Toward environmentally friendly photolithographic materials: A new class of water-soluble photoresists

Article abstract of DOI:10.1021/ma034461r

New water-soluble styrenic polymers bearing two functional groups, pendant ammonium salts of half-esters of malonic acids and acid-labile alkyl esters, were synthesized and evaluated for water-soluble positive-tone photoresist application. These polymers feature two solubility switches: insolubilization of the entire film by baking and selective solubilization upon exposure to UV light. Time-resolved FT-IR measurement of the baked films showed sequential evaporation of ammonia from the films and the decarboxylation of the malonate half-esters. The rates of decarboxylation depend on the structure of substituents at the 2-position of the malonates. The choice of acid-labile esters, the structure of the half-esters, and the polymer compositions were carefully optimized, and high-resolution positive tone images were obtained that were fully processed in aqueous media.

Full text of DOI:10.1021/ma034461r

Macromolecules 2004, 37, 377-384
377
Toward Environmentally Friendly Photolithographic Materials: A New
Class of Water-Soluble Photoresists
Sh in ta r o Ya m a d a ,† Th om a s Mr ozek ,‡ Tim o Ra ger ,§ J or d a n Ow en s,
J ose Ra n gel, a n d C. Gr a n t Willson *
Department of Chemistry and Chemical Engineering, The University of Texas at Austin,
Austin, Texas 78731
J effer y Byer s#
International SEMATECH, Austin, Texas 78741
Received April 10, 2003; Revised Manuscript Received August 12, 2003
ABSTRACT: New water-soluble styrenic polymers bearing two functional groups, pendant ammonium
salts of half-esters of malonic acids and acid-labile alkyl esters, were synthesized and evaluated for water-
soluble positive-tone photoresist application. These polymers feature two solubility switches: insolubi-
lization of the entire film by baking and selective solubilization upon exposure to UV light. Time-resolved
FT-IR measurement of the baked films showed sequential evaporation of ammonia from the films and
the decarboxylation of the malonate half-esters. The rates of decarboxylation depend on the structure of
substituents at the 2-position of the malonates. The choice of acid-labile esters, the structure of the half-
esters, and the polymer compositions were carefully optimized, and high-resolution positive tone images
were obtained that were fully processed in aqueous media.
Sch em e 1. Deca r boxyla tion of Ca r boxylic Acid s a n d
In tr od u ction
Th eir Sa lts (SH ) P r oton -Don a tin g Gr ou p )
Recent economic strategies are becoming more and
more sensitive to environmental issues.1 Efforts have
been made to reduce the amount of waste organic sol-
vent produced by the photolithographic processes. This
would not only have a significantly positive impact on
the environment but also reduce waste disposal costs.2
Replacing the organic solvents in photoresists with
water, i.e., performing the wafer coating from an aque-
ous solution, would greatly reduce this waste stream.
Initial approaches to design of resists for aqueous
processing were based on solubility switching by either
photoinduced cross-linking reaction3 or polarity change4
in polymers. These early systems lacked the required
photosensitivity. To circumvent the sensitivity issue, our
group in collaboration with professor J . M. J . Fre´chet’s
group at University of California at Berkley exploited
the concept of a chemical amplification to develop a
variety of sensitive negative tone systems using cross-
linking mechanisms.5 Alternatively, switching mecha-
nisms featuring polarity change were developed on the
basis of ring-closure reactions of maleic acid6 or the
pinacol rearrangement.7
had to be incorporated into the resist material, which
makes the realization of the design rather challenging.
Our approach involves using the post-application bake
(PAB) as the means to initially insolubilize the water-
borne films and subsequent solubilization of exposed
areas using acid-catalyzed chemical amplification type
deprotection reactions. More specifically, the PAB not
only removes casting solvent (water) from the film but
also renders the film water-insoluble by inducing a
chemical reaction. By incorporating a photoacid genera-
tor (PAG), acids generated only in the exposed areas
catalyze another type of reaction that renders the ex-
posed areas water-soluble. The concept of the two solu-
bility switches has been demonstrated by our group us-
ing a cross-linking-de-cross-linking reaction of vinyl
ethers.7 However, this system showed poor dry-etch
resistance and short shelf life. We therefore sought
another switching mechanism, which has more practical
performance. One approach to achieving the two changes
in solubility is by combining a decarboxylation of car-
boxylic acids and an acid-catalyzed ester cleavage. Ap-
propriately substituted carboxylic acids can be decar-
boxylated either as the free acid or in the salt form.9
â-Keto carboxylic acids, for example, undergo thermal
decarboxylation to produce the ketone and carbon diox-
ide, typically at temperatures below 100 °C. The mech-
anism is believed to involve a six-membered cyclic
transition state (Scheme 1).10
In this paper we present the results of work on a new
approach to the design of water-soluble positive tone
photoresist materials.2,8 By definition, positive tone
imaging generates solubility in exposed areas while the
unexposed areas remain insoluble. All the components
of the photoresist have to be water-soluble in order to
cast a film from water. Therefore, two solubility switches
† Current address: Shipley Company, 455 Forest Street, Mar-
lborough, MA 01752.
‡ Current address: Great Lakes Manufacturing GmbH, Teplitzer-
str. 14-16, 84467 Waldkraiburg, Germany.
§ Current address: MPI for Solid State Research, Heisenberg-
strasse 1, 70569 Stuttgart, Germany.
Current address: International SEMATECH, Austin, TX
78741.
