Phosphate Removal from Aqueous Solutions Using Calcium Silicate Hydrate Prepared from Blast Furnace Slag

Abstract

Discharging phosphorus through wastewaters into waterbodies holds potential risks of eutrophication and adverse effects on aquatic organisms. On the other hand, phosphorus is an indispensable element needed for all life forms and plant growth, while phosphate rock reserves available are anticipated to be depleted in the near future. Herein, we report phosphate removal from aqueous solution using a calcium silicate hydrate (CSH) adsorbent prepared from blast furnace slag (BF slag), a high-volume byproduct produced in iron-making industry. A high-surface-area CSH was synthesized through a facile two-step dissolution-precipitation method under controlled conditions (at 373 K and pH 11.0) using BF slag as a sole metal source. Adsorption experiments demonstrated that the slag-made CSH showed a maximum phosphorus adsorption capacity of 53.1 P-mg/g under standard adsorption conditions (pH 7.0 and 298 K), which was 73 times greater than that of BF slag. Owing to the presence of abundant Ca²⁺ species reactive with phosphate ions, the adsorbent showed excellent phosphorus adsorption performances in a broad range of conditions, particularly even from low-concentrated phosphate solutions and under a wide range of pH conditions. Kinetic analysis demonstrated that the adsorption of phosphate onto slag-made CSH adsorbent can be well fitted with pseudo-second-order kinetic model and Freundlich adsorption isotherms. This study demonstrates that the CSH synthesized from BF slag can be used as a promising adsorbent for efficient bulk wastewater treatment and phosphorus recovery.

1. Introduction

Discharging phosphorus through wastewaters into waterbodies holds potential risks of eutrophication and the serious effects on aquatic organisms and the surrounding ecosystem. On the other hand, phosphorus is a non-substitutable, essential element for all life forms and plant growth. A rapidly growing population in developing countries and concomitant demand for food production are expected to require continuous increase of phosphorus fertilizer supply. However, assuming current rates of use, world phosphate rock reserves and resources are available for the next 90–130 years and would be depleted in the near future.^1,2) However, it has been reported that the actual utilization efficiency of phosphorus is low; only 15–20% of applied phosphorus is used by crops and animals, and the remaining phosphorus unproductively runs off through various waste streams, such as agricultural and municipal wastewaters and industrial effluents (particularly fertilizer industries).³⁾ If the phosphorus in these waste streams is economically recovered, it can contribute to a sustainable management of both water and phosphorus resources. Therefore, technologies that enable the removal and recovery of phosphorus from P-containing wastewater is strongly required.⁴⁾

Technologies to remove phosphorus from wastewater streams developed so far include tertiary filtration, precipitation, coagulation, flocculation, crystallization and biological removal.^5,6,7) Of all the methods, precipitation using coagulants (lime and Al or Fe salts) and alkaline chemicals has extensively been a majority in municipal sewage treatment facilities, by which phosphorus presents in wastewater in the form of orthophosphates and condensed phosphates can be recovered as hydroxyapatite (HAP: Ca₁₀(PO₄)₆(OH)₂) or magnesium ammonium phosphate (MAP: MgNH₄PO₄) precipitates.⁸⁾ However, Mg, Al and Fe salts cannot remove the phosphate below a particular concentration due to low reactivities, requiring large volumes of these chemicals and hence resulting in a large volume of sludge production. Furthermore, because of the high cost of these chemicals, the overall economics of this method cannot compete with the market price of phosphorus, which makes the recycling of phosphorus difficult. Adsorption using high-volume industrial wastes has attracted increasing concerns as an alternative phosphorus recovery process, because of simple manipulation, less sludge production and economic competitiveness. To date, various kinds of industrial wastes were examined as low-cost phosphorus adsorbent, including fly ash,^9,10) red mud,¹¹⁾ oil-shale ash,¹²⁾ acid mine drainage sludge,¹³⁾ and dolomite;¹⁴⁾ however, the adsorption capacities of these adsorbents were considerably low.

