A Comparative Kinetic and Computational Investigation of the Carbon-Sulfur Cross Coupling of Potassium Thioacetate and 2-Bromo Thiophene Using Palladium/Bisphosphine Complexes
Graphical Abstract
The palladium-catalyzed carbon-sulfur cross-coupling reaction involving a 2-bromothiophene derivative and potassium thioacetate as a hydrogen sulfide surrogate was investigated. Two 1,1′-bisphosphino-ferrocene ligands were compared in kinetic experiments and by computational methods to rationalize the different reactivity profiles.
Abstract
We conducted an investigation into the palladium-catalyzed carbon-sulfur cross-coupling reaction involving a 2-bromothiophene derivative and potassium thioacetate as a substitute for hydrogen sulfide. This investigation utilized kinetic and computational methods. We synthesized two palladium complexes supported by the bisphosphane ligands bis(diphenylphosphino)ferrocene (DPPF) and bis(diisopropylphosphino)ferrocene (DiPPF), as well as their tentative intermediates in the catalytic cycle. Reaction rates were measured and then compared to computational predictions.
Introduction
Since its discovery in the late 1970s, the development of palladium-catalyzed carbon-sulfur bond cross coupling1 has seen remarkable progress.2 It has garnered considerable attention from both academia and industry due to its versatility and broad applicability in the synthesis of pharmaceuticals,3 agrochemicals,4 materials,5 and natural products.6 Additionally, the mild reaction conditions and high functional group tolerance offered by these catalytic systems have contributed to their widespread adoption in contemporary synthetic routes.7 This catalytic process involves the reaction of an organic substrate bearing a carbon-halogen bond with a sulfur-containing nucleophile, typically an aryl or alkyl thiol. Under the influence of palladium catalysts, the coupling partners undergo a controlled transformation, forming a new C−S bond in a predictable and efficient manner. In recent years, the application of hydrogen sulfide (H2S) surrogates emerged as a powerful tool for the construction of unsymmetrical thioethers and heterocycles. For this purpose, several surrogates such as silyl protected thiol,8 thiourea9 and thioacetates10 were introduced which allowed for sequential deprotection strategies. Particularly, the cross coupling of potassium thioacetate (KSAc, 1) proved in our hands as a valuable and easy-to-handle H2S surrogate for the concise synthesis of sulfur-rich heteroacenes and polymers.5d, 11 Although some mechanistic and kinetic investigations are reported for the coupling of aryl and alkyl thiolates,12 the coupling of 1 has not been investigated in detail yet. In general, the palladium catalyzed C−S coupling proceeds by the oxidative addition of a usually in situ formed active Pd complex, followed by transmetalation and reductive elimination (Scheme 1).
The general mechanism of the palladium catalyzed C−S cross coupling using thioacetate as H2S surrogate.
Some unproductive resting states for bisphosphine derived Pd complexes12, 13 and for bis-thiolate Pd complexes14 were reported, which renders the C−S cross coupling a particularly challenging endeavor.15
Only recently, we developed a polycondensation of bifunctional benzene and thiophene derivatives, providing access to poly(phenylene) and poly(thiophenylene) sulfides.11c However, we found that the efficiency of the polycondensation was strongly affected by the supporting bisphosphine ligands. Therefore we decided to initiate a comparative study of the catalyzed C−S cross coupling of potassium thioacetate (1) and 2-bromo-3-noctylthiophene (2 a) using 1,1′-diphenylphosphinoferrocene (DPPF, 3 a) and 1,1′-di-isopropylphosphinoferrocene (DiPPF, 3 b) derived Pd(0) complexes.
Results and Discussion
We initiated our study by comparing the reaction progress of the C−S cross coupling of potassium thioacetate (1) and 2 a, using 10 mol % [Pd(dba)2]/3 a or 3 b at 120 °C (Figure 1).
