DISCRETE SEMICONDUCTOR DEVICE RELIABILITY · 2011. 5. 15. · semiconductor device reliability. It...

of 118 /118
DISCRETE SEMICONDUCTOR DEVICE RELIABILITY DTIC ELECTE I% l _ L -i A 3 1 4

Embed Size (px)

Transcript of DISCRETE SEMICONDUCTOR DEVICE RELIABILITY · 2011. 5. 15. · semiconductor device reliability. It...

  • DISCRETESEMICONDUCTORDEVICERELIABILITY

    DTICELECTE

    I% l _ L -i A 3 1 4

  • The information and data contained herein have been compiled from

    government and nongovernment technical reports and from material

    supplied by various manufacturers and are intended to be used forreference purposes. Neither the United States Government nor lIT

    Research Institute warrant the accuracy of this information and data. The

    user is further cautioned tnat the data contained herein may not bc used in

    lieu of other contractually cited references and specifications.

    Publication of this information is not an expression of the opinion of The

    United States Government or of lIT Research Institute as to the quality or

    durability of any product mentioned herein and any use for advertising orpromotional purposes of this information in conjunction with the name of

    The United States Government or lIT Research Institute without writtenpermission is expressly prohibited.

  • Ordering No. DSR-4

    DISCRETESEMICONDUCTOR

    DEVICE RELIABILITY

    1988

    Prepared by:

    Mary G. PrioreThe Reliability Analysis Center

    Under Contract to:

    Rome Air Development CenterGriffiss AFB, NY 13441-5700

    i Reliability Analysis Center

  • The Reliability Analysis Center (RAC) is a Department of Defense Information AnalysisCenter sponsored by the Defense Logistics Agency, managed by the Rome Air DevelopmentCenter (RADC), and operated at RADC by lIT Research Institute (IITRI). RAC is chartered tocollect, analyze and disseminate reliability information pertaining to electronic systems andparts used therein. The present scope includes integrated circuits, hybrids, discretesemiconductors, microwave devices, optoelectronics and non-electronic parts employed inmilitary, space and commercial applications.

    Data is collected on a continuous basis from a broad range of sources, including testinglaboratories, device and equipment manufacturers, government laboratories ano equipmentusers (government and non-government). Automatic distribution lists, voluntary datasubmittals and field failure reporting systems supplement an intensive data solicitationprogram.

    Reliability data and analysis documents covering most of the device types mentionedabove are available from the RAC. Also, RAC provides reliability (.onsulting, training, technicaland bibliographic inquiry services which are discussed at the end of this document.

    REQUEST FOR TECHNICAL ALL OTHER REQUESTSASSISTANCE AND INFORMATION SHOULD BE DIRECTED TO:ON AVAILABLE RAC SERVICESAND PUBLICATIONS MAY BEDIRECTED TO:

    Reliability Analysis Center Rome Air Development CenterP.O. Box 4700 RBE/Preston R. MacDiarmidRome, NY 13440-8200 Griffiss AFB, NY 13441-5700

    Technical Inquiries (315) 337-9933 Telephone: (315) 330-4920Non-Technical Inquiries: (315) 330-4151 Autovon: 587-4920Autovon: 587-4151TeleFax: (315) 337-9932

    © 1988, lIT Research InstituteAll Rights Reserved

    ii

  • SECURITY CLASS F!CA-,ON OF TIS PAGE

    Form Approved

    REPORT DOCUMENTATION PAGE OMB No 0704-0188la REPORT SECGRjTY CASS,-CAT ON lb RESTRICTIVE MARKINGS

    Unclassified2a SECURITY CLASSIFICATON AUTHORITY 3 DISTRIBUTION/AVAILABILITY OF REPORT

    Approved for public release; distribution2b DECLASSIFCAT,ON DOWNGRADING SCHEDULE unlimited. Available from RAC or DTIC

    in microfiche only4 PERFORMING ORGANiZATiON REPORT NuMBER(S) S MONITORING ORGANiZAT;ON REPORT NUMBER(S)

    DSR-4

    6a NAME OF PERFORMING ORGANIZATION 6b OFF CE SYMBOL 7a NAME OF MONITORING ORGANIZATION

    I (if applicable)Reliability Analysis Center I RADC/RAC RADC/RBE

    6c. ADDRESS (City State, and Z:PCode) 7b ADDRESS (Cty, State, and ZIP Code)

    Rome Air Development CenterGriffiss AFB, NY 13441-5700 Griffiss AFB, NY 13441-5700

    8a. NAME O; FUNDING SPONSORING 8b OFFICE SYMBOL 9 PROCUREMENT INSTRUMENT IDENTIFICATION NUMBERORGAN!ZAT.ON (If applicable)

    DLA/DTIC DTIC-Al F30602-87-C-02288c. ADDRESS (City, State and ZIPCode) 10 SOURCE OF FUNDING NUMBERS

    DTIC PROGRAM PROJECT TASK WORK UNITCameron Station ELEMENT NO NO NO ACCESS;ON NO

    Alexandria, VA 2231411 TITLE (Include Security Classification)

    Discrete Semiconductor Device Reliability12. PERSONAL AUTHOR(S)

    Mary G. Priore13a. TYPE OF REPORT j13b T!ME COVERED 14. DATE OF REPORT (Yea, Month, Day) 15 PAGE COUNT

    FROM TO _ 1988, March 2516 SUPPLEMENTARY NOTATION Hard copies available from Reliability Analysis Center, RADC/RAC,Griffiss AFB, NY 13441-5700 (Price $100.00) DTIC will provide microfiche copies to the DoDand its contractors at the standard microfiche price. -

    17 COSATI CODES 18 SUBJECT TERMS (Continue on reverse if necessary and identify by block number)FIELD GROUP SUB-GROUP transistors diodes optoelectronics

    liquid crystal displays photovoltaic cells

    infrared emitting diodes transient suppressor diodes19 ABSTRACT (Continue on reverse if necessary and identify by block number)

    The objective of this publication is to update the discrete semiconductor device reliability dataand information currently available to the reliability community. Field and test failure data

    are presented in detailed and summarized formats on state-of-the-art transistors, diodes,and optoelectronic devices. Significant issues with respect to discrete semiconductor device

    reliability are investigated including the validity of the exponential time to failure distribution,the effect of technological advances on device reliability, and the effects of electrical parameter

    derating. In addition, failure mode and mechanism data is presented for transistors, diodes, liquid

    crystal dizclay qnd photc.voltaic cell array modules.

    20 DSTRiBU-iON AVAILABILITY OP ABSTRACT 21 ABSTRACT SECURITY CLASSIFICATION'JNCLASSIr-ED INLIMITED E0 SAME AS RPT 0 DTIC USERS

    22a NAME OF RESPONSIBLE INDIJIDIJAL 22b TELEPHONE (Include Area Code) 22c OFFICE SYMBOL

    Steven J. Flint, RAC Technical Director

    DO Form 1473, JUN 86 Previous editions are obsolete SECURITY CLASSIFICAT;ON O TH'S PAGE

    iii

  • PREFACE

    This book is the fourth in a series of RAC data publications dealing with discretesemiconductor device reliability. It offers a detailed presentation of transistor, diode andoptoelectronic device failure experience data as well as failure mode and mechanism information.Pertinent analyses have been performed on the available data, investigating trends and presentingconclusions. The data ini this publication covers the 1976-1987 time span and has been collectedfrom both military and commercial data sources.

    DSR-4 is intended to supplement the data and guidelines in various military publications suchas MIL-HDBK-217E and is not intended to be used in lieu of contractually cited references.

    Other available RAC databooks treat the reliability experience of microcircuit, hybrid, andnonelectronic components.

    Major technical contributors to this effort were David Coit, David Mahar and WilliamCrowell. Typing and assembly were performed by Jeanne Lasher.

    Accession -For

    NTIS GRA&IDTIC TABUn]armoupced 0justlflcatil

    Distribution/-

    Availabillty CodesAvaitl and/or

    Dist. Special

    V

  • TABLE OF CONTENTS

    Page

    Preface vS umrn-ry ixIntroduction xiii

    1.0 USER'S GUIDE TO DSR-4 1

    2.0 DATA ANALYSIS 62.1 Data Summaries 62.2 Validity of the Exponential Failure Assumption 102.3 DSR-3/DSR-4 Data Comparison 252.4 Derating Effects 26

    3.0 FAILURE MODE AND MECHANISM DATA 28

    4.0 DETAILED DATA 34

    5.0 DATA SOURCES 99

    APPENDIX A: REFERENCES

    APPENDIX B: ADDITIONAL RAC SERVICES

    LIST OF TABLES

    Table 1: Discrete Semiconductor Device Data Summary xiTable 2: Average Component Failure Rate 5Table 3: Weibull Shape Parameters 11Table 4: Observed Weibull Parameters 23Table 5: (D) Test Results 24Table 6: DSR-3 vs. DSR-4 Component Failure Rate Comparison 25Table 7: Diode Failure Modes 28Table 8: Diode Failure Mechanisms/Location 29Table 9: Transistor Failure Modes 30Table 10: Transistor Failure Mechanism/Location 31Table 11: Liquid Crystal Display Failure Distribution 32Table 12: Photovoltaic Cell Modules Failure Distribution 33Table 13: Discrete Semiconductor Device Field Reliability Data Source Descriptions 1f1

    LIST OF FIGURES

    Figure 1: Average Failure Rate Comparison by Diode Type 8Figure 2: Average Failure Rate Comparison by Transistor Type 8Figure 3: Diode Average Failure Rate by Data Source 9Figure 4: Transistor Average Failure Data by Data Source 9Figure 5: Weibull Plot of High Temperature Operating Life

    Data for GaAs FETs 12Figure 6: Weibull Plot of High Temperature Operating Life

    Data for GaAs Laser Diodes 12

    vii

  • LIST OF FIGURES (Cont'd)

    PageFigiure 7: Weibull Plot of High Temerature Operating Life

    Data for GaAs Laser Diodes 13Figure 8: Weibull Plot of High Temperature Operating Life

    Data for GaAs Laser Diodes 13Figure 9: Weibull Plot of High Temperature Operating Life

    Data for GaAs Laser Diodes 14Figure 10: Weibull Plot of High Temperature Operating Life

    Data fcr GaAs Laser Diodes 14Figure 11: Weibull Plot of High Temperature Operating Life

    Data for GaAs FETs 15Figure 12: Weibull Plot of High Temperature Operating Life

    Data for GaAs FETs 15Figure 13: Weibull Plot of High Temperature Operating Life

    Data for High Power Pulsed IMPATr Diodes 16Figure 14: Weibull Plot of High Temperature Operating Life

    Data for High Power Pulsed IMPATT Diodes 16Figure 15: Weibull Plot of High Temperature Operating Life

    Data for (A lGa) as Lasers 17Figure 16: Weibull Plot of High Temperature Operating Life

    Data for in GaAs P/in P DH Lasers 17Figure 17: Weibull Plot of High Temperature Operating Life

    Data for JAN 2N918 18Figure 18: Weibull Plot of High Temperature Operating life

    Data for GaAs Power FETs 18Figure 19: Weibull Plot of High Temperature Operating Life

    Data for Low Nosie GaAs FETs 19Figure 20: Weibull Plot of High Temperature Operating Life

    Data for Low Noise GaAs FETs 19Figure 21: Weibull Plot of High Temperature Operating Life

    Data for Low Noise GaAs FETs 20Figure 22: Weibull Plot of High Temperature Operating Life

    Data for Low Noise GaAs FETs 20Figure 23: Weibull Plot of High Temperature Operating Life

    Data for Low Noise GaAs FETs 21Figure 24: Weibull Plot of High Temperature Operating Life

    Data for Low Noise GaAs FETs 21Figure 25: Weibull Plot of High Temperature Operating Life

    Data for Low Noise GaAs FETs 22Figure 26: Failure Rate vs. Voltage Stress for High Power

    Bipolar Transistors 27Figure 27: Failure Rate vs. Voltage Stress for Signal and

    Rectifier Diodes 27Figure 28: Diode Failure Mode Distribution 29Figure 29: Transistor Failure Mode Distribution 30Figure 30: Liquid Crystal Display Failure Mode Distribution 32Figure 31: Photovoltaic Cell Array Failure Mode Distribution 33

  • SUMMARY

    The objective of this publication is to update the discrete semiconductor device reliability dataand information currently available to the reliability community.

    Field and test failure data are presented in detailed and summarized formats on state-of-the-arttransistors, diodes, and optoelectronic devices. In addition to the standard component types,particular emphasis was given to the collection of data on the following low population part types:

    GaAs Power FETsVaristor DiodesInfrared LEDsCurrent Regulator DiodesMicrowave Diodes and TransistorsLiquid Crystal DisplaysPhotovoltaic Cell ArraysTransient Suppressor DiodesDiode Array DisplaysMicrowave DiodesMicrowave Transistors

    Table 1 summarizes the data presented in this publication.

