Download pdf - Polymer Degradation and Stabilitykohl.chbe.gatech.edu/sites/default/files/linked_files...continues and the backbone of the polymer unzips. Molecular weights as low as 13,000 g/mol

lable at ScienceDirect

Polymer Degradation and Stability 96 (2011) 686e702

Contents lists avai

Polymer Degradation and Stability

journal homepage: www.elsevier .com/locate/polydegstab

Decomposition of poly(propylene carbonate) with UV sensitive iodonium salts

Todd J. Spencer*, Paul A. KohlGeorgia Institute of Technology, School of Chemical and Biomolecular Engineering, Atlanta, GA 30332-0100, USA

a r t i c l e i n f o

Article history:Received 2 March 2010Received in revised form19 November 2010Accepted 8 December 2010Available online 24 December 2010

Keywords:Poly(propylene carbonate)Acid degradationThermal degradationDiphenyliodoniumMicroelectronicsSacrificial polymer

* Corresponding author. Tel.: þ1 404 805 3207.E-mail address: [email protected] (T.J. Spencer

0141-3910/$ e see front matter � 2011 Elsevier Ltd.doi:10.1016/j.polymdegradstab.2010.12.003

a b s t r a c t

The decomposition characteristics of poly(propylene carbonate) containing a photoacid generator havebeen studied. The influence of casting solvent, photoacid concentration and type, UV exposure dose,substrate surface, and ambient gas were included in this study. Dynamic thermogravimetric analysis wasused to analyze the decomposition characteristics. Kinetic parameters were extracted using the Kissingermethod and the CoatseRedfern method. Fourier Transform Infrared Spectroscopy was used to analyzeeffects of casting solvent. The highest thermal stability was found to occur in high molecular weight,high-purity poly(propylene carbonate) samples. Cyclohexanone and trichloroethylene solvents werefound to increase the thermal stability. Photoacid generators based on diphenyliodonium salts loweredthe onset decomposition temperature and activation energy.

� 2011 Elsevier Ltd. All rights reserved.

1. Introduction

Poly(propylene carbonate) (PPC) is a soft aliphatic thermoplasticformed by the copolymerization of propylene oxide and carbondioxide, as first reported by Inoue et al [1]. The glass transitiontemperature of PPC is near room temperature (Tg z 25 �C to 45 �C)[2,3] and depends on molecular weight and backbone structuralarrangement [4]. PPC has a relatively low decomposition temper-ature, with the five percent mass loss (Td�5%) widely reported to beapproximately 180 �C [5]. This has limited its commercial applica-tions as a replacement for conventional plastics but makes itattractive as a sacrificial material. Numerous efforts have beenmade to increase the thermal stability andmechanical properties ofPPC [3,5e14]. Thermal decomposition in inert environmentsproceeds via unzipping into cyclic propylene carbonate (4-methyl-1,3-dioxolan-2-one) at lower temperatures (ca. 180 �C) followed byrandom chain scission at higher temperatures (ca. 250 �C) [15]. Thedecomposition mechanism depends on molecular weight [3],temperature, ambient gaseous environment [16,17], and additives[15]. Techniques to increase the thermal stability have beenreported [5,6,11e13]. Methods to lower the decompositiontemperature for use as a sacrificial material has also been reported[18e21].

).

All rights reserved.

The low decomposition temperature and volatile products makePPC an ideal sacrificial material for microelectronic applicationswhich are constrained to temperatures between ca. 100 �C and250 �C. This is well within the stability range of epoxy-basedproducts. Gaseous cavities can be created by forming spatial PPCpatterns followed by overcoating with a dielectric material. The PPCdecomposition products can diffuse through the overcoat leavingan embedded gas cavity within the overcoat whose shape isdetermined by the original shape of the PPC pattern. Embeddedcavities are potentially valuable in the packaging of microchips andmicroelectromechanical systems (MEMS).

Air cavities or other gas-filled spaces are valuable in a variety ofsemiconductor manufacturing and microchip packaging technolo-gies including microfluidic channels for microprocessor cooling[22], lab-on-a-chip applications [23], air-insulated electrical signallines [24], and vacuum packaging of MEMS resonators [25]. Aircavities surrounding a core waveguide in fiber optical cables haveshown low losses [26]. Similar low losses were demonstrated inporous fiber structures with subwavelength holes [27]. Thermallydecomposed PPC loaded with aluminum nitride particles has beenstudied for use in tape automated bonding [17,28].

A variety of polycarbonates including PPC [18e20] and poly-norbornene formulations [29], have been previously studied for useas sacrificial materials. Air cavities in a variety of dimensions havebeen reported using polynorbornene [24] and SiO2 overcoats [30].A small quantity of a photoacid generator (PAG), which createsa catalytic amount of an acid upon exposure to ultraviolet (UV)radiation, reduces the decomposition temperature, enabling direct

mailto:[email protected]

www.sciencedirect.com/science/journal/01413910

http://www.elsevier.com/locate/polydegstab

http://dx.doi.org/10.1016/j.polymdegradstab.2010.12.003



T.J. Spencer, P.A. Kohl / Polymer Degradation and Stability 96 (2011) 686e702 687

photopatterning of the material without the need for a hardmask[19]. Small amounts of residue were observed in the cavitiesfollowing decomposition [19,24,25]. Improved decomposition wasreported by using a combination of PAGs generating acids withdifferent vapor pressures [18].

In this work, the thermal decomposition behavior of solvent-cast PPC is evaluated using dynamic thermogravimetric analysis(DTGA) and compared with pure species decomposition properties.PPC from two sources with different molecular weights anddifferent impurities were compared. The effect of nitrogen and airenvironments in the PPC decomposition chamber was studied. Theinfluence of casting solvent on decomposition behavior has beenevaluated using six polar solvents. Acid catalyzed decomposition ofthe PPC film was evaluated using triflic acid and UV sensitivephotoacid generators (PAGs).

The thermal decomposition data as a function of temperature andtime was analyzed using FlynneWalleOzawa and CoatseRedfernmethods to extract values for the activation energy and pre-exponential factor. These kinetic parameters are important forpredictive modeling of air cavity formation and designing controlleddecomposition recipes [30]. With a sufficient amount of photoacidloading, decomposition proceeds at reduced temperatures via chainunzipping. The acid is created upon ultraviolet (UV) irradiation of thePAG.Without irradiation of the PAG or samples without acid loading,the PPC decomposed at temperatures above 180 �C. Residue from thenon-volatile component of the PAG salts was found to be propor-tional to the initial PAG concentration in the PPC. Low concentrationsof PAG showed anomalously slower decomposition rates after UVexposure and resulting in higher final decomposition temperaturesdue to foaming of the sample during decomposition.

2. Background

Poly(propylene carbonate) is synthesized via copolymerizationof propylene oxide and carbon dioxide in the presence of a catalystat elevated pressures. The primary backbone component ispropylene carbonate as shown in Fig. 1. Impurities, especially etherlinkages are present in the backbone as an undesired by-product ofthe synthesis, at typically less than 10 mol percent.

Molecular weights are typically on the order of 104e105 g/mol[31] and yield of the alternating copolymer is highly dependent oncatalyst and cocatalyst selection. Increased polyether contentdepresses the glass transition temperature by as much as 30 �Cwhile the thermal stability increases with increasing number ofether linkages.

Depending on polymerization conditions, catalyst, and cocata-lyst, the local structure of backbone components varies, affectingthermal stability and mechanical properties. Ring-opening ofpropylene oxide may occur at either the a or b-positions. The a-sitemechanism opens the bond between the oxygen atom and thecenter carbon (CH group) of the propylene unit whereas the b-sitemechanism opens the bond between the oxygen and the CH2 groupin the ring. Thus, the methyl group can be on the alpha or betacarbon with respect to the oxygen in the ring-opened propyleneoxide and polymer structure depends on which ring-opening

Fig. 1. General structure of poly(propylene carbonate). Polyether content is typicallyless than 10 mol%, corresponding to x � 0.90.

position dominates. The regiostructure of PPC can exist in threedifferent configurations as shown in Fig. 2. These configurationshave been termed head-to-head (HH), tail-to-tail (TT), and head-to-tail (HT) configurations. An important feature of the HH and TTconfigurations is they coexist with each other in alternating units orwith chains of head-to-tail configurations. Poly(propylenecarbonate) cannot exist exclusively of HH or TT configurationsbecause they must be coupled with another configuration. PPCmade from units in the head-to-tail arrangement, however, canexist solely in this configuration.

Choice of catalyst is important in order to minimize the etherlinkages, maximize yield of the alternating copolymer, and residualcatalyst may influence decomposition. Common catalysts used inPPC polymerization include heterogeneous zinc compounds,aluminum porphyrins, CoIII(salen), CrIII(salen). High copolymeryield results in fewer ether linkages but increases susceptibility todecomposition via chain unzipping. The unzipping mechanismoccurs in highly regular PPC because terminal alcohols allow thecyclic monomer to form and detach from the backbone. Because theendgroup remains terminated with an alcohol, the cyclizationcontinues and the backbone of the polymer unzips.

Molecular weights as low as 13,000 g/mol [32] and as high as144,600 g/mol [15] have been synthesized in the presence of zincglutarate, zinc-cobalt and zincenickel complexes [33]. Highermolecular weights are difficult to synthesize due to spontaneousdepolymerization into the more thermodynamically stable cyclicpropylene carbonate, especially at elevated temperatures [4],although 176,000 g/mol PPC was reported using allyl glycidyl etheras a cross-linker [34]. Catalyst and cocatalyst determine regios-tructure of high molecular weight PPC with cationic initiatorsopening propylene oxide at both a- and b-positions, while anionicinitiators, coordination initiators, and Lewis base cocatalysts openthe epoxide ring primarily in the a-position. Ring-opening at the a-position gives a more regular structure with an increased numberof HT linkages and reduced ether linkages [4,35,36]. Copolymeryield can be improved by vacuum activation of the catalyst prior topolymerization [3].