C−S cross coupling of 2-bromo-3-noctylthiophene (2 a) with KSAc (1) and 10 mol % [Pd(dba)2]/3 a or 3 b (ratio 1 : 1).
As indicated by the data presented in Figure 1, it is apparent that the overall reaction rate of the [Pd(dba)2]/DiPPF (3 b) catalyst is inferior to that of the analogous DPPF (3 a) catalyst. Given the complex nature of the coupling reaction, which involves multiple elemental reactions, we opted to undertake kinetic investigations for each individual step.
First we investigated the oxidative addition of DPPF (3 a) and DiPPF (3 b) derived palladium(0) complexes to 2-bromo-3-noctyl-thiophene (2 a). Since most C−S cross coupling reactions are performed in toluene, we decided to utilize d6-benzene as a substitute in the kinetic experiments by NMR. The kinetic experiment for the oxidative addition was performed under pseudo-first order conditions using [Pd(dba)2] (dba=dibenzylidene acetone) as Pd(0) source in the presence of 1 equiv. of the respective bisphosphine and 10 equiv. of the bromothiophene 2 a yielding the Pd(II) complexes 4 a and 4 b (Scheme 2 and Figure 2).
Kinetic experiment of the oxidative addition of Pd(0)/bisphosphine complexes to 2-bromo-3-noctyl-thiophene (2 a).
a) 31P {1H} NMR spectra of the oxidative addition of the Pd/bisphosphine complexes to 2-bromo-3-noctyl-thiophene (2 a); b) pseudo-first order plot and pseudo-first order rate constants at 60 °C: kox-addpseudo=0.110 h−1 (Pd/3 a); 0.029 h−1 (Pd/3 b) (average of three independent repetitions).
The reaction was kept at 60 °C investigated by 31P {1H} NMR spectroscopy every 60 min. The formation of the Pd/bisphosphine complexes is instantaneous, as evidenced by the initial 31P NMR spectrum of the mixture (Figure 2a). The 31P NMR resonance of the DiPPF complex 3 b/Pd shifted downfield relative to that of 3 a/Pd due to the reduced π-backbonding in alkyl phosphines.16
The pseudo-first order plot reveals that the oxidative addition of the Pd/3 a complex to 2 a is 3.8 times faster than the addition of the more electron rich Pd/3 b (Figure 2b). This finding is somewhat unexpected, since usually more electron-rich Pd/bisphosphine complexes should oxidatively add faster than less electron-rich complexes as Pd/3 a.17 The starting material for the kinetic investigation of the transmetalation was obtained by the reaction of CpPd-cinnamyl complex (5), DPPF (3 a) or DiPPF (3 b) and 2-bromo-3-noctyl-thiophene (2 a) in toluene at 35 °C (not shown, see Supporting Information for details). The Pd(II) complexes 4 a and 4 b were obtained in 88 % and 89 % respectively. With this starting material in hand, the transmetalation was investigated under pseudo-first order kinetics by the reaction of 1 equiv. of 4 a and 4 b with 10 equiv. of potassium thioacetate (KSAc, 1, Scheme 3 and Figure 3).
Kinetic experiment of the transmetalation of 4 a and 4 b with KSAc (1).
a) 31P {1H} NMR spectra of the transmetalation of 4 a and 4 b with KSAc (1); b) pseudo-first order plot and pseudo-first order rate constants at 50 °C and 60 °C respectively: ktranspseudo=0.070 h−1 (4 a); 0.019 h−1 (4 b) (average of three independent repetitions).
