How Protic Solvents Determine the Reaction Mechanisms of

Jul 25, 2019 - jo9b01228_si_001.pdf (900.16 kb) ... Organische ... ür Physikalische und Theoretische Chemie, Universität Regensburg, Universitätsst...
2 downloads 0 Views 2MB Size
Article Cite This: J. Org. Chem. XXXX, XXX, XXX−XXX

pubs.acs.org/joc

How Protic Solvents Determine the Reaction Mechanisms of Diphenylcarbene in Solution Johannes Knorr,†,∇ Pandian Sokkar,§,∇,∥ Paolo Costa,‡,# Wolfram Sander,‡ Elsa Sanchez-Garcia,*,§ and Patrick Nuernberger*,†,⊥ †

Physikalische Chemie II and ‡Organische Chemie II, Ruhr-Universität Bochum, 44780 Bochum, Germany § Computational Biochemistry, Center of Medical Biotechnology, University of Duisburg-Essen, 45117 Essen, Germany

Downloaded via BUFFALO STATE on July 26, 2019 at 02:28:13 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

S Supporting Information *

ABSTRACT: We investigate the effects of small admixtures of protic solvent molecules, such as water and alcohols, on the ultrafast dynamics of diphenylcarbene in acetonitrile at room temperature. Broadband transient absorption measurements and quantum mechanics/molecular mechanics molecular dynamics simulations allow elucidating the dominant reaction mechanism of an intermediate hydrogen-bonded complex between singlet diphenylcarbene and a protic solvent molecule, thus competing with intersystem crossing. Analysis of the data indicates that complex formation is a diffusioncontrolled process with orientational requirements. The reaction path involving a benzhydryl cation is less likely in neat bulkier alcohols, as it requires the interaction of the carbene with a protic solvent molecule being part of a hydrogen-bonded network. The simulations indicate a further reaction path toward O−H insertion and two side reactions depending on the involved protic solvent species. Thus, we established that not only the number but also the chemical nature of the protic solvent molecule determine which reaction path is pursued.



INTRODUCTION Carbenes, which can be generated by photolysis of diazo precursors, are very reactive species due to the divalent carbon atom. An aspect of high relevance for their reactivity is the spin multiplicity. Initially, the singlet carbene is formed in the photolytic step, which may be followed by a rapid intersystem crossing (ISC), a process that strongly depends on the energetic difference of singlet and triplet species. This energy gap is influenced by variation in side groups and aromaticity of the compounds. Investigations of the prototype carbene diphenylcarbene (Ph2C), using techniques like time-resolved and cryogenic spectroscopy approaches as well as computational studies,1−21 showed that hydrogen-bond interactions with solvent molecules also influence the singlet−triplet gap. Recently, we reported on the ultrafast dynamics of Ph2C in MeOH/MeCN (methanol/acetonitrile) solvent mixtures at room temperature by complementing broadband transient absorption measurements with quantum mechanics/molecular mechanics (QM/MM) molecular dynamics (MD) simulations.21 The study highlighted the importance of a hydrogenbonded complex (1Ph2C···HOMe) between 1Ph2C and a MeOH solvent molecule, thereby resembling results from matrix isolation studies at 3 K.19 However, it is still not clear which parameters determine the complexation process and thus carbene reactivity in solution. To address this question, here, we study Ph2C in neat solvents and admixtures of water and 2-BuOH (2-butanol) in © XXXX American Chemical Society

acetonitrile and compare the results to those using methanol as the protic solvent (neat and admixtures). By modifying the MeCN solvent environment by addition of 1% water, we observe the singlet decay via a hydrogen-bonded complex (1Ph2C···HOH), which so far had only been detected in argon matrices at very low temperatures.20 A comparative analysis of the results in neat MeCN and with 1% admixtures of protic solvent species (H2O or MeOH) indicates that the complexation is diffusion-controlled and requires the molecule of the protic solvent to be very close and in the right orientation with respect to the carbene. We previously established that, with MeOH as the solvent, two alcohol molecules are involved in the formation of the benzhydryl cation (Ph2CH+).21 This was surprising since formally only one MeOH molecule should be sufficient for protonation. To determine if this observation also applies to protic solvents other than MeOH, here, we investigate the protonation of 1Ph2C in the much bulkier solvent 2-BuOH. We found that cation formation is slower and the reaction takes place to less extent in 2-BuOH with respect to MeOH. In addition, QM/MM MD simulations indicate the possibility of an alternative reaction pathway in 2-BuOH leading to Ph2CH2 (diphenylmethane). Moreover, we propose further reaction mechanisms of relevance for water admixtures. Received: May 7, 2019

A

DOI: 10.1021/acs.joc.9b01228 J. Org. Chem. XXXX, XXX, XXX−XXX

Article

The Journal of Organic Chemistry



RESULTS Traces of Protic Solvents. We first address mixtures of water and 2-butanol as protic solvents in acetonitrile from a computational perspective. Without any solvent (vacuum, Figure 1), the singlet state of Ph2C is higher in energy than the

Figure 1. Singlet−triplet energy gaps ΔE(S−T) of Ph2C in different solvent environments using the triplet state energy as a reference. The B3LYP-D3/Def2-TZVP level of theory was used for the QM region in QM/MM computations. The values for methanol are taken from ref 21.

