PROCESSING, ASSEMBLY &

~,

,~B; ':;W-~.,;." .. ~:... .,.~oJ •• ,'r'". ,~, ;.~. _- _. .. 1_ .....

A CAHNERS PUBLICA nON

INTERNATIONAL

JULY 1982

Also In this issue. Capital Equipment: Buy or Lease?., "

• Particulate Filtration of ChemicalS, Qlises and Photoresist

• CV Characterization System

July, 1982VoIS, No.7

News

Features

'Iiili!iidiii!tar..- INTERNATIONAL

9 U.S. Makers Are Still Alive in DRAM Market10 A Resist Wish List from SEMICON/West18 VLSI Project Launched in Australia18 Cominco Expands Pure Metals Plant18 Two Toshiba Companies Develop Silicon Crystal of High Purity20 NTT Will Form Joint Enterprise to Make Custom-Made LSls20 Harwell Designs Ion Implantation Cassette System22 NTD Process Under Study for Australian Use23 Intel and Burroughs Corporations Enter Agreement24 Motorola Grant to ASU24 Air Products Licenses LFE Plasma Etch Technology24 Sumitomo Electric Produces Large GaSb Single Crystal26 Harris Forms New Semiconductor Group26 Award Supports High-Purity Materials Research26 Highest Voltaic Efficiency for Solar Cells Attained28 People in the News30 Company Expansions32 Coming Events

35 Airborne-Particle Monitoring Know-HowGood, practical knowledge of airborne-particle monitoring is necessary to successfullyuse this technique in wafer fabrication clean rooms. But airborne-particle monitoring isgenerally not clearly understood by those who apply it in semiconductor processing. Inaddition, all the rules as specified in FED STD 209B and ASTM F-50, among otherstandards, are not clearly defined.

55 Particulate Filtration of Chemicals, Gases and PhotoresistEvery fluid that comes into contact with IC surfaces is a potential source ofcontamination which will affect yields. For this reason filtration of these fluids at thepoint of use is essential to obtain high yields.

67 Capital Equipment: Buy or Lease?The concept of leasing capital equipment is not new. However, in these times ofexpensive money, electronics firms, particularly the newer ones, are finding thatleasing can ease start-up costs and provide tax advantages over outright purchase.

81 A Computer-Controlled CV Characterization SystemRecent developments in computer-controlled instrumentation make possible greatersophistication in CV measurements in the process development environment, withoutsacrificing throughput.

Departments 99108111

New ProductsNew LiteratureCareer Opportunities

113114

Advertisers IndexSource Locator

t:ABP

~BPA

JULY, 1982

COVER: Sampling clean room air with an isokinetic probe and a light scattering particlemonitor is most rewarding when the readings obtained are un-"event"-ful. Goodparticle-monitoring programs can produce dramatic results in increased yields. For moredetails on particle monitoring, see the article that begins on page 35. (Photographcourtesy of Dryden Engineering, Santa Clara, Calif.)

Semiconductor International copyright 1982 by Cahners Publishing Company, Division of Reed Holdings, Inc. Norman l. Cahners, Chairman of the Board; Saul Goldweltz, President;Ronald G. Segel, Financial Vice President and Treasurer. Semicoodllctor Internatiooal is published by the Cahners Magazine Division; J. A. Sheehan, President; William M. Platt, ExecutiveVice President; Stephen Skilnyk, Vice President and Publishing Director. Circulation records maintained at Cahners Publishmg Co., 270 St Paul St, Denver, CO 80206. Second classpostage paid at Denver, CO 80202 and additional mailmg offices. Postmaster; Send Form 3579 to Semiconductor International 270 St. Paul St., Denver, CO 80206. Advertising andeditorial offices; 222 W. Adams St., Chicago, Il 60606. Telephone 312-263-4866. Subscription office: 270 St. Paul St., Denver, CO 80206. Telephone 303-388-4511. SemiconductorInternatiooal is circulated without charge to those qualified. Annual subscription to others: U.S. $35; Canada & MeXICO $40; Foreign $65; Foreign Air SerVice $35 additional; Single copypltce: U.S. $4; Canada & Mexico $5; Foreign $6. Send requests for qualification forms and/or change of address to subscription office.

5

Recent developments in computer-controlled instrumentation make possible greatersophistication in CV measurements in the process development environment, withoutsacrificing throughput.

A Computer-ControlledCV Characterization System

By C. Glenn Shirley,Process Technology Laboratory,SRDL, Motorola, Inc.,Phoenix, Ariz.

Capacitance-voltage (CV) measurements are widelyused to characterize dielectric thin films and theirinterfaces with semiconductors. These applicationsrange from routine monitoring of thermal gateoxides in MaS processing to fundamental academicstudies. In an industrial R&D environment there isan intermediate emphasis. This is process develop­ment. The development of VLSI processes, in partic­ular, requires CV (and other) measurement supportwhich must satisfy requirements different from bothroutine monitoring and academic studies. In processdevelopment, the variety of dielectric materials andapplications requires considerable flexibility in thechoice of measurement parameters. For example, thethin film electrical characterization lab handlesdielectrics ranging from thermal SiOz through CVDand PECVD SiOz and Si3N4 to polyimides. Anotherrequirement is the increasing sophistication of mea­surements required to monitor VLSI process devel­opment. All process development has a third require­ment which tends to run counter to flexibility andsophistication, and that is throughput. On the otherhand, process development monitoring measure­ments seldom require new fundamental knowledge.The central problem is to tailor existing techniquesdescribed in the literature to the process develop­ment environment.

A conventional approach to research CV character­ization involves the use of a lock-in amplifier, a rampgenerator, a signal generator, an X-Y plotter, a ther­mal chuck, etc., synchronized by TTL triggers, and soforth. The various parameters would be set by theoperator using the front-panel dials and switches.This is adequate in an academic setting, but forindustrial process development the requirement ofthroughput makes this analog approach tedious anddifficult to reproduce because of the detailed atten­tion which must be paid to parameter settings. Inpractice, the analog approach forces a choice betweenflexibility and throughput.

