Environ. Sci. Technol. 2001, 35, 2823-2827
Gas-Phase Photocatalytic Oxidation of Dichlorobutenes NOEL A. HAMILL AND CHRISTOPHER HARDACRE* School of Chemistry, David Keir Building, The Queen’s University of Belfast, Belfast BT9 5AG, Northern Ireland. U.K. JOHANNES A. C. BARTH AND ROBERT M. KALIN Environmental Engineering Research Centre, Department of Civil Engineering, David Keir Building, The Queen’s University of Belfast, Belfast BT9 5AG, Northern Ireland. U.K. JOSEPH F. CUNNINGHAM School of Chemistry, University College Cork, Cork, Ireland
Gas-phase photocatalysis of 1,4-dichlorobut-2-enes and 3,4-dichlorobut-1-ene (DCB) has been studied using TiO2 and 3%WO3/TiO2 supported on SiO2. DCB was found to oxidize efficiently over these catalysts; however, only low rates of CO2 formation were observed. With these chlorinated hydrocarbons, the catalysts were found to deactivate over time, probably via the formation of aldol condensation products of chloroacetaldehyde, which is the predominant intermediate observed. The variation in rate and selectivity of the oxidation reactions with O2 concentration is reported and a mechanism is proposed. Using isotope ratio mass spectrometry, the initial step for the DCB removal has been shown not to be a carbon bond cleavage but is likely to be hydroxyl radical addition to the carboncarbon double bond.
Introduction Semiconductor photocatalysis has been extensively researched as a means of wastewater treatment, TiO2 (in particular Degussa P25) being widely accepted as the most effective photocatalyst in this application due to its inertness, cheapness, and photostability. A key advantage of photocatalytic oxidation is the mineralization of pollutants to carbon dioxide, water, and mineral acids (1). The success of any future photocatalytic water treatment system will be partly determined by its ability to provide maximum mineralization without toxic intermediates, which in some cases can have a higher toxicity and/or stricter emission limits than the original pollutant (e.g. during degradation of the pesticide Diuron (2)). A detailed understanding of the reaction mechanism and mineralization yield is therefore essential when investigating photocatalytic removal of a given pollutant. Although aqueous phase reactions have dominated the field of photocatalysis, a number of gas-phase studies have also been performed. These have concentrated on the reactions of small hydrocarbons such as dicholoroethene (3), trichloroethene (4), chloroform (5), propane (6), and ethene (7) as well as benzene (8) and toluene (9). Gas-phase * Corresponding author phone: +44 (0)28 9027 4592; fax: +44 (0)28 9038 2117; e-mail:
[email protected]. 10.1021/es001939n CCC: $20.00 Published on Web 05/26/2001
2001 American Chemical Society
photooxidation has been shown by these studies to be a facile method of destroying hydrocarbons; however, in the case of chlorinated hydrocarbons the mineralization rates are lower than for non-chlorinated hydrocarbons. Here we have studied the gas-phase photooxidation of 1,4- and 3,4-dichlorobutenes (DCB) over silica supported TiO2 and WO3 doped TiO2. This paper is primarily concerned with the mechanistic and kinetic aspects of DCB oxidation and has used isotope ratio mass spectrometry (IRMS) to help elucidate the reaction mechanism. IRMS is used to determine the 13C/12C ratio of carbonaceous materials, δ13C, which is expressed in parts per thousand difference versus Pee Dee Belemnite (PDB), the internationally accepted standard (‰ vs PDB). The δ13C value is indicative of the carbon source (e.g. atmospheric CO2 is ca. -8‰, whereas manufactured hydrocarbons are isotopically lighter at ca. -30‰). Use of stable isotope IRMS is finding numerous applications in medicine (e.g. breath testing) (10), environmental science (e.g. climate reconstruction) (11), and the food/drink industry (12). Dichlorobutenes were studied because they are extensively used in industry, for example in the manufacture of Neoprene, and they are extremely toxic to marine life even at the part per billion level (13). These molecules are easily air stripped, and the emission limits are strict. In this study we show that DCB does undergo gas-phase photooxidation, but the catalysts are prone to deactivation via aldol condensation reactions. A reaction mechanism based upon hydroxyl radical addition, as opposed to hydrogen atom abstraction, is proposed. Stable isotope IRMS has already been used to investigate the biotic degradation of halomethanes (14), but to the best of our knowledge this is the first study of photocatalysis utilizing stable isotope IRMS.
