Article pubs.acs.org/Macromolecules
Thermally Induced Cross-Linking and Degradation Reactions of Benzocyclobutene-Based Polymers Anthony P. Gies,* Liam Spencer, Nathan J. Rau, Praveenkumar Boopalachandran, Mark A. Rickard, Kenneth L. Kearns, and Nolan T. McDougal Department of Core R&D Analytical Sciences, The Dow Chemical Company, 2301 N. Brazosport Blvd., B-1820, Freeport, Texas 77541, United States S Supporting Information *
ABSTRACT: We describe the use of MALDI-TOF mass spectrometry and collision-induced dissociation (CID) fragmentation techniques to examine the thermally induced cross-linking chemistry of dibenzocyclobutene (BCB)2− resorcinol-based materials. The overall goal was to gain a better understanding of benzocyclobutene ring-opening and the subsequent Diels−Alder cycloaddition reactions, which could assist in troubleshooting problems associated with unwanted side-product formation in these materials. Experimental results indicate that relative changes in average molecular mass (e.g., Mn) and signature side-product formation can be used to predict the optimum curing temperature used for cross-linking/polymerization. CID fragmentation studies identified two low kinetic energy degradation pathways for the BCB2−resorcinol-based oligomers examined in this study: (1) ether bond cleavage with an associated 1,3-hydrogen transfer and (2) transannular bond cleavage across the cyclooctadiene linkage. This information was used to develop a general fragmentation model for BCB2−resorcinol-based materials to predict their degradation products and verify their chemical connectivity.
■
INTRODUCTION The thermally initiated polymerization of monomers derived from benzocyclobutenes (BCB) has produced a number of thermoset polymeric systems with unique physical properties.1 One of the more unique polymeric systems is benzocyclobutene-terminated bisphenol A polycarbonates.2−4 This polymer motif is derived from 4-hydroxybenzocyclobutene, which can induce controlled cross-linking at elevated temperatures to produce polycarbonates that have a wide range of properties. The mechanism of BCB cross-linking in these systems has been extensively investigated and shown that BCB homopolymerization can proceed via the thermally induced ring-opening of the benzocyclobutene moiety to generate a reactive o-quinodimethane intermediate, which then undergoes a Diels−Alder cycloaddition to generate several new products. The mechanism of this transformation has been investigated with support for a pericyclic transformation involving sigmatropic shifts and retro-ene reactions.5 To broaden our knowledge about polymerizations with a BCB monomer, we set out to explore the thermal homopolymerization of dibenzocyclobutene-terminated resorcinol (“BCB2−resorcinol”), in addition to the copolymerization with N,N′-(1,3-phenylene)dimaleimide (“1,3-phenylene bismaleimide”). This BCB derivative has substitution on the fourmembered ring which is unique from many previously described polymeric systems.6 These prior reports utilized BCB monomers that possessed substitution on the sixmembered benzene ring and left the four-membered ring unperturbed. More recently, it has been reported that © XXXX American Chemical Society
substitution on the four-membered ring lowers the ringopening temperature of the BCB moiety.7−9 This approach has been utilized to lower the cross-linking temperature required in the preparation of polyacrylate nanoparticles.9 Harth et al.9 reported that two BCB molecules undergo ring-opening at 150 °C to generate an o-quinodimethane intermediate which can then undergo cross-linking C−C bond formation. While results support the formation of the desired cross-link, the exact chemical nature of the newly formed species was not reported. Driven by this finding, we became interested in the potential for these lower temperature conditions to influence the product distribution in polymerization reactions. Furthermore, the substitution of moieties on the four-membered ring may introduce new decomposition mechanisms that could lead to novel cross-linking and further diversify BCB polymer chemistry. To the best of our knowledge, there have been no reports that have examined the reaction chemistry of these types of BCB derivatives. In the course of both homopolymerization and copolymerization studies, we found that the isolated polymers were insoluble in common organic solvents. As a result, the use of “classical” polymer analysis techniques such as size exclusion chromatography (SEC), nuclear magnetic resonance (NMR), and infrared spectroscopy were deemed ineffective and abandoned, in favor of matrix-assisted laser desorption/ Received: January 4, 2017 Revised: February 28, 2017
A
DOI: 10.1021/acs.macromol.7b00018 Macromolecules XXXX, XXX, XXX−XXX
Article
Macromolecules
Scheme 1. Synthesis of the Model Small Molecule BCB2− Resorcinol/N-(p-Tolyl)maleimide Compound
ionization time-of-flight mass spectrometry (MALDI-TOF MS).10−12 It should be noted that the previously mentioned “classical” methods have the disadvantage of being averaging techniques, which provide general information about the “average” polymer mixture as a whole, instead of providing selective information about individual oligomers within the mixture.10−12 Further, these classical averaging techniques are rarely capable of providing information about the various oligomers and impurities that may be present within the polymer mixture. In contrast, mass spectrometry can be used for molecular mass determination, architectural elucidation, end-group analysis, quantification at trace levels, analysis of complex mixtures, and determination of degradation mechansisms.10−16 Moreover, through the use of collision-induced dissociation (CID), additional structural information about polymers can be determined from the fragment peaks of the precursor ions.17−21 Additionally, most of the classical techniques are “relative” methods that rely on calibration standards, which in many cases do not possess properties similar to the specific polymer that is being studied, and inferences must be made to yield “estimated” characterization information. Alternatively, mass spectrometry is an “absolute” method which does not rely upon polymer standards for calibration; this is an important advantage when standards do not exist for the polymer of interest.22,23 In the present study, we combine the evaporation−grinding matrix-assisted laser desorption/ionization (MALDI) sample preparation method (E−G method)17−20a technique developed specifically for the analysis of intractable materialswith TOF/TOF collision-induced dissociation (CID) to examine the thermal curing of BCB-based materials and generate a general fragmentation model to predict their low kinetic energy fragmentation pathways. It was hypothesized that an understanding of the BCB-based homopolymer and copolymeric degradation pathways could assist in studying the thermal curing of BCB-based materials and determining comonomer connectivity within specific copolymer chains of interest and possibly reveal information about the synthesis pathways and side-product formation. In this study, we describe the use of this technique to gain a better understanding of benzocyclobutene ring-opening and the subsequent Diels−Alder cycloaddition reaction to form new C−C bonds and assist in troubleshooting problems associated with unwanted sideproduct formation.
