Introduction

Adsorption is an efficient strategy for removing heavy metals from polluted water bodies (Liu et al. 2015; Thakur et al. 2017). Lignocellulosic biomasses, e.g., wood sawdust, can technically be used to adsorb heavy metals, but the adsorption capacity is very low—about 0.04 and 0.02 mmol g−1 for Cu2+ and Pb2+ adsorption, respectively (Gan et al. 2016; Sciban et al. 2006). This is attributed to (a) the low specific surface area of the material (Sciban et al. 2006), (b) the insufficient exposure of adsorptive sites (hydroxyl groups, –OH) in fibers which feature an integral cell wall structure (Barnett and Bonham, 2004), and (c) the relatively weak affinity or complexation between the lone pair electrons of oxygen atoms in the –OH groups and the metal ions (Shen et al. 2016a), when compared to, for instance, carboxyl or carboxyalkyl (Chen et al. 2009; Yu et al. 2013), phosphorus (Jeon and Kim, 2009), nitrogen (Donia et al. 2012), or sulfur-containing groups (Wu et al. 2012).

To improve the adsorption ability of lignocellulosic materials, the cell wall layer structure is usually disintegrated to expose cellulosic nanofibrils (CNFs) and therefore produce greater specific surface area and more abundant accessible hydroxyl groups (Sehaqui et al. 2014). During this process, lignin, a biopolymer present in lignocelluloses and generally composed of a carbon–oxygen framework with various functional groups (methoxyl, phenol, ether, etc.) is commonly removed during adsorbent production since it has a lower number of surface-accessible acidic hydroxyl groups than cellulose does, per absorbent mass, and therefore is considered as a “filler”. Such removal of lignin further increases the number of accessible hydroxyl groups per unit area in the CNFs. However, both cell wall disintegration and lignin removal require intensive mechanical grinding coupled with chemical purification and consequently, the preparation process of CNFs is highly energy-consumptive (Chen et al. 2013). Additionally, the disposal of lignin and chemicals used for treatment, e.g., the black liquor produced from the pulp industry (Pokhrel and Viraraghavan 2004), is also a big challenge.

There is an alternative for the exposure of hydroxyl groups by reconstructuring the biopolymers into either hydrogels (Shen et al. 2016a; Thakur et al. 2018) or areogels (Kim et al. 2017), in which the re-crystallization of the cellulose trains can be restrained or diminished. Expectedly, hydrogels prepared from lignin by chemically cross-linking with common cross-linkers such as formaldehyde, glutaraldehyde, α-epichlorohydrin, or diepoxy compounds generally exhibit low water content and low adsorption capacity attributed to the lower number of –OH groups, unless another active component is introduced (Thakura and Thakur 2015).

Concluded from above, hydrogels made from partially or entirely delignified lignocelluloses are supposed to possess maximum adsorption capacity compared with the pristine lignocellulosic biomasses. Since 2009, the Rogers group has reported that the dissolution and regeneration of wood in the ionic liquid (IL) 1-ethyl-3-methylimidazolium acetate ([C2mim][OAc]) produced free lignin and cellulose-rich materials (CRMs). The CRMs contained much lower lignin contents compared to the original wood (Cheng et al. 2014; Li et al. 2011a; Sun et al. 2009, 2011; Wang et al. 2012). In later work, they also showed that among all CRM pulps produced using various IL-dissolution methods, the CRM prepared by the dissolution of wood in [C2mim][OAc] at a higher temperature (175 °C) for a shorter time (30 min) and regeneration in acetone/water as an anti-solvent had the highest molecular weight (MW) and the best hydrogel forming ability (Shen et al. 2016b).

However, on the other hand, considering that lignin has a stiff 3D polymeric network and thereby may help support hydrogel structure, lignin-containing pulps (e.g., CRMs) may provide opportunities for not only hydrogel production at higher yields but also improvements of the porosity, and thus adsorption properties of the resulting hydrogel adsorbent compared to the hydrogels based on fully-delignified cellulose pulps. In this study, we initially investigated the effect of lignin on the microstructure of the reconstituted lignocellulose by analyzing the regenerated aqueous suspensions from poplar wood powder (POP), CRM, Kraft pulp (KP), and microcrystalline cellulose (MCC) solutions in [C2mim][OAc] via transmission electron microscopy (TEM). Subsequently, lignocellulosic hydrogel beads were prepared, by dissolving the POP or CRM pulp in [C2mim][OAc], and reconstituted into a bead architecture using water as a coagulation solvent. These lignocellulosic beads as prepared were compared to hydrogel beads prepared from pure cellulose (MCC and KP) in terms of the properties including microstructure, number of hydroxyl groups, density, and porosity.

