The world is experiencing unprecedented economic growth and an increase in the human population, thereby requiring more sustainable utilization of natural resources [1]. Plant nutrient recovery from anthropogenic waste streams, such as municipal or agricultural wastewater [2][3][4], is of major importance since these streams contain considerable amounts of nitrogen (N) and phosphorus (P) in the form of aqueous ammonium (~500 to ~1000 mg/L NH 4 + ) [5,6] or phosphate (~600 to ~1000 mg/L PO 4 3-) [5,7] ions, with concentration depending on the source and pH of the wastewater [5,8]. If untreated, this wastewater can enter the watershed causing undesirable environmental effects including eutrophication [7,9,10]. Phosphorus scarcity and its sustainable use have been of particular interest since P is a key building block of living organisms that can't be manufactured or destroyed [11]. The most utilized technique to recover P in the form of PO 4 3-is its precipitation/crystallization as magnesium or calcium salts [5,12,13].
The resulting solid, low solubility and slow nutrient release PO 4 3-materials are typically struvite (MgNH 4 PO 4 .6H 2 O) while the calcium PO 4 3analog (CaNH 4 PO 4 .7H 2 O) is very unstable and decomposes at room temperature to hydroxyapatite (Ca 5 (PO 4 ) 3 OH) [14]. Other calcium PO 4 3--containing crystalline materials, such as brushite (CaHPO₄⋅2H₂O), or even amorphous calcium phosphate -have been observed under different reaction conditions such as pH, temperature, supersaturation level and Ca/P molar ratio [12]. Calcium chloride has chiefly been used as a calcium source due to its high water solubility [12,[15][16][17]. Cumulative energy demand associated with calcium chloride produced via the Solvay process and made from brine is 11.8 MJ kg -1 , much greater than that for low solubility calcium source, calcium carbonate, which is mined in a powder form with 0.68 MJ kg -1 [18]. As municipal wastewater treatment plants (WWTPs) begin shifting to the new paradigm of water resource recovery rather than waste production [19], sustainability and availability of low environmental footprint calcium sources become an issue and low solubility calcium-containing materials are currently evaluated to assess their potential for PO 4 3-recovery from wastewater. Natural Ca-rich clays, such as sepiolite [20], calcite-coated phyllosilicates [21] or other calcium-rich natural minerals [22] have been utilized for PO 4 3-recovery as well as other abundantly available processed materials or waste [23][24][25]. Very recent developments, however, have been considering the suitability of calcium-rich industrial waste [26]. In particular, every year a large amount of calcium-containing biomass combustion ash is generated (Denmark 31 kton, Austria 133 kton, The Netherlands 234 kton, Italy 250 kton, Sweden 528 kton, Canada >1000 kton) [27] as the use of biomass for electricity generation as well as for heating and cooling has become more prevalent [28]. Depending on the source of biomass ash (BA), it contains a complex and heterogeneous mixture of inorganic crystalline and amorphous minerals [29,30]. Biomass ash is typically enriched in calcium (Ca) with other nutrients, such as potassium (K), magnesium (Mg) as well as some phosphorus (P) and sulfur (S) [31,32]. Also, elements considered as micronutrients, such as iron (Fe), manganese (Mn) and titanium (Ti) are present [33] with small amounts of other heavy metals. The amounts of ash generated are not insignificant and in developing countries can account for up to 60% of the total municipal solid waste [34]. Few reports exist on BA interactions with PO 4
3-[35,36] but already showed the potential. Up to 97% PO 4 3-recovery was observed with BA 9.05 g/L via surface facilitated adsorption [35]. Solution pH dependence was observed when 10 ppm PO 4
3-was adsorbed on BA with low pH values resulting in the highest removal efficiency [36]. Alternatively, lime kiln dust (LKD), a byproduct collected by dust control systems from burning limestone, contains high calcium concentrations and is currently chiefly used as a soil stabilizer. Together, these two constitute high calcium concentration industrial waste that has a direct potential to recover PO 4 3-from wastewater streams without the need for soluble or mined natural calcium sources.
