**4. Discussion**

#### *4.1. Influence of Crystallinity and Fe/Al Content on P Sorption*

During the preparation of the desorption experiments, it was already shown that the different hydroxides had different reactive surface areas and hence, individual P adsorption capacities. It is well known that P adsorption on Fe- and Al-hydroxides occurs via inner-sphere complex formation [15,25,26,28,29,34], but also by surface precipitation [24,29,35,36], at which the crystallinity grade of the hydroxides played an important role. For goethite, several studies described the formation of either monodentate innersphere complexes [36,37] or bidentate complexes [23,38] with a minor fraction of monodentate complexes [23] in the intermediate pH range. Li and Stanforth (2000) [10] observed a more negative surface charge of goethite due to the replacement of surface OH groups by protonated and more acidic P anions, led to further decreasing P adsorption. For gibbsite, the formation of simultaneously existent monodentate and bidentate surface complexes with hydrogen-bonding to outer-sphere complexes was concluded [28,29]. In addition, the formed inner-sphere complex was described as a precursor for Al-P precipitation [28,29,35], decreasing with increasing pH due to the increased solubility of Al-phosphates at pH values around 6 [35].

Studies on poorly crystalline ferrihydrite divided the process of P adsorption into the formation of a monodentate [15,26] or bidentate inner-sphere complexes [22,25], the migration of P to surface sorption sites of decreasing accessibility within the particles [39], and with longer equilibration time also the formation of stable Fe-P precipitate [24,36]. For the amorphous Fe-hydroxide, the preferred formation of bidentate surface complexes as well as the formation of Fe-P precipitate with increasing P concentration and equilibration time was described, whereas for the amorphous Al-hydroxides prevalent monodentate inner-sphere complexes were reasoned. In the hydroxide mixtures, the Fe content is particularly contributed to a stable P fixation by precipitation reactions [24], whereby also P bindings via inner-sphere complexes were formed [40].

Summarized, the higher accessibility of both surface and structural binding sites of amorphous hydroxides led to a higher amount of adsorbed and precipitated P compared to well crystalline hydroxides, underlines the important role of amorphous Al and Fe fractions for the release of labile P in soils [41]. The more rigid, but poorly crystalline character allowed the migration of P into mineral particles, which also enabled a stable and effective P adsorption related to the specific surface area [24]. The decrease of initially P adsorption on the pure amorphous Al-hydroxide compared to the hydroxide mixtures with predominant Al content can, thus, also be attributed to a transitional phase with greater crystallization observed in the sample [24,34], which was already indicated by the XRD measurements in the present study. It is therefore possible that the release of adsorbed P from the inner mineral particle surface by means of anion exchange, in particular of more complex organic anions, is sterically inhibited by the initial crystallization.

As a consequence, the different binding motifs also have an impact on P mobilization. "Nonspecific" physisorption via electrostatic attraction provides lower binding energy and thus easy mobilization of P via ion exchange [5,6] while more "specific" chemisorption results in stronger binding at the particle surface and lower availability of P over time [7,8]. However, the most stable and long-lasting P fixation occurs via precipitation on the particle surface [9,10]. The preferred formation of Fe-P precipitates during adsorption on ferrihydrite and the amorphous hydroxides with predominant Fe amount led to a more stable P binding, and hence, lower desorption capacities than the crystalline hydroxides. With increasing Al content, the influence of the surface complexes also increased, which is why the desorption capacity increased, in this study independently from the type of extracting agent.

