Zhuangli
Zhu
ab,
Zhenhua
Wang
a,
Yajun
Jian
a,
Huaming
Sun
a,
Guofang
Zhang
a,
Jason M.
Lynam
c,
C. Robert
McElroy
c,
Thomas J.
Burden
c,
Rebecca L.
Inight
c,
Ian J. S.
Fairlamb
*c,
Weiqiang
Zhang
*a and
Ziwei
Gao
*a
aKey Laboratory of Applied Surface and Colloid Chemistry, Ministry of Education, School of Chemistry & Chemical Engineering, Shaanxi Normal University, 199 South Chang'an Road, Xi'an, China. E-mail: zwq@snnu.edu.cn; zwgao@snnu.edu.cn
bTianjin Normal University, 393 West Binshui Road, Tianjin, China
cDepartment of Chemistry, University of York, Heslington, York, North Yorkshire YO10 5DD, UK. E-mail: ian.fairlamb@york.ac.uk
First published on 7th January 2021
The dual function and role of iron(0) pentacarbonyl [Fe(CO)5] has been identified in gaseous CO-free carbonylative Suzuki–Miyaura cross-couplings, in which Fe(CO)5 supplied CO in situ, leading to the propagation of catalytically active Pd–Fe nanoparticles. Compared with typical carbonylative reaction conditions, CO gas (at high pressures), specialised exogenous ligands and inert reaction conditions were avoided. Our developed reaction conditions are mild, do not require specialised CO high pressure equipment, and exhibit wide functional group tolerance, giving a library of biaryl ketones in good yields.
Over the last few decades we have seen practical advances in C-SMCCs, especially enabled through the design of new Pd (pre)catalysts.5 However, the toxic and flammable CO gas potentially hinders wider adoption of C-SMCCs, especially so in academic laboratories, for which expertise and specialised equipment might be limited, with health and safety concerns being at the forefront of any decision-making process. It is also known that the CO-insertion efficiency is reduced when employing electron-deficient aryl halides,6 which can typically be overcome by the adoption of an expensive exogenous ligand or high pressure in CO (Fig. 1a).7
Moving the field forward, the development of a carbonylation procedure employing the in situ generation of CO, as a substitute for gaseous CO, is appealing and arguably necessary for general synthetic chemistry laboratories. With this point in mind several research teams have replaced CO gas with safer sources of molecular CO, i.e. in the form of CO-surrogates.8,9 Indeed, leaders in the field are Skrydstrup10et al., who have developed Pd-catalysed carbonylation reactions using ex situ CO generated from CO-surrogates and a specially designed two-chamber reactor (Fig. 1b). Several CO surrogates have been employed, e.g. 9-methyl-9H-fluorene-9-carbonyl chloride (COgen), silacarboxylic acids (SilaCOgen), alcohols, carbon dioxide, metal carbonyls and formic acid. These CO-surrogates degrade in a controllable manner, by different mechanisms, supplying CO gas for Pd-catalysed carbonylation reactions.10,11
With synthetic verstaility in mind, the well-controlled in situ generation of CO is essential for the development of a safe and convenient one-pot C-SMCC methodology. Transition metal carbonyls (TMCs) have been utilised as CO releasing molecules (CO-RMs) for therapeutic delivery of CO in biological systems,12 a field which we have contributed to independently, and for which much is now known about CO-release rates in different media. Many readily available homoleptic TMCs are attractive CO-RMs, since many carbonyls per CO-RM can in principle be delivered in solution. More importantly, the cleavage of the M–CO bond of TMCs can be controlled by the variation of temperature. Ni(CO)4 has been employed in the aminocarbonylation of vinyl halides13a and carbonylative olefination of aryl halides.13b Larhed et al.14 pioneered Pd-catalysed carbonylation of O, N nucleophiles using Cr(CO)6 and Mo(CO)6. The solid CO-RMs were activated by microwave heating, and CO was generated in situ for carbonylation. Later, other solid TMCs were evaluated as CO-RMs for transition metal-catalysed organic transformations.10 In all cases, the metal is thought to be a waste element within the reaction. Indeed, a limitation identified was that the low valence transition metal residue could poison the active Pd catalyst species, with reactions being sluggish, leading to poor yields of the carbonylative products,15a a potential issue for more challenging substrates.15b,c Maes et al. have developed methodologies employing isocyanides as CO-surrogates in Pd-catalysed cross-couplings, offering an alternative approach.16
In light of highly effective and catalytically competent Pd/Fe nanoparticles being discovered for C–C cross-coupling reactions,17 we recognised that iron carbonyls could potentially play the role of CO–RMs, while modifying the properties and catalytic activity of the Pd catalyst through beneficial interactions with Fe, i.e. where the CO ligands are utilised for the carbonylation process and the waste Fe (oxidised) creating a more efficient Pd–Fe catalytic system in situ. Due to the low-price and earth abundance of iron, we envisaged that an economical, safer and environmentally friendly Pd-catalysed C-SMCC methodology using Fe(CO)5 could be developed.