Residual catalyst may impact thermal stability since removal ofcatalyst has been suggested to increase thermal stability by inhib-iting chain scission [15]. The choice of solvent for use in catalystrecovery may also influence thermal stability. Removal of catalystafter polymerization is performed by precipitation in chlorinatedsolvents, typically chloroform [7,8,15,37] or dichloromethane[4,34,35,38,39] and quenching with dilute hydrochloric acid in

Fig. 2. Backbone structure configurations of PPC. It should be noted that Head-to-Headand Tail-to-Tail configurations are effectively the same monomer unit and cannot existin exclusively one configuration. The Head-to-Tail configuration can exist solely asmembers of these other configurations.

Fig. 3. Chain unzipping in high molecular weight PPC with head-to-tail regiostructure.Unzipping can occur via a) alkoxide biting or b) carbonate biting. DFT calculations havesuggested that mechanism b) is more energetically favorable.

T.J. Spencer, P.A. Kohl / Polymer Degradation and Stability 96 (2011) 686e702688

methanol. Acetone has also been reported for purification/catalystprecipitation [14]. Other studies have suggested residual catalystdoes not alter thermal stability.

The mechanical properties of PPC were not evaluated here butare important for sacrificial material patterning via photo-patterning and imprint lithography. Small weight percentages ofPAGs used in this study and residual solvents are not expected tochangemechanical properties significantly from the bulk. However,these properties will be influenced by polyether content, molecularweight, and region-structure. Chemical foaming in the presence oflow quantities of PAG.

The backbone structure and influence of additives have been thesubject of numerous other studies [2,6e9,40e47] and depend onpolyether content [2,43] and backbone structure [4]. Glass transi-tion temperatures of the alternating copolymer are between 35 �Cand 40 �C, but are as low as 8 �C with greater polyether content.Increasing molecular weight withmore HT units increases the glasstransition temperature to 45 �C and improves tensile andcompressive strength [4]. Tensile strength of the alternatingcopolymer is reported in the range of 22 MPa [6,10,48] and 32 MPa[7,8,43], but may be as low as 8 MPa [44,46,49].

The influence of small quantities of additives (less than 10 wt%)on mechanical properties has not been widely reported. Weightpercentages of octadecanoic acid of 0.28e2.27% showed a slightincrease in glass transition temperature. Other hydrogen-bondedcomplexes including clay nanoparticles reported an increase inshear modulus and viscosity at 2.5 wt% loading [13]. Glass transi-tion temperature, tensile strength, and Young’s modulus increasedat loadings between 1 wt% and 10 wt% [6,46]. Chemical foaming[42,50] will change mechanical properties from that of the bulkpolymer due to porosity. Blends with poly(butylene succinate) [51],poly(3-hydroxybutyrate) [47], poly(lactic acid) [52], and glycerolstarch [49] increased storage modulus and tensile strength linearlywith additive concentration. Terpolymerization with bisphenol-A[48], cyclohexene carbonate [2,43], (2-napthyloxy)methyl oxirane[8], and N-(2,3-epoxyl-propyl)carbazole [7] raised tensile strengthand glass transition temperature while maleic anhydride decreasedtensile strength.

The low degradation temperature of PPC has also limited broadcommercial adoption. The onset of degradation is widely reportedas the temperature where 5 percent weight loss occurs (Td�5%) andis highly dependent on heating rate. Isothermal decomposition andslow heating rates (<5 �C/min) were used to obtain a degradationtemperature as low as 180 �C, whereas studies using faster heatingrates (10 �C/min to 25 �C/min) reported degradation temperaturesabove 240 �C [3,5,6,8,10,12,13,15,19,20,32,35,41,43,44,46,49,51,53].Higher decomposition temperature reported using faster heatingrates is an artifact of the measurement technique. Isothermaldegradation in these studies likely occurs near 180 �C.

The decomposition mechanism of the backbone depends ondecomposition environment, molecular weight of the polymer,number of ether linkages, and backbone regiostructure. Inoxygenated atmospheres, decomposition primarily occurs via chainscission into combustion products. In a nitrogen environment, thebackbone primarily decomposes via chain unzipping from the endsof the polymer chain or unzipping following chain scission[16,17,28]. Higher molecular weight structures decompose almostexclusively via the chain unzipping mechanism from terminaleOHgroups [3,16]. Ether linkages inhibit chain unzipping and result inlarger molecular weight decomposition fragments [19]. HT struc-tured backbones most easily undergo decomposition via chainunzipping due to the highly regular structure. Although it is unclearwhich mechanism is dominant, unzipping into cyclic propylenecarbonate can occur via either carbonate backbiting or alkoxidebackbiting [44]. The chain unzipping mechanisms of highly regular

head-to-tail PPC is illustrated in Fig. 3. In the alkoxide backbitingmechanism, a strong nucleophile attacks the carbonyl atom, asshown from the left side (a) of Fig. 3. In the carbonate backbitingmechanism, a weak nucleophile attacks an electrophilic carbon asshown on the right side (b) of Fig. 3. Density functional theory (DFT)simulations suggest carbonate backbiting (b) has a lower activationenergy [54]. HH and TT regiostructures inhibit this mechanism butmay undergo sidegroup rearrangement to allow unzipping toproceed, albeit at a slightly lower rate.

Techniques to raise the decomposition temperature of PPCinclude blending with inorganic and organic filler materials such ascalcium carbonate [9,50], starch [49,55], other polymers [14,47,52],montmorillonite [46,56] and organoclays [13], polymerizing withadditional backbone components [2,7,8,10,39,43,57], and end-capping of carbonate groups [5,12]. The low degradation temper-ature has also been exploited for use in microelectronic cooling andpackaging applications [19,20,24,25,58]. Acid catalyzed degrada-tion lowered the decomposition temperature using UV sensitivePAG [18,19,21]. Degradation in soil and buffer solution has also beenreported [16].

Techniques to improve thermal stability via the addition offillers have been modestly successful. Blends with calciumcarbonate particles showed improved thermal stability with largeparticles, but nanoparticles actually lowered the thermal decom-position temperature. Blends of PPC with dried starch had slightlydepressed decomposition onset temperatures but further decom-position was not observed until higher temperatures [49]. Residuein these blends is attributable to the filler particles.

Blending PPC with other polymers has also been attempted toimprove the thermal stability. Physical blends with poly(butylenesuccinate) (PBS), poly(3-hydroxybutyrate), and poly(lactic acid)with carbon black elevated the onset decomposition temperatureby about 50 �C at heating rates of 20 �C/min. The thermal decom-position profile for PBS blends was segmented with an initialdecomposition proportional to the weight percentage of PPC andthe secondary decomposition due to the PBS content [51].

End-capping of the terminal hydroxyl groups has beendescribed as a way to increase thermal stability to inhibit theunzipping mechanism. Small quantities of maleic anhydride,benzoyl chloride, ethyl silicate, acetic anhydride, and phosphorusoxychloride were used to delay the onset of decomposition [5].End-capping with maleic anhydride has been evaluated as a func-tion of backbone content relative to propylene oxide [10] andblended with ethyl cellulose [53].

Hydrogen bonding between polymer blends and additives mayalso stabilize PPC against thermal decomposition. The addition ofoctadecanoic acid at weight fractions between 0.28 wt% and 2.17 wt% demonstrated improved thermal stability [12]. Calcium stearate,a derivative of octadecanoic acid, also stabilized PPC by coordi-nating with the metal ion [11]. Hydrogen bonding has also beenobserved in polymer blends [14]. Low weight fractions of organicclays capable of hydrogen bonding dispersed in PPC increasedthermal stability [6,13]. Additionally, end-capping agents may be

Fig. 4. Photoacid generator structures; a) 4-methylphenyl[4-(1-methylethyl)phenyl] tetrakis(pentafluorophenyl) borate (Rhodorsil-FABA) b) (4-tert-butylphenyl)iodonium tris(perfluoromethyl sulfonyl) methide (3M-Methide).


hydrolyzed in the presence of water to form hydrogen bondingstructures which augment thermal stability in addition tosuppression of the unzipping mechanism [5].

Small quantities of residue remain from the decomposedphotosensitive PPC formulations. X-ray photoelectron spectroscopy(XPS) results indicated the residue contained primarily fragmentsof the PAG [59]. Energy dispersive x-ray fluorescence (EDS) analysisconfirmed the primary residue resulted from the PAG [25]. Theresidue may also contain higher molecular weight products fromdecomposition such as alkenes and ketones as observed by massspectroscopy [19]. These products may result from dehydrogena-tion in the presence of acid or due to irregularities in the backbone.

Dynamic thermogravimetric analysis (DTGA) is used to assessthe thermal stability of PPC by measuring mass loss as a function oftemperature at a fixed heating rate. In prior studies, heating rateshave been as low as 5 �C/min and as high as 30 �C/min. Rapidheating rates are a poor indicator of overall film stability. Slowheating rates enable clearer differentiation of decompositioncharacteristics.

