Evaluation of the cure kinetics of the wood/pMDI bondline

Micro-ciiclectric nnnlysis (pIIEA) and diKerentia1 scanning caloririictry (DSC) were used to monitor cure of polymel-ic ciiplicnylmethane cliisocyanate (pMDl) resin with wood strands in a snturittcd steam environment. A first-order autocatalyzed kinetic model was employed to tieternline kinetic ~?ar:ln~ete~-s. The kinetics were found to follo~v an Arrhenius relation. A single ramp DSC tccliniy tie and pDEA produced models tliat predicted si~nilar estllts at higher cure temperatures, but the IDEA-based model predicts a lo~iger curc time at low temperatures. Thc isothcrll~al pDEA method yields higher activation energies and Arrhenius frequency factors than models based on single DSC ramps. A ~nodification to ASTM E698 was made to conform to the assntnptioti of autocatalyzed kinetics. Tlie ~nodifieci ASTM E698 method predicted an earlier end of cure than the pI)EA-based moclels anci was in agreetncnt with 13SC' results obtained by partial cure expt'riri~cnts. The activation energies anti freclncncy factors for the dilTerent cure monitoring methocis arc sensitive to clifferent stages of cure. ( . 2001 Elsevier Science Ltd. A11 rights reserveci. tic~~~ri.ordc: 13. Wood: I'olyn~er-ic isocy;rilatc; C'irrc kinetics; Iliclcctr-ic anirlyris Wlieli producing wood co~iipos~tcs, the pniiiary matiufactui~iig objectlvc is to rnlnlinlzc the t111ie to develop adequate rncclianical strength for rcslsting delamination ~13011 openlng the press I-1owevc1-, the wood/adhcs~ve system 15 exposed to I-ap~dly chaiig~iig c o n d ~ t ~ o ~ i s of prcssui-c, steam, and licat d u r ~ n g m,tiiuf;~cturc The pMD1 reacts with tlic water vapor upon C X ~ ~ ~ L I I ~ The e;irIy part of curc is ctom~nated by chain extciislon of pMD1 preppolymers and llttle 15 contributed to strength devclopmcnt [I]. The latter stages of ctrrc arc dominated by the dcvelopmcnt of a polymer network Iluring the development of the tlirce-d~meiislonal network i t d h ~ s ~ v ~ stlength ;rlso develops. Tlie curc of tlic pMI3I acllics~vc In t h ~ s system 1s a separate iss~te than the I I~LLI I -e of tlic a d l i o ~ v c bonct However, the bond nncl cult ,Ire relateel and slmultaileonsly occur rlng proccssc$ Kliict~c mocicllng 1s one mc~ins tll'it car1 be i~sed to op t~~ i i i ze the cure process Controversy abounds ovei tlie ex'lct nature of Ilic curc of polyrne~ic dtplicliylrnethanc dilsocy,tnntc (pMDI) on wood substr'ites pi-oduced under reali\tlc procc\sing env~ronmcnts [2-61 The possible ~ e , r c t ~ o ~ i ptoclucts ~ticlude polyurea, b~urct/polyui-ct, allophaiintc, and polyurctha~ic boilcis. The coinplex nature of the reaction makes ~iicchanlstlc approaches to modeling extremely difficult and favors plieiiorncnologicd methods. However, phenoniciiolog~cal rnctliods at-c only accurate ~f tlic test coiidi11011s ttsed to geiieratc the k ~ n c t ~ c data are real~stlc to those 111 nianufactunng. Few nietliods used t o evaluate cure can be applied 111 ~ -e~ t l l r t~c processing cnv~ronmcnt\ [7-91. Eot pM1)Ibonded wood composttc5 the prim'try challenge 1s to coiitain ,tnd cv ' t l~~atc the cure in a steam cnv~ro~iiiiciit that 1s produced 111 hot-pressed panels. Difl'ercnt~al scanliiiig calor~metcr (1)SC) 2nd rn~cro-dielectric analysts (1.111EA) arc two inctliods wliere volat~le gases call he cont,tlned durlng curc evaluat~on. A saturated cnvlronInelit can he created i ~ ~ s i d c tlie I>SC pan. but incclianlcal prcsstrrc callnot be applied to the sample as in manuft~cturlng Nunicrous studle, have been conclucted U111ig