Effects of Temperature, Oxygen Level, Ionic Strength, and pH on the

Article pubs.acs.org/JPCA

Effects of Temperature, Oxygen Level, Ionic Strength, and pH on the Reaction of Benzene with Hydroxyl Radicals at the Air−Water Interface in Comparison to the Bulk Aqueous Phase Aubrey A. Heath and Kalliat T. Valsaraj*

Downloaded by UNIV OF NEBRASKA-LINCOLN on September 10, 2015 | http://pubs.acs.org Publication Date (Web): July 22, 2015 | doi: 10.1021/acs.jpca.5b05152

Cain Department of Chemical Engineering, Louisiana State University, 212 Jesse Coates Hall, Baton Rouge, Louisiana 70803-7303, United States ABSTRACT: Atmospheric aerosols (e.g., fog droplets) are complex, multiphase mediums. Depending on location, time of day, and/or air mass source, there can be considerable variability within these droplets, relating to temperature, pH, and ionic strength. Due to the droplets’ inherently small size, the reactions that occur within these droplets are determined by bulk aqueous phase and air−water interfacial conditions. In this study, the reaction of benzene and hydroxyl radicals is examined kinetically in a thin-film flow-tube reactor. By varying the aqueous volume (e.g., film thickness) along the length of the reactor, both bulk and interfacial reaction rates are measured from a single system. Temperature, pH, and ionic strength are varied to model conditions typical of fog events. Oxygen-poor conditions are measured to study oxygen’s overall effect on the reaction pathway. Initial rate activation energies and the bulk aqueous phase and interfacial contributions to the overall rate constant are also obtained.

■

INTRODUCTION The air−water interface, composed of bulk interactions (air− sea), bulk air−water dispersion (rain, fog, mist), and bulk-air in contact with surface films (water films on snow, aerosols, or ice), is the largest interface in the environment.1 Although fog particles only range from 1 to 10 μm in diameter, due to their large number density, they provide a high specific surface area.2 The small size associated with fog particles also leads to more complicated chemistry in that the reactions that take place on these particles are determined by both bulk and interfacial conditions, whereas larger droplets (e.g., raindrops) would be predominately affected by bulk phase conditions. Fog is also a complex medium, which, depending on location, time of day, or air mass source, can vary dramatically in temperature (278−293 K in Baton Rouge, LA, USA),3 ionic strength (typical range: 1.1−78 × 10−3 eq·L−1),4−7 and pH (typical range: 2.4−7.2).4−7 It has been shown that aromatic compounds and oxidative species, such as the hydroxyl radical and hydrogen peroxide, have a free energy minimum at the air−water interface;8−10 hence, this reveals that these types of compounds are more concentrated at the interface, as opposed to the bulk gaseous or liquid phases. This interfacial free energy minimum was observed for benzene, the smallest aromatic structure, but was more pronounced for larger polycyclic aromatic hydrocarbons (PAHs).8 The greater surface free energy minimum indicates that it is even more probable for larger PAHs to congregate at the air−water interface than benzene. Moreover, it has been shown that reaction rates are increased at the film surface as opposed to the bulk aqueous phase.11−14 It is suggested that these increased rates occur due to the increased © 2015 American Chemical Society

concentration of compounds at the interface, the large surface diffusion constants,15 and the incomplete solvent-cage effect at the aqueous surface.16,17 The absence of a complete solventcage effect decreases the probability that compounds at the interface will recombine; thus, it is more likely to observe product formation at a higher rate, especially given the high concentration of reactant species.16 Additionally, the lack of a solvent cage at the interface may also enhance photoproduction of some oxidative radicals, such as the hydroxyl radical, from certain oxidative species at the air−water interface, thus leading to their higher surface concentrations and, in turn, higher observed surface reaction rates.18 Conversely, reactions in the bulk solution are restricted by both the solvent-cage effect and small bulk liquid diffusion constants, which results in slower observed reaction rates. Benzene, a United States Environmental Protection Agency (U.S. EPA) listed carcinogen, is a ubiquitous atmospheric pollutant that is released into the environment by industrial emissions, biomass burning, solvent usage, and automobile exhaust.19,20 Given benzene’s inherent stability due to its aromatic structure, the fact that it reacts readily with hydroxyl radicals in both the gas (kOH = 4.3 × 104 μM−1·min−1) and aqueous phases (kOH = 4.7 × 105 μM−1·min−1) is a tribute to the effectiveness of the hydroxyl radical as a benzene scavenger.21,22 The complete reaction pathway of benzene with hydroxyl radicals in the bulk aqueous phase is described in Received: May 29, 2015 Revised: July 7, 2015 Published: July 9, 2015 8527

