DEVELOPMENT OF NANOSILICA-BISMALEIMIDE (BMI) MATRIX RESINS FOR PREPREG TOOLING COMPOSITES:

FORTIFIED TOOLING PREPREG BMI

James M. Nelson1, Andrew M. Hine1, Paul Sedgwick1, Robert H. Lowe1, Emilie Rexeisen2, Rex M. King2, Nicholas Patz3,.

13M Advanced Composite Materials, Industrial Adhesives and Tapes Division, 23M Corporate Research Analytical Laboratory,

3M Center St. Paul, MN 55144, 3Patz Materials and Technologies, Benecia, CA.

ABSTRACT

Bismaleimide nanocomposite matrix resins filled with spherical silica particles were investigated at nanosilica loading levels from 0 to 40 % by weight. The effect of silica concentration on neat resin properties was thoroughly evaluated. Important composite matrix resin mechanical properties including modulus and fracture toughness showed significant, monotonically increasing improvement with increasing nanosilica concentration. Desirable changes in coefficient of thermal expansion, cure exotherm, and hardness were also measured. Silica concentration levels did not adversely affect the cured glass transition temperature or prepreg processability. Properties of carbon fiber laminates made with woven 2x2 twill fabric prepregs at 40 wt% silica loading revealed significant improvements in compression strength, in-plane shear modulus, and composite hardness.

1. INTRODUCTION Considerable current interest exists in the understanding of the effect of silica nanoparticle inclusion on composite matrix resin properties [1-6]. The use of silica nanoparticle technology enables significant composite property improvements through modification of resin properties. A key feature in the implementation of this technology specifically for prepreg resin systems is the ability of nanoscale silica particles to be impregnated of the silica particles into the prepreg structure without filtration. Homogeneous, non-aggregated, and non-agglomerated nanocomposite epoxy prepreg resin are produced by this approach [3-8].

Previous work by the present investigators has demonstrated the benefits of incorporating high loading (up to 45 wt%) of nanoscale silica into epoxy matrix materials for carbon fiber prepreg composites [3-6]. Important composite matrix resin mechanical properties including modulus and fracture toughness increase with increasing nanosilica concentration. Desirable changes in coefficient of thermal expansion, cure exotherm, cure shrinkage and hardness were also measured. Properties of carbon fiber laminates made with unidirectional prepregs of varying silica loading levels revealed significant improvements in compression strength, in-plane shear modulus, and 0° flexure strength.

Recently our development efforts have concentrated on the application area of composite tooling, with particular emphasis on the development of low temperature, out-of-autoclave cure (OOA) epoxy prepreg systems designed for 180 °C cure environments [8]. Properties of carbon fiber laminates made with fabric prepregs at silica loading levels of 43 wt% revealed significant improvements in compression strength, in-plane shear modulus, and 0° flexure modulus. Additionally, properties of particular importance to the area of composite tooling, such as coefficient of thermal expansion, hardness, shrinkage, and exotherm control have been greatly improved through silica incorporation. The effect of reduced coefficient of thermal expansion on thermal distortion of curved parts (“spring-in”) was demonstrated.

This study continues our focus on composite tooling with the examination of the effect of nanosilica modification on bismaleimide (BMI) resin systems and carbon fiber composites. BMI matrix systems represent the other major class of composite tooling resins used in high temperature tooling applications. BMI resins are commonly used in high production rate composite tooling campaigns due to their high glass transition temperatures and their robustness in the thermal cycling of composite tools.

To our knowledge, this appears to be the first evaluation of the effects of nanosilica on a bismaleimide prepreg resin system. Previous work by Vejayakumaran, et al., explored the functionalization of nanosilica with maleimide species with the goal of developing suitable functionality to enable nanosilica incorporporation into a BMI system [9]. In that work, research was limited to the study and characterization of the surface functionalized particles.

