lable at ScienceDirect

Polymer Degradation and Stability 125 (2016) 129e139

Contents lists avai

Polymer Degradation and Stability

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

Thermal decomposition of poly(propylene carbonate): End-capping,additives, and solvent effects

Oluwadamilola Phillips, Jared M. Schwartz, Paul A. Kohl*

School of Chemical and Biomolecular Engineering, Georgia Institute of Technology, Atlanta, GA, 30332-0100, USA

a r t i c l e i n f o

Article history:Received 2 October 2015Received in revised form11 January 2016Accepted 12 January 2016Available online 14 January 2016

Keywords:Polymer degradationPoly(propylene carbonate)Sacrificial polymerPPCPPC polyol

* Corresponding author.E-mail address: kohl@gatech.edu (P.A. Kohl).

http://dx.doi.org/10.1016/j.polymdegradstab.2016.01.00141-3910/© 2016 Elsevier Ltd. All rights reserved.

a b s t r a c t

Two routes to the thermal decomposition of poly(propylene carbonate) (PPC) have been considered-polymer chain-end unzipping and random chain scission. The inhibition and catalysis of each mecha-nism has been studied. End-capping of low molecular weight (2 kDa) PPC has been achieved andconfirmed with NMR analysis resulting in stabilization from unzipping. Citric acid was used as an ad-ditive and shown to stabilize the PPC backbone from random chain scission. The combination of bothstabilizing agents (i.e. unzipping and ransom chain scission) results in the highest thermal stability.Photo-acid and base generators were incorporated into films and used to catalyze decompositionpathways of PPC. Residual solvent effects in PPC, such as methanol used in its purification, have beenshown to affect the thermal stability through interactions with the backbone of PPC. This study dem-onstrates a widened thermal stability window for PPC, such as when used as a sacrificial material.

© 2016 Elsevier Ltd. All rights reserved.

1. Introduction

Poly(propylene carbonate) (PPC) is an amorphous, aliphaticpolymer synthesized by the copolymerization of propylene oxideand carbon dioxide with an organometallic catalyst, typically zincglutarate [1e7]. The polymer is attractive because it is made by thefixation of CO2, and it is biodegradable and biocompatibile [8]. Thelow glass transition temperature makes PPC suitable for low-temperature packaging applications. The glass transition temper-ature of PPC can range from 25 �C to 45 �C depending on molecularweight and regioregularity of the polymer. Molecular weightsranging from 15 kDa to 200 kDa have previously been produced incommercial quantities. Recently lower molecular weight of 2 kDaPPC Polyol has been synthesized [9].

PPC has a relatively low onset temperature for decomposition,widely reported in the range from 180 �C to 240 �C depending onheating rates [5,10e13]. The low decomposition temperaturemakes it ideal for specific sacrificial polymer applications in mi-croelectronics, especially those involving solder, epoxy, andmovable components such as microelectromechanical systems(MEMS). The movable MEMS component can be encapsulated by asacrificial polymer and a photo-definable dielectric as the overcoat

04

in order to protect the movable component during processing.Later, a thermal treatment can be used to decompose the PPC andvolatize the products, which diffuse through the overcoat, freeingthe movable component in an air-cavity [14e17].

A critical step in the fabrication process is controlling the ther-mal decomposition of PPC, so that the MEMS device is released atthe desired point in the fabrication sequence. The thermal behaviorof PPC has been studied; however, conflicting reports have led to aninadequate understanding of the thermal degradation process forPPC. A better understanding of the PPC decomposition process canlead to modification and control of the rate and temperature ofpolymer decomposition making for a better match of the sacrificialpolymer to the overcoat material. For example, benzoyclobutene(BCB) has been used by Uzunlar et Al. as an overcoat material in theair-cavity process [17]. A thermal treatment of the polymer/BCBstructure at 190 �C is done to cure the BCB and make it a me-chanically stable overcoat before significant decomposition of PPCoccurs. Afterwards, the temperature is increased to 240 �C toquickly decompose PPC. At 190 �C BCB needs 1.3 h to cure to 70%cross-linking density. A modest increase of 20 �C in the sacrificialpolymer decomposition temperature can lower the curing time ofBCB by a factor of 20 to just 4 min.

Desirable thermal properties of the sacrificial polymer that areneeded for the process include: (i) thermal stability of the polymerduring overcoat curing, (ii) rapid speed of decomposition at thedegradation temperature, and (iii) minimal residue after

O. Phillips et al. / Polymer Degradation and Stability 125 (2016) 129e139130

decomposition. In this paper, the mechanisms for PPC decompo-sition have been studied and the means for altering the PPCdecomposition temperature investigated. The results are comparedto previous studies to shed light on the conflicting results.

2. Background

Modifying the thermal properties requires a fundamental un-derstanding of PPC degradation pathways. The thermal decompo-sition mechanisms of PPC have been previously proposed[2,8,18,19]. The polymer can be modified and additives can beadded to the PPC polymer to inhibit or facilitate the decomposition.The thermal decomposition occurs via two mechanisms: (i) poly-mer unzipping (i.e. decomposition from the ends of the polymerchain with the reaction proceeding inward), and (ii) random chainscission (i.e. internal repeat units of the polymer randomly undergoreaction), as shown in Fig. 1a and b respectively.

In the unzipping reaction, chain ends become “active” viathermal energy and initiate a reaction with specific carbon sitesacross the polymer backbone, leading to degradation. Cyclic pro-pylene carbonate is the typical product that is formed from thisdegradation mechanism [7,8,20]. It has been widely understoodthat the unzipping reaction proceeds by either an alkoxide or car-bonate backbiting reaction, depending on how the polymer chain is

Fig. 1. a Chain unzipping reaction in PPC occurring via an alkoxide or carbonate backbiting pin PPC via thermally induced cleavage of CeO bonds creating carbon dioxide as one of the

terminated (i.e. terminated in a hydroxyl propyl or carbonatemoiety). In alkoxide backbiting, a carboxylate nucleophile attacks acarbonyl carbon atom. In carbonate backbiting, the weaker nucle-ophile of the alcohol end-group attacks an electrophilic carbonatom from the polymer backbone. It is believed that carbonatebackbiting is the most common route due to its lower activationenergy [4]. Reactive hydroxyl chain ends can be rendered “inactive”in the unzipping reaction by terminating the polymer chain (i.e.“end-capping” the polymer chain) with a less reactive moiety.Dixon et al. first showed the end-capping of PPC with a variety ofelectrophilic organic compounds [21]. The end-cap kineticallystabilizes the polymer ends so that chain unzipping is inhibited.Although reports of PPC end-capping have appeared in the litera-ture [10,22,23], characterization of end-capped PPC has not beenwell-studied because the density of polymer ends is small for highmolecular weight polymers. Analysis has been mostly limited todynamic thermogravimetric analysis (TGA). In this study, lowermolecular weight PPCwith a higher density of end-groups has beenused to characterize the end-groups.

Chain scission is the second thermal decomposition pathway forPPC. It has higher activation energy than chain unzipping and is notvery sensitive to the type of polymer end-group. Chain scissionoccurs through the random, thermally-induced cleavage of CeObonds within the backbone of the polymer, generally creating

athway creating the cyclic propylene carbonate as the product. b Chain scission reactionproducts.