    Significant issues with respect to discrete semiconductor device reliability are investigatedincluding the validity of the exponential time to failure distribution, the effect of technologicaladvances on device reliability, and the effects of electrical parameter derating. In addition, failuremode and mechanism data is presented for transistors, diodes, liquid crystal displays andphotovoltaic cell array modules.

    ix

  • N~~C 'C -n 00 LC 0 OO - 0C0C Nrr tr, c,, cN 0r~l C 1 m r- 0 OC - t w 0' C14 w V) z Z

    L/ - - r,4 r- r- r, 4 4 N O4 NA 0-0 Il 0 -4 r0 -1 NX C1 N c r- N C N -I Cl OC 00r '4 r- :tC Ln 0 - l

    oc = rl 0 0~ - m oN r- ml NC lIt - M - O-- N . m~ Nt C\0 N OC -00

    m0 rr- - rm0 N rA C) O4N \- N 00 - rl

    Cl K

    ;z 'C, ::7 ri- 0 r . - l N C f 000 O t ri mI 'C 4 L/- r- f Cl Nr 'C tr N o- 'f) tr,~ O C N O 0 C

    -t rj,) Cl Vf) i ' 00 CC , OC Cl

    'C 'C oo kT ~ l' C C 0 0 - ~ 0 O lC n f I) 0~~'I Lr. r' ' l C t Cl - 0 - 0qNIf t N ~ C

    ~~~~~~~ N Nz 0 If N

    $. C V w c C.v- v

    0z 0

    -/ -0

    -( -

    CE 0

    ~- xi

  • DC C14 r~ON C1 V) - c 1 0CO '~N r- m C ",n f " l - \40 r-(- C C

    NN 000 Nc 00 CA \0 0- " I Ch 00~ '1,0 r- CD N \0DX ~ -ClC O'.r0ONc . ) g - -q M "l0 ~ct - G-)

    m~ C4 00 ON NT 00

  • INTRODUCTION

    This book provides up-to-date failure rate information on discrcte semiconductor devicez.The data, including both field experience and test data, spans the period from 1976-1987 updatingthe pre-1978 DSR-3 data.

    Data is presented on all transistors, diodes and optoelectronic part types currently covered inMIL-HDBK-217E as well as special or low-population parts including:

    • GaAs Power FETs• Transient Suppressor Diodes• Infrared LEDs• Diode Array Displays• Current Regulator Diodes• Photovoltaic Cells* Liquid Ciystal Displays• Microwave Diodes* Microwave Transistors

    Although liquid crystal displays are not semiconductor devices, the RAC regulaily receivesinquiries from users requesting information as to their reliability and will use this databook as themedium for presenting such information.

    Pertient device and application-specific details characterizing the device failure are presentedin the detailed data section including:

    • Device Type• Package Type• Screen Level• Application Environment• Temperature• Electrical Stress* Number Tested* Number Failed• Part Hours

    Failure mode and mechanism data was compiled for transistors and diodes from reliabilitydemonstration test reports and is presented in both tabular and graphical formats. Data fromindividual data sources is presented for photovoltaic cell modules and liquid crystal displays. Suchdata is critical to Failure Modes, Effects and Criticality Analyses (FMECAs).

    Various graphical analyses and comparisons are performed on the data and results arediscussed from a reliability perspective. In addition, a comparison of DSR-4 field failure ratesagainst DRS-3 field failure rates was conducted to determine the effects of processing andtechnology changes over the last 10-15 years.

    Xiii

  • 1.0 USER'S GUIDE TO DSR-4

    The data and information presented in this publication are intended to aid the reliabilityengineer in assessing the reliability of electronic systems and equipment containing discretesemiconductor devices. The user is cautioned, however, that this data is intended as a supplementto the information in military documents such as MIL-HDBK-217E, Reliability Prediction ofElectronic EqUipment, amd shall nor be used in lieu of contractually cited references.

    Data is presented on the gamut of discrete semiconductor devices including small signal,rectifier, and zener diodes, bipolar and field effect transistors, microwave diodes and transistors.Of particular interest will be the data presented on low population, state-of-the-art devices whichare either not addressed in other reliability-related documents or for which available data is out-of-date and/or scarce.

    This data book is organized into the following four sections:

    2.() Failure Data Analyses3.0 Failure Mode and Mechanism Data4.0 Detailed Failure Data5.0 Data Sources

    Section 2.0 presents the results of various analyses performed on the detailed data presentedin Section 4.0. First, data summaries are presented for generic component categories for whichthere was sufficient field failure data. The data in these analyses included data from five datasources (designated as Sources I through 5) which are described in detail in Section 5.0. Basedon the data in this table, graphical comparisons ' . made between various diode and transistortypes, respectively. These comparisons are intended to assist engineers in design trade-offdecisions.

    A graphical comparison between component failure rates from the various data sources isalso presented. Data Sources 1 through 3 represent airborne applications which employ military-grade (JANTX) components. Data Source 4 provided data from a military naval environment alsoemploying military grade (90% JANTX, 10% JAN) components, and data Source 5 provided datafrom a number of commercial laboratory-type equipments with unscreened components (77.5%plastic encapsulated and 22.5% hermetically sealed).

    The second analysis in Section 2.0 is an investigation of the validity of the exponential time-to-failure assumption, which is the basis for all MIL-HDBK-217E-type semiconductor failure rateprediction models. The validity of this assumption has been challenged often since it assumes aconstant failure rate over time while most failure mechanisms affecting discrete semiconductordevices are reported in the literature as fitting a log-normal distribution.

    The third analysis presented in Section 2.0 investigates the impact of design and processingtechnology advances made over the last 10-15 years on observed field failure rates. To performthis investigation, field data in this publication is compared with field failure data from similarequipments presented in its' predecessor DRS-3 (dated 1979).

    The final analysis presented in Section 2.0 is an investigation of the effects of electrical stresson component failure rates. Several publications such as RADC-TR-84-254, Reliability DeratingProcedures and ESD-TR-85-145, Derated Application of Parts for ESD Systems, stress thebenefits of electrical derating on equipment reliability. These effects are investigated for part typesfor which sufficient stress level information was available.

  • Section 3.0 presents failure mode and mechanism data for transistors and diodes compiledfrom reports of reliability demonstration tests conducted in accordance with MIL-STD-781,Reliability Design, Qualification and Production Acceptance Tests: Exponential Distribution. Thisdata was limited to MIL-STD-781 type data to ensure all tests were conducted and evaluatedagainst similar criteria. An exception was made for two low-population part types: liquid crystaldisplays and photovoltaic modules. Since so little data is available on these parts, other sources ofinformation were allowed. Failure mode and mechanism distributions are critical to theperformance of an accurate Failure Mode Effects and Criticality Analysis (FMECAs)

    Section 4.0 presents the detailed failure data. The data is segregated first by generalcomponent category, then within each category by part number and data type (field or test). Thedevice categories presented are:

    DiodeSmall Signal, SwitchingSmall Signal, General PurposeRectifierRectifier, Fast RecoveryRectifier, Bridge, Full WaveZenerZener, Voltage RegulatorZener, Voltage ReferenceCurrent RegulatorTransient SuppressorMicrowave, TunnelMicrowave, Schottky BarrierMicrowave, PINMicrowave, VaractorMicrowave, GUNN EffectMicrowave, IMPATT

    TransistorBipolar, NPNBipolar, PNPField Effect, Junction, N-channelField Effect, Junction, P-channelField Effect, MOS, N-channelField Effect, MOS, P-channelUnijun -tionMicrowave/RF, Field EffectMicrowave/RF, BipolarMultiple, Complementary PairMultiple, DarlingtonMultiple, Matched PairSpecial Function, Choppcr

    ThyristorThyristorSCRTRIACTriode, Trigger

  • OptoelectronicEmitter, Single LEDEmitter, Single LED, InfraredEmitter, LED ArrayEmitter, Laser DiodeSensor, PhotodiodeSensor, PhototransistorPhotocoupler,Photocoupler, Phototransistor OutputPhotocoupler, Photodarlington OutputPhotocoupler, IC OutputAlphanumeric Display, LED, Segment TypeAlphanumeric Display, LED, Array TypeAlphanumeric Display, Liquid CrystalPhotovoltaic Module

    Pertinent device and application-specific details are presented for each data record. The fieldand associated field values in the detailed data section are explained in Section 4.0.

    Although the majority of data contained in the database is from field operation, there is sometest data presented that is the result of high temperature operating life tests performed by either partmanufacturers or as a result of specialized research projects. Although such tests are notrepresentative of actual usage conditions, it is often the only failure data available for state-of-the-art components and/or low usage part types which have not had sufficient time to accumulate fielddata. This was the case for GUNN diodes, GaAs FETs, and laser diodes. Often such data is usedto app-oximate operating conditions by assuming it represents a ground benign environment atelevated temperatures. The temperature effects are extrapolated to typical operating temperaturesby assuming that the Arrhenius relationship is valid.

    The Arrhenius relationship models the reaction rate of discrete semiconductor failuremechanisms within a specific temperature range. The Arrhenius model is based on empirical dataand predicts that the rate of a given chemical or physical reaction, in this case a failure mechanism,will be exponential with the inverse of temperature. Conceptually, the Arrhenius model is givenby:

    Reaction Rate o exp (-Ea/KT)

    where:

    Ea = activation energy (eV) (Every chemical reaction has a unique activationenergy associated with it)

    K = Boltzman's constant

    = 8.63 x 10- 5 (eV/0 K)

    T = temperature ('K)

    The observed failure rate at an elevated temperature can be extrapolated to usageconditions as follows:

    3

  • X2= -, exp [ -Ea/K (l/r - 1/T2)]

    where:

    X.2 = usage failure rate estimate at T2

    X-1 = test failure rate observed at T1

    a = activation energy

    K = Boltzman's constant

    T2 = usage operating temperature (OK)

    T1 = test temperature ('K)

    The caveat with this approach is that each failure mechanism has a unique activationassociated with it. Furthermore, during the life of discrete semiconductor components, there canbe several such mechanisms proceeding simultaneously, each capable either individually or jointlyof causing a part failure. Reference 15 presents one approach to dealing with this discrepancy.

    It has been found, however, that for general classes of components with similar failuremechanism distributions the cumulative effects of the various reactions can be approximated by asingle Arrhenius model for a specified temperature range. This relationship has been designed asthe "equivalent Arrhenius relationship." This is the approach used in generic failure rate predictionmodels such as those in MIL-HDBK-217E.

    Section 5.0 is a detailed description of the sources of data for this publication.

    4

  • C) C- r - oc

    C'J' 6

    W1

    00

    ~~, WI) ,,I, V:l I

    QQ< ~ W)

    octo

    Q~tuC) Sr l~ S h * vg

    I I I0 .0

    04 od u >

  • 2.0 DATA ANALYSIS

    2.1 DATA SUMMARIES

    The table and figures presented in this section are summaries of the data presented in thedetailed data section, Section 4.0. Only part types for which there was a sufficient quantity ofdata were included in the summaries. In general this excludes low population parts such asmicrowave devices, most optoelectronics, etc.

    Table 2 presents generic component failure rates for the five predominant data sources whichprovided data for this publication. The five sources are described in detail in Section 5.0, DataSources. Briefly, Source 1 data is a conglomerate of data from all 16 MIL-HDBK-217 avionicenvironments, Sources 2 and 3 data are from an airborne, inhabited fighter environment, Source 4data is from a naval submarine application, and Source 5 data is collected from a number oflaboratory equipments in a controlled environment. Source 1 through 4 data contains militarygrade components (i.e., JANTX or JAN screen level and hermetically sealed) and Source 5 datacontains unscreened parts, predominantly plastic encapsulated. The number of data pointsreflected in the failure rate entries is in parenthesis next to the failure rate to allow the user toexercise judgment when evaluating the data.

    Figure 1 is a comparison of diode failure rates for the purpose of design trade-off decisions.Only data from the airborne data sources was used for this comparison in an attempt to minimizeany differences introduced by environmental and application-type effects. Figure 2 presents similarinformation for transistors.

    Figures 3 and 4 compare the average diode and transistor failure rates respectively for the fivedata sources in Table 1. It is evident that components from data Sources 2 and 3, namely the F- 16HUD and F-16 FCC, are consistently higher failure rate items. When the aspects of the variousdata sources are examined closely, this is not surprizing. The F-16 equipments experiencerelatively higher levels of vibration and shock than the other equipments (particularly fromSources 4 and 5). Additionally, fighter aircraft equipment, in particular, experience high levels ofpower and thermal cycling including both a greater quantity of cycles and much shorter cyclingintervals. Furthermore observed failure rates for similar devices can differ for several reasonsincluding the natural variability in the data, different data collection practices, equipment design,and both component and equipment manufacturing and screening procedures. Data Source 5failure data, as explained in Section 5.0, is first year warranty data from commercial applications.As such it is important to point out that long term failure mechanisms such as moisture andcontaminant related failure mechanisms of plastic encapsulated devices are not reflected in the data.

    It should also be noted that in addition to the variation in equipments, there are many sourcesof variability in the reliability data collected at the part level. Some of this variability is due to:

    * Inaccuracies in failure reporting* Inconsistencies in the criteria used to detect failure• Quality differences between manufacturers* Random, induced failures appearing as inherent failures

    RAC has attempted to make a homogenius dataset from all the various sources by choosingonly confirmed failures that were not induced, however the process to combine these sourcescannot be 100% effective. One predominant source of reliability variability in discretesemiconductor devices is due to the various types of interconnects used in a particular packagestyle. For example, DO-7 packages with whisker/S-bend /C-bend construction exhibit intermittentfailures when exposed to shock and vibration. Additionally, DO-35 packages with silver "button"construction exhibit intermittent and opens due to plastic deformation of the button. These are

    6

  • nonscreenable failure modes and unfortunately the particular construction for each package style isnot known and cannot be reported in the data contained in this databook.