Specifically, this paper outlines the development of the 3MTM Fortified Tooling Prepreg Bismaleimide (FTP BMI) resin system. As part of this development, the effect of nanosilica inclusion on the neat resin and prepreg composite properties of a model diallyl bisphenol A /diphenylmethane bismaleimide system was examined. Particular emphasis in this study is placed on very high loading of particles, up to 40% by weight. Fabric carbon fiber reinforcement is used at a nominal fiber volume fraction of 48-64%. As a function of nanosilica concentration, this study examines: the processibility of the resin for prepreg, the quality of nanoparticle dispersion, and neat resin physical and mechanical properties. In addition, carbon fiber laminates using the FTP BMI resin system made of fabric prepregs at 40 wt% silica were compared to an unfilled control and used to establish the viability and characteristics of elevated temperature composites with high nanoparticle loadings.

2. EXPERIMENTAL 2.1 Mater ials and Resin Sample Preparation Resin samples were generated by use and dilution of a nanosilica/diallyl bisphenol A (DAPBA) blend containing 56 wt% nanosilica with additional DAPBA to produce silica containing resins at ca. 56, 46, 33, and 18 wt%. FTP BMI mixtures were created from the DAPBA/nanosilica precursors at a final silica content of 40, 30, 20 and 10 wt % silica through addition of 4,4-bismaleimidodiphenylmethane (BMDM). A control sample containing no silica was also made, allowing comparison to the nanosilica BMI system. Unless otherwise listed, ratios of DABPA: BMDM were 1:1.

BMDM was blended into the pre-warmed (120 °C), nanosilica-filled DABPA and placed in an oven at 120 °C for 1 hour. This blend was mixed with a DAC 600 SpeedMixer (Flacktek, Landrum, SC) at 2350 rpm for 3 minutes to produce well-dispersed blends. These blends were degassed under vacuum for 3-5 minutes prior to being poured into appropriate molds for neat resin tensile testing, dynamic mechanical analysis (DMA), and determination of hardness, coefficient of thermal expansion (CTE), density, and fracture toughness. The samples were cured in a forced air oven for 30 minutes at 150 °C and then for an additional 4 hours at 180 °C. Cured resin plaques were postcured at 220 °C for 6 hours.

2.2 Uncured Resin Test Methods Rheological analyses of these nanosilica-epoxy resin/curative systems were conducted on an ARES rheometer (TA Instruments, New Castle, DE) in parallel plate dynamic mode. Samples 50 mm in diameter (50 mm diameter top plate and bottom plate) and 1 mm thick were heated from 50 °C to 150 °C at 2 °C/min, at a frequency of 1 Hz and a strain of 2%.

The cure exotherm was obtained using a modification of ASTM D3418-08, conducted on a TA Q2000 differential scanning calorimeter (DSC, TA Instruments). Uncured resin samples were heated from -30 °C to 350 °C at 10 °C/min.

Linear shrinkage of the resins during cure was measured using ASTM D2566-86. All interior surfaces of a semi-cylindrical steel trough mold of dimensions 2.54 cm diameter by 25.4 cm length were coated with mold release. The mold was preheated to 150 °C and the liquid resin was poured into the mold. After 30 minutes at 150 °C, the temperature was increased to 180 °C and held at this temperature for an additional 4 hours. Samples were then postcured in the shrinkage mold at 220 °C for 2h. Upon cooling to room temperature, the sample length and the mold length were measured and linear shrinkage was calculated for each resin.

2.3 Cured Resin Test Methods Resin silica content was determined using a 5 to 10 mg cured sample placed in a TA Instruments Q500 thermogravimetric analyzer (TGA, TA Instruments). Samples were heated in air from 30 °C to 850 °C at 20 °C/min. The noncombustible residue was taken to be the resin’s original nanosilica content.

Flexural storage modulus, E', and glass transition temperature, Tg, of cured resins were obtained by dynamic mechanical analysis (DMA) using an RSA2 Solids Analyzer (Rheometrics Scientific, Inc, Piscataway, NJ) in the dual cantilever beam mode. Experiments were performed using a temperature ramp of -30 °C to 350 °C at 5 °C/min, a frequency of 1 Hz, and a strain of 0.03 to 0.10%. The temperature at the peak of the tan delta curve was reported as the Tg.

Density of cured resin specimens was measured by ASTM D792-86, Test Method B where n-heptane (d = 0.684 g/cc) was used as the immersion liquid. For each material, four specimens were measured and the average density was reported in g/cc.

