The CeramicSociety The Ceramic Society
of of Japan Japan
t・t/1,.---:-{{,ge.sijikie.z.
Journal of
the
Ceramic Societv of Japan
tt,ttstsS.tt,S・et
[1] 1-6 09g2)
100
Effectof
on
ofAdditives Alumina
ElectrostaticChucksForce
Toshiya WATANABE R & D Div., TOTO
the Electrostatic Tetsuo KITABAYASHI
and
Ltd.. 2-8-l, Honson
7JL;- tgeig+
t, v
Chigtxsaki-shi, jfunagawa
253
eoesag"{z&(geVMhntzoptee
rcssQte・ akMMft 253 itrvaMA:N fiveeeee{de)geebllvept,
May [Received
Ceramic
electrostatic
chucks
made
of
alumina
1, 1991;Accepted Sptember 19, 1991]
trostatic force for such
practicalapplications. imagine an electrostatic chuck holding a When voltage material as a two-electrodecapacitor. isapplied to the electrodes of the electrostatic chuck as shown in F:ig.1, the electrostatic forcegenerated between the inaterial and the electrostatic chuck can be given by the followingequation:
ceram-
icswere fabricated. Transitionmetal oxides (1wt% Ti02, Cr203,Mn02, CoO, CuO) were added to centrol the resistance of the dielectric insulation layerof the chuck. The electrical resistivity and electrostatic force at R.T. -2000C were measured, and itwas revealed that the electrostatic
force of the
tricalresistance
of
examined, creasing
Let
depends on the elecAmong the additives found to be most effective for inchuck
the ceramic.
Ti02 was the adhesion
forceof the alumina
us
iechuck.
1. much
attention
Al"mina,
applied
chucking
has wafer
e,
volta:ge, respectively.
Thus the largerthe value of the dielectric constant of the insulating layeror the smaller the thickness of layer,the larger the generated electhe insulating trostatic forcesare, The upper limitof the forcecan be determined by the electric breakdowns strength
{ntroduction
as a semiconductor
a)
constant of d and Vrepresentthe dielectrie the insulator, the thickness of the insulator and the where
Recently,electrostatic
(X)2
F-SE
electrostat-
Eiectrostaticchuck, Key-words;Eiectroslatics, Tmnsitionmetal oxide, Resistivity
2-8-1
received
holder in
the dielectric layer.The breakdown strength of these dielectrics isat highestle to 20 kg!crn2, Ifthe
fabrication systems.i} Semiconductor fabricationprocesses such as exposure systems, CVD, and dry etchers are operated in a vacuum environment, so vacuurn chucks which are operated in
of
semiconductor
dielectric constant of an insulator is10 or so, forceobtained by this configuration 100 g/cm2, When this value is will be approximately an atmospheric environment cannot be utilized. by a vacuum chuck, 1 compared with that obtained Therefore,conventionally, a wafer stage, a mechanical holding system, has been used, Electrostatic kg/cm2, itis found to be too small, In an altlernating method, Waring et al.,4) devechucks offer several advantages as compared with mechanical holding systems. For instance,the fiat- loped an electrostatic chuck by making use of Johnsen-Rahbeck's force,which is generated when a ness of the wafer can be corrected by the uniform the semiconductor electrostatic force on the back of wafer. Furthermore, the close contact increasesheat conduction between the semiconductor wafer and wafer stage; heating and cooling of the wafer can 1-----"' thus be ediciently carried out, 2.F[ In recent years, there have been many reports on 3 +++++++S4 the importance of temperature control in dry etching fer elecand CVD processes,2)and expectations relative
the electrostatic
[
trostatic chucking as a means of controlling wafer . temperature are growlng, Because of these advantages, many studies have been conducted on the application of electrostatic chucking.3} However, there have been only a few reports on the practicalapplication of electrostatic
M
Fig. 1. Schematic of presentation ef electrostatic 1:material to be chucked, 2:air gap, 3:insulation layer, 4:inner electrode.
in semiconductor manufacturing processThis is because of the insuMcient value of elec-
chucking
es.
.
chuck.
