DTIC ADA546251: Effect of Loading Rates and Surface Conditions on the Flexural Strength of Borosilicate Glass

0.65 MB·English

by  Defense Technical Information Center

Form Approved
REPORT DOCUMENTATION PAGE
OMB No. 0704-0188
The public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources,
gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of
information, including suggestions for reducing the burden, to Department of Defense, Washington Headquarters Services, Directorate for Information Operations and Reports (0704-0188),
1215 Jefferson Davis Highway, Suite 1204, Arlington, VA 22202-4302. Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to any
penalty for failing to comply with a collection of information if it does not display a currently valid OMB control number.
PLEASE DO NOT RETURN YOUR FORM TO THE ABOVE ADDRESS.
1. REPORT DATE (DD-MM-YYYY) 2. REPORT TYPE 3. DATES COVERED (From - To)
20-03-2011 Journal Paper 01-10-2008 - 30-09-2009
4. TITLE AND SUBTITLE 5a. CONTRACT NUMBER
Effects of Surface Treatment and Interfacial Strength on the Damage 54666EG
Propagation in Layered Transparent Armor Under Impact
5b. GRANT NUMBER
---Effects of Loading Rates and Surface Conditions on Flexural Strength of
W911NF0810533
Borosilicate Glass
5c. PROGRAM ELEMENT NUMBER
6. AUTHOR(S) 5d. PROJECT NUMBER
Weinong Wayne Chen
5e. TASK NUMBER
5f. WORK UNIT NUMBER
7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8. PERFORMING ORGANIZATION
REPORT NUMBER
Purdue University
Sponsored Programs Services
West Lafayette, IN 47907-2108
9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSOR/MONITOR'S ACRONYM(S)
Dr. Ralph Anthenien ARO
U.S. Army Research Office
P.O. Box 12211 11. SPONSOR/MONITOR'S REPORT
Research Triangle Park, NC 27709-2211 NUMBER(S)
54666EG
12. DISTRIBUTION/AVAILABILITY STATEMENT
Approved for public release; federal purpose rights.
13. SUPPLEMENTARY NOTES
14. ABSTRACT
This study evaluates the loading-rate and surface-condition dependence of flexural strength of a borosilicate glass. The tensile
surfaces of glass bending specimens are subjected to 3 different surface treatments. Four-point bending experiments are performed
on those treated samples, at loading rates ranging from 0.7 MPa/s to 4×10^6 MPa/s. The results show that the flexural strength of
the borosilicate glass increases with increasing loading rate. When the surface is chemically etched, the flexural strength increases
by an order of magnitude and fracture initiates from an edge. The high strength measured in the etched glass is still far less than the
intrinsic tensile strength, indicating a flaw-determined failure.
15. SUBJECT TERMS
16. SECURITY CLASSIFICATION OF: 17. LIMITATION OF 18. NUMBER 19a. NAME OF RESPONSIBLE PERSON
a. REPORT b. ABSTRACT c. THIS PAGE ABSTRACT OF
PAGES
19b. TELEPHONE NUMBER (Include area code)
Standard Form 298 (Rev. 8/98)
Reset
Prescribed by ANSI Std. Z39.18
INSTRUCTIONS FOR COMPLETING SF 298
1. REPORT DATE. Full publication date, including 8. PERFORMING ORGANIZATION REPORT NUMBER.
day, month, if available. Must cite at least the year Enter all unique alphanumeric report numbers assigned
and be Year 2000 compliant, e.g. 30-06-1998; by the performing organization, e.g. BRL-1234;
xx-06-1998; xx-xx-1998.
AFWL-TR-85-4017-Vol-21-PT-2.
2. REPORT TYPE. State the type of report, such as
9. SPONSORING/MONITORING AGENCY NAME(S)
final, technical, interim, memorandum, master's
AND ADDRESS(ES). Enter the name and address of the
thesis, progress, quarterly, research, special, group
organization(s) financially responsible for and
study, etc.
monitoring the work.
3. DATES COVERED. Indicate the time during
which the work was performed and the report was 10. SPONSOR/MONITOR'S ACRONYM(S). Enter, if
written, e.g., Jun 1997 - Jun 1998; 1-10 Jun 1996; available, e.g. BRL, ARDEC, NADC.
May - Nov 1998; Nov 1998.
11. SPONSOR/MONITOR'S REPORT NUMBER(S).
4. TITLE. Enter title and subtitle with volume Enter report number as assigned by the sponsoring/
number and part number, if applicable. On classified monitoring agency, if available, e.g. BRL-TR-829; -215.
documents, enter the title classification in
parentheses.
12. DISTRIBUTION/AVAILABILITY STATEMENT. Use
agency-mandated availability statements to indicate the
5a. CONTRACT NUMBER. Enter all contract
public availability or distribution limitations of the
numbers as they appear in the report, e.g.
report. If additional limitations/ restrictions or special
F33615-86-C-5169.
markings are indicated, follow agency authorization
procedures, e.g. RD/FRD, PROPIN, ITAR, etc. Include
5b. GRANT NUMBER. Enter all grant numbers as
copyright information.
they appear in the report, e.g. AFOSR-82-1234.
5c. PROGRAM ELEMENT NUMBER. Enter all 13. SUPPLEMENTARY NOTES. Enter information not
program element numbers as they appear in the included elsewhere such as: prepared in cooperation
report, e.g. 61101A. with; translation of; report supersedes; old edition
number, etc.
5d. PROJECT NUMBER. Enter all project numbers
as they appear in the report, e.g. 1F665702D1257;
14. ABSTRACT. A brief (approximately 200 words)
ILIR.
factual summary of the most significant information.
5e. TASK NUMBER. Enter all task numbers as they
15. SUBJECT TERMS. Key words or phrases
appear in the report, e.g. 05; RF0330201; T4112.
identifying major concepts in the report.