Blast furnace slag (BF slag) also has a potential as an adsorbent for removing phosphate from wastewater. BF slag is a high-volume byproduct resulting from iron-making process, in which approximately 290 kg of BF slag is produced per ton of pig iron in blast furnace. In Japan, BF slag has currently been recycled in civil engineering work as secondary source for hydraulic cement and concrete aggregate, rather than being deposited in landfills.¹⁵⁾ However, finding effective and novel strategies for recycling BF slag has still been a matter of great concern, because of its high-volume production (approx. 24 million tons in 2015 in Japan), continuously increasing production rate of iron and steel as well as the severe environmental regulation against slag-deposition.¹⁵⁾ BF slag contains CaO, SiO₂, Al₂O₃, MgO and some minor metallic elements (such as Fe, Ti and Mn), which potentially act as adsorbents for removing phosphate from aqueous solutions. However, the phosphorus adsorption capacity of BF slag reported in literature is less than 10 mg-P/g, which might be due to its granular form and low surface area.^16,17,18)

Regarding the above issue, we recently reported a facile conversion process of BF slag into a calcium silicate hydrate (CSH) which contains the slag-derived metals in its structure.¹⁹⁾ CSH comprises a disordered three-dimensional network of Ca–O polyhedra that accommodates Si–O tetrahedral chains, together with interspersed water molecules,²⁰⁾ and has recently found applications as functional adsorbent for removal of dyes and heavy metals from aqueous solutions.^19,21) Synthetic CSH compounds has recently been examined in phosphorus recovery from wastewater as crystal seeds, since Ca²⁺ and OH⁻ ions released from CSH react with phosphate species to produce HAP crystals on the surface.^22,23) Since the slag-made CSH has a higher specific surface area and unique composition, it is expected to show a superior performances for phosphorus adsorption compared with pure synthetic analogue previously reported. Furthermore, application of the slag-made CSH as an alternative low-cost adsorbent for phosphorus would not only lead to a cost-effective phosphorus removal and recovery process, but also contribute to the sustainable management of BF slags produced in iron-making industry. In this study, CSH synthesized from waste BF slag is examined in the recovery of phosphate from simulated wastewater solutions in order to evaluate its practical availability as a phosphorus adsorbent. Effect of adsorption conditions including adsorption time, solution pH, phosphorus concentration and adsorbent amount on phosphate adsorption properties are studied in batch adsorption experiments. In addition, a plausible adsorption mechanism is proposed based on kinetic analyses and material characterizations, and a possible environmental application of this material is discussed.

2. Experimental

2.1. Materials

Water-granulated blast furnace slag (BF slag) used as the source for material synthesis was kindly supplied from Nippon Steel & Sumitomo Metal Corp. The chemical composition of BF slag used is listed in Table 1. Other chemicals for material synthesis and adsorption experiments were purchased from Wako Pure Chemical Ind. Ltd. and were used without further purification.

Table 1. Chemical compositions of raw BF slag used in this study (mass%).

CaO	SiO2	Al2O3	MgO	Fe2O3	TiO2	MnO	Total	C/Sa
40.09	34.58	14.78	5.29	1.53	0.78	0.27	97.32b	1.24

^a  CaO/SiO₂ molar ratio.

^b  Other components may include sulphur, chloride and sodium.

2.2. Preparation of SlagCSH

The conversion of BF slag into CSH was performed by a two-step dissolution-precipitation process as schematically illustrated in Fig. 1.¹⁹⁾ Prior to use, the raw BF slag was ball-milled at 650 rpm for 10 min and screened using a 45 μm mesh to facilitate the following dissolution process. In a typical synthesis, 10.0 g of BF slag was dissolved in 200 mL HCl aqueous solution (3.0 mol/L) by stirring at room temperature for 2 h (Fig. 1, Step 1). A transparent yellow solution was obtained, suggesting that all slag-derived metals were fully dissolved. The pH of this solution was adjusted to 11.0±0.1 by adding NaOH aqueous solution (2.0 mol/L). The obtained precipitate was transferred to a sealed Teflon bottle and then aged at 373 K for 6 h under static condition (Step 2), followed by washing with distilled water several times and drying at 373 K overnight to yield 10.9 g of slagCSH (Step 3). Based on the material balance evaluation, the metal recovery rate was calculated to be 85% (based on metals).

Fig. 1.

Comparison of synthetic procedures for calcium silicate hydrate (CSH) and layered double hydroxide (LDH) from BF slag.