Since the mid-1970s when the General-PurposeInterface Bus or GPIB (also known as the IEEE-4BBbus or, according to Hewlett-Packard, as the Hewlett­Packard Interface Bus [HPIBD came into common

JULY, 1982

use, and especially since 1979 when Hewlett-Packardintroduced the HP4275A LCR Meter, it has becomepossible to conveniently build a computer-controlleddigital system which overcomes the problems ofanalog systems, and which affords significant addi­tional advantages. The GPIB has made it easy todefine the hardware aspect of a system becausecomputer and instruments are plug-compatible, andinstruments can be configured either by front-panelsettings or from the computer via the GPIB. Readingstaken by the instruments can also be transmitted tothe computer over the GPIB. Most of the labor indeveloping a GPIB-based system is in software devel­opment.

This report describes the computer-controlled dig­ital CV system designed and built in the Thin FilmSection of Motorola's Process Technology Laboratory(PTL). The discussion is intended to describe thecapabilities of the system, and to indicate what isrequired to duplicate the system in another lab.Detailed software descriptions and lists will not begiven here.

The hardware systemAlthough interconnection of the elements of thehardware system is easy, choice of the elementsthemselves involves some nontrivial considera­tions.

A basic choice is the ~omputer used to control thesystem. We chose the Tektronix 4051 Graphics Sys­tem because it has graphics capability permittingvisual interaction with the data. This frees the user tothink on a higher plane of abstraction. Anotheressential is that the computer be easily programma­ble in a high level language, e.g., BASIC, includingcommunication with peripherals with simple input/output commands. The display peripherals include a1200 baud matrix printer (a slower printer causesinconvenient delays), a plotter, and a hard copy unit.The hard copy unit gives copies of the 4051 screenand serves as a backup to the printer and plotter. Italso provides copies of unusual display conditions.The computer has a built-in tape drive which is usedonly for scratch and backup applications. The maindata storage facility is the Tektronix 4907 dual floppydisc system. This allows one to develop sophisticatedsoftware systems with interacting files and librariesof files for both data and programs. The softwaresystem described in the next section requires the discdrive as a part of the system. Our system has a dual

81

----------------------_._---------

I, L~=:'::::"_J--l-rtAPEFl\.E$,1~_SCVCORREc't__~:, ,

+--+-'---.--<"j ICVFR£O I---i'-- e. G. Glw.C' VS,l"

t,,, r-~:E.'N'lERCEP't

I ,

H: I I-I " ,.__~_A_NA_L---ItT cv ANALYS,."",NTOUTS

I ,1---.1._ PLOTS Of CV TRACES

library of data files named $CVPROC.CV measurements involve a sequence of tempera­

ture-bias stresses and C-V (and G-V) traces which weshall call a procedure. This sequence may be assimple as the standard Na+-shift monitoring proce­dure used for process monitoring, or it can be quitecomplicated. A feature of our software system is thatone can enter, display and edit a CV procedure using$CVPROCDEF before making any measurements.The procedure so defined is then saved in the library$CVPROC under a user-defined name where it isavailable to be called up and executed by the pro­gram $AUTOCV. It is $AUTOCV which controlsinstruments and handles the real-time data acquisi­tion. There are three basic types of steps which maybe sequenced in any way using $CVPROCDEF, andeach type of step has several parameters to bedefined:

(a) Trace (Parameters: Vstart, Vstop, R, F, T). Thiscauses C and G to be measured, displayed on areal-time plot and stored in memory at values of biasstarting from Vstart and ending at Vstop with a ramprate of R (V Is) at a frequency of F(Hz) and tempera­ture T (0C). The trace begins when the temperaturereaches T after changing from its previous value.

(b) Ramp (Parameters: Vstart, Vstop, R, T). This isthe same as the trace option except that measure­ments of C and G are not made, plotted or stored. Theramp step allows gradual voltage transitions ratherthan abrupt transitions between procedure steps, ifdesired.

(c) Soak (Parameters: time, T, Vbias). This stepimmediately applies the specified voltage bias, Vbias,and starts heating or cooling to the specified temper­ature, T. When T is reached, the clock starts and

L The hardware c;pnfigllration.

drive since this permits separation of programs anddata, and it simplifies backup procedures. The soft­ware system requires a real-time clock that can beread from an executing program. We use the clockbuilt into the disc drive system.

Instrumentation includes the Hewlett-PackardHP4275A LCR meter, a digital voltmeter and adigitally controlled thermochuck. All the parametersfor these instruments can be set up under programcontrol through digital interfaces without operatorintervention. Readings are also transmitted throughthese interfaces. The HP4275A LCR meter can mea­sure capacitance, C, and parallel conductance, G (orany other representation of the impedance) at 10discrete frequencies from 10 kHz to 10 MHz in 1-2-4steps at any bias in the range -99.9 V -< V bias -<99.9 V. The actual bias seen by the sample is con­trolled by a Temptronic TP350A thermochuck. Thiscontrols at any temperature in the range -30°C to200°C. The setpoint and current temperature is com­municated through a separate BCD interface (Trans­Era 632-BCD). All of the equipment is protected fromline power fluctuations by an isolation transformer.The hardware configuration is summarized in Fig. l.

The software systemThe software system, diagramed in Fig. 2, resides ondiscs. The system is divided into a data acquisitionsection and a data analysis section. The two sectionscommunicate via disc data files written under auser-defined name into the library $CVLlB. Thediscussion of the two sections will be separate, andwe will follow several specific examples throughvarious stages of data acquisition and analysis,returning to examples at appropriate points in thediscussion.

Data acquisitionThis function is implemented with three programs,$CVPROCDEF, $CUSTOMCV, and $AUTOCV and a

2. The software system: 1) dat.a acquisition program disc;2) lISer data disc; 3) data analysis program .disc.

82 SEMICONDUCTOR INTERNATIONAL

(

IiFt

LOO

0.50

2.50

C&

1.50 ~

'"

2.00

E.. 2

3.00

2

0.50

.~...._..-_...__......._._....._....~<...-_._~-

0,00-0.50

+100

..~......__._...._.~_.-z::

-LOO

:2................._-_._ _ .

