Porosity Characterization of porous SiLK Dielectric Films
C.E. Mohler, E.G. Landes, G.F. Meyers, BJ. Kern, K.B. Ouellette, S. Magonov*
The Dow Chemical Company, Midland, MI 48674
*Veeco Instruments, 112 Robin Hill Road, Santa Barbara, CA 93117
Abstract. The continual drive for faster interconnects in integrated circuits requires the development of new interlayer
dielectric materials with k values less than 2.2. Porous SiLK semiconductor dielectric resin was developed to achieve
this low dielectric constant by introducing nanometer-sized pores into the SiLK matrix. The development of metrology
to characterize the pores in porous SiLK dielectric films is critical for successful adoption of the material in the industry,
both to ensure the film attains the desired dielectric properties and to monitor pore characteristics that may impact the
integration process. Due to the complex nature of the porous structure, multiple on-wafer methods were investigated to
quantify the porosity in porous SiLK dielectric films. The use of ellipsometry, small angle X-ray scattering and atomic
force microscopy to measure void fraction, pore size, size distribution, pore morphology and their uniformity across a
porous SiLK film are described. Advances in these techniques and commercialization of fab-quality X-ray scattering
tools indicates significant progress has been made in the availability of porosity metrology for porous SiLK resin films.
technologies are either proven as reliable fab-quality
metrology, or have significant potential for use in this
capacity.
INTRODUCTION
To achieve the low dielectric constants expected to
be required for interlayer dielectric films in 65 nm
node technology and beyond, materials suppliers have
turned to incorporating nanoporosity into the dielectric
materials. One of the leading nanoporous dielectric
materials under evaluation is porous SiLK resin [1-4],
an organic spin-on polymer made by incorporating
nanometer-size porogens in SiLK matrix precursor.
After spin-coating the porogen is removed from the
film during cure, leaving behind air-filled closed-cell
voids which reduce the dielectric constant of the film
from 2.65 to 2.2. Characterization of the porosity in
porous low-k dielectrics such as porous SiLK resin
presents significant challenges.
On-wafer, nondestructive and rapid metrology is critical for
successful integration of a porous dielectric into high
volume 1C manufacturing. The key metrics of interest
for porosity characterization include void fraction
(pore density) and uniformity, pore size, pore size
distribution and pore connectivity or morphology.
This paper describes the use of spectroscopic
ellipsometry, small angle X-ray scattering and atomic
force microscopy to characterize these critical metrics
of porosity in porous SiLK dielectric films. All three
M
EXPERIMENTAL
Porous SiLK films made from formulations with
porogen loadings from 0 to 30% were coated on 200
mm silicon substrates by spin-coating the formulations
on a TEL ACT-8 SOD spin track. Spin speed was
adjusted to maintain similar film thicknesses for all
formulations, which was nominally 2000 A.
Film thickness and void fraction measurements
were made using a KLA-Tencor ASET F5x
spectroscopic ellipsometer. The reported values are an
average of measurements done using a standard
Prometrix 49 point layout on the wafer with an 8 mm
edge exclusion. The tool recipe was developed using a
BEMA
(Brueggeman
Effective
Medium
Approximation) including void and dispersions for
cured and uncured SiLK matrix and porous SiLK
films. Refractive index, film thickness and void
fraction can then be measured at multiple sites on the
film.
Trademark of The Dow Chemical Company
CP683, Characterization and Metrology for VLSI Technology: 2003 International Conference,
edited by D. G. Seiler, A. C. Diebold, T. J. Shaffner, R. McDonald, S. Zollner, R. P. Khosla, and E. M. Secula
© 2003 American Institute of Physics 0-7354-0152-7/03/$20.00
562
porogen loading of 30% yields films with a dielectric
constant of 2.2.
Atomic force microscopy (AFM) measurements
were made on porous SiLK resin films using a Veeco
Instruments Dimension 5000 scanning probe
microscope. Data was collected over a 250 nm by 250
nm square area, at 512 x 512 pixel resolution. Low
frequency variations in the intensity of the height
image were removed by subtracting a background
image obtained by applying a Gaussian filter to the
original height image. This enabled more reliable
thresholding and conversion to binary images for data
analysis.
Although there is some difference at the highest
porogen loadings studied, the void fraction by
ellipsometry is in reasonable agreement with that
obtained by surface acoustic wave (SAWS) density
measurements [6], and has the advantages of being a
rapid on-wafer
test using production-tested
technology. Improvements which provide a robust
dispersion for porous SiLK resin over a wider range of
porogen loadings may lead to better correlation of the
void fraction at 30% porogen loading. To test the
short-term repeatability of the void fraction
measurement, ten consecutive measurements were
made in the same spot on the porous SiLK wafer with
30% porogen loading. The one-sigma level for the ten
replicates was -0.1%, showing the method has good
sensitivity to changes in void fraction.
