**1. Introduction**

P, is, on the one hand, one of the most important essential nutrients for plant growth, but on the other hand, a finite resource, limiting productivity in agriculture terrestrial ecosystems. Given the scarcity of global phosphorus reserves, and to ensure sustainable soil fertility on agricultural soils, a fundamental understanding of the mechanisms of fixation, recognized as the reduction of solubility of fertilized P in the soil [1], and mobilization of inorganic phosphorus in soils is required. Although both the inorganic and organic P pools contribute to total P availability [2,3], dissolved inorganic P is the only P fraction that can be taken up by plants and microorganisms, thus maintaining ecosystem nutrition and mineralization [4]. Dissolved P has a high affinity for adsorption to the soil matrix, which affects its bioavailability, depending on soil composition and binding motifs. In particular, adsorption, desorption, and precipitation processes on pedogenic mineral surfaces limit its availability, which is why reactions of P with selected hydroxides have been studied in detail in the past.

"Nonspecific" physisorption via electrostatic attraction provides lower binding energy and thus easy mobilization of P by 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 immobilization occurs via precipitation on the particle surface, where especially amorphous Fe-hydroxides play a major role [9–11].

In addition to the fixation of inorganic P on mineral surfaces, the soil organic matter has an important influence on P adsorption and desorption. If both P and organic anions

**Citation:** Gypser, S.; Schütze, E.; Freese, D. Single and Binary Fe- and Al-hydroxides Affect Potential Phosphorus Mobilization and Transfer from Pools of Different Availability. *Soil Syst.* **2021**, *5*, 33. https://doi.org/10.3390/ soilsystems5020033

Academic Editors: Donald S. Ross, Eric O. Young and Deb Jaisi

Received: 9 March 2021 Accepted: 20 May 2021 Published: 21 May 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

are present in the soil solution, the adsorption of P on hydroxide surfaces can be positively influenced by competition for adsorption sites, ligand exchange, and replacement of P by organic anions, dissolution of adsorbents, and changes in the surface charge of the adsorbents. Soil organic matter can also retard the crystal growth of poorly crystalline Fe- and Al-oxides and -hydroxides, which affects their specific P adsorption capacities. Moreover, organic anions can form metal-organic complexes by adsorption on metal ions (e.g., Al3+, Fe3+) [12–17].

P fixation is a global problem for soil fertility and negatively impacts agricultural productivity because recovery by plants in the year of application is often only 10 to 15% and P inputs from fertilizers tend to accumulate in the soil [18]. A meta-analysis of about 2000 datasets from 30 field trials from Germany and Austria showed that yield increase after P application is mainly determined by pH, soil organic matter, fertilizer type, and crop type, whereas plant-available P in the soil seems to be the most important parameter [19]. However, Syers et al. (2008) [20] showed that an irreversible immobilization of fertilized P is not supported by field studies. They divided inorganic soil P into four pools with different availability to plants based on its accessibility to plant roots and extractability using common soil analytical methods. These pools range from soil solution P characterized by immediate accessibility and availability to very low accessible, extractable, and available P, which is very strongly bonded, inaccessible, precipitated, or mineral P. Roberts and Johnston (2015) [19] have summarized that differences in P bioavailability depend on accessibility to plant roots and extractability by soil test reagents. However, when the concept of P transfer within the four pools and partial nutrient balance is considered, P recovery can exceed 70%. In particular, the distribution of P among these different pools leads to the conversion of excess P into very slowly exchangeable P that can only be partially utilized when soil P content is low [21]. These organic and inorganic P pools of different availability were generated due to the weaker bound P in surface complexes or the strongly bound precipitates. Therefore, the fundamental understanding of P binding motifs on contrasting mineral surfaces, and possible changes in binding over time [22–30] allows a more detailed characterization of soils in terms of their potential P fixation capacity, as well as short- and long-term mobilization.

Most methods for the determination of available P are based on the quantification of solubilized P using different extractants, consist of chemical equilibrium-controlled solubility, and release rates-limiting processes. They do not measure the quantity of plantavailable P, but by experimentation, testing, and the application of regression equations, they allow a prediction about a soil P status related to it [31]. However, these observed results are not always applicable to different soil types or arable crops. Due to ad- and desorption processes, dissolution, or mineralization, the pool of plant-available P is strongly time-dependent [20,31]. For this, sequential extraction methods offer an inexpensive and simple approach to determine the amount of long-term mobilized P. The different extractants can be selected according to the objectives and play a role, e.g., during cultivation, in the rhizosphere, or during soil development. While no particulate speciation such as precipitated or re-adsorbed P can be provided without further investigations [1], different soil types or soil components can be investigated concerning the differentiation of potential P mobilization. Based on the empirical assignments, the P status of soils can be characterized according to a concept of different available P pools and their transformation [20,31].

