Ozone membrane contactor to intensify gas/liquid mass transfer and contaminants of emerging concern oxidation

1 Ozone membrane contactor to intensify gas/liquid mass transfer and contaminants of emerging concern oxidation Pedro H. Presumidoa,b,c, Rosa Montesd, José B. Quintanad, Rosario Rodild, Manuel Felicianoc, Gianluca Li Pumae, Ana I. Gomesb,f, Vítor J. P. Vilara,b* aLaboratory of Separation and Reaction Engineering – Laboratory of Catalysis and Materials (LSRE-LCM), Chemical Engineering Department, Faculty of Engineering, University of Porto, Rua Dr. Roberto Frias, 4200-465, Porto, Portugal bAssociate Laboratory in Chemical Engineering (ALiCE), Faculty of Engineering, University of Porto, Rua Dr. Roberto Frias, 4200-465, Porto, Portugal cCentro de Investigação de Montanha (CIMO), Instituto Politécnico de Bragança, Campus de Santa Apolónia, 5300-253, Bragança, Portugal dDepartment of Analytical Chemistry, Nutrition and Food Sciences, Institute of Research on Chemical and Biological Analysis (IAQBUS), Universidade de Santiago de Compostela, Constantino Candeira S/N, 15782, Santiago de Compostela, Spain eEnvironmental Nanocatalysis & Photoreaction Engineering, Department of Chemical Engineering, Loughborough University, Loughborough LE11 3TU, United Kingdom fLaboratory for Process Engineering, Environment, Biotechnology and Energy (LEPABE), Faculty of Engineering, University of Porto, Rua Dr. Roberto Frias, 4200- 465 Porto, Portugal *Corresponding author: E-mail addresses: [email protected] (Vítor J.P. Vilar) 2 Contaminants of emerging concern Table S1. List of the 19 CECs selected for testing and respective abbreviations and chemical composition. Contaminants Groups Contaminants Abbreviation Chemical Composition CAS Short-chain perfluoroalkylated substances Heptafluorobutyric acid HFBA C 4 HF 7 O 2 375-22-4 Nonafluoro-1- butanesulfonic acid PFBS C4HF9O3S 29420-49-3 Perfluorooctanoic acid PFOA C 8 HF 15 O 2 335-67-1 Trifluormethanesulfonic acid TFMS CHF3O3S 1493-13-6 Angiotensin II Receptor Blockers Valsartan VSTN C 24 H 29 N 5 O 3 137862-53-4 Irbesartan ISTN C 25 H 28 N 6 O 138402-11-6 Losartan LSTN C 22 H 23 ClN 6 O 124750-99-8 Flame Retardant Melamine MLN C 3 H 6 N 6 108-78-1 Herbicide Diuron DRN C 9 H 10 C l2 N 2 O 330-54-1 Insect Repellent DEET DEET C 12 H 17 NO 134-62-3 Artificial Sweeteners Acesulfame AC-K C 4 H 4 KNO 4 S 55589-62-3 Saccharin SCH C 7 H 5 NO 3 S 81-07-2 Carbamazepine and Metabolites Carbamazepine CBZ C 15 H 12 N 2 O 298-46-4 10,11 Carbamazepineepoxide CBZ-EPX C15H12N2O2 36507-30-9 Beta blockers Atenolol ATNL C 14 H 22 N 2 O 3 29122-68-7 Bisoprolol BSPL C 18 H 31 NO 4 66722-44-9 Hormones 17-beta-estradiol E2 C 18 H 24 O 2 50-28-2 17-alpha- ethynylestradiol EE2 C20H24O2 57-63-6 Non-Steroidal Anti- Inflammatory Drugs Diclofenac DCF C14H11Cl2NO2 15307-86-5 3 Iodometric method for the determination of ozone concentration in liquid phase The ozone concentration was determined using the oxidation of iodide in excess and measuring the produced iodine concentration by titration with sodium thiosulfate. Eqs. S1 and S2 chemically describe this procedure [1]. A known quantity of a solution of 2% KI (w/v) was taken in a conical flask (50 mL), and then the water containing ozone was sampled directly from the reactor into the same conical flask (about 100 mL) and 2.5 mL of 0.5 M H2SO4 was added. Subsequently, the sample of ozonised water and KI solution was immediately titrated with sodium thiosulfate, previously standardised with potassium dichromate (0.01 M). In addition, a starch indicator solution (1% w/v) was used. The ozone mass was calculated from the amount of thiosulfate consumed in the titration (Eq. S3) and the ozone concentration (g O3 m-3) was calculated from the ozone mass obtained through Eq. S4. where m is the mass of ozone (mg), Cthi is the nominal concentration of thiosulfate (mol L- 1), f is the correction factor for thiosulfate, Vthi is the turning volume in the titration (mL), MO3 