Figure 2. Transient absorption of Ph2CN2 under 285 nm excitation in different solvent environments. (a) MeCN only; (b) MeCN with 1 vol % MeOH; (c) MeCN with 1 vol % H2O; note the linearlogarithmic ordinate. (d) Band integral of a transient absorption signal from 305 to 320 nm representing the dynamics of 3Ph2C. The green curves correspond to fits to the Smoluchowski diffusion model discussed in the Supporting Information. The upper two curves are vertically offset for clarity.

triplet state by 5.1 kcal/mol at the B3LYP-D3/Def2-TZVP level of theory. Upon complex formation with a single molecule of protic solvent, the energy gap is inverted (Figure 1). The ΔE(S−T) gap becomes more prominent in the condensed phase, with values of −6.5 and −7.3 kcal/mol in 100% 2-BuOH and in the 99:1% (v/v) MeCN/2-BuOH mixture, respectively. This is because polar solvents, such as acetonitrile, water, and 2-butanol, interact more favorably with the singlet state than with the triplet. It is noteworthy that 1% of alcohol in MeCN results in a more negative value for ΔE(S−T) than in the neat alcohol environment (Figure 1). The study of the effect of different protic solvent admixtures on the ultrafast dynamics is addressed experimentally by broadband transient absorption measurements on Ph2CN2 in MeCN (Figure 2a) and with 1 vol % admixtures of MeOH (Figure 2b) or H2O (Figure 2c). Note that, due to the limited solubility of Ph2CN2 in H2O, it is not possible to study solvent mixtures with significantly higher H2O contents. Under all three measurement conditions, we exclusively observe positive absorption changes, reflecting the contributions from a shortlived absorption of the excited precursor Ph2CN2* around time zero at ≈335 nm, the subsequent absorption of 1Ph2C at ≈355 nm, and the signature of 3Ph2C appearing after several hundreds of picoseconds at ≈320 nm (see also the alternative representation of the data as transient spectra in Figure S1, Supporting Information). The observed spectral positions agree well with the literature,17,21 and all three measurements lack the absorption of Ph2CH+ at ≈435 nm observed in solutions containing a significant amount of protic solvent. The most obvious change in the dynamics caused by the protic solvent is the strongly reduced absorption intensity of 3Ph2C. In anhydrous MeCN, the time constants for the decay of 1 Ph2C and the rise of 3Ph2C match, corroborating that all singlet molecules undergo ISC to the triplet state.17,21 We chose mixtures of 1 vol % of H2O and MeOH because, although the mole fractions in MeCN are different (0.028 and 0.013, respectively), the polarity of the two binary mixtures is identical within the experimental accuracy of the measurements. When 1% of the protic solvent is added, the polarity values ET(30)22,23 are 48.0 and 48.1 kcal/mol, respectively, as derived using the relations given in ref 24. Generally, the ISC

rate of carbenes in solution becomes smaller with increasing solvent polarity.18,25 Therefore, the associated lifetime of the singlet carbene should be prolonged. From the studies of the Eisenthal group on Ph2C in various aprotic solvents,13 we can estimate that this prolongation should be by a factor of 1.2 relative to neat MeCN [ET(30) = 45.6 kcal/mol] if there are no other processes than 1Ph2C → 3Ph2C. However, both with H2O and MeOH admixtures, the rise time of the triplet is significantly shorter than in pure MeCN, substantiating that the population of 1Ph2C is quenched by a process competing with ISC. As established in our earlier study for MeCN/MeOH mixtures,21 this quenching can be due to the formation of a hydrogen-bonded complex between the carbene and a molecule of water or alcohol, preventing ISC and eventually leading to an ether product. We found a linear dependence of this reaction channel’s rate on the concentration of the admixture. For rationalizing this behavior, we apply Smoluchowski’s theory for diffusion processes in solution,26−30 as frequently applied in fluorescence quenching. Details on the model and the quantities employed for the reaction between Ph2C and H2O (or MeOH) are provided in the Supporting Information. The description within the Smoluchowski model yields a value of (1.3 ± 0.8) Å for the interaction distance of H2O and 1 Ph2C. In the case of MeOH, the interaction distance is (1.4 ± 0.4) Å. These values are rather small and on the order of the calculated hydrogen-bond distance which is ≈1.8 Å for the complex 1Ph2C···HOMe.21 Noticeably, for the bimolecular reaction of the radical Ph2CH• in hexane and cyclohexane, interaction distances below 1 Å have been determined within the Smoluchowski model.31 On the one hand, the small values obtained here suggest that the complex formation occurs by close collision only, and there is no long-range attraction between the singlet carbene and the B

DOI: 10.1021/acs.joc.9b01228 J. Org. Chem. XXXX, XXX, XXX−XXX

Article

The Journal of Organic Chemistry

47.1 kcal/mol]23 and MeOH [ET(30) = 55.4 kcal/mol],22 respectively, with the differences originating from solvatochromism. Additionally, we observe the ground state bleach of Ph2CN2 at the blue-edge of the detection range. The analysis of the transients associated with 1Ph2C and Ph2CH+ (Figure 3b,d) illustrates the solvents’ influence on the characteristic time scales: in MeOH, compared to 2-BuOH, 1Ph2C disappears much faster, whereas Ph2CH+ is formed earlier and decays more rapidly. Next, we quantify the dynamics of the carbene in 2-BuOH, especially the evidently reduced Ph2CH+ formation in 2BuOH. The determination of the protonation fraction according to the procedure described in ref 17 yields f p = 0.06 in 2-BuOH. Thus, the pathway along the 1Ph2C···HOBu complex formation is even more pronounced, as can also be deduced from the actual coexistence of signals from Ph2CH+ [red curve in Figure 3b, decay time (105 ± 11) ps] and 1Ph2C species [blue curves in Figure 3b, decay time (101 ± 3) ps]. For elucidating at the molecular level how the proton transfer and competitive reaction pathways proceed on an ultrafast time scale, we computationally address the reaction of 1 Ph2C under various solvent environments. It is advantageous to both explore cases with neat protic solvents and their traces in MeCN. Therefore, we simulated the following systems: (i) 100% 2-BuOH, (ii) MeCN with 1 vol % 2-BuOH, and (iii) MeCN with 1 vol % H2O. For this purpose, we established the following protocols: in case (i), we took 16 snapshots from the last 2 ns of classical MD simulations and used a model in which 1Ph2C is placed in a 20 Å solvent droplet. We then ran 16 replicas of QM/MM MD simulations per system. The QM region consisted of 1Ph2C and four or five molecules of 2BuOH. The selection criterion for including the solvent molecules in the QM region was that the oxygen atoms should be within 6 Å of the carbene center (note that the derivation of rather small interaction distances from the experimental data supports this restriction). We applied spherical boundary conditions to prevent the flying-away of solvent molecules. The B3LYP-D3/Def2-SVP level of theory was used for the QM calculations and the rest of the system was computed at the MM level. For cases (ii) and (iii), it is more complicated to predict the microenvironment around 1Ph2C, which is crucial for its reactivity. Hence, we performed longer MD simulations (100 ns each) to determine the number of 2-BuOH or H2O molecules diffusing around 1Ph2C within 10 Å distance of the carbene center. During the sampling, one molecule of 2-BuOH was present in a 10 Å radius with respect to the carbene center for 35% of the simulation time, two molecules were found for 15%, and three molecules were present in 4% of the simulation time (Figure S2, Supporting Information). For H 2 O admixtures of case (iii), the populations were found to be 32, 23, 10, and 4% for one to four molecules, respectively (Figure S2, Supporting Information). This information allows estimating an upper limit for the number of water or 2-BuOH molecules that form the microenvironment around the carbene center. Hence, it may be stated that one to three 2-BuOH molecules are likely to be near 1Ph2C, whereas for H2O on average more molecules are close to 1Ph2C. This can be rationalized given the smaller size, higher mole fraction, and larger diffusion coefficient of H2O compared to alcohols. Previously, our study of 1Ph2C in MeCN/MeOH binary mixtures showed that a single MeOH molecule is not sufficient to react with 1Ph2C but still forms the H-bonded complex.21