It is one of a subset of carboxylic acids in which the
corresponding salts decarboxylate faster than the free
acids at the same temperature. For example, 2-cyano-
2-(4-phenylbutanoic acid)11 and phenylbis(phenylthio)-
#
Current address: KLA-Tencor Texas-Finle Division, Austin,
TX 78759.
* To whom correspondence should be addressed: e-mail
willson@che.utexas.edu.
10.1021/ma034461r CCC: $27.50 © 2004 American Chemical Society
Published on Web 12/24/2003

378 Yamada et al.
Macromolecules, Vol. 37, No. 2, 2004
Ta ble 1. Decom p osition Tem p er a tu r e of P olya cr yla tes in
th e P r esen ce of Tr ip h en ylsu lfon iu m
Hexa flu or oa n tim on a te15
Sch em e 2. Design Con cep t of th e Deca r boxyla tion
System (R ) Acid -Clea va ble Gr ou p )
a
b
With exposure at 248 nm. Without exposure at 248 nm.
Molecular weights are reported relative to polystyrene stan-
dards. Polymers containing acidic functional groups were
treated with either diazomethane or trimethylsilyldiazometh-
ane before GPC measurement. Differential scanning calorim-
etry (DSC) measurements and thermal gravimetric analyses
(TGA) were performed by a Perkin-Elmer series 7 thermal
analysis system.
F ilm An a lysis of P olym er s w ith F T-IR. An aqueous
solution of polymers was prepared by simply mixing solid
polymer (0.3 g), water (1.7 g), and ammonium hydroxide
(approximately 30 wt %, 0.12 g). Polymer 6ib gradually
dissolved and a clear solution was obtained after a few hours
stirring. If the polymer was not soluble in aqueous ammonia,
dioxane was used instead. This solution was spin-coated on
gold-coated, double-polished silicon wafers. The films were
baked at various temperatures, and real-time IR spectra were
collected using Nicolet Avatar 360 IR spectrometer.
P h otolith ogr a p h y. Preliminary evaluations were carried
out using a Headway spin-coater, a J BA LS65 1 kW exposure
system with a 248 nm band-pass filter from Acton Research,
and an Optoline quartz mask for contact printing. The resist
was projection-printed on Ultratech XLS 248 nm exposure tool
(NA ) 0.53; σ ) 0.74) at SEMATECH or on an ASM
lithography PAS 5500 248 nm stepper (NA ) 0.63; σ ) 0.6) at
Shipley Co. The exposed films were developed with a Shipley
MF CD-26 developer. Etch rates were measured on an LAM
9400 PTX tool operating at 450 W/45 W with a pressure of 20
mTorr and employing 75 sccm of Cl2.
acetic acid12 decarboxylate more slowly as the free acids
than they do in the presence of amines.13 On the other
hand, most carboxylic acids including malonic acid
decarboxylate faster than their salts.14 On the basis of
this knowledge, we have chosen the ammonium salt of
a half-ester of malonic acid as the core unit for the two
solubility switching, as illustrated in Scheme 2.
The ammonium salts of the half-esters of malonic
acids provide the initial solubility for the polymer and
are comparatively stable in aqueous base solutions.
During the PAB, the ammonia is volatilized and the free
acid decarboxylates. This decarboxylation reaction con-
verts the acidic polymer into a neutral lipophilic species
that is insoluble in aqueous base. Once the film is
insolubilized, photoacid-catalyzed thermolysis of an
ester, the reaction commonly used in chemically ampli-
fied resists, renders exposed areas soluble in aqueous
base developer, providing positive tone images. To
ensure that our new photoresist provides sufficient dry-
etch resistance, a styrene backbone was chosen for this
design. In the following pages, we describe the synthesis
and evaluation of water-soluble polymers designed for
positive tone photoresist systems that can be fully
processed in aqueous media.
Resu lts a n d Discu ssion
Ch oice of Acid -La bile Gr ou p s. The choice of acid-
labile ester group is critical. Although half-esters of
malonic acids are expected to exhibit the switching
properties discussed above, i.e., thermally induced de-
carboxylation and subsequent acid-catalyzed deprotec-
tion of the ester, the selectivity of these two reactions
is an issue. Taking into account that the esters might
be accidentally cleaved via pyrolysis in the course of the
PAB process, we had to find a protecting group that is
thermally stable to survive the bake yet cleavable by
photo acids in order to guarantee an efficient double
switching function. Several polyacrylates with different
ester groups were prepared, and the decomposition
temperature of the polymers in the presence and
absence of photogenerated acids was measured. Three
secondary esters were tested in hope of providing better
thermal stability compared to tert-butyl ester, a com-
monly used acid-labile functionality in commercial 248
nm photoresists. A variety of acid-labile esters were
auditioned, and the results are shown in Table 1.
The cyclohexenyl (III) and 2-cyclohexanonyl (IV)
acrylates decompose at higher temperatures than poly-
(tert-butyl acrylate) (I). They are also stable to 200 °C
in the presence of photogenerated acid. Poly(isobornyl
acrylate) (II) has a relatively low decomposition tem-
perature in the presence of photogenerated acid but is
more thermally stable than poly(tert-butyl acrylate).
Trefonas et al. showed that the activation energy of acid-
Exp er im en ta l Section
Ma ter ia ls. All chemicals were purchased (Aldrich, Acros,
Fluka, Merck) and used as received unless noted otherwise.