    7

  • AVERAGE FAILURE RATE COMPARISON BY DIODE TYPE

    (AIRBORNE ENVIRONMENTS ONLY)

    1.4

    1.0

    FAILURE RATE 08lFailures/1 0E6

    flours)0.6

    0.4

    0.2_

    Small Signal Rectifier F.R. Rectifier Zemer Varistor/ TunnelCOMPONENT TYPE Suppressor

    Reliability Analysis Center. Griffiss AFB, NY 13441

    FIGURE 1: AVERAGE FAILURE RATE COMPARISON BY DIODE TYPE

    AVERAGE FAILURE RATE COMPARISON BY TRANSISTORTYPE

    (AIRBORNE ENVIRONMENTS ONLY)

    0.9

    0.8

    0.7

    0.6

    FAILURE RATE 0.5(Failures/10E6

    flours) 0.4

    0.3

    0.2

    0.1

    0"

    Bipolar < 5W Bipolar, > = 5W Field Effect DarlingtonCOMPONENT TYPE

    Reliability Analysis Center, Griffiss AFB, NY 13441

    FIGURE 2: AVERAGE FAILURE RATE COMPARISON BY TRANSISTORTYPE

    8

  • DIODE* AVERAGE FAILURE RATE BY DATA SOURCE

    1.9 Mil. Airborne Mil. Naval Comm. Laboratory

    1.6

    1.4

    1.2

    FAILURERAtIE

    (Falltures/ 1OE6 1-0Hours)

    0.8

    0.6

    0.4

    0.2

    0

    1 2 3 5

    DATA SOURCE

    SExcluding Microwave Devices

    Reliability Anaysis Cenre, Gnfiss AFB, NY 13441

    FIGURE 3: DIODE AVERAGE FAILURE RATE BY DATA SOURCE

    TRANSISTOR* AVERA4GE FAILURE RATE BY DATA SOURCE6.0

    Mil. Airborne Mil. Naval Comm. Laboratory

    5.0

    4.0

    FAILURERATE,

    (Fa ilures/10E6,,ours) 3.0

    2.0

    1 .0

    01 2 345

    DATA SCIURCEExcluding Micrx-save Dcevices

    Reia.bility Analy.i. C-sta. 0r-M, AFlS. NXV 13441

    FIG;URE 4: TRANSISTOR AVERAGE FAILURE RATE BY DATA SOURCE

    9

  • 2.2 VALIDITY OF THE EXPONENTIAL FAILURE ASSUMPTION

    The primary purpose of failure rate prediction models for electronic components such asthose in MIL-HDBK-217E is to estimate the reliability of electronic equipment and systems. Suchfailure rate prediction models are based upon the assumption of an exponential time-to-failuredistribution, which assumes a constant failure rate over time.

    However, most failure mechanisms of discrete semiconductor devices investigated in theliterature reportedly follow a log-normal failure distribution. An investigation of such data wasconducted to determine the validity of an exponential approximation in this light. The analysispresented in this section was performed by I.I.T. Research Institute under contract to Rome AirDevelopment Center (contract number F30602-85-C-0131).

    There are many practical reasons why the assumption of a constant failure rate is preferred toa time-dependent failure rate for MIL-HDBK-217E type failure rate prediction models.

    Simplicity

    The mean time between failures (MTBF) of a system whose component parts exhibitconstant failure rates is not time dependent. Alternatively, for a system made up ofcomponents having nonconstant failure rates, the system MTBF will be time-dependentand is therefore undefined unless a particular mission time is specified. The assumptionof exponentiality allows for failure rates to be summed in a series reliability network.

    • Precedent

    The exponential assumption is currently used for the electronic components in acceptedmodels such as those in MIL-HDBK-217E.

    " Data Availability

    If any distribution other than exponential is assumed, the parameters of the distributionmust be determined by analysis of cumulative time-to-failure data. This detailedinformation is seldom available for field data sources. The exponential distributionallows population parameter estimates to be made based upon total part operating hoursand total number of failures.

    • Accuracy

    When developing models such as those employed in MIL-HDBK-217E, anyimprovement in model accuracy resulting from the use of a more complex distribution(than exponential) may be insignificant when compared to the inherent variabilityassociated with reliability prediction and the "statistical noise" in the data.

    An analysis of constant versus time-dependent failure rate distributions was undertaken usingobserved time-to-failure data collected for this databook. However, due to the above-mentionedadvantages, it was predetermined that if discrete semiconductor failure rate prediction models couldbe established as accurately with an exponential time-to-failure distribution as by a log-normal orother time dependent failure distribution, the former is preferable.

    The following paragraphs describe the analysis procedure followed.

    All time-to-failure data for discrete semiconductor components was extracted from theavailable literature. This data consisted of life test results at high temperatures. Ideally, it would

    10

  • have been preferable to analyze time-to-failure field data since such data would more closelyapproximate the actual usage environments. However, such data is not available. Hightemperature life test time-to-failure data was available for the following device types:

    Low Noise GaAs FETsHigh Power GaAs FETsGeneral Purpose Transistors (NPN & PNP)GaAs Laser DiodesIMPATF DiodesSchottky Diodes

    Weibull analysis was then applied. The Weibull distribution is particularly useful inanalyzing life data since (1) it has repeatedly been observed to provide a good fit to the data, and(2) it is a flexible distribution which can approximate many other statistical distributions,depending upon the value of 3, the shape parameter. Table 3 gives some shapes of the Weibulldistribution which approximate other common distributions. The form of the Weibull distributionvaries between texts, but a common one is given by the probability density function:

    t3-1t= ~exp (-(-)

    where

    f(t) = Weibull probability density functiona = scale parameter (characteristic life)P = shape parametert = time

    TABLE 3: WEIBULL SHAPE PARAMETERS (REFERENCE 1)

    Shape Parameter. Distribution Type

    3< 1 Gamma (k < 1)3= 1 Exponential3=2 Rayleigh3 3.44 Normal (approx.)

    Twenty-one individual data sets were plotted on Weibull probability paper, and the value of3 was determined. Figures 5 through 25 illustrates the Weibull plots of the data. The results ofthis step of the analysis were encouraging since, as can be seen from the plots, the values of pseemed to center around 1.0. Table 4 presents of a summary of the best fit Weibull parameters.

    The next step of the analysis was to force the best line with 3 = 1.0 through the observed datapoints. This is also illustrated in the figures. The Kolmogorov-Smirnov (K-S) goodness-of-fittest was then applied to the forced line. The intent of this step was to determine the degree of errorresulting from the exponential assumption. To apply the test, the value of the D statistic, thelargest deviation of observed from expected or theoretical value is compared to standard tables ofcritical values at some predetermined level of significance (in this case 0.2). If D exceeds thecritical value, it can be concluded that the observations do not fit the theoretical distribution.

    11

  • 999LJn (TiME/RESS)

    10.0

    'J 1.0

    .1

    0.1~~~~~~~~ 0_______________

    10 102 103 104

    TIML/STRESS AXIS (HOURS)

    FIGURE 5: WEIBULL PLOT OF HIGH TEMPERATURE OPERATING LIFEDATA FOR GaAs FETs

    Iii (TIM E/STRE SS)

    10.0

    * 101 0 o

    0.12

  • 10.0

    - UST FIT 2S~

    10 1004 losTIME/STRIESS AIS (HOURNS)

    FIGURE 7: WEIBULL PLOT OF HIGH TEMPERATURE OPERATING LIFEDATA FOR GaAs LASER DIODES

    Ito

    WSf 14T (3 s

    102 101 104 JosTIULvSTRISS AXIS (HOURS)

    FIGURE 8: WEIBULL PLOT OF HIGH TEMPERATURE OPERATING LIFEDATA FOR GaAs LASER DIODES

    13

  • 11' (lMI/STRE SS)

    S S RT

    1 102 103TfM IITRESS AXIS (HOURS)

    FIGURE 9: WEIBULL PLOT OF HIGH TEMPERATURE OPERATING LIFEDATA FOR GaAs LASER DIODES

    TIMILTiMEESSSTRHOSSS

    1014

  • 999 gn (TIMEJSTRE 5$)

    100

    I~ o

    10 102 101 104

    T1MISTRISS AXIS (HOWRS)

    FIGURE 11: WEIBULL PLOT OF HIGH TEMPERATURE OPERATING LIFEDATA FOR GaAs FETs

    ".9

    10.0

    S 10 016 0

    0.15

  • 9,.,TRSS

    10.0

    - fT RT

    to 102 103 104TIMh/STRESS AXIS (HOURS)

    FIGURE 13: WEIBULL PLOT OF HIGH TEMPERATURE OPERATING LIFEDATA FOR HIGH POWER PULSED IMPATT DIODES

    99. g (TIME/STRESS)

    'a10.0

    *

    .0

    10102 102 104TIMEMNTESS AXIS (HOURS)

    FIGURE 14: WEIBULL PLOT OF HIGH TEMPERATURE OPERATING LIFEDATA FOR HIGH POWER PULSED IMPATT DIODES

    16

  • ".9Li' (TIMEPSTRE SS)

    0.0

    cc 1.0

    0.1

    8 ST FIT

    0.01 ___ _____ ________ _102 101 104 105 104

    TIMUiSTUESS AXIS (HOURS)

    FIGURE 15: WEIBULL PLOT OF HIGH TEMPERATURE OPERATING LIFEDATA FOR (AlGa) AS LASERS

    99.9~ tnTIME/STRESS)

    10.0

    010

    - ST AIt

    010 0 10410 103 TIME/STRESS AXIS (HOURS) 0

    FIGURE 16: WEIBULL PLOT OF HIGH TEMPERATURE OPERATING LIFEDATA FOR IN GaAs P/In P DHI LASERS

    17

  • In (1MUSTRESS)

    10.0

    0.1

    102 101 104 losnummRss AXIS (HOURS)

    FIGURE 17: WEIBULL PLOT OF HIGH TEMPERATURE OPERATING LIFEDATA FOR JAN 2N918

    99.5mmss

    10.0

    0.01 1________________102 IL0. 105

    T1MfISTRESS AXIS (HOURS)

    FIGURE 18: WEIBULL PLOT OF HIGH TEMPERATURE OPERATING LIFEDATA FOR GaAs POWER FETs

    18

  • Ln {TIMUSTRES$)

    ,o0 - __ _ _ _ __ _ _ _ _ __ _ _ __ _ _ _ _ __ _ _ _ _

    - 1.0

    0.1

    -. If qr IT-...• OB I tH:| I3J.O

    0 01 102 10V 10' 10S

    TIME/STRESS AXIS (HOURS)

    FIGURE 19: WEIBULL PLOT OF HIGH TEMPERATURE OPERATING LIFEDATA FOR LOW NOISE GaAs FETs

    99.9 ILn (TIMrISTRES$) -- s

    10.0 -

    0.1

    M71

    - F0 l4KIfO Wl4I !

    0.01 10 I0N 10 lOS

    TImIL,*TInSS AXIS (HOURS)

    FIGURE 20: WEIBULL PLOT OF HIGH TEMPERATURE OPERATING LIFEDATA FOR LOW NOISE GaAs FETs

    19

  • 9 1.9 &A (TIME/STRIESS)

    l00

    - 1.0

    ta

    0.1

    GISTIT I0

    0.01102 101 104 1os

    TiME STRESS AXIS (HOURS)

    FIGURE 21: WEIBULL PLOT OF HIGH TEMPERATURE OPERATING LIFEDATA FOR LOW NOISE GaAs FETs

    99.9

    99g tn (TIM ESTRISS) . '

    10.0

    S1.0

    0.t

    mIST PITj.1 iSm IST ut

    0.01 _________ _________ _________10 10, 103 104 10, 104

    TIMI/STRESS AXIS (HOURS)

    FIGURE 22: WEIBULL PLOT OF HIGH TEMPERATURE OPERATING LIFEDATA FOR LOW NOISE GaAs FETs

    20

  • tn (TIME/STRESS)

    10.0

    zS 1.0

    0.1

    -. SET FIT

    0.01 ________________ _______________102 l0i 10' Jos

    TIME/STRESS AXIS (HOURS)

    FIGURE 23: WEIBULL PLOT OF HIGH TEMPERATURE OPERATING LIFEDATA FOR LOW NOISE GaAs FETs

    ".9

    '0.1

    .010 a 0 U

    TKITESAI HUS

    FIUE2:WwULPO FHGHTMEAUEOEAIGLFDAAFRLWNOS asF.

    0.1 ____ ____ ___ ____ _____21

  • 9999,Ln(T1ME/STRESS)f-

    I--zs

    0.1

    i~~r Ar~ I4

    f-K-. VAs I N 1

    0.01

    10 10,1 10 104 105TIMIFSTRfSS AXIS (HOURS)

    FIGURE 25: WEIBULL PLOT OF HIGH TEMPERATURE OPERATING LIFEDATA FOR LOW NOISE GaAs FETs

    22

  • Otherwise, it can be assumed that the observed distribution is not sirnificantly different from theexp\,nential model. None of the data sets was significantly different from the exponential model at20% significance. This implies that the available data does not indicate deficiencies with theexponential assumption. The results of the Kolmogorov-Smimov test (KS) are presented in Table5.

