3.3.2. Evaluation of Adsorption Kinetics

The kinetics investigation is important to choose the optimal operating condition on practical systems for HMIs removal [63]. To ge<sup>t</sup> a deeper understanding of the adsorption process of Pb(II) on SF@Cu-NFs, three typical kinetic models including pseudo-first-order (Equation (7)), pseudo-second-order (Equation (8)) and intraparticle diffusion (Equation (9)) models were used to analyze the experimental data

$$\ln\left(Q\_{\mathcal{E}} - Q\_{\mathcal{I}}\right) = \ln Q\_{\mathcal{E}} - k\_1 t \tag{7}$$

$$\frac{t}{Q\_t} = \frac{1}{k\_2 Q\_\epsilon^2} + \frac{t}{Q\_\epsilon} \tag{8}$$

$$Q\_t = C + k\_n t^{0.5} \tag{9}$$

where *t* (min) is the adsorption time; *Q*e and *Q*t (mg g<sup>−</sup>1) are the Pb(II) amount adsorbed at equilibrium; and *k*1 (min−1), *k*2 (g mg<sup>−</sup><sup>1</sup> min−1), and *k*n (mg g<sup>−</sup><sup>1</sup> min−1/2) are the rate constants of pseudo-first-order, pseudo-second-order, and intraparticle kinetics models, respectively. The adsorption capacity at different time was indicated in Figure 7a and the kinetic experimental data investigated by the three kinetic models were shown in Figure 7b–d, respectively. The fitting equations and kinetic parameters for the adsorption of Pb(II) by the prepared SF@Cu-NFs, as calculated from the plots of above three models, were listed in Table 1.

**Figure 7.** (**a**) The adsorption capacity at different times. Pb(II) removal by silk fibroin mediated nanoflowers (**b**) the pseudo-first-order; (**c**) the pseudo-second-order; (**d**) the intra-particle diffusion model sorption kinetics curves.

Pseudo-first-order (shown in Figure 7b) and pseudo-second-order (shown in Figure 7c) models were generally used to predict equilibrium adsorption capacity. With high correlation coefficient values of *R*22 =0.99 > *R*21 =0.98, the results indicated both the pseudo-second-order model provided a better fitting effect on the experimental data, demonstrating the calculated value of the pseudo-second-order model was closer to the actual value than that of pseudo-first-order model. As a result, the adsorption capacity of 300 mg L−<sup>1</sup> Pb(II) was then estimated to be 2528.36 mg g<sup>−</sup><sup>1</sup> by the prepared SF@Cu-NFs. As shown in Table 2, the relative error between the calculated capacity of 2528.36 mg g<sup>−</sup><sup>1</sup> and experiment result of 2407.00 mg g<sup>−</sup><sup>1</sup> was evaluated to be 4.8%, which indicating the good agreemen<sup>t</sup> for them.


**Table 2.** Comparison parameters of the pseudo-first-order, the pseudo-second-order and the intra-particle diffusion models for Pb(II) adsorption by SF@Cu-NFs.

The intraparticle diffusion model was usually employed to examine the controlling mechanism such as transfer and chemical reaction for the adsorption process. As shown in Figure 7d, the fitting curve of intraparticle diffusion can be divided into two linear parts, indicating that the adsorption process consists of two steps. The first stage belongs to boundary layer adsorption, which is the diffusion of Pb(II) adsorbate from solution to SF@Cu-NFs surface. In the second stage, Pb(II) ion passes through the boundary layer to further react inside the SF@Cu-NFs adsorbent, which belongs to the intraparticle diffusion. Because the straight lines of the two stages do not pass through the origin of coordinate axis, it shows that the adsorption process is controlled by both intraparticle diffusion and boundary layer diffusion.

