Introduction

The increasing human activity and technology development are beneficial for mankind; the environmental conditions, however, suffer from it more and more visibly. The pollutants resulting from industrial processes pose threat to biota and non-living part of ecosystems (water, soil, etc.). Heavy metals and metalloids are one of the representative groups of pollutants with scientifically evidenced impact on the global and local scales. One of the most abundant metals in industrial run-offs is copper and lead. The main sources of Cu and Pb from industrial activity are mining, power engineering, metal smelting, radiator manufacturing, and alloy industries (Adriano, 2001; Kadirvelu et al. 2001; Hua et al. 2012). Risk for humans is associated with ability of these metals to disperse in water systems and migrate within them, which is one of the most important paths of exposure since metals accumulate easily in living tissues (Liu et al. 2010; Tchounwou et al. 2012; Chowdhury et al. 2016; Łuczyńska et al. 2018). These factors prompt the research on the removal of metals from water systems.

Several techniques are known for metal ions removal from aqueous solutions, including precipitation (Feng et al. 2000), ion exchange (Hubicki and Kołodyńska 2012), membrane filtration (Chitpong and Husson 2017), and reverse osmosis (Qdais and Moussa 2004). All of them are efficient but very expensive when utilized on a large scale (Fu and Wang 2011; Hashim et al. 2011; Rajasulochana and Preethy 2016). Most cost-effective appears to be adsorption (Bakarat 2011); thus, many researchers take great efforts in developing low-cost and environment-friendly adsorbents. Among tested adsorbents, materials derived from organic compounds (Feng et al. 2009, 2010; Senthil Kumar et al. 2012; Lasheen et al. 2012; Taşar et al. 2014), nanoadsorbents (Kyzas and Matis 2015), organic waste (Bożęcka et al. 2016), and mineral adsorbents like raw and synthetic zeolites (Bajda et al. 2004; Choi et al. 2016; Król et al. 2016; Meng et al. 2017) and clay minerals (Srivastava et al. 2005; Doğan et al., 2008; Huang et al. 2015; Yang et al. 2015), clay-bearing mining waste (Helios-Rybicka and Wójcik 2012), dolomite powder (Pehlivan et al. 2009, Gruszecka-Kosowska et al. 2017), calcite sludge (Merrikhpour and Jalali 2012), Ca-Mg phosphate (Ivanets et al. 2017), copper flotation waste (Mikoda et al. 2017), and calcium silicate powder (Ma et al. 2018) were tested in terms of removal of metals from aqueous solutions. It is also advisable that developed adsorbents were easily recoverable. The common practice in regeneration of used adsorbents is washing with different chemicals to obtain “clean” adsorbents. In the study of Li et al. (2016), the modified chitosan has been washed in several cycles with 0.01M EDTA. The outcome of 6 washing cycles was that the adsorption capacity was reduced from 100 to 83%, showing the excellent stability of the adsorbent. The other chemicals used in the regeneration of low-cost adsorbents were alkalized distilled water (pH = 8) for rice husk, which maintained 91% adsorption capacity in the second cycle (Shalaby et al. 2017), and H2SO4/NaOH treatment of chitosan, that led to the activation of chitosan, which caused full recovery of adsorption capacity (Schwarz et al. 2018).

Flue gas desulfurization (FGD) gypsum and air pollution control (APC) residues are the types of wastes commonly generated worldwide since virtually every industrial activity producing gas effluents needs to meet air quality standards. FGD residue is a material that is derived from wet flue gas desulfurization processes (Yan et al. 2015). In China, 43 Mt of FGD gypsum was produced in 2010 (Yan et al. 2015), and in Poland, 3 Mt of this waste was generated in 2016 (GUS 2017). The utilization of FGD gypsum is performed, e.g., as set retarder in Portland cement, as road sub-base material or in agricultural purposes (Chandara et al. 2009; Chen et al. 2009; Hua et al. 2010). APC residue in contrary is the material derived from the air pollution control sections in different facilities, such as cyclones or flue gas washing installations (Amutha-Rani et al., 2008). In the world, APC residues are generated in amount of 1.44 Mt annually (Yang et al. 2017). Both types of these wastes are of particular interest because of their fine grain size and high content of heavy metals, which makes their disposal hazardous. Copper industry in Poland is responsible for generating ca 30 Mt of wastes each year, mostly from ore enrichment (Kotarska 2012). The copper smelting facilities produce flue gases that need to be decontaminated before release to the atmosphere; therefore, air pollution control systems and wet flue gas desulfurization are applied to meet environmental standards. These methods generate ca 214,000 Mg of APC and FGD residues annually in Poland (Górniak-Zimróz 2009). Some of these wastes are utilized as metallurgical flux (ca 20%), and others are landfilled. Thus, the method of utilizing them needs to be developed. The research papers on using of APCR and FGDR as adsorbents in the literature are scarce, the more on their regeneration. The suitability of FGD gypsum as adsorbent of Pb(II) and Cd(II), as well as silicate, selenite, and phosphate ions from both simulated solutions and real wastewater, was previously examined (Bryant et al. 2012; Córdoba et al. 2015; Yan et al. 2015; Kang et al. 2018). The FGD gypsum has been found promising material in metals and anions removal from aqueous solutions. The research on FGD reuse, however, were not carried out by those authors. The FGD gypsum has also been utilized to synthetize hydroxyapatite, which was subsequently tested in metal removal from aqueous solutions with positive results (Yan et al. 2014; Liu et al. 2018).

