Catalytic S N Ar of unactivated aryl chlorides †

We present nucleophilic aromatic substitution of unsubstituted aryl chlorides via a mechanism that is catalytic in [CpRu( p -cymene)]PF 6 and involves a Ru( g 6 -arylchloride) intermediate. From the spectro-scopic evidence we infer that arene exchange is the rate limiting step in this process and develop several new Ru( II ) complexes that lower the activation barrier to arene exchange.

scopic evidence we infer that arene exchange is the rate limiting step in this process and develop several new Ru(II) complexes that lower the activation barrier to arene exchange.
Nucleophilic aromatic substitution (S N Ar) is one of the building blocks of synthetic chemistry, and is used far and wide in industry and academia. However, limitations restrict the use of S N Ar to certain substrates. Activated arenes are required (incorporating an electron withdrawing group to stabilise the negatively charged intermediate, Scheme 1A) and the C-X bond must be polarised; the rate of reaction follows the order X = F 4 Cl c Br. Alternative arylation methods include metal 1a,b and non-metal 1c catalysed coupling reactions of Ar-X and Ar-H.
Fluorobenzene can undergo S N Ar with strong alkoxide nucleophiles 2 but chlorobenzene does not undergo S N Ar (Scheme 1B). One method by which unactivated aryl halides (i.e. those without an electron withdrawing group) can undergo S N Ar is via Z 6 -coordination to a transition metal (e.g. Ru(II), 3 Cr(0), 4 Fe(II), 5 Mn(I), 6 Scheme 1C). It is well established that p-complexation of an arene to a transition metal increases its reactivity towards nucleophilic attack 7 and deprotonation. 8 More recently, activation of a Cr(0) Z 6 -aryl C-H bond towards Pd insertion and subsequent arylation has been achieved. 9 The drawback to these reactions is that the Z 6 -arene-metal bond is strong and stoichiometric metal is required; liberation of the aryl product is carried out by photolysis 10 or oxidation. 11 Semmelhack et al. propose a catalytic cycle in which an unactivated aryl chloride binds Z 6 to Cr(0), facilitating S N Ar, before exchange between the bound product and free aryl chloride takes place, completing the catalytic cycle (Scheme 2). 12 Despite demonstrating an increase in the rate limiting step (arene exchange), no catalytic S N Ar was reported.
Rh(III) catalysed intramolecular S N Ar based on this catalytic cycle has been reported but is limited to aryl fluorides. 13 Intermolecular S N Ar of aryl fluorides (Ru(II) catalysed) has also appeared 14 but requires a large excess of arene and is unsuccessful with aryl chlorides. Anti-Markovnikov hydroamination of styrene via catalytic p-complexation to Ru(II) has also been reported. 15 In this report we present the first example of catalytic S N Ar of unactivated aryl chlorides. Experimental evidence suggests that our method proceeds via a Z 6 -coordination mechanism.
In an effort to reduce the high reaction temperature and long reaction time, several new catalysts were synthesised with a view to lowering the activation barrier of the rate limiting step.
Our initial investigation focussed on the reaction between morpholine and chlorotoluene under a variety of reaction conditions (selected results are given in Table 1, see ESI † for a complete set of reaction conditions). In the absence of a catalyst no product formed, despite attempts over a range of temperatures and in several solvents (Table 1, entries 1-3).
