Adsorption isotherms and mechanisms of Cu(II) sorption onto TEMPO-mediated oxidized cellulose nanofibers

Abstract

2,2,6,6-Tetramethyl-1-piperidinyloxy (TEMPO) mediated oxidized cellulose nanofibers (TOCNF) have shown potential in the removal of metal ions from contaminated water owing to their abundant carboxylate groups on their surface, functioning as metal sorption sites. The current work aims to study the kinetics and thermodynamics of the sorption behavior of Cu(II) onto TOCNF, and verify the correlation between Δ[H⁺] and the corresponding Δ[Cu(II)] in aqueous solution during sorption. Sorption of Cu(II) onto TOCNF was found to be an exothermic process with fast kinetics; reaching equilibrium Cu(II) adsorption in less than 1 min. The sorption data fits well with Langmuir isotherm models. The SEM imaging of the TOCNF after Cu(II) sorption revealed interesting copper-containing nanoparticles, which was further analyzed by using XRD. Besides, a linear correlation between Δ[H⁺] and the corresponding Δ[Cu(II)] in the solution was found, which indicates that the Cu(II) sorption capacity might be well predicted and calculated by Δ[H⁺] or pH variation during Cu(II) ion sorption onto TEMPO oxidized nanocellulose fibers and have potential to develop online sensors for tracking metal ion removal.

---

Introduction

The removal of heavy metals from wastewater has imposed a tremendous challenge to human society due to their hazardous effect on human health and the environment. Among the heavy metals, copper is one of the most common heavy metal pollutants in industrial effluents and is thus the target metal in the current study. Many industrial fields, such as mining, copper-brass plating, paper, dyeing and petroleum copper-ammonium rayon produce effluent containing copper ions.^1–3 The removal of copper like many other heavy metals from effluents is vital not only due to the hazardous effect to environment, but also the possibility to recover and reuse copper in many industrial applications.⁴

In recent years, nanocelluloses and their surface modified versions have received considerable attention as nanobioadsorbents for metal ion removal owing to their excellent metal binding capacities.^5–8 The nanoscaled fibrous structures results in high specific surface area with numerous adsorption active sites and short intraparticle diffusion distance, which provides a great advantage for both metal absorbability and fast kinetics.^9,10 It was reported that the specific surface area of cellulose nanofibers (CNFs) is 229 m² g⁻¹ and the figure rises to 345 m² g⁻¹ (BET method) after chemical surface oxidation.⁷ Besides, the easily modifiable surface of nanocellulose, coupled with its nontoxicity, hydrophilic properties, excellent mechanical properties, low cost, indicate potential applications of heavy metal removal. Although our previous studies have shown that both unmodified cellulose nanocrystals and nanofibers have capacities to immobilize heavy metals, the grafting of metal-binding functional groups, such as carboxylate, amino, and phosphate groups, onto the surface of nanocellulose can remarkably increase the metal removal capacity.^4,6,7,11

TEMPO (2,2,6,6-tetramethyl-1-piperidinyloxy)-mediated oxidation is an efficient and low energy consumption way to selectively convert C6 primary hydroxyls exposed on the surface of crystalline cellulose microfibrils to sodium carboxylate groups at pH 10 and room temperature.¹² The carboxylate content depends on the input amount of NaClO as the primary oxidant.¹³ Isogai et al. reported that TEMPO mediated oxidized cellulose nanofibers (TOCNF) have the capacities to adsorb a variety of heavy meal ions in aqueous solutions.¹⁴ Our previous studies show that the copper sorption capacity of CNFs was improved by tenfold higher to 113 mg g⁻¹ after TEMPO oxidation (sorption pH = 6.2). The copper sorption on TOCNF was found to rise linearly with the carboxylate content.⁷ In addition, copper contained nanoparticles with a size from several tens to hundreds of nanometers could be observed clearly on the surface of the TOCNF after copper sorption,¹⁵ which provides a possibility to recycle copper or convert the TOCNF saturated with copper contained nanoparticles into a variety of value-added products.^16,19

