A Recycling-Oriented Approach to Rare Earth Element Recovery Using Low-Cost Agricultural Waste

Abstract

The exponential increase in electronic waste (e-waste) from end-of-life electrical and electronic equipment presents a growing environmental challenge. E-waste contains high concentrations of rare earth elements (REEs), which are classified as critical raw materials (CRMs). Their removal and recovery from contaminated systems not only mitigate pollution but also support resource sustainability within a circular economy framework. The present study proposed the use of hazelnut shells as a biosorbent to reduce water contamination and recover REEs. The sorption capabilities of this lignocellulosic material were assessed and optimized using the response surface methodology (RSM) combined with a Box–Behnken Design (three factors, three levels). Factors such as pH (4 to 8), salinity (0 to 30), and biosorbent dose (0.25 to 0.75 g/L) were evaluated in a complex mixture containing 9 REEs (Y, La, Ce, Pr, Nd, Eu, Gd, Tb and Dy; equimolar concentration of 1 µmol/L). Salinity was found to be the factor with greater significance for REEs sorption efficiency, followed by water pH and biosorbent dose. At a pH of 7, salinity of 0, biosorbent dose of 0.75 g/L, and a contact time of 48 h, optimal conditions were observed, achieving removals of 100% for Gd and Eu and between 81 and 99% for other REEs. Optimized conditions were also predicted to maximize the REEs concentration in the biosorbent, which allowed us to obtain values (total REEs content of 2.69 mg/g) higher than those in some ores. These results underscore the high potential of this agricultural waste with no relevant commercial value to improve water quality while providing an alternative source of elements of interest for reuse (circular economy).

1. Introduction

The United Nations recognizes water as the core of sustainable development, given its essential role in human survival, food, and energy production. However, only 1% of the earth’s water is available for human consumption and use, which is not equitably distributed and causes socioeconomic problems [1,2]. More than 2 billion people suffer from severe water scarcity and lack access to safely managed drinking water supplies [3]. In addition, it is estimated that over 80% of the world’s wastewater enters waterways untreated [4]. As a universal solvent, water is easily contaminated with toxic substances, either inorganic, organic, or other contaminants [5,6].

Agriculture, mining, and the transformation industry are common anthropogenic sources of water contaminants. Recently, the high obsolescence rate in the technological sector has generated one of the fastest growing waste streams, e-waste [7,8].

E-waste composition is very diverse and differs among products across several categories, but it is known to contain more than 1000 different substances [9,10], making e-waste management a priority. Although new models for the disposal of this waste are being developed, e-waste is a rich source of elements such as gold, silver, copper, lithium, cobalt, platinum group elements, and rare earth elements (REEs), which can be recovered and brought back into the production cycle [8].

The European Commission [11,12] classify REEs as critical raw materials (CRMs) because these elements have a high supply risk (>95% of the global resources have been provided solely by China), and their demand is growing due to their use in electronic devices, industry, and alternative energy production. Rare earth elements represent a group of 17 elements comprising the lanthanide series, plus yttrium (Y) and scandium (Sc). The REEs group can be considered a multi-applicable group due to their diverse chemical, electrical, metallurgical, magnetic, optical, and catalytic properties [13,14], being the so-called “seed of technology” [15].

To date, mine production is the primary source of REEs, which are the main cause of serious environmental problems [16]. The recovery and recycling of REEs from e-waste and other industrial wastewaters [17,18] can respond to supply risk with a cheaper and unexplored supply source while reducing environmental problems and slowing the depletion of natural resources [14]. Currently, only 1% of REEs are recycled [11], mainly due to lack of incentives [19]. Separation and pre-concentration of REEs have been previously evaluated using multiple methods such as liquid–liquid [20] and solid–liquid extraction [21], precipitation [22], ion exchange [23], and biosorption [24].

Biosorption is a promising biotechnological approach used to extract and recover REEs from aquatic systems. This method has been described as efficient and inexpensive, with several biosorbents being tested, namely, fungi, bacteria [25], plant and animal residues [26,27], and agricultural residues [28]. Among biosorbents, nutshells have many advantages over other fruit, barks, and peels, since they are non-perishable, high in polysaccharides, have large porosity, have no commercial value, and are widely available [29]. Over 5.36 million metric tons of tree nuts were produced worldwide in the period 2021/2022, and by 2023/2024, global production of tree nuts had increased slightly, reaching about 5.69 million metric tons [30]. Nut processing generates a large amount of by-products, with the shells representing the greater portion of the nut (almost 67%) [31]. These characteristics have created an interest in using nutshells as a cost-effective sorbent for the removal of elements from contaminated waters. Dias et al. (2021) [32] demonstrated the high efficiency of hazelnut shells in removing heavy metals like mercury and lead from mineral water.

