Active Palladium Colloids via Palladacycle Degradation as Efficient Catalysts for Oxidative Homocoupling and Cross-Coupling of Aryl Boronic Acids.

Active palladium colloids formed upon degradation of a palladacyclic complex (Herrmann-Beller 1) have been isolated for the first time and thoroughly characterized with techniques such as transmission electron microscopy (TEM), high-resolution TEM, X-ray photoelectron spectroscopy (XPS), and extended X-ray absorption fine structure spectroscopy. The synthesized palladium colloids have been utilized as efficient catalysts for the oxidative homocoupling of aryl boronic acids. Cross-coupling of two different aryl boronic acids has also been made possible using these active palladium colloids. This is the first report of this kind of coupling between aryl boronic acids.

Introduction

In the quest for developing highly efficient and robust Pd-based catalysts for C–C bond formation via cross-coupling reactions,[1] palladacycles have played a significant role and, in recent years, found a wide variety of applications, ranging from catalysts[2] for cross-coupling and related reactions to their more recent application as anticancer agents.[3] Early examples for the use of palladacyclic complexes in catalysis employed azobenzene and hydrazobenzene as coordinating ligands, finding applications in processes such as selective reduction of alkenes, alkynes, or nitroalkanes.[4]

Over the years, with the increasing demand for more robust, air-stable, and catalytically active catalysts for addressing synthetically challenging processes, palladacyclic complexes have come forth as a possible solution. Since the reports by Herrmann and Beller (the Herrmann–Beller complex is henceforth referred to as palladacycle 1) of a highly active palladacyclic complex formed by the reaction of Pd(OAc)2 and P(o-Tol)3, an exponential growth in examples related to their use in cross-coupling reactions has been observed.[5] Although the reactivity exhibited by the palladacycles has been exceptional, the current challenge is to apply them toward solving more important synthetic problems.

Although the synthetic utility of palladacycles has been well explored, for many years, the unique reactivity profile of these catalytic systems was attributed to the presence of either the Pd(II)/Pd(IV)[5a,6] or Pd(0)/Pd(II)[7] catalytic cycle. However, the noteworthy work by Jutand and co-workers[8] needs a special mention in this regard, providing useful insight into the possible mechanistic intermediates in these processes.

Another proposal that has gained lot of interest in recent years is the possibility of palladacycles acting as reservoirs for Pd(0).[9] However, one of the strongest arguments to date, based on experimental evidence, is about the presence of soluble palladium colloids/nanoparticles.[10] Loss of activity for an immobilized palladacyclic catalyst by Nowotny,[11a] along with the observation of induction period in the catalytic reaction followed by sigmoidal kinetic profile by Beletskaya and co-workers,[11b] has offered support for such assumptions. Several others have also provided evidence to highlight nanoparticular/colloidal involvement, notably, Gladysz and co-workers,[11c,11d] through their report on transmission electron microscopy (TEM) confirmation of Pd colloids. These initial reports on the presence of Pd nanoparticles/colloids brought about a flurry of activity, leading to the development of palladacyclic complexes, which could act as sources of highly active palladium colloids. Accordingly, a recent review by Alonso and Najera[12] elegantly highlights the advantages and applicability of oxime-derived palladacycles as sources of active palladium colloids/nanoparticles.

Besides these reports, pioneering work by Dupont on the kinetic and mechanistic aspects of the Heck reaction using the palladacycle {Pd[κ1-C, κ1-N-Cd(C6H5)C(Cl)CH2NMe2](μ-Cl)}2 has also revealed the presence of highly active Pd(0) colloids.[13] The presence of Pd colloids was confirmed by performing catalyst-poisoning studies, TEM analysis, and time-resolved ultraviolet–visible analysis.

Although these studies have provided evidence based on TEM, mass spectroscopy (MS), and sometimes extended X-ray absorption fine structure (EXAFS) analysis of the catalytic solutions,[14] no data to date has been presented on the isolation of colloids as well as their applications. To address this gap in the literature, we present herein the isolation of highly active palladium colloids/nanoparticles stabilized on carbon, formed through the degradation of palladacycle 1. Support for the formation of colloids has been provided using techniques such as TEM, high-resolution TEM (HRTEM), X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS), and EXAFS. The isolated palladium colloids have demonstrated excellent activity in the oxidative homocoupling and cross-coupling of aryl boronic acids.