Thermal decomposition of solids as a function of time isdescribed using the Arrhenius equation multiplied by an expres-sion for the fractional decomposition, a, Equation (1).

dadt

¼ A$e�EaR$T $f ðaÞ (1)

Fig. 5. QPAC PPC in (a) air and (c) nitrogen; Novomer PPC

where A is the pre-exponential constant, Ea is the energy of acti-vation, t is time, R is the ideal gas constant, and f(a) can beexpressed by a variety of power law or logarithmic functions. Underdynamic conditions where the sample temperature is being line-arly increased at a heating rate b, Eq. (1) can be expressed asa function of temperature as shown in Equation (2).

dadT

¼ Ab$e

�EaR$T $f ðaÞ (2)

The integrated quantity, g(a) is shown in Equation (3).

gðaÞ ¼Za0

daf ðaÞ ¼

ZT0

Ab$e

�EaR$T (3)

Numerous expressions with associated physical mechanismshave been proposed to describe thermal decomposition curveswith both sigmoidal and deceleration functions. These functionscan be used to describe effects of boundary conditions, diffusionlimitations, and multiple nucleation sites on an individual particle[60], which are not as relevant here.

The ASTM standard technique for dynamic thermal decompo-sition analysis uses the method of Kissinger, which adopts Eq. (3)and defines f(a)¼(1 � a)n, where n is the reaction order. Thus, Eq.(3) can be rearranged into Equation (4).

(Mw ¼ 90,000; PD ¼ 1.5) in (b) air and (d) nitrogen.

Table 1Pure species kinetic parameters calculated using CoatseRedfern technique for Airand Nitrogen environments.

Ea (kJ/mol) A (s�1) R2 amax

QPAC e Air 119 7.22e5 0.9459 0.58QPAC e Nitrogen 147 1.70e15 0.9883 0.41Novomer e Air 120 3.21e5 0.9713 0.57Novomer e Nitrogen 126 2.07e6 0.9569 0.48


ln

b

T2max

!¼ ln

�A$REa

�þ ln

�nð1� amaxÞn�1

�� EaR$Tmax

(4)

where Tmax is the temperature at themaximum decomposition rate(inflection point in dynamic TGA curve) and amax is the degree ofconversion at the inflection point. For a first order reaction (n ¼ 1)[61e63], Eq. (4) reduces to Eq. (5).

ln

b

T2max

!¼ ln

�A$REa

�� EaR$Tmax

(5)

By plotting the left side of Eq. (1) vs. 1/Tmax and performinga linear regression analysis, the slope and intercept can be used todetermine the activation energy and pre-exponential factor,respectively.

The CoatseRedfern method of analysis offers an advantage overother techniques because a single data set is needed instead ofmultiple heating rates, as required by the Kissinger and Ozawamethods. The CoatseRedfern method is expressed in Equation (6).

ln

gðaÞT2max

!¼ ln

�A$Rb$Ea

� �1� 2$R$T

Ea

�� EaR$Tmax

(6)

The algebraic expression g(a) depends on the physical mecha-nism of decomposition [32]. For a first order reaction with a singlenucleation site on an individual particle, g(a) ¼ �ln(1 � a) and Eq.(6) and is expressed in Equation (7).

ln��lnð1� aÞ

T2

�¼ ln

�A$Rb$Ea

� �1� 2$R$T

Ea

�� EaR$T

(7)

Fig. 6. QPAC PPC dissolved in g-butyrolactone decomposed in a nitrogen p

Thus, by plotting ln[�ln(1 � a)/Tmax2 ] vs. �1/T, the slope and

intercept can be used to determine the activation energy and pre-exponential factor. Alternatively, kinetic parameters for higher orderreactions can be determined by fitting for the best regression value atvarious powers. First order (linear) regression of decomposition data isindicative of single reactionmechanisms,whereas non-linear plots areindicative of complex or competing reactions [5]. High molecularweight PPC at low heating rates decomposes primarily by backboneunzipping and thus first order reaction kinetics are considered here.Higher order regression has greater utility when competing mecha-nisms are present (i.e. unzipping vs. chain scission), but only givesa better mathematical fit of the curve and offer little insight intophysical mechanisms. This analysis is not presented here becausedirect comparison of activation energies is impossible with this tech-nique, but has been used in other PPC decomposition studies [5,20].

3. Experimental

3.1. Film preparation

Poly(propylene carbonate) was obtained from QPac-Aldrich andNovomer Inc. and studied as received without purification. Solu-tions of polymer, PAG, and solvent were mixed for one week ona bottle roller. The films were spin-cast to 10 mm thickness on glassslides. The PPC filmwas soft baked at 110 �C for 20min to evaporatethe solvent. Following solvent volatilization, film samples wereremoved from the glass slide using a razor blade and tweezersbefore being loaded into TGA weighing pans.

Filmswere also decomposedunder glass on a hotplate to observefilm decomposition with various PAG loadings. The samples wereplaced under glass and the chamber was purged for 1 h withnitrogen. The hotplate temperature was increased at a rate of 1�C/min to a final temperature of 220 �C. This provided a supplementto quantitative results and allowed for better understanding of thedecomposition mechanism for different photoacid concentrations.

3.2. Thermogravimetric analysis (TGA)

DTGAwas performed using a Seiko TG/DTA 320 system. The TGAscale was re-zeroed for each weighing tin before every experiment.

urged environment at (a) 0.5 �C/min, (b) 2 �C/min, and (c) 10 �C/min.

Fig. 7. Novomer PPC (Mw ¼ 90,000; PD ¼ 1.5) in N2 at heating rates of (a) 0.5 �C/min, (b) 2 �C/min, and (c) 10 �C/min.


Samples between 10 mg and 15 mg were loaded onto the micro-balance and the system purged for 30 min with nitrogen ata flowrate of 100 mL/min. The thin PPC films were cut into largesections (ca. 10 cm2) and folded several times to the size of the TGAweighing tin in order to achieve the large masses (10e15 mg)required for accurateweight loss measurement. Oxygen content fornitrogen purged decompositionwas below 100 ppm. Samples weredecomposed at heating rates of 0.5 �C/min, 2 �C/min, and10 �C/min. The temperaturewas increased from 30 �C to 300 �C and

Fig. 8. Novomer PPC (Mw ¼ 100,000; PD ¼ 1.9) in N2 at heatin

data were sampled at a frequency of 0.2 Hz for most studies. Afterdecomposition, residue in the pan was observed qualitatively.

3.3. Solvent comparison

PPC was dissolved in solvent to allow spincoating of thin filmson a substrate. The primary solvent used in this study wasg-butyrolactone (GBL). Additional polar solvents including acetone,anisole, cyclohexanone, trichloroethylene, and dichloromethane

g rates of (a) 0.5 �C/min, (b) 2 �C/min, and (c) 10 �C/min.

Table 2Kinetic parameters calculated using CoatseRedfern technique for PPC dissolved in g-butyrolactone and for photoacid generators used in this study.


QPAC e 0.5 �C/min 285 2.66e25 0.9334 0.53QPAC e 2 �C/min 407 1.89e39 0.9312 0.57QPAC e 10 �C/mina 233 9.39e19 0.7973 0.55Novomer 90k e 0.5 �C/min 295 5.82e26 0.9989 0.43Novomer 90k e 2 �C/min 247 5.90e20 0.9958 0.47Novomer 90k e 10 �C/min 234 9.36e19 0.9992 0.46Novomer 100k e 0.5 �C/min 325 6.13e29 0.9973 0.60Novomer 100k e 2 �C/min 288 2.11e25 0.9924 0.62Novomer 100k e 10 �C/min 399 2.04e37 0.9515 0.60

a The poor correlation factor for QPac heated at 10 �C/min may be due to non-linear heating in the TGA (seen as a step backwards in temperature in the TGA inFig. 6) or due to the competing chain scission and unzipping mechanisms resultingfrom residual oxygen in the chamber.


were analyzed in this study. All solvents were obtained from Sig-maeAldrich. Samples dissolved in GBL were analyzed at heatingrates of 0.5 �C/min, 2 �C/min, and 10 �C/min. Experiments usingdifferent solvents were heated at 0.5 �C/min for clearer differenti-ation of decomposition effects.

3.4. PPC films with photoacid generator (PAG)

Acid catalyzed decomposition of polycarbonates has previouslybeen reported for photopatterning of sacrificial materials [19]. Inthis study, triflic acid catalyzed decomposition was compared withtwo photoacid generators which generate equivalent strengthsuper-acids (i.e. entirely dissociated). The PAGs investigated in thisstudy were the iodonium salts 4-methylphenyl [4-(1-methylethyl)phenyl] tetrakis(pentafluorophenyl) borate (referred to hereafter asRhodorsil-FABA) and bis(4-tert-butylphenyl)iodonium tris(per-fluoromethyl sulfonyl) methide (referred to hereafter as 3M-Methide). These photoacids were investigated because they hadpreviously been identified as the most effective for acid catalyzeddecomposition of PPC [59]. The structure of the Rhodorsil-FABAPAG is illustrated in Fig. 4a. The structure of the Methide PAG isillustrated in Fig. 4b.

Proper comparison of the PAGs requires equivalent molar acidcontent after UV exposure. The molecular weight of the Rhodorsil-FABA is 1016.26 g/mol and themolecular weight of the 3M-Methideis 804 g/mol. More importantly, the weight difference is primarily

Fig. 9. Thermal decomposition of photoacid generators in nitrogen heated at 0.5�C/min: a) (4-tert-butylphenyl)iodonium tris(perfluoromethyl sulfonyl) methide (3M-Methide); b) 4-methylphenyl[4-(1-methylethyl)phenyl] tetrakis(pentafluorophenyl)borate (Rhodorsil-FABA).

due to the structure of the anion, which correlates to the vaporpressure of the conjugate acid formed upon exposure to UV light.Tetrakis-pentafluorophenyl boric acid (generated by Rhodorsil-FABA) has essentially no vapor pressure at room temperature,whereas triflic acid (generated by 3M-Methide) has a vapor pres-sure of 8 torr at 25 �C. Consequently, it is expected the Rhodorsil-FABA PAG will remain present in ever increasing concentrationduring decomposition, while the 3M-Methide PAG will decrease inconcentration as the reaction proceeds due to evaporation.

