Double Phosphinoboration of CO2 : A Facile Route to Diphospha-Ureas.

The reactions of CO2 with a series of phosphinoboranes, including R2 PBpin (R=Ph, tBu; pin=pinacol), R2 PBMes2 (R=Ph, tBu; Mes=2,4,6-Me3 -C6 H2 ), and R2 PBcat (R=Ph, tBu, Mes; cat=catechol) are described. Although R2 PBpin and R2 PBMes2 afford products of the form R2 PCO2 Bpin (R=Ph 1, tBu 4) and R2 PCO2 BMes2 (R=Ph 2, tBu 3), respectively, R2 PBcat lead to further reaction affording the diphospha-ureas, (R2 P)2 CO (R=Ph 5, tBu 6, Mes 7), together with O(Bcat)2 . Computational studies provide insight into the mechanism, revealing an intermediate derived from double phosphinoboration of CO2 .

The anthropogenic emission of carbon dioxide (CO2) continues to climb, prompting global climate change.1 This issue has spurred widespread interest in various aspects of CO2 chemistry. Even though efforts targeting CO2 capture2 or its conversion to C1 chemical feedstocks3 have garnered much interest in recent years, there is also growing interest in the utility of CO2 as a reagent in synthetic chemistry. Although transition‐metal chemistry has dominated such developments, main‐group reagents have also emerged, providing unique approaches to the reactivity of CO2. A decade ago, Stephan, Erker, and co‐workers reported the use of frustrated Lewis pairs (FLPs) to capture CO2.4 In the intervening years, a variety of B/N,5 B/P,6 Al/P,7 and Si/P5a, 8 systems have been developed to capture or effect stoichiometric or catalytic reduction of CO2. Apart from FLPs, basic main‐group reagents have been exploited in reactions with CO2. For example, the carboxylations of N‐heterocyclic carbenes9 as well as silyl and germyl amides10 have been reported whereas Dielmann and co‐workers have described CO2 capture by highly basic phosphines incorporating imidazolin‐2‐ylidenamino substituents.11

In recent studies, ureas have been prepared from CO2 and aliphatic amines exploiting transition‐metal‐based catalysts12 or cesium‐based catalysts.13 In 2018, we reported indium‐based Lewis acid catalysts for the conversion of silylamines and CO2 to ureas with liberation of silylethers.14 Interestingly, even though the direct reaction of CO2 with amines12c, 15 generally requires high pressure (14–140 atm) and high temperature (130–170 °C), we have recently shown that judicious choice of solvent promotes the reaction of CO2 and silylamines to give silylcarbamates and ureas.16

In contrast to ureas, the phosphorus analogues, diphospha‐ureas, (R2P)2CO are rare. The first example of a diphospha‐urea (R=Ph) was prepared in low yield by the reaction of Ph2PSiMe3 with phosgene at −110 °C.17 Subsequently, the reduction of dimeric phosphaketenes, RPCO was reported to give an alternative route to a diphospha‐urea.18 Although transition‐metal‐supported diphospha‐ureas have also been reported,19 we have recently described a synthetic route exploiting nucleophilic attack of the phosphaketene Ph3GePCO by phosphides and subsequent alkylation.20 In targeting new strategies to CO2 reduction to diphospha‐ureas, we note that we had some time ago reported the reactions of the phosphinoboranes R2PB(C6F5)2 with H2 through a FLP‐type reaction.21 Subsequently, Westcott and co‐workers reported related addition reactions in which Ph2PBpin reacts with unsaturated multiple bonds prompting phosphinoboration of ketones, imines, aldehydes,22 N‐heterocycles,23 and a series of heteroallenes.24 In the present study, we describe the reactions of CO2 with a series of phosphinoboranes, demonstrating that suitable modification of the B/P reagent prompts CO2 reduction to afford a unique and facile route to diphospha‐ureas. Computational studies provide insight into the mechanism, revealing an intermediate derived from double phosphinoboration of CO2.

We previously described the stoichiometric reaction of Ph2PBpin with CO2.24 Based on the spectroscopy alone, the compound was formulated as Ph2PCO2Bpin 1 (Scheme 1). Repetition of the reaction afforded X‐ray quality crystals and the subsequent diffraction study confirmed the formulation (Figure 1). In this product of phosphinoboration of CO2, the newly formed P−C bond was found to be 1.850(2) Å whereas the O−B bond was found to be 1.383(3) Å. The slightly distorted trigonal‐planar geometry at the central carbon atom results in a P‐C‐O angle of 111.12(16)°, whereas the C−O distances of 1.197(3) Å and 1.360(3) Å are consistent with formal double and single bonds, respectively.

Scheme 1

Synthetic pathways to 1–7.

Figure 1

POV‐ray depictions of the molecular structures of (a) 1; (b) 2; hydrogen atoms are omitted for clarity. C: black, O: red, P: orange, B: yellow‐green.

