*Article* **Transformation of Contaminants of Emerging Concern (CECs) during UV-Catalyzed Processes Assisted by Chlorine**

#### **Edyta Kudlek**

Faculty of Energy and Environmental Engineering, Silesian University of Technology, Konarskiego 18, 44-100 Gliwice, Poland; edyta.kudlek@polsl.pl; Tel.: +48-32-237-24-78

Received: 24 November 2020; Accepted: 6 December 2020; Published: 8 December 2020

**Abstract:** Every compound that potentially can be harmful to the environment is called a Contaminant of Emerging Concern (CEC). Compounds classified as CECs may undergo different transformations, especially in the water environment. The intermediates formed in this way are considered to be toxic against living organisms even in trace concentrations. We attempted to identify the intermediates formed during single chlorination and UV-catalyzed processes supported by the action of chlorine and hydrogen peroxide or ozone of selected contaminants of emerging concern. The analysis of post-processing water samples containing benzocaine indicated the formation of seven compound intermediates, while ibuprofen, acridine and β-estradiol samples contained 5, 5, and 3 compound decomposition by-products, respectively. The number and also the concentration of the intermediates decreased with the time of UV irradiation. The toxicity assessment indicated that the UV-catalyzed processes lead to decreased toxicity nature of post-processed water solutions.

**Keywords:** contaminants of emerging concern; UV; advanced oxidation processes; by-products

#### **1. Introduction**

Contaminants of emerging concern (CECs) have been discovered in every type of natural water worldwide [1–3]. Christen et al. [4] and Kim et al. [5] reported that the exposure to CECs from the group of pharmaceuticals and pesticides causes a deterioration of human health, including the appearance of cancer diseases and cognitive effects. This proves that these compounds are harmful to aquatic organisms and the entire ecosystem fed by waters containing CECs [6].

The literature indicates that many CECs are hardly degradable or completely resistant to conventional water and wastewater treatment methods such as coagulation, sedimentation, filtration, and biological-based processes [7,8]. Even adsorption on activated carbon cannot cope with the removal of compounds with low hydrophobicity from different water matrices [9]. Compounds with phenolic and amine groups or other electron-donating functional groups can be decomposed by selective oxidants such as ozone (O3), hydrogen peroxide (H2O2), or chlorine and chlorine dioxide [10,11].

Water treatment plans still use the chlorination process as one of the most effective methods for water disinfection. Chlorine (Cl2), chlorine dioxide (ClO2), or sodium hypochlorite (NaOCl) reagents are used as disinfectants. However, the oxidizing effect can also be used as an effective and simple method for decomposing organic compounds.

The main oxidizing agent is hypochlorous acid (HOCl), which is formed by the aqueous transformation of Cl<sup>2</sup> according to Equation (1) [12]. HOCl in water solution can dissociate to hypochlorite anions (ClO−) (Equation (2)). In neutral water conditions, ClO<sup>−</sup> are less effective oxidants than HOCl [13].

$$\text{Cl}\_2 + \text{H}\_2\text{O} \rightarrow \text{HOCl} + \text{Cl}^- + \text{H}^+ \tag{1}$$

$$\rm{HOCl} \leftrightarrow \rm{ClO^{-}} + \rm{H^{+}} \tag{2}$$

The addition of another oxidant or precursor of reactive radicals to the reaction matrix, such as H2O<sup>2</sup> or O3, can improve the decomposition of organic compounds in two ways: (1) by the direct action with the compound's molecule or (2) by the initiation of the generation process of other high reactive species [14]. Moreover, implementing the chlorination process together with UV light, can improve the decomposition of contaminant. During UV-catalyzed processes, decomposition of compounds is also caused by their direct photodecomposition and the generation of several radicals (Equations (3) and (4)), like HO• , O−• and reactive chlorine species such as chlorine atoms (Cl• ) and (Cl<sup>2</sup> •−) [15–17]. The oxidation potential of Cl• and Cl<sup>2</sup> •− is 2.47 V and 2.00 V, respectively [18]. Compared to the oxidation potential of HO• , which is equal to 2.80 V, they can also be called strong oxidants. Fang et al. [15] pointed out that Cl• can react more effectively with acetic acid, benzoic acid, and phenol than HO• .

$$\text{HOCl/OCl}^{-} + \text{hv} \rightarrow \text{HO}^{\bullet}/\text{O}^{\bullet-} + \text{Cl}^{\bullet} \tag{3}$$

$$\text{Cl}^{\bullet} + \text{Cl}^{-} \leftrightarrow \text{Cl}\_{2}^{\bullet -} \tag{4}$$

The coexistence of different types of oxidants not always has a positive effect on the decomposition of contaminants. HOCl and ClO−, which did not come into reaction with contaminates or decompose to reactive radicals, can act as a scavenger for HO• and Cl• according to Equations (5)–(8) [15].

$$\text{HCO}^{\bullet} + \text{HOCl} \rightarrow \text{ClO}^{\bullet} + \text{H}\_2\text{O} \tag{5}$$

$$\text{Cl}^\bullet + \text{HOCl} \rightarrow \text{ } + \text{H}^+ \tag{6}$$

$$\rm{HO}^{\bullet} + \rm{OCl}^{-} \rightarrow \rm{ClO}^{\bullet} + \rm{OH}^{-} \tag{7}$$

$$\text{Cl}^{\bullet} + \text{OCl}^{-} \rightarrow \text{ClO}^{\bullet} + \text{Cl}^{-} \tag{8}$$

The exact transformation mechanisms of different CECs during reactive chlorine species coexisting with other oxidants and UV radiation are still unclear. Oxidants and radicals' action removes CECs or deactivates pathogenic microorganisms and reacts with other compounds present in the disinfected water matrix, leading to several harmful chlorination by-products. The most known chlorination intermediates are trihalomethanes (THMs), and haloacetic acids (HAAs) [19,20], but also chlorination by-products of several CECs are detected in chlorinated water.

The paper presents an attempt to identify the intermediates of selected CECs formed during UV-catalyzed oxidation processes conducted in the presence of chlorine supported by the action of hydrogen peroxide and ozone. Identification of transformation products arising in real water matrices and the determination of their toxicity will give an accurate picture of their potentially dangerous impact on aquatic ecosystems. The comprehension of the CECs decomposition pathways caused by specific physicochemical factors will allow for the development of effective hybrid methods for eliminating the aquatic environment. A single chlorination process was carried out to estimate the decomposition ability of chlorine. The action of chlorine was supported by UV irradiation compared with O<sup>3</sup> and H2O2. CECs from the group of pharmaceutical compounds, dye additives, and synthetic hormones were introduced into water solutions based on deionized and surface water and subjected to oxidation agents. The generated by-products were extracted from the post-processed water solutions by the use of Solid Phase Extraction and then chromatographically analyzed and identified based on their mass spectra compared to the National Institute of Standards and Technology NIST v17 database. Bacterial-based toxicological tests confirmed the potentially toxic nature of the intermediates.

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

Experiments based on the single chlorination and UV-catalyzed chlorination processes assisted by the action of O<sup>3</sup> and H2O<sup>2</sup> of CECs water solutions were performed. These processes were assessed

both in terms of removing individual compounds and the formation of decomposition by-products and their toxicity to living water organisms.

#### *2.1. Chlorination of CECs*

The first stage of the study was devoted to assessing the effectiveness of the CECs' decomposition during dark chamber chlorination, where chlorine was introduced into the water in the form of NaOCl. The influence of chlorine dose on the decomposition of tested CECs in deionized water solutions and surface water solutions is presented in Figures 1 and 2, respectively. The decomposition of compounds occurs during action with HOCl and ClO<sup>−</sup> according to Equation (9) [13], leading to the formation of new products, which also should be decomposed. effectiveness of the CECs'

$$\text{HOCl/OCl}^- + \text{organic contaminant} \rightarrow \text{product} \tag{9}$$

−

Sivey and Roberts [21] demonstrated that Cl<sup>2</sup> under low pH values and Cl2O could act as active chlorination agents during the chlorination process and interact with compounds in the treated water solutions. All tested CECs have an aromatic ring in their structure. Therefore, their decomposition should occur due to specific reactions on certain moieties bound to the aromatic ring and the electrophilic substitution of chlorine in the *ortho* or *para* position [22]. HOCl/OCl − + organic contaminant → product

**Figure 1.** Change of (**a**) IBU, (**b**) BE, (**c**) ACR, and (**d**) E2 concentration during single chlorination of deionized water solutions.

**Figure 2.** Change of (**a**) IBU, (**b**) BE, (**c**) ACR, and (**d**) E2 concentration during single chlorination of surface water solutions.

− − − − − As expected, the removal degree of all tested compounds increased with the increase of the chlorine concentration. IBU and BE were the least susceptible to the action of chlorine. No increase of IBU concentration was observed after 2 min reaction time for chlorine doses equal to 0.5 and 1.0 mg L −1 in deionized and surface water solutions. The BE removal noted after a 2 min reaction time for the chlorine dose equal to 0.5, 1.0, 2.0, and 3.0 mg L <sup>−</sup><sup>1</sup> was only 0.2, 0.7, 1.1, and 1.6%, respectively, in deionized water solutions. An extension of the reaction time to 20 min by the chlorine dose equal to 0.5 mg L <sup>−</sup><sup>1</sup> allowed for a 3.1% removal of these CECs and an 8% removal for the chlorine content equal to 3.0 mg L −1 . Meanwhile, the IBU concentration decreased only by 9% after 30 min of process elongation by the 3.0 mg L −1 chlorine dose. Similar low IBU decomposition ability under chlorine action was noted by Xiang et al. [23]. The removal degree of this compound did not exceed 3.1% after 20 min of dark chlorination. The obtained results reconfirmed the recalcitrance of IBP to chlorination.

− − Higher decomposition rates were noted for ACR. This compound's concentration in the presence of 3.0 mg L <sup>−</sup><sup>1</sup> of chlorine decreased by over 20% after 10 min of single dark chlorination and over 26% after 30 min of process duration. It can be concluded that single NaOCl as a source of HOCl and ClO<sup>−</sup> is not sufficient for the decomposition of IBU, BE, and ACR.

− − A reverse observation was noted in the case of the hormone E2 chlorination. The addition of only 0.5 mg L <sup>−</sup><sup>1</sup> of chlorine to the compound solution based on deionized water led to over 75% and 98% decomposition after 10 and 30 min, respectively. Meanwhile, a complete removal of E2 was observed after 30 min of reaction with the chlorine dose equal to 2 mg L −1 . Li et al. [24] also observed a complete removal of this hormone during the chlorination process carried out in neutral conditions.

Similar test results to those obtained for deionized water were noted for a test carried out on surface water (Figure 2). Only the removal degrees of ACR increased for all tested chlorine doses at about 5% and reached, for example, for the free chlorine dose equal to 3.0 mg L <sup>−</sup><sup>1</sup> 30%. Therefore, it can be assumed that organic and inorganic matter in surface water promotes the decomposition of this compound under the influence of chlorine. However, further research in this area is required to determine which component of the natural water matrix is responsible for increasing the removal degree of ACR. − −

Special attention should be paid to inorganic compounds occurring in real water solutions, i.e., NO<sup>3</sup> <sup>−</sup> and NH<sup>4</sup> <sup>+</sup>, whose presence was confirmed in the tested water matrixes. Those compounds can influence the decomposition of compounds and can react with reactive forms of chlorine. However, Qiang and Adams [25] indicated a negligible chlorine reactivity with NH<sup>4</sup> <sup>+</sup>, although NO<sup>3</sup> and N<sup>2</sup> can be formed during reaction of HOCl with NH3. This leads to the formation of mono-, diand trichloramine. − −

#### *2.2. Decomposition of CECs in UV-Catalyzed Chlorination Processes*

The unsatisfactory low removal degree of IBU, BE, and ACR noted during single chlorination forces another treatment process. Therefore, processes integrated the action of chlorination with UV irradiation, additionally supported by the presence of H2O<sup>2</sup> or O<sup>3</sup> were implemented. Figures 3 and 4 compared the removal degrees of tested CECs noted in deionized and surface water solution exposed to UV irradiation in the presence of NaOCl and H2O<sup>2</sup> (UV/NaOCl/H2O<sup>2</sup> process) or O<sup>3</sup> (UV/NaOCl/O3), respectively.

**Figure 3.** Change of CECs concentration during the UV/NaOCl/H2O<sup>2</sup> conducted on (**a**) deionized and (**b**) surface water solutions.

**Figure 4.** Change of CECs concentration during the UV/NaOCl/O<sup>3</sup> conducted on (**a**) deionized and (**b**) surface water solutions.

The applied UV light source, during the reactions with HOCl, ClO<sup>−</sup> (Equations (3) and (4)) and O<sup>3</sup> as well as H2O2, induced the formation of several reactive radicals which are responsible for a non-selective decomposition of compounds. It should also be emphasized that UV light as an electromagnetic wave carries energy in the form of photons. The interaction of one or more photons with a given compound leads to the chemical transformation of the bonds between atoms that make up the compound molecule [26]. This process results in the photodecomposition of the compound molecule. The maximum absorbance of IBU, BE, and E2 were below the wavelengths, which can get into the irradiated solution. However, those compounds and others occurring, especially in surface water matrixes, can still absorb UV radiation energy and undergo the photo-decomposition process. The energy needed for the dissociation of an H–OH bound, and the formation of HO• exceeds 5 eV [27]. Therefore, the simultaneous application of UV irradiation with different wavelengths and the effect of chlorine and O<sup>3</sup> or H2O<sup>2</sup> allows for obtaining a required number of free radicals for the decomposition of CECs and their intermediates.

In both UV/NaOCl/H2O<sup>2</sup> and UV/NaOCl/O<sup>3</sup> processes, preferable removal degrees of BE, ACR, and E2 were observed for compounds occurring in deionized water solutions. This difference in process effectiveness was especially notable during the UV/Cl2/H2O<sup>2</sup> process. For example, 2 min of process implementation led to a 60% removal of E2 in the surface water solution, while removing this compound noted in the deionized water matrix exceeded 93%. After 20 min of process duration, complete removal of E2 in deionized water was noted. The concentration of this contaminant in surface water was reduced by 80% (20 min of process duration). The final removal of BE and ACR noted in deionized water samples was equal to 27 and 53%, whereas removing these compounds observed in surface water reached only 22 and 43%, respectively. Only in the case of IBU were higher removal degrees were noted in surface water solutions. For example, after 20 min of UV/NaOCl/H2O<sup>2</sup> process, the concentration of IBU decreased by 35% in deionized water solutions and over 41% in surface water solutions. Previous studies [28] on the influence of organic and inorganic compounds on the decomposition of IBU in UV-catalyzed processes indicated that the presence of Ca2+, Mg2+, NH<sup>+</sup> 4 , Cl<sup>−</sup> , CO2<sup>−</sup> 3 , HCO<sup>−</sup> 3 , HPO2<sup>−</sup> 4 as well as SO2<sup>−</sup> 4 ions increased the decomposition process of this pharmaceutical compound.

Higher compound removal degrees were noted during the UV/NaOCl/O<sup>3</sup> process. E2 occurring in deionized water was completely removed after 5 min of UV irradiation, and after 20 min of process implementation, complete removal of this compound was also noted in the surface water matrix. BE and ACR concentration was reduced in deionized water by over 63% and in the surface water only by 55%. Meanwhile, the removal degree of IBU in surface water reached a value of 65%.

It can be concluded that the higher effectiveness of the UV/Cl2/O<sup>3</sup> against the UV/Cl2/H2O<sup>2</sup> process was the result of the formation of a larger number of radicals during the O<sup>3</sup> self-decomposition in water. Among these radicals, HO• , HO<sup>2</sup> • , HO<sup>3</sup> • , HO<sup>4</sup> • , O<sup>2</sup> •− and O<sup>3</sup> •− can be mentioned [29]. Whereas the irradiation of H2O<sup>2</sup> with UV light leads to the formation of OH• radicals [30]. However, HO• radicals are endowed with the strongest oxidation potential, and they can abate compounds that are resistant to O<sup>3</sup> or H2O<sup>2</sup> decomposition [31]. It should also be mentioned that UV irradiation leads to an increase of the quantum yields and the molar absorption coefficients of OCl, OCl, leading to a higher and faster production of radicals during the decomposition processes [15,32,33].

#### *2.3. Identification of Decomposition By-Products*

The implementation of the tested CECs decomposition processes allows for a decrease of the contaminant's concentrations and leads to the generation of several by-products. Those intermediates were formed during reactions between the parent compounds and chlorine and/or other reactive species. The intermediates were identified based on their mass spectra using the NIST v17 database (Table 1). The identified by-products decomposed more slowly than the parent micropollutants and were detected even after 20 min of UV-catalyzed process implementation. The single chlorination process led to four BE intermediates: Ethyl 4-chlorobenzoate, 4-Chloroaniline, Chlorohydroquinone, and 2,5-Dichlorohydroquinone, while the chlorination of ACR solutions resulted in the formation

of 9-Chloroacridine and Salicylic acid. The subjection of IBU water solutions to single dark chlorination resulted in three intermediates: 1-(4-Isobutylphenyl)ethanol, 4-Acetylbenzoic acid, and 4-Ethylbenzaldehyde. Intermediates detected in each process are summarized in Table 2. The chlorination process led to small modifications in the structures of the compounds and the formation of more oxidized or chlorinated molecules. This is connected because HOCl can react with organic compounds in three types of reactions: oxidation reactions, addition reactions to unsaturated bonds, and electrophilic substitution reactions at nucleophilic sites [34]. Electrophilic substitution reactions are considered the most common mechanisms during chlorination; therefore, chlorine substitution sites will most likely be on the tested compounds' aromatic ring [35].


**Table 1.** Identified CECs by-products during the performed experiments.


\* - — not detected; + — detected.

During the implemented decomposition processes, the IBU intermediates were mainly formed at the first stage by hydroxylation and chlorine substitution. However, the implemented GC-MS analysis does not allow for detecting IBU by-products with chlorine atoms in their structure. Such compounds were identified by Li et al. [36] and Xiang et al. [23]. The possible decomposition pathway of IBU with the identified intermediates is shown in Figure 5.

' **Figure 5.** Possible decomposition of (**1**) IBU with the identified intermediates (**2**) 1-Hydroxyibuprofen, (**3**) 1-(4-Isobutylphenyl)ethanol, (**4**) 4'-Isobutylacetophenone, (**5**) 4-Acetylbenzoic acid, and (**6**) 4-Ethylbenzaldehyde.

• Ethyl 4-hydroxybenzoate and Ethyl 4-chlorobenzoate were formed by the denitration of the BE molecule and the substitution of the nitric group by the hydroxyl group and chlorine, respectively. Other BE intermediates were possibly generated by the attack of OH• and chlorine on the compound's phenolic ring. The possible decomposition pathway of BE is summarized in Figure 6.

**Figure 6.** Possible decomposition of (**1**) BE with the identified intermediates (**2**) Ethyl 4-hydroxybenzoate, (**3**) Ethyl 4-chlorobenzoate, (**4**) 4-Chloroaniline, (**5**) 4-Chlorophenol, (**6**) 3,4-Dichlorophenol, (**7**) Chlorohydroquinone, and (**8**) 2,5-Dichlorohydroquinone.

The ACR decomposition by-products Acridone, Acridine-10-oxide, and 2-Hydroxyacridine were mainly formed by the attack of reactive oxygen species of the compound molecule. Meanwhile, 9-Chloroacridine results from the substitution of chlorine to the phenolic ring (Figure 7). Further hydroxylation and the deamination of the formed intermediates led to carbon atom ring-opening and Salicylic acid formation, which was subject to further decomposition.

**Figure 7.** Possible decomposition of (**1**) ACR with the identified intermediates, (**2**) Acridone, (**3**) Acridine-10-oxide, (**4**) 2-Hydroxyacridine, (**5**) 9-Chloroacridine, and (**6**) Salicylic acid.

— The applied analytical method based on gas chromatography allows only for the detection of three E2 intermediates, resulting from reactive oxygen species action (Figure 8). Steroid hormones are typically composed of four carbon atoms rings—three cyclohexane rings and one cyclopentane ring. The reactive species, in general, firstly attacks the first cyclohexane ring, which leads to the formation of hydroxylated compounds like 2-Hydroxyestradiol. Li et al. [24] pointed out that the decomposition of E2 occurs by the halogenation of the aromatic ring followed by the cleavage of the benzene moiety and chlorine substitution formation generation of THMs and HAAs from phenolic intermediates.

#### *2.4. Toxicological Assessment*

A toxicological assessment is necessary to determine whether the proposed decomposition processes of pollutants do not deteriorate the quality of treated water solutions. It has already been proved that the initial contaminants occurring in non-treated water were sometimes less toxic than the intermediates detected in post-processed water solutions [37]. The preformation of fast toxicological tests like the bacterial-based Micrtox ® gives a quick response about the impact of the post-processed water on living organisms. Studies indicated a good interspecies correlation between saltwater bacteria like the used *Aliivibrio fischeri* and other freshwater bacteria or fishes [38]. The *Aliivibrio fischeri* are also considered extremely sensitive to a wide range of pollutants and reagents occurring in water [39].

**Figure 8.** Possible decomposition of (**1**) IBU with the identified intermediates, (**2**) 2-Hydroxyestradiol, (**3**) Estradiol-3,4-quinone, and (**4**) 4-(1-Hydroxyethyl)phenol.

It should be noted that chlorine is used in water treatment processes to protect water before secondary contamination [40]. Therefore, it should be toxic to pathogens and small test organisms. Before the proper toxicological test, the adopted sample preparation methodology allowed for the exclusion of chlorine influence on test bacteria. Chlorinated water samples without the addition of tested CECs were characterized by a toxicological effect lower than 9% (Figure 9), which classified them, according to the guidelines given by Mahugo Santana et al. [41], as non-toxic. Therefore, the results presented in Figures 10–12 resulted from generated parent compound intermediates.

**Figure 9.** Change in the toxicity of chlorinated (**a**) deionized and (**b**) surface water samples without the addition of CECs.

−

toxic (toxic effect ≤ 25.0%).

− level (50.0% < toxic effect ≤ 75.0%) and the E2 solution toxicity to a low tox

**−1**

**−1**

toxic (toxic effect ≤ 25.0%).

**Figure 10.** Change in the toxicity of (**a**) deionized water and (**b**) surface water solution after single chlorination (30 min) and the addition of CECs (dashed lines indicate the boundaries between the toxicity classes).

**Figure 11.** Change in the toxicity of the tested (**a**) deionized water and (**b**) surface water solution after the UV/NaOCl/H2O<sup>2</sup> process (dashed lines indicate the boundaries between the toxicity classes).

**Figure 12.** Change in the toxicity of the tested (**a**) deionized water and (**b**) surface water solution after the UV/NaOCl/O<sup>3</sup> process (dashed lines indicate the boundaries between the toxicity classes).

ACR-containing surface water samples before the addition of NaOCl had a toxic effect which exceeded 28%, and according to the toxicity classification, it should be assigned as low toxic (25.0% < toxic effect ≤ 50.0%) against the test bacteria. Whereas IBU, BE, and E2 in the tested concentration of 500 µg L <sup>−</sup><sup>1</sup> were non-toxic (toxic effect ≤ 25.0%).

The presence of chlorine during the implementation of a single chlorination process initiated the compounds' decomposition, leading to an increase in toxicity. It was noted that the toxic effect increased with the increase of the chlorine dose in both deionized and surface water solutions (Figure 10). For example, BE and E2 solutions treated by 0.5 and 1.0 mg L−<sup>1</sup> of chlorine were still non-toxic, while the dose of 3.0 mg L−<sup>1</sup> resulted in the increase of the BE solution toxicity to a toxic level (50.0% < toxic effect ≤ 75.0%) and the E2 solution toxicity to a low toxic level. The chlorination process's implementation did not significantly affect the toxicity of the IBU-containing solution, which remained non-toxic for each tested dose of chlorine. The highest toxicity during single chlorination was observed for ACR post-processed solution. The addition of 0.5 or 1.0 mg/L of chlorine to the ACR water solution led to the formation of several toxic intermediates in deionized and surface water solutions, which increased the toxicity to a toxic level. The addition of 2.0 and 3.0 mg/L of chlorine resulted in the generation of highly toxic solutions (toxic effect > 75.0%). The toxicity noted for deionized water solutions containing BE, ACR, and E2 (Figure 10a) was 5%, 8%, and 2%, respectively, higher than the toxicity noted for surface water solutions (Figure 10b).

The post-processed samples obtained during the UV-catalyzed treatment methods were also subjected to toxicological tests (Figures 11 and 12). It was noted that the implementation of UV irradiation supported by the action of 1.0 mg L−<sup>1</sup> chlorine and H2O<sup>2</sup> or O<sup>3</sup> resulted in an increase of the toxicity of all tested compound solutions in the first 5 min of the treatment process compared to the single chlorination process. This phenomenon confirms the generation of toxic by-products identified during the chromatographic analysis. In general, the toxicity of samples collected after the process carried out with the presence of O<sup>3</sup> (UV/NaOCl/O3) was higher than the toxicity of samples subjected to the simultaneous action of UV light, chlorine, and H2O<sup>2</sup> (UV/NaOCl/H2O2).

For example, E2 solutions irradiated for 2, 10, and 20 min in the presence of H2O<sup>2</sup> were characterized by non-toxicity. Only the sample after 5 min of UV/NaOCl/H2O<sup>2</sup> was low toxic. Samples of the same contaminant subjected to the UV/NaOCl/O<sup>3</sup> process were classified after 5 min as near medium toxic, but after 20 min, their toxicity was reduced to a non-toxic level. The chromatographic analysis indicated that the signals caused by the formed intermediates after 10 min of UV irradiation become weaker than those after 5 min. This means that after this time of the process, the concentration of by-products decreased. This was also reflected in the toxicity results. All tested compounds' toxicity increased in the first 5 min of UV-catalyzed process implementation and then decreased after 10 and 20 min of process duration. For example, the toxicity of the ACR-containing surface water solution treated by UV/NaOCl/H2O<sup>2</sup> decreases from a high toxic level to a medium toxic level. Donner et al. [42] noted an increase in the toxicity during the irradiation of carbamazepine solutions with UV light in the first 30 min. ACR is a toxic decomposition by-product of carbamazepine. The occurrence of this compound and other carbamazepine intermediates increased toxicity. Further irradiation of the pharmaceutical solution leads to the decomposition of acridine and decreased solution toxicity.

The BE and E2-containing solutions' toxicity decreased during both UV/NaOCl/H2O<sup>2</sup> and UV/NaOCl/O<sup>3</sup> processes from a low toxic to a non-toxic level. Only IBU solutions subjected to the UV/NaOCl/O<sup>3</sup> were characterized by increasing toxicity during the first 2, 5, and 10 min. Moreover, the surface water IBU solution after a 10 min exposure to the UV/NaOCl/O<sup>3</sup> process was characterized by a toxic level. It can be summarized that the decomposition of compounds does not always have a beneficial impact on the treated water quality. The generated compound oxidation by-products can radically influence water toxicity and force the necessity to implement further and more complex treatment processes.

#### **3. Materials and Methods**

#### *3.1. Water Samples*

The research subject constituted water solutions prepared on deionized water witch a conductivity of 18 MΩ cm−<sup>1</sup> and surface water matrices (conductivity of 0.152 mS cm−<sup>1</sup> , TOC of 1.546 mg L−<sup>1</sup> , COD of 89 mg O<sup>2</sup> L −1 , N-NH<sup>4</sup> of 0.9 mg L−<sup>1</sup> , N-NO<sup>3</sup> of 2.1 mg L−<sup>1</sup> ) spiked with CECs standards. Ibuprofen sodium salt (IBU), benzocaine (BE), acridine (ACR), and β-estradiol (E2) were chosen as

representative compounds from the group of pharmaceutical compounds, dye additives, and synthetic hormones. The concentration of the compounds in the prepared water solutions was set on 500 µg L −1 . Each compound standard solution was prepared by dissolving 10 mg of each analyte in 10 mL of methanol. The use of compound standard solutions allows for the complete dissolution of the analytes and obtains precisely defined CECs concentrations. Standards of the tested CECs with a purity of over 99%, 97%, and 98% were purchased from Sigma-Aldrich (Pozna ´n, Poland). The characteristic of the compound is presented in Table 3. The selected compound concentrations, which exceeded the usual environmental concentrations, allowed for an increase in the accuracy of the analytical measurements. Absorption spectra of all tested compounds were measured using the Spectroquant Pharo 300 UV/Vis spectrophotometer by Merck (Darmstadt, Germany) and compared in Figure 13. The maximum absorbance λ*max* for IBU, BE, ACR, and E2 was estimated for both deionized and surface water matrixes and were set on 274 nm, 286 nm, 340 nm 280 nm, respectively. conductivity of 18 MΩ cm**−1 −1 −1 −1 −1 −1** and β **−1** Aldrich (Poznań, Poland). The *λ* conductivity of 18 MΩ cm**−1 −1 −1 −1 −1 −1** and β **−1** Aldrich (Poznań, Poland). The *λ* conductivity of 18 MΩ cm**−1 −1 −1 −1 −1 −1** and β **−1** Aldrich (Poznań, Poland). The *λ*


**Table 3.** Characteristics of the tested organic compounds [43].

**−1 Figure 13.** UV-VIS spectrum of (**a**) IBU, (**b**) BE, (**c**) ACR, and (**d**) E2 in deionized and surface water samples.

**−1**

−1 −1

The experiments for all CECs were performed separately in neutral conditions. The pH of each tested water solution was adjusted to 7.0 using 0.1 mol L−<sup>1</sup> NaOH or 0.1 mol L−<sup>1</sup> HCL. Preliminary studies indicated that the very low volumes of the added alkali or acid did not influence the decomposition of the tested micropollutants.

#### *3.2. Decomposition Processes*

All prepared water solutions were subjected to the action of the chlorination process. It was carried out using sodium hypochlorite (NaOCl) with a nominal free chlorine content of 6% (*w*/*v*) purchased from Chemoform (Sosnowiec, Poland). The experiment was conducted on four different chlorine doses equal to 0.5, 1.0, 2.0, and 3.0 mg L−<sup>1</sup> and measured as a total chlorine concentration by the use of the HI-93414-02 EPA Compliant Turbidity and Free & Total Chlorine Meter by HANNA Instruments Inc. The chlorine doses were selected as part of preliminary studies considering doses used for the chlorination of tap water under normal and special (emergency water pollution) operating conditions. Therefore, these are doses that can be introduced into the water by any water treatment station that uses chlorination as a water disinfection method.

The single chlorination process was carried out in a dark chamber to omit the influence of any light source on the chlorine caused decomposition of tested compounds. The water samples were also exposed to chlorine's action in the presence of UV irradiation supported by hydrogen peroxide (H2O2) or ozone (O3). The H2O<sup>2</sup> and O<sup>3</sup> dose used in this study was estimated in preliminary tests and set on 3.0 mg L−<sup>1</sup> . The single chlorination process was carried out 2, 10, 20, and 30 min and stopped by sodium thiosulphate (Na2S2O3) at a dose of 100 mg L−<sup>1</sup> , which acts as an excess chlorine removing agent. Na2S2O<sup>3</sup> with a purity of 98% was purchased from Merck KGaA (Darmstadt, Germany). O<sup>3</sup> was generated by an ozonation machine Ozoner FM500 from WRC Multiozon (Gda ´nsk, Poland) and introduced in the tested water samples using a ceramic diffuser with a height of 25 mm, and a diameter of 12 mm. The O<sup>3</sup> concentration in the water solutions was measured using a photometric method on the Spectroquant® by Merc Sp. z o.o. (Warszawa, Poland). The ozonation reaction was stopped after the UV irradiation time by sodium sulfite Na2SO<sup>3</sup> at a concentration of 24 mmol L−<sup>1</sup> . A 150 Watt medium-pressure mercury lap placed in a glass cooling sleeve by Heraeus (Hanau, Germany) was used as the UV light source during all UV-catalyzed decomposition methods. The irradiation time was set as 2, 5, 10, and 20 min, and the radiation flux emitted by the lamp is summarized in Table 4. Figure 14 shows the characteristic of radiation wavelengths emitted by the used UV lamp, and Table 5 shows the energy of light, which reached the water matrix. The energy of light was calculated based on the multiplication of the Planck's constant with the frequency of light.

Experiments for all tested compounds were carried out separately and repeated three times.

**Table 4.** Physicochemical characteristics of the studied CECs (Data achieved from the supplier of Heraeus UV lamp reactors KENDROLAB Sp. z o.o.).




Wavelength λ,

**−1**

**−1**

**Ф, W**

−

generated by an ozonation machine Ozoner FM500 from WRC Multiozon (Gdańsk, Poland) and

on the multiplication of the Planck's

**Figure 14.** Characteristic of radiation wavelengths emitted by the used UV lamp (Data achieved from the supplier of Heraeus UV lamp reactors KENDROLAB Sp. z o.o.).

#### *3.3. Analytical Procedure and Toxicity Assessment*

**Wavelength λ, nm 7 2 3 4 6 0 5 6 7 8 6 2 46**  The analytical procedure of the tested CECs was adopted from previous studies [11] and based on the extraction of analytes from water matrixes by solid-phase extraction (SPE) and their chromatographic analysis. The SPE was performed by the use of Supelclean™ ENVI-8 and Supelclean™ ENVI-18 cartridges obtained from Sigma-Aldrich (Pozna ´n, Poland) with a silica gel base material with C8 (octyl) and silica gel base material with C18 (octadecyl) bonding bed type, respectively. The used SPE procedure allowed for obtaining a recovery of the tested CECs, which exceeded 95%.

The chromatographic analysis was conducted using the GC-MS (EI) chromatograph model 7890B by Perlan Technologies (Warszawa, Poland). The chromatograph was equipped with an SLB TM—5 ms 30 m × 0.25 mm capillary column of 0.25 µm film thickness from Sigma-Aldrich (Pozna ´n, Poland). The applied column oven temperature program was: 80 ◦C (6 min), 5 ◦C/min up to 260 ◦C, 20 ◦C/min up to 300 ◦C (2 min). Helium 5.0 was used as a carrier gas during the analyses. The temperatures of the ion trap, ion source, and column injector were equal to 150 ◦C, 230 ◦C, and 250 ◦C, respectively. All SPE extract were analyzed twice, in the selected ion monitoring (SIM) mode to monitor CECs concentration and in the total ion current (TIC) mode for the identification of generated compound decomposition by-products. The TIC mode was performed in the range from 50 to 400 m/z.

The percentage of removal of each CECs after the implementation of decomposition processes was calculated according to Equation (10), where C<sup>i</sup> and C<sup>p</sup> are the initial and post-processed compound concentrations in mg L −1 , respectively [44]:

$$\text{Removal} \left( \% \right) = \frac{\text{C}\_{\text{i}} - \text{C}\_{\text{P}}}{\text{C}\_{\text{i}}} \cdot 100 \tag{10}$$

The post-processed samples were also subjected to a spectroscopic analysis performed on the Spectroquant Pharo 300 UV/Vis spectrophotometer by Merck (Darmstadt, Germany), which can measure the UV-VIS spectrum of samples in the range from 200 to 600 nm. The spectroscopic measurement indicated only the decrease of the initial compound concentration (it was discussed in Sections 2.1 and 2.2. based on the percentage of removal calculated from the data obtained by the GC-MS analysis). The UV-VIS spectrum did not indicate the formation of intermediates, which was related to their low concentrations. Therefore, the results were not presented in the paper.

The toxicological evaluation, which gives an answer about the potentially toxic nature of the newly generated decomposition by-products, was carried out using the Microtox ® test. The test procedure

is based on the measurement of the changes in the behavior of saltwater bioluminescent bacteria *Aliivibrio fischeri* according to the Screening Test procedure of MicrotoxOmni system, which controls the work of the Microtox analyzer Model 500 by Modern Water (London, United Kingdom). The test results were expressed as a percentage of bacterial bioluminescence inhibition caused by changes in the bioindicators' metabolic processes exposed to a toxicant for 15 min. The results were compared to a reference nontoxic sample (2% NaCl solution). The collected samples were subjected to toxicological test after 24 h incubation in a cooled dark chamber to exclude the possible chlorine impact, which was left in the post-processed water solutions.

Assignment errors marked on figures presented in this paper were estimated based on the standard deviation for three repetitions of each test. The error values for all tested samples did not exceed 2.0%.

#### **4. Conclusions**

Based on the conducted compound decomposition assessments in the selected oxidation processes, it can be concluded that the lowest compound removal degrees were observed during single dark chlorination. Moreover, the decomposition rates of all tested compounds obtained in this process conducted in deionized water matrixes were very similar to those noted in surface water matrixes. The difference in the compound's removal degrees in both matrices did not exceed 2% and was within the measurement error. Further, the occurrence of chlorine in the reaction matrixes leads to the generation of intermediates with chlorine atoms in their structure. It was also noted that the compound concentration decreases with the increase of the chlorine concentration. Higher chlorine concentrations also lead to an increase of the by-products forming. The conduction of UV-catalyzed oxidation processes shows that the combination of the UV radiation with the action of chlorine and O<sup>3</sup> was more effective for compound decomposition than the UV chlorination process supported by the presence of H2O2. The implementation of the UV radiation in oxidants' presence results in the decrease of the number and concentration of formed by-products after 20 min of process elongation. However, it should be noted that during the first 10 min of both UV/NaOCl/H2O<sup>2</sup> and UV/NaOCl/O<sup>3</sup> processes, several oxidized intermediates were formed. The UV-catalyzed processes also lead to the decrease of the toxicity of the post-processed water solutions, which still depend on the type of decomposed compounds and the UV irradiation time.

**Funding:** The studies were performed within the project's framework founded by the Polish Ministry of Science and Higher Education grant no. 08/040/BKM20/0138.

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

#### **References**


**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2020 by the author. 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 (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Cleaner Production of Epoxidized Cooking Oil Using A Heterogeneous Catalyst**

#### **Maria Kura ´nska \* and Magdalena Niemiec**

Department of Chemistry and Technology of Polymers, Cracow University of Technology, Warszawska 24, 31-155 Kraków, Poland; mniemiec@pk.edu.pl

**\*** Correspondence: maria.kuranska@pk.edu.pl; Tel.: +48-126282747

Received: 15 October 2020; Accepted: 28 October 2020; Published: 30 October 2020

**Abstract:** A cleaner solvent-free process of used cooking oil epoxidation has been developed. The epoxidation reactions were carried out using "in situ"-formed peroxy acid. A variety of ion exchange resins with different cross-linking percentages and particle sizes such as Dowex 50WX2 50-100, Dowex 50WX2 100-200, Dowex 50WX2 200-400, Dowex 50WX4 50-100, Dowex 50WX4 100-200, Dowex 50WX4 200-400, Dowex 50WX8 50-100, Dowex 50WX8 100-200, Dowex 50WX8 200-400 were used in the synthesis as heterogeneous catalysts. No significant effect of the size as well as porosity of the catalysts on the properties of the final products was observed. In order to develop a more economically beneficial process, a much cheaper heterogeneous catalyst—Amberlite IR-120—was used and the properties of the epoxidized oil were compared with the bio-components obtained in the reaction catalyzed by the Dowex resins. The epoxidized waste oils obtained in the experiments were characterized by epoxy values in the range of 0.32–0.35 mol/100 g. To reduce the amount of waste, the reusability of the ion exchange resin in the epoxidation reaction was studied. Ten reactions were carried out using the same catalyst and each synthesis was monitored by determination of epoxy value changes vs. time of the reactions. It was noticed that in the case of the reactions where the catalyst was reused for the third and fourth time the content of oxirane rings was higher by 8 and 6%, respectively, compared to the reaction where the catalyst was used only one time. Such an observation has not been reported so far. The epoxidation process with catalyst recirculation is expected to play an important role in the development of a new approach to the environmentally friendly solvent-free epoxidation process of waste oils.

**Keywords:** ion exchange resins; waste cooking oil; reuse of catalyst; epoxidation; Circular Economy

#### **1. Introduction**

The increasing price of petrochemical raw materials, their limited availability and the growing problem of environmental pollution draw the attention of the chemical industry to sustainable development. One of its assumptions is searching for new renewable raw materials that can be successfully used in the synthesis of chemical compounds. An example of such raw materials is vegetable oils, which in terms of chemical structure consist of triglycerides, i.e., esters of glycerol and three fatty acids, mainly unsaturated [1–3]. Waste vegetable oils are also an interesting raw material [3–6]. Syntheses based on such materials are a more ecological solution owing to the possibility of managing waste generated during the frying process. Such an approach implements the requirements of Circular Economy.

Double bonds in fatty acid chains are reactive sites that allow chemical modifications of vegetable oils to increase their possible applications. One of such modifications is the epoxidation process. During this process unsaturated bonds are oxidized to epoxy groups, which are also called oxirane rings [1]. Among several epoxidation methods, the most important process is the use of carboxylic

peracids as an oxidizing agent. Industrially, peracetic acid is the most commonly used material, although the process can also be carried out using performic acid, perfluoroacetic acid, perbenzoic acid, m-chloroperbenzoic acid, and m-nitroperbenzoic acid [7]. Carboxylic peracid is formed by the reaction of a proper organic acid with hydrogen peroxide in the presence of a catalyst, usually in the form of strong mineral acids, acidic ion exchange resins (AIER), or enzymes [8].

The process of epoxidation can be carried out in one or two stages. The one-step method is often called in situ epoxidation. In this process, all components, i.e., vegetable oil, organic acid, hydrogen peroxide and catalyst, are mixed in one reaction vessel. In the two-step method, two reactors are used. In the first one, peracid is obtained and then placed in the second vessel where the main epoxidation of the vegetable oil is carried out [9].

During the epoxidation process of vegetable oils, side reactions may occur, especially when the process is carried out in the presence of strongly acidic catalysts. The type of catalyst has a significant impact on the epoxidation process. On an industrial scale, homogeneous catalysts, mainly strong mineral acids, are most commonly used. The use of such compounds leads to a reduction of the process costs and allows obtaining epoxidized oils with a low content of unsaturated bonds [10]. Examples of mineral acids used in the epoxidation process of vegetable oils, both fresh and waste, are H2SO4, HNO3, H3PO4, HCl. It has been found that among these catalysts, H2SO<sup>4</sup> is the most efficient and effective [7,11,12]. However, the disadvantage of homogeneous catalysts is the relatively low selectivity of converting unsaturated bonds into epoxy groups. This effect is caused by the occurrence of side reactions, especially the oxirane ring-opening reaction, which are catalyzed by strong mineral acids. Moreover, the process in the presence of these catalysts leads to considerable amounts of waste water that is difficult to purify [10].

Improvement of process selectivity, by reducing side reactions, can be achieved using heterogeneous catalysts such as acidic ion exchange resins [13]. The most commonly used resins are Amberlyst 15, Amberlite IR-120, and Dowex 50WX2. These catalysts are copolymers of styrene and divinylbenzene and differ in the form and content of the cross-linking agent. It has been proved that the type of the ion exchange resin that is used has an impact on the epoxidation process of vegetable oils [14]. In addition, increasing the concentration of the heterogeneous catalyst usually results in increasing the conversion of unsaturated bonds into epoxide groups [15–18]. The advantages of using acidic ion exchange resins include also the ease of separating them from the finished product and the possibility of re-use. Even after several uses, the loss of activity of this type of catalyst is observed to be insignificant and the efficiency of the process is slightly reduced [15,19,20].

The oxidation of double bonds in fatty acid chains can also be carried out in the presence of enzymatic catalysts, thanks to which the epoxidation process of vegetable oils is more environmentally friendly. What is more, this method allows reducing the number of side reactions and achieve high efficiency and selectivity of the process [21–23]. However, enzymes are characterized by low stability and their activity decreases with an increasing temperature [23]. In addition, re-use of enzymes for catalytic purposes results in a significant reduction in the efficiency of the epoxidation reaction [21].

Temperature also has an important influence on the process of epoxidation of vegetable oils. According to the literature, as the reaction temperature increases, the time needed to achieve high efficiency decreases [24]. However, too high a temperature intensifies side reactions, mainly the opening of oxirane rings [24].

Other factors that affect the epoxidation reaction include the type of oxidizing agent [7], the molar ratio of reactants [8] and the intensity of mixing [25]. Epoxy compounds derived from vegetable and waste oils have many applications. They can be used directly as plasticizers and stabilizers for plastics [26,27]. Epoxy oils are also a raw material for the preparation of many chemical compounds, such as alcohols, glycols, olefinic and carbonyl compounds, epoxy resins, polyesters, or polyurethanes.

This paper reports on the epoxidation of waste oil from a local restaurant with peroxyacetic acid formed in situ from acetic acid and hydrogen peroxide in the presence of ion exchange resin as a heterogenous catalyst. Emphasis was mainly put on determining the process conditions that are consistent with cleaner production. The following four main aspects were taken into account: easy removal of the catalyst, an inexpensive catalyst ensuring adequate product properties, the lowest possible catalyst concentration ensuring adequate conversion in a relatively short epoxidation reaction time, the possibility of reusing the catalyst. The conversion of waste oil into useful chemicals has attracted significant attention in the fields of green and sustainable chemistry and has prompted the implementation of the Circular Economy rules in the polymer technology.

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

The experimentally determined initial iodine number of the used cooking oil was 104 gI2/100 g meaning 0.41 mol of double bonds per 100 g of the used cooking oil. Acidic Ion Exchange Resin (AIER) is an insoluble gel type catalyst in the form of small yellowish organic polymer beads. AIER offers considerable advantages over conventional chemical methods of epoxidation of vegetable oil by improving the selectivity and reducing undesirable side reactions to a certain level [1,5]. In order to determine the most favorable conditions of used cooking oil epoxidation, the effects of different reaction modifications were studied as described further. Firstly, the effect of the content of a heterogenous catalyst on the epoxidation of waste oil was analyzed. The effect of the catalyst concentration on the in situ epoxidation expressed by the changes of the epoxy value over time is presented in Figure 1.

**Figure 1.** Effect of increasing catalyst concentration on epoxy value of epoxidized used cooking oil.

The oxirane rings content increased monotononically with the reaction time for all reactions. As expected, the efficiency of the reaction is greater when the catalyst concentration is higher. It was observed that after 6 h of the reaction the epoxy values of Epox15 and Epox20 are similar and it is not necessary to increase the catalyst concentration up to 20%. The increase in the intensity of the absorption band during the reaction, characteristic of epoxy groups depending on the catalyst concentration, is presented in Figure 2.

**Figure 2.** Dependence of absorption band characteristic of epoxy groups on catalyst concentration. (**a**) 5%, (**b**) 10%, (**c**) 15%, and (**d**) 20%.

For Epox5, the peaks in the spectrum exhibit lower intensities compared to the corresponding Epox15 (Figure 2c) and Epox20 (Figure 2d) peaks. The absorbance band intensities correlate with the epoxy number values.

As can be seen in Table 1, the iodine value of the epoxidized used cooking oil confirms a successful epoxidation of the double bonds of the waste oil. Epox20 was characterized by the highest epoxy value and in consequence the lowest iodine value. For all reactions, conversion, efficiency, and selectivity were determined. The epoxy numbers of Epox15 and Epox20 are comparable. Epox15 was characterized by higher selectivity than Epox20, which is associated with a stronger tendency for side reactions to occur when a higher concentration of the catalyst was used.


**Table 1.** Characteristics of epoxidized UCOs.

Ev—content of epoxy groups; Iv—iodine value; Hv—hydroxyl value; E—efficiency; C—conversion; S—selectivity; Mn—number average molecular weight; Mw—weight average molecular weight; D—dispersity; η—viscosity.

As the concentration of the catalyst used in the reaction increases, the viscosity of the epoxidized oils is increased. S. Zoran et al. have shown that an increase in the degree of conversion to oxirane rings correlates with an increase in the viscosity of an epoxidized oil, which is a result of side reactions that can also increase viscosity [13]. There are no major changes in the hydroxyl or acid value that can be correlated with the change in the concentration of the catalyst used in the reaction.

Dinda et al. analyzed the influence of the AIER concentration (10–25 wt.%) on the epoxidation reaction of cottonseed oil. They concluded that when the catalyst loading was increased from 10 to 15 wt.%, the oxirane conversion increased and it was on the same level as after an addition of a greater amount of the catalyst [18]. In the case of Jatropha oil epoxidation, Goud at al. obtained similar results. The maximum conversion to oxirane in an epoxidation reaction conducted with 16% Amberlite IR-120 loading was lower by only 2.7% than that achieved with 20% [28]. It can be concluded that the process of epoxidation of waste oil is the same as in the case of fresh oils and does not require higher catalyst loading due to the presence of oxidation products in used cooking oil.

The efficiency (E), conversion (C), and selectivity (S) of all reactions (Table 1) were found according to the following equations:

$$\mathbf{E} = \frac{\mathbf{E}\mathbf{v}}{\mathbf{E}\mathbf{v}\_{\text{max}}} \cdot \mathbf{100\%} \tag{1}$$

Ev—the epoxy number of the ester of a vegetable oil after the epoxidation reaction, mol/100 g of epoxidized oil

Evmax—the epoxy number calculated based on the number of unsaturated bonds, mol/100 g of epoxidized oil

$$\mathcal{C} = \frac{I v\_0 - I v}{I v\_0} \cdot 100\% \tag{2}$$

*Iv*0—the iodine number of the methyl ester/vegetable oil before epoxidation, gI2/100 g of oil. *Iv*—the iodine number of the methyl ester/vegetable oil after epoxidation, gI2/100 g of oil.

$$S = \frac{Ev}{Iv\_0 - Iv} \cdot 100\% \tag{3}$$

In the reactions carried out, the maximum conversion rate was 97.5%. Espinoza Perez et al. have obtained a conversion rate of 98.5% for the rapeseed oil epoxidation reaction [29]. Espinoza Perez et al. showed a significant effect of conducting the reaction in a solvent (toluene) environment that allows higher conversion rates than the analogous reaction carried out without a solvent [29].

Oil epoxidation may affect its average molar mass. In the environment of epoxidation reactions, partial triglyceride breakdown may occur, resulting in lower molecular weight products such as monoand diglycerides, free fatty acids, and glycerin. Newly formed epoxide rings may also be opened, and derivatives of higher molecular weight may be formed, which in turn increases the molecular weight. Both trends result in an increase in the dispersion of the resulting reaction product [5].

Based on our research, we concluded that the content of the heterogeneous catalyst did not have a significant effect on the average molar masses of epoxidized waste oils and the compounds obtained were monodisperse. The low dispersion value of the epoxidized oils obtained indicates a small amount of by-products resulting from reactions such as oligomers.

In order to find the effect of particle size and degree of cross-linking AIER on the properties of the epoxidized oil, the experiments were conducted using three different sizes of resin particles as well as resins containing three different amounts of the cross-linking agent. Nine different Dowex resins were used in the study and the results were compared with commonly used Amberlite IR-120. The characteristics of the heterogeneous catalysts are presented in the Table 2 and photographs of the catalysts are shown in Figure 3.

The progress of the reaction was monitored by determining the epoxy number and by a FTIR analysis during the reaction. The epoxy number changes during the reaction and the changes of characteristic bands in the FTIR spectrum are presented in Figures 4 and 5.


**Table 2.** Characteristics of acidic ion exchange resins (AIER).

**Figure 3.** SEM micrographs of Dowex ion exchange resins.

 Epox\_2\_50-100 Epox\_2\_100-200 Epox\_2\_200-400

0.15

0.20

0.25

0.30

0.35

0.15

0.20

Epoxy value, mol/100g

0.25

0.30

0.35

 Epox\_4\_50-100 Epox\_4\_100-200 Epox\_4\_200-400

**Figure 4.** Influence of different grain size of (**a**) Dowex50 × 2, (**b**) Dowex50 × 4, and (**c**) Dowex50 × 8. −

−

Absorbance **Figure 5.** FTIR spectra of samples taken during epoxidation process of Epox\_8\_200-400.

**6 h 4 h 2 h 0 h** During all processes, the Ev increased as the reaction progressed, which indicates double bond oxidation in fatty acid chains to form oxirane rings (Figure 4). Slightly lower initial epoxy values were noted for the reaction in the presence of a catalyst containing 2% of the cross-linker. However, no significant difference was observed in the final Evs depending on the type of the catalyst used in the synthesis.

4000 3800 3600 3400 3200 3000 2800 2600 2400 2200 2000 1800 1600 1400 1200 1000 800 600 Wavenumber, cm-1 The analysis of FTIR spectra collected during the reaction allowed on-line detection of characteristic groups. In Figure 5, the FTIR spectra of the used cooking oil epoxidation with Dowex50X8 200-400 are shown. In the FTIR spectrum of the waste rapeseed oil, with wave numbers of about 3010 cm−<sup>1</sup> and about 1650 cm−<sup>1</sup> , the bands corresponding to the vibrations of the double bonds between carbon atoms are visible. These signals lose their intensity as the epoxidation reaction progresses. Along with the decrease in the intensity of the bands corresponding to the vibrations of unsaturated bonds, the FTIR spectra of the epoxidized oil samples show signals in the range 800–850 cm−<sup>1</sup> , which are characteristic for epoxy groups.

In all spectra, apart from the signals of epoxy groups, there are also bands at about 2925 and 2850 cm−<sup>1</sup> , characterized by high intensity. They arise as a result of stretching vibrations of bonds between carbon and hydrogen atoms in the groups –CH2– and –CH3, occurring in fatty acid chains. Another characteristic band in the spectra of the epoxidized oils is the signal visible at approx. 1740 cm−<sup>1</sup> , which comes from the stretching vibrations C=O of the ester group.

Ev, Iv, and Hv determinations were carried out for the epoxidized oils obtained (Table 3).


**Table 3.** Characteristics of epoxidized oil obtained using different AIER.

Based on our research it was found that in the case of Dowex resins containing 2% cross-linking agent the highest epoxy value was obtained for the resin with the largest grain size. In the case of resins containing 8% divinylbenzene, the opposite effect was observed. However, the differences in the epoxy numbers are insignificant. Therefore, the results obtained were compared with the epoxy oil (Epox\_15) obtained through a reaction with Amberlite resin. This resin has a much lower price compared to Dowex resins. From the point of view of waste oil recycling, it is important to develop a low-cost process. The epoxy number of Epox\_15 is slightly lower than the epoxy numbers obtained for the oils synthesized in the presence of Dowex resins containing 8% divinylbenzene, and comparable to resins with 4% cross-linking agent. Dinda et al. conducted epoxidation of cottonseed oil using two very different sizes of resin particles under otherwise similar conditions. The particle sizes were greater than 620 µm and smaller than 120 µm. They concluded that both particle sizes gave nearly the same oxirane conversion [18]. Goud et al. also analyzed the intraparticle diffusional limitations by using two widely differing particle sizes of resin, namely >599 and <64 µm under otherwise the same reaction conditions. They concluded that both particle sizes gave practically the same results [30].

During all epoxidation processes, double bond conversion in the range of 89–94% was achieved. The conversion value for the reaction with Amberlite was 94%. Comparing to the literature data and cases where fresh oils were epoxidized, the values obtained in this study are very favorable. According to literature reports, during oil epoxidation using acetic peroxyacid and in the presence of Amberite IR-120 ion exchange resin, it is possible to achieve about 90% efficiency after 7 h of the process at 65 ◦C [15]. The oxidation reaction of unsaturated soybean oil bonds, using the same oxidizing agent and catalyst for 10 h and at 60 ◦C allows achieving about 86.8% double bond conversion and about 86% selectivity [31]. On the other hand, the epoxidation of karanja oil with the use of acetic peracid and Amberlite IR-120 resin leads to approx. 85% yield.

Waste rapeseed oil and its derivatives, in the form of epoxidized oils, were also characterized by a similar molar mass distribution (Figure 6).

**Figure 6.** Chromatograms of bio-polyols obtained with Dowex catalysts.

The highest intensity signal present in all chromatograms corresponds to the weight of triglyceride. Its retention time is about 26 min. Peaks characterized by low intensity, with a retention time of about 24 min, indicate the presence of oligomers in the epoxidized oils. The signal is also visible in the chromatogram of waste rapeseed oil, hence the oligomerization products may have arisen as a result of a processes related to the thermal treatment of vegetable oil. The waste oil and epoxidized oils were characterized by similar average molar masses.

The dispersion of the products was slightly higher than the dispersion of the starting oil and ranged from 1.03 to 1.07. A small dispersion of the masses of the synthesized epoxidized oils is their advantage. In the production on an industrial scale, striving to obtain materials with low dispersion is sought as it has a positive effect on their performance and the possibility of further processing.

The advantage of heterogeneous catalysts should be easiness of separating them from products. Our experiments showed that in the case of the processes catalyzed by Dowex ion exchange resins, due to their small size, it was not possible to remove the catalyst from the epoxidized oil thoroughly using methods available on a laboratory scale. This resulted in turbid products. For the Amberlite IR-120 resin, this problem did not occur. The epoxidation process in the presence of Dowex ion exchange resins is not economical resulting from high prices of catalysts. Catalysts with smaller grain diameters and lower cross-linker content are generally more expensive. In contrast, the use of Amberlite resin allows obtaining epoxidized oils that can be used as plasticizers as well as intermediates for subsequent syntheses whereas the costs of waste oil modification are reduced.

The life of AIER is limited due to deactivation after a few cycles and the disposal of used catalyst is a major problem faced by industry [32]. In order to create a more environmentally friendly process, the possibility of repeated use of the same catalyst has been analyzed in the literature. Such an experiment was described for the epoxidation of fresh vegetable oils such as cotton seed oil [18], castor oil [33]. In the case of used cooking oil, there are oxidation products and other impurities confirmed by high peroxide and anisidine values, as well as the presence of polycyclic aromatic hydrocarbons and polychlorinated biphenyls [34]. In order to analyze such impurities and their

influence on the efficiency of AIER, in this part of our work the same catalyst was used ten times for the epoxidation reactions. The microphotographs of Amberlite after each reaction are shown in Figure 7.

**Figure 7.** SEM microphotographs of Amberlite IR-120 catalyst before epoxidation reaction (0×) and after each reaction (1×–10×).

Regardless of the reaction, contaminants are present on the resin surface. This effect is associated with the method of resin preparation for synthesis. In order to minimize the costs and make the process simpler, we only separated the catalyst by filtration from the epoxidized oil and washed with water in order to remove acetic acid.

It was observed that after the first recycle the epoxy value of sample 2× was 6% lower (Figure 8). However, interesting results not described in the literature before were obtained in the case of the 3× and 4× reactions where the same catalyst was used for the 3rd and 4th time. The epoxide number of the modified oils was 8 and 6% higher, respectively, compared to the 1× reaction. This effect can be associated with the swelling of the catalyst, which results in easier access to the acid sulfonic groups of the resin.

**Figure 8.** Changes in Ev and Iv depending on the number of times the catalyst was used.

In the literature there are descriptions of solid catalyst recycling. Dinda et al. used the same Amberlite IR-120 in four consecutive experiments and concluded that after the first recycle the catalyst activity decreased gradually with the number of recycles [18]. Mungroo et al. [17] found that AIER can be reusable and exhibited a negligible activity loss. After the catalyst had been used four consecutive times, the relative conversion to oxirane and the iodine conversion were 83 and 85%, as compared to 90 and 88.4%, respectively, obtained with the fresh catalyst. Goud et al. [16] based on their studies on the epoxidation of karanja oil found that it is possible to repeat recirculation of the catalyst Amberlite IR-120 four times. After each synthesis, the catalyst was regenerated. The regeneration relied on the filtration of resin, rinsing with water and diethyl ether, and drying in ambient temperature.

#### **3. Materials and Methods**

#### *3.1. Materials*

Glacial acetic acid (min. 99.5–99.9 wt.%), hydrogen peroxide (30 wt.%), were purchased from Avantor Performance Materials Poland S.A (Gliwice, Poland). Ion-exchange resins—Amberlite® IR-120 (Sigma-Aldrich, St. Louis, MO, USA), Dowex 50WX2 50-100, Dowex 50WX2 100-200, Dowex 50WX2 200-400, Dowex 50WX4 50-100, Dowex 50WX4 100-200, Dowex 50WX4 200-400, Dowex 50WX8 50-100, Dowex 50WX8 100-200, Dowex 50WX8 200-400 were purchased from Sigma-Aldrich (St. Louis, MO, USA). Used cooking oil was collected from 3 local restaurants (Kraków, Poland). The iodine and acid values of the used cooking oil mixtures were 104 gI2/g.

#### *3.2. Epoxidation Procedure*

In the experiment, 250 g of used cooking oil and 0.5 mol of acetic acid and 2 mol hydrogen peroxide per mole of unsaturated bonds of oil as well ion exchange resin were added to a reactor. The content of Amberlite 120 was 15 wt.% (with respect to oil mass). The reactions were carried out for 6 h at a temperature of 60–65 ◦C using continuous stirring. Samples were taken out at the 1st, 2nd, 3rd, 4th, 5th and 6th h. After six hours the reaction mixture was separated into two phases, organic and aqueous. The organic phase was washed successively with warm water until it was acid free. The organic phase was distilled under vacuum (10 mbar) for 2 h in order to remove of water [5].

#### *3.3. Methodology of Epoxidized Oil Characterization*

The designation of Iv was done using the Hanus method according to the standard PN-87/C-04281, in which the iodine atoms are added to unsaturated bonds. The unsaturation degree of a given fat is then expressed by the amount of the iodine added.

Ev was determined according to the PN-87/C-89085/13 standard. The method involves a quantitative reaction of hydrogen chloride with a reactive epoxy group in dioxane at room temperature and titration of the hydrogen chloride excess using a solution of sodium hydroxide in methanol in the presence of cresol red as an indicator.

Hv of the polyol was determined according to the standard PN-93/C-89052/03, in which the hydroxy groups of a polyol undergo acetylation using acetic anhydride. The excess of the acetic anhydride is decomposed by a water addition (formation of acetic acid) and followed by titration using a solution of potassium hydroxide in the presence of an indicator.

Viscosity (η) was determined using a rotational rheometer HAAKE MARS III (Thermo Scientific, Waltham, MA, USA) at 25 ◦C. The control rate mode was used in the plate-plate arrangement with the plates having a diameter of 20 mm and rotation speeds of 100 cycles/min.

Number average molecular weight (Mn) and dispersity (Ð) were determined by a gel permeation chromatography (GPC) analysis. GPC measurements were performed using a Knauer chromatograph (Warsaw, Poland). The calibration was performed using polystyrene standards. Tetrahydrofuran (Avantor Performance Materials Poland S.A, Gliwice, Poland) was used as an eluent at a 0.8 mL/min flow rate at room temperature.

FTIR spectroscopy was performed using a FT-IR SPECTRUM 65 spectrometer (Perkin Elmer, Waltham, MA, USA).

#### **4. Conclusions**

The results presented in this paper allow verification of the influence of the heterogeneous catalyst concentration, its structure and particle size, as well as repeated use on the waste oil epoxidation process. No significant differences were observed in the final properties of the epoxidized waste cooking oil depending on the particle size and the cross-linking degree of the ion exchange resins used in the synthesis. The results were compared with the reaction in which Amberlite 120 resin was used, characterized by a larger grain size and a lower price. It was found that the Amberlite 120 resin can be successfully used in used cooking oil epoxidation reactions. Aiming at waste reduction in technological processes, an attempt was made to reuse a heterogeneous catalyst. The catalyst Amberlite IR-120 was reused without further treatment and no significant differences in the epoxy value were observed. Application of the same resin eight times in epoxidation allows obtaining products with epoxy values of 0.3 mol/100 g and higher. It was noticed that in the case of the reactions where the catalyst was reused for the third and fourth time the content of oxirane rings was higher by 8 and 6% compared to reaction where the catalyst was used one time only. Such an observation has not been reported so far.

**Author Contributions:** Conceptualization, M.K.; methodology, M.K.; validation, M.K.; formal analysis, M.K.; investigation, M.K. and M.N.; resources, M.K.; data curation, M.K. and M.N.; writing—original draft preparation, M.K.; writing—review and editing, M.K.; visualization, M.K.; supervision, M.K.; project administration, M.K.; funding acquisition, M.K. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by National Center for Research and Development in Poland under the Lider Program, grant number LIDER/28/0167/L-8/16/NCBR/2017.

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

#### **References**


**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2020 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 (http://creativecommons.org/licenses/by/4.0/).
