**On the Role of the Cathode for the Electro-Oxidation of Perfluorooctanoic Acid**

**Alicia L. Garcia-Costa 1,2,\*, Andre Savall 2, Juan A. Zazo 1, Jose A. Casas <sup>1</sup> and Karine Groenen Serrano 2,\***


**\*** Correspondence: alicial.garcia@uam.es (A.L.G.C.); serrano@chimie.ups-tlse.fr (K.G.S.)

Received: 22 July 2020; Accepted: 6 August 2020; Published: 8 August 2020

**Abstract:** Perfluorooctanoic acid (PFOA), C7F15COOH, has been widely employed over the past fifty years, causing an environmental problem because of its dispersion and low biodegradability. Furthermore, the high stability of this molecule, conferred by the high strength of the C-F bond makes it very difficult to remove. In this work, electrochemical techniques are applied for PFOA degradation in order to study the influence of the cathode on defluorination. For this purpose, boron-doped diamond (BDD), Pt, Zr, and stainless steel have been tested as cathodes working with BDD anode at low electrolyte concentration (3.5 mM) to degrade PFOA at 100 mg/L. Among these cathodic materials, Pt improves the defluorination reaction. The electro-degradation of a PFOA molecule starts by a direct exchange of one electron at the anode and then follows a complex mechanism involving reaction with hydroxyl radicals and adsorbed hydrogen on the cathode. It is assumed that Pt acts as an electrocatalyst, enhancing PFOA defluorination by the reduction reaction of perfluorinated carbonyl intermediates on the cathode. The defluorinated intermediates are then more easily oxidized by HO• radicals. Hence, high mineralization (xTOC: 76.1%) and defluorination degrees (xF −: 58.6%) were reached with Pt working at current density *j* = 7.9 mA/cm2. This BDD-Pt system reaches a higher efficiency in terms of defluorination for a given electrical charge than previous works reported in literature. Influence of the electrolyte composition and initial pH are also explored.

**Keywords:** perfluorooctanoic acid; emerging contaminant; defluorination; platinum; electro-oxidation

#### **1. Introduction**

Perfluoroalkyl substances (PFAS), such as perfluorooctanoic acid (PFOA, C7F15COOH) are widely used in the chemical industry because of their amphiphilicity, stability, and surfactant property. They are employed in the synthesis of fluoropolymers and fluoroelastomers, as surfactants in fire-fighting foams, and in textile and paper industries to produce water and oil repellent surfaces [1]. Nevertheless, despite their practical interest, these substances present a high toxicity due to their potential bioaccumulation, and common occurrence in water resources. PFOA has been recognized as an emerging environmental pollutant and has been included in the European Candidate List of Substances of Very High Concern ("SVHC") [2]. Hence, the current challenge is to develop highly efficient and cost-effective processes for the elimination of perfluoroalkyl substances at source.

The main issue in PFOA degradation is to break the C-F bond, one of the strongest bonds known (≈460 kJ/mol) [3]. This confers a high stability and resistance to PFAS which cannot be degraded by direct hydrolysis, photolysis, or through conventional biological treatments [4]. As a result, PFAS have been detected in natural water streams [5], sediments [6], and even in tap and bottled water in

concentrations up to 640 ng/L [7]. So far, adsorption onto carbonaceous materials [8], alumina [9], or other sorbents [10] have been successfully applied for PFAS removal. Nonetheless, this technology implies the transfer of the pollutant to another phase, the sorbent, which becomes a new residue after use. To overcome this drawback, advanced oxidation processes (AOP) are being explored for PFAS removal. AOP are based on the use of strong oxidizing radicals to degrade, most commonly, organic pollutants in aqueous phase [11]. The most extended AOP are those based on the use of hydroxyl radicals (HO•) to attack organic pollutants by hydrogen abstraction [12]. Consequently, the substitution of all organic hydrogen for fluorine in PFOA makes these compounds inert to this kind of AOP. The non-reactivity of PFOA to HO• attack has been confirmed by various studies [13–15]. As a matter of fact, Maruthamuthu et al. have shown that the reactivity of hydroxyl radicals on acetate decreases considerably with increasing halogen substitution [13]. Using the Fenton process, known to generate hydroxyl radicals by the action of Fe (II) on hydrogen peroxide, no degradation was observed when the Fenton reagent (0.2 mM, Fe2<sup>+</sup>: H2O2, molar ratio = 1:1) was mixed with PFOA (0.02 mM) at room temperature [14]. Similar results were obtained by Santos et al. with only 10% PFOA removal and without any C-F bond cleavage [16].

More recently, photocatalytic treatments have been applied for PFOA degradation. This technology achieved high PFOA removal (xPFOA > 90%) when using modified TiO2 photocatalysts such as Cu-TiO2 [17], Pb-TiO2 [18], and rGO-TiO2 [19].

Besides PFOA removal, defluorination (xF2212−) is a very important parameter to evaluate the process efficiency. xF <sup>−</sup> defined as the ratio of the fluoride concentration (CF <sup>−</sup>, measured) released by PFOA degradation with respect to the initial content of fluoride in the initial amount of PFOA molecule (CF,PFOA 0) is expressed in percentage as shown in Equation (1).

$$\chi\_{\text{F}^{-}} = \frac{\mathbf{C}\_{\text{F}^{-}, \text{measured}}}{\mathbf{C}\_{\text{F}, \text{ PFOA}\_{\text{A}}}} \cdot 100 \tag{1}$$

In photochemical oxidation of PFOA, defluorination is usually low (xF − < 25%), with the average xF <sup>−</sup>/xPFOA ratio around 0.26 [20].

Another technique for PFAS remediation is electrochemical degradation. PFOA electrooxidation has been successfully carried out in different systems using boron-doped diamond (BDD) as anode (Table 1). Under the studied conditions, PFOA removal ranged from 60% to 100%. It should be noted that defluorination values were very different, suggesting that either the operating conditions (electrolyte, pH, etc.,) or the cathode reduction reactions may play a key role in the PFOA degradation mechanism. The cleavage of the C-F bonds to form F− ions is interesting because F− ions readily combine with Ca2<sup>+</sup> to form environmentally harmless CaF2, as reported by Hori et al. [3].

xF <sup>−</sup> and xPFOA ratios obtained by electrochemical treatment (up to 80–85%, Table 1) are higher than those reported in photo-oxidation (<25%). Nonetheless, all previous electrooxidation studies were conducted employing a high supporting electrolyte concentration, which makes difficult to dispose the treated wastewater after reaction. Therefore, this work aims to gain knowledge on the role of the cathode as electrocatalyst in PFOA electrooxidation working at low electrolyte concentration (3.5 mM). For this purpose, BDD was chosen as the anode and BDD, Pt, Zr, and stainless steel were tested as cathodes in the degradation of 100 mg/L PFOA.


**Table 1.** Perfluorooctanoic acid (PFOA) electrooxidation with boron-doped diamond (BDD) anode.

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

In order to test the influence of the cathode material on the degradation process, a BDD anode was successively coupled with cathodes made of BDD, Pt, Zr, and stainless steel. Results of electrolysis runs conducted at 7.9 mA/cm2, at 25 ◦C, for the treatment of 100 mg/L PFOA solutions (namely, 0.242 mol/m3), are presented in Figure 1a. For each couple of electrodes, the curves show that PFOA concentration followed, from CPFOA,0 = 100 mg/L to CPFOA,t ≈ 25 mg/L, a similar decrease, characteristic of a pseudo-first order kinetics. For experiments presented in Figure 1a the applied current density was higher than the limiting current density. Considering a pure mass transport controlled reaction for the first exchange of charge between a molecule of PFOA and the anode surface, the limiting current density calculated using the equation established from the Nernst diffusion model: *j*lim = *n*·*F*·*km*·CPFOA,0 equals to 0.63 A/m<sup>2</sup> for *n* = 1, *F* = 96,485 C/mol, *km* = 2.7·10−<sup>5</sup> m/s (determined experimentally using the ferri/ferro system, as described elsewhere [26]) for a flow rate of 0.360 m3/h [27], CPFOA,0 = 0.242 mol/m3. This value of the limiting current density is more than 100 times lower than that applied during electrolysis (*j* = 79 A/m2). Under these conditions, the decay of concentration from CPFOA,0 to CPFOA,t depends upon the mass transfer coefficient *km*, the surface area (A) of the electrode, and the volume (V) of electrolyte [28], as follows:

$$\mathbf{C}\_{\text{FFOA},t} = \mathbf{C}\_{\text{FFOA},0} \mathbf{e}^{\left(\frac{-t}{\pi}\right)} \tag{2}$$

where the constant of time, τ, is defined by: τ = V/(*km* A). For its calculation we considered the following values V <sup>=</sup> <sup>10</sup>−<sup>3</sup> <sup>m</sup>3, A <sup>=</sup> <sup>63</sup>·10−<sup>4</sup> <sup>m</sup>2, and *km* <sup>=</sup> 2.7·10−<sup>5</sup> <sup>m</sup>/s, obtaining a time constant (τ) equal to 5800 s. According to Equation (2), the PFOA theoretical concentration at t = 2 h is around 29 mg/L, which is in agreement with the experimental results (≈25 mg/L), as shown in Figure 1a.

**Figure 1.** Influence of the cathode material in (**a**) PFOA removal (symbols: experimental data, lines: kinetic fitting), (**b**) TOC depletion, (**c**) released fluorine (symbols: experimental data, lines: kinetic fitting). Operating conditions: [PFOA]0: 100 mg/L, j: 7.9 mA/cm2, electrolyte: 3.5 mM Na2SO4, T: 25 ◦C, pH0: 4.

The excess of charge on the first oxidation step of PFOA is used for other electron transfers and by the action of HO• radicals during the degradation of the numerous intermediates. PFOA removal after 6 h was 100%, 98.1%, 97.9%, and 97.6% for Pt, steel, Zr, and BDD, respectively. Values of the rate constants and regression coefficients are collected in Table 2. It should be noted that the Pt cathode has the best results with a PFOA degradation rate 39% faster than the other tested materials, which exhibit a similar behavior between them. This enhancement is also reflected in the mineralization degree (Figure 1b), with a 76.1% Total Organic Carbon (TOC) removal with the BDD-Pt system.


**Table 2.** PFOA degradation and fluoride release kinetics.

As previously explained, one of the main challenges in PFOA oxidation is the effective breakdown of the C-F bond. PFOA defluorination was followed along the reaction by means of ionic chromatography, as depicted in Figure 1c. The trend for defluorination was Pt > BDD > Zr > steel. In this case, the cathode also played an important role, reaching a 58.6% in the case of Pt, against 42–49% for BDD, Zr, and steel. Moreover, fluoride release follows a first order, as reflected in Figure 1 and Table 2, where the function of Pt as electrocatalyst is confirmed.

Pt is a common catalyst in hydrodehalogenation reaction of organic molecules, because of its capacity to adsorb hydrogen, providing a catalytic site were the dehalogenation takes place [29]. H2 generation by water electrolysis on the cathode's surface may be responsible for PFOA hydrodefluorination, following the reaction mechanism shown in Figure 2.

**Figure 2.** PFOA electrooxidation mechanism.

Because PFOA is inert to hydroxyl radicals, its degradation is initiated on the anode by a direct electron transfer reaction to form a perfluoro radical C7F15COO• (Equation (3)). This radical loses its carboxylic group (Equation (4)) and reacts with HO• leading to the generation of C7F13−CF2OH (Equation (5)), as previously described by Zhang et al. [30].

$$\text{BDD} + \text{C\gamma} \text{F}\_{15} \text{COO}^{-} \rightarrow \text{BDD} + \text{C\gamma} \text{F}\_{15} \text{COO}^{\bullet} + \text{e}^{-} \tag{3}$$

$$\rm C\_7F\_{15}COO^\bullet \rightarrow \rm C\_7F\_{15}\uparrow + CO\_2 \tag{4}$$

$$\rm C\_7F\_{15}\rm \rm ^\bullet + HO^\bullet \to \rm C\_7F\_{15}OH \tag{5}$$

This alcohol then reacts according to three pathways, (for clarity reasons, only the first one (i) is illustrated in Figure 2):

(i) With adsorbed hydrogen generated by water electro-reduction at the cathode, releasing 2 F− (Equation (6)).

$$\rm C\_6F\_{13}CF\_2OH + 4H\_{ads} \rightarrow \rm C\_6F\_{13}CH\_2OH + 2HF \tag{6}$$

As the first carbon in the alkyl chain is now defluorinated, HO• can attack it once again leading to the formation of C6F13COOH. This mechanism is similar to that presented for PFOA photocatalytic degradation byWang et al. [20] and theoretic quantum calculations and experimental data collected by Trojanowicz et al. [31]. Hence, this step depends strongly on the cathode material.