Although these previous studies have covered the topic of copper sorption by TOCNF, the studies focused mainly on its metal sorption capacities and surface property variations during sorption. Meanwhile, to our knowledge, there have been few reports on the kinetic and equilibrium isotherm studies of Cu(II) onto TOCNF, which is the main target of the current studies. It is believed that kinetics and equilibrium isotherm studies are necessary for deeper understanding of Cu(II) sorption behavior onto TOCNF and the correlative mechanism. In the current works, the effects of contact time and temperature on Cu(II) sorption onto TOCNF were studied. Langmuir and Freundlich models are used to describe the sorption isotherms and calculate the thermodynamic values related to the interaction of Cu(II) ions with TOCNFs coupling with different amounts of carboxylate groups. The correlation between Δ[H⁺] and the corresponding Δ[Cu²⁺] in the solution during sorption is also presented.

Experimental

Materials

Cellulose sludge (supplied by Domsjö Fabrikerna AB, Örnsköldsvik, Sweden) was used as the raw material for the preparation of 2,2,6,6-tetramethyl-1-piperidinyloxy mediated oxidized cellulose nanofibers (TOCNF). The cellulose sludge was reported to be high in cellulose (95%) with some hemicellulose (4.75%) and trace amounts of lignin.¹⁷ The cellulose sludge was used without any pretreatment for isolation into nanocellulose. TEMPO free radical and sodium hypochlorite (NaClO) solution (reagent grade, 10–15% chlorine) were purchased from VWR International. Copper nitrate nona-hydrate (Cu(NO₃)₂·9H₂O) was purchased from VWR, Sweden and used for sorption studies. All the chemicals were of reagent grade or higher in purity and were used as received.

Preparation of TEMPO-oxidized cellulose nanofibers (TOCNFs)

TOCNF aqueous suspension was prepared from cellulose sludge according to a method previously reported by Saito et al. and was supplied by EMPA (Swiss Federal Laboratories for Materials Science and Technology).¹⁸ The pulp residue slurry was first mechanically beaten and then dispersed in water containing sodium bromide and TEMPO (1 mmol and 0.1 mmol per gram of pulp residue, respectively). The concentration of the pulp residue in water was 2 wt%. The reaction was performed by the drop-wise addition of sodium hypochlorite to the suspension while maintaining the pH of the reaction at ca. 10 by sodium hydroxide addition. Two different degrees of oxidation were achieved by choosing two different amounts of hypochlorite (2 and 10 mmol per gram of cellulose sludge). After all of the NaClO was consumed, the oxidized pulp residue was filtered and washed several times with deionized water until the filtrate solution was neutral. The purified pulp fibers were then dispersed in water at a concentration of ca. 0.5 wt% and disintegrated in a Microfluidizer M-110Y (Microfluidics Ind., USA) to achieve an aqueous suspension of TOCNF. TOCNF with the carboxylate content of 0.6 mmol L⁻¹ (TOCNF0.6) and 1.5 mmol L⁻¹ (TOCNF1.5) were achieved, as determined by an electric conductivity titration method.

Morphology and specific surface area

The morphologies of TOCNF0.6 and TOCNF1.5 were observed by using FEG-Scanning electron microscopy (Zeiss, Merlin). The images were taken using the secondary electron and InLens secondary electron detector with a voltage of 15 kV and a current of 15 pA. The samples were sputter coated with tungsten and the coatings were less than 10 nm in thickness. The diameter measurements were conducted with the aid of SemAfore software. Each average value of the nanofiber diameter was calculated from 50 random measurements.

The SEM images presented in Fig. 1 show the surface morphologies of oxidized cellulose nanofibers with different functional group contents. As shown in the figure, oxidized cellulose nanofibers with different degrees of oxidation tend to have similar fibrous shapes. Nevertheless, TOCNF1.5 was observed to have a smaller diameter than TOCNF0.6, due to the harsher applied oxidation conditions. As measured by using the SemAfore software, the average diameters of TOCNF0.6 and TOCNF1.5 are 29.8 nm and 18.6 nm respectively. The 50 randomly measured fiber diameters of TOCNF0.6 are distributed in a wider range (20–40 nm) compared with those of TOCNF1.5 (15–24 nm), which indicates better homogeneity for TOCNF structures with a higher degree of oxidation or carboxylation. The calculated specific surface areas (SSA) are 134 m² g⁻¹ and 215 m² g⁻¹ respectively for TOCNF0.6 and TOCNF1.5.¹⁹

Fig. 1 Morphologies of TOCNF0.6 and TOCNF1.5.