The present study aimed to evaluate the removal of REEs from complex mixtures and different relevant scenarios using hazelnut shells and to optimize the removal conditions. The response surface methodology (RSM), aligned with the Box–Behnken design, was used to assess the significance of three factors that are pivotal for biosorption (pH, salinity, and biosorbent dose) in three different levels (pH 4 to pH 8; salinity 0 to salinity 30; and 0.25 to 0.75 g/L). The objective was to promote the application of nutshells (valorization of this waste) as a sustainable water remediation methodology and, simultaneously, as a secondary source of CRMs for the technological industry.

2. Materials and Methods

2.1. Chemicals

All chemicals used in the experiment were of analytical reagent grade and used without further purification. The certified standard solutions of the rare earth elements (yttrium (Y), lanthanum (La), cerium (Ce), praseodymium (Pr), neodymium (Nd), europium (Eu), gadolinium (Gd), terbium (Tb), and dysprosium (Dy)) (1000 mg/L, in HNO₃ 1.4–7%) were acquired from Inorganic Ventures (Christiansburg, VA, USA), Alfa Aesar (Haverhill, MA, USA), PlasmaCAL (SCP science, Quebec, QC, Canada), and Sigma Aldrich (St. Louis, MO, USA). Nitric acid (HNO₃ 65%) and sodium hydroxide (NaOH) (>99%) were purchased from Merck. Prior to use, all glassware was acid-washed with HNO₃ 25% for 24 h and then rinsed with ultrapure water.

2.2. Biosorbent Preparation

Hazelnut shells used as sorbent in the experiments were mechanically grounded and sieved to obtain particles with sizes between 1 and 2 mm. The material was washed abundantly with tap water at a temperature of 373.15 K and then dried in a muffle at 308.15 K for 48 h.

2.3. Biosorbent Characterization

Characterization of the hazelnut shells was performed before and after contact with the metal solution. Fourier transform infrared spectroscopy-attenuated total reflectance (FTIR-ATR) was performed using a Bruker Optic Tensor 27 Attenuated Total Reflectance spectrometer (Billerica, MA, USA). The spectra were recorded with a resolution of 4 cm⁻¹ and 256 scans. Samples were analyzed directly after baseline correction, and spectra were recorded between 600 and 4000 cm⁻¹. Scanning electron microscopy (SEM) in a Hitachi TM4000 Plus voltage of 15 kV (Hitachi, Tokyo, Japan) was used to evaluate the microstructure. The sample preparation was performed by glueing the hazelnut shells powder onto a carbon-taped aluminum sample holder.

2.4. Design of the Sorption Experiments

The ability of hazelnut shells to remove REEs from water was evaluated by contacting 0.25, 0.50, and 0.75 g/L of the biosorbent in an aqueous solution containing a mixture of REEs (Y, La, Ce, Pr, Nd, Eu, Gd, Tb, and Dy (1 µmol/L)) for determined periods of time under isothermal conditions (295 ± 1 K). All sorption experiments were carried out in Schott Duran^® flasks (Wertheim, Germany) of 0.5 L, under magnetic stirring (450 rpm) at room temperature (295.15 K) for a maximum of 168 h. The working solutions were prepared by adding a volume of the commercial stock solution of each REEs to 0.5 L of ultrapure water (salinity 0) or in a solution of seawater with a salinity of 15 or 30 to evaluate the effect of the ionic strength (seawater composition can be seen in Supplementary Material Table S1). The aqueous solution with a salinity of 15 was obtained by diluting seawater in ultrapure water. Experiments were run at three pH levels, 4, 6, and 8, to study the effect of pH on the sorption process. The pH adjustments were performed using solutions of HNO₃ (1 mol/L) or NaOH (1 mol/L). Different hazelnut shell masses were added to the multi-element spiked solutions, which was considered the starting point of the kinetic experiment. Sample collection was performed at the following increasing contact times: 0, 1, 6, 24, 48, 72, 96, and 168 h. The samples were acidified to pH < 2 with HNO₃ 65% and stored at 277.15 K for further quantification.

2.5. Quantification of REEs

The quantification of REEs (Y, La, Ce, Pr, Nd, Eu, Gd, Tb, and Dy) in solution was performed by Inductively Coupled Plasma Mass Spectrometry (ICP-MS), with a Thermo ICP-MS X Series (Waltham, MA, USA) equipped with a Burgener nebulizer. The concentrations were determined using a calibration curve, established with five standards ranging between 0.1 and 100 μg/L, prepared by dilution of the commercial certified stock solution in acid solution (HNO₃ 2% v/v). The lowest standard was considered the limit of quantification of this method, and only calibration curves with correlation coefficients higher than 0.999 were considered acceptable.