Herein, we report a benign C-SMCC reaction methodology, which does not require the use of CO gas or exogenous specialised ligands, where Fe(CO)5 provides CO for the reaction, generating well-defined, stable and catalytically competent Fe/Pd nanoparticles in situ, which can be further recycled in multiple reaction runs (Fig. 1c).
Entry | Catalyst | Ligand | 3a | 3a’ | |
---|---|---|---|---|---|
Solvent/base | %b | ||||
a Reaction conditions: 1a (0.434 mmol), 2a (1.75 eq.), Fe(CO)5 (0.15 mmol), and [Pd] (1 mol%, unless otherwise stated), 80 °C, 12 h. Entries 1–12: anisole and K2CO3 were used. b Detected by 1H NMR and confirmed by purification by silica gel chromatography. c Pd3(OAc)6, referred to as Pd(OAc)2, (0.5 mol% in Pd). | |||||
Effect of catalyst and ligand (1![]() ![]() |
|||||
1 | Pd(OAc)2 | 90 | 7 | ||
2 | Pd(OAc)2c | 86 | 8 | ||
3 | Pd2(dba)3 | 80 | 7 | ||
4 | Pd(acac)2 | 89 | 7 | ||
5 | PdCl2 | 80 | 18 | ||
6 | PdCl2(PPh3)2 | 5 | 94 | ||
7 | Pd(OAc)2 | PPh3 | 64 | 31 | |
8 | Pd(OAc)2 | CyJohnPhos | 31 | 22 | |
9 | Pd(OAc)2 | PCy3 | 71 | 21 | |
10 | Pd(OAc)2 | Dppf | 70 | 25 | |
11 | Pd(OAc)2 | Xphos | 40 | 12 | |
12 | Pd(OAc)2 | Xantphos | 48 | 14 | |
Effect of solvent (3 mL) and base (2 eq.) | |||||
13 | Pd(OAc)2 | Toluene/K2CO3 | 87 | 11 | |
14 | Pd(OAc)2 | Dioxane/K2CO3 | 60 | 7 | |
15 | Pd(OAc)2 | Anisole/K3PO4 | 83 | 10 | |
16 | Pd(OAc)2 | Anisole/DIPEA | 51 | 1 |
The effects of the solvent and base were further examined. It was found that the C-SMCC reaction proceeded well in relatively less polar solvents, such as anisole and toluene. The inorganic bases, K2CO3 and K3PO4, exhibited high substrate conversion and superior selectivity when compared against the organic base, DIPEA (entries 13–16). Other TMCs were also assessed as CORMs. Indeed, the ferrous-system performed considerably better than Mo(CO)6 and Cr(CO)6 (Table S2, see the ESI†). It is of particular note that using gaseous CO (balloon pressure) in place of Fe(CO)5 failed to effect a successful C-SMCC reaction, with only the SMCC biaryl product 3a′ being formed. The outcome demonstrates that Fe(CO)5 not only acts as a CO source, but also functions positively vide infra to enhance this specific Pd-catalysed C-SMCC reaction in some way.