Tetrahydrofuran (THF), diethyl ether, and 1,4-dioxane were
refluxed over sodium benzophenone ketyl and then distilled.
tert-Butyl alcohol was refluxed over CaH2 and distilled. 2,2′-
Azobis(isobutyronitrile) (AIBN) was recrystallized from etha-
nol. All the photoacid generators were obtained from Midori
Kagaku Co. except triphenylsulfonium nonaflate (TPS-Nf),
which was obtained from Clariant Corp. and Shipley Co. The
details of the syntheses of all the monomers and polymers are
described in Supporting Information.
Mea su r em en ts. FT-IR spectra were collected using a
Nicolet Avatar 360 IR spectrometer. Spectra of films and
liquids were collected using doubly polished silicon wafers,
NaCl, KBr, or KRS-5 crystal plate as substrates. KBr pellets
were prepared for solid samples. 1H and 13C spectra were
recorded on a 300 MHz Varian Unity Plus 300 spectrometer
(1H, 300 MHz; 13C, 75 MHz) and were referenced to the solvent
or tetramethylsilane. Mass spectra were measured on
a
Finnigan MAT TSQ-700 spectrometer. Gas chromatographs
were recorded on a Hewlett-Packard 5890 series II with a
capillary column HP-5 (cross-linked 5% phenylmethylsiloxane)
and flame ionization detector. Molecular weights were mea-
sured from THF solutions using a Viscotek gel permeation
chromatograph (GPC) equipped with a set of two columns
(linear mix and 100 Å) from American Polymer Standards.

Macromolecules, Vol. 37, No. 2, 2004
Water-Soluble Photoresists 379
Sch em e 3. Syn th esis of Vin ylben zyl-Su bstitu ted
Der iva tives: (a ) Mg, DMF ; (b) Meld r u m ’s Acid ,
P ip er id in e, AcOH; (c) Na BH4; (d ) R1OH, BHT
(in h ibitor ); (e) AIBN; (f) LDA, MeI; R1 ) tb
(ter t-Bu tyl), ib (Isobor n yl)
presence of piperidine and acetic acid afforded com-
pound 3, which was immediately converted to the
monosubstituted Meldrum’s acid 4 via reduction with
NaBH4.18 No reduction of the styrenic double bond was
observed. At this stage, ring opening was necessary
prior to polymerization in order to circumvent gel
formation.19 Refluxing 4 and the alcohol, i.e., tert-butyl
alcohol or isoborneol in dioxane, respectively, success-
fully achieved this goal. In this case, the presence of
added radical inhibitor 2,6-di-tert-butyl-4-methylphenol
(BHT) was important in order to prevent oligomeriza-
tion. BHT is reported to open the ring,20 but an excess
of alcohol and a catalytic amount of BHT gave excellent
selectivity, and the desired half-esters 5tb and 5ib were
obtained without a measurable amount of BHT adduct.
Polymerization of 5tb and 5ib with AIBN produced the
homopolymers 6tb and 6ib, respectively. The am-
monium salts of these polymers are water-soluble.
Monomer 5ib was selectively methylated with methyl
iodide using 2 equiv of lithium diisopropylamide (LDA).21
A simple alkylation reaction of half-esters of malonates
with 4-vinylbenzyl chloride was also used to prepare
these monomers. However, it required more steps and
tedious purification.
catalyzed deprotection of the isobornyl ester is about 2
times that of tert-butyl ester.16 Isobornyl esters are more
thermally stable than tert-butyl ester under acidic
conditions but are still cleavable with photoacids. From
these results we conclude that both the tert-butyl ester
and the isobornyl ester are candidates for integration
as photo acid labile switching units.
The vinylphenyl-substituted monomers were prepared
by modification of the reported procedure by Fre´chet et
al.13 The Grignard reaction of vinylbenzyl chloride, 9,
with carbon dioxide (50% yield) followed by esterification
with isoborneol and trifluoroacetic anhydride (TFAA)
afforded isobornyl (4-vinylphenyl)acetate, 11ib (30%
yield). The methyl ester, 11m e, was prepared with
methyl iodide using 1,8-diazabicyclo[5.4.0]undec-7-ene
(DBU) as a base (54% yield). The esters 11 were
methylated with methyl iodide and LDA. Carboxylation
of the resulting compounds 12 with LDA and carbon
dioxide afforded monoacids 13. Polymerization of 13
with AIBN afforded the desired homopolymer 14. Al-
ternatively, separation of the acid-labile group from the
thermally sensitive functionality resulted in copolymer-
izing the isobornyl derivative 11ib or 15ib with mono-
mer 13m e to give copolymers 17 and 16, respectively.
The copolymer ratios of polymer 17 and 16 were
P r ep a r a tion of P olym er s. In the design of the
synthesis of the final target polymer, we decided to
choose styrenic monomers because of the relative ease
of their polymerization with conventional free radical
polymerization and their ability to impart dry-etch
resistance in the lithographic process. The styrene unit
was linked to half-esters of malonic acids in a variety
of ways, as shown in Scheme 3 and Scheme 4. Three
patterns of substitution at the R-position of the malonate
(vinylbenzyl, vinylbenzyl + methyl, vinylphenyl + meth-
yl) and two acid-labile groups, tert-butyl and isobornyl,
and a methyl group were prepared for comparison. The
vinylbenzyl-substituted monomers 5tb and 5ib were
prepared from 4-chlorostyrene, 1, in four steps. The
Grignard reagent prepared from 4-chlorostyrene was
reacted with N,N-dimethylsulfonamide (DMF) followed
by hydrolysis with ammonium acetate, affording 4-vi-
nylbenzaldehyde, 2, in 50% yield.17 A Kno¨venagel
condensation of this aldehyde and Meldrum’s acid in the
1
determined by H NMR, and polymer 17 contains 31%
of 11ib and polymer 16 contains 34% 15ib.
