Apparent volumetric shrinkage study of RTM6 resin during the curing process and its effect on the residual stresses in a composite

Apparent volumetric shrinkage study of RTM6 resin during the curing process and its effect on the residual stresses in a composite
of 6
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
   Apparent Volumetric Shrinkage Study of RTM6 ResinDuring the Curing Process and Its Effect on theResidual Stresses in a Composite K. Magniez, Arun Vijayan, Niall Finn Materials Science and Engineering Division, Commonwealth Scientific and Research Organisation (CSIRO),Geelong, Victoria, Australia  A comprehensive characterization of the volumetricshrinkage of a commercially important aerospace resin(RTM6) during the various stages of the curing processwas studied. The apparent volumetric shrinkage, eval-uated from density measurements at room tempera-ture, was correlated with the progress of epoxide con-version. During the entire curing process, the apparentvolume shrinkage was found to be less than 3% andoccurred before vitrification. A slight re-expansion ofthe resin, attributed to self-antiplasticization effects,was observed during postcuring at 180 8 C. It was con-cluded that residual stresses were not generated dueto chemical cross-linking during curing but rather fromthermal contraction occurring during the cooling stageafter cure. A photo-elastic method was used to charac-terize residual stresses during cooling in a deliberately engineered resin rich hole of a carbon fiber/RTM6composite. The residual stress was found to reachapproximately 28 MPa, which is in good agreementwith the value calculated from the shrinkage and elas-tic moduli. It is proposed that this simple method canbe provide insights useful to the design and materialsselection processes by measuring and localizing resid-ual stresses from resin during curing and or thermalcycling.  POLYM. ENG. SCI., 52:346–351, 2012.  ª 2011 Society of Plastics Engineers INTRODUCTION During curing of thermoset resins cross-linking reac-tions between the functional groups will result in the for-mation of a complex three-dimensional network. With theprogress of the reaction, a densification of the network andhence a resin volumetric shrinkage is generally observed.Resin shrinkage often causes manufacturing issuesrelated to both dimensional control and poor surface qualityof the finished composite part [1, 2]. More importantly, if the shrinkage occurs in a confined environment, it is directlyconverted to residual stress [3] leading to the formation of stress-induced voids [4], cracks, and delamination [2],which may impact on the mechanical performance of thecomposite. In order to minimize these effects; it is thereforecrucial to understand the cure-related chemical shrinkage.The parameters inducing volumetric shrinkage have beenattributed to both chemical cross-linking during isothermalcuring and thermal contraction effects during cooling [5].An adapted schematic representation from Ochi [5] of thevolumetric changes of epoxy resins during curing (at a tem-perature below the glass transition) and during cooling isgiven in Fig. 1. Determination of residual stresses arisingfrom volumetric shrinkage of epoxy and unsaturated polyes-ter resins has been extensively reported using methods suchas dilatometry [6–8], torsional parallel plate rheometry [9],dynamic mechanical analysis [10], bending beam [11, 12],and changes in buoyancy of the resin [2, 5, 13, 14].RTM6 is a mono-component resin system which hasbeen commercially developed by Hexcel. This resin systemis currently used for Vacuum Infusion and Resin Transfer Moulding (RTM) processes, allowing the fast productionof more cost-effective high-performance aerospace com-posite parts. The primary purpose of the work presented inthis article was to determine the apparent shrinkage (Fig.1) of this commercially important resin during the variousstages of the curing process. The apparent shrinkage mea-surement in this work is equivalent to the one described inthe standard American Society for Testing and Materials(ASTM) method [15]. The evolution of the volumetricshrinkage of the resin, calculated from changes in density,was correlated to the progress of resin conversion. Finally,the magnitude of the stress generated during the coolingstage was measured using photo-elastic response. MATERIALS AND EXPERIMENTAL PROCEDURES  Materials The high performance HexFlow 1 RTM6 (Hexcel, Aus-tralia) is a one-part epoxy/amine resin system that has been K. Magniez is currently at Institute for Technology Research and Inno-vation, Deakin University, Geelong Victoria Australia. Correspondence to : K. Magniez; e-mail: 10.1002/pen.22088Published online in Wiley Online Library ( V V C 2011 Society of Plastics Engineers POLYMER ENGINEERING AND SCIENCE—-2012  specifically developed to fulfill the requirements of theaerospace industry in advanced RTM processes. The com-position of RTM6 resin is based on a tetra functional ep-oxy system and a blend of aromatic amine hardeners [16]and its glass transition temperature ( T  g ) is 196 8 C [17].  