'---r--..--,---.,....;;;::;.;....--r------,-..L-O.OO1.00E+2

1.50

2.00

0,00

0.50

E .. 3

~ 1.00u

Hit RETURN to continue ...

BIAS (V)

4b. Hysteresis traces showing charge injection generatedfrom procedure in Fig. 4a. Plasma-enhanced CVD Si02•

"'CV PROCEDURE DEFINITION·Had Procedure Name: HYST100MID

Step #1 (TtMe) vstart = P.po vstop = -!Xl.~0 R = 1.00 (Vis)Freq = 1,OE+005(Hz) T = 25.0 (deg OJ

Step #2 (Soa~~ Time = 100 (sec) Temp = 25,0 (deg C) BI,", =Stl)p #3 (Ttace) Vstart = -99J,o vstop = oo.~ R=l.do(\i/~)

Freq =1.0E+OO6 (Hz) T = 25.0 (deg C)

Step #4 (Soa~) Time = 100 (se<;) Temp = 25,0 (deg C) Bias = ~~.~o

Stl)p #5 (Trace) V.tart = ~9.~ Vslop = 0.00 R = 1.00 (Vis)F",q =1 .O~+005 (Hz) T = 25.0 (089 C)

PB2025U DOT2 WAF IS HYSTJOOMJD 27-FEB-BI 18'39'52 (PB2025U/HYSMI21l

4a. Hysteresis procedure.

Example 1. Standard CV qualification. The proce­dure shown in Fig. 3a, named STDCV, is a standardMOS gate oxide qualification procedure. When thisprocedure was executed by $AUTOCV, the data wasstored on a file PEMASK1 in the library $CVLIB.Later, this data was recalled and plotted using theprogram $CVPLOT, yielding the plot in Fig. 3b.Before executing a procedure in $AUTOCV the oper­ator enters a string of characters which appears in theheading of every plot generated from the data file in$CVLIB. The computer automatically appends to thisstring the name of the procedure in $CVPROC whichgenerated the plot, the date and time and the nameof the data file in $CVLIB on which the data is stored.Thus the system is self-documenting as far as ispractical.

Example 2. Hysteresis study of plasma-enhancedCVD SiOz. A commonly-used type of procedure isshown in Fig. 4a. This subjected the sample to thebias-time profile shown as the inset in Fig. 4b, andgenerated the traces shown in the main part of thefigure. The sense of the hysteresis shows that chargeinjection is occurring in this sample. The finishedplot does not show the sense of the hysteresis, butthe plot generated by $CVPLOT reproduces theoriginal traces in the correct temporal sequence, sodirection arrows may be determined by watching theplot as it is reproduced.

Example 3. Frequency study of thin thermal (gate)

0.20

0,40

E+2

L.OO

0.60

o

• 0:60 i.;.:::01:,,1,1\:;;:;;:::;;; ~

. .'.... '.

Temp = ;roo,o (oeg C) Illas = P,OO

-5,00 R = 0,10 (Vis)T = 2M (delle)=200,0 (dell C) Bias

Vstal1 = 5,00 Vstop = -5,00 R = 0,10 (Vis)Freq = UlE+OO6(Hz) T = 25,0 (deg C)

lime = 300 (sec) Temp = ;roo,o (deg C) Bias = -20.00

V$tel1 = 5.00 Vstop = -5.00 R = 0,10 (Vts)Freq = 1.0E+OO6 (Hz) T = 25,0 (deg C)

(Soak)

(Traee)

ev QUAL. PE>17"03·BI-2 MASK I STDev IB·MAR-al 08,41'~2 (PEMASKI)

....

L.OO ;:: :1;::':::;::'"

2.00

Step #5

Step #6

Htt RETURN to continue,.

3a. Standard MOS qualification procedure.

3.00liL0.

U

BIAS (v)

3b. Plot generated by thermal Si02 using procedure inFig. 3a. C: solid lines. G: dotted lines.

holds the temperature for the specified time. At theend of the time the chuck heats or cools to thetemperature of the next step in the procedure underbias, Vbias. To apply bias just while the sample is atthe requested temperature, one can use "dummy"soak steps as shown in Example 5 below.

The parameters in the procedure steps can haveany value within the range of the measurementequipment:

-99.9 V -< Vstart, Vstop, Vbias -< 99.9 VF = 10 kHz - 10 MHz in 1-2-4 steps-30°C<T< 200°Ctime >- 1 sR (ramp rate) can have any value (V Is)

In principle, the ramp rate R can have any value, butthe bias voltage is changed in discrete steps which,for large ramp rates, can be large because of the finite('V 1.5 s) cycle time between measurements. Forsufficiently slow ramp rates (not using the limitingcycle time), 100 points are plotted and stored pertrace.

Once a procedure has been defined by $CVPROC­DEF and stored in $CVPROC, it can be recalled andexecuted any number of times by $AUTOCV. Thus,even quite complicated procedures can be exactlyreproduced with minimal operator intervention.Examples will illustrate the system's flexibility andthe wide variety of questions which can beaddressed.

84 SEMICONDUCTOR INTERNATIONAL

PROCEDURE DEFIN1TION'~'

Old Procedure Name: FREQ200N

Hit RETURN 10 continue ...

