2.1 Biochar and activated carbon materials

Biochar and activated carbon production details are listed in Table 1. The AD biochar was added since it is a modified biochar designed for optimal P sorption and is thus a state-of-the-art biochar material for comparison to the commercially available pine feedstock biochar (BN). The activated carbon (AC) is a specially produced for water purification (product PSC325, particle size < 325 microns, certified by NSF-61 and AWWA-B600). The BN was sieved to a particle size of 850–150 μm. The AD Biochar was used as received (less than 850-μm fraction).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Biochar source and production.

https://doi.org/10.1371/journal.pwat.0000092.t001

2.2 Iron-modified biochar preparation

To modify the biochars and AC, concentrated FeCl₃ solution was acquired from Kemira Inc. (Helsinki, Finland) (product PIX-311, which is a water treatment chemical; 39.8% w/w Fe concentration). For modification, 15 g biochar or AC was suspended in ~500 ml of 18 mΩ ultra-pure water. The solution was stirred for 10 min at 700 rpm using a magnetic stir bar, after which 1M HCl was added dropwise until the suspension reached a pH of ~ 3.5–4. After the pH was stable, FeCl₃ was added to the suspension to create Fe weight percents from 0.5% to 8%; in this paper the focus is on the 2% Fe-modified biochar because it showed maximum increase in P sorption per Fe addition. The modification suspension was reacted for ~15 min and then titrated with 1.0 and 0.1 M NaOH to achieve a final pH of 6.5. Care was taken to slowly bring the solution up to the final pH to promote Fe sorption on the biochar particles as opposed to formation of a separate solid phase. After 10 min of stable pH, the suspension was centrifuged at 3000 relative centrifugal force for 10 min. The sedimented biochar or AC were then resuspended twice in 200 ml of 18 mΩ ultrapure water followed by another centrifugation and separation. The modified biochar and AC were air dried for 72 h. The non-modified biochar was also rinsed three times to make it comparable to the Fe-modified biochar.

2.3 Biochar characterization

Biochars were digested with a modified dry ash method as described by Enders et al. [83]. Briefly, 200 mg of air-dried sample was added to ceramic crucibles, placed in a muffle furnace, and slowly heated to 500°C for 10 h. The ash was transferred to plastic digestion tubes and digested twice with concentrated HNO₃ at 120°C to dryness. Next, 1.0 ml trace metal grade HNO₃ and 4.0 ml reagent grade 30% H₂O₂ was added, and the tubes were placed back in the pre-heated block and processed at 120°C to dryness. The residue was solubilized with 1.4 ml of concentrated HNO₃ and 48.6 ml 18 Ωm water. Samples were vortexed for 1 min and filtered through a 0.45 μm PES syringe filter (Tisch Scientific, Cleves, OH). Samples were analyzed by ICP spectrometer for elemental composition using certified standards (SPEX CertiPrep (Metuchen, NJ USA)). Duplicates were run in all digestion batches and were within 10% of each other. For CNS analysis, samples were air-dried for 72 hours, ground to pass a 250-μm sieve and oven dried for 24 hrs at 65°C.

Biochar and AC samples were analyzed for total carbon, nitrogen, and sulfur on an Elementar vario Max cube (Elementar Americas Inc., Ronkonkoma, NY, USA). pH and electrical conductivity were measured on a 1:10 solid to solution slurry.

