Introduction

The nature of soil P is a problem that has concerned soil scientists for many decades. A puzzling aspect of the problem is that reaction between soil and P does not seem to reach equilibrium, but rather continues at a rate that is proportional to a fractional power of time. This has been measured in terms of the decline in effectiveness of P fertilisers for plant growth (Barrow 1974), and in terms of the declining concentration in the solution (Barrow and Shaw 1975). A further component of the puzzle is that the rate of reaction increases with increasing temperature (Barrow 1974; Barrow and Shaw 1975). This is a component that must be included in any explanation; it also provides a useful experimental tool. By using increased temperatures, the rate of reaction can be increased thereby producing effects in a few days that would take months or years at lower temperatures.

Further, the behaviour of phosphate is not peculiar; many other reactants behave similarly. These include molybdate, fluoride, sulfate, selenate, selenite, zinc, cadmium, nickel, and cobalt. (For references see Barrow (2022).)

Two views about the nature of soil P exist. One can be traced to the work of Hall and Amos (1906). They extracted soil with dilute acid and concluded that “soil contains compounds of phosphoric acid of varying solubility”. This was disputed by Russell and Prescott (1916) who wrote “something more is concerned than a mere mixture of phosphates…”. Nevertheless, the idea that there was a mixture of P compounds persisted and was further developed, for example by Lindsay (1979). Although frequently quoted, there is a serious problem with Lindsay’s work. He dealt with equilibrium and therefore had no means of describing the slow, continuing reaction. According to this view of the nature of soil P, the decreasing effectiveness of P with time is caused by conversion from one compound to another. Such conversions are thought to be reflected by soil P chemical fractionation schemes.

Alternative theories depended on developments in surface chemistry which largely occurred in the 1960s and 1970s. It is central to these theories that the adsorption of P (and other specifically sorbed ions) is strongly dependent on the surface potential of variable charge materials in soil. This potential decreases as the pH increases and also decreases with increasing reaction with (negatively charged) P ions. By taking these effects into account, it is possible to quantitatively describe many aspects of the reaction of specifically sorbed cations and anions with soil (Barrow 1999). However, to explain the continuing reaction, it is necessary to make a further assumption: that after reaction with the surface, the adsorbed ions penetrate the reacting material by solid-state diffusion. The development of this idea is chronicled by Barrow (2022) and evidence is presented there from which it is argued that it is the only one that is consistent with the observations. For example, the activation energy for the forward reaction is similar to that for the back reaction; this would be very unusual for chemical reactions but is characteristic for diffusion. When this assumption is incorporated into models, it quantitatively explains observed effects (Barrow 1999).

However, some prefer to explain the continuing reaction in terms of rearrangement of surface molecules. We cite Frossard et al. (2011) because it is a distillation of many years of work. In it, they conclude that inorganic P ions located on the solid phase of the soil are distributed along a continuum of solubility. Zhang and Selim (2007) proposed that adsorbed ions react with sites that are described as “equilibrium sites”, “kinetic sites”, “consecutive irreversible sites”, and “concurrent irreversible sites”. Their observations are more efficiently described by equations that are consistent with penetration rather than distribution amongst different sites (Barrow 2023). Shi et al. (2008) proposed that for zinc there was a fast reaction associated with monodentate binding sites and a slow reaction associated with bidentate binding sites. Such explanations cannot be comprehensive because they cannot describe the similar behaviour of fluoride, which cannot form bidentate linkages (Barrow 2022). In general, it is difficult to understand how surface reactions could be slow enough to explain the observed behaviour. Others (Penn and Camberato 2019) do not consider the continuing reaction in their review of soil phosphate chemistry.

The proposal that P ions penetrate the reacting surfaces and thereby change their properties is essential to understanding the increased effectiveness of subsequent P applications (Barrow 1999, 2022). In order to provide further evidence, we have exploited a technique pioneered by Strauss et al. (1977). They reacted P with samples of goethite ranging from well-crystallised to poorly-crystallised and treated the samples with 5 M HCl. They found that, for a well-crystallised sample of goethite, most of the P was dissolved when but little of the iron was dissolved, but for a poorly-crystallised sample the P was not fully dissolved until about 60% of the iron was dissolved. They interpreted this, together with other evidence, as showing that P penetrated poorly-crystallised goethite, but mostly remained on the surface of well-crystallised goethite.

