#21945 - DTIC

patterns are illustrated in Figure I. The damage pattern for the BF glass (Figure la) shows alimited number of very distinct cracks; SP glass (Figure I b) exhibits a network of interconnectedcracks. The damage pattern and extent differs. although the same thermal shock procedure wasemployed.3.2 Confined Compression TestsMost compression experiments were conducted at quasistatic strain rates (- 1o-3 s·' ). Severaltests were also performed at higher strain rates to assess strain rate dependence. These wereconducted on pre-damaged specimens using a Hopkinson (Kolsky) bar setup. The quasistaticand intermediate rate (- 1 s' 1) tests were conducted using an MTS servohydraulic machine.Specimens were confined using either a mechanical or hydraulic confinement technique.With the hydraulic confinement technique. a constant confinement pressure is maintainedduring testing. The hydraulic confinement test is a triaxial compression test, commonly used tocharacterize pressure-dependent materials (e.g., sand or concrete) [2,24].A maximumconfinement pressure of 500 MPa is possible with our current setup. Tests were performed atfluid pressures of 25, 50, I 00, 250, 400. and 500 MPa. Each glass specimen was placed in aTeflon shrink tube sleeve for protection from the hydraulic fluid. A piston was used to load thespecimen inside the pressure vessel using two ceramic anvils. The load was measured with aload cell placed inside the pressure vessel and wired directly to provide the equivalent stressacting on the specimen.The axial strain in the specimen was measured using a calibratedextensometer.For mechanical confinement, a 3.2-mm thick high-strength steel sleeve was used to confinethe specimen during testing. Individual confining sleeves were honed to fit each test specimen.6

and minimize the clearance between the test specimen and the inner diameter of the sleeve. Thesleeved specimens were loaded in displacement control. The confinement pressure changesduring compressive loading of the sleeved specimen, and is dependent on the extent of loading.The maximum confining pressure achieved was approximately 1 GPa. Load was measured withthe load cell on the MTS loadframe. Axial strains in the specimen were measured with anextensometer, with arms situated on the loading platens. Strain gages, mounted on the midsection of the confining sleeve, were also used to measure axial and hoop strains in the sleeve.Additional details for the confined compression experiments and techniques are described inRefs. [8, 18,25].3.3 Post-Test Specimen EvaluationSpecimens were evaluated to determine the extent of damage, with increasing load andpressure levels. and the extent of failure.Careful specimen preparation and hand ling wasnecessary to ensure the fracture/failure characteristics were captured without imparting furtherdamage to the specimens, or disrupting loose or interlocking glass fragments.Initial investigations were performed with optical and stereomicroscopy. Higher resolutionmicroscopy, using a scanning electron microscope (SEM). was performed on select BF and SPspecimens with damage features of interest.These were selected based on the opticalmicroscopy evaluations. The SEM investigations were implemented to provide more detailedanalyses in the vicinity of the dominant shear plane. including the morphology of the glassfragments along the resulting shear plane.SEM evaluations required application of a goldcoating to the g lass specimens to limit specimen charging under the electron beam.7

To view the damage in-siLu, it was necessary to remove the steel sleeve or Teflon holderaround the specimen. The technique for removing a section ofthe steel confinement sleeve wasperfected in previous investigations [18]. A longitudinal section of the steel sleeve was carefullyremoved to minimize disturbance to the tested specimen. Removal of a ·'pie section" from thesteel sleeve, as shown in Figure 2a, allowed viewing of in-situ damage along the entire specimenlength. Internal damage was readily visible owing to the transparency of the glass. Cutting ofthe steel sleeve was initially performed with a Dremel tool; electro-discharge machining (EDM)was implemented subsequently to obtain precision cuts with minimal specimen damage.Numerous confined sleeve specimens were evaluated. as the location of the opening did notalways correspond with the dominant shear plane location.For specimens tested in triaxial compression. slitting and opening of the Teflon sleeve, andunderlying brass foil. was necessary to view specimen damage. The thin brass foil was wrappedaround the specimen to prevent damage to the Teflon shrink tubing due to fragmentation of theglass during testing.Perfection of the cutting/opening technique was required to minimizefurther damage to the specimen. lf the position of the anvils within the Teflon sleeve was notmaintained during the sectioning process, shifting and disruption of glass fragments along theshear plane occurred. Figure 2b shows a BF specimen tested in triaxial compression followingcutting and opening of the shrink tube.4.0 ResultsAxial stress vs. axial strain curves were obtained for each confined compression test. Forthe confined sleeve tests, a,xial stress vs. hoop strain plots were also obta ined. The axial stressstrain response of intact and pre-damaged BF and SP glass are compared in Figure 3 for8

