Palladium-Catalyzed C–P Bond-Forming Reactions of Aryl Nonaflates Accelerated by Iodide

An iodide-accelerated, palladium-catalyzed C–P bond-forming reaction of aryl nonaflates is described. The protocol was optimized for the synthesis of aryl phosphine oxides and was found to be tolerant of a wide range of aryl nonaflates. The general nature of this transformation was established with coupling to other P(O)H compounds for the synthesis of aryl phosphonates and an aryl phosphinate. The straightforward synthesis of stable, isolable aryl nonaflates, in combination with the rapid C–P bond-forming reaction allows facile preparation of aryl phosphorus target compounds from readily available phenol starting materials. The synthetic utility of this general strategy was demonstrated with the efficient preparation of an organic light-emitting diode (OLED) material and a phosphonophenylalanine mimic.


■ INTRODUCTION
Aryl phosphorus compounds are important due to their widespread application in organic, medicinal, and materials chemistry. 1,2 As a consequence, carbon−phosphorus bond formation is a highly active area of research in organophosphorus chemistry. Traditionally, aryl C−P bonds were formed via the reaction of Grignard or organolithium reagents with electrophilic phosphorus compounds. 3 In 1981, pioneering work by Hirao and co-workers demonstrated that aryl C−P bonds could be generated by palladium-catalyzed crosscoupling reactions of aryl bromides with P(O)−H compounds (Scheme 1a). 4 Since the discovery of the Hirao reaction, efforts have focused on extending the range of electrophilic aryl substrates, elucidation of the reaction mechanism and optimization of the reaction conditions. 5,6 Despite the availability of aryl halides as coupling reagents, many complex arenes, particularly natural product-based (e.g., steroids and amino acids), exist only in phenolic form. For this reason, Hirao-type reactions using activated sulfonates have been reported. Aryl triflates have been explored as substrates, 5,6e,7 but the high cost and reactive nature of reagents limits applications. This has led to the development of metal-catalyzed aryl C−P bond-forming reactions using mesylates and tosylates. 5,8 For example, the Kwong group has demonstrated the effective phosphorylation of aryl mesylates and tosylates using low catalyst loadings of Pd(OAc) 2 in combination with the CM-Phos ligand (Scheme 1b). 9 Transformations at 110°C and a reaction time of 18 h gave a wide range of phosphonate esters in high yields. Recently, the Ding and Xu groups independently reported Pdcatalyzed C−P bond-forming reactions of aryl fluorosulfonates. 10 Following synthesis of these from phenols and sulfuryl fluoride gas, these compounds were readily coupled with a range of P(O)−H compounds using Pd(OAc) 2 and either dppf or DPEPhos ligands.
Although advances in palladium-catalyzed C−P bondforming reactions with aryl sulfonates have been achieved, we were interested in developing a method with a short reaction time, which avoided the need for additional ligands or gaseous reagents, and that could also be applied for the preparation of a range of aryl phosphorus compounds. Aryl nonafluorobutylsulfonates [nonaflates, ArOSO 2 (CF 2 ) 3 CF 3 ] are easily prepared from phenols and the inexpensive, industrial product nonaflyl fluoride. In addition, these are stable and can be readily purified by flash column chromatography. For these reasons, aryl nonaflates have been used for a wide range of palladium-catalyzed cross-coupling reactions. 11,12 However, utilization of these for analogous C−P bond-forming reactions are relatively rare. Apart from a few specific examples, 13 the only methodology study was reported by the Lipshutz group, who demonstrated the efficient synthesis of triarylphosphine boranes via the reaction of aryl nonaflates with diphenylphosphine-borane. 14 Herein, we disclose a palladium-catalyzed C− P bond-forming reaction with aryl nonaflates that can be accelerated by iodide, resulting in short reaction times (Scheme 1c). The method does not require additional ligands or substrates prepared by gaseous reagents. Furthermore, we demonstrate that the method can be used as part of an effective strategy for the synthesis of important organophosphorus compounds from phenol starting materials.