2.6. Data Analysis

The average amount of REEs sorbed by hazelnut shells per unit mass, ($q$, mg/g), at time t was calculated by global mass balance, as follows:

$$q={\displaystyle \frac{\left(C_0-C_t\right)\times V}{m}}$$

(1)

where $V$ is the solution volume (L), $m$ is the mass of hazelnut shells (g), $C_0$ is the initial concentration of REEs in solution (µg/L), and $C_t$ is the concentration of REE in solution at time t.

The removal efficiency (Removal, %) of REEs by the biosorbent was calculated as follows:

$$Removal \left(\%\right)={\displaystyle \frac{(C_0-C_t)}{C_0}}\times 100$$

(2)

The repeatability was assessed by the coefficient of variation (CV) as follows:

$$CV={\displaystyle \frac{s}{\overline{x}}}\times 100$$

(3)

2.7. Response Surface Methodology

Response surface methodology (RSM) is a statistical tool used to evaluate and represent the relationship between independent factors and the target response. This methodology uses empirical models to explain and predict the behaviour of the particular response based on fitting linear, quadratic polynomial functions and other models to the experimental results [33]. Generally, the optimization is performed in the following three main steps: (i) running the designed experiments; (ii) obtaining the estimated coefficients and the mathematical model; and (iii) plotting response curves and determining optimal conditions [34].

In this work, a three-factor, three-level (−1; 0; +1) Box–Behnken design was used to assess their influence on the removal (%) and uptake (mg/g) of REEs (

Table 1. Experimental conditions for three different factors studied and the three level conditions for the removal of REEs from water using hazelnut shells as a biosorbent.

Variable	Level
	−1	0	+1
pH	4	6	8
Salinity (g/L)	0	15	30
Hazelnut shell dose (g/L)	0.25	0.50	0.75

). Repeated observations at the central point were used to estimate the experimental error. The 15 experiments generated by the Box–Behnken design are shown in

Table 2. Description of the experimental conditions according to the Box–Behnken design. Fixed conditions: temperature of 295.15 K, stirring velocity of 450 rpm, and volume of 0.5 L, along with 168 h of exposure.

Experiment	pH	Salinity (g/L)	Hazelnut Shell Dose (g/L)
1	4	15	0.25
2	4	15	0.75
3	6	30	0.75
4	4	30	0.50
5	8	30	0.50
6	8	15	0.25
7	6	15	0.50
8	4	0	0.50
9	6	15	0.50
10	6	0	0.25
11	8	0	0.50
12	6	15	0.50
13	6	30	0.25
14	8	15	0.75
15	6	0	0.75

.

2.8. Statistical Analysis

A quadratic equation model according to the following Equation (4) was used for predicting and optimizing the response [33,35]:

$$Y={\beta }_0+{\sum }_{i=1}^k{\beta }_i{X}_i+{\sum }_{i=1}^k{\beta }_{ii}{X}_i^2+{\sum }_{i<j}^k{\beta }_{ij}{X}_i{X}_j$$

(4)

where $Y$ refers to the response variable studied, i.e., REEs removal or uptake, ${\beta }_0$ is a constant coefficient, ${\beta }_i$, ${\beta }_{ii}$, and ${\beta }_{ij}$ are the different interaction coefficients of the model, and $X_i$, $X_j$,…, $X_k$ are the codified independent variables related to the factors calculated by the equation represented bellow:

$$X_k={\displaystyle \frac{x_k-x_0}{{∆x}_k}}$$

(5)

where $X_k$ is the codified value of the independent variable $x_k$, $x_0$ is the variable value at its center point, and ${∆}_{x_k}$ is the step change between levels for the k variable. These codified variables allow their comparison in a common basis.

Data treatment was performed using the software Design-Expert (version 13, StatEase statistics made easy, Minneapolis, MN, USA). Analysis of variance (ANOVA) was used to assess the significance of the factors and the interactions between process variables and response. The quality of fit of the polynomial model was expressed by the coefficient of determination R², and statistical significance was checked by the F-test. The residual error, pure error, and lack-of-fit were calculated from the repeated measurements.

3. Results and Discussion

3.1. Removal Percentage of Rare Earth Elements by Hazelnut Shells

The results of the removal (%) of Y, La, Ce, Pr, Nd, Eu, Gd, Tb, and Dy by hazelnut shells at 24, 48, and 72 h are presented in

Table 3. Removal (%) of REEs (Y, La, Ce, Pr, Nd, Eu, Gd, Tb, and Dy) from a mixture by hazelnut shells, under the experimental conditions presented in Table 2 for 24, 48, and 72 h.