3.5. Fourier transform infrared (FTIR) spectroscopy characterization

Fourier Transform Infrared (FTIR) spectroscopy was used in aneffort to identify influences of casting solvent and photoacidgenerators. FTIR data were collected using a Nicolet Magna IR 560spectrometer. Data were collected in transmission mode ina nitrogen purged chamber under a nitrogen flowrate of 50 mL/minwith 2 cm�1 resolution and averaged over 512 scans. Films werecast on a KBr disk at a thickness of approximately 1 um. Backgroundscans of the KBr disk were performed prior to data collection foreach polymer film.

4. Results

4.1. Bulk polymer comparison (no solvent)

QPac and Novomer PPC were used as-received and decomposedin nitrogen and air environments. The heating rate for all sampleswas 2 �C/min. The TGA curves are shown in Fig. 5. Kinetic param-eters of activation energy (Ea) and pre-exponential factor (A) werecalculated for a first order reaction using the CoatseRedfernmethod. These results, including the linear regression correlationvalue (R2) and the decomposition completion fraction at themaximum weight loss rate (amax), are listed in Table 1.

4.2. TGAs of cast films in g-butyrolactone and M.W. comparison

QPac and Novomer PPC dissolved in g-butyrolactone with noadditional additives were spin-coated as films 10 mm thick and thesolvent was evaporated by softbaking on a hotplate at 100 �C. Filmswere decomposed in a nitrogen purged environment at heatingrates of 0.5 �C/min, 2 �C/min, and 10 �C/min. The dynamic ther-mogravimetric analysis plot for the QPac PPC is shown in Fig. 6.DTGA curves of two different molecular weight PPCs of 90,000g/mol and 100,000 g/mol from Novomer are shown in Figs. 7 and 8,respectively. Activation energies (Ea), pre-exponential factor, andcorrelation factor (R2) obtained using the CoatseRedfern methodfor g-butyrolactone dissolved blends at all heating rates are listedin Table 2 as well as the maximum completion fraction (amax).

4.3. TGAs of solvent-cast films with photoacid generator

The two photoacid generators used in this study, (4-tert-butyl-phenyl)iodonium tris(perfluoromethyl sulfonyl) methide (3M-Methide) and 4-methylphenyl[4-(1-methylethyl)phenyl] tetrakis(pentafluorophenyl) borate (Rhodorsil-FABA), were heated at0.5 �C/min and decomposed at in a nitrogen purged environment.

Table 3Kinetic parameters calculated using CoatseRedfern technique for photoacidgenerators used in this study heated at 0.5 �C/min.


Rhodorsil-FABA 155 1.33e13 0.9776 0.693M-Methide 303 5.61e30 0.9876 0.58

Fig. 10. Solvent-cast Novomer (MW ¼ 100k) PPC Weight Fraction vs. Temperature in nitrogen heated at 0.5 �C/min compared with as-received sample: (a) methylene chloride, (b)anisole, (c) acetone, (d) as-received (no solvent added), (e) g-butyrolactone, (f) trichloroethylene, and (g) cyclohexanone.

Table 4Kinetic parameters calculated using CoatseRedfern technique.


Acetonea 431 1.23e43 0.8621 0.72Anisole 529 2.26e56 0.9859 0.67g-butyrolactone 583 4.13e21 0.9466 0.60Trichloroethylene 457 3.87e43 0.9770 0.52Methylene chloride 458 4.71e43 0.9841 0.48No solvent added 469 2.07e47 0.9972 0.60

a The poor correlation factor for the kinetic parameters of the acetone dissolvedsample is due to inversion of the curve in the middle of the decomposition range,seen as a step backwards in the TGA plot in Fig. 10.


The PAGs were used in their as-received state without drying. ThePAGs were examined by thermal analysis either neat or dissolved insolvent and dried in a TGA pan. The DTGA curves for these PAGs areshown in Fig. 9. The activation energy (Ea), pre-exponential factor,and regression correlation (R2) obtained using the CoatseRedfernmethod are listed in Table 3with themaximum completion fraction(amax).

4.4. TGAs of photoacid generators

The thermal decomposition of the PAG is important because itcan facilitate the acid activated decomposition mechanism of thePPC. Acid is generated upon thermal dissociation in a mannersimilar to UV exposure. Both PAGs exhibit five percent mass loss attemperatures near the decomposition temperature of PPC(Td�5%,Methide ¼ 182 �C; Td�5%, FABA ¼ 187 �C). The 3M-Methide PAGcompletely decomposes rapidly and thus may completely volatilizewith little effect on PPC films while Rhodorsil-FABA does notdecompose completely until 210 �C, allowing the thermallygenerated acid to play an active role in decomposition of unexposedfilms.

4.5. TGAs of Novomer films cast from different solvents

DissolvingPPC in solvent allowscontrol offilm thicknessbyvaryingspin speed. PPC dissolves readily in numerous polar solvents includingmethylene chloride, anisole, acetone, g-butyrolactone, trichloroethy-lene, and cyclohexanone. Novomer PPC with a molecular weight of100,000 g/molwas dissolved in each of these solvents, spun as a 10mmthick film, and decomposed in a nitrogen environment at a heatingrate of 0.5 �C/min. TheDTGAcurves for thesefilms are shown in Fig.10.Table 4 shows the kinetic parameters extracted using the Coat-seRedfern method. Residual solvent or interaction with PPC hasa notable effect on the temperature for the onset of decomposition, asreflected in the activation energy and pre-exponential factor.

No easily discernible trend can be seen on how the castingsolvent may impact the thermal decomposition of PPC. Comparisonof physical properties such as dipole moment, heat of vaporization,and boiling point show no pattern to the change in decomposition

temperature. This suggests a more complex interaction betweensolvent and polymer backbone and may be an interaction with theterminal alcohol groups to suppress initiation or at the carbonylsites to inhibit unzipping propagation. End-capping at the chainterminations seem most likely as this has been observed in othersystems [5] where no evidence of hydrogen bonding was reported.All solvents in this study are readily hydrolyzed in the presence ofwater and these products may also be responsible for end-cappingof the polymer backbone. These effects may depend highly onmolecular weight, polydispersity, and degree of film drying.

4.6. TGAs of PPC films with different photoacid generatorconcentrations

The effects of two different photoacid generators (3M-Methideand Rhodorsil-FABA)were studied by varying concentration and UVexposure. The Novomer PPC with a molecular weight of 90,000g/mol and QPac PPC with a molecular weight of 50,000 g/mol wereinvestigated by first dissolving the PAG in g-butyrolactone with thepolymer andmixing on a bottle roller. The photoacid concentrationsby weight were 3.00 percent 3M-Methide and 3.75 percent Rho-dorsil-FABA, which corresponds to an acid loading of approximately1 mol of acid to 440 monomer units of the polymer. The averagechain length of the Novomer PPC with a molecular weight of90,000 g/mol is approximately 880 monomer units, which meansthere are 2 acid molecules per polymer chain. Upon UV exposure or

Fig. 11. Novomer (thin black lines) and QPac (thick aqua lines) PPC with 3.75 wt% Rhodorsil-FABA PAG dissolved in g-butyrolactone: (a),(b), (g), and (h) heated at 0. 5 �C/min, (c), (d),(i), and (j) heated at 2 �C/min, and (e), (f), (k), and (l) heated at 10 �C/min. Curves (aef) were exposed to 1 J/cm2 of UV light at a wavelength of l ¼ 248 nm.


thermal activation, the organic cation of the PAG decomposes. Theremaining anion has a highly dispersed negative charge and thusgenerates a strong Brønsted acid. Since the acid is regenerated aftereach propylene carbonate decomposition, there is sufficient acid toinitiate decomposition from each end of the backbone.

Ether linkages consume the acid upon decomposition byterminating the backbone with an alcohol group which does notallow polymer unzipping or acid regeneration. A lower acidconcentration of 0.200 wt% 3M-Methide or 0.250 wt% Rhodorsil-FABA was also studied. This lower acid concentration correspondsto a mole ratio of 6600 monomer units per acid molecule. All UVexposures used a dose of 1 J/cm2 at l ¼ 248 nm.

Fig. 12. Novomer (thin black lines) and QPac (thick aqua lines) PPC with 3.0 wt% 3M-Methideand (j) heated at 2 �C/min, and (e), (f), (k), and (l) heated at 10 �C/min. Curves (aef) were

The DTGA curves for UV exposed and unexposed films with3.75 wt% Rhodorsil-FABA in QPac and Novomer (M.W. ¼ 90,000g/mol) at heating rates of 0.5 �C/min, 2 �C/min, and 10 �C/min inanitrogen environment are shown in Fig.11. TheUVexposed samplesexhibit a low temperature, gradual decomposition while the unex-posed films show a sharp decomposition at higher temperature. Theshift in the decomposition temperature to lower values reflects theacid activation step. Additionally, the mass loss occurs over a broadtemperature range in the exposed case, compared with a relativelynarrow temperature range in the unexposed case. Decomposition ofQPac and Novomer (M.W. ¼ 90,000) loaded with 3.00 wt% 3M-Methide PAG at heating rates of 0.5 �C/min, 2 �C/min, and 10 �C/min

PAG dissolved in g-butyrolactone: (a), (b), (g), and (h) heated at 0. 5 �C/min, (c), (d), (i),exposed to 1 J/cm2 of UV light at a wavelength of l ¼ 248 nm.