1 NII-Electronic NII-Electronic
Library Service
The CeramicSociety The Ceramic Society
of of Japan Japan
2
Effactof Additis,es on the E]ectrostatic Forceef AluminaElectrostatic Chucks
fiows across a conductor In thiscase, the microgap between the ceramic surface and the substance to be chucked acts as a capacitor, and the charges from the microcurrent accumulate on the surface of the insulation, generating electrostatic force, The electrostatic chuck by Waring et al., was originally intended foruse as a chuck formechanical processing, and SiC was used as the material. When the same electrostatic chuck is used in a semiconductor fabrication system, the current fiowingin the SiC seems to be too large,and it may cause damage to the material to be processed, Balakrishnan5)produced pelletsusing a mixture of MgO and Ti02, sintered them in an atmosphere consisting of hydrogen and nitrogen, and measured the eiectrostatic force of the material. The insulation resistance of the dielectric was approximately 105 9, and they reported that 1kgfcm2 was produced by the application of 100 V. When this processispractiminute and
amount
of current
a semiconductor,
cally
to semiconductor
applied
processing,
[ll"mi"a
However, no has
these conditions
satisfying
Furthermore,the trostaticforceand
electrostatic
the
values
the suriace
have
of
Cast./lb'
l"Dr'yin... 1・'u=ott-=g
.tttttttt'r'irttit/g(T.un ' gz・tenpaF,te) t.t.L tttttt'Lai,・iEglg
l--1..
-r'-ttttttt ilL-Si/'/ter'ing
chuck
must
high degree of fiatnessand good wearproperties,Also, in some CVD systems, wafers are processed at temperatures of 2000 to 3000C; thus high temperature tolerance is required. These requirements could be satisfied by employing a
resistant
ceramic
materials
as electrostatic
In this study,
alumina
transition metal
oxide
Mn02, CoO
CuO be
metal
oxides
oxides
by
and can
to
atmosphere6]・71' resistance,
were
The
teristics were
a
with
as
and
adjustment
electrostatic
of charac-
to investigate the efi'ect of
measured
the resistance of the force of the chuck,
the
enable
electrical
such
to nonstoichiometric hydrogen and nitrogen
converted
in
chucks
Cr203,Ti02, fabricated. The transition
additives
sitering
chucks.
electrostatic
on
ceramics
the
of
test
of electrostatic
Ni
/
...i,
pLat.' r/g/
..!ir't[・dtng,andpolishine
[-ttttsvu'z'//1/・E] pa/,,e'L -Fig. 2.
Flow
chart
of sample
preparation process.
trodes, These pieceswere
stacked on a green sheet formed intoa laminatethrough the applicationof pressure (1kgfcm2) at 700C.By exposingpart of the electrode on the substrate, a terminal forvoltage application was created. The sintering was performed in a hydrogen and nitrogen atmosphere at the 30"C dew point for 2 h. The insulation layeron the electrode was polished to a thickness of 300 ptm with diamondpastes, The size of the sample, includingthe substrate, was 50 × 50 mm × 2 mm. mm The terminal of the electrodes was Ni plated.The average surface roughness of the ceramic material was O.4 um, and the flatness was lessthan 3"m, The shape of the sample isshown in Fig,3. For the sintered samples, the bulk densitywas substrate
and
samples
50mm-
F----
for the in Fig,
force isshown 2, Oxides such as TiOz, Cr203, Mn02, CoO and CuO {Wako Jun-yaku,Supergrade) were added to 93% alumina ceramics (1wt%),and the material was formed intoa O,6 mm-thick green sheet by the conventional doctorblade procedure and cut intopieces. Tungsten paste was printed on the ceramics as elecmeasurement
ir/
aL,to5'F/bSiY..
electrostatic
2. Experimental procedure
The fabrication method
/i8SfiC
at
x2+EIR+H20
chuck
inthis range
the electrostatic
1
the
has not been investigated, When an electrostatic chuck isused to correct the fiatnessof a semiconductor wafer or to transport a wafer,
:,nxi:Ug"]
1
been developed, between the elec-
resistance
Dry
"
not
relationship
tcU..
/t "'-'s'n';l'e-nt [It'il'l/i-e,7er t tttttttt [I'e'L-1:,jll'ng
is necessarily increasedwithout decreasing the high electrostatic force,The appropriate range of the volumetric intrinsicresistance of the dielectriclayer isconsidered to be between 10i4 and so,
, /{nO .....-j,
CuO
TIC2
Cv20]
----r/
resistance
10iO9・cm or
r,ddltl,,,.s
ee'r'5L.I.L..poNder
Kx---------'--Nx"
2
N(..1
N
---LLJ
3
X.. 50mm
--J---iX)I2mM x
Fig.3. Schematicillustration of the electrostatic 1 : innerelectrode (tungstenrnetalization layer), 2 / insulation layer, :S/ electrode terminal Cnickel plating).
chuck
NII-Electronic NII-Electronic
sample.