For comparison, pure CSH was also prepared by a coprecipitation method.¹⁹⁾ 0.1 mol of sodium metasilicate (Na₂SiO₃·9H₂O) and 0.1 mol of calcium chloride (CaCl₂) were dissolved in 200 mL of distilled water and magnetically stirred for 2 h at pH 11.0±0.1 and at 373 K. The resulting precipitate was further aged for 6 h at 373 K in a Teflon bottle, followed by filtration, washing with distilled water and drying at 373 K overnight to afford pureCSH as a white powder. The chemical compositions of slagCSH and pureCSH are listed in Table 2.

Table 2. Chemical compositions of slagCSH and pureCSH (mass%).^a

Sample	Ca	Si	Al	Mg	Fe	Ti	Mn	Physisorbed H2O
slagCSH	19.9	15.1	5.9	3.0	0.23	0.25	0.14	12.0
pureCSH	28.0	20.1	–	–	–	–	–	11.2

^a  Determined by the combination of EDX and thermogravimetric analyses. Other components may include oxygen atoms, hydroxyls and traces of Na and Cl atoms.

2.3. Characterizations

Crystal phase was identified by using X-ray diffraction (XRD) measurement. XRD patterns were recorded using a Rigaku Ultima IV diffractometer with CuKα radiation (λ=1.54056 Å) operated at step size 0.02° over the 2θ range 10–70°. The morphology of the samples were observed by a field emission scanning electron microscopy (FE-SEM) in a JEOL JSM-6500 equipped with an energy dispersive X-ray (EDX) fluorescence spectrometer. Nitrogen adsorption–desorption measurement was performed at 77 K using BELSORP-max system (MicrotracBEL, Corp.) to determine specific surface area. Prior to measurement, the samples were outgassed under vacuum at 473 K to eliminate physisorbed water molecules. The BET (Brunauer-Emmett-Teller) method was applied to calculate specific surface areas.

2.4. Adsorption Test

Adsorption experiments were conducted in a set of quartz vessels placed in an incubator using 200 mL simulated phosphate solutions (0.5–8.0 mmol/L) and adsorbent dosages of 0.1–3.5 g/L at room temperature (298±1 K) and at initial pH of 7.0.²⁴⁾ In the standard conditions, the adsorption tests were performed with an adsorbent dosage of 0.5 g/L and an initial phosphate concentration of 4.0 mmol/L (adjusted using Na₂HPO₄·12H₂O). The suspensions were agitated for 24 h at a stirring speed of 200 rpm. A portion of the solution was collected at predetermined time intervals by filtration, and the residual phosphorus concentration was measured by employing the molybdenum blue ascorbic acid method using a Shimadzu UV-2600 UV-vis spectrophotometer. The amounts of phosphorus adsorbed per unit mass of adsorbent at given time t (q_t (P-mg/g)) and at equilibrium (q_e (P-mg/g)), phosphate removal efficiency (R_eff (%)) were calculated according to the following equations.   

$$q_{\mathrm{t}}=\left(C_0–C_{\mathrm{t}}\right)V/M$$(1)

  

$$q_{\mathrm{e}}=\left(C_0–C_{\mathrm{e}}\right)V/M$$(2)

  

$$R_{\mathrm{e}\mathrm{f}\mathrm{f}}=\left(C_0–C_{\mathrm{t}}\right)/C_0\times 100$$(3)

where C₀ (P-mg/L), C_t (P-mg/L) and C_e (P-mg/L) are the phosphorus concentrations at the initial time, at given time t and at equilibrium conditions, respectively, which are all determined by UV-vis spectrophotometer. V (L) and M (g) are the volume of the phosphate solution and the mass of adsorbent, respectively. Adsorption experiments were performed in different pH conditions and with different adsorbent dosages to study the effects on the adsorption performances, and the obtained adsorption data were analyzed by applying Langmuir and Freundlich isotherm models to determine adsorption kinetic parameters. To check structural and compositional alterations upon the phosphate adsorption, the used adsorbent retrieved from the solution by filtration and drying was further characterized by means of XRD, FE-SEM and EDX.