SMALLAREACV

DOT

Vstart =2.00 Vslop ~ -2.00 R = 0.02 (VIs)Freq = 1.0E+004 (H~) T = 25.0 (deg C)

Vslart = 2.00 Vslop = -2.00 R = 0.02 (Vis)Freq ·=2.0E+ 004 (H~) T = 25.0 (deg C)

VSlari = 2.00 Vslop = - 2.00 R = 0.02 (ViS)Freq = 4.0E+004 (H~) T = 25.0 (dllg C)

VlIlart =.2.00 Vslop = -2.00 R= 0.02 (Vis)Freq = 1.0E+005 (H~) T = 25..0 (deg 0)

Vslart = 2.00 Vslop =-2.00 iii = 0.02 (Vis)Freq = 2.0E+005 (H~) T ~ 25.0 (de9 C)

Vslarl = 2.00 VS10p = -2.00 R = 0.02 (Vis)Freq = 4.0E+005 (H~) T ~ 25.0 (deg C)

VSlari = 2.00 VSlop = -2.00 iii = 0.02 (VIs)Freq = 1.0E+ 006 (H~) T= 2S.0(deg C)

Vslarl= 2:00 Vstop = -2.00. iii = 0.02 (Vis)Freq= 2.0E+OO6 (H~) T = 25.0(~gC)

(Trace)

(Trace)

(Trace)

(Trace)

(Trace)

(Trace)

(Trace)

(Trace)

Slep #6

Step #2

Slep#8

Step #3

Slep #4

. Step #7

LARGEAREACV

DOT

Gi;..ENWAFER ;o?~2 L.RG. DOT FREQ200N 19-MAY-81 H'17:27 (GI;..EN282L.) ·GL.EN WAFER #282 SML. DOT FREQ200N t9-MAY-81 10'39'57 (GL.·ENN28Z)

E+3 E+4E+3

3.00

I • I : : .. ; : I ; : : : : : : : : : ; • ~ ; i ~ ~ ; ; ; ~ i i ; ; ; ; ; ; ; ;

1.50

0.00

1.00

0.50

2.00

........ , .. ~ '.

1.00

..................;: ::11111.0; Ilni H fl.

......

0.00

BIAS (V)

-1.00-2.00

O.OO-'---.,----.--------.--_--.- ..,...---J

2.50

Z.OO

~~ 1.50u

1.00

0.50

0.00

LOO 0

~;:,

'"

1.50

0.50

2.00

..........

1.00

................

0.00

BIAS (V)

-1.00-2.00

1.00

0.00.J.,.,..........---,..,..,-__,....-__...-__,-.-...J

IL

~ 2,00u

BAGl<SIDE IMPEDANCE (GL.EN282L. AND GL.ENN282) FRONTSIDE. IMPEDANCE (GL.ENZ8ZL. AND GLENN2eZl

E ... 4E+5

0.00I •• I I • 1 •• I • I , , , I • I ; : ~ : : ; : ; ~ : : : ; ~ ; ; ; :; ; ; ; :0.00

Z.OO 8.00

1.50 6.00

~'0

Q. too 2u Z

C>

0.50 2.00

3.00

1.00

4.00

0.00

12.00 ::0

.3

'"

=.l;:;.::;;;; Ii;;:;:;; ;;;;;;;;;;;;;;;;;;

.,7,.

I0.001

-2.00 -1.00 0.00

BIAS (V)

1.00 2.00 -2.00 -tOO 0.00

BIAS (V1

wo 2.00

5. Tap: Frequency series procedure.Middle: Tr~ces generated an small and large dots.Bottom: Output of $CVCORRECT. Thermal Si02.

86 SEMICONDUCTOR INTERNATIONAL

"'CV PROCEDURE DEFINITION",

Htl RETURN to continue ..

leaved with soaks at increasing temperatures chosenfor equal intervals in 1IT, with bias and time of stressheld constant. The traces generated when this proce­dure is executed by $AUTOCV are also shown in Fig.6. The separation between successive traces increaseswith temperature. This data will be analysed furtherbelow.

Example 5. Polarization kinetics of polyimide. Theprocedure in Fig. 7 was run 01). a composite film with1.4 J.Lm of polyimide on 1000 A of thermal SiOz. Thisprocedure has standard room temperature tracesinterlaced with soaks for increasing times at fixedbias and temperature. The "dummy" one-secondsoaks are used to turn the bias on only when the

(Seak)

(Trace)

(Soak)

(T,_)

(Seek)

(Trace)

(Soak)

(TraCe)

(Soak)

(Trace)

Slep # 8

Step'" 9

Step #10

Slep #11

Slep # 6$tep # 7

Slep # 2

Step # 3

Old Pl«8duh. Name' ACTlV25

Step # I (TraC$J VSlart~ -25.00. V$top~ 15.00 R~ 0.50 (Vis)Freq~ 1.0E+004 (Hz) T~ 250 (dog C)

Tilne= 100 (sec) Temp~ 25.0 (deg C) 6ias~ -40.00

VSlart~ -25.00 Vstop~ 15.00 R~ 0.50 (Vis)Freq~ 1.0E+004 (Hz) T~ 25.0 (deg C)

Time= 100 (sec) Temp~ 550 (dag C) Blas~ -40.00

Vsl"rt'" -25.00 Vsfop~ 15.00 R= 0.50 (Vis)Freq~ 1.0E+004. (Hz) T= 25.0 (dog C)

Time= 100 (sec) Temp= 95.0 (dog C) Bi"~ -40.00

Vslarl'.o -25.00 Vslop= \5.00 R= 0.50 {Vis)Freq~ tOE +004 (Hz) T=25.0 (dog C)

Time~ 100 (sac) Tarnp~ 14Q.0 (dog C) Bias~ -40.00

Vstart= -25.00 Wtop" 15.00 R~ 0.50 (Vis)F,eq= 1.0E+004 (HZ) T~ 25.0 (deg C)

Time~ 100 (sec) Temp= 200.0 (dog C) 6ias~ -4Q.00

Vstart" ~25.00 VSlop= 15.00 R= 050 (Vis)Freq= 1.0E+004 {Hz) T= 25.0 (deg C)

Step # 4

Step .. 5

PB2025U WAF 19 ACriv25 12.MAY-81 16:40:24IPB2025U/ACTI)

"'CV PROCEDURE DEFINITION'"

(Soak)

(Soak)

(Seak)

(Traca)

(Seak)

(Soak)

(Soak)

(Trace)

(Soak)

(Soak)

(Seek)

(Traca)

Step #10

Step #11

Step #12

Slep #13

Old Procedure Name: GAYSHIFT

Step # I (Trace) Vstart~ 100.00 V$toP~. -100.00 R= 1..00 (Vis)Freq= 1.0E+005 (Hz) T= 25.0 (deg C)

