SLAC-PUB-9990

June 2003

Supercritical Carbon Dioxide Extraction of Porogens for the Preparation of

Ultralow Dielectric Constant Films
T. Rajagopalan, J. Sun*, B. Lahlouh, J. A. Lubguban, D. H. Huang*, N. Biswas, S. L.
Simon* and S. Gangopadhyay
Department of Physics, *Department of Chemical Engineering,
Texas Tech University, Lubbock TX 79409
A.Mallikarjunan
Center for Integrated Electronics, Rensselaer Polytechnic Institute, Troy, NY 12180
H. -C. Kim, W. Volksen, M. F. Toney(*), E. Huang, P. M. Rice, E. Delenia and R. D.
Miller
IBM Almaden Research Center, San Jose, CA
* present address: Stanford Synchrotron Radiation Laboratory, Stanford Linear
Accelerator Center, 2575 Sand Hill Rd, M/S 69, Menlo Park, CA 94025

Abstract

Supercritical carbon dioxide (SCCO2) extraction of a CO2-soluble poly(propylene

glycol) (PPG) porogen from poly(methylsilsesquioxane) (PMSSQ) cured to temperatures
adequate to initiate matrix condensation, but still below the decomposition temperature of
the porogen is demonstrated to produce nanoporous, ultralow dielectric constant thin
films. Both closed and open cell porous structures were prepared simply by varying the
porogen load in the organic/inorganic hybrid films. The porogen loads investigated in the
present work ranged from 25 - 55 wt.%. Structural characterization of the samples
conducted using transmission electron microscope (TEM), small angle X-ray scattering
(SAXS) and Fourier transform infrared spectroscopy (FTIR) confirms the successful
extraction of the porogen from the PMSSQ matrix at relatively low temperatures
(≤200oC). The standard thermal decomposition process is performed at much higher
temperatures (typically in the range of 400oC-450oC). The values of dielectric constants
and refractive indices measured are in good agreement with the structural properties of
these samples.

Work supported in part by the Department of Energy contract DE-AC03-76SF00515.

 With the ever-growing demand for higher speed in computers, the feature dimensions
of integrated circuits (IC) continue to shrink. As a consequence, narrower and thinner
metal lines are being used in ICs and the spacing between the metal lines is also being
reduced. The former increases the wiring resistance (R) whereas the latter increases the
capacitance (C), resulting in an increase in the interconnect (RC) delay. In order to
reduce R, copper is being used instead of aluminum for back-end-of-the-line (BEOL)
wiring. In an effort to reduce the capacitance, the main focus of current research has been
the development of suitable insulating materials with lower dielectric constants (k) than
SiO2. The International Technology Roadmap published by the semiconductor industry
association (2001 Edition) indicates that there will be need for materials with extremely
low k = 1.5-1.9 by year 2003.1 The only viable approach to this goal is the incorporation
of nanoporosity in the insulating matrix.2
One promising class of low-k candidates is spin-on glasses, such as the organosilicate
poly(methylsilsesquioxane) (PMSSQ), having an empirical formula (CH3- SiO1.5)n. This
material has an inherently low value of k, low moisture uptake, and excellent thermal
stability up to 500oC.3 The k value can be further lowered by introducing nanoporosity
through the incorporation and subsequent thermal degradation of pore generating organic
materials, termed porogens, into the PMSSQ matrix.4-6 However, there are some inherent
disadvantages with the thermal degradation route to nanoporosity. The process window
can be narrow since the porogen decomposition must occur below the glass transition
(Tg) of the thermally stable phase (when plasticized with the porogen) so that the
nanopores formed do not collapse. Since many organic polymers have relatively low Tgs
compared to their degradation temperatures, this processing constraint can lead to
incomplete porogen decomposition often resulting in char formation. Therefore, it is of
interest to develop a lower temperature process for the introduction of porosity. Such a
process could also be advantageous in the generation of patterned porous films since
there is a tendency toward degraded resolution (resolution of what) in high temperature
processes. A possible alternative to the purely thermal decomposition method is the use
of supercritical carbon dioxide (SCCO2) treatment to extract CO2-soluble porogens from
the PMSSQ matrix at lower process temperatures. Above its critical temperature (31oC)
and critical pressure (1050 psi), CO2 is supercritical. In this state, it has a very high