5f. WORK UNIT NUMBER. Enter all work unit
16. SECURITY CLASSIFICATION. Enter security
numbers as they appear in the report, e.g. 001;
AFAPL30480105. classification in accordance with security classification
regulations, e.g. U, C, S, etc. If this form contains
6. AUTHOR(S). Enter name(s) of person(s) classified information, stamp classification level on the
responsible for writing the report, performing the top and bottom of this page.
research, or credited with the content of the report.
The form of entry is the last name, first name, middle 17. LIMITATION OF ABSTRACT. This block must be
initial, and additional qualifiers separated by commas, completed to assign a distribution limitation to the
e.g. Smith, Richard, J, Jr.
abstract. Enter UU (Unclassified Unlimited) or SAR
(Same as Report). An entry in this block is necessary if
7. PERFORMING ORGANIZATION NAME(S) AND
the abstract is to be limited.
ADDRESS(ES). Self-explanatory.
Standard Form 298 Back (Rev. 8/98)
J.Am.Ceram.Soc.,92[6]1287–1295(2009)
DOI:10.1111/j.1551-2916.2009.03019.x
ournal r2009TheAmericanCeramicSociety
J
Effect of Loading Rate and Surface Conditions on the Flexural Strength
of Borosilicate Glass
w
Xu Nie and Weinong W. Chen
AAEandMSESchools,PurdueUniversity,WestLafayette,Indiana47907-2045
Andrew A. Wereszczak
OakRidgeNationalLaboratory,OakRidge,Tennessee37831-6068
Douglas W. Templeton
USArmyTankandAutomotiveResearchandDevelopment,Warren,Michigan48397-5000
Thisstudyevaluatestheloadingrateandsurfaceconditionde- andcrackpropagationofglasses.8–11Comparedwithceramics,
pendence of the flexural strength of a borosilicate glass. The glassmaterialsaremoresusceptibletosurfaceflawsduetothe
glass specimens are subjected to three different surface treat- lackofbulkdefects.4Althoughthetheoreticalstrengthofglass
mentsbeforefour-pointbendingteststostudytheeffectofsur- is considered to be in the order of 10 GPa at room tempera-
faceflaws.Quasistatic(MaterialTestSystem810)anddynamic ture,12verylittleworkhasbeenreportedontheachievementof
(Kolskybar)experimentsareperformedatloadingratesrang- this strength under laboratory conditions mostly due to the
ingfrom0.7to4(cid:2)106MPa/s.Theresultsshowthattheflex- improper handling of glass samples and the moisture effect in
uralstrengthoftheborosilicateglasshasastrongdependenceon thetestingenvironment.
the loading rate. A chemically etched surface produces an en- Different methods to improve the surface quality of oxide
hancedflexuralstrengthbyaboutanorderofmagnitude.Scan- glasses have been reviewed in Donald.13 As a conclusion, sur-
ning electron microscopy images on fracture surfaces indicate face etching is considered to be one of the most effective and
that the failure is governed by different types of flaws under simplemethodstoremoveorreducesurfacedefects.Theprin-
differentsurfacetreatmentconditions.Edgefailureisalsoiden- cipleforsurfaceetchingistoeitherremovesurfacecrackscom-
tifiedforsamplespossessinghighflexuralstrength. pletely or blunt the crack tip significantly through material
removal.Bulkglass strengths exceeding 1 GPahave beenpro-
ducedbythismethod.14,15Recentresearchresultsontheballis-
I. Introduction tic response of thin glass plates have revealed a significantly
enhanced impact resistanceinan HFacid-etchedglasstarget.3
RECENT research on the dynamic compressive response of No fracture was initiated on the back side of the glass plate
borosilicateglassshowedthattheglassiscapableofbear- target,whichwasintension,atimpactspeedsofupto700m/s.
ing up to 1.5 GPa uniaxial compression stress before cracks In this study, although the effects of surface modification on
propagateextensivelyinthespecimen.1Thedynamicdeforma- impactresistanceareclearlyshown,thesurfacemorphologiesof
tion and fracture behavior of this glass under multiaxial com- treatedanduntreatedglassandtheirinfluencesonthematerial
pressionwasstudiedwiththespecimenconfinedbysteelcollars strengthwerenotdocumented.Itisofsignificantinteresttoes-
inaKolskybarorsplitHopkinsonpressurebar(SHPB).2How- tablish the relationship between dynamic flexural strength and
ever,ithasbeenshownthatunderballisticimpactloadingcon- surfaceconditionsinordertooptimizetheballisticresistancein
ditions,themostdominantandvitalfailuremodesarespalling lighttransparentarmors.
andbending-inducedtensiononthebacksideoftheglassplate,3 Inthispaper,westudiedtheloadingrateandsurfacecondi-
which points to the need for determining the dynamic tensile tion effects on the flexural strength of a borosilicate glass uti-
failure behavior under high-rate bending. Under flexural load- lizingaKolskybarwithitstestingsectionmodifiedintoafour-
ing, glass strength is typically limited by surface defects intro- pointbendingconfiguration.Thetensilesurfacesofthebending
duced by after-manufacturing handling, but rarely by bulk specimensaremodifiedintothreedifferentsurfaceconditionsas
defects.4Therefore,fortheglassesinblast-resistingwindowap- follows: ground by220-grit sandpapers, ground bya 1500-grit
plications, it is important to understand the loading rate and sandpaper,andpolishedandetchedby5wt%HFacid.Grind-
surfaceconditioneffectsonthetensileortheflexuralstrength. ing directionsareperpendicular tothe surface tensilestressdi-
Under quasistatic loading conditions, intensive research ef- rection.Themorphologyofthetreatedsurfacesischaracterized
fortshavebeenmadetowardexploringtheinfluenceofvarious by scanning electron microscopy (SEM) and atomic force mi-
surfacetreatmentsontheflexuralstrengthsofglassandceramic croscopy (AFM). Detailed surface features and roughness val-
materialsduringthepastdecades.5–11Ithasbeenidentifiedthat uesaregiveninthispaper.Ahigh-speedcameraissynchronized
surfaceconditionsplayasignificantroleinthefracturestrength withtheSHPBtocapturethecrackinitiationandpropagation
in the glass specimens. It was found that the flexural strength
V.Sglavo—contributingeditor decreaseswithincreasingsurfaceroughnessforgroundsamples,
whiletheetchedsamplespossessthehigheststrengthalthoughat
highsurfaceroughness.Thisindicatesthatthesurfaceroughness
may not be the critical parameter that determines the tensile
ManuscriptNo.24917.ReceivedJune29,2008;approvedFebruary5,2009. strengthofglassmaterials.Whentheloadingrateincreases,the
ThisworkwaspartiallysupportedbytheU.S.ArmyResearchOfficeunderGrantNo.
