SynthesisThe commercially available diphenylphosphinic acid, Ph2P(O)OH, was seen as a pnictogen analogue of benzoic acid that had previously found success in the synthesis of iodine(I) complexes. This precursor was seen as an ideal analogue, especially given that the silver(I) derivative, (Ph2P(O)OAg)n (silver(I) diphenylphosphinate), necessary for subsequent iodine(I) complex synthesis had also already been reported45. Nonetheless, a higher-yielding procedure for the synthesis of (Ph2P(O)OAg)n (1) was developed, which gave the species in an 82% yield (Supplementary Note 2).Reaction of (Ph2P(O)OAg)n with a Lewis base (L), followed by I2 (in a 1:1:1 stoichiometry), gave iodine(I) pnictogenates of the form Ph2P(O)O─I─L (Fig. 2; Supplementary Note 3), which were achieved when L = 4-methylpyridine (4-Mepy; 1b), 4-ethylpyridine (4-Etpy; 1c), 4-dimethylaminopyridine (DMAP; 1d), 4-pyrrolidinopyridine (4-pyrpy; 1e), 4-piperidinopyridine (4-pippy; 1f), 4-morpholinopyridine (4-morpy; 1g). It should be noted that the IUPAC names would be: pyridine iodine(1+) diphenyldioxidophosphate(1−) (1a), and so forth, but will be shortened to iodine(I) phosphinates (for R2P(O)O derivatives), or more generally, to iodine(I) pnictogenates.Fig. 2: The general synthetic procedure for the synthesis of iodine(I) pnictogenate complexes.The numbering convention used for the iodine(I) pnictogenates synthesised herein.Interestingly, these phosphorus-based analogues differed in their apparent stabilities in solution, ranging from unstable (1a) to relatively stable in dry solvents (1b, 1c, 1e, 1g), and even demonstrating apparent water-tolerance in non-treated solvents (1d, 1f), which are important traits toward their appeal as prospective iodination reagents.Solution studiesThe 1H NMR studies (Supplementary Figs. 1, 4, 7, 10, 13, 16) revealed predominantly only minor coordination shifts upon coordination of the free ligands (b–g) in comparison to their respective iodine(I) derivatives (1b–1g), with differences in the aromatic substituted pyridines ranging from 0.00 to 0.42 ppm, though most fell under 0.20 ppm, with only 4-morpy demonstrating notable coordination shifts of 0.26 and 0.42 ppm.Prior observations of the carbonyl carbon atom (via 13C NMR studies) or of the fluorinated substituents of the carbonyl substituents (via 19F NMR studies) both found negligible coordination shifts upon iodine(I) formation46,47. However, the inclusion of a phosphorus atom, with its 100% abundance 31P nuclei and close proximity to the iodine atom, was envisioned to provide a valuable new NMR handle for these iodine(I) phosphinates. This proved true with notable, though modest, coordination shifts of 7.1–10.0 ppm in relation to (Ph2P(O)OAg)n (18.4 ppm), though it should be noted that both these species were by necessity recorded in (CD3)2SO, preventing a direct comparison to the iodine(I) phosphinate complexes that were collected in CD2Cl2 (Supplementary Figs. 3, 6, 9, 12, 15, 18). Nevertheless, differences between the Lewis base incorporated in the iodine(I) phosphinates (all recorded in CD2Cl2) were also reflected in the 31P{1H} NMR chemical shifts (Supplementary Note 5), with values ranging from 25.5 ppm (1e) to 28.4 ppm (1b/1c).Nevertheless, the solution studies were primarily focussed on the 15N NMR chemical shifts (determined from 1H–15N HMBC experiments; Supplementary Figs. 2, 5, 8, 11, 14, 17), as these have previously been demonstrated to show characteristic coordination shifts (ΔδN) of >100 ppm upon of the formation of iodine(I) complexes for both the [N─I─N]+ and C(O)O─I─N motifs39,47,48. The coordination shift is defined as the difference between the chemical shift of the coordinating (pyridinic) nitrogen atom in the free ligand (δ[15NLigand]) and the respective iodine(I) complex (δ[15NComplex]) incorporating that same ligand and is usually reported as a positive value.The free ligands (also in CD2Cl2) 4-Mepy (b, −75.1 ppm)46, 4-Etpy (c, −75.6 ppm)49, DMAP (d, −105.3 ppm)13, 4-pyrpy (e, −110.0 ppm)50, 4-pippy (f, −104.2 ppm)39, and 4-morpy (g, −99.1 ppm)47, were all found to demonstrate significant coordination shifts upon formation of the iodine(I) phosphinates (Supplementary Note 5), with ΔδN values of 101.1 (1b; δN = −176.2), 99.6 (1c; δN = −175.2), 117.6 (1d; δN = −222.9), 118.1 (1e; δN = −228.1), 119.2 (1f; δN = −223.4), 116.3 (1g; δN = −215.4) ppm. These large ΔδN values are consistent