*2.4. Adsorption Tests*

All adsorption experiments were performed in an ARGOLAB SKI 4 orbital shaker at 260 rpm and 20 ◦C. The AC was tested at a solid/liquid ratio equal to 0.75 g/L. All tests were conducted in three replicates.

Firstly, equilibrium tests were necessary to find the equilibrium time (teq) for each target pollutant and the pre-treated waste PUF. Flasks of 250 mL were filled with 200 mL of 10 mg/L solution of each pollutant and 5 g of PUF (solid/liquid ratio equal to 25 g/L, chosen according to literature studies in Table 1). Three milliliter aliquots of solution were withdrawn after different time intervals, filtered on 0.45 μm cellulose ester syringe filters, and analyzed to measure the residual pollutant concentration. teq was determined as the time after which no decrease in the residual aqueous concentration was detected. qeq, i.e., the amount of pollutant adsorbed, was calculated as the difference between the initial concentration of pollutant in the liquid phase (CLi) and the residual value (CLf).

The adsorption tests were performed in 50 mL falcon test tubes filled with 40 mL of pollutants solution and 1 g of pre-treated waste PUF (solid/liquid ratio equal to 25 g/L). The pollutant solutions were as follows: methylene blue: 0.5, 1, 2, 5, 7.5, 10, 15, 18, 20 mg/L; phenol: 6, 8, 10, 12, 14, 16, 19, 24, 30 mg/L; mercury: 2, 3, 4, 5, 6.5, 10, 12, 17, 22 mg/L. The tubes were shaken for an interval equal to the teq of each pollutant. The supernatant was separated from the solid phase through a Z20A Hermle centrifuge (Labortechnik GmbH, Wehingen, Germany) at 3500 rpm for 5 min, then filtered on 0.45 μm cellulose ester syringe filters and analyzed. The adsorption tests involved three replicates.

#### *2.5. Isotherm Models*

At a constant temperature, the process of adsorption can be described by an adsorption isotherm. After the equilibrium state has been reached, the concentrations of the adsorbate on the solid phase are plotted against concentrations of adsorbate in liquid phase. Two models were used for the interpretation of experimental data. The Freundlich model is based on Equation (1) [14]:

$$\mathbf{q\_{eq}} = \mathbf{K\_f} \left( \mathbf{C\_{eq}} \right)^{1/n} \tag{1}$$

where qeq is the amount of adsorbate transferred on the sorbent at equilibrium; Kf is the capacity factor, a parameter that characterizes the strength of adsorption, and it is directly proportional to qeq. The exponent 1/n determines the curvature of the isotherm, and it denotes the intensity of adsorption.

The Langmuir model is based on Equation (2) [14]:

$$\mathbf{q}\_{\rm eq} = \frac{\mathbf{q}\_{\rm max} \, b}{1 + b \mathbf{C}\_{\rm e}} \cdot \mathbf{C}\_{\rm e} \tag{2}$$

where qeq is the amount of adsorbate transferred on the sorbent at equilibrium; qmax is the maximum capacity of adsorption at saturation (assuming the formation of a single layer of adsorbed molecules); *b* is the Langmuir constant related to the adsorption energy.

#### **3. Results and Discussion**

#### *3.1. Adsorption Equilibrium Tests*

Considering the results of the equilibrium tests (Figure 1 and Table 3), for all target contaminants, the equilibrium of adsorption was reached more quickly with AC, so that the test was stopped earlier than for reactors with PUF, since no significant changes in liquid concentration were detectable. Compared to PUF, the much shorter teq found for AC is reasonably a consequence of its high specific surface area [14] and of the hydrophobic nature of polyurethane, which could make adsorption slower [32]. The pollutants reached the adsorption equilibrium on AC rather quickly (30–35 min), while waste PUF required much longer times: 60 min for methylene blue, and 135–140 min for phenol and mercury. From these preliminary tests and considering the amounts of pollutant transferred on the solid adsorbent (qeq), methylene blue exhibited the highest affinity, compared to phenol and mercury, both for waste PUF and AC (Table 3).