Time= 1 (seo) Tamp~ 200.00 (deg C) Bias= 0.00

Time~ 60 (sec) Tam))= 200.00 (dog C) Bias~ ~5.00

Tima~ 1 (sec) Temp~ 200.00 (deg 0) Bias= 0.00

Vstart= 100.00 Vslop~ -100.00 R" 1.00 (Vis)Freq~ 1.0E+005(Hz) T~ 25.0 (deg 0)

Time= 1 (sac) Temp= 200.00 (dag C) Bias" -0.00

Time~ 300 (sec) Tarnp~ 200,00 (deg 0)Blas='-5.00

Time~ 1 (sec) Tarnp~ 200.00 (deg C) Blas~ 0,00

Vstart~ \.00.00 VSIOp~ ~ \00.00 R~ 1,00 (Vis)Freq" 1.0E+005 (Hz) T~ 25.0 (dog C)

Time= 1 (sac) Temp~ 200.00 (dog C) Bias~'O.QO

Time= 900 (sec) Temp= 200.00 (deg C) Bias" ~5.00

Time~ 1 (sec) Temp«200.00 (dagC) Bias'" 0..00

Vslart= 100..00 Vstop" ~100.00 R~ 1.00 (Vis)Freq~ 1.0E+005 (Hz) T~ 25.0 (dog C)

Step # 6

Stap # 7

Step # 8

Step # 9

S!ep # 2

Slep # 3

Step # 4Step # 5

0.50

2.50

2.00

'0.c

1.50 :::!O::J

C!)

1.00

3.00

E+2

.............................. 1: Ziti" III" I 0.00

1.00

0.00

0.50

2.00

E+2

2.50

3.00

~ L50u

,2.00 -1.00 0.00 tOo E+ I

BIAS (V)

6. Top: Procedure to detetmine activation energies.Bottom: Trat::es generated by this procedure.Plas11la~etthatlced CVD Si02.

Hit RETURN 10 continue ...

POSITION. INTt:RCt:PTS. REPLOT (P/I/R)PPOSITION POINTER a ~IT 'RETURN" IS TO STOP)

GAY S. WAFER #'4 00T(2,OJ GAYS~IFT 09c JUL·BI 11($8:48 (BAYS!4)

E+I

2.401.00

0.80

1,$:0· l I •. : •.. ' II: \, "' ~ :, 1:::':' ., " , ." • 1""" 0 ..00

7. Kinetics study of polyimide. Top: Procedure withdummy soaks. Middle: Curves generated by procedureshowing cursor positions. Bottom: Cursor coordinates.

»CURSOR POSITIONS«

0.20

G (J.l.Mho)6.40E-0015.86E-0015.12E-0013.12E-001

6:00 E... I4.00

Cap (pF)21.2020.7420.1018.39

2.00

BIAS (V)

__-Ai1=::;:~-'*"'"-1 •0.60 i:::>

'"

0.00

Bias (V)64,8534.9933.96-3.41

-2.00

Position1234

2.20

1.80

i 2.00u

SiOz. The frequency variation of CV plots is notusually done using conventional CV equipment, butwith our equipment it is just as easy to set up afrequency series as any other procedure. The proce­dure at the top of Fig. 5 generated the middle tracesin this figure. The bottom traces will be explainedbelow.

The system is particularly well-suited to a class ofmeasurements which gives flat-band voltage shifts asfunctions of (a) temperature at fixed bias and time ofstress, (b) time of stress at fixed temperature and bias,or (c) bias at fixed time of stress and temperature.These measurements give, respectively, (a) activationenergies, (b) kinetic laws, or (c) stress activation lawsfor charge movement in, or injection into, dielectricfilms. Procedures for this type of measurement canbe quite complicated and nearly impossible to carryout manually without error, let alone reproduceaccurately with different samples. We present exam­ples of measurements of type (a) and (b).

Example 4. Activation energy for charge injectioninto plasma-enhanced CVD SiOz. The procedure inFig. 6 has standard room-temperature traces inter-

JULY, 1982 87

***CUSTOM CV PLOTIING"'*

Is chuck Hi or Lo (H/L}H

Enter plot caption and hit "RETURN" to continueSTANDARD MOS SI02 • .

Dot area (cm.2}.01503

Save procedure (YIN}Y

New procedure filenameMOS1

2.00

Vstart,Vstop,R,F,T4, -4..08,1E4,25Hit "RETURN" to start run ...Trace/Soak/Clear/Quit (T/S/C/Q}STimerremp/Bias20,0,-2Trace/Soak/Clear/Quit (T/S/C/Q)TTrace Params (Prev/New}PHit "RETURN" to start run ...Trace/Soak/Clear/Quit (T/S/C/Q}C

0.50

0.00

l.50 :g

'"1.00 2-

'"/~:~:: :::~:::: :::~::::: :~::::~:::::: :::::::~ :~~ :

., ~ ~ ................•.,•. :;:;I......................................, "'". ., ..

E+3

2.00

i1.50

u 1.00

0.50

0.00

Trace/Soak/plear/Quit (T/S/C/Q}TTrace Params (Prev/New)NVstart,Vstop,R,F,T-~,5,.05,156,25

Plot Scale (Prev/New}NVmin,Vmax,Cmin,Cmax,Gmin,Gmax-5,5,0,2000,0,2000Hit "RETURN" to start rUn ...Trace/Soak/Clear/Quit (T/S/C/Q}STlmelTemp/BIas30,50,2Trace/Soak/Clear/Quit (T/S/C/Q}TTrace Params (Prev/New}N

Solve yourpurchasingproblems ...with the new 1982 U.S.Industrial Directory!Join the thousands of satisfiedusers who havediscovered thatthe U.S. Industrial Directory isthe best source of informationon industrial product manufac­turers and suppliers. Just fillout the order form and mailit to:

U.S. Industrial Directory270 St. Paul StreetDenver, CO 80206USA

-4.00 - 2.00 0.00 2.00 4.00BIAS tV)

,...-----------,The New 1982 U.S. IndustrialDirectory-ORDER FORMSATISFACTION GUARANTEED­30 DAY RETURN PRIVILEGED YESl Send the U.S. Industrial Direc­tory to me immediately! I want to solvepurchasing problems and save time andmoney in 1982. (Please allow 60 t090daysfor delivery.)

sigraturc

date

TOTAL COST FOR ALL FOUR VOLUMES:(payable only In U.S fundS)

D United States $ 7500D Canada and Mexico $ 8000D All othercountnes $12500

PREFERRED METHOD OF PAYMENT:D PAYMENT ENCLOSED D Bill meD Bill my company-PO attached

PLEASE PROVIDE THE INFORMATIONREQUESTED BELOW. THANK YOU.

