*Article* **Green Chemistry for the Transformation of Chlorinated Wastes: Catalytic Hydrodechlorination on Pd-Ni and Pd-Fe Bimetallic Catalysts Supported on SiO<sup>2</sup>**

**Julien G. Mahy 1,2,\*, Thierry Delbeuck <sup>1</sup> , Kim Yên Tran <sup>1</sup> , Benoît Heinrichs <sup>1</sup> and Stéphanie D. Lambert <sup>1</sup>**


**Abstract:** Monometallic catalysts based on Fe, Ni and Pd, as well as bimetallic catalysts based on Fe-Pd and based on Ni-Pd supported on silica, were synthesized using a sol–gel cogelation process. These catalysts were tested in chlorobenzene hydrodechlorination at low conversion to consider a differential reactor. In all samples, the cogelation method allowed very small metallic nanoparticles of 2–3 nm to be dispersed inside the silica matrix. Nevertheless, the presence of some large particles of pure Pd was noted. The catalysts had specific surface areas between 100 and 400 m2/g. In view of the catalytic results obtained, the Pd-Ni catalysts are less active than the monometallic Pd catalyst (<6% of conversion) except for catalysts with a low proportion of Ni (9% of conversion) and for reaction temperatures above 240 ◦C. In this series of catalysts, increasing the Ni content increases the activity but leads to an amplification of the catalyst deactivation phenomenon compared to Pd alone. On the other hand, Pd-Fe catalysts are more active with a double conversion value compared to a Pd monometallic catalyst (13% vs. 6%). The difference in the results obtained for each of the catalysts in the Pd-Fe series could be explained by the greater presence of the Fe-Pd alloy in the catalyst. Fe would have a cooperative effect when associated with Pd. Although Fe is inactive alone for chlorobenzene hydrodechlorination, when Fe is coupled to another metal from the group VIIIb, such as Pd, it allows the phenomenon of Pd poisoning by HCl to be reduced.

**Keywords:** noble metal; environmental catalysis; sol–gel process; porous materials; green chemistry

### **1. Introduction**

Over time, the presence of chlorinated organic compounds in industry increased, both in the chemical and petrochemical sectors and in the agricultural and electronic sectors. However, the discovery of the carcinogenic, mutagenic and toxic effects of these compounds led to their production being reduced and then banned. These products are nevertheless present in residual stocks and in certain chemical industry process fumes [1,2]. It is therefore necessary to develop treatment methods which allow their emissions, as well as the quantities of these compounds which have already been produced, to be eliminated.

Thermal or catalytic incineration [2,3] is the most widespread process for the treatment of these compounds because it is already in place for other wastes and is therefore wellknown and simple to implement. However, the incineration of organochlorine compounds leads to the formation of other harmful compounds such as dioxins. The development of other methods is therefore necessary.

Hydrodechlorination (HDC) seems to be an interesting alternative. This process, the principle of which can be summarized by Equation (1) [4], has the advantages of producing recoverable hydrocarbons, of recovering chlorine in the form of hydrochloric acid and of

**Citation:** Mahy, J.G.; Delbeuck, T.; Tran, K.Y.; Heinrichs, B.; Lambert, S.D. Green Chemistry for the Transformation of Chlorinated Wastes: Catalytic Hydrodechlorination on Pd-Ni and Pd-Fe Bimetallic Catalysts Supported on SiO2. *Gels* **2023**, *9*, 275. https:// doi.org/10.3390/gels9040275

Academic Editors: Francesco Caridi, Giuseppe Paladini and Andrea Fiorati

Received: 3 March 2023 Revised: 21 March 2023 Accepted: 23 March 2023 Published: 25 March 2023

**Copyright:** © 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

not producing dioxins. This process could therefore prove to be attractive from an economic and environmental point of view [2,5].

$$\rm{C\_xH\_yCl\_z + z\ H\_2 \to C\_xH\_{y+z} + z\ HCl} \tag{1}$$

There are a wide variety of chlorinated organic compounds [6–8]. This work focuses on the study of the hydrodechlorination of chlorinated aromatic compounds because these are the most widespread and the most persistent. The study of the HDC reaction for this family of compounds is made through the use of a model molecule, monochlorobenzene [9–11], C6H5Cl, which is the simplest molecule in this family of compounds. The equation corresponding to the reaction carried out is:

$$\rm C\_6H\_5Cl + H\_2 \to C\_6H\_6 + HCl.\tag{2}$$

This reaction must be controlled well to avoid the formation of cyclohexane. Otherwise, the need for a complex separation would arise. In addition, benzene is of much greater economic interest. High selectivity of the monochlorobenzene hydrodechlorination reaction is therefore a very important criterion in this study.

The use of a catalyst is necessary for this reaction to happen. Palladium has been identified as the most active metal for the HDC reaction of chlorobenzene [8,11–13]. However, palladium presents a major problem: the palladium-based catalyst is deactivated by adsorption of the HCl produced during the HDC reaction [14]. Nickel is the metal most often cited after palladium for the HDC of chlorinated compounds [2,11,15,16]. Nickel is less active than palladium but has the advantage of being less sensitive than the latter to deactivation [17]. Some authors have observed an increase in activity and/or stability, i.e., resistance to deactivation, when nickel and palladium are alloyed with other metals such as iron [5]. It is known that certain bimetallic catalysts exhibit superior performance to monometallic catalysts [18–20]. Many authors [12,21,22] demonstrated that the selectivity of non-destructive catalytic reactions is often higher for a bimetallic catalyst than a monometallic catalyst. Additionally, bimetallic catalysts are often more resistant to poisoning than monometallic catalysts [11,14]. Among the supports used, carbon [23,24], alumina [11,14] and silica [10,17] are frequently encountered. Only the latter two allow oxidative regeneration of spent catalysts.

There are several existing methods of synthesizing bimetallic catalysts [12]. Among those, the sol–gel method of cogelation allows the synthesis of the support and the dispersion of the precursors of the active metal sites in the porosity of this support to be combined in a single step to produce Ni-Cu/SiO<sup>2</sup> [22], Cu/SiO<sup>2</sup> [25], Pd/SiO<sup>2</sup> [26–28] and Fe/SiO<sup>2</sup> [28–30]. The sol–gel method is a synthetic process carried out at a low temperature and at a low pressure which leads to the formation of engineered and controlled ceramic nanopowders. This approach could be successfully employed to fabricate oxide, non-oxide and composite nanopowders with high purity [31]. Previous studies which characterized catalysts prepared using this method have demonstrated the method's remarkable properties [25,26,28,32]: very high dispersion of the active metallic phase, formation of nanoalloys, localization of metallic nanoparticles in silica cages allowing sintering of the metal to be avoided while keeping it perfectly accessible, hierarchical porous structure allowing easy distribution of reactants and products and the possible regeneration of the catalyst.

The objective of this work is to evaluate the extent to which the sol–gel method of cogelation for the synthesis of catalysts for the HDC of monochlorobenzene is of interest. The novelty resides in the production of Pd-Ni/SiO<sup>2</sup> and Pd-Fe/SiO<sup>2</sup> catalysts in one step by the cogelation process, something that is not reported in the literature. Different Pd-Ni and Pd-Fe bimetallic catalysts, supported on silica and synthesized by cogelation, were studied and their performances were evaluated. Then, the performances of the catalysts were related to their physico-chemical properties. A series of characterizations were carried out, including X-ray diffraction, transmission electron microscopy, measurement of nitrogen adsorption–desorption isotherms and ICP. Bimetallic catalysts Pd-Ni/SiO<sup>2</sup> and Pd-Fe/SiO<sup>2</sup>

and monometallic catalysts Fe/SiO2, Ni/SiO<sup>2</sup> and Pd/SiO<sup>2</sup> were be synthesized. For each bimetallic catalyst formulation, a series of catalysts with different metal contents was studied, the Pd content remained constant and the value of the second metal varied. Thus, the influence of the presence of the second metal can be studied.

### **2. Results and Discussion**

The amounts of all reagents used for material synthesis are detailed in Table 1.


**Table 1.** Operating synthesis variables for the catalysts Pd-Fe/SiO<sup>2</sup> et Pd-Ni/SiO<sup>2</sup> .

<sup>a</sup> = measured by ICP-AES.

### *2.1. Synthesis*

For all catalysts, no weight loss could be observed during the different steps of synthesis, calcination and reduction, and the wt.% of metal was as expected in each material (Table 1). Therefore, all the reagents reacted as was previously observed with this kind of sol–gel synthesis [26,28].

The gel time for all samples is given in Table 2. The gel time is the time which elapsed between the introduction of the last reagent into the solution and gelation in the oven at 80 ◦C. Gelling is defined as the moment when the liquid no longer flows when the bottle is tilted at an angle of 45◦ .

**Table 2.** Physico-chemical properties of samples Pd-Fe/SiO<sup>2</sup> and Pd-Ni/SiO<sup>2</sup> .


<sup>a</sup> = measured with Scherrer equation [33].

As described in [28,34], EDAS reacts faster than TEOS for the hydrolysis and condensation reactions. The evolution of the gel time observed in Table 2 follows this trend.

Indeed, for example, in the Pd-Fe series the gel time increases from 22 min to 30 min when the EDAS/TEOS ratio decreases from 0.075 to 0.031. The impact of the EDAS on the catalyst texture will be discussed later when analyzing the specific surface area results (see Section 2.4). deed, for example, in the Pd-Fe series the gel time increases from 22 min to 30 min when the EDAS/TEOS ratio decreases from 0.075 to 0.031. The impact of the EDAS on the catalyst texture will be discussed later when analyzing the specific surface area results (see Section 2.4).

As described in [28,34], EDAS reacts faster than TEOS for the hydrolysis and condensation reactions. The evolution of the gel time observed in Table 2 follows this trend. In-

*Gels* **2023**, *9*, x FOR PEER REVIEW 4 of 20

#### *2.2. Composition of the Catalysts 2.2. Composition of the Catalysts*

The ICP results (Table 1) showed that the catalysts were well doped with Pd, Ni and Fe but it does not give information about the nature of these species. XRD measurements were performed to determine the phases of the metallic species. Figures 1 and 2 represent all the XRD patterns of the samples. The ICP results (Table 1) showed that the catalysts were well doped with Pd, Ni and Fe but it does not give information about the nature of these species. XRD measurements were performed to determine the phases of the metallic species. Figures 1 and 2 represent all the XRD patterns of the samples.

**Figure 1.** XRD patterns of (**a**) Ni100, (**b**) Pd33Ni67, (**c**) Pd50Ni50, (**d**) Pd67Ni33 and (**e**) Pd100 samples. Reference peaks corresponding to Pd (JCPD No. 46-1043 [24]), Ni (JCPD No. 04-0850 [16]) and Pd-Ni alloy (JCPD No. 65-9444 [35]) are denoted on the figures. **Figure 1.** XRD patterns of (**a**) Ni100, (**b**) Pd33Ni67, (**c**) Pd50Ni50, (**d**) Pd67Ni33 and (**e**) Pd100 samples. Reference peaks corresponding to Pd (JCPD No. 46-1043 [24]), Ni (JCPD No. 04-0850 [16]) and Pd-Ni alloy (JCPD No. 65-9444 [35]) are denoted on the figures.

In Figure 1, the Pd-Ni/SiO<sup>2</sup> series is represented. In Figure 1a, which represents the Ni100 catalyst, two weak crystallographic reflections are observed coinciding with two of the characteristic lines of Ni at 44.41◦ and 51.90◦ . The low intensities of these peaks and their spreading could be explained by two factors: (i) the very small size of the metallic particles; (ii) the presence of amorphous Ni oxide or hydroxide which cannot be detected. In Figure 1b–d which illustrates the Pd-Ni/SiO<sup>2</sup> catalysts, two crystallographic reflections characteristic of Pd are observed, but none of Ni. The presence of a Pd-Ni alloy would be marked by the presence of a crystallographic reflection between the characteristic lines of Pd and Ni. Their absence at the location of the characteristic peaks of the Pd-Ni alloy and Ni does not mean these two compounds are absent. Indeed, seeing the crystallographic reflection obtained in Figure 1a for the Ni100 sample, one can easily conceive that, with such a weak signal, it is very difficult to detect this compound. Moreover, assuming that a

Pd-Ni alloy has been formed, the signal would be even weaker and would more than likely be confused with the background noise. *Gels* **2023**, *9*, x FOR PEER REVIEW 5 of 20

**Figure 2.** XRD patterns of (**a**) Fe100, (**b**) Pd33Fe67, (**c**) Pd50Fe50 and (**d**) Pd67Fe33 samples. Reference peaks corresponding to Pd are denoted on the figures. **Figure 2.** XRD patterns of (**a**) Fe100, (**b**) Pd33Fe67, (**c**) Pd50Fe50 and (**d**) Pd67Fe33 samples. Reference peaks corresponding to Pd are denoted on the figures.