Fig. 2. Photo-acid generator (PAG) structure of 4-methylphenyl[4-(1-methylethyl)phenyl] tetrakis(pentafluorophenyl) borate (Rhodorsil-Faba) with a molecular weightof 1016 g/mol. Photo-base generator (PBG) structure of 1,5,7 e triaza-bicyclo [4.4.0]dec-5-ene tetraphenylborate (TBD�HBPh4) with a molecular weight of 459 g/mol.

O. Phillips et al. / Polymer Degradation and Stability 125 (2016) 129e139 131

carbon dioxide and acetone as products. Organic additives havebeen used in previous studies to stabilize the polymer to randomchain scission by creating complex interactions between thecarbonyl moieties of the backbone [24e26]. Stearic acid has beenused to improve the thermal stability at loadings between 0.28 wt%and 2.17 wt% [27]. The reports suggest that hydrogen bonding wasthe interaction between carboxylic acid and carbonyl moieties ofthe polymer. Many of the additives have a higher decompositiontemperature (300 �Ce400 �C) than PPC which creates substantialresiduewhen the polymer decomposes and vaporizes. Additionally,stearic acid is a mono-functional carboxylic acid that has only onesite available for interacting with the backbone of the polymer. Inthis study, a biodegradable, tri-functional carboxylic acid with alow decomposition temperature of 175 �C has been used to stabilizePPC from chain scission while leaving minimal residue.

Thus, end-capping (to inhibit unzipping) and organic additives(to inhibit random chain scission) are two approaches to inhibitingPPC decomposition. In addition to inhibiting each of the decom-position mechanisms, it has been shown that each mechanism canbe catalyzed so as to occur more easily at a temperature below thatof the pure polymer. It was previously shown [11,28e30] thatintroducing a photo or thermal acid generator, such as an iodoniumsalt, into the PPC mixture can catalyze the degradation reactions.Upon exposure to UV light, these iodonium salts undergo aphotolysis reaction releasing a strong protonic acid that attacks thepolymer in a “chain scission” manner [11]. Incorporation of aphotoacid generator (PAG) into the PPC can lower the photo-catalyzed PPC decomposition to 100 �C. PAGs can also be ther-mally activated. If the PAG thermally decomposes at temperatureslower than that of the PPC, it has the effect of lowering the overallPPC decomposition temperature. It has also been shown thatintroducing a photo or thermal base generator, such as an aminesalt, can catalyze degradation reactions [31]. From exposure to UVlight, the salt releases a strong amine base that performs a nucle-ophilic attack on the polymer. The method of degradation is not aswidely studied, but it has been suggested that the PBG acts pri-marily in a “backbiting” manner catalyzing the unzipping mecha-nism of PPC [32]. This paper also reports acid-catalyzed and base-catalyzed degradation of end-capped PPC for the first time.

3. Experimental

3.1. Materials

Polypropylene carbonate (PPC) was generously supplied byNovomer Inc. The weight-average molecular weight of PPC mate-rials provided was 137 kDa, 160 kDa, 219 kDa, and 263 kDa. 137 kDaPPC has a polydispersity index (PDI) of 1.16. PPC polyol was alsosupplied by Novomer Inc. at a weight-average molecular weight of2 kDa with a PDI of 1.1. End-capping reagents of nitrophenylchloroformate, vinyl chloroformate, and benzoyl chloride werecommercially purchased from SigmaeAldrich and used as is. Sol-vents such as tetrahydrofuran (THF) and methanol were all pur-chased from BDH, at purity levels > 99%. The PAG investigated inthis study was an iodonium salt, 4-methylphenyl [4-(1-methylethyl) phenyl] tetrakis(pentafluorophenyl) borate (referredto as Rhodorsil-Faba) from Solvay Inc. Photo-base generator (PBG)used in this study is an amine salt 1,5,7 - triaza-bicyclo [4.4.0]dec-5-ene tetraphenylborate (referred to as TBD�HBPh4). PAG and PBGadditives are shown in Fig. 2. Synthesis of PBG used in this studywas made following the Sun et al. procedure [31].

3.2. Characterization

TGAwas performed by using a TGA Q50 from TA instruments to

investigate the thermal decomposition of PPC materials. The plat-inum pan was re-zeroed after each run for accurate measurementsof residue. Samples between 3 and 10mgwere loaded onto the pan.Nitrogen and air environments were used at a flow rate of 40 ml/min for the sample and the balance. PPC films were removed fromsilicon substrates with a razor blade and loaded onto the pan. Dy-namic TGA samples were heated at rates of 1 �C/min and 5 �C/minfrom room temperature to 400 �C. Isothermal TGA experimentswere ramped at heating rate of 10 �C/min to the temperature ofinterest and held at that temperature for two hours.

Nuclear magnetic resonance (NMR) was used to determine thechemical structure of PPC before and after end-capping reactions.NMR measurements were performed using a Varian Mercury Vx400 (400 MHz) tool. Chloroform-D (CDCl3) was used for the NMRsolvent and was supplied by SigmaeAldrich at a 99.8% purity level.The concentration of PPC in NMR sample tubes was held at 46mg in0.75 ml CDCl3. 13C NMR spectra was collected at 1204 scans with arelaxation time of 5s. 1H NMR spectra were collected at 16 scanswith a relaxation time of 1s. The spectra in 1H and 13C NMR werecalibrated and referenced to the chemical shift of the solvent CDCl3,which were 7.26 ppm and 77.16 ppm, respectively.

3.3. Preparation

End-capping reactions with 137 kDa PPC were performed in a20 ml scintillation vial under magnetic stirring. The scintillationvial contained 10 wt% of 137 kDa PPC in THF solution. Equimolarquantities of pyridine to end-capping reagents were also added tothe reaction vessel dropwise. End-capping reagents added to thereaction vessel were dependent on reactivity, and the number ofend-groups for the specified polymer. For example, 1 g of 137 kDaPPC has an estimated amount of 1.46E-05 mol of eOH endgroups.Benzoyl chloride was added at 700 times the molar amount of end-groups of PPC to drive the reaction to the end-capped PPC products.Thus, 0.0102 mol of benzoyl chloride was used. Chloroformates are

Fig. 3. “Neat” Novomer PPC with molecular weights of 2 kDa, 137 kDa, 160 kDa,219 kDa, and 263 kDa at a ramp rate of 1 �C/min in N2 atmosphere.

Fig. 4. Solvent-cast Novomer PPC in GBL with molecular weights of 2 kDa, 137 kDa,160 kDa, 219 kDa, and 263 kDa at a ramp rate of 1 �C/min in N2 atmosphere.

O. Phillips et al. / Polymer Degradation and Stability 125 (2016) 129e139132

generally more reactive; hence, vinyl chloroformate was added at50 times the amount of end-groups of 137 kDa PPC in which0.000730 mol was added to the reaction vessel. Magnetic stirringtimes were held for 48 h to ensure that all ends have been reacted.Pyridinium-chloride precipitation is an indicator that the reactionis underway, because the produced white salt is insoluble in THF.The PPC was precipitated in methanol from the reaction solution at10 times the volume of THF reaction solvent to ensure impuritieshave been removed.

End-capping reactions with 2 kDa PPC were held to at least16 wt% in THF solution to ensure the polymer was at a concentra-tion large enough to precipitate from solution in methanol. End-capping reagents must be added at lower amounts as to notexceed the solubility limit of reaction solvent because of the greaternumber of end-groups. The greater fraction of end-groups in-creases reactivity, thus, less amounts of end-capping reagents wereused. For example, 1 g of 2 kDa PPC has an estimated amount of0.001 mol of eOH end-groups. Vinyl chloroformate was added at 5times the number of end-groups of PPC to drive the reaction to theend-capped PPC products. Thus, 0.005 mol of end-capping reagentwas added to the reaction vessel. Magnetic stirring was held atequally long times for 48 h and subsequent precipitation of thepolymer from methanol.