#### *4.2. Effect of PZC and Electrolytes on P Adsorption*

In addition to crystallinity or Fe/Al ratio, surface properties of the hydroxides also play a major role regarding exchange reactions. Besides the specific reactive surface area, a varying PZC influences sorption processes. The reported PZC from literature for goethite varies from 6.4 to 9.7 [10,42,43], are in the range from 7.5 to 11.3 for gibbsite [44–47], and between 7.6 and 8.0 for ferrihydrite [46,48]. Sujana et al. (2009) [33] reported PZC values of the amorphous Fe- and Al-hydroxide mixtures in a range from 4.8 to 6.1, increasing with an increasing amount of Al. In the present study, the measured PZC values of the amorphous Fe- and Al-hydroxides were significantly higher and in a range from 6.0 to 9.8, also increasing with increasing Al-amount. As the experimental pH was set to a value of 6, the positive net charge and thus a positive electric potential below the PZC values led to a charge surplus, and hence, stronger adsorption of anions. If the pH of the surrounding solution will be increased, the positive potential decreases and becomes negative at pH values above the PZC, constraining further specific adsorption.

But simultaneously with P adsorption, also sorption of protons (H+) takes place [49]. Along with the P adsorption process, a higher surface charge was produced by diffusion of protons from and to the hydroxide surface [43,50]. Similar was observed for background electrolyte solutions such as NaCl, NaNO3, or KNO3. The presence and concentration of electrolytes in the reaction solution can lead to a decrease of the positive electric potential and hence, a weaker P adsorption at low pH (<4) [42,51]. Certainly, the adsorption of electrolytes or P will be mutually affected. Higher adsorption of cations can be supported by a higher surface coverage of the hydroxide with negatively charged P [51]. Although the PZC changes during P adsorption, the amount of bound P can be affected, depending on the surrounding pH. In combination with the different binding motifs, depending on the crystallinity and the Fe-amount of the hydroxides, the later release of P can be influenced.

#### *4.3. Desorption Kinetics*

Similar to the observations for P adsorption kinetics, desorption showed a biphasic behavior with a first rapid and a second slower stage, which was described in previous studies for Fe- and Al-hydroxides by applying organic and inorganic reagents [11,26,52]. A great P release within the first 24 h of desorption time was observed for all investigated hydroxides, independent of their degree of crystallization or the Fe/Al ratio. However, compared to the crystalline hydroxides, ferrihydrite, and the amorphous Fe- and Al-hydroxide mixtures showed in general lower release rates and a continuous P mobilization. This was shown in both the P desorption kinetic curves and the calculated kinetic parameters. This typical time-dependent trend can be attributed to the different P binding mechanisms of the crystalline and amorphous hydroxides [11,24,53] and therefore, an easier release of P from weaker outer-sphere bindings [54] and nonspecific adsorbed P from low-affinity sites, followed by a slower release of specific adsorbed P from high-affinity sites as well as diffusion of structural bound P [55]. Thus, during the adsorption process, related to the specific surface, comparatively more P was bound to goethite or gibbsite than to ferrihydrite, but this P can be released again in the short-term. Meanwhile, the binding to amorphous

hydroxides mainly contributes to a long-term release and a distinct ongoing mobilization over time. The better fit of the Elovich equation suggested that P was desorbed by chemisorption reactions, which was corroborated by previous studies using soil [53,56,57]. For ferrihydrite, in particular, the better applicability of the exponential function suggested that P mobilization is slow at first and stronger with time. This would be in good agreement with an inner-particulate P binding, which precedes the actual desorption with a diffusion phase and the migration of inner-particle bound P.

But independently of the desorption reactions, the high calculated values of both α (goethite and gibbsite) and β (ferrihydrite and amorphous hydroxide mixtures with predominant Fe) do not seem realistic compared to the measured values. For example, a cumulative P desorption of 3.44 ± 0.22 mg g−<sup>1</sup> after 336 h desorption time was measured for ferrihydrite. This equals 0.01 ± 0.00 mg m−2, while the calculated P release constant over time β for the KCl treatment in the Elovich model amounted 392.88 mg min−<sup>1</sup> m−2. Since an empirical model like the Elovich equation describes processes in an ideal system, an interpretation of the values can be difficult or misleading. An application to less ideal or even natural systems such as soils can therefore be problematic [58]. An increase or decrease of the fit parameters can display a change of reaction rates, whereas the slope of the function depends more on the reaction conditions than on their characterization. Therefore, it is possible to overestimate the initial or mid-term release due to either sharp or weak curvature of the P desorption kinetics [58].

