Molecular Simulation of the Catalytic Regeneration of nBuLi through a

Article Cite This: Inorg. Chem. XXXX, XXX, XXX−XXX

pubs.acs.org/IC

Molecular Simulation of the Catalytic Regeneration of nBuLi through a Hydrometalation Route Mal-Soon Lee,† Vassiliki-Alexandra Glezakou,*,† Roger Rousseau,† and B. Peter McGrail‡ †

Physical and Computational Science Directorate and ‡Energy and Environment Directorate, Pacific Northwest National Laboratory, Richland, Washington 99352, United States

Downloaded via UNIV DE BARCELONA on February 7, 2019 at 03:31:44 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

S Supporting Information *

ABSTRACT: Efficient regeneration of organolithium compounds is a challenging aspect in the process of novel organometathetical catalytic cycles. One of these catalytic cycles is a newly suggested method for Mg production from seawater that capitalizes on the rich chemistry of Grignard reagents. The proposed three-step catalytic cycle with Cp2MClL catalyst (M = Ti, Zr; L = select organic ligands) requires the regeneration of nBuLi from Li(s), butene, and H2. The potential of this approach is evaluated with density functional theory-based molecular simulations. The results reveal that the high affinity of Li toward Cl and N results in the formation of alkanes, and the strong coupling between the catalyst and BuLi leads to catalyst deactivation. To improve its catalytic performance, we proposed the use of a diamine cocatalyst and a modified catalyst with a ligand that does not contain N, which would help release BuLi from the vicinity of the catalytic center. Ab initio molecular dynamics simulations at 298 K in explicit solvent (THF) were used to estimate the Gibbs free energetics and equilibrium constants obtained from the vibrational density of states using velocity autocorrelation functions. The results show a marked improvement in the free energetics with lower barriers toward the completion of the catalytic cycle and suppression of deactivation channels.

■

INTRODUCTION

central to the COMET process but also the most challenging aspect. The standard preparation for nBuLi is the reaction of 1bromobutane or 1-chlorobutane with Li metal.3 This reaction is often catalyzed by 1−3% sodium, and the typical solvents used include benzene, cyclohexane, and diethyl ether. When BuBr is the precursor, the product is a homogeneous solution, consisting of a mixed cluster containing both LiBr and nBuLi, together with a small amount of octane. nBuLi forms a weaker complex with LiCl, so that the reaction of BuCl with Li produces a precipitate of LiCl, which can be easily isolated. Using this industrial process to regenerate nBuLi would require the conversion of 1-butene to chloro- or bromobutane. This can in theory be achieved in an atom-economical reaction by addition of hydrochloric acid (HCl) to 1-butene, but this process yields 2-chlorobutane as the major product and 1chlorobutane as a minor product. Quantitative conversion of 1butene to 1-chlorobutane requires a two-step process consisting of hydroboration of the olefin followed by chlorination.4 The limitation of this process is the formation of boric acid as a stoichiometric byproduct, which needs to be reduced to borane, thus adding more unit operations and, consequently, higher energy consumption. The reaction of Li

Magnesium alloys are used in a large range of applications for automotive, aerospace, electronics, etc., due to their light weight, vibration damping capacity, high stiffness, and high recycling capacity. Commercial production of magnesium metal is presently performed through two methods: (1) molten salt electrolysis of MgCl2 and (2) the Pidgeon process reacting ferrosilicon with MgO. Both processes are energy intensive, consuming 5−7 times the theoretical minimum per ton of Mg produced. A research effort at PNNL was undertaken to develop and demonstrate a new process for production of magnesium metal, the catalyzed organo-metathetical (COMET) process for magnesium production from seawater (Figure 1).1 A techno-economic analysis shows that this process can significantly reduce the cost and energy requirements to produce magnesium metal.2 Through this process, anhydrous MgCl2 is converted to dibutylmagnesium (MgBu2) upon addition of nbutyllithium (nBuLi). MgBu2 is then decomposed to magnesium, hydrogen, and butene. Lithium metal is generated from electrolysis of LiCl that is also generated as a byproduct. Key to the most efficient and costeffective process is recovery of lithium, hydrogen, and butene, which are then catalytically converted to nBuLi. Efficient regeneration of nBuLi, the key reactant for the critical metathetical exchange reaction to produce MgBu2, is thus © XXXX American Chemical Society

Received: October 12, 2018

A

DOI: 10.1021/acs.inorgchem.8b02910 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

Figure 1. Process flow diagram for catalyzed organo-metathetical process for Mg production.

