Design Principles for Rational Polyurethane Catalyst Development

May 24, 2019 - Design Principles for Rational Polyurethane Catalyst Development ... conformational energy contributions in the calculated activation f...
0 downloads 0 Views 1MB Size
Article Cite This: J. Org. Chem. 2019, 84, 8202−8209

pubs.acs.org/joc

Design Principles for Rational Polyurethane Catalyst Development Mikko Muuronen,*,† Peter Deglmann,† and Ž eljko Tomović*,‡ †

BASF SE, Carl-Bosch-Str. 38, 67056 Ludwigshafen am Rhein, Germany BASF Polyurethanes GmbH, Elastogranstr. 60, 49448 Lemfoerde, Germany



Downloaded via GUILFORD COLG on July 17, 2019 at 16:46:52 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

S Supporting Information *

ABSTRACT: Tertiary amine catalysts are essential components in manufacturing polyurethane materials. The low-emission requirements for indoor applications are typically achieved by employing tertiary amines with catalytically active N,N-dimethyl groups as the base catalyst and a longer alkyl substituent with a reactive end, that is, alcohol or amine, to incorporate it in the polyurethane matrix. N,N-dimethyl groups are, however, oxidized when exposed to air and lead to undesired formaldehyde emissions. Here, we employ modern quantum chemical methods to understand design principles how the structure of tertiary amine catalysts having N,N-dimethyl groups can be modified to avoid this source of formaldehyde formation but still preserve their catalytic activity. We found the pyrrolidine derivative of commonly used N,N-dimethylated catalysts to be the most promising candidate and developed design principles to rationalize why longer alkyl chains or larger ring sizes inhibit the catalytic activity. The computationally predicted catalyst performances were confirmed experimentally in model polyurethane systems for selected amine catalysts, and emission measurements showed that the formaldehyde emission was completely suppressed when pyrrolidine derivative was used as a catalyst. Our results further illustrate how condensed phase reactions can be predicted using quantum chemical methods and that to account for steric hindrance near the reaction center, it was also necessary to include conformational energy contributions in the calculated activation free energies.

1. INTRODUCTION Polyurethane (PU) is one of the most versatile polymers that appears in an astonishing variety of forms and plays a vital role in many industries, from furniture to footwear and construction to automotive.1−3 PUs show thermoplastic, elastomeric, and thermoset behavior depending on their chemical and morphological structure, and their properties can be easily tailored in an extremely wide range and finetuned for numerous applications by changing the type and functionality of the polyol and isocyanate precursors. Because of their versatility and diverse utility, PUs account for 5 wt % of the total worldwide polymer production and around 16 million tons of PU raw materials are currently produced annually.2 Tertiary amines are essential components in successful manufacturing of PU materials. Urethane formation is typically catalyzed by tertiary amines, such as 1,4-diazabicyclo[2.2.2]octane (DABCO) and 2,2′-bis-(dimethylaminoethylether), or amidines such as 1,8-diazabycyclo[5.4.0]undec-7-ene.4−6 These are, however, compounds with noticeable vapor pressure and, therefore, they contribute significantly to volatile organic compound (VOC) emissions and associated odor. To © 2019 American Chemical Society

overcome the emissions problem, a wide variety of nonvolatile amines containing hydroxyl or reactive amino functional groups as a reaction site with isocyanates have been used, for example, N,N-dimethylethanolamine; N,N-dimethylaminopropylamine; and N,N,N′-trimethyl-N′-hydroxyethyl-bisaminoethylether, see Scheme 1. These so-called reactive catalysts are incorporated into the polymer network and, therefore, improve air quality, fogging, and odor problems. Consequently, there has been a transition from the use of nonreactive catalysts, for example, DABCO, to reactive catalysts for lowering the amine emissions in many industrial applications.7 However, the typically used reactive catalysts in the PU synthesis contain Nmethyl groups and thus lead to undesirable degradation products, such as formaldehyde, when exposed to air.8−10 This is a critical drawback for the use of reactive catalysts as formaldehyde is an indoor air pollutant that can cause respiratory irritation and cancer after long-term exposure.11 The International Agency for Research on Cancer (IARC) Received: May 17, 2019 Published: May 24, 2019 8202

DOI: 10.1021/acs.joc.9b01319 J. Org. Chem. 2019, 84, 8202−8209

Article

The Journal of Organic Chemistry

Scheme 1. (a) Schematic Presentation of Urethane Reaction; (b) Examples of Commonly Used Reactive Catalysts in PU Synthesis