Fig. 13. Novomer (thin black lines) and QPac (thick aqua lines) PPC with 0.25 wt% Rhodorsil-FABA PAG dissolved in g-butyrolactone: (a), (c), (f), and (g) heated at 0. 5 �C/min, (b), (d),(i), and (j) heated at 2 �C/min, and (e), (h), (k), and (l) heated at 10 �C/min. Curves (aee) and (h) were exposed to 1 J/cm2 of UV light at a wavelength of l ¼ 248 nm.


with and without UV exposure is shown in Fig. 12. Shown in Fig. 13are the low acid loading of the Rhodorsil-FABA PAG (0.25 wt%) forthe exposed and unexposed cases at the heating rates listed above forQPac and Novomer (M.W. ¼ 90,000). Fig. 14 shows the low acidloadingof the 3M-MethidePAG (0.20wt%) forQPac and90,000 g/molNovomer with UV exposure and without UV exposure at the sameheating rates.

The 3M-Methide PAG had a lower decomposition onsettemperature than the Rhodorsil-FABA PAG for the UV exposedsamples. Similar mass loss rates were observed in all exposed cases,consistent with a decomposition and steadily increasing rate of

Fig. 14. Novomer (thin black lines) and QPac (thick aqua lines) PPC with 0.20 wt% 3M-Meth(h), and (i) heated at 2 �C/min, and (e), (j), (k), and (l) heated at 10 �C/min. Curves (aee) a

product loss. This is consistent with depolymerization into thecyclic monomer and a mass loss due to the evaporation of theliquid. In the unexposed cases, the films with the FABA PAG hada slightly lower decomposition onset temperature and all samplesexhibited behavior consistent with rapid decomposition and vola-tilization. This is consistent with depolymerization followed byimmediate evaporation although chain scission may be competing.Table 5 summarizes the CoatseRedfern kinetic parameters fora first order reaction and completion fraction at the maximumdecomposition rate (amax) for the UV exposed cases for both basepolymers, both photoacid concentrations of each PAG, and all

ide PAG dissolved in g-butyrolactone: (a), (b), (f), and (g) heated at 0. 5 �C/min, (c), (d),nd (k) were exposed to 1 J/cm2 of UV light at a wavelength of l ¼ 248 nm.

Table 5CoatseRedfern kinetic parameters for a first order reaction for UV exposed PPC filmsloaded with photoacid generator. All films were spun to a thickness of 10 mmwhendissolved in g-butyrolactone. Residual solvent was removed using a softbaketemperature of 100 �C.


3.75 wt. % Rhodorsil-FABAQPac e 0.5 �C/min 98.2 2.90e10 0.9980 0.79QPac e 2 �C/min 82.5 9.96e7 0.9993 0.80QPac e 10 �C/min 76.4 9.86e6 0.9926 0.80Novomer e 0.5 �C/min 79.7 1.12e8 0.9967 0.84Novomer e 2 �C/min 69.4 1.85e6 0.9980 0.83Novomer e 10 �C/min 58.4 3.89e4 0.9979 0.833.00 wt. % 3M-MethideQPac e 0.5 �C/min 99.0 2.14e6 0.9995 0.80QPac e 2 �C/min 83.7 3.43e4 0.9983 0.80QPac e 10 �C/min 80.7 4.05e4 0.9951 0.84Novomer e 0.5 �C/min 78.8 1.01e4 0.9984 0.86Novomer e 2 �C/min 72.0 1.48e3 0.9967 0.91Novomer e 10 �C/min 69.1 4.11e3 0.9966 0.880.250 wt. % Rhodorsil-FABAQPac e 0.5 �C/min 70.1 1.44e0 0.9926 0.84QPac e 2 �C/min 69.4 7.31e0 0.9967 0.81QPac e 10 �C/min 80.1 9.25e2 0.9985 0.76Novomer e 0.5 �C/min 28.3 3.80e-1 0.9574 0.84Novomer e 2 �C/min 44.8 1.34e1 0.9847 0.90Novomer e 10 �C/min 47.8 3.57e0 0.9918 0.780.200 wt. % 3M-MethideQPac e 0.5 �C/min 62.4 2.99e-1 0.9980 0.74QPac e 2 �C/min 73.4 4.05e1 0.9996 0.76QPac e 10 �C/min 83.6 7.41e0 0.9980 0.71Novomer e 0.5 �C/min 26.8 4.57e-1 0.9460 0.89Novomer e 2 �C/min 31.6 9.90e-1 0.9818 0.87Novomer e 10 �C/min

Table 6CoatseRedfern kinetic parameters for a first order reaction for unexposed PPC filmsloaded with photoacid generator. All films were spun to a thickness of 10 mmwhendissolved in g-butyrolactone. Residual solvent was removed using a softbaketemperature of 100 �C.

Unexposed Ea (kJ/mol) A (s�1) R2 amax

3.75 wt. % FABAQPac e 0.5 �C/min 409 1.22e45 0.9744 0.69QPac e 2 �C/min 338 5.82e35 0.9392 0.67QPac e 10 �C/min 335 1.60e35 0.9959 0.70Novomer e 0.5 �C/min 659 1.37e73 0.9751 0.80Novomer e 2 �C/min 854 8.42e92 0.9733 0.78Novomer e 10 �C/min 845 3.32e88 0.9660 0.803.0 wt. % MethideQPac e 0.5 �C/min 489 9.99e48 0.9781 0.79QPac e 2 �C/min 569 7.86e57 0.9956 0.36QPac e 10 �C/min 414 7.85e37 0.9826 0.29Novomer e 0.5 �C/min 747 3.40e78 0.9889 0.80Novomer e 2 �C/min 783 6.01e80 0.9974 0.82Novomer e 10 �C/min 763 8.46e76 0.9821 0.730.25 wt. % FABAQPac e 0.5 �C/min 298 1.03e25 0.9889 0.73QPac e 2 �C/min 463 4.34e29 0.9978 0.70QPac e 10 �C/min 237 3.97e20 0.9651 0.74Novomer e 0.5 �C/min 263 2.17e21 0.9781 0.48Novomer e 2 �C/min 279 1.32e23 0.9112 0.40Novomer e 10 �C/min 312 3.26e26 0.9849 0.570.20 wt. % 3M-MethideQPac e 0.5 �C/min 409 2.37e38 0.9367 0.68QPac e 2 �C/min 485 1.47e43 0.9562 0.65QPac e 10 �C/min 376 4.41e37 0.9757 0.67Novomer e 0.5 �C/min 307 1.41e26 0.9862 0.55Novomer e 2 �C/min 417 2.53e36 0.9737 0.45Novomer e 10 �C/min 460 1.82e41 0.9797 0.66


heating rates. Table 6 lists the CoatseRedfern kinetic parametersand amax for the unexposed cases of these same films.

4.7. TGAs of PPC films with different PAG concentrations andsolvents

The influence of casting solvent on Rhodorsil-FABA PAG wasanalyzed at concentrations of 3.75 wt% and 0.25 wt%. The UVexposure dose was 1 J/cm2 at l ¼ 248 nm. All samples weredecomposed at a heating rate of 0.5 �C/min. Several casting solventswere compared including: g-butyrolactone, acetone, anisole, andtrichloroethylene. Eachmixture had 3.75wt% PAG. TheDTGA curvesfrom these films are shown in Fig. 15. The kinetic parameters werecalculated using the CoatseRedfernmethod. The first order reactionparameters are listed in Table 7 along with the completion fractionat the point of maximum decomposition rate. A second set ofsamples were prepared with 0.25 wt% PAG examining the solventeffect including: g-butyrolactone, acetone, anisole, and methylenechloride. The DTGA curves for these samples are shown in Fig. 16.The CoatseRedfern kinetic parameters are listed in Table 8.

4.8. TGAs of high PAG concentration and triflic acid

Novomermaterial loaded with 10.0 wt%, 3.75 wt%, and 0.25 wt%of Rhodorsil-FABA PAG is compared in Fig.17. The residue remainingat temperatures above 180 �C is primarily attributable to the initialPAG loading. This remaining fraction decomposes between 190 �Cand 230 �C. Although not illustrated here, the decomposition curveof the Rhodorsil-FABA PAG from Fig. 9 almost exactly overlays the“foot” observed in the decomposition range of the PAG.

Fig.18 shows the comparison between the equivalent molar acidloadings (wt% adjusted to achieve the samemolar concentration) of3M-Methide and Rhodorsil-FABA PAGs after UV exposure. PAG

decomposition using triflic acid was examined to compare theeffect of a strong acid which did not need to be UV exposed to formthe acid. A 1.0 wt% solution of triflic acid was and PPC was made byusing previously dried triflic acid and acetone as the solvent, asshown in Fig. 19. It was not possible to use g-butyrolactone as thesolvent because at the softbake temperature, 100 �C, the triflic acidcontaining film catalyzed the decomposition of the PPC polymer.The slightly higher temperature for weight loss with the triflic acidis consistent with the elevated weight loss temperature seen inFig. 15 for the acetone-solvent effect with Rhodorsil-FABA PAG.

The decomposition of PPC via chain unzipping into the monomermeans that the decomposition profile, temperatures, rates, andcalculated activation energies should be similar to that of the puremonomer. If the decomposition products are only the monomer, theTGA curves for PPC should match that of propylene carbonate,perhaps with a slight temperature lag due to the unzipping of themonomer. If, however, the polymer splits into lighter molecularweight fragments (i.e. acetone and carbon dioxide), thenweight lossshould occur more rapidly because these are more volatile products.To compare the monomer with polymer decomposition products,propylene carbonate was volatilized at heating rates of 0.5 �C/min,2 �C/min, and 10 �C/min. Propylene carbonate loaded with 3.75 wt%Rhodorsil-FABA and exposed to UV light was also studied at theseheating rates. The DTGA curves for these experiments are shown inFig. 19. The general shape is nearly identical at each heating rate,except for the foot on the PAG loaded liquid. This foot is present at thesame wt% as the initial PAG loading and exhibits a second decom-position region as the PAGdecomposes. Kinetic parameters extractedusing the CoatseRedfern method are summarized in Table 9.