Library Service
The CeramicSociety The Ceramic Society
of of Japan Japan
Toshiya WATANABE
measured
by
microstructure
means of
of
et
tij'the Cetumic,Se"'ety ,tfYipan100 fo"rnal
al.
Archimedes'method.
the sample,
which
was
The lapped by
diamond pastes,was observed with a scanning electron microscope (Hitachi, S-560).The composition was analyzed by an energy dispersion-typecharacteristicX-ray analyzer (Hitachi, KEVEX-7000). Phase identification of the sintered body was done by X-ray (CuKcr) diffiraction (,XRD), was followedinthe evaluation of elecJIS-C-2141 tricresistance and dielectric breakdown strength, forsamples of 50 mmdi and 2 mm thicknesswith Ag electrodes, The electric resistance was measured by a resistance measuring system CTakedaRika). The relative
TRIO C strength
dielectric constant
(AndoElectric).The was
measured
was
measured
were
according
to
Results and discussion
3.1 Fabricationof
electrostatic
chucks
Figure 4 shows therelationship between the sintering temperature and bulk density.The bulk density was decreased by the addition of an additive, but
3,5
using
Ee..h.2el:m
dielectric breakdown
JIS-C-2110;
between brass electrodes of 20 mmdi, and voltage was applied. The rate the voltage application was 1 kVls, Measurementsof electrostatic forcewere made by brassdisc pulling testsof a 30 mmth and 4 mm-thick which was held by an electrostatic chuck fixedin a vacuum chamber. The roughness of the brass plate was lessthan R.=O.1 "m, and the flatness was less than 1 um. The loading rate was approximately 20 kgfs. samples
3.
3
[']] 1992
sandwiched
3.0
Sinteringtemperature( and Fig,ture. 4. Relationship between bulk desity
SEM photographs ef cross section of alumina ceramics additive (SEM image), CSEM), 2-B:Ti02 (X-rayimage ef Ti), image of Cr}, 3-A:Cr203 (SEM), 3-B:Cr203 (X-ray image of Mn), 4-A : MnOz (SEM}, 4-B : Mn02 (X-ray image of Co), 5-A:CeO (SEM), 5-B:CoO {X-ray 6-A:CuO (SEM), 6-B:CuO (X-rayimage of Cu). Fig. 5. 1 :No 2-A:TiO,
with
transition metal
oxide
℃
)0 sintering
tempera-
additives.
NII-Electronic Library Service
The CeramicSociety The Ceramic Society
of of Japan Japan
4
Efflect of Additives
on
Electrostatic Force
the
Alumina Electrostatic Chucks
of
those of the samples sintered at 1580QCfor2h exceeded the density of 3.5gfcm3, Since it was clarified that sintering at 1580℃ results in the fabrication of a high-density body,all samples fabricated thereafter were sintered at 15800C for2 h, Figure 5 shows the result of the microstructural observation of each sample at 15800C). (sintered The results of the observation showed that the alumina
ceramics
with
and
without
additives
were
,6i
lo16
ld5
Eida,dlVh;-ti'
com-
grainsof 2-6 um in diameter and the grain boundaries.The distribution of additives inceramics varied greatly depend'ing on the type of transition metal oxide. In the case of Crx03, (Fig.5-A, B), the additive unjformly posed
of corundum
Si-rich phase
ld'iE
of
distributed in alumina
inthe case
lol
of
boundary. Identification of the phases by XRD
sample
dum,
a small
amount
ef
,di
L
zCuO
/p.
i " i
--・--2 2,5
2P
1/
MnO! CeO Cep03 3.0
35
1000XT{K")
Fig. 6.
present in the sintered that,in addition to corunmullite was present in all
showed
oTiO?
/
CoO, CuO and Ti02, the detectedas inclusions on the Si-richphase of the grain boundary. However, inthe case of Mn02, most of the additive melted uniformly in the Si-riehphase of the grain grains, and additives were
-
'.9・・''''''.o''''.b tt Noadditives
amics
Resistance-temperature transition metal oxide.
characteristics
of
alurnina
cer-
with
samples.