Measurement of the Volumetric Shrinkage During Cure Nine sets of RTM6 samples (52 mm  3  27 mm  3  13mm in dimension,    5 g in weight) were cured in a con-ventional oven using silicon moulds. A standard precurecycle [17] is performed at 150 8 C for 180 min, and is fol-lowed by postcure at 180 8 C for 90 min (Fig. 2). In thismethod, six samples were precured at 150 8 C for variousperiods of time: 50, 60, 80, 120, 150, or 180 min (sam-ples 1, 2, 3, 4, 5, and 6, respectively). The remainingthree sets were precured for 180 min at 150 8 C then post-cured at 180 8 C for 40, 60, or 90 min (samples 7, 8, and9, respectively).The change in density was calculated from the changein buoyancy by displacing the RTM6 samples in air andwater at ambient temperature (20 8 C) as previouslyreported by Ochi [5, 13, 14] and Eom et al. [4]. The testprocedure consisted of measuring the sample weight inair and then weighing the sample suspended on a wireand immersed in a water bath using an under-hook bal-ance (accuracy within 0.05%) according to the standardASTM test method [18]. The density was converted toshrinkage by using the following  Eq . 5:Volumetricshrinkage ð % Þ ¼ 100   1  r 1 r    (1)where  r 1  is the density of liquid resin at 20 8 C (1.117 g/ cm 3 [17]) and  r  is the density of the cured sample.Additionally, the residual heat for the pre- and post-cured samples was evaluated using a Perkin-Elmer DSC 7differential scanning calorimeter in dynamic mode. Theinstrument was operated under a nitrogen stream at a flowrate of 50 mL/min and was systematically recalibratedbetween experiments using an indium standard. Samplesweighing between 2 and 7 mg were sealed in vented alu-minium pans. Dynamic curing was achieved by heatingthe sample from 25 8 C to 300 8 C at a heating rate of 10 8 Cper minute. The epoxide conversion was determined usingthe assumption that the heat output correlates directlywith the epoxy/amine reaction. Hence, at time  t  , the frac-tional residual epoxide conversion  a t   was calculated usingthe equation: a t   ¼  D  H  t  D  H  total (2)where  D  H  t   is the specific heat flow (due to residual epox-ide conversion after curing time t), and  D  H  total  is the totalspecific heat flow for the neat resin to achieve 100% cure. D  H  total  was found to be 511 J/g. This value is similar tothat reported by Varley [16]. The fractional residual epox-ide conversion  a t   was converted to a percentage conver-sion at time t using the following equation:%Conversion  ¼ 100 ð 1  a t  Þ  (3)Rheological analysis was performed by Varley [19] ona TA Instruments, ARES strain-controlled rheometer. Iso-thermal cure at 150 8 C was performed using a parallelplate configuration (25 mm top plate, 50 mm bottomplate) with a gap of 1 mm, and at a frequency of 5 rad/s.The strain was initially set to 10% but was continuouslyadjusted throughout the cure using a looping sequencefrom the Orchestrator  1 rheology software to preventoverload of the transducer. Stress Measurements Using Birefringence A sample of carbon fiber/RTM6 composite (similar inshape to the one described in the ‘‘Measurement of theVolumetric Shrinkage During Cure’’ section) of approxi-mate dimensions 50 mm  3  2.4 mm  3  10 mm was pre-pared using three plies of woven carbon fabric into whicha 5.5 mm hole had been punched, followed by an infusion FIG. 2. Autocatalytic curing process of RTM6 resin.FIG. 1. Schematic representation of the volumetric changes of epoxyresin during heating, isothermal curing and cooling. DOI 10.1002/pen POLYMER ENGINEERING AND SCIENCE—-2012 347  with RTM6 resin in a silicone mould (Fig. 3). The size of the punched hole was chosen to allow optical measure-ment of the residual stresses generated by the resin thatfills the hole. The sample was cured and postcured usingthe aforementioned experimental method. The sample wasmounted inside a specially made heating chamber mounted on an Olympus BX50 polarized light microscopeequipped with a full-wave retardation plate ( c  ¼  530 nm)and fitted with a digital camera. Monochromatic (green)linearly polarized light was transmitted through a first po-larizer then through the sample and analyzed with a sec-ond polarizer with its plane of polarization perpendicular to the first. Stress-induced birefringence results in changesin the polarization state of the light passing through eachregion