    TABLE 4: OBSERVED WEIBULL PARAMETERS

    Figure # Ref# Temperature (°C)() D" 2 200 (Tc) 1.15 600

    6 3 70 (Tc) .69 4,400

    7 3 55 (Tc) 1.25 8,000

    8 3 70 (Tc) .82 5,200

    9 3 70 (Tc) .95 230

    10 3 70 (Tc) .57 10,000

    11 4 245 (Tc) 1.10 950

    12 4 231 (Ta) 1.60 580

    13 5 90 (Tc) 220 (Tj) 1.15 1,300

    14 5 90 (Tc) 220 (Tj) .75 2,000

    15 6 70 (Ta) .87 8,000

    16 7 20 (Tc) 1.05 6,000

    17 8 300 (Tj) 1.60 4,000

    18 9 228 (Tj) 1.00 2,700

    19 10 200 (Ta) 1.20 1,600

    20 10 200 (Ta) .73 1,500

    21 10 220 (Ta) 1.00 500

    22 10 220 (Ta) 1.32 700

    23 10 85 (Ta) .85 3,100

    24 10 120 (Ta) 1.00 2,100

    25 10 240 (Ta) 1.46 1,200

    NOTES: (1) Tc = case temperature

    Ta = ambient temperatureTj = junction temperature

    Reliability Analysis Center, Griffiss AFB, NY 13441

    23

  • Otherwise, it can be assumed that the observed distribution is not significantly different from theexponential model. None of the data sets was significantly different from the exponential model at20% significance. This implies that the available data does not indicate deficiencies with theexponential assumption. The results of the Kolmogorov-Smirnov test (KS) are presented in Table5.

    TABLE 4: OBSERVED WEIBULL PARAMETERS

    Figure # Ref # Temperature ('C)(1)

    5 2 200 (Tc) 1.15 600

    6 3 70 (Tc) .69 4,400

    7 3 55 (Tc) 1.25 8,000

    8 3 70 (Tc) .82 5,200

    9 3 70 (Tc) .95 230

    10 3 70 (Tc) .57 10,000

    11 4 245 (Tc) 1.10 950

    12 4 231 (Ta) 1.60 580

    13 5 90 (Tc) 220 (Tj) 1.15 1,300

    14 5 90 (Tc) 220 (Tj) .75 2,000

    15 6 70 (Ta) .87 8,00016 7 20 (Tc) 1.05 6,000

    17 8 300 (Tj) 1.60 4,000

    18 9 228 (Tj) 1.00 2,700

    19 10 200 (Ta) 1.20 1,600

    20 10 200 (Ta) .73 1,500

    21 10 220 (Ta) 1.00 500

    22 10 220 (Ta) 1.32 700

    23 10 85 (Ta) .85 3,100

    24 10 120 (Ta) 1.00 2,100

    25 10 240 (Ta) 1.46 1,200

    NOTES: (1) Tc = case temperatureTa = ambient temperature

    Tj = junction temperature

    Reliability Analysis Center, Griffiss AFB, NY 13441

    23

  • TABLE 5: D STATISTIC TEST RESULTS'

    # of Maximum D Statistic (0.2Fi -ure# Ref.# Failures Deviation Significance Level) Conclusion

    52 4 .022 .494 Fits 3=1.06 3 5 .078 .446 Fits 3=1.0

    7 3 5 .054 .446 Fits 3=1.08 3 6 .200 .410 Fits 3=1.0

    9 3 7 .083 .381 Fits 3=1.010 3 4 .116 .494 Fits 3=1.0

    11 4 9 .090 .339 Fits 3 = 1.012 4 7 .227 .381 Fits 3=1.013 5 11 .055 .323 Fits 3=1.0

    14 5 15 .231 .276 Fits 3=1.0

    15 6 74 .118 .124 Fits 3=1.016 7 7 .140 .381 Fits 3=1.0

    17 8 13 .025 .297 Fits 13 = 1.018 9 4 .080 .494 Fits 13 = 1.019 10 11 .250 .323 Fits 1=1.0

    20 X) 13 .040 .297 Fits 1=1.0

    21 10 15 .090 .276 Fits 1 = 1.022 10 14 .060 .274 Fits 1=1.0

    23 10 11 .090 .323 Fits 3=1.0

    24 10 16 .190 .258 Fits 3=1.0

    25 10 10 .250 .322 Fits 13 = 1.0

    Reliability Analysis Center, Griffiss AFB, NY 13441

    Based on the results of the K-S test, it was assumed that the failure distributions of thesemiconductor devices analyzed could be described by a Weibull distribution with a slope of 1.0.Assuming anything other than a constant failure rate would introduce unnecessary complexity intothe prediction models. The observed time-to-failure distributions were accurately represented byan exponential distribution over the range of variables in the data. Additionally, as many differenttime-dependent failure mechanism distributions are summed (since a device can be susceptible toseveral failure mechanisms at one time) an exponential distribution often results.

    24

  • Time-to-failure data was not available for all discrete semiconductor device types. It wastherefore reasonable to assume that the times-to-failure of other discrete semiconductor part typeswould follow the same distribution. There was no evidence in the literature that any discretesemiconductor part types should differ from those with available times-to-failure. Furthermore,tho part types analyzed represent a diverse cross-section of all part types, since they include bothField Effect and Bipolar devices and members from the transistor, diode. and optoelectronicgroups.

    2.3 DSR-3/DSR-4 DATA COMPARISON

    Significant design and processing technology advances have been made over the past 10-15years in the semiconductor industry. To examine the influence of these changes on semiconductordevice failure rates, the data in this publication was compared with data from similar applicationsand part types in DSR-3, its predecessor. The results of this comparison are presented in Table 6.Overall there seems to be a general improvement in observed failure rates. The three cases whereDSR-4 failure rates are worse than the corresponding DSR-3 number represent part types forwhich the failure rate should probably have stabilized and the result is simply a statisticalaberration. Of particular interest is the DSR-3 to DSR-4 commercial device data comparison, sincethese values are from the same data sources thereby negating some of the variations due to datacollecting practice, and application-related effects.

    Table 6: DSR-3* vs. DSR-4** Component Failure Rate

    (failures/10 6 hours) Comparison

    DSR-3 DSR-4 DSR-3 DSR-4Device Type Airborne Airborne Ratio Commercial Commercial Ratio

    DIODESmall Signal Switching 0.046 0.085 0.53 0.038 --- ---Rectifier 5.833 0.2127 27.41 0.1165 0.056 2.06Rectifier, Fast Recovery 0.4629 0.2671 1.73 --- ---Zener 1.4905 0.1433 10.32 0.1772 0.123 1.43Current Regulator --- --- --- --- 0.147 - - -Varistor/Suppressor 34.4827 1.2225 28.30 --.- --- -Tunnel --- 0.2511 --- --- 0.207 - - -Varactor --- --- -- - 0.0886 0.235 0.37

    TRANSISTORBipolar, < 5W 2.6001 0.3771 6.89 0.2234 0.092 2.40Bipolar, >=5 11.9134 0.8475 14.05 0.7242 0.272 2.66Field Effect 8.9285 0.3953 22.59 0.4714 0.189 2.49Unijunction --- --- 0.3831 0.307 1.24Darlington -- - 0.8120 - - - 0.2657 ---.. - - .

    THYRISTOR 1.2432 4.6824 0.26 0.6007 0.175 3.41OPTOELECTRONIC

    LED ....-. ... 0.1656 0.003 46.09

    * DATA VINTAGE 1968-1978** DATA VINTAGE 1976-1987

    - - - INSUFFICIENT DATAReliability Analysis Center, Griffiss AFB, NY 13441

    25

  • 2.4 DERATING EFFECTSThe benefits of electrical parameter derating on discrete semiconductor device reliability

    have been well documented (Ref. 11, 12). Since data collected on the ARN-118 wascomplemented with a detailed MIL-HIDBK-217 part stress analysis performed on that sameequipment, electrical stress levels were available for bipolar transistors and diodes. Theknowvhdge of electrical stress levels combined with detailed part-levei failure experience datafa,'il itated the analosis Of Vcc derating for bipolar transistors and VR derating for small signal andrectifier diodes.

    The analysis was performed by linear regression of failure rate against voltage stress.Tran.,fornations , ere performed on the failure rate and stress variables to provide results in one oft,,vo forms:

    Failure rate - exp (a1 Vs)or

    Failure rate - (vs)a2

    where at and a- are regression constants and Vs is the applied voltage stress level. The form withthe best R- value for e,1ch device type would be determined from the analysis. The effects oftemperature and envirc.iment, which are not the samne for all data within the ARN- 118 dataset sinceit is from several aircrafts, were normalized out assuming current MIL-HDBK-217E factors to becorrect.

    Voltage stress indeed proved to have a significant effect on failure rate for high power(>SW) bipolar transistors. This effect was not evident in the low power bipolar transistor datacollected. However, from a theoretical perspective, derating should also benefit low powerdevices. Most likely, the comparatively lower failure rates and the associated higher failure ratevariability in the data prevented the confinnation of such a factor. A relationship between voltagestress and failure rate is also evident in the data for signal and rectifier diodes.

    The results of the regression analysis for transistors is:

    Failure rate - exp (3.1Vs) R2 = .40 [0 < Vs: 1.01where Vs is Vci\'ceo. The lower and upper 95% confidence interval values about "ai" are 1.72

    and 4.56 respectively. This relationship is depicted in Figure 26.

    The result for small signal and rectifier diodes is:

    Failure rate o (V,) 2-4 R2 =.45 10V < 1.01

    where Vs is the reverse voltage stress. The lower and upper 95% confidence values Thout "a2" are.96 and 2.43 respectively. This relationship is depicted in Figure 27.

    Although insufficient data was available to analyze the effects of derating for device typesother than bipolar transistors, small signal diodes and rectifiers, it is anticipated that givensufficient data for other parts in this publication, the benefits of derating would be furtherexemplified. The relationships presented are only valid within the range of stresses found in thedata analyzed. This would include stress values from .1 to 1 0, and would not include overstresscon(ditions.

    26

  • Failure Rate(p.)

    (Arbitrarylinearscale)

    0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

    (V s ) Voltage Stress Ratio VCE

    VCEO

    Re'iability Analysis Center, Griffiss AFB, NY 13441

    FIGURE 26: FAILURE RATE vs. VOLTAGE STRESS FOR HIGH POWERBIPOLAR TRANSISTORS

    Failure Rate

    Vs2.4(A,,) Xos V

    (Arbitrarylinearscale)

    0.1 02 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

    Voltage Stress (Vs) Ratio

    Reliability Analysis Center, C3iffiss A'113, NY 13441

    FIGURE 27: FAILURE RATE vs. VOLTAGE STRESS FOR SIGNAL ANDRECTIFIER DIODES

    27

  • 3.0 FAILURE MODE AND MECHANISM DATA

    This section presents quantitative distributions of failure modes and failure mechanisms fortransistors, diodes, liquid crystal displays and photovoltaic modules.

    Failure mode here refers to the externally detectable effect of a part failure. Failuremechanism generally refers to the chemical or physical process which caused the part to fail. Anextended definition is used herein, due to limitations in the available data, which includes thephysical location of the defect or failure within a device.

    The transistor and diode data presented in Tables 7 through 10 and Figures 28 and 29 hasbeen compiled from reports of reliability demonstration tests conducted in accordance with MIL-STD-78 1. All testing and device failures were therefore conducted and evaluated against similarcriteria. Only primary, verified failures were included in the data. Additionally no failuresresulting from mishandling or overstress conditions were included. Data from reliabilitydemonstration test reports included in DSR-3 which met the above criteria was included along withnew data collected since that publication.

    Insufficient detail was available to determine if the individual parts represented in theseanaivses were hermetically sealed or plastic encapsulated. Differences in distributions of failuremodes and mechanisms would be expected depending upon many factors, including hermeticity.However, since all diode and transistor failure mode/mechanism data was taken from militaryequipments, it can be assumed that the vast majority of the parts were hermetically sealed.

    Table 11 and Figure 30 present failure mode and mechanism data on a liquid crystaldisplays. This data should be viewed with scrutiny in comparison with other similar parts since itis from a single, reliable data source on a single part type. This data was included here because ofthe general lack of such information within the industry. However, significantly more data mustbe collected before any general conclusions can be made about the failure behavior of liquid crystaldisplays.

    Table 12 and Figure 31 present available failure mode data on photovoltaic cell modules.This data was compiled from 5 experimental test sites across the United States in a programsponsored by the United States Department of Energy (Reference 13) and used with permission ofIEEE.

    The information in these tables and figures, particularly those relating to failure modes, willbe beneficial to engineers performing Failure Modes, Effects and Criticality Analysis (FMECAs) inthe evaluation of system designs.