■
Synthesis and Thermal Curing of the BCB2−Resorcinol/N-(pTolyl)maleimide Copolymer. A 20 mL scintillation vial was charged with BCB2−resorcinol (100 mg, 0.32 mmol) and 1 equiv of N-(ptolyl)maleimide (60 mg, 0.32 mmol) and slowly heated under N2 in an aluminum block to 160 °C for 30 min. After this time, the vial was cooled to room temperature and submitted for mass spectrometric analysis. Synthesis and Thermal Curing of the BCB2−Resorcinol Homopolymer. Three 20 mL scintillation vials were charged with BCB2−resorcinol (100 mg, 0.32 mmol) and slowly heated under N2 in an aluminum block to 130, 160, or 190 °C for 30 min. After this time, the vial was cooled to room temperature and submitted for mass spectrometric analysis. Synthesis and Thermal Curing of the BCB2−Resorcinol/1,3Phenylene Bismaleimide Copolymer. A 20 mL scintillation vial was charged with BCB2−resorcinol (100 mg, 0.32 mmol) and 1 equiv of N,N′-(1,3-phenylene)dimaleimide (“1,3-phenylene bismaleimide”) (85 mg, 0.32 mmol) and slowly heated under N2 in an aluminum block to 160 °C for 30 min. After this time, the vial was cooled to room temperature and submitted for mass spectrometric analysis. Differential Scanning Calorimetry and Thermogravimetric Analysis. Differential scanning calorimetry (DSC) was used to determine the thermal properties of the BCB2−resorcinol monomer after thermal curing at 130, 160, and 190 °C as well as reactions with the 1,3-phenylene bismaleimide. A TA Instruments Q2000 DSC was used to generate heat−cool−heat profiles to determine the onset of reaction, peak temperature of the reaction, and the glass transition temperature of the material after the reaction was completed (Figure 1). These values are listed in Table S1 (Supporting Information). Samples were sealed in hermetic aluminum pans and heated from −80 to 250 °C at 10 °C/min in a nitrogen environment. The sample was held at 250 °C for 10 min between heating runs to ensure that the reaction was complete, which was corroborated by the lack of an exotherm during the second heat. The instrument was calibrated for heat flow and temperature using a pure indium standard. Thermogravimetric analysis (TGA) was also carried out to determine T95 values. T95 was the temperature where 95% of the initial mass remained during the heating profile. These values were collected by heating at 10 °C/min to 700 °C/min in a nitrogen atmosphere and are listed in Table S1. A Discovery Series TGA from TA Instruments was used. A four-point temperature calibration was conducted by determining the Curie temperature of alumel, nickel, Ni0.83Co0.17 alloy, and Ni0.63Co0.37 alloy. Infrared Spectroscopy Analysis. The liquid- and solid-phase mid-infrared spectra of the samples were collected on a Nicolet FTIR spectrometer equipped with a single bounce DuraScope diamond pike
EXPERIMENTAL SECTION
Synthesis of Dibenzocyclobutene-Terminated Resorcinol (“BCB2−Resorcinol”). To a 100 mL round-bottomed flask was added 1-bromobenzocyclobutene24 (2.00 g, 10.9 mmol), 1,3benzenediol (“resorcinol”) (481 mg, 4.37 mmol), K2CO3 (6.04 g, 43.7 mmol), and 50 mL of N,N-dimethylformamide (DMF). The stirring mixture was heated to 50 °C overnight whereupon it was cooled and the solvent removed by rotary evaporation. The residue was partitioned between 50 mL of water and 50 mL of diethyl ether. The organic layer was collected, and the aqueous layer was washed with diethyl ether (2 × 50 mL). The organic fractions were combined and washed with water (3 × 50 mL). After drying the organic fraction with MgSO4, the solution was filtered, and the organic solvent was removed by rotary evaporation and the residue recrystallized from boiling methanol (910 mg, 67% yield). The structure of the BCB2− resorcinol end product is shown in Scheme 1. 1H NMR (CDCl3): δ 3.32 (d, J = 8 Hz, 2H), 3.73 (dd, J = 2, 8 Hz, 2H), 5.68 (dd, J = 2 Hz, 2H), 6.68 (m, 3H), 7.19 (d, J = 8 Hz, 2H), 7.24−7.36 (m, 7 H). 13 C{1H} NMR (CDCl3): δ 39.6. 74.3, 102.5, 107.7, 123.1, 123.5, 127.4, 129.9, 130.1, 142.6, 144.6, 159.3. B
DOI: 10.1021/acs.macromol.7b00018 Macromolecules XXXX, XXX, XXX−XXX
Article
Macromolecules
Figure 1. DSC thermogram of the homopolymerization of BCB2−resorcinol. Daltronics Inc., Billerica, MA) equipped with a 355 nm Nd:YAG laser. Spectra were obtained in the positive ion reflection mode with a mass resolution greater than 20 000 full width at half-maximum height (fwhm), isotopic resolution was observed throughout the entire mass range detected, and the laser intensity was set approximately 10% greater than threshold. Instrument voltages were optimized for each spectrum to achieve the best signal-to-noise ratio. External mass calibration was performed using protein standards (Peptide Mix II) from a Peptide Mass Standard Kit (Bruker Daltronics) and a sevenpoint calibration method using Bradykinin (clip 1−7) (m = 757.40 Da), Angiotensin II (m = 1046.54 Da), Angiotensin I (m = 1296.68 Da), Substance P (m = 1347.74 Da), ACTH (clip 1−17) (m = 2093.09 Da), ACTH (clip 18−39) (m = 2465.20 Da), and Somatostatin 28 (m = 3147.47 Da) to yield monoisotopic mass accuracy better than Δm = ± 0.05 Da. The instrument was calibrated before each measurement to ensure constant experimental conditions. For CID fragmentation experiments, argon was used as a collision gas at pressures of 1.5 × 10−6 Torr and the collision energy amounts to 20 keV.25,26 All spectra were acquired in the reflection mode with a mass resolution greater than 20 000 full width at half-maximum height (fwhm); isotopic resolution was observed throughout the entire mass range detected. MALDI spectra were run in a dithranol (Aldrich) matrix doped with sodium trifluoroacetate (NaTFA; Aldrich). Polymer samples were prepared using the evaporation−grinding method (E−G method)15−18 in which a 2 mg sample of polymer was ground to a fine powder with 60 μL of distilled tetrahydrofuran (THF, Fisher) in an agate mortar and pestle. The molar ratios of matrix:NaTFA:polymer were 25:1:1. The mixture was then ground a second time to ensure homogeneity. A sample of the mixture was then pressed into a sample well, by spatula, on the MALDI sample plate. MS and MS/MS data were processed using Polymerix 3.0 software supplied by Sierra Analytics (Modesto, CA). Electronic Structure Calculations. All calculations were performed by using Gaussian09 software.27 Unless stated otherwise, all electronic structure (DFT, density functional theory) calculations were performed using the B3LYP level of theory with 6-31+G* basis sets. All reported energies are for optimized stable geometries and transition states in vacuo at 0 K and are corrected to account for zero point energies (ZPE). Optimized stable geometries and transition states were confirmed as minima or maxima by ensuring either zero or one imaginary vibrational modes were present, respectively. Optimized geometries are available in the Supporting Information. Transition state geometries were determined using QST2 or QST3 searches from optimized stable geometries for the starting and ending
Table 1. Structural Assignments for the Peaks Observed in the MALDI-TOF Mass Spectra of the Model Small Molecule BCB2−Resorcinol/N-(p-Tolyl)maleimide Compound
ATR accessory (Figures S1−S4). The spectrometer is equipped with a globar light source, a KBr beamsplitter, and deuterated L-alanine lanthanum triglycine sulfate (DLaTGS) detector. The samples were directly placed on the ATR crystal and scanned for mid-infrared experiments (4000−400 cm−1). Typically, 64 scans were collected with a resolution of 4 cm−1. Raman Spectroscopy Analysis. BCB2−resorcinol was flamesealed in a nitrogen-purged glass capillary tube. The capillary tube was rapidly heated to 160 °C with a Linkam THMS600 heating stage and held at that temperature. Raman spectra of the heated material were collected with a Raman RXN1 microscope (Kaiser Optical Systems, Inc.) equipped with a 785 nm Invictus laser. The laser was focused on the sample with a Mk II probe and a 20× microscope objective (≈20 mW at sample), and the backscattered light was collected with the same optics. The total acquisition time of each spectrum was 60 s. The spectra were analyzed with a classical least-squares model using the 1500−1750 cm−1 region. Pure component spectra of BCB2− resorcinol, the intermediate species, and the product were taken from the reaction. The Raman signal was normalized to the peak area from 2994 to 3125 cm−1 (Figures S5−S7). MALDI-TOF/TOF CID Measurements. All samples were analyzed using a Bruker UltrafleXtreme MALDI-TOF/TOF MS (Bruker C
DOI: 10.1021/acs.macromol.7b00018 Macromolecules XXXX, XXX, XXX−XXX
Article
Macromolecules Table 2. Structural Assignments for the Peaks Observed in the MALDI-TOF Mass Spectra of the BCB2−Resorcinol Homopolymer
states. In several instances, a transition state was not located between stable or quasi-stable states and will be discussed further below. For simplicity, chemical structures used for reaction modeling were calculated as neutral molecules or radicals. Even though the mass spectrometric experiments focused on sodiated ions, the CID fragmentation data strongly support charge-remote fragmentation. Therefore, we decided to exclude this complicating factor from our computational model, assuming that sodiation would not significantly alter relative reaction energies. Nomenclature. All figures will show structures and peaks labeled according to the following key: (i) Precursor ion peaks are labeled in the x−y format (x = table number, y = structure number), and the table and structure number are followed by the ion which provides the charge added to the oligomers during the MALDI process (e.g., Na+). (ii) Fragment ions are labeled with the letter “A” or “B” and are numbered in accordance with their respective fragmentation scheme. In the CID fragmentation spectra, precursor ions have their
corresponding molecular mass enclosed in a colored box, while their fragment ion masses/series are underlined in their respective color. For convenience, the precursor ion structures were inserted above the corresponding CID fragmentation mass spectrum, along with mass numbers of ions formed by specific bond fragmentation. (iii) The number of repeat units (n), when appropriate, corresponds to the calculated mass numbers found in Tables 1−3. For example, a precursor ion peak labeled “2−1 Na+” corresponds to structure 2−1 in Table 2 with a charge from a sodium cation (Na+) and some number of BCB2−resorcinol homopolymer repeat units (n). The structures of the species identified in Figures 2−9 are shown in Tables 1−3, without the Na+ ion attached. The mass ranges are listed in the tables for ions having different “n” values. Additionally, it should be noted that calculated masses are listed in Tables 1−3, while observed masses are reported in the figurestheir deviations being primarily linked to the MALDI ionization/calibration process. D
DOI: 10.1021/acs.macromol.7b00018 Macromolecules XXXX, XXX, XXX−XXX
Article
Macromolecules
Table 3. Structural Assignments for the Peaks Observed in the MALDI-TOF Mass Spectra of the BCB2−Resorcinol/1,3Phenylene Bismaleimide Copolymer
■
RESULTS AND DISCUSSION Thermal Curing Studies of BCB2−Resorcinol-Based Polymers. After synthesizing the BCB2−resorcinol monomer, it was examined with differential scanning calorimetry (DSC) and thermogravimetric analysis (TGA) to evaluate its reactivitytheir results are shown in Table S1. It has been well-documented that the exothermic peak observed in DSC
measurements of BCB derivatives is a result of the heat released upon the formation of new C−C bonds.6 This occurs after the ring-opening of the four-membered ring and has been previously used to identify curing temperatures to induce polymerization of BCB derivatives. The DSC thermogram of the BCB2−resorcinol monomer homopolymerization (Figure 1) reveals a small endotherm was observed at 90 °C, which is E
DOI: 10.1021/acs.macromol.7b00018 Macromolecules XXXX, XXX, XXX−XXX
Article
Macromolecules
Figure 2. MALDI-TOF mass spectrum of the BCB2−resorcinol/N-(p-tolyl)maleimide compound, thermally cured at 160 °C.
Figure 3. MALDI-TOF/TOF CID fragmentation spectrum of the sodium cationized BCB2−resorcinol/N-(p-tolyl)maleimide compound (711.2 Da) precursor ion (structure 1−1).
derivatives produced with substitution on the six-membered ring. The next step was to examine the reactivity of BCB2− resorcinol with N-(p-tolyl)maleimide to see if this reaction was a feasible route for making new copolymers that could potentially exhibit novel BCB branching chemistry. While the resulting compound was partially insoluble, it did demonstrate the proof-of-concept that the copolymerization was possible and led to further homopolymerization and copolymerization studies (vide inf ra). Additionally, BCB2−resorcinol homopolymerization was examined at three different curing temperatures (130, 160, and 190 °Cbased on the DSC results) to evaluate its BCB curing chemistry. Similar to the BCB2−resorcinol/N-
consistent with a melting point. A strong exotherm was observed with an onset of 130 °C, a midpoint at 160 °C, a maximum at 190 °C, and a cessation at 220 °C. These results are consistent with other BCB derivatives that possess substitution on the four-membered ring of the BCB molecule.28 In contrast, derivatives that possess no substitution on this ring typically have exothermic onsets greater than 200 °C.6 Furthermore, TGA results showed that the T95 values and thermal stability (as defined by the temperature observed for 5% mass loss, denoted T95) are significantly lower than previously reported systems. These results suggest that the products of the polymerizations with phenolic substitution on the four-membered ring on the BCB are different from those F
DOI: 10.1021/acs.macromol.7b00018 Macromolecules XXXX, XXX, XXX−XXX
Article
Macromolecules Scheme 2. Degradation Pathways of the BCB2−Resorcinol/ N-(p-Tolyl)maleimide Compound Species 1−1
There are two peaks of interest displayed in Figure 2: (1) the predominant peak at 711.3 Da (structure 1−1) and (2) the low intensity peak at 1025.5 Da (structure 1−2); their idealized structures are shown in Table 1. Structure 1−1 represents the BCB2−resorcinol/N-(p-tolyl)maleimide unimer; this is the expected product from the cycloaddition reaction shown in Scheme 1. Structure 1−2 represents the product formed when two BCB2−resorcinol monomers react to form a dibenzocyclooctadiene linkage, as depicted in Figure 2 and Table 1. These initial findings demonstrate the utility of the E-G MALDI sample preparation method for characterizing the various species produced in these insoluble materials. In the next series of experiments, we will discuss the use of CID fragmentation for structural characterization and elucidation of low kinetic energy fragmentation pathways of the BCB2−resorcinol/N-(ptolyl)maleimide compound. Fragmentation Pathways of the BCB2−Resorcinol/N-(pTolyl)maleimide Compound: Species 1−1. Figure 3 shows the MALDI-TOF/TOF CID spectrum for the sodiated BCB2− resorcinol/N-(p-tolyl)maleimide compound (structure 1−1: 711.2466 Da). This figure shows that three major and one minor fragment peaks of interest are observed under low kinetic energy CID conditions: 422.1, 502.2, 312.1, and 524.2 Da. The BCB2−resorcinol/N-(p-tolyl)maleimide monomer appears to have two low KE fragmentation pathways: (1) cleavage of the ether bond, with an associated 1,3-H transfer, yields the 422.1 and 312.1 fragment peaks (Scheme 2A),10,11 and (2) transannular bond cleavage that produces the high intensity protonated 502.2 Da fragment peak (Scheme 2B) and its associated trace level sodiated fragment peak at 524.2 Da (Figure 3).10,11 The additional trace level fragment observed at 132.0 Da is produced through multiple ether chain breaks. Note that these findings are in agreement with bond energy calculations (vide inf ra), that predict that the weakest bonds occur at the aromatic ether linkage and the six-membered ring attaching the BCB2−Resorcinol to the N-(p-tolyl)-maleimide molecule. MALDI-TOF Mass Spectral Analysis of the BCB2− Resorcinol Homopolymerization at 160 °C. MALDI-TOF MS of the BCB2−Resorcinol Homopolymerization at 160 °C. Figure 4 shows the overall MALDI spectrum for the BCB2− Resorcinol homopolymer that has been thermally cured at 160 °C. In this spectrum, a total of five unique species are identified and their structures are given in Table 2 along with the masses of the sodium cationized oligomers. The BCB−BCB endcapped oligomers (structure 2−1) account for the predominant species series (77.0%) observed in this oligomeric mixture. Given that this is the expected product, based on the cycloaddition reaction shown in Scheme 3, this is not surprising. However, inspection of the expanded MALDI spectrum in Figure 4 reveals the presence of three side-product series: (1) structure 2−2 represents a BCB−resorcinol capped series, at 8.9% total ion current (TIC); (2) structure 2−3 is a BCB−cyclooctadiene (“CyOD”) terminated species series, with 5.7% TIC; and (3) structure 2−4 represents a BCB− hydroylated cyclooctadiene (“CyODOH”) capped series, with a 6.7% TIC. There is an additional low molecular mass side product species series (structure 2−5) that is end-capped by two resorcinol groups (“Resor−Resor”) and has a TIC of 1.7%. In an effort to understand the relative degree of polymerization (for comparison of MALDI spectra), the Polymerix 3.0 software package (Sierra Analytics) was used to calculate the average molecular mass of the oligomer observed in Figure 4:
(p-tolyl)maleimide compound, the resulting BCB2−resorcinol homopolymers did not possess solubility in common organic solvents. It should be noted that the copolymerization of BCB2−resorcinol and 1,3-phenylene bismaleimide also yielded intractable end products. Because of their intractability, SEC, NMR, and FTIR were unable to generate usable data for these materials and their use was abandoned, in favor of the evaporation−grinding MALDI sample preparation method (E−G method)a solids analysis technique. In all cases, film studies indicated the presence of this unexpected chemistry and justified further mass spectral characterization, by CID fragmentation, to determine the true structures of the oligomers and co-oligomers. The following studies systematically examine the BCB2−resorcinol/N-(ptolyl)maleimide compound, BCB2−resorcinol homopolymer, and BCB2−resorcinol/1,3-phenylene bismaleimide copolymer, in an effort to gain insight into their polymerization/ degradation pathways. Additionally, it will be demonstrated how general degradation models can be used to predict the chemical connectivity of the BCB-based reaction products. MALDI-TOF Mass Spectral Analysis of the BCB2− Resorcinol/N-(p-Tolyl)maleimide Compound. MALDI-TOF MS of the Model BCB2−Resorcinol/N-(para-Tolyl)-Maleimide Compound. Figure 2 shows the overall MALDI spectrum for the model BCB2−resorcinol/N-(p-tolyl)maleimide compound. This model compound was initially chosen to examine the utility of the evaporation−grinding MALDI sample preparation method (E−G method) for examining the cross-linking chemistry of these materials. G
DOI: 10.1021/acs.macromol.7b00018 Macromolecules XXXX, XXX, XXX−XXX
Article
Macromolecules
Figure 4. MALDI-TOF mass spectrum of the BCB2−resorcinol homopolymer, thermally cured at 160 °C. The red and green asterisks refer to mass peaks assigned to species series 2−2 and 2−3, respectively.
Fragmentation Pathways of the BCB2−Resorcinol Homopolymer: Species 2−1. Figure 5 shows the MALDI-TOF/TOF CID spectrum for the expected BCB−BCB end-capped homopolymer product (structure 2−1: 965.3813 Da). There are essentially two fragmentation pathways that can explain the major fragment peaks in this CID spectrum; these are depicted in Scheme 4. Scheme 4A, the lowest kinetic energy degradation pathway of the BCB−BCB capped oligomer, initiates through a 1,3-H transfer accompanied by cleavage of the aromatic ether bond (on the cyclooctadiene side of the ether linkage). This cleavage explains the fragments produced at 549.2 and 439.2 Da (Scheme 4A1; labeled “A1” in Figure 5) and the fragments at 235.1 and 753.3 Da (Scheme 4A2; labeled “A2” in Figure 5). Further, through an additional cleavage of the second ether bond, the fragment at 541.2 Da (Scheme 4A3; labeled “A3” in Figure 5) is produced by this mechanism. Scheme 4B represents a high kinetic energy degradation pathway that involves transannular bond cleavage across the cyclooctadiene (CyOD) linkage. This mechanism explains the two relatively low intensity fragments observed at 337.1 and 651.3 Da. The
Scheme 3. Synthesis of the BCB2−Resorcinol Homopolymer
Mn = 1107 Da and Mw = 1400 Da. The next step was to use CID fragmentation for structural characterization and elucidation of low kinetic energy fragmentation pathways of the BCB2−resorcinol homopolymer. H
DOI: 10.1021/acs.macromol.7b00018 Macromolecules XXXX, XXX, XXX−XXX
Article
Macromolecules
Figure 5. MALDI-TOF/TOF CID fragmentation spectrum of the sodium cationized BCB2−resorcinol homopolymer, structure 2−1 (965.4 Da) precursor ion.
Scheme 4. Degradation Pathways of the BCB2−Resorcinol Homopolymer Species 2−1
I
DOI: 10.1021/acs.macromol.7b00018 Macromolecules XXXX, XXX, XXX−XXX
Article
Macromolecules Scheme 5. Extended Low Kinetic Energy Degradation Pathways of the BCB2−Resorcinol Homopolymer Species 2−1
Scheme 6. Degradation Pathways of the BCB2−Resorcinol Homopolymer Species 2−3
degradation mechanisms shown in Scheme 4 are consistent with this explanation. The next step in this study was to test this hypothesis by setting up an experiment to examine the effects of synthesis temperature on molecular weight and side-product formation. Additional Studies of BCB2−Resorcinol Homopolymer Thermal Curing Chemistry. BCB2−Resorcinol Homopol-
additional trace level fragments observed at 205.1 and 132.0 Da are produced through multiple chains breaks, primarily involving the 1,3-H transfer reactions shown in Scheme 4A. Comparison of the CID fragment structures in Figure 5 with the side products identified in Table 2 (specifically structures 2−2, 2−3, and 2−5) raises the concern that these side products (Figure 4) are created during the thermal curing process. The J
DOI: 10.1021/acs.macromol.7b00018 Macromolecules XXXX, XXX, XXX−XXX
Article
Macromolecules
Figure 6. MALDI-TOF mass spectrum of the BCB2−resorcinol/1,3-phenylene bismaleimide (1:1) copolymer, cured at 160 °C. The purple and light blue asterisks refer to mass peaks assigned to species series 3−4 and 3−5, respectively.
Figure 7. MALDI-TOF mass spectrum of the BCB2−resorcinol/1,3-phenylene bismaleimide (1:0.5) copolymer, cured at 160 °C. The purple and light blue asterisks refer to mass peaks assigned to species series 3−4 and 3−5, respectively.
ymer Thermally Cured at 130, 160, and 190 °C. Comparison of the Polymerix results on the right-hand side of Table 2 (Figure 4, Figures S8 and S10) reveals three critical pieces of information: (1) the molecular weights continue to increase throughout the curing process, with 190 °C producing the highest Mn and Mw; (2) the abundance of structure 2−1 (BCB−BCB capped series) decreases at the highest temperature; and (3) structures 2−2 and 2−5 (both series have resorcinol end groups) increase in intensity with increasing curing temperature. Based upon the calculated molecular weights, curing at 190 °C would be the optimum synthesis
temperature for this curing process. However, along with optimizing the molecular weight, the other caveat is to minimize the side-product formation. Between 130 and 160 °C the %TIC of structure 2−1 increases from 75.7% to 77.0%, but this change is not statistically significant. However, there is a dramatic decrease between 160 and 190 °C (from 77.0% to 53.7%, respectively). Furthermore, the decrease in structure 2− 1, at 190 °C, is accompanied by a notable increase in side products 2−2 and 2−5 (26.0% and 6.0%, respectively). Based upon the significant amount of side products formed at 190 °C, the optimum curing temperature was determined to be 160 °C. K
DOI: 10.1021/acs.macromol.7b00018 Macromolecules XXXX, XXX, XXX−XXX
Article
Macromolecules Scheme 7. Synthesis of the BCB2−Resorcinol/1,3-Phenylene Bismaleimide Copolymer
observed in this oligomeric mixture. Given the 1:1 mixing ratios of the comonomers, these are the expected products, based upon the cycloaddition reaction shown in Scheme 7. The last three structures only account for minor species series. For example, structure 3−3 is a maleimide−maleimide (“Mal− Mal”) end-capped oligomer series that represents 11.97% of the TIC, and structures 3−4 and 3−5, each containing the degradation “fingerprint” (the cyclooctadiene terminal group), represent 3.41% and 1.88% of the total ion current, respectively. Comparison of the %TICs, for each species series in Figures 6 and 7, reveals the expected changes when modifying the comonomer mixing ratios. For example, when 1:1 mixing ratios are used (Figure 6), the BCB−Mal capped series (structure 3− 1) has the largest %TIC. Similarly, when 1:0.5 mixing ratios are used (Figure 7), the BCB−BCB terminated series (structure 3−2) has the largest %TIC and an associated lowering of the % TIC for structure 3−1. The next step was to use CID fragmentation for structural characterization and elucidation of low kinetic energy fragmentation pathways of the BCB2− resorcinol/1,3-phenylene bismaleimide copolymer and test the utility of the proposed general fragmentation mechanism (Scheme 5). Fragmentation Pathways of the BCB2−Resorcinol/1,3Phenylene Bismaleimide Copolymer: Species 3−1. Figure 8 shows the MALDI-TOF/TOF CID spectrum for the sodiated BCB2−resorcinol/1,3-phenylene bismaleimide copolymer, structure 3−1 (1187.3474 Da), precursor ion. This figure shows the fragmentation data for two isobaric structures. “Isobar A” represents the linear form of structure 3−1, which is shown as the upper boxed structure in Figure 8. “Isobar B” represents the cyclic form of structure 3−1, which is shown in its ring-opened formed as the lower boxed structure in Figure 8. We will use our general fragmentation model (shown in Scheme 5) to explain the low kinetic energy fragmentation products of each isobaric structure. The linear structure 3−1 (isobar A) undergoes ether bond cleavage at four different linkages, as shown in Figure 8. The first cleavage occurs at the ether linkage labeled “A1” in Figure 8 and produces the fragments peaks at 393.1 and 817.3 Da. The second ether bond cleavage (labeled “A2”) forms the fragments at 503.1 and 707.2 Da. The last single bond ether cleavage (labeled “A3”) yields the fragment peak observed at 975.3 Da. Additional fragmentation, which results in the cleavage of two ether linkages, produces the fragments at m = 495.1 and m = 605.2 (labeled “A4” and “A5,” respectively). The cyclic structure 3−1 (isobar B) initially fragments through a ring-opening mechanism that involves cleavage of an ether linkage, with an associated 1,3-H transfer reaction. This ring-opened structure is shown in Figure 8. Subsequent fragmentation reactions occur at the three ether linkages previously described for the linear structure 3−1. For example, cleavage of the ether linkage labeled “B1” produces the fragment peaks at 495.1 and 715.2 Da. Cleavage of the ether bonds labeled “B2” and “B3” forms the fragments at 605.2 and 1077.3 Da, respectively. Additional multichain breaks produce the fragment at 495.1 Da (labeled “B4”). Based upon the low kinetic energy fragmentation pathways of structures 1−1, 2−1, and 3−1, a general fragmentation model was developed to predict the degradation products of these BCB2−resorcinol-based materials (Scheme 5). It is worth noting that this model was consistent with the architectural and end group defects that occur during the synthesis of BCB2− resorcinol-based materials. Also, the preferred fragmentation of
Inspection of the data raises two questions: (1) how are the degradation products formed, and (2) why is structure 2−3 not increasing with the curing temperature? We will use Scheme 5 to probe the first question about the evolution of the degradation products. Beginning with structure 2−1, the predominant species series in the oligomeric mixture, we propose that this structure undergoes a 1,3-H transfer reaction and cleavage of the ether linkage (Scheme 5A) to produce structures 2−2 and 2−3. Structure 2−2 can undergo an additional 1,3-H transfer and ether cleavage degradation process (Scheme 5B) to produce structures 2−3 and 2−5. These low kinetic energy degradation pathways are consistent with all of the major species observed in the MALDI spectra. Scheme 6 will be used to address why the %TIC for structure 2−3 is not observed to increase steadily with increasing curing temperature. During the thermal curing of the BCB2− resorcinol homopolymer, structure 2−3 is produced as a thermal degradation side product (Scheme 5). The double bond contained in the cyclooctatriene end group of structure 2−3 has the ability to react with the BCB2−resorcinol and BCB end groups contained on other oligomers (Scheme 6). Secondary reactions between these structures produce higher molecular weight oligomers of structure 2−3 (Scheme 6). Evidence for this reaction is shown in Figure S12, which displays the structure 2−3 series at much higher molecular weights than observed in the MALDI spectra collected at the 130 and 160 °C curing temperatures. MALDI-TOF Mass Spectral Analysis BCB2−Resorcinol/ 1,3-Phenylene Bismaleimide Copolymer. MALDI-TOF MS of the Model BCB2 Resorcinol/Dimaleimide. Figures 6 and 7 display the overall MALDI spectra for two BCB2−resorcinol/ 1,3-phenylene bismaleimide copolymers, that were synthesized using 1:1 and 1:0.5 mixing ratios of the comonomers. Since both spectra are very similar, we will focus our discussion on Figure 6. In this MALDI spectrum, there is a total of five species of interest; their idealized structures are shown in Table 3, along with the masses of the sodium cationized oligomers. (It should be noted that a number of these species have isobaric structures; vide inf ra.) Structures 3−1 and 3−2 account for the predominant species series (42.62% and 40.12%, respectively) L
DOI: 10.1021/acs.macromol.7b00018 Macromolecules XXXX, XXX, XXX−XXX
Article
Macromolecules
Figure 8. MALDI-TOF/TOF CID fragmentation spectrum of the sodium cationized BCB2−resorcinol/1,3-phenylene bismaleimide copolymer, structure 3−1 (1187.3 Da), precursor ion.
reduce the number of extraneous atoms in the predominant species (structure 2−1), we chose a smaller model compound that contains the core benzocyclobutene reactive moiety responsible for the observed cross-linking chemistry (Figure 9, structure A). Earlier work with benzocyclobutene based chemistry suggests cross-linking proceeds via cyclodimerization resulting in formation of dibenzocyclooctadiene rings.2,5 The dimerization of our model compound via this reaction mechanism2,5 has the potential to result in up to five constitutional isomers with a total of 18 unique structures (Table S2).
the aromatic ether linkage, on the cyclooctadiene side, is illustrated in the numerous “fingerprint” fragment peaks produced in the CID spectra of structures 1−1, 2−1, and 3− 1. Not only can these peaks be used for the identification of end groups and comonomer sequences, but they can also be used to differentiate the different isobars that are present in complex mixtures of BCB2 resorcinol-based materials (vide supra). Computational Study of the Model BCB2−Resorcinol Compound. A density functional theory (DFT) computational study was conducted to gain further insight into the BCB2−resorcinol CID fragmentation (and, by extension, thermal degradation) pathways proposed in this study. To M
DOI: 10.1021/acs.macromol.7b00018 Macromolecules XXXX, XXX, XXX−XXX
Article
Macromolecules
Figure 9. Potential energy surface of major CID fragmentation pathways for model BCB2 dimer calculated at B3LYP/631+G*, in kcal/mol.
Figure 9 shows a calculated potential energy surface for fragmentation of the lowest energy symmetrical isomer of the model BCB2 compound (Figure 9, Table S2 structure E3). However, calculated fragmentation pathways for all isomers are available in the Supporting Information. This reaction coordinate models the transannular and ether cleavage CID pathways discussed in the preceding sections. Upon initial inspection, it is apparent that the transannular pathway has higher energy intermediates than does the ether cleavage. This basic observation is in agreement with the CID results throughout this study which show a prevalence of fragment ions resulting from the ether cleavage relative to those from the transannular cleavage. Upon closer inspection of the hypothesized fragmentation mechanism, the transannular cleavage pathway appears more complex than the competing ether cleavage. Only a single transition state was located for the transannular fragmentation corresponding to the rearrangement between ring-opening and closing of the four-membered ring of the benzocyclobutene molecule. The mechanism proposed by Marks et al.2,5 suggests that the transannular fragmentation begins with a bond cleavage to form a diradical species that rearranges to an intermediate tricyclic structure (Figure 9, structure C). They also postulate that this rearrangement could potentially be concerted without an explicit diradical intermediate. According to our model, a concerted rearrangement would require C−C bond formation across approximately 3.3 Å and disruption of an aromatic system. We were unable to locate any transition state for such a concerted rearrangement. As expected, we found that for homolytic bond cleavage, only the triplet state would converge to a stable diradical structure. A third possibility is that fragmentation proceeds via two sequential
homolytic cleavages of the benzylic bond to directly yield two ring-opened forms of the benzocyclobutene “monomer” (Figure 9, structure B). While the exact fragmentation mechanism is not fully understood, the ring-opened forms of the fragmentation products (B) are 65.6 kcal/mol above the dimer and thus constitute a minimum energy barrier to the transannular fragmentation pathway. The ether cleavage fragmentation pathway is simpler and corresponds to the net loss of an alcohol and the formation of an alkene in the eight-membered ring. Figure 9 depicts the sequential homolytic cleavage of the ether bond followed by H atom abstraction to form the alcohol and alkene. Using DFT, a formal concerted 1,3 H-shift mechanism was not located. The transition state searches (QST2 and QST3) we employed only converge on the lowest energy transition state. None of the constrained geometries of the model cylcooctadiene ring isomers (Table S2) are favorable for a formal concerted 1,3 H-shift, and therefore we would not expect our DFT approach to locate one. This being said, rapid sequential ether bond cleavage and hydrogen atom abstraction in the same orbiting CID complex would still be consistent with our DFT model. To the experiment observer, there is no simple way to distinguish such sequential reaction steps from a formally concerted process in a single CID complex. Our calculations show that homolytic cleavage of the ether bond is a barrier-less or nearly barrier-less process. The H atom abstraction was expected to have a small barrier, but no transition state was conclusively determined. If it exists, the transition state for H atom abstraction is similar in geometry and energy to the radical intermediates, in line with the Hammond postulate. Given that no transition state was found, the products of homolytic ether bond cleavage likely represent N
DOI: 10.1021/acs.macromol.7b00018 Macromolecules XXXX, XXX, XXX−XXX
Article
Macromolecules
(4) Marks, M. J.; Schrock, A. K.; Newman, T. H. Carbonate Polymer Laminate Structure. U.S. Patent 5,198,527, Mar 30, 1993. (5) Marks, M. J.; Erskine, J. S.; McCrery, D. A. Products and Mechanism of the Thermal Crosslinking of BenzocyclobuteneTerminated Bisphenol A Polycarbonates. Macromolecules 1994, 27, 4114−4126. (6) Kirchhoff, R. A.; Bruza, K. Benzocyclobutenes in Polymer Synthesis. Prog. Polym. Sci. 1993, 18 (1), 85−185. (7) Oppolzer, W. Intramolecular Cycloaddition Reactions of orthoQuinodimethanes in Organic Synthesis. Synthesis 1978, 1978 (11), 793−802. (8) Chino, K.; Takata, T.; Endo, T. Polymerization of oQuinodimethanes. III. Polymerization of o-Quinodimethanes Bearing Electron-Withdrawing Groups Formed In Situ by Thermal RingOpening Isomerization of Corresponding Benzocyclobutenes. J. Polym. Sci., Part A: Polym. Chem. 1999, 37, 1555−1563. (9) Dobish, J. N.; Hamilton, S. K.; Harth, E. Synthesis of LowTemperature Benzocyclobutene Cross-Linker and Utilization. Polym. Chem. 2012, 3, 857−860. (10) Barner-Kowollik, C.; Gruendling, T.; Falkenhagen, J.; Weidner, S. Mass Spectrometry in Polymer Chemistry; John Wiley & Sons, Ltd.: West Sussex, England, 2012. (11) Montaudo, G.; Carroccio, S.; Montaudo, M. S.; Puglisi, C.; Samperi, F. Recent Advances in MALDI Mass Spectrometry of Polymers. Macromol. Symp. 2001, 169 (1), 101−112. (12) Montaudo, G.; Samperi, F.; Montaudo, M. S.; Carroccio, S.; Puglisi, C. Current Trends in Matrix-Assisted Laser Desorption/ Ionization of Polymeric Materials. Eur. Mass Spectrom. 2005, 11, 1−14. (13) Montaudo, G.; Samperi, F.; Montaudo, M. S. Characterization of Synthetic Polymers by MALDI-MS. Prog. Polym. Sci. 2006, 31, 277− 357. (14) Pasch, H.; Ghahary, R. Analysis of Complex Polymers by MALDI-TOF Mass Spectrometry. Macromol. Symp. 2000, 152, 267− 278. (15) Gies, A. P.; Hercules, D. M.; Ellison, S. T.; Nonidez, W. K. MALDI-TOF MS Study of Poly(p-phenylene Terephthalamide) Fibers. Macromolecules 2006, 39, 941−947. (16) Gies, A. P.; Hercules, D. M. MALDI−TOF MS Study of Aromatic Polybenzoxazole Fibers. Macromolecules 2006, 39, 2488− 2500. (17) Gies, A. P.; Kliman, M.; McLean, J. A.; Hercules, D. M. Characterization of Branching in Aramid Polymers Studied by MALDI−Ion Mobility/Mass Spectrometry. Macromolecules 2008, 41, 8299−8301. (18) Gies, A. P.; Geibel, J. F.; Hercules, D. M. MALDI-TOF MS Study of Poly(p-phenylene sulfide). Macromolecules 2010, 43, 943− 951. (19) Crecelius, A. C.; Baumgaertel, A.; Schubert, U. S. Tandem Mass Spectrometry of Synthetic Polymers. J. Mass Spectrom. 2009, 44, 1277−1286. (20) Gross, J. H. Mass Spectrometry - A Textbook; Springer: Berlin, 2004. (21) McLafferty, F. W.; Turecek, F. Interpretation of Mass Spectra, 4th ed.; University Science Books: Sausalito, 1993. (22) Murgasova, R.; Hercules, D. M. MALDI of Synthetic PolymersAn Update. Int. J. Mass Spectrom. 2003, 226, 151−162. (23) Gruendling, T.; Weidner, S.; Falkenhagen, J.; Barner-Kowollik, C. Mass Spectrometry in Polymer Chemistry: A State-of-the-Art UpDate. Polym. Chem. 2010, 1, 599−617. (24) Park, J.; Sun, J.; Gilmore, C. D.; Zhang, J. Z.; Hustad, P. D.; Trefonas, P.; O’Connel, K. M. US Provisional Filing 20150210793, July 30, 2015. (25) Altuntas, E.; Krieg, A.; Baumgaertel, A.; Crecelius, A. C.; Schubert, U. S. ESI, APCI, and MALDI Tandem Mass Spectrometry of Poly(methyl acrylate)s: A Comparison Study for the Structural Characterization of Polymers Synthesized via CRP Techniques and the Software Application to Analyze MS/MS data. J. Polym. Sci., Part A: Polym. Chem. 2013, 51, 1595−1605.
a good estimate of the lower boundary of the highest energy species in the reaction resulting in the alkene product. Overall, the calculated fragmentation pathway supports the experimental observations of a low energy ether cleavage and a slightly higher energy transannular cleavage pathway.
■
CONCLUSIONS The results of this study show that the evaporation−grinding MALDI sample preparation method can be used to examine the intact distributions present within complex mixtures of insoluble materials. Experimental results indicate that relative changes in average molecular mass (e.g., Mn) and signature side-product formation can be used to predict the optimum curing temperature used for cross-linking/polymerization. CID fragmentation studies identified two low kinetic energy degradation pathways for the BCB 2 −resorcinol-based oligomers examined in this study: (1) ether bond cleavage with an associated 1,3-hydrogen transfer and (2) transannular bond cleavage across the cyclooctadiene linkage. This information was used to develop a general fragmentation model for BCB2−resorcinol-based materials to predict their degradation products and verify their chemical connectivity. Future studies will include refinement of this general fragmentation model to probe the structure/property relationships of new cross-linked materials.
■
ASSOCIATED CONTENT
S Supporting Information *
. The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.macromol.7b00018. Additional CID fragmentation spectra, FTIR, Raman, and computational data are presented for the BCB2− resorcinol-based materials (PDF)
■
AUTHOR INFORMATION
Corresponding Author
*Tel (979) 238-1778; e-mail
[email protected] (A.P.G.). ORCID
Anthony P. Gies: 0000-0002-3558-593X Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS This work was performed under the auspices of The Dow Chemical Company. Additional gratitude is extended to Prof. David M. Hercules and Prof. Ned A. Porter, at Vanderbilt University, for their helpful discussions on characterization of cross-linked materials.
■
REFERENCES
(1) Hahn, S. F.; Townsend, P. H.; Burdeaux, D. C.; Gilpin, J. A. In Polymeric Materials for Electronics and Interconnection; Lupinski, J. H., Moore, R. S., Eds.; ACS Symposium Series 407; American Chemical Society: Washington, DC, 1989. (2) Marks, M. J.; Sekinger, J. K. Synthesis, Crosslinking, and Properties of Benzocyclobutene-Terminated Bisphenol A Polycarbonates. Macromolecules 1994, 27, 4106−4113. (3) Marks, M. J.; Schrock, A. K.; Newman, T. H. Process for the Preparation of Arylcyclobutene Terminated Condensation Polymers. U.S. Patent 5,171,824, Dec 18, 1992. O
DOI: 10.1021/acs.macromol.7b00018 Macromolecules XXXX, XXX, XXX−XXX
Article
Macromolecules (26) Altuntas, E.; Krieg, A.; Crecelius, A. C.; Schubert, U. S. Tandem Mass Spectrometry of Poly(methylacrylate)s by ESI, APCI and MALDI. Bruker Application Note # ET-37. (27) Gaussian 09, Revision C.01: Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. A.; Nakatsuji, H.; Caricato, M.; Li, X.; Hratchian, H. P.; Izmaylov, A. F.; Bloino, J.; Zheng, G.; Sonnenberg, J. L.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Vreven, T.; Montgomery, J. A., Jr.; Peralta, J. E.; Ogliaro, F.; Bearpark, M.; Heyd, J. J.; Brothers, E.; Kudin, K. N.; Staroverov, V. N.; Kobayashi, R.; Normand, J.; Raghavachari, K.; Rendell, A.; Burant, J. C.; Iyengar, S. S.; Tomasi, J.; Cossi, M.; Rega, N.; Millam, J. M.; Klene, M.; Knox, J. E.; Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Martin, R. L.; Morokuma, K.; Zakrzewski, V. G.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Dapprich, S.; Daniels, A. D.; Farkas, Ö .; Foresman, J. B.; Ortiz, J. V.; Cioslowski, J.; Fox, D. J. Gaussian, Inc.: Wallingford, CT, 2009. (28) Chino, K.; Endo, T. Prediction of Thermal Isomerization Temperatures of Substituted Benzocyclobutenes to o-Quinodimethanes by Semi-Empirical Molecular Orbital Calculation. Lett. Org. Chem. 2011, 8, 138−142.
P
DOI: 10.1021/acs.macromol.7b00018 Macromolecules XXXX, XXX, XXX−XXX