Describing Phosphorus Sorption Processes on Volcanic Soil in the Presence of Copper or Silver Engineered Nanoparticles

: Engineered nanoparticles (ENPs) present in consumer products are being released into the agricultural systems. There is little information about the direct effect of ENPs on phosphorus (P) availability, which is an essential nutrient for crop growth naturally occurring in agricultural soils. The present study examined the effect of 1, 3, and 5% doses of Cu 0 or Ag 0 ENPs stabilized with L-ascorbic acid (suspension pH 2–3) on P ad- and desorption in an agricultural Andisol with total organic matter (T-OM) and with partial removal of organic matter (R-OM) by performing batch experiments. Our results showed that the adsorption kinetics data of H 2 PO 4 − on T-OM and R-OM soil samples with and without ENPs were adequately described by the pseudo-second-order (PSO) and Elovich models. The adsorption isotherm data of H 2 PO 4 − from T-OM and R-OM soil samples following ENPs addition were better ﬁtted by the Langmuir model than the Freundlich model. When the Cu 0 or Ag 0 ENPs doses were increased, the pH value decreased and H 2 PO 4 − adsorption increased on T-OM and R-OM. The H 2 PO 4 − desorption (%) was lower with Cu 0 ENPs than Ag 0 ENPs. Overall, the incorporation of ENPs into Andisols generated an increase in P retention, which may affect agricultural crop production.


Introduction
In the past decade, the incorporation of engineered nanoparticles (ENPs) into consumer products [1,2] has led to a significant increase in their turnover from $250 billion in 2009 to $3 trillion in 2020 [3]. Two of the most widely used ENPs in consumer products are metallic copper (Cu 0 ) and silver (Ag 0 ), due to their antibacterial properties. Cu 0 ENPs are added to biocides, electronics, paints, cosmetics, agrochemicals, ceramics, and film [1,3,4], whereas Ag 0 ENPs are used in textiles, air filters, bandages, paints, food storage containers, agrochemicals, deodorants, toothpaste, and household appliances [5]. Thus, as a consequence of extensive and diverse commercial applications, these ENPs can be released into the environment. Soil is the main sink of disposal for most of the released ENPs [6]. Adverse effects on human health and ecosystems may be expected, making it necessary to improve our current understanding of environmental risks, fate, transformations and aggregation behaviors of metallic ENPs [7].
The geochemistry of metallic Cu 0 and Ag 0 ENPs in soils is complex, due to their chemical transformation between Cu 0 , Cu + and Cu 2+ as well as between Ag 0 and Ag + , respectively [1,4], also due to their strong binding capacity to various soil components like clay minerals, organic matter, microorganisms, among others. Transformations of metallic ENPs in soil include oxidation, dissolution, and sulfidation. Over time, Cu 0 ENPs can be oxidized in the soil to form CuO (tenorite) and Cu 2 O (cuprite) nanoparticles with a core-shell structure. Any of these, both forms of copper oxide nanoparticles, can dissolve and release cuprous and/or cupric ions into solution [8]. Meanwhile, the Ag 0 ENPs show a slow oxidation process, which can be promoted in acid soils. The metallic ENP oxidation in soils can be diminished when organic molecules are used as stabilizing agents [9]. Transformation on metallic ENPs is an important consideration to developing risk assessments of ENPs [4,9].
Several studies have intended to determine the effects caused by ENPs on soil properties. In these studies, it has been shown than due to metallic Cu 0 and Ag 0 , ENPs are characterized by a high surface area and chemical reactivity, variable surface charge and chemical transformation [10]. Once in contact with soil, ENPs may therefore modify their structural and physico-chemical properties such as pH, electric conductivity, redox potential, porosity, and hydraulic conductivity [10][11][12]. This could affect reactions and processes of elements in soil, such as precipitation, dissolution, co-precipitation, complexation, oxidation/reduction, plant uptake, and ad-and desorption. Particularly, ad-and desorption are important because they control the availability and mobility of contaminants and nutrients [10]. In this context, Taghipour and Jalali [13] reported that metal oxide ENPs (Al 2 O 3 and TiO 2 ) caused immobilization of phosphorus (P) in calcareous soils from Hamadan, Western Iran, and reduced the bioavailability of P.
In relation to effects caused by ENPs in volcanic soils, no studies have assessed the influence of metallic ENPs on the adsorption of nutrients. In this context, the aim of this research was to evaluate the effect of Cu 0 or Ag 0 ENPs on phosphorus sorption processes in volcanic soils and its relationship with organic matter content. Overall, the results provide new information about the implication of ENPs for nutrient availability in soils.