(ii) With hydroxyl radicals leading to the formation of COF2, as related by Niu et al. [31] and Zhang et al. [28], following Equations (7)–(9):

$$\rm C\_7F\_{15}OH + HO^\bullet \rightarrow C\_7F\_{15}O^\bullet + H\_2O \tag{7}$$

$$\text{C}\_7\text{F}\_{15}\text{O}^\bullet \rightarrow \text{C}\_6\text{F}\_{13}\text{"}+\text{COF}\_2\tag{8}$$

$$\text{COF}\_2 + \text{H}\_2\text{O} \rightarrow \text{CO}\_2 + 2\text{HF} \tag{9}$$

According to George et al. [30] hydrolysis of carbonyl fluoride COF2 in the aqueous phase is extremely fast since its half-life is 0.7 s at T = 273 K.

(iii) Giving the perfluorocarbonyl fluoride (Equation (10)) for which hydrolysis leads the formation of perfluorocarboxylic acid and HF (Equation (11)) [30,32].

$$\rm C\_7F\_{15}OH \rightarrow \rm C\_6F\_{13}COF + HF \tag{10}$$

$$\rm C\_6F\_{13}COF + H\_2O \rightarrow C\_6F\_{13}COO^- + HF + H^+ \tag{11}$$

Considering this complex reaction mechanism, it should be noted that TOC decay was faster within the first hour of reaction, then it slowed down (Figure 1b). This is related to the generation of short-chain fluorinated acids (decarboxylation step), which are less active to electro-oxidation processes. In fact, pH value in the Pt system decreased from 4 to 3.2 in 120 min, maintaining this pH until the end of the reaction, which evidences the generation of these acidic species.

Data displayed in Figure 1 allow to determine the fluoride concentrations produced with respect to the degraded carbon in the form of CO2 (F−/CO2) or with respect to the PFOA eliminated over time (F−/PFOA). Figure 3 shows that the PFOA defluorination leads to the formation of 1 to 1.5 fluorine ions per removed atom of carbon in the first three hours of electrolysis. According to the proposed mechanism, this value close to the one at the beginning of the electrolysis is related to decarboxylation which leads to the formation of Rf-COF. The kinetics of this step are probably faster than that of the defluorination stages according to Equations (6)–(9). Besides, part of the process can be attributable to the electrocatalytic hydrogenation of the perfluorocarbonyl fluoride C6F13COF that forms simultaneously the hydrofluoric acid and the 1,1-dihydroperfluoroalkyl alcohol C6F13CH2OH. This alcohol is stable but easily oxidizable on the BDD anode [33] (cf. Figure 2). This process is slowed down by the diffusion of the species to the cathode. Not all molecules undergo the loss of two fluoride atoms, which would explain the value of 1.5 instead of the usual ratio 1.9 present in the initial PFOA molecule.

In addition, Figure 3 shows the variation of the ratio between the concentration of fluoride ions released and the concentration of the removed PFOA. This ratio varies from 7.7 to 9 for 360 min of electrolysis. These values highlight the high, yet incomplete PFOA defluorination. Finally, the ratio between the carbon loss (in the form of CO2) and the removed PFOA (CO2/PFOA) is in the order of 6–7, slightly less than 8, i.e., the theoretical value for C7F15COOH, confirming the formation of reaction intermediates. This ratio decreases during electrolysis: the degradation being faster at the beginning of the reaction, until t:150 min.

At this time more than 85% of PFOA has been eliminated. PFOA depletion slows down both the defluorination and carbon skeleton breaking. In addition, shorter molecular chains could display slower kinetics.

**Figure 3.** Variation of fluorine ions (full symbols) and carbon removal (empty symbols) during electrolysis with respect to carbon removal and degraded PFOA. Operating conditions: [PFOA]0: 100 mg/L, j: 7.9 mA/cm2, electrolyte: 3.5 mM Na2SO4, T: 25 ◦C, pH0: 4.

Figure 4 shows the ratio of fluorine and carbon atoms contained in the chemical intermediates. The molar concentration of F and C atoms contained in the intermediates are defined, respectively, as follows:

$$\mathcal{L}\_{\text{F,intermediate}} = 15 \cdot (\mathcal{C}\_{\text{FPCA},0} - \mathcal{C}\_{\text{FPCA},t}) - \mathcal{C}\_{\text{F}^{-}t} \tag{12}$$

$$\mathbf{C\_{C,inermodies}} = \mathbf{TOC\_t} - \mathbf{TOC\_{FPOA,t}} \tag{13}$$

where CPFOA,0 and CPFOA,t refer to the molar concentration of PFOA at initial time and at time t, respectively; CF−,t is the molar concentration of fluorine ions at t; TOCt and TOCPFOA,t are the total carbon molar concentration and the carbon molar concentration in the PFOA, respectively.

From Figure 1, after 360 min of electrolysis, the defluorination rate is 59% whereas more than 98% of PFOA and 76% of TOC have been eliminated. Figure 4 highlights that in this moment, the intermediates still contain 24% of carbon and 41% of fluorine. Xiao et al. reached a 90% defluorination and mineralization working at high temperature (T: 80–120 ◦C), meaning they managed to degrade the short-chain acids [25]. This is in agreement with the results for degradation of phenol in heterogeneous Fenton at high temperature, where maleic, malonic, oxalic, and formic acids can be completely degraded [11], in contrast with room temperature processes [34]. Aiming to verify Xiao et al.'s results, an electrooxidation run at 80 ◦C was performed using Pt cathode. After 30 min reaction there was an overvoltage on the cell due to the damage on the cathode, probably because of the HF attack (Figure S1 of the Supplementary Material). Hence, high temperature electrooxidation could not be performed in our system and further runs were conducted at 25 ◦C.

**Figure 4.** Ratio of fluorine and carbon atoms in the chemical intermediates. Operating conditions: [PFOA]0: 100 mg/L, j: 7.9 mA/cm2, electrolyte: 3.5 mM Na2SO4, T: 25 ◦C, pH0: 4.

After selecting Pt as the best cathode, within the tested materials, different salts were used as electrolyte, NaClO4, KNO3, Na2SO4, Na2S2O8, at 3.5 mM. Results for these experiments can be found in Figure 5. As previously reported by Schaefer et al. [21], the influence of the electrolyte type on PFOA degradation is very low. Still, significant differences were found for TOC removal, where the removal efficiency followed this trend Na2SO4 (76.1%) > Na2S2O8 (72.6%) > KNO3 (70.5%) > NaClO4 (67.1%). Sulfate achieved both a slightly higher mineralization degree and defluorination. This can be explained by the fact that sulfate anions behave as an active electrolyte via the electrochemical generation of the strong oxidizing sulfate radicals (SO4 •−) on a BDD anode [35,36]. Indeed, the oxidation of water at the anode greatly decreases locally the pH at the surface leading to the formation of HSO4 <sup>−</sup> from SO4 2−. Then HSO4 − reacts with HO• radicals to form sulfate radicals [33,37].

$$\mathrm{HSO}\_{4}^{-}+\mathrm{HO}^{\bullet}\rightarrow\mathrm{SO}\_{4}^{\bullet-}+\mathrm{H}\_{2}\mathrm{O}\quad\mathrm{k}=6.9\cdot10^{5}\mathrm{M}^{-1}\mathrm{s}^{-1}\tag{14}$$

SO4 •− radical participates in electron transfer reactions and promotes the decarboxylation of carboxylic acids, contrary to HO• which rather acts in hydrogen abstraction or addition [38]. In addition, sulfate radicals are more stable than hydroxyl radicals (their half-life is 30–40 μs and 10−<sup>3</sup> μs, respectively).

Considering PFOA degradation with sulfate radicals, the literature review by Yang et al. highlights that the decomposition and defluorination efficiencies increase with a decrease in PFOA chain-length [39]. Besides the major role of hydroxyl radicals on PFOA oxidation, the presence of sulfate radicals helps to improve the degradation of the generated intermediates. Qian et al. [40] estimated the constant rate of PFOA degradation with sulfate radicals at 2.59·10<sup>5</sup> <sup>M</sup><sup>−</sup>1s−1. This is consistent with the higher TOC removal observed in our experiments in presence of sulfate. Furthermore, sulfate is a more environmentally friendly electrolyte, in comparison to perchlorate and nitrate, which can be considered pollutants by themselves. Thus, the rest of experiments were carried out using Na2SO4 3.5 mM.

Influence of initial pH (pH0) on PFOA degradation was also evaluated working at the natural pH of PFOA solution (pH: 4) and at pH values of 7 and 9. Results for these experiments are shown in Figure 6. As it may be seen in Figure 6d, reaction media is quickly acidified. This is related to both the generation of short chain acids and the reaction between sulfate radicals and water to produce hydroxyl radicals, which also generates protons, as depicted in Equation (15). PFOA decay (Figure 6a) was similar for all the runs. However, pH0 had a great influence on the initial rate for TOC abatement, related to the higher oxidation potential of sulfate radicals in alkaline media [41]. Despite achieving a higher mineralization degree at pH0: 9, the highest defluorination was reached when starting in acidic media.

$$\text{SO}\_4^{\bullet-} + \text{H}\_2\text{O} \rightarrow \text{H}^+ + \text{HO}^\bullet + \text{SO}\_4^{2-} \tag{15}$$

**Figure 5.** Influence of the electrolyte in PFOA (**a**) degradation, (**b**) mineralization, and (**c**) defluorination in electrooxidation using BDD/Pt electrodes. Operating conditions: [PFOA]0: 100 mg/L, j: 7.9 mA/cm2, electrolyte: 3.5 mM, T: 25 ◦C, pH0: 4.

So far, the role of the cathode as electrocatalyst in the degradation and defluorination of PFOA has been proved. Also, the influence of several operating conditions has been tested, demonstrating an overall great decontamination working at low electrolyte concentration at mild temperature. Nonetheless, in order to compare the obtained results with those reported in literature, we have compared the defluorination degree against the energetic requirements, measured as the applied charge, as shown in Figure 7. As may be seen, both Shaefer et al. [21] and Urtiaga et al. [22] boosted the defluorination degree when increasing the applied charge. However, the results presented in this work using Pt cathode at 7.9 mA/cm2, 3.5 mM Na2SO4 at pH0:4 and T:25 ◦C are the most competitive in terms of PFOA defluorination against electric charge. In this sense, cathode selection becomes a key point for both increasing the activity and reducing the energy requirements in PFOA electrooxidation.

**Figure 6.** Influence of the pH0 in PFOA (**a**) degradation, (**b**) mineralization, and (**c**) defluorination and (**d**) pH evolution in electrooxidation using BDD/Pt electrodes. Operating conditions: [PFOA]0: 100 mg/L, j: 7.9 mA/cm2, Na2SO4: 3.5 mM, T: 25 ◦C.

**Figure 7.** Process comparison in terms of defluorination against electric charge for PFOA electrooxidation with BDD anodes. Cathodes: Schaefer et al.—W [21], Urtiaga et al.—W [22], Zhuo et al.—BDD [23], Ochiai et al.—Pt [24], this work—Pt.

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

#### *3.1. Reactants*

Perfluoroctanoic acid (95 wt.%), Na2SO4, KNO3, NaClO4, Na2S2O8, acetonitrile (ACN), H2SO4, and NaOH were supplied by Sigma-Aldrich (Darmstadt, Germany). All reagents are of analytical grade and they were used as received without further purification. Working standard solutions of PFOA and fluoride (NaF from Sigma-Aldrich, Darmstadt, Germany) were prepared for calibration.

#### *3.2. Experimental Set-Up*

The electrochemical oxidation system consists in a 1-L thermoregulated glass reservoir connected to the cell through a centrifugal pump. PFOA solution was recycled in the system at 360 L/h flow rate and the temperature was set at 25 ± 1 ◦C. The electrochemical cell is a one-compartment flow filter-press reactor which was operated under galvanostatic conditions using an ELCAL 924 power supply (Italy). Electrodes present a 63 cm<sup>2</sup> active surface and the gap between them was set at 10 mm. A detailed scheme of the experimental set-up can be found elsewhere [42]. All experiences were performed with a BDD anode from Adamant Technologies (La Chaux-de-Fonds, Switzerland), which was elaborated by chemical vapor deposition on a conductive substrate of Si. BDD (Adamant Technologies, La Chaux-de-Fonds, Switzerland), Zirconium, Stainless steel, and Pt (5 μm) on titanium substrate (provided by MAGNETO special anodes B.V., Schiedam, Netherlands) were employed as cathodes. Before each electrolysis, the working electrodes were anodically pretreated (40 mA/cm2 for 30 min in 0.1 M H2SO4) to clean their surfaces of any possible adsorbed impurities. Then, the system was rinsed by ultrapure water.

In a typical reaction 1 L PFOA solution (100 mg/L) with 3.5 mM Na2SO4 as electrolyte at the natural pH of the solution (pH:4) was loaded to the reservoir, preheated to 25 ◦C and recycled through the system. Once the selected temperature was reached, the power supply was turned on and current intensity was set at 0.5 A, representing this as the reaction starting time. Samples were taken at regular intervals in the tank. The global volume of samples was less than 10% of the total volume. All runs were performed by triplicate with a deviation lower than 5% in all cases.