Th er m a l Gr a vim etr ic An a lyses. TGA was used to
determine the selectivity between decarboxylation and
Sch em e 4. Syn th esis of Vin ylp h en yl-Su bstitu ted Der iva tives: (a ) Mg, CO2; (b) TF AA, R2OH; (c) LDA, MeI; (d )
LDA, CO2; (e) AIBN, 13m e; (f) AIBN, 13Me; R2 ) ib (Isobor n yl), m e (Meth yl)

380 Yamada et al.
Macromolecules, Vol. 37, No. 2, 2004
F igu r e 1. TGA thermograms of polymers 6tb and 6ib.
Heating rate: 10 °C/min.
decomposition of the ester. Figure 1 shows the thermo-
grams of polymers 6tb and 6ib. These polymers exhibit
significantly different decomposition features. In the
case of 6tb, weight loss starts around 130 °C, and the
first loss is about 35 wt %. This corresponds to the
theoretical weight loss for the combined loss of CO2 (15.9
wt %) and isobutene (20.2 wt %). Polymer 8tb showed
the same result. The selectivity between the decarboxy-
lation of the acid group and thermolysis of the tert-butyl
ester is not as high as we had hoped. This poor
selectivity was also confirmed in a model compound
1
study using H NMR.22 The calculated loss for 6ib is
12% for CO2 and 39% for the protecting group. Thus,
the first weight loss represents only decarboxylation,
and the second weight loss is due to the loss of the
protecting group. The thermogram shows a distinct
division between decarboxylation and ester deprotection.
The selectivity between the thermolysis of the malonic
acid and isobornyl ester is quite high for this new
polymer 6ib. Other polymers bearing isobornyl and
methyl groups showed the same selectivity.
F igu r e 2. FT-IR spectra of ammonium salts of polymers 8tb,
6tb, and 6ib after baking. Total baking time at 165 °C: 10
min.
F ilm Stu d y w ith F T-IR. To more precisely analyze
the decarboxylation and decomposition reactions of tert-
butyl and isobornyl ester, respectively, an FT-IR study
was carried out with spin-coated films of polymers 6tb,
6ib, and 8tb cast from aqueous ammonia. Figure 2
shows the IR spectra of polymers 6tb, 6ib, and 8tb
undergoing baking at 165 °C. Rapid volatilization of
ammonia can be observed as indicated by the disap-
pearance of the carboxylate anion peak at ca. 1580 cm-1
accompanied by the increase in the carboxylic acid
carbonyl stretching band around 1715 cm-1 in all three
cases. The sequential decarboxylation can be observed
as indicated by the decrease in the carbonyl peak
intensity. Whereas primarily the high-frequency part
(1743 cm-1) of the carbonyl peak from 6tb and 8tb
significantly decreases in intensity as baking time
increases, a similar observation can be made only for
the low-frequency part (∼1714 cm-1) of the same IR
peak from 6ib. The thermal behavior of the latter
polymer clearly indicates that decarboxylation is the
only reaction that occurs after evaporation of ammonia.
In the case of 6tb and 8tb, decarboxylation is not the
only reaction. The decrease of the peak intensity at 1743
cm-1 (νCdO of ester) as baking time increases indicates
that the tert-butyl ester is subject to concomitant
decomposition. Moreover, evolution of a new peak at
1812 cm-1 is observed for 6tb but not for 8tb. The new
peak at 1812 cm-1 can be assigned to νCdO of anhydride,
and this indicates that some of the carboxylic acid in
polymer 6tb is subject to anhydride formation during
baking. No peaks that correspond to the anhydride are
observed in 8tb, indicating that increased steric hin-
drance by the extra methyl group at the R-position
effectively inhibits this side reaction.
A more detailed study of the release of ammonia and
of the decarboxylation (Scheme 5) of 6ib was conducted
in which the polymer films were baked as before. The
changes in the IR spectrum were monitored over time.
Figure 3 shows the time evolution of changes in COO-
stretch (1628-1527 cm-1). As ammonia evaporates, the
COO- stretch of the ammonium salt of 6ib decreases.
The ammonia evaporated within 6 s.
Figure 3 shows changes over time in the total CdO
stretches of the carboxylic acid and the ester (1786-
1660 cm-1). As ammonia evaporates, the carboxylate
turns to carboxylic acid, which has an IR peak at 1721
cm-1, and this peak decreases as decarboxylation takes
place. The evaporation of ammonia is much faster than
the decarboxylation. Thus, a steep increase in the peak
intensity is observed during the early stage of baking.
The initial intensity was normalized to unity because
theoretically there was only a CdO stretch of the ester

Macromolecules, Vol. 37, No. 2, 2004
Water-Soluble Photoresists 381
F igu r e 3. Time evolution of changes in CdO stretch (1786-
1660 cm-1) of polymer 6ib-ammonium salt. Inset: baking
temperatures in °C.
Sch em e 5. Tr a n sfor m a tion of Am m on iu m Sa lt of 6ib
F igu r e 4. Arrhenius plots of 6ib: (a) volatilization of am-
monia; (b) decarboxylation.
group before baking. As ammonia evaporates, the
CdO of carboxylic acid increases, and eventually the
peak intensity reaches 2 if decarboxylation reaction
during the evaporation of ammonia is negligible. The
peak intensity then decreases as decarboxylation takes
places. The peak intensity should return to unity after
decarboxylation is completed, but in fact it was less than
unity. This may be due to the time lag between the spin-
coating and the IR measurement, which may have
resulted in some premature volatilization of ammonia
that would affect the initial signal intensity.