culonmetrlc techn~clues to evaluate rcsln curc [3,10.11]. It 11'1s been sliowli that the curc of pMDI 111 a saturated steam eiivlioiiiiient can be illortitored 111 situ ~isliig pIlEA [ I ] ' Ihc change ln d~c lec t r~c slg~ial can then be ,tpplicd to ,I phenomeiiologlcal model of curc A basic ratc ecl~latioii (Eq. ( 1 ) ) rcl,ttcs tlic rate of cul-c at a constant tcml?et.ature to 'I fui~ctron ( / ( r ) ) of curc by a constant (10, where r is tlic degree of cure Tlic rate constant follo\vs the A r r h e i ~ i ~ ~ \ eqnation wlicrc 1.: 1s tlie activatioti energy, A is the Arrhen~us frey~icilcy f'~ctor. I< is the ideal gas constaiit, ' ~ n d T 1s the temperature. The Arrheiiiu\ frcyuciicy f:tctor relates the amount of collisio~is that need to occuiiii a unit time to carry out the rcactlon, and the activatioli energy describes the amount of energy needed to propagate curc Two types of phenorncnologicd kiiietic models ava~ l ~ b l e for modeling thermosetting adhesives arc nth or-der and ,iutocataly/ed 1121 The thennoset no deli arc rclated to the sate ccluation by f ( r ) . Iicact~oiis that have their greatest ratc of curc at tlie onset of the reaction arc charactcri/cd as nth ordcr. Foif ( r ) nn 12th-order reactlo11 is described by Eq. ( 3 ) , wliere 11 1s the ordcr of the reaction. f ( x ) = (1 x)" (3) I11 contrast, autocataly/cd k~netics i \ character~/cd by duldt rcachii1g ' 1 maximuin at 30-40'Xj cure 1121. I11 general, a~itocataly7ed reactions take the followiiig fo rm wl~cre t n dnd n sum to cclual the over-all rcactlon order. Most conimonly, a ~ n a x i i n ~ ~ i i i ratc of curc is rcacl~ed ,tt 45-5S1X, o f c ~ i r c Altllougl~ this bchav~or docs not str-ictly follow t l ~ c spcc~licd enterla, ~t ccrta~nly 1s niorc ,tccuintc than an nth-order ki~ietics approach. Standardi/cd methods exist for evaluating kinetic paramctcr-s of iitllordcr redct1oi1s (ASTM E69X-79 1979). However, these methods 111~1st be ~liodified for c t ~ t o ~ L ~ t a l y / ~ d models 112-141. The ovcl-all oblcctivc In modcli~ig the c u ~ e process of pMIII 1s to optiini/c proccssiiig par;tnicter-s to \hosten prcssiiig cycle4 To o p t ~ m ~ / e a procc\s using pMIII 'idl~esives, t\+o events 111ust be consrtfcrcd in the inodcl ( 1 ) the point at which the bond strength can i ~ ~ 1 s t panel dclamiii,ttion and (2) rcduct~on of fl-cc i\ocy,tiiate level4 for safe storage Tllc important ,i\soc~,ttron between the chemicnl and iiicchan~cnl pi-opcrticj in inodeling call be illustrated wlicrc b o t l ~ c o n d ~ t ~ o n \ 11i~1st be \;it~sIied. 111 this casc, ,I pllc~iotncnologic;il ,tpproacli thut cncoinpasse\ the over:tll proces\ may be iiiorc ~14eful Ili'tn 21 niecl~~tnistic inodel that may not .~ccurately I-el'itc to mcch,~n~c,il PI-opcrty dcvelopiiici~t T o ,~chicvc this ovci2111 goal the spccific oblectivcs ,Ire: I Analyt~cally model the C L I I ~ of pMI>I In a saturated stcam cnv~roiiincnt 2 Fniploy tlie L I ~ C of pDEA i t i cv,tluatii~g curc for ~nocicl developmciit 3 litllrrc ail autoc,itnly/cd inodcl ernploy~iig otily onc rate constant 4 Comp'lre single atid multiple ramp calori~nctirc method\ of modeli~ig. 5 lielate the kinetic ~nodcls to physrcal phenorncnon 3. Methods and materials Aspen (Poplrlu\ ~ I . ~ I ? z I I / o I ( / ~ \ ) [lake\ werc obt,iiticd and p i ~ p ~ ~ i ed for 11IIT.A 111 A cornniercially available pMD1 was sprayed on the aspcii Il'lkes at 3,5,o1 7'!4) levels ((>veil dry mass basis). The llakc-pa~r assemblies were pressed at isothermal teinperature5 of 1 10, 120, 130, and 140 C in the presence of saturated steam. pDEA was