Low-Velocity Impact Response and Ultrasonic NDE of Woven Carbon-Epoxy—Nanoclay Nanocomposites

Low-Velocity Impact Response andUltrasonic NDE of Woven Carbon/Epoxy–Nanoclay NanocompositesM AHESH V.H OSUR,*F ARHAN C HOWDHURY AND S HAIK J EELANICenter for Advanced Materials,Tuskegee University,Tuskegee,AL36088,USAABSTRACT:In the present study,NanomerÕI-28E,organically modifiedmontmorillonite nanoclay supplied by Nanocor Inc.,was used to modify SC-15,atoughened epoxy system using sonication route.Different weight percentage rangingfrom1–3%of nanoclay was used.The modified epoxy was then used to fabricate15-layer plain weave carbon/epoxy composite laminates using a vacuum assistedresin transfer molding(VARTM)method.Samples of size100Â100mm were cutfrom the laminates and were subjected to low-velocity impact loading using aninstrumented drop-weight system(Dynatup Model8210)at three different energylevels of10,20and30J.Transient response of the samples was recorded andanalyzed in terms to load-energy vs time relations.Impact damage was characterizedby utilizing an ultrasonic nondestructive evaluation system(c-scan).Results of thestudy indicate that the infusion of nanoclay in the system reduced the impact damagethough the impact response in terms of the peak load remained mostly unaltered.Reduced damage size was attributed to increased stiffness and resistance to damageprogression of the nanophased laminates.KEY WORDS:nanocomposites,low-velocity impact,ultrasonic NDE.INTRODUCTIONI T IS A general practice to include mineral fillers to polymers in commercial productionfor reasons of cost reduction and stiffness improvement.Most studies dealing with modification of semi-crystalline polymers with inorganic particles report embrittling effects by comparing ultimate elongation and impact strength with those of unfilled resins. Some researchers showed enhancement of toughness in rigid particles filled polyproplylene and polyethelene.In the case of micrometer sized particulates,high filler content,typically higher than20%by volume is generally required to bring about the above stated positive effects of the fillers.This would detrimentally affect some properties of the matrix*Author to whom correspondence should be addressed.E-mail:mhosur@Figures1–9appear in color online:Journal of C OMPOSITE M ATERIALS,Vol.41,No.18/200721950021-9983/07/182195–18$10.00/0DOI:10.1177/0021998307074146ßSAGE Publications2007Los Angeles,London,New Delhi and Singapore2196M.V.H OSUR ET AL. polymers such as processability,appearance,density and aging performance.There is, therefore,a need for new materials where lower particle concentration is desired.With this regard,newly developed nanocomposites would be competitive materials.Nanoscale materials such as nanocomposites provide the opportunity to explore new behavior and functionality beyond those found in conventional materials.Polymeric nanocomposites are materials that are designed and processed from selected constituents.The extremely high surface area is one of the most attractive characteristics of the nanoparticles because it facilitates creating a great interface in a composite.An interphase of1nm thick represents roughly0.3%of the total volume of polymer in the case of micro particle filled composites,whereas it can reach30%of the total volume in the case of nanocomposites[1].Contribution made by the interphase modified by the low nanofiller loading provides possibilities of enhanced performance by influencing the properties of the matrices.Chun et al.reported significant improvement in the tensile performance of polypropylene composites in terms of stiffening,strengthening and toughening with a low-filled content of about0.5%[2].Clay and inorganic reinforcements have been shown to be effective reinforcements in neat polymeric structures[3–8],but very little work has been done to examine advanced fiber-reinforcement composites that utilize nano-composite matrices.Hussain et al.[9]incorporated nano scale Al2O3particles in filament wound carbon fiber/epoxy composites.He observed an increase in modulus,flexural strength, interlaminar shear strength,and fracture toughness when the matrix was filled with10% by volume of alumina particles of25nm diameter.Seferis et al.[10]have shown the ability to incorporate nanosized alumina structures in the matrix and interlayer regions of prepreg-based carbon/epoxy composites.Timmerman et al.[11]studied the influence of nanoclay on the carbon fiber/matrix composites under thermal cyclic loading and reported that the transverse cracking in symmetric carbon fiber/epoxy laminates was significantly reduced when nanoparticle fillers were used.Gopakumar et al.