Sei sulla pagina 1di 7

American Institute of Aeronautics and Astronautics

1
Nanoparticulate Mitigation of Process-Induced Errors in
Carbon Fiber Reinforced Composite Mirrors

Garrett J. Lu
*
and H. Thomas Hahn


University of California, Los Angeles, California, 90095-1597
Jake D. Hochhalter


University of New Mexico, Albuquerque, New Mexico, 87131-001
and
Arup K. Maji


Air Force Research Laboratory, Albuquerque, New Mexico, 87117-5776


The drive to reduce the weight of space-based telescope mirrors has been of great
importance as payload costs represent a significant inhibiting factor for the advancement of
space technologies. Composite replica mirrors are by far the lightest of the proven new
technologies that can be applied to space applications and have the potential to be the
cheapest to manufacture as well. Using carbon fiber reinforced polymers with areal densities
as low as 2 kg/m
2
, it will be possible to fabricate and deploy significantly larger mirrors at a
fraction of the current cost. The ability to successfully fabricate composite mirrors is limited
by its resulting surface roughness. On the microscopic level, fiber print-through is a leading
cause of concern. Print-through occurs when carbon fibers in the matrix are transposed to
the surface of the laminate during autoclaving as a result of cure shrinkage, causing high
frequency errors. Through scatterometry measurements, we can quantify the roughness
values as a function of fabrication parameters and gain a clearer understanding of the
problem. One accepted approach towards alleviating this problem is the introduction of a
resin-rich layer on the surface of the laminate. By increasing the thickness of this resin layer,
fiber print-through roughness can eventually be eliminated completely. However, as the
resin layer thickness increases, significant side effects also begin to emerge, including the
introduction of voids as well as macroscopic surface figure errors due to increased resin flow
during autoclaving. Nanoparticles with low coefficients of thermal expansion dispersed into
a resin-rich layer can aid in the reduction of cure shrinkage. In addition, the size and shape
of the particles serve as a means of buffering the transposition of fibers to the surface of the
laminate. This research project focuses on quantifying the laminate surface roughness and
investigating the use of nanoparticles as a method of fully mitigating fiber print-through.

I. Introduction
OLLOWING the success of the Hubble Space Telescope mission during the past decade, and looking beyond the
James Webb Space Telescope, set to launch in 2011, it is clear that the future of space-based telescopes lies in

*
Graduate Student Researcher, Multifunctional Composites Laboratory, UCLA Department of Mechanical &
Aerospace Engineering, 420 Westwood Plaza 18-121, Los Angeles, CA 90095-1597; AIAA Student Member

Chair & Professor, UCLA Department of Mechanical & Aerospace Engineering, 420 Westwood Plaza 18-121, Los
Angeles, CA 90095-1597; AIAA Senior Member

Graduate Student Researcher, Department of Civil Engineering, MSC 01 1070, University of New Mexico,
Albuquerque, NM 87131-001.

Senior Research Scientist, Air Force Research Laboratory, Space Vehicles Directorate, AFRL/VSSV, Building
472, Albuquerque, NM 87117-5776.
F

American Institute of Aeronautics and Astronautics
2
our ability to minimize the weight and simplify the fabrication techniques of the optical components. Not only will
the mirrors have to function at cryogenic temperatures while in orbit, they will also reach diameters as large as 6
meters or greater
1
. Traditional glass mirrors, such as those found on the Hubble, are bulky and particularly
expensive to carry as payload into space
3
(>$10,000/kg). In addition, they require a sophisticated polishing process
to obtain the needed surface optical quality
2,3
. However, advances in the development of lightweight mirrors can
already be seen on the James Webb Space Telescope. Being jointly developed by Northrop Grumman and Ball
Aerospace, the facesheet will utilize beryllium
2
, a material that is extremely lightweight, but is also toxic and
difficult to handle during fabrication. A relatively new approach to fabricating lightweight mirrors is a process
known as composite mirror replication, in which a carbon fiber reinforced polymer (CFRP) composite facesheet is
used to reproduce a high quality optical surface that is identical to a pre-polished mandrel during the cure cycle
4-6
.
CFRP facesheets currently being produced possess the distinct advantage of having areal densities as low as 2 kg/m
2

and a very low coefficient of thermal expansion
4-6
.