2.4 Phosphorus sorption isotherms

Phosphorus sorption behavior on the biochars was measured using isotherm methods described in Graetz et al. [84]. Briefly, a 1000 mg/L P stock solution was prepared with Na₂HPO₄ ∙7H₂O (Fisher Scientific Inc., Massachusetts, USA) diluted to 100, 75, 50, 20, 10, 5, 2, 1, and 0.5 mg/L P. The pH of the solutions was adjusted to 6.5 with 0.1 and 0.01 M HCl. In this study we chose pH 6.5 because it is a near neutral pH, reflecting the pH of many wastewaters, and because P adsorption on iron oxides is relatively high at that pH and decreases as pH increases [85]. 0.2 g of rinsed biochar or AC for each concentration was placed in 15 ml polypropylene centrifuge tubes. Each sample concentration was run in triplicate. 5.0 ml of 18 mΩ ultra-pure water was added to each tube. The suspensions were allowed to equilibrate for ~30 minutes, then the pH checked again and adjusted to 6.5. The appropriate volume of stock solution was then added to the pH adjusted slurry. The remaining volume of water was added to bring the total solution volume to 10 ml, yielding a 1:50 solid to solution ratio. The tubes were placed on an end-over-end shaker and reacted at lab temperature ~22°C for 24 hours at 30 rpm. Post-incubation samples were allowed to settle for ~30 min and a final pH was measured. The samples were centrifuged at 3000 relative centrifugal force for 10 min, and supernatant was syringe filtered through a 0.45 μm PES filter. Solutions were analyzed by either molybdate blue colorimetric assay for dissolved reactive phosphorus [86], or by ICP-AES as described above. The molybdate blue reagent was prepared as outlined by Pote et al. [87]. Solutions were analyzed on a GENESYS 10S UV-VIS Spectrophotometer (Thermo Scientific, Waltham, MA) at 880 nm wavelength.

The amount of P sorbed (Q) on the biochars and AC was calculated using the following equation: Where Q is phosphorus adsorbed (mg/kg); C_i is initial P concentration (mg/L); C_e is equilibrium concentration (mg/L); V = volume of solution (L); m = mass of biochar (kg). The adsorption isotherms were fit using Freundlich and Langmuir adsorption isotherm models [88], implemented as nonlinear fit optimization algorithms in the software OriginLab (Northampton, MA, USA).

2.5 Phosphorus release

Phosphorus release studies were done on biochar and AC that had no P addition, and biochars incubated in 10 mg/L P solution using the same procedure described in the isotherm experiments. The release studies were done by leaching DI water adjusted to pH 6.5 through a polycarbonate stir-flow cell [89] with a peristaltic pump at 2.0 ml min^-1. The stirred-flow cell was fitted with a 0.45 μm PES filter. The effluent was collected in 6.0 ml increments in a fraction collector. The pH of the effluent was periodically checked in the collection tubes. Solutions were analyzed on the ICP-AES or colorimetric analysis as previously described. Release data are plotted as percent P release as a function of the volume of effluent. Percent P released per volume effluent was calculated as the area of the release concentration curve, summed for each time interval, divided by the total amount of P on the biochar as determined by digestion.

2.6 Phosphorus K-edge XANES analysis methods

X-ray absorption near edge structure (XANES) spectroscopy was used to determine P species present in biochar and AC samples. All samples were ground and sieved to less than 500 μm before mounting to a sample holder. P K-edge XANES spectra of biochars and AC were collected at the Canadian Light Source (CLS) on the Soft X-ray Micro-Characterization Beamline (SXRMB). SXRMB uses an InSb(III) monochromator with a 300 μm x 300 μm beam. The beamline was calibrated to 2158 eV using ZnPO₄ powder. Spectra were collected from 2135–2190 eV with a step size of 1 eV on the pre-edge region (2110–2145 eV), 0.25 eV in the near edge region (2145.25–2180 eV), and 0.5 eV in the post edge region (2180.5–2200 eV). Multiple spectra were collected for each sample.