#### *4.4. Inorganic Extracting Agents*

A further aspect of P release is the mechanism of action and, thus, the efficiency of both organic and inorganic extraction agents. Concerning inorganic anions, ion exchange reactions are the main mechanisms during nutrient mobilization in general and P desorption in particular [1,59]. While hydrated monovalent ions such as K+, Cl<sup>−</sup>, or NO3 − usually form weak non-spherical complexes on oppositely charged surfaces, P can be attached via ligand exchange to the hydroxide surfaces [6]. However, the anion exchange of adsorbed P by Cl− and NO3 − had still a clear effect on the crystalline hydroxides goethite and gibbsite (from 40 to 81% desorption capacity), but its effectiveness was lower for ferrihydrite and the amorphous Fe- and Al-hydroxides.

The successively decreasing desorption and the incomplete release of P, which varies depending on the hydroxide, can be explained, among other factors, by the change of anion exchange sites as a result of P adsorption, which was partly irreversible with respect to adsorption of NO3 − and Cl− [34]. During P desorption using KCl and KNO3, the concentration of total Cl and N decreased with increasing P release, whereby the effect was more pronounced for KNO3. This gives an indication of anion exchange reactions, where the anion concentration can vary greatly depending on the extracting agent and investigated hydroxide. However, in the absence of a clear trend of the anion concentration change in the reaction solution, mainly equilibrium reactions between the solid hydroxide surface and the reaction solution took place.

#### *4.5. Organic Extracting Agents*

If both P and organic anions were present, the adsorption of P can already be affected by the competition for adsorption sites, dissolution of adsorbents, change of the adsorbents surface charge, the formation of new adsorption sites by formation of metal-organic complexes through adsorption of metal ions (e.g., Al3+, Fe3+), as well as the retardation of crystal growth of poorly crystalline Fe- and Al-oxides and hydroxides [14–16,60,61]. If P was already adsorbed and hence, fixed, the further release can be controlled by the dissolution of low soluble minerals, ligand exchange and the replacement P by organic anions, as well as the formation of metal-organic complexes and thus the blocking of adsorption sites [12,17,62–64].

The measurement of total C showed no clear trend supporting concentration changes of organic anions during desorption using His and Mal, even though a stronger decrease was correlated for the 50 mM treatments. It can also be assumed that, in addition to anion exchange, equilibrium reactions took place. If the changes of the Fe and Al concentrations in the reaction solution were considered, in particular, more Fe and Al dissolved from the amorphous hydroxides with increasing desorption. Due to the higher Fe and Al concentrations at the beginning of desorption as well as the low release from the crystalline hydroxides, dissolution of the hydroxides by organic reactants, in particular, was assumed to be improbable. Moreover, the use of the organic reagents His and Mal showed a distinctive higher release from crystalline than from amorphous hydroxides. Certainly, the amount of released P was similar between organic and inorganic extracting agents, and, therefore, a clear beneficial influence of the organic compounds was not detected.

Basak (2019) [13] reported an average amount of P released from rock phosphates by organic acids in the range from 0.015 to 83.5% after 6 d reaction time, increasing with an increasing concentration of the acids. The effectiveness of the used acids followed the order: oxalic acid > citric acid > tartaric acid > formic acid > malic acid > succinic acid > acetic acid. However, Basak (2019) [13] also described a decreasing P release for the increase of the organic acid concentration from 0.3 to 0.5 M. Xu et al. (2004) [65] demonstrated that citric acid has the highest capacity to solubilize P from rock and iron phosphates after 24 h reaction time and that an increase in the concentration of organic acids enhanced significantly P solubilization (except oxalic acid). For P release from Fe phosphates, the effectiveness of the organic acids followed the order: citric acid > oxalic acid > malonic acid > tartaric acid > malic acid > acetic acid (not complete). Especially for citric acid, the amount of released P increased from 13.7 to 67.0 mg g−<sup>1</sup> with an increasing acid concentration from 0.001 to 0.01 M. Wang et al. (2015) [53] studied P release from acidic, neutral, and calcareous soils using low molecular weight organic acids. They reported a high efficiency of oxalic acid (0.6 to 3.2 mg kg<sup>−</sup>1); followed by citric acid (0.6 to 2.8 mg kg−1) on cumulative released organic P, regardless form the soil type. Concerning inorganic P fractions, oxalic acid was more effective on calcareous soils, while citric acid had the highest amounts of P release from neutral and acidic soils.