The original beads prepared were then functionalized by carboxymethylation, magnetization, or combined treatment with magnetization/carboxymethylation, to examine the effect of the presence of lignin on the functionalization efficacy of the lignocellulosic beads. Finally, the adsorption ability of the lignocellulosic and pure cellulosic hydrogel beads (both the original and the functionalized) towards Cu2+ and Pb2+ ions was studied and compared to verify the potential superiority of the lignocellulosic hydrogel beads over the pure cellulosic ones as bio-adsorbents.

Materials and methods

Chemicals and cellulosic materials

Ionic liquid 1-ethyl-3-methylimidazolium acetate ([C2mim][OAc], 99%, 1900 ppm water content measured with a Metrohm 831 Karl Fischer Titrator, Beijing, China), was purchased from Chengjie Chemical Ltd., Shanghai, China. Other chemicals including pyridine, sodium hydroxide (NaOH), ferrous chloride hydrate (FeCl2·4H2O), ferric chloride hydrate (FeCl3·6H2O), absolute ethanol (EtOH), sodium monochloroacetate (NaClCH2COO), cupric nitrate hydrate (Cu(NO3)2·3H2O), lead nitrate (Pb(NO3)2), and nitric acid (HNO3; 65 wt%) were of analytical grade and used as received from Sigma Aldrich (Shanghai, China). Deionized (DI) water with specific resistivity of 7.8 MΩ cm at 25 °C was purchased from Rightleder Environmental Investment Group Inc., Harbin, China. Ultrapure water with specific resistivity of 18.2 MΩ cm at 25 °C was obtained from the Milli-Q Century System (Millipore, Billerica, MA, USA).

Four different polymers (two lignocellulose and two pure cellulose) were used for preparing hydrogels: (1) mechanically ground poplar wood powder (POP), of particle sizes ranging from 80 to 100 mesh, was supplied by Yongxu Co., Ltd., Harbin, China; (2) cellulose-rich material (CRM), which was prepared by dissolving 0.5 g poplar wood powder in 10 g [C2mim][OAc] at 175 °C for 30 min followed by regeneration in 50% v/v acetone/H2O (Li et al. 2011a; Shen et al. 2016b); (3) bleached softwood Kraft pulp (KP),which was provided by Ilim Group (St. Petersburg, Russia) and had a whiteness value of 88–90%; and (4) microcrystalline cellulose (MCC), which was obtained from Sigma Aldrich (Shanghai, China).

Transmission electron microscopy (TEM) of lignocellulose suspensions

The 1% lignocellulose- or cellulose-IL solution was coagulated in ultrapure water, washed, and ultrasonically treated to obtain 0.05% aqueous suspensions. The suspensions were respectively dropped onto conventional carbon-coated copper grids, and were vacuum dried at 25 °C for 1 h. TEM observation of the samples was conducted by using a JEOL JEM-2100 transmission electron microscope (Tokyo, Japan) at 200 kV.

Preparation of hydrogel beads from lignocellulose-IL solution

Note % load = mass of solute (g)/mass of solvent (g). POP (0.7 g) was dissolved in 20 g [C2mim][OAc], in a 100 mL beaker at 175 °C in an oil bath under magnetic stirring for 30 min to obtain a 3.5% solution; CRM (0.4 g), KP (0.4 g), or MCC (1.0 g) was dissolved in 20 g of [C2mim][OAc] at 100 °C for 1 h to obtain solutions of 2% CRM, 2% KP, or 5% MCC, respectively. The polymer concentrations were selected to guarantee both the fluidity of the solution and the strength of the resulting hydrogel beads.

The hot lignocellulose- or cellulose-IL solutions were quickly transferred from the beaker into a 5 mL plastic syringe and injected dropwise (without a needle) into a 500 mL Erlenmeyer flask containing 300 mL DI water (at room temperature, 21 °C), where the hot drops were fully cooled down to form hydrogel beads that won’t stick together. (Please note that during the dropping process, the remaining solution in the beaker was stored in the oven at 80 °C. Water in the flask was frequently refreshed every two or three syringes to keep it cool.) The hydrogel beads in the Erlenmeyer flask were taken out and thoroughly washed repeatedly with DI water to completely remove [C2mim][OAc] IL. The resultant original hydrogel beads were denoted as POP3.5, CRM2, KP2, and MCC5, respectively.