2
diffusivity due its gas-like viscosity coupled with a high solvent power due to its liquid-
like density.7 Moreover it is a well-established fact that CO2 in its supercritical state can
be a good solvent for many low molecular weight molecules.7,8 Furthermore, CO2 is
inexpensive, nonflammable and environmentally benign. These properties of CO2 in its
supercritical state have motivated us to examine the feasibility of extracting porogens
using SCCO2 treatment to form nanopores.
Another key area of interest is the porous morphology in the low-k film, i.e., whether
it is open or closed cell. The creation of closed pores in the matrix is desirable because
pore interconnectivity can result in low mechanical strength and breakdown voltage.6
Highly interconnected pores may also allow contaminants to diffuse between structures,
possibily resulting in electrical shorts.4 Thus, a key goal is the study of SCCO2 processes
to produce nanoporosity in thin films of the order of 1µm thick. This film thickness is
compatible with integration requirements. Generation of nanoporosity in such thin films
has not been achieved by SCCO2 saturation techniques.9,10 In this investigation, two
porogen compositions in PMSSQ were studied. The first composition (75/25,
PMSSQ/porogen), is expected (WHY) to form nanohybrids with a dispersed porogen
morphology while the second (45/55) has been demonstrated to (need a citation if
demonstrated) produce an interconnected nanohybrid. For purely thermal processes, the
final porous morphology is dictated by that of the nanohybrid. Here we report, for the
first time, the successful extraction of phase separated porogens from each sample using
SCCO2. For simplicity, the two nanohybrid samples will be hereafter referred to as
OS7525 and OS4555.
A solution containing appropriate quantities of the linear poly(propylene glycol)
(PPG) porogen (Mn = 20,000 g/mol and PDI = 1.05) and low molecular weight PMSSQ
dissolved in propylene glycol methyl ether acetate (PM acetate) were loaded into a
disposable syringe and passed through a 0.2µm PTFE filter directly onto silicon
substrates. The spin-coated samples were heated briefly to 50oC on a hot plate under
nitrogen atmosphere to remove residual solvent before curing at 200oC to produce the
organic/inorganic polymer hybrid. This curing temperature is still well below the thermal
degradation temperature of the porogen in an inert atmosphere. During matrix curing,
phase separation of the porogen occurs via a nucleation and growth process. The thermal

3
processes (baking/curing) were carried out under nitrogen as porogen decomposition
begins as low as 150oC in the presence of trace amounts of oxygen. The experimental
steps involved in the SCCO2 extraction process are described here. The samples were
placed inside a small cylindrical pressure vessel made of stainless steel with inner
diameter of 2.5 cm, outer diameter of 6.3 cm, and depth of 12.7 cm. The ambient air was
then pumped out using a roughing pump and the vessel backfilled with CO2 at
approximately 300 psi. The process temperature was then ramped slowly to achieve the
desired temperature. The chamber was subsequently pressurized isothermally with CO2
using a gas booster pump (Haskel) driven by air at 70 psi and the pressure of the chamber
was maintained over the duration of the experiment. High purity (99.99%) CO2 was
used in these studies. At the completion of the experiment, the vessel was depressurized
to 500 psi and the temperature was ramped down to room temperature. Depressurization
was then completed at room temperature.
FTIR spectroscopy was used to quantify the degree of porogen extraction. FTIR
spectra were recorded before and after extraction using a Perkin-Elmer Model 1600
spectrometer with a resolution of 8 cm-1 in the range of 400 to 4000 cm-1. The
microstructure of the samples after extraction was examined using transmission electron
microscopy (TEM, TopCon-002B) and small angle x-ray scattering (SAXS). Samples for
TEM were prepared using a FEI 830XL dual beam focused ion beam (FIB) machine. A
thin carbon layer less than 20 nm and a Pt layer of ~50 nm thick were deposited on the
surface to minimize contamination. The pore size and size distribution were determined
by SAXS. SAXS experiments were performed at the IMM-CAT at the Advanced Photon
Source at Argonne National Laboratory. A detailed description of the experimental
11
geometries has been reported elsewhere. The thickness and refractive index of the
samples were measured using ellipsometry after every SCCO2 treatment. Porosity was
estimated applying the Maxwell-Garnett effective medium approximation method.12 For
dielectric measurements, Ti electrodes with an area of 0.196 × 10-6m2 were deposited by
e-beam evaporation on the organic/inorganic hybrid films deposited on heavily doped
silicon substrates.
In order to quantify the amount of porogen remaining after extraction in the films
using FTIR, it is necessary to know which characteristic vibrational bands are assigned to

