Pyrolysis of EVA and its application in recycling of photovoltaic modules

Journal of Enirornental Sciences Vol. 16. No.6.pp.889- -893 .2004ISSN 1001-0427N11- 2629/XArticle ID: 1001-0742(20060889-05CLC number: X79Document code: APyrolysis of EVA and its application in recycling of photovoltaic modulesZENG De-wen'"，Manfred Born'，Karsten W ambach2(1. Institut fir Energieverfahrungtechnik und Chemieingenieurwesen, TU Bergakademie F'reiberg, 09599 Freiberg. Germany. E mail: dewenzeng@hnu. cn; 2. Deutsche Solar AG, Alfred- Lange-Sr.18, 09599 Freiberg, Germany)Abstract: The basic pyrolysis behaviour of ethylene vinyl acetate( EVA) copolymer. which is olten used as a lamination agent in solarmodules, was investigated in thermogravimetry, difrential thermnal analysis(DTA) and thermovolumetry. The TG analysis showed thatthe EVA pyrolysis can be accelerated under the partial oxldizing atmosphere but the end pyrolysis temperature must be higher than innitrogen, to eliminate the coke formed. Meanwhile, a strong exothermal peak occurs at about 450C under the air condition and getsweaker obviously at the oxygen content lower than 10 vol. % . The mass balance of EVA pyrolysis was given through the thermovolumetrywith the output of 10 wt. % permanent gas, 89.9 wt. % condensate and 0.1% residual coke. Besides, the composition of the permanentgas and condensate at different pyrolysis stages were analysed and interpreted on the known pyrolysis mechanism.Keywords: pyrolysis; EVA: kinetic; recyclingdelamination of the Si- wafer from the glass(John, 1997).IntroductionTo understand the delamination hehaviours of the solarWith the increasing use of expensive photovoltaic cell,modules, one must first understand the pyrolysis behavioursmore and more end-of-life solar modules are accumulated .of the laminate in it. Up to now a number of literatureTherefore，the recycling of such“ rubbish ”is meaningfulconceming the pyrolysis of EVA is available ( Haussler,environmentally and economically. Of all these solar1998; McGrattan， 1994; Sultan. 1991; Maurin， 1991;modules，the most are the crytalline silicon PV modules ,Abe. 1978; Munteanu， 1977; Hrdina， 1990; 1999 ).which contain high purity silicon wafers laminated on a glassshowing that the EVA pyrolysis in inner atmosphere undergoesor other plastic plate front and back with ethylene vinyltwo stages. The first stage is related to the removal of theacetate( EVA) copolymer. The cost for the pure Si- waferacetic acid from the main chain, as described in Eq.(1).covers over 60% of the whole model(John, 1997). So, it isThe second stage is the degradation of the remainingeconomically preferable to recycle the Si- wafers withoutpolyethylene co-polyacetylene. The degradation reaction inmechanical damage， so as to use them to produce newthe first stage can be well characterized by thermo gravimetricmodules after the chemical treatment on the surface. To thisanalysis(T(A), from which the content of the vinyI acetateend, pyrolysis is proved to be an ffective method for thein EVA can be calculated( Harussler, 1998; Hrdina, 1998).EVAacetie aucid palyethylene co-polyaretylerne-[(CH:一CH). (CHCH2)、in一， CH,COOH +-[(CH- CH2).-(CH= -CH),- - ]。(1)However, some conclusions in the literatures are notential thermal analysis( DTA)consistent with each other, such as the DTA results of theTGA and DTA were crried out simultaneously in theEVA pyrolysis ( Hrdina，1998; Frsson， 2000 ). Sorme .instrument Netzsch STA 409. At each est 35 mg sampleinfornation about the EVA pyrolysis, which is necessary for(Elvax-150) was placed in an Al2O, pan and heated atthe design of a technical fumace for recycling the solarconstant heating rate under different oxidizing almosphere.modules, such as the gas volume of the EVA pyrolysis, theThe neat EVA used is Elvax-150 produced by Duponinfluence of the healing regimes and the oxidizing almosphereCompany. Its selcted properties are given in Table 1 .on the EVA pyrolysis， especially the EVA in the solarTable 1 Thermal propertes of EVA copolymers measured by DSCmodules，is not clear. Therefore， further systematicElvax' gradeVA. nt.