Based on the work of Woodgate et al., 3a we anticipated that catalysts incorporating a [CpRu] + fragment (Cp = cyclopentadienyl) would expedite the S N Ar reaction, via Z 6 -coordination of chlorotoluene. A number of catalysts were investigated (Table 1, entries 4-6). Each catalyst led to conversion into the desired product, providing that high temperatures were employed, with [CpRu( p-cymene)]PF 6 performing best. In a solvent screen at 180 1C, 1-octanol was found to give the highest conversion (  [12][13][14][15]. No increase in product conversion was observed for a variety of additives, including base, molecular sieves and ligands. In summary, catalytic S N Ar between morpholine and chlorotoluene can be achieved with 90% yield in 1-octanol at 180 1C in 14 days. We sought to understand why such long reaction times and high reaction temperatures were required. Under the optimised reaction conditions, positive mode electrospray mass spectrometry (ESI-MS + ) showed a peak at m/z = 344 with the characteristic Ru isotope pattern, corresponding to the Z 6 -bound product (complex 1b in Scheme 3). 1 H-NMR showed a pair of doublets in the range 5.5-5.7 ppm, also corresponding to 1b; no Z 6 -bound chlorotoluene (1a) was observed by ES-MS + or 1 H-NMR. We postulate, therefore, that the reaction occurs via the mechanism shown in Scheme 2 and that arene exchange is the rate limiting step; S N Ar is fast so that no bound chlorotoluene is observed but arene exchange is slow so that bound product is observed.
The mechanism by which arene exchange takes place in [Cr(CO) 3 (Z 6 -arene)] complexes has been well studied. 16 It was determined that the rate-determining step is an initial change in arene bonding from Z 6 -Z 4 (Scheme 4). While similar analysis has not been carried out for Ru II (Z 6 -arene) complexes, we assume the same mechanism for arene exchange and propose that the success found for our catalytic system with 1-octanol as a solvent is due to coordination of the hydroxyl group during the slow Z 6 -Z 4 step. This maintains an 18 e À complex and lowers the activation energy. This theory is supported by the reasonable product conversion in solvents such as cyclohexanol and DMI (Table 1, entries 8 and 9) and the lack of conversion in the non-coordinating solvent p-xylene (Table 1, entry 10). It has previously been noted that addition of ketones, such as cyclohexanone, lowers the activation barrier to arene exchange in Cr(0) systems via coordination of the carbonyl lone pair during the Z 6 -Z 4 process. 17 To test this theory, several new catalysts were synthesised (2-6), each incorporating a tether, covalently bound to the Cp ring, capable of coordinating to Ru during the Z 6 -Z 4 step (Scheme 4). 18 Our expectation was that intramolecular coordination of the tether to the Ru centre would lower the activation barrier to arene exchange and, subsequently, increase the rate of the S N Ar reaction.
Each complex was synthesised by reaction between [Ru( p-cymene)Cl 2 ] 2 and the corresponding Cp-tether adduct, in the presence of Na 2 CO 3 and ethanol (Scheme 5). 19 The reaction mixtures were treated with NH 4 PF 6 to afford complexes 2-6. The formation of the Cp-tether adducts were specific to the various tethers. For complex 2, CpNa and BrCH 2 CO 2 Me were stirred in THF at 0 1C to give CpCH 2 CO 2 Me in quantitative yield 20 before complexation with [Ru(p-cymene)Cl 2 ] 2 . 21 Similar protocols were used to synthesise the other Cp-tether compounds. In certain cases, it was necessary to synthesise the electrophilic component before reaction with CpNa. For example, in the synthesis of 5, 2-Py(CH 2 ) 2 OH (2-Py = 2-pyridine) was converted into 2-Py(CH 2 ) 2 OSO 2 Me, prior to reaction with CpNa. Each complex 2-6 was purified by column chromatography on silica and fully characterised. For full Experimental detail see ESI. † To investigate whether the presence of the various tethers would decrease the energy barrier for the Z 6 -Z 4 process, exchange experiments were carried out. Each complex was heated in either 1-octanol or cyclohexanone in the presence of 100 equivalents of hexamethylbenzene (C 6 Me 6 ) for 16 h. The extent of exchange between p-cymene and C 6 Me 6 was measured at 3 h and 16 h using ESI-MS + ( Table 2); calibration of the ESI-MS + measurement was required to allow quantitative assessment (see ESI †). Relative to the parent complex, [CpRu( p-cymene)]PF 6 (1), a moderate increase in the extent of arene exchange was observed for 3 and 6 in cyclohexanone after 16 h (51% and 50%, respectively versus 38% for 1). Most notable, however, was the complete arene exchange in cyclohexanone for complex 5, which incorporates a pyridyl tether. In the same solvent, 2 appeared to undergo ester hydrolysis and decarboxylation, giving [(CpMe)Ru( p-cymene)] + , indicated by a m/z peak at 315. This species underwent a similar amount of arene exchange to 1. Finally, exchange for 4 in cyclohexanone could not be determined due to a side reaction with the solvent. Similar exchange behaviour was found when 1-octanol was used as the solvent. Complex 5 again displayed the greatest amount of arene exchange. The tether components in 3 and 5 rapidly reacted with the solvent to form octyl esters, which exchanged at a similar rate to the parent complex (1). Once again, complex 2 decarboxylated whilst 6 showed a slight reduction in the rate of exchange compared to 1.