The aim of this study was to verify the feasibility of air pollution control residues (APCR) and flue gas desulfurization residues (FGDR) from copper foundry in Southwestern Poland as adsorbents of Cu(II) and Pb(II) from simulated aqueous solutions. Sorption capacities with regard to reaction time, initial pH, initial concentration, and adsorbent dose were examined to identify possible mechanisms of adsorption. The experiments were carried out on a pilot scale, since these very materials have not been utilized as adsorbents before. Therefore, the limited amount of experiments has been performed. The further experiments in laboratory scale, as well as semi-industrial scale, are to be undertaken using the input parameters established in this study.

Materials and methods

Material preparation

APCR and FGDR powder samples were received from copper smelter in Głogów (Southwestern Poland), where they were sampled directly from individual places of the process (air pollution control system and wet flue gas desulfurization system, respectively). After transporting to the laboratory, the materials were dried in 70 °C for 24 h using POL-EKO SLN 32 oven. Because of very small grain size (< 0.2 mm), further processing was not required. After such preparation, basic techniques were applied to characterize obtained materials.

Analytical procedures

The chemical composition was determined with the wavelength-dispersive X-ray fluorescence (WDXRF) method, using Rigaku ZSX Primus II apparatus (X-ray tube: end-window with Rh anode, 4 kW power, 60 kV voltage). Calculations of intensities to obtain the chemical composition were performed using standardless fundamental parameters method (Sherman 1955; Shiraiwa and Fujino 1966). The method used in the calculations is based on the theoretical relations between X-ray intensities and concentrations of elements. No external standards are utilized in the analysis. Instead, the sensitivity library derived with the apparatus, which contains the X-ray intensities on some standards with known elemental composition, is used. In this procedure, the sensitivity library provided by Rigaku has been applied.

Powder X-ray diffraction (XRD) patterns were recorded, using Rigaku MiniFlex 600 apparatus with Cu-Kα radiation, 2–75° 2ϴ range with 0.05° step width. Obtained data were analyzed using XRAYAN software accompanied with JCPDS-ICDD 2013 mineral database.

Infrared (FTIR) spectra of the materials before and after sorption experiments (see the “Sorption experiments” section) were recorded by Bruker Tensor 27 spectrometer using DRIFT technique (3 wt% sample/KBr) in 4000–400 cm−1 region at 4 cm−1 resolution. Obtained spectra were processed using OMNIC software.

Specific surface area and pore space volume were determined before and after adsorption experiments (see the “Sorption experiments” section), using the N2 adsorption/desorption method in 77 K temperature with Micromeritics ASAP 2020 apparatus. Five-gram portions of APC and FGD residues were heated in 105 °C for 12 h before taking the measurements. The Brunauer-Emmett-Teller (BET) isotherm was used to characterize the specific surface area (SBET) of examined materials (Brunauer et al. 1938). The total pore volume was calculated for the relative pressure of p/p0 = 0.99. The volume of micropores (d < 2 nm) was calculated, using the Dubinin-Radushkevich method (ISO 15901:3-2007). The volume of mesopores (2 < d < 50 nm) was calculated, using the Barrett-Joyner-Halenda (BJH) equation (Barrett et al. 1951). The volume of macropores (d > 50 nm) was obtained by subtracting the volumes of micropores and mesopores from the total pore space volume.

The metal concentrations in supernatants after sorption experiments (see the “Sorption experiments” section) were determined by flame atomic absorption spectrometry (FAAS) with Thermo Scientific ICE 3000 Series apparatus using following setup: 0.5-nm spectral slit, 75% lamp current, background correction: off, continuous signal, 0.9–1.1 dm3 min−1 fuel flow. Wave lengths used for analysis were 217 nm for Pb and 324.8 nm for Cu. The limit of quantification was 0.033 μg dm−3 for Cu and Pb. The device was calibrated using standard solutions of known Cu and Pb concentrations. The standards were analyzed at least once every ten samples to ensure quality control. Randomly chosen samples were analyzed again to confirm the correctness of measurements.