#### *3.3. Analytical Methods*

Samples were periodically withdrawn from the reactors, filtered through 0.2 μm nylon syringe plug-in filters and immediately analyzed, without any further manipulation. PFOA concentration was measured by high performance liquid chromatography connected with an ultraviolet-visible spectrometry detector (HPLC-UV Agilent 1200 Series HPLC, Santa Clara, USA). An ion-exclusion column (ZORBAX Eclipse Plus C18, 100 mm, 1.8 μm, Agilent, USA) was used as the stationary phase. As mobile phase mixture of ACN/4 mM H2SO4 aqueous solution with a ratio: 3/2 was employed and the column temperature was set to 50 ◦C. A 60% CAN—40% mixture was employed at 0.5 mL/min. The detection UV wavelength was set to 206 nm. Total organic carbon was quantified using a TOC analyzer (Shimadzu TOC-VSCH, Kyoto, Japan). Fluoride was analyzed in an ion chromatograph with chemical suppression (Metrohm 790 IC, Herisau, Switzerland) using a conductivity detector. A Metrosep A supp 5–250 column (25 cm long, 4 mm diameter, Herisau, Switzerland) was used as the stationary phase and 0.7 mL/min of a 3.2 mM/1 mM aqueous solution of Na2CO3 and NaHCO3, respectively, as the mobile phase.

#### **4. Conclusions**

PFOA electro-degradation follows a complex mechanism which involves both oxidation reactions on the anode surface and reduction reactions, responsible for the molecule's defluorination, which take place over the cathode. Electrocatalytic hydrogenation of the unsaturated acyl fluoride RfCOF can be a route for the degradation process. Atomic hydrogen produced in situ at the catalyst surface can form simultaneously the alcohol RFCH2OH and hydrofluoric acid.

In this work, different cathodes have been used, finding that its selection plays a key role in PFOA degradation. In this sense, Pt acts as an electrocatalyst because of its higher capacity to produce in situ atomic hydrogen, which seems efficient in hydrodefluorination. It has been also demonstrated that working at low electrolyte concentration (3.5 mM Na2SO4), complete PFOA removal can be reached with up to 76.1% TOC abatement and 58.6% defluorination working at the natural pH of the solution (pH0: 4). The kind of electrolyte employed did not have a significant impact on the overall reaction. Still, slightly better results were achieved using sulfate because of the generation of sulfate radicals. Regarding the influence of the starting pH, higher TOC removal was obtained working at pH0: 9, while at higher pH values PFOA mineralization was hindered. When comparing the results obtained in this work with those reported in literature, it must be remarked that the employed BDD-Pt system allows a higher defluorination degree with a lower energy consumption. In view to render the process economically viable to treat dilute solutions, further experiments are planned to combine the electrochemical process with a preconcentration step (such as filtration or adsorption).

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4344/10/8/902/s1, Figure S1. Damaged Pt cathode after high temperature PFOA electrooxidation (T: 80 ◦C).

**Author Contributions:** A.L.G.C.: conceptualization, investigation, data curation, methodology, writing: original draft. J.A.Z.: supervision, writing: review. A.S.: supervision, data curation, validation, writing: review. K.G.S.: formal analysis, supervision, validation, writing: review. J.A.C.: formal analysis, funding acquisition, supervision, validation, writing: review. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Spanish Ministerio de Ciencia, Innovación y Universidades through project CTM2016-76454-R and by Comunidad de Madrid by P2018/EMT-4341 REMTAVARES-CM.

**Acknowledgments:** Authors thank the funding received from Ministerio de Ciencia, Innovación y Universidades through research project CTM2016-76454-R and Comunidad de Madrid for P2018/EMT-4341 REMTAVARES-CM. Alicia L. Garcia-Costa would like to thank Campus France for the mobility grant under the Make Our Planet Great Again (MOPGA) program and the Spanish Ministerio de Ciencia, Innovación y Universidades for mobility grant EST2019-013106-I. She would also like to thank both the Spanish Ministerio de Economía y Competitividad and the European Social Fund for the PhD grant BES-2014-067598.

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

#### **References**


© 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/).

### *Article* **Simulated Ageing of Crude Oil and Advanced Oxidation Processes for Water Remediation since Crude Oil Pollution**

**Filomena Lelario 1, Giuliana Bianco 1, Sabino Aurelio Bufo 1,2,\* and Laura Scrano <sup>3</sup>**


**Abstract:** Crude oil can undergo biotic and abiotic transformation processes in the environment. This article deals with the fate of an Italian crude oil under simulated solar irradiation to understand (i) the modification induced on its composition by artificial ageing and (ii) the transformations arising from different advanced oxidation processes (AOPs) applied as oil-polluted water remediation methods. The AOPs adopted were photocatalysis, sonolysis and, simultaneously, photocatalysis and sonolysis (sonophotocatalysis). Crude oil and its water-soluble fractions underwent analysis using GC-MS, liquid-state 1H-NMR, Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR-MS), and fluorescence. The crude oil after light irradiation showed (i) significant modifications induced by the artificial ageing on its composition and (ii) the formation of potentially toxic substances. The treatment produced oil oxidation with a particular effect of double bonds oxygenation. Non-polar compounds present in the water-soluble oil fraction showed a strong presence of branched alkanes and a good amount of linear and aromatic alkanes. All remediation methods utilised generated an increase of C5 class and a decrease of C6-C9 types of compounds. The analysis of polar molecules elucidated that oxygenated compounds underwent a slight reduction after photocatalysis and a sharp decline after sonophotocatalytic degradation. Significant modifications did not occur by sonolysis.

**Keywords:** crude oil; photocatalysis; sonolysis; sonophotocatalysis; FT-ICR/MS; Kendrick plot; van Krevelen diagram; water; pollution; remediation

#### **1. Introduction**

The composition of petroleum crude oil varies widely depending on the source and processing. Oil is a complex organic mixture counting for a high number of chemically distinct components, including unsaturated and saturated hydrocarbons, hetero-atoms (such as N, S, and O) and a minor percentage of metals predominantly vanadium, nickel, iron, and copper. Many oil constituents can be carcinogens, neurotoxins, respiratory irritants, hepatotoxins, nephrotoxins, and mutagens. Their toxic effects can be acute and chronic, causing many direct symptoms and major long-term injuries, including reproductive problems and cancer [1].

The hydrocarbon fraction can be as high as 90% by weight in light oils, compared to about 70% in heavy crude oil. A majority of the heteroatomic free constituents are side-byside paraffinic chains, naphthalene rings, and aromatic rings. Heteroatomic compounds constitute a relatively small portion of crude oils, less than 15%. However, they have significant implications since their presence, composition, and solubility, which depend on the origin of the crude oil, can cause either positive or negative effects in the transformation processes and are of environmental concern [2,3].

A significant consideration of the several processes affecting the crude oil spilt into the environment is needed to clarify the effects of increasingly widespread harmful events and

**Citation:** Lelario, F.; Bianco, G.; Bufo, S.A.; Scrano, L. Simulated Ageing of Crude Oil and Advanced Oxidation Processes for Water Remediation since Crude Oil Pollution. *Catalysts* **2021**, *11*, 954. https://doi.org/ 10.3390/catal11080954

Academic Editor: Fernando J. Beltrán Novillo

Received: 13 July 2021 Accepted: 4 August 2021 Published: 10 August 2021

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

**Copyright:** © 2021 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/).

predict the future fate of the oil. For this reason, the awareness of such phenomena will prove to be a valuable resource in the effort to develop innovative remediation technologies. The ecological impact of oil contamination in different environmental sections (marine, terrestrial and atmospheric) is a source of severe concern. Extraction techniques, transportation and refinery treatments of crude oil can originate pollution phenomena due to the dispersion of these compounds everywhere. These problems have attracted significant attention to understanding the fate of oil in the environment and the natural mechanisms of oil degradation and transformation to suggest a method to reduce the damages caused by original and derivative products [4–6].

As all xenobiotic substances, crude oil undergoes biotic (biotransformation by aquatic organisms such as algae, bacteria) and abiotic (hydrolysis, oxidation, photodegradation) processes, giving rise to many derivatives. In the same way as their parent molecules, these transformation products can lead to the contamination of terrestrial and aquatic environments due to oil deposition on soil and into the surface- and ground-water. Nevertheless, they can be more persistent and toxic than the parent compounds [1–3].

Extensive literature is already available on the microbiological degradation of crude oil, which received considerable attention from researchers. For example, since 1975, the biodegradability of crude oils has been studied and found to be highly dependent on their composition and incubation temperature [5]. Researchers also examined the ability of microorganisms to degrade a high number of hydrocarbons of a different structure in petroleum [6]. Furthermore, many authors have elucidated that the lighter fractions can undergo degradation more rapidly than the heavier ones, e.g., n-alkanes degraded more quickly than branched alkanes, and aromatics with two to three rings readily biodegraded through several pathways [4–8].

Photochemical processes are also essential contributors to pollutants' degradation and the removal of exogenous substances from the environment [9,10], especially in tropical and sub-tropical climates. In those areas, solar irradiation intensity is high, and the lack of nutrients hinders biological processes. Moreover, photochemical reactions are the primary cause of the compositional change of crude oil spilt in a marine environment [11–13]. Photolysis plays an essential role in the mousse formation that begins a few moments after an oil spill [12]. Due to sunlight, the interfacial tension of a crude oil film rapidly decreases, and chocolate mousse starts to form, which leads to the stabilisation of the waterin-oil emulsions [13,14]. The formation of emulsions seems to depend on the amount of asphaltene present in the oil film, and researchers reported that this amount increases upon irradiation [13]. Moreover, an increase in emulsion viscosity occurs due to the structural organisation of the asphaltenes [14].

The oxidised products resulting from the photochemical transformation significantly affect the viscosity, mousse formation, and weathered petroleum's physical properties. Moreover, photo-oxidation can lead to the destruction of existing toxic components, the generation of new toxic constituents and the formation of water-soluble products [10–14].

Since crude oil settles on the surface of water and soil, it undergoes solar irradiation. Solar degradation is a natural way for petroleum decontamination, also suggesting that techniques based on light irradiation could be helpful to the petroleum degradation processes. Light irradiation-based technologies have been improved using catalysts, the most effective and cheapest water purification tool being titanium dioxide (TiO2) [15,16]. Researchers have exploited combinations of different advanced oxidation processes (AOPs) for environmental detoxification in the last years, especially for wastewater treatment. The so-called sonophotocatalysis (SPC), the simultaneous use of ultrasound (US) and photocatalysis (PC) by semiconductors to degrade organic pollutants in water (e.g., the effluent of dye works) has been investigated, but combined AOPs methods were not applied to oil-polluted water remediation to our knowledge [17–22].

Among the analytical techniques available for structurally determining crude oil components or metabolites, gas chromatography combined with mass spectrometry (GC-MS) has been the best choice so far and most widely used [23,24]. The fractionation of crude

oil and subsequent GC-MS analysis has characterised nearly 300 components comprising aliphatic, aromatic, and biomarker compounds [25–28]. However, most crude oil fractions remain unidentified since many components cannot be resolved and appear as "hump" or "unresolved complex mixture (UCM)" in GC chromatograms [29,30].

Compositions of the saturated hydrocarbons have been better characterised by twodimensional gas chromatography coupled to mass spectrometry [29] and liquid chromatographymass spectrometry [31]. However, polar species appear poorly resolved due to their compositional complexity far exceeding the peak capacity of typical analytical techniques. High mass resolving power is necessary for the resolution of many compounds present in crude oil.

The development of Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR MS) had provided the needed ultra-high resolving power (m/Δm**50%** > 100,000, in which Δm**50%** is peak width at half peak height), and the use of electrospray ionization (ESI) mass spectrometry had made possible to detect most polar species. Thus, the coupling of these two techniques, ESI and FT-ICR mass spectrometry, produces a powerful analytical tool for analysing these polar species without a preliminary chromatographic separation [32,33].

This work investigates the modifications that the artificial ageing induced on the composition of the polar fraction of an Italian crude oil (Basilicata region—Southern Italy, Val D'Agri countryside) under solar irradiation. Moreover, it explores the possibility of oil-polluted water remediation using AOPs, such as photocatalysis (UV + TiO2), sonolysis (US, ultrasound irradiation) and the simultaneous use of photocatalysis and sonolysis, i.e., sonophotocatalysis (UV + TiO2 + US).

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

The crude oil sample, collected from the Oil Centre sited in Val D'Agri (Basilicata), underwent simulated solar treatment. Information on the composition of the oil watersoluble fraction obtained through GC-MS, liquid state 1H NMR and FT-ICR-MS was the basis for this investigation. Liquid state 1H NMR spectroscopy accomplished helpful information on the oil composition. This technique recognised amounts of 59%, 19%, and 20% of total hydrocarbons as linear, cyclic (or branched), and aromatic compounds. NMR spectroscopy cannot discriminate branched from cyclic alkanes because both compounds have the same intramolecular environment. Figure S1 compares data obtained by NMR and GC-MS.

