Non-metal-templated approaches to bis(borane) derivatives of macrocyclic dibridgehead diphosphines via alkene metathesis

Two routes to the title compounds are evaluated. First, a ca. 0.01 M CH2Cl2 solution of H3B·P((CH2)6CH=CH2)3 (1·BH3) is treated with 5 mol % of Grubbs' first generation catalyst (0 °C to reflux), followed by H2 (5 bar) and Wilkinson's catalyst (55 °C). Column chromatography affords H3B·P(n-C8H17)3 (1%), H3B·P((CH2)13CH2)(n-C8H17) (8%; see text for tie bars that indicate additional phosphorus–carbon linkages, which are coded in the abstract with italics), H3B·P((CH2)13CH2)((CH2)14)P((CH2)13CH2)·BH3 (6·2BH3, 10%), in,out-H3B·P((CH2)14)3P·BH3 (in,out-2·2BH3, 4%) and the stereoisomer (in,in/out,out)-2·2BH3 (2%). Four of these structures are verified by independent syntheses. Second, 1,14-tetradecanedioic acid is converted (reduction, bromination, Arbuzov reaction, LiAlH4) to H2P((CH2)14)PH2 (10; 76% overall yield). The reaction with H3B·SMe2 gives 10·2BH3, which is treated with n-BuLi (4.4 equiv) and Br(CH2)6CH=CH2 (4.0 equiv) to afford the tetraalkenyl precursor (H2C=CH(CH2)6)2(H3B)P((CH2)14)P(BH3)((CH2)6CH=CH2)2 (11·2BH3; 18%). Alternative approaches to 11·2BH3 (e.g., via 11) were unsuccessful. An analogous metathesis/hydrogenation/chromatography sequence with 11·2BH3 (0.0010 M in CH2Cl2) gives 6·2BH3 (5%), in,out-2·2BH3 (6%), and (in,in/out,out)-2·2BH3 (7%). Despite the doubled yield of 2·2BH3, the longer synthesis of 11·2BH3 vs 1·BH3 renders the two routes a toss-up; neither compares favorably with precious metal templated syntheses.

We subsequently developed an interest in the free dibridgehead diphosphine ligands P((CH 2 ) n ) 3 P (n = 14, 2; 18, 3), prompted in part by the unexpected discovery of the facile demetalations shown in Scheme 1 [5,6,10,22]. Such compounds were previously known only for much smaller ring sizes (n < 4) [23]. These reactions require excesses of certain nucleophiles, and the mechanisms remain under study. The yields are quite good, but the routes are stoichiometric in precious metals. Although the metals can be recovered as species such as K 2 Pt(CN) 4 or RhCl(PMe 3 ) 3 , we have nonetheless sought to develop more economical protocols.
The analogous Fe(CO) 3 adducts are easily prepared [1][2][3][4], but in efforts to date it has not been possible to efficiently remove the dibridgehead diphosphine ligands from the low cost iron fragment. Oxidations that lead to the corresponding dibridgehead diphosphine dioxides (O=)P((CH 2 ) n ) 3 P(=O) have exhibited promise, but purification has been problematic [24]. Indeed, phosphine oxides are everyday precursors to phosphines, so we have considered various non-metal-templated routes to 2·2(=O), 3·2(=O), and related species. However, as described in the discussion section, the yields have not been competitive [25].
Another preliminary point concerns the ability of macrocyclic dibridgehead diphosphorus compounds to exhibit in/out isomerism [26]. As shown in Scheme 1, there are three limiting configurations for 2 and 3: in,in, out,out, and in,out (identical to out,in). The first two, as well as the degenerate in,out pair, can rapidly interconvert by a process termed homeomorphic isomerization [26,27], which is akin to turning the molecules inside out. Readers are referred to earlier publications in this series for additional details [22,25,[28][29][30]. Interconversions between the in,in/out,out and in,out/out,in manifolds require phosphorus inversion and temperatures considerably in excess of 100 °C.
In this paper, we describe two non-metal-templated approaches to 2 that are based upon metatheses of phosphine boranes of alkene containing phosphines. The first involves the monophosphorus precursor H 3 B·P((CH 2 ) 6 CH=CH 2 ) 3 (1·BH 3 ) [31], and the second a diphosphorus precursor in which one of the methylene chains linking the two phosphorus atoms has already been installed. The advantages and limitations of each are analyzed in detail. Some of the results (Scheme 2) have appeared in the supporting information of a preliminary communication [28], and others in a dissertation [32].