			Experiment
			1	2	3	4	5	6	7	8	9	10	11	12	13	14	15
Removal (%)	Y	24 h	0	4	39	3	31	14	22	62	27	46	78	25	13	39	87
		48 h	0	5	53	6	40	30	46	70	44	54	87	39	21	53	91
		72 h	0	5	64	4	49	36	46	75	48	61	91	48	29	64	95
	La	24 h	0	1	30	3	16	8	14	69	17	48	84	17	21	17	82
		48 h	0	3	41	4	28	22	35	80	29	59	91	25	23	33	95
		72 h	0	3	51	3	36	33	35	85	34	66	91	33	33	50	96
	Ce	24 h	0	4	41	4	17	6	18	71	23	48	78	24	19	16	89
		48 h	0	4	53	3	28	21	42	82	38	62	86	34	26	34	93
		72 h	0	4	64	5	36	33	42	87	38	68	87	48	36	52	94
	Pr	24 h	0	4	45	5	17	5	22	72	26	48	77	27	21	17	88
		48 h	0	6	55	5	30	20	44	82	40	62	85	39	27	36	92
		72 h	0	5	64	5	38	32	44	87	42	68	85	47	37	54	94
	Nd	24 h	0	5	45	6	18	5	23	72	26	51	76	27	23	18	86
		48 h	0	6	55	0	31	22	45	82	41	64	85	41	26	36	92
		72 h	0	6	66	4	38	34	45	87	32	68	83	48	36	55	93
	Eu	24 h	0	5	52	6	21	7	27	71	34	51	73	33	26	21	86
		48 h	2	8	61	7	34	22	49	81	47	64	82	48	34	41	99
		72 h	0	8	71	6	42	34	50	86	49	71	86	57	44	58	93
	Gd	24 h	0	3	47	3	21	9	25	70	31	48	76	30	20	23	87
		48 h	0	7	60	7	35	23	48	80	45	61	85	43	28	42	92
		72 h	0	6	68	7	43	36	48	84	48	68	87	53	38	58	94
	Tb	24 h	0	4	51	4	25	10	28	66	33	48	73	33	22	27	86
		48 h	0	6	61	6	38	25	50	77	48	61	84	48	32	46	91
		72 h	0	6	70	6	45	36	50	81	51	66	88	57	43	61	93
	Dy	24 h	0	6	49	3	27	12	29	63	35	46	73	34	22	32	85
		48 h	0	7	61	5	39	28	49	74	50	58	84	49	31	49	90
		72 h	0	8	70	5	47	37	49	79	52	64	88	58	41	63	93

. The highest removals were obtained in experiment 15 (pH 6, salinity 0, and biosorbent dose of 0.75 g/L) at 72 h. More than 93% of each element was removed from the mixture. Reducing the time to 24 h reduced the removal of all elements, although values remained greater than 80%. High element removals (75–91%) were also achieved in experiments 8 (pH 4, salinity 0, and biosorbent dose 0.5 g/L) and 11 (pH 8, salinity 0, and biosorbent dose 0.5 g/L) at 72 h. These results show that the hazelnut shells have a high capacity to simultaneously remove REEs at different pH levels (4, 6, and 8) with a salinity of 0 and using 0.5–0.75 mg/L of biosorbent.

Hazelnut shells have been previously used to remove other elements from water (as Cu, Pb, Cd, Cr, Zn, and Hg) [36,37,38], and some parameters, such as pH and sorbent dose, have been assessed, although individually [32]. It is known that pH affects the protonation of functional groups in the biomass and, thus, can influence the uptake of elements by the biosorbent. When the pH is too acidic, the concentration of protons in the solution increases and, consequently, there will be competition with cations for active sites on the surface of biosorbents. On the other hand, for pH values that are too basic, metal cations tend to precipitate, given the high concentration of hydroxyl anions in the solution. It should be noted that the effect of pH on removal efficiency is dependent on the contaminant to be sorbed and the functional groups of the biosorbent, in particular their point of zero charge (PZC). The point of zero charge is defined by the conditions of the solution for which the surface density of positive charges (cations) equals that of negative charges (anions) [39]. Demirbaş et al. (2008) [40] studied the PZC of hazelnut shells in a pH range from 2 to 10 and observed that the biosorbent had no isoelectric point and had a negative zeta potential value at all pH values studied. These findings were also reported by Al-Ajji and Al-Ghouti (2021) [41].

In contrast to the present work, most of the studies use simplified conditions, such as distilled or deionized water as the contaminated water matrix, which do not consider the possible effect of ionic strength and dissolved organics on the complexation of metals in the solution and, consequently, on the competition for sorption sites. In general, the tests are carried out in individual systems, not considering the possible competition among elements present in multi-element systems [42]. Results show that higher removal percentages (%) are obtained with no salinity, which indicates that hazelnut shells do not present a good performance when the environment shows competition with salt ions.

To evaluate the normal dispersion and model repeatability, the Box–Behnken design (BBD) involves conducting central point replicates, in which all factors are set to their central values (Figure 1).