Yields of the beads were determined to be 74 (3)% for POP3.5 (the moderate yield explained by the highest solution viscosity, and therefore the greatest difficulty during transfer into the syringe), 86 (2)% for CRM2, 89 (3)% for KP2, and 80 (2)% for MCC5 (some MCC drops failed to form beads), respectively.

Functionalization of hydrogel beads

The original hydrogel beads were separated into four groups: one quarter was used for carboxymethylation treatment through a modified procedure previously reported (Qi et al. 2009), one quarter for magnetization through a slightly modified technique reported previously (Luo et al. 2009; Tang et al. 2014), one quarter for magnetization followed by carboxymethylation (details described in Electronic supplementary material), and the last quarter was used as untreated controls. The resultant carboxymethylated, magnetized, and dually modified beads were abbreviated as C (e.g., CPOP3.5), M (e.g., MPOP3.5), and MC-(MCPOP3.5) beads, respectively. Characterization methods including viscosity determination of the lignocellulose-IL solutions, Fourier transform-infrared spectroscopy (FT-IR), powder X-ray diffraction (PXRD), scanning electron microscopy (SEM), hydroxyl number determination, lignin analysis, density and porosity determination, substitution degree determination, and magnetic property test of the hydrogel beads, are described in Electronic supplementary material.

Heavy metal ion adsorption test

Erlenmeyer flasks (100 mL) each containing 25 mL of 0.1 mM Cu(NO3)2·3H2O or Pb(NO3)2 solution (hereafter abbreviated as Cu2+ or Pb2+ solution) were kept in a SHA-C thermostatic incubator shaker (Ruihua Co., Ltd., Jintan, China) at 25 °C. The pH values of Cu2+ or Pb2+ solutions were adjusted to 4.0, 4.5, 5.0, 5.3, or 6.0 by adding 0.1 M HNO3 or 0.1 M NaOH. Wet beads (0.04 g dry mass per 100 mL solution) were added to each conical flask shaken mechanically at 100 rpm for 24 h. The beads were then removed and the concentration of metal ions in the resulting solutions were determined using an atomic absorption spectrophotometer (AAS; Purkinje General TAS-990, Beijing, China) according to the calibration curves determined for the Cu2+ or Pb2+ ions (Fig. S1).

Adsorption isotherms and adsorption kinetics

The equilibrium sorption isotherms (at 25 °C) were obtained by recording the equilibrium ion concentrations of the solutions and the adsorption capacitors of the beads after immersing different amount of the wet beads (ca. 0.004–0.090 g dry mass per 100 mL solution) in 0.1 mM Cu2+ and Pb2+ solutions for 24 h at pH 6 for Cu2+ and pH 5.3 for Pb2, respectively. The equilibrium sorption isotherms were then described via Langmuir, Freundlich and Sips models (Chatterjee et al. 2010; Ngah and Fatinathan 2008), which are presented using the equations below:

$$q_{e} = \frac{{q_{m} K_{L} C_{e} }}{{1 + K_{L} C_{e} }}$$
$$q_{e} = K_{F} C_{e}^{1/n}$$
$$q_{e} = \frac{{q_{\hbox{max} } (K_{S} C_{e} )\beta_{S} }}{{1 + (K_{S} C_{e} )^{{\beta_{S} }} }}$$

where qe (mmol g−1) is the adsorption capacity of the sample at equilibrium, Ce (mM) the concentration of metal ions in solution at equilibrium, qm (mmol g−1), KL (l mmol−1), KF, 1/n, qmax (mmol g−1), KS (l mmol−1) and βS the corresponding equation constants.

The kinetic adsorption experiments were also measured by immersing the wet beads (0.04 g dry mass per 100 mL solution) in a series of 0.1 mM Cu2+ and Pb2+ solutions at pH 6 for Cu2+ and pH 5.3 for Pb2+, respectively, and shaking for different time intervals. Several adsorption kinetic models such as pseudo-first order, pseudo-second order and intraparticle diffusion kinetic models (Ngah and Fatinathan 2008, 2010) were used to understand the adsorption kinetics and the rate limiting step during adsorption process. The kinetic models are given as:

$$ln(q_{e} - q_{t} ) = lnq_{e} - k_{1} t$$
$$\frac{t}{{q_{t} }} = \frac{1}{{k_{2} q_{e}^{2} }} + \frac{t}{{q_{e} }}$$
$$q_{t} = k_{\text{int}} t^{1/2} + C$$

where qt (mmol g−1) is the ion amount adsorbed onto the beads at time t, k1 (h−1), k2 (g mmol−1 h−1), kint (mmol g−1 h1/2) are the rate constants of kinetic models, and C is the intercept which is associated to the boundary layer thickness.