    TABLE 7: DIODE FAILURE MODES

    NORMALIZEDMODE QUANTITY % %

    Open 12 20.7 23.5

    Short 10 17.2 19.6

    Degraded 8 13.8 15.7

    Intermittent 21 36.2 41.2

    Unknown 7 12.1 - -

    Reliability Analysis Center, Griffiss AFB, NY 13441

    28

  • DIODE FAILURE MODE DISTRIBUTION

    Unknown 12.1%Short 17.2%

    Open 20.7%

    Intermittent 36.2%

    Degraded 13.8%

    Reliability Analysis Center, Griffiss AFB, NY 13441

    FIGURE 28: DIODE FAILURE MODE DISTRIBUTION

    TABLE 8: DIODE FAILURE MECHANISMS/LOCATION

    MECHANISM/ NORMALIZED

    LOCATION QUANTITY % %

    Wire Bond 2 5.6 8.7

    Broken Wire 1 2.8 4.3

    Contamination 2 5.6 8.7

    Cracked Die 4 11.1 17.4

    Cracked Glass Case 10 27.8 43.5

    Lifted Wire 1 2.8 4.3

    Package 1 2.8 4.3

    Whisker Alignment 2 5.6 8.7

    Unknown 13 36.1 - -

    Reliability Analysis Center, Griffiss AFB, NY 13441

    29

  • TABLE 9: TRANSISTOR FAILURE MODES

    NORMALIZEDMODE QUANTITY % %

    Open 49 52.1 56.3

    Short 17 18.1 19.5

    Degraded 14 15.0 16.1

    Intermittent 7 7.4 8.0

    Unknown 7 7.4 - -

    Reliability Analysis Center, Griffiss AFB, NY 13441

    TRANSISTOR FAILURE MODE DISTRIBUTION

    S~hort 18.1%

    Unknown 7.4%

    Open 52.1%

    Intermittent 7.4%

    Degraded 15.0%

    Reliability Analysis Center, Griffiss AFB, NY 13441

    FIGURE 29: TRANSISTOR FAILURE MODE DISTRIBUTION

    30

  • TABLE 10: TRANSISTOR FAILURE MECHANISM/LOCATION

    MECHANISM/ NORMALIZED 1LOCATION QUANTITY%%

    Wire Bond 26 27.7 43.3

    Die 7 7.4 11.7

    Contamination 1 1.1 1.7Lifted Die 1 1.1 1.7Metallization 14 14.9 23.3

    Oxide Defect 3 3.2 5.0Interconnect 8 8.5 13.3

    Unknown 34 36.1--

    Reliability Analysis Center, Griffiss AFB, NY 13441

    31

  • TABLE 11: LIQUID CRYSTAL DISPLAY FAILURE DISTRIBUTION

    FAILURE LOCATION ANDASSOCIATED MODES QUANTITY %

    Conductive Epoxy InterconnectDim Rows 44 33.3Flickering Rows 22 16.7

    Missing Rows 5 3.8Integrated Circuit

    Blank Display 9 6.8Improper Characters 16 12.1

    Zebra Strip

    Flashing Dots 5 3.8Missing or Dim Columns 11 8.3

    Miscellaneous Manufacturing Defects 20 15.2

    Reliability Analysis Center, Griffiss AFB, NY 13441

    LIQUID CRYSTAL DISPLAY FAILURE MODE DISTRIBUTION*

    Flickering Rows 16.7%

    Dim Rows 33.3%

    Missing Rows 3.8%

    Improper Characters12.1%

    Missing or Dim Columns8.3%

    . Misc. Mfg. Defects 15.2%

    Flashing Dots 3.8%Blank Display 6.8%

    *Based on testing of 3600 32-Character (5x7 Dot Matrix) Liquid Crystal Display Reliability Analysis Center, Griffiss AFB, NY 13441

    FIGURE 30: LIQUID CRYSTAL DISPLAY FAILURE MODE DISTRIBUTION

    32

  • TABLE 12: PHOTOVOLTAIC CELL MODULES FAILURE DISTRIBUTION

    FAILURE TY-PE QUANTITY%

    Interconnect Fractures 64 1 8.4

    Unsoldered Interconnects 30 8.6

    Cracked Cells 117 33.6Dielectric Breakdown 31 8.9

    Encapsulant Delamination 77 22.1

    Wire and Terminal Corrosion 21 6.1

    Exposed Interconnects 8 2.3

    Reliability Analysis Center, Griffiss AFB, NY 13441

    PHOTOVOLTAIC CELL ARRAY FAILURE MODE DISTRIBUTION*

    Exposed Interconnects2.3%7

    EncapsulantDelamiunation 22.1%

    Cracked Cells 33.6%

    Interconnects 8.6%

    Wire and TerminalCorrosion 6.1%.....

    Interconnect FracturesDielectric Breakdown 18.4%

    8.9%

    (C) 1982 wLF ith permission from Dumnes. L. and S. Shumnka. "PhotovoltaicModule Reliability Improvemenrts Through Application Testing and Failure Analysis.Rla"iyAayi eerGifs ~I Y 1341FF Transactions, on Reliability. Vol R-3, No. 3, August 1982. p. 234ReibltAn yssC trG fi sAI ,N 134

    FIG;URE 31: PHOTOVOLTAIC CELL ARRAY FAILURE MODE DISTRIBUTION

    33

  • 4.0 DETAILED DATA

    This section presents the detailed failure rate data for diodes, transistors, thyristors andoptoelectronics, in that order. Descriptions of the various column entries are given in Section 1.0,"User's Guide to DSR-4." Each of the four part categories (diodes, transistors, thyristors andoptoelectronics) are broken down into subcategories which further define the part type. Forexample, under the diode listing are: small signal switching, small signal general purpose,rectifier, fast recovery, etc. These subcategories are listed below. Within each subcategory, partlevel data is listed in order by test type (life then field) and then by generic part number.

    The data fields and associated field values in the detailed data section are explained below:

    DiodeSmall Signal, SwitchingSmall Signal, General PurposeRectifierRectifier, Fast RecoveryRectifier, Bridge, Full WaveZenerZener, Voltage RegulatorZener, Voltage ReferenceCurrent RegulatorTransient SuppressorMicrowave, TunnelMicrowave, Schottky BarrierMicrowave, PINMicrowave, VaractorMicrowave, GUNN EffectMicrowave, IMPAT

    TransistorBipolar, NPNBipolar, PNPField Effect, Junction, N-channelField Effect, Junction, P-channelField Effect, MOS, N-channelField Effect, MOS, P-channelUnijunctionMicrowave/RF, Field EffectMicrowave/RF, BipolarMultiple, Complementary PairMultiple, DarlingtonMultiple, Matched PairSpecial Function, Chopper

    ThyristorThyristorSCRTRIACTriode, Trigger

    34

  • OptoelectronicEmitter, Single LEDEmitter, Single LED, InfraredEmitter, LED ArrayEmitter, Laser DiodeSensor, PhotodiodeSensor, PhototransistorPhotocoupler,Photocoupler, Phototransistor OutputPhotocoupler, Photodarlington OutputPhotocoupler, IC OutputAlphanumeric Display, LED, Segment TypeAlphanumeric Display, LED, Array TypeAlphanumeric Display, Liquid CrystalPhotovoltaic Module

    The data fields and associated field values in the detailed data section are described below:

    Field Description

    Part Number The basic device part number plus suffix. "--" indicatesunknown part number.

    Slash Number The MIL-slash number according to MIL-S-19500, MilitarySpecification: Semiconductor Device, General Specificationfor. "--" indicates an unknown MIL-slash number.

    Equipment Reference The system/equipment nomenclature from which the datawas obtained. All data sources are described in detail inSection 5.0. For test data, N/A refers to not applicable.

    Package Type The outline package reference designation. "--" indicatesunknown drawing number.

    Semi. Material The semiconductor material the component is fabricatedfrom:

    -- UnknownSi SiliconGe GermaniumGaAs Gallium-ArsenideGaP Gallium PhosphideGaAsP Gallium-Arsenide-PhosphideAlGaAs Aluminum-Gallium-Arsenide

    Quality Level A factor describing both the screen level and packagehermiticity of the device as described in MIL-S-19500including: JANTXV, JANTX, JAN, Lower (CommercialHermetic) and Plastic (Commercial Plastic Encapsulated)"--" indicates unknown quality level. In addition, thefollowing quality levels were employed to furthercharacterize non-MIL-S-19500 part types:

    35

  • Field Description

    Quality Level (cont'd)JTXEQU - Screening equivalent to JANTX level for

    NON-QPL partsJANEQU - Screening equivalent to JAN level for

    NON-QPL parts

    HERMETIC - Hermetically packaged laser diode

    FACET COAT - Laser diode with Facet Coating andnonhermetic package

    App Env The MIL-HDBK-217E-type application environment codedas:

    Application Environment CodeUnknownGround Benign GBGround Fixed GFGround Mobile GMMan Pack MPNaval Sheltered NSNaval Unsheltered NUNaval Undersea Unsheltered NUUNaval Hydrofoil NHAirborne Inhabited AlAirborne Inhabited Cargo AICAirborne Inhabited Trainer AITAirborne Inhabited Bomber AIBAirborne Inhabited Attack AIAAirborne Inhabited Fighter AIFAirborne Uninhabited AUAirborne Uninhabited Cargo AUCAirborne Uninhabited Trainer AUTAirborne Uninhabited Bomber AUBAirborne Uninhabited Attack AUAAirborne Uninhabited Fighter AUFAirborne Rotary Winged ARWMissile Launch MLCannon Launch CLUndersea Launch ULMissile Free Flight MFFAirbreathing Missile Flight MFA

    Temp The temperature in degrees Centigrade associated with thedevice failure. An "A" following the entry indicatesambient temperature, "C" indicates case, and "J"indicates junction temperature. "--" indicates unknowntemperature.

    36

  • Field Description

    Rated Current The current rating in Amperes of the device. This value isgiven for diodes, thyristors and optoelectronic devices asfollows:

    Small Signal, Switching Diode, forward current (If)

    Small Signal, General Purpose Diode, forward current (If)

    Rectifier Diode, forward current (If)

    Zener Diode, dynamic impedance current (Iz)

    Thyristors, maximum static on-state current (If)

    LED, forward current (If)

    Laser Diode, forward current (If)

    Sensor, Photodiode, forward current (If)

    Sensor, Phototransistor, collector current (Ic)

    Photocoupler, forward current (If)

    "--" indicates unknown

    Rated Power The maximum rated power dissipation in Watts. This valueis given for transistors, and photovoltaic cell arrays.indicates unknown.

    Voltage Stress The applied divided by the rated voltage for the device.Stresses given are for the following parameters:

    Small Signal, Switching Diode, working peak reversevoltage (VRWM)

    General Purpose Diode, working peak reverse voltage(VRWM)

    Rectifier Diode, working peak reverse voltage (VRWM)

    "--" indicates unknown

    Current Stress The applied divided by the rated current for the device.Stresses are given for the following parameters:

    Small Signal, Switching Diode, average forward current (If)

    Small Signal, General Purposed Diode, average forwardcurrent (If)

    Rectifier Diode, average forward current (If)

    37

  • Field Description

    Current Stress (cont'd)Thyristor, maximum static on-state current (It)

    Bipolar Transistor, collector current (Ic)

    Field Effect Transistor, reverse gate current (Igss)

    Unijunction Transistor, emitter current (le)

    LED, forward current (If)

    Laser Diode, forward current (If)

    Sensor, Photodiode, forward current (If)

    Sensor, Phototransistor, collector current (Ic)

    Photocoupler, forward current (If)

    "--" indicates unknown

    Freq. Band Operating frequency for microwave devices codedas follows:

    Approximate Band (Code) Frequency (GH 7 )

    Unknown

    < 1GHz 250.0

    Duty Cycle Percentage of time the device is under electrical stress."--" indicates unknown duty cycle.

    Character Count The number of characters present in a optoelectronicalphanumeric display. "--" indicates unknown charactercount.

    38

  • Field Description

    Diode Count The number of LEDs comprising an optoelectronic array oralphanumeric display. "--" indicates unknown diode count.

    Voc Open circuit voltage for photovoltaic modules. indicatesunknown.

    Isc Short circuit current for photovoltaic modules. "--" indicatesunknown.

    Number Tested Quantity of parts under the described test or field conditionsfor that data point. "--" indicates unknown quantity tested.

    Number Failed The quantity of parts in the number tested column whichfailed.

    Part Hours The total number of parts tested multiplied by the totaloperating hours

    Failure Rate The total quantity of failures divided by the totalquantity of part hours for records with failures.Units are failures per 106 operating hours.

    39

  • - - CD 0.0(2 r - '0 Cr- r- O.O 0. 't0 , n CoU t p0(. '0t oj.t -l 0' - ' .0 0.0CDcoC 11, I o0 CD1 :: 00 , C. '0

    CD) C0 .)C D Q >C : )C ) C

    r'- g a, CD C000 0 00 ( 'o ~ -. t C> '~0 A (AN r- co -n ' -t r -'jo C 0''0 0 ( '0 C)C " - N- - C) '. 0. '0 0. 't 0. ro ~

    r- ' - 0D Nl 00j. N 0 (~~ a)O CD..c>I " OlL )P t 0 eN"N- w Ln

    Ln (N,- N( (Ao

    MU 0

    CL x

    L '0 , CD *I CD 0 (N N- (.1 ev 00 C ) C (D~(f( (A (A (A (ADf (A (A( DU 0 * . . ( N ( N ~ (N ( N (. N ( N (

    c >1' ) CD C C) 0 C D C CD C ) 0 ( C>C Q.C D C D CD ) C3 C C.C ) 0 C C )0 DC)C C C>CDC 0 000000000C C D0 0C DC DC DC DC n )C

    C) CD0=1-D''CD CD C(A D C DC o - D Q C D CD C >C ,C )( > D C

    c*i 0 ( A (A (N -0 0 0 0 0 0 0 0 0 0 0 0 0 0

    01 (~0

    -C1 L0 0 0K -C C ~ uz me = C C C -

    -1c -C-

    L.-

    - ) U - - 'Cn 1-A '- '(7 '(7 ' n I-A ',7 I-A ',7 'j '(7 'N- S.. I- I S- N-

    OL-j~ - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

    -2 4) c

    0 xU 000 -- 0 0 0 0 0 0 0 0 .- 0 0 0-J 0. z- 00 m' xC m00 0 0 0 0 0 0 0 0 0 0 0 0z m z00 0 0 0 0 0 0

    (A - - - - - -U -0 -N -I- -uC 3 ~ . . ' . . ' c

    4) 7 7 222~L. LUI 4 I t I t t ' t t

    -- -2,

    441

  • C) CD W)"0 C) 02 M0 c'.4- )0 r- .-. O) 000 82 i

    t- 0' - mrO 0o L0 co1

    C) : : : :,'0 9" ' C)'0 -

    C- oi c> 0i 00 C-n r-- 01\ 0 m0no C C D 0 C

    0- Lh' L o c

    U"0 v0 ('40' U) 00 N- 0

    C) 'r (S. ) w- U). C') A (0.- O- 4-' Ll 4-' V) cc) (NJ

    L "

    (a0( 0 m 0

    =- , o-V = N 0 0 0 0 0 = () ( ) 00 0 0

    (4") Ot"jJ~' N N LA 0 . tLA2'J 0 O

    W~ 4) -o C. l O) 10 m -n LU U).C")A) C'

    0.~-, 0.I 0.