4
the PPG (CH(CH3)CH2O)n and PMSSQ (CH3-SiO1..5)n components. FTIR absorbance
measurements for the PMSSQ and PPG, respectively, show many overlapping vibrational
bands; prominent among these are the OH stretch at 3150-3700 cm-1 and CHn stretch at
2800-3000 cm-1. However, in case of CHn bands, the CH3 asymmetric stretch at 2960
cm-1 is the only band observed in PMSSQ sample, whereas the CH2 asymmetric stretch at
2915 cm-1 is observed in PPG in addition to CH3 bands corresponding to symmetric and
asymmetric stretch. Thus, in a hybrid film, the CH2 contribution arises from PPG. A
ratio of the area of this peak before and after extraction gives a reasonable quantitative
estimate of porogen left in the sample after extraction.
The strongest vibrational bands for PMSSQ were observed from 1000-1200 cm-1.
From the literature13-15, the peaks at 1120 cm-1 and 1030 cm-1 in the as-spun samples can
be assigned to the cage and/or chains with non-bridging oxygen atoms and network
structures of MSSQ, respectively .16,17 In the FTIR spectra of PPG films, we observe the
strongest peak in the region of 1000-1200 cm-1. This band is assigned as the C-O
stretching vibration as observed for alcohols and ethers.18 As a result, the region from
1000-1200 cm-1 in the FTIR spectrum of the organic/inorganic hybrid film contains
overlapping bands for both PPG and PMSSQ and deconvolution of the bands in this
region is not easy. In the present context of the SCCO2 extraction of PPG from PMSSQ
matrix, we monitor both the bands at 2800-3000 cm-1 and 1000-1200 cm-1. In addition,
we also monitor changes in OH band intensity since it is important from the perspective
of lowering the k value.
The procedure used for the analysis of FTIR data has been described elsewhere.19
Fig.1 (a) and Fig.1 (b) show plots of the absorption coefficient versus wave number for
the two spectral regions of interest; 400-1800cm-1 and 2700-4000cm-1 of as-spun OS4555
and for OS4555 films treated with SCCO2 at 7000psi for 14 hours at 160oC. The
intensities of the CH and OH bands decreased significantly after SCCO2 treatment; a
feature consistent with extraction of the porogen. However, we still observe a small peak
at 2915 cm-1 attributed to the asymmetric CH2 stretching of the porogen signifying a
small residue remaining in the sample after extraction. Careful analysis of the integrated
intensities of the CH2 bands before and after extraction yields a value of 8% for the
amount of porogen remaining. The value of n measured for the extracted film is sharply