catalysts 1−3, where N,N-dimethyl groups are substituted with diethyl and dipropyl groups, while catalysts 5−7 were studied in order to understand the catalytic activity of cyclic tertiary amines, that is, pyrrolidine (5), piperidine (6), and morpholine (7). Functionalized imidazole (4) was also studied to understand the catalytic activity of aromatic imines. All catalyst models were further functionalized with a propyl group to mimic a longer alkane chain with reactive hydroxyl or amine group as shown in Scheme 1b. Earlier study showed that measured rate constants of amine-catalyzed urethane reaction in acetonitrile correlate linearly with the amine’s basicity.12 The correlation only exists, however, for sterically similar tertiary amine catalysts and in nonprotic media.12 Therefore, our study focuses on understanding the catalytic activity in a real PU system, where the protic environment may interfere with the relative reactivities of different catalysts. As first step of our computational screening of catalysts, we studied the complete reaction mechanism for the N,Ndimethylaminopropane (1)-catalyzed urethane formation. Our computational protocol used p-methylphenylisocyanate as a model for aromatic isocyanates and 2-methoxyethan-1-ol as a model for polyol with reactive primary hydroxyl groups, for example, Lupranol 2090 used in the experiments later. In order to accurately predict the relative activation free energies in solution, that is, in PU media, for different catalysts, our computational approach relies on extensive conformational sampling and optimization of all structures using dispersion corrected density functional BP86-D313−15 with triple-ζ level def2-TZVP basis sets,16 accurately including solvation using the COSMO-RS17 model as described elsewhere,18 and calculating final energies using high-level random phase approximation (RPA)19 and MP2 methods with large quadruple-ζ level def2-QZVPP basis sets.16 RPA and MP2 were selected because they capture long-range interactions from first principles without empirical parameters and are thus expected to systematically describe the energy differences between different catalysts.20,21 RPA benefits over MP2 further by capturing also the non-pairwise-additive nature of longrange interactions, and the accuracy of it can also be increased by improving it with the approximate exchange kernel (AXK)22 method to approach the accuracy of the “gold standard” CCSD(T) method for organic molecules.23 All discussed data represent free energies in PU media at 60 °C, chosen as an average temperature for reaction typically run as an adiabatic reaction in a wide temperature range. Because of the heterogeneity of the PU matrix and possibility for

classified formaldehyde as cancerogenic to humans, which led to stricter regulations on the emissions of formaldehyde. The recommended limit for formaldehyde in residential indoor air was established at 0.1 mg/m3 or about 0.08 ppm by the World Health Organization (WHO).11 The stringent requirements for indoor air quality have made VOC and formaldehyde emissions increasingly important to the PU industry. Therefore, understanding the rules for PU catalyst design and employing them to develop amine catalysts that exhibit similar or higher activity in urethane formation as N,N-dimethylcontaining catalysts, but are less prone to formaldehyde emission, is highly desired. In this work, we focus on understanding which structural modifications are necessary to avoid formaldehyde emissions but to preserve the catalytic activity of the tertiary amine catalyst in urethane formation, that is, amine-catalyzed addition reaction between alcohol and isocyanate as depicted in Scheme 1a. To achieve this, we explore how structural modification of the N-methyl substituents of commonly used PU catalysts affects their catalytic activity in real PU systems. The initial catalyst screening was performed using modern quantum chemical methods to predict how the activation free energies of the urethane reaction are affected when the methyl groups of the Me2N−R catalyst are replaced by longer alkyl chains or cyclic structures. Finally, the promising candidates were studied experimentally in a model PU reaction and the emissions were measured for selected catalysts.

2. RESULTS AND DISCUSSION We chose seven amines to be screened computationally regarding their catalytic activity, see Figure 1. The studied

Figure 1. Studied amine catalysts.

catalysts were chosen to provide understanding and principles for future catalyst design by modifying their electronic and steric properties: the effect of increasing chain length of alkyl group of tertiary amine catalysts was studied by comparing

Scheme 2. Generally Accepted Reaction Mechanism of the Tertiary Amine Catalyzed Addition Reaction between Alcohol and Aromatic Isocyanates

8203

DOI: 10.1021/acs.joc.9b01319 J. Org. Chem. 2019, 84, 8202−8209

Article

The Journal of Organic Chemistry

Figure 2. Reaction profile for N,N-dimethylaminopropane-catalyzed urethane reaction. Relative Gibbs free energies were computed at the RPA/ def2-QZVPP//BP86-D3/def2-TZVP level of theory.

product, see Figure 2. In the model with the polyol−water mixture medium, our computations predict formation of the amine−alcohol and amine−water precomplexes to be endergonic by 5−9 kJ/mol for N,N-dimethylpropane catalyst at the RPA level. This indicates that tertiary amine catalysts equilibrate with the reaction medium and the activation free energy of the reaction is defined as the free energy difference between separated reactants and the transition state. One should point out here that hydrogen bonding interactions between solvent and solute are taken into account within the used COSMO-RS model by an explicit term in the interaction potential so that differences in hydrogen-bonding affinities for different catalysts should be reproduced in this protocol. In agreement with previous studies,24,25 the transition state for urethane formation is concerted in which the catalyst activates the alcohol via hydrogen-bonding and transfers the proton to the urethane nitrogen spontaneously after the transition state according to dynamic reaction coordinate calculation, which follows the reaction path from transition state to both minima conserving the total energy,26,27 see Figure 2. In the polyol−water mixture at RPA level, the activation free energy is predicted to be 83.7 kJ/mol and the reaction free energy to be −24.8 kJ/mol, having the exergonic character (negative value) and indicating that the equilibrium is on the product side. Additionally, catalyst dissociation from

