ebook img

NASA Technical Reports Server (NTRS) 19980237335: Room Temperature Tensile Behavior and Damage Accumulation of Hi-Nicalon Reinforced SiC Matrix Composites PDF

10 Pages·0.52 MB·English
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview NASA Technical Reports Server (NTRS) 19980237335: Room Temperature Tensile Behavior and Damage Accumulation of Hi-Nicalon Reinforced SiC Matrix Composites

// ROOM TEMPERATURE TENSILE BEHAVIOR AND DAMAGE ACCUMULATION OF HI-NICALON REINFORCED SIC MATRIX COMPOSITES G.N. Morscher*, Case Western Reserve University, Cleveland, OH J.Z. Gyekenyesi ° ,Cleveland State University, Cleveland, OH ABSTRACT Composites consistingofwoven Hi-Nicalon fibers,BN interphases, and different SiC matrices were studied intensionat roomtemperature. Composites with SiC matrices processed byCVI and meltinfiltrationwere compared. Monotonic and load/unload/reload tensile hysteresis experiments were performed. A modal acousticemission (AE) analyzer was usedto monitordamage accumulation duringthe tensiletest. Post test polishingof the tensilegage sectionswas performed todetermine the extent ofcracking. The occurrence andlocationofcrackingcouldeasily bedetermined usingmodal AE. The loss ofmodulus couldalsoeffectivelybedetermined from thechange inthe velocity of sound across thesample. Finally,the stresses where cracks appear tointersect the load- bearing fibers correspondwithhightemperature lowcyclefatigue runout stresses for these materials. INTRODUCTION The hightemperature mechanical propertiesofSiC/SiC composites are limited duetooxidation reactionswhichoccur inthe presence ofspecies suchas oxygen and water vapor. What enables suchreactionstooccuratthe fiber/interphase/matrix region ofacomposite isenvironmental ingressthrough pores inthe matrixand/or through matrix cracks. For SiC/SiC composites where chemical vapor infiltration(CVI) isthe matrix processing route,thefiber-matrix interphase regionshouldbe sealed offfrom the environment even though significantmatrixporosityexists. Therefore, the environmental reactions are dependent on matrixcracking, althoughahighly porous matrix willallow the environment access toinnerregionsofa compositewhichwouldnot beaccessible ifthe matrixwas dense. The compositefabrication methoddictates the amount of porosity which exists inthe matrix. The loadinghistoryofthecomposite, coupled withthe flaw distributioninthe matrix, dictatethe amount and natureofcracking that exists inthe matrix. An understanding ofthe damage accumulation inacomposite is criticalin determiningwhen matrixcrackingoccursand thenature ofmatrix cracking, i.e. local microcracks orthrough-thickness cracks. Guillaumat and Lamon[1.2h1ave characterized "Resident Research Associates atNASA Lewis Research Center; Cleveland, OH the damage accumulation inaNicalon #,C interphase,CVI SiC matrixcompositeas a TM function ofstrain. They foundthat 1. Microcracks emanate from macropores between differentfiberplies and propagate acrossthe plies(s=0.025% to0.12%) 2. Cracks then form andpropagate intransverse (90°) bundlesand dense matrix regions (s=- 0.12% to0.2%) 3. Finally, cracksform inthe matrixsurrounding0° bundles(s> 0.2%) The microcracks formedat largepores inthe matrixandin90° tows are sometimes referred toas tunnelcracksp'_. A slightreduCtioninelasticmodulus wouldbe expected from suchcracking. When the microcracksintersect anddebond the fiber/matrix interphases ofthe load-bearingfibers, significantelasticmodulus reduction and unload/reload hysteresis behaviorwas observed161. This studywillcompare thestress-strainbehavior ofwoven Hi-Nicalon_u_,BN interphase, SiC matrix compositeswithmeltinfiltrated(MI) and CVI SiC matrices. Modal acoustic emission_]willbeused tomonitorthedamage accumulation andunload/reload hysteresis loopswill be performedtodetermine theonset ofdamage inthe 