at SciVerse ScienceDirect

International Journal of Impact Engineering 57 (2013) 108e118

Contents lists available

International Journal of Impact Engineering

journal homepage: www.elsevier .com/locate/ i j impeng

Effects of amino-functionalized MWCNTs on ballistic impact performanceof E-glass/epoxy composites using a spherical projectile

Muhammad Rahman a, Mahesh Hosur b,*, Shaik Zainuddin b, Uday Vaidya c, Arefin Tauhid a,Ashok Kumar d, Jonathan Trovillion d, Shaik Jeelani a,b

aDepartment of Mechanical Engineering, Tuskegee University, Tuskegee, AL 36088, USAbDepartment of Materials Science and Engineering, Tuskegee University, Tuskegee, AL 36088, USAcDepartment of Materials Science and Engineering, University of Alabama at Birmingham, Birmingham, AL 35294, USAdU. S. Army Construction Engineering Research Laboratory, Champaign, IL 61821-9005, USA

a r t i c l e i n f o

Article history:Received 17 May 2012Received in revised form23 December 2012Accepted 12 January 2013Available online 15 February 2013

Keywords:Ballistic impactCarbon nanotubesCalendaring methodDamage area

* Corresponding author. Tel.: þ1 334 724 4220.E-mail addresses: hosur@mytu.tuskegee.edu, mho

0734-743X/$ e see front matter � 2013 Elsevier Ltd.http://dx.doi.org/10.1016/j.ijimpeng.2013.01.011

a b s t r a c t

Effect of adding amino-functionalized multi-walled carbon nanotubes (NH2-MWCNTs) on the ballisticperformance of E-glass/epoxy composites was investigated in this study. E-glass/epoxy panels wereprepared with and without MWCNTs. Two different weight percentages of 0.3 and 0.5% MWCNTs wereused to modify the epoxy resin. MWCNTs were dispersed in epoxy resin through a combination ofsonication and 3-roll mill methods. Laminated composite panels with E-glass plain weave fabricswere made through hand lay-up followed by compression molding process. Samples of size120 � 120 � 5.25 mmwere then cut from the panels and subjected to spherical projectile impacts using agas-gun set-up at different velocities to determine the ballistic limit velocity (VBL). Additionally, ballisticlimit velocity was statistically determined through polynomial and power regression using experimentalresults. Ballistic limit increased by about 15e19 m/s (5e6.5%) for laminates with 0.3 wt.% MWCNTs. Onthe other hand, the performance of laminates with 0.5 wt.% MWCNTs was comparable with that ofcontrol samples. Damage size as determined by ultrasonic c-scan studies was considerably less for thelaminates with MWCNTs at a given impact velocity below and above ballistic limit.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

Fiber reinforced polymer (FRP) matrix composites are increas-ingly used as alternatives to conventional metallic materials invarious armored structures of military vehicles due to their lightweight, high stiffness-to-weight and strength-to-weight. FRP com-posites are preferred over conventional materials in the battlefieldfor light weight feature as this material class increases the mobilityof vehicles to avoid enemy threats. Besides, lower load imposed onthe engine system increases fuel efficiency of the vehicles. However,impact resistance is one of the major concerns for fiber reinforcedcomposite due to its high susceptibility to impact damage.

Glass fiber has been attractive material as reinforcement incomposites due to low cost, high impact strength, high chemicalresistance and excellent insulating properties. Because of goodstiffness, strength, excellent chemical resistance, dimensional sta-bility, and excellent adhesionwith fillers and fibers, epoxy has been

sur@gmail.com (M. Hosur).

All rights reserved.

excellent candidate in glass fiber reinforced composites amongpolymer matrix. Typically, glass fiber reinforced polymer (GFRP)composites are considered as high-ductility advanced compositeswhich are capable of absorbing maximum projectile’s kinetic en-ergy, but have limited ability against deforming or fracturing due toimpact by projectile [1]. Carbon nanotubes (CNTs) have high po-tential to combine the benefits of higher absorption capability aswell as high hardness. They exhibit an exceptionally high aspectratio in combination with low density, high strength, stiffness andhardness, which makes them potential candidate for the rein-forcement of polymer-based composite materials. As a result, car-bon nanotubes reinforced polymer matrix composites might behighly effective as armor structures while high strength combinedwith high ductility might absorb more kinetic energy of the pro-jectile without much deformation. High surface area of CNTs pro-vides desirable interfaces for stress transfer effectively in compositematerial. Ajayan et al. investigated stress transfer mechanism inCNT/epoxy composites [2]. They found multi-walled carbon nano-tubes comparatively more efficient during stress transfer frommatrix to CNTs than single-walled nanotubes. On the contrary,another study found that this high surface area induces undesirable

Table 1Fiber volume fraction of control and NH2-MWCNTs incorporated E-glass/epoxycomposites.

NH2-MWCNTs weight percent content 0.0 wt.% 0.3 wt.% 0.5 wt.%

Fiber volume fraction, Vf 57.54 � 1.3 58.18 � 2.7 58.59 � 2.2Void fraction, Vd 3.94 4.28 5.89

M. Rahman et al. / International Journal of Impact Engineering 57 (2013) 108e118 109

strong attractive forces between CNTs which leads to excessiveagglomeration [3]. Various dispersion methods to de-agglomeratenanotubes in polymer resins, such as stirring, sonication and highshear mixing have been reported in literature [4e6]. Sonication andcalendaringmethods are reported to be highly effectivewith regardto achieving dispersion.