5
reduced (1.146) from 1.442 measured for the as-spun film. The reduction in the n value is
attributed to the extraction of significant amounts of porogen and generation of porosity.
In case of sample OS7525 treated with SCCO2 at 7000 psi for 14 hours at 160oC, the
intensity of the band at 1000-1200 cm-1 and that of the CHn band does not show
appreciable decrease after SCCO2 treatment. There was also only a small reduction in
the intensity of the OH band. The results indicate that SCCO2 treatment is not able to
effectively extract the porogen under these process conditions presumably due to the lack
of interconnectivity of the PPG phase. A possible solution is to increase the temperature
of the extraction process. Although increasing the temperature decreases the solubility of
the porogen both in the SCCO2 and in the matrix, it does increase the diffusivity of the
porogen through the matrix. In addition, at higher temperatures, some limited bond
breaking may occur resulting in a decrease in molecular weight, which would make the
polymer easier to extract. When the OS75/25 as-spun samples were treated with SCCO2
at 200oC we observed an appreciable decrease in the intensity of the bands at both 1000-
1200 cm-1 and 2800-3000 cm-1 as shown in Fig.2 (b) after extraction in comparison to
Fig.2 (a) for as-spun OS7525 films. The amount of porogen left in this sample after
extraction is estimated to be ~15%. Thus for this example, the SCCO2 process has not
extracted the porogens completely even at the higher temperatures for non-
interconnecting morphologies. We are currently trying to further optimize extraction
process by tuning certain process parameters, e.g. the use of a suitable co-solvent along
with SCCO2. We are also in the process of building a flow-through system to facilitate
complete removal of porogens.
The refractive index of the extracted OS7525 sample decreased substantially from
1.441 before extraction to 1.313. It is noteworthy that the treatment temperature is still,
however, much lower than required for complete thermal decomposition of the porogen.
As a control, a sample thermally annealed 14 hours at 200oC in nitrogen ambient showed
no change in the IR absorbance spectra for either of the as-spun OS4555 and OS7525
samples. This confirms that the removal of the porogen is exclusively due to SCCO2
extraction and was not simply due to thermal decomposition and volatilization of PPG. .
The data confirm that extraction of porogens even from non-interconnecting hybrid
morphologies can be accomplished with SCCO2.

6
 As mentioned earlier, in the conventional thermal decomposition process, the porogen
is removed by heating to 400oC-450oC in a nitrogen atmosphere. In addition to the
creation of nanopores, this thermal treatment also further cures and crosslinks the matrix,
increasing the mechanical strength of the film. For this reason, after extraction of the
porogen from our samples, we annealed the films under nitrogen at 430oC for two hours.
The FTIR results for these samples shown in Fig.1(c) and Fig.2 (c) clearly indicate that
there is an increase in the intensity of the peak at 1040 cm-1 corresponding to network
formation with simultaneous decrease in the intensity of the peak at 1140 cm-1
corresponding to cage structure and/or nonbridging oxygen. In addition, the 430oC
annealing treatment also removed the residual OH functionality completely, a critical
issue for lowering the value of k. For comparison, Fig.1 (d) and Fig.2 (d) show FTIR
absorption spectra for OS4555 and OS7525 films after only thermal treatment at 430oC
for two hours (no SCCO2). It is apparent that the thickness normalized intensity of the
CH3 signals at 2970 cm-1 and 1270 cm-1 is higher for the film first treated with SCCO2
and then annealed at 430oC than for the film heated directly to 430oC. It is well known
that the concentration of bonds is proportional to the area under the peak. The OS4555
sample first treated with SCCO2 and then annealed showed ~7% higher concentration of
CH3 than the directly annealed sample. In case of the OS7525 sample, the concentration
was ~ 18% higher. The relative porosity of the films can be calculated from the n values
of the treated films.20 Using the Lorentz-Lorenz equation, and relating that to the total
volume of the pores, the porosity of the films can be evaluated using the following
equation:
é (n 2 − 1) ù é ( nd2 − 1) ù
V = 1 − ê 2p ú ê ( n 2 + 2) ú (1)
ë ( n p + 2) û ë d û
where V is the total pore volume, np is the refractive index for the film after porogen
extraction and nd is the refractive index of the dense matrix of the film (MSSQ matrix in
this case).
The n-values for the samples first extracted with SCCO2 followed by thermal
annealing are 1.137 and 1.274 for the OS4555 and OS7525 samples, respectively.
Porosity estimates for these samples are 62.5% and 27.5% respectively (XRR should be
done to confirm these). The values of n and porosity for the samples which received only