% Mel point. C Heat of fusion. J1ginvestigations of the pyrolysis behaviour of EVA in neat state40W404723under different atmospheres at different heating rates are15063necessary. The results to be obtained should be meaningful242874for the industrial practice of the recyeling of the solar中国煤化工modules.65MYHCNMHG781 Experimental76100821.1Thermo gravimetric analysis (TGA) and differ-Note: # given by Dulpon Company# Corsponding author. Aetual adres: School of Chemistry and Chemical Enginecring. Hunan University , Changsha 410082. China89ZENG De-wen et al .Vol.161.2 Thermo volumetric measurements00F，70.30Thermo volume analysis was caried out in an apparatust 0.250tas shown in Fig.1. The tubular reactor (φ =4 cm) with ai- 0.20constant temperature zone of 10 cm infumace was|H 0.15).10connected with an Erlenmeyer flask and a vesse! throughg A010 F.05flexible tubes . The Erlenmeyer flask was cooled with ice and- 0.00the vessel was flld with saturated salt solution. The pressure20 F-0.05 号￥in the tube was kepl in balance with the atmnosphere through.10the deflection of the salt solution. whose volume equals thal100 200 300 400 500r,cof the permanent pyrolysis gas. During the experiment, 10 gneat EVA(EIvax 150 or 460) were placed in the crucible inFig.2 TGA/DTA of EVA copolymer in nitrogen 曲constantthe reactor, which was then purged with Argon for about 90heating rale 5 K/minmin at a now rate of 30 l/h. After that, the reactor wasBesides. the results of the EVA pyrolysis carried outheated at a certain heating rate .The solution volume and thetemperature of the crucible were noticed every five minules ●under different oxidizing atmosphere( Fig.3).During pyrolysis, the pyrolysis gas with a high boiling point100 (门condensed in the Erlenmeyer flask and the flexible tubes .80 t1.6层↑The weight of condensate was obtained by weighting theErlenmeyer flask and the flexible tubes before and after the姿602.0experiment.日40.80.6日6-200.4量↓.2|2100 200300 400 500 600*0i0.5 vol.% O2 + 99.5 vol.% N25 vol.% O2+ 95 vol.% N210 vol.% O2 + 90 vol.% Nz15vol.% O2+ 85 vol.% NzFig. 1 Sehematic diagram of the experimental apparntus for thermovulumetryFig.3 TCA/DTA of EVA copolymer at cnstant heating1. valve; 2. vsed; 3. U form tube; 4. Elenmeyer flask; 5. ice; 6.rale of 5 K/min under diferent oxidizing atmospherestemperature transmitter;: 7. temperature controller: 8. thermocouple: 9.furmace; 10. sample crucable; 11. rectorIt is very obvious that the oxidizing atmosphere canaccelerate the degradation reaction and the optimal O2 content2 Results and discussionis about 5 vol. %. But, in comparison with that in nitrogen,a shoulder occurs in all the TGA curve in the temperature2.1 TGA and DTArange from 460C to 570C under all different oxidizingThe combined TCA and DTA resuls of the neat EVA inatmospheres. This should be caused by the carbonization ofnitrogen are shown in Fig. 2. The DTA showed threeEVA under oxidizing atmosphere. To eliminale the coke, aendothermic peaks. The first endothermic peak ( 63C)corresponds t0 the melt points of the EVA. The second peakhigher end pyrolysis temperature is necessary, though it may( 334C) reflectse the endothermic degradation reactionbring damage to the Si-wafer. The infuence of highertemperalure on the quality of the Si-wafer should be(Equation I) in the first pyrolysis stage. The second regioninvestigated further. Another disadvantage of oxidizingof weight loss begins with predominantly endothermicatmosphere is that a strong exothermal peak occurs in thereactions and ends with predominantly exothermic reactions .second pyrolysis stage under the air condition， which mayThe results are consislent with the lierature( Hrdina，1998)indicate a buming reaction of the pyrolysis production andbut diferent from the other( Frisson，2000)， where neithermay damage the Si- wafer also. The exothermnal peak getstype nor the source of the EVA was staled. Note that the firstmelt peak(63C) of the EVA used in this work is lower thanobserv中国煤化工than 10 vol. %. Undererent pyrolysis behaviorthat(88C) reported by Hrdina( Hrdina，1998)， because of( solid:TYHC N M H Gryrolysis temperature atthe different lypes of EVA(Table 1) . The three endothermicpeaks ccuring during the EVA-pyrolysis should bhe same degradation rate under nitrogen is in most casesfavourable lo the protection of the Si- wafer in the module fromhigher than that in oxidizing atmosphere, its end pyrolysisthe damage of the thermal stress during the rapid heatingtemperature( approx. 