The collected scans were analyzed using the Athena software program [90]. Replicate spectra from each sample were averaged. Each spectrum was calibrated at 2151.6 eV set as the inflection point of the main edge, allowing for the main edge peak height and shape, and pre-edge or post-edge peaks to be linearly separable elements in linear combination fitting (LCF). The same background subtraction parameters were used for each spectrum to provide consistent background subtraction effects across the samples and standards. Supervised linear combination fitting was performed to determine the speciation of each spectrum. Initially, all standards were used to fit the spectra, followed by a reduction to five standards that were deemed to represent the major species possible in the sample spectra (adsorbed P, apatite, dicalcium phosphate dihydrate (DCPD), DCDP with 50:50 Ca to Mg ratio, and phytic acid as an organic P standard). Standards were iteratively removed if they comprised zero or negative contributions to the sample fit. The fits were not forced to sum to 100% but were checked to ensure that all fits were within 10% of 100%, which is an important criterion for evaluating normalization effects. Accuracy of LCF has been estimated to be between 5% to 15% percent absolute in the sample and standard sets that are carefully calibrated and screened to reduce data collection and processing artifacts [91,92], and likely higher when these artifacts are less well controlled. In fitting of the spectra, it was observed that in some spectra, Ca-P minerals could be substituted for each other with only a slightly decreased fit quality as judged by reduced chi-square. However, in no cases could the Ca-P minerals be eliminated or replaced with another component in a fit. Thus, the fit of these mineral standards is grouped as Ca-P minerals instead of individual Ca-P species.

2.7 SEM analysis of biochar

Scanning electron microscopy (SEM) and X-ray fluorescence analysis of the BN biochar samples were done using a Zeiss Supra 35 FEG-SEM. AD biochar SEM was previously reported in Mood et al. [44]. Biochar samples were mounted on a metal stub and carbon coated and analyzed at an accelerating voltage of 20 kV.

3.1 Biochar characterization

The pH of the biochar and activated carbon are alkaline (Table 2), likely due to the presence of carbonate minerals and alkali salts [93]. The pH of the Fe-modified AD, BN, and AC samples were 8.19, 6.98 and 6.51, respectively. The Fe modification process treats the biochar in an acidic suspension followed by base addition to increase the pH to 6.5, which should dissolve and rinse out many of the salts and carbonates that create alkaline biochar solutions; however, the amount of carbonates removed may vary depending on the mineral size and reactivity. The decreased pH may impact soil solution chemistry when the Fe-modified biochars are amended to soils.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Chemical properties of biochar and activated carbon.

https://doi.org/10.1371/journal.pwat.0000092.t002

Total P in the biochar and AC samples ranged from 314 mg/kg in the BN sample to 16,500 mg/kg in the AD sample (Table 2). Feedstock is the major factor controlling total P. Phosphorus content of plants is variable, and in wood from pine trees (feedstock for the BN biochar), total P may be as low as 100 mg/kg [94]. The AD biochar created from anaerobically digested cow manure has the highest P content due to the high P concentration in the cow manure feedstock.

Plant available P from the biochar and activated carbon samples was assessed using the formic acid leachable P test [95]. The P in the AD biochar sample was 90% plant available. In the BN biochar and AC samples, P was 48% and 69% available, respectively. Thus, the availability of the biochars’ native P is variable, depending on feedstock and production methods, which has been observed in many studies on plant uptake of P from biochar [28,29]. Other element compositions are also reported in Table 2, including N, which is an important macronutrient, and Fe, Ca, and Mg, which are important nutrients for plants and are also reactive components of biochar that promote P sorption.

SEM images of the BN biochar showed some mineral grains of silicates and calcium carbonates amongst the biochar particles (S4 Fig). SEM analysis of the AD biochar by Mood et al. [44] showed the presence of small clusters of Mg and Ca minerals dispersed on the surface of the biochars. SEM images from the Fe-modified biochars did not show any Fe mineral particles, only slightly enriched Fe biochar particle regions (S4 Fig), suggesting the Fe is precipitated or sorbed on the biochar surfaces.