Results and discussion

Regenerated lignocellulosic suspensions before attempting the fabrication of hydrogel beads

For the purpose of comparing the lignin-containing lignocellulosic hydrogel beads with the pure cellulosic ones in their properties and adsorption ability, we initially needed to identify what role, if any, lignin plays in the microstructure of the regenerated lignocellulose fibrils. Therefore, before attempting the fabrication of hydrogel beads, we coagulated poplar wood powder (POP), cellulose-rich material (CRM), Kraft pulp (KP), and microcrystalline cellulose (MCC) IL solutions in ultrapure water as the anti-solvent to produce 0.05 wt% aqueous suspensions. These suspensions were dropped on carbon-coated copper grids, ultrasonicated and vacuum-dried. The remnants after drying of the suspension was analyzed using TEM.

Interestingly, the remnants from lignin-free KP and MCC suspensions exhibited a higher degree of fibril entanglement where individual fibrils in pure suspensions could not be seen (Fig. 1a, b). However, entanglement phenomenon in the remnant from lignin-contained CRM was considerably less (Fig. 1c). The remnant from POP suspension, which has a higher lignin content than CRM as observed from the aromatic skeletal vibrations of lignin at 1589/1506 cm−1 in FT-IR (Labbe et al. 2005) in Fig. S2, displayed a great number of lignin globule aggregates (black spots in Fig. 1d).

Fig. 1
figure 1

TEM micrographs of the dry remnants obtained from the cellulose (a Kraft pulp—KP, b microcrystalline cellulose—MCC) or lignocellulose (c CRM, d poplar wood powder—POP) aqueous suspensions (0.05 wt%) after regenerating IL solutions with ultrapure water, ultrasonically treating, and vacuum drying on carbon-coated copper grids

Full size image

The higher polymer entanglement in pure regenerated cellulose (Scheme 1a) can be attributed to the strong interaction between the pure cellulosic fibrils via hydrogen bonds. It can be inferred that a relatively low content of residual lignin in the CRM, as a three dimensional stiff macromolecular network, may restrict the approach of cellulosic fibrils thereby reducing formation of hydrogen bonds between the hydroxyl groups of the polysaccharide polymers (Scheme 1b). As the result, the entanglement of the polymer chains is inhibited, and the hydroxyl groups remain freely accessible. However, excessively high amount of hydrophobic lignin (e.g., POP) may create large “lignin globules” in POP suspensions (Scheme 1c), thereby covering the polysaccharides, and inhibiting the accessibility of the hydroxyl groups. In addition, the branched molecular structure of residual hemicellulose in CRM or POP may also impede the cellulosic polymers from forming denser fibril aggregates (Duchesne et al. 2001).

Scheme 1
scheme 1

Diagrammatic illustrations of the interaction between the macromolecules in 0.05 wt% aqueous suspensions of the regenerated cellulose (a KP or MCC) or lignocellulose (b: CRM, c poplar wood powder—POP)

Full size image

Preparation and characterization of cellulosic beads prior to derivatization

Having clarified the effect of lignin on the microstructure of the regenerated lignocellulose fibrils, we then proceeded to the preparation and comparison of the hydrogel beads. The cellulosic and lignocellulosic raw powder used could be reconstituted to form hydrogel beads from their [C2mim][OAc] solutions by dropping the solutions into water anti-solvent. The appropriate concentrations of the solutions were found to be 3.5% for POP, 2% for KP, 2% for CRM, and 5% for MCC. At these concentrations the corresponding lignocellulose- and cellulose-IL solutions showed a proper viscosity or fluidity for the following beads preparation (Fig. S3). The pure cellulosic beads (KP2 and MCC5) appeared white, while lignocellulosic beads (POP3.5 and CRM2) looked yellow (Fig. S4, insets) due to the presence of lignin (Table 1). The CRM2, KP2, and MCC5 original beads were round and exhibited a Gaussian-like distribution in diameter, with mean diameters of approximate 3.78, 2.52, and 3.60 mm, respectively (Fig. S4). This suggests that the presence of residual lignin and the higher substance mass are favorable factors in achieving larger diameter values. The 3.5% POP solution was extremely hard to form droplets due to the highest viscosity, resulting in ellipsoid-like POP3.5 beads with an average minor axis as low as 2.24 mm (Fig. S4).