The purpose of this study is to assess the performance of these two calcium-rich industrial waste sources, BA and LKD, for their PO 4 3removal ability and the resulting release of any heavy metals due to the interaction with organic molecules routinely present in anthropogenic wastewater. PO 4 3-sorption isotherms and the resulting kinetics information were obtained using batch experiments. Moreover, trace metal presence was also quantified as industrial solid waste contains metals that may act as micronutrients [31] or exhibit biotoxicity above critical concentrations. Of particular interest was PO 4 3-removal in the presence of low molecular weight dicarboxylic acids. Gluconic, glyceric, glucuronic, glucaric and tartaric acids are generally known as sugar acids and are oxidation products of sugars [37] hence abundantly present in wastewater since the dissolved organic compounds mainly consist of ethanol, tartaric acid/tartrates, microbial biomass (mainly yeasts) and phenolics [38,39]. For example, tartaric acid was previously found to reduce Cr(VI) to Cr(III) in the soil, chiefly in the liquid environment, and to affect the mobility of heavy metals due to their desorption, complexation, and precipitation [40]. Hence, the leaching of the heavy metals during PO 4 3-recovery in the presence of several dicarboxylic acids was investigated and quantified.
BA and LKD were obtained from manufacturers in Lithuania. LKD used in this study was obtained from JSC Naujasis Kalcitas, Naujoji Akmene, Lithuania and BA obtained from JSC Mortar Akmene by burning wood pellets. All reagents for chemical analysis were obtained reagent grade from Fischer Scientific and used as received. Double distilled water was used in all experiments.
Due to the complexity of solid waste materials, model wastewater was utilized. Simulated NH 4
+ and PO values found in municipal, animal and industrial wastewater [41]. The optimal amount of the BA or LKD was determined by increasing the concentration from 200 to 2200 ppm range until PO 4 3-adsorption achieved a steady state. 600 MAP solution was prepared with and without organic acid dissolved by dissolving solid salts or organic acids in deionized water and solid adsorbent was added while stirring. Aliquots of 1 mL were sampled periodically, filtered through a 0.22 µm polyethersulfone filter, and analyzed using ion chromatography.
The X-ray powder diffraction (XRD) data were obtained with a DRON-6 X-ray diffractometer with Bragg-Brentano geometry using Nifiltered CuK α radiation, operating at a voltage of 30 kV and an emission current of 20 mA. The step-scan spanned the angular range from 10 • to 60
The surface morphology of materials was examined using the Thermo Fisher Scientific Phenom scanning electron microscope (SEM) at an accelerating voltage of 15 kV. SEM images were obtained using the secondary electron (SE) mode. The instrument was equipped with an energy dispersive spectrometer (EDS) system, for the analysis of the Xrays emitted by the samples to determine the elemental composition during SEM observations [42].
The BET surface areas of the samples were measured via nitrogen physisorption (-196 • C) using a Micromeritics ASAP 2020 instrument. Approximately 0.45 g of catalyst was used for each measurement. The specific surface area was calculated using the Brunauer-Emmett-Teller (BET) method. Before the measurement, the catalyst was evacuated at 60 • C until the degassing rate < 10 -4 mmHg/min was observed. In addition, Barrett-Joyner-Halenda (BJH) analysis was employed to determine pore size using adsorption and desorption isotherms. This technique characterizes pore size distribution independent of the external area due to the particle size of the sample.
A chemical analysis of the materials, used in the experimental investigation, was carried out. The samples were sieved, extracted with aqua regia and then analyzed with Perkin Elmer Optima 2100 DV ICP-OES spectrometer (Perkin Elmer, USA). Atomic Absorption Spectroscopy (AAS) (Perkin Elmer, USA) was used to find the concentration of Ca and Mg, flame photometer (FP) (Sherwood, UK) of K, UV-VIS spectrophotometer (Shimadzu, Germany) of P and inductively coupled plasma mass spectrometry (ICP-MS) (Thermo Fisher Scientific, USA) of heavy metals.