“Neat” PPC refers to polymers that were used as-received fromthe manufacturer. “AirDry-MeOH” samples are polymers that havebeen precipitated from THF in methanol. The sample was left to dryin a fume hood but contains residual methanol. “VaccDry” samplesare polymers that have been precipitated from THF in methanol.However, the residual methanol has been essentially removed byvacuum drying at 100�C for 12 h “Solvent-cast” samples are for-mulations with 20% weight PPC in GBL. These formulations werecast onto a silicon substrate and soft-baked at 100�C for 3 min. Inthis study, additives were used to catalyze or inhibit the decom-position of PPC. The PPC material was made photosensitive by theaddition of PAG (Rhodorsil-Faba) at a loading of 3 mass parts PAGper 100 parts polymer (pphr) to the formulation. The polymer wasalso made photosensitive by the addition of PBG (TBD�HBPh4) at aloading of 1.35 pphr. The PAG and PBG loadings in PPC formulationscorrespond to approximately 332 monomer units per 1 mol of acidor base. Films were exposed using an Oriel Instrument floodexposure source at 2 J/cm2. PPC formulations were stabilized fromchain scission degradation by the addition of citric acid at 0.26 and3 pphr loadings. Formulations were left on a ball-roller for 3 h toensure complete mixing.

4. Results

4.1. Molecular weight effects

The molecular weight of the PPC can be an important contrib-utor to the overall decomposition and volatilization processbecause the polymer ultimately has to fragment into small, highvapor pressure moieties. A larger number of reactions would berequired to vaporize a high molecular weight polymer. Thedecomposition and volatilization of five PPC polymers with mo-lecular weight ranging from 2 kDa to 263 kDawere evaluated usingTGA. Each of the five PPC samples was analyzed as-received (neat)at a ramp rate of 1 �C/min in N2 atmosphere and the TGA results areshown in Fig. 3. While there is a general trend toward higherdecomposition temperature with higher molecular weights, thetrend does not correspond to simply the number of intramolecularreactions required to obtain small molecular weight products. Forexample, the onset of decomposition of 137 kDa and 2 kDa PPC arenearly the same even though the molecular weight of the 137 kDamaterial is 68 times greater than that of the 2 kDa polymer. The

difference between the molecular weight of the 137 kDa polymerand three higher ones is less than a factor of two, yet there is adramatic shift to higher temperature. Second, the TGA experimentswere performed at a very low scan rate giving time for intra-molecular reactions. It is interesting that the 2 kDa and 137 kDa hadnearly the same on-set temperature but the rate of decompositionof the 2 kDa PPC was significantly faster (sharper slope) than the137 kDa polymer which would be indicative of the number ofintermolecular reactions needed for volatilization. The speed of theunzipping mechanism of PPC is expected to be faster for lowermolecular weight polymers due to the greater number of end-groups. For example, the 2 kDa PPC has only 20 weight averagemonomer units per polymer chain, while the 263 kDa PPC has 2578weight average monomer units per polymer chain.

To further investigate the effect of molecular weight on thedecomposition temperature, five PPC samples were dissolved inGBL, solvent-cast onto a silicon substrate, and soft-baked at 100�Cfor 3 min. The PPC film was removed from the silicon wafer with arazor blade and evaluated using TGAwith the same conditions usedfor Fig. 3. The results are shown in Fig. 4. The decompositiontemperature of the five PPC samples (each with a different mo-lecular weight) was similar to each other and significantly differentfrom the as-received form of the same polymers shown in Fig. 3.The wide range of decomposition temperatures shown in Fig. 3 no

O. Phillips et al. / Polymer Degradation and Stability 125 (2016) 129e139 133

longer remained after this chemical treatment. The various mo-lecular weight PPC samples had only a minimal difference indecomposition temperature and the decomposition temperaturewas like that of the original 2 kDa and 137 kDa PPC in Fig. 3.

It is clear that the solvent-casting process changed the decom-position temperature of some of the materials, mitigating theoriginal differences in decomposition temperature. Solvent castingfrom GBL is purely a physical process of dissolving the PPC in asolvent followed by evaporation of most or all of the GBL,depending on the drying process. It is evident from this that mo-lecular weight has a minimal effect on the stability of PPC films. Atleast two explanations exist for the role of GBL in shifting the PPCdecomposition temperature. First, GBL could be participating indecomposition of PPC, which seems unlikely because GBL has noapparent reactivity with PPC. Second, GBL could remove (bydissolution) the presence of a stabilizing agent which had beenpresent in some of the as-received PPC.

4.2. End-capping

The mechanistic pathway and resulting temperature profile forPPC decomposition can be altered by end-capping with a chemicalmoiety which inhibits the unzipping reaction. The free hydroxylend-groups can react with a chemical moiety to kinetically inhibitthe unzipping reaction resulting in the thermal stabilization of PPC.Pyridine was used as a nucleophilic catalyst in the reaction. Theend-capping process takes place via nucleophilic attack with theend-group of the polymer. Organohalide compounds containingcarbonyl chlorides are excellent end-capping reagents due to thechloride leaving group and the partial positive charge of thecarbonyl group. Benzoyl chloride, 4-nitrophenyl chloroformate, andvinyl chloroformate were used in this study for end-capping PPC.Chloroformates are known to be more reactive end-caps due to theadditional electron-withdrawing oxygen bonded to the carbonylgroup.

Previously, end-capping of the PPC was verified only throughthermogravimetic analysis. Characterization of the end-groups onhigh molecular weight polymers has proved to be difficult becausethe number of polymer ends is small compared to the number ofmonomers in the polymer. 137 kDa PPC has 672 monomers perend-group. In this study, we verified the end-capping process byNMR of the 2 kDa PPC because it has only 10 weight averagemonomer units per end-group.

Fig. 5 shows the dynamic TGA in N2 atmosphere of the

Fig. 5. “VaccDry” Novomer 2 kDa PPC polyol end-capped with benzoyl chloride and 4-nitrophenyl chloroformate with a ramp rate of 5 �C/min in N2 atmosphere.

decomposition of neat and end-capped 2 kDa PPC. The samples arein the “VaccDry” form meaning that the residual methanol wasessentially completely removed from the PPC after end-cappingand precipitation. The temperature was ramped at 5 �C/min. Asseen, the decomposition temperature of PPC end-capped withbenzoyl chloride and nitrophenyl choloroformate showed 20 �Cand 50 �C increase in thermal stability, respectively, whencompared to neat PPC. The temperature shift in the TGA curvessuggests that end-capping the hydroxyl end-groups of the PPCinhibit the unzipping reaction. The degree of endcap stabilization isdependent on the stability (i.e. structure) of the endcap. In thisstudy, the 4-nitrophenyl chloroformate shows a greater degree ofstabilization from unzipping. The same end-capping reactions wererepeated with the higher molecular weight PPC (137 kDa) andshowed similar results, Fig. 6. It is interesting to note that the sta-bilization effect is not as significant as with the lower molecularweight PPC. Previous research has shown that higher molecularweight PPC is less susceptible to the unzipping mechanism becauseit has fewer terminal hydroxyl end-groups [18]. It is likely that the2 kDa PPC is more susceptible to unzipping because it has a greaterfraction of end-groups compared to the highmolecular weight PPC.The 4-nitrophenyl chloroformate shows the highest degree of sta-bilizationwith the 137 k PPC, just as with the 2 k PPC. However, theoverall stabilization effect due to endcap stabilization is lessimportant with the 137 k PPC, vs the 2 k PPC, because the unzippingreaction is not as significant for higher molecular weight materials.Higher molecular weight PPC has fewer ends than low molecularweight PPC making the random chain scission reaction moreimportant.