W911-05-1-0218toPurdueUniversity.WereszczakwassupportedbyWFOsponsorUS strength increases for all the surface conditions. The detailed
ArmyTank-AutomotiveResearch,Development,andEngineeringCenterundercontract experimental setup, procedures, and experimental results are
DE-AC05-00OR22725withUT-Battelle,LLC.
wAuthortowhomcorrespondenceshouldbeaddressed.e-mail:[email protected] presentedinthefollowingsections.
1287
1288 JournaloftheAmericanCeramicSociety—Nieetal. Vol.92,No.6
II. ExperimentalProcedure dureofdeterminingthefracturetoughnessofceramicsatqua-
sistatic loading rates. To load the specimen at high constant
(1) MaterialsandSpecimens
rates while maintaining equilibrated loading across the gauge
TheborosilicateglassusedinthisresearchwasprovidedbyU.S. section, the pulse-shaping technique was used to generate
Army Research Laboratory, Aberdeen Proving Ground, MD, controlled loading profiles. In this research, we adopted the
intheformofflatplates.Thedimensionsandpropertiesofthe four-point bending fixture design following ASTM Standard
as-receivedmaterialweredescribedinapreviouspaperstudying C1161-02c23 for determination of the flexural strength of ad-
thesheareffectsonthecompressivestrengthofthesameglass.1
vanced ceramic materials, and extended the method to high
Theglassplateswerefirstmachinedinto2mm(cid:2)3mm(cid:2)30mm rates.
glass prisms, and then polished to a surface finish of 80/50 In a typical SHPB experiment on brittle materials such as
scratch/dig. All the four edges of the as-received samples were
glasses andceramics, a nearly linear loading pulse isnecessary
chamfered before surface treatments were performed. The
in order to deform the specimen at a nearly constant strain
chamfer angle was 451, with a dimension of approximately rate.24–26 This is usually achieved by placing a thin annealed
0.12 mm. The polished and chamfered samples were then di- copper disk, called a pulse shaper, betweenthe striker and the
vided into three groups. The first group was ground with a
incidentbar.Ramploadingwithdifferentslopesmaybegener-
220-grit sandpaper, and the second group was ground with a
ated by adjusting the thickness and diameter of this pulse
1500-gritsandpaper.Bothgrindingswereperformedonlyonthe
shaper, as well as the striking velocity. In this research, the
tensile surface and were perpendicular to the tensile axis, as specimenisasmallglassbeamthathasaverylowstiffnessalong
shown in Fig. 1. These two groups were denoted as group #1
theloadingaxis.AccordingtotheEulerianelasticbeamtheory,
andgroup#2,respectively.Thespecimensfromthethirdgroup
aconstantloadingrateisensuredonlywhenthedeflectionrate
wereimmersedin5%HFacidfor15min.TheprincipleofHF isconstant.Whenthespecimenstiffnessisverylow,thedeflec-
acidetchingisthatHFintheaqueoussolutionattacksSiO in
2 tion rate is nearly determined by the particle velocity of the
the glass toform water-soluble reaction products. Inthis way,
loadingendoftheincidentbar,whichisinturndeterminedby
theglassmaterialisstrippedofffromthesurfacelayerbylayer
theamplitudeoftheincidentpulse.Thus,toachieveaconstant
sothatthepreexistingsurfaceflawsareeliminated.Theetching deflection rate in the beam specimen, a constant amplitude of
processremoved100–150mmofglassmaterialfromthetensile
theincidentpulseisnecessary.
surface. This group was denoted as group #3. To minimize
Besidesthelowstiffness,thebeamspecimensalsopossesslow
theinteractionbetweentheetchedglassandthemoistureinthe
resonancefrequenciesthatareeasilyexcitedbysuddenacceler-
room air, the etched specimens were subjected to mechanical ationordecelerationoftheloadinggrips.27Althoughitisinev-
loadingwithinafewminutesafteretching.
itablethatthespecimenshavetobeacceleratedtodesiredtesting
conditions and then decelerated back, the process of velocity
changes can be controlled to minimize inertia effects. For ex-
(2) StrengthTestingMethodology ample, if a step incident pulse is used to achieve the necessary
In this study, uniaxial flexural strength experiments are per- constant amplitude of the incident pulse, the sharp rise in the
formedinsteadofuniaxialtensilestrengthtesting,whichisvery stress–time profile will excite resonance in the beam specimen.