    U'~ PI' 0 0 0

    o a, C) C) oC) (A L) U) . 0 0 0 ) (A. '

    CW o/ c) ol C) o 0

    '2 C) I4 C)0 0 0 0

    Uj > Cl) c; C) >; C

    C) c> o c)0 o 0000o0o0 C'. 00000000000c

    c> 00 00000000 oCc) o0 0 0 0 0

    R C,0 o- 0 cD C0 0 C, D o c C) 0

    oD VD C) 0, o~)0 0 0 0 cC0 L( L) 0, Nc 'J '. c) C 0 LI"C) -0) a)o'-o c , c

    on -' 2 LU '0 L2

    o) W x L)cK u((o

    (4 0.§ 4l~ c4 4

    Or 'r -1 It t8 0 "IO~ ) r" O~

    )

    v) C) 't N- C)LU r '.t'.'N"N-N.. -I

    >I) >1 c0 13o 0o D u c

    m >C xc~ z -z) : 4 ) I 0W 4 4 4 4 4x

    00

    o 0

    (o >- m (aLU>

    -4 - 0 44 0 0 0 O- 0 Ln3, in cL o. vn Lon Ln Ln mU aU oU QU LU w - X

    L3L

    LU - !2C)/) C) : n -C)u

    U)u ; E : C u uQ0)

    :3 (8 3V)' (2 z z z c a1 6

    444 Uc dLr czu 8u

  • cc. U.) 000 0 01 n 0 0 0 0CC:: 0,

    7'

    2 o

    ooo o o o ' -J' 00 0 '- - D -

    co ex D I' ) 0 c) C- ) C> '0 0> r, 0 N- C) 0D CD w~ '0 w- C> LA L ) . \ r - 0- '0 (0 InLAL 0 00 N- -'4 (D IT m4 ? 00 cc m- (SI r- 0 0 4 L -'

    C- COo SO- CD0'-.- , r-- 0Ln P > L. ci 1 f C- 0 ' L() L% C)

    U CN . oC A , 1 n wA

    f.LA Nl) L2 0) V') c0 C) 0 0 - 0 t 0 ' .- CS. 0 0 C0 CD C0 0r 0T 0 - 0 0 'T

    (S0 105 LA

    V 0 L" 0n tI, 0' LA N-0' C:, ) C, 0 0 0 '0 C, 0> '0 '4> N- LA LA '4 C',' L , 0 0

    N-C'.).-~~C '0 CS. '0(10 a) "O-N LA.- -- N-0 (1'0'' (0 n C a A. C0)" w -2 " C : D' D C r I-r C >V

    m ) LA fn) (S.

    CD )h, 0 0 0 0 0 C) CDi CSi CDi C

    2 '. 0,0000 0000

    C)!LD CD C C .C C DC~00

    U ' ' ' ' '' ' ' '

    (D CDCDCD CoC)C)CDCDC>CDCD4z C C ( C C ( C CN-N C-N- CNC-CN- C ( CC0 **' C4--. C, 0 0 0 00C DC C DC DL C DC DC C = C jC)C )C DCal01>C DC DC C lC Z )( DC C-. C.).D.. . . .D.. . .CD.. . . . . .D.. . .C.. . . . .

    -0 cl0

    0 0 00D 0 0 0 0 0 0 0D 0) 0 0z 0> C>. CD (SA ( C 0 0n 0 0 0) 0W 0

    V) IT 0 0 0 0 0 0 0 0 - - - - - - 000t LA " L A ' 0 Ln L LA LA V%'

    00

    u LU

    C x~ -) -U -U LU LU- )L UUW N N -N -N -N -N

    LLU

    m. 1

    L 00 aIc o c o cu c 't' I I I~

    N- -m 0I '0 LA C O CO v N COif COi M) CO COj

    L -4 44 4 ------ 444-l7 77 u 77 7

    Ccui cii w i ci ci ciic . .ci ., , ' c i..cw w,.,,wocw w d w w wu 4) ac ,o2 2 0 Z 2 2 L. 2 2 2

    0LUx8 z

    cMl - - --- - t tUtS's r 'C u

    C -wc/h col 0- - - - - - - - 4' N '4'4

    443

  • 00 ~ C 0 4) o00 02 !2 00 '~ ~L'M~ Ino coI

    0c (\0 0 0. I-0 -Wp!I I ( 0

    DC) ( 4

    w 14 -0 'n 'IT IT CD :8 aj C C0' 'r ~o. 'r It .TJ w) ( - o. fl-" -o o )L 1"I ~ ' C> u, esj -0 I rD 00 1 n ) co0 0 0

    4. j C> 4.' L- r c>0 ,L )f1r e j" v

    0. 9 0. L .c . .L 7 .9 7 - r - c

    m ~0 0 0 0-0 j lj (J 0 0 - 0D 0D D 0 C> 0 - 0) 0 - 0CD ) 'a. 0 , 0a C) C) >C c

    't~ ~~ ((5 C>0 (=C C) r(j (IS %0 C') 0' C) '0j ..t)C> I tc: D ca4.'..,- l-u 1 1 in c Lr 0,~' 40' 4-' C'. 10 0-((C 1"I (\ ^(C - rj 2

    mC) l 01 U0. (- 1-

    C~~C C) .' '' C)

    a) L, L Lr 0 0 0 0

    LI) m 'o

    :3 CD, (( D 0Sf LCD f( L. 0 0 0 00D c>5 (A C':JrJ ')' CD 0 0 0 0 0)C:

    cl C) ) CD.'..'' C (DCD, CDJ'J ' * C) (DCDa C D D( D c C D C )C0D c)C ,' )C c D 0 C DC > 5C D ( D( C) Co CD c=, 0~ C).. . . . .

    Co (D CIS C> (D C (C/DQ( D )C DC D!%C)C >C )C

    C) 'a 0c ,c --0 0 00 00 0 0000C D0c D D( c 0 0000000 D C

    -0 0 0 0 00 0 0 0 0 ' 0 0 0 0

    Cc D LA3 D ( C) L. ct-' rL 1-~ Lt P - P - - ' t I q 1. t- r

    o0

    QC4 4 4 4 4 ~ 4 4 4 44 4 4 4C) > '. 0o0 - - - - - ((orSo O o L c c 000 n

    aC) A o 3c L

    U) > >i~0m m. z - 4.

    0.00. (U (

    of 0

    fl C)) 6n unN p) Q I t 10 0m 0.j 0) 0 : D( S rI tI

    = !2 !2 c2 >? !2 > - IS 0 02 ~~~~~~~~ ~~~~~~~~ 0 ' ~ * 0 ( ~ '( '~ 1 . . C(.( (5 u j-.' 4 4 40 U>~ '.J 'JtJC.-~ 000 - - - - - 0 4 U -. 0*-.'i'±-

    Li~ ~~~~~ ~~ ~~~~ 0. z44 4 0 0 4 4 4 4 U- m. - ((z(5( (150. :z- z .

    -3 m In -o Wo------------C) 2F2 2 P- 2 2 2 2 2 c) C -c> 0 4:3 0. -~

    U-j r~ N* * ~ cm'Oc~ N M1I T 'I 'r 0 cD c> o U-r 'T.0 0 0. . .0 0 0 't LUW r- cm w r0.01"I fn . .N

    c' Li C 4 4 '' 4 4 4 n00 O ' ZLi ' sO0 4 44(A;) 2 2 , 2 2 2 2 ,2 00 0 ) * * 0 2 2

    LU 4 4 U U-4 4 4 4 4 U-U- - 4 U-Li i L 0.U-U- i 4 4 4 4

    0, - n 0 0 0 0 0 Ln V . . . . ,a21 m~t~

    0 -0 1.-

    44

  • Go 'r

    U '-CD C C

    0 l U) , N. U 0 ~ 0, ..i 0, .0 r- M C 0 0' CN ) M ) M 0 (NJ % 0 ,~ 0 MILU 0) .0 0w LA P,' r' LA LV U) r .. 0 . A 'N N - 0 L C0)0

    NO Nl . N. .00. . - 0 N9 N N r .~ N N N N . . N N.,V '0 U) n Ln LA' - C, CALA (4 U) fA ) '0 (4 i pn 'nA LA L"

    L 0Ia- =1

    4- C)(( C (4') CD' C4) CD ND CD NN 0 0 0 0 0 0 0 0 0 C N n T CD C

    m o o o o oN .. N .'N .N

    4) Cf 00 0 00 C "'' 0 0 0 0 0 0 0 00iC~ T( pi >fv

    Ch I 0 0m 0 04 0, 0n 0l 02 0l 0n 00 r- 0i (4)(' 4' 4)(l(UjU' 0' ' . ' 0n It ..± . 4 * 0 0 0 0 0

    4)0 4-',o C! CLA C 9999

    ) 0l 0~ 0A 0I 0l 0~ 0> C 0 0o 0D 0t 0 0 t 0 0 0 0 10 10 0 0 10 0 100000 0 00 00 000 0000 000000000000000 00000000000

    78 C> . . . . . . . . . . . . . . . . .D D D C C C D . . . . . . . ..nCD CD CD Co ) 0 0 O O O L L L LA A A A A A A A AfL L

    v 4) LU) CD CD ) CD D D C C (D CDCD D .N N C CN N D(DC>N D DNC CN Nm CD L DC DC ,0 C DC DC DC DC n DC DC DC

    CD C, C) C)CC D 0 C C D ( C D n C C .C DC )C

    ot LU C! C.)9 - C 9u ,c ,c 3 ( -o c D 801,.4 LL ,L ,w n v n t ,L

    0 -c o c o D o c o c 0 4 Q ( I 4 C l l i(

    4) 0. :5 4 4 4 4 4 4 4 4 4 4 4 4

    (4) 0.

    0) : I - .

    0L

    L..4- '-4z

    > 'n x CAn '(7 Cj, ';n CA C 'LA C A L 7 A L ) C71 LA i;L A A L L A L0

    uLUI (Dce 0D

    it w) 00 w w w 0 w w w w 0,0' 0,-g g m 0 w 0 co w w wLA .d 0 .. . r4.4.4--. N tv y 4-.tmOj ~j

    !) !2 V v v ! ZZ222 2 v v ! 2 v v

    0 LU

    L: : 7 ; 1 0 l l , ,0 0

    ,0 ,0.0

    C~ Lu0 (Ui .- 0'0N0 f0'0(I0'0'0'0 0''O'O

    O u- NNN N N

    -445

  • x~ N CU0ix, =~ ,- w .u ON2 .- o- o-t2

    L W L L_ Lii 0D0 00 0 0 0 0 00 D*CDC

    'O NC It- .QmN i - 10 IT' U.' 0' N - p 0. - ~ n F 0 a.'0 - 10 U' - 'rLU m) It ItQ 0-Nl. 0'0, ' . n Ir 10 Nm n NOt CD 'Ic P

    w~ .1 0 0 0 0 0 " fi00ac

    Z - PU-.C 0 .* i.1

    CD 0, o.0 N- - N O A D(DC M o

    V) 'r M CC p1. It'''J P

    Q. =-0.I

    Z C: C C'

    C) C',4 C

    'n PO) Cf\I 144 U)

    cc 4) m4)

    U Cf) L) Cd

    C) -'(a0 -C f

    o 4~1 * 0u~ > * . . .. . . .. . . .. . . .

    oQ.

    c 'a :-

    0

    a) CD CD CD C)QCCD C .C C D C >U%& CD C- CD C CD ) U )1u

    0 4)

    o -4 0 .4 (a. .4 0. 4 o .. 4 U4 4 .4 4 .4 .4 0 .4 .4 0 0 .4 . . 0 .u w (

    -InC

    ji _j _ . 0

    -1 4)It4LU 0) CD)

    Pn (). r_ r-4F- N N - t '-- N 1 m. rI)

    0 . 0.1 0 004m00a0000a 0 C CI 00000000

    LU 4' ) t- 0 f-u Coo C.w-- - - - - - - - - - - - -

    14 * 4)' M -4)2EEZEEEEEEZ is mm~ = 2E x

    4 Lr .40 L -

    LU~ .C0.4

    m z.m-4 x m m =zzzwam

    46

  • w- CD 00 4-0C

    C)' C) C CD2 LU- 02 -U

    L2 0 D CD 0n C C l w r-l t1 >0 l DN. .- ONmrwf-f-Itr n C 2.

    CU .4- u, In r- t - l

    -. ~~ i!. 9 C, 'ou~CJ-~~ -Ul 0,J*-m00c0

    iCA i ~ C D~ C, IW- L ! 4-- L) C n "

    CU 0 (U 0

    L 0 Z,0 00 0 0 (S, 000 L

    'D

    0000000.- Co"

    w) C) C2 oCm) - m la 0 mffl.' r

    0.

