Improved control of chambercondition through new Waferless
Auto CleanSauabh Ullal, Shibu Gangadharan, Erik Edelberg, John Daugherty,
Harmeet Singh, Vahid VahediLam Research Corporation, Fremont, CA 94538
Anna R. Godfrey, Eray S. AydilDepartment of Chemical Engineering
University of California Santa Barbara, Santa Barbara, CA 93106
Lam Research Slide - 2(v2)
Importance of Wall ConditionPlasma reactor walls can play a crucial role in determining the plasma properties,such as the radical concentrations and ion densities, especially at low pressuresThis well-known plasma-wall interaction has caused problems ranging from processdrifts in IC manufacturing to irreproducible data in fundamental studies employingplasma diagnostics.Terms such as “wall conditioning” and “reactor seasoning” have become acceptedlanguage in literature to describe art of avoiding such effects.
Nature of deposition on wallsAmount of halogens on the wallsAmount of oxygen on the wallsOther species present on the wall
– H2O
What is Wall Condition?
Lam Research Slide - 3(v2)
850 1050 1250 1450
0.00
0.02
0.04
0.06
0.08
0.10
SiO2
After Cl2 plasma A fter SF 6/O2 plasma
Abs
orba
nce
W avenumbers (cm-1)
Kota, Coburn,Graves; J. Vac. Sci.Technol. A 16(1);270-277 (1998)
Etch Product re-deposition affects plasma chemistry5 Å thick SiO2 film is deposited on reactorwalls even when Si wafer or O2 are notpresent in the chamber: quartz window isthe source of Si and O.Wall recombination probability of Cl islowered drastically when walls are coatedeven with ~ a monolayer of SiO2.γ calculated by fitting measured Clconcentration to a model is ~ 0.03 on SiO2covered walls.SF6/O2 plasma removes the SiO2 film. γ ~ 1after the SiO2 is removed.Waferless Auto Cleans (fluorine richplasma) are used to clean chamberdeposition
– Eliminate drifts in process due to depositionbuildup
Lam Research Slide - 4(v2)
E = E0 exp {-z/dp}
d
n1
n2
z
E
A
Cl SiCl O
From FTIRSpectrometerTo HgCdTe
Detector
Chamber Wall
Internal Reflection Element
Plasma
MTIR-FTIR Surface ProbeMultiple Total Internal Reflection Fourier Transform InfraRed Spectroscopy is used tomonitor wall condition
Lam Research Slide - 5(v2)
650 850 1050 1250 1450
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
Increasing numberof wafers etched
Abs
orba
nce
Wavenumbers (cm -1)
Cl2/O2 plasma etching Si
...
Clean-1
Waferless Auto Clean development
Clean-1 is not sufficient to cleanaway all the wall deposits andhence leads to build up ofdeposition on the walls of thechamber O back bonded Si -F peakincreases i.e. F incorporationincreasingClean-1 is not suitable for thisetching process
Lam Research Slide - 6(v2)
650 850 1050 1250 1450
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
Abs
orba
nce
Wavenumbers (cm -1)
Cl2/O2 plasma etching Si
...
Clean-2
Waferless Auto Clean development (contd.)
Clean-2 eliminates all the walldeposits and prevents build upof deposition on the walls of thechamber.
Clean-2 is suitable for thisetching process.
Lam Research Slide - 7(v2)
0 2 4 6 8 100
2000
4000
6000
8000
10000
Etc
h R
ate
(Å/m
in)
Wafer Number
Shallow trench etch repeatablity
Clean-2 eliminates all the wall10 wafers etched with Cl2/O2plasma with plasma cleaning ofthe walls in between each waferi.e. Clean-2.Reproducible wall conditionsensured by the plasma clean:SixOyClz film is completelyremoved.Etch rate is very reproducible.However, close observation ofthe trench profiles show asubtle drift in the slope of thesidewalls.
Lam Research Slide - 8(v2)
Wafer 1 Wafer 10
Pre-Strip Stripped StrippedPre-Strip
Etch Profile Drift
Deposition on sidewalls of thetrench decreases from wafer 1to wafer 10
Lam Research Slide - 9(v2)
650 850 1050 1250 1450
0.00
0.05
0.10
0.15
0.20
0.25
Post-etch deposition
on chamber walls
Abs
orba
nce
Wavenumbers (cm -1)
Cl2/O2 plasma etching Si
...
Clean-2
Infrared spectra collected afterthe etching step of each wafer
Monitoring wall deposition
Chamber wall is clean before every wafer is etched i.e. Clean-2 removes all priordeposition from the chamber walls.Amount of deposition on chamber walls due to each wafer increases
Lam Research Slide - 10(v2)
0
4
8
12
16
20
24
Si-O
abs
orbt
ion
inte
nsit
y
Wafer Number0 2 4 6 8 10
0
2000
4000
6000
8000
Etc
h R
ate(
Å/m
in)
0 2 4 6 8 100
5
10
15
20
25 SiCl SiCl2 SiCl3 Ar Is at
Em
issi
on I
nten
sity
,Ion
Flu
x (A
.U.)
Wafer Number
3
4
5
6
7
8
9
SiCl4
SiC
l 4 con
cent
rati
on (A
.U.)