Synthesis of Cu 0 and Ag 0 ENPs
CuCl 2 ·2H 2 O and AgNO 3 were used for the formation of Cu 0 , and Ag 0 ENPs, respectively, and L-ascorbic acid was added as a reducing and capping agent [28]. Cu 0 ENPs (or Ag 0 ENPs) was synthesized by mixing 10.0 mmol·L −1 CuCl 2 ·2H 2 O (or 10.0 mmol·L −1 AgNO 3 ) in 50 mL double-distilled water. An Erlenmeyer flask (100 mL), containing the CuCl 2 ·2H 2 O (or AgNO 3 ) solution, was heated in a water bath at 80 • C with magnetic stirring; 50 mL of L-ascorbic acid (1.0 mol·L −1 ) was added dropwise into the flask while stirring. The aqueous dispersion of stabilized Cu 0 ENPs (or Ag 0 ENPs) obtained was kept at 80 • C for 24 h and it was finally saved to ambient conditions for later research.

Soil Samples
The soil used was an Andisol belonging to Santa Bárbara series from Southern Chile (36 • 50 S; 71 • 55 W). The soil was collected from the top 20 cm depth of the soil horizon. The soil was passed through a <2 mm mesh sieve and freeze-dried (total organic matter soil sample = T-OM). For partial removal of organic matter (OM), the T-OM soil sample was treated several times with H 2 O 2 until adding did not result anymore in air bubbles emanating from the aqueous solution and maintained at 40 • C in a thermoregulated bath [29]. The resulting sample was then washed four times with double-distilled water (partial removal of OM soil sample = R-OM). Finally, both soil samples were freeze-dried and stored at 4 • C.

Characterization of Ag 0 and Cu 0 ENPs
The synthetized Cu 0 and Ag 0 ENPs were characterized using transmission electron microscopy (TEM) on a Hitachi model HT7700 (Hitachi, Tokyo, Japan) with Olympus camera (Veleta 2000 × 2000) using high resolution mode at 120 kV. The TEM images obtained were analyzed manually to calculate the particle size with the ImageJ program (version 1.50i, Wayne Rasband, National Institute of Health, Bethesda, MD, USA). The ultraviolet-visible (UV-Vis) spectra was recorded with a double-beam Rayleigh UV-2601 spectrophotometer (BRAIC Co. Ltd., Beijing, China) using 1 cm path length glass cell. The zeta potential (ZP) of Cu 0 and Ag 0 ENPs (25 mg) was measured in the presence of 10 mL KCl 0.01 M using a Nano ZS apparatus (Malvern Instruments, Worcestershire, UK) at 20 • C and the isoelectric point (IEP) was obtained from graphs of ZP versus pH. The Fourier-transform infrared spectroscopy (FT-IR) were recorded with a 1 mL of ENPs suspension. FT-IR analysis was realized using a Cary 630 spectrometer (Agilent Technologies, Santa Clara, CA, USA). The transmission spectrum was acquired with 4 cm −1 resolution and the operating range was 600 cm −1 to 4000 cm −1 at atmospheric pressure and 20 • C. The pH of the suspensions of ENPs was measured with 10 mL using a pH Meter.

Characterization of Soil Samples
The morphological characteristics of both soil samples were obtained by scanning electron microscopy with a STEM SU-3500 transmission module (Hitachi, Tokyo, Japan) and the QUANTAX 100 energy-dispersive X-ray spectrometer detector (EDX), (Bruker, Berlin, Germany) was used for the semi-quantitative analysis of the elemental composition (Al, Si, and Fe content). 20 mg of each soil sample were deposited onto 300-mesh Formvar/carboncoated grids and were inspected under a high-vacuum. Confocal analysis was performed by laser scanning confocal microscopy (LSCM) using the Olympus Fluoview1000 (Olympus Optical Co., Melville, New York, NY, USA). 50 µL of the suspensions were collocated on a microscope slide with a micropipette and the sample was dried on a stove at 40 • C. The total organic carbon (TOC) of T-OM and R-OM soil samples was calculated using a Shimadzu TOC-V CPH instrument (Shimadzu, Tokyo, Japan). The TOC was transformed into soil organic matter content using the conversion factor of 1.72 [30]. The specific surface area of R-OM and T-OM soils was obtained using the Brunauer, Emmett and Teller (BET) theory. Approximately 200 mg of soil sample was degassed for 2 h at 105 • C and then was conducted using N 2 gas at −196 • C in the relative pressure range (P/P0) of 0.05-0.4. Surface area measurements were made with a Quantachrome Nova 1000e analyzer (Quantachrome Instruments, Boynton Beach, FL, USA). The average pore volume and size were obtained using the Barrett-Joyner-Halenda (BJH) model. For the FT-IR absorption spectrum, soil samples were dried at 50 • C for 12 h to eliminate the interference produced by the absorption of the water molecules. To determine the functional groups in both soil samples, the analysis was performed under similar conditions to the ENPs. Soil pH was determined in 1:2.5 soil: double-distilled water ratio after 5 min shaking and 120 min resting, using the same pH Meter used for ENPs determination. Total P was extracted from the soil samples by alkaline oxidation with sodium hypobromite (NaBrO) [31]. After each extraction, the supernatant was filtered (5C, Advantec) and then the concentration of total P in the supernatant was determined using a spectrophotometer Rayleigh UV-2601 with a wavelength of 880 nm [32]. Exchangeable Al was extracted with KCl (1 M) and measured using a Unicam model Solaar 969 atomic absorption spectrophotometer (AAS) (Unicam Ltd, Cambridge, UK). Exchangeable base cations (Na, K, Mg and Ca) in soils were extracted using NH 4 Ac (1 M, pH 7.0) and were measured by AAS [33]. Effective cation exchange capacity (ECEC) was calculated as the sum of exchangeable Al plus the exchangeable base cations [33].
The ZP and IEP of the soil samples were determined pre-and post-adsorption of H 2 PO 4 − on T-OM and R-OM soil samples in the absence and presence of 5% Cu 0 or Ag 0 ENPs using the high point adsorption isotherms similar to the procedure followed by ENPs.