Total N was determined using the Kjeldahl method, total P 2 O 5 using the spectrophotometric method with spectrophotometer Agilent Technologies Cary 60 UV-Vis (USA) in solid samples before and after the reaction (BA, LKD, 2200/600 ppm BA/MAP, 1800/600 ppm LKD/MAP, 2200/600/1000 ppm BA/MAP/TA and 1800/600/1000 ppm LKD/ MAP/TA). Determination of C org content according to oxidizability by spectrophotometric method with spectrophotometer Agilent Technologies Cary 50 UV-Vis (USA) in the same samples. Water-soluble P 2 O 5 and total S (SO 4 alpha300R confocal Raman microscope using 532 nm laser, Zeiss EC Epiplan-Neofluar x100 / 0.9 objective, G2: 600 g/mm grating, 3 s integration time per point for ex situ study. For in situ studies, 532 nm laser, Zeiss x20 / 0.4 objective, and G2:600 g/mm grating were used. In both studies, the spectral range was 100-4000 cm -1 with the center at 2000 cm -1 and the spectral resolution was ~2 cm -1 . Before each experiment, the instrument was calibrated using a Si wafer. The laser intensity at the sample was ~54 mW. Spectral maps were typically acquired using 2-4 s exposure time per single scan point. To elucidate heterogeneity of a single particle, 2-D spectral maps were acquired in the x-z plane, e.g. parallel to the direction of the beam.
The True Component Analysis (TCA) was used to create spectral intensity distribution images of different spectra components. It is utilizing a linear combination of the spectral components using the basis analysis algorithm via Eq. (1).
where S i → -spectrum i from the spectral dataset, Bmatrix of basis spectra, H i ̅→ -mixing values of spectrum i, E i → -error spectrum. The mixing values are fitted using least-squares minimization methods following the expression (2). 3-removal mechanisms present. Overall, ~95% PO 4 3-removal was achieved with 1800/600 ppm LKD/MAP while ~78% of PO 4
3-was removed with 2200/600 ppm BA/MAP. These two solid adsorbent concentrations were selected as optimal for all kinetic experiments.
The full chemical composition of BA and LKD is shown in Table 1 with a graphical representation of major nutrients (P, K, Ca, Mg and S) as well as residual Al and Fe shown in Fig. 2a. Both are comprised chiefly of Ca 2+ containing chemical compounds with Ca accounting for ~50-92% of major nutrients measured in this study by weight according to Fig. 2a. Other plant nutrients, such as Mg, K and P, comprise the remainder. LKD, in particular, contains high calcium amounts depending on whether high-calcium lime or dolomitic lime is processed [43]. According tot he literature, typical composition of LKD varies from 31% to 55% CaO with significantly smaller concentrations of other elements (1.7-9.9% SiO 2, 0.7-4.1% Al 2 O 3 , 0.03-0.22% K 2 O, 0.5-25% MgO) [44][45][46]. Notable was residual phosphorus content in BA of ~5% by weight of all major nutrients since P is a primary nutrient necessary to the plants. Al and Fe were present in both BA and LKD. It needs to be stressed that soluble Al 3+ form at low pH is toxic to the plants directly or by binding phosphorus into insoluble compounds [47]. However, this effect only becomes apparent when soil pH reaches below 5.5, e.g. in relatively acidic soil [48,49]. Fe is a nutrient that is required for normal plant growth and reproduction [47]. However, soil that is alkaline or has a significant amount of lime added often causes an iron deficiency in the plants as it becomes unavailable for plant adsorption [50].
P increase in 2200/600 ppm BA/MAP and 1800/600 ppm LKD/MAP solid materials was particularly noticeable in Fig. 2b as it increased by ~x3 and fold, respectively. These early findings suggested that BA and LKD can be viable PO 4 3-adsorption materials and yield sizeable concentrations of this nutrient in the solid phase. The adsorption process in an aqueous solution, however, dramatically modified the physical particle properties of both BA and LKD as shown in Fig. 2c andd, respectively. Before the reaction, as-received materials were particular with the measured particle size of 10-20 µm. After the reaction, distinct particle size decreased as materials underwent partial dissolution resulting in flake morphology of the smaller particle agglomerates. Spatially resolved SEM in combination with EDS was further utilized to investigate the distribution of elements within these reacted particles and allowed observing the dispersion of the major elements and their incorporation into the calcium oxide and calcium carbonate minerals. It can be seen in Fig. 3 that in both reacted cases of 2200/600 ppm BA/ MAP and 1800/600 ppm LKD/MAP elements detected under EDS were chiefly dispersed at the micron level with P uniformly distributed across the particle agglomerates mapped. This suggests that reacted materials uniformly adsorb PO 4 3-and can potentially serve for its controlled release.