In Figure 1, the Pd-Ni/SiO2 series is represented. In Figure 1a, which represents the Ni100 catalyst, two weak crystallographic reflections are observed coinciding with two of the characteristic lines of Ni at 44.41° and 51.90°. The low intensities of these peaks and their spreading could be explained by two factors: (i) the very small size of the metallic particles; (ii) the presence of amorphous Ni oxide or hydroxide which cannot be detected. As regards the Pd-Fe/SiO<sup>2</sup> series of catalysts, the results obtained are almost identical to the Pd-Ni/SiO<sup>2</sup> series (Figure 2), except that no compound is detected in the case of the Fe100 sample. As a reminder, the diffractograms in this series extend from 2θ = 0◦ to 80◦ . The crystallographic reflections present for angles of less than approximately 40◦ are due to the support, namely, the silica. The diffractogram corresponding to Fe100 is shown in Figure 2a.

In Figure 1b–d which illustrates the Pd-Ni/SiO2 catalysts, two crystallographic reflections characteristic of Pd are observed, but none of Ni. The presence of a Pd-Ni alloy would be marked by the presence of a crystallographic reflection between the characteristic lines of Pd and Ni. Their absence at the location of the characteristic peaks of the Pd-Ni alloy and Ni does not mean these two compounds are absent. Indeed, seeing the crystallographic reflection obtained in Figure 1a for the Ni100 sample, one can easily conceive that, with such a weak signal, it is very difficult to detect this compound. Moreover, assuming that a Pd-Ni alloy has been formed, the signal would be even weaker and would more than likely be confused with the background noise. As regards the Pd-Fe/SiO2 series of catalysts, the results obtained are almost identical Although a signal can be detected in the case of the Ni100 sample (Figure 1a), no signal is detected in the case of the Fe100 sample (Figure 2a). Two possibilities could explain this phenomenon: (i) the size of the iron particles is very small; (ii) the presence of amorphous iron oxide or hydroxide which cannot be detected. If, with Ni it could already be conceived that the presence of an alloy was possible, with Fe it can be conceived even more easily that a Pd-Fe alloy could be present. Indeed, since Fe was not even detected on the diffractogram for the Fe100 sample (Figure 2a), it is more than likely that if a Pd-Fe alloy existed in the catalysts synthesized for this study, it would not be detected either. Moreover, it seems obvious that if the Fe is not detected in the Fe100 catalyst, it will also not be detected on the other catalysts in the Pd-Fe series, as can be seen on the diffractograms (Figure 2b–d).

to the Pd-Ni/SiO2 series (Figure 2), except that no compound is detected in the case of the Fe100 sample. As a reminder, the diffractograms in this series extend from 2θ = 0° to 80°. The crystallographic reflections present for angles of less than approximately 40° are due to the support, namely, the silica. The diffractogram corresponding to Fe100 is shown in Figure 2a. The size of the metallic particles (dXRD) can be estimated directly from the diffractograms of the catalysts by applying Scherrer's formula based on the width of the peaks (Equation (3)). The particle sizes of the different catalysts are listed in Table 2 for all the samples. The size is in the range of 5–15 nm for the Pd-Ni/SiO<sup>2</sup> series and around 20–25 nm for the Pd-Fe/SiO<sup>2</sup> series.

Although a signal can be detected in the case of the Ni100 sample (Figure 1a), no signal is detected in the case of the Fe100 sample (Figure 2a). Two possibilities could explain this phenomenon: (i) the size of the iron particles is very small; (ii) the presence of amorphous iron oxide or hydroxide which cannot be detected. If, with Ni it could already

### *2.3. Morphology*

Micrographs of some Pd-Ni/SiO<sup>2</sup> and Pd-Fe/SiO<sup>2</sup> catalysts are presented in Figure 3. For most of the samples, similar morphology is obtained with bigger particles (corresponding to silica particles around 20–30 nm) with smaller darker particles inside (metallic particles around 1–3 nm). With the TEM micrographs, one can measure the distribution of the metal particles within the silica matrix and obtain their average metallic size (dTEM), as denoted in Table 2. In these images, the metallic particles are generally fairly well distinguishable (zoom on Figure 3b). By measuring the different particles, a mean particle size can be calculated, dTEM. For Ni100 (Figure 3a) and Fe100 samples, no metallic nanoparticle is observed inside the silica particle. The presence of oxide in these catalysts, despite the reduction, cannot be ruled out in view of the results obtained using X-ray diffraction. As the contrast is much worse on the TEM images for the oxides, it can be deduced that it would not be possible to distinguish the oxides, which could be one of the explanations for the lack of visible particles for these catalysts. For the other samples, the metallic particle sizes are between 1.5 and 2.6 nm. *Gels* **2023**, *9*, x FOR PEER REVIEW 7 of 20

**Figure 3.** TEM micrographs of (**a**) Ni100, (**b**) Pd33Ni67 and (**c**) Pd50Fe50 samples.

It is observed that the average sizes obtained by this method are very different from the sizes determined by X-ray diffraction (Table 2). This can be explained by the presence of very large particles which are relatively rare on the images as can be seen in Figure 3c (red square). cases to a type IV isotherm (macroporous adsorbents [36]). **Figure 3.** TEM micrographs of (**a**) Ni100, (**b**) Pd33Ni67 and (**c**) Pd50Fe50 samples. It is observed that the average sizes obtained by this method are very different from the sizes determined by X-ray diffraction (Table 2). This can be explained by the presence of very large particles which are relatively rare on the images as can be seen in Figure 3c (red square).

### *2.4. Texture 2.4. Texture*

The specific surface area for each sample has been calculated and is denoted in Table 2. Two nitrogen adsorption–desorption isotherms are represented in Figure 4 for the Fe100 and Pd50Ni50 samples, illustrating the two shapes observed for all samples. Indeed, for all samples except for the Fe100 catalyst, the isotherms look like Figure 4b with a sudden The specific surface area for each sample has been calculated and is denoted in Table 2. Two nitrogen adsorption–desorption isotherms are represented in Figure 4 for the Fe100 and Pd50Ni50 samples, illustrating the two shapes observed for all samples. Indeed, for all samples except for the Fe100 catalyst, the isotherms look like Figure 4b with a sudden

increase in adsorbed volume at low relative pressures followed by a plateau, characteristic of microporous adsorbents (type I [36]). In the case of the Fe100 sample (Figure 4a), at low

increase in adsorbed volume at low relative pressures followed by a plateau, characteristic of microporous adsorbents (type I [36]). In the case of the Fe100 sample (Figure 4a), at low relative pressures no sudden increase in absorbed volume is observed, which testifies to the absence of micropores. At relative pressures close to 1, the isotherms correspond in all cases to a type IV isotherm (macroporous adsorbents [36]). *Gels* **2023**, *9*, x FOR PEER REVIEW 8 of 20

**Figure 4.** Nitrogen adsorption–desorption isotherms for (**a**) Fe100 and (**b**) Pd50Ni50 samples. **Figure 4.** Nitrogen adsorption–desorption isotherms for (**a**) Fe100 and (**b**) Pd50Ni50 samples.

All the isotherms present a hysteresis characteristic of the phenomenon of capillary All the isotherms present a hysteresis characteristic of the phenomenon of capillary condensation in the mesopores.

condensation in the mesopores. For the SBET values, it can be observed that in each series the specific surface area increases with the EDAS/TEOS ratio as previously explained in [26,28,29,34]. This is due to the role of the nucleating agent EDAS which reacts faster than TEOS (see gel time in Table 2 and Section 2.1) and is therefore the nucleating agent of the silica particles. As was previously described in [26,28,29,34], when the amount of EDAS increases it leads to a higher specific surface area. For example, in the Pd-Fe/SiO2 series (Table 2), when the EDAS/TEOS ratio is equal to 0.031 for the Pd67Fe33 sample, the SBET value is 300 m2/g, while when the EDAS/TEOS ratio increases to 0.075 for the Pd33Fe67 sample, the SBET is 395 m2/g. For the SBET values, it can be observed that in each series the specific surface area increases with the EDAS/TEOS ratio as previously explained in [26,28,29,34]. This is due to the role of the nucleating agent EDAS which reacts faster than TEOS (see gel time in Table 2 and Section 2.1) and is therefore the nucleating agent of the silica particles. As was previously described in [26,28,29,34], when the amount of EDAS increases it leads to a higher specific surface area. For example, in the Pd-Fe/SiO<sup>2</sup> series (Table 2), when the EDAS/TEOS ratio is equal to 0.031 for the Pd67Fe33 sample, the SBET value is 300 m2/g, while when the EDAS/TEOS ratio increases to 0.075 for the Pd33Fe67 sample, the SBET is 395 m2/g.

#### *2.5. Catalytic Activity 2.5. Catalytic Activity*

It should be noted that the benzene selectivity is equal to 100% for each catalyst evaluated in this study. It should be noted that the benzene selectivity is equal to 100% for each catalyst evaluated in this study.

#### 2.5.1. Pd-Ni/SiO<sup>2</sup> Series 2.5.1. Pd-Ni/SiO2 Series

Figure 5 shows the conversion of chlorobenzene to benzene as a function of time for the different catalysts in the Pd-Ni/SiO<sup>2</sup> series. The mean conversion values on the plateau at 220 ◦C are denoted in Table 3 for all samples. Figure 5 shows the conversion of chlorobenzene to benzene as a function of time for the different catalysts in the Pd-Ni/SiO2 series. The mean conversion values on the plateau at 220 °C are denoted in Table 3 for all samples.

**Figure 5.** Comparison of the conversions of chlorobenzene to benzene as a function of time for the Pd-Ni/SiO2 series. **Figure 5.** Comparison of the conversions of chlorobenzene to benzene as a function of time for the Pd-Ni/SiO<sup>2</sup> series.


**Table 3.** Mean chlorobenzene conversion at 220 °C. **Table 3.** Mean chlorobenzene conversion at 220 ◦C.

It must be remembered that the mass of the catalyst for the Pd-Ni/SiO2 series is 0.2 g compared to the Pd-Fe series and the Pd monometallic sample where it is 0.1 g. This difference in mass is necessary for the Pd-Ni/SiO2 series as otherwise the initial conversion is It must be remembered that the mass of the catalyst for the Pd-Ni/SiO<sup>2</sup> series is 0.2 g compared to the Pd-Fe series and the Pd monometallic sample where it is 0.1 g. This difference in mass is necessary for the Pd-Ni/SiO<sup>2</sup> series as otherwise the initial conversion is too high and leads to a rapid deactivation.

too high and leads to a rapid deactivation. There are two different types of behavior in this same series: (i) a relatively constant activity on each level, that is to say a fairly low deactivation. This behavior is characteristic for Pd100 and Pd67Ni33 catalysts; (ii) a strong deactivation. The increase in temperature increases the activity of the catalyst. The deactivation is such that this increase does not There are two different types of behavior in this same series: (i) a relatively constant activity on each level, that is to say a fairly low deactivation. This behavior is characteristic for Pd100 and Pd67Ni33 catalysts; (ii) a strong deactivation. The increase in temperature increases the activity of the catalyst. The deactivation is such that this increase does not make it possible to obtain a better conversion than the maximum conversion at the previous level. This behavior is characteristic of the Pd33Ni67 and Pd50Ni50 catalysts. ous level. This behavior is characteristic of the Pd33Ni67 and Pd50Ni50 catalysts. Note that the Ni100 catalyst is not shown in Figure 5. In fact, the catalytic experiment

make it possible to obtain a better conversion than the maximum conversion at the previ-

*Gels* **2023**, *9*, x FOR PEER REVIEW 10 of 20

Note that the Ni100 catalyst is not shown in Figure 5. In fact, the catalytic experiment carried out on this catalyst revealed virtually zero activity throughout the test, despite the temperature variations (less than 1%). carried out on this catalyst revealed virtually zero activity throughout the test, despite the temperature variations (less than 1%). Taking into account the mass difference introduced into the reactor between the Pd-

