DALTON FULL PAPER J. Chem. Soc., Dalton Trans., 1999, 2189–2196 2189 Complexes of the lanthanide metals (La–Nd, Sm–Lu) with hypophosphite and phosphite ligands: crystal structures of [Ce(H2PO2)3(H2O)], [Dy(H2PO2)3] and [Pr(H2PO2)(HPO3)- (H2O)]?H2O Joanne A. Seddon, Andrew R. W. Jackson, Roman A. Kresin� ski * and Andrew W. G. Platt School of Sciences, StaVordshire University, College Road, Stoke-on-Trent, StaVordshire, UK ST4 2DE. E-mail: sctrak@staVs.ac.uk Received 5th October 1998, Accepted 29th April 1999 Reactions of lanthanide chlorides LnCl3 with NaH2PO2 in aqueous solution aVorded a series of complexes [Ln(H2PO2)3(H2O)n] (Ln = La–Tb, n = 1; Ce–Nd, Dy–Lu; n = 0) with members falling into at least four separate structural types.The crystal structures of members of two of these types, [Ce(H2PO2)3(H2O)] and [Dy(H2PO2)3], have been determined for the first time, along with that of [Pr(H2PO2)(HPO3)(H2O)]?H2O, one of a new series of complexes [Ln(H2PO2)(HPO3)(H2O)n]?nH2O (Ln = La–Gd, n = 1; Tb–Lu; n = 0) obtained by controlled oxidation of [Ln(H2PO2)3(H2O)n].The hydrated members of this series are notable due to their incorporation of chiral channels inhabited by water molecules, and to their ability reversibly to dehydrate. Steric angle sum calculations have been used to rationalise the morphologies of the new complexes. Tripositive lanthanide ions are influenced only slightly by their own d or f orbitals in their bonding with ligands, these interactions being largely electrostatic.Others 1 have expounded from this that the concept of co-ordination saturation becomes equated with one of steric saturation in the case of lanthanide complexes. Thus, a wide variety of physicochemical properties of lanthanide complexes can vary smoothly when surveyed across the lanthanide series, until, at a certain limit, an abrupt variation is encountered.2 This phenomenon is ascribable to the ionic radius of the lanthanide metal decreasing smoothly up to a point where the metal environment is suYciently crowded as to eject one or more ligand atoms from the primary co-ordination sphere. In a desire to survey systematically the structures of a family of simple lanthanide ion complexes we sought a ligand system which was uncomplicated, but nonetheless relatively unexplored, and this prompted our study of hypophosphite (phosphinate) complexes of lanthanides.Hypophosphitecontaining structures appear to be well distributed around the Periodic Table: structures reported to date include those of d-block,3–5 s-block,6–9 p-block,10,11 and actinide metals,12,13 and an early structure of hypophosphite as its ammonium salt.14 These show the hypophosphite ion to be very flexible in its co-ordination, doubly bridging 11,13 m(k1,k1), triply bridging 3,5 m3(k1,k1,k1), and quadruply bridging 4,6 m4(k1,k1,k1,k1) modes all being represented.Four structures of lanthanide hypophosphite complexes have been elucidated to date.These are consistent with the expectation that later lanthanide ions should be less able to accommodate high co-ordination numbers than the larger, earlier lanthanides. Thus, the closely related complexes [La(H2PO2)3- (H2O)] and [Eu(H2PO2)3(H2O)], belonging to structural types H P O H O Ln Ln H P O H O Ln Ln Ln Ln H P O H O Ln Ln Ln m(k1, k1) m3(k1, k1, k1) m4(k1, k1, k1, k1) denoted by us here as A and A9 respectively, accommodate a water ligand in their co-ordination spheres.15,16 Two of the hypophosphite ligands adopt the m(k1,k1) mode and the third m3(k1,k1,k1), the co-ordination number thus being eight in both structures.Despite their evident close similarity, the A and A9 structures diVer in the disposition of the m3(k1,k1,k1) ligand, and therefore cannot be considered isostructural.16 The transition point between them in the lanthanide series has not yet been located and, in planning so to do, we considered that the close similarity of these structural types may prohibit unambiguous attribution of class to intervening lanthanide analogues simply by means of spectroscopic techniques.We anticipated that powder X-ray diVraction and, given the ability to grow suitable crystals, single-crystal X-ray diVraction, would prove our most valuable tools in achieving this and our other aims, which also include establishing for which metals anhydrous structural types may be formed.An example is the recently discovered17 structural type B adopted by [Pr(H2PO2)3], which achieves 8-coordination by use of the m(k1,k1) mode for one ligand only, the other two being m3(k1,k1,k1). In contrast, the late lanthanide complex [Er(H2PO2)3] is reported 18 to be 6-co-ordinate, having an approximately octahedral geometry about the metal ion, all hypophosphite ligands adopting the m(k1,k1) mode; this structure is denoted by us here as belonging to type C. Our final aim was to explore whether other lanthanide structures might adopt types B, C, or an entirely new structural form, potentially