High Resolution X-ray Tomography of Fiber Reinforced Polymeric Composites

Transcription

1 University of Tennessee, Knoxville Trace: Tennessee Research and Creative Exchange Masters Theses Graduate School High Resolution X-ray Toography of Fiber Reinforced Polyeric Coposites Stephen Andrew Young University of Tennessee - Knoxville Recoended Citation Young, Stephen Andrew, "High Resolution X-ray Toography of Fiber Reinforced Polyeric Coposites. " Master's Thesis, University of Tennessee, This Thesis is brought to you for free and open access by the Graduate School at Trace: Tennessee Research and Creative Exchange. It has been accepted for inclusion in Masters Theses by an authorized adinistrator of Trace: Tennessee Research and Creative Exchange. For ore inforation, please contact trace@utk.edu.

2 To the Graduate Council: I a subitting herewith a thesis written by Stephen Andrew Young entitled "High Resolution X-ray Toography of Fiber Reinforced Polyeric Coposites." I have exained the final electronic copy of this thesis for for and content and recoend that it be accepted in partial fulfillent of the requireents for the degree of Master of Science, with a ajor in Engineering Science. We have read this thesis and recoend its acceptance: Roberto S. Benson, John D. Landes (Original signatures are on file with official student records.) Dayakar Penuadu, Major Professor Accepted for the Council: Dixie L. Thopson Vice Provost and Dean of the Graduate School

3 To the Graduate Council: I a subitting herewith a thesis written by Stephen Andrew Young entitled High Resolution X-ray Toography of Fiber Reinforced Polyeric Coposites. I have exained the final electronic copy of this thesis for for and content and recoend that it be accepted in partial fulfillent of the requireents for the degree of Master of Science, with a ajor in Engineering Science. Dayakar Penuadu Major Professor We have read this dissertation and recoend its acceptance: Roberto S. Benson John D. Landes Accepted for the Council: Carolyn R. Hodges Vice Provost and Dean of the Graduate School

4 HIGH RESOLUTION X-RAY TOMOGRAPHY OF FIBER REINFORCED POLYMERIC COMPOSITES A Thesis Presented for the Master of Science Degree The University of Tennessee, Knoxville Stephen Andrew Young August 2009

5 DEDICATION I dedicate this thesis to y recognition of the Universal Mind, y parents, Dr. John H. Young, Jr. and Sonja Young, and y brothers, John H. Young III, and Paul K. Young, for their unending love, support, and encourageent. Together they have inspired e to live a life with integrity. ii

6 ACKNOWLEDGEMENTS My heartfelt thanks and deepest appreciation go to y advisor, Dr. Dayakar Penuadu, giving e an opportunity and his guidance, patience, and support throughout the course of y study. My wholehearted thanks also go to the distinguished ebers of the Coittee, Dr. Roberto S. Benson and Dr. John Landes for reviewing of this thesis. I would like to express y gratitude to Mr. Jaes T. Pippin for his guidance and support throughout y course of study. In addition I would to express y gratitude to Walter Odo, Dr. Ronald McFadden, Celeste Brooks, Ken Thoas, Douglas E. Fielden, Vlastiil Kunc, Jaret Fratjord, Ashely Stowe and Robin Woracek throughout y course of study. Without their inestiable assistance, this research would not have been possible. iii

7 ABSTRACT A high resolution x-ray toography syste was used to study chopped fiber polyeric coposites ade of polypropylene resin, nickel coated carbon fiber and E- glass fiber. Procedures are developed to obtain icro-structured features of iportance. In-situ tensile testing syste was developed and integrated into the existing hardware for toography equipent to study the evolution of daage and icro-structural features as a function of echanical stress. High resolution x-ray toographic iages of glass fiber were collected and viewed on a icron scale. The radiographs were reconstructed to visualize the fiber content of the saples in three diensional volue. In addition, glass fiber dogbone speciens were tested on a iniature tensile achine using x-ray toography to view deforation of the saples in high resolution. Fractures in the chopped glass coposite were observed for x-ray icroscopy showing the doinant failure echanis of the saple are low interfacial strength and adhesion between the fiber and atrix. Cracks were not observed until after failure by fiber pull-out using the digital icroscopy ethod. Using SEM icroscopy ethod, resin cracking and fiber debonding was observed for a carbon fiber with vinyl ester resin while under tensile loading. Iportant icro-structural inforation relationship with and echanical behavior including variation odulus, yield and ultiate strength are discussed. iv

8 TABLE OF CONTENTS Chapter Page 1 INTRODUCTION 1 2 HIGH RESOLUTION X-RAY TOMOGRAHY OF CHOPPED FIBER POLYMERIC COMPOSITES 2.1 Introduction Materials and experiental setup Coposite Speciens Micro-XCT Experiental Set-up Fiber Volue Fraction Experiental results Fiber Length Distribution, 3D Reconstruction, 9 and Fiber Volue Fraction Effective Moduli Conclusion 16 3 IN-SITU DAMAGE EVOLUTION OF POLYMERIC COMPOSITES USING RESOLUTION X-RAY TOMOGRAPHY, DIGITAL MICROSCOPY, AND LARGE CHAMBER SEM 3.1 Introduction Materials Coposite Speciens Experiental Procedure and Results 21 v

9 3.3.1 Tensile Testing Micro-XCT In-situ Mechanical Testing Syste In-Situ Mechanical Tensile Testing Syste using 25 Digital Optical Microscopy In-situ Mechanical Tensile Testing using Large 32 Chaber Scanning Electron Microscope (LC-SEM) 3.4 Conclusion 36 REFERENCES 38 APPENDIX 45 VITA 63 vi

10 LIST OF FIGURES Figure Page 2.1 Scheatic of the 3 thick pellet injection olding of E-glass 46 fiber/polypropylene ISO-plaque and center-gated disk 2.2 Scheatic diagra of the skin, shell, and core layers of the E-glass fiber 46 polypropylene olding copound at 50x agnification. 2.3 Radiograph of an absorption Radon transfor projection of the E-glass 47 fiber with polypropylene atrix at 4x agnification. 2.4 Illustration of toography raw iage collection. E-Glass fiber 47 with polypropylene atrix coposite saple (0.5 x 0.5 x 10 ) inside saple holder. The X-ray source (100 kv, 10 W) and theroelectrically cooled scintillation CCD detector (2048 x 2048 pixels, 16 bit) at -59 o C with fast readout scintillation crystal. 2.5 Radiograph x-ray at 20x, and (b-c) 2D reconstruction slices 48 of the cross section bit grayscale toography reconstruction and corresponding 48 thresholded iage of E-glass fiber with polypropylene atrix coposite saple. Thresholded and easured E-glass fibers in the area of interest. 2.7 Fiber Length Distribution of E-glass fibers in polypropylene atrix Figure 8: Area Density Distribution for fast-filled and slow-filled 49 E-glass fibers in polypropylene atrix. vii

11 2.9 3D reconstructed iages of the E-glass fiber 50 with polypropylene atrix Flow chart to deterine effective oduli for the E-glass fiber in 50 polypropylene atrix (a) 8-bit grayscale toography reconstruction, (b) corresponding 51 thresholded iage and (c) effective odulus for slow-filled E-glass fiber with polypropylene Figure 12: Effective odulus, E-glass spheres in polypropylene atrix Scheatic of the 3 thick, 45 o [±45, 2s] carbon fiber 53 with vinyl ester resin and tensile saple. 3.2 Scheatic of (a) E-glass fiber/polypropylene dogbone coposite 54 saple and (b) corresponding finite eleent analysis stress concentration distribution. 3.3 In-situ echanical tensile syste with actuator and N (300 lbf) 54 load cell. E-Glass fiber with polypropylene atrix coposite saple (2.3 x 1.4 x ) inside saple holder. The X-ray source (100 kv, 10 W) and theroelectrically cooled scintillation CCD detector (2048 x 2048 pixels, 16 bit) at -59 o C with fast readout scintillation crystal. 3.4 Radiograph of transission Radon transfor projection of the 55 E-glass fiber with polypropylene atrix at 4x agnification (a) before tensile loading (b) during tensile loading and (c) after failure of the tensile saple. viii

12 3.5 Experiental data corresponding to fast-filled E-glass fiber 55 with polypropylene dogbone saple. 3.6 (a) 2-D reconstruction of E-glass fiber with polypropylene resin and 56 (b) corresponding 3-D voluetric rendering after tensile loading. 3.7 Typical experiental data for a tensile test for E-glass/polypropylene Fracture surface of E-glass with polypropylene resin after tensile failure Typical experiental data for a tensile test for carbon fiber 58 with vinyl ester resin Fracture surface of carbon fiber/vinyl ester after tensile failure 59 for (a,b) by fiber pullout and (c) cleavage fracture Experient setup for the in-situ echanical syste in LC-SEM Daage evolution of carbon fiber/vinyl ester tensile saple 60 under tensile loading Load-Displaceent daage evolution of carbon fiber/vinyl ester tensile 60 saple under tensile loading Daage pattern of carbon fiber vinyl ester and (b) fracture surface of 61 carbon fiber/vinyl ester fiber after tensile failure. ix