Results and Discussion

Our interest in exploring the possibility of conversion of palladacycles to active colloids came about as a part of our studies on the application of Herrmann–Beller palladacycle 1 for diarylmethanol synthesis. Most of these processes are catalyzed by transition metals, namely, Fe, Cu, Ti, Rh, and Pd, with the latter showing enhanced reactivity.[15,16] Our study on diarylmethanol synthesis from palladacycle 1 was initiated by performing a screening study for the reaction of an aryl aldehyde with aryl boronic acid using different catalyst systems in PhMe (Scheme ).

Scheme 1

Screening Study for Diarylmethanol Synthesis

Although, the best results were obtained on changing the solvent from PhMe to CH3CN/H2O (for a complete screening study, see the Supporting Information). In all cases, a catalytic amount of CHCl3 was added, in line with recent reports on the beneficial effects of its addition through activation of the Pd-catalyst toward such types of reactions.[17] Chloroform when used as the solvent also gave comparable yields (86%) to those of CH3CN/H2O, and it could also be used as the reaction solvent.

To assess the applicability of the developed protocol, a scope study was undertaken using palladacycle 1 in the CH3CN/H2O solvent system at 80 °C (Scheme ). First, we investigated the effect of electronics on the aryl aldehyde on the catalytic reaction. Electron-accepting substituents favored arylation of the corresponding aryl aldehyde, smoothly furnishing the desired products in good yields (4a–d; Scheme ). Interestingly, electron-releasing substituents led to complete reduction of product formation, possibly due to the poor reactivity of the respective aldehydes. In contrast, employing a variety of differently substituted aryl boronic acids proved to be beneficial in obtaining good yields of diarylmethanols (4e–j; Scheme ).

Scheme 2

Substrate Scope for Palladium-Catalyzed Diarylmethanol Synthesis,

Aldehyde (1.0 mmol), aryl boronic acid (2.0 mmol), palladacycle 1 (1.0 mol %), CH3CN/H2O (1:1 v/v, 2.0 mL), K2CO3 (2.0 equiv.), CHCl3 (0.1 mL) at 80 °C for 24 h.

Isolated yields after column chromatography.

Colloidal/Nanoparticular Pathway Investigation

While following the catalytic reactions, it was observed that in almost all cases black particulate matter was formed, pointing toward the possible formation of palladium colloids/nanoparticles. To investigate this further, we performed a series of catalyst-poisoning experiments.

The first of this series was an experiment that is popular and was commonly employed by Whitesides in the early 80s. It encompasses poisoning the catalytic solutions with Hg, which results in amalgamation of the colloidal palladium species.[18] To our excitement, complete inhibition of the catalytic reaction was observed. This alone cannot be taken as absolute evidence for a nanoparticular pathway, and further analysis was undertaken to ascertain this assumption. Carbon disulfide (addition of excess of CS2) has also been employed as catalyst poison,[18b,18c] resulting in total inhibition of the catalytic reaction. The addition of tetrabutylammonium bromide[18c] as a nanoparticular stabilizer gave identical yields of the coupling product.

The Collman test[19] has been the method of choice for detecting the nanoparticular involvement in palladacyclic systems for the past several years. However, the nonavailability of polyethyleneglycol-bound benzaldehydes or aryl boronic acids prompted us to run the Crabtree test instead.[20] This test involves the specific use of a homogeneous species inhibitor, dibenzo[a,e]cyclooctatetraene (DCT), as the catalyst poison. The employment of such a sterically demanding poison would enable selective inhibition of any homogeneous species present in the catalytic solution via formation of a strong Pd–DCT complex, whereas in heterogeneous systems, coordination would be unfavorable.

The Crabtree test was carried out by addition of DCT (2.0 equiv. with respect to the Pd concentration) to the catalytic reaction 2 h prior to the addition of all substrates. The substrates were then added, and the reaction was continued for 24 h. On completion of the reaction, analysis of the catalytic products showed slight inhibition (39% product obtained), suggesting the possible presence of homogeneous Pd species or soluble Pd colloids (Scheme ).

