Experimental Investigation and Proposal of Artificial Neural Network Models of Lead and Cadmium Heavy Metal Ion Removal from Water Using Porous Nanomaterials

Abstract

This study used porous nanomaterials MCM-41 and SBA-15, as well as their modified species, to remove lead and cadmium ions from water. We used X-ray diffraction (XRD), a scanning electron microscope (SEM), the Brunauer–Emmett–Teller (BET), and the Fourier transform infrared (FT-IR) method to investigate the characteristics of porous nanomaterials. Additionally, atomic absorption spectroscopy (AAS) measured the concentration of lead and cadmium ions. The stratigraphic analysis showed the samples’ isothermal shape to be type IV. This study investigated the amount, absorbent, pH changes, and adsorption time parameters. We observed that the adsorption efficiency of lead by the synthesized samples was higher than that of the adsorption of cadmium. Mesoporous structures also displayed increased adsorption efficiency due to the amino group. Four testing stages were conducted to determine the reproducibility of the adsorption by the synthesized samples, with the results showing no significant changes. As a result of the adsorption process, the structure of the recycled sample ${\mathrm{N}\mathrm{H}}_2-\mathrm{M}\mathrm{C}\mathrm{M}-41$ was preserved. We also used artificial neural networks (ANN) to propose predictive models based on the experimental results. The ANN models were very accurate, such that the mean absolute error (MAE) was less than 2% and the ${\mathrm{R}}^2$ was higher than 0.98.

1. Introduction

Heavy metals refer to a group of metal elements that have a specific gravity (atomic density) greater than 6 $\mathrm{g}/{\mathrm{c}\mathrm{m}}^3$ and an atomic mass greater than 55.8 g/mol [1]. The existence of some of these components in the diet of humans and other living beings is necessary, which is why they are called essential elements [2,3]. In concentrations exceeding permissible limits, these elements can cause various complications for humans and other living beings. Also, they cause pollution and environmental hazards [4,5,6]. Heavy metals include mercury, arsenic, cadmium, lead, nickel, copper, iron, chrome, vanadium, and others with different boiling points [7,8]. Metal industries, electroplating, paint making, battery making, tanning, textile, paper making, and other similar industries are among sources of environmental pollution related to heavy metal emissions [9,10,11]. These cause pollution in the environment by disposing of and releasing elements such as cadmium, mercury, nickel, lead, chrome, copper, silver, and others. Heavy metals are part of the biosphere and are found naturally in the soil and plants. In addition to being dangerous for human health, these metals are non-renewable resources [12,13].

By entering the cells of living organisms, most heavy metals in biological reactions, prevent some of the physical–chemical responses of cells and reduce purification efficiency. In severe cases, they stop the biological activities of the purification systems [14,15,16,17]. Lead and cadmium are metals that cause many unpleasant complications for human health. The regulators have limited Cd concentrations in drinking water to a high of 0.003 ppm [18,19,20]. Additionally, the lead concentration in drinking water can go up to 0.01 ppm [6]. In general, it can be said that one of the reasons for the contamination of drinking water with lead is the corrosion of water pipes. Groundwater pollution is possible due to the infiltration of surface water and the migration of metals in the soil [21,22,23].

In order to maintain the quality of drinking water as well as control pollutant sources, heavy metals need to be removed from water. This critical matter requires that proper research be conducted in this field. Until now, many methods for measuring these elements from water have been presented. For example, some of the measuring methods are spectrophotometry [24], flame atomic absorption spectroscopy (AAS) [25,26], graphic furnace atomic adsorption spectrometry [27,28,29], and electrochemical oxidation [30,31,32,33,34]. Also, among the heavy metal adsorption methods, we can mention precipitation [35,36,37,38], ion exchange [39,40,41], and liquid–liquid extraction [42,43,44].