Adsorption Experiments
Batch experiments were conducted to investigate the adsorption of phosphate (indicated as H 2 PO 4 − ) on T-OM and R-OM soil samples in the absence and presence of 0, 1, 3, and 5% Cu 0 or Ag 0 ENPs doses (% w/w). Cu 0 or Ag 0 ENPs doses were added to 0.5 g (dry weight) of soil samples in polyethylene tubes and mixed with 20 mL H 2 PO 4 − solution. The adsorbed amounts of H 2 PO 4 − (q t , mmol·kg −1 ) were determined as the difference between initial concentration and final concentration of H 2 PO 4 − in the solution (Equation (1)).
where, C o is the initial concentrations of H 2 PO 4 − and C t is the concentrations of H 2 PO 4 For the kinetic study, the initial solution of 6.47 mmol·L −1 of H 2 PO 4 − was adjusted to pH 5.5 ± 0.2 by adding 0.1 M HCl or KOH at ionic strength 0.01 M KCl and 20 ± 2 • C. Samples were taken from the suspension at 2.5, 5, 10, 30, 45, 60, 120, 180, 360, 720, and 1440 min, and H 2 PO 4 − was determined in solution. Furthermore, the initial pH (pH i ) and the final pH (pH f ) were measured after H 2 PO 4 − solution was added to soil samples (time 0 min) and after H 2 PO 4 − adsorption (1440 min), respectively. Adsorption isotherms were obtained by varying the initial H 2 PO 4 − concentrations from 0.016 to 9.71 mmol·L −1 and were initially adjusted to pH 5.5 ± 0.2 and ionic strength 0.01 M KCl. The suspensions were stirred at 200 rpm in an orbital shaker at 20 ± 2 C for 24 h. To determine the effect of copper (Cu 2+ ) or silver cations (Ag + ) or L-ascorbic acid on H 2 PO 4 − adsorption onto T-OM and R-OM soil samples, adsorption isotherms were made in the presence of 3% Cu 2+ or Ag + or L-ascorbic acid (% w/w) under the aforementioned experimental conditions. The desorption experiment was performed once the adsorption isotherm procedure had ended by adding 20 mL of double-distilled water three times, and the samples were then stirred at 200 rpm in an orbital shaker at 20 ± 2 • C for 24 h. The desorption percentages (%) were calculated by the equation used by Silva-Yumi et al. [34] All the samples of the adsorption experiments were first centrifuged at 10,000 rpm for 10 min, using a centrifuge RC-5B Plus (Sorvall, Newtown, CT, USA) and then filtered through 0.22 µm syringe filters. In all experiments, the concentration of H 2 PO 4 − in the supernatant was determined according to the procedure followed for total P. To minimize manipulation errors in the analysis, the adsorption experiments were performed in triplicate.

Data Analysis
The kinetics adsorption (e.g., pseudo-first-order, pseudo-second-order, and Elovich) and isotherm (e.g., Langmuir and Freundlich) models used in this study are presented in Tables 1 and 2, respectively. Table 1. The kinetic models used for the description of phosphate adsorption.

Kinetic Equations Expression Formula Parameters References
Pseudo-first-order (PFO) q t = q e 1 − e −k 1 t q t = amount of anion adsorbed at any time (mmol·kg −1 ). q e = amount of anion adsorbed at equilibrium (mmol·kg −1 ).
. β = number of sites available for the sorption and desorption constant (mmol·kg −1 ).
* From PSO initial adsorption rate (h), can be calculated by multiplying k 2 q 2 t (mmol·kg −1 ·min −1 ). Table 2. The isotherm models used for the description of phosphate adsorption.