In addition, interfacial adhesion between the CNTs and polymer isalso a critical issue. In order to have sufficient stress transfer from thematrix to the CNTs and to efficiently use the potential of CNTs asstructural reinforcement, a strong interfacial adhesion between theCNTs and polymer is desired. The interfacial adhesion between CNTsand matrix was reported to improve by functionalizing the CNTs.Tailored amino, carboxyl or glycidyl groups enable covalent bondingbetween CNTs and epoxy resulting in improved interfacial bonding.The positive effects of functionalized CNTs on the mechanicalproperties are reported by various researchers [7e9]. In our recentstudy, a combined dispersion of 0.3 wt.% NH2-MWCNTs by sonicat-ion and calendaring method lead to an optimum improvement interms of flexure and thermo-mechanical properties [5].

In case of high velocity impact, FRP composites hinder the pro-jectile by absorbing kinetic energy following a variety of energyabsorption mechanisms through the formation of new surfaces inthe surroundings of the impacted area. The main energy absorbingmechanisms are: cone formation on the back face of composites,tensile failure of primary yarns, elastic deformation of secondaryyarns, matrix cracking and delamination, shear plugging of theprojectile into target and frictional energy during penetration [10].So far, most of the studies were about determining how the failuremechanism and penetration phenomenon occur through theoreticaland numerical predictions [11e14]. Till now, very few studieshave been conducted on the control of damage behavior andmechanism of high velocity impact properties of FRP composites

Fig. 1. Schematic of fabrication process

using nanoparticles. Javad et al. carried out high velocity impact testof nanoclay reinforced unsaturated polyester (UP) resin and betterperformance was found for samples containing 1.5 wt.% nanoclay[15]. Avila et al. focused on high velocity impact of two hybridnanocomposites i.e. glass fiber/epoxy/nanoclay and glass fiber/epoxy/nanographite [16]. The addition of nanoclay and nanographiteto glass fiber/epoxy laminates increased the impact resistance. Be-sides, they found a considerable effect of using nanoparticles on thefailure mechanisms of glass fiber/epoxy composites.

To the best of the authors’ knowledge, there are no studies re-ported in the open literature on the effect of NH2-MWCNTs on theballistic response of glass fiber reinforced epoxy composites. Hence,in this study, the effect of amino-functionalized MWCNTs on theballistic impact properties of E-glass/epoxy composite laminates wasinvestigated. Composite samples for ballistic test were fabricatedusing hot press process. Samples of size 120 � 120 � 5.25 mmweresubjected to ballistic impact using a gas-gun set-up with sphericalprojectiles at seven different pressure levels. Dispersion state of CNTsin the epoxy matrix systems was investigated by transmissionelectron microscope (TEM). In addition, effects of interaction ofMWCNTs with epoxy and fiber/matrix bonding were investigatedusing scanning electron microscope (SEM). Projected damage areaafter impact test was evaluated by ultrasonic c-scan. All results werecompared with the control composites containing no MWCNTs.

of E-glass/epoxy nanocomposites.

Table 2Ballistic impact results of control and NH2-MWCNTs incorporated E-glass/epoxy composites.

Sample Specimen no. Air pressure (MPa) Bullet weight (kg) Incident velocity (m/s) Residual velocity (m/s) Remarks

Reference GFRP 01 1.38 0.0021 367.38 189.63 Penetration02 0.83 0.0021 332.01 115.85 Penetration03 0.69 0.0021 324.08 76.52 Penetration04 0.62 0.0021 309.75 71.03 Penetration05 0.55 0.0021 299.08 9.45 Penetration06 0.48 0.0021 290.85 0.00 Embedded07 0.34 0.0021 264.32 0.00 Embedded

0.3 wt.% CNT-doped GFRP 01 1.38 0.0021 369.82 185.97 Penetration02 0.83 0.0021 335.06 89.02 Penetration03 0.69 0.0021 319.51 18.90 Penetration04 0.62 0.0021 306.70 5.18 Penetration05 0.55 0.0021 301.82 0.00 Embedded06 0.48 0.0021 291.16 0.00 Embedded07 0.34 0.0021 269.51 0.00 Embedded

0.5 wt.% CNT-doped GFRP 01 1.38 0.0021 369.82 201.83 Penetration02 0.83 0.0021 331.70 117.98 Penetration03 0.69 0.0021 329.26 108.84 Penetration04 0.62 0.0021 319.82 74.39 Penetration05 0.55 0.0021 300.91 0.00 Embedded06 0.48 0.0021 283.23 0.00 Embedded07 0.34 0.0021 248.17 0.00 Embedded

M. Rahman et al. / International Journal of Impact Engineering 57 (2013) 108e118110

2. Experimental

2.1. Materials

SC-15 epoxy resin, a two-part epoxy resin (Part A: diglycidy-lether of bisphenol A, aliphatic diglycidyl and Part B: cyclo-aliphaticamine hardener), was used in this study. E-glass woven fabric witha density of 2.58 g/cm3 and a single fiber diameter of 14e16 mmwasused. E-glass fibers were sized using epoxy silanes of max. 0.4% byweight to increase the compatibility and to have better adhesionbetween fibers and epoxy matrix. Multi-walled carbon nanotubesfunctionalized using amino groups (eNH2) were purchased fromNanocyl Inc., Belgium. These nanotubes were of diameter 10 nm,average length of several microns and carbon purity >95%. Highspecific surface area and cotton-like entanglements cause the for-mation of agglomerates among nanotubes as reported by Reynaudet al. [17].

