Photocatalytic Degradation of Sulfolane Using a LED-Based Photocatalytic Treatment System

: Sulfolane is an emerging industrial pollutant detected in the environments near many oil and gas plants in North America. So far, numerous advanced oxidation processes have been investigated to treat sulfolane in aqueous media. However, there is only a few papers that discuss the degradation of sulfolane using photocatalysis. In this study, photocatalytic degradation of sulfolane using titanium dioxide (TiO 2 ) and reduced graphene oxide TiO 2 composite (RGO-TiO 2 ) in a light-emitting diode (LED) photoreactor was investigated. The impact of different waters (ultrapure water, tap water, and groundwater) and type of irradiation (UVA-LED and mercury lamp) on photocatalytic degradation of sulfolane were also studied. In addition, a reusability test was conducted for the photocatalyst to examine the degradation of sulfolane in three consecutive cycles with new batches of sulfolane-contaminated water. The results show that LED-based photocatalysis was effective in degrading sulfolane in waters even after three photocatalytic cycles. UVA-LEDs displayed more efﬁcient use of photon energy when compared with the mercury lamps as they have a narrow emission spectrum coinciding with the absorption of TiO 2 . The combination of UVA-LED and TiO 2 yielded better performance than UVA-LED and RGO-TiO 2 for the degradation of sulfolane. Much lower sulfolane degradation rates were observed in tap water and groundwater than ultrapure water. photocatalytic performance occurs when used with light at wavelengths under 387 nm [49]. The light sources used in this study emit the highest percent of light at wavelengths below this threshold and, as a result, were able to achieve sufﬁcient photoactivation of TiO 2 to oxidize sulfolane. of