In a study, Ahmadi et al. [45] investigated the removal of lead ions from water using a nanoclay nanocomposite/chitosan/multi-walled carbon nanotube. In this study, the nanotube was simply prepared via mixing in solvent and interlayer penetration. It was then investigated in a discontinuous process to remove lead from water. The results obtained from the adsorption tests showed that the concentration of the remaining metal ions processed using nanocomposites significantly decreased. In less than 180 min, the pH became equal 7. At an ambient temperature, 98% of the lead ions are removed using nanocomposite compared to pure nanoclay. In a study [46], magnetic iron nanoparticles were used to remove lead from the wastewater of battery industries. The first focus of this study was on the lead contamination of artificial wastewater using magnetic iron nanoparticles. Then, the battery factory effluent was tested as a sample. Also, by changing the physical and chemical parameters such as pH, temperature, the number of nanoparticles, and the initial concentration of lead, Freundlich and Langmuir adsorption isothermal curves were drawn. This research shows a high percentage of lead removal from natural wastewater, with an efficiency of 70% at acidic pH and ambient temperature. Different methods have been proposed and investigated due to the need to remove or reduce the amount of lead and cadmium from water. Still, due to these methods’ time-consuming nature and high cost, researchers have sought to provide more straightforward and less expensive ways. This study used the surface adsorption method with porous nanomaterials to absorb lead and cadmium.

One of the essential features of nanoporous materials is their selectivity of shape and size, which has led to many catalytic applications for purification and separation and made their role in nanotechnology more prominent. In this study, water from mesoporous MCM-41 and SBA-15 samples, functionalized using amine group ($-{\mathrm{N}\mathrm{H}}_2$), has been used to separate and remove ${\mathrm{P}\mathrm{b}}^{2+}$ and ${\mathrm{C}\mathrm{d}}^{2+}$ ions from samples. Amine groups exhibit strong affinity for metal ions, ensuring their efficient capture and removal from the solution, streamlining purification. The use of nanoporous MCM-41 and SBA-15 with tailored pore sizes optimizes ion adsorption, enhancing separation efficiency. Using nanoporous materials with water aligns with the standards for green chemistry, reducing their environmental impact. These versatile materials can be regenerated, enhancing their sustainability and cost-effectiveness. Being successful at a smaller scale, this technique holds promise for application to larger-scale metal ion separation processes. Insights from amine-functionalized nanoporous materials advance our understanding of nanoscale interactions and inform novel applications. Different tests are performed, and the results are compared to achieve the best model. Finally, an artificial neural network is used to predict removal efficiency. The machine learning model is thoroughly explained, and the computational costs are also compared with the time required to carry out experiments.

2. Materials and Methods

2.1. Required Materials for Solutions

All reagents used in the study were of analytical grade and purchased from Sigma Aldrich (Taufkirchen, Germany) or Fisher Scientific (Leicestershire, UK). In order to carry out the experiment, the following materials were required: lead nitric acid, nitrate, cadmium nitrate, sodium hydroxide, cetyl trimethyl ammonium bromide (CTAB), pleuritic surfactant P-123, tetraethyl orthosilicate (TEOS), methanol, hydrochloric acid, N-(2-Aminoethyl)-3-aminopropyltrimethoxysilane (APTMS), n-hexane, and toluene. To prepare a 1000 ppm ${\mathrm{P}\mathrm{b}}^{2+}$ and ${\mathrm{C}\mathrm{d}}^{2+}$ solution, the appropriate amount of lead nitrate and cadmium nitrate was carefully weighed and made up to 100 mL with distilled water in a volumetric flask.

2.2. Synthesis of Mesoporous SBA-15 and $N{H}_2-SBA-15$

Various methods have been proposed for the preparation of SBA-15; the hydrothermal method is mentioned in this study. This method completely dissolved 4 g P-123 Pluronic surfactants in 20 mL of concentrated hydrochloric acid and 104 g of distilled water. The role of the surfactant is to form the main mesoporous template of SBA-15, which has a hexagonal shape. After homogenizing the mixture, an electric stirrer is used to stir the TEOS silica source at 40 °C for 24 h. By increasing the temperature to 90 °C, the mixture is placed in static conditions so that the silicon source is placed onto the surface of the hexagonal mold created by the surfactant. After 24 h, the white substance is separated using filter paper. An oven heated to 550 °C for 6 h was used to remove the surfactant mold.