    (U C) ( C

    o, 0 -

    D- 4-. : :

    x L L x c w4 U..u U L U u L

    C~l

    C) 04- 00

    F5 -j

    4-D CC . . . . . .D.. .C" 'V CC) m c

    n) U T55 .~ " m5. m- " m "s " l -t-r -r - L 2

    I-00

    C)2 !2!C) . > m m m m~i- m~U.i p m ~ u> m .- m u~m >

    U u. C Cl ,- - 7 7 7 7 (/) Cfl fl (A

    m . 4-4 0 0 G O 0 4-

    .0 C)C -C -C -L ~ - 1- I- - C- C- I--

    o w4 -w -D -, C) CD CD 10 -, -0 -, -0 1, -M V, !2 !2 -2 V --.5 -,- ,-0->Cn r- ' m -8c " . jC v ~ nrl n"I

    10 10 Ci) CA (- 10 10 10 10 C/) (A CA C0 10 "40 4A CA CA CD -J C3

    C3 C)o l0,0 ,It I I I It C) It MmI n p -0) I r4 04 P4 r4 N N Nm i

    0 . CU f z

    ~e fl A SC/ 4S CL S - C It47

  • -0 CJU W" 06 U, 02 r-!' 02 0 wN--0 W- 'c-.. N-. Z6JN . 0 .. N!

    w c2 N - pn N! fl Ln 02 C, C2 02 CD. C0 CD C> 0) 0D 0Q r 02 G-.- I 2n Lm n C, .- MrinN N-( CD 000 0 0 0D 00 0t N!

    C:- .40 U, 2. C'J '0 -S-S 0' CJ '0 Z- I mN cD *0 (D - C. C, w- '2S0i -S) -n N- CD U'. r. C') - r - 0, '0S .0 2 ,. 2 '

    Cd, C'. K' - " ! U, 01. C'.)-'~~ .. .'. C'

    '0 0

    0 'O C'N - 02 N! N Dc Dc n! N Go c- 02 0 C') C'j C'.fn ! U't -5 go) 0 rn co0 '00 '0V -U , - 0202 S 0.0 .r'.) U, CO 0 > 0 n (D 0. oN C)' 00 0' C) N- 'r 4 -

    C') 02 0 - C') - C.

    C 0

    LO4

    0' (D0, D;z ;

    V)4

    03 '0 CL CL m m L m

    0 0

    Lu U >0 r I

    LDU 00~ l)M M 0 00 0 0 0 0 0 0 0 0 0.4-.N-L.! v ! v v

    q ~ ~ ~ ~ ~ C !2 !2 !24S -N N N N !2 21 N-N!-N2--. S- -- - -N -N7 7u0777 uu

    4 w- , 10 0 0 0 9w w u u u 00u0

    ca LU )- - I- I - 1 - - -I I - C- C-;- I - I - - C O C O C O C O

    c 'o > 2 2o t ' 2t t ' 2 2 o -o 'o'o 4 4 4 4 4

    a))

    4o 'o 'o al Z8 !s NO NO CO C O O C VC O C C. Z; 0- CO CO O 0, Cl O C It r- o mO COC CD CO C> CO

    LU) U ~ . l m -pn t > N It It N It It It It 'tS- ItLUts I LU LU LU LU UN LU LU LU LU LU LU LU(3z ;r ; I2 m0! ! -- 0 0 C. C. N 'e z z.) '). . . . . . .

    4 . 4 0 4 z --- - 4 4 4 4 4 4 4 4 4 4

    0 0022020220220 02 20 020 -J.~.~- .J00480

  • tc L es

    3 00

    a, '4-

    01,. &t 0 n C 1010 0

    coo0

    4.. V 0" - 0 0 0 0 0 0 0 n O 0D 0l 0m 0 0 0 0 0r 0 l 0 0 0 0 0 0M

    -04

    m w4

    too

    .0

    (0 4

    0 C>4)

    4) 0

    L) cc w< W

    Y) 0

    0 4

    o 4-

    0 (03C) 0)

    4e4) ~ ~ ~ ~ .0. a 4 4 4 4 4 4 4 4 ' 4 4 4 4 4 4 4

    ------------------------- _----------------------u I-------------------------LN

    4).L 00 01 0. Ci I- 0 'C 0. 0i 0i I- 0- 0 0 0 0 U. U 0 0 i ' .~ C I

    >. !2a !2 !2 !2 !2 !2t !2t !2 ! 2t

    4- 01- - 1 1 1

    LU 4) - - - - - - - - - - - - - - I- - -- - - - - - I- - -- - I- --

    P- 0 o .. o 'o 'o 'o - o 'o 'o 0, 0,- ,, 0, , 0, CD -D C: , C ,

    4.49

  • ~1 ") OO' N -

    W. LU 0 0 . . .

    NN-N-2 r .-0O N .(D t-LA "tO A~ L(L w ~ CDC A N, w w 0. N Lnt- o, Nl 1- C- 0 N%-LA rn 61% N N-

    0n CO '0i 'o 0! 0s 0, Nt N>- 0t +- '44C2.4 -4' Cr'4 C

    (m 0

    Io rn O % 0 O s 'I0 C O (D 0P0 O 0 m' n CDO D0 IVO C L 00 0 , 00 N L rn '0 W% N- OA VII 0V%'. -N L A0 4 0, 0- 0- '4

    m )10a. i-I

    'A L) CA

    n (

    )

    4-. 0) U) ^ UulLn r , r r (

    4)> >14 owL oL oc ) -

    .1 -- Ic

    (5) 3E Z;jN N-- N-7 Z4N L A L A 'C ',7 Z- N-; N-7 N-. N -N-N -n '4; N- - - 'N- N - 4

    4C)

    0. 0

    44- m

    Lu - ) 4- 0- - - - 0 -~

    > !2 !22 2 2 !2 2 !2 !2 2 2!R2!R22 2:2 222!2 7 74747447474474 77474 774747747477474 77474 7OD444

    z <

  • '0 LA(-~ c-.

    LW 0, W r-N 1 . - 0DC )c 0 W 0, M C )010 0 n c i .( > L D P nw w L lL

    N.1 0 1 ) - r _P - . r_ P- I( o ..- T o 31; ; o 0,c o C 0 fn C o -I lc o 1 lr-c

    8. 'o ' . CO'o C m c C . p o oc o .In I c o2 lL ; L n IC

    Co 0 N

    0 'J cD a 0 0 0 - c '0 0 o0 '00 w- 0 N -D'o 'oIs '0 _ cl It~ h. OLAN N-C'n, LJ cD _r-0m e'. .- 000 (J 0 0 0 w A L 0 MJ.1 0 00 ' ' j 0 0 0 0 V)'o 'T "j

    c- L~ - , n Cl -J

    C 'A

    '~0

    0.

    01) 0- I

    Cc(

    Ll L

    U1 w U (

    >1>

    0- 0

    (J( u~ Nu N- N-_ N- N P- P, N- N- N-0 N - N- N- N-P- N N- N- N- N(), N- N

    > !2 !) !2

  • N-l L c r C

    af w 0.0.4 Li' -. 5-5 n~

    LW w~ LL D0 0 0 0 0000CDCCDC

    1 Is0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

    CD . 1^O' Nn r0r~ -_ O 00- N ' 0 C.1ONr40 0 . 00- LnCD C, c0NCD m 10 -LU LlLU r-000.N ' J, 10 0t L "IoO05 1

    0 0 C, F4D 00o0l CD -m NO, Q N* N -l 0. r4n " m0 N0l C1 0 r' -- f- Li

    tL = L-~

    a. = 0. =

    ;; L.. 't ' DC n K 0 -

    72N- L ODi' Q~ 0C- CDOr. ~ -0 o ' 0 N O CD 0,fl4P00-t~t0 fl N NI- N-0, 'o.0 0-.t co- a

    u w~ L (A O P . C

    04 04

    C0 0

    (0 Uc u Ix u

    .00

    >1 Go -C 0(1c0 w C 0 w w 0w0L C M4 m M,. . . . . . . . . . . . . . . . . . . . . . .

    4, 4, u 0o >CO ~u

    4- Q 4- L1 . iP - I- f- P- t-CI K 0 0 0

    U 0

    4,c 00 0 0 0 0 0 0 0 0 0 0

    U)~~ C -t * -* u * u 5 * U * U t -t 5 -* U -5 t - U. , U~ U~4, - c c ca - w w w w w w w w w w w

    C-ii

    LUi -4, -li-- I-

    W -K c cc

    0 w &0 CDC

    , 01 Z;,

    L- - 4P.

    0. .- N N LU Z~ -- -- -- -- N -

    P0P P00 0 >~O OOO O00 000 000520

  • M) - 10 -r LA(U L 0

    w~~~0 00 co,, . (

    L ui D q010 0 0o 0n 0000c DLn wC D D

    OLAO (.I AnL ' AW)A ( c' ;2 '0-* CmU 0 rrl-, 0 f l 4n co

    O~~ ~ .. . ..- C9 .0 0 0 LA 0 00-0D10-4 CD t 't. esj -- -O - CD o 0 0 M- (D Q C ( M L A'

    4-Ar rn . 10 '

    (U 0

    c. r- 0j 0 " 0i 0 0> 0v 0M 0 0 -0 -T 0 0 .m 1

    ca 4) &01 -4 -

    CU

    4) U,

    C.) (A

    (D

    (u > (A

    0

    E 0. 0. 4 - -(U QO r LA LA OUD in Q CDOOODCO

    a) > 0 w1 mm u PA- w < ~ l w ~ AU mm J S. 0 U

    4-. -J -

    -U Z.) .J- , -7, , Z; Z; Z; Z; 17) Z; -;i -, 71 Z; Z; 7,I - ~- , . 2.0 0w

    LU

    LU 4) N -N N r -N r -rQJ CU 0 DC C DC

    > U

    (3 u 0000000000000000 U U U

    0 Uo -C

  • 01 .' .- -.m~~1 Lr~ m) t-'.-

    cc ID.1

    .0 0 , 4). 0 0 1LW C 0CL LU C LWI 0

    00 C-~C - 0

    0 D0000 0 C)0N( D( tCm 0 0 0A .'4 0tC o'o 0

    IA ;! 0' IAj L -4 ( 0 0L C, Ln Ln I

    0.2 0.2 ". Ln 10

    ijCDC C

    m 0~) % (a 000? 0 00

    0. CL = ~ t

    L~~~0 'tI , AL I

    3e Lc m1 c c4) ;!. .!2 CA. 4D I A ' . 40 0

    $ 4 . 0, r4 L r 4 ) ,- , Lo 4).

    IT a)) (a 4)4CL a. CO0)101 I 1 I

    U (A cn U (A U) >

    m0 00 00 t 0AL0 00 0 00 00

    0il 4 - ,0 0-- 0 00 04 0

    0- C, (D C. a DCD0 CC)0 C (D W 0( CDJA C ~ ' ) 0

    00 c 10 c 0 1

    0- cc u C

    4- 4- It It 't I4t--P- P P z0 .- 0 1 0- 1 -. -) ( X X X

    >J > co co CIO U) co >1 cc >1 4) CO C- >1' c-I coI ~ 4 ULIJ 10 Q0 10 10 Dc -

    LU czL U U K

    aU C11 0.00mm110 C10 002L 0HJ U - - ~ ~ .U ~ "

    .u.4I1iO0 O O2 ~ )(2 10 0 0 0 9 0 0 0

    CE m. i 0 0 0.m 5 - - - - - - rmui x w CIDcLU

    41 u u r-1 w 2.2: : 2u 0 )

    LU 0 LiLc u.L ChL LU L L, U L

    2 - W) K K '- 4)1i 0. - of w w w aLI - 4)12

    LULILLU cc uI LI LI uLI x uuio cLI w

    0 in Lu L',

    U). ) 2,4ct~.?4U

    U~ > U1 I nL nL I 0>4-., L t "I 4-. L. Lm U., L- LA )U I % LA A 4 A

    0. 0.2 0.2- - - - - - - - - - - -0.2--pn PI

    54

  • 4) ''0. '; 0 C) 1'0 . el 0) 'o C.4- CDL c'. 4 (( ' t 4- r.- 0.(0 LO 0 -. 01r ( -. 0

    0) 10 (D C'o . 0 '0w CDO 0) I. L U CO 0 L- w O 0 CD

    2 0W 0N) , 'o CD0

    CD~~ ~ ~ CNI a)C -C -rlCD0 n %C %C

    s00 ollAJ4 LU. s 2 " p , 2 0 ! S -1

    CD r- w A C> C>t~ CD1 D1 l 4rn N4- L 4- (r. 4-

    L o 0 I.. CD C> CDC D C ) m C

    (0 0 (00 00. = a. = a- =

    L. 't ZLL CD LLC D' I C>C n Dr

    T. ( r~ -1 LA Ln LO ( T . (N0

    L(o L4 (0.....

    C) ~ r ( A , .,, C) , C,, ,C ..

    M) M &- co-

    (0c C c '

    4- 0) (A 0(A u0( A

    o - C ,, .. 4- ).D. 4- )

    C)04 04 04u (A >U)>f

    C) (D C0 0 D0D0 0OD LO O O O Ln (DC>

    C- ( i(N LA LAO

    u. W C)00D0D0D C (D C)

    0) Cie C 9) cc u

    a) CD C,- C D a4- C . 4- C>-

    C 0 ~ 0 4 (em c00 4 LU -C LLI . Li C -U

    L)-

    4 C

    CD 0. 0. SI C co 0. Z 0 w w , c

    Li

    (a 4- w w 4- m4- 00000

    0. .L U, c2 LU L!2 L

    C LU LU Lu LU LU Lu - C) . c)

    L- -J 0-i w )~ w 0i )K( c c c

    3E ( 0) 3E 2rLU LUw

    (A '-C U) (A (W- ) . .