Table 3Calculated initial kinetic energy, residual kinetic energy, absorbed energy and projected

Sample Specimen no. Initial kineticenergy, Uk (J)

Residual kineenergy, Ur (J)

Reference GFRP 01 141.71 37.7602 115.74 14.0903 110.28 6.1404 100.75 5.2905 93.92 0.0906 88.83 0.0007 73.36 0.00

0.3 wt.% CNT-doped GFRP 01 143.60 36.3202 117.88 8.3203 107.19 0.3604 98.77 0.0305 95.66 0.0006 89.01 0.0007 76.27 0.00

0.5 wt.% CNT-doped GFRP 01 143.60 42.7702 115.53 14.6203 107.40 5.8104 113.84 12.4405 95.08 0.0006 84.23 0.0007 64.66 0.00

2.2. Manufacturing process

2.2.1. Dispersion and mixing of NH2-MWCNTs into epoxy resinAt first, NH2-MWCNTs were mixed manually with Part A of

epoxy resin as per calculated weight ratio. The mixture was thensonicated at room temperature for 1 h at 35% amplitude and 20-son/off cycle pulse mode. To further improve the dispersion ofMWCNTs, the sonicated mixture was then passed through a three-roll mill. In three-roll milling process, rollers 1 and 3 rotate in thesame direction whereas the roller 2 placed in between rotates inthe opposite direction thereby inducing high shearing in themixture. The three rollers rotate at different speed. This is facili-tated by a gear assembly which maintains a ratio of 1:3 betweensuccess rollers. Roller speed of the three rolls was maintained at aratio of 1:3:9 with a maximum speed of 180 rpmwasmaintained inall the three passes. Differential speed of and contra-rotation be-tween successive rollers facilitates high shear which helps in

damage area of control and NH2-MWCNTs incorporated E-glass/epoxy composites.

tic Absorbed energy,Uk � Ur (J)

Projected damagearea (cm2)

% difference in damagearea w.r.t. reference

103.95 40.32 e

101.65 44.35 e

104.13 48.38 e

95.44 49.16 e

93.83 56.45 e

88.83 56.45 e

73.36 49.39 e

107.28 32.26 �19.99109.56 40.32 �9.07106.83 40.32 �16.6798.74 44.35 �9.7895.66 36.29 �35.7189.01 32.25 �42.8776.27 48.38 �2.04

100.83 36.29 �9.99100.91 40.32 �9.07101.59 44.35 �8.33101.40 48.38 �1.5995.08 48.38 �14.3084.23 40.32 �28.5864.66 48.38 �2.04

M. Rahman et al. / International Journal of Impact Engineering 57 (2013) 108e118 111

breaking down the bundles of MWCNTs in the resin mixture.Further, a varying gap setting between the rollers and multiplepasses of 20 mm (1st pass), 10 mm (2nd pass) and 5 mm (3rd pass)were used to induce high shear force in the mixture. The inductionof high shear forces further improves the dispersion of CNTs inresin. The hardener, Part B was added as per stoichometric ratio(Part A:Part B ¼ 10:3) to the modified mixture and mixed with ahigh speed mechanical stirrer for 10 min at 800 rpm. The mixture

60 70 80 90 100 110 120 130 140 15060

70

80

90

100

110

120

130

140

150

Abso

rbed

Ene

rgy

(J)

Impact Energy (J)

60 70 80 90 100 110 120 130 140 15060

70

80

90

100

110

120

130

140

150

Abso

rbed

Ene

rgy

(J)

Impact Energy (J)

60 70 80 90 100 110 120 130 140 15060

70

80

90

100

110

120

130

140

150

Abso

rbed

Ene

rgy

(J)

Impact Energy (J)

(a) Control

(b) 0.3 wt. CNT

(c) 0.5 wt. CNT

Fig. 2. Ballistic limit indicated by red arrow for (a) control laminates, (b) 0.3 wt.% NH2-MWCpower regression. (For interpretation of the references to colour in this figure legend, the r

was then placed in a vacuum oven at room temperature for 30 minfor degasification to remove entrapped air bubbles that weregenerated due to intense mechanical mixing.

2.2.2. Manufacturing of fiber reinforced nanocompositesE-glass/epoxy nanocomposites were fabricated by using a

combination of hand lay-up and hot press techniques to ensureproper fiber wetting and uniform resin distribution in the fabrics.

60 70 80 90 100 110 120 130 140 15060

70

80

90

100

110

120

130

140

150

Abso

rbed

Ene

rgy

(J)

Impact Energy (J)

60 70 80 90 100 110 120 130 140 15060

70

80

90

100

110

120

130

140

150

Abso

rbed

Ene

rgy

(J)

Impact Energy (J)

60 70 80 90 100 110 120 130 140 15060

70

80

90

100

110

120

130

140

150

Abso

rbed

Ene

rgy

(J)

Impact Energy (J)

NTs and (c) 0.5 wt.% NH2-MWCNTs incorporated laminates obtained by polynomial andeader is referred to the web version of this article.)

Table 4Ballistic limit velocity (VBL) of control and NH2-MWCNTs incorporated E-glass/epoxycomposites.