3.2 P sorption behavior

Phosphate sorption behavior on the biochars and activated carbon showed high relative amounts of sorption at low concentrations, followed by decreased sorption as concentration increased (Fig 1). The sorption isotherms on BN and AD biochars and the AC sample show typical sorption behavior that occurs when there is a high affinity for sorption, but sorption sites are limited and become saturated as solution P concentration increases [96]. If phosphate removal was dominated by a bulk precipitation process, the sorption isotherm would continuously increase as solution concentration increases, dictated by the solubility of a solid. The isotherm data on the BN biochar (Fig 1C) and the AC (Fig 1A) were fit using the Freundlich isotherm model equations (S1 Table); the Langmuir isotherm model was a poor fit to the data on these samples. On the AD biochar (Fig 1B), the Langmuir isotherm model fit the sorption data slightly better than the Freundlich model, although both fits were good. While both the Freundlich and Langmuir isotherm equations fit non-linear adsorption trends with steep initial slopes, the Langmuir isotherm equation fit provides a sorption maximum value, while the Freundlich isotherm equation does not predict a maximum sorption amount but fits a logarithmic growth sorption curve [97]. Sorption isotherms are useful for predicting concentrations of sorbate at a particular solution concentration, however, interpreting mechanistic reaction processes from the isotherm fit parameters is not possible [98].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Sorption isotherms of phosphorus on unmodified and modified biochar.

Solid lines are fits of isotherms using either the Langmuir (AD) or Freundlich (BN and AC biochar) models (Fit parameters are reported in S1 Table).

https://doi.org/10.1371/journal.pwat.0000092.g001

After Fe modification, P sorption increased in the BN biochar and AC sample (Fig 1C); P sorption maximum in 2% Fe-modified BN biochar increased by nearly 10 times more than the unmodified BN biochar. The untreated AC had a high P sorption capacity and modifying with 2% Fe increased the sorption capacity by ~30% (Fig 1A), which is a much smaller percent increase than observed in the BN biochar.

The unmodified AD biochar had the highest P sorption capacity of all biochars tested in this study, and when it was modified with Fe, P sorption capacity decreased (Fig 1B). Phosphorus sorption behavior of the AD biochar was studied by Mood et al. [44]. They reported that the P sorption capacity was three orders of magnitude greater than observed in this study, however, the incubation pH used in their study was much higher (pH = 10) than this study (pH = 6.5). At higher pH, the solution will be supersaturated with respect to Mg-P and Ca-P minerals, and the P sorption is not limited by surface properties, and thus greater P removal from solution occurs because of mineral precipitation. SEM and X-ray photoemission spectroscopy (XPS) analysis by Mood et al. [44] confirmed the presence of Ca-P minerals. At pH 6.5, phosphate bonding to the positively charged amine functional groups on the AD biochar may be a dominant sorption mechanism in addition to possible Ca or Mg phosphate mineral precipitation. Modifying the AD biochar with 0.5% Fe may have blocked access to the amine functional groups, thus decreasing sorption capacity (Fig 1B). Increasing the Fe modification of the AD biochar from 0.5 to 2% Fe, the sorption capacity increased, but it still did not surpass the P sorption capacity of the unmodified AD biochar. This is likely due to the blocking of the amine functional groups that is countered by addition of more iron-phosphorus adsorption sites than occurred in the 0.5% Fe-modified AD biochar.

Research on biochars made from traditional carbon-rich biomass feedstock such as forest biomass, plant fibers from agricultural residue, and coconut fibers, indicates that they do not sorb large amounts of P. Several papers have provided methods for modifying the biochars to enhance P sorption [15]. Some of the most common modifications include adding Fe to the biochar [50,52,55] or adding Ca or Mg [60,63,64]. Other more complex modification methods have also been developed to produce magnetic biochars [68–70] as well as the AD biochar that was modified by pyrolyzing anaerobic digestate in the presence of ammonia gas to create positively charged amine functional groups that can adsorb P [44]. Iron modification is often proposed because it is an inexpensive method to increase P adsorption on biochars; additionally, the addition of Fe may be beneficial as a soil amendment on the P-loaded biochar. Results from the P sorption isotherm studies show that Fe modification enhances P sorption on biochar or activated carbon made from woody biomass (AC (Fig 1A) and BN (Fig 1C)), but that more specialized biochars, like the AD biochar (Fig 1B), may not need post-production iron modification. The enhanced P sorption on Fe-modified biochars provides a method for recovering P from waste waters and supports creating an enhanced efficiency P fertilizer needed to support a circular bioeconomy that includes P recycling [12,15,16].

