VOLATILE ORGANIC COMPOUND EMISSIONS ARISING FROM THE HOT-PRESSING OF MIXED-HARDWOOD PARTICLEBOARD

COMPOSITES AND MANUFACTURED PRODUCTS VOLATILE ORGANIC COMPOUND EMISSIONS ARISING FROM THE HOT-PRESSING OF MIXED-HARDWOOD PARTICLEBOARD TING JIANG DOUGLAS J. GARDNER* MELISSA G.D. BAUMANN* ABSTRACT This study investigted the effects of dhesive type nd press vribles on the voltile orgnic compound (VOC) emissions rising from hot-pressing mixed-hrdwood prticlebord. Three dhesive types, ure-formldehyde resin (UF), phenol-formldehyde resin (PF), nd polymeric methylene diisocynte resin (pmdi), were evluted in this study. A2 5-1 frctionl fctoril design ws used to evlute the primry effect of five press vribles (press temperture, press time, mt resin content, mt moisture content, nd bord density) nd their interctions. A totl of 27 chemicl compounds were identified nd quntified in the VOC emissions using four nlyticl techniques. Formldehyde, methnol, cetic cid, nd HMw VOCs with hexnl being the predominnt chemicl were the mjor compounds comprising the VOC emissions. The results reveled tht formldehyde nd methnol emissions from UF prticlebord, s well s the methnol emissions from PF prticlebord were the most bundnt components of the VOC emissions, nd they contributedbout 92 nd 72 percent of the totl identified VOCs, respectively. Lower levels of formldehyde nd cetic cid were relesed during the hot-pressing of PF prticlebord. Acetic cid nd HMw VOC emissions were the most bundnt components of the VOC emissions rising from pmdi-bonded prticlebord. pmdi significntly reduced the methnol emissions from the mixed-hrdwood prticlebord. The most significnt press vribles controlling VOC emissions were press time, mt resin content, press temperture, nd interctions mong these three vribles. These press vribles hd different effects on the individully identified compounds bsed on the dhesive type used. In generl, formldehyde emissions rising from hrdwood prticlebord hot-pressing were significntly lower thn those from softwood prticlebord. However, formldehyde emissions from UF-bonded hrdwood prticlebord were significntly higher thn from the softwood UF-bonded prticlebord. Tody, there re 48 prticlebord mills in the United Sttes with totl nnul cpcity of 9,606,000m3 (Anonymous 2000), nd the demnd for prticlebord is projected to continue. With implementtion of the 1990 Clen Air Act, voltile orgnic compound (VOC) emissions nd hzrdous ir pollutnts (HAPS) rising during the hot-pressing of prticlebord re mjor concern of prticlebord mnufcturers. The mesurement of VOC emissions from hotpressinghsbeenthe subjectofrecentreserch(wngndgrdner1999;ncasi 1999, 1996;Wolcottetl. 1996;Broline et l. 1995; Crlson et l. 1995). fctors, such s wood species, dhesive type, nd pressing conditions. The d- hesives nd the wood itselfcontin vol tile compounds, which my be emitted The VOC press emissions in wood composite mnufcturing rise from both the wood prticles nd dhesive binder, nd re dependent on numerous The uthors re, respectively, Reserch Assistnt nd Professor, Univ. of Mine, Advnced Engineered Wood Composites Center, Orono, ME 04469; nd Reserch Chemist, USDA Forest Serv., Forest Prod. Lb., One Gifford Pinchot Dr., Mdison, WI 53705-2398, This study ws funded by USDA/CSREES/NRICGP Grnt 9702592. The uthors thnk Dr. Brry Goodell, s well s Johnn Brzier nd students in Dr. Cole's lb t the Univ. of Mine, for rrnging smple nlysis. The uthors lso thnk Steve Verrill t the USDA Forest Serv., Forest Prod. Lb. for ccommodting prt of the SAS progrm relting to the sttisticl nlysis. The uthors re grteful to Rodmn Industries, Southestern Adhesive Co., Neste Resin Corp., nd Byer Corp., for supplying the experimentl mterils. This pper ws received for publiction in August 2001. Reprint No. 9351. * Forest Products Society Member. Forest Products Society 2002. Forest Prod. J. 52(11/12):66-77. 66 NOVEMBER/DECEMBER 2002

during mnufcturing. Additionlly, therml degrdtion of the wood my contribute to VOC emissions during mnufcturing. Lower levels of VOC emissions re expectedfromhrdwoodmnufcturing processes compred to VOCs from softwood mnufcturingprocesses (NCASI 1996, 1989) due to the lrge mount of voltile extrctives in softwoods. Reserchers evluting emissions of nine different wood species found tht the primry emissions re: terpenes from softwoods nd cetic cid from hrdwoods (Risholm-Sundmn et l. 1998). Crlson et l. (1995) ttributed theemissions of cetic cid, formic cid, methnol, nd furfurl during hot-pressing of prticlebord to xyln decomposition s the thermlly unstble hemicellulose is broken down. Previous reserch on emissions from southern pine press opertions found tht the wood extrctives -pinenend b-pinene re the predominnt VOC emissions (Wng 1999, Crlson et l. 1995, Ingrm et l. 1994), lthough degrdtion nd rection products of the wood nd extrctives, e.g. hexnl, pentnl, nd pentnol, were lsopresent. Adhesive type my hve significnt effect upon emissions during pnel mnufcture both becuse of the chemicls present in the dhesive nd the interction ofthe dhesive with the wood. The primry dhesives used in wood pnel mnufcture re ure-formldehyde (UF), phenol-formldehyde (PF), nd polymeric methylene diisocynte (pmdi). During hot-pressing of wood pnels bonded with UF or PF dhesives, free-formldehyde nd methnol re relesed becuse both UF nd PF