13 CHAPTER 1 INTRODUCTION Polyeric coposites ade of glass or carbon fiber and polypropylene resin have been of significant interest in the autootive field due to their light weight, resistance to corrosion, and high strength. Despite these advantages, the ode of failure of such aterials is typically sudden due to failure in fibers, resin, or their cobination. In this research, high-resolution toography was attepted to evaluate the state of icrostructure and daage evolution as function of echanical stress. High resolution x-ray toographic iages of glass fiber and nickel-coated carbon fiber coposites were collected at exceptional resolution with sub-icron scale. The radiographs will be reconstructed to visualize the fiber arrangeent of the saples in three diensional volue. In addition, glass fiber dogbone speciens were tested on a custo iniature tensile achine using x-ray toography to view deforation of the saples as a function of applied stress. The icrostructure of polyeric coposite aterials can significantly influence its overall functional properties such as its echanical strength, electrical and theral conductivity. Factors including total nuber of fibers in the atrix, diaeter, length, orientation of fibers, interface adhesion with atrix polyer, and pore structure affect the properties. Micro-toography (Micro-XCT) can produce iages with a sufficient resolution to study coposites with features including sall diaeter fibers (e.g. 5 to 15 u). Analysis of 2D or 3D can provide an accurate description of icrostructure, such as fiber volue ratio, fiber distribution and orientation. The evaluation of icro-structure 1

14 under echanical stress is of interest to develop an understanding of fracture and fatigue behavior of such aterials. A high resolution x-ray toography syste was used for the first tie to study chopped fiber polyeric coposites ade of polypropylene resin, nickel coated carbon fiber and E-glass fiber. The research project evaluated challenges of developing procedures to obtain quantitative icro-structural features of significance. In addition, in-situ tensile testing syste was developed and integrated into the existing hardware for toography equipent to study the evolution of daage and icro-structural features as a function of echanical stress with ORNL collaboration. This research will pave the way to quantify daage and develop new class of scale-dependent constitutive odels for coposites aterials for iediate use in structural and transportation applications. 2

15 CHAPTER 2 HIGH RESOLUTION X-RAY TOMOGRAPHY OF CHOPPED FIBER POLYMERIC COMPOSITES This chapter is a slightly revised version of a paper with the sae title subitted for the Journal of Coposites Science and Technology in 2009 by Stephen A. Young, Dayakar Penuadu, and Vlastiil Kunc. My priary contributions to this paper include: (i) developent of the proble into a work, (ii) identification of the study areas, (iii) gathering and reviewing of literature, (iv) sapling, processing, and analyzing reconstructed x-ray iages fro x-ray radiography and toography, (v) pulling various contributions into a single paper, (v) ost of the writing. Abstract Polyeric coposites have been of particular interest in the autootive field due to their low weight, high strength, and resistance to environental degradation. In this research, high resolution x-ray iages of chopped glass fiber coposites are obtained with a high spatial resolution of 3. The radiographs were reconstructed to visualize the fiber and atrix arrangeent in three diensions with an ability to analyze icrostructure including cracks, local fiber volue distribution, fiber orphology after extrusion, and interfacial bonding issues. A siple quantitative approach of using x-ray digital toographs to obtain spatial variation of coposite echanical properties is deonstrated for the E-glass/polypropylene coposite using a coposite sphere 3

16 analytical odel. Iportant icro-structural inforation such as the variation of fiber length distribution with distance fro the injection olding location is included. 2. Introduction This paper presents results on using x-ray coputed icro-toography (Micro- XCT) to evaluate the echanical behavior of short fiber reinforced coposites considering its icrostructure. Pellet injected E-glass coposite fiber speciens with polypropylene resin were evaluated as part of the ongoing research project between University of Tennessee, Oak Ridge National Laboratory (ORNL), Pacific Northwest National Laboratory (PNNL), and Delphi Corporation. The goal of this research is to evaluate and predict the echanical properties of short fiber reinforced coposites in 2D and 3D architecture [1]. Spherical inclusions surrounded by a resin atrix have been used successfully to odel the local and global constitutive behavior of the coposites [2-3]. This analytical odel was ipleented by the authors in this study to integrate 3D icrostructural easureents fro toography to predict spatially resolved coposite echanical properties. The pellet injection olding process influence on echanical properties of coposites and benefits of chopped fibers are well known [4-6]. Past research has not been perfored on evaluating the echanical properties of E-glass and polypropylene chopped fiber coposites considering detailed icro-structure inforation ainly due to a lack of precise non-destructive easureent technique. Hoayonifar [7] has analyzed the atrix-fiber interaction showing that increasing atrix volue fraction causes variation in stress distribution along the fiber length. In the past, autoatic iage analysis have been successfully used to count and accurately easure 4

17 fiber length with straight and curved fibers over large fiber aspect ratios using 2D optical iages. However, these techniques require destructive techniques such as burning resin after extrusion to obtain digital iages of chopped fibers in a dispersed state [8-9]. X-ray iaging is becoing iportant non-destructive approach due to its ability to provide a detailed three diensional visualization of the polyeric coposites for subsequent quantitative analysis to obtain fiber structure and orientation [10]. 2.2 Materials and experiental setup Coposite Speciens The coposite aterial used in this study included polypropylene resin having E- glass fibers anufactured by Montsinger Technologies, Inc. using a fiber elt process. The chopped E-glass fiber roving was 12.7 in length with a diaeter of 17 [1]. Detailed anufacturing aspects of the coposites were presented elsewhere [1] and a brief explanation is included this paper here. As shown in Figure 1, in the Appendix, the ISO-plaque and center-gated disk geoetries were used for the pellet injection olding using two voluetric flow rates (16.4 and c 3 /s) to evaluate effect of the injection speed on the as-fored icrostructure of chopped fiber coposites and corresponding physical properties. The old teperature was held at 78 o C, while the inlet teperature of the elt was 240 o C. The center gated disk is 3 thick and in diaeter. The ISO-plaque is 3 thick, 90 long and 80 wide. The flow direction for the center gated disk and ISO-plaque was radially outward as illustrated in Figure 2.1b. The bulk density of the E- glass fiber coposite with polypropylene resin coposite was g/c 3 [1]. The 5

18 fiber weight fraction corresponded to 40% for E-glass fiber/polypropylene (GF/PP) olding copound which leads to approxiately 20% global fiber volue fraction. Figure 2.1 shows three regions (A,B,C) of interest in our study to evaluate fiber length and fiber orientation easureents. The A, B, C regions corresponds to 6, 34, 64 distance fro the center of the disk and 15, 45, and 75 fro the injection point of the ISO-plaque saple as denoted in the figure. Typical crystallinity values of ISO-plaque olding copounds for the A, B, C regions were observed to be 47.6 %, 45.5 %, and 43.5%. The orphology of fibers and echanical properties of the coposite saples were studied using x-ray iaging to view the chopped fibers coposites at 3 icrons resolution. As shown in Figure 2.2, cubical saples having a size of 2 x 2 x 10 (with the fibers in the transverse direction) were obtained fro the specified A, B, and C region using a diaond saw. Figure 2.2 displays surface iage fro an optical icrograph showing the skin, shell and core layer structure of the GF/PP saple. The velocity at the core is higher than the shell and skin layer due to the velocity profile of the elt in contact with the wall inside the injection olding achine. Prior research found that the higher velocity gradient in the core layer results increased fiber orientation and iproved echanical properties in the direction of flow [11]. One edge noral to the transverse fibers was polished for optical icroscopic observations where photoicrographs were taken to analyze the fiber length and orientation easureents. 6

19 2.2.2 Micro-XCT Experiental Setup The use of Micro-XCT provides the benefit of fiber characterization in a nondestructive anner. The coputed toography used in our research affords high resolution (3 ) to view the spatial variation of the x-ray absorption for various projection views. X-ray absorption depends on several factors including density, length of absorbing aterial transversed by bea, and the x-ray wavelength [10]. The principles of x-ray iaging and coputed toography are covered in detail by Stock, Kak and Slaney, and Banhart [12-14]. The transission intensity through a saple is governed by the Beer-Labert Law shown in Eq. 1. s ( x, y) ds ( x, y) I 0 ( x, y) e (2.1) I 0 where I o is the incident intensity of the x-ray bea before passing through the a distance ds of the saple with attenuation coefficient (x,y). For coposite saples, an average value of (x,y) can be calculated fro the weight fraction of each eleent and ass absorption coefficient, [12]. The attenuation is easured along the x-ray paths using (x,y), to generate 2D projections. For each toography experient, 2D radiographs [Figure 2.3] are taken at any different angular positions. This results in a set of projections that are used for reconstructing a coplete 3D iage representing local attenuation of the saple. Matheatically, this corresponds to inverting the Radon transfor of the projection data. To achieve this, a filtered back projection algorith was 7