Experimental Section The photocatalytic degradation of DCB reaction was performed in a flow reactor using a bed of catalyst shown schematically in Figure 1. The reaction vessel was a Pyrex glass tube (13 mm i.d.) secured vertically to allow operation as a fluidized or fixed bed. A grade 2 glass sinter functioned as an inlet gas distributor and a grade 0 sinter at the top of the reactor stopped carryover of catalyst particles without inhibiting gas flow (sinter separation ) 220 mm). 8 × Phillips TL 8W/05 UV lamps provided the illumination (total light intensity 400 µW cm-2, i.e. 7 × 1014 photons cm-2 s-1 at λmax ) 365 nm). A gas mix of 1000 ppmv 3,4-dichlorobut-1-ene (3,4 DCB) in nitrogen (Intergas) was diluted to the desired gas composition using nitrogen (99.999%, total hydrocarbon < 1 ppmv) and oxygen (99.999%, total hydrocarbon < 1 ppmv) prior to the reactor. The gas mixture also contains cis- and trans-1,4-DCB, due to 3,4-DCB isomerization catalyzed by the steel cylinder wall. This ratio remained constant throughout the experiments, and the variation of each isomer was similar. Throughout this paper DCB refers to the total reactant concentration. All joints in the reactor were sealed using Viton “O”-rings which were PTFE coated to prevent dissolution of the DCB in the rubber. All materials used were either glass or stainless steel to limit reaction on the walls. Unless otherwise stated, the mole fractions of oxygen, water, and DCB used in the feed mix were 0.13, 4.0 × 10-3, and 1.0 × 10-4, respectively, carried at a gas flow rate of 180 cm3 min-1 and regulated using a mass flow controller. This flow rate was chosen in order to reduce DCB conversion and overall mineralization, to allow the reaction intermediates to be studied and therefore the reaction mechanism to be investigated. The water concentration was controlled using VOL. 35, NO. 13, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2823
FIGURE 1. Schematic of the continuous flow gas-phase reactor. a pervaporation membrane. Although the controllable range for this method is only ( 2% humidity, greater accuracy was not required since the effluent stream showed little variation with mole fraction of water. The catalysts used were A, 9.1% w/w Degussa P25 TiO2 supported on Davisil 646 silica (Aldrich), and B, 3 mol % (compared with TiO2) of WO3 deposited on catalyst A. The catalyst charge (1 g) yielded a bed height of 3 cm. Catalyst A was prepared by sonicating the TiO2 (0.1 g) and silica (1 g) in deionized water (3 cm3) for 5 min, followed by drying in an oven at 100 °C for 16 h and then calcination at 550 °C for 24 h. Catalyst B was prepared by incipient wetness, and TiO2 (2 g) was added to ammonium paratungstate (0.192 g) dissolved in deionized water (20 cm3) and dried in an oven at 100 °C. The dried powder was bound to the silica using a similar approach to catalyst A except calcination was at 450 °C for 4 h to prevent sintering of the WO3 into larger crystallites (15). Under these operating conditions, reactant mass transfer had no influence on the reaction kinetics since changing the Reynolds number from 0.88 to 1.64 did not change the reaction rate. The light intensity variation was varied by reducing the number of lamps used. A direct linear relationship is found between light intensity and DCB conversion. This is normally associated with processes where electronhole pair recombination is slow and is indicative of low light intensity (16, 17). Mills and Wang (18) have also shown that reaction rates on supported TiO2 films can be directly proportional to I even at higher intensities, because the densely packed catalyst aggregates screen the incident radiation. From this linear relationship and the opacity of the catalyst bed, it is reasonable to assume that the degradation is photon limited under these conditions. Unless otherwise stated, the data described was obtained where the DCB is in pseudo-steady state, i.e. prior to any catalyst deactivation. The reaction was followed using online GC-ECD detection using a Pye Unicam 300 series GC and 12m OV-1 column. Transformation products were identified and quantified using external standards. Due to difficulties in analysis of organic acids (e.g. chloroacetic acid) by GC, the reactor effluent was flushed through deionized water (7 mL) in a 12 mL Exetainer vial for 30 min. Excess sodium sulfate (5 g) and dimethyl sulfate (100 µL) were added, and the vial was sealed and shaken for 1 h. The headspace was analyzed by GC-ECD. Identification and quantification of the chlorinated methyl esters were correlated with retention times and peak areas with known standards (19). Quantification of CO2 and isotope ratio measurements on DCB and CO2 were obtained by flushing 12 cm3 Exetainer vials with reactor effluent for 1 min and injecting into a Europa ANCA TG-II IRMS. All δ values are (0.3‰. Diffuse reflectance infrared spectroscopy (DRIFTS) spectra were collected on a Perkin-Elmer FTIR on the used and fresh catalysts. Backgrounds were normalized using the fresh catalysts as blanks. The spectra obtained were compared 2824
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 13, 2001
FIGURE 2. Reaction time variation for CO2 ([) and DCB (0) for the reaction of 100 ppmv DCB in humidified air over irradiated 9.1% TiO2/SiO2 (13% oxygen, 0.4% water, balance nitrogen, flow rate ) 180 cm3 min-1 catalyst charge ) 1 g).