With the optimised reaction conditions in hand {0.15 mmol Fe(CO)5, 1 mol% Pd(OAc)2, anisole and K2CO3}, various functional biaryl ketones were synthesized using the new C-SMCC reaction differentiated by changes in aryl substituents, either electron-donating or electron-withdrawing, of aryl iodides (Table 2) and those of phenylboronic acids (Table 3), to evaluate the scope and limitations of the new C-SMCC methodology.
Table 2 shows the effects of substituents in the reactions of aryl iodides. Under the best identified reaction conditions, the para-halogenated (Cl, Br, and F) aryl iodides gave diaryl ketones in good yields (p-F, 3b, 79%, p-Cl, 3f, 73%, and p-Br, 3n, 74%). However, ortho-Cl and Br substituents led to only modest yields of cross-coupled products with 50% (3j) and 30% (3o) biaryl ketone products being generated respectively. We tentatively attribute the lowering of the yield to steric effects, which retard the oxidation addition step.18 The presence of a methyl group at para-, meta- and ortho-positions led to good product yields (p-Me, 3c, 84%, m-Me, 3g, 74%, and o-Me, 3k, 72%). The corresponding biaryl ketone products possessing methoxy groups at para-, meta- and ortho-positions were isolated in good yields, with 74% for 3e, 53% for 3i and 78% for 3m, respectively. Meanwhile, as the length of the alkyl chain was increased to ethyl a 66% yield of 3d was recorded. As for electron-withdrawing substituents, similar trends were noted. For example, the weakly electron-withdrawing ester group, p-CO2Me, gave a good yield of 80% (3q), with the p-CO2Et group giving a 73% yield of 3u. In addition, the stronger electron-withdrawing substituents only lower the yields marginally (p-CF3, 3h, 57%, p-NO2, 3p, 68%, and m-NO2, 3t, 65%). We further succeeded in synthesizing the biaryl ketones containing naphthalene (3r, 55%) and thiophene (3s, 54%), which are key intermediates to interesting luminescent materials.19
As shown in Table 3, different organoboronic acids were utilised in our new C-SMCC methodology. Most of the organoboronic acids underwent carbonylative coupling with good yields. The halogenated aryl iodides were converted into diaryl ketones in very good yields (e.g. p-F, 3b, 84%, p-Cl, 3f, 89%, and p-Br, 3n, 74%). Both electron-donating (p-Me, 3c, 88% and p-OMe, 3e, 80%) and electron-withdrawing substituents (p-CO2Me, 3q, 84%) also gave good yields. The 9-fluorenyl arylboronic acid (4d, 74%) was pleasingly formed with a good yield. Notably, thio-moieties were well tolerated. Thus, the biarylketone products bearing p-MeS and 2-thiophene were both isolated in 65% yield (4a and 3s). A sterically encumbered substituent, o-iPr, gave rise to a 73% yield of 4b, whilst the organoboronic acids bearing several fluorine atoms or a p-SO2Me group gave lower yields of 23% (4c) and 37% (4e), respectively.
Encouraged by the results presented above, we subsequently carried out a C-SMCC reaction on a gram scale to demonstrate its practical utility. With 5 mmol of 1a, product 3a was formed in 86% yield under our optimised conditions (Scheme 1). Furthermore, the phenylenebis–phenylmethanones were synthesized via the C-SMCC reactions of various diiodobenzenes with phenylboronic acid, in up to 86% yield (Scheme 1, 5a–c). The results highlight the significant potential of this C-SMCC for practical applications.
Finally, we have tested the C-SMCC methodology using 4-bromo-anisole and 4-bromo-6-methyl-2-pyrone as exemplar brominated substrates (employing standard conditions as described in Table 2, in the presence and absence of activating DPPF and Xantphos ligands which we hypothesised might assist with C–Br bond activation and CO-migratory insertion steps). However, in the case of 4-bromo-anisole it was left unreacted. For 4-bromo-6-methyl-2-pyrone, a small amount of 4-phenyl-6-methyl-2-pyrone was formed (7%) when the exogenous Xantphos ligand was added (by traditional SMCC reaction).