H2O or MeOH molecules. On the other hand, these values indicate that the Smoluchowski theory is a too simplified approach. The involved particles are not spherical, and the formation of the complex has orientational requirements for which simply diffusing to within a certain distance is not sufficient. Several approaches found in the literature allow taking additional aspects into account,26,27,32 and we have applied two of these approaches to model the dynamics more accurately (see the Supporting Information). The corresponding results imply that the small interaction distances are either due to a finite time required for complex formation even when the reaction partners are sufficiently close or that not all H2O molecules reaching the required interaction distance eventually form a complex. Reaction Mechanisms. After the formation of a hydrogenbonded complex 1Ph2C···HOH or 1Ph2C···HOMe, the protonation of 1Ph2C takes place to yield Ph2CH+ and a hydroxide or alkoxide anion, respectively.17,21,33 For this protonation reaction, a second molecule of water or 2-BuOH (besides the one donating the proton) is required. Kohler and co-workers17 hypothesized that this should be less likely in bulkier protic solvents, and corroborated this by measuring the protonation fraction f p, i.e., the amount of photoexcited diazo molecules that follow the reaction pathway with the observable Ph2CH+ signal. They reported decreasing f p values ranging from 0.30 in neat MeOH via 0.15 (EtOH) and 0.14 (1-PrOH) to 0.09 (2-PrOH). Here, we explore whether this trend is still valid for even larger alcohols and investigate the underlying mechanisms. Figure 3 shows the fate of 1Ph2C in neat 2-BuOH (Figure 3a,b) and neat MeOH (Figure 3c,d). Transient absorption spectra feature the positive signatures of Ph2CN2* and 1Ph2C, while 3Ph2C is not formed. Instead, both measurements unveil the build-up and decay of Ph2CH+ with a characteristic signal in the region around 430 nm.17,21,34−37 Here, this region is centered at 429 and 435 nm for the case of 2-BuOH [ET(30) =

Figure 3. Transient absorption for excitation of Ph2CN2 with 285 nm pump pulses in an alcohol solution. Dynamics observed (a) in 2BuOH and (c) in MeOH. (b) Transients in 2-BuOH taken from (a) for probe wavelengths of 355 nm (signal of 1Ph2C species) and 429 nm (Ph2CH+ signal). (d) Transients in MeOH taken from (c) for probe wavelengths of 355 nm (signal of 1Ph2C species) and 435 nm (Ph2CH+ signal). C

DOI: 10.1021/acs.joc.9b01228 J. Org. Chem. XXXX, XXX, XXX−XXX

Article

The Journal of Organic Chemistry

Type 2 mechanism leads to the formation of products that contain fragments from the same 2-BuOH molecule. Initially, a 2-BuOH molecule forms a H-bonded complex with 1Ph2C. This process is reflected by the experimental data (Figure 2b,c) and is in agreement with the diffusion analysis presented above. An additional interaction with a neighboring 2-BuOH molecule eventually triggers the proton transfer within the complex and the butoxide ion geminately attacks the benzhydryl cation to form the ether product. Therefore, essentially only one 2-BuOH molecule is involved in the reaction and an additional 2-BuOH molecule acts as the catalyst for the reaction. Similar to MeOH and 2-BuOH, we found the type 1 and type 2 mechanisms in the simulations with H2O admixtures (iii). Here, type 1 involves two or three H2O molecules, whereas for 2-BuOH only the participation of two molecules was observed. Interestingly, in the system with water, (iii) we observed a further mechanism (type 3, Scheme 2) in which the carbene is

Hence, for the QM/MM studies, we do not consider systems with only one 2-BuOH or H2O molecule in the proximity of 1 Ph2C but rather concentrate on situations where two protic solvent molecules might act conjointly. We thus took from the MD trajectories, representative structures with two or three 2BuOH molecules near 1Ph2C (for the case of H2O, we considered two, three, or four H2O molecules near the carbene). Similar to protocol (i), we built a solvent droplet model by deleting solvent molecules outside a 20 Å radius from the carbene. Distance restraints were applied between the carbene center and the oxygen atoms of the 2-BuOH or H2O molecules with a weak force constant (0.5 kcal/mol/Å2) and an equilibrium distance of 3 Å to ensure that they do not move far away from the carbene. Then, we performed short (2 ns) MD simulations to sample different orientations of the protic solvent molecules around 1Ph2C. We used 10−20 snapshots from the last 1 ns of these trajectories for QM/MM MD simulations for 20 ps at 300 K of each system. Thus, we extensively studied our systems by performing a total of 33 replicas of QM/MM MD simulations of water−acetonitrile systems (two, three, or four water molecules near the carbene center) and 56 replicas of QM/MM MD simulations involving 2-butanol (neat solvent and mixtures with two or three alcohol molecules near the carbene center). The QM/MM MD simulations revealed that the type 1 and type 2 mechanisms previously reported by us for the reaction of 1Ph2C with MeOH21 also occur with both 2-BuOH and H2O (Scheme 1).