of the sample according to the local stress in thatregion integrated through the depth of the sample alongthe light path. The sample was designed such that thestress would be in the plane and so equal through itsdepth. The in-phase orthogonal polarization componentsof the linearly polarized light pass through the sample atdifferent speeds according to the directions and magni-tudes of the stresses present and the resultant phase shiftresults in elliptical polarization. These localized changesin polarization result in changes in the transmission of thelight through the analyzer so that images of the sampledisplay intensity patterns that depend directly on the stresspatterns in the sample. The polarization direction can berotated several times as the stresses change with time sothat the local intensity oscillates between minima andmaxima giving rise to dark and light fringes representingcontours of equal principal stress differences.The temperature of the sample was measured with athermocouple attached to its side. A hot air gun attachedto the chamber was used to heat the fully cured sampleup to 160 8 C (to reproduce the thermal state of the sampleat the end of the curing process); the heating was stopped,and the sample was allowed to cool naturally while itstemperature was monitored. At every 10 8 C change intemperature, a photograph was taken of the samplethrough the crossed polarizers. As the sample cooleddown, the changing stress was measured by counting theisochromatics (i.e., fringes) passing the centre of the aper-ture using the photographs. RESULTS AND DISCUSSION The percentage epoxide conversion for the RTM6 sam-ples precured at 150 8 C for 50, 60, 80, 120, 150, and 180min and postcured at 180 8 C for 40, 60, and 90 min aregiven in Fig. 4. As expected, the percentage epoxide con-version increased monotonically as a result of cross-link-ing to reach 85% before the postcure process, andincreased from 85% to 95% during the postcure. Thebuild-up of residual stress for an epoxy resin can occur during both the curing and cooling processes, as previ-ously reported by Ochi [5]. During the cure of epoxy res-ins at a temperature below the glass transition ( T  g ), resid-ual stresses will be generated at the vitrification point,and those stresses will increase as a result of coolingshrinkage [14]. The amplitude of the residual stresses af-ter curing will therefore depend on both the evolution of the elastic modulus of the resin after vitrification and thedifference in coefficient of thermal expansion (CTE)between the resin and any constraints; reinforcements andmoulds [5, 13, 14].The evolution of the elastic shear modulus with thecure of RTM6 at 150 8 C is given in Fig. 5. Before gela-tion, the elastic shear modulus was found to be very low FIG. 3. Schematic of the carbon fiber/RTM6 composite sample con-taining a resin rich region (note that the figure is not to scale for visualclarity).FIG. 4. Evolution of the % rate of conversion with curing, showing thegel point  G , and the vitrification point  V  .FIG. 5. Evolution of the elastic shear modulus  G 0 as a function of thecure at 150  8 C, showing the gel point  G , and the vitrification point  V  . 348 POLYMER ENGINEERING AND SCIENCE—-2012 DOI 10.1002/pen  (i.e., \ 1 Pa) as the resin behaved like a Newtonian fluid.After gelation, the resin behaves like an elastic solid, andits shear modulus had increased by several orders of mag-nitude at vitrification (i.e., 10 MPa). Kinetic and rheomet-ric data on isothermally cured RTM6 resin between110 8 C and 180 8 C has previously been reported by Varley[16]. Varley reported gelation time varying with tempera-ture from 358 min at 110 8 C to 18 min at 180 8 C. Vitrifica-tion time varied from 400 min at 110 8 C to 29 min at180 8 C. In the particular case of curing at 150 8 C, the geland vitrification times were found to be 60.5 min and80.6 min, respectively, corresponding to fractional epox-ide conversions of 0.53 and 0.74. These values are con-sistent with those found in this study: 0.5 and 0.72 at cur-ing times of 60 and 80 min, respectively. In the case of RTM6 during isothermal curing at 150 8 C, residualstresses could only be generated at the vitrification timeof 80 min (at 74% epoxide conversion), and before vitrifi-cation, the stresses generated would be negligible.The percentage apparent volume shrinkage as a func-tion of percentage conversion is plotted in Fig. 6. It canbe noticed that during the entire curing process, the per-centage volume shrinkage does not exceed 3%. It can alsobe seen that a slight decrease in the volume shrinkage (aslight re-expansion) was observed during the postcuringprocess at 180 8 C. This decrease can possibly be attributedto