Introduction
Sulfolane (tetrahydrothiophene-1,1-dioxide), an organosulfur compound, is commonly used as an industrial solvent in the Sulfinol process to remove impurities from sour natural gas [1]. It is also used as a solvent for extraction, extractive distillation, production of polymers, and other electrical applications [2]. Sulfolane is highly soluble in water, with a solubility of 1266 g/L at 20 • C [2]. It does not easily volatilize from water or soil, as shown by its low vapor pressure (0.01 kPa at 20 • C) and low Henry's constant (8.9 × 10 −10 atm·m −3 ·mol −1 ). According to adsorption studies conducted on aquifer sediments collected in western Canada, sulfolane displays a low average soil-water partition coefficient (kd) of 0.08 L/kg and an organic carbon partition coefficient (KOC) of log 0.07 [3]. Extensive industrial application of sulfolane in gas processing has led to increasing accounts of accidental release into the environment. Sulfolane contamination varying from a few ppb to hundreds of ppm was reported in groundwater and soil near gas processing plants as well as chemical manufacturing facilities, with the most prevalent cause being the mishandling and improper disposal of process waste and spent sulfolane catalysts [4]. Sulfolane is highly mobile in subsurface waters, eventually reaching ecological and human receptors far from the contaminated source [2]. Toxicity studies conducted on various test species have shown the potential adverse effects sulfolane exposure may have on ecological receptors. Acute doses of sulfolane have been shown to inhibit thermoregulatory response of metabolic rate in rats [5]. Sub-acute levels of sulfolane caused convulsions, inflamed lungs, and death in most test species [6]. A recent study involving zebrafish embryos Graphene/carbon-based TiO2 composites have been reported to display enhanced photocatalytic performance for the degradation of dyes and other compounds when compared with pure TiO2 photocatalysts [36][37][38][39][40]. Li et al. [34] observed that adsorption of reactive black 5 (RB5) by TiO2 was increased from less than 5% to more than 50% with the presence of graphene. A faster RB5 degradation rate was also observed with TiO2-graphene composite. Ruidiaz-Martinez et al. [26] synthesized RGO-TiO2 composites for ethylparaben degradation and found that the photocatalytic degradation rate was increased from 0.031 to 0.096 min −1 when the RGO content in RGO-TiO2 composite was changed from 0% to 7%. However, Yu et al. [35] found that RGO-TiO2 composite containing >3% RGO reduced the catalytic degradation of Bisphenol A as the color of the catalyst becomes darker to shield the light.
The enhancement observed in previous works can be attributed to (1) bandgap reduction and (2) the adsorb and shuttle phenomena. The wavelength of light required to The concentration of total organic carbon (TOC) was measured for photocatalytic degradation of sulfolane using P25. The normalized TOC concentrations, at 30 min intervals and for a total of three hours of irradiation time, are shown in Figure 1b. After three hours, TOC was reduced to 22% of the initial concentration, indicating that 78% mineralization of sulfolane was achieved. Degradation of sulfolane during photocatalytic oxidation took place by way of attack by hydroxyl radicals generated through activation of the photocatalyst by photons (λ ≈ 365 nm) emitted by the UVA-LEDs. The pathway for sulfolane degradation by oxidation is not yet well-studied. Yu et al. [12] proposed that the hydroxyl radical may react with sulfolane to produce intermediate radical species by either the extraction of a hydrogen atom or by reacting with the sulfur atom, triggering the fission of the C-S bonds.