specimens tested with hydraulic confinement; confinement pressures are included on the plot.Each curve corresponds to a single test, and is representative of the behavior of each glass atsimilar confinement pressures.The response differs for intact vs. predamaged glass. but isconsistent for BF and SP glass.The axial stress-strain curves for intact confined glass follow a similar trajectory. Thefailure stress for the intact glass increases with confining pressure. The stress-strain curves forthe intact glasses are generally linear until failure. Failure occurs suddenly as indicated by thevertical arrows in Figure 3. Some deviation from linearity is evident for the intact BF glass. Thecause of the non-linearity was not investigated. but may be related to densification of the glass.Densification ofBF glass has been demonstrated in recent work by Holmquist and Johnson [3].The predamaged glass exhibits a dramatic response difference vs. the intact specimens. Alinear stress-strain response was measured upon initial loading of the confined, predamagedspecimens.The moduli are less than for the corresponding intact specimens, owing to thepresence of cracks in the predamaged specimens. With continued loading. a maximum axialstress is reached; this stress is lower than the failure strength of the corresponding intact glass.The strength then drops to a lower value which is maintained with increasing axial strain, asshown in Figure 3. This indicates load carrying capability even after initial damage occurs.Previous interrupted tests on BF glass specimens with mechanical confinement showed specimenstrength was maintained until a critical stress level was exceeded [ 19]. The residual strengthvalues shown in Figure 3 vary with the confinement pressure; higher confining pressurecorresponds to higher residual strength. Similar residual strengths were measured for BF and SPglass at each confinement pressure tested. The fluctuation in the curves around the residualstrength value likely occurs due to particle movement and extension of pre-existing cracks. as9

discussed in the next section. Additional details on the mechanical response are provided111recent publications [25.26].Experimental data from both types oftests were reduced to equivalent stress versus pressureplots for direct performance comparisons of the two glasses. and to assess the effect of specimencondition (e.g., intact, pre-damaged) on response. The equivalent stress, for axial symmetry, isdefined by:(I)whereCY.is the axial load andCY,is the radial load. For the triaxial compression tests.CY,is thefluid pressure from the hydraulic confinement ( CY 0 in compression, (f 0 in compression).For the confined sleeve tests,CY,is determined analytically based on the hoop strain in the steelsleeve - assuming the sleeve remains elastic [25]. The pressure for the equivalent stress vs.pressure plots is the hydrostatic pressure derived from the confining pressure. The hydrostaticpressure on the specimen is given by Eqn (2). See Chocron. et al. (25] for details of the analysis.P ! (2a3r a.) (2)Maximum equivalent stress vs. hydrostatic pressure comparison plots for intact and predamaged glass are shown in Figures 4 and 5. respectively.The behavior of confined vs.unconfined intact glass is compared in Figure 4 for BF and SP glass. The equivalent stressversus pressure response for intact, unconfined glass lies on a line with a slope of 3. from Eqn.(I) and (2) since if, 0for the unconfined tests. There is considerable scatter in the intactstrength of both glasses, though BF glass is generally stronger than SP glass. The strength (i.e.,equivalent stress) increases with confinement, while the scatter in the intact strength decreaseswith confinement.10