0° bundles as well asthe interfacial propertiesofthe two compositesystems. The findingswillthen be related tothe elevated temperature propertiesofthese twocomposite systems. EXPERIMENTAL Table I:Woven Hi-Nicalon TM. BN-intemhase. SiC Matrix ComPosites Composite Composite Processing Weave vf Details (loading direction) CVI SiC CVD BN interphase* 8 HS 0.15 CVI SiC matrix* MI SiC CVD BN-interphase* 5 HS 0.17 2pm CVI SiC overcoat* Si meltinfiltratedSiC matrix** allcompositeswere0/90layupswith8plies(17endsperinch) *DupontLanxide:BNinterphase;CVISiC **Carborundum:MeltInfiltrationSiC - specificprocessingdetailsareproprietary Table Idescribes the processingapproaches, vendors, andwoven architectures ofthe twocomposites compared inthisstudy. The tensile sample dimensionsand mounting arrangement are shown inFigure 1. The samples were cutfrom 152mm x229 mm plates intodogbone tensilespecimens where the lergth was 203 mm, the widthatthe ends was 12.7 mm andthewidthatthe gagesectionwas -10.2 mm. The plate thicknesses were nominally2.35 and2.2 mmforthe CVI SiC and MI SiCcomposites, respectively. Tensile testswere performedwitha screw-driven universal testing machine#. Glass fiberreinforced epoxy tabs (1.5 mmthick) were mounted on bothsides ofthe ' Nicalon andHi-Nicalon producedby Nippon Carbon, J_pan TM TM ' InstronModel 8562, InstronLtd.,Canton,Mass. sample inthe gripregions. The specimens were grippedwith rigidly mounted hydraulically actuated wedge grips. Aclipon straingage, with a range of2.5% strain and 25.4 mmgage lengthwas used tomeasure strain. Wide-band (50 kHz to2.0 MHz)==, highfidelitysensors (9.2 mm indiameter) were placedjustoutside the tapered regionofthe dogbone sample to recordacoustic emission events. Vacuum grease was usedas acouplant and electricaltape was transducer usedto mountthe sensors tothe wsaemreplree.coArcdoeudsutiscinegmaiss2i-ocnhawnanveel,forrns 25mmI "_--Clip-on-_ =65ram Extensometer Fracture Wave Detector_ (FWD). The FWD consisted ofapersonal computer(Pentium, 120 MHz) witha 12-bit, 30 MHz analogto digital acquisition board. Each sensorwas connected to apreamplifier andfilter triggermodule whichwas fed intothe computer. The preamplifier was setat 20 dB, the filtersignalwas amplified 3 riB,andthe filtertriggerwas amplified point of lead break 21 dB. The loadand strainwere also recorded withthe FWD computer. Figure1:Specimen configuration. The post-test analysis was performed onWave DetectorTM software providedbythe FWD manufacturer. RESULTS AND DISCUSSION Typical monotonic tensile stress-straincurvesare shown inFigure 2for the two composites. The stress-strain behaviorisverysimilartothat shown by otherresearchers for Nicalon/SiC woven compositesIs'l°].Bothcompositesfailed atsimilarstresses (~380 MPa); however, the MI composites always failed inthe gripswhereas the CVI composites always failed inthe gage section.The"knee" inthe stress straincurve occursata lower stress for the CVI SiC composites andthestrain tofailureofthe CVI SiCcomposites is greater than the strain tofailure ofthe MI SiC composite. The mechanical properties of the two composites are listedinTable II. Figures 3a and bshowmonotonic andhysteresis stress-strain curves ofthe two composite systems. The monotonic andhysteresisstress-strain curvesare almost identicalfor bothcomposite systems. The hysteresis stress-strain curveshave slightly larger strainsfor thesame stress compared tothe monotonic stress-strain curves. For the hysteresis experiments, AE activityoccurred inthe gage section upon initialunloading and atthe endofthe subsequent reloading. This indicatesa greater amount of damage inthehysteresis specimensm]andcould accountfor the increased strainincurredfor hysteresis tensile tests. Digital Wave Corporation, Englewood, CO 3 The maximum loopwidthwas determined for each unload/reload hysteresis =_ loopat thestress whichwas halfthe _ maximum stress ofthe specificloop(Figure _ =, 4). The onset of0°cracking isconsidered to _ =_ ....... - ogcrecautrewr thheannthzeeroh.ysTtherisesoiscclouorspawtid~th11is0 MPa i ',". for CVI SiCcomposites and ~ 140 MPa for , the MI SiC composites. Itisalsointeresting 0 ...J .... , .., ..i .... ,.. , .... ,,,i, 010 tonote thatthe loop widthofthe MI SiC 0 0-1 0.2 0.0 OA _l LI 0.7 _0 8_kl, % compositeswere greater than thatoftheCVI SiC compositesathigher stresses. Itwas observed from polishedsectionsoffailed Figure 2: Monotonic stress-strain composites that CVI SiC composites had curves for MI SIC and CVI SIC almost threetimes the numberofcracks of composites. 4OO 40 ,, MI(did not fail in gag_ .... .o CV] Std. :::.... ...-3.;_/ soo ..=:::;.._ 2J / /_. d/t'./" 1_ ,,-'_/"'.-/.:_,_.-'" "-l//,;s:d-'" / +,"/ 100 " / _'/_""" co,.i ,__//_ i_.'/_ "_ ............................ 0 .,. 4 .... , ,i,..i .... L.... • I/'_'O[ t O.2 O`S O.4 O.S O`S O.7 0.0 0 0.1 01 0.3 0.4 0.0 0.1 O.T 0.0 0.0 Strain, % Straht, % (a) (b) Figure 3: Monotonic and unload/reload hysteresis tensile stress-strain curves for MI (a) and CVI (b) composites. MI SiC (Table II). 0._IS The modal AE analysis was 0.04 . usedtodiscernthe locationofevents O.03S which occurred inthe gage sectionof the composites[lq. This isan important i0.0+25 factor todetermine because thenumber 0.02 of events recorded for the monotonic 0.045 experiments was ~ 7000 and 9000 for 0.01 the MI and CVI SiC composites, 0.006 respectively, whereas only- 1500 and 1250 events actually occurred inthe ComlpOelte Stress, MPa gage sectionfor thetwo composites, Figure 4: Hysterssis loop width for the respectively. Modal AE analysis, as two composites versus the peak stress of the hysteresis loop. 4 opposed to traditional AE", involvesthedigitizationofthe realAE waveforms andanalysis ofthe "modes", i.e. extensional (longitudinal)andflexural, of soundtravelthat make up the individualwaveforms. Inordertodetermine the location ofthe events, the change inthe speed ofsound must be knownfor the stress-straincondition. This was done, as described inref. 11, for the extensional wave from AE eventswhich occurred outside ofthetwo sensors during the tensiletest. Ifthespeed ofsoundisknown as afunction ofstrain,then, the elastic modulus asafunction ofstrain can bedetermined. The speed of soundofthe extensional wave, C,, isrelated toelasticmodulus, Eo,by: 1 'lh_ • normce(al) C. =[Eo/ p (1-v2)]lr2 (1) • 0.9 O norm Eo (MI) t.) where p isthe densityand v is 0.8 _rQ • norm, Co (CVI) Poisson's ratio. The Eoltzata 001_ I (_ • a norm Eo (CVI) givenstrain can be normalized _,,o 0.7 II ° goo byEo1_ats=0 and Ceata _ 0.6 o given straincan be normalized z 0.5 by C, ats =0. The normalized 0.4 ] _ t I I Eolr2andnormalized C,1r2 0 0.2 0.4 0.6 0.8 1 shouldhave the same relationshipas shownin Strain,% Figure 5for the unload/reload hysteresis tensile testswhere Figure 5: Normalized Eo'rz,{Eo(s)lEo(c=O)}'=, and Eowas determined from the normalized C., C.(s)IC.(s=O), plotted versus tangent modulus upon initial composite strain for the two composites from unloadingofthe hysteresis unload/reload hysteresis tensile experiments. loopt6].There isexcellent agreement between the reductionin modulusand the reduction inthe speed ofsound for bothcomposites. Infact the modulusand speed ofsound decreases are the same relative tothe initialmodulus (Table II)and speed ofsoundofthe undamaged material for bothcomposites, respectively. This isnot toosurprisingsince both compositesappear tohave reached ornearly reached matrixcrack saturation and bothcomposites failedat similarstress levelswhichwould result ina similaroverallstress distributionon the fibersoverthe entire gage length(assuming aconstant interfacial shear strength). The specificgage events were identifiedfrom (i) the