Phosphate retention by soil in relation to waste disposal

The disposal of large amounts of domestic sewage water and liquid manure, both containing dissolved phosphates, is often problematic. Discharge of these into (shallow and standing) surface waters is highly undesirable, as phosphate is considered to be one of the prime causes of eutrophication. If, on the other hand, these materials are disposed of onto land, losses of P from soil via surface runoff and leaching are likely to increase, thus contributing again to eutrophication phenomena. An additional reason for concern about increased leaching losses in the case discussed is found in the simultaneous addition of large amounts of water. This may rapidly transport the phosphate to deeper layers, enhancing the chance of penetration into ground- or drainage water.In the present text it was attempted to construct a model describing the movement of phosphate through soil following sizable applications of water. Such a model requires in the first place a quantitative insight into the kinetics of the processes governing the retention of phosphate in soil.Since only scant information is available about phosphate retention in soils treated with soluble phosphate compounds, extensive analytical data were collected from a sandy soil profile that had been exposed to regular applications of (raw) sewage water during a period of about 45 years. As a result of these sewage water additions, phosphates accumulated in the soil, mainly in the top 50 em. Fractionation of these accumulated forms of phosphate via selective extraction methods indicated that aluminum- and ironbound phosphates constitute the largest fraction (60-75%). Organicbound phosphates and calcium-bound phosphates were present in smaller quantities. Phosphate analysis of sewage water and effluent leaving the soil via a system of tile drains indicated that, in spite of 45 years prior usage, the removal of phosphate from the sewage water by the soil is still very effective (around 90%). Moreover, some other substances were partly or nearly completely removed from the sewage water, during its percolation through the soil (Chapter 3).In view of the above findings indicating that aluminum (and iron) were the main binding agents for phosphate, the total and oxalate extractable forms of these components were determined in the soil. These results strongly suggested that in the top layers (0-50) cm of the soil treated with sewage water accumulation of aluminum compounds had taken place in the past, most likely due to the supply of sewage water containing industrial waste water. Additional experiments showed, furthermore, that the phosphate binding capacity of different sandy soils could be related to the oxalate extractable forms of aluminum and iron. Under conditions as were found in soils treated with sewage water, the molar ratio of oxalate extractable (Al+Fe) over the phosphate retained tended to a value around three. A determination of the oxalate extractable aluminum plus iron thus presents a simple method to obtain a fair estimate of the phosphate bonding capacity of sandy soils, thereby assuming that the above value of the ratio (Al+Fe)/P remains valid (Chapter 4).Kinetic aspects of the retention processes of dissolved orthophosphate with soil were studied with the help of batch shaking experiments, using samples of the soil treated with sewage water (Chapter 5). The samples were suspended in solutions containing most of the inorganic ions that were present in the sewage water, at the same concentrations. The soil suspensions were brought to pH 6.2 and adjusted when necessary. Determination of dissolved P as a function of time showed that P was removed rapidly from the solution initially, followed by a much slower rate of removal over longer periods. With the help of a graphical procedure the contributions of the fast and slow processes were separated out; adsorption processes were considered to be responsible for the fast removal, whereas the slow rate of removal was associated with the formation of a rather insoluble and non-reactive form of solid phosphate, presumably a (calcium) aluminum-phoshate.The equilibrium adsorption processes could be described with two Langmuir equations, which lead to the introduction, in the model. of two pools of adsorbed phosphate, i.e. X 1 and X 2 . In addition a third pool of solid phosphate was distinguished, viz pool N, comprising the solid phosphate compounds formed during slow retention processes. This division of retained phosphates into an adsorbed fraction and another immobilized form was supported by results of isotopic exchange experiments using 32P. It was found that the adsorbed forms present in the pools X 1 and X 2 were relatively readily accessible to isotopic exchange whereas forms of phosphate present in pool N were rather inaccessible to exchange with 32P labelled species in solution.An important feature associated with the formation of phosphate in pool N was the apparent freeing of adsorption sites, suggesting that during the slow retention process part of the adsorbed phosphate was released from the adsorption sites.The decrease of the phosphate concentration in the soil sus pensions was described in terms of three rate equations. The reactions leading to the formation of the forms of phosphate present in the pools X 1 and X 2 were treated as simple first order reactions, the rate being proportional to the 'excess' concentration. The rate constants k1 and k2 ,defining the loading of pools XX 1 and X 2 , respectively, were derived from adsorption experiments. It was shown that k1 differed significantly