In order to chemically modify the surface of SBA-15 with an amine functional group ($-\mathrm{N}{\mathrm{H}}_2$), 5 g of SBA-15 was added to 150 mL of dry toluene and the mixture was placed under reflux conditions. The mixture was then recirculated for 24 h with 5.5 mL of (3-Aminopropyl) triethoxysilane (APTES). Once the mixture had cooled, it was filtered.

2.3. Synthesis of Mesoporous MCM-41 and ${NH}_2$-MCM-41

In order to synthesize MCM-41 in 240 mL of distilled water and 1.75 mL of sodium hydroxide, CTAB 500 mg was dissolved in 240 mg. After dissolution, 2.5 mL of TEOS was added as a silica source and the solution was stirred for 3 h. To remove the surfactant mold, we put the obtained product in an oven with a temperature of 550 °C for 6 h [47].

In order to synthesize ${\mathrm{N}\mathrm{H}}_2$-MCM-41, the synthesized sample was placed under a vacuum to dry and open the pores; then, 20 mL of the solvent hexane was added to it. Under rapid mixing conditions and the presence of ${\mathrm{N}}_2$, 1.15 mmol of APTMS gas was added as an amine precursor to the mixture drop by drop. The obtained mixture was refluxed for 24 h at 80 °C, and the final product was smooth. It was washed several times using hexane solvent. Finally, it was placed in a desiccator for 24 h to dry. The final product was ${\mathrm{N}\mathrm{H}}_2$-MCM-41 [48].

2.4. The Experiment Procedure

In order to carry out the experiment, certain amounts of the synthesized samples were added to 25 mL solution of 2 ppm ${\mathrm{P}\mathrm{b}}^{2+}$ and ${\mathrm{C}\mathrm{d}}^{2+}$. Then, the obtained mixture was stirred for different amounts of time at various pH levels and filtered using filter paper. The filtered solution was studied to determine the concentration of the desired remaining ions using atomic adsorption spectrometry (AAS). Also, the remaining particles on the filter paper were washed with 10 mL of 3 M nitric acid to check the reproducibility of the synthesized nanoabsorbers.

The characterization methods employed for the porous materials encompassed various techniques. X-ray diffraction (XRD), FT-IR analysis, SEM, and BET were carried out to characterize the mesoporous materials.

2.5. An introduction to Artificial Neural Network (ANN)

In our study, we used ANNs to predict the percentages of adsorption using the data we gathered from our experiments. We took great care in crafting our ANN models, and we fine-tuned their details through a process called hyperparameter tuning. This step helps us to adjust things like batch size, epochs, and activation functions to improve the performance of our models. The detailed information about these models is presented in in the results, where the specifics are listed, including the number of hidden layers and the training parameters. Also, Figure 1 presents a thorough overview of the ANN predictive model with its layers.

Figure 1. The ANN architecture.

3. Results

3.1. X-ray Diffraction (XRD) Pattern of Samples

The results of XRD (Shimadzu, Japan) patterns for all the samples are shown in Figure 2. The results of XRD for MCM-41 and $\mathrm{N}{\mathrm{H}}_2$-MCM-41 show a sharp peak with greater intensity at the angle of 2.2, corresponding to XRD from 100 crystal plates. The sharp peaks at angles of 3.92, 4.52, and 6.02 correspond to the XRD from 110, 200, and 210 crystal plates, respectively, which represents the hexagonal structure of MCM-41 and ${\mathrm{N}\mathrm{H}}_2$-MCM-41. It can also be seen that the design of the synthesized sample is well preserved after chemical modification with the amine group [49].

Figure 2. XRD pattern of nanoporous (a) MCM-41, (b) NH2-MCM-41, (c) SBA-15, and (d) NH2-SBA-15.

In the XRD for SBA-15 and $\mathrm{N}{\mathrm{H}}_2$-SBA-15, the sharp peak at 0.98 corresponds to XRD from the crystal plate of 100. Also, the presence of two other peaks with low intensity at 1.70 and 1.87 is related to crystal plates of 110 and 200, respectively. Using SBA-15 as an example, the three peaks indicate the two-dimensional hexagonal mesopore structure [50].