Dissolution with acid is certainly not new; it was used by Hall and Amos (1906). The difference is that by simultaneously measuring iron and aluminium dissolution, we thought that much could be learned about the nature of soil phosphate.

In the work reported here, we used this technique, and also dissolution using alkali, to study the effects of period of reaction. We included a treatment with alkali because we wanted to compare the effects of continuous treatment with both acid and alkali with the results obtained from soil fractionation techniques in which a treatment with acid follows a treatment with alkali. The P extracted by these treatments has been interpreted as representing different pools of P.

We studied a sample of goethite, a sample of aluminium oxide, and a soil. We reacted P with these materials at 70 °C in order to accelerate the reaction and produce, within periods measured in days, effects equivalent to those requiring years at lower temperature. Our aim was to test whether changes through time in P, Fe and Al were consistent with P diffusion into the adsorbing particles.

Methods

Preparation of materials

Goethite was prepared using the method described by Strauss et al. (1977) for the sample they described as Goe-132. Aluminium oxide was prepared by hydrolyzing aluminium alkoxide by excess water. Details of the methods used are given in the supplementary material. In both cases the materials used were the same as those used by Barrow et al. (2020a). Powder XRD plots showed that the goethite was single-phase. For the aluminium oxide, all peaks with significant intensity matched the cubic form. Details of the methods used are given in Barrow et al. (2020a) and are also provided as supplementary material. .

The soil used

We used a soil which we refer to as Jhargram. It had a low P status; it was collected from uncultivated land from the Regional Research Station of Bidhan Chandra Krishi Viswavidyalaya (22° 26′ 58.99″ N, 86° 59′ 50.64″ E) and was used in Barrow et al. (2020a). It was collected in the dry season in January, 2020 and air dried. It is classified as an Ultic Haplustalf (USDA Soil Staff Survey 1999). Some properties are: pH(1:2.5), 4.10; cation exchange capacity 8.91 (cmol (p+)kg−1); sand, 62%; silt, 23%; clay, 15%. Further properties are given as Supplemental Table S1.

Reaction with P

We loaded samples of goethite and of aluminium oxide with P by a process similar to that of Strauss et al. (1977). We mixed 3 g of the samples at a pH close to 4.0 with 150 mL of a 0.01 M NaNO3 solution containing P as KH2PO4 equivalent to 1250 mg kg−1. For the goethite, this was about 0.3 μmol m−2. The maximum P sorption for a similar sample of goethite was about 8 μmol m−2 (Strauss et al. 1977). The suspensions were incubated at 70 °C for 3 h, 1, 3, 10 and 30 days with vigorous shaking once a day. Previous studies (Barrow et al. 2020b) showed the P molecules were not present as precipitates. After incubation, the suspension was separated by centrifugation; 0.5 mL remained entrapped. The P in the solution was measured. We allowed for the P in the entrapped solution, and calculated the P retained by goethite and by the aluminium oxide. Similarly, a sample of Jhargram soil was reacted with 1250 mg kg−1 P for 1, 3, 10, and 30 days. The samples were dried at 40 °C for 24 h before dissolution studies. All results are expressed as fractions of the P retained.

Dissolution with acid and alkali

We gently shook one set of twelve 45 mg samples with 30 mL of 5 M HCl and a second set of twelve 45 mg samples with 0.1 M NaOH. The temperature was 25 °C. The two sets of twelve samples were analysed after the following periods: 1, 2, 5, 10, 15, 20, 30, 45, 60, 120, 180, and 240 min. The acid concentration used was the same as that of Strauss et al. (1997). These concentrations were chosen because they were found to give rates of dissolution that could be followed within a convenient period. We measured the iron and aluminium concentrations in the solution using Atomic Absorption Spectrophotometer (PerkinElmer).

Phosphate determination

In all cases P concentration was measured following the method of Murphy and Riley (1962).