is the molar mass of ozone (g mol-1), C is the ozone concentration (mg L-1), and V is the volume of the added ozonised water sample (mL). O 3 + 3I−+ 2H+↔O 2 + I 3 −+ H 2 O (S1) I 3 −+2S 2 O 3 2−↔3I−+ S 4 O 6 2− (S2) m = 0.5 × C thi × f × V thi × M O3 (S3) C = m V×103 (S4) 4 Determination of self-decomposition constant of O3 The volumetric mass transfer coefficient (KLa) can be calculated from the mass balance of O3 in the liquid, considering plug flow conditions (Eq. S5): 𝑑𝑑[𝑂𝑂 3 ] 𝑑𝑑𝑑𝑑 = 𝐾𝐾𝐿𝐿𝑎𝑎× ([𝑂𝑂3∗]−[𝑂𝑂3]) −𝑘𝑘𝑑𝑑[𝑂𝑂3] (S5) where [𝑂𝑂3]: dissolved O3 concentration at time t, [𝑂𝑂3∗]: O3 concentration at saturation, KLa: volumetric mass transfer coefficient of O3, and kd: self-decomposition constant of O3. The experiments to determine the self-decomposition constant of O3 were carried out, in batch mode (Fig. S1), with demineralised water at water pH of 3.0 ± 0.2, T = 20ºC, QG = 0.75 Ndm3 min-1, [O3]G = 20 g Nm-3, QL = 50 L h-1. Fig. S1. Schematic representation of laboratory-scale setup of the batch system. Note: DEH – dehumidifier; GEN – ozone generator; GP – gear pump; H-CAT – heat catalyst (ozone destruction); KI – 2% KI solution; MC – ozone measuring cells; MFC – mass flow controller; MS – magnetic stirrer; MSB – magnetic stir bar; O2 – oxygen; O3AN – ozone analyser; TC – temperature controller; Continuous lines – water; Dotted lines – water + ozone gas stream; Dashed lines – ozone gas stream. Ozone generation / measurements O 2 DEH O 3AN H-CAT KI Exhaust GEN GPGP 0 Annular Reaction Zone Water and Gas Streams Ozone Gas 5 After the saturation, the injection of the ozone gas stream in the system was stopped (Fig. S2). Then, the value of kd was calculated by the decrease of ozone concentration in water with time. As reported by other authors [2-4], the decrease of ozone concentration was considered as a first-order process and the self-decomposition constant kd could be calculated from the following equation (Eq. S6): 𝑑𝑑[𝑂𝑂 3 ] 𝑑𝑑𝑑𝑑 = −𝑘𝑘𝑑𝑑× [𝑂𝑂3] (S6) Therefore, integrating Eq. S6 gives Eq. S7: ln [𝑂𝑂 3 ∗] [𝑂𝑂3]= 𝑘𝑘𝑑𝑑×𝑑𝑑 (S7) Fig. S2. Time course of ozone dissolution in water and ozone self-decomposition using the tubular porous borosilicate membrane contactor. Conditions: QL = 50 L h-1; QG = 0.75 Ndm3 min-1; [O3]G = 20.0 ±1.0 g Nm-3; T = 20 ºC; pH = 3; batch mode. A plot of ln([𝑂𝑂3∗]/[𝑂𝑂3]) versus t for the self-decomposition data in Fig. S2 yielded a straight line (R2 = 0.985) with a slope (𝑘𝑘𝑑𝑑) of (131 ± 7) ×10-4 min-1, which was close to the literature values [3, 5]. The KLa value is notably more significant than the kd value (KLa = 0.53 ± 0.02 min-1 >> kd = (131 ± 7) ×10-4 min-1; more than 40 times higher). Thus, the kd value can be considered negligible for these experimental conditions. This means that the mass balance of O3 in the liquid (Eq. S5) can be converted into Eq. S8: 𝑑𝑑[𝑂𝑂 3 ] 𝑑𝑑𝑑𝑑 = 𝐾𝐾𝐿𝐿𝑎𝑎× ([𝑂𝑂3∗]−[𝑂𝑂3]) (S8) 010 20 30 40 50 60 70 80 0 1 2 3 4 5 k d = (131 ± 7) × 10 -4 min -1 R 2 = 0.9850 [O3]L (g m-3) Time (min) K L a = 0.53 ± 0.02 min -1 R 2 = 0.9936 Q G off 6 Dissolved O3 concentration at saturation The ozone concentration at saturation [𝑂𝑂3∗] was 4.65 ± 0.07 g Nm-3 (Fig. S3) for the following operating conditions: [O3]G = 20 g Nm-3; QG = 0.75 Ndm3 min-1; T = 20ºC; pH = 3; batch mode. The ozone saturated value in water is close to the theoretical value found by Henry's law [6]. Fig. S3. Dissolved ozone concentration in the liquid phase as a function of time (a), and the effect of liquid flow rate and Reynolds number on the KLa value (b). Note: 50 � �; 60 � �; and 150 � � L h-1. Conditions: QG = 0.75 Ndm3 min-1; [O3]G = 20.0 ±1.0 g Nm-3; T = 20 ºC; pH = 3; batch mode. 010 20 30 40 50 0 1 2 3 4 5 (b) [O3]L (g m-3) Time (min) (a) 239 298 358 40 50 60 0.0 0.2 0.4 0.6 0.8 1.0 Q L (L h -1 ) K L a (min -1 ) Reynolds number 7 Determination of bubble size distribution, mean bubble size, gas holdup and specific interfacial area The acquisition of bubble images was performed using a high-speed digital