20 used. The nuber of projections has to be sufficiently high to obtain good spatial resolution of reconstructed iage which depends on x-ray optics, scintillator, exposure tie, and saple type. Ring artifacts are reduced by using flat-field corrected iages. The MicroXCT series x-ray transission 3D toographic icroscope developed by Xradia, Inc. ( was used in this research as shown in Figure 2.4. The MicroXCT is a transission-type full field iaging x-ray icroscope. The saple tower allows translations in x,y,z directions and rotation ( ) up to 175 o. The axiu saple size this testing syste accoodates is a cube of 75 and rotation stage liits saples having weight less than 4.5 kg (9.9 lbs). In this research, the x-ray source was positioned at 40 fro the rotational axis and a theroelectrically cooled scintillation CCD detector was used 20 fro the rotational axis. Two agnifications of 4x and 20x with a field of view of 5 x 5 and 1 x 1 were used in the present study. As illustrated in Figure 2.4, the x-ray iages data were acquired by MicroXCT X- ray bea penetrating the saple on a rotating saple stage. High resolution was obtained by axiizing the geoetric agnification of the object on the detector with reduced blur [10]. The flat-field corrected toographic raw iages were reconstructed using grayscale 8-bit iage at 1x and 4x binning with the absorption values ranging fro 0 to Fiber Volue Fraction The areal ethod, where the nuber of black and white pixels (after suitable thresholding) within the specified region of interest (representing the atrix and fiber, 8

21 respectively) were counted using a coputational algorith to deterine the fiber volue fraction by integrating the inforation fro a stack of iages. Since the area associated with each pixel is equal, the fraction of the total pixels for each constituent aterial becoes the fraction for that constituent [15]. Two iage processing progras, IageJ and Iage-Pro Plus, were used for analysis for the reconstructed slices of the coposite saples. IageJ iage processing progra was used to create an.avi file of the reconstructed slices. IagePro Plus was used to calculate volue fraction of the fiber content using series of reconstructed slices of the GF/PP saple. The rule of ixtures was used to deterine the fiber volue fraction for each binary reconstructed slice: Vi c 100% (2.2) V V i where V i = volue fraction of fiber content V = volue fraction of polyer atrix V f = volue fraction percentage of fiber to polyer atrix 2.3 Experiental results Fiber Length Distribution, 3D Reconstruction, and Fiber Volue Fraction Figure 2.5a shows a typical radiograph taken at a agnification of 20x for the slow-filled GF/PP coposite saple with 0.92 field of view taken at 1x binning. It is a transission radiograph for a saple with a size of 0.5 x 0.5 x 0.8 obtained fro the core layer of the saple. The high contrast between the individual fibers can be seen in this radiograph. Figures 2.5b and 2.5c displays 2D reconstructed slices beneath the surface of the saple where the individual fibers clearly show the 9

22 orientation and the geoetric arrangeent of chopped fibers fro the polish side through the thickness of the saple. The high concentration of fiber bundles shown in Figure 2.5b indicates high strength properties for the GF/PP saple. As shown in Figure 2.5b, soe ring artifacts fro reconstruction can also be seen despite the use of filtering during reconstruction. The elliptical features in Figures 2.5b and 2.5c indicate the fibers are found to be curved in and out the x-y plane view potentially highly stressed, and ay be initiation sites for fracture resulting fro conditions used in the injection olding process corresponding to the slow-fill rate. Figure 2.6a shows a typical 2D reconstructed slice of a fast-filled GF/PP saple with high concentration of fibers in the cross section. GF/PP 2D reconstructed slices with a 2 x 2 view area were thresholded and filtered using custo algorith ipleented in Iage-Pro Plus iaging software an exaple shown in Figure 2.6b. In this process, gray scale iages of fibers and resin are transferred to binary for quantitative iage analysis. Figure 2.6b shows that reasonable representation of gray scale iages were obtained with autoated threshold value of intensity to obtain binary iages. Subsequently, using these binary iages, fibers were counted and analyzed as shown in Figs 6b and 6c. Figure 2.7 shows the fiber length distribution (FLD), obtained fro twenty slices, for the fast filled GF/PP ISO-plaque coposite saple at location A, which appear to favor a shorter length resulting fro the plastication process in injection olding probably due to fiber-atrix and fiber-fiber contact, where the brittle behavior of E-glass fiber are prone to breakage [16]. Most fibers are less than 4 in length, indicating that the process conditions associated with fast filled voluetric rates have a 10

23 high ipact on the degradation of fiber length. The ean average fiber length was 0.9, significantly less than the typical critical length value of 1.8 for GF/PP [4]. However this is consistent with prior research [17] that as the fiber volue content of short glass fibers increases the ean length decreases. The noralized weight was deterined by easuring the relative weight of the easured fiber as a fraction of the original pellet length. The fiber lengths were counted in a length range (e.g. 4-5 ). Using a noinal pellet length of 12 and 4.5 long fiber would place 4.5/12 = 0.38 in the 4-5 length range [6]. The fiber count in the region of interest typically yielded 170 to 300 fibers fro a single reconstructed slice. Figure 2.8 displays the fiber area fraction variation for a fast-filled GF/PP saple (2 x 2 x 3). An area of interest (700 pixels by 700 pixels) siilar to Figure 2.6b was selected for 700 reconstructed slices to siulate a volue eleent size of 700 x 700 x 700 voxels to calculate the volue fraction of the GF/PP saple. Using a custo developed iage processing algorith the analysis of the reconstructed slices for this saple, the fiber area percentage was found to vary fro 40% to 47%. The slow-filled GF/PP was siilarly analyzed and showed a uch higher variation of fiber volue fraction fro 35% to 59%. Figures 2.9a and 2.9b show a three diensional reconstruction cross section of a slow filled GF/PP ISO-plaque coposite saple at location C that has been converted to a binary iage for quantitative analysis. Beneath the surface, the fibers can be seen luped together indicating potentially high stress concentrations if this saple were to be subjected to external echanical loading. The high concentration of fibers locally is 11

24 expected to increase the odulus/strength of the aterial and resulting stress concentrations in that region. The broken fiber ends indicate areas of a weaker surrounding atrix and that debonding will occur under tensile load [18]. Our future planned experients to view fracture of individual E-glass fibers and atrix under tensile loading using in-situ x-ray iaging approach should confir such hypothesis and could prove to be very useful for developing suitable process-property relationship for chopped fiber coposites. At present tie, very sall cracks (saller than 3 ) in the fibers and atrix can not be located due to liitation of geoetric agnification of 20x Effective Moduli Hashin and Christensen [2-3] proposed a siple atheatical odel which uses coposite sphere analogy to deterine the effective properties of coposites with binary phases (resin and fiber). The ost effective way to utilize such a odel has been to average icroscale effects and characteristics to predict acroscopic behavior [19]. A representative volue eleent coprises of a spherical inclusion surrounded by a atrix phase which is ultiately encopassed by an equivalent hoogenous ediu. This odel has been reported to provide reasonably good predicative results for odeling the echanical properties of fiber reinforced coposites [3]. In our research, by selecting a voxel inside the 3D reconstructed saple of the coposite, the localized constituent properties now can be analyzed using an analytical odel, such as the one proposed by Hashin. In this study, we used the spatial distribution of the 3-phase odel that uses fiber volue fraction to deterine effective bulk and shear odulus for a given nuber of voxels. 12

25 There are several advantages to using the 3-phase odel as outlined by Christensen including a highly localized interfacial shear stress and bulk odulus deterination [3]. The odel covers the entire volue fraction range in the sense that the spherical inclusions can exist at any level, 0 c 1. A very close prediction of experientally easured effective uni-axial odulus up to a 50% volue fraction of inclusions (c = 0.5) was deonstrated by Christensen [3] for E-glass icrospheres in a polyester atrix. The effective bulk odulus, k, was deterined based on a non-dilute elastic suspension of spherical particles. The bulk odulus has a displaceent condition iposed on a single coposite sphere with upper and lower bounds. Hence, the spherical inclusion is treated as a displaceent field in a representative volue eleent based on the theore of inial potential energy. This procedure will allow calculating the localized bulk odulus in the volue eleent. The bulk odulus is derived as shown in Eq. 3, where k i and k are the inclusion and atrix bulk oduli respectively, is the shear odulus of the atrix, and c is the volue fraction of inclusions [3]. k k 1 (1 c) ( k c( k i i k k ) ) k 4 3 (2.3) As shown in Eq. 4, the effective shear odulus of coposite aterial is obtained fro the positive root where paraeters A, B, and C are defined in the Appendix (See A.1). A 2 2B C 0 (2.4) 13

26 The purpose of Eq. 4 is to calculate the effective shear odulus since the inclusion is coposed of two aterials and does not experience a unifor stress state. [3]. The three phases of the odel is the fiber (inclusion), fiber interface with the resin atrix, and the surrounding resin representing the hoogeneous ediu. Figure 2.10 illustrates the process of selecting a representative volue eleent fro a three diensional reconstruction, the above describe ethod of using 2D reconstructed slices was ipleented. Several local voluetric eleents fro x-ray toography reconstruction iages were selected fro two GF/PP coposite saples at different fill rates to deterine the fiber volue fraction variation. As shown in Eq. 5, for each volue eleent, the echanical properties were deterined in finding the effective bulk, shear, and Young s odulus, E, where E 9k E 3 k (2.5) Figures 2.11a and 2.11b show a grayscale reconstruction iage of the slow-filled GF/PP saple having a cross section of 0.5 x 0.8 thresholded using the Aira 3.0 volue visualization iaging software. Figure 2.11c shows the variation of c and the corresponding effective Young s odulus of the selected volue eleents in Figure 2.11b. Areas of interest siilar to the iages shown in Figures 2.11a and 2.11b with the voxel size of 100 x 100 x 74 were used to calculate the fiber volue fraction representing the volue eleent. Figure 2.12 shows the volue fraction variation and effective odulus for E-glass chopped fibers ebedded in polypropylene atrix for two fill rates for eight areas of interest. The fast filled GF/PP coposite saple is located in location 14