with alkyl halide (1-chlorobutane) yields nBuLi and an equimolar amount of LiCl. The resulting LiCl would have to be recycled to the electrolyzer step, thus making it energy intensive. An alternative process that does not generate LiCl in this step would be clearly optimal for COMET. For this purpose, we propose utilization of Li metal produced from electrolysis of LiCl to regenerate nBuLi via a two-step catalyzed reaction process taking advantage of the hydrogen and 1-butene from the MgBu2 decomposition reaction: Li(s) + 1/2H 2(g) → LiH(s)

(1)

LiH(s) + C4H8(g) → BuLi

(2)

in comparison with other metal chlorides such as TiCl4, FeCl3, and CoCl2.7 On this basis, we formulated a catalytic cycle consisting of three reaction steps with Cp2TiCl2 as a catalyst (see Figure 2),

This process is expected to produce nBuLi without stoichiometric byproducts in the presence of a suitable catalyst. The first step is the reduction of Li in the presence of H2 to yield lithium hydride (eq 1). Howie et al. synthesized lithium hydride (LiH) by compressing elemental lithium and hydrogen.5 They observed that LiH remains stable in the presence of molecular hydrogen and no further reaction between LiH and H2 is observed to form higher hydrogen content metal hydrides at 300 K and up to 160 GPa. This precedent encouraged us to consider a catalytic route to generate butyllithium. The second step is the catalytic reaction of LiH with 1-butene to yield nBuLi, a process known as hydrometalation (eq 2). There is no literature precedent for hydrometalation of alkenes using alkali-metal hydrides. However, Ashby and Smith showed that an alkali-earth-metal hydride (MgH2) undergoes hydrometalation with alkenes in the presence of bis(cyclopentadienyl)titanium dichloride (Cp2TiCl2) catalyst.6 Because organo-Mg and organo-Li have similar chemistries, we propose that hydrometalation of alkenes using LiH may proceed in a fashion analogous to that for MgH2. Studies by Ashby and Noding showed that while some reactivity was reported for transition-metal chlorides in the addition of lithium hydride to 1-octene, Cp2TiCl2 was the only catalyst reported to show higher activity

Figure 2. Proposed catalytic cycle to generate nBuLi.

where transmetalation with LiH is used to generate a titanium hydride followed by hydrometalation. These steps are similar to those during alkene polymerization and are expected to be facile. In what follows, we present mechanistic studies of the proposed catalytic cycle toward favorable formation of nBuLi from 1-butene and LiH by performing first-principles molecular simulations. We also examine the effect of solvents and cocatalyst on the overall catalytic cycle by explicitly including solvent molecules in the simulations. The proposed catalyst shows promise for improved catalytic performance during the cycle.

■

COMPUTATIONAL METHODS

As a first step to study nBuLi generation, we performed quantum chemical calculations where all possible reaction intermediates B


Inorganic Chemistry

l o o ℏω ji ℏω zyz z D(ω)o cothjjj m j 2kBT zz o 2 k T o k { n B ÄÅ ÉÑ| ÅÅ Ñ ij ℏω yzÑÑo Å zzÑÑo − logÅÅÅ2 sinhjjj } dω j 2kBT zzÑÑÑo ÅÅ o k {ÑÖo ÅÇ ~