DOI: 10.1021/acs.jpca.5b05152 J. Phys. Chem. A 2015, 119, 8527−8536

Article


The Journal of Physical Chemistry A

Figure 1. (a) Sketch of the sample flow path and thin film reactor assembly used in all photo-oxidation experiments. Solid and dashed lines represent gas flow and coolant flow, respectively. (b) Cross-section of the thin film reactor.

detail in our previous study.23 In brief, benzene reacts with the hydroxyl radical to form the hydroxycyclohexadienyl radical (CR), which can then react to form two main products: phenol and biphenyl.24−28 Phenol, the predominant oxidation product from this reaction, can be formed either by the disproportionation of CR,26,27,29 the reaction of CR with the hydroxyl radical,29 or the reaction of CR with oxygen.25,28,30,31 Phenol can react further with the hydroxyl radical to form additional oxidation products including 1,4-benzoquinone, hydroquinone, catechol, and resorcinol.32,33 Biphenyl, the less common product, is formed by the dimerization of CR.26,27 The focus of this study is the reaction of benzene with hydroxyl radicals at the air−water interface. Although this reaction has been studied extensively in the gaseous phase,20,25,28,30,34 and somewhat in the liquid phase,23,26,27,35,36 the reaction at the air−water interface at ambient atmospheric conditions has been examined to a lesser degree and predominantly from a theoretical level.37 For this study, hydroxyl radicals are generated by the irradiation of hydrogen peroxide by UVB light in the liquid phase. Both hydrogen peroxide and the hydroxyl radical are naturally present in the atmosphere and show an affinity to concentrate at the interface.9,38−40 The hydroxyl radical is also known to react easily with benzene in both the bulk gaseous (kOH,g = 7.2 × 105 m3·mol−1·s−1)22 and bulk aqueous (kOH,aq = 7.8 × 109 μM−1·s−1)21 phases. Benzene is also the simplest aromatic structure, and it is used in this study as a model compound due to its presence in the atmosphere and its relatively simple reaction pathway.23 As the larger PAHs are more surface active, it is expected that the results derived for benzene will be further enhanced for the larger compounds.8 The effect of interfacial reaction rates is studied by the use of a temperature controlled flow-tube photoreactor assembly. By

adjusting the liquid volume distributed along a known surface area within the reactor, we can in effect adjust the film thickness, or surface area-to-volume ratio; thus, this system gives the advantage that both bulk and interfacial properties can be achieved from a single reactor. Temperature, pH, ionic strength, and oxygen content are varied to model conditions typical of a fog event.

■

EXPERIMENTAL SECTION Reactor Setup. Portions of the reactor apparatus used in these experiments are described in detail in previous publications,12,41 but some changes were made to the original design to better fit the experiments performed in this study. The reactor schematic and cross-section of the photo reactor are shown in Figure 1. UHP-grade air (Air Liquide America L.P., Houston, TX, USA) or UHP-grade nitrogen (Air Liquide America L.P., Houston, TX, USA), supplied from a compressed gas cylinder, was used as the carrier gas. UHP-grade nitrogen was used in the experiments that tested the effects of an oxygen-poor environment. The carrier gas, which was controlled by a mass flow controller (GFC17, Aalborg, Orangeburg, NY, USA), traveled at a flow rate of 70 mL·min−1 toward Chiller #1. Chiller #1, filled with a 50% (v/v) ethylene glycol (Avantor Performance Materials, Inc., Center Valley, PA, USA) and 50% (v/v) deionized water, was a Cole-Parmer Polystat Advanced 15L Heat/Cool Bath (Cole-Parmer Instrument Company, LLC, Vernon Hills, IL, USA), with a possible temperature range of 245−473 K. Chiller #1 was set between 278 and 293 K for the temperature dependent experiments. Inside the chiller bath, two separate flow channels could occur. In Path #1, the carrier gas traveled through a single, double-channel bubbler (Widgett Scientific, Baton Rouge, LA, USA) that contained 8528