Traee/Soak/Clear/Quit (T/S/C/Q}TTrace Pararns (Prev/New}NVstart,Vstop,R,F,T5,-5... 1,1 E6,OPlot Scafe (Prev/New)PHit "RETURN" to start run ...Trace/Soak/Clear/Quit (T/S/C/Q}TTrace Params (Prev/New}NVstart,Vstop,R,F,T5, -5,.1,156,5Hit"RETURN" to sta.rt run .•.Trace/Soak/Clear/Quit (TIS/C/Q)TTrace Pararns (Prev/New}N

2.00

Vstart,Vstop,R,F,T5.,-5,.1,1 E6,10Hit "RETURN" to start run ...Trace/Soak/Clear/Quit (TlS/C/Q}qSave data (Y/N}YFilenameMOS1

E+3

2.00

Illic

r:OlnpClny

1.50

...

.e 1.00<.>

0.50

0.00

,II il ~! ~ n;n ~ nn;; ~~ ~ ~ ~.1.,..•.................... "

.".I.,lt •. ,.I.,I •.. I.'"

1.502'"1.002-.~

0.50

0.00

-4.00 -2.00 0.00 2.00 4.00BIAS (VI

(:lly

('(Jl,fllr'.,; 8. Screen images for $CUSTOMCV real-time data acquisition session.

L. ~

88 SEMICONDUCTOR INTERNATIONAL

stress temperature is reached, and to turn it off at theend of the stress period just before cooling to thetrace temperature commences. The data generated bythis procedure is shown in Fig. 7.

Sometimes, with completely unknown samples, itis useful to do a "seat-of-pants" measurement, that is,to execute a soak or trace, observe the result, andthen decide on the next step. This can be useful, forexample, if one is not sure whether the flat-bandvoltage falls within the trace voltage span chosen.The program $CUSTOMCV allows the operator toperform this type of measurement. Shown in Fig. 8are screen images for a data acquisition session. Thescreen was cleared midway through the session tomake room for further interrogation and plots, butthe earlier data was not lost. The program has twooptions, both exercised during the sample session: (a)the sequence of steps chosen by the operator can bestored in the procedure library, $CVPROC, as theyare executed, under a user-defined name; (b) the dataaccumulated can be saved in the data library $CVLIBfor later analysis using data analysis programs. Sav­ing the procedure serves two purposes: first, a com­plete and detailed record of the sample's history isavailable for display using the printout function ofthe program $CVPROCDEF; second, the procedure isavailable for execution in the usual fashion by$AUTOCV. Thus, a procedure developed "manual­ly" under $CUSTOMCV may be repeated automati­cally any number of times with minimal operator

intervention. Displays generated from the savedprocedure and plot data are shown in Fig. 9.

Data analysisAll data generated by the data aquisition part of thesoftware system is written to data files in the library$CVLIB in a standard format. In addition to the C-Vand G-V trace data, these files contain capacitor areaand frequency data as well as an identifying stringwhich contains date, time, generating procedurename, data file name and user-supplied identifyingcharacters.

Since the plot data is available in digital form, it ispossible to write programs which can read the datafrom the disc and allow the operator to interact withit, often visually, to make desired parameter mea­surements. An analog CV system would requiretedious manual analysis of raw plots. In this subsec­tion we describe some of the analysis programs wehave developed, using further analysis of some ofthe examples described above as illustrations.

$CVPLOT. This simply allows one to plot datastored in $CVLIB either to the screen of the comput­er, or to the plotter. Various automatic and manualscaling options are available. All plots in this reportwith high resolution "tic" marks on axes are outputsof this program. Note that axis scales are multipliedby 10, 100, etc., according to whether E + 1, E + 2,etc., appears on the axis. .

$CVANAL. This program uses just the CV profile

Zurn Custom EngineeredClean Rooms ForTotal Environmental Control

Manufacturing, processing and packagingoperations for critical subminiature partsare far more efficient when done in an..engineered clean environment. We havethe design and engineering experience ...from concept to certification. Totalturnkey anywhere in the free world.

If you need a clean room in compliancewith Federal Standard 209B, we're readyto serve. For more information contactLinear-Flo Systems Co., 8110 N. St. LouisAvenue, Skokie, II- 60076. Or call312/674-1662.

LINEAR·FLO(~)A ZURN COMPANY

JULY, 1982Circle No. 81

89

"'CV PROCEDURE DEFINITION... The user will learn to monitor particular entries inthe difference table as a function or processingvariables according to the procedure definition andhis own interests.

$CVMEAS. This program gives complete flexibilityin measuring intercepts and values of C, G and V atvarious points on the plots, but carries the analysisno further than this. The program displays the CVplots on the computer's screen and then offers twomeasurement options. The first allows the user toplace the cursor by means of the joystick anywhereon the plot and record the value of C, G, and V at thepoint and repeat for any number of points. Thesecond option allows one to place the cursor any­where on the plot and have crosshairs drawn. Allintercepts with the crosshairs, as well as differencematrices of intercepts, are found and printed out.

Example 5 (continued). Polarization kinetics ofpolyimide. The plot in Fig. 7 shows numbered cursorposition marks, and below it the correspondingprintout of coordinates. The cursor can be placedanywhere on the plot, not necessarily on a curve. InFig. 11 the intercept option has been chosen. Thecursor was positioned, crosshairs drawn, and theintercept report generated.

$CVFREQ. Often the frequency variation of thefour quantities, C, G, G/w or D(=G/wC), wherew = 2-d, at various values of bias is useful in deter­mining the dynamical properties of the interfacestates, etc.