3. Results

3.1. Characterization of Slag-made CSH

Figure 2 shows XRD patterns of the slagCSH and pureCSH samples. The slagCSH showed clear diffraction peaks at 16.5, 29.3, 32.0, 43.1, 49.8 and 55.1° which are well consistent with those of pureCSH, verifying a successful formation of CSH. CSH is known to transform into various crystalline calcium silicate phases (such as tobermolite (Ca₅Si₆O₁₆(OH)₂·4H₂O), wollastonite (CaSiO₃) and belite (Ca₂SiO₄)) by thermal treatment.²⁵⁾ Considering the crystalline structures and diffraction intensity of the solid, the product is considered to be a low-crystalline amorphous CSH. Minor peaks seen at 11.2 and 60.8° can be indexed to Mg–Al hydrotalcite, a layered double hydroxide compound (LDH) consisting of Mg(OH)₆ and Al(OH)₆ octahedra, which might be formed in the early stage of precipitation due to their lower equilibrium product constants than that of Ca.²⁶⁾ Formation of Ca-based LDH (i.e. hydrocalumite) which is obtained from BF slag via a similar synthetic route (represented as “previous study” in Fig. 1)^26,27) was not observed because of fast reactivity of Si with Ca to form CSH phase. EDX elemental analysis ascertained that slagCSH contains Ca, Si as well as minor slag-derived metals (for EDX analysis data, see Table 2), whereas pureCSH is composed of Ca and Si. These minor metals are likely to be uniformly incorporated into the calcium silicate network during the alkali-precipitation process. The FE-SEM image of slagCSH showed agglomerated particles consisting of some irregular flakes and fibrils, which is so-called “desert rose”-like morphology (for FE-SEM image, see also inset in Fig. 7(a)), being in line with the typical characteristics of CSH gels.^27,28) The specific surface area (S_BET) and total pore volume (V_total) of slagCSH were calculated to be 219 m²/g and 0.78 cm³/g, respectively, while raw BF slag was considered as non-porous (S_BET = 8.1 m²/g, V_total = 0.016 cm³/g). Such a high surface area and high porosity is attributed to the slit-like pores formed in flake/fibril particles, which are expected to be effective for adsorption of phosphate ions.

Fig. 2.

XRD patterns of (a) slagCSH and (b) pureCSH (♦: CSH, +: Mg–Al hydrotalcite).

Fig. 7.

(Left) XRD patterns of slagCSH (a) before and (b) after phosphate adsorption test (♦: CSH, +: Mg–Al hydrotalcite, inset shows the corresponding FE-SEM images) and (Right) the corresponding EDX spectra.

3.2. Adsorption of Phosphate

In preliminary phosphate adsorption experiments, 4.0 mmol/L phosphate aqueous solution (corresponding to phosphorus concentration of 120 ppm) was used as a model solution, simulating the real wastewater discharged in the sewage sludge dewatering process.²⁴⁾ Figure 3 shows adsorption kinetics in the phosphate adsorption using slagCSH, pureCSH and raw BF slag at 298 K and at initial pH of 7. Raw BF slag showed negligible amount of phosphate adsorbed during 24 h of adsorption (0.73 P-mg/g). On the other hand, slagCSH adsorbent showed rapid adsorption of phosphate within the first 1 h, followed by gradual increase to reach adsorption equilibrium. Phosphorus adsorption capacity of slagCSH after 24 h of adsorption was determined to be 53.11 P-mg/g, which was 73 times greater than that of BF slag. This phosphorus adsorption capacity is markedly higher than those previously reported on solid adsorbents derived from minerals and industrial wastes,^{9,10,11,12,13,14)} demonstrating its excellent adsorption ability toward phosphate ions in water. PureCSH as a reference sample exhibited a higher phosphorus adsorption capacity (86.23 P-mg/g) than slagCSH irrespective of its subpar structural properties (S_BET = 154 m²/g, V_total = 0.62 cm³/g). The reduced phosphorus adsorption capacity of slagCSH is probably due to the following two reasons; i) the presence of impurity metals in CSH network that decreases the proportion of adsorption sites effective for phosphate adsorption and ii) the formation of microcrystalline Mg–Al hydrotalcite phase possessing a lower phosphate adsorption capacity than CSH.²⁸⁾

Fig. 3.

Phosphate adsorption kinetic data for slagCSH, pureCSH and raw BF slag (adsorbent dosage = 0.5 g/L, phosphate concentration = 4.0 mmol/L, 298 K, initial pH 7.0).