The replication method is quite simple in theory, and begins with a lay-up process of the CFRP laminate,
followed by the cure cycle where the laminate is in direct contact with an optical quality mandrel under high
temperature and pressure. Theoretically, the laminate will be an exact replicate of the optical surface, eliminating the
need for a separate polishing process, thus reducing time and cost. This process, however, has actually proven to be
very difficult, with one dominating issue that is preventing complete replication of the mandrel. This problem is
known as fiber print-through, and is a common occurrence in CFRP composites.
Fiber print-through occurs when, during the cure cycle, carbon fibers in the matrix are transposed to the surface
of the mirror causing high frequency surface errors (>20 cycles across the aperture)
7
. This imparts a rough texture to
the surface that scatters light and causes microscopic errors in the optical figure. There are two main reasons why
fiber print-through appears on the surface of a laminate, the first being the reduction of matrix volume due to resin
bleeding. Incorporating anti-bleeding measures such as Teflon tape and a resin dam during the cure cycle can
usually eliminate this issue. The second, more important reason for the appearance of fiber print-through is the cure
shrinkage of the resin during autoclaving. Thermosetting resins solidify due to the cross-linking of polymer chains
during the chemical curing reaction, and a significant volumetric shrinkage occurs because of the formation of these
chemical bonds
8,9
.
II. Mirror Replication Process
CFRP composites are anisotropic due to the fact that its material properties are dominated either by the fibers or
the polymer matrix, depending on the direction. For the purposes of designing a composite space mirror, it is
beneficial to have an isotropic material. Thus, to achieve at least a quasi-isotropic laminate, 12 plies of IM7/977-2
preimpregnated carbon fiber were laid up on top of one
another in a [0,(60)
2
,0]
s
orientation. This lay-up was
chosen because it minimizes the curvature of the part
after autoclaving and is 20-mils thicker than the more
traditional [45,0,90]
s
quasi-isotropic layup
7
. For testing
purposes, 6-inch diameter laminates were fabricated. The
optical quality glass mandrel used was a zerodur flat
with surface finish 5/8 scratch/dig and /10 peak-
valley flatness. Before use, the zerodur mandrel was
coated with a layer of mold release to allow for the easy
removal of the laminate after cure. Loctite-Frekote mold
release was selected due to its low solid content as not to
interfere with the surface roughness specifications.
After the lay-up process, the 12-ply laminate was
placed directly onto the optical surface of the zerodur mandrel along with an aluminum flat placed on the opposite
side of the composite. Teflon tape was then adhered to the outer diameter of the part to minimize resin bleeding, and
~ 2 kg/m
2
15-25 kg/m
2
180 kg/m
2
Areal Density
CFRP Facesheet Beryllium Facesheet Glass Facesheet

Table 1: Areal density comparison of various facesheet materials.
Figure 1: Consolidation and bagging of laminate.

American Institute of Aeronautics and Astronautics
3
a silicone rubber resin dam was fitted around the outer diameter to provide consolidation. During autoclaving, the
desired optical side was placed facing up. Under the assumption that there is no resin bleeding during the cure cycle,
the fibers in the matrix are theoretically allowed to move and settle as the resin begins to lose viscosity. Thus, it
should not matter which way the laminate is cured because the surface resin thickness is the same in both cases.
Evidence from photomicrographs helped support this concept as samples cured either face up or face down had
approximately the same degree of fiber print-through.
7


A. Resin-Rich Layer Application
7

To reduce the effects of fiber print-through, the first approach taken was to apply additional resin layers to the
surface of the laminate in either a post-cure or a co-cure process. The presumption behind this method is that with no
resin bleeding, the additional resin layers will eventually cover up the fiber roughness. The resin layers are a
polycyanate film adhesive in solid form obtained from
YLA Inc., and are manufactured for space-based
applications, providing maximum thermal expansion
compatibility with the CFRP laminate. Each layer of
additional resin provided about 5 mils of thickness on the
optical surface. While it is theoretically possible to fully
eliminate the fiber print-through just by increasing the resin
layer thickness, there are other issues that come in to play.
With just 5~10 mils of extra resin thickness, serious
surface figure errors can development due to the slightest
incline in the autoclave during the cure cycle. In addition,
increasing the resin thickness also increases the chances of
introducing voids to the laminate surface.
III. Results and Discussion
The laminate was examined with an Atomic
Force Microscope (AFM) to carry out preliminary
inspections of the fiber print-through roughness.
Under tapping mode, the AFM tip contacts the
surface of the material intermittently in the
prescribed scan area (50m), and from the
deflection of the tip and the force feedback, a
surface profile is established. From Fig. 3, a very
clear representation of the transposed fibers can
be seen protruding the surface. A detailed
topography analysis of the vertical displacement
of these fibers indicated an average peak-valley
roughness on the order of 300-500 nm. Multiple
tests conducted at different locations on the
laminate surface demonstrated that the roughness
values increased slightly as the measurements
tended towards the outer diameter.