    C) Lr C) Ir n rf-) 0) fn 0) f )

    LU~( (0 (-A A9

    N N- > N-C P. 4-C)4)4 LI( v

    pI- l )p fn U) w) 0 00 0 0-M M CD CD Cl U) CD

    4 Li Li Li Li Li Li C

    ~ ~ ~ ) K KK 55

  • r, LL PJ- to z N-i Pf3 0. t @

    w C)ODO LW 0u 0 0 0D C 0

    2 2

    0 D (S 0 C> (I l N - P-'0N -N-IT @ ' olWL, n .0 [email protected] (S w3 &w : r , [email protected]

    C0. .0. 9

    m 00 N( 4m MI L0 C30 C>t m(S 0 a; 0 00000000CD C

    ~- T) (4 L- C D (iD w -ci C D o C > C > (

    r.- t-0.Q

    4, . ) w4

    CD 4, LD 4=Q4)C )C

    T L C

    0 t

    o 0 4= C.i C)C n C4 , = 0 n )..CC : n L

    40 04-

    LC 0 LA LAAL LA LALALA00 0 n0 0 w w

    cc i C3 0- wC

    x0 0 0 0 LA 0 0 0 W ~ LAW A 0 0 00 LA0t L& cy *- t .? 5 m.. m. tL 0- CS. m5 N- m ~ m.

    4))

    w 0 0

    wi Li

    - L > > )C to> 1- 1- 4 4 4 K K ) )3E 4)-si- 6 4J .4 . 1J J z44 . . 4 ,,,.

    0 wJ a. m m LU 0-JO 9 0...00 u 0.0.0Q0. , ,u-u -C -

    to S.- m

    4) u -- 4)C

    r,- K >t Ut t L t K t U t t t t tt t Itt Ut P, t~ Ut Ut0, o

    m~~~ i

    LU LU J 4 4 4 4 . J! 4 - 0 0, ,0>~~ Z- 4)4 44 44 4

    56N - - N---..-- - -

  • 01N' N-, CD C m . 0 0'0 4- ,f N 0 N > 0O.

    L-- .L 4 1 D 01 C* * 0

    CLC.C CP ' C)w LD W NCD r 0 0 Co Cr.0F 0 - 0 N DC

    - Lh: uj C 1 C

    0 CD CD 0w 0 ~ 0 0 0

    (a 0 (~ 0

    V -t0.0 C') l L.r CN NO C') OND 0 n-

    4.4

    4- V-

    Ca l .0.X 1

    C) ~ C-L) (f)

    .00

    > CDCd

    Q) 0

    C.C)

    pn ~ ~ ~ -0, *C4 -n Cj L , -M"- 0 N N N N 7 M N N N N

    C-'Y

    - 0' -C ' -C -.. ... ...

    o -JC~

    L CL u .

    -CU -C CU

    '01 014 "C" C e 4

    (61 8- CU

    w1 >

    - - M

    w ~ Z ZZ ~ CU Z Z Z 22572

  • 4) "LA N. 4)D Lf% LA% (' 0 (' 4jALc:, L'm 17% It C) C.i)l LA%

    (X CO cc 't C) ('4 L"0 00 0o co

    W10-17.

    C)~ C>AL C)- n -LMC Ln.U'.'CD LV% 0 0 0. ' ' 0 0 '0 M 0 ' Mco r- 8: 10 o 1 0 1 0 1 0 1 l1 T o

    4-' L M 4.J LL 3 J ~ ". JQU% M n m I n m nnrlV no

    2 U- (1 -29U-.. 1 l r

    CD. C)C) 0C' .- .- .4 .'. .4 (41 eN- CX ' SJ (4 (4 ' Ln s~C'j r

    0.L 0. N

    L- CD 4-' t'- A O C) '0) LA'0DC)CJC D Drj - DC ) D C

    (% 0C DC CD Cn (4 00 00 0 1 1 '4 0 4 10 0 0r)0

    9. IT. v.' 0 0 0 0 0 0 0i 0s , r lP-r - r-o 0 0oLi C\; Lii LA ri

    Cc L C) c 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 C

    40 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 00, 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

    000 CD00 0 0 0 0 0 0 0 0 0 0 0 00> DC ( C ) D CD C

    9.- 3000 3 0 0 0 0 0 0 0 0 0 0 0 0 0 0 ' Z00 0 0 0 00 rn 'rC:o~~~C '.'' AL A'L AL AL A AL A L A LA LA LA LA CD LA LA nA LA

    10 10

    C> L) 0 ) 0 DC 0 C DC >( C C 0 CDC0D C ) )C DC

    C) LA ll LfAO 9 9 O U L 9 LfA 4l U- Li Ui Li Lit) Li Li W! Vi V! Li Li

    EL i

    vo)~~~~~C)) )D CD CD CD, IX X- 747t474747A4747474z7z4z z ;

    o ,c -J ., . .J -, z z , - i - -K -C -x .c

  • It ~ Gol 0o 00 co 0CO~( L . C.D -tL

    10 U. 0 01 01 01 01 01 l 0 10 0 0 0 1 10 0 0wL

    'o '0C '. ON O

    CL =

    10 10 pn r'.j 10 '0 '0 '0 '0 '0 10 '0 '0 0 '0 '0 0 '0 0 '0 0 '0 '0 0 0 0, N- tn M~ P)l 00 r(-I LA '0 LA ULA N LA LAN cLAL LA ALn LALALn c c L A ALA LALAr.'SLA LA 'n ri L

    - ' ' C s.) j fm LA N~ Nv C'.) " N m Mi ( L N N CS. N v Nl (NN J N l Nr A.) ., ~ C

    CO, 'o0. C-L'cj -1 " C

    CD a )CL) (A )C >( ( ) C D C

    .0C 9 C 4) LA N9 0 0L L:0 00 .0 0 0 .L 00 . LA 9- .L -C!0

    0) Ln C Dt 0Clu CO v! 9 0 r

    D C 0 D 0 40 C> 0C0 0 0D 0D 0 0D 0 0 0D 0D 0 05 0 0D 0 0 0 0D CD 0D 0D 0D 0D 0, 0D 0 0 0aC)' 0 -u > (A

    ' 0 3 CD 0 0 D 0C0 0 0 0. 0 0 0D 0D 0D 0 0> 0D 0 0D 0D 0D 0D 0 0D 0 0D 0 0D 0D 0. 0 C0 0

    0 l U '-'A L LL V!LA LLAL LA LAW V L LAL LA AL LA LA LALALA L ALL LL LA LALA LA !V !V LA! LA

    Ct CO 0

    C).w0 , , - -PPN.r t N f N N N r- -- N NI- N N N - N9Nt- t

    >. >

    0- X X)CX X X XX X X X X XX X X X XX X X X X XX X XX XX X

    0) (n x - 1- 9. - 9 - - 9 - 9 1 - - 9 9 - 9-Z ',7 ',7 '1- Z- 9c7; 9-, ',7 Z- '47 '( '17 n-

    -. 0 !) 2 !2 !2 -, !. 2 !L 2 !2 !2 ! ! !2 !2 !? !2 V !2 !! V !2 !2 -, !22 - 2 !2

    0 0

    !2 CO! 2 2 !

    cocu77777777777777

    k .- z mz)mm

    2i M) w c Ww e C -. 04ccw Ww MM Ww Mw wW--aw-

    0 zA zA CA wA mA wA mA (A xA mA zA zA C AC AC A C AC A (A C A C A C A C

    4) -C 4)1 -K - C - C - c c 4C.) 0.) r l 4 M l li C4 r I ici m m i Cie m cu ( -i Ci ri ci ci O l

    06

  • m. I'- C')D

    L- '0 D )C

    w 'O 00 0o 0ow wC

    m 0

    0. =0

    c rv

    a 0.a-

    QO-7- -7AL 9 9 9 9 .

    Olr- C 0.N. C.s'O LAN LAD LA -7 -t C ) ' -CDNLCAC.-DC)

    V, V,0 0000 0o 000 0' 00 0000

    0 U V l 4W 0 LAO ..7 LAO LALA 0 a, 0 00O f. OL AO V LA 0 0 LAO V!0 0 W!-0 0 0 LAu DC)C C D DC C ) DC)C , D 'A C 0 N. N L N! 0 N 0 CD(D D0 ( CD(D C)CD D 0NC00D0Nc m 0! N L N N !N ' N 0 '' N !~ N N N N N N

    0 - ) . . . . . .. . . . . . . . . .

    9)W

    o> w(4!m4) o oooo oooO O O O O O O O O O O O O O O<

    oi 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

    Ui0 0 0 0 0 0 00 0 0 0 0 0 0 0

    5I 3 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Z

    01 ~ ~ a ... A L A L A A L A L A L A L A A L A L A L AL A L A L AL A Lhu 2 ! 2 ! 2! 2!2 ! 2V! 2! ! V !R 00 0 0 0 00 0 0 0 0 00 0 0 0 0m 4)4) 000000 000000

    !2 !2 !R!32 ! 2! 2!

    c a 00o c= wzw=lac w4 4 4 4 4 4 4 4 ce~ CCW O Cc

    L) m z

    4) wWC - C - C - C 4 C -0! in V . > - N - N - N - N I - N - N - N - N - - N - N - N - N -

    -C m.C 0 0 0 0

    o~~~~ ~ ~ ~ U n 'n14 4 4 4 4 44444444w'A (4!CV ",Ln eqCJ L ~ ",lelLn fi ey(V CjLnjei ~ wlCWI r~ ~ L ",L~ ~ ~ LQ rv "l

    0 K.;m r l v " 0 c ~ ~ ~ N0 4.0.i fXX XXX Xv XXX m C XXX . ) .X .X XXX NX XXX " 4" rXAXXcl m

    - ) - N- N N - - 9- I- I 9 - - I- N N I - - N- N N - - - N- N N -61N- N

  • (D CD CD£.~L

    m It St S

    C LU-

    CND 101, ON.N N Il- N. r-_ N I- i! N. N. N. N. N.N . .N N

    m 00

    0. =

    - 1D 0. 0 0D - 0C 0 0 0o 0D 0 D 0 D 0 0 D 0 0 D 0 D 0 D 0 0 D 0 D 0 D 0 D 0 D 0 0 D 0 ) 0 D 0 0 a 0

    CO Lt 0. 0. V) '0 u"% co wl L Ln 0' LA L^ LA uL LA co .. LA LA ULA LA LA LA LA LA LA La LA LA V%0 r Is It t 'T 0. 14t 0' ItS t It -r St t I t 0. r N. ' 1-~ t It IT It It '~ t St St 11 0.

    u- CA.

    C.n a, 0 L L A O 0 . 0' Ni 'ON9 09tL L 9 . iStNI"Nf 0~' - 0 . .~ Q CD C CD- C>C-00CDC DQ C ,C , DC ) 0 0-0 0-0 0 ( C D D(DC

    C) 0 f 0

    0r 4) LAL 00) N. 00 LAV L W L V If N. St.- 0 r 0 CLAO.0 000.0W LAU LA OLA 00W! .l

    4) 0 .

    oL >4) ~ 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

    0 00000000000000000000000000000000 .x

    -'0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

    F5 4- c c

    C CC O

    Z) zS

    u- >

    0

    co C :

    0

    a) L^ L" L . " L n , N. m. ^ L . UU n , U

    U'% CAi Ln W V L U L n uAL LAAW% A nAC IN ALAL LMALALALA U)vs LA W% V lALW% ALA L^ L A CAU LAVI

    (A m C fi 2 2 m v "2 2 2ir4 C4cu2 2 Y 42 Yr4CI i Q QMAevMA.... mr4NJI-c w < ~ 4 w

    c N r~ 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 004 i 4 c ( N liC~ ri M)r~

    0 J , " " m Am q v(jm ir N l0 ( .rific v ( me r4 4. . . . . . .. . . .................. C4 r ( evC0 r4 N C 4r V N r . . c S i N rv r Vc

    M- "X maC mi " Z '' v " M '' M "Z "

    2 C62

  • a, rMs co

    0) '0

    LW w A w 0 w-nL Dr nC

    CD CDC ) )rrN 0 r'

    CD It It "I CID '5 CD~! CD"

    L 0

    IT 0 t 0 t z. 0l 0 0 0D 0 0p ) 0 0 0' 0 0' 0' 0) 0

    - ~ ~ ~ ~ c co LAf AL l ~0 ~ ~0 L AN . )N - N- N- N. N N N. N-

    0. f-

    -(0 00 0-J U-0 %. .. .0 0(O . 0 0g

    L C)

    00

    0 CD 00 LA 00 LA C> -±1 L 000 CL LA A ON-N c, 000 LA 0>C C C .c C >C D 0 00.lC>C

    CD C C) CD 0 D C> CD c LA 0 D 0 D 0 D CD CD CD. ?4 D 4 D - D Cl CD F M A D a l CD7 C ) CD9 CD D In CDC

    0) 00 CLi

    .

    V.4

    C) C - - - - - - m n m n n m D D D m m D D D D m D n m n

    I/idi~ N N.N.N. - - N N N-N. - - N N N.N.N. . - N. N- N- N. N. N. N. N- N. N - N-

    C) U)

    00

    Lii

    0o 4 - - - - 1- 7- 7- 7- 7- 7 7. 7- 7 - 7- 7 7 9 97 7 7- 7- 9- 7 7 7 7- 7- 7--(0>~ x z x m 2m222 2 2