|
|
 |
|
   
|
||||||
|
|
FOCUS: Vaccum-assisted resin transfer molding using tackified fiber preforms
INTRODUCTION Vacuum-Assisted Resin Transfer Molding (VARTM processes, such as the Seemann Composites Resin Infusion Molding Process (SCRIMP), are recognized as low-cost methods for manufacturing largesized composites. These methods are widely used in non-aerospace applications such as marine, transportation, and civil infrastructure industries. However, the use of SCRIMP in fabricating aerospacegrade composites has been hindered by the difficulties in achieving mechanical performance and quality comparable to that of an autoclave/ prepreg molded composite structure. Accomplishing this would require a comparable fiber (> 60%) and void (< 1%) content using similar fiber reinforcements. Preforming becomes very crucial since the compression force offered by a vacuum can achieve only a fiber content less than 50%. Tackification and/or other methods are needed to make fiber preforms with a higher fiber content (i.e. > 60%). Since high temperature resin systems are used in the molding process, the fiber preform must maintain a high fiber content throughout the entire mold filling and curing process. A proper permeability is also needed to facilitate mold filling for larger composite structures. The tackification process has been investigated by several researchers (1-5). Kittelson and Hackette (1) suggested an optimal concentration and preforming temperature for tackification using PT500 tackifier. Rohatgi and Lee (3) studied the moldability of the tackified fabrics, and found that the permeability of the tackified preform would increase greatly if tackifier was first dissolved in the solvent, then applied to the fiber reinforcement. Shih and Lee (4, 5) found that this phenomenon can also be observed in the dry powder tackification process, in which the permeability of the tackified preform depends strongly on the location and concentration of the tackifier. The effect of tackifier on mechanical properties of composites has also been studied (1, 5, 6). It was found that the presence of uncatalyzed, chemically incompatible, and/or unevenly distributed tackifier reduces the mechanical properties of molded composites. Adding a rubbery tackifier can improve the toughness of composites (6). None of the studies, however, could distinguish the difference between tackifier and voids when measuring the mechanical properties of the composites. In this work, the feasibility of using tackified preforms in high temperature SCRIMP is investigated. The effects of tackification methods on composite dimension control, void content and mechanical properties are also investigated and compared in both RTM and SCRIMP. EXAs Materials The tackifier PT 500 and the resin PR 500 were evaluated in this work. PT 500 is a dry powder tackifier manufactured by 3M. It has a melting point of about 70*C and its chemistry is similar to 3M's PR 500 resin. PR 500 is a high-performance, one-part epoxy based on the fluorene moiety, which enhances the glass transition temperature of the epoxy resin while providing high ductility and low moisture absorption. PR 500 is a viscous paste at room temperature (viscosity around 1000 poise at 22C) fn, thus needing to be heated to reduce the viscosity for ease in mold filling. The recommended resin injection temperature is about 1600C, at which the curing agent can be melted and not filtrated by the fiber preform during filling. Cure and rheological behaviors of PT 500 tackifier and PR 500 resin were measured using a Differential Scanning Calorimeter (DSC, TA29 10) and a Rheometrics Dynamic Analyzer (RDA II). In DSC, reactions were conducted in volatile aluminum sample pans, which are capable of withstanding at least 2 atm internal pressure after sealing. The sample weight was about 10 mg. The scanning DSC curves (Fig. 1) show that PR 500 (284.4 J/g) has a higher heat of reaction than PT 500 (179 J/gh under a temperature increase rate of 5C/min from 25degC to 300degC. Conversion of the tacller after preforming can be calculated by dividing the exotherm of the reaction through a specific thermal cycle by the total heat of reaction. In RDA II, a pair of parallel plates of diameter 25 mm was used as the sample cell. Samples were placed between these two parallel plates with a gap of approximately 1 mm between them. Under a temperature increase rate of 5C/min, the scanning RDA curves (Fig. 2) show a much higher viscosity for PT 500 than PR 500about two orders of magnitude. The PR 500 resin shows a decreasing viscosity at the early heating period from more than 100 poise at50degCto less than 1 poise at 180degC. Above 180degC, the chemical reaction starts and the resin reacts to the gel point. A reasonably low viscosity for mold filling can be obtained if the temperature is higher than 140degC. The woven carbon fabrics used in this work are an AS4-6K, five- harness fabric, 390 g/m2, (Hexcel, Inc.) and an IM?