difference intime ofarrival between the two sensors (Figure 1)for agiven event and (ii)the change inthe speed of soundasthe damage inthe material increases[11].Figure 6 showsthe AE data for thetwo composites tested monotonicallyintensiontofailure. The AE data showthe same general trend for thetwo composites. Initially,rapidAE activity isoccurring uptothe "knee" inthe stress straincurve. The rate ofAE activitythen decreases substantially. Forthe CVI SiC composite, an increase inthe rate ofAE activityoccurredjustpriorto "In traditional AE, the real sound wave is fittoa damped sine curve and AE parameters such as counts, amplitude, duration, energy, etc.., are tabulated and saved. Resonant frequency transducers are used withtraditionalAE which cannot detect the wide frequency range of the real waveform thatisproduced by a fractureevent. See references 7 and 11forgreater detail concerning the differences between the two types of AE analysis. failure; whereas thisdid nothappen for the MI material. TheonsetofAE activityinthe gage sectionoccurred atlower stresses andstrainsfor the CVI composite(Table II). However, thetotal amount ofAE events recorded was greater for the MI composite. It shouldalsobe notedthat the =loudest"orhighestenergy events occur predominantlyat the lower strainswhere the rate ofAE activitywas mostrapid["]. Table I1: Results from Tensile Tests Composite Ultimate Elastic 1"tAE 1't0° Cracking Avg. Avg. [#tests] Strength Modulus Noise Stress (MPa) Crack I[ (MPa) (GPa) Stress Spacing (MPa) (MPa) [LCF stress]# (mm) CVl Std. 370 + 5 195+5 55 - 110 0.34 36+8 [2m, 2h] [103] MI* >390_+5 214+6 74 ~ 140 0.87 11+3 [4m, 2h] [143] #LowCycle Fatigue (2hrhold, unload/reload cycle) runout stress(500 hours) *did notfail ingage section The tapering offoftheAE activityis _. mostlikelydueto thedecrease inmatrix =. _°°,-°° crackingormatrixcrack saturation (- 0.35% _, strain). The inflectioninthe stress-strain | °°°°'*_ _ Std curveisalsoindicativeof matrixcrack == ooo..Oo-' "=! saturation. The occurrence ofAE activity | i nearthe ultimatestrength for theCVI material isprobablydue tothe initiationof fiberfailures inthe gage sectionm]. This did ' ..... ,....... .i .... J,,, o o.1 o.,1 o.,1 0.4 o.S o.$ o.7 o.II o.il nothappen for the MI material because itdid 81nd..% notfail inthegage. Figure 6:AE activity for monotonic Also notedon Figure6 isthe point stress strain curves of two composites. on the stress-strainandAE event-strain curveswhere theonsetof0°crackingwas determined (Figure 4). Note that ~350 AE events occurred priortothisstressinthe CVI material and~200 AE events occurred priortothisstress inthe MI material. Forthe CVI material,the number ofAE events occurringpriortothe onset of0°cracks accounts for ove_25% ofthe total numberofAE events occurringinthe gagesection. Evidently,a significantnumberofmicrocracks in the matrixand900bundles were formed inthese materials priortotheextension or formation ofcracks which significantlyintersectthe load-nearing fibers. The gage sectionsofthe failed composites were cutand polishedlongitudinallyin thegage section todetermine theamount andextent ofcracking(Figures 7and 8). For bothsystems, each crack couldessentially be tracedcor_pletelythrough the composite from one edge orsideofthecomposite tothe other. The cracksappeared to be transverse withinaply,but could deviate upordown between plies. The numberof crackswere counted over the lengthofthe gage section.ca;ndthe average crack spacing was found tobe 0.34 mm and0.87 mmfor the CVI and MI composites, respectively. In 6 otherwords,nearlythreetimesasmanycracksoccurredinthegagesectionfor the CVI composites compared tothe MI composites. The most striking difference between the two composites isthe porosityofthe CVI matrixcomposite. Inthe CVI composites, similartorefs. 