentres during 1991/1992.At present, Total Quality Management andcustomer-oriented quality improvements are inprogress.

References1. Radford, G.S. 1917. The control of quality.

Industrial Management Vol. 54, p. 100.2. Radford, G. S. 1922. The control of quality of

manufacturing. Ronald Press, New York.3. Shewart, W. A. 1931. The economic control of

quality of manufactured product. Van Nostrand,New York.

4. 1964. ZD Time. Time November 6, pp. 9394.5. Dale, B. G. 1990. Managing Quality

Total Quality Management: an overviewPolicy deploymentInvolve suppliers and customersInvolve all operationsProcess managementPerformance measurementTeamworkEmployee involvement

Quality systems developmentAdvanced quality planningComprehensive quality manualsUse of quality costsInvolvement of non-production operationsFailure Mode and Effects AnalysisStatistical process control

Develop quality manual Process performance dataSelf-inspectionProduct testingBasic quality planningUse of basic statisticsPaperwork controls

SalvageSorting, grading, reblendingCorrective actionsIdentify sources of non-conformance

TotalQuality

Management

QualityAssurance

QualityControl

Inspection

Continuous improvement Empowering people Caring for people Involvement

Compliance to specification Allocating blame

TQM

TQMQA

QCQAI

I

88

External standardization

Standardization is aimed at achieving concensusbetween the parties concerned. Internal standardization concerns the standardsfor parties within Philips.With external standardization, standards aremade with parties outside Philips, to be usedeither worldwide, for a particular region (e.g.Europe), for national use or for use betweenindustrial organizations.In the case of semiconductors, the externalstandardization covers items such as thespecification of a semiconductor, test methods,mechanical outlines, symbols and definitions,data elements for quick reference data, protectionagainst EMC and ESD, quality systems, etc.

Worldwide standardizationWorldwide standardization is carried out throughthe International Electrotechnical Commission(IEC) and the International Organization forStandardization (ISO). The members of theseorganizations are the national standards bodies.Founded in 1906, IEC is the worldwidestandardization body for electrotechnical andelectronic engineering.Founded in 1947, ISO is the worldwidestandardization body for all technologies, exceptelectrotechnical and electronic engineering.In both organizations International Standards areprepared by technical committees. DraftInternational Standards, adopted by thesecommittees, are circulated to the member bodiesfor approval, before acceptance by the IEC/ISOcouncil.The technical committee for semiconductortechnology is IEC/TC47. Philips Semiconductorsparticipates in the following committees andworking groups of IEC:

technical comittee 47:Semiconductor devices

TC47/working group 1:Terminology

TC47/working group 2:Environmental test methods

TC47/working group 5:Wafer level reliability

sub-committee 47A:Integrated Circuits

SC/47A/working group 1:Hybrid integrated circuits

SC/47A/working group 2:Logic digital integrated circuits

SC/47A working group 3:Memories

SC/47A/working group 6:Manufacture approval and TQM concept

SC/47A/working group 8:Reliability characteristics

SC/47A/working group 9:EMC measuring methods and test procedures

sub-committee 47D:Mechanical standardization

sub-committee 47E:Discrete semiconductor devices

sub. committee 3D/WG3:Classification of components

technical committee 91:Surface mounting technology

89

European standardizationCENELEC is a European organization forstandardization and certification of electroniccomponents. Many world standards of IEC 47 havetheir origin in CECC standards. CECC is now part ofCENELEC.Philips Semiconductors participates in the followingsub-committees and working groups of CENELEC: WG Known Good Dies (KGD) WG QAP:

Quality assessment procedures SC91:

Surface-mounted devices TC110:

EMC standards.

USA standardizationThe standardization in the USA is carried out throughthe Electronic Industries Association (EIA). Standardsfor electronic devices are made by the Joint ElectronicDevice Engineering Council (JEDEC).Philips Semiconductors participates in the followingcommittees: IEA/JEDEC Council, JC-11 Mechanical (package outline) standardization, JC-13 Government liaison JC-14 Quality and reliability, JC-15 Electrical & Thermal characterization JC-16 Electrical interface & power supply standards, JC-40 CMOS digital logic, JC-42 Memories.

Availability of external standards is described in quality standard SNW-SQ-023.

90

Failure analysis

Since causes of failure must be defined beforethey can be remedied, failure analysis is a keyelement of any quality-improvement activity. Forsemiconductors particularly, it is also a delicate,complex, time-consuming and, consequentlyexpensive operation. Careful documentation of thecircumstances of failure is essential if the resultsof failure analysis are to be of use. Our complaintsprocedure is designed both to generate theappropriate documentation and to make maximumuse of the results obtained.

Analysis proceduresRejected semiconductors for analysis may comefrom our production, customer returns, or fieldservice. All rejects are first given a full electricaltest; those that fail (less than 50% for somecustomer returns) proceed for further examination.

CorrelationWhere no defect according to the Final TestSpecification is evident, it may be necessary toexamine the coverage of the test program todetect any correlation problem; this often requiresapplication-engineering facilities. Where there is aconsistently high percentage of good devices inreturns, it may be necessary to discuss testcoverage and application conditions with thecustomer.

Examination of defectsIn most cases, it is necessary to decapsulate thesemiconductor to reveal the cause of the defect.However, before decapsulation, non-destructivemethods can be used to gather extra informationabout the defect. For example, analytical electricaltests can often identify the part of the circuitresponsible for the failure, and X-ray or ultrasonicinspection can reveal package defects.Decapsulation without damaging the die or thebond wires, is a delicate operation requiringspecial chemical etching facilities. Once the

interior of the semiconductor is exposed, it can beexamined with an optical microscope or ascanning electron microscope (SEM). Many of ourcentres have voltage-contrast SEM facilities thatallow potential distributions across a die to beobserved during operation. This is especiallyuseful for examining digital ICs. Another techniqueuses highly temperature-sensitive liquid crystals todetect hot spots, which often pinpoint the locationof a defect. Since a defect can often be hiddenunder several layers, the semiconductor may needto be deprocessed to allow access to the failuresite. This may require a combination of wet-etching and plasma-etching. Several analyticaltechniques can then be used, such as Augerspectroscopy for powerful surface-analysis. Manyof our centres also have EDAX SEM facilities todetermine the precise nature of any foreignparticles at the failure site.

Reporting resultsFailure analysis results are reported fully to thedepartments concerned, and, where returns areinvolved, to the customer using the complaintsprocedure. Failure analysis is the major source ofdata for corrective action in production.

91

Failure Modes and Effects Analysis (FMEA)

FMEA is a structured analysis of potential failuremodes and their effects, with the aim of reducingor eliminating failures of products or processes.FMEA identifies corrective actions required toprevent failures from reaching the customer,thereby assuring the highest yield, quality andreliability. As a result it reduces the cost of quality,both internally and at the customer.

FMEA has three main aspects:1. To recognize and evaluate potential failure

modes that could occur in the design ormanufacture of a product.

2. To identify actions that could eliminate orreduce the chances of the potential failureoccurring.

3. To document the process.

For semiconductor devices, it's normal to carry outtwo types of FMEA: Design FMEA, used by Product Design, which

addresses potential product failures. Process FMEA, used by Process

Design/Engineering, which addresses potentialprocess failures (which could of course causeproduct failures).

It's important to remember that these two FMEAsare produced independently. The Design FMEA isnot a precondition for the Process FMEA; lack of aDesign FMEA should never delay work on aProcess FMEA.FMEA is a multifunctional team effort. Its successrequires the input of many disciplines such asAssembly, Test, Quality and Marketing.

TimingThe timing of FMEA is all-important. It's meant tobe a before-the-event action, not an after-the-factexercise. To achieve its greatest value, FMEAmust be carried out before a design or processfailure mode has been unknowingly designed into

the product. Ideally, it should be an integral part ofthe product or process development, carried outbetween ATD (acceptance for type development)and design approval.

Benefits of FMEAFMEA offers the following benefits: Assists in selecting design/development

alternatives with high producibility and reliabilitypotential, during an early phase ofdevelopment.

Ensures that all possible failure modes andtheir effects on the fitness-for-use of theproduct have been considered.

Lists potential failures and identifies the relativeseverity of their effects.

Provides an instant visual record ofimprovements resulting from any correctiveactions taken.

Provides a basis for an additional testprogramme during development andmanufacturing.

Provides historical information for futurereference to aid in the analysis of possiblefailure modes for consideration in intendedproduct/process changes.

Ensures that the responsible development orprocess engineer organizes defect-preventiontechniques for assessment at finalproduct/process review meetings.

Documenting the FMEAThis early warning and preventive techniqueprovides the development or process engineerwith a methodical way of studying the causes andeffects of failures before the design ordevelopment is finalized. All aspects of theanalysis are recorded on an FMEA form (Fig. 1).

92

For each identified potential failure mode anestimate is made of its cause, and the likely effecton the fitness-for-use of the product. A risk prioritynumber (RPN) is then calculated and assigned toeach identified failure mode. RPN is calculated bymultiplying together three parameters (S x O x D)where:

S= severity if the failure occurs,O= occurrence of the failure,D= detection likelihood before product delivery.

A value of 1 (low risk) to 10 (high risk) is assignedto S, O and D, so the RPN can be between 1 (lowrisk) and 1000 (high risk). The resulting RPNs are

then assessed, and engineering judgement isused to decide if the risk is acceptable, whethercorrective actions are necessary and, if so, whatthey should be and in what timescale. Thisinformation is also recorded on the FMEA form.After any necessary corrective actions have beentaken, RPN is again calculated and recorded onthe FMEA form, and a judgement is made toascertain that the new result is acceptable. In thisway the form can show at a glance the dramaticimprovements that can result from FMEA. TheFMEA form is a living document which must beregularly updated to reflect changes in design,process and use of the product.

FMEA Type/process: Report no.:

Date:

Sheet no.:

Function Potentialfailuremode

Potentialcause offailure

Potentialeffect offailure

Present

OS D RPN

Resp.person

Recommendedcorrective

actions

Rec.enddate

Results

OS D RPN

Fig. 1 Typical FMEA form.

FITS is commonly used to express componentreliability (see RELIABILITY), and is defined asthe number of failures occurring in 1 billion (109)hours.Reliability calculations generally involveconsiderations of statistics, time and operatingconditions.Consider a group of operating semiconductors. Atany elapsed time (t), the reliability R(t) of thegroup is given by:

where no is the initial number of semiconductorsin the group and nf is the number of failures inelapsed time (t).However, reliabilty is more commonly expressedas an exponential probability distribution :

where λ, the failure rate, is constant with time.

From these two equations,

When averaged over a long period, thisapproximates to:

And, over the time standard of 109 hours, thisbecomes:

This simple equation is the basis of most reliabilitycalculations. For example, consider a group of

100 semiconductors operated for 3 years understandard conditions, with only 2 failures observed:

In practice, we usually need to predict failure rateover a longer period, say 20 years. We can'tsensibly test over such a long period, so we use atechnique called Accelerated Life-Testing.

Accelerated Life-TestingIn this method of testing, components are made toperform at abnormally high levels of stress tomake them fail earlier (earlier failures mean lowertesting costs and quicker answers). Extrapolationis then used to convert the short life under severeconditions into the expected life under normalconditions. The same simple equation applies,except that the time t now becomes A x t, where Ais the Acceleration Factor (typically between 5 and150, see Acceleration Factors) and t is the timeunder stressed conditions. So, under AcceleratedLife-Tests:

Example: a group of 100 semiconductors havebeen stressed for 1000 hours in a life test. TheAcceleration Factor was 75, and 2 failuresoccurred :

93

Failures In Time Standard (FITS)

R(t) =

R(t) = exp (- λt)

no - nf

no

λ =no - nf

1

δt

δnf

λ =no

1

t

nf

λ = x 109 FITS.

.

.

no

1

t

nf

failure rate λ = x 109 100

1

3 x 365 x 24

2

failure rate λ = x 109 FITS.noAt

nf

failure rate λ =

= 267 FITS.

= 761 FITS.

100 x 75 x 1000

2 x 109

Confidence levelsFrom the failure rate determined from samplemeasurements an estimate of the failure rate ofthe whole population can be made by expressingthe maximum failure rate with a certain confidencelevel. Poisson statistics are used to calculateconversion constants (see Table) to be used whenconfidence levels are taken into account:

Taking the previous example (2 observed failuresconverts to 3.1), failure rate (with a 60%confidence level) would become:

Hence, with a 60% confidence, we can say thatthe actual failure rate will be less than 413 FITS.Similarly, failure rate would become:709 FITS (with a 90% confidence level), or839 FITS (with a 95% confidence level).

94

observedfailures

conversion constant at aconfidence level of:

0123

60%

0.922.023.14.17

90%

2.33.895.326.68

95%

2.994.746.297.75

failure rate λ =

= 413 FITS.

100 x 75 x 1000

3.1 x 109

95

General quality specificationsPhilips Semiconductors has issued a set ofGeneral Quality Specifications aimed at informingcustomers of our quality assurance system and atmaximizing product and service quality. Thesespecifications are:SNWEQ-611A

Small Signal Transistors, Diodes and SensorsSNWEQ611B

Power Transistors, Power Diodes andControlled Rectifiers

SNWEQ611CRF Power transistors, Hybrid Modules and Broad band transistors in metal-ceramic packages

SNWFQ611DIntegrated Circuits.

ScopeThese specifications relate to particular groups ofproducts, and each identifies particular types ofproduct. The specifications set out parts of theQuality Assurance Specifications that must beused by the technical organization within theProduct Centres, suppliers and subcontractors.The General Quality Specifications are inaccordance with ISO 9000 and represent theminimum quality requirements.

OverviewThe General Quality Specifications define thegeneral procedures that must be used for thedevelopment and manufacture of the devices andpackage outlines specified. They each cover themain aspects of product and process quality andreliability in: Development Production Management Defects Inspection and test requirements.

DevelopmentThe related sections outline the generalresponsibilities of the Development Departmentand the procedures for product, process andpackage release that must be followed.

ProductionThe related sections outline the quality assuranceprocedures that must be followed during production.It covers, among others, incoming and in-lineinspection, acceptance testing, qualityassessment, special approvals and audits.

ManagementThe related sections outline the proceduresnecessary to control the process between thesupplier (Development and Production) and thecustomer. The areas covered include qualityindicators and improvement planning, traceability,failure analysis and customer complaints, productor process changes, and customer notification andquality reporting.

Inspection and test requirementsThe related sections cover the in-line inspectionrequirements, Group A tests (acceptance tests perlot), Group B tests (conformance tests per lot),Group C tests (periodic inspection) and Group Dtests (qualification approval).

96

Green Flagship products

When it comes to developing methods of environ-mental monitoring that provide a powerful tool forcomparing the environmental performance ofdifferent products and processes at all stagesof their lifecycles, Philips Semiconductors takesa leading role.But commitment to the environment isn't limitedto a single activity or range of devices. Everybusiness line at Philips Semiconductors isactively working to incorporate Eco-Designprinciples into its products. Our Green Flagshipsolutions show Eco-Design at its very best,demonstrating how care for the environmentcombines perfectly with advanced technologies.Some of these products have obvious 'green'benefits, like the AE1000 self-powered radio,which runs off a hand-cranked rechargeablebattery, while some have implications which areless immediately obvious, yet highly significant.For example, our in-car navigation systems helpdrivers to avoid traffic jams, resulting in fewerexhaust fumes and cleaner air for all of us engineering elegance from PhilipsSemiconductors working to protect the naturalworld. Here are some typical Green Flagshipproducts from Philips Semiconductors: Less is morePhilips Semiconductors has shrunk a completecircuit board into a single chip. The M-AFRIC(Multi-standard Alignment FRee IF IC) is theworld's first IC of this type to appear on themarket. M-AFRIC can be used freely withoutexpensive external elements and supports a widerange of I2C-Bus inputs for control of tuner gain,frequency control and related features. A substantial advantage is the saving in PCBsurface, which allows a cost-effective solution fortelevision, VCR and multimedia applications. Thealignment-free concept allows the IC to be usedin the many different worldwide TV standards(PAL, NTSC, SECAM) and FM radio. M-AFRIC reduces packaging needs, power, heat

and energy consumption due to the reduced sizeof the IC and the dispensing of the need forexternal components such as coils, ceramic filtersand electrolytic capacitors.Winner of Philips Eco-Product 2000 awardPhilips Semiconductors new BISS (BreakthroughIn Small Signal transistors) won the Eco-Product2000 award for best Philips product. One of thekey features is the extremely low emitter-collectorsaturation voltage, which in turn leads to low powerconsumption and exceptionally high collector-current capability. The low power consumption ofthe BISS transistors makes them ideal for battery-powered applications in telecommunications andfor handheld devices. The low dissipation offersadvantages for the automotive segment andhousehold appliances in which power dissipationis often a critical factor. Based on PhilipsSemiconductors Eco-Design principles, thesetransistors significantly reduce energyrequirements and device sizes, making thema highly efficient 'green' solution. Today's telematics reduce tomorrow's trafficjamsUsed in automotive applications, PhilipsSemiconductors SAF3100 is a basic telematicsprocessor telematics being the integration of carnavigation and infotainment systems. Philips haslong been involved in the development of bustechnologies, and the SAF3100 telematics pro-cessor is specially designed to interface with thein-vehicle CAN bus for communication with otherelectronic devices in the car.

97

Reducing packaging, power and costThe OM6211 is the controlling device for an LCDscreen used in cellular phones, and is mounteddirectly on the LCD-cell by means of flip chiptechnology. Its area is 35% smaller than thecompetition's, and 19% smaller than itspredecessor (the OM4081 LCD driver IC). It alsouses half the power required by its predecessor,leading to reductions in size, power and materialsused. The OM6211 eliminates the need for nineexternal capacitors, reducing the manufacturingcosts and power consumption even further.Manufacturers are able to reduce packaging size,cost and power consumption, making the cellularphone cheaper to manufacture and to run. Theenvironment benefits as well as the consumer,with savings in material waste, production andenergy.

CATV amplifier moduleThe CGY887A is a hybrid dynamic range amplifierfor CATV systems operating in the 40 to 870 MHzfrequency range. It features high gain, superiorlinearity and extremely low noise, and uses goldmetallization to ensure excellent reliability.Operating at a supply voltage of 24 V DC, theruggedly constructed module comes in a SOT115Jpackage and employs both GaAs and Si dies todeliver all the required functions, minimizing thenumber of external components.

History of Philips Semiconductors

Philips has been a leader in the semiconductormarket right from the start, 50 years ago. In 1991its semiconductor operations became morefocused by the formation of an independentproduct division called ‘Philips Semiconductors’.Since then Philips Semiconductors has becomehighly successful – the world’s 9th largestsemiconductor supplier.So what happened in those 50 years since theinvention of the transistor? And how did PhilipsSemiconductors reach its top-10 position?

Those first 50 yearsThe invention of the transistor was announced tothe press 50 years ago. It was called the‘transistor’ because it was a resistor orsemiconductor device that could amplify electricalsignals as they passed through it (‘trans’ meaning‘through’ in Latin). The announcement did notcreate much excitement at the time because thedevice was simply seen as a compact and ruggedreplacement for the vacuum tube. Even the BellTelephone Laboratories team of inventors JohnBardeen, Walter Brattain and their group leader,William Shockley, viewed it as such. Nobody hadany idea what a colossal role the device wouldplay in revolutionising electronics over thesubsequent 50 years.

Before the transistorPrior to the invention of the transistor, Philips’work on lamp technology had led to a broadunderstanding of materials, vacuum technologyand glass processing. The birth of the radio, afterthe invention of the electronic valve in 1907,marked Philips’ diversification into electronics. Thefirst Philips radio valve was made in 1917 and thefirst complete radio in 1927. It was during the1920s and 1930s that research took place into X-ray tubes, thermionic valves and, later on, into TVpicture tubes and image intensifiers. Thesedevelopments formed a platform to support the

company’s continued diversification, for exampleinto the fields of medicine and dental surgery.In 1924, this research resulted in a portable (!)X-ray system, called the ‘Metalix’. Later on, imageintensifiers were applied to greatly improve thequality of X-ray pictures. They also increasedsafety by reducing the strength of the X-raydosages necessary for examination.Even during this period, the Company was rapidlybecoming a worldwide operation. In 1924 Philipsbought a 50 percent stake in the British Company,Mullard Radio Valve Co. Ltd. In 1925 it signed acooperation agreement with Radio RöhrenfabrikGmbH, and signed a 20-year contract forexchange of know-how, patents andmanufacturing rights with RCA of America. In1932, Philips acquired a 50 percent share of theFrench company, la Radio Technique. And by1934, Philips had valve factories in nine countries.From its start in 1918, when a customer in TheHague, the Netherlands, ordered just 180 valvesto start up a radio station, the total production hadrisen by 1933 to an amazing 100 million pieces.

Birth of the transistorAt the time of the invention of the transistor,Philips already had a contract of cooperation withBell Telephone Laboratories’ parent company,Western Electric. This opened up the way toPhilips making a contribution to the furtherdevelopment of the transistor. Early Philipsreactions were realistic rather than visionary.A report by the Philips Electron Tubes Division inOctober 1948 concluded: “If it proves possible tomanufacture transistors at a low cost price, with asufficiently long life, good stability and closeproduction tolerances, it is expected that newapplications will be found for such devices wherevacuum tubes have so far not proved suitable. Inthis way, the transistor may become a valuableaddition to the electronic tube”. Although it was aremarkable prediction in the context of an

98

99

apparently endless requirement for vacuum tubes,this professional assessment has since proved tohave been very accurate.

