Report

Sustainable Fe–ppm Pd nanoparticle catalysis of Suzuki-Miyaura cross-couplings in water

See allHide authors and affiliations

Science  04 Sep 2015:
Vol. 349, Issue 6252, pp. 1087-1091
DOI: 10.1126/science.aac6936

Iron lends power to traces of palladium

Palladium (Pd) is a mainstay of chemical catalysis. The precious metal has a knack for forging carbon-carbon (C-C) bonds. Handa et al. now report that when mixed in a specific preparation with iron, just parts per million of Pd suffice to catalyze the C-C bond–forming Suzuki coupling reaction. The addition of surfactants allowed the reaction to proceed in water. The protocol bodes well for conserving Pd in pharmaceutical and agrochemical synthesis.

Science, this issue p. 1087

Abstract

Most of today’s use of transition metal–catalyzed cross-coupling chemistry relies on expensive quantities of palladium (Pd). Here we report that nanoparticles formed from inexpensive FeCl3 that naturally contains parts-per-million (ppm) levels of Pd can catalyze Suzuki-Miyaura reactions, including cases that involve highly challenging reaction partners. Nanomicelles are employed to both solubilize and deliver the reaction partners to the Fe–ppm Pd catalyst, resulting in carbon-carbon bond formation. The newly formed catalyst can be isolated and stored at ambient temperatures. Aqueous reaction mixtures containing both the surfactant and the catalyst can be recycled.

Precious metal catalysis has been and continues to be a predominant means of C-C, C-H, and C-heteroatom bond construction in organic synthesis. In particular, palladium-catalyzed Suzuki-Miyaura, Heck, and Negishi couplings are indispensable, as recognized by the 2010 Nobel Prize (1, 2). However, economically accessible supplies of Pd and other precious metals are dwindling, thus raising concerns about the sustainability of this chemistry (3).

To circumvent this situation, alternative metals such as nickel (4, 5) and copper (6, 7) have been studied, especially as applied to the heavily used, Pd-catalyzed Suzuki-Miyaura reactions (8, 9). Despite varying degrees of success, Pd remains, by far, the most effective metal for such reactions. In trace levels, perhaps as impurities in salts of less expensive metals, Pd could ultimately prove to be both natural and sustainable for use in catalysis. Here we disclose such a discovery: a technique that takes a readily available commercial salt derived from Earth-abundant iron—which naturally contains parts-per-million (ppm) levels of Pd—and processes it, in a single step, into highly active nanoparticles capable of mediating Suzuki-Miyaura cross-couplings in recyclable water as the reaction medium.

At the heart of this advance lies the confluence of several reaction variables: the origin and source of the iron salt, the presence of ppm levels of Pd, the manner through which these are converted to nanoparticles, and the use of aqueous micellar conditions for catalysis. The catalyst preparation calls for the use of inexpensive (97% purity) FeCl3 containing ppm levels of Pd, admixed with a ligand and dissolved specifically in tetrahydrofuran (THF). Treatment of this solution at room temperature with two equivalents of a Grignard reagent, also in THF, quickly affords nanoparticles that, after solvent removal in vacuo, can be used directly in Suzuki-Miyaura reactions. The in situ generation of 5 mole percent (mol %) of these Fe–ppm Pd nanoparticles was found empirically to be sufficient. Next, an aqueous solution containing 2 weight percent (wt %) of our commercially available designer surfactant TPGS-750-M (10) and a base (K3PO4⋅H2O, 1.5 to 2.0 equivalents) is added to the nanoparticles. Reaction partners 1 and 2 as model substrates are then introduced, leading to biaryl product 3 (Fig. 1). The choice of ligand is crucial (Fig. 1), with SPhos (2-dicyclohexylphosphino-2′,6′-dimethoxybiphenyl) and XPhos (2-dicyclohexylphosphino-2′,4′,6′-triisopropylbiphenyl) affording the best results. Vigorous stirring at temperatures between ambient and 45°C, depending upon the extent of crystallinity of the reaction partners, is sufficient to drive couplings to completion, typically in the 12- to 24-hour time frame (at a global concentration of 0.5 M).

Fig. 1 Ligand optimization for Fe–ppm Pd catalysis of Suzuki-Miyaura cross-couplings.

Conditions were as follows: 4-bromoanisole (0.5 mmol), naphthalene-1-boronic acid (0.75 mmol), 5 mol % Fe–ppm Pd nanoparticles, K3PO4•H2O (0.75 mmol), 2 wt % TPGS-750-M (1 ml). Asterisks indicate yields based on gas chromatography–mass spectrometry (GC-MS).

We examined the scope of this technology and found that many representative cases afford good-to-excellent isolated yields (Fig. 2). A broad variety of aromatic and heteroaromatic arrays, with either partner being the aryl halide or boron derivative, can be tolerated. Functional groups including CF3, amines, acetals, amides, aldehydes, esters, ketones, phosphate esters, nitro groups, polyaromatics, sulfonamides, and carbamates are represented among these examples. Several types of heteroaromatic arrays are also amenable, including nitrogen-containing moieties that might present complications as competing ligands for Pd. Both bromides and iodides are excellent educts, and the boron species involved can be any of those commonly employed: boronic acids, Bpin (boronic acid pinacol ester) (11) or N-methyliminodiacetic acid (MIDA) boronates (12, 13), and BF3K salts (14, 15).