Graphene/carbon-based TiO 2 composites have been reported to display enhanced photocatalytic performance for the degradation of dyes and other compounds when compared with pure TiO 2 photocatalysts [36][37][38][39][40]. Li et al. [34] observed that adsorption of reactive black 5 (RB5) by TiO 2 was increased from less than 5% to more than 50% with the presence of graphene. A faster RB5 degradation rate was also observed with TiO 2 -graphene composite. Ruidiaz-Martinez et al. [26] synthesized RGO-TiO 2 composites for ethylparaben degradation and found that the photocatalytic degradation rate was increased from 0.031 to 0.096 min −1 when the RGO content in RGO-TiO 2 composite was changed from 0% to 7%. However, Yu et al. [35] found that RGO-TiO 2 composite containing >3% RGO reduced the catalytic degradation of Bisphenol A as the color of the catalyst becomes darker to shield the light.
The enhancement observed in previous works can be attributed to (1) bandgap reduction and (2) the adsorb and shuttle phenomena. The wavelength of light required to achieve activation of the photocatalyst is determined by the bandgap energy (E g ) of the photocatalyst. The bandgap of P25-TiO 2 is in the range of 3.10-3.2 eV, corresponding to the energy of photons with a wavelength of 387-400 nm [41,42]. Studies have shown that graphene-based composites can be used to reduce the bandgap of TiO 2 and extend its photocatalytic application into longer wavelengths [43,44]. Doping with RGO, a graphene derivative, also led to better charge separation, slower recombination, and more efficient production of radical species on the surface of TiO 2 [35]. In the case of the adsorb and shuttle phenomena, a mechanism is proposed to explain the advantage of TiO 2 coated with carbon-based adsorbents based on a two-phase approach. The first phase acts as an adsorbent, adsorbing the substrate, which then transfers it via diffusion to the second phase, the photoactivated site [39]. On the RGO phase, a large number of π-π conjugated double bonds facilitate the adsorption of organic molecules onto the graphene surface [35]. In addition, RGO-TiO 2 were reported to have a larger BET (Brunauer, Emmett, and Teller) surface area than TiO 2 [26,34], which might favor adsorption.
The advantages of RGO-P25 due to the two phenomena, however, were not observed in this study, as (1) the LED light source applied in this study has a maximum emission at 365 nm with narrow emission band. Most of the photons emitted from this light source were already available for photocatalysis with P25. Therefore, the reduction of bandgap energy of photocatalyst did not significantly increase the number of photons contributing to photocatalytic degradation of sulfolane. (2) Our dark adsorption experiments showed that the adsorption of sulfolane on both P25 and RGO-P25 particles are less than <3%. The adsorption of sulfolane on RGO-P25 was too weak to demonstrate the benefit of adsorbshuttle. A loss of two-dimensional structure of graphene can be caused by aggregation of graphene during the reduction of graphene from graphene oxide and synthesis of RGO-P25 [45,46]. This significantly alters graphene's electrical conductivity, surface area, and optical transparency, potentially blocking active sites on the surface of TiO 2 . Reducing the surface area of TiO 2 exposed to irradiation and water molecules also reduces electron hole and hydroxyl radical formation, therefore reducing photocatalytic activity [47]. Furthermore, Akhavan et al. [48] reported a decrease of carbon content in TiO 2 -RGO composite by increasing the irradiation time from 4 to 24 h, indicating that a long irradiation might lead to a degradation of RGO by TiO 2 and lower its the catalytic efficiency.