The confined compression test results for pre-damaged BF and SP glasses are shown inFigure 5.Both predamaged glasses exhibit similar behavior, and strength, at hydrostaticpressures less than - 1 GPa (the pressure range for the predamaged data in Figure 3). Thedifference becomes more pronounced at higher pressures: greater than 1.0 GPa and 1.2 GPa forSP and BF glass, respectively.When the hydrostatic pressure exceeds a critical pressure. astrength cap occurs. Pre-damaged SP glass exhibits a strength cap at approximately 1.6 GPa.The strength of the pre-damaged BF glass continues to increase until reaching an approximateequivalent stress of 2.1 GPa. Figure 5 shows the strength cap is approximately 0.5 GPa higherfor BF glass. The residual strength difference between the BF and SG glass may help to explainrecent findings by Bless (27] where BF glass demonstrated greater penetration resistance vs. SPglass in depth of penetration experiments using both blunt and sharp projectiles.5.0 DiscussionThe slopes of the equivalent stress vs. pressure plots in Figures 4 and 5 are independent ofthe damage condition (i.e., intact vs. predamaged). However, the intercept varies with the extentof damage. The lower residual strength of predamaged SP glass (vs. predamaged BF glass),shown in Figure 5, may be related to the higher level of initial damage for SP glass (seeFigure 1). Predamaged SP glass specimens exhibited an extensive network of interconnectedcracks. The lower thermal shock resistance and more extensive cracking tor the SP glass isattributed to its high linear expansion coefficient (3x greater than BF glass).The difference in response for the unconfined intact glasses, shown in Figure 4,demonstrates an inherent compositional difference between SP and BF glass. Compositionaldifferences may also indirectly affect the maximum strength (cap) of the predamaged glass.II

5.1 Post-Test Microscopy ObservationsAxial loading of both glasses, whether intact or pre-damaged, results in formation of adominant shear plane. Damage likely initiates from flaws and/or pre-existing cracks due to thethermal shock procedure. Compression loading of the specimen causes slippage. and movementof the material.For Bf glass, specimens exhibited slip along a shear plane located at a55 -70 angle from the compression loading axis. The shear angle for the SP specimens wasslightly less. with an orientation of 50 -60 from the loading axis.Figure 6 showsrespresentative shear planes for BF and SP specimens tested with hydraulic confinement (forconfining pressures up to 250 MPa). The shear plane becomes more pronounced with increasedconfinement pressure.Shear planes were also observed for confined sleeve specimens tested to higher confiningpressures (- 1 GPa max).The sectioned BF and SP specimens shown in Figure 7 arerepresentative of the findings for the mechanical confinement tests. Note the dominant shearplane marked by the solid arrows. and the similarity in the shear angle for the sleeved specimensvs. the specimens tested in triaxial compression (see Figure 6).The maximum confiningpressure for the sleeved specimens shown in Figure 7 was 870 MPa and 650 MPa for BF and SPglass. respectively. These observations indicate that the orientation of the shear failure is (i)independent of the confinement method (i.e . hydraulic vs. mechanical confinement). and (ii)independent of the confinement pressure for both glasses. It also appears independent of strainrate, based on limited high strain rate (- 1 s· 1, - 1400 s· 1) test results for both glasses.Both internal and surface damage were observed during post-test microscopy investigations.This was detected based on evaluation of several specimens at different viewing angles.Specimen BF-37 (Figure 7a) showed that the specimen curvature was maintained without12