Fig. 7 shows the 1H NMR analysis of 2 kDa PPC after the freehydroxyl end-groups have been reacted with benzoyl chloride. Themajor peaks were identified as followed: 1H NMR (CDCl3, d, ppm),1.3 (3H;CH3, peak c), 4.18 (2H;CH2CH, peak b), 4.98 (1H;CH, peak a),0.95 (3H;CH3, peak e), 3.92 (2H;CH2, peak d). Chemical shifts of 1.3,4.18, and 4.8 ppm confirm the existence of carbonate linkage of thePPC polymer, which are denoted as peaks (a)-(c). Chemical shifts of0.95 and 3.92 ppm, denoted as peaks (e) and (d), show the diolinitiator is incorporated into the PPC backbone during synthesis.The chemical shifts of PPC agree with literature3,13,21. The carbonatecontent was calculated to be 91 mol% by integration of peaksfollowing this formula: fcarbonate% ¼ [a þ b]/[a þ b þ d]. This cor-responds to a 9 mol% of the diol initiator (or 1 diol monomer unitper polymer chain of PPC). No polyether content was observed inthe spectrum. We should also note here that 137 K PPC has

Fig. 6. “VaccDry” Novomer 137 kDa PPC end-capped with benzoyl chloride, vinylchloride, and 4-nitrophenyl chloroformate with a ramp rate of 5 �C/min in N2

atmosphere.

Fig. 7. 1H NMR analysis of “AirDry-MeOH” Novomer 2 kDa PPC polyol end-capped with benzoyl chloride in chloroform-D solution.

O. Phillips et al. / Polymer Degradation and Stability 125 (2016) 129e139134

approximately 99.99% carbonate linkages following the same for-mula and the polyether content is undetectable by NMR. In Fig. 7,Peaks (a)e(d) remained unchanged from before and after reactionshowing that the polymer backbone has not been altered from theend-capping reaction for the 2 kDa PPC. However, additional peaksfrom the end-group appear: 1H NMR (CDCl3, d, ppm), 7.96 (1H;CH,hydrogen f), 7.48 (1H;CH hydrogen g), 7.36 (1H;CH, hydrogen h).

Quantitative characterization of the molar ratio of the end-groups to the polymer chain ends can be determined from inte-gration of 1H NMR peaks. The number-average molecular weight ofthe lower molecular weight PPC is 1818.18 g/mol. This correspondsto a monomer to end-group ratio of 8.91:1. In Fig. 7, integration ofthe PPC backbone Peak (a) was compared to integration of theprotons from the end-cap Peak (f). Integration of Peak (a) shows anarea of 4.17 which is normalized to protons from two different siteson the end-group. Thus, the ratio of monomer to end-group is8.34:1. This value is very close to the number of monomers per end-groups (8.91:1) and within expected experimental error. Theseresults reveal that end-capping has been achieved.

13C NMR was used to confirm the site of the end-capping re-action on the PPC polymer chain. Fig. 8 show the 13C NMR analysisof 2 kDa PPC end-capped with benzoyl chloride, respectively. Themajor peaks were identified as followed: 13C NMR (CDCl3, d, ppm),16.3 (CH3, peak A), 69.1 (CH2CH, peak B), 72.5 (CH2CH, peak D),154.7(OCOO, peak E). Chemical shifts of 16.3, 69.1, 72.5, and 154.7 ppm,denoted as Peaks (A)-(E), show the signals from the backbone ofPPC remained before and after the end-capping reaction. Thebenzoyl groups were observed at 128.3, 129.5, 130.1, 133, and165.8 ppm, denoted as Peaks (F)-(I), after the end-capping reaction.The chemical shift of Peak (F) corresponds to the carbonyl carbon ofthe benzoyl group that has been reacted to the PPC ends. Thechemical shift of the carbonyl carbon of benzoyl chloride is168 ppm prior to end-capping and 165.8 ppm after end-capping.After end-capping, the carbonyl carbon of the benzoyl group iscovalently bonded to the oxygen termination of the PPC as opposedto chlorine. As a result, Peak (F) shifts upfield to 165.8 ppm indic-ative of end-capping.

The nature of the unreacted PPC end-groups is of interest. PPCcan be carboxylic acid and/or alcohol terminated, depending on the

synthetic route. The carboxylic acid hydroxyl has a chemical shiftgreater than 10 ppm in the 1H NMR spectrum, which was notobserved. The absence of a carboxylic acid peak suggests that thePPC is alcohol terminated. The alcoholic hydroxyl appears at4e5 ppmwhich is masked by the signal from the backbone protons.The 13C NMR spectrum also suggests that the end-groups arealcohol terminated. After reaction, peak (F) would shift furtherupfield to 161 ppm if the unreacted end groups were carboxylicacid terminated because of its electron withdrawing nature. How-ever, no such chemical shift was observed, suggesting that the2 kDa PPC were alcohol terminated. The analogous NMR analysiscould not be performed on the higher molecular weight PPCbecause the fraction of polymer ends was too low.

4.2.1. Acid/base catalyzed end-capped PPCAcid-catalyzed decomposition of end-capped PPC has also been

studied via TGA. The as-received 137 kDa PPC and benzoyl chlorideend-capped 137 kDa PPC samples were mixed separately with 3pphr of PAG (Rhodorsil-Faba) in GBL and solvent-cast onto a siliconsubstrate. The resulting films were exposed to 248 nm UV light at adosage of 2 J/cm2. The acid-activated PPC-PAG samples wereanalyzed by dynamic TGA and are shown in Fig. 9. Temperaturewasramped at 5 �C/min. As seen by the temperature of 5% weight loss,the onset of the degradation of the acid-catalyzed neat and end-capped PPC are almost identical at 88 �C. However, the end-capped polymer depolymerized at a significantly slower rate asthe temperaturewas raised. This is to be expected as acid-catalyzeddegradation of PPC carries out primarily by chain scission, but alsoby unzipping mechanisms. The onset of degradation is the samedue to the primary acid-catalyzed degradation pathway of chainscission, and the slower rate can be attributed to the inhibition ofthe second and minor acid-catalyzed degradation pathway ofunzipping.

Base-catalyzed decomposition of end-capped PPC was analo-gously studied via TGA in Fig. 10. The as-received 137 kDa PPC andbenzoyl chloride end-capped 137 kDa PPC samples were mixedwith 1.35 pphr of the PBG in order to maintain the same concen-tration of 332 monomer units to 1 mol of acid or base. The filmswere treated with the same UV exposure conditions as the acid-

Fig. 8. 13C NMR analysis of “AirDry-MeOH” Novomer 2 kDa PPC polyol end-capped with benzoyl chloride in chloroform-D solution.

Fig. 9. Solvent-casted Novomer 137 kDa PPC samples in GBL with 3 pphr PAG and a UVexposure dosage of 2 J/cm2. Ramp rate used was 5 �C/min in N2 atmosphere.