Effect of Type of Lamp on Photocatalytic Degradation of Sulfolane
In order to study the effect of different sources of irradiation on the photocatalytic oxidation of sulfolane, the performances of UVA and UVC fluorescent mercury lamps were compared with that of UVA-LEDs. The normalized concentrations of sulfolane over varying energy doses from each of the irradiation sources tested are shown in Figure 2. The results show that all irradiation sources used were able to degrade over 99% of sulfolane with less than 600 Joules (J) of photon energy. The bandgap energy of TiO 2 defines that optimal Catalysts 2021, 11, 624 5 of 14 photocatalytic performance occurs when used with light at wavelengths under 387 nm [49]. The light sources used in this study emit the highest percent of light at wavelengths below this threshold and, as a result, were able to achieve sufficient photoactivation of TiO 2 to oxidize sulfolane.
In order to study the effect of different sources of irradiation on the photocatalytic oxidation of sulfolane, the performances of UVA and UVC fluorescent mercury lamps were compared with that of UVA-LEDs. The normalized concentrations of sulfolane over varying energy doses from each of the irradiation sources tested are shown in Figure 2.
The results show that all irradiation sources used were able to degrade over 99% of sulfolane with less than 600 Joules (J) of photon energy. The bandgap energy of TiO2 defines that optimal photocatalytic performance occurs when used with light at wavelengths under 387 nm [49]. The light sources used in this study emit the highest percent of light at wavelengths below this threshold and, as a result, were able to achieve sufficient photoactivation of TiO2 to oxidize sulfolane. To quantitatively compare the three lamps, the photon energy required to achieve one order reduction of sulfolane was calculated for each lamp and summarized in Table   Figure 2. Photocatalytic degradation of sulfolane by 2 g/L of P25 using UVA-LEDs, UVA mercury lamps, and UVC mercury lamps.
To quantitatively compare the three lamps, the photon energy required to achieve one order reduction of sulfolane was calculated for each lamp and summarized in Table 1. It is noted that the photon energy calculated here refers to the energy entering the reaction vessel and not that which is emitted by the lamps. For the UVA-LED reactor, 360 J of photon energy was needed for a one log reduction of sulfolane. UVA mercury and UVC mercury systems required 443 and 496 J of photon energy to remove 90% of sulfolane from solution, which corresponds to 23% and 38% more than UVA-LED, respectively. Yu et al. [30] observed similar results from photocatalytic degradation of a recalcitrant pesticide, 2,4-D. They found that UVA mercury lamps took 26% more photon energy than UVA-LED to obtain a one order reduction in 2,4-D concentration. Two possible theories have been proposed by the author to explain this result. First, UVA-LEDs emitted photons at a narrower range of wavelengths within the required range for photoactivation of TiO 2 , which led to more efficient photon utilization. Second, UVA-LED, UVA, and UVC mercury lamps have the maximum emissions at 365, 350, and 254 nm, respectively. A shorter wavelength is associated with a higher photon energy requirement to degrade the contaminant.  only the milli-Q water spiked with sulfolane reached non-detectable levels of sulfolane in the 3 h of irradiation time, while 29% and 32% of sulfolane was remaining in tap water and groundwater at the end of the reaction time, respectively. Degradation of sulfolane in all matrices followed first-order kinetics, with milli-Q water showing the highest rate of degradation. For all matrices tested, no significant decrease in sulfolane concentration (<3%) was observed during the initial 30 min of dark time.  Figure 3 shows the results for sulfolane degradation observed in milli-Q water, groundwater, and tap water. Degradation of sulfolane was observed in all matrices tested, however, only the milli-Q water spiked with sulfolane reached non-detectable levels of sulfolane in the 3 h of irradiation time, while 29% and 32% of sulfolane was remaining in tap water and groundwater at the end of the reaction time, respectively. Degradation of sulfolane in all matrices followed first-order kinetics, with milli-Q water showing the highest rate of degradation. For all matrices tested, no significant decrease in sulfolane concentration (<3%) was observed during the initial 30 min of dark time. Natural organic matter (NOM) and inorganic ions in natural waters may interfere with photocatalytic degradation [50]. Bicarbonate (HCO 3 − ) and carbonate ions (CO 3 2− ) can serve as hydroxyl radical scavengers, which react with OH · to produce CO 3 · , a weak and highly selective oxidant [51]. This reduces the amount of OH · available for reacting with sulfolane. HCO 3 − can also react with h + formed by TiO 2 , adsorbing to the TiO 2 surface and forming a negatively charged layer. This can inhibit the interaction between the TiO 2 surface and the target compound [52]. Studies conducted to understand the effect of inorganic ions on photocatalytic degradation of dyes and other model organic compounds found that, similar to HCO 3 − ions, SO 4 2− and Cl − can adsorb onto the TiO 2 surface and trap h + and OH · [53]. SO 4 2− , Cl − , and HCO 3 − concentrations are significantly higher in the groundwater and tap water tested, when compared to milli-Q water. It is likely that during the initial 30 min of dark time, inorganic ions coated the TiO 2 surface, resulting in reduced photocatalytic degradation of sulfolane. Studies on cations like Ca 2+ and Mg 2+ showed negligible effects on photocatalytic degradation of organic pollutant [54]. This is likely due to their inability to scavenge oxidant species as they already exist at their maximum oxidation state.
Since DIPA and sulfolane are commonly used together in industrial processes, DIPA is often present in sulfolane-contaminated waters [55]. The photocatalytic degradation of samples containing 50 mg/L of sulfolane and varying concentrations of DIPA was studied in this section. The results presented in Figure 4 show that sulfolane degradation was observed in samples with all levels of DIPA concentrations. However, a lag in sulfolane degradation can be observed in the samples containing 25 and 50 mg/L of DIPA. This lag can be explained by the competing effect of DIPA. In this case, the co-contaminant may scavenge photons, absorption sites, and/or photo-induced holes or hydroxyl radicals.
UV-Visible spectroscopy of DIPA showed that no absorption peaks were observed above 300 nm; therefore, photon scavenging by DIPA can be disregarded. Some studies have used TiO 2 photocatalysts to treat aqueous solutions containing DIPA. Mahirah et al. [56] demonstrated that degradation of DIPA by TiO 2 took place mainly by photocatalytic oxidation, and adsorption of DIPA onto the surface of the photocatalyst was minimal in comparison. Since DIPA can be degraded through photocatalysis, it may be using up the hydroxyl radicals and holes produced during photoactivation of TiO 2 . In this case, DIPA may be getting preferentially degraded in the mixtures containing 25 mg/L, and greater, of DIPA.