one adopting the comparatively rare 1 7-co-ordination. Results and discussion Synthetic studies The reaction of aqueous solutions of lanthanide chlorides with sodium hypophosphite gave rise to the corresponding lanthanide hypophosphites.The compounds are all poorly soluble in water, but dissolve in dilute HCl. Their spectroscopic and other properties, and subsequent structural studies (discussed below), showed that the compounds form four distinct series of structural types depending on the metal and the reaction conditions.These and their interconversions are summarised in Table 1. In some cases air oxidation results in the formation of2190 J. Chem. Soc., Dalton Trans., 1999, 2189–2196 Table 1 Formation of the various classes of lanthanide hypophosphite complexes La Ce Pr Nd (Pm) Sm Eu Gd Tb Dy Ho Er Tm Yb Lu AA 9 B C ! ! b ! b ! b ! ! c ! ! c ! ! ! ! ! ! !, Formed on mixing reagents; b, formed from type A by digestive dehydration; c, formed on forced dehydration of type A9.phosphite, manifest in the 31P NMR spectra of the praseodymium and neodymium products completely dissolved in aqueous HCl. This oxidation is suppressed by carrying out the reaction under nitrogen or by buVering the solution to pH 1.4. Digestion of the type A complexes on a water-bath over one day leads to the formation of anhydrous type B complexes for Ce, Pr and Nd.Interestingly no type B complex is observed for lanthanum and the type A9 compounds do not undergo similar reactions. On prolonged digestion open to the air well defined mixed hypophosphite–phosphite complexes are formed. Two distinct forms of the mixed complex are observed, type 1, [Ln(H2PO2)(HPO3)(H2O)]?H2O (Ln = Pr–Gd) and an anhydrous type 2, [Ln(H2PO2)(HPO3)] (Ln = Tb–Lu), but we have never unambiguously identified the lanthanum and cerium type 1 analogues.Over prolonged periods evidence of further oxidation to phosphate is observed in the 31P NMR spectra of solutions of the products which show, in addition to HPO3 22 and H2PO2 2, a singlet due to PO4 32. The substitution of lanthanide chlorides by the nitrates, whilst giving the same hypophosphite complexes, leads to the acceleration of the formation of type 1 and 2 materials; in these cases the oxidation of hypophosphite by nitrate was confirmed by the identification of the main gaseous nitrogen containing product as N2O contaminated with smaller amounts of NO, the infrared spectrum of the gas collected from the reaction mixture being identical with that expected from the literature.19 Of note here is the observation that the cerium type B complex does not easily appear to undergo oxidation under these conditions.In view of the formation of anhydrous hypophosphites for the heavier lanthanides it seemed reasonable to suppose that at some point the type A9 and type C structures might possess similar stabilities and thus be interconvertible.Indeed, reaction of [Tb(H2PO2)3(H2O)], the last type A9 complex, with refluxing HC(OMe)3 gives a type C complex which is considerably more soluble in water than the other type C compounds and can be precipitated therefrom as type A9 by the addition of acetone. When Ln = Eu a type C compou is also formed, which rapidly rehydrates in air.Attempts to convert type 1 into type 2 materials by dehydration with HC(OMe)3 have not been successful. Spectroscopic and thermal studies All type A, A9, B and C complexes produced possess spectral and thermal properties consistent with their known15–18 structures. Thus the types A and A9 have very similar infrared spectra. Both are indicative of the presence of two or more diVerent types of hypophosphite environment, and the presence of water (Fig. 1). Types A9 have only three P–H stretching bands, presumably due to overlap.Thermal analysis shows an endotherm at around 210 8C, associated with a loss of ca. 5% mass, and consistent in all cases with loss of water. At around 310 8C there is a further compound exothermic event, resulting in a further 2% loss of mass; infrared investigation of the gaseous products of this decomposition revealed that phosphine 20 is evolved, and we believe the change to be attributable to the known21 disproportionation of hypophosphite, occurring concomitantly with its partial oxidation. Type B complexes possess infrared spectra indicative of the absence of water ligands and the presence of two or more diVerent types of hypophosphite. They lose no significant mass below 310 8C at which point an exothermic loss of 5% mass occurs.This is also attributable to the disproportionation of hypophosphite and, in common with the cases of A and A9 above, is associated with another, concomitant, event, although much less in evidence here than in A or A9.Type C complexes possess uncomplicated infrared spectra indicative of rather uniform hypophosphite environments. They lose no significant mass below 340 8C, at which point there is an exothermic gain in mass of