resins contin smll mounts of methnol nd free-formldehyde (Pizzi 1983). Using verge test dt from four UF-bonded prticlebordmills,reserchers(ncasi 1999) hve extrpolted tht n nnul production of 265,000 m 3 (150 million ft. 2, 3/4 in. bsis)relesespproximtely 60 tons/yr. of totl HAPS, 40 tons/yr. of methnol, nd 20 tons/yr. of formldehyde emissions from prticlebord hot-pressing. pmdi is non free-formldehyde contining dhesive. It forms chemicl bonds by recting with ctive hydrogen toms nd wter on the wood surfces nd cures very fst. pmdi hs become replcement for conventionl formldehyde dhesives within woodbsed pnel mnufcturing industries since the lte 1980s. However, there is possibility tht some lower moleculr weight pmdi my be relesed during hot-pressing, nd there re concerns bout worker helth when using pmdi. Informtion bout othervoc emissions from pmdi pnel production is limited. VOCpressemissionsresignificntly ffectedbypressingconditions. Wolcott et l. (1996) studied the pressing vribles ffecting formldehyde nd methnol emissions from UF-bonded prticlebord with mixture of Dougls-fir nd southern pine furnish. They found tht formldehyde emissions incresed with incresing press time, plten temperture, moisture content, resin level, nd formldehyde-to-ure (F/U) mole rtio. Methnol emissions incresed with incresing press time nd moisture content (MC). Broline et l. (1995) employed two-levelhlf-frctionl fctoril design to determine the effects of process vribles on formldehyde, methnol, ndmmonipress emissions from UF-bonded mixed-pine prticlebord production. Five press vribles (MC, resin loding, resin mole rtio, press time, nd pnel density) were investigted. They found tht MC, resin loding, resin mole rtio, nd press time were significntfctors impcting VOC emissions. Crlson et l. (1995) investigted the effect of press vribles on VOC emissions from lbortory PFbonded spen oriented strndbord (OSB). They reported tht formldehyde emissions incresed with incresing press temperture, mt MC, resin tretment, resin free formldehyde, nd pressingtime. Above ll, wood species, dhesives, nd press conditions re significnt fctors ffecting VOC emissions from prticlebordhot-pressing.however,mny of the previous studies focused on the VOC press emissions from softwood furnish. There is lck of informtion on the type nd quntity of VOC press emissions from hrdwood furnish, nd very little work hs been done in evluting the effects of dhesives nd pressing vribles onvoc press emissions. The objectives of this reserch focusedon: 1) Identifiction ndquntifiction of hrdwoodvocpressemissions; 2) Evlution ofvoc emissions from hrdwood-dhesive nd hrdwood furnish subjected to therml tretment; 3) Evlution ofthe effect ofdhesive type on hrdwood VOC press emissions; 4) Evlution of the effect of five mjor press vribles nd their interctions on hrdwood VOC press emissions. MATERIALS AND METHODS MATERIALS Hrdwood prticles used in this study were obtined from Rodmn Industry, Mrinette, Wisconsin. They were composed of 60 to 70 percent northern hrdwoods, 30 to 40 percent spen, nd smll mount of softwood. The MC of the prticles ws bout 12 percent. The dhesives selected were UF resin (product no. 9-2011) from Southestern Adhesive Compny, PF resin (product no. NRC139b) from Neste Resin Corportion, nd pmdi resin (product no. G541) from Byer Corportion. The chemicls used in this study for the collection nd nlysis of VOCs were methylene chloride, formldehyde, methnol, formic cid, ceticcid, propionic cid, butyric cid, hexnl, -pinene, b-pinene, octnl, D-limonene, borneol, nphthlene, hexdecne, heptdecne, nd octdecne. Methylene chloride ws used for extrcting the VOCs from the hot-press process strem, nd performing liquidliquid extrctions with the resulting VOC-wter solutions. The other listed chemicls were used s stndrds in the vrious chemicl nlyses to be described in subsequent experimentl sections. PANEL MANUFACTURE The resin ws pplied to the prticle furnish in drum blender with spinning disk tomizer t 15,000 rpm. The trget size of the prticlebord ws 305 by 305 by 19 mm (12 by 12 by 3/4 in.) All mts were hnd formed in 305- by 305-mm (12- by 12-in.) deckle box, nd were pressed in the hot-press mintined t temperture of ± 1 C round the trgetsetpoint. VOC EMISSION COLLECTION During hot-pressing, VOC emissions were collected using collection system comprised of cul plte, two 250-mL scrubbers, nd vcuum pump. The two 250-mL scrubbers were chilled in cooler bth to pproximtely 5 C. The FOREST PRODUCTS JOURNAL VOL. 52. NO. 1 1/12 67

first scrubber contined 100 ml fresh distilled wter, nd the second scrubber contined 150 ml methylene chloride. Using vcuum oil pump, VOC emissions were pulled through the cul plte nd bsorbed in the two scrubbers. The collection of VOC emissions begn when the cul plte ws closed. The recovery efficiency of the collecting system ws greter thn 90 percent, bsed on previous experiments (O Neill 2000). Afterthe hot-pressingcycle ws completed nd the press ws opened, 20-mL wter solution ws tken from the first scrubber for determintion of low moleculr weight (LMw) compounds in the VOC emissions. The remining wter solution ws combined with methylene chloride solution from the second scrubber, nd ws extrcted in seprtory funnel. The combined solution ws extrcted twice with 30 ml methylene chloride. The methylene