Table 1 Lignin content, hydroxyl number, density, porosity, and water content of the hydrogel beads
Full size table

The freeze-dried beads exhibited a porous, low crystalline (Fig. S5), three-dimensional architecture with thin lamellar walls (Fig. 2). The macro-pores were interconnected, allowing access of the adsorbate (e.g., metal ion solution) to the adsorptive sites. The lamellar walls of the lignocellulosic beads (POP3.5 and CRM2) were in ordered alignment and the interior had a wood cell-like porous structure (Fig. 2a, b). Compared to the lignocellulosic beads, the pore structure of the pure cellulosic beads (KP2 and MCC5) appeared less ordered and some even collapsed (Fig. 2c, d), which may be due to formation of hydrogen bonds between polysaccharide polymers. At higher magnifications (5000–10,000 ×), micro-pores can be observed on the rough walls of the lignocellulosic beads, while the cellulosic beads exhibited smooth and dense walls. The ordered macro-porous structure and presence of micro-pores in the walls of lignocellulosic beads can be attributed to the stiff macromolecular network of lignin in the beads that can support the micro- and/or macro-porous structure and restrict the formation of hydrogen bonds. Such increase in the amount of the nano-scale pores due to the presence of lignin was also found in literature, which was observed by the specific surface area for nanofibrillated cellulose suspensions (Ferrer et al. 2012) and calorimetric thermoporometry for delignified biomass (Driemeier et al. 2016).

Fig. 2
figure 2

SEM micrographs of POP3.5 (Column a), CRM2 (Column b), KP2 (Column c), and MCC5 beads (Column d) at different magnifications (the scale bars are 500, 50, 5 or 2 μm, respectively)

Full size image

The CRM2 beads had a lignin content of 12.0 (4)% that was comparable to the raw CRM powder (12.2 (3)%) (Shen et al. 2016b), which were approximately half that of the original wood (22.3 (6)%; Table 1). The lower lignin content was mainly attributed to the leaching of lignin from the IL-dissolved wood powder to the acetone/water solution during preparation of CRM powder (175 °C). The second dissolution from CRM powder to CRM2 hydrogel beads did not cause further reduction in lignin content, because the IL dissolution was performed under more mild condition (100 °C) and the dissolved hydrophobic lignin would only be slightly leached into water. Compared to the CRM2 beads, POP3.5 had a higher lignin content (18.4 (4)%), which can be explained by the fact that POP3.5 beads only underwent one IL dissolution process and were reconstituted in water.

Lignin in the lignocellulosic beads is approximately the “native lignin” after dissolution in the non-derivative IL solvent and reconstitution in water anti-solvent, which has a high degree of polymerization and thus a low phenolic hydroxyl number. The hydrogel beads prepared in this study had more accessible –OH groups than the pristine wood powder (Table 1), which was attributed to the exposure of the hydroxyl groups by disintegration of the cell wall structure and delignification of the resulting cell wall compositions.

Among all beads, the CRM2 beads possessed the largest number of –OH groups (28.7 mmol KOH g−1), followed by POP3.5 (25.8 mmol KOH g−1), KP2 (23.2 mmol KOH g−1), and MCC5 (19.4 mmol KOH g−1). This suggests that an appropriate amount of lignin in the beads can support the porous structure to access the exposed hydroxyl groups (e.g., CRM2) on both carbohydrates and lignin, which is consistent with the emergence of the micropores as observed by SEM (Fig. 2b). However, excess lignin may agglomerate to envelop the polysaccharides (Scheme 1c), thereby inhibiting the access of the hydroxyl groups on the pore walls (e.g., POP3.5). For pure cellulosic beads, the higher –OH number of KP2 than MCC5 may mainly be explained by the presence of amorphous substance in KP2 that has more accessible hydroxyl groups compared to MCC5 that only the hydroxyl groups on the surface of microcrystalline are available.

The density of the beads was primarily related to the cellulose or lignocellulose load in the IL solution. Therefore, it is unsurprising that the MCC5 and POP3.5 beads had high densities (0.113 and 0.099 g cm−3, respectively)and thus low porosities (93.2% and 94.2%, respectively; Table 1). The CRM2 beads possessed the lowest density (0.042 g cm−3) and thus the highest porosity (97.0%). At a comparable concentration level, the KP2 exhibited higher density than the CRM2, which can be attributed to formation of more hydrogen bonds, resulting in the collapse of the pores in the KP2 beads. The water uptake ability of the beads seemed closely related to their porosity, with the water content of the CRM2 beads being the highest (97.0%; Table 1).