Small angle X-ray measurements were done at the
Advanced Photon Source, Argonne National
Laboratories. A normal beam transmission mode
geometry was used, with sample exposure times of 10
minutes at energies of 15 or 18 KeV. Average pore
size and size distributions were calculated using rigid
sphere models which incorporate both form and
structure factor contributions [5].
In addition to an average void fraction for a porous
SiLK film, the uniformity of void fraction across the
film is also of interest since this is directly related to
the uniformity of the film dielectric properties. With
commercial spectroscopic ellipsometers, mapping of
the void fraction across a wafer can be easily
automated. A map of the void fraction for a porous
SiLK dielectric film measured using the KLA-Tencor
ASET F5x is shown in Figure 1. The mean void
fraction for the entire wafer was 23.6%, with a
standard deviation of 0.3% across the 200 mm film,
indicating excellent uniformity of dielectric constant.
VOID FRACTION BY ELLIPSOMETRY
Since the density of pores in porous SiLK resin
films is directly related to the dielectric constant that
the film can achieve, the pore density (or void
fraction) is a critical metric of the dielectric
performance of the film. Since the optical constants
(refractive index, n, and extinction coefficient, k) are
also related to the film density, ellipsometry can be
used as an on-wafer probe of the film void fraction by
following changes in the film's refractive index. To
illustrate this, the refractive index at 633 nm and void
fraction calculated for porous SiLK resin films as a
function of porogen loading is shown in Table 1. As
expected the films become less dense as the porogen
loading of the film increases, which in turn results in a
decreased refractive index.
0.241
TABLE 1. Ellipsometric Results on porous SiLK resin
Porogen
Loading
0%
10%
20%
30%
Thickness
(A)
Refractive
Index
(633 nm)
F5 Void
Fraction
SAWS
Void
Fraction
2601 ± 10
2068 ± 5
1977 ± 11
2001 ±11
1.6471.004
1.588±.006
1.570±.004
1.527±.005
12.3±1.0
13.9±0.6
20.1±0.9
—
8.9
15.2
25.0
FIGURE 1. Void fraction map for a porous SiLK resin
film, using a KLA-Tencor ASET F5x. The standard
deviation of the void fraction across the 200 mm wafer is
0.3%, indicating a high uniformity of pore density.
PORE SIZE BY SMALL ANGLE X-RAY
SCATTERING
The film thickness also decreases with increasing
porogen loading, reflecting a small but gradual
decrease in porogen efficiency at higher loading
levels. Even with some loss of porogen efficiency a
Pore size and pore size distribution are key
characteristics of porous SiLK resin films, since pore
563
into a pore size distribution, using the approach of
Pedersen [5].
size may impact the ease of integration of the
dielectric layer in interconnect structures. Adequate
barrier coverage may be challenging if large pores are
present in the trench/via sidewalls or trench bottom;
there is also concern that large pores may potentially
facilitate migration of process chemicals or metals
through the dielectric, ultimately leading to electrical
and reliability failures in the device. Although
definitive information correlating pore size with
successful integration of porous dielectrics is not yet
available, pore size and size distribution measurements
are anticipated to be one of the most critical needs for
on-wafer metrology of porous dielectric films. Small
angle X-ray scattering (SAXS) has emerged as one of
the few non-destructive techniques available that is
sensitive to pore size and pore size distribution in thin
films. Recently an on-wafer SAXS method has been
developed which can successfully measure pore size
and pore size distribution for porous SiLK resin films
coated on silicon substrates [7].
Figure 3 shows the pore size distributions
calculated from the scattering pattern for porous SiLK
films with porogen loadings from 10 to 30%. The
average pore diameters for the samples are listed in
Table 2. Over the range of porogen loadings used in
this study the average pore diameter appears to be
independent of porogen loading within the error of the
data analysis. No evidence of porogen agglomeration
at higher porogen loadings is observed. Any slight
decrease in pore size at higher porogen loadings might
be attributed to multiple scattering effects in the film,
expected to be more important at higher porogen
levels. The shape of the pore size distribution can be
adequately modeled using either a gaussian or lognormal function over the range of porogen loadings
studied, and shows only a slight increase in
distribution width with loading. There appears to be a
non-zero amount of very small pores at 30% porogen
loading, however during the fitting procedure the
probability was not limited at the small end of the pore
size range so it is possible this observation is not
related to a property of the material.