-12k, five- harness fabric, 370 g/m2 (textile Products Inc). Preforming The PT 500 tackifier powder at 596 to 14% by weight was applied uniformly on the fabric surface by a sifter. The powder-coated fabric was placed under an IR lamp for several seconds at 1000C to fuse the tackifier to the fabric surface. Nine layers of coated fabrics were placed in a 30-ton press and compressed to 3.15 mm (i.e. fiber content 60% by volume) using different thermal cycles. A spacer was used to control the preform porosity. The location of the tackifier in the fiber reinforcement is controlled by the preforming conditions (3, 4). When the tackified fabrics were preformed at high temperatures, e.g. 177egC for 2 hours, the tackifier was pulled inside the fiber tows by the capillary force because of its low viscosity. On the other hand, when the tackified fabrics were preformed at low temperatures first, e.g. 80degC for 24 hours followed by 177C for 2 hours, the tackifier remained outside the fiber tows. In both cases, the tackifier was fully cured at the end of preforming. The effect of tackifier on preform dimension control was evaluated by both U-shape bending and lateral compression tests. For the U- shape bending test, eight plies of the tackified fabrics were fixed in a Ushaped bending tool (8) under a pre-specified temperature and pressure. After cooling, the preform was taken out of the tool and the springback angle was measured. For the lateral compression test, the fiber preform was compressed by a pneumatically operated press (4) for a specific pressure and thermal cycle. The pressure was then released and the springback of the preform was measured. Nine layers of fabrics were used in each experiment. SCRIMP Molding A schematic of the high temperature SCRIMP is shown in hag. 3. Experiments were based on a highly permeable medium as the resin distribution system (9, 10). As seen from the schematic, only one side of the mold is rigid, while the other side is formed by a vacuum bag. A 30 cm x 30 cm X 1.27 cm aluminum plate was used as the rigid mold. Two grooves, 20.32 cm long and 1 cm wide each and 20.32 cm apart, were cut on the mold surface as the resin inlet and outlet grooves. A hole was drilled from the side of the plate into each groove to provide the tubing connection to both inlet and outlet. A 15.24 cm X 20.32 cm fiber preform was placed between the two grooves on the mold surface. A layer of high temperature peel ply (De-Comp Composites, Inc.) and a layer of highly permeable medium were placed on top of the preform under a high temperature vacuum film (Airtech, Inc.). The edge of the bagging was sealed by high temperature tacky tapes (Schnee-Morehead, Inc.). Two sides of the preform were carefully sealed in order to prevent race tracking, which would result in dry spot formation (11). The mold setup was then placed in an oven to achieve the desirable molding temperature. Metal tubes with 0.95 cm diameter were used to facilitate resin infusion at high temperatures. The PR 500 resin was first degassed at 106degC, then placed in a resin pot kept constant at 1400C to maintain a reasonably low viscosity and reactivity (4). The inlet and outlet temperatures were controlled at 160degC to both ensure that the solid cure agent in the resin was totally melted before entering the preform, and to prevent the resin from freezing at the outlet, thereby blocking the vacuum path. The mold temperature was set at 1800C to provide low viscosity and high reactivity for resin infusion and curing. Two flasks were connected to the outlet to collect and recycle the exces\s resin. The composite was cured at 180degC for 2 hours after the resin gelled. RTM A schematic of the experimental setup is shown in Fig. 4. A high- pressure tank was used as the resin reservoir. A nitrogen gas cylinder was connected to the top of the tank as the pressure source, allowing the resin to be pushed into the mold through a pipe connection. The bottom part of the mold consisted of double O-ring grooves and a cavity of 15.24 cm in diameter and 0.483 cm in depth. Two Viton O-rings were installed around the mold cavity to ensure the sealing of the mold cavity at high temperatures. The border of the mold was machined with a tapered angle for ease of de-molding. A caul plate was used to change the thickness of the molded part. The inlet pipe was connected to the bottom of the mold at the edge of the cavity, and the outlet pipe was attached to the center of the cavity. A circular preform of 13.34 cm outside diameter and 1.9 cm inside diameter was placed at the center of the mold. Using this design, the incoming resin flows around the edge of the preform before flowing into it. During this process, the high mold temperature ensures that the cure agent in the PR 500 resin melts before flowing into the preform. The outlet pipe was connected