1and2, crackingappeared toemanate primarilyfrom the large,jagged regions (figure7b) ofthe pores whichexisted between 90°tows andin-between pliesoron the surface atthe intersectionofa 0 and90°tow. Forthe MI composites, suchpores andjagged regionsdidnot existand itwas impossibletodiscernwhere the originofthe cracks occurred. Presumablythey occurredat surface flaws, large matrixregions, and/or inthe 90° bundles. The interfacial shear (a) (b) stress,_,was determined from the Figure 7: Lower magnification (a) and higher hysteresis loopwidth,6s,,=, and magnification (b) optical micrographs of polished average matrixcrack spacing, longitudinal section of CVI composite tested intension to failure. Arrows indicate cracks. dc,fOrthe composites from the relationshipl6]: = (Gp=/28s,,=x[)bz(1-alv_)=(RJdc)/{4v_E_](2) where, apisthe hysteresis loop peak stress,b=andat are coefficientsgiven by HutchinsonandJensen112]R,fisthe average fiber diameter (13 microns), and Emisthe elasticmodulus ofthe matrix(400 GPa). Inordertoapproximate dcfor each hysteresis loop,the average crack spacing was assumed tobe indirectlyproportionalto theAE activity(similarto Figure 6). Forthe apof agivenhysteresis loop,the finaldcmeasured afterfracture was multipliedby the finalnumberofAE events inthegage anddivided by the numberofAE events which occurred inthe gage up toGp. The larger86.==anddefor the MI composites resulted ina• value lessthan one thirdthat ofthe CVI composites. Thiscouldbe dueto differingresidual stressstatesfrom the different matrixprocessing routesused tofabricate the composites. However, itis mostlikelythat thedifference in was due totheformation ofa carbon layerbetween the fibers andthe BN interphase in the MI composites. The processingtemperature for MI SiC isat leastthat ofthe melting temperature of Si(~ 1410°C). The Hi-NicalonTMfibers undergosome decomposition above about 1400°C which results intheformation ofa carbonaceous layeratthe surface ofthefiber_131T.ransmission electron microscopystudies on similarcomposites have shownthata thincarbon layerdoes existbetween the fiberand theBN for Hi- 7 Nicalon_/BN/MSIiCcompositeasndhasnot been observedfor Hi-Nicalon_U/BN/CVI SiC compositesI"1. The nature ofdamage accumulation isexpected to influence theelevated temperature lifepropertiesofCMC's. Some data existszls-_6f)or low-cycle fatigue, LCF, tests performed on these same types ofcompositesat 815°(3 inair. The LCF cycle consisted ofa loadingstep tothe peak stress,a2 hour holdatthe peak stressfollowed byan unloadingstep down toaminimum stress value. The run-outcondition for these experiments was setat 250 cycles(500 hoursatthe peak stress). Itwas found that the run- out stressesfor the CVI and MI compositeswere 103 MPa and 143 (a) (b) MPa respectively. These are Figure 8: Lower magnification (a) and higher nearly identicalwiththe onset of0° magnification (b) optical micrographs of polished crackingasdetermined from the longitudinal section of MI composite tested in hysteresistensile tests (Table II). tension to failure. Arrows indicate cracks. Evidently,for thiscomposite life criteria,the stresswhere the onsetof crackingoccurswas notlifelimiting. Instead, itwas the stresswhere crack growthorcrack formation resultedinthe intersectionofthe load- bearing fibers. The LCF peak stress was about twicethe:stress atwhich crackingwas firstdetected inthegage sectionofthese composites. Tqisofcourse assumes thatthe onsetofcrackingat815°C isthe same asat roomtemperature. Lipetzkyetaitl°] have shown thatfor aNicalon-CVI SiC composite themodulus,proportionallimitanddamage regime slopeofthe tensile stress-straincurve upto 1200°C isessentially unchanged. Since these properties do notchange itisreasonable toassume thatthe onsetof cracking isessentiallythe same overthistemperature ra_lge. CONCLUSION The mechanical propertiesanddamage accumU,ationofHi-Nicalon, BN