Article



reactor, underneath the glass tube. A separate cooling loop utilizing the chiller’s coolant fluid was also placed underneath the glass tube to further decrease the temperature. This twochiller setup allowed for Chiller #2 to be kept several degrees colder than Chiller #1 (set to the reaction temperature), providing ample temperature control. Similarly, in order to decrease the intensity of the UVB bulbs, which was also high, several light filtration methods were used within the reactor. A 3/16 in. borosilicate glass pane (McMaster-Carr, Atlanta, GA, USA) was placed between the upper and lower boxes both to further offset this temperature increase and to filter out some of the strong UVB rays. Likewise, the glass tube and PFA foil further served as barriers to the strong UVB lights. The combination of the borosilicate glass pane, the KG-33 glass tube, and the PFA foil lowers the percent transmittance, which in turn decreases the intensity that the surface of the glass slide receives. Solution Preparation. Several aqueous samples at various film thicknesses were tested in this study. All experiments were at 293.0 ± 0.5 K and under air saturated conditions unless otherwise noted. For the first set of experiments that strictly tested the effect of film thickness (δ = 77.2, 86.6, 135.1, 193.1, and 386.2 μm), LC−MS grade water was used as received. Similarly, for the experiments that compared the effect of film thickness (δ = 77.2, 86.6, 135.1, and 193.1 μm) at different temperatures (278.0 ± 0.5 K and 283.0 ± 0.5 K), LC−MS grade water was also used as received. In order to test the effect of ionic strength at various film thicknesses (δ = 77.2 and 193.1 μm), ammonium sulfate (purim, p.a. ≥99.0%, Sigma-Aldrich, St. Louis, MO, USA) was added to LC−MS grade water, resulting with a final concentration of 0.025 M. For the experiments testing the effect of pH at various film thicknesses (δ = 77.2 and 193.1 μm), either sulfuric acid (ACS grade, BDH Acids-Columbus Chemical, Columbus, OH, USA) or ammonium hydroxide (ACS grade, FisherChemicals, Fair Lawn, NJ, USA) was added to LC−MS grade water. pH was measured for each solution with a pH meter (Oakton Acorn Series pH 6, Oakton Instruments, Vernon Hills, IL, USA) and a ROSS Ultra Combination pH Electrode (ThermoScientific, Beverly, MA, USA). For the experiments that tested the effects of oxygenpoor conditions at various film thicknesses (δ = 77.2 and 193.1 μm), LC−MS grade water was degassed with nitrogen for at least 2 h before being added to the glass slide. In order to create the initial H2O2 stock solution, 30% (w/w) hydrogen peroxide (EMD Chemicals, Inc., Gibbstown, NJ, USA) was diluted to a 1% (w/w) solution in LC−MS grade water. The actual concentration of the H2O2 stock solution was determined by iodometric titration with sodium thiosulfate (EMD Chemicals, Inc., Gibbstown, NJ, USA), standardized with potassium iodate (Mallinckrodt Chemical Works, New York, USA), and a 1% (w/w) starch (Acros Organics, New Jersey, USA) indicator solution. Experimental Procedure. For each experiment, a certain volume, depending on what particular film thickness was being tested, of an aqueous solution (described above) was added to the glass slide. Hydrogen peroxide (H2O2) was also added to the slide in the liquid phase so that its final concentration was 29.4 mM. The volume of H2O2 added to the slide was always 10% of the total slide volume. Dilution of the aqueous phase by H2O2 addition was accounted for in the solution calculations. Once the slide was loaded with the liquid solution, it was placed inside the reactor. The reactor was then sealed, and benzene was allowed to enter the reactor through the gas phase