self-antiplasticization effects as previously reported byVenditti et al. [20]. These effects arise from changes infree volume where the unreacted end groups attached to anetwork are able to fill the spaces.During cooling, the resin shrinkage is dependent on theCTE of the resin and will be converted to stress if theresin is constrained by a material with a different CTE. Ina carbon fiber reinforced composite, temperature gradientsmay therefore generate residual stresses between the vari-ous components. The linear CTE for fully cured RTM6samples at 25 8 C and 180 8 C are, respectively, 54.5  3 10 2 6 K 2 1 and 62.5  3  10 2 6 K 2 1 [21], which correspondsto a linear shrinkage of 0.91% in cooling from 180 8 Cdown to 25 8 C. The modulus reported for fully curedRTM6 resin is 2890 MPa [17], and therefore, the coolingshrinkage of the resin can be converted into a linear stressnearing 26 MPa. Assuming RTM6 to be isotropic, thevolumetric thermal expansion coefficient  O V  is veryclosely approximated as three times the linear CTE,  O  L ,(see equation below). O V  ¼  1  L 3 q  L 3 q T  ¼  1  L 3 q  L 3 q  L q  L q T    ¼  1  L 3  3  L 2 q  L q T    ¼ 31  L q  L q T  ffi 3 O  L (4)The percentage volumetric shrinkage of RTM6 oncooling is then approximately 2.73%. Carbon fibers arenot isotropic, and their CTE is negative in the longitudi-nal direction ( 2 0.6  3  10 2 6 K 2 1 ). The transverse CTE of carbon fibers is difficult to find in the literature due to thedifficulty of the measurement, but it has been estimated tobe 15  3  10 2 6 K 2 1 [22]. The longitudinal CTE of carbonfibers is almost negligible compared with that of the resin,and therefore, cooling of a carbon fiber/RTM6 compositefrom the cure temperature can be expected to generatesome residual stresses.An estimation of the magnitude of stress generatedduring cooling in a carbon fiber/RTM6 composite wasperformed using its photo-elastic response. The photo-elastic response (isochromatic fringes) in a resin rich FIG. 6. Evolution of the apparent volumetric shrinkage with the extentof epoxide reaction, showing the gel point  G , and the vitrificationpoint  V  .FIG. 7. Schematic of the three-point-bending apparatus fitted to themicroscope setting. A micrometer was used to deform the RTM6 curedsample only in the plane while recording the fringe pattern on the ten-sion side of the beam. DOI 10.1002/pen POLYMER ENGINEERING AND SCIENCE—-2012 349  pocket was recorded to analyze the stress distribution pat-tern. To calibrate the system, a sample of cured RTM6 of approximate dimensions 3.5 mm  3  9 mm  3  100 mmwas mounted in a three-point-bending apparatus fitted tothe microscope (Fig. 7). A micrometer was used todeform the sample only in the plane by known amountswhile recording the fringe pattern on the tension side of the RTM6 beam. Some examples of fringe patterns asstress increases are shown in Fig. 8. It can be seen that asthe deflection increases the number of fringes, and their spatial density increase as the stress at the edge and thestress gradient toward middle of the sample increase. Thedevice was then moved from the microscope to an Instrontensile test device and the deflection was repeated withthe load cell pushing against the centre pin carriage of thethree-point bending device to obtain a load-deflectioncurve for the beam. The peak stress  r  at the edge of thebeam was calculated from simple beam theory [23] andcorrelated with the fringe number   n  obtained from themicroscope to obtain the photo-elastic calibration plot for a pure RTM6 beam (Fig. 9).The gradient of the plot in Fig. 9 provides a photo-elastic constant,  S , used to obtain the stress  r  from thefringe number   n : s ¼ n  St  (5)The measurements for calibration of the photo-elasticproperties of RTM6 also gave a value for the elastic modu-lus of 3240 MPa (in flexure), which was slightly higher thanthe value of 2890 MPa reported elsewhere [17]. As linear shrinkage was measured to be 0.91%, the stress caused bythe differential in CTE between the carbon fibers and theresin would be approximately 29.5 MPa (26.3 MPa if weuse the reported value in [17]). Using the photo-elasticmethod, when the sample cooled from 160 8 C to 20 8 C, 2.7 FIG. 8. Birefringence images with increasing stress in three-point-bending of RTM6 sample. The imageshows an increasing number of dark fringes from 1 to 4, clockwise from top left when the peak stresses atthe edge of the sample are 7 MPa, 14 MPa, 21 MPa, and 28 MPa.FIG. 9. Photo-elastic calibration for a pure RTM6 beam in a three-point-bending test. Stress multiplied by thickness ( t  ) versus Fringe num-ber,  n . 350 POLYMER ENGINEERING AND SCIENCE—-2012 DOI 10.1002/pen
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks