Research Article Cite This: ACS Catal. 2018, 8, 8006−8013
pubs.acs.org/acscatalysis
Mechanism of Isobutylene Polymerization: Quantum Chemical Insight into AlCl3/H2O‑Catalyzed Reactions Minh Nguyen Vo,† Yasemin Basdogan,† Bridget S. Derksen,† Nico Proust,‡ G. Adam Cox,‡ Cliff Kowall,‡ John A. Keith,† and J. Karl Johnson*,† †
Downloaded via KAOHSIUNG MEDICAL UNIV on August 1, 2018 at 18:33:32 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
Department of Chemical & Petroleum Engineering, Swanson School of Engineering, University of Pittsburgh, 3700 O’Hara Street, Pittsburgh, Pennsylvania 15261, United States ‡ The Lubrizol Corporation, 29400 Lakeland Boulevard, Wickliffe, Ohio 44092, United States S Supporting Information *
ABSTRACT: The production of polyisobutylene with Lewis acid catalysts has been in widespread use for over 60 years, but no validated molecular-level understanding of the reaction mechanism exists. We have computed initiation and propagation reaction pathways for isobutylene polymerization under industrially relevant conditions with an AlCl3/H2O initiator from density functional theory calculations. The initiator/catalyst complex we identified is fundamentally different from the putative complex identified in the literature, which typically assumes that the AlCl3OH2 complex is the active catalyst. We found that the reaction pathway with the AlCl3OH2 complex is infeasible due to unreasonably high energy barriers. Our calculations indicate that a minimum of two AlCl3 groups and one H2O molecule is required to initiate the reaction and that the complex must produce a highly acidic proton. It is the extreme acidity of the complex that is crucial for successful initiation of the reaction. The active catalyst moiety we identified produces low-energy-barrier pathways for both initiation and propagation steps. This complex was identified using the growing-string method to identify possible reaction pathways with various AlCl3/ H2O complexes. The initiation reaction with our proposed complex was observed to occur naturally in an ab initio molecular dynamics simulation under typical operating conditions, confirming the activity of the complex. KEYWORDS: cationic polymerization, AlCl3 catalyst, Brønsted acid, density functional theory, growing string method, superacidity
■
INTRODUCTION Polyisobutylenes (PIBs) are highly versatile materials, having properties that can be tuned for specific applications by varying the molecular weight (Mn) and chain termination. Highmolecular-weight PIBs (Mn > 105 g mol−1) have high viscosity and are commonly used for the synthesis of rubber products, such as chewing gum and car tires. Medium-molecular-weight PIBs (5 × 103 < Mn < 3 × 104 g mol−1) vary from viscous liquids to tacky semisolids and can be found in sealant and caulking products. Low-molecular-weight PIBs (Mn < 5 × 103 g mol−1) are used as precursors for making adhesives and lubricants and as additives for motor oils and fuels. PIBs are produced through exothermic cationic polymerization. The molecular weight distribution is typically controlled through the reaction temperature. Low-molecular-weight PIBs are polymerized between −40 and 10 °C, and high-molecularweight PIBs require even lower temperatures of about −100 to −90 °C.1 Cationic polymerization typically involves the use of a Lewis acid as the catalyst (e.g., AlCl3, BF3, TiCl4).2 Another important component is the proton donor, such as water, hydrogen halide, or alcohol. The choice of catalyst can greatly affect the terminal groups on PIBs. The AlCl3-catalyzed process produces conventional PIBs, which contain up to 90% of internal double bond end groups (trisubstituted, tetrasub© XXXX American Chemical Society
stituted), as shown in Figure 1a. In contrast, the BF3-catalyzed process produces PIBs having a high content of terminal
Figure 1. (a) Terminal group of conventional PIB on catalysis with AlCl3. (b) highly reactive PIB terminal group on catalysis with BF3.
vinylidene (exo) groups, as shown in Figure 1b. PIBs having a high percentage of exo groups are known as highly reactive (HR) PIBs. With the terminal vinylidene group, HR PIBs react readily with maleic anhydride to form PIB succinic anhydride, which can react with oligoalkylenamines to produce dispersants.3,4 Conventional PIBs have low reactivity because of the internal double bond and therefore require a promoter (e.g., Cl2) to react with maleic anhydride. Received: April 16, 2018 Revised: July 18, 2018
8006
DOI: 10.1021/acscatal.8b01494 ACS Catal. 2018, 8, 8006−8013
Research Article
ACS Catalysis
highly acidic proton that gives an acceptably low barrier to the initiation reaction. It is the extreme acidity of the complex that is the distinguishing characteristic of a viable catalyst; this superacidity requirement has been overlooked in describing the reaction pathway for Lewis acid cationic polymerization until now.
It is well-known that neat AlCl3 and BF3 are not active initiators. The presence of adventitious water or another proton donor is essential for catalyzing the polymerization reaction.3−6 The putative mechanism reported in the literature proceeds as follows:2,4,7,8 AlCl3 + H 2O F AlCl3·OH 2
■
(1)
METHODS We have carried out both gas-phase and condensed-phase calculations in this work. The gas-phase calculations were used to identify potential reaction pathways and reaction energies. Condensed-phase calculations were carried out to explore explicit solvent effects and reaction events through molecular dynamics. Gas-Phase Calculations. Gas-phase electronic structure calculations were performed with the ORCA program.11 Optimizations were performed at the BP8612/def2-SVP13 level of theory with the RI-J approximation14 and Grimme’s D3 dispersion corrections.15 Frequency calculations were performed at the same level of theory and basis set to verify that initial and final state geometries were at local minima, that the transition state had only one imaginary frequency, and to compute zero point energies and thermal corrections. Vibrational frequencies lower than 60 cm−1 were raised to 60 cm−1, following the approximation proposed by Truhlar et al.16 to correct for the well-known breakdown of the harmonic oscillator model for free energies of low-frequency vibration modes. Single-point energy calculations were performed using the BP86/def2-TZVP13 level of theory. The gas-phase Gibbs free energy (Gg) for each species under standard conditions (T = 298 K, P = 1 bar) was computed from
AlCl3·H 2O + (CH3)2 CCH 2 → (CH3)3 C+(AlCl3OH)− (2)
[(CH3)3 C(CH 2C(CH3)2 )n ]+ (AlCl3OH)− + (CH3)2 CCH 2 → [(CH3)3 C(CH 2C(CH3)2 )n + 1]+ (AlCl3OH)−
(3)
The first step in this process involves a reaction between a Lewis acid (AlCl3) and a proton donor (H2O) to form an initiator complex (AlCl3OH2), which is often represented as [H]+[AlCl3OH]−. Consequently, AlCl3OH2 is believed to be the active catalyst. Next, the initiator complex donates a proton to an isobutylene (IB) to form the tert-butyl cation (IBH+) and the counterion (AlCl3OH−), as indicated in eq 2. The reaction propagates as additional IBs are added to the chain, as shown in eq 3. The termination step can occur through chain breaking or chain transfer, depending on the operating temperature and choice of catalyst.9 PIBs have been produced industrially for over 60 years, but no validated molecular-level understanding of the reaction mechanism exists. To the best of our knowledge, there has been only one molecular-level study of the cationic polymerization of IB in the literature. Chaffey-Millar and Kühn studied the PIB initiation mechanism with manganese complexes as catalysts using density functional theory coupled with secondorder Møller−Plesset (MP2) perturbation theory to compute reaction pathways and barriers.10 However, that study failed to identify any feasible pathways for the proposed initiation mechanisms. The focus of this work is to identify initiation and propagation pathways for AlCl3-catalyzed production of PIB under typical industrial conditions. The use of halogenated solvents such as dichloromethane and chloroform is common in academic laboratories (on small scales) to facilitate the solvation of the desired reagents and promote the polymerization reaction in the liquid phase at very low temperatures and atmospheric pressures. However, the use of chlorinated solvents is unusual for production scales in industry because they present significant health and environmental hazards and require an additional separation process for solvent recovery. Industrial IB polymerization is usually carried out at temperatures of about 10 °C in pressurized reactors, using a feed of isobutene and a mixture of butanes. The C4 alkanes act as a solvent. Reaction pressures are adjusted to maintain a liquidphase environment in the reactor. The AlCl3 catalyst is added as a fine (roughly micrometer size) powder, and trace amounts of water are required to carry out the reaction. We note that if too little or too much water is added the reaction will not proceed at an appreciable rate. In this work, we will show that a complex of one AlCl3 and one H2O cannot be the active catalyst for IB polymerization, in contradiction to the putative mechanism. More importantly, we have identified an initiator complex consisting of two AlCl3 groups combined with one H2O that facilitates low-barrier reaction pathways for both the initiation and propagation steps in the polymerization of IB. Crucially, this complex contains a
Gg = ESCF + ZPE + ΔH0 → 298 − TS
(4)
where ESCF is the single-point electronic energy calculated from Kohn−Sham density functional theory (DFT) or a higher level of theory, ZPE is the zero-point energy, ΔH0→298 is the change in enthalpy due to changing the temperature from 0 to 298 K, and S is the entropy at 298 K and a pressure of 1 bar. We used standard-state conditions in keeping with common practice. Moreover, 298 K is close to the practical reaction temperature of 283 K. All gas-phase calculations used the standard ideal gas, rigid rotor, and harmonic oscillator approximations. The Gibbs free energies were used mainly to assess stabilities of initiator complexes and compute pKa values. As previously mentioned, the polymerization reaction is carried out with IB in a solvent consisting of a mixture of C4 alkanes.17 The dielectric constants for these solvents are very small (e.g., butane’s dielectric constant is ε = 1.4). Thus, gasphase reaction energies (ε = 1) were used to approximate the thermodynamics of the actual processes. However, we report electronic energies (0 K), rather than Gibbs free energies, in the potential energy surface diagrams below. This is because the ideal gas entropy, used in the Gibbs free energy calculations, is not necessarily an accurate approximation to the entropy in the condensed phase. Our approach is also supported by recent work showing that 0 K reaction barriers, like those presented here, are often in good agreement with Gibbs free energy barriers for condensed-phase reactions computed by the potential of mean force approach.18 Reaction Pathway Search and Activation Energy Estimation. The growing string method (GSM), a transition 8007
DOI: 10.1021/acscatal.8b01494 ACS Catal. 2018, 8, 8006−8013
Research Article
ACS Catalysis state locating technique developed by Zimmerman,19−21 was used to identify reaction pathways. The GSM method has been applied to a variety of different systems and has been shown to be both accurate and efficient.19−27 GSM calculations were performed with ORCA using the BP86 functional with D3 dispersion correction and with the 6-31G** basis set28,29 to provide the quantum mechanical gradients. We found that the single-ended GSM is a useful tool for identifying potential reaction pathways because it does not require the user to know the configuration of the product of the reaction and only takes the reactant configuration and a set of bonds to break and/or form as the input. The product configurations typically identified with single-ended GSM were not at local minima. Consequently, the product configurations were optimized in ORCA with the same functional and basis set and used as inputs in the double-ended GSM method to generate the potential energy surfaces reported here. Single-point energy calculations on GSM reaction coordinates were performed using the BP86/def2-TZVP model chemistry. We have also computed single-point energies on some reaction pathways using the B3LYP functional and MP2. These results are compared with the BP86 results in Figures S1−S3 and Table S1 of the Supporting Information. Overall, the trends predicted from BP86 are in reasonable agreement with those computed from higher levels of theory. Ab Initio Molecular Dynamics (AIMD) Simulations. Born−Oppenheimer AIMD simulations were performed with the CP2K software package,30 using the QUICKSTEP method31 and a time step of 0.5 fs. We used the Perdue− Burke−Ernzerhof (PBE) generalized gradient functional.32 Goedecker−Teter−Hutter (GTH) pseudopotentials33,34 and short-range double-ζ basis sets with polarization35 were employed. Grimme’s D3 dispersion correction was included. Periodic unit cells containing up to 10 IBs and various initiator complexes were created. The systems were equilibrated for 10−15 ps in the isothermal−isobaric ensemble to achieve liquid densities consistent with experimental conditions. The barostat for the isothermal−isobaric ensemble simulation was set at 10 bar, and the Nosé thermostat36 was set at 300 K, which gave liquid states for all systems studied. The equilibrated configuration was then used to perform isothermal−isochoric ensemble simulations of about 15 ps at 300 K, with some simulations at elevated temperatures (up to 1500 K). The AIMD simulations were only used to observe spontaneous reaction events. Reaction energetics were not computed because these require computationally intensive methods (e.g., potential of mean force).
Figure 2. Putative initiator complex AlCl3OH2.
Figure 3. Potential energy surface for the initiation reaction with AlCl3OH2.
initiation reaction generated with the double-ended GSM is shown in Figure 3. As a test of the reliability of the GSM method for calculating transition states, we have used the initial and final structures from Figure 3 as starting and ending points for a nudged elastic band (NEB) method calculation,38,39 because the NEB method is more widely used than GSM. The results of our NEB calculations are plotted in Figure S4 of the Supporting Information, where it can be seen that the results from GSM and NEB are in good agreement. This demonstrates the reliability of the GSM method for our systems. The initiation reaction proceeds by proton transfer from the H2O group to the IB to form the carbenium ion (IBH+). After the proton transfer, the newly formed carbenium ion and the OH− group form a C−O bond, as shown in the inset to Figure 3, to mitigate charge separation. As the polymerization reaction occurs very rapidly at low temperature, one would expect the initiation reaction to have a relatively low activation barrier (Ea = ETS − EIS, where TS and IS stand for transition and initial states, respectively). However, our calculated value at 0 K is Ea = 90.3 kJ/mol, which is substantially higher than what would be expected for a reaction that is rapid at low temperatures. The high Ea value calculated for this initiator complex implies that the reaction rate would be extremely small under the operating conditions. We stress that the pathway we have identified is the best pathway we
■
RESULTS AND DISCUSSION Reaction Pathway with AlCl3OH2. The putative initiator complex most commonly identified in the literature is AlCl3OH2, which consists of one molecule of AlCl3 and one molecule of H2O.2,7,37 The most energetically favorable complex of AlCl3OH2 is shown in Figure 2. The Al−O bond is a dative bond (1.96 Å), as the oxygen atom shares its lone pair of electrons with the empty orbital on the Al atom. The gas-phase standard state (298 K, 1 bar) Gibbs free energy (ΔGgas) and enthalpy of formation (ΔHgas) of the AlCl3OH2 complex relative to separated AlCl3 and H2O are −97.6 and −110.0 kJ/mol, respectively. We have identified reaction pathways for the initiation reaction of IB with AlCl3OH2 using the single-ended GSM method, refining the best pathway using double-ended GSM. The potential energy surface for the 8008
DOI: 10.1021/acscatal.8b01494 ACS Catal. 2018, 8, 8006−8013
Research Article
ACS Catalysis
energy and enthalpy of forming 1 from AlCl3OH2 and AlCl3 are ΔG g a s (AlCl 3 HOHAlCl 3 ) = −165 kJ/mol and ΔHgas(AlCl3HOHAlCl3) = −194.9 kJ/mol, respectively. This shows that 1 is thermodynamically more stable than AlCl3OH2 (ΔGgas = −97.6 kJ/mol, ΔHgas = −110.0 kJ/mol). The reaction barrier to form 1 from AlCl3OH2 + AlCl3 is 28.6 kJ/mol (see Figure S2 of the Supporting Information for the reaction pathway). The driving force to activate the strong O−H bond of water in nonpolar solvents and form 1 is partially due to the formation of a bridging OH− group held between two Lewis acid AlCl3 molecules, where the Al−O bond distances in 1 are 1.87 and 1.86 Å, showing nearly equivalent chemical bonding. The proton also becomes situated distally from the O atom between chlorine atoms from adjacent AlCl3 groups, thus resulting in a charge-neutral complex that contains a highly reactive proton (Figure 4). Refer to Figure S5 of the Supporting Information for bond distances for complex 1. The initiation reaction pathway with 1 is shown in Figure 5. The reaction proceeds as the proton migrates toward the
found after extensive testing, including many GSM calculations and AIMD simulations at various temperatures for one AlCl3 and one H2O molecule. The overall reaction is exothermic with a reaction energy (ΔErxn) value of −39.8 kJ/mol at 0 K. This large exothermic value is a result of the formation of a strong C−O bond between the IBH+ and the resulting anion. This C−O bond distance (1.51 Å) is similar to the C−O bond in tert-butyl alcohol (1.44 Å). Meanwhile, the Al−O bond remains a dative bond and is slightly shortened to 1.93 Å, relative to the Al−O bond in AlCl3OH2 of 1.96 Å. This indicates that tert-butyl alcohol is an intermediate product for this reaction pathway. As we shall see, the formation of tertbutyl alcohol impedes the polymerization reaction. For the propagation reaction to occur, the C−O bond must first be broken, resulting in an even larger reaction barrier of 217 kJ/ mol, shown in Figure S6 of the Supporting Information. This very high barrier is inconsistent with the observed reaction rate under process conditions. Given that we investigated all combinations of AlCl3 interacting with a single water molecule, we conclude that a complex consisting of only one AlCl3 molecule and one H2O molecule cannot be the correct initiator complex. It is unlikely that water clusters will form under the reaction conditions because only trace amounts of water are added. Moreover, water molecules are much more likely to bind to AlCl3 than to form clusters, because H2O−AlCl3 interactions are much more thermodynamically favorable than H2O−H2O interactions (see Table S2 of the Supporting Information). Nevertheless, we have constructed reaction pathways for the AlCl3OH2 initiation reaction with additional explicit water molecules in order to see if proton shuttling would reduce the barrier obtained in Figure 3. However, we were unable find any pathways with lower barriers (see Figures S7 and S8 and the associated discussion in the Supporting Information). Reaction Pathway with AlCl3HOHAlCl3. Having established that the putative initiator complex is not correct, we examined multiple (AlCl3)n/(H2O)m complexes, where n is the number of AlCl3 units and m is the number of water molecules in a complex (see Figures S7−S11 and Table S2 in the Supporting Information). We considered n = 1−3 and m = 1− 6. Among all the complexes we studied, we found that AlCl3HOHAlCl3 (1), as identified in Figure 4, gave the lowest barrier pathways for both initiation and propagation steps for IB polymerization. The calculated standard state Gibbs free
Figure 5. Potential energy surface for the IB initiation reaction with 1.
alkene group on the incoming IB (see the animation of this reaction in the Supporting Information). After the proton transfer to a nearby IB molecule, the resulting IBH+ remains loosely coordinated with the anion [AlCl3OHAlCl3]− to reduce charge separation. In contrast to the case involving proton transfer from AlCl3OH2, the product does not result in a strong C−O bond formation as was seen in Figure 3. This allows the IBH+ to react facilely with another IB in the propagation step. There is negligible change to Al−O bond distances after the proton transfer. The Ea value for the initiation reaction with 1 is 21.6 kJ/mol, and the overall reaction is exothermic, with ΔErxn = −47.7 kJ/mol. We have computed the standard-state gas-phase Gibbs free energy of activation, ΔG⧧ at the BP86/def2-TZVP and MP2/def2-TZVP levels of theory, obtaining activation energies of 13.0 and 10.9 kJ/mol, respectively. The low values of Ea and ΔG⧧ indicate that the initiation rate will be significant at the low temperatures of the operating conditions, consistent with the high reaction rate observed for the polymerization reaction. A key point to understanding why 1 is a viable initiation catalyst while other complexes, even those with the same chemical formula, are ineffective (see Figures S8−S10 of the
Figure 4. Identified initiator complex AlCl3HOHAlCl3 (1). 8009
DOI: 10.1021/acscatal.8b01494 ACS Catal. 2018, 8, 8006−8013
Research Article
ACS Catalysis
of this reaction mechanism can be found in the Supporting Information. The overall reaction is exothermic with ΔErxn = −52.5 kJ/mol. No experimental data for IB polymerization reaction kinetics have been reported, but it has been noted that the energy barrier for the addition of an ion pair to a monomer in a low-polarity medium should be less than 25 kJ/mol.2 Hence, our calculated propagation reaction barrier is consistent with the general rule from the literature. The second propagation pathway is shown in Figure 7. The Ea value for the second propagation step is 5.0 kJ/mol. It is
Supporting Information) is the superacidity of 1. We have computed pKa values in both aqueous and gas phases for five complexes (see Table S3 of the Supporting Information). The calculations indicate that 1 is a superacid, having an aqueous pKa of −30.1. This extremely negative value suggests that 1 would not be stable in water, but the zwitterionic state formed by compensating the energy of breaking the O−H bond from water with forming Al−O and H−Cl bonds should be feasible in nonpolar environments such as those used in the PIB reaction mixture (see the discussion in the Supporting Information). Indeed, the combination of the superacidity of 1 with the aprotic and nonpolar character of this reaction environment allows ion pair complexes to form that are energetically favorable adducts for catalytic reaction mechanisms. We note that the bridging of a proton between two chlorine atoms in 1 is very similar to the proton coordination observed in halogenated carborane superacids, which have been shown to be extremely strong Brønsted acids.40,41 The structural similarity between 1 and the known carborane catalysts is indirect support for the plausibility of the existence of our calculated structure. Thus, we found that the combination of two Lewis acids with water can form a Brønsted superacid capable of protonating IB in nonpolar environments. The propagation step was identified by starting from the product structure of the initiation reaction shown in Figure 5 and adding an IB molecule initially located at various starting positions and orientations, followed by GSM calculations. The lowest energy pathway we have identified is shown in Figure 6.
Figure 7. Potential energy surface for the second propagation reaction with 1.
interesting to note that the reaction barrier is reduced by about half with each additional IB (initiation Ea = 21.6 kJ/mol, first propagation Ea = 9.5, and second propagation Ea = 5.0 kJ/ mol). This is likely due to the charge delocalization on the carbocation as the PIB chain increases. The second propagation reaction is also exothermic (ΔErxn = −21.9 kJ/ mol). Figure 8 summarizes our proposed reaction mechanism with 1. AIMD Simulations. We performed AIMD simulations to observe the initiation reaction with various initiator complexes. We readily observed the reaction starting from 1 in a simulation at 300 K. The proton transfer took place after 5 ps of simulation time. Snapshots from the AIMD simulation are shown in Figure 9, where the actual proton transfer took place in a window of about 125 fs. The reaction in the AIMD simulation proceeds in a fashion similar to the reaction pathway generated with the GSM method, which confirms the reliability of the GSM-generated reaction pathway. The AIMD simulation demonstrated that 1 is an active initiator for the IB polymerization reaction. In contrast, simulations involving other initiator complexes, including AlCl3OH2, failed to produce a proton transfer event, at either 300 or 1500 K. This corroborates the high-energy barriers associated with breaking the O−H bond in the AlCl3OH2 complex. Polar Solvent Effects. Although this study is directed toward studying initiation and propagation reactions under industrial conditions, we note that IB polymerization is sometimes carried out in halogenated solvents at much lower temperatures (around −80 °C). Under these conditions of a high dielectric solvent and cryogenic temperatures, the active
Figure 6. Potential energy surface for the first propagation reaction starting from 1 as the initiator complex.
This reaction proceeds via the anti addition mechanism, where the IB molecule approaches the IBH+ on the opposite side of the negatively charged initiator complex. We have also identified a syn addition pathway for this complex, but the barrier is somewhat higher (see Figure S12 in the Supporting Information). The Ea value for the propagation reaction is 9.5 kJ/mol, which we believe is associated with the structural reorganization, since the gas-phase addition of IB to IBH+ is barrierless (see Figure S13). After the C−C bond is formed, the new carbocation migrates again and remains close to the anion complex to mitigate the charge separation. An animation 8010
DOI: 10.1021/acscatal.8b01494 ACS Catal. 2018, 8, 8006−8013
Research Article
ACS Catalysis
Figure 8. Summary of the proposed initiation and propagation mechanisms with 1..
Figure 10. Effect of solvation on the initiation reactions: initiation reaction (a) with AlCl3OH2 and (b) with complex 1.
13.4 kJ/mol, indicating that 1 is likely to be an active catalyst in polar solvents as well. We note that ΔErxn for 1 decreases from −47.7 to −91.3 kJ/mol in dichloromethane (Figure 10b), whereas there is almost no change in ΔErxn for AlCl3OH2 (Figure 10a). This is because the products of the reaction with 1, IBH+, and [AlCl3OHAlCl3]− (Figure 5) have more charge separation in comparison to the product of the initiation reaction with AlCl3OH2 (Figure 3).
Figure 9. AIMD simulation snapshots showing the progression of the initiation reaction of IB with 1 from Figure 4. The proton of interest is shown in blue for clarity, and nonreacting IB molecules are shown as stick models.
■
CONCLUSION We have presented the first successful molecular-level study of catalyzed IB polymerization. Our calculations overturn the widespread view that the AlCl3OH2 complex is an effective proton donor and is the active catalyst, because both the initiation and propagation steps have excessively high barriers. We used the GSM method to identify the AlCl3-based initiator complex 1 (Figure 4), which has a low-energy-barrier pathway for initiation and which is an active catalyst for the propagation reaction. The most important feature of 1 is that it is a chargeneutral complex with a superacidic proton that is suitable for proton transfer in nonpolar solvents. The proton in 1 is labile because proton transfer results in the stable ion pair [AlCl3OHAlCl3−][IBH+], which can then seed a growing polymer chain. The presence of similar ion pairs in this reaction then facilitate low reaction barriers for the propagation steps. The low reaction barriers for the initiation and propagation reactions with 1 are consistent with experimental conditions (at approximately 10 °C in pressurized reactors) under which the polymerization reaction is carried out. We conclude that a complex containing at least
catalyst and reaction pathways may be different from those we have identified above. To explore the effect of using a polar solvent, we carried out calculations using polar and aprotic dichloromethane as the solvent (ε = 9.08). Solvation effects were approximated with the conductor-like screening (COSMO) continuum solvation model42 as implemented in ORCA. Results from our calculations for the initiation reactions with AlCl3OH2 and 1 are shown in Figure 10. We see that the reaction barriers for both catalysts are lower in the polar solvent than in the nonpolar solvent. This is because the polar solvent better stabilizes the charge separation during the proton transfer in comparison to the nonpolar solvent. Consequently, the Ea value for the initiation reaction with AlCl3OH2 decreases from 90.3 to 64.4 kJ/mol. This barrier is still very high in comparison with the reaction temperature, meaning that our conclusion that AlCl3OH2 is not a viable catalyst for IB polymerization under industrial conditions remains valid for polymerization in halogenated solvents at low temperatures. The reaction barrier for 1 decreases from 21.6 to 8011
DOI: 10.1021/acscatal.8b01494 ACS Catal. 2018, 8, 8006−8013
Research Article
ACS Catalysis
(7) Kennedy, J. P.; Gillham, J. K. In Fortschritte Der HochpolymerenForschung; Springer: 1972; pp 1−33. (8) Kennedy, J. P. Novel Sequential Copolymers by Elucidating the Mechanism of Initiation and Termination of Carbocationic Polymerizations. J. Polym. Sci., Polym. Symp. 1976, 56, 1−11. (9) Kennedy, J.; Squires, R. Fundamental Studies on Cationic Polymerization IvHomo-and Co-Polymerizations with Various Catalysts. Polymer 1965, 6, 579−587. (10) Chaffey-Millar, H.; Kühn, F. E. Quantum Chemical Insights into the Initiation Mechanism of Transition Metal Catalysed Polymerisation of Isobutene. Appl. Catal., A 2010, 384, 154−164. (11) Neese, F. The Orca Program System. WIREs Comput. Mol. Sci. 2012, 2, 73−78. (12) Perdew, J. P. Density-Functional Approximation for the Correlation Energy of the Inhomogeneous Electron Gas. Phys. Rev. B: Condens. Matter Mater. Phys. 1986, 33, 8822. (13) Weigend, F.; Ahlrichs, R. Balanced Basis Sets of Split Valence, Triple Zeta Valence and Quadruple Zeta Valence Quality for H to Rn: Design and Assessment of Accuracy. Phys. Chem. Chem. Phys. 2005, 7, 3297−3305. (14) Izsák, R.; Neese, F. An Overlap Fitted Chain of Spheres Exchange Method. J. Chem. Phys. 2011, 135, 