Fig. 10. Solvent-casted Novomer 137 kDa PPC samples in GBL with 1.35 pphr PBG and aUV exposure dosage of 2 J/cm2. Ramp rate used was 5 �C/min in N2 atmosphere.

O. Phillips et al. / Polymer Degradation and Stability 125 (2016) 129e139 135

catalyzed films and run with TGA. The 137 kDa PPC neat and end-capped samples were drastically different. The 5 weight % tem-perature of the base-catalyzed neat 137 kDa PPC is 70 �C, while theend-capped PPC is 174 �C. Base-catalyzed degradation of the end-capped PPC samples performed poorly as expected. PBG catalyzesprimarily the “backbiting” or unzipping mechanism of PPC. Thus, itis demonstrated here that the end-capped samples that are base-catalyzed are not as effective due to the inactive ends.

4.2.2. Solvent effectsSolvent-casting PPC from a variety of solvents was previously

shown to change the decomposition temperature of PPC [30]. Toinvestigate the effect of the casting solvent on the decompositiontemperature, end-capped polymers with residual solvent wereinvestigated. After end-capping, the PPC was precipitated from THFby the addition of methanol. The precipitated PPC product was thenair-dried which left a small amount of residual methanol in the PPC,identified as “AirDry-MeOH”. The thermal stability of as-received,end-capped with 4-nitrophenyl chloroformate and end-cappedPPC dried of residual methanol was compared and is shown in

Fig. 11. Novomer 137 kDa PPC samples end-capped with 4-nitrophenyl chloroformatewith “VaccDry” and “AirDry-MeOH” conditions.

O. Phillips et al. / Polymer Degradation and Stability 125 (2016) 129e139136

Fig. 11 for the 137 kDa PPC. End-capping the PPC inhibits theunzipping reaction of the polymer. The 50% weight loss tempera-ture of the dry end-capped PPC shifted to 240�C compared to theas-received polymer at 220�C. However, leaving the residualmethanol solvent in the end-capped PPC shifted the decompositiontemperature an additional 20�C to 260�C. The effect of solventcasting, as shown in Fig. 4, is simply to remove the residual, sta-bilizing solvent from the PPC. Thus, the stabilizing effect of residualmethanol adds to the end-capping stabilizing effect. End-cappingstabilizes PPC from unzipping whereas methanol appears to sta-bilize the random chain scission reaction.

The end-capping and methanol stabilization effects wererepeated on 2 kDa PPC to confirm that the end-groups were presentbefore and after the different processing steps in Fig. 12. The dryend-capped 2 kDa MW PPC shifted the 50% weight loss tempera-ture to 270�C compared to the as-received polymer at 220�C.Leaving the residual methanol solvent in the end-capped PPCshifted the decomposition temperature an additional 30�C to300�C. The end-capped PPC samplewith the residual methanol wasthen solvent-cast onto a silicon substrate from GBL or THF, soft-baked at 100�C for 5 min, and measured by TGA. As expected, the“VaccDry” end-capped PPC samples and the GBL or THF solvent castsamples produced similar 50 weight% decomposition temperaturesat approximately 270�C. The residual methanol that stabilized PPCfrom random chain scission was removed by the solvent-casting orvacuum drying procedure.

The presence of methanol in the “AirDry-MeOH” PPC sampleswas confirmed using NMR analysis. A peak at 3.46 ppm was iden-tified in the 1H NMR spectrum. The 3.46 ppm peak is consistentwith that of methanol and corresponds to a concentration of 28mol%. The monomer to end-group ratio was checked for this samplewith methanol present. The ratio of monomer to end-group in 1HNMR spectrum of “AirDry-MeOH” sample was found to be 8.2:1.The complete removal of methanol in “VaccDry” samples wasconfirmed by NMR analysis. Analogous integration measurementswere done in the 1H NMR spectrum of “VaccDry” sample gave amonomer to end-group ratio of 8.54:1. Thus, the NMR spectrareveal no significant change in the monomer to end-group ratiobetween “AirDry-MeOH” and “VaccDry” samples. The polymerend-groups are not affected by different processing techniqueseven though the TGA results show a significant shift in thermalstability for “AirDry-MeOH” samples with residual methanol. Thus,it is proposed that methanol is the stabilizing agent that inhibitsrandom chain scission. The stabilization may occur through com-plex interactions with carbonyl sites on the PPC backbone.

Fig. 12. Novomer 2 kDa PPC polyol samples end-capped with 4-nitrophenyl chlor-oformate with “Solvent cast”, “VaccDry”, and “AirDry-MeOH” conditions.

It is suggested thatmethanol only inhibits chain scission and notunzipping since the stabilization effect is also seen with PPC thathas been end-capped. Solvent-casting generally removes themethanol by dilution and evaporation during spin coating and soft-baking. PPC samples that were dried by forcibly removing meth-anol under vacuum or precipitation in a non-polar solvent mayprovide more accurate stability measurements of the polymer infuture use.

4.3. Chain scission stabilization

Monofunctional fatty acids have been used in the past to sta-bilize PPC by inhibiting chain scission degradation [27]. Fatty acids,such as stearic acid, typically have a very high decompositiontemperature and leave substantial residue which is not suitable fora sacrificial polymer in some electronic applications. Citric acid, atri-functional carboxylic acid, was used to kinetically stabilize PPCfrom chain scission. Citric acid has a decomposition temperature of175 �C [32e35], which is below the onset of decomposition of PPC(i.e. 180 �C). In addition, the citric acid vaporizes mostly to carbondioxide, acetone, and hydrogen, each of which are volatile. How-ever, citric acid can decompose to anhydride derivatives whichhave a lower vapor pressure. Sustained heating can further breakthis compound to more volatile products. Citric acid will likelyproduce minimal residue after decomposition from the polymermatrix.

A sample of 137 kDa PPC and 3 pphr of citric acid were mixed inGBL and solvent-cast onto a silicon substrate. Fig. 13 shows thedynamic TGA of this mixture. The PPC filmwith 3 pphr citric acid asan additive showed a 35 �C increase in thermal stability with a 50%weight loss temperature of 253 �C compared to the neat sample. Asexpected, excess citric acid (2.74 wt %) vaporized at 175 �C. Theamount of citric acid added to the film (3 wt%) was subtracted fromthe excess citric acid shown from the TGA profile suggesting thatonly 0.26 wt% of citric acid was actually needed for stabilizationfrom random chain scission mechanism. An exact amount of 0.26pphr was blended with 137 kDa PPC and cast on a silicon substrate,also shown in Fig. 13. The PPC film with 0.26 pphr citric acid dis-played only a 20 �C increase in thermal stability compared to the as-received polymer. The 50% weight loss temperature was 238 �C andthe evaporation of excess citric acid was not seen.

PPC is typically heated to higher temperatures to rapidlydegrade and volatize the sacrificial polymer leaving an air-cavity.An isothermal TGA at 180 �C of the film was done to quantify theamount of residue and the time to completely vaporize PPC. After

Fig. 13. Solvent-casted Novomer 137 kDa PPC samples in GBL with 0.26 and 3 pphrcitric acid.

O. Phillips et al. / Polymer Degradation and Stability 125 (2016) 129e139 137

2 h the residue is minimal (0.34 weight %) and not visible from theTGA pan.