3,2 Electrical propertiesof
fabricated
chucks
Noadditives FErot-eusg・bl.ele TiOl-
Table 1 shows the electrical properties of the samconstant ples. The relative dielectric of all samples at 1 kHz was around 9. To determinethe insulation resistance, the Ieak current was measured after 10 min of voltage application, which is the time forthe current te become stable, The resistance value of the TiOrcontaing sample was 10i4n・cm, while those of the other samples were too high to measure at room temperature. Then the resistance measurements were
high temperatures,The results of for samples heated to in Fig,6. By the addition of Ti02,
conducted
the resistance
2000C are shown .resistance
over
changes
observed.
other
types of additives, of no more
than a
eflect oi additives was
three orders
However,
were was
o
cuoh MnO?CeOi Cke03''''''''''''P
t)'`if': /o/i .,i4'tf ,-'
..-, ,=.., --::"p.'J.-・・e"''-' ltX
o
looo
on
with
the
single
of
order
3000
Fig. 7. Electrostatic force of alumina electrostatic transitionmetal oxide. The adhesion forceis measttred tage applicatien for 20 s.
chuck
with
after
vol-
containing
samples
effect
magnitude
2oeo Voltage{V}
at
measurements
-
-
the additives magnitude. No
of
of
dielectric breakdownstrength
the samples
iorceshown
observed.
3.3 Electrostatic force of chucks fabricated Figure 7 shows the results of measurements electrostatic force.The average surface roughness
of
isabout R.=1 pm, The electrostatic inFig.7 was obtained after 20 s of volThe Ti02-added sampre exhibited a large force.Similarly,the samples inadditives produced largerforcesthan
tage application. remarkably of
cluding
other
the sample without Figure 8 shows
an
additive.
betweenthe electrostaticforceand the durationof voltage applicaTable 1 , Electrical properties of alumina with transitien metal tion at room temperature for samples with and oxide. without an additive. The applied voltage was 1000 V. In samples to which Ti02 was added, there was a Dt'eleot,ric ]roak d:'fi//i At/ / ti '., /ta/tS Hosistivj.ty Co/tstant clear dependence of electrostatic " t. 1lg /' forceon the duraClktrn) cocm) ClltHz) {kV) tion of voltage application; i.e., the electrostatic Nu itdriir/ives L).2・e,pO.zg./・V.2O,u 4x loJ/lxlOM >loi5 /・1."12,512.i11.3/・ force increased with an increase in the durationof '1/;xl・:n4/Lx/o'/1,lxlom4tlxloT4 ・L・1-Cp 1O /'">'/ois>lcls>lotE>lo/S voltage application. When the duration of applica・:uc'
the
relationship
es
/Iolt・Rge
e{nc/./ c,)c
cr :・s]
/1
,O1/'1.5
tion was extended greatly (approxirnately 30 min), the electrostatic ierce becarne saturated between 1
kgfcm2
and
veltage
was
1.2kg/cm2.Even terminated
and
when
the
application electrocles
NII-Electronic NII-Electronic
oi were
Library Service
The CeramicSociety The Ceramic Society
of of Japan Japan
Teshiya WATANABE
fij'the Ceramic'Socr'ely ",urnal
et al.
100
,V'fopan
/1.]1992
5
se2,o8gg1 F
i.o='aeo
loooEg-es'iS
5o 800FE?.M.
,vldO 6eosts-5'mo
Fig. 9.
o
Fig, 8, voltage
50
force did not reture to zero immediately, but graduallydeclined,This type of phenomena were not observed in the samples without
electrostatic
the effect
applied
of
the resistance
additives
oxide
force of alumiat elevated tem-
the
electrostatic
electrostatic
transitionmetal voltage is 1000 V.
with
on
of
the electrostatic
of
TiOz
added with
force. With the sample
alumina,
consisting
force in-
the electrostatic
ti.ncreasing durationof
applica-
voltage
in Fig.8. The saturated value of electrostaticforce was above 1 kgfcm2 (whenapplied voltage 1kV) and this large forcecan not be exby by plained the usual Coulomb forcerepresented Eq. (1),but itis attributable to Johnsen-Rahbeck's force.Therefore the time dependenceshould be exof the charge plained by the gradual accumulation from the inner electrode te the surface of the chuck resulting in a gradual increasein electrostatic force, tion as shown
on
force, the durationtime dependence of the electrostatic iorce were measured at high temthe sample was heated perature. In the experiment, in a vacuum chamber. The temperature was measured by a thermocouple attached to the sample, The the electrostatic
in Fig. 9 were obtained at 2000C. It is that the adhesion forceof the Ti02-containing sample heated at 2000C showed a lesstime depenresults
perature.The
creased
an additive.