involved in the catalytic cycle with different catalysts were optimized using the Gaussian 09 software package.8 Optimizations were performed using density functional theory (DFT) with a PBE0 hybrid exchange-correlation functional9−11 along with the 6-31G** basis sets12−14 for hydrogen, carbon, nitrogen, phosphorus, and chlorine and the Stuttgart RSC 1997 basis sets and corresponding effective core potentials for titanium and zirconium, obtained from the EMSL basis set exchange.15 Harmonic vibrational frequencies were calculated at the optimized geometries using the same level of theory to estimate the zero-point energy (ZPE) and the thermal contributions (298 K and 1 atm) to the free energy. To study solvent effects, we also optimized structures in tetrahydrofuran (THF), toluene, 1,4-dioxane, and diethyl ether using the polarizable continuum model (PCM)16 of solvation. The quantum chemistry calculations were performed not only to obtain reasonable starting structures for subsequent molecular dynamics simulations but also to make an initial assessment of the reactive intermediates. The continuum calculations were also done to assess the effects of the dielectric on the catalytic cycle for the various systems. After optimization, the thermodynamic stability of intermediates in the catalytic cycle was studied via climbing image nudged elastic band (CI-NEB) simulations17 and obtained energy barriers that connect the catalytic steps. These calculations were performed with the CP2K package18−20 using the Perdew, Burke, and Ernzerhof (PBE) exchange-correlation functional.10,11 Grimme’s third-generation corrections (DFT-D3)21 were used to account for dispersion. The core electrons were described by the norm-conserving pseudopotentials,22 while the valence wave functions were expanded in terms of double-ζquality basis sets.23 An additional auxiliary plane wave basis set with a 400 Ry cutoff was used for the calculation of the electrostatic terms. For the CI-NEB minimization, molecular-dynamics-based optimization was employed at each reaction path, where simulations were started at 20 K followed by slow annealing with convergence criteria of maximum displacement of 0.01 au, ∼0.005 Å and maximum force of 0.002 au, ∼2.2 kcal/(mol Å). Given the difference between the underlying theoretical and computational schemes of quantum chemical and condensed phase codes (e.g., G09 and CP2K), in section S1 of the Supporting Information we have included a detailed comparison of structures and reaction energetics of selected systems that showcases the qualitative consistency between these approaches. The enhancement of nBuLi regeneration was investigated with ab initio molecular dynamics (AIMD) simulations conducted within the canonical NVT ensemble at 298 K with a time step of 1.0 fs with the most probable catalyst/cocatalyst in an explicit solvent. Simulations were performed using a cubic cell with a side length of 25.52 Å, which contains ∼1100 atoms consisting of catalyst, cocatalyst, BuLi, and 80 THF molecules. The XYZ coordinates of the optimized structure of this system can be found in section S3 in the Supporting Information. In general, each of the trajectories was equilibrated for 20−25 ps, where equilibration was confirmed by monitoring the potential energy to reach a plateau and fluctuate about an average value. Additionally, we monitored the velocity distribution to make sure that all atom types exhibited a Maxwell−Boltzmann distribution. Statistical data were compiled from an additional 30 ps of trajectories after equilibration. The velocities obtained from these trajectories were used to calculate the vibrational density of states (VDOS), using the Fourier transform of the velocity autocorrelation function:

D(ω) =

∫0

∞

e−iωt ⟨v(τ )· v(τ + t )⟩ dt

S = 3kB

∫0

Article

∞

(4)

where D(ω) is the vibrational density of states calculated with eq 3. This allows us to capture low-frequency collective dynamics from the liquid. Previous studies from this group have shown that the QHA provides very good estimates of entropic contributions, especially where solvent effects can be important.24,26

■

RESULTS AND DISCUSSION We performed quantum chemical calculations to obtain the energetics of the critical steps of the catalytic cycle with five different catalysts, shown in Figure 3. The effect of different

Figure 3. Catalytic systems examined with various ligands.

solvents was screened in the form of a continuum dielectric on the overall catalytic cycle. The proposed catalytic scheme (Figure 2), consisting of three basic reactions (reactions 1−3), was evaluated by determining the relative free energies for each reaction step. We first examined the catalytic system of Cp2TiCl2 (Figure 3, 1) since it has been shown to be effective for hydrometalation.7 From enthalpy (ΔH) calculations, we find that reactions 1 and 2 are exothermic (Table 1). The overall trend, with or without solvent, shows significant stabilization of reaction 1 and only a modest stabilization for reactions 2 and 3 (ΔG = −4 to −5 kcal/mol; ΔH = −17 to −18 kcal/mol). Reaction 3, the catalyst regeneration step, results in stabilization of Cp2TiHCl and BuLi by ΔG ≈ −9 kcal/mol in the gas phase. In the presence of solvent, reaction 3 becomes less favorable. Although the high affinity of Li/Cl helps the initial complexation of LiH to the Cp2TiCl2 to form Cp2TiHCl, it can also lead to formation of alkanes in addition to alkenes, as has been observed experimentally. One possible route to explain this is shown in Figure 4. The addition of 2 equiv of butyllithium is a known reaction, which results in the decomposition of the in situ formed dibutyltitanocene.28 This decomposition produces 1 equiv of butene and 1 equiv of butane. Quantum chemical calculations for the reaction of butyltitanocene and butyllithium to give dibutyltitanocene and lithium chloride indeed show a decrease in the free energies by ∼17−30 kcal/mol in different solvents, indicating strong stabilization of dibutyltitanocene, as seen in experiments.28 The resulting catalyst may no longer be able to perform further additions, thus limiting the amount of alkane formed through this route. In addition, we examined the alkane formation by evaluating the reactivity of monochloride titanocenes with 1-butene and lithium hydride. Similar reactivity in comparison to Cp2TiCl2 was observed computationally. This indicates that alkane formation is still favorable with the modified ligand. While the previously invoked mechanism for butane formation is still possible by displacement of the L ligand by BuLi, another route that may potentially lead to formation of butane is the acid/base