Statistical analysis Controllaminates

0.3 wt.%laminate

0.5 wt.%laminate

Polynomialregression

Energy at ballistic limit (J) 91 103 92Ballistic velocity limit (m/s) 294 313 296

Powerregression

Energy at ballistic limit (J) 92 101 95Ballistic velocity limit (m/s) 296 311 300

M. Rahman et al. / International Journal of Impact Engineering 57 (2013) 108e118112

At first, twelve layers of E-glass fabric were stacked properly. Theorientation of fiber was kept constant during hand lay-up process.The entire set-up was then kept in a hot press maintained at 60 �C.The consolidation of the laminates was carried out by applying apressure of 1.43 MPa for 6 h. One reason to carry a higher tem-perature curing of SC-15 epoxy resinwas to achieve good fiber wet-out by reducing the viscosity of nanophased resin. The other reasonis to reduce the gel time. Because gel time decreases as the curetemperature increases which results in cure reaction occurringmore rapidly [18]. After completion of curing, the laminate wastaken out from the press and thermally post cured at 100 �C for 5 hin a mechanical convection oven. The final thickness of the panelwas found to be 5.25 mm. The schematic of the whole process isshown in Fig. 1.

Fig. 3. Optical microscope of fractured samples showing (a) fiberematrix delamina-tion and (b) fiber breakage.

2.3. Testing procedure

2.3.1. Calculation of fiber volume fractionVolume fraction of fiber, matrix and void content of the fabri-

cated E-glass/epoxy composites was determined by matrix diges-tion test according to ASTMD3171-99(2004) [19]. Samples cut fromthe panels as per standard were weighed accurately and weresubmerged in a bath containing 80% concentrated nitric acid forabout 5 h at a temperature maintained 75 �C. After 5 h, the matrixwas digested completely in the acid. The fibers were washed byusing acetone and water, and kept in a convection oven maintainedat 100 �C for 1 h to dry. After drying, the weight of the fiber wasmeasured and the corresponding volume fractions were calculatedusing the following equations:

Fiber volume fraction; Vf ¼ W=Fw=C

� 100% (1)

Matrix volume fraction; Vm ¼ ðw�WÞ=MV

� 100% (2)

Void volume fraction; Vv ¼ 100��Vm þ Vf

�(3)

where,W is the weight of fiber in the composite, w is the weight ofthe initial composite specimen, F is the fiber density, M is thedensity of the matrix, V is the volume of the composite and C is thecomposite density.

2.3.2. High velocity impact testingHigh velocity or ballistic impact tests were carried out using a

gas-gun test set-up. The gas-gun consists of a 3 m long barrel, afast acting high pressure firing valve, and capture chamber. Thesample of dimension 120 mm � 120 mm � 5.25 mmwas mountedin a fully clamped boundary condition between two rigidaluminum plates in the capture chamber. Spherical steel pro-jectiles with a diameter of 7.90 mm and weighing 2.10 g wereused. Polyurethane sabots with a diameter of 40 mm assistedspherical projectiles through the barrel. The projectile was sepa-rated from the launching sabot by a sabot stripper plate beforeimpacting the target. Initial velocity and residual velocity of theprojectile were measured using two chronographs mounted at thebottom of the capture chamber with a transparent optical window.Helium gas was used as propellant. Impact velocity was varied byvarying the pressure of gas in the firing chamber. Seven differentpressures setting varying from 0.34 to 1.38 MPa (50e200 psi) wereused for the test. Incident velocity levels ranged from 240 m/s to380 m/s and accordingly initial kinetic energy levels ranging from60 J to 150 J.

2.3.3. Micrographic analysisDispersion state of MWCNTs in epoxy resin was investigated

by Transmission Electron Microscopy (TEM) using a Zeiss EM10Transmission Electron Microscope operated at 60 kV. The analysisof fracture surfaces was carried out using a Zeiss EVO50 scanningelectron microscope (SEM) at 20 kV accelerating voltage. Specimensurfaces were coated with a thin gold film to protect the fracturesurfaces from beam damage and to prevent charge build up. Thefailure pattern was investigated using optical microscopy.

M. Rahman et al. / International Journal of Impact Engineering 57 (2013) 108e118 113

2.3.4. Ultrasonic non-destructive evaluationAn ultrasonic c-scan inspection was carried out to obtain

quantitative information of damage of the impacted laminates us-ing an ultrasonic pulse-receiver unit with ODIS software made bySonix Inc. The scanning was carried out in pulse-echo immersionmode using a 5-MHz 50 mm point focus transducer. In the currentstudy, scanning was done in the xey direction with an electronicgate set on the back surface signal. Scan data was collected at every0.05 mm interval. Software associated with the set-up has a pro-vision to see real-time digitized oscilloscope where the signals aremonitored. Both amplitude as well as time-of-flight informationwas collected and stored as image files in TIF or GIF format.

3. Results and discussion

3.1. Fiber volume fraction

Fiber volume fraction and void content in control, 0.3 and0.5 wt.% CNTs incorporated E-glass/epoxy composites calculated bymatrix digestion test is summarized in Table 1. The average fibervolume fraction of the laminates was 57e59% with void content of4e5%, respectively. It can be seen thatwith increase ofMWCNTswt.%loading, the percentage of void content was also increased and can

Fig. 4. Pictures of top and bottom surface, and ultrasonic time-of-flight c-scan imag

be attributed majorly due to increase of resin viscosity, which wehave previously reported [5]. With increasing CNTs wt.% loading, theresin viscosity increased in comparison to control resin. The increasein resin viscosity may hinder the removal of entrapped bubbles andvolatile impurities from the systems during processing.

