Title:

The dissolution of mineral phosphate in soil

The use of cheap, sparingly soluble calcium phosphate fertilizers is increasitgly widespread, particularly in the extensive agriculture systems of the tropics where very high yields are not sought, and phosphate deficiency is a major limitation to crop production. At present there is little quantitative understanding of the factors determining the rates of dissolution of calcium phosphates in soils. Existing quantitative treatments are inadequate, being either empirical or based on oversimplified theory. By developing a precise model of the dissolution process, it should be possible to shortcut the usual practice of running extensive field trials to establish the responses over a wide range of soil conditions and management practices. In this thesis a model which makes no arbitrary assumptions is developed for predicting the rates of dissolution of dicalcium phosphate dihydrate (DCPD) in soils. DCPD is the initial reaction product of the dissolution of many phosphatic fertilizers, and is an important fertilizer in its own right; the mechanisms governing its dissolution in soils are basically the same for other, more complex calcium phosphates. The simple case of a planar layer of DCPD in contact with soil is considered first to introduce the principles of the model. This is the simplest system for measuring experimentally the solute concentration profiles close to the dissolving surface, in order to test the model. The model is then extended to describe the dissolution of granules of DCPD in soil. The model comprises numerical solutions of mathematical equations describing the diffusion and reaction of calcium, phosphate and base in soil. The concentrations of calcium, phosphate and hydrogen ions in the soil solution at the mineral/soil boundary are found (a) from the ion activity product of DCPD and (b) by equating the fluxes of calcium, phosphate and base across the boundary (1 mol of DCPD gives 1 mol each of calcium, phosphate and base). In the granular system, the diminution of the granules as they dissolve, and the effect of neighbouring particles on each other are allowed for. The solute concentration profiles predicted for the planar system agreed with experimentally measured profiles; and the predicted net rates of dissolution of granules of DCPD agreed with the rates determined by a radioactivetracer technique, in which ^{45}Ca dissolved from labelled DCPD is recovered from the soil with an extractant, saturated with respect to DCPD. Thus all important processes have been accounted for in the model. Since the theory is nonspecific, the model should apply equally well to most other soils. The model has nine input parameters : the concentrations of calcium and phosphate in the native soil solution, the native soil pH, the phosphate and lime potential buffer capacities of the soil, the moisture status, the diffusion impedance factor, and the rate of application and particle size of the DCPD. A sensitivity analysis of the model showed that the rate is particularly dependent on particle size, rate of application, and the pH and concentration of calcium in the soil solution. If the granules are so stages the rate of dissolution is independent of the soil buffer terms. But for typical rates and methods of application, neighbouring granules will influence each other, and the consequent interactions between the rate determining variables are complex. The extension of the model to describe the dissolution of carbonateapatites, and hence rock phosphates, is discussed.