(3)

where v is the velocity and the angular brackets indicate the statistical average over time. The relative stability at 298 K is evaluated from the Gibbs free energy difference (ΔG) using ΔG = ΔH − TΔS, where ΔH is the average energy difference at 298 K corrected for the PV term by subtracting the RT term24 and ΔS is the entropy difference between the two end states. The entropy for each state was obtained using a quasi-harmonic approximation (QHA):24−27 C


Article

Inorganic Chemistry

Table 1. Free Energy and Enthalpy (in Parentheses) Changes (kcal/mol) at 298 K for the Catalytic Cycle Shown in Figure 2 with Cp2TiCl2 Catalysts in the Gas Phase and in Different Solvents entry

catalyst

reaction

gas phase

THF

toluene

Et2O

1,4-dioxane

1

Cp2TiCl2

1 2 3

−33.9 (−32.6) −4.5 (−17.8) −8.6 (−4.9)

−34.6 (−33.6) −4.2 (−16. 6) 1.8 (4.9)

−33.3 (−33. 9) −5.2 (−16.5) −1.4 (0.1)

−34.4 (−33.5) −4.0 (−16. 8) 0.3 (3.3)

−33.3 (−33.9) −5.2 (−16.5) −1.7 (−0.3)

Figure 4. Alkene and alkane formations in the presence of alkyl lithium during the catalytic cycle.

Table 2. Free Energy Changes (kcal/mol) at 298 K for the Catalytic Cycle (Figure 2) with Modified Catalysts in the Gas Phase and in Different Solvents entry

catalyst

2

Cp2ZrHNC4

3

Cp2TiCl(NPPh3)

4

2,4-Me2Cp2TiClNC4(CH3)2

5

2,5-Me2Cp2TiClNC4(CH3)2

reacn

gas phase

THF

toluene

Et2O

1,4-dioxane

2 3 1 2 3 1 2 3 1 2 3

−9.3 −3.9 −29.2 −0.1 −13.0 −33.0 −0.7 −12.5 −36.9 3.6 −16.8

−5.6 3.6 −31.4 4.0 −6.0 −34.7 −0.4 −1.6 −38.8 4.8 −6.8

−7.2 0.7 −30.3 2.5 −9.1 −34.1 −0.5 −6.0 −38.1 4.5 −11.1

−6.1 2.4 −28.7 1.2 −5.0 −34.2 −0.7 −3.0 −38.5 4.6 −8.4

−7.4 0.4 −30.5 2.7 −9.6 −34.0 −0.5 −6.5 −38.0 4.5 −11.5

Figure 5. Free energy reaction profile for reactions 1-3 for catalyst 3 in THF. Structures at the transition state are also shown. Energies are given in kcal/mol. Color code of circles: Ti (green), Li (gray), C (black), H (white), N (blue).

reaction between BuLi and the intermediate titanium hydride species proposed during the catalytic cycle, as shown in Figure 4b. The free energetics indicate stabilization of product by ΔG = −29 to −39 kcal/mol in solvents. These observations imply that BuLi near the catalyst results in formation of alkanes through unwanted side reactions, which could potentially deactivate the catalyst. According to the proposed catalytic cycle, one chlorine ligand on the catalyst does not participate in any of the

intermediate steps. Thus, substitution with a stronger binding ligand is proposed to prevent reaction with the formed alkyl lithium and formation of dibutyl titanocene. Steric effects were assessed through a series of catalysts of the type Cp2MClL (M = Ti, Zn) by varying the steric bulk of ligand L, as shown in Figure 3 systems 2−5. Table 2 shows the estimated free energies at 298 K in the gas phase and in different solvents. Reaction 2 for catalyst 2 (in solvent) is exergonic by ΔG = −6 to −7 kcal/mol but endergonic for D