Fig. 2 Couplings between aryl halides (Ar-X) and aryl (Ar') or alkenyl boron derivatives.

Unless otherwise noted, conditions for these couplings were as follows: Ar-X (0.5 mmol), Ar'-BRn (0.6 mmol), FeCl3 (5 mol %), SPhos (5 mol %), MeMgCl (10 mol %), K3PO4•H2O (0.75 mmol), TPGS-750-M (2 wt %, 1 ml), 45°C. Room temperature, rt. Asterisks indicate the use of Ar-B(OH)2 or Ar-B(MIDA) (1.2 mmol) and K3PO4•H2O (1.5 mmol). Reported yields are for isolated, chromatographically purified materials. 320 ppm Pd is required (the general procedure is described in detail in the supplementary materials).

The composition of the iron salt plays a role in the activity of the resulting nanoparticles, as does the manner in which the salt is reduced. Attempts to use Fe(acac)3 (acac, acetylacetonate) and iron pyrophosphate [Fe4(P2O7)3], as well as several other salts (table S2), led to a far less reactive catalyst (also formed in situ) than that derived from FeCl3. Analysis of FeCl3 from a commercially available source by inductively coupled plasma (ICP) atomic absorption spectrometry showed that ~300 to 350 ppm Pd was present. Sources that contained less precious metal (16, 17) led to incomplete reactions under otherwise identical conditions. Doping alternative sources of FeCl3 (≥97% purity) with 350 ppm Pd(OAc)2 (OAc, acetate) yielded nanoparticles of identical activity, as assessed by using the model reaction that produced 3 (table S14) (1720). However, attempts to use these ppm levels of Pd in the absence of preformed iron-based nanoparticles led to virtually no reaction, suggesting that the release of Pd into the aqueous medium is not responsible for the catalysis observed. Although the use of Grignard reagents MeMgX (X = Cl or Br) and i-PrMgCl yielded material of comparable activity, both PhMgCl and n-hexyl-MgBr, among other reductants (e.g., NaBH4, polymethylhydrosiloxane), led to nanoparticles of inferior quality (table S4).

Successful couplings require the presence of both Fe and Pd within these nanocomposites, as determined by several control reactions (Fig. 3). Reactions that were attempted using 400 ppm Pd(OAc)2, with SPhos as a ligand in various ratios, did not lead to product formation. The use of 5 mol % pure FeCl3 with 400 ppm Pd(OAc)2 and 500 ppm SPhos, without prior treatment with MeMgCl, afforded none of the biaryl product. However, upon reduction of 5 mol % pure FeCl3 with 10 mol % MeMgCl in the presence of 320 ppm Pd(OAc)2 and 5 mol % SPhos, a highly active nanocatalyst was generated that mediated the desired coupling to deliver pure product in a 95% isolated yield.

Fig. 3 Control reactions documenting the importance of both Fe and Pd in catalyst formation.

Details are provided in the supplementary materials.

Doping pure FeCl3 with 500 ppm of other metals, such as NiCl2, Ni(acac)2, CoCl3, MnCl2, Cu(OAc)2, or CuBr2, led to catalysts that produced variable levels of product formation. In all cases, the yields were ≤38%, as compared with 95% obtained in the presence of added Pd(OAc)2 (table S5).

Solid iron nanoparticles formed from FeCl3 and MeMgCl were collected and analyzed by transmission electron microscopy (TEM) and x-ray photoelectron spectroscopy (XPS) (figs. S8 to S11). As illustrated in fig. S11, XPS analysis revealed that most of this nanomaterial is raft-shaped; composed of large amounts of carbon (57.4%), oxygen (23.6%), magnesium (6.5%), and chlorine (9.8%); and characterized by an essentially 1:1 ratio between iron (1.4%) and phosphorus (in SPhos; 1.3%). The high levels of carbon and oxygen are associated with residual solvent (THF) integrated within these clusters; the C-O signal appears as a shoulder in the C1s spectrum (286.5 eV) (figs. S9 to S11). Only 1.4% iron, in the form of iron oxides (Fe 2p3, 710.86 eV), was present in the nanoparticles produced via reduction of FeCl3 with MeMgCl in THF.

Cryogenic TEM (cryo-TEM) analysis revealed the aggregation of rafts of metal nanoparticles, either inside or around the nanomicellar surface (Fig. 4, A to C). Scanning electron microscopy (SEM), together with energy-dispersive x-ray (EDX) analyses, showed the presence of hybrid nanoparticles containing both iron and ligand (Fig. 4, D and E, and figs. S12 to S14). Further analyses by atomic force microscopy (AFM) (Fig. 4F) revealed an atypical arrangement of iron atoms intermixed mainly with Mg. These particles have a long shelf life (≥1 month at room temperature) and are virtually identical in catalytic activity to those prepared and used in situ.