Further developmentsTwo years after the transistors invention, the firstgrown-junction transistor was produced. Itcomprised a sandwich structure and could handlesignificantly more power than its predecessor.Junction transistors were also made in whichsmall spheres of indium were placed on eitherside of a slice of germanium and heated until theindium alloyed into the germanium. The alloyprocess was subsequently replaced by diffusion.By 1951, Philips had delivered its first germanium-crystal diodes, and in 1952 the first mass-produced transistors left the production lines of arented building in Nijmegen, the Netherlands. In1953 a specialized factory was built in Nijmegento cater for this important new activity. Plants inHamburg, Germany, and in Micham, UK, soonfollowed. By 1954, Philips transistors were to befound in radios, and fast-switching transistorswere being supplied to manufacturers of electronicadding machines.

From transistors to circuitsAlthough they were initially more expensive thanvalves, transistors required no time to warm upand used 90 percent less energy. This made itpossible to manufacture more sophisticatedproducts. Electronic circuits could now be muchmore complex, using more active components,since a transistor was small, cheap, reliable anddissipated much less power than a valve. Soonmany more components became available tochoose from than had been available in the valveera. They satisfied demand in application domainslike radio and TV, and were also used in the earlycomputers for which much larger circuits wererequired.

New technologyIn 1959, diode production at Philips wastransferred to Stadskanaal, the Netherlands. Inthe same year, Philips opened their first transistorfactory in Switzerland. Others followed inBrussels, Belgium (1960), and in Klagenfurt,Austria (1961).Philips wafer production started in 1962, withsilicon slices of 19 mm diameter containing 1,000transistors. In 1964 Philips produced its firstintegrated circuit (for a hearing aid) in Nijmegen.The next design was for the extraction of soundinformation from a TV signal. In 1965, productionof ICs for colour television started inSouthampton, UK. That same year, the Philipsdivisions, Electron Tubes, and IndustrialComponents and Materials, merged to form theElectronic Components and Materials (ELCOMA)division.It is interesting to note the comparative sales ofvalves and semiconductors at this stage. In 1953-54, semiconductor sales had amounted to just 1.5percent of radio and TV valve sales. But by 1958-59 the figure had increased to 23 percent. In1963-64 it was up to 65 percent and by 1968 itstood at 95 percent.In 1966, production of consumer ICs began inHamburg, and Philips also cooperated withvarious other partners to start a joint venturecompany called Faselec AG in Switzerland tomanufacture ICs for clocks and watches (thePhilips factory was added to Faselec in 1969).Also in 1966, Philips Electronic Building Elements(PEBEI) Ltd. was established in Kaohsiung,Taiwan, for the production of ICs. In 1969,Electronic Devices Limited (EDL) was founded inHong Kong as a joint venture between Philips andT. Zau Sr., to produce discrete semiconductors(EDL became 100 per cent Philips-owned in 1997).Philips Research had been deeply involved insemiconductor technologies since the early years.

100

One of its most significant contributions came in1966 with the LOCOS (LOCal Oxidation OfSilicon) process. It offered a means of improvingthe isolation between transistors on a chip,resulting in greater packing density. The processis very important for the efficient integration ofmillions of transistors on one chip andmanufacturers all over the world are still using it.In 1970, MOS (Metal Oxide Semiconductor) ICactivities were started in Nijmegen, while furtherMOS facilities were also opened in Southamptonin 1975 and in Hamburg in 1977. In 1974, an ICtest and assembly facility was opened in Bangkok,Thailand.In 1975, Philips acquired one of the pioneering UScompanies, Signetics, which had wafer fabs inAlbuquerque, Orem and Sunnyvale. Theyproduced ICs in bipolar analog, bipolar digital andMOS technologies.

Division renamedAt this point, Philips split its semiconductoractivities into two parts Discrete Semiconductorsand Integrated Circuits. In 1988, the ElectronicComponents and Materials Division, to which bothof these activities belonged, was renamed asPhilips Components. Philips-ownedsemiconductor companies in several countrieswere brought under the Philips banner (e.g. Valvoin Germany, RTC in France and Mullard in theUK). The rebranding was completed in 1992 withthe renaming of Signetics.

Mega projectMuch of the cooperation going on at this time wasin conjunction with JESSI (Joint EuropeanSubmicron Silicon), the coordinated effort of ninemajor European IC manufacturers. However, in1990, Philips decided to withdraw from the so-called Mega Project, an initiative by JESSI toproduce high-density memories. This decisioninvolved stopping production of 1 Mbit SRAMs at

a state-of-the-art IC fab (MOS 3) in Nijmegen.Although the plant had only been completed in1987, the dramatic fall in memory prices provedthe wisdom of that decision and the plant hassince become an important production centre forconsumer and telecom ICs.

New eraIn that same year, Philips was in a financial crisis.Philips Components had become a very largeorganization. Philips President J.D. Timmerannounced a major restructuring programme tobring the necessary performance improvements.As a result of the growing importance of the twosemiconductor business units and the decreasingsynergy with other activities of PhilipsComponents, a new product division called PhilipsSemiconductors was born on January 1, 1991.Around one third of the total 75,000 employees ofPhilips Components transferred to the newdivision.The new organization was formed under theleadership of its first CEO, Heinz Hagmeister. Itcomprised six core product groups: ConsumerICs, Industrial ICs, Transistors and Diodes, PowerDevices, Standard Products and ApplicationProducts. Regional sales and marketingorganizations (RSOs) were announced. InEurope, the region was subdivided into salesareas North, West, Central and South, and in theFar East by country. The new organizationbrought many benefits: countries were able toshare front-line technical resources; the amount oflogistical and administrative interfacing betweencountries and worldwide staff departments wasreduced; sales forces could concentrate on thebusiness of selling, focusing on major accounts,while the distributor networks enhanced service tosmaller accounts. Moreover, rapid and structuredmarket feedback became possible anddeployment of policy became quicker and moreefficient.

101

Quality journeyCustomers soon began to experience improvedservice levels, and Philips Semiconductors beganto climb back to an increasingly competitiveposition. All plants were set the goal ofcertification to ISO 9000 standards, a target thatwas achieved as rapidly as November 1992. Thefollowing year, tough new goals were set and allplants were given the task of achieving the FordTQE quality standard by the end of 1993. The PD-wide QIC (Quality Improvement Competition) wasalso inaugurated in that year. Teams from all overthe world took part, signalling the involvement ofthousands of employees. By 1997, no fewer than667 teams comprising over 5,000 peopleparticipated in QIC. This meant that almost 20percent of the workforce were actively engaged inthe dedicated improvement process throughteamwork.

Consolidation and growthBoosted by a growing confidence, PhilipsSemiconductors began a period of cooperationand growth that encompassed agreements withother major manufacturers and expansion andrenewal at plants around the world. Over theperiod 1991 to 1996 the Company implementedwafer fabrication capabilities equivalent to 1.25million 8 inch wafers per year. These came fromfive new facilities: Caen and Limeil in France,Nijmegen in the Netherlands, Hazel Grove in theUK and Albuquerque in the USA, plus significantupgrades to a further five existing fabs.In 1991, a new IC and Application Centre wasopened in Southampton, UK. And in 1992, a newbipolar ICs fab, built as a joint venture with theShanghai No.7 Radio Factory, was opened inChina. In 1993, a joint venture with Motorola sawthe building of a new assembly and test facility forsmall signal transistors and diodes in Seremban,Malaysia. 1994 saw the completion of a newassembly and test facility in Thailand. And to meet

growing demand, a brand new plant was built inthe Philippines to house assembly and testfacilities for various discrete products.The first systems laboratory outside Europe wasopened in Sunnyvale in 1994, but the biggestannouncement that year concerned theinvestment of 500 million Dutch Guilders in asubmicron 8 inch wafer facility (MOS4YOU) inNijmegen. The plant would feature an advancedsubmicron process (0.5 micron and below)developed in Crolles near Grenoble, France, in ajoint project with SGS Thomson that was startedin 1992.At the end of 1994, an announcement was madethat Philips and IBM would cooperate in themanufacture of wafers at IBMs Böblingen facilityin Germany, initially producing 0.8 micron line-width logic products. It was a strange coincidencethat the two companies should work together 25years after they worked independently on animportant early innovation in IC technology. CalledIntegrated Injection Logic, this process offered thepotential for bipolar circuit speed with MOS circuitdensity. It was cross-licensed and adopted for usein microprocessors, custom gate-array chips andmemories. Today, the Böblingen facility is 100%Philips owned.Finally, in 1996, Philips Semiconductors jointlyestablished a major new software centre inBangalore, India, together with other Philipsproduct divisions. That same year, a newinternational production centre for discretesemiconductors was opened at Cabuyao in thePhilippines. Work on building a further test andassembly facility in the Philippines, this time forICs, was started in 1998 at Calamba, just south ofManila.

Growing customer confidenceBetween 1992 and 1995, customers began toreaffirm their confidence in the new PD andscores of supplier awards were won by Philips

102

Semiconductors around the world. In 1993, thenew CEO, Doug Dunn, was able to declare thatthe year had been a financial success for the PD.1994 proved to be a recordbreaking year forDiscrete Semiconductors. For the first time in itshistory over 10 billion parts were sold in a singleyear and sales exceeded 1 billion US dollars. Infact, Discrete Semiconductors made a substantialcontribution to the PDs recovery and financialsuccess. 1995 proved to be a record year for thePD. The semiconductor industry was powering itsrecovery from a recession and PhilipsSemiconductors grew an unprecedented 23percent.

Climbing even higherAfter another successful year in 1996, PhilipsSemiconductors returned to the worlds top tensemiconductor manufacturers, based on sales.It was seen as the result of hard work.This sentiment was clearly expressed by its newlyappointed third Chairman and CEO, Arthur vander Poel: Jumping back into the top ten is a justreward for all the hard work put into improving ourperformance and competitive position byemployees across the PD. In 1997 and 1998,Philips Semiconductors further consolidated itstop-ten position. In 1999 Philips acquired VLSITechnology, by which the sixth place wasreached.

Philips Semiconductors todayNew applications have indeed been found for thetransistor and the products its invention enabled.Although very hard to detect in everyday life,these little devices play a colossal role in peopleslives. And Philips Semiconductors has a solidpresence in their application domains.Now more than one in every three televisionsmade around the world is based on a PhilipsSemiconductors one-chip tv.

One out of every two telephones in the world usesa Philips Semiconductors line interface IC. Andmost of todays leading car manufacturers usePhilips Semiconductors car immobilisertechnology.Now, 50 years after the invention of the transistor,Philips Semiconductors is the tenth largestsemiconductors supplier in the world. It employsapproximately 30,000 people worldwide.Together, they produce around 70 million ICs anddiscrete semiconductors every day.

CHAR. PATTERN CHAR. PATTERN

1234567890ABCDEFGHIJKL

MNOPQRSTUVWXYZ-•

SPACE*$/+%

103

Identification labeling

The symbol denotes a unique start/stopcharacter which must be the first and lastcharacter of every barcode.

Fig. 1 Barcode patterns of the characters that can berepresented by CODE 39.

start character

stop character

leading quietzone

trailingquietzoneA B C∗ ∗

Fig. 2. Barcode representation of the characters ABC.

The identification label on our semiconductorpacking box is a combination of human-readabletext and machine-readable information. The labelidentifies the product, gives traceability andprovides additional information. All the informationis given in both human-readable and in 2D-codedform, some fields are also given in barcode. The2D symbol enables all the label information to beread in one action via our traceability informationsystem ROOTS. The 2D code is defined in DataMatrix ECC-200, the barcode is code 39.

What is CODE 39?CODE 39 derives its name from the structure ofeach coded character. Each character isrepresented by nine bars (four white and fiveblack) as shown in Fig. 1. Three of the nine barsare wide (binary value 1), the other six are narrow(binary value 0).Note that every barcode pattern starts and endswith the unique () character. Figure 2 gives anenlarged example of the bar-code patternrepresenting ABC.

104

Philips Semiconductorsidentification labelOne of our identification labels is shown in Fig. 3.The composition of the label is defined in the labelstandard SNW-SQ-404. The label contains thefollowing data (bottom to top):

Fig. 3. A typical identification label.

P H I L I P S S E M I C O N D U C T O R SASSEMBLED IN TAIWAN B.O.C.DIFFUSED IN EEC

LOT2(30T)DATE2(30D)QTY2 (30Q)LOT(1T)

DATE(9D)

ORIG(1L)PMC(31T)MSL(31P)

Z371DSD4

0965209944376033950

9931

QTY(Q) 500

TYPE(30P) SAA4977H/V1CODENO(1P) 9352 375 20518

80

80

Philips 12NC product code number (CODENO) Type number (TYPE) Packaging quantity (QTY) Production date (DATE) Traceability lot-ID (LOT) Product origin code (ORIG) Product manufacturing codes (PMC) Moisture sensitivity level (MSL) Country of origin (MADE IN), which can be

shown in 2 lines (as in Fig. 3):ASSEMBLED IN / DIFFUSED IN.

It can also contain any of the following data: Re-approval date (REDATE) Second production date (DATE 2) Second traceability lot-ID (LOT 2) Customer reference (CUSTOMER) Customer-specific information (CUST INFO) Additional product information (PROD INFO).

Identification labeling is described in qualitystandard SNW-SQ-404.

105

ISO (the International Organization for Standar-dization) is a worldwide federation of nationalbodies. International Standards are prepared byISO technical committees. Draft InternationalStandards adopted by these committees arecirculated to the member bodies for approvalbefore acceptance by the ISO council. The workof ISO covers all fields of standardization with theexception of electrical and electronic engineeringwhich, by agreement, are the responsibility of theInternational Electro-technical Commission (IEC).ISO 9000 guarantees conformance tospecifications and procedures, but does nothamper improvement actions. As shown underBusiness Excellence (page 11), ISO 9000 is thestepping stone to improvements.

Evolution of the ISO standardOn December 15, 2000, the third revision of ISO9000 was issued. Compared with the previous(1994) version, the 2000 revision implemented anumber of basic changes:- the new norm is applicable to all organizations

delivering any type of product, and is independent of the size of the organization

- the new norm has a wider attention to all aspects of the business

- there is a strong relationship with the benefit,of the stakeholders

- the organization should describe how it works rather than what it does

- the business should be able to show continuous improvement in a closed-loop cycle

- the new norm is more user-friendly and less descriptive.

In the new ISO norm, eight quality managementprinciples have been defined to lead theorganization towards improved performance:- customer focus- leadership- involvement of people- process approach- system approach to management- continual improvement

ISO 9000

- factual approach to decision making- mutually beneficial supplier relationships.In order to maintain certification, organizationsshould transfer to the new norm before December15, 2003.

ISO 9000 in Philips SemiconductorsPhilips Semiconductors follows the Philips policythat every operational unit should be ISO 9000-certified. In 1991, Philips Semiconductors starteda programme for the certification of all BusinessLines, Wafer Fabs, Assembly & Test operations,Sales and other operational units in Europe, theUSA and the Far East. The programme wascompleted by the end of 1992 for all Wafer Fabsand Assembly & Test factories and for somesupport centres.To obtain ISO certification, each centre is given aninitial preparation period, followed by an audit onall aspects of the norm. The audit covers thecentres total organization including management,logistics, purchasing, marketing and finance.Certification is granted by independent third-partyregistrars such as DNV, Lloyds, SGS Yarsley orKEMA. Philips has signed volume agreementswith the three best-performing registrars.Certification can only be granted afterdemonstrating that the quality system is inagreement with all the ISO requirements. Theregistrar will make periodic surveillances to ensurethat the quality system is kept up to ISOstandards. In general, the following rules areapplied regarding Quality systems and Certification:- quality systems (including quality manuals) are

set up according to ISO guidelines for entire organizations such as Business Lines, Wafer fabs, Assembly & Test factories and Technologygroups

- certification programmes are usually carried out on site level, where several units are combined.

It is common practice to combine ISO 9000certification with QS-9000 (see QS-9000/Automotive quality standards) or ISO/TS 16949,when applicable.

106

Table 1 Philips Semiconductors ISO 9000 certifications

Albuquerque, USA

Bangkok, Thailand

Barcelona, Spain

Cabuyao, Philippines

Caen, France

Calamba, Philippines

Eindhoven, The Netherlands

Eindhoven, The Netherlands

Hamburg, Germany

Hamburg, Germany

Hazel Grove, UK

Hong Kong

Hong Kong

Kaohsiung, Taiwan

London, UK

Milan, Italy

Nijmegen, The Netherlands

Paris, France

Singapore

Southampton, UK

Stadskanaal, The Netherlands

Sunnyvale, USA

Sunnyvale, USA

Taipei, Taiwan

Tokyo, Japan

Veldhoven, The Netherlands

Vienna, Austria

Zürich, Switzerland


Plant ICs

IC assembly

sales

module/metal-encapsulated discretes assembly

plant ICs

IC assembly

sales

headquarters

plant (IC/discretes)

sales

power semiconductors

plastic-encapsulated discretes assembly

sales

IC assembly

sales

sales

plant (IC/discretes)

sales

sales

teletext/digital audio ICs (CD, DCC, etc)

medium-power rectifier diodes

ICs

sales

consumer ICs

sales

central stores

sales

telecom and logic ICs

sales

June 1992

June 1991

October 1993

April 1991

May 1991

November 1999

October 1996

September 1998

December 1991

March 1995

November 1991

May 1991

June 1995

June 1991

May 1993

October 1995

September 1992

July 1995

June 1995

March 1993

September 1990

October 1991

December 1992

January 1996

October 1995

December 1991

October 1994

January 1993

January 1995

Centre Certified unit Certification date

Note: All centres are recertified every 3 years, prompted through a contractual agreement by the certification body.The certificates are available on Philips Intranet.

107

Marking of ICs

As Fig. 1 shows, the top side of an integratedcircuit from Philips Semiconductors containsidentification marking. This marking normallycomprises three lines (A, B, and C) of text andthe Philips logo.Line A contains the commercial type number andoptional designations for crystal technology and forpackage type.Line B contains the diffusion lot number and a2-digit assembly sequence number, for traceabilityto the assembly batch.Line C contains three code letters designating thediffusion centre, the assembly centre, and thecentre of responsibility, followed by the assemblyyear and week date code (YYWW), the masklayout version and the release status code (blankfor released products, X for development samplesor Y for qualification samples). For the code lettersof the centres see Product manufacturing codes.A typical example of IC marking is:Line A: 74HL33534DLine B: K3P08604Line C: Hnn9322 A X

On very small packages the marking is condensedby truncating the information into two lines.

Pin 16

Pin 1

9

8

Line A

Line B

Line C

Fig. 1 The IC marking format.

108

Mean Time Between Failures (MTBF)

The period of time that a piece of electronicequipment will run without failure is a criticalparameter. The MTBF (called MTTF, mean time tofailure, for non-repairable equipment) is arelatively simple calculation to make, yet it causesdifficulties for many electronics engineers. Thepurpose of this section of the handbook is to givea simple explanation of MTBF (MTTF)calculation.

Failure rateAfter testing electronic components, failure rate isgiven by:

number of failuresfailure rate =

quantity tested x test time

Failures can be complete, partial, sudden, gradualor intermittent, and so may not show up at allduring the test time. For this reason we usuallyquote assessed figures, based on a confidencelevel of 60%. The relationship between observedfailures and assessed failures is:

observed failures assessed failures0 < 0.91 < 2.02 < 3.13 < 4.24 < 5.2 etc.

Example500 components were tested for 10,000 hours, 1 failure was observed:

Failure rate is usually expressed in FITS (failuresin time standard, 1 FIT = 10-9 per hour). So, anassessed failure rate of 0.4 x 10-6 per hour = 400FITS.

Calculation of MTBFWhen the components are built into a piece ofelectronic equipment, the assessed failure rates ofeach individual component add up to give theassessed equipment failure rate.

ExampleConsider a piece of electronic equipmentcontaining 319 electronic components, as shownin the Table below. For each component type theassessed failure rate will be known:

observedfailure = = 0.2 x 10-6 per hour, orrate

1

500 x 10000

MTBF = = 49505 hours1

equipmentfailure rate

109

20200

assessedfailure < = 0.4 x 10-6 per hour.rate

2

500 x 10000

ICdiodetransistorresistorcapacitorcoil

Equipment failure rate = 20200 FITS

9209

14012021

5001202002020

300

450024001800280024006300

quantityin

equipment

component assessedfailure rate

(FITS)

combinedfailure rate

(FITS)

So, based on continuous working for 8 hours aday, 5 days a week, the equipment should runsuccessfully for 23 years.

109

Moisture sensitivity level (MSL)

If packed or stored incorrectly, moisture-sensitiveplastic SMDs can be easily damaged by exposureto the high temperatures associated with soldering.If any moisture is present in the plastic packageduring soldering, it may turn into steam and expandrapidly. Under certain circumstances the forcecreated by this expansion can cause internaldelamination and, in the most severe conditions,cause internal or external package cracks (thepopcorn effect). This effect can be more prominentwith infra-red or vapour-phase reflow solderingmethods. The effect is less in wave soldering,which only exposes the devices to hightemperatures for a very short time.To minimize this problem, Philips Semiconductorsdelivers moisture-sensitive ICs in a resealablemoisture-resistant packing (see Drypack).

Determining moisture sensitivity levelNot all plastic packages are equally sensitive tomoisture. Each has its own moisture sensitivitylevel (MSL) which is influenced by:- chip size- package body size- package material properties.Philips Semiconductors determines MSL by testingbatches of each package type. After moisturizingthe package to a predetermined level, it is heatedto high (soldering) temperature and then cooled.The package is then checked for functionality and,if necessary, the test is repeated at a higher levelof moisturization.All Philips Semiconductors test centres performthese MSL tests and classify a MSL for eachpackage type. Moisture sensitivity levels rangefrom MSL = 1 (device not sensitive to moisture)to MSL = 6 (device very sensitive to moisture).