difficult to conduct accurately on brittle materials. The mea- Thisinducesmixingofinertiaforceswiththemechanicalstress
suredflexuralstrengthvaluesareusedtomakeinferencesonthe in the measured axial load signals, making the resultant data
uniaxial tensile strength of glass. Four-point bending flexural verydifficulttoexplain.Theinertiaforcesalsodestroytheforce
testswereconductedtoevaluatethetensilestrengthsofthethree equilibrium between the loading and the supporting points on
groupsofborosilicateglasssamples.Asetoffour-pointbending thespecimen,makingthedatareductionuncertainandinaccu-
fixtures were designed according to ASTM C1161-02c. The rate.Thehigh-frequencyoscillationsthatrideonaconventional
spacing between the two loading pins and that between two SHPBincidentpulsecanalsocauseinertiaproblems,inaddition
support pins are 10 and 20 mm, respectively. Pin rollers were tononconstantloading rates.Inthisresearch,inordertogen-
made of hardened M-2 steel with a hardness of RC 60. These erateanincidentpulsewithbothagradualinitialincreaseanda
rollerswerefixedtothealuminumbackfixture.Thetemperature constant plateau over most of the loading duration without
ofthetestingenvironmentwas261C,andtherelativehumidity high-frequencyoscillations,athincopperdiskof0.2mmthick-
was34%.Thelow-rate(0.7, 50, and2500MPa/s)experiments nessand3mmdiameterwasplacedbetweenthestrikerandthe
wereperformedonacloseloop-controlledservohydraulictest- incidentbarasthepulseshaper.Variationsinthepulse-shaper
ing machine (Material Test System [MTS] 810, MTS Corp., dimensionsandstrikingvelocitycanaltertheinitialloadingrate.
Eden, Prairie, MN), while the high-rate experiments were car- The maximum achievable loading rate without sacrificing the
riedoutonamodifiedSHPBsetup.SHPBisawell-established force equilibrium across the sample was found to be 4(cid:2)106
apparatus commonly utilized in the high-strain-rate character- MPa/s.Also,inboththeservo-hydraulicandtheSHPBexper-
ization of materials to provide a complete family of dynamic iments,athinTeflonfoilwasplacedbetweenthesampleandthe
stress–strain curves as a function of strain rates.16 This tech- roller surfaces to minimize friction and contact-induced stress
niquewasinitiallydesignedbyKolsky17forthecharacterization concentration. The use of a Teflon foil may exert an effect on
ofthedynamicflowbehaviorofductilematerialsatstrainrates wave propagation. However, due to the small thickness of the
up to 104 s(cid:3)1. The modified versions of this device have been foil,thedurationofeffectisveryshort,whichisabsorbedinthe
widelyusedincharacterizingthedynamicpropertiesofdifferent dynamicequilibriumprocessoftheloading.
materials, such as brittle materials1,18 and soft materials,19,20
duringthepastyears.ArecentmodificationofSHPB21extends
the range of characterization to the dynamic fracture tough- (3) GlassSurfaceCharacterizationandFractography
nessofbrittlematerialsinthefour-pointbendingconfiguration. Surfaceroughnessvaluehaslongbeenconsideredasanimpor-
This experimental method was designed based on the ASTM tantparameterdescribingthesurfacequalitiesandtherefore,for
StandardC1421-01b,22whichspecifiesthestandardizedproce- brittlematerials,acriticalparameterthatcanaffecttheflexural
strength.28,29 Two measures of surface quality that are com-
monly used are the arithmetic mean roughness value (R ) and
Grinding direction a
Tensile axis perpendicular to themaximumpeak-to-valleyheight(Rt).Theformerprovidesa
grinding direction general picture of surface flaws, while the latter points out the
2mm mostsevereflawonthesurfaceandisperhapsmorerelevantto
30mm the strength response in a brittle material. In this research, we
3mm
utilizebothSEMandAFMforsurfacecharacterization.SEM
Fig.1. Surfacegrindingofbendingsamples. images provide a surface view over a relatively large area,
June2009 RateandSurfaceEffectsonGlassFailure 1289
and AFM quantifies the selected surface with 3D images and thesurfacepropertyofthesample.Opticalmicroscopywasalso
roughnessvalues.Itshouldbenotedthattheroughnessvalues used to analyze the fractography of fracture surfaces using the
obtained bya profilometer measurement mayvary from those methods outlined in ASTM C1322. In particular, we identified
obtainedbyAFMduetothedifferencesintheareascoveredby thelocationoffailureinitiationunderdifferentloadingratesand
thetwoinstruments.AFMimagesprovideamoredetailedsur- surfaceconditions.Fracturesurfacesfrombrokensamplesunder
faceflawdistributionthanthezigzagcurvesfromprofilometers. six different combinations of testing conditions (two loading
The limited scan area of AFM image may result in surface ratesandthreesurfacetreatments)wereexamined.
roughness values that are not representative of the entire sur-
face. However, the trends of surface roughness variation mea-
suredamongdifferentsurfaceconditionsareconsistentbyboth III. ExperimentalResultsandDiscussions
methods.8Intheresearchreportedinthisarticle,AFMwasused
for surface morphology measurements and comparisons. The (1) GlassSurfaceFeatures
maximum scanning capability of the AFM we used is 0.1 AtypicalsetofSEMandAFMimagesforallthethreegroups
mm(cid:2)0.1 mm, which is a relatively small area compared with ofglasssamplesaredisplayedinFig.2.Thecorrespondingsur-
theentireglasssurfaceinvestigated.Foreachexaminedsample,a faceroughnessvaluesaresummarizedinTableI.Forthepur-
minimumof10AFMimagesweretakenfromdifferentpartsof poseofcomparison,asetofas-polishedsurfaceimagesarealso
the surface to avoid localized results. Surface roughness values showninFig.3.Itisevidentthatthepolishedsurfacesshowan
(R and R) of theseimages werethen collected by a computer extremelysmoothandflatfeaturewithveryfewidentifiablesur-
a t
programandthecalculatedaveragevalueswereconsideredtobe face flaws at the magnification of (cid:2)1800 for the SEM image.
Fig.2. Glasssamplesurfacemorphologyof(a)groundby220-gritsandpaper;(b)groundby1500-gritsandpaper;(c)polishedandetchedbyHFacid.