To obtain pore size information, the X-ray
scattering patterns are circularly averaged to produce
intensity (I) vs. scattering vector (q) data where q is
related to the wavelength and scattering angle, q =
4n/X sin(Q/2).
Figure 2 shows a typical SAXS
scattering pattern for a porous SiLK film. The
scattering pattern shows no significant dependence on
position on the wafer, indicating excellent uniformity
of pore size and distribution across the film. The
position of the discontinuity in the scattering intensity
at intermediate q values can be used to estimate the
0.15H
• 10%
O 20%
30%
0.10-
tram
0.05-
2
Q_
0.00 -I
I
I
I
I
I
25rnn
0
5
10
15
20
*3)mn
Diameter (nm)
* 75irro
FIGURE 3. Pore size distributions for porous SiLK resin
*
tHDQ
1
tun
films with porogen loadings from 10-30%.
-ray
0001
001
01
1
TABLE 2. Pore size by on-wafer SAXS, as a
function of porogen loading
Porogen
Avg Pore
Diameter (nm)
Loading
FIGURE 2. Small angle X-ray scattering pattern for a
porous SiLK resin film, as a function of position on a 200
mm wafer. The similarity of the scattering patterns indicates
excellent uniformity of pore size and pore size distribution
across the film.
10%
average pore size, since for spheres this feature occurs
roughly at q~7t/d. If a spherical pore geometry is
assumed, a hard sphere model can be used to
transform the entire scattering distribution observed
9.3 ± 0.5
20%
9.2 ±0.5
30%
8.6 ± 0.5
While small spot size and measurement speed are
important areas for improvement, SAXS shows
564
geometry is unaffected by the porogen loading and the
small standard deviation indicates narrow and similar
shape distributions for all three films. The average
pore size may be slightly lower than that measured by
SAXS, but agrees within the error of the measurement.
Better agreement between methods may be obtained
with improved sampling statistics and refinements of
the image analysis procedure.
enough promise for pore size and pore size distribution
metrology that fab-quality tools are currently being
commercialized.
PORE MORPHOLOGY BY ATOMIC
FORCE MICROSCOPY
Since ellipsometry and SAXS depend on
mathematical models to derive information about
porosity in porous dielectric films, they do not give
direct information about the pore morphology. Pore
connectivity and geometry are key properties of
porous dielectrics for which on-wafer metrology is
expected to be necessary. It has been shown that
closed porosity increases barrier performance [8].
Information on pore geometry should also improve the
models used for analysis of SAXS data, leading to
improved accuracy of pore size measurements.
Atomic force microscopy was investigated to
determine its ability to detect and quantify pore
morphology in porous SiLK resin films. While this
technique may be able to image only pores near the
film surface and not throughout the entire film, it is the
pore morphology near the film surface that is
important in determining adequacy of barrier coverage
and whether process chemicals can enter the film. All
AFM images were collected using low amplitude, hard
tapping conditions which improved the visualization
of pores. Figure 4 shows the height images of porous
SiLK resin films at various porogen loadings, together
with the binary image created by image analysis.
FIGURE 4. AFM images of porous SiLK films at 30, 20
and 10% porogen loadings using Veeco D5000 SPM. The
height image is on the left, the binary image is on the right.
Subtraction of a background from the raw height
image to account for variations in surface topography
allowed the image to be more readily converted into a
binary format. The pores are easily visualized in the
binary image and all three films show discrete closed
pore morphology. This result serves as visual support
for the ability to deposit a continuous barrier on porous
SiLK resin films [9]. Using the binary image several
aspects of the pore morphology can be quantified,
including area fraction of porosity, pore size and
distribution, and pore roundness and distibution.
These results are listed in Table 3.
TABLE 3.
AFM Measurements of pore
morphology in porous SiLK resin films
Porogen
Area
Pore
Avg Pore
Loading
Fraction
Diameter
Roundness
(nm)
(%)
0.62 ± 0.12
10%
5.1
6.9 ± 2.3
0.62 ± 0.13
20%
9.3
7.4 ± 2.1
0.64 ± 0.12
8.2 ± 2.5
15.9
30%
The area fraction is systematically lower than void
fraction measured by ellipsometry or SAWS, but is
highly correlated with void fraction by both methods
(R2 >0.97) and shows promise as a metric of pore
density. A quantitative indicator of pore roundness
can also be extracted from the images. The pores in
all three porous SiLK films are slightly more oblong
than a sphere (pore roundness factor <1.0). The
Figure 5 shows the height and binary AFM images
of a porous SiLK resin film with 30% porogen
loading, but processed at 135°C higher than the
optimal bake temperature. The area fraction and pore
roundness for this sample are listed in Table 4,
compared to the 30% porogen loading sample
processed under optimal conditions (Figure 4). The
area fraction of pores is similar for both samples, but a
565
comparison of Figure 4 and Figure 5 clearly shows
more pore connectivity for the sample processed with
a large temperature difference from optimal. This is
also reflected in a decreased value of the pore
roundness factor.