ca. 4% occurring over ca. 40 8C. Presumably this is air oxidation of hypophosphite, no loss of phosphine being observed; this fact can be rationalised by the structural studies discussed below. For all the above type A, A9, B and C complexes, their 31P NMR spectra measured in aqueous HCl show only hypophosphite ion in solution. Infrared spectra of the new type 1 complexes show intense broad bands in the appropriate regions indicating the presence of water.The P–H region is simple relative to type A complexes, but the P–O region is rather more complex. Thermal analysis reveals an endothermic loss of mass of ca. 11% at 130 8C, which corresponds to a loss of two water molecules per metal atom, followed by no further events up to 460 8C.Samples heated to 270 8C and allowed to stand in air appear to rehydrate to regenerate materials with type 1 infrared spectra. Most notably, however, 31P NMR spectra of types 1 in aqueous HCl indicate the presence of both phosphite and hypophosphite, suggesting an empirical formulation of Ln(H2PO2)(HPO3)(H2O)2 for type 1 complexes. Proton coupled 31P NMR spectra of the complexes dissolved in concentrated HCl clearly show the presence of both phosphite (doublet) and hypophosphite (triplet).The integrated peak area ratios of these, measured using diVering pulse delays and pulse widths, remains unaltered at 1 : 1, implying that rapid relaxation is occurring due to the presence of the paramagnetic lanthanide ions, and thus the measured ratio is a good reflection of the relative amounts of phosphite and hypo- Fig. 1 Typical infrared spectra of all structural types.J. Chem. Soc., Dalton Trans., 1999, 2189–2196 2191 Table 2 Crystal data for lanthanide hypophosphite complexes Formula M Crystal system Space group a/Å b/Å c/Å a/8 b/8 g/8 Dc/g cm23 Colour F(000) Size/mm q Range/8 hkl ranges Total data Unique data Merging R m/mm21 tmin, tmax R (all data) wR2 (all data) Goodness of fit [Ce(H2PO2)3(H2O)] 353.09 Triclinic P1� 7.1729(10) 7.9827(9) 8.8710(6) 110.643(12) 98.101(13) 104.970(11) 2.642 None 334 0.20 × 0.25 × 0.30 2.54–24.92 28 £ h £ 5, 29 £ k £ 8, 210 £ l £ 10 1964 1274 0.0580 5.661 0.532, 1.000 0.0374 0.0922 1.112 [Dy(H2PO2)3] 357.46 Monoclinic C2/m 14.368(3) 5.7340(10) 12.1230(10) 122.33(2) 2.813 Lilac 660 0.35 × 0.24 × 0.24 1.99–24.84 216 £ h £ 16, 26 £ k £ 5, 210 £ l £ 13 1816 717 0.0495 9.398 0.584, 1.000 0.0309 0.0850 1.129 [Pr(H2PO2)(HPO3)(H2O)]?H2O 1287.62 Orthorhombic P212121 6.6558(5) 7.1539(5) 16.5506(18) 2.713 Light green 608 0.25 × 0.28 × 0.30 3.10–25.03 27 £ h £ 7, 27 £ k £ 6, 218 £ l £ 17 3274 1220 0.0670 6.574 0.740, 1.000 0.0430 0.1029 1.051 phosphite present.The spectral properties of type 2 materials are similar to those of type 1 mentioned above, with the exception that no bands arising from water are evident in their infrared spectra. Their thermogravimetric traces are almost featureless below 460 8C. The infrared spectra of the materials obtained by thermal dehydration of type 1 complexes are very similar to those of type 2, and we believe type 2 complexes to be the anhydrous analogues of type 1, the absence of water being necessitated by smaller ionic radii of the later lanthanides, in the same way as type C are analogues of type A or A9.We therefore tentatively assign the formulation Ln(H2PO2)(HPO3) to type 2. There is some evidence here of the existence of alternative structural types which might be denoted 19 and 29, by analogy with A9. This arises from the interesting fact that type 1 materials when heated to 160 8C are completely dehydrated, and upon rehydration in air revert to type 1.When heated to 270 8C, however, they possess diVerent IR spectra after complete rehydration. These transformations are irreversible, but the products still contain the same 1 : 1 ratio of hypophosphite to phosphite (NMR evidence). Structural studies Powder X-ray diVraction patterns were simulated for all of types A, A9, B and C based upon the literature.15–18 Experimental traces were recorded for all the new and known type A, A9, B and C complexes and showed four families of patterns, in accordance with the categories assigned to each complex by spectroscopic and thermal means.Good agreement was also obtained between each pattern prediction and the experimental trace in all cases, the worst fit being for types C. Patterns were also recorded for all type 1 materials, and also showed a familial resemblance. As a final verification of the compositions of these materials, and in order to clarify the discrepancy between the predicted and observed powder patterns of type C complexes, we sought to examine one of each type by single-crystal X-ray diVraction and found Ce to give satisfactory crystals of type A, Dy of C, and Pr of types B and 1.No satisfactory crystals were obtained of any type A9 complex except for Eu, or of any type 2 complex. Each of the above crystals aVorded a structural solution. Details salient