Barcol hardness (HB) was measured according to ASTM D2583-95 (Reapproved 2001). A Barcol Impressor (Model GYZJ-934-1, available from Barber-Colman Company, Leesburg, VA)

was used to make measurements. Vickers hardness was determined for each specimen; between 5 and 10 measurements were made and the average value was reported.

Coefficient of thermal expansion (CTE) measurements were performed using a TMA Q400 (TA Instruments) with a macroexpansion probe. A force of 1.0 N was applied and the specimen lengths were measured at room temperature. The specimens were cycled 10 times from 25 °C to 180 °C. The CTE was recorded as a curve fit from 0 °C to 180 °C on the fifth heat.

Fracture toughness was measured according to ASTM D5045-99 using a compact tension geometry, wherein the specimens had nominal dimensions of 3.18 cm by 3.05 cm by 0.64 cm with W = 2.54 cm, a = 1.27 cm, and B = 0.64 cm. A modified loading rate of 1.3 mm/min (0.050 in/min) was used.

The tensile strengths, failure strains, and moduli of the resins at room temperature were measured according to ASTM D638 using a “Type I” specimen. The loading rate was 1.3 mm/min (0.05 in/min). Five specimens were tested for each silica concentration level.

Nanoindentation studies were performed using an MTS Nanoindenter XP with a DCM module using Continuous Stiffness Measurement (CSM). Load and displacement of the indenter probe into the surface was used to calculate the sample modulus and hardness over hundreds of depths for a single indentation. Each sample was loaded to a maximum force of ca. 17 mN. A Berkovich diamond probe was used to determine the modulus and hardness. Data was averaged over indentation depths from 500-1000 nm.

A Hitachi H-9000 transmission electron microscope (TEM) was used to examine prepared samples. Cured samples for TEM observation were microtomed at room temperature. All samples were cut at the thickness of 87 nm so that a direct comparison could be made between the different wt% particle loadings.

2.4 Carbon Fiber Composite Sample Preparation Fabric prepreg tape for the FTP BMI (40 wt% Si) resin system was produced using T300-6K and T700-12K 2x2 twill carbon fabric. An unfilled control resin was used to generate a control prepreg on the same fabrics. The 127 cm wide fabric prepregs had aresin content of 50 wt%.

Composite laminates were prepared for FTP BMI and the control prepreg using typical vacuum bag techniques to achieve porosity-free samples. Laminates were heated from room temperature to 180 °C at 5 °C/min using 0.6 MPa of pressure. The laminates were cured at 180 °C for four hours, then were allowed to slowly cool to below 37 °C before removal. The resulting laminates underwent a free standing postcure at 220 °C for 4 hours and then were allowed to slowly cool to below 37 °C before removal.

Two types of laminates were made from each 2x2 twill prepreg. Values for n correspond to and 670 (12k) gsm fabrics, respectively: a) [0]4 for compression on 370 (6k) gsm fabric and b) [0]4 cut at 45°, for in-plane shear. Nominal cured ply thicknesses for the two prepregs were 0.35, and 0.64 mm, respectively. A wet diamond saw was used to cut specimens. Compression specimen ends were surface-ground to ensure squareness and parallelism.

2.5 Carbon Fiber Composite Test Methods Scanning electron transmission images were taken using a Hitachi S-4700 Field Emission Scanning Electron Microscope (FESEM) in compositional imaging mode. Compositional imaging uses backscattered electrons and is affected by composition with areas of higher average atomic number appearing brighter in compositional images.

Compression strength of the composite laminates was measured according to the Suppliers of Advanced Composite Materials Association recommended method SRM 1R-94 “Recommended Test Method for Compressive Properties of Oriented Fiber-Resin Composites.” Tabs were cut from twelve-ply laminates of a common commercial carbon fiber prepreg tape made using a [0, 90]3s lay-up. The tabs were bonded using a scrimmed epoxy film adhesive AF163-2 (3M, Saint Paul, MN) so that a consistent gage section of 4.75 mm was obtained. A “Modified ASTM D695” test fixture (Wyoming Test Fixtures, Inc., Salt Lake City, UT) was used with bolt torques of 113 N-cm. A spherically-seated lower platen and a fixed upper platen were used to compress the specimens at a rate of 1.27 mm/min. Nine specimens of each laminate were tested.