Wafer to wafer drift in plasma properties
With every successive wafer processed in the chamber– the amount of wall deposition due to each wafer increases;– etch rate and SiCl4 concentration in reactor exhaust remain constant;– emissions from SiClx etch products increase;– Ion density and Ar (750.4nm) emission remain constant;
⇒ SiClx concentration in gas phase increases.
Lam Research Slide - 11(v2)
0
4
8
12
16
20
24
Si-O
int
ensi
ty, S
iCl 4 c
once
ntra
tion
Wafer Number0 2 4 6 8 10
0
2000
4000
6000
8000
Discrepancy
{
Etc
h R
ate(
Å/m
in)
Decreasing deposition on the waferClean-2 Etch products generatedeither deposit on the surfacesexposed to the plasma or getpumped out of the chamber.E.R. constant ⇒ the amount of etchproducts generated is not changing.Downstream FTIR ⇒ SiCl4 in theexhaust of the reactor does notchange.MTIR-FTIR surface probe ⇒ theamount of wall deposition due toeach wafer increases with everysuccessive wafer processed.Etch product deposition on waferdecreases to be consistent withexpected mass balance.
Lam Research Slide - 12(v2)
Wafer 1 Wafer 10
Pre-Strip Stripped StrippedPre-Strip
Etch Profile Drift
Deposition on sidewalls of thetrench decreases from wafer 1to wafer 10
Lam Research Slide - 13(v2)
14:00 14:04 14:08 14:12
0
500
1000
1500
2000
2500 Cl2 plasma on
SiF
4 Con
cent
rati
on (A
.U.)
Time (hh:mm) 680 685 690 695 700 705 710 715 7200
200
400
600
800
F emission lines
Em
issi
on I
nten
sity
Wavelength (nm)
Residual Fluorine level in the plasma
SiF4 is detected in the exhaust of the reactor; F emissions in the gas phase eventhough no fluorine containing species are introduced into the chamber.The decay in the SiF4 concentration in the exhaust indicates that an exhaustibleamount of fluorine is present in the reactor.F is liberated from reactor walls
Lam Research Slide - 14(v2)
0.01
0.02
0.03
SiF
4 con
cent
ratio
n (A
.U.)
W afer No.0 5 10 15 20
20
22
24
26
28
30
Si-
O A
bsor
banc
e In
tens
ity
Cl2/O2 plasma etching SiClean-2
... ...
{{
Wafer to wafer drift due to residual fluorine drift
Amount of SiF4 in exhaust of the reactorindicates the residual Fluorine level inthe chamber.More Fluorine is incorporated in thechamber when multiple un-optimizedclean steps are run.Fluorine level in the chamber is notreplenished to the same level by thesucceeding Clean-2.Residual Fluorine level in the chamberdecreases as more wafers are processedin the chamber and finally reaches asteady state.
Lam Research Slide - 15(v2)
0.04
0.06
0.08
0.10
0.12
SiF
4 con
cent
ratio
n (A
.U.)
W afer No.0 2 4 6 8 10
0
4
8
12
16
20
24
Int
egra
ted
Si-O
abs
orba
nce Cl2/O2 plasma etching Si
...
Clean-3
Optimized WAC: No wafer to wafer drift
Fluorine level in the chamber is replenished tothe same level by the succeeding Clean-3’s.Residual Fluorine level in the chamber isconstant as more wafers are processed in thechamber.Deposition on the chamber wall is notchanging.Deposition on the sidewalls of the trench isalso invariant.
Lam Research Confidential Slide - 16
Chamber Condition changes due toWet Cleans
Lam Research Slide - 17(v2)
Wet CleanWet clean are performed periodically on etch systemsWet clean can alter chamber condition due to introduction of moisture and otheratmospheric species in the plasma chamber
Typically, few seasoning wafers are required to restore on-wafer performance afterwet clean
– Process dependent (usually not an issue)– 65nm and below technology node may require tighter control of CD after wet clean– WAC and recovery optimization may be required to control chamber condition after wet
clean
Lam Research Slide - 18(v2)
Impact of Wet Clean and Recovery process on CD
6 nm CD shiftdue to wet clean
Immediately afterWet Clean
Steady State Immediately afterWet Clean
Steady State
No significant Shift
Un-Optimized Recovery Optimized Recovery
This particular etch process is very sensitive to chamber condition– Process stability?
Lam Research Slide - 19(v2)
Un-OptimizedRecovery
OptimizedRecovery
Immediately afterWet Clean
Steady State
Impact of Wet Clean and Recovery process on Profile
Moisture may act as oxygen source
Lam Research Slide - 20(v2)
PCA shows small drift without optimized recovery after wet cleanDrift in chamber condition not observed through other diagnosticsPrinciple component analysis shows a slight drift in tool parameters over ~ 60seasoning wafers after wet clean
Recovery complete
Prin
ciple
Com
pone
nt 1
Lam Research Slide - 21(v2)
Optimization of WAC results in no drift in tool parameters after wet clean
No drift in tool condition with optimized recovery after wet clean
Prin
ciple
Com
pone
nt 1
Lam Research Slide - 22(v2)
SummaryChamber condition control is critical for repeatable etch performance
– Control of deposition on chamber walls– Control of reactive species on chamber walls (halogens, oxygen)
Waferless Auto Cleans can be optimized to achieve repeatable chamber conditions– This will be essential as CD size continues to shrink