Statistics

Observations were described using various forms of the rescaled Mitscherlich equation of Barrow and Mendoza (1990). The statistical properties of this equation were discussed by Ratkowsky (1983) who attributed it to Weibull (1951). It was also previously used by Strauss et al. (1977). The general form of the equation is:

$$Y=\mathrm{a}\hbox{--} b\ \exp \left(-{cx}^n\right)$$
(1)

where Y is the observation being described (P, iron, or aluminium dissolved as appropriate) and x is the independent variable (time, or iron or aluminium dissolved again as appropriate).

When n = 1, the equation is the original Mitscherlich equation; when n > 1 a sigmoid relationship results; when n < 1 a relationship which is less steep than the original Mitscherlich equation results. When Y represents a fraction with a maximum value of unity, a = 1.

The equations were fitted using SigmaPlot 10 (Systat Software, Inc.).

Results

In all cases, the concentrations of HCl and of NaOH used were effective in dissolving a large amounts of iron, aluminium and P (Figs. 1, 2 and 4) within convenient experimental periods ranging up to 4 hours. Further, the re-scaled Mitscherlich equation (Eq. (1) were able to closely describe a large range of responses shapes (Figs. 1, 2 and 4).

Fig. 1
figure 1

Effect of the indicated periods of incubation at 70 °C on the dissolution of P and of the relevant metals by 5 M HCl (a, b) and by 0.5 M NaOH (c, d). P had been added as KH2PO4 at 1250 mg kg−1 of goethite and of aluminium oxide prior to incubation. The times of dissolution range from 1 to 240 minutes. The data are described by equations of the form: P = 1 – b exp(−cxn). where P is the P dissolved, and x the Fe or Al dissolved, as appropriate. For part a, values of n ranged from 1.96–2.99. For part b they were all 1. For part c, they were all 2. For part d they were all 1

Full size image
Fig. 2
figure 2

Effect of the indicated periods of incubation at 70 °C on the dissolution of P and of Fe and Al from soil by 5 M HCl (a, b) and by 0.5 M NaOH (c, d). P had been added as KH2PO4 at 1250 mg kg−1 of soil prior to incubation. The times of dissolution range from 1 to 240 minutes. The data are described by equations of the form: P = 1 – b exp(−cxn). where P is the P dissolved, and x the Fe or Al dissolved, as appropriate. For part a, values of n ranged from 3.1–4.6. For part b they were all 1. For part c, they ranged from 2.2–4.4. For part d they were all 1

Full size image

Dissolution of phosphates

After 1 d incubation at 70 °C, goethite had sorbed 95% of the added P. This increased to 99.9% after 30 d. For the aluminium oxide, the analogous values were 74 and 99.6%. For goethite, with increasing periods of reaction with P, more iron had to be dissolved for any given amount of P dissolved (Fig. 1). Further, the relationships between P dissolved and Fe dissolved were clearly sigmoid. This is accommodated by values of n of the re-scaled Mitscherlich equations (Eq. 1) that were > 1. Similarly for the aluminium oxide, with increasing periods of the reaction with P, more Al had to be dissolved for any given amount of P dissolved (Fig. 1). In this case there was no evidence that the responses were sigmoid.

For the soil, the relations between P dissolved and Fe/Al dissolved were remarkably similar to those for the pure compounds (Fig. 2). As for the pure compounds, with increasing periods of incubation, more Fe and Al was dissolved for any given fraction of P dissolved. Further, the relations between P dissolved and Fe dissolved were also sigmoid; indeed the sigmoid nature was even more marked.

Reaction with P also decreased the rate at which Fe dissolved from goethite and the rate at which Al dissolved from the aluminium oxide (Fig. 3). This effect was small when dissolution was by acid but large when dissolution was by alkali.

Fig. 3
figure 3

Effect of incubation for the indicate periods at 70 °C on the rate of dissolution of Fe and of Al by 5 M HCl (a, b) and by 0.5 M NaOH (c, d). P had been added as KH2PO4 at 1250 mg kg−1 of goethite and of aluminium oxide prior to incubation. The data are described by equations of the form: M = a – b exp(−ct). where M is the fraction of the metal dissolved, and t the period of treatment with acid or alkali

Full size image

For the soil, the rate at which Fe and Al dissolved also decreased with increasing period of prior incubation with P (Fig. 4). The effect differed somewhat from those observed with pure compounds in that the effects were well marked for the acid dissolution treatments.