Article

Inorganic Chemistry reaction 3 by ΔG = 1−4 kcal/mol, indicating that BuLi regeneration is not favorable. Thus, we exclude catalyst 2 from further investigation. In contrast, catalyst 3 in solvent shows stabilization of BuLi formation (reaction 3, ΔG = −5 to −10 kcal/mol) in addition to strong stabilization of reaction 1 (ΔG = −29 to −31 kcal/mol). Structural examinations from molecular simulations with catalyst 3 indicates large steric hindrance between the phenyl rings and the butyl lignad, which further hinders interactions between LiH and the metal center. Catalyst 4 with a 2,4-dimethylpyrrole ligand shows slight stabilization of both reactions 2 and 3. Catalyst 5 with a 2,5-dimethylpyrrole ligand shows endergonic behavior for reaction 2 but strong stabilization for reaction 3. On the basis of these results, catalyst 3 would be the most promising for BuLi generation. We proceed to examine the thermodynamic stability of intermediates, by calculating the energy barriers for each reaction step from CI-NEB calculations. As shown in Figure 5, in contrast to a barrierless reaction 1, large barriers of ∼20 and ∼37 kcal/mol for reactions 2 and 3, respectively, were predicted, indicating that the final step to form nBuLi is the slowest. Conversely, for catalyst 5, we observe an inversion of the relative barriers: the energy barrier of reaction 3 becomes ∼10 kcal/mol in comparison to that for reaction 2, ∼28 kcal/mol. Although catalyst 5 shows lower energy barriers in comparison to those of catalyst 3, molecular dynamics simulations also indicate a strong affinity of Li toward Cl and N (in N-containing ligands), with Li always remaining inside the first coordination shell of the Ti center and the N-containing ligands. Furthermore, the final structure of reaction 3 shows a supercomplex structure where BuLi remains close to the regenerated catalyst. The observation of a strong coupling between the catalyst and BuLi indicates that a modified catalyst may have improved performance. To this end, we proposed a catalyst with N-free ligands to prevent potential interactions between Li and N. The ligand used is 2,4-dimethylcyclopentane, shown in Figure 6a. We also used dimethylaminobutyllithium instead of plain

energy barriers for reaction 2 and 3 that are the ratedetermining steps. In comparison to the original catalyst Cp2TiCl(NPPh3), the barriers determined were ∼38 kcal/mol for reaction 2 and ∼18 kcal/mol for reaction 3. Although the barrier in reaction 2 has increased, reaction 3 shows a significant reduction by about 50%, indicating that use of an amine group may have a favorable effect and render the transmetalation step feasible. On the basis of these encouraging results, we further modified our model system with a carbene ligand and with addition of a third component (1,4-butanediamine) to act as cocatalyst (see Figure 7). To test this idea, we first calculated

Figure 7. Modified catalyst and proposed cocatalyst. The color code for circles is the same as in Figure 5.

the energetics of each reaction step and the associated energy barriers. Energies for each reaction step with/without solvent give behavior similar to that seen with dimethylaminobutyllithium. Using the catalyst having a carbene ligand, we also calculated the barriers for reactions 2 and 3 which are ∼23 and ∼17 kcal/mol, respectively. The barriers are reasonably low for both reactions in comparison to those with N-containing ligands. This gives us confidence to further examine the newly proposed catalyst and cocatalyst by AIMD, which allows us to consider dynamic and dissociative phenomena that may not be well described with continuum solvent models. AIMD simulations of the newly suggested catalyst/cocatalyst system were performed at 298 K in explicit THF solvent in order to evaluate the cocatalyst ability to assist in BuLi removal away from the catalytic center. The simulations were executed using three different starting configurations: (A) the BuLi is still within the first coordination sphere of the catalyst while the cocatalyst is further out, (B) BuLi and cocatalyst are within the first coordination shell, and (C) BuLi and cocatalyst are away from the catalyst as shown in Figure 8. From simulation A, we note that the cocatalyst moves toward BuLi, closing the distance between Li and N of the cocatalyst from 8.7 to 6.2 Å. In simulation B, the Ti−H distance in the catalyst remains the same, indicating that the presence of cocatalyst prevents deprotonation of Ti which further prohibits the formation of butane observed with catalysts as shown in Figure 3. Simulation B also shows an increase in the distance between catalyst and Li, but a decrease in the distance between Li and cocatalyst, indicating that the cocatalyst with an amine group helps in removal of BuLi from the immediate coordination

Figure 6. Proposed new catalyst.