Scheme 2. Type 3 Reaction Mechanism Observed only During the QM/MM MD Simulations of 1Ph2C in MeCN with H2O

Scheme 1. Type 1 and Type 2 Reaction Mechanisms Observed During the QM/MM MD Simulations of 1Ph2C in MeCN with 2-BuOH and H2O. For Simplicity, only 2BuOH Reactions are Shown

inserted into the O−H bond of H2O. This mechanism involves more than one H2O molecule. The first step is the formation of a hydrogen-bonded complex between H2O and 1Ph2C, with the corresponding H2O molecule also interacting with other H2O molecule(s) in a H-bond network. As opposed to type 1 and type 2 mechanisms, the reaction is initiated by a nucleophilic attack on the carbene center, i.e., the oxygen atom of H2O attacks from above the plane of the H-bond. Assuming that the carbene center of 1Ph2C possesses sp2 hybridization with an empty p-orbital, the type 3 mechanism is initiated by the overlap of nonbonded orbitals of oxygen with the empty p-orbital of the carbene, leading to the formation of an oxonium ylide. In the next step, proton transfer occurs to the neighboring H2O to form a hydronium cation. A proton transfer to the carbanion completes the formal O−H insertion reaction. In addition to the three mechanisms toward O−H insertion, during the simulations, we also observed two side reactions, that is, oxidation of 2-BuOH to 2-butanone and the orthohydroxylation in H2O (Scheme 3). The 2-BuOH oxidation mechanism is very similar to the type 1 mechanism, with a difference in the final reaction step. Instead of an attack of the butoxide anion, a hydride anion transfer occurs between the butoxide anion and the benzhydryl cation, leading to the formation of the ketone and diphenylmethane. The orthohydroxylation reaction found with H2O is also very similar to the type 1 mechanism, but instead of the hydroxide anion attacking the carbocation center of the benzhydryl cation, it attacks the ortho position of the phenyl ring. All reactions (mechanisms of type 1 to 3 and side reactions) occur with varying frequency depending on the solvent

In the type 1 mechanism, the O−H insertion product contains fragments from distinct 2-BuOH molecules. In the first step, 1Ph2C forms a H-bond with a 2-BuOH molecule that is part of a H-bond network with additional 2-BuOH molecules. Thus, 1Ph2C is protonated, yielding a Ph2CH+/ butoxide anion pair. The butoxide anion is stabilized by the Hbond network and quickly receives a proton from another 2BuOH molecule, thereby propagating the charge. Subsequently, this secondary butoxide anion attacks the benzhydryl cation to form the ether product. In principle, any number of proton transfer reactions (charge propagation) is possible across the H-bond network of 2BuOH molecules. However, in the QM/MM MD simulations (56 trajectories, 20 ps each) involving 2-BuOH, we observed only two 2-BuOH molecules involved in proton transfer. D

DOI: 10.1021/acs.joc.9b01228 J. Org. Chem. XXXX, XXX, XXX−XXX

Article

The Journal of Organic Chemistry

Furthermore, we discuss if the transient absorption signals contain indications for the side reactions disclosed in the simulations. With 1% of water or 2-BuOH added to MeCN, the probability to find several molecules of these protic solvents in close proximity to 1Ph2C and thus to interact conjointly is experimentally very low. Thus, we do not expect to detect signs of the side reactions in these admixtures. In the 2-BuOH oxidation reaction found in the simulations with neat 2-BuOH, 1Ph2C is always surrounded by a sufficient amount of alcohol molecules. The putative product diphenylmethane would only absorb in the deep-UV, which is outside our spectral observation window.38 Trying to detect the ketone’s vibrational signal might be a possibility, but it will clash with the solvent’s absorption. Lacking a direct transient signal sensitive to the 2-BuOH oxidation reaction, we point out one factor: Kohler and co-workers report that the protonation fraction shrinks from 0.30 in MeOH to 0.09 in 2-PrOH, and in turn the decay time constant associated with Ph2CH+ steadily increases from 30.5 ps (MeOH) to 130 ps (2-PrOH).17 In our analysis (Figure 3a), we determined a protonation fraction of 0.06 for 2-BuOH, but the Ph2CH+ signal disappears with a comparably low time constant of 105 ps. While this is only an indirect indication, the fact that the lifetime of Ph2CH+ is shorter than expected could be related to a further process that quenches Ph2CH+ beyond the attack of the butoxide anion, putatively the hydride transfer seen in the simulations.

Scheme 3. Side Reactions Observed in the QM/MM MD Simulations of 1Ph2C in 2-BuOH and H2O Environments

conditions, as summarized in Table 1 for the different microenvironments.



DISCUSSION The experiments with 1 vol % of protic solvents in acetonitrile indicated that the ISC pathway is strongly quenched by another reaction mechanism that does not give rise to a signal from an intermediate benzhydryl cation. Thus, complex formation and subsequent O−H insertion according to the type 2 mechanism identified in the simulations occur. Type 2 prevails for 2-BuOH admixtures (Table 1). A starting situation involving several molecules of water may be rare in the experiments, because the water molecules have to diffuse within a short distance on the order of the length of a H-bond to interact with the carbene center. Therefore, no indication of the type 1 mechanism is evident from the experiments (Figure 2c). The analysis based on the Smoluchowski model confirms the contribution of diffusion to the reaction rate and provides insights into complex formation. The obtained short interaction distance of H2O and 1Ph2C points toward further relevant effects, suggesting that not every encounter of the carbene with a water or alcohol molecule leads to complex formation, which is related to orientational mismatching. The simulations support this statement, that is, in the case of two H2O molecules close to the carbene center (Table 1), only one out of ten trajectories leads to a reaction despite the high mobility of H2O. Experiments in neat 2-BuOH and neat MeOH showed that the type 1 mechanism is 5 times less likely in 2-BuOH compared to MeOH. This can be understood by the necessity of a hydrogen-bonded network to stabilize the alkoxide anion, a situation more distinct in less bulky alcohols. Consequently, the type 2 mechanism is more pronounced for 2-BuOH, which is reflected in the similar time constants found for the decay of 1 Ph2C···HOBu and Ph2CH+.



