tool comparison at gc stack etch in lam tcp using plasma ... · investigate impacts of process...

12.12.2000

APCAPC

Christoph SteuerTU DresdenInstitut fürOberflächen- undMikrostrukturphysik

ProjectPULSAR

2nd Workshop on Self Excited Electron Plasma ResonanceSpectroscopy, 11th-12th December 2000, Dresden, Germany

Tool Comparison at GC Stack Etch at LAMTCP Using Plasma Parameters (SEERS)

Tool Comparison at GC Stack Etch in LAM TCPUsing Plasma Parameters (SEERS)

Christoph Steuerdiploma thesis

Institut für Oberflächen- und Mikrostrukturphysik, TU Dresdenin cooperation with Infineon Technologies GmbH & Co. OHG

Center for Development and Innovation

Das diesem Bericht zugrundeliegende Vorhaben wurde mit Miitteln des SächsischenStaatsministeriums für Wirtschaft und Arbeit (Förderkennzeichen 5706) gefördert.

Die Verantwortung für den Inhalt dieser Veröffentlichung liegt beim Autor.

12.12.2000

APCAPC


ProjectPULSAR



Motivation

! Comparison of two plasma etch chambers LAM TCP 9400 SE atGate Conductor Stack Etch

! Critical etching of gate oxid found at finished products depending onchamber " find reason

! Investigate impacts of process parameter variations on in-situplasma parameters electron collision rate and electron density

! Find correlations between plasma parameters and measurementsas etch rate, uniformity

! Monitoring of high volume production

12.12.2000

APCAPC


ProjectPULSAR



Principle plasma etch (LAM TCP)

! LAM TCP 9400 SE:! 1 - wafer - plane - chamber! low pressure at chamber! inductive RF power coupling (= Top

Power)! additional power capacitively coupled

(Bottom Power)! reactive gases flow as mixture into

chamber

! plasma:# low pressure# high density# asymmetric RF discharge

123

107

854

69

12.12.2000

APCAPC


ProjectPULSAR



GC Stack - Principle

Si3N4

WSix

poly - Si

SiO2

wafer

+ +

-- +

+ condensor

gate

gate connection (gate conductor GC)

wafer surface

gate oxide ( GOX insulator)

channel

12.12.2000

APCAPC


ProjectPULSAR



SEERS - electrical model of RF discharge

12.12.2000

APCAPC


ProjectPULSAR



Principle of HERCULES measurement system

12.12.2000

APCAPC


ProjectPULSAR



Plan of experiments

parameter from to step width Test Etchduration (s)

pressure 5 mTorr 30 mTorr 1; 2; 5 mTorr 30

top power 50 W 400 W 50 W 20

bottom power 0 W 300 W 25 W 20

Cl2-flow 0 sccm 80 sccm 5 sccm 25

HCl- flow 0 sccm 160 sccm 10 sccm 25

O2- flow 0 sccm 50 sccm 1; 5 sccm 25

NF3- flow 0 sccm 50 sccm 1; 2; 5 sccm 25

! parameter variations:# changing one gas flow while leaving all

others constant# pressure variation# power grid, remaining sizes under

process conditions

! process characteristics:# process chemistry at main etch dominated by

chlorine# competition between chlorine and fluorine

species# competition between deposition and erosion

depending on process conditions

! measured values:# electron collision rate# electron density# etch rate, uniformity at

several tests

! wafer types:# mono Si wafer, raw# poly Si wafer for etch

rate & uniformity

12.12.2000

APCAPC


ProjectPULSAR



Results of parameter variations:Electron collision rate depending on pressure

! Pressure variation: collision rate shows nonlinear behaviour! Distinction of domination by ohmic heating / stochastic heating possible

(= different modi of power conversion into plasma)! Saturation / maximum at even higher pressures estimated

Mean electron collision rate vs. pressure, chamber 1

2,0E+07

4,0E+07

6,0E+07

8,0E+07

1,0E+08

1,2E+08

1,4E+08

0 10 20 30 40

pressure [mTorr]

Elec

tron

collis

ion

rate

[1/s

]

12.12.2000

APCAPC


ProjectPULSAR



Results of parameter variations: Correlationbetween uniformity and electron collision rate

! Nonlinear behaviour of uniformity depending on meanelectron collision rate

! No explicit function at whole parameter range

uniformity vs. mean electroncollision rate; pressure variation,

Chamber 1

0,0

2,0

4,0

6,0

8,0

10,0

12,0

0,0E+00 5,0E+07 1,0E+08 1,5E+08

Electron collision rate [1/s]

unifo

rmity

[%]

12.12.2000

APCAPC


ProjectPULSAR



Results of parameter variations: Correlationbetween electron collision rate and NF3 flow

! Using chamber several hours with chlorine chemistry exclusively! after that input of NF3

! Competition between chlorine and fluorine species (atoms, ions, radicals)! Explanation for shown behaviour ?