480C), however, is much lower than570C under oxidizing atmosphere. To decrease thestage芳芳数据No.6Pyrolysis of EVA and is application in recycling of photovoltaic modules91exothermal effect and the end pyrolysis temperature，pyrolysiswith the increasing of the heating rate. In the second stageunder inner or weak oxidizing atmosphere is highlythe releasing rate of the permanent pyrolysis gas is a litlerecommended .smaller than that in the first stage but both of them are2.2 Thermovolumetry and pyrolysis mass balancecomparable. Although the max weight loss ( approx. 782.2.1 Infuence of heating rate and EVA-types on thewt. %) occurs in the second stage( Fig. 2), the permanentpyrolysis gas volume and gas releasing ratepyrolysis gas amount in this stage is smaller than in the firstThe results of thermovolumetry ( Fig.4) are consistentstage .with that of TGA( Fig.2). The pyrolysis gas begins to releaseThe amount of the permanent pyrolysis gas dependsobviously at about 275C and its releasing rate shows two strongly on the vinyl acetate content in EVA, as shown inpeaks also. The first rate peak moves from 310C to 340CFig.4. More permanent gas is produced from the pyrolysis ofdepending on the heating rates and the second peak appearsEVA with higher content of VA.at about 460C . The permanent gas volume increased slightly2880Elvax-150, (33 VA mass. %;24 t-o- 4K/min→0- 5 K/min--- 3 K/min- 4 K/min-0- 2 K/min20600Elvax-460. (18 VA mass. %):400-一5 K/min8-2000 88100 200 300 400 500 6000 100 200 300 400 500 600r,crig.4 Inluence of the heaing rate on the volume and the releasing rate of permanent prolysis gas2.2.2Production compositions and mass balance ofand CO in the permanent gas should be the secondaryEVA pyrolysisdegradation productions of the acetic acid， the weight ofThe mass balance of the EVA pyrolysis in nitrogen iswhich，according to the literalure ( Haussler， 1998;shown in Fig.5. The main product of the EVA pyrolysis isMunteanu, 1977), equals the weight loss of EVA in the firstthe low volatile condensate that accounls for aboul 90% ofstage(about 22 w1. % according to Fig.2). From Fig.5 andthe EVA. This places a special requirement for the bumer ofTable 2 it can be calculated that about 5.17 g acetic acidthe pyrolysis gas and condensate ，which bums the pyrolysisfrom 100 EVA is degraded to CH, CO2 and CO. The rest 17organic productions to fulfl the heat requirement in theg acetic acid should stay in the condensate and accounts forpyrolysis process in industry. The litle residue coke(0.117/89.9= 17 wt. %. The composition of the acetic acid inwt. %) is favourable for the surface chemical lreatments ofthe condensale is determined experimentally to be about 12the Si- wafer after the pyrolysis .wt. %，being lower than the theoretically caleulaled , becauseof the incomplete condensation of the acetic acid in the100g EVAexperimental process .Different gas compositions are expected in the twoPermanent gasdegradation stages because of the different reactionCondensale 89.9 gReidue0.1 g10 g(7.22 L)mechanisms. Table 4 shows the experimental results of theFig.5 Mass balance of EVA pyrolysiscompositions of pernanent gas at different pyrolysis stages. Inthe first stage the main compositions are CH. CO2 and COThe compositions of the permanent gas and condensateresulting from the degradation of the acetic acid .of the EVA pyrolysis, which are the average values of over 20volue is much lower .experiments at diferent heating rates, are given in Table 2Notewl中国煤化工in lhe second stage.and Table 3，respectively. The compositions of CH. CO2TYHCNMHGTable2 Composition and heating value Hu of the permanent pyrolysis gas(vol. %)SubstanceHcoCO2C2H%CHC,H2C,H。