7
thermal annealing were 1.143 and 1.257 and 60.9% and 31.7% respectively. From a
practical point of view, annealing after extraction provides an ancillary benefit in that
most of the porogen has already been removed by extraction minimizing mass transport
issues and for interconnecting morphologies an efficacious path for expelling the
thermalized fragments is provided.
The dielectric constant results for all of the samples studied are collected in Table I.
The results include samples which (i) were only annealed at 430oC in N2 atmosphere (ii)
those which were treated with SCCO2 only, and (iii) samples which were both treated
with SCCO2 and thermally annealed at 430oC in N2 atmosphere. In the case of the
OS4555 films, the measured k value was essentially the same for SCCO2 extraction and
annealing as for thermal annealing at 430oC alone (1.47). After only SCCO2 extraction at
160oC, the value of k was 1.87. For the OS7525 sample, SCCO2 treatment followed by
annealing resulted in a value of k of 2.29, whereas thermal annealing alone at
430oCproduced a k= 2.47. The film sample which received only the SCCO2 treatment
showed a measured k= 2.81. In the case of the samples with a dispersed morphology for
the nanohybrids, the lowest values of k are obtained for films first treated with SCCO2
and then annealed at 430oC. The decrease in the k-value after annealing to 430oC for
both compositions is attributed to the completed removal of the porogens and also to
further curing of the MSSQ matrix
The cross-sectional TEM images of the nanoporous thin films prepared with only
SCCO2 treatments are shown in Fig. 3. The images suggest that pore structure is strongly
dependent on the porogen loading. Uniformly distributed small pores throughout the
entire thickness of the film are observed for OS7525 while increase in pore size with
localized pore distribution is observed for OS4555. Although TEM provides useful
information regarding the pore structure, the extraction of quantitative values for pore
size and size distribution from these images is difficult due to several factors including
overlapped pore structure in the specimen and TEM resolution limitation.
SAXS was used to quantify the porous structure. Fig. 4a shows the background
subtracted, circularly averaged, and transmitted beam intensity normalized SAXS profiles
(was the MSSQ subtracted?). The open and closed symbols correspond to the porous
films prepared by SCCO2 and thermal method, respectively. The scattering intensities

8
increase and profiles shift to a lower scattering angle with increasing porogen loading. It
is noted that no significant differences in SAXS profile were observed between two pore
generation methods for both OS7525 and OS4555 (I disagree! The two lower curves are
different! Also see caption). To determine the pore distribution, we have quantitatively
modeled the SAXS data based on Pedersen’s approach.21 The scattered intensity, I(q), is
written as,
∞
I = c ò n( r ) f ( qr ) S ( qr ) dr (2)
0

where c is a constant, n(r) is the pore size distribution, f(qr) is the spherical form factor,
and S(qr) is the structure factor for the monodisperse hard sphere model.11,21 The best fit
of the data was obtained using log-normal pore size distributions as shown in Fig. 4b.
The size distributions in Fig. 4b show that the average pore size increases and the size
distribution broadens with increasing porosity. This is in good qualitative agreement
with the images obtained by TEM.
In summary, supercritical CO2 extraction of a PPG porogen has been demonstrated to
lower k by creating nanoporosity in both connecting and non-interconnecting hybrid
compositions of thin MSSQ films with thicknesses ranging from 0.6 to 1µm. It is quite
remarkable that SCCO2 treatment is able to extract the porogen from the MSSQ matrix
even in a sample with aninitial non-interconnecting hybrid morphology. These
o
extractions were performed using SCCO2 at much lower temperatures (160 C for the
interconnecting and 200oC for dispersed nanohybrid compositions) compared to the
conventional thermal sacrificial porogen removal process. For the dispersed morphology,
the dielectric constant is lower for films undergoing SCCO2 extraction followed by
annealing at 430oC compared to those directly annealed at 430oC without SCCO2
treatment. For interconnected hybrid compositions, both processes (extraction plus
annealing and only annealing) resulted in the same value of k.

Acknowledgements
Profs. Gangopadhyay and Simon gratefully acknowledge the financial support of this
work from the Semiconductor Research Corporation and from the National Science
Foundation, CMS-0210230. The authors would also like to acknowledge initial support

9
provided by the State of Texas Advanced Technology Program (ATP grant # 003644-
0229-1999).

10
References
1
International Technology Roadmap for Semiconductors, 2001 Ed.,(Semiconductor

Industry Association, 2001).

2
Cattien V.Nguyen, Craig J. Hawker, James L.Hedrick, Richard L.Jaffe, Robert

D.Miller, Julius F.Remenar, Hoe-Woo Rhee, Michael F.Toney, Mikael Trollsas, Willi

Volksen and Do Y.Yoon, Electrochemical Soc.Proc. 99-7, 38 (1999).

