X-ray crystallography of calcium phosphates: the structures of dicalcium phosphate and monocalcium phosphate monohydrate

Abstract

One of the major fields of research in modern soil chemistry is concerned with the nature and proportions of the various phosphate minerals present in the soil, and applied as fertilizers. Phosphorous is an essential plant food whose chief natural source is the calcium phosphate mineral, apatite. Unfortunately apatite itself cannot easily be absorbed by the plant-roots and chemical treatment is essential to produce a suitable fertilizer. The best methods of chemical treatment, from the point of view of industrial practicability, and availability as plant foods of the treated phosphate rock, have recently been the subject of much discussion among agricultural chemists. The complex nature of many calcium phosphate precipitates and the difficulty and tedium of adequate field studies hae led to a demand for detailed knowledge of the structures of certain calcium orthophosphates, in the hope that this might lead to a more exact knowledge of the phases likely to be present in fertilizers and in the soil, and, in addition, throw light on the possible ease of transformation reactions between different but related structures.

A programme of research into these structures has been undertaken in collaboration with the Research Department of Scottish Agricultural Industries. The structures of fluorapatite (1) and of dicalcium phosphate dihydrate (2) have already been determined.

This thesis describes structural determinations on two further orthophosphates of particular interest in fertilizer manufacture, namely anhydrous dicalcium phosphate, CaHPO₄, and monocalcium phosphate monohydrate, Ca(H₂PO₄)₂H₂O. The former is the phase most likely to be present in compound fertilizers whose manufacture involves the use of free ammonia or nitric acid, such as ammoniated superphosphate and "Nitro-Phosphate". These preparations are of especial interest in view of the current world shortage of sulphur, since sulphuric acid is not necessary for their manufacture. The alternatives mentioned are expensive, however, and for economy it is desirable not to carry the conversion of apatite to dicalcium phosphate any further than is necessary to obtain a product with similar properties to the latter. Hence the structure of dicalcium phosphate, considered in relation to that of apatite, will throw light on the possible nature of the phases present in precipitated solids with Ca/P ratios between 1.66 (apatite) and 1.0 (dicalciun phosphate).

Eisenberger (3) and Arnold (4) have suggested, for example, that there might be a series of solid solutions over part at least of this range.

Monoocalcium phosphate monohydrate, on the other hand, is the phase present (5) in the ordinary superphosphate fertilizer produced by the action of sulphuric acid on apatite rock. The possibility of a structural analogy with dicalcium phosphate dihydrate has recently been pointed out (6) on morphological evidence. In any case it is to be anticipated from chemical theory and the known structures of related compounds that both calcium phosphates mentioned consist of a network of tetrahedra formed by the PO₄'" ion, alternating with Ca++ ions in the centre of a coordination shell of between six and nine oxygen atoms. The structure will be further bound by hydrogen bonds between oxygens of neighbouring tetrahedra, and with any water molecules. In dicalcium phosphate there is one hydrogen to every four oxygens, so that only half the oxygens can be linked by hydrogen bonds, but in the monocalcium phosphate there are sufficient hydrogen atoms to enable every oxygen to take part in hydrogen bonding.

The crystal structures of several orthophosphates have ,previously been determined, the phosphate ion in each being found to be tetrahedral. The mean P - O distance reported has varied slightly. In KH₂PO₄ (7) it was 1.55A; in BP0₄ (8) it was

1.54A, and in Ag₃PO₄ (9) it was 1.61A. Recently, the structure of the parent orthophosphoric acid, H₃PO₄, has been published (10), the mean P - O distance being 1.56A. Individual bond lengths and angles have usually been found to vary appreciably; some of these variations may be significant and related to packing distortions.

The structure of fluorapatite, Ca₅F(PO₄)₃ which is isomorphous with hydroxyapatite, Ca₅OH(PO₄)3, has been determined, (11, 1). The crystal possesses hexagonal symmetry. Chains of Ca atoms lie along three-fold axes in the c direction, each coordinated by nine oxygens belonging to phosphate tetrahedra, which link the chains transversely. These linkages produce a hexagonal network, like a honeycomb, with open channels through in the c direction. Further Ca⁺⁺ ions fit into the walls of these channels, being partly coordinated by six oxygens belonging to the columns, and partly by the F or OH ions which occupy the centre of the open channels, lying six-fold screw axes. The total calcium coordination is thus seven, as compared to nine for the "chain" calciums.

The only other calcium orthophosphate structure fully determined is that of dicalcium phosphate dihydrate, CaHPO₄,2H₂O, a preliminary account of which has been published (2). The crystals are monoclinic and the structure is similar to that of the almost isomorphous gypsum, CaSO₄,2H₂O (12). It consists of sheets in the a c plane, containing parallel chains of = PO₄ = Ca = PO₄ = Ca =. Adjoining chains are related by centres of symmetry, and differ in height along the b axis by 2.3A to give a corrugated effect.

The sheets are repeated at a distance of b/2 by vertical screw axes, and are separated by water molecules which link the sheets in the b direction. Tile calcium ion has a coordination of eight (six oxygens from PO₄ tetrahedra, and two water molecules).

The unit cells of both dicalcium phosphate and monocalcium phosphate monohydrate have been determined (6) and accurate intensity measurements of their respective X-ray powder patterns, using a micro-densitometer, have also been published (13).

In the present investigation, the structural determinations have been tackled, as far as possible, independently of chemical postulates. Only the numbers and scattering powers of the atoms involved, and, in the case of CaHPO₄, the approximate length of the PO₄ bond, were assumed.