chloride extrction solutionws usedfor the determintion of high moleculr weight (HMw) compounds in the VOC emissions. VOC EMISSION ANALYSIS Four nlyticl techniques (Fig. 1) were utilized to identify ndquntify the chemicl compounds contined in the VOCemissions. Gs chromtogrphy/mss spectrometry (GUMS). GC/MS ws used to nlyze the HMw VOC emissions in the methylene chloride solution. The GC/ MS nlysis ws performed using Hewlett-Pckrd GC-MS system consisting of n HP 6890 gs chromtogrph nd n HP qudrpole mss selective detector. A 30-m HP-5 column with 0.25-mm ID, nd 5 percent crosslinked phenymethylsiloxne ws used to seprte the chemicl compounds. The GC injector temperture ws set t 270 C. The crrier used ws helium with flow rte of 0.7 ml/min. The GC oven temperture ws strted t 40 C nd held for 4 minutes. Then the temperture ws progrmmed to 280 C t 10 C per minute nd held for 8 minutes. The ionizer voltge of the MS detector ws set t 70 ev, nd the temperture ws set t 200 C. The scn rnge of moleculr weights ws 40 to 500 mu. The totl HMw VOC emissions were clculted by combining the estimted mount of the 50 lrgest peks, which ccounted for more thn 97 percent of totl VOCs detected in the GC/MS totl Figure 1. Anlyticl techniques for mesuring VOC emissions including extrct from the wter scrubber. TABLE 1. Press vribles nd their levels during hot-pressing of hrdwood prticlebord. Tretment levels nd Press Press Resin Mt Bord dhesives temp. time content MC density Low level ( C) (min.) - - - - - - - - - - (%) - - - - - - - - - - (g/cm 3 ) UF 165 3 4 12 0.71 PF 165 4 3 13 0.71 pmdi 165 3 2 12 0.71 Center point UF 182 5 6 13 0.11 PF 182 6 5 14 0.11 pmdi 182 5 3.5 13 0.71 FH1 182 5 0 13 0.11 FH2 182 6 0 14 0.11 Highlevel UF 199 7 8 14 0.83 PF 199 8 7 15 0.83 pmdi 199 7 5 14 0.83 Press conditions of net hrdwood. ion chromtogrph. The quntity of sions cn be found in the work of HMwVOCs wsclculted usingexter- O Neill(2000). nl stndrds. Ten chemicls (hexnl, Gs chromtogrphy/flme ioniz octnl,-pinene, b-pinene, limonene, tion detector (GC/FID). A GC/FID borneol, nphthlene, hexdecne,hep- technique ws used to nlyze the wter tdecne,ndoctdecne) wereselected solution smples. The nlysis ws pers the externl stndrds in this study. formed using Hewlett-Pckrd 6850 Ech stndrd compound ws selected gs chromtogrph nd n FID. The colbsedonmoleculrformuls,functionl umn used ws n HP-WAX cpillry groups, nd rnge of diverse chemicl column (15 m by 320 µm by 0.5 µm) structure.moredetilsonthequntific- bonded with polyethylene glycol. The tion methodology for the VOC emis- injector temperture ws set t 200 C 68 NOVEMBER/DECEMBER 2002

TABLE 2. The min VOCs rising during the hot-pressing ofmixed-hrdwood prticlebord. Compounds Boiling points Anlysis method Retention time Externl stndrd ( C) (min.) Formldehyde -19.5 ASTM D-5582 Formldehyde Methnol 64.6 Formic cid 100.7 Acetic cid 117.9 Hexnl 131 Propionic cid 140.7 -Pinene 156.2 Butyric cid 163.5 b-pinene 166 2-Pentyl furn 64 to 66 t 23 mm Octnl 171 3-Crene N/A Limonene 178 (E)-2-octenl N/A Nonnl 191 Borneol 209 2-Methylnphthlene 241 Tetrdecne 253.7 Longifolene 150 to 151 t 36 mm Hexdecne 287 Heptdecne 301.8 Octdecne 316.1 Eicosne 342.7 Heneicosne 356.5 Docosne 368.6 Tricosne 380.2 Tetrcosne 391.3 N/A = not vilble, with n injection volume of 1.0 µl. The crrier used ws helium with flow rte of 4.7 ml/min. The GC oven temperture ws initilly held t 50 C for 2 minutes. Then the temperture ws progrmmed to 200 C t 25 C per minute nd held for 2 minutes. The FID ws set t 300 C, n H 2 feed rte of 30 ml/ min, n ir feed rte of 400 ml/min., nd helium mkeup feed rte of 25 ml/min. Methnol ws identified nd quntified with externl stndrds in the GCnlysis. High performnce liquid chromtogrphy (HPLC) nlysis. The nlysis of LMw cids in the wter solutions ws crried out using n HPLC with UV VIS spectrophotometer(hitchil-4500 diode rry). An Adsorbosphere C18 5µ HPLC column (150 mm by 4.6 mm) ws used, nd 0.1 mm H 3PO 4 ws selected s the mobile phse with flow rte of 0.5 ml/min. The bsorbnce t 210 nm ws mesured to quntify the FOREST PRODUCTS JOURNAL VOL. 52, NO. 11/12 GC -1.46 Methnol HPLC -3.51 Formic cid HPLC -3.95 Acetic cid GC/MS -4.05 Hexnl HPLC -4.89 Propionic cid GC/MS -7.13 Alph-Pinene HPLC -6.55 Butyric cid GC/MS ~8.04 Bet-Pinene GC/MS ~8.39 Hexnl GC/MS ~8.65 Octnl GC/MS ~8.72 Limonene GC/MS ~9.08 Limonene GC/MS ~9.68 Octnl GC/MS ~10.49 Octnl GC/MS ~11.54 Borneol GC/MS ~13.58 Nphthlene GC/MS ~14.76 Hexdecne GC/MS ~14.99 Hexdecne GC/MS ~17.24 Hexdecne GC/MS ~18.37 Heptdecne GC/MS ~19.45 Octdecne GC/MS ~21.45 Octdecne GC/MS ~22.39 Octdecne GC/MS ~23.29 Octdecne GC/MS ~24.13 Octdecne GC/MS ~24.96 Octdecne LMw cids. Four chemicls (formic cid, cetic cid, propionic cid, nd butyric cid) were selected s externl stndrds inthe HPLC nlysis. Chromotropic cid method (ASTM 0-5582). An ultrviolet visible spectrophotometer(hewlettpckrd 8452A diode rry) ws used to determine the mountofformldehyde in the wter solution, ccording toastmd-5582. EXPERIMENTAL DESIGN AND STATISTICAL ANALYSIS A two-level hlf-frctionl fctoril design (2 5-1 ) ws used to evlute the mjoreffectsoffivepress vribles. One center point with triplictes ws included in the design to check linerity, experimentl error, nd to provide hotpressingvriblespproximtingindustril production (Montgomery 1991). The five press vribles included press