Isotherm Equations Expression Formula Parameters References
Langmuir q e = q m K L C e 1+K L C e q e = Amount of adsorbed anion per unit mass of the adsorbent at equilibrium (mmol·kg −1 ). q max = Maximum adsorption capacity (mmol·kg −1 ). C e = concentration of anion at equilibrium in the solution (mmol·L −1 ). K L = Constant related to the affinity (L·mmol −1 ). K F = Freundlich adsorption coefficient ((mmol·kg −1 ) (L·kg −1 ) 1/n ). n = Adsorption intensity (1 < n < 10). [34,37] Freundlich The data were evaluated through the Chi-square (χ 2 ), adding the coefficient of determination (r 2 ) (Equations (2) and (3)). The lowest χ 2 and highest r 2 values were used as the best fit [37]. The statistical analysis of the adsorption data was conducted using Origin Pro 8.0.
(q e,cal − q e,mean ) 2 + (q e,cal − q e,exp ) 2 (2) where, q e,mean is the average value of experimental adsorption capacity (mmol·kg −1 ), q e,cal is the equilibrium capacity from a model (mmol·kg −1 ) and q e,exp is the experimental adsorption capacity.

Characterization of Cu 0 and Ag 0 ENPs and Soils
The size, morphology, surface charge and the presence of functional groups on the surface of prepared ENPs were determined by TEM images, UV-Vis, ZP and FT-IR analyses. TEM images showed that both ENPs had spherical morphology ( Figure S1a,b in Supplementary Materials). Cu 0 ENPs had a diameter between 8 and 29 nm, whereas Ag 0 ENPs showed a diameter between 7 and 27 nm ( Figure S2a,b). The UV-Vis spectra of Cu 0 and Ag 0 ENPs showed an extended peak in the range of 342-512 and 337-474 nm, respectively ( Figure S3). The FT-IR spectra for pure L-ascorbic acid showed a band corresponding to a stretching vibration carbon-carbon double bond at 1674 cm −1 and the peak of enol hydroxyl at 1322 cm −1 ( Figure S4a). After the reduction of Cu 2+ and Ag + by L-ascorbic acid, the peaks disappeared and new peaks at 3481 cm −1 and 1636 cm −1 were observed ( Figure S4b,c), which were associated with the conjugated hydroxyl and carbonyl groups, respectively. The pH of Cu 0 and Ag 0 ENPs suspension was 2.46 and 2.35, respectively. The IEP of Cu 0 ENPs was 2.7, whereas Ag 0 ENPs had a negatively charged surface in the studied pH range ( Figure S5).
Physico-chemical properties of the soils untreated (T-OM) and treated with H 2 O 2 (R-OM) are shown in Table 3. The T-OM and R-OM were a typical Andisol exhibiting acidic characteristics showing pH values of 5.40 (strongly acidic) for T-OM and 6.20 (slightly acidic) for R-OM. Total P and OM in T-OM were 1.8 and 3.1 times higher as compared to R-OM, whereas the Al and Fe contents for R-OM were 1.2 and 1.4 times higher than T-OM. The SEM images revealed a decreased number of aggregates in R-OM compared to T-OM (Figure 1a,b). The contrasting OM content was also indicated in confocal images (Figure 1c,d) by a higher green fluorescence intensity for T-OM as compared to R-OM images. The IEP for T-OM was 3.2, while it was 5.7for R-OM. Furthermore, the BET-specific surface area and pore volume increased 1.4 and 11.5 times for R-OM in comparison to T-OM. The FT-IR analysis ( Figure S6) showed that R-OM had bands at 1003 cm −1 and 913 cm −1 corresponding to alumina and silica-rich allophane, respectively, while T-OM only showed the band at 1003 cm −1 [29]. T-OM had more effective cation exchange capacity (ECEC) than R-OM (Table 3). Table 3. Physico-chemical properties of soil with total organic matter (T-OM) and with partial removal of matter (R-OM). Parameter 1766

Adsorption Kinetics
The kinetic studies are shown in Figure 3. We observed that increasing in contac time at pH 5.5 as well as in the presence of Cu 0 or Ag 0 ENPs there was a subsequent in crease in the adsorption of H2PO4 − in T-OM and R-OM soil samples. It was also shown