Batch kinetic studies

CuNO₃ stock solution with a Cu(II) concentration of 400 mg L⁻¹ was prepared from Cu(NO₃)₂·9H₂O in Millipore water. Batch kinetic studies were performed by shaking 50 mL of Cu(II) solution containing 0.025 g TOCNF0.6 (or TOCNF1.5) operated at the temperature of 25 °C. To prevent copper hydroxide precipitation during sorption study, the pH of the metal solution was maintained at pH 5 by the addition of HNO₃ (0.01 mol L⁻¹) or NaOH (0.015 mol L⁻¹) solution (ion product < K_sp). The effect of initial Cu(II) concentrations were studied at 2 mg L⁻¹, 50 mg L⁻¹ and 200 mg L⁻¹.

The samples after sorption for a certain contact time (1 min, 30 min, 1 h, 3 h, 6 h, 9 h, 12 h) were then filtered through a commercial vacuum filter with the membrane pore size of 0.45 μm (DVPP, Millipore). The filtrate obtained was used for inductively coupled plasma-optical emission spectrometric (ICP-OES) analysis. An ICP-OES, Optima 2000 DV, Perkin Elmer, (USA) with a radial torch was used to determine the Cu(II) concentrations of the filtrates. The amount of Cu(II) sorption q_t (mg g⁻¹) at certain contact time t, was obtained as follows:

where C₀ (mg L⁻¹) and C_t (mg L⁻¹) are the Cu(II) concentrations in the solution at time t = 0 and at the given time t, respectively. V (L) is the volume of the solution and W (g) is the weight of adsorbent in the solution. Control experiments using CuNO₃ solution showed that the DVPP membrane used did not adsorb Cu(II) ions.

Studies on the effect of temperature

The batch temperature experiments were conducted in a similar way to the kinetic studies. The effect of temperature was studied at an initial Cu(II) concentration of 50 mg L⁻¹ (pH: 5). The sorption experiments were conducted in thermal baths maintained at 25, 40 and 55 °C respectively for 3 h and 5 replicates were used at each temperature. The samples after adsorption and vacuum filtration were analyzed using (ICP-OES). The adsorbed amount of Cu(II) was calculated using eqn (1).

Batch isotherm studies

The batch adsorption isotherm studies were performed in a similar way to batch pH studies using Cu(II) initial concentration (C₀) of 2 mg L⁻¹ to 240 mg L⁻¹ (10 data points) at 25 °C. The adsorption contact time for all the samples were 3 hours. The initial and equilibrium Cu(II) concentrations were detected by using ICP-OES. The adsorbed amount at equilibrium (q_e) was calculated by using the following equation:where C_e (mg L⁻¹) is the Cu(II) equilibrium concentration in the solution at equilibrium time.

The pHs of all samples before sorption was controlled in the range of 5–5.4 by the addition of HNO₃ (0.01 mol L⁻¹) or NaOH (0.015 mol L⁻¹). The pHs of all the samples after Cu(II) sorption were detected by pH meter (Checker, Hanna Instrument, USA). All the pH data before and after sorption were carefully recorded and used to study the correlation between the concentration variations of hydrogen ions and Cu(II) ions in the solution during Cu(II) sorption.

Surface characterizations of TOCNF after Cu(II) sorption

TOCNF1.5 filter cakes obtained from batch kinetic studies (initial Cu(II) conc. = 200 mg g⁻¹) were subjected to SEM and X-ray diffraction (XRD) to analyze the surface of TOCNF after Cu(II) sorption and the chemical composition of adsorbed copper. XRD analysis was conducted at the same temperature by step scanning using Siemens X-ray diffractometer D5000 (Berlin, Germany). The angle of incident monochromatic X-ray was in the range of 2θ = 10–80° and the step size was 0.0263. The wavelength of the monochromatic X-ray was 1.540598 Å (Kα1)°.

Results and discussion

Effect of contact time and temperature

The rate of the Cu(II) sorption and equilibrium sorption time (t_e) were determined for TEMPO-oxidized cellulose nanofibers with the carboxylate content of 0.6 mmol L⁻¹ (TOCNF0.6) and 1.5 mmol L⁻¹ (TOCNF1.5) at different Cu(II) initial concentrations. As expected, higher initial Cu(II) concentration leads to higher amount of copper uptake on the adsorbents.