interfacial reactions, all the free energies were calculated in two separate media to estimate to what extent obtained results depend on specific environments: (1) in a mixture of polyol and water and (2) in a mixture of polyol and 4,4methylenediphenylisocyanate. For more information, see Experimental Section and Supporting Information. 2.1. Mechanism of Urethane Formation. The mechanism of the base-catalyzed urethane reaction is generally accepted, that is, a tertiary amine activates the alcohol for nucleophilic attack via hydrogen bonding and forms the urethane after transferring the alcohol proton, see Scheme 2. This is supported by kinetic studies, indicating that the tertiary amine catalyst is only partially protonated in the transition state,12 and by earlier computational studies, suggesting that the ammonium cation is formed after the transition state and that it spontaneously transfers the proton to the benzylic nitrogen to form the urethane.24,25 A mechanism where the amine catalyst acts as a nucleophile can represent an option for strong Brønsted base catalysts such as cyclic guanidines6 but is not considered realistic for the tertiary amines studied here. According to the generally accepted mechanism, our mechanistic scenario considers the formation of amine− alcohol and amine−water precomplexes from solvated species, formation of the transition state with the subsequent protontransfer step, and dissociation of the catalyst from the urethane 8204

DOI: 10.1021/acs.joc.9b01319 J. Org. Chem. 2019, 84, 8202−8209

Article

The Journal of Organic Chemistry

Table 1. Activation Free Energies (ΔG⧧) in kJ/mol and Relative Rate Constants (krelative) for Catalysts 1−7 at the RPA Level with and without the Contribution of Conformational Entropy (ΔGconf)a

the hydrogen-bonded urethane complex is exergonic by approximately −12 kJ/mol, and thus catalyst turn over frequency is not limited by strong interactions between the catalyst and urethane product either. A change of the reaction medium is not expected to drastically change the relative free energies in the reaction profile. In less protic environments, however, a faster reaction is expected, that is, the predicted activation free energy in water-free polyol−MDI mixture is 81.8 kJ/mol, which corresponds to 2-fold increase in the reaction rate obtained from the Eyring equation and was also experimentally observed as discussed later. The reaction mechanism is the same at the MP2 level, which predicts a slightly more exergonic formation of the amine−alcohol precomplex (−1.7 kJ/mol) and a lower activation free energy of 69.8 kJ/mol. Activation free energies calculated at RPA level are, however, expected to be more accurate: MP2 is known to overestimate the stability of dispersion-dominated unsaturated complexes28−30 and, therefore, predicts too low activation energies here where attractive weak interactions dominate the transition state but are not present in isolated reactants. To confirm the accuracy of the RPA method, we also computed the activation free energy barrier for catalyst 1 by employing the AXK method, which should increase the accuracy of RPA in organic reactions.22,23 RI-AXK predicts an activation free energy of 86.4 kJ/mol, which is 3 and 16 kJ/mol higher than predicted by RPA and MP2, respectively. The relative activation free energies for different catalysts are nevertheless expected to be more accurate due to systematic cancellation of errors in similar transition states. For all results in different media and with both levels of computation, see Tables S1 and S2 in Supporting Information. 2.2. Computational Screening of Catalysts. The reaction mechanism discussed above was applied for screening of all catalysts in Figure 1, that is, all water−amine and alcohol−amine precomplexes, and all transition state structures were optimized and studied. At RPA level and in both media, the precomplex formation is predicted to be endergonic for all catalysts, and at MP2 level, only catalyst 1 was predicted to form a weak alcohol−amine precomplex, see Tables S1 and S2 (Supporting Information). Because of the discovered large differences in the activation free energies between RPA and MP2 levels for catalyst 1, all discussed activation free energies in the text are at RPA level in PU media with 3 wt % of water. First, we compared catalysts 1−3 to identify how the tertiary amine activities are affected by substituting the N-methyl groups (1) with N-ethyl (2) and N-propyl (3) groups, see Table 1 and Figure 3. The predicted activation free energies for catalysts 1, 2, and 3 are 83.7, 86.0, and 84.5 kJ/mol, when only the transition state structures with the lowest energies are considered. These data indicate that the activity of the catalyst is not much affected by modifying the alkyl chain and can be rationalized with sterically nonconstrained transition states for all catalysts, see Figure 3. These estimates, however, do not account for the change of conformational entropy between the isolated reactants and the transition state, that is, different amount of easily accessible conformers in the corresponding structures, which is required to accurately model how increasing length of the alkyl chain affects the activity of the amine catalyst.18,31 The effect of the conformational entropy on the activation free energy (ΔGconf) was calculated according to Deglmann et al.18 by optimizing the full conformational space at BP86-D3/def2-TZVP level for the degrees of freedom

without ΔGconf

a

with ΔGconf

catalyst

ΔG⧧

krelative

ΔG⧧

krelative

1 2 3 4 5 6 7

83.7 86.0 84.5 94.1 86.1 87.6 92.4

100 42 75 2 42 24 4

86.0 90.9 91.1 92.2 85.6 88.2 93.6

100 17 16 11 116 45 7

All energies are in kJ/mol.