For the DPPF derivative 4 a the transmetalation was too fast to monitor, so that the reaction was conducted at 50 °C, whereas the reaction of 4 b was performed at 60 °C, because it was too slow at 50 °C. Although, the pseudo-first order rate constants cannot directly be compared, it is evident that the DPPF derivative undergoes significantly faster transmetalation compared to the DiPPF derivative 4 b. This may be attributed to the increased steric bulk of the DiPPF ligand compared to the DPPF (see below). Furthermore, in non-polar solvents transmetalation often occurs via an associative mechanism by formation of an ate-complex, followed by elimination of the halide. This association step is particularly less favored for electron-rich Pd complexes, resulting in a reduced reaction rate.18 This is further supported by the comparable short reaction time (3 h versus 12 h) at room temperature when the reaction was carried out in a polar solvent like acetonitrile for the synthesis of the reductive elimination precursors 6 a and 6 b.
The corresponding thioacetate complexes 6 a and 6 b were synthesized by the reaction of 4 a and 4 b with 10 equiv. KSAc (1) in MeCN at room temperature. Both products were obtained after 3 h as yellow solids in 89 % yield. The molecular structure of 6 a was determined by X-ray crystallography19 and exhibits nearly perfect square planar coordination of the Pd center (sum of angles 360.3°) typical for a d8 electron configuration (Figure 4)
Molecular structure of 6 a; hydrogen atoms were omitted for clarity.18
To quantify the proximity of the coordination to a perfect square planar coordination, the geometry indexes τ4=0.14520 and τ′4=0.14721 were calculated, ranging from 0 for square planar to 1 for tetrahedral coordination. In contrast to this, deviations from this coordination can be found in the bond distances of the two phosphorous and the sulfur atom to the palladium, which range between 2.3184(3) Å and 2.3672(3) Å in comparison to the significant shorter bond length to the carbon atom with 2.0358(11) Å. Furthermore, the biting angle of the coordinating bisphosphane ligand is with 101.717(11)° significantly greater than in a perfect square coordination. The monodentate κS-thioacetate coordination is known in DPPF−Pd(SAc)2 complexes22 and is also observed in 6 a. However, comparable solid-state structures bearing e. g. an aryl substituent are so far not reported. Notably, the acetyl group is located cis to the thiophene ring system, avoiding interference with the noctyl substituent.
The reductive elimination was investigated by the reaction of 6 a and 6 b in the presence of 1 equiv. PPh3 in d6-benzene at 70 °C, using as internal standard tBu2PCl in a sealed capillary (Scheme 4 and Figure 5).
Kinetic experiment for the reductive elimination of 6 a and 6 b in the presence of 1 equiv. PPh3.
a) 31P {1H} NMR spectra of the reductive elimination of 6 a and 6 b; b) first order plot and first order rate constants at 70 °C: kred-el=0.078 h−1 (6 a); 0.027 h−1 (6 b) (average of three independent repetitions).
Again, the rate constant for the DPPF complex is approx. 3-fold higher than for the corresponding DiPPF complex (Figure 4). This reactivity is often observed23 since the less electron-rich ligand is better suited to stabilize the Pd(0) center. However, significant steric strain can also often accelerate the reductive examination, but this is apparently not the case. The reductive elimination of 6 a shows the smallest reaction rate constants and must be considered as the rate-determining step. In contrast, the rate constants of the DiPPF/Pd(0) catalyzed reaction are of the same magnitude. Overall, the reaction rates for the individual reaction steps are considerably higher for the DPPF complex than for the isostructural DiPPF derivative.
The DPPF complex outperforms the DiPPF complex under these typical reaction conditions. The Pd/DPPF catalyzed reaction is essentially complete within 2.5 h, whereas the Pd/DiPPF catalyzed reaction requires 6 h to achieve comparable conversion.