Reaction rate constants were calculated using the IR
concentration data. Although evaporation of ammonia
and decarboxylation (A f B f C) are sequential
reactions, the rate constants could be calculated inde-
pendently because volatilization of ammonia was so
much faster than decarboxylation.23 The time evolution
of the IR signal intensities for both reactions is well
described by simple exponentials, as expected for first-
order kinetics. Arrhenius plots were generated for the
temperature-dependent rate constants (Figure 4a,b),
and the activation energies of both reactions were
calculated. The activation energy of ammonia volatiliza-
tion is 9.2 kcal/mol, and the activation energy of
decarboxylation is 39.6 kcal/mol. Wallraff et al. reported
kinetics studies on thermal decomposition of poly(tert-
butyl methacrylate).24 The activation energy calculated
for the thermal decomposition of poly(tert-butyl meth-
acrylate) was 40.9 kcal/mol, close to that of decarboxy-
lation. This explains the low reaction selectivity of the
previous polymer bearing tert-butyl ester (6tb).
The decarboxylation kinetics of polymers 6ib , 8ib ,
14ib, 14m e, 16, and 17 were studied using the same
IR equipment. The ammonium salts of polymers 8ib and
14ib were not water-soluble; hence, all the polymers
were dissolved in 1,4-dioxane. These solutions were
spin-coated on gold-coated, double-polished silicon wa-
fers. The films were baked at various temperatures, and
IR spectra were collected as described previously. Figure
5 shows the time evolution of changes in CdO stretches
F igu r e 5. Time evolution of changes in CdO stretch (1786-
1660 cm-1) of polymers 6ib, 8ib, 14ib, 14m e, 16, and 17.
of carboxylic acid and ester (1786-1660 cm-1) for the
six polymers with baking at 180 °C.25 IR data of the six
polymers at different baking temperatures are shown
in Figures 12-17 in the Supporting Information. The
results are summarized in Table 2. Decarboxylation of
polymers 6ib and 8ib followed first-order decay kinetics.
No anhydride formation was observed with polymer 8ib
while 6ib showed anhydride peak at higher baking
temperature than 170 °C. Interestingly, the kinetics of
the thermolysis of polymers 14m e, 16, and 17 showed
autocatalytic-type behavior. The rate of decarboxylation
slowly increases as baking time increases. From the
mechanical point of view, the decarboxylation reaction
follows first-order kinetics. However, the mechanism is
apparently dependent upon the polymer structure.
Various factors can affect the rate of decarboxylation,
especially in a film. For example, the Tg and polarity of
the polymer change drastically as decarboxylation pro-
ceeds. The polymer films are initially highly polar, and
their Tg values are high. As decarboxylation takes place,
the Tg and polarity decrease. If the Tg changes, the
mobility of the polymer chain changes, and this could
influence the reaction rates. The bulkiness of the ester
group may also affect the ease of formation of the six-
membered ring to induce decarboxylation. The decar-

382 Yamada et al.
Macromolecules, Vol. 37, No. 2, 2004
Ta ble 2. Com p a r ison of P olym er s 6ib, 8ib, 14ib, 14m e, 16, a n d 1726
anhydride
formationa
order of
kinetics
T(de-CO c at
)
2
180 °C (min)
polymer
Ea (kcal/mol)
Ae
aq solf
6ib
8ib
14ib
14m e
16
Y
N
N
N
N
N
1st
1st
1st or 0
autob
autob
autob
8.3
27.9
2.5
2.9
1.7
29.4
30.3
36.2d
5.09 × 1013
1.16 × 1015
3.53 × 1017
S
I
I
S
S
S
17
1.6
a
b
d
Y ) observed, N ) not observed. Autocatalytic type behavior. c Time for complete decarboxylation. Calculated from first-order
fitting. e Arrhenius preexponential factor. f Aqueous solubility: S ) soluble; I ) insoluble.
F igu r e 7. TGA thermogram of PAGs.
F igu r e 6. Change in normalized film thickness of polymer
6ib. Baking time: 5 min.
boxylation rate dramatically increased when one me-
thylene unit was removed from polymer 8ib. Complete
decarboxylation occurs 2-3 times as fast as in that of
polymer 6ib.
Lit h ogr a p h ic E va lu a t ion s. Ba k in g Test a n d
Eva lu a tion of P AG. On the basis of the results of the
thermal analysis and aqueous solubility of the polymers,
lithographic evaluation of polymer 6ib was performed.
An aqueous solution of the ammonium salt of polymer
6ib was coated on silicon wafers, and the baking
temperature was varied in order to determine the
threshold temperature for effective film retention to-
ward an aqueous developer. All the films were baked
for 5 min and immersed in TMAH (2.38 wt %) for 30 s
for development. Figure 6 shows the normalized film
thickness after development. One can see that the
polymer films become insoluble in the developer when
baked above 150 °C. This positive result proves that the
selectivity between decarboxylation and deprotection of
the isobornyl ester, as shown by the TGA and IR
analysis, is sufficiently high to perform the first ther-
mally induced solubility switch. It is also seen from
Figure 6 that the film has swollen by approximately 20%
after development; this phenomenon will be discussed
in detail later. When the polymer was mixed with a
water-soluble PAG, 4-methoxyphenyldimethylsulfonium
triflate (MPDMS-Tf), there was no film retention under
the same baking conditions.