The obtained adsorption kinetic data was then analysed by pseudo-first-order and pseudo-second-order kinetic models to provide better understanding toward the adsorption process. The pseudo-first-order kinetic model can be expressed as follows:   

$$log(q_e-q_t)=log{q}_e-\left(\frac{k_1}{2.303}\right)t$$(4)

where q_e and q_t (P-mg/g) are the amounts of phosphorus adsorbed per unit mass of adsorbent at equilibrium and at specific adsorption time t, respectively. k₁ (min⁻¹) is the first order rate constant for the present adsorption process. The q_e and k₁ values were calculated from the intercept and the slope of the plots of log(q_e–q_t) versus t, respectively, and were listed in Table 3. The calculated q_e values do not agree with the experimental q_e values, and coefficient (R²) values significantly deviate from unity, suggesting that phosphate adsorption onto CSH compounds does not follow pseudo-first-order model. On the contrary, pseudo-second-order model provided a better fitting of the adsorption data, which can be expressed by the following equation:   

$$\frac{t}{q_t}=\frac{1}{k_2{q}_e^2}+\left(\frac{1}{q_e}\right)t$$(5)

where k₂ (g/P-mg/min) is the rate constant of the pseudo-second-order adsorption and q_e (P-mg/g) is the amount of phosphorus adsorbed per unit mass of adsorbent at equilibrium. Kinetic parameters q_e and k₂ were calculated from the slope and the intercept of the plots of t/q_e versus t, respectively, and were listed in Table 3. The close agreement between the calculated q_e values and the experimental q_e values, and the proximity of correlation coefficient (R²) to unity (R² > 0.98) suggest that adsorption of phosphate onto CSH materials better follows pseudo-second-order kinetic model.

Table 3. Fitting parameters of phosphate adsorption kinetic data for slagCSH, pureCSH and raw BF slag.

Adsorbent	qe (exp) (P-mg/g)	Pseudo-first-order			Pseudo-second-order
		qe (P-mg/g)	k1 (min−1)	R2	qe (P-mg/g)	k2 (g/P-mg/min)	R2
slagCSH	53.11	24.83	18.0×10−4	0.901	45.68	5.56×10−4	0.998
pureCSH	86.23	36.95	32.5×10−4	0.909	83.47	4.35×10−4	0.998
raw BF slag	0.73	0.48	54.6×10−4	0.817	0.78	2.07×10−2	0.983

3.3. Effect of Solution pH

The solution pH frequently provides critical impacts on phosphate adsorption performances. Figure 4(a) shows the effect of initial solution pH on phosphorus adsorption capacity, q_e, and final solution pH. A considerable increase of phosphorus adsorption capacity was observed in acidic solution conditions (initial pH < 7) and the maximum phosphorus capacity of 97.62 P-mg/g was attained when initial solution pH was 3.0. It should be noted that slagCSH retained its great phosphorus adsorption capacity even under basic pH conditions (initial pH > 7). Figure 4(b) shows the effect of initial solution pH on the amounts of Ca²⁺ and Si⁴⁺ eluted during 24 h of adsorption. A substantial amount of Ca²⁺ and Si⁴⁺ ions was detected from the reaction solution over the whole pH range, and an increase of solution pH to basic condition (pH = 8–9) was observed after 24 h of adsorption, indicating an elution of Ca²⁺ and Si⁴⁺ ions due to the dissolution of CSH network and a concomitant release of OH⁻ ions. In the presence of alkaline (OH⁻), insoluble calcium phosphate species are readily formed as a result of reaction between eluted Ca²⁺ ions and phosphate ions, hence leading to prominent phosphorus capacities.²⁹⁾ Particularly, at initial solution pH of 3.0, 20.7 ppm of Ca²⁺ and 21.3 ppm of Si⁴⁺ eluted were detected, which correspond to 10.4 mol% and 14.1 mol% fractions of total Ca and Si content in the slagCSH adsorbent, respectively. Under acidic pH conditions where the dissolution of CSH network is promoted, slagCSH supplies a larger amount of Ca²⁺ ions which can act as crystal seeds, thereby resulting in an increased phosphorus capacity.

Fig. 4.

(a) Effect of initial solution pH on phosphorus adsorption capacity and final pH value of the solution after 24 h. (b) Effect of initial solution pH on the amounts of Ca²⁺ and Si⁴⁺ eluted during 24 h (adsorbent dosage = 0.5 g/L, phosphate concentration = 4.0 mmol/L, 298 K).