CONCLUSIONS The ultrafast reaction of 1Ph2C photogenerated from diphenyldiazomethane in MeCN solution can be controlled by the addition of protic solvent molecules. Our joint experimental and theoretical study elucidates to which extent the reaction pathways are related to the nature of the protic admixtures. Without the latter, the primary step performed by all 1Ph2C molecules is intersystem crossing, whereas competing reaction mechanisms set in if protic molecules are within reach of the carbene center. The reaction pathway along the intermediate Ph2CH+, requiring at least two protic solvent molecules, becomes less likely with bulkier alcohol species, and could become completely inaccessible for very large alcohols. The dynamics of the reaction pathway comprising a strongly H-bonded complex between 1Ph2C and a single protic solvent molecule is governed by translational diffusion of the reaction partners. Our transient absorption data also suggest that the relative orientation of the carbene and the solvent molecule, and thus rotational diffusion, are also crucial for the complexation process. The simulations further unveiled the possibility of side reactions beyond the O−H insertion. Hence, our studies establish that different protic partners react with

Table 1. Reaction Pathways Observed in the QM/MM MD Simulations number of trajectories where the reaction occurs MeCN (vol %) 99 99 99 99 99

2-BuOH (vol %)

H2O (vol %)

protic solvent molecules near 1PhC

total trajectories

type 1

type 2

type 3

side

1 1 1

n/a 2 3 2 3 4

18 18 20 10 10 13

1 2 2 0 2 5

12 9 13 0 1 4

0 0 0 1 1 1

5a 0 1a 0 0 1b

100 1 1

a

BtOH oxidation. bo-Hydroxylation. E

DOI: 10.1021/acs.joc.9b01228 J. Org. Chem. XXXX, XXX, XXX−XXX

Article

The Journal of Organic Chemistry 1

cutoff distance. Periodic boundary conditions were employed to eliminate surface effects and the particle mesh Ewald summation was used to compute long-range electrostatics. The NAMD program v2.11 was used for all classical MD simulations.48 QM/MM Optimizations. The ChemShell program v3.6.052,53 was used for all QM/MM calculations. The built-in DL_POLY54 code was employed for the MM part and Turbomole6655 was used for the QM region. For the S−T energy gap calculations, 10 snapshots were randomly selected from the last 1 ns of the MD trajectories of Ph2C with 2BuOH, the MeCN/2-BuOH mixture, and the MeCN/H2O mixture (Table S1, Supporting Information). All solvent molecules located farther than a 25 Å radius from the carbene center were deleted. We note that, although the initial structures contained a racemic mixture of S-2-BuOH and R-2-BuOH, the QM/MM snapshots did not necessarily have an equal number of 2-BuOH isomers, which should not affect the quality of the results. The resulting droplet was partitioned into an active region (all atoms within 15 Å of the carbene center) and a frozen region (the remaining area of the system outside the active region). Ph2C was included in the QM region (B3LYP-D3/def2-SVP level of theory)43−45,56 and all solvent molecules were treated at the classical level using the Chemistry at HARvard Macromolecular Mechanics force field.49 Geometry optimizations were performed using the DLFIND code57 of ChemShell. The same initial structure was used for both singlet (S) and triplet (T) optimizations. Single point energy calculations at the B3LYP-D3/def2-TZVP level of theory were performed on the optimized structures and the S−T energy gap was calculated using the expression

Ph2C to a different degree and disclose the interactions underlying the pursued reaction mechanisms.



EXPERIMENTAL SECTION

Spectroscopy Setup and Sample Preparation. Details on the experimental setup and the measurement routine have been described elsewhere.21 Briefly, the apparatus allows us to record transient absorption data of Ph 2 C generated via UV photolysis of diphenyldiazomethane (Ph2CN2, synthesized as outlined previously19,21), i.e., the diazo precursor. All radiation employed originates from a regenerative Ti:Sapphire amplifier system (1 kHz, 800 nm, 100 fs), the output of which is used to pump a commercial noncollinear optical parametric amplifier generating 285 nm pump pulses (130 nJ, 40 fs at the sample position) as well as to produce supercontinuum probe pulses in a CaF2 plate.39,40 The pump and probe beams are focused to diameters of 50 and 40 μm at the sample position, respectively, under magic angle polarization configuration (54.7°).39,41 Recording changes in optical density from pairs of subsequent laser shots is enabled using a charge coupled device-based grating spectrometer with an acquisition rate of 1 kHz, mechanically chopping every second pump pulse, and perpetually pumping the sample volume through a flow cell (quartz, 200 μm path length) to exchange the probed sample volume between subsequent pump− probe pairs. Sample solutions investigated in Figure 2a,c were prepared by dissolving 22 mg of Ph2CN2 either in 20 mL anhydrous MeCN or in a mixture of 20 mL anhydrous MeCN and 200 μL Millipore water. The latter solution provides a MeCN/H2O volume mixing ratio of approximately 99:1 (v/v). The sample solution investigated in Figure 2b was prepared by dissolving Ph2CN2 in 19.8 mL MeCN and 200 μL MeOH. The solution provides a MeCN/ MeOH volume mixing ratio of 99:1 (v/v). The samples investigated in Figure 3 were prepared by dissolving Ph2CN2 in 20 mL anhydrous 2-BuOH or 20 mL MeOH. Due to short measurement times, photolysis of the precursor was kept to a minimum, which was verified by observing only minor changes between linear absorption spectra recorded before and after the measurement routine. To account for slightly different sample concentrations, absorption change signals of all data sets were scaled to the very early dynamics of the excited precursor at 335 nm observed in 2-BuOH. Additionally, the chirp of the supercontinuum probe was removed from transient absorption data.39,42 Computational Details. Force Field Parametrization. The force field parameters for Ph2C and MeCN were taken from previous studies.21 Harmonic dihedral restraints were applied to keep the relative orientation of the phenyl rings. Dihedral restrains were applied to the bonds connecting the carbene center (using the B3LYP-D3/def2-TZVP43−47 geometry for reference angles), with a force constant of 2.0 kcal/mol. The restraints were supplied using the “extra bonds” feature implemented in the nanoscale molecular dynamics (NAMD) program.48 The parameters for 2-BuOH were generated using the CGenFF tool.49−51 The same parameters were used for both S-2-BuOH and R-2-BuOH. MD Simulations. Ph2C was solvated in cubic boxes (with a minimum thickness of 30 Å in all directions), representing different solvent environments; (i) a racemic mixture of S-2-BuOH and R-2BuOH (i.e., the system contained equal amounts of S-2-BuOH and R2-BuOH molecules), (ii) 99% (v/v) MeCN + 0.5% S-2-BuOH + 0.5% R-2-BuOH and (iii) 99% MeCN + 1% water. In all cases, 5000 steps of energy minimization followed by 100 ps NVT MD simulation and 500 ps NPT MD simulation runs were performed for equilibration (1 fs time step, the positions of Ph2C atoms were fixed in space). The systems were then subjected to production MD runs with a time step of 2 fs and the position restraints removed. The harmonic dihedral restraints were kept. The MD simulations were performed in all cases using the Langevin thermostat at 300 K. For the constant pressure ensembles, the Langevin pressure coupling was used at 1 atm. Shortrange Lennard-Jones and Coulomb interactions were cutoff at 10 Å (pair list cutoff = 14 Å) and a switching function was used between 9 and 10 Å to avoid sudden jumps in the potential and forces at the