3.2. FT-IR Infrared Spectroscopy Analysis

In the FT-IR spectrum of the MCM-41 sample in Figure 3, the broadband in the region of 3450 ${\mathrm{c}\mathrm{m}}^{-1}$ is related to the OH group. As can be seen in Figure 3, for Si-O-Si vibrations [51], the region at 1090 ${\mathrm{c}\mathrm{m}}^{-1}$ indicates asymmetric stretching, the region at 816 ${\mathrm{c}\mathrm{m}}^{-1}$ indicates symmetric extension, and bending vibrations can be seen at the 459 ${\mathrm{c}\mathrm{m}}^{-1}$ region. Also, the very weak peak in the 2950 ${\mathrm{c}\mathrm{m}}^{-1}$ region corresponds to the C-H group of the sample, which has been removed from the absorbent structure due to calcination. The peak seen in the region of 1640 ${\mathrm{c}\mathrm{m}}^{-1}$ is related to the H-O-H bending vibration of water molecules adsorbed onto the adsorbent surface.

Figure 3. FT-IR spectrum of the sample.

According to the FT-IR spectrum of the ${\mathrm{N}\mathrm{H}}_2$-MCM-41 sample in Figure 3, the broad bands seen in the 696 and 1560 ${\mathrm{c}\mathrm{m}}^{-1}$ regions are related to the bending vibration of the H-N group and the symmetric stretching vibration of the ${\mathrm{N}\mathrm{H}}_2$ group [52].

Si-O-Si bonds’ symmetric and asymmetric stretching vibrations are detected in the FT-IR spectrum of the SBA-15 sample at 810 and 965 ${\mathrm{c}\mathrm{m}}^{-1}$, respectively. The peak at 1629 ${\mathrm{c}\mathrm{m}}^{-1}$, which is relatively sharp, is related to the bending vibrations of H-O-H of water molecules absorbed by the absorbent surface [53].

In the FT-IR spectrum of the aminated sample (${\mathrm{N}\mathrm{H}}_2$-SBA-15), the peak at 1577 ${\mathrm{c}\mathrm{m}}^{-1}$ is related to the bending vibration of the amine group, and the peaks at 2860 and 2960 ${\mathrm{c}\mathrm{m}}^{-1}$ are related to the C-H stretching vibration of the propyl group, which indicates the presence of the ${-\mathrm{N}\mathrm{H}}_2$ the group in the structure [54].

Examination of the FT-IR spectrum of the synthesis samples shows that the surfactant used as a template has been completely removed from the absorbent structure via calcination. Also, the appearance of peaks related to the amine functional group in certain areas shows that the functionalization of the adsorbents has been conducted correctly.

3.3. Examination of SEM Images

The results obtained from the electron microscope (JSM-6060LV JEOL Ltd., Tokyo, Japan) are shown in Figure 4. The SEM images of MCM-41 and ${\mathrm{N}\mathrm{H}}_2$-MCM-41 samples are shown in Figure 4. The image obtained from SEM showed that the synthetic mesoporous structure is in the form of regular spheres, and the size of the crystals was estimated to be between 57 and 60 nm, respectively. Also, the SEM image of mesoporous SBA-15 and ${\mathrm{N}\mathrm{H}}_2$-SBA-15 is given in Figure 4c,d, which show string-like morphology in accordance with the results of other sources [55].

Figure 4. SEM image of nanoporous (a) MCM-41, (b) ${\mathrm{N}\mathrm{H}}_2$-MCM-41, (c) SBA-15 and (d) ${\mathrm{N}\mathrm{H}}_2$-SBA-15.

3.4. Surface Analysis (BET)

BET analysis isothermal nitrogen adsorption–desorption for MCM-41 and ${\mathrm{N}\mathrm{H}}_2$-MCM-41 is shown in Figure 5. The temperatures obtained for these samples were isothermal type IV of the classification of the International Union of Pure and Applied Chemistry (IUPAC), which shows that the synthesized material is mesoporous and has the characteristics of MCM-41.

Figure 5. Isothermal nitrogen adsorption and desorption.