Determination of MSL is covered in qualitystandards:- SNW-FQ-225A

(SMD preconditioning specification)- SNW-FQ-225B

(Moisture sensitivity level assessment method).The MSL corresponds with a certain out of bagtime during which the product can be safely usedwithout damage during soldering. For these out ofbag times see Drypack.

110

Philips Business Excellence (PBE)

Fig. 1. The Business Excellence Model.

Leadership100

Keyperformance

results150

Processes140

People90

Policy &strategy

80

Partnerships& resources

90

Peopleresults

90

Customerresults

200

Societyresults

60

Enablers Results500 points 500 points

Innovation and Learning

Forwards from PQA-90Organizations that have already achieved thePQA-90 Award are continuing their drive forbusiness excellence by working towards achievingprestigious external awards such as the MalcolmBaldridge National Quality Award (USA), theDeming Prize (Asia Pacific) and the EuropeanQuality Award (TEQA). Each of these awardsrecognizes exceptional achievement, and in thecase of the USA and European awards, they arebased on a comprehensive model for businessexcellence.As preparation for an external award, Philips hasadopted Philips Business Excellence (PBE) whichis a core part of the BEST program.In PBE, organizations will be assessed using thesame comprehensive Business Excellence Model(Fig. 1) and criteria as that used to assessorganizations for the European Quality Award.

The Business Excellence Model (Fig. 1)The Business Excellence Model recognizes thatcustomer satisfaction, people (employees)satisfaction and impact on society are achievedthrough leadership-driven people management,

policy and strategy, resources and processes, allof which ultimately lead to excellence in businessresults.Each of the nine parameters of the model can beused to assess an organization’s progresstowards excellence. The scoring points shown foreach parameter equate to the percentages usedfor the European Quality Award (i.e. 100 = 10%etc.), such that the total points allocated (1,000)equates to 100%. The model is split (500 scoringpoints each) equally between the “enabler”parameters (concerned with how an organizationapproaches its business in each of the areasshown) and the “results” parameters (concernedwith what an organization is achieving and hasalready achieved).Each of the nine parameters is described below: Leadership How leaders develop and facilitate

the achievement of the mission and vision,develop values required for long-term successand implement these via appropriate actionsand behaviours, and are personally involved inensuring that the organization’s managementsystem is developed and implemented.

111

People How the organization manages,develops and releases the knowledge and fullpotential of its people at an individual, team-based and organization-wide level, and plansthese activities in order to support its policy andstrategy and the effective operation of itsprocesses.

Policy and strategy How the organizationimplements its mission and vision via a clearstakeholder focused strategy, supported byrelevant policies, plans, objectives, targets andprocesses.

Partnerships and resources How theorganization plans and manages its externalpartnerships and internal resources in order tosupport its policy and strategy and the effectiveoperation of its processes.

Processes How the organization designs,manages and improves its processes in order tosupport its policy and strategy and fully satisfy,and generate increasing value for, its customersand other stakeholders.

People results What the organization isachieving in relation to its people.

Customer results What the organization isachieving in relation to its external customers.

Society results What the organization isachieving in relation to local, national andinternational society as appropriate.

Key performance results What the organizationis achieving in relation to its plannedperformance.

Assessment and points scoringWithin the enabler parameters, assessmentaddresses the excellence of the approaches usedand the extent of the deployment of theseapproaches, both vertically through all levels of theorganization and horizontally across all areas andactivities. Each of the enablers parameters is scoredon the combination of two factors: the degree ofexcellence of the approach, and the extent of the

deployment of the approach.Within the results parameters, assessmentaddresses the organizations trends andachievements in terms of the actual performancecompared to targets set and, wherever possible,compared to competitors (particularly the best inclass) for the results under review. A key distinctionwithin the results parameters is the recognition ofboth direct feedback data from the relevantstakeholder, and the internal measurement ofpredictive performance measurements. These twoareas are sometimes described as the leading andlagging indicators of performance. Each of theresults parameters is scored on the combination oftwo factors: the degree of excellence of the results,and the scope of the results.

Implementation of PBE in PhilipsSemiconductorsPBE is implemented for the BUs and sharedresources of Philips Semiconductors and for thepoduct division as a whole.Yearly assessments are carried out for these unitsby a team of high-level assessors which includesassessors from other product divisions.

Corporate awardsIn the Philips BEST program, achieving certainmaturity levels in PBE is recognized by PBEAwards. These levels are: bronze (500) silver (600) gold (700).

The first Philips (bronze) PBE award was achievedby BU MMP (now DMI) in November 2000.

112

Philips Business Excellence policy

113

Philips Quality

Philips Quality defines the partners, frameworkand conditions for Building the Winning Company.

Philips Quality partnersIt is the interaction between four partners whichshapes Philips Quality: Customers who buy the companys products

and services People, all of us, who are the company Leaders who merge the interests of all who

have a stake in the company Suppliers who provide materials and services.

Philips Quality FrameworkPhilips Quality provides the framework forinteraction between the partners. Policy Deployment communicates and

translates company objectives throughsuccessive organizational layers, therebyempowering every individual to contribute tothe common goal. It provides the framework for

the interaction between the leaders and thepeople.

Process Management puts all tasks in theperspective of the challenge to surpasscustomers expectations. It provides theframework for the interaction betweencustomers, the company and suppliers.

Continuous Improvement means striving forperfection in a systematic and coherent way.Interaction between the four partners providesmomentum to the improvement task.

Achieving Philips QualityAchieving Philips Quality requires managingchange. An important task of managers is tocreate the conditions leading to change inprocesses: Organise, Communicate, Learn, Recognize,

Imagine Measure, Assure, Analyze, Audit, Benchmark.

114

Philips Values

The Five Corporate Philips Values are listed heretogether with some practical ideas on how we putthem into everyday practice.

1. Delight customersListen to customers and actively seek theiropinion.We add value through all our actions askyourself Am I actively improving Philips productsor services through this actionMake the customer visible, especially to those notin day-to-day contact with customers.Be a business partner not simply a supplier lookat all your actions from the customers point ofview.Treat both internal and external suppliers as avalued part of the customers supply chain.

2. Value people as our greatestresource

Ensure every employee has equal opportunity forrecognition and career development.Monitor, coach and support people look foropportunities to encourage personal development,and act on them.Listen and communicate to all levels make timeto hear views and give out information.Be willing and able to reward fairly define howwe expect others to behave and what we wantthem to achieve and make sure we are consistentin our rewards.

3. Deliver quality and excellence inall actions

Understand quality, ask yourself How can Ideliver quality in my own role and encourageothers to do the same?Demonstrate commitment to quality let otherssee us delivering quality, show them how to followsuit and reward achievements.

4. Achieve premium return onequity

Set clear financial targets and make sure yourcolleagues understand them and know how toachieve them.Commit to these targets do not move thegoalposts.Maintain continuous cost evaluation andeducate others to do the same.Make our financial targets your number onepriority.

5. Encourage entrepreneurialbehaviour at all levels

Encourage freedom of ideas within clearly definedborders educate your colleagues to understandwhat makes an idea a truly practical suggestion.Use realistic reporting work on an idea beforesubmitting it and encourage others to do thesame.Agree risk assessment look at the risks andpresent the pros and cons formally, in writing.Be a supporter you are part of the team, sosupport with actions not just words.Select, reward and promote people whodemonstrate entrepreneurial behaviour.

115

PPM

Fig. 1. Reject-level values depend on the point of measurement. Lots of most semiconductors will be inspected at anumber of points between our production and our customers' assembly. The value obtained at the last point is evidently thecritical one: values obtained earlier can be quality indicators only.

Component manufacturer

Component production

Acceptance sample inspection Incoming inspection

Equipment production

Incoming, defect level

Conformance to specification Fitness for use

Stores & delivery

Line fall-off(gross line reject level)

Analysis &confirmation process

Net line reject level

Acceptance testprocess average

OEM

PPM – literally 'parts per million' – is the measureof quality used in conjunction with Zero Defects-oriented quality-improvement activities. PPM is ameasure of actual quality, as distinct from thelimiting quality set by AQL-based samplinginspection which is given as a percentage.The use of ppm – usually numbers in the region10 to 1000 – is considered to make people moreconscious of quality levels, and to distinguish ppmfrom AQL (given in %).

WHICH PPM?Many ppm values are usually measured for agiven product during the course of its manufactureand use, Fig. 1. Principal among these areprocess average reject level or estimated processquality, which may be given for inoperatives, or formechanical/visual or electrical rejects. Theselevels are calculated from the manufacturers'

Acceptance Testing results: defect level: derived from customers' receiving

inspection (where this is still carried out). receiving inspection level: the reject level

arising from customers' receiving inspection. line reject level: the reject level resulting from

customers' assembly-line testing; generallydivided into gross and net values; the net valuebeing that agreed between customer andmanufacturer after failure analysis.

Thus, ppm figures can relate to conformance tospecification: inoperatives only (electrical and/or machanical

visual) all electrical defects mechanical/visual defects all defects combined.

116

They can also relate to fitness for use : gross line rejects NET (OEM) NET confirmed electrical defects mechanical/visual defects all defects combined application OEM test/inspection stages.

OUR PPM VALUESWe record and publish several process averagevalues for our semiconductors, derived fromAcceptance test results. In any production period(week, month, quarter, etc.) m lots are submittedfor Acceptance testing, a sample of size n is takenfrom each lot, and x rejects are found duringtesting. The number of rejects found is used toestimate the number of rejects in the whole lot :

Xn = Nn (xn/nn)

where Nn and Xn are the lot size and estimatedrejects in the whole lot, respectively.Then, for each test, the process average

X1 + X2 + .... + XmP = x 106 (ppm).

N1 + N2 + .... + Nm

Thus, for the electrical test, n will be the sampletested and x the number of rejects found in thattest; these are converted to Xn for a batch sizeNn. Process average values for mechanical/visualare similarly derived.The operation of our outgoing PPM ProcessAverage system is such that the results of all theQ & R sample tests per type and per family areaccumulated from all lots and only exclude clearrogue lots. This yields EPQ and ensures that theresults represent the true process average as

would be received by the customer, andeliminates problems which arise when trying tomeasure AOQ (Average Outgoing Quality) whenan acceptance sampling scheme is based on zeroacceptance number.

VerificationPPM values, especially low values, are extremelydifficult to confirm. Reliable verification requiresstable application conditions, fixed assemblymethods and, above all, large quantities.

117

PQA-90

PQA-90 stands for Philips Quality Award (forProcess Management) for the nineties. It sets outstandards for quality improvement that unitsthroughout the concern worldwide must achievethrough the nineties. The award is granted by thePhilips Group Management Committee.PQA-90 provides a structure for managingimprovement by systematically evaluatingimprovement organization and activities.

The PQA-90 AwardThe PQA-90 Award provides recognition foroutstanding performances in managing businessprocesses and thereby satisfying our customers.Its purpose is to support and stimulateorganizations in their efforts to achieve PhilipsQuality. The PQA-90 criteria (see below) serve asa working tool for planning, training andassessment, and provide a framework for movingtowards world-class quality. In addition, thecriteria provide a common language tocommunicate requirements throughout theorganization and with our partners.

The PQA-90 criteriaAn essential step in the systematic improvementof a process is to check against criteria. ThePQA-90 criteria are compatible with the standardsdefined in other supplier awards and with those ofprestigious awards such as the Deming Prize(Japan), the Malcolm Baldrige National QualityAward (USA) and The European Quality Award(TEQA). The PQA-90 criteria are defined in sixcategories: Role of Management

provide leadership and set conditions deploy policy monitor progress and initiate actions influence norms and values by personal

attitude.

Improvement Process process management organizing the improvement process continuous learning problem-solving discipline.

Quality System quality procedures document control internal audits corrective actions.

Relationship with Customers customer-needs are guidance for action customer partnerships customer interface.

Relationship with Suppliers supplier assessment preferred suppliers supplier-partners.

Results customer satisfaction performance of suppliers process control.

The PQA-90 assessment processIn the PQA-90 process, the progress of units ischecked (see Fig. 1) via self-evaluation, peerauditing (cross-functional) and managementassessment (cross-business group), therebycreating opportunities for shared learning and forpreparing groups for external assessment.

Philips PQA-90 Awards and certification dates areshown in Table 1.

118

Positive?

Positive?

Positive?

Assessmentworkshop

Improve!

Improve!

Self-evaluation

Prepare for e.g.TEQA

Malcolm Baldrige Deming Prize

Continuousimprovement

Awardceremony

Nomination byCQC-member

Self evaluation

PQA-90

no

yes

no

no

yes

yes

Criteriaunderstood?

Positive?

Competing forworld-class awards?

Award olderthan 3 years?

Approvalby CQC?

no

Peer audit

Managementassessment

yes

yes

yes

yes

yes

yes

mostlyno

no

no

no

no

Applying forPQA-90 Award?

Fig. 1 The PQA-90assessment process.

119

Table 1 Philips Semiconductors PQA-90 awards

Hazel Grove, UK

Sunnyvale, USA (ASLBG / MCO / CPG / Finance

Kaohsiung, Taiwan

Bangkok, Thailand

Caen, France

Barcelona, Spain (sales)

Sunnyvale, USA (sales)

London, UK (sales)

Hong Kong (EDL)


Cabuyao, Philippines

Southampton, UK

Albuquerque, USA

Nijmegen (T&D), The Netherlands

Nijmegen (CIC), The Netherlands

Hamburg, Germany

Hong Kong (sales)

Stadskanaal, The Netherlands

Nijmegen (ITEC), The Netherlands

Nijmegen (MOS-3), The Netherlands

February 1996

March 1996

June 1996

June 1996

December 1996

February 1997

April 1997

April 1997

May 1997

June 1997

June 1997

June 1997

November 1997

December 1997

December 1997

January 1998

January 1998

October 1998

February 1999

March 1999

Centre Certification date

120

Codes for ICs

Diffusion Assembly Responsible Centre

B San AntonioC/F/V C Albuquerque (fab 4/6/8)

D D Hamburg G Southampton

H/P/T/U H Nijmegen (CS/MOS-3/MOS-4/MOS-2) I Gratkorn, Austria

J/M J Caen (fab 5/6) M Zürich P Calamba S KaohsiungY ShanghaiZ Singapore6 Limeilb Böblingenk Hsin-Chu m Taipei n Bangkokr r Sunnyvale

Product manufacturing codes

The product manufacturing codes marked on theenvelopes of discrete semiconductors and ICs arelisted in full in specification SNWSZ602.For discrete semiconductors the code is a singleupper-case or lower-case letter identifying thecentre where the component is manufactured.For ICs the code is a single letter or figureidentifying where diffusion and/or assembly and/orfinal (QA) test is carried out.The following lists give the codes for the mostimportant Manufacturing Centres.

Codes for discrete semiconductors

Code Manufacturing CentreD Philips Semiconductors, Hamburg,

Germany

E Philips Semiconductors, Hazel Grove, UK

F Philips Semiconductors, Stadskanaal, The Netherlands

H Philips Semiconductors, Nijmegen, The Netherlands

P Electronic Devices Ltd. (EDL), Hong Kong

W Philips Semiconductors Guangdong, China

m Philips Semiconductors Philippines Inc., Cabuyao, Philippines

t PSS, Seremban, Malaysia.

121

Product Quality & Reliability Assurance database (PQRA)

Fig. 1. Typical quality data from PQRA.

As part of our Total Quality Management (TQM),Philips Semiconductors is committed to sharingQuality & Reliability (Q&R) data with ourcustomers. To be helpful to the customer, thedata shared must be correct, up-to-date andmeaningful. To issue Q&R data in the mostprofessional, efficient and effective way, we havedeveloped a database (PQRA), loaded with thelatest Q&R data: Product Quality Assurance data Product Reliability Assurance data.

Quality dataPQRA data provides Estimated Process Quality(EPQ) figures (in ppm, including or excludingrogue lots), either for product families or forspecific products. These figures show productquality after final test, calculated from actual QAtest results.

Figure 1 shows typical quality data. It contains: product type and family test type (e.g. Electrical) month and year when tested number of lots tested in the month Lot Acceptance Rate (LAR) total quantity in all lots in the month EPQ expressed in ppm.

122

Reliability dataPQRA data provides reliability data for productfamilies or specific products. The informationprovided includes FITS (Failures in TimeStandard), FPM (Failures Per Million) and rawbatch data.

Figure 2 shows a typical FITS/FPM output.

Fig. 2. Typical reliability data (FITS and FPM).

Accessing the PQRA databaseThe database can be accessed through thePhilips Intranet by employees of PhilipsSemiconductors. A password is required foraccess. For information to customers, the localsales office can access the PQRA database fora report, and communicate this to the customer.

123

QS-9000/automotive quality standards

QS-9000The QS-9000 Quality Systems Requirements(QSR) were developed by the Chrysler/Ford/General Motors Supplier Quality RequirementsTask Force as a standard covering qualityassessment for suppliers of automotive parts.Previously, each automotive manufacturer haddeveloped its own expectations for supplier qualitysystems and the associated assessmentdocuments. The most recent issue of QS-9000 (Rev. 3) wasissued in 1998. It follows the ISO 9000:1994structure.Together with the QS-9000 requirements, a QualitySystem Assessment document was issued.Fulfilling QS-9000 requirements meanscompliance with a number of linked documents:- Failure Mode and Effects Analysis (FMEA)- Statistical Process Control (SPC)- Advanced Product Quality Planning and Control

plan (APQP)- Production Part Approval Process (PPAP)- Measurement System Analysis (MSA).

QS-9000 requirements for the semiconductorindustrySpecific semiconductor requirements are notincluded in the QS-9000 documents mentionedabove, and some of the requirements formechanical parts are not applicable to thesemiconductor industry. For this reason theAutomotive Electronics Council (AEC) wasestablished to provide the appropriate automotivequality requirements for the semiconductorindustry. The AEC produced the SemiconductorSupplement to the QS-9000 QSR. Since then, anew standard (AEC A-100) has been producedfor the assessment of quality systems in thesemiconductor industry.

ISO/TS 16949In the late nineties, the International AutomotiveTask Force (IATF) was formed with the mission todevelop a global standard for automotive qualitysystems. It is a cooperation between ISO andrepresentatives from the North American andEuropean Automotive organizations. Next to thebig three, represented by the InternationalAutomotive Oversight Bureau (IAOB), the followingEuropean organizations are represented:ANFIA (Italy), VDA (Germany), CCA/ FIEV (France)and SMMT (UK). The first document was published in 1999 andwas called ISO/TS 16949:1999. It used the ISO9000:1994 structure. At that time, ISO/TS 16949registration was optional, as an alternative to QS-9000. On March 1, 2002 the second revision willbe issued, based on the ISO 9000:2000 structure. No more revisions of QS-9000 will be issued.Organizations which are QS-9000 certified andwant to continue their automotive registration,should transfer to the new norm by December 15,2003, the same deadline as for ISO 9000:2000.

Automotive quality standards in PhilipsSemiconductorsPhilips Semiconductors has been working withQS-9000 since September 1994, when a trialaudit was carried out in Albuquerque attended bymany representatives from the automotiveindustry. The QS-9000 requirements were seenas a good opportunity to drive continuousimprovement in all Philips Semiconductors sitesafter having achieved the TQE level. Certification is done by accredited third-partyregistrars that mostly combine QS-9000 auditswith ISO 9000 audits. Registrars used by Philips Semiconductors allhave ISO 9000 and QS-9000 accreditation.

124

Our policy with respect to certification accordingto QS-9000 or ISO/TS 16949 is as follows:- all shared resource production units (Wafer fabs,

Assembly & Test Units) should be certified - Business Units/Lines delivering to the

automotive industry should also be certified - External wafer foundries, acting as

subcontrac-tors, should also be certified according to QS-9000 or ISO/TS 16949. Preferred IC Assembly & Test subcontractors have to reach SAC level 1 certification (see Semiconductor Assembly Council).

Table 1 Philips Semiconductors QS-9000 certifications

Centre Certified Unit Certification date

Nijmegen, The Netherlands Site (covers multiple BLs and Fabs) Jan 1998Hamburg, Germany Site (covers multiple BLs and Fabs) Dec 1997Sunnyvale, USA Site (covers multiple BLs) July 1996Albuquerque, USA Site (covers multiple BLs and Fabs) May 1996Caen, France Site (covers multiple BLs and Fabs) May 1999Zürich, Switzerland Site (covers multiple BLs) Aug 1998Hazel Grove, UK Power semiconductors Dec 1997Stadskanaal, The Netherlands Medium-power rectifier diodes Oct 1997EDL, Hong Kong Plastic-encapsulated discretes assembly Jan 1998PSPI Cabuyao, The Philippines Module/metal-encapsulated discretes assembly Mar 1998PSS Seremban, Malaysia Plastic-encapsulated discretes assembly Aug 1998SMST Böblingen, Germany IC Wafer fabrication June 1998PSK Kaohsiung, Taiwan IC Assembly & Test Jan 1998PST Bangkok, Thailand IC Assembly & Test Sept 1997PSC Calamba, Philippines IC Assembly & Test Dec 1999

Note: All centres are recertified every 3 years, prompted through a contractual agreement by thecertification body. The certificates can be downloaded from the Philips Semiconductors Intranet.