It was found in Fig. 2(a and b) that Cu(II) uptake amounts go up dramatically at the first minute for both TOCNF0.6 and TOCNF1.5 and keep rather constant at the following time points. The equilibrium sorption kinetic curves of TOCNF are different from the most common obtained ascending kinetic curves as a function of time.

Fig. 2 (a and b) Effect of contact time on sorption at varying initial Cu(II) concentrations (pH = 5, T = 25 °C). (c) Effect of temperature on Cu(II) sorption capacity (pH: 5, 5 measurements for each temperature).

The fast kinetics of sorption on TOCNF regardless of surface carboxylation degree might be due to two factors: the nanoscale diameter and hydrophilic property. It was expected that the external diffusion (transport of adsorbate from the bulk liquid phase to the boundary layer of external surface) is fast, since the sorption experiments were conducted under strong stirring. Hence, the sorption rate is mainly determined by surface diffusion (transport in the adsorbed state along the internal surface) and rate surface sorption reaction (energetic interaction between the adsorbate and the final sorption sites) rate.¹⁰ On the basis of the model solutions for surface diffusion provided by Suzuki and Kawazoe, the following equations for the minimal sorption equilibrium time, t_min, was derived:²⁰

where r is the adsorbent particle radius, D_S is the surface diffusion coefficient, T_B,min is the minimal dimensionless time necessary for approaching the equilibrium. In accordance with the eqn (3), the particle diameter of the adsorbent has a strong influence on the required equilibration time. The equation offers the insight that the nano-scaled diameter of TEMPO oxidized cellulose nanofibers is a critical factor influencing the fast rate of equilibrium. Besides, owing to the abundance of carboxyl and carboxylate groups on the surface, oxidized cellulose nanofibers are very hydrophilic and well dispersed in Cu(II) solution, which results in short diffusion distance. The fast kinetics also indicates the fast surface sorption reaction between Cu(II) ions and TOCNF. The fast kinetics for TOCNF to capture copper ions is of significance from the perspective of industrial application.

The effect of temperature on the sorption capacity of Cu(II) onto TOCNF0.6 and TOCNF1.5 is shown in Fig. 2(c). It displays that Cu(II) ion sorption capacities decreases by 2.8% (TOCNF0.6) and 8.9% (TOCNF1.5) with increasing the temperature from 25 °C to 55 °C. It is shown that temperature variation has a bigger impact to Cu(II) sorption of the oxidized nanofibers with higher carboxylation. The downward trend in sorption capacities indicates that capture or sorption of copper ions on the energetically active sorption sites on the surface of the oxidized cellulose nanofibers is an exothermic process (or enthalpy driven process). When sorption takes place, the surface free energy is reduced from the initial surface tension at the TOCNF-solution interface to the new surface tension at the TOCNF–Cu(II)-solution interface.

Adsorption isotherm

The well-known equations proposed by Langmuir (1918) and Freundlich (1906) are the most widely and frequently used two-parameter adsorption isotherms. Many three-parameter isotherms, such as Redlich–Peterson isotherm, are derived from Langmuir isotherm and Freundlich isotherm.¹⁰ The Langmuir isotherm model is representative of monolayer adsorption occurring on energetically homogenous surface on which the adsorbate molecules are not interactive. The equation is expressed as:²¹where q_e (mg g⁻¹) is the equilibrium sorption capacity and Q (mg g⁻¹) is the maximum sorption amount of Cu(II) ion per unit weight of TOCNF to form complete monolayer coverage on the surface at equilibrium Cu(II) ion concentration C_e (mg L⁻¹). b is the Langmuir constant and is related to the affinity of binding sites.

Fig. 3(a) presents the experimental equilibrium isotherms for sorption of Cu(II) on to TOCNF with different carboxylations, 0.6 mmol g⁻¹ (TOCNF0.6) and 1.5 mmol g⁻¹ (TOCNF1.5), respectively. TOCNF with higher carboxylation has higher Cu(II) sorption capacity and the sorption amount gap between the two adsorbents widens with the rising of Cu(II) concentration.