### 3.3.3. Adsorption Isotherm Experiment and Adsorption Thermodynamics

At the above optimal condition of pH = 5, the adsorption thermodynamics were further investigated by changing the initial Pb(II) concentrations with 1 × 10−<sup>3</sup> g mL−<sup>1</sup> SF@Cu-NFs adsorbent. The adsorption performances of Pb(II) on SF@Cu-NFs were studied by the following Langmuir (Equation (10)), Freundlich (Equation (11)) and Temkin (Equation (12)) models.

$$\frac{C\_{\text{e}}}{Q\_{\text{e}}} = \frac{C\_{\text{e}}}{Q\_{\text{max}}} + \frac{1}{K\_{\text{L}}Q\_{\text{max}}} \tag{10}$$

$$
\ln Q\_{\rm e} = \ln K\_{\rm F} + \frac{1}{n} \ln \mathcal{C}\_{\rm e} \tag{11}
$$

$$Q\_{\varepsilon} = A \ln \mathcal{C}\_{\varepsilon} + B \tag{12}$$

where *K*L, *K*F, and A are the equilibrium constants of Langmuir, Freundlich and Temkin adsorption, respectively; *C*e is the equilibrium concentration of Pb(II); *Q*e and *Q*max are the amount of equilibrium

adsorption capacity and the maximum adsorption capacity of Pb(II), respectively; The value of *n* > 1 suggests a normal Langmuir isotherm and *n* < 1 suggests the cooperative adsorption, respectively [64].

The adsorption of Pb(II) was investigated with different initial concentrations of 5–500 mg L−<sup>1</sup> at different temperatures of 298 K. Meanwhile, due to the high adsorption ability of SF@Cu-NFs, almost all the Pb(II) in the solution were adsorbed completely in the low concentration of 5–50 mg L−1. The free concentration of Pb(II) in the final solution could not be effectively detected and the equilibrium concentration of Pb(II) was thereafter regarded to be 0 during this concentration stage. As a result, the Langmuir, Freundlich, and Temkin adsorption isotherms of Pb(II) adsorption were efficiently fitted in high Pb(II) concentration of 80–500 mg L−<sup>1</sup> and shown in Figure S4a–c, respectively.

For clarity, the fitting equations and parameters of Langmuir model, Freundlich model and Temkin model for Pb(II) by SF@Cu-NFs were summarized and listed in Table 3. It is found that Langmuir model provided better fitting to the equilibrium data than that of Freundlich and Temkin models with a higher correlation coefficient of 0.98, indicating that the adsorption of Pb(II) on the prepared SF@Cu-NFs belonged to monolayer adsorption instead of multilayer adsorption. Since the Langmuir model suggested that molecules are adsorbed uniformly, it can be deduced that the prepared SF@Cu-NFs were fairly homogeneous with SF protein assembly.

**Table 3.** Comparison parameters of Langmuir, Freundlich and Temkin models for Pb(II) adsorption by SF@Cu-NFs.


In order to evaluate the treat ability of the prepared SF@Cu-NFs for Pb(II) adsorption, the maximum adsorption capacity in this work was compared the results obtained by some other adsorbents which were reported previously. The results were listed in Table S2. As compared, SF@Cu-NFs indicated as an excellent adsorbent for Pb(II) treatment with the *Qmax* as high as 1908 mg g<sup>−</sup>1, which was about 3–20 folds than that of the other adsorbents. As a result, SF@Cu-NFs was suggested to be a candidate for Pb(II) removal in wastewater with much higher adsorption performance.

In order to obtain the experimental parameters of adsorption thermodynamics on the Pb(II) adsorption by SF@Cu-NFs, the adsorption capacity and equilibrium constant at different temperatures were shown in Figure 8a,b, respectively. Then the thermodynamic data were calculated assuming the temperature-constant entropy and enthalpy of adsorption and according to the following temperature-related equations of Equation (13) to Equation (15).

$$
\Delta\_{\rm I} G\_{\rm m(T)}^{\rm 0} = -RT \ln K\_{\rm T}^{\rm 0} \tag{13}
$$

$$
\Delta\_r G\_{\text{m(T)}}^{\text{0}} = \Delta\_r H\_{\text{m}}^{\text{0}} - T \Delta\_r S\_{\text{m}}^{\text{0}} \tag{14}
$$

$$K\_\Gamma^0 = \frac{Q\_\varepsilon}{C\_\varepsilon} \tag{15}$$

**Figure 8.** Effect of temperature on the equilibrium adsorption of Pb(II) (**a**) adsorption capacity and (**b**) adsorption constant.

The results of <sup>Δ</sup>*rG*θm(T), <sup>Δ</sup>*rH*θm and <sup>Δ</sup>*rS*θm were listed in Table S3 and several conclusions could be obtained.