125

Quality Function Deployment (QFD)

With Quality Function Deployment therequirements of the customers are deployed to thedevelopment and manufacture of new products.QFD is used to translate customer requirements(expressed in customer language) into productspecifications or design parameters. It alsoprovides the opportunity to assess possibleimprovements and set targets.The word 'Quality' in QFD means 'the ability tosatisfy stated or implied needs', so it is used in thesense of satisfying customer needs orrequirements, rather than being specific to aQuality Department or Quality System.With QFD, customer requirements areimplemented at the concept phase ofdevelopment, where the product design exists onlyon paper, so customer needs can be implementedwithout expensive equipment or process changes.This is an ideal opportunity to fully implement anyapplication aspects. The aim of QFD is to getbetter products sooner: better products because

they will better fulfil the stated and implied needsof the customer: sooner because in thespecification stage all parties are involved andpriorities have been set for improvements towardscustomer requirements.To be successful, QFD must tackle the real needsof the customer. These can only be determined byface-to-face customer visits or detailed customerresearch.QFD requires multi-discipline team effort, bringingtogether the skills of Marketing, Development andManufacture.During QFD, each important specificationparameter of the product is benchmarked againstcompetitor products.

The House of QualityThe main tool of QFD is the House of Quality(Fig.1). At the centre of this 'house' (under theroof) is the interaction matrix, where the customerrequirements are applied to the design

What

How

Target values

Interactingdesign req's

Targetgroups

Product evaluations and project objectives

Fig. 1 The House of Quality.

Customerimportance

ratings

Competitivebench-marking

Competitiveanalysis

data

ProjectobjectivesDesign parameters

Tech

nica

l ana

lysis

and

deve

lopm

ent t

arge

ts

Interaction matrix

Unit of measurement

Technical importancerating

Technicalperformance data

Design targets

Customerrequirements

Priority setting

Technicalbenchmarking

126

parameters influencing those requirements. Here,the translation is made from customer language todesign language. Figure 2 takes the House of Quality model into aworking scenario. Here, the customer iscategorized as a group of 'sound freaks', having aspecific set of requirements imposing different'weight' factors.The 'What' section, at the left of the House ofQuality, lists four customer needs for an audio set.Under the roof (the How section) are listed fourrelevant design parameters, and the interactionmatrix compares What with How.In the roof the interaction between designparameters are indicated. In the example shown,a higher number of controls has a positive (+)influence on harmonic distortion.The relevant interaction between What and Howis assessed, with interaction factors of 9 (strong),3 (medium), 1 (weak) or none at all.The product of demanded weight (from the

customer) and interaction factor gives theweighted sum of the design parameter, which canthen be ranked in accordance with the need toimprove it.For example, referring to Fig. 2, the customer hasdemanded a weight of 6.25 for natural sound. Thishas been assessed in terms of the relevant designparameters and given an interaction of 9 (strong)against frequency response and harmonicdistortion. The product (9 x 6.25) gives a weightedsum of 56.25, placing these two designparameters in ranking position 1 for improvementto satisfy the customers' needs.Below the 'House of Quality', the question of howfar each design parameter must be improved isanswered by benchmarking the parameter againstthe products of our main competitors. Forexample, in Fig. 2, our product has a frequencyresponse of 0.7 dB whereas Company A's producthas a better figure of 0.5 dB. A target value of 0.5dB is therefore set for improvement.

127

Fig. 2 House of Quality scenario 'Sound freaks'.

Tech

nica

l ben

chm

arkin

g(p

rese

nt v

alue

)

Natural sound

Adjust. freq. band

Easy to operate

Radiates quality

Weighted sum

Ranking

Unit of measurement

Our product

Company A's product

Company B's product

Target values

Demandedweight

6.25

12.5

0

2

How

Target group:Sound freaks

What

Legend

∆

+

= 9

= 3

= 1

= negative

= positive

56.25

1

dB

0.7

0.5

1.2

0.5

56.25

1

dB

-40

-60

-60

-60

∆

39.5

2

#

9

12

19

17

6

3

Type

pvc

pvc

al.

pvc

Freq

. res

pons

e

Harm

onic

dist

orsio

n

Num

ber o

f con

trols

Fron

t mat

eria

l

+

128

Quality Improvement Competition (QIC)

Employees who work with a product or processevery day are the most qualified to change orimprove it. Such is the philosophy behind theworldwide Quality Improvement Competition (QIC,pronounced quick) of Philips Semiconductors.Launched in 1993, QIC encourages employeesworldwide to form or join a Quality Improvementteam with the following aims: to encourage improved levels of customer

service and quality throughout the Company to promote structured teamwork and the

effective use of improvement tools to recognize and reward success.

Competition structureProject suggestions must come from individualemployees, must be supported by localmanagement, and must have a direct link toQuality Improvement.After an initial screening process, projects areofficially registered with the competitioncoordinators, an official registration certificate isissued, and the project team begins its work.At the end of the calendar year, each project teammust present its results at a local final, chaired bylocal plant management. The local winners arethen entitled to compete in one of three regionalfinals, whose jury is chaired by a director of Sales.Regional finalists receive special certificates andgain recognition throughout the company byarticles in the magazine Signal. In regional andworld finals a major customer is invited toparticipate on the jury.From the regional finals, twelve teams arenominated for the world final, held in the first halfof the following year, and chaired by the GeneralManager International Marketing & Sales ofPhilips Semiconductors. The overall winning teamreceives the prestigious QIC trophy, based on thePhilips Quality Triangle design. In addition, thefinalist teams themselves select the public prize.

Project assessment criteriaTo ensure a properly structured competition, andto let participants know what is expected of themand how they will be assessed, scoring is done onseven project assessment criteria based on ateamwork approach to problem solving andincremental improvement:1. Project selection 4. Results2. Project analysis 5. Project learning3. Solutions 6.Working as a team

Results to dateThe QIC has proved to be an excellent way ofpromoting the involvement of employees inQuality Improvement activities.In the 2000/2001 competition, the number ofteams competing was 830, showing an increasingparticipation. The final took place in Cape TownSouth Africa, where the twelve team finalists werefrom the USA (2), Germany, Philippines, Thailand,UK (2), the Netherlands, Taiwan, Malaysia, Franceand China. The winning team was LPG Lead FrameTesting from PST Bangkok, Thailand. Their projectof parallel product testing in minimum test timeincluded new testing methodology which reducestest time and cost by as much as 50 per cent.

Cost-saving benefitsAlthough the competition is intended to stimulatequality through employee teamwork, resultingimprovements also lead to major cost-savings.It is estimated that the annual cost-savings achievedby the finalists in the inaugural competition werearound 20 million US dollars. And when savingsby the other participating teams were also takeninto account, the competition achieved an overallproductivity improvement of around 20% for thesubjects under improvement.

129

Year Location Teams QIC award Vox Populi award1993 Singapore 235 Heat diffusers, Manila1994 Monte Carlo 273 Tropical Depression, Manila Probe Fighters, Zürich1995 Orlando 345 WIRA, Malaysia Forming Busters, Malaysia1996 Shanghai 527 Team 2000, Sunnyvale RF Power Vision, Manila1997 Rome 667 E.T.F.II, Hazel Grove Pocket Team, Bangkok1998 Las Vegas 714 PCF Vision 2000, Hamburg Climber, Hong Kong1999 Bangkok 676 BonA, Nijmegen Q Formers, Calamaba2000 Cape Town 830 LPG Lead Frame Testing, BEST in CSS, Chengdu

Bangkok

130

Quality standards for customers

This section contains a survey of Philips Semiconductors qualitystandards which are made available for our customers.

Title Related Quality 12 NC numberStandard

Philips Semiconductors Quality Manual SNMSQ001 9397 750 05298

Product Release Procedure SNWSQ002 9397 750 00188

Supplier Quality System, Part AIntroduction and Requirements SNWSQ003A 9397 750 02738

Supplier Quality System Part BChecklist and Scoring System SNWSQ003B 9397 750 02739

General quality specification for Discrete Semiconductors,Small Signal Transistors and Diodes and Sensors SNWEQ611A 9397 750 00314

General Quality Specification for Discrete Semiconductors,RF Power Transistors, Hybrid Modules and Broad BandTransistors in Metal-Ceramic packages SNWEQ611C 9397 750 01492

General Quality specification for Integrated Circuits SNWFQ611D 9397 750 05459

Quality complaint procedure for Discrete Semiconductors SNWEQ632 9398 510 36011

Philips Semiconductors Requirements for Packing,Traceability, Labeling, Transport & Storage SNWSQ401 9397 750 00747

SNWSQ402SNWSQ404SNWSQ405SNWSQ407SNWSQ623

Packing Information for ICs and Discrete Semiconductors 9397 750 01124

Guidelines for using Failure Mode and Effectanalysis Techniques SNWSQ670 9398 510 85011

131

Quality techniques and tools

Problems prevent us from performing our dailywork as we want to. So to improve the perfor-mance of our daily activities, resulting in improvedsatisfaction of our customers, we must solve thoseproblems. The best way to solve problems is byusing the quality techniques and tools described inthis section.

PDCA cycleFor problem solving and continuous improvementthe Plan–Do–Check–Action (PDCA) cycle (Fig. 1)is used in its never-ending rotation. This cycle wasintroduced by Dr. W. Edwards Deming, and isdefined as follows: Plan: With a study of the current situation, the

facts are gathered to be used in formulating aplan for problem solving. Determine the goalsand methods for a change or a test aimed atimprovement.

Do: Once a plan has been finalized, nextcomes the job of implementation. For solvingthe problems, use the problem-solvingtechniques mentioned later on. Carefullyformulate the results and conclusions.

Check: What was learned? Check the resultsand conclusions in comparison with the plan tosee whether they have really solved theproblem. Find out whether the solution bringsthe expected improvement.

Action: Adopt the change, or abandon it, or runthrough the cycle again possibly under different

Action

Plan Do

Check

Fig. 1 The PDCA cycle.

conditions. If a result deviates from theexpectation, find and correct (or remove) thecause. When problem-solving activities havebeen successful, a final action such asmethodological standardization is taken toensure that the new methods introduced will bepractised on a continuous basis for sustainedimproved performance. After that, start againwith the Plan stage.

Problem solvingFigure 2 shows a step-by-step summary of how aproblem can be attacked. In this summary, qualitytools are shown in use. Assessment (Identify the Problem): Use the

Pareto principle (defined later in this section) tohighlight major problem areas and to select thenext problem to work on.

Define the Problem: Frequently a cause orsolution is stated as "the problem". Differencesbetween problems, causes and solutionsshould be understood to avoid skipping most ofthe problem-solving process. A problem can bedescribed by the observed facts (is) and by thecomparable facts (is not) related to What,Where, When and Magnitude. Observed factswould normally include answers to thefollowing: On what object (product, unit, etc.) isthe defect observed and what is wrong(defect)? What standard exists and what is thedeviation from standard?

132

Brainstorming

Start

Make

Good?

Process

Scrap

Yes

No

Check Sheet

1 2 3 4 5

A

B

C

D

Examine how possible causes and effects are

interrelated

Determine whereproblem lies

Dissect and Analyze

Fishbone diagram

Histogram Stratification Relationship(Scatter plot)

Time Sequence Study

Take Corrective Measures

StudyResults

Control

Paretodiagram

X

R

Control Chart ParetoGraph

Flowcharting

Fig. 2 Constant quality improvement using statistical quality control.

133

Ford Motor Company. It involved 8 steps (1 to 8),which are also used as the sequence for the 8-Dreport (step 0 was added later). The steps are:

Step 0: Prepare for the 8-D process.In response to a system, evaluate the need forthe 8-D process. If necessary, provide anEmergency Response Action to protect thecustomer and initiate the 8-D process.8-D application criteria are: the symptoms have been defined and

quantified the 8-D customer(s) who experienced the

symptom(s) and the affected parties (whenappropriate), have been identified

measurements taken to quantify thesymptom(s) demonstrate that a performancegap exists and/or priority (severity, urgency,growth) of the symptom warrants initiation ofthe process

the cause is unknown management is committed to dedicate

necessary resources to fix the problem at theroot cause level and to prevent recurrence

symptom complexity exceeds the ability ofone person to resolve.

Step 1: Establish teamEstablish a small group of people with theprocess and/or product knowledge, allocatedtime, authority and skill in the required technicaldisciplines to solve the problem and implementcorrective actions. The group must have adesignated Champion and Team Leader. Thegroup begins the team building process.

Step 2: Describe the problemDescribe the internal/external customerproblem by identifying what is wrong withwhat and detail the problem in quantifiableterms.

Where is the object with the defect and whereon the object does the defect appear? Whenwas the defect first observed (calendar time)?;when in the life-cycle of the object did thedefect occur and in what pattern? How much ofthe object is defective and how manyunits/objects are defective? What is the trend?

Consideration: Brainstorm possible causes ofthe problem and list ideas on a fishbonediagram (Fig. 5). Are any of the ideas related?Flowcharting (workflow, process flow, etc.) priorto brainstorming may help participant-understanding and point to contributing causes.

Investigation: Determine the most likelycause. Make use of check sheets (Fig. 3) tocollect data if time allows. Otherwise use othermethods such as voting to decide on the mostlikely cause.

Analysis: Decide on the most appropriateaction to reduce and (if possible) eliminate theproblem.

Initiation: Put the action into operation. Collectdata as required. Refer to higher managementif necessary.

Verification: Verify the result. Plot charts andgraphs to highlight features.

Implementation: If the solution is a success,then the appropriate action should beincorporated into the quality improvementprogramme.

Control and improvement: Check thatprocedures are being adhered to and that thenew level of performance is being maintained.Aim for further improvement.

8-D method for team-orientedproblem solvingAfter a general description of problem solving, aspecific method should be used for analyzingcustomer quality problems and reporting theresults of the analysis. The 8-D method or TOPS(team-oriented problem solving) was developed by

134

Step 3: Develop Interim Containment Action(ICA)Define, verify and implement the InterimContainment Action (ICA) to isolate effects ofthe problem from any internal/external customeruntil Permanent Corrective Actions (PCAs) areimplemented. Validate the effectiveness of thecontainment actions.

Step 4: Define and verify Root Cause andEscape PointIsolate and verify the Root Cause by testingeach possible cause against the problemdescription and test data. Also isolate and verifythe place in the process where the effect of theRoot Cause should have been detected andcontained (Escape Point).

Step 5: Choose and verify PermanentCorrective Actions (PCAs) for Root Causeand Escape PointSelect the best Permanent Corrective Action toremove the Root Cause. Also select the bestPermanent Corrective Action to eliminateEscape. Verify that both decisions will besuccessful when implemented without causingundesirable effects.

Step 6: Implement and validate PermanentCorrective Actions (PCAs)Plan and implement selected PermanentCorrective Actions. Remove the InterimContainment Action. Monitor the long-termresults.

Step 7: Prevent recurrenceModify the necessary systems includingpolicies, practices and procedures, to preventrecurrence of this problem and similar ones.Make recommendations for systemicimprovements, as necessary.

Step 8: Recognise team and individualcontributionsComplete the team experience, sincerelyrecognise both team and individualcontributions, and celebrate.

Fact gatheringAll improvement starts with knowing the factsabout a problem. The effects of improvement canonly be demonstrated by facts. Bits of information,which together make up facts, are called data. Sofact gathering can also be called data gathering.Accurate data is essential to data-based decisionmaking. Measurability is important in collectingdata. The more you use measurable data, thebetter your decision will be. Three types of dataexist: Counted Data: These are noted as being

present or absent and are generally answers to"how many" or "how often".

Measured Data (often called measuredvariables): These are answers to questions like"how long", "what volume", "how much time"."how far", etc.

Location Data: These answer the simplequestion "where?".

Before starting the data gathering, there should bea clear understanding (a plan) of which data isneeded, and also the Why, When, Where, Whoand How of the fact-gathering must be clear. Fact-gathering can be done by surveys, interviews,statistical tabulations and checksheets .Checksheets are the most common technique.

Checksheets (Fig. 3)Checksheets are one of the most effective andfrequently used techniques for the collection ofdata. They provide a systematic method forcollecting data which then serves as the basis foranalyzing a problem, displaying the data ingraphical form, and for presentation of a solution.

They are also a means by which, when it isneeded, more than one person can collect thesame data in the same way. There are three kindsof checksheets used to record counted,measured, and location data.

A problem location or defect location checksheetis a picture, illustration or map on which data iscollected. Recording data in this way oftensimplifies the collection process and also helps usbetter see the problem. An example of a defectlocation checksheet is a picture of asemiconductor wafer with defective die sites notedor an illustration of accident locations which canhelp employees to analyze accident causes tomake an area safer. The simplest checksheetsare for counted data. In this type of checksheet(Fig. 3) data is collected by making marks for eachoccurrence, usually within predefined timeperiodsand then converting into a meaningful table.

BrainstormingUsing a group of people to generate as manyideas as possible is called brainstorming. Thisworks best with a group of 6 to 12 persons. Thetopic for brainstorming must be clear and wellunderstood by everybody taking part in thebrainstorming session. Guidelines forbrainstorming are:

Set an appropriate meeting place. Selectinga venue that is comfortable, casual, and theright size will greatly enhance a brainstormingsession.

Generate a large number of ideas. Don'tinhibit yourself or others, just let the ideas flowout. Say whatever comes into your mind andencourage others to do the same. Theimportant thing is quantity of ideas.

Encourage free-wheeling. Even though anidea may appear to be half-baked or silly, it hasvalue. It may provoke thoughts from othermembers. Sometimes, making a sillysuggestion can spur another idea you didn'tknow you had.

Don't criticize. This is the most importantguideline. There will be ample time later to siftthrough the ideas to select the good ones.

Fig. 3 Typical 'counted data' checksheet.

From:

Type of fault

A

B

C

D

Total no. of faults

No. produced

% faulty

Week 1 Week 2 Week 3

To:

Type of fault

A

B

C

D

Total no. of faults

No. produced

% faulty

Week 1

17

12

3

26

58

1532

3.79

Week 2

14

13

8

22

57

1511

3.77

Week 3

22

12

5

26

65

1634

3.98

135

136

creative. Let it do its work by giving it time.Don't stop your brainstorming sessions toosoon; let some time go by to allow those ideasto develop by themselves.

Pareto analysisHaving obtained the data of a problem, you canthen use the Pareto principle to decide how bestto use your resources, or how to concentrate onthe most important facts. The Pareto principlederives its name from Vilfredo Pareto, a 19thcentury economist, who applied the concept toincome distributions. His observations led him tostate that 80% of wealth is controlled by 20% ofthe people (the "80–20" principle). The name"Pareto" and the universal applicability of theconcept are credited to Dr. Joe M. Duran, aleading American consultant who used thephilosophy "the important few and the trivialmany". The Pareto principle states that only a fewcauses are responsible for most of the defects.

During the session, you should not criticizeideas because this may inhibit other members.When you criticize the half-baked ideas, youthrow away the building blocks for the greatones.

Encourage everyone to participate.Everyone thinks and has ideas, so alloweveryone to speak up. Speaking in turn helps;solicit ideas clockwise around the group.Encourage everyone to share his or her ideas.

Record all ideas. Appoint a recorder to notedown everything suggested. The ideas shouldnot be edited: rather, they should be jotteddown just as they are mentioned. Keep apermanent record that can be read at futuremeetings. You may want to read through thelist and take an "inventory" a few times; thisprocess sometimes stimulates more ideas.

Let ideas incubate. Once you've started brainstorming, ideas will come more easily.You are freeing your subconscious mind to be

120

100

80

60

40

20

total numberof defects

defecttyping

filing

omissions

wrong telephone

n

umberdiary

routineothers

Fig. 4 Typical Pareto diagram.

137

In a Pareto diagram (Fig. 4) the total number ofdefects is entered, and the number of defects areshown for each cause. The Pareto diagram is a specific type of columngraph in which the vertical columns (the causes)are arranged in descending order from left to rightto picture the frequency with which relatedcategories occur. The one exception to thedescending order is the 'others' category, acollection of very minor categories which,regardless of size, always appears on the far rightof the diagram. The Pareto diagram is primarilyused to distinguish the vital few categories(causes) from the trivial many to aid in settingpriorities by choosing those causes forimprovement which can give the highestimprovement results.

Fishbone diagram (Fig. 5)After using Pareto to select a problem to improve,the next step is to find the causes of that problemusing cause-and-effect analysis, which is astructured analysis used to separate and define

causes. The effects are the symptoms which letus know that we have a problem. A fishbone (alsocalled cause-and-effect or Ishikawa: named afterProfessor Kaoru Ishikawa) diagram can beconstructed for the analysis (Fig. 5). Theconstruction is a four-step process: First, the problem or "effect" is named and

placed in a box on the right, and a long processarrow is drawn pointing to the box.

Second, the major categories of causes aredecided. These major categories are placedparallel to and some distance from the mainprocess arrow. The boxes are then connectedby arrows slanting toward the main arrow.These branches (zones) represent groups ofpossible causes. With technical problems(shown in Fig. 5) these branches could be:Methods, Manpower, Material and Machines.For each branch, sub-branches are drawnwhich give possible causes of the problem. Forthe branch "Manpower", for example, the sub-branches could be training, motivation,workmanship, supervision, etc.

EFFECT

MATERIALS MANPOWER

Workmanship

Training

MACHINES METHODS

Supervision

Motivation

Fig. 5 Typical fishbone diagram.

138

Histogram (Fig. 6)If data is tabulated and arranged according tosize, the result is called a frequency distribution.The frequency distribution will indicate where mostof the data is grouped and will show how muchvariation there is. A histogram (Fig. 6) is a columngraph depicting the frequency distribution of datacollected on a given variable. To construct ahistogram, you need data, the more data youhave, the more accurate your histogram will be.A minimum acceptable amount of data is from 30to 50 measurements. The different values aregrouped in classes, by setting class boundaries.Often the range (largest measurement minus thesmallest measurement) of data divided by 10 isused to obtain the width of the intervals (classwidth) to be plotted on the horizontal axis of thehistogram, each interval being one column wide.The more data you have, the larger the numberyou should divide by to determine the interval (e.g.if over 250, divide by 20 instead of 10).