**Table 3.** Details and results of the adsorption equilibrium tests performed on PUF and AC (CLi: initial concentration in the liquid phase; CLf: final concentration in the liquid phase; teq: equilibrium time; qeq: amount of contaminant transferred on the sorbent).

### *3.2. Adsorption Batch Tests*

The results of the adsorption batch tests (Figure 2 and Table 4) showed that the Freundlich isotherm model better fitted, compared to Langmuir model, the data related to waste PUF with an adequate correction factor (R<sup>2</sup> = 0.93) for all the three pollutants. The adsorption capacity of waste PUF was moderate for methylene blue and mercury (Kf values around 0.02), while it was considerably lower for phenol (Kf around <sup>1</sup> × <sup>10</sup><sup>−</sup>3). Indeed, the maximum removal efficiency achieved from the batch tests by waste

PUF (Table 5) was also rather limited: 12.2% for phenol and 37–38% for methylene blue and mercury.

**Figure 1.** Results of the adsorption equilibrium tests performed on waste PUF and AC with (**a**) methylene blue, (**b**) phenol, and (**c**) mercury.

**Figure 2.** Results of the adsorption batch tests performed with waste PUF and AC in contact with (**a**) methylene blue, (**b**) phenol, and (**c**) mercury (Ce: equilibrium concentration in the liquid phase; qe: equilibrium concentration in the solid phase).

The results of the adsorption tests performed on AC were described with higher accuracy by the Langmuir model for methylene blue (R2 = 0.99) and mercury (R<sup>2</sup> = 0.95). Only in the case of phenol was the Freundlich model more adequate in describing the adsorption by AC (R<sup>2</sup> = 0.88) than the Langmuir model (R<sup>2</sup> = 0.59). The maximum removal efficiency achieved by AC for methylene blue was 99.9%, leading to very low residual concentrations in the liquid phase (CLf = 0.04 mg/L). The Freundlich model had inadequate experimental results obtained for methylene blue (R<sup>2</sup> = 0.14), probably because when the concentrations at equilibrium are much lower than the initial concentrations, the adsorption is generally well described by a linear model. The Freundlich isotherm, which is in exponential form, cannot describe the linear range at very low concentrations. On the contrary, this limit case is well described by Langmuir model and when *b* · *CLf << 1*, it is equivalent to a linear isotherm. The higher qmax found for AC applied to the adsorption of methylene blue (135.13 mg/g), compared to qmax of phenol (26.11 mg/g) and mercury (0.05 mg/g), was realistically expected since the considered commercial AC is commonly applied for decolorization purposes.


**Table 4.** Values of Freundlich (Kf, n) and Langmuir (qmax, b) isotherm models' parameters resulting from the interpolation of the experimental data derived from batch adsorption tests with waste PUF and AC.

**Table 5.** Maximum removal efficiencies achieved in batch adsorption tests performed with waste PUF and AC (CLi: initial concentration in the liquid phase; CLf: final concentration in the liquid phase).


Unfortunately, because of the different level of correction factors, a direct comparison of the two adsorbents was not possible. However, since the differences between the values of Kf and qmax obtained from waste PUF and AC were of several orders of magnitude almost in every case, it was evident that there was a considerable gap in favor of AC towards the adsorption of the considered target pollutants.

The results of this study were compared to literature data related to other novel and low-cost "non-conventional" (i.e., not commercial) materials tested for the adsorption of mercury (Table 6), phenol (Table 7) and methylene blue (Table 8). These materials, although at an experimental level, all underwent treatments aimed at improving their adsorption performances (e.g., activation for biomass-based sorbents, modification by addition of reagents for other materials). Literature data referred to mercury adsorption (Table 6) exhibited qmax in the range of 1.8–13 mg/g from the Langmuir model, and Kf between 0.02 and 19 L/mg from the Freundlich model, with correction factor values exceeding 0.9 for both isotherm models in all studies. Literature data on phenol adsorption (Table 7) found qmax values in the range of 38–285 mg/g from the Langmuir model, and Kf between 0.19 and 7.40 L/mg, with correction factor values exceeding 0.9 for both isotherm models in all studies. Methylene blue adsorption literature studies (Table 8) found typical values of qmax in the range of 29–2639 mg/g for the Langmuir model, and Kf between 0.82 and 1746 L/mg, with correction factor values around 0.8–0.9 for both isotherm models in all studies.