Scheme 3

(a) Poison Tests for Confirming the Nanoparticular or Colloidal Catalyst and (b) Crabtree Test

All of the above results strongly suggest the presence of active palladium nanoparticles or colloids. However, the presence of homotopic Pd catalysts also cannot be excluded, and further analysis is needed to ascertain the exact nature of the catalytic species.

TEM Studies of Catalytic Solution

To ascertain the involvement of palladium colloids/nanoparticles, a detailed TEM study was undertaken. This was done by directly subjecting the general catalytic solution to TEM analysis after a reaction time of 24 h, represented in Scheme , involving the reaction between 4-chlorobenzaldehyde and phenylboronic acid, by simple deposition of the solution on carbon-coated copper grids. Palladium colloids approximately 2–4 nm in size were observed to be present in the catalytic solution (Figure a).

Figure 1

(a) TEM image of the catalytic solution, (b) TEM image of the poly(ethylene) glycol (PEG)-stabilized colloids. (c) HRTEM image of mercury-engulfed Pd colloids.

Further confirmation for the involvement of palladium colloids was obtained by trapping these colloids in the catalytic reaction with PEG[21] (added at the beginning of the reaction). Because of the stabilized nature of the colloids, they appear to be well-dispersed compared with the nonstabilized colloidal particles. The size of the colloidal particles was found to be in the approximate range of 3–5 nm (Figure b). Similar observations were also made during the HRTEM analysis of mercury-engulfed Pd colloids obtained after the mercury drop test. Energy-dispersive X-ray analysis clearly indicated the presence of mercury, with the size of the colloids increasing up to ∼4.0 nm (Figure c).

Isolation and TEM Characterization of Palladium Colloids

Next, we turned our attention toward the possible isolation of active palladium colloids from the catalytic solution. For this to be achieved, we first tried to find the substrate combination that led to the formation of active palladium colloids. Experiments were performed with palladacycle 1, used in stoichiometric amounts under different conditions, which have been summarized in Table . Only the combination of palladacycle 1 with 4-chlorobenzaldehyde in CHCl3 and added K2CO3 base at 80 °C resulted in the formation of a black powder. The solution was then centrifuged to obtain the black powder, which was washed several times with deionized water and after suspension in dry ethanol was collected and dried in vacuo (see isolation reaction in Table ).

Table 1

Experiments for Ascertaining Substrate Combination for Active Palladium Colloid Formation

1	ald	boronic acid	CHCl3	K2CO3	CH3CN/H2O	observation
√	√	×	√	√	×	black particles
√	√	×	×	√	√	no change
√	×	√	√	√	×	no change
√	×	×	√	√	×	no change

The powder thus obtained was characterized thoroughly by TEM, HRTEM, and energy-dispersive X-ray spectroscopy (EDS) (Figure ). TEM analysis of the powder (A) suggested the average size of colloids to be in the range of 2–4 nm, which is in good accordance with the values obtained from TEM analysis of the catalytic solution. To obtain a more precise value for the size of the colloidal particles, HRTEM analysis was performed on the powder sample, with image (B) of the palladium colloids indicating the formation of uniformly distributed spherical structures of ∼2 nm diameter.

Figure 2

(A) TEM image of Pd colloids, (B) HRTEM image of Pd colloids (∼2 nm diameter), and (C) EDS analysis of Pd colloids.

Confirmation of the presence of Pd colloids was further obtained by EDS. The spectra thus recorded allowed the identification of characteristic Pd lines at Lα1 = 2.838 keV and Lβ1 = 3.03 keV, which are commonly assigned to palladium metal, confirming the successful formation of the palladium colloids. A closer look at the EDS spectra also reveals the presence of elemental carbon (Kα = 0.277), which could possibly be obtained through the degradation pathway of palladacycle 1 or the aldehyde. Copper lines are observed as a part of the grid used for the analysis.