performed by a Micromet Eumetr~c Systein I11 dielectric analy7cr with an 1I)EX scnson in ordcr to observe the change in cond~icttvity while flake palrs werc pressed in an enclosed steam-producing press [I] . A criterion was established for the oiiret and ciid of cure The onset of curc was take11 'is the inaximum coiiductivity, IT,,,,,, and the end was taken 'is the absolute minlmuin value of the \lope to obta i~i an ovciall chaiigc 111 conductivity, Aa The miniilluili v'tluc for tile slope was calculated by i1unicrically taking the dciivcttive of the log a ( / ) cur-ve w ~ t h respect to t i ~ n c Where, the ciid of cure occurs wlieil the slope is 0 01 log(sieincn\)s ' The degree of curc at LI time, t , call then be cCilciilated by log a,,,,,, log IT(/) x( / ) = -----log AIT In preparation SolIISC analysrs, pMD1 was appltcci to two 'Ispen Ilakes '11 ' 1 7% Icvcl. A 25-mg s a i ~ ~ p l c was I-crnovcd with :I small hole puncli fro111 the Ilctke pair. F o r ,~rialysis, the sainplc wa\ placed 111 a hcr~nctically waled, stainless-steel DSC pa11 with 2 111 of di\trlled water to 171-ovide enough moisture to create a sat~lrated stearn cnviro~imcnt The samples were analy/ed in tlic IISC while they werc heated fi-oin 30 to 200 C at heatiiig rates (1)) of 1 , 5, 10, 15. arid 20 C m~ri. A fi-csh specimen was urcd for each DSC r~tmp. The cxotlicriinc heat geiier'ltion was then measui-cd foi each I-amp The degree of curc w'ts calcul,ttcd from the cxothcriu as wherc Q(t) I \ the residual heat at time, / and Q,, 1s the total heat of reaction Q,, was calculated by n~~mcrically IiitcgiLlting the power spectra ~vith I-cspcct to tllne using the tr,tpc/oicinl inctliod. I"irt~al curc DSC cxpenments 'tlid lap-shear expertmelit\ werc prepared s~multancously 111 an enclosed stearn-productng press [ I ] The samples werc pressed at ' 1 spectficd tetnpcl-atulc ctnd time to obtatn 0-100'Xr curc '1s detctm~~-tcd by dtclect~tc t~ictlysts A 25-mg satiiplc wcti theti rernoved fro111 ' 1 fl'lke pair 'ind ptcparcd as d~scussed ,ibovc for DSC A ramp of 20 Clmtn w,ts perfol-n~cd and the exotliertntc heat me'isureci The x was then calculated '1s the ratto of heat gelicratloli of ' 1 pal t~ally cured sample to that of an L I I ~ C L I ~ C ~ sample at the \ame l ieat~ng rate Lap-shear speci~nc~is were pulled 111 tension 011 a untversal testing macht~ic at a rate of 1 27 mtn/mln ~lntt l fatl~lre Fa~lures were noted as adlies~ve, colics~vc, o r wood fa l ures An autocataly/cd k~nettcs rnodel (Eq. (4)) w'is used for both so thermal and ramp curlng s tud~es The collected data were used as Input Into the model to cstlmate klnettc parameters In the Arrhen~us rcl,tttonsh~p By ~ntegtatlng Eq. ( I ) cornpansons w c ~ c made bctwcctl parameters atid results previously obtatncd for pMD1 curc [I]. 4. Results and discussion A model of autocataly/cd kt~iettcs was used to dctcrmtnc k~nctlc parameters of the pDEA results Stilrtlng wlth the model for autocataly/cd k~nettcs 111 Eel (4), a I~t ic~tr expressloll may be obtatned Assum~ng a firstorder I-cactton whcrc 111 + 11 = 1, tlie equation can be rearranged and expressed ' 1s T h ~ s fortn can now be ~ t i cd to clctertii~tic I ' and r l . The rc l a t~onsh~p of I \ to tlic cure tctnpcraturc call be ~ ~ s c d to dctcrm~nc Art-hentus paratnctcrs As a n nltcrnat~vc to t h ~ s ~iiethod, the Art licntus rclatton cat1 be subit~tuted d ~ r cctly uito Eq. (7) to y~cld val~clatc the use of first-order ktnet~cs over Iitgl~etorders. . . I bus Ecl. (8) will be changed to Thts wtll alio t~iflucnce the rcactlon order where This t e ~ u l t for second-order ktlietlcs wets 'tchlevcd by Lam 1141 Ltliear regresstons were pelfortned to compare first'ind second-older kttlctlcs (Figs I c l~id 2) A h~gher coctlic~cnt of deterni~nat~on ( R ~ ) was achlcvecl f o ~ the ii~st-order model than second ordcl Otily the I~ncar pottlon of the ddtd wds lit to d l i ~ ~ e ~ i r reg ession to dctcr ln~~ie rate constants and reactlon orders Neat the end of cure the I-eactton becomes dtffuston controlled, whtch tc not accurately represented In thts model The values for rz and in were observed to be dependent on temperature (Flg 3) The overall reactlon order docs not change, but rn and II d o affect the rates of convcrslon (Eq (4)), whlch may have an Impact on ramp cure studles A Ilneal rcgrcsslon was pcrforlned to calculate the Arllicnius parameters (Fig. 4 and Table 1). lntegrat~ng Eel ( I ) (Fig 5 ) can fac~lttate a cotnpanson between the pledtcted and expert~nclital results for curc Altho~tgh tlie pledtctcd results are close to what 1s exper~~nentally observed, the lnodel pred~cts an earlletend of cure then wli,lt 1s expertmentally observed Furthct, the tiiodel 1s not seiis~ttvc to the rcstn load foltlie selected ~ c s ~ n levels Tlie expc~~inentally observed dtiTcrcticcs In re\lti lo,ids ,111se ldte In cure where dtfl'uslon control llkely occu14 [ I ] b R~ = 0 999 1, = ItlA + I (8) 2 1 kc]. (8) call tlicii be fit to the calculated curc w~t l i 111~1lt1ple $. u l~nea r regrcsslon Because the reactton order of 73 O ?? nl iil = I was assumed 111 t h ~ s d e r ~ v a t ~ o ~ i , '11i ;tlterti;~t~vc 1 t~ ie t l i~cl fatdc tc rm~n~t ig I-eactlon orders 1s by Jind~ng tlic ~iiax~mutii rate of cure. Take the dcrtvattvc of Eel. (4) with 2 rcipcct to (t), ytcld~ng the iecond d c r ~ v a t ~ v c of 7 wtth 3 respect to ttme, and evaluate at Y = 2 ,,,,,. whcrc y,,,, 1s the dcgrcc of cure at the maxtmutii rate of cure, 4 , p -. --i--7 1 D 6 4 2 o 2 4 6 a Y,,, ,, = /?I. (9) in((l r x ) / ( x ) C ' o m p , ~ ~ s o ~ i s to second-orcier k~nettcs can be made by I lg 1 I i l \ l o l d o i L~~~~~~ plot fo I 120 (l ,ot~cll , , 511/;, 1 , ~ ~ ) ~ wit^, substttutlng ill + r l = 2 S~lch a conipartsoti 1s uicd to I ~ i ~ c ~ r ~ Icglc\\lon loi d , ~ t , ~ oht,rineci hg p I ) I A 4 2 Mzrltrple , ~ n ~ p r l m u ~ n degree of cure (Y,,,~,) occurring a t IIIIIC (f,,,,) , d r The ASTM E69X-79 method 1s 3 very accurate and / (~, , , . , .~=[: y l "(1 Y)" = 1: Ae ' ' I di ( I ? ) cffcctt.vc method for cvaluat~ng the k~ilcttcs of tlicl-mally r instable systems L12.151 K ~ n e t ~ c parameters call bc estlu~~~~~ a constant lieat lng rate (11 = d ~ , d , ) , m,~tcd froin mult~ple ramps w ~ t h dtJTerent heat~ng rates Eor ASTM k698-79, a first-order kirlet~c model 1s fit to d Y A 2 lc : 'pe ' " I d 7 (13) the c x o t h e r ~ n ~ c temperature peak (T,,) In tlic 1)SC data The bas15 for t h ~ s model 1s that the peak occurs at ;t constant cure, Y,,,,~,, and that an nth-order reactloll 15 The r~ght-hand ~ntegral can be solved colls~dcnng a subfollowed 1141. Both assumptions are often v~olatcd by s t ~ t u t ~ o n Solf r = EIR7 and u,, = EIRT,, autocataly~cd react~ons. The v~olattonc are ~llustrated by f ( ~ , , l , i Y ) the temperature dcpcndencc of Y,,,,, that was observed forthe I-amps (Table 2). T'herefor-e, the ASTM method was mod~lied to lit autocatalyzed k ~ n e t ~ c s Starttng wlth lil-st-ordcr kinetics and Eq. (4), and consider~ng the maxValues for p(u) have been tal?ulated and for ~ t s linear i~iterpolation 1161. I'nme 1121 ~1t1117ed Doyle's so lut~on to obta111 a hnear form for 60 2 14) 2 20