The fact that waste PUF did not show similar adsorption performances in the present study means that the tested material was not yet ready to provide competitive adsorption performances. Indeed, the gap was not so large when comparing the Freundlich parameters obtained from waste PUF (Kf = 0.019 L/mg) and other non-commercial adsorbents in contact with mercury solutions (Kf mostly in the range 0.02–4.50 L/mg, with one exception).


**Table 6.** Performances of some non-commercial adsorbents tested for the removal of mercury.

**Table 7.** Performances of some non-commercial adsorbents tested for the removal of phenol.


**Table 8.** Performances of some non-commercial adsorbents tested for the removal of methylene blue.


#### **4. Conclusions**

Investigating any possible opportunities for the recovery of plastics is a key step for supporting the European Circular Economy strategies. This research provides preliminary results about the adsorption properties of waste PUF deriving from the shredding of EoL refrigerators. In this study, waste PUF performances for the removal of methylene blue, phenol, and mercury from aqueous phases were compared with the ones of a commercial AC. Adsorption batch tests allowed to determine the adsorption isotherm parameters. The Freundlich isotherm model better fitted (R<sup>2</sup> = 0.93), compared to the Langmuir model (R<sup>2</sup> < 0.60), the adsorption of methylene blue, phenol, and mercury on waste PUF. In the considered experimental conditions, waste PUF showed a constrained affinity in adsorbing the target pollutants. The obtained Freundlich adsorption parameter Kf was around 0.02 L/mg for mercury and methylene blue, and 0.001 L/mg for phenol. These values were three or four orders of magnitude lower compared to commercial AC, and rather low when compared to the average adsorption capacities of non-commercial adsorbents according to the literature. Moreover, the long time required to reach the adsorption equilibrium (60–140 min depending on the pollutant) in the considered experimental conditions makes waste PUF direct application as an adsorbent rather challenging, especially in fixed-bed columns wherein short equilibrium times are desirable to design columns of reasonable height.

However, summarizing the results obtained in this study, it must be considered that waste PUF is a material deriving from a waste treatment process totally unintended for any adsorption application, and with a minimal preparation consisting only of sieving and washing. The results of this study can support the design of other pre-treatments aimed at overcoming the adsorption limits of the waste PUF "as such". For instance, reducing the particle size of waste PUF, and thus increasing the available specific surface area, would benefit the rate of adsorption. After these additional studies, waste PUF could be applied for "rough-cut" wastewater treatment. When industrial wastewater with high pollution loads is delivered to treatment plants, a rough removal of contamination can be conducted with a relatively low-performant adsorbent such as PUF, prior to a second- more advanced purification process. Additionally, considering the comparison with the performances of other non-conventional (i.e., non-commercial) adsorbents, PUF exhibited the most promising affinity towards mercury. Therefore, further research could be conducted aiming at a feasible application of PUF for mercury removal.

**Author Contributions:** Conceptualization, methodology, experimental investigation, data curation, and writing—original draft preparation: V.S.; conceptualization, methodology, supervision, writing review and editing, project administration, and funding acquisition: S.F. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was performed in the framework of the project "Material recovery from WEEE" (in Italian: "Recupero di materia da RAEE/R1-R2") funded by the Italian Ministry for Environmental Transition, and ongoing between January 2019 and August 2021. The project was coordinated by Politecnico di Torino (Polytechnic University of Turin) and involved as industrial partners, among others, IREN Group and AMIAT. Specifically, this study was based on the activities of work package 1 of the project, dedicated to the recovery of waste PUF from the shredding of end-of-life refrigerators.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The data presented in this study are available on reasonable request from the corresponding author.

**Conflicts of Interest:** The authors declare no conflict of interest.