Preparation of reagents

All chemicals used in this study were of analytical grade (> 99% purity; Chempur, Poland). Stock solutions (10,000 mg dm−3) were prepared by dissolving appropriate amounts of Pb(NO3)2 (CAS: 10099-74-8) and CuSO4 · 5H2O (CAS: 7758-99-8) in de-ionized water (Milli-Q, Millipore). The stock solutions were further dissolved for obtaining binary aqueous solutions of Cu(II) and Pb(II) in five different concentrations: 50, 100, 500, 1000, and 2500 mg dm−3 of each metal ion.

Sorption experiments

The sorption experiments were carried out with batch technique. 0.6-g portions of APC and FGD residues were exposed to 30 cm3 of mixed Cu/Pb solutions. For adsorbent dose test, the same volume of solution was mixed with 0.3, 0.6, and 1.5 g of tested materials (doses of 10, 20, and 50 g dm−3). The effect of pH was examined using aqueous solutions of pH 2–5 for Cu(II) and 2–8 for Pb(II), adjusted with 0.1 M HNO3 and 0.1 M NaOH. The duration of the tests used to investigate the effect of pH was 120 min. The pH ranges for sorption experiments in case of Cu(II) and Pb(II) were different to avoid formation of Cu and Pb hydroxides, which occurs when pH exceeds 5.5 and 8.0 for Cu and Pb, respectively (Ayres et al. 1994). Kinetic studies were carried out using five different contact time periods, i.e., 30, 60, 120, 180, and 240 min. Adsorption isotherms were investigated with five initial concentration values, i.e., 50, 100, 500, 1000, and 2500 mg dm−3 for both Cu(II) and Pb(II). Thermodynamics studies were performed in three different temperatures, i.e., 20, 40, and 60 °C. In each experiment, mixtures were shaken at 250 rpm with orbital shaker. After shaking, the solutions were centrifuged, and supernatants were filtered through Whatman® qualitative cellulose filter paper (25-mm diameter, circles). Subsequently, supernatants were acidified to avoid precipitation of metals from solution (Jakimowicz-Hnatyszak and Rubel 1998) and stored in the fridge (4 °C) prior to Cu and Pb analysis (see the “Analytical procedures” section).

Results and discussion

Adsorbent characterization

The XRD analysis (Fig. 1) revealed that the two examined materials differed in mineral composition. The pattern of FGDR sample (Fig. 1(a)) revealed mixed mineral composition of gypsum, hannebachite, calcite, and portlandite. The sole mineral phase observed in APCR sample was gypsum (Fig. 1(b)). Chemical analysis carried out with WDXRF (Table 1) confirmed that chemical composition was in accordance with mineral composition, with CaO and SO3 prevailing in both material samples (43 and 33.8% respectively for APCR; 54.5 and 37.5% respectively for FGDR). The APCR sample contains substantial amounts of Fe2O3, As2O3, ZnO, and F (8.24%, 2.94%, 2.28%, and 2.77%, respectively). The total amounts of oxides are not 100% since the elements that were abundant in amount of < 0.1% were not enclosed in Table 1.

Fig. 1
figure 1

X-ray diffraction patterns of FGDR (a) and APCR (b) samples

Full size image
Table 1 Chemical (oxide) composition of APCR and FGDR samples determined by WDXRF (all results in wt%)
Full size table

The FTIR results for tested adsorbent samples (Fig. 2) were fairly accordant with determined phase composition. In case of APCR before experiments (Fig. 2a), characteristic bands for gypsum (3530, 3400, 1685, 1621, 1116, 668, 601, and 457 cm−1), calcite (872 cm−1), and hannebachite (1475, 1430, and 1108 cm−1) were found. The spectrum of post-adsorption APCR sample (Fig. 2b) was similar, with intensity of some bands reduced. The FGDR sample before experiments spectrum (Fig. 2c) revealed bands characteristic for all minerals determined using XRD (3645 cm−1—portlandite; 3400, 1630, 654, and 490 cm−1—gypsum; 1492, 1475, 1430, 1217, 1108, 984, and 942 cm−1—hannebachite; 2512 and 872 cm−1—calcite). Post-adsorption spectrum for FGDR sample (Fig. 2d) reflected several changes in intensity of bands. The increase was noted for 1430–1492, 1108, and 872 cm−1 bands. On the other hand, bands associated with portlandite (3645 cm−1) almost disappeared. However, the shifting of FTIR bands was not observed after the adsorption process, which suggested that the process did not cause the formation of additional metal-ion associations and had no influence on the structure of the adsorbent.