Fig. 3 (a) Langmuir Cu(II) adsorption isotherms (W: 0.025 g; V: 50 mL, C₀: 2–240 mg L⁻¹; pH: 5–5.3, contact time: 3 h). (b) The linearized Langmuir adsorption isotherm for Cu(II) sorption. (c) The linearized Freundlich adsorption isotherm for Cu(II) sorption.

By means of eqn (5), the linearized plot of (C_e/q_e) versus C_e is obtained as shown in Fig. 3(b). Q and b are the two parameters in the Langmuir isotherm and can be computed from the slopes and intercepts by using eqn (5).

The Langmuir constant b is related to the energy of adsorption and Gibbs free energy change (ΔG°) could be evaluated by the following eqn (6):^22,23

ΔG° = −RT ln b (6)

The computed values of Q, b, ΔG° and R² (correlation coefficient) are listed in Table 1.

Table 1 Langmuir isotherm constants for Cu(II) sorption

Adsorbents	Q (mg g^−1)	b (L g^−1)	ΔG° (kJ mol^−1)	R^2
TOCNF0.6	43.1	74.3	−9.3	0.9693
TOCNF1.5	73.0	170.4	−12.7	0.9939

---

The negative values of ΔG° indicate that the sorption of Cu(II) onto oxidized cellulose nanofibers is a spontaneous process. Besides, the thermodynamic potential variation for Cu(II) sorption on the surface of TOCNF1.5 (ΔG = −12.7 kJ mol⁻¹) is more dramatic than the sorption on the surface of TOCNF0.6 (ΔG = −9.3 kJ mol⁻¹), which, from the perspective of sorption energy, interprets the difference in the maximum sorption capacities (Q) of oxidized cellulose nanofibers with different carboxylations.

The Freundlich model describes the sorption on an energetically heterogeneous surface on which the adsorbate molecules are interactive and the adsorbate loading increases infinitely with the enhancement in the concentration. The Freundlich isotherm is given as follows (eqn (7)):²⁴

q_e = K_FC_eⁿ (7)

where K_F and n (dimensionless) are the Freundlich parameters. K_F is the adsorption coefficient characterizing the strength of the adsorption. The higher the value K_F is, the higher adsorbent loading can be achieved. The exponent n is indicative of the energetic heterogeneity. A linear form of the Freundlich expression is expressed as:

ln q_e = ln K_F + n ln C_e (8)

Therefore, K_F and n can be computed from the linear plot of ln q_e versus C_e that is shown in Fig. 3(c).

The calculated values of K_F and n are listed in the Table 2. However, the low correlation coefficients (R² = 0.8339 & 0.8045) indicate the poor correlation between Freundlich isotherm data and the experimental data. The comparison with the models shows that experimental data fits well with Langmuir isotherm with the correlation coefficients higher than 0.96.

Table 2 Freundlich isotherm constants for Cu(II) sorption

Adsorbent	Qf (mg g^−1)	n	R^2
TOCNF0.6	14.7	0.19	0.8045
TOCNF1.5	14.9	0.36	0.8339

---

Surface characterizations of TOCNF after Cu(II) adsorption

Since TOCNF with higher carboxylation degree has better Cu(II) sorption behavior, TOCNF1.5 was selected as the adsorbent for the following studies, including surface characterizations and adsorption mechanisms.

As displayed in Fig. 4(a and b), copper-containing nanoparticles (NPs) were observed clearly on the surface of the oxidized nanofibers after sorption, which complied with the previous studies (see ESI† for the EDS results†). Besides, the NPs distributed on the surface of TOCNF1.5 have a rather narrow size distribution from 200 nm to 300 nm. It was also reported in the previous study that the surface of TOCNF after copper sorption exhibited superhydrophilicity and no contact angle could be detected after 0.2 s, which might be due to the stronger polarity of adsorbed copper ions and copper-containing nanoparticles.¹⁹

Fig. 4 SEM images (a and b) and XRD (c) pattern of TOCNF1.5 filter cake after Cu(II) sorption.