Figure 7 shows the TGA thermogram of MPDMS-Tf,
triphenylsulfonium triflate (TPS-Tf) and -nonaflate
(TPS-Nf), which are widely used in various chemically
amplified resist systems. MPDM-Tf decomposes at lower
temperatures than TPS-Nf or TPS-Tf, starting slowly
around 150 °C. The thermal decomposition of the PAG
generates acid, which in turn cleaves the isobornyl
protecting group rendering the film soluble. This finding
led us to omit MPDMS-Tf from further experimentation.
TPS-Tf and TPS-Nf are both thermally stable, but TPS-
Nf and TPS-Tf are not as water-soluble as MPDMS-Tf.
TPS-Nf and TPS-Tf are slightly soluble in water, at least
F igu r e 8. Change of normalized thickness of polymer 6ib
with TPS-Nf; baked at 165 °C.
2.0 wt % of the PAGs based on a matrix polymer.
Therefore, it was possible to dissolve the PAG in an
aqueous solution of polymer 6ib. An aqueous solution
of ammonium salt of polymer 6ib and TPS-Nf (1.5 wt
% to polymer) was spin-coated on a silicon wafer and
baked at 165 °C for 2-10 min. A dramatic change in
solubility can be seen between baking times of 3.5 and
4 min (Figure 8). Decarboxylation gives a decrease in
acid concentration, and the solubility threshold is
around 4 min baking at 165 °C. Longer baking times
minimize film swelling.
The same solution was spin-coated on a gold-coated,
double-polished silicon wafer. The film was baked at 165
°C for 5 min, exposed at 248 nm, and postexposure
baked at 140 °C for 1 min. Figure 9 shows FT-IR spectra
of the films before exposure and after PEB. The increase
in peak intensity at 3432 cm-1 (OH stretch) and 1710
cm-1 (CdO stretch of acid) and decrease in peak
intensity at 2953 cm-1 (CH2) and 1733 cm-1 (CdO
stretch of ester) clearly demonstrate acid-catalyzed
deprotection of the isobornyl ester.
Im a gin g Test. Imaging tests at SEMATECH were
conducted on a photoresist sample composed of polymer
6ib, TPS-Nf (1.5 wt %), water, and ammonium hydrox-
ide. Silicon wafers were unprimed or primed with either
antireflective coat DUV-30 (60 Å thickness) or HMDS.
The best adhesion was obtained with DUV-30. Baking

Macromolecules, Vol. 37, No. 2, 2004
Water-Soluble Photoresists 383
F igu r e 11. Top-down and cross-section SEM pictures of
positive tone images of polymer 18.
able around 1800 cm-1 for 6ib as shown in Figure 2,
an anhydride peak emerged with higher baking tem-
perature and longer baking time. Therefore, the cause
of cross-linking must be the anhydride formation. Figure
2 also shows that a significant amount of carboxylic acid
is left after baking for 5-10 min at 165 °C. With baking
at 165 °C for 5 min, the small amount of cross-linking
helped the film render insoluble in the developer.
However, the small amount of cross-linking and the
residual carboxylic acid caused swelling of the baked
film as shown in Figures 6 and 8 and smaller features
as shown in Figure 10 (III). Baking at 165 °C for 10
min produced too much cross-linking, and thus no
solubility change was obtained in the exposed areas.
F igu r e 9. FT-IR spectra of polymer 6ib with TPS-Nf before
and after exposure and subsequent PEB. Exposure energy: 30
mJ /cm2. Inset: spectra range from 1500 to 2000 cm-1
.
Op tim iza tion of P olym er s. Copolymerization of two
(or more) monomers with each of them being responsible
for only one of the two solubility switches will offer
additional variables (i.e., the comonomer ratios) for
improving the imaging quality, i.e., eliminating the
swelling phenomenon. Because of its fast autocatalytic
decarboxylation, its high aqueous solubility and its
reduced tendency for anhydride formation (Table 2),
monomer 13m e seems to be particularly suited for
imparting the initial solubility switching mechanism
into the target copolymer without causing an observable
swelling problem. Monomer 11ib, which obviously is
equivalent to monomer 15ib when incorporated in the
polymer, is supposed to represent the second solubility
switch. Imaging experiments with 17 revealed that the
concept of double solubility switching is transferable to
copolymers. However, the decarboxylation of 17 was so
fast that the resulting film after PAB became hydro-
phobic. Therefore, further optimization on polymer
composition was conducted in order to impart the initial
aqueous solubility, the two sequential solubility switches,
and sufficient polarity after decarboxylation. Two ad-
ditional monomers, monomer 10 and 4-acetoxystyrene
(AcOSt), were introduced into polymer 17 for this
optimization (Scheme 6). The acetoxy group of AcOSt
is easily hydrolyzed by aqueous ammonia to generate
phenolic groups that not only provide a moderate
dissolution rate in the TMAH developer but also im-
prove adhesion of the photoresist.27 One drawback using
phenolic group is that the initial solubility of the
polymer in aqueous ammonia is decreased, and this
narrows the options for optimizing the composition.