In-plane shear modulus was determined by the procedure of ASTM D3518. Eight specimens were tested from each panel. A biaxial extensometer was employed. Following the standard, the shear modulus was taken to be the chord modulus between 2,000 and 6,000 micro-shear-strain.

3. RESULTS

3.1 Effect of Silica Concentration on Processability and Par t Fabr ication Resin viscosity during curing is an important criterion for prepreg resin systems. Rheological analysis using viscosity vs. temperature profiles reveals any possible changes in resin cure profile as a result of increasing silica content. The prepreg manufacturing process requires a resin system with viscosity capable of film formation at a temperature well below the cure temperature. The resin must also flow during cure sufficiently to allow air to be removed from the laminate and to fully wet out all of the fibers. In our approach, the nanosilica dispersion is created in the DAPBA portion of the resin system. Initially, DABPA/nanosilica mixtures were analyzed as a function of silica content displaying increased viscosity with increasing silica concentration as shown in Table 1.

Table 1 DAPBA Processing Data

Silica (wt%)

Complex Viscosity @ 71°C( x 10-1 Pa-s)

Complex Viscosity @ 130°C ( x 10-2 Pa-s)

0 0.2 0.2 18 0.4 0.3 25 0.8 0.6 36 2.0 1.4 56 6.0 4.8

During cure, the resultant BMI resins were examined rheometrically as a function of temperature. Table 2 illustrates the viscosity of each of the resins at the film-forming temperature of 71 °C. Excellent low-void laminates were made from the 40 wt% filled formulation. Representative viscosity profiles for the resin/curative blends are shown in Figure 1 and key viscosities listed in Table 2.

Figure 1. Nanosilica BMI Viscosity vs. temperature profiles

An important finding of this work is the reduction of cure exotherm with increasing silica content. The cure exotherm as a function of silica content is shown in Table 2. A 40 % reduction of cure exotherm is measured by the addition of 40 wt% nanosilica. In previous studies of dicy- and DDS-cured nanosilica-filled epoxy resins, the cure exotherm was found to be reduced proportionally to the organic weight fraction. Nanosilica lowers the extent of exotherm during cure by simply reducing the amount of curable resin present [3-6, 8]. This may be very important for the fabrication of thick parts where heat management during cure is crucial. Also shown in Table 2 are the shrinkage values which show the trend of reduced shrinkage with increased silica content. Both of these features are desirable for composite tool fabrication.

To probe the effect of silica content on nanoparticle dispersion, TEM micrographs were obtained as a function of silica wt% as displayed in Figure 2. These micrographs demonstrate that a non-agglomerated, non-aggregated dispersion was produced over the range of 10-40 wt% silica.

Table 2. Effect of Nanosilica Concentration on BMI Resin Processing and Properties

Property Silica (wt%) 0 10 20 30 40

Complex Viscosity @ 71°C (Pa-s) 0.4 1.5 2.3 3.1 3.9 Minimum Complex Viscosity (x 10-1 Pa-s) 0.4 1.2 2.0 2.7 3.3

Cure Exotherm (J/g) 233 208 184 160 139 Cure Shrinkage (%) 0.66 0.60 0.50 0.42 0.33

Tg (°C) 268.3 269.1 268.5 269.3 270.1 Tensile Modulus (GPa) 4.0 4.6 5.6 6.2 8.3 Tensile Strength (MPa) 89.7 100 58.1 79.5 70.6

Tensile Strain (%) 1.7 1.5 1.2 1.0 0.9 Fracture Toughness (MPa-m1/2) 0.44 0.64 0.77 0.87 0.96

Barcol Hardness (HB) 55 62 72 76 81 Density (g/cc) 1.23 1.30 1.37 1.45 1.51

Figure 2. TEM Micrographs of nanosilica dispersion at various nanosilica levels. a) 40 wt%; b) 30 wt%; c) 20 wt%; d) 10 wt% (All at 10,000x instrument magnification)

3.2 Effect of Silica Concentration on Cured Resin Properties

Reduced coefficient of thermal expansion is desirable for tooling materials in order to reduce thermal stresses and part distortion. As shown in Table 3, incorporation of 40 wt% silica lowered the CTE by 40%. As seen in Table 3 increasing silica incorporation leads to an increase in surface Barcol hardness. Nanoindentation [10, 11] was used to further confirm the higher hardness of the nanosilica-BMI system. Results are summarized in Table 3. The high hardness enhances durability and part surface quality, while imparting significant scratch resistance, an

important property for composites used for the manufacture of composite tooling. High glass transition temperatures, higher hardness, lower CTE, and lower cure exotherm are also particularly important in composite tooling applications areas including tooling, as well as many aerospace applications.