4.4. Decomposition atmosphere

Although the ambient atmosphere within the TGA is notinvolved in the PPC decomposition reaction [2,10,22] (i.e. noinvolvement of nitrogen or oxygen in the primary unzipping orchain scission reactions), there have been some reports of shifts inthe TGA curves due to the ambient gas [30]. As-received PPCsamples were decomposed in nitrogen and air environments at aheating rate of 5 �C/min. The TGA curves in Fig. 14 show a signifi-cant thermal stability shift for the air ambient for all PPC molecularweights. Notably, the thermal shift of 137 kDa PPC had a 35 �C in-crease in thermal stability. Interestingly, the 160 kDa has only aminimal shift in thermal stability of 3 �C. However, the 50 wt%thermal decomposition temperature of the 137 kDa and 160 kDaPPC in air are nearly identical at 253 �C.

The 137 kDa PPC end-capped with vinyl chloroformate with“VaccDry” processing was also decomposed in nitrogen and airenvironments at a heating rate of 5 �C/min. The end-capped sampleshowed a 20 �C shift in thermal stability in air with a 50 wt%decomposition temperature of 256 �C, as compared to its thermalprofile in nitrogen environment - 50% weight loss at 236 �C. Thisresult suggests that the stabilization of PPC material in air is notrelated to any interaction with end-groups of the polymer.

The stabilizing effect of oxygen ambient (from air) was furtherinvestigated. PPC films with 3 pphr citric acid (i.e. already stabilizedfrom chain scission degradation) showed no difference in thermalstability in nitrogen and air ambient conditions. The nitrogen andair ambient had nearly identical 50% weight loss temperatures,256 �C. It appears that oxygen may react with specific in-termediates during the PPC decomposition process slowing thedecomposition reaction. Further investigation of this effect will benecessary to understand the role of oxygen.

4.5. End-capping and organic additives

Previous studies have focused on stabilizing PPC by consideringonly a single route to decomposition. This has led to confusing andsometimes contradictory conclusions. In this study, we haveconsidered both routes to decomposition, and the stabilization orcatalysis of each route. Polymer unzipping can be kinetically sta-bilized through covalent end-capping, and chain scission can bestabilized by additives. Inhibiting both mechanisms leads to the

Fig. 14. “Neat” Novomer PPC with molecular weights of 2 kDa, 137 kDa, 160 kDa at aramp rate of 5 �C/min in N2 and ambient air atmosphere.

highest decomposition temperature PPC, thereby expanding thethermal window. Fig. 15 compares the dynamic TGA profiles of thedecomposition of end-capped 137 kDa PPC with benzoyl chloride,137 kDa PPC with 3 pphr citric acid, and end-capped 137 kDa PPCwith 3 pphr citric acid as an additive. The plots were normalizedafter the initial loss of excess citric acid for comparison of thethermal stability. There is 20 �C and 35 �C increase in thermalstability by using end-capping or hydrogen bonding, however,when both decomposition mechanisms are suppressed, a 60 �Cincrease in thermal stability (50 weight % loss temperature of277 �C) is achieved.

4.6. Discussion

Many observations of decomposition temperature trends forPPC were made in the past including the effects of casting solvent,molecular weight, end-capping, and preparation. Molecular weighthas been shown to have a considerable effect on the thermal sta-bility of PPC 5, 19]. In the previous study, PPC was shown to shift tohigher decomposition temperature with increasing molecularweight. For example, a 144 kDa PPC compared to a 56 kDa showeda þ20 �C shift in thermal stability. Similar results were producedhere in Fig. 3 as the thermal stability appeared to trend with highermolecular weight with as-received PPC polymers. However, theprocess of dissolving the polymers in GBL and casting films showedthat thermal stability actually does not depend on molecularweight, as shown in Fig. 4. The highest molecular weight, solvent-cast PPC (263 kDa) displayed less than 8 �C shift in thermal stabilitycompared to the lowest molecular weight PPC (2 kDa). These re-sults suggest that impurities from synthesis can act as stabilizingagents and may be present in as-received polymers. The solvent-casting of samples can remove some impurities due to dilutionand evaporation during the spin-coating and soft-bake procedures.

Impurities have been the subject of investigations of thermalstability, specifically catalyst residues from polymerization[18,22,36e39]. Residual solvent from polymerization or purifica-tion procedures can affect the thermal stability of PPC [22,30] andare typically not reported in many studies. Methanol is one of thefrequently used solvents to precipitate and purify PPC from reactionsolution. It has been shown in this study to be an impurity that canact as a stabilizing agent. In Figs. 11 and 12, PPC containing residualmethanol had a higher thermal stability than PPC where themethanol was removed by vacuum evaporation. NMR analysis wasused to confirm the existence of methanol as an impurity. The PPC

Fig. 15. Novomer 137 kDa PPC “Neat”, “VaccDry” benzoyl chloride end-capped 137 kDaPPC with benzoyl chloride, and solvent-casted benzoyl chloride end-capped 137 kDaPPC with 3 pphr of citric acid. Ramp rate used was 5 �C/min in N2 atmosphere.

O. Phillips et al. / Polymer Degradation and Stability 125 (2016) 129e139138

material was end-capped, thus, any additional increase in thermalstability is the result of inhibition of chain scission through in-teractions with carbonyl moieties in the backbone of the polymer.The workup of these polymers from residual solvent has beendemonstrated to be an important factor to the overall thermalstability of PPC polymers.

End-capping has been shown in many previous studies throughthermal analysis to stabilize PPC by inhibiting the unzippingmechanism of hydroxyl terminated PPC. Characterization of theseend-groups has not been fully studied due to the low concentrationof end-groups. Low molecular weight PPC (2 kDa) was used tocharacterize the end-groups. 1H and 13C NMR analysis confirmedfor the first time that end-capping occurred with PPC, Figs. 7 and 8.Repeating the same experiments with higher molecular weightpolymers and more desirable mechanical properties showedsimilar thermal effects. TGA curves in Fig. 6 show the thermal shiftsin decomposition temperature as compared to the neat polymerwith hydroxyl end-groups with 137 kDa PPCmaterial. Interestingly,the end-capped 2 kDa PPC in Fig. 5 showed a much greater shift inthermal stability than 137 kDa PPC. The thermal decomposition oflow vs highmolecular weight PPC has been previously studied withpyrolysis-gas chromatography emass spectrometry. It has beenreported by X.H. Li et al. [18] that lower molecular weight PPC ismore susceptible to unzipping due to the larger fraction of end-groups. Furthermore, higher molecular weight PPC is more proneto chain scission due to the lower fraction of end-groups andgreater number of repeat units in the backbone. Each end-cap hasshown a different degree of stabilization from unzipping whenused with the 2 kDa PPC. It is suggested here that the thermalstability and electronic properties of the end-group can become amore significant factor with lower molecular weight PPC that de-grades primarily through unzipping.

Acid-catalyzed decomposition of PPC has also been the subjectof studies [28,29,40]. The mechanism of degradation has beenpreviously proposed by J. P. Jayachandaran et al. [11] from productformation analysis using a GCeMS system. The generated acid fromthe PAG after UV exposure protonates the carbonyl oxygen groupsacross the backbone of the polymer leading to a degradationpathway similar to chain scission. It was also shown that unzippingcan proceed from some acid-catalyzed intermediates leading topropylene carbonate as reaction products. Fig. 9 reveals the acid-catalyzed decomposition of an end-capped and neat PPC samples.The onset temperature is similar due to initiation of chain scissionas the primary degradation pathway. However, the end-capped PPCsamples displayed a much slower rate (broader slope) because ofthe inhibition of the second, and minor, acid-catalyzed degradationpathway.