3
Cattien V.Nguyen, K.R.Carter, Craig J. Hawker, James L.Hedrick, Richard L.Jaffe,

Robert D.Miller, Julius F.Remenar, Hoe-Woo Rhee, P.M.Rice, Michael F.Toney,

Mikael Trollsas, and Do Y.Yoon, Chem.Mater. 11, 3080 (1999).

4
Julius F.Remenar, Craig J. Hawker, James L.Hedrick, Sung Mog Kim, Robert

D.Miller, Cattien V.Nguyen, Mikael Trollsas, and Do Y.Yoon,

Mat.Res.Soc.Symp.Proc. 511, 69 (1998).

5
Cattien V.Nguyen, Craig J. Hawker, Robert D.Miller, E.Huang, James L.Hedrick,

R.Gauderon, J.G.Hilborn, Macromolecules, 33, 4281 (2000).

6
Jorg Bolze, Moonhor Ree, Hwa Shik Youn, Sang-Hyon Chu and Kookheon Char,

Langmuir, 17, 6683 (2001).

7
M. A. McHugh and V. J. Krukonis, "Supercritical Fluid Extraction," Butterworths,

Stoneham, Mass. (1986).

8
Jacqueline A.Behles and Joseph M.DeSimone, Pure Appl. Chem., 73(8), 1281 (2001).
9
S. K. Geol, E. J. Beckman, Polymer Engineering and Science, 34, 1148(1994).
10
S. K. Geol, E. J. Beckman, AlChE Journal, 41, 357(1995).
11
E. Huang, M. F. Toney, W. Volksen, D. Mecerreyes, P. Brock, H.-C. Kim, C. J.

Hawker, J. L. Hedrick, V. Y. Lee, T. Magbitang, R. D. Miller, L. B. Lurio, Appl.

Phys. Lett. 81(12), 2232, (2002).

11
12
K. P. Mogilnikov, V. G. Polovinkin, F. N. Dultsev, and M. R. Baklanov, Mat. Res.

Soc. Proc. 565, 81 (1999).

13
W.C.Chen, A.C.Lin, B.T.Dai and M.S.Tsai, J. of Electrochem. Soc., 146, 3004

(1999).
14
W.C.Chen and C.T.Yen, J. Vac. Sci. Technol B, 18, 201 (2000).
15
Long-Hua Lee, Wen-Chang Chen and Wei-Chih Liu, J.Polym Sci Part A: Polym.

Chem 40 1560 (2002).

16
C.Y. Wang, Z.X. Shen, J.Z. Zheng Appl. Spectrosc., 54, 209 (2000).
17
C.Y. Wang, Z.X. Shen, J.Z. Zheng, Appl. Spectrosc., 55(10), 1347 (2001).
18
George Socrates, Infrared Characteristics Group Frequencies, (J. Wiley, New York

1994) Chap.
19
J. Lubguban, T. Rajagopalan, N. Mehta, B. Lahlouh, S.L. Simon and S.

Gangopadhyay, J. Appl. Phys. 92(2), 1033 (2002).

20
M. R. Baklanov, K. P. Mogilnikov, Mat. Res. Soc. Symp.,612, D4.2.1 (2000).
21
J. S. Pedersen, J. Appl. Crystallogr. 27, 595 (1994).

12
Figure Captions

Fig.1 FTIR absorption spectra of OS4555 films (a) as spun, (b) SCCO2 treated at

7000psi, 160oC, 14hours, (c) first SCCO2 treated at 7000psi, 160oC, 14 hours and

then thermally annealed at 430oC for two hours and (d) only thermally annealed at

430oC for two hours.

Fig.2 FTIR absorption spectra of OS4555 films (a) as spun, (b) SCCO2 treated at

7000psi, 200oC, 14hours, (c) first SCCO2 treated at 7000psi, 200oC, 14 hours and

then thermally annealed at 430oC for two hours and (d) only thermally annealed at

430oC for two hours.

Fig.3 Cross-sectional TEM images of nanoporous PMSSQ films, (a) OS7525, (b)

OS4555. Scale bar=150nm.