3.2. High velocity impact characterization

3.2.1. Absorbed energyHigh velocity impact properties were analyzed in terms of

absorbed energy efficiency and ballistic limit velocity of fabricatedlaminates to figure out the effect of NH2-MWCNTs on ballisticproperties of E-glass/epoxy composites. Typical room temperatureinitial and final velocity determined by chronographs equipped inthe testing device, are shown in Table 2. The summary of thecalculated absorbed energy results at all of the impact velocitylevels is shown in Table 3. Table 3 also provides the information onthe projected damage calculated based on the ultrasonic c-scanimages. For this calculation, damage is assumed to be elliptical inshape. Major and minor axes were measured to calculate the area.The percentage changes in the damage area for samples with NH2-MWCNTs as compared with control (ones with no NH2-MWCNTs)samples were calculated.

es showing projected damage of samples tested at 1.38 MPa pressure setting.

M. Rahman et al. / International Journal of Impact Engineering 57 (2013) 108e118114

Absorbed energy is one of the main parameters to assess andevaluate the degree of damage and process in a composite lami-nates after an impact incident. Impact energy is defined as the ki-netic energy of the projectile that is transferred to composites justbefore impact while absorbed energy is the amount of energydissipated within composites at the end of an impact event.Absorbed energy can be calculated in several ways. One of the waysis to measure the integral area under loadedeflection curve whichis treated as the total absorbed energy by a composite specimen. Inthis study, total absorbed energy was determined by the differenceof initial kinetic energy just before impact and final kinetic energyjust after impact according to Eq. (4).

Ea ¼ 12mv2 � 1

2mu2 (4)

where, m, v, and u are the mass, incident impact velocity and re-sidual velocity of the projectile, respectively and Ea is the absorbedenergy by the target.

The positive effect of NH2-MWCNTs in epoxy resin is clearlyevident in case of absorbed energy improvements (Table 3). Energyabsorption capability by composites has been improved at 0.3 wt.%loading of CNTs with respect to control laminates and thendecreased. For example, at 1.38 MPa release pressure level, theaverage absorbed energy was 103.95 J, 107.28 J and 100.83 J forcontrol, 0.3 wt.% and 0.5 wt.% samples, respectively. The

Fig. 5. Pictures of top and bottom surface, and ultrasonic time-of-flight c-scan imag

corresponding initial velocity and residual velocity were 141.71 m/sand 37.76 m/s, 143.60 m/s and 36.32 m/s, 143.60 m/s and 42.77 m/sfor control, 0.3 wt.% and 0.5 wt.% samples, respectively. A similartrend was also observed in case of all other release pressure level.

As shown in Table 3, the results clearly show that, in some casesabsorbed energy is lower than impact energy and in some casesabsorbed energy is same as impact energy. So some of the test dataare below equal energy line where absorbed energy equals toimpact energy and in other cases it is coincident on the line. Thisclearly indicates that there are complete penetrations in all thesamples in which absorbed energy is less than impact energy. Theexcess energy i.e. impact energy minus absorbed energy, defined asresidual energy was retained by the projectile after completelypenetrating the sample. The other case indicates that the projectileis caught by the laminates and the projectile velocity becomes zerobefore all the fibers are broken. It indicates the absorbed energy andimpact energy is same and the partial penetration has occurred.

3.2.2. Ballistic limit velocity (VBL)If the projectile’s initial kinetic energy is less than or equal to the

absorbed energy of the target, then the projectile can either getstuck within the target or rebound. Complete penetration takesplace with certain residual velocity if the projectile’s initial kineticenergy is more than the absorbed energy. When the projectilepenetrates the target completely with zero residual velocity theninitial velocity of the projectile of a given mass is referred to as the

es showing projected damage of samples tested at 0.62 MPa pressure setting.

M. Rahman et al. / International Journal of Impact Engineering 57 (2013) 108e118 115

ballistic limit of a target. In other word, ballistic limit is oftendefined as the minimum impact velocity that will result in com-plete perforationwith zero exit velocity. At ballistic limit, the targetabsorbs total impact energy of the projectile. As a result, absorbedenergy by the target is exactly same as the initial kinetic energy ofthe projectile from which the ballistic limit velocity of a target canbe calculated. As shown in Table 2, the velocity range whereembedded phenomenon occurred was found in between 290 and299 m/s for control laminates while it became in between 301 and307 m/s for 0.3 wt.% CNTs laminates. Surprisingly, it was found inthe range of 301e319 m/s in 0.5 wt.% loading.