In Fig. 4(c), the most intensive peaks observed at 2θ = 15° and 22.5° are due to the cellulose structure, namely phases Iα and Iβ.²⁵ According to XRD patterns, the copper nanoparticles formed after sorption contain both Cu(II) and Cu(I). The presence of CaCO₃ can be caused by the hydrolysis of CO₂ from air.²⁶ The presence of Cu₂O on the surface of cellulose nanofibers suggests the occurrence of redox reaction during Cu(II) adsorption. Adsorbed Cu(II) ions can be reduced to Cu(I) ions by the aldehyde groups that are formed by oxidation of some of the C6 hydroxyl groups during production of TOCNF.¹³ Although the diffraction peaks (2θ = 43.3° and 50.5°) that represent Cu(0) were not detected, the formation of Cu(0) nanoparticles during adsorption cannot be ruled out, since Cu(0) nanoparticles are easily oxidized under ambient-air conditions.^27,28 Unlike the formations of Au(0), Ag(0) and Pb(0) nanoparticles, oxygen-free environments or chemical stabilizers are normally required to synthesize Cu(0) nanoparticles.²⁶ In accordance with SEM and XRD patterns, a two step-process might lead to the formation of the NPs. Copper ions were first captured by the functional groups on the surface by monolayer sorption (according to Langmuir isotherm) and subsequently self-assembled to copper-containing nanoparticles during drying in air.^29,30

In addition, copper oxides nanoparticles were found with biocidal effects that inhibit the growth of microbes such as bacteria, fungus and algae. Therefore, TOCNF during the process of Cu(II) removal might inhibit the growth of microbes at the same time, which is beneficial to water purification.¹⁶

Correlation between the concentration variations of hydrogen ions (Δ[H⁺]) and Cu(II) ions (Δ[Cu(II)]) in the solution during sorption

The initial pHs of the solutions were adjusted in the range of 5–5.4 to prevent formation of copper hydroxide precipitate. The initial pHs (before sorption) and final pHs after isotherm sorption experiments were plotted in Fig. 5.

Fig. 5 Solution pH variations during sorption at respective Cu(II) equilibrium concentrations (C_e).

Δ[H⁺] during sorption can be computed from the values of pH change and Cu(II) concentration reduction (Δ[Cu(II)]) can be calculated from the data of ICP-OES. Fig. 6(a) shows the Δ[H⁺] and the corresponding Δ[Cu(II)] at increasing Cu(II) equilibrium concentrations (C_e). It is shown in Fig. 5 and 6(a) that the pH of the solution increases (or [H⁺] decrease) during sorption under the condition that C_e is below 5 mg L⁻¹, however, the solution pH declines (or [H⁺] increase) during sorption when C_e is above 5 mg L⁻¹.

Fig. 6 (a) Δ[H⁺] and Δ[Cu(II)] during sorption as a function of Cu(II) equilibrium concentration (adsorbent: TOCNF1.5). (b) Correlation between Δ[Cu(II)] and the corresponding Δ[H⁺] during Cu(II) ion sorption.

This might be caused by the existence of the two states of the functional groups on the adsorbent at the given pH range (5–5.4). Carboxyl groups (or carboxylate) oxidized from C6 primary hydroxyls on the surface of cellulose belong to weak acid, which partially ionize in solutions (eqn (9)).¹³ The pK_a value of carboxyl groups on nanocellulose is considered to be 4.8.³¹ The degree of protolysis (α) of carboxyl dissociation can be calculated by using the eqn (10) at any pH conditions.

Surface|–COOH ⇔ surface|–COO⁻ + H⁺ (9)

It was calculated that, at pH 5, the percentage of the charged species (–COO⁻) of functional groups is 61% and the percentage of neutral species (–COOH) is 39%, if the impact of Cu(II) ions is neglected.

Generally, positively charged Cu(II) ions interact preferentially with the negative charged species of functional groups due to electrostatic attraction and the different steps during the process are schematically shown in Fig. 7. When the concentration of copper ions is relatively low (C_e < 5 mg L⁻¹), Cu(II) tend to interact with the unprotonated carboxylate groups (–COO⁻) on the oxidized cellulose nanofibers as displayed in Fig. 7(a). Due to the reduction of Cu(II) ion amount in the solution, the hydrolysis effect caused by Cu(II) ion weakens and lead to pH enhancement as displayed in the hydrolysis equilibrium (eqn (11)).³² In this case, hydrolysis equilibrium is shifting to the left side, coinciding with lowering the amount of hydrogen ions in the solution.