interphase woven composites withCVI SiCand MI SiC rrlatriceswere determined and compared. The ultimateand elasticpropertiesofthe twocompositeswere similar; however, the onsetofcrackingoccurred atlower stresses andstrainsfor the CVI compositesystem. This wasdue tothe large,jagged poreswhichexist inthe CVI SiC matrixandwere the site ofinitialcrackingfor these composites. MI SiC matrix composites didnot have such pores. Modal acousticemissionwas found tobe auseful techniqueto monitortheonset andcontinuum ofdamage which occurredinthegage section ofthe tensilespecimens tested. The decrease inthe speed ofsoundas increasingdamage occurredwithincreasingstress couldbe usedtodetermine the elastic moduluswithincreasingdamage accumulation. Fewer cracks occurred inthe gage sectionfor MI compositesdueto the lower interfacialshear strengthofthese composites. Itwaspresumed thatthe lower interfacial strengthofthe MI composites was duetothe formation ofacarbon layer atthefiber matrix interphasedue tothe decompositionofthe fibersduringprocessing. Finally,the stresses for bothcomposite systemswhere cracks intersectedthe load-bearing bundles (0°cracking) couldbecorrelated tothe life-limitingstresses from 8 elevatedtemperaturleow cycle fatiguetests performed inairat 815°C; even though crackinghad occurred inthe compositesatstresses whichwere halfthe stressfor 0° cracking. ACKNOWLEDGEMENT This work was supported by the HITEMP program at NASA Lewis Research Center, Cleveland, OH. REFERENCES 1. L.Guillaumat and J. Lamon, "Multi-fissuration de Composites SiC/SiC," inRevue des Comoosites etdes Materiaux Avances, vol. 3, pp. 159-171 (1993) 2. L.Guillaumat andJ. Lamon, "Probabilistic-Statistical Simulationof the Non-Linear Mechanical Behavior ofaWoven SiC/SiC Composite," Comp. Sci.Tech., 56,803- 808 (1996) 3. D.S. Beyerle, S.M. Spearing, andA.G. Evans, "Damage Mechanisms and the Mechanical Properties ofa Laminated 0/90 Ceramic/Matrix Composite," J. Am. Ceram. Soc., 75, 3321-30 (1992) 4. S. Ho andZ. Suo, "Tunneling Cracks inConstrained Layers," J.Appl. Mech., 60, 890-94 (1993) 5. Z.C. Xia, R.R. Cart, andJ.W. Hutchinson,"Transverse Cracking inFiber Reinforced BrittleMatrix,Cross-Ply Laminates," Acta Metall. Mater., 41 [8] 2365-76 (1993). 6. J.M. Domergue, F.E. Heredia, andA.G. Evans, "Hysteresis Loopsand the Inelastic Deformation of0/90 Ceramic Matrix Composites," J. Am. Ceram. Soc., 79[1] 161-70 (1996) 7. M.R. Gorman, "New Technology for Wave Based Acoustic EmissionandAcousto- Ultrasonics," AMD-Vol. 188, Wave Prooaaation and Emerging Technologies. ASME, pp.47-59 (1994) 8. M. Leparoux, L.Vandenbulcke, S. Goujard, C. Robin-Brosse, and J.M. Domergue, "Mechanical Behavior of2D-SiC/BN/SiC Processed by ICVI," Proc. ICCM-10, Vol. IV, 633-640 (1995). 9. C. DroillardandJ. Lamon, "Fracture Toughness of2-D Woven SiC/SiC CVI- Composites withMultilayered Interphases," J.Am. Ceram. Soc., 79[4] 849-58 (1996) 10. P. Lipetzky, G.J. Dvorak, N.S. Stoloff,=Tensile Properties ofa SiCf/SiC Composite," Mater. Sci. Eng. A216 pp. 11-19 (1996). 11. G.N. Morscher, "Modal AcousticEmissionof Damage Accumulation inaWoven SiC/SiC Composite," submittedto Comp. Sci.Tech. 12. J.W. Hutchinson and H.Jensen, "Models of FiberDebonding and Pull-Out inBrittle Composites withFriction," Mech. Mater., 9, 139-63 (1990) 13. G. Chollon, R. Pailler, R. Naislain, F. Laanani, M. Monthioux, and P. Olry, "Thermal Stabilityof aPCS-derived SiC Fibrewitha LowOxygen Content (Hi-Nicalon)," J. Mater. Sci., 32, 327-47 (1997) 14. J. Brennan, unpublished research. 15. D. Brewer, NASA Lewis Research Center, unpublished research. 16. G.N. Morscher, "Embrittlement ofHi-Nicalon BN, CVI and MI SiC Matrix TM, Composites inAir at Elevated Temperatures, "to be presented atthe 1998 Annual American Ceramic Society Meeting.

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.