Probing substituent effects, the phosphinoborane Mes2PB(C6F5)2 21 was prepared, however it showed no reaction with 4 atm of CO2. In contrast, Ph2PBMes2 25 reacted with CO2 slowly over 3 days to give upon workup 2 as a white solid in 86 % yield. The 31P{1H} NMR spectrum of 2 showed a single resonance at δ=−1.5 ppm with coupling to the carbon of CO2 of J=19 Hz in the 13C‐labelled product. The corresponding 13C{1H} NMR spectrum revealed a resonance at 178.1 ppm with the same coupling constant in addition to resonances attributable to the remaining carbon atoms. The 11B{1H} NMR spectroscopic data showed a single resonance at 53.1 ppm. These data were consistent with the formulation of 2 as Ph2PCO2BMes2. This assignment was confirmed crystallographically (Figure 1). The geometry of 2 is similar to that seen in 1 with the newly formed P−C and O−B distances being 1.8590(14) and 1.4099(18) Å, respectively.

In a similar fashion, the species tBu2PBMes2 and tBu2PBpin were prepared (see the Supporting Information) and reacted with CO2 to give the analogous phosphinoboration products tBu2PCO2BMes2 (3) and tBu2PCO2Bpin (4), which were isolated in 82 and 89 % yield, respectively (Scheme 1). Interestingly, the 31P{1H} NMR spectrum of 3 showed a resonance at 47.3 ppm with a coupling to the carbon atom of 40 Hz, whereas the spectrum of 4 showed a resonance at 48.2 ppm with a coupling constant to the carbon atom of 35 Hz in the 13C‐labelled products, consistent with significant increases in the P−C bond strengths.

The related phosphinoboranes R2PBcat (R=Ph,22 tBu, Mes) were also prepared (see the Supporting Information) and their reactions with CO2 were examined. These reactions gave products quite distinct from 1–4. The reaction of Ph2PBcat with CO2 proceeds over several days at ambient temperature to give a yellow solution. Monitoring of the solution in situ by NMR spectroscopy revealed the appearance of a 11B{1H} NMR signal at 24.4 ppm. In addition, performing the reaction with 13CO2, a doublet 31P{1H} resonance at 30.5 ppm showed P−C coupling of 70 Hz, whereas the corresponding triplet 13C{1H} NMR resonance appeared at 239.0 ppm with the same coupling constant. These data are consistent with the generation of the diphospha‐urea, (Ph2P)2CO (5) and the corresponding by‐product O(Bcat)2. Efforts to isolate 5 were plagued by its thermal and photochemical instability. It is noteworthy that this species has been previously prepared by Becher and Langer using phosgene and Ph2PSiMe3. These authors also noted its instability.17

The corresponding reaction of tBu2PBcat with CO2 also proceeds over several days at ambient temperature to give a yellow solution. The product solution displayed a 11B{1H} NMR signal at 22.5 ppm whereas the 31P{1H} NMR resonance appears at 75.5 ppm. Again, use of 13CO2 reveals P−C coupling of 83 Hz, whereas the corresponding triplet 13C{1H} NMR resonance appears at 245.5 ppm. These data are consistent with the generation of the diphospha‐urea, (tBu2P)2CO (6). Although compound 6 crystallizes from pentane, X‐ray data collection was complicated by the photochemical instability of 6, despite collection at −123 °C in the dark. Nonetheless, the preliminary X‐ray data confirmed the connectivity (see the Supporting Information), although prudence precludes a discussion of the metric data.

The corresponding reaction of Mes2PBcat under CO2 proceeds similarly. After 6 days at ambient temperature, workup afforded a yellow solid 7 in 34 % yield of isolated material. This product exhibited a 31P{1H} NMR resonance at 14.4 ppm with P−C coupling of 73 Hz and the corresponding triplet 13C{1H} NMR signal at 227.5 ppm in the 13C‐labelled product. This species was crystallized from pentane and the X‐ray data were collected at −123 °C in the dark. The structural data affirmed the formulation of 7 as (Mes2P)2CO (Figure 2), revealing P−C(O) bond lengths of 1.914(2) and 1.905(2) Å and a C−O bond length of 1.192(2) Å. The resulting P‐C‐P angle was 102.29(10)° and the P‐C‐O angles were 130.03(16)° and 127.65(16)°. These parameters are comparable to those recently reported for tBu2PC(O)P(GePh3)Me (P−C: 1.887(3), 1.848(3); C−O: 1.219(3) Å) and (Ph3Ge)2PC(O)P(tBu)2 (P−C: 1.887(3), 1.892(3); C−O: 1.212(3) Å).

Figure 2

POV‐ray depiction of the molecular structure of 7; hydrogen atoms are omitted for clarity. C: black, O: red, P: orange.