27 A on the ISO-plaque as shown in Figure 2.1a. This saple had a fiber volue fraction E range of 0.39 c and an effective odulus range of The slow- filled GF/PP coposite saple is located in location C on the center-gated disk shown in Figure 2.1b. The fiber volue fraction ranges were 0.35 c and effective E odulus range of The fast filled saple had a sall fiber volue E fraction and strength variation between fiber and atrix through the thickness of the saple. The slow-filled coposite has a wider range of fiber volue fraction variation and effective odulus indication stronger and weaker interface between fiber and atrix through the thickness of the saple. A Poisson ratio value of 0.21 was assued for E- glass spherical inclusion and 0.32 for polypropylene atrix [20]. The results shown in Figure 2.12 indicate that the elastic odulus increased with fiber content for both fastfilled and slow-filled saples. Based on the results, additional factors such as yield stress and fracture toughness are known to increase with glass fiber content [18]. In addition, the strain to failure will decrease with increased aount of fibers, although the increased brittle behavior of the coposite is due to a higher aount of E-glass fiber content [16]. Using a 3-phase analytical odel, the effect of the fiber volue fraction is related to an increase in echanical properties as shown in Figure The advantage this odel allows the study of the icrostructure of the coposite for local areas of interest, providing a good prediction of echanical properties including the effective shear and elastic odulus based on fiber volue fraction, c. E 15

28 2.4 Conclusion A novel experiental technique was developed to evaluate the icrostructure of chopped fiber polyeric coposites, ade of polypropylene resin with E-glass fiber using a high resolution x-ray toography syste. A ethod was ipleented to characterize the icrostructure quantitatively using fiber volue fraction fro three diensional x-ray iages. Using a three phase analytical odel and spatially resolved icrostructural features, spatial variation of elastic odulus coposite saples are predicted. Using x-ray toography, actual variation of fiber volue fraction for local regions of interest fro the bulk coposite specien were easured along with spatial variation of oduli. The effect of the fiber volue fraction is directly related to echanical properties. The slow filled GF/PP had a higher fiber volue fraction variation than fast-filled. This research deonstrates the use of high resolution in nondestructive techniques to evaluate echanical properties as a function of coposite aterials icrostructure. In the future, scale-dependent constitutive odels for coposites aterials fro such easured data is expected to provide iportant insight into optial process conditions and final properties of chopped fiber polyeric coposites. 16

29 CHAPTER 3 IN-SITU DAMAGE EVOLUTION OF POLYMERIC COMPOSITES USING HIGH RESOLUTION X-RAY TOMOGRAPHY, DIGITAL MICROSCOPY, AND LARGE CHAMER SEM This chapter is a slightly different version of a paper with the sae title that will be subitted for the Journal of Coposites Science and Technology in 2009 by Stephen A. Young, Robin Woracek, Dayakar Penuadu, Jaret Frafjord, Ashley Stowe, and Vlastiil Kunc. My priary contributions to this paper include: (i) developent of the proble into a work, (ii) identification of the study areas, (iii) gathering and reviewing of literature, (iv) sapling, processing, and analyzing reconstructed x-ray iages fro x-ray radiography, digital icroscopy and scanning electron icroscopy, (v) pulling various contributions into a single paper, (vi) ost of the writing. Abstract Polyeric coposites have been of particular interest in the autootive field due to their low weight, high strength, and resistance to environental degradation. In this research, high resolution x-ray iages of chopped glass fiber tensile coposites are obtained with a high spatial resolution of 3 using a unique in-situ tensile testing syste. The radiographs were reconstructed to visualize the daage evolution as a function of applied on the coposite aterial in three diensions with an ability to 17

30 analyze icrostructure including cracks, fiber orphology, and interfacial bonding. Insitu tensile testing of chopped glass fiber and continuous carbon fiber are evaluated using a chaber SEM icroscopy at agnification not possible using x-ray icroscopy to study icrostructure in great detail. Fractures in the chopped glass coposite were observed for x-ray icroscopy showing the doinant failure echanis of the saple is low interfacial strength and adhesion between the fiber and atrix. Cracks were not observed until after failure by fiber pull-out using the digital icroscopy ethod. Using SEM icroscopy ethod, resin cracking and fiber debonding was observed for a carbon fiber with vinyl ester resin while under tensile loading. Iportant icro-structural inforation relationship with and echanical behavior including variation odulus, yield and ultiate strength are discussed. 3.1 Introduction The Part 1 of the present work [23] dealt with developent of a novel technique to evaluate the icrostructure of chopped polyeric using x-ray coputed icrotoography (Micro-XCT) and deterine the spatial resolved echanical properties of the coposite. A three phase analytical odel and icrostructural spatially resolved icroechanical features were evaluated as a function of the fiber volue fraction to predict the echanical properties of the coposite. In this study, a unique in-situ tensile testing syste, developed by the authors, integrated 3D icrostructural easureents fro toography to evaluate spatially resolved coposite echanical properties. Past research has not been perfored on evaluating in-situ deforation and daage behavior of E-glass and polypropylene chopped fiber coposites considering 18

31 detailed icro-structure inforation ainly due to a lack of precise non-destructive easureent technique. Tensile fractography analysis has indicated the tensile fractures in the coposites are doinated by interfacial failure [24]. Lindhagen and Bergund [25] found using in-situ icroscopy of glass fiber that cracks in the coposite specien result fro high areas of stress concentration and ajor daage points of initiation were transversely oriented fibers and fiber bundles. In-situ SEM icroscopy has shown in good detail the interfacial adhesion of the polyer atrix onto the fiber. Using electron icroscopy ethods such as low voltage, high resolution scanning electron icroscopy (LV-SEM) has shown different icrostructural behavior of polypropylene coposites tend to have ore brittle behavior which ay be accounted for in ters of the tiedependence of daage developent [26]. Polarized light icroscopy has been used to observed daage zones of E-glass and polypropylene to successfully exaine failure echaniss such as crack propagation [27]. Digital video icroscopy echanical testing has given access the plastic response of polyers under uniaxial tension perfored locally at the center of the neck for dogbone polyeric aterials [28]. Although digital icroscopy and SEM ethods have been used to visualize the surface of the polyeric coposites surfaces to study the deforation of glass fiber coposites, these techniques require destructive techniques deforing the saple to failure state of strain in order to obtain digital iages of the fibers beneath the coposite surface [25, 27]. MicroXCT iaging is becoing an iportant non-destructive approach due to its ability to provide a detailed three diensional visualization of the polyeric coposites 19

32 for subsequent quantitative analysis to obtain fiber structure and orientation [29]. Micro- XCT has been used to obtain direct observation and easureent of fiber length, width, and volue distribution for fiber-reinforced polyeric coposite however in-situ tensile testing has not been used. [30]. X-ray scattering and optical icroscopy has been used to study the deforation echanis for polypropylene where the deforation behavior is thought to depend on crystal phase, spherulite size and laellar arrangeent [31]. In addition, x-ray scattering using stretch hold techniques have been unsuccessfully able to observe icro-features such as stress relaxation [32]. There is a growing interest in carbon fiber with vinyl ester resin (CFVE), a continuous fiber polyeric coposite, due to its superior echanical properties for naval applications, relative ease of fabrication using VARTM technique, and its resistance to environental degradation. The icrostructure of CFVE were also evaluated in this research as part of the ongoing research sponsored by the United States Office of Naval Research (ONR). Shivakuar has predicted elastic behavior based on the siple icroechanical equations of the CFVE coposites with experiental data validation with good agreeent [33]. In-situ observations of carbon fiber coposite speciens using SEM has shown successfully the onset of failure showing that the interfacial failure is the doinant failure echanis for this aterial [34]. In our study, a vacuu-suitable echanical testing syste was fabricated to perfor unique in-situ studies on polyeric continuous coposite saples using high resolution x-ray iaging digital icroscopy and large chaber scanning electron icroscope (LC-SEM). 20

33 3.2 Materials Coposite Speciens The aterials and processing of the pellet injection olded polypropylene resin having chopped E-glass fiber (GFPP) are described in Part 1 [23]. The CFVE coposite specien was ade of carbon stitch bonded fabric designated by LT650-C10-R2VE supplied by the Devold AMT AS, Sweden. This was an equibiaxial fabric produced using Toray s Toraya T700 12K carbon fiber tow with vinyl ester copatible sizing. The T700 fiber had a tensile strength at 4.9 GPa, a tensile odulus of 230 GPa and elongation of 2.1%. The atrix used a Dow Cheical DERAKANE 510A-40, a broinated vinyl ester, resin and coposite aterial was fabricated using the VARTM process. The fiber volue fraction was found to be 58% by the area density ethod and includes 2.2% weight of polyester stitch [33]. The dogbone saples were cut fro 45 o oriented CFVE aterial and achined to for the saple as shown in Figure Experiental Procedure and Results Tensile Testing The coposite dogbone saples were loaded according to three unique experiental set ups discussed later in this paper. Loading was introduced by eans of custo developed in-situ echanical testing systes under displaceent control where loads were onotonically increased until failure. Digital iages were captured using a large depth of focus digital icroscope in an attept to observe cracking in gage length section the saple while under tensile loading. The load, displaceent, and strain data were continuously recorded until the saple failed. The Young s odulus was 21