Adsorption Kinetics
The kinetic studies are shown in Figure 3. We observed that increasing in contact time at pH 5.5 as well as in the presence of Cu 0 or Ag 0 ENPs there was a subsequent increase in the adsorption of H 2 PO 4 − in T-OM and R-OM soil samples. It was also shown that adsorption comprised a fast initial phase at 45 min, followed by a slower rate stage until equilibrium was reached at 360 min for T-OM and at 720 min for R-OM, whereas in the presence of ENPs for most systems it was reached at 720 min. Based on the    To determine the kinetic constants and understand the adsorption mechanisms, the experimental kinetics data were modeled by the pseudo-second-order (PSO) Elovich (Table 5) and pseudo-first-order (PFO) (Table S1) models. PSO and PFO models describe the kinetics of the adsorbate on an adsorbent based on chemical-adsorption and physical-adsorption, respectively, with respect to the adsorbent capacity [36]. On the other hand, the Elovich model describes the sorption of adsorbate onto a heterogeneous surface [38,39]. Table 5. Pseudo-second-order and Elovich parameters (± standard error) obtained from H2PO4 − adsorption kinetics at pH 5.5 ± 0.2 for the soil with total organic matter (T-OM) and with partial removal of organic matter (R-OM) in the absence and presence of different doses of Cu 0 or Ag 0 ENPs.  To determine the kinetic constants and understand the adsorption mechanisms, the experimental kinetics data were modeled by the pseudo-second-order (PSO) Elovich (Table 5) and pseudo-first-order (PFO) (Table S1) models. PSO and PFO models describe the kinetics of the adsorbate on an adsorbent based on chemical-adsorption and physical-adsorption, respectively, with respect to the adsorbent capacity [36]. On the other hand, the Elovich model describes the sorption of adsorbate onto a heterogeneous surface [38,39].
Based on the higher r 2 and the lower χ 2 values, the PSO model fitted to the adsorption kinetics data better than the PFO model. According to the PSO model, the amount of H 2 PO 4 − adsorbed at equilibrium (q e,cal ) in T-OM and R-OM soil samples increased with ENPs contents and it was higher in R-OM than T-OM, except for 3 and 5% Ag 0 ENPs doses. The kinetic rate (k 2 ) did not show a clear trend at low ENPs contents. However, it increased in T-OM with 5% Ag 0 ENPs and with 3 and 5% Cu 0 ENPs in R-OM as compared to the soils without ENPs. Similar behavior was observed for the initial adsorption rate (h) in the presence of ENPs leading to increases by adding 3% Cu 0 and 5% Ag 0 ENPs for T-OM and R-OM soil samples and 5% Cu 0 for R-OM and 3% Ag 0 ENPs for T-OM.
Experimental kinetic data at pH 5.5 in T-OM and R-OM soil samples without and with increasing Cu 0 or Ag 0 ENPs content also adequately fitted the Elovich model (r 2 = 0.927 − 0.998 and χ 2 = 9 − 279). This means that the H 2 PO 4 − adsorption happened on a heterogeneous substrate [38]. The initial rate (α) and the surface coverage (β) obtained from this model showed a similar tendency to h and k 2 , respectively, calculated from the PSO model. Thus, both PSO and Elovich models were capable of describing the kinetics of H 2 PO 4 − adsorption on volcanic soils. Similar results have been obtained by other researchers for an acid soil [40] and for adsorbents such as biochar [38] and chitosan [41].

Adsorption Isotherms
The isotherm adsorptions at pH 5.5 ( Figure 4) showed that the amount of H 2 PO 4 − adsorbed was slightly higher in R-OM than T-OM and H 2 PO 4 − adsorption increased with increasing ENPs contents. In general, all adsorption isotherm described curves type L [42]. This means that a high affinity of H 2 PO 4 − anions exist in both soils. In particular, in T-OM samples, the curve reached a strict asymptotic plateau, while in R-OM the curve did not reach it. This difference indicated that the number of adsorption sites in the T-OM sample for H 2 PO 4 − is limited; on the contrary, the R-OM sample had a greater number of adsorption sites for H 2 PO 4 − . At the same time, by increasing the Cu 0 or Ag 0 ENPs content, the curves showed a much less strict plateau for both soil samples, suggesting that the number of available adsorption sites for H 2 PO 4 − increased [42,43]. Table 5. Pseudo-second-order and Elovich parameters (± standard error) obtained from H 2 PO 4 − adsorption kinetics at pH 5.5 ± 0.2 for the soil with total organic matter (T-OM) and with partial removal of organic matter (R-OM) in the absence and presence of different doses of Cu 0 or Ag 0 ENPs.

Pseudo-Second-Order Elovich
ENPs Doses (%) q e,exp (mmol·kg −1 ) q e,cal (mmol·kg −1 )  The adsorption isotherm data were fitted by Langmuir and Freundlich models ( Table  6), which have been frequently used to explain H2PO4 − adsorption on different soils [44,45]. Table 6. Langmuir and Freundlich parameters (± standard error) obtained from H2PO4 − adsorption isotherms at pH 5.5 ± 0.2 and desorption (%) for the soil with total organic matter (T-OM) and with partial removal of organic matter (R-OM) in the absence and presence of different doses of Cu 0 or Ag 0 ENPs.  The adsorption isotherm data were fitted by Langmuir and Freundlich models (Table 6), which have been frequently used to explain H 2 PO 4 − adsorption on different soils [44,45]. Table 6. Langmuir and Freundlich parameters (± standard error) obtained from H 2 PO 4 − adsorption isotherms at pH 5.5 ± 0.2 and desorption (%) for the soil with total organic matter (T-OM) and with partial removal of organic matter (R-OM) in the absence and presence of different doses of Cu 0 or Ag 0 ENPs.