to data collection and structural refinement are presented in Table 2 (and elsewhere for [Pr(H2PO2)3]).17 The structure of the cerium type A complex was confirmed to be [Ce(H2PO2)3(H2O)] and is depicted in Fig. 2. It is, as indicated by supporting methods, virtually isostructural 15 with [La(H2PO2)3(H2O)], the structure of which is outlined briefly above. The Ln–O (hypophosphite) distance therein is 2.560[7] Å, but only 2.492[2] Å ([smean] = i = n i = 1 S(si 2)/n) in the cerium analogue (Table 3). There appears to be some compensation for this contraction in an extension of the hypophosphite ligands; the P–O (hypophosphite) lengths in [Ce(H2PO2)3(H2O)] average 1.495[2] Å as compared with 1.478[6] Å in the lanthanum analogue, the O–P–O angles likewise extending to average 117.77[17]8 as compared to 115.6[5]8 in [La(H2PO2)3(H2O)]. The hydrogen atoms of the waters were not locateable, but their presence was inferred from two, and only two, close contacts between O(1) and hypophosphite atoms O(21)iv and O(22)v which subtend almost tetrahedral angles for O(21)iv ? ? ? O(1)– Ce and O(21)iv ? ? ? O(1) ? ? ? O(22)v.The O(22)v ? ? ? O(1)–Ce Fig. 2 A view22 of [Ce(H2PO2)3(H2O)] along the unit cell b axis. The Ce, P, and H atoms are represented by spheres of decreasing relative arbitrary size. Only the water H atoms are shown, for reasons of clarity, along with their hydrogen-bonding contacts. Symmetry operators used throughout: i = 2x, 2y, 2z; ii = x 1 1, y, z; iii = 1 2 x, 1 2 y, 1 2 z; iv = 1 2 x, 2y, 2z; v = x 1 1, y 1 1, z; vi = 2x, 2y, 1 2 z; vii = x, 2y, z; viii = x, 1 2 y, z; ix = 2 2 x, y 1 1– 2, 1– 2 2 z; x = 2 2 x, y 2 1– 2, 1– 2 2 z; xi = x 2 1, y, z; xii = x 2 1– 2, 11– 2 2 y, 2z; xiii = 11– 2 2 x, 1 2 y, z 2 1– 2; xiv = 11– 2 2 x, 2 2 y, z 2 1– 2; xv = 1 2 x, 1 2 y, 2z; xvi = 2x, y 2 1, 1 2 z; xvii = x, y 2 1, z; xviii = 2x, y, 1 2 z; xix = 2x, 1 2 y, 1 2 z; xx = 2x, y 2 1, 2z; xxi = 2x, y, 2z; xxii = 2x, 1 2 y, 2z; xxiii = 11– 2 2 x, 1 2 y, z 1 1– 2; xxiv = 11– 2 2 x, 2 2 y, z 1 1– 2; xxv = x 1 1– 2, 11– 2 2 y, 2z; xxvi = x, y 1 1, z.2192 J.Chem. Soc., Dalton Trans., 1999, 2189–2196 angle, however, is far larger, which implies that, assuming an sp3 hybridisation at O(1), the attitude of the lone pairs of O(1) to Ce is not one of direct alignment. This is, however, a common phenomenon, manifest also in other structures,23–25 and probably implies a degree of interaction between Ce and both lone pairs; indeed, in this very structure, the orientations of the hypophosphite ligands also are not consistent with an sp3 hybridisation at oxygen atoms and direct alignment between only one oxygen lone pair and Ce.The overall co-ordination geometry (Fig. 3) around Ce conforms most closely to bicapped trigonal prismatic (BTP), although polytopal analysis shape parameters 26 indicate that a degree of square antiprismatic (SAP) character is also present (Table 4; DOD = dodecahedral) which is manifest mainly in the elongation of the O(1) ? ? ? O(31) distance so as to destroy the coplanarity of the sides of the prism.The important bond distances and angles of the type B complex [Pr(H2PO2)3] have been communicated elsewhere.17 Its overall co-ordination geometry conforms fairly closely to square antiprismatic (Table 4). The structure reported 18 for [Er(H2PO2)3] and that determined by us for our dysprosium type C complex are diVerent, Fig. 3 The co-ordination polyhedron in [Ce(H2PO2)3(H2O)].Table 3 Selected interatomic separations (Å) and angles (8) for [Ce(H2PO2)3(H2O)] Ce–O(11) Ce–O(31) Ce–O(12)i Ce–O(22)i P(1)–O(11) P(2)–O(22) P(3)–O(31) O(1) ? ? ? O(21)iv O(31)–Ce–O(11) O(11)–Ce–O(32)iii O(11)–Ce–O(22)i O(31)–Ce–O(21) O(32)iii–Ce–O(21) O(31)–Ce–O(1) O(32)iii–Ce–O(1) O(21)–Ce–O(1) O(11)–Ce–O(12)i O(22)i–Ce–O(12)ii O(1)–Ce–O(12)i O(11)–Ce–O(12)ii O(22)i–Ce–O(12)ii O(1)–Ce–O(12)ii O(11)–P(1)–O(12) O(31)–P(3)–O(32) Ce–O(1) ? ? ? O(22)v 2.432(4) 2.416(5) 2.574(4) 2.482(5) 1.483(5) 1.496(5) 1.484(5) 2.714(7) 86.59(17) 79.92(17) 81.64(16) 78.99(18) 143.03(16) 102.88(18) 72.20(17) 140.46(15) 119.21(15) 77.19(15) 77.56(15) 144.17(15) 130.51(14) 67.61(14) 115.59(28) 118.48(32) 142.04(22) Ce–O(21) Ce–O(1) Ce–O(12)ii Ce–O(32)iii P(1)–O(12) P(2)–O(21) P(3)–O(32) O(1) ? ? ? O(22)v O(31)–Ce–O(32)iii O(31)–Ce–O(22)i O(32)iii–Ce–O(22)i O(11)–Ce–O(21) O(22)i–Ce–O(21) O(11)–Ce–O(1) O(22)i–Ce–O(1) O(31)–Ce–O(12)i O(32)iii–Ce–O(12)i O(21)–Ce–O(12)i O(31)–Ce–O(12)ii O(32)iii–Ce–O(12)ii O(21)–Ce–O(12)ii O(12)i–Ce–O(12)ii O(22)–P(2)–O(21) Ce–O(1) ? ? ? O(21)iv O(21)iv ? ? ? O(1) ? ? ? O(22)v 2.502(5) 2.515(5) 2.607(4) 2.434(5) 1.514(5) 1.504(5) 1.489(5) 2.834(7) 75.64(18) 150.06(18) 75.24(16) 72.10(15) 122.41(15) 146.97(16) 74.52(15) 132.11(16) 143.38(15) 73.06(15) 71.36(15) 119.25(15) 76.11(14) 64.69(16) 119.19(28) 110.75(19) 107.06(21) but we note that the veracity of the former has been queried elsewhere.23 Although the structures do not agree, the axial lengths of the