Table 3. Nanosilica-DABPA/BMI Mechanical Property Data

Silica (wt%)

CTE (μm/m/°C)

Nanoindentor Hardness

(GPa)

Nanoindentor Modulus

(GPa) 0 40 0.3 4.0 40 24 0.6 10.0

Density measurements were conducted on cured resin samples ranging from 0-40 wt% nanosilica and results are displayed in Table 2. The inclusion of nanosilica in resins increases the density of the resultant system because the density of silica is higher than that of the base resin. The measured densities and weight fractions of silica can be used to verify the nominal density of these particles is between 2.2 and 2.3 g/cc. Typical carbon fiber prepregs have fiber volume fractions of about 60%, so the increase in density of prepreg-based composites with nanosilica modification is a few percent. As will be seen, the accompanying gain in composite properties offer composite designers latitude in eliminating carbon fiber and other weight- and cost-saving strategies. These can result in an overall reduction in part weight for equal strength or stiffness.

Cured resin tensile tests were performed to directly measure the resin tensile modulus. Table 2 lists the tensile modulus as well as the average stress and strain at failure. At the 40 wt% level of nanosilica, the tensile modulus is 108 % higher than the control resin. The variability of failure stress and strain can be attributed to the flaw-sensitive nature of the test. In general, increased silica content appeared to produce similar strength levels with reduced failure strains for this base resin system as the modulus increased.

Glass transition temperatures were also measured by DMA and found to be uninfluenced by the presence 10-40 wt% nanosilica (Table 2). Resin modulus values obtained via nanoindentation methods support the measured increase in modulus at high silica content (Table 3).

Results of resin fracture testing are given in Table 2. The critical plane-strain stress intensity factor, KIC, increased with increasing nanosilica content. The incorporation of 40 wt% silica increased KIC by about 105 %. Fracture toughness values remain considerably higher than the unfilled control at silica levels between 10 and 40 wt%.

3.3 FTP BMI Composite Mechanical Proper ties Previously, the effect of nanosilica concentration on composite mechanical properties was examined using both 121 °C and 180 °C-cured systems suitable for unidirectional tape prepreg manufacturing [1-4]. These studies were extended to a tooling-specific fabric-based prepregs with epoxy resin systems designed with composite tooling in mind [8]. These examinations revealed significant composite mechanical property enhancements, as discussed in the

introduction. This section shows the translation of the enhanced BMI properties discussed above into key properties in composite laminates

Incorporation of nanomaterials into fiber-reinforced composite structures can be difficult due to particle filtration by the reinforcing fibers. To probe the dispersion of nanosilica in these composites, field emission scanning electron microscopy was used to examine polished cross sections of the silica-containing laminates. Images illustrating the homogeneous dispersion of silica in a FTP BMI laminate are shown in Figure 6.

Figure 6. Field emission SEM image showing homogeneous distribution of 86 nm diameter silica particles between individual T700 carbon fiber (diameter ca. 7 microns) in a T700-12k/FTP BMI

laminate.

Composite laminate data was generated on T700-12K 2x2 twill fabric prepreg for the FTP BMI versus a the unfilled control,with the exception of compression strength values which were conducted using T300-6k fabric laminates. Composite data for the study is compiled in Table 4.

Compression strength results for FTP BMI system and the controlwere measured using laminates of significantly different fiber volume fraction. It is notable that even at much lower fiber volume fraction the nanosilica-modified composite had equal strength to the control. If the strength values are normalized to equal fiber volume fraction, the change from the control to the FTP BMI material is 30 %. This improvement is consistent with previous studies on the effect of nanosilica inclusion on composite laminate properties. Previous work in our laboratories has shown that the incorporation of silica into epoxy-carbon fiber systems at up to 45 wt% silica loadings produces significant improvements in compression strength [3-6].