4.9. Kinetic analysis by Kissinger method

The kinetic parameters were calculated using the Kissingermethod for all polymer blends evaluated in this study at multiple

Fig. 15. Novomer PPC with 3.75 wt% Rhodorsil-FABA PAG dissolved in different solvents: a) g-butyrolactone (UV exposed); b) anisole (UV exposed); c) acetone (UV exposed); d)trichloroethylene (UV exposed); e) acetone (unexposed); f) anisole (unexposed); g) trichloroethylene (unexposed); h) g-butyrolactone (unexposed).


heating rates. The results of these calculations are summarized inTable 10. Additionally, the kinetic parameters extracted for themonomer with PAG and without PAG are included. Without PAG,the QPac PPC had a higher activation energy (165 kJ/mol) than theNovomer materials (136 kJ/mol and 142 kJ/mol). The mixturescontaining QPac or Novomer with a high concentration of PAGbefore UV exposure had activation energies between 154 kJ/moland 166 kJ/mol. UV exposure of the blends with a high concen-tration of PAG had lower activation energies between 45 kJ/mol and71 kJ/mol. QPac showed a higher activation energy than theNovomer PPC. Low acid concentrations exhibited a mixed effect onthe activation energies for both exposed and unexposed films withno easily distinguished trend.

Films decomposed under nitrogen exhibited widely varyingbehavior depending on the acid type and concentration. Sampleswithout PAG and those which contained PAG but were not UVexposed demonstrated similar behavior. The PPC decomposed intoa clear liquid product which quickly evaporated at the decompo-sition temperature. This decomposition occurred at approximately

Table 7Calculated activation energies and pre-exponential factors for Novomer PPC thinfilms for different casting solvents loaded with 0.25 wt% Rhodorsil-FABA PAG andexposed to UV irradiation (1 J/cm2; l ¼ 248 nm) and unexposed samples. All valuescalculated using the CoatseRedfern method. Regression correlation factor (R2) andcompletion fraction at the point of maximum decomposition rate (amax) are alsolisted.

3.75% FABA Ea (kJ/mol) A (s�1) R2 amax

ExposedAcetone 45.5 4.66e-3 0.9502 0.82Anisole 75.0 6.53e2 0.9934 0.80g-butyrolactone 75.8 1.74e3 0.9898 0.84Trichloroethylenea 37.1 3.57e-2 0.9095 0.80UnexposedAcetone 411 3.80e40 0.9983 0.74Anisole 381 5.81e36 0.9891 0.80g-butyrolactone 659 1.37e73 0.9751 0.80Trichloroethylene 410 1.30e40 0.9685 0.74

a Poor regression correlation is due to the segmented nature of the curve shownin Fig. 15.

190 �C and appears to be consistent with depolymerization intoa liquid product, which is likely propylene carbonate.

Films loaded with higher concentrations of photoacid (3.0 wt%3M-Methide, 3.75 wt% Rhodorsil-FABA, and 10 wt% Rhodorsil-FABA) exhibited a somewhat similar decomposition mechanism asthe unloaded and unexposed samples. The onset of decompositionwas much lower (by about 100 �C) and the reaction occurred overa wide temperature range (from 100 �C to 140 �C). The filmsdecomposed into a liquid with a dark tint which subsequentlyvolatilized and left a dark-brown and/or black residue in the TGApans. These samples were also observed to form bubbles in the pan,which was not seen with PPC films in the absence of PAG.

Films loaded with a low concentration of PAG (0.20 wt% and0.25 wt% of 3M-Methide and Rhodorsil-FABA, respectively) showedvery different decomposition characteristics than the unloaded,unexposed, or high PAG concentration samples. At these lowconcentrations, the nucleation and growth of gas bubbles in thepolymerwere clearly visiblewhen thefilmwasUV irradiated. In thiscase, the entrapped bubbles continuously expanded after forma-tion, which caused the polymer to foam and expand in volume.

4.10. FTIR analysis

FTIR was used to examine the thin polymer films deposited onKBr. Fig. 20 shows the spectra of PPC films cast in different solventsafter solvent evaporation. Fig. 21 shows the PPC spectra for filmscast in g-butyrolactone in the presence of a small quantity ofRhodorsil-FABA. Additional spectra were gathered for UV exposedfilms and UV exposed films heated to 120 �C. However, thesespectra exactly overlay the unexposed spectra shown in Fig. 21 butwith a lower intensity due to the decreased film thickness. Whenthe peaks are normalized for intensity around the C]O peak at1750 cm�1, they accurately overlay each other and are therefore notshown because no significant information can be discerned fromthem. Similar results were experienced in attempts to track filmdecomposition using UVeVisible light spectroscopy. Completedecomposition of the film leaves very weak, nearly indistinguish-able peaks, as has previously been reported [19].

Fig. 16. Novomer PPC with 0.25 wt% Rhodorsil-FABA PAG dissolved in different solvents: a) acetone (UV exposed); b) methylene chloride (UV exposed); c) anisole (UV exposed); d)g-butyrolactone (UV exposed); e) acetone (unexposed); f) methylene chloride (unexposed); g) anisole (unexposed); h) g-butyrolactone (unexposed).


Films with and without PAG and in different solvents exhibitedvery similar spectral peaks, as would be expected since the residualsolvent is mostly removed and the PAG and only accounts for a fewweight percent of the mixture. The broad peak at ca. 3000 cm�1 isattributable to CeH bond stretching in the propylene units. Theremay also be an OH peak hidden within this peak. The peak at1760 cm�1 is attributed to the C]O of the carbonate groups. Thebroad peak at 1250 cm�1 is due to the CeO bond of the carbonategroups but also may be due to any irregularities (ether linkages) inthe backbone of the polymer. The peak at 1475 cm�1 is due to CeCbonding in the propylene groups. The additional peaks in the1350e1450 cm�1 range are due to scissoring and bending of theCeH bonds.

As seen in Fig. 20, the FTIR spectra of solvent-cast PPC films donot distinguish the essential differences between various solventsused to cast the films. The CeH stretching (2850e2960 cm�1), andbending and scissoring (1350e1470 cm�1) modes exhibit slightvariations between solvents. Very slight differences are seen at1575 cm�1, which may be attributed to carboxylate groups presentat the ends of the polymer chains.

Table 8Calculated activation energies and pre-exponential factors for Novomer PPC thinfilms for different casting solvents loaded with 0.25 wt% Rhodorsil-FABA PAG andexposed to UV irradiation (1 J/cm2; l ¼ 248 nm) and unexposed samples. All valuescalculated using the CoatseRedfern method. Regression correlation factor (R2) andcompletion fraction at the point of maximum decomposition rate (amax) are alsolisted.

0.25% FABA Ea (kJ/mol) A (s�1) R2 amax

ExposedAcetone 63.4 3.79e0 0.9655 0.72Anisole 57.5 2.82e-1 0.9883 0.60g-butyrolactone 28.3 3.80e-1 0.9574 0.84Methylene chloride 48.8 8.42e-3 0.9919 0.61UnexposedAcetone 334 1.46e30 0.9935 0.87Anisole 353 1.44e32 0.9957 0.86g-butyrolactone 263 2.17e21 0.9781 0.48Methylene chloride 282 1.87e24 0.9946 0.87

The influence of the Rhodorsil-FABA on the FTIR spectra isshown in Fig. 21. Novomer PPC was dissolved in g-butyrolactoneand cast on KBr slides from solutions containing different weightpercentages of Rhodorsil-FABA PAG. Additionally, the PAG alonewas dissolved in g-butyrolactone and cast on KBr. The primary peakin the Rhodorsil-FABA spectrum is seen at 1470 cm�1, CeC bonding.Secondary peaks are seen at 975 cm�1 from the methyl groups andat 1514 cm�1 from benzene ring stretching. The peaks at 1470 cm�1

and 975 cm�1 are obscured by PPC peaks. The benzene ring peak at1514 cm�1 is with higher PAG loadings and is the only cleardifference between the spectra. No evidence of hydrogen bondingas observed in other PPC studies at 1550 cm�1 was observed here.

5. Discussion

5.1. Enthalpy of vaporization by Kissinger method

The enthalpy of vaporization of propylene carbonate has beenreported to be between 49 kJ/mol [64] and 51 kJ/mol [65]. Thepropylene carbonate data analyzed using the Kissinger methodyielded an activation energy in the same range as the as theseliterature values. The similar activation energies for the monomerwith no additives (50.5 kJ/mol) and with 3.75% Rhodorsil-FABA PAG(48.8 kJ/mol) indicate that the photoacid does not decompose themonomer into smaller fragment products and has little effect onthe overall vapor pressure.

The activation energies calculated by the Kissinger method forthe Novomer PPC with UV exposed photoacid at high concentra-tions (3.0% 3M-Methide or 3.75% Rhodorsil-FABA) are very close tothat of the heat of vaporization of the cyclic monomer. This suggeststhe formation of propylene carbonate via the chain unzippingmechanism and subsequent volatilization. The chain unzippingmechanism has been reported to be due to the formation ofterminaleOH groups upon acid activation [19] and is the dominantdecomposition mechanism in high molecular weight PPC [15].Because the Novomer polymer had both higher molecular weightand higher regularity of repeat units (HT), the lower activation

Fig. 17. Exposed and unexposed Novomer with Rhodorsil-FABA at different concentrations. Shown in (a) and (d) are 3.75 wt%. Shown in (b) and (c) are 10.0 wt%. Shown in (e) and (f)are 0.25 wt%.


energy of the Novomer material was consistent with chain unzip-ping into the monomer.