The use of real standards introduced by an electrospray ion source allowed the calibration of mass spectra. Recalibration was necessary for the identified homologous series in each sample [34]. A troubling complication in structural studies of crude oil has been its enormous complexity on a molecular scale. The ultrahigh-resolution of FT-ICR spectra can be highly complex: these spectra typically comprise many peaks at each "Nominal mass" and thousands of peaks in a whole spectrum. Each peak could represent a chemically diverse compound. This complexity poses an investigative challenge to the study of spectra for structural interpretation.

The univocal assignment of elementary composition, merely based on the high resolution and accuracy of the instrument, is not possible for all mass values. For values higher than 400–500 Da, it is necessary to validate the result differently. The Kendrick plot (Kendrick mass defect vs Kendrick Nominal mass or KMD vs KNM) offers an outstanding vehicle to visualise and categorise all of the peaks in a mass spectrum. Kendrick mass defect (KMD) breakdown has been effectively applied to ultra-high resolution mass spectra, consenting to categorise peaks into complex spectra based on their homologous similarities across a selected type of masses [35]. Bi-dimensional plots can discern compounds differing by masses associated with a structural unit (e.g., CH2, COOH, CH2O, etc.). In this drawing, the signals of structurally related moieties all lie on horizontal or diagonal straight lines. Such a method permits the extraction of peaks that are homogeneously associated. The method can effectively recognise groups of associated compounds in FT-ICR-MS of

petroleum samples [36]. The compounds of the same homologous series (having a different number of groups CH2) will fall in a single horizontal line of the diagram (KNM), with peaks separated from 14 Da and no difference of KMD.

Similarly, the signals relating to compounds of the same class but of different types will occupy points on a vertical line of the diagram, separated by a difference of 0.013 in the Kendrick mass defect. The conversion of mass spectra from the IUPAC mass scale (based on the 12C atomic mass as exactly 12 Da) to the Kendrick mass scale is the first step. The Kendrick mass scale poses CH2 = 14.0000 Da rather than 14.01565 Da. The Kendrick mass comes from the IUPAC mass, as shown in Equation (1) [35,36]:

$$\text{Kendrick mass} = \text{IIUPAC mass} \times \text{(14.00000/14.01565)} \tag{1}$$

Members of a homologous series (specifically, compounds that comprehend the same heteroatom and number of rings plus double bonds, but a different number of CH2 groups) have the same KMDs. They are thus quickly organised and selected from a list of all detected ion masses, as shown in Equation (2):

$$\text{KMD} = \text{KNM} - \text{KEM} \tag{2}$$

where KEM is the Kendrick exact mass.

By rounding the Kendrick mass up to the nearest whole number, the nominal Kendrick mass conveniently arises. Next, homologous series are parted based on even and odd Kendrick Nominal mass and KMD, as described elsewhere [36,37]. Finally, the Kendrick masses are sorted based on Kendrick mass defect and nominal-Z value and exported into an Excel spreadsheet in the second step. Then, a molecular formula calculator programme, limited to molecular formulas consisting of up to 12C 0–80 and 16O 0–10, assigns elemental compositions. Since members of a homologous series diverge only by integer multiples of CH2, the assignment of a single unit of such a series typically suffices to identify all higher-mass members of that series [36].

We also used the van Krevelen diagram for examining ultra-high resolution mass spectra. This kind of layout is used broadly in the geochemistry literature to study the evolution of coals or oil samples [38–40]. The molar hydrogen-to-carbon ratios (H/C) constitute the ordinate, and the molar oxygen-to-carbon ratio (O/C), the abscissa. As a result, each class of compounds plots in a specific location on the diagram. Researchers well recognised that they can identify the type of compounds from the position of their representative points in the van Krevelen plot [41–43].

In general, the chemical formula CcH2c(Z)NnOoSs can identify the crude oil composition. That is because the hydrogen deficiency index <Z> of the molecule is the same for all members of a homologous "type" series (i.e., the fixed number of rings plus double bonds). Every two-units decrease in <Z> value represents the addition of one ring or a double bond. Therefore, number-average molecular weight, *Mn*, and weight-average molecular weight *Mw* have a synthetic definition as:

$$\mathbf{M}\mathbf{n} = \Sigma \mathbf{i}\mathbf{M}\mathbf{i}/\Sigma\mathbf{N}\mathbf{i} \tag{3}$$

and

$$\mathbf{M}w = \Sigma \mathbf{N} \mathbf{i} \mathbf{M} \mathbf{i}^2 / \Sigma \mathbf{N} \mathbf{i} \mathbf{M} \mathbf{i} \tag{4}$$

where Ni is the relative abundance of ions of mass Mi [34].

The <Z> number plays an essential role for the general molecular formula CcH2c(Z)X of the corresponding neutral species, in which X denotes the constituent heteroatom (Nn, Oo, and Ss).

#### *2.1. Ageing Study of Crude Oil by FT-ICR-MS*

A solar simulator (Suntest®), equipped with a xenon lamp as the light source used for the ageing treatment, provided information about the crude oil's photochemical behaviour.

GC-MS spectra showed that the fraction present in the highest percentage shifted from the C8–C11 fractions to the C13 (Figure S2) in the irradiated sample. We observe an increased amount of the C13–C23 and a decreased amount of the C7–C12 fractions. In the natural (not irradiated) oil, the C8–C11 fractions represented 54.8% of all the compounds detected. Figure S2B depicts the distribution of the compounds as a function of their chemical type. The GC-MS analysis of the mixture deriving from solar simulator irradiation showed an increase in the relative amounts of both linear alkanes and aromatic compounds. At the same time, we observed a sharp decrease in the relative amounts of branched chains. After irradiation, we did not find cyclic alkanes and alkenes.

After the irradiation, the compositional analysis of the linear alkanes highlighted several changes (Figure S2C) compared to the not irradiated sample. Undecane was the hydrocarbon found in the highest percentage in the crude oil, while pentadecane was in the irradiated oil. A decrease in the C7–C12 and an increase in C13–C25 fractions is evident. All our analytical determinations agree with reducing the numberof branched alkanes in crude oil after irradiation. Figure S2D illustrates the modifications of the composition of this fraction. After the irradiation, branched alkanes underwent a sharp reduction and only C8, C9, C11, C12, and C13 fractions were present. Cyclic alkanes were not present after the solar simulator experiment (Figure S2E).

The percentage area of the aromatic compounds did not vary with solar irradiation (Figure S2B). However, a sharp decrease in benzene-like structures and an increase in naphthalenic ones have been observed (Figure S2F).

Figure 1a,b show the FT ICR-MS spectra of untreated and treated crude oil, respectively. The spectra show the distribution of multiple ions with a single charge comprised between m/z 150 and m/z 1400. Figure 1(c1–c3) show the scale-expanded segment of the mass spectrum in Figure 1a, revealing an average period of nominal 14 mass units. The signal intensities increased after light irradiation. The shift of maximum apex was not negligible in the treated crude oil sample, which was also more viscous. Thanks to the high accuracy of mass and the excellent resolution power of FT-ICR-MS, it was possible to carry out the non-ambiguous determination of the elementary composition of multiple isobaric picks.

The chemical formula CcH2c(Z)X generally expresses the composition of a hydrocarbon molecule; where, <c> is the number of carbon atoms, <Z> is the hydrogen deficiency (a measure of aromatic character), and X represents the constituent heteroatom (N, S, O) in the molecule. The heteroatom of interest is oxygen in this study. For simplification, Kendrick and van Krevelen diagrams of natural and irradiated crude oil shown in the figures report only the O3 class, which contains the most numerous groups of detected ions. Table 1 illustrates an example of homologous series extracted by the mass spectrum of the untreated sample, with a degree of unsaturation Z = −20 and class of oxygen O3, containing the most numerous groups of detected ions.

The compounds of the same homologous series, having a different number of CH2 groups, fall in a single horizontal line of the Kendrick plot with peaks separated from 14 Da and no difference in the Kendrick mass defect (Table 1, Figure 2). The compounds of the same class but different typology settle down on a vertical line of the diagram separated from a difference of 0.0134 in the value of KMD. Figure 2 compares Kendrick plots for positive-ion ESI FT-ICR mass spectra of natural (-) and irradiated (**X**) crude oil samples. Due to the high number of signals, the figure reports only the O3 class, containing the most numerous group of detected ions. Kendrick plot of crude oil sample for O3 class shows many compounds with a high degree of unsaturation (high value of KMD). In the low values of KMD, the highest percentage of compounds has a small alkylation series (limited number of -CH2- moieties).

**Figure 1.** (**a**) FT ICR-MS spectrum of the untreated oil sample; (**b**) FT ICR-MS spectrum of the same sample irradiated by xenon-lamp. The spectrum shows the distribution of multiple ions with a single charge comprised between m/z 150 and 1400; **c1**, **c2**, **c3** insets = mass scale-expanded segments of the full range crude oil mass spectrum in Figure 1a, revealing periodicities of 14.016 Da from compound series differing in the number of CH2 groups and 2.016 Da from compound series differing in the number of rings plus double bonds.

**Figure 2.** Kendrick mass plot of the O3 species found in natural (-) and irradiated oil (**X**). This plot illustrates the increase in the number of rings plus double bonds as the KMD increases (*y*-axis) and the alkylation series along the *x*-axis.


**Table 1.** Homologous series of O3 class with <Z> = −20.

This plot can visually sort up to thousands of compounds horizontally according to the number of CH2 groups and vertically according to class (heteroatom composition) and type (rings plus double bonds). Since these two classes have the same number of oxygen atoms, they have identical O/C ratios but distinguish themselves by different H/C ratios.

The attained results elucidate the transformation of oil components following irradiation. After irradiation with the xenon lamp (Suntest®), a slight shift of the peak to the higher masses appears in the recorded mass spectra, according to Griffiths et al. findings [31]. Therefore, it seems that a phenomenon of molecular polymerisation prevails on the destruction of the tri-, tetra- and penta-aromatic groups. Furthermore, since the increase in unsaturation correlates with the higher toxicity [44], our results could indicate higher toxicity for the oil after irradiation.

The plot of Figure 2 highlights the increase in the number of double bonds' rings as the KMD increases (*y*-axis) and the alkylation series along the *x*-axis. The solar irradiation causes a diminution of rings or double bonds (picks rarefaction in samples irradiated), a consequent Kendrick Nominal mass raising of 2 Da, and the Kendrick mass defect diminution. The irradiated crude oil sample shows an expansion of alkylation in compounds with a high degree of unsaturation and a reduced unsaturation number for molecules with a low alkylation degree.

Figure 3 shows the van Krevelen plot for the class of O3 compounds found in the natural crude oil. The compounds in homologous series, corresponding to varying degrees of alkylation, appear along lines that intersect the value of 2 on the H/C axis. Similarly, a vertical line connects homologous series differing in degree of unsaturation. In agreement with the results in the Kendrick plot, most compounds have a low number of oxygen atoms and a high degree of unsaturation.

**Figure 3.** van Krevelen plot of the O3 species found in natural (-) and irradiated oil (**X**). The compounds in homologous series, corresponding to varying degrees of alkylation, appear along lines that intersect the value of 2 on the H/C axis. Similarly, a vertical line connects homologous series differing in degree of unsaturation.

As the H/C ratio increases, the number of rings plus double bonds decreases. Thus, a slight shift to a lower H/C ratio (i.e., a higher number of rings plus double bonds) occurred. Figure 3 shows a minor shift of the data to the right due to increased oxidation and slight dehydration (the picks shift to the lower left) of hydrocarbons. Kendrick mass defect analysis has dramatically facilitated the interpretation of mass spectra, but it is still challenging to derive details for molecules that contribute to complex ultrahigh-resolution mass spectra.

Figure 4 shows the distribution of compounds associated with their number of oxygen atoms in natural and irradiated samples. In both samples, the number of total oxygenated compounds increases. The augmentation of oxygenated compounds should mainly refer to the O3 and O4 types present in the investigated model. The irradiation of crude oil in the solar simulator produces oil oxidation with a particular effect of double bonds oxygenation.

**Figure 4.** FT-ICR compositional analysis of natural and irradiated crude oil samples as a function of the number of oxygen atoms.

Figure 5 shows oxygen class Z-distributions for natural and irradiated samples, confirming a diminution of hydrogen deficiency index (~15–30% less) and augmentation of oxygen number after the light irradiation. Therefore, the light irradiation induces a manifest photo-oxidation of the crude oil composition. These results highlight toxicity as most of the new oxidised compounds are water-soluble, available in higher concentrations to the living organisms and probably more reactive and biologically active than their parent compounds [43,44].

**Figure 5.** Oxygen class <Z>-distributions for natural and irradiated samples.

#### *2.2. Remediation of Oil-Polluted Water*

Since crude oil lies over the surface of water and soil, it suffers solar irradiation. Solar degradation is one of the natural ways for petroleum decontamination, and, as a consequence, techniques based on light irradiation could be advantageous in the petroleum degradation processes. Enhanced light irradiation-based technologies are available, adopting different approaches for the scope [10–12,45].