temperture, press time, mt resin content, mt moisture content, nd bord density. Tble 1 shows the press vri- bles nd their levels (low, center, nd high) forthree dhesives, ndpress vribles t center point for net hrdwood furnish. There re totl of 19 experimentl runs for ech dhesive. The run number ws rndomly chosen during the experiment. Sttisticl nlysis of the experimentl results ws crried out for ll clssified compounds from the VOC emissions. In the nlysis of vrince (ANOVA) of the dt, the vrible with probbility vlue less thn 0.05 ws considered sttisticlly significnt. Bsed on the ssumption of no interctions between vribles, the reltive effect of ech vrible ws clculted to esily compre effects on VOC emissions mong the press vribles nd dhesives. where Coef = the reltive effect of press vrible i; DY= themen effectofvrible i. The Fisher LSD multiple comprison test ws used for the comprison of VOC emissions from three hrdwood-dhesive furnishes nd nethrdwoodfurnish. RESULTS AND DISCUSSION VOC EMISSIONS CLASSIFICATION Twenty-seven chemicl compounds were identified in the VOC emissions rising from the hot-pressing of mixedhrdwood prticlebord (Tble 2). However, the compounds emitted were nlyzed by severl different techniques. Clssifictions ofcompounds in the discussions tht follow re bsed on the chemicl species nd the technique used to identify nd quntify them. HMW VOCs. The HMw VOCs re compounds tht were identified in the methylene chloride trp nd extrcts. These compounds were identified by GC-MS. Twenty-two chemicl compounds were included in this clssifiction, with the min compounds being hexnl, pinenes, 2-pentyl-furn, 2-octenl, nonnl, 2-methyl-nphthlene, nd longer chin lknes (C 14 - C 24). Under the press conditions ofthis study, hexnl emissions ccounted for 20 to 51 percent of HMw VOCs, nd it ws the predominnt single chemicl. Becuse of the limittion of the GC/MS nlyticl technique, the compounds tht elute erlier thn the methylene chloride do not pper in the GC/MS to- 69

TABLE 3. Summry of experimentl results of VOC emissions from hrdwood-dhesive furnish nd net hrdwood. voc Amount of VOC emissions compounds UF PF pmdi FH1 FH2 - - - - - - - - - - - - - - - - - - - - (mg/kg OD bord) - - - - - - - - - - - - - - - - - - - - Formldehyde Averge 396.6 8.3 0.3 1.0 1.2 SD b 17.5 1.3 0.3 0.5 0.4 Methnol Averge 328.2 216.2 5.1 12.3 9.7 SD 61.8 50.0 2.1 0.9 0.9 Acetic cid Averge 31.7 19.8 26.2 29.5 36.9 SD 0.5 3.2 2.5 1.5 3.1 LMw cids (C1-C4) Averge 32.5 29.4 27.6 32.0 41.9 SD 0.4 2.2 3.3 0.8 3.5 Hexnl Averge 11.5 7.6 12.2 30.1 28.7 SD 1.5 2.9 7.0 1.6 7.6 HMw VOCs Averge 34.4 29.8 33.6 66.2 69.7 SD 1.5 8.6 10.9 3.3 21.8 Totl VOCs c 791.8 283.7 66.7 111.4 122.4 Refer to the press conditions in Tble 1. b SD = stndrd devition. c Totl identified VOCs including formldehyde, methnol, LMw cids (C1-C4), nd HMw VOCs. TABLE 4. Multiple comprisons of VOC emissions between UF prticlebord nd PF prticlebord, pmdi prticlebord, FH1, nd FH2. PF pmdi FH1 FH2 Formldehyde Methnol Acetic cid NS LMw cids NS NS Hexnl NS NS HMw VOCs NS NS =comprison significnt t 0.05 level; NS = not significnt. tlion chromtogrph nd were not included s HMw VOCs. Formldehyde. Formldehydews identified through chromtropic cid nlysis of the queous impinger solutions. Methnol. Methnolwsidentified through GC-FID nlysis ofthe queous impingersolutions.althoughothercompounds were identified in the GC-FID nlysis, only methnol ws reported quntittively. LMwAcids. Low moleculrweight orgnic cids (C1 toc4) were identified ndquntifiedusingthehplc nlysis. Acetic cid, primry compound in the LMw cids, ccounted for bout 33 to 99 percent of the LMw cids occurring t the press conditions exmined in this study. THE COMPONENTS OF VOC EMISSIONS In determining the components of VOC emissions, there re two min points of interest: the type of significnt VOC emissions for ech dhesive used, nd the effects of ech dhesive on the identified VOCs in the wood-dhesive furnish. To determine the effect of the dhesivetypeonpressemissions,pnels were pressed t the center point conditions shown in Tble 1. Pnels were pressed with ech of the dhesive types (UF, PF, nd pmdi) nd with wood prticles to which no resin hd been dded (FH1 nd FH2). The FH1 wood-only pnels nd the UF- nd pmdi-bonded pnels hd similr press conditions (press temperture: 182 C,press time: 5 min., MtMC: 13%, borddensity: 0.77 g/cm3 (48 lb./ft. 3 ), nd the FH2 wood-only pnels were pressed under conditions similr to the PF-bonded pnels (press temperture: 182 C, press time: 6 min., Mt MC: 14%; bord density: 0.77 g/cm 3 (48 pcf). Tble 3 summrizes the VOC emissions from hrdwood pnels bonded with UF, PF, nd pmdi, nd the hrdwood pnels pressed with no dded dhesive (FH1 nd FH2). The results of multiple comprisons mong them re summrized in Tbles 4 to 6. Effect of UF resin. For the UFbondedhrdwoodpnels,formldehyde nd methnol were the lrgest components of VOC emissions during pressing, ccounting for bout 92 percent of the totl identified VOCs. Comprison with the emissions from the FH1 pnel with no dded dhesive shows tht more thn 95 percent of these two emissions cme from the UF dhesive. Formldehydeemissions rise fromthree sources: 1)excessfree-formldehydeintheresin; 2) therml decomposition of the wood; nd 3) hydrolysis of the UF resin during the heting undermoist conditions in the pnel (Mrutzky nd Mrgosin 