3.4. Effect of Adsorbent Dosage

Figure 5 illustrates phosphate removal efficiency (R_eff) plotted against the adsorbent dosage. The R_eff value increased by increasing adsorbent dosage due to the availability of more adsorption sites, reaching 97–99% removal rate from every simulated phosphate solutions with concentrations of 0.5–8.0 mmol/L (corresponding to 15–250 ppm as phosphorus). From 4.0 mmol/L (= 120 ppm as phosphorus) phosphate solution simulating the real wastewater in sewage treatment process, 99% phosphate removal efficiency was achieved when 2.5 g/L adsorbent dosage was applied, showing a great adsorption performance of slagCSH to remove phosphate from real wastewater streams. Even from low-concentrated phosphate solution (0.5 mmol/L), 97% of phosphate removal efficiency was achieved with 1.5 g/L adsorbent dosage.

Fig. 5.

Adsorption of phosphate on slagCSH as a function of adsorbent dosage (298 K, initial pH 7.0).

3.5. Study of Adsorption Mechanism

To study the adsorption mechanism, the phosphate adsorption data collected above was fitted to Langmuir and Freundlich adsorption isotherm models (Fig. 6). The Langmuir isotherm equation is represented as follows.   

$$q_e=\frac{b{C}_e}{1+b{C}_e}{q}_m$$(6)

where C_e (P-mg/L) is the equilibrium concentration of phosphorus, and q_e (P-mg/g) is the amount of phosphorus adsorbed per unit mass of adsorbent at equilibrium. q_m (P-mg/g) is the Langmuir constant representing maximum adsorption capacity assuming monolayer coverage of adsorbate over a homogenous adsorbent surface, and b is a kinetic parameter representing adsorption energy of the adsorbent for the adsorbate molecule. The kinetic parameters q_m and b were calculated from the slope and intercept of the linear plot of 1/q_e versus 1/C_e, respectively (Fig. 6(a)), which were listed in Table 4. Langmuir theory predicted that the maximum adsorption capacity for phosphorus (q_m) onto slagCSH is 75.7 P-mg/g, showing a deviation from the equilibrium adsorption capacity obtained from the adsorption experiment (q_e (exp)) (Table 3).

Fig. 6.

(a) Langmuir and (b) Freundlich adsorption isotherms of phosphate on slagCSH. Symbols represent experimental data and lines represent fitting data calculated by each model equation.

Table 4. Langmuir and Freundlich parameters and correlation constants for phosphate adsorption onto slagCSH.

Langmuir isotherm model		Freundlich isotherm model
b (L/P-mg)	0.191	1/n	0.222
qm (P-mg/g)	75.70	Kf (P-mg/g)	27.71
R2	0.836	R2	0.957

Compared to Langmuir isotherm model, Freundlich isotherm model provided a better fitting for the adsorption data. Freundlich isotherm model is an empirical adsorption model assuming adsorption of adsorbate on heterogeneous surface, which can be represented by the following equation.   

$$ln{q}_e=\frac{1}{n}ln{C}_e+ln{K}_f$$(7)

where q_e (P-mg/g) is the amount of phosphorus adsorbed per unit mass of adsorbent, C_e (P-mg/L) is the equilibrium concentration of phosphorus, and K_f (P-mg/g) is a Freundlich constant related to adsorption capacity and 1/n is an empirical parameter related to adsorption intensity or surface heterogeneity. As shown in Fig. 6(b), the plot of ln q_e versus ln C_e provided a linear correlation with a high R² value (0.957), suggesting that the phosphate adsorption onto slagCSH better follows Freundlich adsorption isotherm. The value of 1/n was determined to be 0.222, which was less than unity (Table 4), representing that adsorbent has a heterogeneous energy distribution of adsorption sites. As shown in Fig. 3, the adsorption of phosphate onto slagCSH exhibits a rapid adsorption within the first 1 h and then slowly reaches equilibrium, representing that two different adsorption processes are working successively. In the initial stage, diffusion and adsorption of phosphate ions from the solution onto the adsorbent surface (probably surface-exposed Ca²⁺ sites) rapidly takes place through the pores of the CSH structures. As the adsorption proceeds, the degree of surface site occupation increases, and thus the binding affinity towards phosphate ions gradually decreases, but in contrast, Ca²⁺ ions as crystal seeds are gradually released from the adsorbent into the solution. Therefore, in the later stage, adsorption likely proceeds via a gradual layer-by-layer adsorption process on the adsorbent surface and a calcium phosphate precipitation process in the aqueous phase, both of which are most likely working simultaneously.