Fig. 4
figure 4

Effect of incubation for the indicated periods at 70 °C on the rate of dissolution of Fe and of Al by 5 M HCl (a, b) and by 0.5 M NaOH (c, d) from treated soil. P had been added as KH2PO4 at 1250 mg kg−1 of soil prior to incubation. The data are described by equations of the form: M = a – b exp(−ctn). where M is the amount of the metal dissolved, and t the period of treatment with acid or alkali. Values of n were: a, 0.15–0.30; b, 0.5–0.99; c, 0.60–0.99; d, 0.66–1.5

Full size image

The effects on rate of dissolution were a consequence of the addition of P; they did not occur when the soil was incubated at the same temperature without added P (Fig. 5).

Fig. 5
figure 5

Effect of period of incubation in the absence of added P on the dissolution by of Fe and Al by 5 M HCl from the soil

Full size image

Disscussion

If inorganic P ions were located on the solid phase of the soil, it should not require the dissolution of large amounts of Fe and Al before the phosphate was dissolved. We observed that, not only were large amounts required, but the amounts required to dissolve a given amount of P increased with time. This occurred for three different materials, goethite, aluminium oxide and a soil. It is consistent with the explanation that, with increasing period of incubation, P had diffused into the material on which it was initially adsorbed.

The curves we used to describe these effects were smooth; there was no indication that there was anything but one process involved. That is, there was no evidence to support the theory that phosphate was present as discrete compounds. This was called the precipitate-particulate theory by Barrow (2020). This paper presented arguments against this theory.

The results were consistent with the conclusion that the acid and the alkali treatments reacted with the same material. This differs from the interpretation of fractionation procedures in which alkali and acid treatments are applied consecutively and the P dissolved is interpreted as coming from different fractions. This idea was introduced by Chang and Jackson (1957). In their method, P extracted with 0.1 M NaOH is interpreted as iron phosphate and P extracted with a subsequent treatment with 0.25 M H2SO4 is interpreted as calcium phosphate. Although the designations in the individual methods differ, it is a common theme of subsequently published methods that the P extracted by the different reagents is in different forms. For example, the method of Chen et al. (2000) is a modification of the method of Hedley et al. (1982). It includes a treatment with 0.1 M NaOH. This is interpreted as dissolving P held strongly by chemisorption to iron and aluminium components of soil surfaces. A subsequent treatment with 1 M HCl is interpreted as extracting P mainly from apatite type minerals along with the occluded form in highly weathered soil. Our results are not consistent with such interpretations.

For goethite and for soil, the relationships betwen the Fe dissolved and the P dissolved were distinctly sigmoid. A similar observation was made by Strauss et al. (1977) for goethite. They explained it by suggesting that the material between the domains dissolved quickly, giving rise to particles with a larger total surface area; subsequent dissolution was therefore quicker.

In general, the dissolution behaviour for soil was similar to that of the phosphated pure compounds. This supports the idea that these compounds were good models at least for the soil used. It is also relevant that the technique of increasing the temperature to accelerate the reaction had similar effects on soil and on the inorganic compounds. We think this means that any effects of high temperature on organic compounds in soil had little influence. We have used increasing temperatures to accelerate the reaction many times (Barrow 2022) since it was observed (Barrow 1974). It is logically impossible to prove that the effects are identical to those obtained at lower temperature and longer time. It is only possible to record that we have seen no evidence that they are different.

The rate of dissolution of the phosphated pure compounds decreased with time. That is, they became more stable. Strauss et al. (1977) showed that P did not penetrate into the crystal lattice of goethite but rather into the spaces between crystal domains. We interpret the increased stability as indicating that, as the P penetrated further, the domains became more tightly bound to each other. For soil, the rate of dissolution of both iron and aluminium also decreased with time. This might be interpreted as indicating that P had acted with both iron and aluminium compounds in soil rather than with one or the other. Alternatively it might be interpreted as indicating that P had reacted with mixed oxides, that is, oxides that contain both Fe and Al atoms.

Conclusions

We suggest that our results using dissolution, either with acid or with alkali, show that, with increasing time, P penetrates further into the substrate rather than changing from one pool to another. We also suggest that dissolution, combined with measurement of iron and or aluminium, is a better way to follow time trends than chemical fractionation schemes.