1994). Methnol emissions my be ccounted for by the methnol present in the formldehyde used to mke the resin (Crlson et l. 1995) nd by the therml decomposition of wood to mke methnol (commonly known s wood lcohol). Therefore, these much higher levels of formldehyde nd methnol emissions from the UF prticlebord, reltive to pnels with no resin, re expected. Acetic cidemissions duringpressing of the UF-bonded pnel did not differ significntly from the emissions from the hrdwood pnel with no resin (Tble 4). This is becuse cetic cid is formed from the hydrolysis of cetyl groups contined in the hemicellulose of the hrdwood prticles (Crlson et l. 1995). Xyln, the mjor hemicellulose component in hrdwood, is highly cetylted t C-2 or C-3 (or both). On the verge, there re 7 cetyl groups for every 10 bckbone units (Sjostrom 1981). 70 NOVEMBER/DECEMBER 2002

Thus, reltively high emission of ce- TABLE 5. Multiple comprisons of VOC emissions between PF prticlebord nd UF prticlebord, tic cid from hrdwoods ws expected pmdi prticlebord, FH1, nd FH2. (Sollidy et l. 1999, Risholm-Sundmn UF pmdi FH1 FH2 et l. 1998). Formldehyde HMw VOC emissions were the lrg- Methnol est type of VOCs originting in the hrd- Acetic cid wood prticles, nd they were signifi- LMw cids NS NS NS cntly reduced when the UF resin ws Hexnl NS NS dded (Tble 4). Hexnl nd some HMw VOCs NS NS other components in the HMw VOCs, = comprison significnt t 0.05 level; NS = not significnt. for exmple pentyl-furns, rise from the decomposition products of the hrdwood furnish (Bumnn et l. 1999, Fengel nd Wegener 1984). These com- TABLE 6. Multiple comprisons of VOC emissions between pmdi prticlebord nd UF prticle- ponents re relesed t lter time in the bord, PF prticlebord, FH1, nd FH2. press run becuse they hve higher boil- UF PF FH1 FH2 ing points. The reduction in HMw VOC Formldehyde NS emissions when UF resin is dded my Methnol be due to the UF creting brrier on the Acetic cid NS wood prticles nd restricting the emis- LMw cids NS sion of these higher boiling compounds. Hexnl NS" NS Effect of PF resin. For the PF hrd. HMw VOCs NS NS wood-dhesive furnish, methnol ws =comprison significnt t 0.05 level; NS = not significnt. the lrgest component of VOC hotpressingemissions,ccountingforbout 76 percent of the totl identified VOCs. nish becuse the pthwys of the emis- could occur during hot-pressing. Fur Comprison with the pnels pressed sions with higher boiling points might ther studies could ddress why pmdi with no resin (FH2) shows tht 95 percent of the methnol emissions cn be be blocked fter the PF resin cures. ffects methnol, but not formldehyde The presence of the PF resin decresed emissions. ttributed to PF resin, with the remining being present in the wood. Forml the rte of cetic cid nd HMw VOCs VOC EMISSIONS AS A relese. FUNCTION OF ADHESIVE TYPE dehyde emissions comprised only bout 3 percent of the totl VOCs from the Effect of pmdi. For hrdwood pn- The results of the VOC emissions pnel, with bout 85 percent of the emissions ttributed to the PF resin. Methnd HMw VOCs were the lrgest com- wood prticlebord re summrized in els bonded with the pmdi, cetic cid from the hot-pressing of mixed-hrdnol nd formldehyde were emitted ponents of VOC emissions. They c- Tbles 7 nd 8. from both the wood nd the resin with count for bout 45 nd 50 percent of the Formldehyde. Formldehyde the mjor source being the resin due to totl identified VOC emissions, respec- emissions from UF-bonded pnels were the free formldehyde nd methnol in tively. Methnol nd free-formldehyde the lrgest component of VOC emisre not present in pmdi resin, so the sions mong the three dhesive types the resin. The lower levels of formldehyde could possibly be ttributed to only contribution to these emissions ws (UF, PF, nd pmdi). The emissions lower level of free formldehyde (< from the wood prticles. Methnol nd rnged from 52.4 to 640 mg/kg (oven- 0.1%) nd n dditive in the PF resin HMw VOCs were significntly de- dry weight [OD]), nd were significresed in the pmdi-bonded pnels rel- cntly higher thn those from pnels (Wng 1999). The dditive contined n mine functionl group, which tive to the pnels with no dhesive p- bonded with either PF or pmdi resin could rect with formldehyde to reduce the emissions during the hot-presss crrier for some VOC emissions sions for pnels bonded with PF nd plied (FH1). Moisture in the pnel cts (Tbles 4 nd 7). Formldehyde emising process. (Wng 1999, Wolcott et l. 1996), but pmdi resin rnged from 2.7 to 10.6 The cetic cid emissions from the pmdi rects rpidly with wter in the mg/kg (OD), nd from 0.02 to 0.85 PF pnels, however, were significntly furnish pnel (Wendleretl. 1995).This mg/kg (OD), respectively. The pmdi lower thn those from pnels pressed could cuse the reduction of the VOC pnels hd significntly lower formlemissions, especilly formldehyde nd dehyde emissions compred to both the with no dhesive. This my be ttributed to the higher ph vlue (12.2 to 12.7) of methnol emissions. The fst curing of UF nd PF pnels, due to the lck of thepfresisnusedinthis study.,thebsic pmdi my be the second fctor, block- formldehyde in pmdi resin. components contined in the PF probbly rect with cetic cid, nd reduced the lter stges of hot-pressing process. were the second lrgest components of ing pthwys of VOC emissions during Methnol. Methnol emissions the cetic cid emissions. HMw VOCs The third possible reson is tht pmdi VOC emissions from pnels bonded emissions from the PF prticlebord rects with hydroxyl groups. Becuse with either UF or PF resins. Pnels pressing were lso significntly lower methnol contins hydroxyl group, the bonded with pmdi hd significntly thn those from the net hrdwood fur- rection between pmdi nd methnol lower methnol emissions. The meth- FOREST PRODUCTS JOURNAL VOL. 52, NO. 11/1 2 71