F-C.Hro N-C.Hrolu°Average value8.4322.122.7126.1112.2210.925.2610.410.171.6139.6Note: * Hu in M/m'892 .ZENG De-wen et al .Vol. 16Table 3 Composition of the condensate from EVA pyrolysls( wt. % )2.3 Reaction kineticElementcHBecause thal the kinetic parameters vary with the changeComposition81.212.20.0.2of amount of vinyl aceltate in EVA ( Hrdina， 1998)， thNotes: H2O: < 2 m1.哭，CH,COOH: 12 w1. %，aliphatic hydrucarbon; 80determination of the pyrolysis kinetic pararmeters of the EVAw1.%used in solar modules， is necessary, although some kineticThat means that the acetice group was not removed complelelyparameters of the EVA pyrolysis were published for otherfrom the main chain or the acetic acid generated in the firstgoals( Hrdina, 1999; Salin, 1993).stage did not diffuse cornpletely from the molten polymer.Table 4 Gas compositions( vol. %) and heating value Hu( MJ/m' ) at difterent EVA prolysis stage”SubstanceH2CH。CO2C2HC,H。C,HI-ButanN-ButanHuSage I3.15.268.7.20.80.50.70.41.20.30Stagr_ I14.9.415.915.510.053Note:，The heating rate is 5 K/minFor the kinetic analysis using TGA weight loss dala300 rseverl methods are available. One of them is he constant250heating rate method( Nam, 1992; Moddeman. 1986). In thease of EVA degradation, the fllowing kinetic equation is200 t Avarage value: 216 kJ/molapplied:150da/dt = Ag(a)exp(- E1kT).where a is the mass fraction remaining，t is the time, A is享s(the pre-exponenl, E is the activation energy, g(a)=(1 -a)"，n is the reaction order, k is the gas constant.024681012141618202Taking the logarithm of Eq.(2) we haveConversion, 0U%ln(da/dt) = In(A) + In[g(a)]- E/kT. (3)Fig.7 Activity energy as function of conversion al the firstA plot of In(da/dt) versus I/T for various constantpyrolysis stuge of EVAlevels of conversion a should then yield a straight line withslope of - E1k. The trealment of the experimental data inβ1T° = (A.R)1E.exp(- E_/RT.).(4)Fig.6 with the Eq.(3) yields the reaction activation energieswhere β= dT/dt is the heating rate， and the subscripl mat different conversion rate a shown in Fig.7. The averagedenotes values at the maximum decomposition rate .activation energies in the first pyrolysis stage is 216 kJ/molThe nalural log of Eq. (4) is laken such thatand obviously higher than those( 163-- 186 kJ/mol) from thelog(β1T%) = K- E/RTm .(5)literatures( Hrdina, 1998; Salin, 1993).Substituting the experimental data(Fig. 6) in Eq.(5)00 r，80yields the linear relation between the lwo termns log( p1Tm)60nd 1/Tm as shown in Fig.8. It can be seen that th10公/minactivation energies for the first slage calculated with different日2020 C /minmethods are identical . The difference from the other literature459.2T. 334.4-艾(Hrdina，1998; Nam，1992) can be atributed to the acetic旨-10 34.354.3C 468.3C .acid content in EVA. which is 22 w1.% for our system-20s心/min10 c /minmatching 30 wt. % and 12 wt. % in the EVA used by Namg-3020 c /min476.00(Nam，1993) and Hrdina( Hrdina, 1998) ，respectively .4950300350450500r;c-9.60 CFirst stage、 -.10.0K= Eg/R= -25300Fig.6 TGA of EVA in nirogen at dflerent constant冒-10.4heating rates-10.6K=-E/R=436000 examine the correctness of the above results,中国煤化工another method, which was derived by Kissinger( Kissinger ，MHCNMHG1957), is applied to calculate the reaction activity energy.-1.Kissinger applied the maximum decomposition rate conditions0.00136 0.00140 001600.00164to Eq.(2). i.e. d(da/d)/dt = 0, and after differentiation1/Tm, 1/KfoundFig.8 TGA dala poted acording t0o Kisinger' s methodNo.6Pyrolysof EVA and is application in recyeling of photovoltaic modules893The influence of the temperalure on the reaction rate canBergakademie Freiberg. The research was supported bbe known from the activation energy . The influence is largerBundesministerium fir Bildung und Forschung andin the second stage than in the first, for example, when theBundesministerium fur Forschung und Technik， Germany,temperature is 20 K higher, the reaction rate increase 4 timesunder contract number: InnoRegio 04020105.and 6 times in the first and second slage, respectively. So,References :in the industrial practice the solar modules should be heatedas evenly as possible, because the slowest reaction districtAbeS. Mtsubara H. Tamura T. 1978. Erolved as analysis of ethyle-vinylacetate copolymers[J]. Bunscki Kagaku, 27: T34- -T38.always decides the reaction time needed for the wholeFisson L, Lieten K. Bruton T et al.. 2000. Recent improvements in industrilmodule .PV module reeling [ C ].16th European photovolaic solar energyconference, Glagow，uK. May 1-5. 