The mechanism of the reactions of Ph2PBpin, Ph2PBcat, and tBu2PBcat with CO2 were probed by state‐of‐the‐art DFT calculations at the PW6B95‐D3/def2‐QZVP+COSMO‐RS//TPSS‐D3/def2TZVP+COSMO level of theory26 in CH2Cl2 solution. As shown in Scheme 2, the direct addition of the P−B single bond of Ph2PBpin to a C=O bond of CO2 is −2.3 kcal mol−1 exergonic over a moderate barrier of 23.4 kcal mol−1 via the transition structure (TS) of TS1, affording the product of Ph2PC(O)OBpin 1′ with a cis‐arrangement of the boryl and phosphanyl substituents, which is slightly (0.3 kcal mol−1) more favorable in solution than the crystallographically observed trans‐conformer 1. Further addition of another Ph2PBpin equivalent to the C=O bond of 1′ is −3.3 kcal mol−1 exergonic but over a sizable barrier of 28.2 kcal mol−1 via TS2 to give the complex (Ph2P)2C(OBpin)2 (8) that may further eliminate O(Bpin)2 over a lower barrier of 24.1 kcal mol−1 via TS3 to form the diphospha‐urea (Ph2P)2CO (5). Direct access to diphospha‐urea 5 through P−B/C−O bond metathesis between Ph2PBpin and 1′ is prevented by an even higher barrier of 31.6 kcal mol−1 via TS4. These computations are consistent with the observed slow formation of 1 instead of the more desirable 5 at ambient temperature.

Scheme 2

DFT‐computed free‐energy paths in CH2Cl2 solution (in kcal mol−1, at 298 K and 1 m reference concentration) for the reactions of Ph2PBpin (Ph2PBcat and tBu2PBcat values in parentheses and in brackets, respectively, for comparison) with CO2.

We note that prolonged exposure of 1 to excess Ph2PBpin in pentane at −35 °C afforded crystals of (Ph2P)2C(OBpin)2 8, albeit in low yield. Unfortunately, efforts to isolate 8 or measure its HRMS and NMR spectroscopic data proved fruitless due to its instability in the solid state at ambient temperature and in solution at −35 °C. However, an X‐ray diffraction study of 8 confirmed the formulation (Figure 3) and the pseudo‐tetrahedral nature of the central carbon atom. The P−C bond lengths were 1.9116(17) and 1.9388(18) Å, whereas the newly formed B−O bond lengths were 1.370(2) and 1.364(2) Å.

Figure 3

POV‐ray depiction of the molecular structure of 8; hydrogen atoms are omitted for clarity. C: black, O: red, P: orange, B: yellow‐green.

The isolation of 8, although supporting the computed mechanism, is surprising given its lower dissociation barrier to the conversion to 5. Nonetheless, this is consistent with the thermal and photochemical instability of 5 as well as the observation that compound 8 is stable in the solid state at −35 °C but immediately degrades in solution, as evidenced by spectroscopic data.

As shown in Scheme 2, very similar reaction paths are also found for the reactions of Ph2PBcat and tBu2PBcat with CO2, despite evidently changed energetics. The introduction of the more Lewis acidic boryl Bcat leads to nearly the same barrier of 23.3 kcal mol−1 for the first P−B/C=O addition (dominating electrophilic attack at P). On one hand, lower barriers of 22.1 and 19.6 kcal mol−1 for the second P−B/C=O addition and the subsequent O(Bcat)2 elimination steps were computed. This is in accord with the observed facile formation of diphospha‐urea 5 with the rate‐limiting step being the first P−B/C=O addition over a barrier of 23.3 kcal mol−1. On the other hand, when the more bulky and Lewis basic phosphanyl tBu2P is further introduced, the first P−B/C=O addition becomes almost neutral in free energy over an evidently lower barrier of 17.1 kcal mol−1, whereas further P−B/C=O addition followed by O(Bcat)2 elimination is now prevented by a high overall barrier of 34.9 kcal mol−1 via the high‐lying 8‐like intermediate. In this case, TS4‐like P−B/C−O bond metathesis over a barrier of 31.5 kcal mol−1 after the first P−B/C=O addition is more likely to afford the slower formation of diphospha‐urea 6, as was observed experimentally.

In summary, we have described the reactions of CO2 with a series of phosphinoboranes, illustrating a sequence of B/P additions to the carbonyl fragments. This provides access to species of the form R2PCO2BR′2 initially whereas Bcat‐containing B/P reagents afford a unique and facile route to diphospha‐ureas. The computed mechanism proceeds through a double‐addition product that was supported by experimental data. With this facile synthetic route in hand, we are now exploring the chemistry of diphospha‐ureas as well as other applications of phosphinoboration reactions.27

Conflict of interest

The authors declare no conflict of interest.

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supplementary

Click here for additional data file.