15

10

5

frequency

0 5 10 15 20 25 30 35 40 45 50 55 60 65reading

Third, the completion of the diagram is done bybrainstorming for causes. The causes arewritten on the chart, clustered around the majorcategory or subdivision which they influence.These minor causes are connected by arrowspointing to the main process arrow. The causesshould be divided and subdivided to show, asaccurately as possible, how they interact.

Fourth, the most likely causes are circled. Thisis usually done after all possible ideas havebeen posted on the diagram. Only then is eachidea critically evaluated. The most likely onesare circled for special attention (see 'Motivation'circled under 'Manpower').

Separate diagrams may be needed if the definedproblem is not specific enough, so causing somemajor categories of the diagram to be overloaded.This indicates the need for additional diagrams.The diagramming exercise will give a betterunderstanding of the real causes of the problemand can be a basis for further measurements orcorrective action.

Fig. 6 Typical histogram.

139

Fig. 7 The control chart, a running plot of average values from regular sample measurement, is fundamental toStatistical Process Control.

upper control limit

lower control limit

target value

observed valuesaveragevalue

x

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15number of samples

The number of values in each class is shown inthe histogram, where each is represented by abar. The result shows a representation of theoccurring values of the variable. The shape orcurve formed by the tops of the columns has aspecial meaning. This curve can be associatedwith statistical distributions that in turn can beanalyzed using mathematical tools. The variousshapes which can occur are given names such asnormal, bimodal (or multi-peaked) or skewed. Aspecial significance can sometimes be attached tothe causes of these shapes. A normal distributioncauses the distribution to have a "bell" shape andis often referred to as a "bell-shaped curve".Histograms enable us to do three things: Spot abnormalities in a product or process.

Absence of a normal distribution is anindication of some abnormality in the variablebeing measured.

Compare actual measurements withrequired standards. These standards can beindicated by dotted vertical lines imposed overthe histogram.

Identify sources of variation. The presenceof more than one source of variation in thepopulation of the histogram may produce amulti-peaked curve.

Control chart (Fig. 7)

The control chart was invented in 1924 by Dr.Walter A. Shewart. Control charts are discussedbriefly in the Statistical Process Control section ofthis handbook. Control charts (Fig. 7) are tools tobe used to achieve stable processes bymonitoring the variability of significant processparameters. At stated intervals some values of arepeated measurement are averaged. Theaverage value is plotted on the control chart,which indicates control limits. Attribute charts (P,% defective; nP, number of defectives; U, defectsper unit; C, number of defects) are used whenvariability data is not available or is difficult toobtain. These charts may be used for monitoring aprocess, however, they are more of aprocess/product appraisal tool than a processcontrol tool. The purpose of control charts forvariables is to compare the process behaviouragainst its inherent variability as determined bycontrol limits. It is a tool for deciding when toadjust a process and when to leave it alone. Itseffectiveness is enhanced by careful selection ofthe sampling method parameters (the followingpoints apply mainly to Shewart Control charts): When the sample is chosen so that variation

within the sample is significantly less thanvariation between samples, then the controllimits for the average value (x) will beextremely narrow compared to individual

sample averages, resulting in too many out-of-control conditions.

Another problem may occur when non-homogeneous groups are included in a sample(stream effect). In this case, the control limitsfor the average value (x) will be too widecompared to the individual sample averages,resulting in the control chart not being sensitiveto process changes.

Sample sizes for x and R are usually between3 and 6. A sample size of 4 or 5 is usuallyselected as a compromise between the amountof information gained and the cost of obtainingthe information.

Sample sizes for x and S are usually between7 and 10. A sample size of 7 or 8 is usuallyselected to obtain reasonably sensitive controllimits. A standard deviation chart shouldreplace the range chart any time the samplesize exceeds 10.

For calculation of centreline and control limitsfor R or S and x, data is collected preferably inchronological order for 25 or more samples.For most control charts, the control limits arecalculated on a basis of the average plus/minus3 times the standard deviation of the statisticused.

It is customary to place the charts for x and R oneabove the other so that average and range for anyone sample are in the same vertical line. Alwaysinterpret the R or S part of the chart first and whenthey are in control, then interpret the x part of thechart. When the R and S points are out of control,the control limits for x chart are not reliable. 'Out ofControl' can be determined by observing thefollowing rules: One or more plotted points outside the control

limits. If a sample average falls outside thelimit lines, it is evidence that a general changeaffecting all pieces has occurred betweensamples. If a sample range falls outside limits,

it is evidence that the uniformity of the processhas changed.

A trend of five or more consecutive plottedpoints that are all increasing or all decreasing.

A run of five or more consecutive plotted pointsthat are all above or all below the centre line.

Any other non-random pattern. Note thatJuran/Shewart have a few other indicators ofnon-randomness (2 out of 3 successive pointsat 2 standard deviations or beyond from centreline; 4 out of 5 successive points at 1 standarddeviation or beyond from centre line; not onlyone point out of 10 or nine out of 10 within oneSigma of the centre line) but these definitionsare difficult for a typical operator to apply.

If any of these occurs, this is an indication that theprocess has changed and action must be taken.

Scatter diagram (Fig. 8)A scatter diagram (Fig. 8) can be used to test therelationship between two variables (X and Y). Inthe case of defects and causes, the Y-axis couldbe the number of defects, and the X-axis the valueof a variable which could cause the defects (e.g. atemperature). If the plotted measurements show aline, there is correlation. Without correlation thepoints will be without a pattern.Numerous problems encountered in quality controlrequire the estimation of relationships betweentwo or more variables. Often interest centres onfinding an equation relating one particular variableto another set of one or more variables. 'Leastsquares' is a statistical technique for estimatingthe parameters of an equation relating a particularvariable to a set of variables. Some authors referto this as least squares or curve fitting, whereasmany practitioners refer to it as regressionanalysis and call the resulting equation aregression equation. If the scatter diagram isplotted on linear-linear graph paper and a straightline results, the line would be represented by theequation Y = AX + C where A is the slope and C is

140

the Y intercept. A relationship may be non-linear.If a scatter plot is done on other forms ofprobability graph paper (e.g. log-normal, Weibull)and the plotted measurements form a line, theequation is represented by the mathematical

model upon which the probability graph paper isconstructed. Experience has shown that mostcontinuous characteristics follow one of severalcommon probability distributions, i.e. the "normal",the "exponential" and the "Weibull".

Fig. 8 Typical scatter diagram.

8.25 – 8.49

8.00 – 8.24

7.75 – 7.99

7.50 – 7.74

7.25 – 7.49

7.00 – 7.24

6.75 – 6.99

6.50 – 6.74

6.25 – 6.49

6.00 – 6.24

Frequency

X

0.5

– 0

.74

0.75

– 0

.99

1.00

– 1

.24

1.25

– 1

.49

1.50

– 1

.74

1.75

– 1

.99

2.00

– 2

.24

2.25

– 2

.49

2.50

– 2

.74

2.75

– 2

.99

Freq

uenc

y

Y

6

5

6

9

13

12

7

7

2

3

701 3 5 10 13 11 11 6 8 2

Y

X

141

142

Paynter chart (Fig. 9)A Paynter chart is a tool to track in time whethercorrective actions have proved effective in solvinga problem, usually in a manufacturing process.It is used in combination with the 8-D method toshow which corrective actions have been takenand when.Figure 9 shows a typical Paynter chart for adiscrete semiconductor. Here, in August 1994,a batch of discrete semiconductors displayed 20failures due to deformed leads. Investigationsshowed that the reel unit was creating leadinterruption, and causing lead deformation.A corrective action (A1) was taken to install a newreel unit, and new batches were tested monthly.In November 1994, two further failures deformed-lead failures occurred, and a further correctiveaction (A3) was taken to modify the tape carrier.No further failures have occurred (up to July1995), but the monthly tests can continue until it isfelt that further testing in unnecessary.The same chart shows that a batch in September1994 displayed 14 failures due to cracked body.A corrective action (A2) was taken to modify thehandler to put less stress on the body.Subsequent monthly testing has shown no furtherfailures.

Deformed lead

Cracked body

Corrective actions:A1: Install new reel unit to prevent lead interruptionA2: Modify handler to put less stress on bodyA3: Modify tape carrier to improve lead feed.

9408

20A1

0

9409

0

14

9410

0

0A2

9411

2

0

9412

0A3

0

9501

0

0

9502

0

0

9503

0

0

9504

0

0

9505

0

0

9506

0

0

9507

0

0

Fig. 9 Typical Paynter Chart.

143

Spider graph (Radar chart) (Fig. 10)The Spider graph derives its name from its shape,which resembles a spiders web. It provides aninstant visual indication of the rating by one keycustomer of our performance criteria, comparedwith the customers best in class supplier.Figure 10 shows our supplier performance ratingfor one key customer over a 9-month period (3quarters) based on five performance criteria service, product quality, cost of use, product rangeand overall performance.In the first quarter the key customer rated us at60%, 91%, 30%, 50% and 61% respectively in thefive performance criteria categories comparedwith their best in class supplier ratings of 100%,100%, 87%, 100% and 86% respectively. Thiswas considered unacceptable and a QualityImprovement Programme was initiated.The results of this programme are shown by themuch improved customer ratings in the Spidergraphs for quarters 2 and 3.The aim of Philips Semiconductors is to obtainSpider graphs from each strategic/key customerand improve our performance to the superiorlevel, so becoming each customers first choice asa semiconductor supplier.

1st quarter UNACCEPTABLE

Fig. 10 Typical Spider graph showing the supplier performancerating of Philips Semiconductors for one key customer for threesuccessive quarters.

20 40 60 80 100

86

71

70

91

87

50

75

Service100

Product quality 100

Product range100

Overall performance

Cost of use

2nd quarter ACCEPTABLE

20 40 60 80 100

89

73

76

8080

60

80

Service88

Product quality 94

Product range100

Overall performance

Cost of use

3rd quarter SUPERIOR

20 40 60 80 100

89

61

92

77

30

50

Service96

Product quality 94

Product range100

Overall performance

Cost of use

144

Quality testing

The Total Quality Management (TQM) system ofPhilips Semiconductors ensures that quality isbuilt-in during the design, development andmanufacture of semiconductors. In the QualityAssurance part of the TQM system, quality testingverifies continuously product conformance to thespecifications, and product reliability.

The conformance test programmes for ourproducts are: Acceptance tests

These acceptance tests on finished productsverify conformance to the Final DeviceSpecification. The test results are used forquality feedback and corrective actions. Theinspection and test requirements are detailedin the General Quality Specifications.

Monitoring testsThese measure and monitor the conformanceof final products to the required level ofreliability. Their purpose is to identify reliabilityperformance trends and to collect data offailure rates and failure modes.

Qualification testsThese reliability tests assess new or modifiedproducts, or manufacturing processes.

Tables 1 to 3 give an example of the tests used forsmall-signal transistors and diodes. For furtherdetails, refer to the relevant quality standards:SNWEQ/611A, SNWEQ/611B,SNWEQ611C and SNWFQ611D,see Quality standards.

Examination or test

Sub-group Description Normal Reduced

Requirements

Lot-size

A1

A2a

A2b

A3

A4

A5

Visual/Mechanical Inoperative

Electrical Inoperative

Electrical Primary DC

Electrical Other DC

Electrical AC

Visual inspection

1K 10K10K 35K35K150K150K500K

1K 10K10K 35K35K150K150K500K

1K 10K10K 35K35K150K150K500K

0/2000/3150/5000/800

0/320/500/800/80

0/800/1250/2000/315

0/800/1250/2000/315

0/130/200/320/32

0/320/500/800/125

Table 1 Acceptance tests

145

Table 2 Monitoring tests

Examination or test

Sub-group Description

Inspectionrequirements

n c

C1C2aC2bC2cC3C4C5C6C7C8C9C15

Dimensions Characteristic inspectionComplementary characteristicsVerification of maximum ratings (where appropriate)Robustness of terminations (other than B3)Soldering heat & solderability with and without ageingTemperature cycling, 200 cyclesMechanical treatment (shock and/or acceleration and/or vibration)Reverse bias tropical at 85 °C/85% RH, 1000 h, with biasEndurance at maximum ratings, performed per test, 1000 hStorage at high temperature, 1000 hAutoclave, 96 h, 121 °C, 100% RH, no bias

222222222222454545454545

00000000000

0

Examination or test

Sub-group Description n c

Inspectionrequirements

D2D5D8D9D10D11D12D13D14D15_

Electrical characteristics inspectionTemperature cycling, 1000 cycles air-to-airEndurance at maximum ratings performed per test >1000 hStorage at high temperatures > 1000 hStorage at low temperatures > 1000 hHAST test (unsaturated) 133°C, 85% RH, 96 h with biasThermal shock, liquid to liquid, 100 cyclesPassive flammabilityESD investigationAutoclave, 144h, 121°C, 100% RH, no biasAdhesion strength test SMD

4577777777777745307745

0111111011

Table 3 Qualification tests

146

Release of new products

ATS

Type study stage

ATD

Development stage

CQS

Qualification stage

RFS

Production stage

DOD

Discontinuation stage

WIT

For product development and product release of semiconductors, all Product Groups use the same basicflow chart (Fig. 1).

Fig. 1 Product life cycle flow chart.

Prod

uctio

n pe

riod

Deve

lopm

ent p

erio

d

147

The product life cycle flow chart (Fig. 1) containsfive milestones which are described below.

Acceptance for Type Study (ATS)The ATS milestone is the point where a feasibilitystudy is started to determine whether thedevelopment of a particular product has technicaland commercial feasibility.

Acceptance for Type Development(ATD)The ATD milestone is the point where thefeasibility study is concluded and assessed. If theresult has been successful and provisions havebeen made for financial allocations, it becomesthe starting point of type development. Typedevelopment can also include a Pilot Productionstage.The ATD can also be the starting point of thedevelopment period, in cases where a feasibilitystudy is not necessary.

Availability of CustomerQualification Samples (CQS)The objective of this milestone is to enable thesupply of representative samples of a new type tocustomers for their qualification.The CQS milestone is reached when the finaldesign is achieved and there is reasonableconfidence that products meet the quality andreliability requirements.Customer Qualification Samples must bemanufactured on the production lines which will beused for the full production.Commercial delivery of products before RFS isonly possible when all relevant conditions are met.These have to be further detailed and controlledby local procedures.

Release For Supply (RFS)The RFS milestone is the official and formalrelease of the product as an irreversiblecommitment to the market. It is the point in the lifecycle where the device specification is frozen andproduct responsibility is handed over from thedevelopment to the manufacturing department.

Discontinuation Of Delivery (DOD)The DOD milestone is the point where thedecision is taken to withdraw the type from themarket. Customers receive notification.

Withdrawn (WIT)The type is withdrawn from the market.

During the development stage (before RFS)experiments and tests are carried out to prove: conformance to specifications reproducibility reliability.During this stage samples can be delivered tocustomers.

At RFS, the development is complete, andcommercial and technical data is available.

After RFS the product is commercially availableand will be included in a data handbook andthe catalogue. Eventual changes must followthe international change procedure, includingthe notification of customers.

The product release process is covered in qualitystandard SNW-SQ-002.

Reliability is the ability of an item to perform arequired function under stated conditions for astated period of time (IEC 271 definition).Thus, reliability is a measure of the qualityremaining after some time, exposed to particularoperating stresses.

Reliability and failure rateLike other measures of quality, reliability is aprobability: the probability of a componentsurviving for a given time. For semiconductors,reliability is generally quoted in terms of failurerate: determined by the number of failuresobserved in a sample of semiconductors operatedat a stated stress level for a fixed time. Reliabilitycan be calculated from failure rate :

R(t) = exp(-λt)

where λ is the failure rate, which is assumed to beconstant.

Fig. 1. The bathtub curve of failure rate with time ischaracterized by the initial early failure period,intervening constant-failure rate period, and a finalwearout or end-of-life period.

Failure rate is not usually constant, however, butvaries with time in the way typified by the familiarbathtub curve, Fig. 1. This translates into acorresponding reliability curve, Fig. 2.

Fig. 2. The curve for reliability versus time, derived fromthe bathtub curve for failure rate, exhibits the same threeregions.

Real improvements in the quality of oursemiconductors in recent years has resulted inbetter reliability, especially during the early-failureperiod, so that their reliability bathtub curve is nowless pronounced at the beginning, as shown inFig. 3.

Fig. 3. Improved semiconductor conformity resultingfrom our quality-improvement activities has resulted inimproved reliability, especially during the early-failureperiod, and a flatter bathtub curve. Reliability is higheruntil wearout sets in.

148

Reliability

log t

1

R(t)

0.5

0

log t

1

R(t)

0.5

0

λ

λ

R

Constant failurerate

Early failure period

Wearout

λ

log t

149

MTBF, MTTF and call rateReliability in finished equipment depends on thereliability of the variety of components that itcomprises, under the operating conditionsdetermined by the design of the equipment as awhole, and the environment in which it operates.Two measures of equipment reliability arecommon: Mean Time Between Failures, MTBF,(used where equipment is repaired as it fails) andMean Time To Failure, MTTF, (used where repairis not carried out). Both are defined as the ratio ofthe cumulative observed operating time to thenumber of failures. Thus, MTTF is 1/λ.Call rate, a common indicator of the reliability ofconsumer equipment, is the number of servicecalls required during the guarantee period perhundred equipments sold. Since semiconductorsoperate mainly in the early-failure period (if therehas been no burn-in) during the guarantee period,the improvements in our conformity and early-failure rate greatly reduce guarantee costs.

Observed and assessed reliabilityFailure rate, reliability, MTBF and MTTF areclassed as either observed or assessed. Theobserved value is that calculated from therecorded time and observed failures. Assessedvalues are those corrected to a given confidencelevel.

150

Return shipments

PD quality standard SNWSQ636 'Procedure forreturn shipments' categorizes return shipmentsinto three groups:

Commercial returns Discrepancy returns Technical returns.

Each of these groups involves differentresponsibilities and goods-flow procedures asshown in Figs 1, 2 and 3 below, which use thefollowing key:

= Return Number issue

= Goods flow

= Information

= Regional Sales Office RSO

Fig. 1 Commercial returns.

CustomerLocal point

RSOstore

Internalcustomers

IPMM/Logistics

Commercial returns (Fig. 1)These are goods returned from the customer because his stock level is too high.

151

Fig. 2 Discrepancy returns.

Local point

RSOstore

Internalcustomers

Fig. 3 Technical returns.

Samples

CustomerLocal point

RSOstore

Factory

IPMM/Logistics

Technical returns (Fig. 3)These are goods which do not meet specification or are over the maximum agreed age. For goods whichdo not meet specification, the samples sent to the factory will be analyzed. If the analysis justifies thecomplaint, corrective action will be taken.

Internalcustomers

Discrepancy returns (Fig. 2)These are goods not ordered, or with wrong 12 NC number, or of wrong quantity, or with the wrong label.

Customer

152

Sampling on the Fly

Many of the techniques and procedures used tomonitor and protect product quality today havehardly changed since Quality Control was firstintroduced over 50 years ago. Electroniccomponent quality levels have changed, though:by around a thousand times. Process averages of1% or more that used to be the industry standardhave been replaced by reject levels around 10 to100 ppm. Quality procedures that were designed

for the old 1% levels can present a significanthazard when applied to modern production. Thisis especially true for semiconductors. WithSampling on the Fly, we ensure not only that noneof the valuable data needed for the initiation ofcorrective action is lost, but also that the very lowreject levels currently achieved in production arenot put at risk during inspection.

Fig. 1. Compared with the traditional system (a) where batches are sampled manually following final electrical tests,Sampling on the Fly (b) reduces manual handling with associated ESD hazard and the risk of mixing good and rejectsemiconductors.

(b)

SOTF

Pass

Pass

Reject

Pass

Fail

Sample

Finalelectrical test

Electricalacceptance test

Batchinformation

Controlprogram

AssemblyPass

Fail

Reject

(a)

Manualsampling

Finalelectrical testAssembly Electrical

acceptance test

153

SAMPLING ON THE FLYApplying modern data processing and automatichandling techniques to Quality-ControlAcceptance testing can result in bothimprovements in the validity of the procedure anda reduction in the hazards associated withadditional handling. This is the basis of ourdevelopment of Sampling on the Fly* (SOTF), apowerful extension of conventional samplinginspection.SOTF allows the Group A electrical tests to beintegrated with the 100% final electricalinspection. The procedures are compared inFig. 1, where the reduction in quality-hazardingmanual handling due to SOTF is evident. A furtherbenefit is the rigorously-stochastic software-controlled sampling that is an integral feature ofthe SOTF system. Finally, data recorded from thesampling inspection is available for StatisticalProcess Control.

SOTF test systemTo accommodate SOTF procedures, theautomatic test equipment used for the 100%electrical test is augmented by an active terminal.This runs the sampling software and integratesthe electrical Acceptance Test sequence with thefinal electrical test.

Test initiationAt the commencement of the final electrical testrun on each batch the following data is enteredinto the terminal: device type die lot number assembly batch card number gross batch quantity week code number sampling status (normal, tightened or reduced) test station number budget yield.

From this data the SOTF program is selectedautomatically; it estimates the quantity of gooddevices after the test, and initiates the samplingroutine. The yield revealed by the test at 10%,50% and 90% of the batch is checked and the sampling interval simultaneously adjusted toensure that the required sample is distributed overthe full quantity of good devices.

AdvantagesBesides the primary benefits of SOTF: even distribution of samples over all good

devices in a batch reduced h andling and so reduced hazard;

we have taken the opportunity to collect furtherdata for use in our quality-improvementprogramme.After each SOTF run, the terminal generates asummary sheet detailing the attributes (yields,defect rate) of both the 100% and samplingelectrical test and the electrical variables in thesample. This shows the distribution of staticparameters over the batch. Full device traceabilityenables this data to be used for the continualprocess improvement required for progresstowards zero defects.