144105. (15) Grimme, S.; Furche, F.; Ahlrichs, R. An Improved Method for Density Functional Calculations of the Frequency-Dependent Optical Rotation. Chem. Phys. Lett. 2002, 361, 321−328. (16) Ribeiro, R. F.; Marenich, A. V.; Cramer, C. J.; Truhlar, D. G. Use of Solution-Phase Vibrational Frequencies in Continuum Models for the Free Energy of Solvation. J. Phys. Chem. B 2011, 115, 14556− 14562. (17) William, J. Process for the Production of Polyisobutylene. U.S. Patent 2,957,930, October 25, 1960. (18) Groenenboom, M. C.; Keith, J. A. Quantum Chemical Analyses of Bh4− and Bh3oh− Hydride Transfers to Co2 in Aqueous Solution with Potentials of Mean Force. ChemPhysChem 2017, 18, 3148−3152. (19) Zimmerman, P. Reliable Transition State Searches Integrated with the Growing String Method. J. Chem. Theory Comput. 2013, 9, 3043−3050. (20) Zimmerman, P. M. Single-Ended Transition State Finding with the Growing String Method. J. Comput. Chem. 2015, 36, 601−11. (21) Zimmerman, P. M. Growing String Method with Interpolation and Optimization in Internal Coordinates: Method and Examples. J. Chem. Phys. 2013, 138, 184102. (22) Peters, B.; Heyden, A.; Bell, A. T.; Chakraborty, A. A Growing String Method for Determining Transition States: Comparison to the Nudged Elastic Band and String Methods. J. Chem. Phys. 2004, 120, 7877−7886. (23) Sun, Z.; Winschel, G. A.; Zimmerman, P. M.; Nagorny, P. Enantioselective Synthesis of Piperidines through the Formation of Chiral Mixed Phosphoric Acid Acetals: Experimental and Theoretical Studies. Angew. Chem., Int. Ed. 2014, 53, 11194−11198. (24) Pendleton, I. M.; Pérez-Temprano, M. n. H.; Sanford, M. S.; Zimmerman, P. M. Experimental and Computational Assessment of Reactivity and Mechanism in C (Sp3)−N Bond-Forming Reductive Elimination from Palladium (Iv). J. Am. Chem. Soc. 2016, 138, 6049− 6060. (25) Behn, A.; Zimmerman, P. M.; Bell, A. T.; Head-Gordon, M. Efficient Exploration of Reaction Paths Via a Freezing String Method. J. Chem. Phys. 2011, 135, 224108. (26) Behn, A.; Zimmerman, P. M.; Bell, A. T.; Head-Gordon, M. Incorporating Linear Synchronous Transit Interpolation into the Growing String Method: Algorithm and Applications. J. Chem. Theory Comput. 2011, 7, 4019−4025. (27) Gomes, J.; Head-Gordon, M.; Bell, A. T. Reaction Dynamics of Zeolite-Catalyzed Alkene Methylation by Methanol. J. Phys. Chem. C 2014, 118, 21409−21419. (28) Hariharan, P. C.; Pople, J. A. The Influence of Polarization Functions on Molecular Orbital Hydrogenation Energies. Theor. Chim. Acta 1973, 28, 213−222.
two AlCl3 groups and one H2O molecule is necessary to initiate the reaction. We have investigated the initiation and propagation reactions for polymerization of IB assuming a homogeneous reaction. However, in practice, micrometer-size particles of AlCl3 are suspended in the IB mixture in the production of PIB,17 which means that the actual reaction mechanism may actually be heterogeneous. Given that the low-energy surface of AlCl3 is chemically inert,43 we postulate that defective surfaces of AlCl3 particles facilitate the formation of surface sites that resemble 1, which can provide low-barrier initiation pathways. Investigation of surface-mediated reactions is beyond the scope of this work. We note that our observations about the need for a superacid to initiate proton transfer in nonpolar environments are consistent with recent work on cationic polymerization using heteropolyacids as catalysts.6
■
ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acscatal.8b01494. Reaction and activation energies at different levels of theory, bond distances, reaction pathways of other initiator complexes, NEB calculations, binding energies of water to AlCl3 clusters, calculated pKa values, AIMD simulation summary, and complex coordinates (PDF) Reaction mechanism animation (AVI) Reaction mechanism animation (AVI)
■
AUTHOR INFORMATION
Corresponding Author
*E-mail for J.K.J.:
[email protected]. ORCID
John A. Keith: 0000-0002-6583-6322 J. Karl Johnson: 0000-0002-3608-8003 Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS We thank Paul Zimmerman for helpful discussions regarding the Growing String Method, Giannis Mpourmpakis for fruitful discussion on superacidity, and Abhishek Bagusetty for help with the NEB calculations. We acknowledge the Center for Research Computing at the University of Pittsburgh for computing resources and technical support. This work was funded by the Lubrizol Corporation.