While the enhanced release of inorganic P was likely due to accelerating desorption and dissolution processes, the acid strength contributed to the release of organic P. Gypser et al. (2019) [11] showed a clear lower effect of inorganic constituents than organic acids on P release from Fe- and Al-hydroxides using a concentration of 0.01 M at pH 6 over 1344 h reaction time. While P desorption using CaCl2 and CaSO4 amounted between 0.0 and 57.4%, humic acid showed a desorption capacity in the range from 0.3 to 87.2%. Citric acid had the highest P release in the range from 6.7 to 90.5%. Moreover, desorption increased with increasing crystallinity grade and Al content of the hydroxides. However, the reaction time is a crucial factor in P release, also with respect to P that was strongly bound on amorphous hydroxides. Thus, desorption reactions were still detected after an experimental reaction time of 8 weeks [11]. Taghipour and Jalali (2013) [66] also reported a lower efficiency of malic acid in comparison to citric and oxalic acid for calcareous soils. In particular, the chemical structure, type, and location of the functional groups of the ligands of organic acids influence the efficiency of P mobilization, where di- and tri-carboxylic acids were more effective during mobilization than mono-carboxylic acids [63]. In terms of carboxylation, the P release capacity of Mal (di-carboxylic) is expected to be higher than that of His (mono-carboxylic). In the present study, the release capacities of both compounds were equal for most hydroxides or only slightly higher for Mal. Therefore, the effect related to the number of carboxylic groups may have been relativized by at least one other mechanism.

Although the above-mentioned previous studies have shown that organic compounds support P release, this effect could not be observed in comparison with the inorganic compounds. One possible reason can be organic molecules acting as P adsorbing surfaces in some circumstances [67], and form loosely surface-bound complexes with already released P. Especially His has its isoelectric point in a neutral pH range (7.47 [68]) and can act both as a proton donor as well as an acceptor. In addition, it has a simple aromatic ring and is therefore considered more stable than Mal [69]. This is relevant, considering that P could be adsorbed on His and thus remained in soils, but can be released over a longer time. Another reason, which is essentially also related to the complex structure of organic molecules, is the formation of a "physical barrier" on the mineral surface and hence, limiting P desorption [65,70].

As mentioned above, several processes can be considered for the release of initially adsorbed P by organic constituents. These mechanisms can take place separately or in combination. A mechanism that can take place both when P is adsorbed and when P and organic anions occur simultaneously in the solution, is the dissolution of the adsorbent, which is responsible for a very effective P release. In particular, the dissolution reveals a clear relationship between released P and major components of the adsorbent [71]. In the present study, the dissolved concentrations of Fe and Al in comparison to the initial Fe and Al contents of the hydroxides (see Table 2) were too low to indicate the dissolution of the hydroxides during P release. Furthermore, CTotal fluctuated around the initial concentration of CTotal during the experiments with 5 mM His and Mal, while CTotal decreased using 50 mM His and Mal. The concentrations of Fe and Al slightly increased at the beginning of P desorption and fluctuated around zero during ongoing P mobilization, which rather indicated an equilibrium reaction between solid and liquid phase than the formation of metal-organic complexes in the reaction solution. Thus, ligand exchange and the replacement of P by organic anions were concluded to be the predominant mechanisms of P release by His and Mal. At the same time, the sorption of anions can induce adsorption of H+ and thus, explain the increasing pH in the reaction solution.