Monophosphorus precursors
As reported earlier [31], the alkene containing phosphine P((CH 2 ) 6 CH=CH 2 ) 3 (1) can be prepared in 87% yield from the reaction of PCl 3 and MgBr(CH 2 ) 6 CH=CH 2 . Following the addition of H 3 B·SMe 2 , the phosphine borane 1·BH 3 can be isolated in 65-85% yields [31], as shown in Scheme 2. It is critical to avoid an excess of H 3 B·SMe 2 , as this brings the C=C units into play. In fact, when substoichiometric amounts of H 3 B·SMe 2 are added to THF solutions of purified 1·BH 3 , gels immediately form.
A ca. 0.01 M CH 2 Cl 2 solution of 1·BH 3 and a ca. 0.002 M CH 2 Cl 2 solution of Grubbs' first generation catalyst (3 mol %) were combined at 0 °C. The mixture was warmed to room temperature, and a second charge of Grubbs' catalyst added (2 mol %). The sample was refluxed, and then filtered through silica gel. The filtrate was concentrated and treated with H 2 (5 bar) and Wilkinson's catalyst (55 °C). The mixture was taken to dryness and the residue tediously chromatographed on a silica gel column. Numerous fractions were collected and analyzed by TLC. The mass recovery from the column was 33% of theory (for complete metathesis).
More than ten mobile products could be discerned, but only five could be isolated in pure form and ultimately identified. These are described in order of elution. Each was analyzed by NMR ( 1 H, 31 P{ 1 H}, 13 C{ 1 H}; always CDCl 3 ) and IR spectroscopy,  mass spectrometry, and microanalysis, as summarized in the experimental section. The 13 C{ 1 H} NMR spectra proved to be most diagnostic of structure, and were analyzed in detail. The 31 P{ 1 H} NMR spectra were all very similar (broad apparent doublets due to phosphorus boron coupling).
First, traces of a colorless oil were obtained. The 1 H NMR spectrum showed a characteristic triplet at 0.83 ppm consistent with a terminal methyl group. The 13 C{ 1 H} NMR spectrum exhibited eight signals, two of which were phosphorus coupled doublets. One of the singlets (14.0 ppm) was typical of a terminal methyl group. Based upon these data, and the integration of the 1 H NMR spectrum, the oil was assigned as the hydrogenated phosphine borane H 3 B·P(n-C 8 H 17 ) 3 (4·BH 3 ), a known compound [33]. The yield was only 1%.
Next, another colorless oil eluted. The 1 H NMR and 13 C{ 1 H} NMR spectra again exhibited signals characteristic of a methyl group (0.86 ppm, t; 14.0 ppm, s). Integration of the 1 H NMR spectrum established a 14:1 area ratio for the methylene (1.62-1.19 ppm) and methyl signals. The 13 C{ 1 H} NMR spectrum featured one set of seven signals and another set of eight with an intensity ratio of approximately 2:1. The less intense set resembled the signals arising from the n-octyl groups in 4·BH 3 . The more intense set was very similar to the signals arising from the cyclic substructures of 6·2BH 3 (described below) and a phosphine borane reported earlier [34]. The mass spectrum exhibited an intense ion at m/z 340 (5 + , 93%), and no ions of higher mass. Hence, the oil was assigned as the monocyclic intramolecular metathesis product (5·BH 3 ; see Scheme 2). The yield was 8%.
The third product was also a colorless oil. The 13 C{ 1 H} NMR spectrum exhibited seven signals, three of which were phosphorus coupled doublets (second spectrum from top, Figure 1). Analogous coupling patterns are found with the free dibridgehead diphosphines 2 and 3 in Scheme 1. No NMR signals diagnostic of methyl groups were present, and further analysis is presented along with that for an isomer below.
A white powder was obtained next. The 13 C{ 1 H} NMR spectrum exhibited fourteen signals, half of which were approximately twice as intense as the others. Two signals of each set exhibited phosphorus coupling. The overall pattern was quite similar to those shown by metal complexes with cis or trans coordinating diphosphine ligands of the formula (6) [6,7,12,13,35]. This suggested the diphosphine diborane structure 6·2BH 3 (see Scheme 2), which is derived from one metathesis involving alkenyl moieties on different phosphorus atoms, and two metatheses of alkenyl moieties on identical phosphorus atoms. The yield was 10%. The structure has been confirmed by an independent synthesis (detachment of the diphosphine from a platinum complex followed by borane addition) and a crystal structure [6].
Finally, another white powder was obtained. As with the previous oil isolated above, the 13 C{ 1 H} NMR spectrum exhib-ited seven signals, three of which were phosphorus coupled doublets (third spectrum from top, Figure 1). Both spectra were consistent with dibridgehead diphosphine diboranes H 3 B·P((CH 2 ) 14 ) 3 P·BH 3 (2·2BH 3 ) derived from threefold intermolecular metatheses of 1·BH 3 . Based upon independent syntheses from the dibridgehead diphosphines 2 obtained in Scheme 1 [6], they were assigned as in,out-2·2BH 3 (4%) and the stereoisomer (in,in/out,out)-2·2BH 3 (2%), as shown in Scheme 2. The depiction of the latter as an out,out (vs in,in) isomer in Scheme 2 is arbitrary, but represents the form found in a confirming crystal structure [6].
Parallel reactions were conducted with Grubbs' second generation catalyst and the nitro-Grela catalyst [36]. However, the combined yields of 2 diminished.