As ballistic limit or V50 as is generally referred to as a particularvalue of impact velocity at which at least 50% of the samples testedwill be completely penetrated by the projectile. Unless a largenumber of samples are tested close to the ballistic limit, it isparticularly difficult to determine the same. Hence, polynomial andpower regressions were carried out to get a statistical value forballistic limit velocity as shown in Fig. 2. An equal energy line andthe line obtained from the fit by polynomial and power regressionintersects at some point where impact and absorbed energy aresame. The farthest intersection point from the origin beyond whichit will not intersect any more, can be used as statistically deter-mined ballistic limit of the fabricated laminates. At the intersectionpoint, the impact energy is determined and the ballistic limit iscalculated from the kinetic energy theorem for a givenmass. In case

Fig. 6. Pictures of top and bottom surface, and ultrasonic time-of-flight c-scan imag

of polynomial regression analysis, the energy at ballistic limit isfound to be 91 J for control laminates, 103 J for 0.3 wt.% laminatesand 92 J for 0.5 wt.% laminates. Consecutively, by power regression,the energy at ballistic limit is found to be 93 J for control laminates,101 J for 0.3 wt.% laminates and 95 J for 0.5 wt.% laminates. Usingkinetic energy formula, the apparent ballistic limit has been foundto be 294 m/s for control laminates, 313 m/s for 0.3 wt.% laminatesand 296 m/s for 0.5 wt.% laminates by polynomial regression asshown in Table 4. On the other hand, by power regression theballistic limit has been found to be 296 m/s for control laminates,311 m/s for 0.3 wt.% laminates and 300 m/s for 0.5 wt.% laminates(Table 4). Thus, the improvement in ballistic limit has been found toincrease by about 6% for 0.3 wt.% loading with respect to controllaminates determined by statistical analysis. Statistical analysisclosely agreed with the experimental analysis.

3.3. Ultrasonic evaluation of impact damage

All the samples were subjected to ultrasonic c-scan evaluationafter the impact tests to determine the extent of damage andcorrelate the incident impact energy to the damage. The tests werecarried out in an immersion mode by pulse-echo technique. In thissystem the specimenwas submerged into awater tank. An ultrasonicpulse was sent from the submerged piezoelectric transducer into thespecimen. Ultrasonic pulse travels through the water column. Part of

es showing projected damage of samples tested at 0.34 MPa pressure setting.

Fig. 7. TEM micrographs of (a) 0.3 wt.%, and (b) 0.5 wt.% epoxy nanocomposites.

M. Rahman et al. / International Journal of Impact Engineering 57 (2013) 108e118116

the signal gets reflected back from the front surface of the samplewhile remaining in transmitted into the sample. The portion of thepulse that is traveling through the sample will undergo inherentattenuation within the sample. If the location of the sample whereultrasonic pulse enters the sample is defect-free, the pulse travelsthrough the thickness of the sample and again gets reflected back.On the other hand, if there is a defect or damage in the path of ul-trasonic pulse, then that acts as a reflector and ultrasonic pulse getsreflected back from defect or damage. If the defect or damage issubstantial, then most of the transmitted pulse will not reach theback surface. Collecting the back surface signal, hence from all thescanned points and converting the digitized data in an image formwill create a c-scan, which gives the projected information of thedamage in the sample. Projected damage calculated based on the c-scan time-of-flight data is presented in the last column of Table 3. Ifthe impact energy is less, the projectile bounces of the sample.However, in the current study, the lowest incident energy was highenough to create damage in the sample and projectiles actuallypenetrated the samples either partially or completely. When there ispartial penetration, the projectile gets caught in the sample trans-ferring its energy completely to the sample. Some of the incidentenergy of the sample is absorbed in elastic deformation of thesample while the remaining is spent in creating damage in thesample. Major mode of damage in woven fabric composites isdelamination, in addition to fiber breakage at the back surface aswell as matrix cracking as shown in Fig. 3. In general, the interfacewhere the projectile gets arrested has the largest delamination. Onthe other hand, when the projectile completely penetrates thesample, the damage gets muchmore localized and is generally lowerthan the case where there is partial penetration of the sample as isseen for the samples 1e3 in all cases. When the samples withMWCNTs are comparedwith thosewithoutMWCNTs on one-on-onebasis at a given pressure setting, it is generally noticed that thedamage size is much lower in case of samples with MWCNTs as canbe seen from Figs. 4e6 which give front and back surface images aswell as time-of-flight c-scan images of samples tested at 0.34, 0.62and 1.38 MPa pressure settings, respectively. This is more so whenthere is partial penetration, i.e. below the ballistic limit. This is due tothe fact that the samples with MWCNTs have higher bending stiff-ness and absorb more energy in terms of elastic deformation thanthat of the control samples. Hence, the amount of energy that isrequired to create damage is relatively more. This results in lowerdamage area. Even after complete penetration, the damage area isless for samples with MWCNTs.

3.4. Effect of MWCNTs on ballistic properties and damage area

As previously stated, GFRP composites are considered as high-ductility advanced composites where possible damage mecha-nism by high velocity impacts are tensile failure and deformation offiber yarns, delamination and matrix cracking [10]. In other word,these are the main energy absorbing mechanisms by GFRP com-posites. So, ballistic properties and damage mechanisms of thecomposites are strongly dependent on the fracture toughness andbending properties of the target including elastic modulus, tensilestrength and failure strain of fiber and matrix as well as interfacialbonding between them.