to a flask, which in turn was connected to a vacuum pump. The preform was placed in the mold cavity and compressed to a desirable fiber volume fraction. The mold was then heated to a high temperature (1600C) before mold filling. The PR 500 resin was degassed in a vacuum oven at 1060C for two hours to remove the dissolved gas and water vapor. The mold was also evacuated for 2 hours. The degassed resin was poured into the reservoir in the high- pressure tank and heated to a temperature at which the viscosity was sufficiently low, then injected into the evacuated mold with a medium pressure of 2.8 x 105 N/m2 (40 psi). After the mold was filled, a process known as packingand-bleeding (11) (i.e., burping) was performed. In this process, the outlet of the mold was closed after mold filling was completed, then the injection pressure was adjusted to a higher level of 106 N/m2 (150 psi). Afterwards, the outlet was opened to purge the voids, then closed again to increase the pressure inside the mold. This procedure was performed several times until no bubbles from the outlet were observed. Finally, the sample was cured for two hours at 1771C. No post-cure is needed. Void Content The void content of the molded composites was measured by microscopy and image analysis techniques. The void content was measured by scanning over the entire surface of the cross section area. Each measured void content was obtained by analyzing an average of 100 digitized images at the magnification of 200x using the commercial software OMNIM= 4. Information on void size, shape, and distribution was gathered by this method. Mechanical Properties The apparent interlaminar shear strength of the composites was measured according to ASTM D2322. The molded composites were cut into sample specimens sized 6.4 mm X 19.2 mm x 3.2 mm. A span-to- thickness ratio of 4 was chosen for the carbon composite samples. Ten specimens were tested for each condition to obtain an average value. The apparent shear strength was calculated by the following equation: RESULTS AND DISCUSSION The PT 500 tackification techniques were recently studied by Shih, Lee, and Coyle (3, 4). Their experimental results showed that tackifier location and distribution depend strongly on tackifier powder size, tackifier concentration, application methods, and preforming conditions. If the tackifier is uncured or partially cured and is initially located on the fabric surface, it remains outside the fiber tows when the preforming temperature is low (e.g. 80degC), but is pulled into the fiber tows by the capillary force when the preforming temperature is high (e.g. 160'C). If the tackifier is initially inside the fiber tows or cured beyond gelation, the preforming conditions do not affect the tackifier location. The tackifier location also has a great influence on the preform permeability. For fabrics preformed at 80degC for 24 hours (i.e. tackifier remains outside the fiber tows), the preform permeability decreases as the tackifier concentration increases. For fabrics preformed at 160*C for 20 minutes (i.e. tackifier stays inside the fiber tows), the preform permeability increases as the tackifier concentration increases. The permeability reaches a maximum value at 170% tackifier concentration, at which the permeability is 4.5 times that of the untackified fiber preform, then decreases afterwards. This is because the tackifier inside the fiber tows shrinks the fiber tows and increases the space outside the tows. Therefore, the permeability of the fiber preform increases (12). As more tackifier is added inside the fiber tows, the stronger force holds the fiber filaments together. This either shrinks the tow further or prevents the fiber tow from deformation during compression. Both effects tend to increase the permeability of the fiber preform. Adding too much tackifier (> 17%) causes some tackifier to stay outside the fiber tows and, consequently, the permeability of the fiber preform starts to decrease. On the other hand, if the tackifier remains on the surface of the fiber mat, it blocks the resin flow and, consequently, reduces the permeability. Since vacuum is the only driving force in SCRIMP, the permeability of the fiber preform together with the resin viscosity dictates the maximum moldable composite size. Several preforming conditions were evaluated in high temperature SCRIMP based on AS4 and IM7 carbon fabrics with PR 500 resin. Table I indicates that the PT 500 tackifier needs to be completely cured in order to maintain its dimensional control ability; otherwise, the tackifier becomes soft or dissolved in the hot resin and loses its ability for preform dimension control during molding. Comparing the two carbon fabrics used in this study, AS4 and IM7 have the same weaving style (fiveharness woven) but different numbers of filaments (6k and 12k). Another major difference between the two is the sizing on the