Table 2 summarizes EDS measured weight concentrations of major elements and their corresponding oxides in the solid material after the reaction of 600 ppm MAP with 2200/600 ppm BA/MAP and 1800/ 600 ppm LKD/MAP. In particular, P 2 O 5 concentrations of up to 36% were detected for 2200/600 ppm BA/MAP, quite comparable to those currently utilized in mineral phosphorus fertilizers. Notable is the difference in concentrations as measured using bulk techniques in Fig. 2b with EDS which largely measures surface information from a few micro electron beam/sample interaction volumes. This can be rationalized by the mass transfer limitation of the process whereby PO 4 3-is chiefly concentrated on the particle surface with the diffusion being limited.
X-ray diffraction (XRD) measurements were used to elucidate the crystallinity of BA and LKD powders. The results are shown in Fig. 4a andb. This mineralogical characterization using XRD patterns showed distinct crystalline phases. The obtained data confirmed the presence of several major crystalline phases in the as-received materials. The main crystalline phase in BA was CaO [51][52][53] with the main peaks at 2θ of 32 and 37 • and CaSiO 3 [54] and quartz in a complex mixture. The predominant crystalline phase in LKD was calcite -CaCO 3 [55] -with the main peak at 2Θ of 29 • , accompanied by smaller quantities of quartz -SiO 2 [56] and calcium hydroxide Ca(OH) 2 [57]. The XRD analysis of reacted materials showed two different behaviors. BA crystalline structure was significantly perturbed with the peaks almost disappearing or significantly changing suggesting strong dissolution or amorphization of the particles, consistent with the SEM data in Fig. 2d. LKD, on the other hand, exhibited only minor crystalline changes upon PO 4
3-adsorption and largely retained its crystalline order. Interestingly, XRD did not provide conclusive evidence in the formation of any crystalline phosphates implying amorphous internalized or adsorbed PO 4 3-nature. This interesting behavior was further elucidated using BET analysis where, after the adsorption experiments, there was a significant increase in the measured BET surface area. In particular, it increased from 2 and 5 m 2 /g to 35 and 29 m 2 /g for BA and 2200/600 ppm BA/MAP and LKD and 1800/600 ppm LKD/MAP, respectively. While as received BA was a non-porous material as it can be seen in Fig. 4d, it developed a significant amount of porosity with a pore size of up to 1000 Å, similar behavior to that observed for LKD.
The amorphous nature of the material formed warranted a detailed analysis using Raman spectroscopy which can provide detailed information on the structure of the adsorbed PO 4
3-ion [58,59]. Raman peak assignments of the post PO 4
3-adsorption will be based on the existing fundamental data [60,61] based on changes in ion symmetry through the series PO ) of CaCO 3 , as received BA did not exhibit any peaks in Raman spectra (not shown). After 2200/600 ppm BA/MAP and 1800/600 ppm LKD/MAP reaction, however, new peaks were observed in both cases. Representative spectral components derived from TCA and their corresponding spectral maps are shown in Fig. 5 for (a) 2200/600 ppm BA/MAP and (b) 1800/600 ppm LKD/MAP. The most relevant was phosphate peak likely originating from hydroxyapatite, Ca 5 (PO 4 ) 3 (OH), which can be characterized by the characteristic vibration band of PO 4
3-groups 960-962 cm -1 , while amorphous calcium PO 4 3-exhibits a most characteristic shift by a 10 cm -1 towards ~950 cm -1 [62]. Another peak at 1020-1080 cm -1 observed typically overlaps with that of ν 1 (CO 3
) of CaCO 3 [63,64]. The peaks at 848-853 cm -1 are more difficult to assign since they can likely be associated with P-OH or P-(OH) 2 stretches although they miss the corresponding complementary vibrations originating from the PO 4
3-ion symmetry split upon adsorption [65]. Further, since they comprise rather a large portion of the reacted BA material surface as shown in Fig. 5a, they likely can be related to more complex calcium silicates rather than PO 4 3-ion vibration [66]. Other peaks below 750 cm -1 were non-specific or of low intensity supporting the amorphous nature of adsorbed PO 4 3-ion and integral CO 3 2-remaining originating from CaCO 3 in LKD as shown in Fig. 5b.