Recyclability of TiO2-P25
P25 photocatalyst was used for three consecutive photocatalytic cycles to study the performance and reusability of P25. As shown in Figure 5, degradation of sulfolane was achieved in all three photocatalytic cycles, indicating that the TiO2 was still active during the final cycle. However, a slight drop in degradation rate suggests that the activity of the photocatalyst decreased in each consecutive cycle. The photocatalyst loading, degradation rate, and removal of sulfolane are shown in Table 2. About 10% and 18% of photocatalyst was lost during the filtering and cleaning process used to recover the P25 after cycles 1 and 2, respectively. Loss of photocatalyst surface may contribute to the decrease in degradation rate. Another factor may be the deactivation of photocatalyst surface by the products of sulfolane mineralization. Abdullah et al. [57] found that sulfate ions can reduce photocatalytic degradation rate by covering photoactive sites on the surface of the TiO2, deactivating the photocatalyst for degradation of organic molecules. Sulfate ions can be produced as a result of oxidation of sulfolane by hydroxyl radicals [13,14]. As a result, deactivation of photoactive sites by oxidation by-products is a probable cause for the increase in residual sulfolane concentration observed after each consecutive photocatalytic cycle.

Recyclability of TiO 2 -P25
P25 photocatalyst was used for three consecutive photocatalytic cycles to study the performance and reusability of P25. As shown in Figure 5, degradation of sulfolane was achieved in all three photocatalytic cycles, indicating that the TiO 2 was still active during the final cycle. However, a slight drop in degradation rate suggests that the activity of the photocatalyst decreased in each consecutive cycle. The photocatalyst loading, degradation rate, and removal of sulfolane are shown in Table 2. About 10% and 18% of photocatalyst was lost during the filtering and cleaning process used to recover the P25 after cycles 1 and 2, respectively. Loss of photocatalyst surface may contribute to the decrease in degradation rate. Another factor may be the deactivation of photocatalyst surface by the products of sulfolane mineralization. Abdullah et al. [57] found that sulfate ions can reduce photocatalytic degradation rate by covering photoactive sites on the surface of the TiO 2 , deactivating the photocatalyst for degradation of organic molecules. Sulfate ions can be produced as a result of oxidation of sulfolane by hydroxyl radicals [13,14]. As a result, deactivation of photoactive sites by oxidation by-products is a probable cause for the increase in residual sulfolane concentration observed after each consecutive photocatalytic cycle.

Chemicals
Sulfolane with 99% purity, ACS grade dichloromethane (DCM), and DIPA with 99% purity were obtained from Sigma Aldrich, Canada (Oakville, ON, Canada). Powdered TiO2-P25 was procured from Evonik Canada Inc. (Burlington ON, Canada). Graphene oxide suspension was obtained from ACS Chemical Inc. (Point Pleasant, NJ, USA). RGO-P25 composites were synthesized in the lab using the hydrothermal method described by Li et al. [34] and Zhang et al. [58]. Milli-Q water was used in most experiments, except where groundwater and tap water were used. Properties for the tap water and groundwater used in this study are listed in Table 3.

Chemicals
Sulfolane with 99% purity, ACS grade dichloromethane (DCM), and DIPA with 99% purity were obtained from Sigma Aldrich, Canada (Oakville, ON, Canada). Powdered TiO 2 -P25 was procured from Evonik Canada Inc. (Burlington ON, Canada). Graphene oxide suspension was obtained from ACS Chemical Inc. (Point Pleasant, NJ, USA). RGO-P25 composites were synthesized in the lab using the hydrothermal method described by Li et al. [34] and Zhang et al. [58]. Milli-Q water was used in most experiments, except where groundwater and tap water were used. Properties for the tap water and groundwater used in this study are listed in Table 3.

Photoreactor
An LED photoreactor, originally fabricated by Yu et al. [30], was used for most of the photocatalytic experiments. This cylindrical photoreactor is lined on the inside with 90 UV LEDs (λ max = 365 nm, full-width half-maximum = 12 nm, NSHU551B), arranged in 6 rows of 15 LEDs, and fixed with a fan. Figure 6a shows a vertical cross-section of this LED photoreactor. An LZC-ORG reactor (Luzchem Research Inc., Ottawa, ON, Canada), with the capacity to hold 10 fluorescent lamps, was obtained from Luzchem Research Inc. for the experiments using mercury lamps (Figure 6b). Ten black lamps (LZC-UVA16) centered at 350 nm, and germicidal lamps (LZC-UVC16), at 254 nm, were installed in this reactor for the UVA and UVC mercury lamp experiments, respectively. The number of photons entering the reaction beaker per unit time in both reactors were determined using ferrioxalate actinometry [59]. They were 9.06 × 10 16 , 1.74 × 10 17 , and 1.85 × 10 17 photon/s for the LED photoreactor, the Luzchem reactor equipped with UVA lamps, and the Luzchem reactor equipped with UVC lamps, respectively.