Characterization of modified hydrogel beads

Functionalization efficacy during carboxymethylation

Carboxymethylation of the beads proceeded via etherification of the accessible hydroxyl groups of polysaccharides (mainly cellulose) with sodium monochloroacetate under alkaline condition (Scheme S1), resulting in formation of carboxymethylated beads, CPOP3.5, CCRM2, CKP2 and CMCC5. The degree of substitution (DS), the average number of substituent (here, carboxymethyl-) groups attached per base unit, was positively related to the number of accessible hydroxyl groups (Fig. 3a). Thus, the substitution efficacy of CRM2 was almost three times higher than that of MCC5, which further suggested the support effect of lignin in the CRM beads, allowing better access of monochloroacetate to the hydroxyl groups.

Fig. 3
figure 3

a Effect of the hydroxyl number of the original beads on the degree of substitution (DS) of the carboxymethylated beads; and b effect of the density of the original beads on the content of Fe3O4 incorporated in the magnetized beads

Full size image

Functionalization efficacy during magnetization

To facilitate the removal of beads from solutions, we attempted the in situ synthesis of Fe3O4—one of the commonly-used magnetic adsorbents (Banerjee and Chen 2007; Mahdavi et al. 2012), into the structure of “original” beads with the precursor solution containing 15/30 mM FeCl2·4H2O/FeCl3·6H2O via a hydrothermal method. The resultant magnetized beads, MPOP3.5, MCRM2, MKP2 and MMCC5, became black and the Fe3O4 content in the beads was determined by weighing the dry masses of the beads before and after magnetite incorporation. The results showed that the Fe3O4 content was reduced with increasing density of the original beads (Fig. 3b). The MCRM2 beads had the highest Fe3O4 load, which is also consistent with their highest porosity and –OH number. MPOP3.5 had a slightly higher Fe3O4 content than MKP2 despite of the higher density, which can be attributed to the presence of the micropores in the MPOP3.5.

Magnetized/carboxymethylated hydrogel beads

After dual functionalization, the magnetized/carboxymethylated beads were of brown color, different from the black color of the “magnetized only” beads (Fig. 4a, insets), which is consistent with the results reported in the literature (Nata et al. 2011; Tang et al. 2012). Under external magnetic field, magnetized/carboxymehylated beads exhibited a comparable saturation magnetization (Ms) value to that obtained for the “magnetized only” beads (Fig. 4a). This result revealed that the removal of Fe3O4 nanoparticles (Fig. S6) from the hydrogel beads during subsequent carboxymethylation was negligible, suggesting a strong interaction between the iron oxide and the lignocellulose polymers. The Fe3O4 aggregates were successfully precipitated on the bead’s surface and lamellar walls as verified by SEM (Fig. 4b, c), exhibiting sizes ranging from tens to hundreds of nanometers.

Fig. 4
figure 4

a The field-dependent magnetization curves, and b, c the micrographs of the inner structure of the MCCRM2 bead’s lamellar walls at × 500 and × 5000 magnifications, respectively

Full size image

Powder X-ray Diffraction (PXRD) of the magnetized beads exhibited (220), (311), (400), (422), and (511) planes at 2θ values of 30.0°, 35.3°, 43.0°, 53.5° and 57.0°, confirming the formation and introduction of black crystalline Fe3O4 in the beads (Fig. 5a) (Liu et al. 2016). The peaks’ intensities of the magnetized/carboxymethylated beads hardly changed compared to those of the “magnetized only” beads, implying again the firm deposition of Fe3O4.

Fig. 5
figure 5

a PXRD patterns and b FT-IR spectra of the functionalized beads [“(P)” represents “powder”]; and c the possible modes of covalent binding of monochloroacetate onto Fe3O4 incorporated in the beads according to Togashi et al. (2011)

Full size image

The success of etherification by sodium monochloroacetate was confirmed by the emergence of the 1595 cm−1 peak assigned to the carbonyl (C=O) asymmetric vibration of the carboxylate anions (–COO−) in the FT-IR spectrum (Fig. 5b), in agreement with the literature value of 1605 cm−1 for carboxymethylated cellulose (Qi et al. 2009). Interestingly, after magnetization and carboxymethylation, the intensity of the carboxylate vibration peak in the functionalized beads (MCCRM2) was considerably enhanced compared to the “carboxymethylated only” beads (Fig. 5b), suggesting that the number of carboxylate groups as adsorptive sites increased after combinative functionalization. This was most likely due to the occurrence of carboxymethylation not only on hydroxyl sites of carbohydrates, but also on the incorporated Fe3O4 nanoparticles in the beads via binding of hydrolyzed monochloroacetate to the atomic surface of Fe3O4 (Fig. 5c) (Togashi et al. 2011). Additionally, the hemicellulose peak at 1725 cm−1 (Sun et al. 2009) disappeared, suggesting the leaching out of hemicellulose after functionalization (Fig. 5b). This result indicates that functionalized lignocellulosic hydrogel beads consist mainly of lignin and cellulose.