Langmuir Freundlich Desorption
ENPs Doses (%) q e,exp (mmol·kg −1 ) The Freundlich model fitted the experimental data of T-OM and R-OM soil samples better than the Langmuir model (Table 6). However, in the presence of ENPs in T-OM and R-OM soil samples, the Langmuir model, except for R-OM-1% Cu 0 ENPs and T-OM-1% Ag 0 ENPs systems, showed a better fit to the experimental data (r 2 = 0.926 − 0.982 and χ 2 = 151 − 1000). According to the Langmuir model, the maximum H 2 PO 4 − adsorption capacity (q max ) in R-OM and T-OM soils increased with ENPs contents, and it was higher on R-OM than T-OM, except for 5% Ag 0 ENPs dose, in contrast to the affinity coefficient (K L ).

Desorption
The desorption (%) depends on the chemical nature and energy of the bonds between soil components and phosphate [46]. In this sense, after the soil samples were treated with double-distilled water repeatedly (three times), H 2 PO 4 − desorption was about 3.2 times higher from T-OM than R-OM (Table 6). In the presence of ENPs, the desorption from R-OM and T-OM soils decreased with increasing Cu 0 ENPs doses as well as from T-OM with 3 and 5% Ag 0 ENPs. In contrast, with increasing Ag 0 ENPs content, desorption from R-OM was greater than without ENPs.

Characterization of Cu 0 and Ag 0 ENPs and Soil Samples Studied
The particle size average of Cu 0 (19 nm) and Ag 0 ENPs (17 nm) was low due to L-ascorbic acid coating, which provides colloidal stability to the nanoparticles by electrostatic repulsion. The stability effect of the L-ascorbic acid coating could be attributed to the presence of a polyhydroxyl structure on the surface of both nanoparticles [28]. This was supported by the high negative values of ZP, which is normally related to the negatively charged groups of the capping agents [28,47]. Similar results using organic molecules as reducing and capping agents for the preparation of ENPs have been reported previously [28,[47][48][49][50].
The organic matter in volcanic soils is highly stabilized [51], whereby after repeated treatment with H 2 O 2 , only a part of the OM was removed from soil, accounting a 14.1% of OM (T-OM), obtaining a soil sample with 4.6% of OM (R-OM) ( Table 3). The partial removal of OM significantly changed the aggregate structure of the soil because OM acts as a binding agent [52]. In addition, T-OM had more aggregates, a higher P concentration and an effective cation exchange capacity (ECEC) as compared to R-OM. In this sense, it is knowing that the functional groups of OM such as carboxyl, alcoholic hydroxyl, and phenolic hydroxyl contribute to the aggregation of soil particles, formation of humic (organic matter)-Al (Fe)-phosphate complexes and cations adsorption [52,53]. Likewise, R-OM samples had a higher IEP and BET-specific surface area than T-OM. This can be explained by the exposure of ≡Fe-OH and ≡Al-OH active sites from amorphous components of the soil, which decreased the negative charges of the surface and increased BET-specific surface area [34]. In general, allophane and ferrihydrite minerals can interact with negatively charged ENPs through attraction (Van der Waals) forces contributing to their retention in the soil [54].