only LC−MS water (Burdick & Jackson, Muskegon, MI, USA). Path #1 was in use during all nonexperiment times to ensure equilibrium and consistency inside the reactor. Path #2 was made up of two bubblers in series. The first bubbler was a traditional single-channel bubbler filled only with LC−MS water to ensure 100% relative humidity in the incoming air stream, thus preventing dry-out of the liquid phase inside the reactor. The second bubbler was a double-channel bubbler filled with 20% LC−MS water and 80% benzene (anhydrous, 99.8%, Sigma-Aldrich, Co., St. Louis, MO, USA). As the carrier gas stream traveled through the bubbler, benzene was bubbled into the gas phase such that the stream was saturated with both benzene and water. All tubing exiting the chiller was insulated with 1/2 in. thick foam rubber pipe insulation (McMaster-Carr, Atlanta, GA, USA) to minimize heat loss as the stream traveled from the chiller toward the photoreactor assembly. The photoreactor assembly was made up of a 5 cm × 101 cm Kimble KG-33 glass tube coated internally with PFA foil (0.508 mm thickness, McMaster-Carr, Atlanta, GA, USA), a copper tube heat exchanger, and a chemically etched borosilicate glass slide, on which the liquid phase reaction took place. The glass tube and copper heat exchanger setup were described in previous publications,12,41 except that this time a T-type thermocouple wire (Omega Engineering, Inc., Stamford, CT, USA) was placed outside of the glass tube to measure this temperature directly. Additionally, no fans were required for temperature control. The 1000 mm × 40 mm × 2 mm borosilicate glass slide was chemically etched with hydrofluoric acid (ACS grade, BDH Acids-Columbus Chemical, Columbus, OH, USA) and treated with dichlorodimethylsilane (5% in toluene, Supelco, Bellefonte, PA, USA) and a 20% sodium hydroxide (EMD Chemicals Inc., Gibbstown, NJ, USA) solution in LC−MS water in order to create a hydrophilic inner trough and a hydrophobic outer frame with a surface area of 259 cm2.41 By adjusting the aqueous volume placed on the hydrophilic inner trough, a certain surface area-to-volume ratio could be obtained; hence, 1 mL of water corresponds to a film thickness of 38.6 μm. Two insulated aluminum boxes were used to house either the horizontal flow tube reactor (the lower box) or the four UVB light bulbs (15 W, 275−390 nm, UVP LLC, Upland, CA, USA) (the upper box). The light bulbs’ wavelength range (maximum emission wavelength at 302 nm) spanned the highly energetic UVB range of the solar spectrum; hence, this range compares to the solar radiation in the atmosphere. A 22 cm × 100 cm rectangle was cut out of the top of the lower box such that the simulated light could penetrate into the reactor. Since the UVB light bulbs generate substantial heat, the upper box was kept at room temperature by channeling a compressed air stream chilled by using a vortex tube (Arizona Vortex Tube Manufacturing Company, Wickenburg, AZ, USA) throughout the upper box with copper tubing. The vortex tube can decrease the temperature of compressed air at ambient temperature by at least 25 K, which helps to significantly offset the temperature increase caused by the UVB bulbs. In order to further combat the temperature increase, especially in the experiments where the temperature was 283 K or less, Chiller #2 was used. Chiller #2 was identical to Chiller #1, except a copper cooling coil was inside the circulating bath. Compressed air traveled through the cooling coil and was channeled into the lower box of the reactor, where it was split into five separate streams along the length of the 8529


Article



remained constant throughout each experiment, as benzene was constantly replenished in the gas phase with the carrier gas through the double-channel bubbler. Thus, we assume that CA = CA0 = constant, further simplifying the rate expression. Values of the rate constants were obtained by plotting the concentration of phenol or biphenyl versus time and modeling the data using the “Exponential Rise to Maximum: Single, 2 Parameter” and “Polynomial: Linear” expressions, respectively, in SigmaPlot 12.5. Effect of Film Thickness and Temperature. The effect of film thickness was studied by comparing k′D and k′p at several film thicknesses (δ = 77.2, 86.6, 135.1, 193.1, and 386.2 μm) in pure LC−MS grade water at 293 K under air saturated conditions. Rate constants for k′D and k′p for the effect of film thickness experiments are shown in Table 1. From Table 1, it is

for 96 min at the target reaction temperature so that the system had ample time to reach equilibrium. Then, the UVB lights were turned on for a time interval between 0 and 180 min (10 time points in total). After this particular time interval, the UVB lights were turned off and the slide was taken out of the reactor. The entire liquid contents of the slide were quickly washed into a volumetric flask using HPLC grade acetonitrile (EMD Chemical Inc., Gibbstown, NJ, USA). Using acetonitrile, as opposed to water as the slide wash solvent, helped minimize the potential loss of benzene through evaporation during this process, but it is still possible that some evaporation could have occurred. After the contents were removed from the slide, the liquid phase was analyzed immediately by HPLC. Each time point was repeated in triplicate to ensure consistency for all parameters measured; one data point each for phenol, biphenyl, and benzene was obtained from a single slide volume. The gas phase concentration of benzene was also measured for each experiment. The outlet stream from the photoreactor was channeled into a 25 mL solution of HPLC-grade acetonitrile that was contained in a 40 mL amber borosilicate vial (Qorpak, Bridgeville, PA, USA) for 15 min. The amber vial was held in an ice-chilled aqueous solution to minimize benzene evaporation throughout the sample collection. Samples were drawn up from the acetonitrile trap via syringe, diluted further with acetonitrile so that they were not too concentrated for analysis, and then injected immediately onto the HPLC. All liquid phase and gas phase samples were analyzed for benzene, phenol, and biphenyl using an Agilent 1100 HPLC system outfitted with a diode array detector (G1315A) and a programmable fluorescence detector (1046A), as described in detail in our previous study.23