Heavy metal adsorption

For evaluating the effects of lignin in the beads -“original” (O), “carboxymethylated” (C), “magnetized” (M), and “magnetized/carboxymethylated” (MC), on the adsorption ability, wet beads (0.01 g dry mass) were immersed in 25 mL of 0.1 mM Cu(NO3)2·3H2O or Pb(NO3)2 solution with pH value varying from 4.0 to 6.0, followed by incubating at 25 °C for 24 h under shaking. After adsorption, the beads were removed and the absorbance of the ions in the solution was determined using an atomic absorption spectrophotometer (AAS).

The original hydrogel beads exhibited an ascending order of adsorption ability for Cu2+ and Pb2+: MCC5, KP2, POP3.5, and CRM2 (Fig. 6), under the optimum conditions (pH 6 for Cu2+ and pH 5.3 for Pb2; Fig. S7). This is consistent with the increased order in the number of hydroxyl groups (Table 1) on both carbohydrates and lignin that are the adsorptive site for the metal ions. Both lignocellulosic hydrogel beads (POP3.5 and CRM2) adsorbed more metal ions than the pure cellulosic beads (MCC5 and KP2), which can be explained by the presence of micropores on the pore walls (Fig. 2) and the greater porosity of the lignocellulosic beads (Table 1). POP3.5 had higher lignin content but lower adsorption ability than CRM2, suggesting again that an excessive amount of hydrophobic lignin likely agglomerates and resultantly encapsulates polysaccharide macromolecules and thus occludes the hydroxyl groups.

Fig. 6
figure 6

The adsorption ability of the original (O), carboxymethylated (C), magnetic (M), and magnetized/carboxymethylated (MC) hydrogel beads for a Cu2+ and b Pb2+ ions [standard deviation < 0.004 mmol g−1; 0.04 g dry weight of prepared gel beads in 100 mL of 0.1 mM Cu(NO3)2·3H2O or Pb(NO3)2 solution; 24 h vibration, 25 °C, pH 6.0 for Cu2+ and 5.3 for Pb2+]

Full size image

Carboxymethylation improved the adsorption for Cu2+ and Pb2+ compared to the original beads. For instance, CCRM2 had adsorption for ions approximately 2.5 times the amount of the original CRM2 due to the high degree of substitution, while the adsorption ability of CMCC5 was around 1.6 times that of its original MCC5 (Fig. 6). The improved adsorption capacity can be explained by the higher coordinating affinity of the carboxylate anions with the divalent metal ions than the hydroxyl groups (Kikuchi et al. 2006). Magnetization by in situ synthesizing Fe3O4 nanoparticles in the original beads also resulted in an increase in the adsorption for metal ions, especially for Cu2+, most likely caused by the higher adsorption (mmol g−1) of Fe3O4 for Cu2+ than for Pb2+ ions (Mahdavi et al. 2012).

The adsorptive capacity of the beads successively treated with magnetization and carboxymethylation exceeded the sum of the adsorption of the beads individually treated. For instance, CCRM2 and MCRM2 had more than double the capacity of the original CRM2; after combinative treatment, the resultant magnetized/carboxymethylated beads loaded with carboxylate Fe3O4 exhibited an adsorption capacity around 0.13 mmol g−1 for both Cu2+ and Pb2+ ions (Fig. 6), which were 6 times of the adsorption of CRM2. This may be explained by two reasons: (1) the number of carboxylate groups as adsorptive sites increased in the “MC” beads due to the carboxymethylation of both the hydroxyl groups on lignocellulose polymers and on Fe3O4 particles; and (2) Fe3O4 particles in the “MC” beads are covered with Fe–OH or Fe–O− groups in water, which can directly coordinate with Pb2+ and Cu2+ ions and form mono- and binuclear inner-sphere complexes (Horst et al. 2015; Mahdavi et al. 2012; Upadhyay et al. 2014). The adsorption was confirmed by the EDX spectra (Fig. S8).

Adsorption modeling analysis

The correlation between the adsorption capacity of the beads and the equilibrium concentration of the metal solution after adsorption was studied for MCCRM2 which had the best adsorption ability among all the beads. It was found that the amount of metal adsorbed by the beads increased as the metal concentration in the solution increased, until the adsorption approached an equilibrium state (Fig. 7). The regression analyses of our experimental data for both Cu2+ and Pb2+ were shown to well correspond to the Langmuir relationship (R2 = 0.985 and 0.996, Table 2), indicating homogeneous, monolayer adsorption. The calculated maximum adsorption capacity (qm) reached 0.144 mmol g−1 and 0.161 mmol g−1 for Cu2+ and Pb2+, respectively.