Experimental Procedure
All experiments were conducted in batch mode in either the LED or Luzchem photoreactors, as shown in Figure 6. Reactions were carried out with 80 mL of solution in either a 100 mL glass beaker (for reactions in UVA-LED) or 150 mL quartz beaker (for UVA and UVC fluorescent mercury lamps). The 50 mg/L solution of sulfolane was prepared using pure sulfolane and the water of interest along with 2 g/L of powdered photocatalyst (either P25 or RGO-P25). A 30 min dark experiment was conducted at the beginning of every experiment to observe the amount of sulfolane adsorbed to the photocatalyst in the absence of light. The solution was continuously stirred throughout the dark and irradiated experiments to homogenize the solution and to ensure adequate distribution of sulfolane and photocatalyst throughout the field of irradiation. Samples were collected at predetermined time intervals, immediately filtered using 0.2 µm polytetrafluoroethylene (PTFE) filters and stored in the dark before extraction and analysis. Additional parameters for each set of experiments can be found in Table 4. It is noted that all experiments were conducted in duplicate, and the average concentration with the error bar indicating the variation between the duplicates were reported in all figures.

Analytical Methods
Extraction of sulfolane was carried out using 1.5 mL of the water sample and 1.5 mL of DCM in a liquid-liquid extraction. The samples were shaken in a mechanical shaker for 30 min. The DCM phase was then extracted and analyzed. An Agilent 6890 gas chromatograph (GC) (Agilent, Santa Clara, CA, USA), equipped with a flame ionization detector (FID) and an auto sampler, was used to analyze the extracted sulfolane samples in DCM. The Agilent ChemStation software, (Agilent, Santa Clara, CA, USA) was used to carry out data acquisition and calibration. Chromatographic separation was achieved on a fused silica capillary column (ZB 5MSI, Phenomenex, Torrance, CA, USA) upon injection of the DCM samples into the GC. High-purity helium, with a head pressure of 250 kPa, was used as the carrier gas. Sample injection was adjusted to splitless mode with a 1.00 mL injection volume and the injection port temperature was 165 • C. The initial temperature was set to 90 • C, held constant for 2 min, then ramped up to 175 • C at a rate of 10 • C/min, where it was held constant for 3 min. The FID detector temperature was set to 330 • C.

Experimental Procedure
All experiments were conducted in batch mode in either the LED or Luzchem photoreactors, as shown in Figure 6. Reactions were carried out with 80 mL of solution in either a 100 mL glass beaker (for reactions in UVA-LED) or 150 mL quartz beaker (for UVA and UVC fluorescent mercury lamps). The 50 mg/L solution of sulfolane was prepared using pure sulfolane and the water of interest along with 2 g/L of powdered photocatalyst (either P25 or RGO-P25). A 30 min dark experiment was conducted at the beginning of every experiment to observe the amount of sulfolane adsorbed to the photocatalyst in the absence of light. The solution was continuously stirred throughout the dark and irradiated experiments to homogenize the solution and to ensure adequate distribution of sulfolane and photocatalyst throughout the field of irradiation. Samples were collected at predetermined time intervals, immediately filtered using 0.2 μm polytetrafluoroethylene

Conclusions
In this paper, a comprehensive study on the photocatalytic degradation of sulfolane in water using UVA-LED was conducted. The results showed that UVA-LED-based photocatalysis is a promising technology to treat sulfolane in aqueous media as it does not require large amounts of chemicals and utilizes less photon energy than mercury lamps. TiO 2 integrated with UVA-LED can degrade more than 99% of sulfolane from water over three hours of irradiation. Extended irradiation can lead to complete mineralization of sulfolane. A recyclability study on TiO 2 showed that it can be used for multiple cycles without significantly losing its photocatalytic activity. Compared to UVA mercury lamps, UVA-LEDs saved 23% of photon energy consumption in achieving one log reduction of sulfolane. TiO 2 -RGO composite synthesized through the hydrothermal method did not display enhanced photocatalytic performance for the degradation of sulfolane due to its weak adsorption on TiO 2 -RGO composite.
Photocatalytic degradation of sulfolane in groundwater and tap water were observed to be slower than that in ultrapure water due to the high sulfate, chloride, and bicarbonate content in the tap water and groundwater. The sulfolane degradation rate decreased by 43% and 74% in tap water and groundwater respectively, when compared with milli-Q water. The presence of DIPA (>25 mg/L), a co-contaminant frequently detected in the sulfolane-contaminated water, also inhibited photocatalytic degradation of sulfolane.