The adsorbent retrieved from reaction solution was further analysed by XRD, FE-SEM and EDX. The P-adsorbed slagCSH sample retained its original crystalline structure attributed to low-crystalline CSH phase, and no crystalline phases attributed to calcium phosphate compounds were observed in XRD patterns (Fig. 7(b), Left). FE-SEM images verified taht the P-adsorbed slagCSH still retained “desert rose”-like morphology which is identical to that of original slagCSH (inset of Fig. 7(b)), confirming high stability of the particles. The elemental mapping images verify that used slagCSH contains substantial amount of phosphorus which is derived from the adsorbed phosphate ions, together with Ca, Si, Al, Mg and Ti as the original components of slagCSH, with homogeneous distributions (Fig. 8). Detailed analysis by means of EDX (Fig. 7(b), Right) revealed that the phosphorus content after 24 h of adsorption in the recovered slagCSH reached 3.4 wt%, and the corresponding P/Ca molar ratio was determined to be 0.22 (mol/mol%), which was far less than 1.0. These combined analyses conclusively suggest that the phosphate ions are fixed onto the adsorbent surface in the form of amorphous calcium phosphate species, where the surface exposed Ca²⁺ sites and Ca²⁺ ions eluted into aqueous solution are the main active species.^22,23)

Fig. 8.

Elemental mapping images of slagCSH after phosphate adsorption.

3.6. Future Perspectives

Owing to an excellent and stable adsorption property especially even from low-concentrated aqueous phosphate solutions, the CSH synthesized from BF slag can be applied as an effective phosphate adsorbent in the real municipal sewage treatment facilities. Since the synthetic process is facile and requires BF slag as a sole metal source, the cost for adsorbent synthesis can be reduced which may be promising for large-scale implementation. Furthermore, the slagCSH recovered after phosphorus adsorption contains not only P (3.4 wt% as phosphorus) but also Ca, Si, Mg, Fe and Mn that are important minerals for plant growth, suggesting the possibility of used adsorbent as a phosphate fertiliser in agricultural field. The potential environmental risks might be negligible, since raw BF slag has been used as a chemical fertilizer, especially effective for rice cultivation, over the ages. From these perspectives, we believe that phosphorus removal from wastewater streams using slagCSH and reuse of the recovered adsorbent as a phosphate fertilizer in agricultural field would be a reasonable option to contribute to a sustainable management of BF slag and phosphorus resources,⁴⁾ although further studies on cost reduction and waste management problems for adsorbent synthesis in large-scale are needed to substantiate this technology.

4. Conclusions

This study demonstrated that the calcium silicate hydrate (CSH) synthesized from BF slag can be used as a promising adsorbent to remove phosphate from water in a broad range of conditions, particularly even from low-concentrated aqueous phosphate solutions and under a wide range of pH conditions. Under standard adsorption conditions (pH 7.0 and 298 K), the maximum phosphorus adsorption capacity was determined to be 53.1 P-mg/g, which was 73 times greater than that of BF slag. This outstanding phosphorus capacity was attributed to the inherent high-surface-area of CSH and the strong affinity of Ca²⁺ species abundant in the CSH network with phosphate ions. Kinetic analysis demonstrated that the adsorption of phosphate onto slag-made CSH adsorbent can be well fitted with pseudo-second-order kinetic model and Freundlich adsorption isotherms.

Acknowledgement

This work was financially supported by the ISIJ Research Promotion Grant (Tekkou Kenkyu Shinko Josei) from the Iron and Steel Institute of Japan. Y.K. acknowledges the Grant-in-Aid awarded from the Steel Foundation for Environmental Protection Technology, the JFE 21st Century Foundation and the Japan Association for Chemical Innovation (JACI). Y.K. also thanks the financial support from Frontier Research Base for Global Young Researchers, Osaka University.