Figure 3. Optimized transition states for catalysts 1−3. All geometries were optimized at the BP86-D3/def2-TZVP/COSMO(ε = ∞) level.

that rotate more freely in isolated species compared in the transition states. The comparison was made relative to catalyst 1 and the bonds included in the full conformational sampling are shown in blue in Figure 4: these rotations account for the

Figure 4. Schematic presentation of the transition state: blue bonds represent the degrees of freedom included in the full conformational sampling to compute the effect of conformational entropy on the activation free energies.

degrees of freedom that are most affected between studied catalysts from isolated reactants to the transition state. For more comprehensive discussion about methods used to calculate ΔGconf, see Supporting Information. When conformational entropies are included in the studied activation free energies (ΔGconf), the predicted activation free energies are 86.0, 90.9, and 91.1 kJ/mol for catalysts 1, 2, and 3, respectively, see Table 1. According to the Eyring equation, these correspond to relative rate-constants of 100, 17, and 16 for catalysts 1, 2, and 3 and indicate strongly that substitution of N-methyl groups with longer alkyl-chain leads to a decreased catalytic activity. The origin of this is the larger loss of conformational entropy for longer alkyl chain in the transition state, that is, less N-alkyl rotamers are energetically 8205

DOI: 10.1021/acs.joc.9b01319 J. Org. Chem. 2019, 84, 8202−8209

Article

The Journal of Organic Chemistry

energies or not; this points out the importance of this often neglected contribution which should maybe be included more frequently, especially when alkyl substituents close to the reactive center exhibit structural differences. In summary, the systematic trends indicate catalysts with shorter alkyl chains to be more active than with longer ones and cyclic catalysts with smaller rings to be more active compared to larger ones according to Bent’s rule. RPA and MP2 levels both predict that cyclic amines such as pyrrolidine can exhibit higher activity than N-methyl containing tertiary amines. RPA predicts relative catalyst activities according to 5 (krelative = 116) > 1 (krelative = 100) > 6 (krelative = 45) > 2 (krelative = 17) > 3 (krelative = 16) > 4 (krelative = 11) > 7 (krelative = 7) in PU media with 3 wt % of water, and solvation is not expected to have a big effect (Table S1). The high activity of pyrrolidine catalyst is also supported by experimental studies showing good catalytic activity of methylpyrrolidine in urethane formation in acetonitrile,12 as well as by preliminary reports on activity of pyrrolidine derivatives in PU systems.33−35 2.3. Experimental Screening of Catalysts. To validate our computational predictions and identify a catalyst with similar activity to 1, we selected structures 1, 2, 4, 5, and 6 for further investigation in our experimental study, see Figure 6. For the experimentally studied catalysts, the propyl group was

accessible than for the free amine; obviously, this effect significantly helps to disfavor larger alkyl groups even more than just steric repulsion in the one best transition state conformer. The activation free energies obtained at the MP2 level are consistently lower for all catalysts compared to the RPA level, but both methods agree on the relative activity differences between the catalysts, see Table S2. Next, we studied the cyclic tertiary amines 5−7 and the imidazole derivative 4, see Table 1 and Figure 5. The predicted

Figure 5. Optimized transition states for catalysts 4−6. All geometries were optimized at the BP86-D3/def2-TZVP/COSMO(ε = ∞) level.

reactivity of the cyclic tertiary amines seems to follow Bent’s rule:32 Five-membered heterocycle pyrrolidine (5) shows the highest reactivity among cyclic amines with an activation free energy of 86.1 kJ/mol when contribution of conformational entropy is not included. Slightly lower activity is predicted for the six-membered piperidine catalyst (6, 87.6 kJ/mol) and the lowest for the morpholine derivate (7, 92.4 kJ/mol). These data can be rationalized purely with the increase of the electronic inductive effect of the heterocycle, that is, increase of the ring size (5 → 6) affects the electronics of the nitrogen lowering the basicity of the lone electron pair required for catalysis. Most strongly this is observed by comparing catalysts 6 and 7 where the electronegativity of the ring differs without changing any sterics. The increased sterics are further expected to affect the amount of populated transition state conformers, and therefore conformational entropy was also included in the activation free energies as previously described. This leads to stabilization of transition state for catalyst 5 (85.6 kJ/mol), whereas transition states for larger catalysts 6 (88.2 kJ/mol) and 7 (93.6 kJ/mol) are destabilized. The imidazole derivative 4 with sp2 hybridized lone electron pair is predicted to be the least-active catalyst in the study with an activation free energy of 94.1 kJ/mol when ΔGconf is not included, but however, the contribution of conformational entropy lowers the activation free energy of it to 92.2 kJ/mol, and therefore, it is predicted to be more active catalyst than the morpholine derivate 7. For further discussion about the negative sign of ΔGconf for catalysts 4 and 5, see Supporting Information. Overall, the predictions indicate that pyrrolidine derivative 5 should exhibit slightly higher catalytic activity than N,Ndimethyl derivative 1 and larger cycles to be less active: the relative rate-constants for catalysts 1, 4, 5, 6, and 7 are 100, 11, 116, 45, and 7 at the RPA level when conformational entropy is included, see Table 1. The results obtained at the MP2 level are consistent with RPA for the relative differences between cyclic catalysts (Table S2).The relative reactivities differ, however, even qualitatively depending whether the effect of conformational entropy is included in the activation free