Example 6. Frequency study of thin thermal Si02 on

0.50

1.00

0.00

1.50

E+3

4.00-2.00

(Trace)

(Trace)

(Soak)

(Trace)

(Soak)

(Trace)

..1.. ,_ _. _ _.., _

-4'.00

·.~·J·l·J·.·I·I·iTI·I·I.I·rl·,TI'I·l

E+3

STANDARD MOS SI02 MOSI 22-JUL-81 14: 18:45 (MOSI)

1.50

1.00

0.50

0.00

Hit RETURN 10 conlinue ...

Slap #- 2

Step #- 3

Slep #- 8

Step #- 7

Step #- 6 . (Trace)

Step # 4

Step #- 5

0.00 2.00

BIAS (v)

9. Printout of procedure saved, and plot of data savedduring $CUSTOMCV session in Fig. 8.

Old Procedure Nama: MOSl

Step #- 1 (Trace) Vstart~ -5.00 Vstop~ 5.00 R~ 0.05 (Vis)Fraq~ 1.0E+OO6 (Hz) T~ 25.0 (<lag C)

nme~ 30 (sec) Temp~ 50.0 (cleg C) Blas~ 2.00

Vstart~ 4.00 Vslop~ -4.00 R~ 0.08 (Vis)Freq~ 1.0E+ 004 (Hz) T~ 25.0 (dell C)

nme~ 20 (sec) Temp~ 0.0 (OOg C) Blas~ -2.00

Vstart~ 4.00 Vstop~ -4.00 R~ 0.08 (Vis)Freq~ 1.0E+OO4 (HZ) T~ 25.0 (dell C)

Vstart~ 5.00 Vslop~ -5.00 R~ 0.10 (Vis)Freq~ 1.0E+OO6 (Hz) T~ 0.0 (deg C)

Vstart~ 5.00 VSlop~ -5.00 R~ 0.10 (VIs)Freq~ 1.00'+006 (Hz) T~ 5.0 (deg C)

Vstart~ 5.00 VSIOP~ -5.00 R~ 0.10 (Vis)Fraq~ l.OE+OO6 (HZ) T~ 10.0 (deg C)

.....,,..,O;QO

•'.000-;00;0.00

••.00(i.OO..,',00'.00'.00

L for examples 1 (top) and 4

FLATBANO- AND, THflESi;IOtP VOLTAGeT-rK,e" \<::lb ..,

I ~'a25 - &,65;! - 6.5'& - 5.993 "" 7.17 _ 6.58-'4 - 8.'11 - 7.5-15 -'9'$ >-; 9.276. ':',14,39 -13.16

FLA~VOLTAGE DIFFERENCE MAT.AlX

3 4. 5 ~ '7i'Mn 3.6:2 8.14 0,00

:~ ~:~ :'~ ~:~p~ U$ e:,2B 0.00;':':2;71 0,00 4.51 ·U.OO-7.n .-4.51 0.00 0,00:

1.1i&E.ti03{em'2) 1103.3 (mII'2)crn~ ... ;l:,46E+Ofl4:{Wcm'Zl. =-2:48,OOfpi)CrnIl1 .. 4.'oe+t'lOa,.,-lcnf2j, "'" ,,29.15{pt)Qb .. 1.56'E.+(1()4 tprk#f:2). = 110-69 (pt)

P,tyP$:{N" 1:24E+014 {aT,3}jMax ~i;lI,)IW ~dlh: .. 22::l98 (Anga)Pbi(S) .. 4.04 WOIlSI

T~"'974i~1Qietc:ons ... 3,80_."'"Phl(Ml,.'" 3.20 (V~

CN QUAt.. P'E.17~l·i ~SK 1 STOCV HI·MAFI-81 0&;41;52 {F'EMASK1JAREA 1.448E~OO2icm·2i 2244,4 (mW2}

RAW PATA emu; - 2,59E+OO4 (pllcm·2).- 375.10 (pt)Cinirl '" 1.48E+OO4 {plICl'lf21.- 21~.10 (pt)CfC. ~ 2.35E'-I-004 {plIOtrf21.a 340.45 (pfj

SUBSTRATE N-lype [N= LOOE ... (116 (ern" ~~,]MILK~ ISy&fwidl" =- 3007 (At!g:$)Ptri(SJ -" 3.45 (VolIsl

DJElECTlilC Thicknen'" 1299 CArlOs)Oielconsoe. 3,8(1

GATE METAl AluminumPl;li(M) - 3;20: (Volt6)

Fl.,AJDAN.O AND Tt':me:SHOL.O VOLTAGES

Trace" Vfll Vi1 -am -2.562 -0,67 -~,23

3 -(1,1tJ -2,67

FLATaAND VOLTAGE OIFFERENCE MA-TRllt1 2 3 4 5 6: 7 8

1 ,ttOO nti6 0.$ Q.QO 0.00 0-.00 0,00 .Q,OOi-O,Sf)' 0.00 -0,56 O,OO~ 0,00 0.00 0-.00 0,003-0,09 0,56 Q.OO lJ,OO 0.00 0,00 (11,00 0.1;10

SUBSTRATe '

?B~ WAf. 19 Acn\.l25 1;2,Ji1AY"61 16:40:24- (PB202SU1ACT11

AR£A

FlAW DATA

GATE METAl

data, ignoring the G-V data. It finds Cmax and Cmin,computes Cflal-band, substrate type and concentrationand dielectric (usually Si02) thickness, and findsflatband and threshold voltages. The algorithm usedfor this is given in Appendix A.

Example 1 (continued). Standard CV qualification.The report generated by $CVANAL for the dataplotted in Fig. 3 is shown in Fig. 10. A useful item isthe matrix of all possible flatband voltage differencesbetween traces. For example, the entry in row 2,column 1 is to be interpreted as a Na+ shift becausethe procedure STDCV (Fig. 3) is a standard MOS CVqualification.

Example 4 (continued). Activation energy forcharge injection into plasma-enhanced CVD Si02.When the data in Fig. 6 are analysed by $CVANAL,we obtain the analysis in Fig. 10, in which theoff-diagonal elements in the matrix [(1,2), (2,3), etc.]give the flat band voltage shifts between successivetraces generated by the procedure ACTIV25 (Fig. 6).These shifts are caused by soaks at successivelyhigher temperatures. Thus an Arrhenius plot of theoff-diagonal matrix elements using the soak temper­atures in ACTIV25 gives the activation energy of themechanism causing the shift.