Tartaric acid (2,3-dihydroxysuccinic acid, further in the text referred to as TA) is a carboxylic acid that exists in various plants, such as grapes and tamarinds [67]. It represents a useful dicarboxylic acid molecule which in some wastewater streams has been found with concentrations of up to 1300 ppm [68] and has been shown, in general, to inhibit PO 4 3sorption and crystallization of the solid products, such as struvite [5]. Fig. 6a shows the concentration profile (q, mg/g) of PO 4 3-ions adsorbed from 600 MAP solution by BA, LKD or LKD in the presence of 500 ppm TA.
It was observed that most of the PO 4 3-uptake by 1800 ppm LKD, 1800 ppm LKD in the presence of 500 ppm tartaric acid and 2200 ppm BA occurs within ~15 min. The equilibrium amount of PO 4 3-removed from solution was 63% on LKD, 53% on LKD in the presence of tartaric acid and 42% BA. After the initial 25 min, the uptake further increased albeit at a much slower rate. The LKD exhibited the fastest initial rate of PO 4 3-removal with the PO 4 3-almost completely removed after 90 min. In the case of BA and LKD with tartaric acid, the rate of PO 4 3-removal was slower and achieved much lower equilibrium values.
The pH of the solution was measured to provide critical information on PO 4 3-adsorption and the following precipitation reactions and the data are shown in Fig. 6b. In particular, pH quickly rose to 8 and 9 for 2200 ppm BA and 1800 ppm LKD, respectively. At these pH values, hydroxyapatite has been shown to form [69]. Only for the highest concentration of OA did pH remain low around 3. Oxalic acid had the lowest pKa value among the organic acids tested. In this case, the highest pH value of 3 was achieved after 120 min with 2200 ppm BA while with 1800 ppm LKD it was 7.5 showing that BA had a much lower buffering capacity to provide alkaline ions needed to maintain elevated pH. This suggests that the presence of the organic acids will strongly influence the PO 4
3-ion removal by modulating solution pH, especially for BA and much less for LKD. However, only OA at the highest concentrations can inhibit phosphate adsorption due to the pH as the rest of the acids at 1000 ppm exhibited much higher pH values.
Further understanding of the sorption kinetics involved was pursued since it is necessary for the design of any large-scale adsorption facilities. Many models are used to fit PO 4 3-adsorption experiments to obtain quantitative kinetic parameters. Furthermore, to examine the mechanism of the adsorption processes, such as mass transfer and chemical reaction, a suitable kinetic model is needed to be applied to obtain the rate constants. In this work, two kinetic models [70][71][72] were used to model the adsorption kinetics of PO 4 3-onto BA and LKD. The kinetic equations including pseudo-second-order model [72], and Elovich model [71] used are described as follows Pseudosecondorder equation :
Elovich equation :
where q t is the adsorbed amount at time t, mg/g, q e is the adsorbed amount at equilibrium, mg/g, k 2 is the rate constant of pseudo-secondorder adsorption, g/(mg h); the parameter a e is the initial adsorption rate, mg/(g h), and b e is related to extent of surface coverage and activation energy for chemisorption, g/mg. The relative parameters (the derived rate constants together with the correlation coefficient R 2 ) of the pseudo-second-order and Elovich are in Table 3. Constant k 2 was calculated from the intercept of the line by plotting t/q t versus t in the pseudo-second-order model, while the initial adsorption rate a e was calculated from the intercept of the line obtained by plotting q t versus lnt in the Elovich equation. Figs. 7 and8 show the corresponding plots in linearized form. Notably, pseudo-first order fits resulted in low R 2 values likely since pseudo-first order model is only valid in scenarios where the change in the bulk solution concentration of the adsorbate is small [70,73]. As such, the physical insight gained from the pseudo-second order fitting is that the adsorption is surface-reaction limited. Compared to previously reported natural minerals, such as Ca-rich sepiolite, which exhibits a q e value of 32 mg P/g, and goethite which exhibited 34 mg P/g, both LKD and BA show higher phosphate adsorption capacities (Table 4) [20,74]. The higher q e value is indicative of an adsorbent that can be utilized more efficiently given that a smaller mass of adsorbent can bind a higher concentration of PO 4 3-. The addition of tartaric acid leads to a reduction in the q e for LKD, indicating that phosphate adsorption is hindered by the presence of the acid, due to surface dissolution of the adsorbent. In addition to the lower q e value, a slightly lower pseudo-second-order rate constant is also observed for LKD in the presence of tartaric acid due to the suppressed PO 4 3adsorption.