Having established that arene exchange could be accelerated by the presence of a coordinating tether, we proceeded to calculate half lives (t 1/2 ) for the initial p-cymene complexes of 1, 3, 5 and 6 under the exchange conditions described in Table 2, with cyclohexanone as the solvent. The results confirm that the pyridyl complex 5 exchanges an order of magnitude faster than the parent complex, 1 (t 1/2 = 2.2 AE 0.1 h versus 34 AE 0.7 h, see ESI † for details). Complexes 3 and 6 each have shorter t 1/2 than 1 (23 AE 0.3 h and 28 AE 0.4 h, respectively), confirming that the rate of arene exchange is accelerated by the coordinating tether.
Our hypothesis stated that faster arene exchange will lead to higher conversion in the S N Ar reaction between morpholine and chlorotoluene. To test this, we carried out the S N Ar reaction with three of the new potential catalysts, 1, 3 and 5 (Table 3). As a comparison, 1 with an equivalent of free pyridine was also tested.  After 24 h, each catalyst gave a similar conversion. After 4 d, the catalysts incorporating a tether showed a small improvement in conversion but after 7 d the parent compound returned the highest value, due, in the case of 5, to degradation of the catalytic species (indicated by loss of Ru species in the mass spectrum). Once again, mass spectrometric analysis after 24 h shows a m/z peak corresponding to the Z 6 -bound product in each of the reactions. Ultimately, it appears that although the rate of arene exchange is accelerated in the p-cymene -C 6 Me 6 experiment described above, this acceleration does not translate into a higher conversion in the S N Ar reaction. Two explanations present themselves: (1) the Z 6 -bound product is more kinetically inert than [CpRu(p-cymene)] + , so that arene exchange is slower, (2) the Z 6 -bound product is more thermodynamically stable than [CpRu(Z 6 -chlorotoluene)] + , leading to the build-up of the product bound complex and long reaction times. Of course, a combination of both factors may be present. [(Z 6 -Dimethylaniline)Cr(CO) 3 ] is known to be more thermodynamically stable than the chlorobenzene analogue 16a,17 and a similar trend is likely in the case of [CpRu(Z 6 -arene)] + -the aryl ring in the product (e.g. 1b in Scheme 3) is more electron rich than p-cymene. In an attempt to destabilise the Z 6 -bound product, the nucleophile was changed from morpholine to 2,2,6,6-tetramethylmorpholine. Formation of product (either free or bound) was not observed, likely due to the steric influence of the nucleophile. Efforts are ongoing to understand and overcome the limiting factors of the reaction.
In conclusion, we have demonstrated for the first time catalytic S N Ar of unactivated aryl chlorides, albeit at high temperatures and with long reaction times. In an attempt to move to milder reaction conditions, several new Ru complexes were synthesised, incorporating tethers capable of lowering the activation barrier to dissociation. Despite achieving an 18-fold increase in the rate of arene exchange, the rate of S N Ar was not significantly increased in our test reactions. Once an efficient catalytic system based on Z 6 -arene exchange is realised, we envisage many applications in efficient organic synthesis.