At relative pressures ($P/P_0$) less than 0.6, a linear increase in the volume of absorbed nitrogen can be seen, and this volume increase is very intense in the relative pressure range between 0.68–0.73. This sharp increase in the immersed volume is related to the stomatal condensation phenomenon inside the mesoporous pores of ${\mathrm{N}\mathrm{H}}_2$-MCM-41.

Figure 5 shows an analysis of the surface of SBA-15 and SBA-15 samples as adsorption–desorption isotherms. They correspond to porous structures with pores between 2 and 30 nm and are classified as type IV isotherms according to the IUPAC classification [56,57].

3.5. Examining Parameters Affecting Adsorption

The effects of various factors, including the amount of adsorbent, adsorption time, and pH, were investigated to find the optimal conditions for achieving the highest adsorption efficiency. The single factorial optimization method is applied to achieve optimal conditions.

3.5.1. Examining the Effect of the Absorbent Amount

To investigate the effect of adsorbent amount on the adsorption of 2 ppm ${\mathrm{P}\mathrm{b}}^{2+}$ and ${\mathrm{C}\mathrm{d}}^{2+}$ ions, different amounts of adsorbents were used. According to the obtained results in Figure 6, the adsorption of ${\mathrm{P}\mathrm{b}}^{2+}$ and ${\mathrm{C}\mathrm{d}}^{2+}$ is optimized for MCM-41 in the amounts of 30 mg and 30 mg, ${\mathrm{N}\mathrm{H}}_2$-MCM-41 in the amounts of 30 mg and 40 mg, SBA-15 in the amounts of 30 mg and 30 mg, and ${\mathrm{N}\mathrm{H}}_2$-SBA-15 in 30 mg and 30 mg amounts, respectively. Figure 6 shows the results well. According to the obtained results, it can be seen that with the increase in the amount of absorbent due to the rise in the amount of active surface for adsorption, the adsorption process trends upward, except for $\mathrm{N}{\mathrm{H}}_2$-SBA-15. Also, the presence of the amine functional group increases the adsorption. The use of many adsorbents, including the amino group, led to the formation of hybrid materials, and these organic–inorganic hybrid materials have a high adsorption capacity for heavy metal ions.

Figure 6. Adsorption percentage change (a) for $\mathrm{P}{\mathrm{b}}^{+2}$ and (b) $\mathrm{C}{\mathrm{d}}^{+2}$ according to the amount of adsorbent.

3.5.2. Investigation of Adsorption Time

To check the optimal time for absorption, 25 mg of a solution containing 2 ppm of ${\mathrm{P}\mathrm{b}}^{2+}$ and ${\mathrm{C}\mathrm{d}}^{2+}$ was stirred for 5 to 25 min, and the species was absorbed. According to Figure 7, the results of the adsorption efficiency of ${\mathrm{P}\mathrm{b}}^{2+}$ and ${\mathrm{C}\mathrm{d}}^{2+}$ by MCM-41 went from 85.4% to 92.5% and from 82.3% to 89.9%, ${\mathrm{N}\mathrm{H}}_2$-MCM-41 went from 95.1% to 97.1% and from 91.3% to 96.9%, SBA-15 increased from 94.1% to 96.8% and from 91.3% to 96.9%, and ${\mathrm{N}\mathrm{H}}_2$-SBA-15 went from 95.3% to 97.1% and 93.1% to 97.1%, respectively. In general, the results show that increasing the adsorption time increases the duration of interaction between the adsorbent and the desired ions. Finally, the adsorption efficiency rises so that the increase in adsorption efficiency is more significant than the increase in the adsorption process time.

Figure 7. Adsorption percentage change (a) for $\mathrm{P}{\mathrm{b}}^{+2}$ and (b) $\mathrm{C}{\mathrm{d}}^{+2}$ according to interaction time.

3.5.3. Examining the Effect of pH on Adsorption

To investigate the effect of pH on the adsorption of ${\mathrm{P}\mathrm{b}}^{2+}$ and ${\mathrm{C}\mathrm{d}}^{2+}$ ions, 72 solutions of 25 mL containing 2 ppm of these two ions were adjusted using nitric acid and sodium hydroxide solutions at different pHs, and the adsorption process was carried out. The results show that the studied ions in the pH range of 6–8 can have quantitative optimal adsorption. The results can be seen in Figure 8. The results show that the adsorption rate is affected by pH, and that the adsorption efficiency is lower at low pHs in all samples. This is because at low pH, the amine groups on the surface of the adsorbent, which form chelate with metal ions, are protonated and lose their ability to form complexes with metal ions [58].