Mean electron collision rate vs.NF3-flow, chamber 1 & 2

0,0E+00

5,0E+07

1,0E+08

1,5E+08

0 10 20 30 40 50NF3-flow [sccm]

elec

tron

collis

ion

rate

[1/s

]

Ch.1 Ch.2

Mean electron density vs. NF3-flow, chamber 1 & 2

7,50E+09

1,00E+10

1,25E+10

1,50E+10

1,75E+10

2,00E+10

2,25E+10

2,50E+10

0 10 20 30 40 50NF3-flow [sccm]

elec

tron

dens

ity [1

/cm

³]

Ch.1 Ch.2

12.12.2000

APCAPC


ProjectPULSAR



Results of parameter variations: Etch rate andplasma parameters depending on pressure

! Partly correlationbetween in-situ and in-line measurementspossible (right part)

! Example: mean etchrate (blue) and quotientelectron density /electron collision rate(purple) at pressurevariation

! some more sizes seemto influence etch rate

ER & f(CR, ED) vs. pressurechamber 2

160

170

180

190

200

0 10 20 30 40pressure [mTorr]

Etch

rate

[nm

/min

]

10,00

20,00

30,00

40,00

50,00

60,00

f(CR

, ED

) [ar

b.u.

]

Mean_Etch Rate ED/CR

12.12.2000

APCAPC


ProjectPULSAR



Results of chamber comparison:Long term drift of chamber conditions

Control tests during whole experimental time:

mean collision rate[107 s-1]

mean electron density[1010 cm-3]

relative standard deviationcollision rate [%]

cham

ber 1

0

4

8

12

1,2

1,6

2,0

2,4

0%

25%

50%

75%

100%

cham

ber 2

0

4

8

12

1,2

1,6

2,0

2,4

0%

25%

50%

75%

100%

test No. test No. test No.

! All values at bothchambers showdrifts about wholetime exceptedmean electrondensity of ch. 2

! Ch.2: deviations ofmean values arehigher withincreasingtendency

! During eachparametervariation just smalldrifts

12.12.2000

APCAPC


ProjectPULSAR



Chamber comparison by electron collision ratedepending on top power and bottom power

! No significant qualitative differences! Quantitative difference in collision rate! Interesting: local minimum along increasing bottom power

" separating ohmic + stochastic heating

reference chamber striking chamber

Mean Electron Collision Rate vs. VariationTop and Bottom Power, Chamber 1, t = 86 rfh

Mean Electron Collision Rate vs. VariationTop and Bottom Power, Chamber 2, t = 73 rfh

12.12.2000

APCAPC


ProjectPULSAR



Chamber comparison by electron densitydepending on top power and bottom power

! Significant different shape and quantity at electron density area! Hint for cause of chamber difference: power coupling " hypothesis: top

power


Mean Electron Density vs. VariationTop and Bottom Power, Chamber 1, t = 86 rfh

Mean Electron Density vs. VariationTop and Bottom Power, Chamber 2, t = 73 rfh

12.12.2000

APCAPC


ProjectPULSAR



electr

on de

nsity

[109 c

m-3

]

chamber 2

top *

botto

m po

w.

(top p

ow.)2

(pres

sure

)2

botto

m po

w.

top po

w.

press

ure

HCl-f

low

(botto

m po

wer)2

NF3-f

low

8

-2

42

6

0

2018

121416

10

-4

electr

on de

nsity

[109 c

m-3]

ow

chamber 1

pres

sure

top po

w.

botto

m po

w. (pre

ssur

e)2

top*b

ottom

pow.