from k2 viz, k1/k2 ≈ 50. Since it was the main objective to study the retention of phosphate, only super ficial attention was paid to desorption processes. The rate constants for desorption, kd1 and kd2 were therefore given the same values as the constants k1 and k2 , although there are indications that the ensuing assumption of reversible adsorption is an oversimplification. The loading of pool N was defined as a second order equation, the rate being proportional to the 'excess' adsorbed phosphate present in pool X 2 , implying that P present in pool N is formed at the cost of P present in pool X 2 . Since the formation of P in pool N must involve a P-binding agent with limited supply for a given soil, a delimiting factor was introduced in the form of an available capacity. This factor was related to the oxalate extrac table amounts of Al and Fe present in the soil. The rate constant of this slow reaction k ' n , is presumably independent of the 'excess' amount adsorbed in pool X 2 and the available capacity of the P binding agent.On the basis of the presumed three pools of solid phosphate and a pool of dissolved phosphate a simulation model was developed; the transformation processes between the different pools were formulated on the basis of the rate equations written in the simulation language CSMPIII. Computed results were compared with experimental data of phosphate concentrations in solution of the soil suspensions to check the model. Finally short soil columns were assembled in the laboratory and subjected to percolation with a phosphate solution. The phosphate concentrations in the effluent leaving the column were used to test the computed result obtained with the model under conditions of liquid flow. To this purpose the model was extended with equations describing transport of phosphate by convective and diffusion/dispersion flow (cf. Chaper 5). The computed effluent concentration agreed satisfactorily with measured concentrations.Since the cheek with column experiments indicated that the model could be used under conditions of liquid flow, it was applied to predict the effects of longterm additions of soluble (ortho)phosphate to the soil (Chapter 6). As for the present purpose the model required initially a long computing time, special procedures were introduced to reduce the latter to an acceptable level.The reliability of the model for longterm predictions was checked against available field data of the scil treated with sewage water. To this purpose the distribution profile of accumulated (AI+Fe)-bound phosphate found in the field was compared with the computed data obtained after a load of phosphate was supplied identical to the (Al+Fe)-bound P present in the field. This load was equivalent to 4460 kg P/ha; in the program it was supplied in separate portions of 19.6 kg P/ha in the form of a solution containing 9.8 mg P/l. An application regime was sustained consisting of 10 applications per year, as is the normal practice used at the sewage farm. The agreement between the 2 sets of data was good from a practical standpoint, showing that the accumulated phosphate was mainly present in the top 30 cm with very little movement to deeper layers. On a more detailed scale, the field profile was definitely more spread-out than the computed one, very probably because the dispersive properties of the field soil differed from the ones present in the model, where only one dimensional leaching was considered. The downward transport of P appeared to be subcritical with respect to the chosen values of the rate constants and the yearly dose.A partly loaded soil was used to study the effects of different application regimes, yearly loads, different fluxes and reduced values of the rate constants on the downward movement of P. To this purpose, the fraction of added phosphate that moved below a depth of 50 em and the fraction of P lost from the soil via leaching (viz. at a depth of 1 meter) were used as criterions. The total amount of P supplied during these cases was always identical to the additional amount that could be bound in the top 50 em. It could be shown that a variation in yearly load of 200 to 780 kg/ha, added in portions of different size, had only a small effect on the downward transport of P. A variation in the waterflux between 10 and 50 cm/day had the same minor effect on downward transport. A reduction of the rate constants controlling the loading of the pools X 1 and X 2 below 10% of their original values leads to higher leaching losses. A reduction of the rate constant defining the loading of pool N below 50% of the original value significantly increased the downward movement of P.These calculations have shown that an application regime and yearly dose as are in use at the sewage farm studied, guarantee an effective removal of phosphate from solution. Of particular importance is the introduction of a sufficient waiting period between two doses, in order to allow for transformation of the adsorbed P to a less reactive form. This then leads to a regeneration of adsorption sites which is of great importance to remove the phosphate effectively from the sewage water during its relatively short residence time in the soil.Finally the model was used to establish limits with respect to the application regime. These results indicated that the penetration depth of phosphate in the soil could be controlled on the basis of (a) the size of the single dose in relation to the capacity of the fast loading pools X 1 and X 2 and (b) the waiting period between successive doses, provided one does not surpass the maximum retention capacity of the relevant soil layer.