2003 ConclusionsHausler L. Pompe G, Albrecht V t al.. 1998. Determination of vinyl acetateThe basic pyrolysis behavior of neat ethylene-content in EVA copolymers. Psibitie and limits in the use of MS coupledanalysis methods[J]. Journal of Thermal Analysis and Calorimetry, 52 :vinylacetate( EVA) copolymer was investigated by TGA andDTA at different heating rale under different oxidizingHrdina K E. Halloran J w. Olivein A et al.. 1998. Chemistry of removal ofatmosphere. The DTA of EVA gave three endothermic peaks,ethylene vinyl acetate binders[J]. Jourmnal of Materials Science. 33: 2795-which will prolect the Si-wafer in the solar modules from be2803.Hrdina K E. HalloranJ W. Kaviany M et al., 999. Defet formation duringdamaged by the thermal stress during the rapid heating stagebinder removal in ethylene vinyl acetate flled Bystem[J]. Journal of Materialsof the pyrolysis of solar modules. The partial oxidizingScience, 34: 3281-3290.atmosphere can accelerate the pyrolysis reactions but resultsJohn R B. IgorI A. 1997. Psbilitiy of recyeling silicon PV modules{[C]. 26thin the formation of an abundant amount of coke . To eliminatehe residual coke thoroughly, a higher end pyrolysisKisinger H E，1957. Reaction kinetics in difereniud thermal anaiysis [ J].Analytical Chemistry, 29: 1702- 1706.temperalure ( 5709) is necessary. Besides， a strongMaurin M B, Dittert L w，Hussain A A. 1991. Thermogravimetrie analysis ofexothermal peak was observed in the EVA pyrolysis under aircondition. This peak may indicate a combustion reaction.the prolysis products[J]. Thernochimica Acta, 186; 97- 102.The strong exothermal peak disappears when the 02 content isMeCGraitan B J，1994. Examining the decomposition of elhyie-riryi acetatecopolymers using TC/GC/IR[J]. Applied Spetoscopy. 48: 1472- 1476.lower than 10 vol. %. The mass balance of EVA pyrolysisModdeman W E, Bowling w C. Tibitts E E. et al..1986. Thermal slability andhas been made according to the thernovolumetry，which gavecompatibility of plylteretherkeone( PEEK) with an oxidzer and prlecniethe output of 10 wt. % permanent gas， 89.9 wt. %blend[J]. Polym Eng Sci, 26: 1469- - 1477.condensate and 0.1 wt. % residual coke. The differentMuntearu D. Turcu s.1977. sudy of the mechanism of thermal dempositionreactions of the graf copolyners of plyetbylene with inyl acetate[J].compositions of the pernanent gas in different pyrolysis stagesMateriale Plastice，, 14: 144一147.result in different heating values of the gas ，corresponding toiam 」D, Seferis J C，1992. A compusilion methodology for mulisagethe known pyrolysis mechanism. Furthermore， the gasdegradation of ploymers[J]. J Polym Sei B: Polyn Phys, 30: 601- 608composition in the second stage indicated that the acetic acidSalinL M. Seferis J C. 1993. Kinetic analysis of higlh-resolution TCA variablewas not completely released from the matrix plastic in the firstheaing rale data[J]. J Appl Polym Sci, 47: 847- 856.Sulan B A. Soernik E, 1991. Thermal degradation of EVA and EBA.pyrolysis stage. The kinetic parameter of the EVA ( Elvax-comparison. 1. Volatile decompsition products[J小]. Jourmal of Applied150) pyrolysis was derived from the TC dataIn short, thePolymer Science. 43: 1737- 1745 .pyrolysis behaviours of EVA. such as the thermal effects ，WambachK，1998. Reeyeling of PV modules[C]. 2th world conference andmelting point, pyrolysis gas amount, kinetic parameter ete ..exkhibition on pholovoltaic solar energy conversin. Vienna， July 6- 11,are strongly influenced by the content of vinyI acetalte in it.Acknowledgements: This work was carried out while D( Received for reriew December 17. 2003. Accepled March 24. 2004)Zeng was working as a postdoclor in the Institul fuirEnergieverfahrungstechnik und Chemieingenieurwesen， TU中国煤化工MYHCNMHG