Ad-and Desorption of Phosphate on Soils
The phosphate adsorption isotherms on T-OM and R-OM soil samples in the absence of ENPs were best fitted to the Freundlich model (Table 6), which reflected the heterogenic nature of soil components. The intensity of adsorption (n) and relative adsorption capacity (K F ) for R-OM were higher than T-OM. The difference between K F and n for two soil samples may be due to the higher OM content of T-OM, since OM could block adsorption-specific sites leading to a lower availability of surface-reactive sites and weak interaction with H 2 PO 4 − [55]. The OM can act by preventing the irreversible retention of H 2 PO 4 − and increasing the nutrient recovery. We found that, after partial OM removal, the H 2 PO 4 − desorption from R-OM was lower than from T-OM (Table 6), indicating a strong interaction between the phosphate and mineral components of R-OM [15,16,23]. These results are supported by the higher BET-specific surface area and lower negative surface charge of R-OM as compared to T-OM. Similar results were obtained by Zeng et al. [56] for H 2 PO 4 − desorption in volcanic soils exhibiting contrasting OM contents. However, these findings were in contrast to the results reported by Debicka et al. [57] by removing the OM from sandy soil resulted in decreases of K F and n values. Contrasting results could be attributed to the particularly components in each soil. According to the FAO-WRB soil classification, sandy soils such as Brunic Arenosols are mainly characterized by minerals such as hematite, goethite, and maghemite [57,58]. On the contrary, Santa Barbara soil is formed by minerals such as allophane (>50%), followed by 1-5% halloysite and vermiculite [59]. In this context, Parfitt [60] found that phosphate was adsorbed in the order hematiteg oethite < ferrihydrite < allophane. Moreover, H 2 PO 4 − can be rapidly and strongly adsorbed on the most reactive aluminol (≡Al-OH) groups of the allophane by ligand exchange forming monodentate or/and binuclear complexes.
According to the PSO model, the higher H 2 PO 4 − adsorption (q e,cal ) was in the R-OM as compared to T-OM (Table 5), which could due to the destruction of OM in T-OM, leading to a larger pore volume and BET-surface area. In addition, R-OM improved the accessibility to active sites for H 2 PO 4 − according to the higher values of α and h obtained for R-OM (Table 5) [57]. The h parameter can be associated to the chemical and/or hydrogen bonding interaction between H 2 PO 4 − and surface hydroxyls in soil samples at the initial adsorption process [16]. Moreover, considering the Elovich model and increase in pH f values after H 2 PO 4 − adsorption with respect to pH i (Table 4), we might suggest that H 2 PO 4 − adsorption in T-OM and R-OM soil samples was performed mainly through ligand exchange (chemi-adsorption) onto Fe/Al (hydr)oxides forming monodentate or bidentate complexes. The pH changes were consistent with the studies carried out by Vistoso et al. [24], who reported that H 2 PO 4 − was adsorbed through ligand exchange mechanism in volcanic soils with contrasting properties.
The H 2 PO 4 − adsorption on T-OM was pH-dependent in contrast to R-OM ( Figure 2). In this context, the IEP of T-OM was 3.2 whereas it was 5.7 for R-OM. Therefore, in acidic pH H 2 PO 4 − solution the surface hydroxyl (-OH) groups in R-OM were more protonated than in T-OM, causing a favorable effect on electrostatic interaction and ligand exchange [61]. However, at alkaline pH H 2 PO 4 − solution, mainly for T-OM, there was a decrease in the ligand exchange and an increase in electrostatic repulsion due to deprotonation from soil superficial groups. Likewise, at a higher pH, the competition between OHand H 2 PO 4 − on the T-OM surface would also reduce the H 2 PO 4 − adsorption [62].