Fig. 2
figure 2

FTIR spectra of APCR sample before and after adsorption (a) and FGDR sample before and after adsorption (b)

Full size image

The textural parameters (Table 2) reflected the feasibility of utilizing tested materials as adsorbents. Both materials were characterized by relatively high specific surface area (66.8 and 28.8 m2 g−1 for APCR and FGDR, respectively) when compared to other waste materials from copper enrichment facilities (copper flotation waste—6.24 m2 g−1) (Mikoda and Gruszecka-Kosowska 2018); however, the surface area is low when compared to other efficient adsorbents such as activated carbon or natural bentonite (1239 and 1000 m2 g−1, respectively) (Patnukao et al. 2008; Iskander et al. 2011). Low pore space volume (0.158 and 0.142 cm3 g−1 for APCR and FGDR, respectively) is common for microporous materials. However, as per pores dimension, mesopores dominated in APCR sample, while meso- and macropores were prevailing in FGDR sample. After the experiments, samples exhibited slightly different properties. In case of APCR, a decrease of surface area was observed along with the decrease of mesopores percentage and increase of macropores volume, which implied that the mesopores could be occupied by adsorbed ions. The surface area of FGDR sample also slightly decreased, but in this case, the volume of mesopores increased at the expense of macropores, suggesting that adsorbed particles could occupy mesopores. The analysis of N2 adsorption/desorption isotherms (Fig. 3) revealed that in case of both adsorbents, the isotherms exhibited type IV of adsorption isotherm and H3 hysteresis loop, that are typical for mesoporous and slit-shape pored materials (Sing et al. 1985).

Table 2 Textural parameters of APCR and FGDR samples before and after adsorption
Full size table
Fig. 3
figure 3

The N2 adsorption/desorption isotherms for APCR and FGDR samples

Full size image

Effect of pH

The pH value is defined as one of the most important factors affecting the adsorption process due to its influence on speciation of metals and the surface properties of the adsorbent (Mohammadi et al. 2015; Yan et al. 2015). In this study, pH ranging from 2 to 5 was used for Cu(II) and from 2 to 8 range was applied for Pb(II) (see the “Sorption experiments” section). The pH value was surprisingly found as not governing the adsorption of metals on APCR and FGDR samples (Fig. 4), while other studies show dependence of adsorption on pH (e.g., Yan et al. 2015; Mikoda et al. 2017; Gruszecka-Kosowska et al. 2017). The sorption values maintained at about 78 mmol kg−1 Cu(II) and about 24 mmol kg−1 Pb(II) for both analyzed material samples at initial concentration of Pb and Cu equal to 100 mg dm−3. For that reason, other experiments were conducted at initial pH 3.0 (acidic), because this value is below the hydroxide formation point for both ions, which should secure the metals dissolved in the solution. Thus, research in real effluent will be performed in higher pH values. Before, repetition of the experiment on pilot scale will be made to investigate the reason of metals adsorption being not dependent on initial pH values.

Fig. 4
figure 4

Effect of pH on adsorption of metal ions onto APCR (a) and FGDR (b) samples

Full size image

Effect of adsorbent dose

Both examined ions exhibited similar behavior in relation to adsorbent dose (Fig. 5). The amount of adsorbed Cu(II) and Pb(II) decreased rapidly (exponentially) with increasing adsorbent dose. The biggest leap was noted between 10 and 20 g dm−3 doses, where amounts of adsorbed ions decreased twofold. For Cu(II), this decrease tempo was maintained with increasing dose (20 to 50 g dm−3), whereas for Pb(II), it even increased to threefold. Namasivayam and Kumuthu (1998) stated that this phenomenon might be due to “over-crowding” of adsorbent particles, which means that there is “too much” adsorbent to the extent that the process is stopped. The results showed that assumed absorbent doses were too high. This part of experiments will be repeated before industrial scale research; however, it seems that low doses of adsorbent should be effective in Pb and Cu removal. It means that relatively low amounts of material would be enough for wastewater purification. This might reduce the additional waste generation in case if adsorbent regeneration would be technically not possible or economically unprofitable.

Fig. 5
figure 5

Effect of adsorbent concentration on adsorption of metal ions onto APCR (a) and FGDR (b) samples

Full size image

Equilibrium sorption modeling

Modeling of adsorption can be performed in different ways. Most popular method is describing the process using isotherms, being the relation of amount of metal adsorbed as a function of equilibrium concentration at the same temperature. The isotherms are useful in understanding the mechanism of adsorption and in designing optimal process conditions (Matusik 2014; Mohammadi et al. 2015; Yan et al. 2015). Freundlich, Langmuir, and Dubinin-Radushkevich (D-R) models (Eqs. 1–3) are often used and well described in literature.