■
REFERENCES
(1) Kennedy, J. P.; Rengachary, S. In Fortschritte Der HochpolymerenForschung; Springer: 1974; pp 1−48. (2) Odian, G. Principles of Polymerization, 4th ed.; Wiley: Hoboken, NJ, 2004; p 372. (3) Kostjuk, S. V. Recent Progress in the Lewis Acid Co-Initiated Cationic Polymerization of Isobutylene and 1, 3-Dienes. RSC Adv. 2015, 5, 13125−13144. (4) Dimitrov, P.; Emert, J.; Faust, R. Polymerization of Isobutylene by Alcl3/Ether Complexes in Nonpolar Solvent. Macromolecules 2012, 45, 3318−3325. (5) Evans, A. G.; Meadows, G. W. The Polymerization of Isobutene by Boron Trifluoride. Trans. Faraday Soc. 1950, 46, 327−331. (6) Burrington, J. D.; Johnson, J. R.; Pudelski, J. K. Cationic Polymerization Using Heteropolyacid Salt Catalysts. Top. Catal. 2003, 23, 175−181. 8012
DOI: 10.1021/acscatal.8b01494 ACS Catal. 2018, 8, 8006−8013
Research Article
ACS Catalysis (29) Francl, M. M.; Pietro, W. J.; Hehre, W. J.; Binkley, J. S.; Gordon, M. S.; DeFrees, D. J.; Pople, J. A. Self-Consistent Molecular Orbital Methods. Xxiii. A Polarization-Type Basis Set for Second-Row Elements. J. Chem. Phys. 1982, 77, 3654−3665. (30) Hutter, J.; Iannuzzi, M.; Schiffmann, F.; VandeVondele, J. Cp2k: Atomistic Simulations of Condensed Matter Systems. WIREs Comput. Mol. Sci. 2014, 4, 15−25. (31) VandeVondele, J.; Krack, M.; Mohamed, F.; Parrinello, M.; Chassaing, T.; Hutter, J. Quickstep: Fast and Accurate Density Functional Calculations Using a Mixed Gaussian and Plane Waves Approach. Comput. Phys. Commun. 2005, 167, 103−128. (32) Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized Gradient Approximation Made Simple. Phys. Rev. Lett. 1996, 77, 3865. (33) Goedecker, S.; Teter, M.; Hutter, J. Separable Dual-Space Gaussian Pseudopotentials. Phys. Rev. B: Condens. Matter Mater. Phys. 1996, 54, 1703. (34) Krack, M. Pseudopotentials for H to Kr Optimized for Gradient-Corrected Exchange-Correlation Functionals. Theor. Chem. Acc. 2005, 114, 145−152. (35) VandeVondele, J.; Hutter, J. Gaussian Basis Sets for Accurate Calculations on Molecular Systems in Gas and Condensed Phases. J. Chem. Phys. 2007, 127, 114105. (36) Nosé, S. A Unified Formulation of the Constant Temperature Molecular Dynamics Methods. J. Chem. Phys. 1984, 81, 511−519. (37) Pepper, D. Ionic Polymerisation. Q. Rev., Chem. Soc. 1954, 8, 88−121. (38) Henkelman, G.; Uberuaga, B. P.; Jónsson, H. A Climbing Image Nudged Elastic Band Method for Finding Saddle Points and Minimum Energy Paths. J. Chem. Phys. 2000, 113, 9901−9904. (39) Henkelman, G.; Jónsson, H. Improved Tangent Estimate in the Nudged Elastic Band Method for Finding Minimum Energy Paths and Saddle Points. J. Chem. Phys. 2000, 113, 9978−9985. (40) Kostetskyy, P.; Zervoudis, N. A.; Mpourmpakis, G. Carboranes: The Strongest Brønsted Acids in Alcohol Dehydration. Catal. Sci. Technol. 2017, 7, 2001−2011. (41) Nava, M.; Stoyanova, I. V.; Cummings, S.; Stoyanov, E. S.; Reed, C. A. The Strongest Brønsted Acid: Protonation of Alkanes by H (Chb11f11) at Room Temperature. Angew. Chem., Int. Ed. 2014, 53, 1131−1134. (42) Klamt, A.; Schüürmann, G. Cosmo: A New Approach to Dielectric Screening in Solvents with Explicit Expressions for the Screening Energy and Its Gradient. J. Chem. Soc., Perkin Trans. 2 1993, 2, 799−805. (43) Murthy, J. K.; Gross, U.; Rüdiger, S.; Rao, V. V.; Kumar, V. V.; Wander, A.; Bailey, C.; Harrison, N.; Kemnitz, E. Aluminum Chloride as a Solid Is Not a Strong Lewis Acid. J. Phys. Chem. B 2006, 110, 8314−8319.
8013
DOI: 10.1021/acscatal.8b01494 ACS Catal. 2018, 8, 8006−8013