5.2. Photoacid consumption by impurities

For the QPac material, fragmentation into larger molecularweight products likely competed with chain unzipping. Polyethergroups present in the QPac PPC prevented acid regeneration byforming an alcohol termination unable to propagate the unzippingmechanism. The quenching of the unzipping mechanism by theether groups is likely the reason for the higher activation energies

Fig. 18. Exposed Novomer PPC with (a) 3.0 wt% 3M-Methide and (b) 3.75 wt% Rh

calculated using kinetic analysis. This is particularly true at lowconcentration of PAG where acid regeneration is critical becausethe number of acid molecules is less than the number of polymerchains. The higher molecular weight decomposition products thatresult from the ether impurities may also contribute to the higheractivation energy.

5.3. Additional decomposition products

Deviations from the enthalpy of the monomer may also beattributed to the formation of hydrolysis products. 1,2-propanediol

odorsil-FABA and (c) an unexposed, air-dried film with a 1.0 wt% triflic acid.

Fig. 19. Dynamic TGAs of cyclic propylene carbonate at heating rates of (a) and (b) 0.5 �C/min; (c) and (d) 2 �C/min; and (e) and (f) 10 �C/min. Curves a, c, and e are the monomeronly. Curves b, d, and f contain 3.75 wt% Rhodorsil-FABA photoacid generator (PAG) and were exposed to 1 J/cm2 of UV irradiation (l ¼ 248 nm).

Table 10Kinetic parameters calculated by Kissinger method for all solutions analyzed atmultiple heating rates in this study. All polymer and polymer/PAG solutions weredissolved in g-butyrolactone, cast as a film on a glass slide, and heated to 100 �C toevaporate solvent. All data collected at heating rates of 0.5 �C/min, 2 �C/min, and10 �C/min.


has been observed in PPC pyrolysates via FTIR [15] and may form inthe presence of water, although most of the water was likelyremoved during film drying on a hotplate. It is possible that somewater remained bound within the polymer matrix and may havecontributed to formation of minor products including 1,2-pro-panediol. Activation energies calculated using the techniques citedin this paper cannot distinguish between the cyclic monomer andthe alcohol product because 1,2-propanediol has an enthalpy ofvaporization close to that of propylene carbonate [66].

Decomposition of the monomer into carbon dioxide andacetone with additional heating has been reported, but this doesnot occur until higher temperatures than evaluated in this study.These products may result from decomposition in an oxygenatedatmosphere or at extremely fast ramp rates, but appear to be onlyminor products in this study. The TGA curves shown in Fig. 19suggest that the addition of photoacid to cyclic propylenecarbonate does not dimerize or degrade the monomer. Thus, it islikely that the PAG has little or no impact after degradation of thepolymer into cyclic propylene carbonate.

Ea (kJ/mol) A (s�1) R2

QPac 165 1.19e16 0.9956QPac þ 0.20% Methide 121 1.81e10 0.9935QPac þ 0.25% FABA 130 1.12e11 0.9956QPac þ 3.0% Methide 157 1.34e15 0.9922QPac þ 3.75% FABA 158 2.51e15 0.9999

5.4. Residual solvent influence on decomposition

The exact influence of residual solvent on the decompositionbehavior is very subtle. Some of the solvents evaluated in this study

Table 9Kinetic parameters calculated by CoatseRedfern method of propylene carbonatewithout PAG and propylene carbonate with 3.75 wt% Rhodorsil-FABA PAG exposedto 1 J/cm2 of UV radiation at l ¼ 248 nm.


Propylene carbonate (No PAG)0.5 �C/min 59.82 4.19 0.9948 0.872 �C/min 59.52 8.75 0.9984 0.8810 �C/min 50.02 2.18 0.9987 0.88Propylene carbonate (3.75 wt% FABA þ UV exposure)0.5 �C/min 60.90 5.55 0.9920 0.882 �C/min 59.71 8.95 0.9983 0.8910 �C/min 50.09 2.20 0.9987 0.88

are capable of hydrogen bonding and thus stabilizing the PPCbackbone, but no evidence of this is observed in the FTIR spectra.Additionally, the solvents evaluated in this study, with the notableexception of g-butyrolactone, are not capable of undergoing poly-merization. Although g-butyrolactone has been copolymerizedwith PPC [47] and may undergo homopolymerization [67], it isdifficult to polymerize except under controlled conditions [68],despite high ring strain [69]. No change in the viscosity of g-butyrolactone was observed when photoacid was added to thesolvent and UV exposed in the absence of PPC. Deviations inthermal stability due to residual solvent or interaction will dependhighly on molecular weight, polydispersity, and film dryingconditions and temperatures.

QPac þ 0.20% Methide þ 1J/cm�2 UV 134 1.02e12 0.9944QPac þ 0.25% FABA þ 1J/cm�2 UV 142 1.79e12 0.9938QPac þ 3.0% Methide þ 1J/cm�2 UV 58.6 3.71e4 0.9992QPac þ 3.75% FABA þ 1J/cm�2 UV 71.1 3.32e2 0.9930Novomer (MW ¼ 90 kg/mol) 136 2.39e12 0.9377Novomer (MW ¼ 100 kg/mol) 142 8.96e12 0.9983Novomer þ 0.20% Methide 134 2.49e11 0.9999Novomer þ 0.25% FABA 165 9.92e14 0.9972Novomer þ 3.0% Methide 166 5.81e15 0.9981Novomer þ 3.75% FABA 154 3.66e14 0.9989Novomer þ 0.20% Methide þ 1J/cm�2 UV 184 1.63e16 0.9487Novomer þ 0.25% FABA þ 1J/cm�2 UV 150 1.34e13 1.000Novomer þ 3.0% Methide þ 1J/cm�2 UV 56.0 3.15e4 0.9383Novomer þ 3.75% FABA þ 1J/cm�2 UV 45.3 4.48e4 0.9986Propylene carbonate 50.3 1.89e3 0.9986Propylene carbonateþ3.75% FABA þ UV 48.8 1.25e3 0.9981

Fig. 20. FTIR absorbance spectra for Novomer PPC dissolved in g-butyrolactone, trichloroethylene, methylene chloride, and acetone.

Fig. 21. FTIR absorbance spectra for Novomer PPC dissolved in g-butyrolactone with various weight percentage Rhodorsil-FABA PAG.


5.5. Confounding effects on kinetic analysis

Without adequate photoacid loading (i.e. low weight percent-ages of PAG), decomposition proceeds slowly via the unzippingmechanism before proceeding via random chain scission at highertemperatures and subsequent unzipping. This results ina segmented and complex degradation curve as seen in Figs. 13 and14. The irregularity in the decomposition curves at these acidloadings levels is also likely the result of bubble entrapment ofdecomposition products during heating. Such effects are notaccounted for in the kinetic models. TGA data and results of theopen-face decomposition observed under glass both indicatecomplex, multiphase behavior of the decomposition mechanism inthe presence of PAG. Intumescent flame retardants (IFRs) aremulticomponent systems where an acidic source chars a carbon-ization agent to form a low thermal conductivity layer whilea blowing agent is converted to a gas and expands with heat.Consequently heat transfer into the sample andmass transfer out ofthe polymer are inhibited. The photoacid in our system coupledwith the polymer backbone fragments potentially provide acid andcarbonization agent, while the decomposition products act asa blowing agent. Chemical foaming of PPC has been reported toraise the thermal stability as measured by TGA [41,42]. The foamingeffect observed for the UV exposed acid samples decomposedunder glass are consistent with this expectation. Under sufficientacid loading, the polymer matrix decomposes to the liquid productand does not allow bubble entrapment. At low acid concentrations,the decomposition of the polymer matrix is incomplete allowingliquid decomposition products to convert into vapor which foamsthe polymer. These effects are responsible for the elevated activa-tion energies at low acid loading as well as the complex andsegmented degradation curves at these acid concentrations.

Differences in the calculated activation energies may be influ-enced by numerous other factors not previously discussed.Although every effort was made to mix a homogenous blend ofsolid polymer with photoacid generators, the local concentration ofphotoacid generator may have been greater than or less than thenominal concentration of the blend. The oxygen content in the TGAatmosphere may also have impacted the decomposition behavior.Although the TGAwas purged with nitrogen for 1 h before heating,reducing oxygen content to below 100 ppm, some experiments at10 �C/min exhibited characteristics consistent with an oxygenatedenvironment (i.e. rapid decomposition with the absence of the“foot” observed in nitrogen environments).

6. Conclusions

The acid catalyzed decomposition of PPC significantly reducesthe activation energy as calculated by CoatseRedfern and Kissingermethods. The activation energies calculated using these techniquesare similar to those calculated for cyclic propylene carbonate whensufficient quantities of acid are present. In these cases, thedecomposition appears to occur via chain unzipping into cyclicpropylene carbonate. Under insufficient acid loading, decomposi-tion occurs via a composite decomposition mechanism where partof the polymer decomposes at low temperature and the remainingportion decomposes at higher temperatures.