Table 1. k′p and k′D at Five Different Film Thicknesses in a Pure LC−MS Grade Water Solution under Air Saturated Conditions at 293 K 77.2 86.6 135.1 193.1 386.2

1.27 4.94 3.20 3.87 2.22

± ± ± ± ±

0.56 3.00 1.02 1.93 1.34

× × × × ×

k′D (min−1) −3

10 10−4 10−4 10−4 10−4

5.10 3.40 1.80 1.70 1.50

± ± ± ± ±

0.50 1.50 0.60 0.70 0.30

× × × × ×

10−3 10−3 10−3 10−3 10−3

apparent that, for all film thicknesses, k′D is higher than k′p. This agrees with previous data for the reactions of phenol and benzene with hydroxyl radicals in the bulk aqueous phase.21 However, for both k′D and k′p, the rate constant is much higher for the thinner films than the thicker films. We observed a 472% and 240% increase in k′p and k′D, respectively, when the film thickness was reduced from 386.2 to 77.2 μm. It has been shown that rate constants like k′p and k′D can be described as a combination of bulk phase and interfacial rate constants.17,44 The more detailed evaluation of k′p and k′D is shown in eq 3 and eq 4, respectively. K 1 k′p = k′1p + k′2p σA KWA δ (3)

■

RESULTS AND DISCUSSION Data Analysis. Simplified kinetic expressions were derived for phenol and biphenyl, eq 1 and eq 2, respectively, from the rate expressions for the reaction of benzene with hydroxyl radicals, which was outlined in detail in our previous study.23 k p[OH]′ss k ′p CB = (1 − e−kD[OH] ′ss t ) = (1 − e−k ′D t ) CA0 kD[OH]′ss k′D (1)

CC = k′C CA0t CA0

k′p (min−1)

thickness (μm)

k′D = k′1D + k′2D

(2)

where CA0 is the initial concentration of benzene in water, CB is the concentration of phenol in water, CC is the concentration of biphenyl in water, [OH]′ss is the steady state hydroxyl radical aqueous concentration in the presence of benzene, t is time, kp and kD are the rate constants of the reaction of benzene and phenol, respectively, with hydroxyl radicals, k′p and k′D are the rate constants of the reaction of benzene and phenol, respectively, with the steady state hydroxyl radical concentration absorbed into the rate constant, and k′C is the apparent rate constant for biphenyl formation from dimerization of CR. Several assumptions were used to obtain the above simplified kinetic expressions from the full kinetic expressions. First, it was assumed that the rate of disproportionation for the formation of phenol was negligible due to steric hindrance.27 Second, as in previous works, it was assumed that the radical species, the hydroxyl radical and the hydroxycyclohexadienyl radical, CR, were both at a steady state throughout the reaction.23,42,43 Finally, it was assumed that the concentration of benzene

K σA 1 KWA δ

(4)

where k′1p and k′1D are the bulk phase contributions to k′p and k′D, respectively, k′2p and k′2D are the interfacial contributions to k′p and k′D, respectively, KσA is the air−water interface partition constant, KWA is the air−bulk water partition constant, and δ is the film thickness. KWA for benzene at 293 K was estimated from the temperature dependent expression, KWA−1 = KAW = (1/RT)eA−(B/T), where R is gas constant (8.205 × 10−5 atm·m3·K−1·mol−1), T is temperature in kelvins, and A and B are constants for a given compound.2 For benzene, A and B are 5.534 and 3194, respectively.2 This results with a KWA of 5.15 for benzene at 293 K. To our knowledge, no temperature dependent expression exists for KσA. However, several studies have calculated this value at 298 K,44−46 and Braunt and Conklin found values for KσA at several different temperatures, including 291.2 K, at which KσA was found to be 0.62 μm.45 Since this is the closest temperature to our experimental temperature value of 293 K, this value for KσA was used to estimate the k′2p and k′2D calculations. 8530


Article


Again, we observe higher values for k′1D than k′1p, the bulk phase contribution to the rate constants, but this agrees with the literature data on the reaction of phenol and benzene with hydroxyl radicals in the bulk aqueous phase.21 Conversely, within the bounds of error, there is no significant difference between k′2D and k′2p. That said, the rate constants corresponding to the interface are 4−5 orders of magnitude higher than the rate constants associated with the bulk aqueous phase. This indicates that when the film thickness is low (

Effects of Temperature, Oxygen Level, Ionic Strength, and pH on the

Recommend Documents