Figure 6. IR kinetic measurements: (a) depletion of NCO during the reaction of Lupranol 2090 and Lupranat MM 103 with 0.5 wt % of water and (b) without water at 60 °C, in the presence of equimolar amount of 1, 2, 4, 5, and 6 as catalysts. The first IR measurement was performed 15 s after the beginning of stirring of reaction mixture. 8206

DOI: 10.1021/acs.joc.9b01319 J. Org. Chem. 2019, 84, 8202−8209

Article

The Journal of Organic Chemistry

increased ring size and by inclusion of electronegative atoms or substituents in the ring. The predicted relative activities of different catalysts were confirmed experimentally, showing the potential of modern quantum chemical methods in predicting catalytic activity in real PU foam system even when the relative activation free energies are distributed inside few kJ/mol. These further highlight the importance of including conformational entropy contribution in the computed activation free energies especially when comparing catalysts with different alkyl substituents attached to the reactive center as it was shown here to affect even the qualitative trends between different catalysts. Finally, pyrrolidine derivatives are proposed as alternative tertiary amine catalysts for PU production with high catalytic activity and without being a source of formaldehyde emissions because of their optimal steric and electronic parameters.

replaced with aminopropyl to form a reactive catalyst that is incorporated in the PU matrix. As a model reaction, bulk polymerization of trifunctional polyether polyol Lupranol 2090, containing reactive primary hydroxyl groups, water, and Lupranat MM 103, a carbodiimide-modified diphenylmethane-4,4′-diisocyanate (MDI) with average functionality 2,2, was investigated. Lupranol 2090 and Lupranat MM 103 are used on industrial scale to produce various PUs such as microcellular elastomers, flexible and semi-rigid foams, as well as for compact PU parts used, among others, in many different indoor applications. 3-(Dimethylamino)propylamine was used as a reference catalyst according to catalyst 1 used in the modeling, and all studied catalysts were used in equimolar amounts. The reaction progress was estimated in IR kinetic studies by measuring the disappearance of the NCO stretching vibration at 2271 cm−1 as a function of time.36 Quantification was carried out relying on the unchanged absorption of CH2 groups at 2865 cm−1, which were used as an internal standard in IR measurements. The relative ordering of the catalysts in the kinetic experiment is according to the predicted values at the RPA level: pyrrolidine derivative (5) shows increased reactivity compared to N,N-dimethyl derivative (1), piperidine derivative (6) and N,N-diethyl derivative (2) show a similar reactivity, and imidazole derivative (4) is the slowest of the studied catalysts. We also performed identical kinetic experiments at 60 °C without water: the relative catalytic activities of N,N-dimethyl derivative (1), N,N-diethyl derivative (2), and piperidine derivative (6) are slightly closer to each other than when compared in protic media, but the ordering was not affected, see Figure 6. The amine catalysts are also more active in nonprotic environment as was also computationally predicted. To address the temperature dependence of relative catalytic activities, the corresponding kinetics were also measured at 120 °C without water (see Figure S1 in the Supporting Information). The relative ordering of the catalysts is not changed in these experiments, and therefore, it can be considered rather independent of the changes in the PU environment and the computated activation free energies to be valid for a wide range of PU formulations. Finally, we applied the same model reaction to prepare molded PU foams having density 360 g/L with equimolar amounts of reactive catalysts 1 and 5, which are incorporated in the PU, and measured their formaldehyde emissions using the chamber method described in Experimental Section. The measured formaldehyde emission from the foam prepared with catalyst 1 is, as expected, very high, 244 μg/m3. The formaldehyde emission is completely avoided (below detection limit of 11 μg/m3) by using the cyclic amine catalyst based on pyrrolidine ring (5), which makes this catalyst suitable for use in the production of PU materials for indoor applications.