TABLE 7. Summry of experimentl results of VOC emissions in 2 5-1 experimentl design. Press conditions Run Press Press Resin Mt Bord Formldehyde Methnol Acetic cid no. temp. time content MC density UF PF pmdi UF PF pmdi UF PF pmdi 1 - - - - + 52 2.7 0.02 5 65 0.8 0.2 0.4 0.8 2 + - - - - 174 5.6 0.07 164 152 4.8 15.0 9.4 12.0 3 - + - - - 326 10.5 0.15 210 229 10.4 33.6 17.5 30.2 4 + + - - + 397 5.6 0.51 248 330 15.5 83.8 30.5 50.8 5 - - + - - 141 6.5 0.09 24 338 1.7 0.8 1.7 4.9 6 + - + - + 250 6.4 0.09 227 329 3.1 12.0 15.4 12.1 7 - + + - + 417 14.5 0.20 436 548 7.9 38.0 16.5 29.1 8 + + + - - 565 12.1 0.90 233 222 9.6 64.4 61.6 48.0 9 - - - + - 121 5.1 0.09 17 136 2.6 2.0 3.0 2.8 10 + - - + + 225 4.1 0.16 119 154 5.6 17.4 8.5 13.8 11 - + - + + 388 8.3 0.16 115 142 13.1 53.9 18.9 39.3 12 + + - + - 604 6.8 0.34 203 112 9.7 78.1 75.3 37.3 13 - - + + + 144 6.1 0.03 22 254 0.9 1.3 2.5 2.5 14 + - + + - 285 6.0 0.10 303 296 2.1 18.9 13.3 10.9 15 - + + + - 525 12.1 035 277 370 8.2 40.5 15.9 26.2 16 + + + + + 640 10.6 0.85 312 176 5.4 107.5 73.8 44.4 17 0 0 0 0 0 385 9.9 0.31 390 213 2.7 32.0 16.3 23.6 18 0 0 0 0 0 389 7.6 0.23 329 168 6.6 32.0 22.4 28.5 19 0 0 0 0 0 417 7.5 0.33 266 268 6.0 31.1 20.8 26.4 Referto Tble 1, the symbols +, -, nd 0 represent high level, low level, nd center points, respectively. Forexmple, in run 5, the symbols men press temperture = 165 C, press time = 3 minutes, mt resin content = 8%, mt MC = 12%, nd bord density = 0.71 g/cm 3. TABLE 8. Summry of experimentl results of VOC emissions in 2 5-1 experimentl design (continued). Press conditions LMw cids (C1 to C4) Hexnl Hmw VOCs Run Press Press Resin Mt Bord no. temp. time content MC density UF PF pmdi UF PF pmdi UF PF pmdi 1 - - - - + 0.8 1.5 0.9 1.2 6.4 6.1 4.5 21.0 15.5 2 + - - - - 15.2 16.1 13.4 26.8 10.6 31.7 52.8 30.0 69.0 3 - + - - - 34.1 22.7 31.4 29.7 12.8 28.6 62.5 51.1 67.0 4 + + - - + 86.7 57.1 57.7 27.8 7.2 29.7 73.3 25.3 81.8 5 - - + - - 1.0 2.9 5.1 1.3 20.9 9.8 4.4 48.8 21.1 6 + - + - + 12.3 28.4 12.8 3.5 21.9 17.0 11.4 58.5 36.9 7 - + + - + 38.3 29.5 29.4 9.3 29.7 22.6 38.8 87.6 54.4 8 + + + - - 85.5 119.1 55.7 9.4 25.9 15.5 42.2 67.5 43.8 9 - - - + - 2.0 6.6 2.9 7.6 7.4 12.6 16.3 20.5 26.2 10 + - - + + 47.6 11.0 14.1 12.7 3.2 19.7 25.9 10.9 47.0 11 - + - + + 54.2 24.2 40.7 27.7 11.0 29.7 65.8 47.5 73.9 12 + + - + - 99.6 107.2 42.6 25.6 8.2 14.0 68.6 37.3 34.6 13 - - + + + 1.4 7.2 2.8 1.7 20.0 12.6 7.3 43.4 26.6 14 + - + + - 19.5 22.0 11.2 6.7 23.2 13.4 19.1 54.6 30.2 15 - + + + - 41.6 29.1 27.6 10.1 27.0 19.1 37.0 70.3 43.2 16 + + + + + 109.6 131.9 52.2 8.1 2.7 11.3 38.8 11.0 36.1 17 0 0 0 0 0 32.7 27.4 24.0 12.5 5.9 12.7 36.1 25.0 29.7 18 0 0 0 0 0 32.7 31.7 30.5 12.1 10.9 18.9 33.3 39.7 46.0 19 0 0 0 0 0 32.1 29.1 28.3 9.7 6.1 5.0 33.8 24.8 25.3 Refer to Tble I, the symbols +, -, nd 0 represent high level, low level, nd center points, respectively. For exm le, in run 5, the symbols men press temperture = 165 C. press time = 3 minutes, mt resin content = 8%, mt MC = 12%. nd bord density = 0.71 g/cm 3. 72 NOVEMBER/DECEMBER 2002

TABLE 9. - Effects ofpress prmeters nd their interctions on VOC emissions from UF-, PF-, nd pmdi-bonded prticlebord. Press prmeters nd their interctions VOC emissions Temp. Time Resin MC Density Temp. * time Time * resin Time * MC ( C) (min.) - - - - - - - - (%) - - - - - - - - (g/cm 3 ) Formldehyde (ASTM D 5582) UF b PF pmdi Methnol UF PF pmdi Acetic cid UF PF pmdi LMw cids (C1 to C4) UF PF pmdi Hexnl UF PF pmdi HMw VOCs UF PF pmdi Temp. = press temperture; time = press time; resin = resin content; MC = mt MC; nd density = bord density. b = 95 percent significnt. nol emissions from pnels bonded with PF pnels were significntly lower thn the emissions of HMw VOCs by block- UF resin rnged from 5.4 to 436 mg/kg those from either the UF or pmdi pn- ing the emission pthwys fter resin (OD), nd with PF resin rnged from els (Tble 5). This reduction in cetic curing. 64.7 to 547.7 mg/kg (OD). Methnol is cid emissions for pnels bonded with VOC EMISSIONS AS A FUNCTION present in formldehyde solutions used PF resin is due to the higher ph vlue of OF PRESS VARIABLES in UF nd PF dhesive mnufcture, PF resins (12.2 to 12.7), which would Formldehyde. Tble 9 nd Fig- thus ccounting for the significntly hve rected with the cetic cid. The ures 2 nd 3 summrize the ANOVA rehigher methnol emissions from pnels significnt difference in cetic cid emissions between UF-bonded pnels nd nd their interction on formldehyde sults of the effect of five press vribles pressed with these two resins. Under similr hot-pressing conditions, the UFbonded pnels emitted significntly pressconditions,is notredilyexplined were significntly incresed by incres- pmdi-bonded pnels, under the sme emissions. Formldehyde emissions higher levels of methnol thn the PF becuse the cetic cid emissions re ttributed to the hrdwood furnish, not the els bonded