I

i90 SEMICONDUCTOR INTERNATIONAL I______________L

2.401.00

E+ I

P-type SiOz. The procedure in Fig. 5 produced thedata shown in Fig. 12. After plotting the data, theuser can "section" the CV and GV traces at a set ofbias values. A vertical line is drawn at each of thesevalues, and C, G and f are found for each slice-curveintersection. The user then has the option of plottingany of the four quantities vs. log (w); Fig. 13. Eachcurve in the plots in Fig. 13 corresponds to a slice,i.e., a value of bias.

$CVCORRECT. Sometimes the impedance of thecountercontact in an MOS measurement has anappreciable effect on the reading. For example, forthin (= 200 A) thermal gate oxides the MOS capaci­tor impedance can be small enough to be comparableto the backside contact impedance. If C-V measure­ments are made on two MOS capacitors of differentarea, with the same counter contact in each case, thenthe capacitance and equivalent parallel conductanceof the MOS capacitor (per unit area) can be separatedfrom the (total) capacitance and equivalent parallelconductance of the backside contact. This separationassumes negligible fringing effects for the MOScapacitors. The theoretical details of the separationtechnique are given in Appendix B.

Example 3 (continued). Frequency study of thinthermal (gate) SiOz. The two sets of curves generatedon a large MOS capacitor, and a small MOS capacitorare used as inputs to $CVCORRECT. See Fig. 5. Theresulting MOS capacitor impedance and backsidecontact impedance are then written into the library$CVLIB. The resulting data files can be read andanalysed using any of the above CV analysis soft­ware. In particular, $CVPLOT was used to generateplots of the separate impedances at the bottom of Fig.5. These results show that most of the frequencydispersion in the original data is due to the backcontact.

$CVXFER. This is a utility which permits data to bewritten to and read from magnetic tape for long-termarchival storage.

0.20

6O.OOE+OooO.oOE+oOOa.DOE +000O.OOE ~ 000

9 "00 oe00 0.00,0 0.00.0 0.0

60.000000.0

6.00 E+ I

5O.OOE +000OOOE'OOOO.OOE+OOOo OOE +000

.0:00E+000o,ooE+ooo0,001:+000O,OOE+ooo

6 700 0:00,0 0.00,0 0.00,0 0,0

6 , 6 9 "0.00 IlOO 0,00 000 000000 0,00 000 0.00 0000.00 0,00 000 0.00 0.000.00 0,00 000 0.00 0.00

6 7 B ,"000 000 0.00 0.00 000

3O.OOE! 000O.OOE .000O.ooE~ 000o.ooE+ooo

_;..-~~::::::::;;;1fi!""'---10.60 ~:;:::>

• 5HI 0.00.0 0.00.0 0-000 O~O

· 5~23.54 0.00- 4.~5 0.00

1.62- 0.00000 0.00

· 50.00 0.00

ODO 2.00 4.00

BIAS (V)

-2.00

;----, CURSOR INTERCEPTS«

C and G Intercepts lor BIaS '" 0,34

1.80

Cap (pF) G (I"Mho)20,25 2ooE·00216.4i 2 QOE;OO218.42 200E-002

<I 16.45 200E-002

Bias Inlerceplglor C= 16.92 (pF) arod G"-3,75E.OOl I"MtJoC·Blas (V] G·8ias (V)

-7-43 0.0011.56 0.0017 74 0.001611 0.00

11. Application of $CVMEAS. Data with crosshairsdrawn (top) and corresponding intercept report.

2.20

POSITION. INTERCEPTS. REPLOT (P/URlJHIT "RETURN" WHEN CURSOR lS POSITIONEDCOPY (YIN) Y

GAY S. WAFER #14 DOT{2.01 GAYSHIFT 09-JUL-81 "'36'48 (GAYSI4)

0.80

C·Ir\lercopl diflerence matrix for Bias - o 34(Vj, , 3

D,O 18 1.B-1.8 00 0.0-1.8 0;0 1).<)

4 -1,8 {l,D 1M

(Nn1efcept_r~ malm: lor BHI$ t34V

O,OOE+OOO O.OOE +000O,OOE+ooO O,OOE_oooO.OOE+ooo 0 OOE; 1000

4 O.OOE~OOO O,OOE+ooo

Bias Jl'IIercepl matrix lor C 18-_!l6 (pF)i 2 3

0.00 -16.99 ~2S.fB

tB.99 -0,00 ~ 6-.f6Z5-.}6 £,1$ 0.00

4 :23.54 4,55 - f,,62

EJias /r'l1ef'Cl!iPl matrix lor G ~ :17~E-001 (IlMhQ,j, 2 3

000 -0.00 0.00

2- 2.00

<> 1----_Y-+#-------------1 0.40 '"

12. Application of $CVFREQ. CV and GV data atvarious frequencies, generated by the procedure in Fig. 5,is sectioned at a series of biases (vertical lines) prior togenerating frequency plots.

E+3

2.50

..~

;:;1.50

1.00

0.50

-1.00 0.00BIAS (v)

LaO

ConclusionsDevelopments in computers and instrumentation inthe past few years have made it easy to assemble thehardware for a digital computer-controlled CV sys­tem to support process development. The key tomaking such a system flexible, but convenient, andto exploiting the full potential of the hardware is todevelop a carefully thought-out software system. Theemphasis of this report has been the description ofboth the data acquisition software and data analysissoftware.

The software has evolved over a period inresponse to practical needs, so the system describedhere is a well-honed tool. Nevertheless, softwarechanges are continually being made, albeit at aslower rate now, and this will continue. Even in awell-developed system the computer's time will bedivided among program development, data analysis,and data acquisition tasks. Considerable improve­ment in efficiency can be realized if a second com­puter is mainly dedicated to the first two tasks, whilethe first computer is dedicated entirely to data acqui­sition.

92 SEMICONDUCTOR INTERNATIONAL