4. EXPERIMENTAL SECTION 4.1. Computational Details. All computations were performed using Turbomole 7.2.1 program package.37 Structures were optimized at the density functional theory (DFT) level using the BP8613 functional with def2-TZVP basis sets,16,38 combined with the BJdamped D3-dispersion correction, denotedD3 in the text.14,15 DFT computations were sped-up by employing multipole-accelerated resolution-of-the-identity approximation for Coulomb term (MARIJ)39 with the corresponding auxiliary basis sets.40 Resolution-ofidentity (RI) RPA41 and RI-MP242 calculations were performed with def2-QZVPP basis sets and with corresponding auxiliary basis sets.43,44 For RPA, gas-phase TPSS45 orbitals were used, and the core orbitals were kept frozen for computation of correlation energy in RPA and MP2 computations. Solvation effects were accounted for during structure optimizations using the COSMO solvation model with a dielectric constant of infinity.46,47 Solvation free energies were calculated at 60 °C using the COSMO-RS17 model in COSMOTherm (version 2017) with parameter file BP_TZVPD_FINE_C30_1501.ctd, based on BP86/def2-TZVPD48 level. The best conformer of each studied structure was obtained from extensive conformational searches performed at the level of optimization with respect to all rotatable bonds. Pictures of the computed structures were generated using CYLview.49 For more information on the thermodynamic cycle used to obtain Gibbs free energies in solution, description of the COSMO-RS environment, and how conformational entropy was accounted for, see Supporting Information. 4.2. General Information. 3-(Dimethylamino)propylamine, 3(diethylamino)propylamine, N-(3-aminopropyl)imidazole, and N-(3aminopropyl)piperidine were obtained from Aldrich, N-(3aminopropyl)pyrrolidine from TCI. All reagents were used as received. Lupranol 2090 (BASF, Germany) is a trifunctional polyether polyol, which contains primary hydroxyl groups and OH number 28 mg KOH/g. Lupranat MM 103 (BASF, Germany) is a carbodiimidemodified MDI, with average functionality 2,2 and NCO content 29,5%. IR spectra were recorded on a Bruker Tensor 27 spectrophotometer using a ZnSe crystal ATR accessory. 4.3. General Procedure for the Synthesis of PU Foam and Conversion Monitoring. Lupranol 2090 (49.25 g), 0.25 g of water, and 0.5 g of 3-(dimethylamino)propylamine (1) were mixed, and stoichiometric amount of Lupranat MM 103 (8.5 g) was added at room temperature. After stirring for 5 s, a small sample of reaction mixture was removed from the reaction flask and placed on the ATR crystal heated to 60 °C. The reaction progress was estimated by measuring the disappearance of the NCO stretching vibrations at 2271 cm−1 as a function of time. Quantification was carried out, relying on the unchanged absorption of CH2 groups at 2865 cm−1. The same procedure was used for conversion monitoring in the presence of equimolar amount of 1, 2, 4, 5, and 6. 4.4. Emission Method. Formaldehyde emission was measured using a small chamber test. PU foam sample (50 × 50 × 25 mm3, density 360 g/L) was placed in a 4.7 L chamber and heated to 65 °C

3. CONCLUSIONS We used modern quantum chemical methods to study consistently how structural modification of the N,N-dimethyl groups of widely employed reactive PU catalysts affects their catalytic activity. The computational study provided us with general design principles: (1) the catalyst reactivity decreases with increased chain length, that is, from methyl to ethyl and propyl, mainly because of increased conformational entropy; and (2) aliphatic nitrogen-heterocycles provide a catalytically active alternative for tertiary amines containing N,N-dimethyl groups. The activity of the aliphatic heterocycles follows Bent’s rule, suggesting that the catalyst reactivity is decreased with 8207

DOI: 10.1021/acs.joc.9b01319 J. Org. Chem. 2019, 84, 8202−8209

Article

The Journal of Organic Chemistry with air flowing over it, with an air exchange rate of 1.28 h−1 and 10% relative humidity. The exhaust airstream (6 L) containing formaldehyde from PU foam is sampled after a preconditioning time of 2 h. Formaldehyde is collected on a dinitrophenyl hydrazine (DNPH) column (cartridge). After absorption, the DNPH cartridge is eluted with 5 mL acetonitrile/water mixture (1:1) and obtained solution analyzed by HPLC to quantify the formaldehyde in the sample. The results are expressed as emission per volume of air sampled (μg/m3). The method’s limit of detection for formaldehyde is ≤11 μg/m3.