References

• 1)   N.  Gilbert: Nature, 461 (2009), 716.
• 2)   K.  Ashley,  D.  Cordell and  D.  Mavinic: Chemosphere, 84 (2011), 737.
• 3)   D.  Cordell,  J.-O.  Drangert and  S.  White: Glob. Environ. Chang., 19 (2009), 292.
• 4)   P. J. A.  Withers,  J. J.  Elser,  J.  Hilton,  H.  Ohtake,  W. J.  Schipper and  K. C.  van Dijk: Green Chem., 17 (2015), 2087.
• 5)   H.  Ohtake,  K.  Takahashi,  Y.  Tsuzuki and  K.  Toda: Water Res., 19 (1985), 1587.
• 6)   G. K.  Morse,  S. W.  Brett,  J. A.  Guy and  J. N.  Lester: Sci. Total Environ., 212 (1998), 69.
• 7)   L. E.  de-Bashan and  Y.  Bashan: Water Res., 38 (2004), 4222.
• 8)   P.  Cornel and  C.  Schaum: Water Sci. Technol., 59 (2009), 1069.
• 9)   E.  Oguz: Colloids Surf. A, 262 (2005), 113.
• 10)   M. Y.  Can and  E.  Yildiz: J. Hazard. Mater., 135 (2006), 165.
• 11)   W.  Huang,  S.  Wang,  Z.  Zhu,  L.  Li,  X.  Yao,  V.  Rudolph and  F.  Haghseresht: J. Hazard. Mater., 158 (2008), 35.
• 12)   A.  Kaasik,  C.  Vohla,  R.  Mõtlep,  Ü.  Mander and  K.  Kirsimäe: Water Res., 42 (2008), 1315.
• 13)   P. L.  Sibrell,  G. A.  Montgomery,  K. L.  Ritenour and  T. W.  Tucker: Water Res., 43 (2009), 2240.
• 14)   S.  Karaca,  A.  Gürses,  M.  Ejder and  M.  Açıkyıldız: J. Hazard. Mater., 128 (2006), 273.
• 15)  The Japan Iron and Steel Federation and Nippon Slag Association: The Annual Report of Statistics of Iron and Steel Slag 2015, http://www.slg.jp, (accessed 2016-08-08).
• 16)   H.  Yamada,  M.  Katayama,  K.  Saito and  M.  Hara: Water Res., 20 (1986), 547.
• 17)   E.  Oguz: J. Hazard. Mater., 114 (2004), 131.
• 18)   L. J.  Westholm: Water, 2 (2010), 826.
• 19)   Y.  Kuwahara,  S.  Tamagawa,  T.  Fujitani and  H.  Yamashita: J. Mater. Chem. A, 1 (2013), 7199.
• 20)   J. J.  Chen,  J. J.  Thomas,  H. F. W.  Taylor and  H. M.  Jennings: Cem. Concr. Res., 34 (2004), 1499.
• 21)   M. J.  Cairns,  T.  Borrmann,  W. H.  Höll and  J. H.  Johnston: Microporous Mesoporous Mater., 95 (2006), 126.
• 22)   D. C.  Southam,  T. W.  Lewis,  A. J.  McFarlane,  T.  Borrmann and  J. H.  Johnston: J. Colloid Interface Sci., 319 (2008), 489.
• 23)   K.  Okano,  S.  Miyamaru,  A.  Kitao,  H.  Takano,  T.  Aketo,  M.  Toda,  K.  Honda and  H.  Ohtake: Sep. Purif. Technol., 144 (2015), 63.
• 24)   Y.  Kuwahara,  S.  Tamagawa,  T.  Fujitani and  H.  Yamashita: Bull. Chem. Soc. Jpn., 89 (2016), 472.
• 25)   A.  Meiszterics,  L.  Rosta,  H.  Peterlik,  J.  Rohonczy,  S.  Kubuki,  P.  Henits and  K.  Sinkó: J. Phys. Chem. A, 114 (2010), 10403.
• 26)   Y.  Kuwahara and  H.  Yamashita: ISIJ Int., 55 (2015), 1531.
• 27)   Y.  Kuwahara,  T.  Ohmichi,  T.  Kamegawa,  K.  Mori and  H.  Yamashita: J. Mater. Chem., 20 (2010), 5052.
• 28)   J. H.  Johnston,  T.  Borrmann,  D.  Rankin,  M.  Cairns,  J. E.  Grindrod and  A.  McFarlane: Curr. Appl. Phys., 8 (2008), 504.
• 29)   W.  Guan,  F.  Ji,  Q.  Chen,  P.  Yan and  L.  Pei: Materials, 6 (2013), 2846.