■ RESULTS AND DISCUSSION
Initial studies focused on the reaction of diphenylphosphine oxide with the nonactivated starting material, p-tolyl nonaflate (1a) ( Table 1). As previous work has shown palladium acetate as an effective catalyst for C−P bond formation, 6 this was used in combination with triethylamine, originally utilized as a base by the Hirao group. 4 In addition, dimethylformamide (DMF) was chosen as an effective solvent for working with aryl nonaflates. Using 1 equiv of diphenylphosphine oxide, at a reaction temperature of 90°C, showed only 70% conversion by 1 H nuclear magnetic resonance (NMR) spectroscopy, resulting in a 41% isolated yield (entry 1). Mechanistic work of palladium-catalyzed C−P bond-forming reactions by the Stawinski, 6a,c Montchamp, 6f,g and Keglevich groups 6h,i have shown that excess amounts of the P(O)H coupling partner are required to reduce the palladium(II) catalyst and act as a ligand (see Scheme 6). Therefore, using 1.5 equiv of diphenylphosphine oxide and an increased reaction temperature of 110°C, allowed full conversion after 24 h and a 58% isolated yield (entry 2). A further increase of reaction temperature to 120°C resulted in further improvement in isolated yield (79%); however, the reaction still required 24 h to reach completion (entry 3). As ionic additives such as chloride and acetate ions are well known to promote Pdmediated cross-coupling reactions, 15,16 and facilitate the Hirao reaction, 6a,c,e these were investigated to improve the reaction time. Interestingly, the addition of stoichiometric quantities of NaOAc or NaCl (entries 4 and 5) led to no improvement in the reaction time and gave phosphine oxide 2a in lower isolated yields. In contrast, the addition of NaI (1 equiv) resulted in a significantly faster reaction time of 4 h, which gave 2a in 78% yield (entry 6). This effect was observed to a lesser extent using 0.1 equiv of NaI (entry 7). In this case, the reaction was complete after 8 h.
Having identified rapid and efficient conditions for the synthesis of 2a, the scope of the iodide-accelerated reaction was investigated for the coupling of diphenylphosphine oxide with various aryl nonaflates (Scheme 2). Using NaI (1 equiv) throughout, the process was found to be compatible with a wide range of substituents and functional groups, forming the majority of diphenylphosphine oxides after 4 h reaction times. Some variations to the standard conditions were observed. For  24  90  41  2  24  110  58  3  24  120  79  4 NaOAc (1)  22  120  55  5 NaCl (1)  32  120  64  6 NaI (1) 4 120 78 7 NaI (0.1) 8 120 76 example, the reaction of naphthyl analogue 1f was found to proceed at 90°C and was complete after 3 h, while aryl nonaflates with ortho-substituents (1c) or with strong electrondonating groups (1h) required slightly longer reaction times. Although the reaction conditions tolerated chloride substituents (1p), attempted coupling of 3-bromophenyl nonaflate (1q) with diphenylphosphine oxide (1.5 equiv) gave a mixture of compounds. Analysis of the reaction mixture by 1 H NMR spectroscopy showed the presence of bis-phosphine oxide 2q as the major product, along with mono-phosphine oxide byproducts. As a selective reaction was not possible, 1q was allowed to react with 3 equiv of diphenylphosphine oxide, which gave bis-phosphine oxide 2q in 57% yield. Pyridin-2-yl nonaflate (1r) was also a substrate for this transformation, giving clean conversion to 2r in 60% yield. From the series of nonaflates investigated, only a p-nitrophenyl analogue failed to generate the desired product. In this case, the reaction conditions led to decomposition of the nonaflate. Using p-tolyl nonaflate (1a) as a standard substrate, the study then investigated the use of the reaction for the preparation of other aryl C−P bonds (Scheme 3). In a similar manner to the synthesis of diphenylphosphine oxide 2a, the iodide-accelerated reaction with Pd(OAc) 2 permitted the synthesis of phosphine oxide 3a. While Pd(OAc) 2 did allow the preparation of other aryl C−P-containing compounds, the reactions were less efficient, leading to the products in moderate yields (40−50%). For this reason, a brief screen for alternative catalysts was performed that identified Pd(PPh 3 ) 4 as an effective substitute. 17 Reaction of 1a with di-n-butylphosphine oxide in the presence of Pd(PPh 3 ) 4 and NaI gave dialkylphosphine oxide 3b in 58% yield, after a reaction time of 5 h. Reaction of 1a with the more reactive coupling partners, ethyl phenylphosphinate and diethyl phosphite was found to proceed at 80°C and after reaction times of 4 and 6 h, respectively, gave phosphinate 3c and phosphonate 3d in good yields.