TABLE 1. Summary of the Time To Reach Steady State and Approximate Relative Mole Fraction for Molecules Detected in the Effluent Stream, for the Reaction of 100 ppmv DCB in Humidified Air over Irradiated 9.1% TiO2/SiO2a molecule
time/min
mole fraction/%
DCB chloroacetaldehyde 2,4-dichlorocrotonaldehyde chloroacetic acid 1,2-dichloroethane CO2
140 90 120 90 140 70
1.3 75 13 2.9 0.8 7.0
a 13% oxygen, 0.4% water, balance nitrogen, flow rate ) 180 cm3 min-1 catalyst charge ) 1 g.
with authentic samples of DCB and reaction intermediates in the liquid phase.
Results and Discussion In the absence of TiO2, trace amounts of reaction did occur. This was attributed to the acid catalyzed reactions on silica due to the large surface area available for reaction on the inlet and outlet glass sinters. This was verified by using silica as a catalyst bed (i.e. catalyst A without TiO2). In this case, the DCB conversion increased to 10%, forming predominantly chloroacetaldehyde as well as traces of 2,3-dichloropropanoic acid which were observed. Figure 2 shows the variation of DCB and CO2 with illumination time upon irradiation in the presence of catalyst A. Carbon dioxide was found to increase with reaction time and reached steady state after 70 min corresponding to 22 ppmv, i.e. only 5.6% mineralization. At this point, only a small concentration of DCB was observed ( 20%. The DCB variation is consistent with the general trends described for chlorinated and non-chlorinated hydrocarbons by Einaga et al. (8). In gas-phase photooxidation of chlorinated hydrocarbons, chlorine radicals are thought to act as augmentative oxidants, which reduce the oxygen concentration required for reaction. The variation of 1,2-dichloroethane formation in this study mirrors that of intermediates formedin the photooxidation of TCE (4). The increase in oxygenated products indicates increased oxygen trapping of photogenerated radicals in line with higher oxygen partial pressure. Reaction Pathway. Coupled with the identification of the intermediates and the variation with oxygen concentration, VOL. 35, NO. 13, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2825
FIGURE 5. Variation in δ13C of reactively formed CO2 as a function of illumination time for the photooxidation of 100 ppmv DCB in humidified air over ([) 9.1% TiO2/SiO2, (2) WO3 doped TiO2/SiO2, and (O) poisoned 9.1% TiO2/SiO2. isotope ratio mass spectrometry has been used to investigate the reaction pathway of DCB in the gas phase. By measuring the isotope ratio of the carbon in the starting material and comparing it with the ratio found in the products, the important reaction steps can be elucidated. The 13C/12C isotope ratio, δ13C, of the DCB was -26.7‰ and was not found to change during the course of the reaction. Since the δ13C for DCB does not change, this implies that carbon bond cleavage is not the initial step in DCB removal. If bond breaking were rate determining, the δ13C for the DCB would be expected to increase, i.e. become enriched in 13C, since those bonds containing 12C would break preferentially and 12C related DCB would be removed faster than 13C. This result is consistent with hydroxyl radical addition to the double bond being the initial step, as opposed to H abstraction, and is in agreement with gas-phase radical mechanisms where H abstraction in alkenes, by OH•, constitutes