The repeatability of the central point results was assessed using the coefficient of variation (CV), with the highest variation being approximately 20% for Nd at 72 h, indicating that the repeatability of the model is acceptable.

3.2. Time Exposure Impact

Assessing the efficacy of the sorption process primarily depends on the contact time between the biosorbent and the REEs-enriched solution. Decreasing the contact time can be beneficial for practical industrial purposes, as it can potentially reduce cost [43]. The selection of the contact time depends on the main objective of the application. Non-parametric statistical tests (Friedman test with Dunn’s multiple comparisons test) indicated that, for all REEs, except Ce and Nd, time extensions from 24 h to 48 h were significant for the removal efficiencies (p-values < 0.05) (

Table 4. Friedman test with Dunn’s multiple comparisons test applied to REEs in the mixture for exposure times of 24, 48, 72, and 96 h. Significant values are shown as underlined bold values.

	Friedman Test with Dunn’s Multiple Comparisons Test (p-Value ≤ 0.05)
Element	24 h vs. 48 h	24 h vs. 72 h	24 h vs. 96 h	48 h vs. 72 h	48 h vs. 96 h	72 h vs. 96 h
Y	0.0433	0.0002	<0.0001	0.9438	0.0973	>0.9999
La	0.0281	0.0003	<0.0001	>0.9999	0.3374	>0.9999
Ce	0.3960	0.0006	<0.0001	0.2418	0.0533	>0.9999
Pr	0.0433	0.0008	<0.0001	>0.9999	0.0973	>0.9999
Nd	0.2034	0.0019	<0.0001	0.8254	0.1177	>0.9999
Eu	0.0225	0.0003	<0.0001	>0.9999	0.4626	>0.9999
Gd	0.0433	<0.0001	<0.0001	0.5381	0.2034	>0.9999
Tb	0.0225	<0.0001	<0.0001	0.7188	0.3374	>0.9999
Dy	0.0433	0.0001	<0.0001	0.7188	0.0655	>0.9999

). Time extension from 24 h to 72 or 96 h was also significant for all REEs removal (p-values < 0.05). Extending the contact time from 48 h to 72 h was not significant for the removal response.

Based on these results, 48 h appears to be the maximum contact time, as further contact times showed no significant differences in the removal (%) of the elements.

3.3. Response Surface Models and Statistical Analysis

To evaluate which factors (linear, quadratic, and interaction) significantly influenced the removal of REEs by hazelnut shells, ANOVA (p-values < 0.05) was used (

Table 5. Factors with the corresponding p-value for 48 h of contact. Bold green values mean positive significance, and underlined red values mean negative significance.

		p-Value for Quadratic Model
		pH	Salinity	HN Mass	pH2	Salinity2	HN Mass2	pH-Salinity	pH-HN Mass	Salinity-HN Mass
48 h	Nd	0.0496	0.0014	0.0659	0.0328	0.0090	0.2348	0.2881	0.7254	0.9332
	Gd	0.0375	0.0021	0.0411	0.0294	0.0096	0.1692	0.3744	0.6139	0.9762
	Eu	0.0503	0.0018	0.0440	0.0140	0.0085	0.2326	0.3025	0.5624	0.7491
	La	0.0369	0.0007	0.0751	0.1568	0.0028	0.3679	0.5553	0.7031	0.4612
	Dy	0.0162	0.0031	0.0399	0.0222	0.0188	0.1708	0.3210	0.5267	0.9328
	Tb	0.0245	0.0029	0.0477	0.0221	0.0147	0.1726	0.3193	0.5243	0.9792
	Pr	0.0697	0.0017	0.0635	0.0408	0.0080	0.2310	0.3942	0.6762	0.9440
	Y	0.0066	0.0015	0.0221	0.0671	0.0131	0.1770	0.4627	0.4152	0.8326
	Ce	0.0679	0.0017	0.0801	0.0536	0.0076	0.3051	0.4312	0.7217	0.8979

). The values in underlined red color correspond to a negative impact, meaning that the response (removal (%)) will decrease if that factor increases. The values in a bold green color correspond to a positive impact, meaning that if the factor increases, the removal (%) also increases. The black values represent factors that are not significant (p-value ≥ 0.05).

It is possible to observe that the most significant factor is salinity with a negative impact, which means that, when salinity increases, the removal percentage (%) of REEs decreases. The second most significant factor is pH with a positive impact; thus, when pH increases, the removal percentage (%) increases as well. The hazelnut shell mass is the least significant factor among the three studied, but when significant, it has a positive impact.

Before the application of any sorbent in the real world, it is necessary to evaluate its performance in complex mixtures and scenarios with different salinity, pH, and temperature (key parameters that influence the sorbent’s efficiency). Salinity is an interesting parameter since it can decrease the electrostatic forces between sorbent and sorbate due to competition among cations present in the solution [44]. This factor could explain the decrease in the removal percentage (%) observed in the present study, when salinity was at the highest value (salinity 30) compared to ultrapure water (salinity 0). The abundance of cations such as Ca²⁺, Mg²⁺, and Na⁺ may inhibit metal’s sorption [32].