The study next investigated the combination of the mild conditions for nonaflate synthesis with the accelerated aryl C− P bond-forming reaction for the simple conversion of phenols to aryl phosphorus-containing targets (Scheme 4). Pyrene nonaflate 5 was prepared in 87% yield by the treatment of 1hydroxypyrene (4) with nonaflyl fluoride, under basic conditions. Reaction of 5 with diphenylphosphine oxide, using Pd(OAc) 2 and NaI gave phosphine oxide 6, a blue lightemitting diode material in 73% yield. 18 In a similar manner, commercially available L-tyrosine derivative 7 was converted to the corresponding aryl nonaflate 8 under mild conditions, in 94% yield. Iodide-accelerated phosphorylation of 8, performed at a 1 mmol scale, was found to proceed at 80°C, and after a reaction time of 6 h, gave phosphonate ester 9 in 72% yield. With this transformation, a lower loading of the palladium catalyst was investigated. Using 5 mol % Pd(PPh 3 ) 4 showed no significant difference in reaction efficiency. Again, at a 1 mmol scale, the transformation was complete in 7 h and produced phosphonate ester in 65% yield. Acid-mediated deprotection allowed the isolation of phosphonophenylalanine 10, a compound used for various medicinal chemistry applications, such as a component of peptides that act as thrombin inhibitors and as competitive N-methyl-D-aspartic acid antagonists. 7a, 19 Having demonstrated the utility of this method, the possible role of iodide in accelerating the C−P bond-forming process was considered. Initially, the different rates observed during the reaction of p-tolyl nonaflate (1a) with diphenylphosphine oxide in the presence of NaI (0, 0.1, and 1 equiv) were further investigated. A conversion graph generated by 1 H NMR spectroscopy confirmed that while the reaction with NaI (1 equiv) was complete after 4 h (∼95% conversion), only 12% conversion was observed at the same time during the reaction without NaI (Figure 1).

Scheme 3. Reaction Scope for the Synthesis of Various Aryl Phosphorus Compounds a,b
Halide and acetate additives have been shown to promote Pd-catalyzed cross-coupling reactions by the formation of more nucleophilic anionic palladium complexes. 15, 16 However, no accelerating effects were observed when acetate or chloride ions were employed during this transformation (Table 1). It has also been proposed that iodide accelerating effects during Pd-catalyzed cross-coupling reactions are due to the faster oxidative addition of aryl iodide intermediates formed in situ via a Finkelstein reaction. 20 Using the optimized conditions for the coupling of p-tolyl nonaflate (1a) with diphenylphosphine oxide, control experiments were conducted to determine whether p-tolyl iodide (11) was an intermediate (Scheme 5).
Repeating the reaction under the same conditions, but in the absence of either diphenylphosphine oxide or Pd(OAc) 2 , no iodide could be detected (by 1 H NMR spectroscopy), even after 24 h. A final experiment to probe the mechanism investigated the use of p-tolyl iodide (11) as the starting material. Previous work using aryl iodides as substrates for palladium-catalyzed C−P cross-coupling reactions reported lower yields compared to other halide leaving groups. 21 It was proposed that this was due to competing reduction of the ArPdI intermediate. Reaction of p-tolyl iodide (11) using our non-catalyzed, standard conditions was found to be fast, with completion observed after 1.5 h. However, this gave phosphine oxide 2a in only 34% isolated yield. This is in contrast to ptolyl nonaflate (1a), which under the same conditions required a reaction time of 24 h but gave 2a in 79% yield (Table 1, entry 3). This difference in reaction times and isolated yields of 2a suggest that an aryl iodide and the subsequent oxidative addition product, ArPdI are not intermediates during the reaction with aryl nonaflates and that the reaction of these proceeds via an alternative mechanism in the presence of iodide.