To further strengthen our claim for the presence of Pd colloids in the catalytic solution, powder XRD analysis was carried out on the isolated black powder. XRD analysis (Figure ) revealed the palladium colloids to be present in a single phase (111 of Bragg reflection), obtained at a 2θ value of 40. A slight hump in the region of 20–30 (2θ) also points toward the presence of carbon, although its exact nature cannot be ascertained at this moment and the employment of high-end characterization techniques, such as XPS or EXAFS, could help shed light on this aspect.

Figure 3

Powder XRD analysis of Pd colloids.

XPS and EXAFS Characterization of Pd Colloids

To verify the observation of carbon, to understand its origin, as well as to accurately determine the composition of the isolated Pd colloids, XPS was utilized. The XPS spectral analysis revealed only palladium, carbon, and oxygen, whereas no phosphorus species were observed, suggesting the absence of phosphine on the surface of Pd colloids (Figure ).

Figure 4

X-ray photoelectron spectral analysis of the synthesized nanoparticles.

The overall composition of the nanoparticular material was revealed to be 71.5% C, 22.4% O, and 3.4% Pd. On closer inspection of the palladium peaks (deconvolution of the peaks), it was observed that Pd is present as a mixture of PdO (66%), Pd(0) (22%), and PdX2 (halides possibly coming from the aldehyde).

Next, X-ray absorption near-edge structure analysis and comparison of Pd colloids with standards, using the Athena[22] program, suggest that the colloids are mostly composed of Pd(0) species along with some Pd(II) (Figures and 6). The comparison of Pd(OAc)2 and Pd2(DBA)3 with Pd foil and palladacycle 1 suggest that E0 for the colloids (24341 eV) lies within the region of Pd(0) (24350 eV) and Pd(II) (24353.3 eV). Thus, the sample of colloids can be considered as mixture of Pd (0) and Pd(II). The EXAFS region of Pd colloids in R-space is similar to that of Pd foil, except that the amplitude was reduced. This can be attributed to the termination effect.[23] The Fourier transforms (FTs) of the EXAFS data for the Pd colloids are presented in Figure . In the R-space, two main peaks were observed: one in the range of 1.2–2.0 Å and the other, more prominent one, in the range of 2.3–2.7 Å. The latter can be attributed to the Pd–Pd shell. For the former, it is difficult to distinguish between the Pd–O, Pd–C, and Pd–P shells in the region of 1.2–2.0 Å. Our mechanistic investigations from other methods prompted us to use the oxygen shell for fitting purposes. FT spectra and best fits are represented in Figure .

Figure 5

Comparison of Pd compounds and Pd nanoparticles plotted in R-space.

Figure 6

Comparison of Pd compounds and Pd nanoparticles plotted in energy.

Figure 7

FT spectra for palladium colloids.

The fitting of Pd colloids using Pd foil and Pd–O shell of Pd(OAc)2 as theory yielded a good fit, with a residual factor (R-value) of 0.006. During the fitting, the data and theory were k-weighted and Fourier-transformed using a Hanning window function in the k-range from 3 to 13 Å–1.

On the basis of the fitting values, it is fair to assume that the prior peak around 2.0 Å arises from the Pd–O shell. However, the possibility of a carbon atom cannot be ruled out at this step. From the fitting, it was possible to calculate the coordination numbers with respect to the central Pd atom. It was found that there were roughly six Pd atoms present around the central Pd at the distance of about 2.78 ± 0.005 Å. About 1.4 ± 0.5 atoms of oxygen (or carbon) were also present in the lattice at a distance of 1.97 ± 0.013 Å. FTs of Pd colloids show characteristic signatures of the Pd foil in the FCC structure, which was used as a reference and theory for calculation purposes.

The percent composition of Pd(0)/Pd(II) can be obtained by the calibration method using EXAFS analysis. However, discussion on such experiments is beyond the scope of this article. We are currently carrying out detailed studies on the subjected and will be reporting the results of the same shortly. Besides the two main peaks mentioned above, an additional peak below 1.2 Å was observed, which mostly comes from background. It was not possible to completely subtract this peak using AUTOBKG without affecting the amplitude of the peak for the Pd–Pd shell; hence, it was left untouched.

All of the above characterization details have enabled us to predict with greater certainty the formation of highly active Pd colloids through the possible degradation/conversion of palladacycle 1. Although this is the first time that the isolation and complete characterization of colloids has been undertaken, it is still not clear how or by which pathway the palladacycle degrades/converts to active colloids.