Ad-and Desorption of Phosphate on Soils in the Presence of Cu 0 or Ag 0 ENPs
The increasing phosphate adsorption with increasing ENPs content in soils indicated that in the presence of ENPs, the number of adsorption sites increased. Although, there was a decrease in the initial adsorption rate (h) with 1% ENPs content, which implied that during the first few minutes ENPs compete with H 2 PO 4 − for the adsorption sites of the soil surface. Additionally, h strongly increased with 3 and 5% ENPs content, suggesting that ENPs also contributed to new adsorption sites for H 2 PO 4 − [63,64]. Accordingly, Duncan and Owens [63] indicated that CeO 2 ENPs can be adsorbed on soil adsorption sites before H 2 PO 4 − and Sun et al. [64] determined a similar trend for h with increasing carbon nanotubes (CNTs) content after studying the effects of CNTs with outer diameter of 25 nm and inner diameter of 5 nm on H 2 PO 4 − adsorption in sediments. The adsorption isotherms of H 2 PO 4 − on T-OM and R-OM following Cu 0 or Ag 0 ENPs addition fitted to the Langmuir model (Table 6). Similarly, Sun et al. [64] found that in the presence of CNTs the isotherms for Cd(II) on sediment showed a better fit to the Langmuir than the Freundlich model; however due to the adsorption sites of sediments with CNTs are heterogeneous, they used the Freundlich to describe their results. Therefore, the fit of adsorption data to the Langmuir model in the presence of ENPs should be more studied.
Adsorption enhancement was larger through Cu 0 than Ag 0 ENPs. According to Afshinnia and Baalousha [65], the decrease in the zeta potential after H 2 PO 4 − adsorption on T-OM and R-OM soil samples with ENPs could be associated with H 2 PO 4 − adsorption/complexation onto the ENPs surface ( Figure S7). In this context, Niaura et al. [66] indicated that H 2 PO 4 − was adsorbed through monodentate surface coordination on Cu 0 ENPs, while on Ag 0 ENPs it was performed through hydrogen bonding [66,67]. Although both coated ENPs had a low rate of oxidation and dissolution [68], it was probable that these processes could be favored by an acidic soil pH as well as a consequence of the ionic exchange between H 2 PO 4 − and L-ascorbic acid on the surface of the ENPs, being similar to the mechanism observed for citric acid [50]. Under such conditions, Cu 0 could be oxidized to Cu 2+ (E • Cu2+/Cu0 = 0.337 V) and the amount of phosphate adsorbed in T-OM and R-OM soil samples increased ( Figure S8) because Cu 2+ could be linked to H 2 PO 4 − and hydroxyl groups of OM via a cation bridge [69]. Furthermore, this could be attributed to the formation of complexes between Cu 2+ and H 2 PO 4 − and the precipitate of Cu 3 (PO 4 ) 2 (Ksp = 2.07 × 10 −33 ) [70]. Meanwhile, in the case that Ag + ions were released from Ag 0 ENPs into solution (E • Ag+/Ag0 = 0.799 V), the formation of AgCl precipitate was more favorable (Ksp = 1.77 × 10 −10 ) than a Ag 3 PO 4 formation (Ksp = 8.89 × 10 −17 ) [71,72].
On the other hand, the presence of L-ascorbic acid free in soil solution slightly competes with H 2 PO 4 − for available adsorption sites, decreasing H 2 PO 4 − adsorption on T-OM and R-OM soil samples ( Figure S8). However, as a consequence of the addition of Cu 0 or Ag 0 ENPs suspensions to soil samples, the pH i values decreased, being less acidic in T-OM as compared to R-OM (Table 4), which was consistent with the buffering capacity of OM [73]. An acid pH can be associated with a decrease in the electrostatic repulsion between H 2 PO 4 − and the negatively charged surface of the organic matter (-COOH, -OH) due to a decrease in the number of deprotonated surface groups [74]. Furthermore, the protonation of surface hydroxyl groups of Fe/Al (hydr)oxides might be favored by acid pH values, promoting the H 2 PO 4 − adsorption through a ligand exchange [24,75,76]. In the same way, it has been reported that below 4.5 of pH values the mineral dissolution is favored, promoting the precipitation reactions between H 2 PO 4 − and cations in solution (Al 3+ and Fe 3+ ) [77], and to form H 2 PO 4 − -cation-organic matter complexes [53]. The increase of the H 2 PO 4 − adsorption at a low pH has been demonstrated on pillared bentonites [75], AgNPs-tea activated carbon [76], sediments [78] and in Andisol soils [24]. Future research should be addressed to corroborate whether, in the presence of both ENPs, one of these mechanisms was prevalent for H 2 PO 4 − adsorption on T-OM and R-OM soil samples, or whether several mechanisms acted together.
The H 2 PO 4 − adsorption in the presence of ENPs through chemical interactions onto a heterogeneous surface was indicated by the adequate fits of the kinetic data to the PSO and Elovich models (Table 5). In addition, the desorption behavior supported the adsorption mechanisms proposed in the presence of ENPs. With Cu 0 ENPs, the desorption of H 2 PO 4 − from T-OM and R-OM soil samples was smaller than Ag 0 ENPs. These results can be supported by a chemisorption-like interaction between H 2 PO 4 − and Cu 0 ENPs. Similarly, desorption studies of U(VI) on the soil in the presence of nano-crystalline goethite showed that U(VI) was more resistant to released due to an increase in the inner-sphere complexes on the soil surface [79]. In addition, Elkhatib et al. [80] revealed that sorption of Hg(II) on arid soils in the presence of water treatment residual nanoparticles occurred mainly through inner-sphere complexes, which enhanced Hg immobilization in the arid soils. The high desorption of H 2 PO 4 − in R-OM following Ag 0 ENPs addition needs further investigation. One possible explanation for this is that the Ag 0 ENPs were attached to the potential H 2 PO 4 − adsorption sites, such as allophane and Fe oxides, leading to a blocking effect for H 2 PO 4 − on this soil with lower levels of OM. Then, the H 2 PO 4 − physi-adsorbed (through hydrogen bonding) on the surface of the attached Ag 0 ENPs was more desorbable.

Conclusions
Our study demonstrated that the phosphate adsorption process in the presence of ENPs was dependent on the amount of ENPs and soil organic matter content. The addition of Cu 0 caused a higher increase in phophate adsorpion on T-OM and R-OM as compared to the Ag 0 ENPs. The Elovich and pseudo-second-order (PSO) models correctly described the kinetic adsorption of phosphate on T-OM and R-OM soil samples without and with ENPs.
The phosphate adsorption with both ENPs was better described by the Langmuir isotherm model than the Freundlich model. According to the Langmuir model, by increasing the ENPs content from 0 to 5%, the maximum adsorption capacity (q max ) of H 2 PO 4 − for T-OM ranged from 216.1 to 316.4 mmol·kg −1 following the Cu 0 ENPs addition and to 332.8 mmol·kg −1 using Ag 0 ENPs. Meanwhile, with the increase from 0 to 5% of ENPs, the q max of H 2 PO 4 − for R-OM ranged from 224.7 to 440.2 mmol·kg −1 with Cu 0 ENPs and to 301.4 mmol·kg −1 with Ag 0 ENPs. Phosphate desorption in T-OM and R-OM soils following Cu 0 ENPs addition was lower than Ag 0 ENPs. In the future, more attention should be pointed globally to management agriculture practices based on nanotechnology, because the incorporation of ENPs into the soil have the potential to reduce the already limited crop phosphorus availability.