First, with <sup>Δ</sup>*rG*θm(T) < 0 for both 100 and 500 mg L−<sup>1</sup> Pb(II), the adsorption process of Pb(II) onto SF@Cu-NFs was indicated to be spontaneous. However, the absolute value of <sup>Δ</sup>*rG*θm(T) for Pb(II) adsorption was noted to decrease from 9.07 kJ mol−<sup>1</sup> to 3.63 kJ mol−<sup>1</sup> with the increase in temperature from 298 K to 328 K, which indicating that lower temperature was favored for the removal of Pb(II) by SF@Cu-NFs.

Second, with <sup>Δ</sup>*rH*θm < 0 for both 100 and 500 mg L−<sup>1</sup> Pb(II), the adsorption process of Pb(II) onto SF@Cu-NFs was indicated to be exothermic. However, the absolute value of <sup>Δ</sup>*rH*θm for Pb(II) adsorption was noted to decrease from 64.30 kJ mol−<sup>1</sup> to 17.91 kJ mol−<sup>1</sup> with the increase in Pb(II) concentration from 100 mg L−<sup>1</sup> to 500 mg <sup>L</sup>−1, which indicating that higher concentration would result in a mutual repulsion between mutual Pb(II).

Third, with the absolute value of Δ*G* < 40 kJ mol−<sup>1</sup> at different Pb(II) concentrations and different temperatures, the observations on the adsorption of Pb(II) by the SF@Cu-NFs in present study was an obvious physical adsorption process.

### 3.3.4. Investigation of Adsorption Selectivity for Pb(II)

Selectivity is one of the primary criteria for good adsorbents for the removal of trace amounts of heavy metals in the presence of other competing metal ions. In this work, the adsorption selectivity of SF@Cu-NFs was studied for three HMIs of Pb(II), Cd(II), and Ni(II) under the condition of 20 mL 100 mg L−<sup>1</sup> HMIs with 3 mg SF@Cu-NFs adsorbent. The removal efficiency for different HMIs at different concentration were shown in Figure S5. For the three HMIs, all the adsorption processes

were indicated to be rapid within the first 5 min and thereafter relatively slower by achieving the equilibrium in 90 min. The heavy metal ions adsorption efficiency (*AE*) can be calculated as

$$AE(\%) = \frac{(C\_0 - C\_c)}{C\_0} \times 100\% \tag{16}$$

The adsorption efficiencies of Cd(II), Ni(II) and Pb(II) were calculated to be 23.77%, 18.76%, and 99.75%, indicating the much higher adsorption performance for Pb(II) by the prepared nanoflower. The selective factor (sf) was defined to evaluate the adsorbent selectivity as

$$sf = \frac{AE\_a}{AE\_b} \tag{17}$$

where *AEa* and *AEb* were adsorption efficiencies for the superior and inferior adsorption HMIs, respectively. For the prepared SF@Cu-NFs, its selective factors of Pb(II) were calculated to be 4.2 relative to Cd(II) and 5.3 relative to Ni(II), which proved the excellent adsorption selectivity for Pb(II) by SF@Cu-NFs.

### *3.4. SF@Cu-NFs Adsorption Mechanism for Pb(II)*

### 3.4.1. Verification of Pb(II) Adsorption by SF@Cu-NFs

In order to access the interactions between SF@Cu-NFs adsorbent and Pb(II) ion, SF@Cu-NFs after Pb(II) adsorption was furthermore investigated by zeta potential, FTIR, and XRD measurements, which were respectively shown in Figure 9a,b and Figure S6.

**Figure 9.** (**a**) Surface zeta potential measurement of SF@Cu-NFs at different Pb2+ concentrations; (**b**) FTIR spectra of SF@Cu-NFs before (spectrum i) and after (spectrum ii) Pb(II) adsorption.

The average zeta potentials of SF@Cu-NFs adsorbent were analyzed through DLS measurements. With the addition of di fferent Pb(II) concentrations, the surface zeta potential measurement of SF@Cu-NFs was presented in Figure 9a. At the optimum pH = 5, the surface of the SF@Cu-NFs adsorbent was negatively charged and had an average zeta potential of about −12 mV without Pb(II). With the addition of Pb(II), an increase in the zeta potential was produced, which showed fast in lower Pb(II) concentration (indicated as the blue area) and thereafter varies rather slowly in higher Pb(II)concentration (indicated as the green area). This is a typical two-site adsorption behavior corresponding to two-type interaction dominance [65].