with prior observations for analogous iodine(I) carboxylate complexes39,46,50, though the iodine(I) pnictogenates demonstrated slightly larger values by comparison, e.g., PhC(O)O─I─(DMAP) (ΔδN = 100.9) versus 1d (ΔδN = 117.6) and PhC(O)O─I─(4-pippy) (ΔδN = 104.9) versus 1f (ΔδN = 119.2).Unfortunately, complex 1a could not be isolated in a pure form, as observed by the additional peaks in the 1H and 31P NMR spectra. Despite the 15N NMR chemical shift of −163.7 ppm observed in the reaction mixture giving a characteristic ΔδN value of 96.0 ppm (cf. py (a) = −67.7 ppm)13, which is indicative of successful iodine(I) formation, the possibility of similar chemical shifts being observed for the protonated free ligands means these observations are ambiguous15.Solid-state studiesThe solid-state structures for 1b–1g were definitively established via single-crystal X-ray diffraction (SCXRD; Fig. 3), with the solid-state structure of 1a understandably remaining elusive due to its high reactivity as observed in solution by NMR studies. In iodine(I) carboxylates, the relationship between the O─I and I─N bond lengths is such that strong Lewis bases generally cause shorter I─N (and concomitantly longer O─I) distances, and vice versa for weaker Lewis bases39, with retention of the overall intramolecular O···N distance that remains fairly constant (4.45 Å)47,51. These iodine(I) phosphinates also demonstrated fairly consistent intramolecular O···N distances for the halogen-bonded iodine atoms, despite the variation of their stabilising Lewis bases incorporated, with values ranging from 4.411(4) (1b) to 4.457(4) Å (1e/1f) (average = 4.43 Å). The I─N bond lengths did demonstrate the general trend of being shorter for the stronger Lewis bases, e.g., the shortest and longest I─N distances of 2.182(2) and 2.249(3) Å were for the DMAP (1d) and 4-Etpy (1c) derivatives, respectively. These inversely coincided with the range of O─I bond lengths of 1d (2.251(2) Å) to 1e (2.198(3) Å; N.B. the O─I distance for 1c was 2.204(2) Å, which is crystallography indistinguishable to 1e to a 3σ tolerance). The O─I─N angles were all effectively linear (180°), with an overall average of 176.3°, and the largest deviation from linearity observed of 5.1° (1f) being negligible. The formally inequivalent P=O and P─O bonds were always visibly differentiated in all the iodine(I) phosphinate derivatives, with narrow ranges of 1.481(3)–1.493(2) and 1.532(4)–1.546(2) Å, respectively, indicating distinct double- and single-bond character in their bonding.Fig. 3: The solid-state structures of 1b–1g.Structures annotated with their O─I─N bond angles (black), as well as their O─I (red) and I─N (blue) bond lengths in Ã… (thermal ellipsoids at 50% probability; minor disordered positions and solvates omitted for clarity).In comparison to the closest matching iodine(I) carboxylate complex (Supplementary Note 7), PhC(O)O─I─L, for which only three solid-state structures of different Lewis bases (L) are known (L = py, DMAP, 4-morpy), the P(O)O─I─L complexes demonstrated generally shorter I─N (and longer O─I) bond lengths. This can be seen in the direct comparison of the O─I/I─N bond lengths of PhC(O)O─I─(DMAP) (2.210(3)/2.241(3) Å) with 1d (2.251(2)/2.182(2) Å)39, and PhC(O)O─I─(4-morpy) (2.230(4)/2.232(4) Å) with 1g (2.247(4)/2.206(4) Å)39, as well as the approximate comparison of PhC(O)O─I─(py) (O─I/I─N = 2.169(5)/2.292(6) and 2.159(5)/2.299(7) Å) with 1b (2.236(2)/2.222(3) Å) and 1c (2.204(2)/2.249(3) Å) for which no direct analogue is available for comparison39,40; in all these comparisons, the Ph2P(O)O─I derivatives are shorter beyond a 3σ tolerance when compared to the PhC(O)O─I derivatives. When compared to the only other known non-carbonyl O─I─L derivatives, that is, the sulfonate iodine(I) derivatives of the form RS(O)2O─I─L44, 1b–1g possess longer I─N (and shorter O─I) bond lengths. This is illustrated by the direct comparison of the O─I/I─N bond lengths of MeS(O)2O─I─(4-Mepy) (2.347(4)/2.154(4) Å) and (tolyl)S(O)2O─I─(4-Mepy) (2.358(5)/2.142(4) Å) with 1b (2.236(2)/2.222(3) Å), and of MeS(O)2O─I─(DMAP) (2.331(2)/2.140(3) Å) and (tolyl)S(O)2O─I─(DMAP) (2.339(2)/2.142(2) Å) with 1d (2.251(2)/2.182(2) Å). These solid-state comparisons indicate that the iodine(I) pnictogenate complexes 1b–1g lie in the penumbra between the less reactive carbonyl and unwieldy more reactive sulfonate O─I─N complexes, making