ΔE(S − T) = < ES,QM − E T,QM>

(1)

where ES,QM is the QM energy of 1Ph2C and ET,QM is the QM energy of 3Ph2C. Since we use an electrostatic embedding, these energies include the contribution of the MM point charges. The ΔE(S−T) values were averaged over 10 snapshots. QM/MM MD Simulations. Snapshots from the classical MD simulations were used for QM/MM MD simulations. In general, the setup was similar to that of QM/MM optimizations. However, a solvent layer of 20 Å was used and spherical boundary conditions were employed to restrain all molecules to the spherical volume (r = 20 Å). Unlike in the QM/MM optimizations, the system was not partitioned into active and frozen regions. The Nose−Hover thermostat was employed for keeping the temperature at 300 K. The time step was 1 fs and no cutoffs were used. The simulations were performed for 20 ps or until the completion of the reaction. The B3LYP-D3/def2-SVP level of theory was used for the QM region.



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.joc.9b01228. Details on diffusion models, histogram of solvent molecules near the carbene center, table with S−T gaps, Cartesian coordinates for the QM calculations in the different environments (PDF)



AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected] (E.S.-G.). *E-mail: [email protected] (P.N.). ORCID

Wolfram Sander: 0000-0002-1640-7505 Elsa Sanchez-Garcia: 0000-0002-9211-5803 Patrick Nuernberger: 0000-0002-4690-0229 F

DOI: 10.1021/acs.joc.9b01228 J. Org. Chem. XXXX, XXX, XXX−XXX

Article

The Journal of Organic Chemistry Present Addresses

(13) Sitzmann, E. V.; Langan, J. G.; Griller, D.; Eisenthal, K. B. Effects of solvent polarity and structure on intersystem crossing in diphenylcarbenes. A picosecond laser study on dimesitylcarbene. Chem. Phys. Lett. 1989, 161, 353−360. (14) Chateauneuf, J. E. Picosecond spectroscopic detection of diphenylcarbenium ion in the photolysis of diphenyldiazomethane in aliphatic alcohols. J. Chem. Soc. Chem. Commun. 1991, 1437−1438. (15) Steenken, S. Production of carbenium ions from carbenes by protonation. Pure Appl. Chem. 1998, 70, 2031−2038. (16) Portella-Oberli, M. T.; Jeannin, C.; Soep, B.; Zerza, G.; Chergui, M. Femtosecond study of the rise and decay of carbenes in solution. Chem. Phys. Lett. 1998, 296, 323−328. (17) Peon, J.; Polshakov, D.; Kohler, B. Solvent reorganization controls the rate of proton transfer from neat alcohol solvents to singlet diphenylcarbene. J. Am. Chem. Soc. 2002, 124, 6428−6438. (18) Wang, J.; Kubicki, J.; Peng, H.; Platz, M. S. Influence of Solvent on Carbene Intersystem Crossing Rates. J. Am. Chem. Soc. 2008, 130, 6604−6609. (19) Costa, P.; Sander, W. Hydrogen Bonding Switches the Spin State of Diphenylcarbene from Triplet to Singlet. Angew. Chem. 2014, 126, 5222−5225. (20) Costa, P.; Fernandez-Oliva, M.; Sanchez-Garcia, E.; Sander, W. The Highly Reactive Benzhydryl Cation Isolated and Stabilized in Water Ice. J. Am. Chem. Soc. 2014, 136, 15625−15630. (21) Knorr, J.; Sokkar, P.; Schott, S.; Costa, P.; Thiel, W.; Sander, W.; Sanchez-Garcia, E.; Nuernberger, P. Competitive solventmolecule interactions govern primary processes of diphenylcarbene in solvent mixtures. Nat. Commun. 2016, 7, No. 12968. (22) Reichardt, C.; Welton, T. Solvents and Solvent Effects in Organic Chemistry; Wiley-VCH, 2010. (23) Marcus, Y. The properties of organic liquids that are relevant to their use as solvating solvents. Chem. Soc. Rev. 1993, 22, 409−416. (24) Ortega, J.; Ràfols, C.; Bosch, E.; Rosés, M. Solute−solvent and solvent−solvent interactions in binary solvent mixtures. Part 3. The ET(30) polarity of binary mixtures of hydroxylic solvents. J. Chem. Soc., Perkin Trans. 2 1996, 2, 1497−1503. (25) Wang, J.; Kubicki, J.; Hilinski, E. F.; Mecklenburg, S. L.; Gustafson, T. L.; Platz, M. S. Ultrafast Study of 9-Diazofluorene: Direct Observation of the First Two Singlet States of Fluorenylidene. J. Am. Chem. Soc. 2007, 129, 13683−13690. (26) Collins, F. C.; Kimball, G. E. Diffusion-controlled reaction rates. J. Colloid Sci. 1949, 4, 425−437. (27) Rice, S. A. Diffusion-Limited Reactions; Elsevier Science, 1985. (28) Lakowicz, J. R. Topics in Fluorescence Spectroscopy - Volume 2: Principles; Kluwer Academic Publishers, 2002. (29) Megerle, U.; Wenninger, M.; Kutta, R. J.; Lechner, R.; König, B.; Dick, B.; Riedle, E. Unraveling the flavin-catalyzed photooxidation of benzylic alcohol with transient absorption spectroscopy from subpico- to microseconds. Phys. Chem. Chem. Phys. 2011, 13, 8869−8880. (30) Kumpulainen, T.; Lang, B.; Rosspeintner, A.; Vauthey, E. Ultrafast Elementary Photochemical Processes of Organic Molecules in Liquid Solution. Chem. Rev. 2017, 117, 10826−10939. (31) Arita, T.; Kajimoto, O.; Terazima, M.; Kimura, Y. Experimental verification of the Smoluchowski theory for a bimolecular diffusioncontrolled reaction in liquid phase. J. Chem. Phys. 2004, 120, 7071− 7074. (32) Sveshnikov, B. Y. The Quenching of the luminescence of solutions by foreign substances. Sov. Phys. Usp. 1962, 4, 776. (33) Kirmse, W.; Guth, M.; Steenken, S. Production of αSiloxycarbenium Ions by Protonation of Photochemically Generated α-Siloxycarbenes. Formation Mechanism and Reactivities with Nucleophiles. J. Am. Chem. Soc. 1996, 118, 10838−10849. (34) Ammer, J.; Sailer, C. F.; Riedle, E.; Mayr, H. Photolytic Generation of Benzhydryl Cations and Radicals from Quaternary Phosphonium Salts: How Highly Reactive Carbocations Survive Their First Nanoseconds. J. Am. Chem. Soc. 2012, 134, 11481−11494. (35) Fingerhut, B. P.; Sailer, C. F.; Ammer, J.; Riedle, E.; de VivieRiedle, R. Buildup and Decay of the Optical Absorption in the