Based on these results and previous mechanistic studies by the Stawinski, 6a,c Montchamp, 6f,g and Keglevich groups, 6h,i that implicated the role of the tautomer form of diphenylphosphine oxide as a reducing agent to form Pd(0), as a ligand and as the nucleophilic coupling partner, we propose the following catalytic cycle (Scheme 6). Initially, the active palladium species I is formed by reduction and coordination with the tautomeric form of the excess P(O)H coupling reagent (30 mol % required for 10 mol % Pd catalyst). Following oxidative addition of the aryl nonaflate by Pd(0) species I, the presence of NaI may result in the formation of sodium nonaflate and a coordinatively unsaturated Pd(0) complex II. With the iodide anion weakly bound, this may accelerate coordination and subsequent reaction with the phosphorus nucleophile, 22 and following reductive elimination, allow overall faster access to the coupled product. There are other possible roles of iodide that could result in accelerated reactions. For example, the larger trans effect of the iodide when complexed to a Pd intermediate, in comparison to the other ionic additives, could also lead to an accelerated transformation through faster substitution reactions. 16

■ CONCLUSIONS
In summary, aryl nonaflates, which are isolable intermediates, readily prepared from abundant phenols, were found to be effective substrates for palladium-catalyzed C−P bond-forming reactions. Optimization studies revealed that the addition of NaI resulted in accelerated reactions, allowing the rapid synthesis of a wide range of aryl phosphine oxides. Extension of the process with other P(O)H coupling reagents resulted in the synthesis of further aryl phosphorus compounds, such as an aryl phosphinate and aryl phosphonates. This included the three-step synthesis of pharmaceutically relevant, phosphonophenylalanine 10 from a commercially available tyrosine derivative in 60% overall yield. Preliminary mechanistic studies suggested that the addition of iodide may accelerate the reaction via a coordinatively unsaturated Pd(0) complex or through the trans effect of a Pd−I intermediate. Investigation of further applications of this transformation is currently underway.

■ EXPERIMENTAL SECTION
All reagents and starting materials, including methyl (2S)-2-[(benzyloxycarbonyl)amino]-3-(4-hydroxyphenyl)propanoate (7), were obtained from commercial sources and used as received, unless otherwise stated. Anhydrous dichloromethane was purified using a PureSolv 500 MD solvent purification system. All reactions were performed under an atmosphere of air unless otherwise stated. All reactions performed at elevated temperatures were heated using an oil bath. Dry glassware was oven-dried at 140°C for a minimum of 16 h, cooled to room temperature in vacuo, and then purged with argon. Brine is defined as a saturated aqueous solution of sodium chloride. Merck aluminum-backed plates precoated with silica gel 60 (UV 254 ) were used for thin-layer chromatography and were visualized under UV light (254/365 nm) and then stained with iodine, potassium permanganate, vanillin, or ninhydrin solution. Flash column chromatography was carried out using Merck Geduran Si 60 (40− 63 μm). 1 H and 13 C NMR spectra were recorded on Bruker DPX 400, Bruker AVI 400, and Bruker AVIII 400 ( 1 H 400 MHz; 13 C 101 MHz) spectrometers or a Bruker AVIII 500 ( 1 H 500 MHz; 13 C 126 MHz) spectrometer with chemical shift values reported in ppm relative to tetramethylsilane (δ H 0.00 and δ C 0.0), CDCl 3 (δ H 7.26 and δ C 77.2) or 3-(trimethylsilyl)propionic-2,2,3,3-d 4 acid sodium salt in D 2 O (δ H 0.00 and δ C 0.0). Assignments of 1 H and 13 C NMR signals are based on COSY, DEPT, HSQC, and HMBC experiments. Mass spectra were obtained using a JEOL JMS-700 spectrometer or a Bruker microTOFq high-resolution mass spectrometer. Melting points were determined on a Gallenkamp melting point apparatus and are uncorrected. Infrared spectra were recorded neat on a Shimadzu FTIR-84005 spectrometer. Optical rotations were determined as solutions irradiating with the sodium D line (λ = 598 nm) using an Autopol V polarimeter.