The relationship between independent variables and the response was drawn by second-order polynomial equations. For a robust model, only variables with p-values < 0.05 (

Table 6. Reduced equations with significant factors and the respective coefficient of determination (R²) and adjusted coefficient of determination (adjusted R²) for each REE in the mixture.

Reduced Equation	Nº of Equation	R2	Adjusted R2
$\begin{array}{cc}\hfill \% Removal of Y& \\ & =2.77+8.15\times pH-4.42\times salinity\hfill \\ & +47.8\times mass{+9.66\times 10}^{-2}\times {salinity}^2\hfill \end{array}$	Equation (6)	0.8615	0.8061
$\begin{array}{cc}\hfill \% Removal of La& \\ & =48.9+5.42\times pH-6.17\times salinity\hfill \\ & +{1.42\times 10}^{-1}\times {salinity}^2\hfill \end{array}$	Equation (7)	0.8673	0.8311
$\% Removal of Ce=80.8-5.69\times salinity+{1.30\times 10}^{-1}\times {salinity}^2$	Equation (8)	0.7150	0.6675
$\begin{array}{cc}\hfill \% Removal of Pr& \\ & =-86.8+53.6\times pH-5.29\times salinity\hfill \\ & -{{4.06\times pH}^2+1.19\times 10}^{-1}\times {salinity}^2\hfill \end{array}$	Equation (9)	0.8387	0.7742
$\begin{array}{cc}\hfill \% Removal of Nd& \\ & =-97.8+57.1\times pH-5.22\times salinity-4.31\times {pH}^2\hfill \\ & +1.15{\times 10}^{-1}\times {salinity}^2\hfill \end{array}$	Equation (10)	0.8440	0.7817
$\begin{array}{cc}\hfill \% Removal of Eu& \\ & =-149+68.2\times pH-4.94\times salinity+42.5\times mass\hfill \\ & -5.26\times {pH}^2+{1.12\times 10}^{-1}\times {salinity}^2\hfill \end{array}$	Equation (11)	0.9063	0.8542
$\begin{array}{cc}\hfill \% Removal of Gd& \\ & =-124+57.7\times pH-4.93\times salinity+44.4\times mass\hfill \\ & -4.33\times {pH}^2+{1.12\times 10}^{-1}\times {salinity}^2\hfill \end{array}$	Equation (12)	0.8978	0.8410
$\begin{array}{cc}\hfill \% Removal of Tb& \\ & =-142+63.1\times pH-4.48\times salinity+42.4\times mass\hfill \\ & -{{4.72\times pH}^2+1.00\times 10}^{-1}\times {salinity}^2\hfill \end{array}$	Equation (13)	0.8870	0.8242
$\begin{array}{cc}\hfill \% Removal of Dy& \\ & =-144+62.2\times pH-4.17\times salinity+43.8\times mass\hfill \\ & -{{4.60\times pH}^2+9.20\times 10}^{-2}\times {salinity}^2\hfill \end{array}$	Equation (14)	0.8879	0.8256

) were considered in the reduced model. The validity of the model fit is verified with the closeness between the coefficient of determination (R²) and the adjusted coefficient of determination (adjusted R²); this closeness should be less than 0.2, which was the case in all cases examined in this work.

3.4. 3D Responses for the Removal of REEs by Hazelnut Shells

For easy visual understanding of the interactions of the three factors on the response (removal (%)) of REEs by the hazelnut shells, 3D response surface plots based on the second-order polynomial equations presented in Table 6 were created.

Figure 2, Figure 3 and Figure 4 show the 3D response surfaces for the removal of REEs obtained by plotting two factors and keeping one factor constant. Figure 2A, Figure 3A and Figure 4A show the interactive effects of pH and salinity on the removal of Nd, La, and Y, respectively, while keeping the biosorbent dose constant (0.50 g/L). The fixation of the biosorbent dose is related to the statistical analysis (not significant, p-value ≥ 0.05) for some of the models that describe the removal. For Nd, it can be seen that, for 48 h (Figure 2A), the decrease in salinity has a positive impact on the removal process. Regarding pH factor, it is possible to see that the central value (pH 6) gives the highest response when combined with the lower salinity. Most of the REEs studied in this work had 3D surfaces similar to the ones presented for Nd. For La (Figure 3A), higher removal (%) is obtained with higher values of pH and when in contact with lower salinities. Yttrium (Figure 4A) shows a similar behavior to La; however, for the highest salinity value (salinity 30), the removal (%) along the studied pH varied less than 20%. For Y, the behavior was different. For the highest salinity (30), the removal (%) along the studied pH range varied from 0 to 40% (the removal will be higher for higher pH values). The same is observed when the pH is constant at 4; the removal (%) along the studied salinity range varied from 0 to 60%. The highest removal occurs at the lowest salinity, aligned with the highest pH value.