BuLi (Figure 6b) to facilitate the detachment of BuLi from the vicinity of the catalyst. The calculated reaction energies for the catalytic reactions 1−3 with the new catalyst are summarized in Table 3. All of the reactions are also thermodynamically favorable in different dielectric media. We also calculated the Table 3. Computational Evaluation of Reaction Energy (kcal/mol) with Proposed Catalyst without N and Tertiary Diamine reacn

gas phase

THF

toluene

Et2O

1,4-dioxane

1 2 3

−25 −3 −10

−27 −4 −1

−26 −2 −4

−27 −2 −2

−26 −2 −5 E


Article

Inorganic Chemistry

Figure 8. Configurations for AIMD simulations. The Li−N (cocatalyst) and Li−H distances are used as criteria to evaluate the BuLi abstraction using a N-containing cocatalyst. The color code for circles is the same as in Figure 5.

Figure 9. Schematic of thermodynamic quantities for catalyst/cocatalyst systems from AIMD, at T = 298 K.

Table 4. Free Energies (kcal/mol) and Equilibrium Constants Obtained from the AIMD Simulations with THF Solvent at T = 298 K Showing the Effect of Different Ligands and Cocatalyst on the Catalytic Cycle system no cocatalyst cocatalyst, N-ligand cocatalyst, carbene ligand (no N)

reaction

ΔS

ΔH

ΔG

Cat:BuLi → Cat + BuLi Cat:BuLi + Co-cat → Cat:BuLi:co-cat Cat:BuLi:co-Cat → Cat + BuLi:co-cat Cat:BuLi + co-Cat → Cat:BuLi:co-Cat Cat:BuLi:co-Cat → Cat + BuLi:co-Cat

−0.012 −0.007 0.001 0.03 0.005

−13.6 −24.0 23.4 19.8 −26.8

−11.3 −21.9 23.1 10.9 −28.3

sphere of the catalyst. Finally, simulation C indicates that, once the BuLi/cocatalyst complex is formed, it concertedly moves away from the catalyst, increasing the distance between catalyst and Li from 8.7 to 11.2 Å. From the data obtained from the AIMD simulations, we also calculated the vibrational density of states using the Fourier transform of the velocity autocorrelation function (eq 3). This enabled us to obtain estimates of entropy (ΔS) using eq 4 for each of these processes, while the enthalpies (ΔH) are obtained as averages of the total energies at 298 K. We then combined these thermodynamic quantities, to compute the Gibbs free energy using ΔG = ΔH − TΔS, as shown in Figure 9. These values are subsequently used to estimate the equilibrium constant K = exp(−ΔG/RT) = 5.8 × 1020 for the catalyst regeneration. Thus, our free energy estimates suggest a substantial exergonic driver for the capture and removal of the product from the immediate coordination of the catalyst as long as the appropriate cocatalyst is used. To understand the small entropy change at the intermediate step, we also calculated the entropy contributions from each component of the catalyst, cocatalyst, BuLi, and solvent separately. This can be achieved by calculating the projected VDOS for each of these components. The calculations show that the increase in entropy is mainly due to reorganization of the solvent in the system. This finding underscores the importance of AIMD models with explicit solvent, as this

K 1.8 1.2 1.3 3.0 5.8

× × × × ×

108 1016 10−16 10−8 1020

would be lost in the continuum solvent model. We postulate that the endergonic nature of this step can be easily tuned by screening the acidity/basicity through molecular modification of the amine and/or change in the concentration of the cocatalyst. To verify these results, we repeated the calculations with the N-containing catalyst 5 with and without the cocatalyst. The results are summarized in Table 4, and the free energy profile for the different systems is shown in Figure S3 of the Supporting Information. As shown in Figure S2 of the Supporting Information, once the cocatalyst is added, the catalytic system (now with N-containing ligands) forms a highly stable supercomplex. The dissociative state of catalyst and BuLi:cocatalyst complex is an endergonic process. Furthermore, a positive value of ΔG for reaction 2 in Table 4 indicates that the cocatalyst will not be able to change the chemistry enough to make this route work. Although we do not have an estimate of the activation free energy for this reaction, the trajectory implies a small barrier and quick formation of a very stable supercomplex. Finally, the last system in Table 4 shows the formation of an activated complex catalyst with a carbene ligand and the cocatalyst that quickly dissociates to the catalyst and a separated BuLi:cocatalyst complex. The relative energetics of this last case provides strong evidence that ligand modification (replacement of N-ligand) and a cocatalyst are necessary to F


Inorganic Chemistry

■

change the chemical profile that will allow the catalytic route to work.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.inorgchem.8b02910. Comparison of structures and energetics between G09 and CP2K, AIMD simulations with N-containing catalyst and cocatalyst and distance changes as a function of time during AIMD simulations, and optimized structure of BuLi, catalyst without N ligand, and cocatalyst in THF solvent shown in Figure 8A. (PDF)