The accidental dispersion of crude oil in water bodies forms a characteristic thin layer of not water-miscible compounds and a deeper layer of solubilised substances, which cannot easily separate from the aqueous solvent. In this direction, our approach was to prepare a water/oil suspension and investigate the efficiency of different cleaning methods. The water-soluble fraction of crude oil was undergone degradation by photocatalysis, sonolysis, and sonophotocatalysis, i.e., the simultaneous use of UV, titanium dioxide, and ultrasound emitter (UV + TiO2 + US). GC-MS, liquid-state NMR, fluorescence, and high-resolution mass spectrometry (FT-ICR) analyses elucidated the chemical nature of water-soluble organic compounds after degradation processes and liquid-liquid extractions (LLEs). The results obtained in this study are concisely readable in Table 2.

**Table 2.** Synthetic results obtained from the different photodegradation processes of crude oil and oil water-soluble fraction (WSF) under investigation.


2.2.1. Photocatalytic Degradation

In the photocatalytic process, the water/oil suspension was treated for 1 h with UV irradiation in the presence of titanium dioxide. GC-MS analysis of WSF (Figure S3) evidenced increased C5 compounds from 67% in not-treated WSF to 89% in the irradiated sample. Moreover, the amount of C6, C7, C8, and C9 compounds decreased. The analysis of chemical classes occurring in the irradiated WSF showed increased branched and cyclic

alkanes, from 50% to 65% (branched) and from 4% to 7% (cyclic), respectively. On the other hand, the number of linear alkanes underwent a slight decrease (from 22% to 14%), and the aromatic compounds had a sharp decline (from 23% to 13%).

1H-NMR spectra (Figures S4 and S5) confirmed a slight increase of linear and cyclic alkanes, and a sharp decrease in aromatics, as evidenced in the chromatographic analysis.

FT-ICR MS analysis showed a minor decrease in the total number of oxygenated compounds. The O1 and O2 classes prevailed over the other types (Figure 6).

**Figure 6.** FT-ICR MS analysis of natural and 1-h photocatalysed WSF of crude oil as a function of the number of oxygen atoms.

Comparison of Kendrick plots constructed for the untreated sample (Figure 7a) and the photodegraded model (Figure 7b) shows an increase in the number of compounds with low molecular weight and low degree of unsaturation. The formation of several homologous series, with KDM values of 0.124, 0.137, 0.150, and so on, is underlined in the Kendrick diagram plotted for the treated sample. The unsaturation degree of these homologous series falls in the range Z = −16 to Z = −20.

**Figure 7.** Kendrick mass plots of the O1-O10 species found in the untreated crude oil WSF (**a**) and the 1-h photocatalysed sample (**b**).

The van Krevelen diagram (Figure 8) shows an increase in O1 class, reduced O/C ratio, and decreased unsaturated compounds in the treated sample compared to the untreated one. After photocatalysis, the number of compounds with a low number of oxygen atoms increased. As shown in Figure 6, the O1 and O2 classes prevailed over the other types, resulting in a decrease in the O/C ratio.

**Figure 8.** The van Krevelen plots of the O1–O10 species found in the untreated crude oil WSF (**a**) and the 1-h photocatalysed sample (**b**).

From fluorescence spectra (Figure S6), it was possible to argue aromatic compounds' decrease after 1-h photocatalytic treatment. Essentially, the absolute intensity of the peak at 347 nm decreases from 92.07 mAU for the natural sample to 49.50 mAU for the treated sample, with a reduction of 46%.

#### 2.2.2. Ultrasonic Irradiation

In the sonolytic process, the water/oil suspension received 1-h ultrasound irradiation. GC-MS analysis (Figure S7) indicated that C5 compounds increased from 67% to 91% at the end of sonolysis, evidencing a behaviour analogous to photocatalysis. Furthermore, the C6, C7 and C8 compounds decreased, similarly to the photocatalytic process; otherwise, the C9 class disappeared. The analysis of functional groups evidenced that branched alkanes increased from 50% to 54% in the sonolysed WSF (from 50% to 65% in photocatalysis). Cyclic alkanes underwent a minor increase from 4% to 5% (like photocatalysis), but aromatic compounds slightly increased from 23% to 24% (decreased dramatically to 13% in photocatalysis). The number of linear alkanes decreased from 22% to 17% (22% to 14% in photocatalysis).

Liquid-state 1H-NMR spectra (Figures S8 and S9) evidenced a relatively equal amount of the three classes of compounds in the not-treated and sonolysed samples. In conclusion, no significant differences emerged in the composition of WSF before and after the processes of sonication and photocatalysis, with a unique exception for aromatic compounds, as mentioned above in the case of photocatalysis.

From FTICR MS analysis, the total number of oxygenated compounds registered a low increase in the sonicated sample. O1, O2 and O7 classes increased, but the other oxygenated types decreased (Figure 9).

**Figure 9.** FT-ICR analysis of natural and 1-h sonicated WSF of crude oil samples asa function of the number of oxygen atoms.

Analysis of the Kendrick plot (Figure 10) after the degradation treatment shows a sharp increase in the number of compounds with low molecular weight. Seventy percent of compounds stay in the range m/z 159–597, and many homologous series with a high degree of unsaturation are visible.

**Figure 10.** Kendrick mass plot of the O1–O10 species found in the not-treated WSF of crude oil (**a**) and after 1-h of US treatment (**b**).

The van Krevelen diagram (Figure 11) substantiates any differences between the natural and treated WSF samples.

The fluorescence study (Figure S10) confirms that the decrease of aromatic compounds is not so evident with US treatment. After 1-h of the sonolytic process, the absolute intensity of the maximum peak displays an insignificant drop from 99.96 mAU for the natural sample to 93.37 mAU for the treated one.

**Figure 11.** The van Krevelen plots of the O1-O10 species found in the not-treated WSF of crude oil (**a**) and after 1-h US treatment (**b**).

2.2.3. Sonophotocatalytic Degradation

The contemporary use of UV irradiation, titanium dioxide and ultrasound irradiation to treat the oil aqueous suspension shows results mainly similar to those obtained with sonolysis or photocatalysis.

GC-MS analysis (Figure S11) demonstrated that after 1-h of treatment, C5 compounds increased from 67% in not-treated WSF to 91% in the treated sample, while C6, C7 and C8 compounds decreased; C9 compounds were not detected (like the simple US). The analysis of functional groups in the sonophotocatalytic degradation evidenced an increase from 50% to 64% of branched alkanes (similar to photocatalysis) and from 4% to 9% of cyclic alkanes (higher than the other technologies). The number of linear alkanes underwent a slight decrease (from 22% to 19%, similar to the other technologies). In comparison, aromatic compounds showed the sharpest decline (from 23% to 7%), also proved by integrating NMR spectra (Figures S12 and S13). In the natural WSF, the aromatics alkanes occupied 19% of the whole spectral area, whilst in the treated sample, this amount decreases up to 3.3%. On the other hand, the amount of linear and cyclic alkanes increases by about 7–8%.

Figure 12 shows the trend of oxygenated compounds after sonophotocatalytic treatment. In this case, the total number of oxygenated compounds decreased from 1203 (not treated WSF) to 993 (treated WSF).

**Figure 12.** FT-ICR MS analysis of natural and 1-h sonophotocatalysed WSF sample of crude oil as a function of the number of oxygen atoms.

Comparison of Kendrick plots (Figure 13) obtained for the not-treated and treated samples showed an increased number of compounds with low molecular weight, especially in the range m/z 169–369, and compounds with a low unsaturation degree.

**Figure 13.** Kendrick mass plots of the O1-O10 species found in the not-treated WSF of crude oil (**a**) and after 1-h of sonophotocatalytic treatment (**b**).

The van Krevelen diagram (Figure 14) let us see an intensification of signals relative to oxygenated compounds with an O/C ratio in the range 0.10–0.25.

**Figure 14.** The van Krevelen plots of the O1-O10 species found in the not-treated WSF of crude oil (**a**) and after 1-h of sonophotocatalytic treatment (**b**).

Fluorescence spectra (Figure S14) substantiated the decreasing of aromatic compounds after 1-h sonophotocatalytic treatment. The absolute intensity of the peak at 347 nm decreased from 89.92 mAU for the natural sample to 47.95 mAU for the treated sample, reducing by 48%.

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

#### *3.1. Crude Oil and Chemicals*

The director of the Eni-Cova Oil Plant in Val d'Agri (Basilicata Region, Southern Italy) kindly provided the oil sample taken from the first step of oil purification after extraction, including dehydration and degasification. Table 3 accounts for the elemental composition reported in the label accompanying the sample delivered for this research.

**Table 3.** Elemental composition (%) of the oil sample taken from the first step of oil purification after extraction, including dehydration and degasification a.


<sup>a</sup> Metals (Ni and V) < 1000 ppm. <sup>b</sup> Obtained as the complement to 100.

All chemicals used were of analytical grade. TiO2 Degussa P-25, obtained as a gift from Evonik (Hanau, Germany), was the catalyst adopted. Table 4 reports a summary scheme of the investigation executed.


**Table 4.** Experiments performed and analytical methods used in this study.

#### *3.2. Photodegradation Apparatus*

The Suntest CPS+ (Heraeus Industrietechnik GmbH, Hanau, Germany), equipped with a xenon lamp of 1.1 kW, protected employing a quartz plate (total passing wavelength: 300 nm < λ < 800 nm), was the solar simulator adopted for photochemical reactions. The temperature of the irradiation chamber was 25 ◦C, maintained through both a thermostatic bath and a conditioned airflow. During the experiments, the crude oil samples were kept up in the horizontal position, creating a homogeneous film of 0.5 cm thickness.

#### *3.3. Photodegradation Process and Sample Preparation for ESI FT-ICR MS*

The protocol used for oil ageing experiments was: (i) the irradiation of the natural crude oil (10 mL) for a week in the borosilicate planar reactor; (ii) crude oil samples preparation by dissolving ~30 mg of material in 30 mL of toluene; (iii) withdrawal of 1 mL solution and its dilution with 0.5 mL methanol; addition of either 10 μL acetic acid (for positive ion ESI) or 10 μL ammonium hydroxide (for negative ion ESI) to facilitate protonation or deprotonation in the electrospray ionisation process, respectively.

#### *3.4. Ultrasonic Irradiation of WSF Samples*

A crude oil/water suspension was arranged in a borosilicate decanter (5 L) equipped with a Teflon tap at the bottom. The decanter was filled with 3.5 L of ultrapure water; the crude oil was added at the ratio of 1/20 (oil/water), and the suspension was magnetically stirred and then kept in the dark for 30 days at constant temperature (25 ◦C) to reach equilibrium and the separation of oil phase on the surface of the aqueous phase. Aqueous samples were drawn off through the Teflon tap without disturbing the oil/water separation surface. The collected aqueous sample (500 mL) underwent cotton filtration, to avoid the formation of an emulsion in the solution. The ultrasonic degradation tool was the immersible ultrasonic emitter Sinaptec Nexus P198-R (Sinaptec, Lezennes, France), an ultrasonic module furnished with a titanium sonotrode (S23-10-1/2, Sinaptec), an electrical signal of frequency close to 20 kHz, and a voltage of about 1 kV. In this configuration, the electric power provided by the generator (Nexus P198-R, same manufacturer) is adjustable between 7 W and 100 W, as indicated on a digital display panel. However, this electrical measurement does not determine with high precision the acoustic power dissipated in the liquid. The experimental temperature was fixed to 25 ◦C.

#### *3.5. Photocatalytic and Sonophotocatalytic Degradation of WSF Samples*

The photocatalytic method to degrade the water-soluble fraction of crude oil utilises a 125 W high-pressure mercury lamp (Philips-HPK, Philips, Turnhout, Belgium) that provides its maximum energy at 365 nm, with a range of emission from 195 to 580 nm. The catalyst was TiO2 (80% anatase–20% rutile). The simultaneous use of the mercury lamp, titanium dioxide and the ultrasound emitter (UV + TiO2 + US) permitted the sonophotocatalytic degradation. The experimental temperature was 25 ◦C for photocatalysis and sonophotocatalysis trials.

#### *3.6. Liquid–Liquid Extraction (LLE)*

The experimental protocol was (i) to collect samples of the oil WSF after 15, 30, 45, and 60 min of treatment; (ii) to extract in triplicate 30 mL of each sample in a separatory funnel (50 mL) with 3 mL dichloromethane for GC-MS analysis and (iii) another 30 mL with the same solvent for 1H-NMR spectroscopy; (iv) to perform fluorescence analysis using 5 mL of the aqueous solution without extraction.

The internal standard used for assessing the reproducibility of WSF extraction was 1 mL of 1,3-dibromopropane (26.7 mg L−1) added to the volume of dichloromethane needed for each liquid–liquid extraction. In addition, it was necessary to add 1.0 mL of 1-bromododecane (29.0 mg L<sup>−</sup>1) to evaluate the GC-MS analysis reproducibility at the end of each extraction. Thus, the injection volume was 1 μL extract for each GC-MS run.