[12]studied the influence of clay exfoliation on the physical properties of montomorillonite/ployethelene composites. They reported that nanoscale dimensions of the dispersed clay particles in the nanocomposites led to significant improvement in viscous and elastic properties and stiffness,which was attributed to the enhanced phase adhesion.In order to integrate the nanoscale materials as part of larger material components,it is essential to determine the influence of the nano phased particles on the properties of the larger material systems. It has been demonstrated that effective dispersion of anisotropic nanoparticles can substantially improve properties of the polymeric matrices.This dispersion of nanoparticles highly depends on processing techniques such as solution blending,shear mixing,in situ polymerization,ultrasonic cavitation,high pressure mixing[13–19]. Nanoparticles,like nanoclays,used in carbon fiber-reinforced polymeric composites,at very low concentrations seem to improve mechanical and thermo mechanical properties of the composites[20].However,impact resistance of such polymer composites is probably one of the most important and least understood phenomena.Very few studies on the effect of nanoclay platelets in polymer composites have been carried out.Low velocity impact testing of composites is a very crucial area for investigation.A major concern that limits the usage of composites is their susceptibility to damage due to impact loading.There are practical situations like tool drops,runway debris,bird strikes,hailstorms,and ballistic loading,which induce considerable damage to the composite posites are inherently weak in transverse direction,i.e.,the stiffness and the strength in through-the-thickness direction are poor since no fibers are present in that direction.Low-velocity impact is considered potentially dangerous mainly because the damage might be leftWoven Carbon/Epoxy–Nanoclay Nanocomposites2197 undetected,as the surface may appear to be undamaged.Understanding the causes for the formation of such damages and improving the damage resistance characteristics of composites are very important.When subjected to impact loading,the energy is absorbed in the form of creation of new surfaces.The failure mechanisms include indentation, matrix cracking,delamination,ply splitting,and fiber fracture[21–24].This will considerably reduce the residual mechanical properties of the laminate.The worst scenario occurs when the damage is at subsurface levels.It is known that the residual compressive strength,which is the most affected mechanical property,is reduced to almost up to50%[25–28].In many situations,the level of impact at which visible damage is formed is much higher than the level at which substantial loss of residual properties occurs. Even when no visible impact damage is observed at the surface(energies below barely visible impact damage,BVID),matrix cracking and interlaminar failure can occur,and the load carrying capacity of the composite laminates is considerably reduced.Visible damage occurs if an impact is above a threshold impact energy,which depends on the laminate stiffness.Though there have been a number of studies on the impact response of composites,to the best of the authors’knowledge,there are no studies reported in the open literature on the effects of nanoclay on the impact response of woven composites.Hence,in the present study,low velocity impact response of carbon/nanoclay-epoxy composites was investi-gated.In addition to that,the damage area of these composite laminates was also studied through ultrasonic c-scan images to observe the effect of different weight percentages of nanoclay platelets on the polymer composites.EXPERIMENTALResin PreparationUltrasonic cavitation technique is one of the most efficient means to disperse nanoparticles into a polymer[29].Sonics Vibra Cell ultrasonic processor(Ti-horn, 20kHz,100W/cm2)was used to obtain a homogeneous mixture of epoxy resin and Nanocor NanomerÕI-28E nanoclay,a surface modified montmorillonite clay.SC-15 epoxy,manufactured by Applied Poleramic,Inc.,comes with two parts:part A and part B. At first,the part A of SC-15epoxy(mixture of diglycidylether of bisphenol A,60–70%, aliphatic diglycidylether,10–20%)was sonicated with nanoclay at an amplitude of55% with a total mixing time of30min.The distance of one movement up and down of the ultrasonic probe is called its amplitude.The amplitude can be set at a certain percentage of maximum amplitude of the probe.In the current study,a12.7mm diameter high intensity probe which has maximum amplitude of124m m was used.Amplitude setting at55% corresponds to amplitude of68.2m m.Nanoclay pick up moisture from air very easily.To get rid of the moisture,nanoclay was heated to100 C for2h before sonication.A