■

REFERENCES

(1) McGrail, P. B. N., Satish, K.; Motkuri, R. K.; Glezakou, V.-A.; Koech, P. K.; Adint, T. T.; Fifield, L. S.; Fernandez, C. A.; Liu, J. System and process for production of magnesium metal and magnesium hydride from magnesium-containing salts and brines. US Patent 9,499,880, 2016. (2) Liu, J.; Bearden, M. D.; Fernandez, C. A.; Fifield, L. S.; Nune, S. K.; Motkuri, R. K.; Koech, P. K.; McGrail, B. P. Techno-Economic Analysis of Magnesium Extraction from Seawater via a Catalyzed Organo-Metathetical Process. JOM 2018, 70, 431. (3) Gilman, H.; Beel, J. A.; Brannen, C. G.; Bullock, M. W.; Dunn, G. E.; Miller, L. S. The Preparation of n-Butyllithium. J. Am. Chem. Soc. 1949, 71, 1499−1500. (4) Brown, H. C.; Lane, C. F. Organoboranes for synthesis. 10.1 the base-induced reaction of bromine with organoboranes. A convenient procedure for the conversion of alkenes into alkyl bromides via hydroboration. Tetrahedron 1988, 44, 2763−2772. (5) Howie, R. T.; Narygina, O.; Guillaume, C. L.; Evans, S.; Gregoryanz, E. High-pressure synthesis of lithium hydride. Phys. Rev. B: Condens. Matter Mater. Phys. 2012, 86, No. 064108. (6) Ashby, E. C.; Smith, T. Hydrometallation of alkenes and alkynes with magnesium hydride. J. Chem. Soc., Chem. Commun. 1978, 0, 30b−31. (7) Ashby, E. C.; Noding, S. A. Hydrometalation. 6. Evaluation of lithium hydride as a reducing agent and hydrometalation agent. J. Org. Chem. 1980, 45, 1041−1044. (8) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. A.; Nakatsuji, H.; Caricato, M.; Li, X.; Hratchian, H. P.; Izmaylov, A. F.; Bloino, J.; Zheng, G.; Sonnenberg, J. L.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Vreven, T.; Montgomery, J. A., Jr.; Peralta, J. E.; Ogliaro, F.; Bearpark, M. J.; Heyd, J.; Brothers, E. N.; Kudin, K. N.; Staroverov, V. N.; Kobayashi, R.; Normand, J.; Raghavachari, K.; Rendell, A. P.; Burant, J. C.; Iyengar, S. S.; Tomasi, J.; Cossi, M.; Rega, N.; Millam, N. J.; Klene, M.; Knox, J. E.; Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Martin, R. L.; Morokuma, K.; Zakrzewski, V. G.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Dapprich, S.; Daniels, A. D.; Farkas, Ö .; Foresman, J. B.; Ortiz, J. V.; Cioslowski, J.; Fox, D. J. Gaussian 09; Gaussian, Inc.: Wallingford, CT, 2009. (9) Adamo, C.; Barone, V. Toward reliable density functional methods without adjustable parameters: The PBE0 model. J. Chem. Phys. 1999, 110, 6158−6170. (10) Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized Gradient Approximation Made Simple. Phys. Rev. Lett. 1996, 77, 3865−3868. (11) Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized Gradient Approximation Made Simple [Phys. Rev. Lett. 77, 3865 (1996)]. Phys. Rev. Lett. 1997, 78, 1396−1396. (12) Ditchfield, R.; Hehre, W. J.; Pople, J. A. Self-Consistent Molecular-Orbital Methods. IX. An Extended Gaussian-Type Basis for Molecular-Orbital Studies of Organic Molecules. J. Chem. Phys. 1971, 54, 724−728. (13) 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. (14) Hariharan, P. C.; Pople, J. A. The influence of polarization functions on molecular orbital hydrogenation energies. Theor. Chim. Acta 1973, 28, 213−222. (15) Feller, D. The role of databases in support of computational chemistry calculations. J. Comput. Chem. 1996, 17, 1571−1586. (16) Scalmani, G.; Frisch, M. J. Continuous surface charge polarizable continuum models of solvation. I. General formalism. J. Chem. Phys. 2010, 132, 114110.