The data in Table 4 show that the green metrics for the two synthetic methodologies are broadly similar. Surprisingly, this includes atom economy (AE), with our work only being slightly poorer due to one fragment of “Fe(CO)2” not making it into the final product (note: the final destination for Fe is in the form of Fe2O3vide infra which is required for Pd catalyst enhancement). However, the AE is dominated by the iodine and boronic acid leaving groups, a necessary and mandatory requirement for the arylation process. More significant however is the Reaction Mass Efficiency (RME) being significantly better for our work. This is because the CO–RM, Fe(CO)5, is an efficient and mild CO source, while the reported reaction conditions employ a 23 molar excess under significant pressure. The same effect can be seen in the Process Mass Intensity (PMI) reaction chemicals. These results are likely more significant, as 3× refills and purges of CO are necessary for the reported21 pressurised reaction vessel. The inclusion of this practical and necessary requirement gives an RME of 6.7% and a PMI of 14.9. Neither reaction scores well on health and safety, due to the need for liberation of CO (which is formally a necessary requirement for a carbonylation reaction). Lastly, the lower temperature and absence of the need to pressurize give an improved energy demand for our C-SMCC methodology.
To further characterize the catalytically active Pd–Fe particles, Transmission Electron Microscopy (TEM) measurements were conducted. The experiments showed the presence of 2 nm sized Pd–Fe nanoparticles over a large area (histogram inset in Fig. 2a). Lattice fringes were observed in the high-resolution TEM images (Fig. 2b), showing that the inter-planar distances approximate to 0.226 nm and 0.198 nm, corresponding to the (200) and (111) planes of the Pd metal. The assignment was confirmed using Fast Fourier Transform (FFT) and inverse FFT images (inset in Fig. 2b).
The chemical state of the Pd–Fe nanoparticle surface was analysed by XPS (Fig. 3 and ESI S3†). Fig. 3 shows the XPS core level spectra for Pd 3d and Fe 2p, characterized by spin–orbit splitting (Pd 3d5/2, 3d3/2 and Fe 2p3/2, 2p1/2 components), where the peaks at 335.6 and 340.8 eV result from Pd 3d5/2 and Pd 3d3/2, respectively. The Pd species are characterised as being in an oxidation state of zero. FeIII was confirmed based on the peaks at 710.6 and 724.1 eV, corresponding to Fe 2p3/2 and Fe 2p1/2, respectively. The side peaks were attributed to the oxide; thus Fe exists in the form of Fe2O3. The EDAX mapping analysis confirmed the composition of the Pd–Fe nanoparticles deriving from Fe, Pd and O, which is fully consistent with our deduction (ESI S4†). SEM images show the morphology of the Pd–Fe nanoparticles (Fig. S5, see the ESI†).
![]() | ||
Fig. 3 XPS images of the Pd–Fe catalyst nanoparticle material, with Fe2O3 and Pd0 prominent species. |
Finally, the recyclability of the Pd–Fe nanoparticles was assessed. The Pd–Fe nanoparticles were first isolated from a reaction of 1a + 2a → 3a, by filtration, and then sequentially washed with water and ethyl acetate. The C-SMCC reaction 1a + 2a → 3a was then recharged with fresh Fe(CO)5 and the isolated catalyst. At each cycle we found that the isolated catalyst can be recycled giving 3a, with the yields of 3a going from 90 to 50% over the 4 cycles. The deactivation of the heterogeneous Pd–Fe nanocatalyst species can be attributed to an excess Fe residue being deposited onto the heterometallic surface, either blocking the catalytically active sites or causing a restructuring of the catalyst surface (Fig. 4A).15,21 Interestingly, in the absence of Fe(CO)5, the isolated Pd–Fe nanoparticles also independently catalyzed the efficient SMCC reaction of 1a + 2a → 3a′ (Fig. 4B). A small amount of carbonylative product 3a is formed in the first cycle, formed from residual CO from the Pd–Fe nanoparticle catalyst. Crucially, the yield of 3a′ was over 80% for the subsequent cycles (2–5). These results are in accordance with the unique cooperative activity of a Pd–Fe bimetallic catalytic system reported by Lipshutz et al. for standard SMCC reactions (Fig. 4B).17
Footnote |
† Electronic supplementary information (ESI) available: Experimental details and characterization. See DOI: 10.1039/d0gc03036h |
This journal is © The Royal Society of Chemistry 2021 |