fiber surface. The SEM micrographs in Flg. 5 show the sizing on the two fabrics at 2000X. A thick coating of G- prime sizing is observed on the fiber surface of AS4-6k fabric, while the IM7-12k fabric is relatively sizing-free. This difference in sizing has a strong impact on preform dimension control. Table 2 compares the effect of sizing on fiber preform dimension control when both fabrics are preformed at the same preforming conditions. No tackifier was used in either case. For the U-shape bending test, AS4 preforms show much better springback control than IM7 preforms. The same trend is observed in the lateral compression test. In this test, nine layers of the fabrics were compressed to 3.15 mm during preforming. AS4 preforms showed only 0.01 mm of springback while IM7 preforms showed 0.20 mm of springback. Apparently, the G-prime sizing on AS4 fabrics functions like tackifier, improving the springback control by making the fabrics stiffer and providing more interlaminar adhesion. To further investigate the moldability of these two fabrics, 10% by weight of the tackifier was added to the fabrics to improve the springback control. The tackifier was totally cured either inside or outside of fiber tows by controlling the preforming temperatures. The thicknesses of both fiber preforms and SCRIMP molded composites were measured and compared in Mg. 6. The thickness of the SCRIMP molded composites tends to be larger than that of the fiber preform due to springback of the preform during molding. Results show that, although both fiber preforms can reach 60% by volume of the fiber content, molded composites based on AS4 show very little springback after molding. On the other hand, molded composites based on IM7 show strong springback after SCRIMP molding. The fiber content decreased from 60% to 57% (tackifier inside fiber tows) or 54% (tackifier outside fiber tows) depending on the tackifier location. As mentioned before, the G-prime sizing on the AS4 fabric functions like tackifier, improving the springback control. Without the G-prime sizing (i.e. IM7 preforms), the tackifier location plays a significant role on the springback control. The composite with tackifier inside the fiber tows shows better springback control than the one with tackifier outside the fiber tows. Rohatgi and Lee (2) explained this phenomenon by the consolidation of fiber layers. The micrographs in their work showed that it requires a much larger force to compress the intralayer gaps inside the fiber tows than the interlayer gaps between the fiber tows. Therefore, in order to provide enough holding force to prevent the springback of the compressed intralayer gaps, the majority of the tackifier should stay inside the fiber tows. From the above experimental results and our previous study (2, 12), we conclude that fiber preforms with tackifier inside the fiber tows provide better dimension control and higher permeability than preforms with tackifier outside the fiber tows. However, tackifier inside the fiber tows tends to result in more voids in molded composites. Figure 7 shows the effect of tackifier on void location. If the tackifier stays outside the fiber tows, It tends to cause macrovoid formation outside the fiber tows. On the other hand, if the tackifier stays inside the fiber tows, it tends to cause microvoid formation inside the fiber tows. Macrovoids can be easily purged by packing-and-bleeding, while microvoids are very difficult to remove. To study the effect of tackifier on the void content and mechanical properties of the composites, test samples with the same dimensions were required. RTM composites were therefore molded, because a rigid mold provides consistent dimensions. The resin wa\s degassed first and the mold cavity was evacuated at 10.1 mm-Hg vacuum before mold filling. Packingand-bleeding was carried out several times after mold filling in order to purge the voids. Experimental results of the void content and interlaminar shear strength (ISS) of RTM composites are shown in Fig. & Generally speaking, the void content of the molded composites is less than 1% in most cases. For the same taclfier content, the composite with tackifier inside the fiber tows (i.e. preformed at 160*C for 2 hours) showed many more voids than the composites with tacMer outside the fiber tows (i.e. preformed at 80degC, 24 hours). For composites with taclfier inside the fiber tows, the void content increases as the tacIdfier content increases. For the USS, composites with tackifier inside the fiber tows showed lower ILSS as the tackifier content increased, while the ones with tackifier outside the fiber tows showed similar ILSS until the tackifier content reached 14% (i.e. void content - 0.5%). The ISS correlated well with the void content, i.e. for a void content less than 0.5%, there was little change of ILSS. Above 0.5%, the ILSS decreased as the void content increased. To study