Cu²⁺ + nH₂O ⇔ Cu(OH)_n²⁻ⁿ + nH⁺ (11)

Fig. 7 Schematics of Cu(II) adsorption mechanism.

However, under the condition of higher Cu(II) ion concentration (C_e > 5 mg L⁻¹) as shown in Fig. 7(b), Cu(II) ions start to interact with the protonated carboxylate groups due to the limited amount of unprotonated carboxylate groups, H⁺ are desorbed from the original protonated functional groups and released into the solution ending up with pH decline. In this case, the amount of hydrogen ions released from functional groups is the primary factor to influence pH change, rather than weakening Cu(II) hydrolysis. Hence, pH reduction occurs with the enhancement of Cu(II) ion sorption capacity (C_e > 5 mg L⁻¹).

It was found that Δ[H⁺] follows a linear correlation (R² = 0.9734) with Δ[Cu(II)] when C_e is in the range of 5 to 172 mg L⁻¹. The linear eqn (12) indicates that hydrogen ions are not exchanged by Cu(II) ions in the reverse ratio of their charge ratio.

According to Fig. 6(b) and eqn (12), around 42 copper ions anchored on to the surface of TOCNF1.5 coincide with one hydrogen ion released into the solution.

Δ[Cu²⁺] = 41.82Δ[H⁺] + 0.3657 (C_e > 5 mg L⁻¹) (12)

However, as shown in Fig. 6(b), when the C_e is lower 5 mg L⁻¹, the correlation of Δ[H⁺] and Δ[Cu(II)] during sorption follow another trend which needs to be clarified accurately with more collected data.

The linear correlation between Δ[Cu(II)] and the corresponding Δ[H⁺] indicates that, under a given pH and temperature condition, sorption capacity might be well predicted and calculated by Δ[H⁺] or pH variation during Cu(II) ion sorption onto TEMPO oxidized nanocellulose fibers. This can be used for development of online sensors for tracking the sorption and removal of Cu(II) using TOCNF and may also be extended to other systems involving metal ions and nanocellulose.

Conclusions

The study revealed that sorption of Cu(II) ions onto TOCNF has fast kinetics (t_e < 1 min) probably contributed by two factors: the nanoscale diameter and hydrophilic property. An increase in temperature leads to a decrease in Cu(II) sorption capacity, which indicates the sorption of Cu(II) onto TOCNF is an exothermic process or enthalpy driven process.

The equilibrium sorption data could be well described by Langmuir isotherms with a high correlation coefficient (R² > 0.99 for TOCNF1.5) and supports a monolayer sorption behavior. The maximum sorption capacities for TOCNF0.6 and TOCNF1.5 are 43 mg g⁻¹ and 73 mg g⁻¹ respectively, which can be interpreted from the higher surface free energy reduction of TOCNF1.5 (ΔG = −12.7 kJ mol⁻¹) during Cu(II) sorption compared to that of TOCNF0.6 (ΔG = −9.3 kJ mol⁻¹). Monolayer adsorbed copper ions on the surface subsequently self-assembled to Cu(I) & Cu(II)-containing NPs (d: 200–300 nm), most probably during the drying step. Cu(II) preferentially interact with charged sites on TOCNF viz. unprotonated –COOH via electrostatic interactions. A linear correlation found between Δ[H⁺] and the corresponding Δ[Cu(II)] during sorption indicates that, at given pH and temperature condition, the sorption capacity might be well predicted and calculated by Δ[H⁺] or pH variation during Cu(II) ion sorption onto TEMPO oxidized nanocellulose fibers. This correlation may be used for developing online sensors for water purification involving metal ions and TOCNF.

Acknowledgements

The authors gratefully acknowledge the financial support of the European Commission, under the NanoSelect Project, EU FP7-NMP4-SL-2012-280519. Houssine Sehaqui, Swiss Federal Laboratories for Materials Science and Technology (EMPA), is acknowledged for supplying TEMPO-modified cellulose nanofibers.