The pattern of HMI adsorption onto solid adsorbents can be attributable to the groups and bonds present on the material surface. In order to elucidate the active interaction site, FTIR spectrophotometry was performed to investigate the changes of functional groups of SF@Cu-NFs adsorbent before and after Pb(II) adsorption, which were shown as spectrum (i) and spectrum (ii) in Figure 9b, respectively. As it can be seen, the two FITR domains of SF@Cu-NFs after Pb(II) adsorption showed di fferent groups and bonds, including (1) decreasing peaks at 1638, 1536, and 1421 cm–1 at SF domain, indicating the interaction between Pb(II) and functional N–H, C–O, and C–N groups; (2) increasing peaks at 1035, 603, and 563 cm–1 at Cu domain, indicating the interaction between Pb(II) and P-O groups. Because there was no new functional group appearing in the SF@Cu-NFs adsorbent after Pb(II) adsorption, it can be determined that the interaction between Pb(II) and SF@Cu-NFs belongs to physical but not chemical adsorption. The two bands around 2800 cm<sup>−</sup><sup>1</sup> are corresponding to stretching vibration of saturated C-H. Generally, the group with high electronegativity has strong ability of electron absorption. When it is connected with the number of carbon atoms on the carbonyl group of alkyl ketone, the electron cloud will shift from oxygen atom to the middle of double bond due to the induction e ffect. These increase the force constant of C=O bond, increases the vibration frequency of C=O, and shifts the absorption peak to a higher wave number. This result is also consistent to the adsorption energy obtained in thermodynamic investigation, which was calculated to be Δ G < 40 kJ mol−1.

As presented in Figure S6, XRD patterns of SF@Cu-NFs were measured before and after Pb(II) uptake. Compared to the di ffraction peaks of hybrid nanoflowers before adsorbing Pb(II), there were new several miscellaneous di ffraction peaks at 2θ values of 21.5, 26.2, 27.5, 30.0, which confirmed hybrid nanoflowers successfully adsorbed heavy metal ion Pb(II).

### 3.4.2. Mechanism Analysis of Pb(II) Adsorption by SF@Cu-NFs

Based on the results of adsorption property and adsorption characterization mentioned above, the mechanism is proposed to illustrate the elimination performance for Pb(II) by the prepared SF@Cu-NFs. The mechanism diagram is schematically presented in Figure 10, in which the chemical structure of SF protein was referenced from the previous report [66] and the electronic structure of Cu3(PO4)2·3H2O was calculated by Material Studio 7.0. Blue, yellow, red, and grey spheres designate Cu, P, O, and H atoms, respectively. The adsorption of SF@Cu-NFs for Pb(II) removal was originated from two types of adsorption sites and two kinds of interaction dominance, which can be ascribed to the individual organic SF protein and inorganic Cu3(PO4)2 crystal. Correspondingly, two stages of fast adsorption and slow adsorption of Pb(II) by the prepared SF@Cu-NFs was revealed and described as follows.

Fast adsorption stage of Pb(II). For this stage, the flower 'stamen' of organic SF protein was designated as responsible adsorption site for fast adsorption of Pb(II) (shown as upper part in Figure 10). This kind adsorption was originated from multiple coordinative interaction produced between Pb(II) and abundant N, O elements. This interaction showed strong due to the numerous amide groups provided by SF protein. Meanwhile, the fast adsorption occurred in the shorter adsorption time (shown as the first linear part by intraparticle kinetic investigation in Figure 7c) and in the lower adsorbent concentration (shown as the first increasing part by zeta potential measurement in Figure 9a).

**Figure 10.** Proposed adsorption mechanism of Pb(II) by SF@Cu-NFs. Upper part indicates the fast adsorption of Pb(II) by the organic SF component and lower part indicates the slow adsorption of Pb(II) by the inorganic Cu3(PO4)2 component.

Slow adsorption stage of Pb(II). For this stage, the flower 'petal' of inorganic Cu3(PO4)2 crystal was designated as responsible adsorption site for slow adsorption of Pb(II) (shown as lower part in Figure 10). This kind adsorption was originated from unique coordinative interaction produced between Pb(II) and O element. This interaction showed weak due to the powerful restriction from the strong ion bond by Cu(II) elements in Cu3(PO4)2 crystal. Meanwhile, the slow adsorption occurred in the longer adsorption time (shown as the second linear part by intraparticle kinetic investigation in Figure 7c) and in the higher adsorbent concentration (shown as the second increasing part by zeta potential measurement in Figure 9a).