them attractive prospects as potential iodination reagents.Interestingly, despite their aforementioned slight intolerance to water, only the monohydrated solid-state structures could be obtained for 1b, 1c, and 1g (and also potentially 1d, as an unknown solvate(s) in its solid-state structure had to be accounted for with Platon Squeeze) upon relaxation of the Schlenk handling techniques52. Attempts to obtain the non-hydrated solid-state structures for those complexes were unsuccessful, though such hydrated structures have been observed previously for even more unstable iodine(I) complexes46. For the solid-state structures of 1b, 1c, and 1g, were all observed as 1:1 hydrates with a pair of H2O molecules bridging across two neighbouring P=O groups in a planar, diamond-shaped motif (Fig. 4). The possibility of the hydrogen-bonding inducing structural anomalies was considered, as had been observed previously in [N─I─N]- complexes53, in comparison to the three non-hydrated solid-state structures 1d, 1e, and 1f, though no structural anomalies were observed to support similar behaviour for the iodine(I) phosphinate complexes.Fig. 4: The packing of two molecules of 1b·H2O.The intermolecular hydrogen bonding motif of 1b with two molecules of H2O (hydrogen bonds shown in green).Computational studiesThe reliability of computational methods, which had been successfully utilised previously to excellently reproduce the molecular geometries obtained from SCXRD studies of [N─I─N]+ and RC(O)O─I─N (iodine(I) carboxylate) halogen-bonded iodine(I) complexes13,15,23,39,46,49,54,55,56,57, was also explored to see if they remained broadly applicable for this iodine(I) pnictogenate complexes (Supplementary Note 6). The geometry optimisations were performed at the M06-2X/def2-TZVP level of theory58 in the SPARTAN’20 programme59 with CH2Cl2 (dielectric = 8.82) as the solvent (Supplementary Notes 8, 9), and using a conductor-like polarisable continuum model (C-PCM)60,61. Additionally, this enabled the pyridine derivative 1a (Fig. 5), for which the solid-state structure could not be obtained, to also be explored structurally.Fig. 5: The calculated structure of 1a.Calculated using Spartan’20 at the M06-2X/def2-TZVP level of theory. Annotated with its O─I─N bond angle (black), as well as its O─I (red) and I─N (blue) bond lengths in Ã….In terms of the general geometric features, such as the intramolecular O···N distances, the computationally calculated models, 1acalc–1gcalc, showed excellent agreement with the SCXRD-determined structures, with an average O···N distance of 4.42 Å (cf. SCXRD = 4.43 Å). This was also observed for the Ph2P(O)O as a whole, with the averages of the calculated P─O and P=O bond lengths of 1.55 and 1.49 Å, respectively, matching well with the experimentally determined values (cf. 1.54 and 1.49 Å, respectively). Nevertheless, for the halogen-bonded bond lengths of interest, the accuracy was slightly diminished in the computational models, with differences ranging from 0.007–0.074 Å (O─I) and 0.004–0.036 Å (I─N). It should also be noted that in all but one case, the O─I bond lengths were underestimated, and the I─N bond lengths overestimated in the calculated models, with the one exception (1e) agreeing extremely well (O─I/I─N for 1e: calc. = 2.2054/2.2119 Å; expt. = 2.198(3)/2.216(3) Å), such that these generalisations are not appropriate (to a 3σ tolerance, assuming approximate parity in e.s.d. values).Whilst toxic, arsenic-based iodine(I) complexes would be pariahs as iodination reagents, especially toward the synthesis of pharmaceutical compounds, they were of interest to test the fundamental scope of iodine(I) complex synthesis on Group 15 elements. To this end, cacodylic acid (Me2As(O)OH) was also investigated in an analogous fashion to Ph2P(O)OH. Unfortunately, there was no literature procedure for the preparation of (Me2As(O)OAg)n, nor could this material be prepared in a similar fashion to the carboxylic and phosphinic acids without significant contamination, and was therefore ultimately abandoned. However, it should be noted that the corresponding As(O)O─I─N motif was proposed to be viable based on computational DFT studies (Fig. 6), which demonstrated comparable bond lengths/angles to known iodine(I) complexes.Fig. 6: The calculated structure of Me2As(O)O─I─(DMAP).Calculated using Spartan’20 at the M06-2X/def2-TZVP level of theory. Annotated with