To obtain more insight into the Pd-catalyzed C−S cross coupling, we investigated the kinetic experiments by computational chemistry using density functional theory (DFT). For simplicity reasons, we utilized 2-bromo-3-methylthiophene (2 b), potassium thioacetate and potassium bromide implicitly solvated by three benzene molecules (KSAc ⋅ 3 C6H6, KBr ⋅ 3 C6H6) for the computations, in order to avoid the critical handling of non-solvated spherical ions. All calculations were performed using the ORCA 5.0.3 quantum chemistry package.24 The ground structures and transition states were optimized using the PBE025 functional with Ahlrich's def2-SVP basis,26 auxiliary basis set and D4 dispersion correction.27 Minima and saddle points were characterized by frequency calculations, exhibiting the absence of imaginary frequencies for ground state structures and one imaginary frequency indicating the reaction coordinate of a transition state. Thermostatistical corrections at 60 °C on this level of theory were utilized with single point calculations using the PBE0 functional with the def2-QZVPP basis set, D4 dispersion correction and SMD28 solvent treatment (benzene) for final calculation of free energies. The results of the computational studies are summarized in Figure 6.
Calculated reaction path with PBE0-D4/def2-QZVPP/CPCM(benzene)//PBE0-D4/def2-SVP for the C−S cross coupling of 2-bromo-3-methyl-thiophene (2 a) with KSAc (1) using DPPF and DiPPF derived palladium catalysts [Pd/3 a] and [Pd/3 b]. Free reaction energies in kcal ⋅ mol−1.
In general, the C−S cross coupling is by −18.5 kcal ⋅ mol−1 exergonic without considering the formation of potassium bromide. The barrier for oxidative addition to 2 b is for the DPPF-derived complex [Pd/3 a] by 3.1 kcal ⋅ mol−1 (TS1a) lower than for the more electron-rich DiPPF-derived complex [Pd/3 b] (TS1b). This finding is in line with the kinetic data of a higher rate for the oxidative addition of [Pd/3 a] to 2 a (compare Figure 1). The reason for this marked difference in reaction rate may be found in the differences in steric restrictions of the bisphosphines. The steric constraints were analyzed by the SambVca 2.1 method29 in order to visualize and quantify the steric differences by the buried volume (%Vbur, Figure 7). The %Vbur of the DiPPF ligand (3 b) is by 3.5 % larger than 3 a, indicating a considerably higher steric demand compared to the DPPF ligand. This increased steric demand reflects in the displacement of the Pd center by 0.46 Å in the TS1 in order to achieve oxidative addition (compare Figure 6b). The formation of the oxidative addition products 4 a and 4 b are both exergonic by −28.3 kcal ⋅ mol−1 and −33.5 kcal ⋅ mol−1 respectively.
a) Left: Calculated structures of the Pd(0)/DPPF ([Pd/3 a]) and of the Pd(0)/DiPPF ([Pd/3 b]) complexes (Pd atom omitted for clarity); right: steric maps and buried volume of the Pd(0)/DPPF ([Pd/3 a]) and of the Pd(0)/DiPPF ([Pd/3 b]) complexes generated by SambVca 2.1;29 b) structure overlay of the ground (beige) and transition state (cyan) of the oxidative addition (TS1); the 2 b fragment was omitted for clarity.