34 deterined fro the linear part of stress-strain curve and the yield strength was deterined using a 0.2% offset ethod. Fig 3.2a and Figure 3.2b show a typical dogbone coposite saple and corresponding predicated stress concentration distribution of the saple under uniaxial tensile loading using ABAQUS Finite Eleent Analysis. As shown in Figure 3.2b, the stress concentration is greatest at the center of the neck of the dogbone coposite saple Micro-XCT In-situ Mechanical Testing Syste The benefits of the use of Micro-XCT for fiber characterization in a nondestructive anner are described in Part 1 [23]. The MicroXCT series x-ray transission 3D toographic icroscope developed by Xradia, Inc. ( and echanical tensile testing syste was used in this research as shown in Figure 3.3. The saple tower with in-situ tensile syste allows translations in x,y,z directions and rotation ( ) up to 160 o. In this research, the x-ray source was positioned at 40 fro the rotational axis and a theroelectrically cooled scintillation CCD detector was used 20 fro the rotational axis. A agnification of 4x with a field of view of 5 x 5 were used in the present study. As illustrated in Figure 3.3, the x-ray iages data were acquired by MicroXCT X-ray bea penetrating the saple on a rotating saple stage. The flat-field corrected toographic raw iages were reconstructed using grayscale 8-bit iage at 1x binning with the absorption values ranging fro 0 to 255. Two fast-filled GFPP saples were loaded using the in-situ echanical tensile syste experiental set up with actuator and N (300 lb-f) load cell shown in Figure 3.3, where two tensile loading techniques were ipleented for the GFPP 22

35 saples. For the AF3D-1 saple, radiographs siilar to Figure 3.4a were taken at 5 seconds exposure tie, under displaceent control in 22.2 N (5 lb-f) increents, while under tensile loading until the saple failed. For the AF3D-2 saple, radiographs were taken continuously while under tensile loading until a crack initiation site was observed siilar to as shown in Figure 3.4b. The saple was then held at a constant displaceent to collect toography projections of the AF3D-2 saple taken on the Xradia syste using 1000 projections in a 160 degree view angle over a period of tie of three hours. Toographic iages were reconstructed using TXM Reconstructer software to reduce data fro raw iages to sinogras to reconstructed iages using a parallel bea algorith, to view the fiber orphology of the tensile saple beneath the surface The reconstructed iages were analyzed using Kitware Volview. Following toography collection the saple was loaded to failure. In Figure 3.5 it can be seen that the state of stress corresponding to failure (load to cause failure) is substantially different between the tensile saples for the tensile testing techniques discussed above. Figure 3.5 also shows salls changes in the initial stiffness (slope of the stress-strain curve) corresponding to the reported tensile Young s odulus (Table 1) for the two loading techniques applied, indicating loading rate on the interpreted odulus was inial. Strain to failure decreased significantly with the use of the toography indicating a brittle character and weakening of the atrix. For all practical purposes, these data are expected to deviate significantly fro acroscopic stress-strain behavior of GFPP fro the techniques used for the x-ray toography testing due to the effects of stress relaxation in the coposite saples. 23

36 Relaxation is expected for these aterials and holding the displaceent constant for a period of tie prior to unloading allows for dissipation of relaxation displaceents. Fro Figure 3.5 it can be shown using the above described loading techniques that stress relaxation significantly changes the icrostructure substantially weakening fiber and atrix interface. Saple AF3D-1 experienced shorter relaxation periods than AF3D-2, however the stress-strain curve clears shows the decrease in strength due to the radiographs where saple AF3D-2 decreased 37%, significantly weakening the icrostructure of the saple during the toography collection. This stress relaxation behavior suggests that shear stress in the atrix near the fiber breaks relax, having distribution effect stress profiles in neighboring fibers. This daage accuulation would in a general sense be siilar to that of debonding [35]. Figure 3.4a shows a typical transission radiograph taken at a agnification of 4x for the fast-filled GFPP coposite saple with 4.69 field of view. Figure 3.4b and Figure 3.4c show the radiograph of the saple during loading and after failure. The high contrast between the individual fibers can be seen in these radiographs indicating a large concentration of fibers in the saple. Although noticeable fracture in the saple could be observed during the loading as shown in Figure 3.4b, very sall cracks (saller than 3 ) in the fibers and atrix can not be located due to liitation of geoetric agnification of 4x. However, by rotating the stage the failure of the saple can be clearly observed as shown in Figure 3.4c. The failed saple in Figure 3.4c shows typical GFPP coposite fiber pull-out behavior, evidenced by a high concentration of exposed broken fibers indicating the glass fibers were perpendicular to the coposite during 24

37 loading. In addition, bent fibers also shown indicate a decrease in fracture resistance in the GFPP coposite saple [36]. Fig 6a displays a typical 2D reconstructed slice beneath the surface of the saple where the individual fibers clearly show the orientation and the geoetric arrangeent of chopped fibers through the thickness of the saple. The high concentration of fiber bundles shown in Figure 3.6b indicates crack initiation sites for the saple [25]. As shown in Fig 6b, 3D reconstructed fibers of the GFPP can be observed yet the contrast between the fiber and atrix was not great enough to threshold the iage accurately to deterine the echanical properties such as the fiber volue fraction. It can be concluded that the advantage of in-situ echanical syste using attenuation based x-rays for GFPP coposites provides a nondestructive technique to visualize the fiber orphology and ability to predict zones of fracture. For future experients, a higher optical agnification is required (500x-1000x) to view the cracks and daage evolution of the coposite. This will allow observation of echanical deforation including fiber pull-out, fiber volue fraction redistribution, crack propagation, crazing, and fiber debonding. Also the duration of tie during which tie the displaceent was held constant corresponding to toography collection, will need to be significantly reduced to prevent a significant stress relaxation effects on the coposite saples In-Situ Mechanical Tensile Testing Syste using Digital Optical Microscopy The Keyence VHX-600 digital icroscope developed by Keyence, Inc. ( was used in this research. This digital icroscope affords the 25

38 ability to view in real-tie high resolution key features of interest on the surface of the coposite saples by using color CCD caera high density pixels. Two GFPP (one fast-fill, one slow-filled) and two CFVE tensile saples were loaded using in-situ echanical tensile syste with an axial force capacity of 90 kn. The testing syste uses custo developed LabView based data acquisition and control software for perforing both stress and strain controlled tests. A very precise sall load cell with full scale capacity of 111 N (25 lb-f) was used to carefully perfor the tensile testing. An.avi video file, using Keyence digital iage software, recorded the saples during tensile loading at a continuous rate until noting an abrupt drop in their aplitudes. At high agnification (500x-1000x), cracks could not be observed until the saple failed. Typical experiental results for the GFPP shown in Figure 3.7 indicate brittle behavior and sall changes in the initial stiffness corresponding to tensile Young s odulus for both fill rates. The state of stress corresponded to a failure decrease (engineering strain = 19%) for slow-filled GFPP copared to the fast-filled GFPP coposite. This difference in echanical failure ay be attributed to anufacturing process effects on slow and fast-filled injection olded coposites found in Part 1 [23]. The interpreted odulus suggest no significant changes in reported acroscopic stressstrain behavior of GFPP [37]. The tensile strength, strain to failure decreased with the use of the slow-fill rate indicating that a ore brittle behavior when copared to the fastfill case. It is interesting to note the considerable difference in echanical behavior between the two saples after failure. The fast-filled coposite (AF3D-3) tend to fail abruptly indicating fiber pull-out and the slow-filled coposite (AS3I) appears to show 26

39 fiber debonding behavior corresponding to point C as shown in Figure 3.7. Due to the external stress applied to the slow-filled GFPP coposite, the interfacial debonding is nonlinear due to the effect of the Poisson contraction of the fiber, subjected to uniaxial tension [38]. As shown in Figure 3.7, point D on the stress-strain curve for the slowfilled dogbone saple indicates that cracking of the atrix occuring in the fiber bundles oriented at other angles to the applied stress. This rapid change in the stress-strain curve slope is due to cracking between fiber bundles. As shown in Figure 3.7 it can be concluded for the tensile test data that the fast-filled GFPP coposite saple indicates a higher strength but weaker interface, and the slow-filled GFPP coposite is a tougher aterial however at the expense of strength. Figure 3.8 shows a typical GFPP coposite saple after failure. The coposite failed due to debonding, fracture, and pullout behavior which these failure echaniss indicate shear yielding and plastic deforation in the coposite [39]. The fibers are randoly dispersed without preferred orientation. In the daage zones, the doinant echanis is poor adhesion where the cracks toward the interface obstacle by the fibers. Hence, less energy is required to pullout due to poor adhesive interfacial strength. These observations agree with prior research that E-glass has a very poor interfacial adhesion for polypropylene [40]. Fiber breakage observed in the fracture zone where failure initiation in highly stress concentration areas, indicating the location of local crack propagation toward the end of the glass fibers. This behavior is consistent with prior research that glass fiber ends location for crack initiation sites [27]. In addition, the GFPP appears to have a poorly filled resin where the presence of cavities of the surface 27