Therefore, this study aims to investigate whether the crystallinity, as well as the Fe/Al ratio of hydroxides, affects the potential mobilization of adsorbed P. The results will hopefully guide as to whether the composition of soil in terms of pedogenic hydroxides affects the moderately to non-labile P reserves. For this purpose, desorption kinetics are created in a batch setup using synthetic Fe- and Al-hydroxides as well as inorganic and organic extraction agents at two concentrations.

The amount of low and very low available inorganic P that can be desorbed by increasing concentrations of organic and inorganic extractants has to be determined, and the time-depended mobilization process will be evaluated.

### **2. Materials and Methods**

#### *2.1. Preparation of the Fe- and Al-Hydroxides*

Goethite and gibbsite were used as model substances of crystalline Fe- and Alhydroxides. In addition to ferrihydrite, these minerals were the main model minerals of previous studies. However, two- or multi-component hydroxide systems are more common in soils, but they have barely been included so far, especially in desorption studies. Therefore, synthesized Fe- and Al-hydroxide mixtures with varying Fe- and Al-ratio represented this binary amorphous fraction appearing in soils. Poorly crystalline ferrihydrite was used as a transitional Fe-hydroxide, bridging between the initial amorphous hydroxide structure and crystalline goethite during pedogenesis.

The synthetic hydroxides investigated in this study were goethite (99%, Alfa Aesar, Haverhill, MA, USA), gibbsite (analytical grade, Merck Millipore, Merck KGaA, Darmstadt, Germany), ferrihydrite (prepared according to [32]), and mixed Fe:Al-hydroxide (prepared according to [33]).

For the preparation of 2-line-ferrihydrite, a 1 M KOH was added to 500 mL of a 0.2 M Fe(NO3)3·9 H2O-solution, until a pH of 7.5 was reached. The developed precipitate was washed with ultrapure water to remove remaining salts, centrifuged for 5 min at 12,134× *g* (Avanti J-25 Centrifuge, Beckman Coulter, Brea, CA, USA), frozen and freeze-dried, and stored in a desiccator.

The Fe:Al-hydroxide mixtures were prepared by mixing 0.1 M Fe(NO3)3·9 H2O and 0.1 M Al(NO3)3·9 H2O in molar ratios of 1:0, 5:1. 1:1, 1:5, and 0:1, and adjusted to pH 6 with 5 M KOH. The solutions were equilibrated for 1 h. Subsequently, the precipitate was washed with ultrapure water, centrifuged for 5 min at 12,134× *g*, dried at 60 ◦C, and ground into a powder. All chemicals used for the preparation were of analytical grade.

Prior to desorption experiments, 20 g of each hydroxide was adjusted to pH 6 in 50 mL ultrapure water with 0.1 M HCl or KOH, respectively, and dried at 40 ◦C for 5 days. P was adsorbed by adding 200 mL of a 0.3 M P solution to 17 g of the dried hydroxides. The P solution consisted of a Na2HPO4/KH2PO4 buffer solution (pH 6) with additional KH2PO4 to achieve the desired P concentration of 0.3 M. Subsequently, the hydroxide-P solution mixtures were shaken horizontally with 200 Motions min−<sup>1</sup> for 24 h. Afterward, they were centrifuged for 15 min at 21,572× *g* until the supernatant was clear.

#### *2.2. Characterization of the Fe- and Al-Hydroxides*

The elemental composition of the hydroxides was verified using SEM-EDX, scanning electron microscopy (DSM 962, Zeiss, Oberkochen, Germany) with energy dispersive X-ray spectroscopy (X-Max 50 mm<sup>2</sup> with INCA, Oxford Instruments, Abingdon, Great Britain). The determination of the crystallization, as well as the poorly crystalline and amorphous structures was performed using X-ray diffraction (Empyrean powder diffractometer, PANalytical, Almelo, Netherlands) (for the results of gibbsite, ferrihydrite and the Fe-Al-hydroxide mixtures see [24], for the results of goethite see [23]).

Specific surface areas were determined in duplicate with an Autosorb-1 (Quantachrome, Odelzhausen, Germany) using a multi-point BET-measurement (Brunauer– Emmett–Teller) and N2 as adsorptive. An outgas test was performed to verify the completed outgas procedure for each hydroxide. The specific surface area was substantially higher for the amorphous Fe-hydroxides and decreased with increasing crystallinity grade, as well as an increasing amount of Al for the mixtures (Table 1). The specific surface area was in the same range for the crystalline and the amorphous Al-hydroxides.


**Table 1.** Point of zero charge (PZC), specific surface area (SSA), and the amount of total Fe, Al, and P of crystalline and amorphous Fe- and Al-hydroxides.