#

Department of Chemistry, University of Ottawa, Ottawa, ON, Canada K1N 6N5 (P.C.). ⊥ Institut für Physikalische und Theoretische Chemie, Universität Regensburg, Universitätsstraße 31, 93053 Regensburg, Germany (P.N.). ∥ Faculty of Allied Health Sciences, Chettinad Academy of Research and Education, Kelambakkam 603103, India (P.S.). Author Contributions ∇

J.K. and P.S. contributed equally to this work.

Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS



REFERENCES

We thank Dr. Sebastian Schott and Prof. Dr. Tobias Brixner (University of Würzburg) for their support and Dr. Joel Mieres-Perez (University of Duisburg-Essen) for useful discussions. We acknowledge funding by the German Research Foundation under Germany′s Excellence Strategy EXC-2033 (Project ID:390677874) and the Emmy-Noether program (P.N., grant number NU 263/3-1). E.S.-G. acknowledges the support of the Boehringer Ingelheim Foundation (Plus-3 Program) and the computational time provided by the Computing and Data Facility of the Max Planck Society.

(1) Eisenthal, K. B.; Turro, N. J.; Aikawa, M.; Butcher, J. A.; DuPuy, C.; Hefferon, G.; Hetherington, W.; Korenowski, G. M.; McAuliffe, M. J. Dynamics and energetics of the singlet-triplet interconversion of diphenylcarbene. J. Am. Chem. Soc. 1980, 102, 6563−6565. (2) Platz, M. S.; Senthilnathan, V. P.; Wright, B. B.; McCurdy, C. W. Reactions of triplet diphenylcarbene by hydrogen atom tunneling in rigid media. J. Am. Chem. Soc. 1982, 104, 6494−6501. (3) Sitzmann, E. V.; Langan, J.; Eisenthal, K. B. Picosecond laser studies of the charge-transfer reaction of excited triplet diphenylcarbene with electron donors. Chem. Phys. Lett. 1983, 102, 446−450. (4) Wang, Y.; Sitzmann, E. V.; Novak, F.; Dupuy, C.; Eisenthal, K. B. Reactions of excited triplet diphenylcarbene studied with picosecond lasers. J. Am. Chem. Soc. 1982, 104, 3238−3239. (5) Sitzmann, E. V.; Wang, Y.; Eisenthal, K. B. Picosecond laser studies on the reaction of excited triplet diphenylcarbene with alcohols. J. Phys. Chem. A 1983, 87, 2283−2285. (6) Sitzmann, E. V.; Langan, J.; Eisenthal, K. B. Intermolecular effects on intersystem crossing studied on the picosecond timescale: the solvent polarity effect on the rate of singlet-to-triplet intersystem crossing of diphenylcarbene. J. Am. Chem. Soc. 1984, 106, 1868−1869. (7) Griller, D.; Nazran, A. S.; Scaiano, J. C. Reaction of diphenylcarbene with methanol. J. Am. Chem. Soc. 1984, 106, 198− 202. (8) Eisenthal, K. B.; Moss, R. A.; Turro, N. J. Divalent Carbon Intermediates: Laser Photolysis and Spectroscopy. Science 1984, 225, 1439−1445. (9) Hadel, L. M.; Platz, M. S.; Scaiano, J. C. Study of hydrogen atom abstraction reactions of triplet diphenylcarbene in solution. J. Am. Chem. Soc. 1984, 106, 283−287. (10) Leyva, E.; Barcus, R. L.; Platz, M. S. Reinvestigation of the chemistry of arylcarbenes in polycrystalline alcohols at 77K. Secondary photochemistry of matrix-isolated carbenes. J. Am. Chem. Soc. 1986, 108, 7786−7788. (11) Turro, N. J.; Cha, Y.; Gould, I. R. Temperature dependence of the reactions of singlet and triplet diphenylcarbene. Tetrahedron Lett. 1985, 26, 5951−5954. (12) Langan, J. G.; Sitzmann, E. V.; Eisenthal, K. B. Inverse deuterium isotope effect in the intersystem crossing of diphenylcarbene. Chem. Phys. Lett. 1986, 124, 59−62. G