Comparing the correlation coefficients of two kinetic models, it is revealed that the pseudo-second-order model best fits the adsorption kinetic of PO 4 3-on LKD (R 2 = 0.99), LKD with tartaric acid (R 2 = 0.99), and BA (R 2 = 0.97).
Due to the marked difference observed in the equilibrium amount and the corresponding kinetics of PO 4
3-adsorption, sorption experiments were also performed utilizing other common organic acids present in the wastewater solutions from natural or anthropogenic sources. Those included, in addition to tartaric acid, citric, glycolic, oxalic and citric acids in the concentration range from 50 to 1000 ppm in the initially simulated wastewater also containing 600 ppm of MAP. The equilibrium removal results (e.g. maximum amount of PO 4 3removed after the equilibrium was achieved) are tabulated in Table 5. Interesting dynamics were observed where citric and oxalic acids strongly inhibited the adsorption of PO 4 3-on 1800 ppm of LKD while glycolic acid did not result in strong inhibiting effects. Acid concentration dependence was also observed within the range of acids analyzed with 1000 ppm resulting in the greatest inhibition. Interestingly, oxalic acid exhibited a dramatic inhibition of PO 4 3-removal in the presence of 2200 ppm of BA. Two phenomena can be proposed to explain these data. First, solution pH, as suggested by Fig. 6b, remains at pH = 3 in the presence of the lowest pKa oxalic acid at 1000 ppm thus inhibiting precipitation. This effect is much less pronounced for LKD suggesting LKD possesses a much greater capacity to maintain the alkalinity of the solution pH. Second, organic acids strongly bind to the surface sites 3-removal, likely calcium atomscalcium is known to forming binding complexes with organic acids, such as oxalic [81], and thus inhibits the overall process. Furthermore, citric, tartaric, and oxalic acids can lead to the surface dissolution of the adsorbent, which would lead to lower PO 4 3-removal.
Raman characterization was performed of the 1800 ppm LKD surfaces reacted with 600 ppm MAP in the presence of tartaric, citric, glycolic and oxalic acids at and high initial concentrations of acids (50, 500 and 1000 ppm). The results are shown in Fig. 9 for several spots analyzed within each sorbent material. The corresponding complex spectra are grouped according to the corresponding main vibrations to exhibit the trends. The notable peak at 948-952 cm -1 is due to the adsorbed PO 4
3- [82]. Other peaks were assigned to those due to the inorganic ions HPO 4 -and CO chromatography measurements. Similarly, tartaric acid interacted strongly with LKD at 1000 ppm concentration in the initial solution. Glycolic acid did not inhibit PO 4
3-adsorption and did not exhibit strong binding with the LKD. Citric acid showed very different and unexpected behavior. While it strongly inhibited PO 4 3-adsorption from solution at higher concentrations of 1000 ppm, it was not measured in detectable amounts on the sample surface [88]. The lack of citric acid in the final product indicates that the PO 4 3-adsorption was hindered due to the surface dissolution of LKD by citric acid rather than competitive adsorption, hence, it is not present adsorbed on the solid surface. If citric acid had adsorbed on the surface and blocked the PO 4 3-adsorption sites, Raman spectroscopy would have detected the relevant C-H vibrations from the citrate ions, and the lack of such peaks indicates that the citric acid causes dissolution to hinder PO 4 3-adsorption on LKD. From pKa values for the first deprotonation of each acid oxalic acid (1.2), citric acid (3.1), and tartaric acid (2.98), show stronger acidity compared to glycolic acid (3.98) [89][90][91][92]. However, calcium tartrate is a low-solubility mineral that is known to precipitate at low Ca 2+ concentrations, and thus, its presence in the product along with LKD is expected as it can bind to surface Ca sites [93]. Similarly, the Raman spectra shown in Fig. 9c contained peaks at 830, 911, 1480, 1630 cm -1 which have been reported for calcium oxalate dihydrate crystals, revealing that calcium oxalate formation can block PO 4 3-adsorption [94]. Therefore, a combination of organic acid adsorption on Ca-sites and adsorbent dissolution has led to the suppression of PO 4 3-removal by LKD.