> The point of zero charge (PZC) of each hydroxide was determined using potentiometric titration. 0.5 g of each hydroxide mixture was weighed into 100 mL PE-cups in triplicate. 30 mL of different KCL solutions (0.02, 0.2, and 2 M) were added separately to each sample. The solutions were diluted with ultrapure water to a total volume of 60 mL, leading to final concentrations of 0.01, 0.1, and 1 M KCl, respectively. The hydroxide/KCl mixtures were equilibrated for 4 days at 21 ◦C and shaken for1hd-1 to reach an equilibration pH value prior to the titration procedure. In the beginning, the pH of the suspensions was increased with 1 mL of 5 M KOH, followed by titration with fixed amounts of 1 M HCl. The amount of adsorbed H+ on the hydroxide surface at each pH was determined by subtracting the titration curve of the blank KCl solutions from the titration curve of the suspension. The PZC derived from the titration curves was highest for amorphous hydroxides with a predominant Al- amount with a value of 9.8 (Table 1). The PZC decreased with increasing Fe-amount and was lowest for the pure amorphous Fe-hydroxide with a value of 6.0. Within the group of the crystalline hydroxides, gibbsite, and goethite were in a similar range with PZC values of 8.5 and 8.8, respectively. Ferrihydrite offered a slightly lower value of 7.1.

> The total amount of Fe, Al, and adsorbed P was determined in duplicate by digestion of 0.02 g hydroxide in 50 mL ultrapure water with 1 mL aqua regia. The poorly crystalline ferrihydrite adsorbed initially 30.92 mg g−<sup>1</sup> P, which was 4-fold higher than for goethite or gibbsite (Table 1). The amorphous Fe:Al-hydroxide mixtures had adsorbed P concentrations in the range from 31.34 to 64.65 mg g<sup>−</sup>1, increasing with a predominant amount of Al. Thus, the amorphous hydroxides showed a higher P adsorption than the crystalline hydroxides. Related to the specific surface area, P adsorption values of 80.81 and 40.25 mg m−<sup>2</sup> were obtained for the amorphous Al-hydroxides, and 8.21 mg m−<sup>2</sup> was obtained for gibbsite. P adsorption values in the range from 0.12 to 0.43 mg m−<sup>2</sup> were lower for Fe-hydroxides due to the substantially higher SSA.

#### *2.3. Desorption Experiments*

Desorption experiments were performed in a batch setup at room temperature. The investigations were carried out in quadruplicate using 50 mL PE-centrifuge tubes. Each batch contained a hydroxide-solution mixture, consisting of 0.8 g hydroxide and 40 mL reaction solution, resulting in a solid-solution ratio of 1:50.

KCl and KNO3 were selected as inorganic compounds for desorption experiments since they are an essential part of the soil solution in agricultural soils, and components of mineral fertilizers. Histidine (C6H9N3O2, His) and malic acid (C4H6O5, Mal) were chosen as organic extractants. These reaction solutions were used with concentrations of 5 mM and 50 mM, adjusted to a pH of 6 with 0.1 M KOH, HCl, or HNO3, respectively.

At the beginning of desorption, the samples were shaken horizontally at 200 Motions min−<sup>1</sup> for 24 h, afterward once a week for 1 h. For sample taking, the hydroxide-solution mixtures were centrifuged for 20 min at 21,572× *g*. The clear supernatant was carefully decanted and filtrated using P-poor Whatman 512 1/1 folded filter papers. Afterward, 40 mL of the fresh reaction solution was added to continue desorption. Sample taking was done after 2, 6, 24, 48, 168, and 336 h desorption time. After a desorption time of 2 h, the pH of the sample solution was measured again in two randomly selected samples of each hydroxide for all treatments.

Concentrations of dissolved total P, Fe, and Al were determined by using ICP–AES (Unicam iCAP6000 Duo, Thermo Fisher Scientific, Waltham, MA, USA), total Cl was determined by using ion chromatography (Dionex DX 500 + DX 120, Thermo Fisher Scientific), total N and C were measured with a TOC-Analyzer (TOC-VCPH and TOC5000, Shimadzu, Kyoto, Japan). Repeated washing of all used materials with ultrapure water and immediate freezing of the sample solutions prevented microbial activity.

#### *2.4. Kinetics of P Desorption*

The cumulative P desorption depending on time was calculated, and different linearized kinetic models were applied to the data (Table 2). The aim was to fit the experimental data to an appropriate kinetic model and to analyze the influence of organic and inorganic solutions on desorption kinetics from contrasting Fe- and Al-hydroxides. The coefficients of determination (R2), as well as the standard errors (S.E.) were tested using linear regression analysis to determine their applicability on the kinetics using SigmaPlot 12.0 (Systat Software Inc., San Jose, CA, USA). When not stated otherwise, the *p*-value was <0.05.