the effect of tackifier on the ILSS of the composites, it is necessary to make void-free samples since the presence of voids has a strong effect on the composite ILSS (13, 14). RTM molding experiments were carried out by molding very small tackified fiber preforms (7.62 cm in diameter). The mold was sealed by double O- rings and pre-vacuumed at a pressure of 10.1 mm-Hg for at least three hours. Again, the resin was degassed before mold filling. More than 10 runs of packing-and-bleeding were performed and the mold filling pressure was maintained at 106 N/m2 (150 psi) during the curing stage. The void content and 11,SS of the resulting composites are given in Table 3. To observe the effect on the composite ILSS, 14% by weight of the tackifier was used. For composites with tacldfier outside the fiber tows, the void content is almost zero (0.05%), and the composite ILSS is very close to that without adding tackifier. This result indicates that the PT 500 tackifier does not affect the IISS of PR 500 based composites as long as the void content is low. On the other hand, for composites with tackifier inside the fiber tows, there are still 0.5% microvoids even after purging and the ILSS is much lower than the composite without voids. Again, this confirms that if the tackifier is located inside the fiber tows, more microvoids result and they tend to reduce the ILSS of the molded composites. The void content and ILSS of composites molded by SCRIMP are given in Table 4. For fabrics preformed without tackifier, the SCRIMP experiment was carried out in a rigid mold for thickness control to achieve the same fiber content as other samples. Compared to RTM composites, the SCRIMP composites show the same interlaminar shear strength when the sample thickness and void content are similar. Again, when the tackifier is located inside the fiber tows, the composite samples have more microvoids and lower ILSS. An interesting observation in high temperature SCRIMP is that since both sides of the fiber preform (except inlet and outlet sides) are tightly sealed by tacky tapes, no leakage flow occurs around the sides of the preform. If the resin is too viscous or the preform permeability is too low, the incoming resin flow rate can become much higher than the outgoing flow rate. The resin then accumulates inside the vacuum bag. When the vacuum level decreases inside the bagging, a resin rich area forms above the highly permeable medium. As the outgoing resin starts to gel, this resin rich area keeps growing until the pressure in the vacuum bag equalizes with the atmospheric pressure. A thick, resin-rich layer can form between the peel ply and the fiber preform as shown in Mg. 9. To solve this problem, the inlet valve should be turned off before the resin near the outlet starts to gel. CONCLUSIONS This study showed that high performance, aerospace-grade composites can be manufactured by low cost SCRIMP with the assistance of tackification. Fiber content and mechanical properties of the composites molded by SCRIMP are similar to those produced by RIM. The preforming conditions (i.e. tackifier content and location) are crucial to the final composite qualities (i.e. fiber content, void content, and mechanical properties). To produce composites with high fiber content and low void content, the tackifier needs to be totally cured and located outside the fiber tows. ACKNOWLEDGEMENT The authors would like to thank Bell Helicopter Textron Inc. for financial support and material donation. NSF (DMI-9616456 and EEC- 9612323) also partially funded this research. We greatly appreciate the technical assistance and valuable suggestions from Mr. Lawrence Coyle of Bell Helicopter. REFERENCES 1. J. L. Kittelson and S. C. Hackett, Proc. 39th Int'l SAMPE Symposium, 83 ( 1994). 2. V. Rohatgi and L. J. Lee, Polym. Compos., 31, 720 (1997). 3. C. H. Shih, L. J. Lee, and L. W. Coyle, 55th Annual Forum of the American Helicopter Society, Montreal, Quebec, Canada (May 25- 27, 1999). 4. C. H. Shih and L. J. Lee, J. Compos. Mater., 34, 000 (2001). 5. G. Brinkman, C. Cadenas-Montes. R. Phillips, A. Arnold, and K Bowman, Proc. 40th Int'l SAMPE Symposium, 1523 (1995). 6. R W. Hillermeier, B. S. Hayes, and J. C. Seferis, J. Adv. Mater., 81(4), 52 (1999). 7. 3M Product Introductory Data Sheet 8. J. Xu and L. J. Lee, J. Adv. Mater., 26, 33 (1994). 9. W. H. Seemann, U.S. Pat. 4,902,215 (1990). 10. X. Sun, S. Lt, and L. J. Lee, Cartpos., 19, 807 (1998). 11. K. Han and L. J. Lee, Compos. Mater., 30, 1458 (1996). 12. C. H. Shih and L. J. Lee, Polym Compos., 19, 626 (1998). 13. S. R Ghiorse, SAM-E Q., 24, 54 (1993). 14. A, A- Goodwin, C. A. Howe, and PL J. Paton, Int. Conf. Compos. Mater., Proc., l Ith, 4, 11 (1997). CHIH-HSIN SHIH, QINGFANG UU, and L. JAMES LEE* Departnent of Chemical Engineering The Ohio State University Columbus, Ohio 43210 *Corresponding author. Society of Plastics Engineers
|
|
 |
|