A final metal reflective coating can be applied to the laminate surface
via magnetron plasma sputtering once the part has been cured.
Magnetron sputtering is a high-rate, high-energy sputtering source that
results in 90% reflectivity
7
. This process does not distort the laminate
surface because of its relatively thin deposition layer. Thus,
measurements can be taken to directly determine the affects of fiber
print-through without any data contamination from the reflective
coating. Additionally, an alternative method of applying a reflective
coating via a CuZr nanolaminate is currently being investigated.
Figure 4: Mirror with an Al coating.
Figure 2: Laminate with carbon foam backing
and additional polycyanate resin film.
Figure 3: Atomic force miscroscopy of fiber print-through.

American Institute of Aeronautics and Astronautics
4
A. Scatterometry Measurements
Quantification of the optical surface was conducted using
scatterometry measurements. The sample is held by a housing and
leveled on the focus plane. In scatterometry, a reference detector is
used to determine the amount of energy in the beam used as the
light source. With the beams initial energy and diameter known, it
is possible to measure and make use of the scatter at a range of
angles relative to the sample surface after the beam is reflected. The
scatter energy, at multiple angles in the plane of incidence, is
measured and recorded with a photo-multiplier. In scatterometry,
there is a measured incoming light direction e
i
, and an outgoing
reflected direction e
s
, each defined relative to a small surface
element. A bi-directional reflectance distribution function (BRDF)
is defined as the ratio of the reflected radiance in direction e
s
(measured in a small solid angle dO), to the amount of
irradiance that reaches the light source in direction e
i
. The amount of light reflected in the e
s
direction is given by
dP and the amount of light scattered in the e
i
direction is given by P
i
-cos(u
s
). Therefore, the BRDF is given by the
following equation:
10

(1)
The energy scatter due to the surface is measured in one-dimensional slices. Shown in Fig. 6 are two separately
oriented one-dimensional slices that quantify the effects of fiber print-through according to Eq. (1). The scatter
measurements and roughness values can be found in Table 2.
One implication in scatterometry is that there are limits to what type of surface can be measured. In
scatterometry no smooth surface limit is set. However, the Rayleigh smooth surface criterion states that Eq. (2) must
be satisfied, and for mid-IR to UV the results are most accurate when the left hand side is <0.1:
10


(2)
Where: o rms surface roughness
u
i
angle of incidence (7.5)
wavelength of source (632.8 nm)


ZD4
0-mil resin thickness
1.E-05
1.E-04
1.E-03
1.E-02
1.E-01
1.E+00
1.E+01
1.E+02
-60 -50 -40 -30 -20 -10 0 10 20 30 40 50 60
ANGLE RELATIVE TO SURFACE NORMAL(DEG)
B
R
D
F

(
1
/
s
r
)
Horizontal Fiber Orientation
Vertical Fiber Orientation

Figure 6: Anisotropy due to fiber orientation observed in two BRDF plots.
7

Scattered Energy, e
s

Incoming
Energy, e
i

Figure 5: Light scatter due to CFRP.
e.

American Institute of Aeronautics and Astronautics
5
For the scatterometry test set-up conducted, the criterion set forth in Eq. (2) was not met for samples shown in
Table 2 with rms roughness > 50 . However, the scatterometry test is still a viable one for determining whether a
surface is isotropic. In the observation of fiber print-through, the ability to quantify anisotropy has proven to be
equally purposeful as an rms roughness magnitude.
There are two very distinct scatter patterns that appear as a result of two different orientations of the laminate
surface as seen in Fig. 6. When the fibers are oriented vertically during the test, the scatter from the fiber print-
through is the dominating cause of the roughness values being measured. However, when the fibers are oriented in
the horizontal direction, the scatter is in the vertical direction, perpendicular to the plane of measurement. Thus, a
higher value of BRDF is measured in a more narrow range of angles. From this, it can be concluded that the sample
surface roughness is anisotropic, dominated by the fiber print-through. Shown in Fig. 7 below is a typical elongated
scatter pattern produced by the fiber print-through.


Figure 7: Scatter pattern of a CFRP laminate (vertical fiber orientation) with no resin thickness.
7

Table 2: Summary of roughness results.
7

Laminate ID Fiber Orientation
Roughness
()
Resin Rich Layer
Thickness
(mil)
Cure Pressure
(psi)
ZD1 Horizontal 112.86 10 45
ZD1 Vertical 113.41 10 45
ZD2 Horizontal 175.66 5 45
ZD2 Vertical 184.28 5 45
ZD3
(Spherical) Vertical 166.61 5 45
ZD3
(Spherical) Horizontal 167.19 5 45
ZD4 Horizontal 124.22 0 45
ZD4 Vertical 267.98 0 45
ZD6 Horizontal 173.49 5 100
ZD6 Vertical 178.68 5 100
ZD7 Bottom Right 184.79 10 100
ZD7 Center 159.08 10 100
Note: Roughness values were calculated over a bandwidth of 0.65m 73m

As shown in Table 2, with increasing resin layer thickness, the roughness values decreased in both the horizontal
and vertical orientations. Anisotropy was reduced significantly with the 5 mils of resin thickness, but scatter pattern
differences were still noticeable. In addition, anisotropy in the surface roughness was virtually eliminated at ~10
mils of resin thickness, although at the expense of other problems previously stated. It should also be noted that
increasing the cure pressure to 100 psi (ZD6) did not significantly change the overall roughness values. In ZD7, a
wide variation in roughness is seen as when comparing a measurement in the center as opposed to a measurement in
the outer radius. A spherical sample tested with 5 meters radius of curvature (ZD3) did not show an increase in
roughness value either.


American Institute of Aeronautics and Astronautics
6
IV. Incorporation of Nanoparticles
It has been shown in a variety of applications that nanoparticles in a polymer matrix resist and reduce the effects
of cure shrinkage in the resin
11-13
. Nanoparticles in general maintain a lower coefficient of thermal expansion, and
act in a similar manner to that of tiles in mortar. As has been demonstrated in advance integrated circuits,
nanoparticles are small enough to fill gaps as narrow as 50-60 nm in the complex circuitry. During the fabrication of
ICs, the addition of a large fraction of nanoparticles that shrink less within the spin-on polymer material allows the
shrinkage of the higher CTE filler around the particles to be compensated
13
. Likewise, in the case of fiber print-
through, the nanoparticles can theoretically fill the gaps between the fibers, not only reducing the cure shrinking of
its surrounding resin, but also providing a means of support for the matrix to reduce the peak-valley roughness.


A number of nanoparticles exist today that have the potential to be
used as filler material in a resin matrix. Aluminum oxide (Al
2
O
3
) and
silicon carbide (SiC) both can be manufactured in the nanometer
range, and possess a lower coefficient of thermal expansion compared
to that of epoxy resin as shown in Table 3. From the previous
roughness measurements, it is clear that the surface of the laminate
contains fiber print-through on the order of hundreds of nanometers to
the single micron range. Thus, in order to fit between the valleys of
the print-through, the nanoparticles should be at or less than 100 nm.
Transmission electron microscopy of Al
2
O
3
nanoparticles seen in Fig.
9 have shown that the average particle size is on that order or even
less. Further work is currently being conducted in this area for the full
mitigation of fiber print-through in CFRP composite mirrors.




Figure 8: Theoretical effects of nanoparticles on fiber print-through.
Table 3: Comparison of coefficients of thermal expansion.
Figure 9: Transmission electron
microscopy images of Al
2
O
3
.