Notes and references

1. J. Wang and C. Chen, Biotechnol. Adv., 2006, 24, 427–451.
2. J. W. Moore and S. Ramamoorthy, Heavy metals in natural waters, Springer-Verlag, New York, USA, 1983.
3. Z. Aksu, Y. Sag and T. Kutsal, Environ. Technol., 1992, 13, 579–586.
4. S. Hokkanen, E. Repo, T. Suopajärvi, H. Liimatainen, J. Niinimaa and M. Sillanpää, Cellulose, 2014, 21, 1471–1487.
5. H. Ma, B. S. Hsiao and B. Chu, ACS Macro Lett., 2012, 1, 213–216.
6. P. Liu, H. Sehaqui, P. Tingaut, A. Wichser, K. Oksman and A. P. Mathew, Cellulose, 2014, 21, 449–461.
7. H. Sehaqui, U. P. de Larraya, P. Liu, N. Pfenninger, A. P. Mathew, T. Zimmermann and P. Tingaut, Cellulose, 2014, 21, 2831–2844.
8. A. Demirbas, J. Hazard. Mater., 2008, 157, 220–229.
9. X. Qu, P. J. J. Alvarez and Q. Li, Water Res., 2013, 47, 3931–3946.
10. E. Worch, Adsorption Technology in Water Treatment: Fundamentals, Processes, and Modeling, De Gruyter, 2012.
11. S. Chen, Y. Zou, Z. Yan, W. Shen, S. Shi, X. Zhang and H. Wang, J. Hazard. Mater., 2009, 161, 1355–1359.
12. A. E. J. de Nooy, A. C. Besemer and H. van Bekkum, Carbohydr. Res., 1995, 269, 89–98.
13. A. Isogai, T. Saito and H. Fukuzumi, Nanoscale, 2011, 3, 71–85.
14. T. Saito and A. Isogai, Carbohydr. Polym., 2005, 61, 183–190.
15. Z. Hu, S. Ballinger, R. Pelton and E. D. Cranston, J. Colloid Interface Sci., 2015, 439, 139–148.
16. N. C. Cady, J. L. Behnke and A. D. Strickland, Adv. Funct. Mater., 2011, 21, 2506–2514.
17. M. Jonoobi, A. P. Mathew and K. Oksman, Ind. Crops Prod., 2012, 40, 232–238.
18. T. Saito, S. Kimura, Y. Nishiyama and A. Isogai, Biomacromolecules, 2007, 8, 2485–2491.
19. P. Liu, K. Oksman and A. P. Mathew, J. Colloid Interface Sci., 2016, 464, 175–182.
20. M. Suzuki and K. Kawazoe, J. Inst. Ind. Sci., Univ. Tokyo, 1974, 26, 275–277.
21. I. Langmuir, J. Am. Chem. Soc., 1918, 40, 1361–1403.
22. C. Septhum, S. Rattanaphani, J. B. Bremner and V. Rattanaphani, J. Hazard. Mater., 2007, 148, 185–191.
23. K. K. Panday, G. Prasad and V. N. Singh, Water Res., 1985, 19, 869–873.
24. H. Freundlich, 1906, 1906, 57, 385–470.
25. H. S. Barud, C. A. Ribeiro, M. S. Crespi, M. A. U. Martines, J. Dexpert-Ghys, R. F. C. Marques, Y. Messaddeq and S. J. L. Ribeiro, J. Therm. Anal. Calorim., 2007, 87, 815–818.
26. R. J. B. Pinto, M. C. Neves, C. P. Neto and T. Trindade, Eur. J. Inorg. Chem., 2012, 5043–5049.
27. P. Kanninen, C. Johans, J. Merta and K. Kontturi, J. Colloid Interface Sci., 2008, 318, 88–95.
28. H. Zhu, C. Zhang and Y. Yin, Nanotechnology, 2005, 16, 3079.
29. J. B. P. Ricardo, N. C. Márcia, P. N. Carlos and T. Tito, in Nanocomposites - New Trends and Developments, ed. Farzad Ebrahimi, InTec, 2012, p. 77.
30. J. Wang and C. Chen, Biotechnol. Adv., 2009, 27, 195–226.
31. L. Wågberg, G. Decher, M. Norgren, T. Lindström, M. Ankerfors and K. Axnäs, Langmuir, 2008, 24, 784–795.
32. A. J. Paulson and D. R. Kester, J. Solution Chem., 1980, 9, 269–277.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra22397d

---