1290 JournaloftheAmericanCeramicSociety—Nieetal. Vol.92,No.6
TableI. SurfaceRoughnessValuesforDifferentGroups 0.006 0.03
ofSamples
0.004 0.02
Samplegroup#1 Samplegroup#2 Samplegroup#3 V)
Ra(mm) 0.414 0.0585 0.236 al (V) 0.002 0.01 gnal (
Rt(mm) 3.8 0.725 2.25 sign 0.000 0.00 on si
cident –0.002 –0.01 smissi
n n
I a
–0.004 Incident signal –0.02 Tr
Transmission signal
–0.006 –0.03
0.0000 0.0004 0.0008 0.0012 0.0016
Time (s)
Fig.5. OriginaloscilloscoperecordsofasplitHopkinsonpressurebar
Fig.3. Glasssurfacemorphologyofas-polishedsamples. dynamicbendingexperiment.
Thisnearlyflaw-freesurfaceprovidesanidealbasefor further 800
surfacemodifications.BothSEMandAFMimagesfromFig.2
revealed two types of major flaws on sample surfaces from 700 Front end Force
Back end Force
groups #1 and #2. Grooves are evident along the grinding di-
600
rection,suggestingseveredeformationcausedbyparticleabra-
sion. However, the real fracture-initiating flaws are thought to 500
N)
besharpindentationpitsindicatedbythewhitearrowsinFigs. e ( 400
2(a)and(b).Theseflawsarecreatedbyparticleindentationon c
the initially polished surfaces during the grinding process, fol- For 300
lowed by the formation of new cracks underneath due to the
alkali ion-assisted surface reaction between moisture and the 200
silicanetwork.30Whiletheglasssampleisloadedinfour-point
100
bending,stressconcentrationsaroundthecracktipsincreasethe
localstressestomuchhigherlevelsanddrivethecracktoprop- 0
agatethroughthespecimen.Thisfailuremodeisactivatedeven 0 100 200 300 400
thoughtheoverallstresslevelinthesampleisstillverylow.The Time (microsecond)
SEMimagesofFigs.2(a)and(b)demonstrateaclearcompar-
ison in the sizedifferences in those indentation pits created by Fig.6. Force equilibrium histories between the front and back end
spans.
differentgritsizesandpaper particles.Therelativelysmallflaw
size in Fig. 2(b) indicates a less damaged surface and thuspo-
tentiallyahighertensilestrength.Itcanbeseenthatthesurface the surface morphology transformed into a bumpy pattern
roughness values (both R and R) for group #2 samples are as shown in Fig. 2(c). It is believed that a simultaneous acid
a t
lowerthanthoseofgroup#1samples.Figure2(c)showsasur- attack on a series of surface flaws eventually yielded this ob-
facemorphologyafter15minofHFacidetchingonthetensile servedsurfacepattern.Forasharp,strength-dominatingsurface
surfaces of polished samples. It can be seen that the polishing crack, the crack tip radius can be in the range of nanometers.
tracesarealmostcompletelyremovedbyHFacidetching,leav- Afteretching,theaverageradiusincreasestoaround20mm.The
ingroundedpitsasthedominantsurfacefeatures. flaw tip rounding achieved by acid attack significantly relieves
Asimplifiedmodelthatdescribesthesurfaceetchingeffecton the stress concentrations at the crack tip when the surface is
theflexurestrengthofglassmaterialsisschematicallyshownin subjectedtotension.
Fig.4.31Similarapproacheswereusedinamorerecentstudyon
thebendingstrengthofetchedsoda-limeglassrods.32According
(2) Four-PointBendingExperiments
to this model, an idealized surface crack is uniformly attacked
byacidateverypointsothatthiscrackeventuallydevelopsinto Four-point bending experiments on glass samples subjected to
asemicircle.Basedonthemodel,thefinalradiiofthesepitsare different tensile surface treatments were conducted at four
determinedbythedepthoforiginalcracks.Hence,theobserved
variationinthepitsradiimaybedirectlyrelatedtothevariation
1200 6x106
in the initial crack size. Sharp surface flaws are considered to Stress history
exist widely in glass materials due to mechanical machining, Pa) 1000 Loading rate history 5x106
grinding, and polishing, as well as after-manufacturing han- M s)
dling.Inourstudy,theas-receivedglasssurfaceswerepolished ss ( 800 4x106 Pa/
and were flat and smooth as shown in Fig. 3. After etching, e M
Original glass Surface after Surface after nsile str 600 3x106 g rate (
surface short etch l0 “full” etch urface te 240000 12xx110066 Loadin
S
0 0
l0 0 50 100 150 200 250 300 350
l0 Time (us)
Fig.7. LoadingratehistoryinasplitHopkinsonpressurebarexper-
Fig.4. HFacidetchingofanidealizedsurfaceflaw. iment.