#### *3.7. Analysis of Fluorescence*

A research-grade spectrofluorometer FP-6500 Jasco (Jasco Corporation, Cremella, Italy) was available for fluorescence analysis. This analysis was necessary to appreciate the aromatic compounds remaining in the water-soluble fraction of crude oil. The spectrofluorometer FP-6500 Jasco, adopting as emitting source a DC-powered 150 W xenon lamp (in a sealed housing), employs a photometric rationing system, which utilises a second photomultiplier tube to monitor and compensate for any variations in the intensity of the xenon source, thus ensuring maximum analytical stability. Furthermore, a concave holographic grating monochromator with optimised blaze angles provides maximum sensitivity over the entire wavelength range; (220–750 nm; 1 nm resolution).

The fluorescence optical path adopted was 1 cm in quartz cells (volume ca 5 cm3) at 237 and 320 nm excitation and 347 and 360 nm emission.

#### *3.8. 1H-NMR Analysis*

A Varian Oxford AS400 spectrometer (Palo Alto, CA, USA), operating at 400 MHz, was enough for the 1H-NMR spectra recording. The set temperature for the used 5 mm non-gradient broadband inverse (BBI) probe was 25 ◦C.

All the 1H-NMR spectra have tetramethylsilane (TMS) as reference under the acquisition parameters shown in Table 5.

In degradation experiments, liquid–liquid extraction with chloroform permitted to isolate the water-soluble fraction of crude oil. After the complete evaporation of the solvent in a rotary evaporator, the addition of 500 μL deuterated chloroform (CDCl3) permitted to recuperate the residual organic mixture.


**Table 5.** 1H-NMR acquisition parameters.

*3.9. Mass Spectrometry of Polar Components*

The instrument available to determine polar components was the micro ESI/FT-ICR/MS 7 T Thermo Electron (Waltham, MA, USA).

The method used for the routine analyses permitted a mass accuracy better than 2 ppm by external calibration, using the mixture of caffeine, MRFA, and Ultramark. The technique separated more than 6000 ion signals belonging to chemically different elemental compositions with a 200,000 resolving power (m/Δm**50%** at *m*/*z* 400) in positive electrospray mode. The robustness of this equipment, combined with unprecedented ease of use, ultra-high mass accuracy, high sensitivity, and excellent resolving power, make it an ideal instrument for analysis. The infusion of the samples at a flow rate of 5 μL/min permitted the best result in terms of spectrum resolution. ESI conditions were: needle voltage, +4.5 kV; heated capillary current 4 A; tube lens voltage 135.12; temperature 300 ◦C; N2 speed 2.33 u.a.; aux gas flow rate 0.73; scansions per second 1000.

#### **4. Conclusions**

In this study, we tried to characterise the ageing process of crude oil simulating solar irradiation on a thin layer of an oil sample. As a result, FT-ICR MS evidenced an augmentation of compounds with low molecular weight and a slight increase of the number of oxygen atoms in the oxygenated species, as depicted in Kendrick and van Krevelen diagrams. Furthermore, the simulated ageing produced the oxidation of crude oil with a particular effect on double bonds' oxygenation, as confirmed by the disappearance of alkenes in gas chromatographic analysis. The observed results seem to be recognisable because the energy irradiated with the xenon lamp could be enough for catalysing the reaction of olefins with the atmospheric oxygen following a bridge mechanism.

We experimented with different solutions for the cleaning treatment of oil-polluted water (photocatalysis, sonolysis, and sonophotocatalysis). GC-MS analyses of the watersoluble fraction of crude oil for both natural and treated samples discovered that only a few compounds are detectable in the aqueous solution, principally C5-organic chains (~50%). Low amounts of C6, C7, C8 and C9 chains were also present. Both GC-MS and liquid state 1H-NMR signals showed that the branched alkanes were the principal chemical class in the soluble fraction of oil, followed by a small amount of linear and aromatic alkanes.

With all the degradation methods utilised, an increase of the C5-class and a decrease of C6–C9 types of compounds was evident. Furthermore, the FT-ICR comparative analyses of oxygenated species elucidated that the total number of O-compounds in the treated WSF samples is different for all of the degradation methods experimented. The number of the oxygenated compounds slightly decreased with photocatalysis compared to the non-treated sample. An opposite trend appeared with the sonolysis treatment. The sonophotocatalytic method showed a sharp reduction in the number of oxygenated compounds, probably due to the volatilisation of small molecules formed during the oxidation process. It is conceivable that ultrasound can promote this volatilisation. The degradation of the watersoluble fraction of crude oil performed with photocatalysis and sonophotocatalysis led to an apparent decrease of aromatic compounds of 46% and 48%, respectively, for the two techniques, as also confirmed by the fluorescence analysis. With the use of sonolysis, there was no effect on the number of aromatic compounds. Nevertheless, all the degradation methods applied were capable of increasing the number of cyclic alkanes. Therefore, we could speculate that ultrasound in sonophotocatalytic technology can affect the rate of the

photocatalytic degradation of the organic pollutants due to a synergistic effect typically observed with an increase of the degradation process efficiency.

In conclusion, our results confirm the photo-oxidation effect caused by light irradiation either on crude oil (simulated ageing) or on the soluble oil fraction. Naturally, the behaviour of each oil type could be different, and then it is not possible to generalise our findings to all cases of oil spilling and environmental remediation. Therefore, it is necessary to check case by case before reaching specific solutions for more efficient remediation processes to avoid making the situation worse.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/article/10 .3390/catal11080954/s1 Figure S1: Percentage of compounds in crude oil as recognized by GC-MS (blue column) and 1H-NMR (red column). Figure S2: GC-MS compositional analysis of crude oil (blue column) and solar simulator irradiated crude oil (red column), as a function of the number of carbon atoms (A); composition of crude oil as a function of the type of compounds, LH: linear aliphatic hydrocarbons; BH: branched aliphatic hydrocarbons; CH: cyclic aliphatic hydrocarbons; AH: aromatic hydrocarbons; AL: alkenes (B); composition of the linear aliphatic hydrocarbons as a function of the number of carbon atoms (C); composition of the branched aliphatic hydrocarbons fraction as a function of the number of carbon atoms (D); composition of the cyclic hydrocarbons as a function of the number of carbon atoms (E); composition of the aromatic hydrocarbons fraction as a function of the number of carbon atoms (F). Figure S3: GC-MS compositional analysis of WSF crude oil before (red column) and after (blue column) photocatalysis: distribution of hydrocarbons as a function of the number of carbon atoms (A) and distribution of the compounds as a function of chemical species (B). Figure S4: 1H-NMR spectra of WSF crude oil before (A) and after (B) photocatalysis. Figure S5: 1H-NMR compositional analysis of WSF crude oil before (red column) and after (blue column) photocatalysis: distribution of the compounds as a function of chemical species. Figure S6: Fluorescence spectra of WSF crude oil before (blu line) and after (red line) photocatalysis. Figure S7: GC-MS compositional analysis of WSF crude oil before (red column) and after (blue column) sonolysis: distribution of hydrocarbons as a function of the number of carbon atoms (A) and distribution of the compounds as a function of chemical species (B). Figure S8: 1H-NMR spectra of WSF crude oil before (A) and after (B) sonolysis. Figure S9: 1H-NMR compositional analysis of WSF crude oil before (red column) and after (blue column) sonolysis: distribution of the compounds as a function of chemical species. Figure S10: Fluorescence spectra of WSF crude oil before (blu line) and after (red line) sonolysis. Figure S11: GC-MS compositional analysis of WSF crude oil before (red column) and after (blue column) sonophotocatalysis: distribution of hydrocarbons as a function of the number of carbon atoms (A) and distribution of the compounds as a function of chemical species (B). Figure S12: 1H-NMR spectra of WSF crude oil before (A) and after (B) sonophotocatalysis. Figure S13: 1H-NMR compositional analysis of WSF crude oil before (red column) and after (blue column) sonophotocatalysis: distribution of the compounds as a function of chemical species. Figure S14: Fluorescence spectra of WSF crude oil before (blu line) and after (red line) sonophotocatalysis.

**Author Contributions:** Conceptualisation, F.L., S.A.B. and L.S.; Data curation, F.L., G.B. and S.A.B.; Formal analysis, L.S.; Investigation, L.S.; Methodology, G.B.; Resources, S.A.B.; Writing—original draft, F.L. and L.S.; Writing—review and editing, S.A.B. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Data Availability Statement:** Data supporting results reported in this paper can be found at the Department of Sciences, University of Basilicata, Via dell'Ateneo Lucano 10, 85100, Potenza, Italy. Due to the multitude of files produced for this investigation, a link will be provided under specific requests.

**Acknowledgments:** We acknowledge the Director of ENI-COVA Oil Plant in Val d'Agri (Basilicata Region, Southern Italy), who kindly provided the oil sample. Special thanks go to Mauro Tummolo, and to Jean-Marc Chovelon, University Claud Bernard Lyon 1, for the sonolysis and sonophotocatalysis data.

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

#### **References**


### *Article* **Visible Light Responsive Strontium Carbonate Catalyst Derived from Solvothermal Synthesis**

#### **Pornnaphat Wichannananon 1,2, Thawanrat Kobkeatthawin <sup>2</sup> and Siwaporn Meejoo Smith 2,\***


Received: 24 July 2020; Accepted: 11 September 2020; Published: 17 September 2020

**Abstract:** A single crystalline phase of strontium carbonate (SrCO3) was successfully obtained from solvothermal treatments of hydrated strontium hydroxide in ethanol (EtOH) at 100 ◦C for 2 h, using specific Sr:EtOH mole ratios of 1:18 or 1:23. Other solvothermal treatment times (0.5, 1.0 and 3 h), temperatures (80 and 150 ◦C) and different Sr:EtOH mole ratios (1:13 and 1:27) led to formation of mixed phases of Sr-containing products, SrCO3 and Sr(OH)2 xH2O. The obtained products (denoted as 1:18 SrCO3 and 1:23 SrCO3), containing a single phase of SrCO3, were further characterized in comparison with commercial SrCO3, and each SrCO3 material was employed as a photocatalyst for the degradation of methylene blue (MB) in water under visible light irradiation. Only the 1:23 SrCO3 sample is visible light responsive (Eg = 2.62 eV), possibly due to the presence of ethanol in the structure, as detected by thermogravimetric analysis. On the other hand, the band gap of 1:18 SrCO3 and commercial SrCO3 are 4.63 and 3.25 eV, respectively, and both samples are UV responsive. The highest decolourisation efficiency of MB solutions was achieved using the 1:23 SrCO3 catalyst, likely due to its narrow bandgap. The variation in colour removal results in the dark and under visible light irradiation, with radical scavenging tests, suggests that the high decolourisation efficiency was mainly due to a generated hydroxyl-radical-related reaction pathway. Possible degradation products from MB oxidation under visible light illumination in the presence of SrCO3 are aromatic sulfonic acids, dimethylamine and phenol, as implied by MS direct injection measurements. Key findings from this work could give more insight into alternative synthesis routes to tailor the bandgap of SrCO3 materials and possible further development of cocatalysts and composites for environmental applications.

**Keywords:** strontium carbonate (SrCO3); solvothermal method; photocatalysis; visible light

#### **1. Introduction**

Textile industries employ over 10,000 dyes and pigments in the manufacturing of cotton, leather, clothes, wool, silk and nylon products [1–3]. An estimated 700,000 tons or more of synthetic dyes are thought to be annually discharged into the environment [4], causing serious water pollution as many of these dyes are toxic, highly water soluble and highly stable against degradation by sunlight or increased temperature [5]. Therefore, effective treatments of dye-contaminated water have continuingly received great attention by academic and industrial sectors. Various wastewater treatment methods have been applied to remove toxic dyes from wastewater, such as coagulation–flocculation, adsorption, membrane separation, biodegradation and oxidation processes [6]. Among these methods, photocatalytic oxidation processes have been proven to be simple and effective at organic dye decomposition, forming relatively low toxic by-products with potential mineralization to generate CO2 and H2O [7–9]. In this process, under light irradiation a semiconducting catalyst absorbs photon energy promoting electron transfer from the valence band (VB) to the conduction band (CB), resulting in electron-hole pair generation. The generated holes (h+) further react with water molecules while the electrons (CB) react with oxygen, resulting in formation of active hydroxyl (•OH) and superoxide (•O2 <sup>−</sup>) radicals, respectively. The •OH radicals subsequently attack organic pollutants in water leading to oxidative degradation of pollutants.

Wide bandgap TiO2 [10] and ZnO [11] semiconducting materials have proven to be efficient catalysts for the photo-oxidation of organic pollutants in water. However, these require high-energy ultraviolet irradiation, which requires special and costly safety protocols to be in place for the use of these materials in wastewater treatment. An attractive alternative is to use harmless visible light sources in the photoreactor, employing a visible light responsive photocatalyst for pollutant degradation. Such visible light responsive photocatalysts need to promote the photo-oxidative degradation of pollutants using sunlight (7% UV and 44% visible light emission, and other low-energy radiations [12]) to ensure wastewater treatment is a sustainable process. Strontium carbonate (SrCO3) is a common starting material for the manufacture of colourants in fireworks, glass cathode-ray tubes and computer monitors [13,14]. While commercially available SrCO3 material is commonly derived from celestine (SrSO4) mineral via calcination followed by Na2CO3 treatment (the black ash method) [15], synthetic SrCO3 can be obtained using calcination and wet chemical methods under ambient [16,17] or high-pressure [18] atmospheres. Table 1 summarizes the key features (synthesis conditions, characteristics and the bandgap energy) of synthetic SrCO3 in the literature. Methylene blue, a cationic organic dye and a common colouring agent used in cotton, wood, silk [19] cosmetics, and textile [20] dying is a frequently utilized representative dye pollutant mimicking those present in industrial effluents. Song and coworkers reported effective methylene blue (MB) degradation under visible light irradiation (λ > 400 nm) after 3 h treatment with SrCO3 obtained from the calcination of synthetic Sr(OH)2 [21], while Molduvan and coworkers reported the removal of MB from aqueous solutions using a commercial natural activated plant-based carbon [22]. Other works have utilized SrCO3 as a cocatalyst incorporated in photocatalyst composites, e.g., Ag2CO3/SrCO3 [23], TiO2/SrCO3 [24] and SrTiO3/SrCO3 [25], to expand the photoresponsive range of the material and to improve its catalytic activity and reaction selectivity.


**Table 1.** Synthesis, key characteristics and bandgap energy of synthetic SrCO3.

This work investigated the effects of precursor concentrations (Sr:ethanol mole ratios), solvothermal temperatures and treatment times on the properties of SrCO3 materials and their photocatalytic degradation of MB in water under visible light irradiation, as a function of pH and temperature. Kinetic and mechanistic studies of the MB degradation process were carried out through reaction rate determination and identification of the end-products. The photocatalytic performance of synthesized SrCO3 was compared with that of commercially available material, in order to derive insights into the relationships between properties and catalytic activity.

#### **2. Results and Discussions**

#### *2.1. E*ff*ects of Synthesis Conditions*

Solvothermal treatments of strontium nitrate in ethanol (EtOH) were carried out at various temperatures (80, 100 and 150 ◦C), treatment times (0.5, 1, 2 and 3 h) and Sr:EtOH mole ratios (1:13, 1:18, 1:23 and 1:27). From powder X-ray diffraction (PXRD) results in Figure 1, a single phase of SrCO3 was obtained from two conditions: 2 h solvothermal treatment at 100 ◦C using a Sr:EtOH mole ratio of 1:18 or 1:23. These samples are denoted as 1:18 SrCO3 and 1:23 SrCO3 in further discussions. Notably, mixed phases of SrCO3 and hydrated strontium hydroxides (Sr(OH)2·xH2O, where x is the number of molar coefficient of water in strontium hydroxide solid) were obtained from all other synthesis conditions (results shown in Supplementary Materials: Figures S1 and S2). Typical diffraction peaks correspond well with (110), (111), (021), (002), (012), (130), (220), (221), (132) and (113) orthorhombic SrCO3 lattice planes [34,36], whereas other diffraction peaks match with those of previously reported Sr(OH)2·H2O [37] and Sr(OH)2·8H2O phases [38]. The formation of Sr(OH)2 xH2O is possibly due to adsorbed alcohol, promoting the addition of OH functional groups on the solid surface [39], upon solvothermal crystallization of Sr-containing products.

**Figure 1.** PXRD patterns of Sr-containing samples derived from 2 h solvothermal treatments of hydrated strontium hydroxide in ethanol at various Sr:EtOH mole ratios (1:13, 1:18, 1:23 or 1:27) at 100 ◦C.

FTIR spectra of the prepared Sr-containing samples are shown in Figure 2. The absorption bands located within 1700–400 cm−<sup>1</sup> regions were attributed to the vibrations in CO3 <sup>2</sup><sup>−</sup> groups. The strong broad absorption at 1470 cm−<sup>1</sup> was considered to be due to an asymmetric stretching vibration, and the sharp absorption bands at 800 cm−<sup>1</sup> and 705 cm−<sup>1</sup> can be specified to the bending out-of-plane vibration and in-plane vibration, respectively. The weak peak at 1770 cm−<sup>1</sup> indicated a combination of vibration modes of the CO3 <sup>2</sup><sup>−</sup> groups and Sr2<sup>+</sup>. The sharp peak at 3500 cm−<sup>1</sup> was assigned to the stretching mode of –OH- in Sr(OH)2, and the broad absorption peak around 2800 cm−<sup>1</sup> was assigned to the stretching mode of H2O in Sr(OH)2·H2O and Sr(OH)2·8H2O. These results are consistent with the commercial SrCO3 and the 1:18 SrCO3 and 1:23 SrCO3 samples being of similar chemical composition.

**Figure 2.** FTIR spectra of Sr-containing samples derived from 2 h solvothermal treatment of hydrated strontium hydroxide in ethanol at various Sr:EtOH mole ratios (1:13, 1:18, 1:23 or 1:27) at 100 ◦C.

SEM images of the obtained SrCO3 materials (derived from Sr:EtOH mole ratios of 1:18 or 1:23) are compared with those of commercial SrCO3 in Figure 3. Whisker-like SrCO3 and spherical particles were obtained under these respective synthesis conditions. Figure 3c highlights the relatively large rod-like particles of commercial SrCO3. Variation in particle sizes was observed in solvothermally obtained SrCO3, with particle sizes being smaller for the 1:18 SrCO3 samples. Notably, commercial SrCO3 contains much larger particles than those of the synthesized material. From literature [26,27], SrCO3 production plants utilize two common methods, the black ash method and the soda method, in conversion of celestine ore (SrSO4) to SrCO3 (Table 1). The black ash method involves high-temperature calcination of the ore to obtain SrS, with crystalline SrCO3 solid being formed after dissolving the SrS in aqueous Na2CO3, followed by precipitation. The soda method produces SrCO3 through the two-step decomposition reaction between celestine and aqueous Na2CO3, to obtain precipitated SrCO3. From this information, as the formation of commercial SrCO3 does not require high temperatures (>150 ◦C) for solvent evaporation and precipitation of SrCO3, the larger grain size of the commercial SrCO3 sample is probably due to the fast solvent evaporation during the precipitation processes.

**Figure 3.** SEM images of SrCO3 derived from 2 h solvothermal treatment at 100 ◦C, using Sr:EtOH mole ratios of (**a**) 1:18 and (**b**) 1:23 compared with (**c**) commercial SrCO3.

Thermogravimetric analysis (TGA) plots (Figure 4) suggest thermal stability of all SrCO3 samples up to 600 ◦C. Slight weight loss (<1%) was likely due to moisture or solvent residue [40]. The 1:23 SrCO3 sample gives a relatively high weight loss of 0.21%, which corresponds to the removal of surface adsorbed moisture and ethanol (weight loss upon heating up to 400 ◦C) and the loss of ethanol from the SrCO3 lattice at ca. 450 ◦C. Decomposition of SrCO3 takes place at temperatures above 800 ◦C as a result of conversion to SrO.

**Figure 4.** Thermogravimetric analysis (TGA) plots of SrCO3 samples prepared using Sr:EtOH mole ratios of 1:18 and 1:23, compared with that of commercial SrCO3.

Based on the PXRD and TGA results, chemical transformation of hydrated strontium hydroxide in the presence of ethanol under solvothermal treatments leads to the formation of SrCO3 and ethanol incorporated SrCO3 materials, as proposed by the reactions below. In general, CO2 in air can react with strontium hydroxide to form SrCO3, which precipitates after the sonication step and solvothermal treatments. Ethoxide could be formed under basic conditions, resulting in an CH3CH2O···Sr2+···OCH2CH3 intermediate, which is subsequently transformed to ethanol incorporated in SrCO3. Note that the amount of ethanol incorporated within the SrCO3 is sufficiently low, such that a single phase of SrCO3 was observed in PXRD pattern of the 1:23 SrCO3 sample.

$$\text{Sr(OH)}\_{2} + \text{CO}\_{2} \rightarrow \text{SrCO}\_{3} + \text{H}\_{2}\text{O}$$

$$\text{Sr(OH)}\_{2} + \text{CO}\_{2} \rightleftharpoons \text{Sr}^{2+} + \text{HCO}\_{3}^{-} + \text{OH}^{-}$$

$$\mathrm{HCO\_3^- + OH^- \to CO\_3^{2-} + H\_2O}$$

$$\mathrm{CH\_3CH\_2OH + OH^- \rightleftharpoons CH\_3CH\_2O^- + H\_2O}$$

$$\mathrm{Sr(OH)\_2 + H\_2O \rightleftharpoons Sr^{2+} (aq) + OH^- (aq)}$$

$$\mathrm{Sr^{2+} + 2CH\_3CH\_2O^- \to Sr^{2+} \cdots 2OCH\_2CH\_3}$$

$$\mathrm{Sr^{2+} \cdot \mathrm{OCH\_2CH\_3} + HCO\_3^- \to SrCO\_3 \cdots HOCH\_2CH\_3}$$

#### *2.2. Optical Properties*

UV–VIS diffuse reflectance spectra of the 1:18 SrCO3, 1:23 SrCO3 and commercial SrCO3 in Figure 5a showed that the characteristic absorption edge of the 1:23 SrCO3 sample is located in the visible light region (473 nm), whereas the spectral response of other SrCO3 samples was observed in the UV region, with absorption band edges of 268 and 381 nm for the 1:18 SrCO3 sample and commercial SrCO3, respectively. The band gap energy values suggested by Kubelka–Munk plots (Figure 5b) are 4.63 and 3.25 eV for 1:18 SrCO3 and commercial SrCO3, respectively. By contrast, the bandgap energy of the 1:23 SrCO3 sample is 2.62 eV, and its visible response is possibly due to the presence of incorporated ethanol in the solid sample, as suggested by TGA results.

**Figure 5.** (**a**) UV-visible diffuse reflectance spectra and (**b**) Kubelka–Munk plots of the SrCO3 synthesized at Sr:EtOH mole ratios of 1:18 and 1:23, compared with those of commercial SrCO3.

#### *2.3. Decolourisation of Methylene Blue (MB)*

Figure 6a illustrates the colour removal efficiencies of 10 ppm MB aqueous solutions in the dark and under visible light irradiation after 1 h treatment with SrCO3. Similar colour removal efficiencies from treatment of MB(aq) with 1:18 SrCO3 in the dark and under light illumination suggested major adsorption processes occurred due to the wide bandgap of the 1:18 SrCO3 sample. On the other hand, the visible responsive 1:23 SrCO3 and commercial SrCO3 gave higher colour removal efficiencies under irradiation conditions than those from dark experiments, implying both adsorption and photodegradation of MB are of importance. Therefore, from these catalyst screening tests, the colour removal efficiencies of aqueous MB solutions strongly depend on the bandgap energy of SrCO3 materials and that the 1:23 SrCO3 is the most active catalyst. Figure 6b demonstrates that only low colour removal efficiencies occur due to adsorption (in the dark) and photolysis (irradiation and no SrCO3). Treatments of dye solutions with 1:23 SrCO3 is much less effective (low colour removal efficiency) under dark conditions in comparison to decolourisation under visible light irradiation. These results suggest that the main process of MB colour removal is caused by photocatalytic treatment by using the SrCO3 photocatalyst rather than adsorption.

**Figure 6.** (**a**) Colour removal efficiencies of 10 ppm methylene blue (MB) aqueous solution in the dark, and under visible light irradiation in the presence of SrCO3. (**b**) Absorption spectra of MB in the dark or under visible light irradiation by SrCO3 (1:23 and 1:18). All decolourisation experiments were performed at 30 ◦C using SrCO3 with catalyst loadings of 4.0 g·L−<sup>1</sup> with 1 h treatment.

The percentage of MB colour removal after treatment with SrCO3 photocatalyst (sample 1:23) is shown in Figure 7a. When a suspension of SrCO3 in 10 ppm fresh MB solution was kept in the dark for 3 h, the concentration of dye slightly decreased, while the colour of the dye solution remained unchanged. It was observed that the absorption capacity of MB on the SrCO3 surface is negligible because the specific area of the prepared SrCO3 photocatalyst is low (9.23 m2·g−1). Upon visible irradiation, the prepared SrCO3 gave a high percentage of MB colour removal (>99% after 3 h visible irradiation).

**Figure 7.** (**a**) Colour removal efficiencies of 10 ppm MB aqueous solution (pH 5.5) as a function of time in the dark or under visible light irradiation; adsorption of MB into SrCO3 (loading 4 g·L<sup>−</sup>1) in the dark, MB photolysis and photocatalysis of MB under visible light illumination catalyzed by SrCO3 (loading 4 g·L<sup>−</sup>1). (**b**) Effects of a scavenger (tert-BuOH) on the colour removal efficiency of 10 ppm MB after 3 h treatment with the 1:23 SrCO3 (4 g·L<sup>−</sup>1).

In order to prove that hydroxyl radicals (•OH) are the active species in the photocatalytic degradation process, experiments were conducted in the presence of a radical scavenging reagent. One such reagent, tert-butyl alcohol (tert-BuOH), if present, should significantly inhibit the oxidation of MB [41]. The result in Figure 7b indicates that after treatment for 3 h, adding tert-BuOH resulted in poor colour removal efficiencies (6.90%), whereas in the absence of the reagent very high colour removal efficiencies (>99%) were achieved. The formation of a product arising from the reaction

between tert-BuOH and •OH as ascribed through a radical pathway [41] thus resulted in the poor activity, confirming that hydroxyl radicals are the important active species assisting MB degradation.

The effect of pH on the MB decolourisation under visible light irradiation was examined over a range of pH 3–9. The colour removal efficiency reached 73% after 1 h treatment at pH 3, while lower colour removal efficiencies were obtained at pH 5.5 (51%), pH 7 (42%) and pH 9 (29%) over the same time period, as shown in Figure 8a. In addition, the natural logarithm of the MB concentrations was plotted as a function of irradiation time, affording a linear relationship, as presented in Figure 8b. Using the first-order model, the highest rate constant of MB colour removal was obtained at pH 3, with the degradation being slowest at pH 9. The decreasing rate constants of MB decolourisation with increasing pH may be the result of the presence of carbonate (CO3 <sup>2</sup>−) and hydroxide (OH−) ions, which are radical scavengers [42,43]. At pH 5.5–10, the low colour removal efficiencies may be due to the following reactions.

$$\mathrm{CO\_3^{2-}} + \bullet\mathrm{OH} \rightarrow \mathrm{CO\_3^{\bullet-}} + \mathrm{OH^-}$$

$$\cdot \text{OH}^- + \bullet \text{OH} \rightarrow \text{H}\_2\text{O} + \text{O}^-$$

**Figure 8.** (**a**) Colour removal efficiencies of 10 ppm MB aqueous solution with time, using SrCO3 as photocatalyst. (**b**) Kinetics of MB decolourisation catalyzed by SrCO3 as a function of pH. All decolourisation experiments were performed using SrCO3 with catalyst loading of 4.0 g·L−<sup>1</sup> at 30 ◦C from pH 3–9.

The effect of temperature on the degradation of MB as a function of time is discussed in Figure 9. From Figure 9a, it can be observed that higher temperatures result in higher MB colour removal efficiencies. Under visible light irradiation, the MB colour removal efficiency reached 100% after 1 h treatment at 70 ◦C. In all cases MB concentrations decrease with irradiation time. The linear plots between the natural logarithm of the MB concentration versus irradiation time are shown in Figure 9b, which indicate that the decolourisation process follows first-order kinetics. The rate constants of MB decolourisation increased with temperature, indicating that MB removal by 1:23 SrCO3 is overall endothermic. The 1:23 SrCO3 sample is rather stable during the photocatalytic MB degradation reaction, as only negligible concentrations of Sr (<10 ppm) were detected in the treated MB solution.

**Figure 9.** (**a**) Colour removal efficiencies of 10 ppm MB aqueous solution (pH 5.5) over time using SrCO3 as photocatalyst. (**b**) Kinetics of MB decolourisation catalyzed by SrCO3 as a function of temperature. All decolourisation experiments were performed using 1:23 SrCO3 with catalyst loading of 4.0 g·L−<sup>1</sup> at temperatures 20–70 ◦C.

#### *2.4. Degradation Products*

Figure 10 highlights mass spectra generated from the MB degradation products with the mass-to-charge ratios (m/z) of 77, 122, 234, 284 and 303, reported with the possible fragmented ions shown accordingly.

**Figure 10.** Mass spectra of intermediates from the MB degradation after treatment for (**a**) 10 min, (**b**) 25 min and (**c**) 60 min. All experiments were performed by suspending 1:23 SrCO3 in 10 ppm MB (4 g·L−<sup>1</sup> of MB solution) followed by visible light illumination.

The proposed reaction pathway of MB photooxidation over SrCO3 photocatalyst is outlined in Figure 11. The detected degradation products, as identified from fragments based on m/z ratio, are illustrated in blue, while undetectable but expected intermediates [44,45] are presented in black. These results are in general agreement with previous works that report the generated intermediates during the MB photodegradation process [44,45].

**Figure 11.** Proposed photocatalytic degradation pathway of MB. Detected degradation products are illustrated in blue, while expected but undetectable [44] species are presented in black.

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

#### *3.1. Chemicals*

All reagents were used without further purification. Chemicals of HPLC grade were acetic acid (C2H3O2, Merck, Darmstadt, Germany) and acetonitrile (C2H2N, J.T. Baker, CA, USA). Chemicals of AR grade were ethanol (C2H5OH, Merck, Germany), potassium bromide (KBr, Merck, Germany), tert-butanol (C4H10O, Merck, Germany), ammonium acetate (C2H2ONH4, Rankem, Gurugram, India), sodium hydroxide (NaOH, Rankem, India), methylene blue (C16H18N3Cl, Fluka, Saint Louis, MO, USA), strontium carbonate (SrCO3, Fluka, USA), concentrated hydrochloric acid (HCl, Lab Scan, Ireland), mercury(II) sulphate (HgSO4, QRëc, Newzaland), nitric acid (HNO3, Mallinckrodt Chemicals, Phillipsburg, NJ, USA), potassium dichromate (K2Cr2O7, Unilab, Mandaluyong, Philippines), potassium hydrogen phthalate (C8H5KO4, Univar, Redmond to Downers Grove, IL, USA), silver sulphate (AgSO4, Carlo Erba, Barcelona, Spain), strontium hydroxide octahydrate (Sr(OH)2·8H2O, Sigma Aldrich, Saint Louis, MO, USA) and concentrated sulfuric acid (H2SO4, Lab supplies, Spain).

#### *3.2. Synthesis of Strontium Carbonate (SrCO3)*

Strontium carbonate (SrCO3) was synthesized by a solvothermal method modified from the procedure of Zhang et al. [34]. A suspension of 20 g Sr(OH)2·8H2O in ethanol (100 mL) was sonicated in an ultrasonic bath for 20 min, followed by solvothermal treatment in an autoclave at 80, 100, 120 or 150 ◦C for 2 h. The reaction mixtures were left at room temperature to cool down to room temperature. Then, the precipitates were washed with deionized water to remove Sr(OH)2·xH2O, dried and kept in a dry condition at room temperature. After obtaining the optimal treatment temperature, the reaction time was investigated through the above procedure by fixing the treatment temperature at 100 ◦C and varying reaction time between 0.5, 1, 2 or 3 h. The strontium-based samples were prepared by varying the Sr(OH)2·8H2O: ethanol mole ratio as either 1:13, 1:18, 1:23 or 1:27, and then the above procedures were followed using a treatment temperature of 100 ◦C for 2 h.

#### *3.3. Materials Characterisation*

The crystallinity and the phase structure of the samples were investigated using X-ray diffractometry (PXRD, Bruker AXS model D8 advance). The measurements were examined with CuKα radiation between 2<sup>θ</sup> values of 10–80 degrees, at a scan rate of 0.075 degree·min−<sup>1</sup> using accelerating voltage and currents of 40 kV and 40 mA, respectively. Chemical composition and bonding information were probed using Fourier transform infrared spectrophotometry (FT-IR, Elmer model lamda 800). Diffusion reflectance spectra were measured on a UV–VIS spectrophotometer (Agilent Cary 5000) using a scanning rate of 200–1100 nm. Sample morphologies were investigated using scanning electron microscopy (SEM). The thermal decomposition of SrCO3 was monitored using a thermogravimetric analyzer (TGA, TA instruments SDT 2960 Simultaneous DSC-TGA).

#### *3.4. Catalyst Performance Examinations*

SrCO3 samples were dispersed in 10 mL of 10 ppm MB aqueous solution in order to observe the change in colour under dark and visible light irradiation conditions. Before illumination, the suspensions were stirred in the dark for 5 min. Then, suspensions were irradiated using an LED (16 × 12 V EnduraLED 10 W MR16 dimmable 4000 K with λ > 400 nm) [46]. The colour removal efficiency of MB was monitored as a function of degradation time by measuring the absorbance of the dye solution after treatment. In order to terminate the reaction, the photocatalyst was filtered off using a syringe filter (0.45 μm). The absorbance of the dye was then measured, and the concentration of remaining MB was quantified using the absorbance at maximum wavelength (around 664.5 nm) using the Beer Lambert law.

The colour removal efficiency of MB was calculated via Equation (1):

$$\text{Color removal efficiency} = \left(\frac{\text{C}\_0 - \text{C}\_l}{\text{C}\_0}\right) \times 10\tag{1}$$

where *C*<sup>0</sup> is the concentration of fresh MB solution, and *Ct* is the concentration of dye residue after treatment at *t* minutes.

Leaching of strontium ions may be a major cause of photocatalyst deactivation. Therefore, the amount of strontium ions in the filtered MB solution was quantified by flame atomic absorption spectrometry (FAAS, Perkin Elmer, Waltham, MA, USA).

A mass spectrometer (micro TOF MS, Bruker, Billerica, MA, USA) equipped with electrospray ionization (ESI) source was employed to detect MB degradation products. For this, direct injection of the treated MB solution (with 1:23 SrCO3) under visible light irradiation was carried out, with fragments examined over the range m/z 50–700.

#### **4. Conclusions**

In this work, a solvothermal method without any calcination step was employed to prepare a single crystalline phase of strontium carbonate (SrCO3). Ethanol incorporated SrCO3, a visible light responsive SrCO3 material having a bandgap energy of 2.62 eV, was obtained from the solvothermal treatment of hydrated strontium hydroxide in ethanol at Sr:EtOH of 1:23. Nevertheless, the synthesis conditions strongly influence the bandgap energy of SrCO3, as UV responsive SrCO3 material can also be obtained by varying the precursor concentration. The narrow bandgap SrCO3 material can be utilized as a photocatalyst for decolourisation of methylene blue in water under visible light irradiation. Effective decolourisation of 10 ppm methylene blue aqueous solutions was achieved with >99% colour removal efficiencies after 3 h treatment, under visible light irradiation over the 1:23 photocatalyst, using a catalyst loading of 4 g·L−1. The decolourisation is mainly due to photocatalytic processes. The rate constant values showed a direct correlation with temperature, but decolourisation was most rapid at low pH. In addition to the conventional uses of SrCO3 in pyrotechnics and frit manufacturing, synthesized SrCO3 materials have their place as semiconductors and cocatalysts employed in energy

and environmental applications. The key findings of this work highlight that incorporated ethanol in the SrCO3 structure results in a narrowing of the energy bandgap in SrCO3, with the material being a visible light responsive semiconductor and active photocatalyst in dye degradation. Results from this work may suggest alternative synthesis routes to obtain visible responsive SrCO3 materials, for further development of new composites and cocatalysts in broader applications.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4344/10/9/1069/s1. Figure S1: PXRD patterns of Sr-containing samples derived from solvothermal treatments of hydrated strontium hydroxide in ethanol (a) at various solvothermal temperatures, 2 h, Sr:EtOH mole ratios of 1:23 and (b) at various solvothermal treatment times, 100 ◦C, Sr:EtOH mole ratios of 1:23; Figure S2: FTIR spectra of Sr-containing samples derived from solvothermal treatment of hydrated strontium hydroxide in ethanol (a) at various solvothermal temperatures, 2 h, Sr:EtOH mole ratios of 1:23 and (b) at various solvothermal treatment times, 100 ◦C, Sr:EtOH mole ratios of 1:23.

**Author Contributions:** Formal acquisition, investigation and writing—original draft, P.W. writing—review, editing, T.K.; funding acquisition, writing—review, editing and supervision, S.M.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** M.Sc. Student scholarship (for P.W.) was provided by the Center of Excellence for Innovation in Chemistry (PERCH-CIC). This work was partially supported by the Thailand Research Fund (Grant No. RSA5980027 and IRN62W0005) for T.K. and S.M.S., the National Research Council of Thailand for P.W, and by the CIF, Faculty of Science, Mahidol University.

**Acknowledgments:** The authors are grateful for partial financial support from the Thailand Research Fund (Grant No. RSA5980027 and IRN62W0005), the National Research Council of Thailand, and the CIF, Faculty of Science, Mahidol University. PP is thankful for an M.Sc. student scholarship from the Center of Excellence for Innovation in Chemistry (PERCH-CIC).

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

#### **References**


© 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/).

MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Catalysts* Editorial Office E-mail: catalysts@mdpi.com www.mdpi.com/journal/catalysts

MDPI St. Alban-Anlage 66 4052 Basel Switzerland

Tel: +41 61 683 77 34 Fax: +41 61 302 89 18

www.mdpi.com