40 can be seen. The failure pattern of the coposite saple indicate the following: i) the echanical failure can be attributed to the existence of longer fibers which restricts the atrix oveent. ii) The individual fibers and atrix fail by fiber debonding and pullout as can be observed. These largely sooth and clean glass fibers resulting fro the fiber pull-out indicates poor adhesion and weak interface between the fiber and atrix [37]. iii). Since the interface of the fiber and atrix is weak then this is a shear stress concentration parallel to the fiber and atrix causing interfacial shear debonding [41]. iv). The fractured fiber ends and fiber debonding viewed in Figure 3.8, indicate regions here where the stress is largely concentrated, where near fiber ends, debonded areas of the interface of the fiber and atrix are fored at relatively lower levels of stress than the other regions of the coposite [42]. v). Also, the bundles of fibers shown indicate factors such as low wettability and fiber-fiber contact during injection olding contributed to the failure. It is worth addressing the low adhesion between the fiber and atrix interface for the GFPP coposite specien used in the present study. The failure behavior of the coposite indicates poor interface shear strength in the saple. The void content and low interfacial strength ay explain the lower odulus observed for slow-filled GFPP coposite since it has significantly higher and low stress distribution copared to the relative consistent strength of the fast-filled saple. The injection olding teperature should be increased to decrease the void content which would decrease the fiber pull-out sites. In addition, decreasing the pellet fiber length fro 12 ay iprove the echanical properties of the coposite specien. Though it is intuitive to increase the 28

41 fiber lengths to iprove the echanical properties of the GFPP coposite specien, the 12 long pellets in injection olding used in this paper have a bending tendency during plastication, which leads to a decrease in tensile strength and odulus [42]. An increase in tensile strength and odulus was found to be effective up to 9 pellets for GFPP coposite aterial [43]. Using a continuous loading technique for the GFPP coposites significantly increase the odulus and yield strength as shown in Table 1. Although the surface of the coposite provided good detail of the echanical failure properties using digital icroscopy, this ethod is liited to two diensional observations. Copared to x-ray iaging, we are unable to view the fiber orphology beneath the surface to qualitatively evaluate changes in the icrostructure of the glass fiber until after failure. In addition, although the icroscope had 1000x optical agnification capability, it did not offer any unique observations above 500x. Hence, an iproved light source is needed to view cracks that develop on the surface while under tensile loading. Future planned experients will address these liitations by the use of SEM with high optical agnification to view the surface in high resolution. Typically, the strength of carbon fibers are approxiately twice that of glass fiber, however lower elastic odulus and tensile strength is expected for 45 o oriented dogbone saples where the deforation echanis is doinated by the effect of flaws in the atrix leading to fiber pull-out [44]. Figure 3.9 and Table 3.1 shows typical tensile stress-strain curves of CFVE coposite saples, where the saple exhibit linear stressstrain behavior with plateau region indicating a progressive rather than catastrophe failure. 29

42 The stress then decrease rapidly after the plateau region for all three saples. After which the load drops with increasing displaceent, an indication of brittle behavior, and finally fors a long tail due to fiber debonding and pullout. The oduli of the stress-strain curve for all three saples are close, although CFVE-2 has a higher proportional liit and a higher failure strain. CFVE-1 and CFVE-3 has siilar ultiate strengths and ay be lower than CFVE-2 due to anufacturing defects such a achining and curing, which leads to strength degradation. In addition, the curves exhibit curves exhibit a plateaulike shape beyond the proportional liit and the linear curves typical of brittle behavior. These data suggest that a rather significant weakening of the interface bond between the fiber and atrix fro the alignent and anufacturing, which correlates fiber degradation of the fibers. The stress-strain curve of CFVE-2 saple exhibits a stronger and tougher behavior than CFVE-1 and CFVE-3 saples. This observed increase in strength of CFVE-2 infers the load can still be transferred effective fro atrix to carbon fiber, assuing its interfacial bond strength between fiber and atrix to be appropriate. The saples CFVE- 1 and CFVE-2 using the digital icroscope (CFVE-3 is discussed in the next section) siilarly to the GFPP coposites, cracks could not be observed during the tensile testing fro the viewing angle shown until after the saples failed. In Figure 3.10 and Figure 3.10b, typical failure pattern of the CFVE saples show several fibers are present during the pull-out because the high concentration of the fibers which leads to daaged fibers that are perpendicular to crack propagation directions [45]. Figure 3.10a and Figure 3.10b shows the orphology of the carbon fiber, where long fiber pullout is observed 30

43 indicating oderately weak interfacial strength. Furtherore as shown in Figure 3.10a and Figure 3.10b, the sooth and clean fibers indicate a weak adhesion of the vinyl resin to carbon fibers which is not surprising since there are known issues of carbon fiber not adhering to the vinyl resin well [44]. Fig 10a shows local strongly bonded and isaligned fibers indicating there is stress and strain agnification in the atrix which is axiu between the fibers corresponding to the plateau-like region and knees as shown in Figure 3.9. The 45 o fibers are isaligned after pullout, which suggest the fiberatrix bonding ay have been draatically lowered and fibers significantly weakened fro local weak adhesion between fiber and atrix. Beyond the elastic liit, the applied load results in unifor plastic deforation, till the axiu load is reached. As observed in Figure 3.10b, CFVE-1 failure did not fail along the 45 o alignent of fibers, which ay explain the lower bound odulus shown in Figure 3.9. The CFVE-2 saple failed along the 45 o alignent which is typical for this [ ± 45, 2s] lay-up [33]. As shown in Figure 3.10c the fracture shown indicate that fibers aligned in the crack direction and adhesion interfacial failure for this results in a cleavage fracture. This cleavage fracture was cracking of the atrix obstacled by the carbon fibers. Figure 3.10c shows the icrocracks which are fored in the rather narrow stress range corresponding to the plateau-like doain in the stress-strain curve of CFVE-1 (Figure 3.9), run across the thickness of the saple in planes roughly perpendicular to the load axis. Although the surface of the stitched carbon vinyl resin in could be viewed in great detail, the liitation of the polarized light and optical agnification prevented observation of the crack developing for the CFVE coposite saples. The liitations 31

44 included two diensional projection of the video, and liited viewing angle having the inability to rotate the saple to different view angles in order to observe cracking of the CFVE saple in real-tie. For future planned experients, non-destructing testing such as Micro-XCT would be needed to view any possible defects in the CFVE coposite and predict daage zones using failure criterion such as fiber orphology and fiber volue fraction In-situ Mechanical Tensile Testing using Large Chaber Scanning Electron Microscope (LC-SEM) The LC-SEM anufactured by VisiTec ( shown in Figure 3.11 was used to investigate the CFVE-3 saple to obtain icro-structured features of iportance, including the evolution of daage and icrostructural features as a function of echanical stress. The CFVE-3 saple was loaded using the above entioned in-situ tensile testing syste used for digital icroscopy and vacuu suitable for LC-SEM testing. A load cell with a full scale capacity of N (750 lb-f) was used for the tensile testing. As shown in Figure 3.11, the LC-SEM is equipped with the following: secondary electron detector, backscattered electron detector (4 quadrants), energy dispersive x-ray spectroetry (EDS), and variable pressure ode. In-situ LC-SEM affords the ability of testing larger speciens having geoetry and diensions siilar to those used in traditional echanical testing laboratories. The LC-SEM eliinates the need for using artificially sall speciens, reducing unwanted size effects associated with applied deforation on the icrostructure. Deforation echaniss, such as crack propagation, as well as other icroscopic features, such as fiber debonding, and atrix 32

45 yielding were exained in-situ under tensile stress at desired agnifications ranging fro 125x to 10,000x. Siilar to the digital icroscopy ethod, a video using digital iage software was recorded the saple while under tensile loading at constant displaceent rate until echanical failure of the saple. In order to find the echanis of fracture, Figure 3.12(a-d) represents optical icrographs captured while under uniaxial tensile loading. The corresponding loaddisplaceent curve is shown in Figure Figure 3.12 shows gage length area of the CFVE-3 saple where the daage evolution can be qualitatively evaluated. As shown in Figure 3.12a, the white patches are vinyl ester resin and the dark region represented the carbon fiber concentration of the coposite saple. These white patch features suggest these zones has already undergone soe degradation during the curing and anufacturing of the saple thus allowing ore decohesion to occur between the fibers and the atrix. Hence, the white patches are indication of failure initiation and final fracture zones leading to pre-ature fracture during a tensile test [33]. Initial failure is shown in Figure 3.12b, where resin cracking was first noticed at 305 N tensile loading as at point O indicates where the load transferred to the fibers has peaked. Five additional cracks appear on the surface as this crack propagated weakening the fiber and atrix interface causing the crack to at point P to appear as shown in Figure 3.12c. These icrocracks shown in Figure 3.12b and Figure 3.12c suggest the crack propagated or followed along the fiber atrix interface in regions of the dense packing of fibers, where it can be seen that any initial cracks were generated siultaneously where the daage occurs alost instantaneously. This indicates isolated interfacial cracks that grow 33