two unit cells do, only the b angles being diVerent.The two structures share the common characteristic of consisting a framework of metal atoms linked by m(k1,k1)- hypophosphite bridges extending along the principal axes of the lattice. The metal atoms (of which there are two crystallographically unique) in [Dy(H2PO2)3] are thereby six-coordinate. The co-ordination geometry is near-octahedral (Fig. 4), the trans angles being perfectly 1808 and the cis angles ranging close around 908, both extremes of this range being found at Dy(2).The mean Dy–O distance is 2.253[1] Å (Table 5), shorter than in any other lanthanide hypophosphite complex (leaving aside [Er(H2PO2)3]).18 The presence of two waters per metal ion for the Type 1 complexes is substantiated by the emergent structure of [Pr- (H2PO2)(HPO3)(H2O)]?H2O (a formulation to be preferred over our former interpretation).17 Only one of these waters is associated directly with the Pr.The metal is once again 8-co-ordinate by means of additional contributions from m(k1,k1)-hypophosphite and a m4(k2,k1,k1,k1)-phosphite (Fig. 5). As in [Ce(H2PO2)3(H2O)], the water H atoms were not definitively located, but were clearly implied by interoxygen distances and angles: as a result it is clear that the structural units are bound together by means of the non-ligand water molecules, which bridge between hypophosphite O atoms as hydrogen-bonding acids, in turn acting as hydrogen-bonding bases to the ligand waters (Fig. 6). The result is a region of the structure inhabited by water molecules associated by hydrogen-bonding and extending along the screw axes. The space group itself being chiral, this area resembles a ‘screw thread’ of hydrogen bonds. Again, as in [Ce(H2PO2)3(H2O)], the geometry at the ligand water is far from tetrahedral. The O(2) ? ? ? O(1) ? ? ? O(2)xii angle is almost tetrahedral, but the corresponding angles between Pr and O(2) or O(2)xii (these are the O atoms which act as hydrogen-bonding bases to H(13) and H(14) respectively) are obtuse; as in [Ce(H2PO2)3(H2O)], no other close contacts are made with O(1) to imply a demand on the remaining lone pair, the implication being, in view of the angles at O(1), that it interacts somewhat with Pr.The geometry at O(2) is more regular (Table 6). The mean hypophosphite P–O Fig. 4 A view22 of [Dy(H2PO2)3]. The Dy, P and O atoms are represented by spheres of decreasing relative arbitrary size, and H atoms are omitted. H P O O O Ln Ln Ln Ln m4(k2, k1, k1, k1)J.Chem. Soc., Dalton Trans., 1999, 2189–2196 2193 Table 4 Polytopal a analysis shape parameters for 8-co-ordinate lanthanide hypophosphite complexes [Ce(H2PO2)3(H2O)] [Pr(H2PO2)3] b [Pr(H2PO2)(HPO3)(H2O)]?H2O BTP SAP DOD d 1(57)3/8 d 2(68)4/8 d 2(58)3/8 d 1(67)4/8 f 7–1–2–8/8 f 5–3–4–6/8 13.8 2.9 45.6 42.9 16.0 16.6 8.0 7.6 50.0 57.4 24.2 37.7 34.1 16.0 50.7 45.6 5.1 20.4 21.8 0.0 48.2 48.2 14.1 14.1 0.0 0.0 52.5 52.5 24.5 24.5 29.5 29.5 29.5 29.5 0.0 0.0 a Vertices are assigned, for the above three structures respectively, as follows: 1 [O(12)ii, O(11), O(21)], 2 [O(21), O(21), O(22)], 3 [O(32)iii, O(31), O(23)xi], 4 [O(22)i, O(12)xv, O(1)], 5 [O(31), O(32)xix, O(11)], 6 [O(12)i, O(22)iv, O(12)ix], 7 [O(1), O(22), O(22)ix], 8 [O(11), O(12)xi, O(21)x].b See ref. 17. distance of 1.491[4] Å in this structure appears to be shorter than the 1.513[2] Å seen in [Pr(H2PO2)3], but this is to be interpreted in the light of the fact that both the co-ordination mode of the hypophosphites and the nature of the coligands is diVerent in both structures.In the phosphite ligand the P(2)–O(23) distance is shorter than the distances to O(21) and O(22), which situation has been observed before 27,28 and is analogous to the asymmetry of the m3(k1,k1,k1)-hypophosphite ligands, caused by diVerent ligating demands on the two O atoms.In [Pr(H2- PO2)(HPO3)(H2O)]?H2O the co-ordination polyhedron is a dodecahedron, perhaps rather distorted in the direction of SAP by the loss of planarity of the O(11), O(23)xi, O(1), O(12)ix trapezoid (Fig. 7). The degree of steric crowding in the above complexes can be compared by the use of steric angle sum (SAS) calculations,1 which express the fraction of surface area of a notional 1 Å Fig. 5 A view22 of [Pr(H2PO2)(HPO3)(H2O)]?H2O.The Pr, P, O and H atoms are depicted with decreasing arbitrary radii. Only H atoms, and their hydrogen-bonds to surrounding atoms, are shown for purposes of clarity. Table 5 Selected interatomic separations (Å) and angles (8) for [Dy(H2PO2)3] Dy(1)–O(11) Dy(2)–O(12) P(1)–O(12) P(2)–O(21) O(11)–Dy(1)–O(21) O(21)vi–Dy(1)–O(21)vii O(11)–Dy(1)–O(11)vi O(12)–Dy(2)–O(31) O(31)–Dy(2)–O(31)vii O(12)i–Dy(2)–O(12) O(12)–P(1)–O(11) O(31)–P(3)–O(31)viii 2.243(6) 2.246(6) 1.439(7) 1.486(5) 91.3(2) 89.5(2) 180.0 88.4(2) 90.1(2) 180.0 121.1(5) 117.0(4) Dy(1)–O(21) Dy(2)–O(31) P(1)–O(11) P(3)–O(31) O(11)–Dy(1)–O(21)vi O(21)–Dy(1)–O(21)vii O(21)–Dy(1)–O(21)vi