With the well-characterized active palladium colloids in hand, we performed a catalytic reaction to obtain diarylmethanol, to prove that the same species is responsible for the catalytic activity (Scheme ). To our excitement, the catalytic reaction with palladium colloids proceeded smoothly and furnished the coupled product in comparable yields (4b) in some cases, whereas the catalyst performed better than palladacycle 1 in certain other cases (4c and 4g). On the basis of the composition determined by XPS analysis of the carbon-stabilized Pd colloids, 10 mg (0.32 mol% of Pd) of the catalyst was able to efficiently catalyze the transformation at a relatively lower Pd concentration as compared to palladacycle 1.

Scheme 4

Diarylmethanol Synthesis Using Pd Colloids

Yields in brackets correspond to those obtained using 1.0 mol% palladacycle 1 under identical conditions

Oxidative Homocoupling of Aryl Boronic Acids

With the success obtained with the isolated Pd colloids in catalyzing the synthesis of diarylmethanol, we envisaged their use for tackling a more challenging synthetic process, oxidative homocoupling of aryl boronic acids. Homocoupling of aryl boronic acids has proven to be a very useful synthetic methodology for obtaining symmetrical biaryls, with several reports published over the last decade detailing the homocoupling of aryl boronic acids as the most convenient pathway.[24,25] Although, this transformation has been carried out using different metals, palladium-catalyzed oxidative homocoupling is at the forefront.[26,27] Recent advancements have also highlighted the possibility of performing the catalytic transformation under base- and ligand-free conditions.[28] However, most of these reports have been found to employ homogeneous catalysts that are difficult to recover, and the catalyst loadings have also been high.

In this regard, nanoparticular catalysis[29] has provided researchers with a sustainable alternative in terms of catalyst concentration and recoverability. However, only a few reports have emerged in the literature on their application in oxidative homocoupling of aryl boronic acids.[30] Among other catalysts, Au nanosized catalysts have also been employed in recent years. However, the relatively expensive nature of Au+ and Au3+ salts[31] make the protocol synthetically less attractive.[32] On closer inspection of the these reports, it is also evident that the protocols have been restricted to aryl boronic acids, whereas heteroaryl boronic acids provided poor yields. Similarly, electron-withdrawing substituents on the aryl boronic acids have proven to be sluggish substrates for homocoupling in general. With this background, we envisaged the possibility of employing isolated palladium colloids as catalysts to address these issues with the homocoupling process.

As a starting point of the study, 3-nitrophenyl boronic acid was chosen as the substrate and an extensive screening study was performed (Scheme ). Initially, the same conditions as that used for diarylmethanol synthesis were employed (Scheme ), without the addition of aldehyde as the coupling partner (40%). Encouraging results obtained under the conditions prompted us to explore the reaction further. Eventually, optimum product formation was observed on employment of air as the oxidant in the CH3CN/H2O solvent system at 80 °C. It is to be noted that the removal of CHCl3 (that was added in catalytic amounts for diarylmethanol synthesis) promoted the homocoupling reaction to proceed efficiently.

Scheme 5

Screening Studies for Oxidative Homocoupling of Aryl Boronic Acids

However, at this moment, the exact reason for such an observation is not known. With these results in hand, palladium-catalyzed oxidative homocoupling of a wide variety of aryl and heteroaryl boronic acids was undertaken. From the table given below (Scheme ), it can be observed that the palladium colloids are able to catalyze oxidative homocoupling of aryl boronic acids in very good yields. Electronic effects on the aryl boronic acids were found to have no effect on the catalytic activity of the colloids, with electron-withdrawing as well as electron-donating substituted boronic acids exhibiting comparable results. Heteroaryl boronic acids when employed as substrates also furnished good yields of the homocoupled products. Similar observations were made on using challenging, bulkier aryl boronic acids, with good-to-excellent yields of the homocoupled products obtained under the catalytic conditions.