Taking into account the mass difference introduced into the reactor between the Pd-Ni/SiO<sup>2</sup> series and the monometallic Pd catalyst, it is observed that the presence of Ni in the catalyst does not improve the conversion of chlorobenzene into benzene. Indeed, given that in the case of the Pd100 sample a mass of catalyst twice as small as the mass introduced into the reactor for the Pd-Ni/SiO<sup>2</sup> series samples was introduced into the reactor, one should obtain a significantly lower conversion of chlorobenzene to benzene than in the case of bimetallic catalysts. However, in Figure 5, it is observed that the conversion values of chlorobenzene into benzene for the Pd100 sample are similar to those obtained for the Pd50Ni50 sample, whereas the mass present in the reactor for the Pd100 catalyst is double that for the Pd50Ni50 sample. Ni/SiO2 series and the monometallic Pd catalyst, it is observed that the presence of Ni in the catalyst does not improve the conversion of chlorobenzene into benzene. Indeed, given that in the case of the Pd100 sample a mass of catalyst twice as small as the mass introduced into the reactor for the Pd-Ni/SiO2 series samples was introduced into the reactor, one should obtain a significantly lower conversion of chlorobenzene to benzene than in the case of bimetallic catalysts. However, in Figure 5, it is observed that the conversion values of chlorobenzene into benzene for the Pd100 sample are similar to those obtained for the Pd50Ni50 sample, whereas the mass present in the reactor for the Pd100 catalyst is double that for the Pd50Ni50 sample. When the nickel content increases in the bimetallic catalysts, the deactivation of the

When the nickel content increases in the bimetallic catalysts, the deactivation of the bimetallic catalyst takes place more and more rapidly. Therefore, the Pd67Ni33 catalyst is the most attractive of the Pd-Ni/SiO<sup>2</sup> catalysts tested. Seeing the increase in conversion with temperature, it could be thought that this catalyst would only become more interesting than the monometallic Pd catalyst at higher temperatures. bimetallic catalyst takes place more and more rapidly. Therefore, the Pd67Ni33 catalyst is the most attractive of the Pd-Ni/SiO2 catalysts tested. Seeing the increase in conversion with temperature, it could be thought that this catalyst would only become more interesting than the monometallic Pd catalyst at higher temperatures. The differences in activity and behavior between the different Pd-Ni/SiO2 catalysts

The differences in activity and behavior between the different Pd-Ni/SiO<sup>2</sup> catalysts could be the consequence of the formation of different alloys, the nature of which differs according to the quantity of Ni present in the bimetallic catalyst. could be the consequence of the formation of different alloys, the nature of which differs according to the quantity of Ni present in the bimetallic catalyst. According to the phase diagram of the Pd-Ni alloy (Figure 6), at the temperatures of

According to the phase diagram of the Pd-Ni alloy (Figure 6), at the temperatures of the hydrodechlorination reaction of chlorobenzene to benzene (between 453 K and 513 K), Pd and Ni form a solid solution, and this is the case in the entire range of concentration between pure Pd and pure Ni. According to this phase diagram, the Pd-Ni/SiO<sup>2</sup> catalysts studied in this work therefore form an alloy with an FCC structure (face centered cubic). However, since the metal particles present in the Pd-Ni/SiO<sup>2</sup> catalysts are very small in size (dTEM, Table 2), the metals may not follow exactly the same phase diagram [37]. Nevertheless, in Figure 1, the main palladium lines appear. According to Scherrer's formula as applied to these peaks (Equation (3)), particle sizes between 4 and 13 nm are obtained (dXRD, Table 2). It could therefore be that the large particles observed using TEM (Figure 3c) are particles of pure palladium. It is also hypothesized that the small particles which are not detectable by using X-ray diffraction consist of a mixture of the two metals, Pd and Ni. the hydrodechlorination reaction of chlorobenzene to benzene (between 453 K and 513 K), Pd and Ni form a solid solution, and this is the case in the entire range of concentration between pure Pd and pure Ni. According to this phase diagram, the Pd-Ni/SiO2 catalysts studied in this work therefore form an alloy with an FCC structure (face centered cubic). However, since the metal particles present in the Pd-Ni/SiO2 catalysts are very small in size (dTEM,Table 2), the metals may not follow exactly the same phase diagram [37]. Nevertheless, in Figure 1, the main palladium lines appear. According to Scherrer's formula as applied to these peaks (Equation (3)), particle sizes between 4 and 13 nm are obtained (dXRD, Table 2). It could therefore be that the large particles observed using TEM (Figure 3c) are particles of pure palladium. It is also hypothesized that the small particles which are not detectable by using X-ray diffraction consist of a mixture of the two metals, Pd and Ni.

**Figure 6.** Phase diagram of the Pd-Ni alloy, reproduced from [38]. **Figure 6.** Phase diagram of the Pd-Ni alloy, reproduced from [38].

The Pd-Ni alloy is an alloy that has been studied very little in the literature on the hydrodechlorination reaction of chlorinated compounds. These catalysts are reported, as is the case with this study, to provide good activity, which, however, is lower than in the case of Pd-Fe catalysts. However, Pd-Ni catalysts have a selectivity close to or equal to 100% for the hydrodechlorination of chlorinated compounds [12,39]. The Pd-Ni alloy is an alloy that has been studied very little in the literature on the hydrodechlorination reaction of chlorinated compounds. These catalysts are reported, as is the case with this study, to provide good activity, which, however, is lower than in the case of Pd-Fe catalysts. However, Pd-Ni catalysts have a selectivity close to or equal to 100% for the hydrodechlorination of chlorinated compounds [12,39]. An explanation for the better activity of catalysts containing a large amount of Ni has

An explanation for the better activity of catalysts containing a large amount of Ni has been given by Simagina et al. [39]. The cause may be a Pd segregation phenomenon on the surface of the bimetallic particles. At a high Ni concentration, the Pd is essentially found isolated in small aggregates on the surface of the Ni-rich particles and is therefore more active than in a case where it is found in larger aggregates. Indeed, according to Sr˛ebowata et al. [ ´ 40], there would be a segregation of Pd at the surface of the Pd-Ni particles. been given by Simagina et al. [39]. The cause may be a Pd segregation phenomenon on the surface of the bimetallic particles. At a high Ni concentration, the Pd is essentially found isolated in small aggregates on the surface of the Ni-rich particles and is therefore more active than in a case where it is found in larger aggregates. Indeed, according to Śrębowata et al. [40], there would be a segregation of Pd at the surface of the Pd-Ni particles.

The deactivation of catalysts containing Ni has already been observed in previous works [2,41]. The cause of this deactivation seems to be the adsorption of HCl. The adsorbed HCl modifies the cubic crystalline structure of Ni to form a hexagonal structure of the NiCl<sup>2</sup> type which induces an irreversible deactivation of the catalyst. The deactivation of Pd-based catalysts supported on alumina is usually caused by coking, which leads to a rapid decrease in catalytic activity [12,42]. Nevertheless, in this work it can be assumed that coking is a minor phenomenon on the silica support [12,19] and that catalyst deactivation is mainly due to Ni deactivation by chlorine atoms. This would explain the increased deactivation of the catalysts when the Ni content increases. The deactivation of catalysts containing Ni has already been observed in previous works [2,41]. The cause of this deactivation seems to be the adsorption of HCl. The adsorbed HCl modifies the cubic crystalline structure of Ni to form a hexagonal structure of the NiCl2 type which induces an irreversible deactivation of the catalyst. The deactivation of Pd-based catalysts supported on alumina is usually caused by coking, which leads to a rapid decrease in catalytic activity [12,42]. Nevertheless, in this work it can be assumed that coking is a minor phenomenon on the silica support [12,19] and that catalyst deactivation is mainly due to Ni deactivation by chlorine atoms. This would explain the increased deactivation of the catalysts when the Ni content increases.

#### 2.5.2. Pd-Fe/SiO<sup>2</sup> Series 2.5.2. Pd-Fe/SiO2 Series

Figure 7 shows the conversion of chlorobenzene to benzene for the different catalysts tested in the Pd-Fe/SiO<sup>2</sup> series. The mean conversion values on the plateau at 220 ◦C are denoted in Table 3 for all samples. Figure 7 shows the conversion of chlorobenzene to benzene for the different catalysts tested in the Pd-Fe/SiO2 series. The mean conversion values on the plateau at 220 °C are denoted in Table 3 for all samples.

**Figure 7.** Comparison of the conversions of chlorobenzene to benzene as a function of time for the Pd-Fe/SiO2 series. **Figure 7.** Comparison of the conversions of chlorobenzene to benzene as a function of time for the Pd-Fe/SiO<sup>2</sup> series.

Again, the same two types of behavior observed for the Pd-Ni/SiO2 series are present, namely: (i) a relatively constant activity on each level, i.e., a fairly low deactivation. This behavior is characteristic of Pd100, Pd67Fe33 and Pd50Fe50 catalysts; (ii) a strong deactivation. The increase in temperature increases the activity of the catalyst. The deactivation is such that this increase does not make it possible to obtain a better conversion than the Again, the same two types of behavior observed for the Pd-Ni/SiO<sup>2</sup> series are present, namely: (i) a relatively constant activity on each level, i.e., a fairly low deactivation. This behavior is characteristic of Pd100, Pd67Fe33 and Pd50Fe50 catalysts; (ii) a strong deactivation. The increase in temperature increases the activity of the catalyst. The deactivation is such that this increase does not make it possible to obtain a better conversion than the maximum conversion at the previous level. This behavior is characteristic of the Pd33Fe67 catalyst, other than between the first two levels at 180 ◦C and 200 ◦C.

It can be noted that the Fe100 catalyst is not shown in Figure 7. Indeed, the catalytic experiment carried out on this catalyst revealed zero activity throughout the test despite the temperature variations. It can be noted that the Fe100 catalyst is not shown in Figure 7. Indeed, the catalytic experiment carried out on this catalyst revealed zero activity throughout the test despite the temperature variations.

maximum conversion at the previous level. This behavior is characteristic of the Pd33Fe67

*Gels* **2023**, *9*, x FOR PEER REVIEW 12 of 20

catalyst, other than between the first two levels at 180 °C and 200 °C.

In Figure 7, it is observed that the presence of Fe in the presence of Pd leads to different results depending on the quantity of Fe present in the bimetallic catalyst: (i) When compared to the Pd100 catalyst, the presence of a small amount of iron (Pd67Fe33) improves the performance of the catalyst without affecting its deactivation; (ii) the presence of a quantity of Fe equivalent to the quantity of Pd in the catalyst Pd50-Fe50 reduces catalytic performance at a low temperature (180 ◦C and 200 ◦C). The conversion values of chlorobenzene into benzene are only higher than those of the Pd100 sample from the third stage, namely 220 ◦C. Deactivation remains low; (iii) the presence of a greater quantity of Fe than of Pd (Pd33Fe67) improves catalytic performance, compared to the Pd100 and Pd67Fe33 catalysts. On the other hand, the deactivation of the catalyst is much stronger than in the other Pd-Fe/SiO<sup>2</sup> catalysts. In Figure 7, it is observed that the presence of Fe in the presence of Pd leads to different results depending on the quantity of Fe present in the bimetallic catalyst: (i) When compared to the Pd100 catalyst, the presence of a small amount of iron (Pd67Fe33) improves the performance of the catalyst without affecting its deactivation; (ii) the presence of a quantity of Fe equivalent to the quantity of Pd in the catalyst Pd50-Fe50 reduces catalytic performance at a low temperature (180 °C and 200 °C). The conversion values of chlorobenzene into benzene are only higher than those of the Pd100 sample from the third stage, namely 220 °C. Deactivation remains low; (iii) the presence of a greater quantity of Fe than of Pd (Pd33Fe67) improves catalytic performance, compared to the Pd100 and Pd67Fe33 catalysts. On the other hand, the deactivation of the catalyst is much stronger than in the other Pd-Fe/SiO2 catalysts.

According to the literature [43], the difference in the behavior of a catalyst with the presence of a second metal can be explained by the formation of alloys. The change in the electronic state of the catalyst would then be responsible for the modification of the catalytic properties [44]. According to the literature [43], the difference in the behavior of a catalyst with the presence of a second metal can be explained by the formation of alloys. The change in the electronic state of the catalyst would then be responsible for the modification of the catalytic properties [44].