The rate of transmetalation is strongly dependent on the solvent polarity. This is demonstrated by the synthesis of the thioacetates 6 a and 6 b from the bromides 4 a and 4 b and KSAc (1) in acetonitrile within 3 h (compare Scheme 2), and presumably proceeds through autoionization of the bromides 4 a and 4 b and addition of the thioacetate ion. However, the cationic Pd-species are less stabilized in non-polar solvents like benzene in our kinetic experiments. The formation of such species is by 16.5 kcal ⋅ mol−1 and 20.5 kcal ⋅ mol−1 endergonic with respect to 4 a and 4 b and, thus, very unlikely so that the addition/elimination mechanism through an “ate” intermediate appears reasonable. The estimation of the transition state of the transmetalation was found to be very challenging due to the formation of KBr. This interfered with our approach of the explicit solvent approximation. Consequently, we were forced to include explicit solvent molecules in order to account for the solvation of the potassium cation and bromide as [K ⋅ 3C6H6]+ and [KBr ⋅ 3C6H6]. This approach provided significantly more reasonable energies, but rendered the transition state search as highly challenging. However, the transition state for the transmetalation was approximated by the climbing image, including a second iteration with tighter convergence (ZOOM-NEB-CI keyword). Solvation of the potassium anion was accounted by the inclusion of one benzene molecule in the transition state.30 However, the subsequent transition state searches did not converge due to the shallow potential energy surface caused by the number of conformers in the coordination of potassium by benzene. The complexes 4 a and 4 b have two sterically different half spaces, one with the thiophene‘s methyl group and one with the sulfur atom pointing inwards. Both faces appear feasible for the addition of the KSAc molecule in the transmetalation step. The addition of KSAc from the sterically less occupied half space afforded only 4.6 kcal ⋅ mol−1 for trans-TS2a and 9.0 kcal ⋅ mol−1 for trans-TS2b, whereas the addition from the more sterically hindered side required 13.3 kcal ⋅ mol−1 and 12.7 kcal ⋅ mol−1 for cis-TS2a and cis-TS2b respectively. Taking into account that we approximated the noctyl side chain by a methyl group in the computations, these energy differences may be even more pronounced. The cis and trans orientation of acetyl group with respect to the methyl-substituent in the thioacetates 6 a and 6 b are almost equal in energy (cis-6 a: −30.6 kcal ⋅ mol−1); trans-6 a: −30.0 kcal ⋅ mol−1 and cis-6 b: −31.6 kcal ⋅ mol−1; trans-6 b: −31.5 kcal ⋅ mol−1). The barriers for the reductive elimination from the trans conformation are by 4.2 kcal ⋅ mol−1 and 4.5 kcal ⋅ mol−1 lower than for the elimination from the corresponding cis isomers. Since the formation of the trans conformers is kinetically more favorable and the reaction path through the barrier of 23.4 kcal ⋅ mol−1 for trans-TS3a and 26.7 kcal ⋅ mol−1 for trans-TS3b is most likely the preferred reaction path. For both complexes, this constitutes the rate determining step of the C−S cross coupling sequence. Whereas the computation reproduces this finding for the diphenylphosphino derived catalyst [Pd/3 a], the kinetic data shows similar reaction rates for the diisopropyphosphino derivative [Pd/3 b]. To elucidate variations in reaction barriers, we examined structural features such as the P−Pd−P bite angle (see Table 1) and energies by distortion/interaction analysis31 (see Figure 8).
|
P−Pd−P angle (bite angle) [°] |
|
|---|---|---|
compound |
DPPF (3 a) |
DiPPF (3 b) |
L−PdCl232 |
99.1 |
103.6 |
L−Pd |
124.9 |
131.4 |
TS1 |
107.8 |
109.4 |
4 |
99.4 |
104.7 |
trans-TS2 |
101.4 |
100.7 |
cis-TS2 |
96.8 |
99.5 |
trans-6 |
99.0 |
106.1 |
cis-6 |
100.3 |
106.6 |
trans-TS3 |
103.2 |
108.5 |
cis-TS3 |
105.7 |
112.4 |
Calculated energies in kcal ⋅ mol−1 of the distortion/interaction analysis for a) 3 a/Pd and b) 3 b/Pd derived catalysts (PBE0/def2-QZVPP, solvent and dispersion corrections were not considered; E(dist): distortion energy of Pd and organic fragment; E(int.): interaction energy; E(act.): activation energy).