Figure 8. Adsorption percentage change (a) for $\mathrm{P}{\mathrm{b}}^{+2}$ and (b) $\mathrm{C}{\mathrm{d}}^{+2}$ according to pH changes.

3.6. Adsorbent Repeatability

To investigate the reproducibility of adsorbents, the mesoporous materials that were used once in the adsorption process were isolated. They were washed with 10 mL of 3 M nitric acid and used in four stages. According to the adsorption percentage results shown in Table 1, it can be seen that, after repeating the adsorption process four times, the adsorption percentage results did not change much. This shows that the synthesized adsorbents have repeatability.

Table 1. Checking the reproducibility of absorbents.

MCM-41		${\mathbf{N}\mathbf{H}}_2$-MCM-41		SBA-15		${\mathbf{N}\mathbf{H}}_2$-SBA-15
$\%{\mathbf{P}\mathbf{b}}^{2+}$	$\%{\mathbf{C}\mathbf{d}}^{2+}$	$\%{\mathbf{P}\mathbf{b}}^{2+}$	$\%{\mathbf{C}\mathbf{d}}^{2+}$	$\%{\mathbf{P}\mathbf{b}}^{2+}$	$\%{\mathbf{C}\mathbf{d}}^{2+}$	$\%{\mathbf{P}\mathbf{b}}^{2+}$	$\%{\mathbf{C}\mathbf{d}}^{2+}$
77	77	86	83	95	95.5	95	95
79	78	79	80	94.5	93	92	95.7
81.5	81.4	76	85	91	90	95	96.3
83	83	82	78	86	85.5	95	96.1

3.7. XRD of the Recovered Sample

To ensure that the structure of the recycled piece was removed after the adsorption process, the recycled model of ${\mathrm{N}\mathrm{H}}_2$-MCM-41 was studied via XRD. The XRD pattern of this catalyst is shown in Figure 9. According to this pattern, the peak intensity crystal plates decrease after the adsorption process, but the hexagonal structure of ${\mathrm{N}\mathrm{H}}_2$-MCM-41 is still preserved. The index peak in the plate of 100 at an angle of 2.12 and the peaks at higher grades confirm this claim.

Figure 9. XRD pattern of recovered ${\mathrm{N}\mathrm{H}}_2$-MCM-41 sample.

4. Artificial Neural Network

Artificial neural networks (ANNs) have emerged as powerful tools for modeling and predicting complex systems in various fields, including wastewater treatment research [59,60,61,62,63,64,65,66,67]. ANNs are computational models that mimic the structure and function of the human brain, consisting of interconnected nodes that process and transmit information through a series of layers [68]. In wastewater treatment research, ANNs have been used to model and predict various aspects of the treatment process, such as nutrient removal, sludge production, and effluent quality. ANNs have several advantages over traditional modeling techniques, including their ability to handle large amounts of data, adapt to changes in the system, and learn from experience [69]. As such, ANNs have become an increasingly popular tool in wastewater treatment research, offering new insights into the complex processes involved in treating wastewater and improving the efficiency and effectiveness of treatment systems [70,71,72,73].

In this study, we used ANN to predict the percentage of adsorption of $\mathrm{P}{\mathrm{b}}^{+2}$ and $\mathrm{C}{\mathrm{d}}^{2+}$. Predictive models are developed based on the data collected in the experiments. Table 2 lists the specifications of these models. A hyperparameter tuning procedure is used in order to optimize the results of models.

Table 2. ANN models’ specifications.