June2009 RateandSurfaceEffectsonGlassFailure 1291
1400 differentloadingratestoinvestigatetherateeffectsonflexural
strength. Three of the loading rates (0.7, 50, and 2500 MPa/s)
Pa) 1200 GGrroouunndd bbyy 212500 0G Gritr ist asnadn dp appaeprer were achieved by a servohydraulic testing machine, while a
M Polished and etched by HF acid modified SHPB was used to perform experiments at the high-
h ( 1000 estloadingrate—4(cid:2)106MPa/s.Atypicalsetoforiginaloscil-
ngt loscope recordings of the incident, reflected, and transmitted
stre 800 signalsfromanSHPBexperimentareshowninFig.5.Itshould
e benotedthat the transmitted signal was collected bya pair of
sil 600 semiconductorstraingauges,whosesensitivityisapproximately
n
e 70 times higher than that of the resistor strain gauges on the
ce t 400 incidentbar.Thismodificationistoincreasethesignal-to-noise
a
urf 200 ratioofweaktransmittedsignals.Inadynamicfour-pointbend-
S ingexperiment,forceequilibriumbetweenthetwopairsofload-
ing/supporting points may become an issue due to inertia and
0
1E–3 0.1 10 1000 100000 wave-propagationeffects. Ifa specimen isloadedwitha tradi-
tional trapezoidal-like Hopkinson bar incident pulse, sample
Loading rate (MPa/s)
vibration/resonance is expected because of the high-frequency
Fig.8. Loading-rate effects on flexural strength of borosilicate glass components in the incident pulse, which imposes difficulties
withdifferentsurfaceconditions. in interpreting the collected force signals. Inthis case, numeri-
(a)
(b)
(c)
Fig.9. Fracturesurfaceandfailureoriginsofdifferentgroupsofglasssamples.Flawsizeismeasuredintermsoflengthanddepth:(a)groundby
220-gritsandpapers(approximateflawsize125mm(cid:2)29mm);(b)groundby1500-gritsandpapers(approximateflawsize57mm(cid:2)11mm);(c)polished
andetchedbyHFacid(approximateflawsize12mm(cid:2)4mm).
1292 JournaloftheAmericanCeramicSociety—Nieetal. Vol.92,No.6
process. The nearly overlapping force histories recorded at the
loadingpointsandsupportingpointsindicatethatthespecimen
wassubjectedtoanequilibratedloadinghistory.Inertiaeffects
wereminimized.Figure7showsthetimehistoryoftheloading
rate.Thisisobtainedbydifferentiatingthestress–timeprofilein
the transmission bar, which is also shown. It is seen that after
increasinginitially,theloadingratewasmaintainedtoarounda
constant value, which is important when the strength results
needtobereportedasafunctionofloadingrates.
In a four-point bending experiment on a beam with a rect-
angularcrosssection,themaximumsurfacetensilestressiscal-
culatedas
3PðtÞðL(cid:3)lÞ
sðtÞ¼ (1)
2bd2
wherelistheloadspan(incidentbarside),Listhesupportspan
(transmissionbarside),andbanddarethespecimenwidthand
height,respectively.P(t)isthemeasuredforcehistoryandthe
Fig.10. SemiellipticalsubsurfacemircocrackinducedbyKnoopinden-
flexurestrengthiscalculatedusingthemaximumachievedvalue
tation.
ofP.IntheexperimentsperformedontheservohydraulicMTS
machine, this force history is recorded by a load cell, while
cal methods are often required in sample stress analysis. It is inSHPBexperimentstheforcehistoryinthespecimeniscalcu-
certainlydesiredthatinstrength-characterizationexperimentsat latedby
high rates, inertia effects are minimized. In this study, we suc-
cessfullyavoidedthepotentialvibration/resonancethroughcon- PðtÞ¼EAe ðtÞ (2)
T
troloftheincidentpulseshape.Thisisachievedbyextendingthe
risetimeoftheincident(loading)pulsefromaround10toB160 whereEandAaretheYoung’smodulusandthecross-sectional
ms, with a smooth transition to a plateau. This loading profile areaofthetransmissionbar,respectively;e isthestrainsignal
T
eliminatesmostofthehigh-frequencycomponentsinaconven- measuredonthetransmissionbar.Whenthespecimenfractures,
tional SHPB incident pulse and thus significantly reduces the theforcehistoryreaches its peak.The flexuralstrengthisthen
possibilityofexcitingresonanceinthebeamspecimenorcaus- calculatedfromthispeakloadusingEq.(1).
ingwavedispersionalongthebars.Theplateaufollowedbythe Thevariationinflexuralstrengthasafunctionoftheloading
initialrampisnecessaryforcreatingaconstantloadingrateat rateforallsurfaceconditionsisshowninFig.8.Itisclearthat
thetensilesurface.Figure6showstheforceequilibriumbetween theflexuralstrengthofborosilicateglassincreaseswithincreas-
theincidentandtransmittedloadspans.Forcevibrationinthe ingloadingratesintheloading-raterangeachievedinthisstudy,
incidentbarsidewaslimitedtotheverybeginningoftheloading regardless of the surface conditions. Below the rate of 2500
(a)
1 2 3
(b)
1 2 3
Fig.11. High-speedcameraimagesshowingthetypicaldynamicfailureof(a)samplegroundbysandpapers;(b)sampleetchedbyHFacid.White
arrowindicatesthethroughcrack.
June2009 RateandSurfaceEffectsonGlassFailure 1293
MPa/s,theflexural strengthisroughly a linear functionofthe surface images taken on a specimen after the experiment at a
logarithmloadingrate,whereasnotmuchratesensitivityisob- 4(cid:2)106 MPa/s loading rate are shown in Fig. 12. The sample
servedabove2500MPa/s.Forthegroup#1and#2samples,the crosssectionis2mm(cid:2)3mm.Forsamplesgroundbya220-grit
particle size difference in sandpapers imposes only relatively sandpaper,fracturewasinitiatedfromsurface-locatedstrength-
smallvariationsintheflexuralstrength.Theaveragestrengthof limitingflaws.Theflawislikelyassociatedwitharelativelylarge
group#2samplesis60%–90%(dependingontheloadingrates) anddeepmachininggrooveorpit.ForHFacid-etchedsamples
higher than that of group #1 samples. This strength improve- andsamplesgroundbya1500-gritsandpaper,fractureismore
mentisachievedbyanapproximately80%reductioninsurface likely to initiate from an edge-located, strength-limiting flaw.
roughness. For group #3 samples, which were etched by HF SimilarresultsatalowerloadingratearealsoshowninFig.13.