The interactive effects of pH and biosorbent dose on the removal of Nd, La, and Y, while keeping the salinity constant (salinity 15), can be seen in Figure 2B, Figure 3B, and Figure 4B, respectively. The 3D surface of Nd (Figure 2B) shows that, only for pH levels between 5 and 7, in all mass ranges studied, the removal (%) is higher (near 40%). For La (Figure 3B), the 3D surface also shows that biosorbent dose does not have significant influence on the response, since the removal (%) is enhanced when the pH increases, independently of the mass used. Yttrium (Figure 4B) shows the positive effect on removal efficiency when increasing both biosorbent dose and pH, where the removal response increases from 0 to almost 60%.

Figure 2C, Figure 3C and Figure 4C show the interactive effects of biosorbent dose and salinity on the removal of Nd, La, and Y, respectively, while the pH is fixed at the central point (pH 6). Neodymium and La have similar 3D surfaces, where it is possible to observe that the biosorbent dose has no significant influence on the response (removal (%)). On the other hand, the decrease in salinity (from 30 to 0) increases the removal response by three to four times. The 3D surface for Y shows a minor influence of the biosorbent dose, where the response increases c.a. 20%.

3.5. Optimization of Operational Parameters Affecting REEs Removal by Response Surface Methodology

Optimization studies on the use of biosorbents to remove some REEs, such as Pr, La, and Ce have been reported in the literature using crab shells and orange peels [45] and brown alga C. indica by CS₂ [46], in which the response studied was the uptake by those biosorbents. The DoE helped localize the levels of the factors that contributed to the maximum Pr biosorption of 57.8 mg/g for crab shells (pH = 5.0, biomass dose = 2.5 g/L, initial Pr concentration = 250 mg/L, temperature = 50 °C, contact time = 35 min) and 49.9 mg/g for orange peels (pH 5.0, biomass dose = 2.0 g/L, initial Pr concentration = 200 mg/L, temperature = 50 °C, contact time = 50 min). For C. indica, maximum La biosorption of 179 mg/g was obtained at the following levels: pH 5.2, biomass dose = 0.5 g/L, initial La concentration = 250 mg/L, temperature = 55 °C; for Ce, the conditions were pH 5.0, biomass dose = 0.7 g/L, initial Ce concentration = 227 mg/L, and temperature = 55 °C, with a maximum biosorption of 168 mg/g. To the best of our knowledge, no optimization study was carried out with hazelnut shells.

The values of the optimized variables for the removal of REEs by hazelnut shells are presented in

Table 7. Optimal condition values for the maximum removal (%) and uptake (mg/g) of REEs from the mixture at 48 h of exposure.

Optimal Condition for Removal (%)
	pH	Salinity	Sorbent Dose (g/L)	Removal (%)	Uptake (mg/g)	Total Uptake (mg/g)
Y	7.0	0	0.75	96	0.103	1.52
La				87	0.187
Ce				81	0.103
Pr				89	0.192
Nd				90	0.141
Eu				100	0.205
Gd				100	0.194
Tb				99	0.180
Dy				99	0.220
Optimal Condition for Uptake (mg/g)
	pH	Salinity	Sorbent dose (g/L)	Removal (%)	Uptake (mg/g)	Total Uptake (mg/g)
Y	7.0	0	0.25	72	0.170
La				87	0.304
Ce				81	0.170
Pr				89	0.324
Nd				90	0.312	2.69
Eu				81	0.360
Gd				79	0.383
Tb				78	0.302
Dy				77	0.363

. Not surprisingly, since these elements have similar chemical properties, the conditions that maximize removal are the same for all REEs (pH 7.0, salinity 0, and biosorbent dose of 0.75 g/L). With these experimental conditions, it was possible to reach virtually 100% removal of Gd and Eu and more than 80% of the remaining REEs. These efficiencies are higher than those reported by Fabre et al. (2021) [47] on the removal of Nd by marine macroalgae (71 and 77% with Ulva sp. and Gracilaria sp., respectively, at 72 h) and by Ferreira et al. [43] on the removal of Dy by the same macroalgae (59 and 83% for Ulva sp. and Gracilaria sp., respectively, at 96 h). Raji et al. (2018) [48] also reported a removal efficiency of 99% for Dy using multi-walled carbon nanotube (MWCNT), but in a mono-elementary solution. In this sense, the results of the present study highlight the use of hazelnut shells as a great option to remove REEs from contaminated solutions.