Fig. 4 Catalyst characterization.

(A to C) Cryo-TEM images of Fe-Pd nanorods in aqueous TPGS-750-M. (D and E) SEM images of the solid nanomaterial. (F) AFM image of the solid nanomaterial.

Thermogravimetric analysis (TGA) of nanomaterial revealed about a 40% total drop in weight between 60° and 145°C, corresponding to the loss of THF bound within the catalyst. Material heated beyond 145°C was stable up to 380°C. However, when the catalyst was preheated at 80°C under a vacuum for 12 hours (fig. S6), a loss of catalytic activity was observed, indicating the importance of THF in maintaining the nanocage structure.

Upon completion of a Suzuki-Miyaura coupling, in-flask extraction with a single organic solvent (e.g., ethyl acetate or methyl tert-butyl ether) at ambient temperatures produces crude material that can be further purified by standard means (fig. S2). The remaining aqueous mixture containing both nanomicelles and nanoparticles of iron can then be recycled, with a modest augmentation of Pd [i.e., 160 ppm Pd(OAc)2] at every other recycling to compensate for losses during extraction. Although the external addition of this Pd salt extends the catalyst activity, the manner in which it is reduced to active Pd(0) and how it is incorporated into the Fe nanoparticles remain unclear. Either the same or different educts can be used in these couplings, indicating the robustness of the process. Alternatively, with solid products, dilution with water could be followed by simple filtration to produce the targeted material directly. The diluted filtrate could be augmented with TPGS-750-M to the original level (2 wt %) and reused, thereby creating little to no wastewater stream. The environmental factor (E factor) (21), a metric of “greenness” that has previously been applied to micellar catalysis (22), is very low (E factor = 3).

We used ICP to analyze the palladium content (<10 ppm) of a product formed via the technology presented here, and we compared the result with that quantified following a traditional Suzuki-Miyaura coupling in organic solvent (fig. S4). Residual palladium in the product derived from a standard coupling in dioxane was far higher than that observed using our nanoparticle approach.

Prospects for incorporating this water-based nanomicelle-nanometal technology into a one-pot sequence of reactions are shown in Fig. 5A. Heteroaryl iodide 4, containing carbamate and trimethylsilyl (TMS) protecting groups, was generated in situ and then subjected to cross-coupling with alkenyl tetrafluoroborate salt 5, using the Fe–ppm Pd nanoparticle protocol. The coupling product 6 was then exposed to aqueous base to remove the TMS groups and effect elimination to 7, followed by butoxycarbonyl (Boc) deprotection to 8. Final aryl amination with bromobenzene to 9 provided entry to the bioactive class of 2,4,5-substituted pyrazol-3-one compounds in a one-pot sequence with an overall isolated yield of 68% (23).

Fig. 5 Further applications of Fe–ppm Pd–catalyzed couplings.

(A) Sequential reactions, including a Suzuki-Miyaura coupling using Fe–ppm Pd nanoparticles as the catalyst (TMSI, TMS iodide; DIPEA, diisopropylethylamine; cBRIDP, di-t-butyl(2,2-diphenyl-1-methyl-1-cyclopropyl)phosphine; KO-t-Bu, potassium t-butoxide; TIPSOH, triisopropylsilyl alcohol). (B) A representative example suggestive of the extension of this approach to Sonogashira couplings (OTBS, t-butyldimethylsilyloxy).

In addition, testing the potential for this mixed-metal catalyst system to effect other important Pd-catalyzed reactions, such as Sonogashira couplings, was carried out (in the absence of added copper), following the example illustrated in Fig. 5B. The prognosis for a similar outcome is good.

Supplementary Materials

www.sciencemag.org/content/349/6252/1087/suppl/DC1

Materials and Methods

Figs. S1 to S18

Tables S1 to S15

References (2434)

REFERENCES AND NOTES

  1. Acknowledgments: We thank Novartis for financial support; J. Feng for technical assistance; M. Cornish for acquisition of the AFM images; S. Kraemer for obtaining the cryo-TEM, SEM, and EDX data; and J. Matthey for providing Pd salts. Parts of this work were carried out in the Characterization Facility, University of Minnesota, which receives partial support from NSF through the Materials Research Science and Engineering Center (MRSEC) program. This work also made use of the University of California–Santa Barbara (UCSB) Materials Research Laboratory Central Facilities, supported by NSF’s MRSEC program under award no. DMR-1121053. ICP-MS analyses were provided by J. Reilly (Novartis, Cambridge, MA). We also acknowledge support from NIH in the form of a Shared Instrument Grant to UCSB(1S10OD012077-01A1). A preliminary patent covering this chemistry has been filed by the University of California–Santa Barbara. The experimental data reported in this paper are available in the supplementary materials.
View Abstract

Subjects

Navigate This Article