O(12)i–Dy(2)–O(31) O(31)i–Dy(2)–O(31)vii O(31)–Dy(2)–O(31)i O(21)–P(2)–O(21)viii 2.256(4) 2.259(5) 1.465(7) 1.489(5) 88.7(2) 90.5(2) 180.0 91.6(2) 89.9(2) 180.0 116.6(4) radius sphere, at a metal’s centre, which is covered by conic projections of the ligating atoms.Computationally this is done by summing all of the individual atoms’ contributions but, since overlap between the individual atoms’ projections is not counted twice, an overlap correction is then applied.Values calculated for the above complexes are presented in Table 7 along with those for some related complexes. The table is arranged according to atomic weight of the metal, and the eVects of the lanthanide contraction are immediately evident in the general increase in SAS values for the later lanthanides. The lanthanum complex has the smallest value, and there is no overlap correction required.In the case of the isostructural type A cerium complex a very small overlap correction is present, which arises from a close approach of atoms O(12)i and O(12)ii which reside in two m3(k1,k1,k1)-hypophosphite ligands. This contact may be expected to become closer in the later lanthanide structures, and suggests that the structural change from A to A9 occurs when it reaches a limiting value. The SAS calculation for the type A9 [Eu(H2PO2)3(H2O)] requires an overlap correction, but this arises mainly from a contact between the Fig. 6 The packing arrangement of [Pr(H2PO2)(HPO3)(H2O)]?H2O seen 22 along the unit cell a axis. Some bonds are omitted for clarity. The symbol X denotes the approximate location of a screw axis consisting the centre of a ‘screw thread’ of water molecules. Fig. 7 The co-ordination around the Pr atom in [Pr(H2PO2)(HPO3)- (H2O)]?H2O.2194 J. Chem. Soc., Dalton Trans., 1999, 2189–2196 Table 6 Selected interatomic separation (Å) and angles (8) for [Pr(H2PO2)(HPO3)(H2O)]?H2O Pr–O(11) Pr–O(22) Pr–O(12)ix Pr–O(22)ix P(1)–O(11) P(2)–O(21) P(2)–O(23) O(1) ? ? ? O(2)xii O(2) ? ? ? O(11)xiv O(23)xi–Pr–O(21)x O(21)x–Pr–O(22)ix O(21)x–Pr–O(1) O(23)xi–Pr–O(12)ix O(22)ix–Pr–O(12)ix O(23)xi–Pr–O(11) O(22)ix–Pr–O(11) O(12)ix–Pr–O(11) O(21)x–Pr–O(22) O(1)–Pr–O(22) O(11)–Pr–O(22) O(21)x–Pr–O(21) O(1)–Pr–O(21) O(11)–Pr–O(21) O(11)–P(1)–O(12) O(23)–P(2)–O(22) Pr–O(1) ? ? ? O(2) O(2) ? ? ? O(1) ? ? ? O(2)xii O(1) ? ? ? O(2) ? ? ? O(11)xiv 2.479(6) 2.570(5) 2.479(6) 2.395(5) 1.478(6) 1.540(6) 1.490(6) 2.827(9) 2.866(8) 90.82(20) 171.07(21) 79.98(23) 132.99(21) 82.07(20) 77.29(23) 94.55(20) 147.52(21) 66.84(19) 123.70(22) 77.55(20) 122.41(15) 137.49(20) 77.99(22) 118.86(33) 113.05(34) 133.35(30) 101.48(26) 112.18(32) Pr–O(21) Pr–O(1) Pr–O(21)x Pr–O(23)xi P(1)–O(12) P(2)–O(22) O(1) ? ? ? O(2) O(2) ? ? ? O(12)xiii O(23)xi–Pr–O(22)ix O(23)xi–Pr–O(1) O(22)ix–Pr–O(1) O(21)x–Pr–O(12)ix O(1)–Pr–O(12)ix O(21)x–Pr–O(11) O(1)–Pr–O(11) O(23)xi–Pr–O(22) O(22)ix–Pr–O(22) O(12)ix–Pr–O(22) O(23)xi–Pr–O(21) O(22)ix–Pr–O(21) O(12)ix–Pr–O(21) O(22)–Pr–O(21) O(23)–P(2)–O(21) O(21)–P(2)–O(22) Pr–O(1) ? ? ? O(2)xii O(1) ? ? ? O(2) ? ? ? O(12)xiii O(11)xiv ? ? ? O(2) ? ? ? O(12)xiii 2.639(6) 2.475(7) 2.390(6) 2.367(6) 1.505(7) 1.547(6) 2.784(9) 2.794(8) 80.27(20) 70.85(21) 96.31(23) 103.82(20) 68.29(21) 84.06(20) 143.93(23) 147.80(19) 121.53(13) 76.93(20) 135.61(19) 65.63(18) 71.17(19) 56.01(18) 114.37(32) 104.86(33) 123.82(27) 88.62(27) 114.58(29) m3(k1,k1,k1) and the water ligands, rather than between like ligands.The type B [Pr(H2PO2)3] contains no strong contacts such as would necessitate an overlap correction, despite possessing a larger SAS value than [Ce(H2PO2)3(H2O)] and being presumably more sterically crowded. The distribution of ligand atoms17 around the metal ion in the type B materials is evidently quite eYcient, which is in accordance with the observed SAP co-ordination geometry.This might explain why type B forms by dehydration of type A, even in the presence of water. The complex [Dy(H2PO2)3] contains two crystallographically inequivalent dysprosium ions, and thus has two ascribable SAS values, the comparatively small magnitudes of which reflect the fact that this structure is a 6-co-ordinate one. This allows the ligands to move well away from one another, which is manifest in the lack of any overlap correction and the lower density of [Dy(H2PO2)3] compared to 17 [Pr(H2PO2)3]. Furthermore, it might also aVord type C complexes their resistance to disproportionation of hypophosphite, which occurs readily in types A, A9 and B at elevated temperatures.Since one of the products of this process is phosphine, interphosphorus oxygen transfer is a requisite, but there are no P ? ? ? O approaches in [Dy(H2PO2)3] closer than 3.508(6) Å.This can be compared with values of 3.126(5) Å in [Ce(H2PO2)3(H2O)] and 2.918(4) Å in [Pr(H2- PO2)3], both of which fall well inside the sum of van der Waals Table 7 Steric angle sums (SASs) for lanthanide hypophosphite complexes SAS Complex Uncorrected Corrected [La(H2PO2)3(H2O)] a [Ce(H2PO2)3(H2O)] [Pr(H2PO2)3] b [Pr(H2PO2)(HPO3)H2O]?H2O [Eu(H2PO2)3(H2O)] c [Dy(H2PO2)3] 0.653 0.691 0.698 0.705 0.738 0.6503, 0.6486 0.653 0.691 0.698 0.699 0.733 0.6503, 0.6486 a See ref. 15. b See ref. 17. c See ref. 16. radii of P and O for these type A and B complexes both of which undergo thermal disproportionation. Conclusion Whilst the hypophosphite–lanthanide system has not proven uncomplicated in its behaviour, relatively simple analyses, such as SAS calculations, appear to have aVorded some fundamental insights to the structural preferences and reactivities of lanthanide hypophosphite complexes. These studies have established the classes of structures adopted by these materials and the ranges over which each is formed in the lanthanide series. The structure of type A has been verified by examination of a new member, and that of type C established correctly for the first time.No evidence for a 7-co-ordinate structural type, between the former 8- and the latter 6-co-ordinate complexes, has been found. A new class of mixed-ligand hypophosphite– phosphite complex, containing interesting chiral channels of water molecules, has been synthesized and characterised structurally.Experimental Instrumentation JEOL JNM-FX270 and FX90Q (NMR), ATI Mattson Genesis single beam (FTIR) Philips 1050 Cu-Ka (powder diVraction) and Netzch STA 409 EP (TGA) instruments were used for routine analysis. Analysis Oxidisable phosphorus was determined by treatment of samples with an excess of bromine (generated quantitatively from bromate and bromide). The sample was stirred with bromine for 3 h in a stoppered flask, cooled in ice and an excess of potassium iodide added.The iodine liberated was titrated with standard sodium thiosulfate solution. For parity of reporting for pure hypophosphite and mixed-ligand systems, the results are expressed in terms of calculated molar mass. Percentage phosphorus values were determined by DCP (direct current plasma) atomic absorption spectroscopy.J. Chem. Soc., Dalton Trans., 1999, 2189–2196 2195 Syntheses [Ln(H2PO2)3(H2O)n] (types A and A9, n 5 1; C, n 5 0), general procedure.Sodium hypophosphite (ca. 1.00 g) was dissolved in pH 1.4 (KCl–HCl) buVer solution (10 ml). Hydrated lanthanide chloride (ca. 1.20 g) in the buVer solution (10 ml) was added. The mixture was allowed to stand in air at room temperature for one hour. A solid precipitated from solution where Ln = La–Eu, Dy–Lu. For Ln = Gd or Tb addition of a seed crystal resulted in rapid precipitation from the supersaturated solution. The resulting solids, bearing the characteristic colour of the Ln31 ions, were collected by suction filtration, washed with water, and dried at the pump. Yields 61– 89%.Representative analyses: [Ce(H2PO2)3(H2O)], found P 26.1, H8CeO7P3 requires 26.3%; [Dy(H2PO2)3], found P 25.6. H6DyO6P3 requires 26.0%; [Sm(H2PO2)3(H2O)], M found 366, theoretical 363; [Eu(H2PO2)3(H2O)], M found 362, theoretical 365. [Ln(H2PO2)3] (type B). Stoichiometric amounts of sodium hypophosphite and the lanthanide chloride were dissolved separately in the minimum amount of water.On mixing a precipitate was formed immediately, and the reaction mixture heated under reflux under nitrogen for several hours. Needle-like crystals formed of the colour of the corresponding Ln31 ion, which were collected by suction filtration, washed with water, and dried at the pump. Yields 51–69%. [Pr(H2PO2)3)], M found 336, theoretical 336. [Tb(H2PO2)3] (type C) from [Tb(H2PO2)3(H2O)] (type A9). A small quantity of [Tb(H2PO2)3(H2O)] was stirred under reflux with excess of trimethyl orthoformate in a nitrogen atmosphere for 2 h.The resulting material was filtered in a dry-box, washed with dry thf and allowed to dry. [Ln(H2PO2)(HPO3)(H2O)n]?H2O (type 1, n 5 1; 2, n 5 0). Air oxidation of the lanthanide hypophosphite complexes, monitored by IR spectroscopy, was achieved by in situ air oxidation on heating the lanthanide chloride with sodium hypophosphite in aqueous solution on a steam-bath for several hours.Needle like crystals were formed for the praseodymium and neodymium complex, whilst for the later lanthanides the complexes were obtained as fine powders. The products were filtered oV, washed with water and dried in air at the pump. Representative analyses: [Nd(H2PO2)(HPO3)(H2O)]?H2O, M found 301, theoretical 325; [Tm(H2PO2)(HPO3)], M found 316, theoretical 314; [Lu(H2PO2)(HPO3)], M found 325, theoretical 320. Crystallography DiVraction data were collected, using Mo-Ka radiation, according to previously published procedures.29 Non-H atoms were modelled 30 anisotropically, and H atoms isotropically, these being placed in theoretical positions.Following convergence of full-matrix least-squares refinement30 using merged data, these were corrected 31 for absorption eVects and refinement resumed to convergence. [Ce(H2PO2)3(H2O)]. The structure was solved by isomorphous replacement.15 A common refined