Fig.4 Small angle x-ray scattering (SAXS) profiles of nanoporous

poly(methylsilsesquioxane) films. (a) closed and open symbols present thermal

and SCCO2 processes, respectively. (b) pore size distribution obtained from the

best fits (solid line in (a)). This figure is hard to follow. Edit out the superions

and plot with lines.

13
 List of Tables

Table I Measured values of refractive index, estimated values of porosity and

dielectric constant for OS4555 and OS7525 as-spun, SCCO2 treated and

SCCO2 treated + annealed and only thermally annealed films.

14
 Table I Measured values of refractive index, estimated values of porosity and
dielectric constant for as-spun, SCCO2 treated, SCCO2 treated +
annealed and only annealed OS4555 and OS7525 films

Sample Treatment Thickness Refractive Percent k value

index Porosity ± 0.1
Name (µm)
MSSQ As-spun pure organosilicate MSSQ 0.4799 1.4099 3.1
film cured at 200oC ±0.0042 ±0.0004
OS4555 As spun sample cured at 200oC 0.9205 1.4423 ….. ….
As-spun ±0.013 ±0.0006
OS4555 SCCO2 treated at 7000psi160oC for 0.9119 1.146 ….. 1.87
Press. 14 h ±0.0011 ±0.003
OS4555 SCCO2 treated first + Annealed at 0.6051 1.137 62.5 1.47
Press.& 430oC in 600mTorr N2 for 4h ±0.0009 ±0.003 0.005
ann
OS4555 Only annealed at 430oC in 0.6310 1.143 60.9 1.47
Ann 600mTorr N2 for 4h ±0.0011 ±0.003 ±0.005
OS7525 As-spun sample cured at 200oC 0.5779 1.441 ….. ….
As-spun ±0.0037 ±0.0008
OS7525 SCCO2 treated at 7000psi 0.5690 1.313 ….. 2.81
Press. 200oC for 14 h ±0.0014 ±0.003
OS7525 SCCO2 Treated first + Annealed at 0.4435 1.274 27.5 2.29
Press. & 430oC in 600mTorr N2 for 4h ±0.0009 ±0.003 0.002
Ann
OS7525 Only annealed at 430oC in 0.4502 1.257 31.7 2.47
Ann 600mTorr N2 for 4h ±0.0011 ±0.003 ±0.002

15
 6000 1600
4500 1200
3000 800
1500 (d) 400

Absorption Coefficient (cm-1)

0 0

6000 1600
4500 1200
3000 800
1500 (c)
400
0 0

6000 1600
4500 1200
3000 800
1500 (b)
400
0 0

6000 1600
4500 1200
3000 (a) 800
1500 400
0 0
500 1000 1500 2500 3000 3500 4000

Wavenumber (cm-1)
Fig.1 FTIR absorption spectra of OS4555 films (a) as spun, (b) SCCO2 treated at

7000psi, 160oC, 14hours, (c) first SCCO2 treated at 7000psi, 160oC, 14 hours and

then thermally annealed at 430oC for two hours and (d) only thermally annealed at

430oC for two hours.

16
 10000 1600
7500 1200
5000 800
(d)

Absorption Coefficient (cm-1)

2500 400
0 0
10000 1600
7500 1200
5000 (c) 800
2500 400
0 0

6000 1600
1200
4000 (b) 800
2000 400
0 0

6000 1600
1200
4000 (a) 800
2000 400
0 0
500 1000 1500 2500 3000 3500 4000
Wavenumber (cm-1)
Fig.2 FTIR spectra of OS7525 films (a) as spun, (b) SCCO2 treated at 7000psi,

200oC, 14 hours, (c) first SCCO2 treated at 7000psi, 200oC, 14 hours and

then thermally annealed at 430oC for two hours and (d) Only thermally

annealed at 430oC for two hours.

17
Fig.3 Cross-sectional TEM images of nanoporous PMSSQ films, (a) OS7525,

(b) OS4555. Scale bar=25nm.

18
Fig.4 Small angle x-ray scattering (SAXS) profiles of nanoporous

poly(methylsilsesquioxane) films. (a) closed and open symbols present

thermal and SCCO2 processes, respectively. (b) pore size distribution

obtained from the best fits (solid line in (a)).

19