pulser cycle(turning on and off-time ratio of2:1)was used to control the temperature of the mixture.It was recommended to keep the temperature of the mixture at around40–50 C for good mixing,which was monitored using an infrared thermometer.The pulser cycle helps to achieve this goal.In addition,a water bath was also used.After sonication,the homogeneous mixture was cooled down to room temperature in a refrigerating cooler maintained at5 C as shown in Figure1(b).The sonicated part A with nanoclay was then mixed with part B of SC-15epoxy(epoxy hardener,cycloaliphaticamine 70–90%and polyoxylalkylamine 10–30%)at a ratio of 10:3by weight using a high speed mechanical stirrer for about 5min.Figure 1(c)illustrates the mechanical mixing of part A and part B.The mechanical mixing introduces air bubbles into the resin.This warranted a need of removing the trapped air and reaction volatiles from the reaction mixture.This was done using high vacuum for about 30min.The resin mixture was kept in a sealed desiccating chamber,as can be seen in Figure 1(d),and connected to a vacuum suction pump.The high vacuum pulls out the air from the mixture and the resin is degasified at the lower layer.Two distinct layers are evident in Figure 1(d).Nanocomposite FabricationCarbon/nanoclay-epoxy composites were manufactured by both VARTM processes.A 15-layer plain weave carbon fabric was used to fabricate the composite laminates.The fabric had T-300carbon fibers with 3k fiber tows in both warp and fill directions.Nominal thickness of the laminates was 3.3mm.Arrangement of the fabrication process is detailed schematically in Figure 2.The VARTM process uses vacuum pressure to remove air from the fabric lay-up before and while the matrix resin is introduced to the fabric reinforcement.The pressure difference between the atmosphere and the vacuum is the driving force for infusion of the resin into thelay-up.(a)(d)(b)(c)Figure 1.(a)Vibra-cell ultrasonic processor.(b)Cooling of sonicated Part A.(c)Mechanical mixing of part A and part B of SC-15epoxy .(d)Degasification of resin mixture.2198M.V.H OSUR ET AL .Woven Carbon/Epoxy–Nanoclay Nanocomposites2199Suction lineVacuum baggingFigure2.VARTM layup sequence.For fabricating the laminate,freekote(mold releasing agent)was sprayed on the mold. The required number of layers was carefully placed on the mold.Then a sealant tape was tacked on the surface of the mold about25–50mm from the perimeter of the fabric layers. Resin supply tubes were connected to the system with the mold end of the tube connected to a spiral wrap along with a distribution mesh that lies on the top and bottom of the preform.This facilitates easy flow of resin over the top and through the thickness of the laminate when vacuum is applied.Tubes linking the vacuum pump and the spiral wrap were also connected.Resin traps were placed between the vacuum pump and the mold to collect the excess resin.Finally,a vacuum bag was placed on the mold pressed firmly against the sealant tape to provide a vacuum tight system.The preform was left to debulk under vacuum.After debulking,the nanophased SC-15resin system was infused impregnating the fabric as the resin flow advanced towards the vacuum side.The resin inlet valve was closed when resin reaches the suction side,and the infused laminate was left to cure at room temperature.Vacuum was maintained until the end of cure to remove any volatiles generated during the polymerization,in addition to maintaining the pressure of one atmosphere.The room temperature cured material was taken out from the vacuum bagging and trimmed.Test samples were machined for low velocity impact characterization.They were thermally post cured at100 C for5h in a mechanical convection oven.Low Velocity Impact TestingAll the impact tests in this study were conducted using an impact drop tower device DYNATUP Model8210manufactured by GRC Instruments.DYNATUP,as shown in Figure3,is equipped with as Impulse data acquisition system,version3.Impulse v.3can acquire8192data ing this machine,impact energy and velocity can be varied by changing the mass and height of the dropping weight.The Specimen is held with clamped edge conditions in the fixture placed at the bottom of the drop tower,which provides aclamped circular support span of 75mm in diameter.The weight of crosshead was maintained at 6.62kg and it was guided through two smooth guide columns.The impactor end is fitted with an instrumented tup of 15.56kN capacity that records the transient response of the specimens.In all 8192data points are collected during the impact event.Samples were impacted with 12.5mm diameter instrumented tup having an hemispherical end.Transient response of the samples includes velocity,deflection,load