acid,theaverageflexuralstrengthis700%–1500%higherthan Thestressconcentrationsontheedgeflawsbecomemoredom-
thatofgroup#1samplesoverthesameloading-raterange.The inant as the surface tensile strength increases. Previous studies
averageflexuralstrengthofetchedsamplesachievedatthehigh- onthebendingstrengthofborosilicateglass33showedthatthe
estloadingrateis1.1GPa.Thesurfaceroughnessvaluesofthe edge defects should account for the low-strength behavior of
etched samples are actually higher than those of fine-ground unchamferedsamples.However,inourcurrentstudy,thefailure
samples(group#2).Thisobservationindicatesthatthesurface of glass samples in four-point bending tests is actually a com-
roughness value alone is not sufficient to relate the flexural petingprocessbetweenthesurfaceflaws,whichwereintention-
strength to the surface quality of glass materials. SEM images ally introduced during various surface modifications, and
on fracture surfaces and therefore examined to further investi-
gate the failure mechanism of different groups of specimens.
(a)
Failureoriginsinsandpaper-groundglasssamplesareobserved
to locate at the subsurface microcrack front, as is evident in C
Figs. 9(a) and (b), which are most likely being induced by in-
dentation of abrasive particles.Such semielliptical microcracks
areverysimilarinshapetothoseresultingfromKnoopinden-
tation,asshowninFig.10.Hence,theflexuralstrengthofsand-
paper-ground samples is governed by the sharp subsurface
flaws.Theobserveddifferencesinstrengthunderdifferentgrind-
ingconditionscouldalsobeexplainedbythedifferencesinthe
flawsize.Figure9(c)showsthefracturesurfaceofanHFacid-
etched sample. In this particular case, failure is initiated from
right underneath a deep surface pit (not a sharp crack front).
The acid-etching-induced surface flaw is much smaller in size
T
compared with those semielliptical cracks, and no subsurface
1mm
cracksarevisible.Thesesurfacepitsarebluntinnaturesothat
the stress intensity is much less than that around a sharp sub-
surfacecrackfront,whichresultsinthehighstrengthofetched (b)
samples.
A further look at the loading rate dependence on flexural
C
strengthrevealsthatforsandpaper-groundsamples,theflexural
strengthincreasedbyabout90%whentheloading ratevaried
from 0.7(cid:2)106 to 4(cid:2)106 MPa/s, whereas for HF acid-etched
samples,thusstrengthincreaseisabout200%.Becausetheonly
differencesbetweenthosetwogroupsofsamplesarethecritical
flaw types, the difference in the rate dependence of strength is
consideredtobethedifferenceintheshapesofdefects.Undera
sharp indentation pit, cracks are formed and connected to the
pitbeforeglobalmechanicalloading.Ontheotherhand,under
apitbluntedbyHFacid,newcrackswillhavetoformunderthe
loading.Ahigherloadingratewillcausemultiplecrackstoform
simultaneously,whichmayaccountforthedifferenceintherate 1mm T
dependency.
Tomonitorthedynamicdeformationandfailureprocessesin (c)
theglassspecimen,ahigh-speeddigitalcameraalongwithapair
ofstrobelightswassynchronizedwiththeSHPBtorecordthe
dynamicbendingdeformationandfailureathighloadingrates. C
Thecameratriggerwascarefullysetsuchthattheentiredefor-
mation and fracture processes could be visualized. Thirty-two
framesweretakenineachexperimentataframerateof50000
fps.Thefractureofgroup#1and#2sampleswasfoundtobe
verysimilar,forwhichatypicalimageisshowninFig.11(a).A
dominantcrackwasinitiatedfromthetensionsideandpropa-
gated to the compression side, leading to the fracture of the
beamspecimen.However,whenthetensilesurfaceofthebeam
sampleis polishedandetched(group#3), thefailureprocess is
quitedifferent,asshowninFig.11(b).Whilethespecimenwas
intensively bent, multiple cracks were initiated and propagated 1mm
T
throughthesample.Thisoccurredinan‘‘explosive’’mannerand
disassembled the sample into many tiny pieces. The dynamic
Fig.12. Fracture surface morphology of samples tested at 4(cid:2)106
compressivestrengthofthisborosilicateglassisaround1.5GPa,
MPa/s. (a) Sample ground by 220-grit sandpaper; (b) sample ground
whichwasmeasuredinapreviousresearch.1 by1500-gritsandpaper;(c)samplepolishedandthenetchedbyHFacid.
The tested samples were collected and the fracture surfaces ‘‘C’’and‘‘T’’represent‘‘compressionside’’and‘‘tensionside,’’respec-
were examined under an optical microscope. A set of fracture tively.Arrowsindicatecrackinitiationsites.
1294 JournaloftheAmericanCeramicSociety—Nieetal. Vol.92,No.6
(a) higherthanthecurrentlymeasuredvalues.Thefailureinitiation
sites have raised challenges in the accurate characterization of
C the flexural strength ofborosilicateglass under desiredsurface
conditions.Furtherexpansionofthecurrentexperimentaltech-
niqueisdesiredinordertoeliminateedgeflaws.
IV. Summary
Quasistatic and dynamic flexural experiments were conducted
on a borosilicate glass under different surface conditions. The
dynamic experiments were carried out on a modified SHPB
setup.Apulse-shapingtechniquewasusedtosubjectthespec-
imen to constant loading rates. The surface conditions of the
1mm T tensilesurfacesofthesampleswerevariedtostudytheireffects
ontheflexuralstrengthoftheglass.Asampleflexuralstrength
of 1.1 GPa at high loading rates was measured on specimens
(b) where the tensile surfaces were chemically etched. The flexural
strengthoftheborosilicateglassincreaseswithincreasingload-
C ing rate for all the surface conditions studied. For the ground
samples,the flexuralstrength decreases withincreasing surface
roughness, while the etched samples possess high strength al-
though at high surface roughness. SEM image analysis of the
fracture surfaces shows that small, blunt surface pits are the
failureinitiationsitesforacid-etchedsamples,whilelarge,sharp
sub-surfacecracksareidentifiedasfractureoriginsforthesand-
paper-groundsamples.