On the Fe-Pd phase diagram [38] (Figure 8), by tracing a horizontal line at 400 ◦C (673 K), there are several possible alloys depending on the atomic percentage of Pd: a mixture of Fe + FePd, the FePd alloy or the FePd<sup>3</sup> alloy. It is hypothesized that the nature of the particles detected on micrographs (Figure 3) follows this phase diagram, although the metallic particles are very small in size [37]. On the Fe-Pd phase diagram [38] (Figure 8), by tracing a horizontal line at 400 °C (673 K), there are several possible alloys depending on the atomic percentage of Pd: a mixture of Fe + FePd, the FePd alloy or the FePd3 alloy. It is hypothesized that the nature of the particles detected on micrographs (Figure 3) follows this phase diagram, although the metallic particles are very small in size [37].

**Figure 8.** Phase diagram of the Pd-Fe alloy, reproduced from [38]. **Figure 8.** Phase diagram of the Pd-Fe alloy, reproduced from [38].

Pure palladium particles were detected by using XRD (Figure 2). According to Scherrer's formula (Equation (3)), the dimensions of these pure palladium particles are between 22 and 25 nm (Table 2). These particles would thus correspond to the large particles observed by TEM. As these are quite rare (Figure 3c), it is considered that the hydrodechlorination activity of Pd-Fe/SiO2 catalysts mainly result from the small metal particles (Table 2). Pure palladium particles were detected by using XRD (Figure 2). According to Scherrer's formula (Equation (3)), the dimensions of these pure palladium particles are between 22 and 25 nm (Table 2). These particles would thus correspond to the large particles observed by TEM. As these are quite rare (Figure 3c), it is considered that the hydrodechlorination activity of Pd-Fe/SiO<sup>2</sup> catalysts mainly result from the small metal particles (Table 2).

For a percentage close to 33 atomic % in Pd, which would correspond to the catalyst Pd33Fe67, the palladium and the iron are organized in a Fe + FePd mixture. The same For a percentage close to 33 atomic % in Pd, which would correspond to the catalyst Pd33Fe67, the palladium and the iron are organized in a Fe + FePd mixture. The same analysis can be carried out for the two other catalysts, Pd50Fe50 and Pd67Fe33, therefore corresponding, respectively, to 50% and 67% atomic Pd. In these two cases, the alloys present are in different proportions or in different natures. For the Pd50-Fe50 sample, the FePd alloy would be almost alone with a small proportion of monometallic Fe, while for the Pd67Fe33 sample a mixture of alloys with FePd and FePd<sup>3</sup> structure is present.

In view of these results (Figure 7), it seems that the FePd-type alloy only allows a better conversion of chlorobenzene into benzene at higher temperatures than for the Pd100 sample. The presence of iron in the unalloyed state, although not active if used alone in a catalyst, could greatly favor the conversion of chlorobenzene into benzene but nevertheless leads to accelerated deactivation of the catalyst. Finally, the FePd<sup>3</sup> structural alloy shows better activity than the FePd structural alloy while keeping a reasonable deactivation rate.

Zhang et al. [45] suggest an explanation for the roles played by the two metals. The Pd would act as a catalyst while the Fe would act as an electron donor (reducer). According to Kim et al. [46], the presence of Fe improves the activity of Pd-based catalysts. Tests revealed an alloy with an FePd structure. This alloy would be responsible for the improvement of the catalytic activity because of the cooperative effect of Fe during the hydrogen transfer step.

According to some authors [9,47,48], the particles of Pd-Fe catalysts would be enriched at the surface with Fe atoms and therefore the formation of FeCl<sup>3</sup> on the surface of the catalysts would be thermodynamically favored compared to the formation of PdCl2. In this case, the poisoning of Pd sites by HCl is reduced by the elimination of chlorine in the form of FeCl3. In this way, the stability of the bimetallic catalysts is higher than that of the monometallic catalyst based on Pd, as is the case here for the catalysts Pd67Fe33 and Pd50Fe50. Too high a quantity of Fe can nevertheless lead to a phenomenon of encapsulation of the Pd active sites by FeCl3. This would lead to lower stability than in the case of Pd alone, as is the case with Pd33Fe67 [43].

Lingaiah et al. [47] also highlight the importance of particle sizes in the context of the HDC of chlorobenzene. Low dispersion would improve the activity of the catalysts tested, thus demonstrating that the hydrodechlorination reaction of chlorobenzene is sensitive to the surface structure of the catalysts. In addition, interest in low dispersion would be based on an improvement in catalytic stability. Large Pd particles would be less prone to quenching by HCl. These results show that the process of drying using microwaves (MW) is of particular interest. This technique would favor the increase of particle sizes compared to traditional thermal drying methods. In the case of this present study, the metal particles remain small in size, and the activity remains high.

Lingaiah et al. [47] have also shown that even catalysts with high dispersion can exhibit activity and stability comparable to catalysts with low dispersion. Their findings also show a higher activity of monometallic Pd catalysts compared to those of bimetallic catalysts. However, Babu et al. [5] highlight the possibility of forming particular Pd species which are more active than others. It could therefore be thought that the species formed in this work are much more active in the case of bimetallic alloys than for monometallic Pd.

### *2.6. Comparison with the Literature*

In Table 4, the present study is compared to the literature by summarizing the most important parameters and results of each study. First, it can be observed that it is quite difficult to make a comparison because many conditions such as the temperature, the amount of the catalyst, or the H<sup>2</sup> fraction differ greatly from one study to another. From Table 4, it can be also seen that some studies do not give all the information needed for a proper comparison.

Second, it is observed that the catalysts developed in this paper allow a 100% selectivity at a reasonably low temperature (220 ◦C) with a dilute amount of H<sup>2</sup> (2.6 vol%). Conversion is lower, but the idea of this study was to stay at lower conversion in order to consider a differential reactor. The benzene selectivity is often missing in the other studies and, as expected, conversion increases when the temperature is higher.


**Table 4.** Comparison of the chlorobenzene (CB) catalytic activity with the literature.

### **3. Conclusions**

Two series of bimetallic catalysts, Pd-Ni/SiO<sup>2</sup> and Pd-Fe/SiO2, were synthesized by the sol–gel process for the catalysis of the hydrodechlorination of chlorobenzene into benzene. This synthesis process uses two silicon alkoxides, TEOS and EDAS. EDAS is a modified alkoxide with a diamine function allowing the metal to be highly dispersed in the silica matrix. In all cases, the sol–gel process allows very small metallic nanoparticles in the order of 2–3 nm to be dispersed inside the silica matrix. Nevertheless, the presence of some large particles of pure Pd in the analyzed catalysts can be noted. The catalysts have a specific surface between 100 and 400 m2/g.

All bimetallic catalysts can catalyze the hydrodechlorination of chlorobenzene. In order to consider a differential reactor, the work was carried out at low conversion conditions. The Pd-Ni combination is nevertheless less active than the Pd-Fe combination. Monometallic catalysts Ni/SiO<sup>2</sup> and Fe/SiO<sup>2</sup> are inactive for the hydrodechlorination reaction of chlorobenzene, with no conversion.

Pd-Ni/SiO<sup>2</sup> catalysts are less active than monometallic Pd catalyst (mean conversion of 6%). For catalysts with a low proportion of Ni such as Pd67Ni33 (mean conversion of 9%), one can only foresee a possible improvement in activity compared to that of the monometallic Pd catalyst for reaction temperatures above 240 ◦C. In this series of catalysts, increasing the Ni content increases the activity but leads to an amplification of the catalyst deactivation phenomenon compared to Pd alone (mean conversion drops to 4% with the Pd50Ni50 sample). The increase in activity could be due to an overall effect, i.e., the Ni would dilute the Pd into small aggregates, and thus increase the surface area of active Pd. The deactivation would be due to the formation of NiCl2.

The Pd-Fe catalysts are overall more active than the monometallic Pd catalyst. The difference in the results obtained for each of the catalysts in the series could be explained by the significant presence of FePd alloy in the catalyst. The combined action of Fe and Pd would be responsible for this increase in activity. Fe would have a cooperative effect when associated with Pd, although it is inactive alone. The presence of Fe could also reduce the phenomenon of Pd poisoning by HCl. The best Pd-Fe catalyst (Pd67Fe33) doubled the mean conversion to 13% compared to the Pd monometallic catalyst (mean conversion of 6%).

### **4. Materials and Methods**

### *4.1. Catalyst Synthesis*

Reagents: Fe(III) acetylacetonate (Fe(acac)3, (CH3COCH=C(O)CH3)3Fe, Sigma-Aldrich, 97%, St. Louis, MO, USA); Pd (II) acetylacetonate (Pd(acac)2, CH3COCH=C(O)CH3)2Pd, Sigma-Aldrich, 99%); Ni(II) acetylacetonate (Ni(acac)2, CH3COCH=C(O)CH3)2Ni, Sigma-Aldrich, 97%); Ammonia solution (NH4OH, Sigma-Aldrich, 25%); distilled water; tetraethoxysilane (TEOS, Si(OC2H5)4, Sigma-Aldrich, 98%); N-[3-(triméthoxysilyl)propyl]ethylenediamine (EDAS, (OCH3)3-Si-(CH2)3-NH-(CH2)2-NH, Sigma-Aldrich, 97%). The amounts of all reagents used for material synthesis are detailed in Table 1.

Pd(acac)<sup>2</sup> and the salt of the second metal, (Ni(acac)<sup>2</sup> or Fe(acac)3), were mixed with EDAS and half the volume of ethanol (or only one metallic salt if a monometallic catalyst was prepared). The mixture was then stirred at room temperature until the solution became clear. After the addition of TEOS, a 0.54 M aqueous ammoniacal solution diluted in the remaining half of the ethanol was added with vigorous stirring. The synthesis bottle was then hermetically sealed and placed in the oven at 80 ◦C for 3 days in order to undergo gelling and aging of the gel.

The wet gels were then dried under vacuum at 80 ◦C for 3 days, during which the pressure was slowly reduced to 1200 Pa in order to avoid the gel exploding. The gels were then dried at 150 ◦C at 1200 Pa for 12 h. The dried gels were then calcined in air (0.1 mmol/s) for 12 h at 400 ◦C with a ramp of 120 ◦C/h (air flow rate 0.02 mmol/s). The calcined gels used for the catalytic tests were finally reduced in situ under hydrogen (0.05 mmol/s) for 6 h at 400 ◦C and 1.25 bar. The catalysts used for the characterization were reduced under hydrogen (0.25 mmol/s) for 6 h at 400 ◦C and at atmospheric pressure. The synthesis process is represented in Figure 9.

**Figure 9.** Schematic diagram of the catalyst preparation**. Figure 9.** Schematic diagram of the catalyst preparation.

#### *4.2. Characterizations 4.2. Characterizations*

Bragg angle (rad).

Nitrogen adsorption–desorption isotherms were obtained thanks to an ASAP 2420 multi-sampler adsorption–desorption volumetric device from Micromeritics. Nitrogen adsorption–desorption isotherms were obtained thanks to an ASAP 2420 multi-sampler adsorption–desorption volumetric device from Micromeritics.

The crystallographic properties were observed through the X-ray diffraction (XRD) patterns recorded with a Siemens D5000 powder diffractometer using Cu-Kα radiation for the Ni series and using Fe-Kα radiation for the Fe series. The Scherrer formula (Equation (3)) was used to determine the size of the metal crystallites (i.e., *d*XRD) [33]: The crystallographic properties were observed through the X-ray diffraction (XRD) patterns recorded with a Siemens D5000 powder diffractometer using Cu-K<sup>α</sup> radiation for the Ni series and using Fe-K<sup>α</sup> radiation for the Fe series. The Scherrer formula (Equation (3)) was used to determine the size of the metal crystallites (i.e., *d*XRD) [33]:

where *d*XRD is the crystallite size of metals (nm), *B* the peak full width at half maximum after correction of the instrumental broadening (rad), *λ* the wavelength (nm) and *θ* the

$$d\_{XRD} = 0.9 \frac{\lambda}{(B \cos(\theta))} \tag{3}$$

where *d*XRD is the crystallite size of metals (nm), *B* the peak full width at half maximum after correction of the instrumental broadening (rad), *λ* the wavelength (nm) and *θ* the Bragg angle (rad).