P–H distance (1.309(14) Å) was used, and water H atoms were restrained to sit 30% along the length of their putative hydrogen-bonding vectors and, as a result, the refinement was unstable and required slight damping.Final error limits were estimated using undamped cycles of refinement. The largest hole and peak in the final Fourier-diVerence map were 21.08 and 1.40 e Å23 (at Ce). [Dy(H2PO2)3]. Data were collected as for triclinic symmetry, and transformed to the above system. The structure was solved by trial placement of heavy atoms at special positions.A common refined P–H distance (1.245(15) Å) was used for the refinement. The largest hole and peak in the final Fourierdi Verence map were 20.86 and 0.74 e Å23 (ca. 1.2 Å from Dy(1) and O(11)). [Pr(H2PO2)(HPO3)(H2O)]?H2O. The structure was solved by Patterson methods.32 A common refined P–H distance (1.266(26) Å) was used, which was considered a reasonable step in the light of the similarity of the IR stretching frequencies.Some systematic error in the data necessitated that phosphite O atoms be weakly restrained to be isotropic to prevent ‘nonpositive definite’ collapse. Water H atoms were restrained to sit 30% along the length of their putative hydrogen-bonding vectors and, as a result, the refinement was unstable and required damping. Final error limits were estimated using undamped cycles of refinement. The largest hole and peak in the final Fourier-diVerence map were 21.29 and 3.99 e Å23 (at Pr).CCDC reference number 186/1450. Acknowledgements The authors gratefully acknowledge the use of the EPSRC’s X-Ray Crystallography Service, and Chemical Database Service at Daresbury. Thanks are due to Stephanie Barnett for simulations of powder patterns, and to all in the Divison of Ceramics for use of apparatus. References 1 X.-Z. Zhang, A.-L. Guo, Y.-T. Xu, X.-F. Li and P.-N. Sun, Polyhedron, 1987, 6, 1041. 2 F. A. Hart, in Comprehensive Coordination Chemistry, eds. G.Wilkinson, R. D. Gillard and J. A. McCleverty, Pergamon Press, Oxford, 1987, vol. 3. 3 M. D. Marcos, R. Ibanez, P. Amoros and A. LeBail, Acta Crystallogr., Sect. C, 1991, 47, 1152. 4 T. J. R. Weakley, Acta Crystallogr., Sect. B, 1979, 35, 42. 5 M. D. Marcos, P. Amoros, F. Sapina, A. Beltran-Porter, R. Martinez-Manez and J. P. Attfield, Inorg. Chem., 1993, 32, 5044. 6 T. Matsuzaki and Y. Itaka, Acta Crystallogr., Sect. B, 1969, 25, 1932. 7 B. O. Loopstra, Int. Estab. Nucl. Energy Res.Publ., 1958, 15, 64. 8 A. R. Pedrazuela, S. Garcia-Blanco and L. Rivoir, An. Soc. Esp. Fis. Quim., 1953, 49, 255. 9 T. Akimoto, Ph.D. Thesis, University of Tokyo, 1965. 10 T. J. R. Weakley, J. Chem. Soc. Pak., 1983, 5, 279. 11 T. J. R. Weakley and W. W. L. Watt, Acta Crystallogr., Sect. B, 1979, 35, 3023. 12 P. A. Tanner, T. Sze, T. C. Mak and W. Yip, J. Cryst. Spectrosc. Res., 1992, 22, 25. 13 P. A. Tanner, S.-T Hung, T. C. W. Mak and W. Ru-Ji, Polyhedron, 1992, 11, 817. 14 W. H. Zachariasen and R. C. L. Mooney, J. Chem. Phys., 1934, 2, 34. 15 V. M. Ionov, L. A. Aslanov, V. B. Rybakov and M. A. Porai-Koshits, Kristallografiya, 1973, 18, 403. 16 V. M. Ionov, L. A. Aslanov, M. A. Porai-Koshits and V. B. Rybakov, Kristallografiya, 1973, 18, 405. 17 J. A. Seddon, A. R. W. Jackson, R. A. Kresinski and A. W. G. Platt, Polyhedron, 1996, 15, 1899. 18 L. A. Aslanov, V. M. Ionov, M. A. Porai-Koshits, V. G. Lebedev, B. N. Kulikovskii, O. N. Gilyarov and T. L. Novoderezhkina, Izv. Acad. Nauk. SSSR, Neorg. Mater., 1975, 11, 96. 19 J. Laane and J. R. Ohlsen, Prog. Inorg. Chem., 1980, 27, 465. 20 K. Nakamoto, Infrared and Raman Spectra of Inorganic and Coordination Compounds, 4th edn., Wiley, New York, 1986. 21 J. W. Mellor, Comprehensive Treatise on Inorganic and Theoretical Chemistry, Longmans, New York, London, 1971, vol. 8, suppl. 3. 22 E. K. Davies, SNOOPI, Chemical Crystallography Laboratory, University of Oxford, 1982. 23 The United Kingdom Chemical Database Service, D. A Fletcher, R. F. McMeeking and D. Parkin, J. Chem. Inf. Comput. Sci., 1996, 36, 746. 24 D. L. Faithfull, J. M. Harrowfield, M. I. Ogden, B. W. Skelton, K. Third and A. H. White, Aust. J. Chem., 1992, 45, 583. 25 M. C. Favas, D. L. Kepert, B. W. Skelton and A. H. White, J. Chem. Soc., Dalton Trans., 1980, 454. 26 M. A. Porai-Koshits and L. A. Aslanov, J. Struct. Chem., 1978, 13, 244.2196 J. Chem. Soc., Dalton Trans., 1999, 2189–2196 27 Y.-P. Zhang, H. Hu and A. Clearfield, Inorg. Chim. Acta, 1992, 193, 35. 28 J.-D. Foulon, N. Tijani, J. Durand, M. Rafiq and L. Cot, Acta Crystallogr., Sect. C, 1993, 49, 1. 29 J. A. Darr, S. R. Drake, M. B. Hursthouse and K. M. A. Malik, Inorg. Chem., 1993, 32, 5704. 30 G. M. Sheldrick, SHELXL 93, Program for Crystal Structure Refinement, University of Göttingen, 1993. 31 N. Walker and D. Stuart, Acta Crystallogr., Sect. A, 1983, 39, 158. 32 G. M. Sheldrick, SHELXS 86, Program for the Solution of Crystal Structures, University of Göttingen, 1986. Paper 8/07727D