Freundlich model is based on the assumption that the adsorption process is infinite and the surface of the adsorbent is heterogenous. This equation is purely theory-based, since the saturation phenomena occur in actual solid/liquid mixtures (Król et al. 2016).

On the other hand, Langmuir theory applies to homogenous adsorbent surfaces. The model assumes that there is no interaction between adsorbed molecules and that the particles are bonded with the same mechanism and activation energy. The process is limited in this model by the number of active sites, leading to creation of a single layer of molecules on the adsorbent’s surface (Yan et al. 2014; Maziarz and Matusik 2016).

The D-R model gives a basic idea about the mechanism of adsorption, which is not given by Langmuir or Freundlich equations. The parameters of the model help to verify whether the adsorption is of physical or chemical nature; the determination is based on the mean adsorption energy (E) [kJ mol−1]; if E < 8, the process is of physical nature; if 8 < E < 16, the adsorption takes place via ion exchange; if E > 16, then the process is of chemical nature (Yan et al. 2015).

In this study, Freundlich, Langmuir, and Dubinin-Radushkevich (D-R) models were used. For the Freundlich model of sorption, formula (1) (Freundlich 1906) was applied. For the Langmuir model of sorption, formula (2) (Langmuir 1918) was used. For D-R isotherm, formula (3) (Dubinin et al. 1949) was applied:

$$ {q}_{\mathrm{eq}}={K}_F{C}_{\mathrm{eq}}^{1/n} $$
(1)
$$ {q}_{\mathrm{eq}}=\frac{K_L{q}_m{C}_{\mathrm{eq}}}{1+{K}_L{C}_{\mathrm{eq}}} $$
(2)
$$ \ln {q}_{\mathrm{eq}=}\ln {q}_m-{\beta \varepsilon}^2 $$
(3)

where qeq is the sorption capacity at equilibrium [mmol kg−1], Ceq is the equilibrium concentration [mmol dm−3], KL is the Langmuir adsorption constant [dm3 mmol−1], qm is the maximum adsorption capacity [mmol kg−1], KF is the Freundlich adsorption capacity [mmol kg−1], n is the Freundlich dimensionless constant, β is the D-R constant [mol2 J−2], and ε is the Polanyi potential (Eq. 4):

$$ \varepsilon = RT\ln \left(1+\frac{1}{C_{eq}}\right) $$
(4)

where R is the gas constant (8.31446 J mol−1 K−1), T is the absolute temperature [K], and Ceq is the equilibrium concentration [mmol dm−3].

The mean adsorption energy was calculated with the following formula (5):

$$ E=\frac{1}{\sqrt{-2\beta }} $$
(5)

where E is the mean adsorption energy [kJ mol−1] and β is the D-R constant [mol2 J−2].