The sizes of the metallic nanoparticles were estimated by transmission electron microscopy (TEM) with a Philips CM100 device, with measurements being taken on approximately one hundred particles. Each sample was prepared before being analyzed using TEM. This preparation was necessary to disperse the samples and thus allow their analysis under the electron microscope. The preparation procedure was as follows:


The grids had a diameter of 3.05 mm and included 400 mesh (37 µm thick).

The sections were then placed directly on the sample holder of the transmission electron microscope.

Real palladium, nickel and iron contents in the catalysts were measured using inductively coupled plasma atomic emission spectroscopy (ICP-AES) on a Spectroflam from Spectro Analytical Instruments. Samples were dissolved in concentrated acids (18 M H2SO4, 22 M HF, 14 M HNO3). Palladium, nickel and iron contents were obtained by comparison with standard solutions in the same medium.

### *4.3. Catalytic Experiments*

The installation used for the hydrodechlorination of chlorobenzene is represented in Figure 10. The reactor was removed and charged with 0.2 g of Pd-Ni/SiO<sup>2</sup> catalyst or 0.1 g of Pd-Fe/SiO<sup>2</sup> catalyst. It was then moved back to its original location. The system was purged with helium. After a purge time of 5 to 10 min, the reduction procedure could be started: a temperature ramp of 5 ◦C/min up to a plateau of 400 ◦C was maintained for 6 h, the 2 stages being carried out under a flow rate of 4 *N*L/h of hydrogen. The reduction temperature was determined by temperature programmed reduction (TPR) (see Supplementary Materials). The temperature of the oven was then brought back to 180 ◦C, the starting temperature of the actual test cycle. The hydrogen flow was reduced to 1 *N*L/h and a flow of 37 *N*L/h of helium was added.

The temperature program was made up of different levels of 180 ◦C, 200 ◦C, 220 ◦C, 240 ◦C and finally a return to 220 ◦C. Each level was maintained for a period of 2 h except for the last which was maintained for a minimum of 6 h. The temperature ramps were set at a rate of 5 ◦C/min. An illustration of this temperature program can be seen in Figure 11. The start of the program coincided with the pivoting of the valve, allowing continuous injection of chlorobenzene.

Throughout the experiment, the chromatograph analyzed the reactor effluents at regular intervals, i.e., approximately every 12 min. Each experiment was carried out three times to assess reproducibility. The effluent was analyzed by gas chromatography (ThermoFinnigan with FID) using a Porapak Q5 packed column.

**References** 

https://doi.org/10.1134/S0023158420060130.

**1990**, *94*, 2493–2504.

The sizes of the metallic nanoparticles were estimated by transmission electron microscopy (TEM) with a Philips CM100 device, with measurements being taken on approximately one hundred particles. Each sample was prepared before being analyzed using TEM. This preparation was necessary to disperse the samples and thus allow their analy-

(ii) embedded catalysts were placed under vacuum for a few hours to allow the resin to

(iv) very thin slices of polymerized resin containing the catalyst were cut using a crystal

The sections were then placed directly on the sample holder of the transmission elec-

Real palladium, nickel and iron contents in the catalysts were measured using inductively coupled plasma atomic emission spectroscopy (ICP-AES) on a Spectroflam from Spectro Analytical Instruments. Samples were dissolved in concentrated acids (18 M H2SO4, 22 M HF, 14 M HNO3). Palladium, nickel and iron contents were obtained by com-

The installation used for the hydrodechlorination of chlorobenzene is represented in Figure 10. The reactor was removed and charged with 0.2 g of Pd-Ni/SiO2 catalyst or 0.1 g of Pd-Fe/SiO2 catalyst. It was then moved back to its original location. The system was purged with helium. After a purge time of 5 to 10 min, the reduction procedure could be started: a temperature ramp of 5 °C/min up to a plateau of 400 °C was maintained for 6 h, the 2 stages being carried out under a flow rate of 4 *N*L/h of hydrogen. The reduction temperature was determined by temperature programmed reduction (TPR) (see Supplementary Materials). The temperature of the oven was then brought back to 180 °C, the starting temperature of the actual test cycle. The hydrogen flow was reduced to 1 *N*L/h

The grids had a diameter of 3.05 mm and included 400 mesh (37 µm thick).

sis under the electron microscope. The preparation procedure was as follows: (i) the catalysts to be observed were embedded in the EPON 812 resin;

(iii) the catalysts then remained for 3 days at 60 °C for the resin to polymerize;

penetrate the pores of the catalyst;

(v) this thin slice was then placed on a copper grid.

parison with standard solutions in the same medium.

and a flow of 37 *N*L/h of helium was added.

knife;

tron microscope.

*4.3. Catalytic Experiments* 

**Figure 11.** Temperature profile used for the catalytic experiments. **Figure 11.** Temperature profile used for the catalytic experiments.

Throughout the experiment, the chromatograph analyzed the reactor effluents at regular intervals, i.e., approximately every 12 min. Each experiment was carried out three times to assess reproducibility. The effluent was analyzed by gas chromatography (Ther-**Supplementary Materials:** The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/gels9040275/s1, Figure S1. TPR of Pd50-Ni50 sample; Figure S2. TPR of Pd50-Fe50 sample.

moFinnigan with FID) using a Porapak Q5 packed column. **Supplementary Materials:** The following supporting information can be downloaded at: www.mdpi.com/xxx/s1, Figure S1. TPR of Pd50-Ni50 sample; Figure S2. TPR of Pd50-Fe50 sample. **Author Contributions:** J.G.M.: Methodology, Investigation, Formal analysis, Writing—original **Author Contributions:** J.G.M.: Methodology, Investigation, Formal analysis, Writing—original draft, writing—review and editing. T.D., K.Y.T. and B.H.: Conceptualization, investigation, formal analysis, writing—review and editing. S.D.L.: Methodology, formal analysis, writing—review and editing, funding acquisition and project administration. All authors have read and agreed to the published version of the manuscript.

draft, writing—review and editing. T.D., K.Y.T. and B.H.: Conceptualization, investigation, formal **Funding:** This research received no external funding.

analysis, writing—review and editing. S.D.L.: Methodology, formal analysis, writing—review and editing, funding acquisition and project administration. All authors have read and agreed the pub-**Institutional Review Board Statement:** Not applicable.

lished version of the manuscript. **Informed Consent Statement:** Not applicable.

**Funding:** This research received no external funding. **Institutional Review Board Statement:** Not applicable. **Data Availability Statement:** The raw/processed data required to reproduce these findings cannot be shared at this time as these data are part of an ongoing study.

**Informed Consent Statement:** Not applicable. **Data Availability Statement:** The raw/processed data required to reproduce these findings cannot be shared at this time as these data are part of an ongoing study. **Acknowledgments:** Julien G. Mahy and Stéphanie D. Lambert thank the F.R.S.-FNRS for his Postdoctoral Researcher position and her Research Director position, respectively. J.G.M. is grateful to the Rotary for a District 2160 grant, to the University of Liège and the FNRS for financial support for a postdoctoral stay in INRS Centre Eau, Terre, Environnement in Québec, Canada.

doctoral Researcher position and her Research Director position, respectively. J.G.M. is grateful to the Rotary for a District 2160 grant, to the University of Liège and the FNRS for financial support

for a postdoctoral stay in INRS Centre Eau, Terre, Environnement in Québec, Canada.

**Acknowledgments:** Julien G. Mahy and Stéphanie D. Lambert thank the F.R.S.-FNRS for his Post-**Conflicts of Interest:** The authors declare no conflict of interest.

2. Ryaboshapka, D.A.; Lokteva, E.S.; Golubina, E.V.; Kharlanov, A.N.; Maslakov, K.I.; Kamaev, A.O.; Shumyantsev, A.V.; Lipatova, I.A.; Shkol'nikov, E.I. Gas-Phase Hydrodechlorination of Chlorobenzene over Alumina-Supported Nickel Catalysts: Effect of Support Structure and Modification with Heteropoly Acid HSiW. *Kinet. Catal.* **2021**, *62*, 127–145.

3. Ritter, E.R.; Bozzelli, J.W.; Dean, A.M. Kinetic Study on Thermal Decomposition of Chlorobenzene Diluted in H2. *J. Phys. Chem.* 

1. McGrath, M. Catalysts and Systems for the Removal of Volatile Oganic Compounds. *Appl. Catal. B* **1995**, *5*, N25.

### **References**


**Disclaimer/Publisher's Note:** The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

**Maria A. Morosanova and Elena I. Morosanova \***

Analytical Chemistry Division, Chemistry Department, Lomonosov Moscow State University, 119234 Moscow, Russia

**\*** Correspondence: emorosanova@gmail.com

**Abstract:** Chromogenic enzymatic reactions are very convenient for the determination of various biochemically active compounds. Sol-gel films are a promising platform for biosensor development. The creation of sol-gel films with immobilized enzymes deserves attention as an effective way to create optical biosensors. In the present work, the conditions are selected to obtain sol-gel films doped with horseradish peroxidase (HRP), mushroom tyrosinase (MT) and crude banana extract (BE), inside the polystyrene spectrophotometric cuvettes. Two procedures are proposed: the use of tetraethoxysilane-phenyltriethoxysilane (TEOS-PhTEOS) mixture as precursor, as well as the use of silicon polyethylene glycol (SPG).In both types of films, the enzymatic activity of HRP, MT, and BE is preserved. Based on the kinetics study of enzymatic reactions catalyzed by sol-gel films doped with HRP, MT, and BE, we found that encapsulation in the TEOS-PhTEOS films affects the enzymatic activity to a lesser extent compared to encapsulation in SPG films. Immobilization affects BE significantly less than MT and HRP. The Michaelis constant for BE encapsulated in TEOS-PhTEOS films almost does not differ from the Michaelis constant for a non-immobilized BE. The proposed sol-gel films allow determining hydrogen peroxide in the range of 0.2–3.5 mM (HRP containing film in the presence of TMB), and caffeic acid in the ranges of 0.5–10.0 mM and 2.0–10.0 mM (MT- and BEcontaining films, respectively). BE-containing films have been used to determine the total polyphenol content of coffee in caffeic acid equivalents; the results of the analysis are in good agreement with the results obtained using an independent method of determination. These films are highly stable and can be stored without the loss of activity for 2 months at +4 ◦C and 2 weeks at +25 ◦C.

**Citation:** Morosanova, M.A.; Morosanova, E.I. Sol-Gel Films Doped with Enzymes and Banana Crude Extract as Sensing Materials for Spectrophotometric Determination. *Gels* **2023**, *9*, 240. https://doi.org/ 10.3390/gels9030240

Academic Editors: Francesco Caridi, Giuseppe Paladini and Andrea Fiorati

Received: 28 February 2023 Revised: 16 March 2023 Accepted: 16 March 2023 Published: 18 March 2023

**Copyright:** © 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

**Keywords:** sol-gel films; optical sensors; tetraethoxysilane; phenyltriethoxysilane; silicon polyethylene glycol; peroxidase; tyrosinase; crude banana extract; determination of total polyphenol content

### **1. Introduction**

Sol-gel materials are widely used in analytical practice. The immobilization of proteins in sol-gel matrices has been attempted in a number of works, given the good properties of the resulting materials: preservation of the biocatalytic properties of the encapsulated proteins and excellent optical properties. Since the first reported case of a successful immobilization of active alkaline phosphatase via the sol-gel method [1], sol-gel silicate has become a desired immobilization matrix for the design of active biocomposite materials. Sol-gel materials as a matrix for immobilized proteins or biomolecules can have many applications, such as stationary phase [2], drug delivery materials [3], and coatings [4,5]. Often sol-gel materials with immobilized enzymes are created for analytical enzymatic applications [1,6–15].

Silica sol-gel films present a major advantage comparing to other sol-gel materials for enzyme immobilization: the closeness of the immobilized enzymes to the solid-solution interface. This increases the accessibility of the enzymes to substrates from the aqueous phase, making silica sol-gel film a promising platform for biosensor development. Solgel films are particularly often employed for both optical [6–12], and electrochemical biosensors [6,13–15].

Chromogenic enzymatic reactions are very convenient for the determination of various biochemically active compounds. In our opinion, the use of sol-gel films to create optical biosensors deserves attention as an effective way to simplify analysis for its further implementation in field conditions.

To obtain sol-gel materials doped with biomolecules, the latter are often encapsulated in matrices of Ormosils—materials obtained by the hydrolysis of tetraethoxysilane in the presence of organic silicon alkoxides, primarily 3-aminopropyltriethoxysilane [16,17]. However, the alcohol formed during the hydrolysis and condensation of the alkoxide precursors could negatively affect the entrapped enzymes' activity [17]. This is an important problem that must be solved before employing the sol-gel process as a universal method of protein or other biomolecules encapsulation. In the literature, special schemes are described for the synthesis of sol-gel materials doped with enzymes to preserve the enzymatic activity of immobilized enzymes [18–23]. The approaches described can be divided into two groups: the use of tetraethoxysilane and its derivatives as precursors while using various ways to minimize the contact of the enzyme with ethanol released upon hydrolysis of the precursors and the use of silicate and glycerates as precursors, the hydrolysis of which does not release ethanol.

The approaches in the first group include the modification of the procedure while using standard alkoxide precursors [9,10,18]. To minimize the contact of the enzyme with ethanol, a two-stage synthesis scheme is proposed: the first stage is the hydrolysis of tetraethoxysilane or its mixtures with organic silicon alkoxides and the preparation of the sol; the second stage is the addition of the enzyme to the sol-gel solution and the formation of gels. Sometimes even the removal of the alcohol by rotavaporization method is performed before the second stage [18]. The technology of "kinetic doping" is also proposed where the nascent sol-gel film is submerged into enzyme-containing buffer solution which provides alcohol dilution [9,10].

The second group implies the use of different precursors: sodium silicate [19,20] or glycerol-derived silicates [21–23], which provide an alcohol-free sol. Both these routes possess some limitations in their application: the glycerol-derived silicate precursors need to be synthesized and sodium silicate precursors produce high sodium concentration levels in the sol.

Using crude extracts as enzyme sources in biocomposite sol-gel materials seems a promising approach. It was shown for crude extracts with polyphenol oxidase activity that, other conditions being equal, the enzymatic activity of sol-gel materials doped with the extracts is significantly higher than the activity of sol-gel materials doped with commercial tyrosinase [11,12]. Based on our experience of studying crude extracts, we also noted that the use of plant and mushroom extracts as enzyme sources presents a number of advantages compared to purified enzymes: higher interference thresholds, better stability, and lower cost [24–26]. In our opinion, the study of the crude extracts' properties and the study of the possibilities of their inclusion in sol-gel films will contribute to the development of methods for the field determination of biochemically important analytes.

Historically, sol-gel films for optical applications were prepared on glass slides, which then could be put inside the cuvettes. However, a more efficient and practical way to create an optical biocomposite sensor is also described [27,28]: the sol-enzyme mixture is put directly into the dispensable polystyrene cuvettes and the film is formed on the cuvette inner side. This approach provides an easy way to fixate the film position in relation to the optical path and also to precisely measure the amount of the enzyme and sol.

The goal of this work was to develop methods for the synthesis of transparent sol-gel films doped with the most well-studied enzymes (horseradish peroxidase and mushroom tyrosinase), as well as with crude banana extract as a source of polyphenol oxidase, on the inner surface of plastic cuvettes with the use of tetraethoxysilane and organic alkoxides or silicon glycerate as precursors, studying the effect of immobilization on the activity of these enzymes, and evaluating the analytical performance of synthesized sol-gel films.

### **2. Results**

### *2.1. Synthesis of Sol-Gel Films with Immobilized Enzymes and Banana Extract*

The validity of the sol-gel route presented in this work to preserve the enzyme activity during the immobilization process has been studied on two of the enzymes that are most widely used in analytical methods: horseradish peroxidase (HRP) and mushroom tyrosinase (MT), and also the widely used crude plant extract—banana extract (BE) as a source of polyphenol oxidase [23]. Most sol-gel immobilization studies use HRP [9,10,15,18,19,21]; polyphenol oxidase (tyrosinase) is studied less frequently [11–14,29]. We have optimized the sol-gel matrix using HRP, and then studied the performance of HRP, MT, and BE in the selected conditions. We have developed two approaches: using sol-gel films based on mixtures of TEOS with derivatives (i.e., Ormosils) and using glycerol precursors; then we studied the effect of immobilization on the activity of enzymes and the possibility of analytical use of the synthesized films.

### 2.1.1. Sol-Gel Films Based on Alkoxide Precursors (TEOS Films)

Sol-gel synthesis consists of successively carrying out the following stages: hydrolysis of precursors, polymerization (transformation of a sol into a gel) and, if necessary, drying of the gels under different conditions. Typically, hydrolysis is carried out in the presence of a related alcohol; when using TEOS, in the presence of ethanol. The properties of sol-gel materials depend on the nature of the precursors, the ratio of components in the hydrolyzing mixture, the nature of the gelation catalyst, and special additives to control the porosity of the materials. To obtain sol-gel materials doped with analytical reagents, we have developed a synthesis scheme [30,31], based on the hydrolysis of tetraethoxysilane in an aqueous-ethanol medium in the presence of hydrochloric acid as a catalyst and cetylpyridinium chloride as a pore former [32]. This scheme was used as the basis for the development of a method for the synthesis of sol-gel films doped with enzymes.

Studies show that when enzymes are included in Ormosils, i.e., sol-gel materials obtained from modified alkoxide precursors, their activity decreases to a lesser extent than in the case of standard TEOS materials. Most often, methyl- and amino-derivatives of tetraethoxysilane are used as precursors for the immobilization of enzymes in sol–gel materials [16], while phenyl derivatives are used much less frequently.

According to the literature, adding the enzyme solution to the already formed sol and not to the hydrolyzing mixture seems like a promising approach. This method was proposed using sodium silicate as a precursor [19]. We decided to take this approach when we used tetraethoxysylane (TEOS) and its mixtures with phenyltriethoxysilane (PhTEOS) and 3-aminopropyltriethoxysilane (AmTEOS) as sol-gel precursors. These TEOS-based sol-gel matrices were studied using horseradish peroxidase (HRP) as a model enzyme. HRP is widely used in enzyme immobilization studies and comparing the results of the present study using HRP was likely to be easier.

In order to select the conditions for the synthesis of transparent HRP-containing solgel films on the inner surface of plastic cuvettes, the influence of the nature and ratio of precursors, concentrations of hydrochloric acid and cetylpyridinium chloride was studied. TEOS, AmTEOS, and PhTEOS were used as precursors. The mixtures of precursors with water were prepared with 3:1 ratio of precursors:water and were mixed under the influence of ultrasound with a sound energy density of 0.24–0.38 W/mL at the hydrolyzing stage. The transparent homogeneous sols were formed in 45–90 min. The resulting sols are stable and can be stored at +4 ◦C for a week. To obtain gels, the sols were mixed with HRP solution in a buffer (pH 6.0) in a ratio of 1:0.8. The loss of fluidity of this mixture (i.e., gel formation) occurs after 2–10 min, depending on the composition of the sol. To prepare the sol-gel films, 0.5–0.9 mL of the mixture was placed in cuvettes, distributed on one of the inner sides, and after 2–10 min, films with an approximate thickness of 1.5–2.5 mm were formed. The stability of the sol-gel films, the enzymatic activity of the immobilized enzymes, and their storage stability were evaluated to choose the synthesis conditions.

When 1.0–4.5% *w*/*w* AmTEOS was added to TEOS, the films lost transparency compared to films prepared with only TEOS. For TEOS-AmTEOS films, the 10–100 fold increase in the concentration of hydrochloric acid led to the increase in transparency, but it was accompanied by significant reduction in the gelation time (less than 1 min), which made it difficult to obtain films by our method. made it difficult to obtain films by our method. The use of TEOS-PhTEOS mixtures with a PhTEOS content of 1–10% *w/w* made it possible to obtain films that are transparent in the visible light range. Films produced at PhTEOS contents greater than 2% cracked on the next day after the preparation, but at

resulting sols are stable and can be stored at +4 °C for a week. To obtain gels, the sols were mixed with HRP solution in a buffer (pH 6.0) in a ratio of 1:0.8. The loss of fluidity of this mixture (i.e., gel formation) occurs after 2–10 min, depending on the composition of the sol. To prepare the sol-gel films, 0.5–0.9 mL of the mixture was placed in cuvettes, distributed on one of the inner sides, and after 2–10 min, films with an approximate thickness of 1.5–2.5 mm were formed. The stability of the sol-gel films, the enzymatic activity of the immobilized enzymes, and their storage stability were evaluated to choose

When 1.0–4.5% *w*/*w* AmTEOS was added to TEOS, the films lost transparency compared to films prepared with only TEOS. For TEOS-AmTEOS films, the 10–100 fold increase in the concentration of hydrochloric acid led to the increase in transparency, but it was accompanied by significant reduction in the gelation time (less than 1 min), which

*Gels* **2023**, *9*, 240 4 of 15

the synthesis conditions.

The use of TEOS-PhTEOS mixtures with a PhTEOS content of 1–10% *w/w* made it possible to obtain films that are transparent in the visible light range. Films produced at PhTEOS contents greater than 2% cracked on the next day after the preparation, but at 1–2% *w*/*w* they remained stable and did not crack. When using mixtures of TEOS-AmTEOS-PhTEOS with 1% *w*/*w* of PhTEOS and 0.4% *w*/*w* of AmTEOS, a lack of film transparency was observed, which led us to stop using AmTEOS and to concentrate on studying TEOS-PhTEOS mixtures. 1–2% *w*/*w* they remained stable and did not crack. When using mixtures of TEOS-AmTEOS-PhTEOS with 1% *w*/*w* of PhTEOS and 0.4% *w*/*w* of AmTEOS, a lack of film transparency was observed, which led us to stop using AmTEOS and to concentrate on studying TEOS-PhTEOS mixtures. In the literature, enzyme-doped sol-gel materials obtained by various methods are usually characterized by the retention of the immobilized enzyme in the matrix and its

In the literature, enzyme-doped sol-gel materials obtained by various methods are usually characterized by the retention of the immobilized enzyme in the matrix and its activity [9,19,21]. We investigated the effect of PhTEOS content on enzyme retention using HRP as a model enzyme. Under the described above conditions, HRP-doped films were synthesized with different PhTEOS content in the mixture of precursors (0, 1, and 2%)— HRP-PhTEOS0, HRP-PhTEOS1, and HRP-PhTEOS2 films. Enzyme retention was studied by determining the enzyme activity in buffer solutions obtained by washing the sol-gel films. The results are shown in Figure 1. The introduction of PhTEOS into the precursor mixture increased the retention of peroxidase by the sol-gel matrix. Thus, it can be concluded that peroxidase retention is significantly improved when PhTEOS is introduced into the matrix, and the enzyme is washed out slightly less from films containing 1% PhTEOS than from the films containing 2% PhTEOS. The obtained HRP retention values in the films after three washes are shown in Table 1. Comparison with literature data shows that the enzyme washing out values in our experiments are somewhat greater than for the previously proposed methods [19,21]. However, our film preparation method is simple, involves the use of commercially available precursors, and preserves significant activity of the immobilized enzyme. activity [9,19,21]. We investigated the effect of PhTEOS content on enzyme retention using HRP as a model enzyme. Under the described above conditions, HRP-doped films were synthesized with different PhTEOS content in the mixture of precursors (0, 1, and 2%)—HRP-PhTEOS0, HRP-PhTEOS1, and HRP-PhTEOS2 films. Enzyme retention was studied by determining the enzyme activity in buffer solutions obtained by washing the sol-gel films. The results are shown in Figure 1. The introduction of PhTEOS into the precursor mixture increased the retention of peroxidase by the sol-gel matrix. Thus, it can be concluded that peroxidase retention is significantly improved when PhTEOS is introduced into the matrix, and the enzyme is washed out slightly less from films containing 1% PhTEOS than from the films containing 2% PhTEOS. The obtained HRP retention values in the films after three washes are shown in Table 1. Comparison with literature data shows that the enzyme washing out values in our experiments are somewhat greater than for the previously proposed methods [19,21]. However, our film preparation method is simple, involves the use of commercially available precursors, and preserves significant activity of the immobilized enzyme.

**Figure 1.** HRP activity (% in relation to the initial TEOS-PhTEOS sol-gel film activity) in the washing buffer solutions (pH 6.0).


### **Table 1.** Retention of HRP in various sol-gel films after washing.

It is widely known that the activity of an entrapped enzyme is usually only a fraction of its activity in free solution [9]. The relative activity of the immobilized enzyme was calculated as a percentage of the activity of a similar amount of the enzyme in solution. The initial rates comparison allowed calculating the film loaded enzymes' relative activity (Table 2). The relative activities were in the range of 6.6–7.4%, which is similar to the results observed for other sol-gel matrices described in the literature [9,19]. The highest relative activity was observed for 1% PhTEOS sol-gel film. For further experiments, we chose a film obtained by adding 1% PhTEOS to a mixture of precursors—HRP-PhTEOS1.

**Table 2.** Relative activity of the immobilized enzymes (% of the activity of the same amount of native enzyme) for the different sol-gel films.


HRP—horseradish peroxidase, MT—mushroom tyrosinase, BE—crude banana extract.

Three types of films were prepared for the following studies using this sol-gel matrix: HRP-PhTEOS1 with immobilized HRP, MT-PhTEOS1 with immobilized MT, BE-PhTEOS1 with immobilized BE. The relative activities were also calculated for MT and BE (Table 2) and they were similar to the HRP activities obtained in our study and other works [9,19]. No available data on the relative activity of immobilized tyrosinase or crude extracts were found in the literature.

HRP-PhTEOS1, MT-PhTEOS1, and BE-PhTEOS1 films were used to study the kinetics of enzymatic reactions.

2.1.2. Sol-Gel Films Based on Silicon Polyethylene Glycol (SPG Films)

Another approach for creating biocomposite sol-gel films is the employment of silicon polyethylene glycol (SPG). SPG rapidly hydrolyzes and forms gels in aqueous media without the need for any catalyst, such as hydrochloric acid, to form silica hydrogels, which are transparent, and physically stable [21–23]. This approach was tested with many biological molecules, such as peroxidase, catalase, various oxidases, etc. [21]. We decided to synthesize glycerol-containing-precursors based sol-gel films and compare them to our TEOS-PhTEOS films.

Unlike alkoxide-based films, no ethanol is generated when using SPG, so SPG films are easier to prepare. In order to obtain SPG films, the precursor was mixed with a solution of the enzyme/extract in a pH 6.0 buffer solution in a ratio of 1:2. To form sol-gel films, 0.6 mL of the mixture was placed in cuvettes, distributed on one of its inner sides, and after 90 min, films with an approximate thickness of 1.5–2.5 mm were formed. Gelation occurred in the absence of a catalyst, and film formation took longer than in the case of TEOS-PhTEOS films.

Three types of films were prepared for the following studies using this sol-gel matrix: HRP-SPG with immobilized HRP, MT-SPG with immobilized MT, BE-SPG with immobilized BE. The relative activity of enzymes in these films is given in Table 2, and it is comparable to TEOS-PhTEOS based films. HRP-SPG, MT-SPG, and BE-SPG films were also used to study the kinetics of enzymatic reactions.

### *2.2. Study of the Kinetics of Enzymatic Reactions in the Presence of Sol-Gel Films Doped with HRP, MT, and BE*

The effect of the sol-gel process on the enzyme activity was investigated by comparing the kinetic parameters (Michaelis constants) of the reactions catalyzed by native and immobilized enzyme. The initial rates of the HRP, MT, and BE catalyzed reactions were measured as the absorbance increase over time. The Michaelis constants were obtained through the fitting of the data to Michaelis–Menten kinetic analysis using Lineweaver–Burk plots. The Michaelis constants were calculated for hydrogen peroxide in the presence of constant TMB concentration (0.009%) in the case of HRP and for caffeic acid in the cases of MT and BE.

The enzymes included in the sol-gel films obtained in this work retain their activity, and the kinetics of the reactions catalyzed can be fitted to the Michaelis–Menten equation. Figure 2 shows, for example, kinetic curves for different concentrations of hydrogen peroxide in the presence of HRP-PhTEOS1 and HRP-SPG films.

For the evaluation of the immobilized enzymes' properties, we studied their interaction with substrates (hydrogen peroxide in the case of HRP and caffeic acid in the case of MT and BE) and calculated the kinetic parameters of the enzymatic reaction (Michaelis constants). Figure 3 shows the dependence of the reaction rate on the concentration of hydrogen peroxide in the presence of HRP-PhTEOS1 and HRP-SPG films. When Lineweaver–Burk coordinates are used, these dependencies become linear and allow the calculation of the Michaelis constants (Table 3). Both in our experiments and in the literature data [11,18,19] the Michaelis constant values (KM) of the immobilized enzymes were higher than those of the native enzymes, indicating the presence of partitioning and diffusional effects in the pores of the sol-gel matrix. Table 3 shows that, when using SPG-based films, an even greater increase in K<sup>M</sup> values is observed, meaning that such films are better fit for the determination of high concentrations of substrates. This can be explained by the greater steric hindrance because of bulkier sol-gel precursor molecules. The obtained data indicate that for all the studied sol-gel films, the inclusion of HRP, MT, and BE does not hinder their enzymatic activity and allows their use for enzymatic reactions. TEOS-PhTEOS-based films seem to be more promising for the development of methods for determining low contents of analytes-substrates.

**Figure 2.** Kinetic curves of the TMB oxidation (absorbance at 650 nm over time) in the presence of different hydrogen peroxide concentrations (indicated in the legends) and HRP-PhTEOS1 (**a**) or HRP-SPG (**b**) sol-gel films. **Figure 2.** Kinetic curves of the TMB oxidation (absorbance at 650 nm over time) in the presence of different hydrogen peroxide concentrations (indicated in the legends) and HRP-PhTEOS1 (**a**) or HRP-SPG (**b**) sol-gel films.

For the evaluation of the immobilized enzymes' properties, we studied their interaction with substrates (hydrogen peroxide in the case of HRP and caffeic acid in the case of MT and BE) and calculated the kinetic parameters of the enzymatic reaction (Michaelis constants). Figure 3 shows the dependence of the reaction rate on the concentration of hydrogen peroxide in the presence of HRP-PhTEOS1 and HRP-SPG films. When Lineweaver–Burk coordinates are used, these dependencies become linear and allow the calculation of the Michaelis constants (Table 3). Both in our experiments and in the literature data [11,18,19] the Michaelis constant values (KM) of the immobilized enzymes were higher than those of the native enzymes, indicating the presence of partitioning and diffusional effects in the pores of the sol-gel matrix. Table 3 shows that, when using SPG-based films, an even greater increase in KM values is observed, meaning that such films are better fit for the determination of high concentrations of substrates. This can be We have studied crude banana extract in the present immobilization study, because earlier we have established that crude plant extracts have higher interference thresholds than purified enzymes [24]. There are data indicating that crude extracts are more robust and endure sol-gel immobilization better: in some cases, the extract can withstand the immobilization procedure that inhibits the corresponding purified enzyme activity [11]. In the present study, we observed that for crude banana extract the Michaelis constant remained almost the same after the immobilization in PhTEOS1 film (2.4 mM in solution vs. 2.8 mM in film). A similar effect was described earlier for desert truffle tyrosinase extract [11]: the Michaelis constant even slightly decreased upon immobilization (0.5 mM in solution vs.0.2 mM in film). This can be possibly explained by the presence of other plant cell fragments in the crude extracts which create a better environment for the enzymes inside the sol-gel matrices.

> explained by the greater steric hindrance because of bulkier sol-gel precursor molecules. The obtained data indicate that for all the studied sol-gel films, the inclusion of HRP, MT, and BE does not hinder their enzymatic activity and allows their use for enzymatic reac

Horseradish peroxidase

Mushroom tyrosinase

tions. TEOS-PhTEOS-based films seem to be more promising for the development of

Na silicate 0.163 0.985 6 [19] TEOS 0.55 2.38 4.3 [18]

TEOS+ 1% PhTEOS 0.4 1.4 3.5 Present work SPG 0.4 9.9 24.8

TEOS+ 1% PhTEOS 1.1 4.1 3.7 Present work SPG 1.1 10.1 9.2

TEOS + colloidal silica 0.9 no activity - [11]

**Table 3.**The influence of immobilization in sol-gel film on the Michaelis constants (KM).

methods for determining low contents of analytes-substrates.

**Enzyme Source Film Precursor(s) KM, mM KM Ratio Reference Solution Film** 

**Figure 3.** The dependence of the reaction speed (min<sup>−</sup>1) in the presence of HRP-PhTEOS1 film (blue) and HRP-PGS film (red) on the concentration of hydrogen peroxide (*n* = 3). **Figure 3.** The dependence of the reaction speed (min−<sup>1</sup> ) in the presence of HRP-PhTEOS1 film (blue) and HRP-PGS film (red) on the concentration of hydrogen peroxide (*n* = 3).


We have studied crude banana extract in the present immobilization study, because **Table 3.** The influence of immobilization in sol-gel film on the Michaelis constants (KM).

Based on the enzyme kinetics study of the immobilized enzymes we have chosen HRP-PhTEOS1, MT-PhTEOS1, and BE-PhTEOS1 films for the analytical application.

### *2.3. Analytical Application HRP-PhTEOS1, MT-PhTEOS1, and BE-PhTEOS1 Films*

We studied the possibility of the analytical use of the proposed sol-gel films doped with enzymes and banana extract. Hydrogen peroxide was used as analyte for HRP-PhTEOS1 film in the presence of TMB, and caffeic acid was used for MT-PhTEOS1 and BE-PhTEOS1 films. Immobilized enzymes catalyze the corresponding chromogenic reactions: hydrogen peroxide reduction with TMB oxidation and caffeic acid oxidation by air oxygen. The reaction rates were used as an analytical signal; the dependence of the absorbance on time was studied for different concentrations of analytes. Analytical ranges—analyte concentration ranges with a linear dependence of the reaction rate on analyte concentrations—are given in Table 4. Comparison of detection limits (LOD) for the sol-gel film encapsulated enzymes and banana extract, with LODs for non-immobilized enzymes and extract, demonstrate only 2–3 fold loss of sensitivity (Table 4). Such effect is likely attributed to steric hindrances arising during immobilization. The simplicity of determinations using sol-gel films doped

with enzymes and banana extract should be noted: it is simply needed to place 3.0 mL of a sample in a cuvette containing a sol-gel film (in the case of determining hydrogen peroxide, 0.4 mL of 0.08% TMB solution should also be added), and monitor the change in absorbance at 650 nm for the determination of hydrogen peroxide and 400 nm for the determination of caffeic acid. Such measurements can be carried out using various portable photometers, which opens the prospect of mass analyses for the determination of biochemically active analytes in the field.


**Table 4.** Analytical parameters of the procedures using HRP-PhTEOS1, MT-PhTEOS1, and BE-PhTEOS1 sol-gel films.

In this paper, to demonstrate the analytical capabilities of sol-gel films, we present the results of the total polyphenol content determination in coffee (in caffeic acid equivalents) using an immobilized banana extract (BE-PhTEOS1 film). Total polyphenol content determination is often used in food quality control [24].

The recovery study of the caffeic acid determination using BE-PhTEOS1 film shows that the RSD values are comparable to those for banana extract in solution and equal 7–10% (*n* = 3).

The results of TPC determination in coffee compared with the results of independent methods are given in Table 5.

**Table 5.** Results of TPC determination in coffee using BE-PhTEOS1 film and by independent methods (*n* = 3, *P* = 0.95).


The good agreement between the different procedures indicates the good accuracy of the TPC determination with BE-PhTEOS1 film. No significant difference was found between the four values using Student test (*p* > 0.28 for all the pairs).

The stability and lifetime of the immobilized BE was investigated by measuring the sol-gel film activity using 2.0 mM caffeic acid solution. 95% activity of immobilized BE was retained after 2 months storage at +4 ◦C (BE solution lost its activity after 4 days). 90% activity of immobilized BE was retained after 2 weeks storage at +25 ◦C.

These novel enzyme-doped silica matrixes provide promising platforms for development of various on-site analytical procedures. In the present work we proposed a procedure using crude plant extract immobilized in TEOS-PhTEOS sol-gel film (BE-PhTEOS1 film) for spectrophotometric determination of total polyphenol content using a standard curve of caffeic acid. The detection limit for caffeic acid equals 0.7 mM, while LOD values of other enzymatic methods of TPC determination lie in the 0.01–0.5 mM range [24]. However, the sol-gel films are ready-to-use and offer the possibility of storage at a room temperature. Using crude extract as the enzyme source in these sol-gel materials allows low-cost analysis which makes the process suitable for wide screening tests.

### **3. Conclusions**

We have chosen the conditions for the synthesis of sol-gel films doped with HRP, MT, and BE using TEOS-PhTEOS mixture or SPG as precursors, on the inner side surface of polystyrene cuvettes. When using TEOS and PhTEOS precursors, the film preparation consists of two stages: the preparation of the sol under the influence of ultrasound for 90 min, which leads to the evaporation of a significant part of the formed alcohol, and the subsequent mixing of the sol with an enzyme solution. In the case of SPG films, the enzyme solution is mixed directly with the precursor. Basing on the study of the activity of the immobilized enzymes and the immobilized extract, we have concluded that for both types of films, the enzymatic activity is preserved, and the kinetics of the catalyzed reactions can be described by the Michaelis–Menten equation. The relative activity of the immobilized enzymes is comparable for both types of films and is about 10% of the activity of the non-immobilized enzyme. Thus, the preservation of enzyme activity in the proposed procedures is comparable to those described in the literature, which can also sometimes be significantly more complicated in execution.

When HRP and MT are included in alkoxide-based films, the Michaelis constants increase 3–4 fold, and in SPG-based films—10–20 fold. Compared to these purified enzymes, the crude banana extract demonstrates that it can better withstand the effect of immobilization: for BE the Michaelis constant almost does not change in the alkoxide-based films, and it increases only 3 fold in SPG-based films.

The analytical capabilities of sol-gel films doped with enzymes and banana extract are demonstrated: the analytical range for the hydrogen peroxide determination is 0.2–3.5 mM using HRP-PhTEOS1 film in the presence of TMB, and the analytical ranges for caffeic acid determination are 0.5–10.0 mM and 2.0–10.0 mM using MT-PhTEOS1 and BE-PhTEOS1 films, respectively. The sensitivity of the determination is decreased only 2–3 fold compared to non-immobilized enzymes, while the use of disposable cuvettes with a sol-gel film on the inner side surface greatly simplifies the determination procedure and makes it possible to carry out the determination in field conditions.

BE-PhTEOS1 films have been used to determine the total polyphenol content of coffee in caffeic acid equivalents. The lifetime of BE-PhTEOS1 is 2 months at +4 ◦C storage and 2 weeks at +25 ◦C storage, which is a significant improvement of the shelf life compared to the non-immobilized enzymes and extracts.

### **4. Materials and Methods**

### *4.1. Materials*

Tetraethoxysilane (TEOS), phenyltriethoxysilane (PhTEOS), 3-aminopropyltriethoxysilane (AmTEOS), and hydrogen peroxide were purchased from Sigma-Aldrich (St. Louis, MO, USA). Silicon polyethylene glycol (SPG) was synthesized under the supervision of Dr. T.G. Khonina according to the method described earlier [22].

Caffeic acid, cetylpyridinium chloride, and 3,30 ,5,50 -tetramethylbenzidine (TMB) were purchased from Acros Organics (Carlsbad, CA, USA). Phosphate buffer (pH 6.0) was prepared using sodium monophosphate and potassium diphosphate.

Horseradish peroxidase (HRP) (250 U/mg) was purchased from Biozyme laboratories (UK). Mushroom tyrosinase (MT) from *A. niger* (3900 U/mg) was purchased from Sigma (St. Louis, MO, USA).

Banana extract (BE) was prepared similarly to [24]: 100.0 g of homogenized banana pulp tissue was stirred in 200.0 mL of phosphate buffer (pH 6.0) at 0 ◦C for 30 min, and then filtered twice through a paper filter. The protein content of the banana extract was determined by the Biuret method. Total protein content equaled 3.8 mg/mL for banana pulp crude extract. The activity of the crude banana extract used in this work has been determined by comparing the reaction speed of catechol oxidation in the presence of the crude extract and the commercial mushroom tyrosinase. Crude banana extract activity was found to be 292 ± 6 U/mL (*n* = 3, *P* = 0.95).

Polystyrene cuvettes (10 × 10 × 45 mm) with caps were purchased from Sarstedt (Numbrecht, Germany).

### *4.2. Synthesis of Sol-Gel Films Doped with Mushroom Tyrosinase, Horseradish Peroxidase and Banana Extract*

4.2.1. Sol-Gel Films Based on TEOS and Its Mixtures with PhTEOS and AmTEOS (TEOS film)

In vials, a certain volume of AmTEOS or PhTEOS was added to a certain volume of TEOS. An aqueous solution of cetylpyridinium chloride and hydrochloric acid as a catalyst was added to the resulting mixture. The silane mixture: water ratio was 3:1. The mixture was stirred under the influence of ultrasound with the sound energy density of 0.28–0.34 W/mL for 90 min. In a plastic cuvette, 0.3 mL of the sol was mixed with 0.3 mL of the enzyme/crude extract solution in buffer. The cuvettes were capped, shaken, and then placed on their side. After 2–10 min, a transparent sol–gel film formed on the inner wall surface of the cuvette. The cuvettes with films were stored at +4 ◦C.

### 4.2.2. Synthesis of HRP-PhTEOS1, MT-PhTEOS1, BE-PhTEOS1 Films

1.5 mL of aqueous solution containing 0.5 mMcetylpyridinium chloride and 1.6 mM hydrochloric acid was added to 4.5 mL of TEOS and 0.05 mL of PhTEOS. The mixture was stirred under the influence of ultrasound with a sound energy density of 0.3 W/mL for 90 min. In a plastic cuvette, 0.3 mL of the sol was mixed with 0.3 mL of HRP solution in buffer (pH 6.0) to obtain HRP-PhTEOS1 film, with 0.3 mL MT solution in buffer (pH 6) to obtain MT-PhTEOS1 film, 0.3 mL of BE to obtain BE-PhTEOS1 film. The cuvettes were capped, shaken, and then placed on their side. After 2–10 min, a transparent sol–gel film formed on the inner wall surface of the cuvette. The cuvettes with films were stored at +4 ◦C.

### 4.2.3. Sol-Gel Films Based on SPG

In a cuvette 0.2 mL of SPG was mixed with 0.4 mL of HRP solution in buffer (pH 6.0) to obtain an HRP-SPG film, with 0.4 mL of MT solution in buffer (pH 6.0) to obtain a MT-SPG film, with 0.4 mL of BE to obtain a BE-SPG film. The cuvettes were capped, shaken, and then placed on their side. After 60–90 min, a transparent sol–gel film formed on the inner wall surface of the cuvette. The cuvettes with films were stored at +4 ◦C.

### *4.3. Study of the Immobilized HRP, MT, and BE Activity and Properties in Sol-Gel Films* 4.3.1. Relative Activity of Immobilized Enzymes

To study the activity of HRP immobilized in sol-gel films, 3.0 mL of a hydrogen peroxide solution of various concentrations and 0.4 mL of a 0.08% TMB solution were added to the cuvette with film. Absorbance was measured at 650 nm for 10 min every 10 s. Enzyme activity was determined as the initial reaction rate. The relative activity was determined from the dependence of the HRP activity in solution on the HRP amount. This dependence was obtained by the following procedure: 3.0 mL of 8.0 mM hydrogen peroxide solution was mixed with 0.4 mL of 0.08% TMB solution and 0.4 mL of HRP solution with different amounts of enzyme, and absorbance was measured at 650 nm.

To study activity of MT and BE immobilized in sol-gel films, 3.0 mL of caffeic acid solution of various concentrations was added to the cuvette with film and the absorbance was measured at 400 nm for 10–15 min every 10 s. Enzyme activity was determined as the initial reaction rate. The relative activity was determined from the dependence of the MT/BE activity in the solution on the MT/BE amount. This dependence was obtained by the following procedure: 1.0 mL of MT/BE solution with different amounts of enzyme was added to 2.0 mL of 5.0 mM caffeic acid solution, and the absorbance was measured at 400 nm.

### 4.3.2. HRP Retention on PhTEOS0, PhTEOS1, PhTEOS2 Films

To study the retention of peroxidase in films of various compositions—TEOS (Ph-TEOS0), TEOS + 1%PhTEOS (PhTEOS1), TEOS + 2%PhTEOS (PhTEOS2)—2.0 mL of a buffer solution (pH 6.0) was added to the cuvettes with films and left for 30 min. After that, the buffer solution was decanted and its enzyme activity was determined according to the described above procedure.

### 4.3.3. Sol-Gel Films Stability Studies

The films doped with enzymes and banana extract were stored in closed cuvettes at +25 ◦C and at +4 ◦C; their stability was checked by measuring the activity according to the described above procedure.

### 4.3.4. Study of the Kinetics of Enzymatic Reactions in the Presence of Immobilized HRP, MT, and BE

To study the activity of HRP immobilized in sol-gel films, 3.0 mL of a hydrogen peroxide solution of various concentrations and 0.4 mL of a 0.08% TMB solution were added to the cuvette. Absorbance was measured at 650 nm for 10 min every 10 s. To calculate the Michaelis constant, the dependence of the reaction rate (min−<sup>1</sup> ) on the concentration of hydrogen peroxide was plotted in Lineweaver–Burk coordinates.

To study the activity of MT and BE immobilized in sol-gel films, 3.0 mL of caffeic acid solution of various concentrations was added to the cuvette and the absorbance was measured at 400 nm for 10–15 min every 10 s. To calculate the Michaelis constant, the dependence of the reaction rate (min−<sup>1</sup> ) on the concentration of caffeic acid was plotted in Lineweaver–Burk coordinates.

### *4.4. Calibration Curves Using HRP-PhTEOS1, MT-PhTEOS1, BE-PhTEOS1 Films*

To obtain a calibration curve for hydrogen peroxide, 3.0 mL of a hydrogen peroxide solution of various concentrations and 0.4 mL of a 0.08% TMB solution were added to a cuvette with HRP-PhTEOS1 film. The difference in absorbance at 650 nm, measured after 1 and 2 min from the reaction start, was used as analytical signal.

To obtain a calibration curve for caffeic acid, 3.0 mL of caffeic acid solution of various concentrations was added to a cuvette with MT-PhTEOS1 or BE-PhTEOS1 films. The reaction rate, i.e., the rate of increase in absorbance at 400 nm, was used as analytical signal.

The limit of detection (LOD) was calculated as 3 standard deviation of the blank absorbance (*n* = 3) divided by the slope value. The limit of quantitation (LOQ) was calculated as 3·LOD.

### *4.5. Total Polyphenol Content Determination*

1.0 g of coffee sample was mixed with 100.0 mL of boiling water, and filtered after 15 min. After cooling to the room temperature, 3.0 mL of the sample solution was added to the cuvette with the BE-PhTEOS1 film, and the reaction rate was used as the analytical signal. The total polyphenol content (TPC) in caffeic acid equivalents was determined in the treated sample using the standard addition method and using the calibration curve for caffeic acid in the range of 2.0–10.0 mM.

The procedure for TPC determination with Folin reagent was carried out similarly to [24].

### *4.6. Instrumentation*

Sols were prepared under the ultrasound radiation using the ultrasound equipment UZH-02 (SonoTech, Russia). The sound energy density (W/mL) was defined as the ratio of the power absorbed in the reactor to the volume of liquid in the reactor. To determine the power, the time was measured until a certain mass of water was heated to a certain temperature.

Spectra of colored products of enzymatic oxidation of phenolic compounds were recorded with SPECTROstar Nano spectrophotometer (BMG Labtech, Ortenberg, Germany). Spectra were analyzed with MARS software (BMG Labtech, Ortenberg, Germany) and statistical analysis was carried out using MS Excel.

**Author Contributions:** Conceptualization, E.I.M.; Validation, E.I.M. and M.A.M.; Formal Analysis, M.A.M.; Investigation, M.A.M..; Resources, E.I.M.; Writing—Original Draft Preparation, M.A.M.; Writing—Review and Editing, E.I.M.; Supervision, E.I.M.; Funding Acquisition, E.I.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** The study was funded by MedEcoTest Ltd. (Grant N 407/14).

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

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** No data is available.

**Acknowledgments:** The authors would like to thank T.G. Khonina (I.Ya. Postovsky Institute of Organic Synthesis, Yekaterinburg, Russia) for providing silicon polyethylene glycol, and MSU students S. Terekhov, T. Ibragimov, A. Bolbat, and D. Anisimov for their participation in the preliminary experiments.

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

### **References**


**Disclaimer/Publisher's Note:** The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.