with ll three dhesive types. ing press time nd resin content for pn- pnels. Interestingly, multiple comprisons of the dt given in Tble 6 indicte resin. Significnt increses in formldehyde tht ppliction of pmdi resin resulted HMw VOCs. HMw VOC emissions from three types ofpnels (UF, PF, were observed for the UF- nd pmdi- emissions with incresing temperture in decresed emissions from the hrdwood furnish. nd pmdi) were emissions from the bonded pnels, but not the PF-bonded Acetic cid. Acetic cid emissions wood rther thn the dhesives, nd pnels. Longer press times, higher resin rnged from 0.2 to 107.5 mg/kg (OD) for UF-bonded pnels, 0.4 to 73.8 mg/kg (OD) for PF-bonded pnels, nd 0.8 to 50.8 mg/kg (OD) for pmdibonded pnels (Tble 7) under the press conditions of this study. The results from the multiple comprison indicted tht the cetic cid emissions from the rnged from 4.4 to 73.3 mg/kg (OD), 11 to 87.6 mg/kg (OD), nd 15.5 to 81.8 mg/kg (OD), respectively. As shown in Tbles 4, 5, 6, nd 8, under similr press conditions (center points), there ws no significnt difference in HMw VOC emissions mong these three pnels. All three dhesives could possibly decrese contents, nd higher press tempertures (except for the PF-bonded pnels) cn cuse higher levels of formldehyde emissions from both dhesive nd hrdwood furnish. Figure 2 shows tht incresing press time lone ccounted for pproximtely 50 percent ofthe totl effects, while MC hd smll significnt FOREST PRODUCTS JOURNAL VOL. 52, NO. 11/12 73

Figure 2. Reltive effects of press vribles on formldehyde emissions during hot-pressing of pnels bonded with UF, PF, nd pmdl (reltive effect: single effect neglecting interctions; l = 95% significnt). Figure 3. Temperture-time interction effect on formldehyde emissions during hot-pressing of pmdi-bonded pnels. theeffectoffivepress vribles nd their interction on methnol emissions. Emissions of methnol during pressing ofuf-bonded pnels were significntly incresed by incresing press temperture, press time nd resin content. Sig nificntincresesinmethnolemissions were observed from PF-bonded pnels by incresing resin content nd from pmdi-bondedpnelsbyincresingpress time or decresing resin content. The effect nd pnel density showed no effect. There were significnt interctions of press temperture-press time nd press time-resin content on the formldehyde emissions from pmdi-bonded pnels (Fig. 3), nd significnt interc- tion of press time-mc on the formlde- hydeemissions fromuf-bondedpnels. Methnol. Tble 9 nd Figures 4 nd 5 summrize the ANOVA results of 74 negtive dependence ofmethnol emissions on resin content forpmdi-bonded pnels cn be ttributed to the rection of pmdi with methnol nd with the wter, which cts s crrier for the methnolemissions(ingrmetl. 1994). The combined effects of press time nd resin content contributed bout 60 percent of the totl effects on the methnol emissions for ll three dhesive types (Fig. 4). Becuse methnol emissions cme from both dhesive (excluding pmdi dhesive) nd hrdwood furnish, it is resonble tht the methnol emissions incresed with incresing press time, resin content (excluding the pmdi dhesive), nd press temperture (excluding the PF dhesive). There ws one significnt interction term of press temperture-press time found for the methnol emissions from UF-bonded pnels (Fig.5). LMw cids. Tble 9 nd Figure 6 summrizethe ANOVAresultsoftheeffect of five press vribles on emissions of orgnic cids. Acetic cid emissions significntly incresed t incresing press temperture nd press time for ll three dhesive types, nd the reltive effects of these two press vribles ccounted for bout 80 percent of the totl effects (Fig. 6). A possible explntion for this result cn be ttributed to cetyl groups contined in the hemicellulose of the hrdwoodprticles hydrolyzing into cetic cidduringhot-pressing.becuse the speed of hydrolysis rection increses s the rection temperture is rised, cetic cid emissions incresed with incresing press temperture. Under the sme pressing temperture, cetic cid emissions incresed with incresing press time. Emissions oflmw cids incresed with incresing mt MC for pnels pressed with UF resin. Resin content did not hve significnt effect on the cetic cidemissions, since cetic cid ws ttributed entirely to the hrdwoodprticles. Asignificntinterction term of press temperture-press time ws found for the cetic cid emissions from the UF- nd PF-bonded pnels. HMw VOCs. Tble 9 nd Figures 7 nd 8 summrize the ANOVA results of the effect of five press vribles nd their interction on HMw VOC emissions. The HMw VOC emissions were significntly reduced with decresing press time nd incresing resin content for both the UF- nd pmdi-bonded pn- NOVEMBER/DECEMBER 2002