(top p

ow.)2

(bott

om po

w.)2

Cl2-f

low

HCl-fl

ow

86

42

0

-4

-2

12

10

Results of chamber comparison - Pareto

proof by Pareto comparison:! sequence and size of

influences change chamberby chamber

! Top power has biggestinfluence on plasmaparameters at strikingchamber

! Pressure plays second roleafter powers

! Gas flows have just verylittle influence


12.12.2000

APCAPC


ProjectPULSAR



Application of electron density on RF power inputcalibration

! Idea: calibration of powerdissipation by electron densitymeasurement

! Problem:# strong nonlinearity, set point

near local minimum atstriking chamber

# to achieve target value(reference chamber) halvingactual value is necessary

! Strong nonliearity possibly caused by interaction of TCP power matchboxand bottom power matchbox

! Because of this strong nonlinear effect electron density could not be usedfor chamber matching,

! Hardware reason of local maximum must be fixed before chamber tuning

reference chamber

striking chamber

12.12.2000

APCAPC


ProjectPULSAR



Results of production monitoring - main etch! More than 25 000 product

wafer were monitored! Good results and high yields at

all time

! Main etch step! Means and relative standard

deviations of wafers! Two logic products! Products clearly

distinguishable " recipe andwafer impact

! Chamber drift during wet cleancycle visible " influence ofchamber

electron collision rate vs. wafer (mean)

5

10

15

20

25

30

35

40

0 5000 10000 15000 20000 25000

wafer

colli

sion

rate

[107 s

-1]

one point - one wafer

Logic product 1

Logic product 2

electron collision rate vs. wafer (rel. std. dev.)

0,0

0,2

0,4

0,6

0,8

0 5000 10000 15000 20000 25000wafer

rel.

std.

dev.

one point - one wafer

Logic product 2

Logic product 1

12.12.2000

APCAPC


ProjectPULSAR



Results production monitoring - Over Etch

! Conditioning effectsremarkable after every shortclean

! Memory products causesignificant chamberconditioning effects

! Logic products lessconditioning

! Means of over etch steps(productive)

! Several selected memoryand logic products shown

! Products distinguishableeach by each " influence ofwafer and process

electron collision rate vs. wafer [mean]

0

50

100

150

200

3000 3200 3400 3600 3800 4000wafer

colli

sion

rate

[107 s

-1]

DRAM product 1 DRAM product 2 DRAM Product 3DRAM Product 4 Logic Product 1 Logic Product 2 one point - one wafer

electron density vs. wafer [mean]

5

10

15

20

2530

35

40

45

50

3000 3200 3400 3600 3800 4000wafer

el. d

ensi

ty [1

09 cm

-3]

DRAM Product 1 DRAM Product 2 DRAM Product 3DRAM Product 4 Logic Product 1 Logic Product 2 one point - one wafer

12.12.2000

APCAPC


ProjectPULSAR



Summary

! Plasma parameter measurement was used sccessfully to identify the mainreason chamber performance difference – the top power coupling# Detection by electron density and electron collision rate# Proofed by Pareto analysis

! " Recommendation for maintenance team:# Check whole RF input parts and connections and power coupling into the

chamber

! Changes in pressure, powers, gas flows have nonlinear effects on electrondensity and electron collision rate

! Competing processes / reactions between chlorine and fluorine species! Distinct between ohmic heating and stochastic heating shown by plasma

parameters " effects on etch rate and uniformity

12.12.2000

APCAPC


ProjectPULSAR



Summary (cont.)

! Drifting chamber conditions proven at both chambers by means ofelectron density and electron collision rate

! Production monitoring shows:# long time drift during wet clean cycle at main etch step# short time drift during short clean cycle at over etch step# memory products and their processes cause conditioning effects# products and groups of them distinguishable inside plasma parameters

12.12.2000

APCAPC


ProjectPULSAR



Questions and possibel aims

! Reason is narrowed down - confirmation by maintenance teamnecessary

! Detailed hardware reason not known! Provide plasma parameter measurements including RF bias voltage

(depending on tool supplier)! Sequence inverted gas flow variation might confirm chamber chemistry

influence on plasma parameter (for example: saturation behaviour ofchamber wall)

12.12.2000

APCAPC


ProjectPULSAR



Acknowledgement

! Acknowledge for all help, support and stimulating discussions to:# Profes. C. Laubschat (IOMP) and J.W. Bartha (IHM) - TU Dresden# A. Steinbach, S. Bernhardt, L. Christoph

Center for Development and Innovation# L. Dittmar, F. Zschorlich, J. Bullmann - ETCH# M. Klick et al. - ASI GmbH Berlin# U. Nehring - Product Engineering# M. Pierschel, T. Werner, H. Wendel, S. Mothes

tool comparison at gc stack etch in lam tcp using plasma ... · investigate impacts of process...

Documents