Base-catalyzed decomposition of PPC has not been well-studiedin literature, but it has been suggested that PBG-caused degrada-tion occurs primarily by the “backbiting” or unzipping reaction. Sunet al. monitored the thermolysis of PPC with 3 wt % of TBD by FTIRand GPC and revealed the formation of cyclic propylene carbonate[32]. In the previous study, it was proposed that the strong aminebase performs a nucleophilic attack of the carbonyl group thatproduces an alkoxide. This intermediate further backbites leadingto the cyclic products in a similar way to the unzipping mechanism.PPC samples that have already been rendered “inactive” by end-capped chain ends should not be as susceptible to base-catalyzeddecompositions. Fig. 10 demonstrates the idea as the base-catalyzed degradation of end-capped PPC samples were not veryeffective compared to the neat PPC samples with free hydroxyl end-groups.

Organic compounds such as octadecanoic acid have been usedin previous studies to stabilize PPC from chain scission throughinteractions with carbonyl moieties [27]. The high decomposition

temperature of the octadecanic acid (350�C) leaves substantialresidue after the decomposition of the polymermatrix, which is notdesirable in some applications. In this study, a tri-functional car-boxylic acid additive, citric acid, was used to mitigate the residueproblem of octadecanoic acid. Citric acid has a lower decompositiontemperature (175�C). The thermal decomposition of 137 kDa PPCwith 3 pphr citric acid had a thermal stability higher than end-capped 137 kDa PPC samples. Studies have used octadecanoicacid and poly(ester) amide to stabilize PPC, and all suggested thathydrogen bonding is the reason [24,25,27]. However, confirmingthe presence of hydrogen bonding with the carbonyl group of PPChas not been easily demonstrated. It is clear from this study thatthese organic additives are inhibiting chain scission, due to theadded thermal stability with end-capped samples as shown inFig. 15. This implies that there is a complex interaction withcarbonyl moieties across the backbone of the polymer. However,further investigation is needed to determine the exact mechanismof inhibition that these additives provide.

5. Conclusion

Thermal stabilization of PPC as a sacrificial material for an air-cavity MEMS packaging process has been demonstrated. End-capping of PPC has been achieved and confirmed with NMR anal-ysis that has not been seen before in literature. Chain scission in-hibition was demonstrated through the use of citric acid additiveswith low residue after decomposition of the polymer matrix.Combing both methods of stabilizing interactions of the backboneof the polymer with citric acid and end-capping of chain ends in aPPC film provided thermal stability that surpassed bothmethods bythemselves. Residual solvents in PPC such as methanol, have beendemonstrated to be an impurity that can affect the thermal stabilitythrough interactions with the backbone of PPC. This study haswidened the thermal process window of PPC as a sacrificial mate-rial while maintaining low-residue for use in MEMS packaging.

Acknowledgment

The technical discussions and financial support of Promerus LLC,and materials donations from Novomer Inc. are gratefullyacknowledged. We gratefully acknowledge the synthesis of photo-base generators by the C. G. Willson Group at the University ofTexas.

References

[1] S.D. Allen, D.R. Moore, E.B. Lobkovsky, G.W. Coates, High-activity, single-sitecatalysts for the alternating copolymerization of CO2 and propylene oxide,J. Am. Chem. Soc. 124 (2002) 14284e14285. http://www.ncbi.nlm.nih.gov/pubmed/12452684.

[2] M.H. Chisholm, D. Navarro-Llobet, Z. Zhou, Poly(propylene carbonate). 1. Moreabout poly(propylene carbonate) formed from the copolymerization of pro-pylene oxide and carbon dioxide employing a zinc glutarate catalyst, Mac-romolecules 35 (2002) 6494e6504, http://dx.doi.org/10.1021/ma020348þ.

[3] L.J. Gao, M. Xiao, S.J. Wang, F.G. Du, Y.Z. Meng, Copolymerization of carbondioxide and propylene oxide with zinc glutarate as catalyst in the presence ofcompounds containing active hydrogen, J. Appl. Polym. Sci. 104 (2007) 15e20,http://dx.doi.org/10.1002/app.

[4] G. a Luinstra, G.R. Haas, F. Molnar, V. Bernhart, R. Eberhardt, B. Rieger, On theformation of aliphatic polycarbonates from epoxides with chromium(III) andaluminum(III) metal-salen complexes, Chemistry 11 (2005) 6298e6314,http://dx.doi.org/10.1002/chem.200500356.

[5] Q. Zhu, Y. Meng, S. Tjong, X. Zhao, Y. Chen, Thermally stable and high mo-lecular weight poly(propylene carbonate)s from carbon dioxide and propyl-ene oxide, Polym. Int. 51 (2002) 1079e1085, http://dx.doi.org/10.1002/pi.847.

[6] D.J. Darensbourg, Chemistry of carbon dioxide relevant to its utilization: apersonal perspective, Inorg. Chem. 49 (2010) 10765e10780, http://dx.doi.org/10.1021/ic101800d.

[7] M.R. Kember, A. Buchard, C.K. Williams, Catalysts for CO2/epoxide copoly-merisation, Chem. Commun. 47 (2011) 141e163, http://dx.doi.org/10.1039/

O. Phillips et al. / Polymer Degradation and Stability 125 (2016) 129e139 139

c0cc02207a.[8] G. Luinstra, Poly(propylene carbonate), old copolymers of propylene oxide

and carbon dioxide with new interests: catalysis and material properties,Polym. Rev. 48 (2008) 192e219, http://dx.doi.org/10.1080/15583720701834240.

[9] A. Michel, S. Allen, Polycarbonate polyol compositions, U. S. Pat. Appl. U. S.2013033 (2013).

[10] S. Peng, Y. An, C. Chen, B. Fei, Y. Zhuang, L. Dong, Thermal degradation kineticsof uncapped and end-capped poly(propylene carbonate), Polym. Degrad. Stab.80 (2003) 141e147, http://dx.doi.org/10.1016/S0141-3910(02)00395-6.

[11] J.P. Jayachandran, H. a Reed, H. Zhen, L.F. Rhodes, C.L. Henderson,S.A. Bidstrup, et al., Air-channel fabrication for microelectromechanical sys-tems via sacrificial photosensitive polycarbonates, J. MicroelectromechanicalSyst. 12 (2003) 147e159, http://dx.doi.org/10.1109/JMEMS.2003.809963.

[12] X.H. Li, Y.Z. Meng, G.Q. Chen, R.K.Y. Li, Thermal properties and rheologicalbehavior of biodegradable aliphatic polycarbonate derived from carbon di-oxide and propylene oxide, J. Appl. Polym. Sci. 94 (2004) 711e716, http://dx.doi.org/10.1002/app.20938.

[13] C. Guo, L. Zhou, J. Lv, Improved thermal stability and mechanical propertiesofpoly(propylene carbonate) by reactive blendingwith maleic anhydride, Polym.Polym. Compos 21 (2013) 449e456, http://dx.doi.org/10.1002/app.

[14] P. Monajemi, P.J. Joseph, P.A. Kohl, F. Ayazi, Characterization of a polymer-based MEMS packaging technique, in: 11th Int. Symp. Adv. Packag. Mater.Process. Prop. Interface, IEEE, n.d., 2006, pp. 139e144, http://dx.doi.org/10.1109/ISAPM.2006.1666023.