Diphosphorus precursors
Since the yields of the cage like diphosphine diboranes 2·2BH 3 in Scheme 2 were -as expected -very low, alternative strategies were considered. The poor mass balance was attributed, at least in part, to the formation of oligomeric products that were retained on the column. Improvements might be expected from precursors in which one of the methylene chains tethering the two phosphorus atoms was pre-formed. Thus, we set out to prepare a tetraalkenyl metathesis precursor as shown in Scheme 3.
Thus, despite the low yield of the final step in Scheme 3, reasonable quantities of the diphosphine diborane 11·2BH 3 could be stockpiled. As shown in Scheme 5, 11·2BH 3 was subjected to a metathesis/hydrogenation/column chromatography sequence similar to that for 1·BH 3 in Scheme 2. However, a tenfold higher dilution was used in the metathesis step (0.0010 M as compared to 0.010 M). Figure 1 shows a 13 C{ 1 H} NMR spectrum of the crude product after hydrogenation stacked above spectra of the three products that could be isolated after the rather tedious column chromatography: the dibridgehead diphosphine diborane in,out-2·2BH 3 , its constitutional isomer 6·2BH 3 , and its stereoisomer (in,in/ out,out)-2·2BH 3 . It can be inferred from the top spectrum that the three products were the major components and moreover present in approximately equal amounts. However, the isolated yields were affected by the challenging separation. In particular, in,out-2·2BH 3 and 6·2BH 3 eluted very closely, rendering some mixed fractions unavoidable and lowering the amounts of pure products.
Compared to the metathesis/hydrogenation sequence for 1·BH 3 (Scheme 2) the yields of in,out-2·2BH 3 and (in,in/out,out)-2·2BH 3 (Scheme 5) are higher but still poor. Taking into account the overall yields (three steps from PCl 3 and BrMg(CH 2 ) 6 CH=CH 2 in the first synthesis vs seven steps from 1,14-tetradecanedioic acid in the second), the latter route does not offer any advantage, even if one were to improve the conversion of 10·2BH 3 to 11·2BH 3 . Scheme 6: Schematic comparison of the key alkene metathesis steps in Scheme 2 and Scheme 5.