Table 4. Summary of FTP BMI Carbon-fiber Composite Properties

Property Control FV FTP BMI FV

Silica (wt%) 0 - 40 - In-plane Shear Modulus (GPa) 4.5 58 5.8 63 Compression Strength (GPa) 0.7 61 0.7 48.1 Modulus Nanoindentation:

Resin Region (GPa) 4.8 59 15.3 61

Modulus Nanoindentation: Fiber Region (GPa)

14.7 59 16.8 61

Hardness Nanoindentation: Resin Region (GPa)

0.3 59 0.8 61

Vickers Hardness: Resin Region (HV)

41 59 56 61

Vickers Hardness: Fiber Region (HV)

92 59 93 61

z-azis CTE µm/m/°C 33 59 28 59

Additionally, in-plane shear modulus increased with increased nanosilica content. At 40 wt% nanosilica, the increase over the unfilled control was 29%. However, there is a mismatch in the fiber volume fraction for these panels. If strength values are normalized to equal fiber volume, the change from the control to the FTP BMI material is 18 %. The tensile modulus of the neat matrix resin reported in Table 2 increased about 100% for this loading. The matrix shear modulus trend is expected to follow that of the matrix tensile modulus closely. As previously discussed, micromechanical considerations preclude one-to-one translation of matrix stiffness to composite shear stiffness due to the presence of fibers [3,4].

Although differences in fiber volume fraction make quantifying the amount of property improvemnent difficult, these improvements in compression strength and in-plane shear are consistent with previous studies on the effect of nanosilica inclusion on composite laminate properties. Previous work in our laboratories has shown that the incorporation of silica into epoxy-carbon fiber systems at up to 45 wt% silica loadings produces significant improvements in compression strength, in-plane shear modulus, and 0° flexure strain [3-8]. Further composite laminate testing is needed to complete this study and will be the topic of future publications.

As previously mentioned in the resin data section, increasing silica incorporation leads to an increase in surface Barcol hardness for the bulk resin (Table 1). In an effort to determine if the enhanced neat resin hardness transfers into improved composite laminate hardness, Vickers hardness and the determination of hardness by nanoindentation was explored to further confirm the higher hardness of the FTP BMI system. Resin rich and fiber dominated areas of the laminate were examined and results are summarized in Table 4.

Vickers hardness for the resin-rich regions of the nanosilica-containing laminate displayed a 38% increase in hardness in comparison to the control. At these volume fractions, the fiber rich regions showed nearly identical Vickers hardness values. Similar determinations of

nanohardness via nanoindentation revealed significant hardness improvements for the FTP BMI laminate in the resin-rich regions of 300 %. The high hardness enhances tool durability and surface quality. In a related examination of nanosilica epoxy tooling resins, this increase in hardness led to increased scratch resistance as measured by nanoscratching techniques [8]. Further exploration of the use of nanoscratching techniques to probe scratch resistance will be included in future publications.

0.0%

0.2%

0.4%

0.6%

0 50 100 150 200

Change in t hic kness

Temperature ° C

Control

0.56 %

0.48%

FTP BMI Cha

nge i

n T

hick

ness

%

Figure 7. The Change in Thickness for FTP BMI Laminates vs. the Control Laminate as a Function of Temperature.

The incorporation of silica also influences dimensional stability of fiber reinforced composite structures, particularly the through-thickness (z-axis) coefficient of thermal expansion (CTE). The z-axis CTE was measured for the FTP BMI versus the control and the change in thickness as a function of temperature is shown in Figure 7. Average CTE values for these systems are listed in Table 4. The presence of nanosilica in the FTP BMI resin results in lower CTE which will lead to less thermal distortion of curved laminates commonly referred to as “spring-in” [8, 12].

4. FUTURE WORK The effect of nanosilica concentration resin properties of a model matrix system based on a DAPBA/BMDM system was studied. The neat resin properties for silica concentrations from 0 to 40 % by weight were evaluated. Bismaleimide nanocomposite resins with high weight fractions of nanosilica offer significant enhancements in both resin modulus and fracture toughness. In addition, other resin properties desirable for fiber-reinforced composites were improved. The incorporation of nanosilica loading of 40 wt% produced a prepreg resin with

suitable characteristics for successful panel fabrication. Carbon fiber prepreg composite laminates incorporating 40 wt% nanosilica in the FTP BMI resin displayed improvements in composite properties such as compression strength and in-plane shear over an unfilled control. The presence of nanosilica enhanced other important composite tooling properties such as hardness and reduced the coefficient of thermal expansion. The present study has demonstrated the viability and utility of incorporation of nanosilica into high temperature composite matrix materials at high concentration.