[15] R. Saha, N. Fritz, S.A. Bidstrup-Allen, P. a. Kohl, Packaging-compatible waferlevel capping of MEMS devices, Microelectron. Eng. 104 (2013) 75e84, http://dx.doi.org/10.1016/j.mee.2012.11.010.

[16] P.J. Joseph, P. Monajemi, F. Ayazi, P. a. Kohl, Wafer-level packaging ofmicromechanical resonators, IEEE Trans. Adv. Packag. 30 (2007) 19e26,http://dx.doi.org/10.1109/TADVP.2006.890220.

[17] E. Uzunlar, P. a. Kohl, Size-compatible, polymer-based air-gap formationprocesses, and polymer residue analysis for wafer-level MEMS packagingapplications, J. Electron. Packag. 137 (2015) 041001, http://dx.doi.org/10.1115/1.4030952.

[18] X.H. Li, Y.Z. Meng, Q. Zhu, S.C. Tjong, Thermal decomposition characteristics ofpoly(propylene carbonate) using TG/IR and Py-GC/MS techniques, Polym.Degrad. Stab. 81 (2003) 157e165, http://dx.doi.org/10.1016/S0141-3910(03)00085-5.

[19] X.L. Lu, Q. Zhu, Y.Z. Meng, Kinetic analysis of thermal decomposition ofpoly(propylene carbonate), Polym. Degrad. Stab. 89 (2005) 282e288, http://dx.doi.org/10.1016/j.polymdegradstab.2004.12.025.

[20] M.H. Chisholm, D. Navarro-Llobet, Z. Zhou, Poly(propylene carbonate). 1.More about poly(propylene carbonate) formed from the copolymerization ofpropylene oxide and carbon dioxide employing a zinc glutarate catalyst,Macromolecules 35 (2002) 6494e6504, http://dx.doi.org/10.1021/ma020348þ.

[21] D.D. Dixon, M.E. Ford, G.J. Mantell, Thermal stabilization of poly(alkylenecarbonate)s, J. Polym. Sci. Polym. Lett. Ed. 18 (1980) 131e134.

[22] C. Barreto, E. Hansen, S. Fredriksen, Novel solventless purification of poly(-propylene carbonate): tailoring the composition and thermal properties ofPPC, Polym. Degrad. Stab. 97 (2012) 893e904, http://dx.doi.org/10.1016/j.polymdegradstab.2012.03.033.

[23] C. Guo, L. Zhou, J. Lv, Improved thermal stability and mechanical propertiesofpoly(propylene carbonate) by reactive blendingwith maleic anhydride, Polym.Polym. Compos. (2013), http://dx.doi.org/10.1002/app.

[24] Y. Qin, L. Chen, X. Wang, X. Zhao, F. Wang, Enhanced mechanical performance

of poly(propylene carbonate) via hydrogen bonding interaction with o-lauroylchitosan, Carbohydr. Polym. 84 (2011) 329e334, http://dx.doi.org/10.1016/j.carbpol.2010.11.045.

[25] L. Chen, Y. Qin, X. Wang, Y. Li, X. Zhao, F. Wang, Toughening of poly(propylenecarbonate) by hyperbranched poly(ester-amide) via hydrogen bondinginteraction, Polym. Int. 60 (2011) 1697e1704, http://dx.doi.org/10.1002/pi.3132.

[26] T. Yu, Y. Zhou, K. Liu, Y. Zhao, E. Chen, F. Wang, et al., Improving thermalstability of biodegradable aliphatic polycarbonate by metal ion coordination,Polym. Degrad. Stab. 94 (2009) 253e258, http://dx.doi.org/10.1016/j.polymdegradstab.2008.10.021.

[27] T. Yu, Y. Zhou, Y. Zhao, K. Liu, E. Chen, D. Wang, et al., Hydrogen-bondedthermostable liquid crystalline complex formed by biodegradable polymerand amphiphilic molecules, Macromolecules 41 (2008) 3175e3180, http://dx.doi.org/10.1021/ma7020562.

[28] M.G. Cupta, Photoacid Generators for Catalytic Decomposition of Poly-carbonate, Georgia Institute of Technology, 2006.

[29] C.E. White, A. Balogun, C.L. Henderson, Effects of the photoacid generator typeon the imaging and thermal decomposition properties of photodefinable,thermally sacrificial poly(propylene carbonate) materials, J. Appl. Polym. Sci.102 (2006) 266e271, http://dx.doi.org/10.1002/app.23724.

[30] T.J. Spencer, P. a. Kohl, Decomposition of poly(propylene carbonate) with UVsensitive iodonium salts, Polym. Degrad. Stab. 96 (2011) 686e702, http://dx.doi.org/10.1016/j.polymdegradstab.2010.12.003.

[31] X. Sun, J.P. Gao, Z.Y. Wang, Bicyclic guanidinium tetraphenylborate: a pho-tobase generator and a photocatalyst for living anionic ring-opening poly-merization and cross-linking of polymeric materials containing ester andhydroxy groups, J. Am. Chem. Soc. 130 (2008) 8130e8131, http://dx.doi.org/10.1021/ja802816g.

[32] X. Sun, Development of Tetraphenylborate-based Photobase Generators andSacrificial Polycarbonates for Radiation Curing and Photoresist Applications,Carleton University, 2008.

[33] D. Wyrzykowski, E. Hebanowska, G. Nowak-Wiczk, M. Makowski,L. Chmurzynski, Thermal behaviour of citric acid and isomeric aconitic acids,J. Therm. Anal. Calorim. 104 (2011) 731e735, http://dx.doi.org/10.1007/s10973-010-1015-2.

[34] J.W. Fischer, L.H. Merwin, R.A. Nissan, NMR investigation of the thermolysis ofcitric acid, Appl. Spectrosc. 49 (1995) 120e126.

[35] M.M. Barbooti, D.A. Al-Sammerrai, Thermal decomposition of citric acid,Thermochim. Acta. 98 (1986) 119e126.

[36] J. Kodiyan, S. Jae, J. Hae, D. Woo, I. Yang, B. Yeoul, Thermal and weatheringdegradation of poly (propylene carbonate), Polym. Degrad. Stab. 95 (2010)1039e1044, http://dx.doi.org/10.1016/j.polymdegradstab.2010.03.006.

[37] W. Kuran, P. G�orecki, Degradation and depolymerization of poly(propylenecarbonate) by diethylzinc, Die Makromol. Chem. 184 (1983) 907e912, http://dx.doi.org/10.1002/macp.1983.021840502.

[38] B. Liu, L. Chen, M. Zhang, A. Yu, Degradation and stabilization of poly ( pro-pylene carbonate ), Macromol. Rapid Commun. 23 (2002) 881e884.

[39] J. Seo, G. Jeon, E.S. Jang, S.B. Khan, H. Han, Preparation and properties of poly(propylene carbonate) and nanosized ZnO composite films for packagingapplications, J. Appl. Polym. Sci. 122 (2011) 1101e1108, http://dx.doi.org/10.1002/app.

[40] C.E. White, C.L. Henderson, C.E. White, C.L. Henderson, Development ofimproved photosensitive polycarbonate systems for the fabrication ofmicrofluidic devices development of improved photosensitive polycarbonatesystems for the fabrication of microfluidic devices, J. Vac. Sci. Technol. B 21(2003) 2926, http://dx.doi.org/10.1116/1.1622934.