Monomer 10 does not undergo decarboxylation and
compromises this problem, but the amount should be
minimized in order to avoid swelling problem as seen
in Figure 10 (III). An imaging experiment with a
photoresist sample composed of polymer 18 (10/11ib/
13m e/AcOSt ) 3/23/49/24), TPS-Nf (2.0 wt %), water,
and ammonium hydroxide was conducted, and 0.4 µm
resolution positive tone images were obtained, as shown
in Figure 11. These results demonstrate that a photo-
resist based on the tetrapolymer 18 is capable of
generating high-resolution images and represent an
improvement of the simple homopolymer version, i.e.,
F igu r e 10. SEM pictures of positive tone images with polymer
6ib. ARC: DUV 30, 61 nm; film thickness ) 0.6 µm. PAB:
165 °C/5 min. Exp: 248 nm stepper. PEB: 140 °C/5 min.
Developer: MF CD-26/30 s.
Ta ble 3. Com p a r ison of Etch Ra tes
APEX-E
(DUV)
SPR-510L
(I-line)
polymer 6ib
etch rate (Å/min)
758
698
443
ratio
1.09
1.00
0.63
Ta ble 4. Ch a n ge of Molecu la r Weigh t by Ba k in g
Mn
Mw
PDI
polymer 6ib
resist film (140 °C/5 min)
resist film (165 °C/5 min)
11 800
11 900
11 400
24 600
27 000
61 500
2.09
2.27
5.40
at 165 °C for 5 min produced a 0.6 µm thick film. The
insolubilized film was exposed to 248 nm radiation (20
mJ /cm2) and baked at 140 °C for 1 min. Development
with Shipley MF CD-26 (TMAH 2.38 wt %) for 30 s
produced 1 µm resolution positive tone images, as shown
in Figure 10.
The etch rate of photoresist film containing polymer
6ib was measured in Cl2 plasma and compared to that
of commercially available resists (Table 3). Its etch
stability is not as good as the conventional I-line resist,
SPR-510L, but it is comparable to the DUV resist APEX-
E.
With longer baking time, e.g. 10 min, a smaller
increase in thickness was observable after development
(Figure 8). However, no image was obtained. This
suggests the possibility of some side reactions, such as
thermally induced cross-linking, during the decarboxy-
lation process. Table 4 summarizes the change in
molecular weight of the films after baking. The initial
polymer 6ib has Mw of 24 600. Baking at 140 °C induces
no significant change in molecular weight. However, the
GPC chromatogram showed a tailing in higher molec-
ular weight region after baking at 165 °C for 5 min, and
the resulting Mw increased significantly. The molecular
weight of the film baked at 165 °C for 10 min was not
measured because the film was not soluble in the GPC
solvent. Although there was no anhydride peak observ-

384 Yamada et al.
Sch em e 6. Syn th esis of Tetr a p olym er
Macromolecules, Vol. 37, No. 2, 2004
(5) (a) Lin, Q.; Steinha¨usler, T.; Simpson, L.; Wilder, M.; Me-
deiros, D. R.; Willson, C. G.; Havard, J . M.; Fre´chet, J . M. J .
Chem. Mater. 1997, 9, 1725. (b) Havard, J . M.; Shim, S. Y.;
Fre´chet, J . M. J .; Lin, Q.; Medeiros, D. R.; Willson, C. G.;
Byers, J . D. Chem. Mater. 1999, 11, 719. (c) Havard, J . M.;
Yoshida, M.; Pasini, D.; Vladimirov, N.; Fre´chet, J . M. J .;
Medeiros, D. R.; Patterson, K.; Yamada, S.; Willson, C. G.;
Byers, J . D. J . Polym. Sci., Part A 1999, 37, 1225. (d) Havard,
J . M.; Vladimirov, N.; Frechet, J . M. J .; Yamada, S.; Willson,
C. G.; Byers, J . D. Macromolecules 1999, 32, 86.
(6) Darling, G. D.; Vekselman, A. M.; Yamada, S. Proc. SPIE
1998, 3333, 438.
(7) Yamada, S.; Medeiros, D. R.; Patterson, K.; J en, W. K.; Rager,
T.; Lin, Q.; Lenci, C.; Byers, J . D.; Havard, J . M.; Pasini, D.;
Fre´chet, J . M. J .; Willson, C. G. Proc. SPIE 1998, 3333, 245.
(8) Yamada, S.; Owens, J .; Rager, T.; Nielsen, M.; Byers, J . D.;
Willson, C. G. Proc. SPIE 2000, 3999, 365.
(9) March, J . In Advanced Organic Chemistry, Reactions, Mech-
anisms, and Structure, 4th ed.; J ohn Wiley and Sons: New
York, 1992; p 627.
(10) Westheimer, F. H.; J ones, W. A. J . Am. Chem. Soc. 1941, 63,
3283.
(11) Cram, D. J .; Haberfield, P. J . Am. Chem. Soc. 1961, 83, 2354.
(12) Takagi, W.; Ueyama, K.; Minamida, I.; Kim, Y. H.; Ikeda,
Y.; Oae, S. Bull. Chem. Soc. J pn. 1966, 39, 917.
(13) (a) Willson, C. G.; Cameron, J . F.; MacDonald, S. A.; Niesert,
C. P.; Fre´chet, J . M. J .; Leung, M. K.; Ackman, A. Proc. SPIE
1993, 1926, 354. (b) Fre´chet, J . M. J .; Leung, M.; Urankar,
E. J . Chem. Mater. 1997, 9, 2887.
6ib. Further adjustments of the copolymer ratio, the
formulation of the photoresist, and the process param-
eters, should provide even better imaging quality.
Con clu sion
(14) Hall, G. A. J . Am. Chem. Soc. 1949, 71, 2691.