(10) Barman, B. N. Formation and Determination of Aldehydes in Amine Catalysts, Adhesives & Sealants Industry, 2015; pp 36−42. (11) Salthammer, T.; Mentese, S.; Marutzky, R. Formaldehyde in the Indoor Environment. Chem. Rev. 2010, 110, 2536−2572. (12) Schwetlick, K.; Noack, R.; Stebner, F. Three fundamental mechanisms of base-catalyzed reactions of isocyanates with hydrogenacidic compounds. J. Chem. Soc., Perkin Trans. 2 1994, 599−608. (13) Becke, A. D. Density-functional exchange-energy approximation with correct asymptotic behavior. Phys. Rev. A: At., Mol., Opt. Phys. 1988, 38, 3098−3100. (14) Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. A Consistent and Accurate Ab Initio Parametrization of Density Functional Dispersion Correction (DFT-D) for the 94 Elements H-Pu. J. Chem. Phys. 2010, 132, 154104. (15) Grimme, S.; Ehrlich, S.; Goerigk, L. Effect of the damping function in dispersion corrected density functional theory. J. Comput. Chem. 2011, 32, 1456−1465. (16) 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. (17) Klamt, A.; Eckert, F. COSMO-RS: a novel and efficient method for the a priori prediction of thermophysical data of liquids. Fluid Phase Equilib. 2000, 172, 43−72. (18) Deglmann, P.; Mueller, I.; Becker, F.; Schaefer, A.; Hungenberg, K.-D.; Weiss, H. Prediction of Propagation Rate Coefficients in Free Radical Solution Polymerization Based on Accurate Quantum Chemical Methods: Vinylic and Related Monomers, Including Acrylates and Acrylic Acid. Macromol. React. Eng. 2009, 3, 496−515. (19) Eshuis, H.; Bates, J. E.; Furche, F. Electron Correlation Methods Based on the Random Phase Approximation. Theor. Chem. Acc. 2012, 131, 1084. (20) Chen, G. P.; Voora, V. K.; Agee, M. M.; Balasubramani, S. G.; Furche, F. Random-Phase Approximation Methods. Annu. Rev. Phys. Chem. 2017, 68, 421−445. (21) Eshuis, H.; Furche, F. A Parameter-Free Density Functional That Works for Noncovalent Interactions. J. Phys. Chem. Lett. 2011, 2, 983−989. (22) Bates, J. E.; Furche, F. Communication: Random Phase Approximation Renormalized Many-Body Perturbation Theory. J. Chem. Phys. 2013, 139, 171103. (23) Chen, G. P.; Agee, M. M.; Furche, F. Performance and Scope of Perturbative Corrections to Random-Phase Approximation Energies. J. Chem. Theory Comput. 2018, 14, 5701−5714. (24) Wen, Z.; Hu, W.; Chi, X.; Wang, X.; Tian, D.; Wang, M.; Liu, J.; Ma, X.; Pang, A. DFT Study of the Catalytic Mechanism for Urethane Formation in the Presence of Basic Catalyst 1,4diazabicyclo[2.2.2]octane. Commun. Comput. Chem. 2014, 2, 22−35. (25) Hatanaka, M. DFT analysis of catalytic urethanation. Bull. Chem. Soc. Jpn. 2011, 84, 933−935. (26) Stewart, J. J. P.; Davis, L. P.; Burggraf, L. W. Semi-epirical calculation of molecular trajectories: Method and application to some simple molecular systems. J. Comput. Chem. 1987, 8, 1117−1123. (27) Depending on conformation of the transition state, a high energy zwitter-ionic intermediate could exist. These, however, are expected to transfer the proton spontaneously and not limit the reactivity. (28) Goerigk, L.; Grimme, S. A thorough benchmark of density functional methods for general main group thermochemistry, kinetics, and noncovalent interactions. Phys. Chem. Chem. Phys. 2011, 13, 6670−6688. (29) Sinnokrot, M. O.; Valeev, E. F.; Sherrill, C. D. Estimates of the ab initio limit for pi-pi interactions: the benzene dimer. J. Am. Chem. Soc. 2002, 124, 10887−10893. (30) Janowski, T.; Pulay, P. A benchmark comparison of σ/σ and π/ π dispersion: the dimers of naphthalene and decalin, and coronene and perhydrocoronene. J. Am. Chem. Soc. 2012, 134, 17520−17525.

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.joc.9b01319. Cartesian coordinates for all optimized catalysts, reactants, and transition states (TXT) Additional tables and figures including IR kinetic measurements, optimized transition states, and Gibbs free energies and relative rate constants; more detailed description of free energies in solution and the solvation environment; and computation of conformational entropy (PDF)



AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected] (M.M.). *E-mail: [email protected] (Ž .T.). ORCID

Mikko Muuronen: 0000-0001-9647-7070 Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS Professor Filipp Furche and Guo Chen are greatly acknowledged for sharing RI-AXK code and instructions on performing the AXK computations. For laboratory support, we thank Natascha Mitko, Anja Gerber, and Astrid Güngördü.



REFERENCES

(1) Randal, D.; Lee, S. The Polyurethane Book; John Wiley & Sons: New York, 2002. (2) Engels, H.-W.; Pirkl, H.-G.; Albers, R.; Albach, R. W.; Krause, J.; Hoffmann, A.; Casselmann, H.; Dormish, J. Polyurethanes: Versatile Materials and Sustainable Problem Solvers for Today’s Challenges. Angew. Chem., Int. Ed. 2013, 52, 9422−9441. (3) Akindoyo, J. O.; Beg, M. D. H.; Ghazali, S.; Islam, M. R.; Jeyaratnam, N.; Yuvaraj, A. R. Polyurethane types, synthesis and applications - a review. RSC Adv. 2016, 6, 114453−114482. (4) Delebecq, E.; Pascault, J.-P.; Boutevin, B.; Ganachaud, F. On the Versatility of Urethane/Urea Bonds: Reversibility, Blocked Isocyanate, and Non-isocyanate Polyurethane. Chem. Rev. 2013, 113, 80− 118. (5) Sardon, H.; Pascual, A.; Mecerreyes, D.; Taton, D.; Cramail, H.; Hedrick, J. L. Synthesis of Polyurethanes Using Organocatalysis: A Perspective. Macromolecules 2015, 48, 3153−3165. (6) Alsarraf, J.; Ammar, Y. A.; Robert, F.; Cloutet, E.; Cramail, H.; Landais, Y. Cyclic Guanidines as Efficient Organocatalysts for the Synthesis of Polyurethanes. Macromolecules 2012, 45, 2249−2256. (7) See for example JEFFCAT catalysts at www.huntsman.com. (8) Barman, B. N.; Grigsby, R. A. Inhibition of amine oxidation. WO 2011084865A1, 2011. (9) Barman, B. N. Accurate determination of aldehydes in amine catalysts or amines by 2,4-dinitrophenylhydrazine derivatization. J. Chromatogr. A 2014, 1327, 19−26. 8208