Fig. 7
figure 7

Non-linear Langmuir, Freundlich and Sips isotherms for the adsorption of a Cu2+ and b Pb2+ ions onto the MCCRM2 beads

Full size image
Table 2 Constant parameters and correlation coefficients calculated using various isotherm models for the adsorption of Cu2+ and Pb2+ ions on MCCRM2
Full size table

The Langmuir constant KL derived from the Langmuir model is related to the affinity of the binding sites to the metal ions, that is, generally the higher adsorption, the larger KL value. However, although the beads had a higher adsorption for Cu2+ ions than for Pb2+ at low concentrations (consistent with the higher KL for Cu2+), their adsorption at high concentrations or their qm for Cu2+ was lower (Table 2). Volesky (2003) has reported this phenomenon, i.e., bio-adsorbents adsorbing better at low concentrations may be inferior at higher concentrations, and vice versa. In the case of high metal concentrations, the influence of the adsorbate nature including distribution coefficient (i.e., metal mobility in solution) and hydrated radius on the adsorption of the adsorbents may become larger. Pb2+ ions had a greater distribution coefficient (Li et al. 2011b) and a smaller hydrated radius (Qiu and Zheng 2009) than Cu2+ ions, hence justifying its growing uptake at high concentrations.

The Freundlich model assumes that the adsorbent has energetically heterogeneous surface and has different affinity for adsorption, in which the 1/n value is related to the degree of surface heterogeneity, i.e., smaller 1/n indicates more heterogeneous surface (Ngah and Fatinathan 2010). However, the low R2 values illuminate again homogeneous adsorption of the beads towards metal ions. In the Sips isotherm model, since the βS values are close to 1 (Table 2), the equations nearly reduce to the Langmuir equation.

From Fig. 8a, b and Table 3 it could be seen that the normalized standard deviation of pseudo-second order model was smaller than that of pseudo-first order model. Therefore, pseudo-second order model was more suitable to predict the ions adsorption process of beads in our experiments. As a result, it can be concluded that the chemical adsorption is dominant in the adsorption process, rather than physical adsorption described by the pseudo-first order model.

Fig. 8
figure 8

a Pseudo-first order, b pseudo-second order and c intraparticle diffusion kinetics plots for the adsorption of Cu2+ and Pb2+ ions onto the MCCRM2 beads

Full size image
Table 3 Comparison of pseudo-first order, pseudo-second order and intraparticle diffusion kinetic models for the adsorption of Cu2+ and Pb2+ ions on MCCRM2
Full size table

Intraparticle diffusion model was used to analyze the adsorption process, as shown in Fig. 8c. If the plot of qt versus t1/2 is multi-linear, other rate-limiting step along with intraparticle diffusion resistance also plays a major role on the adsorption process (Song et al. 2016). The overall rate of adsorption on the beads can be described by the following three steps (Ren et al. 2016): (1) surface diffusion of the sorbate from the bulk solution to the external surface of sorbent, which was controlled by the chemical adsorption, (2) intraparticle or pore diffusion of the sorbate molecules in the interior of sorbent, and (3) adsorption–desorption balance. Therefore, the overall rate of the adsorption process by the functionalized lignocellulosic beads was controlled by both surface diffusion and intraparticle diffusion.

Conclusions

Lignocellulosic beads (CRM2 and POP3.5) exhibited the presence of micropores, and higher hydroxyl number and porosity compared to pure cellulosic beads (KP2 and MCC5) due to the porous structure support effect of the stiff lignin network, and thus higher functionalization efficacy after carboxymethylation and magnetization. Therefore, functionalized lignocellulosic beads had higher adsorption capacity for heavy metal ions than the analogous cellulosic beads, with maximum adsorption capacities (qm) of 0.144 mmol g−1 and 0.161 mmol g−1 for Cu2+ and Pb2+, respectively, derived from the Langmuir model. However, excessive lignin may agglomerate and occlude the adsorptive sites, thereby leading to a lower adsorption ability of the beads prepared directly from wood either before or after functionlization. Since lignocellulosic hydrogel beads can be obtained in a simpler, more economical and environmentally-friendly approach than pure cellulose beads, the preparation of lignocellulosic adsorbents may be more practical and competitive.