46 and connect as the tensile load is increased [34]. Furtherore, this cracking fro the applied load is progressively transferred to the fibers up to coposite failure which occurs at a strain. As shown in Figure 3.12c, the atrix icrocracking corresponding to the plateau region indicates that the fibers, which are now largely debonded fro the atrix, carry the applied load alone and the failure occurs. The icrocracking results in abrupt failure of the coposite as shown in Figure 3.12d. Interfacial failure is the doinating failure echanis for this saple and the features are illustrated in Figure 3.12c and Figure 3.12d. These cracks shown in Figure 3.12c and Figure 3.12d indicate that debonding of the fiber fro atrix probably begin before the plateau-like region show in Figure 3.13 is reached. As shown in Figure 3.13, the plateau region indicates that the fiber and atrix interface could not sustain any additional load. The debonding is the Poisson contraction of the carbon fiber against the atrix cure shrinkage, assuing that the fiber did not break first. It is interesting to note since the carbon fibers are aligned 45 o to the tensile axis, the width of the gage length area is decreased fro this alignent approaching 305 N loading, although the axiu load is still in the proportional liit. A possible reason for this failure echanis is due to the local strongly bonding between the fiber and atrix effective carry the shear load across the gage length. The cracking of the atrix occurs while under tensile load and the width of the observed cracks increases with increase in the applied load together debonding and pullout of the fibers fro the atrix as shown in Figure 3.12d. This weakening corresponding to plateau region and long tail can viewed on the stress-strain curve as 34

47 shown in Figure As shown in Figure 3.12d and Figure 3.13, beyond point D the pullout is only under the resistance of friction appear to correspond to the drop following the plateau-like region. It is worth noting, the cracks developed around the white patches confiring these were crack failure initiation sites as predicted. The LC-SEM experiental test successfully shows the daage evolution as a function of the tensile loading correlating the load-displaceent curve and icrographs. As shown in Figure 3.14a and Figure 3.14b, the fracture surface after failure was investigated using LC-SEM to qualitatively evaluate the interfacial debonding. In this case locally sall pullout is observed and atrix appears between the fibers, which ay be inferred that the fiber-atrix interfacial bond is strong locally. Using SEM icrscopy show degraded echanical properties based on a local bonded interface and degradation of fibers due to achining. As shown in Figure 3.14, it can be clearly seen that the interface failed. The daage pattern of the saple is shown where typical fracture surface for a specien with 45 o fiber orientation were caused by fracture icroechanics resulting fro interfacial debonding in the carbon fiber/vinyl ester. The atrix plastically defored locally allowing fibers to bridge the atrix crack ore aligned to the tensile axis corresponding to the plateau region and pullout regions in Figure In addition, the cracks tend to run along the atrix between the fibers which indicate a brittle icrostructure for the coposite specien. Figure 3.14a shows sooth and clean carbon fibers although having local areas of strong interfacial adhesion. This indicate local atrix near the interface between the fiber and atrix failed. SEM evaluation indicates the doinating echanis is 35

48 adhesive failure for weaker interface and cohesive failure for the strong interface. On fracture, corresponding beyond the elastic liit of the load-displaceent curve, the atrix which is porous and cracked gets disintegrated leaving the fibers exposed. The brittle behavior corresponding to the region following point D in Figure 3.13, the local yielding foring the long tail fro pullout as shown in Figure 3.14a. As shown in Fig 14b the fractured surface of an individual fiber using high agnification (10000x) be can observed indicating the brittle behavior of the CFVE saple caused by fiber pull-out. Since the effective load carried by the fibers depends priarily on their strength and effective area, iproveents in adhesion will be needed to provide suitable fiberatrix interfacial bonding for effective transfer of load for atrix to fibers. Due to liitation of one viewing angle, it is unknown whether resin cracking, crack propagations occurred on the other side of the gage length area. Although the cracks developent and icrostructure of the surface of the CFVE saple was observed in great detail, siilar to the digital icroscopy experient, the liitations are the restricted two diensional viewing and destroying the saple in order to view the fiber and atrix interface. Future planned experients would be able to rotate the saple using a rotating stage to view different sides of the gage length area. In addition, the CFVE saples should be coated to prevent charging, lower the use of the voltage (below 1 kev) or test at variable pressure [25]. 3.4 Conclusion Three unique in-situ experiental tensile syste using Micro-XCT, digital icroscopy, and LC-SEM were used to investigate the icrostructure of the E-glass fiber 36

49 with polypropylene resin and carbon fiber with vinyl ester resin coposites while under tensile loading. Changes in icrostructure beneath the surface of the GFPP coposites was able to be observed under loading, however due to geoetric agnification liitation cracks in the atrix and fiber was not able to be observed. Radiograph and toography collection caused stress relaxation which significantly reduce the strength of the GFPP coposites. Using digital icroscopy cracks were not able to be observed during tensile testing. However, key features of interest such as fiber pull-out, fiber bundling, fracture fiber ends were observed fro failure for both GFPP and CFVE coposites. The doinant failure echanis for GFPP was low adhesion between the fiber and atrix. SEM provided unique observation of the stress daage evolution was successfully observed on the surface of the CFVE coposite where crack initiation and crack propagation while the saple was under tensile loading. The doinant failure echanis for CFVE saples was interfacial failure. 37

50 REFERENCES 38

51 [1] Nguyen BN, et al. Fiber Length and Orientation in Long-Fiber Injection-Molded Theroplastics Part 1: Modeling of Microstructure and Elastic Properties. Journal of Coposites Materials 2008; 42: [2] Hashin Z. The Elastic Moduli of Heterogeneous Materials. Journal of Applied Mechanics 1962; 29: [3] Christensen RM. Mechanical Properties of Coposite Materials. John Wiley, New York: 1979a. [4] Matthews FL, Rawlings RD. Coposites aterials: Engineering and Science. Woodhead Publishing Liited, Cabridge: [5] Montsinger LV. Theroplastic Coposite Made by Rotational Shear. U. S. Patent 6,258,453: [6] Velez-Garcia G, Kunc V. Experiental Methods to Evaluate Fiber Length and Orientation Distribuation of Long Glass Fibers in Injected Molded Theroplastics. Fiber Society Conference, Knoxville, Tennessee. October 10-12, [7] Hoayonifar M, Zebarjad SM. Investigation of the effect of atrix volue on fiber stress distribution in polypropylene fiber coposite using a siulation ethod. Materials and Design. 2007; 28: [8] Davidson NC, Clarke AR. Extending the dynaic range of fibre length and fibre aspect by autoated iage analysis. Journal of Microscopy. 1999; 196: [9] Eberhardt CN, Clarke AR. Autoated reconstruction of curvlinear fibres fro 3D datasets acquired by X-ray icrotoography. Journal of Microscropy. 2002; 206:

52 [10] Scott D, et al. High Resolution 3D Toography for Advanced Package Failure Analysis White Paper. Xradia, Inc. Concord, CA. [11] Patcharaphun S, Mennig G. Properties Enhanceent of Short Glass Fiber- Reinforced Theroplastics via Sandwich Injection Molding. Polyer Coposites. 2005; 26: [12] Stock SR. X-Ray Methods for Mapping Deforation and Daage. In Microechanics Experiental Techniques. J.W.D. Sharpe (Ed.) ASME. AMD. 1990; 102: [13] Kak C, Slaney M. Principles of Coputerized Toogographic Iaging. IEEE Press: [14] Banhart J. Advanced Toographic Methods in Mateials Research and Engineering. Oxford University Press, New York: [15] Cann MT, Adas DO, and Schneider, CL. Characterization of Fiber Volue Fraction Gradients in Coposite Lainates. Journal of Coposite Materials 2008; 42(5): [16] Ota WN, Aico SC, Satyanarayana KG. Studies on the cobined effect of injection teperature and fiber content on the properties of polypropylene-glass fiber coposites. Coposites Science and Technology 2005; 65: [17] Fu SY, et al. Fracture behavior of short glass fiber and short carbon fiber reinforced polypropylene coposites. Int. J. of Materials and Product Technology 2002; 17(1/2):

53 [18] Zebarjad SM, et al. Role of the interface on the deforation echanis of glass fiber/polypropylene coposites. Journal of Materials Science Letters 2002; 21: [19] Hashin Z, Herakovich CT. Mechanical Properties of Coposite Materials: Recent Advances. Pergaon Press, New York: [20] Gol dan AY, et al. Prediction of the Deforation Properties of Polyeric and Coposite Materials. Aerican Cheical Society, Washington, DC: [21] Xradia, Inc. MicroXCT User s Manual 2.0 Issue 1.1 (2005). [22] Nguyen BN, Holbery JD, and Kunc V. Engineering Property Prediction Tools for Tailored Polyer Coposite Structures: FY 2005 Progress Report. Oak Ridge National Laboratory, Oak Ridge, TN. Retrieved July 8, 2009 fro < [23] Young S, Penuadu D, and Kunc V. High Resolution X-ray Toography of Chopped Fiber Polyeric Coposites. Coposites Science and Technology (subitted). [24] Liu H, Liao K. Tensile Behavior and Morphology Studies of Glass-Fiber- Reinforced Polyeric In Situ Hybrid Coposites. Journal of Applied Polyer Science 2004; 94: [25] Lindhagen J, Bergund L. Microscopial Daage Mechaniss in Glass Fiber Reinforced Polypropylene. Journal of Applied Polyer Science 1998; 69:

54 [26] Pluer CJG, et al. Microdeforation in Heterogeneous Polyers, Revealed by Electron Microscopy. Macrool. Syp. 2004; 214: [27] Zebarjad SM. The influence of glass fiber on fracture behavior of isotactic polypropylene. Materials and Design 2003; 24: [28] Addiego F, Abdessela D, G Sell C, and Hiver JM. Characterization of volue strain at large deforation under uniaxial tension in high-density polyethylene. Polyer 2006; 47: [29] Stock SR. Recent Advances in X-ray Microtoography Applied to Materials. International Materials Reviews 2008; 53(3): [30] Eberhardt CN, Clarke AR. Autoated Reconstruction of Curvilinear Fibers fro 3D Datasets Acquired by X-ray Microtoography. Journal of Microscopy 2002; 206(1): [31] Davies R. J., et. al. The Use of Synchrotron X-ray Scattering Coupled with In Situ Mechanical Testing for Studying Deforation and Structural Change in Isotactic Polypropylene. Colloid Poly Sci 2004; 282: [32] Stribeck N, Nochel U, Funari SS, and Schubert T. Tensile Tests of Polypropylene Monitered by SAXS. Coparing the Stretch-Hold Technique to the Dynaic Technique. Journal of Polyer Science: Part B: Polyer Physics 2008; 46: [33] Shivakuar KN. Carbon/vinyl ester coposites for enhanced perforance in arine applications. J. Reinf Plast Cop 2006; 25(10):

55 [34] Hobbiebrunken T, Hojo M, Adachi T, De Jong C, and Fiedler B. Evaluation of Interfacial Strength in CF/expoxies Using FEM and In-Situ Experients. Coposites: Part A 2006; 37: [35] Cardon AH. Recent Developents in Durability Analysis in Coposite Systes. Taylor and Francis, New York: [36] Rizov V, et al. Fracture Toughness of Discontinuous Long Glass Fiber Reinforced Polypropylene: An Approach Based on a Nuerical Prediction of Fiber Orientation in Injection Molding. Polyers and Polyer Coposites 2005; 13(2): [37] Zebarjad SM. Investigation of deforation echanis in polypropylene/glass fiber coposite. Journal of Applied Polyer Science 2003; 87(13): [38] Ki JK, Mai YW. Engineering Interfaces in Fiber Reinforced Coposites. Elseverier, New York: [39] Kuar KS, Bhatnagar N, and Ghosh AK. Mechanical Properties of Injection Molded Long Fiber Polypropylene Coposites, Part 2: Ipact and Fracture Toughness. Poly. Copos 2008; 29: [40] Tselios CH, Bikiaris D, Savidis P, and Panayiotou C. Glass-fiber Reinforceent of In Situ Copatibilized Polypropylene/Polyethylene Blends. Journal of Materials Science 1999; 34: [41] Li H. Stress transfer and daage evolution siulations of fiber-reinforced polyeratrix coposites. Materials Science and Engineering A-Structural Materials Properties Microstructure and Processing 2006; 425(1-2):

56 [42] Kuar, KS et al. Mechanical Properties of Injection Molded Long Fiber Polypropylene Coposites, Part 2: Ipact and Fracture Toughness. Poly. Copos. 2008; 29: [43] Kuar KS, et al. Developent of long glass fiber reinforced coposites: Mechanical and orphological characteristics. Journal of Reinforced Plastics and Coposites 2007; 26(3): [44] Wonderly C, et al. Coparison of echanical properties of glass fiber/vinyl ester and carbon fiber/vinyl ester coposites. Coposites Part B-Engineering 2005; 36(5): [45] Fu, SY. Cobined effect of fiber content and icrostructure on the fracture toughness of SGF and SCF reinforced polypropylene. Journal of Materials Science 2002; 37: [46] Sawyer, LC, et al. Polyer Microscopy: Characterization and Evaluation of Materials. Alden Press, Oxford:

57 APPENDIX 45

58 Figure 2.1: Scheatic of the 3 thick pellet injection olding of E-glass fiber/polypropylene (a) ISO-plaque and (b) center-gated disk [1, 22]. Figure 2.2: Scheatic diagra of the skin, shell, and core layers of the E-glass fiber polypropylene olding copound at 50x agnification [6]. 46

59 Figure 2.3: Radiograph of an absorption Radon transfor projection of the E-glass fiber with polypropylene atrix at 4x agnification. Figure 2.4: Illustration of toography raw iage collection. E-Glass fiber with polypropylene atrix coposite saple (0.5 x 0.5 x 10 ) inside saple holder. The X-ray source (100 kv, 10 W) and theroelectrically cooled scintillation CCD detector (2048 x 2048 pixels, 16 bit) at -59 o C with fast readout scintillation crystal. 47

60 (a) (b) (c) Figure 2.5: (a) Radiograph x-ray at 20x, and (b-c) 2D reconstruction slices of the cross section. (a) (b) (c) (d) Figure 2.6: (a-b) 8-bit grayscale toography reconstruction and corresponding thresholded iage of E-glass fiber with polypropylene atrix coposite saple. (c-d) Thresholded and easured E-glass fibers in the area of interest. 48

61 Figure 2.7: Fiber Length Distribution of E-glass fibers in polypropylene atrix. Figure 2.8: Area Density Distribution for fast-filled and slow-filled E-glass fibers in polypropylene atrix. 49

62 (a) (b) Figure 2.9 (a,b) 3D reconstructed iages of the E-glass fiber with polypropylene atrix. Figure 2.10: Flow chart to deterine effective oduli for the E-glass fiber in polypropylene atrix. 50

63 (a) (b) (c) Figure 2.11: (a) 8-bit grayscale toography reconstruction, (b) corresponding thresholded iage and (c) effective odulus for slow-filled E-glass fiber with polypropylene. 51

64 Figure 2.12: Effective odulus, E-glass spheres in polypropylene atrix A.1 3-Phase Model The 3-phase odel is derived fro Christensen and Lo s odel [3]. The solution for the effective shear odulus,, is given by the solution of the quadratic equation shown in Eq. 4 where subscript refers to the atrix and constants A, B, C are defined by the following relationships: A 8 i c 10/ i c 7 / 3 (6) 252 i 1 2 c 5/ 3 50 i 1 ( ) 2 c 4(7 10 ) 2 3 B i i c 5 / c 10 / 3 i (3 i ) 2 1 c ( c 7 / 3 7) 2 3 (7) 52

65 C 4 i c 10 / i c 7 / 3 (8) 252 i 1 2 c 5 / 3 25 i 1 ( 2 7) 2 c (7 5 ) 2 3 with (49 i i 50 i ) 1 35 ( i 2 ) 35(2 i ) (9) i i i (10) (8 10 ) (7 i 3 5 ) (11) Figure 3.1: Scheatic of the 3 thick, 45 o [±45, 2s] carbon fiber with vinyl ester resin and tensile saple. 53

66 Figure 3.2: Scheatic of (a) E-glass fiber/polypropylene dogbone coposite saple and (b) corresponding finite eleent analysis stress concentration distribution. Figure 3.3: In-situ echanical tensile syste with actuator and N (300 lbf) load cell. E-Glass fiber with polypropylene atrix coposite saple (2.3 x 1.4 x ) inside saple holder. The X-ray source (100 kv, 10 W) and theroelectrically cooled scintillation CCD detector (2048 x 2048 pixels, 16 bit) at -59 o C with fast readout scintillation crystal. 54

67 (a) (b) (c) Figure 3.4: Radiograph of transission Radon transfor projection of the E-glass fiber with polypropylene atrix at 4x agnification (a) before tensile loading (b) during tensile loading and (c) after failure of the tensile saple. Figure 3.5: Experiental data corresponding to fast-filled E-glass fiber with polypropylene dogbone saple. 55

68 Figure 3.6: (a) 2-D reconstruction of E-glass fiber with polypropylene resin and (b) corresponding 3-D voluetric rendering after tensile loading. 56

69 Figure 3.7: Typical experiental data for a tensile test for E-glass/polypropylene. Figure 3.8: Fracture surface of E-glass with polypropylene resin after tensile failure. 57

70 Figure 3.9: Typical experiental data for a tensile test for carbon fiber with vinyl ester resin. (a) (b) 58

71 (c) Fig 3.10: Fracture surface of carbon fiber/vinyl ester after tensile failure for (a,b) by fiber pullout and (c) cleavage fracture. Figure 3.11: Experient setup for the in-situ echanical syste in LC-SEM. 59

72 Figure 3.12: Daage evolution of carbon fiber/vinyl ester tensile saple under tensile loading. Figure 3.13 Load-Displaceent daage evolution of carbon fiber/vinyl ester tensile saple under tensile loading. 60

73 (a) Figure 3.14: (a) Daage pattern of carbon fiber vinyl ester and (b) fracture surface of carbon fiber/vinyl ester fiber after tensile failure. (b) 61