The crystal structures of 3 a/PdCl232a and 3 b/PdCl232b reveal bite angles of 99.1° and 103.6°, respectively. In the Pd(0) species, these angles increase to 124.9° and 131.4°, inducing a more constrained geometry. This alleviation of strain appears to account for the exergonic nature of the oxidative addition (refer to Figure 6). In both transition states (TS1), the bite angle decreases to 107.8° and 109.4°, compelling the Pd complex into a more reactive conformation with the Pd atom displaced by 0.34 Å and 0.46 Å for 3 a/Pd and 3 b/Pd, respectively (compare Figure 7b). This geometric alteration is more pronounced for 3 d and is evident in the distortion energy (ΔE(dist. Pd-frag.)) associated with this step. The interplay of the bite angle and ΔE(dist.) in palladium complexes was discussed earlier.31c, 31e, 31g In Figure 8, a comparison of the energy decomposition schemes for TS1 reveals a slight increase in distortion energy for 3 b/Pd. However, the predominant contribution to the barrier arises from the elongation of the C−Br bond in TS1b, suggesting a late transition state compared to the 3 a/Pd system. Moreover, TS1a experiences stabilization through two non-covalent interactions of 2 a with the 3 a/Pd complex (Figure 9 left).
Non-covalent interactions plots (NCI plots) of the transition states.33
In TS1a, the thiophene ring forms attractive non-covalent interactions with the phenyl rings of the DPPF ligand, whereas these interactions are absent for the DiPPF ligand. Additionally, the sulfur atom in TS1a engages in binding interactions with the Pd center, providing further stabilization. In TS1b, the sulfur atom lacks interaction with the Pd center but exhibits weak repulsive interactions with the DiPPF ligand. The products of the oxidative addition, 4 a and 4 b, demonstrate nearly perfect alignment of the bite angle with literature-precedented dichloro complexes. The most significant distortion energies for the L−Pd-fragment occur in the transition state of the transmetalation step, as expected for the associative mechanism, necessitating substantial structural reorganization from square-planar to strongly distorted trigonal bipyramidal. The bite angles in the resulting thioacetate complexes, 6 a and 6 b, fall within the range of square planar complexes. However, the angle is slightly enlarged for the DiPPF derivatives of 6 a by 3°. Reductive elimination from the DPPF complexes of 6 a require higher degrees of distortion in the product fragment compared to the DiPPF-derived structures of 6 b (compare Figure 8a and b, trans/cis-TS3). Nevertheless, the transition states of DPPF-derived complexes exhibit stabilizing π–π and π*–π (C=O to arene) interactions, while the DiPPF-derived structures feature destabilizing S−O and π–O interactions (compare Figure 9).
Conclusions
In summary, we investigated the palladium catalyzed cross coupling of a thiophene electrophile with potassium thioacetate as sulfur nucleophile by kinetic and computational methods. The reaction rates proved to be highly dependent on the supporting bisphosphine ligand. The less electron-releasing bis(diphenylphosphino)ferrocene (DPPF) ligand displayed highest rates in the oxidative addition to 2-bromo-3-noctyl thiophene and is by a factor of 3.8 faster than the more electron-rich bis(diisopropylphoshpino)ferrocene ligand. This may be attributed to the significant necessary displacement of the palladium atom in the oxidative addition (compare Figure 6b). The rate determining step is for both catalyst systems the reductive elimination. Steric influences of the ligand have significant impact on the rate of reductive elimination,22 rendering the more bulky one as kinetically less efficient. Furthermore, transmetalation has a significant impact on the overall reaction rates. For both investigated complexes, the transmetalation has a comparable rate as the reductive elimination. Computational investigations have unveiled that the reduced reaction barriers observed in the DPPF ligand-supported catalyst stem from the presence of stabilizing non-covalent contacts, including π–π and π*–π interactions between the ligand and the substrates.
Supporting Information
Details of synthetic procedures, NMR spectra, computational methods and crystallographic results can be found in the Supporting Information. The authors have cited additional references within the Supporting Information.34, 35
Acknowledgments
The German Science Foundation (DFG) is gratefully acknowledged for financial support (PA 1562/14-1). The authors would like to thank the Paderborn Center for Parallel Computing (PC2) for providing computing time support for this project. Open Access funding enabled and organized by Projekt DEAL.
Conflict of interests
The authors declare no conflict of interest.
Open Research
Data Availability Statement
The data that support the findings of this study are available in the supplementary material of this article.