video camera (Photron, FASTCAM SA-Z) in three different axial zones of the reactor (top, middle and bottom), under the same conditions of KLa determinations. During image acquisition, a light source and a black sheet (behind the reactor) were used to improve the image's quality and differentiate the boundaries of the gas bubbles. The digital camera was positioned at approximately 30 cm from the reactor wall. Thus, in the axial zone, a filming of 30 s was performed and 10 frames were taken as images (same position as the camera, but at different times of image acquisition). The best images acquired, with at least 500 marked bubbles, were treated, analysed and the results were extracted using the ImageJ software coupled with the "Analyse Particles" tool. The numerical descriptors resulting from the software image analysis are the projected bubble area (Ap). Considering that the bubbles are homogeneous, without deformations and symmetrical, it can be considered for the calculations that the bubbles generated by the ozone contactor are spherical. Therefore, the equivalent spherical bubble diameter (deq) can be calculated by Eq. S9. Fig. S4a shows the average distribution of equivalent bubble diameters. The bubble diameter distribution is detailed using histograms with 5 classes. The first frequency peak appears with bubbles <0.1 mm (representing more than 60% of the bubble size) and the second between 0.1 and 0.3 mm (>30%). The average size of the ozone bubbles formed by the new ozone membrane contactor is approximately 0.20 ± 0.04 mm (Conditions: QL = 150 L h-1; QG = 0.75 Ndm3 min-1; T = 20 ºC). 𝑑𝑑𝑒𝑒𝑒𝑒=�𝐴𝐴𝑝𝑝 𝜋𝜋 (S9) 8 Fig. S4. a) Bubble size distribution in the tubular porous borosilicate membrane contactor. Conditions: QL = 150 L h-1; QG = 0.75 Ndm3 min-1; T = 20 ºC. b) Influence of the gas flow rate (QG) on the gas holdup εG (%). With the same digital images, the volume fraction of the gas phase, also known as gas holdup (εG), was calculated according to Eq. S10 and values between 9.0 and 13.5% were obtained (see Fig. S4b). The methodology involves measuring the height of the liquid without the presence of gas (h0) and the corresponding level (h) when the gas is continuously introduced into the reactor at a given flow rate. The gas-phase volume fraction was measured at least 10 times with a mean error of less than 2%. 𝜀𝜀𝐺𝐺= ℎ−ℎ 0 ℎ (S10) Images obtained allowed the determination of the bubble size distribution and mean bubble size (Conditions: QL = 150 L h-1; QG = 0.75 Ndm3 min-1; T = 20 ºC), corresponding to Sauter mean diameter (d32 = 0.38 mm), gas holdup (εG = 13.5%) and specific interfacial area (a = 2131.6 m-1). 0.0 0.2 0.4 0.6 0.8 0 4 8 12 16 < 0.1 0.1 ≥ d < 0.3 0.3 ≥ d < 0.5 0.5 ≥ d < 0.7 0.7 ≥ d < 1.0 0 50 100 150 200 250 300 350 400 Bubble count D eq (mm) (a) ε G (%) Q G (Ndm 3 min -1 ) (b) 9 Liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis of contaminants of emerging concern A sample volume of 45 μL was directly injected into a Luna C18 100A column (50 mm × 2 mm, 3 µm particle size) supplied by Phenomenex (Torrance, CA, USA) maintained at a constant temperature of 30 °C and 0.2 mL min-1 flow rate. The mobile phases consisted of (A) 1 mM ammonium fluoride in ultrapure water and (B) 1 mM ammonium fluoride in MeOH. The gradient was as follows: 0-1 min, 0% B; 1-8 min, linear gradient to 100% B; 7-13 min, 100% B and finally 13-20 min, 0% B. Nitrogen was used as nebulising and drying gas and Argon was used as collision gas. The analyses were determined in the ESI positive and negative polarities, and the acquisition mode was multiple-reaction monitoring (MRM). Two MRM transitions were used as a quantifier (Q) and qualifier (q) for each compound except in the case of PFBA, where only one intense MRM transition was observed (see Table S2 for detailed information). The quantification was performed by the standard addition method using standards. The method performance parameters are given in Table S3.