Polymer purity and molecular weight impact the observedactivation energy. Higher molecular weight products of the alter-nating copolymer most likely decompose via the chain unzippingmechanism. The high molecular weight, high-purity polymer inthis study (Novomer) decomposed with UV exposed PAG and hadactivation energies close to that of cyclic propylene carbonate. Thesimilar activation energies suggest the acid generated upon UV


exposure is sufficient to initiate chain unzipping, which is nearlyidentical to the reported values of the heat of vaporization. Thissuggests that upon acid activation, chain unzipping occurs at orbelow the temperature where cyclic propylene carbonate exhibitssignificant vapor pressure. Chain unzipping may occur at roomtemperature after acid activation but is not observed in thermog-ravimetric analysis because of the negligible vapor pressure ofpropylene carbonate at room temperature. However, this isunlikely (or unzipping proceeds very slowly) as no liquid product isobserved upon UV exposure without heating.

Residual solvent clearly impacts the decomposition, especially inthe presence of UV activated PAG. The amount of solvent remainingin the polymer matrix was not measured and likely varied betweensamples and casting solvents. Because of the uncertainty regardingthe solvent contentwithin the solvent-cast films, it is difficult to saywith certainty the role the solvent plays in increasing thermalstability. Other effects such as free volume/plasticization andabsorbed energy in the form of heat of vaporization may alsocontribute to the observed thermal stability.

Acknowledgements

The authors acknowledge the contributions of Kristie DeLisoand Annapoorani Sundaramoorthi who prepared polymer solu-tions and processed some films for TGA.

References

[1] Inoue S, Koinuma H, Tsuruta T. J Polym Sci B Polym Phys 1969;7:287e92.[2] Thorat SD, Phillips PJ, Semenov V, Gakh A. J Appl Polym Sci 2003;89

(5):1163e76.[3] Zhu Q, Meng YZ, Tjong SC, Zhao XS, Chen YL. Polym Int 2002;51:1079e85.[4] Tao Y, Wang X, Chen X, Zhao X, Wang F. J Polym Sci A Polym Chem

2008;46:4451e8.[5] Peng S, An Y, Chen C, Fei B, Zhuang Y, Dong L. Polym Degrad Stab

2003;80:141e7.[6] Du L, Qu B, Meng Y, Zhu Q. Compos Sci Technol 2006;66:913e8.[7] Gao LJ, Du FG, Xiao M, Wang SJ, Meng YZ. J Appl Polym Sci 2008;108

(6):3626e31.[8] Gao LJ, Xiao M, Wang SJ, Meng YZ. J Polym Sci B Polym Phys 2008;108

(2):1037e43.[9] Li XH, Tjong SC, Meng YZ, Zhu Q. J Polym Sci B Polym Phys 2003;41:1806e13.

[10] Song PF, Xiao M, Du FG, Wang SJ, Gan LQ, Liu GQ, et al. J Appl Polym Sci2008;109:4121e9.

[11] Yu T, Yong Z, Liu K, Zhao Y, Chen E, Wang F, et al. Polym Degrad Stab 2009;94(2):253e8.

[12] Yu T, Zhou Y, Zhao Y, Liu K, Chen E, Wang DY, et al. Macromolecules2008;41:3175e80.

[13] Zhang Z, Lee J-H, Lee S-H, Heo S-B, Pittman Jr CU. Polymer 2008;49:2947e56.[14] Zhang Z, Mo Z, Zhang H, Zhang Y, Na T, An Y, et al. J Polym Sci B Polym Phys

2002;40:1957e64.[15] Li XH, Meng YZ, Zhu Q, Tjong SC. Polym Degrad Stab 2003;81:157e65.[16] Du LC, Meng YZ, Wang SJ, Tjong SC. J Appl Polym Sci 2004;92:1840e6.[17] Yan H, Cannon WR, Shanefield D. Ceram Int 1998;24:433e9.[18] Cupta MG, Joseph PJ, Kohl PA. J Appl Polym Sci 2007;105(5):2655e62.[19] Jayachandran PJ, Reed HA, Zhen H, Rhodes LF, Henderson CL, Bidstrup

Allen SA, et al. J Microelectromech Syst 2003;12(2):147e59.[20] Reed HA, White CE, Rao V, Bidstrup Allen SA, Henderson CL, Kohl PA.

J Micromech Microeng 2001;11:733e7.

[21] White CE, Balogun A, Henderson CL. J Appl Polym Sci 2006;102:266e71.[22] Dang B, Joseph P, Bakir M, Spencer T, Kohl P, Meindl J. San Fransisco, CA;

2005.[23] Metz S, Jiguet S, Bertsch A, Renaud P. Lab Chip 2004;4:114e20.[24] Spencer TJ, Joseph PJ, Kim TH, Swaminathan M, Kohl PA. IEEE Trans Microw

Theory Tech 2007;55(9):1919e25.[25] Joseph PJ, Monajemi P, Ayazi F, Kohl PA. IEEE Trans Adv Packag 2007;30

(1):19e26.[26] Lu JY, Yu CP, Chang HC, Chen HW, Li YT, Pan CL, et al. Appl Phys Lett 2008;92

(064105):1e3.[27] Hassani A, Dupuis A, Skorobogatiy M. Appl Phys Lett 2008;92(071101):1e3.[28] Yan H, Cannon WR, Shanefield DJ. J Am Ceram Soc 1993;76:166e72.[29] Wedlake M, Kohl PA. J Mater Res 2002;17(3):632e40.[30] Wu X, Reed HA, Wang Y, Rhodes LF, Elce E, Ravikiran R, et al. J Electrochem

Soc 2003;150(9):H205e13.[31] Jones III CA, DeSimone JM. John Wiley & Sons, Inc.; 2002. pp. 111e126.[32] Liu B, Zhao X, Wang X, Wang F. J Appl Polym Sci 2003;90:947e53.[33] Chen S, Xiao Z, Ma M. J Appl Polym Sci 2008;107:3871e7.[34] Tao Y, Wang X, Zhao X, Li J, Wang F. J Polym Sci A Polym Chem

2006;44:5329e36.[35] Qin Y, Wang X, Zhang S, Zhao X, Wang F. J Polym Sci A Polym Chem

2008;46:5959e67.[36] Darensbourg DJ, Lewis SJ, Rodgers JL, Yarbrough JC. Inorg Chem 2003;42

(2):581e9.[37] Meng YZ, Du LC, Tiong SC, Zhu Q, Hay AS. J Polym Sci A Polym Chem

2002;40:3579e91.[38] Li B, Wu GP, Ren WM, Wang YM, Rao DY, Lu XB. J Polym Sci A Polym Chem

2008;46:6102e13.[39] Lu L, Huang K. Polym Int 2005;54:870e4.[40] Aikawa S, Yoshida Y, Nishiyama S, Noguchi H, Shoji A. Mol Cryst Liq Cryst

2006;445:315[606]e321[611].[41] Guan LT, Du FG, Wang GZ, Chen YK, Xiao M, Wang SJ, et al. J Polym Res

2007;14:245e51.[42] Guan LT, Xiao M, Meng YZ. Polym Eng Sci; 2006:153e9.[43] Liu Q, Zou Y, Bei Y, Qi G, Meng Y. Mater Lett 2008;62:3294e6.[44] Luinstra GA. Polym Rev 2008;48:192e219.[45] Luinstra GA, Molnar F. Macromol Symp 2007;259:203e9.[46] Shi X, Gan Z. Eur Polym J 2007;43:4852e8.[47] Yang DZ, Hu P. J Appl Polym Sci 2008;109:1635e42.[48] Huang Y, Wang J, Liao B, Chen M, Cong G. J Appl Polym Sci 1998;64

(12):2457e65.[49] Ma X, Chang PR, Yu J, Wang N. Carbohydr Polym 2008;71:229e34.[50] Jiao J, Xiao M, Shu D, Li L, Meng YZ. J Appl Polym Sci 2006;102:5240e7.[51] Pang MZ, Qiao JJ, Wang SJ, Xiao M, Meng YZ. J Appl Polym Sci

2008;107:2854e60.[52] Ning W, Zhang X, Yu J, Jianming F. Polym Int 2008;57:1027e35.[53] Zhang Z, Zhang H, Zhang Q, Zhou Q, Zhang H, Mo Z, et al. J Appl Polym Sci

2006;100:584e92.[54] Luinstra GA, Haas GR, Molnar F, Bernhart V, Eberhart R, Rieger B. Chem Eur J

2005;11(21):6298e314.[55] Joshi SS, Mebel AM. Polymer 2008;48(13):3893e901.[56] Zhang Z, Shi Q, Peng J, Song J, Chen Q, Yang J, et al. Polymer

2006;47:8548e55.[57] Lu L, Liu S, Huang K. J Appl Polym Sci 2008;107:872e80.[58] Bhusari D, Reed HA, Wedlake M, Padovani AM, Bidstrup Allen SA, Kohl PA.

J Microelectromech Syst 2001;10(3):400e8.[59] Cupta M. Georgia Tech; 2005. p. 137.[60] Chen Y, Wang Q. Polym Degrad Stab 2007;92:280e91.[61] Ozawa T. J Therm Anal 1970;2:301e24.[62] Ozawa T. Bull Chem Soc Jpn 1965;38:1881e6.[63] Flynn JH, Wall LA. J Polym Sci B Polym Lett 1966;4(5):323e8.[64] Nasirzadeh K, Neueder R, Kunz W. J Chem Eng Data 2005;50:26e8.[65] Chernyak Y, Clements JH. J Chem Eng Data 2004;49:1180e4.[66] Verevkin SP. Fluid Phase Equilib 2004;224:23e9.[67] Moore T, Adhikari R, Gunatillake P. Biomaterials 2005;26:3771e82.[68] van der Mee L, Helmich F, de Bruijn R, Vekemans JAJM, Palmans ARA,

Meijer EW. Macromolecules 2006;39:5021e7.[69] Houk KN, Jabbari A, Hall Jr HK, Aleman C. J Org Chem 2008;73(7):2674e8.