DOI: 10.1021/acs.joc.9b01228 J. Org. Chem. XXXX, XXX, XXX−XXX

Article

The Journal of Organic Chemistry Ultrafast Photo-Generation and Reaction of Benzhydryl Cations in Solution. J. Phys. Chem. A 2012, 116, 11064−11074. (36) Sailer, C. F.; Thallmair, S.; Fingerhut, B. P.; Nolte, C.; Ammer, J.; Mayr, H.; Pugliesi, I.; de Vivie-Riedle, R.; Riedle, E. A Comprehensive Microscopic Picture of the Benzhydryl Radical and Cation Photogeneration and Interconversion through Electron Transfer. ChemPhysChem 2013, 14, 1423−1437. (37) Riedle, E.; Roos, M. K.; Thallmair, S.; Sailer, C. F.; Krebs, N.; Fingerhut, B. P.; de Vivie-Riedle, R. Ultrafast photochemistry with two product channels: Wavepacket motion through two distinct conical intersections. Chem. Phys. Lett. 2017, 683, 128−134. (38) Kortüm, G.; Dreesen, G. Ü ber die Konstitutionsabhängigkeit der Schwingungsstruktur im Absorptionsspektrum von aromatischen Kohlenwasserstoffen. Chem. Ber. 1951, 84, 182−203. (39) Megerle, U.; Pugliesi, I.; Schriever, C.; Sailer, C. F.; Riedle, E. Sub-50 fs broadband absorption spectroscopy with tunable excitation: putting the analysis of ultrafast molecular dynamics on solid ground. Appl. Phys. B 2009, 96, 215−231. (40) Marefat Khah, A.; Grimmelsmann, L.; Knorr, J.; Nuernberger, P.; Hättig, C. How a linear triazene photoisomerizes in a volumeconserving fashion. Phys. Chem. Chem. Phys. 2018, 20, 28075−28087. (41) Schott, S.; Steinbacher, A.; Buback, J.; Nuernberger, P.; Brixner, T. Generalized magic angle for time-resolved spectroscopy with laser pulses of arbitrary ellipticity. J. Phys. B: At., Mol. Opt. Phys. 2014, 47, No. 124014. (42) Schott, S.; Ress, L.; Hrušaḱ , J.; Nuernberger, P.; Brixner, T. Identification of photofragmentation patterns in trihalide anions by global analysis of vibrational wavepacket dynamics in broadband transient absorption data. Phys. Chem. Chem. Phys. 2016, 18, 33287− 33302. (43) Becke, A. D. Density-functional thermochemistry. III. The role of exact exchange. J. Chem. Phys. 1993, 98, 5648−5652. (44) Lee, C.; Yang, W.; Parr, R. G. Development of the ColleSalvetti correlation-energy formula into a functional of the electron density. Phys. Rev. B 1988, 37, 785−789. (45) Grimme, S.; Ehrlich, S.; Goerigk, L. Effect of the damping function in dispersion corrected density functional theory. J. Comput. Chem. 2011, 32, 1456−1465. (46) 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. (47) Vosko, S. H.; Wilk, L.; Nusair, M. Accurate spin-dependent electron liquid correlation energies for local spin density calculations: a critical analysis. Can. J. Phys. 1980, 58, 1200−1211. (48) Phillips, J. C.; Braun, R.; Wang, W.; Gumbart, J.; Tajkhorshid, E.; Villa, E.; Chipot, C.; Skeel, R. D.; Kalé, L.; Schulten, K. Scalable molecular dynamics with NAMD. J. Comput. Chem. 2005, 26, 1781− 1802. (49) Vanommeslaeghe, K.; Hatcher, E.; Acharya, C.; Kundu, S.; Zhong, S.; Shim, J.; Darian, E.; Guvench, O.; Lopes, P.; Vorobyov, I.; Mackerell, A. D., Jr CHARMM general force field: A force field for drug-like molecules compatible with the CHARMM all-atom additive biological force fields. J. Comput. Chem. 2010, 31, 671−690. (50) Vanommeslaeghe, K.; MacKerell, A. D. Automation of the CHARMM General Force Field (CGenFF) I: Bond Perception and Atom Typing. J. Chem. Inf. Model. 2012, 52, 3144−3154. (51) Vanommeslaeghe, K.; Raman, E. P.; MacKerell, A. D. Automation of the CHARMM General Force Field (CGenFF) II: Assignment of Bonded Parameters and Partial Atomic Charges. J. Chem. Inf. Model. 2012, 52, 3155−3168. (52) Sherwood, P.; de Vries, A. H.; Guest, M. F.; Schreckenbach, G.; Catlow, C. R. A.; French, S. A.; Sokol, A. A.; et al. QUASI: A general purpose implementation of the QM/MM approach and its application to problems in catalysis. J. Mol. Struct.: THEOCHEM 2003, 632, 1− 28. (53) Metz, S.; Kästner, J.; Sokol, A. A.; Keal, T. W.; Sherwood, P. ChemShell-a modular software package for QM/MM simulations: ChemShell. WIREs Comput. Mol. Sci. 2014, 4, 101−110.

(54) Todorov, I. T.; Smith, W. DL_POLY_3: the CCP5 national UK code for molecular-dynamics simulations. Philos. Trans. R. Soc., A 2004, 362, 1835−1852. (55) Ahlrichs, R.; Bär, M.; Häser, M.; Horn, H.; Kölmel, C. Electronic structure calculations on workstation computers: The program system turbomole. Chem. Phys. Lett. 1989, 162, 165−169. (56) Eichkorn, K.; Weigend, F.; Treutler, O.; Ahlrichs, R. Auxiliary basis sets for main row atoms and transition metals and their use to approximate Coulomb potentials. Theor. Chem. Acc. 1997, 97, 119− 124. (57) Kästner, J.; Carr, J. M.; Keal, T. W.; Thiel, W.; Wander, A.; Sherwood, P. DL-FIND: An Open-Source Geometry Optimizer for Atomistic Simulations. J. Phys. Chem. A 2009, 113, 11856−11865.

H

DOI: 10.1021/acs.joc.9b01228 J. Org. Chem. XXXX, XXX, XXX−XXX