and energy as a function of time.For each type of laminates,at least three samples were subjected to impact at 10,20,and 30J.The machine is also fitted with a velocity detector that measures the velocity of the tup just before it strikes the specimen.It is also fitted with pneumatic rebound brakes,which prevents multiple impacts on the specimen.Data acquisition system records load vs.time data for each test.Along with this data and the velocity at impact data,the software calculates deflection,specimen velocity and energy absorbed by the specimen.The data is analyzed in terms of peak load and absorbed energy.The absorbed energy is calculated as the difference of total energy (at the end of the event)and the energy at peak load.The impact energy is,in general,mainly absorbed in the form of elastic deformation,plastic deformation and through various damage modes.As composite materials have no plastic deformation,all the energy is absorbed through elastic deformation and through different failure modes.Hence,in the current study,absorbed energy is attributed to the energy spent in creating damage.The 100mm Â100mm composite samples were placed between the pneumatic clamps.Heights were adjusted depending on the desired energy level.To calculate the height of the indenter from which it should be dropped to achieve 10,20,and 30J of energy,thePneumatic clamphTup, m 1Insert, m 2Main massFigure 3.Experimental set-up for low-velocity impact test.2200M.V.H OSUR ET AL .Woven Carbon/Epoxy–Nanoclay Nanocomposites2201 equation E¼mgh is used,where E is the impact energy,m is the mass of the impactor,g is the acceleration due to gravity,and h is the drop height.The height is the distance between the tip of the indenter and the top surface of the sample held between the pneumatic clamps.The two known parameters are the mass which is6.62kg and the acceleration due to gravity which is9.81m/s2.Now the values for energies are put separately as10,20,and 30J to get their respective heights.Once the height required to attain a particular energy level is known,the indenter is moved accordingly to that height before it is dropped on the specimen for the test.Ultrasonic C-ScanUltrasonic inspection of the laminates was conducted using an ultrasonic pulse-receiver unit by Sonix Inc.with FlexSCAN-C TM software.The scanning was done in pulse-echo immersion mode using1MHz50mm-focus transducer.The1MHz sensor had a piezoelectric element size of19mm.Scanning was done with the impacted surface facing the sensor to obtain the projected damage.Gate was set on the back surface echo.All the laminates were subjected to ultrasonic nondestructive evaluation both before and after impact testing.The ultrasonic testing before impact loading was carried out to ensure that there was no fabrication defect in the sample.Post impact ultrasonic testing was conducted to evaluate the extent of damage in the sample.From the c-scan images,the damage area as projected onto a plane was measured.RESULTS AND DISCUSSIONLow Velocity Impact CharacterizationThe specimens were tested under three energy levels namely10,20,and30J.Figure4 shows the load and energy vs.time response of control samples,1%,2%,and3% nanoclay samples at these three energy levels.Table1gives the average values of the peak load and absorbed energy of the three types of samples impacted at the three energy levels. It was observed that the average peak load,as shown in Table1,at which the control samples failed,was 1.40kN for10J, 1.42kN for20J,and 1.45kN for30J.The corresponding absorbed energy was4.97J,15.84,J and13.18J.As expected there was an increase in the peak load as the energy levels were increased.A similar trend was also seen in the case of different weight percentages of nanoclay samples.The average peak load at which the1%nanoclay samples failed,was1.42kN for10J,1.45kN for20J,and1.50kN for30J whereas the average absorbed energy was5.36J,12.88J and8.64J.The average peak load at which the2%nanoclay samples failed,was1.35kN for10J,1.43kN for20J, and1.46kN for30J and the corresponding average absorbed energy was5.34,15.42,and 9.45J.The average peak load,as shown in Figure4(d),at which the3%nanoclay samples failed,was1.42kN for10J,1.52kN for20J,and1.41kN for30J and the corresponding average absorbed energy was5.98J,14.94J and8.55J.In most of the cases the load vs.time plot showed a smooth rise up to the maximum load indicating an elastic response up to that point.Whenever there was some oscillations until the load reaches the peak value indicates that there is perforation of the top surface followed by the oscillatory drop in load implying the progressive local failure of thelaminate [30].The slope of the load–time curve,which is designated as the contact