High-speed camera images reveal that a dominating crack
fromatensilesurfaceflawgovernsthefractureinitiationinmost
ground samples. Multiple crack initiation was observed in the
failure of the samples with etched tensile surfaces. Optical mi-
1mm T croscopyimagesshowedthatflawscausedfromchamferingand
locatedatspecimenedgesarethecrackinitiationsitesforspec-
imens possessinghigh flexural strengths. Thisfact suggests the
(c)
need to extend the current four-point bending experiment to
equibiaxial flexural experiments to circumvent the effects of
C edgesonflexuralstrength.
References
1X.Nie,W.Chen,X.Sun,andD.Templeton,‘‘DynamicFailureofBorosilicate
GlassUnderCompression/ShearLoading,’’J.Am.Ceram.Soc.,90[8]2556–62
(2007).
2K.A.Dannemann,A.E.Nicholls,C.E.AndersonJr.,I.S.Chocron,andJ.D.
Walker,‘‘Compressiontestingandresponseofborosilicateglass:intactanddam-
aged’’; Proceedings of Advanced Ceramics and Composites Conference, Cocoa
Beach,FL,January23–27,2006.
3A.S.Vlasov,E.L.Zilberbrand,A.A.Kozhushko,A.I.Kozachuk,andA.B.
1mm Sinani,‘‘BehaviorofStrengthenedGlassUnderHigh-VelocityImpact,’’Strength
T Mater.,34[3]266–8(2002).
4M.Hara,‘‘SomeAspectsofStrengthCharacteristicsofGlass,’’Glastech.Ber.,
61,191–6(1988).
Fig.13. Fracturesurfacemorphologyofsamplestestedat0.7MPa/s. 5B.P.Bandyopadhyay,‘‘TheEffectsofGrindingParametersontheStrength
(a)Samplegroundby220-gritsandpaper;(b)samplegroundby1500- andSurfaceFinishofTwoSiliconNitrideCeramics,’’J.Mater.Process.Technol.,
gritsandpaper;(c) samplepolishedandthenetchedbyHFacid. ‘‘C’’ 53,533–43(1995).
and‘‘T’’represent‘‘compressionside’’and‘‘tensionside,’’respectively. 6D.M.Liu,C.T.Fu,andL.J.Lin,‘‘InfluenceofMachiningontheStrengthof
Arrowsindicatecrackinitiationsites. SiC–Al2O3–Y2O3Ceramic,’’Ceram.Int.,22,267–70(1996).
7M.Guazzato,M.Albakry,L.Quach,andM.V.Swain,‘‘InfluenceofGrinding,
Sandblasting,PolishingandHeatTreatmentontheFlexuralStrengthofaGlass-
the edge flawsfromchamfering. Chamfering onthe edge both InfiltratedAlumina-ReinforcedDentalCeramic,’’Biomaterials,25,2153–60(2004).
reducestheflawsizeintroducedbymachiningandremovesthe 8J.E.RitterJr.andC.L.Sherburne,‘‘DynamicandStaticFatigueofSilicate
Glasses,’’J.Am.Ceram.Soc.,54[12]601–5(1971).
highly stress-concentrated corners. While the corners are re- 9J.E.RitterandM.R.Lin,‘‘EffectofPolymerCoatingsontheStrengthand
moved, newflawscreatedby the chamfering procedure are lo- FatigueBehaviorofIndentedSoda-LimeGlass,’’GlassTechnol.,32[2]51–4(1991).
cated in less stress-concentrated areas. When the surface flaws 10J.J.MecholskyJr.,S.W.Freiman,andR.W.Rice,‘‘EffectofGrindingon
becamelesssevere,theglassfailureunderwentatransitionfrom FlawGeometryandFractureofGlass,’’J.Am.Ceram.Soc.,60[3–4]114–7.
11G.ScottGlaesemann,K.Jakus,andJ.E.RitterJr.,‘‘StrengthVariabilityof
a surface flaw-dominated mode to an edge flaw-dominated
IndentedSoda-LimeGlass,’’J.Am.Ceram.Soc.,70[6]441–4(1987).
mode. The chamfering procedure increased the stress level for 12C.Gurney,‘‘SourceofWeaknessinGlass,’’Proc.R.Soc.Lond.,282[1388]
thistransition.OncomparingFigs.12and13,itisevidentthat 24–33(1964).
the surface quality level required for this transition increased 13I.W.Donald,‘‘MethodsforImprovingtheMechanicalPropertiesofOxide
Glasses,’’J.Mater.Sci.,24,4177–208(1989).
withdecreasingloadingrate.Thiscouldbeattributedtothefact 14C. Symmers, J. B. Ward, and B. Sugarman, ‘‘Studies of the Mechanical
thattheultimatetensilestrengthatalowerratedecreasesdras- StrengthofGlass,’’Phys.Chem.Glasses,3,76–83(1962).
tically for all the surface conditions investigated, and thus a 15C.K.SahaandA.R.Cooper,‘‘EffectofEtchedDepthonGlassStrength,’’
higher surface quality level is required to satisfy the transition J.Am.Ceram.Soc.,67[8]C158–60(1984).
16G.T.Gray,‘‘ClassicSplitHopkinsonPressureBarTechnique’’;pp.462–76,in
stress.Becauseoftheprematurefailurefromtheedgeflaws,the
ASMHandbook,Vol.8,MechanicalTestingandEvaluation.EditedbyH.Kuhn
intrinsictensilestrengthoftheetchedsamplesisexpectedtobe andD.Medlin.ASMInternational,MaterialsPark,OH,2000.