3.3 Phosphorus K-edge XANES spectra

Phosphorus K-edge XANES spectra from the biochars and activated carbon are shown in Fig 2 and fit results are reported in S3 Table. Linear combination fitting of P XANES has a fair amount of uncertainty because of the lack of differences in spectral features between molecular species and spectral artifacts that may cause errors in the LCF results [91,99–101]. This uncertainty is especially problematic in natural samples because the standard set may be missing species that are present in the sample, but the fit can still have good fit statistics [101,102]. The standards used in fitting the biochar and activated carbons are shown in S3 Fig. For some species, there are distinguishing features that can be used for species identification, such as the post-edge shoulder at 2155 eV in the spectra of the Ca-P minerals. While other species are distinguished only by the intensity of the main-edge peak. Because the biochar and AC sample spectra are comprised of several species, the LCF results are used to categorize the P as either Ca-P or adsorbed/organic P rather than assigning specific species. In all but one sample, either hydroxyapatite or DCPD were fit as one of the species present that were categorized as Ca-P minerals. Ca-P species fit results ranged from 28% to 83% of the total P species in the biochar samples (S3 Table).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Phosphorus K-edge XANES spectra (solid lines) from biochar and activated carbon samples, and best fits from LCF using standards (dashed lines).

Fit results are reported in S3 Table. Vertical lines indicate distinctive XANES spectral features.

https://doi.org/10.1371/journal.pwat.0000092.g002

The P K-edge XANES spectra from the biochar and activated carbon samples are shown in Fig 2. There are distinct differences between the untreated and Fe-modified samples, especially in the post-edge shoulder at 2155 eV and the small peak at 2163 eV. Both features are indicative of Ca-P minerals; these peaks are most prominent in hydroxyapatite (S1 Fig). The unmodified BN, AD and AC samples all had strong Ca-P minerals peaks. Despite the uncertainty in LCF of P in natural samples, these peaks are distinctive, and their presence clearly shows that Ca-P minerals are a major species occurring in the biochars. The AC and BN samples had the most Ca-P mineral present in the LCF results (82%, S3 Table). Calcium phosphate minerals, like hydroxyapatite, have low solubility [103,104]. Thus, it is anticipated that P release from these biochars would be less than P sorbed on mineral surfaces [105,106]. The occurrence of P in Ca-P species in biochar is likely a result of the presence of Ca in the ash, which comes from Ca in the biomass feedstock. The Ca in the feedstock can either be plant-based Ca that gets released during pyrolysis (Ca is a significant element of biological tissues), or as mineral dust in the biomass feedstock taken in from soil attached to the biomass feedstock or mixed in with the biomass when it is collected from the environment during harvesting.

The P K-edge XANES spectra from the Fe-modified AD and BN biochars and the AC sample all show decreased peak intensity for the Ca-P peaks at 2155 eV and 2163 eV compared to the non-Fe modified samples (Fig 2). In the AC sample, there is nearly a complete loss of the Ca-P mineral features in the spectrum after Fe modification. The LCF results of the AC samples show that the Ca-P mineral species decreased from 82% in the unmodified AC sample to only 28% in the Fe-modified AC sample, indicating a major shift in species to P sorption on iron oxides in the Fe-modified AC sample. Possible sorption mechanisms for P on the Fe-modified biochar are: 1) ternary biochar-Fe-P complexes similar to how P adsorbs on natural organic matter [16,107]; 2) phosphate adsorption on iron oxide phases that may exist either on the surfaces of the biochar as surface precipitates; or 3) P sorption on small, possibly nano-sized, precipitates co-mingled with the biochar. Fe-modified BN and AD biochar also had a decrease in Ca-P species in them, as indicated by the decrease in the peaks at 2155 eV and 2163 eV, suggesting that P is sorbing on the Fe phases on the biochar in these samples.