After the water treatment process, it is important to recover the REEs from the enriched hazelnut shells, which can be performed using diluted acid solutions or by calcining the biomass [49]. In this sense, it is also of interest to optimize the concentration of REEs in the biosorbent (uptake). Table 7 shows that, by using the best removal conditions, the highest uptake was verified for Dy (0.22 mg/g), with a total accumulation of REEs of 1.52 mg/g. By optimizing the uptake (pH 7.0, salinity 0, and biosorbent dose of 0.25 g/L), it is possible to obtain a total accumulation of REEs of 2.69 mg/g (highest value for Gd, 0.38 mg/g). This REEs content in hazelnut shells is higher than those in common ores [50], showing that it may be a secondary source of these elements, which are currently classified as critical raw materials.

3.6. Hazelnut Shell Removal of Rare Earth Elements Under Optimal Operational Conditions in Presence and Absence of Potentially Toxic Elements

To validate the optimal conditions predicted by the models (Table 7), the removal of REEs was experimentally evaluated using those optimal operating conditions. Overall, the removal performance showed a good agreement between the predicted and obtained values, with some variation observed for the medium and heavy REEs (Eu, Gd, Tb, and Dy (Figure 5A)).

The effect of competing ions on the removal of REEs was also evaluated using the same optimal operating conditions. The results showed no loss of efficiency for REEs removal when in a more complex solution in the presence of potentially toxic elements (PTEs). Yet, no selectivity was observed, as the hazelnut shell was able to remove 89% of Cu, 93% of Cd, and 84% of Pb (Figure 5B).

In comparison with the literature, few studies have evaluated the capacity of hazelnut shells or any nutshell to remove REEs with or without competing ions (Supplementary Material Table S2). Some studies evaluated the removal of PTEs in mono- and multi-element solutions [51,52], yet with no selectivity observed.

As global nut production continues to grow, the associated generation of nutshell waste correspondingly increases. More than 1195 t of hazelnut shells are produced annually [53]. These results show the importance of studying the capacity of agricultural food to reduce waste disposal and add value to nutshells.

3.7. Hazelnut Shell Characterization

FTIR spectroscopy was used to identify the main functional groups present on the hazelnut shells’ surface before and after exposure to REEs solutions and REEs + PTEs solution (Figure 6A). The hazelnut shells showed common peaks associated with the main polymers present in most nutshells, which are cellulose, hemicellulose, and lignin [53]. The broad absorption band observed between 3000 and 3500 cm⁻¹ corresponds to the O–H stretching vibrations, indicative of hydroxyl groups present in lignin, cellulose, and hemicellulose [54]. The bands observed between 1020 and 1035 cm⁻¹, along with the peak near 1370 cm⁻¹, are associated with C–O stretching vibrations of alcohols, carboxylic acids, and esters, as well as C–H bending of alkenes, respectively, which are functional groups typically found in cellulose and hemicellulose [55]. The peak around 1735 cm⁻¹ is attributed to the stretching vibrations of C=O bonds present in the acetyl and ester groups of hemicellulose and lignin [55]. A characteristic absorption band at 1607 cm⁻¹ corresponds to the C=C stretching vibrations of aromatic rings, which is a distinctive feature of the structure of lignin [55]. Moreover, there is a noticeable decrease in the sharpness of the peaks near 2850 cm⁻¹ after exposure, which can be attributed to CH and CH₂ stretching vibrations of methyl and methylene groups [54].

SEM images were obtained at the same scale (Figure 6B–D), allowing the comparison of surface structures in the hazelnut shells before and after exposure. The surface of hazelnut shells exhibited a pronounced porous structure. The exposure to REEs and REEs + PTEs did not lead to any morphological alterations.

4. Conclusions

The present study proposed the valorization of food waste, generated in large quantities every year, as a biosorbent to decontaminate water and allow the recovery of elements of interest such as REEs. Results demonstrated the high capability of hazelnut shells in removing REEs from multi-contaminated solutions. Response surface methodology allied with the Box–Behnken design selected for the present study allowed the identification of the factors that are significant and that most contribute to removal efficiency. Optimized conditions predicted by the developed model equations allow for removal efficiencies between 80% and 100%, which is very promising from the perspective of industrial application. Water salinity was the factor with the greatest impact on removal, with higher salinities decreasing removal (%). In the opposite direction, the biosorbent dose was the factor with the lowest significance. The factors evaluated in this work are common to biosorption studies and are easy to control. The large range of conditions considered here provides a better understanding of the expected results, allowing for reduced periods of time and lower resource demand for experimental trials. The sorption capacity of hazelnut shells did not decrease in the presence of PTEs. This work highlights the importance of the study of agricultural waste as a sorbent to reduce residues of the agricultural industry, promoting the removal and recovery of technologically critical elements, aiming at a circular economic process.