One of the obvious factors that contribute to the improvementsin ballistic impact properties is the increased load transfer capa-bility between the epoxy matrix and nanotubes. It has been re-ported that effective stress transfer between epoxy matrix andamino-functionalized MWCNTs contributed in enhancing me-chanical properties of fiber reinforced composites [20]. However,an effective stress transfer between epoxy and nanotubes isstrongly dependent on improved dispersion. Fig. 7 shows the

degree of dispersion at both of the loading that reveals clearly abetter dispersion of 0.3 wt.% CNTs in epoxy resin. An improveddispersion of CNTs in matrix facilitates to enhance interfacial re-actions and forms covalent bond between them. This covalent bondinduces different cross-linking regions into the epoxy matrix.Generally after mixing epoxy Part A and NH2-MWCNTs, the inter-facial reaction takes place between amine functional groups ofMWCNTs and epoxide groups of DGEBA resin. Two ring openingreactions followed by a cross-linking reaction create interlockingstructure in the resin blend which facilitates impediment of themobility of polymer chains in the system [5]. A bridging betweenepoxy matrix and MWCNTs, and thus, a better adhesion betweenthem has been observed due to cross-link interaction as found bySEM investigation of cured epoxy/MWCNT composites (Fig. 8).When crack propagates in nanocomposites through a nanotube,crack tips cannot break the strong MWCNTs due to the bridging bycross-linking. The energy of the tips is significantly reduced by thelarge quantity of nanotubes pull-out as shown in Fig. 8. As a result,crack tips are then forced to stop or frequently change their crackpropagation line. This results in an increase to fracture toughness of

Fig. 9. Interfacial bonding between fiber and matrix in (a) neat sample and (b) 0.3 wt.%NH2-MWCNTs incorporated sample.

Fig. 8. SEM Micrograph showing CNT pull-out and bridging between NH2-MWCNTsand epoxy resin.

M. Rahman et al. / International Journal of Impact Engineering 57 (2013) 108e118 117

the composites [21]. Due to this phenomenon, crack initiation andpropagation become difficult within thematrix and at fiberematrixinter-phase than that of without reinforced CNTs samples as shownin Fig. 9(b) that result in a higher energy absorption capability inthe composites. Due to increased fracture toughness, less delami-nation and matrix cracking might occur and a decrease in damagearea as well as an increase in absorption capability and ballisticlimit of GFRP composites was observed. However, no such bondingwas observed in control GFRP samples as shown in Fig. 9(a).

The noticeable increase in absorbed energy, ballistic limit andprojected damage area at 0.3 wt.% loading thus can be attributed tothe better dispersion of MWCNTs and better interfacial interactionamong the amino-functionalized MWCNTs, epoxy matrix and epoxysilanes of glass fiber interfaces. A decrease in the ballistic perfor-mance of samples with 0.5 wt.% MWNCTs was observed. It is wellknown that high surface area of nanotubes gives it desirable inter-face to transfer stresses. However, it creates a strong attractive forcebetween them leading to excessive CNTs agglomeration. In addition,incorporation of nanotubes in polymer matrix has a major influenceon the viscosity of nanocomposites and hence, mechanical proper-ties also. It was reported that the incorporation of more than 0.3 wt.%loading increases the viscosity of matrix which is reported as animpediment to dispersion [5]. A poor dispersion has also beenobserved in case of 0.5 wt.% CNTs loading from the TEM investigation(Fig. 7(b)). Increased resin viscosity facilitates increased void con-tents in the systems during processing. A strong correlation betweenlarge voids and a detrimental effect on the initiation and propagationof static failures of unidirectional fiber composite was found bychambers et al. [22]. The void content has severe impact on thematrix-dominated properties, while the fiber-dominated properties

are less significantly affected [23]. High viscosity also increases thesurface tension of resin systems. In case of laminate fabrication, thisis one of the main problems as it causes poor wetting of the glassfibers that leads to poor adhesion between glass fibers and epoxymatrix during fabrication. As a result, a potential decrease has beenobserved in the ballistic limit velocity, absorbed energy and pro-jected damage area of the composites at 0.5 wt.% loading.

3.5. Statistical significance of the test

Statistical analysis has been carried out to determine the sta-tistical significance of the difference among the means of the threegroups by one-way analysis of variance (ANOVA) as the experi-mental values were close to each other. In statistics, if the p-valueobtained from ANOVA is less than 0.05, the data from differentgroups can be considered statistically distinct at a 95% confidencelevel. Otherwise they are belongs to the same group. The p-value ofthe absorbed energy are found to be 0.2659, the mean propertiesfrom one group of composite to another is not significantlydifferent at the 95.0% confidence level. However, the p-value of theprojected damage area data is found to be 0.0163, the mean fromone group of composites to another are significantly different at the95.0% confidence level. To compare the mean projected damagearea for each of the group to that of the other group, Tukey’sHonestly Significant Difference (HSD) test has been adopted. Fromthis statistical test, it was obtained that the properties of 0.3 wt.%samples are always significantly different than 0.4 wt.% samples

M. Rahman et al. / International Journal of Impact Engineering 57 (2013) 108e118118

and control samples. Again, control samples and 0.4 wt.% samplesare not significantly different from each other.

4. Conclusions

In this study, NH2-MWCNTs were incorporated in E-glass/epoxycomposites to enhance ballistic impact performance. MWCNTsincorporation to the epoxy resin system was carried out through acombination of sonication and three-roll mill methods. Based on theexperimental analysis, it can be reported that the addition ofMWCNTs at 0.3 wt.% loading increased the ballistic limit velocity byabout 6% whereas higher loading of MWCNTs did not increase theballistic limit. Energy absorption capability has also been increasedat 0.3 wt.% loading of CNTs. The addition of MWCNTs to the com-posites increased the damage tolerance considerably, exhibitinglower damage size. These phenomena were attributed to the in-crease in bending strength and stiffness of the samples. Amino-functionalized MWCNTs interact with epoxy system create ahigher cross-link density of polymer network. Modification of epoxymatrix and improved fiberematrix interfacial shear strength mightincrease the fracture toughness. The advantages become two-fold.Increased bending stiffness results in higher energy absorption inthe elastic deformation whereas increased fracture toughness re-duces the extent of delamination and matrix cracking. Overall, thiswork showed that the ballistic performance of E-glass/epoxy sam-ples can be enhanced by adding a very small percentage of amino-functionalized MWCNTs. At higher loading, the effect is reduceddue to poor dispersion that causes a poor interfacial interactionbetween fiber and matrix. Further, due to increased viscosity properwetting of fibers does not take place and hence, the interfacial shearproperties alongwith othermechanical properties are not enhanced.