3.4 P release behavior

Recovery of P on biochar creates a pathway for nutrient recycling when the biochar is used as a soil amendment and may promote additional soil health benefits [30,77], such as increasing soil carbon, providing required nutrients, and promoting soil structure that allows better water retention, root development and gas exchange. Phosphorus in soils is not typically readily available because it strongly bonds to iron and aluminum oxides in acidic soils and forms calcium phosphate minerals in alkaline soils [108]. Thus, P release from the Fe-modified biochar needs to be measured to understand how much of it is available for plant uptake.

The percentage of total P released from the different biochars and activated carbon are shown in Fig 3A and 3B (total P is listed in Tables 2 and S2). An important factor in P sorption and release behavior is solution pH. The influent solution pH was adjusted to 6.5 in these experiments, however, the effluent pHs were different: AC sample pHs ranged from 8.8–9.5; Fe-modified AC pHs ranged from 4.5–4.8; BN sample pHs ranged from 9.3–9.4; Fe-modified BN pHs ranged from 6.1–6.8; AD sample pHs ranged from 6.9–9.2; and Fe-modified AD sample pHs ranged from 4.6–7.1. In all Fe-treated biochar samples, the effluent pHs were 2–3 units lower than they were in the untreated samples. This occurs because the Fe-modification procedure used an acidic Fe solution to incubate the biochars before titrating the suspensions to pH 6.5, which leached out many of the alkaline salts and carbonate minerals. Additionally, the release of phosphate adsorbed on iron oxides is a ligand exchange reaction, thus consuming hydroxide ions and decreasing pH. The varying pH of the biochar and AC samples is an important factor in the behavior of the P release curves. The change in effluent pH in the stirred-flow reactor simulates the effects that would occur in biochar-amended soils when P desorbs and leaches out of the soils into surface or groundwaters.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3.

Percent P released into solution from unmodified (panel a) and Fe-modified (panel b) biochars and activated carbon. Fe-modified chars were incubated in a 10 mg/L P solution prior to the stirred-flow reactor leaching experiments (P loading reported in S2 Table).

https://doi.org/10.1371/journal.pwat.0000092.g003

The unmodified BN biochar released the most P (Fig 3A); ~35% of the total P was released after 0.35 L of leaching, and the positive slope at the end of the experiment suggests P release will continue with more leaching. The AD biochar released very little P (<1% of its total P), indicating that the P is strongly bound in the AD biochar as a poorly soluble species. Phosphorus release from the AC sample was only about 13% of total P after 0.35 L, but even after 0.35 L of leaching, the slope remains positive, suggesting that P leaching will continue. The P species in the AC sample may be a low solubility species that has slow release.

The percentage of P released from the Fe-modified BN biochar and AC sample (Fig 3B) were much less than in the unmodified samples (Fig 3A). Iron-modified AD biochar released slightly more P (~4% after 0.7 L) compared to the unmodified AD biochar (<1%). The greatest amount of P release was from the Fe-modified AC sample, which released ~14% of the P after ~0.7 L of leaching, and the steep slope indicates that additional leaching will release more P. The Fe-modified BN and AD biochar samples both released ~4% of the sorbed P after 0.7 L, and the slope also suggests that P release will continue. The continuous P release in the Fe-modified samples suggests that their use as a slow release fertilizer is promising. However, complete assessment of P availability would require a plant growth trial since plant rhizosphere reactions may enhance P mobility by release of low molecular weight organic acids [109].

Release of Ca and Mg cations during the stirred-flow experiments was also measured (S1 and S2 Figs). Initially, Ca and Mg concentrations were high in all the biochar leaching experiments, followed by a tailing off to a constant low concentration. There are direct relationships between Ca concentrations and P release in the AC sample and AD biochar, which likely reflects the solubility control from calcium phosphate minerals. The Mg release in the AD biochar was inhibited until Ca concentrations decreased. It has been observed that the presence of Mg inhibits apatite formation, favoring meta-stable Ca-P minerals because of the isomorphic substitution of Mg for Ca [110]. Upon Fe amendment, P release was not related to Ca and Mg in the BN and AD biochars. However, there remained a significant inhibition of P release concentration in the AC sample when the Ca and Mg concentrations were high (S2 Fig).