CONCLUSIONS In this article we present molecular level arguments for the choice of the ligand and cocatalyst effects in the regeneration of butyllithium from the decomposition products of a transmetalation reaction between MgCl2 and nBuLi. Calculations of the thermodynamics of the three reaction steps reveal that isolation of nBuLi in this catalytic route is a critical bottleneck due to the strong coupling of BuLi with the catalyst that leads to deactivation. To overcome the impediment resulting from the high affinity of Li toward N-containing catalyst ligands, we propose the use of N-free ligands for the catalyst. Estimated Gibbs free energies and equilibrium constants indicate that, while the modification of catalyst alone is not sufficient to enable the suggested catalytic cycle, N-bearing cocatalysts may contribute toward considerable improvements in the reaction free energetics.

■

Article

AUTHOR INFORMATION

Corresponding Author

*E-mail for V.-A.G.: [email protected]. ORCID

Mal-Soon Lee: 0000-0001-6851-177X Vassiliki-Alexandra Glezakou: 0000-0001-6028-7021 Roger Rousseau: 0000-0003-1947-0478 Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS M.-S.L., V.-A.G., and B.P.M. were in part supported by the US Department of Energy (DOE), Advanced Research Projects Agency for Energy, performed at the Pacific Northwest National Laboratory (PNNL). V.-A.G. and R.R. were in part supported by the DOE, Office of Science, Office of Basic Energy Sciences-CGBS Catalysis Program. V.-A.G. gratefully acknowledges support from the DOE, Office of Science, Office of Basic Energy Sciences-Separations Program. Computational resources were provided by the PNNL Research Computing Cluster and the National Energy Research Scientific Computing Center (NERSC), a U.S. Department of Energy Office of Science User Facility operated under Contract No. DE-AC0205CH11231. The authors acknowledge useful discussions with P. P. Koech, T. Adint, and J. Page regarding the choice of potential catalytic systems. G


Article

Inorganic Chemistry (17) 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. (18) Lippert, G.; Hutter, J.; Parrinello, M. A hybrid Gaussian and plane wave density functional scheme. Mol. Phys. 1997, 92, 477−487. (19) 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. (20) Hutter, J.; Iannuzzi, M.; Schiffmann, F.; VandeVondele, J. CP2K: atomistic simulations of condensed matter systems. Wiley Interdisciplinary Reviews: Computational Molecular Science 2014, 4, 15−25. (21) Grimme, S.; Hujo, W.; Kirchner, B. Performance of dispersioncorrected density functional theory for the interactions in ionic liquids. Phys. Chem. Chem. Phys. 2012, 14, 4875−4883. (22) Goedecker, S.; Teter, M.; Hutter, J. Separable dual-space Gaussian pseudopotentials. Phys. Rev. B: Condens. Matter Mater. Phys. 1996, 54, 1703−1710. (23) VandeVondele, J.; Hutter, J. Gaussian basis sets for accurate calculations on molecular systems in gas and condensed phases. J. Chem. Phys. 2007, 127, 114105. (24) Alexopoulos, K.; Lee, M.-S.; Liu, Y.; Zhi, Y.; Liu, Y.; Reyniers, M.-F.; Marin, G. B.; Glezakou, V.-A.; Rousseau, R.; Lercher, J. A. Anharmonicity and Confinement in Zeolites: Structure, Spectroscopy, and Adsorption Free Energy of Ethanol in H-ZSM-5. J. Phys. Chem. C 2016, 120, 7172−7182. (25) De Sousa, R. L.; Leite Alves, H. W. Ab Initio Calculation of the Dynamical Properties of PPP and PPV. Braz. J. Phys. 2006, 36, 501− 504. (26) Glezakou, V.-A.; Rousseau, R.; Dang, L. X.; McGrail, B. P. Structure, dynamics and vibrational spectrum of supercritical CO2/ H2O mixtures from ab initio molecular dynamics as a function of water cluster formation. Phys. Chem. Chem. Phys. 2010, 12, 8759− 8771. (27) Ndongmouo, U. F. T.; Lee, M.-S.; Rousseau, R.; Baletto, F.; Scandolo, S. Finite-temperature effects on the stability and infrared spectra of HCl(H2O)6 clusters. J. Phys. Chem. A 2007, 111, 12810− 12815. (28) McDermott, J. X.; Whitesides, G. M. Preparation and reactions of a titanium(IV) metallocycle. Formation of cyclopentanone from ethylene and carbon monoxide. J. Am. Chem. Soc. 1974, 96, 947−948.

H


Molecular Simulation of the Catalytic Regeneration of nBuLi through a

Recommend Documents