stiffness,increases with the increasing amount of energy.The initial knee found in the load–time–energy plot is due to the inertia effect of the tup and the sample.Once the inertia of the tup and samples matched,a smooth load rise is seen.Figure 5(a)–(c)shows a comparison of load–time response for control,1%,2%,and 3%nanoclay samples at 10,20,and 30J energy levels respectively.The slope of the load–time curve,which is designated as the contact stiffness,remains the same.The peak load has an increased value for 1%nanoclay sample as compared to control and other percentages of nanoclay samples.It is observed that the peak load,as shown in00.51.01.5246810051015202530 J 20 J 10 J30 J 20 J 10 J30 J 20 J 10 J30 J 20 J 10 JTime (msec)246810Time (msec)246810Time (msec)0246810Time (msec)L o a d (k N )E n e r g y (J )05101520E n e r g y (J )05101520E n e r g y (J )05101520E n e r g y (J )(a)00.51.01.5L o a d (k N )(c)0.51.01.5L o a d (k N )(b)00.51.01.5L o a d (k N )(d)Figure 4.Load and energy vs time response of VARTM:(a)control samples.(b)1%nanoclay samples.(c)2%nanoclay samples,and (d)3%nanoclay samples at three different energy levels.Table 1.Average peak load and absorbed energy.Damage area,mm 2Control sample1%nanoclay2%nanoclay3%nanoclay10JPeak load (kN)1.41 1.42 1.35 1.42Absorbed energy (J) 4.97 5.36 5.34 5.9820JPeak load (kN)1.42 1.45 1.43 1.52Absorbed energy (J)15.8412.8815.4214.9430JPeak load (kN)1.45 1.50 1.46 1.41Absorbed energy (J)13.188.649.458.552202M.V.H OSUR ET AL .2.(a )1.1.0.L o a d (k N )2.(c )1.1.0.L o a d (k N )2.(b )1.1.0.L o a d (k N )T i m e (m s e c )T i m e (m s e c )T i m e (m s e c )F i g u r e 5.L o a d –t i m e r e s p o n s e o f c o n t r o l ,1%,2%,a n d 3%n a n o c l a y s a m p l e s a t 10,20,a n d 30J e n e r g y l e v e l s r e s p e c t i v e l y .Woven Carbon/Epoxy–Nanoclay Nanocomposites22032204M.V.H OSUR ET AL. Figure5(a),at which the samples failed,was1.35,1.44,1.39,and1.36kN for control,1%, 2%and3%nanoclay samples,respectively.As can be seen in Figure5(b),the peak load at which the samples failed was1.40,1.49,1.46and1.43kN for control,1%,2%,and3% nanoclay samples,respectively.Also,at30J impact energy,the peak load,as shown in Figure5(c),was1.37,1.54,1.48,and1.42for control,1%,2%and3%nanoclay samples, respectively.It is clearly seen from here that1%nanoclay samples had the highest peak load than control and other nanoclay samples.But there was no significant effect of nanoclay seen in all three energy levels.Quite a few oscillations were observed untill the load reaches the peak value.The loading and unloading portion of the load–time curves is not smooth and symmetric which indicates the presence of progressive damage.In the case of composite laminates subjected to impact loading at different energy levels, the peak load generally increases with impact energy until a threshold is reached beyond which it either remains constant or will reduces(when there is severe damage)with increasing energy.For this to happen,the requirements is to start at a very low energy and gradually increase the impact energy level.In the current study,impact energy was chosen to be10,20,and30J.In the control samples,the first indication of damage was seen at 10J.With the increase in the energy levels,however,the peak load did not increase much indicating that the peak load was close to the threshold level when the samples were tested at10J.As was seen in the study,damage level increased at20and30J and hence the peak load did not increase.A close look at Figure4reveals that the shape of the load–time curves is different for30J as compared to10and20J for all the samples.This is related to the damage in the samples.The composite laminates are brittle in nature and respond elastically until they reach the peak load.If the impact energy is higher than the energy absorbed until the peak load,the additional energy is taken up in the creation of damage with a small amount of energy lost in friction between the sample and the impactor.After the peak load is reached,there is a drop in the load indicating the creation of damage.If there was no damage in the sample,the load–time response will be symmetric about the peak load.The drop in the peak load is not too sharp indicating qualitatively that the damage created is not significantly high so as to reduce the stiffness of the laminate. This is characteristic of woven fabric composites unlike unidirectional laminates. During impact loading,the initial damage is at the contact zone.However,that damage is highly localized and does not affect the residual properties to a large extent. As the impact energy is increased,the laminate undergoes large deformation.The next failure that takes place will be the tensile failure