5. ACKNOWLEDGEMENTS We would like to thank Mary Buckett and Mary Swierczek of 3M’s Corporate Research Analytical Laboratory for their assistance with the SEM/TEM images, Rachel Wilkerson for neat resin testing and Daniel Quinn for neat resin sample preparation

6. REFERENCES 1. Hsieh T.H., Kinloch, A.J., Masania, K., Lee, J.S., Taylor, A.C., Sprenger, S. “The toughness of epoxy polymers and fibre composites modified with rubber microparticles and silica nanoparticles.” Journal of Material Science 45 (2010): 1193-1210.

2. Tsai, J-L., Hsiao, H., Cheng, Y-L. “Investigating mechanical behaviors of silica nanoparticle reinforced composites” Journal of Composite Materials 44 (2010): 505-524.

3. Hackett, S.C., Nelson, J.M., Hine, A.M., Sedgwick, P., Lowe, R.H., Goetz, D.P., Schultz, W.J. “Improved carbon fiber composite compression strength and shear stiffness through matrix modification with nanosilica” Proc. American Society for Composites 25th Annual Technical Conference. Dayton, OH, Sept. 20-22, 2010.

4. Hackett, S.C., Nelson, J.M., Hine, A.M., Sedgwick, P., Lowe, R.H., Goetz, D.P., Schultz, W.J. “The Effect of Nanosilica Concentration on the Enhancement of Epoxy Matrix Resins for Prepreg Composites” Proc. Society for the Advancement of Materials and Process Engineering, SAMPE 2010, Salt Lake City, Oct. 11-14, 2010.

5. Nelson, J.M., Hackett, S.C., Hine, A.M., Sedgwick, P., Lowe, R.H., Quinn, D.J.; Goetz, D.P., Schultz, W.J. “Development of Nanosilica-Epoxy Matrix Resins for High Temperature Prepreg Composites” Proc. Society for the Advancement of Materials and Process Engineering, SAMPE 2011, Long Beach, CA., May. 23-26, 2011.

6. Nelson, J.M., Hackett, S.C., Hine, A.M., Sedgwick, P., Lowe, R.H., Quinn, D.J.; Goetz, D.P., Schultz, W.J. “Development of Nanosilica-Epoxy Matrix Resins for High Temperature Prepreg Composites” American Chemical Society, Division of Polymer Chemistry, Symposium Proceeding, Denver CO, August, 29-31, 2011.

7. US Patent 5,648,407. “Curable resin sols and fiber-reinforced composites derived therefrom.” (1997).

8. Nelson, J.M., Hine, A.M., Sedgwick, P., Lowe, R.H., Quinn, D.J.; Goetz, D.P.,. Patz, N., Rexeisen, E., King, R.M. “Development of Nanosilica-Epoxy Matrix Resins for Prepreg

Composite Tooling Applications” Proc. Society for the Advancement of Materials and Process Engineering, SAMPE 2012, Long Beach, CA., May. 21-24, 2012.

9. Vejayakumaran, P., Rahman, I. A., Sipaut, C. S., Ismail, J. and Chee, C.K. “Structural and thermal characterizations of silica nanoparticles grafted with pendant maleimide and epoxide groups.” J. Colloid Interface Sci. 328, 81-91, 2008.

10.Beake, B.D., Leggett, G.L.,”Nanoindentation and Nanoscratch Testing of Uniaxially and Biaxially Drawn Poly(ethylene terephthalate) Film” Polymer 43 (2002): 319-327.

11. Wang, H., Feng, J., Hu, X., Ng, K.M., “Tribological Behaviors of Aligned Carbon Nanotube/Fullerene-Epoxy Nanocomposites” Polymer Engineering and Science (2008), 1467-1475.

12. Radford, D.W., Rennick, T.S. “Separating sources of manufacturing distortion in laminated composites.” Reinforced Plastics and Composites 19 (2000):621-641.