Acknowledgements

Authors would like to acknowledge the support of ERDC-CERL(W9132T-11-C-0027) and NSF-EPSCoR (grant #EPS-1158862) forfunding this work.

References

[1] Grujicic M, Bell WC, Thompson LL, Koudela KL, Cheeseman BA. Ballistic-pro-tection performance of carbon-nanotube-doped poly-vinyl-ester-epoxy ma-trix composite armor reinforced with E-glass fiber mats. Mater Sci Eng 2008;479:10e22.

[2] Schadler LS, Giannaris SC, Ajayan PM. Load transfer in carbon nanotube epoxycomposites. Appl Phys Lett 1998;73:3842e4.

[3] Gojny FH, Wichmann M, Fiedler B, Schulte K. Influence of different carbonnanotubes on the mechanical properties of epoxy matrix composites e Acomparative study. Compos Sci Technol 2005;65:2300e13.

[4] Liao YH, Olivier MT, Liang ZY, Zhang C, Wang B. Investigation of the dispersionprocess of SWNTs/SC-15 epoxy resin nanocomposites. Mater Sci Eng A 2004;385:175e81.

[5] Rahman MM, Zainuddin S, Hosur M, Malone JE, Salam MBA, Kumar A, et al.Improvements in mechanical and thermo-mechanical properties of e-glass/epoxy composites using amino-functionalized multi-walled carbon nanotubes.Compos Struct 2012. http://dx.doi.org/10.1016/j.compstruct.2012.03.014.

[6] Song YS, Youn JR. Influence of dispersion sates of carbon nanotubes onphysical properties of epoxy nanocomposites. Carbon 2005;43:1378e85.

[7] Sahoo NG, Rana S, Cho JW, Li L, Chan SH. Polymer nanocomposites based onfunctionalized carbon nanotubes. Prog Polym Sci 2010;35:837e67.

[8] Yaping Z, Aibo Z, Quinghua C, Jiaoxia Z, Rongchang N. Functionalized effect oncarbon nanotube/epoxy nano-composites. Mater Sci Eng A 2006;435e436:145e9.

[9] Ma PC, Mo SY, Tang BZ, Kim JK. Dispersion, interfacial interaction and reag-glomeration of functionalized carbon nanotubes in epoxy composites. Carbon2010;48:1824e34.

[10] Naik NK, Shrirao P. Composite structures under ballistic impact. ComposStruct 2004;66:579e90.

[11] Cantwell WJ, Morton J. Impact perforation of carbon fibre reinforced plastic.Compos Sci Technol 1990;38(2):119e41.

[12] Lee SWR, Sun CT. Dynamic penetration of graphite/epoxy laminates impactedby a blunt-ended projectile. Compos Sci Technol 1993;49(4):369e80.

[13] Jenq ST, Jing HS, Chung C. predicting the ballistic limit for plain woven glass/epoxy composite laminate. Int J Impact Eng 1994;15:451e64.

[14] Ulven C, Vaidya UK, Hosur MV. Effect of projectile shape during ballisticperforation of VARTM carbon/epoxy composite panels. Compos Struct 2003;61:143e50.

[15] Esfahani JM, Sabet AR, Esfandeh M. Assessment of nanocomposites based onunsaturated polyester resin/nanoclay under impact loading. Polym AdvTechnol 2012;23(4):817e24.

[16] Ávila AF, Neto AS, Nascimento JH. Hybrid nanocomposites for midrange bal-listic protection. Int J Impact Eng 2011;38(8):669e76.

[17] Reynaud E, Gauthier C, Perez J. Nanophases in polymer. Rev Metall 1999;96:169e76.

[18] Park S, Bae K, Seo M. A study on rheological behavior of MWCNTs/epoxycomposites. J Ind Eng Chem 2010;16:337e9.

[19] ASTM D 3171-99. Standard test method for constituent content of compositematerials. West Conshohocken, PA: Annual Book of ASTM Standards; 2004.

[20] Shen J, Huang W, Wu L, Hu Y, Ye M. Thermo-physical properties of epoxynanocomposites reinforced with amino-functionalized multi-walled carbonnanotubes. Compos Part A 2007;38:1331e6.

[21] Rahman M, Hosur M, Zainuddin S, Jajam K, Tippur H, Jeelani S. Mechanicalcharacterization of epoxy composites modified with reactive polyol diluentand randomly oriented amino-functionalized MWCNTs. Polym Test 2012;31:1083e93.

[22] Chambers AR, Earl JS, Squires CS, Suhot MA. The effect of voids on the flexuralfatigue performance of unidirectional carbon fibre composites developed forwind turbine applications. Int J Fatigue 2006;28:1389e98.

[23] Zhu H, Wu B, Li D, Zhang D, Chen Y. Influence of voids on the tensile per-formance of carbon/epoxy fabric laminates. J Mater Sci Technol 2011;27(1):69e73.