(15) Polyacrylates (20 wt %) and TPS-SbF6 (PAG, 1 wt % to
polymer) were dissolved in THF. The solutions were spin-
coated on silicon wafers and baked (PAB) at 100 °C for 60 s.
The baked films were exposed to 248 nm light (20 mJ /cm2)
and scraped off the wafers for thermal gravimetric analysis.
(16) Trefonas, P.; Szmanda, C.; Barclay, G.; Blacksmith, R.;
Kavanagh, R.; Monaghan, M.; Coley, S.; Mao, Z.; Taylor, G.
Proc. Soc. Plast. Eng., Oct. 6-8, McAfee, NJ 1997, 11, 44.
(17) Dale, W. J .; Starr, L.; Strobel, C. W. J . Org. Chem. 1961, 26,
2225.
(18) (a) Hutchins, R. O.; Rotstein, D.; Natale, N.; Fanelli, J . J .
Org. Chem. 1976, 41, 3228. (b) Wright, A. D.; Haslego, M.
L.; Smith, F. X. Tetrahedron Lett. 1979, 25, 2325.
(19) The gel formation was observed only with the monosubsti-
tuted Meldrum’s acid, 4. Disubstituted Meldrum’s acid (not
reported here) and the opened forms (5tb, 5ib) were polym-
erized without a gel formation. This could be due to the highly
acidic R-hydrogen of Meldrum’s acid (pKa ) 4.8). We believe
that the R-hydrogen could be easily abstracted during the
radical polymerization, forming a stable radical, and this
radical initiated the growth of polymer chain, causing the
gel formation.
An aqueous processable positive tone resist material
was designed using a combination of a decarboxylation
reaction and an acid-catalyzed reaction of esters. This
aqueous resist system possesses both high etching
resistance and high resolution. The thermal stability of
the ester group to decarboxylation conditions is critical
for the system to function as a positive tone resist. In
this regard, isobornyl esters showed stability that is
superior to tert-butyl esters. The kinetics of the decar-
boxylation reaction were studied with time-resolved FT-
IR, and it was found that the rate of decarboxylation is
highly dependent upon substituents on the malonate
moiety. Steric hindrance at the R-position played an
important role to prevent anhydride formation during
decarboxylation while increasing the hydrophobicity of
the polymers. On the basis of these findings, we opti-
mized a composition of tetrapolymer, and high-resolu-
tion imaging was demonstrated with DUV exposure.
(20) J unek, H.; Ziegler, E.; Herzog, U.; Kroboth, H. Synthesis
1976, 332.
Ack n ow led gm en t. We thank International SE-
MATECH and Shipley Co. for kindly providing us the
opportunity to perform the imaging experiments at their
facilities as well as the helpful assistance. Financial
funding from Semiconductor Research Corp., Interna-
tional SEMATECH, Intel Corp., and the Welch Founda-
tion is gratefully acknowledged. Special thanks to Prof.
J . M. J . Fre´chet for valuable technical discussions.
(21) (a) Ihara, M.; Takahashi, M.; Niitsuma, H.; Taniguchim, N.;
Yasui, K.; Fukumoto, K. J . Org. Chem. 1989, 54, 5413. (b)
Ihara, M.; Takahashi, M.; Niitsuma, H.; Taniguchim, N.;
Yasui, K.; Fukumoto, K. J . Chem. Soc. PT1 1991, 525.
(22) Yamada, S.; Rager, T.; Owens, J .; Byers, J . D.; Nielsen, M.;
Willson, C. G. ACS Polym. Prepr. 2000, 87.
(23) Deprotection of the ester was assumed negligible, and this
was validated by TGA, IR spectroscopy, and 1H NMR study
(ref 22), all of which indicated no thermal deprotection.
(24) Wallraff, G.; Hutchinson, J .; Hinsberg, W.; Houle, F.; Seidel,
P.; J ohnson, R.; Oldham, W. J . Vac. Sci. Technol. B 1994,
12, 3857.
(25) The normalized intensity after completion of decarboxylation
varies because the initial intensity contains some inherent
errors. At the very early stage of baking, the casting solvent
evaporates, the wafers take about 10 s to reach thermal
equilibrium, and signal-to-noise ratio of spectra is much
higher than that for longer baking time. These factors all
affect the final intensity, and thus the final normalized
intensity also changes.
Su p p or tin g In for m a tion Ava ila ble: Details of monomer
and polymer syntheses and characterization. This material is
available free of charge via the Internet at http://pubs.acs.org.
Refer en ces a n d Notes
(1) (a) Environmental issues caused by the “greenhouse” effect:
Bradford, D. F. Nature (London) 2001, 410, 649. (b) Solvents
and birth defects: Seppa, N. Sci. News Onl. 1999, 155. (c)
Environmental issues in electronics manufacturing: Ellis, B.
Circuit World 2000, 26 (2), 17.
(2) Yamada, S. Dissertation, The University of Texas at Austin,
2000.
(26) Arrhenius plots of polymer 6ib, 8ib, and 14ib are shown in
Figure 18 in the Supporting Information.
(27) Willson, C. G. In Thompson, L. F., Willson, C. G., Bowden,
M. J ., Eds.; Introduction to Microlithography, 2nd ed.;
American Chemical Society: Washington, DC, 1994; p 139.
(3) Ichimura, K. J . Polym. Sci., Part A: Polym. Chem. 1982, 20,
1411.
(4) Taylor, L. D.; Kolesinski, H. S.; Edwards; B.; Haubs, M.;
Ringsdorf, H. J . Poly. Sci., Polym. Lett. 1988, 26, 177.
MA034461R