Scheme 6

Palladium Colloids for Oxidative Homocoupling of Aryl Boronic Acids

Interestingly, the more reactive alkyl boronic acid, cyclohexyl boronic acid, which is prone to protodeborylation, also underwent the homocoupling reaction, furnishing dicylohexane in decent yields. Interestingly, 1,4-phenylenediboronic acid when subjected to homocoupling conditions furnished a polymeric material, which is insoluble in most organic solvents (characterization was not possible). Such a species has been described in the literature to be formed as a part of the homocoupling process, and it has been assumed that the material in hand is that reported.[33]

Another interesting result was obtained during the oxidative homocoupling of 2-bromo phenylboronic acid under oxidative coupling conditions (Scheme ). The homocoupling reaction when monitored by gas chromatography–mass spectrometry suggested the formation of three different products, with complete comparison of the starting boronic acid. Instead of exclusively furnishing the desired 2,2′-dibromo biphenyl, the coupling process provided two more products, namely, 2-bromobiphenyl and 1,1′ and 2′,1″-terphenyl in a ratio of 10:30:60 (the possible pathway for 6c has been represented in Scheme ). Although, this result reiterates the active nature of the palladium colloids, its efficiency in catalyzing a subsequent cross-coupling reaction via C–Br activation could have further synthetic relevance. The possibility of formation of products 6a and 6c could be understood from the proposed mechanism that has been put forth on the basis of some literature evidence for protodeboronation[34] and protodehalogenation[35] pathways that are known to take place during the catalytic processes.

Scheme 7

Oxidative Homocoupling of 2-Bromophenyl Boronic Acid

Cross-Coupling of Aryl Boronic Acids

The success obtained with oxidative homocoupling of aryl boronic acids (including electron-poor substituents) prompted us to further investigate the applicability of the isolated palladium colloids. Crossed homocoupling of substrates also provides researchers with an attractive alternative to the general cross-coupling processes. Success in such processes could be achieved by the judicious choice of electronically different substrates. However, to date crossed homocoupling has seldom been reported, given the difficulty in preferentially obtaining the cross-coupled product over the commonly observed homocoupling product. An example for oxidative cross-coupling of two different Grignard reagents was recently reported by Cahiez and co-workers under manganese-catalyzed conditions.[36] Similarly, gold(I)-catalyzed cross-coupling of two different alkynes has also been reported by Shi and co-workers for the synthesis of unsymmetrical 1,3-diynes.[37] Herein, we present the first report on cross-coupling of aryl boronic acids using synthesized palladium colloids under aerobic conditions.

At the outset of our studies, we decided to employ one of the coupling partners as an electron-poor boronic acid (3-nitro phenylboronic acid) that could serve as the limiting reagent. A more reactive electron-rich boronic acid (4-methoxyphenyl boronic acid) was then added in excess (1:1.3) under aerobic conditions using oxygen as the oxidant. To our surprise, the cross-coupled product (4-methoxy-3-nitrobiphenyl) was obtained in good yields (68%), providing the possibility of obtaining differently substituted biaryls via palladium colloid-catalyzed oxidative coupling of air-stable and cheaply available aryl boronic acids (Scheme ). Therefore, with the proper selection of coupling partners it could be possible to achieve good selectivity toward promoting the cross-coupling process over the commonly observed homocoupling of aryl boronic acids.

Scheme 8

Cross-Coupling of Different Aryl Boronic Acids for Accessing Unsymmetrical Biaryls

This is an attractive synthetic alternative for obtaining unsymmetrical biaryls compared with using hazardous Grignard reagents. Although at this moment only representative examples have been highlighted, the full potential of the synthetic protocol along with mechanistic studies will be reported elsewhere.

Conclusions

In conclusion, herein, we report evidence for the degradation/conversion of palladacycle 1 to active palladium colloids, which were isolated and thoroughly characterized by techniques such as TEM, HRTEM, XPS, and EXAFS. The distribution of these palladium colloids, with an average size of ∼2 nm, on carbon was established by the above-mentioned techniques. The isolated palladium colloids were shown to be active catalysts in a variety of coupling processes, including the synthesis of diarylmethanol and oxidative homocoupling of aryl boronic acids. One of the first reports on the cross-coupling of aryl boronic acids was also made possible by the unique reactivity exhibited by the palladium colloids.