154

Self-qualification

Self-qualification is a service performed by PhilipsSemiconductors to provide customers withinformation concerning the qualification of majorchanges to any process, material or product. Thepurpose is to provide formally-documented,detailed information in advance of the qualificationof a product or process, and to conduct athorough and well-documented programme withresults that will help secure customer approval.This activity is intended to minimize the impact oncustomers' resources, and to expedite thechange-approval process so that timely and well-coordinated implementation can be accomplished.Customers participating in self-qualification aregiven an opportunity to check the proposedqualification beforehand against their ownstandards and application-specific conditions.Customers may audit the project, and are giventhe option of modifying or enhancing thequalification plan according to their needs.

Self-qualification proposalThe self-qualification proposal informs customersof major process or product changes bydescribing those changes in detail, and byproposing a stress plan to be used in allqualification activities. An interim report is issuedwhen there is a need to modify or change anyinformation prior to final report approval, or whena customer specifically requests immediateinformation prior to the publication of the finalreport. The final report describes findings relativeto the proposal, and is aimed at satisfying bothinternal and external customers' reliabilityrequirements. In some cases the final report maybe published as a stand-alone document, havingno proposal prior to its writing.

Programme reviewA Quality Review Board carefully monitors theprogramme to ensure that all requirements are met.

Each proposal is published at least 30 calendardays prior to initiation of qualification activities.The final report is released within 30 calendardays after completion of the qualification project(when the last test group has completed itsappropriate stress duration), and at least 30 daysprior to the expected date of implementation.The Quality Review Board assesses each self-qualification activity for conformance and accuracyin the following areas: Qualification schedule in proposal and interim

reports (optional) List of affected products Process/product attributes Benefits derived from the process/product

change Other reliability considerations, adding to the

report's credibility Product selection for stress (what and why) Plan modifications (if appropriate) Definition of stresses and stress conditions Qualification samples (including sample sizes) Sample description and history Burn-in schematics with drawings Preliminary stress results (where applicable) Detailed stress results (presented in final

report) Schematics (to accompany text in

"Qualification Plan" section). These may eitherbe included in the report or offered "uponrequest"

Individual qualification summary pages(Reliability Engineering Project Summarydocuments)

Failure Analysis report Applicable Structural Similarity rules and

results.

155

Semiconductor Assembly Council (SAC)

SAC is a worldwide organization of semiconductorassembly subcontractors, suppliers and end-users, established to certify subcontract qualitysystems and control practices. It represents about95% of the non-Japanese semiconductor industry,and has some Japanese companies among itsmembers as well.The main function of SAC is to providesubcontractor quality system certification. It alsopromotes teamwork with suppliers and customers,and provides a forum for discussion of non-proprietary technology issues and best practices,giving an opportunity to benchmark in the industry.SAC certification is carried out by an audit team ofsemiconductor specialists coming from within theindustry. The audit team represent at least two butusually three semiconductor companies. The SACAudit Committee continually reviews data fromSAC-certified subcontractors in order to monitortheir continued compliance to the certification.SAC procedures also include a de-certificationprocedure in case of any serious lapse in qualitystandard by a certified subcontractor.

SAC in Philips SemiconductorsPhilips Semiconductors requires that all itsassembly subcontractors are SAC certified or areplanning to achieve SAC certification. This is asignificant part of our Supplier Rating System forcontract assembly. SAC certification is a means ofdeveloping the quality system of oursubcontractors. We are required to do this by theQS-9000 Quality System Requirements that wehave implemented in all Philips Semiconductorfacilities.As a SAC member, Philips Semiconductors hasan influence on the audit standard, and alsoreceives the audit reports done by SAC on eachassembly subcontractor. Its therefore notnecessary for Philips Semiconductors to carry outits own quality system audit on subcontractors,thereby saving time and money.

156

Six Sigma

The Six Sigma definition used by PhilipsSemiconductors originated within Motorola. InJanuary 1987, Motorola set as one of its corporategoals to achieve Six Sigma capability within 5years (i.e. by 1992). When Motorola Inc. won theUSA Malcolm Baldrige National Quality Award in1988, other companies began to benchmarkthemselves against Motorola. As a result, theMotorola concept of Six Sigma began to beadopted by others (e.g. IBM, Ford). The Motorolagoal applies to all areas of the business, not justproduct and process quality, and is orientedtoward approaching the standard of zero defects.The strategy includes the six steps to Six Sigmafor manufacturing and another six steps to SixSigma for non-manufacturing.

The 1991 Philips SPC ConferenceSix Sigma Definition"A total business culture involving the dynamicprocess of identifying key product/servicecharacteristics (as defined by both internal andexternal customers) and controlling elementsfollowed by the determination of process capabilityand the continuous reduction of variability leadingto the achievement of 3.4 ppm (Cp = 2.0, Cpk =1.5)."

Refer to the Statistical Process Control section ofthis handbook for concepts such as Sigma,Process Capability and Capability Indices Cp andCpk.

Philips Semiconductors quality improvement goalsinclude "improve process capability to achieve 6Sigma performance by the end of 1994"

Application of Six Sigma withinmanufacturingWithin manufacturing, variation of a process ismeasured in standard deviations (Sigma) from the

mean. The normal variation, defined as processwidth, is ± 3 Sigma around the mean (Fig. 1).Approximately 2700 parts per million (ppm) will falloutside the normal variation of ±3 Sigma. For aproduct to be manufactured virtually defect-free, itmust be designed to accept characteristics whichare significantly more than ± 3 Sigma away fromthe mean.The way to achieve this is to reduce variability,resulting in a smaller value of Sigma. It can beshown that a design which can accept twice thenormal variation of the process (± 6 Sigma) canbe expected to have no more than 3.4 ppm foreach characteristic (Fig. 2), even if the processmean were to shift ± 1.5 Sigma.The ± 1.5 Sigma shift allowance is the Motorolacontribution to the definition of Six Sigma. Adesign specification width of ±6 Sigma and aprocess width of ± 3 Sigma produces a Cp of 12/6= 2, as Cp is (by definition) the specification widthdivided by the process width. Cpk is the distanceof the process mean to the nearest specificationlimit divided by half the process width. In the caseof 1.5 Sigma shift, Cpk is (6 - 1.5)/3 = 1.5.

Application of Six Sigma in bothnon-manufacturing andmanufacturingThe concept of defect rate per millionopportunities for error (DPMO) can be used inboth manufacturing and non-manufacturing areas.The DPMO is an attributes measurement(compared with the variables measurements usedto calculate Cp and Cpk). An example of DPMO isthe parts per million (ppm) measurement used forelectrical and visual mechanical quality (refer tothe PPM section of this handbook). A typicalexample of calculating DPMO in a non-manufacturing area is:A product or service contains 80 opportunities for

error in 2500 units processed. If 45 defects arediscovered, the defect rate is:

DPMO = = 225.45 x 106

2500 x 80

-6σ -5σ -4σ -3σ -2σ -1σ +1σ +2σ+3σ+4σ+5σ+6σx

LowerSpec Limit

UpperSpec Limit

SPEC LIMIT ±1 sigma ±2 sigma ±3 sigma ±4 sigma ±5 sigma ±6 sigma

PERCENT 30.23 69.13 93.32 99.379 99.9767 99.99966

DEFECTIVE PPM 697,700 308,700 66,810 6,210 233 3.4

Normal distribution shifted 1.5σ

-6σ -5σ -4σ -3σ -2σ -1σ +1σ +2σ +3σ +4σ +5σ +6σx

±3σ99.73%

LowerSpec Limit

UpperSpec Limit

SPEC LIMIT ±1 sigma ±2 sigma ±3 sigma ±4 sigma ±5 sigma ±6 sigma

PERCENT 68.27 95.45 99.73 99.9937 99.999943 99.9999998

DEFECTIVE PPM 317,300 45,500 2,700 63 0.57 0.002

Normal distribution centred

Fig. 1 Typical process variations with a centred distribution between Six Sigma limits. Only 2 devices per billion fail tomeet the specification target.

Fig. 2 Effects of a 1.5 Sigma shift, where only 3.4 ppm fail to meet specification.

157

158

Milestone

Deliverable

Project management plan S S C, S C, S I, SSoftware requirementsspecification D I, SSoftware acceptance testspecification ISoftware architectural design I, CSoftware detailed design CSource code CUser documentation D ISoftware integration test report CRelease note SPerformance indicators CSoftware project evaluation CSoftware project archive S

Software Quality

Software qualityAt a time where applications and components GODIGITAL the quality of software itself and thecapability of software development centres getmore and more important. Many organizationswithin Philips Semiconductors are active insoftware design. These organizations meet eachother in the SMM (Software Managers Meeting)and the SPI (Software Process Improvement) taskforce.

Software releaseThe whole Software Development Processincluding software release is described in theOverall System Realization Process (OSRP).under pww.osrp.sc.philips.com.The table shows the milestones in the softwarerelease and the deliverables per milestone.

Plan

ning

Defin

ition

High

leve

lDe

sign

Impl

emen

tatio

n

Inte

grat

ion

& Te

st

Rele

ase

Arch

iving

Legend :D : Draft (created but not yet cross-checked or inspected)C : Cross-checked I : InspectedS : Signed (by the persons responsible)

159

Software capability (CMM)

The capability of software development centres is assessed against the Capability Maturity Model (CMM), asdescribed in the publication CMU/SEI-93-TR-24 d.d. February 1993 by the Software Engineering Institute ofthe Carnegie Mellon University.

In this model starting organizations are at CMM level 1. The highest level is 5.The levels are called 1: initial, 2: repeatable, 3: defined, 4: managed and 5: optimizing.The policy of Philips Semiconductors is for all centres to reach at least level 2 in 2001. Some softwarecentres are certified on level 3.

The requirements for the first levels are given below :

CMM level 2 CMM level 3

Software configuration management Peer reviews

Software quality assurance Intergroup communication

Software subcontractor management Software product engineering

Software project tracking and oversight Integrated software management

Software project planning Training program

Requirements management Organization process definition Organization process focus

160

Statistical Process Control (SPC)

No machine, operator, batch of parts or rawmaterial can ever be consistently precise: therewill always be variations about any specification.Thus, the properties of a product from anyproduction line will vary about the target (median)value; the further characteristic values are fromthe median value, the less often they shouldoccur, as shown in Fig. 1.

Fig. 1 A typical distribution of the values of the productcharacteristic. Ideally, the peak (median) value shouldcoincide with the target value for the product, and veryfew products should be out of limit.

Analysis of the causes of variation (e.g. operatorerrors, setting errors, faulty raw material spreads)reveals two distinct categories (Table 1): random(chance) causes and assignable causes (causesthat can be identified and corrected). If all theassignable causes of variation can be identifiedand eliminated, those that remain will be random:due to the natural limitations on accuracy, such astolerance and noise. When this condition isreached, the process is said to be under control.

Table 1Causes of product variation

random (chance) causes

many separate causes

any one cause resultsin only a small variation(but many causestogether can yieldsubstantial variations)limit limittarget

value

Frequencyof

occurrence

assignable causes

one or few separate causes

any one cause can resultin a large variation

scrap. Moreover, the most frequently-occuringproduct properties do not coincide with the targetspecification. By adjustment of conditions, productproperties are shifted to peak on the target value;by reducing equipment tolerances and rawmaterial spreads, the random variations of theproduct characteristics are progressively reduced,Fig. 2(b), the yield improves, and the costsdecrease. When process spreads have beenreduced so that virtually the whole output of aprocess is within specification limits, Fig. 2(c), thatprocess is said to be capable.

Process control systemIn its basic form, Statistical Process Control (SPC)works by treating a production process as acontrol system, and adding that essential element,the feedback loop. Analysis of product propertiesat each stage of the process reveals assignablevariations, and records the extent of the randomvariations so that corrections can be applied tobring the process under control. Data generatedby SPC is also fed forward to warn ofextraordinary spreads or deviations, so thatcorrective action can be planned and anynecessary screening instituted.The principal elements of the type of process-control feedback system using SPC that weoperate are shown in Fig. 3.1. The process itself is an interaction of people,equipment, materials, methods and environment,each of which exerts a characteristic influence onthe product.2. Measurement monitors output and variation,providing performance and capability information.3. Feedback translates measurement informationinto action to influence the process towardsimproved output.4. Feedforward informs the following stage aboutthe distribution of product properties so thatcompensating measures can be planned orscreening instituted.

Fig. 2 Analysis of the causes of variations in productcharacteristics from a process (a) can be used toprogressively improve the process (b) so that less scrap(shaded areas) is produced. A process producingnegligible scrap for economic purposes (c) is said to beunder (statistical) control.

Process capabilityOnce a process has been brought to a state ofstatistical control, it then becomes possible toengineer it so that its output meets thespecification. In Fig. 2(a), the shaded areasrepresent out-of-limit products, in other words,

target

minimumlimit

maximumlimit

freq.

freq.

freq.

(a)

(b)

(c)

161

162

Control chartsCentral to the operation of SPC, the control chartalso known as the 'Shewart control chart',compares product characteristics graphically withcalculated control limits. Its principal function isthe detection of assignable causes of variation inthe process.

Fig. 4 The control chart, a running plot of averagevalues from regular sample measurement, isfundamental to Statistical Process Control.

A typical SPC control chart is a running plot ofthe average value of measurements carried outon samples of products, as shown in Fig. 4 withcontrol limits marked. Where control limits arechosen to be + or 3 times the standard deviationσ of the characteristic being monitored, if randomcauses of variation only are present (with noextraordinary trends), 99.7% of plotted values willfall within the limits. This property can be used asthe criterion of process capability. The majority ofvariations outside 3σ control limits will be due toassignable causes, which should be investigated.Besides the average values from samples, a chartof the range (maximum less minimum value) mayalso be kept. Variations in the average value ofrange R indicate changes in the uniformity of rawmaterials, or the state of maintenance of theprocess equipment or test equipment. As withchanges in average values x, so variations in Rshould be investigated: causes of deteriorationsmust be corrected; improvements may indicate

Fig. 3 Statistical Process Control applies feedback around a process to eliminate assignable causes of variation andreduce the effect of random causes. Information generated can also be used to predict product quality for use at laterstages in production.

1. Processpeople materials equipment

methods environment

2. Measurement

Product

3. Feedbackfor correction

of processvariations

2. Informationconcerning

performance andcapability

4. Feedforwardof process quality

performance

Tofollowing

stage

Upper control limit

Lower control limit

Target value

Observed valuesAveragevalue

x

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15Sample number

beneficial changes that could be applied to otherprocesses.Since control charts are usually available to eachoperator or displayed at each work station, theyare an excellent means of involving everyonedirectly in the quality aspects of their activity.

Capability criteriaA useful way of relating the results obtained fromSPC to the quality of finished products is to relatethe specification limits (upper and lower) to thestandard deviation σ and to the deviation of theprocess mean M to the target value T.The potential process capability index (Cp)defines a process in terms of its parameter spreadwith respect to the defined limits of a specification.It is a function of two variables: the width of thespecification and the process spread, where theprocess spread is measured by 6 Sigma (± 3Sigma).

Cp = specification width

= USL LSL

process spread 6 Sigmawhere USL and LSL are the upper and lowerspecification limits.

Note that Cp measurement does not take intoconsideration where the distribution is located inrelation to the specification.

The process capability index (Cpk) measures theactual process capability by taking intoconsideration where the distribution is located inrelation to the specification. For a processparameter distribution that is normal, stable andcentred, Cp = Cpk. If the distribution is not centred,the process capability index Cpk = minimum valueof

USL Mand

M LSL

3σ 3σ

or (mathematically the same)

Cpk = USL LSL 2 M T

6σ

( M T is the absolute value of M T).

The capability index shows whether a process iscapable of producing products within specification.If the distance of the process mean M to thenearest specification limit is 3σ(USL M = 3σ), then Cpk = 1.In that case, if the distribution is assumed normal(Gaussian), the chance of products out ofspecification is Pz = 0.00135, or 1350 ppm. Thisvalue applies to one specification limit, so for asymmetrical situation (T=M), the process averagewould be 2700 ppm.

163

Fig. 5 Lower specification limit (LSL) and upper specification limit (USL) can be defined in terms of the standard deviationσ. M is the process mean, and T the target value (mid-way between USL and LSL).

LSL USLT M

,

Values of Pz are tabulated in Table 2, where:

Z = USL M

orM LSL

σ σ

With this symbol Z, the capability index can becalculated as Cpk = Zmin/3.

Current market requirements for process averageare such that most customers regard Cpk = 1.33 (Z = 4) as a minimum value for the capabilityindex, the goal being Cpk = 1.5 (Z = 4.5 / 3.4 ppm).

General notes on Cpk The major use of Cpk is to monitor

improvement. Consequently, the most validcomparisons that can be made with Cpk arethose against itself over time. Suppliercomparison is not a recommended use of Cpk.Cpk can be affected by the specification limits,measurement equipment and accuracy.

Cpk variation may be due to insufficientsampling. A minimum sample size of 5 isrecommended for calculating x and R perparameter per lot. To establish x and R controllimits, a minimum of 10 lots is recommended. ACpk calculation for a parameter of a single lotusing a small sample size (e.g. 5) is subject tovariation due to insufficient sample size. Toobtain a valid Cpk, a minimum of 100measurements (for example 20 samples of 5each) is required.

A negative Cpk would occur when the mean ofthe population is outside the specification limit.By definition in such cases Cpk = 0.

In general, calculation of a composite Cpkvalue for a product group, product, process orprocess node, is of limited value due to lack offocus on individual characteristics for futureimprovement. However this technique is usedoccasionally for reporting to management orcustomers. If a composite Cpk index is desired,it is usually represented by the minimum of allindices or a weighted average of the individualcharacteristic indices.

BenefitsEarly warning of deviations provided by SPCbefore large quantities of defective items havebeen produced reduces scrap and so helpsachieve delivery schedules. Finally, the availabilityof information in a clear and relevant form makesfor more informed commercial and managerialdecisions.Although SPC has long been in use in oursemiconductor factories, its application is nowbeing extended and intensified as part of thedevelopment of our Quality Assurance system,and the broader Quality Awareness programmespursued throughout our organization. StatisticalProcess Control is a powerful tool for monitoringmanufacturing processes, revealing variations inproduct characteristics and identifying the causes.The improved control of a process that resultsfrom continued use of SPC increases yield,reduces rework, reduces inspection effort, andlowers costs. The increased volume of informationthat SPC generates, together with the associatedinterpretation and organization of data, providesearly warning of impending quality problems, andmakes for more accurate decisions concerning thechanges required to maintain product quality.

164

Use of this table for values of Z greater than 4 should berestricted to cases where the distribution has accuratelybeen measured.

Table 2Values of Pz for selected values of Z assuming

a normal distribution

Z

6.0

5.5

5.0

4.5

4.0

3.93.83.73.63.5

3.43.33.23.13.0

2.92.82.72.62.5

Pz

10-9

2 x 10-8

3 x 10-7

3.4 x 10-6

0.00003

0.000050.000070.000110.000160.00023

0.000340.000480.000690.000970.00135

0.00190.00260.00350.00470.0062

Z

2.42.32.22.12.0

1.91.81.71.61.5

1.41.31.21.11.0

0.90.80.70.60.5

0.40.30.20.10.0

Pz

0.00820.01070.01390.01790.0228

0.02870.03590.04460.05480.0668

0.08080.09680.11510.13570.1587

0.18410.21190.24200.27430.3085

0.34460.38210.42070.46020.5000

165

Structural similarity

When a semiconductor device design is modified,a major process step is changed, or a new typehas to be qualified, structural similarity can betaken into account in deciding the amount ofcharacterization and qualification (reliabilitytesting) that needs to be carried out. Structuralsimilarity determines the extent to which testresults from a specific device or family can beconsidered representative for other similar types.When applied to reliability aspects, structuralsimilarity is indispensible in predicting thereliability performance of types similar to thosethat have been specifically subjected to reliabilitytesting. Not only is reliability testing of every typeexpensive, it's also unnecessary, since devicetype itself is less important than the commonstructural aspects and generic data of the group oftypes. For example, humidity testing explores thepackage, the lead finishing, the passivation of thedie, the metal deposition and the spread betweendiffusion batches. Humidity test results are hardlyinfluenced by individual features of a device type(with the exception of operating voltage), so oncereliability tests have been completed on one type,it's not necessary to test other structurally-similartypes in that specific family.Structural similarity is also used to determinewhich types are grouped for reliability monitoringas part of the periodic investigation ofconformance to reliability requirements. Theworst-case device (i.e. the most-complex or thelargest die size) in a structurally-similar group isusually subject to stressing. A good reliabilityperformance means a high probability of a goodperformance from other types in the same group,and obviates the need to test these other types.Structural similarity is test-dependent, the allowedgrouping of types being unique to a specific test.For example, a solderability test may only need tobe done on one sample of all the types from oneassembly line having a similar pin-count, the sameenvelope and the same lead-finish. Here,

electrical performance and the wafer-fab processare not relevant. However, other tests may useelectrical performance or the wafer-fab process asprime selection criteria for structurally-similargrouping.For semiconductors, and depending on the testconcerned, devices are generally categorized instructurally-similar groups based on wafer processor envelope family, or a combination of both.

Structural similarity groupingbased on wafer processWafer processes are subdivided in a hierarchyof three levels: 1st level group is determined by the main

technology used (e.g. CMOS, NMOS, bipolar,etc.).

2nd level group is determined by theproduction centre.

3rd level group is determined by the processvariant (e.g. a difference in design rules, metallayers or shrink versions).

Wafer processes are considered to have structuralsimililarity provided that they are classified in thesame group for all three levels.