its O─I─N bond angle (black), as well as its O─I (red) and I─N (blue) bond lengths in Ã….Reactivity studiesThe reactivity of 1b–1g as potential iodination reagents was also of particular interest, given that they represented a distinctly-different motif compared to prior O─I─N complexes primarily based on iodine(I) carboxylates (RC(O)O─I─L; L = Lewis base), in both the hybridisation and electronegativity of the oxygen-adjacent substituent (with sp3 P versus sp2 C). The utility of antipyrine’s straightforward conversion to iodo-antipyrine has been previously demonstrated41,62,63 and provides a useful benchmark in quantifying the reactivity of these iodine(I) phosphinates. The main limitation found for 1b–1g was an inability to isolate and purify the bulk quantities necessary to proceed with such reactivity tests, with only 1d and 1f being found amenable. The importance of purifying the bulk material is due to the potential for the elemental iodine used in the synthesis of the iodine(I) pnictogenates also accomplishing the same iodination of antipyrine, and in better yield than observed under the same conditions for PhC(O)O─I─(DMAP)41.These reactivity tests were performed over three reaction times (0.5, 2, and 22 h; Supplementary Note 4) in triplicate for 1d and 1f, which gave effectively indiscernible average percentage yields over the three reaction times tested (Table 1). When compared to [I(py)2]BF4 (Barluenga’s reagent), which can be considered the industry standard for such iodinations, [I(py)2]BF4 outperformed at 2 h with an average yield of 93%, though was identical at 0.5 h (65%) to 1d (64%) and 1f (65%). In comparison to I2, the iodine(I) phosphinates performed better at shorter reaction times (2 h) with yields of 75/73% (1d/1f) to the 55% observed for I2, though I2 gave higher yields (90%) at 22 h (cf. 77/76% for 1d/1f). Most interestingly, though, was a marked improvement in comparison to the structurally similar iodine(I) carboxylates, with 1d (75%) and 1f (73%) at 2 h giving double the yields of PhC(O)O─I─(DMAP) (37%). Altogether, these comparisons show that the iodine(I) pnictogenates (1d and 1f) can give respectable iodination yields that seem to crest at ~75–80% on a timescale analogous to currently utilised iodination reagents, and though they perform slightly worse than Barluenga’s reagent overall, they greatly outperform iodine(I) carboxylates.Table 1 The average percentage conversions of antipyrine to iodo-antipyrine for complexes 1d and 1f over various reaction times, as well as other prior literature examples for comparisonThese surprising reactivity results support the potential of iodine(I) phosphinates as viable iodination reagents, which is a tantalising prospect if this can be combined with their other advantages. The structural studies of 1b–1g had indicated that the tetrahedrally coordinated phosphorus atom of the Ph2P(O) substituent would project one of the Ph rings away from the iodine atom, whilst the other Ph ring would be proximal to the iodine atom (Fig. 7); a trait not possible in iodine(I) carboxylate complexes at the sp2-hybridised carbonyl carbon atom. In the iodine(I) pnictogenates this could potentially be utilised, most likely in analogous R2P(O) groups bearing bulkier R groups, to facially block one side of the iodine atom, which could be relevant toward future enantioselective applications. The ability to introduce chirality at the phosphorus atom, e.g., with enantiopure BINOL ([1,1’-binaphthalene]-2,2’-diol) derivatives of the form (BINOL)P(O)O─I─L (Fig. 8), much closer to the iodine atom than other prior examples50,64, is also a highly attractive prospect given that these proof-of-concept iodine(I) pnictogenate complexes demonstrate water tolerance in the solid state and respectable reactivity as iodination reagents.Fig. 7: The spacefill representation of 1d.The orientation of the two phenyl substituents of 1d, with one proximal to the iodine atom (inset: the same representation of 1d viewed along the N─I─O bond with the DMAP ligand simplified for clarity).Fig. 8: The calculated structure of (S-BINOL)P(O)O─I─(py).Calculated using Spartan’20 at the M06-2X/def2-TZVP level of theory. Annotated with its O─I─N bond angle (black), as well as its O─I (red) and I─N (blue) bond lengths in Ã… (inset: the spacefill representation of the same complex to better display the axial chirality of the S-BINOL substituent).