Discussion
As contrasted in Scheme 6, Scheme 2 and Scheme 5 present two conceptually related routes to the isomeric title compound 2·2BH 3 . In the first, two trialkenylphosphine boranes (1·BH 3 = I) must undergo metathesis. The first productive step is intermolecular, giving a diphosphorus compound with a P(CH 2 ) 6 CH=CH(CH 2 ) 6 P tether II that is positioned for subsequent intramolecular ring closing steps. Those involving alkenyl groups from different phosphorus atoms are productive (leading to 2·2BH 3 via hydrogenation of IIIa), and those involving groups from the same phosphorus atoms are non-productive (leading to 6·2BH 3 via hydrogenation of IVa). In the second, the starting material has a preformed P(CH 2 ) 14 P tether (11·2BH 3 = V), and the four alkenyl groups have reactivity options (→ IIIb or IVb) analogous to those of intermediate II with the P(CH 2 ) 6 CH=CH(CH 2 ) 6 P tether. Importantly, all of these steps are presumed to be largely under kinetic control, consistent with experience with the types of metatheses in Scheme 1 [1][2][3][4][5][6][7][8][9][10][11][12][13]34].
Although the second route intuitively seems more favorable, after the initial intermolecular metathesis of 1·BH 3 (I), both require an equivalent series of steps to reach (after hydrogenation) 2·2BH 3 . One reason 1·BH 3 is an inferior substrate is that following the initial generation of a P(CH 2 ) 6 CH=Ru species, two P(CH 2 ) 6 CH=CH 2 moieties remain available for nonproductive intramolecular ring closing metathesis (giving VI). In contrast, with the analogous intermediate derived from 11·2BH 3 (V), there is only one P(CH 2 ) 6 CH=CH 2 moiety that can give non-productive chemistry. It is also worth noting that high dilution provides less of an advantage in Scheme 2, as one wants to favor intermolecular over intramolecular metatheses in the first step. In Scheme 5, one wants to avoid intermolecular metatheses at all stages.
At present, we have no rationale for the in,out vs (in,in/out,out) isomer ratios for 2·2BH 3 . However, it is easy to map the sequence leading to each, as shown in Scheme 7. When there is only one tether between the two phosphorus atoms, the phosphorus-boron bonds can be arrayed in an anti fashion, as depicted in VII. When subsequent metatheses join alkenyl groups in the syn positions on each phosphorus atom (front to front and rear to rear), (in,in/out,out)-2·2BH 3 must result (as drawn in Scheme 7, the out,out isomer would be the kinetic Scheme 8: Another non-metal-templated approach to dibridgehead diphosphorus compounds. product). When the first metathesis does not join the syn positions, as in VIII (front to rear), one phosphorus-boron bond must subsequently be rotated by 180° to create a syn orientation for the second metathesis.
Of course, if the first metathesis step does not require a syn relationship (per VIII), the same possibility can be entertained for the second (see IX). This would lead to an isomeric bicyclic compound with "crossed chains". We have sought to access such species by conducting metatheses of substrates of the types in Scheme 1 that give thirty-three membered macrocycles (n = 30) [7]. However, none have so far been detected. Other types of crossed chain in/out isomer systems have in fact been realized [25,30].
However, as shown in Scheme 8, it has proved possible to synthesize the diphosphine dioxides 14, in which the two phosphorus atoms are tethered by a methylene chain, in two steps in 66-68% overall yields from diethyl phosphonate ((O=)PH(OEt) 2 ), Grignard reagents BrMg(CH 2 ) m CH=CH 2 , base (NaH), and appropriate α,ω-dibromides Br(CH 2 ) n Br [25]. Following metathesis and hydrogenation, these afford dibridgehead diphosphine oxides 15 and 16 in 14-19% yields. This is slightly better than the combined yield of in,outand (in,in/ out,out)-2·2BH 3 in Scheme 5, although the data are not strictly comparable as the ring sizes differ. It has not yet proved possible to efficiently separate the in/out isomers of 15 and 16. However, byproducts derived from metatheses of alkenyl groups on the same phosphorus atom -such as 17 (comparable to 6·2BH 3 ) -appear to form in much smaller amounts.
To our knowledge, only one macrocyclic dibridgehead diphosphine diborane has been previously reported, (in,in/out,out)-18·2BH 3 in Scheme 9 [50,51]. This features triarylphosphorus bridgeheads and p-phenylene containing tethers that are long enough to allow rapid homeomorphic isomerization. The precursor 18·2(=O) was prepared by a threefold Williamson ether synthesis in surprisingly high yields (61% in,in/out,out and in,out combined) [50,51], likely aided by the geminal dialkyl effect associated with the quaternary centers [52]. Finally, it should be noted that a number of alkene containing phosphine boranes have been employed in metathesis reactions [53,54]. In particular, the tetraalkenyl diphosphine diborane 19·2BH 3 in Scheme 10 represents a downsized version of 11·2BH 3 . A species analogous to 6·2BH 3 , 20·2BH 3 , is obtained in much higher yield than any of the products in Scheme 5 [53]. Hence, selectivities can strongly depend upon the lengths of the methylene segments in the precursor.

Conclusion
In conclusion, this work constitutes a further installment in the evolution of synthetic strategies for dibridgehead diphosphorus compounds that employ alkene metathesis. The new approaches (Scheme 2; Scheme 3 and Scheme 5) lack metal templates, which differentiates them from the routes presented in Scheme 1. However, neither is competitive with Scheme 1, despite eliminating the requirement for stoichiometric amounts of precious metals. Furthermore, preassembling a diphosphine diborane substrate per Scheme 3 and Scheme 5 is not competitive with the "shotgun" approach in Scheme 2, and both routes require comparably demanding preparative column chromatography. Hence, the most promising direction for future research would seem to be templated syntheses via non-precious metals [55]. This remains an area of ongoing investigation in our laboratory and further results will be reported in due course.

Supporting Information
Supporting Information File 1 Additional experimental data.