Structural similarity groupingbased on envelope familyEnvelopes are subdivided in a hierarchy of threelevels: 1st level group is determined by the main

technology and body shape (e.g. glass bead,DIL, SO, SIL, QFP etc.).

2nd level group is determined by theproduction centre.

3rd level group is determined by the envelopeand process variant (e.g. significant differencein pin-count or pitch, lead frame material or die-attach process).

Envelopes are considered to have structuralsimilarity provided that they are classified in thesame group for all three levels.

166

167

Supplier quality system

PurposePhilips Semiconductors purchases hardware,software, processed materials and services on alarge scale, for the manufacture of its high-qualitysemiconductor products.Suppliers to Philips Semiconductors must be fullyresponsible for the quality of their products,thereby consistently ensuring compliance with thequality standards and conditions agreed upon.The suppliers' quality assurance system shouldmeet at least one of the ISO 9000-seriesstandards. For a number of purchased products,specific requirements for defect prevention andcontinuous quality improvement are applicable.The full requirements are defined in the PhilipsSemiconductors Supplier Quality System. Thissystem contains the Quality System requirements(SNWSQ003A) as well as an audit checklistand scoring system (SNWSQ003B). It is anessential tool in the selection and qualificationprocess of the suppliers. It also facilitates thedevelopment of long-term, mutually supportive,relationships that permit reduction of incominginspection, inventory and lead-times. In short:increased quality at reduced cost.The supplier Quality System is described inthe quality standards SNW-SQ-003A andSNW-SQ-003B.

Certification requirementsPhilips Semiconductors certification of suppliers'quality assurance systems is an essential steptowards becoming a fully certified supplier. Thesupplier certification requirements and proceduresdepend on the specific nature of the relevantproduction products. These products are classifiedunder two categories:

Critical production productsThose products which have a direct or indirectinfluence on the quality of the processes and/orend-products affecting them in form, fit and/orfunction.

Non-critical production productsAll other production products.

Suppliers of critical production products must be incompliance with, and certified to, the PhilipsSemiconductors Supplier Quality Systemrequirements, and in addition, to the relevantstandard of the ISO 9000-series.All other suppliers, i.e. suppliers of non-criticalproduction products, must still be in compliancewith the relevant standard of the ISO 9000-series.In both cases the certification must be granted byan accredited third-party inspectorate.

168

Thermal resistance

The thermal characteristics of semiconductors area major consideration for both manufacturers andusers because high junction temperatures canhave an adverse effect on device performanceand long-term stability.When the semiconductor dissipates power it getshot. The lower the thermal resistance, the quickerthe die can dissipate this heat via the lead-frameto the heatsink.The thermal resistance Rth of the package ismeasured in °C/W or K/W (K in kelvin) and is akey parameter in the calculation of junctiontemperature Tj, using the equation:

Tj = RthP + Ta,

where P is the dissipated power in watts and Ta isthe ambient temperature.Practical measurements of Tj are made toMIL-STD 883C, method 1021.1, using thetemperature-sensitive parameter (TSP) technique.

Package designFor good thermal performance, it's essential thatthermal resistance is kept low by good packagedesign. Some elements of package design affectthermal resistance more than others: Die size has a large effect on thermal

resistance. In general, the smaller the die size,the higher the thermal resistance.

Die attach methods and materials must becarefully selected for maximum reliability, sincethey can affect thermal resistance.

Lead-frame material, particularly for plasticpackages, has a significant effect on thermalresistance. The higher the material's thermalconductivity, the lower will be its thermalresistance, due to the heat-spreading effect ofthe lead-frame. An alloy 42 lead-frame has ahigher thermal resistance than a copper alloylead-frame. For hermetic packages, which donot use copper lead-frames, the lead-frame

material has a less-significant effect on thermalresistance because the thermal conductivity ofceramic is much higher than that of mouldedplastic.

Lead-frame design, particularly for plasticpackages, must maximize thermal dissipationwherever possible. The design is usuallydetermined by die size and pad layout.However, large pads and support structureshelp to lower thermal resistance.

Bond wires, due to their small diameter, donot provide a significant thermal path andtherefore have little effect on thermalresistance.

Package body material can affect thermalresistance. However, material selection isdetermined more by reliability andmanufacturing constraints than by thermalconsiderations.

Internal heat spreaders can help to reducethermal resistance in plastic packages byimproving the heat distribution through thepackage. They are mainly used in powerdevices for consumer applications. Typically,tests indicate that internal heat spreaders canreduce thermal resistance by:

- 35 to 40% in PDIL 64 packages- 23 to 32% in PLCC 68 packages- 9 to12% in SOL 20, 24 and 28 packages.

External heatsinks can help to reduce thermalresistance in power semiconductors. Threevarieties of heatsink are in common use: flat-plate heatsinks (including chassis), diecastfinned heatsinks and extruded finnedheatsinks.Heatsink thermal resistance is a function ofsurface finish. A painted surface will have agreater emissivity than a bright unpainted one.

169

Total Quality Excellence (TQE)

Q1

POTE

NTIA

L LO

NG T

ERM

PREF

ERRE

DLO

NG T

ERM

Supply Base Management process

TOTAL QUALITY EXCELLENCE

SuperiorSuperior Superior

Q1 Good toexcellent

Good toexcellent

Good toexcellent

Q101 Requiresupgrade

Requiresupgrade

Requiresupgrade

SHO

RT T

ERM

UNSATISFACTORY

RATINGS

PRODUCTQUALITY

ENGINEERING DELIVERY COMMERCIAL

CONT

INUO

US IM

PROV

EMEN

T IN

ALL

ASP

ECTS

OF

BUSI

NESS

TQE at the Ford motor company is the highestrecognition that full-service suppliers can achieve.It represents a superior level of excellence andcontinuous improvement in everything suppliersdo to meet Ford customers' needs andexpectations.Criteria are based on the Supply BaseManagement process which focuses on productquality, engineering, delivery, and commercialperformance.

It requires a commitment to ensure excellenceand continuous improvement in all aspects ofbusiness.It involves executives, management, and itsemployees in creating a culture that emphasizes:leadership, information analysis, strategic qualityplanning, human resource utilization, qualityassurance of products and services, qualityresults, and customer satisfaction.

170

TQE minimum criteriaUnlike Ford Q1, which is awarded on a plantbasis, TQE is awarded on a commodity basis.All supplier plants producing the commodity musthold Q1 for at least one year, and be free ofvalidated initial sample rejections, ownernotifications, and recalls for the petitionedcommodity from at least six months prior tosubmission of the TQE petition through the awardapproval.Evidence of continuous improvement is requiredin all four major areas of evaluation: product quality engineering delivery commercial.

The TQE award requires also that suppliersprovide evidence of how they manage continuousimprovement.Areas of consideration and examples ofcontinuous improvement include, but are notlimited to: leadership information analysis strategic quality planning human resource utilization quality assurance of products and services quality results customer satisfaction.

In November 1991, Philips Semiconductors as atotal organization entered the TQE programmewith Ford Electronic Division. All product groupsand engineering, delivery and commercialfunctions have now reached, and are maintainingTQE level.The TQE Award (see photo) was granted toPhilips Semiconductors on November 3rd 1995.

171

Traceability / ROOTS

Its essential that the manufacturing history andthe location of all products manufactured byPhilips Semiconductors can be traced at any time. Traceability starts during diffusion, by identifyingthe diffusion batch. At assembly, the traceabilityidentifiers: date code and PMC (ProductManufacturing Codes for the diffusion andassembly site) are added. The traceability data ismarked on the product and printed on theidentification label.Traceability is particularly important if qualityproblems occur. For example, if a quality problemis detected after the delivery of products to awarehouse or customer, Philips Semiconductorswill take containment action to limit the damageeffects of such faulty products. If the problem isdetected while products are in the warehouse, thecontainment action taken is to block the affectedproducts to prevent delivery to customers. If theproblem occurs after delivery to the customer, thecontainment action taken is to alert the customerand provide specific information to trace the faultyproducts.

Backward traceabilityIn case of a customer complaint, a product can betraced back (from its lot-ID) to the productioncentres where diffusion and assembly took place.

Forward traceabilityIf a potential quality problem is detected in acertain lot, forward traceability identifiescustomers who have received products from thatlot and/or in which warehouses these products arestored.

Traceability system (ROOTS)ROOTS (Rapid On-line Overall TraceabilitySystem) is operational from the end of 1999.Thesystem contains a database with manufacturingand delivery data.

All manufacturing and warehouse systems will beconnected to this database to feed the traceabilitydata to the system. This is partly completed andwill be finished during 2000. The database willalso include the manufacturing data of thesubcontractors. Users can reach the database through Intranetand can define questions (reports) to the databaseabout manufacturing history of lots and aboutcustomers to which products of a lot wereshipped.The ROOTS dataflow schedule (Fig. 1) on thefollowing page shows the data collection points(oval-shaped) for ROOTS in the life cycle of aproduct. The main ones are:DIFS = diffusion startDIFE = diffusion endPTSO = pre-test outASMS = assembly startASMO = assembly outFTSO = final test outWHSE = transfer to distribution warehouseSHIP = shipment to customer. Further information on ROOTS can be obtainedfrom the ROOTS website:http://pww.sc.philips.com/qms/roots/index.htm

The traceability process is described in qualitystandard SNW-SQ-405.

172

DirectMaterials

DirectMaterials

PurchasedICs

MergePack

Diffusion orPre-Test Extra

Step

MultichipModule Merge

Assembly orFinal TestExtra Step

Diffusion(Wafer Fab)

IndustrialWarehouse

RegionalDistributionWarehouse

Pre-Test(Esort)

Assembly

Final Test

Tested Die/Wafer Sales

Customer

Subcontracted Grinding, Bumping, Sawing

Etch, Deposit, Passivate waferCritical Material: Silicon

Critical Materials: Lead Frame,Bonding Wire, Die AttachAdhesive, Moulding Compound

Electrically probe wafer

SLDI (Sliced wafer diffused with circuitry)

CEPT (Circuit Element Pre-Tested)Die Level

ICAM (IC Assembled / Marked)

ICFT (IC Final Tested)

Direct Shipment

PEP (Packed End Product)

Lead Die = Lot ID

DIF Batch ID + DC

Subcontracted Grinding, Burn-in, Extra Testing

>2 Lots

DIFX

DIFS

DIFE

PTSO

ASMS

ASMO

FTSO

MRGA

ASMX

MRGP

SHIP

SHIP

WHSE

Fig.1 ROOTS dataflow schedule.

173

Vision & Mission statement

The management charter of PhilipsSemiconductors (PSC) defines Vision & Missionas follows:

VisionPSC’s long-term vision is reflected in thestatements: we will be the customers’ first choice in Audio,

Video, Communications and combinationsthereof, for systems in the home and on themove

we will be recognized as a leader for complete,affordable and easy-to-use systems-on-siliconas well as for reliable, cost-effective supply ofmultimarket ICs and discrete semiconductors.

MissionPSC’s mission is to consistently increase valuefor its stakeholders based upon: full partnership relations with selected

customers hardware and software skills and business

models matched to the ‘Digital Era’ a platform-based system-on-silicon approach

for PSC’s key businesses premier league technology capabilities leading high-volume, low-cost manufacturing

capabilities being ‘employer of choice’ for customer-

oriented entrepreneurial people a high-quality, learning organization.

174

Wafer Level Reliability constitutes a set ofreliability tests which are done on wafer level, andwhich address mainly wearout mechanisms. Thetests are usually done on test structures, whichare designed to accelerate specific failuremechanisms. Some examples of mechanisms areelectromigration, hot carrier degradation, gateoxide breakdown, mobile ion instability, stressmigration, and junction spiking. The mostimportant mechanisms are described brieflybelow.

ElectromigrationThe electron current in metal interconnects exertsa force on the metal atoms which acts in thedirection of the electron flow. This causesdisplacement of the atoms, which can result invoids at places where the electron flux diverges,and in hillocks at places where the electron fluxconverges. Catastrophic failures occur when ametal line is completely open or when a shortcircuit between two metal lines has developed.Electromigration is tested by forcing a stresscurrent through metal lines at elevatedtemperature. The acceleration factor (A) is givenby Blacks equation:

where J1 and J 2 are the current densities at useand stress conditions respectively, T1 and T2 arethe corresponding absolute temperatures and k =Boltzmanns constant (8.6 x 10-5eV/K). Theactivation energy (Ea ) depends on details of thebackend process, and lies usually between 0.5and 1 eV. For AC currents the degradation isreduced compared to DC, due to duty cycling andhealing effects. For frequencies >1 kHzelectromigration does not cause measurabledegradation.

Wafer level reliability

Hot carrier degradationThe electric field in the channel of a MOStransistor in saturation reaches a maximum valuenear the drain. In this small region of high electricfield, electrons or holes are accelerated tovelocities higher than the thermal velocity. Undercertain bias conditions these hot carriers drift tothe Si-SiO2 interface where they can generateinterface traps, or pass the energy barrier at theinterface and get trapped in the oxide. The resultis a shift in the characteristics of the transistor.The degradation increases with decreasingchannel length, and is a concern for sub-microntechnologies.Hot carrier degradation is tested on MOStransistors under worst-case bias conditions.The acceleration factors (A) are given by theequations:

for NMOST

for PMOST

Vds1 and Vds2 are the drain-source voltagesunder use and stress conditions respectively; Leff1 and Leff2 are the effective channel lengths ofthe transistor in the product and test structurerespectively; B , c and C are process-dependentconstants.Under AC bias conditions the transistor stays inthe worst-case situation for only a fraction of thetime. This gives a cycle time reduction factor ofthe degradation, which can amount to more thanan order of magnitude.

J 2

J1

A = · exp · Ea 1 1

k T1 T2

A = exp B · *

n

c1 1

Vds1 Vds2

Leff1

Leff2

A = exp B ·1 1

Vds1 Vds2

Leff1

Leff2*exp C ·

175

Gate oxide breakdown Thin gate oxides suffer from electrical breakdownunder electrical field stress. There are severaltheories for the breakdown mechanism, but thereis no general agreement yet.Gate oxide breakdown is tested on MOScapacitors by applying electrical stress. In the EBDtest the oxide is stressed with a ramped voltageuntil breakdown and the breakdown field (EBD) isdetermined. The QBD test applies a constant or steppedcurrent until breakdown, and the total injectedcharge to breakdown (QBD) is determined. The TDDB (time-dependent dielectric breakdown)test stresses the oxide with a constant voltageuntil breakdown. This test allows extrapolation oftest results to use conditions. The mostconservative acceleration factor is given by theE-model:

where E1 and E2 are the oxide fields in use and instress conditions respectively, and γ is a constant.

A = exp ( γ · ( E2 E1 ))

176

Weibull

f(t)

t/n0 0.5 1 1.5 2

0.5

1

2

2.42.8

3.44

5.76.1β=

β < 1 early failuresβ = 1 constant failure rateβ > 1 wear out

Weibull probability paper and the associatedanalysis procedures are used extensively toanalyse the results of reliability testing. Givensufficient results, it is possible to obtain indicationsof the number and characteristics of failuremechanisms quickly, and to predict useful life.During investigations into fatigue phenomena inmetals, W. Weibull arrived in 1939 at the formulafor a family of distributions named after him. Thecurve of the Weibull function varies according tothe numerical values of the parameters, Fig. 1,especially shape factor ß. When ß is unity, theWeibull function reduces to the eponentialdistribution. When ß = 3.44, the distribution isnormal (Gaussian). In its most general form, theWeibull formula is:

R(t) = exp (t g)/n g) ß

Here, R(t) is the instantaneous reliability aftertime t, n is the characteristic life (the time to 63.2%failures), and g is the time during which no failuresoccur.In practice, Weibull methods make it possible todetermine in a straightforward way whichdistribution best fits a set of data. In practice, thisis usually done using Weibull probability paper, asshown in Fig. 2, in which the vertical axis isproportional to InIn1/R(t). The horizontal axis is alog scale of time. Where, for a given set of testconditions, points plotted on the chart lie on astraight line, it is likely that a single failure mode isinvolved and a Weibull distribution applies.

Fig. 1 The Weibull family of distributions varies with thevalues of the parameters used, especially shape factorß. When ß is unity, the distribution is exponential.

1 2 3 4 5^β

99.99990

50

20

10

5

2

1

0.5

0.2

0.110 102 103 104 105

age at failure (h)

cumulative failures

(%)

Article & Source

132.9/85 HEF 4000 family

132.9/85 high-speed CMOS

85/85 THB HEF 4000 family

85/85 THB high-speed CMOS

177

Weibull analysis of the bathtub curve thecharacteristic plot of failure rate versus time, Fig. 3 of a semiconductor might reveal an initial regionwhere ß<1 (early failures), with a final regionwhere ß>1 (wearout), Fig. 4. Plotted on a Weibullchart, these would resolve into two linesrepresenting the dominant early-failure andwearout mechanisms, Fig. 5.The Weibull shape factor ß is the directionalcoefficient of the line that best fits cumulativefailure/life points plotted onto a Weibull chart.

With charts provided with a ß estimation point andscale, Fig. 6, ß may be found by drawing a verticalto the (straight) line of best fit of the plotted pointsthat passes through the ß estimation point, andreading ß from the scale. Characteristic life h is, ofcourse, the intersection between the best fit lineand the 63.2% cumulative failure level. Since theorigin of the chart is at a low value of cumulativefailures, it is usually sufficient to take theintersection between the best fit line and the timeaxis as the value of g.

Fig. 2 Weibull plots of cumulative percentage failure against time. This example, taken from Technical Publication 249,compares the results of standard (85/85) and accelerated (132.9/85) humidity tests on CMOS ICs.

Failurerate

λ early failures

'constant failure rate'

wearout

time t

178

Fig. 3 The variation of failure rate with the life of adevice is typified by the bathtub curve. (Note: theconstant failure rate region (dashed) is a combination ofthe end of the early failures period and the beginning ofthe wearout period.)

Fig. 4 The bathtub curve of a given device type might beresolved into two Weibull distributions for the earlyfailures and for wearout.

Fig. 5 A Weibull plot of the results of a life test on asample of the devices of Fig. 4 would reveal two distinctregions corresponding to the two types of failure.

Fig. 6 Use of the ß estimation point. A vertical to the lineof best fit intersects the ß scale at the valuecorresponding to the failure distribution of the resultsplotted.

f(t)

β < 1(early failures)

β > 1(wearout)

t/n

%failure

β < 1

β > 1

log t

β estimation point

1 2 3 4 5 β^

99.9

99

90

70

50

30

20

10

5

3

2

1

.5

.3

.2

.15321 4 6 7 8 9 1

Cum

ulat

ive p

erce

nt fa

ilure

(fo

r N <

50

use

plot

ling

posit

ion

table

)

π Estimator

Article & Source

179

Fig. 6 Use of the ß estimation point. A vertical to the lineof best fit intersects the ß scale at the value correspon-ding to the failure distribution of the results plotted.

181

Part 3

Keywords

182

Keywords

Aaccelerated test:A test in which the applied stress level is chosento exceed that stated in the reference conditionsin order to shorten the time required to observethe stress response of the item, or magnify theresponse in a given time. To be valid, an accele-rated test shall not alter the basic modes andmechanisms of failure, or their relative prevalence.IEC 50 (191).

acceptance:A conclusion that a batch, lot or quantity ofproduct, material or service satisfies therequirement criteria based on the informationobtained from the sample(s).

acceptance criteria (in statistics):Specification criteria for acceptance of individual(quality) characteristics.

acceptance inspection:Inspection to determine whether an item, lotor service delivered or offered for delivery isacceptable.

alert: see customer alert.

approval:Declaration by a body vested with the necessaryauthority that a set of published criteria has beenfulfilled.

assessed reliability:The reliability of an item determined by a limitingvalue or values of the confidence interval asso-ciated with a stated confidence level, based on thesame data as the observed reliability of nominallyidentical items. IEC 50 (191).

audits:A Quality Audit is an independent, in-depthexamination of the documentation and implemen-tation of instructions and quality methods,especially for a production line or factory, but alsofor stores and other activities.Such audits often reveal deficiencies that havebeen ignored or missed by the people operatingthe department concerned, or stimulate thesolution of problems that limit achievable qualitylevels.Audits are usually performed by one or more smallteams two members per team is typical whospend a few days (depending on the size of theoperation) observing working practices andchecking documentation. Audits may be per-formed either by customers, third parties,government agencies or by corporate qualitymanagement. Before the actual audit, the QualityManual and other publications describing theactivity are usually examined by the auditor.A typical audit might start by the audit team beingintroduced to key people; the auditors may thendecide to divide activities between teams,allocating particular departments to each team.Following their examination of the working areasand documentation, the audit teams may interviewindividuals to discuss specific problems that havebeen revealed. After a final discussion, theauditors then prepare a report for circulation tokey people in the organization audited. After sometime, there may be a follow-up meeting to assessthe value of the audit recommendations.Since audits are regularly performed by majorcustomers or government agencies, it is essentialthat only accurate information is provided, whetherdirectly or by publications. To organize this in aprofessional way, required customer-audits shouldbe notified to the PD-auditor who will coordinatethe required actions.

183

average outgoing quality (AOQ):The expected average quality level of outgoingproduct for a given value of incoming productquality. Unless otherwise specified, the averageoutgoing quality (AOQ) is computed over allaccepted lots plus not-accepted lots after the latterhave been inspected 100% and the nonconformingitems have been replaced by good items.

Bbackward traceability:Tracing back a product to the production centresand to the manufacturing dates for each processstep.

batch:A definite quantity of some commodity manufac-tured or produced under conditions which arepresumed uniform. (ISO Guide 30)

bathtub curve:A plot of failure rate against time that resemblesa cross-section of a bath.