DOI: 10.1021/acs.joc.9b01319 J. Org. Chem. 2019, 84, 8202−8209

Article

The Journal of Organic Chemistry (31) Jensen, J. H. Predicting accurate absolute binding energies in aqueous solution: thermodynamic considerations for electronic structure methods. Phys. Chem. Chem. Phys. 2015, 17, 12441−12451. (32) Bent, H. A. An Appraisal of Valence-bond Structures and Hybridization in Compounds of the First-row elements. Chem. Rev. 1961, 61, 275−311. (33) Weider, R.; Scholz, U. Use of N-(aminoalkyl)pyrrolidines as catalysts for the polyisocyanate polyaddition process. U.S. Patent 5134217, 1992. (34) Schmitz, S.; Glos, M.; Hubel, R.; Terheiden, A. Amines suitable for the use in the production of polyurethanes. WO 2014121959A1, 2014. (35) Guenther, T.; Fiedel, M.; Glos, M.; Hubel, R.; Vidakovic, M.; Wettey, C. Nitrogen-containing compounds suitable for use in the production of polyurethanes. WO 2016020200A1, 2016. (36) Madra, H.; Tantekin-Ersolmaz, S. B.; Guner, F. S. Monitoring of oil-based polyurethane synthesis by FTIR-ATR. Polym. Test. 2009, 28, 773−779. (37) Furche, F.; Ahlrichs, R.; Hättig, C.; Klopper, W.; Sierka, M.; Weigend, F. Turbomole. Wiley Interdiscip. Rev.: Comput. Mol. Sci. 2014, 4, 91−100. (38) Weigend, F.; Furche, F.; Ahlrichs, R. Gaussian Basis Sets of Quadruple zeta Valence Quality for Atoms H-Kr. J. Chem. Phys. 2003, 119, 12753. (39) Sierka, M.; Hogekamp, A.; Ahlrichs, R. Fast evaluation of the Coulomb potential for electron densities using multipole accelerated resolution of identity approximation. J. Chem. Phys. 2003, 118, 9136− 9148. (40) Weigend, F. Accurate Coulomb-fitting basis sets for H to Rn. Phys. Chem. Chem. Phys. 2006, 8, 1057−1065. (41) Eshuis, H.; Yarkony, J.; Furche, F. Fast Computation of Molecular Random Phase Approximation Correlation Energies Using Resolution of the Identity and Imaginary Frequency Integration. J. Chem. Phys. 2010, 132, 234114. (42) Weigend, F.; Häser, M. RI-MP2: first derivatives and global consistency. Theor. Chem. Acc. 1997, 97, 331−340. (43) Weigend, F.; Häser, M.; Patzelt, H.; Ahlrichs, R. RI-Mp2: Optimized Auxiliary Basis Sets and Demonstration of Efficiency. Chem. Phys. Lett. 1998, 294, 143−152. (44) Hättig, C. Optimization of Auxiliary Basis Sets for RI-MP2 and RI-CC2 Calculations: Core−Valence and Quintuple-ζ Basis Sets for H to Ar and QZVPP Basis Sets for Li to Kr. Phys. Chem. Chem. Phys. 2005, 7, 59−66. (45) Tao, J.; Perdew, J.; Staroverov, V.; Scuseria, G. Climbing the Density Functional Ladder: Nonempirical Meta-Generalized Gradient Approximation Designed for Molecules and Solids. Phys. Rev. Lett. 2003, 91, 146401. (46) Schäfer, A.; Klamt, A.; Sattel, D.; Lohrenz, J. C. W.; Eckert, F. COSMO Implementation in TURBOMOLE: Extension of an efficient quantum chemical code towards liquid systems. Phys. Chem. Chem. Phys. 2000, 2, 2187−2193. (47) 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, 799−805. (48) Rappoport, D.; Furche, F. Property-optimized Gaussian basis sets for molecular response calculations. J. Chem. Phys. 2010, 133, 134105. (49) Legault, C. Y. CYLview, b.; Université de Sherbrooke, 2009 (http://www.cylview.org).

8209

DOI: 10.1021/acs.joc.9b01319 J. Org. Chem. 2019, 84, 8202−8209