In order to examine
chuck
force Balakrishnan5) reported the large electrostatic forcecaused by the Johnsen-Rahbeck's etibct, but he has not reported about the duration time dependence
Relationship between electrostatic forceand durationof The applied voltage is1000 V.
the
Durationtime dependence of
na electrostatic
3.4 Effecl/of
100 Time(sec)15e
application.
grounded,
1000
100
Time(see)
4oo22200
shown
obvious
dent characteristics, period of time with
Itbecame a smaller
saturated saturated
F Johnsen'Rahbeck's
in a
shorter
value.
Ba]altrlshntLn et
m
a].''t''
furc/t
1/
ee.:.2R
F=
.=
O.5E
=.
(vt・ti).
i
-t Time
m Fig, 1O.
Schematic presentation for relationship
between charge
distributionin the
layer and dielectric
electrostatic
force.
NII-Electronic NII-Electronic
Library Service
The CeramicSociety The Ceramic Society
of of Japan Japan
6
Effectof Adclitiveson
the
Electrostatic Ferce of ALumina Electrostatic Chucks
ing atmosphere. Electricaland electrostatic force itis assumed that the time required for the acwere estimated and the following depends on the resistance of the dielectric characteristics layer.Figure 1 O schows schematic illustration of such results were obtained. of Ti02 contributed to the relationship betweenelectrostatic force and charge Cl) The addition ceramics, in the dielectric layer. reduction of the resistance of alumina distribution while other additives made no contribution to the High temperature measurement of electrostatic reduction, force shows the relationship between the response the time (2) Withanelectrostaticchuckingsubstratecontime and the resistance oi the samples: i.e., the electrostatic force requierd forthe electrostatic forceto become saturatsiting of TiOz-added alumina, increasedwith increasing duration of voltage applicaed was foundto be much shorter at higher temperain tion. ture,The response time reduces with decreasing resistivity of the ceramics at higher temperature. As (3) When the temperature of the substrate was increasedto 2000C the electrostatic forcewas found shown inthe Fig. 6, the resistivity of alumina ceramicswithout additives was about 10iff9・cm and this to increase inproportionalto the decreasingamount with a distinctreduction in the value is assumed to be too high for the charge to of resistance response tlme. transferto the surface of the ceramic, therefore the From these results, itwas concluded that a large electrostatic force isdescribed by the usual Coulomb On the other hand, the electrostatic force dependent on the duration of volforcerepresented by Eq, (1), tage application can be generatedforan electrostatic Ti02 added sample having the resistivity below 10i4 chuck made of alurnina centaining Ti02 which has a 9・cm is considered to have the transitional characrelatively high resistance (approximately 10it9). force from the usual teristicsof electrostatic Coulomb forcetoward the largerJohnsen-Rahbeck's References force.Otheradditives except Ti02 was not so effk)c1) G. A. Wardly, Rev.Sch,instn{m.,44, 1506-09 (1973), tiveto control theresistance of the aiumina ceramics 2) S.Tach,J IEE]bpan,108. 1195-98 (1988). as compared with TiO2,Therefore they do not im3) M. Nakamura, T, Kurimoto, H, Yano and K. Yanagida, elecprove the electrostatic force of the alumina EDD-88-r 43, 17-26 (1988>and
cumlation
trostatic chuck,
"Handbeek
4.
Conclusions
Ceramicelectrostatic chucks foradhesion ments were fabricated by adding 1 wt% the additive
93%
alumina
of 4) The Instituteof Electrestatics Japan, Elecrostatics" {1981)pp. 672-75. 5} C. Balakrishnan, Br,J.APPI.Rlays., 4, 211-13 C1950). 6) M.A.KhMa,Z.M.HanatiandA.K.Mohamed,ThermochimicaActa, 56, 291-98 (1982), I.Yamai and H. Saito,Yiogyo-Kbeokai-Shi, B3, 7) J.Takahashi, 589-94 (1975).
Ti02, Cr203, MnO, ceramics
and
CoO
sintering
and
them
experiof
each of
CuO, to inreduc-
NII-Electronic NII-Electronic
Library Service