Ccall rate:The call rate of a set is the number of repairs perguarantee period per 100 sets sold expressed asa percentage (%).

concession:Written authorization to use or release a quantityof material or components already produced butwhich do not conform to the specifiedrequirements.

conformity:The fulfilment of a specified requirement by aquality characteristic of an item or service, theassessment of which does not depend essentiallyon the passage of time.

control chart:A chart, with upper and/or lower control limits,on which values of some statistical measure fora series of samples or sub-groups are plotted. The chart frequently shows a central line to assistdetection of a trend of plotted values towardseither control limit.

control limits:The levels indicating the upper and lower boun-daries within which a particular variable mayfluctuate in a given period without any actionsrequired to be taken.

corrective action:An action required by a user from its supplier tocorrect quality hazards and so prevent recurrence.

critical defect:A defect that, according to judgement andexperience, is likely to result in hazardous orunsafe conditions for individuals using, maintain-ing, or depending upon the considered product, orthat is likely to prevent performance of the functionof a major end item.

customer alert:Notification to the customer about a quality problemoccurring in one of the lots of a certain product.

184

Ddefect:The nonfulfilment of intended usage requirements.(ISO 3402).

defect (major):A defect, other than critical, that is likely to resultin a failure or to reduce the usability of theconsidered product for its intended purpose.

defect (minor):A defect that is not likely to reduce materially theusability of the considered product for its intendedpurpose, or that is a departure from establishedspecifications having little bearing on the effectiveuse or operation of this product.

derating:1. Reduction of the intensity of stress for thepurpose of gaining an advantage at another point,e.g. improvement of reliability.2. Translation of the test result under acceleratedconditions to a result under normal operatingconditions, by the use of calculation models.

design review:A formal, documented, comprehensive andsystematic examination of a design to evaluatethe design requirements and the capability of thedesign to meet these requirements and to identifyproblems and propose solutions. (ISO 8402).

double sampling:Sampling inspection in which the inspection of thefirst sample of the size given by the sampling planleads to a decision to accept a lot, not to accept it,or to take a second sample of the size given bythe sampling plan; the inspection of the secondsample then leads to a decision of acceptance ornon-acceptance.

Eearly failure period:That possible early period, beginning at a statedtime and during which the failure rate decreasesrapidly in comparison with that of the subsequentperiod. IEC 50 (191).

endurance test (in quality):An experiment carried out over a period of time toinvestigate how the properties of an item areaffected by application of stated stresses and theirduration. IEC 50 (191).

environment (in environmental testing):All external physical conditions that may influencethe performance of an item.

estimated process quality:The ppm level detected in sampling inspectionrecalculated in relation to lot size.

Ffailure:The termination of the ability of an item to performa required function. IEC 50 (191).(catastrophic): Failure which is both sudden andcomplete. IEC 50 (191).(complete): Failure resulting from deviations incharacteristic(s) beyond specified limits such as tocause complete lack of the required function. IEC50 (191).(critical): Failure which is likely to cause injury topersons or significant damage to material. IEC 50(191).(gradual): Failure that could be anticipated byprior examination or monitoring. IEC 50 (191).(major): Failure, other than a critical failure, whichis likely to reduce the ability of a more complex

185

item to perform its required function. IEC 50 (191).(minor): Failure, other than a critical failure, whichdoes not reduce the ability of a more complex itemto perform its required function. IEC 50 (191).(misuse): Failure attributable to the application ofstresses beyond the stated capabilities of theitem. IEC 50 (191).(partial): Failure resulting from deviations incharacteristic(s) beyond specified limits, but notsuch as to cause complete lack of the requiredfunction. IEC 50 (191).(sudden): Failure that could not be anticipated byprior examination or monitoring. IEC 50 (191).

failure mechanism:The physical, chemical or other process whichresults in failure. IEC 50 (191).

failure rate:For the stated period in the life of an item, the ratioof the total number of failures in a sample to thecumulative time on that sample. The failure rate isto be associated with particular and stated timeintervals (or summation of intervals) in the life ofthe items, and with stated conditions. IEC 50(191).

fall-off rate:The observed number of failures in the productionline, expressed in ppm. IEC 50 (191).

field call rate:See call rate.

field data:Data from observations during field use. IEC 50(191).

final inspection:Lot inspection carried out at the end of aproduction line.

fitness for use:The ability of a product, a process or a service tofulfil a defined purpose under specific conditions.

FITS (failures in time standard):failures in 109 hours.

forward traceability:Tracing a lot to the customers and/or stores whichhave received products from that lot, and tracingthe corresponding shipment data.

GHIincoming inspection:Lot inspection by a consumer on a lot delivered oroffered for delivery.

in-process (in-line) inspection:Product inspection carried out at various discretestages in manufacture.

inspection:Activities such as measuring, examining, testing,or gauging one or more characteristics of aproduct or service and comparing these withspecified requirements to determine conformity.(ISO 8402).

inspection (100%):Inspection of every item of product, process orservice, i.e. the whole (as contrasted with anyform of sampling inspection).

186

inspection by attributes:A method which consists in taking note, for everyitem of a population or of a sample taken from thispopulation, of the presence or absence of acertain qualitative characteristic (attribute) and incounting how many items have or do not have thischaracteristic.

inspection by variables:A method which consists of measuring aquantitative characteristic for each item of apopulation or of a sample taken from thispopulation.

inspection level:An index of the relative amount of inspection of asampling scheme, chosen in advance and relatingthe size of samples to the lot size, so that a lower(higher) intensity can be selected if pastexperience shows that this will be satisfactory.

JJust In Time (JIT):JIT production is a pull-through system asopposed to producing for stock. Upstreamoperations build only as much product asdownstream operations request. Also, purchasedparts or materials are received only in thequantities that the downstream operations need.The primary advantage is low inventory costs.Stocks are a major problem: they absorb capital,are subject to handling errors and damage, anddelay the discovery of production problems.The system demands that product is producedright first time and without delay. This requireswell optimized processes for manufacturing,planning and purchasing and requires a high levelof workforce involvement.

KLliability (product or service):A generic term used to describe the onus on aproducer or others to make restitution for lossrelated to personal injury, property damage orother harm caused by a product or service. Thelimits on liability may vary from country to countryaccording to national legislation. (ISO 8402).

lot: See batch.

lot-by-lot inspection:Inspection of products submitted in a series oflots.

lot tolerance percent defective (LTPD):A quality level which in a sampling plancorresponds to a specified and relatively lowprobability of acceptance (usually 10%).

Mmark of conformity:A mark attesting that a product or a service is inconformity with specific standards or technicalspecifications.

market-driven quality (MDQ):The partnership programme of IBM with itssuppliers. The programme is aimed at 'Quality,driven by market needs, that achieves totalcustomer satisfaction through the delivery oftimely, defect-free solutions that offer the bestvalue to customers'.IBM's market-driven principles are:

187

1. Make the customer the final arbiter.2. Understand our market.3. Commit to leadership in the markets we choose

to serve.4. Execute with excellence across our enterprise.

material specification:The document that describes in detail thematerials, components, or supplies used inmanufacturing the item.

maverick lot: A lot having an actual or potential problem thatmay go undetected until its use in the finalapplication. The problem can be in the field ofquality, reliability or functionality.

mean time between failures (MTBF):For a stated period in the life of an item, the meanvalue of the length of time between consecutivefailures, computed as the ratio of the cumulativeobserved time to the number of failures, understated conditions. IEC 50 (191).

mean time to failure (MTTF):For a stated period in the life of an item, the ratioof the cumulative time for a sample to the totalnumber of failures in the sample during the period,under stated conditions. IEC 50 (191).

merge (traceability): The function in which multiple lots are combinedinto one lot.

Nnonconformity:The nonfulfilment of specified requirements. (ISO3402).

normal inspection:The inspection which is used when there is noreason to think that the quality level of theproduction differs from the acceptable levelprovided for.

Oobserved reliability (of non-repaired items):For a stated period of time, the ratio of the numberof items which performed their functionssatisfactorily at the end of the period to the totalnumber of items in the sample at the beginning ofthe period. IEC 50 (191).

Pparameter:A variable in a system whose magnitude isdetermined by influences outside that system.

performance test:A test for assessing a performance characteristicdirectly or through simulation of the influencingfactors occuring in use, sometimes under moresevere conditions.

population:The totality of items under consideration.

predicted failure rate:For the stated conditions of use, and taking intoaccount the design of an item, the failure ratecomputed from the observed, assessed orextrapolated failure rates of its parts.

probability:A real number in the scale 0 to 1 attached to arandom event. It can be related to a long runrelative frequency or occurrence or degree of belief

188

that an event will occur. The scale 0 to 1 can beexpressed as a percentage. 100% is certainty.

probability of acceptance:When using a given sampling plan, the probabilitythat a lot will be accepted when the lot or processis of a given quality.

probability of rejection:The probability that a lot of a given quality will berejected by a given sampling plan.

process:The method of operation in any particular stage ofany element, group of elements or total aspect ofproduction or service.

process average:The process level averaged over a defined timeperiod or quantity of production.

process capability:A measure of inherent process variability.

process inspection:Inspection of a process by examination of theprocess itself or of the product characteristics atthe appropriate stage(s) of the process.

process quality control:That part of quality control that is concerned withmaintaining process variability within the requiredlimits.

process under statistical control:A process, the mean and variability of whichremain stable with no adverse trends.

product-hold: The function to prevent delivery of products from acertain lot.

product qualification package:At the introduction of new products, or aftersignificant changes in products (or processes), a set of quality and reliability information must bemade available. This information can be based ongeneric data of the product family supported bytype-specific data indicating that set targets will bereached. This product qualification package mustbe suitable for giving to customers.

Qqualification approval:The status given to a manufacturer's productionunit, whose product has been shown to meet allthe requirements of the product specification andquality plan.

quality:The totality of features and characteristics of aproduct or service that bear on its ability to satisfystated or implied needs. (ISO 8402).

quality assurance:All those planned and systematic actionsnecessary to provide adequate confidence that aproduct or service will satisfy requirements forquality. (ISO 8402).

quality audit:A systematic and independent examination todetermine whether quality activities and relatedresults comply with planned arrangements andwhether these arrangements are implementedeffectively and are suitable to achieve objectives.(ISO 8402).

quality control:The operational techniques and activities that areused to fulfil requirements for quality. (ISO 8402).

189

quality level:Any relative quality measure obtained by comparingobserved values with the relevant requirements.

quality management:That aspect of the overall management functionthat determines and implements the quality policy.(ISO 8402).

quality manual:A document setting out the general qualitypolicies, procedures and practices of anorganization.

quality operating system (QOS):A Ford methodology focused on ContinuousImprovement, QOS requires the gathering ofrecent data on business key parameters, whichare relevant for the department concerned. Qualityoperating systems are used for each level of thebusiness chain. The data is analyzed andpresented in such a way that it can be quicklyreviewed by management. In this way QOSprovides a systematic method of monitoringperformance improvement on key parameters,and at the same time allows for a data-drivenmanagement approach towards problemrecognition.

quality plan:A document setting out the specific qualitypractices, resources and sequence of activitiesrelevant to a particular product, service, contractor project. (ISO 8402).

quality policy:The overall quality intentions and direction of anorganization as regards quality, as formallyexpressed by top management. (ISO 8402).

quality system:The organizational structure, responsibilities,

procedures, processes and resources forimplementing quality management. (ISO 8402).

Rrange:The difference between the greatest and thesmallest observed values of a quantitativecharacteristic.

rejection:A conclusion that a quantity of a product, materialor service has not been shown to satisfy therequirement criteria based on the informationobtained from the sample(s).

reliability:The ability of an item to perform a requiredfunction under stated conditions for a statedperiod of time. IEC 50 (191).

repeatability (of measurements):The closeness of the agreement between theresults of successive measurements of the samequantity carried out by the same method, by thesame observer, with the same measuringinstruments, in the same laboratory at quite shortintervals of time. (ISO GUIDE 30).

risk:The combined effect of the probability of occur-rence of an undesirable event, and the magnitudeof the event.

rogue lot (batch):A lot rejected under circumstances indicating afundamental processing error, such as mishandling.A rogue lot is usually defined as being a lot forwhich the sampling inspection result indicated aless than 5% chance of acceptance according tothe relevant sampling system.

190

Ssample (of a reference material):A representative quantity of material extractedfrom a batch of reference material. (ISO GUIDE30).

sampling inspection:The inspection of products, processes or servicesusing samples (as distinct from 100% inspection).

sampling plan: A specific plan which states sample size(s) to beused and the associated acceptance criteria.

sampling system:A collection of sampling schemes, e.g. oneindexedby lot-size ranges, inspection levels and AQLs.

screening inspection:Complete inspection, i.e. 100% examination of aquantity of material or items of a product, withrejection of all items or portions foundnonconforming.

screening test:A test, or combination of tests, intended to removeunsatisfactory items or those likely to exhibit earlyfailures. IEC 50 (191).

Shewart control chart:1. A chart for controlling a process by attributes

using percent nonconformity 2. A chart for controlling a process by variables

using charts for controlling the central locationand the dispersion.

Shewart control limits:In a control chart, the limit below which (upperlimit) or above which (lower limit) or the limits

between which the statistic under considerationlies with a very high probability when the processis under control.

specification:The document that prescribes the requirementswith which the product or service has to conform.(ISO 8402).

split (traceability): The function in which a single lot is divided intotwo or more separate lots.

standard deviation:The square root of the variance (see variance).

state of statistical control:A state in which the variations among theobserved sampling results can be attributed to asystem of chance causes which does not appearto change with time.

statistical process control (SPC):The application of statistical quality control toindividual process stages.

statistical quality control:That part of quality control in which statisticaltechniques are used.

Ttarget specification:The document that describes the primary purposeof an item and gives the essential guidanceconcerning such matters as its style, grade,performance, appearance, conditions of use(including health and safety considerations),characteristics, packaging, conformity, reliability,maintenance, etc.

test specification:The document that describes in detail themethods of conducting tests including, ifnecessary, the criteria for assessing the result.

tightened inspection:The inspection, more severe than the normalinspection, to be applied when the inspectionresults of a number of lots indicate that the qualitylevel of the production is below specifications.

traceability:The ability to trace the history, application orlocation of an item or activity, or similar items oractivities, by means of recorded identification.(ISO 3402).(backward): A product is traced back to theproduction centres for diffusion, assembly, testingand packout to the manufacturing processes andquality data. This is needed in the case of acustomer complaint.(forward): The customers who received productsof a certain lot, are traced. This is needed for analert, when customers must be warned for a(potential) quality problem in a certain lot.

type approval:Approval of a certain product or group of productsconsidered by the approval body asrepresentative for the continuous production.

Uuseful life:The period from a stated time, during which, understated conditions, an item has an acceptablefailure rate, or until an unrepairable failure occurs.IEC 50 (191).

Vvariance:The sum of the squares of the difference betweenthe values of the number of observations and thearithmetic means of these observations divided bythe number of observations.

verification of reject:The process whereby, following examination bythe supplier, an item rejected as unserviceable atsome stage in its use is agreed to contain one ormore defects according to the agreed specification.

verification sampling:A sampling scheme to ascertain whether theproducer's sampling procedures are inaccordance with his declared sampling scheme.

Wwaiver: 1. For material or products see concession.2. For procedures, a written authorization todeviate from a specified requirement. This waivermust be temporary, the end date being part of thewritten authorization.

warning limits:In a control chart, the limit below which (upperlimit) or above which (lower limit) or the limitsbetween the statistic under consideration lieswith a high probability when the process is undercontrol.

wear-out failure:Failure whose probability of occurrence increaseswith the passage of time and which occurs as aresult of processes which are characteristic of thepopulation.

191

192

XYZzero-defect philosophy:The principle that no defect level is acceptable,but that all defect causes should be traced andeliminated in the quest for perfect products. Notethat 'zero' can never be a target for process-average reject levels since such targets, bydefinition, must be both achievable andmeasurable.

193

Index

abbreviations 50accelerated life testing 93accelerated test 182acceleration factors 53acceptable quality level 56acceptance 182acceptance criteria 182acceptance inspection 182acceptance tests 144activation energies 53Albuquerque, USA 28alert 182analysis 61, 90AOQ 183appraisal costs 66approval 182AQL 56Arrhenius equation 53assembly council 155assembly quality control 58assessed reliability 182audits 182automotive quality standards 123average outgoing quality 183

backward traceability 183, 191Bangalore, India 42Bangkok, Thailand 40batch 183, 189bathtub curve 183 BEST 11, 59Böblingen, Germany 17brainstorming 135business excellence 11business excellence policy 12

Cabuyao, Philippines 36Caen, France 16Calamba, Philippines 37call rate 183catastrophic failure 184CHAMP 61checksheets 134chemical content of semiconductors 60 CMM 159complaint analysis 61complaint processing 61complete failure 184concession 183confidence level 64, 93conformity 183contents 1control chart 139, 162, 183control limits 183corrective action 183cost of quality 66

CPCN 68critical defect 183critical failure 184 customer alert 183customer notification 68customer-specific labeling 69

defect 184derating 184design review 184discontinuation of delivery 68DOD 68double sampling 184drypack 70

early failure period 184eco-design 81eco-product award 96ecovision 83Eindhoven, The Netherlands 43electromagnetic compatibility 72electromigration 174electrostatic discharge 73EMC 72endurance test 184environment 184environmental care 80environmental policy and goals 82, 83ESD 73estimated process quality 184evolution of quality 84excellence 11external standardization 88

fact gathering 134failure 184failure analysis 90failure costs 66failure mechanism 185failure modes and effects analysis 91failure rate 108, 185failures in time standard 93, 185fall-off rate 185field call rate 185field data 185final inspection 185fishbone diagram 137Fishkill, USA 29fitness for use 185FITS 93, 185FMEA 91forward traceability 185, 191

gate oxide breakdown 172general quality specifications 95goals 83

194

misuse failure 185monitoring tests 144moisture sensitivity level 109 MSL 109MTBF 108, 187MTTF 187

Nijmegen, The Netherlands 21nonconformity 187normal inspection 187notification of changes 68

observed reliability 187ODC-free 81organization of PSC 5organized for quality 4

parameter 187Pareto analysis 136partial failure 185Paynter chart 142PBE 11, 110PDCA cycle 131performance test 187Philips Business Excellence 110Philips Business Excellence policy 112Philips Quality 113Philips Values 114policy 12, 82population 187PPM 115PQA-90 117PQRA 121predicted failure rate 187preventive cost 66probability 187probability of acceptance 188probability of rejection 188problem solving 126, 131process 188process average 188process capability 188process inspection 188process quality control 188process survey tools 59process under statistical control 188product-hold 188product/manufacturing centres 14product manufacturing codes 118product pruning 68product qualification package 188Product Quality & Reliability Assurance database 121product release pruning 68QFD 125QIC 128

gradual failure 184Gratkorn, Austria 41green flagship products 96 Guangdong, China 34

Hamburg, Germany 18hazardous material 81Hazel Grove, United Kingdom 26headquarter audits 59histogram 138history of Philips Semiconductors 98Hong Kong, China 35hot carrier degradation 174house of quality 125Hsinchu, Taiwan 43humidity 54

identification labeling 103in-line inspection 185in-process inspection 185incoming inspection 185inspection 185inspection by attributes 186 inspection by variables 186inspection level 186Introduction 2ISO 9000 11, 105

JIT 186just in time 186Kaohsiung, Taiwan 38keywords 181, 182knowledge management 59

label coordinators 69labeling 69, 103labels 69liability 186Limeil, France 41lot 186lot-by-lot inspection 186lot tolerance percent defective 186LTPD 186

major failure 184mark of conformity 186market-driven quality 186marking of ICs 107material specification 187maverick lot 187MDQ 186mean time between failures 108, 187mean time to failure 187merge 187minor failure 185mission statement 173

195

QOS 189QS-9000 123qualification approval 188qualification tests 144quality 188quality assurance 188quality audit 188quality control 188quality function deployment 125quality history 84quality improvement competition 128quality/improvement managers 44quality level 189quality management 189quality manual 189quality operating system 189quality organization 6quality plan 189quality policy 189quality standards for customers 130quality system 189quality techniques and tools 131quality testing 144

radar chart 143range 189rejection 189release of new products 146reliability 148, 189repeatability 189return shipments 150risk 189rogue lot 189ROOTS 171

SAC 155sales quality managers 10sample 190sampling on the fly 152sampling inspection 190sampling plan 190sampling system 190San Antonio, USA 30San Jose, USA 31scatter diagram 140screening inspection 190screening test 190self-qualification 154semiconductor assembly council 155separate units, joint ventures and subcontractors 41

Serembam, Malaysia 42Shanghai, China 41Shewart control chart 190Shewart control limits 190Singapore 42

six sigma 156software capability 159software quality 158software release 158Sophia Antipolis, France 41Southampton, United Kingdom 27SPC 160, 190specification 190spider graph 143split 190Stadskanaal, The Netherlands 24standard deviation 190standardization 88state of statistical control 190statistical process control 160, 190statistical quality control 190structural similarity 166sudden failure 185Sunnyvale, USA 32supplier quality system 167

Taipei, Taiwan 39target specification 190team-oriented problem solving 133 Tempe, USA 43test specification 191testing 58, 144 thermal cycling 54thermal resistance 168tightened inspection 191total quality excellence 169TQE 169traceability 171, 187, 191type approval 191

useful life 191

values variance 191verification of reject 191verification sampling 191 vision & mission statement 173

wafer level reliability 174waiver 191warning limits 191wear-out failure 191Weibull 176 worldwide standardization 88zero-defect philosophy 192 Zürich, Switzerland 25

8-D method 133

Philips Quality Reference Handbook 2002.pdf

Documents

Transcript of Philips Quality Reference Handbook 2002.pdf