made at the DuPont-Northwestern-Dow Collaborative
Access Team Synchrotron Research Center (Advanced
Photon Source), which is supported by the E.I. DuPont
de Nemours & Co., The Dow Chemical Company, the
U.S. National Science Foundation (Grant DMR9304725), the State of Illinois through the Department
of Commerce and the Board of Higher Education
Grant IBHE HECA NWU 96. Use of the Advanced
Photon Source was supported by the U.S. DOE Basic
Energy Sciences, Office of Energy Research (Contract
No.W-31-102-Eng-38).
REFERENCES
1. Silvis, H. C.; Hahnfeld, J. L.; Niu, Q. J.; Radler, M. J.,
Abstracts of Papers, 224th ACS National Meeting,
Boston, MA, U.S., August 18-22, 2002 (2002), PMSE114. American Chemical Society, Washington, D. C.
FIGURE 5. Height and binary AFM images of a porous
SiLK resin film with 30% porogen loading, baked at 135°C
higher than the optimal bake temperature.
2. Waeterloos, J. J.,;Struyf, H.; Van Aelst, J.; Das, A.;
Caluwaerts, R.; Alaerts, C.; Boullart, W.; Tokei, Z. S.;
lacopi, F.; Van Hove, M.; Maex, K., Advanced
Metallization Conference 2001, Conference Proceedings,
Montreal, Canada, Oct. 8-11, 2001 (2002), 19-24.
Materials Research Society, Warrendale, Pa.
TABLE 4.
AFM measurements of pore
morphology in porous SiLK resin films using
different process conditions
Temp
Area
Avg Pore
above
Fraction
Roundness
optimal
(%)
0°C
135 °C
15.9
16.0
3. Strittmatter, R. J., Proceedings of the Materials Research
Society, April 2003, to be published.
0.62 ± 0.12
0.55 ± 0.13
4. Tyberg, Christy S. et al, Abstracts of Papers, 224th ACS
National Meeting, Boston, MA, U.S., August 18-22,
2002 (2002), PMSE-115. American Chemical Society,
Washington, D. C.
CONCLUSIONS
Spectroscopic ellipsometry, small angle X-ray
scattering and atomic force microscopy were shown to
have significant capabilities as on-wafer techniques to
characterize the porosity in porous SiLK dielectric
films. Information on void fraction, pore size, pore
size distribution and pore morphology in porous SiLK
films can be obtained using a combination of all three
technologies. Commercialization of fab-quality X-ray
scattering tools and improvements in AFM image
analysis are positive indicators that production-worthy
metrology will soon be available for porosity
characterization of porous SiLK dielectric films.
5. Pederson, J. S., J. Appl. Cryst. 27, 595, (1994).
6. Fraunhofer USA, Center for Coatings and Laser
Applications, Michigan State University, East Lansing,
MI 48824
7. Landes, B.C.; Kern, B.., Ouellette, K.; Yontz, D.,
Godschalx; Niu, J.; Lyons, J; King, D.; DeLong, M.;
Mohler, C.; Hahnfeld, J.; International SEMATECH
Ultra Low-k Workshop, June 6-7, 2002.
8. lacopi, F.; Tokei, Z.; Stucchi, M.; Brongersma, S.H.;
VanHaeren, D.; Maex, K.; Microelectronic Engineering,
65, 123(2003).
9. Tokei, Z.; lacopi, F.; Richard, O.; Waeterloos, J.;
Rozeveld, S.; Beach, E.; Mebarki, B.; Mandrekar, T.;
Guggilla, S.; Maex, K.; Proceedings of the Metals and
Advanced Metals Conference 2003, to be published.
ACKNOWLEDGMENTS
We thank Arun Srivatsa and Thierry N guy en of
KLA-Tencor for consultation on recipes for the
ellipsometric measurements.
We thank Thomas
Schuelke and Mahmut Yaran of Fraunhofer USA for
the SAWS data. Synchrotron measurements were
566