The isotherms along with Langmuir, Freundlich, and Dubinin-Radushkevich model values were shown on Fig. 6a–f. The isotherm parameters and constants shown in Table 3 revealed that the data for Pb(II) (Fig. 6b) and Cu(II) (Fig. 6a) adsorbed on FGDR sample followed the Langmuir model; Cu(II) data followed the model very well (R2 = 0.99). The model was found to illustrate moderately the Pb(II) adsorption data onto FGDR (R2 = 0.90). The Langmuir model was perfectly fitted also for Cu(II) sorption onto APCR sample (Fig. 6a; R2 = 1). For Pb(II) adsorption on FGDR, Freundlich equation gave much better correlation (R2 = 0.84) than of Langmuir model (R2 = 0.05); however, both R2 values show that this model is not good for displaying the Pb(II) adsorption onto FGDR (Fig. 6c, d). The Dubinin-Radushkevich model (Fig. 6e, f) was found to be moderately fitted to Cu(II) adsorption data onto both adsorbents (R2 equal to 0.96 and 0.95 for APCR and FGDR, respectively) and not fitting the Pb(II) adsorption data for both materials (0.81 and 0.07 R2 values for APCR and FGDR, respectively). The maximum adsorption capacities of Cu(II) were 676.0 mmol kg−1 for APCR and 1554 mmol kg−1 for FGDR, which after calculation were equal to 42.9 and 98.8 mg g−1, respectively. In case of Pb(II), the capacities were 600.6 mmol kg−1 for APCR and 602 mmol kg−1 for FGDR (124.4 and 124.7 mg g−1 respectively after calculation). Langmuir model capacities qm for Cu(II) were fairly accordant with obtained real data (588 and 1667 mmol kg−1 for APCR and FGDR, respectively), and far better than Freundlich KF constants (302 and 849 mmol kg−1 for APCR and FGDR, respectively). On the other hand, Freundlich (KF; 4540 and 310 mmol kg−1 for APCR and FGDR, respectively) and Langmuir (qm; 1111 and 200 mmol kg−1 for APCR and FGDR, respectively) model capacities for Pb(II) are strongly inadequate, even when fairly good correlation was found. Good accordance of data with Langmuir model may suggest that the process was unidirectional (Merrikhpour and Jalali 2012). The Langmuir constant b was used to calculate dimensionless separation factor (RL = (1 + bC0)−1; C0—initial concentration [mg dm−3]) that is used to assess the favorability of adsorption (Yan et al. 2015). The results were shown on Fig. 7a for APCR and Fig. 7b for FGDR as function of RL against C0. All RL values fell into 0–1 range, which implied that the adsorption was favorable (Meitei and Prasad 2014). Moreover, the RL values from Fig. 7 were mostly higher for Pb(II) than for Cu(II), suggesting higher affinity between Pb(II) and both adsorbents. Similar results were obtained by Yan et al. (2014, 2015) using FGD gypsum to remove Pb(II) ions from aqueous solutions. The explanation of this phenomenon can be as follows: (i) the smaller hydrated radius of Pb(II) (0.261 nm) than of Cu(II) (0.295 nm), that can suggest higher accessibility of Pb(II) for FGDR and APCR; (ii) higher electronegativity of Pb(II) (2.33) than of Cu(II) (1.90), that suggests stronger attraction of Pb(II) to the surface of the adsorbents; (iii) the ionic radius of Pb(II) (0.122 nm) is larger than of Ca(II) (0.097 nm), while that of Cu(II) (0.072 nm) is smaller than that of Ca(II), which suggests that Cu(II) is less likely to be incorporated to the structure of APCR and FGDR (Yan et al. 2014, 2015). The favorability of adsorption can also be assessed with Freundlich’s model exponent (n). Values of 1/n (Table 3) fell into 0–1 range for both adsorbents and both ions, indicating favorable adsorption of Cu and Pb onto APCR and FGDR. Also, the KF constant gives some information about binding affinity of ions (Ma et al. 2011). In case of APCR, the KF value for Pb(II) was higher, implying stronger affinity of Pb(II) to APCR. This finding was compliant with comparison of RL values from Fig. 6. On the other hand, KF values for FGDR suggest that Cu(II) was bonded stronger to the adsorbent, contrary to RL results and results obtained by Yan et al. (2015).

Fig. 6
figure 6

Langmuir, Freundlich, and Dubinin-Radushkevich (D-R) isotherm plots for the adsorption of Cu(II) onto APCR (a), Pb(II) onto APCR (b), Cu(II) onto FGDR (c), Pb(II) onto FGDR (d), Cu(II) and Pb(II) onto APCR (e), and Cu(II) and Pb(II) onto FGDR (f)

Full size image
Table 3 Langmuir, Freundlich, and Dubinin-Radushkevich constants and correlation coefficients for adsorption of metal ions onto APCR and FGDR samples
Full size table
Fig. 7
figure 7

Separation factor, RL, for the adsorption of Pb(II) and Cu(II) at different initial concentrations (C0) onto APCR (A) and FGDR (A) samples

Full size image

The data for D-R model (Table 3) suggested that the mean adsorption energy (E) fell into 3–5 kJ mol−1 range for all samples. The E values were below 8 kJ mol−1, so the adsorption presumably was of physical nature and metals were weakly bonded with the adsorbent, which again was contrary to results of Yan et al. (2015). The reason of these differences might be due to different mineral composition (relatively pure FGD gypsum versus mixed calcium sulfate-carbonate-hydroxide FGDR sample).

Thermodynamic studies were conducted in 20, 40, and 60 °C. The input parameters were 500 mg dm−3 initial metal concentration, pH 3.0, 60-min reaction time. The Gibbs free energy change (ΔG°, kJ mol−1), enthalpy change (ΔH°, kJ mol−1), and entropy change (ΔS°, kJ mol−1 K−1) were calculated using van’t Hoff’s equations (Eqs. 6–8; Milonjić 2007; Yan et al. 2015; Frantz et al. 2017):

$$ K=\frac{C_s}{C_{\mathrm{eq}}} $$
(6)
$$ \Delta {G}^{{}^{\circ}}={R}_T\ln K $$
(7)
$$ \ln K=\frac{\Delta {S}^{{}^{\circ}}}{R}-\frac{\Delta {H}^{{}^{\circ}}}{RT} $$
(8)

where K is the dimensionless adsorption equilibrium constant, R is the gas constant (8.31446 J mol−1 K−1), T is the absolute temperature [K], Cs is the equilibrium sorption capacity [mmol dm−3], and Ceq is the equilibrium concentration [mmol dm−3].