American Institute of Aeronautics and Astronautics
7
V. Conclusion
An investigation was conducted to quantify and further understand the effects of fiber print-through on replicated
CFRP composite mirrors. 12-ply laminates were fabricated in the [0,(60)
2
,0]
s
orientation and were examined using
atomic force microscopy and scatterometry to determine the roughness imparted by the fiber print-through. Peak-
valley values were determined to be on the range of 300-500 nm on the AFM, while scatterometry results concluded
that the anisotropy in roughness values when changing measurement orientations was a clear product of the fiber
print-through. Additional resin layers reduced and eventually eliminated this anisotropy, but resulted in a series of
problems, including macroscopic surface figure errors and voids. Through the incorporation of nanoparticles with a
thin layer of additional resin, fiber print-through can theoretically be eliminated. Additional studies are currently
being conducted on multiple types of nanoparticles, including the distribution of particles within a liquid resin
matrix to be applied prior to the curing cycle. Increasing the particle loading as well as using alternative methods of
dispersion may also aid in this goal.
Acknowledgments
Garrett J. Lu and H. Thomas Hahn would like to acknowledge and thank the Air Force Office of Scientific
Research (AFOSR) for their generous support through grant number F49620-02-1-0414. Garrett J. Lu would also
like to express his gratitude to Leslie Sasa, Audrey Pool ONeal, and Hyuncheol Kim for their time, support, and
knowledge. All of the initial fabrication, analysis, and measurements were conducted at the Air Force Research
Laboratory in Albuquerque, NM under the exceptional guidance of Jake D. Hochhalter, Jack Massarello, and Arup
K. Maji. The authors would also like to thank Brad Huschka and Jeffrey Welsh from Kirtland Air Force Base for
their support and technical expertise.
References
1
Burge, J. H., Angel, J. R. P., Cuerden, B., Martin, H. M., Miller, S. M., and Sandler, D. G., Lightweight Mirror Technology
Using A Thin Facesheet With Active Rigid Support, Space Telescopes and Instruments V, 1998, pp. 690-701.
2
Seery, B. D., The James Webb Space Telescope (JWST): Hubbles Scientific and Technological Successor, NASA
Goddard Space Flight Center, 2001.
3
Dressler, A., et al., Hubble Space Telescope and Beyond, Space Telescope Science Institute, Baltimore, MD, 1996, URL:
http://saturn1.hst.nasa.gov/ngst/.
4
Chen, P. C., et. al., "Advances in Very Lightweight Composite Mirror Technology," Opt. Eng., Vol. 39, Sept. 2000, pp.
2320-2329.
5
Chen, P. C., Romeo, R. C., "Fabrication and Testing of Very Lightweight Composite Mirrors, Proc. SPIE 3356, 1998, pp.
938-945
6
Chen, P. C., Saha, T. T., Smith, A. M., Romeo, R. C., "Progress in Very Lightweight Optics Using Graphite Fiber
Composite Materials, Opt. Eng. 37(2), 1998, pp. 666-676
7
Hochhalter, J., Maji, A., and Reicher, D., Process Induced Errors in Replicated Carbon Fiber Reinforced Polymer Mirrors,
ASCE Space 2004, March 2004, ISBN 0-7844-0722-3, pp. 875-882.
8
Marissen, I. R., Curing Shrinkage and Residual Stresses in Visco-Elastic Thermosetting Resins and Composites, Delft
University of Technology, ISBN 90-9013686-X, May 2000
9
Connell, S. J., Abusafieh, A., Lightweight Space Mirrors from Carbon Fiber Composites, SAMPE Journal 38.4, 2002, pp.
46-55.
10
Stover, J. C., Optical Scattering: Measurement and Analysis, 2
nd
ed., SPIE The International Society for Optical
Engineering, May 1995
11
Davidson, R. S., Nanotechnology The Application of Nanosized Particles, The International Centre for Coating
Technologies, Citiflour Ltd., Dec. 2003.
12
Angeletakis, C., Nguyen, D., TRAN, C., New Low Shrinkage Composite With Submicron Fillers, Kerr Corporation,
Orange, CA, 32
nd
Annual Meeting and Exhibition of the American Association for Dental Research, Mar. 2003.
13
Rittner, M., Electronics and Optoelectronics Dielectric Nanoparticles Could Fill Trenches in Advanced ICs,
Nanoparticle News, Vol. 6, Number 11, Dec. 2003, pp. 1-2

Potrebbero piacerti anche