The chemical composition of the resulting effluents after the reaction of 2200/600 ppm BA/MAP and 1800/600 ppm LKD/MAP without organic acids is shown in Table 6. It can be seen that, most importantly, ~44 and 9 mg/l of P are still present in the solution, consistent with data in Fig. 1 (expressed as PO 4 3-and measured using ion chromatography). This is higher than that required by some USA states with heavy nutrient loading due to the agriculture, such as Wisconsin, of ~1 mg/L. This suggests the effluent needs to undergo secondary treatment to remove the residual phosphorus. This suggests that original solid material, while reconstructing physically as shown in Fig. 2c andd SEM images, largely retains its primary elemental composition.
Accumulation of heavy metals in the soil through leaching from the rapidly expanding anthropogenic sources, including but not limited to land application of fertilizers, animal manures, sewage sludge, wastewater irrigation, is of major environmental concern [95]. The corresponding concentration of heavy metals in as received BA and LKD powders was measured and is shown in Fig. 10. Six major heavy metals were detected including cadmium (Cd), lead (Pb), nickel (Ni), copper (Cu), zinc (Zn) and chromium (Cr). BA contained ~x10 higher concentrations of microelements and heavy metals with ~700 mg kg -1 of Zn alone, consistent with recent reports [96,97]. While some metals, such as Cu and Zn, are essential to plant growth and only toxic to plants at high concentrations, others such as Cd, Pb, Cr or Ni, have a chiefly toxic effect on living organisms. Both BA and LKD contained Cd, Pb, Cr or Ni from few to tens of ppm.
The measured concentrations of leached metals varied between LKD and BA with BA releasing more metals, especially Zn. Notably, the mobilization of the heavy metals was essentially unchanged when LKD and BA were reacted with organic acids with and without the presence of MAP (not shown), which confirms that the surface dissolution of the adsorbent occurs as discussed above, and is the main driver for releasing Bank for wastewater effluents, showing that LKD can be utilized successfully for PO 4 3-removal in wastewater with similar organic acid conditions without interfering with legal limits of heavy metal contamination [98,99]. In the case of BA, the release of Zn was shown to be above the EPA regulated level of 0.497 ppm when citric and oxalic acids were present in 1000 ppm concentrations. However, tartaric and glycolic acids did not cause Zn release to exceed the legal limit.
Both LKD and BA are generated in large quantities and contain important major nutrients as well as regulated heavy metals. This manuscript represents work to evaluate their feasibility in PO 4 3-ion adsorption from aqueous solutions. Batch experiments showed that PO 4
3-adsorption proceeds in two regimes while kinetics were shown to fit pseudo second order Estimated sorption capacity was 55.6 and 35.3 mg P/g for LKD and BA, respectively, comparable or higher than for other natural Ca-containing natural minerals reported in the literature. The presence of organic acids, found in wastewater, except glycolic, inhibited the amount of PO 4
3-ion adsorbed for the relevant acid concentration of up to 1000 ppm. Heavy metal, including Zn, Cu, Cr, Ni, Pb, and Cd, leaching was observed into the aqueous solution largely below the maximum allowable concentrations regulated by EPA. A notable exception was Zn from BA due to its high initial concentration in the material. Interestingly, the amount of the metals released varied only slightly between the experimental conditions (with or without PO 4 3-ion or organic acid) suggesting that the chief cause was a strong physical dissolution of the original Ca-containing material rather than competitive PO 4 3-ion/organic acid adsorption. The data presented here suggest that PO 4 3-ion adsorption on calcium-containing waste materials takes place efficiently via chemisorption with intra-particle diffusion and external mass transfer being rate-limiting steps. This is likely related to the change in particle shape