of the back surface due to flexure. In the case of unidirectional laminates as the fibers are provided in only one direction, the ply will split along the fibers.This leads to the creation of large delaminations at the interfaces immediately preceding the last layer.However,in the case of woven fabric composites,the plies will have fibers in both directions.In addition,they are woven together.Even if a fiber tow fails in tension,the damage is localized and will not propagate in the plane of the ply as the damage front is resisted by not only the fibers in the transverse direction but also by the interlacing.Hence,the sharp drop is not observed in the case of woven composites.In the current study,there was a barely visible dent at the surface of the samples when impacted at10J.However,at20J the dent is quite visible at the impact location with fiber fracture at the front(impact)surface as well as fiber fracture with prominent cracks on the back surface.However,at30J,there was clear penetration through the entire thickness for all samples.As long as there is no penetration,the impact duration will be long.Hence,itis seen in the load–time plots of samples impacted at10and20J,the duration was longer than for those impacted at30J.Further,in the event of an impact,energy is absorbed by a material in the form of elastic,plastic deformations and through failure modes.As indicated earlier,in the case of composites,there is hardly any plastic deformation.Hence the energy absorption is predominantly through the creation of various damage modes.As can be seen from Table1,the absorbed energy for the control samples at10J was less when compared to all the nanophased laminates which implies that nanophased laminates were able to absorb more energy elastically at all energy levels.At higher energy levels,the absorbed energy was higher in the case of the control samples which would mean than the damaged area was also higher as will be demonstrated in the next section.At30J,there was clear penetration of the samples.When this happens,the entire kinetic energy of the tup is not transferred to the sample.As a consequence,the absorbed energy will be less.The absorbed energy for all the nanophased samples at30J was less compared to control samples.This would mean that the control samples had higher damage than the nanophased samples.It can be inferred that the inclusion of nanoclay into the samples resisted the propagation of cracks and contained the growth of damage.Ultrasonic NDENanoclay seems to have a marginal effect on peak load and initial slope of the load–time graph but the effect is readily seen when the damage areas are investigated.Ultrasonic c-scan studies were used to find out the extent of damage for all the samples.In the ultrasonic testing,ultrasound traverses through the water media in the case of immersion technique and gets reflected back at the front surface of the sample.Part of the ultrasonic energy gets transmitted into the sample.If there is no defect in the path of the ultrasound, it traverses through the thickness of the material while suffering an attenuation due to the absorption of ultrasound by the material and then gets reflected back again at the back surface of the sample.Signals of ultrasound that are reflected off the front and back surfaces of the sample are called front wall and back wall echoes.If a defect or damage is present in the sample,then it acts as a reflector of the ultrasound and a defect echo is generated.Depending on the nature of defect or damage,there may be a presence or a total absence of back surface echo.In the current study,back surface signal was continuously monitored and collected over an area of50Â50mm at an internal of0.25mm.The amplitude of this signal is stored in terms of decibels(dB)and plotted as a c-scan image.Presence of damage due to impact in the sample reduces the back wall signal to almost zero.In the current study,the term damage is used to include the cumulative effects of all damage modes which include matrix cracking,fiber fracture as well as penetration.Some representative ultrasonic c-scan images of control and1%nanoclay samples with the photographs of front and back surface of the laminates impacted at10,20,and30J, respectively,are presented in Figures6(a)–(c)through Figures9(a)–(c)respectively. Impact testing was not carried out beyond30J as there was total penetration of the sample at30J.The horizontal and vertical stripes seen in the ultrasonic c-scans are impressions left by the distribution meshes in VARTM manufacturing creating waviness in the surface.This waviness is generally produced in VARTM manufactured composites which is captured by the c-scans.In Figure6,the samples,tested at10J,did not show any。