As shown in Table 4, ΔG° absolute values for Cu(II) adsorption onto APCR decreased with temperature rising, suggesting that lower temperature favored the process. On the other hand, ΔG° absolute values for APCR were increasing for Pb(II) adsorption with the temperature increase, implying different behavior of this ion in thermodynamic context. The ΔG° absolute values for adsorption of ions onto FGDR were in indirect proportion with the temperature for both Cu(II) and Pb(II), which implied that lower temperature was favoring the adsorption process as well. Generally, lower temperature was more advantageous for adsorption on both materials. Negative ΔG° and positive ΔH° values for Cu(II) adsorption onto both adsorbents indicated that the process was spontaneous and endothermic. On the other hand, ΔG° values for Pb(II) adsorption onto both adsorbents were positive and ΔH° values were negative, suggesting not spontaneous and exothermic adsorption process. Enthalpy change ΔH° is used as an indicator of sorption mechanism (2.1–20.9 kJ mol−1—physisorption; 20.9–418.4 kJ mol−1—chemisorption) (Saǧ and Kutsal 2000). In case of both investigated adsorbent samples and Cu(II) ion, the values of enthalpy change exceeded 20.9 kJ mol−1, which implied that the process was of chemical nature. On the other hand, the absolute enthalpy change values for Pb(II) adsorption onto both adsorbents fell into 2.1–20.9 kJ mol−1 range, which suggest the physisorption as the prevailing mechanism. Positive values of ΔS° for Cu(II) adsorption onto both adsorbents implicated the increase of randomness of solid-liquid interface, which aided the adsorption process.

Table 4 Thermodynamic parameters for metal ions adsorption onto APCR and FGDR samples
Full size table

Sorption kinetics

The kinetics of adsorption process is of great importance in terms of technological optimization of the process efficiency (Mohammadi et al. 2015). Figure 8 shows that the equilibrium was reached at 30 min of reaction time; thus, in case of materials used, the contact time could be shorter. The rapid adsorption of metal ions and silicate anions has been proved for FGD gypsum in the studies of other authors (Yan et al. 2015; Kang et al. 2018). This finding is essential for industrial practices, since shorter contact time ensures lower costs for the removal process, which is a strong point in using those materials in real wastewater treatment. The most commonly used models are pseudo-first-order model, described by the equation (Eq. 9) and pseudo-second-order model, described by the equation (Eq. 10) (Matusik 2014):

$$ \log \left({q}_{\mathrm{eq}}-{q}_t\right)=\log {q}_{\mathrm{eq}}-\frac{k_1t}{2.303} $$
(9)
$$ \frac{t}{q_t}=\frac{1}{k_2{q}_{\mathrm{eq}}^2}+\frac{t}{q_{\mathrm{eq}}} $$
(10)

where qeq is the sorption capacity at equilibrium [mmol kg−1], qt is the sorption capacity at time t [mmol kg−1], k1 is the first-order rate constant [min−1], and k2 is the second-order rate constant [kg mmol−1 min−1].

Fig. 8
figure 8

Effect of contact time on adsorption of metal ions onto APCR (a) and FGDR (b) samples

Full size image

The constants and correlation coefficients of kinetic models (Table 5) showed that Cu(II) and Pb(II) adsorption on APCR and FGDR was perfectly fitted to pseudo-second-order equation (R2 = 1), which was also illustrated on Fig. 9(b). Some authors (Ho and McKay 1998; Merrikhpour and Jalali 2012; Cherifi et al. 2013) imply that the fact that the experimental data were perfectly expressed by pseudo-second-order model could be a sign of chemical adsorption as prevailing in the process. These statements, however, are contradictory to E values (Table 3) as indicators of physical nature of the process; therefore, the authors believe that chemisorption is out of question as possible adsorption mechanism for Cu(II) and Pb(II) onto APCR and FGDR.

Table 5 Correlation coefficients and adsorption rate constants for kinetic models of metal ions adsorption onto APCR and FGDR samples
Full size table
Fig. 9
figure 9

Pseudo-first-order (a) and pseudo-second-order (b) kinetic models of adsorption of metal ions onto APCR and FGDR samples