Initially, it was expected that a higher concentration of the respective reactant would also increase P release. In the present study, the opposite was observed. A possible explanation could be an increasing P co-adsorption of already desorbed P, and the formation of outer-sphere complexes by electrostatic interaction. Due to the addition of negative charge by adsorption of organic anions, the electrostatic repulsive force decreased and induced H<sup>+</sup> adsorption as mentioned above. This more positive surface charge influenced the sorption behavior of P on the mineral surface. The similar was observed for Ca2+ [42,72], and Na+ [30] as background electrolytes. Duputel et al. (2013) [73] reported a decrease of available P for citrate concentrations below 20 μM due to large adsorption of citrate, enhancing Ca2+ adsorption and facilitate P binding through Ca-bridging. Hence, it can be assumed that a purely additive effect is invalid or limited to a small range [34].

#### **5. Conclusions**

The Fe- and Al-hydroxides showed different capacities to retain inorganic P, depending on the crystallinity of the hydroxides and, thus, the specific P binding motifs, govern the extent and strength of desorption.

The poorly crystalline and amorphous hydroxides especially contribute to a stable fixation of P. In addition, the proportion of Fe and Al plays a considerable role. Precipitation of poorly soluble Fe-phosphates inhibits or prevents effective short- to medium-term P mobilization from amorphous Fe-hydroxides. Al-hydroxides adsorb more P related to the specific reactive surface area than Fe-hydroxides, but they also show a substantially greater P release over time. Certainly, the release of adsorbed P from the inner mineral particle surface through complex organic anions can be sterically inhibited by an initial crystallization process.

In the present study, desorption using His and Mal might not be expected to substantially influence P desorption compared to selected inorganic constituents of the soil solution. An increase in concentration tends to have a detrimental effect on P release as well. It was suggested that organic molecules act as P adsorbing surfaces, which is relevant, considering that P could be adsorbed on His, but can be released over a longer time. Another reason, which is essentially also related to the complex structure of organic molecules, is the formation of a "physical barrier" on the mineral surface and, hence, limiting P desorption.

The assumption that the majority of P fertilization is stably bound to components of the soil and thus permanently unavailable to plants can therefore not be supported. The implication of these results for a sustainable P management in agricultural soils is that the consideration of the reactions of the P ions with soil particles should be included in balancing of potential plant-available P. There will be a transfer of inorganic P from very low/low available P pools (amorphous hydroxides) and low/readily available P pools (crystalline Fe-hydroxides, Al-hydroxides), but with time as one decisive factor. Therefore, a combination of extraction methods should be chosen for a comprehensive characterization of the current soil P status, and a prediction of the potential recovery of P reserves. Thus, at least the questions of the right rate and time of fertilizer application can be taken into account to supply plants with previously unused P reserves, and to reduce fertilizer or to use them more efficiently.

However, P adsorption and desorption will vary greatly in natural systems compared to purely artificial systems, as various physico–chemical properties significantly limit the assumption of pure additive effects. For an accurate calculation, a transfer from the lab to the field and the consideration of further factors, such as soil organic matter, pH, or crop type is necessary. The purpose should be the establishment of advanced methods and protocols to evaluate these implications and to link soil research with agronomic implementations.

**Author Contributions:** Conceptualization, E.S., S.G. and D.F.; methodology, E.S., S.G. and D.F.; investigation, E.S. and S.G.; characterization and chemical analysis, batch experiments, E.S. and S.G.; batch experiments, E.S.; kinetic modeling, S.G.; validation, S.G.; resources, S.G. and D.F.; writing original draft preparation, S.G.; writing—review and editing, S.G. and D.F.; visualization, S.G.; supervision, D.F.; project administration, D.F.; funding acquisition, D.F. and S.G. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the German Federal Ministry of Education and Research (BMBF) in the BonaRes project InnoSoilPhos (No. 031B509C).

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Soil Systems* Editorial Office E-mail: soilsystems@mdpi.com www.mdpi.com/journal/soilsystems

MDPI St. Alban-Anlage 66 4052 Basel Switzerland

Tel: +41 61 683 77 34 Fax: +41 61 302 89 18

www.mdpi.com

ISBN 978-3-0365-2433-7