els. However, s shown in Figure 7, resincontenthd significntly positive effecton the HMw VOC emissions from the PF-bonded pnels, lthough ll three dhesives decresed the emissions of HMw VOCs from hrdwood prticles t the level of center points. These results re further illustrted in Figure 8: HMw VOCs emissions rpidly decresed with incresing resin content for the UFbonded pnels. The emissions from the PF-bonded pnels slightly decresed with incresing resin content up to 6 percent resin content, then shrply incresed s resin content ws rised. For the pmdi-bonded pnels, HMw VOCs shrply decresed with incresing resin content below 4.5 percent resin content, then incresed slightly s resin content incresed. These results indicted tht the three dhesives hd different effects on the HMw VOCs emissions mong the different rnge of resin contents. As shown in Tble 8, the HMw VOC emissions t higher levels of resin content (runs 5, 6, 7, 8, 13, 14, 15, nd 16) were higher thn those t lower levels ofresin content (runs 1, 2, 3, 4, 9, 10, 11, nd 12) for the PF-bonded pnels, except for run 16. This is becuse 20 to 51 percent of HMw VOC emissions were hexnl emissions tht were higher t higher levels of resin content. Press temperture hd significnt effect on the HMw VOC emissions only from the UFbonded pnels. There ws significnt interction of press temperture-press time on the HMw VOC emissions from thepmdi-bondedpnels. It ws surprising tht gretereffects of MC on emissions were not observed. However, this lck of dependence upon MC concurs with previous reserch by Crlson et l. (1995). Crlson, while mesuring emissions from pressing of PF-bonded oriented strndbord, found tht if the mt MC ws bove 8 percent, chnges in mt MC ppered to hve less of n effect on the formldehyde emissions. Becusethe MC levels inthis study rnged from 10 to 14 percent for UF nd pmdi pnels, nd 11 to 15 percent for PF pnels, it is possible tht there ws sufficient moisture in the pnels, even t the low point, tht inhibiting of the crrier effects of the moisture were notobserved. Figure 4. Reltive effects of press vribles on methnol emissions during hotpressing of pnels bonded with UF, PF, nd pmdl (reltive effect: single effect neglecting interctions; * = 95% significnt). Figure 5. Temperture-time interction effect on methnol emissions during hotpressing of UF-bonded pnels. COMPARISON OF FORMALDEHYDE EMISSIONS BETWEEN HARDWOOD AND SOFTWOOD Using published dt of VOC press emissions from southern pine prticlebord (Wng 1999), the formldehyde emissions from hrdwood nd softwood prticlebord were compred (Tble 10). The results show thtformldehyde emissions from hrdwood prticlebord bonded with the UF resin were much higher thn those from softwood prticlebord bonded with the UFresin. This might be ttributed to resin btch differences between the pine nd mixedhrdwood studies or the cetic cid from the hrdwood furnish contributing to UF resin hydrolysis, nd subsequently, lrgeremissionsofformldehyde. However, the formldehyde emissions from hrdwood prticlebord were significntly lower thn those from softwood FOREST PRODUCTS JOURNAL VOL. 52, NO. 11/12 75

prticlebord for the other dhesive typesexmined. CONCLUSIONS The VOC emissions rising from the hot-pressing of mixed-hrdwood prticlebord with three dhesive types were identified nd quntified using four nlyticl techniques. A two-level hlf-frctionl fctoril experimentl design ws utilized to determine the primry effect offive press vribles nd the interctions between ny two press vribles on the VOC emissions. Twenty-seven compounds were identified in the hrdwood VOC emissions, nd the primry components in the VOC emissions were formldehyde, methnol, cetic cid, nd HMw VOCs in whichhexnl ws the predominnt single chemicl. Among the three dhesive types,formldehydendmethnolemissions from the UF prticlebord were the most bundnt components of the VOCemissions,ccountingforbout92 percent of the totl identified VOCs in the UF prticlebord. Methnol emissions from the PFprticlebord were the second lrgest component of the VOC emissions,contributingbout75 percent of the totl identified VOCs in the PF prticlebord. More thn 95 percent of these emissions cme from the dhesives used. Lower levels of formldehyde nd cetic cid emissions were foundinthepfprticlebord.compred with UF nd PF, pmdi hd much lower levels of formldehyde emissions, nd lso significntly reduced the methnol emissions from the hrdwood prticles. Acetic cid nd HMw VOC emissions were the lrgest components of VOC emissions from pmdi prticlebord. Press time, resincontent, press temperture, nd interctions mong these three vribles re the most significnt press vribles controlling VOC emissions rising during the hot-pressing of hrdwood prticlebord. Most of the identified VOCs were significntly incresed with incresing press time, mt resin content, ndpress temperture. LITERATURE CITED Figure 6. Reltive effects of press vribles on cetic cid emissions during hotpressing of pnels bonded with UF, PF, nd pmdl (reltive effect: single effect neglecting interctions; * = 95% significnt). Figure 7. Reltive effects of press vribles on HMw VOC emissions during hotpressing of pnels bonded with UF, PF, nd pmdl (reltive effect: single effect neglecting interctions; * = 95% significnt). 76 NOVEMBER/DECEMBER 2002

TABLE 10. Comprison of formldehyde press emissions between hrdwood (HW) nd softwood (SW). Press conditions Resin type UF UF PF PF pmdi pmdi FH FH Wood type HW b SW b HW SW HW SW HW SW Press time (min.) 5 6 6 8 5 6 5 6 Resin content (%) 6 7 5 6 3.5 4.5 0 0 Mt MC (%) 13 12 14 13 13 12 13 12 Formldehyde (mg/kg OD bord) 385 244 10 11 0.3 35 1.4 42 389 257 8 13 0.2 32 0.4 41 417 249 7 12 0.3 32 1.1 40 Averge 397 250 8 12 0.3 33 1.0 41 All pnels were pressed t 182 C with pnel density of 0.77 g/cm 3. b Results from two different studies. Figure 8. HMw VOC emissions during hot-pressing s function of resin content for pnels bonded with UF, PF, nd pmdi. FOREST PRODUCTS JOURNAL VOL. 52, NO. 11/12 77