Full size image

Research perspectives

The preliminary research conducted in a batch mode and laboratory scale showed that APCR and FGDR are promising materials to be considered as adsorbents in heavy metal removal from aqueous solutions. The results were unusual at some points. For instance, the pH of the solution and the reaction time were found to not govern the adsorption process, which is contrary to other works describing the metal removal by low-cost adsorbents (e.g., Yan et al. 2015). The pilot-scale research was designed to prove the feasibility of the waste materials for metal removal. The drawbacks recorded during the experiments will be eliminated by repetition of research with inadequate results. Secondly, the adsorbent doses used in this study appear to be too high, since only in case of Cu(II) for APCR, the 100% adsorption capacity has been reached, while in case of other ions, the capacity reached 30–50% when the 10 g dm−3 dose has been applied. Therefore, the repetition of experiments using much lower doses is also planned. Encouraging is the fact that lower adsorbent doses indicate the low amounts of adsorbent needed to clean a wastewater in industrial scale process. The obvious next step is using the tested adsorbents in fixed bed adsorption column experiments. The fixed bed adsorption is far more realistic in industrial practice since large streams of wastewater are generated in manufactures, and the batch mode is more efficient for certain effluent concentrations (Dichiara et al. 2015); it is not designed to recover the industrially generated amounts of wastewater (Unuabonah et al. 2010). The information obtained during the fixed bed column experiments is helpful in designing the real-case scenario that gives proper characteristic of tested adsorbent in terms of its potential in purification of industrial effluents and verifies the feasibility of the material to be utilized in a large-scale wastewater treatment.

Comparison of adsorption capacity of examined materials with other mineral adsorbents

The results of adsorption capacity for APCR and FGDR from copper foundry in Southwestern Poland were compared to low-cost mineral adsorbents analyzed in other research (Table 6). The literature research showed that the Cu(II) capacity of APCR was higher than that of dolomite powder (8.26 mg g−1), zeolite (10.9 mg g−1), Mg-zeolite (15.2 mg g−1), and bog iron ore (25.2 mg g−1) and lower than that of copper flotation waste (58.8 mg g−1), zeolite A (62.4 mg g−1), FGDR (98.8 mg g−1), Ca-Mg phosphate (240.2 mg g−1), calcium silicate powder (680.9 mg g−1), and waste calcite sludge (1067.2 mg g−1). The capacities of both adsorbents analyzed in this study for Pb(II) (since they are virtually the same) were found to be higher than that of dolomite powder (21.7 mg g−1), zeolite (28.3 mg g−1), colemanite ore waste (33.6 mg g−1), Mg-zeolite (58.5 mg g−1), Jordanian sorbent (66.3 mg g−1), copper flotation waste (83.3 mg g−1), and bog iron ore (97 mg g−1) and lower than that of FGD gypsum (161.3 mg g−1), zeolite A (227.7 mg g−1), Ca-Mg phosphate (246.6 mg g−1), synthetic hydroxyapatite (277.8 mg g−1), and waste calcite sludge (534.2 mg g−1). The study revealed that APCR and FGDR samples were competitive as adsorbents of Cu(II) and Pb(II) from aqueous solutions when compared to other low-cost adsorbents. Presumably, the materials will also be competitive to other materials taking under consideration contact time of the adsorption process and adsorbent dose used for removal of metals from aqueous solutions.

Table 6 Comparison of adsorption capacity for heavy metal ions of APCR and FGDR with other adsorbents
Full size table

Conclusions

This work was aimed at the evaluation of APC and FGD residues from copper foundry in Southwestern Poland for heavy metal removal from simulated binary aqueous solutions with help of batch experiments using various input parameters. Conclusions that can be drawn from this study are the following:

  1. a.

    The residues from air pollution control (APCR) and flue gas desulfurization (FGDR) from copper smelting facility could be considered as adsorbents of heavy metals from aqueous solutions.

  2. b.

    The adsorption process on APCR and FGDR samples was not dependent on time and pH values in the tested range, which was in opposition to data for other adsorption materials.

  3. c.

    The increase of adsorbent dose suppressed the adsorption process in the tested range. Thus, lower doses of the adsorbent will be used in further research and lower amounts of material would be used in comparison to other materials.

  4. d.

    Equilibrium studies revealed that adsorption data followed Langmuir model quite well, except for Pb(II) adsorption on FGDR, which did not follow any of the models. The mean adsorption energy suggested that the process took place as a physisorption.

  5. e.

    Sorption kinetics of both Cu(II) and Pb(II) on both materials was expressed perfectly (R2 = 1) using pseudo-second-order equation.

  6. f.

    Thermodynamic studies showed that the process was spontaneous and endothermic for Cu(II) and not spontaneous and exothermic and Pb(II). Lower temperature was found more appropriate for the adsorption.

  7. g.

    The maximum sorption capacities of both materials were equal to 42.9 mg g−1 Cu(II) and 124.4 mg g−1 Pb(II) for APCR and 98.8 mg g−1 Cu(II) and 124.7 mg g−1 Pb(II) for FGDR, which was comparable to other low-cost adsorbents.