Model	Hidden Layers	Batch Size	Epochs
$\mathrm{P}{\mathrm{b}}^{2+}$ adsorption	(32, 64, 128, 128, 64, 32)	4	35,000
$\mathrm{C}{\mathrm{d}}^{2+}$ adsorption	(32, 64, 128, 64, 32)	8	25,000

The hyperparameter tuning is a method with which the contributing factors, e.g., batch size, number of epochs, and activation functions, can be optimized. The results for the $\mathrm{P}{\mathrm{b}}^{+2}$ adsorption are presented in Figure 10. We used 27 data points to evaluate the model. The input parameters for this model are the amount of absorbent, time, and pH. The results show that the model is very accurate, and it has an MAE of 1.01% and an ${\mathrm{R}}^2$ of 0.99.

Figure 10. The ANN model for $\mathrm{P}{\mathrm{b}}^{2+}$ adsorption.

Similarly, the model for the $\mathrm{C}{\mathrm{d}}^{2+}$ adsorption is presented in Figure 11. The results are also very accurate. This model shows an MAE of 1.21%. The R-squared of the model is also 0.98. The use of machine learning algorithms has proven to be very accurate and efficient. It should be noted that the computational costs of the machine learning algorithms are not even comparable to those of the numerical models, making them very cost-effective.

Figure 11. The ANN model for $\mathrm{C}{\mathrm{d}}^{2+}$ adsorption.

5. Conclusions

Due to the importance of removing heavy metal elements from aqueous solutions, various studies have been conducted in this field. Heavy metals pollute the water that humans and other living beings consume and cause severe pollution of soil and agricultural lands. Using mesoporous materials makes it possible to remove heavy metals such as lead and cadmium. We have delved deep into the details of adsorption efficiency in order to pinpoint the ideal conditions for effective water purification from metal ions. Our investigation meticulously explored the impacts of varying factors, such as adsorbent quantity, adsorption time, and pH levels, all in pursuit of achieving optimal performance. Taking a closer look at the influence of adsorbent quantity, we discovered specific optimal amounts for the efficient adsorption of $\mathrm{P}{\mathrm{b}}^{2+}$ and $\mathrm{C}{\mathrm{d}}^{2+}$ ions. It turned out that for MCM-41, ${\mathrm{N}\mathrm{H}}_2-\mathrm{M}\mathrm{C}\mathrm{M}-41$, SBA-15, and ${\mathrm{N}\mathrm{H}}_2-\mathrm{S}\mathrm{B}\mathrm{A}-15$, 30 mg was the magic number. This amount seemed to trigger enhanced active surface availability, leading to a more robust adsorption process, with an interesting exception in the case of NH₂-SBA-15, where things became more intricate. The presence of amino groups further played a role in boosting adsorption efficiency thanks to their knack for forming hybrid materials. Shifting our focus to adsorption time, we conducted meticulous stirring experiments within a 5- to 25-min window to find the optimal duration for solutions containing 2 ppm of $\mathrm{P}{\mathrm{b}}^{2+}$ and $\mathrm{C}{\mathrm{d}}^{2+}$ ions. The outcomes highlighted time ranges for each material that led to notable efficiency gains. The basis of success was extending the interaction time between our adsorbents and the ions, which directly correlated with improved adsorption efficiency. We also delved into the intriguing influence of pH on the adsorption process. We found that, within the pH range of 6–8, our samples exhibited top-notch quantitative adsorption. As the pH levels dropped, the amine groups on the adsorbent’s surface took on a protonated form, which affected their ability to form complexes with metal ions. As a result, adsorption efficiency dwindled at lower pH levels. Furthermore, we explored the resilience and reusability of our tailored adsorbents. Through multiple rounds of adsorption and reuse, we observed minimal shifts in adsorption percentages, reaffirming the robustness and consistency of our materials. X-ray diffraction analysis added another layer of proof, showcasing that the structural integrity remained intact post adsorption. This characteristic can be attributed to the surface area of the metal ions between the pores. The XRD pattern of the recycled piece of ${\mathrm{N}\mathrm{H}}_2$-MCM-41 also shows that the mesoporous structure is preserved after adsorption. Finally, by using the artificial neural networks, we proposed models for predicting the percentage of adsorption for both $\mathrm{P}{\mathrm{b}}^{2+}$ and $\mathrm{C}{\mathrm{d}}^{2+}$. The results of the ANN were very satisfactory, with an MAE of less than 2%. The computational cost of these models is also very efficient compared to both experimental and numerical models.