Amongst the Fe-modified samples, the Fe-modified AC sample had the most P release in the stirred flow experiments (Fig 3B), which corresponds to the XANES spectra results showing it had the biggest decrease in the poorly soluble Ca-P minerals. While sorbed P is more available for release to solution than many Ca-P species, its desorbability should still be low [111]. The release curves of P from the Fe-modified BN and AD samples were less than the Fe-modified AC sample. The XANES spectra showed that in the Fe-modified BN and AD samples there was still 37–65% Ca-P minerals (S3 Table), which is more than remained in the Fe-modified AC sample (28% Ca-P species, decreased from 83% in the non-Fe-modified AC sample).

3.5 Phosphorus sorption and release mechanisms

The highest phosphorus sorption capacity by unmodified biochars occurred on the AD biochar produced from anaerobic digest solid separates under ammonia gas pyrolysis. This biochar has positively charged amine functional groups that can adsorb the phosphate anions, thus facilitating a high sorption capacity [44]. However, the P XANES spectra from the AD biochar sample showed that the majority of P in this biochar exists as Ca-P mineral phases, which is in agreement with the findings of Mood et al. [44], who characterized the P sorption behavior of AD biochar at a higher equilibrium pH than used in this experiment.

Based on the P release experiments, the Ca-P species in the AD biochar are not soluble, which agrees with the expected solubility of Ca-P minerals such as hydroxyapatite [104]. In contrast, Kizito et al. [112] measured P sorption from anaerobically digested swine manure effluent onto biochars and observed that the sorbed P was completely reversible. Based on sorption and molecular analysis, Kizito et al. [112] proposed that phosphate was precipitating as Ca-P minerals, which generally have low solubility when they exists in the most stable forms (i.e., hydroxyapatite). However, biochar produced from animal manures may have meta-stable Ca-P minerals that have higher solubility than the hydroxyapatite. The meta-stable Ca-P minerals are favored because high concentrations of Mg in the dairy manure may isomorphically substitute for Ca inhibiting hydroxyapatite formation [110], and also because dissolved organic matter inhibits hydroxyapatite formation [113,114]. These mechanisms could have inhibited the formation of the most stable Ca-P minerals in the biochar produced by Kizito et al. [112], but, due to differences in pyrolysis methods, may not have inhibited formation of more crystalline Ca-P minerals in the AD biochar, explaining why it had very low P release behavior. Iron modification decreased the amount of P sorption on the AD biochar, and P XANES indicated that more of the P in this sample existed as P sorbed as Fe-ternary complexes or sorbed on Fe-oxide nanoparticles on the biochar surface rather than Ca-P species. When Fe modified, the AD biochar released more P in the stirred-flow reactor compared to the unmodified AD biochar. Thus, this may be a suitable strategy to create a biochar that has both slow and fast release P for use as a soil amendment.

The BN biochar had low P sorption capacity, and Fe modification increased the sorption capacity. The most prominent P species in the BN biochar was a Ca-P species, which likely formed from the Ca released from Ca carbonates in the ash fraction of the sample. Iron modification of the BN biochar increased the amount of P-Fe-biochar ternary complexes. Phosphorus release concentrations from the Fe-modified BN sample were much less than in the unmodified BN sample.

Results from this study show that Ca-P minerals are common in biochar, likely due to the alkaline pH and high amounts of Ca salts in the ash contained within the biochar. Magnesium phosphate minerals may also be present in biochars, particularly in biochars with high Mg concentrations [41,53,58,59,64], such as the AD biochar sample. Although Ca-P minerals typically have low solubility, the solubility varies, depending on the crystallinity of the Ca-P species [103]. Thus, some Ca-P minerals may be soluble and would be a good P fertilizer in the biochar-amended soils.