disruption of specimen fragments.Also. several specimens that were gold coated for SEMevaluation appeared more intact than identical uncoated specimens; see Ref. [26] for anillustration. The gold coating reduces the transparency of the glass, highlighting cracks thatoriginated from or extend to the surface.Without the coating, it could not be determinedwhether the damage occurs on the surface or is internal.SEM evaluations provided detailed views of the damage in the vicinity of the dominantshear plane for select predamaged specimens. For BF glass. a dusting of very fine ( 1 1-lm)particles was detected in the vicinity of the dominant shear plane.Fine particles are likelydislodged during axial loading as the specimen shifts to relieve the pressure.Highermagnification microscopy revealed the fine particles are actually aggregates of particles; fewindividual particles were observed. Generally, the aggregates were composed of rounded orspherical particles, as illustrated in Figure 8 for two representative BF specimens.Similarparticle morphologies were also observed by Nie and Chen [14] in BF glass specimens followinghigh strain rate compression tests conducted with a double pulse to simulate fracture andcompaction.For BF specimens tested at a higher confining pressure, a hackle region wasdetected on some of the particles/aggregates. This is characteristic of brittle fracture, and isillustrated in Figure 8b at two different locations. marked by arrows.The two differentorientations of the hackle lines indicate crack branching likely occurred.The SP specimens exhibited greater fragmentation than the BF specimens followingconfinement testing. Loose particles were also detected in the vicinity of the dominant shearplane in predamaged SP glass. However. these particles were generally large and angular, asshown in Figure 9. with fewer round particles. The shape and size of the SP particles are readilycompared with the BF particles in Figure 8; note the scale difference. The absence of very fine13

particulates in the SP glass indicates less comminution than the BF glass, possibly due to thehigh density of pre-existing cracks. Whiskers were detected in some SP specimens followingcompression testing; see arrow in Figure 9a. These have also been reported by Bless [17] insoda-lime glass targets following high speed projectile impact.This phenomenon was notexplored; further investigation is necessary to determine the growth kinetics.Deformed and compacted glass particles were observed in the vicinity ofthe shear plane forsome BF glass specimens. This is illustrated in Figure I 0 for Specimen BF-83 following fivecompression load/reload cycles with mechanical confinement (max. confining pressure:- 395 MPa). The surface characteristics, and absence of loose particles, are evidence that particlesliding or rubbing occurred. resulting in compaction. A similar morphology was observed byShockey, et al. [1 0,11] inthe vicinity of the projectile following laboratory scale impactexperiments on borosilicate glass.For several of the BF specimens. it appears that the pressure was high enough for sinteringof the particles to occur.This is illustrated in Figure I 1 for a BF specimen tested withmechanical confinement at a strain rate of 2 s·' ; the maximum hydrostatic pressure in thespecimen approached 2.0 GPa. At lower magnifications (Figure II a), the compacted regionshows individual particles.magnifications (Figure I 1b).The compacted particles appear sintered when viewed at higherCompacted regions were not observed for SP glass followingsimilar testing. For SP glass, the pressure is relieved with extension of pre-existing cracks andshear plane fonnation.14

5.2 Damage MechaoismOptical and SEM findings were compared to the mechanical response data, and used tosupport a hypothesis for the damage mechanism and continued load carrying capability ofpredamaged glass, as shown in Figure 3. Pressure buildup in the confined specimen. resultingfrom axial loading, causes movement and shifting within the glass specimen to relieve thepressure. This results in dislodgement of glass particles, as shown in Figures 8 and 9 for BF andSP glass, respectively. Damage likely initiates at dominant flaws (e.g. pre-existing cracks due tothe thennal shock treatment); propagation proceeds with continued loading.The differentorientations of the hackle regions, shown in Figure 8b. imply that damage propagation proceedsin different planes and directions. Eventually, separate damage regions link together. Initialfailure occurs by slippage along a dominant shear plane, as shown in Figures 6 and 7, when thespecimen can no longer support the maximum applied load. Subsequent loading of the confinedspecimen after formation of the dominant shear plane causes further shifting of the parlicles andfragments, and the creation of secondary damage regions and additional shear planes asillustrated in Figure 7.Shifting of the glass material duringa. ialloadingrequires movement ofparticles within theconfined specimen. As the axial load is increased. the particles attempt to slide over each other.This requires considerable activation energy, based on the observed particle morphologies (seeFigures 8 and 9). Initial movement is limited by frictional resistance between the particles. Withcontinued loading,

the onset and extent of damage in Borofloat glass were evaluated. 2.0 Materials Borofloat 33 (BF) and Starphire (SP) glasses were evaluated. Both glasses were obtained from Swift Glass (Elmira, NY). BF is a borosilicate glass manufactured by Schott Glass using a float process. SP float glass is a crys